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"
@@ -1,290931 +1,3 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 6.691332179234268,
-  "eval_steps": 500,
-  "global_step": 41550,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.0001610370787873908,
-      "grad_norm": 9.459293323743623e-06,
-      "learning_rate": 2e-05,
-      "loss": 46.0,
-      "step": 1
-    },
-    {
-      "epoch": 0.0003220741575747816,
-      "grad_norm": 1.8449256458552554e-05,
-      "learning_rate": 4e-05,
-      "loss": 46.0,
-      "step": 2
-    },
-    {
-      "epoch": 0.0004831112363621724,
-      "grad_norm": 1.7989095795201138e-05,
-      "learning_rate": 6e-05,
-      "loss": 46.0,
-      "step": 3
-    },
-    {
-      "epoch": 0.0006441483151495632,
-      "grad_norm": 1.1759363587771077e-05,
-      "learning_rate": 8e-05,
-      "loss": 46.0,
-      "step": 4
-    },
-    {
-      "epoch": 0.000805185393936954,
-      "grad_norm": 1.139960022555897e-05,
-      "learning_rate": 0.0001,
-      "loss": 46.0,
-      "step": 5
-    },
-    {
-      "epoch": 0.0009662224727243448,
-      "grad_norm": 1.4804975762672257e-05,
-      "learning_rate": 0.00012,
-      "loss": 46.0,
-      "step": 6
-    },
-    {
-      "epoch": 0.0011272595515117356,
-      "grad_norm": 2.1032992663094774e-05,
-      "learning_rate": 0.00014,
-      "loss": 46.0,
-      "step": 7
-    },
-    {
-      "epoch": 0.0012882966302991264,
-      "grad_norm": 1.5429051927640103e-05,
-      "learning_rate": 0.00016,
-      "loss": 46.0,
-      "step": 8
-    },
-    {
-      "epoch": 0.001449333709086517,
-      "grad_norm": 1.1477978659968358e-05,
-      "learning_rate": 0.00018,
-      "loss": 46.0,
-      "step": 9
-    },
-    {
-      "epoch": 0.001610370787873908,
-      "grad_norm": 1.7123571524280123e-05,
-      "learning_rate": 0.0002,
-      "loss": 46.0,
-      "step": 10
-    },
-    {
-      "epoch": 0.0017714078666612987,
-      "grad_norm": 1.0772208952403162e-05,
-      "learning_rate": 0.00019999999999998722,
-      "loss": 46.0,
-      "step": 11
-    },
-    {
-      "epoch": 0.0019324449454486896,
-      "grad_norm": 9.808137292566244e-06,
-      "learning_rate": 0.0001999999999999488,
-      "loss": 46.0,
-      "step": 12
-    },
-    {
-      "epoch": 0.0020934820242360804,
-      "grad_norm": 1.8400718545308337e-05,
-      "learning_rate": 0.00019999999999988484,
-      "loss": 46.0,
-      "step": 13
-    },
-    {
-      "epoch": 0.0022545191030234712,
-      "grad_norm": 2.247749398520682e-05,
-      "learning_rate": 0.00019999999999979523,
-      "loss": 46.0,
-      "step": 14
-    },
-    {
-      "epoch": 0.002415556181810862,
-      "grad_norm": 1.1134417945868336e-05,
-      "learning_rate": 0.0001999999999996801,
-      "loss": 46.0,
-      "step": 15
-    },
-    {
-      "epoch": 0.002576593260598253,
-      "grad_norm": 1.4492045920633245e-05,
-      "learning_rate": 0.0001999999999995393,
-      "loss": 46.0,
-      "step": 16
-    },
-    {
-      "epoch": 0.0027376303393856437,
-      "grad_norm": 1.4358376574818976e-05,
-      "learning_rate": 0.00019999999999937293,
-      "loss": 46.0,
-      "step": 17
-    },
-    {
-      "epoch": 0.002898667418173034,
-      "grad_norm": 1.1229038136661984e-05,
-      "learning_rate": 0.00019999999999918098,
-      "loss": 46.0,
-      "step": 18
-    },
-    {
-      "epoch": 0.003059704496960425,
-      "grad_norm": 1.850031731009949e-05,
-      "learning_rate": 0.0001999999999989634,
-      "loss": 46.0,
-      "step": 19
-    },
-    {
-      "epoch": 0.003220741575747816,
-      "grad_norm": 1.7780901544028893e-05,
-      "learning_rate": 0.00019999999999872027,
-      "loss": 46.0,
-      "step": 20
-    },
-    {
-      "epoch": 0.0033817786545352066,
-      "grad_norm": 1.23179188449285e-05,
-      "learning_rate": 0.00019999999999845152,
-      "loss": 46.0,
-      "step": 21
-    },
-    {
-      "epoch": 0.0035428157333225975,
-      "grad_norm": 1.1472806363599375e-05,
-      "learning_rate": 0.00019999999999815716,
-      "loss": 46.0,
-      "step": 22
-    },
-    {
-      "epoch": 0.0037038528121099883,
-      "grad_norm": 9.506068636255804e-06,
-      "learning_rate": 0.00019999999999783724,
-      "loss": 46.0,
-      "step": 23
-    },
-    {
-      "epoch": 0.003864889890897379,
-      "grad_norm": 1.6990292351692915e-05,
-      "learning_rate": 0.0001999999999974917,
-      "loss": 46.0,
-      "step": 24
-    },
-    {
-      "epoch": 0.00402592696968477,
-      "grad_norm": 1.920577051350847e-05,
-      "learning_rate": 0.00019999999999712059,
-      "loss": 46.0,
-      "step": 25
-    },
-    {
-      "epoch": 0.004186964048472161,
-      "grad_norm": 1.57214544742601e-05,
-      "learning_rate": 0.00019999999999672388,
-      "loss": 46.0,
-      "step": 26
-    },
-    {
-      "epoch": 0.004348001127259552,
-      "grad_norm": 2.9738939701928757e-05,
-      "learning_rate": 0.00019999999999630155,
-      "loss": 46.0,
-      "step": 27
-    },
-    {
-      "epoch": 0.0045090382060469425,
-      "grad_norm": 1.2327154763625003e-05,
-      "learning_rate": 0.00019999999999585364,
-      "loss": 46.0,
-      "step": 28
-    },
-    {
-      "epoch": 0.004670075284834333,
-      "grad_norm": 1.3533114724850748e-05,
-      "learning_rate": 0.00019999999999538014,
-      "loss": 46.0,
-      "step": 29
-    },
-    {
-      "epoch": 0.004831112363621724,
-      "grad_norm": 1.5398421965073794e-05,
-      "learning_rate": 0.00019999999999488103,
-      "loss": 46.0,
-      "step": 30
-    },
-    {
-      "epoch": 0.004992149442409115,
-      "grad_norm": 1.6816700735944323e-05,
-      "learning_rate": 0.00019999999999435636,
-      "loss": 46.0,
-      "step": 31
-    },
-    {
-      "epoch": 0.005153186521196506,
-      "grad_norm": 1.3670231965079438e-05,
-      "learning_rate": 0.00019999999999380604,
-      "loss": 46.0,
-      "step": 32
-    },
-    {
-      "epoch": 0.005314223599983897,
-      "grad_norm": 1.3132093954482116e-05,
-      "learning_rate": 0.00019999999999323017,
-      "loss": 46.0,
-      "step": 33
-    },
-    {
-      "epoch": 0.0054752606787712875,
-      "grad_norm": 2.7774303816840984e-05,
-      "learning_rate": 0.00019999999999262868,
-      "loss": 46.0,
-      "step": 34
-    },
-    {
-      "epoch": 0.005636297757558678,
-      "grad_norm": 2.077547469525598e-05,
-      "learning_rate": 0.0001999999999920016,
-      "loss": 46.0,
-      "step": 35
-    },
-    {
-      "epoch": 0.005797334836346068,
-      "grad_norm": 1.980315391847398e-05,
-      "learning_rate": 0.00019999999999134892,
-      "loss": 46.0,
-      "step": 36
-    },
-    {
-      "epoch": 0.005958371915133459,
-      "grad_norm": 2.2662210540147498e-05,
-      "learning_rate": 0.00019999999999067067,
-      "loss": 46.0,
-      "step": 37
-    },
-    {
-      "epoch": 0.00611940899392085,
-      "grad_norm": 2.1207339159445837e-05,
-      "learning_rate": 0.00019999999998996683,
-      "loss": 46.0,
-      "step": 38
-    },
-    {
-      "epoch": 0.006280446072708241,
-      "grad_norm": 1.7205984477186576e-05,
-      "learning_rate": 0.00019999999998923738,
-      "loss": 46.0,
-      "step": 39
-    },
-    {
-      "epoch": 0.006441483151495632,
-      "grad_norm": 2.1629894035868347e-05,
-      "learning_rate": 0.00019999999998848234,
-      "loss": 46.0,
-      "step": 40
-    },
-    {
-      "epoch": 0.006602520230283022,
-      "grad_norm": 2.7965374101768248e-05,
-      "learning_rate": 0.00019999999998770166,
-      "loss": 46.0,
-      "step": 41
-    },
-    {
-      "epoch": 0.006763557309070413,
-      "grad_norm": 2.5096620447584428e-05,
-      "learning_rate": 0.00019999999998689542,
-      "loss": 46.0,
-      "step": 42
-    },
-    {
-      "epoch": 0.006924594387857804,
-      "grad_norm": 3.5594304790720344e-05,
-      "learning_rate": 0.0001999999999860636,
-      "loss": 46.0,
-      "step": 43
-    },
-    {
-      "epoch": 0.007085631466645195,
-      "grad_norm": 3.1788349588168785e-05,
-      "learning_rate": 0.00019999999998520618,
-      "loss": 46.0,
-      "step": 44
-    },
-    {
-      "epoch": 0.007246668545432586,
-      "grad_norm": 2.8515927624539472e-05,
-      "learning_rate": 0.00019999999998432315,
-      "loss": 46.0,
-      "step": 45
-    },
-    {
-      "epoch": 0.007407705624219977,
-      "grad_norm": 2.7847892852150835e-05,
-      "learning_rate": 0.0001999999999834145,
-      "loss": 46.0,
-      "step": 46
-    },
-    {
-      "epoch": 0.007568742703007367,
-      "grad_norm": 3.0626124498667195e-05,
-      "learning_rate": 0.00019999999998248033,
-      "loss": 46.0,
-      "step": 47
-    },
-    {
-      "epoch": 0.007729779781794758,
-      "grad_norm": 2.7933489036513492e-05,
-      "learning_rate": 0.00019999999998152052,
-      "loss": 46.0,
-      "step": 48
-    },
-    {
-      "epoch": 0.007890816860582149,
-      "grad_norm": 3.10891009576153e-05,
-      "learning_rate": 0.0001999999999805351,
-      "loss": 46.0,
-      "step": 49
-    },
-    {
-      "epoch": 0.00805185393936954,
-      "grad_norm": 1.9566661649150774e-05,
-      "learning_rate": 0.00019999999997952412,
-      "loss": 46.0,
-      "step": 50
-    },
-    {
-      "epoch": 0.00821289101815693,
-      "grad_norm": 5.5264958064071834e-05,
-      "learning_rate": 0.00019999999997848754,
-      "loss": 46.0,
-      "step": 51
-    },
-    {
-      "epoch": 0.008373928096944322,
-      "grad_norm": 2.4125354684656486e-05,
-      "learning_rate": 0.00019999999997742532,
-      "loss": 46.0,
-      "step": 52
-    },
-    {
-      "epoch": 0.008534965175731712,
-      "grad_norm": 3.317361915833317e-05,
-      "learning_rate": 0.00019999999997633757,
-      "loss": 46.0,
-      "step": 53
-    },
-    {
-      "epoch": 0.008696002254519103,
-      "grad_norm": 3.8641785067738965e-05,
-      "learning_rate": 0.00019999999997522417,
-      "loss": 46.0,
-      "step": 54
-    },
-    {
-      "epoch": 0.008857039333306494,
-      "grad_norm": 1.588473423907999e-05,
-      "learning_rate": 0.0001999999999740852,
-      "loss": 46.0,
-      "step": 55
-    },
-    {
-      "epoch": 0.009018076412093885,
-      "grad_norm": 3.6028392059961334e-05,
-      "learning_rate": 0.00019999999997292065,
-      "loss": 46.0,
-      "step": 56
-    },
-    {
-      "epoch": 0.009179113490881276,
-      "grad_norm": 3.70649722754024e-05,
-      "learning_rate": 0.0001999999999717305,
-      "loss": 46.0,
-      "step": 57
-    },
-    {
-      "epoch": 0.009340150569668667,
-      "grad_norm": 3.086036304011941e-05,
-      "learning_rate": 0.0001999999999705147,
-      "loss": 46.0,
-      "step": 58
-    },
-    {
-      "epoch": 0.009501187648456057,
-      "grad_norm": 3.073871630476788e-05,
-      "learning_rate": 0.00019999999996927337,
-      "loss": 46.0,
-      "step": 59
-    },
-    {
-      "epoch": 0.009662224727243448,
-      "grad_norm": 5.74203186261002e-05,
-      "learning_rate": 0.00019999999996800642,
-      "loss": 46.0,
-      "step": 60
-    },
-    {
-      "epoch": 0.009823261806030839,
-      "grad_norm": 1.878445800684858e-05,
-      "learning_rate": 0.0001999999999667139,
-      "loss": 46.0,
-      "step": 61
-    },
-    {
-      "epoch": 0.00998429888481823,
-      "grad_norm": 3.903669130522758e-05,
-      "learning_rate": 0.00019999999996539575,
-      "loss": 46.0,
-      "step": 62
-    },
-    {
-      "epoch": 0.01014533596360562,
-      "grad_norm": 6.173171277623624e-05,
-      "learning_rate": 0.000199999999964052,
-      "loss": 46.0,
-      "step": 63
-    },
-    {
-      "epoch": 0.010306373042393012,
-      "grad_norm": 3.220455255359411e-05,
-      "learning_rate": 0.0001999999999626827,
-      "loss": 46.0,
-      "step": 64
-    },
-    {
-      "epoch": 0.010467410121180402,
-      "grad_norm": 6.513421976706013e-05,
-      "learning_rate": 0.00019999999996128778,
-      "loss": 46.0,
-      "step": 65
-    },
-    {
-      "epoch": 0.010628447199967793,
-      "grad_norm": 4.503427044255659e-05,
-      "learning_rate": 0.00019999999995986726,
-      "loss": 46.0,
-      "step": 66
-    },
-    {
-      "epoch": 0.010789484278755184,
-      "grad_norm": 5.680751200998202e-05,
-      "learning_rate": 0.00019999999995842113,
-      "loss": 46.0,
-      "step": 67
-    },
-    {
-      "epoch": 0.010950521357542575,
-      "grad_norm": 5.3842857596464455e-05,
-      "learning_rate": 0.00019999999995694943,
-      "loss": 46.0,
-      "step": 68
-    },
-    {
-      "epoch": 0.011111558436329966,
-      "grad_norm": 3.435022517805919e-05,
-      "learning_rate": 0.00019999999995545212,
-      "loss": 46.0,
-      "step": 69
-    },
-    {
-      "epoch": 0.011272595515117357,
-      "grad_norm": 4.8579728172626346e-05,
-      "learning_rate": 0.00019999999995392925,
-      "loss": 46.0,
-      "step": 70
-    },
-    {
-      "epoch": 0.011433632593904747,
-      "grad_norm": 6.448462227126583e-05,
-      "learning_rate": 0.00019999999995238077,
-      "loss": 46.0,
-      "step": 71
-    },
-    {
-      "epoch": 0.011594669672692137,
-      "grad_norm": 3.0377606890397146e-05,
-      "learning_rate": 0.00019999999995080667,
-      "loss": 46.0,
-      "step": 72
-    },
-    {
-      "epoch": 0.011755706751479527,
-      "grad_norm": 4.2671086703194305e-05,
-      "learning_rate": 0.000199999999949207,
-      "loss": 46.0,
-      "step": 73
-    },
-    {
-      "epoch": 0.011916743830266918,
-      "grad_norm": 5.8799378166440874e-05,
-      "learning_rate": 0.0001999999999475817,
-      "loss": 46.0,
-      "step": 74
-    },
-    {
-      "epoch": 0.012077780909054309,
-      "grad_norm": 3.1305702577810735e-05,
-      "learning_rate": 0.00019999999994593085,
-      "loss": 46.0,
-      "step": 75
-    },
-    {
-      "epoch": 0.0122388179878417,
-      "grad_norm": 4.199046816211194e-05,
-      "learning_rate": 0.00019999999994425437,
-      "loss": 46.0,
-      "step": 76
-    },
-    {
-      "epoch": 0.01239985506662909,
-      "grad_norm": 4.404900028021075e-05,
-      "learning_rate": 0.0001999999999425523,
-      "loss": 46.0,
-      "step": 77
-    },
-    {
-      "epoch": 0.012560892145416482,
-      "grad_norm": 4.4940319639863446e-05,
-      "learning_rate": 0.0001999999999408247,
-      "loss": 46.0,
-      "step": 78
-    },
-    {
-      "epoch": 0.012721929224203872,
-      "grad_norm": 5.616362614091486e-05,
-      "learning_rate": 0.00019999999993907143,
-      "loss": 46.0,
-      "step": 79
-    },
-    {
-      "epoch": 0.012882966302991263,
-      "grad_norm": 4.128489672439173e-05,
-      "learning_rate": 0.00019999999993729257,
-      "loss": 46.0,
-      "step": 80
-    },
-    {
-      "epoch": 0.013044003381778654,
-      "grad_norm": 3.650110375019722e-05,
-      "learning_rate": 0.00019999999993548814,
-      "loss": 46.0,
-      "step": 81
-    },
-    {
-      "epoch": 0.013205040460566045,
-      "grad_norm": 3.355491207912564e-05,
-      "learning_rate": 0.0001999999999336581,
-      "loss": 46.0,
-      "step": 82
-    },
-    {
-      "epoch": 0.013366077539353436,
-      "grad_norm": 3.100166577496566e-05,
-      "learning_rate": 0.00019999999993180247,
-      "loss": 46.0,
-      "step": 83
-    },
-    {
-      "epoch": 0.013527114618140827,
-      "grad_norm": 4.128339787712321e-05,
-      "learning_rate": 0.00019999999992992127,
-      "loss": 46.0,
-      "step": 84
-    },
-    {
-      "epoch": 0.013688151696928217,
-      "grad_norm": 5.978300396236591e-05,
-      "learning_rate": 0.00019999999992801443,
-      "loss": 46.0,
-      "step": 85
-    },
-    {
-      "epoch": 0.013849188775715608,
-      "grad_norm": 6.463456520577893e-05,
-      "learning_rate": 0.00019999999992608203,
-      "loss": 46.0,
-      "step": 86
-    },
-    {
-      "epoch": 0.014010225854502999,
-      "grad_norm": 4.458018156583421e-05,
-      "learning_rate": 0.00019999999992412402,
-      "loss": 46.0,
-      "step": 87
-    },
-    {
-      "epoch": 0.01417126293329039,
-      "grad_norm": 3.9894523069960997e-05,
-      "learning_rate": 0.00019999999992214042,
-      "loss": 46.0,
-      "step": 88
-    },
-    {
-      "epoch": 0.01433230001207778,
-      "grad_norm": 6.879306602058932e-05,
-      "learning_rate": 0.00019999999992013123,
-      "loss": 46.0,
-      "step": 89
-    },
-    {
-      "epoch": 0.014493337090865172,
-      "grad_norm": 7.311927038244903e-05,
-      "learning_rate": 0.00019999999991809642,
-      "loss": 46.0,
-      "step": 90
-    },
-    {
-      "epoch": 0.014654374169652562,
-      "grad_norm": 7.103729149093851e-05,
-      "learning_rate": 0.00019999999991603603,
-      "loss": 46.0,
-      "step": 91
-    },
-    {
-      "epoch": 0.014815411248439953,
-      "grad_norm": 6.073687472962774e-05,
-      "learning_rate": 0.00019999999991395005,
-      "loss": 46.0,
-      "step": 92
-    },
-    {
-      "epoch": 0.014976448327227344,
-      "grad_norm": 8.019880624487996e-05,
-      "learning_rate": 0.0001999999999118385,
-      "loss": 46.0,
-      "step": 93
-    },
-    {
-      "epoch": 0.015137485406014735,
-      "grad_norm": 7.244279549922794e-05,
-      "learning_rate": 0.0001999999999097013,
-      "loss": 46.0,
-      "step": 94
-    },
-    {
-      "epoch": 0.015298522484802126,
-      "grad_norm": 6.675118493149057e-05,
-      "learning_rate": 0.00019999999990753854,
-      "loss": 46.0,
-      "step": 95
-    },
-    {
-      "epoch": 0.015459559563589517,
-      "grad_norm": 8.598087151767686e-05,
-      "learning_rate": 0.0001999999999053502,
-      "loss": 46.0,
-      "step": 96
-    },
-    {
-      "epoch": 0.015620596642376907,
-      "grad_norm": 4.9705315177561715e-05,
-      "learning_rate": 0.00019999999990313622,
-      "loss": 46.0,
-      "step": 97
-    },
-    {
-      "epoch": 0.015781633721164298,
-      "grad_norm": 5.5819342378526926e-05,
-      "learning_rate": 0.00019999999990089667,
-      "loss": 46.0,
-      "step": 98
-    },
-    {
-      "epoch": 0.015942670799951687,
-      "grad_norm": 3.597228351281956e-05,
-      "learning_rate": 0.00019999999989863155,
-      "loss": 46.0,
-      "step": 99
-    },
-    {
-      "epoch": 0.01610370787873908,
-      "grad_norm": 5.974910527584143e-05,
-      "learning_rate": 0.0001999999998963408,
-      "loss": 46.0,
-      "step": 100
-    },
-    {
-      "epoch": 0.01626474495752647,
-      "grad_norm": 5.3069412388140336e-05,
-      "learning_rate": 0.00019999999989402448,
-      "loss": 46.0,
-      "step": 101
-    },
-    {
-      "epoch": 0.01642578203631386,
-      "grad_norm": 5.418806176749058e-05,
-      "learning_rate": 0.00019999999989168252,
-      "loss": 46.0,
-      "step": 102
-    },
-    {
-      "epoch": 0.01658681911510125,
-      "grad_norm": 9.603131184121594e-05,
-      "learning_rate": 0.000199999999889315,
-      "loss": 46.0,
-      "step": 103
-    },
-    {
-      "epoch": 0.016747856193888643,
-      "grad_norm": 5.194047116674483e-05,
-      "learning_rate": 0.0001999999998869219,
-      "loss": 46.0,
-      "step": 104
-    },
-    {
-      "epoch": 0.016908893272676032,
-      "grad_norm": 9.761226829141378e-05,
-      "learning_rate": 0.00019999999988450315,
-      "loss": 46.0,
-      "step": 105
-    },
-    {
-      "epoch": 0.017069930351463425,
-      "grad_norm": 6.508623482659459e-05,
-      "learning_rate": 0.00019999999988205884,
-      "loss": 46.0,
-      "step": 106
-    },
-    {
-      "epoch": 0.017230967430250814,
-      "grad_norm": 5.57337116333656e-05,
-      "learning_rate": 0.00019999999987958895,
-      "loss": 46.0,
-      "step": 107
-    },
-    {
-      "epoch": 0.017392004509038207,
-      "grad_norm": 0.00010947446571663022,
-      "learning_rate": 0.00019999999987709344,
-      "loss": 46.0,
-      "step": 108
-    },
-    {
-      "epoch": 0.017553041587825596,
-      "grad_norm": 5.486145164468326e-05,
-      "learning_rate": 0.00019999999987457235,
-      "loss": 46.0,
-      "step": 109
-    },
-    {
-      "epoch": 0.017714078666612988,
-      "grad_norm": 0.00012351380428299308,
-      "learning_rate": 0.00019999999987202566,
-      "loss": 46.0,
-      "step": 110
-    },
-    {
-      "epoch": 0.017875115745400377,
-      "grad_norm": 0.00010108689457410946,
-      "learning_rate": 0.00019999999986945337,
-      "loss": 46.0,
-      "step": 111
-    },
-    {
-      "epoch": 0.01803615282418777,
-      "grad_norm": 8.059964602580294e-05,
-      "learning_rate": 0.00019999999986685548,
-      "loss": 46.0,
-      "step": 112
-    },
-    {
-      "epoch": 0.01819718990297516,
-      "grad_norm": 0.00013831921387463808,
-      "learning_rate": 0.00019999999986423204,
-      "loss": 46.0,
-      "step": 113
-    },
-    {
-      "epoch": 0.01835822698176255,
-      "grad_norm": 7.420378824463114e-05,
-      "learning_rate": 0.00019999999986158295,
-      "loss": 46.0,
-      "step": 114
-    },
-    {
-      "epoch": 0.01851926406054994,
-      "grad_norm": 4.6814933739369735e-05,
-      "learning_rate": 0.00019999999985890828,
-      "loss": 46.0,
-      "step": 115
-    },
-    {
-      "epoch": 0.018680301139337333,
-      "grad_norm": 9.741364192450419e-05,
-      "learning_rate": 0.00019999999985620802,
-      "loss": 46.0,
-      "step": 116
-    },
-    {
-      "epoch": 0.018841338218124722,
-      "grad_norm": 6.949760427232832e-05,
-      "learning_rate": 0.0001999999998534822,
-      "loss": 46.0,
-      "step": 117
-    },
-    {
-      "epoch": 0.019002375296912115,
-      "grad_norm": 6.653879972873256e-05,
-      "learning_rate": 0.00019999999985073074,
-      "loss": 46.0,
-      "step": 118
-    },
-    {
-      "epoch": 0.019163412375699504,
-      "grad_norm": 0.0001317582791671157,
-      "learning_rate": 0.00019999999984795367,
-      "loss": 46.0,
-      "step": 119
-    },
-    {
-      "epoch": 0.019324449454486897,
-      "grad_norm": 8.747701940592378e-05,
-      "learning_rate": 0.00019999999984515106,
-      "loss": 46.0,
-      "step": 120
-    },
-    {
-      "epoch": 0.019485486533274286,
-      "grad_norm": 0.00013079583004582673,
-      "learning_rate": 0.0001999999998423228,
-      "loss": 46.0,
-      "step": 121
-    },
-    {
-      "epoch": 0.019646523612061678,
-      "grad_norm": 8.933498611440882e-05,
-      "learning_rate": 0.000199999999839469,
-      "loss": 46.0,
-      "step": 122
-    },
-    {
-      "epoch": 0.019807560690849067,
-      "grad_norm": 8.507933671353385e-05,
-      "learning_rate": 0.00019999999983658957,
-      "loss": 46.0,
-      "step": 123
-    },
-    {
-      "epoch": 0.01996859776963646,
-      "grad_norm": 0.00011648085637716576,
-      "learning_rate": 0.00019999999983368456,
-      "loss": 46.0,
-      "step": 124
-    },
-    {
-      "epoch": 0.02012963484842385,
-      "grad_norm": 0.00012115373829146847,
-      "learning_rate": 0.00019999999983075394,
-      "loss": 46.0,
-      "step": 125
-    },
-    {
-      "epoch": 0.02029067192721124,
-      "grad_norm": 0.00017015583580359817,
-      "learning_rate": 0.00019999999982779772,
-      "loss": 46.0,
-      "step": 126
-    },
-    {
-      "epoch": 0.02045170900599863,
-      "grad_norm": 6.490958912763745e-05,
-      "learning_rate": 0.00019999999982481592,
-      "loss": 46.0,
-      "step": 127
-    },
-    {
-      "epoch": 0.020612746084786023,
-      "grad_norm": 0.00011056676157750189,
-      "learning_rate": 0.00019999999982180854,
-      "loss": 46.0,
-      "step": 128
-    },
-    {
-      "epoch": 0.020773783163573412,
-      "grad_norm": 8.274184801848605e-05,
-      "learning_rate": 0.00019999999981877556,
-      "loss": 46.0,
-      "step": 129
-    },
-    {
-      "epoch": 0.020934820242360805,
-      "grad_norm": 7.937616464914754e-05,
-      "learning_rate": 0.00019999999981571695,
-      "loss": 46.0,
-      "step": 130
-    },
-    {
-      "epoch": 0.021095857321148194,
-      "grad_norm": 0.0001607205340405926,
-      "learning_rate": 0.00019999999981263277,
-      "loss": 46.0,
-      "step": 131
-    },
-    {
-      "epoch": 0.021256894399935587,
-      "grad_norm": 7.551934686489403e-05,
-      "learning_rate": 0.000199999999809523,
-      "loss": 46.0,
-      "step": 132
-    },
-    {
-      "epoch": 0.021417931478722976,
-      "grad_norm": 0.00015242550580296665,
-      "learning_rate": 0.00019999999980638763,
-      "loss": 46.0,
-      "step": 133
-    },
-    {
-      "epoch": 0.021578968557510368,
-      "grad_norm": 6.205275713000447e-05,
-      "learning_rate": 0.00019999999980322667,
-      "loss": 46.0,
-      "step": 134
-    },
-    {
-      "epoch": 0.021740005636297757,
-      "grad_norm": 0.00013843170017935336,
-      "learning_rate": 0.0001999999998000401,
-      "loss": 46.0,
-      "step": 135
-    },
-    {
-      "epoch": 0.02190104271508515,
-      "grad_norm": 8.66490590851754e-05,
-      "learning_rate": 0.00019999999979682795,
-      "loss": 46.0,
-      "step": 136
-    },
-    {
-      "epoch": 0.02206207979387254,
-      "grad_norm": 0.00012498012802097946,
-      "learning_rate": 0.0001999999997935902,
-      "loss": 46.0,
-      "step": 137
-    },
-    {
-      "epoch": 0.02222311687265993,
-      "grad_norm": 0.00010799634037539363,
-      "learning_rate": 0.00019999999979032683,
-      "loss": 46.0,
-      "step": 138
-    },
-    {
-      "epoch": 0.02238415395144732,
-      "grad_norm": 0.00010536867921473458,
-      "learning_rate": 0.00019999999978703793,
-      "loss": 46.0,
-      "step": 139
-    },
-    {
-      "epoch": 0.022545191030234713,
-      "grad_norm": 0.00013301914441399276,
-      "learning_rate": 0.00019999999978372336,
-      "loss": 46.0,
-      "step": 140
-    },
-    {
-      "epoch": 0.022706228109022102,
-      "grad_norm": 0.00018815284420270473,
-      "learning_rate": 0.00019999999978038323,
-      "loss": 46.0,
-      "step": 141
-    },
-    {
-      "epoch": 0.022867265187809495,
-      "grad_norm": 0.00010110333823831752,
-      "learning_rate": 0.00019999999977701752,
-      "loss": 46.0,
-      "step": 142
-    },
-    {
-      "epoch": 0.023028302266596884,
-      "grad_norm": 0.00014375909813679755,
-      "learning_rate": 0.0001999999997736262,
-      "loss": 46.0,
-      "step": 143
-    },
-    {
-      "epoch": 0.023189339345384273,
-      "grad_norm": 9.653297456679866e-05,
-      "learning_rate": 0.0001999999997702093,
-      "loss": 46.0,
-      "step": 144
-    },
-    {
-      "epoch": 0.023350376424171666,
-      "grad_norm": 0.00013295281678438187,
-      "learning_rate": 0.00019999999976676676,
-      "loss": 46.0,
-      "step": 145
-    },
-    {
-      "epoch": 0.023511413502959055,
-      "grad_norm": 0.00010267561447108164,
-      "learning_rate": 0.00019999999976329867,
-      "loss": 46.0,
-      "step": 146
-    },
-    {
-      "epoch": 0.023672450581746447,
-      "grad_norm": 0.00014794350136071444,
-      "learning_rate": 0.00019999999975980494,
-      "loss": 46.0,
-      "step": 147
-    },
-    {
-      "epoch": 0.023833487660533836,
-      "grad_norm": 0.00012542445620056242,
-      "learning_rate": 0.00019999999975628567,
-      "loss": 46.0,
-      "step": 148
-    },
-    {
-      "epoch": 0.02399452473932123,
-      "grad_norm": 0.0002975241222884506,
-      "learning_rate": 0.00019999999975274077,
-      "loss": 46.0,
-      "step": 149
-    },
-    {
-      "epoch": 0.024155561818108618,
-      "grad_norm": 7.225852459669113e-05,
-      "learning_rate": 0.0001999999997491703,
-      "loss": 46.0,
-      "step": 150
-    },
-    {
-      "epoch": 0.02431659889689601,
-      "grad_norm": 0.00010266884055454284,
-      "learning_rate": 0.00019999999974557422,
-      "loss": 46.0,
-      "step": 151
-    },
-    {
-      "epoch": 0.0244776359756834,
-      "grad_norm": 0.00019384155166335404,
-      "learning_rate": 0.00019999999974195255,
-      "loss": 46.0,
-      "step": 152
-    },
-    {
-      "epoch": 0.024638673054470792,
-      "grad_norm": 9.250854782294482e-05,
-      "learning_rate": 0.00019999999973830526,
-      "loss": 46.0,
-      "step": 153
-    },
-    {
-      "epoch": 0.02479971013325818,
-      "grad_norm": 0.00014043178816791624,
-      "learning_rate": 0.00019999999973463242,
-      "loss": 46.0,
-      "step": 154
-    },
-    {
-      "epoch": 0.024960747212045574,
-      "grad_norm": 9.878585115075111e-05,
-      "learning_rate": 0.00019999999973093396,
-      "loss": 46.0,
-      "step": 155
-    },
-    {
-      "epoch": 0.025121784290832963,
-      "grad_norm": 0.00014434207696467638,
-      "learning_rate": 0.00019999999972720992,
-      "loss": 46.0,
-      "step": 156
-    },
-    {
-      "epoch": 0.025282821369620356,
-      "grad_norm": 0.00012252044689375907,
-      "learning_rate": 0.00019999999972346023,
-      "loss": 46.0,
-      "step": 157
-    },
-    {
-      "epoch": 0.025443858448407745,
-      "grad_norm": 0.00013778013817500323,
-      "learning_rate": 0.00019999999971968499,
-      "loss": 46.0,
-      "step": 158
-    },
-    {
-      "epoch": 0.025604895527195137,
-      "grad_norm": 0.00012422144936863333,
-      "learning_rate": 0.00019999999971588418,
-      "loss": 46.0,
-      "step": 159
-    },
-    {
-      "epoch": 0.025765932605982526,
-      "grad_norm": 0.00015130244719330221,
-      "learning_rate": 0.0001999999997120577,
-      "loss": 46.0,
-      "step": 160
-    },
-    {
-      "epoch": 0.02592696968476992,
-      "grad_norm": 0.0001149854579125531,
-      "learning_rate": 0.0001999999997082057,
-      "loss": 46.0,
-      "step": 161
-    },
-    {
-      "epoch": 0.026088006763557308,
-      "grad_norm": 0.0001454713783459738,
-      "learning_rate": 0.00019999999970432808,
-      "loss": 46.0,
-      "step": 162
-    },
-    {
-      "epoch": 0.0262490438423447,
-      "grad_norm": 0.00012921511370223016,
-      "learning_rate": 0.00019999999970042487,
-      "loss": 46.0,
-      "step": 163
-    },
-    {
-      "epoch": 0.02641008092113209,
-      "grad_norm": 0.00023755831352900714,
-      "learning_rate": 0.00019999999969649604,
-      "loss": 46.0,
-      "step": 164
-    },
-    {
-      "epoch": 0.026571117999919482,
-      "grad_norm": 0.00018338421068619937,
-      "learning_rate": 0.00019999999969254163,
-      "loss": 46.0,
-      "step": 165
-    },
-    {
-      "epoch": 0.02673215507870687,
-      "grad_norm": 0.000321844156133011,
-      "learning_rate": 0.00019999999968856163,
-      "loss": 46.0,
-      "step": 166
-    },
-    {
-      "epoch": 0.026893192157494264,
-      "grad_norm": 0.0002377462078584358,
-      "learning_rate": 0.00019999999968455605,
-      "loss": 46.0,
-      "step": 167
-    },
-    {
-      "epoch": 0.027054229236281653,
-      "grad_norm": 0.0003046017372980714,
-      "learning_rate": 0.00019999999968052485,
-      "loss": 46.0,
-      "step": 168
-    },
-    {
-      "epoch": 0.027215266315069046,
-      "grad_norm": 0.00015924294712021947,
-      "learning_rate": 0.00019999999967646806,
-      "loss": 46.0,
-      "step": 169
-    },
-    {
-      "epoch": 0.027376303393856435,
-      "grad_norm": 0.00023570118355564773,
-      "learning_rate": 0.00019999999967238566,
-      "loss": 46.0,
-      "step": 170
-    },
-    {
-      "epoch": 0.027537340472643827,
-      "grad_norm": 0.00018532936519477516,
-      "learning_rate": 0.0001999999996682777,
-      "loss": 46.0,
-      "step": 171
-    },
-    {
-      "epoch": 0.027698377551431216,
-      "grad_norm": 0.00019276805687695742,
-      "learning_rate": 0.00019999999966414413,
-      "loss": 46.0,
-      "step": 172
-    },
-    {
-      "epoch": 0.02785941463021861,
-      "grad_norm": 0.0001177889498649165,
-      "learning_rate": 0.00019999999965998497,
-      "loss": 46.0,
-      "step": 173
-    },
-    {
-      "epoch": 0.028020451709005998,
-      "grad_norm": 0.0001398147433064878,
-      "learning_rate": 0.00019999999965580022,
-      "loss": 46.0,
-      "step": 174
-    },
-    {
-      "epoch": 0.02818148878779339,
-      "grad_norm": 0.00016932954895310104,
-      "learning_rate": 0.00019999999965158985,
-      "loss": 46.0,
-      "step": 175
-    },
-    {
-      "epoch": 0.02834252586658078,
-      "grad_norm": 0.000123539415653795,
-      "learning_rate": 0.00019999999964735393,
-      "loss": 46.0,
-      "step": 176
-    },
-    {
-      "epoch": 0.028503562945368172,
-      "grad_norm": 0.0002068693720502779,
-      "learning_rate": 0.00019999999964309236,
-      "loss": 46.0,
-      "step": 177
-    },
-    {
-      "epoch": 0.02866460002415556,
-      "grad_norm": 0.00022907050151843578,
-      "learning_rate": 0.0001999999996388052,
-      "loss": 46.0,
-      "step": 178
-    },
-    {
-      "epoch": 0.028825637102942954,
-      "grad_norm": 0.00013423155178315938,
-      "learning_rate": 0.00019999999963449247,
-      "loss": 46.0,
-      "step": 179
-    },
-    {
-      "epoch": 0.028986674181730343,
-      "grad_norm": 0.00021165514772292227,
-      "learning_rate": 0.00019999999963015417,
-      "loss": 46.0,
-      "step": 180
-    },
-    {
-      "epoch": 0.029147711260517736,
-      "grad_norm": 0.0002548027550801635,
-      "learning_rate": 0.00019999999962579023,
-      "loss": 46.0,
-      "step": 181
-    },
-    {
-      "epoch": 0.029308748339305125,
-      "grad_norm": 0.00017856726481113583,
-      "learning_rate": 0.0001999999996214007,
-      "loss": 46.0,
-      "step": 182
-    },
-    {
-      "epoch": 0.029469785418092517,
-      "grad_norm": 0.0002464374992996454,
-      "learning_rate": 0.0001999999996169856,
-      "loss": 46.0,
-      "step": 183
-    },
-    {
-      "epoch": 0.029630822496879906,
-      "grad_norm": 0.0002536777756176889,
-      "learning_rate": 0.00019999999961254489,
-      "loss": 46.0,
-      "step": 184
-    },
-    {
-      "epoch": 0.0297918595756673,
-      "grad_norm": 0.00021144087077118456,
-      "learning_rate": 0.00019999999960807857,
-      "loss": 46.0,
-      "step": 185
-    },
-    {
-      "epoch": 0.029952896654454688,
-      "grad_norm": 0.00031986809335649014,
-      "learning_rate": 0.00019999999960358667,
-      "loss": 46.0,
-      "step": 186
-    },
-    {
-      "epoch": 0.03011393373324208,
-      "grad_norm": 0.0001996238570427522,
-      "learning_rate": 0.00019999999959906917,
-      "loss": 46.0,
-      "step": 187
-    },
-    {
-      "epoch": 0.03027497081202947,
-      "grad_norm": 0.0002590011281426996,
-      "learning_rate": 0.0001999999995945261,
-      "loss": 46.0,
-      "step": 188
-    },
-    {
-      "epoch": 0.030436007890816862,
-      "grad_norm": 0.00011825704132206738,
-      "learning_rate": 0.0001999999995899574,
-      "loss": 46.0,
-      "step": 189
-    },
-    {
-      "epoch": 0.03059704496960425,
-      "grad_norm": 0.00025153084425255656,
-      "learning_rate": 0.00019999999958536312,
-      "loss": 46.0,
-      "step": 190
-    },
-    {
-      "epoch": 0.03075808204839164,
-      "grad_norm": 0.0004842008638661355,
-      "learning_rate": 0.00019999999958074326,
-      "loss": 46.0,
-      "step": 191
-    },
-    {
-      "epoch": 0.030919119127179033,
-      "grad_norm": 0.000218238215893507,
-      "learning_rate": 0.00019999999957609778,
-      "loss": 46.0,
-      "step": 192
-    },
-    {
-      "epoch": 0.031080156205966422,
-      "grad_norm": 0.00028900333563797176,
-      "learning_rate": 0.0001999999995714267,
-      "loss": 46.0,
-      "step": 193
-    },
-    {
-      "epoch": 0.031241193284753815,
-      "grad_norm": 0.00026324810460209846,
-      "learning_rate": 0.00019999999956673005,
-      "loss": 46.0,
-      "step": 194
-    },
-    {
-      "epoch": 0.03140223036354121,
-      "grad_norm": 0.0004234739753883332,
-      "learning_rate": 0.0001999999995620078,
-      "loss": 46.0,
-      "step": 195
-    },
-    {
-      "epoch": 0.031563267442328596,
-      "grad_norm": 0.0003671174345072359,
-      "learning_rate": 0.00019999999955725995,
-      "loss": 46.0,
-      "step": 196
-    },
-    {
-      "epoch": 0.031724304521115985,
-      "grad_norm": 0.0003977952292189002,
-      "learning_rate": 0.0001999999995524865,
-      "loss": 46.0,
-      "step": 197
-    },
-    {
-      "epoch": 0.031885341599903375,
-      "grad_norm": 0.0003290062304586172,
-      "learning_rate": 0.00019999999954768748,
-      "loss": 46.0,
-      "step": 198
-    },
-    {
-      "epoch": 0.03204637867869077,
-      "grad_norm": 0.00023715550196357071,
-      "learning_rate": 0.00019999999954286283,
-      "loss": 46.0,
-      "step": 199
-    },
-    {
-      "epoch": 0.03220741575747816,
-      "grad_norm": 0.0002557702246122062,
-      "learning_rate": 0.00019999999953801263,
-      "loss": 46.0,
-      "step": 200
-    },
-    {
-      "epoch": 0.03236845283626555,
-      "grad_norm": 0.00030496579711325467,
-      "learning_rate": 0.0001999999995331368,
-      "loss": 46.0,
-      "step": 201
-    },
-    {
-      "epoch": 0.03252948991505294,
-      "grad_norm": 0.00019574211910367012,
-      "learning_rate": 0.00019999999952823537,
-      "loss": 46.0,
-      "step": 202
-    },
-    {
-      "epoch": 0.032690526993840334,
-      "grad_norm": 0.000399620970711112,
-      "learning_rate": 0.00019999999952330838,
-      "loss": 46.0,
-      "step": 203
-    },
-    {
-      "epoch": 0.03285156407262772,
-      "grad_norm": 0.00023076729848980904,
-      "learning_rate": 0.00019999999951835577,
-      "loss": 46.0,
-      "step": 204
-    },
-    {
-      "epoch": 0.03301260115141511,
-      "grad_norm": 0.0003438382700551301,
-      "learning_rate": 0.00019999999951337758,
-      "loss": 46.0,
-      "step": 205
-    },
-    {
-      "epoch": 0.0331736382302025,
-      "grad_norm": 0.0002556856779847294,
-      "learning_rate": 0.00019999999950837377,
-      "loss": 46.0,
-      "step": 206
-    },
-    {
-      "epoch": 0.0333346753089899,
-      "grad_norm": 0.00024224602384492755,
-      "learning_rate": 0.00019999999950334437,
-      "loss": 46.0,
-      "step": 207
-    },
-    {
-      "epoch": 0.033495712387777286,
-      "grad_norm": 0.0002187650097766891,
-      "learning_rate": 0.00019999999949828938,
-      "loss": 46.0,
-      "step": 208
-    },
-    {
-      "epoch": 0.033656749466564675,
-      "grad_norm": 0.00031273020431399345,
-      "learning_rate": 0.0001999999994932088,
-      "loss": 46.0,
-      "step": 209
-    },
-    {
-      "epoch": 0.033817786545352065,
-      "grad_norm": 0.00029702080064453185,
-      "learning_rate": 0.00019999999948810263,
-      "loss": 46.0,
-      "step": 210
-    },
-    {
-      "epoch": 0.03397882362413946,
-      "grad_norm": 0.00018055284454021603,
-      "learning_rate": 0.00019999999948297085,
-      "loss": 46.0,
-      "step": 211
-    },
-    {
-      "epoch": 0.03413986070292685,
-      "grad_norm": 0.00024684282834641635,
-      "learning_rate": 0.0001999999994778135,
-      "loss": 46.0,
-      "step": 212
-    },
-    {
-      "epoch": 0.03430089778171424,
-      "grad_norm": 0.0003250887675676495,
-      "learning_rate": 0.00019999999947263052,
-      "loss": 46.0,
-      "step": 213
-    },
-    {
-      "epoch": 0.03446193486050163,
-      "grad_norm": 0.00041803737985901535,
-      "learning_rate": 0.00019999999946742199,
-      "loss": 46.0,
-      "step": 214
-    },
-    {
-      "epoch": 0.034622971939289024,
-      "grad_norm": 0.0002714908914640546,
-      "learning_rate": 0.0001999999994621878,
-      "loss": 46.0,
-      "step": 215
-    },
-    {
-      "epoch": 0.03478400901807641,
-      "grad_norm": 0.00026197044644504786,
-      "learning_rate": 0.00019999999945692807,
-      "loss": 46.0,
-      "step": 216
-    },
-    {
-      "epoch": 0.0349450460968638,
-      "grad_norm": 0.0002161135635105893,
-      "learning_rate": 0.00019999999945164275,
-      "loss": 46.0,
-      "step": 217
-    },
-    {
-      "epoch": 0.03510608317565119,
-      "grad_norm": 0.0003438770945649594,
-      "learning_rate": 0.0001999999994463318,
-      "loss": 46.0,
-      "step": 218
-    },
-    {
-      "epoch": 0.03526712025443859,
-      "grad_norm": 0.00020931630569975823,
-      "learning_rate": 0.0001999999994409953,
-      "loss": 46.0,
-      "step": 219
-    },
-    {
-      "epoch": 0.035428157333225976,
-      "grad_norm": 0.00029885736876167357,
-      "learning_rate": 0.00019999999943563315,
-      "loss": 46.0,
-      "step": 220
-    },
-    {
-      "epoch": 0.035589194412013365,
-      "grad_norm": 0.00047396900481544435,
-      "learning_rate": 0.00019999999943024543,
-      "loss": 46.0,
-      "step": 221
-    },
-    {
-      "epoch": 0.035750231490800755,
-      "grad_norm": 0.00031950193806551397,
-      "learning_rate": 0.0001999999994248321,
-      "loss": 46.0,
-      "step": 222
-    },
-    {
-      "epoch": 0.03591126856958815,
-      "grad_norm": 0.00022150597942527384,
-      "learning_rate": 0.0001999999994193932,
-      "loss": 46.0,
-      "step": 223
-    },
-    {
-      "epoch": 0.03607230564837554,
-      "grad_norm": 0.0003227836568839848,
-      "learning_rate": 0.0001999999994139287,
-      "loss": 46.0,
-      "step": 224
-    },
-    {
-      "epoch": 0.03623334272716293,
-      "grad_norm": 0.0002998760901391506,
-      "learning_rate": 0.0001999999994084386,
-      "loss": 46.0,
-      "step": 225
-    },
-    {
-      "epoch": 0.03639437980595032,
-      "grad_norm": 0.0004136904899496585,
-      "learning_rate": 0.0001999999994029229,
-      "loss": 46.0,
-      "step": 226
-    },
-    {
-      "epoch": 0.036555416884737714,
-      "grad_norm": 7.639649993507192e-05,
-      "learning_rate": 0.0001999999993973816,
-      "loss": 46.0,
-      "step": 227
-    },
-    {
-      "epoch": 0.0367164539635251,
-      "grad_norm": 0.00030812519253231585,
-      "learning_rate": 0.00019999999939181474,
-      "loss": 46.0,
-      "step": 228
-    },
-    {
-      "epoch": 0.03687749104231249,
-      "grad_norm": 0.0002964360173791647,
-      "learning_rate": 0.00019999999938622224,
-      "loss": 46.0,
-      "step": 229
-    },
-    {
-      "epoch": 0.03703852812109988,
-      "grad_norm": 0.0004796150024048984,
-      "learning_rate": 0.0001999999993806042,
-      "loss": 46.0,
-      "step": 230
-    },
-    {
-      "epoch": 0.03719956519988728,
-      "grad_norm": 0.0005432307370938361,
-      "learning_rate": 0.00019999999937496052,
-      "loss": 46.0,
-      "step": 231
-    },
-    {
-      "epoch": 0.037360602278674666,
-      "grad_norm": 0.00046832027146592736,
-      "learning_rate": 0.00019999999936929123,
-      "loss": 46.0,
-      "step": 232
-    },
-    {
-      "epoch": 0.037521639357462055,
-      "grad_norm": 0.0006342283450067043,
-      "learning_rate": 0.0001999999993635964,
-      "loss": 46.0,
-      "step": 233
-    },
-    {
-      "epoch": 0.037682676436249445,
-      "grad_norm": 0.0002783329109661281,
-      "learning_rate": 0.00019999999935787593,
-      "loss": 46.0,
-      "step": 234
-    },
-    {
-      "epoch": 0.03784371351503684,
-      "grad_norm": 0.00032400648342445493,
-      "learning_rate": 0.0001999999993521299,
-      "loss": 46.0,
-      "step": 235
-    },
-    {
-      "epoch": 0.03800475059382423,
-      "grad_norm": 0.0004526137199718505,
-      "learning_rate": 0.00019999999934635824,
-      "loss": 46.0,
-      "step": 236
-    },
-    {
-      "epoch": 0.03816578767261162,
-      "grad_norm": 0.0005123643204569817,
-      "learning_rate": 0.000199999999340561,
-      "loss": 46.0,
-      "step": 237
-    },
-    {
-      "epoch": 0.03832682475139901,
-      "grad_norm": 0.00032366730738431215,
-      "learning_rate": 0.00019999999933473816,
-      "loss": 46.0,
-      "step": 238
-    },
-    {
-      "epoch": 0.0384878618301864,
-      "grad_norm": 0.0005605536862276495,
-      "learning_rate": 0.00019999999932888973,
-      "loss": 46.0,
-      "step": 239
-    },
-    {
-      "epoch": 0.03864889890897379,
-      "grad_norm": 0.00045590754598379135,
-      "learning_rate": 0.00019999999932301572,
-      "loss": 46.0,
-      "step": 240
-    },
-    {
-      "epoch": 0.03880993598776118,
-      "grad_norm": 0.0004321521264500916,
-      "learning_rate": 0.0001999999993171161,
-      "loss": 46.0,
-      "step": 241
-    },
-    {
-      "epoch": 0.03897097306654857,
-      "grad_norm": 0.0005606827326118946,
-      "learning_rate": 0.00019999999931119092,
-      "loss": 46.0,
-      "step": 242
-    },
-    {
-      "epoch": 0.03913201014533596,
-      "grad_norm": 0.0004983293474651873,
-      "learning_rate": 0.00019999999930524006,
-      "loss": 46.0,
-      "step": 243
-    },
-    {
-      "epoch": 0.039293047224123356,
-      "grad_norm": 0.00033221906051039696,
-      "learning_rate": 0.0001999999992992637,
-      "loss": 46.0,
-      "step": 244
-    },
-    {
-      "epoch": 0.039454084302910745,
-      "grad_norm": 0.00033477682154625654,
-      "learning_rate": 0.00019999999929326169,
-      "loss": 46.0,
-      "step": 245
-    },
-    {
-      "epoch": 0.039615121381698135,
-      "grad_norm": 0.000464974669739604,
-      "learning_rate": 0.0001999999992872341,
-      "loss": 46.0,
-      "step": 246
-    },
-    {
-      "epoch": 0.039776158460485524,
-      "grad_norm": 0.00038884178502485156,
-      "learning_rate": 0.0001999999992811809,
-      "loss": 46.0,
-      "step": 247
-    },
-    {
-      "epoch": 0.03993719553927292,
-      "grad_norm": 0.0006443243473768234,
-      "learning_rate": 0.00019999999927510214,
-      "loss": 46.0,
-      "step": 248
-    },
-    {
-      "epoch": 0.04009823261806031,
-      "grad_norm": 0.0007490874268114567,
-      "learning_rate": 0.00019999999926899776,
-      "loss": 46.0,
-      "step": 249
-    },
-    {
-      "epoch": 0.0402592696968477,
-      "grad_norm": 0.0005477413069456816,
-      "learning_rate": 0.00019999999926286778,
-      "loss": 46.0,
-      "step": 250
-    },
-    {
-      "epoch": 0.04042030677563509,
-      "grad_norm": 0.0007731926161795855,
-      "learning_rate": 0.00019999999925671223,
-      "loss": 46.0,
-      "step": 251
-    },
-    {
-      "epoch": 0.04058134385442248,
-      "grad_norm": 0.0004357139696367085,
-      "learning_rate": 0.00019999999925053105,
-      "loss": 46.0,
-      "step": 252
-    },
-    {
-      "epoch": 0.04074238093320987,
-      "grad_norm": 0.0008482228149659932,
-      "learning_rate": 0.00019999999924432432,
-      "loss": 46.0,
-      "step": 253
-    },
-    {
-      "epoch": 0.04090341801199726,
-      "grad_norm": 0.000651440757792443,
-      "learning_rate": 0.00019999999923809195,
-      "loss": 46.0,
-      "step": 254
-    },
-    {
-      "epoch": 0.04106445509078465,
-      "grad_norm": 0.0006377010722644627,
-      "learning_rate": 0.000199999999231834,
-      "loss": 46.0,
-      "step": 255
-    },
-    {
-      "epoch": 0.041225492169572046,
-      "grad_norm": 0.0008143018931150436,
-      "learning_rate": 0.00019999999922555047,
-      "loss": 46.0,
-      "step": 256
-    },
-    {
-      "epoch": 0.041386529248359435,
-      "grad_norm": 0.0005103488801978528,
-      "learning_rate": 0.00019999999921924133,
-      "loss": 46.0,
-      "step": 257
-    },
-    {
-      "epoch": 0.041547566327146825,
-      "grad_norm": 0.000732742715626955,
-      "learning_rate": 0.0001999999992129066,
-      "loss": 46.0,
-      "step": 258
-    },
-    {
-      "epoch": 0.041708603405934214,
-      "grad_norm": 0.0006536121945828199,
-      "learning_rate": 0.00019999999920654627,
-      "loss": 46.0,
-      "step": 259
-    },
-    {
-      "epoch": 0.04186964048472161,
-      "grad_norm": 0.00036585298948921263,
-      "learning_rate": 0.00019999999920016035,
-      "loss": 46.0,
-      "step": 260
-    },
-    {
-      "epoch": 0.042030677563509,
-      "grad_norm": 0.0008182922028936446,
-      "learning_rate": 0.00019999999919374887,
-      "loss": 46.0,
-      "step": 261
-    },
-    {
-      "epoch": 0.04219171464229639,
-      "grad_norm": 0.0006231138831935823,
-      "learning_rate": 0.00019999999918731174,
-      "loss": 46.0,
-      "step": 262
-    },
-    {
-      "epoch": 0.04235275172108378,
-      "grad_norm": 0.0006047601345926523,
-      "learning_rate": 0.00019999999918084903,
-      "loss": 46.0,
-      "step": 263
-    },
-    {
-      "epoch": 0.04251378879987117,
-      "grad_norm": 0.0005017635994590819,
-      "learning_rate": 0.00019999999917436073,
-      "loss": 46.0,
-      "step": 264
-    },
-    {
-      "epoch": 0.04267482587865856,
-      "grad_norm": 0.0005903529818169773,
-      "learning_rate": 0.00019999999916784682,
-      "loss": 46.0,
-      "step": 265
-    },
-    {
-      "epoch": 0.04283586295744595,
-      "grad_norm": 0.0010839580791071057,
-      "learning_rate": 0.00019999999916130734,
-      "loss": 46.0,
-      "step": 266
-    },
-    {
-      "epoch": 0.04299690003623334,
-      "grad_norm": 0.0006588880205526948,
-      "learning_rate": 0.00019999999915474225,
-      "loss": 46.0,
-      "step": 267
-    },
-    {
-      "epoch": 0.043157937115020736,
-      "grad_norm": 0.0007730197394266725,
-      "learning_rate": 0.0001999999991481516,
-      "loss": 46.0,
-      "step": 268
-    },
-    {
-      "epoch": 0.043318974193808125,
-      "grad_norm": 0.0005474932258948684,
-      "learning_rate": 0.00019999999914153532,
-      "loss": 46.0,
-      "step": 269
-    },
-    {
-      "epoch": 0.043480011272595515,
-      "grad_norm": 0.0003796566743403673,
-      "learning_rate": 0.00019999999913489344,
-      "loss": 46.0,
-      "step": 270
-    },
-    {
-      "epoch": 0.043641048351382904,
-      "grad_norm": 0.000516220519784838,
-      "learning_rate": 0.00019999999912822598,
-      "loss": 46.0,
-      "step": 271
-    },
-    {
-      "epoch": 0.0438020854301703,
-      "grad_norm": 0.0006308844895102084,
-      "learning_rate": 0.0001999999991215329,
-      "loss": 46.0,
-      "step": 272
-    },
-    {
-      "epoch": 0.04396312250895769,
-      "grad_norm": 0.0006311295437626541,
-      "learning_rate": 0.00019999999911481426,
-      "loss": 46.0,
-      "step": 273
-    },
-    {
-      "epoch": 0.04412415958774508,
-      "grad_norm": 0.0005519656115211546,
-      "learning_rate": 0.00019999999910807,
-      "loss": 46.0,
-      "step": 274
-    },
-    {
-      "epoch": 0.04428519666653247,
-      "grad_norm": 0.0007428042590618134,
-      "learning_rate": 0.0001999999991013002,
-      "loss": 46.0,
-      "step": 275
-    },
-    {
-      "epoch": 0.04444623374531986,
-      "grad_norm": 0.0009881554869934916,
-      "learning_rate": 0.00019999999909450474,
-      "loss": 46.0,
-      "step": 276
-    },
-    {
-      "epoch": 0.04460727082410725,
-      "grad_norm": 0.00044621757115237415,
-      "learning_rate": 0.0001999999990876837,
-      "loss": 46.0,
-      "step": 277
-    },
-    {
-      "epoch": 0.04476830790289464,
-      "grad_norm": 0.0005154189420863986,
-      "learning_rate": 0.0001999999990808371,
-      "loss": 46.0,
-      "step": 278
-    },
-    {
-      "epoch": 0.04492934498168203,
-      "grad_norm": 0.0008768675616011024,
-      "learning_rate": 0.00019999999907396485,
-      "loss": 46.0,
-      "step": 279
-    },
-    {
-      "epoch": 0.045090382060469426,
-      "grad_norm": 0.0010444631334394217,
-      "learning_rate": 0.00019999999906706703,
-      "loss": 46.0,
-      "step": 280
-    },
-    {
-      "epoch": 0.045251419139256815,
-      "grad_norm": 0.000744470045901835,
-      "learning_rate": 0.0001999999990601436,
-      "loss": 46.0,
-      "step": 281
-    },
-    {
-      "epoch": 0.045412456218044205,
-      "grad_norm": 0.001204045256599784,
-      "learning_rate": 0.00019999999905319463,
-      "loss": 46.0,
-      "step": 282
-    },
-    {
-      "epoch": 0.045573493296831594,
-      "grad_norm": 0.0008198407595045865,
-      "learning_rate": 0.00019999999904622,
-      "loss": 46.0,
-      "step": 283
-    },
-    {
-      "epoch": 0.04573453037561899,
-      "grad_norm": 0.0010659419931471348,
-      "learning_rate": 0.0001999999990392198,
-      "loss": 46.0,
-      "step": 284
-    },
-    {
-      "epoch": 0.04589556745440638,
-      "grad_norm": 0.0010541955707594752,
-      "learning_rate": 0.00019999999903219404,
-      "loss": 46.0,
-      "step": 285
-    },
-    {
-      "epoch": 0.04605660453319377,
-      "grad_norm": 0.0006653243908658624,
-      "learning_rate": 0.00019999999902514263,
-      "loss": 46.0,
-      "step": 286
-    },
-    {
-      "epoch": 0.04621764161198116,
-      "grad_norm": 0.0009864730527624488,
-      "learning_rate": 0.00019999999901806564,
-      "loss": 46.0,
-      "step": 287
-    },
-    {
-      "epoch": 0.046378678690768546,
-      "grad_norm": 0.0012773185735568404,
-      "learning_rate": 0.00019999999901096306,
-      "loss": 46.0,
-      "step": 288
-    },
-    {
-      "epoch": 0.04653971576955594,
-      "grad_norm": 0.0012416749959811568,
-      "learning_rate": 0.0001999999990038349,
-      "loss": 46.0,
-      "step": 289
-    },
-    {
-      "epoch": 0.04670075284834333,
-      "grad_norm": 0.001295020105317235,
-      "learning_rate": 0.00019999999899668116,
-      "loss": 46.0,
-      "step": 290
-    },
-    {
-      "epoch": 0.04686178992713072,
-      "grad_norm": 0.0009199907653965056,
-      "learning_rate": 0.0001999999989895018,
-      "loss": 46.0,
-      "step": 291
-    },
-    {
-      "epoch": 0.04702282700591811,
-      "grad_norm": 0.0011596897384151816,
-      "learning_rate": 0.00019999999898229684,
-      "loss": 46.0,
-      "step": 292
-    },
-    {
-      "epoch": 0.047183864084705505,
-      "grad_norm": 0.000495227228384465,
-      "learning_rate": 0.00019999999897506626,
-      "loss": 46.0,
-      "step": 293
-    },
-    {
-      "epoch": 0.047344901163492895,
-      "grad_norm": 0.0010823584161698818,
-      "learning_rate": 0.00019999999896781013,
-      "loss": 46.0,
-      "step": 294
-    },
-    {
-      "epoch": 0.047505938242280284,
-      "grad_norm": 0.0009814115474000573,
-      "learning_rate": 0.0001999999989605284,
-      "loss": 46.0,
-      "step": 295
-    },
-    {
-      "epoch": 0.04766697532106767,
-      "grad_norm": 0.001199794583953917,
-      "learning_rate": 0.00019999999895322105,
-      "loss": 46.0,
-      "step": 296
-    },
-    {
-      "epoch": 0.04782801239985507,
-      "grad_norm": 0.0008057034574449062,
-      "learning_rate": 0.00019999999894588813,
-      "loss": 46.0,
-      "step": 297
-    },
-    {
-      "epoch": 0.04798904947864246,
-      "grad_norm": 0.0009269247530028224,
-      "learning_rate": 0.0001999999989385296,
-      "loss": 46.0,
-      "step": 298
-    },
-    {
-      "epoch": 0.04815008655742985,
-      "grad_norm": 0.0007923051598481834,
-      "learning_rate": 0.0001999999989311455,
-      "loss": 46.0,
-      "step": 299
-    },
-    {
-      "epoch": 0.048311123636217236,
-      "grad_norm": 0.0011324144434183836,
-      "learning_rate": 0.00019999999892373577,
-      "loss": 46.0,
-      "step": 300
-    },
-    {
-      "epoch": 0.04847216071500463,
-      "grad_norm": 0.0009203523513861,
-      "learning_rate": 0.00019999999891630045,
-      "loss": 46.0,
-      "step": 301
-    },
-    {
-      "epoch": 0.04863319779379202,
-      "grad_norm": 0.0010944072855636477,
-      "learning_rate": 0.00019999999890883956,
-      "loss": 46.0,
-      "step": 302
-    },
-    {
-      "epoch": 0.04879423487257941,
-      "grad_norm": 0.0010690286289900541,
-      "learning_rate": 0.00019999999890135307,
-      "loss": 46.0,
-      "step": 303
-    },
-    {
-      "epoch": 0.0489552719513668,
-      "grad_norm": 0.0013335959520190954,
-      "learning_rate": 0.00019999999889384095,
-      "loss": 46.0,
-      "step": 304
-    },
-    {
-      "epoch": 0.049116309030154195,
-      "grad_norm": 0.0011828442802652717,
-      "learning_rate": 0.00019999999888630328,
-      "loss": 46.0,
-      "step": 305
-    },
-    {
-      "epoch": 0.049277346108941585,
-      "grad_norm": 0.0008028900483623147,
-      "learning_rate": 0.00019999999887874,
-      "loss": 46.0,
-      "step": 306
-    },
-    {
-      "epoch": 0.049438383187728974,
-      "grad_norm": 0.0007478521438315511,
-      "learning_rate": 0.0001999999988711511,
-      "loss": 46.0,
-      "step": 307
-    },
-    {
-      "epoch": 0.04959942026651636,
-      "grad_norm": 0.0012638323241844773,
-      "learning_rate": 0.00019999999886353664,
-      "loss": 46.0,
-      "step": 308
-    },
-    {
-      "epoch": 0.04976045734530376,
-      "grad_norm": 0.0005974912201054394,
-      "learning_rate": 0.0001999999988558966,
-      "loss": 46.0,
-      "step": 309
-    },
-    {
-      "epoch": 0.04992149442409115,
-      "grad_norm": 0.000533314305357635,
-      "learning_rate": 0.0001999999988482309,
-      "loss": 46.0,
-      "step": 310
-    },
-    {
-      "epoch": 0.05008253150287854,
-      "grad_norm": 0.0014910665340721607,
-      "learning_rate": 0.00019999999884053965,
-      "loss": 46.0,
-      "step": 311
-    },
-    {
-      "epoch": 0.050243568581665926,
-      "grad_norm": 0.0012139989994466305,
-      "learning_rate": 0.00019999999883282281,
-      "loss": 46.0,
-      "step": 312
-    },
-    {
-      "epoch": 0.05040460566045332,
-      "grad_norm": 0.0012169424444437027,
-      "learning_rate": 0.00019999999882508034,
-      "loss": 46.0,
-      "step": 313
-    },
-    {
-      "epoch": 0.05056564273924071,
-      "grad_norm": 0.0011955752270296216,
-      "learning_rate": 0.0001999999988173123,
-      "loss": 46.0,
-      "step": 314
-    },
-    {
-      "epoch": 0.0507266798180281,
-      "grad_norm": 0.0008034955244511366,
-      "learning_rate": 0.00019999999880951865,
-      "loss": 46.0,
-      "step": 315
-    },
-    {
-      "epoch": 0.05088771689681549,
-      "grad_norm": 0.001763079664669931,
-      "learning_rate": 0.00019999999880169944,
-      "loss": 46.0,
-      "step": 316
-    },
-    {
-      "epoch": 0.051048753975602885,
-      "grad_norm": 0.0016992834862321615,
-      "learning_rate": 0.0001999999987938546,
-      "loss": 46.0,
-      "step": 317
-    },
-    {
-      "epoch": 0.051209791054390275,
-      "grad_norm": 0.0011860596714541316,
-      "learning_rate": 0.0001999999987859842,
-      "loss": 46.0,
-      "step": 318
-    },
-    {
-      "epoch": 0.051370828133177664,
-      "grad_norm": 0.0010058664483949542,
-      "learning_rate": 0.00019999999877808817,
-      "loss": 46.0,
-      "step": 319
-    },
-    {
-      "epoch": 0.05153186521196505,
-      "grad_norm": 0.0011413118336349726,
-      "learning_rate": 0.00019999999877016655,
-      "loss": 46.0,
-      "step": 320
-    },
-    {
-      "epoch": 0.05169290229075245,
-      "grad_norm": 0.0014378430787473917,
-      "learning_rate": 0.00019999999876221933,
-      "loss": 46.0,
-      "step": 321
-    },
-    {
-      "epoch": 0.05185393936953984,
-      "grad_norm": 0.0013385158963501453,
-      "learning_rate": 0.00019999999875424656,
-      "loss": 46.0,
-      "step": 322
-    },
-    {
-      "epoch": 0.05201497644832723,
-      "grad_norm": 0.0008365577086806297,
-      "learning_rate": 0.00019999999874624816,
-      "loss": 46.0,
-      "step": 323
-    },
-    {
-      "epoch": 0.052176013527114616,
-      "grad_norm": 0.0011793512385338545,
-      "learning_rate": 0.00019999999873822417,
-      "loss": 46.0,
-      "step": 324
-    },
-    {
-      "epoch": 0.05233705060590201,
-      "grad_norm": 0.001501126098446548,
-      "learning_rate": 0.00019999999873017456,
-      "loss": 46.0,
-      "step": 325
-    },
-    {
-      "epoch": 0.0524980876846894,
-      "grad_norm": 0.0008644497720524669,
-      "learning_rate": 0.00019999999872209937,
-      "loss": 46.0,
-      "step": 326
-    },
-    {
-      "epoch": 0.05265912476347679,
-      "grad_norm": 0.0007261348073370755,
-      "learning_rate": 0.00019999999871399862,
-      "loss": 46.0,
-      "step": 327
-    },
-    {
-      "epoch": 0.05282016184226418,
-      "grad_norm": 0.0009311382309533656,
-      "learning_rate": 0.00019999999870587225,
-      "loss": 46.0,
-      "step": 328
-    },
-    {
-      "epoch": 0.052981198921051575,
-      "grad_norm": 0.0009469041833654046,
-      "learning_rate": 0.00019999999869772027,
-      "loss": 46.0,
-      "step": 329
-    },
-    {
-      "epoch": 0.053142235999838965,
-      "grad_norm": 0.0008246630313806236,
-      "learning_rate": 0.0001999999986895427,
-      "loss": 46.0,
-      "step": 330
-    },
-    {
-      "epoch": 0.053303273078626354,
-      "grad_norm": 0.001087374985218048,
-      "learning_rate": 0.00019999999868133955,
-      "loss": 46.0,
-      "step": 331
-    },
-    {
-      "epoch": 0.05346431015741374,
-      "grad_norm": 0.0014433509204536676,
-      "learning_rate": 0.0001999999986731108,
-      "loss": 46.0,
-      "step": 332
-    },
-    {
-      "epoch": 0.05362534723620113,
-      "grad_norm": 0.0008655768469907343,
-      "learning_rate": 0.00019999999866485645,
-      "loss": 46.0,
-      "step": 333
-    },
-    {
-      "epoch": 0.05378638431498853,
-      "grad_norm": 0.0005709116230718791,
-      "learning_rate": 0.00019999999865657654,
-      "loss": 46.0,
-      "step": 334
-    },
-    {
-      "epoch": 0.05394742139377592,
-      "grad_norm": 0.0012545662466436625,
-      "learning_rate": 0.000199999998648271,
-      "loss": 46.0,
-      "step": 335
-    },
-    {
-      "epoch": 0.054108458472563306,
-      "grad_norm": 0.0007371182437054813,
-      "learning_rate": 0.00019999999863993986,
-      "loss": 46.0,
-      "step": 336
-    },
-    {
-      "epoch": 0.054269495551350695,
-      "grad_norm": 0.0013507921248674393,
-      "learning_rate": 0.00019999999863158313,
-      "loss": 46.0,
-      "step": 337
-    },
-    {
-      "epoch": 0.05443053263013809,
-      "grad_norm": 0.0010357642313465476,
-      "learning_rate": 0.0001999999986232008,
-      "loss": 46.0,
-      "step": 338
-    },
-    {
-      "epoch": 0.05459156970892548,
-      "grad_norm": 0.0007762422901578248,
-      "learning_rate": 0.0001999999986147929,
-      "loss": 46.0,
-      "step": 339
-    },
-    {
-      "epoch": 0.05475260678771287,
-      "grad_norm": 0.001174192875623703,
-      "learning_rate": 0.00019999999860635938,
-      "loss": 46.0,
-      "step": 340
-    },
-    {
-      "epoch": 0.05491364386650026,
-      "grad_norm": 0.0011004736879840493,
-      "learning_rate": 0.00019999999859790027,
-      "loss": 46.0,
-      "step": 341
-    },
-    {
-      "epoch": 0.055074680945287655,
-      "grad_norm": 0.001205814303830266,
-      "learning_rate": 0.0001999999985894156,
-      "loss": 46.0,
-      "step": 342
-    },
-    {
-      "epoch": 0.055235718024075044,
-      "grad_norm": 0.0008946635643951595,
-      "learning_rate": 0.0001999999985809053,
-      "loss": 46.0,
-      "step": 343
-    },
-    {
-      "epoch": 0.05539675510286243,
-      "grad_norm": 0.0007763797184452415,
-      "learning_rate": 0.00019999999857236943,
-      "loss": 46.0,
-      "step": 344
-    },
-    {
-      "epoch": 0.05555779218164982,
-      "grad_norm": 0.0015236377948895097,
-      "learning_rate": 0.00019999999856380792,
-      "loss": 46.0,
-      "step": 345
-    },
-    {
-      "epoch": 0.05571882926043722,
-      "grad_norm": 0.001082977163605392,
-      "learning_rate": 0.00019999999855522084,
-      "loss": 46.0,
-      "step": 346
-    },
-    {
-      "epoch": 0.05587986633922461,
-      "grad_norm": 0.0005746458773501217,
-      "learning_rate": 0.00019999999854660816,
-      "loss": 46.0,
-      "step": 347
-    },
-    {
-      "epoch": 0.056040903418011996,
-      "grad_norm": 0.0004423210339155048,
-      "learning_rate": 0.00019999999853796991,
-      "loss": 46.0,
-      "step": 348
-    },
-    {
-      "epoch": 0.056201940496799385,
-      "grad_norm": 0.0007441586349159479,
-      "learning_rate": 0.00019999999852930605,
-      "loss": 46.0,
-      "step": 349
-    },
-    {
-      "epoch": 0.05636297757558678,
-      "grad_norm": 0.0011474787024781108,
-      "learning_rate": 0.00019999999852061658,
-      "loss": 46.0,
-      "step": 350
-    },
-    {
-      "epoch": 0.05652401465437417,
-      "grad_norm": 0.0006653806194663048,
-      "learning_rate": 0.00019999999851190152,
-      "loss": 46.0,
-      "step": 351
-    },
-    {
-      "epoch": 0.05668505173316156,
-      "grad_norm": 0.0015253883320838213,
-      "learning_rate": 0.00019999999850316087,
-      "loss": 46.0,
-      "step": 352
-    },
-    {
-      "epoch": 0.05684608881194895,
-      "grad_norm": 0.0003858486597891897,
-      "learning_rate": 0.00019999999849439464,
-      "loss": 46.0,
-      "step": 353
-    },
-    {
-      "epoch": 0.057007125890736345,
-      "grad_norm": 0.0011664539342746139,
-      "learning_rate": 0.00019999999848560281,
-      "loss": 46.0,
-      "step": 354
-    },
-    {
-      "epoch": 0.057168162969523734,
-      "grad_norm": 0.0007604304701089859,
-      "learning_rate": 0.00019999999847678538,
-      "loss": 46.0,
-      "step": 355
-    },
-    {
-      "epoch": 0.05732920004831112,
-      "grad_norm": 0.0009285574778914452,
-      "learning_rate": 0.00019999999846794233,
-      "loss": 46.0,
-      "step": 356
-    },
-    {
-      "epoch": 0.05749023712709851,
-      "grad_norm": 0.0010224482975900173,
-      "learning_rate": 0.00019999999845907374,
-      "loss": 46.0,
-      "step": 357
-    },
-    {
-      "epoch": 0.05765127420588591,
-      "grad_norm": 0.0008152456721290946,
-      "learning_rate": 0.00019999999845017952,
-      "loss": 46.0,
-      "step": 358
-    },
-    {
-      "epoch": 0.0578123112846733,
-      "grad_norm": 0.0008902782574295998,
-      "learning_rate": 0.0001999999984412597,
-      "loss": 46.0,
-      "step": 359
-    },
-    {
-      "epoch": 0.057973348363460686,
-      "grad_norm": 0.0013972967863082886,
-      "learning_rate": 0.0001999999984323143,
-      "loss": 46.0,
-      "step": 360
-    },
-    {
-      "epoch": 0.058134385442248075,
-      "grad_norm": 0.0006916502024978399,
-      "learning_rate": 0.0001999999984233433,
-      "loss": 46.0,
-      "step": 361
-    },
-    {
-      "epoch": 0.05829542252103547,
-      "grad_norm": 0.0008311839774250984,
-      "learning_rate": 0.0001999999984143467,
-      "loss": 46.0,
-      "step": 362
-    },
-    {
-      "epoch": 0.05845645959982286,
-      "grad_norm": 0.0012427597539499402,
-      "learning_rate": 0.0001999999984053245,
-      "loss": 46.0,
-      "step": 363
-    },
-    {
-      "epoch": 0.05861749667861025,
-      "grad_norm": 0.0006254159961827099,
-      "learning_rate": 0.00019999999839627672,
-      "loss": 46.0,
-      "step": 364
-    },
-    {
-      "epoch": 0.05877853375739764,
-      "grad_norm": 0.00047978339716792107,
-      "learning_rate": 0.0001999999983872033,
-      "loss": 46.0,
-      "step": 365
-    },
-    {
-      "epoch": 0.058939570836185035,
-      "grad_norm": 0.0009679915965534747,
-      "learning_rate": 0.00019999999837810436,
-      "loss": 46.0,
-      "step": 366
-    },
-    {
-      "epoch": 0.059100607914972424,
-      "grad_norm": 0.0010024794610217214,
-      "learning_rate": 0.00019999999836897977,
-      "loss": 46.0,
-      "step": 367
-    },
-    {
-      "epoch": 0.05926164499375981,
-      "grad_norm": 0.0008527908357791603,
-      "learning_rate": 0.00019999999835982962,
-      "loss": 46.0,
-      "step": 368
-    },
-    {
-      "epoch": 0.0594226820725472,
-      "grad_norm": 0.000915585202164948,
-      "learning_rate": 0.00019999999835065385,
-      "loss": 46.0,
-      "step": 369
-    },
-    {
-      "epoch": 0.0595837191513346,
-      "grad_norm": 0.000650250818580389,
-      "learning_rate": 0.0001999999983414525,
-      "loss": 46.0,
-      "step": 370
-    },
-    {
-      "epoch": 0.05974475623012199,
-      "grad_norm": 0.0009959996677935123,
-      "learning_rate": 0.00019999999833222556,
-      "loss": 46.0,
-      "step": 371
-    },
-    {
-      "epoch": 0.059905793308909376,
-      "grad_norm": 0.001148334820754826,
-      "learning_rate": 0.000199999998322973,
-      "loss": 46.0,
-      "step": 372
-    },
-    {
-      "epoch": 0.060066830387696765,
-      "grad_norm": 0.000527723110280931,
-      "learning_rate": 0.00019999999831369486,
-      "loss": 46.0,
-      "step": 373
-    },
-    {
-      "epoch": 0.06022786746648416,
-      "grad_norm": 0.0009392919600941241,
-      "learning_rate": 0.00019999999830439113,
-      "loss": 46.0,
-      "step": 374
-    },
-    {
-      "epoch": 0.06038890454527155,
-      "grad_norm": 0.0005747764371335506,
-      "learning_rate": 0.0001999999982950618,
-      "loss": 46.0,
-      "step": 375
-    },
-    {
-      "epoch": 0.06054994162405894,
-      "grad_norm": 0.0005832109018228948,
-      "learning_rate": 0.0001999999982857069,
-      "loss": 46.0,
-      "step": 376
-    },
-    {
-      "epoch": 0.06071097870284633,
-      "grad_norm": 0.0007533972384408116,
-      "learning_rate": 0.00019999999827632637,
-      "loss": 46.0,
-      "step": 377
-    },
-    {
-      "epoch": 0.060872015781633725,
-      "grad_norm": 0.0007178978412412107,
-      "learning_rate": 0.00019999999826692024,
-      "loss": 46.0,
-      "step": 378
-    },
-    {
-      "epoch": 0.061033052860421114,
-      "grad_norm": 0.0006495073321275413,
-      "learning_rate": 0.00019999999825748855,
-      "loss": 46.0,
-      "step": 379
-    },
-    {
-      "epoch": 0.0611940899392085,
-      "grad_norm": 0.0006937084253877401,
-      "learning_rate": 0.00019999999824803121,
-      "loss": 46.0,
-      "step": 380
-    },
-    {
-      "epoch": 0.06135512701799589,
-      "grad_norm": 0.0013413194101303816,
-      "learning_rate": 0.00019999999823854832,
-      "loss": 46.0,
-      "step": 381
-    },
-    {
-      "epoch": 0.06151616409678328,
-      "grad_norm": 0.0008035088540054858,
-      "learning_rate": 0.00019999999822903984,
-      "loss": 46.0,
-      "step": 382
-    },
-    {
-      "epoch": 0.06167720117557068,
-      "grad_norm": 0.0006494863773696125,
-      "learning_rate": 0.00019999999821950575,
-      "loss": 46.0,
-      "step": 383
-    },
-    {
-      "epoch": 0.061838238254358066,
-      "grad_norm": 0.0010893800063058734,
-      "learning_rate": 0.00019999999820994606,
-      "loss": 46.0,
-      "step": 384
-    },
-    {
-      "epoch": 0.061999275333145455,
-      "grad_norm": 0.0008796124602667987,
-      "learning_rate": 0.00019999999820036077,
-      "loss": 46.0,
-      "step": 385
-    },
-    {
-      "epoch": 0.062160312411932844,
-      "grad_norm": 0.0009478065185248852,
-      "learning_rate": 0.00019999999819074994,
-      "loss": 46.0,
-      "step": 386
-    },
-    {
-      "epoch": 0.06232134949072024,
-      "grad_norm": 0.0006546524236910045,
-      "learning_rate": 0.00019999999818111347,
-      "loss": 46.0,
-      "step": 387
-    },
-    {
-      "epoch": 0.06248238656950763,
-      "grad_norm": 0.0006173010915517807,
-      "learning_rate": 0.00019999999817145139,
-      "loss": 46.0,
-      "step": 388
-    },
-    {
-      "epoch": 0.06264342364829502,
-      "grad_norm": 0.0004461584612727165,
-      "learning_rate": 0.00019999999816176374,
-      "loss": 46.0,
-      "step": 389
-    },
-    {
-      "epoch": 0.06280446072708241,
-      "grad_norm": 0.0008034147322177887,
-      "learning_rate": 0.00019999999815205048,
-      "loss": 46.0,
-      "step": 390
-    },
-    {
-      "epoch": 0.0629654978058698,
-      "grad_norm": 0.0011473468039184809,
-      "learning_rate": 0.0001999999981423116,
-      "loss": 46.0,
-      "step": 391
-    },
-    {
-      "epoch": 0.06312653488465719,
-      "grad_norm": 0.000513106700964272,
-      "learning_rate": 0.0001999999981325472,
-      "loss": 46.0,
-      "step": 392
-    },
-    {
-      "epoch": 0.06328757196344459,
-      "grad_norm": 0.0006003491580486298,
-      "learning_rate": 0.00019999999812275713,
-      "loss": 46.0,
-      "step": 393
-    },
-    {
-      "epoch": 0.06344860904223197,
-      "grad_norm": 0.0006133901188150048,
-      "learning_rate": 0.0001999999981129415,
-      "loss": 46.0,
-      "step": 394
-    },
-    {
-      "epoch": 0.06360964612101937,
-      "grad_norm": 0.0014315887819975615,
-      "learning_rate": 0.00019999999810310027,
-      "loss": 46.0,
-      "step": 395
-    },
-    {
-      "epoch": 0.06377068319980675,
-      "grad_norm": 0.0012847103644162416,
-      "learning_rate": 0.00019999999809323346,
-      "loss": 46.0,
-      "step": 396
-    },
-    {
-      "epoch": 0.06393172027859415,
-      "grad_norm": 0.0006227092235349119,
-      "learning_rate": 0.00019999999808334104,
-      "loss": 46.0,
-      "step": 397
-    },
-    {
-      "epoch": 0.06409275735738154,
-      "grad_norm": 0.0009588095708750188,
-      "learning_rate": 0.00019999999807342303,
-      "loss": 46.0,
-      "step": 398
-    },
-    {
-      "epoch": 0.06425379443616892,
-      "grad_norm": 0.00042919005500152707,
-      "learning_rate": 0.00019999999806347943,
-      "loss": 46.0,
-      "step": 399
-    },
-    {
-      "epoch": 0.06441483151495632,
-      "grad_norm": 0.0005253739072941244,
-      "learning_rate": 0.00019999999805351025,
-      "loss": 46.0,
-      "step": 400
-    },
-    {
-      "epoch": 0.06457586859374372,
-      "grad_norm": 0.0010851717088371515,
-      "learning_rate": 0.00019999999804351545,
-      "loss": 46.0,
-      "step": 401
-    },
-    {
-      "epoch": 0.0647369056725311,
-      "grad_norm": 0.0006280020461417735,
-      "learning_rate": 0.00019999999803349506,
-      "loss": 46.0,
-      "step": 402
-    },
-    {
-      "epoch": 0.0648979427513185,
-      "grad_norm": 0.0008929861360229552,
-      "learning_rate": 0.00019999999802344903,
-      "loss": 46.0,
-      "step": 403
-    },
-    {
-      "epoch": 0.06505897983010588,
-      "grad_norm": 0.001072326092980802,
-      "learning_rate": 0.00019999999801337745,
-      "loss": 46.0,
-      "step": 404
-    },
-    {
-      "epoch": 0.06522001690889327,
-      "grad_norm": 0.0006430553621612489,
-      "learning_rate": 0.0001999999980032803,
-      "loss": 46.0,
-      "step": 405
-    },
-    {
-      "epoch": 0.06538105398768067,
-      "grad_norm": 0.0004205092554911971,
-      "learning_rate": 0.00019999999799315754,
-      "loss": 46.0,
-      "step": 406
-    },
-    {
-      "epoch": 0.06554209106646805,
-      "grad_norm": 0.0009531372343190014,
-      "learning_rate": 0.00019999999798300916,
-      "loss": 46.0,
-      "step": 407
-    },
-    {
-      "epoch": 0.06570312814525545,
-      "grad_norm": 0.000906719418708235,
-      "learning_rate": 0.0001999999979728352,
-      "loss": 46.0,
-      "step": 408
-    },
-    {
-      "epoch": 0.06586416522404284,
-      "grad_norm": 0.0006057941354811192,
-      "learning_rate": 0.00019999999796263565,
-      "loss": 46.0,
-      "step": 409
-    },
-    {
-      "epoch": 0.06602520230283022,
-      "grad_norm": 0.0012407077010720968,
-      "learning_rate": 0.00019999999795241048,
-      "loss": 46.0,
-      "step": 410
-    },
-    {
-      "epoch": 0.06618623938161762,
-      "grad_norm": 0.0007258470286615193,
-      "learning_rate": 0.00019999999794215976,
-      "loss": 46.0,
-      "step": 411
-    },
-    {
-      "epoch": 0.066347276460405,
-      "grad_norm": 0.0007349825464189053,
-      "learning_rate": 0.0001999999979318834,
-      "loss": 46.0,
-      "step": 412
-    },
-    {
-      "epoch": 0.0665083135391924,
-      "grad_norm": 0.0008235921850427985,
-      "learning_rate": 0.00019999999792158146,
-      "loss": 46.0,
-      "step": 413
-    },
-    {
-      "epoch": 0.0666693506179798,
-      "grad_norm": 0.0005888417363166809,
-      "learning_rate": 0.00019999999791125395,
-      "loss": 46.0,
-      "step": 414
-    },
-    {
-      "epoch": 0.06683038769676718,
-      "grad_norm": 0.0008665561908856034,
-      "learning_rate": 0.0001999999979009008,
-      "loss": 46.0,
-      "step": 415
-    },
-    {
-      "epoch": 0.06699142477555457,
-      "grad_norm": 0.0012398657854646444,
-      "learning_rate": 0.00019999999789052208,
-      "loss": 46.0,
-      "step": 416
-    },
-    {
-      "epoch": 0.06715246185434197,
-      "grad_norm": 0.0007625618600286543,
-      "learning_rate": 0.00019999999788011778,
-      "loss": 46.0,
-      "step": 417
-    },
-    {
-      "epoch": 0.06731349893312935,
-      "grad_norm": 0.0009577570017427206,
-      "learning_rate": 0.00019999999786968786,
-      "loss": 46.0,
-      "step": 418
-    },
-    {
-      "epoch": 0.06747453601191675,
-      "grad_norm": 0.0010338622378185391,
-      "learning_rate": 0.00019999999785923236,
-      "loss": 46.0,
-      "step": 419
-    },
-    {
-      "epoch": 0.06763557309070413,
-      "grad_norm": 0.0006397149409167469,
-      "learning_rate": 0.00019999999784875127,
-      "loss": 46.0,
-      "step": 420
-    },
-    {
-      "epoch": 0.06779661016949153,
-      "grad_norm": 0.0005648143123835325,
-      "learning_rate": 0.00019999999783824456,
-      "loss": 46.0,
-      "step": 421
-    },
-    {
-      "epoch": 0.06795764724827892,
-      "grad_norm": 0.0004593702033162117,
-      "learning_rate": 0.00019999999782771227,
-      "loss": 46.0,
-      "step": 422
-    },
-    {
-      "epoch": 0.0681186843270663,
-      "grad_norm": 0.0006899412255734205,
-      "learning_rate": 0.00019999999781715442,
-      "loss": 46.0,
-      "step": 423
-    },
-    {
-      "epoch": 0.0682797214058537,
-      "grad_norm": 0.0006216235924512148,
-      "learning_rate": 0.00019999999780657092,
-      "loss": 46.0,
-      "step": 424
-    },
-    {
-      "epoch": 0.0684407584846411,
-      "grad_norm": 0.0006858006818220019,
-      "learning_rate": 0.00019999999779596187,
-      "loss": 46.0,
-      "step": 425
-    },
-    {
-      "epoch": 0.06860179556342848,
-      "grad_norm": 0.0012070111697539687,
-      "learning_rate": 0.00019999999778532717,
-      "loss": 46.0,
-      "step": 426
-    },
-    {
-      "epoch": 0.06876283264221587,
-      "grad_norm": 0.0007498610648326576,
-      "learning_rate": 0.00019999999777466695,
-      "loss": 46.0,
-      "step": 427
-    },
-    {
-      "epoch": 0.06892386972100326,
-      "grad_norm": 0.0006476924754679203,
-      "learning_rate": 0.00019999999776398108,
-      "loss": 46.0,
-      "step": 428
-    },
-    {
-      "epoch": 0.06908490679979065,
-      "grad_norm": 0.0010762411402538419,
-      "learning_rate": 0.00019999999775326962,
-      "loss": 46.0,
-      "step": 429
-    },
-    {
-      "epoch": 0.06924594387857805,
-      "grad_norm": 0.000566616712603718,
-      "learning_rate": 0.00019999999774253255,
-      "loss": 46.0,
-      "step": 430
-    },
-    {
-      "epoch": 0.06940698095736543,
-      "grad_norm": 0.000477853900520131,
-      "learning_rate": 0.00019999999773176992,
-      "loss": 46.0,
-      "step": 431
-    },
-    {
-      "epoch": 0.06956801803615283,
-      "grad_norm": 0.0006761650438420475,
-      "learning_rate": 0.0001999999977209817,
-      "loss": 46.0,
-      "step": 432
-    },
-    {
-      "epoch": 0.06972905511494021,
-      "grad_norm": 0.0007774282712489367,
-      "learning_rate": 0.00019999999771016787,
-      "loss": 46.0,
-      "step": 433
-    },
-    {
-      "epoch": 0.0698900921937276,
-      "grad_norm": 0.0011541653657332063,
-      "learning_rate": 0.00019999999769932845,
-      "loss": 46.0,
-      "step": 434
-    },
-    {
-      "epoch": 0.070051129272515,
-      "grad_norm": 0.0012715215561911464,
-      "learning_rate": 0.00019999999768846341,
-      "loss": 46.0,
-      "step": 435
-    },
-    {
-      "epoch": 0.07021216635130238,
-      "grad_norm": 0.0007645237492397428,
-      "learning_rate": 0.0001999999976775728,
-      "loss": 46.0,
-      "step": 436
-    },
-    {
-      "epoch": 0.07037320343008978,
-      "grad_norm": 0.001217718468979001,
-      "learning_rate": 0.00019999999766665658,
-      "loss": 46.0,
-      "step": 437
-    },
-    {
-      "epoch": 0.07053424050887717,
-      "grad_norm": 0.0006087617948651314,
-      "learning_rate": 0.0001999999976557148,
-      "loss": 46.0,
-      "step": 438
-    },
-    {
-      "epoch": 0.07069527758766456,
-      "grad_norm": 0.0011646051425486803,
-      "learning_rate": 0.00019999999764474738,
-      "loss": 46.0,
-      "step": 439
-    },
-    {
-      "epoch": 0.07085631466645195,
-      "grad_norm": 0.0016641434049233794,
-      "learning_rate": 0.00019999999763375436,
-      "loss": 46.0,
-      "step": 440
-    },
-    {
-      "epoch": 0.07101735174523933,
-      "grad_norm": 0.0011168629862368107,
-      "learning_rate": 0.00019999999762273577,
-      "loss": 46.0,
-      "step": 441
-    },
-    {
-      "epoch": 0.07117838882402673,
-      "grad_norm": 0.0006337081431411207,
-      "learning_rate": 0.00019999999761169163,
-      "loss": 46.0,
-      "step": 442
-    },
-    {
-      "epoch": 0.07133942590281413,
-      "grad_norm": 0.0006407672772184014,
-      "learning_rate": 0.00019999999760062181,
-      "loss": 46.0,
-      "step": 443
-    },
-    {
-      "epoch": 0.07150046298160151,
-      "grad_norm": 0.0007342903991229832,
-      "learning_rate": 0.00019999999758952647,
-      "loss": 46.0,
-      "step": 444
-    },
-    {
-      "epoch": 0.0716615000603889,
-      "grad_norm": 0.0006470965454354882,
-      "learning_rate": 0.00019999999757840548,
-      "loss": 46.0,
-      "step": 445
-    },
-    {
-      "epoch": 0.0718225371391763,
-      "grad_norm": 0.000804529117885977,
-      "learning_rate": 0.00019999999756725888,
-      "loss": 46.0,
-      "step": 446
-    },
-    {
-      "epoch": 0.07198357421796368,
-      "grad_norm": 0.0006461947341449559,
-      "learning_rate": 0.00019999999755608675,
-      "loss": 46.0,
-      "step": 447
-    },
-    {
-      "epoch": 0.07214461129675108,
-      "grad_norm": 0.0006891589728184044,
-      "learning_rate": 0.000199999997544889,
-      "loss": 46.0,
-      "step": 448
-    },
-    {
-      "epoch": 0.07230564837553846,
-      "grad_norm": 0.0013634568313136697,
-      "learning_rate": 0.00019999999753366564,
-      "loss": 46.0,
-      "step": 449
-    },
-    {
-      "epoch": 0.07246668545432586,
-      "grad_norm": 0.0006031526136212051,
-      "learning_rate": 0.00019999999752241669,
-      "loss": 46.0,
-      "step": 450
-    },
-    {
-      "epoch": 0.07262772253311325,
-      "grad_norm": 0.0006277126958593726,
-      "learning_rate": 0.00019999999751114215,
-      "loss": 46.0,
-      "step": 451
-    },
-    {
-      "epoch": 0.07278875961190064,
-      "grad_norm": 0.001542499870993197,
-      "learning_rate": 0.00019999999749984202,
-      "loss": 46.0,
-      "step": 452
-    },
-    {
-      "epoch": 0.07294979669068803,
-      "grad_norm": 0.0007459594053216279,
-      "learning_rate": 0.00019999999748851629,
-      "loss": 46.0,
-      "step": 453
-    },
-    {
-      "epoch": 0.07311083376947543,
-      "grad_norm": 0.0008035858627408743,
-      "learning_rate": 0.00019999999747716496,
-      "loss": 46.0,
-      "step": 454
-    },
-    {
-      "epoch": 0.07327187084826281,
-      "grad_norm": 0.0007284557796083391,
-      "learning_rate": 0.00019999999746578805,
-      "loss": 46.0,
-      "step": 455
-    },
-    {
-      "epoch": 0.0734329079270502,
-      "grad_norm": 0.0009163875947706401,
-      "learning_rate": 0.00019999999745438555,
-      "loss": 46.0,
-      "step": 456
-    },
-    {
-      "epoch": 0.07359394500583759,
-      "grad_norm": 0.0004265153256710619,
-      "learning_rate": 0.00019999999744295743,
-      "loss": 46.0,
-      "step": 457
-    },
-    {
-      "epoch": 0.07375498208462498,
-      "grad_norm": 0.00032776183797977865,
-      "learning_rate": 0.00019999999743150373,
-      "loss": 46.0,
-      "step": 458
-    },
-    {
-      "epoch": 0.07391601916341238,
-      "grad_norm": 0.0007664909935556352,
-      "learning_rate": 0.00019999999742002442,
-      "loss": 46.0,
-      "step": 459
-    },
-    {
-      "epoch": 0.07407705624219976,
-      "grad_norm": 0.000509080127812922,
-      "learning_rate": 0.00019999999740851954,
-      "loss": 46.0,
-      "step": 460
-    },
-    {
-      "epoch": 0.07423809332098716,
-      "grad_norm": 0.0005612930981442332,
-      "learning_rate": 0.00019999999739698905,
-      "loss": 46.0,
-      "step": 461
-    },
-    {
-      "epoch": 0.07439913039977455,
-      "grad_norm": 0.0009621938224881887,
-      "learning_rate": 0.00019999999738543295,
-      "loss": 46.0,
-      "step": 462
-    },
-    {
-      "epoch": 0.07456016747856194,
-      "grad_norm": 0.0008162377635017037,
-      "learning_rate": 0.00019999999737385128,
-      "loss": 46.0,
-      "step": 463
-    },
-    {
-      "epoch": 0.07472120455734933,
-      "grad_norm": 0.0004411375557538122,
-      "learning_rate": 0.000199999997362244,
-      "loss": 46.0,
-      "step": 464
-    },
-    {
-      "epoch": 0.07488224163613671,
-      "grad_norm": 0.00037210877053439617,
-      "learning_rate": 0.00019999999735061114,
-      "loss": 46.0,
-      "step": 465
-    },
-    {
-      "epoch": 0.07504327871492411,
-      "grad_norm": 0.0011236772406846285,
-      "learning_rate": 0.00019999999733895268,
-      "loss": 46.0,
-      "step": 466
-    },
-    {
-      "epoch": 0.07520431579371151,
-      "grad_norm": 0.0008054023492150009,
-      "learning_rate": 0.00019999999732726862,
-      "loss": 46.0,
-      "step": 467
-    },
-    {
-      "epoch": 0.07536535287249889,
-      "grad_norm": 0.0004958999343216419,
-      "learning_rate": 0.00019999999731555896,
-      "loss": 46.0,
-      "step": 468
-    },
-    {
-      "epoch": 0.07552638995128629,
-      "grad_norm": 0.0007969607831910253,
-      "learning_rate": 0.0001999999973038237,
-      "loss": 46.0,
-      "step": 469
-    },
-    {
-      "epoch": 0.07568742703007368,
-      "grad_norm": 0.0011302282800897956,
-      "learning_rate": 0.0001999999972920629,
-      "loss": 46.0,
-      "step": 470
-    },
-    {
-      "epoch": 0.07584846410886106,
-      "grad_norm": 0.0009448499768041074,
-      "learning_rate": 0.00019999999728027645,
-      "loss": 46.0,
-      "step": 471
-    },
-    {
-      "epoch": 0.07600950118764846,
-      "grad_norm": 0.0007362283649854362,
-      "learning_rate": 0.00019999999726846442,
-      "loss": 46.0,
-      "step": 472
-    },
-    {
-      "epoch": 0.07617053826643584,
-      "grad_norm": 0.0009826513705775142,
-      "learning_rate": 0.0001999999972566268,
-      "loss": 46.0,
-      "step": 473
-    },
-    {
-      "epoch": 0.07633157534522324,
-      "grad_norm": 0.0010123316897079349,
-      "learning_rate": 0.00019999999724476357,
-      "loss": 46.0,
-      "step": 474
-    },
-    {
-      "epoch": 0.07649261242401063,
-      "grad_norm": 0.0007657674723304808,
-      "learning_rate": 0.00019999999723287475,
-      "loss": 46.0,
-      "step": 475
-    },
-    {
-      "epoch": 0.07665364950279802,
-      "grad_norm": 0.0010322603629902005,
-      "learning_rate": 0.00019999999722096034,
-      "loss": 46.0,
-      "step": 476
-    },
-    {
-      "epoch": 0.07681468658158541,
-      "grad_norm": 0.0012446970213204622,
-      "learning_rate": 0.00019999999720902035,
-      "loss": 46.0,
-      "step": 477
-    },
-    {
-      "epoch": 0.0769757236603728,
-      "grad_norm": 0.0008477270603179932,
-      "learning_rate": 0.00019999999719705472,
-      "loss": 46.0,
-      "step": 478
-    },
-    {
-      "epoch": 0.07713676073916019,
-      "grad_norm": 0.0005673026898875833,
-      "learning_rate": 0.00019999999718506352,
-      "loss": 46.0,
-      "step": 479
-    },
-    {
-      "epoch": 0.07729779781794759,
-      "grad_norm": 0.0006731751491315663,
-      "learning_rate": 0.00019999999717304674,
-      "loss": 46.0,
-      "step": 480
-    },
-    {
-      "epoch": 0.07745883489673497,
-      "grad_norm": 0.0004351187963038683,
-      "learning_rate": 0.00019999999716100434,
-      "loss": 46.0,
-      "step": 481
-    },
-    {
-      "epoch": 0.07761987197552236,
-      "grad_norm": 0.0007636580266989768,
-      "learning_rate": 0.00019999999714893636,
-      "loss": 46.0,
-      "step": 482
-    },
-    {
-      "epoch": 0.07778090905430976,
-      "grad_norm": 0.000572064018342644,
-      "learning_rate": 0.0001999999971368428,
-      "loss": 46.0,
-      "step": 483
-    },
-    {
-      "epoch": 0.07794194613309714,
-      "grad_norm": 0.0005930215120315552,
-      "learning_rate": 0.00019999999712472363,
-      "loss": 46.0,
-      "step": 484
-    },
-    {
-      "epoch": 0.07810298321188454,
-      "grad_norm": 0.0006548817036673427,
-      "learning_rate": 0.00019999999711257886,
-      "loss": 46.0,
-      "step": 485
-    },
-    {
-      "epoch": 0.07826402029067192,
-      "grad_norm": 0.000415973539929837,
-      "learning_rate": 0.0001999999971004085,
-      "loss": 46.0,
-      "step": 486
-    },
-    {
-      "epoch": 0.07842505736945932,
-      "grad_norm": 0.0005942594725638628,
-      "learning_rate": 0.00019999999708821253,
-      "loss": 46.0,
-      "step": 487
-    },
-    {
-      "epoch": 0.07858609444824671,
-      "grad_norm": 0.0005471553304232657,
-      "learning_rate": 0.000199999997075991,
-      "loss": 46.0,
-      "step": 488
-    },
-    {
-      "epoch": 0.0787471315270341,
-      "grad_norm": 0.0005777325131930411,
-      "learning_rate": 0.00019999999706374385,
-      "loss": 46.0,
-      "step": 489
-    },
-    {
-      "epoch": 0.07890816860582149,
-      "grad_norm": 0.000728434999473393,
-      "learning_rate": 0.0001999999970514711,
-      "loss": 46.0,
-      "step": 490
-    },
-    {
-      "epoch": 0.07906920568460889,
-      "grad_norm": 0.0009583527571521699,
-      "learning_rate": 0.0001999999970391728,
-      "loss": 46.0,
-      "step": 491
-    },
-    {
-      "epoch": 0.07923024276339627,
-      "grad_norm": 0.0007965164259076118,
-      "learning_rate": 0.00019999999702684885,
-      "loss": 46.0,
-      "step": 492
-    },
-    {
-      "epoch": 0.07939127984218367,
-      "grad_norm": 0.0005470888572745025,
-      "learning_rate": 0.00019999999701449933,
-      "loss": 46.0,
-      "step": 493
-    },
-    {
-      "epoch": 0.07955231692097105,
-      "grad_norm": 0.0009533630218356848,
-      "learning_rate": 0.0001999999970021242,
-      "loss": 46.0,
-      "step": 494
-    },
-    {
-      "epoch": 0.07971335399975844,
-      "grad_norm": 0.0010155423078686,
-      "learning_rate": 0.0001999999969897235,
-      "loss": 46.0,
-      "step": 495
-    },
-    {
-      "epoch": 0.07987439107854584,
-      "grad_norm": 0.0009387807222083211,
-      "learning_rate": 0.00019999999697729715,
-      "loss": 46.0,
-      "step": 496
-    },
-    {
-      "epoch": 0.08003542815733322,
-      "grad_norm": 0.0007800189196132123,
-      "learning_rate": 0.00019999999696484528,
-      "loss": 46.0,
-      "step": 497
-    },
-    {
-      "epoch": 0.08019646523612062,
-      "grad_norm": 0.001141066080890596,
-      "learning_rate": 0.00019999999695236777,
-      "loss": 46.0,
-      "step": 498
-    },
-    {
-      "epoch": 0.08035750231490801,
-      "grad_norm": 0.0007430302212014794,
-      "learning_rate": 0.0001999999969398647,
-      "loss": 46.0,
-      "step": 499
-    },
-    {
-      "epoch": 0.0805185393936954,
-      "grad_norm": 0.0011654486879706383,
-      "learning_rate": 0.000199999996927336,
-      "loss": 46.0,
-      "step": 500
-    },
-    {
-      "epoch": 0.08067957647248279,
-      "grad_norm": 0.001330543658696115,
-      "learning_rate": 0.0001999999969147817,
-      "loss": 46.0,
-      "step": 501
-    },
-    {
-      "epoch": 0.08084061355127017,
-      "grad_norm": 0.0007218665559776127,
-      "learning_rate": 0.00019999999690220184,
-      "loss": 46.0,
-      "step": 502
-    },
-    {
-      "epoch": 0.08100165063005757,
-      "grad_norm": 0.000574540754314512,
-      "learning_rate": 0.00019999999688959636,
-      "loss": 46.0,
-      "step": 503
-    },
-    {
-      "epoch": 0.08116268770884497,
-      "grad_norm": 0.0006847270997241139,
-      "learning_rate": 0.0001999999968769653,
-      "loss": 46.0,
-      "step": 504
-    },
-    {
-      "epoch": 0.08132372478763235,
-      "grad_norm": 0.0005952648934908211,
-      "learning_rate": 0.00019999999686430862,
-      "loss": 46.0,
-      "step": 505
-    },
-    {
-      "epoch": 0.08148476186641974,
-      "grad_norm": 0.0008884788258001208,
-      "learning_rate": 0.00019999999685162636,
-      "loss": 46.0,
-      "step": 506
-    },
-    {
-      "epoch": 0.08164579894520714,
-      "grad_norm": 0.000789961835835129,
-      "learning_rate": 0.00019999999683891853,
-      "loss": 46.0,
-      "step": 507
-    },
-    {
-      "epoch": 0.08180683602399452,
-      "grad_norm": 0.000833467289339751,
-      "learning_rate": 0.00019999999682618507,
-      "loss": 46.0,
-      "step": 508
-    },
-    {
-      "epoch": 0.08196787310278192,
-      "grad_norm": 0.0014145354507490993,
-      "learning_rate": 0.00019999999681342604,
-      "loss": 46.0,
-      "step": 509
-    },
-    {
-      "epoch": 0.0821289101815693,
-      "grad_norm": 0.0005128522752784193,
-      "learning_rate": 0.0001999999968006414,
-      "loss": 46.0,
-      "step": 510
-    },
-    {
-      "epoch": 0.0822899472603567,
-      "grad_norm": 0.0005069162580184639,
-      "learning_rate": 0.00019999999678783117,
-      "loss": 46.0,
-      "step": 511
-    },
-    {
-      "epoch": 0.08245098433914409,
-      "grad_norm": 0.0009712486644275486,
-      "learning_rate": 0.00019999999677499535,
-      "loss": 46.0,
-      "step": 512
-    },
-    {
-      "epoch": 0.08261202141793147,
-      "grad_norm": 0.0004906330723315477,
-      "learning_rate": 0.0001999999967621339,
-      "loss": 46.0,
-      "step": 513
-    },
-    {
-      "epoch": 0.08277305849671887,
-      "grad_norm": 0.0009142711060121655,
-      "learning_rate": 0.0001999999967492469,
-      "loss": 46.0,
-      "step": 514
-    },
-    {
-      "epoch": 0.08293409557550627,
-      "grad_norm": 0.0008989947964437306,
-      "learning_rate": 0.00019999999673633428,
-      "loss": 46.0,
-      "step": 515
-    },
-    {
-      "epoch": 0.08309513265429365,
-      "grad_norm": 0.0005218035075813532,
-      "learning_rate": 0.0001999999967233961,
-      "loss": 46.0,
-      "step": 516
-    },
-    {
-      "epoch": 0.08325616973308105,
-      "grad_norm": 0.0009906719205901027,
-      "learning_rate": 0.00019999999671043228,
-      "loss": 46.0,
-      "step": 517
-    },
-    {
-      "epoch": 0.08341720681186843,
-      "grad_norm": 0.000553377321921289,
-      "learning_rate": 0.0001999999966974429,
-      "loss": 46.0,
-      "step": 518
-    },
-    {
-      "epoch": 0.08357824389065582,
-      "grad_norm": 0.0003130970289930701,
-      "learning_rate": 0.00019999999668442788,
-      "loss": 46.0,
-      "step": 519
-    },
-    {
-      "epoch": 0.08373928096944322,
-      "grad_norm": 0.0007157115614973009,
-      "learning_rate": 0.00019999999667138731,
-      "loss": 46.0,
-      "step": 520
-    },
-    {
-      "epoch": 0.0839003180482306,
-      "grad_norm": 0.0005627658101730049,
-      "learning_rate": 0.00019999999665832113,
-      "loss": 46.0,
-      "step": 521
-    },
-    {
-      "epoch": 0.084061355127018,
-      "grad_norm": 0.0005468367016874254,
-      "learning_rate": 0.00019999999664522934,
-      "loss": 46.0,
-      "step": 522
-    },
-    {
-      "epoch": 0.08422239220580538,
-      "grad_norm": 0.0007199125830084085,
-      "learning_rate": 0.000199999996632112,
-      "loss": 46.0,
-      "step": 523
-    },
-    {
-      "epoch": 0.08438342928459278,
-      "grad_norm": 0.0007692417129874229,
-      "learning_rate": 0.000199999996618969,
-      "loss": 46.0,
-      "step": 524
-    },
-    {
-      "epoch": 0.08454446636338017,
-      "grad_norm": 0.000630500609986484,
-      "learning_rate": 0.00019999999660580046,
-      "loss": 46.0,
-      "step": 525
-    },
-    {
-      "epoch": 0.08470550344216755,
-      "grad_norm": 0.0014079947723075747,
-      "learning_rate": 0.0001999999965926063,
-      "loss": 46.0,
-      "step": 526
-    },
-    {
-      "epoch": 0.08486654052095495,
-      "grad_norm": 0.0006229651044122875,
-      "learning_rate": 0.00019999999657938655,
-      "loss": 46.0,
-      "step": 527
-    },
-    {
-      "epoch": 0.08502757759974235,
-      "grad_norm": 0.000505558040458709,
-      "learning_rate": 0.0001999999965661412,
-      "loss": 46.0,
-      "step": 528
-    },
-    {
-      "epoch": 0.08518861467852973,
-      "grad_norm": 0.0007336870185099542,
-      "learning_rate": 0.00019999999655287028,
-      "loss": 46.0,
-      "step": 529
-    },
-    {
-      "epoch": 0.08534965175731712,
-      "grad_norm": 0.0004319110594224185,
-      "learning_rate": 0.00019999999653957373,
-      "loss": 46.0,
-      "step": 530
-    },
-    {
-      "epoch": 0.0855106888361045,
-      "grad_norm": 0.0007156056235544384,
-      "learning_rate": 0.00019999999652625162,
-      "loss": 46.0,
-      "step": 531
-    },
-    {
-      "epoch": 0.0856717259148919,
-      "grad_norm": 0.0006044222391210496,
-      "learning_rate": 0.0001999999965129039,
-      "loss": 46.0,
-      "step": 532
-    },
-    {
-      "epoch": 0.0858327629936793,
-      "grad_norm": 0.0004857086460106075,
-      "learning_rate": 0.00019999999649953056,
-      "loss": 46.0,
-      "step": 533
-    },
-    {
-      "epoch": 0.08599380007246668,
-      "grad_norm": 0.0006408842164091766,
-      "learning_rate": 0.00019999999648613167,
-      "loss": 46.0,
-      "step": 534
-    },
-    {
-      "epoch": 0.08615483715125408,
-      "grad_norm": 0.000600326107814908,
-      "learning_rate": 0.00019999999647270716,
-      "loss": 46.0,
-      "step": 535
-    },
-    {
-      "epoch": 0.08631587423004147,
-      "grad_norm": 0.0006083932239562273,
-      "learning_rate": 0.00019999999645925703,
-      "loss": 46.0,
-      "step": 536
-    },
-    {
-      "epoch": 0.08647691130882885,
-      "grad_norm": 0.00048264962970279157,
-      "learning_rate": 0.00019999999644578132,
-      "loss": 46.0,
-      "step": 537
-    },
-    {
-      "epoch": 0.08663794838761625,
-      "grad_norm": 0.0007165013230405748,
-      "learning_rate": 0.00019999999643228005,
-      "loss": 46.0,
-      "step": 538
-    },
-    {
-      "epoch": 0.08679898546640363,
-      "grad_norm": 0.000687831430695951,
-      "learning_rate": 0.00019999999641875317,
-      "loss": 46.0,
-      "step": 539
-    },
-    {
-      "epoch": 0.08696002254519103,
-      "grad_norm": 0.0006070000817999244,
-      "learning_rate": 0.00019999999640520067,
-      "loss": 46.0,
-      "step": 540
-    },
-    {
-      "epoch": 0.08712105962397843,
-      "grad_norm": 0.0006539839087054133,
-      "learning_rate": 0.00019999999639162258,
-      "loss": 46.0,
-      "step": 541
-    },
-    {
-      "epoch": 0.08728209670276581,
-      "grad_norm": 0.0005195296835154295,
-      "learning_rate": 0.00019999999637801893,
-      "loss": 46.0,
-      "step": 542
-    },
-    {
-      "epoch": 0.0874431337815532,
-      "grad_norm": 0.0006608666735701263,
-      "learning_rate": 0.00019999999636438967,
-      "loss": 46.0,
-      "step": 543
-    },
-    {
-      "epoch": 0.0876041708603406,
-      "grad_norm": 0.001367369550280273,
-      "learning_rate": 0.00019999999635073482,
-      "loss": 46.0,
-      "step": 544
-    },
-    {
-      "epoch": 0.08776520793912798,
-      "grad_norm": 0.000743956072255969,
-      "learning_rate": 0.00019999999633705433,
-      "loss": 46.0,
-      "step": 545
-    },
-    {
-      "epoch": 0.08792624501791538,
-      "grad_norm": 0.0008423933759331703,
-      "learning_rate": 0.00019999999632334828,
-      "loss": 46.0,
-      "step": 546
-    },
-    {
-      "epoch": 0.08808728209670276,
-      "grad_norm": 0.0011517110979184508,
-      "learning_rate": 0.00019999999630961664,
-      "loss": 46.0,
-      "step": 547
-    },
-    {
-      "epoch": 0.08824831917549016,
-      "grad_norm": 0.0010252471547573805,
-      "learning_rate": 0.0001999999962958594,
-      "loss": 46.0,
-      "step": 548
-    },
-    {
-      "epoch": 0.08840935625427755,
-      "grad_norm": 0.00048051943304017186,
-      "learning_rate": 0.00019999999628207658,
-      "loss": 46.0,
-      "step": 549
-    },
-    {
-      "epoch": 0.08857039333306493,
-      "grad_norm": 0.0007046059472486377,
-      "learning_rate": 0.00019999999626826813,
-      "loss": 46.0,
-      "step": 550
-    },
-    {
-      "epoch": 0.08873143041185233,
-      "grad_norm": 0.0010297683766111732,
-      "learning_rate": 0.0001999999962544341,
-      "loss": 46.0,
-      "step": 551
-    },
-    {
-      "epoch": 0.08889246749063973,
-      "grad_norm": 0.0005623582983389497,
-      "learning_rate": 0.00019999999624057449,
-      "loss": 46.0,
-      "step": 552
-    },
-    {
-      "epoch": 0.08905350456942711,
-      "grad_norm": 0.0010210892651230097,
-      "learning_rate": 0.00019999999622668924,
-      "loss": 46.0,
-      "step": 553
-    },
-    {
-      "epoch": 0.0892145416482145,
-      "grad_norm": 0.0010434873402118683,
-      "learning_rate": 0.00019999999621277844,
-      "loss": 46.0,
-      "step": 554
-    },
-    {
-      "epoch": 0.08937557872700189,
-      "grad_norm": 0.0007033632718957961,
-      "learning_rate": 0.00019999999619884205,
-      "loss": 46.0,
-      "step": 555
-    },
-    {
-      "epoch": 0.08953661580578928,
-      "grad_norm": 0.0005892971530556679,
-      "learning_rate": 0.00019999999618488005,
-      "loss": 46.0,
-      "step": 556
-    },
-    {
-      "epoch": 0.08969765288457668,
-      "grad_norm": 0.0006412552902474999,
-      "learning_rate": 0.00019999999617089243,
-      "loss": 46.0,
-      "step": 557
-    },
-    {
-      "epoch": 0.08985868996336406,
-      "grad_norm": 0.0007516163750551641,
-      "learning_rate": 0.00019999999615687926,
-      "loss": 46.0,
-      "step": 558
-    },
-    {
-      "epoch": 0.09001972704215146,
-      "grad_norm": 0.0005776139441877604,
-      "learning_rate": 0.0001999999961428405,
-      "loss": 46.0,
-      "step": 559
-    },
-    {
-      "epoch": 0.09018076412093885,
-      "grad_norm": 0.0007281535072252154,
-      "learning_rate": 0.00019999999612877608,
-      "loss": 46.0,
-      "step": 560
-    },
-    {
-      "epoch": 0.09034180119972623,
-      "grad_norm": 0.0009508816292509437,
-      "learning_rate": 0.00019999999611468612,
-      "loss": 46.0,
-      "step": 561
-    },
-    {
-      "epoch": 0.09050283827851363,
-      "grad_norm": 0.0005823606043122709,
-      "learning_rate": 0.00019999999610057054,
-      "loss": 46.0,
-      "step": 562
-    },
-    {
-      "epoch": 0.09066387535730101,
-      "grad_norm": 0.000553961144760251,
-      "learning_rate": 0.0001999999960864294,
-      "loss": 46.0,
-      "step": 563
-    },
-    {
-      "epoch": 0.09082491243608841,
-      "grad_norm": 0.0005303358775563538,
-      "learning_rate": 0.00019999999607226264,
-      "loss": 46.0,
-      "step": 564
-    },
-    {
-      "epoch": 0.0909859495148758,
-      "grad_norm": 0.0010254705557599664,
-      "learning_rate": 0.00019999999605807027,
-      "loss": 46.0,
-      "step": 565
-    },
-    {
-      "epoch": 0.09114698659366319,
-      "grad_norm": 0.001085357740521431,
-      "learning_rate": 0.00019999999604385234,
-      "loss": 46.0,
-      "step": 566
-    },
-    {
-      "epoch": 0.09130802367245058,
-      "grad_norm": 0.0015264761168509722,
-      "learning_rate": 0.00019999999602960877,
-      "loss": 46.0,
-      "step": 567
-    },
-    {
-      "epoch": 0.09146906075123798,
-      "grad_norm": 0.0005250285030342638,
-      "learning_rate": 0.00019999999601533964,
-      "loss": 46.0,
-      "step": 568
-    },
-    {
-      "epoch": 0.09163009783002536,
-      "grad_norm": 0.0004584493872243911,
-      "learning_rate": 0.0001999999960010449,
-      "loss": 46.0,
-      "step": 569
-    },
-    {
-      "epoch": 0.09179113490881276,
-      "grad_norm": 0.0019098113989457488,
-      "learning_rate": 0.00019999999598672457,
-      "loss": 46.0,
-      "step": 570
-    },
-    {
-      "epoch": 0.09195217198760014,
-      "grad_norm": 0.0005570544744841754,
-      "learning_rate": 0.00019999999597237867,
-      "loss": 46.0,
-      "step": 571
-    },
-    {
-      "epoch": 0.09211320906638754,
-      "grad_norm": 0.0012722942046821117,
-      "learning_rate": 0.00019999999595800714,
-      "loss": 46.0,
-      "step": 572
-    },
-    {
-      "epoch": 0.09227424614517493,
-      "grad_norm": 0.0006153949652798474,
-      "learning_rate": 0.00019999999594361002,
-      "loss": 46.0,
-      "step": 573
-    },
-    {
-      "epoch": 0.09243528322396231,
-      "grad_norm": 0.0004970795125700533,
-      "learning_rate": 0.0001999999959291873,
-      "loss": 46.0,
-      "step": 574
-    },
-    {
-      "epoch": 0.09259632030274971,
-      "grad_norm": 0.0006614192971028388,
-      "learning_rate": 0.000199999995914739,
-      "loss": 46.0,
-      "step": 575
-    },
-    {
-      "epoch": 0.09275735738153709,
-      "grad_norm": 0.0006567550590261817,
-      "learning_rate": 0.0001999999959002651,
-      "loss": 46.0,
-      "step": 576
-    },
-    {
-      "epoch": 0.09291839446032449,
-      "grad_norm": 0.0005471435142681003,
-      "learning_rate": 0.0001999999958857656,
-      "loss": 46.0,
-      "step": 577
-    },
-    {
-      "epoch": 0.09307943153911188,
-      "grad_norm": 0.0007232374045997858,
-      "learning_rate": 0.00019999999587124053,
-      "loss": 46.0,
-      "step": 578
-    },
-    {
-      "epoch": 0.09324046861789927,
-      "grad_norm": 0.0005138620617799461,
-      "learning_rate": 0.00019999999585668983,
-      "loss": 46.0,
-      "step": 579
-    },
-    {
-      "epoch": 0.09340150569668666,
-      "grad_norm": 0.0010566923301666975,
-      "learning_rate": 0.00019999999584211357,
-      "loss": 46.0,
-      "step": 580
-    },
-    {
-      "epoch": 0.09356254277547406,
-      "grad_norm": 0.0005584961618296802,
-      "learning_rate": 0.0001999999958275117,
-      "loss": 46.0,
-      "step": 581
-    },
-    {
-      "epoch": 0.09372357985426144,
-      "grad_norm": 0.00048247986705973744,
-      "learning_rate": 0.0001999999958128842,
-      "loss": 46.0,
-      "step": 582
-    },
-    {
-      "epoch": 0.09388461693304884,
-      "grad_norm": 0.0009872997179627419,
-      "learning_rate": 0.00019999999579823117,
-      "loss": 46.0,
-      "step": 583
-    },
-    {
-      "epoch": 0.09404565401183622,
-      "grad_norm": 0.0004425085207913071,
-      "learning_rate": 0.0001999999957835525,
-      "loss": 46.0,
-      "step": 584
-    },
-    {
-      "epoch": 0.09420669109062361,
-      "grad_norm": 0.00032448922866024077,
-      "learning_rate": 0.00019999999576884826,
-      "loss": 46.0,
-      "step": 585
-    },
-    {
-      "epoch": 0.09436772816941101,
-      "grad_norm": 0.0007217460079118609,
-      "learning_rate": 0.0001999999957541184,
-      "loss": 46.0,
-      "step": 586
-    },
-    {
-      "epoch": 0.0945287652481984,
-      "grad_norm": 0.0011977762915194035,
-      "learning_rate": 0.00019999999573936297,
-      "loss": 46.0,
-      "step": 587
-    },
-    {
-      "epoch": 0.09468980232698579,
-      "grad_norm": 0.0009963751072064042,
-      "learning_rate": 0.00019999999572458194,
-      "loss": 46.0,
-      "step": 588
-    },
-    {
-      "epoch": 0.09485083940577319,
-      "grad_norm": 0.0005563003360293806,
-      "learning_rate": 0.0001999999957097753,
-      "loss": 46.0,
-      "step": 589
-    },
-    {
-      "epoch": 0.09501187648456057,
-      "grad_norm": 0.00044900630018673837,
-      "learning_rate": 0.00019999999569494307,
-      "loss": 46.0,
-      "step": 590
-    },
-    {
-      "epoch": 0.09517291356334796,
-      "grad_norm": 0.0005432058242149651,
-      "learning_rate": 0.00019999999568008525,
-      "loss": 46.0,
-      "step": 591
-    },
-    {
-      "epoch": 0.09533395064213535,
-      "grad_norm": 0.0005411191377788782,
-      "learning_rate": 0.00019999999566520184,
-      "loss": 46.0,
-      "step": 592
-    },
-    {
-      "epoch": 0.09549498772092274,
-      "grad_norm": 0.0011375902686268091,
-      "learning_rate": 0.0001999999956502928,
-      "loss": 46.0,
-      "step": 593
-    },
-    {
-      "epoch": 0.09565602479971014,
-      "grad_norm": 0.0010755470721051097,
-      "learning_rate": 0.0001999999956353582,
-      "loss": 46.0,
-      "step": 594
-    },
-    {
-      "epoch": 0.09581706187849752,
-      "grad_norm": 0.0007898997864685953,
-      "learning_rate": 0.00019999999562039802,
-      "loss": 46.0,
-      "step": 595
-    },
-    {
-      "epoch": 0.09597809895728492,
-      "grad_norm": 0.0004766711499541998,
-      "learning_rate": 0.0001999999956054122,
-      "loss": 46.0,
-      "step": 596
-    },
-    {
-      "epoch": 0.09613913603607231,
-      "grad_norm": 0.000768249505199492,
-      "learning_rate": 0.00019999999559040083,
-      "loss": 46.0,
-      "step": 597
-    },
-    {
-      "epoch": 0.0963001731148597,
-      "grad_norm": 0.0007913981680758297,
-      "learning_rate": 0.00019999999557536384,
-      "loss": 46.0,
-      "step": 598
-    },
-    {
-      "epoch": 0.09646121019364709,
-      "grad_norm": 0.0006753549678251147,
-      "learning_rate": 0.00019999999556030127,
-      "loss": 46.0,
-      "step": 599
-    },
-    {
-      "epoch": 0.09662224727243447,
-      "grad_norm": 0.0004898831248283386,
-      "learning_rate": 0.00019999999554521307,
-      "loss": 46.0,
-      "step": 600
-    },
-    {
-      "epoch": 0.09678328435122187,
-      "grad_norm": 0.0013990583829581738,
-      "learning_rate": 0.00019999999553009932,
-      "loss": 46.0,
-      "step": 601
-    },
-    {
-      "epoch": 0.09694432143000926,
-      "grad_norm": 0.0006502924952656031,
-      "learning_rate": 0.00019999999551495993,
-      "loss": 46.0,
-      "step": 602
-    },
-    {
-      "epoch": 0.09710535850879665,
-      "grad_norm": 0.0009294247720390558,
-      "learning_rate": 0.000199999995499795,
-      "loss": 46.0,
-      "step": 603
-    },
-    {
-      "epoch": 0.09726639558758404,
-      "grad_norm": 0.0005790565046481788,
-      "learning_rate": 0.00019999999548460444,
-      "loss": 46.0,
-      "step": 604
-    },
-    {
-      "epoch": 0.09742743266637144,
-      "grad_norm": 0.0005814313190057874,
-      "learning_rate": 0.0001999999954693883,
-      "loss": 46.0,
-      "step": 605
-    },
-    {
-      "epoch": 0.09758846974515882,
-      "grad_norm": 0.0006459489814005792,
-      "learning_rate": 0.00019999999545414654,
-      "loss": 46.0,
-      "step": 606
-    },
-    {
-      "epoch": 0.09774950682394622,
-      "grad_norm": 0.0006937396829016507,
-      "learning_rate": 0.0001999999954388792,
-      "loss": 46.0,
-      "step": 607
-    },
-    {
-      "epoch": 0.0979105439027336,
-      "grad_norm": 0.0005056326044723392,
-      "learning_rate": 0.00019999999542358627,
-      "loss": 46.0,
-      "step": 608
-    },
-    {
-      "epoch": 0.098071580981521,
-      "grad_norm": 0.00040239939698949456,
-      "learning_rate": 0.00019999999540826771,
-      "loss": 46.0,
-      "step": 609
-    },
-    {
-      "epoch": 0.09823261806030839,
-      "grad_norm": 0.0016964153619483113,
-      "learning_rate": 0.0001999999953929236,
-      "loss": 46.0,
-      "step": 610
-    },
-    {
-      "epoch": 0.09839365513909577,
-      "grad_norm": 0.000712358858436346,
-      "learning_rate": 0.0001999999953775539,
-      "loss": 46.0,
-      "step": 611
-    },
-    {
-      "epoch": 0.09855469221788317,
-      "grad_norm": 0.0005845819250680506,
-      "learning_rate": 0.00019999999536215857,
-      "loss": 46.0,
-      "step": 612
-    },
-    {
-      "epoch": 0.09871572929667057,
-      "grad_norm": 0.0007552969036623836,
-      "learning_rate": 0.00019999999534673767,
-      "loss": 46.0,
-      "step": 613
-    },
-    {
-      "epoch": 0.09887676637545795,
-      "grad_norm": 0.0006136150914244354,
-      "learning_rate": 0.00019999999533129118,
-      "loss": 46.0,
-      "step": 614
-    },
-    {
-      "epoch": 0.09903780345424534,
-      "grad_norm": 0.0009305122075602412,
-      "learning_rate": 0.00019999999531581907,
-      "loss": 46.0,
-      "step": 615
-    },
-    {
-      "epoch": 0.09919884053303273,
-      "grad_norm": 0.0005232894327491522,
-      "learning_rate": 0.00019999999530032138,
-      "loss": 46.0,
-      "step": 616
-    },
-    {
-      "epoch": 0.09935987761182012,
-      "grad_norm": 0.0003367464814800769,
-      "learning_rate": 0.0001999999952847981,
-      "loss": 46.0,
-      "step": 617
-    },
-    {
-      "epoch": 0.09952091469060752,
-      "grad_norm": 0.00024847566965036094,
-      "learning_rate": 0.0001999999952692492,
-      "loss": 46.0,
-      "step": 618
-    },
-    {
-      "epoch": 0.0996819517693949,
-      "grad_norm": 0.0008768857805989683,
-      "learning_rate": 0.00019999999525367471,
-      "loss": 46.0,
-      "step": 619
-    },
-    {
-      "epoch": 0.0998429888481823,
-      "grad_norm": 0.0006605791859328747,
-      "learning_rate": 0.00019999999523807467,
-      "loss": 46.0,
-      "step": 620
-    },
-    {
-      "epoch": 0.10000402592696968,
-      "grad_norm": 0.0005555636016651988,
-      "learning_rate": 0.000199999995222449,
-      "loss": 46.0,
-      "step": 621
-    },
-    {
-      "epoch": 0.10016506300575707,
-      "grad_norm": 0.0007445701630786061,
-      "learning_rate": 0.00019999999520679774,
-      "loss": 46.0,
-      "step": 622
-    },
-    {
-      "epoch": 0.10032610008454447,
-      "grad_norm": 0.0008318775217048824,
-      "learning_rate": 0.00019999999519112086,
-      "loss": 46.0,
-      "step": 623
-    },
-    {
-      "epoch": 0.10048713716333185,
-      "grad_norm": 0.0015548253431916237,
-      "learning_rate": 0.00019999999517541844,
-      "loss": 46.0,
-      "step": 624
-    },
-    {
-      "epoch": 0.10064817424211925,
-      "grad_norm": 0.0005522758001461625,
-      "learning_rate": 0.00019999999515969038,
-      "loss": 46.0,
-      "step": 625
-    },
-    {
-      "epoch": 0.10080921132090664,
-      "grad_norm": 0.0005080297123640776,
-      "learning_rate": 0.00019999999514393673,
-      "loss": 46.0,
-      "step": 626
-    },
-    {
-      "epoch": 0.10097024839969403,
-      "grad_norm": 0.0008117557154037058,
-      "learning_rate": 0.0001999999951281575,
-      "loss": 46.0,
-      "step": 627
-    },
-    {
-      "epoch": 0.10113128547848142,
-      "grad_norm": 0.0007852627313695848,
-      "learning_rate": 0.00019999999511235265,
-      "loss": 46.0,
-      "step": 628
-    },
-    {
-      "epoch": 0.1012923225572688,
-      "grad_norm": 0.0006700888625346124,
-      "learning_rate": 0.00019999999509652224,
-      "loss": 46.0,
-      "step": 629
-    },
-    {
-      "epoch": 0.1014533596360562,
-      "grad_norm": 0.0005911216721870005,
-      "learning_rate": 0.00019999999508066622,
-      "loss": 46.0,
-      "step": 630
-    },
-    {
-      "epoch": 0.1016143967148436,
-      "grad_norm": 0.0007278225966729224,
-      "learning_rate": 0.0001999999950647846,
-      "loss": 46.0,
-      "step": 631
-    },
-    {
-      "epoch": 0.10177543379363098,
-      "grad_norm": 0.00048417862853966653,
-      "learning_rate": 0.00019999999504887738,
-      "loss": 46.0,
-      "step": 632
-    },
-    {
-      "epoch": 0.10193647087241837,
-      "grad_norm": 0.0011535398662090302,
-      "learning_rate": 0.00019999999503294457,
-      "loss": 46.0,
-      "step": 633
-    },
-    {
-      "epoch": 0.10209750795120577,
-      "grad_norm": 0.0007467381074093282,
-      "learning_rate": 0.0001999999950169862,
-      "loss": 46.0,
-      "step": 634
-    },
-    {
-      "epoch": 0.10225854502999315,
-      "grad_norm": 0.0016067279502749443,
-      "learning_rate": 0.00019999999500100218,
-      "loss": 46.0,
-      "step": 635
-    },
-    {
-      "epoch": 0.10241958210878055,
-      "grad_norm": 0.0008178745047189295,
-      "learning_rate": 0.0001999999949849926,
-      "loss": 46.0,
-      "step": 636
-    },
-    {
-      "epoch": 0.10258061918756793,
-      "grad_norm": 0.0019085468957200646,
-      "learning_rate": 0.0001999999949689574,
-      "loss": 46.0,
-      "step": 637
-    },
-    {
-      "epoch": 0.10274165626635533,
-      "grad_norm": 0.0012237601913511753,
-      "learning_rate": 0.00019999999495289662,
-      "loss": 46.0,
-      "step": 638
-    },
-    {
-      "epoch": 0.10290269334514272,
-      "grad_norm": 0.0004765917547047138,
-      "learning_rate": 0.0001999999949368103,
-      "loss": 46.0,
-      "step": 639
-    },
-    {
-      "epoch": 0.1030637304239301,
-      "grad_norm": 0.0006971553666517138,
-      "learning_rate": 0.00019999999492069828,
-      "loss": 46.0,
-      "step": 640
-    },
-    {
-      "epoch": 0.1032247675027175,
-      "grad_norm": 0.0005272507551126182,
-      "learning_rate": 0.00019999999490456075,
-      "loss": 46.0,
-      "step": 641
-    },
-    {
-      "epoch": 0.1033858045815049,
-      "grad_norm": 0.000627867178991437,
-      "learning_rate": 0.00019999999488839757,
-      "loss": 46.0,
-      "step": 642
-    },
-    {
-      "epoch": 0.10354684166029228,
-      "grad_norm": 0.000540842127520591,
-      "learning_rate": 0.0001999999948722088,
-      "loss": 46.0,
-      "step": 643
-    },
-    {
-      "epoch": 0.10370787873907968,
-      "grad_norm": 0.0008780147763900459,
-      "learning_rate": 0.00019999999485599448,
-      "loss": 46.0,
-      "step": 644
-    },
-    {
-      "epoch": 0.10386891581786706,
-      "grad_norm": 0.0004521933733485639,
-      "learning_rate": 0.00019999999483975454,
-      "loss": 46.0,
-      "step": 645
-    },
-    {
-      "epoch": 0.10402995289665445,
-      "grad_norm": 0.0009476662380620837,
-      "learning_rate": 0.00019999999482348896,
-      "loss": 46.0,
-      "step": 646
-    },
-    {
-      "epoch": 0.10419098997544185,
-      "grad_norm": 0.000842555717099458,
-      "learning_rate": 0.00019999999480719785,
-      "loss": 46.0,
-      "step": 647
-    },
-    {
-      "epoch": 0.10435202705422923,
-      "grad_norm": 0.0008570015779696405,
-      "learning_rate": 0.00019999999479088112,
-      "loss": 46.0,
-      "step": 648
-    },
-    {
-      "epoch": 0.10451306413301663,
-      "grad_norm": 0.0007610553875565529,
-      "learning_rate": 0.00019999999477453878,
-      "loss": 46.0,
-      "step": 649
-    },
-    {
-      "epoch": 0.10467410121180402,
-      "grad_norm": 0.0007040970376692712,
-      "learning_rate": 0.00019999999475817085,
-      "loss": 46.0,
-      "step": 650
-    },
-    {
-      "epoch": 0.1048351382905914,
-      "grad_norm": 0.0014141713036224246,
-      "learning_rate": 0.00019999999474177736,
-      "loss": 46.0,
-      "step": 651
-    },
-    {
-      "epoch": 0.1049961753693788,
-      "grad_norm": 0.0005423424299806356,
-      "learning_rate": 0.00019999999472535826,
-      "loss": 46.0,
-      "step": 652
-    },
-    {
-      "epoch": 0.10515721244816618,
-      "grad_norm": 0.0005851828027516603,
-      "learning_rate": 0.00019999999470891354,
-      "loss": 46.0,
-      "step": 653
-    },
-    {
-      "epoch": 0.10531824952695358,
-      "grad_norm": 0.0004984669503755867,
-      "learning_rate": 0.00019999999469244326,
-      "loss": 46.0,
-      "step": 654
-    },
-    {
-      "epoch": 0.10547928660574098,
-      "grad_norm": 0.0010419270256534219,
-      "learning_rate": 0.00019999999467594734,
-      "loss": 46.0,
-      "step": 655
-    },
-    {
-      "epoch": 0.10564032368452836,
-      "grad_norm": 0.0008180539007298648,
-      "learning_rate": 0.00019999999465942586,
-      "loss": 46.0,
-      "step": 656
-    },
-    {
-      "epoch": 0.10580136076331575,
-      "grad_norm": 0.001139450934715569,
-      "learning_rate": 0.00019999999464287877,
-      "loss": 46.0,
-      "step": 657
-    },
-    {
-      "epoch": 0.10596239784210315,
-      "grad_norm": 0.0004230669583193958,
-      "learning_rate": 0.00019999999462630612,
-      "loss": 46.0,
-      "step": 658
-    },
-    {
-      "epoch": 0.10612343492089053,
-      "grad_norm": 0.0006402652943506837,
-      "learning_rate": 0.00019999999460970785,
-      "loss": 46.0,
-      "step": 659
-    },
-    {
-      "epoch": 0.10628447199967793,
-      "grad_norm": 0.0006499659502878785,
-      "learning_rate": 0.000199999994593084,
-      "loss": 46.0,
-      "step": 660
-    },
-    {
-      "epoch": 0.10644550907846531,
-      "grad_norm": 0.0007256578537635505,
-      "learning_rate": 0.0001999999945764345,
-      "loss": 46.0,
-      "step": 661
-    },
-    {
-      "epoch": 0.10660654615725271,
-      "grad_norm": 0.0008143671439029276,
-      "learning_rate": 0.00019999999455975944,
-      "loss": 46.0,
-      "step": 662
-    },
-    {
-      "epoch": 0.1067675832360401,
-      "grad_norm": 0.0008687827503308654,
-      "learning_rate": 0.0001999999945430588,
-      "loss": 46.0,
-      "step": 663
-    },
-    {
-      "epoch": 0.10692862031482749,
-      "grad_norm": 0.0007232081843540072,
-      "learning_rate": 0.00019999999452633256,
-      "loss": 46.0,
-      "step": 664
-    },
-    {
-      "epoch": 0.10708965739361488,
-      "grad_norm": 0.0009402253199368715,
-      "learning_rate": 0.00019999999450958072,
-      "loss": 46.0,
-      "step": 665
-    },
-    {
-      "epoch": 0.10725069447240226,
-      "grad_norm": 0.0006775706424377859,
-      "learning_rate": 0.00019999999449280326,
-      "loss": 46.0,
-      "step": 666
-    },
-    {
-      "epoch": 0.10741173155118966,
-      "grad_norm": 0.00037028861697763205,
-      "learning_rate": 0.00019999999447600024,
-      "loss": 46.0,
-      "step": 667
-    },
-    {
-      "epoch": 0.10757276862997706,
-      "grad_norm": 0.0005190723459236324,
-      "learning_rate": 0.00019999999445917163,
-      "loss": 46.0,
-      "step": 668
-    },
-    {
-      "epoch": 0.10773380570876444,
-      "grad_norm": 0.000885549234226346,
-      "learning_rate": 0.00019999999444231739,
-      "loss": 46.0,
-      "step": 669
-    },
-    {
-      "epoch": 0.10789484278755183,
-      "grad_norm": 0.0013564539840444922,
-      "learning_rate": 0.0001999999944254376,
-      "loss": 46.0,
-      "step": 670
-    },
-    {
-      "epoch": 0.10805587986633923,
-      "grad_norm": 0.00039320086943916976,
-      "learning_rate": 0.00019999999440853218,
-      "loss": 46.0,
-      "step": 671
-    },
-    {
-      "epoch": 0.10821691694512661,
-      "grad_norm": 0.0006643789820373058,
-      "learning_rate": 0.00019999999439160115,
-      "loss": 46.0,
-      "step": 672
-    },
-    {
-      "epoch": 0.10837795402391401,
-      "grad_norm": 0.0006942615727894008,
-      "learning_rate": 0.00019999999437464455,
-      "loss": 46.0,
-      "step": 673
-    },
-    {
-      "epoch": 0.10853899110270139,
-      "grad_norm": 0.001256827381439507,
-      "learning_rate": 0.00019999999435766237,
-      "loss": 46.0,
-      "step": 674
-    },
-    {
-      "epoch": 0.10870002818148879,
-      "grad_norm": 0.0006793045904487371,
-      "learning_rate": 0.00019999999434065457,
-      "loss": 46.0,
-      "step": 675
-    },
-    {
-      "epoch": 0.10886106526027618,
-      "grad_norm": 0.0003982323396485299,
-      "learning_rate": 0.0001999999943236212,
-      "loss": 46.0,
-      "step": 676
-    },
-    {
-      "epoch": 0.10902210233906356,
-      "grad_norm": 0.0004949101130478084,
-      "learning_rate": 0.0001999999943065622,
-      "loss": 46.0,
-      "step": 677
-    },
-    {
-      "epoch": 0.10918313941785096,
-      "grad_norm": 0.00018608546815812588,
-      "learning_rate": 0.00019999999428947765,
-      "loss": 46.0,
-      "step": 678
-    },
-    {
-      "epoch": 0.10934417649663836,
-      "grad_norm": 0.000787493190728128,
-      "learning_rate": 0.00019999999427236748,
-      "loss": 46.0,
-      "step": 679
-    },
-    {
-      "epoch": 0.10950521357542574,
-      "grad_norm": 0.0006306565483100712,
-      "learning_rate": 0.00019999999425523171,
-      "loss": 46.0,
-      "step": 680
-    },
-    {
-      "epoch": 0.10966625065421313,
-      "grad_norm": 0.000702961056958884,
-      "learning_rate": 0.00019999999423807034,
-      "loss": 46.0,
-      "step": 681
-    },
-    {
-      "epoch": 0.10982728773300052,
-      "grad_norm": 0.0007224611472338438,
-      "learning_rate": 0.0001999999942208834,
-      "loss": 46.0,
-      "step": 682
-    },
-    {
-      "epoch": 0.10998832481178791,
-      "grad_norm": 0.0010548344580456614,
-      "learning_rate": 0.00019999999420367085,
-      "loss": 46.0,
-      "step": 683
-    },
-    {
-      "epoch": 0.11014936189057531,
-      "grad_norm": 0.00038637136458419263,
-      "learning_rate": 0.0001999999941864327,
-      "loss": 46.0,
-      "step": 684
-    },
-    {
-      "epoch": 0.11031039896936269,
-      "grad_norm": 0.0010405262000858784,
-      "learning_rate": 0.00019999999416916894,
-      "loss": 46.0,
-      "step": 685
-    },
-    {
-      "epoch": 0.11047143604815009,
-      "grad_norm": 0.000544868700671941,
-      "learning_rate": 0.00019999999415187963,
-      "loss": 46.0,
-      "step": 686
-    },
-    {
-      "epoch": 0.11063247312693748,
-      "grad_norm": 0.0003214654861949384,
-      "learning_rate": 0.0001999999941345647,
-      "loss": 46.0,
-      "step": 687
-    },
-    {
-      "epoch": 0.11079351020572487,
-      "grad_norm": 0.0006614724989049137,
-      "learning_rate": 0.0001999999941172242,
-      "loss": 46.0,
-      "step": 688
-    },
-    {
-      "epoch": 0.11095454728451226,
-      "grad_norm": 0.0009841398568823934,
-      "learning_rate": 0.00019999999409985806,
-      "loss": 46.0,
-      "step": 689
-    },
-    {
-      "epoch": 0.11111558436329964,
-      "grad_norm": 0.0010768999345600605,
-      "learning_rate": 0.00019999999408246635,
-      "loss": 46.0,
-      "step": 690
-    },
-    {
-      "epoch": 0.11127662144208704,
-      "grad_norm": 0.0007441352354362607,
-      "learning_rate": 0.00019999999406504905,
-      "loss": 46.0,
-      "step": 691
-    },
-    {
-      "epoch": 0.11143765852087444,
-      "grad_norm": 0.0003263895632699132,
-      "learning_rate": 0.00019999999404760613,
-      "loss": 46.0,
-      "step": 692
-    },
-    {
-      "epoch": 0.11159869559966182,
-      "grad_norm": 0.0004391400143504143,
-      "learning_rate": 0.00019999999403013763,
-      "loss": 46.0,
-      "step": 693
-    },
-    {
-      "epoch": 0.11175973267844921,
-      "grad_norm": 0.00041094841435551643,
-      "learning_rate": 0.00019999999401264354,
-      "loss": 46.0,
-      "step": 694
-    },
-    {
-      "epoch": 0.11192076975723661,
-      "grad_norm": 0.0005844959523528814,
-      "learning_rate": 0.00019999999399512386,
-      "loss": 46.0,
-      "step": 695
-    },
-    {
-      "epoch": 0.11208180683602399,
-      "grad_norm": 0.0009718054789118469,
-      "learning_rate": 0.00019999999397757857,
-      "loss": 46.0,
-      "step": 696
-    },
-    {
-      "epoch": 0.11224284391481139,
-      "grad_norm": 0.00024566747015342116,
-      "learning_rate": 0.0001999999939600077,
-      "loss": 46.0,
-      "step": 697
-    },
-    {
-      "epoch": 0.11240388099359877,
-      "grad_norm": 0.0004258676781319082,
-      "learning_rate": 0.00019999999394241123,
-      "loss": 46.0,
-      "step": 698
-    },
-    {
-      "epoch": 0.11256491807238617,
-      "grad_norm": 0.0005756361060775816,
-      "learning_rate": 0.00019999999392478915,
-      "loss": 46.0,
-      "step": 699
-    },
-    {
-      "epoch": 0.11272595515117356,
-      "grad_norm": 0.0007779954466968775,
-      "learning_rate": 0.00019999999390714148,
-      "loss": 46.0,
-      "step": 700
-    },
-    {
-      "epoch": 0.11288699222996094,
-      "grad_norm": 0.00047630540211685,
-      "learning_rate": 0.00019999999388946825,
-      "loss": 46.0,
-      "step": 701
-    },
-    {
-      "epoch": 0.11304802930874834,
-      "grad_norm": 0.0009862382430583239,
-      "learning_rate": 0.00019999999387176938,
-      "loss": 46.0,
-      "step": 702
-    },
-    {
-      "epoch": 0.11320906638753574,
-      "grad_norm": 0.00047935088514350355,
-      "learning_rate": 0.00019999999385404493,
-      "loss": 46.0,
-      "step": 703
-    },
-    {
-      "epoch": 0.11337010346632312,
-      "grad_norm": 0.0006019410793669522,
-      "learning_rate": 0.0001999999938362949,
-      "loss": 46.0,
-      "step": 704
-    },
-    {
-      "epoch": 0.11353114054511051,
-      "grad_norm": 0.0003566310042515397,
-      "learning_rate": 0.00019999999381851925,
-      "loss": 46.0,
-      "step": 705
-    },
-    {
-      "epoch": 0.1136921776238979,
-      "grad_norm": 0.0008112428477033973,
-      "learning_rate": 0.00019999999380071804,
-      "loss": 46.0,
-      "step": 706
-    },
-    {
-      "epoch": 0.1138532147026853,
-      "grad_norm": 0.0008270882535725832,
-      "learning_rate": 0.0001999999937828912,
-      "loss": 46.0,
-      "step": 707
-    },
-    {
-      "epoch": 0.11401425178147269,
-      "grad_norm": 0.0005021857796236873,
-      "learning_rate": 0.00019999999376503879,
-      "loss": 46.0,
-      "step": 708
-    },
-    {
-      "epoch": 0.11417528886026007,
-      "grad_norm": 0.0010833428241312504,
-      "learning_rate": 0.00019999999374716078,
-      "loss": 46.0,
-      "step": 709
-    },
-    {
-      "epoch": 0.11433632593904747,
-      "grad_norm": 0.0007432840066030622,
-      "learning_rate": 0.00019999999372925716,
-      "loss": 46.0,
-      "step": 710
-    },
-    {
-      "epoch": 0.11449736301783486,
-      "grad_norm": 0.0003462954191491008,
-      "learning_rate": 0.00019999999371132793,
-      "loss": 46.0,
-      "step": 711
-    },
-    {
-      "epoch": 0.11465840009662225,
-      "grad_norm": 0.0006248669815249741,
-      "learning_rate": 0.00019999999369337313,
-      "loss": 46.0,
-      "step": 712
-    },
-    {
-      "epoch": 0.11481943717540964,
-      "grad_norm": 0.0008630806696601212,
-      "learning_rate": 0.00019999999367539278,
-      "loss": 46.0,
-      "step": 713
-    },
-    {
-      "epoch": 0.11498047425419702,
-      "grad_norm": 0.0026565478183329105,
-      "learning_rate": 0.00019999999365738678,
-      "loss": 46.0,
-      "step": 714
-    },
-    {
-      "epoch": 0.11514151133298442,
-      "grad_norm": 0.0003844198363367468,
-      "learning_rate": 0.0001999999936393552,
-      "loss": 46.0,
-      "step": 715
-    },
-    {
-      "epoch": 0.11530254841177182,
-      "grad_norm": 0.0009668955462984741,
-      "learning_rate": 0.000199999993621298,
-      "loss": 46.0,
-      "step": 716
-    },
-    {
-      "epoch": 0.1154635854905592,
-      "grad_norm": 0.0006512926192954183,
-      "learning_rate": 0.00019999999360321522,
-      "loss": 46.0,
-      "step": 717
-    },
-    {
-      "epoch": 0.1156246225693466,
-      "grad_norm": 0.0004899580380879343,
-      "learning_rate": 0.00019999999358510687,
-      "loss": 46.0,
-      "step": 718
-    },
-    {
-      "epoch": 0.11578565964813398,
-      "grad_norm": 0.0007209365139715374,
-      "learning_rate": 0.0001999999935669729,
-      "loss": 46.0,
-      "step": 719
-    },
-    {
-      "epoch": 0.11594669672692137,
-      "grad_norm": 0.0008001966052688658,
-      "learning_rate": 0.00019999999354881334,
-      "loss": 46.0,
-      "step": 720
-    },
-    {
-      "epoch": 0.11610773380570877,
-      "grad_norm": 0.00044169294415041804,
-      "learning_rate": 0.00019999999353062818,
-      "loss": 46.0,
-      "step": 721
-    },
-    {
-      "epoch": 0.11626877088449615,
-      "grad_norm": 0.00034914870047941804,
-      "learning_rate": 0.00019999999351241746,
-      "loss": 46.0,
-      "step": 722
-    },
-    {
-      "epoch": 0.11642980796328355,
-      "grad_norm": 0.0007880452321842313,
-      "learning_rate": 0.0001999999934941811,
-      "loss": 46.0,
-      "step": 723
-    },
-    {
-      "epoch": 0.11659084504207094,
-      "grad_norm": 0.0006567139644175768,
-      "learning_rate": 0.00019999999347591915,
-      "loss": 46.0,
-      "step": 724
-    },
-    {
-      "epoch": 0.11675188212085832,
-      "grad_norm": 0.0012865960597991943,
-      "learning_rate": 0.00019999999345763164,
-      "loss": 46.0,
-      "step": 725
-    },
-    {
-      "epoch": 0.11691291919964572,
-      "grad_norm": 0.00031495318398810923,
-      "learning_rate": 0.00019999999343931851,
-      "loss": 46.0,
-      "step": 726
-    },
-    {
-      "epoch": 0.1170739562784331,
-      "grad_norm": 0.0009946618229150772,
-      "learning_rate": 0.00019999999342097978,
-      "loss": 46.0,
-      "step": 727
-    },
-    {
-      "epoch": 0.1172349933572205,
-      "grad_norm": 0.0006088796071708202,
-      "learning_rate": 0.00019999999340261545,
-      "loss": 46.0,
-      "step": 728
-    },
-    {
-      "epoch": 0.1173960304360079,
-      "grad_norm": 0.0006799606489948928,
-      "learning_rate": 0.00019999999338422554,
-      "loss": 46.0,
-      "step": 729
-    },
-    {
-      "epoch": 0.11755706751479528,
-      "grad_norm": 0.0011438957881182432,
-      "learning_rate": 0.00019999999336581,
-      "loss": 46.0,
-      "step": 730
-    },
-    {
-      "epoch": 0.11771810459358267,
-      "grad_norm": 0.0005068652098998427,
-      "learning_rate": 0.00019999999334736893,
-      "loss": 46.0,
-      "step": 731
-    },
-    {
-      "epoch": 0.11787914167237007,
-      "grad_norm": 0.0005817121127620339,
-      "learning_rate": 0.00019999999332890223,
-      "loss": 46.0,
-      "step": 732
-    },
-    {
-      "epoch": 0.11804017875115745,
-      "grad_norm": 0.0015141000039875507,
-      "learning_rate": 0.00019999999331040994,
-      "loss": 46.0,
-      "step": 733
-    },
-    {
-      "epoch": 0.11820121582994485,
-      "grad_norm": 0.0008203965262509882,
-      "learning_rate": 0.00019999999329189206,
-      "loss": 46.0,
-      "step": 734
-    },
-    {
-      "epoch": 0.11836225290873223,
-      "grad_norm": 0.0004771222302224487,
-      "learning_rate": 0.00019999999327334855,
-      "loss": 46.0,
-      "step": 735
-    },
-    {
-      "epoch": 0.11852328998751963,
-      "grad_norm": 0.0003979514294769615,
-      "learning_rate": 0.0001999999932547795,
-      "loss": 46.0,
-      "step": 736
-    },
-    {
-      "epoch": 0.11868432706630702,
-      "grad_norm": 0.0005276133888401091,
-      "learning_rate": 0.0001999999932361848,
-      "loss": 46.0,
-      "step": 737
-    },
-    {
-      "epoch": 0.1188453641450944,
-      "grad_norm": 0.0005074460059404373,
-      "learning_rate": 0.00019999999321756456,
-      "loss": 46.0,
-      "step": 738
-    },
-    {
-      "epoch": 0.1190064012238818,
-      "grad_norm": 0.0010084678651764989,
-      "learning_rate": 0.0001999999931989187,
-      "loss": 46.0,
-      "step": 739
-    },
-    {
-      "epoch": 0.1191674383026692,
-      "grad_norm": 0.0007157850777730346,
-      "learning_rate": 0.0001999999931802472,
-      "loss": 46.0,
-      "step": 740
-    },
-    {
-      "epoch": 0.11932847538145658,
-      "grad_norm": 0.00042946788016706705,
-      "learning_rate": 0.00019999999316155017,
-      "loss": 46.0,
-      "step": 741
-    },
-    {
-      "epoch": 0.11948951246024397,
-      "grad_norm": 0.00038639921694993973,
-      "learning_rate": 0.00019999999314282755,
-      "loss": 46.0,
-      "step": 742
-    },
-    {
-      "epoch": 0.11965054953903136,
-      "grad_norm": 0.000835128768812865,
-      "learning_rate": 0.00019999999312407928,
-      "loss": 46.0,
-      "step": 743
-    },
-    {
-      "epoch": 0.11981158661781875,
-      "grad_norm": 0.000891783507540822,
-      "learning_rate": 0.00019999999310530545,
-      "loss": 46.0,
-      "step": 744
-    },
-    {
-      "epoch": 0.11997262369660615,
-      "grad_norm": 0.0004395773576106876,
-      "learning_rate": 0.000199999993086506,
-      "loss": 46.0,
-      "step": 745
-    },
-    {
-      "epoch": 0.12013366077539353,
-      "grad_norm": 0.0005717150052078068,
-      "learning_rate": 0.000199999993067681,
-      "loss": 46.0,
-      "step": 746
-    },
-    {
-      "epoch": 0.12029469785418093,
-      "grad_norm": 0.0004644012951757759,
-      "learning_rate": 0.00019999999304883036,
-      "loss": 46.0,
-      "step": 747
-    },
-    {
-      "epoch": 0.12045573493296832,
-      "grad_norm": 0.0005582737503573298,
-      "learning_rate": 0.00019999999302995416,
-      "loss": 46.0,
-      "step": 748
-    },
-    {
-      "epoch": 0.1206167720117557,
-      "grad_norm": 0.00038077603676356375,
-      "learning_rate": 0.00019999999301105237,
-      "loss": 46.0,
-      "step": 749
-    },
-    {
-      "epoch": 0.1207778090905431,
-      "grad_norm": 0.0012594073778018355,
-      "learning_rate": 0.00019999999299212496,
-      "loss": 46.0,
-      "step": 750
-    },
-    {
-      "epoch": 0.12093884616933048,
-      "grad_norm": 0.0008036632789298892,
-      "learning_rate": 0.00019999999297317197,
-      "loss": 46.0,
-      "step": 751
-    },
-    {
-      "epoch": 0.12109988324811788,
-      "grad_norm": 0.0005055126966908574,
-      "learning_rate": 0.00019999999295419336,
-      "loss": 46.0,
-      "step": 752
-    },
-    {
-      "epoch": 0.12126092032690527,
-      "grad_norm": 0.0005579818389378488,
-      "learning_rate": 0.00019999999293518917,
-      "loss": 46.0,
-      "step": 753
-    },
-    {
-      "epoch": 0.12142195740569266,
-      "grad_norm": 0.0011566895991563797,
-      "learning_rate": 0.0001999999929161594,
-      "loss": 46.0,
-      "step": 754
-    },
-    {
-      "epoch": 0.12158299448448005,
-      "grad_norm": 0.0005442249239422381,
-      "learning_rate": 0.000199999992897104,
-      "loss": 46.0,
-      "step": 755
-    },
-    {
-      "epoch": 0.12174403156326745,
-      "grad_norm": 0.0005158275598660111,
-      "learning_rate": 0.000199999992878023,
-      "loss": 46.0,
-      "step": 756
-    },
-    {
-      "epoch": 0.12190506864205483,
-      "grad_norm": 0.0016388717340305448,
-      "learning_rate": 0.00019999999285891647,
-      "loss": 46.0,
-      "step": 757
-    },
-    {
-      "epoch": 0.12206610572084223,
-      "grad_norm": 0.0006762741249985993,
-      "learning_rate": 0.0001999999928397843,
-      "loss": 46.0,
-      "step": 758
-    },
-    {
-      "epoch": 0.12222714279962961,
-      "grad_norm": 0.0009991052793338895,
-      "learning_rate": 0.00019999999282062654,
-      "loss": 46.0,
-      "step": 759
-    },
-    {
-      "epoch": 0.122388179878417,
-      "grad_norm": 0.001074435655027628,
-      "learning_rate": 0.00019999999280144318,
-      "loss": 46.0,
-      "step": 760
-    },
-    {
-      "epoch": 0.1225492169572044,
-      "grad_norm": 0.00041439803317189217,
-      "learning_rate": 0.00019999999278223424,
-      "loss": 46.0,
-      "step": 761
-    },
-    {
-      "epoch": 0.12271025403599178,
-      "grad_norm": 0.000713004672434181,
-      "learning_rate": 0.00019999999276299968,
-      "loss": 46.0,
-      "step": 762
-    },
-    {
-      "epoch": 0.12287129111477918,
-      "grad_norm": 0.000530066667124629,
-      "learning_rate": 0.00019999999274373956,
-      "loss": 46.0,
-      "step": 763
-    },
-    {
-      "epoch": 0.12303232819356656,
-      "grad_norm": 0.000889762828592211,
-      "learning_rate": 0.0001999999927244538,
-      "loss": 46.0,
-      "step": 764
-    },
-    {
-      "epoch": 0.12319336527235396,
-      "grad_norm": 0.00025027673109434545,
-      "learning_rate": 0.0001999999927051425,
-      "loss": 46.0,
-      "step": 765
-    },
-    {
-      "epoch": 0.12335440235114135,
-      "grad_norm": 0.0018331586616113782,
-      "learning_rate": 0.00019999999268580557,
-      "loss": 46.0,
-      "step": 766
-    },
-    {
-      "epoch": 0.12351543942992874,
-      "grad_norm": 0.0006177728064358234,
-      "learning_rate": 0.00019999999266644304,
-      "loss": 46.0,
-      "step": 767
-    },
-    {
-      "epoch": 0.12367647650871613,
-      "grad_norm": 0.0011881053214892745,
-      "learning_rate": 0.00019999999264705493,
-      "loss": 46.0,
-      "step": 768
-    },
-    {
-      "epoch": 0.12383751358750353,
-      "grad_norm": 0.001661331276409328,
-      "learning_rate": 0.0001999999926276412,
-      "loss": 46.0,
-      "step": 769
-    },
-    {
-      "epoch": 0.12399855066629091,
-      "grad_norm": 0.0010164374252781272,
-      "learning_rate": 0.00019999999260820193,
-      "loss": 46.0,
-      "step": 770
-    },
-    {
-      "epoch": 0.1241595877450783,
-      "grad_norm": 0.000998982461169362,
-      "learning_rate": 0.00019999999258873703,
-      "loss": 46.0,
-      "step": 771
-    },
-    {
-      "epoch": 0.12432062482386569,
-      "grad_norm": 0.000523116672411561,
-      "learning_rate": 0.00019999999256924654,
-      "loss": 46.0,
-      "step": 772
-    },
-    {
-      "epoch": 0.12448166190265308,
-      "grad_norm": 0.001023276592604816,
-      "learning_rate": 0.00019999999254973044,
-      "loss": 46.0,
-      "step": 773
-    },
-    {
-      "epoch": 0.12464269898144048,
-      "grad_norm": 0.002048916881904006,
-      "learning_rate": 0.00019999999253018875,
-      "loss": 46.0,
-      "step": 774
-    },
-    {
-      "epoch": 0.12480373606022786,
-      "grad_norm": 0.0005882385303266346,
-      "learning_rate": 0.00019999999251062148,
-      "loss": 46.0,
-      "step": 775
-    },
-    {
-      "epoch": 0.12496477313901526,
-      "grad_norm": 0.00048309596604667604,
-      "learning_rate": 0.00019999999249102862,
-      "loss": 46.0,
-      "step": 776
-    },
-    {
-      "epoch": 0.12512581021780264,
-      "grad_norm": 0.0005158385611139238,
-      "learning_rate": 0.00019999999247141014,
-      "loss": 46.0,
-      "step": 777
-    },
-    {
-      "epoch": 0.12528684729659004,
-      "grad_norm": 0.0012636081082746387,
-      "learning_rate": 0.00019999999245176608,
-      "loss": 46.0,
-      "step": 778
-    },
-    {
-      "epoch": 0.12544788437537743,
-      "grad_norm": 0.000460075621958822,
-      "learning_rate": 0.00019999999243209643,
-      "loss": 46.0,
-      "step": 779
-    },
-    {
-      "epoch": 0.12560892145416483,
-      "grad_norm": 0.0013009477406740189,
-      "learning_rate": 0.0001999999924124012,
-      "loss": 46.0,
-      "step": 780
-    },
-    {
-      "epoch": 0.12576995853295223,
-      "grad_norm": 0.0004484928213059902,
-      "learning_rate": 0.00019999999239268034,
-      "loss": 46.0,
-      "step": 781
-    },
-    {
-      "epoch": 0.1259309956117396,
-      "grad_norm": 0.0005908760358579457,
-      "learning_rate": 0.00019999999237293388,
-      "loss": 46.0,
-      "step": 782
-    },
-    {
-      "epoch": 0.126092032690527,
-      "grad_norm": 0.0013063220540061593,
-      "learning_rate": 0.00019999999235316188,
-      "loss": 46.0,
-      "step": 783
-    },
-    {
-      "epoch": 0.12625306976931439,
-      "grad_norm": 0.0008520261035300791,
-      "learning_rate": 0.00019999999233336424,
-      "loss": 46.0,
-      "step": 784
-    },
-    {
-      "epoch": 0.12641410684810178,
-      "grad_norm": 0.0007049093255773187,
-      "learning_rate": 0.00019999999231354098,
-      "loss": 46.0,
-      "step": 785
-    },
-    {
-      "epoch": 0.12657514392688918,
-      "grad_norm": 0.0006392010254785419,
-      "learning_rate": 0.00019999999229369217,
-      "loss": 46.0,
-      "step": 786
-    },
-    {
-      "epoch": 0.12673618100567655,
-      "grad_norm": 0.000791541300714016,
-      "learning_rate": 0.00019999999227381777,
-      "loss": 46.0,
-      "step": 787
-    },
-    {
-      "epoch": 0.12689721808446394,
-      "grad_norm": 0.0006850518984720111,
-      "learning_rate": 0.00019999999225391775,
-      "loss": 46.0,
-      "step": 788
-    },
-    {
-      "epoch": 0.12705825516325134,
-      "grad_norm": 0.0010444442741572857,
-      "learning_rate": 0.00019999999223399215,
-      "loss": 46.0,
-      "step": 789
-    },
-    {
-      "epoch": 0.12721929224203873,
-      "grad_norm": 0.0006679598009213805,
-      "learning_rate": 0.00019999999221404093,
-      "loss": 46.0,
-      "step": 790
-    },
-    {
-      "epoch": 0.12738032932082613,
-      "grad_norm": 0.0009414281230419874,
-      "learning_rate": 0.00019999999219406415,
-      "loss": 46.0,
-      "step": 791
-    },
-    {
-      "epoch": 0.1275413663996135,
-      "grad_norm": 0.0002177072165068239,
-      "learning_rate": 0.00019999999217406176,
-      "loss": 46.0,
-      "step": 792
-    },
-    {
-      "epoch": 0.1277024034784009,
-      "grad_norm": 0.0007464716909453273,
-      "learning_rate": 0.00019999999215403378,
-      "loss": 46.0,
-      "step": 793
-    },
-    {
-      "epoch": 0.1278634405571883,
-      "grad_norm": 0.0006135865696705878,
-      "learning_rate": 0.0001999999921339802,
-      "loss": 46.0,
-      "step": 794
-    },
-    {
-      "epoch": 0.1280244776359757,
-      "grad_norm": 0.001909434562548995,
-      "learning_rate": 0.00019999999211390104,
-      "loss": 46.0,
-      "step": 795
-    },
-    {
-      "epoch": 0.12818551471476308,
-      "grad_norm": 0.0012111793039366603,
-      "learning_rate": 0.00019999999209379624,
-      "loss": 46.0,
-      "step": 796
-    },
-    {
-      "epoch": 0.12834655179355048,
-      "grad_norm": 0.0005831873277202249,
-      "learning_rate": 0.0001999999920736659,
-      "loss": 46.0,
-      "step": 797
-    },
-    {
-      "epoch": 0.12850758887233785,
-      "grad_norm": 0.0005240371683612466,
-      "learning_rate": 0.00019999999205350992,
-      "loss": 46.0,
-      "step": 798
-    },
-    {
-      "epoch": 0.12866862595112524,
-      "grad_norm": 0.0006707817665301263,
-      "learning_rate": 0.0001999999920333284,
-      "loss": 46.0,
-      "step": 799
-    },
-    {
-      "epoch": 0.12882966302991264,
-      "grad_norm": 0.0012076569255441427,
-      "learning_rate": 0.00019999999201312122,
-      "loss": 46.0,
-      "step": 800
-    },
-    {
-      "epoch": 0.12899070010870003,
-      "grad_norm": 0.0012680863728746772,
-      "learning_rate": 0.00019999999199288852,
-      "loss": 46.0,
-      "step": 801
-    },
-    {
-      "epoch": 0.12915173718748743,
-      "grad_norm": 0.0006465308251790702,
-      "learning_rate": 0.00019999999197263014,
-      "loss": 46.0,
-      "step": 802
-    },
-    {
-      "epoch": 0.1293127742662748,
-      "grad_norm": 0.0006390257622115314,
-      "learning_rate": 0.0001999999919523462,
-      "loss": 46.0,
-      "step": 803
-    },
-    {
-      "epoch": 0.1294738113450622,
-      "grad_norm": 0.00044906462426297367,
-      "learning_rate": 0.0001999999919320367,
-      "loss": 46.0,
-      "step": 804
-    },
-    {
-      "epoch": 0.1296348484238496,
-      "grad_norm": 0.001070193131454289,
-      "learning_rate": 0.0001999999919117016,
-      "loss": 46.0,
-      "step": 805
-    },
-    {
-      "epoch": 0.129795885502637,
-      "grad_norm": 0.0004957982455380261,
-      "learning_rate": 0.00019999999189134084,
-      "loss": 46.0,
-      "step": 806
-    },
-    {
-      "epoch": 0.12995692258142438,
-      "grad_norm": 0.00028306854073889554,
-      "learning_rate": 0.00019999999187095453,
-      "loss": 46.0,
-      "step": 807
-    },
-    {
-      "epoch": 0.13011795966021175,
-      "grad_norm": 0.001261686789803207,
-      "learning_rate": 0.00019999999185054264,
-      "loss": 46.0,
-      "step": 808
-    },
-    {
-      "epoch": 0.13027899673899915,
-      "grad_norm": 0.0004846264491789043,
-      "learning_rate": 0.00019999999183010513,
-      "loss": 46.0,
-      "step": 809
-    },
-    {
-      "epoch": 0.13044003381778654,
-      "grad_norm": 0.00039904803270474076,
-      "learning_rate": 0.00019999999180964203,
-      "loss": 46.0,
-      "step": 810
-    },
-    {
-      "epoch": 0.13060107089657394,
-      "grad_norm": 0.0005873935297131538,
-      "learning_rate": 0.00019999999178915335,
-      "loss": 46.0,
-      "step": 811
-    },
-    {
-      "epoch": 0.13076210797536134,
-      "grad_norm": 0.0003998268221039325,
-      "learning_rate": 0.00019999999176863905,
-      "loss": 46.0,
-      "step": 812
-    },
-    {
-      "epoch": 0.1309231450541487,
-      "grad_norm": 0.000923747313208878,
-      "learning_rate": 0.00019999999174809917,
-      "loss": 46.0,
-      "step": 813
-    },
-    {
-      "epoch": 0.1310841821329361,
-      "grad_norm": 0.0010215610964223742,
-      "learning_rate": 0.0001999999917275337,
-      "loss": 46.0,
-      "step": 814
-    },
-    {
-      "epoch": 0.1312452192117235,
-      "grad_norm": 0.00045427761506289244,
-      "learning_rate": 0.0001999999917069426,
-      "loss": 46.0,
-      "step": 815
-    },
-    {
-      "epoch": 0.1314062562905109,
-      "grad_norm": 0.00033573489054106176,
-      "learning_rate": 0.00019999999168632596,
-      "loss": 46.0,
-      "step": 816
-    },
-    {
-      "epoch": 0.1315672933692983,
-      "grad_norm": 0.0006197026814334095,
-      "learning_rate": 0.0001999999916656837,
-      "loss": 46.0,
-      "step": 817
-    },
-    {
-      "epoch": 0.13172833044808568,
-      "grad_norm": 0.000529855489730835,
-      "learning_rate": 0.00019999999164501585,
-      "loss": 46.0,
-      "step": 818
-    },
-    {
-      "epoch": 0.13188936752687305,
-      "grad_norm": 0.001919926144182682,
-      "learning_rate": 0.0001999999916243224,
-      "loss": 46.0,
-      "step": 819
-    },
-    {
-      "epoch": 0.13205040460566045,
-      "grad_norm": 0.00046961347106844187,
-      "learning_rate": 0.0001999999916036033,
-      "loss": 46.0,
-      "step": 820
-    },
-    {
-      "epoch": 0.13221144168444784,
-      "grad_norm": 0.0006622943910770118,
-      "learning_rate": 0.0001999999915828587,
-      "loss": 46.0,
-      "step": 821
-    },
-    {
-      "epoch": 0.13237247876323524,
-      "grad_norm": 0.00031556421890854836,
-      "learning_rate": 0.00019999999156208845,
-      "loss": 46.0,
-      "step": 822
-    },
-    {
-      "epoch": 0.13253351584202264,
-      "grad_norm": 0.000540242122951895,
-      "learning_rate": 0.00019999999154129264,
-      "loss": 46.0,
-      "step": 823
-    },
-    {
-      "epoch": 0.13269455292081,
-      "grad_norm": 0.0003218321071472019,
-      "learning_rate": 0.0001999999915204712,
-      "loss": 46.0,
-      "step": 824
-    },
-    {
-      "epoch": 0.1328555899995974,
-      "grad_norm": 0.000582866370677948,
-      "learning_rate": 0.00019999999149962418,
-      "loss": 46.0,
-      "step": 825
-    },
-    {
-      "epoch": 0.1330166270783848,
-      "grad_norm": 0.0018018016126006842,
-      "learning_rate": 0.00019999999147875158,
-      "loss": 46.0,
-      "step": 826
-    },
-    {
-      "epoch": 0.1331776641571722,
-      "grad_norm": 0.0010835527209565043,
-      "learning_rate": 0.00019999999145785336,
-      "loss": 46.0,
-      "step": 827
-    },
-    {
-      "epoch": 0.1333387012359596,
-      "grad_norm": 0.0008114787633530796,
-      "learning_rate": 0.00019999999143692956,
-      "loss": 46.0,
-      "step": 828
-    },
-    {
-      "epoch": 0.13349973831474696,
-      "grad_norm": 0.0005753064760938287,
-      "learning_rate": 0.00019999999141598015,
-      "loss": 46.0,
-      "step": 829
-    },
-    {
-      "epoch": 0.13366077539353435,
-      "grad_norm": 0.0004097472701687366,
-      "learning_rate": 0.00019999999139500515,
-      "loss": 46.0,
-      "step": 830
-    },
-    {
-      "epoch": 0.13382181247232175,
-      "grad_norm": 0.0005311646382324398,
-      "learning_rate": 0.00019999999137400458,
-      "loss": 46.0,
-      "step": 831
-    },
-    {
-      "epoch": 0.13398284955110915,
-      "grad_norm": 0.0011267259251326323,
-      "learning_rate": 0.00019999999135297838,
-      "loss": 46.0,
-      "step": 832
-    },
-    {
-      "epoch": 0.13414388662989654,
-      "grad_norm": 0.0002796841145027429,
-      "learning_rate": 0.0001999999913319266,
-      "loss": 46.0,
-      "step": 833
-    },
-    {
-      "epoch": 0.13430492370868394,
-      "grad_norm": 0.0008160819415934384,
-      "learning_rate": 0.00019999999131084924,
-      "loss": 46.0,
-      "step": 834
-    },
-    {
-      "epoch": 0.1344659607874713,
-      "grad_norm": 0.0005949228652752936,
-      "learning_rate": 0.00019999999128974627,
-      "loss": 46.0,
-      "step": 835
-    },
-    {
-      "epoch": 0.1346269978662587,
-      "grad_norm": 0.00039833923801779747,
-      "learning_rate": 0.0001999999912686177,
-      "loss": 46.0,
-      "step": 836
-    },
-    {
-      "epoch": 0.1347880349450461,
-      "grad_norm": 0.0007659844122827053,
-      "learning_rate": 0.00019999999124746353,
-      "loss": 46.0,
-      "step": 837
-    },
-    {
-      "epoch": 0.1349490720238335,
-      "grad_norm": 0.0008190494845621288,
-      "learning_rate": 0.0001999999912262838,
-      "loss": 46.0,
-      "step": 838
-    },
-    {
-      "epoch": 0.1351101091026209,
-      "grad_norm": 0.0006771478219889104,
-      "learning_rate": 0.00019999999120507843,
-      "loss": 46.0,
-      "step": 839
-    },
-    {
-      "epoch": 0.13527114618140826,
-      "grad_norm": 0.0004955170443281531,
-      "learning_rate": 0.0001999999911838475,
-      "loss": 46.0,
-      "step": 840
-    },
-    {
-      "epoch": 0.13543218326019565,
-      "grad_norm": 0.0004837427695747465,
-      "learning_rate": 0.00019999999116259096,
-      "loss": 46.0,
-      "step": 841
-    },
-    {
-      "epoch": 0.13559322033898305,
-      "grad_norm": 0.0005543114384636283,
-      "learning_rate": 0.00019999999114130883,
-      "loss": 46.0,
-      "step": 842
-    },
-    {
-      "epoch": 0.13575425741777045,
-      "grad_norm": 0.0011339414631947875,
-      "learning_rate": 0.0001999999911200011,
-      "loss": 46.0,
-      "step": 843
-    },
-    {
-      "epoch": 0.13591529449655784,
-      "grad_norm": 0.0006278124637901783,
-      "learning_rate": 0.00019999999109866776,
-      "loss": 46.0,
-      "step": 844
-    },
-    {
-      "epoch": 0.1360763315753452,
-      "grad_norm": 0.0016938516637310386,
-      "learning_rate": 0.00019999999107730884,
-      "loss": 46.0,
-      "step": 845
-    },
-    {
-      "epoch": 0.1362373686541326,
-      "grad_norm": 0.0019337739795446396,
-      "learning_rate": 0.00019999999105592436,
-      "loss": 46.0,
-      "step": 846
-    },
-    {
-      "epoch": 0.13639840573292,
-      "grad_norm": 0.0006349372561089694,
-      "learning_rate": 0.00019999999103451424,
-      "loss": 46.0,
-      "step": 847
-    },
-    {
-      "epoch": 0.1365594428117074,
-      "grad_norm": 0.0010687074391171336,
-      "learning_rate": 0.00019999999101307853,
-      "loss": 46.0,
-      "step": 848
-    },
-    {
-      "epoch": 0.1367204798904948,
-      "grad_norm": 0.0008426251588389277,
-      "learning_rate": 0.00019999999099161724,
-      "loss": 46.0,
-      "step": 849
-    },
-    {
-      "epoch": 0.1368815169692822,
-      "grad_norm": 0.0006499296869151294,
-      "learning_rate": 0.00019999999097013036,
-      "loss": 46.0,
-      "step": 850
-    },
-    {
-      "epoch": 0.13704255404806956,
-      "grad_norm": 0.0011350644053891301,
-      "learning_rate": 0.00019999999094861786,
-      "loss": 46.0,
-      "step": 851
-    },
-    {
-      "epoch": 0.13720359112685696,
-      "grad_norm": 0.0006878915010020137,
-      "learning_rate": 0.00019999999092707978,
-      "loss": 46.0,
-      "step": 852
-    },
-    {
-      "epoch": 0.13736462820564435,
-      "grad_norm": 0.000902946456335485,
-      "learning_rate": 0.00019999999090551608,
-      "loss": 46.0,
-      "step": 853
-    },
-    {
-      "epoch": 0.13752566528443175,
-      "grad_norm": 0.0003760393592528999,
-      "learning_rate": 0.00019999999088392682,
-      "loss": 46.0,
-      "step": 854
-    },
-    {
-      "epoch": 0.13768670236321914,
-      "grad_norm": 0.0011185002513229847,
-      "learning_rate": 0.00019999999086231195,
-      "loss": 46.0,
-      "step": 855
-    },
-    {
-      "epoch": 0.1378477394420065,
-      "grad_norm": 0.0007393692503683269,
-      "learning_rate": 0.00019999999084067152,
-      "loss": 46.0,
-      "step": 856
-    },
-    {
-      "epoch": 0.1380087765207939,
-      "grad_norm": 0.0006073627737350762,
-      "learning_rate": 0.00019999999081900545,
-      "loss": 46.0,
-      "step": 857
-    },
-    {
-      "epoch": 0.1381698135995813,
-      "grad_norm": 0.0011971439234912395,
-      "learning_rate": 0.0001999999907973138,
-      "loss": 46.0,
-      "step": 858
-    },
-    {
-      "epoch": 0.1383308506783687,
-      "grad_norm": 0.0007797310245223343,
-      "learning_rate": 0.00019999999077559654,
-      "loss": 46.0,
-      "step": 859
-    },
-    {
-      "epoch": 0.1384918877571561,
-      "grad_norm": 0.0002941296552307904,
-      "learning_rate": 0.0001999999907538537,
-      "loss": 46.0,
-      "step": 860
-    },
-    {
-      "epoch": 0.13865292483594346,
-      "grad_norm": 0.0004795315326191485,
-      "learning_rate": 0.00019999999073208526,
-      "loss": 46.0,
-      "step": 861
-    },
-    {
-      "epoch": 0.13881396191473086,
-      "grad_norm": 0.000500028021633625,
-      "learning_rate": 0.00019999999071029122,
-      "loss": 46.0,
-      "step": 862
-    },
-    {
-      "epoch": 0.13897499899351826,
-      "grad_norm": 0.0007521376828663051,
-      "learning_rate": 0.00019999999068847163,
-      "loss": 46.0,
-      "step": 863
-    },
-    {
-      "epoch": 0.13913603607230565,
-      "grad_norm": 0.0016052303835749626,
-      "learning_rate": 0.0001999999906666264,
-      "loss": 46.0,
-      "step": 864
-    },
-    {
-      "epoch": 0.13929707315109305,
-      "grad_norm": 0.0011634090915322304,
-      "learning_rate": 0.0001999999906447556,
-      "loss": 46.0,
-      "step": 865
-    },
-    {
-      "epoch": 0.13945811022988042,
-      "grad_norm": 0.0007818127050995827,
-      "learning_rate": 0.00019999999062285918,
-      "loss": 46.0,
-      "step": 866
-    },
-    {
-      "epoch": 0.1396191473086678,
-      "grad_norm": 0.00027503236196935177,
-      "learning_rate": 0.00019999999060093718,
-      "loss": 46.0,
-      "step": 867
-    },
-    {
-      "epoch": 0.1397801843874552,
-      "grad_norm": 0.0010879694018512964,
-      "learning_rate": 0.00019999999057898957,
-      "loss": 46.0,
-      "step": 868
-    },
-    {
-      "epoch": 0.1399412214662426,
-      "grad_norm": 0.00039905880112200975,
-      "learning_rate": 0.00019999999055701637,
-      "loss": 46.0,
-      "step": 869
-    },
-    {
-      "epoch": 0.14010225854503,
-      "grad_norm": 0.0005459731328301132,
-      "learning_rate": 0.00019999999053501758,
-      "loss": 46.0,
-      "step": 870
-    },
-    {
-      "epoch": 0.1402632956238174,
-      "grad_norm": 0.0011519825784489512,
-      "learning_rate": 0.0001999999905129932,
-      "loss": 46.0,
-      "step": 871
-    },
-    {
-      "epoch": 0.14042433270260476,
-      "grad_norm": 0.000506752694491297,
-      "learning_rate": 0.00019999999049094322,
-      "loss": 46.0,
-      "step": 872
-    },
-    {
-      "epoch": 0.14058536978139216,
-      "grad_norm": 0.0006271168240346014,
-      "learning_rate": 0.00019999999046886764,
-      "loss": 46.0,
-      "step": 873
-    },
-    {
-      "epoch": 0.14074640686017956,
-      "grad_norm": 0.0011113008949905634,
-      "learning_rate": 0.00019999999044676648,
-      "loss": 46.0,
-      "step": 874
-    },
-    {
-      "epoch": 0.14090744393896695,
-      "grad_norm": 0.0007349987863563001,
-      "learning_rate": 0.00019999999042463973,
-      "loss": 46.0,
-      "step": 875
-    },
-    {
-      "epoch": 0.14106848101775435,
-      "grad_norm": 0.0005502145504578948,
-      "learning_rate": 0.00019999999040248736,
-      "loss": 46.0,
-      "step": 876
-    },
-    {
-      "epoch": 0.14122951809654172,
-      "grad_norm": 0.0005725485389120877,
-      "learning_rate": 0.0001999999903803094,
-      "loss": 46.0,
-      "step": 877
-    },
-    {
-      "epoch": 0.1413905551753291,
-      "grad_norm": 0.0006727438885718584,
-      "learning_rate": 0.00019999999035810585,
-      "loss": 46.0,
-      "step": 878
-    },
-    {
-      "epoch": 0.1415515922541165,
-      "grad_norm": 0.00040779856499284506,
-      "learning_rate": 0.00019999999033587672,
-      "loss": 46.0,
-      "step": 879
-    },
-    {
-      "epoch": 0.1417126293329039,
-      "grad_norm": 0.0006494661211036146,
-      "learning_rate": 0.00019999999031362198,
-      "loss": 46.0,
-      "step": 880
-    },
-    {
-      "epoch": 0.1418736664116913,
-      "grad_norm": 0.001008855295367539,
-      "learning_rate": 0.00019999999029134165,
-      "loss": 46.0,
-      "step": 881
-    },
-    {
-      "epoch": 0.14203470349047867,
-      "grad_norm": 0.0003528682282194495,
-      "learning_rate": 0.0001999999902690357,
-      "loss": 46.0,
-      "step": 882
-    },
-    {
-      "epoch": 0.14219574056926607,
-      "grad_norm": 0.001214552205055952,
-      "learning_rate": 0.00019999999024670418,
-      "loss": 46.0,
-      "step": 883
-    },
-    {
-      "epoch": 0.14235677764805346,
-      "grad_norm": 0.0005489051109179854,
-      "learning_rate": 0.00019999999022434707,
-      "loss": 46.0,
-      "step": 884
-    },
-    {
-      "epoch": 0.14251781472684086,
-      "grad_norm": 0.0011040157405659556,
-      "learning_rate": 0.00019999999020196436,
-      "loss": 46.0,
-      "step": 885
-    },
-    {
-      "epoch": 0.14267885180562825,
-      "grad_norm": 0.0006406849715858698,
-      "learning_rate": 0.00019999999017955605,
-      "loss": 46.0,
-      "step": 886
-    },
-    {
-      "epoch": 0.14283988888441565,
-      "grad_norm": 0.0006732973852194846,
-      "learning_rate": 0.00019999999015712217,
-      "loss": 46.0,
-      "step": 887
-    },
-    {
-      "epoch": 0.14300092596320302,
-      "grad_norm": 0.0010554519249126315,
-      "learning_rate": 0.00019999999013466268,
-      "loss": 46.0,
-      "step": 888
-    },
-    {
-      "epoch": 0.14316196304199041,
-      "grad_norm": 0.0007054017623886466,
-      "learning_rate": 0.00019999999011217757,
-      "loss": 46.0,
-      "step": 889
-    },
-    {
-      "epoch": 0.1433230001207778,
-      "grad_norm": 0.0011370867723599076,
-      "learning_rate": 0.00019999999008966688,
-      "loss": 46.0,
-      "step": 890
-    },
-    {
-      "epoch": 0.1434840371995652,
-      "grad_norm": 0.0006134258583188057,
-      "learning_rate": 0.0001999999900671306,
-      "loss": 46.0,
-      "step": 891
-    },
-    {
-      "epoch": 0.1436450742783526,
-      "grad_norm": 0.000407539278967306,
-      "learning_rate": 0.00019999999004456873,
-      "loss": 46.0,
-      "step": 892
-    },
-    {
-      "epoch": 0.14380611135713997,
-      "grad_norm": 0.0005476608639582992,
-      "learning_rate": 0.00019999999002198125,
-      "loss": 46.0,
-      "step": 893
-    },
-    {
-      "epoch": 0.14396714843592737,
-      "grad_norm": 0.0007749724318273365,
-      "learning_rate": 0.00019999998999936818,
-      "loss": 46.0,
-      "step": 894
-    },
-    {
-      "epoch": 0.14412818551471476,
-      "grad_norm": 0.0010029029799625278,
-      "learning_rate": 0.00019999998997672952,
-      "loss": 46.0,
-      "step": 895
-    },
-    {
-      "epoch": 0.14428922259350216,
-      "grad_norm": 0.000552076380699873,
-      "learning_rate": 0.00019999998995406525,
-      "loss": 46.0,
-      "step": 896
-    },
-    {
-      "epoch": 0.14445025967228955,
-      "grad_norm": 0.0011516448576003313,
-      "learning_rate": 0.00019999998993137542,
-      "loss": 46.0,
-      "step": 897
-    },
-    {
-      "epoch": 0.14461129675107692,
-      "grad_norm": 0.0010751265799626708,
-      "learning_rate": 0.00019999998990865997,
-      "loss": 46.0,
-      "step": 898
-    },
-    {
-      "epoch": 0.14477233382986432,
-      "grad_norm": 0.0006675152108073235,
-      "learning_rate": 0.00019999998988591891,
-      "loss": 46.0,
-      "step": 899
-    },
-    {
-      "epoch": 0.14493337090865172,
-      "grad_norm": 0.0003535577270668,
-      "learning_rate": 0.0001999999898631523,
-      "loss": 46.0,
-      "step": 900
-    },
-    {
-      "epoch": 0.1450944079874391,
-      "grad_norm": 0.0012404737062752247,
-      "learning_rate": 0.0001999999898403601,
-      "loss": 46.0,
-      "step": 901
-    },
-    {
-      "epoch": 0.1452554450662265,
-      "grad_norm": 0.0006735806819051504,
-      "learning_rate": 0.00019999998981754224,
-      "loss": 46.0,
-      "step": 902
-    },
-    {
-      "epoch": 0.14541648214501388,
-      "grad_norm": 0.0003508478112053126,
-      "learning_rate": 0.00019999998979469883,
-      "loss": 46.0,
-      "step": 903
-    },
-    {
-      "epoch": 0.14557751922380127,
-      "grad_norm": 0.0003961780748795718,
-      "learning_rate": 0.0001999999897718298,
-      "loss": 46.0,
-      "step": 904
-    },
-    {
-      "epoch": 0.14573855630258867,
-      "grad_norm": 0.0009000244317576289,
-      "learning_rate": 0.0001999999897489352,
-      "loss": 46.0,
-      "step": 905
-    },
-    {
-      "epoch": 0.14589959338137606,
-      "grad_norm": 0.001379994209855795,
-      "learning_rate": 0.000199999989726015,
-      "loss": 46.0,
-      "step": 906
-    },
-    {
-      "epoch": 0.14606063046016346,
-      "grad_norm": 0.0010532027808949351,
-      "learning_rate": 0.0001999999897030692,
-      "loss": 46.0,
-      "step": 907
-    },
-    {
-      "epoch": 0.14622166753895086,
-      "grad_norm": 0.0007648330647498369,
-      "learning_rate": 0.0001999999896800978,
-      "loss": 46.0,
-      "step": 908
-    },
-    {
-      "epoch": 0.14638270461773822,
-      "grad_norm": 0.000269240525085479,
-      "learning_rate": 0.0001999999896571008,
-      "loss": 46.0,
-      "step": 909
-    },
-    {
-      "epoch": 0.14654374169652562,
-      "grad_norm": 0.0006921233725734055,
-      "learning_rate": 0.0001999999896340782,
-      "loss": 46.0,
-      "step": 910
-    },
-    {
-      "epoch": 0.14670477877531302,
-      "grad_norm": 0.0009213497396558523,
-      "learning_rate": 0.00019999998961103005,
-      "loss": 46.0,
-      "step": 911
-    },
-    {
-      "epoch": 0.1468658158541004,
-      "grad_norm": 0.000873835408128798,
-      "learning_rate": 0.00019999998958795628,
-      "loss": 46.0,
-      "step": 912
-    },
-    {
-      "epoch": 0.1470268529328878,
-      "grad_norm": 0.00044415402226150036,
-      "learning_rate": 0.00019999998956485692,
-      "loss": 46.0,
-      "step": 913
-    },
-    {
-      "epoch": 0.14718789001167518,
-      "grad_norm": 0.0006393120274879038,
-      "learning_rate": 0.00019999998954173194,
-      "loss": 46.0,
-      "step": 914
-    },
-    {
-      "epoch": 0.14734892709046257,
-      "grad_norm": 0.0005304512451402843,
-      "learning_rate": 0.00019999998951858138,
-      "loss": 46.0,
-      "step": 915
-    },
-    {
-      "epoch": 0.14750996416924997,
-      "grad_norm": 0.0004301891603972763,
-      "learning_rate": 0.0001999999894954052,
-      "loss": 46.0,
-      "step": 916
-    },
-    {
-      "epoch": 0.14767100124803736,
-      "grad_norm": 0.0005527819157578051,
-      "learning_rate": 0.0001999999894722035,
-      "loss": 46.0,
-      "step": 917
-    },
-    {
-      "epoch": 0.14783203832682476,
-      "grad_norm": 0.0008386385161429644,
-      "learning_rate": 0.00019999998944897614,
-      "loss": 46.0,
-      "step": 918
-    },
-    {
-      "epoch": 0.14799307540561213,
-      "grad_norm": 0.0010009764228016138,
-      "learning_rate": 0.0001999999894257232,
-      "loss": 46.0,
-      "step": 919
-    },
-    {
-      "epoch": 0.14815411248439952,
-      "grad_norm": 0.0014237198047339916,
-      "learning_rate": 0.00019999998940244468,
-      "loss": 46.0,
-      "step": 920
-    },
-    {
-      "epoch": 0.14831514956318692,
-      "grad_norm": 0.0007006244268268347,
-      "learning_rate": 0.00019999998937914054,
-      "loss": 46.0,
-      "step": 921
-    },
-    {
-      "epoch": 0.14847618664197432,
-      "grad_norm": 0.0006828503683209419,
-      "learning_rate": 0.00019999998935581084,
-      "loss": 46.0,
-      "step": 922
-    },
-    {
-      "epoch": 0.1486372237207617,
-      "grad_norm": 0.0011275885626673698,
-      "learning_rate": 0.00019999998933245553,
-      "loss": 46.0,
-      "step": 923
-    },
-    {
-      "epoch": 0.1487982607995491,
-      "grad_norm": 0.0003434883547015488,
-      "learning_rate": 0.00019999998930907457,
-      "loss": 46.0,
-      "step": 924
-    },
-    {
-      "epoch": 0.14895929787833648,
-      "grad_norm": 0.0004500861978158355,
-      "learning_rate": 0.0001999999892856681,
-      "loss": 46.0,
-      "step": 925
-    },
-    {
-      "epoch": 0.14912033495712387,
-      "grad_norm": 0.0009388361359015107,
-      "learning_rate": 0.00019999998926223598,
-      "loss": 46.0,
-      "step": 926
-    },
-    {
-      "epoch": 0.14928137203591127,
-      "grad_norm": 0.0006658335332758725,
-      "learning_rate": 0.00019999998923877826,
-      "loss": 46.0,
-      "step": 927
-    },
-    {
-      "epoch": 0.14944240911469867,
-      "grad_norm": 0.0005450811004266143,
-      "learning_rate": 0.00019999998921529498,
-      "loss": 46.0,
-      "step": 928
-    },
-    {
-      "epoch": 0.14960344619348606,
-      "grad_norm": 0.0003031016094610095,
-      "learning_rate": 0.00019999998919178612,
-      "loss": 46.0,
-      "step": 929
-    },
-    {
-      "epoch": 0.14976448327227343,
-      "grad_norm": 0.0008631189703010023,
-      "learning_rate": 0.00019999998916825162,
-      "loss": 46.0,
-      "step": 930
-    },
-    {
-      "epoch": 0.14992552035106083,
-      "grad_norm": 0.0008258898742496967,
-      "learning_rate": 0.00019999998914469152,
-      "loss": 46.0,
-      "step": 931
-    },
-    {
-      "epoch": 0.15008655742984822,
-      "grad_norm": 0.0003272114263381809,
-      "learning_rate": 0.00019999998912110587,
-      "loss": 46.0,
-      "step": 932
-    },
-    {
-      "epoch": 0.15024759450863562,
-      "grad_norm": 0.00045941982534714043,
-      "learning_rate": 0.00019999998909749463,
-      "loss": 46.0,
-      "step": 933
-    },
-    {
-      "epoch": 0.15040863158742301,
-      "grad_norm": 0.000885982473846525,
-      "learning_rate": 0.00019999998907385775,
-      "loss": 46.0,
-      "step": 934
-    },
-    {
-      "epoch": 0.15056966866621038,
-      "grad_norm": 0.0003253989852964878,
-      "learning_rate": 0.0001999999890501953,
-      "loss": 46.0,
-      "step": 935
-    },
-    {
-      "epoch": 0.15073070574499778,
-      "grad_norm": 0.0009691553423181176,
-      "learning_rate": 0.00019999998902650725,
-      "loss": 46.0,
-      "step": 936
-    },
-    {
-      "epoch": 0.15089174282378517,
-      "grad_norm": 0.0006153081776574254,
-      "learning_rate": 0.0001999999890027936,
-      "loss": 46.0,
-      "step": 937
-    },
-    {
-      "epoch": 0.15105277990257257,
-      "grad_norm": 0.0006243716343306005,
-      "learning_rate": 0.00019999998897905435,
-      "loss": 46.0,
-      "step": 938
-    },
-    {
-      "epoch": 0.15121381698135997,
-      "grad_norm": 0.0013074682792648673,
-      "learning_rate": 0.00019999998895528953,
-      "loss": 46.0,
-      "step": 939
-    },
-    {
-      "epoch": 0.15137485406014736,
-      "grad_norm": 0.0004748936044052243,
-      "learning_rate": 0.0001999999889314991,
-      "loss": 46.0,
-      "step": 940
-    },
-    {
-      "epoch": 0.15153589113893473,
-      "grad_norm": 0.0005445227143354714,
-      "learning_rate": 0.00019999998890768308,
-      "loss": 46.0,
-      "step": 941
-    },
-    {
-      "epoch": 0.15169692821772213,
-      "grad_norm": 0.00029829939012415707,
-      "learning_rate": 0.00019999998888384145,
-      "loss": 46.0,
-      "step": 942
-    },
-    {
-      "epoch": 0.15185796529650952,
-      "grad_norm": 0.0007962873787619174,
-      "learning_rate": 0.00019999998885997423,
-      "loss": 46.0,
-      "step": 943
-    },
-    {
-      "epoch": 0.15201900237529692,
-      "grad_norm": 0.0008284706273116171,
-      "learning_rate": 0.00019999998883608142,
-      "loss": 46.0,
-      "step": 944
-    },
-    {
-      "epoch": 0.15218003945408431,
-      "grad_norm": 0.0007207043236121535,
-      "learning_rate": 0.00019999998881216303,
-      "loss": 46.0,
-      "step": 945
-    },
-    {
-      "epoch": 0.15234107653287168,
-      "grad_norm": 0.0007585242274217308,
-      "learning_rate": 0.00019999998878821905,
-      "loss": 46.0,
-      "step": 946
-    },
-    {
-      "epoch": 0.15250211361165908,
-      "grad_norm": 0.00043909219675697386,
-      "learning_rate": 0.00019999998876424943,
-      "loss": 46.0,
-      "step": 947
-    },
-    {
-      "epoch": 0.15266315069044648,
-      "grad_norm": 0.0002628343063406646,
-      "learning_rate": 0.00019999998874025424,
-      "loss": 46.0,
-      "step": 948
-    },
-    {
-      "epoch": 0.15282418776923387,
-      "grad_norm": 0.0010836570290848613,
-      "learning_rate": 0.00019999998871623347,
-      "loss": 46.0,
-      "step": 949
-    },
-    {
-      "epoch": 0.15298522484802127,
-      "grad_norm": 0.0008976529352366924,
-      "learning_rate": 0.0001999999886921871,
-      "loss": 46.0,
-      "step": 950
-    },
-    {
-      "epoch": 0.15314626192680864,
-      "grad_norm": 0.0010474735172465444,
-      "learning_rate": 0.00019999998866811512,
-      "loss": 46.0,
-      "step": 951
-    },
-    {
-      "epoch": 0.15330729900559603,
-      "grad_norm": 0.00025471372646279633,
-      "learning_rate": 0.00019999998864401756,
-      "loss": 46.0,
-      "step": 952
-    },
-    {
-      "epoch": 0.15346833608438343,
-      "grad_norm": 0.00031341315479949117,
-      "learning_rate": 0.00019999998861989438,
-      "loss": 46.0,
-      "step": 953
-    },
-    {
-      "epoch": 0.15362937316317082,
-      "grad_norm": 0.0015352104092016816,
-      "learning_rate": 0.00019999998859574562,
-      "loss": 46.0,
-      "step": 954
-    },
-    {
-      "epoch": 0.15379041024195822,
-      "grad_norm": 0.0009060533484444022,
-      "learning_rate": 0.00019999998857157128,
-      "loss": 46.0,
-      "step": 955
-    },
-    {
-      "epoch": 0.1539514473207456,
-      "grad_norm": 0.00029884095420129597,
-      "learning_rate": 0.00019999998854737132,
-      "loss": 46.0,
-      "step": 956
-    },
-    {
-      "epoch": 0.15411248439953298,
-      "grad_norm": 0.0008722096681594849,
-      "learning_rate": 0.0001999999885231458,
-      "loss": 46.0,
-      "step": 957
-    },
-    {
-      "epoch": 0.15427352147832038,
-      "grad_norm": 0.00042822243995033205,
-      "learning_rate": 0.00019999998849889463,
-      "loss": 46.0,
-      "step": 958
-    },
-    {
-      "epoch": 0.15443455855710778,
-      "grad_norm": 0.0003030253865290433,
-      "learning_rate": 0.0001999999884746179,
-      "loss": 46.0,
-      "step": 959
-    },
-    {
-      "epoch": 0.15459559563589517,
-      "grad_norm": 0.0005117275868542492,
-      "learning_rate": 0.00019999998845031557,
-      "loss": 46.0,
-      "step": 960
-    },
-    {
-      "epoch": 0.15475663271468257,
-      "grad_norm": 0.0003974712744820863,
-      "learning_rate": 0.00019999998842598764,
-      "loss": 46.0,
-      "step": 961
-    },
-    {
-      "epoch": 0.15491766979346994,
-      "grad_norm": 0.0008037528023123741,
-      "learning_rate": 0.00019999998840163413,
-      "loss": 46.0,
-      "step": 962
-    },
-    {
-      "epoch": 0.15507870687225733,
-      "grad_norm": 0.0004638684040401131,
-      "learning_rate": 0.00019999998837725503,
-      "loss": 46.0,
-      "step": 963
-    },
-    {
-      "epoch": 0.15523974395104473,
-      "grad_norm": 0.0005257294396869838,
-      "learning_rate": 0.00019999998835285032,
-      "loss": 46.0,
-      "step": 964
-    },
-    {
-      "epoch": 0.15540078102983212,
-      "grad_norm": 0.0006085250061005354,
-      "learning_rate": 0.00019999998832842002,
-      "loss": 46.0,
-      "step": 965
-    },
-    {
-      "epoch": 0.15556181810861952,
-      "grad_norm": 0.0010381838073953986,
-      "learning_rate": 0.00019999998830396413,
-      "loss": 46.0,
-      "step": 966
-    },
-    {
-      "epoch": 0.1557228551874069,
-      "grad_norm": 0.001192597672343254,
-      "learning_rate": 0.00019999998827948263,
-      "loss": 46.0,
-      "step": 967
-    },
-    {
-      "epoch": 0.15588389226619428,
-      "grad_norm": 0.0005393070168793201,
-      "learning_rate": 0.00019999998825497554,
-      "loss": 46.0,
-      "step": 968
-    },
-    {
-      "epoch": 0.15604492934498168,
-      "grad_norm": 0.0009535938152112067,
-      "learning_rate": 0.00019999998823044287,
-      "loss": 46.0,
-      "step": 969
-    },
-    {
-      "epoch": 0.15620596642376908,
-      "grad_norm": 0.001277977367863059,
-      "learning_rate": 0.0001999999882058846,
-      "loss": 46.0,
-      "step": 970
-    },
-    {
-      "epoch": 0.15636700350255647,
-      "grad_norm": 0.000667413289193064,
-      "learning_rate": 0.0001999999881813007,
-      "loss": 46.0,
-      "step": 971
-    },
-    {
-      "epoch": 0.15652804058134384,
-      "grad_norm": 0.0013546354603022337,
-      "learning_rate": 0.00019999998815669126,
-      "loss": 46.0,
-      "step": 972
-    },
-    {
-      "epoch": 0.15668907766013124,
-      "grad_norm": 0.0005856928182765841,
-      "learning_rate": 0.00019999998813205618,
-      "loss": 46.0,
-      "step": 973
-    },
-    {
-      "epoch": 0.15685011473891863,
-      "grad_norm": 0.0004939138889312744,
-      "learning_rate": 0.00019999998810739554,
-      "loss": 46.0,
-      "step": 974
-    },
-    {
-      "epoch": 0.15701115181770603,
-      "grad_norm": 0.0005340013885870576,
-      "learning_rate": 0.0001999999880827093,
-      "loss": 46.0,
-      "step": 975
-    },
-    {
-      "epoch": 0.15717218889649343,
-      "grad_norm": 0.00032319730962626636,
-      "learning_rate": 0.00019999998805799742,
-      "loss": 46.0,
-      "step": 976
-    },
-    {
-      "epoch": 0.15733322597528082,
-      "grad_norm": 0.0008735068258829415,
-      "learning_rate": 0.00019999998803326002,
-      "loss": 46.0,
-      "step": 977
-    },
-    {
-      "epoch": 0.1574942630540682,
-      "grad_norm": 0.0009990877006202936,
-      "learning_rate": 0.00019999998800849698,
-      "loss": 46.0,
-      "step": 978
-    },
-    {
-      "epoch": 0.15765530013285559,
-      "grad_norm": 0.00027033319929614663,
-      "learning_rate": 0.00019999998798370835,
-      "loss": 46.0,
-      "step": 979
-    },
-    {
-      "epoch": 0.15781633721164298,
-      "grad_norm": 0.0009129714453592896,
-      "learning_rate": 0.00019999998795889413,
-      "loss": 46.0,
-      "step": 980
-    },
-    {
-      "epoch": 0.15797737429043038,
-      "grad_norm": 0.0018556106369942427,
-      "learning_rate": 0.0001999999879340543,
-      "loss": 46.0,
-      "step": 981
-    },
-    {
-      "epoch": 0.15813841136921777,
-      "grad_norm": 0.000484311836771667,
-      "learning_rate": 0.00019999998790918888,
-      "loss": 46.0,
-      "step": 982
-    },
-    {
-      "epoch": 0.15829944844800514,
-      "grad_norm": 0.0010563847608864307,
-      "learning_rate": 0.00019999998788429788,
-      "loss": 46.0,
-      "step": 983
-    },
-    {
-      "epoch": 0.15846048552679254,
-      "grad_norm": 0.0013567782007157803,
-      "learning_rate": 0.00019999998785938129,
-      "loss": 46.0,
-      "step": 984
-    },
-    {
-      "epoch": 0.15862152260557993,
-      "grad_norm": 0.0008116239914670587,
-      "learning_rate": 0.00019999998783443905,
-      "loss": 46.0,
-      "step": 985
-    },
-    {
-      "epoch": 0.15878255968436733,
-      "grad_norm": 0.00039403728442266583,
-      "learning_rate": 0.00019999998780947126,
-      "loss": 46.0,
-      "step": 986
-    },
-    {
-      "epoch": 0.15894359676315473,
-      "grad_norm": 0.0007989925798028708,
-      "learning_rate": 0.00019999998778447788,
-      "loss": 46.0,
-      "step": 987
-    },
-    {
-      "epoch": 0.1591046338419421,
-      "grad_norm": 0.00031061729532666504,
-      "learning_rate": 0.00019999998775945888,
-      "loss": 46.0,
-      "step": 988
-    },
-    {
-      "epoch": 0.1592656709207295,
-      "grad_norm": 0.0004982459358870983,
-      "learning_rate": 0.0001999999877344143,
-      "loss": 46.0,
-      "step": 989
-    },
-    {
-      "epoch": 0.1594267079995169,
-      "grad_norm": 0.0007720077410340309,
-      "learning_rate": 0.00019999998770934413,
-      "loss": 46.0,
-      "step": 990
-    },
-    {
-      "epoch": 0.15958774507830428,
-      "grad_norm": 0.0003660463262349367,
-      "learning_rate": 0.0001999999876842484,
-      "loss": 46.0,
-      "step": 991
-    },
-    {
-      "epoch": 0.15974878215709168,
-      "grad_norm": 0.0008429509471170604,
-      "learning_rate": 0.00019999998765912703,
-      "loss": 46.0,
-      "step": 992
-    },
-    {
-      "epoch": 0.15990981923587907,
-      "grad_norm": 0.0009334971546195447,
-      "learning_rate": 0.00019999998763398007,
-      "loss": 46.0,
-      "step": 993
-    },
-    {
-      "epoch": 0.16007085631466644,
-      "grad_norm": 0.00027421597042120993,
-      "learning_rate": 0.0001999999876088075,
-      "loss": 46.0,
-      "step": 994
-    },
-    {
-      "epoch": 0.16023189339345384,
-      "grad_norm": 0.0013092216104269028,
-      "learning_rate": 0.00019999998758360937,
-      "loss": 46.0,
-      "step": 995
-    },
-    {
-      "epoch": 0.16039293047224124,
-      "grad_norm": 0.0007798975566402078,
-      "learning_rate": 0.0001999999875583856,
-      "loss": 46.0,
-      "step": 996
-    },
-    {
-      "epoch": 0.16055396755102863,
-      "grad_norm": 0.00038831288111396134,
-      "learning_rate": 0.0001999999875331363,
-      "loss": 46.0,
-      "step": 997
-    },
-    {
-      "epoch": 0.16071500462981603,
-      "grad_norm": 0.00036058248952031136,
-      "learning_rate": 0.00019999998750786137,
-      "loss": 46.0,
-      "step": 998
-    },
-    {
-      "epoch": 0.1608760417086034,
-      "grad_norm": 0.00032413488952443004,
-      "learning_rate": 0.0001999999874825608,
-      "loss": 46.0,
-      "step": 999
-    },
-    {
-      "epoch": 0.1610370787873908,
-      "grad_norm": 0.0008596943807788193,
-      "learning_rate": 0.0001999999874572347,
-      "loss": 46.0,
-      "step": 1000
-    },
-    {
-      "epoch": 0.1611981158661782,
-      "grad_norm": 0.0003194579330738634,
-      "learning_rate": 0.00019999998743188297,
-      "loss": 46.0,
-      "step": 1001
-    },
-    {
-      "epoch": 0.16135915294496558,
-      "grad_norm": 0.00046182089135982096,
-      "learning_rate": 0.00019999998740650567,
-      "loss": 46.0,
-      "step": 1002
-    },
-    {
-      "epoch": 0.16152019002375298,
-      "grad_norm": 0.0010411414550617337,
-      "learning_rate": 0.00019999998738110274,
-      "loss": 46.0,
-      "step": 1003
-    },
-    {
-      "epoch": 0.16168122710254035,
-      "grad_norm": 0.0009827236644923687,
-      "learning_rate": 0.00019999998735567424,
-      "loss": 46.0,
-      "step": 1004
-    },
-    {
-      "epoch": 0.16184226418132774,
-      "grad_norm": 0.0006024896283634007,
-      "learning_rate": 0.00019999998733022015,
-      "loss": 46.0,
-      "step": 1005
-    },
-    {
-      "epoch": 0.16200330126011514,
-      "grad_norm": 0.00046483069309033453,
-      "learning_rate": 0.00019999998730474048,
-      "loss": 46.0,
-      "step": 1006
-    },
-    {
-      "epoch": 0.16216433833890254,
-      "grad_norm": 0.0009994750143960118,
-      "learning_rate": 0.0001999999872792352,
-      "loss": 46.0,
-      "step": 1007
-    },
-    {
-      "epoch": 0.16232537541768993,
-      "grad_norm": 0.0002608858922030777,
-      "learning_rate": 0.0001999999872537043,
-      "loss": 46.0,
-      "step": 1008
-    },
-    {
-      "epoch": 0.1624864124964773,
-      "grad_norm": 0.0008573723607696593,
-      "learning_rate": 0.00019999998722814783,
-      "loss": 46.0,
-      "step": 1009
-    },
-    {
-      "epoch": 0.1626474495752647,
-      "grad_norm": 0.00032435834873467684,
-      "learning_rate": 0.00019999998720256575,
-      "loss": 46.0,
-      "step": 1010
-    },
-    {
-      "epoch": 0.1628084866540521,
-      "grad_norm": 0.0008361474028788507,
-      "learning_rate": 0.0001999999871769581,
-      "loss": 46.0,
-      "step": 1011
-    },
-    {
-      "epoch": 0.1629695237328395,
-      "grad_norm": 0.0008367006666958332,
-      "learning_rate": 0.00019999998715132482,
-      "loss": 46.0,
-      "step": 1012
-    },
-    {
-      "epoch": 0.16313056081162688,
-      "grad_norm": 0.0007053214940242469,
-      "learning_rate": 0.00019999998712566598,
-      "loss": 46.0,
-      "step": 1013
-    },
-    {
-      "epoch": 0.16329159789041428,
-      "grad_norm": 0.000897809921298176,
-      "learning_rate": 0.00019999998709998153,
-      "loss": 46.0,
-      "step": 1014
-    },
-    {
-      "epoch": 0.16345263496920165,
-      "grad_norm": 0.0006495975540019572,
-      "learning_rate": 0.0001999999870742715,
-      "loss": 46.0,
-      "step": 1015
-    },
-    {
-      "epoch": 0.16361367204798904,
-      "grad_norm": 0.0011052775662392378,
-      "learning_rate": 0.00019999998704853584,
-      "loss": 46.0,
-      "step": 1016
-    },
-    {
-      "epoch": 0.16377470912677644,
-      "grad_norm": 0.001142477267421782,
-      "learning_rate": 0.00019999998702277462,
-      "loss": 46.0,
-      "step": 1017
-    },
-    {
-      "epoch": 0.16393574620556384,
-      "grad_norm": 0.00043422140879556537,
-      "learning_rate": 0.0001999999869969878,
-      "loss": 46.0,
-      "step": 1018
-    },
-    {
-      "epoch": 0.16409678328435123,
-      "grad_norm": 0.0006619326304644346,
-      "learning_rate": 0.00019999998697117538,
-      "loss": 46.0,
-      "step": 1019
-    },
-    {
-      "epoch": 0.1642578203631386,
-      "grad_norm": 0.0006206837715581059,
-      "learning_rate": 0.00019999998694533733,
-      "loss": 46.0,
-      "step": 1020
-    },
-    {
-      "epoch": 0.164418857441926,
-      "grad_norm": 0.0008341065840795636,
-      "learning_rate": 0.00019999998691947374,
-      "loss": 46.0,
-      "step": 1021
-    },
-    {
-      "epoch": 0.1645798945207134,
-      "grad_norm": 0.00038806578959338367,
-      "learning_rate": 0.0001999999868935845,
-      "loss": 46.0,
-      "step": 1022
-    },
-    {
-      "epoch": 0.1647409315995008,
-      "grad_norm": 0.0005078114336356521,
-      "learning_rate": 0.00019999998686766972,
-      "loss": 46.0,
-      "step": 1023
-    },
-    {
-      "epoch": 0.16490196867828819,
-      "grad_norm": 0.0003059906011912972,
-      "learning_rate": 0.0001999999868417293,
-      "loss": 46.0,
-      "step": 1024
-    },
-    {
-      "epoch": 0.16506300575707555,
-      "grad_norm": 0.0013124411925673485,
-      "learning_rate": 0.00019999998681576332,
-      "loss": 46.0,
-      "step": 1025
-    },
-    {
-      "epoch": 0.16522404283586295,
-      "grad_norm": 0.00030253586010076106,
-      "learning_rate": 0.00019999998678977172,
-      "loss": 46.0,
-      "step": 1026
-    },
-    {
-      "epoch": 0.16538507991465035,
-      "grad_norm": 0.0005021268152631819,
-      "learning_rate": 0.00019999998676375452,
-      "loss": 46.0,
-      "step": 1027
-    },
-    {
-      "epoch": 0.16554611699343774,
-      "grad_norm": 0.0009989166865125299,
-      "learning_rate": 0.00019999998673771177,
-      "loss": 46.0,
-      "step": 1028
-    },
-    {
-      "epoch": 0.16570715407222514,
-      "grad_norm": 0.001155222998932004,
-      "learning_rate": 0.0001999999867116434,
-      "loss": 46.0,
-      "step": 1029
-    },
-    {
-      "epoch": 0.16586819115101253,
-      "grad_norm": 0.0005994467646814883,
-      "learning_rate": 0.00019999998668554942,
-      "loss": 46.0,
-      "step": 1030
-    },
-    {
-      "epoch": 0.1660292282297999,
-      "grad_norm": 0.0009421958820894361,
-      "learning_rate": 0.00019999998665942985,
-      "loss": 46.0,
-      "step": 1031
-    },
-    {
-      "epoch": 0.1661902653085873,
-      "grad_norm": 0.0006251431186683476,
-      "learning_rate": 0.00019999998663328472,
-      "loss": 46.0,
-      "step": 1032
-    },
-    {
-      "epoch": 0.1663513023873747,
-      "grad_norm": 0.0008859604713506997,
-      "learning_rate": 0.00019999998660711395,
-      "loss": 46.0,
-      "step": 1033
-    },
-    {
-      "epoch": 0.1665123394661621,
-      "grad_norm": 0.0003262088866904378,
-      "learning_rate": 0.0001999999865809176,
-      "loss": 46.0,
-      "step": 1034
-    },
-    {
-      "epoch": 0.1666733765449495,
-      "grad_norm": 0.0009352091583423316,
-      "learning_rate": 0.00019999998655469568,
-      "loss": 46.0,
-      "step": 1035
-    },
-    {
-      "epoch": 0.16683441362373685,
-      "grad_norm": 0.0007035856833681464,
-      "learning_rate": 0.00019999998652844815,
-      "loss": 46.0,
-      "step": 1036
-    },
-    {
-      "epoch": 0.16699545070252425,
-      "grad_norm": 0.00041801491170190275,
-      "learning_rate": 0.000199999986502175,
-      "loss": 46.0,
-      "step": 1037
-    },
-    {
-      "epoch": 0.16715648778131165,
-      "grad_norm": 0.0001697216648608446,
-      "learning_rate": 0.00019999998647587627,
-      "loss": 46.0,
-      "step": 1038
-    },
-    {
-      "epoch": 0.16731752486009904,
-      "grad_norm": 0.0004610596806742251,
-      "learning_rate": 0.00019999998644955195,
-      "loss": 46.0,
-      "step": 1039
-    },
-    {
-      "epoch": 0.16747856193888644,
-      "grad_norm": 0.0004097422061022371,
-      "learning_rate": 0.00019999998642320204,
-      "loss": 46.0,
-      "step": 1040
-    },
-    {
-      "epoch": 0.1676395990176738,
-      "grad_norm": 0.0007244375301524997,
-      "learning_rate": 0.00019999998639682654,
-      "loss": 46.0,
-      "step": 1041
-    },
-    {
-      "epoch": 0.1678006360964612,
-      "grad_norm": 0.0004293345846235752,
-      "learning_rate": 0.00019999998637042544,
-      "loss": 46.0,
-      "step": 1042
-    },
-    {
-      "epoch": 0.1679616731752486,
-      "grad_norm": 0.0008935158839449286,
-      "learning_rate": 0.0001999999863439987,
-      "loss": 46.0,
-      "step": 1043
-    },
-    {
-      "epoch": 0.168122710254036,
-      "grad_norm": 0.0006440468714572489,
-      "learning_rate": 0.00019999998631754643,
-      "loss": 46.0,
-      "step": 1044
-    },
-    {
-      "epoch": 0.1682837473328234,
-      "grad_norm": 0.0007334767142310739,
-      "learning_rate": 0.00019999998629106853,
-      "loss": 46.0,
-      "step": 1045
-    },
-    {
-      "epoch": 0.16844478441161076,
-      "grad_norm": 0.0005875652423128486,
-      "learning_rate": 0.00019999998626456505,
-      "loss": 46.0,
-      "step": 1046
-    },
-    {
-      "epoch": 0.16860582149039816,
-      "grad_norm": 0.0005739011103287339,
-      "learning_rate": 0.00019999998623803595,
-      "loss": 46.0,
-      "step": 1047
-    },
-    {
-      "epoch": 0.16876685856918555,
-      "grad_norm": 0.0009351703920401633,
-      "learning_rate": 0.0001999999862114813,
-      "loss": 46.0,
-      "step": 1048
-    },
-    {
-      "epoch": 0.16892789564797295,
-      "grad_norm": 0.0002408776490483433,
-      "learning_rate": 0.00019999998618490102,
-      "loss": 46.0,
-      "step": 1049
-    },
-    {
-      "epoch": 0.16908893272676034,
-      "grad_norm": 0.0005334711167961359,
-      "learning_rate": 0.00019999998615829515,
-      "loss": 46.0,
-      "step": 1050
-    },
-    {
-      "epoch": 0.16924996980554774,
-      "grad_norm": 0.00042552888044156134,
-      "learning_rate": 0.00019999998613166368,
-      "loss": 46.0,
-      "step": 1051
-    },
-    {
-      "epoch": 0.1694110068843351,
-      "grad_norm": 0.0005725707160308957,
-      "learning_rate": 0.00019999998610500662,
-      "loss": 46.0,
-      "step": 1052
-    },
-    {
-      "epoch": 0.1695720439631225,
-      "grad_norm": 0.0008814763859845698,
-      "learning_rate": 0.00019999998607832397,
-      "loss": 46.0,
-      "step": 1053
-    },
-    {
-      "epoch": 0.1697330810419099,
-      "grad_norm": 0.0015081973979249597,
-      "learning_rate": 0.00019999998605161573,
-      "loss": 46.0,
-      "step": 1054
-    },
-    {
-      "epoch": 0.1698941181206973,
-      "grad_norm": 0.0009094245033338666,
-      "learning_rate": 0.0001999999860248819,
-      "loss": 46.0,
-      "step": 1055
-    },
-    {
-      "epoch": 0.1700551551994847,
-      "grad_norm": 0.0003089603269472718,
-      "learning_rate": 0.00019999998599812247,
-      "loss": 46.0,
-      "step": 1056
-    },
-    {
-      "epoch": 0.17021619227827206,
-      "grad_norm": 0.0007297676638700068,
-      "learning_rate": 0.00019999998597133742,
-      "loss": 46.0,
-      "step": 1057
-    },
-    {
-      "epoch": 0.17037722935705946,
-      "grad_norm": 0.000567425973713398,
-      "learning_rate": 0.00019999998594452683,
-      "loss": 46.0,
-      "step": 1058
-    },
-    {
-      "epoch": 0.17053826643584685,
-      "grad_norm": 0.0005964718293398619,
-      "learning_rate": 0.0001999999859176906,
-      "loss": 46.0,
-      "step": 1059
-    },
-    {
-      "epoch": 0.17069930351463425,
-      "grad_norm": 0.0003678709326777607,
-      "learning_rate": 0.00019999998589082877,
-      "loss": 46.0,
-      "step": 1060
-    },
-    {
-      "epoch": 0.17086034059342164,
-      "grad_norm": 0.0006606096285395324,
-      "learning_rate": 0.00019999998586394137,
-      "loss": 46.0,
-      "step": 1061
-    },
-    {
-      "epoch": 0.171021377672209,
-      "grad_norm": 0.0007891838322393596,
-      "learning_rate": 0.00019999998583702838,
-      "loss": 46.0,
-      "step": 1062
-    },
-    {
-      "epoch": 0.1711824147509964,
-      "grad_norm": 0.0009180408087559044,
-      "learning_rate": 0.00019999998581008978,
-      "loss": 46.0,
-      "step": 1063
-    },
-    {
-      "epoch": 0.1713434518297838,
-      "grad_norm": 0.0003136332961730659,
-      "learning_rate": 0.0001999999857831256,
-      "loss": 46.0,
-      "step": 1064
-    },
-    {
-      "epoch": 0.1715044889085712,
-      "grad_norm": 0.0006892234669066966,
-      "learning_rate": 0.00019999998575613578,
-      "loss": 46.0,
-      "step": 1065
-    },
-    {
-      "epoch": 0.1716655259873586,
-      "grad_norm": 0.0008585926843807101,
-      "learning_rate": 0.00019999998572912042,
-      "loss": 46.0,
-      "step": 1066
-    },
-    {
-      "epoch": 0.171826563066146,
-      "grad_norm": 0.000655372510664165,
-      "learning_rate": 0.00019999998570207944,
-      "loss": 46.0,
-      "step": 1067
-    },
-    {
-      "epoch": 0.17198760014493336,
-      "grad_norm": 0.0003225669206585735,
-      "learning_rate": 0.00019999998567501285,
-      "loss": 46.0,
-      "step": 1068
-    },
-    {
-      "epoch": 0.17214863722372076,
-      "grad_norm": 0.0006178176845423877,
-      "learning_rate": 0.0001999999856479207,
-      "loss": 46.0,
-      "step": 1069
-    },
-    {
-      "epoch": 0.17230967430250815,
-      "grad_norm": 0.00044223055010661483,
-      "learning_rate": 0.00019999998562080293,
-      "loss": 46.0,
-      "step": 1070
-    },
-    {
-      "epoch": 0.17247071138129555,
-      "grad_norm": 0.0013964077224954963,
-      "learning_rate": 0.00019999998559365958,
-      "loss": 46.0,
-      "step": 1071
-    },
-    {
-      "epoch": 0.17263174846008295,
-      "grad_norm": 0.001068666111677885,
-      "learning_rate": 0.0001999999855664906,
-      "loss": 46.0,
-      "step": 1072
-    },
-    {
-      "epoch": 0.1727927855388703,
-      "grad_norm": 0.0015525006456300616,
-      "learning_rate": 0.00019999998553929605,
-      "loss": 46.0,
-      "step": 1073
-    },
-    {
-      "epoch": 0.1729538226176577,
-      "grad_norm": 0.00045818460057489574,
-      "learning_rate": 0.00019999998551207594,
-      "loss": 46.0,
-      "step": 1074
-    },
-    {
-      "epoch": 0.1731148596964451,
-      "grad_norm": 0.0006378164980560541,
-      "learning_rate": 0.00019999998548483018,
-      "loss": 46.0,
-      "step": 1075
-    },
-    {
-      "epoch": 0.1732758967752325,
-      "grad_norm": 0.0015530576929450035,
-      "learning_rate": 0.00019999998545755884,
-      "loss": 46.0,
-      "step": 1076
-    },
-    {
-      "epoch": 0.1734369338540199,
-      "grad_norm": 0.0005802001105621457,
-      "learning_rate": 0.00019999998543026193,
-      "loss": 46.0,
-      "step": 1077
-    },
-    {
-      "epoch": 0.17359797093280727,
-      "grad_norm": 0.0009448960772715509,
-      "learning_rate": 0.00019999998540293942,
-      "loss": 46.0,
-      "step": 1078
-    },
-    {
-      "epoch": 0.17375900801159466,
-      "grad_norm": 0.0003484654880594462,
-      "learning_rate": 0.0001999999853755913,
-      "loss": 46.0,
-      "step": 1079
-    },
-    {
-      "epoch": 0.17392004509038206,
-      "grad_norm": 0.000427607330493629,
-      "learning_rate": 0.0001999999853482176,
-      "loss": 46.0,
-      "step": 1080
-    },
-    {
-      "epoch": 0.17408108216916945,
-      "grad_norm": 0.0005763288354501128,
-      "learning_rate": 0.00019999998532081828,
-      "loss": 46.0,
-      "step": 1081
-    },
-    {
-      "epoch": 0.17424211924795685,
-      "grad_norm": 0.000552642741240561,
-      "learning_rate": 0.00019999998529339336,
-      "loss": 46.0,
-      "step": 1082
-    },
-    {
-      "epoch": 0.17440315632674425,
-      "grad_norm": 0.0006849284982308745,
-      "learning_rate": 0.00019999998526594288,
-      "loss": 46.0,
-      "step": 1083
-    },
-    {
-      "epoch": 0.17456419340553161,
-      "grad_norm": 0.0011226956266909838,
-      "learning_rate": 0.0001999999852384668,
-      "loss": 46.0,
-      "step": 1084
-    },
-    {
-      "epoch": 0.174725230484319,
-      "grad_norm": 0.00029608685872517526,
-      "learning_rate": 0.0001999999852109651,
-      "loss": 46.0,
-      "step": 1085
-    },
-    {
-      "epoch": 0.1748862675631064,
-      "grad_norm": 0.00017822741938289255,
-      "learning_rate": 0.0001999999851834378,
-      "loss": 46.0,
-      "step": 1086
-    },
-    {
-      "epoch": 0.1750473046418938,
-      "grad_norm": 0.0009377987589687109,
-      "learning_rate": 0.00019999998515588495,
-      "loss": 46.0,
-      "step": 1087
-    },
-    {
-      "epoch": 0.1752083417206812,
-      "grad_norm": 0.001035149092786014,
-      "learning_rate": 0.00019999998512830648,
-      "loss": 46.0,
-      "step": 1088
-    },
-    {
-      "epoch": 0.17536937879946857,
-      "grad_norm": 0.0003625023819040507,
-      "learning_rate": 0.00019999998510070242,
-      "loss": 46.0,
-      "step": 1089
-    },
-    {
-      "epoch": 0.17553041587825596,
-      "grad_norm": 0.0008139334968291223,
-      "learning_rate": 0.00019999998507307275,
-      "loss": 46.0,
-      "step": 1090
-    },
-    {
-      "epoch": 0.17569145295704336,
-      "grad_norm": 0.00025367780472151935,
-      "learning_rate": 0.00019999998504541752,
-      "loss": 46.0,
-      "step": 1091
-    },
-    {
-      "epoch": 0.17585249003583076,
-      "grad_norm": 0.000908449525013566,
-      "learning_rate": 0.00019999998501773664,
-      "loss": 46.0,
-      "step": 1092
-    },
-    {
-      "epoch": 0.17601352711461815,
-      "grad_norm": 0.0009912943933159113,
-      "learning_rate": 0.0001999999849900302,
-      "loss": 46.0,
-      "step": 1093
-    },
-    {
-      "epoch": 0.17617456419340552,
-      "grad_norm": 0.0005765684763900936,
-      "learning_rate": 0.00019999998496229816,
-      "loss": 46.0,
-      "step": 1094
-    },
-    {
-      "epoch": 0.17633560127219292,
-      "grad_norm": 0.0009983826894313097,
-      "learning_rate": 0.00019999998493454052,
-      "loss": 46.0,
-      "step": 1095
-    },
-    {
-      "epoch": 0.1764966383509803,
-      "grad_norm": 0.001541944220662117,
-      "learning_rate": 0.0001999999849067573,
-      "loss": 46.0,
-      "step": 1096
-    },
-    {
-      "epoch": 0.1766576754297677,
-      "grad_norm": 0.0009314665803685784,
-      "learning_rate": 0.0001999999848789485,
-      "loss": 46.0,
-      "step": 1097
-    },
-    {
-      "epoch": 0.1768187125085551,
-      "grad_norm": 0.0005268935929052532,
-      "learning_rate": 0.00019999998485111407,
-      "loss": 46.0,
-      "step": 1098
-    },
-    {
-      "epoch": 0.17697974958734247,
-      "grad_norm": 0.0011514154030010104,
-      "learning_rate": 0.00019999998482325405,
-      "loss": 46.0,
-      "step": 1099
-    },
-    {
-      "epoch": 0.17714078666612987,
-      "grad_norm": 0.0004244883602950722,
-      "learning_rate": 0.00019999998479536843,
-      "loss": 46.0,
-      "step": 1100
-    },
-    {
-      "epoch": 0.17730182374491726,
-      "grad_norm": 0.001184743014164269,
-      "learning_rate": 0.00019999998476745722,
-      "loss": 46.0,
-      "step": 1101
-    },
-    {
-      "epoch": 0.17746286082370466,
-      "grad_norm": 0.0012123126070946455,
-      "learning_rate": 0.00019999998473952044,
-      "loss": 46.0,
-      "step": 1102
-    },
-    {
-      "epoch": 0.17762389790249206,
-      "grad_norm": 0.000795738014858216,
-      "learning_rate": 0.00019999998471155803,
-      "loss": 46.0,
-      "step": 1103
-    },
-    {
-      "epoch": 0.17778493498127945,
-      "grad_norm": 0.0005148123600520194,
-      "learning_rate": 0.00019999998468357005,
-      "loss": 46.0,
-      "step": 1104
-    },
-    {
-      "epoch": 0.17794597206006682,
-      "grad_norm": 0.0003261953534092754,
-      "learning_rate": 0.00019999998465555647,
-      "loss": 46.0,
-      "step": 1105
-    },
-    {
-      "epoch": 0.17810700913885422,
-      "grad_norm": 0.0004914383171126246,
-      "learning_rate": 0.0001999999846275173,
-      "loss": 46.0,
-      "step": 1106
-    },
-    {
-      "epoch": 0.1782680462176416,
-      "grad_norm": 0.0007221323903650045,
-      "learning_rate": 0.00019999998459945253,
-      "loss": 46.0,
-      "step": 1107
-    },
-    {
-      "epoch": 0.178429083296429,
-      "grad_norm": 0.0003358956892043352,
-      "learning_rate": 0.00019999998457136215,
-      "loss": 46.0,
-      "step": 1108
-    },
-    {
-      "epoch": 0.1785901203752164,
-      "grad_norm": 0.0006616240134462714,
-      "learning_rate": 0.0001999999845432462,
-      "loss": 46.0,
-      "step": 1109
-    },
-    {
-      "epoch": 0.17875115745400377,
-      "grad_norm": 0.0006990287220105529,
-      "learning_rate": 0.00019999998451510463,
-      "loss": 46.0,
-      "step": 1110
-    },
-    {
-      "epoch": 0.17891219453279117,
-      "grad_norm": 0.0008044660207815468,
-      "learning_rate": 0.0001999999844869375,
-      "loss": 46.0,
-      "step": 1111
-    },
-    {
-      "epoch": 0.17907323161157856,
-      "grad_norm": 0.00043473043479025364,
-      "learning_rate": 0.00019999998445874474,
-      "loss": 46.0,
-      "step": 1112
-    },
-    {
-      "epoch": 0.17923426869036596,
-      "grad_norm": 0.0006207488477230072,
-      "learning_rate": 0.00019999998443052637,
-      "loss": 46.0,
-      "step": 1113
-    },
-    {
-      "epoch": 0.17939530576915336,
-      "grad_norm": 0.0006231630686670542,
-      "learning_rate": 0.00019999998440228245,
-      "loss": 46.0,
-      "step": 1114
-    },
-    {
-      "epoch": 0.17955634284794073,
-      "grad_norm": 0.0007432869169861078,
-      "learning_rate": 0.00019999998437401293,
-      "loss": 46.0,
-      "step": 1115
-    },
-    {
-      "epoch": 0.17971737992672812,
-      "grad_norm": 0.0006049188668839633,
-      "learning_rate": 0.0001999999843457178,
-      "loss": 46.0,
-      "step": 1116
-    },
-    {
-      "epoch": 0.17987841700551552,
-      "grad_norm": 0.000956145056989044,
-      "learning_rate": 0.0001999999843173971,
-      "loss": 46.0,
-      "step": 1117
-    },
-    {
-      "epoch": 0.1800394540843029,
-      "grad_norm": 0.0004712227964773774,
-      "learning_rate": 0.00019999998428905076,
-      "loss": 46.0,
-      "step": 1118
-    },
-    {
-      "epoch": 0.1802004911630903,
-      "grad_norm": 0.0006539911846630275,
-      "learning_rate": 0.00019999998426067884,
-      "loss": 46.0,
-      "step": 1119
-    },
-    {
-      "epoch": 0.1803615282418777,
-      "grad_norm": 0.001276426948606968,
-      "learning_rate": 0.00019999998423228134,
-      "loss": 46.0,
-      "step": 1120
-    },
-    {
-      "epoch": 0.18052256532066507,
-      "grad_norm": 0.001008278806693852,
-      "learning_rate": 0.00019999998420385824,
-      "loss": 46.0,
-      "step": 1121
-    },
-    {
-      "epoch": 0.18068360239945247,
-      "grad_norm": 0.001220381585881114,
-      "learning_rate": 0.00019999998417540954,
-      "loss": 46.0,
-      "step": 1122
-    },
-    {
-      "epoch": 0.18084463947823987,
-      "grad_norm": 0.0005268471431918442,
-      "learning_rate": 0.00019999998414693527,
-      "loss": 46.0,
-      "step": 1123
-    },
-    {
-      "epoch": 0.18100567655702726,
-      "grad_norm": 0.002254362916573882,
-      "learning_rate": 0.00019999998411843536,
-      "loss": 46.0,
-      "step": 1124
-    },
-    {
-      "epoch": 0.18116671363581466,
-      "grad_norm": 0.0006368353497236967,
-      "learning_rate": 0.0001999999840899099,
-      "loss": 46.0,
-      "step": 1125
-    },
-    {
-      "epoch": 0.18132775071460203,
-      "grad_norm": 0.0004503419331740588,
-      "learning_rate": 0.00019999998406135881,
-      "loss": 46.0,
-      "step": 1126
-    },
-    {
-      "epoch": 0.18148878779338942,
-      "grad_norm": 0.00037043457268737257,
-      "learning_rate": 0.00019999998403278215,
-      "loss": 46.0,
-      "step": 1127
-    },
-    {
-      "epoch": 0.18164982487217682,
-      "grad_norm": 0.0008587737684138119,
-      "learning_rate": 0.00019999998400417986,
-      "loss": 46.0,
-      "step": 1128
-    },
-    {
-      "epoch": 0.18181086195096421,
-      "grad_norm": 0.0005351615254767239,
-      "learning_rate": 0.00019999998397555202,
-      "loss": 46.0,
-      "step": 1129
-    },
-    {
-      "epoch": 0.1819718990297516,
-      "grad_norm": 0.0004402895283419639,
-      "learning_rate": 0.0001999999839468986,
-      "loss": 46.0,
-      "step": 1130
-    },
-    {
-      "epoch": 0.18213293610853898,
-      "grad_norm": 0.0004181397089269012,
-      "learning_rate": 0.00019999998391821952,
-      "loss": 46.0,
-      "step": 1131
-    },
-    {
-      "epoch": 0.18229397318732637,
-      "grad_norm": 0.0014928559539839625,
-      "learning_rate": 0.00019999998388951488,
-      "loss": 46.0,
-      "step": 1132
-    },
-    {
-      "epoch": 0.18245501026611377,
-      "grad_norm": 0.0009738048538565636,
-      "learning_rate": 0.00019999998386078464,
-      "loss": 46.0,
-      "step": 1133
-    },
-    {
-      "epoch": 0.18261604734490117,
-      "grad_norm": 0.000441311567556113,
-      "learning_rate": 0.0001999999838320288,
-      "loss": 46.0,
-      "step": 1134
-    },
-    {
-      "epoch": 0.18277708442368856,
-      "grad_norm": 0.00042914546793326735,
-      "learning_rate": 0.00019999998380324738,
-      "loss": 46.0,
-      "step": 1135
-    },
-    {
-      "epoch": 0.18293812150247596,
-      "grad_norm": 0.001419546315446496,
-      "learning_rate": 0.00019999998377444035,
-      "loss": 46.0,
-      "step": 1136
-    },
-    {
-      "epoch": 0.18309915858126333,
-      "grad_norm": 0.0010176844662055373,
-      "learning_rate": 0.00019999998374560773,
-      "loss": 46.0,
-      "step": 1137
-    },
-    {
-      "epoch": 0.18326019566005072,
-      "grad_norm": 0.00042564354953356087,
-      "learning_rate": 0.00019999998371674952,
-      "loss": 46.0,
-      "step": 1138
-    },
-    {
-      "epoch": 0.18342123273883812,
-      "grad_norm": 0.000891973206307739,
-      "learning_rate": 0.0001999999836878657,
-      "loss": 46.0,
-      "step": 1139
-    },
-    {
-      "epoch": 0.18358226981762552,
-      "grad_norm": 0.0004178571398369968,
-      "learning_rate": 0.0001999999836589563,
-      "loss": 46.0,
-      "step": 1140
-    },
-    {
-      "epoch": 0.1837433068964129,
-      "grad_norm": 0.001021531643345952,
-      "learning_rate": 0.0001999999836300213,
-      "loss": 46.0,
-      "step": 1141
-    },
-    {
-      "epoch": 0.18390434397520028,
-      "grad_norm": 0.0007500174688175321,
-      "learning_rate": 0.00019999998360106073,
-      "loss": 46.0,
-      "step": 1142
-    },
-    {
-      "epoch": 0.18406538105398768,
-      "grad_norm": 0.000505536503624171,
-      "learning_rate": 0.00019999998357207456,
-      "loss": 46.0,
-      "step": 1143
-    },
-    {
-      "epoch": 0.18422641813277507,
-      "grad_norm": 0.0006568719400092959,
-      "learning_rate": 0.00019999998354306274,
-      "loss": 46.0,
-      "step": 1144
-    },
-    {
-      "epoch": 0.18438745521156247,
-      "grad_norm": 0.0003531081310939044,
-      "learning_rate": 0.00019999998351402537,
-      "loss": 46.0,
-      "step": 1145
-    },
-    {
-      "epoch": 0.18454849229034986,
-      "grad_norm": 0.0006274929619394243,
-      "learning_rate": 0.0001999999834849624,
-      "loss": 46.0,
-      "step": 1146
-    },
-    {
-      "epoch": 0.18470952936913723,
-      "grad_norm": 0.001293198554776609,
-      "learning_rate": 0.00019999998345587386,
-      "loss": 46.0,
-      "step": 1147
-    },
-    {
-      "epoch": 0.18487056644792463,
-      "grad_norm": 0.0007399072637781501,
-      "learning_rate": 0.00019999998342675967,
-      "loss": 46.0,
-      "step": 1148
-    },
-    {
-      "epoch": 0.18503160352671202,
-      "grad_norm": 0.0011488314485177398,
-      "learning_rate": 0.00019999998339761992,
-      "loss": 46.0,
-      "step": 1149
-    },
-    {
-      "epoch": 0.18519264060549942,
-      "grad_norm": 0.0017289138631895185,
-      "learning_rate": 0.00019999998336845458,
-      "loss": 46.0,
-      "step": 1150
-    },
-    {
-      "epoch": 0.18535367768428682,
-      "grad_norm": 0.00041705454350449145,
-      "learning_rate": 0.0001999999833392636,
-      "loss": 46.0,
-      "step": 1151
-    },
-    {
-      "epoch": 0.18551471476307418,
-      "grad_norm": 0.0011145389871671796,
-      "learning_rate": 0.00019999998331004707,
-      "loss": 46.0,
-      "step": 1152
-    },
-    {
-      "epoch": 0.18567575184186158,
-      "grad_norm": 0.0004316457489039749,
-      "learning_rate": 0.00019999998328080494,
-      "loss": 46.0,
-      "step": 1153
-    },
-    {
-      "epoch": 0.18583678892064898,
-      "grad_norm": 0.000757015950512141,
-      "learning_rate": 0.0001999999832515372,
-      "loss": 46.0,
-      "step": 1154
-    },
-    {
-      "epoch": 0.18599782599943637,
-      "grad_norm": 0.0006713431212119758,
-      "learning_rate": 0.0001999999832222439,
-      "loss": 46.0,
-      "step": 1155
-    },
-    {
-      "epoch": 0.18615886307822377,
-      "grad_norm": 0.0006979407626204193,
-      "learning_rate": 0.00019999998319292496,
-      "loss": 46.0,
-      "step": 1156
-    },
-    {
-      "epoch": 0.18631990015701116,
-      "grad_norm": 0.00021065879263915122,
-      "learning_rate": 0.00019999998316358046,
-      "loss": 46.0,
-      "step": 1157
-    },
-    {
-      "epoch": 0.18648093723579853,
-      "grad_norm": 0.0019219517707824707,
-      "learning_rate": 0.00019999998313421032,
-      "loss": 46.0,
-      "step": 1158
-    },
-    {
-      "epoch": 0.18664197431458593,
-      "grad_norm": 0.000829343858640641,
-      "learning_rate": 0.00019999998310481462,
-      "loss": 46.0,
-      "step": 1159
-    },
-    {
-      "epoch": 0.18680301139337332,
-      "grad_norm": 0.00031882067560218275,
-      "learning_rate": 0.00019999998307539333,
-      "loss": 46.0,
-      "step": 1160
-    },
-    {
-      "epoch": 0.18696404847216072,
-      "grad_norm": 0.000313698808895424,
-      "learning_rate": 0.00019999998304594642,
-      "loss": 46.0,
-      "step": 1161
-    },
-    {
-      "epoch": 0.18712508555094812,
-      "grad_norm": 0.0007100233342498541,
-      "learning_rate": 0.00019999998301647396,
-      "loss": 46.0,
-      "step": 1162
-    },
-    {
-      "epoch": 0.18728612262973549,
-      "grad_norm": 0.0006322844419628382,
-      "learning_rate": 0.00019999998298697585,
-      "loss": 46.0,
-      "step": 1163
-    },
-    {
-      "epoch": 0.18744715970852288,
-      "grad_norm": 0.0016928017139434814,
-      "learning_rate": 0.0001999999829574522,
-      "loss": 46.0,
-      "step": 1164
-    },
-    {
-      "epoch": 0.18760819678731028,
-      "grad_norm": 0.001201709033921361,
-      "learning_rate": 0.00019999998292790288,
-      "loss": 46.0,
-      "step": 1165
-    },
-    {
-      "epoch": 0.18776923386609767,
-      "grad_norm": 0.0006109843961894512,
-      "learning_rate": 0.00019999998289832804,
-      "loss": 46.0,
-      "step": 1166
-    },
-    {
-      "epoch": 0.18793027094488507,
-      "grad_norm": 0.0002959135454148054,
-      "learning_rate": 0.00019999998286872756,
-      "loss": 46.0,
-      "step": 1167
-    },
-    {
-      "epoch": 0.18809130802367244,
-      "grad_norm": 0.0003422789741307497,
-      "learning_rate": 0.00019999998283910152,
-      "loss": 46.0,
-      "step": 1168
-    },
-    {
-      "epoch": 0.18825234510245983,
-      "grad_norm": 0.0006142971687950194,
-      "learning_rate": 0.00019999998280944987,
-      "loss": 46.0,
-      "step": 1169
-    },
-    {
-      "epoch": 0.18841338218124723,
-      "grad_norm": 0.00048411451280117035,
-      "learning_rate": 0.00019999998277977262,
-      "loss": 46.0,
-      "step": 1170
-    },
-    {
-      "epoch": 0.18857441926003463,
-      "grad_norm": 0.0003633377200458199,
-      "learning_rate": 0.00019999998275006977,
-      "loss": 46.0,
-      "step": 1171
-    },
-    {
-      "epoch": 0.18873545633882202,
-      "grad_norm": 0.001585592282935977,
-      "learning_rate": 0.00019999998272034132,
-      "loss": 46.0,
-      "step": 1172
-    },
-    {
-      "epoch": 0.18889649341760942,
-      "grad_norm": 0.000719084870070219,
-      "learning_rate": 0.00019999998269058732,
-      "loss": 46.0,
-      "step": 1173
-    },
-    {
-      "epoch": 0.1890575304963968,
-      "grad_norm": 0.0004340797895565629,
-      "learning_rate": 0.00019999998266080767,
-      "loss": 46.0,
-      "step": 1174
-    },
-    {
-      "epoch": 0.18921856757518418,
-      "grad_norm": 0.0013872854178771377,
-      "learning_rate": 0.00019999998263100244,
-      "loss": 46.0,
-      "step": 1175
-    },
-    {
-      "epoch": 0.18937960465397158,
-      "grad_norm": 0.0007887212559580803,
-      "learning_rate": 0.00019999998260117162,
-      "loss": 46.0,
-      "step": 1176
-    },
-    {
-      "epoch": 0.18954064173275897,
-      "grad_norm": 0.0005852034664712846,
-      "learning_rate": 0.0001999999825713152,
-      "loss": 46.0,
-      "step": 1177
-    },
-    {
-      "epoch": 0.18970167881154637,
-      "grad_norm": 0.00038101040991023183,
-      "learning_rate": 0.0001999999825414332,
-      "loss": 46.0,
-      "step": 1178
-    },
-    {
-      "epoch": 0.18986271589033374,
-      "grad_norm": 0.00033767044078558683,
-      "learning_rate": 0.0001999999825115256,
-      "loss": 46.0,
-      "step": 1179
-    },
-    {
-      "epoch": 0.19002375296912113,
-      "grad_norm": 0.0005349040147848427,
-      "learning_rate": 0.0001999999824815924,
-      "loss": 46.0,
-      "step": 1180
-    },
-    {
-      "epoch": 0.19018479004790853,
-      "grad_norm": 0.000546657363884151,
-      "learning_rate": 0.0001999999824516336,
-      "loss": 46.0,
-      "step": 1181
-    },
-    {
-      "epoch": 0.19034582712669593,
-      "grad_norm": 0.0005343665834516287,
-      "learning_rate": 0.00019999998242164923,
-      "loss": 46.0,
-      "step": 1182
-    },
-    {
-      "epoch": 0.19050686420548332,
-      "grad_norm": 0.00037214389885775745,
-      "learning_rate": 0.00019999998239163925,
-      "loss": 46.0,
-      "step": 1183
-    },
-    {
-      "epoch": 0.1906679012842707,
-      "grad_norm": 0.0005355009925551713,
-      "learning_rate": 0.00019999998236160368,
-      "loss": 46.0,
-      "step": 1184
-    },
-    {
-      "epoch": 0.1908289383630581,
-      "grad_norm": 0.0005049539031460881,
-      "learning_rate": 0.0001999999823315425,
-      "loss": 46.0,
-      "step": 1185
-    },
-    {
-      "epoch": 0.19098997544184548,
-      "grad_norm": 0.0003507023211568594,
-      "learning_rate": 0.00019999998230145575,
-      "loss": 46.0,
-      "step": 1186
-    },
-    {
-      "epoch": 0.19115101252063288,
-      "grad_norm": 0.002438133815303445,
-      "learning_rate": 0.00019999998227134339,
-      "loss": 46.0,
-      "step": 1187
-    },
-    {
-      "epoch": 0.19131204959942028,
-      "grad_norm": 0.00040006329072639346,
-      "learning_rate": 0.00019999998224120538,
-      "loss": 46.0,
-      "step": 1188
-    },
-    {
-      "epoch": 0.19147308667820764,
-      "grad_norm": 0.0005568754859268665,
-      "learning_rate": 0.00019999998221104185,
-      "loss": 46.0,
-      "step": 1189
-    },
-    {
-      "epoch": 0.19163412375699504,
-      "grad_norm": 0.0005926373996771872,
-      "learning_rate": 0.0001999999821808527,
-      "loss": 46.0,
-      "step": 1190
-    },
-    {
-      "epoch": 0.19179516083578244,
-      "grad_norm": 0.000892647251021117,
-      "learning_rate": 0.00019999998215063797,
-      "loss": 46.0,
-      "step": 1191
-    },
-    {
-      "epoch": 0.19195619791456983,
-      "grad_norm": 0.0004454712907318026,
-      "learning_rate": 0.00019999998212039764,
-      "loss": 46.0,
-      "step": 1192
-    },
-    {
-      "epoch": 0.19211723499335723,
-      "grad_norm": 0.0007348660146817565,
-      "learning_rate": 0.0001999999820901317,
-      "loss": 46.0,
-      "step": 1193
-    },
-    {
-      "epoch": 0.19227827207214462,
-      "grad_norm": 0.0014717044541612267,
-      "learning_rate": 0.00019999998205984015,
-      "loss": 46.0,
-      "step": 1194
-    },
-    {
-      "epoch": 0.192439309150932,
-      "grad_norm": 0.0005465009598992765,
-      "learning_rate": 0.00019999998202952307,
-      "loss": 46.0,
-      "step": 1195
-    },
-    {
-      "epoch": 0.1926003462297194,
-      "grad_norm": 0.0009091340471059084,
-      "learning_rate": 0.00019999998199918034,
-      "loss": 46.0,
-      "step": 1196
-    },
-    {
-      "epoch": 0.19276138330850678,
-      "grad_norm": 0.0007343650213442743,
-      "learning_rate": 0.00019999998196881206,
-      "loss": 46.0,
-      "step": 1197
-    },
-    {
-      "epoch": 0.19292242038729418,
-      "grad_norm": 0.0005318308249115944,
-      "learning_rate": 0.00019999998193841813,
-      "loss": 46.0,
-      "step": 1198
-    },
-    {
-      "epoch": 0.19308345746608158,
-      "grad_norm": 0.00048128454363904893,
-      "learning_rate": 0.00019999998190799862,
-      "loss": 46.0,
-      "step": 1199
-    },
-    {
-      "epoch": 0.19324449454486894,
-      "grad_norm": 0.001856165356002748,
-      "learning_rate": 0.00019999998187755354,
-      "loss": 46.0,
-      "step": 1200
-    },
-    {
-      "epoch": 0.19340553162365634,
-      "grad_norm": 0.0003363307623658329,
-      "learning_rate": 0.00019999998184708285,
-      "loss": 46.0,
-      "step": 1201
-    },
-    {
-      "epoch": 0.19356656870244374,
-      "grad_norm": 0.0004334888653829694,
-      "learning_rate": 0.00019999998181658658,
-      "loss": 46.0,
-      "step": 1202
-    },
-    {
-      "epoch": 0.19372760578123113,
-      "grad_norm": 0.0006357750971801579,
-      "learning_rate": 0.00019999998178606466,
-      "loss": 46.0,
-      "step": 1203
-    },
-    {
-      "epoch": 0.19388864286001853,
-      "grad_norm": 0.0015225925017148256,
-      "learning_rate": 0.0001999999817555172,
-      "loss": 46.0,
-      "step": 1204
-    },
-    {
-      "epoch": 0.1940496799388059,
-      "grad_norm": 0.00039018222014419734,
-      "learning_rate": 0.00019999998172494414,
-      "loss": 46.0,
-      "step": 1205
-    },
-    {
-      "epoch": 0.1942107170175933,
-      "grad_norm": 0.0012539116432890296,
-      "learning_rate": 0.00019999998169434546,
-      "loss": 46.0,
-      "step": 1206
-    },
-    {
-      "epoch": 0.1943717540963807,
-      "grad_norm": 0.0006164974183775485,
-      "learning_rate": 0.0001999999816637212,
-      "loss": 46.0,
-      "step": 1207
-    },
-    {
-      "epoch": 0.19453279117516808,
-      "grad_norm": 0.0003595839662011713,
-      "learning_rate": 0.00019999998163307137,
-      "loss": 46.0,
-      "step": 1208
-    },
-    {
-      "epoch": 0.19469382825395548,
-      "grad_norm": 0.001383608439937234,
-      "learning_rate": 0.0001999999816023959,
-      "loss": 46.0,
-      "step": 1209
-    },
-    {
-      "epoch": 0.19485486533274288,
-      "grad_norm": 0.0003669859142974019,
-      "learning_rate": 0.00019999998157169485,
-      "loss": 46.0,
-      "step": 1210
-    },
-    {
-      "epoch": 0.19501590241153025,
-      "grad_norm": 0.0006808997713960707,
-      "learning_rate": 0.00019999998154096823,
-      "loss": 46.0,
-      "step": 1211
-    },
-    {
-      "epoch": 0.19517693949031764,
-      "grad_norm": 0.0005866778665222228,
-      "learning_rate": 0.000199999981510216,
-      "loss": 46.0,
-      "step": 1212
-    },
-    {
-      "epoch": 0.19533797656910504,
-      "grad_norm": 0.0008350455318577588,
-      "learning_rate": 0.00019999998147943816,
-      "loss": 46.0,
-      "step": 1213
-    },
-    {
-      "epoch": 0.19549901364789243,
-      "grad_norm": 0.0006567607051692903,
-      "learning_rate": 0.00019999998144863473,
-      "loss": 46.0,
-      "step": 1214
-    },
-    {
-      "epoch": 0.19566005072667983,
-      "grad_norm": 0.0005408382276073098,
-      "learning_rate": 0.00019999998141780574,
-      "loss": 46.0,
-      "step": 1215
-    },
-    {
-      "epoch": 0.1958210878054672,
-      "grad_norm": 0.00021963795006740838,
-      "learning_rate": 0.0001999999813869511,
-      "loss": 46.0,
-      "step": 1216
-    },
-    {
-      "epoch": 0.1959821248842546,
-      "grad_norm": 0.0008670611423440278,
-      "learning_rate": 0.0001999999813560709,
-      "loss": 46.0,
-      "step": 1217
-    },
-    {
-      "epoch": 0.196143161963042,
-      "grad_norm": 0.0005301130004227161,
-      "learning_rate": 0.00019999998132516508,
-      "loss": 46.0,
-      "step": 1218
-    },
-    {
-      "epoch": 0.19630419904182939,
-      "grad_norm": 0.0004676864482462406,
-      "learning_rate": 0.0001999999812942337,
-      "loss": 46.0,
-      "step": 1219
-    },
-    {
-      "epoch": 0.19646523612061678,
-      "grad_norm": 0.0006992176058702171,
-      "learning_rate": 0.00019999998126327674,
-      "loss": 46.0,
-      "step": 1220
-    },
-    {
-      "epoch": 0.19662627319940415,
-      "grad_norm": 0.00042340063373558223,
-      "learning_rate": 0.00019999998123229414,
-      "loss": 46.0,
-      "step": 1221
-    },
-    {
-      "epoch": 0.19678731027819155,
-      "grad_norm": 0.0007891609566286206,
-      "learning_rate": 0.00019999998120128596,
-      "loss": 46.0,
-      "step": 1222
-    },
-    {
-      "epoch": 0.19694834735697894,
-      "grad_norm": 0.0007281369180418551,
-      "learning_rate": 0.00019999998117025216,
-      "loss": 46.0,
-      "step": 1223
-    },
-    {
-      "epoch": 0.19710938443576634,
-      "grad_norm": 0.0013537160120904446,
-      "learning_rate": 0.00019999998113919278,
-      "loss": 46.0,
-      "step": 1224
-    },
-    {
-      "epoch": 0.19727042151455373,
-      "grad_norm": 0.0006485064513981342,
-      "learning_rate": 0.00019999998110810784,
-      "loss": 46.0,
-      "step": 1225
-    },
-    {
-      "epoch": 0.19743145859334113,
-      "grad_norm": 0.0009129114914685488,
-      "learning_rate": 0.00019999998107699728,
-      "loss": 46.0,
-      "step": 1226
-    },
-    {
-      "epoch": 0.1975924956721285,
-      "grad_norm": 0.0018358170054852962,
-      "learning_rate": 0.0001999999810458611,
-      "loss": 46.0,
-      "step": 1227
-    },
-    {
-      "epoch": 0.1977535327509159,
-      "grad_norm": 0.00032996985828503966,
-      "learning_rate": 0.00019999998101469935,
-      "loss": 46.0,
-      "step": 1228
-    },
-    {
-      "epoch": 0.1979145698297033,
-      "grad_norm": 0.0010607584845274687,
-      "learning_rate": 0.00019999998098351203,
-      "loss": 46.0,
-      "step": 1229
-    },
-    {
-      "epoch": 0.1980756069084907,
-      "grad_norm": 0.0004708537016995251,
-      "learning_rate": 0.0001999999809522991,
-      "loss": 46.0,
-      "step": 1230
-    },
-    {
-      "epoch": 0.19823664398727808,
-      "grad_norm": 0.0007520978688262403,
-      "learning_rate": 0.00019999998092106055,
-      "loss": 46.0,
-      "step": 1231
-    },
-    {
-      "epoch": 0.19839768106606545,
-      "grad_norm": 0.000604619737714529,
-      "learning_rate": 0.0001999999808897964,
-      "loss": 46.0,
-      "step": 1232
-    },
-    {
-      "epoch": 0.19855871814485285,
-      "grad_norm": 0.0009547881199978292,
-      "learning_rate": 0.0001999999808585067,
-      "loss": 46.0,
-      "step": 1233
-    },
-    {
-      "epoch": 0.19871975522364024,
-      "grad_norm": 0.0008918555686250329,
-      "learning_rate": 0.00019999998082719135,
-      "loss": 46.0,
-      "step": 1234
-    },
-    {
-      "epoch": 0.19888079230242764,
-      "grad_norm": 0.00038530738675035536,
-      "learning_rate": 0.00019999998079585045,
-      "loss": 46.0,
-      "step": 1235
-    },
-    {
-      "epoch": 0.19904182938121504,
-      "grad_norm": 0.0005893002962693572,
-      "learning_rate": 0.00019999998076448394,
-      "loss": 46.0,
-      "step": 1236
-    },
-    {
-      "epoch": 0.1992028664600024,
-      "grad_norm": 0.0002431150496704504,
-      "learning_rate": 0.00019999998073309182,
-      "loss": 46.0,
-      "step": 1237
-    },
-    {
-      "epoch": 0.1993639035387898,
-      "grad_norm": 0.0013590792659670115,
-      "learning_rate": 0.00019999998070167413,
-      "loss": 46.0,
-      "step": 1238
-    },
-    {
-      "epoch": 0.1995249406175772,
-      "grad_norm": 0.0006564633222296834,
-      "learning_rate": 0.00019999998067023083,
-      "loss": 46.0,
-      "step": 1239
-    },
-    {
-      "epoch": 0.1996859776963646,
-      "grad_norm": 0.001041515963152051,
-      "learning_rate": 0.00019999998063876192,
-      "loss": 46.0,
-      "step": 1240
-    },
-    {
-      "epoch": 0.199847014775152,
-      "grad_norm": 0.0007092384621500969,
-      "learning_rate": 0.00019999998060726745,
-      "loss": 46.0,
-      "step": 1241
-    },
-    {
-      "epoch": 0.20000805185393936,
-      "grad_norm": 0.000286154419882223,
-      "learning_rate": 0.00019999998057574738,
-      "loss": 46.0,
-      "step": 1242
-    },
-    {
-      "epoch": 0.20016908893272675,
-      "grad_norm": 0.0004713652888312936,
-      "learning_rate": 0.0001999999805442017,
-      "loss": 46.0,
-      "step": 1243
-    },
-    {
-      "epoch": 0.20033012601151415,
-      "grad_norm": 0.0015317403012886643,
-      "learning_rate": 0.00019999998051263042,
-      "loss": 46.0,
-      "step": 1244
-    },
-    {
-      "epoch": 0.20049116309030154,
-      "grad_norm": 0.000713564979378134,
-      "learning_rate": 0.00019999998048103357,
-      "loss": 46.0,
-      "step": 1245
-    },
-    {
-      "epoch": 0.20065220016908894,
-      "grad_norm": 0.0008483220590278506,
-      "learning_rate": 0.00019999998044941113,
-      "loss": 46.0,
-      "step": 1246
-    },
-    {
-      "epoch": 0.20081323724787634,
-      "grad_norm": 0.001816748408600688,
-      "learning_rate": 0.00019999998041776305,
-      "loss": 46.0,
-      "step": 1247
-    },
-    {
-      "epoch": 0.2009742743266637,
-      "grad_norm": 0.0003024954639840871,
-      "learning_rate": 0.00019999998038608942,
-      "loss": 46.0,
-      "step": 1248
-    },
-    {
-      "epoch": 0.2011353114054511,
-      "grad_norm": 0.0003261783276684582,
-      "learning_rate": 0.00019999998035439016,
-      "loss": 46.0,
-      "step": 1249
-    },
-    {
-      "epoch": 0.2012963484842385,
-      "grad_norm": 0.0006194661254994571,
-      "learning_rate": 0.00019999998032266532,
-      "loss": 46.0,
-      "step": 1250
-    },
-    {
-      "epoch": 0.2014573855630259,
-      "grad_norm": 0.0017727031372487545,
-      "learning_rate": 0.0001999999802909149,
-      "loss": 46.0,
-      "step": 1251
-    },
-    {
-      "epoch": 0.2016184226418133,
-      "grad_norm": 0.0008676075958646834,
-      "learning_rate": 0.00019999998025913888,
-      "loss": 46.0,
-      "step": 1252
-    },
-    {
-      "epoch": 0.20177945972060066,
-      "grad_norm": 0.0007908233674243093,
-      "learning_rate": 0.00019999998022733725,
-      "loss": 46.0,
-      "step": 1253
-    },
-    {
-      "epoch": 0.20194049679938805,
-      "grad_norm": 0.0010229131439700723,
-      "learning_rate": 0.00019999998019551004,
-      "loss": 46.0,
-      "step": 1254
-    },
-    {
-      "epoch": 0.20210153387817545,
-      "grad_norm": 0.0005363317322917283,
-      "learning_rate": 0.0001999999801636572,
-      "loss": 46.0,
-      "step": 1255
-    },
-    {
-      "epoch": 0.20226257095696284,
-      "grad_norm": 0.00044931445154361427,
-      "learning_rate": 0.00019999998013177882,
-      "loss": 46.0,
-      "step": 1256
-    },
-    {
-      "epoch": 0.20242360803575024,
-      "grad_norm": 0.0004354381235316396,
-      "learning_rate": 0.00019999998009987481,
-      "loss": 46.0,
-      "step": 1257
-    },
-    {
-      "epoch": 0.2025846451145376,
-      "grad_norm": 0.0011329345870763063,
-      "learning_rate": 0.0001999999800679452,
-      "loss": 46.0,
-      "step": 1258
-    },
-    {
-      "epoch": 0.202745682193325,
-      "grad_norm": 0.0013573268661275506,
-      "learning_rate": 0.00019999998003599002,
-      "loss": 46.0,
-      "step": 1259
-    },
-    {
-      "epoch": 0.2029067192721124,
-      "grad_norm": 0.0004111604648642242,
-      "learning_rate": 0.00019999998000400923,
-      "loss": 46.0,
-      "step": 1260
-    },
-    {
-      "epoch": 0.2030677563508998,
-      "grad_norm": 0.00033504419843666255,
-      "learning_rate": 0.00019999997997200285,
-      "loss": 46.0,
-      "step": 1261
-    },
-    {
-      "epoch": 0.2032287934296872,
-      "grad_norm": 0.001046760007739067,
-      "learning_rate": 0.00019999997993997085,
-      "loss": 46.0,
-      "step": 1262
-    },
-    {
-      "epoch": 0.2033898305084746,
-      "grad_norm": 0.000282672350294888,
-      "learning_rate": 0.0001999999799079133,
-      "loss": 46.0,
-      "step": 1263
-    },
-    {
-      "epoch": 0.20355086758726196,
-      "grad_norm": 0.0005610276130028069,
-      "learning_rate": 0.00019999997987583013,
-      "loss": 46.0,
-      "step": 1264
-    },
-    {
-      "epoch": 0.20371190466604935,
-      "grad_norm": 0.0006805678131058812,
-      "learning_rate": 0.00019999997984372137,
-      "loss": 46.0,
-      "step": 1265
-    },
-    {
-      "epoch": 0.20387294174483675,
-      "grad_norm": 0.00043429923243820667,
-      "learning_rate": 0.00019999997981158703,
-      "loss": 46.0,
-      "step": 1266
-    },
-    {
-      "epoch": 0.20403397882362415,
-      "grad_norm": 0.0004495600878726691,
-      "learning_rate": 0.00019999997977942707,
-      "loss": 46.0,
-      "step": 1267
-    },
-    {
-      "epoch": 0.20419501590241154,
-      "grad_norm": 0.0007008801912888885,
-      "learning_rate": 0.00019999997974724153,
-      "loss": 46.0,
-      "step": 1268
-    },
-    {
-      "epoch": 0.2043560529811989,
-      "grad_norm": 0.0013177571818232536,
-      "learning_rate": 0.0001999999797150304,
-      "loss": 46.0,
-      "step": 1269
-    },
-    {
-      "epoch": 0.2045170900599863,
-      "grad_norm": 0.0006462169112637639,
-      "learning_rate": 0.00019999997968279365,
-      "loss": 46.0,
-      "step": 1270
-    },
-    {
-      "epoch": 0.2046781271387737,
-      "grad_norm": 0.0005713836289942265,
-      "learning_rate": 0.00019999997965053132,
-      "loss": 46.0,
-      "step": 1271
-    },
-    {
-      "epoch": 0.2048391642175611,
-      "grad_norm": 0.0008084340370260179,
-      "learning_rate": 0.0001999999796182434,
-      "loss": 46.0,
-      "step": 1272
-    },
-    {
-      "epoch": 0.2050002012963485,
-      "grad_norm": 0.003203538479283452,
-      "learning_rate": 0.0001999999795859299,
-      "loss": 46.0,
-      "step": 1273
-    },
-    {
-      "epoch": 0.20516123837513586,
-      "grad_norm": 0.0016390079399570823,
-      "learning_rate": 0.00019999997955359077,
-      "loss": 46.0,
-      "step": 1274
-    },
-    {
-      "epoch": 0.20532227545392326,
-      "grad_norm": 0.0006589957047253847,
-      "learning_rate": 0.00019999997952122603,
-      "loss": 46.0,
-      "step": 1275
-    },
-    {
-      "epoch": 0.20548331253271065,
-      "grad_norm": 0.0004350185627117753,
-      "learning_rate": 0.00019999997948883574,
-      "loss": 46.0,
-      "step": 1276
-    },
-    {
-      "epoch": 0.20564434961149805,
-      "grad_norm": 0.0006646730471402407,
-      "learning_rate": 0.00019999997945641983,
-      "loss": 46.0,
-      "step": 1277
-    },
-    {
-      "epoch": 0.20580538669028545,
-      "grad_norm": 0.0010368756484240294,
-      "learning_rate": 0.00019999997942397836,
-      "loss": 46.0,
-      "step": 1278
-    },
-    {
-      "epoch": 0.20596642376907284,
-      "grad_norm": 0.001357734901830554,
-      "learning_rate": 0.00019999997939151127,
-      "loss": 46.0,
-      "step": 1279
-    },
-    {
-      "epoch": 0.2061274608478602,
-      "grad_norm": 0.0006798878312110901,
-      "learning_rate": 0.00019999997935901857,
-      "loss": 46.0,
-      "step": 1280
-    },
-    {
-      "epoch": 0.2062884979266476,
-      "grad_norm": 0.0006989031098783016,
-      "learning_rate": 0.0001999999793265003,
-      "loss": 46.0,
-      "step": 1281
-    },
-    {
-      "epoch": 0.206449535005435,
-      "grad_norm": 0.0006288923905231059,
-      "learning_rate": 0.0001999999792939564,
-      "loss": 46.0,
-      "step": 1282
-    },
-    {
-      "epoch": 0.2066105720842224,
-      "grad_norm": 0.0011941434349864721,
-      "learning_rate": 0.00019999997926138695,
-      "loss": 46.0,
-      "step": 1283
-    },
-    {
-      "epoch": 0.2067716091630098,
-      "grad_norm": 0.00041245456668548286,
-      "learning_rate": 0.00019999997922879188,
-      "loss": 46.0,
-      "step": 1284
-    },
-    {
-      "epoch": 0.20693264624179716,
-      "grad_norm": 0.0009822394931688905,
-      "learning_rate": 0.00019999997919617124,
-      "loss": 46.0,
-      "step": 1285
-    },
-    {
-      "epoch": 0.20709368332058456,
-      "grad_norm": 0.0004020059132017195,
-      "learning_rate": 0.00019999997916352497,
-      "loss": 46.0,
-      "step": 1286
-    },
-    {
-      "epoch": 0.20725472039937196,
-      "grad_norm": 0.00035606411984190345,
-      "learning_rate": 0.00019999997913085313,
-      "loss": 46.0,
-      "step": 1287
-    },
-    {
-      "epoch": 0.20741575747815935,
-      "grad_norm": 0.0002625926863402128,
-      "learning_rate": 0.00019999997909815568,
-      "loss": 46.0,
-      "step": 1288
-    },
-    {
-      "epoch": 0.20757679455694675,
-      "grad_norm": 0.0003559159522410482,
-      "learning_rate": 0.00019999997906543264,
-      "loss": 46.0,
-      "step": 1289
-    },
-    {
-      "epoch": 0.20773783163573412,
-      "grad_norm": 0.00042531246435828507,
-      "learning_rate": 0.00019999997903268405,
-      "loss": 46.0,
-      "step": 1290
-    },
-    {
-      "epoch": 0.2078988687145215,
-      "grad_norm": 0.0011229532537981868,
-      "learning_rate": 0.0001999999789999098,
-      "loss": 46.0,
-      "step": 1291
-    },
-    {
-      "epoch": 0.2080599057933089,
-      "grad_norm": 0.0007212982163764536,
-      "learning_rate": 0.00019999997896710995,
-      "loss": 46.0,
-      "step": 1292
-    },
-    {
-      "epoch": 0.2082209428720963,
-      "grad_norm": 0.00037363465526141226,
-      "learning_rate": 0.00019999997893428454,
-      "loss": 46.0,
-      "step": 1293
-    },
-    {
-      "epoch": 0.2083819799508837,
-      "grad_norm": 0.0004636388330254704,
-      "learning_rate": 0.00019999997890143354,
-      "loss": 46.0,
-      "step": 1294
-    },
-    {
-      "epoch": 0.20854301702967107,
-      "grad_norm": 0.00041166823939420283,
-      "learning_rate": 0.00019999997886855692,
-      "loss": 46.0,
-      "step": 1295
-    },
-    {
-      "epoch": 0.20870405410845846,
-      "grad_norm": 0.0011840173974633217,
-      "learning_rate": 0.00019999997883565472,
-      "loss": 46.0,
-      "step": 1296
-    },
-    {
-      "epoch": 0.20886509118724586,
-      "grad_norm": 0.0007331250235438347,
-      "learning_rate": 0.00019999997880272693,
-      "loss": 46.0,
-      "step": 1297
-    },
-    {
-      "epoch": 0.20902612826603326,
-      "grad_norm": 0.0004854659200645983,
-      "learning_rate": 0.00019999997876977356,
-      "loss": 46.0,
-      "step": 1298
-    },
-    {
-      "epoch": 0.20918716534482065,
-      "grad_norm": 0.0006746966973878443,
-      "learning_rate": 0.00019999997873679454,
-      "loss": 46.0,
-      "step": 1299
-    },
-    {
-      "epoch": 0.20934820242360805,
-      "grad_norm": 0.00042606753413565457,
-      "learning_rate": 0.00019999997870378996,
-      "loss": 46.0,
-      "step": 1300
-    },
-    {
-      "epoch": 0.20950923950239542,
-      "grad_norm": 0.00040148888365365565,
-      "learning_rate": 0.00019999997867075982,
-      "loss": 46.0,
-      "step": 1301
-    },
-    {
-      "epoch": 0.2096702765811828,
-      "grad_norm": 0.0003798995167016983,
-      "learning_rate": 0.00019999997863770404,
-      "loss": 46.0,
-      "step": 1302
-    },
-    {
-      "epoch": 0.2098313136599702,
-      "grad_norm": 0.0013018834870308638,
-      "learning_rate": 0.00019999997860462268,
-      "loss": 46.0,
-      "step": 1303
-    },
-    {
-      "epoch": 0.2099923507387576,
-      "grad_norm": 0.0006651169969700277,
-      "learning_rate": 0.0001999999785715157,
-      "loss": 46.0,
-      "step": 1304
-    },
-    {
-      "epoch": 0.210153387817545,
-      "grad_norm": 0.0007748394273221493,
-      "learning_rate": 0.00019999997853838316,
-      "loss": 46.0,
-      "step": 1305
-    },
-    {
-      "epoch": 0.21031442489633237,
-      "grad_norm": 0.00041452041477896273,
-      "learning_rate": 0.00019999997850522497,
-      "loss": 46.0,
-      "step": 1306
-    },
-    {
-      "epoch": 0.21047546197511977,
-      "grad_norm": 0.0011760251363739371,
-      "learning_rate": 0.00019999997847204126,
-      "loss": 46.0,
-      "step": 1307
-    },
-    {
-      "epoch": 0.21063649905390716,
-      "grad_norm": 0.0009818397229537368,
-      "learning_rate": 0.0001999999784388319,
-      "loss": 46.0,
-      "step": 1308
-    },
-    {
-      "epoch": 0.21079753613269456,
-      "grad_norm": 0.000671629561111331,
-      "learning_rate": 0.00019999997840559699,
-      "loss": 46.0,
-      "step": 1309
-    },
-    {
-      "epoch": 0.21095857321148195,
-      "grad_norm": 0.0003047497302759439,
-      "learning_rate": 0.00019999997837233643,
-      "loss": 46.0,
-      "step": 1310
-    },
-    {
-      "epoch": 0.21111961029026932,
-      "grad_norm": 0.00030269133276306093,
-      "learning_rate": 0.0001999999783390503,
-      "loss": 46.0,
-      "step": 1311
-    },
-    {
-      "epoch": 0.21128064736905672,
-      "grad_norm": 0.0009661979274824262,
-      "learning_rate": 0.0001999999783057386,
-      "loss": 46.0,
-      "step": 1312
-    },
-    {
-      "epoch": 0.2114416844478441,
-      "grad_norm": 0.0008302502683363855,
-      "learning_rate": 0.00019999997827240129,
-      "loss": 46.0,
-      "step": 1313
-    },
-    {
-      "epoch": 0.2116027215266315,
-      "grad_norm": 0.0007613637717440724,
-      "learning_rate": 0.00019999997823903838,
-      "loss": 46.0,
-      "step": 1314
-    },
-    {
-      "epoch": 0.2117637586054189,
-      "grad_norm": 0.0009449763456359506,
-      "learning_rate": 0.00019999997820564986,
-      "loss": 46.0,
-      "step": 1315
-    },
-    {
-      "epoch": 0.2119247956842063,
-      "grad_norm": 0.0008858305518515408,
-      "learning_rate": 0.0001999999781722358,
-      "loss": 46.0,
-      "step": 1316
-    },
-    {
-      "epoch": 0.21208583276299367,
-      "grad_norm": 0.001000791322439909,
-      "learning_rate": 0.0001999999781387961,
-      "loss": 46.0,
-      "step": 1317
-    },
-    {
-      "epoch": 0.21224686984178107,
-      "grad_norm": 0.0008186783525161445,
-      "learning_rate": 0.00019999997810533077,
-      "loss": 46.0,
-      "step": 1318
-    },
-    {
-      "epoch": 0.21240790692056846,
-      "grad_norm": 0.002846318995580077,
-      "learning_rate": 0.00019999997807183993,
-      "loss": 46.0,
-      "step": 1319
-    },
-    {
-      "epoch": 0.21256894399935586,
-      "grad_norm": 0.000925748492591083,
-      "learning_rate": 0.00019999997803832345,
-      "loss": 46.0,
-      "step": 1320
-    },
-    {
-      "epoch": 0.21272998107814325,
-      "grad_norm": 0.0009930927772074938,
-      "learning_rate": 0.00019999997800478135,
-      "loss": 46.0,
-      "step": 1321
-    },
-    {
-      "epoch": 0.21289101815693062,
-      "grad_norm": 0.0015030791983008385,
-      "learning_rate": 0.00019999997797121372,
-      "loss": 46.0,
-      "step": 1322
-    },
-    {
-      "epoch": 0.21305205523571802,
-      "grad_norm": 0.0009468320058658719,
-      "learning_rate": 0.00019999997793762042,
-      "loss": 46.0,
-      "step": 1323
-    },
-    {
-      "epoch": 0.21321309231450541,
-      "grad_norm": 0.0008606510818935931,
-      "learning_rate": 0.00019999997790400156,
-      "loss": 46.0,
-      "step": 1324
-    },
-    {
-      "epoch": 0.2133741293932928,
-      "grad_norm": 0.0002674665302038193,
-      "learning_rate": 0.0001999999778703571,
-      "loss": 46.0,
-      "step": 1325
-    },
-    {
-      "epoch": 0.2135351664720802,
-      "grad_norm": 0.00042528240010142326,
-      "learning_rate": 0.00019999997783668708,
-      "loss": 46.0,
-      "step": 1326
-    },
-    {
-      "epoch": 0.21369620355086758,
-      "grad_norm": 0.000520424684509635,
-      "learning_rate": 0.00019999997780299143,
-      "loss": 46.0,
-      "step": 1327
-    },
-    {
-      "epoch": 0.21385724062965497,
-      "grad_norm": 0.0005076273810118437,
-      "learning_rate": 0.00019999997776927017,
-      "loss": 46.0,
-      "step": 1328
-    },
-    {
-      "epoch": 0.21401827770844237,
-      "grad_norm": 0.0007098311325535178,
-      "learning_rate": 0.00019999997773552334,
-      "loss": 46.0,
-      "step": 1329
-    },
-    {
-      "epoch": 0.21417931478722976,
-      "grad_norm": 0.001110561192035675,
-      "learning_rate": 0.00019999997770175093,
-      "loss": 46.0,
-      "step": 1330
-    },
-    {
-      "epoch": 0.21434035186601716,
-      "grad_norm": 0.002431141911074519,
-      "learning_rate": 0.0001999999776679529,
-      "loss": 46.0,
-      "step": 1331
-    },
-    {
-      "epoch": 0.21450138894480453,
-      "grad_norm": 0.0009852462681010365,
-      "learning_rate": 0.0001999999776341293,
-      "loss": 46.0,
-      "step": 1332
-    },
-    {
-      "epoch": 0.21466242602359192,
-      "grad_norm": 0.0011847583809867501,
-      "learning_rate": 0.00019999997760028007,
-      "loss": 46.0,
-      "step": 1333
-    },
-    {
-      "epoch": 0.21482346310237932,
-      "grad_norm": 0.0006205018726177514,
-      "learning_rate": 0.00019999997756640526,
-      "loss": 46.0,
-      "step": 1334
-    },
-    {
-      "epoch": 0.21498450018116672,
-      "grad_norm": 0.0006113162962719798,
-      "learning_rate": 0.00019999997753250486,
-      "loss": 46.0,
-      "step": 1335
-    },
-    {
-      "epoch": 0.2151455372599541,
-      "grad_norm": 0.0008273277198895812,
-      "learning_rate": 0.00019999997749857887,
-      "loss": 46.0,
-      "step": 1336
-    },
-    {
-      "epoch": 0.2153065743387415,
-      "grad_norm": 0.0008812635205686092,
-      "learning_rate": 0.0001999999774646273,
-      "loss": 46.0,
-      "step": 1337
-    },
-    {
-      "epoch": 0.21546761141752888,
-      "grad_norm": 0.0002859734231606126,
-      "learning_rate": 0.0001999999774306501,
-      "loss": 46.0,
-      "step": 1338
-    },
-    {
-      "epoch": 0.21562864849631627,
-      "grad_norm": 0.0016791780944913626,
-      "learning_rate": 0.0001999999773966473,
-      "loss": 46.0,
-      "step": 1339
-    },
-    {
-      "epoch": 0.21578968557510367,
-      "grad_norm": 0.0007860228652134538,
-      "learning_rate": 0.00019999997736261892,
-      "loss": 46.0,
-      "step": 1340
-    },
-    {
-      "epoch": 0.21595072265389106,
-      "grad_norm": 0.001189288916066289,
-      "learning_rate": 0.00019999997732856497,
-      "loss": 46.0,
-      "step": 1341
-    },
-    {
-      "epoch": 0.21611175973267846,
-      "grad_norm": 0.0004669920017477125,
-      "learning_rate": 0.00019999997729448538,
-      "loss": 46.0,
-      "step": 1342
-    },
-    {
-      "epoch": 0.21627279681146583,
-      "grad_norm": 0.00029777351301163435,
-      "learning_rate": 0.00019999997726038025,
-      "loss": 46.0,
-      "step": 1343
-    },
-    {
-      "epoch": 0.21643383389025322,
-      "grad_norm": 0.0005833743489347398,
-      "learning_rate": 0.00019999997722624949,
-      "loss": 46.0,
-      "step": 1344
-    },
-    {
-      "epoch": 0.21659487096904062,
-      "grad_norm": 0.0005694436840713024,
-      "learning_rate": 0.00019999997719209313,
-      "loss": 46.0,
-      "step": 1345
-    },
-    {
-      "epoch": 0.21675590804782802,
-      "grad_norm": 0.0004959586076438427,
-      "learning_rate": 0.0001999999771579112,
-      "loss": 46.0,
-      "step": 1346
-    },
-    {
-      "epoch": 0.2169169451266154,
-      "grad_norm": 0.0007592539768666029,
-      "learning_rate": 0.00019999997712370364,
-      "loss": 46.0,
-      "step": 1347
-    },
-    {
-      "epoch": 0.21707798220540278,
-      "grad_norm": 0.0009125350625254214,
-      "learning_rate": 0.00019999997708947052,
-      "loss": 46.0,
-      "step": 1348
-    },
-    {
-      "epoch": 0.21723901928419018,
-      "grad_norm": 0.0004352336982265115,
-      "learning_rate": 0.0001999999770552118,
-      "loss": 46.0,
-      "step": 1349
-    },
-    {
-      "epoch": 0.21740005636297757,
-      "grad_norm": 0.0004969755536876619,
-      "learning_rate": 0.00019999997702092745,
-      "loss": 46.0,
-      "step": 1350
-    },
-    {
-      "epoch": 0.21756109344176497,
-      "grad_norm": 0.0008792067528702319,
-      "learning_rate": 0.00019999997698661755,
-      "loss": 46.0,
-      "step": 1351
-    },
-    {
-      "epoch": 0.21772213052055236,
-      "grad_norm": 0.002492264611646533,
-      "learning_rate": 0.00019999997695228203,
-      "loss": 46.0,
-      "step": 1352
-    },
-    {
-      "epoch": 0.21788316759933976,
-      "grad_norm": 0.0009090930689126253,
-      "learning_rate": 0.00019999997691792093,
-      "loss": 46.0,
-      "step": 1353
-    },
-    {
-      "epoch": 0.21804420467812713,
-      "grad_norm": 0.0006379113183356822,
-      "learning_rate": 0.0001999999768835342,
-      "loss": 46.0,
-      "step": 1354
-    },
-    {
-      "epoch": 0.21820524175691453,
-      "grad_norm": 0.000359359139110893,
-      "learning_rate": 0.0001999999768491219,
-      "loss": 46.0,
-      "step": 1355
-    },
-    {
-      "epoch": 0.21836627883570192,
-      "grad_norm": 0.001027589780278504,
-      "learning_rate": 0.00019999997681468404,
-      "loss": 46.0,
-      "step": 1356
-    },
-    {
-      "epoch": 0.21852731591448932,
-      "grad_norm": 0.000839768152218312,
-      "learning_rate": 0.00019999997678022053,
-      "loss": 46.0,
-      "step": 1357
-    },
-    {
-      "epoch": 0.2186883529932767,
-      "grad_norm": 0.0015528691001236439,
-      "learning_rate": 0.00019999997674573146,
-      "loss": 46.0,
-      "step": 1358
-    },
-    {
-      "epoch": 0.21884939007206408,
-      "grad_norm": 0.00020052723994012922,
-      "learning_rate": 0.00019999997671121678,
-      "loss": 46.0,
-      "step": 1359
-    },
-    {
-      "epoch": 0.21901042715085148,
-      "grad_norm": 0.0004890299751423299,
-      "learning_rate": 0.0001999999766766765,
-      "loss": 46.0,
-      "step": 1360
-    },
-    {
-      "epoch": 0.21917146422963887,
-      "grad_norm": 0.0008463022531941533,
-      "learning_rate": 0.00019999997664211063,
-      "loss": 46.0,
-      "step": 1361
-    },
-    {
-      "epoch": 0.21933250130842627,
-      "grad_norm": 0.0009772107005119324,
-      "learning_rate": 0.00019999997660751916,
-      "loss": 46.0,
-      "step": 1362
-    },
-    {
-      "epoch": 0.21949353838721367,
-      "grad_norm": 0.0022033867426216602,
-      "learning_rate": 0.0001999999765729021,
-      "loss": 46.0,
-      "step": 1363
-    },
-    {
-      "epoch": 0.21965457546600103,
-      "grad_norm": 0.0009736161446198821,
-      "learning_rate": 0.00019999997653825946,
-      "loss": 46.0,
-      "step": 1364
-    },
-    {
-      "epoch": 0.21981561254478843,
-      "grad_norm": 0.002472112188115716,
-      "learning_rate": 0.00019999997650359123,
-      "loss": 46.0,
-      "step": 1365
-    },
-    {
-      "epoch": 0.21997664962357583,
-      "grad_norm": 0.0005026605795137584,
-      "learning_rate": 0.00019999997646889738,
-      "loss": 46.0,
-      "step": 1366
-    },
-    {
-      "epoch": 0.22013768670236322,
-      "grad_norm": 0.00116539909504354,
-      "learning_rate": 0.00019999997643417792,
-      "loss": 46.0,
-      "step": 1367
-    },
-    {
-      "epoch": 0.22029872378115062,
-      "grad_norm": 0.0004651888448279351,
-      "learning_rate": 0.0001999999763994329,
-      "loss": 46.0,
-      "step": 1368
-    },
-    {
-      "epoch": 0.22045976085993801,
-      "grad_norm": 0.0014744664076715708,
-      "learning_rate": 0.0001999999763646623,
-      "loss": 46.0,
-      "step": 1369
-    },
-    {
-      "epoch": 0.22062079793872538,
-      "grad_norm": 0.00045267806854099035,
-      "learning_rate": 0.00019999997632986604,
-      "loss": 46.0,
-      "step": 1370
-    },
-    {
-      "epoch": 0.22078183501751278,
-      "grad_norm": 0.000523657537996769,
-      "learning_rate": 0.00019999997629504423,
-      "loss": 46.0,
-      "step": 1371
-    },
-    {
-      "epoch": 0.22094287209630017,
-      "grad_norm": 0.0012504342012107372,
-      "learning_rate": 0.0001999999762601968,
-      "loss": 46.0,
-      "step": 1372
-    },
-    {
-      "epoch": 0.22110390917508757,
-      "grad_norm": 0.0005932914209552109,
-      "learning_rate": 0.00019999997622532383,
-      "loss": 46.0,
-      "step": 1373
-    },
-    {
-      "epoch": 0.22126494625387497,
-      "grad_norm": 0.0027143647894263268,
-      "learning_rate": 0.0001999999761904252,
-      "loss": 46.0,
-      "step": 1374
-    },
-    {
-      "epoch": 0.22142598333266234,
-      "grad_norm": 0.000593595381360501,
-      "learning_rate": 0.00019999997615550102,
-      "loss": 46.0,
-      "step": 1375
-    },
-    {
-      "epoch": 0.22158702041144973,
-      "grad_norm": 0.0009034270187839866,
-      "learning_rate": 0.00019999997612055124,
-      "loss": 46.0,
-      "step": 1376
-    },
-    {
-      "epoch": 0.22174805749023713,
-      "grad_norm": 0.0006523123593069613,
-      "learning_rate": 0.00019999997608557583,
-      "loss": 46.0,
-      "step": 1377
-    },
-    {
-      "epoch": 0.22190909456902452,
-      "grad_norm": 0.005185622721910477,
-      "learning_rate": 0.00019999997605057486,
-      "loss": 46.0,
-      "step": 1378
-    },
-    {
-      "epoch": 0.22207013164781192,
-      "grad_norm": 0.0007409183890558779,
-      "learning_rate": 0.0001999999760155483,
-      "loss": 46.0,
-      "step": 1379
-    },
-    {
-      "epoch": 0.2222311687265993,
-      "grad_norm": 0.0009297197684645653,
-      "learning_rate": 0.00019999997598049612,
-      "loss": 46.0,
-      "step": 1380
-    },
-    {
-      "epoch": 0.22239220580538668,
-      "grad_norm": 0.001152454991824925,
-      "learning_rate": 0.00019999997594541836,
-      "loss": 46.0,
-      "step": 1381
-    },
-    {
-      "epoch": 0.22255324288417408,
-      "grad_norm": 0.00031069456599652767,
-      "learning_rate": 0.000199999975910315,
-      "loss": 46.0,
-      "step": 1382
-    },
-    {
-      "epoch": 0.22271427996296148,
-      "grad_norm": 0.0006136883166618645,
-      "learning_rate": 0.00019999997587518605,
-      "loss": 46.0,
-      "step": 1383
-    },
-    {
-      "epoch": 0.22287531704174887,
-      "grad_norm": 0.001425751717761159,
-      "learning_rate": 0.0001999999758400315,
-      "loss": 46.0,
-      "step": 1384
-    },
-    {
-      "epoch": 0.22303635412053624,
-      "grad_norm": 0.0005713800201192498,
-      "learning_rate": 0.00019999997580485133,
-      "loss": 46.0,
-      "step": 1385
-    },
-    {
-      "epoch": 0.22319739119932364,
-      "grad_norm": 0.0012481024023145437,
-      "learning_rate": 0.0001999999757696456,
-      "loss": 46.0,
-      "step": 1386
-    },
-    {
-      "epoch": 0.22335842827811103,
-      "grad_norm": 0.0007147170254029334,
-      "learning_rate": 0.00019999997573441427,
-      "loss": 46.0,
-      "step": 1387
-    },
-    {
-      "epoch": 0.22351946535689843,
-      "grad_norm": 0.0009140543988905847,
-      "learning_rate": 0.00019999997569915734,
-      "loss": 46.0,
-      "step": 1388
-    },
-    {
-      "epoch": 0.22368050243568582,
-      "grad_norm": 0.0003687191638164222,
-      "learning_rate": 0.0001999999756638748,
-      "loss": 46.0,
-      "step": 1389
-    },
-    {
-      "epoch": 0.22384153951447322,
-      "grad_norm": 0.0006373602082021534,
-      "learning_rate": 0.0001999999756285667,
-      "loss": 46.0,
-      "step": 1390
-    },
-    {
-      "epoch": 0.2240025765932606,
-      "grad_norm": 0.0003636250621639192,
-      "learning_rate": 0.00019999997559323298,
-      "loss": 46.0,
-      "step": 1391
-    },
-    {
-      "epoch": 0.22416361367204798,
-      "grad_norm": 0.0018738931976258755,
-      "learning_rate": 0.00019999997555787368,
-      "loss": 46.0,
-      "step": 1392
-    },
-    {
-      "epoch": 0.22432465075083538,
-      "grad_norm": 0.00021359573293011636,
-      "learning_rate": 0.00019999997552248876,
-      "loss": 46.0,
-      "step": 1393
-    },
-    {
-      "epoch": 0.22448568782962278,
-      "grad_norm": 0.0005083411233499646,
-      "learning_rate": 0.00019999997548707826,
-      "loss": 46.0,
-      "step": 1394
-    },
-    {
-      "epoch": 0.22464672490841017,
-      "grad_norm": 0.0004379588644951582,
-      "learning_rate": 0.00019999997545164217,
-      "loss": 46.0,
-      "step": 1395
-    },
-    {
-      "epoch": 0.22480776198719754,
-      "grad_norm": 0.0023488837759941816,
-      "learning_rate": 0.0001999999754161805,
-      "loss": 46.0,
-      "step": 1396
-    },
-    {
-      "epoch": 0.22496879906598494,
-      "grad_norm": 0.001294159796088934,
-      "learning_rate": 0.0001999999753806932,
-      "loss": 46.0,
-      "step": 1397
-    },
-    {
-      "epoch": 0.22512983614477233,
-      "grad_norm": 0.000325981731293723,
-      "learning_rate": 0.00019999997534518032,
-      "loss": 46.0,
-      "step": 1398
-    },
-    {
-      "epoch": 0.22529087322355973,
-      "grad_norm": 0.0004426139057613909,
-      "learning_rate": 0.00019999997530964185,
-      "loss": 46.0,
-      "step": 1399
-    },
-    {
-      "epoch": 0.22545191030234712,
-      "grad_norm": 0.0008275063592009246,
-      "learning_rate": 0.0001999999752740778,
-      "loss": 46.0,
-      "step": 1400
-    },
-    {
-      "epoch": 0.2256129473811345,
-      "grad_norm": 0.000515216845087707,
-      "learning_rate": 0.00019999997523848813,
-      "loss": 46.0,
-      "step": 1401
-    },
-    {
-      "epoch": 0.2257739844599219,
-      "grad_norm": 0.00035010234569199383,
-      "learning_rate": 0.00019999997520287285,
-      "loss": 46.0,
-      "step": 1402
-    },
-    {
-      "epoch": 0.22593502153870929,
-      "grad_norm": 0.0015862504951655865,
-      "learning_rate": 0.000199999975167232,
-      "loss": 46.0,
-      "step": 1403
-    },
-    {
-      "epoch": 0.22609605861749668,
-      "grad_norm": 0.0005513600190170109,
-      "learning_rate": 0.00019999997513156558,
-      "loss": 46.0,
-      "step": 1404
-    },
-    {
-      "epoch": 0.22625709569628408,
-      "grad_norm": 0.001568580511957407,
-      "learning_rate": 0.0001999999750958735,
-      "loss": 46.0,
-      "step": 1405
-    },
-    {
-      "epoch": 0.22641813277507147,
-      "grad_norm": 0.002001952612772584,
-      "learning_rate": 0.00019999997506015587,
-      "loss": 46.0,
-      "step": 1406
-    },
-    {
-      "epoch": 0.22657916985385884,
-      "grad_norm": 0.0004378009762149304,
-      "learning_rate": 0.00019999997502441266,
-      "loss": 46.0,
-      "step": 1407
-    },
-    {
-      "epoch": 0.22674020693264624,
-      "grad_norm": 0.0003171944117639214,
-      "learning_rate": 0.00019999997498864382,
-      "loss": 46.0,
-      "step": 1408
-    },
-    {
-      "epoch": 0.22690124401143363,
-      "grad_norm": 0.0013251467607915401,
-      "learning_rate": 0.0001999999749528494,
-      "loss": 46.0,
-      "step": 1409
-    },
-    {
-      "epoch": 0.22706228109022103,
-      "grad_norm": 0.0007552897441200912,
-      "learning_rate": 0.0001999999749170294,
-      "loss": 46.0,
-      "step": 1410
-    },
-    {
-      "epoch": 0.22722331816900843,
-      "grad_norm": 0.0005726873641833663,
-      "learning_rate": 0.00019999997488118378,
-      "loss": 46.0,
-      "step": 1411
-    },
-    {
-      "epoch": 0.2273843552477958,
-      "grad_norm": 0.0004938723286613822,
-      "learning_rate": 0.00019999997484531257,
-      "loss": 46.0,
-      "step": 1412
-    },
-    {
-      "epoch": 0.2275453923265832,
-      "grad_norm": 0.00035536818904802203,
-      "learning_rate": 0.00019999997480941577,
-      "loss": 46.0,
-      "step": 1413
-    },
-    {
-      "epoch": 0.2277064294053706,
-      "grad_norm": 0.00028908287640661,
-      "learning_rate": 0.0001999999747734934,
-      "loss": 46.0,
-      "step": 1414
-    },
-    {
-      "epoch": 0.22786746648415798,
-      "grad_norm": 0.0012565568322315812,
-      "learning_rate": 0.0001999999747375454,
-      "loss": 46.0,
-      "step": 1415
-    },
-    {
-      "epoch": 0.22802850356294538,
-      "grad_norm": 0.0004596236685756594,
-      "learning_rate": 0.0001999999747015718,
-      "loss": 46.0,
-      "step": 1416
-    },
-    {
-      "epoch": 0.22818954064173275,
-      "grad_norm": 0.000986226019449532,
-      "learning_rate": 0.0001999999746655726,
-      "loss": 46.0,
-      "step": 1417
-    },
-    {
-      "epoch": 0.22835057772052014,
-      "grad_norm": 0.0005775231984443963,
-      "learning_rate": 0.00019999997462954785,
-      "loss": 46.0,
-      "step": 1418
-    },
-    {
-      "epoch": 0.22851161479930754,
-      "grad_norm": 0.000912361079826951,
-      "learning_rate": 0.00019999997459349745,
-      "loss": 46.0,
-      "step": 1419
-    },
-    {
-      "epoch": 0.22867265187809493,
-      "grad_norm": 0.00042985662003047764,
-      "learning_rate": 0.00019999997455742152,
-      "loss": 46.0,
-      "step": 1420
-    },
-    {
-      "epoch": 0.22883368895688233,
-      "grad_norm": 0.0005762381479144096,
-      "learning_rate": 0.00019999997452131994,
-      "loss": 46.0,
-      "step": 1421
-    },
-    {
-      "epoch": 0.22899472603566973,
-      "grad_norm": 0.0005960650159977376,
-      "learning_rate": 0.0001999999744851928,
-      "loss": 46.0,
-      "step": 1422
-    },
-    {
-      "epoch": 0.2291557631144571,
-      "grad_norm": 0.0006273018661886454,
-      "learning_rate": 0.00019999997444904003,
-      "loss": 46.0,
-      "step": 1423
-    },
-    {
-      "epoch": 0.2293168001932445,
-      "grad_norm": 0.00036709007690660655,
-      "learning_rate": 0.0001999999744128617,
-      "loss": 46.0,
-      "step": 1424
-    },
-    {
-      "epoch": 0.2294778372720319,
-      "grad_norm": 0.0005468791932798922,
-      "learning_rate": 0.00019999997437665775,
-      "loss": 46.0,
-      "step": 1425
-    },
-    {
-      "epoch": 0.22963887435081928,
-      "grad_norm": 0.0006789955659769475,
-      "learning_rate": 0.00019999997434042824,
-      "loss": 46.0,
-      "step": 1426
-    },
-    {
-      "epoch": 0.22979991142960668,
-      "grad_norm": 0.0006706491694785655,
-      "learning_rate": 0.0001999999743041731,
-      "loss": 46.0,
-      "step": 1427
-    },
-    {
-      "epoch": 0.22996094850839405,
-      "grad_norm": 0.0005731749115511775,
-      "learning_rate": 0.00019999997426789238,
-      "loss": 46.0,
-      "step": 1428
-    },
-    {
-      "epoch": 0.23012198558718144,
-      "grad_norm": 0.0005523664294742048,
-      "learning_rate": 0.00019999997423158608,
-      "loss": 46.0,
-      "step": 1429
-    },
-    {
-      "epoch": 0.23028302266596884,
-      "grad_norm": 0.000494952779263258,
-      "learning_rate": 0.00019999997419525416,
-      "loss": 46.0,
-      "step": 1430
-    },
-    {
-      "epoch": 0.23044405974475624,
-      "grad_norm": 0.0006475234404206276,
-      "learning_rate": 0.00019999997415889664,
-      "loss": 46.0,
-      "step": 1431
-    },
-    {
-      "epoch": 0.23060509682354363,
-      "grad_norm": 0.0004894636804237962,
-      "learning_rate": 0.00019999997412251355,
-      "loss": 46.0,
-      "step": 1432
-    },
-    {
-      "epoch": 0.230766133902331,
-      "grad_norm": 0.000386284664273262,
-      "learning_rate": 0.00019999997408610485,
-      "loss": 46.0,
-      "step": 1433
-    },
-    {
-      "epoch": 0.2309271709811184,
-      "grad_norm": 0.0005470376927405596,
-      "learning_rate": 0.00019999997404967056,
-      "loss": 46.0,
-      "step": 1434
-    },
-    {
-      "epoch": 0.2310882080599058,
-      "grad_norm": 0.0005402215174399316,
-      "learning_rate": 0.00019999997401321066,
-      "loss": 46.0,
-      "step": 1435
-    },
-    {
-      "epoch": 0.2312492451386932,
-      "grad_norm": 0.00040768837789073586,
-      "learning_rate": 0.0001999999739767252,
-      "loss": 46.0,
-      "step": 1436
-    },
-    {
-      "epoch": 0.23141028221748058,
-      "grad_norm": 0.0004655256343539804,
-      "learning_rate": 0.00019999997394021412,
-      "loss": 46.0,
-      "step": 1437
-    },
-    {
-      "epoch": 0.23157131929626795,
-      "grad_norm": 0.002065239707008004,
-      "learning_rate": 0.00019999997390367745,
-      "loss": 46.0,
-      "step": 1438
-    },
-    {
-      "epoch": 0.23173235637505535,
-      "grad_norm": 0.0010129864094778895,
-      "learning_rate": 0.00019999997386711515,
-      "loss": 46.0,
-      "step": 1439
-    },
-    {
-      "epoch": 0.23189339345384274,
-      "grad_norm": 0.0005043009296059608,
-      "learning_rate": 0.00019999997383052728,
-      "loss": 46.0,
-      "step": 1440
-    },
-    {
-      "epoch": 0.23205443053263014,
-      "grad_norm": 0.0005455774371512234,
-      "learning_rate": 0.00019999997379391385,
-      "loss": 46.0,
-      "step": 1441
-    },
-    {
-      "epoch": 0.23221546761141754,
-      "grad_norm": 0.0008954782970249653,
-      "learning_rate": 0.00019999997375727479,
-      "loss": 46.0,
-      "step": 1442
-    },
-    {
-      "epoch": 0.23237650469020493,
-      "grad_norm": 0.00048723258078098297,
-      "learning_rate": 0.00019999997372061016,
-      "loss": 46.0,
-      "step": 1443
-    },
-    {
-      "epoch": 0.2325375417689923,
-      "grad_norm": 0.000462414143839851,
-      "learning_rate": 0.0001999999736839199,
-      "loss": 46.0,
-      "step": 1444
-    },
-    {
-      "epoch": 0.2326985788477797,
-      "grad_norm": 0.00027866853633895516,
-      "learning_rate": 0.00019999997364720406,
-      "loss": 46.0,
-      "step": 1445
-    },
-    {
-      "epoch": 0.2328596159265671,
-      "grad_norm": 0.0005929471226409078,
-      "learning_rate": 0.00019999997361046264,
-      "loss": 46.0,
-      "step": 1446
-    },
-    {
-      "epoch": 0.2330206530053545,
-      "grad_norm": 0.0005723999929614365,
-      "learning_rate": 0.00019999997357369561,
-      "loss": 46.0,
-      "step": 1447
-    },
-    {
-      "epoch": 0.23318169008414188,
-      "grad_norm": 0.00035127042792737484,
-      "learning_rate": 0.000199999973536903,
-      "loss": 46.0,
-      "step": 1448
-    },
-    {
-      "epoch": 0.23334272716292925,
-      "grad_norm": 0.0011589425848796964,
-      "learning_rate": 0.00019999997350008476,
-      "loss": 46.0,
-      "step": 1449
-    },
-    {
-      "epoch": 0.23350376424171665,
-      "grad_norm": 0.0015349846798926592,
-      "learning_rate": 0.00019999997346324095,
-      "loss": 46.0,
-      "step": 1450
-    },
-    {
-      "epoch": 0.23366480132050405,
-      "grad_norm": 0.001190618029795587,
-      "learning_rate": 0.00019999997342637157,
-      "loss": 46.0,
-      "step": 1451
-    },
-    {
-      "epoch": 0.23382583839929144,
-      "grad_norm": 0.0009072727989405394,
-      "learning_rate": 0.00019999997338947655,
-      "loss": 46.0,
-      "step": 1452
-    },
-    {
-      "epoch": 0.23398687547807884,
-      "grad_norm": 0.0006187950493767858,
-      "learning_rate": 0.00019999997335255597,
-      "loss": 46.0,
-      "step": 1453
-    },
-    {
-      "epoch": 0.2341479125568662,
-      "grad_norm": 0.00041065565892495215,
-      "learning_rate": 0.00019999997331560974,
-      "loss": 46.0,
-      "step": 1454
-    },
-    {
-      "epoch": 0.2343089496356536,
-      "grad_norm": 0.0014450005255639553,
-      "learning_rate": 0.000199999973278638,
-      "loss": 46.0,
-      "step": 1455
-    },
-    {
-      "epoch": 0.234469986714441,
-      "grad_norm": 0.0004300459986552596,
-      "learning_rate": 0.00019999997324164062,
-      "loss": 46.0,
-      "step": 1456
-    },
-    {
-      "epoch": 0.2346310237932284,
-      "grad_norm": 0.000641741615254432,
-      "learning_rate": 0.00019999997320461764,
-      "loss": 46.0,
-      "step": 1457
-    },
-    {
-      "epoch": 0.2347920608720158,
-      "grad_norm": 0.0007824747590348125,
-      "learning_rate": 0.00019999997316756907,
-      "loss": 46.0,
-      "step": 1458
-    },
-    {
-      "epoch": 0.23495309795080319,
-      "grad_norm": 0.0004166323342360556,
-      "learning_rate": 0.0001999999731304949,
-      "loss": 46.0,
-      "step": 1459
-    },
-    {
-      "epoch": 0.23511413502959055,
-      "grad_norm": 0.0008934013312682509,
-      "learning_rate": 0.00019999997309339514,
-      "loss": 46.0,
-      "step": 1460
-    },
-    {
-      "epoch": 0.23527517210837795,
-      "grad_norm": 0.0007518811617046595,
-      "learning_rate": 0.00019999997305626978,
-      "loss": 46.0,
-      "step": 1461
-    },
-    {
-      "epoch": 0.23543620918716535,
-      "grad_norm": 0.0011339082848280668,
-      "learning_rate": 0.00019999997301911883,
-      "loss": 46.0,
-      "step": 1462
-    },
-    {
-      "epoch": 0.23559724626595274,
-      "grad_norm": 0.0009774438804015517,
-      "learning_rate": 0.0001999999729819423,
-      "loss": 46.0,
-      "step": 1463
-    },
-    {
-      "epoch": 0.23575828334474014,
-      "grad_norm": 0.0009151012054644525,
-      "learning_rate": 0.00019999997294474015,
-      "loss": 46.0,
-      "step": 1464
-    },
-    {
-      "epoch": 0.2359193204235275,
-      "grad_norm": 0.00034659350058063865,
-      "learning_rate": 0.00019999997290751241,
-      "loss": 46.0,
-      "step": 1465
-    },
-    {
-      "epoch": 0.2360803575023149,
-      "grad_norm": 0.0022008137311786413,
-      "learning_rate": 0.0001999999728702591,
-      "loss": 46.0,
-      "step": 1466
-    },
-    {
-      "epoch": 0.2362413945811023,
-      "grad_norm": 0.0004182037664577365,
-      "learning_rate": 0.00019999997283298018,
-      "loss": 46.0,
-      "step": 1467
-    },
-    {
-      "epoch": 0.2364024316598897,
-      "grad_norm": 0.0003482593165244907,
-      "learning_rate": 0.00019999997279567563,
-      "loss": 46.0,
-      "step": 1468
-    },
-    {
-      "epoch": 0.2365634687386771,
-      "grad_norm": 0.0003351763589307666,
-      "learning_rate": 0.00019999997275834552,
-      "loss": 46.0,
-      "step": 1469
-    },
-    {
-      "epoch": 0.23672450581746446,
-      "grad_norm": 0.001521661994047463,
-      "learning_rate": 0.00019999997272098982,
-      "loss": 46.0,
-      "step": 1470
-    },
-    {
-      "epoch": 0.23688554289625186,
-      "grad_norm": 0.0005002696998417377,
-      "learning_rate": 0.0001999999726836085,
-      "loss": 46.0,
-      "step": 1471
-    },
-    {
-      "epoch": 0.23704657997503925,
-      "grad_norm": 0.0005701422924175858,
-      "learning_rate": 0.00019999997264620164,
-      "loss": 46.0,
-      "step": 1472
-    },
-    {
-      "epoch": 0.23720761705382665,
-      "grad_norm": 0.0021671743597835302,
-      "learning_rate": 0.00019999997260876912,
-      "loss": 46.0,
-      "step": 1473
-    },
-    {
-      "epoch": 0.23736865413261404,
-      "grad_norm": 0.0006085827481001616,
-      "learning_rate": 0.00019999997257131105,
-      "loss": 46.0,
-      "step": 1474
-    },
-    {
-      "epoch": 0.2375296912114014,
-      "grad_norm": 0.0009190710843540728,
-      "learning_rate": 0.00019999997253382736,
-      "loss": 46.0,
-      "step": 1475
-    },
-    {
-      "epoch": 0.2376907282901888,
-      "grad_norm": 0.0007860172190703452,
-      "learning_rate": 0.0001999999724963181,
-      "loss": 46.0,
-      "step": 1476
-    },
-    {
-      "epoch": 0.2378517653689762,
-      "grad_norm": 0.0005892843473702669,
-      "learning_rate": 0.00019999997245878323,
-      "loss": 46.0,
-      "step": 1477
-    },
-    {
-      "epoch": 0.2380128024477636,
-      "grad_norm": 0.0005900952382944524,
-      "learning_rate": 0.00019999997242122275,
-      "loss": 46.0,
-      "step": 1478
-    },
-    {
-      "epoch": 0.238173839526551,
-      "grad_norm": 0.0016260374104604125,
-      "learning_rate": 0.00019999997238363668,
-      "loss": 46.0,
-      "step": 1479
-    },
-    {
-      "epoch": 0.2383348766053384,
-      "grad_norm": 0.0005029069725424051,
-      "learning_rate": 0.00019999997234602503,
-      "loss": 46.0,
-      "step": 1480
-    },
-    {
-      "epoch": 0.23849591368412576,
-      "grad_norm": 0.0009254749165847898,
-      "learning_rate": 0.00019999997230838777,
-      "loss": 46.0,
-      "step": 1481
-    },
-    {
-      "epoch": 0.23865695076291316,
-      "grad_norm": 0.0003914689878001809,
-      "learning_rate": 0.00019999997227072492,
-      "loss": 46.0,
-      "step": 1482
-    },
-    {
-      "epoch": 0.23881798784170055,
-      "grad_norm": 0.0007488206611014903,
-      "learning_rate": 0.00019999997223303648,
-      "loss": 46.0,
-      "step": 1483
-    },
-    {
-      "epoch": 0.23897902492048795,
-      "grad_norm": 0.0002851858444046229,
-      "learning_rate": 0.00019999997219532245,
-      "loss": 46.0,
-      "step": 1484
-    },
-    {
-      "epoch": 0.23914006199927534,
-      "grad_norm": 0.0003480883315205574,
-      "learning_rate": 0.0001999999721575828,
-      "loss": 46.0,
-      "step": 1485
-    },
-    {
-      "epoch": 0.2393010990780627,
-      "grad_norm": 0.0012112134136259556,
-      "learning_rate": 0.00019999997211981758,
-      "loss": 46.0,
-      "step": 1486
-    },
-    {
-      "epoch": 0.2394621361568501,
-      "grad_norm": 0.0008839130750857294,
-      "learning_rate": 0.0001999999720820268,
-      "loss": 46.0,
-      "step": 1487
-    },
-    {
-      "epoch": 0.2396231732356375,
-      "grad_norm": 0.0004892251454293728,
-      "learning_rate": 0.00019999997204421036,
-      "loss": 46.0,
-      "step": 1488
-    },
-    {
-      "epoch": 0.2397842103144249,
-      "grad_norm": 0.00029318995075300336,
-      "learning_rate": 0.00019999997200636832,
-      "loss": 46.0,
-      "step": 1489
-    },
-    {
-      "epoch": 0.2399452473932123,
-      "grad_norm": 0.0005369277787394822,
-      "learning_rate": 0.00019999997196850074,
-      "loss": 46.0,
-      "step": 1490
-    },
-    {
-      "epoch": 0.24010628447199966,
-      "grad_norm": 0.000695554306730628,
-      "learning_rate": 0.00019999997193060755,
-      "loss": 46.0,
-      "step": 1491
-    },
-    {
-      "epoch": 0.24026732155078706,
-      "grad_norm": 0.0009490074007771909,
-      "learning_rate": 0.00019999997189268872,
-      "loss": 46.0,
-      "step": 1492
-    },
-    {
-      "epoch": 0.24042835862957446,
-      "grad_norm": 0.0004491792933549732,
-      "learning_rate": 0.00019999997185474438,
-      "loss": 46.0,
-      "step": 1493
-    },
-    {
-      "epoch": 0.24058939570836185,
-      "grad_norm": 0.0015042545273900032,
-      "learning_rate": 0.00019999997181677437,
-      "loss": 46.0,
-      "step": 1494
-    },
-    {
-      "epoch": 0.24075043278714925,
-      "grad_norm": 0.002967716893181205,
-      "learning_rate": 0.00019999997177877878,
-      "loss": 46.0,
-      "step": 1495
-    },
-    {
-      "epoch": 0.24091146986593664,
-      "grad_norm": 0.0004654759541153908,
-      "learning_rate": 0.00019999997174075762,
-      "loss": 46.0,
-      "step": 1496
-    },
-    {
-      "epoch": 0.241072506944724,
-      "grad_norm": 0.0005797880003228784,
-      "learning_rate": 0.00019999997170271086,
-      "loss": 46.0,
-      "step": 1497
-    },
-    {
-      "epoch": 0.2412335440235114,
-      "grad_norm": 0.0014245264464989305,
-      "learning_rate": 0.00019999997166463847,
-      "loss": 46.0,
-      "step": 1498
-    },
-    {
-      "epoch": 0.2413945811022988,
-      "grad_norm": 0.0009832995710894465,
-      "learning_rate": 0.0001999999716265405,
-      "loss": 46.0,
-      "step": 1499
-    },
-    {
-      "epoch": 0.2415556181810862,
-      "grad_norm": 0.0005896422662772238,
-      "learning_rate": 0.00019999997158841697,
-      "loss": 46.0,
-      "step": 1500
-    },
-    {
-      "epoch": 0.2417166552598736,
-      "grad_norm": 0.000540532055310905,
-      "learning_rate": 0.00019999997155026783,
-      "loss": 46.0,
-      "step": 1501
-    },
-    {
-      "epoch": 0.24187769233866097,
-      "grad_norm": 0.0005968528566882014,
-      "learning_rate": 0.0001999999715120931,
-      "loss": 46.0,
-      "step": 1502
-    },
-    {
-      "epoch": 0.24203872941744836,
-      "grad_norm": 0.0003688789438456297,
-      "learning_rate": 0.00019999997147389275,
-      "loss": 46.0,
-      "step": 1503
-    },
-    {
-      "epoch": 0.24219976649623576,
-      "grad_norm": 0.0022087355609983206,
-      "learning_rate": 0.0001999999714356668,
-      "loss": 46.0,
-      "step": 1504
-    },
-    {
-      "epoch": 0.24236080357502315,
-      "grad_norm": 0.0012558646267279983,
-      "learning_rate": 0.00019999997139741527,
-      "loss": 46.0,
-      "step": 1505
-    },
-    {
-      "epoch": 0.24252184065381055,
-      "grad_norm": 0.000379395904019475,
-      "learning_rate": 0.00019999997135913814,
-      "loss": 46.0,
-      "step": 1506
-    },
-    {
-      "epoch": 0.24268287773259792,
-      "grad_norm": 0.0005132973310537636,
-      "learning_rate": 0.00019999997132083544,
-      "loss": 46.0,
-      "step": 1507
-    },
-    {
-      "epoch": 0.24284391481138531,
-      "grad_norm": 0.0005915425135754049,
-      "learning_rate": 0.00019999997128250713,
-      "loss": 46.0,
-      "step": 1508
-    },
-    {
-      "epoch": 0.2430049518901727,
-      "grad_norm": 0.0006312188343144953,
-      "learning_rate": 0.0001999999712441532,
-      "loss": 46.0,
-      "step": 1509
-    },
-    {
-      "epoch": 0.2431659889689601,
-      "grad_norm": 0.0008536765235476196,
-      "learning_rate": 0.00019999997120577373,
-      "loss": 46.0,
-      "step": 1510
-    },
-    {
-      "epoch": 0.2433270260477475,
-      "grad_norm": 0.0006685783737339079,
-      "learning_rate": 0.0001999999711673686,
-      "loss": 46.0,
-      "step": 1511
-    },
-    {
-      "epoch": 0.2434880631265349,
-      "grad_norm": 0.0004960522637702525,
-      "learning_rate": 0.00019999997112893795,
-      "loss": 46.0,
-      "step": 1512
-    },
-    {
-      "epoch": 0.24364910020532227,
-      "grad_norm": 0.0005449623567983508,
-      "learning_rate": 0.00019999997109048162,
-      "loss": 46.0,
-      "step": 1513
-    },
-    {
-      "epoch": 0.24381013728410966,
-      "grad_norm": 0.0006992222042754292,
-      "learning_rate": 0.00019999997105199974,
-      "loss": 46.0,
-      "step": 1514
-    },
-    {
-      "epoch": 0.24397117436289706,
-      "grad_norm": 0.0009419902344234288,
-      "learning_rate": 0.0001999999710134923,
-      "loss": 46.0,
-      "step": 1515
-    },
-    {
-      "epoch": 0.24413221144168445,
-      "grad_norm": 0.0003883341560140252,
-      "learning_rate": 0.0001999999709749592,
-      "loss": 46.0,
-      "step": 1516
-    },
-    {
-      "epoch": 0.24429324852047185,
-      "grad_norm": 0.0006075865821912885,
-      "learning_rate": 0.00019999997093640053,
-      "loss": 46.0,
-      "step": 1517
-    },
-    {
-      "epoch": 0.24445428559925922,
-      "grad_norm": 0.00044282834278419614,
-      "learning_rate": 0.0001999999708978163,
-      "loss": 46.0,
-      "step": 1518
-    },
-    {
-      "epoch": 0.24461532267804662,
-      "grad_norm": 0.0008768010884523392,
-      "learning_rate": 0.00019999997085920642,
-      "loss": 46.0,
-      "step": 1519
-    },
-    {
-      "epoch": 0.244776359756834,
-      "grad_norm": 0.0017720020841807127,
-      "learning_rate": 0.00019999997082057095,
-      "loss": 46.0,
-      "step": 1520
-    },
-    {
-      "epoch": 0.2449373968356214,
-      "grad_norm": 0.0008989129564724863,
-      "learning_rate": 0.00019999997078190993,
-      "loss": 46.0,
-      "step": 1521
-    },
-    {
-      "epoch": 0.2450984339144088,
-      "grad_norm": 0.0010180702665820718,
-      "learning_rate": 0.00019999997074322327,
-      "loss": 46.0,
-      "step": 1522
-    },
-    {
-      "epoch": 0.24525947099319617,
-      "grad_norm": 0.0005450667231343687,
-      "learning_rate": 0.00019999997070451104,
-      "loss": 46.0,
-      "step": 1523
-    },
-    {
-      "epoch": 0.24542050807198357,
-      "grad_norm": 0.00030308065470308065,
-      "learning_rate": 0.00019999997066577323,
-      "loss": 46.0,
-      "step": 1524
-    },
-    {
-      "epoch": 0.24558154515077096,
-      "grad_norm": 0.0004089412686880678,
-      "learning_rate": 0.0001999999706270098,
-      "loss": 46.0,
-      "step": 1525
-    },
-    {
-      "epoch": 0.24574258222955836,
-      "grad_norm": 0.0007387937512248755,
-      "learning_rate": 0.00019999997058822076,
-      "loss": 46.0,
-      "step": 1526
-    },
-    {
-      "epoch": 0.24590361930834576,
-      "grad_norm": 0.00023531209444627166,
-      "learning_rate": 0.00019999997054940616,
-      "loss": 46.0,
-      "step": 1527
-    },
-    {
-      "epoch": 0.24606465638713312,
-      "grad_norm": 0.0010702188592404127,
-      "learning_rate": 0.00019999997051056595,
-      "loss": 46.0,
-      "step": 1528
-    },
-    {
-      "epoch": 0.24622569346592052,
-      "grad_norm": 0.0003514440613798797,
-      "learning_rate": 0.00019999997047170015,
-      "loss": 46.0,
-      "step": 1529
-    },
-    {
-      "epoch": 0.24638673054470792,
-      "grad_norm": 0.00024456362007185817,
-      "learning_rate": 0.00019999997043280873,
-      "loss": 46.0,
-      "step": 1530
-    },
-    {
-      "epoch": 0.2465477676234953,
-      "grad_norm": 0.0003912733809556812,
-      "learning_rate": 0.00019999997039389175,
-      "loss": 46.0,
-      "step": 1531
-    },
-    {
-      "epoch": 0.2467088047022827,
-      "grad_norm": 0.00044792823609896004,
-      "learning_rate": 0.00019999997035494916,
-      "loss": 46.0,
-      "step": 1532
-    },
-    {
-      "epoch": 0.2468698417810701,
-      "grad_norm": 0.000588231545407325,
-      "learning_rate": 0.00019999997031598099,
-      "loss": 46.0,
-      "step": 1533
-    },
-    {
-      "epoch": 0.24703087885985747,
-      "grad_norm": 0.0015653539448976517,
-      "learning_rate": 0.0001999999702769872,
-      "loss": 46.0,
-      "step": 1534
-    },
-    {
-      "epoch": 0.24719191593864487,
-      "grad_norm": 0.0003892197273671627,
-      "learning_rate": 0.00019999997023796784,
-      "loss": 46.0,
-      "step": 1535
-    },
-    {
-      "epoch": 0.24735295301743226,
-      "grad_norm": 0.0002733547589741647,
-      "learning_rate": 0.00019999997019892285,
-      "loss": 46.0,
-      "step": 1536
-    },
-    {
-      "epoch": 0.24751399009621966,
-      "grad_norm": 0.0002564641763456166,
-      "learning_rate": 0.00019999997015985227,
-      "loss": 46.0,
-      "step": 1537
-    },
-    {
-      "epoch": 0.24767502717500706,
-      "grad_norm": 0.0014239680022001266,
-      "learning_rate": 0.00019999997012075613,
-      "loss": 46.0,
-      "step": 1538
-    },
-    {
-      "epoch": 0.24783606425379442,
-      "grad_norm": 0.0004348398360889405,
-      "learning_rate": 0.00019999997008163437,
-      "loss": 46.0,
-      "step": 1539
-    },
-    {
-      "epoch": 0.24799710133258182,
-      "grad_norm": 0.000407269224524498,
-      "learning_rate": 0.00019999997004248703,
-      "loss": 46.0,
-      "step": 1540
-    },
-    {
-      "epoch": 0.24815813841136922,
-      "grad_norm": 0.0006929379305802286,
-      "learning_rate": 0.00019999997000331407,
-      "loss": 46.0,
-      "step": 1541
-    },
-    {
-      "epoch": 0.2483191754901566,
-      "grad_norm": 0.0007344314944930375,
-      "learning_rate": 0.00019999996996411556,
-      "loss": 46.0,
-      "step": 1542
-    },
-    {
-      "epoch": 0.248480212568944,
-      "grad_norm": 0.0005696816951967776,
-      "learning_rate": 0.00019999996992489143,
-      "loss": 46.0,
-      "step": 1543
-    },
-    {
-      "epoch": 0.24864124964773138,
-      "grad_norm": 0.00044112952309660614,
-      "learning_rate": 0.00019999996988564168,
-      "loss": 46.0,
-      "step": 1544
-    },
-    {
-      "epoch": 0.24880228672651877,
-      "grad_norm": 0.0008369360584765673,
-      "learning_rate": 0.00019999996984636638,
-      "loss": 46.0,
-      "step": 1545
-    },
-    {
-      "epoch": 0.24896332380530617,
-      "grad_norm": 0.00041345180943608284,
-      "learning_rate": 0.00019999996980706546,
-      "loss": 46.0,
-      "step": 1546
-    },
-    {
-      "epoch": 0.24912436088409357,
-      "grad_norm": 0.0006784378783777356,
-      "learning_rate": 0.00019999996976773892,
-      "loss": 46.0,
-      "step": 1547
-    },
-    {
-      "epoch": 0.24928539796288096,
-      "grad_norm": 0.0013008955866098404,
-      "learning_rate": 0.00019999996972838683,
-      "loss": 46.0,
-      "step": 1548
-    },
-    {
-      "epoch": 0.24944643504166836,
-      "grad_norm": 0.0007540644728578627,
-      "learning_rate": 0.00019999996968900912,
-      "loss": 46.0,
-      "step": 1549
-    },
-    {
-      "epoch": 0.24960747212045573,
-      "grad_norm": 0.0006829394842498004,
-      "learning_rate": 0.00019999996964960583,
-      "loss": 46.0,
-      "step": 1550
-    },
-    {
-      "epoch": 0.24976850919924312,
-      "grad_norm": 0.0015458069974556565,
-      "learning_rate": 0.00019999996961017695,
-      "loss": 46.0,
-      "step": 1551
-    },
-    {
-      "epoch": 0.24992954627803052,
-      "grad_norm": 0.0007320896256715059,
-      "learning_rate": 0.00019999996957072248,
-      "loss": 46.0,
-      "step": 1552
-    },
-    {
-      "epoch": 0.2500905833568179,
-      "grad_norm": 0.001418554806150496,
-      "learning_rate": 0.00019999996953124237,
-      "loss": 46.0,
-      "step": 1553
-    },
-    {
-      "epoch": 0.2502516204356053,
-      "grad_norm": 0.001416199142113328,
-      "learning_rate": 0.0001999999694917367,
-      "loss": 46.0,
-      "step": 1554
-    },
-    {
-      "epoch": 0.2504126575143927,
-      "grad_norm": 0.0007409258396364748,
-      "learning_rate": 0.00019999996945220544,
-      "loss": 46.0,
-      "step": 1555
-    },
-    {
-      "epoch": 0.2505736945931801,
-      "grad_norm": 0.0005304866353981197,
-      "learning_rate": 0.00019999996941264856,
-      "loss": 46.0,
-      "step": 1556
-    },
-    {
-      "epoch": 0.25073473167196747,
-      "grad_norm": 0.0006201507640071213,
-      "learning_rate": 0.00019999996937306608,
-      "loss": 46.0,
-      "step": 1557
-    },
-    {
-      "epoch": 0.25089576875075487,
-      "grad_norm": 0.00029246992198750377,
-      "learning_rate": 0.00019999996933345803,
-      "loss": 46.0,
-      "step": 1558
-    },
-    {
-      "epoch": 0.25105680582954226,
-      "grad_norm": 0.00046166894026100636,
-      "learning_rate": 0.0001999999692938244,
-      "loss": 46.0,
-      "step": 1559
-    },
-    {
-      "epoch": 0.25121784290832966,
-      "grad_norm": 0.002228065626695752,
-      "learning_rate": 0.00019999996925416515,
-      "loss": 46.0,
-      "step": 1560
-    },
-    {
-      "epoch": 0.25137887998711705,
-      "grad_norm": 0.001267742714844644,
-      "learning_rate": 0.00019999996921448031,
-      "loss": 46.0,
-      "step": 1561
-    },
-    {
-      "epoch": 0.25153991706590445,
-      "grad_norm": 0.0004505883844103664,
-      "learning_rate": 0.00019999996917476986,
-      "loss": 46.0,
-      "step": 1562
-    },
-    {
-      "epoch": 0.2517009541446918,
-      "grad_norm": 0.00058057124260813,
-      "learning_rate": 0.00019999996913503383,
-      "loss": 46.0,
-      "step": 1563
-    },
-    {
-      "epoch": 0.2518619912234792,
-      "grad_norm": 0.0009239677456207573,
-      "learning_rate": 0.0001999999690952722,
-      "loss": 46.0,
-      "step": 1564
-    },
-    {
-      "epoch": 0.2520230283022666,
-      "grad_norm": 0.0006732263718731701,
-      "learning_rate": 0.000199999969055485,
-      "loss": 46.0,
-      "step": 1565
-    },
-    {
-      "epoch": 0.252184065381054,
-      "grad_norm": 0.0004894639714621007,
-      "learning_rate": 0.0001999999690156722,
-      "loss": 46.0,
-      "step": 1566
-    },
-    {
-      "epoch": 0.2523451024598414,
-      "grad_norm": 0.0012960133608430624,
-      "learning_rate": 0.00019999996897583378,
-      "loss": 46.0,
-      "step": 1567
-    },
-    {
-      "epoch": 0.25250613953862877,
-      "grad_norm": 0.0005833309260196984,
-      "learning_rate": 0.00019999996893596978,
-      "loss": 46.0,
-      "step": 1568
-    },
-    {
-      "epoch": 0.25266717661741617,
-      "grad_norm": 0.0004100526275578886,
-      "learning_rate": 0.00019999996889608017,
-      "loss": 46.0,
-      "step": 1569
-    },
-    {
-      "epoch": 0.25282821369620356,
-      "grad_norm": 0.0006177093018777668,
-      "learning_rate": 0.000199999968856165,
-      "loss": 46.0,
-      "step": 1570
-    },
-    {
-      "epoch": 0.25298925077499096,
-      "grad_norm": 0.0012867854675278068,
-      "learning_rate": 0.00019999996881622418,
-      "loss": 46.0,
-      "step": 1571
-    },
-    {
-      "epoch": 0.25315028785377836,
-      "grad_norm": 0.0005361064104363322,
-      "learning_rate": 0.0001999999687762578,
-      "loss": 46.0,
-      "step": 1572
-    },
-    {
-      "epoch": 0.2533113249325657,
-      "grad_norm": 0.00029784213984385133,
-      "learning_rate": 0.00019999996873626584,
-      "loss": 46.0,
-      "step": 1573
-    },
-    {
-      "epoch": 0.2534723620113531,
-      "grad_norm": 0.000444651348516345,
-      "learning_rate": 0.00019999996869624824,
-      "loss": 46.0,
-      "step": 1574
-    },
-    {
-      "epoch": 0.2536333990901405,
-      "grad_norm": 0.0005632179090753198,
-      "learning_rate": 0.00019999996865620507,
-      "loss": 46.0,
-      "step": 1575
-    },
-    {
-      "epoch": 0.2537944361689279,
-      "grad_norm": 0.0005233383853919804,
-      "learning_rate": 0.00019999996861613632,
-      "loss": 46.0,
-      "step": 1576
-    },
-    {
-      "epoch": 0.2539554732477153,
-      "grad_norm": 0.0007815441931597888,
-      "learning_rate": 0.00019999996857604196,
-      "loss": 46.0,
-      "step": 1577
-    },
-    {
-      "epoch": 0.2541165103265027,
-      "grad_norm": 0.0005321700591593981,
-      "learning_rate": 0.000199999968535922,
-      "loss": 46.0,
-      "step": 1578
-    },
-    {
-      "epoch": 0.25427754740529007,
-      "grad_norm": 0.0005488034221343696,
-      "learning_rate": 0.00019999996849577646,
-      "loss": 46.0,
-      "step": 1579
-    },
-    {
-      "epoch": 0.25443858448407747,
-      "grad_norm": 0.00032734667183831334,
-      "learning_rate": 0.0001999999684556053,
-      "loss": 46.0,
-      "step": 1580
-    },
-    {
-      "epoch": 0.25459962156286486,
-      "grad_norm": 0.00039045518497005105,
-      "learning_rate": 0.00019999996841540857,
-      "loss": 46.0,
-      "step": 1581
-    },
-    {
-      "epoch": 0.25476065864165226,
-      "grad_norm": 0.0007887433748692274,
-      "learning_rate": 0.00019999996837518624,
-      "loss": 46.0,
-      "step": 1582
-    },
-    {
-      "epoch": 0.25492169572043966,
-      "grad_norm": 0.0002828974393196404,
-      "learning_rate": 0.00019999996833493832,
-      "loss": 46.0,
-      "step": 1583
-    },
-    {
-      "epoch": 0.255082732799227,
-      "grad_norm": 0.0010236426023766398,
-      "learning_rate": 0.00019999996829466482,
-      "loss": 46.0,
-      "step": 1584
-    },
-    {
-      "epoch": 0.2552437698780144,
-      "grad_norm": 0.0007546267588622868,
-      "learning_rate": 0.00019999996825436568,
-      "loss": 46.0,
-      "step": 1585
-    },
-    {
-      "epoch": 0.2554048069568018,
-      "grad_norm": 0.001640638685785234,
-      "learning_rate": 0.00019999996821404097,
-      "loss": 46.0,
-      "step": 1586
-    },
-    {
-      "epoch": 0.2555658440355892,
-      "grad_norm": 0.0006235822220332921,
-      "learning_rate": 0.00019999996817369065,
-      "loss": 46.0,
-      "step": 1587
-    },
-    {
-      "epoch": 0.2557268811143766,
-      "grad_norm": 0.0017373906448483467,
-      "learning_rate": 0.00019999996813331477,
-      "loss": 46.0,
-      "step": 1588
-    },
-    {
-      "epoch": 0.255887918193164,
-      "grad_norm": 0.0007070084102451801,
-      "learning_rate": 0.00019999996809291325,
-      "loss": 46.0,
-      "step": 1589
-    },
-    {
-      "epoch": 0.2560489552719514,
-      "grad_norm": 0.00045126734767109156,
-      "learning_rate": 0.00019999996805248617,
-      "loss": 46.0,
-      "step": 1590
-    },
-    {
-      "epoch": 0.25620999235073877,
-      "grad_norm": 0.0006181900971569121,
-      "learning_rate": 0.00019999996801203348,
-      "loss": 46.0,
-      "step": 1591
-    },
-    {
-      "epoch": 0.25637102942952616,
-      "grad_norm": 0.0002809020515996963,
-      "learning_rate": 0.0001999999679715552,
-      "loss": 46.0,
-      "step": 1592
-    },
-    {
-      "epoch": 0.25653206650831356,
-      "grad_norm": 0.002091732807457447,
-      "learning_rate": 0.0001999999679310513,
-      "loss": 46.0,
-      "step": 1593
-    },
-    {
-      "epoch": 0.25669310358710096,
-      "grad_norm": 0.00030604659696109593,
-      "learning_rate": 0.00019999996789052187,
-      "loss": 46.0,
-      "step": 1594
-    },
-    {
-      "epoch": 0.2568541406658883,
-      "grad_norm": 0.0003619135241024196,
-      "learning_rate": 0.0001999999678499668,
-      "loss": 46.0,
-      "step": 1595
-    },
-    {
-      "epoch": 0.2570151777446757,
-      "grad_norm": 0.0003118929744232446,
-      "learning_rate": 0.00019999996780938612,
-      "loss": 46.0,
-      "step": 1596
-    },
-    {
-      "epoch": 0.2571762148234631,
-      "grad_norm": 0.0004634863289538771,
-      "learning_rate": 0.00019999996776877988,
-      "loss": 46.0,
-      "step": 1597
-    },
-    {
-      "epoch": 0.2573372519022505,
-      "grad_norm": 0.0014591977233067155,
-      "learning_rate": 0.00019999996772814802,
-      "loss": 46.0,
-      "step": 1598
-    },
-    {
-      "epoch": 0.2574982889810379,
-      "grad_norm": 0.00050339475274086,
-      "learning_rate": 0.0001999999676874906,
-      "loss": 46.0,
-      "step": 1599
-    },
-    {
-      "epoch": 0.2576593260598253,
-      "grad_norm": 0.0008291835547424853,
-      "learning_rate": 0.00019999996764680754,
-      "loss": 46.0,
-      "step": 1600
-    },
-    {
-      "epoch": 0.2578203631386127,
-      "grad_norm": 0.0006652891752310097,
-      "learning_rate": 0.0001999999676060989,
-      "loss": 46.0,
-      "step": 1601
-    },
-    {
-      "epoch": 0.25798140021740007,
-      "grad_norm": 0.002328990027308464,
-      "learning_rate": 0.00019999996756536468,
-      "loss": 46.0,
-      "step": 1602
-    },
-    {
-      "epoch": 0.25814243729618747,
-      "grad_norm": 0.0004221296403557062,
-      "learning_rate": 0.00019999996752460483,
-      "loss": 46.0,
-      "step": 1603
-    },
-    {
-      "epoch": 0.25830347437497486,
-      "grad_norm": 0.0019949637353420258,
-      "learning_rate": 0.00019999996748381942,
-      "loss": 46.0,
-      "step": 1604
-    },
-    {
-      "epoch": 0.2584645114537622,
-      "grad_norm": 0.000833825848530978,
-      "learning_rate": 0.00019999996744300843,
-      "loss": 46.0,
-      "step": 1605
-    },
-    {
-      "epoch": 0.2586255485325496,
-      "grad_norm": 0.00023156612587627023,
-      "learning_rate": 0.0001999999674021718,
-      "loss": 46.0,
-      "step": 1606
-    },
-    {
-      "epoch": 0.258786585611337,
-      "grad_norm": 0.00037104793591424823,
-      "learning_rate": 0.0001999999673613096,
-      "loss": 46.0,
-      "step": 1607
-    },
-    {
-      "epoch": 0.2589476226901244,
-      "grad_norm": 0.0005041220574639738,
-      "learning_rate": 0.0001999999673204218,
-      "loss": 46.0,
-      "step": 1608
-    },
-    {
-      "epoch": 0.2591086597689118,
-      "grad_norm": 0.0009729259763844311,
-      "learning_rate": 0.0001999999672795084,
-      "loss": 46.0,
-      "step": 1609
-    },
-    {
-      "epoch": 0.2592696968476992,
-      "grad_norm": 0.0005305565427988768,
-      "learning_rate": 0.00019999996723856942,
-      "loss": 46.0,
-      "step": 1610
-    },
-    {
-      "epoch": 0.2594307339264866,
-      "grad_norm": 0.0005919802933931351,
-      "learning_rate": 0.00019999996719760485,
-      "loss": 46.0,
-      "step": 1611
-    },
-    {
-      "epoch": 0.259591771005274,
-      "grad_norm": 0.0008214295958168805,
-      "learning_rate": 0.00019999996715661466,
-      "loss": 46.0,
-      "step": 1612
-    },
-    {
-      "epoch": 0.25975280808406137,
-      "grad_norm": 0.0016582038952037692,
-      "learning_rate": 0.0001999999671155989,
-      "loss": 46.0,
-      "step": 1613
-    },
-    {
-      "epoch": 0.25991384516284877,
-      "grad_norm": 0.0005017261137254536,
-      "learning_rate": 0.0001999999670745575,
-      "loss": 46.0,
-      "step": 1614
-    },
-    {
-      "epoch": 0.26007488224163616,
-      "grad_norm": 0.001049538841471076,
-      "learning_rate": 0.00019999996703349055,
-      "loss": 46.0,
-      "step": 1615
-    },
-    {
-      "epoch": 0.2602359193204235,
-      "grad_norm": 0.0008739667246118188,
-      "learning_rate": 0.00019999996699239801,
-      "loss": 46.0,
-      "step": 1616
-    },
-    {
-      "epoch": 0.2603969563992109,
-      "grad_norm": 0.0003842436126433313,
-      "learning_rate": 0.00019999996695127984,
-      "loss": 46.0,
-      "step": 1617
-    },
-    {
-      "epoch": 0.2605579934779983,
-      "grad_norm": 0.0005943210562691092,
-      "learning_rate": 0.0001999999669101361,
-      "loss": 46.0,
-      "step": 1618
-    },
-    {
-      "epoch": 0.2607190305567857,
-      "grad_norm": 0.0008918180246837437,
-      "learning_rate": 0.00019999996686896675,
-      "loss": 46.0,
-      "step": 1619
-    },
-    {
-      "epoch": 0.2608800676355731,
-      "grad_norm": 0.0005695180152542889,
-      "learning_rate": 0.00019999996682777183,
-      "loss": 46.0,
-      "step": 1620
-    },
-    {
-      "epoch": 0.2610411047143605,
-      "grad_norm": 0.0008891428587958217,
-      "learning_rate": 0.00019999996678655128,
-      "loss": 46.0,
-      "step": 1621
-    },
-    {
-      "epoch": 0.2612021417931479,
-      "grad_norm": 0.001256552874110639,
-      "learning_rate": 0.00019999996674530514,
-      "loss": 46.0,
-      "step": 1622
-    },
-    {
-      "epoch": 0.2613631788719353,
-      "grad_norm": 0.0011620331788435578,
-      "learning_rate": 0.00019999996670403344,
-      "loss": 46.0,
-      "step": 1623
-    },
-    {
-      "epoch": 0.26152421595072267,
-      "grad_norm": 0.00022432550031226128,
-      "learning_rate": 0.00019999996666273612,
-      "loss": 46.0,
-      "step": 1624
-    },
-    {
-      "epoch": 0.26168525302951007,
-      "grad_norm": 0.0008183136233128607,
-      "learning_rate": 0.0001999999666214132,
-      "loss": 46.0,
-      "step": 1625
-    },
-    {
-      "epoch": 0.2618462901082974,
-      "grad_norm": 0.0006643966771662235,
-      "learning_rate": 0.0001999999665800647,
-      "loss": 46.0,
-      "step": 1626
-    },
-    {
-      "epoch": 0.2620073271870848,
-      "grad_norm": 0.0006474255351349711,
-      "learning_rate": 0.0001999999665386906,
-      "loss": 46.0,
-      "step": 1627
-    },
-    {
-      "epoch": 0.2621683642658722,
-      "grad_norm": 0.0021476037800312042,
-      "learning_rate": 0.0001999999664972909,
-      "loss": 46.0,
-      "step": 1628
-    },
-    {
-      "epoch": 0.2623294013446596,
-      "grad_norm": 0.0009265922126360238,
-      "learning_rate": 0.0001999999664558656,
-      "loss": 46.0,
-      "step": 1629
-    },
-    {
-      "epoch": 0.262490438423447,
-      "grad_norm": 0.0010417335433885455,
-      "learning_rate": 0.00019999996641441471,
-      "loss": 46.0,
-      "step": 1630
-    },
-    {
-      "epoch": 0.2626514755022344,
-      "grad_norm": 0.00042905265581794083,
-      "learning_rate": 0.00019999996637293826,
-      "loss": 46.0,
-      "step": 1631
-    },
-    {
-      "epoch": 0.2628125125810218,
-      "grad_norm": 0.0004169638850726187,
-      "learning_rate": 0.00019999996633143617,
-      "loss": 46.0,
-      "step": 1632
-    },
-    {
-      "epoch": 0.2629735496598092,
-      "grad_norm": 0.0007529768045060337,
-      "learning_rate": 0.0001999999662899085,
-      "loss": 46.0,
-      "step": 1633
-    },
-    {
-      "epoch": 0.2631345867385966,
-      "grad_norm": 0.0016266483580693603,
-      "learning_rate": 0.00019999996624835525,
-      "loss": 46.0,
-      "step": 1634
-    },
-    {
-      "epoch": 0.263295623817384,
-      "grad_norm": 0.0009302504477091134,
-      "learning_rate": 0.00019999996620677636,
-      "loss": 46.0,
-      "step": 1635
-    },
-    {
-      "epoch": 0.26345666089617137,
-      "grad_norm": 0.00047719170106574893,
-      "learning_rate": 0.00019999996616517192,
-      "loss": 46.0,
-      "step": 1636
-    },
-    {
-      "epoch": 0.2636176979749587,
-      "grad_norm": 0.000791926053352654,
-      "learning_rate": 0.0001999999661235419,
-      "loss": 46.0,
-      "step": 1637
-    },
-    {
-      "epoch": 0.2637787350537461,
-      "grad_norm": 0.00034146930556744337,
-      "learning_rate": 0.00019999996608188622,
-      "loss": 46.0,
-      "step": 1638
-    },
-    {
-      "epoch": 0.2639397721325335,
-      "grad_norm": 0.0005008805310353637,
-      "learning_rate": 0.000199999966040205,
-      "loss": 46.0,
-      "step": 1639
-    },
-    {
-      "epoch": 0.2641008092113209,
-      "grad_norm": 0.00041181285632774234,
-      "learning_rate": 0.00019999996599849815,
-      "loss": 46.0,
-      "step": 1640
-    },
-    {
-      "epoch": 0.2642618462901083,
-      "grad_norm": 0.0003014621906913817,
-      "learning_rate": 0.00019999996595676574,
-      "loss": 46.0,
-      "step": 1641
-    },
-    {
-      "epoch": 0.2644228833688957,
-      "grad_norm": 0.0006959281163290143,
-      "learning_rate": 0.00019999996591500772,
-      "loss": 46.0,
-      "step": 1642
-    },
-    {
-      "epoch": 0.2645839204476831,
-      "grad_norm": 0.0006425845203921199,
-      "learning_rate": 0.00019999996587322406,
-      "loss": 46.0,
-      "step": 1643
-    },
-    {
-      "epoch": 0.2647449575264705,
-      "grad_norm": 0.0005940453265793622,
-      "learning_rate": 0.00019999996583141487,
-      "loss": 46.0,
-      "step": 1644
-    },
-    {
-      "epoch": 0.2649059946052579,
-      "grad_norm": 0.0006798694957979023,
-      "learning_rate": 0.00019999996578958006,
-      "loss": 46.0,
-      "step": 1645
-    },
-    {
-      "epoch": 0.2650670316840453,
-      "grad_norm": 0.0007211341289803386,
-      "learning_rate": 0.00019999996574771966,
-      "loss": 46.0,
-      "step": 1646
-    },
-    {
-      "epoch": 0.26522806876283267,
-      "grad_norm": 0.0005820472142659128,
-      "learning_rate": 0.00019999996570583365,
-      "loss": 46.0,
-      "step": 1647
-    },
-    {
-      "epoch": 0.26538910584162,
-      "grad_norm": 0.0013958399649709463,
-      "learning_rate": 0.00019999996566392208,
-      "loss": 46.0,
-      "step": 1648
-    },
-    {
-      "epoch": 0.2655501429204074,
-      "grad_norm": 0.0003573969879653305,
-      "learning_rate": 0.00019999996562198487,
-      "loss": 46.0,
-      "step": 1649
-    },
-    {
-      "epoch": 0.2657111799991948,
-      "grad_norm": 0.0004746705817524344,
-      "learning_rate": 0.0001999999655800221,
-      "loss": 46.0,
-      "step": 1650
-    },
-    {
-      "epoch": 0.2658722170779822,
-      "grad_norm": 0.0003913062100764364,
-      "learning_rate": 0.00019999996553803372,
-      "loss": 46.0,
-      "step": 1651
-    },
-    {
-      "epoch": 0.2660332541567696,
-      "grad_norm": 0.0010346017079427838,
-      "learning_rate": 0.00019999996549601974,
-      "loss": 46.0,
-      "step": 1652
-    },
-    {
-      "epoch": 0.266194291235557,
-      "grad_norm": 0.0005458653904497623,
-      "learning_rate": 0.00019999996545398018,
-      "loss": 46.0,
-      "step": 1653
-    },
-    {
-      "epoch": 0.2663553283143444,
-      "grad_norm": 0.0004970126319676638,
-      "learning_rate": 0.000199999965411915,
-      "loss": 46.0,
-      "step": 1654
-    },
-    {
-      "epoch": 0.2665163653931318,
-      "grad_norm": 0.00036395867937244475,
-      "learning_rate": 0.00019999996536982425,
-      "loss": 46.0,
-      "step": 1655
-    },
-    {
-      "epoch": 0.2666774024719192,
-      "grad_norm": 0.0016682945424690843,
-      "learning_rate": 0.0001999999653277079,
-      "loss": 46.0,
-      "step": 1656
-    },
-    {
-      "epoch": 0.2668384395507066,
-      "grad_norm": 0.0003370000922586769,
-      "learning_rate": 0.00019999996528556596,
-      "loss": 46.0,
-      "step": 1657
-    },
-    {
-      "epoch": 0.2669994766294939,
-      "grad_norm": 0.0012163551291450858,
-      "learning_rate": 0.00019999996524339841,
-      "loss": 46.0,
-      "step": 1658
-    },
-    {
-      "epoch": 0.2671605137082813,
-      "grad_norm": 0.0010939136845991015,
-      "learning_rate": 0.00019999996520120528,
-      "loss": 46.0,
-      "step": 1659
-    },
-    {
-      "epoch": 0.2673215507870687,
-      "grad_norm": 0.0007409893441945314,
-      "learning_rate": 0.00019999996515898655,
-      "loss": 46.0,
-      "step": 1660
-    },
-    {
-      "epoch": 0.2674825878658561,
-      "grad_norm": 0.00038251234218478203,
-      "learning_rate": 0.00019999996511674221,
-      "loss": 46.0,
-      "step": 1661
-    },
-    {
-      "epoch": 0.2676436249446435,
-      "grad_norm": 0.0004333512915764004,
-      "learning_rate": 0.0001999999650744723,
-      "loss": 46.0,
-      "step": 1662
-    },
-    {
-      "epoch": 0.2678046620234309,
-      "grad_norm": 0.0010560126975178719,
-      "learning_rate": 0.0001999999650321768,
-      "loss": 46.0,
-      "step": 1663
-    },
-    {
-      "epoch": 0.2679656991022183,
-      "grad_norm": 0.000265865441178903,
-      "learning_rate": 0.00019999996498985568,
-      "loss": 46.0,
-      "step": 1664
-    },
-    {
-      "epoch": 0.2681267361810057,
-      "grad_norm": 0.0004550080338958651,
-      "learning_rate": 0.00019999996494750896,
-      "loss": 46.0,
-      "step": 1665
-    },
-    {
-      "epoch": 0.2682877732597931,
-      "grad_norm": 0.000625417975243181,
-      "learning_rate": 0.00019999996490513666,
-      "loss": 46.0,
-      "step": 1666
-    },
-    {
-      "epoch": 0.2684488103385805,
-      "grad_norm": 0.0003973632992710918,
-      "learning_rate": 0.00019999996486273877,
-      "loss": 46.0,
-      "step": 1667
-    },
-    {
-      "epoch": 0.2686098474173679,
-      "grad_norm": 0.0009851433569565415,
-      "learning_rate": 0.0001999999648203153,
-      "loss": 46.0,
-      "step": 1668
-    },
-    {
-      "epoch": 0.2687708844961552,
-      "grad_norm": 0.0006015888066031039,
-      "learning_rate": 0.00019999996477786618,
-      "loss": 46.0,
-      "step": 1669
-    },
-    {
-      "epoch": 0.2689319215749426,
-      "grad_norm": 0.0022469062823802233,
-      "learning_rate": 0.0001999999647353915,
-      "loss": 46.0,
-      "step": 1670
-    },
-    {
-      "epoch": 0.26909295865373,
-      "grad_norm": 0.0010461158817633986,
-      "learning_rate": 0.00019999996469289124,
-      "loss": 46.0,
-      "step": 1671
-    },
-    {
-      "epoch": 0.2692539957325174,
-      "grad_norm": 0.0011682932963594794,
-      "learning_rate": 0.00019999996465036536,
-      "loss": 46.0,
-      "step": 1672
-    },
-    {
-      "epoch": 0.2694150328113048,
-      "grad_norm": 0.0020578380208462477,
-      "learning_rate": 0.0001999999646078139,
-      "loss": 46.0,
-      "step": 1673
-    },
-    {
-      "epoch": 0.2695760698900922,
-      "grad_norm": 0.0016831484390422702,
-      "learning_rate": 0.00019999996456523684,
-      "loss": 46.0,
-      "step": 1674
-    },
-    {
-      "epoch": 0.2697371069688796,
-      "grad_norm": 0.0006138753378763795,
-      "learning_rate": 0.0001999999645226342,
-      "loss": 46.0,
-      "step": 1675
-    },
-    {
-      "epoch": 0.269898144047667,
-      "grad_norm": 0.00041789753595367074,
-      "learning_rate": 0.00019999996448000595,
-      "loss": 46.0,
-      "step": 1676
-    },
-    {
-      "epoch": 0.2700591811264544,
-      "grad_norm": 0.0006504368502646685,
-      "learning_rate": 0.0001999999644373521,
-      "loss": 46.0,
-      "step": 1677
-    },
-    {
-      "epoch": 0.2702202182052418,
-      "grad_norm": 0.0007486810791306198,
-      "learning_rate": 0.00019999996439467265,
-      "loss": 46.0,
-      "step": 1678
-    },
-    {
-      "epoch": 0.2703812552840291,
-      "grad_norm": 0.0007228718604892492,
-      "learning_rate": 0.0001999999643519676,
-      "loss": 46.0,
-      "step": 1679
-    },
-    {
-      "epoch": 0.2705422923628165,
-      "grad_norm": 0.00026839703787118196,
-      "learning_rate": 0.000199999964309237,
-      "loss": 46.0,
-      "step": 1680
-    },
-    {
-      "epoch": 0.2707033294416039,
-      "grad_norm": 0.00033148820511996746,
-      "learning_rate": 0.00019999996426648076,
-      "loss": 46.0,
-      "step": 1681
-    },
-    {
-      "epoch": 0.2708643665203913,
-      "grad_norm": 0.00047075640759430826,
-      "learning_rate": 0.00019999996422369895,
-      "loss": 46.0,
-      "step": 1682
-    },
-    {
-      "epoch": 0.2710254035991787,
-      "grad_norm": 0.00040921446634456515,
-      "learning_rate": 0.00019999996418089153,
-      "loss": 46.0,
-      "step": 1683
-    },
-    {
-      "epoch": 0.2711864406779661,
-      "grad_norm": 0.0005522763822227716,
-      "learning_rate": 0.00019999996413805853,
-      "loss": 46.0,
-      "step": 1684
-    },
-    {
-      "epoch": 0.2713474777567535,
-      "grad_norm": 0.0018814326031133533,
-      "learning_rate": 0.0001999999640951999,
-      "loss": 46.0,
-      "step": 1685
-    },
-    {
-      "epoch": 0.2715085148355409,
-      "grad_norm": 0.0008086210582405329,
-      "learning_rate": 0.00019999996405231573,
-      "loss": 46.0,
-      "step": 1686
-    },
-    {
-      "epoch": 0.2716695519143283,
-      "grad_norm": 0.003679858520627022,
-      "learning_rate": 0.00019999996400940596,
-      "loss": 46.0,
-      "step": 1687
-    },
-    {
-      "epoch": 0.2718305889931157,
-      "grad_norm": 0.0006994935101829469,
-      "learning_rate": 0.00019999996396647055,
-      "loss": 46.0,
-      "step": 1688
-    },
-    {
-      "epoch": 0.2719916260719031,
-      "grad_norm": 0.001574354711920023,
-      "learning_rate": 0.00019999996392350958,
-      "loss": 46.0,
-      "step": 1689
-    },
-    {
-      "epoch": 0.2721526631506904,
-      "grad_norm": 0.0005769500276073813,
-      "learning_rate": 0.000199999963880523,
-      "loss": 46.0,
-      "step": 1690
-    },
-    {
-      "epoch": 0.2723137002294778,
-      "grad_norm": 0.00021050751092843711,
-      "learning_rate": 0.00019999996383751083,
-      "loss": 46.0,
-      "step": 1691
-    },
-    {
-      "epoch": 0.2724747373082652,
-      "grad_norm": 0.0010245003504678607,
-      "learning_rate": 0.00019999996379447304,
-      "loss": 46.0,
-      "step": 1692
-    },
-    {
-      "epoch": 0.2726357743870526,
-      "grad_norm": 0.0017468957230448723,
-      "learning_rate": 0.0001999999637514097,
-      "loss": 46.0,
-      "step": 1693
-    },
-    {
-      "epoch": 0.27279681146584,
-      "grad_norm": 0.0009225663961842656,
-      "learning_rate": 0.0001999999637083207,
-      "loss": 46.0,
-      "step": 1694
-    },
-    {
-      "epoch": 0.2729578485446274,
-      "grad_norm": 0.0004138411895837635,
-      "learning_rate": 0.00019999996366520617,
-      "loss": 46.0,
-      "step": 1695
-    },
-    {
-      "epoch": 0.2731188856234148,
-      "grad_norm": 0.000979991746135056,
-      "learning_rate": 0.00019999996362206603,
-      "loss": 46.0,
-      "step": 1696
-    },
-    {
-      "epoch": 0.2732799227022022,
-      "grad_norm": 0.0005196937127038836,
-      "learning_rate": 0.00019999996357890029,
-      "loss": 46.0,
-      "step": 1697
-    },
-    {
-      "epoch": 0.2734409597809896,
-      "grad_norm": 0.0005258449236862361,
-      "learning_rate": 0.00019999996353570895,
-      "loss": 46.0,
-      "step": 1698
-    },
-    {
-      "epoch": 0.273601996859777,
-      "grad_norm": 0.0002927335153799504,
-      "learning_rate": 0.00019999996349249203,
-      "loss": 46.0,
-      "step": 1699
-    },
-    {
-      "epoch": 0.2737630339385644,
-      "grad_norm": 0.00042445422150194645,
-      "learning_rate": 0.0001999999634492495,
-      "loss": 46.0,
-      "step": 1700
-    },
-    {
-      "epoch": 0.2739240710173517,
-      "grad_norm": 0.0007652249187231064,
-      "learning_rate": 0.00019999996340598137,
-      "loss": 46.0,
-      "step": 1701
-    },
-    {
-      "epoch": 0.2740851080961391,
-      "grad_norm": 0.00042803556425496936,
-      "learning_rate": 0.00019999996336268766,
-      "loss": 46.0,
-      "step": 1702
-    },
-    {
-      "epoch": 0.2742461451749265,
-      "grad_norm": 0.0005782022490166128,
-      "learning_rate": 0.00019999996331936833,
-      "loss": 46.0,
-      "step": 1703
-    },
-    {
-      "epoch": 0.2744071822537139,
-      "grad_norm": 0.0004274678649380803,
-      "learning_rate": 0.00019999996327602345,
-      "loss": 46.0,
-      "step": 1704
-    },
-    {
-      "epoch": 0.2745682193325013,
-      "grad_norm": 0.000336539902491495,
-      "learning_rate": 0.00019999996323265292,
-      "loss": 46.0,
-      "step": 1705
-    },
-    {
-      "epoch": 0.2747292564112887,
-      "grad_norm": 0.0009749602759256959,
-      "learning_rate": 0.00019999996318925684,
-      "loss": 46.0,
-      "step": 1706
-    },
-    {
-      "epoch": 0.2748902934900761,
-      "grad_norm": 0.0003062020696233958,
-      "learning_rate": 0.00019999996314583516,
-      "loss": 46.0,
-      "step": 1707
-    },
-    {
-      "epoch": 0.2750513305688635,
-      "grad_norm": 0.0005401447997428477,
-      "learning_rate": 0.0001999999631023879,
-      "loss": 46.0,
-      "step": 1708
-    },
-    {
-      "epoch": 0.2752123676476509,
-      "grad_norm": 0.0007868324755690992,
-      "learning_rate": 0.000199999963058915,
-      "loss": 46.0,
-      "step": 1709
-    },
-    {
-      "epoch": 0.2753734047264383,
-      "grad_norm": 0.0003401754656806588,
-      "learning_rate": 0.00019999996301541654,
-      "loss": 46.0,
-      "step": 1710
-    },
-    {
-      "epoch": 0.2755344418052256,
-      "grad_norm": 0.0010305638425052166,
-      "learning_rate": 0.00019999996297189246,
-      "loss": 46.0,
-      "step": 1711
-    },
-    {
-      "epoch": 0.275695478884013,
-      "grad_norm": 0.0006922843167558312,
-      "learning_rate": 0.0001999999629283428,
-      "loss": 46.0,
-      "step": 1712
-    },
-    {
-      "epoch": 0.2758565159628004,
-      "grad_norm": 0.0005645108758471906,
-      "learning_rate": 0.00019999996288476752,
-      "loss": 46.0,
-      "step": 1713
-    },
-    {
-      "epoch": 0.2760175530415878,
-      "grad_norm": 0.00034462963230907917,
-      "learning_rate": 0.00019999996284116665,
-      "loss": 46.0,
-      "step": 1714
-    },
-    {
-      "epoch": 0.2761785901203752,
-      "grad_norm": 0.0004854914150200784,
-      "learning_rate": 0.00019999996279754023,
-      "loss": 46.0,
-      "step": 1715
-    },
-    {
-      "epoch": 0.2763396271991626,
-      "grad_norm": 0.0007340558222495019,
-      "learning_rate": 0.00019999996275388816,
-      "loss": 46.0,
-      "step": 1716
-    },
-    {
-      "epoch": 0.27650066427795,
-      "grad_norm": 0.0007356297574006021,
-      "learning_rate": 0.00019999996271021054,
-      "loss": 46.0,
-      "step": 1717
-    },
-    {
-      "epoch": 0.2766617013567374,
-      "grad_norm": 0.0003644342941697687,
-      "learning_rate": 0.00019999996266650732,
-      "loss": 46.0,
-      "step": 1718
-    },
-    {
-      "epoch": 0.2768227384355248,
-      "grad_norm": 0.0006618080660700798,
-      "learning_rate": 0.00019999996262277847,
-      "loss": 46.0,
-      "step": 1719
-    },
-    {
-      "epoch": 0.2769837755143122,
-      "grad_norm": 0.0002794034662656486,
-      "learning_rate": 0.00019999996257902405,
-      "loss": 46.0,
-      "step": 1720
-    },
-    {
-      "epoch": 0.2771448125930996,
-      "grad_norm": 0.0007610495667904615,
-      "learning_rate": 0.00019999996253524405,
-      "loss": 46.0,
-      "step": 1721
-    },
-    {
-      "epoch": 0.27730584967188693,
-      "grad_norm": 0.0004926969413645566,
-      "learning_rate": 0.00019999996249143843,
-      "loss": 46.0,
-      "step": 1722
-    },
-    {
-      "epoch": 0.2774668867506743,
-      "grad_norm": 0.0005112563376314938,
-      "learning_rate": 0.0001999999624476072,
-      "loss": 46.0,
-      "step": 1723
-    },
-    {
-      "epoch": 0.2776279238294617,
-      "grad_norm": 0.0007417351589538157,
-      "learning_rate": 0.00019999996240375044,
-      "loss": 46.0,
-      "step": 1724
-    },
-    {
-      "epoch": 0.2777889609082491,
-      "grad_norm": 0.0008114571101032197,
-      "learning_rate": 0.00019999996235986803,
-      "loss": 46.0,
-      "step": 1725
-    },
-    {
-      "epoch": 0.2779499979870365,
-      "grad_norm": 0.0004694694362115115,
-      "learning_rate": 0.00019999996231596004,
-      "loss": 46.0,
-      "step": 1726
-    },
-    {
-      "epoch": 0.2781110350658239,
-      "grad_norm": 0.0013439524918794632,
-      "learning_rate": 0.00019999996227202643,
-      "loss": 46.0,
-      "step": 1727
-    },
-    {
-      "epoch": 0.2782720721446113,
-      "grad_norm": 0.0005587266059592366,
-      "learning_rate": 0.00019999996222806727,
-      "loss": 46.0,
-      "step": 1728
-    },
-    {
-      "epoch": 0.2784331092233987,
-      "grad_norm": 0.0008996734395623207,
-      "learning_rate": 0.00019999996218408248,
-      "loss": 46.0,
-      "step": 1729
-    },
-    {
-      "epoch": 0.2785941463021861,
-      "grad_norm": 0.0003521718899719417,
-      "learning_rate": 0.00019999996214007212,
-      "loss": 46.0,
-      "step": 1730
-    },
-    {
-      "epoch": 0.2787551833809735,
-      "grad_norm": 0.0021810263860970736,
-      "learning_rate": 0.00019999996209603613,
-      "loss": 46.0,
-      "step": 1731
-    },
-    {
-      "epoch": 0.27891622045976083,
-      "grad_norm": 0.000892314943484962,
-      "learning_rate": 0.00019999996205197456,
-      "loss": 46.0,
-      "step": 1732
-    },
-    {
-      "epoch": 0.27907725753854823,
-      "grad_norm": 0.0002361432125326246,
-      "learning_rate": 0.0001999999620078874,
-      "loss": 46.0,
-      "step": 1733
-    },
-    {
-      "epoch": 0.2792382946173356,
-      "grad_norm": 0.0004186244332231581,
-      "learning_rate": 0.0001999999619637747,
-      "loss": 46.0,
-      "step": 1734
-    },
-    {
-      "epoch": 0.279399331696123,
-      "grad_norm": 0.00048368930583819747,
-      "learning_rate": 0.00019999996191963633,
-      "loss": 46.0,
-      "step": 1735
-    },
-    {
-      "epoch": 0.2795603687749104,
-      "grad_norm": 0.0003011162916664034,
-      "learning_rate": 0.00019999996187547239,
-      "loss": 46.0,
-      "step": 1736
-    },
-    {
-      "epoch": 0.2797214058536978,
-      "grad_norm": 0.0010776141425594687,
-      "learning_rate": 0.00019999996183128285,
-      "loss": 46.0,
-      "step": 1737
-    },
-    {
-      "epoch": 0.2798824429324852,
-      "grad_norm": 0.0005428716540336609,
-      "learning_rate": 0.00019999996178706773,
-      "loss": 46.0,
-      "step": 1738
-    },
-    {
-      "epoch": 0.2800434800112726,
-      "grad_norm": 0.0012752683833241463,
-      "learning_rate": 0.00019999996174282697,
-      "loss": 46.0,
-      "step": 1739
-    },
-    {
-      "epoch": 0.28020451709006,
-      "grad_norm": 0.0007044172962196171,
-      "learning_rate": 0.00019999996169856065,
-      "loss": 46.0,
-      "step": 1740
-    },
-    {
-      "epoch": 0.2803655541688474,
-      "grad_norm": 0.0008881782414391637,
-      "learning_rate": 0.00019999996165426874,
-      "loss": 46.0,
-      "step": 1741
-    },
-    {
-      "epoch": 0.2805265912476348,
-      "grad_norm": 0.0005630353698506951,
-      "learning_rate": 0.00019999996160995125,
-      "loss": 46.0,
-      "step": 1742
-    },
-    {
-      "epoch": 0.28068762832642213,
-      "grad_norm": 0.0005720113986171782,
-      "learning_rate": 0.00019999996156560814,
-      "loss": 46.0,
-      "step": 1743
-    },
-    {
-      "epoch": 0.28084866540520953,
-      "grad_norm": 0.0006910123047418892,
-      "learning_rate": 0.00019999996152123944,
-      "loss": 46.0,
-      "step": 1744
-    },
-    {
-      "epoch": 0.2810097024839969,
-      "grad_norm": 0.0009649637504480779,
-      "learning_rate": 0.00019999996147684516,
-      "loss": 46.0,
-      "step": 1745
-    },
-    {
-      "epoch": 0.2811707395627843,
-      "grad_norm": 0.0015174208674579859,
-      "learning_rate": 0.00019999996143242526,
-      "loss": 46.0,
-      "step": 1746
-    },
-    {
-      "epoch": 0.2813317766415717,
-      "grad_norm": 0.0003052711545024067,
-      "learning_rate": 0.00019999996138797977,
-      "loss": 46.0,
-      "step": 1747
-    },
-    {
-      "epoch": 0.2814928137203591,
-      "grad_norm": 0.0021164272911846638,
-      "learning_rate": 0.0001999999613435087,
-      "loss": 46.0,
-      "step": 1748
-    },
-    {
-      "epoch": 0.2816538507991465,
-      "grad_norm": 0.00044605176663026214,
-      "learning_rate": 0.000199999961299012,
-      "loss": 46.0,
-      "step": 1749
-    },
-    {
-      "epoch": 0.2818148878779339,
-      "grad_norm": 0.0008384662214666605,
-      "learning_rate": 0.00019999996125448974,
-      "loss": 46.0,
-      "step": 1750
-    },
-    {
-      "epoch": 0.2819759249567213,
-      "grad_norm": 0.0014187163906171918,
-      "learning_rate": 0.00019999996120994188,
-      "loss": 46.0,
-      "step": 1751
-    },
-    {
-      "epoch": 0.2821369620355087,
-      "grad_norm": 0.0004106323467567563,
-      "learning_rate": 0.00019999996116536843,
-      "loss": 46.0,
-      "step": 1752
-    },
-    {
-      "epoch": 0.2822979991142961,
-      "grad_norm": 0.000659812125377357,
-      "learning_rate": 0.00019999996112076936,
-      "loss": 46.0,
-      "step": 1753
-    },
-    {
-      "epoch": 0.28245903619308343,
-      "grad_norm": 0.00042445125291123986,
-      "learning_rate": 0.00019999996107614471,
-      "loss": 46.0,
-      "step": 1754
-    },
-    {
-      "epoch": 0.28262007327187083,
-      "grad_norm": 0.0009564983774907887,
-      "learning_rate": 0.00019999996103149448,
-      "loss": 46.0,
-      "step": 1755
-    },
-    {
-      "epoch": 0.2827811103506582,
-      "grad_norm": 0.0003712513134814799,
-      "learning_rate": 0.00019999996098681865,
-      "loss": 46.0,
-      "step": 1756
-    },
-    {
-      "epoch": 0.2829421474294456,
-      "grad_norm": 0.001536081195808947,
-      "learning_rate": 0.0001999999609421172,
-      "loss": 46.0,
-      "step": 1757
-    },
-    {
-      "epoch": 0.283103184508233,
-      "grad_norm": 0.0016187526052817702,
-      "learning_rate": 0.0001999999608973902,
-      "loss": 46.0,
-      "step": 1758
-    },
-    {
-      "epoch": 0.2832642215870204,
-      "grad_norm": 0.00040868844371289015,
-      "learning_rate": 0.00019999996085263755,
-      "loss": 46.0,
-      "step": 1759
-    },
-    {
-      "epoch": 0.2834252586658078,
-      "grad_norm": 0.00116222002543509,
-      "learning_rate": 0.00019999996080785935,
-      "loss": 46.0,
-      "step": 1760
-    },
-    {
-      "epoch": 0.2835862957445952,
-      "grad_norm": 0.00030192965641617775,
-      "learning_rate": 0.0001999999607630555,
-      "loss": 46.0,
-      "step": 1761
-    },
-    {
-      "epoch": 0.2837473328233826,
-      "grad_norm": 0.0005130195058882236,
-      "learning_rate": 0.00019999996071822613,
-      "loss": 46.0,
-      "step": 1762
-    },
-    {
-      "epoch": 0.28390836990217,
-      "grad_norm": 0.0002736506867222488,
-      "learning_rate": 0.00019999996067337112,
-      "loss": 46.0,
-      "step": 1763
-    },
-    {
-      "epoch": 0.28406940698095734,
-      "grad_norm": 0.00039211285184137523,
-      "learning_rate": 0.00019999996062849054,
-      "loss": 46.0,
-      "step": 1764
-    },
-    {
-      "epoch": 0.28423044405974474,
-      "grad_norm": 0.0006182106444612145,
-      "learning_rate": 0.00019999996058358432,
-      "loss": 46.0,
-      "step": 1765
-    },
-    {
-      "epoch": 0.28439148113853213,
-      "grad_norm": 0.00035284028854221106,
-      "learning_rate": 0.00019999996053865254,
-      "loss": 46.0,
-      "step": 1766
-    },
-    {
-      "epoch": 0.28455251821731953,
-      "grad_norm": 0.00034616593620739877,
-      "learning_rate": 0.00019999996049369515,
-      "loss": 46.0,
-      "step": 1767
-    },
-    {
-      "epoch": 0.2847135552961069,
-      "grad_norm": 0.0005541214486584067,
-      "learning_rate": 0.0001999999604487122,
-      "loss": 46.0,
-      "step": 1768
-    },
-    {
-      "epoch": 0.2848745923748943,
-      "grad_norm": 0.00048042586422525346,
-      "learning_rate": 0.00019999996040370363,
-      "loss": 46.0,
-      "step": 1769
-    },
-    {
-      "epoch": 0.2850356294536817,
-      "grad_norm": 0.0005554311792366207,
-      "learning_rate": 0.00019999996035866943,
-      "loss": 46.0,
-      "step": 1770
-    },
-    {
-      "epoch": 0.2851966665324691,
-      "grad_norm": 0.00022341113071888685,
-      "learning_rate": 0.00019999996031360971,
-      "loss": 46.0,
-      "step": 1771
-    },
-    {
-      "epoch": 0.2853577036112565,
-      "grad_norm": 0.0006562733906321228,
-      "learning_rate": 0.00019999996026852433,
-      "loss": 46.0,
-      "step": 1772
-    },
-    {
-      "epoch": 0.2855187406900439,
-      "grad_norm": 0.0004613788623828441,
-      "learning_rate": 0.0001999999602234134,
-      "loss": 46.0,
-      "step": 1773
-    },
-    {
-      "epoch": 0.2856797777688313,
-      "grad_norm": 0.0006948672817088664,
-      "learning_rate": 0.00019999996017827683,
-      "loss": 46.0,
-      "step": 1774
-    },
-    {
-      "epoch": 0.28584081484761864,
-      "grad_norm": 0.0021478149574249983,
-      "learning_rate": 0.00019999996013311472,
-      "loss": 46.0,
-      "step": 1775
-    },
-    {
-      "epoch": 0.28600185192640604,
-      "grad_norm": 0.0006172028952278197,
-      "learning_rate": 0.00019999996008792696,
-      "loss": 46.0,
-      "step": 1776
-    },
-    {
-      "epoch": 0.28616288900519343,
-      "grad_norm": 0.001169553492218256,
-      "learning_rate": 0.00019999996004271364,
-      "loss": 46.0,
-      "step": 1777
-    },
-    {
-      "epoch": 0.28632392608398083,
-      "grad_norm": 0.0011354945600032806,
-      "learning_rate": 0.0001999999599974747,
-      "loss": 46.0,
-      "step": 1778
-    },
-    {
-      "epoch": 0.2864849631627682,
-      "grad_norm": 0.0019422700861468911,
-      "learning_rate": 0.0001999999599522102,
-      "loss": 46.0,
-      "step": 1779
-    },
-    {
-      "epoch": 0.2866460002415556,
-      "grad_norm": 0.00022413030092138797,
-      "learning_rate": 0.00019999995990692008,
-      "loss": 46.0,
-      "step": 1780
-    },
-    {
-      "epoch": 0.286807037320343,
-      "grad_norm": 0.001351299462839961,
-      "learning_rate": 0.0001999999598616044,
-      "loss": 46.0,
-      "step": 1781
-    },
-    {
-      "epoch": 0.2869680743991304,
-      "grad_norm": 0.00043627008562907577,
-      "learning_rate": 0.00019999995981626306,
-      "loss": 46.0,
-      "step": 1782
-    },
-    {
-      "epoch": 0.2871291114779178,
-      "grad_norm": 0.00038289171061478555,
-      "learning_rate": 0.00019999995977089616,
-      "loss": 46.0,
-      "step": 1783
-    },
-    {
-      "epoch": 0.2872901485567052,
-      "grad_norm": 0.0005784197128377855,
-      "learning_rate": 0.00019999995972550365,
-      "loss": 46.0,
-      "step": 1784
-    },
-    {
-      "epoch": 0.28745118563549255,
-      "grad_norm": 0.0012538628652691841,
-      "learning_rate": 0.00019999995968008558,
-      "loss": 46.0,
-      "step": 1785
-    },
-    {
-      "epoch": 0.28761222271427994,
-      "grad_norm": 0.00039487218600697815,
-      "learning_rate": 0.0001999999596346419,
-      "loss": 46.0,
-      "step": 1786
-    },
-    {
-      "epoch": 0.28777325979306734,
-      "grad_norm": 0.0005266358493827283,
-      "learning_rate": 0.0001999999595891726,
-      "loss": 46.0,
-      "step": 1787
-    },
-    {
-      "epoch": 0.28793429687185473,
-      "grad_norm": 0.0002815193438436836,
-      "learning_rate": 0.00019999995954367774,
-      "loss": 46.0,
-      "step": 1788
-    },
-    {
-      "epoch": 0.28809533395064213,
-      "grad_norm": 0.0004312753735575825,
-      "learning_rate": 0.0001999999594981573,
-      "loss": 46.0,
-      "step": 1789
-    },
-    {
-      "epoch": 0.2882563710294295,
-      "grad_norm": 0.0004867517272941768,
-      "learning_rate": 0.0001999999594526112,
-      "loss": 46.0,
-      "step": 1790
-    },
-    {
-      "epoch": 0.2884174081082169,
-      "grad_norm": 0.0015913225943222642,
-      "learning_rate": 0.00019999995940703954,
-      "loss": 46.0,
-      "step": 1791
-    },
-    {
-      "epoch": 0.2885784451870043,
-      "grad_norm": 0.0006699271034449339,
-      "learning_rate": 0.00019999995936144228,
-      "loss": 46.0,
-      "step": 1792
-    },
-    {
-      "epoch": 0.2887394822657917,
-      "grad_norm": 0.0006766504375264049,
-      "learning_rate": 0.00019999995931581946,
-      "loss": 46.0,
-      "step": 1793
-    },
-    {
-      "epoch": 0.2889005193445791,
-      "grad_norm": 0.00027622951893135905,
-      "learning_rate": 0.00019999995927017102,
-      "loss": 46.0,
-      "step": 1794
-    },
-    {
-      "epoch": 0.2890615564233665,
-      "grad_norm": 0.0013221491826698184,
-      "learning_rate": 0.00019999995922449697,
-      "loss": 46.0,
-      "step": 1795
-    },
-    {
-      "epoch": 0.28922259350215385,
-      "grad_norm": 0.0005121470312587917,
-      "learning_rate": 0.00019999995917879734,
-      "loss": 46.0,
-      "step": 1796
-    },
-    {
-      "epoch": 0.28938363058094124,
-      "grad_norm": 0.0005169783835299313,
-      "learning_rate": 0.0001999999591330721,
-      "loss": 46.0,
-      "step": 1797
-    },
-    {
-      "epoch": 0.28954466765972864,
-      "grad_norm": 0.001369944540783763,
-      "learning_rate": 0.0001999999590873213,
-      "loss": 46.0,
-      "step": 1798
-    },
-    {
-      "epoch": 0.28970570473851603,
-      "grad_norm": 0.00030911123030819,
-      "learning_rate": 0.00019999995904154488,
-      "loss": 46.0,
-      "step": 1799
-    },
-    {
-      "epoch": 0.28986674181730343,
-      "grad_norm": 0.0007034396403469145,
-      "learning_rate": 0.00019999995899574286,
-      "loss": 46.0,
-      "step": 1800
-    },
-    {
-      "epoch": 0.2900277788960908,
-      "grad_norm": 0.0007518244092352688,
-      "learning_rate": 0.00019999995894991526,
-      "loss": 46.0,
-      "step": 1801
-    },
-    {
-      "epoch": 0.2901888159748782,
-      "grad_norm": 0.0005977744003757834,
-      "learning_rate": 0.00019999995890406205,
-      "loss": 46.0,
-      "step": 1802
-    },
-    {
-      "epoch": 0.2903498530536656,
-      "grad_norm": 0.0005462935660034418,
-      "learning_rate": 0.00019999995885818325,
-      "loss": 46.0,
-      "step": 1803
-    },
-    {
-      "epoch": 0.290510890132453,
-      "grad_norm": 0.0005249497480690479,
-      "learning_rate": 0.00019999995881227886,
-      "loss": 46.0,
-      "step": 1804
-    },
-    {
-      "epoch": 0.2906719272112404,
-      "grad_norm": 0.000635012227576226,
-      "learning_rate": 0.00019999995876634886,
-      "loss": 46.0,
-      "step": 1805
-    },
-    {
-      "epoch": 0.29083296429002775,
-      "grad_norm": 0.0003316863439977169,
-      "learning_rate": 0.0001999999587203933,
-      "loss": 46.0,
-      "step": 1806
-    },
-    {
-      "epoch": 0.29099400136881515,
-      "grad_norm": 0.0018930124351754785,
-      "learning_rate": 0.00019999995867441212,
-      "loss": 46.0,
-      "step": 1807
-    },
-    {
-      "epoch": 0.29115503844760254,
-      "grad_norm": 0.0013742187293246388,
-      "learning_rate": 0.00019999995862840536,
-      "loss": 46.0,
-      "step": 1808
-    },
-    {
-      "epoch": 0.29131607552638994,
-      "grad_norm": 0.0005785526591353118,
-      "learning_rate": 0.000199999958582373,
-      "loss": 46.0,
-      "step": 1809
-    },
-    {
-      "epoch": 0.29147711260517734,
-      "grad_norm": 0.0006674519390799105,
-      "learning_rate": 0.00019999995853631504,
-      "loss": 46.0,
-      "step": 1810
-    },
-    {
-      "epoch": 0.29163814968396473,
-      "grad_norm": 0.0011885145213454962,
-      "learning_rate": 0.00019999995849023146,
-      "loss": 46.0,
-      "step": 1811
-    },
-    {
-      "epoch": 0.2917991867627521,
-      "grad_norm": 0.0010971532901749015,
-      "learning_rate": 0.00019999995844412232,
-      "loss": 46.0,
-      "step": 1812
-    },
-    {
-      "epoch": 0.2919602238415395,
-      "grad_norm": 0.0006365017616190016,
-      "learning_rate": 0.00019999995839798757,
-      "loss": 46.0,
-      "step": 1813
-    },
-    {
-      "epoch": 0.2921212609203269,
-      "grad_norm": 0.00039623305201530457,
-      "learning_rate": 0.00019999995835182722,
-      "loss": 46.0,
-      "step": 1814
-    },
-    {
-      "epoch": 0.2922822979991143,
-      "grad_norm": 0.0014342054491862655,
-      "learning_rate": 0.0001999999583056413,
-      "loss": 46.0,
-      "step": 1815
-    },
-    {
-      "epoch": 0.2924433350779017,
-      "grad_norm": 0.00039575566188432276,
-      "learning_rate": 0.00019999995825942975,
-      "loss": 46.0,
-      "step": 1816
-    },
-    {
-      "epoch": 0.29260437215668905,
-      "grad_norm": 0.0017069606110453606,
-      "learning_rate": 0.00019999995821319265,
-      "loss": 46.0,
-      "step": 1817
-    },
-    {
-      "epoch": 0.29276540923547645,
-      "grad_norm": 0.0016336991684511304,
-      "learning_rate": 0.00019999995816692993,
-      "loss": 46.0,
-      "step": 1818
-    },
-    {
-      "epoch": 0.29292644631426384,
-      "grad_norm": 0.00032223513699136674,
-      "learning_rate": 0.0001999999581206416,
-      "loss": 46.0,
-      "step": 1819
-    },
-    {
-      "epoch": 0.29308748339305124,
-      "grad_norm": 0.00021341664250940084,
-      "learning_rate": 0.00019999995807432768,
-      "loss": 46.0,
-      "step": 1820
-    },
-    {
-      "epoch": 0.29324852047183864,
-      "grad_norm": 0.0006592419813387096,
-      "learning_rate": 0.00019999995802798818,
-      "loss": 46.0,
-      "step": 1821
-    },
-    {
-      "epoch": 0.29340955755062603,
-      "grad_norm": 0.000712876149918884,
-      "learning_rate": 0.00019999995798162309,
-      "loss": 46.0,
-      "step": 1822
-    },
-    {
-      "epoch": 0.29357059462941343,
-      "grad_norm": 0.00023029805743135512,
-      "learning_rate": 0.0001999999579352324,
-      "loss": 46.0,
-      "step": 1823
-    },
-    {
-      "epoch": 0.2937316317082008,
-      "grad_norm": 0.0007901446660980582,
-      "learning_rate": 0.0001999999578888161,
-      "loss": 46.0,
-      "step": 1824
-    },
-    {
-      "epoch": 0.2938926687869882,
-      "grad_norm": 0.0005559787969104946,
-      "learning_rate": 0.00019999995784237423,
-      "loss": 46.0,
-      "step": 1825
-    },
-    {
-      "epoch": 0.2940537058657756,
-      "grad_norm": 0.00031766644679009914,
-      "learning_rate": 0.00019999995779590676,
-      "loss": 46.0,
-      "step": 1826
-    },
-    {
-      "epoch": 0.294214742944563,
-      "grad_norm": 0.0003780285478569567,
-      "learning_rate": 0.00019999995774941368,
-      "loss": 46.0,
-      "step": 1827
-    },
-    {
-      "epoch": 0.29437578002335035,
-      "grad_norm": 0.0007771283271722496,
-      "learning_rate": 0.000199999957702895,
-      "loss": 46.0,
-      "step": 1828
-    },
-    {
-      "epoch": 0.29453681710213775,
-      "grad_norm": 0.0005054048378951848,
-      "learning_rate": 0.00019999995765635075,
-      "loss": 46.0,
-      "step": 1829
-    },
-    {
-      "epoch": 0.29469785418092515,
-      "grad_norm": 0.0005131353973411024,
-      "learning_rate": 0.0001999999576097809,
-      "loss": 46.0,
-      "step": 1830
-    },
-    {
-      "epoch": 0.29485889125971254,
-      "grad_norm": 0.00022505295055452734,
-      "learning_rate": 0.00019999995756318545,
-      "loss": 46.0,
-      "step": 1831
-    },
-    {
-      "epoch": 0.29501992833849994,
-      "grad_norm": 0.00022824318148195744,
-      "learning_rate": 0.0001999999575165644,
-      "loss": 46.0,
-      "step": 1832
-    },
-    {
-      "epoch": 0.29518096541728733,
-      "grad_norm": 0.0005779191269539297,
-      "learning_rate": 0.00019999995746991774,
-      "loss": 46.0,
-      "step": 1833
-    },
-    {
-      "epoch": 0.29534200249607473,
-      "grad_norm": 0.0006373411160893738,
-      "learning_rate": 0.00019999995742324552,
-      "loss": 46.0,
-      "step": 1834
-    },
-    {
-      "epoch": 0.2955030395748621,
-      "grad_norm": 0.00031253762426786125,
-      "learning_rate": 0.0001999999573765477,
-      "loss": 46.0,
-      "step": 1835
-    },
-    {
-      "epoch": 0.2956640766536495,
-      "grad_norm": 0.0007603811682201922,
-      "learning_rate": 0.00019999995732982427,
-      "loss": 46.0,
-      "step": 1836
-    },
-    {
-      "epoch": 0.2958251137324369,
-      "grad_norm": 0.0003260927915107459,
-      "learning_rate": 0.00019999995728307523,
-      "loss": 46.0,
-      "step": 1837
-    },
-    {
-      "epoch": 0.29598615081122426,
-      "grad_norm": 0.0005162190063856542,
-      "learning_rate": 0.00019999995723630064,
-      "loss": 46.0,
-      "step": 1838
-    },
-    {
-      "epoch": 0.29614718789001165,
-      "grad_norm": 0.0004505754041019827,
-      "learning_rate": 0.00019999995718950043,
-      "loss": 46.0,
-      "step": 1839
-    },
-    {
-      "epoch": 0.29630822496879905,
-      "grad_norm": 0.0008108963374979794,
-      "learning_rate": 0.0001999999571426746,
-      "loss": 46.0,
-      "step": 1840
-    },
-    {
-      "epoch": 0.29646926204758645,
-      "grad_norm": 0.0007972571766003966,
-      "learning_rate": 0.0001999999570958232,
-      "loss": 46.0,
-      "step": 1841
-    },
-    {
-      "epoch": 0.29663029912637384,
-      "grad_norm": 0.0012649011332541704,
-      "learning_rate": 0.00019999995704894622,
-      "loss": 46.0,
-      "step": 1842
-    },
-    {
-      "epoch": 0.29679133620516124,
-      "grad_norm": 0.0007034620502963662,
-      "learning_rate": 0.0001999999570020436,
-      "loss": 46.0,
-      "step": 1843
-    },
-    {
-      "epoch": 0.29695237328394863,
-      "grad_norm": 0.0008080760599114001,
-      "learning_rate": 0.00019999995695511543,
-      "loss": 46.0,
-      "step": 1844
-    },
-    {
-      "epoch": 0.29711341036273603,
-      "grad_norm": 0.00039131162338890135,
-      "learning_rate": 0.00019999995690816164,
-      "loss": 46.0,
-      "step": 1845
-    },
-    {
-      "epoch": 0.2972744474415234,
-      "grad_norm": 0.0004916760371997952,
-      "learning_rate": 0.00019999995686118227,
-      "loss": 46.0,
-      "step": 1846
-    },
-    {
-      "epoch": 0.2974354845203108,
-      "grad_norm": 0.0005050976760685444,
-      "learning_rate": 0.0001999999568141773,
-      "loss": 46.0,
-      "step": 1847
-    },
-    {
-      "epoch": 0.2975965215990982,
-      "grad_norm": 0.00033225907827727497,
-      "learning_rate": 0.00019999995676714673,
-      "loss": 46.0,
-      "step": 1848
-    },
-    {
-      "epoch": 0.29775755867788556,
-      "grad_norm": 0.0008810764993540943,
-      "learning_rate": 0.00019999995672009057,
-      "loss": 46.0,
-      "step": 1849
-    },
-    {
-      "epoch": 0.29791859575667295,
-      "grad_norm": 0.0004395331779960543,
-      "learning_rate": 0.00019999995667300882,
-      "loss": 46.0,
-      "step": 1850
-    },
-    {
-      "epoch": 0.29807963283546035,
-      "grad_norm": 0.00041944763506762683,
-      "learning_rate": 0.00019999995662590148,
-      "loss": 46.0,
-      "step": 1851
-    },
-    {
-      "epoch": 0.29824066991424775,
-      "grad_norm": 0.00198348262347281,
-      "learning_rate": 0.0001999999565787685,
-      "loss": 46.0,
-      "step": 1852
-    },
-    {
-      "epoch": 0.29840170699303514,
-      "grad_norm": 0.0006780798430554569,
-      "learning_rate": 0.00019999995653161,
-      "loss": 46.0,
-      "step": 1853
-    },
-    {
-      "epoch": 0.29856274407182254,
-      "grad_norm": 0.0005949756596237421,
-      "learning_rate": 0.00019999995648442584,
-      "loss": 46.0,
-      "step": 1854
-    },
-    {
-      "epoch": 0.29872378115060993,
-      "grad_norm": 0.00024164844944607466,
-      "learning_rate": 0.00019999995643721612,
-      "loss": 46.0,
-      "step": 1855
-    },
-    {
-      "epoch": 0.29888481822939733,
-      "grad_norm": 0.0006569984834641218,
-      "learning_rate": 0.00019999995638998082,
-      "loss": 46.0,
-      "step": 1856
-    },
-    {
-      "epoch": 0.2990458553081847,
-      "grad_norm": 0.0007791115203872323,
-      "learning_rate": 0.00019999995634271985,
-      "loss": 46.0,
-      "step": 1857
-    },
-    {
-      "epoch": 0.2992068923869721,
-      "grad_norm": 0.0005597388371825218,
-      "learning_rate": 0.00019999995629543338,
-      "loss": 46.0,
-      "step": 1858
-    },
-    {
-      "epoch": 0.29936792946575946,
-      "grad_norm": 0.0009192762663587928,
-      "learning_rate": 0.00019999995624812126,
-      "loss": 46.0,
-      "step": 1859
-    },
-    {
-      "epoch": 0.29952896654454686,
-      "grad_norm": 0.0004769730439875275,
-      "learning_rate": 0.00019999995620078356,
-      "loss": 46.0,
-      "step": 1860
-    },
-    {
-      "epoch": 0.29969000362333426,
-      "grad_norm": 0.0004369158123154193,
-      "learning_rate": 0.00019999995615342027,
-      "loss": 46.0,
-      "step": 1861
-    },
-    {
-      "epoch": 0.29985104070212165,
-      "grad_norm": 0.0014577005058526993,
-      "learning_rate": 0.00019999995610603136,
-      "loss": 46.0,
-      "step": 1862
-    },
-    {
-      "epoch": 0.30001207778090905,
-      "grad_norm": 0.00032134598586708307,
-      "learning_rate": 0.00019999995605861687,
-      "loss": 46.0,
-      "step": 1863
-    },
-    {
-      "epoch": 0.30017311485969644,
-      "grad_norm": 0.00032133038621395826,
-      "learning_rate": 0.0001999999560111768,
-      "loss": 46.0,
-      "step": 1864
-    },
-    {
-      "epoch": 0.30033415193848384,
-      "grad_norm": 0.0007147595752030611,
-      "learning_rate": 0.00019999995596371113,
-      "loss": 46.0,
-      "step": 1865
-    },
-    {
-      "epoch": 0.30049518901727124,
-      "grad_norm": 0.0004683503066189587,
-      "learning_rate": 0.00019999995591621982,
-      "loss": 46.0,
-      "step": 1866
-    },
-    {
-      "epoch": 0.30065622609605863,
-      "grad_norm": 0.0004294590326026082,
-      "learning_rate": 0.00019999995586870298,
-      "loss": 46.0,
-      "step": 1867
-    },
-    {
-      "epoch": 0.30081726317484603,
-      "grad_norm": 0.0006481785676442087,
-      "learning_rate": 0.00019999995582116052,
-      "loss": 46.0,
-      "step": 1868
-    },
-    {
-      "epoch": 0.3009783002536334,
-      "grad_norm": 0.00149905018042773,
-      "learning_rate": 0.00019999995577359245,
-      "loss": 46.0,
-      "step": 1869
-    },
-    {
-      "epoch": 0.30113933733242076,
-      "grad_norm": 0.0002647887449711561,
-      "learning_rate": 0.0001999999557259988,
-      "loss": 46.0,
-      "step": 1870
-    },
-    {
-      "epoch": 0.30130037441120816,
-      "grad_norm": 0.0018763943808153272,
-      "learning_rate": 0.00019999995567837955,
-      "loss": 46.0,
-      "step": 1871
-    },
-    {
-      "epoch": 0.30146141148999556,
-      "grad_norm": 0.0003123406204394996,
-      "learning_rate": 0.00019999995563073472,
-      "loss": 46.0,
-      "step": 1872
-    },
-    {
-      "epoch": 0.30162244856878295,
-      "grad_norm": 0.00041101162787526846,
-      "learning_rate": 0.00019999995558306428,
-      "loss": 46.0,
-      "step": 1873
-    },
-    {
-      "epoch": 0.30178348564757035,
-      "grad_norm": 0.00036199679016135633,
-      "learning_rate": 0.00019999995553536824,
-      "loss": 46.0,
-      "step": 1874
-    },
-    {
-      "epoch": 0.30194452272635774,
-      "grad_norm": 0.0006806057062931359,
-      "learning_rate": 0.0001999999554876466,
-      "loss": 46.0,
-      "step": 1875
-    },
-    {
-      "epoch": 0.30210555980514514,
-      "grad_norm": 0.0007571068708784878,
-      "learning_rate": 0.0001999999554398994,
-      "loss": 46.0,
-      "step": 1876
-    },
-    {
-      "epoch": 0.30226659688393254,
-      "grad_norm": 0.00042474898509681225,
-      "learning_rate": 0.00019999995539212657,
-      "loss": 46.0,
-      "step": 1877
-    },
-    {
-      "epoch": 0.30242763396271993,
-      "grad_norm": 0.00032792898127809167,
-      "learning_rate": 0.0001999999553443282,
-      "loss": 46.0,
-      "step": 1878
-    },
-    {
-      "epoch": 0.30258867104150733,
-      "grad_norm": 0.0013749575009569526,
-      "learning_rate": 0.00019999995529650417,
-      "loss": 46.0,
-      "step": 1879
-    },
-    {
-      "epoch": 0.3027497081202947,
-      "grad_norm": 0.00045742292422801256,
-      "learning_rate": 0.00019999995524865459,
-      "loss": 46.0,
-      "step": 1880
-    },
-    {
-      "epoch": 0.30291074519908207,
-      "grad_norm": 0.0005735918530263007,
-      "learning_rate": 0.00019999995520077936,
-      "loss": 46.0,
-      "step": 1881
-    },
-    {
-      "epoch": 0.30307178227786946,
-      "grad_norm": 0.00024322968965861946,
-      "learning_rate": 0.00019999995515287858,
-      "loss": 46.0,
-      "step": 1882
-    },
-    {
-      "epoch": 0.30323281935665686,
-      "grad_norm": 0.0006605838425457478,
-      "learning_rate": 0.0001999999551049522,
-      "loss": 46.0,
-      "step": 1883
-    },
-    {
-      "epoch": 0.30339385643544425,
-      "grad_norm": 0.0010462775826454163,
-      "learning_rate": 0.0001999999550570002,
-      "loss": 46.0,
-      "step": 1884
-    },
-    {
-      "epoch": 0.30355489351423165,
-      "grad_norm": 0.0007160481764003634,
-      "learning_rate": 0.00019999995500902265,
-      "loss": 46.0,
-      "step": 1885
-    },
-    {
-      "epoch": 0.30371593059301905,
-      "grad_norm": 0.00022240905673243105,
-      "learning_rate": 0.00019999995496101946,
-      "loss": 46.0,
-      "step": 1886
-    },
-    {
-      "epoch": 0.30387696767180644,
-      "grad_norm": 0.00023290369426831603,
-      "learning_rate": 0.00019999995491299072,
-      "loss": 46.0,
-      "step": 1887
-    },
-    {
-      "epoch": 0.30403800475059384,
-      "grad_norm": 0.0012489889049902558,
-      "learning_rate": 0.00019999995486493635,
-      "loss": 46.0,
-      "step": 1888
-    },
-    {
-      "epoch": 0.30419904182938123,
-      "grad_norm": 0.00040387638728134334,
-      "learning_rate": 0.00019999995481685638,
-      "loss": 46.0,
-      "step": 1889
-    },
-    {
-      "epoch": 0.30436007890816863,
-      "grad_norm": 0.0003309334279038012,
-      "learning_rate": 0.00019999995476875084,
-      "loss": 46.0,
-      "step": 1890
-    },
-    {
-      "epoch": 0.30452111598695597,
-      "grad_norm": 0.00021112282411195338,
-      "learning_rate": 0.0001999999547206197,
-      "loss": 46.0,
-      "step": 1891
-    },
-    {
-      "epoch": 0.30468215306574337,
-      "grad_norm": 0.0014652787940576673,
-      "learning_rate": 0.00019999995467246296,
-      "loss": 46.0,
-      "step": 1892
-    },
-    {
-      "epoch": 0.30484319014453076,
-      "grad_norm": 0.0016692185308784246,
-      "learning_rate": 0.00019999995462428063,
-      "loss": 46.0,
-      "step": 1893
-    },
-    {
-      "epoch": 0.30500422722331816,
-      "grad_norm": 0.0006531266262754798,
-      "learning_rate": 0.0001999999545760727,
-      "loss": 46.0,
-      "step": 1894
-    },
-    {
-      "epoch": 0.30516526430210555,
-      "grad_norm": 0.001210064860060811,
-      "learning_rate": 0.00019999995452783917,
-      "loss": 46.0,
-      "step": 1895
-    },
-    {
-      "epoch": 0.30532630138089295,
-      "grad_norm": 0.0005247330409474671,
-      "learning_rate": 0.00019999995447958003,
-      "loss": 46.0,
-      "step": 1896
-    },
-    {
-      "epoch": 0.30548733845968035,
-      "grad_norm": 0.0012705554254353046,
-      "learning_rate": 0.00019999995443129533,
-      "loss": 46.0,
-      "step": 1897
-    },
-    {
-      "epoch": 0.30564837553846774,
-      "grad_norm": 0.0013942826772108674,
-      "learning_rate": 0.00019999995438298504,
-      "loss": 46.0,
-      "step": 1898
-    },
-    {
-      "epoch": 0.30580941261725514,
-      "grad_norm": 0.0005664772470481694,
-      "learning_rate": 0.00019999995433464914,
-      "loss": 46.0,
-      "step": 1899
-    },
-    {
-      "epoch": 0.30597044969604253,
-      "grad_norm": 0.0010166022693738341,
-      "learning_rate": 0.00019999995428628763,
-      "loss": 46.0,
-      "step": 1900
-    },
-    {
-      "epoch": 0.30613148677482993,
-      "grad_norm": 0.00048385770060122013,
-      "learning_rate": 0.00019999995423790055,
-      "loss": 46.0,
-      "step": 1901
-    },
-    {
-      "epoch": 0.30629252385361727,
-      "grad_norm": 0.0007341625750996172,
-      "learning_rate": 0.00019999995418948786,
-      "loss": 46.0,
-      "step": 1902
-    },
-    {
-      "epoch": 0.30645356093240467,
-      "grad_norm": 0.0019976783078163862,
-      "learning_rate": 0.00019999995414104958,
-      "loss": 46.0,
-      "step": 1903
-    },
-    {
-      "epoch": 0.30661459801119206,
-      "grad_norm": 0.0008493874338455498,
-      "learning_rate": 0.0001999999540925857,
-      "loss": 46.0,
-      "step": 1904
-    },
-    {
-      "epoch": 0.30677563508997946,
-      "grad_norm": 0.0024145091883838177,
-      "learning_rate": 0.00019999995404409621,
-      "loss": 46.0,
-      "step": 1905
-    },
-    {
-      "epoch": 0.30693667216876686,
-      "grad_norm": 0.00027187468367628753,
-      "learning_rate": 0.00019999995399558115,
-      "loss": 46.0,
-      "step": 1906
-    },
-    {
-      "epoch": 0.30709770924755425,
-      "grad_norm": 0.000542456575203687,
-      "learning_rate": 0.0001999999539470405,
-      "loss": 46.0,
-      "step": 1907
-    },
-    {
-      "epoch": 0.30725874632634165,
-      "grad_norm": 0.00044663192238658667,
-      "learning_rate": 0.00019999995389847423,
-      "loss": 46.0,
-      "step": 1908
-    },
-    {
-      "epoch": 0.30741978340512904,
-      "grad_norm": 0.0008326354436576366,
-      "learning_rate": 0.00019999995384988237,
-      "loss": 46.0,
-      "step": 1909
-    },
-    {
-      "epoch": 0.30758082048391644,
-      "grad_norm": 0.0005407662829384208,
-      "learning_rate": 0.00019999995380126496,
-      "loss": 46.0,
-      "step": 1910
-    },
-    {
-      "epoch": 0.30774185756270384,
-      "grad_norm": 0.0008776931208558381,
-      "learning_rate": 0.0001999999537526219,
-      "loss": 46.0,
-      "step": 1911
-    },
-    {
-      "epoch": 0.3079028946414912,
-      "grad_norm": 0.00022998114582151175,
-      "learning_rate": 0.0001999999537039533,
-      "loss": 46.0,
-      "step": 1912
-    },
-    {
-      "epoch": 0.30806393172027857,
-      "grad_norm": 0.0004771912354044616,
-      "learning_rate": 0.00019999995365525903,
-      "loss": 46.0,
-      "step": 1913
-    },
-    {
-      "epoch": 0.30822496879906597,
-      "grad_norm": 0.0014869171427562833,
-      "learning_rate": 0.0001999999536065392,
-      "loss": 46.0,
-      "step": 1914
-    },
-    {
-      "epoch": 0.30838600587785336,
-      "grad_norm": 0.0011783756781369448,
-      "learning_rate": 0.0001999999535577938,
-      "loss": 46.0,
-      "step": 1915
-    },
-    {
-      "epoch": 0.30854704295664076,
-      "grad_norm": 0.001212630420923233,
-      "learning_rate": 0.00019999995350902276,
-      "loss": 46.0,
-      "step": 1916
-    },
-    {
-      "epoch": 0.30870808003542816,
-      "grad_norm": 0.0020968548487871885,
-      "learning_rate": 0.00019999995346022616,
-      "loss": 46.0,
-      "step": 1917
-    },
-    {
-      "epoch": 0.30886911711421555,
-      "grad_norm": 0.0010417501907795668,
-      "learning_rate": 0.00019999995341140396,
-      "loss": 46.0,
-      "step": 1918
-    },
-    {
-      "epoch": 0.30903015419300295,
-      "grad_norm": 0.0009635143796913326,
-      "learning_rate": 0.00019999995336255616,
-      "loss": 46.0,
-      "step": 1919
-    },
-    {
-      "epoch": 0.30919119127179034,
-      "grad_norm": 0.0006447298219427466,
-      "learning_rate": 0.00019999995331368276,
-      "loss": 46.0,
-      "step": 1920
-    },
-    {
-      "epoch": 0.30935222835057774,
-      "grad_norm": 0.0010088207200169563,
-      "learning_rate": 0.00019999995326478378,
-      "loss": 46.0,
-      "step": 1921
-    },
-    {
-      "epoch": 0.30951326542936514,
-      "grad_norm": 0.0005225542699918151,
-      "learning_rate": 0.00019999995321585918,
-      "loss": 46.0,
-      "step": 1922
-    },
-    {
-      "epoch": 0.3096743025081525,
-      "grad_norm": 0.0003586713573895395,
-      "learning_rate": 0.000199999953166909,
-      "loss": 46.0,
-      "step": 1923
-    },
-    {
-      "epoch": 0.3098353395869399,
-      "grad_norm": 0.0003926061326637864,
-      "learning_rate": 0.00019999995311793326,
-      "loss": 46.0,
-      "step": 1924
-    },
-    {
-      "epoch": 0.30999637666572727,
-      "grad_norm": 0.0008712766575627029,
-      "learning_rate": 0.00019999995306893187,
-      "loss": 46.0,
-      "step": 1925
-    },
-    {
-      "epoch": 0.31015741374451467,
-      "grad_norm": 0.0004809441161341965,
-      "learning_rate": 0.0001999999530199049,
-      "loss": 46.0,
-      "step": 1926
-    },
-    {
-      "epoch": 0.31031845082330206,
-      "grad_norm": 0.000390014291042462,
-      "learning_rate": 0.00019999995297085234,
-      "loss": 46.0,
-      "step": 1927
-    },
-    {
-      "epoch": 0.31047948790208946,
-      "grad_norm": 0.0004524428804870695,
-      "learning_rate": 0.0001999999529217742,
-      "loss": 46.0,
-      "step": 1928
-    },
-    {
-      "epoch": 0.31064052498087685,
-      "grad_norm": 0.0022342661395668983,
-      "learning_rate": 0.00019999995287267044,
-      "loss": 46.0,
-      "step": 1929
-    },
-    {
-      "epoch": 0.31080156205966425,
-      "grad_norm": 0.0007858508615754545,
-      "learning_rate": 0.0001999999528235411,
-      "loss": 46.0,
-      "step": 1930
-    },
-    {
-      "epoch": 0.31096259913845165,
-      "grad_norm": 0.002166828140616417,
-      "learning_rate": 0.00019999995277438615,
-      "loss": 46.0,
-      "step": 1931
-    },
-    {
-      "epoch": 0.31112363621723904,
-      "grad_norm": 0.00030961562879383564,
-      "learning_rate": 0.0001999999527252056,
-      "loss": 46.0,
-      "step": 1932
-    },
-    {
-      "epoch": 0.31128467329602644,
-      "grad_norm": 0.0007051625871099532,
-      "learning_rate": 0.0001999999526759995,
-      "loss": 46.0,
-      "step": 1933
-    },
-    {
-      "epoch": 0.3114457103748138,
-      "grad_norm": 0.0006752362241968513,
-      "learning_rate": 0.00019999995262676777,
-      "loss": 46.0,
-      "step": 1934
-    },
-    {
-      "epoch": 0.3116067474536012,
-      "grad_norm": 0.0004780669114552438,
-      "learning_rate": 0.00019999995257751046,
-      "loss": 46.0,
-      "step": 1935
-    },
-    {
-      "epoch": 0.31176778453238857,
-      "grad_norm": 0.00041254935786128044,
-      "learning_rate": 0.00019999995252822754,
-      "loss": 46.0,
-      "step": 1936
-    },
-    {
-      "epoch": 0.31192882161117597,
-      "grad_norm": 0.001679743523709476,
-      "learning_rate": 0.00019999995247891903,
-      "loss": 46.0,
-      "step": 1937
-    },
-    {
-      "epoch": 0.31208985868996336,
-      "grad_norm": 0.0017879728693515062,
-      "learning_rate": 0.00019999995242958493,
-      "loss": 46.0,
-      "step": 1938
-    },
-    {
-      "epoch": 0.31225089576875076,
-      "grad_norm": 0.001382913556881249,
-      "learning_rate": 0.00019999995238022524,
-      "loss": 46.0,
-      "step": 1939
-    },
-    {
-      "epoch": 0.31241193284753815,
-      "grad_norm": 0.003139752196148038,
-      "learning_rate": 0.00019999995233083994,
-      "loss": 46.0,
-      "step": 1940
-    },
-    {
-      "epoch": 0.31257296992632555,
-      "grad_norm": 0.0009320085518993437,
-      "learning_rate": 0.00019999995228142905,
-      "loss": 46.0,
-      "step": 1941
-    },
-    {
-      "epoch": 0.31273400700511295,
-      "grad_norm": 0.0003911656385753304,
-      "learning_rate": 0.00019999995223199258,
-      "loss": 46.0,
-      "step": 1942
-    },
-    {
-      "epoch": 0.31289504408390034,
-      "grad_norm": 0.0010725856991484761,
-      "learning_rate": 0.00019999995218253052,
-      "loss": 46.0,
-      "step": 1943
-    },
-    {
-      "epoch": 0.3130560811626877,
-      "grad_norm": 0.00045514703379012644,
-      "learning_rate": 0.0001999999521330428,
-      "loss": 46.0,
-      "step": 1944
-    },
-    {
-      "epoch": 0.3132171182414751,
-      "grad_norm": 0.00111088075209409,
-      "learning_rate": 0.00019999995208352958,
-      "loss": 46.0,
-      "step": 1945
-    },
-    {
-      "epoch": 0.3133781553202625,
-      "grad_norm": 0.0006126022199168801,
-      "learning_rate": 0.0001999999520339907,
-      "loss": 46.0,
-      "step": 1946
-    },
-    {
-      "epoch": 0.31353919239904987,
-      "grad_norm": 0.000604628527071327,
-      "learning_rate": 0.00019999995198442626,
-      "loss": 46.0,
-      "step": 1947
-    },
-    {
-      "epoch": 0.31370022947783727,
-      "grad_norm": 0.0004172946501057595,
-      "learning_rate": 0.0001999999519348362,
-      "loss": 46.0,
-      "step": 1948
-    },
-    {
-      "epoch": 0.31386126655662466,
-      "grad_norm": 0.0006414538365788758,
-      "learning_rate": 0.00019999995188522057,
-      "loss": 46.0,
-      "step": 1949
-    },
-    {
-      "epoch": 0.31402230363541206,
-      "grad_norm": 0.0002925368899013847,
-      "learning_rate": 0.0001999999518355793,
-      "loss": 46.0,
-      "step": 1950
-    },
-    {
-      "epoch": 0.31418334071419945,
-      "grad_norm": 0.0005588184576481581,
-      "learning_rate": 0.00019999995178591247,
-      "loss": 46.0,
-      "step": 1951
-    },
-    {
-      "epoch": 0.31434437779298685,
-      "grad_norm": 0.0006587941315956414,
-      "learning_rate": 0.00019999995173622005,
-      "loss": 46.0,
-      "step": 1952
-    },
-    {
-      "epoch": 0.31450541487177425,
-      "grad_norm": 0.0017494851490482688,
-      "learning_rate": 0.000199999951686502,
-      "loss": 46.0,
-      "step": 1953
-    },
-    {
-      "epoch": 0.31466645195056164,
-      "grad_norm": 0.0011188029311597347,
-      "learning_rate": 0.0001999999516367584,
-      "loss": 46.0,
-      "step": 1954
-    },
-    {
-      "epoch": 0.314827489029349,
-      "grad_norm": 0.0006506229401566088,
-      "learning_rate": 0.00019999995158698919,
-      "loss": 46.0,
-      "step": 1955
-    },
-    {
-      "epoch": 0.3149885261081364,
-      "grad_norm": 0.0010482812067493796,
-      "learning_rate": 0.00019999995153719438,
-      "loss": 46.0,
-      "step": 1956
-    },
-    {
-      "epoch": 0.3151495631869238,
-      "grad_norm": 0.0005690926918759942,
-      "learning_rate": 0.00019999995148737396,
-      "loss": 46.0,
-      "step": 1957
-    },
-    {
-      "epoch": 0.31531060026571117,
-      "grad_norm": 0.0004758228315040469,
-      "learning_rate": 0.00019999995143752796,
-      "loss": 46.0,
-      "step": 1958
-    },
-    {
-      "epoch": 0.31547163734449857,
-      "grad_norm": 0.0005149694625288248,
-      "learning_rate": 0.00019999995138765637,
-      "loss": 46.0,
-      "step": 1959
-    },
-    {
-      "epoch": 0.31563267442328596,
-      "grad_norm": 0.0005577944102697074,
-      "learning_rate": 0.0001999999513377592,
-      "loss": 46.0,
-      "step": 1960
-    },
-    {
-      "epoch": 0.31579371150207336,
-      "grad_norm": 0.0010247377213090658,
-      "learning_rate": 0.0001999999512878364,
-      "loss": 46.0,
-      "step": 1961
-    },
-    {
-      "epoch": 0.31595474858086076,
-      "grad_norm": 0.0005274456925690174,
-      "learning_rate": 0.000199999951237888,
-      "loss": 46.0,
-      "step": 1962
-    },
-    {
-      "epoch": 0.31611578565964815,
-      "grad_norm": 0.0008289752877317369,
-      "learning_rate": 0.00019999995118791404,
-      "loss": 46.0,
-      "step": 1963
-    },
-    {
-      "epoch": 0.31627682273843555,
-      "grad_norm": 0.0006728868465870619,
-      "learning_rate": 0.0001999999511379145,
-      "loss": 46.0,
-      "step": 1964
-    },
-    {
-      "epoch": 0.3164378598172229,
-      "grad_norm": 0.00047442642971873283,
-      "learning_rate": 0.00019999995108788932,
-      "loss": 46.0,
-      "step": 1965
-    },
-    {
-      "epoch": 0.3165988968960103,
-      "grad_norm": 0.0007199350511655211,
-      "learning_rate": 0.00019999995103783856,
-      "loss": 46.0,
-      "step": 1966
-    },
-    {
-      "epoch": 0.3167599339747977,
-      "grad_norm": 0.0003270877932664007,
-      "learning_rate": 0.00019999995098776222,
-      "loss": 46.0,
-      "step": 1967
-    },
-    {
-      "epoch": 0.3169209710535851,
-      "grad_norm": 0.0014233592664822936,
-      "learning_rate": 0.00019999995093766026,
-      "loss": 46.0,
-      "step": 1968
-    },
-    {
-      "epoch": 0.3170820081323725,
-      "grad_norm": 0.0005007522413507104,
-      "learning_rate": 0.00019999995088753272,
-      "loss": 46.0,
-      "step": 1969
-    },
-    {
-      "epoch": 0.31724304521115987,
-      "grad_norm": 0.0006731337634846568,
-      "learning_rate": 0.00019999995083737958,
-      "loss": 46.0,
-      "step": 1970
-    },
-    {
-      "epoch": 0.31740408228994726,
-      "grad_norm": 0.0010688817128539085,
-      "learning_rate": 0.00019999995078720084,
-      "loss": 46.0,
-      "step": 1971
-    },
-    {
-      "epoch": 0.31756511936873466,
-      "grad_norm": 0.0017510128673166037,
-      "learning_rate": 0.0001999999507369965,
-      "loss": 46.0,
-      "step": 1972
-    },
-    {
-      "epoch": 0.31772615644752206,
-      "grad_norm": 0.000877570768352598,
-      "learning_rate": 0.0001999999506867666,
-      "loss": 46.0,
-      "step": 1973
-    },
-    {
-      "epoch": 0.31788719352630945,
-      "grad_norm": 0.00034507166128605604,
-      "learning_rate": 0.00019999995063651107,
-      "loss": 46.0,
-      "step": 1974
-    },
-    {
-      "epoch": 0.31804823060509685,
-      "grad_norm": 0.0005786537076346576,
-      "learning_rate": 0.00019999995058622995,
-      "loss": 46.0,
-      "step": 1975
-    },
-    {
-      "epoch": 0.3182092676838842,
-      "grad_norm": 0.0009048181818798184,
-      "learning_rate": 0.00019999995053592327,
-      "loss": 46.0,
-      "step": 1976
-    },
-    {
-      "epoch": 0.3183703047626716,
-      "grad_norm": 0.0012877408880740404,
-      "learning_rate": 0.00019999995048559094,
-      "loss": 46.0,
-      "step": 1977
-    },
-    {
-      "epoch": 0.318531341841459,
-      "grad_norm": 0.001697192550636828,
-      "learning_rate": 0.00019999995043523306,
-      "loss": 46.0,
-      "step": 1978
-    },
-    {
-      "epoch": 0.3186923789202464,
-      "grad_norm": 0.0004150728927925229,
-      "learning_rate": 0.0001999999503848496,
-      "loss": 46.0,
-      "step": 1979
-    },
-    {
-      "epoch": 0.3188534159990338,
-      "grad_norm": 0.0007999094668775797,
-      "learning_rate": 0.00019999995033444048,
-      "loss": 46.0,
-      "step": 1980
-    },
-    {
-      "epoch": 0.31901445307782117,
-      "grad_norm": 0.0008080822299234569,
-      "learning_rate": 0.0001999999502840058,
-      "loss": 46.0,
-      "step": 1981
-    },
-    {
-      "epoch": 0.31917549015660857,
-      "grad_norm": 0.0006414721719920635,
-      "learning_rate": 0.00019999995023354554,
-      "loss": 46.0,
-      "step": 1982
-    },
-    {
-      "epoch": 0.31933652723539596,
-      "grad_norm": 0.0012701860396191478,
-      "learning_rate": 0.00019999995018305967,
-      "loss": 46.0,
-      "step": 1983
-    },
-    {
-      "epoch": 0.31949756431418336,
-      "grad_norm": 0.0016254791989922523,
-      "learning_rate": 0.0001999999501325482,
-      "loss": 46.0,
-      "step": 1984
-    },
-    {
-      "epoch": 0.31965860139297075,
-      "grad_norm": 0.000467137957457453,
-      "learning_rate": 0.00019999995008201113,
-      "loss": 46.0,
-      "step": 1985
-    },
-    {
-      "epoch": 0.31981963847175815,
-      "grad_norm": 0.0004153347690589726,
-      "learning_rate": 0.00019999995003144847,
-      "loss": 46.0,
-      "step": 1986
-    },
-    {
-      "epoch": 0.3199806755505455,
-      "grad_norm": 0.000923548883292824,
-      "learning_rate": 0.00019999994998086022,
-      "loss": 46.0,
-      "step": 1987
-    },
-    {
-      "epoch": 0.3201417126293329,
-      "grad_norm": 0.0004752822278533131,
-      "learning_rate": 0.00019999994993024638,
-      "loss": 46.0,
-      "step": 1988
-    },
-    {
-      "epoch": 0.3203027497081203,
-      "grad_norm": 0.0006223291275091469,
-      "learning_rate": 0.00019999994987960693,
-      "loss": 46.0,
-      "step": 1989
-    },
-    {
-      "epoch": 0.3204637867869077,
-      "grad_norm": 0.0008064461289905012,
-      "learning_rate": 0.0001999999498289419,
-      "loss": 46.0,
-      "step": 1990
-    },
-    {
-      "epoch": 0.3206248238656951,
-      "grad_norm": 0.0010274579981341958,
-      "learning_rate": 0.00019999994977825127,
-      "loss": 46.0,
-      "step": 1991
-    },
-    {
-      "epoch": 0.32078586094448247,
-      "grad_norm": 0.0006910872762091458,
-      "learning_rate": 0.00019999994972753505,
-      "loss": 46.0,
-      "step": 1992
-    },
-    {
-      "epoch": 0.32094689802326987,
-      "grad_norm": 0.0013477810425683856,
-      "learning_rate": 0.00019999994967679323,
-      "loss": 46.0,
-      "step": 1993
-    },
-    {
-      "epoch": 0.32110793510205726,
-      "grad_norm": 0.00043071535765193403,
-      "learning_rate": 0.0001999999496260258,
-      "loss": 46.0,
-      "step": 1994
-    },
-    {
-      "epoch": 0.32126897218084466,
-      "grad_norm": 0.0011416514171287417,
-      "learning_rate": 0.0001999999495752328,
-      "loss": 46.0,
-      "step": 1995
-    },
-    {
-      "epoch": 0.32143000925963205,
-      "grad_norm": 0.0008783790399320424,
-      "learning_rate": 0.00019999994952441422,
-      "loss": 46.0,
-      "step": 1996
-    },
-    {
-      "epoch": 0.3215910463384194,
-      "grad_norm": 0.00042994358227588236,
-      "learning_rate": 0.00019999994947357,
-      "loss": 46.0,
-      "step": 1997
-    },
-    {
-      "epoch": 0.3217520834172068,
-      "grad_norm": 0.0008973407093435526,
-      "learning_rate": 0.0001999999494227002,
-      "loss": 46.0,
-      "step": 1998
-    },
-    {
-      "epoch": 0.3219131204959942,
-      "grad_norm": 0.00044807515223510563,
-      "learning_rate": 0.00019999994937180482,
-      "loss": 46.0,
-      "step": 1999
-    },
-    {
-      "epoch": 0.3220741575747816,
-      "grad_norm": 0.00028068304527550936,
-      "learning_rate": 0.00019999994932088383,
-      "loss": 46.0,
-      "step": 2000
-    },
-    {
-      "epoch": 0.322235194653569,
-      "grad_norm": 0.0004950006259605289,
-      "learning_rate": 0.00019999994926993725,
-      "loss": 46.0,
-      "step": 2001
-    },
-    {
-      "epoch": 0.3223962317323564,
-      "grad_norm": 0.0008484144927933812,
-      "learning_rate": 0.0001999999492189651,
-      "loss": 46.0,
-      "step": 2002
-    },
-    {
-      "epoch": 0.32255726881114377,
-      "grad_norm": 0.0011145814787596464,
-      "learning_rate": 0.0001999999491679673,
-      "loss": 46.0,
-      "step": 2003
-    },
-    {
-      "epoch": 0.32271830588993117,
-      "grad_norm": 0.000589332077652216,
-      "learning_rate": 0.00019999994911694394,
-      "loss": 46.0,
-      "step": 2004
-    },
-    {
-      "epoch": 0.32287934296871856,
-      "grad_norm": 0.0019545366521924734,
-      "learning_rate": 0.00019999994906589499,
-      "loss": 46.0,
-      "step": 2005
-    },
-    {
-      "epoch": 0.32304038004750596,
-      "grad_norm": 0.0008821808150969446,
-      "learning_rate": 0.00019999994901482044,
-      "loss": 46.0,
-      "step": 2006
-    },
-    {
-      "epoch": 0.32320141712629336,
-      "grad_norm": 0.0005089048063382506,
-      "learning_rate": 0.0001999999489637203,
-      "loss": 46.0,
-      "step": 2007
-    },
-    {
-      "epoch": 0.3233624542050807,
-      "grad_norm": 0.0002948503533843905,
-      "learning_rate": 0.00019999994891259454,
-      "loss": 46.0,
-      "step": 2008
-    },
-    {
-      "epoch": 0.3235234912838681,
-      "grad_norm": 0.0019477446330711246,
-      "learning_rate": 0.00019999994886144321,
-      "loss": 46.0,
-      "step": 2009
-    },
-    {
-      "epoch": 0.3236845283626555,
-      "grad_norm": 0.0003937899600714445,
-      "learning_rate": 0.0001999999488102663,
-      "loss": 46.0,
-      "step": 2010
-    },
-    {
-      "epoch": 0.3238455654414429,
-      "grad_norm": 0.00036119233118370175,
-      "learning_rate": 0.00019999994875906377,
-      "loss": 46.0,
-      "step": 2011
-    },
-    {
-      "epoch": 0.3240066025202303,
-      "grad_norm": 0.00027369082090444863,
-      "learning_rate": 0.00019999994870783562,
-      "loss": 46.0,
-      "step": 2012
-    },
-    {
-      "epoch": 0.3241676395990177,
-      "grad_norm": 0.0005119135603308678,
-      "learning_rate": 0.0001999999486565819,
-      "loss": 46.0,
-      "step": 2013
-    },
-    {
-      "epoch": 0.3243286766778051,
-      "grad_norm": 0.001278889481909573,
-      "learning_rate": 0.00019999994860530262,
-      "loss": 46.0,
-      "step": 2014
-    },
-    {
-      "epoch": 0.32448971375659247,
-      "grad_norm": 0.00034376498661004007,
-      "learning_rate": 0.0001999999485539977,
-      "loss": 46.0,
-      "step": 2015
-    },
-    {
-      "epoch": 0.32465075083537986,
-      "grad_norm": 0.0003255880146753043,
-      "learning_rate": 0.00019999994850266722,
-      "loss": 46.0,
-      "step": 2016
-    },
-    {
-      "epoch": 0.32481178791416726,
-      "grad_norm": 0.00047895527677610517,
-      "learning_rate": 0.00019999994845131108,
-      "loss": 46.0,
-      "step": 2017
-    },
-    {
-      "epoch": 0.3249728249929546,
-      "grad_norm": 0.0009383424185216427,
-      "learning_rate": 0.0001999999483999294,
-      "loss": 46.0,
-      "step": 2018
-    },
-    {
-      "epoch": 0.325133862071742,
-      "grad_norm": 0.0008149516070261598,
-      "learning_rate": 0.00019999994834852213,
-      "loss": 46.0,
-      "step": 2019
-    },
-    {
-      "epoch": 0.3252948991505294,
-      "grad_norm": 0.0013515559257939458,
-      "learning_rate": 0.00019999994829708926,
-      "loss": 46.0,
-      "step": 2020
-    },
-    {
-      "epoch": 0.3254559362293168,
-      "grad_norm": 0.000812262820545584,
-      "learning_rate": 0.00019999994824563077,
-      "loss": 46.0,
-      "step": 2021
-    },
-    {
-      "epoch": 0.3256169733081042,
-      "grad_norm": 0.0009132091654464602,
-      "learning_rate": 0.0001999999481941467,
-      "loss": 46.0,
-      "step": 2022
-    },
-    {
-      "epoch": 0.3257780103868916,
-      "grad_norm": 0.002700040815398097,
-      "learning_rate": 0.00019999994814263701,
-      "loss": 46.0,
-      "step": 2023
-    },
-    {
-      "epoch": 0.325939047465679,
-      "grad_norm": 0.0010764887556433678,
-      "learning_rate": 0.0001999999480911018,
-      "loss": 46.0,
-      "step": 2024
-    },
-    {
-      "epoch": 0.3261000845444664,
-      "grad_norm": 0.0014558681286871433,
-      "learning_rate": 0.0001999999480395409,
-      "loss": 46.0,
-      "step": 2025
-    },
-    {
-      "epoch": 0.32626112162325377,
-      "grad_norm": 0.00041977630462497473,
-      "learning_rate": 0.00019999994798795446,
-      "loss": 46.0,
-      "step": 2026
-    },
-    {
-      "epoch": 0.32642215870204117,
-      "grad_norm": 0.0005945920129306614,
-      "learning_rate": 0.00019999994793634245,
-      "loss": 46.0,
-      "step": 2027
-    },
-    {
-      "epoch": 0.32658319578082856,
-      "grad_norm": 0.00208011525683105,
-      "learning_rate": 0.0001999999478847048,
-      "loss": 46.0,
-      "step": 2028
-    },
-    {
-      "epoch": 0.3267442328596159,
-      "grad_norm": 0.0008152250084094703,
-      "learning_rate": 0.00019999994783304157,
-      "loss": 46.0,
-      "step": 2029
-    },
-    {
-      "epoch": 0.3269052699384033,
-      "grad_norm": 0.0003730282187461853,
-      "learning_rate": 0.00019999994778135274,
-      "loss": 46.0,
-      "step": 2030
-    },
-    {
-      "epoch": 0.3270663070171907,
-      "grad_norm": 0.00042543260497041047,
-      "learning_rate": 0.0001999999477296383,
-      "loss": 46.0,
-      "step": 2031
-    },
-    {
-      "epoch": 0.3272273440959781,
-      "grad_norm": 0.002453569555655122,
-      "learning_rate": 0.00019999994767789828,
-      "loss": 46.0,
-      "step": 2032
-    },
-    {
-      "epoch": 0.3273883811747655,
-      "grad_norm": 0.00020690720702987164,
-      "learning_rate": 0.00019999994762613267,
-      "loss": 46.0,
-      "step": 2033
-    },
-    {
-      "epoch": 0.3275494182535529,
-      "grad_norm": 0.0013863329077139497,
-      "learning_rate": 0.00019999994757434147,
-      "loss": 46.0,
-      "step": 2034
-    },
-    {
-      "epoch": 0.3277104553323403,
-      "grad_norm": 0.00252534169703722,
-      "learning_rate": 0.00019999994752252466,
-      "loss": 46.0,
-      "step": 2035
-    },
-    {
-      "epoch": 0.3278714924111277,
-      "grad_norm": 0.0031253830529749393,
-      "learning_rate": 0.00019999994747068228,
-      "loss": 46.0,
-      "step": 2036
-    },
-    {
-      "epoch": 0.32803252948991507,
-      "grad_norm": 0.0006806250894442201,
-      "learning_rate": 0.00019999994741881427,
-      "loss": 46.0,
-      "step": 2037
-    },
-    {
-      "epoch": 0.32819356656870247,
-      "grad_norm": 0.0010512153385207057,
-      "learning_rate": 0.0001999999473669207,
-      "loss": 46.0,
-      "step": 2038
-    },
-    {
-      "epoch": 0.32835460364748986,
-      "grad_norm": 0.001149252406321466,
-      "learning_rate": 0.0001999999473150015,
-      "loss": 46.0,
-      "step": 2039
-    },
-    {
-      "epoch": 0.3285156407262772,
-      "grad_norm": 0.0019053128780797124,
-      "learning_rate": 0.0001999999472630567,
-      "loss": 46.0,
-      "step": 2040
-    },
-    {
-      "epoch": 0.3286766778050646,
-      "grad_norm": 0.0009342010016553104,
-      "learning_rate": 0.00019999994721108633,
-      "loss": 46.0,
-      "step": 2041
-    },
-    {
-      "epoch": 0.328837714883852,
-      "grad_norm": 0.0003408027987461537,
-      "learning_rate": 0.00019999994715909038,
-      "loss": 46.0,
-      "step": 2042
-    },
-    {
-      "epoch": 0.3289987519626394,
-      "grad_norm": 0.0008491163607686758,
-      "learning_rate": 0.00019999994710706882,
-      "loss": 46.0,
-      "step": 2043
-    },
-    {
-      "epoch": 0.3291597890414268,
-      "grad_norm": 0.0006491068634204566,
-      "learning_rate": 0.00019999994705502167,
-      "loss": 46.0,
-      "step": 2044
-    },
-    {
-      "epoch": 0.3293208261202142,
-      "grad_norm": 0.0005705612129531801,
-      "learning_rate": 0.0001999999470029489,
-      "loss": 46.0,
-      "step": 2045
-    },
-    {
-      "epoch": 0.3294818631990016,
-      "grad_norm": 0.0004783225595019758,
-      "learning_rate": 0.00019999994695085055,
-      "loss": 46.0,
-      "step": 2046
-    },
-    {
-      "epoch": 0.329642900277789,
-      "grad_norm": 0.0003770444600377232,
-      "learning_rate": 0.00019999994689872663,
-      "loss": 46.0,
-      "step": 2047
-    },
-    {
-      "epoch": 0.32980393735657637,
-      "grad_norm": 0.0007112284074537456,
-      "learning_rate": 0.0001999999468465771,
-      "loss": 46.0,
-      "step": 2048
-    },
-    {
-      "epoch": 0.32996497443536377,
-      "grad_norm": 0.001417339313775301,
-      "learning_rate": 0.00019999994679440196,
-      "loss": 46.0,
-      "step": 2049
-    },
-    {
-      "epoch": 0.3301260115141511,
-      "grad_norm": 0.00069203827297315,
-      "learning_rate": 0.0001999999467422012,
-      "loss": 46.0,
-      "step": 2050
-    },
-    {
-      "epoch": 0.3302870485929385,
-      "grad_norm": 0.0004047035181429237,
-      "learning_rate": 0.00019999994668997491,
-      "loss": 46.0,
-      "step": 2051
-    },
-    {
-      "epoch": 0.3304480856717259,
-      "grad_norm": 0.00046402288717217743,
-      "learning_rate": 0.00019999994663772298,
-      "loss": 46.0,
-      "step": 2052
-    },
-    {
-      "epoch": 0.3306091227505133,
-      "grad_norm": 0.000519428460393101,
-      "learning_rate": 0.0001999999465854455,
-      "loss": 46.0,
-      "step": 2053
-    },
-    {
-      "epoch": 0.3307701598293007,
-      "grad_norm": 0.0007428214885294437,
-      "learning_rate": 0.00019999994653314239,
-      "loss": 46.0,
-      "step": 2054
-    },
-    {
-      "epoch": 0.3309311969080881,
-      "grad_norm": 0.0007421969785355031,
-      "learning_rate": 0.0001999999464808137,
-      "loss": 46.0,
-      "step": 2055
-    },
-    {
-      "epoch": 0.3310922339868755,
-      "grad_norm": 0.0006881988374516368,
-      "learning_rate": 0.0001999999464284594,
-      "loss": 46.0,
-      "step": 2056
-    },
-    {
-      "epoch": 0.3312532710656629,
-      "grad_norm": 0.001343220821581781,
-      "learning_rate": 0.0001999999463760795,
-      "loss": 46.0,
-      "step": 2057
-    },
-    {
-      "epoch": 0.3314143081444503,
-      "grad_norm": 0.0006997298332862556,
-      "learning_rate": 0.000199999946323674,
-      "loss": 46.0,
-      "step": 2058
-    },
-    {
-      "epoch": 0.33157534522323767,
-      "grad_norm": 0.0005575785762630403,
-      "learning_rate": 0.00019999994627124292,
-      "loss": 46.0,
-      "step": 2059
-    },
-    {
-      "epoch": 0.33173638230202507,
-      "grad_norm": 0.00034918313031084836,
-      "learning_rate": 0.00019999994621878626,
-      "loss": 46.0,
-      "step": 2060
-    },
-    {
-      "epoch": 0.3318974193808124,
-      "grad_norm": 0.00030340024386532605,
-      "learning_rate": 0.00019999994616630396,
-      "loss": 46.0,
-      "step": 2061
-    },
-    {
-      "epoch": 0.3320584564595998,
-      "grad_norm": 0.0005032563931308687,
-      "learning_rate": 0.0001999999461137961,
-      "loss": 46.0,
-      "step": 2062
-    },
-    {
-      "epoch": 0.3322194935383872,
-      "grad_norm": 0.0008769478881731629,
-      "learning_rate": 0.00019999994606126266,
-      "loss": 46.0,
-      "step": 2063
-    },
-    {
-      "epoch": 0.3323805306171746,
-      "grad_norm": 0.0005646526697091758,
-      "learning_rate": 0.0001999999460087036,
-      "loss": 46.0,
-      "step": 2064
-    },
-    {
-      "epoch": 0.332541567695962,
-      "grad_norm": 0.0005444358685053885,
-      "learning_rate": 0.00019999994595611896,
-      "loss": 46.0,
-      "step": 2065
-    },
-    {
-      "epoch": 0.3327026047747494,
-      "grad_norm": 0.0009491333621554077,
-      "learning_rate": 0.00019999994590350873,
-      "loss": 46.0,
-      "step": 2066
-    },
-    {
-      "epoch": 0.3328636418535368,
-      "grad_norm": 0.003124666167423129,
-      "learning_rate": 0.00019999994585087288,
-      "loss": 46.0,
-      "step": 2067
-    },
-    {
-      "epoch": 0.3330246789323242,
-      "grad_norm": 0.0009676054469309747,
-      "learning_rate": 0.00019999994579821144,
-      "loss": 46.0,
-      "step": 2068
-    },
-    {
-      "epoch": 0.3331857160111116,
-      "grad_norm": 0.0006022557499818504,
-      "learning_rate": 0.0001999999457455244,
-      "loss": 46.0,
-      "step": 2069
-    },
-    {
-      "epoch": 0.333346753089899,
-      "grad_norm": 0.0006285131676122546,
-      "learning_rate": 0.0001999999456928118,
-      "loss": 46.0,
-      "step": 2070
-    },
-    {
-      "epoch": 0.3335077901686863,
-      "grad_norm": 0.0010589825687929988,
-      "learning_rate": 0.0001999999456400736,
-      "loss": 46.0,
-      "step": 2071
-    },
-    {
-      "epoch": 0.3336688272474737,
-      "grad_norm": 0.0002169949293602258,
-      "learning_rate": 0.00019999994558730975,
-      "loss": 46.0,
-      "step": 2072
-    },
-    {
-      "epoch": 0.3338298643262611,
-      "grad_norm": 0.000426082726335153,
-      "learning_rate": 0.00019999994553452036,
-      "loss": 46.0,
-      "step": 2073
-    },
-    {
-      "epoch": 0.3339909014050485,
-      "grad_norm": 0.0018415822414681315,
-      "learning_rate": 0.00019999994548170532,
-      "loss": 46.0,
-      "step": 2074
-    },
-    {
-      "epoch": 0.3341519384838359,
-      "grad_norm": 0.0005673546111211181,
-      "learning_rate": 0.00019999994542886475,
-      "loss": 46.0,
-      "step": 2075
-    },
-    {
-      "epoch": 0.3343129755626233,
-      "grad_norm": 0.0007914051529951394,
-      "learning_rate": 0.00019999994537599856,
-      "loss": 46.0,
-      "step": 2076
-    },
-    {
-      "epoch": 0.3344740126414107,
-      "grad_norm": 0.0019066901877522469,
-      "learning_rate": 0.00019999994532310676,
-      "loss": 46.0,
-      "step": 2077
-    },
-    {
-      "epoch": 0.3346350497201981,
-      "grad_norm": 0.0008363793604075909,
-      "learning_rate": 0.00019999994527018937,
-      "loss": 46.0,
-      "step": 2078
-    },
-    {
-      "epoch": 0.3347960867989855,
-      "grad_norm": 0.0004497472255025059,
-      "learning_rate": 0.0001999999452172464,
-      "loss": 46.0,
-      "step": 2079
-    },
-    {
-      "epoch": 0.3349571238777729,
-      "grad_norm": 0.0014955061487853527,
-      "learning_rate": 0.00019999994516427784,
-      "loss": 46.0,
-      "step": 2080
-    },
-    {
-      "epoch": 0.3351181609565603,
-      "grad_norm": 0.001897261361591518,
-      "learning_rate": 0.0001999999451112837,
-      "loss": 46.0,
-      "step": 2081
-    },
-    {
-      "epoch": 0.3352791980353476,
-      "grad_norm": 0.0014495557406917214,
-      "learning_rate": 0.00019999994505826393,
-      "loss": 46.0,
-      "step": 2082
-    },
-    {
-      "epoch": 0.335440235114135,
-      "grad_norm": 0.0004804792406503111,
-      "learning_rate": 0.00019999994500521855,
-      "loss": 46.0,
-      "step": 2083
-    },
-    {
-      "epoch": 0.3356012721929224,
-      "grad_norm": 0.0007584398263134062,
-      "learning_rate": 0.0001999999449521476,
-      "loss": 46.0,
-      "step": 2084
-    },
-    {
-      "epoch": 0.3357623092717098,
-      "grad_norm": 0.00047130463644862175,
-      "learning_rate": 0.00019999994489905106,
-      "loss": 46.0,
-      "step": 2085
-    },
-    {
-      "epoch": 0.3359233463504972,
-      "grad_norm": 0.0008734721341170371,
-      "learning_rate": 0.00019999994484592892,
-      "loss": 46.0,
-      "step": 2086
-    },
-    {
-      "epoch": 0.3360843834292846,
-      "grad_norm": 0.0008928874740377069,
-      "learning_rate": 0.00019999994479278117,
-      "loss": 46.0,
-      "step": 2087
-    },
-    {
-      "epoch": 0.336245420508072,
-      "grad_norm": 0.0017567990580573678,
-      "learning_rate": 0.00019999994473960783,
-      "loss": 46.0,
-      "step": 2088
-    },
-    {
-      "epoch": 0.3364064575868594,
-      "grad_norm": 0.0010819401359185576,
-      "learning_rate": 0.00019999994468640893,
-      "loss": 46.0,
-      "step": 2089
-    },
-    {
-      "epoch": 0.3365674946656468,
-      "grad_norm": 0.0021045005414634943,
-      "learning_rate": 0.00019999994463318439,
-      "loss": 46.0,
-      "step": 2090
-    },
-    {
-      "epoch": 0.3367285317444342,
-      "grad_norm": 0.0004154415219090879,
-      "learning_rate": 0.00019999994457993428,
-      "loss": 46.0,
-      "step": 2091
-    },
-    {
-      "epoch": 0.3368895688232215,
-      "grad_norm": 0.0018240243662148714,
-      "learning_rate": 0.00019999994452665857,
-      "loss": 46.0,
-      "step": 2092
-    },
-    {
-      "epoch": 0.3370506059020089,
-      "grad_norm": 0.0012899583671241999,
-      "learning_rate": 0.00019999994447335726,
-      "loss": 46.0,
-      "step": 2093
-    },
-    {
-      "epoch": 0.3372116429807963,
-      "grad_norm": 0.0009415480890311301,
-      "learning_rate": 0.00019999994442003037,
-      "loss": 46.0,
-      "step": 2094
-    },
-    {
-      "epoch": 0.3373726800595837,
-      "grad_norm": 0.00041233381489291787,
-      "learning_rate": 0.0001999999443666779,
-      "loss": 46.0,
-      "step": 2095
-    },
-    {
-      "epoch": 0.3375337171383711,
-      "grad_norm": 0.0005574105307459831,
-      "learning_rate": 0.00019999994431329978,
-      "loss": 46.0,
-      "step": 2096
-    },
-    {
-      "epoch": 0.3376947542171585,
-      "grad_norm": 0.00043876664130948484,
-      "learning_rate": 0.0001999999442598961,
-      "loss": 46.0,
-      "step": 2097
-    },
-    {
-      "epoch": 0.3378557912959459,
-      "grad_norm": 0.00038646222674287856,
-      "learning_rate": 0.00019999994420646683,
-      "loss": 46.0,
-      "step": 2098
-    },
-    {
-      "epoch": 0.3380168283747333,
-      "grad_norm": 0.0008918823441490531,
-      "learning_rate": 0.00019999994415301195,
-      "loss": 46.0,
-      "step": 2099
-    },
-    {
-      "epoch": 0.3381778654535207,
-      "grad_norm": 0.00026927090948447585,
-      "learning_rate": 0.00019999994409953148,
-      "loss": 46.0,
-      "step": 2100
-    },
-    {
-      "epoch": 0.3383389025323081,
-      "grad_norm": 0.00041321897879242897,
-      "learning_rate": 0.0001999999440460254,
-      "loss": 46.0,
-      "step": 2101
-    },
-    {
-      "epoch": 0.3384999396110955,
-      "grad_norm": 0.0005375874461606145,
-      "learning_rate": 0.00019999994399249376,
-      "loss": 46.0,
-      "step": 2102
-    },
-    {
-      "epoch": 0.3386609766898828,
-      "grad_norm": 0.00026700831949710846,
-      "learning_rate": 0.0001999999439389365,
-      "loss": 46.0,
-      "step": 2103
-    },
-    {
-      "epoch": 0.3388220137686702,
-      "grad_norm": 0.000915630254894495,
-      "learning_rate": 0.00019999994388535366,
-      "loss": 46.0,
-      "step": 2104
-    },
-    {
-      "epoch": 0.3389830508474576,
-      "grad_norm": 0.0006769729661755264,
-      "learning_rate": 0.0001999999438317452,
-      "loss": 46.0,
-      "step": 2105
-    },
-    {
-      "epoch": 0.339144087926245,
-      "grad_norm": 0.0007546727429144084,
-      "learning_rate": 0.00019999994377811118,
-      "loss": 46.0,
-      "step": 2106
-    },
-    {
-      "epoch": 0.3393051250050324,
-      "grad_norm": 0.0009583216742612422,
-      "learning_rate": 0.00019999994372445155,
-      "loss": 46.0,
-      "step": 2107
-    },
-    {
-      "epoch": 0.3394661620838198,
-      "grad_norm": 0.00047133295447565615,
-      "learning_rate": 0.0001999999436707663,
-      "loss": 46.0,
-      "step": 2108
-    },
-    {
-      "epoch": 0.3396271991626072,
-      "grad_norm": 0.00082333420868963,
-      "learning_rate": 0.0001999999436170555,
-      "loss": 46.0,
-      "step": 2109
-    },
-    {
-      "epoch": 0.3397882362413946,
-      "grad_norm": 0.0006518145091831684,
-      "learning_rate": 0.00019999994356331908,
-      "loss": 46.0,
-      "step": 2110
-    },
-    {
-      "epoch": 0.339949273320182,
-      "grad_norm": 0.0005809074500575662,
-      "learning_rate": 0.00019999994350955707,
-      "loss": 46.0,
-      "step": 2111
-    },
-    {
-      "epoch": 0.3401103103989694,
-      "grad_norm": 0.0006611874559894204,
-      "learning_rate": 0.00019999994345576948,
-      "loss": 46.0,
-      "step": 2112
-    },
-    {
-      "epoch": 0.3402713474777568,
-      "grad_norm": 0.00037456219433806837,
-      "learning_rate": 0.00019999994340195624,
-      "loss": 46.0,
-      "step": 2113
-    },
-    {
-      "epoch": 0.3404323845565441,
-      "grad_norm": 0.0002828915312420577,
-      "learning_rate": 0.00019999994334811744,
-      "loss": 46.0,
-      "step": 2114
-    },
-    {
-      "epoch": 0.3405934216353315,
-      "grad_norm": 0.00040224610711447895,
-      "learning_rate": 0.00019999994329425306,
-      "loss": 46.0,
-      "step": 2115
-    },
-    {
-      "epoch": 0.3407544587141189,
-      "grad_norm": 0.0005563198355957866,
-      "learning_rate": 0.00019999994324036306,
-      "loss": 46.0,
-      "step": 2116
-    },
-    {
-      "epoch": 0.3409154957929063,
-      "grad_norm": 0.0008193428511731327,
-      "learning_rate": 0.00019999994318644748,
-      "loss": 46.0,
-      "step": 2117
-    },
-    {
-      "epoch": 0.3410765328716937,
-      "grad_norm": 0.0004005931841675192,
-      "learning_rate": 0.00019999994313250633,
-      "loss": 46.0,
-      "step": 2118
-    },
-    {
-      "epoch": 0.3412375699504811,
-      "grad_norm": 0.0003109571698587388,
-      "learning_rate": 0.00019999994307853954,
-      "loss": 46.0,
-      "step": 2119
-    },
-    {
-      "epoch": 0.3413986070292685,
-      "grad_norm": 0.0004226868913974613,
-      "learning_rate": 0.0001999999430245472,
-      "loss": 46.0,
-      "step": 2120
-    },
-    {
-      "epoch": 0.3415596441080559,
-      "grad_norm": 0.0006647769478149712,
-      "learning_rate": 0.0001999999429705292,
-      "loss": 46.0,
-      "step": 2121
-    },
-    {
-      "epoch": 0.3417206811868433,
-      "grad_norm": 0.001019866205751896,
-      "learning_rate": 0.00019999994291648566,
-      "loss": 46.0,
-      "step": 2122
-    },
-    {
-      "epoch": 0.3418817182656307,
-      "grad_norm": 0.0006915233097970486,
-      "learning_rate": 0.0001999999428624165,
-      "loss": 46.0,
-      "step": 2123
-    },
-    {
-      "epoch": 0.342042755344418,
-      "grad_norm": 0.0012263247044757009,
-      "learning_rate": 0.00019999994280832178,
-      "loss": 46.0,
-      "step": 2124
-    },
-    {
-      "epoch": 0.3422037924232054,
-      "grad_norm": 0.0009923606412485242,
-      "learning_rate": 0.0001999999427542014,
-      "loss": 46.0,
-      "step": 2125
-    },
-    {
-      "epoch": 0.3423648295019928,
-      "grad_norm": 0.0009882006561383605,
-      "learning_rate": 0.0001999999427000555,
-      "loss": 46.0,
-      "step": 2126
-    },
-    {
-      "epoch": 0.3425258665807802,
-      "grad_norm": 0.00047184983850456774,
-      "learning_rate": 0.00019999994264588398,
-      "loss": 46.0,
-      "step": 2127
-    },
-    {
-      "epoch": 0.3426869036595676,
-      "grad_norm": 0.0005706138908863068,
-      "learning_rate": 0.00019999994259168685,
-      "loss": 46.0,
-      "step": 2128
-    },
-    {
-      "epoch": 0.342847940738355,
-      "grad_norm": 0.0003796783566940576,
-      "learning_rate": 0.00019999994253746411,
-      "loss": 46.0,
-      "step": 2129
-    },
-    {
-      "epoch": 0.3430089778171424,
-      "grad_norm": 0.0002847843279596418,
-      "learning_rate": 0.00019999994248321581,
-      "loss": 46.0,
-      "step": 2130
-    },
-    {
-      "epoch": 0.3431700148959298,
-      "grad_norm": 0.0005586830084212124,
-      "learning_rate": 0.0001999999424289419,
-      "loss": 46.0,
-      "step": 2131
-    },
-    {
-      "epoch": 0.3433310519747172,
-      "grad_norm": 0.0005941084818914533,
-      "learning_rate": 0.0001999999423746424,
-      "loss": 46.0,
-      "step": 2132
-    },
-    {
-      "epoch": 0.3434920890535046,
-      "grad_norm": 0.0003910789673682302,
-      "learning_rate": 0.00019999994232031728,
-      "loss": 46.0,
-      "step": 2133
-    },
-    {
-      "epoch": 0.343653126132292,
-      "grad_norm": 0.0002927309542428702,
-      "learning_rate": 0.00019999994226596658,
-      "loss": 46.0,
-      "step": 2134
-    },
-    {
-      "epoch": 0.3438141632110793,
-      "grad_norm": 0.0016613490879535675,
-      "learning_rate": 0.00019999994221159032,
-      "loss": 46.0,
-      "step": 2135
-    },
-    {
-      "epoch": 0.3439752002898667,
-      "grad_norm": 0.00047319408622570336,
-      "learning_rate": 0.00019999994215718841,
-      "loss": 46.0,
-      "step": 2136
-    },
-    {
-      "epoch": 0.3441362373686541,
-      "grad_norm": 0.00018736837955657393,
-      "learning_rate": 0.00019999994210276095,
-      "loss": 46.0,
-      "step": 2137
-    },
-    {
-      "epoch": 0.3442972744474415,
-      "grad_norm": 0.0004510493890848011,
-      "learning_rate": 0.00019999994204830787,
-      "loss": 46.0,
-      "step": 2138
-    },
-    {
-      "epoch": 0.3444583115262289,
-      "grad_norm": 0.0006545883952639997,
-      "learning_rate": 0.0001999999419938292,
-      "loss": 46.0,
-      "step": 2139
-    },
-    {
-      "epoch": 0.3446193486050163,
-      "grad_norm": 0.0008365369867533445,
-      "learning_rate": 0.00019999994193932495,
-      "loss": 46.0,
-      "step": 2140
-    },
-    {
-      "epoch": 0.3447803856838037,
-      "grad_norm": 0.00029896796331740916,
-      "learning_rate": 0.0001999999418847951,
-      "loss": 46.0,
-      "step": 2141
-    },
-    {
-      "epoch": 0.3449414227625911,
-      "grad_norm": 0.001084184623323381,
-      "learning_rate": 0.00019999994183023963,
-      "loss": 46.0,
-      "step": 2142
-    },
-    {
-      "epoch": 0.3451024598413785,
-      "grad_norm": 0.0005514733493328094,
-      "learning_rate": 0.00019999994177565856,
-      "loss": 46.0,
-      "step": 2143
-    },
-    {
-      "epoch": 0.3452634969201659,
-      "grad_norm": 0.0004598818486556411,
-      "learning_rate": 0.00019999994172105194,
-      "loss": 46.0,
-      "step": 2144
-    },
-    {
-      "epoch": 0.34542453399895323,
-      "grad_norm": 0.0004823756462428719,
-      "learning_rate": 0.00019999994166641972,
-      "loss": 46.0,
-      "step": 2145
-    },
-    {
-      "epoch": 0.3455855710777406,
-      "grad_norm": 0.0004527256533037871,
-      "learning_rate": 0.00019999994161176186,
-      "loss": 46.0,
-      "step": 2146
-    },
-    {
-      "epoch": 0.345746608156528,
-      "grad_norm": 0.00024040110292844474,
-      "learning_rate": 0.00019999994155707845,
-      "loss": 46.0,
-      "step": 2147
-    },
-    {
-      "epoch": 0.3459076452353154,
-      "grad_norm": 0.0005315089947544038,
-      "learning_rate": 0.00019999994150236941,
-      "loss": 46.0,
-      "step": 2148
-    },
-    {
-      "epoch": 0.3460686823141028,
-      "grad_norm": 0.00047090795123949647,
-      "learning_rate": 0.0001999999414476348,
-      "loss": 46.0,
-      "step": 2149
-    },
-    {
-      "epoch": 0.3462297193928902,
-      "grad_norm": 0.0005423012189567089,
-      "learning_rate": 0.0001999999413928746,
-      "loss": 46.0,
-      "step": 2150
-    },
-    {
-      "epoch": 0.3463907564716776,
-      "grad_norm": 0.000755565008148551,
-      "learning_rate": 0.0001999999413380888,
-      "loss": 46.0,
-      "step": 2151
-    },
-    {
-      "epoch": 0.346551793550465,
-      "grad_norm": 0.0002905244182329625,
-      "learning_rate": 0.0001999999412832774,
-      "loss": 46.0,
-      "step": 2152
-    },
-    {
-      "epoch": 0.3467128306292524,
-      "grad_norm": 0.0005571144865825772,
-      "learning_rate": 0.0001999999412284404,
-      "loss": 46.0,
-      "step": 2153
-    },
-    {
-      "epoch": 0.3468738677080398,
-      "grad_norm": 0.00036992679815739393,
-      "learning_rate": 0.0001999999411735778,
-      "loss": 46.0,
-      "step": 2154
-    },
-    {
-      "epoch": 0.3470349047868272,
-      "grad_norm": 0.001313650282099843,
-      "learning_rate": 0.0001999999411186896,
-      "loss": 46.0,
-      "step": 2155
-    },
-    {
-      "epoch": 0.34719594186561453,
-      "grad_norm": 0.0012090746313333511,
-      "learning_rate": 0.00019999994106377584,
-      "loss": 46.0,
-      "step": 2156
-    },
-    {
-      "epoch": 0.34735697894440193,
-      "grad_norm": 0.0005325045203790069,
-      "learning_rate": 0.00019999994100883647,
-      "loss": 46.0,
-      "step": 2157
-    },
-    {
-      "epoch": 0.3475180160231893,
-      "grad_norm": 0.000600369821768254,
-      "learning_rate": 0.00019999994095387148,
-      "loss": 46.0,
-      "step": 2158
-    },
-    {
-      "epoch": 0.3476790531019767,
-      "grad_norm": 0.00028068615938536823,
-      "learning_rate": 0.0001999999408988809,
-      "loss": 46.0,
-      "step": 2159
-    },
-    {
-      "epoch": 0.3478400901807641,
-      "grad_norm": 0.0009161363705061376,
-      "learning_rate": 0.00019999994084386478,
-      "loss": 46.0,
-      "step": 2160
-    },
-    {
-      "epoch": 0.3480011272595515,
-      "grad_norm": 0.0017773328581824899,
-      "learning_rate": 0.000199999940788823,
-      "loss": 46.0,
-      "step": 2161
-    },
-    {
-      "epoch": 0.3481621643383389,
-      "grad_norm": 0.00041139687527902424,
-      "learning_rate": 0.00019999994073375565,
-      "loss": 46.0,
-      "step": 2162
-    },
-    {
-      "epoch": 0.3483232014171263,
-      "grad_norm": 0.0003068914811592549,
-      "learning_rate": 0.0001999999406786627,
-      "loss": 46.0,
-      "step": 2163
-    },
-    {
-      "epoch": 0.3484842384959137,
-      "grad_norm": 0.0003006128827109933,
-      "learning_rate": 0.00019999994062354416,
-      "loss": 46.0,
-      "step": 2164
-    },
-    {
-      "epoch": 0.3486452755747011,
-      "grad_norm": 0.0007143109105527401,
-      "learning_rate": 0.00019999994056840004,
-      "loss": 46.0,
-      "step": 2165
-    },
-    {
-      "epoch": 0.3488063126534885,
-      "grad_norm": 0.0005406810087151825,
-      "learning_rate": 0.00019999994051323033,
-      "loss": 46.0,
-      "step": 2166
-    },
-    {
-      "epoch": 0.34896734973227583,
-      "grad_norm": 0.0006577089661732316,
-      "learning_rate": 0.00019999994045803498,
-      "loss": 46.0,
-      "step": 2167
-    },
-    {
-      "epoch": 0.34912838681106323,
-      "grad_norm": 0.0008731005946174264,
-      "learning_rate": 0.00019999994040281407,
-      "loss": 46.0,
-      "step": 2168
-    },
-    {
-      "epoch": 0.3492894238898506,
-      "grad_norm": 0.0011856661876663566,
-      "learning_rate": 0.00019999994034756755,
-      "loss": 46.0,
-      "step": 2169
-    },
-    {
-      "epoch": 0.349450460968638,
-      "grad_norm": 0.0011892268666997552,
-      "learning_rate": 0.00019999994029229546,
-      "loss": 46.0,
-      "step": 2170
-    },
-    {
-      "epoch": 0.3496114980474254,
-      "grad_norm": 0.0003948225639760494,
-      "learning_rate": 0.00019999994023699774,
-      "loss": 46.0,
-      "step": 2171
-    },
-    {
-      "epoch": 0.3497725351262128,
-      "grad_norm": 0.00041553142364136875,
-      "learning_rate": 0.00019999994018167442,
-      "loss": 46.0,
-      "step": 2172
-    },
-    {
-      "epoch": 0.3499335722050002,
-      "grad_norm": 0.0009228391572833061,
-      "learning_rate": 0.00019999994012632555,
-      "loss": 46.0,
-      "step": 2173
-    },
-    {
-      "epoch": 0.3500946092837876,
-      "grad_norm": 0.0006806949386373162,
-      "learning_rate": 0.00019999994007095106,
-      "loss": 46.0,
-      "step": 2174
-    },
-    {
-      "epoch": 0.350255646362575,
-      "grad_norm": 0.0005127970362082124,
-      "learning_rate": 0.000199999940015551,
-      "loss": 46.0,
-      "step": 2175
-    },
-    {
-      "epoch": 0.3504166834413624,
-      "grad_norm": 0.0007062973454594612,
-      "learning_rate": 0.0001999999399601253,
-      "loss": 46.0,
-      "step": 2176
-    },
-    {
-      "epoch": 0.35057772052014974,
-      "grad_norm": 0.00033460246049799025,
-      "learning_rate": 0.00019999993990467402,
-      "loss": 46.0,
-      "step": 2177
-    },
-    {
-      "epoch": 0.35073875759893713,
-      "grad_norm": 0.0003914613334927708,
-      "learning_rate": 0.00019999993984919716,
-      "loss": 46.0,
-      "step": 2178
-    },
-    {
-      "epoch": 0.35089979467772453,
-      "grad_norm": 0.0007057320908643305,
-      "learning_rate": 0.00019999993979369468,
-      "loss": 46.0,
-      "step": 2179
-    },
-    {
-      "epoch": 0.3510608317565119,
-      "grad_norm": 0.0005471279146149755,
-      "learning_rate": 0.00019999993973816664,
-      "loss": 46.0,
-      "step": 2180
-    },
-    {
-      "epoch": 0.3512218688352993,
-      "grad_norm": 0.0006510689272545278,
-      "learning_rate": 0.000199999939682613,
-      "loss": 46.0,
-      "step": 2181
-    },
-    {
-      "epoch": 0.3513829059140867,
-      "grad_norm": 0.0006412953371182084,
-      "learning_rate": 0.00019999993962703373,
-      "loss": 46.0,
-      "step": 2182
-    },
-    {
-      "epoch": 0.3515439429928741,
-      "grad_norm": 0.0005228205700404942,
-      "learning_rate": 0.00019999993957142887,
-      "loss": 46.0,
-      "step": 2183
-    },
-    {
-      "epoch": 0.3517049800716615,
-      "grad_norm": 0.0008050735341385007,
-      "learning_rate": 0.00019999993951579846,
-      "loss": 46.0,
-      "step": 2184
-    },
-    {
-      "epoch": 0.3518660171504489,
-      "grad_norm": 0.0014420299557968974,
-      "learning_rate": 0.00019999993946014243,
-      "loss": 46.0,
-      "step": 2185
-    },
-    {
-      "epoch": 0.3520270542292363,
-      "grad_norm": 0.0005427033174782991,
-      "learning_rate": 0.0001999999394044608,
-      "loss": 46.0,
-      "step": 2186
-    },
-    {
-      "epoch": 0.3521880913080237,
-      "grad_norm": 0.0015302340034395456,
-      "learning_rate": 0.0001999999393487536,
-      "loss": 46.0,
-      "step": 2187
-    },
-    {
-      "epoch": 0.35234912838681104,
-      "grad_norm": 0.0017362490762025118,
-      "learning_rate": 0.00019999993929302077,
-      "loss": 46.0,
-      "step": 2188
-    },
-    {
-      "epoch": 0.35251016546559844,
-      "grad_norm": 0.0018614483997225761,
-      "learning_rate": 0.00019999993923726237,
-      "loss": 46.0,
-      "step": 2189
-    },
-    {
-      "epoch": 0.35267120254438583,
-      "grad_norm": 0.00042189229861833155,
-      "learning_rate": 0.00019999993918147835,
-      "loss": 46.0,
-      "step": 2190
-    },
-    {
-      "epoch": 0.3528322396231732,
-      "grad_norm": 0.0005222007748670876,
-      "learning_rate": 0.00019999993912566874,
-      "loss": 46.0,
-      "step": 2191
-    },
-    {
-      "epoch": 0.3529932767019606,
-      "grad_norm": 0.0005323282093741,
-      "learning_rate": 0.00019999993906983355,
-      "loss": 46.0,
-      "step": 2192
-    },
-    {
-      "epoch": 0.353154313780748,
-      "grad_norm": 0.00101379188708961,
-      "learning_rate": 0.00019999993901397275,
-      "loss": 46.0,
-      "step": 2193
-    },
-    {
-      "epoch": 0.3533153508595354,
-      "grad_norm": 0.0004553424078039825,
-      "learning_rate": 0.00019999993895808638,
-      "loss": 46.0,
-      "step": 2194
-    },
-    {
-      "epoch": 0.3534763879383228,
-      "grad_norm": 0.00032894639298319817,
-      "learning_rate": 0.0001999999389021744,
-      "loss": 46.0,
-      "step": 2195
-    },
-    {
-      "epoch": 0.3536374250171102,
-      "grad_norm": 0.0012497679563239217,
-      "learning_rate": 0.00019999993884623683,
-      "loss": 46.0,
-      "step": 2196
-    },
-    {
-      "epoch": 0.3537984620958976,
-      "grad_norm": 0.001451124087907374,
-      "learning_rate": 0.00019999993879027365,
-      "loss": 46.0,
-      "step": 2197
-    },
-    {
-      "epoch": 0.35395949917468494,
-      "grad_norm": 0.0010124671971425414,
-      "learning_rate": 0.00019999993873428488,
-      "loss": 46.0,
-      "step": 2198
-    },
-    {
-      "epoch": 0.35412053625347234,
-      "grad_norm": 0.000441828859038651,
-      "learning_rate": 0.00019999993867827055,
-      "loss": 46.0,
-      "step": 2199
-    },
-    {
-      "epoch": 0.35428157333225974,
-      "grad_norm": 0.00082436139928177,
-      "learning_rate": 0.00019999993862223055,
-      "loss": 46.0,
-      "step": 2200
-    },
-    {
-      "epoch": 0.35444261041104713,
-      "grad_norm": 0.0008977174293249846,
-      "learning_rate": 0.00019999993856616502,
-      "loss": 46.0,
-      "step": 2201
-    },
-    {
-      "epoch": 0.35460364748983453,
-      "grad_norm": 0.0006015023100189865,
-      "learning_rate": 0.00019999993851007388,
-      "loss": 46.0,
-      "step": 2202
-    },
-    {
-      "epoch": 0.3547646845686219,
-      "grad_norm": 0.0005927614984102547,
-      "learning_rate": 0.00019999993845395714,
-      "loss": 46.0,
-      "step": 2203
-    },
-    {
-      "epoch": 0.3549257216474093,
-      "grad_norm": 0.0011519651161506772,
-      "learning_rate": 0.00019999993839781482,
-      "loss": 46.0,
-      "step": 2204
-    },
-    {
-      "epoch": 0.3550867587261967,
-      "grad_norm": 0.0005731416749767959,
-      "learning_rate": 0.00019999993834164686,
-      "loss": 46.0,
-      "step": 2205
-    },
-    {
-      "epoch": 0.3552477958049841,
-      "grad_norm": 0.0007182052358984947,
-      "learning_rate": 0.00019999993828545337,
-      "loss": 46.0,
-      "step": 2206
-    },
-    {
-      "epoch": 0.3554088328837715,
-      "grad_norm": 0.0006719041266478598,
-      "learning_rate": 0.00019999993822923423,
-      "loss": 46.0,
-      "step": 2207
-    },
-    {
-      "epoch": 0.3555698699625589,
-      "grad_norm": 0.0005202871398068964,
-      "learning_rate": 0.0001999999381729895,
-      "loss": 46.0,
-      "step": 2208
-    },
-    {
-      "epoch": 0.35573090704134624,
-      "grad_norm": 0.0008943456341512501,
-      "learning_rate": 0.00019999993811671923,
-      "loss": 46.0,
-      "step": 2209
-    },
-    {
-      "epoch": 0.35589194412013364,
-      "grad_norm": 0.00039654961437918246,
-      "learning_rate": 0.00019999993806042333,
-      "loss": 46.0,
-      "step": 2210
-    },
-    {
-      "epoch": 0.35605298119892104,
-      "grad_norm": 0.0008803161908872426,
-      "learning_rate": 0.00019999993800410182,
-      "loss": 46.0,
-      "step": 2211
-    },
-    {
-      "epoch": 0.35621401827770843,
-      "grad_norm": 0.0012736518401652575,
-      "learning_rate": 0.00019999993794775472,
-      "loss": 46.0,
-      "step": 2212
-    },
-    {
-      "epoch": 0.35637505535649583,
-      "grad_norm": 0.0010360858868807554,
-      "learning_rate": 0.00019999993789138204,
-      "loss": 46.0,
-      "step": 2213
-    },
-    {
-      "epoch": 0.3565360924352832,
-      "grad_norm": 0.0007000893820077181,
-      "learning_rate": 0.00019999993783498377,
-      "loss": 46.0,
-      "step": 2214
-    },
-    {
-      "epoch": 0.3566971295140706,
-      "grad_norm": 0.00021741408272646368,
-      "learning_rate": 0.00019999993777855988,
-      "loss": 46.0,
-      "step": 2215
-    },
-    {
-      "epoch": 0.356858166592858,
-      "grad_norm": 0.00039426906732842326,
-      "learning_rate": 0.0001999999377221104,
-      "loss": 46.0,
-      "step": 2216
-    },
-    {
-      "epoch": 0.3570192036716454,
-      "grad_norm": 0.001243783626705408,
-      "learning_rate": 0.00019999993766563534,
-      "loss": 46.0,
-      "step": 2217
-    },
-    {
-      "epoch": 0.3571802407504328,
-      "grad_norm": 0.0012748971348628402,
-      "learning_rate": 0.0001999999376091347,
-      "loss": 46.0,
-      "step": 2218
-    },
-    {
-      "epoch": 0.3573412778292202,
-      "grad_norm": 0.0005534195224754512,
-      "learning_rate": 0.0001999999375526084,
-      "loss": 46.0,
-      "step": 2219
-    },
-    {
-      "epoch": 0.35750231490800755,
-      "grad_norm": 0.000428906234446913,
-      "learning_rate": 0.00019999993749605656,
-      "loss": 46.0,
-      "step": 2220
-    },
-    {
-      "epoch": 0.35766335198679494,
-      "grad_norm": 0.0006642548833042383,
-      "learning_rate": 0.00019999993743947912,
-      "loss": 46.0,
-      "step": 2221
-    },
-    {
-      "epoch": 0.35782438906558234,
-      "grad_norm": 0.0005308387917466462,
-      "learning_rate": 0.00019999993738287607,
-      "loss": 46.0,
-      "step": 2222
-    },
-    {
-      "epoch": 0.35798542614436973,
-      "grad_norm": 0.0007748939679004252,
-      "learning_rate": 0.00019999993732624746,
-      "loss": 46.0,
-      "step": 2223
-    },
-    {
-      "epoch": 0.35814646322315713,
-      "grad_norm": 0.000501910166349262,
-      "learning_rate": 0.00019999993726959323,
-      "loss": 46.0,
-      "step": 2224
-    },
-    {
-      "epoch": 0.3583075003019445,
-      "grad_norm": 0.0004448570543900132,
-      "learning_rate": 0.0001999999372129134,
-      "loss": 46.0,
-      "step": 2225
-    },
-    {
-      "epoch": 0.3584685373807319,
-      "grad_norm": 0.0002656817378010601,
-      "learning_rate": 0.00019999993715620796,
-      "loss": 46.0,
-      "step": 2226
-    },
-    {
-      "epoch": 0.3586295744595193,
-      "grad_norm": 0.0006572982529178262,
-      "learning_rate": 0.00019999993709947695,
-      "loss": 46.0,
-      "step": 2227
-    },
-    {
-      "epoch": 0.3587906115383067,
-      "grad_norm": 0.0004981006495654583,
-      "learning_rate": 0.00019999993704272035,
-      "loss": 46.0,
-      "step": 2228
-    },
-    {
-      "epoch": 0.3589516486170941,
-      "grad_norm": 0.0004337487043812871,
-      "learning_rate": 0.00019999993698593813,
-      "loss": 46.0,
-      "step": 2229
-    },
-    {
-      "epoch": 0.35911268569588145,
-      "grad_norm": 0.0003623554075602442,
-      "learning_rate": 0.00019999993692913033,
-      "loss": 46.0,
-      "step": 2230
-    },
-    {
-      "epoch": 0.35927372277466885,
-      "grad_norm": 0.00041682831943035126,
-      "learning_rate": 0.00019999993687229694,
-      "loss": 46.0,
-      "step": 2231
-    },
-    {
-      "epoch": 0.35943475985345624,
-      "grad_norm": 0.0004244146402925253,
-      "learning_rate": 0.00019999993681543796,
-      "loss": 46.0,
-      "step": 2232
-    },
-    {
-      "epoch": 0.35959579693224364,
-      "grad_norm": 0.000629731104709208,
-      "learning_rate": 0.00019999993675855337,
-      "loss": 46.0,
-      "step": 2233
-    },
-    {
-      "epoch": 0.35975683401103103,
-      "grad_norm": 0.0007296414696611464,
-      "learning_rate": 0.0001999999367016432,
-      "loss": 46.0,
-      "step": 2234
-    },
-    {
-      "epoch": 0.35991787108981843,
-      "grad_norm": 0.0006106806686148047,
-      "learning_rate": 0.00019999993664470743,
-      "loss": 46.0,
-      "step": 2235
-    },
-    {
-      "epoch": 0.3600789081686058,
-      "grad_norm": 0.0006707884604111314,
-      "learning_rate": 0.00019999993658774605,
-      "loss": 46.0,
-      "step": 2236
-    },
-    {
-      "epoch": 0.3602399452473932,
-      "grad_norm": 0.00045559322461485863,
-      "learning_rate": 0.00019999993653075908,
-      "loss": 46.0,
-      "step": 2237
-    },
-    {
-      "epoch": 0.3604009823261806,
-      "grad_norm": 0.0005608388455584645,
-      "learning_rate": 0.00019999993647374652,
-      "loss": 46.0,
-      "step": 2238
-    },
-    {
-      "epoch": 0.360562019404968,
-      "grad_norm": 0.00019335455726832151,
-      "learning_rate": 0.00019999993641670836,
-      "loss": 46.0,
-      "step": 2239
-    },
-    {
-      "epoch": 0.3607230564837554,
-      "grad_norm": 0.00037204186082817614,
-      "learning_rate": 0.00019999993635964463,
-      "loss": 46.0,
-      "step": 2240
-    },
-    {
-      "epoch": 0.36088409356254275,
-      "grad_norm": 0.00035147430025972426,
-      "learning_rate": 0.00019999993630255526,
-      "loss": 46.0,
-      "step": 2241
-    },
-    {
-      "epoch": 0.36104513064133015,
-      "grad_norm": 0.0006060279556550086,
-      "learning_rate": 0.00019999993624544033,
-      "loss": 46.0,
-      "step": 2242
-    },
-    {
-      "epoch": 0.36120616772011754,
-      "grad_norm": 0.0017185431206598878,
-      "learning_rate": 0.0001999999361882998,
-      "loss": 46.0,
-      "step": 2243
-    },
-    {
-      "epoch": 0.36136720479890494,
-      "grad_norm": 0.003380146576091647,
-      "learning_rate": 0.00019999993613113368,
-      "loss": 46.0,
-      "step": 2244
-    },
-    {
-      "epoch": 0.36152824187769234,
-      "grad_norm": 0.0014559919945895672,
-      "learning_rate": 0.00019999993607394196,
-      "loss": 46.0,
-      "step": 2245
-    },
-    {
-      "epoch": 0.36168927895647973,
-      "grad_norm": 0.0008151179645210505,
-      "learning_rate": 0.00019999993601672465,
-      "loss": 46.0,
-      "step": 2246
-    },
-    {
-      "epoch": 0.3618503160352671,
-      "grad_norm": 0.00041224207961931825,
-      "learning_rate": 0.00019999993595948173,
-      "loss": 46.0,
-      "step": 2247
-    },
-    {
-      "epoch": 0.3620113531140545,
-      "grad_norm": 0.0012288088910281658,
-      "learning_rate": 0.00019999993590221323,
-      "loss": 46.0,
-      "step": 2248
-    },
-    {
-      "epoch": 0.3621723901928419,
-      "grad_norm": 0.002109048655256629,
-      "learning_rate": 0.00019999993584491913,
-      "loss": 46.0,
-      "step": 2249
-    },
-    {
-      "epoch": 0.3623334272716293,
-      "grad_norm": 0.0003549403336364776,
-      "learning_rate": 0.00019999993578759942,
-      "loss": 46.0,
-      "step": 2250
-    },
-    {
-      "epoch": 0.36249446435041666,
-      "grad_norm": 0.0005241574835963547,
-      "learning_rate": 0.00019999993573025413,
-      "loss": 46.0,
-      "step": 2251
-    },
-    {
-      "epoch": 0.36265550142920405,
-      "grad_norm": 0.0009149570250883698,
-      "learning_rate": 0.00019999993567288324,
-      "loss": 46.0,
-      "step": 2252
-    },
-    {
-      "epoch": 0.36281653850799145,
-      "grad_norm": 0.0008694474818184972,
-      "learning_rate": 0.00019999993561548677,
-      "loss": 46.0,
-      "step": 2253
-    },
-    {
-      "epoch": 0.36297757558677884,
-      "grad_norm": 0.0006183795048855245,
-      "learning_rate": 0.0001999999355580647,
-      "loss": 46.0,
-      "step": 2254
-    },
-    {
-      "epoch": 0.36313861266556624,
-      "grad_norm": 0.0026527612935751677,
-      "learning_rate": 0.00019999993550061702,
-      "loss": 46.0,
-      "step": 2255
-    },
-    {
-      "epoch": 0.36329964974435364,
-      "grad_norm": 0.003594527952373028,
-      "learning_rate": 0.00019999993544314376,
-      "loss": 46.0,
-      "step": 2256
-    },
-    {
-      "epoch": 0.36346068682314103,
-      "grad_norm": 0.0005805824184790254,
-      "learning_rate": 0.0001999999353856449,
-      "loss": 46.0,
-      "step": 2257
-    },
-    {
-      "epoch": 0.36362172390192843,
-      "grad_norm": 0.0006647557020187378,
-      "learning_rate": 0.00019999993532812042,
-      "loss": 46.0,
-      "step": 2258
-    },
-    {
-      "epoch": 0.3637827609807158,
-      "grad_norm": 0.00038508547004312277,
-      "learning_rate": 0.00019999993527057038,
-      "loss": 46.0,
-      "step": 2259
-    },
-    {
-      "epoch": 0.3639437980595032,
-      "grad_norm": 0.0014680176973342896,
-      "learning_rate": 0.00019999993521299474,
-      "loss": 46.0,
-      "step": 2260
-    },
-    {
-      "epoch": 0.3641048351382906,
-      "grad_norm": 0.0010807851795107126,
-      "learning_rate": 0.0001999999351553935,
-      "loss": 46.0,
-      "step": 2261
-    },
-    {
-      "epoch": 0.36426587221707796,
-      "grad_norm": 0.00028498127358034253,
-      "learning_rate": 0.00019999993509776666,
-      "loss": 46.0,
-      "step": 2262
-    },
-    {
-      "epoch": 0.36442690929586535,
-      "grad_norm": 0.0010145478881895542,
-      "learning_rate": 0.00019999993504011423,
-      "loss": 46.0,
-      "step": 2263
-    },
-    {
-      "epoch": 0.36458794637465275,
-      "grad_norm": 0.000994094298221171,
-      "learning_rate": 0.00019999993498243622,
-      "loss": 46.0,
-      "step": 2264
-    },
-    {
-      "epoch": 0.36474898345344015,
-      "grad_norm": 0.0034477945882827044,
-      "learning_rate": 0.0001999999349247326,
-      "loss": 46.0,
-      "step": 2265
-    },
-    {
-      "epoch": 0.36491002053222754,
-      "grad_norm": 0.0006467928760685027,
-      "learning_rate": 0.00019999993486700336,
-      "loss": 46.0,
-      "step": 2266
-    },
-    {
-      "epoch": 0.36507105761101494,
-      "grad_norm": 0.00040097400778904557,
-      "learning_rate": 0.00019999993480924856,
-      "loss": 46.0,
-      "step": 2267
-    },
-    {
-      "epoch": 0.36523209468980233,
-      "grad_norm": 0.0007467146497219801,
-      "learning_rate": 0.00019999993475146814,
-      "loss": 46.0,
-      "step": 2268
-    },
-    {
-      "epoch": 0.36539313176858973,
-      "grad_norm": 0.0005757774924859405,
-      "learning_rate": 0.00019999993469366217,
-      "loss": 46.0,
-      "step": 2269
-    },
-    {
-      "epoch": 0.3655541688473771,
-      "grad_norm": 0.0007973664905875921,
-      "learning_rate": 0.00019999993463583056,
-      "loss": 46.0,
-      "step": 2270
-    },
-    {
-      "epoch": 0.3657152059261645,
-      "grad_norm": 0.0010491148568689823,
-      "learning_rate": 0.00019999993457797338,
-      "loss": 46.0,
-      "step": 2271
-    },
-    {
-      "epoch": 0.3658762430049519,
-      "grad_norm": 0.00041741313179954886,
-      "learning_rate": 0.0001999999345200906,
-      "loss": 46.0,
-      "step": 2272
-    },
-    {
-      "epoch": 0.36603728008373926,
-      "grad_norm": 0.0014166667824611068,
-      "learning_rate": 0.00019999993446218222,
-      "loss": 46.0,
-      "step": 2273
-    },
-    {
-      "epoch": 0.36619831716252665,
-      "grad_norm": 0.001200210303068161,
-      "learning_rate": 0.00019999993440424825,
-      "loss": 46.0,
-      "step": 2274
-    },
-    {
-      "epoch": 0.36635935424131405,
-      "grad_norm": 0.0004238256369717419,
-      "learning_rate": 0.00019999993434628867,
-      "loss": 46.0,
-      "step": 2275
-    },
-    {
-      "epoch": 0.36652039132010145,
-      "grad_norm": 0.0013004005886614323,
-      "learning_rate": 0.0001999999342883035,
-      "loss": 46.0,
-      "step": 2276
-    },
-    {
-      "epoch": 0.36668142839888884,
-      "grad_norm": 0.0008044597343541682,
-      "learning_rate": 0.00019999993423029276,
-      "loss": 46.0,
-      "step": 2277
-    },
-    {
-      "epoch": 0.36684246547767624,
-      "grad_norm": 0.0007422963972203434,
-      "learning_rate": 0.0001999999341722564,
-      "loss": 46.0,
-      "step": 2278
-    },
-    {
-      "epoch": 0.36700350255646363,
-      "grad_norm": 0.0003879069408867508,
-      "learning_rate": 0.00019999993411419444,
-      "loss": 46.0,
-      "step": 2279
-    },
-    {
-      "epoch": 0.36716453963525103,
-      "grad_norm": 0.0020238852594047785,
-      "learning_rate": 0.0001999999340561069,
-      "loss": 46.0,
-      "step": 2280
-    },
-    {
-      "epoch": 0.3673255767140384,
-      "grad_norm": 0.0018274880712851882,
-      "learning_rate": 0.00019999993399799377,
-      "loss": 46.0,
-      "step": 2281
-    },
-    {
-      "epoch": 0.3674866137928258,
-      "grad_norm": 0.0013155628694221377,
-      "learning_rate": 0.00019999993393985503,
-      "loss": 46.0,
-      "step": 2282
-    },
-    {
-      "epoch": 0.36764765087161316,
-      "grad_norm": 0.0008399191428907216,
-      "learning_rate": 0.00019999993388169072,
-      "loss": 46.0,
-      "step": 2283
-    },
-    {
-      "epoch": 0.36780868795040056,
-      "grad_norm": 0.0004666945606004447,
-      "learning_rate": 0.0001999999338235008,
-      "loss": 46.0,
-      "step": 2284
-    },
-    {
-      "epoch": 0.36796972502918796,
-      "grad_norm": 0.0008019223459996283,
-      "learning_rate": 0.00019999993376528528,
-      "loss": 46.0,
-      "step": 2285
-    },
-    {
-      "epoch": 0.36813076210797535,
-      "grad_norm": 0.000614597403910011,
-      "learning_rate": 0.00019999993370704418,
-      "loss": 46.0,
-      "step": 2286
-    },
-    {
-      "epoch": 0.36829179918676275,
-      "grad_norm": 0.0001731208903947845,
-      "learning_rate": 0.00019999993364877745,
-      "loss": 46.0,
-      "step": 2287
-    },
-    {
-      "epoch": 0.36845283626555014,
-      "grad_norm": 0.0010255471570417285,
-      "learning_rate": 0.00019999993359048513,
-      "loss": 46.0,
-      "step": 2288
-    },
-    {
-      "epoch": 0.36861387334433754,
-      "grad_norm": 0.00031396085978485644,
-      "learning_rate": 0.00019999993353216725,
-      "loss": 46.0,
-      "step": 2289
-    },
-    {
-      "epoch": 0.36877491042312494,
-      "grad_norm": 0.0005155057879164815,
-      "learning_rate": 0.00019999993347382376,
-      "loss": 46.0,
-      "step": 2290
-    },
-    {
-      "epoch": 0.36893594750191233,
-      "grad_norm": 0.00048207244253717363,
-      "learning_rate": 0.00019999993341545468,
-      "loss": 46.0,
-      "step": 2291
-    },
-    {
-      "epoch": 0.3690969845806997,
-      "grad_norm": 0.0020481268875300884,
-      "learning_rate": 0.00019999993335706,
-      "loss": 46.0,
-      "step": 2292
-    },
-    {
-      "epoch": 0.3692580216594871,
-      "grad_norm": 0.0010415689321234822,
-      "learning_rate": 0.0001999999332986397,
-      "loss": 46.0,
-      "step": 2293
-    },
-    {
-      "epoch": 0.36941905873827446,
-      "grad_norm": 0.00044290709774941206,
-      "learning_rate": 0.00019999993324019383,
-      "loss": 46.0,
-      "step": 2294
-    },
-    {
-      "epoch": 0.36958009581706186,
-      "grad_norm": 0.000494443520437926,
-      "learning_rate": 0.0001999999331817224,
-      "loss": 46.0,
-      "step": 2295
-    },
-    {
-      "epoch": 0.36974113289584926,
-      "grad_norm": 0.0002783818927127868,
-      "learning_rate": 0.0001999999331232253,
-      "loss": 46.0,
-      "step": 2296
-    },
-    {
-      "epoch": 0.36990216997463665,
-      "grad_norm": 0.00034031918039545417,
-      "learning_rate": 0.00019999993306470267,
-      "loss": 46.0,
-      "step": 2297
-    },
-    {
-      "epoch": 0.37006320705342405,
-      "grad_norm": 0.0005188029608689249,
-      "learning_rate": 0.00019999993300615443,
-      "loss": 46.0,
-      "step": 2298
-    },
-    {
-      "epoch": 0.37022424413221144,
-      "grad_norm": 0.0005571770598180592,
-      "learning_rate": 0.00019999993294758057,
-      "loss": 46.0,
-      "step": 2299
-    },
-    {
-      "epoch": 0.37038528121099884,
-      "grad_norm": 0.0008157939882948995,
-      "learning_rate": 0.00019999993288898115,
-      "loss": 46.0,
-      "step": 2300
-    },
-    {
-      "epoch": 0.37054631828978624,
-      "grad_norm": 0.0016090160934254527,
-      "learning_rate": 0.0001999999328303561,
-      "loss": 46.0,
-      "step": 2301
-    },
-    {
-      "epoch": 0.37070735536857363,
-      "grad_norm": 0.0010384847410023212,
-      "learning_rate": 0.0001999999327717055,
-      "loss": 46.0,
-      "step": 2302
-    },
-    {
-      "epoch": 0.37086839244736103,
-      "grad_norm": 0.000959428318310529,
-      "learning_rate": 0.00019999993271302926,
-      "loss": 46.0,
-      "step": 2303
-    },
-    {
-      "epoch": 0.37102942952614837,
-      "grad_norm": 0.0019294407684355974,
-      "learning_rate": 0.0001999999326543274,
-      "loss": 46.0,
-      "step": 2304
-    },
-    {
-      "epoch": 0.37119046660493576,
-      "grad_norm": 0.000621182844042778,
-      "learning_rate": 0.00019999993259560002,
-      "loss": 46.0,
-      "step": 2305
-    },
-    {
-      "epoch": 0.37135150368372316,
-      "grad_norm": 0.0004290766955818981,
-      "learning_rate": 0.00019999993253684703,
-      "loss": 46.0,
-      "step": 2306
-    },
-    {
-      "epoch": 0.37151254076251056,
-      "grad_norm": 0.001392365782521665,
-      "learning_rate": 0.00019999993247806841,
-      "loss": 46.0,
-      "step": 2307
-    },
-    {
-      "epoch": 0.37167357784129795,
-      "grad_norm": 0.00041785926441662014,
-      "learning_rate": 0.0001999999324192642,
-      "loss": 46.0,
-      "step": 2308
-    },
-    {
-      "epoch": 0.37183461492008535,
-      "grad_norm": 0.0008172749658115208,
-      "learning_rate": 0.0001999999323604344,
-      "loss": 46.0,
-      "step": 2309
-    },
-    {
-      "epoch": 0.37199565199887275,
-      "grad_norm": 0.0004754325782414526,
-      "learning_rate": 0.00019999993230157903,
-      "loss": 46.0,
-      "step": 2310
-    },
-    {
-      "epoch": 0.37215668907766014,
-      "grad_norm": 0.0016481417696923018,
-      "learning_rate": 0.00019999993224269804,
-      "loss": 46.0,
-      "step": 2311
-    },
-    {
-      "epoch": 0.37231772615644754,
-      "grad_norm": 0.0010458268225193024,
-      "learning_rate": 0.00019999993218379144,
-      "loss": 46.0,
-      "step": 2312
-    },
-    {
-      "epoch": 0.37247876323523493,
-      "grad_norm": 0.0037686415016651154,
-      "learning_rate": 0.0001999999321248593,
-      "loss": 46.0,
-      "step": 2313
-    },
-    {
-      "epoch": 0.37263980031402233,
-      "grad_norm": 0.0004300907894503325,
-      "learning_rate": 0.00019999993206590153,
-      "loss": 46.0,
-      "step": 2314
-    },
-    {
-      "epoch": 0.37280083739280967,
-      "grad_norm": 0.0015523259062319994,
-      "learning_rate": 0.00019999993200691817,
-      "loss": 46.0,
-      "step": 2315
-    },
-    {
-      "epoch": 0.37296187447159707,
-      "grad_norm": 0.0010225278092548251,
-      "learning_rate": 0.0001999999319479092,
-      "loss": 46.0,
-      "step": 2316
-    },
-    {
-      "epoch": 0.37312291155038446,
-      "grad_norm": 0.00028549617854878306,
-      "learning_rate": 0.00019999993188887465,
-      "loss": 46.0,
-      "step": 2317
-    },
-    {
-      "epoch": 0.37328394862917186,
-      "grad_norm": 0.00028623847174458206,
-      "learning_rate": 0.0001999999318298145,
-      "loss": 46.0,
-      "step": 2318
-    },
-    {
-      "epoch": 0.37344498570795925,
-      "grad_norm": 0.0005056145018897951,
-      "learning_rate": 0.00019999993177072876,
-      "loss": 46.0,
-      "step": 2319
-    },
-    {
-      "epoch": 0.37360602278674665,
-      "grad_norm": 0.0005586351617239416,
-      "learning_rate": 0.0001999999317116174,
-      "loss": 46.0,
-      "step": 2320
-    },
-    {
-      "epoch": 0.37376705986553405,
-      "grad_norm": 0.0007560873636975884,
-      "learning_rate": 0.0001999999316524805,
-      "loss": 46.0,
-      "step": 2321
-    },
-    {
-      "epoch": 0.37392809694432144,
-      "grad_norm": 0.0012680510990321636,
-      "learning_rate": 0.00019999993159331797,
-      "loss": 46.0,
-      "step": 2322
-    },
-    {
-      "epoch": 0.37408913402310884,
-      "grad_norm": 0.0006473853136412799,
-      "learning_rate": 0.00019999993153412983,
-      "loss": 46.0,
-      "step": 2323
-    },
-    {
-      "epoch": 0.37425017110189623,
-      "grad_norm": 0.0005554733797907829,
-      "learning_rate": 0.00019999993147491613,
-      "loss": 46.0,
-      "step": 2324
-    },
-    {
-      "epoch": 0.37441120818068363,
-      "grad_norm": 0.000544844544492662,
-      "learning_rate": 0.0001999999314156768,
-      "loss": 46.0,
-      "step": 2325
-    },
-    {
-      "epoch": 0.37457224525947097,
-      "grad_norm": 0.0005107149481773376,
-      "learning_rate": 0.0001999999313564119,
-      "loss": 46.0,
-      "step": 2326
-    },
-    {
-      "epoch": 0.37473328233825837,
-      "grad_norm": 0.0010599495144560933,
-      "learning_rate": 0.00019999993129712142,
-      "loss": 46.0,
-      "step": 2327
-    },
-    {
-      "epoch": 0.37489431941704576,
-      "grad_norm": 0.0009644674137234688,
-      "learning_rate": 0.00019999993123780531,
-      "loss": 46.0,
-      "step": 2328
-    },
-    {
-      "epoch": 0.37505535649583316,
-      "grad_norm": 0.0015917709097266197,
-      "learning_rate": 0.00019999993117846362,
-      "loss": 46.0,
-      "step": 2329
-    },
-    {
-      "epoch": 0.37521639357462055,
-      "grad_norm": 0.0010225436417385936,
-      "learning_rate": 0.00019999993111909634,
-      "loss": 46.0,
-      "step": 2330
-    },
-    {
-      "epoch": 0.37537743065340795,
-      "grad_norm": 0.001246629050001502,
-      "learning_rate": 0.00019999993105970348,
-      "loss": 46.0,
-      "step": 2331
-    },
-    {
-      "epoch": 0.37553846773219535,
-      "grad_norm": 0.0007145669078454375,
-      "learning_rate": 0.000199999931000285,
-      "loss": 46.0,
-      "step": 2332
-    },
-    {
-      "epoch": 0.37569950481098274,
-      "grad_norm": 0.0005153944948688149,
-      "learning_rate": 0.0001999999309408409,
-      "loss": 46.0,
-      "step": 2333
-    },
-    {
-      "epoch": 0.37586054188977014,
-      "grad_norm": 0.0007187469163909554,
-      "learning_rate": 0.00019999993088137125,
-      "loss": 46.0,
-      "step": 2334
-    },
-    {
-      "epoch": 0.37602157896855753,
-      "grad_norm": 0.0010105695109814405,
-      "learning_rate": 0.00019999993082187598,
-      "loss": 46.0,
-      "step": 2335
-    },
-    {
-      "epoch": 0.3761826160473449,
-      "grad_norm": 0.0009219853673130274,
-      "learning_rate": 0.00019999993076235515,
-      "loss": 46.0,
-      "step": 2336
-    },
-    {
-      "epoch": 0.37634365312613227,
-      "grad_norm": 0.0004764607292599976,
-      "learning_rate": 0.00019999993070280868,
-      "loss": 46.0,
-      "step": 2337
-    },
-    {
-      "epoch": 0.37650469020491967,
-      "grad_norm": 0.0025382298044860363,
-      "learning_rate": 0.00019999993064323665,
-      "loss": 46.0,
-      "step": 2338
-    },
-    {
-      "epoch": 0.37666572728370706,
-      "grad_norm": 0.0008136255200952291,
-      "learning_rate": 0.000199999930583639,
-      "loss": 46.0,
-      "step": 2339
-    },
-    {
-      "epoch": 0.37682676436249446,
-      "grad_norm": 0.0017399373464286327,
-      "learning_rate": 0.00019999993052401577,
-      "loss": 46.0,
-      "step": 2340
-    },
-    {
-      "epoch": 0.37698780144128186,
-      "grad_norm": 0.00034965010127052665,
-      "learning_rate": 0.00019999993046436693,
-      "loss": 46.0,
-      "step": 2341
-    },
-    {
-      "epoch": 0.37714883852006925,
-      "grad_norm": 0.002027729758992791,
-      "learning_rate": 0.00019999993040469255,
-      "loss": 46.0,
-      "step": 2342
-    },
-    {
-      "epoch": 0.37730987559885665,
-      "grad_norm": 0.0010564856929704547,
-      "learning_rate": 0.00019999993034499253,
-      "loss": 46.0,
-      "step": 2343
-    },
-    {
-      "epoch": 0.37747091267764404,
-      "grad_norm": 0.0006513954722322524,
-      "learning_rate": 0.0001999999302852669,
-      "loss": 46.0,
-      "step": 2344
-    },
-    {
-      "epoch": 0.37763194975643144,
-      "grad_norm": 0.0009405898745171726,
-      "learning_rate": 0.0001999999302255157,
-      "loss": 46.0,
-      "step": 2345
-    },
-    {
-      "epoch": 0.37779298683521884,
-      "grad_norm": 0.0003593849833123386,
-      "learning_rate": 0.0001999999301657389,
-      "loss": 46.0,
-      "step": 2346
-    },
-    {
-      "epoch": 0.3779540239140062,
-      "grad_norm": 0.0039969636127352715,
-      "learning_rate": 0.0001999999301059365,
-      "loss": 46.0,
-      "step": 2347
-    },
-    {
-      "epoch": 0.3781150609927936,
-      "grad_norm": 0.0006771524786017835,
-      "learning_rate": 0.00019999993004610848,
-      "loss": 46.0,
-      "step": 2348
-    },
-    {
-      "epoch": 0.37827609807158097,
-      "grad_norm": 0.0008385548717342317,
-      "learning_rate": 0.0001999999299862549,
-      "loss": 46.0,
-      "step": 2349
-    },
-    {
-      "epoch": 0.37843713515036836,
-      "grad_norm": 0.0017256089486181736,
-      "learning_rate": 0.00019999992992637572,
-      "loss": 46.0,
-      "step": 2350
-    },
-    {
-      "epoch": 0.37859817222915576,
-      "grad_norm": 0.0014234265545383096,
-      "learning_rate": 0.00019999992986647093,
-      "loss": 46.0,
-      "step": 2351
-    },
-    {
-      "epoch": 0.37875920930794316,
-      "grad_norm": 0.0008165540639311075,
-      "learning_rate": 0.00019999992980654057,
-      "loss": 46.0,
-      "step": 2352
-    },
-    {
-      "epoch": 0.37892024638673055,
-      "grad_norm": 0.00033828490995801985,
-      "learning_rate": 0.00019999992974658462,
-      "loss": 46.0,
-      "step": 2353
-    },
-    {
-      "epoch": 0.37908128346551795,
-      "grad_norm": 0.0009571529226377606,
-      "learning_rate": 0.00019999992968660303,
-      "loss": 46.0,
-      "step": 2354
-    },
-    {
-      "epoch": 0.37924232054430534,
-      "grad_norm": 0.0018818129319697618,
-      "learning_rate": 0.00019999992962659588,
-      "loss": 46.0,
-      "step": 2355
-    },
-    {
-      "epoch": 0.37940335762309274,
-      "grad_norm": 0.0006002320442348719,
-      "learning_rate": 0.00019999992956656315,
-      "loss": 46.0,
-      "step": 2356
-    },
-    {
-      "epoch": 0.3795643947018801,
-      "grad_norm": 0.0019165269332006574,
-      "learning_rate": 0.0001999999295065048,
-      "loss": 46.0,
-      "step": 2357
-    },
-    {
-      "epoch": 0.3797254317806675,
-      "grad_norm": 0.0013940882636234164,
-      "learning_rate": 0.00019999992944642086,
-      "loss": 46.0,
-      "step": 2358
-    },
-    {
-      "epoch": 0.3798864688594549,
-      "grad_norm": 0.0002520715643186122,
-      "learning_rate": 0.0001999999293863113,
-      "loss": 46.0,
-      "step": 2359
-    },
-    {
-      "epoch": 0.38004750593824227,
-      "grad_norm": 0.000614637159742415,
-      "learning_rate": 0.0001999999293261762,
-      "loss": 46.0,
-      "step": 2360
-    },
-    {
-      "epoch": 0.38020854301702967,
-      "grad_norm": 0.00031276774825528264,
-      "learning_rate": 0.00019999992926601548,
-      "loss": 46.0,
-      "step": 2361
-    },
-    {
-      "epoch": 0.38036958009581706,
-      "grad_norm": 0.0005117102991789579,
-      "learning_rate": 0.00019999992920582914,
-      "loss": 46.0,
-      "step": 2362
-    },
-    {
-      "epoch": 0.38053061717460446,
-      "grad_norm": 0.00078467873390764,
-      "learning_rate": 0.00019999992914561724,
-      "loss": 46.0,
-      "step": 2363
-    },
-    {
-      "epoch": 0.38069165425339185,
-      "grad_norm": 0.0007993248291313648,
-      "learning_rate": 0.0001999999290853797,
-      "loss": 46.0,
-      "step": 2364
-    },
-    {
-      "epoch": 0.38085269133217925,
-      "grad_norm": 0.0006927712238393724,
-      "learning_rate": 0.00019999992902511662,
-      "loss": 46.0,
-      "step": 2365
-    },
-    {
-      "epoch": 0.38101372841096665,
-      "grad_norm": 0.0004904004745185375,
-      "learning_rate": 0.0001999999289648279,
-      "loss": 46.0,
-      "step": 2366
-    },
-    {
-      "epoch": 0.38117476548975404,
-      "grad_norm": 0.0007624586578458548,
-      "learning_rate": 0.00019999992890451363,
-      "loss": 46.0,
-      "step": 2367
-    },
-    {
-      "epoch": 0.3813358025685414,
-      "grad_norm": 0.0006003980524837971,
-      "learning_rate": 0.00019999992884417372,
-      "loss": 46.0,
-      "step": 2368
-    },
-    {
-      "epoch": 0.3814968396473288,
-      "grad_norm": 0.0012511751847341657,
-      "learning_rate": 0.00019999992878380824,
-      "loss": 46.0,
-      "step": 2369
-    },
-    {
-      "epoch": 0.3816578767261162,
-      "grad_norm": 0.00030171748949214816,
-      "learning_rate": 0.00019999992872341715,
-      "loss": 46.0,
-      "step": 2370
-    },
-    {
-      "epoch": 0.38181891380490357,
-      "grad_norm": 0.000552336685359478,
-      "learning_rate": 0.0001999999286630005,
-      "loss": 46.0,
-      "step": 2371
-    },
-    {
-      "epoch": 0.38197995088369097,
-      "grad_norm": 0.0003679750661831349,
-      "learning_rate": 0.00019999992860255823,
-      "loss": 46.0,
-      "step": 2372
-    },
-    {
-      "epoch": 0.38214098796247836,
-      "grad_norm": 0.0015138769522309303,
-      "learning_rate": 0.00019999992854209035,
-      "loss": 46.0,
-      "step": 2373
-    },
-    {
-      "epoch": 0.38230202504126576,
-      "grad_norm": 0.002658956218510866,
-      "learning_rate": 0.0001999999284815969,
-      "loss": 46.0,
-      "step": 2374
-    },
-    {
-      "epoch": 0.38246306212005315,
-      "grad_norm": 0.0006245792610570788,
-      "learning_rate": 0.00019999992842107783,
-      "loss": 46.0,
-      "step": 2375
-    },
-    {
-      "epoch": 0.38262409919884055,
-      "grad_norm": 0.0011079673422500491,
-      "learning_rate": 0.00019999992836053322,
-      "loss": 46.0,
-      "step": 2376
-    },
-    {
-      "epoch": 0.38278513627762795,
-      "grad_norm": 0.0005209127557463944,
-      "learning_rate": 0.00019999992829996293,
-      "loss": 46.0,
-      "step": 2377
-    },
-    {
-      "epoch": 0.3829461733564153,
-      "grad_norm": 0.0004823681665584445,
-      "learning_rate": 0.00019999992823936712,
-      "loss": 46.0,
-      "step": 2378
-    },
-    {
-      "epoch": 0.3831072104352027,
-      "grad_norm": 0.00042470896732993424,
-      "learning_rate": 0.0001999999281787457,
-      "loss": 46.0,
-      "step": 2379
-    },
-    {
-      "epoch": 0.3832682475139901,
-      "grad_norm": 0.0006351933116093278,
-      "learning_rate": 0.00019999992811809865,
-      "loss": 46.0,
-      "step": 2380
-    },
-    {
-      "epoch": 0.3834292845927775,
-      "grad_norm": 0.00033180080936290324,
-      "learning_rate": 0.00019999992805742601,
-      "loss": 46.0,
-      "step": 2381
-    },
-    {
-      "epoch": 0.38359032167156487,
-      "grad_norm": 0.0039007551968097687,
-      "learning_rate": 0.00019999992799672782,
-      "loss": 46.0,
-      "step": 2382
-    },
-    {
-      "epoch": 0.38375135875035227,
-      "grad_norm": 0.001388771110214293,
-      "learning_rate": 0.000199999927936004,
-      "loss": 46.0,
-      "step": 2383
-    },
-    {
-      "epoch": 0.38391239582913966,
-      "grad_norm": 0.001418697414919734,
-      "learning_rate": 0.0001999999278752546,
-      "loss": 46.0,
-      "step": 2384
-    },
-    {
-      "epoch": 0.38407343290792706,
-      "grad_norm": 0.0015510203083977103,
-      "learning_rate": 0.00019999992781447962,
-      "loss": 46.0,
-      "step": 2385
-    },
-    {
-      "epoch": 0.38423446998671446,
-      "grad_norm": 0.00042005235445685685,
-      "learning_rate": 0.000199999927753679,
-      "loss": 46.0,
-      "step": 2386
-    },
-    {
-      "epoch": 0.38439550706550185,
-      "grad_norm": 0.0018052227096632123,
-      "learning_rate": 0.00019999992769285281,
-      "loss": 46.0,
-      "step": 2387
-    },
-    {
-      "epoch": 0.38455654414428925,
-      "grad_norm": 0.0005109263001941144,
-      "learning_rate": 0.00019999992763200102,
-      "loss": 46.0,
-      "step": 2388
-    },
-    {
-      "epoch": 0.3847175812230766,
-      "grad_norm": 0.0006831470527686179,
-      "learning_rate": 0.00019999992757112364,
-      "loss": 46.0,
-      "step": 2389
-    },
-    {
-      "epoch": 0.384878618301864,
-      "grad_norm": 0.0006294698105193675,
-      "learning_rate": 0.00019999992751022067,
-      "loss": 46.0,
-      "step": 2390
-    },
-    {
-      "epoch": 0.3850396553806514,
-      "grad_norm": 0.001542579149827361,
-      "learning_rate": 0.00019999992744929208,
-      "loss": 46.0,
-      "step": 2391
-    },
-    {
-      "epoch": 0.3852006924594388,
-      "grad_norm": 0.00169956567697227,
-      "learning_rate": 0.00019999992738833794,
-      "loss": 46.0,
-      "step": 2392
-    },
-    {
-      "epoch": 0.38536172953822617,
-      "grad_norm": 0.0014051354955881834,
-      "learning_rate": 0.00019999992732735815,
-      "loss": 46.0,
-      "step": 2393
-    },
-    {
-      "epoch": 0.38552276661701357,
-      "grad_norm": 0.0006234274478629231,
-      "learning_rate": 0.0001999999272663528,
-      "loss": 46.0,
-      "step": 2394
-    },
-    {
-      "epoch": 0.38568380369580096,
-      "grad_norm": 0.00031594731262885034,
-      "learning_rate": 0.00019999992720532185,
-      "loss": 46.0,
-      "step": 2395
-    },
-    {
-      "epoch": 0.38584484077458836,
-      "grad_norm": 0.0011688338126987219,
-      "learning_rate": 0.00019999992714426533,
-      "loss": 46.0,
-      "step": 2396
-    },
-    {
-      "epoch": 0.38600587785337576,
-      "grad_norm": 0.000736875575967133,
-      "learning_rate": 0.00019999992708318317,
-      "loss": 46.0,
-      "step": 2397
-    },
-    {
-      "epoch": 0.38616691493216315,
-      "grad_norm": 0.0012262999080121517,
-      "learning_rate": 0.00019999992702207544,
-      "loss": 46.0,
-      "step": 2398
-    },
-    {
-      "epoch": 0.38632795201095055,
-      "grad_norm": 0.0007673557847738266,
-      "learning_rate": 0.0001999999269609421,
-      "loss": 46.0,
-      "step": 2399
-    },
-    {
-      "epoch": 0.3864889890897379,
-      "grad_norm": 0.0011410334846004844,
-      "learning_rate": 0.00019999992689978318,
-      "loss": 46.0,
-      "step": 2400
-    },
-    {
-      "epoch": 0.3866500261685253,
-      "grad_norm": 0.0007803708431310952,
-      "learning_rate": 0.00019999992683859867,
-      "loss": 46.0,
-      "step": 2401
-    },
-    {
-      "epoch": 0.3868110632473127,
-      "grad_norm": 0.00037778247497044504,
-      "learning_rate": 0.00019999992677738855,
-      "loss": 46.0,
-      "step": 2402
-    },
-    {
-      "epoch": 0.3869721003261001,
-      "grad_norm": 0.0013648406602442265,
-      "learning_rate": 0.00019999992671615284,
-      "loss": 46.0,
-      "step": 2403
-    },
-    {
-      "epoch": 0.3871331374048875,
-      "grad_norm": 0.00034797709668055177,
-      "learning_rate": 0.00019999992665489154,
-      "loss": 46.0,
-      "step": 2404
-    },
-    {
-      "epoch": 0.38729417448367487,
-      "grad_norm": 0.000945565989241004,
-      "learning_rate": 0.00019999992659360465,
-      "loss": 46.0,
-      "step": 2405
-    },
-    {
-      "epoch": 0.38745521156246227,
-      "grad_norm": 0.00022659880050923675,
-      "learning_rate": 0.00019999992653229213,
-      "loss": 46.0,
-      "step": 2406
-    },
-    {
-      "epoch": 0.38761624864124966,
-      "grad_norm": 0.0005075627123005688,
-      "learning_rate": 0.00019999992647095404,
-      "loss": 46.0,
-      "step": 2407
-    },
-    {
-      "epoch": 0.38777728572003706,
-      "grad_norm": 0.0009664405952207744,
-      "learning_rate": 0.00019999992640959036,
-      "loss": 46.0,
-      "step": 2408
-    },
-    {
-      "epoch": 0.38793832279882445,
-      "grad_norm": 0.00041972025064751506,
-      "learning_rate": 0.00019999992634820108,
-      "loss": 46.0,
-      "step": 2409
-    },
-    {
-      "epoch": 0.3880993598776118,
-      "grad_norm": 0.0007176906801760197,
-      "learning_rate": 0.0001999999262867862,
-      "loss": 46.0,
-      "step": 2410
-    },
-    {
-      "epoch": 0.3882603969563992,
-      "grad_norm": 0.0008887439616955817,
-      "learning_rate": 0.00019999992622534576,
-      "loss": 46.0,
-      "step": 2411
-    },
-    {
-      "epoch": 0.3884214340351866,
-      "grad_norm": 0.0003332272171974182,
-      "learning_rate": 0.00019999992616387969,
-      "loss": 46.0,
-      "step": 2412
-    },
-    {
-      "epoch": 0.388582471113974,
-      "grad_norm": 0.0008063961286097765,
-      "learning_rate": 0.00019999992610238802,
-      "loss": 46.0,
-      "step": 2413
-    },
-    {
-      "epoch": 0.3887435081927614,
-      "grad_norm": 0.0011459635570645332,
-      "learning_rate": 0.0001999999260408708,
-      "loss": 46.0,
-      "step": 2414
-    },
-    {
-      "epoch": 0.3889045452715488,
-      "grad_norm": 0.0006852532969787717,
-      "learning_rate": 0.00019999992597932793,
-      "loss": 46.0,
-      "step": 2415
-    },
-    {
-      "epoch": 0.38906558235033617,
-      "grad_norm": 0.0004779025912284851,
-      "learning_rate": 0.00019999992591775948,
-      "loss": 46.0,
-      "step": 2416
-    },
-    {
-      "epoch": 0.38922661942912357,
-      "grad_norm": 0.0016908718971535563,
-      "learning_rate": 0.00019999992585616544,
-      "loss": 46.0,
-      "step": 2417
-    },
-    {
-      "epoch": 0.38938765650791096,
-      "grad_norm": 0.0002849120646715164,
-      "learning_rate": 0.00019999992579454584,
-      "loss": 46.0,
-      "step": 2418
-    },
-    {
-      "epoch": 0.38954869358669836,
-      "grad_norm": 0.0003883201570715755,
-      "learning_rate": 0.0001999999257329006,
-      "loss": 46.0,
-      "step": 2419
-    },
-    {
-      "epoch": 0.38970973066548575,
-      "grad_norm": 0.001718754880130291,
-      "learning_rate": 0.00019999992567122977,
-      "loss": 46.0,
-      "step": 2420
-    },
-    {
-      "epoch": 0.3898707677442731,
-      "grad_norm": 0.0005050842883065343,
-      "learning_rate": 0.00019999992560953335,
-      "loss": 46.0,
-      "step": 2421
-    },
-    {
-      "epoch": 0.3900318048230605,
-      "grad_norm": 0.0005970178754068911,
-      "learning_rate": 0.00019999992554781132,
-      "loss": 46.0,
-      "step": 2422
-    },
-    {
-      "epoch": 0.3901928419018479,
-      "grad_norm": 0.0003626322722993791,
-      "learning_rate": 0.00019999992548606376,
-      "loss": 46.0,
-      "step": 2423
-    },
-    {
-      "epoch": 0.3903538789806353,
-      "grad_norm": 0.0005492300260812044,
-      "learning_rate": 0.00019999992542429052,
-      "loss": 46.0,
-      "step": 2424
-    },
-    {
-      "epoch": 0.3905149160594227,
-      "grad_norm": 0.0004774110275320709,
-      "learning_rate": 0.00019999992536249176,
-      "loss": 46.0,
-      "step": 2425
-    },
-    {
-      "epoch": 0.3906759531382101,
-      "grad_norm": 0.00033885202719829977,
-      "learning_rate": 0.00019999992530066733,
-      "loss": 46.0,
-      "step": 2426
-    },
-    {
-      "epoch": 0.39083699021699747,
-      "grad_norm": 0.0003523860650602728,
-      "learning_rate": 0.00019999992523881736,
-      "loss": 46.0,
-      "step": 2427
-    },
-    {
-      "epoch": 0.39099802729578487,
-      "grad_norm": 0.0005243658670224249,
-      "learning_rate": 0.00019999992517694178,
-      "loss": 46.0,
-      "step": 2428
-    },
-    {
-      "epoch": 0.39115906437457226,
-      "grad_norm": 0.0021557840518653393,
-      "learning_rate": 0.0001999999251150406,
-      "loss": 46.0,
-      "step": 2429
-    },
-    {
-      "epoch": 0.39132010145335966,
-      "grad_norm": 0.0014437641948461533,
-      "learning_rate": 0.00019999992505311386,
-      "loss": 46.0,
-      "step": 2430
-    },
-    {
-      "epoch": 0.391481138532147,
-      "grad_norm": 0.0017680978635326028,
-      "learning_rate": 0.00019999992499116146,
-      "loss": 46.0,
-      "step": 2431
-    },
-    {
-      "epoch": 0.3916421756109344,
-      "grad_norm": 0.0006148244137875736,
-      "learning_rate": 0.0001999999249291835,
-      "loss": 46.0,
-      "step": 2432
-    },
-    {
-      "epoch": 0.3918032126897218,
-      "grad_norm": 0.0006314045749604702,
-      "learning_rate": 0.00019999992486717996,
-      "loss": 46.0,
-      "step": 2433
-    },
-    {
-      "epoch": 0.3919642497685092,
-      "grad_norm": 0.0009844782762229443,
-      "learning_rate": 0.00019999992480515083,
-      "loss": 46.0,
-      "step": 2434
-    },
-    {
-      "epoch": 0.3921252868472966,
-      "grad_norm": 0.001140897162258625,
-      "learning_rate": 0.00019999992474309606,
-      "loss": 46.0,
-      "step": 2435
-    },
-    {
-      "epoch": 0.392286323926084,
-      "grad_norm": 0.0005091571947559714,
-      "learning_rate": 0.00019999992468101572,
-      "loss": 46.0,
-      "step": 2436
-    },
-    {
-      "epoch": 0.3924473610048714,
-      "grad_norm": 0.00027712646988220513,
-      "learning_rate": 0.0001999999246189098,
-      "loss": 46.0,
-      "step": 2437
-    },
-    {
-      "epoch": 0.39260839808365877,
-      "grad_norm": 0.0008853934123180807,
-      "learning_rate": 0.00019999992455677824,
-      "loss": 46.0,
-      "step": 2438
-    },
-    {
-      "epoch": 0.39276943516244617,
-      "grad_norm": 0.0011501384433358908,
-      "learning_rate": 0.00019999992449462112,
-      "loss": 46.0,
-      "step": 2439
-    },
-    {
-      "epoch": 0.39293047224123356,
-      "grad_norm": 0.000260513334069401,
-      "learning_rate": 0.00019999992443243842,
-      "loss": 46.0,
-      "step": 2440
-    },
-    {
-      "epoch": 0.39309150932002096,
-      "grad_norm": 0.0013478387845680118,
-      "learning_rate": 0.0001999999243702301,
-      "loss": 46.0,
-      "step": 2441
-    },
-    {
-      "epoch": 0.3932525463988083,
-      "grad_norm": 0.0021682940423488617,
-      "learning_rate": 0.00019999992430799618,
-      "loss": 46.0,
-      "step": 2442
-    },
-    {
-      "epoch": 0.3934135834775957,
-      "grad_norm": 0.00033404288114979863,
-      "learning_rate": 0.00019999992424573671,
-      "loss": 46.0,
-      "step": 2443
-    },
-    {
-      "epoch": 0.3935746205563831,
-      "grad_norm": 0.0008781529031693935,
-      "learning_rate": 0.0001999999241834516,
-      "loss": 46.0,
-      "step": 2444
-    },
-    {
-      "epoch": 0.3937356576351705,
-      "grad_norm": 0.0006088555091992021,
-      "learning_rate": 0.0001999999241211409,
-      "loss": 46.0,
-      "step": 2445
-    },
-    {
-      "epoch": 0.3938966947139579,
-      "grad_norm": 0.00035477944766171277,
-      "learning_rate": 0.00019999992405880462,
-      "loss": 46.0,
-      "step": 2446
-    },
-    {
-      "epoch": 0.3940577317927453,
-      "grad_norm": 0.00033839233219623566,
-      "learning_rate": 0.00019999992399644275,
-      "loss": 46.0,
-      "step": 2447
-    },
-    {
-      "epoch": 0.3942187688715327,
-      "grad_norm": 0.0007759662694297731,
-      "learning_rate": 0.00019999992393405526,
-      "loss": 46.0,
-      "step": 2448
-    },
-    {
-      "epoch": 0.3943798059503201,
-      "grad_norm": 0.0020294003188610077,
-      "learning_rate": 0.00019999992387164216,
-      "loss": 46.0,
-      "step": 2449
-    },
-    {
-      "epoch": 0.39454084302910747,
-      "grad_norm": 0.00037049237289465964,
-      "learning_rate": 0.00019999992380920353,
-      "loss": 46.0,
-      "step": 2450
-    },
-    {
-      "epoch": 0.39470188010789486,
-      "grad_norm": 0.00031506139202974737,
-      "learning_rate": 0.00019999992374673922,
-      "loss": 46.0,
-      "step": 2451
-    },
-    {
-      "epoch": 0.39486291718668226,
-      "grad_norm": 0.0017361241625621915,
-      "learning_rate": 0.0001999999236842494,
-      "loss": 46.0,
-      "step": 2452
-    },
-    {
-      "epoch": 0.3950239542654696,
-      "grad_norm": 0.000610153132583946,
-      "learning_rate": 0.00019999992362173394,
-      "loss": 46.0,
-      "step": 2453
-    },
-    {
-      "epoch": 0.395184991344257,
-      "grad_norm": 0.0006328023737296462,
-      "learning_rate": 0.0001999999235591929,
-      "loss": 46.0,
-      "step": 2454
-    },
-    {
-      "epoch": 0.3953460284230444,
-      "grad_norm": 0.0005528049659915268,
-      "learning_rate": 0.00019999992349662625,
-      "loss": 46.0,
-      "step": 2455
-    },
-    {
-      "epoch": 0.3955070655018318,
-      "grad_norm": 0.0006624308880418539,
-      "learning_rate": 0.000199999923434034,
-      "loss": 46.0,
-      "step": 2456
-    },
-    {
-      "epoch": 0.3956681025806192,
-      "grad_norm": 0.0005569378845393658,
-      "learning_rate": 0.0001999999233714162,
-      "loss": 46.0,
-      "step": 2457
-    },
-    {
-      "epoch": 0.3958291396594066,
-      "grad_norm": 0.0006670751608908176,
-      "learning_rate": 0.00019999992330877277,
-      "loss": 46.0,
-      "step": 2458
-    },
-    {
-      "epoch": 0.395990176738194,
-      "grad_norm": 0.0006159030017443001,
-      "learning_rate": 0.00019999992324610375,
-      "loss": 46.0,
-      "step": 2459
-    },
-    {
-      "epoch": 0.3961512138169814,
-      "grad_norm": 0.0028524715453386307,
-      "learning_rate": 0.00019999992318340913,
-      "loss": 46.0,
-      "step": 2460
-    },
-    {
-      "epoch": 0.39631225089576877,
-      "grad_norm": 0.0006164934602566063,
-      "learning_rate": 0.0001999999231206889,
-      "loss": 46.0,
-      "step": 2461
-    },
-    {
-      "epoch": 0.39647328797455617,
-      "grad_norm": 0.0015587115194648504,
-      "learning_rate": 0.00019999992305794312,
-      "loss": 46.0,
-      "step": 2462
-    },
-    {
-      "epoch": 0.3966343250533435,
-      "grad_norm": 0.0005712852580472827,
-      "learning_rate": 0.00019999992299517171,
-      "loss": 46.0,
-      "step": 2463
-    },
-    {
-      "epoch": 0.3967953621321309,
-      "grad_norm": 0.0016132298624143004,
-      "learning_rate": 0.0001999999229323747,
-      "loss": 46.0,
-      "step": 2464
-    },
-    {
-      "epoch": 0.3969563992109183,
-      "grad_norm": 0.0005810867296531796,
-      "learning_rate": 0.00019999992286955212,
-      "loss": 46.0,
-      "step": 2465
-    },
-    {
-      "epoch": 0.3971174362897057,
-      "grad_norm": 0.00031309822225011885,
-      "learning_rate": 0.00019999992280670396,
-      "loss": 46.0,
-      "step": 2466
-    },
-    {
-      "epoch": 0.3972784733684931,
-      "grad_norm": 0.0009651644504629076,
-      "learning_rate": 0.00019999992274383015,
-      "loss": 46.0,
-      "step": 2467
-    },
-    {
-      "epoch": 0.3974395104472805,
-      "grad_norm": 0.0007141558453440666,
-      "learning_rate": 0.00019999992268093079,
-      "loss": 46.0,
-      "step": 2468
-    },
-    {
-      "epoch": 0.3976005475260679,
-      "grad_norm": 0.00030265108216553926,
-      "learning_rate": 0.00019999992261800583,
-      "loss": 46.0,
-      "step": 2469
-    },
-    {
-      "epoch": 0.3977615846048553,
-      "grad_norm": 0.0005087955505587161,
-      "learning_rate": 0.00019999992255505524,
-      "loss": 46.0,
-      "step": 2470
-    },
-    {
-      "epoch": 0.3979226216836427,
-      "grad_norm": 0.001312676933594048,
-      "learning_rate": 0.00019999992249207908,
-      "loss": 46.0,
-      "step": 2471
-    },
-    {
-      "epoch": 0.39808365876243007,
-      "grad_norm": 0.00132571323774755,
-      "learning_rate": 0.00019999992242907734,
-      "loss": 46.0,
-      "step": 2472
-    },
-    {
-      "epoch": 0.39824469584121747,
-      "grad_norm": 0.0004774648405145854,
-      "learning_rate": 0.00019999992236605,
-      "loss": 46.0,
-      "step": 2473
-    },
-    {
-      "epoch": 0.3984057329200048,
-      "grad_norm": 0.0006745936698280275,
-      "learning_rate": 0.00019999992230299704,
-      "loss": 46.0,
-      "step": 2474
-    },
-    {
-      "epoch": 0.3985667699987922,
-      "grad_norm": 0.002986040897667408,
-      "learning_rate": 0.00019999992223991851,
-      "loss": 46.0,
-      "step": 2475
-    },
-    {
-      "epoch": 0.3987278070775796,
-      "grad_norm": 0.0011012536706402898,
-      "learning_rate": 0.00019999992217681437,
-      "loss": 46.0,
-      "step": 2476
-    },
-    {
-      "epoch": 0.398888844156367,
-      "grad_norm": 0.0002778563357423991,
-      "learning_rate": 0.00019999992211368467,
-      "loss": 46.0,
-      "step": 2477
-    },
-    {
-      "epoch": 0.3990498812351544,
-      "grad_norm": 0.0005057700909674168,
-      "learning_rate": 0.00019999992205052932,
-      "loss": 46.0,
-      "step": 2478
-    },
-    {
-      "epoch": 0.3992109183139418,
-      "grad_norm": 0.0008647431968711317,
-      "learning_rate": 0.00019999992198734841,
-      "loss": 46.0,
-      "step": 2479
-    },
-    {
-      "epoch": 0.3993719553927292,
-      "grad_norm": 0.0019028813112527132,
-      "learning_rate": 0.00019999992192414187,
-      "loss": 46.0,
-      "step": 2480
-    },
-    {
-      "epoch": 0.3995329924715166,
-      "grad_norm": 0.0002257141750305891,
-      "learning_rate": 0.00019999992186090979,
-      "loss": 46.0,
-      "step": 2481
-    },
-    {
-      "epoch": 0.399694029550304,
-      "grad_norm": 0.0005610088701359928,
-      "learning_rate": 0.0001999999217976521,
-      "loss": 46.0,
-      "step": 2482
-    },
-    {
-      "epoch": 0.39985506662909137,
-      "grad_norm": 0.002196467248722911,
-      "learning_rate": 0.00019999992173436878,
-      "loss": 46.0,
-      "step": 2483
-    },
-    {
-      "epoch": 0.4000161037078787,
-      "grad_norm": 0.00036643995554186404,
-      "learning_rate": 0.0001999999216710599,
-      "loss": 46.0,
-      "step": 2484
-    },
-    {
-      "epoch": 0.4001771407866661,
-      "grad_norm": 0.00037171339499764144,
-      "learning_rate": 0.0001999999216077254,
-      "loss": 46.0,
-      "step": 2485
-    },
-    {
-      "epoch": 0.4003381778654535,
-      "grad_norm": 0.0003727784496732056,
-      "learning_rate": 0.0001999999215443653,
-      "loss": 46.0,
-      "step": 2486
-    },
-    {
-      "epoch": 0.4004992149442409,
-      "grad_norm": 0.0017312564887106419,
-      "learning_rate": 0.00019999992148097965,
-      "loss": 46.0,
-      "step": 2487
-    },
-    {
-      "epoch": 0.4006602520230283,
-      "grad_norm": 0.0003423290909267962,
-      "learning_rate": 0.00019999992141756838,
-      "loss": 46.0,
-      "step": 2488
-    },
-    {
-      "epoch": 0.4008212891018157,
-      "grad_norm": 0.0002671041584108025,
-      "learning_rate": 0.00019999992135413152,
-      "loss": 46.0,
-      "step": 2489
-    },
-    {
-      "epoch": 0.4009823261806031,
-      "grad_norm": 0.000793307728599757,
-      "learning_rate": 0.00019999992129066905,
-      "loss": 46.0,
-      "step": 2490
-    },
-    {
-      "epoch": 0.4011433632593905,
-      "grad_norm": 0.0008093641954474151,
-      "learning_rate": 0.00019999992122718099,
-      "loss": 46.0,
-      "step": 2491
-    },
-    {
-      "epoch": 0.4013044003381779,
-      "grad_norm": 0.0028547835536301136,
-      "learning_rate": 0.00019999992116366734,
-      "loss": 46.0,
-      "step": 2492
-    },
-    {
-      "epoch": 0.4014654374169653,
-      "grad_norm": 0.00037326946039684117,
-      "learning_rate": 0.00019999992110012808,
-      "loss": 46.0,
-      "step": 2493
-    },
-    {
-      "epoch": 0.40162647449575267,
-      "grad_norm": 0.0007795938872732222,
-      "learning_rate": 0.00019999992103656325,
-      "loss": 46.0,
-      "step": 2494
-    },
-    {
-      "epoch": 0.40178751157454,
-      "grad_norm": 0.0005893181078135967,
-      "learning_rate": 0.00019999992097297282,
-      "loss": 46.0,
-      "step": 2495
-    },
-    {
-      "epoch": 0.4019485486533274,
-      "grad_norm": 0.0003579497570171952,
-      "learning_rate": 0.00019999992090935677,
-      "loss": 46.0,
-      "step": 2496
-    },
-    {
-      "epoch": 0.4021095857321148,
-      "grad_norm": 0.000713040994014591,
-      "learning_rate": 0.00019999992084571516,
-      "loss": 46.0,
-      "step": 2497
-    },
-    {
-      "epoch": 0.4022706228109022,
-      "grad_norm": 0.0009842911968007684,
-      "learning_rate": 0.00019999992078204793,
-      "loss": 46.0,
-      "step": 2498
-    },
-    {
-      "epoch": 0.4024316598896896,
-      "grad_norm": 0.0007644824800081551,
-      "learning_rate": 0.00019999992071835512,
-      "loss": 46.0,
-      "step": 2499
-    },
-    {
-      "epoch": 0.402592696968477,
-      "grad_norm": 0.001677367021329701,
-      "learning_rate": 0.0001999999206546367,
-      "loss": 46.0,
-      "step": 2500
-    },
-    {
-      "epoch": 0.4027537340472644,
-      "grad_norm": 0.0011205823393538594,
-      "learning_rate": 0.0001999999205908927,
-      "loss": 46.0,
-      "step": 2501
-    },
-    {
-      "epoch": 0.4029147711260518,
-      "grad_norm": 0.0016218560049310327,
-      "learning_rate": 0.0001999999205271231,
-      "loss": 46.0,
-      "step": 2502
-    },
-    {
-      "epoch": 0.4030758082048392,
-      "grad_norm": 0.00045920678530819714,
-      "learning_rate": 0.00019999992046332792,
-      "loss": 46.0,
-      "step": 2503
-    },
-    {
-      "epoch": 0.4032368452836266,
-      "grad_norm": 0.0013049228582531214,
-      "learning_rate": 0.00019999992039950715,
-      "loss": 46.0,
-      "step": 2504
-    },
-    {
-      "epoch": 0.403397882362414,
-      "grad_norm": 0.0003285286366008222,
-      "learning_rate": 0.00019999992033566076,
-      "loss": 46.0,
-      "step": 2505
-    },
-    {
-      "epoch": 0.4035589194412013,
-      "grad_norm": 0.0005516027449630201,
-      "learning_rate": 0.00019999992027178878,
-      "loss": 46.0,
-      "step": 2506
-    },
-    {
-      "epoch": 0.4037199565199887,
-      "grad_norm": 0.004015455488115549,
-      "learning_rate": 0.0001999999202078912,
-      "loss": 46.0,
-      "step": 2507
-    },
-    {
-      "epoch": 0.4038809935987761,
-      "grad_norm": 0.0006361909327097237,
-      "learning_rate": 0.00019999992014396804,
-      "loss": 46.0,
-      "step": 2508
-    },
-    {
-      "epoch": 0.4040420306775635,
-      "grad_norm": 0.000788997276686132,
-      "learning_rate": 0.00019999992008001927,
-      "loss": 46.0,
-      "step": 2509
-    },
-    {
-      "epoch": 0.4042030677563509,
-      "grad_norm": 0.002198253758251667,
-      "learning_rate": 0.0001999999200160449,
-      "loss": 46.0,
-      "step": 2510
-    },
-    {
-      "epoch": 0.4043641048351383,
-      "grad_norm": 0.0007323686149902642,
-      "learning_rate": 0.00019999991995204496,
-      "loss": 46.0,
-      "step": 2511
-    },
-    {
-      "epoch": 0.4045251419139257,
-      "grad_norm": 0.0005261424812488258,
-      "learning_rate": 0.0001999999198880194,
-      "loss": 46.0,
-      "step": 2512
-    },
-    {
-      "epoch": 0.4046861789927131,
-      "grad_norm": 0.0007068556733429432,
-      "learning_rate": 0.00019999991982396826,
-      "loss": 46.0,
-      "step": 2513
-    },
-    {
-      "epoch": 0.4048472160715005,
-      "grad_norm": 0.0014703463530167937,
-      "learning_rate": 0.00019999991975989153,
-      "loss": 46.0,
-      "step": 2514
-    },
-    {
-      "epoch": 0.4050082531502879,
-      "grad_norm": 0.0029368470422923565,
-      "learning_rate": 0.0001999999196957892,
-      "loss": 46.0,
-      "step": 2515
-    },
-    {
-      "epoch": 0.4051692902290752,
-      "grad_norm": 0.0005351447034627199,
-      "learning_rate": 0.00019999991963166126,
-      "loss": 46.0,
-      "step": 2516
-    },
-    {
-      "epoch": 0.4053303273078626,
-      "grad_norm": 0.0005643073818646371,
-      "learning_rate": 0.00019999991956750775,
-      "loss": 46.0,
-      "step": 2517
-    },
-    {
-      "epoch": 0.40549136438665,
-      "grad_norm": 0.0005366371478885412,
-      "learning_rate": 0.00019999991950332864,
-      "loss": 46.0,
-      "step": 2518
-    },
-    {
-      "epoch": 0.4056524014654374,
-      "grad_norm": 0.0009746613213792443,
-      "learning_rate": 0.0001999999194391239,
-      "loss": 46.0,
-      "step": 2519
-    },
-    {
-      "epoch": 0.4058134385442248,
-      "grad_norm": 0.0007656855159439147,
-      "learning_rate": 0.0001999999193748936,
-      "loss": 46.0,
-      "step": 2520
-    },
-    {
-      "epoch": 0.4059744756230122,
-      "grad_norm": 0.0004048139089718461,
-      "learning_rate": 0.0001999999193106377,
-      "loss": 46.0,
-      "step": 2521
-    },
-    {
-      "epoch": 0.4061355127017996,
-      "grad_norm": 0.002632913878187537,
-      "learning_rate": 0.00019999991924635622,
-      "loss": 46.0,
-      "step": 2522
-    },
-    {
-      "epoch": 0.406296549780587,
-      "grad_norm": 0.0005818718345835805,
-      "learning_rate": 0.00019999991918204913,
-      "loss": 46.0,
-      "step": 2523
-    },
-    {
-      "epoch": 0.4064575868593744,
-      "grad_norm": 0.0002545182651374489,
-      "learning_rate": 0.00019999991911771642,
-      "loss": 46.0,
-      "step": 2524
-    },
-    {
-      "epoch": 0.4066186239381618,
-      "grad_norm": 0.0008372089941985905,
-      "learning_rate": 0.00019999991905335815,
-      "loss": 46.0,
-      "step": 2525
-    },
-    {
-      "epoch": 0.4067796610169492,
-      "grad_norm": 0.0005814693286083639,
-      "learning_rate": 0.00019999991898897427,
-      "loss": 46.0,
-      "step": 2526
-    },
-    {
-      "epoch": 0.4069406980957365,
-      "grad_norm": 0.0012209018459543586,
-      "learning_rate": 0.0001999999189245648,
-      "loss": 46.0,
-      "step": 2527
-    },
-    {
-      "epoch": 0.4071017351745239,
-      "grad_norm": 0.0006315638311207294,
-      "learning_rate": 0.00019999991886012974,
-      "loss": 46.0,
-      "step": 2528
-    },
-    {
-      "epoch": 0.4072627722533113,
-      "grad_norm": 0.00044594131759367883,
-      "learning_rate": 0.00019999991879566907,
-      "loss": 46.0,
-      "step": 2529
-    },
-    {
-      "epoch": 0.4074238093320987,
-      "grad_norm": 0.00035840366035699844,
-      "learning_rate": 0.0001999999187311828,
-      "loss": 46.0,
-      "step": 2530
-    },
-    {
-      "epoch": 0.4075848464108861,
-      "grad_norm": 0.0013362610479816794,
-      "learning_rate": 0.00019999991866667097,
-      "loss": 46.0,
-      "step": 2531
-    },
-    {
-      "epoch": 0.4077458834896735,
-      "grad_norm": 0.00030423692078329623,
-      "learning_rate": 0.0001999999186021335,
-      "loss": 46.0,
-      "step": 2532
-    },
-    {
-      "epoch": 0.4079069205684609,
-      "grad_norm": 0.00022467058442998677,
-      "learning_rate": 0.0001999999185375705,
-      "loss": 46.0,
-      "step": 2533
-    },
-    {
-      "epoch": 0.4080679576472483,
-      "grad_norm": 0.0010856473818421364,
-      "learning_rate": 0.00019999991847298185,
-      "loss": 46.0,
-      "step": 2534
-    },
-    {
-      "epoch": 0.4082289947260357,
-      "grad_norm": 0.0006642889929935336,
-      "learning_rate": 0.00019999991840836763,
-      "loss": 46.0,
-      "step": 2535
-    },
-    {
-      "epoch": 0.4083900318048231,
-      "grad_norm": 0.0009202100918628275,
-      "learning_rate": 0.0001999999183437278,
-      "loss": 46.0,
-      "step": 2536
-    },
-    {
-      "epoch": 0.4085510688836104,
-      "grad_norm": 0.001075320178642869,
-      "learning_rate": 0.0001999999182790624,
-      "loss": 46.0,
-      "step": 2537
-    },
-    {
-      "epoch": 0.4087121059623978,
-      "grad_norm": 0.0011671797838062048,
-      "learning_rate": 0.0001999999182143714,
-      "loss": 46.0,
-      "step": 2538
-    },
-    {
-      "epoch": 0.4088731430411852,
-      "grad_norm": 0.00038806599332019687,
-      "learning_rate": 0.00019999991814965474,
-      "loss": 46.0,
-      "step": 2539
-    },
-    {
-      "epoch": 0.4090341801199726,
-      "grad_norm": 0.00021556422871071845,
-      "learning_rate": 0.00019999991808491256,
-      "loss": 46.0,
-      "step": 2540
-    },
-    {
-      "epoch": 0.40919521719876,
-      "grad_norm": 0.0013159997761249542,
-      "learning_rate": 0.00019999991802014476,
-      "loss": 46.0,
-      "step": 2541
-    },
-    {
-      "epoch": 0.4093562542775474,
-      "grad_norm": 0.0002375345939071849,
-      "learning_rate": 0.00019999991795535137,
-      "loss": 46.0,
-      "step": 2542
-    },
-    {
-      "epoch": 0.4095172913563348,
-      "grad_norm": 0.0005722291534766555,
-      "learning_rate": 0.00019999991789053237,
-      "loss": 46.0,
-      "step": 2543
-    },
-    {
-      "epoch": 0.4096783284351222,
-      "grad_norm": 0.0011116081150248647,
-      "learning_rate": 0.00019999991782568781,
-      "loss": 46.0,
-      "step": 2544
-    },
-    {
-      "epoch": 0.4098393655139096,
-      "grad_norm": 0.0003885219630319625,
-      "learning_rate": 0.0001999999177608176,
-      "loss": 46.0,
-      "step": 2545
-    },
-    {
-      "epoch": 0.410000402592697,
-      "grad_norm": 0.0011035461211577058,
-      "learning_rate": 0.00019999991769592182,
-      "loss": 46.0,
-      "step": 2546
-    },
-    {
-      "epoch": 0.4101614396714844,
-      "grad_norm": 0.0005727497627958655,
-      "learning_rate": 0.00019999991763100048,
-      "loss": 46.0,
-      "step": 2547
-    },
-    {
-      "epoch": 0.4103224767502717,
-      "grad_norm": 0.0010675222147256136,
-      "learning_rate": 0.0001999999175660535,
-      "loss": 46.0,
-      "step": 2548
-    },
-    {
-      "epoch": 0.4104835138290591,
-      "grad_norm": 0.000688559259288013,
-      "learning_rate": 0.00019999991750108096,
-      "loss": 46.0,
-      "step": 2549
-    },
-    {
-      "epoch": 0.4106445509078465,
-      "grad_norm": 0.0010269100312143564,
-      "learning_rate": 0.0001999999174360828,
-      "loss": 46.0,
-      "step": 2550
-    },
-    {
-      "epoch": 0.4108055879866339,
-      "grad_norm": 0.0005960292182862759,
-      "learning_rate": 0.00019999991737105908,
-      "loss": 46.0,
-      "step": 2551
-    },
-    {
-      "epoch": 0.4109666250654213,
-      "grad_norm": 0.0004713739617727697,
-      "learning_rate": 0.00019999991730600974,
-      "loss": 46.0,
-      "step": 2552
-    },
-    {
-      "epoch": 0.4111276621442087,
-      "grad_norm": 0.0007000573677942157,
-      "learning_rate": 0.00019999991724093478,
-      "loss": 46.0,
-      "step": 2553
-    },
-    {
-      "epoch": 0.4112886992229961,
-      "grad_norm": 0.0014249715022742748,
-      "learning_rate": 0.00019999991717583427,
-      "loss": 46.0,
-      "step": 2554
-    },
-    {
-      "epoch": 0.4114497363017835,
-      "grad_norm": 0.0004502889350987971,
-      "learning_rate": 0.00019999991711070814,
-      "loss": 46.0,
-      "step": 2555
-    },
-    {
-      "epoch": 0.4116107733805709,
-      "grad_norm": 0.0005693979328498244,
-      "learning_rate": 0.0001999999170455564,
-      "loss": 46.0,
-      "step": 2556
-    },
-    {
-      "epoch": 0.4117718104593583,
-      "grad_norm": 0.002164100529626012,
-      "learning_rate": 0.0001999999169803791,
-      "loss": 46.0,
-      "step": 2557
-    },
-    {
-      "epoch": 0.4119328475381457,
-      "grad_norm": 0.000281511020148173,
-      "learning_rate": 0.00019999991691517619,
-      "loss": 46.0,
-      "step": 2558
-    },
-    {
-      "epoch": 0.412093884616933,
-      "grad_norm": 0.0004598365630954504,
-      "learning_rate": 0.00019999991684994766,
-      "loss": 46.0,
-      "step": 2559
-    },
-    {
-      "epoch": 0.4122549216957204,
-      "grad_norm": 0.0005771976429969072,
-      "learning_rate": 0.00019999991678469356,
-      "loss": 46.0,
-      "step": 2560
-    },
-    {
-      "epoch": 0.4124159587745078,
-      "grad_norm": 0.0004161202523391694,
-      "learning_rate": 0.00019999991671941386,
-      "loss": 46.0,
-      "step": 2561
-    },
-    {
-      "epoch": 0.4125769958532952,
-      "grad_norm": 0.001491861417889595,
-      "learning_rate": 0.00019999991665410857,
-      "loss": 46.0,
-      "step": 2562
-    },
-    {
-      "epoch": 0.4127380329320826,
-      "grad_norm": 0.0007350731175392866,
-      "learning_rate": 0.0001999999165887777,
-      "loss": 46.0,
-      "step": 2563
-    },
-    {
-      "epoch": 0.41289907001087,
-      "grad_norm": 0.0006727223517373204,
-      "learning_rate": 0.0001999999165234212,
-      "loss": 46.0,
-      "step": 2564
-    },
-    {
-      "epoch": 0.4130601070896574,
-      "grad_norm": 0.000969410699326545,
-      "learning_rate": 0.00019999991645803914,
-      "loss": 46.0,
-      "step": 2565
-    },
-    {
-      "epoch": 0.4132211441684448,
-      "grad_norm": 0.0005818248027935624,
-      "learning_rate": 0.00019999991639263148,
-      "loss": 46.0,
-      "step": 2566
-    },
-    {
-      "epoch": 0.4133821812472322,
-      "grad_norm": 0.0004061332147102803,
-      "learning_rate": 0.00019999991632719822,
-      "loss": 46.0,
-      "step": 2567
-    },
-    {
-      "epoch": 0.4135432183260196,
-      "grad_norm": 0.0004701008729171008,
-      "learning_rate": 0.00019999991626173935,
-      "loss": 46.0,
-      "step": 2568
-    },
-    {
-      "epoch": 0.41370425540480693,
-      "grad_norm": 0.001591149833984673,
-      "learning_rate": 0.0001999999161962549,
-      "loss": 46.0,
-      "step": 2569
-    },
-    {
-      "epoch": 0.4138652924835943,
-      "grad_norm": 0.0022589925210922956,
-      "learning_rate": 0.00019999991613074485,
-      "loss": 46.0,
-      "step": 2570
-    },
-    {
-      "epoch": 0.4140263295623817,
-      "grad_norm": 0.0006316992803476751,
-      "learning_rate": 0.0001999999160652092,
-      "loss": 46.0,
-      "step": 2571
-    },
-    {
-      "epoch": 0.4141873666411691,
-      "grad_norm": 0.0003919966402463615,
-      "learning_rate": 0.00019999991599964795,
-      "loss": 46.0,
-      "step": 2572
-    },
-    {
-      "epoch": 0.4143484037199565,
-      "grad_norm": 0.000711875210981816,
-      "learning_rate": 0.00019999991593406112,
-      "loss": 46.0,
-      "step": 2573
-    },
-    {
-      "epoch": 0.4145094407987439,
-      "grad_norm": 0.0007769575458951294,
-      "learning_rate": 0.0001999999158684487,
-      "loss": 46.0,
-      "step": 2574
-    },
-    {
-      "epoch": 0.4146704778775313,
-      "grad_norm": 0.000219284716877155,
-      "learning_rate": 0.00019999991580281067,
-      "loss": 46.0,
-      "step": 2575
-    },
-    {
-      "epoch": 0.4148315149563187,
-      "grad_norm": 0.00028979365015402436,
-      "learning_rate": 0.00019999991573714705,
-      "loss": 46.0,
-      "step": 2576
-    },
-    {
-      "epoch": 0.4149925520351061,
-      "grad_norm": 0.0017525262665003538,
-      "learning_rate": 0.00019999991567145787,
-      "loss": 46.0,
-      "step": 2577
-    },
-    {
-      "epoch": 0.4151535891138935,
-      "grad_norm": 0.0005864475388079882,
-      "learning_rate": 0.00019999991560574305,
-      "loss": 46.0,
-      "step": 2578
-    },
-    {
-      "epoch": 0.4153146261926809,
-      "grad_norm": 0.0004776372807100415,
-      "learning_rate": 0.00019999991554000264,
-      "loss": 46.0,
-      "step": 2579
-    },
-    {
-      "epoch": 0.41547566327146823,
-      "grad_norm": 0.0011292042909190059,
-      "learning_rate": 0.00019999991547423664,
-      "loss": 46.0,
-      "step": 2580
-    },
-    {
-      "epoch": 0.41563670035025563,
-      "grad_norm": 0.0006078517180867493,
-      "learning_rate": 0.00019999991540844506,
-      "loss": 46.0,
-      "step": 2581
-    },
-    {
-      "epoch": 0.415797737429043,
-      "grad_norm": 0.0007042553625069559,
-      "learning_rate": 0.0001999999153426279,
-      "loss": 46.0,
-      "step": 2582
-    },
-    {
-      "epoch": 0.4159587745078304,
-      "grad_norm": 0.0004453930596355349,
-      "learning_rate": 0.00019999991527678508,
-      "loss": 46.0,
-      "step": 2583
-    },
-    {
-      "epoch": 0.4161198115866178,
-      "grad_norm": 0.0010095108300447464,
-      "learning_rate": 0.0001999999152109167,
-      "loss": 46.0,
-      "step": 2584
-    },
-    {
-      "epoch": 0.4162808486654052,
-      "grad_norm": 0.00038195581873878837,
-      "learning_rate": 0.00019999991514502274,
-      "loss": 46.0,
-      "step": 2585
-    },
-    {
-      "epoch": 0.4164418857441926,
-      "grad_norm": 0.0011428914731368423,
-      "learning_rate": 0.00019999991507910317,
-      "loss": 46.0,
-      "step": 2586
-    },
-    {
-      "epoch": 0.41660292282298,
-      "grad_norm": 0.001429154654033482,
-      "learning_rate": 0.000199999915013158,
-      "loss": 46.0,
-      "step": 2587
-    },
-    {
-      "epoch": 0.4167639599017674,
-      "grad_norm": 0.0011013075709342957,
-      "learning_rate": 0.00019999991494718726,
-      "loss": 46.0,
-      "step": 2588
-    },
-    {
-      "epoch": 0.4169249969805548,
-      "grad_norm": 0.0007707034819759429,
-      "learning_rate": 0.0001999999148811909,
-      "loss": 46.0,
-      "step": 2589
-    },
-    {
-      "epoch": 0.41708603405934214,
-      "grad_norm": 0.0004365165368653834,
-      "learning_rate": 0.00019999991481516895,
-      "loss": 46.0,
-      "step": 2590
-    },
-    {
-      "epoch": 0.41724707113812953,
-      "grad_norm": 0.002883465960621834,
-      "learning_rate": 0.00019999991474912141,
-      "loss": 46.0,
-      "step": 2591
-    },
-    {
-      "epoch": 0.41740810821691693,
-      "grad_norm": 0.000611049123108387,
-      "learning_rate": 0.0001999999146830483,
-      "loss": 46.0,
-      "step": 2592
-    },
-    {
-      "epoch": 0.4175691452957043,
-      "grad_norm": 0.0003664041869342327,
-      "learning_rate": 0.00019999991461694955,
-      "loss": 46.0,
-      "step": 2593
-    },
-    {
-      "epoch": 0.4177301823744917,
-      "grad_norm": 0.0004098225326742977,
-      "learning_rate": 0.0001999999145508252,
-      "loss": 46.0,
-      "step": 2594
-    },
-    {
-      "epoch": 0.4178912194532791,
-      "grad_norm": 0.0005654104752466083,
-      "learning_rate": 0.0001999999144846753,
-      "loss": 46.0,
-      "step": 2595
-    },
-    {
-      "epoch": 0.4180522565320665,
-      "grad_norm": 0.0005321140633895993,
-      "learning_rate": 0.00019999991441849976,
-      "loss": 46.0,
-      "step": 2596
-    },
-    {
-      "epoch": 0.4182132936108539,
-      "grad_norm": 0.0006334095960482955,
-      "learning_rate": 0.00019999991435229868,
-      "loss": 46.0,
-      "step": 2597
-    },
-    {
-      "epoch": 0.4183743306896413,
-      "grad_norm": 0.0011608528438955545,
-      "learning_rate": 0.00019999991428607198,
-      "loss": 46.0,
-      "step": 2598
-    },
-    {
-      "epoch": 0.4185353677684287,
-      "grad_norm": 0.0007042776560410857,
-      "learning_rate": 0.00019999991421981966,
-      "loss": 46.0,
-      "step": 2599
-    },
-    {
-      "epoch": 0.4186964048472161,
-      "grad_norm": 0.0010931257857009768,
-      "learning_rate": 0.00019999991415354176,
-      "loss": 46.0,
-      "step": 2600
-    },
-    {
-      "epoch": 0.41885744192600344,
-      "grad_norm": 0.0002743923105299473,
-      "learning_rate": 0.0001999999140872383,
-      "loss": 46.0,
-      "step": 2601
-    },
-    {
-      "epoch": 0.41901847900479083,
-      "grad_norm": 0.0011098834220319986,
-      "learning_rate": 0.0001999999140209092,
-      "loss": 46.0,
-      "step": 2602
-    },
-    {
-      "epoch": 0.41917951608357823,
-      "grad_norm": 0.0004687729524448514,
-      "learning_rate": 0.00019999991395455453,
-      "loss": 46.0,
-      "step": 2603
-    },
-    {
-      "epoch": 0.4193405531623656,
-      "grad_norm": 0.000279287516605109,
-      "learning_rate": 0.00019999991388817425,
-      "loss": 46.0,
-      "step": 2604
-    },
-    {
-      "epoch": 0.419501590241153,
-      "grad_norm": 0.0020967002492398024,
-      "learning_rate": 0.00019999991382176839,
-      "loss": 46.0,
-      "step": 2605
-    },
-    {
-      "epoch": 0.4196626273199404,
-      "grad_norm": 0.0010082245571538806,
-      "learning_rate": 0.0001999999137553369,
-      "loss": 46.0,
-      "step": 2606
-    },
-    {
-      "epoch": 0.4198236643987278,
-      "grad_norm": 0.0008503978606313467,
-      "learning_rate": 0.00019999991368887984,
-      "loss": 46.0,
-      "step": 2607
-    },
-    {
-      "epoch": 0.4199847014775152,
-      "grad_norm": 0.0002749088453128934,
-      "learning_rate": 0.0001999999136223972,
-      "loss": 46.0,
-      "step": 2608
-    },
-    {
-      "epoch": 0.4201457385563026,
-      "grad_norm": 0.000738668255507946,
-      "learning_rate": 0.00019999991355588894,
-      "loss": 46.0,
-      "step": 2609
-    },
-    {
-      "epoch": 0.42030677563509,
-      "grad_norm": 0.00029768518288619816,
-      "learning_rate": 0.0001999999134893551,
-      "loss": 46.0,
-      "step": 2610
-    },
-    {
-      "epoch": 0.4204678127138774,
-      "grad_norm": 0.0007780184969305992,
-      "learning_rate": 0.00019999991342279567,
-      "loss": 46.0,
-      "step": 2611
-    },
-    {
-      "epoch": 0.42062884979266474,
-      "grad_norm": 0.0016570388106629252,
-      "learning_rate": 0.00019999991335621064,
-      "loss": 46.0,
-      "step": 2612
-    },
-    {
-      "epoch": 0.42078988687145213,
-      "grad_norm": 0.0006330210599116981,
-      "learning_rate": 0.00019999991328960003,
-      "loss": 46.0,
-      "step": 2613
-    },
-    {
-      "epoch": 0.42095092395023953,
-      "grad_norm": 0.0019042609492316842,
-      "learning_rate": 0.0001999999132229638,
-      "loss": 46.0,
-      "step": 2614
-    },
-    {
-      "epoch": 0.4211119610290269,
-      "grad_norm": 0.0017024396220222116,
-      "learning_rate": 0.00019999991315630198,
-      "loss": 46.0,
-      "step": 2615
-    },
-    {
-      "epoch": 0.4212729981078143,
-      "grad_norm": 0.0008206578204408288,
-      "learning_rate": 0.00019999991308961457,
-      "loss": 46.0,
-      "step": 2616
-    },
-    {
-      "epoch": 0.4214340351866017,
-      "grad_norm": 0.00285553396679461,
-      "learning_rate": 0.00019999991302290155,
-      "loss": 46.0,
-      "step": 2617
-    },
-    {
-      "epoch": 0.4215950722653891,
-      "grad_norm": 0.0012596884043887258,
-      "learning_rate": 0.00019999991295616295,
-      "loss": 46.0,
-      "step": 2618
-    },
-    {
-      "epoch": 0.4217561093441765,
-      "grad_norm": 0.0003684552211780101,
-      "learning_rate": 0.00019999991288939875,
-      "loss": 46.0,
-      "step": 2619
-    },
-    {
-      "epoch": 0.4219171464229639,
-      "grad_norm": 0.0010181218385696411,
-      "learning_rate": 0.00019999991282260897,
-      "loss": 46.0,
-      "step": 2620
-    },
-    {
-      "epoch": 0.4220781835017513,
-      "grad_norm": 0.0006476754206232727,
-      "learning_rate": 0.00019999991275579358,
-      "loss": 46.0,
-      "step": 2621
-    },
-    {
-      "epoch": 0.42223922058053864,
-      "grad_norm": 0.0003088137018494308,
-      "learning_rate": 0.0001999999126889526,
-      "loss": 46.0,
-      "step": 2622
-    },
-    {
-      "epoch": 0.42240025765932604,
-      "grad_norm": 0.0006333135534077883,
-      "learning_rate": 0.00019999991262208605,
-      "loss": 46.0,
-      "step": 2623
-    },
-    {
-      "epoch": 0.42256129473811344,
-      "grad_norm": 0.0005531442002393305,
-      "learning_rate": 0.00019999991255519387,
-      "loss": 46.0,
-      "step": 2624
-    },
-    {
-      "epoch": 0.42272233181690083,
-      "grad_norm": 0.0017164757009595633,
-      "learning_rate": 0.0001999999124882761,
-      "loss": 46.0,
-      "step": 2625
-    },
-    {
-      "epoch": 0.4228833688956882,
-      "grad_norm": 0.000631219707429409,
-      "learning_rate": 0.0001999999124213327,
-      "loss": 46.0,
-      "step": 2626
-    },
-    {
-      "epoch": 0.4230444059744756,
-      "grad_norm": 0.0006477307761088014,
-      "learning_rate": 0.00019999991235436376,
-      "loss": 46.0,
-      "step": 2627
-    },
-    {
-      "epoch": 0.423205443053263,
-      "grad_norm": 0.0008212728425860405,
-      "learning_rate": 0.0001999999122873692,
-      "loss": 46.0,
-      "step": 2628
-    },
-    {
-      "epoch": 0.4233664801320504,
-      "grad_norm": 0.001346821547485888,
-      "learning_rate": 0.00019999991222034908,
-      "loss": 46.0,
-      "step": 2629
-    },
-    {
-      "epoch": 0.4235275172108378,
-      "grad_norm": 0.0004094088508281857,
-      "learning_rate": 0.00019999991215330335,
-      "loss": 46.0,
-      "step": 2630
-    },
-    {
-      "epoch": 0.4236885542896252,
-      "grad_norm": 0.00031145557295531034,
-      "learning_rate": 0.000199999912086232,
-      "loss": 46.0,
-      "step": 2631
-    },
-    {
-      "epoch": 0.4238495913684126,
-      "grad_norm": 0.0009608006803318858,
-      "learning_rate": 0.00019999991201913507,
-      "loss": 46.0,
-      "step": 2632
-    },
-    {
-      "epoch": 0.42401062844719994,
-      "grad_norm": 0.0003583583456929773,
-      "learning_rate": 0.00019999991195201254,
-      "loss": 46.0,
-      "step": 2633
-    },
-    {
-      "epoch": 0.42417166552598734,
-      "grad_norm": 0.0005778919439762831,
-      "learning_rate": 0.00019999991188486443,
-      "loss": 46.0,
-      "step": 2634
-    },
-    {
-      "epoch": 0.42433270260477474,
-      "grad_norm": 0.0006798491813242435,
-      "learning_rate": 0.0001999999118176907,
-      "loss": 46.0,
-      "step": 2635
-    },
-    {
-      "epoch": 0.42449373968356213,
-      "grad_norm": 0.0005650821258313954,
-      "learning_rate": 0.00019999991175049142,
-      "loss": 46.0,
-      "step": 2636
-    },
-    {
-      "epoch": 0.42465477676234953,
-      "grad_norm": 0.001622366369701922,
-      "learning_rate": 0.0001999999116832665,
-      "loss": 46.0,
-      "step": 2637
-    },
-    {
-      "epoch": 0.4248158138411369,
-      "grad_norm": 0.0005355989560484886,
-      "learning_rate": 0.000199999911616016,
-      "loss": 46.0,
-      "step": 2638
-    },
-    {
-      "epoch": 0.4249768509199243,
-      "grad_norm": 0.0009966815123334527,
-      "learning_rate": 0.0001999999115487399,
-      "loss": 46.0,
-      "step": 2639
-    },
-    {
-      "epoch": 0.4251378879987117,
-      "grad_norm": 0.0016135798068717122,
-      "learning_rate": 0.00019999991148143823,
-      "loss": 46.0,
-      "step": 2640
-    },
-    {
-      "epoch": 0.4252989250774991,
-      "grad_norm": 0.0004172810586169362,
-      "learning_rate": 0.00019999991141411095,
-      "loss": 46.0,
-      "step": 2641
-    },
-    {
-      "epoch": 0.4254599621562865,
-      "grad_norm": 0.00042144497274421155,
-      "learning_rate": 0.00019999991134675806,
-      "loss": 46.0,
-      "step": 2642
-    },
-    {
-      "epoch": 0.42562099923507385,
-      "grad_norm": 0.0007188401068560779,
-      "learning_rate": 0.0001999999112793796,
-      "loss": 46.0,
-      "step": 2643
-    },
-    {
-      "epoch": 0.42578203631386125,
-      "grad_norm": 0.0018502577440813184,
-      "learning_rate": 0.00019999991121197552,
-      "loss": 46.0,
-      "step": 2644
-    },
-    {
-      "epoch": 0.42594307339264864,
-      "grad_norm": 0.0006879601860418916,
-      "learning_rate": 0.00019999991114454585,
-      "loss": 46.0,
-      "step": 2645
-    },
-    {
-      "epoch": 0.42610411047143604,
-      "grad_norm": 0.0016043668147176504,
-      "learning_rate": 0.0001999999110770906,
-      "loss": 46.0,
-      "step": 2646
-    },
-    {
-      "epoch": 0.42626514755022343,
-      "grad_norm": 0.0005683312192559242,
-      "learning_rate": 0.00019999991100960976,
-      "loss": 46.0,
-      "step": 2647
-    },
-    {
-      "epoch": 0.42642618462901083,
-      "grad_norm": 0.0005000039236620069,
-      "learning_rate": 0.00019999991094210332,
-      "loss": 46.0,
-      "step": 2648
-    },
-    {
-      "epoch": 0.4265872217077982,
-      "grad_norm": 0.00046185965766198933,
-      "learning_rate": 0.00019999991087457127,
-      "loss": 46.0,
-      "step": 2649
-    },
-    {
-      "epoch": 0.4267482587865856,
-      "grad_norm": 0.00037300106487236917,
-      "learning_rate": 0.00019999991080701363,
-      "loss": 46.0,
-      "step": 2650
-    },
-    {
-      "epoch": 0.426909295865373,
-      "grad_norm": 0.0010627803858369589,
-      "learning_rate": 0.0001999999107394304,
-      "loss": 46.0,
-      "step": 2651
-    },
-    {
-      "epoch": 0.4270703329441604,
-      "grad_norm": 0.0007060438510961831,
-      "learning_rate": 0.00019999991067182156,
-      "loss": 46.0,
-      "step": 2652
-    },
-    {
-      "epoch": 0.4272313700229478,
-      "grad_norm": 0.001122227986343205,
-      "learning_rate": 0.00019999991060418715,
-      "loss": 46.0,
-      "step": 2653
-    },
-    {
-      "epoch": 0.42739240710173515,
-      "grad_norm": 0.0006794400978833437,
-      "learning_rate": 0.0001999999105365271,
-      "loss": 46.0,
-      "step": 2654
-    },
-    {
-      "epoch": 0.42755344418052255,
-      "grad_norm": 0.0012592078419402242,
-      "learning_rate": 0.00019999991046884154,
-      "loss": 46.0,
-      "step": 2655
-    },
-    {
-      "epoch": 0.42771448125930994,
-      "grad_norm": 0.0008367216214537621,
-      "learning_rate": 0.00019999991040113032,
-      "loss": 46.0,
-      "step": 2656
-    },
-    {
-      "epoch": 0.42787551833809734,
-      "grad_norm": 0.00032703651231713593,
-      "learning_rate": 0.0001999999103333935,
-      "loss": 46.0,
-      "step": 2657
-    },
-    {
-      "epoch": 0.42803655541688473,
-      "grad_norm": 0.0006271525053307414,
-      "learning_rate": 0.0001999999102656311,
-      "loss": 46.0,
-      "step": 2658
-    },
-    {
-      "epoch": 0.42819759249567213,
-      "grad_norm": 0.000993541325442493,
-      "learning_rate": 0.00019999991019784312,
-      "loss": 46.0,
-      "step": 2659
-    },
-    {
-      "epoch": 0.4283586295744595,
-      "grad_norm": 0.000801052141468972,
-      "learning_rate": 0.00019999991013002952,
-      "loss": 46.0,
-      "step": 2660
-    },
-    {
-      "epoch": 0.4285196666532469,
-      "grad_norm": 0.0006492906832136214,
-      "learning_rate": 0.00019999991006219037,
-      "loss": 46.0,
-      "step": 2661
-    },
-    {
-      "epoch": 0.4286807037320343,
-      "grad_norm": 0.0004494022286962718,
-      "learning_rate": 0.00019999990999432557,
-      "loss": 46.0,
-      "step": 2662
-    },
-    {
-      "epoch": 0.4288417408108217,
-      "grad_norm": 0.0007963523967191577,
-      "learning_rate": 0.00019999990992643522,
-      "loss": 46.0,
-      "step": 2663
-    },
-    {
-      "epoch": 0.42900277788960905,
-      "grad_norm": 0.0004384623025543988,
-      "learning_rate": 0.00019999990985851925,
-      "loss": 46.0,
-      "step": 2664
-    },
-    {
-      "epoch": 0.42916381496839645,
-      "grad_norm": 0.001220785896293819,
-      "learning_rate": 0.00019999990979057767,
-      "loss": 46.0,
-      "step": 2665
-    },
-    {
-      "epoch": 0.42932485204718385,
-      "grad_norm": 0.0003058039874304086,
-      "learning_rate": 0.00019999990972261052,
-      "loss": 46.0,
-      "step": 2666
-    },
-    {
-      "epoch": 0.42948588912597124,
-      "grad_norm": 0.00038542156107723713,
-      "learning_rate": 0.0001999999096546178,
-      "loss": 46.0,
-      "step": 2667
-    },
-    {
-      "epoch": 0.42964692620475864,
-      "grad_norm": 0.0006260395748540759,
-      "learning_rate": 0.00019999990958659945,
-      "loss": 46.0,
-      "step": 2668
-    },
-    {
-      "epoch": 0.42980796328354604,
-      "grad_norm": 0.001564955455251038,
-      "learning_rate": 0.0001999999095185555,
-      "loss": 46.0,
-      "step": 2669
-    },
-    {
-      "epoch": 0.42996900036233343,
-      "grad_norm": 0.0005347688565962017,
-      "learning_rate": 0.00019999990945048597,
-      "loss": 46.0,
-      "step": 2670
-    },
-    {
-      "epoch": 0.4301300374411208,
-      "grad_norm": 0.0005981026333756745,
-      "learning_rate": 0.00019999990938239084,
-      "loss": 46.0,
-      "step": 2671
-    },
-    {
-      "epoch": 0.4302910745199082,
-      "grad_norm": 0.0013980268267914653,
-      "learning_rate": 0.00019999990931427012,
-      "loss": 46.0,
-      "step": 2672
-    },
-    {
-      "epoch": 0.4304521115986956,
-      "grad_norm": 0.0004649643669836223,
-      "learning_rate": 0.0001999999092461238,
-      "loss": 46.0,
-      "step": 2673
-    },
-    {
-      "epoch": 0.430613148677483,
-      "grad_norm": 0.00039003134588710964,
-      "learning_rate": 0.00019999990917795189,
-      "loss": 46.0,
-      "step": 2674
-    },
-    {
-      "epoch": 0.43077418575627036,
-      "grad_norm": 0.0007489864365197718,
-      "learning_rate": 0.00019999990910975438,
-      "loss": 46.0,
-      "step": 2675
-    },
-    {
-      "epoch": 0.43093522283505775,
-      "grad_norm": 0.000439871393609792,
-      "learning_rate": 0.00019999990904153128,
-      "loss": 46.0,
-      "step": 2676
-    },
-    {
-      "epoch": 0.43109625991384515,
-      "grad_norm": 0.0006495008710771799,
-      "learning_rate": 0.00019999990897328257,
-      "loss": 46.0,
-      "step": 2677
-    },
-    {
-      "epoch": 0.43125729699263254,
-      "grad_norm": 0.0021862033754587173,
-      "learning_rate": 0.00019999990890500827,
-      "loss": 46.0,
-      "step": 2678
-    },
-    {
-      "epoch": 0.43141833407141994,
-      "grad_norm": 0.00034401609445922077,
-      "learning_rate": 0.00019999990883670839,
-      "loss": 46.0,
-      "step": 2679
-    },
-    {
-      "epoch": 0.43157937115020734,
-      "grad_norm": 0.00048437921213917434,
-      "learning_rate": 0.00019999990876838291,
-      "loss": 46.0,
-      "step": 2680
-    },
-    {
-      "epoch": 0.43174040822899473,
-      "grad_norm": 0.0007042984361760318,
-      "learning_rate": 0.00019999990870003185,
-      "loss": 46.0,
-      "step": 2681
-    },
-    {
-      "epoch": 0.43190144530778213,
-      "grad_norm": 0.0007430159021168947,
-      "learning_rate": 0.00019999990863165515,
-      "loss": 46.0,
-      "step": 2682
-    },
-    {
-      "epoch": 0.4320624823865695,
-      "grad_norm": 0.0010032978607341647,
-      "learning_rate": 0.0001999999085632529,
-      "loss": 46.0,
-      "step": 2683
-    },
-    {
-      "epoch": 0.4322235194653569,
-      "grad_norm": 0.00039155816193670034,
-      "learning_rate": 0.00019999990849482504,
-      "loss": 46.0,
-      "step": 2684
-    },
-    {
-      "epoch": 0.4323845565441443,
-      "grad_norm": 0.0009293517796322703,
-      "learning_rate": 0.0001999999084263716,
-      "loss": 46.0,
-      "step": 2685
-    },
-    {
-      "epoch": 0.43254559362293166,
-      "grad_norm": 0.00036867070593871176,
-      "learning_rate": 0.00019999990835789253,
-      "loss": 46.0,
-      "step": 2686
-    },
-    {
-      "epoch": 0.43270663070171905,
-      "grad_norm": 0.001242671045474708,
-      "learning_rate": 0.0001999999082893879,
-      "loss": 46.0,
-      "step": 2687
-    },
-    {
-      "epoch": 0.43286766778050645,
-      "grad_norm": 0.000723321340046823,
-      "learning_rate": 0.00019999990822085765,
-      "loss": 46.0,
-      "step": 2688
-    },
-    {
-      "epoch": 0.43302870485929384,
-      "grad_norm": 0.0004949741996824741,
-      "learning_rate": 0.0001999999081523018,
-      "loss": 46.0,
-      "step": 2689
-    },
-    {
-      "epoch": 0.43318974193808124,
-      "grad_norm": 0.0011749679688364267,
-      "learning_rate": 0.00019999990808372038,
-      "loss": 46.0,
-      "step": 2690
-    },
-    {
-      "epoch": 0.43335077901686864,
-      "grad_norm": 0.0015893633244559169,
-      "learning_rate": 0.00019999990801511334,
-      "loss": 46.0,
-      "step": 2691
-    },
-    {
-      "epoch": 0.43351181609565603,
-      "grad_norm": 0.0004216721863485873,
-      "learning_rate": 0.00019999990794648074,
-      "loss": 46.0,
-      "step": 2692
-    },
-    {
-      "epoch": 0.43367285317444343,
-      "grad_norm": 0.00045768602285534143,
-      "learning_rate": 0.00019999990787782253,
-      "loss": 46.0,
-      "step": 2693
-    },
-    {
-      "epoch": 0.4338338902532308,
-      "grad_norm": 0.0012356709921732545,
-      "learning_rate": 0.0001999999078091387,
-      "loss": 46.0,
-      "step": 2694
-    },
-    {
-      "epoch": 0.4339949273320182,
-      "grad_norm": 0.0006851233774796128,
-      "learning_rate": 0.0001999999077404293,
-      "loss": 46.0,
-      "step": 2695
-    },
-    {
-      "epoch": 0.43415596441080556,
-      "grad_norm": 0.000262343673966825,
-      "learning_rate": 0.0001999999076716943,
-      "loss": 46.0,
-      "step": 2696
-    },
-    {
-      "epoch": 0.43431700148959296,
-      "grad_norm": 0.0004097023338545114,
-      "learning_rate": 0.0001999999076029337,
-      "loss": 46.0,
-      "step": 2697
-    },
-    {
-      "epoch": 0.43447803856838035,
-      "grad_norm": 0.000551571836695075,
-      "learning_rate": 0.00019999990753414752,
-      "loss": 46.0,
-      "step": 2698
-    },
-    {
-      "epoch": 0.43463907564716775,
-      "grad_norm": 0.0016339104622602463,
-      "learning_rate": 0.00019999990746533573,
-      "loss": 46.0,
-      "step": 2699
-    },
-    {
-      "epoch": 0.43480011272595515,
-      "grad_norm": 0.00032277987338602543,
-      "learning_rate": 0.00019999990739649835,
-      "loss": 46.0,
-      "step": 2700
-    },
-    {
-      "epoch": 0.43496114980474254,
-      "grad_norm": 0.0004354232514742762,
-      "learning_rate": 0.00019999990732763538,
-      "loss": 46.0,
-      "step": 2701
-    },
-    {
-      "epoch": 0.43512218688352994,
-      "grad_norm": 0.0006771563785150647,
-      "learning_rate": 0.0001999999072587468,
-      "loss": 46.0,
-      "step": 2702
-    },
-    {
-      "epoch": 0.43528322396231733,
-      "grad_norm": 0.0003972511040046811,
-      "learning_rate": 0.00019999990718983263,
-      "loss": 46.0,
-      "step": 2703
-    },
-    {
-      "epoch": 0.43544426104110473,
-      "grad_norm": 0.000676590483635664,
-      "learning_rate": 0.00019999990712089288,
-      "loss": 46.0,
-      "step": 2704
-    },
-    {
-      "epoch": 0.4356052981198921,
-      "grad_norm": 0.0007195529760792851,
-      "learning_rate": 0.00019999990705192754,
-      "loss": 46.0,
-      "step": 2705
-    },
-    {
-      "epoch": 0.4357663351986795,
-      "grad_norm": 0.00038460109499283135,
-      "learning_rate": 0.00019999990698293658,
-      "loss": 46.0,
-      "step": 2706
-    },
-    {
-      "epoch": 0.43592737227746686,
-      "grad_norm": 0.0006472237873822451,
-      "learning_rate": 0.00019999990691392004,
-      "loss": 46.0,
-      "step": 2707
-    },
-    {
-      "epoch": 0.43608840935625426,
-      "grad_norm": 0.0003614487068261951,
-      "learning_rate": 0.0001999999068448779,
-      "loss": 46.0,
-      "step": 2708
-    },
-    {
-      "epoch": 0.43624944643504165,
-      "grad_norm": 0.0008919125539250672,
-      "learning_rate": 0.00019999990677581017,
-      "loss": 46.0,
-      "step": 2709
-    },
-    {
-      "epoch": 0.43641048351382905,
-      "grad_norm": 0.0006342657143250108,
-      "learning_rate": 0.00019999990670671684,
-      "loss": 46.0,
-      "step": 2710
-    },
-    {
-      "epoch": 0.43657152059261645,
-      "grad_norm": 0.0006811715429648757,
-      "learning_rate": 0.00019999990663759794,
-      "loss": 46.0,
-      "step": 2711
-    },
-    {
-      "epoch": 0.43673255767140384,
-      "grad_norm": 0.002743551740422845,
-      "learning_rate": 0.0001999999065684534,
-      "loss": 46.0,
-      "step": 2712
-    },
-    {
-      "epoch": 0.43689359475019124,
-      "grad_norm": 0.00041297305142506957,
-      "learning_rate": 0.0001999999064992833,
-      "loss": 46.0,
-      "step": 2713
-    },
-    {
-      "epoch": 0.43705463182897863,
-      "grad_norm": 0.0009428643388673663,
-      "learning_rate": 0.00019999990643008758,
-      "loss": 46.0,
-      "step": 2714
-    },
-    {
-      "epoch": 0.43721566890776603,
-      "grad_norm": 0.001723370049148798,
-      "learning_rate": 0.00019999990636086626,
-      "loss": 46.0,
-      "step": 2715
-    },
-    {
-      "epoch": 0.4373767059865534,
-      "grad_norm": 0.0008770760032348335,
-      "learning_rate": 0.0001999999062916194,
-      "loss": 46.0,
-      "step": 2716
-    },
-    {
-      "epoch": 0.43753774306534077,
-      "grad_norm": 0.000561458757147193,
-      "learning_rate": 0.0001999999062223469,
-      "loss": 46.0,
-      "step": 2717
-    },
-    {
-      "epoch": 0.43769878014412816,
-      "grad_norm": 0.0038393663708120584,
-      "learning_rate": 0.00019999990615304883,
-      "loss": 46.0,
-      "step": 2718
-    },
-    {
-      "epoch": 0.43785981722291556,
-      "grad_norm": 0.0010003532515838742,
-      "learning_rate": 0.0001999999060837251,
-      "loss": 46.0,
-      "step": 2719
-    },
-    {
-      "epoch": 0.43802085430170296,
-      "grad_norm": 0.0006829585181549191,
-      "learning_rate": 0.00019999990601437585,
-      "loss": 46.0,
-      "step": 2720
-    },
-    {
-      "epoch": 0.43818189138049035,
-      "grad_norm": 0.000540735840331763,
-      "learning_rate": 0.00019999990594500098,
-      "loss": 46.0,
-      "step": 2721
-    },
-    {
-      "epoch": 0.43834292845927775,
-      "grad_norm": 0.0006235248292796314,
-      "learning_rate": 0.00019999990587560055,
-      "loss": 46.0,
-      "step": 2722
-    },
-    {
-      "epoch": 0.43850396553806514,
-      "grad_norm": 0.0007193930214270949,
-      "learning_rate": 0.00019999990580617447,
-      "loss": 46.0,
-      "step": 2723
-    },
-    {
-      "epoch": 0.43866500261685254,
-      "grad_norm": 0.0008322568028233945,
-      "learning_rate": 0.0001999999057367228,
-      "loss": 46.0,
-      "step": 2724
-    },
-    {
-      "epoch": 0.43882603969563994,
-      "grad_norm": 0.0005510931368917227,
-      "learning_rate": 0.00019999990566724557,
-      "loss": 46.0,
-      "step": 2725
-    },
-    {
-      "epoch": 0.43898707677442733,
-      "grad_norm": 0.0006119512254372239,
-      "learning_rate": 0.0001999999055977427,
-      "loss": 46.0,
-      "step": 2726
-    },
-    {
-      "epoch": 0.4391481138532147,
-      "grad_norm": 0.000798740133177489,
-      "learning_rate": 0.00019999990552821428,
-      "loss": 46.0,
-      "step": 2727
-    },
-    {
-      "epoch": 0.43930915093200207,
-      "grad_norm": 0.0006452187662944198,
-      "learning_rate": 0.00019999990545866022,
-      "loss": 46.0,
-      "step": 2728
-    },
-    {
-      "epoch": 0.43947018801078946,
-      "grad_norm": 0.0004959151847288013,
-      "learning_rate": 0.0001999999053890806,
-      "loss": 46.0,
-      "step": 2729
-    },
-    {
-      "epoch": 0.43963122508957686,
-      "grad_norm": 0.0007036104216240346,
-      "learning_rate": 0.00019999990531947539,
-      "loss": 46.0,
-      "step": 2730
-    },
-    {
-      "epoch": 0.43979226216836426,
-      "grad_norm": 0.00037318517570383847,
-      "learning_rate": 0.00019999990524984456,
-      "loss": 46.0,
-      "step": 2731
-    },
-    {
-      "epoch": 0.43995329924715165,
-      "grad_norm": 0.0006354129291139543,
-      "learning_rate": 0.00019999990518018815,
-      "loss": 46.0,
-      "step": 2732
-    },
-    {
-      "epoch": 0.44011433632593905,
-      "grad_norm": 0.00037247739965096116,
-      "learning_rate": 0.00019999990511050615,
-      "loss": 46.0,
-      "step": 2733
-    },
-    {
-      "epoch": 0.44027537340472644,
-      "grad_norm": 0.002498733112588525,
-      "learning_rate": 0.00019999990504079854,
-      "loss": 46.0,
-      "step": 2734
-    },
-    {
-      "epoch": 0.44043641048351384,
-      "grad_norm": 0.0013193635968491435,
-      "learning_rate": 0.00019999990497106534,
-      "loss": 46.0,
-      "step": 2735
-    },
-    {
-      "epoch": 0.44059744756230124,
-      "grad_norm": 0.0005198375438340008,
-      "learning_rate": 0.00019999990490130655,
-      "loss": 46.0,
-      "step": 2736
-    },
-    {
-      "epoch": 0.44075848464108863,
-      "grad_norm": 0.0006304277339950204,
-      "learning_rate": 0.00019999990483152215,
-      "loss": 46.0,
-      "step": 2737
-    },
-    {
-      "epoch": 0.44091952171987603,
-      "grad_norm": 0.0007082073716446757,
-      "learning_rate": 0.00019999990476171219,
-      "loss": 46.0,
-      "step": 2738
-    },
-    {
-      "epoch": 0.44108055879866337,
-      "grad_norm": 0.00026121517294086516,
-      "learning_rate": 0.00019999990469187658,
-      "loss": 46.0,
-      "step": 2739
-    },
-    {
-      "epoch": 0.44124159587745077,
-      "grad_norm": 0.00047844467917457223,
-      "learning_rate": 0.00019999990462201542,
-      "loss": 46.0,
-      "step": 2740
-    },
-    {
-      "epoch": 0.44140263295623816,
-      "grad_norm": 0.0006873977254144847,
-      "learning_rate": 0.00019999990455212864,
-      "loss": 46.0,
-      "step": 2741
-    },
-    {
-      "epoch": 0.44156367003502556,
-      "grad_norm": 0.0006066896021366119,
-      "learning_rate": 0.00019999990448221628,
-      "loss": 46.0,
-      "step": 2742
-    },
-    {
-      "epoch": 0.44172470711381295,
-      "grad_norm": 0.0006010742508806288,
-      "learning_rate": 0.00019999990441227833,
-      "loss": 46.0,
-      "step": 2743
-    },
-    {
-      "epoch": 0.44188574419260035,
-      "grad_norm": 0.0017767059616744518,
-      "learning_rate": 0.00019999990434231476,
-      "loss": 46.0,
-      "step": 2744
-    },
-    {
-      "epoch": 0.44204678127138775,
-      "grad_norm": 0.0010924983071163297,
-      "learning_rate": 0.00019999990427232563,
-      "loss": 46.0,
-      "step": 2745
-    },
-    {
-      "epoch": 0.44220781835017514,
-      "grad_norm": 0.0006658118800260127,
-      "learning_rate": 0.0001999999042023109,
-      "loss": 46.0,
-      "step": 2746
-    },
-    {
-      "epoch": 0.44236885542896254,
-      "grad_norm": 0.002850266871973872,
-      "learning_rate": 0.00019999990413227056,
-      "loss": 46.0,
-      "step": 2747
-    },
-    {
-      "epoch": 0.44252989250774993,
-      "grad_norm": 0.000484477641293779,
-      "learning_rate": 0.00019999990406220462,
-      "loss": 46.0,
-      "step": 2748
-    },
-    {
-      "epoch": 0.4426909295865373,
-      "grad_norm": 0.0004909904673695564,
-      "learning_rate": 0.0001999999039921131,
-      "loss": 46.0,
-      "step": 2749
-    },
-    {
-      "epoch": 0.44285196666532467,
-      "grad_norm": 0.0005222524632699788,
-      "learning_rate": 0.00019999990392199598,
-      "loss": 46.0,
-      "step": 2750
-    },
-    {
-      "epoch": 0.44301300374411207,
-      "grad_norm": 0.00028756973915733397,
-      "learning_rate": 0.00019999990385185325,
-      "loss": 46.0,
-      "step": 2751
-    },
-    {
-      "epoch": 0.44317404082289946,
-      "grad_norm": 0.00036780882510356605,
-      "learning_rate": 0.00019999990378168496,
-      "loss": 46.0,
-      "step": 2752
-    },
-    {
-      "epoch": 0.44333507790168686,
-      "grad_norm": 0.004427620675414801,
-      "learning_rate": 0.00019999990371149105,
-      "loss": 46.0,
-      "step": 2753
-    },
-    {
-      "epoch": 0.44349611498047425,
-      "grad_norm": 0.0005069668404757977,
-      "learning_rate": 0.00019999990364127156,
-      "loss": 46.0,
-      "step": 2754
-    },
-    {
-      "epoch": 0.44365715205926165,
-      "grad_norm": 0.0011849005240947008,
-      "learning_rate": 0.00019999990357102645,
-      "loss": 46.0,
-      "step": 2755
-    },
-    {
-      "epoch": 0.44381818913804905,
-      "grad_norm": 0.0012834552908316255,
-      "learning_rate": 0.00019999990350075576,
-      "loss": 46.0,
-      "step": 2756
-    },
-    {
-      "epoch": 0.44397922621683644,
-      "grad_norm": 0.0005128609482198954,
-      "learning_rate": 0.00019999990343045948,
-      "loss": 46.0,
-      "step": 2757
-    },
-    {
-      "epoch": 0.44414026329562384,
-      "grad_norm": 0.0005335664027370512,
-      "learning_rate": 0.00019999990336013758,
-      "loss": 46.0,
-      "step": 2758
-    },
-    {
-      "epoch": 0.44430130037441123,
-      "grad_norm": 0.000431609369115904,
-      "learning_rate": 0.00019999990328979013,
-      "loss": 46.0,
-      "step": 2759
-    },
-    {
-      "epoch": 0.4444623374531986,
-      "grad_norm": 0.00036994690890423954,
-      "learning_rate": 0.00019999990321941703,
-      "loss": 46.0,
-      "step": 2760
-    },
-    {
-      "epoch": 0.44462337453198597,
-      "grad_norm": 0.0005365213728509843,
-      "learning_rate": 0.00019999990314901838,
-      "loss": 46.0,
-      "step": 2761
-    },
-    {
-      "epoch": 0.44478441161077337,
-      "grad_norm": 0.000801724148914218,
-      "learning_rate": 0.00019999990307859413,
-      "loss": 46.0,
-      "step": 2762
-    },
-    {
-      "epoch": 0.44494544868956076,
-      "grad_norm": 0.0003944375494029373,
-      "learning_rate": 0.00019999990300814427,
-      "loss": 46.0,
-      "step": 2763
-    },
-    {
-      "epoch": 0.44510648576834816,
-      "grad_norm": 0.00028719662805087864,
-      "learning_rate": 0.00019999990293766883,
-      "loss": 46.0,
-      "step": 2764
-    },
-    {
-      "epoch": 0.44526752284713556,
-      "grad_norm": 0.0023350659757852554,
-      "learning_rate": 0.0001999999028671678,
-      "loss": 46.0,
-      "step": 2765
-    },
-    {
-      "epoch": 0.44542855992592295,
-      "grad_norm": 0.0006715567433275282,
-      "learning_rate": 0.00019999990279664115,
-      "loss": 46.0,
-      "step": 2766
-    },
-    {
-      "epoch": 0.44558959700471035,
-      "grad_norm": 0.0004399584431666881,
-      "learning_rate": 0.00019999990272608892,
-      "loss": 46.0,
-      "step": 2767
-    },
-    {
-      "epoch": 0.44575063408349774,
-      "grad_norm": 0.0007615103386342525,
-      "learning_rate": 0.00019999990265551107,
-      "loss": 46.0,
-      "step": 2768
-    },
-    {
-      "epoch": 0.44591167116228514,
-      "grad_norm": 0.000605251407250762,
-      "learning_rate": 0.00019999990258490766,
-      "loss": 46.0,
-      "step": 2769
-    },
-    {
-      "epoch": 0.4460727082410725,
-      "grad_norm": 0.0006432253867387772,
-      "learning_rate": 0.00019999990251427864,
-      "loss": 46.0,
-      "step": 2770
-    },
-    {
-      "epoch": 0.4462337453198599,
-      "grad_norm": 0.0014349092962220311,
-      "learning_rate": 0.000199999902443624,
-      "loss": 46.0,
-      "step": 2771
-    },
-    {
-      "epoch": 0.44639478239864727,
-      "grad_norm": 0.0006780263502150774,
-      "learning_rate": 0.0001999999023729438,
-      "loss": 46.0,
-      "step": 2772
-    },
-    {
-      "epoch": 0.44655581947743467,
-      "grad_norm": 0.0009366684826090932,
-      "learning_rate": 0.000199999902302238,
-      "loss": 46.0,
-      "step": 2773
-    },
-    {
-      "epoch": 0.44671685655622206,
-      "grad_norm": 0.0011787187540903687,
-      "learning_rate": 0.00019999990223150662,
-      "loss": 46.0,
-      "step": 2774
-    },
-    {
-      "epoch": 0.44687789363500946,
-      "grad_norm": 0.000608434434980154,
-      "learning_rate": 0.0001999999021607496,
-      "loss": 46.0,
-      "step": 2775
-    },
-    {
-      "epoch": 0.44703893071379686,
-      "grad_norm": 0.000304481916828081,
-      "learning_rate": 0.00019999990208996704,
-      "loss": 46.0,
-      "step": 2776
-    },
-    {
-      "epoch": 0.44719996779258425,
-      "grad_norm": 0.0007709647761657834,
-      "learning_rate": 0.00019999990201915882,
-      "loss": 46.0,
-      "step": 2777
-    },
-    {
-      "epoch": 0.44736100487137165,
-      "grad_norm": 0.0014972627395763993,
-      "learning_rate": 0.00019999990194832508,
-      "loss": 46.0,
-      "step": 2778
-    },
-    {
-      "epoch": 0.44752204195015904,
-      "grad_norm": 0.0008488981984555721,
-      "learning_rate": 0.0001999999018774657,
-      "loss": 46.0,
-      "step": 2779
-    },
-    {
-      "epoch": 0.44768307902894644,
-      "grad_norm": 0.000668447813950479,
-      "learning_rate": 0.00019999990180658074,
-      "loss": 46.0,
-      "step": 2780
-    },
-    {
-      "epoch": 0.4478441161077338,
-      "grad_norm": 0.0007189158932305872,
-      "learning_rate": 0.00019999990173567018,
-      "loss": 46.0,
-      "step": 2781
-    },
-    {
-      "epoch": 0.4480051531865212,
-      "grad_norm": 0.0006612858851440251,
-      "learning_rate": 0.000199999901664734,
-      "loss": 46.0,
-      "step": 2782
-    },
-    {
-      "epoch": 0.4481661902653086,
-      "grad_norm": 0.0015983371995389462,
-      "learning_rate": 0.00019999990159377226,
-      "loss": 46.0,
-      "step": 2783
-    },
-    {
-      "epoch": 0.44832722734409597,
-      "grad_norm": 0.00026204640744253993,
-      "learning_rate": 0.0001999999015227849,
-      "loss": 46.0,
-      "step": 2784
-    },
-    {
-      "epoch": 0.44848826442288336,
-      "grad_norm": 0.0010673885699361563,
-      "learning_rate": 0.000199999901451772,
-      "loss": 46.0,
-      "step": 2785
-    },
-    {
-      "epoch": 0.44864930150167076,
-      "grad_norm": 0.0005019505624659359,
-      "learning_rate": 0.00019999990138073342,
-      "loss": 46.0,
-      "step": 2786
-    },
-    {
-      "epoch": 0.44881033858045816,
-      "grad_norm": 0.0012449523201212287,
-      "learning_rate": 0.0001999999013096693,
-      "loss": 46.0,
-      "step": 2787
-    },
-    {
-      "epoch": 0.44897137565924555,
-      "grad_norm": 0.00023666986089665443,
-      "learning_rate": 0.00019999990123857958,
-      "loss": 46.0,
-      "step": 2788
-    },
-    {
-      "epoch": 0.44913241273803295,
-      "grad_norm": 0.0006680472288280725,
-      "learning_rate": 0.00019999990116746427,
-      "loss": 46.0,
-      "step": 2789
-    },
-    {
-      "epoch": 0.44929344981682034,
-      "grad_norm": 0.0005869470769539475,
-      "learning_rate": 0.00019999990109632334,
-      "loss": 46.0,
-      "step": 2790
-    },
-    {
-      "epoch": 0.44945448689560774,
-      "grad_norm": 0.000649125431664288,
-      "learning_rate": 0.00019999990102515685,
-      "loss": 46.0,
-      "step": 2791
-    },
-    {
-      "epoch": 0.4496155239743951,
-      "grad_norm": 0.0003559860051609576,
-      "learning_rate": 0.00019999990095396472,
-      "loss": 46.0,
-      "step": 2792
-    },
-    {
-      "epoch": 0.4497765610531825,
-      "grad_norm": 0.0005704216309823096,
-      "learning_rate": 0.00019999990088274703,
-      "loss": 46.0,
-      "step": 2793
-    },
-    {
-      "epoch": 0.4499375981319699,
-      "grad_norm": 0.000314653676468879,
-      "learning_rate": 0.00019999990081150375,
-      "loss": 46.0,
-      "step": 2794
-    },
-    {
-      "epoch": 0.45009863521075727,
-      "grad_norm": 0.0011541357962414622,
-      "learning_rate": 0.00019999990074023483,
-      "loss": 46.0,
-      "step": 2795
-    },
-    {
-      "epoch": 0.45025967228954467,
-      "grad_norm": 0.0017858966020867229,
-      "learning_rate": 0.00019999990066894036,
-      "loss": 46.0,
-      "step": 2796
-    },
-    {
-      "epoch": 0.45042070936833206,
-      "grad_norm": 0.000754722161218524,
-      "learning_rate": 0.00019999990059762032,
-      "loss": 46.0,
-      "step": 2797
-    },
-    {
-      "epoch": 0.45058174644711946,
-      "grad_norm": 0.0005892252083867788,
-      "learning_rate": 0.0001999999005262746,
-      "loss": 46.0,
-      "step": 2798
-    },
-    {
-      "epoch": 0.45074278352590685,
-      "grad_norm": 0.00027911565848626196,
-      "learning_rate": 0.00019999990045490334,
-      "loss": 46.0,
-      "step": 2799
-    },
-    {
-      "epoch": 0.45090382060469425,
-      "grad_norm": 0.0007249664631672204,
-      "learning_rate": 0.00019999990038350646,
-      "loss": 46.0,
-      "step": 2800
-    },
-    {
-      "epoch": 0.45106485768348165,
-      "grad_norm": 0.0006351073388941586,
-      "learning_rate": 0.00019999990031208402,
-      "loss": 46.0,
-      "step": 2801
-    },
-    {
-      "epoch": 0.451225894762269,
-      "grad_norm": 0.000732273212634027,
-      "learning_rate": 0.00019999990024063596,
-      "loss": 46.0,
-      "step": 2802
-    },
-    {
-      "epoch": 0.4513869318410564,
-      "grad_norm": 0.00045403256081044674,
-      "learning_rate": 0.00019999990016916232,
-      "loss": 46.0,
-      "step": 2803
-    },
-    {
-      "epoch": 0.4515479689198438,
-      "grad_norm": 0.00037547352258116007,
-      "learning_rate": 0.00019999990009766306,
-      "loss": 46.0,
-      "step": 2804
-    },
-    {
-      "epoch": 0.4517090059986312,
-      "grad_norm": 0.0005481303087435663,
-      "learning_rate": 0.00019999990002613825,
-      "loss": 46.0,
-      "step": 2805
-    },
-    {
-      "epoch": 0.45187004307741857,
-      "grad_norm": 0.0012723577674478292,
-      "learning_rate": 0.00019999989995458782,
-      "loss": 46.0,
-      "step": 2806
-    },
-    {
-      "epoch": 0.45203108015620597,
-      "grad_norm": 0.0003037889546249062,
-      "learning_rate": 0.0001999998998830118,
-      "loss": 46.0,
-      "step": 2807
-    },
-    {
-      "epoch": 0.45219211723499336,
-      "grad_norm": 0.0007623456185683608,
-      "learning_rate": 0.00019999989981141016,
-      "loss": 46.0,
-      "step": 2808
-    },
-    {
-      "epoch": 0.45235315431378076,
-      "grad_norm": 0.0004901849897578359,
-      "learning_rate": 0.00019999989973978294,
-      "loss": 46.0,
-      "step": 2809
-    },
-    {
-      "epoch": 0.45251419139256815,
-      "grad_norm": 0.0006916733109392226,
-      "learning_rate": 0.00019999989966813014,
-      "loss": 46.0,
-      "step": 2810
-    },
-    {
-      "epoch": 0.45267522847135555,
-      "grad_norm": 0.0007174148922786117,
-      "learning_rate": 0.00019999989959645174,
-      "loss": 46.0,
-      "step": 2811
-    },
-    {
-      "epoch": 0.45283626555014295,
-      "grad_norm": 0.0003123388160020113,
-      "learning_rate": 0.0001999998995247477,
-      "loss": 46.0,
-      "step": 2812
-    },
-    {
-      "epoch": 0.4529973026289303,
-      "grad_norm": 0.00042951980140060186,
-      "learning_rate": 0.0001999998994530181,
-      "loss": 46.0,
-      "step": 2813
-    },
-    {
-      "epoch": 0.4531583397077177,
-      "grad_norm": 0.0006232656305655837,
-      "learning_rate": 0.00019999989938126293,
-      "loss": 46.0,
-      "step": 2814
-    },
-    {
-      "epoch": 0.4533193767865051,
-      "grad_norm": 0.0007511017029173672,
-      "learning_rate": 0.00019999989930948216,
-      "loss": 46.0,
-      "step": 2815
-    },
-    {
-      "epoch": 0.4534804138652925,
-      "grad_norm": 0.0006081481114961207,
-      "learning_rate": 0.00019999989923767577,
-      "loss": 46.0,
-      "step": 2816
-    },
-    {
-      "epoch": 0.45364145094407987,
-      "grad_norm": 0.0004693980736192316,
-      "learning_rate": 0.00019999989916584377,
-      "loss": 46.0,
-      "step": 2817
-    },
-    {
-      "epoch": 0.45380248802286727,
-      "grad_norm": 0.0008029431337490678,
-      "learning_rate": 0.0001999998990939862,
-      "loss": 46.0,
-      "step": 2818
-    },
-    {
-      "epoch": 0.45396352510165466,
-      "grad_norm": 0.00046594266314059496,
-      "learning_rate": 0.00019999989902210307,
-      "loss": 46.0,
-      "step": 2819
-    },
-    {
-      "epoch": 0.45412456218044206,
-      "grad_norm": 0.000332236522808671,
-      "learning_rate": 0.00019999989895019428,
-      "loss": 46.0,
-      "step": 2820
-    },
-    {
-      "epoch": 0.45428559925922946,
-      "grad_norm": 0.00030856754165142775,
-      "learning_rate": 0.00019999989887825993,
-      "loss": 46.0,
-      "step": 2821
-    },
-    {
-      "epoch": 0.45444663633801685,
-      "grad_norm": 0.0003584769438020885,
-      "learning_rate": 0.0001999998988063,
-      "loss": 46.0,
-      "step": 2822
-    },
-    {
-      "epoch": 0.4546076734168042,
-      "grad_norm": 0.0002391423040535301,
-      "learning_rate": 0.00019999989873431445,
-      "loss": 46.0,
-      "step": 2823
-    },
-    {
-      "epoch": 0.4547687104955916,
-      "grad_norm": 0.0005808458663523197,
-      "learning_rate": 0.0001999998986623033,
-      "loss": 46.0,
-      "step": 2824
-    },
-    {
-      "epoch": 0.454929747574379,
-      "grad_norm": 0.000656791205983609,
-      "learning_rate": 0.0001999998985902666,
-      "loss": 46.0,
-      "step": 2825
-    },
-    {
-      "epoch": 0.4550907846531664,
-      "grad_norm": 0.0007843867060728371,
-      "learning_rate": 0.00019999989851820425,
-      "loss": 46.0,
-      "step": 2826
-    },
-    {
-      "epoch": 0.4552518217319538,
-      "grad_norm": 0.0003235080512240529,
-      "learning_rate": 0.0001999998984461163,
-      "loss": 46.0,
-      "step": 2827
-    },
-    {
-      "epoch": 0.4554128588107412,
-      "grad_norm": 0.0005541152786463499,
-      "learning_rate": 0.0001999998983740028,
-      "loss": 46.0,
-      "step": 2828
-    },
-    {
-      "epoch": 0.45557389588952857,
-      "grad_norm": 0.0004606999282259494,
-      "learning_rate": 0.0001999998983018637,
-      "loss": 46.0,
-      "step": 2829
-    },
-    {
-      "epoch": 0.45573493296831596,
-      "grad_norm": 0.0011908607557415962,
-      "learning_rate": 0.00019999989822969897,
-      "loss": 46.0,
-      "step": 2830
-    },
-    {
-      "epoch": 0.45589597004710336,
-      "grad_norm": 0.0006549574318341911,
-      "learning_rate": 0.00019999989815750867,
-      "loss": 46.0,
-      "step": 2831
-    },
-    {
-      "epoch": 0.45605700712589076,
-      "grad_norm": 0.0007461561472155154,
-      "learning_rate": 0.00019999989808529276,
-      "loss": 46.0,
-      "step": 2832
-    },
-    {
-      "epoch": 0.45621804420467815,
-      "grad_norm": 0.00095327541930601,
-      "learning_rate": 0.00019999989801305128,
-      "loss": 46.0,
-      "step": 2833
-    },
-    {
-      "epoch": 0.4563790812834655,
-      "grad_norm": 0.0013621571706607938,
-      "learning_rate": 0.00019999989794078417,
-      "loss": 46.0,
-      "step": 2834
-    },
-    {
-      "epoch": 0.4565401183622529,
-      "grad_norm": 0.0004566084826365113,
-      "learning_rate": 0.00019999989786849152,
-      "loss": 46.0,
-      "step": 2835
-    },
-    {
-      "epoch": 0.4567011554410403,
-      "grad_norm": 0.0011674485867843032,
-      "learning_rate": 0.00019999989779617323,
-      "loss": 46.0,
-      "step": 2836
-    },
-    {
-      "epoch": 0.4568621925198277,
-      "grad_norm": 0.0005980342975817621,
-      "learning_rate": 0.00019999989772382938,
-      "loss": 46.0,
-      "step": 2837
-    },
-    {
-      "epoch": 0.4570232295986151,
-      "grad_norm": 0.0032675962429493666,
-      "learning_rate": 0.00019999989765145988,
-      "loss": 46.0,
-      "step": 2838
-    },
-    {
-      "epoch": 0.4571842666774025,
-      "grad_norm": 0.0006431429646909237,
-      "learning_rate": 0.00019999989757906483,
-      "loss": 46.0,
-      "step": 2839
-    },
-    {
-      "epoch": 0.45734530375618987,
-      "grad_norm": 0.0008999565034173429,
-      "learning_rate": 0.00019999989750664417,
-      "loss": 46.0,
-      "step": 2840
-    },
-    {
-      "epoch": 0.45750634083497727,
-      "grad_norm": 0.00036903267027810216,
-      "learning_rate": 0.00019999989743419794,
-      "loss": 46.0,
-      "step": 2841
-    },
-    {
-      "epoch": 0.45766737791376466,
-      "grad_norm": 0.0006419433047994971,
-      "learning_rate": 0.00019999989736172607,
-      "loss": 46.0,
-      "step": 2842
-    },
-    {
-      "epoch": 0.45782841499255206,
-      "grad_norm": 0.0025293666403740644,
-      "learning_rate": 0.00019999989728922864,
-      "loss": 46.0,
-      "step": 2843
-    },
-    {
-      "epoch": 0.45798945207133945,
-      "grad_norm": 0.002030485775321722,
-      "learning_rate": 0.0001999998972167056,
-      "loss": 46.0,
-      "step": 2844
-    },
-    {
-      "epoch": 0.4581504891501268,
-      "grad_norm": 0.0009818007238209248,
-      "learning_rate": 0.00019999989714415697,
-      "loss": 46.0,
-      "step": 2845
-    },
-    {
-      "epoch": 0.4583115262289142,
-      "grad_norm": 0.00035288967774249613,
-      "learning_rate": 0.00019999989707158273,
-      "loss": 46.0,
-      "step": 2846
-    },
-    {
-      "epoch": 0.4584725633077016,
-      "grad_norm": 0.0026120112743228674,
-      "learning_rate": 0.00019999989699898292,
-      "loss": 46.0,
-      "step": 2847
-    },
-    {
-      "epoch": 0.458633600386489,
-      "grad_norm": 0.0004638070531655103,
-      "learning_rate": 0.0001999998969263575,
-      "loss": 46.0,
-      "step": 2848
-    },
-    {
-      "epoch": 0.4587946374652764,
-      "grad_norm": 0.0006521355826407671,
-      "learning_rate": 0.0001999998968537065,
-      "loss": 46.0,
-      "step": 2849
-    },
-    {
-      "epoch": 0.4589556745440638,
-      "grad_norm": 0.00038754616980440915,
-      "learning_rate": 0.00019999989678102988,
-      "loss": 46.0,
-      "step": 2850
-    },
-    {
-      "epoch": 0.45911671162285117,
-      "grad_norm": 0.0005511146737262607,
-      "learning_rate": 0.00019999989670832768,
-      "loss": 46.0,
-      "step": 2851
-    },
-    {
-      "epoch": 0.45927774870163857,
-      "grad_norm": 0.0006465655169449747,
-      "learning_rate": 0.00019999989663559988,
-      "loss": 46.0,
-      "step": 2852
-    },
-    {
-      "epoch": 0.45943878578042596,
-      "grad_norm": 0.0006740029202774167,
-      "learning_rate": 0.0001999998965628465,
-      "loss": 46.0,
-      "step": 2853
-    },
-    {
-      "epoch": 0.45959982285921336,
-      "grad_norm": 0.001330913626588881,
-      "learning_rate": 0.0001999998964900675,
-      "loss": 46.0,
-      "step": 2854
-    },
-    {
-      "epoch": 0.4597608599380007,
-      "grad_norm": 0.00043661901145242155,
-      "learning_rate": 0.00019999989641726292,
-      "loss": 46.0,
-      "step": 2855
-    },
-    {
-      "epoch": 0.4599218970167881,
-      "grad_norm": 0.00043866271153092384,
-      "learning_rate": 0.00019999989634443278,
-      "loss": 46.0,
-      "step": 2856
-    },
-    {
-      "epoch": 0.4600829340955755,
-      "grad_norm": 0.0036805500276386738,
-      "learning_rate": 0.000199999896271577,
-      "loss": 46.0,
-      "step": 2857
-    },
-    {
-      "epoch": 0.4602439711743629,
-      "grad_norm": 0.0005626160418614745,
-      "learning_rate": 0.00019999989619869562,
-      "loss": 46.0,
-      "step": 2858
-    },
-    {
-      "epoch": 0.4604050082531503,
-      "grad_norm": 0.001611289568245411,
-      "learning_rate": 0.00019999989612578867,
-      "loss": 46.0,
-      "step": 2859
-    },
-    {
-      "epoch": 0.4605660453319377,
-      "grad_norm": 0.0005461085238493979,
-      "learning_rate": 0.0001999998960528561,
-      "loss": 46.0,
-      "step": 2860
-    },
-    {
-      "epoch": 0.4607270824107251,
-      "grad_norm": 0.0008688995148986578,
-      "learning_rate": 0.00019999989597989796,
-      "loss": 46.0,
-      "step": 2861
-    },
-    {
-      "epoch": 0.46088811948951247,
-      "grad_norm": 0.00169644714333117,
-      "learning_rate": 0.00019999989590691421,
-      "loss": 46.0,
-      "step": 2862
-    },
-    {
-      "epoch": 0.46104915656829987,
-      "grad_norm": 0.0005847355350852013,
-      "learning_rate": 0.00019999989583390488,
-      "loss": 46.0,
-      "step": 2863
-    },
-    {
-      "epoch": 0.46121019364708726,
-      "grad_norm": 0.0006305525312200189,
-      "learning_rate": 0.00019999989576086996,
-      "loss": 46.0,
-      "step": 2864
-    },
-    {
-      "epoch": 0.46137123072587466,
-      "grad_norm": 0.00047217195970006287,
-      "learning_rate": 0.00019999989568780942,
-      "loss": 46.0,
-      "step": 2865
-    },
-    {
-      "epoch": 0.461532267804662,
-      "grad_norm": 0.0029122272972017527,
-      "learning_rate": 0.0001999998956147233,
-      "loss": 46.0,
-      "step": 2866
-    },
-    {
-      "epoch": 0.4616933048834494,
-      "grad_norm": 0.0008619397412985563,
-      "learning_rate": 0.0001999998955416116,
-      "loss": 46.0,
-      "step": 2867
-    },
-    {
-      "epoch": 0.4618543419622368,
-      "grad_norm": 0.0006143774371594191,
-      "learning_rate": 0.00019999989546847427,
-      "loss": 46.0,
-      "step": 2868
-    },
-    {
-      "epoch": 0.4620153790410242,
-      "grad_norm": 0.0007360770250670612,
-      "learning_rate": 0.00019999989539531136,
-      "loss": 46.0,
-      "step": 2869
-    },
-    {
-      "epoch": 0.4621764161198116,
-      "grad_norm": 0.00042803262476809323,
-      "learning_rate": 0.00019999989532212283,
-      "loss": 46.0,
-      "step": 2870
-    },
-    {
-      "epoch": 0.462337453198599,
-      "grad_norm": 0.0009471254306845367,
-      "learning_rate": 0.00019999989524890874,
-      "loss": 46.0,
-      "step": 2871
-    },
-    {
-      "epoch": 0.4624984902773864,
-      "grad_norm": 0.0013519171625375748,
-      "learning_rate": 0.00019999989517566907,
-      "loss": 46.0,
-      "step": 2872
-    },
-    {
-      "epoch": 0.46265952735617377,
-      "grad_norm": 0.0003170229902025312,
-      "learning_rate": 0.00019999989510240379,
-      "loss": 46.0,
-      "step": 2873
-    },
-    {
-      "epoch": 0.46282056443496117,
-      "grad_norm": 0.0005881020915694535,
-      "learning_rate": 0.00019999989502911288,
-      "loss": 46.0,
-      "step": 2874
-    },
-    {
-      "epoch": 0.46298160151374856,
-      "grad_norm": 0.000855987542308867,
-      "learning_rate": 0.0001999998949557964,
-      "loss": 46.0,
-      "step": 2875
-    },
-    {
-      "epoch": 0.4631426385925359,
-      "grad_norm": 0.0006057831342332065,
-      "learning_rate": 0.00019999989488245435,
-      "loss": 46.0,
-      "step": 2876
-    },
-    {
-      "epoch": 0.4633036756713233,
-      "grad_norm": 0.0009846428874880075,
-      "learning_rate": 0.00019999989480908668,
-      "loss": 46.0,
-      "step": 2877
-    },
-    {
-      "epoch": 0.4634647127501107,
-      "grad_norm": 0.0002604479086585343,
-      "learning_rate": 0.0001999998947356934,
-      "loss": 46.0,
-      "step": 2878
-    },
-    {
-      "epoch": 0.4636257498288981,
-      "grad_norm": 0.00033883212017826736,
-      "learning_rate": 0.00019999989466227457,
-      "loss": 46.0,
-      "step": 2879
-    },
-    {
-      "epoch": 0.4637867869076855,
-      "grad_norm": 0.001364146126434207,
-      "learning_rate": 0.00019999989458883012,
-      "loss": 46.0,
-      "step": 2880
-    },
-    {
-      "epoch": 0.4639478239864729,
-      "grad_norm": 0.00047338224248960614,
-      "learning_rate": 0.00019999989451536005,
-      "loss": 46.0,
-      "step": 2881
-    },
-    {
-      "epoch": 0.4641088610652603,
-      "grad_norm": 0.000514526735059917,
-      "learning_rate": 0.00019999989444186443,
-      "loss": 46.0,
-      "step": 2882
-    },
-    {
-      "epoch": 0.4642698981440477,
-      "grad_norm": 0.0002897829399444163,
-      "learning_rate": 0.00019999989436834319,
-      "loss": 46.0,
-      "step": 2883
-    },
-    {
-      "epoch": 0.4644309352228351,
-      "grad_norm": 0.0007724882452748716,
-      "learning_rate": 0.00019999989429479636,
-      "loss": 46.0,
-      "step": 2884
-    },
-    {
-      "epoch": 0.46459197230162247,
-      "grad_norm": 0.0005975780077278614,
-      "learning_rate": 0.00019999989422122394,
-      "loss": 46.0,
-      "step": 2885
-    },
-    {
-      "epoch": 0.46475300938040986,
-      "grad_norm": 0.00030973469256423414,
-      "learning_rate": 0.0001999998941476259,
-      "loss": 46.0,
-      "step": 2886
-    },
-    {
-      "epoch": 0.4649140464591972,
-      "grad_norm": 0.000649851921480149,
-      "learning_rate": 0.0001999998940740023,
-      "loss": 46.0,
-      "step": 2887
-    },
-    {
-      "epoch": 0.4650750835379846,
-      "grad_norm": 0.000337887613568455,
-      "learning_rate": 0.00019999989400035307,
-      "loss": 46.0,
-      "step": 2888
-    },
-    {
-      "epoch": 0.465236120616772,
-      "grad_norm": 0.0003283452242612839,
-      "learning_rate": 0.00019999989392667828,
-      "loss": 46.0,
-      "step": 2889
-    },
-    {
-      "epoch": 0.4653971576955594,
-      "grad_norm": 0.0003954330750275403,
-      "learning_rate": 0.00019999989385297788,
-      "loss": 46.0,
-      "step": 2890
-    },
-    {
-      "epoch": 0.4655581947743468,
-      "grad_norm": 0.0007709976052865386,
-      "learning_rate": 0.0001999998937792519,
-      "loss": 46.0,
-      "step": 2891
-    },
-    {
-      "epoch": 0.4657192318531342,
-      "grad_norm": 0.0003327724989503622,
-      "learning_rate": 0.00019999989370550033,
-      "loss": 46.0,
-      "step": 2892
-    },
-    {
-      "epoch": 0.4658802689319216,
-      "grad_norm": 0.0004651611379813403,
-      "learning_rate": 0.0001999998936317231,
-      "loss": 46.0,
-      "step": 2893
-    },
-    {
-      "epoch": 0.466041306010709,
-      "grad_norm": 0.001808120054192841,
-      "learning_rate": 0.00019999989355792033,
-      "loss": 46.0,
-      "step": 2894
-    },
-    {
-      "epoch": 0.4662023430894964,
-      "grad_norm": 0.000348433677572757,
-      "learning_rate": 0.00019999989348409196,
-      "loss": 46.0,
-      "step": 2895
-    },
-    {
-      "epoch": 0.46636338016828377,
-      "grad_norm": 0.0008953766664490104,
-      "learning_rate": 0.00019999989341023798,
-      "loss": 46.0,
-      "step": 2896
-    },
-    {
-      "epoch": 0.46652441724707117,
-      "grad_norm": 0.0004735529946628958,
-      "learning_rate": 0.00019999989333635844,
-      "loss": 46.0,
-      "step": 2897
-    },
-    {
-      "epoch": 0.4666854543258585,
-      "grad_norm": 0.0022318630944937468,
-      "learning_rate": 0.00019999989326245328,
-      "loss": 46.0,
-      "step": 2898
-    },
-    {
-      "epoch": 0.4668464914046459,
-      "grad_norm": 0.00036539489519782364,
-      "learning_rate": 0.00019999989318852251,
-      "loss": 46.0,
-      "step": 2899
-    },
-    {
-      "epoch": 0.4670075284834333,
-      "grad_norm": 0.0010025915689766407,
-      "learning_rate": 0.00019999989311456618,
-      "loss": 46.0,
-      "step": 2900
-    },
-    {
-      "epoch": 0.4671685655622207,
-      "grad_norm": 0.0004213416832499206,
-      "learning_rate": 0.00019999989304058424,
-      "loss": 46.0,
-      "step": 2901
-    },
-    {
-      "epoch": 0.4673296026410081,
-      "grad_norm": 0.0013688647886738181,
-      "learning_rate": 0.00019999989296657668,
-      "loss": 46.0,
-      "step": 2902
-    },
-    {
-      "epoch": 0.4674906397197955,
-      "grad_norm": 0.000347552151652053,
-      "learning_rate": 0.00019999989289254356,
-      "loss": 46.0,
-      "step": 2903
-    },
-    {
-      "epoch": 0.4676516767985829,
-      "grad_norm": 0.0003829733468592167,
-      "learning_rate": 0.00019999989281848483,
-      "loss": 46.0,
-      "step": 2904
-    },
-    {
-      "epoch": 0.4678127138773703,
-      "grad_norm": 0.0007045012316666543,
-      "learning_rate": 0.0001999998927444005,
-      "loss": 46.0,
-      "step": 2905
-    },
-    {
-      "epoch": 0.4679737509561577,
-      "grad_norm": 0.0012553539127111435,
-      "learning_rate": 0.0001999998926702906,
-      "loss": 46.0,
-      "step": 2906
-    },
-    {
-      "epoch": 0.46813478803494507,
-      "grad_norm": 0.0005822149687446654,
-      "learning_rate": 0.00019999989259615508,
-      "loss": 46.0,
-      "step": 2907
-    },
-    {
-      "epoch": 0.4682958251137324,
-      "grad_norm": 0.0007119124638848007,
-      "learning_rate": 0.00019999989252199397,
-      "loss": 46.0,
-      "step": 2908
-    },
-    {
-      "epoch": 0.4684568621925198,
-      "grad_norm": 0.0009871729416772723,
-      "learning_rate": 0.00019999989244780727,
-      "loss": 46.0,
-      "step": 2909
-    },
-    {
-      "epoch": 0.4686178992713072,
-      "grad_norm": 0.0006980373291298747,
-      "learning_rate": 0.000199999892373595,
-      "loss": 46.0,
-      "step": 2910
-    },
-    {
-      "epoch": 0.4687789363500946,
-      "grad_norm": 0.0002902986598201096,
-      "learning_rate": 0.0001999998922993571,
-      "loss": 46.0,
-      "step": 2911
-    },
-    {
-      "epoch": 0.468939973428882,
-      "grad_norm": 0.0006203086813911796,
-      "learning_rate": 0.0001999998922250936,
-      "loss": 46.0,
-      "step": 2912
-    },
-    {
-      "epoch": 0.4691010105076694,
-      "grad_norm": 0.0005238538724370301,
-      "learning_rate": 0.00019999989215080453,
-      "loss": 46.0,
-      "step": 2913
-    },
-    {
-      "epoch": 0.4692620475864568,
-      "grad_norm": 0.0011008810251951218,
-      "learning_rate": 0.00019999989207648985,
-      "loss": 46.0,
-      "step": 2914
-    },
-    {
-      "epoch": 0.4694230846652442,
-      "grad_norm": 0.0006297993822954595,
-      "learning_rate": 0.00019999989200214957,
-      "loss": 46.0,
-      "step": 2915
-    },
-    {
-      "epoch": 0.4695841217440316,
-      "grad_norm": 0.00040685865678824484,
-      "learning_rate": 0.0001999998919277837,
-      "loss": 46.0,
-      "step": 2916
-    },
-    {
-      "epoch": 0.469745158822819,
-      "grad_norm": 0.0005493556964211166,
-      "learning_rate": 0.00019999989185339224,
-      "loss": 46.0,
-      "step": 2917
-    },
-    {
-      "epoch": 0.46990619590160637,
-      "grad_norm": 0.0005613280809484422,
-      "learning_rate": 0.0001999998917789752,
-      "loss": 46.0,
-      "step": 2918
-    },
-    {
-      "epoch": 0.4700672329803937,
-      "grad_norm": 0.000449535611551255,
-      "learning_rate": 0.00019999989170453255,
-      "loss": 46.0,
-      "step": 2919
-    },
-    {
-      "epoch": 0.4702282700591811,
-      "grad_norm": 0.00031375791877508163,
-      "learning_rate": 0.00019999989163006432,
-      "loss": 46.0,
-      "step": 2920
-    },
-    {
-      "epoch": 0.4703893071379685,
-      "grad_norm": 0.0007060517091304064,
-      "learning_rate": 0.00019999989155557047,
-      "loss": 46.0,
-      "step": 2921
-    },
-    {
-      "epoch": 0.4705503442167559,
-      "grad_norm": 0.0008331080316565931,
-      "learning_rate": 0.00019999989148105103,
-      "loss": 46.0,
-      "step": 2922
-    },
-    {
-      "epoch": 0.4707113812955433,
-      "grad_norm": 0.0007294303504750133,
-      "learning_rate": 0.000199999891406506,
-      "loss": 46.0,
-      "step": 2923
-    },
-    {
-      "epoch": 0.4708724183743307,
-      "grad_norm": 0.0011409181170165539,
-      "learning_rate": 0.00019999989133193536,
-      "loss": 46.0,
-      "step": 2924
-    },
-    {
-      "epoch": 0.4710334554531181,
-      "grad_norm": 0.0006150680710561574,
-      "learning_rate": 0.00019999989125733916,
-      "loss": 46.0,
-      "step": 2925
-    },
-    {
-      "epoch": 0.4711944925319055,
-      "grad_norm": 0.0014988044276833534,
-      "learning_rate": 0.00019999989118271735,
-      "loss": 46.0,
-      "step": 2926
-    },
-    {
-      "epoch": 0.4713555296106929,
-      "grad_norm": 0.0005371063598431647,
-      "learning_rate": 0.00019999989110806995,
-      "loss": 46.0,
-      "step": 2927
-    },
-    {
-      "epoch": 0.4715165666894803,
-      "grad_norm": 0.00030780400265939534,
-      "learning_rate": 0.00019999989103339693,
-      "loss": 46.0,
-      "step": 2928
-    },
-    {
-      "epoch": 0.4716776037682676,
-      "grad_norm": 0.0006254728650674224,
-      "learning_rate": 0.00019999989095869833,
-      "loss": 46.0,
-      "step": 2929
-    },
-    {
-      "epoch": 0.471838640847055,
-      "grad_norm": 0.0007008228567428887,
-      "learning_rate": 0.00019999989088397414,
-      "loss": 46.0,
-      "step": 2930
-    },
-    {
-      "epoch": 0.4719996779258424,
-      "grad_norm": 0.0005706832162104547,
-      "learning_rate": 0.00019999989080922434,
-      "loss": 46.0,
-      "step": 2931
-    },
-    {
-      "epoch": 0.4721607150046298,
-      "grad_norm": 0.002139221178367734,
-      "learning_rate": 0.00019999989073444895,
-      "loss": 46.0,
-      "step": 2932
-    },
-    {
-      "epoch": 0.4723217520834172,
-      "grad_norm": 0.0012941916938871145,
-      "learning_rate": 0.000199999890659648,
-      "loss": 46.0,
-      "step": 2933
-    },
-    {
-      "epoch": 0.4724827891622046,
-      "grad_norm": 0.0009738983935676515,
-      "learning_rate": 0.00019999989058482143,
-      "loss": 46.0,
-      "step": 2934
-    },
-    {
-      "epoch": 0.472643826240992,
-      "grad_norm": 0.00020528545428533107,
-      "learning_rate": 0.00019999989050996925,
-      "loss": 46.0,
-      "step": 2935
-    },
-    {
-      "epoch": 0.4728048633197794,
-      "grad_norm": 0.0004141554527450353,
-      "learning_rate": 0.00019999989043509149,
-      "loss": 46.0,
-      "step": 2936
-    },
-    {
-      "epoch": 0.4729659003985668,
-      "grad_norm": 0.0006022877059876919,
-      "learning_rate": 0.00019999989036018813,
-      "loss": 46.0,
-      "step": 2937
-    },
-    {
-      "epoch": 0.4731269374773542,
-      "grad_norm": 0.0015486562624573708,
-      "learning_rate": 0.0001999998902852592,
-      "loss": 46.0,
-      "step": 2938
-    },
-    {
-      "epoch": 0.4732879745561416,
-      "grad_norm": 0.0005348848062567413,
-      "learning_rate": 0.00019999989021030464,
-      "loss": 46.0,
-      "step": 2939
-    },
-    {
-      "epoch": 0.4734490116349289,
-      "grad_norm": 0.0005023290286771953,
-      "learning_rate": 0.0001999998901353245,
-      "loss": 46.0,
-      "step": 2940
-    },
-    {
-      "epoch": 0.4736100487137163,
-      "grad_norm": 0.002133709844201803,
-      "learning_rate": 0.00019999989006031876,
-      "loss": 46.0,
-      "step": 2941
-    },
-    {
-      "epoch": 0.4737710857925037,
-      "grad_norm": 0.001634693006053567,
-      "learning_rate": 0.00019999988998528742,
-      "loss": 46.0,
-      "step": 2942
-    },
-    {
-      "epoch": 0.4739321228712911,
-      "grad_norm": 0.0002572726516518742,
-      "learning_rate": 0.0001999998899102305,
-      "loss": 46.0,
-      "step": 2943
-    },
-    {
-      "epoch": 0.4740931599500785,
-      "grad_norm": 0.0014945456059649587,
-      "learning_rate": 0.00019999988983514797,
-      "loss": 46.0,
-      "step": 2944
-    },
-    {
-      "epoch": 0.4742541970288659,
-      "grad_norm": 0.0008017955697141588,
-      "learning_rate": 0.00019999988976003986,
-      "loss": 46.0,
-      "step": 2945
-    },
-    {
-      "epoch": 0.4744152341076533,
-      "grad_norm": 0.00044260459253564477,
-      "learning_rate": 0.00019999988968490614,
-      "loss": 46.0,
-      "step": 2946
-    },
-    {
-      "epoch": 0.4745762711864407,
-      "grad_norm": 0.0005199291626922786,
-      "learning_rate": 0.00019999988960974684,
-      "loss": 46.0,
-      "step": 2947
-    },
-    {
-      "epoch": 0.4747373082652281,
-      "grad_norm": 0.0008907295414246619,
-      "learning_rate": 0.00019999988953456194,
-      "loss": 46.0,
-      "step": 2948
-    },
-    {
-      "epoch": 0.4748983453440155,
-      "grad_norm": 0.0004547477583400905,
-      "learning_rate": 0.00019999988945935146,
-      "loss": 46.0,
-      "step": 2949
-    },
-    {
-      "epoch": 0.4750593824228028,
-      "grad_norm": 0.0004121024685446173,
-      "learning_rate": 0.00019999988938411537,
-      "loss": 46.0,
-      "step": 2950
-    },
-    {
-      "epoch": 0.4752204195015902,
-      "grad_norm": 0.0003815851523540914,
-      "learning_rate": 0.00019999988930885368,
-      "loss": 46.0,
-      "step": 2951
-    },
-    {
-      "epoch": 0.4753814565803776,
-      "grad_norm": 0.0004361814062576741,
-      "learning_rate": 0.0001999998892335664,
-      "loss": 46.0,
-      "step": 2952
-    },
-    {
-      "epoch": 0.475542493659165,
-      "grad_norm": 0.0012374628568068147,
-      "learning_rate": 0.00019999988915825353,
-      "loss": 46.0,
-      "step": 2953
-    },
-    {
-      "epoch": 0.4757035307379524,
-      "grad_norm": 0.0008402442908845842,
-      "learning_rate": 0.00019999988908291506,
-      "loss": 46.0,
-      "step": 2954
-    },
-    {
-      "epoch": 0.4758645678167398,
-      "grad_norm": 0.0012711393646895885,
-      "learning_rate": 0.000199999889007551,
-      "loss": 46.0,
-      "step": 2955
-    },
-    {
-      "epoch": 0.4760256048955272,
-      "grad_norm": 0.00039713442674838006,
-      "learning_rate": 0.00019999988893216135,
-      "loss": 46.0,
-      "step": 2956
-    },
-    {
-      "epoch": 0.4761866419743146,
-      "grad_norm": 0.000489401223603636,
-      "learning_rate": 0.00019999988885674607,
-      "loss": 46.0,
-      "step": 2957
-    },
-    {
-      "epoch": 0.476347679053102,
-      "grad_norm": 0.0008890616591088474,
-      "learning_rate": 0.00019999988878130522,
-      "loss": 46.0,
-      "step": 2958
-    },
-    {
-      "epoch": 0.4765087161318894,
-      "grad_norm": 0.0005563640152104199,
-      "learning_rate": 0.00019999988870583879,
-      "loss": 46.0,
-      "step": 2959
-    },
-    {
-      "epoch": 0.4766697532106768,
-      "grad_norm": 0.0015805872390046716,
-      "learning_rate": 0.00019999988863034676,
-      "loss": 46.0,
-      "step": 2960
-    },
-    {
-      "epoch": 0.4768307902894641,
-      "grad_norm": 0.0005280330660752952,
-      "learning_rate": 0.00019999988855482913,
-      "loss": 46.0,
-      "step": 2961
-    },
-    {
-      "epoch": 0.4769918273682515,
-      "grad_norm": 0.0005939175607636571,
-      "learning_rate": 0.00019999988847928588,
-      "loss": 46.0,
-      "step": 2962
-    },
-    {
-      "epoch": 0.4771528644470389,
-      "grad_norm": 0.0005209036753512919,
-      "learning_rate": 0.00019999988840371707,
-      "loss": 46.0,
-      "step": 2963
-    },
-    {
-      "epoch": 0.4773139015258263,
-      "grad_norm": 0.0004026646784041077,
-      "learning_rate": 0.00019999988832812267,
-      "loss": 46.0,
-      "step": 2964
-    },
-    {
-      "epoch": 0.4774749386046137,
-      "grad_norm": 0.0004014724399894476,
-      "learning_rate": 0.00019999988825250263,
-      "loss": 46.0,
-      "step": 2965
-    },
-    {
-      "epoch": 0.4776359756834011,
-      "grad_norm": 0.0016275751404464245,
-      "learning_rate": 0.00019999988817685703,
-      "loss": 46.0,
-      "step": 2966
-    },
-    {
-      "epoch": 0.4777970127621885,
-      "grad_norm": 0.00042475457303225994,
-      "learning_rate": 0.00019999988810118585,
-      "loss": 46.0,
-      "step": 2967
-    },
-    {
-      "epoch": 0.4779580498409759,
-      "grad_norm": 0.0003831503272522241,
-      "learning_rate": 0.00019999988802548905,
-      "loss": 46.0,
-      "step": 2968
-    },
-    {
-      "epoch": 0.4781190869197633,
-      "grad_norm": 0.0003797115059569478,
-      "learning_rate": 0.00019999988794976666,
-      "loss": 46.0,
-      "step": 2969
-    },
-    {
-      "epoch": 0.4782801239985507,
-      "grad_norm": 0.0003801226557698101,
-      "learning_rate": 0.00019999988787401866,
-      "loss": 46.0,
-      "step": 2970
-    },
-    {
-      "epoch": 0.4784411610773381,
-      "grad_norm": 0.0017341175116598606,
-      "learning_rate": 0.00019999988779824504,
-      "loss": 46.0,
-      "step": 2971
-    },
-    {
-      "epoch": 0.4786021981561254,
-      "grad_norm": 0.0006976477452553809,
-      "learning_rate": 0.0001999998877224459,
-      "loss": 46.0,
-      "step": 2972
-    },
-    {
-      "epoch": 0.4787632352349128,
-      "grad_norm": 0.0014606605982407928,
-      "learning_rate": 0.00019999988764662113,
-      "loss": 46.0,
-      "step": 2973
-    },
-    {
-      "epoch": 0.4789242723137002,
-      "grad_norm": 0.0003381700662430376,
-      "learning_rate": 0.00019999988757077075,
-      "loss": 46.0,
-      "step": 2974
-    },
-    {
-      "epoch": 0.4790853093924876,
-      "grad_norm": 0.00047852908028289676,
-      "learning_rate": 0.0001999998874948948,
-      "loss": 46.0,
-      "step": 2975
-    },
-    {
-      "epoch": 0.479246346471275,
-      "grad_norm": 0.0004297027480788529,
-      "learning_rate": 0.00019999988741899324,
-      "loss": 46.0,
-      "step": 2976
-    },
-    {
-      "epoch": 0.4794073835500624,
-      "grad_norm": 0.0004784474731422961,
-      "learning_rate": 0.0001999998873430661,
-      "loss": 46.0,
-      "step": 2977
-    },
-    {
-      "epoch": 0.4795684206288498,
-      "grad_norm": 0.00032422482036054134,
-      "learning_rate": 0.00019999988726711334,
-      "loss": 46.0,
-      "step": 2978
-    },
-    {
-      "epoch": 0.4797294577076372,
-      "grad_norm": 0.00032233702950179577,
-      "learning_rate": 0.000199999887191135,
-      "loss": 46.0,
-      "step": 2979
-    },
-    {
-      "epoch": 0.4798904947864246,
-      "grad_norm": 0.0008858292130753398,
-      "learning_rate": 0.00019999988711513108,
-      "loss": 46.0,
-      "step": 2980
-    },
-    {
-      "epoch": 0.480051531865212,
-      "grad_norm": 0.0009945160709321499,
-      "learning_rate": 0.00019999988703910156,
-      "loss": 46.0,
-      "step": 2981
-    },
-    {
-      "epoch": 0.48021256894399933,
-      "grad_norm": 0.00042676139855757356,
-      "learning_rate": 0.0001999998869630464,
-      "loss": 46.0,
-      "step": 2982
-    },
-    {
-      "epoch": 0.4803736060227867,
-      "grad_norm": 0.0012016601394861937,
-      "learning_rate": 0.0001999998868869657,
-      "loss": 46.0,
-      "step": 2983
-    },
-    {
-      "epoch": 0.4805346431015741,
-      "grad_norm": 0.000258055079029873,
-      "learning_rate": 0.00019999988681085936,
-      "loss": 46.0,
-      "step": 2984
-    },
-    {
-      "epoch": 0.4806956801803615,
-      "grad_norm": 0.0005329045816324651,
-      "learning_rate": 0.00019999988673472747,
-      "loss": 46.0,
-      "step": 2985
-    },
-    {
-      "epoch": 0.4808567172591489,
-      "grad_norm": 0.0013322388986125588,
-      "learning_rate": 0.00019999988665856996,
-      "loss": 46.0,
-      "step": 2986
-    },
-    {
-      "epoch": 0.4810177543379363,
-      "grad_norm": 0.0009696754277683794,
-      "learning_rate": 0.00019999988658238687,
-      "loss": 46.0,
-      "step": 2987
-    },
-    {
-      "epoch": 0.4811787914167237,
-      "grad_norm": 0.0004098875797353685,
-      "learning_rate": 0.0001999998865061782,
-      "loss": 46.0,
-      "step": 2988
-    },
-    {
-      "epoch": 0.4813398284955111,
-      "grad_norm": 0.0009699148940853775,
-      "learning_rate": 0.0001999998864299439,
-      "loss": 46.0,
-      "step": 2989
-    },
-    {
-      "epoch": 0.4815008655742985,
-      "grad_norm": 0.0005095636588521302,
-      "learning_rate": 0.00019999988635368402,
-      "loss": 46.0,
-      "step": 2990
-    },
-    {
-      "epoch": 0.4816619026530859,
-      "grad_norm": 0.0008789250277914107,
-      "learning_rate": 0.00019999988627739855,
-      "loss": 46.0,
-      "step": 2991
-    },
-    {
-      "epoch": 0.4818229397318733,
-      "grad_norm": 0.0006461109733209014,
-      "learning_rate": 0.00019999988620108744,
-      "loss": 46.0,
-      "step": 2992
-    },
-    {
-      "epoch": 0.48198397681066063,
-      "grad_norm": 0.0009350635809823871,
-      "learning_rate": 0.00019999988612475077,
-      "loss": 46.0,
-      "step": 2993
-    },
-    {
-      "epoch": 0.482145013889448,
-      "grad_norm": 0.001425218302756548,
-      "learning_rate": 0.00019999988604838852,
-      "loss": 46.0,
-      "step": 2994
-    },
-    {
-      "epoch": 0.4823060509682354,
-      "grad_norm": 0.0005072914063930511,
-      "learning_rate": 0.00019999988597200065,
-      "loss": 46.0,
-      "step": 2995
-    },
-    {
-      "epoch": 0.4824670880470228,
-      "grad_norm": 0.0014354382874444127,
-      "learning_rate": 0.00019999988589558722,
-      "loss": 46.0,
-      "step": 2996
-    },
-    {
-      "epoch": 0.4826281251258102,
-      "grad_norm": 0.001478009158745408,
-      "learning_rate": 0.00019999988581914817,
-      "loss": 46.0,
-      "step": 2997
-    },
-    {
-      "epoch": 0.4827891622045976,
-      "grad_norm": 0.0011180785950273275,
-      "learning_rate": 0.0001999998857426835,
-      "loss": 46.0,
-      "step": 2998
-    },
-    {
-      "epoch": 0.482950199283385,
-      "grad_norm": 0.0007119972142390907,
-      "learning_rate": 0.00019999988566619326,
-      "loss": 46.0,
-      "step": 2999
-    },
-    {
-      "epoch": 0.4831112363621724,
-      "grad_norm": 0.00042794988257810473,
-      "learning_rate": 0.00019999988558967746,
-      "loss": 46.0,
-      "step": 3000
-    },
-    {
-      "epoch": 0.4832722734409598,
-      "grad_norm": 0.00032044705585576594,
-      "learning_rate": 0.000199999885513136,
-      "loss": 46.0,
-      "step": 3001
-    },
-    {
-      "epoch": 0.4834333105197472,
-      "grad_norm": 0.0009639356285333633,
-      "learning_rate": 0.000199999885436569,
-      "loss": 46.0,
-      "step": 3002
-    },
-    {
-      "epoch": 0.48359434759853454,
-      "grad_norm": 0.0007271425565704703,
-      "learning_rate": 0.0001999998853599764,
-      "loss": 46.0,
-      "step": 3003
-    },
-    {
-      "epoch": 0.48375538467732193,
-      "grad_norm": 0.0011597374686971307,
-      "learning_rate": 0.00019999988528335817,
-      "loss": 46.0,
-      "step": 3004
-    },
-    {
-      "epoch": 0.4839164217561093,
-      "grad_norm": 0.000668774067889899,
-      "learning_rate": 0.00019999988520671437,
-      "loss": 46.0,
-      "step": 3005
-    },
-    {
-      "epoch": 0.4840774588348967,
-      "grad_norm": 0.0010206762235611677,
-      "learning_rate": 0.00019999988513004496,
-      "loss": 46.0,
-      "step": 3006
-    },
-    {
-      "epoch": 0.4842384959136841,
-      "grad_norm": 0.0016464662039652467,
-      "learning_rate": 0.00019999988505335,
-      "loss": 46.0,
-      "step": 3007
-    },
-    {
-      "epoch": 0.4843995329924715,
-      "grad_norm": 0.0003405835013836622,
-      "learning_rate": 0.00019999988497662938,
-      "loss": 46.0,
-      "step": 3008
-    },
-    {
-      "epoch": 0.4845605700712589,
-      "grad_norm": 0.0005689465324394405,
-      "learning_rate": 0.0001999998848998832,
-      "loss": 46.0,
-      "step": 3009
-    },
-    {
-      "epoch": 0.4847216071500463,
-      "grad_norm": 0.0002660571481101215,
-      "learning_rate": 0.0001999998848231114,
-      "loss": 46.0,
-      "step": 3010
-    },
-    {
-      "epoch": 0.4848826442288337,
-      "grad_norm": 0.0010441048070788383,
-      "learning_rate": 0.00019999988474631405,
-      "loss": 46.0,
-      "step": 3011
-    },
-    {
-      "epoch": 0.4850436813076211,
-      "grad_norm": 0.0018408495234325528,
-      "learning_rate": 0.00019999988466949109,
-      "loss": 46.0,
-      "step": 3012
-    },
-    {
-      "epoch": 0.4852047183864085,
-      "grad_norm": 0.0004564946866594255,
-      "learning_rate": 0.0001999998845926425,
-      "loss": 46.0,
-      "step": 3013
-    },
-    {
-      "epoch": 0.48536575546519584,
-      "grad_norm": 0.0008664042106829584,
-      "learning_rate": 0.00019999988451576835,
-      "loss": 46.0,
-      "step": 3014
-    },
-    {
-      "epoch": 0.48552679254398323,
-      "grad_norm": 0.0003404630406294018,
-      "learning_rate": 0.0001999998844388686,
-      "loss": 46.0,
-      "step": 3015
-    },
-    {
-      "epoch": 0.48568782962277063,
-      "grad_norm": 0.0006898824940435588,
-      "learning_rate": 0.00019999988436194324,
-      "loss": 46.0,
-      "step": 3016
-    },
-    {
-      "epoch": 0.485848866701558,
-      "grad_norm": 0.0003574106376618147,
-      "learning_rate": 0.0001999998842849923,
-      "loss": 46.0,
-      "step": 3017
-    },
-    {
-      "epoch": 0.4860099037803454,
-      "grad_norm": 0.0004943932290188968,
-      "learning_rate": 0.00019999988420801577,
-      "loss": 46.0,
-      "step": 3018
-    },
-    {
-      "epoch": 0.4861709408591328,
-      "grad_norm": 0.001738535356707871,
-      "learning_rate": 0.00019999988413101362,
-      "loss": 46.0,
-      "step": 3019
-    },
-    {
-      "epoch": 0.4863319779379202,
-      "grad_norm": 0.0010401951149106026,
-      "learning_rate": 0.00019999988405398588,
-      "loss": 46.0,
-      "step": 3020
-    },
-    {
-      "epoch": 0.4864930150167076,
-      "grad_norm": 0.0011530905030667782,
-      "learning_rate": 0.00019999988397693258,
-      "loss": 46.0,
-      "step": 3021
-    },
-    {
-      "epoch": 0.486654052095495,
-      "grad_norm": 0.0005458248779177666,
-      "learning_rate": 0.00019999988389985367,
-      "loss": 46.0,
-      "step": 3022
-    },
-    {
-      "epoch": 0.4868150891742824,
-      "grad_norm": 0.0005783428205177188,
-      "learning_rate": 0.00019999988382274914,
-      "loss": 46.0,
-      "step": 3023
-    },
-    {
-      "epoch": 0.4869761262530698,
-      "grad_norm": 0.0030912079382687807,
-      "learning_rate": 0.00019999988374561905,
-      "loss": 46.0,
-      "step": 3024
-    },
-    {
-      "epoch": 0.48713716333185714,
-      "grad_norm": 0.0007660651463083923,
-      "learning_rate": 0.00019999988366846335,
-      "loss": 46.0,
-      "step": 3025
-    },
-    {
-      "epoch": 0.48729820041064453,
-      "grad_norm": 0.0005894253263249993,
-      "learning_rate": 0.00019999988359128203,
-      "loss": 46.0,
-      "step": 3026
-    },
-    {
-      "epoch": 0.48745923748943193,
-      "grad_norm": 0.00044560522655956447,
-      "learning_rate": 0.00019999988351407515,
-      "loss": 46.0,
-      "step": 3027
-    },
-    {
-      "epoch": 0.4876202745682193,
-      "grad_norm": 0.0002980520948767662,
-      "learning_rate": 0.00019999988343684266,
-      "loss": 46.0,
-      "step": 3028
-    },
-    {
-      "epoch": 0.4877813116470067,
-      "grad_norm": 0.0011412062449380755,
-      "learning_rate": 0.00019999988335958458,
-      "loss": 46.0,
-      "step": 3029
-    },
-    {
-      "epoch": 0.4879423487257941,
-      "grad_norm": 0.0021029768977314234,
-      "learning_rate": 0.0001999998832823009,
-      "loss": 46.0,
-      "step": 3030
-    },
-    {
-      "epoch": 0.4881033858045815,
-      "grad_norm": 0.0005195364938117564,
-      "learning_rate": 0.00019999988320499164,
-      "loss": 46.0,
-      "step": 3031
-    },
-    {
-      "epoch": 0.4882644228833689,
-      "grad_norm": 0.0009377730893902481,
-      "learning_rate": 0.00019999988312765677,
-      "loss": 46.0,
-      "step": 3032
-    },
-    {
-      "epoch": 0.4884254599621563,
-      "grad_norm": 0.0007240478880703449,
-      "learning_rate": 0.00019999988305029632,
-      "loss": 46.0,
-      "step": 3033
-    },
-    {
-      "epoch": 0.4885864970409437,
-      "grad_norm": 0.0002842667163349688,
-      "learning_rate": 0.00019999988297291022,
-      "loss": 46.0,
-      "step": 3034
-    },
-    {
-      "epoch": 0.48874753411973104,
-      "grad_norm": 0.00034056042204611003,
-      "learning_rate": 0.0001999998828954986,
-      "loss": 46.0,
-      "step": 3035
-    },
-    {
-      "epoch": 0.48890857119851844,
-      "grad_norm": 0.0004168835876043886,
-      "learning_rate": 0.00019999988281806135,
-      "loss": 46.0,
-      "step": 3036
-    },
-    {
-      "epoch": 0.48906960827730583,
-      "grad_norm": 0.00039020171971060336,
-      "learning_rate": 0.0001999998827405985,
-      "loss": 46.0,
-      "step": 3037
-    },
-    {
-      "epoch": 0.48923064535609323,
-      "grad_norm": 0.0003701295063365251,
-      "learning_rate": 0.00019999988266311008,
-      "loss": 46.0,
-      "step": 3038
-    },
-    {
-      "epoch": 0.4893916824348806,
-      "grad_norm": 0.0005590972141362727,
-      "learning_rate": 0.00019999988258559605,
-      "loss": 46.0,
-      "step": 3039
-    },
-    {
-      "epoch": 0.489552719513668,
-      "grad_norm": 0.0006196689209900796,
-      "learning_rate": 0.00019999988250805643,
-      "loss": 46.0,
-      "step": 3040
-    },
-    {
-      "epoch": 0.4897137565924554,
-      "grad_norm": 0.0006001039873808622,
-      "learning_rate": 0.0001999998824304912,
-      "loss": 46.0,
-      "step": 3041
-    },
-    {
-      "epoch": 0.4898747936712428,
-      "grad_norm": 0.002684792736545205,
-      "learning_rate": 0.00019999988235290037,
-      "loss": 46.0,
-      "step": 3042
-    },
-    {
-      "epoch": 0.4900358307500302,
-      "grad_norm": 0.0006069751107133925,
-      "learning_rate": 0.00019999988227528397,
-      "loss": 46.0,
-      "step": 3043
-    },
-    {
-      "epoch": 0.4901968678288176,
-      "grad_norm": 0.0007200596737675369,
-      "learning_rate": 0.00019999988219764197,
-      "loss": 46.0,
-      "step": 3044
-    },
-    {
-      "epoch": 0.490357904907605,
-      "grad_norm": 0.00042322510853409767,
-      "learning_rate": 0.00019999988211997436,
-      "loss": 46.0,
-      "step": 3045
-    },
-    {
-      "epoch": 0.49051894198639234,
-      "grad_norm": 0.0005256300792098045,
-      "learning_rate": 0.00019999988204228117,
-      "loss": 46.0,
-      "step": 3046
-    },
-    {
-      "epoch": 0.49067997906517974,
-      "grad_norm": 0.0006421637954190373,
-      "learning_rate": 0.00019999988196456236,
-      "loss": 46.0,
-      "step": 3047
-    },
-    {
-      "epoch": 0.49084101614396713,
-      "grad_norm": 0.000260350905591622,
-      "learning_rate": 0.000199999881886818,
-      "loss": 46.0,
-      "step": 3048
-    },
-    {
-      "epoch": 0.49100205322275453,
-      "grad_norm": 0.0004371090035419911,
-      "learning_rate": 0.000199999881809048,
-      "loss": 46.0,
-      "step": 3049
-    },
-    {
-      "epoch": 0.4911630903015419,
-      "grad_norm": 0.0016571134328842163,
-      "learning_rate": 0.00019999988173125243,
-      "loss": 46.0,
-      "step": 3050
-    },
-    {
-      "epoch": 0.4913241273803293,
-      "grad_norm": 0.0003236646589357406,
-      "learning_rate": 0.0001999998816534313,
-      "loss": 46.0,
-      "step": 3051
-    },
-    {
-      "epoch": 0.4914851644591167,
-      "grad_norm": 0.00041940153460018337,
-      "learning_rate": 0.0001999998815755845,
-      "loss": 46.0,
-      "step": 3052
-    },
-    {
-      "epoch": 0.4916462015379041,
-      "grad_norm": 0.0014828043058514595,
-      "learning_rate": 0.00019999988149771217,
-      "loss": 46.0,
-      "step": 3053
-    },
-    {
-      "epoch": 0.4918072386166915,
-      "grad_norm": 0.0022164764814078808,
-      "learning_rate": 0.00019999988141981422,
-      "loss": 46.0,
-      "step": 3054
-    },
-    {
-      "epoch": 0.4919682756954789,
-      "grad_norm": 0.0005107664619572461,
-      "learning_rate": 0.00019999988134189066,
-      "loss": 46.0,
-      "step": 3055
-    },
-    {
-      "epoch": 0.49212931277426625,
-      "grad_norm": 0.0004044045927003026,
-      "learning_rate": 0.0001999998812639415,
-      "loss": 46.0,
-      "step": 3056
-    },
-    {
-      "epoch": 0.49229034985305364,
-      "grad_norm": 0.0009663673117756844,
-      "learning_rate": 0.00019999988118596677,
-      "loss": 46.0,
-      "step": 3057
-    },
-    {
-      "epoch": 0.49245138693184104,
-      "grad_norm": 0.00046986175584606826,
-      "learning_rate": 0.00019999988110796645,
-      "loss": 46.0,
-      "step": 3058
-    },
-    {
-      "epoch": 0.49261242401062844,
-      "grad_norm": 0.001860951422713697,
-      "learning_rate": 0.0001999998810299405,
-      "loss": 46.0,
-      "step": 3059
-    },
-    {
-      "epoch": 0.49277346108941583,
-      "grad_norm": 0.002688717795535922,
-      "learning_rate": 0.000199999880951889,
-      "loss": 46.0,
-      "step": 3060
-    },
-    {
-      "epoch": 0.49293449816820323,
-      "grad_norm": 0.0012286669807508588,
-      "learning_rate": 0.00019999988087381187,
-      "loss": 46.0,
-      "step": 3061
-    },
-    {
-      "epoch": 0.4930955352469906,
-      "grad_norm": 0.0008862640825100243,
-      "learning_rate": 0.00019999988079570918,
-      "loss": 46.0,
-      "step": 3062
-    },
-    {
-      "epoch": 0.493256572325778,
-      "grad_norm": 0.000542375142686069,
-      "learning_rate": 0.00019999988071758084,
-      "loss": 46.0,
-      "step": 3063
-    },
-    {
-      "epoch": 0.4934176094045654,
-      "grad_norm": 0.00142467871773988,
-      "learning_rate": 0.00019999988063942693,
-      "loss": 46.0,
-      "step": 3064
-    },
-    {
-      "epoch": 0.4935786464833528,
-      "grad_norm": 0.000658223289065063,
-      "learning_rate": 0.00019999988056124745,
-      "loss": 46.0,
-      "step": 3065
-    },
-    {
-      "epoch": 0.4937396835621402,
-      "grad_norm": 0.00035880127688869834,
-      "learning_rate": 0.00019999988048304235,
-      "loss": 46.0,
-      "step": 3066
-    },
-    {
-      "epoch": 0.49390072064092755,
-      "grad_norm": 0.0008620198932476342,
-      "learning_rate": 0.00019999988040481168,
-      "loss": 46.0,
-      "step": 3067
-    },
-    {
-      "epoch": 0.49406175771971494,
-      "grad_norm": 0.0012930762022733688,
-      "learning_rate": 0.00019999988032655538,
-      "loss": 46.0,
-      "step": 3068
-    },
-    {
-      "epoch": 0.49422279479850234,
-      "grad_norm": 0.0018414922524243593,
-      "learning_rate": 0.00019999988024827352,
-      "loss": 46.0,
-      "step": 3069
-    },
-    {
-      "epoch": 0.49438383187728974,
-      "grad_norm": 0.0003766584559343755,
-      "learning_rate": 0.00019999988016996604,
-      "loss": 46.0,
-      "step": 3070
-    },
-    {
-      "epoch": 0.49454486895607713,
-      "grad_norm": 0.0010134689509868622,
-      "learning_rate": 0.000199999880091633,
-      "loss": 46.0,
-      "step": 3071
-    },
-    {
-      "epoch": 0.49470590603486453,
-      "grad_norm": 0.0007995095220394433,
-      "learning_rate": 0.00019999988001327432,
-      "loss": 46.0,
-      "step": 3072
-    },
-    {
-      "epoch": 0.4948669431136519,
-      "grad_norm": 0.00047520571388304234,
-      "learning_rate": 0.00019999987993489006,
-      "loss": 46.0,
-      "step": 3073
-    },
-    {
-      "epoch": 0.4950279801924393,
-      "grad_norm": 0.0032507237046957016,
-      "learning_rate": 0.0001999998798564802,
-      "loss": 46.0,
-      "step": 3074
-    },
-    {
-      "epoch": 0.4951890172712267,
-      "grad_norm": 0.001113923848606646,
-      "learning_rate": 0.00019999987977804476,
-      "loss": 46.0,
-      "step": 3075
-    },
-    {
-      "epoch": 0.4953500543500141,
-      "grad_norm": 0.001107540330849588,
-      "learning_rate": 0.00019999987969958373,
-      "loss": 46.0,
-      "step": 3076
-    },
-    {
-      "epoch": 0.4955110914288015,
-      "grad_norm": 0.0029324879869818687,
-      "learning_rate": 0.0001999998796210971,
-      "loss": 46.0,
-      "step": 3077
-    },
-    {
-      "epoch": 0.49567212850758885,
-      "grad_norm": 0.001303511206060648,
-      "learning_rate": 0.00019999987954258486,
-      "loss": 46.0,
-      "step": 3078
-    },
-    {
-      "epoch": 0.49583316558637625,
-      "grad_norm": 0.0004946336848661304,
-      "learning_rate": 0.00019999987946404705,
-      "loss": 46.0,
-      "step": 3079
-    },
-    {
-      "epoch": 0.49599420266516364,
-      "grad_norm": 0.0004824888310395181,
-      "learning_rate": 0.00019999987938548364,
-      "loss": 46.0,
-      "step": 3080
-    },
-    {
-      "epoch": 0.49615523974395104,
-      "grad_norm": 0.0008335686870850623,
-      "learning_rate": 0.0001999998793068946,
-      "loss": 46.0,
-      "step": 3081
-    },
-    {
-      "epoch": 0.49631627682273843,
-      "grad_norm": 0.0003709551237989217,
-      "learning_rate": 0.00019999987922828002,
-      "loss": 46.0,
-      "step": 3082
-    },
-    {
-      "epoch": 0.49647731390152583,
-      "grad_norm": 0.0006122171762399375,
-      "learning_rate": 0.0001999998791496398,
-      "loss": 46.0,
-      "step": 3083
-    },
-    {
-      "epoch": 0.4966383509803132,
-      "grad_norm": 0.0009241683874279261,
-      "learning_rate": 0.00019999987907097402,
-      "loss": 46.0,
-      "step": 3084
-    },
-    {
-      "epoch": 0.4967993880591006,
-      "grad_norm": 0.000317135127261281,
-      "learning_rate": 0.0001999998789922826,
-      "loss": 46.0,
-      "step": 3085
-    },
-    {
-      "epoch": 0.496960425137888,
-      "grad_norm": 0.0007719341083429754,
-      "learning_rate": 0.00019999987891356562,
-      "loss": 46.0,
-      "step": 3086
-    },
-    {
-      "epoch": 0.4971214622166754,
-      "grad_norm": 0.0006923080654814839,
-      "learning_rate": 0.00019999987883482305,
-      "loss": 46.0,
-      "step": 3087
-    },
-    {
-      "epoch": 0.49728249929546275,
-      "grad_norm": 0.0008883123518899083,
-      "learning_rate": 0.00019999987875605487,
-      "loss": 46.0,
-      "step": 3088
-    },
-    {
-      "epoch": 0.49744353637425015,
-      "grad_norm": 0.0022773360833525658,
-      "learning_rate": 0.0001999998786772611,
-      "loss": 46.0,
-      "step": 3089
-    },
-    {
-      "epoch": 0.49760457345303755,
-      "grad_norm": 0.002116635674610734,
-      "learning_rate": 0.00019999987859844174,
-      "loss": 46.0,
-      "step": 3090
-    },
-    {
-      "epoch": 0.49776561053182494,
-      "grad_norm": 0.00032865567482076585,
-      "learning_rate": 0.00019999987851959677,
-      "loss": 46.0,
-      "step": 3091
-    },
-    {
-      "epoch": 0.49792664761061234,
-      "grad_norm": 0.000617999816313386,
-      "learning_rate": 0.00019999987844072621,
-      "loss": 46.0,
-      "step": 3092
-    },
-    {
-      "epoch": 0.49808768468939973,
-      "grad_norm": 0.0005839301156811416,
-      "learning_rate": 0.00019999987836183004,
-      "loss": 46.0,
-      "step": 3093
-    },
-    {
-      "epoch": 0.49824872176818713,
-      "grad_norm": 0.0005078520043753088,
-      "learning_rate": 0.0001999998782829083,
-      "loss": 46.0,
-      "step": 3094
-    },
-    {
-      "epoch": 0.4984097588469745,
-      "grad_norm": 0.0005395420012064278,
-      "learning_rate": 0.000199999878203961,
-      "loss": 46.0,
-      "step": 3095
-    },
-    {
-      "epoch": 0.4985707959257619,
-      "grad_norm": 0.000586939335335046,
-      "learning_rate": 0.00019999987812498803,
-      "loss": 46.0,
-      "step": 3096
-    },
-    {
-      "epoch": 0.4987318330045493,
-      "grad_norm": 0.0004774358822032809,
-      "learning_rate": 0.0001999998780459895,
-      "loss": 46.0,
-      "step": 3097
-    },
-    {
-      "epoch": 0.4988928700833367,
-      "grad_norm": 0.0005348059348762035,
-      "learning_rate": 0.00019999987796696534,
-      "loss": 46.0,
-      "step": 3098
-    },
-    {
-      "epoch": 0.49905390716212406,
-      "grad_norm": 0.0002666429500095546,
-      "learning_rate": 0.00019999987788791565,
-      "loss": 46.0,
-      "step": 3099
-    },
-    {
-      "epoch": 0.49921494424091145,
-      "grad_norm": 0.0008934899815358222,
-      "learning_rate": 0.00019999987780884034,
-      "loss": 46.0,
-      "step": 3100
-    },
-    {
-      "epoch": 0.49937598131969885,
-      "grad_norm": 0.0008154212846420705,
-      "learning_rate": 0.00019999987772973942,
-      "loss": 46.0,
-      "step": 3101
-    },
-    {
-      "epoch": 0.49953701839848624,
-      "grad_norm": 0.0005573169910348952,
-      "learning_rate": 0.0001999998776506129,
-      "loss": 46.0,
-      "step": 3102
-    },
-    {
-      "epoch": 0.49969805547727364,
-      "grad_norm": 0.00031111351563595235,
-      "learning_rate": 0.0001999998775714608,
-      "loss": 46.0,
-      "step": 3103
-    },
-    {
-      "epoch": 0.49985909255606104,
-      "grad_norm": 0.0006274578045122325,
-      "learning_rate": 0.0001999998774922831,
-      "loss": 46.0,
-      "step": 3104
-    },
-    {
-      "epoch": 0.5000201296348484,
-      "grad_norm": 0.00036200587055645883,
-      "learning_rate": 0.00019999987741307985,
-      "loss": 46.0,
-      "step": 3105
-    },
-    {
-      "epoch": 0.5001811667136358,
-      "grad_norm": 0.0010207627201452851,
-      "learning_rate": 0.00019999987733385096,
-      "loss": 46.0,
-      "step": 3106
-    },
-    {
-      "epoch": 0.5003422037924232,
-      "grad_norm": 0.0009090094245038927,
-      "learning_rate": 0.0001999998772545965,
-      "loss": 46.0,
-      "step": 3107
-    },
-    {
-      "epoch": 0.5005032408712106,
-      "grad_norm": 0.0009704349213279784,
-      "learning_rate": 0.0001999998771753164,
-      "loss": 46.0,
-      "step": 3108
-    },
-    {
-      "epoch": 0.500664277949998,
-      "grad_norm": 0.0020493874326348305,
-      "learning_rate": 0.00019999987709601072,
-      "loss": 46.0,
-      "step": 3109
-    },
-    {
-      "epoch": 0.5008253150287854,
-      "grad_norm": 0.0015200057532638311,
-      "learning_rate": 0.00019999987701667946,
-      "loss": 46.0,
-      "step": 3110
-    },
-    {
-      "epoch": 0.5009863521075728,
-      "grad_norm": 0.0005671605467796326,
-      "learning_rate": 0.00019999987693732258,
-      "loss": 46.0,
-      "step": 3111
-    },
-    {
-      "epoch": 0.5011473891863601,
-      "grad_norm": 0.0006468218052759767,
-      "learning_rate": 0.00019999987685794015,
-      "loss": 46.0,
-      "step": 3112
-    },
-    {
-      "epoch": 0.5013084262651476,
-      "grad_norm": 0.0005289054824970663,
-      "learning_rate": 0.0001999998767785321,
-      "loss": 46.0,
-      "step": 3113
-    },
-    {
-      "epoch": 0.5014694633439349,
-      "grad_norm": 0.000521992624271661,
-      "learning_rate": 0.00019999987669909846,
-      "loss": 46.0,
-      "step": 3114
-    },
-    {
-      "epoch": 0.5016305004227223,
-      "grad_norm": 0.0007570886518806219,
-      "learning_rate": 0.0001999998766196392,
-      "loss": 46.0,
-      "step": 3115
-    },
-    {
-      "epoch": 0.5017915375015097,
-      "grad_norm": 0.0005296626477502286,
-      "learning_rate": 0.00019999987654015436,
-      "loss": 46.0,
-      "step": 3116
-    },
-    {
-      "epoch": 0.5019525745802971,
-      "grad_norm": 0.0005140362773090601,
-      "learning_rate": 0.00019999987646064396,
-      "loss": 46.0,
-      "step": 3117
-    },
-    {
-      "epoch": 0.5021136116590845,
-      "grad_norm": 0.0008891879697330296,
-      "learning_rate": 0.00019999987638110792,
-      "loss": 46.0,
-      "step": 3118
-    },
-    {
-      "epoch": 0.5022746487378719,
-      "grad_norm": 0.00046948957606218755,
-      "learning_rate": 0.0001999998763015463,
-      "loss": 46.0,
-      "step": 3119
-    },
-    {
-      "epoch": 0.5024356858166593,
-      "grad_norm": 0.0006360865663737059,
-      "learning_rate": 0.0001999998762219591,
-      "loss": 46.0,
-      "step": 3120
-    },
-    {
-      "epoch": 0.5025967228954467,
-      "grad_norm": 0.0005822768434882164,
-      "learning_rate": 0.0001999998761423463,
-      "loss": 46.0,
-      "step": 3121
-    },
-    {
-      "epoch": 0.5027577599742341,
-      "grad_norm": 0.0007643274147994816,
-      "learning_rate": 0.00019999987606270789,
-      "loss": 46.0,
-      "step": 3122
-    },
-    {
-      "epoch": 0.5029187970530214,
-      "grad_norm": 0.0021854920778423548,
-      "learning_rate": 0.00019999987598304388,
-      "loss": 46.0,
-      "step": 3123
-    },
-    {
-      "epoch": 0.5030798341318089,
-      "grad_norm": 0.0004112437309231609,
-      "learning_rate": 0.0001999998759033543,
-      "loss": 46.0,
-      "step": 3124
-    },
-    {
-      "epoch": 0.5032408712105962,
-      "grad_norm": 0.0009620593045838177,
-      "learning_rate": 0.00019999987582363914,
-      "loss": 46.0,
-      "step": 3125
-    },
-    {
-      "epoch": 0.5034019082893836,
-      "grad_norm": 0.0005368393030948937,
-      "learning_rate": 0.00019999987574389835,
-      "loss": 46.0,
-      "step": 3126
-    },
-    {
-      "epoch": 0.503562945368171,
-      "grad_norm": 0.0003768324968405068,
-      "learning_rate": 0.00019999987566413197,
-      "loss": 46.0,
-      "step": 3127
-    },
-    {
-      "epoch": 0.5037239824469584,
-      "grad_norm": 0.001006724894978106,
-      "learning_rate": 0.00019999987558434,
-      "loss": 46.0,
-      "step": 3128
-    },
-    {
-      "epoch": 0.5038850195257458,
-      "grad_norm": 0.0004461338394321501,
-      "learning_rate": 0.00019999987550452242,
-      "loss": 46.0,
-      "step": 3129
-    },
-    {
-      "epoch": 0.5040460566045332,
-      "grad_norm": 0.0007448829710483551,
-      "learning_rate": 0.00019999987542467927,
-      "loss": 46.0,
-      "step": 3130
-    },
-    {
-      "epoch": 0.5042070936833206,
-      "grad_norm": 0.0013202362461015582,
-      "learning_rate": 0.00019999987534481052,
-      "loss": 46.0,
-      "step": 3131
-    },
-    {
-      "epoch": 0.504368130762108,
-      "grad_norm": 0.0020129114855080843,
-      "learning_rate": 0.00019999987526491615,
-      "loss": 46.0,
-      "step": 3132
-    },
-    {
-      "epoch": 0.5045291678408954,
-      "grad_norm": 0.000902953848708421,
-      "learning_rate": 0.00019999987518499622,
-      "loss": 46.0,
-      "step": 3133
-    },
-    {
-      "epoch": 0.5046902049196827,
-      "grad_norm": 0.00033976256963796914,
-      "learning_rate": 0.00019999987510505067,
-      "loss": 46.0,
-      "step": 3134
-    },
-    {
-      "epoch": 0.5048512419984702,
-      "grad_norm": 0.0006377190002240241,
-      "learning_rate": 0.00019999987502507954,
-      "loss": 46.0,
-      "step": 3135
-    },
-    {
-      "epoch": 0.5050122790772575,
-      "grad_norm": 0.0003616051108110696,
-      "learning_rate": 0.0001999998749450828,
-      "loss": 46.0,
-      "step": 3136
-    },
-    {
-      "epoch": 0.5051733161560449,
-      "grad_norm": 0.00026360744959674776,
-      "learning_rate": 0.0001999998748650605,
-      "loss": 46.0,
-      "step": 3137
-    },
-    {
-      "epoch": 0.5053343532348323,
-      "grad_norm": 0.0005565241444855928,
-      "learning_rate": 0.00019999987478501257,
-      "loss": 46.0,
-      "step": 3138
-    },
-    {
-      "epoch": 0.5054953903136197,
-      "grad_norm": 0.00028414541156962514,
-      "learning_rate": 0.00019999987470493906,
-      "loss": 46.0,
-      "step": 3139
-    },
-    {
-      "epoch": 0.5056564273924071,
-      "grad_norm": 0.0004502649244386703,
-      "learning_rate": 0.00019999987462483994,
-      "loss": 46.0,
-      "step": 3140
-    },
-    {
-      "epoch": 0.5058174644711945,
-      "grad_norm": 0.0003624082892201841,
-      "learning_rate": 0.00019999987454471523,
-      "loss": 46.0,
-      "step": 3141
-    },
-    {
-      "epoch": 0.5059785015499819,
-      "grad_norm": 0.0014462817925959826,
-      "learning_rate": 0.00019999987446456494,
-      "loss": 46.0,
-      "step": 3142
-    },
-    {
-      "epoch": 0.5061395386287693,
-      "grad_norm": 0.0007111776503734291,
-      "learning_rate": 0.00019999987438438902,
-      "loss": 46.0,
-      "step": 3143
-    },
-    {
-      "epoch": 0.5063005757075567,
-      "grad_norm": 0.00025288472534157336,
-      "learning_rate": 0.00019999987430418753,
-      "loss": 46.0,
-      "step": 3144
-    },
-    {
-      "epoch": 0.506461612786344,
-      "grad_norm": 0.0018452502554282546,
-      "learning_rate": 0.00019999987422396047,
-      "loss": 46.0,
-      "step": 3145
-    },
-    {
-      "epoch": 0.5066226498651314,
-      "grad_norm": 0.0004520209622569382,
-      "learning_rate": 0.00019999987414370777,
-      "loss": 46.0,
-      "step": 3146
-    },
-    {
-      "epoch": 0.5067836869439188,
-      "grad_norm": 0.0011336015304550529,
-      "learning_rate": 0.0001999998740634295,
-      "loss": 46.0,
-      "step": 3147
-    },
-    {
-      "epoch": 0.5069447240227062,
-      "grad_norm": 0.0006521551986224949,
-      "learning_rate": 0.00019999987398312567,
-      "loss": 46.0,
-      "step": 3148
-    },
-    {
-      "epoch": 0.5071057611014936,
-      "grad_norm": 0.0005709989927709103,
-      "learning_rate": 0.00019999987390279618,
-      "loss": 46.0,
-      "step": 3149
-    },
-    {
-      "epoch": 0.507266798180281,
-      "grad_norm": 0.00048226548824459314,
-      "learning_rate": 0.00019999987382244113,
-      "loss": 46.0,
-      "step": 3150
-    },
-    {
-      "epoch": 0.5074278352590684,
-      "grad_norm": 0.00076228694524616,
-      "learning_rate": 0.0001999998737420605,
-      "loss": 46.0,
-      "step": 3151
-    },
-    {
-      "epoch": 0.5075888723378558,
-      "grad_norm": 0.0005003261612728238,
-      "learning_rate": 0.00019999987366165422,
-      "loss": 46.0,
-      "step": 3152
-    },
-    {
-      "epoch": 0.5077499094166432,
-      "grad_norm": 0.0019028366077691317,
-      "learning_rate": 0.00019999987358122238,
-      "loss": 46.0,
-      "step": 3153
-    },
-    {
-      "epoch": 0.5079109464954306,
-      "grad_norm": 0.00042662062332965434,
-      "learning_rate": 0.00019999987350076493,
-      "loss": 46.0,
-      "step": 3154
-    },
-    {
-      "epoch": 0.508071983574218,
-      "grad_norm": 0.0017045423155650496,
-      "learning_rate": 0.00019999987342028192,
-      "loss": 46.0,
-      "step": 3155
-    },
-    {
-      "epoch": 0.5082330206530054,
-      "grad_norm": 0.0005582640296779573,
-      "learning_rate": 0.0001999998733397733,
-      "loss": 46.0,
-      "step": 3156
-    },
-    {
-      "epoch": 0.5083940577317927,
-      "grad_norm": 0.0018368320306763053,
-      "learning_rate": 0.00019999987325923906,
-      "loss": 46.0,
-      "step": 3157
-    },
-    {
-      "epoch": 0.5085550948105801,
-      "grad_norm": 0.0006485346821136773,
-      "learning_rate": 0.00019999987317867926,
-      "loss": 46.0,
-      "step": 3158
-    },
-    {
-      "epoch": 0.5087161318893675,
-      "grad_norm": 0.000958534306846559,
-      "learning_rate": 0.00019999987309809382,
-      "loss": 46.0,
-      "step": 3159
-    },
-    {
-      "epoch": 0.5088771689681549,
-      "grad_norm": 0.0006661777733825147,
-      "learning_rate": 0.00019999987301748284,
-      "loss": 46.0,
-      "step": 3160
-    },
-    {
-      "epoch": 0.5090382060469423,
-      "grad_norm": 0.0002982582664117217,
-      "learning_rate": 0.00019999987293684623,
-      "loss": 46.0,
-      "step": 3161
-    },
-    {
-      "epoch": 0.5091992431257297,
-      "grad_norm": 0.0004241916467435658,
-      "learning_rate": 0.00019999987285618405,
-      "loss": 46.0,
-      "step": 3162
-    },
-    {
-      "epoch": 0.5093602802045171,
-      "grad_norm": 0.000682660611346364,
-      "learning_rate": 0.00019999987277549624,
-      "loss": 46.0,
-      "step": 3163
-    },
-    {
-      "epoch": 0.5095213172833045,
-      "grad_norm": 0.0004575554048642516,
-      "learning_rate": 0.00019999987269478286,
-      "loss": 46.0,
-      "step": 3164
-    },
-    {
-      "epoch": 0.5096823543620919,
-      "grad_norm": 0.0004946583649143577,
-      "learning_rate": 0.00019999987261404387,
-      "loss": 46.0,
-      "step": 3165
-    },
-    {
-      "epoch": 0.5098433914408793,
-      "grad_norm": 0.00042522800504229963,
-      "learning_rate": 0.00019999987253327932,
-      "loss": 46.0,
-      "step": 3166
-    },
-    {
-      "epoch": 0.5100044285196667,
-      "grad_norm": 0.002267090603709221,
-      "learning_rate": 0.00019999987245248915,
-      "loss": 46.0,
-      "step": 3167
-    },
-    {
-      "epoch": 0.510165465598454,
-      "grad_norm": 0.0002657459699548781,
-      "learning_rate": 0.00019999987237167337,
-      "loss": 46.0,
-      "step": 3168
-    },
-    {
-      "epoch": 0.5103265026772414,
-      "grad_norm": 0.00038011313881725073,
-      "learning_rate": 0.000199999872290832,
-      "loss": 46.0,
-      "step": 3169
-    },
-    {
-      "epoch": 0.5104875397560288,
-      "grad_norm": 0.00034682228579185903,
-      "learning_rate": 0.00019999987220996505,
-      "loss": 46.0,
-      "step": 3170
-    },
-    {
-      "epoch": 0.5106485768348162,
-      "grad_norm": 0.0005717247840948403,
-      "learning_rate": 0.0001999998721290725,
-      "loss": 46.0,
-      "step": 3171
-    },
-    {
-      "epoch": 0.5108096139136036,
-      "grad_norm": 0.0006464279722422361,
-      "learning_rate": 0.00019999987204815436,
-      "loss": 46.0,
-      "step": 3172
-    },
-    {
-      "epoch": 0.510970650992391,
-      "grad_norm": 0.0006448480999097228,
-      "learning_rate": 0.00019999987196721062,
-      "loss": 46.0,
-      "step": 3173
-    },
-    {
-      "epoch": 0.5111316880711784,
-      "grad_norm": 0.0004309078212827444,
-      "learning_rate": 0.00019999987188624129,
-      "loss": 46.0,
-      "step": 3174
-    },
-    {
-      "epoch": 0.5112927251499658,
-      "grad_norm": 0.0004590658063534647,
-      "learning_rate": 0.00019999987180524634,
-      "loss": 46.0,
-      "step": 3175
-    },
-    {
-      "epoch": 0.5114537622287532,
-      "grad_norm": 0.00035977925290353596,
-      "learning_rate": 0.00019999987172422584,
-      "loss": 46.0,
-      "step": 3176
-    },
-    {
-      "epoch": 0.5116147993075406,
-      "grad_norm": 0.0006298014195635915,
-      "learning_rate": 0.00019999987164317972,
-      "loss": 46.0,
-      "step": 3177
-    },
-    {
-      "epoch": 0.511775836386328,
-      "grad_norm": 0.0014333116123452783,
-      "learning_rate": 0.000199999871562108,
-      "loss": 46.0,
-      "step": 3178
-    },
-    {
-      "epoch": 0.5119368734651153,
-      "grad_norm": 0.0005512171774171293,
-      "learning_rate": 0.0001999998714810107,
-      "loss": 46.0,
-      "step": 3179
-    },
-    {
-      "epoch": 0.5120979105439027,
-      "grad_norm": 0.00156623893417418,
-      "learning_rate": 0.00019999987139988782,
-      "loss": 46.0,
-      "step": 3180
-    },
-    {
-      "epoch": 0.5122589476226901,
-      "grad_norm": 0.00032754268613643944,
-      "learning_rate": 0.0001999998713187393,
-      "loss": 46.0,
-      "step": 3181
-    },
-    {
-      "epoch": 0.5124199847014775,
-      "grad_norm": 0.0008100890554487705,
-      "learning_rate": 0.0001999998712375652,
-      "loss": 46.0,
-      "step": 3182
-    },
-    {
-      "epoch": 0.5125810217802649,
-      "grad_norm": 0.000321648403769359,
-      "learning_rate": 0.00019999987115636554,
-      "loss": 46.0,
-      "step": 3183
-    },
-    {
-      "epoch": 0.5127420588590523,
-      "grad_norm": 0.0003590634441934526,
-      "learning_rate": 0.00019999987107514024,
-      "loss": 46.0,
-      "step": 3184
-    },
-    {
-      "epoch": 0.5129030959378397,
-      "grad_norm": 0.0005317406030371785,
-      "learning_rate": 0.00019999987099388937,
-      "loss": 46.0,
-      "step": 3185
-    },
-    {
-      "epoch": 0.5130641330166271,
-      "grad_norm": 0.001840433687902987,
-      "learning_rate": 0.0001999998709126129,
-      "loss": 46.0,
-      "step": 3186
-    },
-    {
-      "epoch": 0.5132251700954145,
-      "grad_norm": 0.0013532418524846435,
-      "learning_rate": 0.00019999987083131084,
-      "loss": 46.0,
-      "step": 3187
-    },
-    {
-      "epoch": 0.5133862071742019,
-      "grad_norm": 0.0003129248507320881,
-      "learning_rate": 0.00019999987074998315,
-      "loss": 46.0,
-      "step": 3188
-    },
-    {
-      "epoch": 0.5135472442529893,
-      "grad_norm": 0.0009054663823917508,
-      "learning_rate": 0.0001999998706686299,
-      "loss": 46.0,
-      "step": 3189
-    },
-    {
-      "epoch": 0.5137082813317766,
-      "grad_norm": 0.003508378518745303,
-      "learning_rate": 0.00019999987058725105,
-      "loss": 46.0,
-      "step": 3190
-    },
-    {
-      "epoch": 0.513869318410564,
-      "grad_norm": 0.0005252668634057045,
-      "learning_rate": 0.0001999998705058466,
-      "loss": 46.0,
-      "step": 3191
-    },
-    {
-      "epoch": 0.5140303554893514,
-      "grad_norm": 0.0011838222853839397,
-      "learning_rate": 0.00019999987042441657,
-      "loss": 46.0,
-      "step": 3192
-    },
-    {
-      "epoch": 0.5141913925681388,
-      "grad_norm": 0.0006155192386358976,
-      "learning_rate": 0.00019999987034296094,
-      "loss": 46.0,
-      "step": 3193
-    },
-    {
-      "epoch": 0.5143524296469262,
-      "grad_norm": 0.0009439602144993842,
-      "learning_rate": 0.0001999998702614797,
-      "loss": 46.0,
-      "step": 3194
-    },
-    {
-      "epoch": 0.5145134667257136,
-      "grad_norm": 0.00026224739849567413,
-      "learning_rate": 0.00019999987017997289,
-      "loss": 46.0,
-      "step": 3195
-    },
-    {
-      "epoch": 0.514674503804501,
-      "grad_norm": 0.001287496299482882,
-      "learning_rate": 0.00019999987009844045,
-      "loss": 46.0,
-      "step": 3196
-    },
-    {
-      "epoch": 0.5148355408832884,
-      "grad_norm": 0.0007224658620543778,
-      "learning_rate": 0.00019999987001688245,
-      "loss": 46.0,
-      "step": 3197
-    },
-    {
-      "epoch": 0.5149965779620758,
-      "grad_norm": 0.0005090706399641931,
-      "learning_rate": 0.0001999998699352988,
-      "loss": 46.0,
-      "step": 3198
-    },
-    {
-      "epoch": 0.5151576150408631,
-      "grad_norm": 0.0020414614118635654,
-      "learning_rate": 0.00019999986985368964,
-      "loss": 46.0,
-      "step": 3199
-    },
-    {
-      "epoch": 0.5153186521196506,
-      "grad_norm": 0.00012341848923824728,
-      "learning_rate": 0.00019999986977205483,
-      "loss": 46.0,
-      "step": 3200
-    },
-    {
-      "epoch": 0.5154796891984379,
-      "grad_norm": 0.0013383296318352222,
-      "learning_rate": 0.00019999986969039443,
-      "loss": 46.0,
-      "step": 3201
-    },
-    {
-      "epoch": 0.5156407262772253,
-      "grad_norm": 0.0005204256158322096,
-      "learning_rate": 0.00019999986960870844,
-      "loss": 46.0,
-      "step": 3202
-    },
-    {
-      "epoch": 0.5158017633560127,
-      "grad_norm": 0.0005140191060490906,
-      "learning_rate": 0.00019999986952699684,
-      "loss": 46.0,
-      "step": 3203
-    },
-    {
-      "epoch": 0.5159628004348001,
-      "grad_norm": 0.001439216430298984,
-      "learning_rate": 0.0001999998694452597,
-      "loss": 46.0,
-      "step": 3204
-    },
-    {
-      "epoch": 0.5161238375135875,
-      "grad_norm": 0.0003279148950241506,
-      "learning_rate": 0.00019999986936349693,
-      "loss": 46.0,
-      "step": 3205
-    },
-    {
-      "epoch": 0.5162848745923749,
-      "grad_norm": 0.000539019179996103,
-      "learning_rate": 0.00019999986928170854,
-      "loss": 46.0,
-      "step": 3206
-    },
-    {
-      "epoch": 0.5164459116711623,
-      "grad_norm": 0.0005979888956062496,
-      "learning_rate": 0.0001999998691998946,
-      "loss": 46.0,
-      "step": 3207
-    },
-    {
-      "epoch": 0.5166069487499497,
-      "grad_norm": 0.0006228465936146677,
-      "learning_rate": 0.00019999986911805505,
-      "loss": 46.0,
-      "step": 3208
-    },
-    {
-      "epoch": 0.5167679858287371,
-      "grad_norm": 0.0012169572291895747,
-      "learning_rate": 0.00019999986903618985,
-      "loss": 46.0,
-      "step": 3209
-    },
-    {
-      "epoch": 0.5169290229075244,
-      "grad_norm": 0.00021819921676069498,
-      "learning_rate": 0.0001999998689542991,
-      "loss": 46.0,
-      "step": 3210
-    },
-    {
-      "epoch": 0.5170900599863119,
-      "grad_norm": 0.00023319286992773414,
-      "learning_rate": 0.00019999986887238278,
-      "loss": 46.0,
-      "step": 3211
-    },
-    {
-      "epoch": 0.5172510970650992,
-      "grad_norm": 0.00025865319184958935,
-      "learning_rate": 0.00019999986879044084,
-      "loss": 46.0,
-      "step": 3212
-    },
-    {
-      "epoch": 0.5174121341438866,
-      "grad_norm": 0.0011366064427420497,
-      "learning_rate": 0.00019999986870847331,
-      "loss": 46.0,
-      "step": 3213
-    },
-    {
-      "epoch": 0.517573171222674,
-      "grad_norm": 0.0016338558634743094,
-      "learning_rate": 0.00019999986862648017,
-      "loss": 46.0,
-      "step": 3214
-    },
-    {
-      "epoch": 0.5177342083014614,
-      "grad_norm": 0.0006788591272197664,
-      "learning_rate": 0.00019999986854446147,
-      "loss": 46.0,
-      "step": 3215
-    },
-    {
-      "epoch": 0.5178952453802488,
-      "grad_norm": 0.0004349009250290692,
-      "learning_rate": 0.00019999986846241713,
-      "loss": 46.0,
-      "step": 3216
-    },
-    {
-      "epoch": 0.5180562824590362,
-      "grad_norm": 0.0006168187246657908,
-      "learning_rate": 0.00019999986838034722,
-      "loss": 46.0,
-      "step": 3217
-    },
-    {
-      "epoch": 0.5182173195378236,
-      "grad_norm": 0.0015551609685644507,
-      "learning_rate": 0.00019999986829825173,
-      "loss": 46.0,
-      "step": 3218
-    },
-    {
-      "epoch": 0.518378356616611,
-      "grad_norm": 0.0014264252968132496,
-      "learning_rate": 0.00019999986821613063,
-      "loss": 46.0,
-      "step": 3219
-    },
-    {
-      "epoch": 0.5185393936953984,
-      "grad_norm": 0.0006923124892637134,
-      "learning_rate": 0.00019999986813398394,
-      "loss": 46.0,
-      "step": 3220
-    },
-    {
-      "epoch": 0.5187004307741857,
-      "grad_norm": 0.0003963627968914807,
-      "learning_rate": 0.00019999986805181163,
-      "loss": 46.0,
-      "step": 3221
-    },
-    {
-      "epoch": 0.5188614678529732,
-      "grad_norm": 0.0020369815174490213,
-      "learning_rate": 0.00019999986796961374,
-      "loss": 46.0,
-      "step": 3222
-    },
-    {
-      "epoch": 0.5190225049317605,
-      "grad_norm": 0.0008632676908746362,
-      "learning_rate": 0.00019999986788739026,
-      "loss": 46.0,
-      "step": 3223
-    },
-    {
-      "epoch": 0.519183542010548,
-      "grad_norm": 0.001770115690305829,
-      "learning_rate": 0.0001999998678051412,
-      "loss": 46.0,
-      "step": 3224
-    },
-    {
-      "epoch": 0.5193445790893353,
-      "grad_norm": 0.0012730071321129799,
-      "learning_rate": 0.00019999986772286654,
-      "loss": 46.0,
-      "step": 3225
-    },
-    {
-      "epoch": 0.5195056161681227,
-      "grad_norm": 0.0003461350570432842,
-      "learning_rate": 0.00019999986764056627,
-      "loss": 46.0,
-      "step": 3226
-    },
-    {
-      "epoch": 0.5196666532469101,
-      "grad_norm": 0.0003870897344313562,
-      "learning_rate": 0.0001999998675582404,
-      "loss": 46.0,
-      "step": 3227
-    },
-    {
-      "epoch": 0.5198276903256975,
-      "grad_norm": 0.0012966131325811148,
-      "learning_rate": 0.00019999986747588894,
-      "loss": 46.0,
-      "step": 3228
-    },
-    {
-      "epoch": 0.5199887274044849,
-      "grad_norm": 0.002284219255670905,
-      "learning_rate": 0.0001999998673935119,
-      "loss": 46.0,
-      "step": 3229
-    },
-    {
-      "epoch": 0.5201497644832723,
-      "grad_norm": 0.0004675232048612088,
-      "learning_rate": 0.00019999986731110924,
-      "loss": 46.0,
-      "step": 3230
-    },
-    {
-      "epoch": 0.5203108015620597,
-      "grad_norm": 0.001080109621398151,
-      "learning_rate": 0.000199999867228681,
-      "loss": 46.0,
-      "step": 3231
-    },
-    {
-      "epoch": 0.520471838640847,
-      "grad_norm": 0.0012963133631274104,
-      "learning_rate": 0.00019999986714622719,
-      "loss": 46.0,
-      "step": 3232
-    },
-    {
-      "epoch": 0.5206328757196345,
-      "grad_norm": 0.0005809199064970016,
-      "learning_rate": 0.00019999986706374773,
-      "loss": 46.0,
-      "step": 3233
-    },
-    {
-      "epoch": 0.5207939127984218,
-      "grad_norm": 0.0006185378879308701,
-      "learning_rate": 0.0001999998669812427,
-      "loss": 46.0,
-      "step": 3234
-    },
-    {
-      "epoch": 0.5209549498772093,
-      "grad_norm": 0.0011855832999572158,
-      "learning_rate": 0.0001999998668987121,
-      "loss": 46.0,
-      "step": 3235
-    },
-    {
-      "epoch": 0.5211159869559966,
-      "grad_norm": 0.0011884800624102354,
-      "learning_rate": 0.00019999986681615588,
-      "loss": 46.0,
-      "step": 3236
-    },
-    {
-      "epoch": 0.521277024034784,
-      "grad_norm": 0.00159457977861166,
-      "learning_rate": 0.00019999986673357407,
-      "loss": 46.0,
-      "step": 3237
-    },
-    {
-      "epoch": 0.5214380611135714,
-      "grad_norm": 0.0015858509577810764,
-      "learning_rate": 0.00019999986665096667,
-      "loss": 46.0,
-      "step": 3238
-    },
-    {
-      "epoch": 0.5215990981923588,
-      "grad_norm": 0.0010136772179976106,
-      "learning_rate": 0.00019999986656833366,
-      "loss": 46.0,
-      "step": 3239
-    },
-    {
-      "epoch": 0.5217601352711462,
-      "grad_norm": 0.0006406564498320222,
-      "learning_rate": 0.0001999998664856751,
-      "loss": 46.0,
-      "step": 3240
-    },
-    {
-      "epoch": 0.5219211723499336,
-      "grad_norm": 0.0010428943205624819,
-      "learning_rate": 0.0001999998664029909,
-      "loss": 46.0,
-      "step": 3241
-    },
-    {
-      "epoch": 0.522082209428721,
-      "grad_norm": 0.0010522321099415421,
-      "learning_rate": 0.0001999998663202811,
-      "loss": 46.0,
-      "step": 3242
-    },
-    {
-      "epoch": 0.5222432465075083,
-      "grad_norm": 0.00025458712480030954,
-      "learning_rate": 0.00019999986623754574,
-      "loss": 46.0,
-      "step": 3243
-    },
-    {
-      "epoch": 0.5224042835862958,
-      "grad_norm": 0.0012281094677746296,
-      "learning_rate": 0.0001999998661547848,
-      "loss": 46.0,
-      "step": 3244
-    },
-    {
-      "epoch": 0.5225653206650831,
-      "grad_norm": 0.001638661720789969,
-      "learning_rate": 0.0001999998660719982,
-      "loss": 46.0,
-      "step": 3245
-    },
-    {
-      "epoch": 0.5227263577438706,
-      "grad_norm": 0.0005527132307179272,
-      "learning_rate": 0.00019999986598918606,
-      "loss": 46.0,
-      "step": 3246
-    },
-    {
-      "epoch": 0.5228873948226579,
-      "grad_norm": 0.00024619928444735706,
-      "learning_rate": 0.00019999986590634827,
-      "loss": 46.0,
-      "step": 3247
-    },
-    {
-      "epoch": 0.5230484319014453,
-      "grad_norm": 0.000301736785331741,
-      "learning_rate": 0.00019999986582348492,
-      "loss": 46.0,
-      "step": 3248
-    },
-    {
-      "epoch": 0.5232094689802327,
-      "grad_norm": 0.0031265667639672756,
-      "learning_rate": 0.000199999865740596,
-      "loss": 46.0,
-      "step": 3249
-    },
-    {
-      "epoch": 0.5233705060590201,
-      "grad_norm": 0.0005060398252680898,
-      "learning_rate": 0.00019999986565768146,
-      "loss": 46.0,
-      "step": 3250
-    },
-    {
-      "epoch": 0.5235315431378075,
-      "grad_norm": 0.0008124821470119059,
-      "learning_rate": 0.0001999998655747413,
-      "loss": 46.0,
-      "step": 3251
-    },
-    {
-      "epoch": 0.5236925802165948,
-      "grad_norm": 0.0007707438198849559,
-      "learning_rate": 0.00019999986549177557,
-      "loss": 46.0,
-      "step": 3252
-    },
-    {
-      "epoch": 0.5238536172953823,
-      "grad_norm": 0.0006837815162725747,
-      "learning_rate": 0.00019999986540878426,
-      "loss": 46.0,
-      "step": 3253
-    },
-    {
-      "epoch": 0.5240146543741696,
-      "grad_norm": 0.0007320574368350208,
-      "learning_rate": 0.00019999986532576733,
-      "loss": 46.0,
-      "step": 3254
-    },
-    {
-      "epoch": 0.5241756914529571,
-      "grad_norm": 0.00021528542856685817,
-      "learning_rate": 0.0001999998652427248,
-      "loss": 46.0,
-      "step": 3255
-    },
-    {
-      "epoch": 0.5243367285317444,
-      "grad_norm": 0.0006836784305050969,
-      "learning_rate": 0.00019999986515965672,
-      "loss": 46.0,
-      "step": 3256
-    },
-    {
-      "epoch": 0.5244977656105319,
-      "grad_norm": 0.0013585075503215194,
-      "learning_rate": 0.000199999865076563,
-      "loss": 46.0,
-      "step": 3257
-    },
-    {
-      "epoch": 0.5246588026893192,
-      "grad_norm": 0.0006389369955286384,
-      "learning_rate": 0.00019999986499344373,
-      "loss": 46.0,
-      "step": 3258
-    },
-    {
-      "epoch": 0.5248198397681066,
-      "grad_norm": 0.00033288082340732217,
-      "learning_rate": 0.0001999998649102988,
-      "loss": 46.0,
-      "step": 3259
-    },
-    {
-      "epoch": 0.524980876846894,
-      "grad_norm": 0.002165345475077629,
-      "learning_rate": 0.0001999998648271283,
-      "loss": 46.0,
-      "step": 3260
-    },
-    {
-      "epoch": 0.5251419139256814,
-      "grad_norm": 0.0005821749218739569,
-      "learning_rate": 0.00019999986474393224,
-      "loss": 46.0,
-      "step": 3261
-    },
-    {
-      "epoch": 0.5253029510044688,
-      "grad_norm": 0.0004455771413631737,
-      "learning_rate": 0.00019999986466071054,
-      "loss": 46.0,
-      "step": 3262
-    },
-    {
-      "epoch": 0.5254639880832561,
-      "grad_norm": 0.0023695342242717743,
-      "learning_rate": 0.00019999986457746327,
-      "loss": 46.0,
-      "step": 3263
-    },
-    {
-      "epoch": 0.5256250251620436,
-      "grad_norm": 0.0007352940738201141,
-      "learning_rate": 0.00019999986449419042,
-      "loss": 46.0,
-      "step": 3264
-    },
-    {
-      "epoch": 0.5257860622408309,
-      "grad_norm": 0.0006059968145564198,
-      "learning_rate": 0.00019999986441089193,
-      "loss": 46.0,
-      "step": 3265
-    },
-    {
-      "epoch": 0.5259470993196184,
-      "grad_norm": 0.0005157870473340154,
-      "learning_rate": 0.00019999986432756787,
-      "loss": 46.0,
-      "step": 3266
-    },
-    {
-      "epoch": 0.5261081363984057,
-      "grad_norm": 0.00043126571108587086,
-      "learning_rate": 0.00019999986424421823,
-      "loss": 46.0,
-      "step": 3267
-    },
-    {
-      "epoch": 0.5262691734771932,
-      "grad_norm": 0.0004135149938520044,
-      "learning_rate": 0.00019999986416084298,
-      "loss": 46.0,
-      "step": 3268
-    },
-    {
-      "epoch": 0.5264302105559805,
-      "grad_norm": 0.0007121902890503407,
-      "learning_rate": 0.00019999986407744213,
-      "loss": 46.0,
-      "step": 3269
-    },
-    {
-      "epoch": 0.526591247634768,
-      "grad_norm": 0.00033011010964401066,
-      "learning_rate": 0.0001999998639940157,
-      "loss": 46.0,
-      "step": 3270
-    },
-    {
-      "epoch": 0.5267522847135553,
-      "grad_norm": 0.001519715297035873,
-      "learning_rate": 0.00019999986391056366,
-      "loss": 46.0,
-      "step": 3271
-    },
-    {
-      "epoch": 0.5269133217923427,
-      "grad_norm": 0.00042124453466385603,
-      "learning_rate": 0.00019999986382708605,
-      "loss": 46.0,
-      "step": 3272
-    },
-    {
-      "epoch": 0.5270743588711301,
-      "grad_norm": 0.0007642715936526656,
-      "learning_rate": 0.00019999986374358284,
-      "loss": 46.0,
-      "step": 3273
-    },
-    {
-      "epoch": 0.5272353959499174,
-      "grad_norm": 0.0008970482740551233,
-      "learning_rate": 0.000199999863660054,
-      "loss": 46.0,
-      "step": 3274
-    },
-    {
-      "epoch": 0.5273964330287049,
-      "grad_norm": 0.0010133379837498069,
-      "learning_rate": 0.0001999998635764996,
-      "loss": 46.0,
-      "step": 3275
-    },
-    {
-      "epoch": 0.5275574701074922,
-      "grad_norm": 0.00031207629945129156,
-      "learning_rate": 0.0001999998634929196,
-      "loss": 46.0,
-      "step": 3276
-    },
-    {
-      "epoch": 0.5277185071862797,
-      "grad_norm": 0.0002669797686394304,
-      "learning_rate": 0.000199999863409314,
-      "loss": 46.0,
-      "step": 3277
-    },
-    {
-      "epoch": 0.527879544265067,
-      "grad_norm": 0.0014513595961034298,
-      "learning_rate": 0.0001999998633256828,
-      "loss": 46.0,
-      "step": 3278
-    },
-    {
-      "epoch": 0.5280405813438545,
-      "grad_norm": 0.0011736878659576178,
-      "learning_rate": 0.000199999863242026,
-      "loss": 46.0,
-      "step": 3279
-    },
-    {
-      "epoch": 0.5282016184226418,
-      "grad_norm": 0.0012734222691506147,
-      "learning_rate": 0.0001999998631583436,
-      "loss": 46.0,
-      "step": 3280
-    },
-    {
-      "epoch": 0.5283626555014292,
-      "grad_norm": 0.0006362651474773884,
-      "learning_rate": 0.00019999986307463563,
-      "loss": 46.0,
-      "step": 3281
-    },
-    {
-      "epoch": 0.5285236925802166,
-      "grad_norm": 0.0008188053616322577,
-      "learning_rate": 0.00019999986299090204,
-      "loss": 46.0,
-      "step": 3282
-    },
-    {
-      "epoch": 0.528684729659004,
-      "grad_norm": 0.00035949761513620615,
-      "learning_rate": 0.0001999998629071429,
-      "loss": 46.0,
-      "step": 3283
-    },
-    {
-      "epoch": 0.5288457667377914,
-      "grad_norm": 0.0005487981252372265,
-      "learning_rate": 0.0001999998628233581,
-      "loss": 46.0,
-      "step": 3284
-    },
-    {
-      "epoch": 0.5290068038165787,
-      "grad_norm": 0.001972783589735627,
-      "learning_rate": 0.00019999986273954776,
-      "loss": 46.0,
-      "step": 3285
-    },
-    {
-      "epoch": 0.5291678408953662,
-      "grad_norm": 0.0007641033735126257,
-      "learning_rate": 0.0001999998626557118,
-      "loss": 46.0,
-      "step": 3286
-    },
-    {
-      "epoch": 0.5293288779741535,
-      "grad_norm": 0.0013937512412667274,
-      "learning_rate": 0.00019999986257185026,
-      "loss": 46.0,
-      "step": 3287
-    },
-    {
-      "epoch": 0.529489915052941,
-      "grad_norm": 0.0006470530643127859,
-      "learning_rate": 0.0001999998624879631,
-      "loss": 46.0,
-      "step": 3288
-    },
-    {
-      "epoch": 0.5296509521317283,
-      "grad_norm": 0.000434192072134465,
-      "learning_rate": 0.00019999986240405037,
-      "loss": 46.0,
-      "step": 3289
-    },
-    {
-      "epoch": 0.5298119892105158,
-      "grad_norm": 0.0006054903496988118,
-      "learning_rate": 0.00019999986232011204,
-      "loss": 46.0,
-      "step": 3290
-    },
-    {
-      "epoch": 0.5299730262893031,
-      "grad_norm": 0.00040821536094881594,
-      "learning_rate": 0.00019999986223614812,
-      "loss": 46.0,
-      "step": 3291
-    },
-    {
-      "epoch": 0.5301340633680905,
-      "grad_norm": 0.0009323583799414337,
-      "learning_rate": 0.00019999986215215858,
-      "loss": 46.0,
-      "step": 3292
-    },
-    {
-      "epoch": 0.5302951004468779,
-      "grad_norm": 0.0008833401370793581,
-      "learning_rate": 0.00019999986206814345,
-      "loss": 46.0,
-      "step": 3293
-    },
-    {
-      "epoch": 0.5304561375256653,
-      "grad_norm": 0.0007340603042393923,
-      "learning_rate": 0.00019999986198410277,
-      "loss": 46.0,
-      "step": 3294
-    },
-    {
-      "epoch": 0.5306171746044527,
-      "grad_norm": 0.003398546949028969,
-      "learning_rate": 0.00019999986190003644,
-      "loss": 46.0,
-      "step": 3295
-    },
-    {
-      "epoch": 0.53077821168324,
-      "grad_norm": 0.0012400518171489239,
-      "learning_rate": 0.00019999986181594453,
-      "loss": 46.0,
-      "step": 3296
-    },
-    {
-      "epoch": 0.5309392487620275,
-      "grad_norm": 0.00046752297203056514,
-      "learning_rate": 0.00019999986173182706,
-      "loss": 46.0,
-      "step": 3297
-    },
-    {
-      "epoch": 0.5311002858408148,
-      "grad_norm": 0.000741161871701479,
-      "learning_rate": 0.00019999986164768397,
-      "loss": 46.0,
-      "step": 3298
-    },
-    {
-      "epoch": 0.5312613229196023,
-      "grad_norm": 0.0015461597358807921,
-      "learning_rate": 0.00019999986156351524,
-      "loss": 46.0,
-      "step": 3299
-    },
-    {
-      "epoch": 0.5314223599983896,
-      "grad_norm": 0.00042216136353090405,
-      "learning_rate": 0.00019999986147932098,
-      "loss": 46.0,
-      "step": 3300
-    },
-    {
-      "epoch": 0.5315833970771771,
-      "grad_norm": 0.000270228018052876,
-      "learning_rate": 0.0001999998613951011,
-      "loss": 46.0,
-      "step": 3301
-    },
-    {
-      "epoch": 0.5317444341559644,
-      "grad_norm": 0.00042081010178662837,
-      "learning_rate": 0.0001999998613108556,
-      "loss": 46.0,
-      "step": 3302
-    },
-    {
-      "epoch": 0.5319054712347518,
-      "grad_norm": 0.0019501334754750133,
-      "learning_rate": 0.00019999986122658456,
-      "loss": 46.0,
-      "step": 3303
-    },
-    {
-      "epoch": 0.5320665083135392,
-      "grad_norm": 0.000719082192517817,
-      "learning_rate": 0.00019999986114228787,
-      "loss": 46.0,
-      "step": 3304
-    },
-    {
-      "epoch": 0.5322275453923265,
-      "grad_norm": 0.00044744223123416305,
-      "learning_rate": 0.00019999986105796565,
-      "loss": 46.0,
-      "step": 3305
-    },
-    {
-      "epoch": 0.532388582471114,
-      "grad_norm": 0.0014701862819492817,
-      "learning_rate": 0.00019999986097361778,
-      "loss": 46.0,
-      "step": 3306
-    },
-    {
-      "epoch": 0.5325496195499013,
-      "grad_norm": 0.0014685611240565777,
-      "learning_rate": 0.00019999986088924433,
-      "loss": 46.0,
-      "step": 3307
-    },
-    {
-      "epoch": 0.5327106566286888,
-      "grad_norm": 0.0005893473862670362,
-      "learning_rate": 0.00019999986080484529,
-      "loss": 46.0,
-      "step": 3308
-    },
-    {
-      "epoch": 0.5328716937074761,
-      "grad_norm": 0.0017547404859215021,
-      "learning_rate": 0.00019999986072042068,
-      "loss": 46.0,
-      "step": 3309
-    },
-    {
-      "epoch": 0.5330327307862636,
-      "grad_norm": 0.0003632925800047815,
-      "learning_rate": 0.00019999986063597042,
-      "loss": 46.0,
-      "step": 3310
-    },
-    {
-      "epoch": 0.5331937678650509,
-      "grad_norm": 0.0006466899649240077,
-      "learning_rate": 0.0001999998605514946,
-      "loss": 46.0,
-      "step": 3311
-    },
-    {
-      "epoch": 0.5333548049438384,
-      "grad_norm": 0.002408044645562768,
-      "learning_rate": 0.00019999986046699317,
-      "loss": 46.0,
-      "step": 3312
-    },
-    {
-      "epoch": 0.5335158420226257,
-      "grad_norm": 0.0004881194035988301,
-      "learning_rate": 0.00019999986038246617,
-      "loss": 46.0,
-      "step": 3313
-    },
-    {
-      "epoch": 0.5336768791014131,
-      "grad_norm": 0.0006727608852088451,
-      "learning_rate": 0.00019999986029791355,
-      "loss": 46.0,
-      "step": 3314
-    },
-    {
-      "epoch": 0.5338379161802005,
-      "grad_norm": 0.0008111980860121548,
-      "learning_rate": 0.00019999986021333537,
-      "loss": 46.0,
-      "step": 3315
-    },
-    {
-      "epoch": 0.5339989532589878,
-      "grad_norm": 0.0002524749143049121,
-      "learning_rate": 0.00019999986012873155,
-      "loss": 46.0,
-      "step": 3316
-    },
-    {
-      "epoch": 0.5341599903377753,
-      "grad_norm": 0.001971939578652382,
-      "learning_rate": 0.00019999986004410217,
-      "loss": 46.0,
-      "step": 3317
-    },
-    {
-      "epoch": 0.5343210274165626,
-      "grad_norm": 0.002227312419563532,
-      "learning_rate": 0.00019999985995944718,
-      "loss": 46.0,
-      "step": 3318
-    },
-    {
-      "epoch": 0.5344820644953501,
-      "grad_norm": 0.0003603290824685246,
-      "learning_rate": 0.0001999998598747666,
-      "loss": 46.0,
-      "step": 3319
-    },
-    {
-      "epoch": 0.5346431015741374,
-      "grad_norm": 0.0008477758965454996,
-      "learning_rate": 0.0001999998597900604,
-      "loss": 46.0,
-      "step": 3320
-    },
-    {
-      "epoch": 0.5348041386529249,
-      "grad_norm": 0.00035513032344169915,
-      "learning_rate": 0.00019999985970532862,
-      "loss": 46.0,
-      "step": 3321
-    },
-    {
-      "epoch": 0.5349651757317122,
-      "grad_norm": 0.0005001723184250295,
-      "learning_rate": 0.00019999985962057125,
-      "loss": 46.0,
-      "step": 3322
-    },
-    {
-      "epoch": 0.5351262128104997,
-      "grad_norm": 0.0004099126090295613,
-      "learning_rate": 0.00019999985953578827,
-      "loss": 46.0,
-      "step": 3323
-    },
-    {
-      "epoch": 0.535287249889287,
-      "grad_norm": 0.0016978277126327157,
-      "learning_rate": 0.00019999985945097973,
-      "loss": 46.0,
-      "step": 3324
-    },
-    {
-      "epoch": 0.5354482869680744,
-      "grad_norm": 0.0008843669784255326,
-      "learning_rate": 0.0001999998593661456,
-      "loss": 46.0,
-      "step": 3325
-    },
-    {
-      "epoch": 0.5356093240468618,
-      "grad_norm": 0.0014790837885811925,
-      "learning_rate": 0.00019999985928128585,
-      "loss": 46.0,
-      "step": 3326
-    },
-    {
-      "epoch": 0.5357703611256491,
-      "grad_norm": 0.00046818848932161927,
-      "learning_rate": 0.0001999998591964005,
-      "loss": 46.0,
-      "step": 3327
-    },
-    {
-      "epoch": 0.5359313982044366,
-      "grad_norm": 0.000849020027089864,
-      "learning_rate": 0.00019999985911148954,
-      "loss": 46.0,
-      "step": 3328
-    },
-    {
-      "epoch": 0.5360924352832239,
-      "grad_norm": 0.0014687919756397605,
-      "learning_rate": 0.000199999859026553,
-      "loss": 46.0,
-      "step": 3329
-    },
-    {
-      "epoch": 0.5362534723620114,
-      "grad_norm": 0.00035368630778975785,
-      "learning_rate": 0.0001999998589415909,
-      "loss": 46.0,
-      "step": 3330
-    },
-    {
-      "epoch": 0.5364145094407987,
-      "grad_norm": 0.0006590307457372546,
-      "learning_rate": 0.00019999985885660318,
-      "loss": 46.0,
-      "step": 3331
-    },
-    {
-      "epoch": 0.5365755465195862,
-      "grad_norm": 0.001852670218795538,
-      "learning_rate": 0.00019999985877158983,
-      "loss": 46.0,
-      "step": 3332
-    },
-    {
-      "epoch": 0.5367365835983735,
-      "grad_norm": 0.0005677458830177784,
-      "learning_rate": 0.00019999985868655092,
-      "loss": 46.0,
-      "step": 3333
-    },
-    {
-      "epoch": 0.536897620677161,
-      "grad_norm": 0.000881219923030585,
-      "learning_rate": 0.00019999985860148645,
-      "loss": 46.0,
-      "step": 3334
-    },
-    {
-      "epoch": 0.5370586577559483,
-      "grad_norm": 0.0004763822944369167,
-      "learning_rate": 0.00019999985851639634,
-      "loss": 46.0,
-      "step": 3335
-    },
-    {
-      "epoch": 0.5372196948347358,
-      "grad_norm": 0.0012226671678945422,
-      "learning_rate": 0.00019999985843128064,
-      "loss": 46.0,
-      "step": 3336
-    },
-    {
-      "epoch": 0.5373807319135231,
-      "grad_norm": 0.001961508998647332,
-      "learning_rate": 0.00019999985834613936,
-      "loss": 46.0,
-      "step": 3337
-    },
-    {
-      "epoch": 0.5375417689923104,
-      "grad_norm": 0.001269041677005589,
-      "learning_rate": 0.00019999985826097246,
-      "loss": 46.0,
-      "step": 3338
-    },
-    {
-      "epoch": 0.5377028060710979,
-      "grad_norm": 0.002586194546893239,
-      "learning_rate": 0.00019999985817577997,
-      "loss": 46.0,
-      "step": 3339
-    },
-    {
-      "epoch": 0.5378638431498852,
-      "grad_norm": 0.0011924192076548934,
-      "learning_rate": 0.0001999998580905619,
-      "loss": 46.0,
-      "step": 3340
-    },
-    {
-      "epoch": 0.5380248802286727,
-      "grad_norm": 0.0016669401666149497,
-      "learning_rate": 0.0001999998580053182,
-      "loss": 46.0,
-      "step": 3341
-    },
-    {
-      "epoch": 0.53818591730746,
-      "grad_norm": 0.0013443110510706902,
-      "learning_rate": 0.00019999985792004896,
-      "loss": 46.0,
-      "step": 3342
-    },
-    {
-      "epoch": 0.5383469543862475,
-      "grad_norm": 0.000323209329508245,
-      "learning_rate": 0.0001999998578347541,
-      "loss": 46.0,
-      "step": 3343
-    },
-    {
-      "epoch": 0.5385079914650348,
-      "grad_norm": 0.0014050803147256374,
-      "learning_rate": 0.00019999985774943365,
-      "loss": 46.0,
-      "step": 3344
-    },
-    {
-      "epoch": 0.5386690285438223,
-      "grad_norm": 0.0005120033165439963,
-      "learning_rate": 0.00019999985766408758,
-      "loss": 46.0,
-      "step": 3345
-    },
-    {
-      "epoch": 0.5388300656226096,
-      "grad_norm": 0.0016074093291535974,
-      "learning_rate": 0.00019999985757871596,
-      "loss": 46.0,
-      "step": 3346
-    },
-    {
-      "epoch": 0.538991102701397,
-      "grad_norm": 0.0015373494243249297,
-      "learning_rate": 0.00019999985749331872,
-      "loss": 46.0,
-      "step": 3347
-    },
-    {
-      "epoch": 0.5391521397801844,
-      "grad_norm": 0.0005608833744190633,
-      "learning_rate": 0.0001999998574078959,
-      "loss": 46.0,
-      "step": 3348
-    },
-    {
-      "epoch": 0.5393131768589717,
-      "grad_norm": 0.0008417388307861984,
-      "learning_rate": 0.00019999985732244743,
-      "loss": 46.0,
-      "step": 3349
-    },
-    {
-      "epoch": 0.5394742139377592,
-      "grad_norm": 0.0005155186518095434,
-      "learning_rate": 0.0001999998572369734,
-      "loss": 46.0,
-      "step": 3350
-    },
-    {
-      "epoch": 0.5396352510165465,
-      "grad_norm": 0.0020404248498380184,
-      "learning_rate": 0.0001999998571514738,
-      "loss": 46.0,
-      "step": 3351
-    },
-    {
-      "epoch": 0.539796288095334,
-      "grad_norm": 0.001458734623156488,
-      "learning_rate": 0.00019999985706594856,
-      "loss": 46.0,
-      "step": 3352
-    },
-    {
-      "epoch": 0.5399573251741213,
-      "grad_norm": 0.0004648598260246217,
-      "learning_rate": 0.00019999985698039777,
-      "loss": 46.0,
-      "step": 3353
-    },
-    {
-      "epoch": 0.5401183622529088,
-      "grad_norm": 0.00084132474148646,
-      "learning_rate": 0.00019999985689482137,
-      "loss": 46.0,
-      "step": 3354
-    },
-    {
-      "epoch": 0.5402793993316961,
-      "grad_norm": 0.0010218317620456219,
-      "learning_rate": 0.00019999985680921938,
-      "loss": 46.0,
-      "step": 3355
-    },
-    {
-      "epoch": 0.5404404364104836,
-      "grad_norm": 0.0008872864418663085,
-      "learning_rate": 0.00019999985672359178,
-      "loss": 46.0,
-      "step": 3356
-    },
-    {
-      "epoch": 0.5406014734892709,
-      "grad_norm": 0.001526973326690495,
-      "learning_rate": 0.00019999985663793858,
-      "loss": 46.0,
-      "step": 3357
-    },
-    {
-      "epoch": 0.5407625105680582,
-      "grad_norm": 0.0005437143845483661,
-      "learning_rate": 0.0001999998565522598,
-      "loss": 46.0,
-      "step": 3358
-    },
-    {
-      "epoch": 0.5409235476468457,
-      "grad_norm": 0.002817622385919094,
-      "learning_rate": 0.00019999985646655544,
-      "loss": 46.0,
-      "step": 3359
-    },
-    {
-      "epoch": 0.541084584725633,
-      "grad_norm": 0.0006917702266946435,
-      "learning_rate": 0.00019999985638082546,
-      "loss": 46.0,
-      "step": 3360
-    },
-    {
-      "epoch": 0.5412456218044205,
-      "grad_norm": 0.0003476533165667206,
-      "learning_rate": 0.00019999985629506987,
-      "loss": 46.0,
-      "step": 3361
-    },
-    {
-      "epoch": 0.5414066588832078,
-      "grad_norm": 0.0006579328910447657,
-      "learning_rate": 0.0001999998562092887,
-      "loss": 46.0,
-      "step": 3362
-    },
-    {
-      "epoch": 0.5415676959619953,
-      "grad_norm": 0.0011675585992634296,
-      "learning_rate": 0.00019999985612348197,
-      "loss": 46.0,
-      "step": 3363
-    },
-    {
-      "epoch": 0.5417287330407826,
-      "grad_norm": 0.000983578385785222,
-      "learning_rate": 0.00019999985603764961,
-      "loss": 46.0,
-      "step": 3364
-    },
-    {
-      "epoch": 0.5418897701195701,
-      "grad_norm": 0.0011690407991409302,
-      "learning_rate": 0.00019999985595179167,
-      "loss": 46.0,
-      "step": 3365
-    },
-    {
-      "epoch": 0.5420508071983574,
-      "grad_norm": 0.001205615932121873,
-      "learning_rate": 0.00019999985586590811,
-      "loss": 46.0,
-      "step": 3366
-    },
-    {
-      "epoch": 0.5422118442771449,
-      "grad_norm": 0.0004708147025667131,
-      "learning_rate": 0.000199999855779999,
-      "loss": 46.0,
-      "step": 3367
-    },
-    {
-      "epoch": 0.5423728813559322,
-      "grad_norm": 0.0011563218431547284,
-      "learning_rate": 0.00019999985569406426,
-      "loss": 46.0,
-      "step": 3368
-    },
-    {
-      "epoch": 0.5425339184347195,
-      "grad_norm": 0.0012656064936891198,
-      "learning_rate": 0.00019999985560810392,
-      "loss": 46.0,
-      "step": 3369
-    },
-    {
-      "epoch": 0.542694955513507,
-      "grad_norm": 0.0007580489618703723,
-      "learning_rate": 0.000199999855522118,
-      "loss": 46.0,
-      "step": 3370
-    },
-    {
-      "epoch": 0.5428559925922943,
-      "grad_norm": 0.0025987382978200912,
-      "learning_rate": 0.0001999998554361065,
-      "loss": 46.0,
-      "step": 3371
-    },
-    {
-      "epoch": 0.5430170296710818,
-      "grad_norm": 0.000891050323843956,
-      "learning_rate": 0.00019999985535006939,
-      "loss": 46.0,
-      "step": 3372
-    },
-    {
-      "epoch": 0.5431780667498691,
-      "grad_norm": 0.0007159209926612675,
-      "learning_rate": 0.0001999998552640067,
-      "loss": 46.0,
-      "step": 3373
-    },
-    {
-      "epoch": 0.5433391038286566,
-      "grad_norm": 0.0004322503518778831,
-      "learning_rate": 0.00019999985517791838,
-      "loss": 46.0,
-      "step": 3374
-    },
-    {
-      "epoch": 0.5435001409074439,
-      "grad_norm": 0.0007701522554270923,
-      "learning_rate": 0.00019999985509180449,
-      "loss": 46.0,
-      "step": 3375
-    },
-    {
-      "epoch": 0.5436611779862314,
-      "grad_norm": 0.0008689900278113782,
-      "learning_rate": 0.00019999985500566498,
-      "loss": 46.0,
-      "step": 3376
-    },
-    {
-      "epoch": 0.5438222150650187,
-      "grad_norm": 0.00033185360371135175,
-      "learning_rate": 0.0001999998549194999,
-      "loss": 46.0,
-      "step": 3377
-    },
-    {
-      "epoch": 0.5439832521438062,
-      "grad_norm": 0.000633304298389703,
-      "learning_rate": 0.00019999985483330922,
-      "loss": 46.0,
-      "step": 3378
-    },
-    {
-      "epoch": 0.5441442892225935,
-      "grad_norm": 0.0004575739149004221,
-      "learning_rate": 0.00019999985474709298,
-      "loss": 46.0,
-      "step": 3379
-    },
-    {
-      "epoch": 0.5443053263013808,
-      "grad_norm": 0.0004649245529435575,
-      "learning_rate": 0.00019999985466085106,
-      "loss": 46.0,
-      "step": 3380
-    },
-    {
-      "epoch": 0.5444663633801683,
-      "grad_norm": 0.000492118822876364,
-      "learning_rate": 0.00019999985457458364,
-      "loss": 46.0,
-      "step": 3381
-    },
-    {
-      "epoch": 0.5446274004589556,
-      "grad_norm": 0.0006083305343054235,
-      "learning_rate": 0.00019999985448829058,
-      "loss": 46.0,
-      "step": 3382
-    },
-    {
-      "epoch": 0.5447884375377431,
-      "grad_norm": 0.001091295969672501,
-      "learning_rate": 0.0001999998544019719,
-      "loss": 46.0,
-      "step": 3383
-    },
-    {
-      "epoch": 0.5449494746165304,
-      "grad_norm": 0.00028013280825689435,
-      "learning_rate": 0.00019999985431562767,
-      "loss": 46.0,
-      "step": 3384
-    },
-    {
-      "epoch": 0.5451105116953179,
-      "grad_norm": 0.0005511103081516922,
-      "learning_rate": 0.00019999985422925783,
-      "loss": 46.0,
-      "step": 3385
-    },
-    {
-      "epoch": 0.5452715487741052,
-      "grad_norm": 0.000784877862315625,
-      "learning_rate": 0.00019999985414286236,
-      "loss": 46.0,
-      "step": 3386
-    },
-    {
-      "epoch": 0.5454325858528927,
-      "grad_norm": 0.00091799400979653,
-      "learning_rate": 0.00019999985405644134,
-      "loss": 46.0,
-      "step": 3387
-    },
-    {
-      "epoch": 0.54559362293168,
-      "grad_norm": 0.0020278797019273043,
-      "learning_rate": 0.00019999985396999473,
-      "loss": 46.0,
-      "step": 3388
-    },
-    {
-      "epoch": 0.5457546600104675,
-      "grad_norm": 0.002359907841309905,
-      "learning_rate": 0.0001999998538835225,
-      "loss": 46.0,
-      "step": 3389
-    },
-    {
-      "epoch": 0.5459156970892548,
-      "grad_norm": 0.002854179823771119,
-      "learning_rate": 0.0001999998537970247,
-      "loss": 46.0,
-      "step": 3390
-    },
-    {
-      "epoch": 0.5460767341680421,
-      "grad_norm": 0.0015953172696754336,
-      "learning_rate": 0.00019999985371050127,
-      "loss": 46.0,
-      "step": 3391
-    },
-    {
-      "epoch": 0.5462377712468296,
-      "grad_norm": 0.0018778997473418713,
-      "learning_rate": 0.00019999985362395225,
-      "loss": 46.0,
-      "step": 3392
-    },
-    {
-      "epoch": 0.5463988083256169,
-      "grad_norm": 0.0003722886322066188,
-      "learning_rate": 0.00019999985353737768,
-      "loss": 46.0,
-      "step": 3393
-    },
-    {
-      "epoch": 0.5465598454044044,
-      "grad_norm": 0.0006690325681120157,
-      "learning_rate": 0.00019999985345077746,
-      "loss": 46.0,
-      "step": 3394
-    },
-    {
-      "epoch": 0.5467208824831917,
-      "grad_norm": 0.0020559511613100767,
-      "learning_rate": 0.00019999985336415166,
-      "loss": 46.0,
-      "step": 3395
-    },
-    {
-      "epoch": 0.5468819195619792,
-      "grad_norm": 0.00043758476385846734,
-      "learning_rate": 0.00019999985327750027,
-      "loss": 46.0,
-      "step": 3396
-    },
-    {
-      "epoch": 0.5470429566407665,
-      "grad_norm": 0.00037907183286733925,
-      "learning_rate": 0.0001999998531908233,
-      "loss": 46.0,
-      "step": 3397
-    },
-    {
-      "epoch": 0.547203993719554,
-      "grad_norm": 0.0006615728489123285,
-      "learning_rate": 0.00019999985310412073,
-      "loss": 46.0,
-      "step": 3398
-    },
-    {
-      "epoch": 0.5473650307983413,
-      "grad_norm": 0.00042850023601204157,
-      "learning_rate": 0.00019999985301739256,
-      "loss": 46.0,
-      "step": 3399
-    },
-    {
-      "epoch": 0.5475260678771288,
-      "grad_norm": 0.00036770780570805073,
-      "learning_rate": 0.0001999998529306388,
-      "loss": 46.0,
-      "step": 3400
-    },
-    {
-      "epoch": 0.5476871049559161,
-      "grad_norm": 0.0006704438710585237,
-      "learning_rate": 0.00019999985284385944,
-      "loss": 46.0,
-      "step": 3401
-    },
-    {
-      "epoch": 0.5478481420347034,
-      "grad_norm": 0.0007254494121298194,
-      "learning_rate": 0.00019999985275705447,
-      "loss": 46.0,
-      "step": 3402
-    },
-    {
-      "epoch": 0.5480091791134909,
-      "grad_norm": 0.00037542221252806485,
-      "learning_rate": 0.00019999985267022392,
-      "loss": 46.0,
-      "step": 3403
-    },
-    {
-      "epoch": 0.5481702161922782,
-      "grad_norm": 0.0014997926773503423,
-      "learning_rate": 0.00019999985258336778,
-      "loss": 46.0,
-      "step": 3404
-    },
-    {
-      "epoch": 0.5483312532710657,
-      "grad_norm": 0.0007830832619220018,
-      "learning_rate": 0.00019999985249648602,
-      "loss": 46.0,
-      "step": 3405
-    },
-    {
-      "epoch": 0.548492290349853,
-      "grad_norm": 0.000924643303733319,
-      "learning_rate": 0.0001999998524095787,
-      "loss": 46.0,
-      "step": 3406
-    },
-    {
-      "epoch": 0.5486533274286405,
-      "grad_norm": 0.000445937184849754,
-      "learning_rate": 0.00019999985232264575,
-      "loss": 46.0,
-      "step": 3407
-    },
-    {
-      "epoch": 0.5488143645074278,
-      "grad_norm": 0.0025469991378486156,
-      "learning_rate": 0.00019999985223568723,
-      "loss": 46.0,
-      "step": 3408
-    },
-    {
-      "epoch": 0.5489754015862153,
-      "grad_norm": 0.0009231626754626632,
-      "learning_rate": 0.0001999998521487031,
-      "loss": 46.0,
-      "step": 3409
-    },
-    {
-      "epoch": 0.5491364386650026,
-      "grad_norm": 0.0008242886979132891,
-      "learning_rate": 0.0001999998520616934,
-      "loss": 46.0,
-      "step": 3410
-    },
-    {
-      "epoch": 0.54929747574379,
-      "grad_norm": 0.0005769507843069732,
-      "learning_rate": 0.00019999985197465808,
-      "loss": 46.0,
-      "step": 3411
-    },
-    {
-      "epoch": 0.5494585128225774,
-      "grad_norm": 0.00034231721656396985,
-      "learning_rate": 0.0001999998518875972,
-      "loss": 46.0,
-      "step": 3412
-    },
-    {
-      "epoch": 0.5496195499013647,
-      "grad_norm": 0.00037883000914007425,
-      "learning_rate": 0.0001999998518005107,
-      "loss": 46.0,
-      "step": 3413
-    },
-    {
-      "epoch": 0.5497805869801522,
-      "grad_norm": 0.0007620594697073102,
-      "learning_rate": 0.00019999985171339861,
-      "loss": 46.0,
-      "step": 3414
-    },
-    {
-      "epoch": 0.5499416240589395,
-      "grad_norm": 0.00036850827746093273,
-      "learning_rate": 0.0001999998516262609,
-      "loss": 46.0,
-      "step": 3415
-    },
-    {
-      "epoch": 0.550102661137727,
-      "grad_norm": 0.0005711701815016568,
-      "learning_rate": 0.00019999985153909764,
-      "loss": 46.0,
-      "step": 3416
-    },
-    {
-      "epoch": 0.5502636982165143,
-      "grad_norm": 0.0005954890511929989,
-      "learning_rate": 0.00019999985145190876,
-      "loss": 46.0,
-      "step": 3417
-    },
-    {
-      "epoch": 0.5504247352953018,
-      "grad_norm": 0.0002647766668815166,
-      "learning_rate": 0.00019999985136469429,
-      "loss": 46.0,
-      "step": 3418
-    },
-    {
-      "epoch": 0.5505857723740891,
-      "grad_norm": 0.0007883899961598217,
-      "learning_rate": 0.00019999985127745423,
-      "loss": 46.0,
-      "step": 3419
-    },
-    {
-      "epoch": 0.5507468094528766,
-      "grad_norm": 0.0006732104811817408,
-      "learning_rate": 0.00019999985119018856,
-      "loss": 46.0,
-      "step": 3420
-    },
-    {
-      "epoch": 0.5509078465316639,
-      "grad_norm": 0.0014344204682856798,
-      "learning_rate": 0.00019999985110289727,
-      "loss": 46.0,
-      "step": 3421
-    },
-    {
-      "epoch": 0.5510688836104513,
-      "grad_norm": 0.0004768389044329524,
-      "learning_rate": 0.00019999985101558045,
-      "loss": 46.0,
-      "step": 3422
-    },
-    {
-      "epoch": 0.5512299206892387,
-      "grad_norm": 0.0003428664058446884,
-      "learning_rate": 0.000199999850928238,
-      "loss": 46.0,
-      "step": 3423
-    },
-    {
-      "epoch": 0.551390957768026,
-      "grad_norm": 0.0006917240680195391,
-      "learning_rate": 0.00019999985084086995,
-      "loss": 46.0,
-      "step": 3424
-    },
-    {
-      "epoch": 0.5515519948468135,
-      "grad_norm": 0.00025905275833792984,
-      "learning_rate": 0.00019999985075347631,
-      "loss": 46.0,
-      "step": 3425
-    },
-    {
-      "epoch": 0.5517130319256008,
-      "grad_norm": 0.0005896909278817475,
-      "learning_rate": 0.00019999985066605707,
-      "loss": 46.0,
-      "step": 3426
-    },
-    {
-      "epoch": 0.5518740690043883,
-      "grad_norm": 0.0004898286424577236,
-      "learning_rate": 0.00019999985057861226,
-      "loss": 46.0,
-      "step": 3427
-    },
-    {
-      "epoch": 0.5520351060831756,
-      "grad_norm": 0.0007520309882238507,
-      "learning_rate": 0.00019999985049114183,
-      "loss": 46.0,
-      "step": 3428
-    },
-    {
-      "epoch": 0.5521961431619631,
-      "grad_norm": 0.0006808222969993949,
-      "learning_rate": 0.00019999985040364583,
-      "loss": 46.0,
-      "step": 3429
-    },
-    {
-      "epoch": 0.5523571802407504,
-      "grad_norm": 0.0019085080130025744,
-      "learning_rate": 0.0001999998503161242,
-      "loss": 46.0,
-      "step": 3430
-    },
-    {
-      "epoch": 0.5525182173195379,
-      "grad_norm": 0.0005711892154067755,
-      "learning_rate": 0.00019999985022857702,
-      "loss": 46.0,
-      "step": 3431
-    },
-    {
-      "epoch": 0.5526792543983252,
-      "grad_norm": 0.00027607844094745815,
-      "learning_rate": 0.0001999998501410042,
-      "loss": 46.0,
-      "step": 3432
-    },
-    {
-      "epoch": 0.5528402914771126,
-      "grad_norm": 0.0005318095209077001,
-      "learning_rate": 0.0001999998500534058,
-      "loss": 46.0,
-      "step": 3433
-    },
-    {
-      "epoch": 0.5530013285559,
-      "grad_norm": 0.0006763770361430943,
-      "learning_rate": 0.0001999998499657818,
-      "loss": 46.0,
-      "step": 3434
-    },
-    {
-      "epoch": 0.5531623656346873,
-      "grad_norm": 0.002392539056017995,
-      "learning_rate": 0.00019999984987813222,
-      "loss": 46.0,
-      "step": 3435
-    },
-    {
-      "epoch": 0.5533234027134748,
-      "grad_norm": 0.0018446936737746,
-      "learning_rate": 0.00019999984979045704,
-      "loss": 46.0,
-      "step": 3436
-    },
-    {
-      "epoch": 0.5534844397922621,
-      "grad_norm": 0.005663339979946613,
-      "learning_rate": 0.00019999984970275628,
-      "loss": 46.0,
-      "step": 3437
-    },
-    {
-      "epoch": 0.5536454768710496,
-      "grad_norm": 0.0006960775936022401,
-      "learning_rate": 0.00019999984961502988,
-      "loss": 46.0,
-      "step": 3438
-    },
-    {
-      "epoch": 0.5538065139498369,
-      "grad_norm": 0.0007040626369416714,
-      "learning_rate": 0.00019999984952727794,
-      "loss": 46.0,
-      "step": 3439
-    },
-    {
-      "epoch": 0.5539675510286244,
-      "grad_norm": 0.000625553191639483,
-      "learning_rate": 0.0001999998494395004,
-      "loss": 46.0,
-      "step": 3440
-    },
-    {
-      "epoch": 0.5541285881074117,
-      "grad_norm": 0.0009212561999447644,
-      "learning_rate": 0.00019999984935169723,
-      "loss": 46.0,
-      "step": 3441
-    },
-    {
-      "epoch": 0.5542896251861992,
-      "grad_norm": 0.00029020055080763996,
-      "learning_rate": 0.00019999984926386848,
-      "loss": 46.0,
-      "step": 3442
-    },
-    {
-      "epoch": 0.5544506622649865,
-      "grad_norm": 0.0005106495227664709,
-      "learning_rate": 0.00019999984917601414,
-      "loss": 46.0,
-      "step": 3443
-    },
-    {
-      "epoch": 0.5546116993437739,
-      "grad_norm": 0.0007982651004567742,
-      "learning_rate": 0.00019999984908813421,
-      "loss": 46.0,
-      "step": 3444
-    },
-    {
-      "epoch": 0.5547727364225613,
-      "grad_norm": 0.0014645976480096579,
-      "learning_rate": 0.00019999984900022865,
-      "loss": 46.0,
-      "step": 3445
-    },
-    {
-      "epoch": 0.5549337735013486,
-      "grad_norm": 0.0020567302126437426,
-      "learning_rate": 0.00019999984891229752,
-      "loss": 46.0,
-      "step": 3446
-    },
-    {
-      "epoch": 0.5550948105801361,
-      "grad_norm": 0.0002986452600453049,
-      "learning_rate": 0.0001999998488243408,
-      "loss": 46.0,
-      "step": 3447
-    },
-    {
-      "epoch": 0.5552558476589234,
-      "grad_norm": 0.0004691271751653403,
-      "learning_rate": 0.00019999984873635848,
-      "loss": 46.0,
-      "step": 3448
-    },
-    {
-      "epoch": 0.5554168847377109,
-      "grad_norm": 0.0008632902172394097,
-      "learning_rate": 0.00019999984864835056,
-      "loss": 46.0,
-      "step": 3449
-    },
-    {
-      "epoch": 0.5555779218164982,
-      "grad_norm": 0.000558181491214782,
-      "learning_rate": 0.00019999984856031709,
-      "loss": 46.0,
-      "step": 3450
-    },
-    {
-      "epoch": 0.5557389588952857,
-      "grad_norm": 0.001122056506574154,
-      "learning_rate": 0.00019999984847225797,
-      "loss": 46.0,
-      "step": 3451
-    },
-    {
-      "epoch": 0.555899995974073,
-      "grad_norm": 0.0004552182799670845,
-      "learning_rate": 0.00019999984838417326,
-      "loss": 46.0,
-      "step": 3452
-    },
-    {
-      "epoch": 0.5560610330528605,
-      "grad_norm": 0.0004523425595834851,
-      "learning_rate": 0.00019999984829606297,
-      "loss": 46.0,
-      "step": 3453
-    },
-    {
-      "epoch": 0.5562220701316478,
-      "grad_norm": 0.0008734787115827203,
-      "learning_rate": 0.0001999998482079271,
-      "loss": 46.0,
-      "step": 3454
-    },
-    {
-      "epoch": 0.5563831072104352,
-      "grad_norm": 0.0007075812900438905,
-      "learning_rate": 0.00019999984811976563,
-      "loss": 46.0,
-      "step": 3455
-    },
-    {
-      "epoch": 0.5565441442892226,
-      "grad_norm": 0.0007711488287895918,
-      "learning_rate": 0.00019999984803157852,
-      "loss": 46.0,
-      "step": 3456
-    },
-    {
-      "epoch": 0.55670518136801,
-      "grad_norm": 0.0006179328775033355,
-      "learning_rate": 0.00019999984794336585,
-      "loss": 46.0,
-      "step": 3457
-    },
-    {
-      "epoch": 0.5568662184467974,
-      "grad_norm": 0.0018959949957206845,
-      "learning_rate": 0.0001999998478551276,
-      "loss": 46.0,
-      "step": 3458
-    },
-    {
-      "epoch": 0.5570272555255847,
-      "grad_norm": 0.0007022268255241215,
-      "learning_rate": 0.00019999984776686373,
-      "loss": 46.0,
-      "step": 3459
-    },
-    {
-      "epoch": 0.5571882926043722,
-      "grad_norm": 0.0006481180316768587,
-      "learning_rate": 0.00019999984767857427,
-      "loss": 46.0,
-      "step": 3460
-    },
-    {
-      "epoch": 0.5573493296831595,
-      "grad_norm": 0.0015879544662311673,
-      "learning_rate": 0.00019999984759025923,
-      "loss": 46.0,
-      "step": 3461
-    },
-    {
-      "epoch": 0.557510366761947,
-      "grad_norm": 0.0003428160853218287,
-      "learning_rate": 0.00019999984750191857,
-      "loss": 46.0,
-      "step": 3462
-    },
-    {
-      "epoch": 0.5576714038407343,
-      "grad_norm": 0.000689804379362613,
-      "learning_rate": 0.00019999984741355236,
-      "loss": 46.0,
-      "step": 3463
-    },
-    {
-      "epoch": 0.5578324409195217,
-      "grad_norm": 0.0007661819108761847,
-      "learning_rate": 0.0001999998473251605,
-      "loss": 46.0,
-      "step": 3464
-    },
-    {
-      "epoch": 0.5579934779983091,
-      "grad_norm": 0.002222961513325572,
-      "learning_rate": 0.00019999984723674308,
-      "loss": 46.0,
-      "step": 3465
-    },
-    {
-      "epoch": 0.5581545150770965,
-      "grad_norm": 0.0003630129504017532,
-      "learning_rate": 0.00019999984714830004,
-      "loss": 46.0,
-      "step": 3466
-    },
-    {
-      "epoch": 0.5583155521558839,
-      "grad_norm": 0.0004200133844278753,
-      "learning_rate": 0.00019999984705983142,
-      "loss": 46.0,
-      "step": 3467
-    },
-    {
-      "epoch": 0.5584765892346713,
-      "grad_norm": 0.0006303141708485782,
-      "learning_rate": 0.0001999998469713372,
-      "loss": 46.0,
-      "step": 3468
-    },
-    {
-      "epoch": 0.5586376263134587,
-      "grad_norm": 0.0006812093779444695,
-      "learning_rate": 0.00019999984688281742,
-      "loss": 46.0,
-      "step": 3469
-    },
-    {
-      "epoch": 0.558798663392246,
-      "grad_norm": 0.0017715313006192446,
-      "learning_rate": 0.00019999984679427198,
-      "loss": 46.0,
-      "step": 3470
-    },
-    {
-      "epoch": 0.5589597004710335,
-      "grad_norm": 0.0007096350891515613,
-      "learning_rate": 0.000199999846705701,
-      "loss": 46.0,
-      "step": 3471
-    },
-    {
-      "epoch": 0.5591207375498208,
-      "grad_norm": 0.0007101345108821988,
-      "learning_rate": 0.00019999984661710443,
-      "loss": 46.0,
-      "step": 3472
-    },
-    {
-      "epoch": 0.5592817746286083,
-      "grad_norm": 0.0002391544112470001,
-      "learning_rate": 0.0001999998465284822,
-      "loss": 46.0,
-      "step": 3473
-    },
-    {
-      "epoch": 0.5594428117073956,
-      "grad_norm": 0.0005278324242681265,
-      "learning_rate": 0.00019999984643983445,
-      "loss": 46.0,
-      "step": 3474
-    },
-    {
-      "epoch": 0.559603848786183,
-      "grad_norm": 0.0004809069214388728,
-      "learning_rate": 0.00019999984635116108,
-      "loss": 46.0,
-      "step": 3475
-    },
-    {
-      "epoch": 0.5597648858649704,
-      "grad_norm": 0.001162560423836112,
-      "learning_rate": 0.0001999998462624621,
-      "loss": 46.0,
-      "step": 3476
-    },
-    {
-      "epoch": 0.5599259229437578,
-      "grad_norm": 0.0012873654486611485,
-      "learning_rate": 0.00019999984617373752,
-      "loss": 46.0,
-      "step": 3477
-    },
-    {
-      "epoch": 0.5600869600225452,
-      "grad_norm": 0.00207755109295249,
-      "learning_rate": 0.00019999984608498736,
-      "loss": 46.0,
-      "step": 3478
-    },
-    {
-      "epoch": 0.5602479971013326,
-      "grad_norm": 0.000612785283010453,
-      "learning_rate": 0.0001999998459962116,
-      "loss": 46.0,
-      "step": 3479
-    },
-    {
-      "epoch": 0.56040903418012,
-      "grad_norm": 0.0010419205063953996,
-      "learning_rate": 0.00019999984590741025,
-      "loss": 46.0,
-      "step": 3480
-    },
-    {
-      "epoch": 0.5605700712589073,
-      "grad_norm": 0.0006163626676425338,
-      "learning_rate": 0.0001999998458185833,
-      "loss": 46.0,
-      "step": 3481
-    },
-    {
-      "epoch": 0.5607311083376948,
-      "grad_norm": 0.0008884693379513919,
-      "learning_rate": 0.00019999984572973073,
-      "loss": 46.0,
-      "step": 3482
-    },
-    {
-      "epoch": 0.5608921454164821,
-      "grad_norm": 0.0009190840064547956,
-      "learning_rate": 0.0001999998456408526,
-      "loss": 46.0,
-      "step": 3483
-    },
-    {
-      "epoch": 0.5610531824952696,
-      "grad_norm": 0.0005326207028701901,
-      "learning_rate": 0.00019999984555194887,
-      "loss": 46.0,
-      "step": 3484
-    },
-    {
-      "epoch": 0.5612142195740569,
-      "grad_norm": 0.00141628657002002,
-      "learning_rate": 0.00019999984546301955,
-      "loss": 46.0,
-      "step": 3485
-    },
-    {
-      "epoch": 0.5613752566528443,
-      "grad_norm": 0.0010659357067197561,
-      "learning_rate": 0.0001999998453740646,
-      "loss": 46.0,
-      "step": 3486
-    },
-    {
-      "epoch": 0.5615362937316317,
-      "grad_norm": 0.0016936389729380608,
-      "learning_rate": 0.0001999998452850841,
-      "loss": 46.0,
-      "step": 3487
-    },
-    {
-      "epoch": 0.5616973308104191,
-      "grad_norm": 0.0022782955784350634,
-      "learning_rate": 0.00019999984519607802,
-      "loss": 46.0,
-      "step": 3488
-    },
-    {
-      "epoch": 0.5618583678892065,
-      "grad_norm": 0.0005731368437409401,
-      "learning_rate": 0.0001999998451070463,
-      "loss": 46.0,
-      "step": 3489
-    },
-    {
-      "epoch": 0.5620194049679939,
-      "grad_norm": 0.00043743624701164663,
-      "learning_rate": 0.000199999845017989,
-      "loss": 46.0,
-      "step": 3490
-    },
-    {
-      "epoch": 0.5621804420467813,
-      "grad_norm": 0.000992362853139639,
-      "learning_rate": 0.0001999998449289061,
-      "loss": 46.0,
-      "step": 3491
-    },
-    {
-      "epoch": 0.5623414791255686,
-      "grad_norm": 0.00025123724481090903,
-      "learning_rate": 0.00019999984483979761,
-      "loss": 46.0,
-      "step": 3492
-    },
-    {
-      "epoch": 0.5625025162043561,
-      "grad_norm": 0.00043814961099997163,
-      "learning_rate": 0.00019999984475066354,
-      "loss": 46.0,
-      "step": 3493
-    },
-    {
-      "epoch": 0.5626635532831434,
-      "grad_norm": 0.0011627791682258248,
-      "learning_rate": 0.00019999984466150385,
-      "loss": 46.0,
-      "step": 3494
-    },
-    {
-      "epoch": 0.5628245903619309,
-      "grad_norm": 0.0005311044515110552,
-      "learning_rate": 0.00019999984457231857,
-      "loss": 46.0,
-      "step": 3495
-    },
-    {
-      "epoch": 0.5629856274407182,
-      "grad_norm": 0.0007569093722850084,
-      "learning_rate": 0.0001999998444831077,
-      "loss": 46.0,
-      "step": 3496
-    },
-    {
-      "epoch": 0.5631466645195056,
-      "grad_norm": 0.0007590854656882584,
-      "learning_rate": 0.00019999984439387122,
-      "loss": 46.0,
-      "step": 3497
-    },
-    {
-      "epoch": 0.563307701598293,
-      "grad_norm": 0.00037324574077501893,
-      "learning_rate": 0.00019999984430460918,
-      "loss": 46.0,
-      "step": 3498
-    },
-    {
-      "epoch": 0.5634687386770804,
-      "grad_norm": 0.0006846066098660231,
-      "learning_rate": 0.00019999984421532153,
-      "loss": 46.0,
-      "step": 3499
-    },
-    {
-      "epoch": 0.5636297757558678,
-      "grad_norm": 0.0003338441310916096,
-      "learning_rate": 0.00019999984412600826,
-      "loss": 46.0,
-      "step": 3500
-    },
-    {
-      "epoch": 0.5637908128346552,
-      "grad_norm": 0.0008743259240873158,
-      "learning_rate": 0.00019999984403666943,
-      "loss": 46.0,
-      "step": 3501
-    },
-    {
-      "epoch": 0.5639518499134426,
-      "grad_norm": 0.0003305981808807701,
-      "learning_rate": 0.000199999843947305,
-      "loss": 46.0,
-      "step": 3502
-    },
-    {
-      "epoch": 0.56411288699223,
-      "grad_norm": 0.002117388416081667,
-      "learning_rate": 0.00019999984385791496,
-      "loss": 46.0,
-      "step": 3503
-    },
-    {
-      "epoch": 0.5642739240710174,
-      "grad_norm": 0.00046136858873069286,
-      "learning_rate": 0.00019999984376849935,
-      "loss": 46.0,
-      "step": 3504
-    },
-    {
-      "epoch": 0.5644349611498047,
-      "grad_norm": 0.0002929664624389261,
-      "learning_rate": 0.00019999984367905811,
-      "loss": 46.0,
-      "step": 3505
-    },
-    {
-      "epoch": 0.5645959982285922,
-      "grad_norm": 0.000627997680567205,
-      "learning_rate": 0.0001999998435895913,
-      "loss": 46.0,
-      "step": 3506
-    },
-    {
-      "epoch": 0.5647570353073795,
-      "grad_norm": 0.0005872087203897536,
-      "learning_rate": 0.0001999998435000989,
-      "loss": 46.0,
-      "step": 3507
-    },
-    {
-      "epoch": 0.5649180723861669,
-      "grad_norm": 0.0006196696776896715,
-      "learning_rate": 0.00019999984341058087,
-      "loss": 46.0,
-      "step": 3508
-    },
-    {
-      "epoch": 0.5650791094649543,
-      "grad_norm": 0.0008690227405168116,
-      "learning_rate": 0.00019999984332103726,
-      "loss": 46.0,
-      "step": 3509
-    },
-    {
-      "epoch": 0.5652401465437417,
-      "grad_norm": 0.0003540632897056639,
-      "learning_rate": 0.00019999984323146807,
-      "loss": 46.0,
-      "step": 3510
-    },
-    {
-      "epoch": 0.5654011836225291,
-      "grad_norm": 0.0020956171210855246,
-      "learning_rate": 0.0001999998431418733,
-      "loss": 46.0,
-      "step": 3511
-    },
-    {
-      "epoch": 0.5655622207013165,
-      "grad_norm": 0.0014627899508923292,
-      "learning_rate": 0.0001999998430522529,
-      "loss": 46.0,
-      "step": 3512
-    },
-    {
-      "epoch": 0.5657232577801039,
-      "grad_norm": 0.0012361931148916483,
-      "learning_rate": 0.0001999998429626069,
-      "loss": 46.0,
-      "step": 3513
-    },
-    {
-      "epoch": 0.5658842948588912,
-      "grad_norm": 0.0008946381858550012,
-      "learning_rate": 0.00019999984287293531,
-      "loss": 46.0,
-      "step": 3514
-    },
-    {
-      "epoch": 0.5660453319376787,
-      "grad_norm": 0.000591730757150799,
-      "learning_rate": 0.00019999984278323816,
-      "loss": 46.0,
-      "step": 3515
-    },
-    {
-      "epoch": 0.566206369016466,
-      "grad_norm": 0.00047976497444324195,
-      "learning_rate": 0.00019999984269351539,
-      "loss": 46.0,
-      "step": 3516
-    },
-    {
-      "epoch": 0.5663674060952534,
-      "grad_norm": 0.000979891512542963,
-      "learning_rate": 0.00019999984260376703,
-      "loss": 46.0,
-      "step": 3517
-    },
-    {
-      "epoch": 0.5665284431740408,
-      "grad_norm": 0.0014252635883167386,
-      "learning_rate": 0.00019999984251399305,
-      "loss": 46.0,
-      "step": 3518
-    },
-    {
-      "epoch": 0.5666894802528282,
-      "grad_norm": 0.0005568196647800505,
-      "learning_rate": 0.00019999984242419352,
-      "loss": 46.0,
-      "step": 3519
-    },
-    {
-      "epoch": 0.5668505173316156,
-      "grad_norm": 0.00042711684363894165,
-      "learning_rate": 0.0001999998423343684,
-      "loss": 46.0,
-      "step": 3520
-    },
-    {
-      "epoch": 0.567011554410403,
-      "grad_norm": 0.0013744635507464409,
-      "learning_rate": 0.00019999984224451764,
-      "loss": 46.0,
-      "step": 3521
-    },
-    {
-      "epoch": 0.5671725914891904,
-      "grad_norm": 0.00048566493205726147,
-      "learning_rate": 0.00019999984215464132,
-      "loss": 46.0,
-      "step": 3522
-    },
-    {
-      "epoch": 0.5673336285679778,
-      "grad_norm": 0.0003043602337129414,
-      "learning_rate": 0.00019999984206473936,
-      "loss": 46.0,
-      "step": 3523
-    },
-    {
-      "epoch": 0.5674946656467652,
-      "grad_norm": 0.004065882880240679,
-      "learning_rate": 0.00019999984197481186,
-      "loss": 46.0,
-      "step": 3524
-    },
-    {
-      "epoch": 0.5676557027255525,
-      "grad_norm": 0.0005343631491996348,
-      "learning_rate": 0.00019999984188485872,
-      "loss": 46.0,
-      "step": 3525
-    },
-    {
-      "epoch": 0.56781673980434,
-      "grad_norm": 0.0004673006769735366,
-      "learning_rate": 0.00019999984179488003,
-      "loss": 46.0,
-      "step": 3526
-    },
-    {
-      "epoch": 0.5679777768831273,
-      "grad_norm": 0.0004798337467946112,
-      "learning_rate": 0.00019999984170487571,
-      "loss": 46.0,
-      "step": 3527
-    },
-    {
-      "epoch": 0.5681388139619147,
-      "grad_norm": 0.00035520331584848464,
-      "learning_rate": 0.00019999984161484582,
-      "loss": 46.0,
-      "step": 3528
-    },
-    {
-      "epoch": 0.5682998510407021,
-      "grad_norm": 0.0007749026408419013,
-      "learning_rate": 0.0001999998415247903,
-      "loss": 46.0,
-      "step": 3529
-    },
-    {
-      "epoch": 0.5684608881194895,
-      "grad_norm": 0.0017666397616267204,
-      "learning_rate": 0.0001999998414347092,
-      "loss": 46.0,
-      "step": 3530
-    },
-    {
-      "epoch": 0.5686219251982769,
-      "grad_norm": 0.002026507630944252,
-      "learning_rate": 0.00019999984134460251,
-      "loss": 46.0,
-      "step": 3531
-    },
-    {
-      "epoch": 0.5687829622770643,
-      "grad_norm": 0.0007767747738398612,
-      "learning_rate": 0.0001999998412544702,
-      "loss": 46.0,
-      "step": 3532
-    },
-    {
-      "epoch": 0.5689439993558517,
-      "grad_norm": 0.002654020907357335,
-      "learning_rate": 0.00019999984116431235,
-      "loss": 46.0,
-      "step": 3533
-    },
-    {
-      "epoch": 0.5691050364346391,
-      "grad_norm": 0.0006586925592273474,
-      "learning_rate": 0.00019999984107412888,
-      "loss": 46.0,
-      "step": 3534
-    },
-    {
-      "epoch": 0.5692660735134265,
-      "grad_norm": 0.0027254067827016115,
-      "learning_rate": 0.0001999998409839198,
-      "loss": 46.0,
-      "step": 3535
-    },
-    {
-      "epoch": 0.5694271105922138,
-      "grad_norm": 0.0005620275624096394,
-      "learning_rate": 0.00019999984089368516,
-      "loss": 46.0,
-      "step": 3536
-    },
-    {
-      "epoch": 0.5695881476710013,
-      "grad_norm": 0.000472288578748703,
-      "learning_rate": 0.0001999998408034249,
-      "loss": 46.0,
-      "step": 3537
-    },
-    {
-      "epoch": 0.5697491847497886,
-      "grad_norm": 0.0004454654990695417,
-      "learning_rate": 0.00019999984071313905,
-      "loss": 46.0,
-      "step": 3538
-    },
-    {
-      "epoch": 0.569910221828576,
-      "grad_norm": 0.0006553643615916371,
-      "learning_rate": 0.0001999998406228276,
-      "loss": 46.0,
-      "step": 3539
-    },
-    {
-      "epoch": 0.5700712589073634,
-      "grad_norm": 0.0007052412838675082,
-      "learning_rate": 0.00019999984053249055,
-      "loss": 46.0,
-      "step": 3540
-    },
-    {
-      "epoch": 0.5702322959861508,
-      "grad_norm": 0.0002986126928590238,
-      "learning_rate": 0.00019999984044212791,
-      "loss": 46.0,
-      "step": 3541
-    },
-    {
-      "epoch": 0.5703933330649382,
-      "grad_norm": 0.0007034339942038059,
-      "learning_rate": 0.00019999984035173969,
-      "loss": 46.0,
-      "step": 3542
-    },
-    {
-      "epoch": 0.5705543701437256,
-      "grad_norm": 0.0005461368709802628,
-      "learning_rate": 0.00019999984026132584,
-      "loss": 46.0,
-      "step": 3543
-    },
-    {
-      "epoch": 0.570715407222513,
-      "grad_norm": 0.0002621538587845862,
-      "learning_rate": 0.00019999984017088642,
-      "loss": 46.0,
-      "step": 3544
-    },
-    {
-      "epoch": 0.5708764443013004,
-      "grad_norm": 0.00048391424934379756,
-      "learning_rate": 0.0001999998400804214,
-      "loss": 46.0,
-      "step": 3545
-    },
-    {
-      "epoch": 0.5710374813800878,
-      "grad_norm": 0.0005650295643135905,
-      "learning_rate": 0.00019999983998993082,
-      "loss": 46.0,
-      "step": 3546
-    },
-    {
-      "epoch": 0.5711985184588751,
-      "grad_norm": 0.0007414202555082738,
-      "learning_rate": 0.0001999998398994146,
-      "loss": 46.0,
-      "step": 3547
-    },
-    {
-      "epoch": 0.5713595555376626,
-      "grad_norm": 0.0006256731576286256,
-      "learning_rate": 0.0001999998398088728,
-      "loss": 46.0,
-      "step": 3548
-    },
-    {
-      "epoch": 0.5715205926164499,
-      "grad_norm": 0.0012972938129678369,
-      "learning_rate": 0.0001999998397183054,
-      "loss": 46.0,
-      "step": 3549
-    },
-    {
-      "epoch": 0.5716816296952373,
-      "grad_norm": 0.00039018644019961357,
-      "learning_rate": 0.00019999983962771243,
-      "loss": 46.0,
-      "step": 3550
-    },
-    {
-      "epoch": 0.5718426667740247,
-      "grad_norm": 0.0012930320808663964,
-      "learning_rate": 0.00019999983953709384,
-      "loss": 46.0,
-      "step": 3551
-    },
-    {
-      "epoch": 0.5720037038528121,
-      "grad_norm": 0.0007005230290815234,
-      "learning_rate": 0.00019999983944644966,
-      "loss": 46.0,
-      "step": 3552
-    },
-    {
-      "epoch": 0.5721647409315995,
-      "grad_norm": 0.0004933098680339754,
-      "learning_rate": 0.00019999983935577986,
-      "loss": 46.0,
-      "step": 3553
-    },
-    {
-      "epoch": 0.5723257780103869,
-      "grad_norm": 0.0012687075650319457,
-      "learning_rate": 0.0001999998392650845,
-      "loss": 46.0,
-      "step": 3554
-    },
-    {
-      "epoch": 0.5724868150891743,
-      "grad_norm": 0.00034164986573159695,
-      "learning_rate": 0.00019999983917436354,
-      "loss": 46.0,
-      "step": 3555
-    },
-    {
-      "epoch": 0.5726478521679617,
-      "grad_norm": 0.0007510171853937209,
-      "learning_rate": 0.00019999983908361698,
-      "loss": 46.0,
-      "step": 3556
-    },
-    {
-      "epoch": 0.5728088892467491,
-      "grad_norm": 0.0008446076535619795,
-      "learning_rate": 0.0001999998389928448,
-      "loss": 46.0,
-      "step": 3557
-    },
-    {
-      "epoch": 0.5729699263255364,
-      "grad_norm": 0.0005185237969271839,
-      "learning_rate": 0.00019999983890204708,
-      "loss": 46.0,
-      "step": 3558
-    },
-    {
-      "epoch": 0.5731309634043239,
-      "grad_norm": 0.0003346470766700804,
-      "learning_rate": 0.00019999983881122373,
-      "loss": 46.0,
-      "step": 3559
-    },
-    {
-      "epoch": 0.5732920004831112,
-      "grad_norm": 0.00042650633258745074,
-      "learning_rate": 0.0001999998387203748,
-      "loss": 46.0,
-      "step": 3560
-    },
-    {
-      "epoch": 0.5734530375618986,
-      "grad_norm": 0.0009748564916662872,
-      "learning_rate": 0.00019999983862950026,
-      "loss": 46.0,
-      "step": 3561
-    },
-    {
-      "epoch": 0.573614074640686,
-      "grad_norm": 0.002157721435651183,
-      "learning_rate": 0.00019999983853860012,
-      "loss": 46.0,
-      "step": 3562
-    },
-    {
-      "epoch": 0.5737751117194734,
-      "grad_norm": 0.0015437881229445338,
-      "learning_rate": 0.0001999998384476744,
-      "loss": 46.0,
-      "step": 3563
-    },
-    {
-      "epoch": 0.5739361487982608,
-      "grad_norm": 0.0020530105102807283,
-      "learning_rate": 0.0001999998383567231,
-      "loss": 46.0,
-      "step": 3564
-    },
-    {
-      "epoch": 0.5740971858770482,
-      "grad_norm": 0.0010107549605891109,
-      "learning_rate": 0.00019999983826574617,
-      "loss": 46.0,
-      "step": 3565
-    },
-    {
-      "epoch": 0.5742582229558356,
-      "grad_norm": 0.001300931558944285,
-      "learning_rate": 0.00019999983817474366,
-      "loss": 46.0,
-      "step": 3566
-    },
-    {
-      "epoch": 0.574419260034623,
-      "grad_norm": 0.001504823681898415,
-      "learning_rate": 0.00019999983808371557,
-      "loss": 46.0,
-      "step": 3567
-    },
-    {
-      "epoch": 0.5745802971134104,
-      "grad_norm": 0.00021979343728162348,
-      "learning_rate": 0.00019999983799266186,
-      "loss": 46.0,
-      "step": 3568
-    },
-    {
-      "epoch": 0.5747413341921978,
-      "grad_norm": 0.0015059680445119739,
-      "learning_rate": 0.00019999983790158256,
-      "loss": 46.0,
-      "step": 3569
-    },
-    {
-      "epoch": 0.5749023712709851,
-      "grad_norm": 0.00039003693382255733,
-      "learning_rate": 0.0001999998378104777,
-      "loss": 46.0,
-      "step": 3570
-    },
-    {
-      "epoch": 0.5750634083497725,
-      "grad_norm": 0.0007288960041478276,
-      "learning_rate": 0.0001999998377193472,
-      "loss": 46.0,
-      "step": 3571
-    },
-    {
-      "epoch": 0.5752244454285599,
-      "grad_norm": 0.0005308515974320471,
-      "learning_rate": 0.0001999998376281911,
-      "loss": 46.0,
-      "step": 3572
-    },
-    {
-      "epoch": 0.5753854825073473,
-      "grad_norm": 0.002361439634114504,
-      "learning_rate": 0.00019999983753700947,
-      "loss": 46.0,
-      "step": 3573
-    },
-    {
-      "epoch": 0.5755465195861347,
-      "grad_norm": 0.0008066802402026951,
-      "learning_rate": 0.0001999998374458022,
-      "loss": 46.0,
-      "step": 3574
-    },
-    {
-      "epoch": 0.5757075566649221,
-      "grad_norm": 0.0005559417768381536,
-      "learning_rate": 0.00019999983735456933,
-      "loss": 46.0,
-      "step": 3575
-    },
-    {
-      "epoch": 0.5758685937437095,
-      "grad_norm": 0.0020220763981342316,
-      "learning_rate": 0.00019999983726331087,
-      "loss": 46.0,
-      "step": 3576
-    },
-    {
-      "epoch": 0.5760296308224969,
-      "grad_norm": 0.0006484172772616148,
-      "learning_rate": 0.00019999983717202682,
-      "loss": 46.0,
-      "step": 3577
-    },
-    {
-      "epoch": 0.5761906679012843,
-      "grad_norm": 0.0004538370994850993,
-      "learning_rate": 0.00019999983708071718,
-      "loss": 46.0,
-      "step": 3578
-    },
-    {
-      "epoch": 0.5763517049800717,
-      "grad_norm": 0.00043682201066985726,
-      "learning_rate": 0.00019999983698938193,
-      "loss": 46.0,
-      "step": 3579
-    },
-    {
-      "epoch": 0.576512742058859,
-      "grad_norm": 0.0015078972792252898,
-      "learning_rate": 0.00019999983689802112,
-      "loss": 46.0,
-      "step": 3580
-    },
-    {
-      "epoch": 0.5766737791376464,
-      "grad_norm": 0.0007493096636608243,
-      "learning_rate": 0.00019999983680663467,
-      "loss": 46.0,
-      "step": 3581
-    },
-    {
-      "epoch": 0.5768348162164338,
-      "grad_norm": 0.0013236848171800375,
-      "learning_rate": 0.00019999983671522265,
-      "loss": 46.0,
-      "step": 3582
-    },
-    {
-      "epoch": 0.5769958532952212,
-      "grad_norm": 0.001121731474995613,
-      "learning_rate": 0.00019999983662378503,
-      "loss": 46.0,
-      "step": 3583
-    },
-    {
-      "epoch": 0.5771568903740086,
-      "grad_norm": 0.0005497890524566174,
-      "learning_rate": 0.00019999983653232181,
-      "loss": 46.0,
-      "step": 3584
-    },
-    {
-      "epoch": 0.577317927452796,
-      "grad_norm": 0.0013116998597979546,
-      "learning_rate": 0.000199999836440833,
-      "loss": 46.0,
-      "step": 3585
-    },
-    {
-      "epoch": 0.5774789645315834,
-      "grad_norm": 0.0018959867302328348,
-      "learning_rate": 0.0001999998363493186,
-      "loss": 46.0,
-      "step": 3586
-    },
-    {
-      "epoch": 0.5776400016103708,
-      "grad_norm": 0.0016209406312555075,
-      "learning_rate": 0.0001999998362577786,
-      "loss": 46.0,
-      "step": 3587
-    },
-    {
-      "epoch": 0.5778010386891582,
-      "grad_norm": 0.0005310988635756075,
-      "learning_rate": 0.000199999836166213,
-      "loss": 46.0,
-      "step": 3588
-    },
-    {
-      "epoch": 0.5779620757679456,
-      "grad_norm": 0.0009252220625057817,
-      "learning_rate": 0.0001999998360746218,
-      "loss": 46.0,
-      "step": 3589
-    },
-    {
-      "epoch": 0.578123112846733,
-      "grad_norm": 0.001061987248249352,
-      "learning_rate": 0.00019999983598300504,
-      "loss": 46.0,
-      "step": 3590
-    },
-    {
-      "epoch": 0.5782841499255204,
-      "grad_norm": 0.0008281744085252285,
-      "learning_rate": 0.00019999983589136266,
-      "loss": 46.0,
-      "step": 3591
-    },
-    {
-      "epoch": 0.5784451870043077,
-      "grad_norm": 0.000869329203851521,
-      "learning_rate": 0.00019999983579969467,
-      "loss": 46.0,
-      "step": 3592
-    },
-    {
-      "epoch": 0.5786062240830951,
-      "grad_norm": 0.0005601634038612247,
-      "learning_rate": 0.0001999998357080011,
-      "loss": 46.0,
-      "step": 3593
-    },
-    {
-      "epoch": 0.5787672611618825,
-      "grad_norm": 0.00048407120630145073,
-      "learning_rate": 0.00019999983561628195,
-      "loss": 46.0,
-      "step": 3594
-    },
-    {
-      "epoch": 0.5789282982406699,
-      "grad_norm": 0.0009586061351001263,
-      "learning_rate": 0.0001999998355245372,
-      "loss": 46.0,
-      "step": 3595
-    },
-    {
-      "epoch": 0.5790893353194573,
-      "grad_norm": 0.0009653475135564804,
-      "learning_rate": 0.00019999983543276682,
-      "loss": 46.0,
-      "step": 3596
-    },
-    {
-      "epoch": 0.5792503723982447,
-      "grad_norm": 0.00044590633478946984,
-      "learning_rate": 0.0001999998353409709,
-      "loss": 46.0,
-      "step": 3597
-    },
-    {
-      "epoch": 0.5794114094770321,
-      "grad_norm": 0.0006686762790195644,
-      "learning_rate": 0.00019999983524914935,
-      "loss": 46.0,
-      "step": 3598
-    },
-    {
-      "epoch": 0.5795724465558195,
-      "grad_norm": 0.001065866556018591,
-      "learning_rate": 0.0001999998351573022,
-      "loss": 46.0,
-      "step": 3599
-    },
-    {
-      "epoch": 0.5797334836346069,
-      "grad_norm": 0.00033228585380129516,
-      "learning_rate": 0.00019999983506542948,
-      "loss": 46.0,
-      "step": 3600
-    },
-    {
-      "epoch": 0.5798945207133943,
-      "grad_norm": 0.0004213275678921491,
-      "learning_rate": 0.00019999983497353117,
-      "loss": 46.0,
-      "step": 3601
-    },
-    {
-      "epoch": 0.5800555577921817,
-      "grad_norm": 0.00023344768851529807,
-      "learning_rate": 0.00019999983488160723,
-      "loss": 46.0,
-      "step": 3602
-    },
-    {
-      "epoch": 0.580216594870969,
-      "grad_norm": 0.0004481116484384984,
-      "learning_rate": 0.00019999983478965772,
-      "loss": 46.0,
-      "step": 3603
-    },
-    {
-      "epoch": 0.5803776319497564,
-      "grad_norm": 0.0009275152697227895,
-      "learning_rate": 0.00019999983469768263,
-      "loss": 46.0,
-      "step": 3604
-    },
-    {
-      "epoch": 0.5805386690285438,
-      "grad_norm": 0.0010258641559630632,
-      "learning_rate": 0.0001999998346056819,
-      "loss": 46.0,
-      "step": 3605
-    },
-    {
-      "epoch": 0.5806997061073312,
-      "grad_norm": 0.0012958927545696497,
-      "learning_rate": 0.0001999998345136556,
-      "loss": 46.0,
-      "step": 3606
-    },
-    {
-      "epoch": 0.5808607431861186,
-      "grad_norm": 0.0005233039846643806,
-      "learning_rate": 0.0001999998344216037,
-      "loss": 46.0,
-      "step": 3607
-    },
-    {
-      "epoch": 0.581021780264906,
-      "grad_norm": 0.0007316289120353758,
-      "learning_rate": 0.00019999983432952622,
-      "loss": 46.0,
-      "step": 3608
-    },
-    {
-      "epoch": 0.5811828173436934,
-      "grad_norm": 0.00045096175745129585,
-      "learning_rate": 0.0001999998342374231,
-      "loss": 46.0,
-      "step": 3609
-    },
-    {
-      "epoch": 0.5813438544224808,
-      "grad_norm": 0.0005383588722907007,
-      "learning_rate": 0.00019999983414529444,
-      "loss": 46.0,
-      "step": 3610
-    },
-    {
-      "epoch": 0.5815048915012682,
-      "grad_norm": 0.0005897500668652356,
-      "learning_rate": 0.00019999983405314016,
-      "loss": 46.0,
-      "step": 3611
-    },
-    {
-      "epoch": 0.5816659285800555,
-      "grad_norm": 0.00036973372334614396,
-      "learning_rate": 0.0001999998339609603,
-      "loss": 46.0,
-      "step": 3612
-    },
-    {
-      "epoch": 0.581826965658843,
-      "grad_norm": 0.001100813620723784,
-      "learning_rate": 0.00019999983386875486,
-      "loss": 46.0,
-      "step": 3613
-    },
-    {
-      "epoch": 0.5819880027376303,
-      "grad_norm": 0.000461019721115008,
-      "learning_rate": 0.00019999983377652378,
-      "loss": 46.0,
-      "step": 3614
-    },
-    {
-      "epoch": 0.5821490398164177,
-      "grad_norm": 0.0005715559236705303,
-      "learning_rate": 0.00019999983368426712,
-      "loss": 46.0,
-      "step": 3615
-    },
-    {
-      "epoch": 0.5823100768952051,
-      "grad_norm": 0.001074679777957499,
-      "learning_rate": 0.00019999983359198488,
-      "loss": 46.0,
-      "step": 3616
-    },
-    {
-      "epoch": 0.5824711139739925,
-      "grad_norm": 0.0018154182471334934,
-      "learning_rate": 0.000199999833499677,
-      "loss": 46.0,
-      "step": 3617
-    },
-    {
-      "epoch": 0.5826321510527799,
-      "grad_norm": 0.0009163039503619075,
-      "learning_rate": 0.0001999998334073436,
-      "loss": 46.0,
-      "step": 3618
-    },
-    {
-      "epoch": 0.5827931881315673,
-      "grad_norm": 0.001382135204039514,
-      "learning_rate": 0.00019999983331498456,
-      "loss": 46.0,
-      "step": 3619
-    },
-    {
-      "epoch": 0.5829542252103547,
-      "grad_norm": 0.0013830214738845825,
-      "learning_rate": 0.00019999983322259993,
-      "loss": 46.0,
-      "step": 3620
-    },
-    {
-      "epoch": 0.5831152622891421,
-      "grad_norm": 0.0014277211157605052,
-      "learning_rate": 0.0001999998331301897,
-      "loss": 46.0,
-      "step": 3621
-    },
-    {
-      "epoch": 0.5832762993679295,
-      "grad_norm": 0.000550763972569257,
-      "learning_rate": 0.0001999998330377539,
-      "loss": 46.0,
-      "step": 3622
-    },
-    {
-      "epoch": 0.5834373364467168,
-      "grad_norm": 0.0009363066637888551,
-      "learning_rate": 0.00019999983294529249,
-      "loss": 46.0,
-      "step": 3623
-    },
-    {
-      "epoch": 0.5835983735255043,
-      "grad_norm": 0.0004841428017243743,
-      "learning_rate": 0.0001999998328528055,
-      "loss": 46.0,
-      "step": 3624
-    },
-    {
-      "epoch": 0.5837594106042916,
-      "grad_norm": 0.00046791156637482345,
-      "learning_rate": 0.00019999983276029287,
-      "loss": 46.0,
-      "step": 3625
-    },
-    {
-      "epoch": 0.583920447683079,
-      "grad_norm": 0.0010113624157384038,
-      "learning_rate": 0.00019999983266775467,
-      "loss": 46.0,
-      "step": 3626
-    },
-    {
-      "epoch": 0.5840814847618664,
-      "grad_norm": 0.0011995943496003747,
-      "learning_rate": 0.00019999983257519089,
-      "loss": 46.0,
-      "step": 3627
-    },
-    {
-      "epoch": 0.5842425218406538,
-      "grad_norm": 0.00030865235021337867,
-      "learning_rate": 0.00019999983248260148,
-      "loss": 46.0,
-      "step": 3628
-    },
-    {
-      "epoch": 0.5844035589194412,
-      "grad_norm": 0.002150277839973569,
-      "learning_rate": 0.0001999998323899865,
-      "loss": 46.0,
-      "step": 3629
-    },
-    {
-      "epoch": 0.5845645959982286,
-      "grad_norm": 0.0004402558843139559,
-      "learning_rate": 0.00019999983229734595,
-      "loss": 46.0,
-      "step": 3630
-    },
-    {
-      "epoch": 0.584725633077016,
-      "grad_norm": 0.0005481082480400801,
-      "learning_rate": 0.00019999983220467976,
-      "loss": 46.0,
-      "step": 3631
-    },
-    {
-      "epoch": 0.5848866701558034,
-      "grad_norm": 0.0013047531247138977,
-      "learning_rate": 0.00019999983211198798,
-      "loss": 46.0,
-      "step": 3632
-    },
-    {
-      "epoch": 0.5850477072345908,
-      "grad_norm": 0.0029339916072785854,
-      "learning_rate": 0.00019999983201927064,
-      "loss": 46.0,
-      "step": 3633
-    },
-    {
-      "epoch": 0.5852087443133781,
-      "grad_norm": 0.000449870916781947,
-      "learning_rate": 0.00019999983192652766,
-      "loss": 46.0,
-      "step": 3634
-    },
-    {
-      "epoch": 0.5853697813921656,
-      "grad_norm": 0.00034026397042907774,
-      "learning_rate": 0.00019999983183375912,
-      "loss": 46.0,
-      "step": 3635
-    },
-    {
-      "epoch": 0.5855308184709529,
-      "grad_norm": 0.0006399744306690991,
-      "learning_rate": 0.00019999983174096497,
-      "loss": 46.0,
-      "step": 3636
-    },
-    {
-      "epoch": 0.5856918555497403,
-      "grad_norm": 0.0007925445679575205,
-      "learning_rate": 0.00019999983164814525,
-      "loss": 46.0,
-      "step": 3637
-    },
-    {
-      "epoch": 0.5858528926285277,
-      "grad_norm": 0.00047588645247742534,
-      "learning_rate": 0.0001999998315552999,
-      "loss": 46.0,
-      "step": 3638
-    },
-    {
-      "epoch": 0.5860139297073151,
-      "grad_norm": 0.000949476205278188,
-      "learning_rate": 0.00019999983146242896,
-      "loss": 46.0,
-      "step": 3639
-    },
-    {
-      "epoch": 0.5861749667861025,
-      "grad_norm": 0.0011420522350817919,
-      "learning_rate": 0.00019999983136953246,
-      "loss": 46.0,
-      "step": 3640
-    },
-    {
-      "epoch": 0.5863360038648899,
-      "grad_norm": 0.0017230648081749678,
-      "learning_rate": 0.00019999983127661034,
-      "loss": 46.0,
-      "step": 3641
-    },
-    {
-      "epoch": 0.5864970409436773,
-      "grad_norm": 0.0022951019927859306,
-      "learning_rate": 0.0001999998311836626,
-      "loss": 46.0,
-      "step": 3642
-    },
-    {
-      "epoch": 0.5866580780224647,
-      "grad_norm": 0.0002443125413265079,
-      "learning_rate": 0.0001999998310906893,
-      "loss": 46.0,
-      "step": 3643
-    },
-    {
-      "epoch": 0.5868191151012521,
-      "grad_norm": 0.000896506302524358,
-      "learning_rate": 0.00019999983099769039,
-      "loss": 46.0,
-      "step": 3644
-    },
-    {
-      "epoch": 0.5869801521800394,
-      "grad_norm": 0.0013104716781526804,
-      "learning_rate": 0.0001999998309046659,
-      "loss": 46.0,
-      "step": 3645
-    },
-    {
-      "epoch": 0.5871411892588269,
-      "grad_norm": 0.00042626337381079793,
-      "learning_rate": 0.00019999983081161581,
-      "loss": 46.0,
-      "step": 3646
-    },
-    {
-      "epoch": 0.5873022263376142,
-      "grad_norm": 0.0006456349510699511,
-      "learning_rate": 0.00019999983071854012,
-      "loss": 46.0,
-      "step": 3647
-    },
-    {
-      "epoch": 0.5874632634164016,
-      "grad_norm": 0.0021495043765753508,
-      "learning_rate": 0.00019999983062543884,
-      "loss": 46.0,
-      "step": 3648
-    },
-    {
-      "epoch": 0.587624300495189,
-      "grad_norm": 0.0008493894129060209,
-      "learning_rate": 0.00019999983053231194,
-      "loss": 46.0,
-      "step": 3649
-    },
-    {
-      "epoch": 0.5877853375739764,
-      "grad_norm": 0.0003227617416996509,
-      "learning_rate": 0.0001999998304391595,
-      "loss": 46.0,
-      "step": 3650
-    },
-    {
-      "epoch": 0.5879463746527638,
-      "grad_norm": 0.000657593656796962,
-      "learning_rate": 0.0001999998303459814,
-      "loss": 46.0,
-      "step": 3651
-    },
-    {
-      "epoch": 0.5881074117315512,
-      "grad_norm": 0.0006709296721965075,
-      "learning_rate": 0.00019999983025277776,
-      "loss": 46.0,
-      "step": 3652
-    },
-    {
-      "epoch": 0.5882684488103386,
-      "grad_norm": 0.001672015176154673,
-      "learning_rate": 0.0001999998301595485,
-      "loss": 46.0,
-      "step": 3653
-    },
-    {
-      "epoch": 0.588429485889126,
-      "grad_norm": 0.0018978442531079054,
-      "learning_rate": 0.00019999983006629363,
-      "loss": 46.0,
-      "step": 3654
-    },
-    {
-      "epoch": 0.5885905229679134,
-      "grad_norm": 0.0015170890837907791,
-      "learning_rate": 0.0001999998299730132,
-      "loss": 46.0,
-      "step": 3655
-    },
-    {
-      "epoch": 0.5887515600467007,
-      "grad_norm": 0.001059291884303093,
-      "learning_rate": 0.00019999982987970713,
-      "loss": 46.0,
-      "step": 3656
-    },
-    {
-      "epoch": 0.5889125971254882,
-      "grad_norm": 0.0006420931313186884,
-      "learning_rate": 0.00019999982978637554,
-      "loss": 46.0,
-      "step": 3657
-    },
-    {
-      "epoch": 0.5890736342042755,
-      "grad_norm": 0.003045084420591593,
-      "learning_rate": 0.0001999998296930183,
-      "loss": 46.0,
-      "step": 3658
-    },
-    {
-      "epoch": 0.589234671283063,
-      "grad_norm": 0.0008750835550017655,
-      "learning_rate": 0.00019999982959963546,
-      "loss": 46.0,
-      "step": 3659
-    },
-    {
-      "epoch": 0.5893957083618503,
-      "grad_norm": 0.0018582548946142197,
-      "learning_rate": 0.00019999982950622705,
-      "loss": 46.0,
-      "step": 3660
-    },
-    {
-      "epoch": 0.5895567454406377,
-      "grad_norm": 0.0004779713926836848,
-      "learning_rate": 0.00019999982941279302,
-      "loss": 46.0,
-      "step": 3661
-    },
-    {
-      "epoch": 0.5897177825194251,
-      "grad_norm": 0.0008654078119434416,
-      "learning_rate": 0.00019999982931933342,
-      "loss": 46.0,
-      "step": 3662
-    },
-    {
-      "epoch": 0.5898788195982125,
-      "grad_norm": 0.002062864601612091,
-      "learning_rate": 0.00019999982922584822,
-      "loss": 46.0,
-      "step": 3663
-    },
-    {
-      "epoch": 0.5900398566769999,
-      "grad_norm": 0.0011476227082312107,
-      "learning_rate": 0.0001999998291323374,
-      "loss": 46.0,
-      "step": 3664
-    },
-    {
-      "epoch": 0.5902008937557872,
-      "grad_norm": 0.0005998503183946013,
-      "learning_rate": 0.00019999982903880098,
-      "loss": 46.0,
-      "step": 3665
-    },
-    {
-      "epoch": 0.5903619308345747,
-      "grad_norm": 0.000645753345452249,
-      "learning_rate": 0.000199999828945239,
-      "loss": 46.0,
-      "step": 3666
-    },
-    {
-      "epoch": 0.590522967913362,
-      "grad_norm": 0.0010744401952251792,
-      "learning_rate": 0.00019999982885165142,
-      "loss": 46.0,
-      "step": 3667
-    },
-    {
-      "epoch": 0.5906840049921495,
-      "grad_norm": 0.000499027140904218,
-      "learning_rate": 0.00019999982875803824,
-      "loss": 46.0,
-      "step": 3668
-    },
-    {
-      "epoch": 0.5908450420709368,
-      "grad_norm": 0.0015712347812950611,
-      "learning_rate": 0.00019999982866439946,
-      "loss": 46.0,
-      "step": 3669
-    },
-    {
-      "epoch": 0.5910060791497243,
-      "grad_norm": 0.0006408823537640274,
-      "learning_rate": 0.0001999998285707351,
-      "loss": 46.0,
-      "step": 3670
-    },
-    {
-      "epoch": 0.5911671162285116,
-      "grad_norm": 0.00043742373236455023,
-      "learning_rate": 0.00019999982847704515,
-      "loss": 46.0,
-      "step": 3671
-    },
-    {
-      "epoch": 0.591328153307299,
-      "grad_norm": 0.00101959565654397,
-      "learning_rate": 0.0001999998283833296,
-      "loss": 46.0,
-      "step": 3672
-    },
-    {
-      "epoch": 0.5914891903860864,
-      "grad_norm": 0.0004744932521134615,
-      "learning_rate": 0.00019999982828958844,
-      "loss": 46.0,
-      "step": 3673
-    },
-    {
-      "epoch": 0.5916502274648738,
-      "grad_norm": 0.0006279168301261961,
-      "learning_rate": 0.00019999982819582165,
-      "loss": 46.0,
-      "step": 3674
-    },
-    {
-      "epoch": 0.5918112645436612,
-      "grad_norm": 0.0002817786007653922,
-      "learning_rate": 0.00019999982810202933,
-      "loss": 46.0,
-      "step": 3675
-    },
-    {
-      "epoch": 0.5919723016224485,
-      "grad_norm": 0.0009364317520521581,
-      "learning_rate": 0.0001999998280082114,
-      "loss": 46.0,
-      "step": 3676
-    },
-    {
-      "epoch": 0.592133338701236,
-      "grad_norm": 0.0011621599551290274,
-      "learning_rate": 0.00019999982791436786,
-      "loss": 46.0,
-      "step": 3677
-    },
-    {
-      "epoch": 0.5922943757800233,
-      "grad_norm": 0.0005299728363752365,
-      "learning_rate": 0.00019999982782049872,
-      "loss": 46.0,
-      "step": 3678
-    },
-    {
-      "epoch": 0.5924554128588108,
-      "grad_norm": 0.0007590660243295133,
-      "learning_rate": 0.00019999982772660402,
-      "loss": 46.0,
-      "step": 3679
-    },
-    {
-      "epoch": 0.5926164499375981,
-      "grad_norm": 0.002326404443010688,
-      "learning_rate": 0.00019999982763268368,
-      "loss": 46.0,
-      "step": 3680
-    },
-    {
-      "epoch": 0.5927774870163856,
-      "grad_norm": 0.0001865627127699554,
-      "learning_rate": 0.00019999982753873775,
-      "loss": 46.0,
-      "step": 3681
-    },
-    {
-      "epoch": 0.5929385240951729,
-      "grad_norm": 0.0034428501967340708,
-      "learning_rate": 0.00019999982744476626,
-      "loss": 46.0,
-      "step": 3682
-    },
-    {
-      "epoch": 0.5930995611739603,
-      "grad_norm": 0.0014843136304989457,
-      "learning_rate": 0.00019999982735076916,
-      "loss": 46.0,
-      "step": 3683
-    },
-    {
-      "epoch": 0.5932605982527477,
-      "grad_norm": 0.0003272859612479806,
-      "learning_rate": 0.00019999982725674647,
-      "loss": 46.0,
-      "step": 3684
-    },
-    {
-      "epoch": 0.5934216353315351,
-      "grad_norm": 0.0004735429538413882,
-      "learning_rate": 0.00019999982716269817,
-      "loss": 46.0,
-      "step": 3685
-    },
-    {
-      "epoch": 0.5935826724103225,
-      "grad_norm": 0.00019508595869410783,
-      "learning_rate": 0.0001999998270686243,
-      "loss": 46.0,
-      "step": 3686
-    },
-    {
-      "epoch": 0.5937437094891098,
-      "grad_norm": 0.001732675707899034,
-      "learning_rate": 0.00019999982697452482,
-      "loss": 46.0,
-      "step": 3687
-    },
-    {
-      "epoch": 0.5939047465678973,
-      "grad_norm": 0.001608144724741578,
-      "learning_rate": 0.00019999982688039973,
-      "loss": 46.0,
-      "step": 3688
-    },
-    {
-      "epoch": 0.5940657836466846,
-      "grad_norm": 0.0004894894664175808,
-      "learning_rate": 0.00019999982678624908,
-      "loss": 46.0,
-      "step": 3689
-    },
-    {
-      "epoch": 0.5942268207254721,
-      "grad_norm": 0.0029790301341563463,
-      "learning_rate": 0.0001999998266920728,
-      "loss": 46.0,
-      "step": 3690
-    },
-    {
-      "epoch": 0.5943878578042594,
-      "grad_norm": 0.0005888977902941406,
-      "learning_rate": 0.00019999982659787093,
-      "loss": 46.0,
-      "step": 3691
-    },
-    {
-      "epoch": 0.5945488948830469,
-      "grad_norm": 0.0012244251556694508,
-      "learning_rate": 0.0001999998265036435,
-      "loss": 46.0,
-      "step": 3692
-    },
-    {
-      "epoch": 0.5947099319618342,
-      "grad_norm": 0.0008973320946097374,
-      "learning_rate": 0.00019999982640939043,
-      "loss": 46.0,
-      "step": 3693
-    },
-    {
-      "epoch": 0.5948709690406216,
-      "grad_norm": 0.00048477097880095243,
-      "learning_rate": 0.0001999998263151118,
-      "loss": 46.0,
-      "step": 3694
-    },
-    {
-      "epoch": 0.595032006119409,
-      "grad_norm": 0.00031924486393108964,
-      "learning_rate": 0.00019999982622080756,
-      "loss": 46.0,
-      "step": 3695
-    },
-    {
-      "epoch": 0.5951930431981964,
-      "grad_norm": 0.0010715728858485818,
-      "learning_rate": 0.0001999998261264777,
-      "loss": 46.0,
-      "step": 3696
-    },
-    {
-      "epoch": 0.5953540802769838,
-      "grad_norm": 0.0007987231947481632,
-      "learning_rate": 0.00019999982603212225,
-      "loss": 46.0,
-      "step": 3697
-    },
-    {
-      "epoch": 0.5955151173557711,
-      "grad_norm": 0.002042955718934536,
-      "learning_rate": 0.00019999982593774124,
-      "loss": 46.0,
-      "step": 3698
-    },
-    {
-      "epoch": 0.5956761544345586,
-      "grad_norm": 0.0008362018852494657,
-      "learning_rate": 0.00019999982584333463,
-      "loss": 46.0,
-      "step": 3699
-    },
-    {
-      "epoch": 0.5958371915133459,
-      "grad_norm": 0.0007073442684486508,
-      "learning_rate": 0.0001999998257489024,
-      "loss": 46.0,
-      "step": 3700
-    },
-    {
-      "epoch": 0.5959982285921334,
-      "grad_norm": 0.0005143919261172414,
-      "learning_rate": 0.0001999998256544446,
-      "loss": 46.0,
-      "step": 3701
-    },
-    {
-      "epoch": 0.5961592656709207,
-      "grad_norm": 0.0019854875281453133,
-      "learning_rate": 0.0001999998255599612,
-      "loss": 46.0,
-      "step": 3702
-    },
-    {
-      "epoch": 0.5963203027497082,
-      "grad_norm": 0.0005943268770352006,
-      "learning_rate": 0.00019999982546545223,
-      "loss": 46.0,
-      "step": 3703
-    },
-    {
-      "epoch": 0.5964813398284955,
-      "grad_norm": 0.0009752835612744093,
-      "learning_rate": 0.0001999998253709176,
-      "loss": 46.0,
-      "step": 3704
-    },
-    {
-      "epoch": 0.596642376907283,
-      "grad_norm": 0.00037144351517781615,
-      "learning_rate": 0.00019999982527635745,
-      "loss": 46.0,
-      "step": 3705
-    },
-    {
-      "epoch": 0.5968034139860703,
-      "grad_norm": 0.0006946514477021992,
-      "learning_rate": 0.00019999982518177162,
-      "loss": 46.0,
-      "step": 3706
-    },
-    {
-      "epoch": 0.5969644510648577,
-      "grad_norm": 0.00024010856577660888,
-      "learning_rate": 0.00019999982508716024,
-      "loss": 46.0,
-      "step": 3707
-    },
-    {
-      "epoch": 0.5971254881436451,
-      "grad_norm": 0.0014027707511559129,
-      "learning_rate": 0.00019999982499252326,
-      "loss": 46.0,
-      "step": 3708
-    },
-    {
-      "epoch": 0.5972865252224324,
-      "grad_norm": 0.0010134363546967506,
-      "learning_rate": 0.00019999982489786073,
-      "loss": 46.0,
-      "step": 3709
-    },
-    {
-      "epoch": 0.5974475623012199,
-      "grad_norm": 0.0005614749970845878,
-      "learning_rate": 0.00019999982480317256,
-      "loss": 46.0,
-      "step": 3710
-    },
-    {
-      "epoch": 0.5976085993800072,
-      "grad_norm": 0.0012409163173288107,
-      "learning_rate": 0.0001999998247084588,
-      "loss": 46.0,
-      "step": 3711
-    },
-    {
-      "epoch": 0.5977696364587947,
-      "grad_norm": 0.000538306834641844,
-      "learning_rate": 0.00019999982461371945,
-      "loss": 46.0,
-      "step": 3712
-    },
-    {
-      "epoch": 0.597930673537582,
-      "grad_norm": 0.0003873163368552923,
-      "learning_rate": 0.0001999998245189545,
-      "loss": 46.0,
-      "step": 3713
-    },
-    {
-      "epoch": 0.5980917106163695,
-      "grad_norm": 0.0013742198934778571,
-      "learning_rate": 0.00019999982442416397,
-      "loss": 46.0,
-      "step": 3714
-    },
-    {
-      "epoch": 0.5982527476951568,
-      "grad_norm": 0.0007378747104667127,
-      "learning_rate": 0.0001999998243293478,
-      "loss": 46.0,
-      "step": 3715
-    },
-    {
-      "epoch": 0.5984137847739442,
-      "grad_norm": 0.0006354287033900619,
-      "learning_rate": 0.00019999982423450608,
-      "loss": 46.0,
-      "step": 3716
-    },
-    {
-      "epoch": 0.5985748218527316,
-      "grad_norm": 0.0002755013410933316,
-      "learning_rate": 0.00019999982413963877,
-      "loss": 46.0,
-      "step": 3717
-    },
-    {
-      "epoch": 0.5987358589315189,
-      "grad_norm": 0.0007118434878066182,
-      "learning_rate": 0.00019999982404474584,
-      "loss": 46.0,
-      "step": 3718
-    },
-    {
-      "epoch": 0.5988968960103064,
-      "grad_norm": 0.00195452687330544,
-      "learning_rate": 0.00019999982394982733,
-      "loss": 46.0,
-      "step": 3719
-    },
-    {
-      "epoch": 0.5990579330890937,
-      "grad_norm": 0.0009835156379267573,
-      "learning_rate": 0.0001999998238548832,
-      "loss": 46.0,
-      "step": 3720
-    },
-    {
-      "epoch": 0.5992189701678812,
-      "grad_norm": 0.0013090168358758092,
-      "learning_rate": 0.00019999982375991352,
-      "loss": 46.0,
-      "step": 3721
-    },
-    {
-      "epoch": 0.5993800072466685,
-      "grad_norm": 0.000989032443612814,
-      "learning_rate": 0.00019999982366491822,
-      "loss": 46.0,
-      "step": 3722
-    },
-    {
-      "epoch": 0.599541044325456,
-      "grad_norm": 0.0005709793185815215,
-      "learning_rate": 0.0001999998235698973,
-      "loss": 46.0,
-      "step": 3723
-    },
-    {
-      "epoch": 0.5997020814042433,
-      "grad_norm": 0.0016295472159981728,
-      "learning_rate": 0.00019999982347485083,
-      "loss": 46.0,
-      "step": 3724
-    },
-    {
-      "epoch": 0.5998631184830308,
-      "grad_norm": 0.0006179817137308419,
-      "learning_rate": 0.00019999982337977874,
-      "loss": 46.0,
-      "step": 3725
-    },
-    {
-      "epoch": 0.6000241555618181,
-      "grad_norm": 0.003281452227383852,
-      "learning_rate": 0.00019999982328468106,
-      "loss": 46.0,
-      "step": 3726
-    },
-    {
-      "epoch": 0.6001851926406055,
-      "grad_norm": 0.0005204345216043293,
-      "learning_rate": 0.00019999982318955777,
-      "loss": 46.0,
-      "step": 3727
-    },
-    {
-      "epoch": 0.6003462297193929,
-      "grad_norm": 0.0012670471332967281,
-      "learning_rate": 0.0001999998230944089,
-      "loss": 46.0,
-      "step": 3728
-    },
-    {
-      "epoch": 0.6005072667981802,
-      "grad_norm": 0.0008828460704535246,
-      "learning_rate": 0.00019999982299923446,
-      "loss": 46.0,
-      "step": 3729
-    },
-    {
-      "epoch": 0.6006683038769677,
-      "grad_norm": 0.0009197549661621451,
-      "learning_rate": 0.00019999982290403438,
-      "loss": 46.0,
-      "step": 3730
-    },
-    {
-      "epoch": 0.600829340955755,
-      "grad_norm": 0.0016467836685478687,
-      "learning_rate": 0.00019999982280880874,
-      "loss": 46.0,
-      "step": 3731
-    },
-    {
-      "epoch": 0.6009903780345425,
-      "grad_norm": 0.0011391020379960537,
-      "learning_rate": 0.00019999982271355748,
-      "loss": 46.0,
-      "step": 3732
-    },
-    {
-      "epoch": 0.6011514151133298,
-      "grad_norm": 0.0006474884576164186,
-      "learning_rate": 0.00019999982261828064,
-      "loss": 46.0,
-      "step": 3733
-    },
-    {
-      "epoch": 0.6013124521921173,
-      "grad_norm": 0.00021935020049568266,
-      "learning_rate": 0.0001999998225229782,
-      "loss": 46.0,
-      "step": 3734
-    },
-    {
-      "epoch": 0.6014734892709046,
-      "grad_norm": 0.0004743691242765635,
-      "learning_rate": 0.00019999982242765017,
-      "loss": 46.0,
-      "step": 3735
-    },
-    {
-      "epoch": 0.6016345263496921,
-      "grad_norm": 0.0007327792700380087,
-      "learning_rate": 0.00019999982233229652,
-      "loss": 46.0,
-      "step": 3736
-    },
-    {
-      "epoch": 0.6017955634284794,
-      "grad_norm": 0.0007064550300128758,
-      "learning_rate": 0.0001999998222369173,
-      "loss": 46.0,
-      "step": 3737
-    },
-    {
-      "epoch": 0.6019566005072668,
-      "grad_norm": 0.000860275118611753,
-      "learning_rate": 0.0001999998221415125,
-      "loss": 46.0,
-      "step": 3738
-    },
-    {
-      "epoch": 0.6021176375860542,
-      "grad_norm": 0.0012082658940926194,
-      "learning_rate": 0.0001999998220460821,
-      "loss": 46.0,
-      "step": 3739
-    },
-    {
-      "epoch": 0.6022786746648415,
-      "grad_norm": 0.00036825332790613174,
-      "learning_rate": 0.00019999982195062607,
-      "loss": 46.0,
-      "step": 3740
-    },
-    {
-      "epoch": 0.602439711743629,
-      "grad_norm": 0.0013934002490714192,
-      "learning_rate": 0.00019999982185514448,
-      "loss": 46.0,
-      "step": 3741
-    },
-    {
-      "epoch": 0.6026007488224163,
-      "grad_norm": 0.0005578524433076382,
-      "learning_rate": 0.00019999982175963727,
-      "loss": 46.0,
-      "step": 3742
-    },
-    {
-      "epoch": 0.6027617859012038,
-      "grad_norm": 0.00038510331069119275,
-      "learning_rate": 0.00019999982166410448,
-      "loss": 46.0,
-      "step": 3743
-    },
-    {
-      "epoch": 0.6029228229799911,
-      "grad_norm": 0.0005473191267810762,
-      "learning_rate": 0.0001999998215685461,
-      "loss": 46.0,
-      "step": 3744
-    },
-    {
-      "epoch": 0.6030838600587786,
-      "grad_norm": 0.0015968807274475694,
-      "learning_rate": 0.00019999982147296213,
-      "loss": 46.0,
-      "step": 3745
-    },
-    {
-      "epoch": 0.6032448971375659,
-      "grad_norm": 0.0008202851167879999,
-      "learning_rate": 0.00019999982137735255,
-      "loss": 46.0,
-      "step": 3746
-    },
-    {
-      "epoch": 0.6034059342163534,
-      "grad_norm": 0.0010265555465593934,
-      "learning_rate": 0.00019999982128171738,
-      "loss": 46.0,
-      "step": 3747
-    },
-    {
-      "epoch": 0.6035669712951407,
-      "grad_norm": 0.0016084174858406186,
-      "learning_rate": 0.0001999998211860566,
-      "loss": 46.0,
-      "step": 3748
-    },
-    {
-      "epoch": 0.6037280083739281,
-      "grad_norm": 0.0033480769488960505,
-      "learning_rate": 0.00019999982109037022,
-      "loss": 46.0,
-      "step": 3749
-    },
-    {
-      "epoch": 0.6038890454527155,
-      "grad_norm": 0.0015542751643806696,
-      "learning_rate": 0.0001999998209946583,
-      "loss": 46.0,
-      "step": 3750
-    },
-    {
-      "epoch": 0.6040500825315028,
-      "grad_norm": 0.0006006431649439037,
-      "learning_rate": 0.00019999982089892075,
-      "loss": 46.0,
-      "step": 3751
-    },
-    {
-      "epoch": 0.6042111196102903,
-      "grad_norm": 0.0010201828554272652,
-      "learning_rate": 0.00019999982080315759,
-      "loss": 46.0,
-      "step": 3752
-    },
-    {
-      "epoch": 0.6043721566890776,
-      "grad_norm": 0.0005846228450536728,
-      "learning_rate": 0.00019999982070736887,
-      "loss": 46.0,
-      "step": 3753
-    },
-    {
-      "epoch": 0.6045331937678651,
-      "grad_norm": 0.00048275847802869976,
-      "learning_rate": 0.00019999982061155453,
-      "loss": 46.0,
-      "step": 3754
-    },
-    {
-      "epoch": 0.6046942308466524,
-      "grad_norm": 0.0007435891893692315,
-      "learning_rate": 0.00019999982051571458,
-      "loss": 46.0,
-      "step": 3755
-    },
-    {
-      "epoch": 0.6048552679254399,
-      "grad_norm": 0.0011720845941454172,
-      "learning_rate": 0.00019999982041984905,
-      "loss": 46.0,
-      "step": 3756
-    },
-    {
-      "epoch": 0.6050163050042272,
-      "grad_norm": 0.0019239976536482573,
-      "learning_rate": 0.00019999982032395795,
-      "loss": 46.0,
-      "step": 3757
-    },
-    {
-      "epoch": 0.6051773420830147,
-      "grad_norm": 0.0009448094642721117,
-      "learning_rate": 0.00019999982022804122,
-      "loss": 46.0,
-      "step": 3758
-    },
-    {
-      "epoch": 0.605338379161802,
-      "grad_norm": 0.0015097275609150529,
-      "learning_rate": 0.00019999982013209892,
-      "loss": 46.0,
-      "step": 3759
-    },
-    {
-      "epoch": 0.6054994162405894,
-      "grad_norm": 0.0013230973854660988,
-      "learning_rate": 0.000199999820036131,
-      "loss": 46.0,
-      "step": 3760
-    },
-    {
-      "epoch": 0.6056604533193768,
-      "grad_norm": 0.0014520409749820828,
-      "learning_rate": 0.00019999981994013754,
-      "loss": 46.0,
-      "step": 3761
-    },
-    {
-      "epoch": 0.6058214903981641,
-      "grad_norm": 0.001978206681087613,
-      "learning_rate": 0.00019999981984411842,
-      "loss": 46.0,
-      "step": 3762
-    },
-    {
-      "epoch": 0.6059825274769516,
-      "grad_norm": 0.0010053090518340468,
-      "learning_rate": 0.00019999981974807372,
-      "loss": 46.0,
-      "step": 3763
-    },
-    {
-      "epoch": 0.6061435645557389,
-      "grad_norm": 0.0006428470369428396,
-      "learning_rate": 0.00019999981965200346,
-      "loss": 46.0,
-      "step": 3764
-    },
-    {
-      "epoch": 0.6063046016345264,
-      "grad_norm": 0.001770240836776793,
-      "learning_rate": 0.0001999998195559076,
-      "loss": 46.0,
-      "step": 3765
-    },
-    {
-      "epoch": 0.6064656387133137,
-      "grad_norm": 0.0008203188772313297,
-      "learning_rate": 0.0001999998194597861,
-      "loss": 46.0,
-      "step": 3766
-    },
-    {
-      "epoch": 0.6066266757921012,
-      "grad_norm": 0.000977416057139635,
-      "learning_rate": 0.00019999981936363902,
-      "loss": 46.0,
-      "step": 3767
-    },
-    {
-      "epoch": 0.6067877128708885,
-      "grad_norm": 0.0004567127616610378,
-      "learning_rate": 0.00019999981926746636,
-      "loss": 46.0,
-      "step": 3768
-    },
-    {
-      "epoch": 0.606948749949676,
-      "grad_norm": 0.0009826109744608402,
-      "learning_rate": 0.0001999998191712681,
-      "loss": 46.0,
-      "step": 3769
-    },
-    {
-      "epoch": 0.6071097870284633,
-      "grad_norm": 0.0003219879581592977,
-      "learning_rate": 0.00019999981907504425,
-      "loss": 46.0,
-      "step": 3770
-    },
-    {
-      "epoch": 0.6072708241072506,
-      "grad_norm": 0.0004293559759389609,
-      "learning_rate": 0.00019999981897879482,
-      "loss": 46.0,
-      "step": 3771
-    },
-    {
-      "epoch": 0.6074318611860381,
-      "grad_norm": 0.0004879342159256339,
-      "learning_rate": 0.00019999981888251978,
-      "loss": 46.0,
-      "step": 3772
-    },
-    {
-      "epoch": 0.6075928982648254,
-      "grad_norm": 0.0018218603217974305,
-      "learning_rate": 0.00019999981878621913,
-      "loss": 46.0,
-      "step": 3773
-    },
-    {
-      "epoch": 0.6077539353436129,
-      "grad_norm": 0.0005505846929736435,
-      "learning_rate": 0.0001999998186898929,
-      "loss": 46.0,
-      "step": 3774
-    },
-    {
-      "epoch": 0.6079149724224002,
-      "grad_norm": 0.0007565902196802199,
-      "learning_rate": 0.0001999998185935411,
-      "loss": 46.0,
-      "step": 3775
-    },
-    {
-      "epoch": 0.6080760095011877,
-      "grad_norm": 0.0012273786123842,
-      "learning_rate": 0.00019999981849716365,
-      "loss": 46.0,
-      "step": 3776
-    },
-    {
-      "epoch": 0.608237046579975,
-      "grad_norm": 0.0007383571937680244,
-      "learning_rate": 0.00019999981840076062,
-      "loss": 46.0,
-      "step": 3777
-    },
-    {
-      "epoch": 0.6083980836587625,
-      "grad_norm": 0.0007197211380116642,
-      "learning_rate": 0.00019999981830433203,
-      "loss": 46.0,
-      "step": 3778
-    },
-    {
-      "epoch": 0.6085591207375498,
-      "grad_norm": 0.0005319115007296205,
-      "learning_rate": 0.0001999998182078778,
-      "loss": 46.0,
-      "step": 3779
-    },
-    {
-      "epoch": 0.6087201578163373,
-      "grad_norm": 0.0008898625383153558,
-      "learning_rate": 0.00019999981811139804,
-      "loss": 46.0,
-      "step": 3780
-    },
-    {
-      "epoch": 0.6088811948951246,
-      "grad_norm": 0.001702308771200478,
-      "learning_rate": 0.00019999981801489263,
-      "loss": 46.0,
-      "step": 3781
-    },
-    {
-      "epoch": 0.6090422319739119,
-      "grad_norm": 0.0004603904963005334,
-      "learning_rate": 0.00019999981791836164,
-      "loss": 46.0,
-      "step": 3782
-    },
-    {
-      "epoch": 0.6092032690526994,
-      "grad_norm": 0.0015911172376945615,
-      "learning_rate": 0.00019999981782180506,
-      "loss": 46.0,
-      "step": 3783
-    },
-    {
-      "epoch": 0.6093643061314867,
-      "grad_norm": 0.0006629017880186439,
-      "learning_rate": 0.00019999981772522287,
-      "loss": 46.0,
-      "step": 3784
-    },
-    {
-      "epoch": 0.6095253432102742,
-      "grad_norm": 0.0025001366157084703,
-      "learning_rate": 0.00019999981762861512,
-      "loss": 46.0,
-      "step": 3785
-    },
-    {
-      "epoch": 0.6096863802890615,
-      "grad_norm": 0.0017279554158449173,
-      "learning_rate": 0.00019999981753198175,
-      "loss": 46.0,
-      "step": 3786
-    },
-    {
-      "epoch": 0.609847417367849,
-      "grad_norm": 0.0007181591936387122,
-      "learning_rate": 0.00019999981743532277,
-      "loss": 46.0,
-      "step": 3787
-    },
-    {
-      "epoch": 0.6100084544466363,
-      "grad_norm": 0.005317715462297201,
-      "learning_rate": 0.0001999998173386382,
-      "loss": 46.0,
-      "step": 3788
-    },
-    {
-      "epoch": 0.6101694915254238,
-      "grad_norm": 0.0013797798892483115,
-      "learning_rate": 0.00019999981724192804,
-      "loss": 46.0,
-      "step": 3789
-    },
-    {
-      "epoch": 0.6103305286042111,
-      "grad_norm": 0.0013191585894674063,
-      "learning_rate": 0.00019999981714519233,
-      "loss": 46.0,
-      "step": 3790
-    },
-    {
-      "epoch": 0.6104915656829986,
-      "grad_norm": 0.0009662522352300584,
-      "learning_rate": 0.00019999981704843097,
-      "loss": 46.0,
-      "step": 3791
-    },
-    {
-      "epoch": 0.6106526027617859,
-      "grad_norm": 0.000845549104269594,
-      "learning_rate": 0.00019999981695164402,
-      "loss": 46.0,
-      "step": 3792
-    },
-    {
-      "epoch": 0.6108136398405732,
-      "grad_norm": 0.000900446146260947,
-      "learning_rate": 0.0001999998168548315,
-      "loss": 46.0,
-      "step": 3793
-    },
-    {
-      "epoch": 0.6109746769193607,
-      "grad_norm": 0.0003443977329879999,
-      "learning_rate": 0.00019999981675799337,
-      "loss": 46.0,
-      "step": 3794
-    },
-    {
-      "epoch": 0.611135713998148,
-      "grad_norm": 0.00047858565812930465,
-      "learning_rate": 0.00019999981666112967,
-      "loss": 46.0,
-      "step": 3795
-    },
-    {
-      "epoch": 0.6112967510769355,
-      "grad_norm": 0.0012598923640325665,
-      "learning_rate": 0.00019999981656424032,
-      "loss": 46.0,
-      "step": 3796
-    },
-    {
-      "epoch": 0.6114577881557228,
-      "grad_norm": 0.0006759954849258065,
-      "learning_rate": 0.00019999981646732538,
-      "loss": 46.0,
-      "step": 3797
-    },
-    {
-      "epoch": 0.6116188252345103,
-      "grad_norm": 0.0004393019189592451,
-      "learning_rate": 0.0001999998163703849,
-      "loss": 46.0,
-      "step": 3798
-    },
-    {
-      "epoch": 0.6117798623132976,
-      "grad_norm": 0.0004934060852974653,
-      "learning_rate": 0.0001999998162734188,
-      "loss": 46.0,
-      "step": 3799
-    },
-    {
-      "epoch": 0.6119408993920851,
-      "grad_norm": 0.00030627084197476506,
-      "learning_rate": 0.0001999998161764271,
-      "loss": 46.0,
-      "step": 3800
-    },
-    {
-      "epoch": 0.6121019364708724,
-      "grad_norm": 0.0005866055726073682,
-      "learning_rate": 0.00019999981607940983,
-      "loss": 46.0,
-      "step": 3801
-    },
-    {
-      "epoch": 0.6122629735496599,
-      "grad_norm": 0.00043013528920710087,
-      "learning_rate": 0.00019999981598236693,
-      "loss": 46.0,
-      "step": 3802
-    },
-    {
-      "epoch": 0.6124240106284472,
-      "grad_norm": 0.0009788851020857692,
-      "learning_rate": 0.00019999981588529844,
-      "loss": 46.0,
-      "step": 3803
-    },
-    {
-      "epoch": 0.6125850477072345,
-      "grad_norm": 0.0005991174839437008,
-      "learning_rate": 0.00019999981578820437,
-      "loss": 46.0,
-      "step": 3804
-    },
-    {
-      "epoch": 0.612746084786022,
-      "grad_norm": 0.0006116996519267559,
-      "learning_rate": 0.0001999998156910847,
-      "loss": 46.0,
-      "step": 3805
-    },
-    {
-      "epoch": 0.6129071218648093,
-      "grad_norm": 0.0005804227548651397,
-      "learning_rate": 0.00019999981559393944,
-      "loss": 46.0,
-      "step": 3806
-    },
-    {
-      "epoch": 0.6130681589435968,
-      "grad_norm": 0.00042501665302552283,
-      "learning_rate": 0.00019999981549676855,
-      "loss": 46.0,
-      "step": 3807
-    },
-    {
-      "epoch": 0.6132291960223841,
-      "grad_norm": 0.0005513046635314822,
-      "learning_rate": 0.00019999981539957213,
-      "loss": 46.0,
-      "step": 3808
-    },
-    {
-      "epoch": 0.6133902331011716,
-      "grad_norm": 0.0009094532579183578,
-      "learning_rate": 0.00019999981530235007,
-      "loss": 46.0,
-      "step": 3809
-    },
-    {
-      "epoch": 0.6135512701799589,
-      "grad_norm": 0.0009424593299627304,
-      "learning_rate": 0.00019999981520510242,
-      "loss": 46.0,
-      "step": 3810
-    },
-    {
-      "epoch": 0.6137123072587464,
-      "grad_norm": 0.0018491118680685759,
-      "learning_rate": 0.00019999981510782918,
-      "loss": 46.0,
-      "step": 3811
-    },
-    {
-      "epoch": 0.6138733443375337,
-      "grad_norm": 0.0002931191702373326,
-      "learning_rate": 0.00019999981501053036,
-      "loss": 46.0,
-      "step": 3812
-    },
-    {
-      "epoch": 0.6140343814163212,
-      "grad_norm": 0.0005261205369606614,
-      "learning_rate": 0.00019999981491320592,
-      "loss": 46.0,
-      "step": 3813
-    },
-    {
-      "epoch": 0.6141954184951085,
-      "grad_norm": 0.0008702565683051944,
-      "learning_rate": 0.0001999998148158559,
-      "loss": 46.0,
-      "step": 3814
-    },
-    {
-      "epoch": 0.6143564555738958,
-      "grad_norm": 0.0009721557726152241,
-      "learning_rate": 0.00019999981471848028,
-      "loss": 46.0,
-      "step": 3815
-    },
-    {
-      "epoch": 0.6145174926526833,
-      "grad_norm": 0.0006699041114188731,
-      "learning_rate": 0.00019999981462107906,
-      "loss": 46.0,
-      "step": 3816
-    },
-    {
-      "epoch": 0.6146785297314706,
-      "grad_norm": 0.0013978193746879697,
-      "learning_rate": 0.00019999981452365227,
-      "loss": 46.0,
-      "step": 3817
-    },
-    {
-      "epoch": 0.6148395668102581,
-      "grad_norm": 0.0007404517964459956,
-      "learning_rate": 0.00019999981442619987,
-      "loss": 46.0,
-      "step": 3818
-    },
-    {
-      "epoch": 0.6150006038890454,
-      "grad_norm": 0.002723252633586526,
-      "learning_rate": 0.00019999981432872183,
-      "loss": 46.0,
-      "step": 3819
-    },
-    {
-      "epoch": 0.6151616409678329,
-      "grad_norm": 0.001904648612253368,
-      "learning_rate": 0.00019999981423121825,
-      "loss": 46.0,
-      "step": 3820
-    },
-    {
-      "epoch": 0.6153226780466202,
-      "grad_norm": 0.0015114553971216083,
-      "learning_rate": 0.00019999981413368906,
-      "loss": 46.0,
-      "step": 3821
-    },
-    {
-      "epoch": 0.6154837151254077,
-      "grad_norm": 0.0004959972575306892,
-      "learning_rate": 0.00019999981403613428,
-      "loss": 46.0,
-      "step": 3822
-    },
-    {
-      "epoch": 0.615644752204195,
-      "grad_norm": 0.0012471459340304136,
-      "learning_rate": 0.0001999998139385539,
-      "loss": 46.0,
-      "step": 3823
-    },
-    {
-      "epoch": 0.6158057892829824,
-      "grad_norm": 0.0011846354464069009,
-      "learning_rate": 0.00019999981384094791,
-      "loss": 46.0,
-      "step": 3824
-    },
-    {
-      "epoch": 0.6159668263617698,
-      "grad_norm": 0.0012976614525541663,
-      "learning_rate": 0.00019999981374331638,
-      "loss": 46.0,
-      "step": 3825
-    },
-    {
-      "epoch": 0.6161278634405571,
-      "grad_norm": 0.0011470116442069411,
-      "learning_rate": 0.0001999998136456592,
-      "loss": 46.0,
-      "step": 3826
-    },
-    {
-      "epoch": 0.6162889005193446,
-      "grad_norm": 0.0016877774614840746,
-      "learning_rate": 0.00019999981354797646,
-      "loss": 46.0,
-      "step": 3827
-    },
-    {
-      "epoch": 0.6164499375981319,
-      "grad_norm": 0.0009053684771060944,
-      "learning_rate": 0.00019999981345026808,
-      "loss": 46.0,
-      "step": 3828
-    },
-    {
-      "epoch": 0.6166109746769194,
-      "grad_norm": 0.0007914634770713747,
-      "learning_rate": 0.00019999981335253416,
-      "loss": 46.0,
-      "step": 3829
-    },
-    {
-      "epoch": 0.6167720117557067,
-      "grad_norm": 0.0006173317669890821,
-      "learning_rate": 0.0001999998132547746,
-      "loss": 46.0,
-      "step": 3830
-    },
-    {
-      "epoch": 0.6169330488344942,
-      "grad_norm": 0.0023323465138673782,
-      "learning_rate": 0.0001999998131569895,
-      "loss": 46.0,
-      "step": 3831
-    },
-    {
-      "epoch": 0.6170940859132815,
-      "grad_norm": 0.0005408467259258032,
-      "learning_rate": 0.00019999981305917873,
-      "loss": 46.0,
-      "step": 3832
-    },
-    {
-      "epoch": 0.617255122992069,
-      "grad_norm": 0.00047445937525480986,
-      "learning_rate": 0.00019999981296134242,
-      "loss": 46.0,
-      "step": 3833
-    },
-    {
-      "epoch": 0.6174161600708563,
-      "grad_norm": 0.00043043107143603265,
-      "learning_rate": 0.00019999981286348049,
-      "loss": 46.0,
-      "step": 3834
-    },
-    {
-      "epoch": 0.6175771971496437,
-      "grad_norm": 0.0006480773445218801,
-      "learning_rate": 0.00019999981276559297,
-      "loss": 46.0,
-      "step": 3835
-    },
-    {
-      "epoch": 0.6177382342284311,
-      "grad_norm": 0.0007769428775645792,
-      "learning_rate": 0.00019999981266767986,
-      "loss": 46.0,
-      "step": 3836
-    },
-    {
-      "epoch": 0.6178992713072184,
-      "grad_norm": 0.001116651576012373,
-      "learning_rate": 0.00019999981256974117,
-      "loss": 46.0,
-      "step": 3837
-    },
-    {
-      "epoch": 0.6180603083860059,
-      "grad_norm": 0.0008682135958224535,
-      "learning_rate": 0.00019999981247177686,
-      "loss": 46.0,
-      "step": 3838
-    },
-    {
-      "epoch": 0.6182213454647932,
-      "grad_norm": 0.00042617073631845415,
-      "learning_rate": 0.00019999981237378697,
-      "loss": 46.0,
-      "step": 3839
-    },
-    {
-      "epoch": 0.6183823825435807,
-      "grad_norm": 0.00023524355492554605,
-      "learning_rate": 0.0001999998122757715,
-      "loss": 46.0,
-      "step": 3840
-    },
-    {
-      "epoch": 0.618543419622368,
-      "grad_norm": 0.0008973352960310876,
-      "learning_rate": 0.00019999981217773042,
-      "loss": 46.0,
-      "step": 3841
-    },
-    {
-      "epoch": 0.6187044567011555,
-      "grad_norm": 0.0010380677413195372,
-      "learning_rate": 0.0001999998120796637,
-      "loss": 46.0,
-      "step": 3842
-    },
-    {
-      "epoch": 0.6188654937799428,
-      "grad_norm": 0.0005477048107422888,
-      "learning_rate": 0.00019999981198157144,
-      "loss": 46.0,
-      "step": 3843
-    },
-    {
-      "epoch": 0.6190265308587303,
-      "grad_norm": 0.0002454533241689205,
-      "learning_rate": 0.00019999981188345358,
-      "loss": 46.0,
-      "step": 3844
-    },
-    {
-      "epoch": 0.6191875679375176,
-      "grad_norm": 0.001563397003337741,
-      "learning_rate": 0.0001999998117853101,
-      "loss": 46.0,
-      "step": 3845
-    },
-    {
-      "epoch": 0.619348605016305,
-      "grad_norm": 0.0013099611969664693,
-      "learning_rate": 0.00019999981168714105,
-      "loss": 46.0,
-      "step": 3846
-    },
-    {
-      "epoch": 0.6195096420950924,
-      "grad_norm": 0.0002848474832717329,
-      "learning_rate": 0.0001999998115889464,
-      "loss": 46.0,
-      "step": 3847
-    },
-    {
-      "epoch": 0.6196706791738797,
-      "grad_norm": 0.0006217787740752101,
-      "learning_rate": 0.00019999981149072615,
-      "loss": 46.0,
-      "step": 3848
-    },
-    {
-      "epoch": 0.6198317162526672,
-      "grad_norm": 0.0007854134892113507,
-      "learning_rate": 0.00019999981139248032,
-      "loss": 46.0,
-      "step": 3849
-    },
-    {
-      "epoch": 0.6199927533314545,
-      "grad_norm": 0.0008232304826378822,
-      "learning_rate": 0.00019999981129420886,
-      "loss": 46.0,
-      "step": 3850
-    },
-    {
-      "epoch": 0.620153790410242,
-      "grad_norm": 0.0016802066238597035,
-      "learning_rate": 0.00019999981119591181,
-      "loss": 46.0,
-      "step": 3851
-    },
-    {
-      "epoch": 0.6203148274890293,
-      "grad_norm": 0.0023175273090600967,
-      "learning_rate": 0.0001999998110975892,
-      "loss": 46.0,
-      "step": 3852
-    },
-    {
-      "epoch": 0.6204758645678168,
-      "grad_norm": 0.00041702997987158597,
-      "learning_rate": 0.00019999981099924098,
-      "loss": 46.0,
-      "step": 3853
-    },
-    {
-      "epoch": 0.6206369016466041,
-      "grad_norm": 0.0015941159799695015,
-      "learning_rate": 0.00019999981090086717,
-      "loss": 46.0,
-      "step": 3854
-    },
-    {
-      "epoch": 0.6207979387253916,
-      "grad_norm": 0.0007031516870483756,
-      "learning_rate": 0.00019999981080246775,
-      "loss": 46.0,
-      "step": 3855
-    },
-    {
-      "epoch": 0.6209589758041789,
-      "grad_norm": 0.0005478821694850922,
-      "learning_rate": 0.00019999981070404274,
-      "loss": 46.0,
-      "step": 3856
-    },
-    {
-      "epoch": 0.6211200128829663,
-      "grad_norm": 0.0006148735410533845,
-      "learning_rate": 0.00019999981060559214,
-      "loss": 46.0,
-      "step": 3857
-    },
-    {
-      "epoch": 0.6212810499617537,
-      "grad_norm": 0.006427546963095665,
-      "learning_rate": 0.00019999981050711595,
-      "loss": 46.0,
-      "step": 3858
-    },
-    {
-      "epoch": 0.621442087040541,
-      "grad_norm": 0.0005981564172543585,
-      "learning_rate": 0.00019999981040861418,
-      "loss": 46.0,
-      "step": 3859
-    },
-    {
-      "epoch": 0.6216031241193285,
-      "grad_norm": 0.0006622473010793328,
-      "learning_rate": 0.00019999981031008676,
-      "loss": 46.0,
-      "step": 3860
-    },
-    {
-      "epoch": 0.6217641611981158,
-      "grad_norm": 0.0008345595560967922,
-      "learning_rate": 0.0001999998102115338,
-      "loss": 46.0,
-      "step": 3861
-    },
-    {
-      "epoch": 0.6219251982769033,
-      "grad_norm": 0.0006166543462313712,
-      "learning_rate": 0.0001999998101129552,
-      "loss": 46.0,
-      "step": 3862
-    },
-    {
-      "epoch": 0.6220862353556906,
-      "grad_norm": 0.0003607094695325941,
-      "learning_rate": 0.00019999981001435105,
-      "loss": 46.0,
-      "step": 3863
-    },
-    {
-      "epoch": 0.6222472724344781,
-      "grad_norm": 0.0033777060452848673,
-      "learning_rate": 0.0001999998099157213,
-      "loss": 46.0,
-      "step": 3864
-    },
-    {
-      "epoch": 0.6224083095132654,
-      "grad_norm": 0.0007539305952377617,
-      "learning_rate": 0.0001999998098170659,
-      "loss": 46.0,
-      "step": 3865
-    },
-    {
-      "epoch": 0.6225693465920529,
-      "grad_norm": 0.0010992271127179265,
-      "learning_rate": 0.00019999980971838497,
-      "loss": 46.0,
-      "step": 3866
-    },
-    {
-      "epoch": 0.6227303836708402,
-      "grad_norm": 0.0008015542407520115,
-      "learning_rate": 0.00019999980961967842,
-      "loss": 46.0,
-      "step": 3867
-    },
-    {
-      "epoch": 0.6228914207496276,
-      "grad_norm": 0.0003821809950750321,
-      "learning_rate": 0.00019999980952094628,
-      "loss": 46.0,
-      "step": 3868
-    },
-    {
-      "epoch": 0.623052457828415,
-      "grad_norm": 0.000647315348032862,
-      "learning_rate": 0.00019999980942218853,
-      "loss": 46.0,
-      "step": 3869
-    },
-    {
-      "epoch": 0.6232134949072023,
-      "grad_norm": 0.0005045785801485181,
-      "learning_rate": 0.0001999998093234052,
-      "loss": 46.0,
-      "step": 3870
-    },
-    {
-      "epoch": 0.6233745319859898,
-      "grad_norm": 0.0006298076477833092,
-      "learning_rate": 0.00019999980922459626,
-      "loss": 46.0,
-      "step": 3871
-    },
-    {
-      "epoch": 0.6235355690647771,
-      "grad_norm": 0.0006782549316994846,
-      "learning_rate": 0.00019999980912576172,
-      "loss": 46.0,
-      "step": 3872
-    },
-    {
-      "epoch": 0.6236966061435646,
-      "grad_norm": 0.0040741171687841415,
-      "learning_rate": 0.00019999980902690162,
-      "loss": 46.0,
-      "step": 3873
-    },
-    {
-      "epoch": 0.6238576432223519,
-      "grad_norm": 0.0009928788058459759,
-      "learning_rate": 0.0001999998089280159,
-      "loss": 46.0,
-      "step": 3874
-    },
-    {
-      "epoch": 0.6240186803011394,
-      "grad_norm": 0.0011796929175034165,
-      "learning_rate": 0.0001999998088291046,
-      "loss": 46.0,
-      "step": 3875
-    },
-    {
-      "epoch": 0.6241797173799267,
-      "grad_norm": 0.001887347549200058,
-      "learning_rate": 0.0001999998087301677,
-      "loss": 46.0,
-      "step": 3876
-    },
-    {
-      "epoch": 0.6243407544587141,
-      "grad_norm": 0.0020020632073283195,
-      "learning_rate": 0.0001999998086312052,
-      "loss": 46.0,
-      "step": 3877
-    },
-    {
-      "epoch": 0.6245017915375015,
-      "grad_norm": 0.0009481040760874748,
-      "learning_rate": 0.00019999980853221708,
-      "loss": 46.0,
-      "step": 3878
-    },
-    {
-      "epoch": 0.6246628286162889,
-      "grad_norm": 0.0003913877881132066,
-      "learning_rate": 0.0001999998084332034,
-      "loss": 46.0,
-      "step": 3879
-    },
-    {
-      "epoch": 0.6248238656950763,
-      "grad_norm": 0.0009294972405768931,
-      "learning_rate": 0.00019999980833416414,
-      "loss": 46.0,
-      "step": 3880
-    },
-    {
-      "epoch": 0.6249849027738636,
-      "grad_norm": 0.0006162965437397361,
-      "learning_rate": 0.00019999980823509926,
-      "loss": 46.0,
-      "step": 3881
-    },
-    {
-      "epoch": 0.6251459398526511,
-      "grad_norm": 0.00369398039765656,
-      "learning_rate": 0.0001999998081360088,
-      "loss": 46.0,
-      "step": 3882
-    },
-    {
-      "epoch": 0.6253069769314384,
-      "grad_norm": 0.0015181349590420723,
-      "learning_rate": 0.00019999980803689273,
-      "loss": 46.0,
-      "step": 3883
-    },
-    {
-      "epoch": 0.6254680140102259,
-      "grad_norm": 0.0005074756918475032,
-      "learning_rate": 0.00019999980793775107,
-      "loss": 46.0,
-      "step": 3884
-    },
-    {
-      "epoch": 0.6256290510890132,
-      "grad_norm": 0.00031046729418449104,
-      "learning_rate": 0.0001999998078385838,
-      "loss": 46.0,
-      "step": 3885
-    },
-    {
-      "epoch": 0.6257900881678007,
-      "grad_norm": 0.0006027091876603663,
-      "learning_rate": 0.00019999980773939097,
-      "loss": 46.0,
-      "step": 3886
-    },
-    {
-      "epoch": 0.625951125246588,
-      "grad_norm": 0.000491827551741153,
-      "learning_rate": 0.0001999998076401725,
-      "loss": 46.0,
-      "step": 3887
-    },
-    {
-      "epoch": 0.6261121623253754,
-      "grad_norm": 0.0005214366246946156,
-      "learning_rate": 0.00019999980754092846,
-      "loss": 46.0,
-      "step": 3888
-    },
-    {
-      "epoch": 0.6262731994041628,
-      "grad_norm": 0.0005825932021252811,
-      "learning_rate": 0.00019999980744165883,
-      "loss": 46.0,
-      "step": 3889
-    },
-    {
-      "epoch": 0.6264342364829502,
-      "grad_norm": 0.0003587489773053676,
-      "learning_rate": 0.00019999980734236358,
-      "loss": 46.0,
-      "step": 3890
-    },
-    {
-      "epoch": 0.6265952735617376,
-      "grad_norm": 0.0003709504962898791,
-      "learning_rate": 0.00019999980724304278,
-      "loss": 46.0,
-      "step": 3891
-    },
-    {
-      "epoch": 0.626756310640525,
-      "grad_norm": 0.0005646224017255008,
-      "learning_rate": 0.00019999980714369636,
-      "loss": 46.0,
-      "step": 3892
-    },
-    {
-      "epoch": 0.6269173477193124,
-      "grad_norm": 0.0005972803337499499,
-      "learning_rate": 0.00019999980704432435,
-      "loss": 46.0,
-      "step": 3893
-    },
-    {
-      "epoch": 0.6270783847980997,
-      "grad_norm": 0.0011230631498619914,
-      "learning_rate": 0.00019999980694492673,
-      "loss": 46.0,
-      "step": 3894
-    },
-    {
-      "epoch": 0.6272394218768872,
-      "grad_norm": 0.000535583880264312,
-      "learning_rate": 0.00019999980684550352,
-      "loss": 46.0,
-      "step": 3895
-    },
-    {
-      "epoch": 0.6274004589556745,
-      "grad_norm": 0.0016877740854397416,
-      "learning_rate": 0.00019999980674605472,
-      "loss": 46.0,
-      "step": 3896
-    },
-    {
-      "epoch": 0.627561496034462,
-      "grad_norm": 0.0014109365874901414,
-      "learning_rate": 0.00019999980664658034,
-      "loss": 46.0,
-      "step": 3897
-    },
-    {
-      "epoch": 0.6277225331132493,
-      "grad_norm": 0.001895105349831283,
-      "learning_rate": 0.00019999980654708034,
-      "loss": 46.0,
-      "step": 3898
-    },
-    {
-      "epoch": 0.6278835701920367,
-      "grad_norm": 0.000813856371678412,
-      "learning_rate": 0.00019999980644755473,
-      "loss": 46.0,
-      "step": 3899
-    },
-    {
-      "epoch": 0.6280446072708241,
-      "grad_norm": 0.0017831121804192662,
-      "learning_rate": 0.00019999980634800358,
-      "loss": 46.0,
-      "step": 3900
-    },
-    {
-      "epoch": 0.6282056443496115,
-      "grad_norm": 0.00043440848821774125,
-      "learning_rate": 0.0001999998062484268,
-      "loss": 46.0,
-      "step": 3901
-    },
-    {
-      "epoch": 0.6283666814283989,
-      "grad_norm": 0.0013220850378274918,
-      "learning_rate": 0.00019999980614882442,
-      "loss": 46.0,
-      "step": 3902
-    },
-    {
-      "epoch": 0.6285277185071863,
-      "grad_norm": 0.0003818328841589391,
-      "learning_rate": 0.00019999980604919646,
-      "loss": 46.0,
-      "step": 3903
-    },
-    {
-      "epoch": 0.6286887555859737,
-      "grad_norm": 0.0005465371068567038,
-      "learning_rate": 0.0001999998059495429,
-      "loss": 46.0,
-      "step": 3904
-    },
-    {
-      "epoch": 0.628849792664761,
-      "grad_norm": 0.003256515832617879,
-      "learning_rate": 0.00019999980584986377,
-      "loss": 46.0,
-      "step": 3905
-    },
-    {
-      "epoch": 0.6290108297435485,
-      "grad_norm": 0.0007257511606439948,
-      "learning_rate": 0.000199999805750159,
-      "loss": 46.0,
-      "step": 3906
-    },
-    {
-      "epoch": 0.6291718668223358,
-      "grad_norm": 0.0015296990750357509,
-      "learning_rate": 0.0001999998056504287,
-      "loss": 46.0,
-      "step": 3907
-    },
-    {
-      "epoch": 0.6293329039011233,
-      "grad_norm": 0.0016780991572886705,
-      "learning_rate": 0.00019999980555067274,
-      "loss": 46.0,
-      "step": 3908
-    },
-    {
-      "epoch": 0.6294939409799106,
-      "grad_norm": 0.0003995900333393365,
-      "learning_rate": 0.0001999998054508912,
-      "loss": 46.0,
-      "step": 3909
-    },
-    {
-      "epoch": 0.629654978058698,
-      "grad_norm": 0.0007608133601024747,
-      "learning_rate": 0.00019999980535108407,
-      "loss": 46.0,
-      "step": 3910
-    },
-    {
-      "epoch": 0.6298160151374854,
-      "grad_norm": 0.0023052096366882324,
-      "learning_rate": 0.00019999980525125136,
-      "loss": 46.0,
-      "step": 3911
-    },
-    {
-      "epoch": 0.6299770522162728,
-      "grad_norm": 0.0012004564050585032,
-      "learning_rate": 0.00019999980515139303,
-      "loss": 46.0,
-      "step": 3912
-    },
-    {
-      "epoch": 0.6301380892950602,
-      "grad_norm": 0.0005191197851672769,
-      "learning_rate": 0.00019999980505150912,
-      "loss": 46.0,
-      "step": 3913
-    },
-    {
-      "epoch": 0.6302991263738476,
-      "grad_norm": 0.0022934270091354847,
-      "learning_rate": 0.0001999998049515996,
-      "loss": 46.0,
-      "step": 3914
-    },
-    {
-      "epoch": 0.630460163452635,
-      "grad_norm": 0.00094611756503582,
-      "learning_rate": 0.00019999980485166453,
-      "loss": 46.0,
-      "step": 3915
-    },
-    {
-      "epoch": 0.6306212005314223,
-      "grad_norm": 0.0015847343020141125,
-      "learning_rate": 0.0001999998047517038,
-      "loss": 46.0,
-      "step": 3916
-    },
-    {
-      "epoch": 0.6307822376102098,
-      "grad_norm": 0.0008317606407217681,
-      "learning_rate": 0.00019999980465171754,
-      "loss": 46.0,
-      "step": 3917
-    },
-    {
-      "epoch": 0.6309432746889971,
-      "grad_norm": 0.0027236042078584433,
-      "learning_rate": 0.00019999980455170563,
-      "loss": 46.0,
-      "step": 3918
-    },
-    {
-      "epoch": 0.6311043117677846,
-      "grad_norm": 0.0015027662739157677,
-      "learning_rate": 0.00019999980445166817,
-      "loss": 46.0,
-      "step": 3919
-    },
-    {
-      "epoch": 0.6312653488465719,
-      "grad_norm": 0.0009333167690783739,
-      "learning_rate": 0.0001999998043516051,
-      "loss": 46.0,
-      "step": 3920
-    },
-    {
-      "epoch": 0.6314263859253593,
-      "grad_norm": 0.00038940150989219546,
-      "learning_rate": 0.0001999998042515164,
-      "loss": 46.0,
-      "step": 3921
-    },
-    {
-      "epoch": 0.6315874230041467,
-      "grad_norm": 0.0011956420494243503,
-      "learning_rate": 0.00019999980415140215,
-      "loss": 46.0,
-      "step": 3922
-    },
-    {
-      "epoch": 0.6317484600829341,
-      "grad_norm": 0.0017112076748162508,
-      "learning_rate": 0.00019999980405126228,
-      "loss": 46.0,
-      "step": 3923
-    },
-    {
-      "epoch": 0.6319094971617215,
-      "grad_norm": 0.0005666279466822743,
-      "learning_rate": 0.00019999980395109682,
-      "loss": 46.0,
-      "step": 3924
-    },
-    {
-      "epoch": 0.6320705342405089,
-      "grad_norm": 0.0002572282974142581,
-      "learning_rate": 0.0001999998038509058,
-      "loss": 46.0,
-      "step": 3925
-    },
-    {
-      "epoch": 0.6322315713192963,
-      "grad_norm": 0.0008163744932971895,
-      "learning_rate": 0.00019999980375068915,
-      "loss": 46.0,
-      "step": 3926
-    },
-    {
-      "epoch": 0.6323926083980836,
-      "grad_norm": 0.0007077203481458127,
-      "learning_rate": 0.0001999998036504469,
-      "loss": 46.0,
-      "step": 3927
-    },
-    {
-      "epoch": 0.6325536454768711,
-      "grad_norm": 0.0013686062302440405,
-      "learning_rate": 0.00019999980355017905,
-      "loss": 46.0,
-      "step": 3928
-    },
-    {
-      "epoch": 0.6327146825556584,
-      "grad_norm": 0.0018887012265622616,
-      "learning_rate": 0.00019999980344988564,
-      "loss": 46.0,
-      "step": 3929
-    },
-    {
-      "epoch": 0.6328757196344458,
-      "grad_norm": 0.0004644394794013351,
-      "learning_rate": 0.0001999998033495666,
-      "loss": 46.0,
-      "step": 3930
-    },
-    {
-      "epoch": 0.6330367567132332,
-      "grad_norm": 0.001086959382519126,
-      "learning_rate": 0.00019999980324922199,
-      "loss": 46.0,
-      "step": 3931
-    },
-    {
-      "epoch": 0.6331977937920206,
-      "grad_norm": 0.0012090452946722507,
-      "learning_rate": 0.00019999980314885178,
-      "loss": 46.0,
-      "step": 3932
-    },
-    {
-      "epoch": 0.633358830870808,
-      "grad_norm": 0.0019170187879353762,
-      "learning_rate": 0.000199999803048456,
-      "loss": 46.0,
-      "step": 3933
-    },
-    {
-      "epoch": 0.6335198679495954,
-      "grad_norm": 0.0011698042508214712,
-      "learning_rate": 0.00019999980294803458,
-      "loss": 46.0,
-      "step": 3934
-    },
-    {
-      "epoch": 0.6336809050283828,
-      "grad_norm": 0.0004384565108921379,
-      "learning_rate": 0.00019999980284758756,
-      "loss": 46.0,
-      "step": 3935
-    },
-    {
-      "epoch": 0.6338419421071702,
-      "grad_norm": 0.0014051563339307904,
-      "learning_rate": 0.00019999980274711498,
-      "loss": 46.0,
-      "step": 3936
-    },
-    {
-      "epoch": 0.6340029791859576,
-      "grad_norm": 0.00027268854319117963,
-      "learning_rate": 0.0001999998026466168,
-      "loss": 46.0,
-      "step": 3937
-    },
-    {
-      "epoch": 0.634164016264745,
-      "grad_norm": 0.0009903289610520005,
-      "learning_rate": 0.000199999802546093,
-      "loss": 46.0,
-      "step": 3938
-    },
-    {
-      "epoch": 0.6343250533435324,
-      "grad_norm": 0.00039567198837175965,
-      "learning_rate": 0.00019999980244554363,
-      "loss": 46.0,
-      "step": 3939
-    },
-    {
-      "epoch": 0.6344860904223197,
-      "grad_norm": 0.0019074567826464772,
-      "learning_rate": 0.00019999980234496867,
-      "loss": 46.0,
-      "step": 3940
-    },
-    {
-      "epoch": 0.6346471275011071,
-      "grad_norm": 0.0004465160018298775,
-      "learning_rate": 0.0001999998022443681,
-      "loss": 46.0,
-      "step": 3941
-    },
-    {
-      "epoch": 0.6348081645798945,
-      "grad_norm": 0.000665780738927424,
-      "learning_rate": 0.00019999980214374194,
-      "loss": 46.0,
-      "step": 3942
-    },
-    {
-      "epoch": 0.6349692016586819,
-      "grad_norm": 0.002296403981745243,
-      "learning_rate": 0.00019999980204309017,
-      "loss": 46.0,
-      "step": 3943
-    },
-    {
-      "epoch": 0.6351302387374693,
-      "grad_norm": 0.0014289953978732228,
-      "learning_rate": 0.00019999980194241283,
-      "loss": 46.0,
-      "step": 3944
-    },
-    {
-      "epoch": 0.6352912758162567,
-      "grad_norm": 0.0019106982508674264,
-      "learning_rate": 0.00019999980184170989,
-      "loss": 46.0,
-      "step": 3945
-    },
-    {
-      "epoch": 0.6354523128950441,
-      "grad_norm": 0.00034020375460386276,
-      "learning_rate": 0.00019999980174098135,
-      "loss": 46.0,
-      "step": 3946
-    },
-    {
-      "epoch": 0.6356133499738315,
-      "grad_norm": 0.00044324115151539445,
-      "learning_rate": 0.00019999980164022723,
-      "loss": 46.0,
-      "step": 3947
-    },
-    {
-      "epoch": 0.6357743870526189,
-      "grad_norm": 0.001357447705231607,
-      "learning_rate": 0.0001999998015394475,
-      "loss": 46.0,
-      "step": 3948
-    },
-    {
-      "epoch": 0.6359354241314062,
-      "grad_norm": 0.0007523288950324059,
-      "learning_rate": 0.00019999980143864217,
-      "loss": 46.0,
-      "step": 3949
-    },
-    {
-      "epoch": 0.6360964612101937,
-      "grad_norm": 0.0007136139320209622,
-      "learning_rate": 0.00019999980133781123,
-      "loss": 46.0,
-      "step": 3950
-    },
-    {
-      "epoch": 0.636257498288981,
-      "grad_norm": 0.0017579509876668453,
-      "learning_rate": 0.0001999998012369547,
-      "loss": 46.0,
-      "step": 3951
-    },
-    {
-      "epoch": 0.6364185353677684,
-      "grad_norm": 0.00034304361906833947,
-      "learning_rate": 0.0001999998011360726,
-      "loss": 46.0,
-      "step": 3952
-    },
-    {
-      "epoch": 0.6365795724465558,
-      "grad_norm": 0.0014558410039171576,
-      "learning_rate": 0.00019999980103516492,
-      "loss": 46.0,
-      "step": 3953
-    },
-    {
-      "epoch": 0.6367406095253432,
-      "grad_norm": 0.0004298131098039448,
-      "learning_rate": 0.0001999998009342316,
-      "loss": 46.0,
-      "step": 3954
-    },
-    {
-      "epoch": 0.6369016466041306,
-      "grad_norm": 0.0006449212087318301,
-      "learning_rate": 0.0001999998008332727,
-      "loss": 46.0,
-      "step": 3955
-    },
-    {
-      "epoch": 0.637062683682918,
-      "grad_norm": 0.0004265451862011105,
-      "learning_rate": 0.00019999980073228822,
-      "loss": 46.0,
-      "step": 3956
-    },
-    {
-      "epoch": 0.6372237207617054,
-      "grad_norm": 0.0030151167884469032,
-      "learning_rate": 0.00019999980063127815,
-      "loss": 46.0,
-      "step": 3957
-    },
-    {
-      "epoch": 0.6373847578404928,
-      "grad_norm": 0.002641751430928707,
-      "learning_rate": 0.00019999980053024246,
-      "loss": 46.0,
-      "step": 3958
-    },
-    {
-      "epoch": 0.6375457949192802,
-      "grad_norm": 0.0006884320755489171,
-      "learning_rate": 0.0001999998004291812,
-      "loss": 46.0,
-      "step": 3959
-    },
-    {
-      "epoch": 0.6377068319980675,
-      "grad_norm": 0.000506981392391026,
-      "learning_rate": 0.00019999980032809435,
-      "loss": 46.0,
-      "step": 3960
-    },
-    {
-      "epoch": 0.637867869076855,
-      "grad_norm": 0.001656600390560925,
-      "learning_rate": 0.00019999980022698187,
-      "loss": 46.0,
-      "step": 3961
-    },
-    {
-      "epoch": 0.6380289061556423,
-      "grad_norm": 0.001254578004591167,
-      "learning_rate": 0.0001999998001258438,
-      "loss": 46.0,
-      "step": 3962
-    },
-    {
-      "epoch": 0.6381899432344297,
-      "grad_norm": 0.0005974529194645584,
-      "learning_rate": 0.00019999980002468015,
-      "loss": 46.0,
-      "step": 3963
-    },
-    {
-      "epoch": 0.6383509803132171,
-      "grad_norm": 0.001432158867828548,
-      "learning_rate": 0.0001999997999234909,
-      "loss": 46.0,
-      "step": 3964
-    },
-    {
-      "epoch": 0.6385120173920045,
-      "grad_norm": 0.00041471130680292845,
-      "learning_rate": 0.00019999979982227606,
-      "loss": 46.0,
-      "step": 3965
-    },
-    {
-      "epoch": 0.6386730544707919,
-      "grad_norm": 0.0004181638068985194,
-      "learning_rate": 0.00019999979972103562,
-      "loss": 46.0,
-      "step": 3966
-    },
-    {
-      "epoch": 0.6388340915495793,
-      "grad_norm": 0.0012539547169581056,
-      "learning_rate": 0.0001999997996197696,
-      "loss": 46.0,
-      "step": 3967
-    },
-    {
-      "epoch": 0.6389951286283667,
-      "grad_norm": 0.000602704705670476,
-      "learning_rate": 0.00019999979951847795,
-      "loss": 46.0,
-      "step": 3968
-    },
-    {
-      "epoch": 0.6391561657071541,
-      "grad_norm": 0.0007977702771313488,
-      "learning_rate": 0.00019999979941716074,
-      "loss": 46.0,
-      "step": 3969
-    },
-    {
-      "epoch": 0.6393172027859415,
-      "grad_norm": 0.0003835688694380224,
-      "learning_rate": 0.00019999979931581793,
-      "loss": 46.0,
-      "step": 3970
-    },
-    {
-      "epoch": 0.6394782398647288,
-      "grad_norm": 0.0012510454980656505,
-      "learning_rate": 0.00019999979921444952,
-      "loss": 46.0,
-      "step": 3971
-    },
-    {
-      "epoch": 0.6396392769435163,
-      "grad_norm": 0.0006620041676796973,
-      "learning_rate": 0.0001999997991130555,
-      "loss": 46.0,
-      "step": 3972
-    },
-    {
-      "epoch": 0.6398003140223036,
-      "grad_norm": 0.001054050400853157,
-      "learning_rate": 0.00019999979901163593,
-      "loss": 46.0,
-      "step": 3973
-    },
-    {
-      "epoch": 0.639961351101091,
-      "grad_norm": 0.00041232031071558595,
-      "learning_rate": 0.0001999997989101907,
-      "loss": 46.0,
-      "step": 3974
-    },
-    {
-      "epoch": 0.6401223881798784,
-      "grad_norm": 0.0028472826816141605,
-      "learning_rate": 0.00019999979880871993,
-      "loss": 46.0,
-      "step": 3975
-    },
-    {
-      "epoch": 0.6402834252586658,
-      "grad_norm": 0.0006197105394676328,
-      "learning_rate": 0.00019999979870722354,
-      "loss": 46.0,
-      "step": 3976
-    },
-    {
-      "epoch": 0.6404444623374532,
-      "grad_norm": 0.0006238263449631631,
-      "learning_rate": 0.00019999979860570158,
-      "loss": 46.0,
-      "step": 3977
-    },
-    {
-      "epoch": 0.6406054994162406,
-      "grad_norm": 0.0005026328144595027,
-      "learning_rate": 0.000199999798504154,
-      "loss": 46.0,
-      "step": 3978
-    },
-    {
-      "epoch": 0.640766536495028,
-      "grad_norm": 0.0004826491349376738,
-      "learning_rate": 0.0001999997984025808,
-      "loss": 46.0,
-      "step": 3979
-    },
-    {
-      "epoch": 0.6409275735738154,
-      "grad_norm": 0.0003820724377874285,
-      "learning_rate": 0.00019999979830098206,
-      "loss": 46.0,
-      "step": 3980
-    },
-    {
-      "epoch": 0.6410886106526028,
-      "grad_norm": 0.0014432755997404456,
-      "learning_rate": 0.00019999979819935768,
-      "loss": 46.0,
-      "step": 3981
-    },
-    {
-      "epoch": 0.6412496477313901,
-      "grad_norm": 0.0013067788677290082,
-      "learning_rate": 0.00019999979809770773,
-      "loss": 46.0,
-      "step": 3982
-    },
-    {
-      "epoch": 0.6414106848101775,
-      "grad_norm": 0.002178050111979246,
-      "learning_rate": 0.0001999997979960322,
-      "loss": 46.0,
-      "step": 3983
-    },
-    {
-      "epoch": 0.6415717218889649,
-      "grad_norm": 0.0004842475464101881,
-      "learning_rate": 0.00019999979789433106,
-      "loss": 46.0,
-      "step": 3984
-    },
-    {
-      "epoch": 0.6417327589677523,
-      "grad_norm": 0.000640039099380374,
-      "learning_rate": 0.0001999997977926043,
-      "loss": 46.0,
-      "step": 3985
-    },
-    {
-      "epoch": 0.6418937960465397,
-      "grad_norm": 0.0008534935768693686,
-      "learning_rate": 0.00019999979769085198,
-      "loss": 46.0,
-      "step": 3986
-    },
-    {
-      "epoch": 0.6420548331253271,
-      "grad_norm": 0.0008801156654953957,
-      "learning_rate": 0.00019999979758907404,
-      "loss": 46.0,
-      "step": 3987
-    },
-    {
-      "epoch": 0.6422158702041145,
-      "grad_norm": 0.0003403981390874833,
-      "learning_rate": 0.00019999979748727052,
-      "loss": 46.0,
-      "step": 3988
-    },
-    {
-      "epoch": 0.6423769072829019,
-      "grad_norm": 0.0014114284422248602,
-      "learning_rate": 0.00019999979738544142,
-      "loss": 46.0,
-      "step": 3989
-    },
-    {
-      "epoch": 0.6425379443616893,
-      "grad_norm": 0.0005789893912151456,
-      "learning_rate": 0.0001999997972835867,
-      "loss": 46.0,
-      "step": 3990
-    },
-    {
-      "epoch": 0.6426989814404767,
-      "grad_norm": 0.0006970431422814727,
-      "learning_rate": 0.00019999979718170638,
-      "loss": 46.0,
-      "step": 3991
-    },
-    {
-      "epoch": 0.6428600185192641,
-      "grad_norm": 0.00064929632935673,
-      "learning_rate": 0.00019999979707980049,
-      "loss": 46.0,
-      "step": 3992
-    },
-    {
-      "epoch": 0.6430210555980514,
-      "grad_norm": 0.00061324069974944,
-      "learning_rate": 0.000199999796977869,
-      "loss": 46.0,
-      "step": 3993
-    },
-    {
-      "epoch": 0.6431820926768388,
-      "grad_norm": 0.00045321404468268156,
-      "learning_rate": 0.00019999979687591188,
-      "loss": 46.0,
-      "step": 3994
-    },
-    {
-      "epoch": 0.6433431297556262,
-      "grad_norm": 0.0008014366612769663,
-      "learning_rate": 0.0001999997967739292,
-      "loss": 46.0,
-      "step": 3995
-    },
-    {
-      "epoch": 0.6435041668344136,
-      "grad_norm": 0.0003293314657639712,
-      "learning_rate": 0.00019999979667192092,
-      "loss": 46.0,
-      "step": 3996
-    },
-    {
-      "epoch": 0.643665203913201,
-      "grad_norm": 0.001550797838717699,
-      "learning_rate": 0.00019999979656988703,
-      "loss": 46.0,
-      "step": 3997
-    },
-    {
-      "epoch": 0.6438262409919884,
-      "grad_norm": 0.00034418125869706273,
-      "learning_rate": 0.00019999979646782758,
-      "loss": 46.0,
-      "step": 3998
-    },
-    {
-      "epoch": 0.6439872780707758,
-      "grad_norm": 0.000606780347879976,
-      "learning_rate": 0.0001999997963657425,
-      "loss": 46.0,
-      "step": 3999
-    },
-    {
-      "epoch": 0.6441483151495632,
-      "grad_norm": 0.000773335515987128,
-      "learning_rate": 0.00019999979626363185,
-      "loss": 46.0,
-      "step": 4000
-    },
-    {
-      "epoch": 0.6443093522283506,
-      "grad_norm": 0.00028685005963779986,
-      "learning_rate": 0.00019999979616149558,
-      "loss": 46.0,
-      "step": 4001
-    },
-    {
-      "epoch": 0.644470389307138,
-      "grad_norm": 0.0018651674035936594,
-      "learning_rate": 0.00019999979605933373,
-      "loss": 46.0,
-      "step": 4002
-    },
-    {
-      "epoch": 0.6446314263859254,
-      "grad_norm": 0.0023687323555350304,
-      "learning_rate": 0.0001999997959571463,
-      "loss": 46.0,
-      "step": 4003
-    },
-    {
-      "epoch": 0.6447924634647128,
-      "grad_norm": 0.0006122443592175841,
-      "learning_rate": 0.00019999979585493325,
-      "loss": 46.0,
-      "step": 4004
-    },
-    {
-      "epoch": 0.6449535005435001,
-      "grad_norm": 0.0003398013941477984,
-      "learning_rate": 0.00019999979575269463,
-      "loss": 46.0,
-      "step": 4005
-    },
-    {
-      "epoch": 0.6451145376222875,
-      "grad_norm": 0.0005419240333139896,
-      "learning_rate": 0.00019999979565043038,
-      "loss": 46.0,
-      "step": 4006
-    },
-    {
-      "epoch": 0.6452755747010749,
-      "grad_norm": 0.0004764005425386131,
-      "learning_rate": 0.00019999979554814057,
-      "loss": 46.0,
-      "step": 4007
-    },
-    {
-      "epoch": 0.6454366117798623,
-      "grad_norm": 0.000622907595243305,
-      "learning_rate": 0.00019999979544582517,
-      "loss": 46.0,
-      "step": 4008
-    },
-    {
-      "epoch": 0.6455976488586497,
-      "grad_norm": 0.0005630090017803013,
-      "learning_rate": 0.00019999979534348415,
-      "loss": 46.0,
-      "step": 4009
-    },
-    {
-      "epoch": 0.6457586859374371,
-      "grad_norm": 0.0012180102057754993,
-      "learning_rate": 0.00019999979524111753,
-      "loss": 46.0,
-      "step": 4010
-    },
-    {
-      "epoch": 0.6459197230162245,
-      "grad_norm": 0.0009136570733971894,
-      "learning_rate": 0.0001999997951387253,
-      "loss": 46.0,
-      "step": 4011
-    },
-    {
-      "epoch": 0.6460807600950119,
-      "grad_norm": 0.0005233532865531743,
-      "learning_rate": 0.00019999979503630753,
-      "loss": 46.0,
-      "step": 4012
-    },
-    {
-      "epoch": 0.6462417971737993,
-      "grad_norm": 0.0008663739426992834,
-      "learning_rate": 0.00019999979493386414,
-      "loss": 46.0,
-      "step": 4013
-    },
-    {
-      "epoch": 0.6464028342525867,
-      "grad_norm": 0.0011348624248057604,
-      "learning_rate": 0.00019999979483139514,
-      "loss": 46.0,
-      "step": 4014
-    },
-    {
-      "epoch": 0.646563871331374,
-      "grad_norm": 0.001946570468135178,
-      "learning_rate": 0.00019999979472890055,
-      "loss": 46.0,
-      "step": 4015
-    },
-    {
-      "epoch": 0.6467249084101614,
-      "grad_norm": 0.0005715051665902138,
-      "learning_rate": 0.0001999997946263804,
-      "loss": 46.0,
-      "step": 4016
-    },
-    {
-      "epoch": 0.6468859454889488,
-      "grad_norm": 0.000543768925126642,
-      "learning_rate": 0.0001999997945238346,
-      "loss": 46.0,
-      "step": 4017
-    },
-    {
-      "epoch": 0.6470469825677362,
-      "grad_norm": 0.000579813786316663,
-      "learning_rate": 0.00019999979442126325,
-      "loss": 46.0,
-      "step": 4018
-    },
-    {
-      "epoch": 0.6472080196465236,
-      "grad_norm": 0.0011095544323325157,
-      "learning_rate": 0.00019999979431866626,
-      "loss": 46.0,
-      "step": 4019
-    },
-    {
-      "epoch": 0.647369056725311,
-      "grad_norm": 0.0003147822862956673,
-      "learning_rate": 0.00019999979421604373,
-      "loss": 46.0,
-      "step": 4020
-    },
-    {
-      "epoch": 0.6475300938040984,
-      "grad_norm": 0.0009261103696189821,
-      "learning_rate": 0.00019999979411339559,
-      "loss": 46.0,
-      "step": 4021
-    },
-    {
-      "epoch": 0.6476911308828858,
-      "grad_norm": 0.0006047890055924654,
-      "learning_rate": 0.00019999979401072183,
-      "loss": 46.0,
-      "step": 4022
-    },
-    {
-      "epoch": 0.6478521679616732,
-      "grad_norm": 0.0015719252405688167,
-      "learning_rate": 0.0001999997939080225,
-      "loss": 46.0,
-      "step": 4023
-    },
-    {
-      "epoch": 0.6480132050404606,
-      "grad_norm": 0.003473407356068492,
-      "learning_rate": 0.00019999979380529753,
-      "loss": 46.0,
-      "step": 4024
-    },
-    {
-      "epoch": 0.648174242119248,
-      "grad_norm": 0.0007352083339355886,
-      "learning_rate": 0.000199999793702547,
-      "loss": 46.0,
-      "step": 4025
-    },
-    {
-      "epoch": 0.6483352791980354,
-      "grad_norm": 0.0004880555206909776,
-      "learning_rate": 0.00019999979359977088,
-      "loss": 46.0,
-      "step": 4026
-    },
-    {
-      "epoch": 0.6484963162768227,
-      "grad_norm": 0.001536658382974565,
-      "learning_rate": 0.00019999979349696916,
-      "loss": 46.0,
-      "step": 4027
-    },
-    {
-      "epoch": 0.6486573533556101,
-      "grad_norm": 0.0005246897926554084,
-      "learning_rate": 0.00019999979339414186,
-      "loss": 46.0,
-      "step": 4028
-    },
-    {
-      "epoch": 0.6488183904343975,
-      "grad_norm": 0.0006400058628059924,
-      "learning_rate": 0.00019999979329128893,
-      "loss": 46.0,
-      "step": 4029
-    },
-    {
-      "epoch": 0.6489794275131849,
-      "grad_norm": 0.0013222291599959135,
-      "learning_rate": 0.00019999979318841043,
-      "loss": 46.0,
-      "step": 4030
-    },
-    {
-      "epoch": 0.6491404645919723,
-      "grad_norm": 0.0004871092096436769,
-      "learning_rate": 0.00019999979308550633,
-      "loss": 46.0,
-      "step": 4031
-    },
-    {
-      "epoch": 0.6493015016707597,
-      "grad_norm": 0.0011869113659486175,
-      "learning_rate": 0.00019999979298257662,
-      "loss": 46.0,
-      "step": 4032
-    },
-    {
-      "epoch": 0.6494625387495471,
-      "grad_norm": 0.002591888653114438,
-      "learning_rate": 0.00019999979287962133,
-      "loss": 46.0,
-      "step": 4033
-    },
-    {
-      "epoch": 0.6496235758283345,
-      "grad_norm": 0.00039742272929288447,
-      "learning_rate": 0.00019999979277664047,
-      "loss": 46.0,
-      "step": 4034
-    },
-    {
-      "epoch": 0.6497846129071219,
-      "grad_norm": 0.00040701148100197315,
-      "learning_rate": 0.00019999979267363397,
-      "loss": 46.0,
-      "step": 4035
-    },
-    {
-      "epoch": 0.6499456499859092,
-      "grad_norm": 0.00034645546111278236,
-      "learning_rate": 0.00019999979257060192,
-      "loss": 46.0,
-      "step": 4036
-    },
-    {
-      "epoch": 0.6501066870646967,
-      "grad_norm": 0.0005740129272453487,
-      "learning_rate": 0.00019999979246754424,
-      "loss": 46.0,
-      "step": 4037
-    },
-    {
-      "epoch": 0.650267724143484,
-      "grad_norm": 0.0013352630194276571,
-      "learning_rate": 0.00019999979236446098,
-      "loss": 46.0,
-      "step": 4038
-    },
-    {
-      "epoch": 0.6504287612222714,
-      "grad_norm": 0.0011539635015651584,
-      "learning_rate": 0.00019999979226135214,
-      "loss": 46.0,
-      "step": 4039
-    },
-    {
-      "epoch": 0.6505897983010588,
-      "grad_norm": 0.0005610228981822729,
-      "learning_rate": 0.00019999979215821768,
-      "loss": 46.0,
-      "step": 4040
-    },
-    {
-      "epoch": 0.6507508353798462,
-      "grad_norm": 0.0004671792557928711,
-      "learning_rate": 0.00019999979205505763,
-      "loss": 46.0,
-      "step": 4041
-    },
-    {
-      "epoch": 0.6509118724586336,
-      "grad_norm": 0.0005615011905319989,
-      "learning_rate": 0.000199999791951872,
-      "loss": 46.0,
-      "step": 4042
-    },
-    {
-      "epoch": 0.651072909537421,
-      "grad_norm": 0.0011466463329270482,
-      "learning_rate": 0.00019999979184866075,
-      "loss": 46.0,
-      "step": 4043
-    },
-    {
-      "epoch": 0.6512339466162084,
-      "grad_norm": 0.0005127380136400461,
-      "learning_rate": 0.0001999997917454239,
-      "loss": 46.0,
-      "step": 4044
-    },
-    {
-      "epoch": 0.6513949836949958,
-      "grad_norm": 0.0005644343327730894,
-      "learning_rate": 0.00019999979164216146,
-      "loss": 46.0,
-      "step": 4045
-    },
-    {
-      "epoch": 0.6515560207737832,
-      "grad_norm": 0.0009933200199157,
-      "learning_rate": 0.00019999979153887348,
-      "loss": 46.0,
-      "step": 4046
-    },
-    {
-      "epoch": 0.6517170578525705,
-      "grad_norm": 0.0008299188921228051,
-      "learning_rate": 0.00019999979143555985,
-      "loss": 46.0,
-      "step": 4047
-    },
-    {
-      "epoch": 0.651878094931358,
-      "grad_norm": 0.0002949825720861554,
-      "learning_rate": 0.00019999979133222064,
-      "loss": 46.0,
-      "step": 4048
-    },
-    {
-      "epoch": 0.6520391320101453,
-      "grad_norm": 0.0014827088452875614,
-      "learning_rate": 0.0001999997912288558,
-      "loss": 46.0,
-      "step": 4049
-    },
-    {
-      "epoch": 0.6522001690889327,
-      "grad_norm": 0.0003980410401709378,
-      "learning_rate": 0.0001999997911254654,
-      "loss": 46.0,
-      "step": 4050
-    },
-    {
-      "epoch": 0.6523612061677201,
-      "grad_norm": 0.001480201375670731,
-      "learning_rate": 0.00019999979102204942,
-      "loss": 46.0,
-      "step": 4051
-    },
-    {
-      "epoch": 0.6525222432465075,
-      "grad_norm": 0.0009770323522388935,
-      "learning_rate": 0.00019999979091860784,
-      "loss": 46.0,
-      "step": 4052
-    },
-    {
-      "epoch": 0.6526832803252949,
-      "grad_norm": 0.001219235360622406,
-      "learning_rate": 0.00019999979081514066,
-      "loss": 46.0,
-      "step": 4053
-    },
-    {
-      "epoch": 0.6528443174040823,
-      "grad_norm": 0.0005848513683304191,
-      "learning_rate": 0.00019999979071164787,
-      "loss": 46.0,
-      "step": 4054
-    },
-    {
-      "epoch": 0.6530053544828697,
-      "grad_norm": 0.0008878202643245459,
-      "learning_rate": 0.0001999997906081295,
-      "loss": 46.0,
-      "step": 4055
-    },
-    {
-      "epoch": 0.6531663915616571,
-      "grad_norm": 0.00041669944766908884,
-      "learning_rate": 0.00019999979050458553,
-      "loss": 46.0,
-      "step": 4056
-    },
-    {
-      "epoch": 0.6533274286404445,
-      "grad_norm": 0.0007548247813247144,
-      "learning_rate": 0.00019999979040101598,
-      "loss": 46.0,
-      "step": 4057
-    },
-    {
-      "epoch": 0.6534884657192318,
-      "grad_norm": 0.00044422075734473765,
-      "learning_rate": 0.0001999997902974208,
-      "loss": 46.0,
-      "step": 4058
-    },
-    {
-      "epoch": 0.6536495027980193,
-      "grad_norm": 0.0004771124222315848,
-      "learning_rate": 0.00019999979019380004,
-      "loss": 46.0,
-      "step": 4059
-    },
-    {
-      "epoch": 0.6538105398768066,
-      "grad_norm": 0.0005559561541303992,
-      "learning_rate": 0.0001999997900901537,
-      "loss": 46.0,
-      "step": 4060
-    },
-    {
-      "epoch": 0.653971576955594,
-      "grad_norm": 0.0031795059330761433,
-      "learning_rate": 0.00019999978998648174,
-      "loss": 46.0,
-      "step": 4061
-    },
-    {
-      "epoch": 0.6541326140343814,
-      "grad_norm": 0.0013497446198016405,
-      "learning_rate": 0.0001999997898827842,
-      "loss": 46.0,
-      "step": 4062
-    },
-    {
-      "epoch": 0.6542936511131688,
-      "grad_norm": 0.004771994426846504,
-      "learning_rate": 0.00019999978977906107,
-      "loss": 46.0,
-      "step": 4063
-    },
-    {
-      "epoch": 0.6544546881919562,
-      "grad_norm": 0.0005237840232439339,
-      "learning_rate": 0.00019999978967531233,
-      "loss": 46.0,
-      "step": 4064
-    },
-    {
-      "epoch": 0.6546157252707436,
-      "grad_norm": 0.0022218686062842607,
-      "learning_rate": 0.00019999978957153803,
-      "loss": 46.0,
-      "step": 4065
-    },
-    {
-      "epoch": 0.654776762349531,
-      "grad_norm": 0.0006077251164242625,
-      "learning_rate": 0.0001999997894677381,
-      "loss": 46.0,
-      "step": 4066
-    },
-    {
-      "epoch": 0.6549377994283184,
-      "grad_norm": 0.003656107233837247,
-      "learning_rate": 0.00019999978936391258,
-      "loss": 46.0,
-      "step": 4067
-    },
-    {
-      "epoch": 0.6550988365071058,
-      "grad_norm": 0.0009901061421260238,
-      "learning_rate": 0.0001999997892600615,
-      "loss": 46.0,
-      "step": 4068
-    },
-    {
-      "epoch": 0.6552598735858931,
-      "grad_norm": 0.0008633272955194116,
-      "learning_rate": 0.00019999978915618478,
-      "loss": 46.0,
-      "step": 4069
-    },
-    {
-      "epoch": 0.6554209106646806,
-      "grad_norm": 0.0005279080360196531,
-      "learning_rate": 0.0001999997890522825,
-      "loss": 46.0,
-      "step": 4070
-    },
-    {
-      "epoch": 0.6555819477434679,
-      "grad_norm": 0.0006885799812152982,
-      "learning_rate": 0.0001999997889483546,
-      "loss": 46.0,
-      "step": 4071
-    },
-    {
-      "epoch": 0.6557429848222553,
-      "grad_norm": 0.0006982874474488199,
-      "learning_rate": 0.0001999997888444011,
-      "loss": 46.0,
-      "step": 4072
-    },
-    {
-      "epoch": 0.6559040219010427,
-      "grad_norm": 0.0008390130242332816,
-      "learning_rate": 0.00019999978874042204,
-      "loss": 46.0,
-      "step": 4073
-    },
-    {
-      "epoch": 0.6560650589798301,
-      "grad_norm": 0.0003052345127798617,
-      "learning_rate": 0.00019999978863641734,
-      "loss": 46.0,
-      "step": 4074
-    },
-    {
-      "epoch": 0.6562260960586175,
-      "grad_norm": 0.0006468112114816904,
-      "learning_rate": 0.00019999978853238708,
-      "loss": 46.0,
-      "step": 4075
-    },
-    {
-      "epoch": 0.6563871331374049,
-      "grad_norm": 0.000711954606231302,
-      "learning_rate": 0.00019999978842833122,
-      "loss": 46.0,
-      "step": 4076
-    },
-    {
-      "epoch": 0.6565481702161923,
-      "grad_norm": 0.00034803905873559415,
-      "learning_rate": 0.00019999978832424976,
-      "loss": 46.0,
-      "step": 4077
-    },
-    {
-      "epoch": 0.6567092072949797,
-      "grad_norm": 0.0005067437305115163,
-      "learning_rate": 0.00019999978822014266,
-      "loss": 46.0,
-      "step": 4078
-    },
-    {
-      "epoch": 0.6568702443737671,
-      "grad_norm": 0.0004636379308067262,
-      "learning_rate": 0.00019999978811601003,
-      "loss": 46.0,
-      "step": 4079
-    },
-    {
-      "epoch": 0.6570312814525544,
-      "grad_norm": 0.0005774891469627619,
-      "learning_rate": 0.00019999978801185178,
-      "loss": 46.0,
-      "step": 4080
-    },
-    {
-      "epoch": 0.6571923185313419,
-      "grad_norm": 0.0004242188297212124,
-      "learning_rate": 0.00019999978790766795,
-      "loss": 46.0,
-      "step": 4081
-    },
-    {
-      "epoch": 0.6573533556101292,
-      "grad_norm": 0.002381916856393218,
-      "learning_rate": 0.0001999997878034585,
-      "loss": 46.0,
-      "step": 4082
-    },
-    {
-      "epoch": 0.6575143926889166,
-      "grad_norm": 0.0035492258612066507,
-      "learning_rate": 0.0001999997876992235,
-      "loss": 46.0,
-      "step": 4083
-    },
-    {
-      "epoch": 0.657675429767704,
-      "grad_norm": 0.0010473319562152028,
-      "learning_rate": 0.00019999978759496285,
-      "loss": 46.0,
-      "step": 4084
-    },
-    {
-      "epoch": 0.6578364668464914,
-      "grad_norm": 0.0007973707397468388,
-      "learning_rate": 0.00019999978749067664,
-      "loss": 46.0,
-      "step": 4085
-    },
-    {
-      "epoch": 0.6579975039252788,
-      "grad_norm": 0.0007441894267685711,
-      "learning_rate": 0.00019999978738636482,
-      "loss": 46.0,
-      "step": 4086
-    },
-    {
-      "epoch": 0.6581585410040662,
-      "grad_norm": 0.000384774204576388,
-      "learning_rate": 0.0001999997872820274,
-      "loss": 46.0,
-      "step": 4087
-    },
-    {
-      "epoch": 0.6583195780828536,
-      "grad_norm": 0.0011437773937359452,
-      "learning_rate": 0.0001999997871776644,
-      "loss": 46.0,
-      "step": 4088
-    },
-    {
-      "epoch": 0.6584806151616409,
-      "grad_norm": 0.002420299919322133,
-      "learning_rate": 0.0001999997870732758,
-      "loss": 46.0,
-      "step": 4089
-    },
-    {
-      "epoch": 0.6586416522404284,
-      "grad_norm": 0.002654311014339328,
-      "learning_rate": 0.00019999978696886163,
-      "loss": 46.0,
-      "step": 4090
-    },
-    {
-      "epoch": 0.6588026893192157,
-      "grad_norm": 0.0023667917121201754,
-      "learning_rate": 0.0001999997868644218,
-      "loss": 46.0,
-      "step": 4091
-    },
-    {
-      "epoch": 0.6589637263980032,
-      "grad_norm": 0.0006622622604481876,
-      "learning_rate": 0.00019999978675995644,
-      "loss": 46.0,
-      "step": 4092
-    },
-    {
-      "epoch": 0.6591247634767905,
-      "grad_norm": 0.000863416469655931,
-      "learning_rate": 0.00019999978665546545,
-      "loss": 46.0,
-      "step": 4093
-    },
-    {
-      "epoch": 0.659285800555578,
-      "grad_norm": 0.00053859205218032,
-      "learning_rate": 0.00019999978655094888,
-      "loss": 46.0,
-      "step": 4094
-    },
-    {
-      "epoch": 0.6594468376343653,
-      "grad_norm": 0.001705170376226306,
-      "learning_rate": 0.00019999978644640672,
-      "loss": 46.0,
-      "step": 4095
-    },
-    {
-      "epoch": 0.6596078747131527,
-      "grad_norm": 0.0013821489410474896,
-      "learning_rate": 0.00019999978634183897,
-      "loss": 46.0,
-      "step": 4096
-    },
-    {
-      "epoch": 0.6597689117919401,
-      "grad_norm": 0.0011414185864850879,
-      "learning_rate": 0.0001999997862372456,
-      "loss": 46.0,
-      "step": 4097
-    },
-    {
-      "epoch": 0.6599299488707275,
-      "grad_norm": 0.0017310691764578223,
-      "learning_rate": 0.00019999978613262665,
-      "loss": 46.0,
-      "step": 4098
-    },
-    {
-      "epoch": 0.6600909859495149,
-      "grad_norm": 0.0008604437462054193,
-      "learning_rate": 0.0001999997860279821,
-      "loss": 46.0,
-      "step": 4099
-    },
-    {
-      "epoch": 0.6602520230283022,
-      "grad_norm": 0.0005920581752434373,
-      "learning_rate": 0.00019999978592331194,
-      "loss": 46.0,
-      "step": 4100
-    },
-    {
-      "epoch": 0.6604130601070897,
-      "grad_norm": 0.000564577232580632,
-      "learning_rate": 0.00019999978581861623,
-      "loss": 46.0,
-      "step": 4101
-    },
-    {
-      "epoch": 0.660574097185877,
-      "grad_norm": 0.0005751004791818559,
-      "learning_rate": 0.00019999978571389487,
-      "loss": 46.0,
-      "step": 4102
-    },
-    {
-      "epoch": 0.6607351342646645,
-      "grad_norm": 0.0014589970232918859,
-      "learning_rate": 0.00019999978560914796,
-      "loss": 46.0,
-      "step": 4103
-    },
-    {
-      "epoch": 0.6608961713434518,
-      "grad_norm": 0.0027970534283667803,
-      "learning_rate": 0.00019999978550437543,
-      "loss": 46.0,
-      "step": 4104
-    },
-    {
-      "epoch": 0.6610572084222393,
-      "grad_norm": 0.0016785872867330909,
-      "learning_rate": 0.00019999978539957735,
-      "loss": 46.0,
-      "step": 4105
-    },
-    {
-      "epoch": 0.6612182455010266,
-      "grad_norm": 0.0010435240110382438,
-      "learning_rate": 0.00019999978529475362,
-      "loss": 46.0,
-      "step": 4106
-    },
-    {
-      "epoch": 0.661379282579814,
-      "grad_norm": 0.0005667787627317011,
-      "learning_rate": 0.0001999997851899043,
-      "loss": 46.0,
-      "step": 4107
-    },
-    {
-      "epoch": 0.6615403196586014,
-      "grad_norm": 0.0005285568186081946,
-      "learning_rate": 0.0001999997850850294,
-      "loss": 46.0,
-      "step": 4108
-    },
-    {
-      "epoch": 0.6617013567373888,
-      "grad_norm": 0.000425788079155609,
-      "learning_rate": 0.0001999997849801289,
-      "loss": 46.0,
-      "step": 4109
-    },
-    {
-      "epoch": 0.6618623938161762,
-      "grad_norm": 0.002144289668649435,
-      "learning_rate": 0.00019999978487520283,
-      "loss": 46.0,
-      "step": 4110
-    },
-    {
-      "epoch": 0.6620234308949635,
-      "grad_norm": 0.0004664475563913584,
-      "learning_rate": 0.00019999978477025114,
-      "loss": 46.0,
-      "step": 4111
-    },
-    {
-      "epoch": 0.662184467973751,
-      "grad_norm": 0.0006001853034831583,
-      "learning_rate": 0.00019999978466527386,
-      "loss": 46.0,
-      "step": 4112
-    },
-    {
-      "epoch": 0.6623455050525383,
-      "grad_norm": 0.0008905772701837122,
-      "learning_rate": 0.00019999978456027096,
-      "loss": 46.0,
-      "step": 4113
-    },
-    {
-      "epoch": 0.6625065421313258,
-      "grad_norm": 0.0006841265712864697,
-      "learning_rate": 0.0001999997844552425,
-      "loss": 46.0,
-      "step": 4114
-    },
-    {
-      "epoch": 0.6626675792101131,
-      "grad_norm": 0.0012838974362239242,
-      "learning_rate": 0.00019999978435018844,
-      "loss": 46.0,
-      "step": 4115
-    },
-    {
-      "epoch": 0.6628286162889006,
-      "grad_norm": 0.005266777705401182,
-      "learning_rate": 0.0001999997842451088,
-      "loss": 46.0,
-      "step": 4116
-    },
-    {
-      "epoch": 0.6629896533676879,
-      "grad_norm": 0.0018772606272250414,
-      "learning_rate": 0.00019999978414000352,
-      "loss": 46.0,
-      "step": 4117
-    },
-    {
-      "epoch": 0.6631506904464753,
-      "grad_norm": 0.0008285978110507131,
-      "learning_rate": 0.00019999978403487266,
-      "loss": 46.0,
-      "step": 4118
-    },
-    {
-      "epoch": 0.6633117275252627,
-      "grad_norm": 0.0004435323644429445,
-      "learning_rate": 0.00019999978392971622,
-      "loss": 46.0,
-      "step": 4119
-    },
-    {
-      "epoch": 0.6634727646040501,
-      "grad_norm": 0.0010455332230776548,
-      "learning_rate": 0.00019999978382453422,
-      "loss": 46.0,
-      "step": 4120
-    },
-    {
-      "epoch": 0.6636338016828375,
-      "grad_norm": 0.0004350183589849621,
-      "learning_rate": 0.00019999978371932657,
-      "loss": 46.0,
-      "step": 4121
-    },
-    {
-      "epoch": 0.6637948387616248,
-      "grad_norm": 0.001673016115091741,
-      "learning_rate": 0.00019999978361409334,
-      "loss": 46.0,
-      "step": 4122
-    },
-    {
-      "epoch": 0.6639558758404123,
-      "grad_norm": 0.001579698990099132,
-      "learning_rate": 0.00019999978350883452,
-      "loss": 46.0,
-      "step": 4123
-    },
-    {
-      "epoch": 0.6641169129191996,
-      "grad_norm": 0.0006469959043897688,
-      "learning_rate": 0.00019999978340355011,
-      "loss": 46.0,
-      "step": 4124
-    },
-    {
-      "epoch": 0.6642779499979871,
-      "grad_norm": 0.0008787863189354539,
-      "learning_rate": 0.0001999997832982401,
-      "loss": 46.0,
-      "step": 4125
-    },
-    {
-      "epoch": 0.6644389870767744,
-      "grad_norm": 0.0007085479446686804,
-      "learning_rate": 0.00019999978319290449,
-      "loss": 46.0,
-      "step": 4126
-    },
-    {
-      "epoch": 0.6646000241555619,
-      "grad_norm": 0.0006078773876652122,
-      "learning_rate": 0.0001999997830875433,
-      "loss": 46.0,
-      "step": 4127
-    },
-    {
-      "epoch": 0.6647610612343492,
-      "grad_norm": 0.0033882097341120243,
-      "learning_rate": 0.00019999978298215648,
-      "loss": 46.0,
-      "step": 4128
-    },
-    {
-      "epoch": 0.6649220983131366,
-      "grad_norm": 0.0006846666801720858,
-      "learning_rate": 0.0001999997828767441,
-      "loss": 46.0,
-      "step": 4129
-    },
-    {
-      "epoch": 0.665083135391924,
-      "grad_norm": 0.00039883560384623706,
-      "learning_rate": 0.0001999997827713061,
-      "loss": 46.0,
-      "step": 4130
-    },
-    {
-      "epoch": 0.6652441724707114,
-      "grad_norm": 0.00034085099468939006,
-      "learning_rate": 0.00019999978266584253,
-      "loss": 46.0,
-      "step": 4131
-    },
-    {
-      "epoch": 0.6654052095494988,
-      "grad_norm": 0.00025625157286413014,
-      "learning_rate": 0.00019999978256035334,
-      "loss": 46.0,
-      "step": 4132
-    },
-    {
-      "epoch": 0.6655662466282861,
-      "grad_norm": 0.00106684525962919,
-      "learning_rate": 0.00019999978245483857,
-      "loss": 46.0,
-      "step": 4133
-    },
-    {
-      "epoch": 0.6657272837070736,
-      "grad_norm": 0.002219436690211296,
-      "learning_rate": 0.00019999978234929819,
-      "loss": 46.0,
-      "step": 4134
-    },
-    {
-      "epoch": 0.6658883207858609,
-      "grad_norm": 0.0026410138234496117,
-      "learning_rate": 0.00019999978224373224,
-      "loss": 46.0,
-      "step": 4135
-    },
-    {
-      "epoch": 0.6660493578646484,
-      "grad_norm": 0.0004199472314212471,
-      "learning_rate": 0.0001999997821381407,
-      "loss": 46.0,
-      "step": 4136
-    },
-    {
-      "epoch": 0.6662103949434357,
-      "grad_norm": 0.000729235471226275,
-      "learning_rate": 0.00019999978203252353,
-      "loss": 46.0,
-      "step": 4137
-    },
-    {
-      "epoch": 0.6663714320222232,
-      "grad_norm": 0.00041112376493401825,
-      "learning_rate": 0.0001999997819268808,
-      "loss": 46.0,
-      "step": 4138
-    },
-    {
-      "epoch": 0.6665324691010105,
-      "grad_norm": 0.0013306516921147704,
-      "learning_rate": 0.00019999978182121245,
-      "loss": 46.0,
-      "step": 4139
-    },
-    {
-      "epoch": 0.666693506179798,
-      "grad_norm": 0.0004246679600328207,
-      "learning_rate": 0.0001999997817155185,
-      "loss": 46.0,
-      "step": 4140
-    },
-    {
-      "epoch": 0.6668545432585853,
-      "grad_norm": 0.0003912248939741403,
-      "learning_rate": 0.00019999978160979899,
-      "loss": 46.0,
-      "step": 4141
-    },
-    {
-      "epoch": 0.6670155803373726,
-      "grad_norm": 0.001465828507207334,
-      "learning_rate": 0.00019999978150405387,
-      "loss": 46.0,
-      "step": 4142
-    },
-    {
-      "epoch": 0.6671766174161601,
-      "grad_norm": 0.002138945274055004,
-      "learning_rate": 0.00019999978139828315,
-      "loss": 46.0,
-      "step": 4143
-    },
-    {
-      "epoch": 0.6673376544949474,
-      "grad_norm": 0.002108029555529356,
-      "learning_rate": 0.00019999978129248684,
-      "loss": 46.0,
-      "step": 4144
-    },
-    {
-      "epoch": 0.6674986915737349,
-      "grad_norm": 0.0013364724582061172,
-      "learning_rate": 0.0001999997811866649,
-      "loss": 46.0,
-      "step": 4145
-    },
-    {
-      "epoch": 0.6676597286525222,
-      "grad_norm": 0.0006084447377361357,
-      "learning_rate": 0.0001999997810808174,
-      "loss": 46.0,
-      "step": 4146
-    },
-    {
-      "epoch": 0.6678207657313097,
-      "grad_norm": 0.0013225043658167124,
-      "learning_rate": 0.00019999978097494432,
-      "loss": 46.0,
-      "step": 4147
-    },
-    {
-      "epoch": 0.667981802810097,
-      "grad_norm": 0.0004937726189382374,
-      "learning_rate": 0.00019999978086904563,
-      "loss": 46.0,
-      "step": 4148
-    },
-    {
-      "epoch": 0.6681428398888845,
-      "grad_norm": 0.0010499687632545829,
-      "learning_rate": 0.00019999978076312133,
-      "loss": 46.0,
-      "step": 4149
-    },
-    {
-      "epoch": 0.6683038769676718,
-      "grad_norm": 0.0005146781331859529,
-      "learning_rate": 0.00019999978065717144,
-      "loss": 46.0,
-      "step": 4150
-    },
-    {
-      "epoch": 0.6684649140464592,
-      "grad_norm": 0.00031648215372115374,
-      "learning_rate": 0.000199999780551196,
-      "loss": 46.0,
-      "step": 4151
-    },
-    {
-      "epoch": 0.6686259511252466,
-      "grad_norm": 0.0007224585278891027,
-      "learning_rate": 0.00019999978044519493,
-      "loss": 46.0,
-      "step": 4152
-    },
-    {
-      "epoch": 0.6687869882040339,
-      "grad_norm": 0.001101408852264285,
-      "learning_rate": 0.00019999978033916825,
-      "loss": 46.0,
-      "step": 4153
-    },
-    {
-      "epoch": 0.6689480252828214,
-      "grad_norm": 0.0010674436343833804,
-      "learning_rate": 0.00019999978023311598,
-      "loss": 46.0,
-      "step": 4154
-    },
-    {
-      "epoch": 0.6691090623616087,
-      "grad_norm": 0.0031106816604733467,
-      "learning_rate": 0.00019999978012703813,
-      "loss": 46.0,
-      "step": 4155
-    },
-    {
-      "epoch": 0.6692700994403962,
-      "grad_norm": 0.0010225242003798485,
-      "learning_rate": 0.00019999978002093466,
-      "loss": 46.0,
-      "step": 4156
-    },
-    {
-      "epoch": 0.6694311365191835,
-      "grad_norm": 0.004867867100983858,
-      "learning_rate": 0.00019999977991480564,
-      "loss": 46.0,
-      "step": 4157
-    },
-    {
-      "epoch": 0.669592173597971,
-      "grad_norm": 0.0024208661634474993,
-      "learning_rate": 0.000199999779808651,
-      "loss": 46.0,
-      "step": 4158
-    },
-    {
-      "epoch": 0.6697532106767583,
-      "grad_norm": 0.0013799263397231698,
-      "learning_rate": 0.00019999977970247074,
-      "loss": 46.0,
-      "step": 4159
-    },
-    {
-      "epoch": 0.6699142477555458,
-      "grad_norm": 0.00043896023998968303,
-      "learning_rate": 0.00019999977959626492,
-      "loss": 46.0,
-      "step": 4160
-    },
-    {
-      "epoch": 0.6700752848343331,
-      "grad_norm": 0.0008262143819592893,
-      "learning_rate": 0.0001999997794900335,
-      "loss": 46.0,
-      "step": 4161
-    },
-    {
-      "epoch": 0.6702363219131205,
-      "grad_norm": 0.0014996931422501802,
-      "learning_rate": 0.00019999977938377645,
-      "loss": 46.0,
-      "step": 4162
-    },
-    {
-      "epoch": 0.6703973589919079,
-      "grad_norm": 0.0005781179643236101,
-      "learning_rate": 0.00019999977927749385,
-      "loss": 46.0,
-      "step": 4163
-    },
-    {
-      "epoch": 0.6705583960706952,
-      "grad_norm": 0.0006504025659523904,
-      "learning_rate": 0.00019999977917118565,
-      "loss": 46.0,
-      "step": 4164
-    },
-    {
-      "epoch": 0.6707194331494827,
-      "grad_norm": 0.0010935229947790504,
-      "learning_rate": 0.00019999977906485182,
-      "loss": 46.0,
-      "step": 4165
-    },
-    {
-      "epoch": 0.67088047022827,
-      "grad_norm": 0.0011264767963439226,
-      "learning_rate": 0.00019999977895849243,
-      "loss": 46.0,
-      "step": 4166
-    },
-    {
-      "epoch": 0.6710415073070575,
-      "grad_norm": 0.0012723105028271675,
-      "learning_rate": 0.00019999977885210742,
-      "loss": 46.0,
-      "step": 4167
-    },
-    {
-      "epoch": 0.6712025443858448,
-      "grad_norm": 0.0024030390195548534,
-      "learning_rate": 0.00019999977874569685,
-      "loss": 46.0,
-      "step": 4168
-    },
-    {
-      "epoch": 0.6713635814646323,
-      "grad_norm": 0.0013126195408403873,
-      "learning_rate": 0.00019999977863926067,
-      "loss": 46.0,
-      "step": 4169
-    },
-    {
-      "epoch": 0.6715246185434196,
-      "grad_norm": 0.0024808617308735847,
-      "learning_rate": 0.0001999997785327989,
-      "loss": 46.0,
-      "step": 4170
-    },
-    {
-      "epoch": 0.6716856556222071,
-      "grad_norm": 0.0005823391838930547,
-      "learning_rate": 0.0001999997784263115,
-      "loss": 46.0,
-      "step": 4171
-    },
-    {
-      "epoch": 0.6718466927009944,
-      "grad_norm": 0.0003966906515415758,
-      "learning_rate": 0.00019999977831979855,
-      "loss": 46.0,
-      "step": 4172
-    },
-    {
-      "epoch": 0.6720077297797818,
-      "grad_norm": 0.0016826341161504388,
-      "learning_rate": 0.00019999977821325997,
-      "loss": 46.0,
-      "step": 4173
-    },
-    {
-      "epoch": 0.6721687668585692,
-      "grad_norm": 0.0005524946027435362,
-      "learning_rate": 0.00019999977810669582,
-      "loss": 46.0,
-      "step": 4174
-    },
-    {
-      "epoch": 0.6723298039373565,
-      "grad_norm": 0.0005389831494539976,
-      "learning_rate": 0.00019999977800010606,
-      "loss": 46.0,
-      "step": 4175
-    },
-    {
-      "epoch": 0.672490841016144,
-      "grad_norm": 0.0015466846525669098,
-      "learning_rate": 0.00019999977789349072,
-      "loss": 46.0,
-      "step": 4176
-    },
-    {
-      "epoch": 0.6726518780949313,
-      "grad_norm": 0.0007032767753116786,
-      "learning_rate": 0.00019999977778684979,
-      "loss": 46.0,
-      "step": 4177
-    },
-    {
-      "epoch": 0.6728129151737188,
-      "grad_norm": 0.0026330510154366493,
-      "learning_rate": 0.00019999977768018324,
-      "loss": 46.0,
-      "step": 4178
-    },
-    {
-      "epoch": 0.6729739522525061,
-      "grad_norm": 0.001228445558808744,
-      "learning_rate": 0.0001999997775734911,
-      "loss": 46.0,
-      "step": 4179
-    },
-    {
-      "epoch": 0.6731349893312936,
-      "grad_norm": 0.0008239775779657066,
-      "learning_rate": 0.00019999977746677335,
-      "loss": 46.0,
-      "step": 4180
-    },
-    {
-      "epoch": 0.6732960264100809,
-      "grad_norm": 0.0013179155066609383,
-      "learning_rate": 0.00019999977736003007,
-      "loss": 46.0,
-      "step": 4181
-    },
-    {
-      "epoch": 0.6734570634888684,
-      "grad_norm": 0.0008079595281742513,
-      "learning_rate": 0.00019999977725326112,
-      "loss": 46.0,
-      "step": 4182
-    },
-    {
-      "epoch": 0.6736181005676557,
-      "grad_norm": 0.0025764591991901398,
-      "learning_rate": 0.00019999977714646664,
-      "loss": 46.0,
-      "step": 4183
-    },
-    {
-      "epoch": 0.673779137646443,
-      "grad_norm": 0.0010873764986172318,
-      "learning_rate": 0.00019999977703964651,
-      "loss": 46.0,
-      "step": 4184
-    },
-    {
-      "epoch": 0.6739401747252305,
-      "grad_norm": 0.0004788876394741237,
-      "learning_rate": 0.0001999997769328008,
-      "loss": 46.0,
-      "step": 4185
-    },
-    {
-      "epoch": 0.6741012118040178,
-      "grad_norm": 0.0034970359411090612,
-      "learning_rate": 0.00019999977682592953,
-      "loss": 46.0,
-      "step": 4186
-    },
-    {
-      "epoch": 0.6742622488828053,
-      "grad_norm": 0.0005447661969810724,
-      "learning_rate": 0.00019999977671903262,
-      "loss": 46.0,
-      "step": 4187
-    },
-    {
-      "epoch": 0.6744232859615926,
-      "grad_norm": 0.0012893053935840726,
-      "learning_rate": 0.00019999977661211014,
-      "loss": 46.0,
-      "step": 4188
-    },
-    {
-      "epoch": 0.6745843230403801,
-      "grad_norm": 0.0009679575450718403,
-      "learning_rate": 0.00019999977650516206,
-      "loss": 46.0,
-      "step": 4189
-    },
-    {
-      "epoch": 0.6747453601191674,
-      "grad_norm": 0.000922391249332577,
-      "learning_rate": 0.00019999977639818838,
-      "loss": 46.0,
-      "step": 4190
-    },
-    {
-      "epoch": 0.6749063971979549,
-      "grad_norm": 0.0014050399186089635,
-      "learning_rate": 0.00019999977629118912,
-      "loss": 46.0,
-      "step": 4191
-    },
-    {
-      "epoch": 0.6750674342767422,
-      "grad_norm": 0.002332976320758462,
-      "learning_rate": 0.00019999977618416424,
-      "loss": 46.0,
-      "step": 4192
-    },
-    {
-      "epoch": 0.6752284713555297,
-      "grad_norm": 0.0005326683167368174,
-      "learning_rate": 0.0001999997760771138,
-      "loss": 46.0,
-      "step": 4193
-    },
-    {
-      "epoch": 0.675389508434317,
-      "grad_norm": 0.001163173234090209,
-      "learning_rate": 0.00019999977597003773,
-      "loss": 46.0,
-      "step": 4194
-    },
-    {
-      "epoch": 0.6755505455131043,
-      "grad_norm": 0.00047027115942910314,
-      "learning_rate": 0.00019999977586293607,
-      "loss": 46.0,
-      "step": 4195
-    },
-    {
-      "epoch": 0.6757115825918918,
-      "grad_norm": 0.0009730842430144548,
-      "learning_rate": 0.00019999977575580884,
-      "loss": 46.0,
-      "step": 4196
-    },
-    {
-      "epoch": 0.6758726196706791,
-      "grad_norm": 0.0007298915879800916,
-      "learning_rate": 0.00019999977564865603,
-      "loss": 46.0,
-      "step": 4197
-    },
-    {
-      "epoch": 0.6760336567494666,
-      "grad_norm": 0.0015910177025943995,
-      "learning_rate": 0.00019999977554147758,
-      "loss": 46.0,
-      "step": 4198
-    },
-    {
-      "epoch": 0.6761946938282539,
-      "grad_norm": 0.00048593347310088575,
-      "learning_rate": 0.00019999977543427356,
-      "loss": 46.0,
-      "step": 4199
-    },
-    {
-      "epoch": 0.6763557309070414,
-      "grad_norm": 0.0005243184277787805,
-      "learning_rate": 0.00019999977532704393,
-      "loss": 46.0,
-      "step": 4200
-    },
-    {
-      "epoch": 0.6765167679858287,
-      "grad_norm": 0.0009467373602092266,
-      "learning_rate": 0.0001999997752197887,
-      "loss": 46.0,
-      "step": 4201
-    },
-    {
-      "epoch": 0.6766778050646162,
-      "grad_norm": 0.001184506225399673,
-      "learning_rate": 0.0001999997751125079,
-      "loss": 46.0,
-      "step": 4202
-    },
-    {
-      "epoch": 0.6768388421434035,
-      "grad_norm": 0.0005141497240401804,
-      "learning_rate": 0.00019999977500520147,
-      "loss": 46.0,
-      "step": 4203
-    },
-    {
-      "epoch": 0.676999879222191,
-      "grad_norm": 0.0008613636018708348,
-      "learning_rate": 0.00019999977489786947,
-      "loss": 46.0,
-      "step": 4204
-    },
-    {
-      "epoch": 0.6771609163009783,
-      "grad_norm": 0.0016730212373659015,
-      "learning_rate": 0.00019999977479051188,
-      "loss": 46.0,
-      "step": 4205
-    },
-    {
-      "epoch": 0.6773219533797656,
-      "grad_norm": 0.0006290310993790627,
-      "learning_rate": 0.0001999997746831287,
-      "loss": 46.0,
-      "step": 4206
-    },
-    {
-      "epoch": 0.6774829904585531,
-      "grad_norm": 0.0006873703096061945,
-      "learning_rate": 0.0001999997745757199,
-      "loss": 46.0,
-      "step": 4207
-    },
-    {
-      "epoch": 0.6776440275373404,
-      "grad_norm": 0.0006282299873419106,
-      "learning_rate": 0.0001999997744682855,
-      "loss": 46.0,
-      "step": 4208
-    },
-    {
-      "epoch": 0.6778050646161279,
-      "grad_norm": 0.0007294894312508404,
-      "learning_rate": 0.00019999977436082556,
-      "loss": 46.0,
-      "step": 4209
-    },
-    {
-      "epoch": 0.6779661016949152,
-      "grad_norm": 0.00033628210076130927,
-      "learning_rate": 0.00019999977425333998,
-      "loss": 46.0,
-      "step": 4210
-    },
-    {
-      "epoch": 0.6781271387737027,
-      "grad_norm": 0.0007842194754630327,
-      "learning_rate": 0.0001999997741458288,
-      "loss": 46.0,
-      "step": 4211
-    },
-    {
-      "epoch": 0.67828817585249,
-      "grad_norm": 0.00039297586772590876,
-      "learning_rate": 0.00019999977403829203,
-      "loss": 46.0,
-      "step": 4212
-    },
-    {
-      "epoch": 0.6784492129312775,
-      "grad_norm": 0.0003269016451667994,
-      "learning_rate": 0.0001999997739307297,
-      "loss": 46.0,
-      "step": 4213
-    },
-    {
-      "epoch": 0.6786102500100648,
-      "grad_norm": 0.0021821712143719196,
-      "learning_rate": 0.00019999977382314173,
-      "loss": 46.0,
-      "step": 4214
-    },
-    {
-      "epoch": 0.6787712870888523,
-      "grad_norm": 0.0005154683603905141,
-      "learning_rate": 0.0001999997737155282,
-      "loss": 46.0,
-      "step": 4215
-    },
-    {
-      "epoch": 0.6789323241676396,
-      "grad_norm": 0.0015087351202964783,
-      "learning_rate": 0.00019999977360788903,
-      "loss": 46.0,
-      "step": 4216
-    },
-    {
-      "epoch": 0.6790933612464269,
-      "grad_norm": 0.0030605087522417307,
-      "learning_rate": 0.00019999977350022431,
-      "loss": 46.0,
-      "step": 4217
-    },
-    {
-      "epoch": 0.6792543983252144,
-      "grad_norm": 0.0005566399777308106,
-      "learning_rate": 0.00019999977339253398,
-      "loss": 46.0,
-      "step": 4218
-    },
-    {
-      "epoch": 0.6794154354040017,
-      "grad_norm": 0.0006102929473854601,
-      "learning_rate": 0.00019999977328481804,
-      "loss": 46.0,
-      "step": 4219
-    },
-    {
-      "epoch": 0.6795764724827892,
-      "grad_norm": 0.001174689969047904,
-      "learning_rate": 0.00019999977317707656,
-      "loss": 46.0,
-      "step": 4220
-    },
-    {
-      "epoch": 0.6797375095615765,
-      "grad_norm": 0.001205551903694868,
-      "learning_rate": 0.0001999997730693094,
-      "loss": 46.0,
-      "step": 4221
-    },
-    {
-      "epoch": 0.679898546640364,
-      "grad_norm": 0.0017714441055431962,
-      "learning_rate": 0.0001999997729615167,
-      "loss": 46.0,
-      "step": 4222
-    },
-    {
-      "epoch": 0.6800595837191513,
-      "grad_norm": 0.004116755444556475,
-      "learning_rate": 0.00019999977285369843,
-      "loss": 46.0,
-      "step": 4223
-    },
-    {
-      "epoch": 0.6802206207979388,
-      "grad_norm": 0.00029588156030513346,
-      "learning_rate": 0.0001999997727458545,
-      "loss": 46.0,
-      "step": 4224
-    },
-    {
-      "epoch": 0.6803816578767261,
-      "grad_norm": 0.0010703310836106539,
-      "learning_rate": 0.00019999977263798503,
-      "loss": 46.0,
-      "step": 4225
-    },
-    {
-      "epoch": 0.6805426949555136,
-      "grad_norm": 0.0008275836589746177,
-      "learning_rate": 0.00019999977253008991,
-      "loss": 46.0,
-      "step": 4226
-    },
-    {
-      "epoch": 0.6807037320343009,
-      "grad_norm": 0.0009265264379791915,
-      "learning_rate": 0.00019999977242216924,
-      "loss": 46.0,
-      "step": 4227
-    },
-    {
-      "epoch": 0.6808647691130882,
-      "grad_norm": 0.00046720009413547814,
-      "learning_rate": 0.00019999977231422298,
-      "loss": 46.0,
-      "step": 4228
-    },
-    {
-      "epoch": 0.6810258061918757,
-      "grad_norm": 0.00045534424134530127,
-      "learning_rate": 0.00019999977220625109,
-      "loss": 46.0,
-      "step": 4229
-    },
-    {
-      "epoch": 0.681186843270663,
-      "grad_norm": 0.00221816822886467,
-      "learning_rate": 0.00019999977209825363,
-      "loss": 46.0,
-      "step": 4230
-    },
-    {
-      "epoch": 0.6813478803494505,
-      "grad_norm": 0.0007783073815517128,
-      "learning_rate": 0.00019999977199023058,
-      "loss": 46.0,
-      "step": 4231
-    },
-    {
-      "epoch": 0.6815089174282378,
-      "grad_norm": 0.0011567584006115794,
-      "learning_rate": 0.00019999977188218192,
-      "loss": 46.0,
-      "step": 4232
-    },
-    {
-      "epoch": 0.6816699545070253,
-      "grad_norm": 0.005762670189142227,
-      "learning_rate": 0.00019999977177410764,
-      "loss": 46.0,
-      "step": 4233
-    },
-    {
-      "epoch": 0.6818309915858126,
-      "grad_norm": 0.0010815223213285208,
-      "learning_rate": 0.0001999997716660078,
-      "loss": 46.0,
-      "step": 4234
-    },
-    {
-      "epoch": 0.6819920286646001,
-      "grad_norm": 0.0031720888800919056,
-      "learning_rate": 0.00019999977155788236,
-      "loss": 46.0,
-      "step": 4235
-    },
-    {
-      "epoch": 0.6821530657433874,
-      "grad_norm": 0.0006679006037302315,
-      "learning_rate": 0.00019999977144973135,
-      "loss": 46.0,
-      "step": 4236
-    },
-    {
-      "epoch": 0.6823141028221748,
-      "grad_norm": 0.0002713054418563843,
-      "learning_rate": 0.0001999997713415547,
-      "loss": 46.0,
-      "step": 4237
-    },
-    {
-      "epoch": 0.6824751399009622,
-      "grad_norm": 0.0004060633364133537,
-      "learning_rate": 0.00019999977123335248,
-      "loss": 46.0,
-      "step": 4238
-    },
-    {
-      "epoch": 0.6826361769797495,
-      "grad_norm": 0.0012862241128459573,
-      "learning_rate": 0.00019999977112512466,
-      "loss": 46.0,
-      "step": 4239
-    },
-    {
-      "epoch": 0.682797214058537,
-      "grad_norm": 0.0013272602809593081,
-      "learning_rate": 0.00019999977101687125,
-      "loss": 46.0,
-      "step": 4240
-    },
-    {
-      "epoch": 0.6829582511373243,
-      "grad_norm": 0.0010185547871515155,
-      "learning_rate": 0.00019999977090859222,
-      "loss": 46.0,
-      "step": 4241
-    },
-    {
-      "epoch": 0.6831192882161118,
-      "grad_norm": 0.0009247910347767174,
-      "learning_rate": 0.0001999997708002876,
-      "loss": 46.0,
-      "step": 4242
-    },
-    {
-      "epoch": 0.6832803252948991,
-      "grad_norm": 0.0007376632420346141,
-      "learning_rate": 0.0001999997706919574,
-      "loss": 46.0,
-      "step": 4243
-    },
-    {
-      "epoch": 0.6834413623736866,
-      "grad_norm": 0.0004836310399696231,
-      "learning_rate": 0.00019999977058360161,
-      "loss": 46.0,
-      "step": 4244
-    },
-    {
-      "epoch": 0.6836023994524739,
-      "grad_norm": 0.000720982498023659,
-      "learning_rate": 0.00019999977047522024,
-      "loss": 46.0,
-      "step": 4245
-    },
-    {
-      "epoch": 0.6837634365312614,
-      "grad_norm": 0.0019374929834157228,
-      "learning_rate": 0.00019999977036681322,
-      "loss": 46.0,
-      "step": 4246
-    },
-    {
-      "epoch": 0.6839244736100487,
-      "grad_norm": 0.0006270456360653043,
-      "learning_rate": 0.00019999977025838067,
-      "loss": 46.0,
-      "step": 4247
-    },
-    {
-      "epoch": 0.684085510688836,
-      "grad_norm": 0.0009443956078030169,
-      "learning_rate": 0.0001999997701499225,
-      "loss": 46.0,
-      "step": 4248
-    },
-    {
-      "epoch": 0.6842465477676235,
-      "grad_norm": 0.0007357310969382524,
-      "learning_rate": 0.00019999977004143873,
-      "loss": 46.0,
-      "step": 4249
-    },
-    {
-      "epoch": 0.6844075848464108,
-      "grad_norm": 0.0008445387938991189,
-      "learning_rate": 0.00019999976993292936,
-      "loss": 46.0,
-      "step": 4250
-    },
-    {
-      "epoch": 0.6845686219251983,
-      "grad_norm": 0.0009629289270378649,
-      "learning_rate": 0.0001999997698243944,
-      "loss": 46.0,
-      "step": 4251
-    },
-    {
-      "epoch": 0.6847296590039856,
-      "grad_norm": 0.0006416314863599837,
-      "learning_rate": 0.00019999976971583382,
-      "loss": 46.0,
-      "step": 4252
-    },
-    {
-      "epoch": 0.6848906960827731,
-      "grad_norm": 0.001572751789353788,
-      "learning_rate": 0.00019999976960724772,
-      "loss": 46.0,
-      "step": 4253
-    },
-    {
-      "epoch": 0.6850517331615604,
-      "grad_norm": 0.000453545042546466,
-      "learning_rate": 0.00019999976949863597,
-      "loss": 46.0,
-      "step": 4254
-    },
-    {
-      "epoch": 0.6852127702403479,
-      "grad_norm": 0.00030573667027056217,
-      "learning_rate": 0.0001999997693899986,
-      "loss": 46.0,
-      "step": 4255
-    },
-    {
-      "epoch": 0.6853738073191352,
-      "grad_norm": 0.0007804098422639072,
-      "learning_rate": 0.00019999976928133568,
-      "loss": 46.0,
-      "step": 4256
-    },
-    {
-      "epoch": 0.6855348443979227,
-      "grad_norm": 0.0004297084524296224,
-      "learning_rate": 0.00019999976917264715,
-      "loss": 46.0,
-      "step": 4257
-    },
-    {
-      "epoch": 0.68569588147671,
-      "grad_norm": 0.0005432379548437893,
-      "learning_rate": 0.00019999976906393303,
-      "loss": 46.0,
-      "step": 4258
-    },
-    {
-      "epoch": 0.6858569185554974,
-      "grad_norm": 0.0017453789478167892,
-      "learning_rate": 0.0001999997689551933,
-      "loss": 46.0,
-      "step": 4259
-    },
-    {
-      "epoch": 0.6860179556342848,
-      "grad_norm": 0.0005355547182261944,
-      "learning_rate": 0.000199999768846428,
-      "loss": 46.0,
-      "step": 4260
-    },
-    {
-      "epoch": 0.6861789927130721,
-      "grad_norm": 0.0028631852474063635,
-      "learning_rate": 0.0001999997687376371,
-      "loss": 46.0,
-      "step": 4261
-    },
-    {
-      "epoch": 0.6863400297918596,
-      "grad_norm": 0.0020064173731952906,
-      "learning_rate": 0.0001999997686288206,
-      "loss": 46.0,
-      "step": 4262
-    },
-    {
-      "epoch": 0.6865010668706469,
-      "grad_norm": 0.0022759444545954466,
-      "learning_rate": 0.0001999997685199785,
-      "loss": 46.0,
-      "step": 4263
-    },
-    {
-      "epoch": 0.6866621039494344,
-      "grad_norm": 0.0007143968250602484,
-      "learning_rate": 0.0001999997684111108,
-      "loss": 46.0,
-      "step": 4264
-    },
-    {
-      "epoch": 0.6868231410282217,
-      "grad_norm": 0.000550659082364291,
-      "learning_rate": 0.00019999976830221752,
-      "loss": 46.0,
-      "step": 4265
-    },
-    {
-      "epoch": 0.6869841781070092,
-      "grad_norm": 0.00155602116137743,
-      "learning_rate": 0.00019999976819329865,
-      "loss": 46.0,
-      "step": 4266
-    },
-    {
-      "epoch": 0.6871452151857965,
-      "grad_norm": 0.001805823645554483,
-      "learning_rate": 0.00019999976808435417,
-      "loss": 46.0,
-      "step": 4267
-    },
-    {
-      "epoch": 0.687306252264584,
-      "grad_norm": 0.0010955483885481954,
-      "learning_rate": 0.0001999997679753841,
-      "loss": 46.0,
-      "step": 4268
-    },
-    {
-      "epoch": 0.6874672893433713,
-      "grad_norm": 0.002118520438671112,
-      "learning_rate": 0.00019999976786638844,
-      "loss": 46.0,
-      "step": 4269
-    },
-    {
-      "epoch": 0.6876283264221587,
-      "grad_norm": 0.001822443213313818,
-      "learning_rate": 0.00019999976775736714,
-      "loss": 46.0,
-      "step": 4270
-    },
-    {
-      "epoch": 0.6877893635009461,
-      "grad_norm": 0.0005718372412957251,
-      "learning_rate": 0.0001999997676483203,
-      "loss": 46.0,
-      "step": 4271
-    },
-    {
-      "epoch": 0.6879504005797334,
-      "grad_norm": 0.0029179533012211323,
-      "learning_rate": 0.00019999976753924784,
-      "loss": 46.0,
-      "step": 4272
-    },
-    {
-      "epoch": 0.6881114376585209,
-      "grad_norm": 0.0024086886551231146,
-      "learning_rate": 0.00019999976743014983,
-      "loss": 46.0,
-      "step": 4273
-    },
-    {
-      "epoch": 0.6882724747373082,
-      "grad_norm": 0.0011706891236826777,
-      "learning_rate": 0.00019999976732102618,
-      "loss": 46.0,
-      "step": 4274
-    },
-    {
-      "epoch": 0.6884335118160957,
-      "grad_norm": 0.000658955832477659,
-      "learning_rate": 0.00019999976721187695,
-      "loss": 46.0,
-      "step": 4275
-    },
-    {
-      "epoch": 0.688594548894883,
-      "grad_norm": 0.0004229762125760317,
-      "learning_rate": 0.00019999976710270213,
-      "loss": 46.0,
-      "step": 4276
-    },
-    {
-      "epoch": 0.6887555859736705,
-      "grad_norm": 0.00032889386056922376,
-      "learning_rate": 0.0001999997669935017,
-      "loss": 46.0,
-      "step": 4277
-    },
-    {
-      "epoch": 0.6889166230524578,
-      "grad_norm": 0.0006168846739456058,
-      "learning_rate": 0.0001999997668842757,
-      "loss": 46.0,
-      "step": 4278
-    },
-    {
-      "epoch": 0.6890776601312453,
-      "grad_norm": 0.001164330285973847,
-      "learning_rate": 0.00019999976677502406,
-      "loss": 46.0,
-      "step": 4279
-    },
-    {
-      "epoch": 0.6892386972100326,
-      "grad_norm": 0.0009718163637444377,
-      "learning_rate": 0.00019999976666574686,
-      "loss": 46.0,
-      "step": 4280
-    },
-    {
-      "epoch": 0.68939973428882,
-      "grad_norm": 0.0008400626247748733,
-      "learning_rate": 0.00019999976655644405,
-      "loss": 46.0,
-      "step": 4281
-    },
-    {
-      "epoch": 0.6895607713676074,
-      "grad_norm": 0.00071126245893538,
-      "learning_rate": 0.00019999976644711565,
-      "loss": 46.0,
-      "step": 4282
-    },
-    {
-      "epoch": 0.6897218084463947,
-      "grad_norm": 0.00064807542366907,
-      "learning_rate": 0.00019999976633776166,
-      "loss": 46.0,
-      "step": 4283
-    },
-    {
-      "epoch": 0.6898828455251822,
-      "grad_norm": 0.0005875331116840243,
-      "learning_rate": 0.0001999997662283821,
-      "loss": 46.0,
-      "step": 4284
-    },
-    {
-      "epoch": 0.6900438826039695,
-      "grad_norm": 0.0006677934434264898,
-      "learning_rate": 0.00019999976611897687,
-      "loss": 46.0,
-      "step": 4285
-    },
-    {
-      "epoch": 0.690204919682757,
-      "grad_norm": 0.0015023647574707866,
-      "learning_rate": 0.0001999997660095461,
-      "loss": 46.0,
-      "step": 4286
-    },
-    {
-      "epoch": 0.6903659567615443,
-      "grad_norm": 0.0013992848107591271,
-      "learning_rate": 0.00019999976590008974,
-      "loss": 46.0,
-      "step": 4287
-    },
-    {
-      "epoch": 0.6905269938403318,
-      "grad_norm": 0.0016881533665582538,
-      "learning_rate": 0.00019999976579060779,
-      "loss": 46.0,
-      "step": 4288
-    },
-    {
-      "epoch": 0.6906880309191191,
-      "grad_norm": 0.0013599763624370098,
-      "learning_rate": 0.00019999976568110022,
-      "loss": 46.0,
-      "step": 4289
-    },
-    {
-      "epoch": 0.6908490679979065,
-      "grad_norm": 0.0005329299601726234,
-      "learning_rate": 0.00019999976557156704,
-      "loss": 46.0,
-      "step": 4290
-    },
-    {
-      "epoch": 0.6910101050766939,
-      "grad_norm": 0.0009675774490460753,
-      "learning_rate": 0.00019999976546200828,
-      "loss": 46.0,
-      "step": 4291
-    },
-    {
-      "epoch": 0.6911711421554813,
-      "grad_norm": 0.000509542238432914,
-      "learning_rate": 0.00019999976535242395,
-      "loss": 46.0,
-      "step": 4292
-    },
-    {
-      "epoch": 0.6913321792342687,
-      "grad_norm": 0.0023436935152858496,
-      "learning_rate": 0.00019999976524281402,
-      "loss": 46.0,
-      "step": 4293
-    },
-    {
-      "epoch": 0.691493216313056,
-      "grad_norm": 0.0007163815316744149,
-      "learning_rate": 0.00019999976513317846,
-      "loss": 46.0,
-      "step": 4294
-    },
-    {
-      "epoch": 0.6916542533918435,
-      "grad_norm": 0.002450268715620041,
-      "learning_rate": 0.00019999976502351735,
-      "loss": 46.0,
-      "step": 4295
-    },
-    {
-      "epoch": 0.6918152904706308,
-      "grad_norm": 0.0003730957396328449,
-      "learning_rate": 0.00019999976491383065,
-      "loss": 46.0,
-      "step": 4296
-    },
-    {
-      "epoch": 0.6919763275494183,
-      "grad_norm": 0.003522138111293316,
-      "learning_rate": 0.0001999997648041183,
-      "loss": 46.0,
-      "step": 4297
-    },
-    {
-      "epoch": 0.6921373646282056,
-      "grad_norm": 0.00035113058402203023,
-      "learning_rate": 0.0001999997646943804,
-      "loss": 46.0,
-      "step": 4298
-    },
-    {
-      "epoch": 0.6922984017069931,
-      "grad_norm": 0.0015195432351902127,
-      "learning_rate": 0.00019999976458461686,
-      "loss": 46.0,
-      "step": 4299
-    },
-    {
-      "epoch": 0.6924594387857804,
-      "grad_norm": 0.0033547542989253998,
-      "learning_rate": 0.00019999976447482775,
-      "loss": 46.0,
-      "step": 4300
-    },
-    {
-      "epoch": 0.6926204758645678,
-      "grad_norm": 0.0008956454694271088,
-      "learning_rate": 0.00019999976436501306,
-      "loss": 46.0,
-      "step": 4301
-    },
-    {
-      "epoch": 0.6927815129433552,
-      "grad_norm": 0.0013032873393967748,
-      "learning_rate": 0.00019999976425517279,
-      "loss": 46.0,
-      "step": 4302
-    },
-    {
-      "epoch": 0.6929425500221426,
-      "grad_norm": 0.0015084196347743273,
-      "learning_rate": 0.0001999997641453069,
-      "loss": 46.0,
-      "step": 4303
-    },
-    {
-      "epoch": 0.69310358710093,
-      "grad_norm": 0.0011208360083401203,
-      "learning_rate": 0.00019999976403541541,
-      "loss": 46.0,
-      "step": 4304
-    },
-    {
-      "epoch": 0.6932646241797173,
-      "grad_norm": 0.001271988032385707,
-      "learning_rate": 0.00019999976392549832,
-      "loss": 46.0,
-      "step": 4305
-    },
-    {
-      "epoch": 0.6934256612585048,
-      "grad_norm": 0.0007630119216628373,
-      "learning_rate": 0.00019999976381555564,
-      "loss": 46.0,
-      "step": 4306
-    },
-    {
-      "epoch": 0.6935866983372921,
-      "grad_norm": 0.0014788095140829682,
-      "learning_rate": 0.0001999997637055874,
-      "loss": 46.0,
-      "step": 4307
-    },
-    {
-      "epoch": 0.6937477354160796,
-      "grad_norm": 0.0014191108057275414,
-      "learning_rate": 0.00019999976359559352,
-      "loss": 46.0,
-      "step": 4308
-    },
-    {
-      "epoch": 0.6939087724948669,
-      "grad_norm": 0.002827615011483431,
-      "learning_rate": 0.00019999976348557405,
-      "loss": 46.0,
-      "step": 4309
-    },
-    {
-      "epoch": 0.6940698095736544,
-      "grad_norm": 0.0012578010791912675,
-      "learning_rate": 0.00019999976337552902,
-      "loss": 46.0,
-      "step": 4310
-    },
-    {
-      "epoch": 0.6942308466524417,
-      "grad_norm": 0.0002686958760023117,
-      "learning_rate": 0.00019999976326545834,
-      "loss": 46.0,
-      "step": 4311
-    },
-    {
-      "epoch": 0.6943918837312291,
-      "grad_norm": 0.0004917251062579453,
-      "learning_rate": 0.0001999997631553621,
-      "loss": 46.0,
-      "step": 4312
-    },
-    {
-      "epoch": 0.6945529208100165,
-      "grad_norm": 0.0008446917636319995,
-      "learning_rate": 0.0001999997630452403,
-      "loss": 46.0,
-      "step": 4313
-    },
-    {
-      "epoch": 0.6947139578888039,
-      "grad_norm": 0.0010952099692076445,
-      "learning_rate": 0.00019999976293509284,
-      "loss": 46.0,
-      "step": 4314
-    },
-    {
-      "epoch": 0.6948749949675913,
-      "grad_norm": 0.0010970853036269546,
-      "learning_rate": 0.00019999976282491982,
-      "loss": 46.0,
-      "step": 4315
-    },
-    {
-      "epoch": 0.6950360320463786,
-      "grad_norm": 0.0005779225029982626,
-      "learning_rate": 0.0001999997627147212,
-      "loss": 46.0,
-      "step": 4316
-    },
-    {
-      "epoch": 0.6951970691251661,
-      "grad_norm": 0.0006116090225987136,
-      "learning_rate": 0.000199999762604497,
-      "loss": 46.0,
-      "step": 4317
-    },
-    {
-      "epoch": 0.6953581062039534,
-      "grad_norm": 0.0010594524210318923,
-      "learning_rate": 0.00019999976249424718,
-      "loss": 46.0,
-      "step": 4318
-    },
-    {
-      "epoch": 0.6955191432827409,
-      "grad_norm": 0.002785651246085763,
-      "learning_rate": 0.00019999976238397178,
-      "loss": 46.0,
-      "step": 4319
-    },
-    {
-      "epoch": 0.6956801803615282,
-      "grad_norm": 0.0014585310127586126,
-      "learning_rate": 0.00019999976227367077,
-      "loss": 46.0,
-      "step": 4320
-    },
-    {
-      "epoch": 0.6958412174403157,
-      "grad_norm": 0.0005903081619180739,
-      "learning_rate": 0.00019999976216334418,
-      "loss": 46.0,
-      "step": 4321
-    },
-    {
-      "epoch": 0.696002254519103,
-      "grad_norm": 0.0004984555998817086,
-      "learning_rate": 0.000199999762052992,
-      "loss": 46.0,
-      "step": 4322
-    },
-    {
-      "epoch": 0.6961632915978904,
-      "grad_norm": 0.0017128164181485772,
-      "learning_rate": 0.0001999997619426142,
-      "loss": 46.0,
-      "step": 4323
-    },
-    {
-      "epoch": 0.6963243286766778,
-      "grad_norm": 0.002269329270347953,
-      "learning_rate": 0.00019999976183221084,
-      "loss": 46.0,
-      "step": 4324
-    },
-    {
-      "epoch": 0.6964853657554652,
-      "grad_norm": 0.0019107286352664232,
-      "learning_rate": 0.00019999976172178184,
-      "loss": 46.0,
-      "step": 4325
-    },
-    {
-      "epoch": 0.6966464028342526,
-      "grad_norm": 0.002972388407215476,
-      "learning_rate": 0.00019999976161132728,
-      "loss": 46.0,
-      "step": 4326
-    },
-    {
-      "epoch": 0.69680743991304,
-      "grad_norm": 0.0006845807074569166,
-      "learning_rate": 0.0001999997615008471,
-      "loss": 46.0,
-      "step": 4327
-    },
-    {
-      "epoch": 0.6969684769918274,
-      "grad_norm": 0.0017358340555801988,
-      "learning_rate": 0.00019999976139034137,
-      "loss": 46.0,
-      "step": 4328
-    },
-    {
-      "epoch": 0.6971295140706147,
-      "grad_norm": 0.0009827794274315238,
-      "learning_rate": 0.00019999976127981,
-      "loss": 46.0,
-      "step": 4329
-    },
-    {
-      "epoch": 0.6972905511494022,
-      "grad_norm": 0.001294991234317422,
-      "learning_rate": 0.00019999976116925306,
-      "loss": 46.0,
-      "step": 4330
-    },
-    {
-      "epoch": 0.6974515882281895,
-      "grad_norm": 0.0012318383669480681,
-      "learning_rate": 0.00019999976105867048,
-      "loss": 46.0,
-      "step": 4331
-    },
-    {
-      "epoch": 0.697612625306977,
-      "grad_norm": 0.00047702481970191,
-      "learning_rate": 0.00019999976094806237,
-      "loss": 46.0,
-      "step": 4332
-    },
-    {
-      "epoch": 0.6977736623857643,
-      "grad_norm": 0.0006607566610909998,
-      "learning_rate": 0.00019999976083742862,
-      "loss": 46.0,
-      "step": 4333
-    },
-    {
-      "epoch": 0.6979346994645517,
-      "grad_norm": 0.00367356208153069,
-      "learning_rate": 0.0001999997607267693,
-      "loss": 46.0,
-      "step": 4334
-    },
-    {
-      "epoch": 0.6980957365433391,
-      "grad_norm": 0.0009460444562137127,
-      "learning_rate": 0.00019999976061608438,
-      "loss": 46.0,
-      "step": 4335
-    },
-    {
-      "epoch": 0.6982567736221265,
-      "grad_norm": 0.002952640177682042,
-      "learning_rate": 0.00019999976050537387,
-      "loss": 46.0,
-      "step": 4336
-    },
-    {
-      "epoch": 0.6984178107009139,
-      "grad_norm": 0.0006764120189473033,
-      "learning_rate": 0.00019999976039463774,
-      "loss": 46.0,
-      "step": 4337
-    },
-    {
-      "epoch": 0.6985788477797013,
-      "grad_norm": 0.0006747061852365732,
-      "learning_rate": 0.00019999976028387602,
-      "loss": 46.0,
-      "step": 4338
-    },
-    {
-      "epoch": 0.6987398848584887,
-      "grad_norm": 0.0006493888795375824,
-      "learning_rate": 0.00019999976017308875,
-      "loss": 46.0,
-      "step": 4339
-    },
-    {
-      "epoch": 0.698900921937276,
-      "grad_norm": 0.00040915925637818873,
-      "learning_rate": 0.00019999976006227583,
-      "loss": 46.0,
-      "step": 4340
-    },
-    {
-      "epoch": 0.6990619590160635,
-      "grad_norm": 0.00402578292414546,
-      "learning_rate": 0.00019999975995143733,
-      "loss": 46.0,
-      "step": 4341
-    },
-    {
-      "epoch": 0.6992229960948508,
-      "grad_norm": 0.0024500973522663116,
-      "learning_rate": 0.00019999975984057327,
-      "loss": 46.0,
-      "step": 4342
-    },
-    {
-      "epoch": 0.6993840331736382,
-      "grad_norm": 0.0026648384518921375,
-      "learning_rate": 0.00019999975972968356,
-      "loss": 46.0,
-      "step": 4343
-    },
-    {
-      "epoch": 0.6995450702524256,
-      "grad_norm": 0.002576041966676712,
-      "learning_rate": 0.00019999975961876833,
-      "loss": 46.0,
-      "step": 4344
-    },
-    {
-      "epoch": 0.699706107331213,
-      "grad_norm": 0.0008924486464820802,
-      "learning_rate": 0.00019999975950782745,
-      "loss": 46.0,
-      "step": 4345
-    },
-    {
-      "epoch": 0.6998671444100004,
-      "grad_norm": 0.0014739540638402104,
-      "learning_rate": 0.00019999975939686098,
-      "loss": 46.0,
-      "step": 4346
-    },
-    {
-      "epoch": 0.7000281814887878,
-      "grad_norm": 0.00282071391120553,
-      "learning_rate": 0.00019999975928586893,
-      "loss": 46.0,
-      "step": 4347
-    },
-    {
-      "epoch": 0.7001892185675752,
-      "grad_norm": 0.0007176279905252159,
-      "learning_rate": 0.00019999975917485126,
-      "loss": 46.0,
-      "step": 4348
-    },
-    {
-      "epoch": 0.7003502556463626,
-      "grad_norm": 0.00041312022949568927,
-      "learning_rate": 0.00019999975906380798,
-      "loss": 46.0,
-      "step": 4349
-    },
-    {
-      "epoch": 0.70051129272515,
-      "grad_norm": 0.0025801435112953186,
-      "learning_rate": 0.00019999975895273914,
-      "loss": 46.0,
-      "step": 4350
-    },
-    {
-      "epoch": 0.7006723298039373,
-      "grad_norm": 0.001084522227756679,
-      "learning_rate": 0.00019999975884164473,
-      "loss": 46.0,
-      "step": 4351
-    },
-    {
-      "epoch": 0.7008333668827248,
-      "grad_norm": 0.004320534411817789,
-      "learning_rate": 0.0001999997587305247,
-      "loss": 46.0,
-      "step": 4352
-    },
-    {
-      "epoch": 0.7009944039615121,
-      "grad_norm": 0.00176756433211267,
-      "learning_rate": 0.00019999975861937906,
-      "loss": 46.0,
-      "step": 4353
-    },
-    {
-      "epoch": 0.7011554410402995,
-      "grad_norm": 0.0005683852941729128,
-      "learning_rate": 0.00019999975850820784,
-      "loss": 46.0,
-      "step": 4354
-    },
-    {
-      "epoch": 0.7013164781190869,
-      "grad_norm": 0.0003886757476720959,
-      "learning_rate": 0.000199999758397011,
-      "loss": 46.0,
-      "step": 4355
-    },
-    {
-      "epoch": 0.7014775151978743,
-      "grad_norm": 0.0002913382777478546,
-      "learning_rate": 0.00019999975828578862,
-      "loss": 46.0,
-      "step": 4356
-    },
-    {
-      "epoch": 0.7016385522766617,
-      "grad_norm": 0.0010053608566522598,
-      "learning_rate": 0.00019999975817454059,
-      "loss": 46.0,
-      "step": 4357
-    },
-    {
-      "epoch": 0.7017995893554491,
-      "grad_norm": 0.0011389650171622634,
-      "learning_rate": 0.000199999758063267,
-      "loss": 46.0,
-      "step": 4358
-    },
-    {
-      "epoch": 0.7019606264342365,
-      "grad_norm": 0.0009005354950204492,
-      "learning_rate": 0.00019999975795196778,
-      "loss": 46.0,
-      "step": 4359
-    },
-    {
-      "epoch": 0.7021216635130239,
-      "grad_norm": 0.0006556838634423912,
-      "learning_rate": 0.000199999757840643,
-      "loss": 46.0,
-      "step": 4360
-    },
-    {
-      "epoch": 0.7022827005918113,
-      "grad_norm": 0.0007158315274864435,
-      "learning_rate": 0.0001999997577292926,
-      "loss": 46.0,
-      "step": 4361
-    },
-    {
-      "epoch": 0.7024437376705986,
-      "grad_norm": 0.0007340789888985455,
-      "learning_rate": 0.00019999975761791664,
-      "loss": 46.0,
-      "step": 4362
-    },
-    {
-      "epoch": 0.7026047747493861,
-      "grad_norm": 0.0005704319919459522,
-      "learning_rate": 0.00019999975750651505,
-      "loss": 46.0,
-      "step": 4363
-    },
-    {
-      "epoch": 0.7027658118281734,
-      "grad_norm": 0.001266828621737659,
-      "learning_rate": 0.00019999975739508788,
-      "loss": 46.0,
-      "step": 4364
-    },
-    {
-      "epoch": 0.7029268489069608,
-      "grad_norm": 0.0019355610711500049,
-      "learning_rate": 0.0001999997572836351,
-      "loss": 46.0,
-      "step": 4365
-    },
-    {
-      "epoch": 0.7030878859857482,
-      "grad_norm": 0.00064177653985098,
-      "learning_rate": 0.00019999975717215675,
-      "loss": 46.0,
-      "step": 4366
-    },
-    {
-      "epoch": 0.7032489230645356,
-      "grad_norm": 0.001236030482687056,
-      "learning_rate": 0.0001999997570606528,
-      "loss": 46.0,
-      "step": 4367
-    },
-    {
-      "epoch": 0.703409960143323,
-      "grad_norm": 0.0017155156238004565,
-      "learning_rate": 0.00019999975694912325,
-      "loss": 46.0,
-      "step": 4368
-    },
-    {
-      "epoch": 0.7035709972221104,
-      "grad_norm": 0.0027009411714971066,
-      "learning_rate": 0.0001999997568375681,
-      "loss": 46.0,
-      "step": 4369
-    },
-    {
-      "epoch": 0.7037320343008978,
-      "grad_norm": 0.0006052309763617814,
-      "learning_rate": 0.00019999975672598734,
-      "loss": 46.0,
-      "step": 4370
-    },
-    {
-      "epoch": 0.7038930713796852,
-      "grad_norm": 0.0006586650270037353,
-      "learning_rate": 0.000199999756614381,
-      "loss": 46.0,
-      "step": 4371
-    },
-    {
-      "epoch": 0.7040541084584726,
-      "grad_norm": 0.000682335696183145,
-      "learning_rate": 0.00019999975650274908,
-      "loss": 46.0,
-      "step": 4372
-    },
-    {
-      "epoch": 0.70421514553726,
-      "grad_norm": 0.0015550417592749,
-      "learning_rate": 0.00019999975639109155,
-      "loss": 46.0,
-      "step": 4373
-    },
-    {
-      "epoch": 0.7043761826160474,
-      "grad_norm": 0.0010525945108383894,
-      "learning_rate": 0.00019999975627940842,
-      "loss": 46.0,
-      "step": 4374
-    },
-    {
-      "epoch": 0.7045372196948347,
-      "grad_norm": 0.0004901184001937509,
-      "learning_rate": 0.00019999975616769974,
-      "loss": 46.0,
-      "step": 4375
-    },
-    {
-      "epoch": 0.7046982567736221,
-      "grad_norm": 0.0007887351093813777,
-      "learning_rate": 0.00019999975605596541,
-      "loss": 46.0,
-      "step": 4376
-    },
-    {
-      "epoch": 0.7048592938524095,
-      "grad_norm": 0.00023493637854699045,
-      "learning_rate": 0.0001999997559442055,
-      "loss": 46.0,
-      "step": 4377
-    },
-    {
-      "epoch": 0.7050203309311969,
-      "grad_norm": 0.0015376863302662969,
-      "learning_rate": 0.00019999975583242,
-      "loss": 46.0,
-      "step": 4378
-    },
-    {
-      "epoch": 0.7051813680099843,
-      "grad_norm": 0.000744383898563683,
-      "learning_rate": 0.00019999975572060892,
-      "loss": 46.0,
-      "step": 4379
-    },
-    {
-      "epoch": 0.7053424050887717,
-      "grad_norm": 0.0011030557798221707,
-      "learning_rate": 0.00019999975560877221,
-      "loss": 46.0,
-      "step": 4380
-    },
-    {
-      "epoch": 0.7055034421675591,
-      "grad_norm": 0.0008488588500767946,
-      "learning_rate": 0.00019999975549690993,
-      "loss": 46.0,
-      "step": 4381
-    },
-    {
-      "epoch": 0.7056644792463465,
-      "grad_norm": 0.0008094905642792583,
-      "learning_rate": 0.00019999975538502205,
-      "loss": 46.0,
-      "step": 4382
-    },
-    {
-      "epoch": 0.7058255163251339,
-      "grad_norm": 0.0013153315521776676,
-      "learning_rate": 0.0001999997552731086,
-      "loss": 46.0,
-      "step": 4383
-    },
-    {
-      "epoch": 0.7059865534039212,
-      "grad_norm": 0.0008418290526606143,
-      "learning_rate": 0.0001999997551611695,
-      "loss": 46.0,
-      "step": 4384
-    },
-    {
-      "epoch": 0.7061475904827087,
-      "grad_norm": 0.0007125269039534032,
-      "learning_rate": 0.00019999975504920485,
-      "loss": 46.0,
-      "step": 4385
-    },
-    {
-      "epoch": 0.706308627561496,
-      "grad_norm": 0.0017784794326871634,
-      "learning_rate": 0.00019999975493721462,
-      "loss": 46.0,
-      "step": 4386
-    },
-    {
-      "epoch": 0.7064696646402834,
-      "grad_norm": 0.0009435164392925799,
-      "learning_rate": 0.00019999975482519876,
-      "loss": 46.0,
-      "step": 4387
-    },
-    {
-      "epoch": 0.7066307017190708,
-      "grad_norm": 0.0021223300136625767,
-      "learning_rate": 0.0001999997547131573,
-      "loss": 46.0,
-      "step": 4388
-    },
-    {
-      "epoch": 0.7067917387978582,
-      "grad_norm": 0.0007367211510427296,
-      "learning_rate": 0.00019999975460109024,
-      "loss": 46.0,
-      "step": 4389
-    },
-    {
-      "epoch": 0.7069527758766456,
-      "grad_norm": 0.0016931348945945501,
-      "learning_rate": 0.0001999997544889976,
-      "loss": 46.0,
-      "step": 4390
-    },
-    {
-      "epoch": 0.707113812955433,
-      "grad_norm": 0.0009905026527121663,
-      "learning_rate": 0.0001999997543768794,
-      "loss": 46.0,
-      "step": 4391
-    },
-    {
-      "epoch": 0.7072748500342204,
-      "grad_norm": 0.00047286628978326917,
-      "learning_rate": 0.00019999975426473557,
-      "loss": 46.0,
-      "step": 4392
-    },
-    {
-      "epoch": 0.7074358871130078,
-      "grad_norm": 0.0011455706553533673,
-      "learning_rate": 0.00019999975415256613,
-      "loss": 46.0,
-      "step": 4393
-    },
-    {
-      "epoch": 0.7075969241917952,
-      "grad_norm": 0.003250241745263338,
-      "learning_rate": 0.00019999975404037112,
-      "loss": 46.0,
-      "step": 4394
-    },
-    {
-      "epoch": 0.7077579612705825,
-      "grad_norm": 0.0009388535981997848,
-      "learning_rate": 0.0001999997539281505,
-      "loss": 46.0,
-      "step": 4395
-    },
-    {
-      "epoch": 0.7079189983493699,
-      "grad_norm": 0.003122034016996622,
-      "learning_rate": 0.0001999997538159043,
-      "loss": 46.0,
-      "step": 4396
-    },
-    {
-      "epoch": 0.7080800354281573,
-      "grad_norm": 0.0007923035882413387,
-      "learning_rate": 0.00019999975370363249,
-      "loss": 46.0,
-      "step": 4397
-    },
-    {
-      "epoch": 0.7082410725069447,
-      "grad_norm": 0.0013338671997189522,
-      "learning_rate": 0.0001999997535913351,
-      "loss": 46.0,
-      "step": 4398
-    },
-    {
-      "epoch": 0.7084021095857321,
-      "grad_norm": 0.0017197271808981895,
-      "learning_rate": 0.00019999975347901211,
-      "loss": 46.0,
-      "step": 4399
-    },
-    {
-      "epoch": 0.7085631466645195,
-      "grad_norm": 0.0010688776383176446,
-      "learning_rate": 0.00019999975336666353,
-      "loss": 46.0,
-      "step": 4400
-    },
-    {
-      "epoch": 0.7087241837433069,
-      "grad_norm": 0.0008712644339539111,
-      "learning_rate": 0.00019999975325428934,
-      "loss": 46.0,
-      "step": 4401
-    },
-    {
-      "epoch": 0.7088852208220943,
-      "grad_norm": 0.0006837205146439373,
-      "learning_rate": 0.00019999975314188956,
-      "loss": 46.0,
-      "step": 4402
-    },
-    {
-      "epoch": 0.7090462579008817,
-      "grad_norm": 0.0012785367434844375,
-      "learning_rate": 0.00019999975302946422,
-      "loss": 46.0,
-      "step": 4403
-    },
-    {
-      "epoch": 0.7092072949796691,
-      "grad_norm": 0.0025434347335249186,
-      "learning_rate": 0.00019999975291701323,
-      "loss": 46.0,
-      "step": 4404
-    },
-    {
-      "epoch": 0.7093683320584565,
-      "grad_norm": 0.0009659446659497917,
-      "learning_rate": 0.00019999975280453666,
-      "loss": 46.0,
-      "step": 4405
-    },
-    {
-      "epoch": 0.7095293691372438,
-      "grad_norm": 0.0008181935409083962,
-      "learning_rate": 0.00019999975269203453,
-      "loss": 46.0,
-      "step": 4406
-    },
-    {
-      "epoch": 0.7096904062160312,
-      "grad_norm": 0.0018432833021506667,
-      "learning_rate": 0.0001999997525795068,
-      "loss": 46.0,
-      "step": 4407
-    },
-    {
-      "epoch": 0.7098514432948186,
-      "grad_norm": 0.0020430530421435833,
-      "learning_rate": 0.00019999975246695343,
-      "loss": 46.0,
-      "step": 4408
-    },
-    {
-      "epoch": 0.710012480373606,
-      "grad_norm": 0.0013659343821927905,
-      "learning_rate": 0.00019999975235437449,
-      "loss": 46.0,
-      "step": 4409
-    },
-    {
-      "epoch": 0.7101735174523934,
-      "grad_norm": 0.0007254041847772896,
-      "learning_rate": 0.00019999975224176995,
-      "loss": 46.0,
-      "step": 4410
-    },
-    {
-      "epoch": 0.7103345545311808,
-      "grad_norm": 0.0003318408562336117,
-      "learning_rate": 0.00019999975212913983,
-      "loss": 46.0,
-      "step": 4411
-    },
-    {
-      "epoch": 0.7104955916099682,
-      "grad_norm": 0.0016765260370448232,
-      "learning_rate": 0.0001999997520164841,
-      "loss": 46.0,
-      "step": 4412
-    },
-    {
-      "epoch": 0.7106566286887556,
-      "grad_norm": 0.0005592864472419024,
-      "learning_rate": 0.00019999975190380278,
-      "loss": 46.0,
-      "step": 4413
-    },
-    {
-      "epoch": 0.710817665767543,
-      "grad_norm": 0.0014356225728988647,
-      "learning_rate": 0.00019999975179109587,
-      "loss": 46.0,
-      "step": 4414
-    },
-    {
-      "epoch": 0.7109787028463304,
-      "grad_norm": 0.0015646725660189986,
-      "learning_rate": 0.00019999975167836337,
-      "loss": 46.0,
-      "step": 4415
-    },
-    {
-      "epoch": 0.7111397399251178,
-      "grad_norm": 0.0006083520711399615,
-      "learning_rate": 0.00019999975156560526,
-      "loss": 46.0,
-      "step": 4416
-    },
-    {
-      "epoch": 0.7113007770039051,
-      "grad_norm": 0.0007206778391264379,
-      "learning_rate": 0.00019999975145282157,
-      "loss": 46.0,
-      "step": 4417
-    },
-    {
-      "epoch": 0.7114618140826925,
-      "grad_norm": 0.0005120691494084895,
-      "learning_rate": 0.0001999997513400123,
-      "loss": 46.0,
-      "step": 4418
-    },
-    {
-      "epoch": 0.7116228511614799,
-      "grad_norm": 0.001734325778670609,
-      "learning_rate": 0.0001999997512271774,
-      "loss": 46.0,
-      "step": 4419
-    },
-    {
-      "epoch": 0.7117838882402673,
-      "grad_norm": 0.0011700481409206986,
-      "learning_rate": 0.00019999975111431692,
-      "loss": 46.0,
-      "step": 4420
-    },
-    {
-      "epoch": 0.7119449253190547,
-      "grad_norm": 0.0006509965169243515,
-      "learning_rate": 0.00019999975100143082,
-      "loss": 46.0,
-      "step": 4421
-    },
-    {
-      "epoch": 0.7121059623978421,
-      "grad_norm": 0.001998740714043379,
-      "learning_rate": 0.00019999975088851916,
-      "loss": 46.0,
-      "step": 4422
-    },
-    {
-      "epoch": 0.7122669994766295,
-      "grad_norm": 0.00327190849930048,
-      "learning_rate": 0.00019999975077558192,
-      "loss": 46.0,
-      "step": 4423
-    },
-    {
-      "epoch": 0.7124280365554169,
-      "grad_norm": 0.0019606011919677258,
-      "learning_rate": 0.00019999975066261905,
-      "loss": 46.0,
-      "step": 4424
-    },
-    {
-      "epoch": 0.7125890736342043,
-      "grad_norm": 0.0006763917626813054,
-      "learning_rate": 0.00019999975054963058,
-      "loss": 46.0,
-      "step": 4425
-    },
-    {
-      "epoch": 0.7127501107129917,
-      "grad_norm": 0.0006627399125136435,
-      "learning_rate": 0.00019999975043661654,
-      "loss": 46.0,
-      "step": 4426
-    },
-    {
-      "epoch": 0.7129111477917791,
-      "grad_norm": 0.0005783293163403869,
-      "learning_rate": 0.0001999997503235769,
-      "loss": 46.0,
-      "step": 4427
-    },
-    {
-      "epoch": 0.7130721848705665,
-      "grad_norm": 0.001110888784751296,
-      "learning_rate": 0.00019999975021051165,
-      "loss": 46.0,
-      "step": 4428
-    },
-    {
-      "epoch": 0.7132332219493538,
-      "grad_norm": 0.001701586996205151,
-      "learning_rate": 0.0001999997500974208,
-      "loss": 46.0,
-      "step": 4429
-    },
-    {
-      "epoch": 0.7133942590281412,
-      "grad_norm": 0.001331550651229918,
-      "learning_rate": 0.0001999997499843044,
-      "loss": 46.0,
-      "step": 4430
-    },
-    {
-      "epoch": 0.7135552961069286,
-      "grad_norm": 0.0012772154295817018,
-      "learning_rate": 0.00019999974987116236,
-      "loss": 46.0,
-      "step": 4431
-    },
-    {
-      "epoch": 0.713716333185716,
-      "grad_norm": 0.0005927607417106628,
-      "learning_rate": 0.00019999974975799475,
-      "loss": 46.0,
-      "step": 4432
-    },
-    {
-      "epoch": 0.7138773702645034,
-      "grad_norm": 0.0010581662645563483,
-      "learning_rate": 0.00019999974964480152,
-      "loss": 46.0,
-      "step": 4433
-    },
-    {
-      "epoch": 0.7140384073432908,
-      "grad_norm": 0.0010330479126423597,
-      "learning_rate": 0.0001999997495315827,
-      "loss": 46.0,
-      "step": 4434
-    },
-    {
-      "epoch": 0.7141994444220782,
-      "grad_norm": 0.00033568430808372796,
-      "learning_rate": 0.00019999974941833833,
-      "loss": 46.0,
-      "step": 4435
-    },
-    {
-      "epoch": 0.7143604815008656,
-      "grad_norm": 0.00046789299813099205,
-      "learning_rate": 0.00019999974930506832,
-      "loss": 46.0,
-      "step": 4436
-    },
-    {
-      "epoch": 0.714521518579653,
-      "grad_norm": 0.0006808977341279387,
-      "learning_rate": 0.00019999974919177271,
-      "loss": 46.0,
-      "step": 4437
-    },
-    {
-      "epoch": 0.7146825556584404,
-      "grad_norm": 0.003396762302145362,
-      "learning_rate": 0.00019999974907845152,
-      "loss": 46.0,
-      "step": 4438
-    },
-    {
-      "epoch": 0.7148435927372278,
-      "grad_norm": 0.0006392574869096279,
-      "learning_rate": 0.00019999974896510475,
-      "loss": 46.0,
-      "step": 4439
-    },
-    {
-      "epoch": 0.7150046298160151,
-      "grad_norm": 0.0003603216609917581,
-      "learning_rate": 0.00019999974885173235,
-      "loss": 46.0,
-      "step": 4440
-    },
-    {
-      "epoch": 0.7151656668948025,
-      "grad_norm": 0.00169837917201221,
-      "learning_rate": 0.0001999997487383344,
-      "loss": 46.0,
-      "step": 4441
-    },
-    {
-      "epoch": 0.7153267039735899,
-      "grad_norm": 0.005692393984645605,
-      "learning_rate": 0.0001999997486249108,
-      "loss": 46.0,
-      "step": 4442
-    },
-    {
-      "epoch": 0.7154877410523773,
-      "grad_norm": 0.0006954838754609227,
-      "learning_rate": 0.00019999974851146165,
-      "loss": 46.0,
-      "step": 4443
-    },
-    {
-      "epoch": 0.7156487781311647,
-      "grad_norm": 0.0010099103674292564,
-      "learning_rate": 0.00019999974839798691,
-      "loss": 46.0,
-      "step": 4444
-    },
-    {
-      "epoch": 0.7158098152099521,
-      "grad_norm": 0.00042093818774446845,
-      "learning_rate": 0.00019999974828448656,
-      "loss": 46.0,
-      "step": 4445
-    },
-    {
-      "epoch": 0.7159708522887395,
-      "grad_norm": 0.00670892046764493,
-      "learning_rate": 0.00019999974817096062,
-      "loss": 46.0,
-      "step": 4446
-    },
-    {
-      "epoch": 0.7161318893675269,
-      "grad_norm": 0.0010251224739477038,
-      "learning_rate": 0.0001999997480574091,
-      "loss": 46.0,
-      "step": 4447
-    },
-    {
-      "epoch": 0.7162929264463143,
-      "grad_norm": 0.0010518493363633752,
-      "learning_rate": 0.00019999974794383194,
-      "loss": 46.0,
-      "step": 4448
-    },
-    {
-      "epoch": 0.7164539635251016,
-      "grad_norm": 0.000520506757311523,
-      "learning_rate": 0.00019999974783022921,
-      "loss": 46.0,
-      "step": 4449
-    },
-    {
-      "epoch": 0.716615000603889,
-      "grad_norm": 0.0029253712855279446,
-      "learning_rate": 0.0001999997477166009,
-      "loss": 46.0,
-      "step": 4450
-    },
-    {
-      "epoch": 0.7167760376826764,
-      "grad_norm": 0.00048601144226267934,
-      "learning_rate": 0.00019999974760294696,
-      "loss": 46.0,
-      "step": 4451
-    },
-    {
-      "epoch": 0.7169370747614638,
-      "grad_norm": 0.001080303336493671,
-      "learning_rate": 0.00019999974748926747,
-      "loss": 46.0,
-      "step": 4452
-    },
-    {
-      "epoch": 0.7170981118402512,
-      "grad_norm": 0.0003332775959279388,
-      "learning_rate": 0.00019999974737556234,
-      "loss": 46.0,
-      "step": 4453
-    },
-    {
-      "epoch": 0.7172591489190386,
-      "grad_norm": 0.0013118890346959233,
-      "learning_rate": 0.00019999974726183164,
-      "loss": 46.0,
-      "step": 4454
-    },
-    {
-      "epoch": 0.717420185997826,
-      "grad_norm": 0.0007557226344943047,
-      "learning_rate": 0.00019999974714807534,
-      "loss": 46.0,
-      "step": 4455
-    },
-    {
-      "epoch": 0.7175812230766134,
-      "grad_norm": 0.00036685456871055067,
-      "learning_rate": 0.00019999974703429344,
-      "loss": 46.0,
-      "step": 4456
-    },
-    {
-      "epoch": 0.7177422601554008,
-      "grad_norm": 0.001007641083560884,
-      "learning_rate": 0.00019999974692048596,
-      "loss": 46.0,
-      "step": 4457
-    },
-    {
-      "epoch": 0.7179032972341882,
-      "grad_norm": 0.0009070277446880937,
-      "learning_rate": 0.0001999997468066529,
-      "loss": 46.0,
-      "step": 4458
-    },
-    {
-      "epoch": 0.7180643343129756,
-      "grad_norm": 0.0016711081843823195,
-      "learning_rate": 0.00019999974669279418,
-      "loss": 46.0,
-      "step": 4459
-    },
-    {
-      "epoch": 0.7182253713917629,
-      "grad_norm": 0.0012451380025595427,
-      "learning_rate": 0.0001999997465789099,
-      "loss": 46.0,
-      "step": 4460
-    },
-    {
-      "epoch": 0.7183864084705504,
-      "grad_norm": 0.002102638129144907,
-      "learning_rate": 0.00019999974646500005,
-      "loss": 46.0,
-      "step": 4461
-    },
-    {
-      "epoch": 0.7185474455493377,
-      "grad_norm": 0.0018069209763780236,
-      "learning_rate": 0.0001999997463510646,
-      "loss": 46.0,
-      "step": 4462
-    },
-    {
-      "epoch": 0.7187084826281251,
-      "grad_norm": 0.0016833161935210228,
-      "learning_rate": 0.00019999974623710354,
-      "loss": 46.0,
-      "step": 4463
-    },
-    {
-      "epoch": 0.7188695197069125,
-      "grad_norm": 0.0006691787275485694,
-      "learning_rate": 0.00019999974612311687,
-      "loss": 46.0,
-      "step": 4464
-    },
-    {
-      "epoch": 0.7190305567856999,
-      "grad_norm": 0.0006645992980338633,
-      "learning_rate": 0.00019999974600910464,
-      "loss": 46.0,
-      "step": 4465
-    },
-    {
-      "epoch": 0.7191915938644873,
-      "grad_norm": 0.0024473455268889666,
-      "learning_rate": 0.0001999997458950668,
-      "loss": 46.0,
-      "step": 4466
-    },
-    {
-      "epoch": 0.7193526309432747,
-      "grad_norm": 0.004251326899975538,
-      "learning_rate": 0.00019999974578100335,
-      "loss": 46.0,
-      "step": 4467
-    },
-    {
-      "epoch": 0.7195136680220621,
-      "grad_norm": 0.001146490452811122,
-      "learning_rate": 0.0001999997456669143,
-      "loss": 46.0,
-      "step": 4468
-    },
-    {
-      "epoch": 0.7196747051008495,
-      "grad_norm": 0.0014277849113568664,
-      "learning_rate": 0.00019999974555279967,
-      "loss": 46.0,
-      "step": 4469
-    },
-    {
-      "epoch": 0.7198357421796369,
-      "grad_norm": 0.0005654064007103443,
-      "learning_rate": 0.00019999974543865947,
-      "loss": 46.0,
-      "step": 4470
-    },
-    {
-      "epoch": 0.7199967792584242,
-      "grad_norm": 0.0011332668364048004,
-      "learning_rate": 0.00019999974532449363,
-      "loss": 46.0,
-      "step": 4471
-    },
-    {
-      "epoch": 0.7201578163372117,
-      "grad_norm": 0.0005477855447679758,
-      "learning_rate": 0.00019999974521030223,
-      "loss": 46.0,
-      "step": 4472
-    },
-    {
-      "epoch": 0.720318853415999,
-      "grad_norm": 0.0022119577042758465,
-      "learning_rate": 0.00019999974509608522,
-      "loss": 46.0,
-      "step": 4473
-    },
-    {
-      "epoch": 0.7204798904947864,
-      "grad_norm": 0.00034289906034246087,
-      "learning_rate": 0.00019999974498184262,
-      "loss": 46.0,
-      "step": 4474
-    },
-    {
-      "epoch": 0.7206409275735738,
-      "grad_norm": 0.0017581023275852203,
-      "learning_rate": 0.00019999974486757443,
-      "loss": 46.0,
-      "step": 4475
-    },
-    {
-      "epoch": 0.7208019646523612,
-      "grad_norm": 0.0008935669902712107,
-      "learning_rate": 0.00019999974475328066,
-      "loss": 46.0,
-      "step": 4476
-    },
-    {
-      "epoch": 0.7209630017311486,
-      "grad_norm": 0.0005180076695978642,
-      "learning_rate": 0.00019999974463896124,
-      "loss": 46.0,
-      "step": 4477
-    },
-    {
-      "epoch": 0.721124038809936,
-      "grad_norm": 0.0009835090022534132,
-      "learning_rate": 0.00019999974452461627,
-      "loss": 46.0,
-      "step": 4478
-    },
-    {
-      "epoch": 0.7212850758887234,
-      "grad_norm": 0.0005806116969324648,
-      "learning_rate": 0.0001999997444102457,
-      "loss": 46.0,
-      "step": 4479
-    },
-    {
-      "epoch": 0.7214461129675108,
-      "grad_norm": 0.0012206181418150663,
-      "learning_rate": 0.00019999974429584953,
-      "loss": 46.0,
-      "step": 4480
-    },
-    {
-      "epoch": 0.7216071500462982,
-      "grad_norm": 0.0011013572802767158,
-      "learning_rate": 0.00019999974418142776,
-      "loss": 46.0,
-      "step": 4481
-    },
-    {
-      "epoch": 0.7217681871250855,
-      "grad_norm": 0.0010482007637619972,
-      "learning_rate": 0.0001999997440669804,
-      "loss": 46.0,
-      "step": 4482
-    },
-    {
-      "epoch": 0.721929224203873,
-      "grad_norm": 0.0012515585403889418,
-      "learning_rate": 0.00019999974395250748,
-      "loss": 46.0,
-      "step": 4483
-    },
-    {
-      "epoch": 0.7220902612826603,
-      "grad_norm": 0.0009090963285416365,
-      "learning_rate": 0.00019999974383800892,
-      "loss": 46.0,
-      "step": 4484
-    },
-    {
-      "epoch": 0.7222512983614477,
-      "grad_norm": 0.0033045061863958836,
-      "learning_rate": 0.00019999974372348476,
-      "loss": 46.0,
-      "step": 4485
-    },
-    {
-      "epoch": 0.7224123354402351,
-      "grad_norm": 0.0010728840716183186,
-      "learning_rate": 0.00019999974360893503,
-      "loss": 46.0,
-      "step": 4486
-    },
-    {
-      "epoch": 0.7225733725190225,
-      "grad_norm": 0.0017847331473603845,
-      "learning_rate": 0.0001999997434943597,
-      "loss": 46.0,
-      "step": 4487
-    },
-    {
-      "epoch": 0.7227344095978099,
-      "grad_norm": 0.0009227863629348576,
-      "learning_rate": 0.00019999974337975879,
-      "loss": 46.0,
-      "step": 4488
-    },
-    {
-      "epoch": 0.7228954466765973,
-      "grad_norm": 0.0019123065285384655,
-      "learning_rate": 0.00019999974326513224,
-      "loss": 46.0,
-      "step": 4489
-    },
-    {
-      "epoch": 0.7230564837553847,
-      "grad_norm": 0.00047635557712055743,
-      "learning_rate": 0.00019999974315048014,
-      "loss": 46.0,
-      "step": 4490
-    },
-    {
-      "epoch": 0.7232175208341721,
-      "grad_norm": 0.00041297514690086246,
-      "learning_rate": 0.0001999997430358024,
-      "loss": 46.0,
-      "step": 4491
-    },
-    {
-      "epoch": 0.7233785579129595,
-      "grad_norm": 0.0012477511772885919,
-      "learning_rate": 0.00019999974292109912,
-      "loss": 46.0,
-      "step": 4492
-    },
-    {
-      "epoch": 0.7235395949917468,
-      "grad_norm": 0.000492687860969454,
-      "learning_rate": 0.0001999997428063702,
-      "loss": 46.0,
-      "step": 4493
-    },
-    {
-      "epoch": 0.7237006320705343,
-      "grad_norm": 0.001497862278483808,
-      "learning_rate": 0.00019999974269161572,
-      "loss": 46.0,
-      "step": 4494
-    },
-    {
-      "epoch": 0.7238616691493216,
-      "grad_norm": 0.0021726367995142937,
-      "learning_rate": 0.00019999974257683563,
-      "loss": 46.0,
-      "step": 4495
-    },
-    {
-      "epoch": 0.724022706228109,
-      "grad_norm": 0.00046507263323292136,
-      "learning_rate": 0.00019999974246202992,
-      "loss": 46.0,
-      "step": 4496
-    },
-    {
-      "epoch": 0.7241837433068964,
-      "grad_norm": 0.0007227102178148925,
-      "learning_rate": 0.00019999974234719865,
-      "loss": 46.0,
-      "step": 4497
-    },
-    {
-      "epoch": 0.7243447803856838,
-      "grad_norm": 0.0007895276648923755,
-      "learning_rate": 0.00019999974223234177,
-      "loss": 46.0,
-      "step": 4498
-    },
-    {
-      "epoch": 0.7245058174644712,
-      "grad_norm": 0.0011881771497428417,
-      "learning_rate": 0.0001999997421174593,
-      "loss": 46.0,
-      "step": 4499
-    },
-    {
-      "epoch": 0.7246668545432586,
-      "grad_norm": 0.003271875437349081,
-      "learning_rate": 0.00019999974200255122,
-      "loss": 46.0,
-      "step": 4500
-    },
-    {
-      "epoch": 0.724827891622046,
-      "grad_norm": 0.0007997031207196414,
-      "learning_rate": 0.00019999974188761758,
-      "loss": 46.0,
-      "step": 4501
-    },
-    {
-      "epoch": 0.7249889287008333,
-      "grad_norm": 0.0009179948829114437,
-      "learning_rate": 0.00019999974177265832,
-      "loss": 46.0,
-      "step": 4502
-    },
-    {
-      "epoch": 0.7251499657796208,
-      "grad_norm": 0.0017261349130421877,
-      "learning_rate": 0.00019999974165767345,
-      "loss": 46.0,
-      "step": 4503
-    },
-    {
-      "epoch": 0.7253110028584081,
-      "grad_norm": 0.0012372866040095687,
-      "learning_rate": 0.00019999974154266302,
-      "loss": 46.0,
-      "step": 4504
-    },
-    {
-      "epoch": 0.7254720399371956,
-      "grad_norm": 0.0008584472816437483,
-      "learning_rate": 0.000199999741427627,
-      "loss": 46.0,
-      "step": 4505
-    },
-    {
-      "epoch": 0.7256330770159829,
-      "grad_norm": 0.0012302917893975973,
-      "learning_rate": 0.00019999974131256534,
-      "loss": 46.0,
-      "step": 4506
-    },
-    {
-      "epoch": 0.7257941140947703,
-      "grad_norm": 0.0025850438978523016,
-      "learning_rate": 0.00019999974119747812,
-      "loss": 46.0,
-      "step": 4507
-    },
-    {
-      "epoch": 0.7259551511735577,
-      "grad_norm": 0.0013695707311853766,
-      "learning_rate": 0.0001999997410823653,
-      "loss": 46.0,
-      "step": 4508
-    },
-    {
-      "epoch": 0.7261161882523451,
-      "grad_norm": 0.0006834747036918998,
-      "learning_rate": 0.00019999974096722686,
-      "loss": 46.0,
-      "step": 4509
-    },
-    {
-      "epoch": 0.7262772253311325,
-      "grad_norm": 0.000519837427418679,
-      "learning_rate": 0.00019999974085206285,
-      "loss": 46.0,
-      "step": 4510
-    },
-    {
-      "epoch": 0.7264382624099199,
-      "grad_norm": 0.0011834946926683187,
-      "learning_rate": 0.00019999974073687325,
-      "loss": 46.0,
-      "step": 4511
-    },
-    {
-      "epoch": 0.7265992994887073,
-      "grad_norm": 0.0006753116031177342,
-      "learning_rate": 0.00019999974062165804,
-      "loss": 46.0,
-      "step": 4512
-    },
-    {
-      "epoch": 0.7267603365674946,
-      "grad_norm": 0.0022536867763847113,
-      "learning_rate": 0.00019999974050641725,
-      "loss": 46.0,
-      "step": 4513
-    },
-    {
-      "epoch": 0.7269213736462821,
-      "grad_norm": 0.0021636977326124907,
-      "learning_rate": 0.00019999974039115086,
-      "loss": 46.0,
-      "step": 4514
-    },
-    {
-      "epoch": 0.7270824107250694,
-      "grad_norm": 0.00033525892649777234,
-      "learning_rate": 0.0001999997402758589,
-      "loss": 46.0,
-      "step": 4515
-    },
-    {
-      "epoch": 0.7272434478038569,
-      "grad_norm": 0.0008815217297524214,
-      "learning_rate": 0.00019999974016054128,
-      "loss": 46.0,
-      "step": 4516
-    },
-    {
-      "epoch": 0.7274044848826442,
-      "grad_norm": 0.0015479987487196922,
-      "learning_rate": 0.0001999997400451981,
-      "loss": 46.0,
-      "step": 4517
-    },
-    {
-      "epoch": 0.7275655219614316,
-      "grad_norm": 0.0007409134414047003,
-      "learning_rate": 0.00019999973992982932,
-      "loss": 46.0,
-      "step": 4518
-    },
-    {
-      "epoch": 0.727726559040219,
-      "grad_norm": 0.0018958671716973186,
-      "learning_rate": 0.00019999973981443497,
-      "loss": 46.0,
-      "step": 4519
-    },
-    {
-      "epoch": 0.7278875961190064,
-      "grad_norm": 0.0006282523390837014,
-      "learning_rate": 0.00019999973969901503,
-      "loss": 46.0,
-      "step": 4520
-    },
-    {
-      "epoch": 0.7280486331977938,
-      "grad_norm": 0.00042072057840414345,
-      "learning_rate": 0.00019999973958356946,
-      "loss": 46.0,
-      "step": 4521
-    },
-    {
-      "epoch": 0.7282096702765812,
-      "grad_norm": 0.0006787588354200125,
-      "learning_rate": 0.00019999973946809832,
-      "loss": 46.0,
-      "step": 4522
-    },
-    {
-      "epoch": 0.7283707073553686,
-      "grad_norm": 0.0011091604828834534,
-      "learning_rate": 0.00019999973935260154,
-      "loss": 46.0,
-      "step": 4523
-    },
-    {
-      "epoch": 0.7285317444341559,
-      "grad_norm": 0.0007960863877087831,
-      "learning_rate": 0.0001999997392370792,
-      "loss": 46.0,
-      "step": 4524
-    },
-    {
-      "epoch": 0.7286927815129434,
-      "grad_norm": 0.0013086351100355387,
-      "learning_rate": 0.00019999973912153128,
-      "loss": 46.0,
-      "step": 4525
-    },
-    {
-      "epoch": 0.7288538185917307,
-      "grad_norm": 0.0007039692136459053,
-      "learning_rate": 0.00019999973900595774,
-      "loss": 46.0,
-      "step": 4526
-    },
-    {
-      "epoch": 0.7290148556705182,
-      "grad_norm": 0.0005595731781795621,
-      "learning_rate": 0.00019999973889035864,
-      "loss": 46.0,
-      "step": 4527
-    },
-    {
-      "epoch": 0.7291758927493055,
-      "grad_norm": 0.0006170624401420355,
-      "learning_rate": 0.0001999997387747339,
-      "loss": 46.0,
-      "step": 4528
-    },
-    {
-      "epoch": 0.729336929828093,
-      "grad_norm": 0.0011318547185510397,
-      "learning_rate": 0.0001999997386590836,
-      "loss": 46.0,
-      "step": 4529
-    },
-    {
-      "epoch": 0.7294979669068803,
-      "grad_norm": 0.0026329183019697666,
-      "learning_rate": 0.00019999973854340768,
-      "loss": 46.0,
-      "step": 4530
-    },
-    {
-      "epoch": 0.7296590039856677,
-      "grad_norm": 0.0013854143908247352,
-      "learning_rate": 0.00019999973842770618,
-      "loss": 46.0,
-      "step": 4531
-    },
-    {
-      "epoch": 0.7298200410644551,
-      "grad_norm": 0.0020008685532957315,
-      "learning_rate": 0.00019999973831197907,
-      "loss": 46.0,
-      "step": 4532
-    },
-    {
-      "epoch": 0.7299810781432425,
-      "grad_norm": 0.0018082942115142941,
-      "learning_rate": 0.00019999973819622636,
-      "loss": 46.0,
-      "step": 4533
-    },
-    {
-      "epoch": 0.7301421152220299,
-      "grad_norm": 0.0003881430020555854,
-      "learning_rate": 0.0001999997380804481,
-      "loss": 46.0,
-      "step": 4534
-    },
-    {
-      "epoch": 0.7303031523008172,
-      "grad_norm": 0.000548550917301327,
-      "learning_rate": 0.0001999997379646442,
-      "loss": 46.0,
-      "step": 4535
-    },
-    {
-      "epoch": 0.7304641893796047,
-      "grad_norm": 0.000778438348788768,
-      "learning_rate": 0.00019999973784881473,
-      "loss": 46.0,
-      "step": 4536
-    },
-    {
-      "epoch": 0.730625226458392,
-      "grad_norm": 0.0023691642563790083,
-      "learning_rate": 0.00019999973773295965,
-      "loss": 46.0,
-      "step": 4537
-    },
-    {
-      "epoch": 0.7307862635371795,
-      "grad_norm": 0.0009080860763788223,
-      "learning_rate": 0.00019999973761707898,
-      "loss": 46.0,
-      "step": 4538
-    },
-    {
-      "epoch": 0.7309473006159668,
-      "grad_norm": 0.0006811475032009184,
-      "learning_rate": 0.0001999997375011727,
-      "loss": 46.0,
-      "step": 4539
-    },
-    {
-      "epoch": 0.7311083376947543,
-      "grad_norm": 0.0007978445501066744,
-      "learning_rate": 0.00019999973738524086,
-      "loss": 46.0,
-      "step": 4540
-    },
-    {
-      "epoch": 0.7312693747735416,
-      "grad_norm": 0.0005806766566820443,
-      "learning_rate": 0.0001999997372692834,
-      "loss": 46.0,
-      "step": 4541
-    },
-    {
-      "epoch": 0.731430411852329,
-      "grad_norm": 0.0007834543939679861,
-      "learning_rate": 0.00019999973715330036,
-      "loss": 46.0,
-      "step": 4542
-    },
-    {
-      "epoch": 0.7315914489311164,
-      "grad_norm": 0.0005286663072183728,
-      "learning_rate": 0.00019999973703729173,
-      "loss": 46.0,
-      "step": 4543
-    },
-    {
-      "epoch": 0.7317524860099038,
-      "grad_norm": 0.000779558380600065,
-      "learning_rate": 0.0001999997369212575,
-      "loss": 46.0,
-      "step": 4544
-    },
-    {
-      "epoch": 0.7319135230886912,
-      "grad_norm": 0.0010596184292808175,
-      "learning_rate": 0.00019999973680519766,
-      "loss": 46.0,
-      "step": 4545
-    },
-    {
-      "epoch": 0.7320745601674785,
-      "grad_norm": 0.005791361443698406,
-      "learning_rate": 0.00019999973668911224,
-      "loss": 46.0,
-      "step": 4546
-    },
-    {
-      "epoch": 0.732235597246266,
-      "grad_norm": 0.0008321021450683475,
-      "learning_rate": 0.0001999997365730012,
-      "loss": 46.0,
-      "step": 4547
-    },
-    {
-      "epoch": 0.7323966343250533,
-      "grad_norm": 0.0005218035657890141,
-      "learning_rate": 0.0001999997364568646,
-      "loss": 46.0,
-      "step": 4548
-    },
-    {
-      "epoch": 0.7325576714038408,
-      "grad_norm": 0.0017712480621412396,
-      "learning_rate": 0.00019999973634070238,
-      "loss": 46.0,
-      "step": 4549
-    },
-    {
-      "epoch": 0.7327187084826281,
-      "grad_norm": 0.001056362991221249,
-      "learning_rate": 0.00019999973622451456,
-      "loss": 46.0,
-      "step": 4550
-    },
-    {
-      "epoch": 0.7328797455614156,
-      "grad_norm": 0.0016243801219388843,
-      "learning_rate": 0.00019999973610830115,
-      "loss": 46.0,
-      "step": 4551
-    },
-    {
-      "epoch": 0.7330407826402029,
-      "grad_norm": 0.0028840191662311554,
-      "learning_rate": 0.00019999973599206218,
-      "loss": 46.0,
-      "step": 4552
-    },
-    {
-      "epoch": 0.7332018197189903,
-      "grad_norm": 0.0006374493823386729,
-      "learning_rate": 0.0001999997358757976,
-      "loss": 46.0,
-      "step": 4553
-    },
-    {
-      "epoch": 0.7333628567977777,
-      "grad_norm": 0.0019701439887285233,
-      "learning_rate": 0.0001999997357595074,
-      "loss": 46.0,
-      "step": 4554
-    },
-    {
-      "epoch": 0.733523893876565,
-      "grad_norm": 0.001144581357948482,
-      "learning_rate": 0.00019999973564319162,
-      "loss": 46.0,
-      "step": 4555
-    },
-    {
-      "epoch": 0.7336849309553525,
-      "grad_norm": 0.0007458243635483086,
-      "learning_rate": 0.00019999973552685022,
-      "loss": 46.0,
-      "step": 4556
-    },
-    {
-      "epoch": 0.7338459680341398,
-      "grad_norm": 0.0025849926751106977,
-      "learning_rate": 0.00019999973541048326,
-      "loss": 46.0,
-      "step": 4557
-    },
-    {
-      "epoch": 0.7340070051129273,
-      "grad_norm": 0.0012840854469686747,
-      "learning_rate": 0.00019999973529409072,
-      "loss": 46.0,
-      "step": 4558
-    },
-    {
-      "epoch": 0.7341680421917146,
-      "grad_norm": 0.0008985294844023883,
-      "learning_rate": 0.00019999973517767256,
-      "loss": 46.0,
-      "step": 4559
-    },
-    {
-      "epoch": 0.7343290792705021,
-      "grad_norm": 0.0006069798255339265,
-      "learning_rate": 0.0001999997350612288,
-      "loss": 46.0,
-      "step": 4560
-    },
-    {
-      "epoch": 0.7344901163492894,
-      "grad_norm": 0.002930633258074522,
-      "learning_rate": 0.00019999973494475945,
-      "loss": 46.0,
-      "step": 4561
-    },
-    {
-      "epoch": 0.7346511534280769,
-      "grad_norm": 0.0014089002506807446,
-      "learning_rate": 0.0001999997348282645,
-      "loss": 46.0,
-      "step": 4562
-    },
-    {
-      "epoch": 0.7348121905068642,
-      "grad_norm": 0.0005846631829626858,
-      "learning_rate": 0.000199999734711744,
-      "loss": 46.0,
-      "step": 4563
-    },
-    {
-      "epoch": 0.7349732275856516,
-      "grad_norm": 0.000707046187017113,
-      "learning_rate": 0.00019999973459519784,
-      "loss": 46.0,
-      "step": 4564
-    },
-    {
-      "epoch": 0.735134264664439,
-      "grad_norm": 0.00043157863547094166,
-      "learning_rate": 0.0001999997344786261,
-      "loss": 46.0,
-      "step": 4565
-    },
-    {
-      "epoch": 0.7352953017432263,
-      "grad_norm": 0.0039930869825184345,
-      "learning_rate": 0.00019999973436202878,
-      "loss": 46.0,
-      "step": 4566
-    },
-    {
-      "epoch": 0.7354563388220138,
-      "grad_norm": 0.00042738113552331924,
-      "learning_rate": 0.00019999973424540584,
-      "loss": 46.0,
-      "step": 4567
-    },
-    {
-      "epoch": 0.7356173759008011,
-      "grad_norm": 0.0006795520312152803,
-      "learning_rate": 0.00019999973412875734,
-      "loss": 46.0,
-      "step": 4568
-    },
-    {
-      "epoch": 0.7357784129795886,
-      "grad_norm": 0.0004120956582482904,
-      "learning_rate": 0.00019999973401208326,
-      "loss": 46.0,
-      "step": 4569
-    },
-    {
-      "epoch": 0.7359394500583759,
-      "grad_norm": 0.0004701597208622843,
-      "learning_rate": 0.00019999973389538353,
-      "loss": 46.0,
-      "step": 4570
-    },
-    {
-      "epoch": 0.7361004871371634,
-      "grad_norm": 0.0009315081988461316,
-      "learning_rate": 0.00019999973377865824,
-      "loss": 46.0,
-      "step": 4571
-    },
-    {
-      "epoch": 0.7362615242159507,
-      "grad_norm": 0.00033041753340512514,
-      "learning_rate": 0.00019999973366190737,
-      "loss": 46.0,
-      "step": 4572
-    },
-    {
-      "epoch": 0.7364225612947382,
-      "grad_norm": 0.0014328625984489918,
-      "learning_rate": 0.00019999973354513085,
-      "loss": 46.0,
-      "step": 4573
-    },
-    {
-      "epoch": 0.7365835983735255,
-      "grad_norm": 0.0002727169485297054,
-      "learning_rate": 0.00019999973342832878,
-      "loss": 46.0,
-      "step": 4574
-    },
-    {
-      "epoch": 0.736744635452313,
-      "grad_norm": 0.00032052627648226917,
-      "learning_rate": 0.00019999973331150112,
-      "loss": 46.0,
-      "step": 4575
-    },
-    {
-      "epoch": 0.7369056725311003,
-      "grad_norm": 0.0007307630148716271,
-      "learning_rate": 0.00019999973319464784,
-      "loss": 46.0,
-      "step": 4576
-    },
-    {
-      "epoch": 0.7370667096098876,
-      "grad_norm": 0.0003718679363373667,
-      "learning_rate": 0.00019999973307776897,
-      "loss": 46.0,
-      "step": 4577
-    },
-    {
-      "epoch": 0.7372277466886751,
-      "grad_norm": 0.0006524814525619149,
-      "learning_rate": 0.00019999973296086452,
-      "loss": 46.0,
-      "step": 4578
-    },
-    {
-      "epoch": 0.7373887837674624,
-      "grad_norm": 0.0005162352463230491,
-      "learning_rate": 0.00019999973284393448,
-      "loss": 46.0,
-      "step": 4579
-    },
-    {
-      "epoch": 0.7375498208462499,
-      "grad_norm": 0.0009220059146173298,
-      "learning_rate": 0.0001999997327269788,
-      "loss": 46.0,
-      "step": 4580
-    },
-    {
-      "epoch": 0.7377108579250372,
-      "grad_norm": 0.00027430258342064917,
-      "learning_rate": 0.00019999973260999756,
-      "loss": 46.0,
-      "step": 4581
-    },
-    {
-      "epoch": 0.7378718950038247,
-      "grad_norm": 0.0005880841636098921,
-      "learning_rate": 0.00019999973249299074,
-      "loss": 46.0,
-      "step": 4582
-    },
-    {
-      "epoch": 0.738032932082612,
-      "grad_norm": 0.00032640728750266135,
-      "learning_rate": 0.0001999997323759583,
-      "loss": 46.0,
-      "step": 4583
-    },
-    {
-      "epoch": 0.7381939691613995,
-      "grad_norm": 0.0015201785136014223,
-      "learning_rate": 0.00019999973225890027,
-      "loss": 46.0,
-      "step": 4584
-    },
-    {
-      "epoch": 0.7383550062401868,
-      "grad_norm": 0.00042293520527891815,
-      "learning_rate": 0.00019999973214181665,
-      "loss": 46.0,
-      "step": 4585
-    },
-    {
-      "epoch": 0.7385160433189742,
-      "grad_norm": 0.0008818499627523124,
-      "learning_rate": 0.00019999973202470742,
-      "loss": 46.0,
-      "step": 4586
-    },
-    {
-      "epoch": 0.7386770803977616,
-      "grad_norm": 0.0009171415586024523,
-      "learning_rate": 0.0001999997319075726,
-      "loss": 46.0,
-      "step": 4587
-    },
-    {
-      "epoch": 0.7388381174765489,
-      "grad_norm": 0.001969479024410248,
-      "learning_rate": 0.0001999997317904122,
-      "loss": 46.0,
-      "step": 4588
-    },
-    {
-      "epoch": 0.7389991545553364,
-      "grad_norm": 0.0025983729865401983,
-      "learning_rate": 0.00019999973167322618,
-      "loss": 46.0,
-      "step": 4589
-    },
-    {
-      "epoch": 0.7391601916341237,
-      "grad_norm": 0.0014474327908828855,
-      "learning_rate": 0.00019999973155601463,
-      "loss": 46.0,
-      "step": 4590
-    },
-    {
-      "epoch": 0.7393212287129112,
-      "grad_norm": 0.0032787180971354246,
-      "learning_rate": 0.0001999997314387774,
-      "loss": 46.0,
-      "step": 4591
-    },
-    {
-      "epoch": 0.7394822657916985,
-      "grad_norm": 0.0005762985674664378,
-      "learning_rate": 0.00019999973132151463,
-      "loss": 46.0,
-      "step": 4592
-    },
-    {
-      "epoch": 0.739643302870486,
-      "grad_norm": 0.001087381155230105,
-      "learning_rate": 0.00019999973120422623,
-      "loss": 46.0,
-      "step": 4593
-    },
-    {
-      "epoch": 0.7398043399492733,
-      "grad_norm": 0.000422496726969257,
-      "learning_rate": 0.00019999973108691225,
-      "loss": 46.0,
-      "step": 4594
-    },
-    {
-      "epoch": 0.7399653770280608,
-      "grad_norm": 0.004855965729802847,
-      "learning_rate": 0.0001999997309695727,
-      "loss": 46.0,
-      "step": 4595
-    },
-    {
-      "epoch": 0.7401264141068481,
-      "grad_norm": 0.0004002965579275042,
-      "learning_rate": 0.00019999973085220752,
-      "loss": 46.0,
-      "step": 4596
-    },
-    {
-      "epoch": 0.7402874511856355,
-      "grad_norm": 0.0005231106770224869,
-      "learning_rate": 0.00019999973073481675,
-      "loss": 46.0,
-      "step": 4597
-    },
-    {
-      "epoch": 0.7404484882644229,
-      "grad_norm": 0.000597749138250947,
-      "learning_rate": 0.0001999997306174004,
-      "loss": 46.0,
-      "step": 4598
-    },
-    {
-      "epoch": 0.7406095253432102,
-      "grad_norm": 0.00393320620059967,
-      "learning_rate": 0.00019999973049995845,
-      "loss": 46.0,
-      "step": 4599
-    },
-    {
-      "epoch": 0.7407705624219977,
-      "grad_norm": 0.001538130920380354,
-      "learning_rate": 0.0001999997303824909,
-      "loss": 46.0,
-      "step": 4600
-    },
-    {
-      "epoch": 0.740931599500785,
-      "grad_norm": 0.0006161427590996027,
-      "learning_rate": 0.00019999973026499777,
-      "loss": 46.0,
-      "step": 4601
-    },
-    {
-      "epoch": 0.7410926365795725,
-      "grad_norm": 0.0028469713870435953,
-      "learning_rate": 0.000199999730147479,
-      "loss": 46.0,
-      "step": 4602
-    },
-    {
-      "epoch": 0.7412536736583598,
-      "grad_norm": 0.0007435024017468095,
-      "learning_rate": 0.0001999997300299347,
-      "loss": 46.0,
-      "step": 4603
-    },
-    {
-      "epoch": 0.7414147107371473,
-      "grad_norm": 0.0004241475253365934,
-      "learning_rate": 0.00019999972991236478,
-      "loss": 46.0,
-      "step": 4604
-    },
-    {
-      "epoch": 0.7415757478159346,
-      "grad_norm": 0.0007312520174309611,
-      "learning_rate": 0.00019999972979476923,
-      "loss": 46.0,
-      "step": 4605
-    },
-    {
-      "epoch": 0.7417367848947221,
-      "grad_norm": 0.0013212421908974648,
-      "learning_rate": 0.00019999972967714812,
-      "loss": 46.0,
-      "step": 4606
-    },
-    {
-      "epoch": 0.7418978219735094,
-      "grad_norm": 0.0008603931055404246,
-      "learning_rate": 0.00019999972955950143,
-      "loss": 46.0,
-      "step": 4607
-    },
-    {
-      "epoch": 0.7420588590522967,
-      "grad_norm": 0.000343397434335202,
-      "learning_rate": 0.00019999972944182912,
-      "loss": 46.0,
-      "step": 4608
-    },
-    {
-      "epoch": 0.7422198961310842,
-      "grad_norm": 0.0011578669073060155,
-      "learning_rate": 0.00019999972932413122,
-      "loss": 46.0,
-      "step": 4609
-    },
-    {
-      "epoch": 0.7423809332098715,
-      "grad_norm": 0.0035870710853487253,
-      "learning_rate": 0.00019999972920640773,
-      "loss": 46.0,
-      "step": 4610
-    },
-    {
-      "epoch": 0.742541970288659,
-      "grad_norm": 0.000637476216070354,
-      "learning_rate": 0.00019999972908865866,
-      "loss": 46.0,
-      "step": 4611
-    },
-    {
-      "epoch": 0.7427030073674463,
-      "grad_norm": 0.0010959397768601775,
-      "learning_rate": 0.00019999972897088398,
-      "loss": 46.0,
-      "step": 4612
-    },
-    {
-      "epoch": 0.7428640444462338,
-      "grad_norm": 0.0009431284852325916,
-      "learning_rate": 0.00019999972885308367,
-      "loss": 46.0,
-      "step": 4613
-    },
-    {
-      "epoch": 0.7430250815250211,
-      "grad_norm": 0.00114214897621423,
-      "learning_rate": 0.00019999972873525781,
-      "loss": 46.0,
-      "step": 4614
-    },
-    {
-      "epoch": 0.7431861186038086,
-      "grad_norm": 0.0010769155342131853,
-      "learning_rate": 0.0001999997286174063,
-      "loss": 46.0,
-      "step": 4615
-    },
-    {
-      "epoch": 0.7433471556825959,
-      "grad_norm": 0.0019252679776400328,
-      "learning_rate": 0.00019999972849952928,
-      "loss": 46.0,
-      "step": 4616
-    },
-    {
-      "epoch": 0.7435081927613834,
-      "grad_norm": 0.0005160552100278437,
-      "learning_rate": 0.00019999972838162663,
-      "loss": 46.0,
-      "step": 4617
-    },
-    {
-      "epoch": 0.7436692298401707,
-      "grad_norm": 0.0005786482943221927,
-      "learning_rate": 0.00019999972826369836,
-      "loss": 46.0,
-      "step": 4618
-    },
-    {
-      "epoch": 0.743830266918958,
-      "grad_norm": 0.0003764034481719136,
-      "learning_rate": 0.0001999997281457445,
-      "loss": 46.0,
-      "step": 4619
-    },
-    {
-      "epoch": 0.7439913039977455,
-      "grad_norm": 0.0006265260744839907,
-      "learning_rate": 0.00019999972802776507,
-      "loss": 46.0,
-      "step": 4620
-    },
-    {
-      "epoch": 0.7441523410765328,
-      "grad_norm": 0.0012461300939321518,
-      "learning_rate": 0.00019999972790976002,
-      "loss": 46.0,
-      "step": 4621
-    },
-    {
-      "epoch": 0.7443133781553203,
-      "grad_norm": 0.0006371387280523777,
-      "learning_rate": 0.0001999997277917294,
-      "loss": 46.0,
-      "step": 4622
-    },
-    {
-      "epoch": 0.7444744152341076,
-      "grad_norm": 0.00048348153359256685,
-      "learning_rate": 0.00019999972767367318,
-      "loss": 46.0,
-      "step": 4623
-    },
-    {
-      "epoch": 0.7446354523128951,
-      "grad_norm": 0.0005813002935610712,
-      "learning_rate": 0.00019999972755559137,
-      "loss": 46.0,
-      "step": 4624
-    },
-    {
-      "epoch": 0.7447964893916824,
-      "grad_norm": 0.0007331267697736621,
-      "learning_rate": 0.00019999972743748394,
-      "loss": 46.0,
-      "step": 4625
-    },
-    {
-      "epoch": 0.7449575264704699,
-      "grad_norm": 0.001337914727628231,
-      "learning_rate": 0.00019999972731935093,
-      "loss": 46.0,
-      "step": 4626
-    },
-    {
-      "epoch": 0.7451185635492572,
-      "grad_norm": 0.002132835565134883,
-      "learning_rate": 0.00019999972720119232,
-      "loss": 46.0,
-      "step": 4627
-    },
-    {
-      "epoch": 0.7452796006280447,
-      "grad_norm": 0.0005022824043408036,
-      "learning_rate": 0.0001999997270830081,
-      "loss": 46.0,
-      "step": 4628
-    },
-    {
-      "epoch": 0.745440637706832,
-      "grad_norm": 0.00043734745122492313,
-      "learning_rate": 0.00019999972696479833,
-      "loss": 46.0,
-      "step": 4629
-    },
-    {
-      "epoch": 0.7456016747856193,
-      "grad_norm": 0.001717623439617455,
-      "learning_rate": 0.00019999972684656294,
-      "loss": 46.0,
-      "step": 4630
-    },
-    {
-      "epoch": 0.7457627118644068,
-      "grad_norm": 0.000820228538941592,
-      "learning_rate": 0.00019999972672830196,
-      "loss": 46.0,
-      "step": 4631
-    },
-    {
-      "epoch": 0.7459237489431941,
-      "grad_norm": 0.0009194695157930255,
-      "learning_rate": 0.00019999972661001537,
-      "loss": 46.0,
-      "step": 4632
-    },
-    {
-      "epoch": 0.7460847860219816,
-      "grad_norm": 0.001057522022165358,
-      "learning_rate": 0.0001999997264917032,
-      "loss": 46.0,
-      "step": 4633
-    },
-    {
-      "epoch": 0.7462458231007689,
-      "grad_norm": 0.00047464677481912076,
-      "learning_rate": 0.00019999972637336545,
-      "loss": 46.0,
-      "step": 4634
-    },
-    {
-      "epoch": 0.7464068601795564,
-      "grad_norm": 0.0032051538582891226,
-      "learning_rate": 0.00019999972625500207,
-      "loss": 46.0,
-      "step": 4635
-    },
-    {
-      "epoch": 0.7465678972583437,
-      "grad_norm": 0.00036359333898872137,
-      "learning_rate": 0.0001999997261366131,
-      "loss": 46.0,
-      "step": 4636
-    },
-    {
-      "epoch": 0.7467289343371312,
-      "grad_norm": 0.0029360989574342966,
-      "learning_rate": 0.00019999972601819857,
-      "loss": 46.0,
-      "step": 4637
-    },
-    {
-      "epoch": 0.7468899714159185,
-      "grad_norm": 0.0008089731563813984,
-      "learning_rate": 0.00019999972589975843,
-      "loss": 46.0,
-      "step": 4638
-    },
-    {
-      "epoch": 0.747051008494706,
-      "grad_norm": 0.001520228455774486,
-      "learning_rate": 0.00019999972578129267,
-      "loss": 46.0,
-      "step": 4639
-    },
-    {
-      "epoch": 0.7472120455734933,
-      "grad_norm": 0.0003285162674728781,
-      "learning_rate": 0.00019999972566280133,
-      "loss": 46.0,
-      "step": 4640
-    },
-    {
-      "epoch": 0.7473730826522806,
-      "grad_norm": 0.005414806772023439,
-      "learning_rate": 0.0001999997255442844,
-      "loss": 46.0,
-      "step": 4641
-    },
-    {
-      "epoch": 0.7475341197310681,
-      "grad_norm": 0.00030580884777009487,
-      "learning_rate": 0.00019999972542574188,
-      "loss": 46.0,
-      "step": 4642
-    },
-    {
-      "epoch": 0.7476951568098554,
-      "grad_norm": 0.0019031764240935445,
-      "learning_rate": 0.00019999972530717374,
-      "loss": 46.0,
-      "step": 4643
-    },
-    {
-      "epoch": 0.7478561938886429,
-      "grad_norm": 0.00041826689266599715,
-      "learning_rate": 0.00019999972518858002,
-      "loss": 46.0,
-      "step": 4644
-    },
-    {
-      "epoch": 0.7480172309674302,
-      "grad_norm": 0.00042645016219466925,
-      "learning_rate": 0.00019999972506996072,
-      "loss": 46.0,
-      "step": 4645
-    },
-    {
-      "epoch": 0.7481782680462177,
-      "grad_norm": 0.0012782086851075292,
-      "learning_rate": 0.00019999972495131582,
-      "loss": 46.0,
-      "step": 4646
-    },
-    {
-      "epoch": 0.748339305125005,
-      "grad_norm": 0.0008948395261541009,
-      "learning_rate": 0.00019999972483264531,
-      "loss": 46.0,
-      "step": 4647
-    },
-    {
-      "epoch": 0.7485003422037925,
-      "grad_norm": 0.0014121216954663396,
-      "learning_rate": 0.00019999972471394922,
-      "loss": 46.0,
-      "step": 4648
-    },
-    {
-      "epoch": 0.7486613792825798,
-      "grad_norm": 0.0008100031409412622,
-      "learning_rate": 0.0001999997245952275,
-      "loss": 46.0,
-      "step": 4649
-    },
-    {
-      "epoch": 0.7488224163613673,
-      "grad_norm": 0.0008760725613683462,
-      "learning_rate": 0.00019999972447648024,
-      "loss": 46.0,
-      "step": 4650
-    },
-    {
-      "epoch": 0.7489834534401546,
-      "grad_norm": 0.000714444147888571,
-      "learning_rate": 0.00019999972435770738,
-      "loss": 46.0,
-      "step": 4651
-    },
-    {
-      "epoch": 0.7491444905189419,
-      "grad_norm": 0.000542790221516043,
-      "learning_rate": 0.00019999972423890888,
-      "loss": 46.0,
-      "step": 4652
-    },
-    {
-      "epoch": 0.7493055275977294,
-      "grad_norm": 0.0007445058436132967,
-      "learning_rate": 0.0001999997241200848,
-      "loss": 46.0,
-      "step": 4653
-    },
-    {
-      "epoch": 0.7494665646765167,
-      "grad_norm": 0.0006571363192051649,
-      "learning_rate": 0.00019999972400123515,
-      "loss": 46.0,
-      "step": 4654
-    },
-    {
-      "epoch": 0.7496276017553042,
-      "grad_norm": 0.0005258733872324228,
-      "learning_rate": 0.0001999997238823599,
-      "loss": 46.0,
-      "step": 4655
-    },
-    {
-      "epoch": 0.7497886388340915,
-      "grad_norm": 0.0020241141319274902,
-      "learning_rate": 0.00019999972376345904,
-      "loss": 46.0,
-      "step": 4656
-    },
-    {
-      "epoch": 0.749949675912879,
-      "grad_norm": 0.001398624968715012,
-      "learning_rate": 0.00019999972364453256,
-      "loss": 46.0,
-      "step": 4657
-    },
-    {
-      "epoch": 0.7501107129916663,
-      "grad_norm": 0.0015605477383360267,
-      "learning_rate": 0.00019999972352558053,
-      "loss": 46.0,
-      "step": 4658
-    },
-    {
-      "epoch": 0.7502717500704538,
-      "grad_norm": 0.0008498406386934221,
-      "learning_rate": 0.0001999997234066029,
-      "loss": 46.0,
-      "step": 4659
-    },
-    {
-      "epoch": 0.7504327871492411,
-      "grad_norm": 0.0027355796191841364,
-      "learning_rate": 0.00019999972328759965,
-      "loss": 46.0,
-      "step": 4660
-    },
-    {
-      "epoch": 0.7505938242280285,
-      "grad_norm": 0.00039430696051567793,
-      "learning_rate": 0.00019999972316857086,
-      "loss": 46.0,
-      "step": 4661
-    },
-    {
-      "epoch": 0.7507548613068159,
-      "grad_norm": 0.0004359922604635358,
-      "learning_rate": 0.0001999997230495164,
-      "loss": 46.0,
-      "step": 4662
-    },
-    {
-      "epoch": 0.7509158983856032,
-      "grad_norm": 0.0014977410901337862,
-      "learning_rate": 0.0001999997229304364,
-      "loss": 46.0,
-      "step": 4663
-    },
-    {
-      "epoch": 0.7510769354643907,
-      "grad_norm": 0.008042525500059128,
-      "learning_rate": 0.0001999997228113308,
-      "loss": 46.0,
-      "step": 4664
-    },
-    {
-      "epoch": 0.751237972543178,
-      "grad_norm": 0.000974364927969873,
-      "learning_rate": 0.0001999997226921996,
-      "loss": 46.0,
-      "step": 4665
-    },
-    {
-      "epoch": 0.7513990096219655,
-      "grad_norm": 0.0005854789051227272,
-      "learning_rate": 0.0001999997225730428,
-      "loss": 46.0,
-      "step": 4666
-    },
-    {
-      "epoch": 0.7515600467007528,
-      "grad_norm": 0.0028308595065027475,
-      "learning_rate": 0.00019999972245386037,
-      "loss": 46.0,
-      "step": 4667
-    },
-    {
-      "epoch": 0.7517210837795403,
-      "grad_norm": 0.004498977214097977,
-      "learning_rate": 0.00019999972233465237,
-      "loss": 46.0,
-      "step": 4668
-    },
-    {
-      "epoch": 0.7518821208583276,
-      "grad_norm": 0.0010365810012444854,
-      "learning_rate": 0.0001999997222154188,
-      "loss": 46.0,
-      "step": 4669
-    },
-    {
-      "epoch": 0.7520431579371151,
-      "grad_norm": 0.002472150372341275,
-      "learning_rate": 0.00019999972209615963,
-      "loss": 46.0,
-      "step": 4670
-    },
-    {
-      "epoch": 0.7522041950159024,
-      "grad_norm": 0.0008548650075681508,
-      "learning_rate": 0.00019999972197687482,
-      "loss": 46.0,
-      "step": 4671
-    },
-    {
-      "epoch": 0.7523652320946898,
-      "grad_norm": 0.002625685650855303,
-      "learning_rate": 0.00019999972185756447,
-      "loss": 46.0,
-      "step": 4672
-    },
-    {
-      "epoch": 0.7525262691734772,
-      "grad_norm": 0.00300764967687428,
-      "learning_rate": 0.0001999997217382285,
-      "loss": 46.0,
-      "step": 4673
-    },
-    {
-      "epoch": 0.7526873062522645,
-      "grad_norm": 0.0010463440557941794,
-      "learning_rate": 0.00019999972161886692,
-      "loss": 46.0,
-      "step": 4674
-    },
-    {
-      "epoch": 0.752848343331052,
-      "grad_norm": 0.0005545472959056497,
-      "learning_rate": 0.00019999972149947978,
-      "loss": 46.0,
-      "step": 4675
-    },
-    {
-      "epoch": 0.7530093804098393,
-      "grad_norm": 0.004115791991353035,
-      "learning_rate": 0.00019999972138006703,
-      "loss": 46.0,
-      "step": 4676
-    },
-    {
-      "epoch": 0.7531704174886268,
-      "grad_norm": 0.001644573058001697,
-      "learning_rate": 0.0001999997212606287,
-      "loss": 46.0,
-      "step": 4677
-    },
-    {
-      "epoch": 0.7533314545674141,
-      "grad_norm": 0.0012830966152250767,
-      "learning_rate": 0.00019999972114116474,
-      "loss": 46.0,
-      "step": 4678
-    },
-    {
-      "epoch": 0.7534924916462016,
-      "grad_norm": 0.0034075479488819838,
-      "learning_rate": 0.00019999972102167522,
-      "loss": 46.0,
-      "step": 4679
-    },
-    {
-      "epoch": 0.7536535287249889,
-      "grad_norm": 0.001458398182876408,
-      "learning_rate": 0.0001999997209021601,
-      "loss": 46.0,
-      "step": 4680
-    },
-    {
-      "epoch": 0.7538145658037764,
-      "grad_norm": 0.0004869127878919244,
-      "learning_rate": 0.00019999972078261937,
-      "loss": 46.0,
-      "step": 4681
-    },
-    {
-      "epoch": 0.7539756028825637,
-      "grad_norm": 0.003625623881816864,
-      "learning_rate": 0.00019999972066305304,
-      "loss": 46.0,
-      "step": 4682
-    },
-    {
-      "epoch": 0.754136639961351,
-      "grad_norm": 0.0018892440712079406,
-      "learning_rate": 0.00019999972054346113,
-      "loss": 46.0,
-      "step": 4683
-    },
-    {
-      "epoch": 0.7542976770401385,
-      "grad_norm": 0.004187189042568207,
-      "learning_rate": 0.0001999997204238436,
-      "loss": 46.0,
-      "step": 4684
-    },
-    {
-      "epoch": 0.7544587141189258,
-      "grad_norm": 0.001565721700899303,
-      "learning_rate": 0.0001999997203042005,
-      "loss": 46.0,
-      "step": 4685
-    },
-    {
-      "epoch": 0.7546197511977133,
-      "grad_norm": 0.0005464204004965723,
-      "learning_rate": 0.00019999972018453182,
-      "loss": 46.0,
-      "step": 4686
-    },
-    {
-      "epoch": 0.7547807882765006,
-      "grad_norm": 0.004974219016730785,
-      "learning_rate": 0.0001999997200648375,
-      "loss": 46.0,
-      "step": 4687
-    },
-    {
-      "epoch": 0.7549418253552881,
-      "grad_norm": 0.0008043889538384974,
-      "learning_rate": 0.00019999971994511762,
-      "loss": 46.0,
-      "step": 4688
-    },
-    {
-      "epoch": 0.7551028624340754,
-      "grad_norm": 0.0011184316826984286,
-      "learning_rate": 0.00019999971982537215,
-      "loss": 46.0,
-      "step": 4689
-    },
-    {
-      "epoch": 0.7552638995128629,
-      "grad_norm": 0.0008623666362836957,
-      "learning_rate": 0.00019999971970560107,
-      "loss": 46.0,
-      "step": 4690
-    },
-    {
-      "epoch": 0.7554249365916502,
-      "grad_norm": 0.0012672202428802848,
-      "learning_rate": 0.0001999997195858044,
-      "loss": 46.0,
-      "step": 4691
-    },
-    {
-      "epoch": 0.7555859736704377,
-      "grad_norm": 0.0011607082560658455,
-      "learning_rate": 0.00019999971946598212,
-      "loss": 46.0,
-      "step": 4692
-    },
-    {
-      "epoch": 0.755747010749225,
-      "grad_norm": 0.0004718489362858236,
-      "learning_rate": 0.00019999971934613428,
-      "loss": 46.0,
-      "step": 4693
-    },
-    {
-      "epoch": 0.7559080478280124,
-      "grad_norm": 0.0014752136776223779,
-      "learning_rate": 0.00019999971922626082,
-      "loss": 46.0,
-      "step": 4694
-    },
-    {
-      "epoch": 0.7560690849067998,
-      "grad_norm": 0.0024576419964432716,
-      "learning_rate": 0.00019999971910636174,
-      "loss": 46.0,
-      "step": 4695
-    },
-    {
-      "epoch": 0.7562301219855871,
-      "grad_norm": 0.0015845508314669132,
-      "learning_rate": 0.0001999997189864371,
-      "loss": 46.0,
-      "step": 4696
-    },
-    {
-      "epoch": 0.7563911590643746,
-      "grad_norm": 0.0010749782668426633,
-      "learning_rate": 0.00019999971886648687,
-      "loss": 46.0,
-      "step": 4697
-    },
-    {
-      "epoch": 0.7565521961431619,
-      "grad_norm": 0.0005168747738935053,
-      "learning_rate": 0.00019999971874651103,
-      "loss": 46.0,
-      "step": 4698
-    },
-    {
-      "epoch": 0.7567132332219494,
-      "grad_norm": 0.0013822518521919847,
-      "learning_rate": 0.00019999971862650958,
-      "loss": 46.0,
-      "step": 4699
-    },
-    {
-      "epoch": 0.7568742703007367,
-      "grad_norm": 0.0003715934290084988,
-      "learning_rate": 0.00019999971850648258,
-      "loss": 46.0,
-      "step": 4700
-    },
-    {
-      "epoch": 0.7570353073795242,
-      "grad_norm": 0.00046432638191618025,
-      "learning_rate": 0.00019999971838642995,
-      "loss": 46.0,
-      "step": 4701
-    },
-    {
-      "epoch": 0.7571963444583115,
-      "grad_norm": 0.0005687046214006841,
-      "learning_rate": 0.00019999971826635174,
-      "loss": 46.0,
-      "step": 4702
-    },
-    {
-      "epoch": 0.7573573815370989,
-      "grad_norm": 0.0012473512906581163,
-      "learning_rate": 0.00019999971814624795,
-      "loss": 46.0,
-      "step": 4703
-    },
-    {
-      "epoch": 0.7575184186158863,
-      "grad_norm": 0.0014611040242016315,
-      "learning_rate": 0.0001999997180261185,
-      "loss": 46.0,
-      "step": 4704
-    },
-    {
-      "epoch": 0.7576794556946737,
-      "grad_norm": 0.0023170479107648134,
-      "learning_rate": 0.0001999997179059635,
-      "loss": 46.0,
-      "step": 4705
-    },
-    {
-      "epoch": 0.7578404927734611,
-      "grad_norm": 0.0008970486233010888,
-      "learning_rate": 0.00019999971778578293,
-      "loss": 46.0,
-      "step": 4706
-    },
-    {
-      "epoch": 0.7580015298522484,
-      "grad_norm": 0.0011431307066231966,
-      "learning_rate": 0.00019999971766557675,
-      "loss": 46.0,
-      "step": 4707
-    },
-    {
-      "epoch": 0.7581625669310359,
-      "grad_norm": 0.0024071645457297564,
-      "learning_rate": 0.00019999971754534497,
-      "loss": 46.0,
-      "step": 4708
-    },
-    {
-      "epoch": 0.7583236040098232,
-      "grad_norm": 0.0013848600210621953,
-      "learning_rate": 0.00019999971742508757,
-      "loss": 46.0,
-      "step": 4709
-    },
-    {
-      "epoch": 0.7584846410886107,
-      "grad_norm": 0.000974328548181802,
-      "learning_rate": 0.0001999997173048046,
-      "loss": 46.0,
-      "step": 4710
-    },
-    {
-      "epoch": 0.758645678167398,
-      "grad_norm": 0.0009354418143630028,
-      "learning_rate": 0.00019999971718449603,
-      "loss": 46.0,
-      "step": 4711
-    },
-    {
-      "epoch": 0.7588067152461855,
-      "grad_norm": 0.0008783229277469218,
-      "learning_rate": 0.00019999971706416187,
-      "loss": 46.0,
-      "step": 4712
-    },
-    {
-      "epoch": 0.7589677523249728,
-      "grad_norm": 0.0020138919353485107,
-      "learning_rate": 0.00019999971694380212,
-      "loss": 46.0,
-      "step": 4713
-    },
-    {
-      "epoch": 0.7591287894037602,
-      "grad_norm": 0.001696997438557446,
-      "learning_rate": 0.00019999971682341675,
-      "loss": 46.0,
-      "step": 4714
-    },
-    {
-      "epoch": 0.7592898264825476,
-      "grad_norm": 0.001785373780876398,
-      "learning_rate": 0.0001999997167030058,
-      "loss": 46.0,
-      "step": 4715
-    },
-    {
-      "epoch": 0.759450863561335,
-      "grad_norm": 0.001679008244536817,
-      "learning_rate": 0.0001999997165825693,
-      "loss": 46.0,
-      "step": 4716
-    },
-    {
-      "epoch": 0.7596119006401224,
-      "grad_norm": 0.0025074633304029703,
-      "learning_rate": 0.00019999971646210714,
-      "loss": 46.0,
-      "step": 4717
-    },
-    {
-      "epoch": 0.7597729377189097,
-      "grad_norm": 0.0010412584524601698,
-      "learning_rate": 0.0001999997163416194,
-      "loss": 46.0,
-      "step": 4718
-    },
-    {
-      "epoch": 0.7599339747976972,
-      "grad_norm": 0.00029251494561322033,
-      "learning_rate": 0.00019999971622110607,
-      "loss": 46.0,
-      "step": 4719
-    },
-    {
-      "epoch": 0.7600950118764845,
-      "grad_norm": 0.001483130268752575,
-      "learning_rate": 0.00019999971610056716,
-      "loss": 46.0,
-      "step": 4720
-    },
-    {
-      "epoch": 0.760256048955272,
-      "grad_norm": 0.0016655530780553818,
-      "learning_rate": 0.00019999971598000263,
-      "loss": 46.0,
-      "step": 4721
-    },
-    {
-      "epoch": 0.7604170860340593,
-      "grad_norm": 0.0029624358285218477,
-      "learning_rate": 0.00019999971585941254,
-      "loss": 46.0,
-      "step": 4722
-    },
-    {
-      "epoch": 0.7605781231128468,
-      "grad_norm": 0.0006819451809860766,
-      "learning_rate": 0.0001999997157387968,
-      "loss": 46.0,
-      "step": 4723
-    },
-    {
-      "epoch": 0.7607391601916341,
-      "grad_norm": 0.0012939812149852514,
-      "learning_rate": 0.00019999971561815552,
-      "loss": 46.0,
-      "step": 4724
-    },
-    {
-      "epoch": 0.7609001972704215,
-      "grad_norm": 0.0005127053591422737,
-      "learning_rate": 0.00019999971549748864,
-      "loss": 46.0,
-      "step": 4725
-    },
-    {
-      "epoch": 0.7610612343492089,
-      "grad_norm": 0.0016179886879399419,
-      "learning_rate": 0.00019999971537679612,
-      "loss": 46.0,
-      "step": 4726
-    },
-    {
-      "epoch": 0.7612222714279963,
-      "grad_norm": 0.0036728864070028067,
-      "learning_rate": 0.00019999971525607805,
-      "loss": 46.0,
-      "step": 4727
-    },
-    {
-      "epoch": 0.7613833085067837,
-      "grad_norm": 0.0023441568482667208,
-      "learning_rate": 0.00019999971513533438,
-      "loss": 46.0,
-      "step": 4728
-    },
-    {
-      "epoch": 0.761544345585571,
-      "grad_norm": 0.00042874025530181825,
-      "learning_rate": 0.0001999997150145651,
-      "loss": 46.0,
-      "step": 4729
-    },
-    {
-      "epoch": 0.7617053826643585,
-      "grad_norm": 0.0007069238345138729,
-      "learning_rate": 0.00019999971489377023,
-      "loss": 46.0,
-      "step": 4730
-    },
-    {
-      "epoch": 0.7618664197431458,
-      "grad_norm": 0.0003816660610027611,
-      "learning_rate": 0.00019999971477294975,
-      "loss": 46.0,
-      "step": 4731
-    },
-    {
-      "epoch": 0.7620274568219333,
-      "grad_norm": 0.0009648569393903017,
-      "learning_rate": 0.0001999997146521037,
-      "loss": 46.0,
-      "step": 4732
-    },
-    {
-      "epoch": 0.7621884939007206,
-      "grad_norm": 0.0006991874543018639,
-      "learning_rate": 0.00019999971453123205,
-      "loss": 46.0,
-      "step": 4733
-    },
-    {
-      "epoch": 0.7623495309795081,
-      "grad_norm": 0.0007834224379621446,
-      "learning_rate": 0.0001999997144103348,
-      "loss": 46.0,
-      "step": 4734
-    },
-    {
-      "epoch": 0.7625105680582954,
-      "grad_norm": 0.0010299012064933777,
-      "learning_rate": 0.00019999971428941195,
-      "loss": 46.0,
-      "step": 4735
-    },
-    {
-      "epoch": 0.7626716051370828,
-      "grad_norm": 0.0006626648246310651,
-      "learning_rate": 0.00019999971416846354,
-      "loss": 46.0,
-      "step": 4736
-    },
-    {
-      "epoch": 0.7628326422158702,
-      "grad_norm": 0.0015821504639461637,
-      "learning_rate": 0.0001999997140474895,
-      "loss": 46.0,
-      "step": 4737
-    },
-    {
-      "epoch": 0.7629936792946576,
-      "grad_norm": 0.0056922342628240585,
-      "learning_rate": 0.00019999971392648983,
-      "loss": 46.0,
-      "step": 4738
-    },
-    {
-      "epoch": 0.763154716373445,
-      "grad_norm": 0.0005951990024186671,
-      "learning_rate": 0.00019999971380546463,
-      "loss": 46.0,
-      "step": 4739
-    },
-    {
-      "epoch": 0.7633157534522323,
-      "grad_norm": 0.0007348938379436731,
-      "learning_rate": 0.0001999997136844138,
-      "loss": 46.0,
-      "step": 4740
-    },
-    {
-      "epoch": 0.7634767905310198,
-      "grad_norm": 0.000635460251942277,
-      "learning_rate": 0.0001999997135633374,
-      "loss": 46.0,
-      "step": 4741
-    },
-    {
-      "epoch": 0.7636378276098071,
-      "grad_norm": 0.0010311596561223269,
-      "learning_rate": 0.00019999971344223538,
-      "loss": 46.0,
-      "step": 4742
-    },
-    {
-      "epoch": 0.7637988646885946,
-      "grad_norm": 0.0004321200249250978,
-      "learning_rate": 0.0001999997133211078,
-      "loss": 46.0,
-      "step": 4743
-    },
-    {
-      "epoch": 0.7639599017673819,
-      "grad_norm": 0.001043844036757946,
-      "learning_rate": 0.0001999997131999546,
-      "loss": 46.0,
-      "step": 4744
-    },
-    {
-      "epoch": 0.7641209388461694,
-      "grad_norm": 0.00042273811413906515,
-      "learning_rate": 0.00019999971307877582,
-      "loss": 46.0,
-      "step": 4745
-    },
-    {
-      "epoch": 0.7642819759249567,
-      "grad_norm": 0.0007260072161443532,
-      "learning_rate": 0.00019999971295757142,
-      "loss": 46.0,
-      "step": 4746
-    },
-    {
-      "epoch": 0.7644430130037441,
-      "grad_norm": 0.0029898914508521557,
-      "learning_rate": 0.00019999971283634143,
-      "loss": 46.0,
-      "step": 4747
-    },
-    {
-      "epoch": 0.7646040500825315,
-      "grad_norm": 0.0030045099556446075,
-      "learning_rate": 0.00019999971271508586,
-      "loss": 46.0,
-      "step": 4748
-    },
-    {
-      "epoch": 0.7647650871613189,
-      "grad_norm": 0.00179480598308146,
-      "learning_rate": 0.00019999971259380468,
-      "loss": 46.0,
-      "step": 4749
-    },
-    {
-      "epoch": 0.7649261242401063,
-      "grad_norm": 0.0046968418173491955,
-      "learning_rate": 0.00019999971247249793,
-      "loss": 46.0,
-      "step": 4750
-    },
-    {
-      "epoch": 0.7650871613188936,
-      "grad_norm": 0.0012994071003049612,
-      "learning_rate": 0.00019999971235116557,
-      "loss": 46.0,
-      "step": 4751
-    },
-    {
-      "epoch": 0.7652481983976811,
-      "grad_norm": 0.0007074291352182627,
-      "learning_rate": 0.0001999997122298076,
-      "loss": 46.0,
-      "step": 4752
-    },
-    {
-      "epoch": 0.7654092354764684,
-      "grad_norm": 0.0038666988257318735,
-      "learning_rate": 0.00019999971210842406,
-      "loss": 46.0,
-      "step": 4753
-    },
-    {
-      "epoch": 0.7655702725552559,
-      "grad_norm": 0.0013941572979092598,
-      "learning_rate": 0.0001999997119870149,
-      "loss": 46.0,
-      "step": 4754
-    },
-    {
-      "epoch": 0.7657313096340432,
-      "grad_norm": 0.0005347354453988373,
-      "learning_rate": 0.00019999971186558018,
-      "loss": 46.0,
-      "step": 4755
-    },
-    {
-      "epoch": 0.7658923467128306,
-      "grad_norm": 0.0004978392389602959,
-      "learning_rate": 0.00019999971174411983,
-      "loss": 46.0,
-      "step": 4756
-    },
-    {
-      "epoch": 0.766053383791618,
-      "grad_norm": 0.0011792602017521858,
-      "learning_rate": 0.00019999971162263391,
-      "loss": 46.0,
-      "step": 4757
-    },
-    {
-      "epoch": 0.7662144208704054,
-      "grad_norm": 0.0006544819916598499,
-      "learning_rate": 0.0001999997115011224,
-      "loss": 46.0,
-      "step": 4758
-    },
-    {
-      "epoch": 0.7663754579491928,
-      "grad_norm": 0.0014281094772741199,
-      "learning_rate": 0.00019999971137958525,
-      "loss": 46.0,
-      "step": 4759
-    },
-    {
-      "epoch": 0.7665364950279802,
-      "grad_norm": 0.0017949396278709173,
-      "learning_rate": 0.00019999971125802255,
-      "loss": 46.0,
-      "step": 4760
-    },
-    {
-      "epoch": 0.7666975321067676,
-      "grad_norm": 0.00041266894550062716,
-      "learning_rate": 0.00019999971113643424,
-      "loss": 46.0,
-      "step": 4761
-    },
-    {
-      "epoch": 0.766858569185555,
-      "grad_norm": 0.0004257208784110844,
-      "learning_rate": 0.00019999971101482034,
-      "loss": 46.0,
-      "step": 4762
-    },
-    {
-      "epoch": 0.7670196062643424,
-      "grad_norm": 0.0015154016437008977,
-      "learning_rate": 0.00019999971089318085,
-      "loss": 46.0,
-      "step": 4763
-    },
-    {
-      "epoch": 0.7671806433431297,
-      "grad_norm": 0.0005804282845929265,
-      "learning_rate": 0.00019999971077151575,
-      "loss": 46.0,
-      "step": 4764
-    },
-    {
-      "epoch": 0.7673416804219172,
-      "grad_norm": 0.0006973495474085212,
-      "learning_rate": 0.00019999971064982509,
-      "loss": 46.0,
-      "step": 4765
-    },
-    {
-      "epoch": 0.7675027175007045,
-      "grad_norm": 0.0024062858428806067,
-      "learning_rate": 0.00019999971052810878,
-      "loss": 46.0,
-      "step": 4766
-    },
-    {
-      "epoch": 0.7676637545794919,
-      "grad_norm": 0.0005186183843761683,
-      "learning_rate": 0.0001999997104063669,
-      "loss": 46.0,
-      "step": 4767
-    },
-    {
-      "epoch": 0.7678247916582793,
-      "grad_norm": 0.0024316725321114063,
-      "learning_rate": 0.00019999971028459944,
-      "loss": 46.0,
-      "step": 4768
-    },
-    {
-      "epoch": 0.7679858287370667,
-      "grad_norm": 0.0034432136453688145,
-      "learning_rate": 0.00019999971016280638,
-      "loss": 46.0,
-      "step": 4769
-    },
-    {
-      "epoch": 0.7681468658158541,
-      "grad_norm": 0.0011938678799197078,
-      "learning_rate": 0.0001999997100409877,
-      "loss": 46.0,
-      "step": 4770
-    },
-    {
-      "epoch": 0.7683079028946415,
-      "grad_norm": 0.0003913584805559367,
-      "learning_rate": 0.00019999970991914343,
-      "loss": 46.0,
-      "step": 4771
-    },
-    {
-      "epoch": 0.7684689399734289,
-      "grad_norm": 0.0005460192332975566,
-      "learning_rate": 0.00019999970979727358,
-      "loss": 46.0,
-      "step": 4772
-    },
-    {
-      "epoch": 0.7686299770522163,
-      "grad_norm": 0.000703003432136029,
-      "learning_rate": 0.00019999970967537816,
-      "loss": 46.0,
-      "step": 4773
-    },
-    {
-      "epoch": 0.7687910141310037,
-      "grad_norm": 0.00302697136066854,
-      "learning_rate": 0.0001999997095534571,
-      "loss": 46.0,
-      "step": 4774
-    },
-    {
-      "epoch": 0.768952051209791,
-      "grad_norm": 0.00032928737346082926,
-      "learning_rate": 0.00019999970943151047,
-      "loss": 46.0,
-      "step": 4775
-    },
-    {
-      "epoch": 0.7691130882885785,
-      "grad_norm": 0.001323476666584611,
-      "learning_rate": 0.00019999970930953824,
-      "loss": 46.0,
-      "step": 4776
-    },
-    {
-      "epoch": 0.7692741253673658,
-      "grad_norm": 0.0007389159873127937,
-      "learning_rate": 0.00019999970918754042,
-      "loss": 46.0,
-      "step": 4777
-    },
-    {
-      "epoch": 0.7694351624461532,
-      "grad_norm": 0.0032452831510454416,
-      "learning_rate": 0.000199999709065517,
-      "loss": 46.0,
-      "step": 4778
-    },
-    {
-      "epoch": 0.7695961995249406,
-      "grad_norm": 0.000530002755112946,
-      "learning_rate": 0.00019999970894346797,
-      "loss": 46.0,
-      "step": 4779
-    },
-    {
-      "epoch": 0.769757236603728,
-      "grad_norm": 0.0011758236214518547,
-      "learning_rate": 0.00019999970882139337,
-      "loss": 46.0,
-      "step": 4780
-    },
-    {
-      "epoch": 0.7699182736825154,
-      "grad_norm": 0.0005026006838306785,
-      "learning_rate": 0.00019999970869929317,
-      "loss": 46.0,
-      "step": 4781
-    },
-    {
-      "epoch": 0.7700793107613028,
-      "grad_norm": 0.0009113785345107317,
-      "learning_rate": 0.00019999970857716737,
-      "loss": 46.0,
-      "step": 4782
-    },
-    {
-      "epoch": 0.7702403478400902,
-      "grad_norm": 0.0008467392181046307,
-      "learning_rate": 0.000199999708455016,
-      "loss": 46.0,
-      "step": 4783
-    },
-    {
-      "epoch": 0.7704013849188776,
-      "grad_norm": 0.00046924632624723017,
-      "learning_rate": 0.000199999708332839,
-      "loss": 46.0,
-      "step": 4784
-    },
-    {
-      "epoch": 0.770562421997665,
-      "grad_norm": 0.0009634041343815625,
-      "learning_rate": 0.0001999997082106364,
-      "loss": 46.0,
-      "step": 4785
-    },
-    {
-      "epoch": 0.7707234590764523,
-      "grad_norm": 0.004751934669911861,
-      "learning_rate": 0.00019999970808840821,
-      "loss": 46.0,
-      "step": 4786
-    },
-    {
-      "epoch": 0.7708844961552398,
-      "grad_norm": 0.0010547551792114973,
-      "learning_rate": 0.00019999970796615447,
-      "loss": 46.0,
-      "step": 4787
-    },
-    {
-      "epoch": 0.7710455332340271,
-      "grad_norm": 0.0007109145517461002,
-      "learning_rate": 0.00019999970784387506,
-      "loss": 46.0,
-      "step": 4788
-    },
-    {
-      "epoch": 0.7712065703128145,
-      "grad_norm": 0.0013763969764113426,
-      "learning_rate": 0.00019999970772157012,
-      "loss": 46.0,
-      "step": 4789
-    },
-    {
-      "epoch": 0.7713676073916019,
-      "grad_norm": 0.0005206355708651245,
-      "learning_rate": 0.00019999970759923956,
-      "loss": 46.0,
-      "step": 4790
-    },
-    {
-      "epoch": 0.7715286444703893,
-      "grad_norm": 0.0009797917446121573,
-      "learning_rate": 0.0001999997074768834,
-      "loss": 46.0,
-      "step": 4791
-    },
-    {
-      "epoch": 0.7716896815491767,
-      "grad_norm": 0.0005716772866435349,
-      "learning_rate": 0.00019999970735450165,
-      "loss": 46.0,
-      "step": 4792
-    },
-    {
-      "epoch": 0.7718507186279641,
-      "grad_norm": 0.0004491658473853022,
-      "learning_rate": 0.0001999997072320943,
-      "loss": 46.0,
-      "step": 4793
-    },
-    {
-      "epoch": 0.7720117557067515,
-      "grad_norm": 0.000580527528654784,
-      "learning_rate": 0.00019999970710966137,
-      "loss": 46.0,
-      "step": 4794
-    },
-    {
-      "epoch": 0.7721727927855389,
-      "grad_norm": 0.0009397721150889993,
-      "learning_rate": 0.00019999970698720288,
-      "loss": 46.0,
-      "step": 4795
-    },
-    {
-      "epoch": 0.7723338298643263,
-      "grad_norm": 0.0032832517754286528,
-      "learning_rate": 0.00019999970686471872,
-      "loss": 46.0,
-      "step": 4796
-    },
-    {
-      "epoch": 0.7724948669431136,
-      "grad_norm": 0.0010393010452389717,
-      "learning_rate": 0.00019999970674220902,
-      "loss": 46.0,
-      "step": 4797
-    },
-    {
-      "epoch": 0.7726559040219011,
-      "grad_norm": 0.001162507920525968,
-      "learning_rate": 0.0001999997066196737,
-      "loss": 46.0,
-      "step": 4798
-    },
-    {
-      "epoch": 0.7728169411006884,
-      "grad_norm": 0.0024015961680561304,
-      "learning_rate": 0.00019999970649711276,
-      "loss": 46.0,
-      "step": 4799
-    },
-    {
-      "epoch": 0.7729779781794758,
-      "grad_norm": 0.000556223327293992,
-      "learning_rate": 0.00019999970637452628,
-      "loss": 46.0,
-      "step": 4800
-    },
-    {
-      "epoch": 0.7731390152582632,
-      "grad_norm": 0.0019684734288603067,
-      "learning_rate": 0.00019999970625191418,
-      "loss": 46.0,
-      "step": 4801
-    },
-    {
-      "epoch": 0.7733000523370506,
-      "grad_norm": 0.0017118286341428757,
-      "learning_rate": 0.00019999970612927646,
-      "loss": 46.0,
-      "step": 4802
-    },
-    {
-      "epoch": 0.773461089415838,
-      "grad_norm": 0.0032506200950592756,
-      "learning_rate": 0.0001999997060066132,
-      "loss": 46.0,
-      "step": 4803
-    },
-    {
-      "epoch": 0.7736221264946254,
-      "grad_norm": 0.0006733370828442276,
-      "learning_rate": 0.0001999997058839243,
-      "loss": 46.0,
-      "step": 4804
-    },
-    {
-      "epoch": 0.7737831635734128,
-      "grad_norm": 0.0020027526188641787,
-      "learning_rate": 0.00019999970576120983,
-      "loss": 46.0,
-      "step": 4805
-    },
-    {
-      "epoch": 0.7739442006522002,
-      "grad_norm": 0.00076530990190804,
-      "learning_rate": 0.00019999970563846974,
-      "loss": 46.0,
-      "step": 4806
-    },
-    {
-      "epoch": 0.7741052377309876,
-      "grad_norm": 0.0016813938273116946,
-      "learning_rate": 0.0001999997055157041,
-      "loss": 46.0,
-      "step": 4807
-    },
-    {
-      "epoch": 0.774266274809775,
-      "grad_norm": 0.002859602915123105,
-      "learning_rate": 0.00019999970539291283,
-      "loss": 46.0,
-      "step": 4808
-    },
-    {
-      "epoch": 0.7744273118885623,
-      "grad_norm": 0.0013228071620687842,
-      "learning_rate": 0.00019999970527009598,
-      "loss": 46.0,
-      "step": 4809
-    },
-    {
-      "epoch": 0.7745883489673497,
-      "grad_norm": 0.0003916521091014147,
-      "learning_rate": 0.00019999970514725352,
-      "loss": 46.0,
-      "step": 4810
-    },
-    {
-      "epoch": 0.7747493860461371,
-      "grad_norm": 0.0007609453168697655,
-      "learning_rate": 0.00019999970502438546,
-      "loss": 46.0,
-      "step": 4811
-    },
-    {
-      "epoch": 0.7749104231249245,
-      "grad_norm": 0.002951554022729397,
-      "learning_rate": 0.00019999970490149183,
-      "loss": 46.0,
-      "step": 4812
-    },
-    {
-      "epoch": 0.7750714602037119,
-      "grad_norm": 0.0015105771599337459,
-      "learning_rate": 0.00019999970477857257,
-      "loss": 46.0,
-      "step": 4813
-    },
-    {
-      "epoch": 0.7752324972824993,
-      "grad_norm": 0.00033626353251747787,
-      "learning_rate": 0.00019999970465562776,
-      "loss": 46.0,
-      "step": 4814
-    },
-    {
-      "epoch": 0.7753935343612867,
-      "grad_norm": 0.00082097650738433,
-      "learning_rate": 0.00019999970453265733,
-      "loss": 46.0,
-      "step": 4815
-    },
-    {
-      "epoch": 0.7755545714400741,
-      "grad_norm": 0.0006431712536141276,
-      "learning_rate": 0.00019999970440966132,
-      "loss": 46.0,
-      "step": 4816
-    },
-    {
-      "epoch": 0.7757156085188615,
-      "grad_norm": 0.0008429883164353669,
-      "learning_rate": 0.0001999997042866397,
-      "loss": 46.0,
-      "step": 4817
-    },
-    {
-      "epoch": 0.7758766455976489,
-      "grad_norm": 0.0008026235736906528,
-      "learning_rate": 0.00019999970416359247,
-      "loss": 46.0,
-      "step": 4818
-    },
-    {
-      "epoch": 0.7760376826764362,
-      "grad_norm": 0.0010946061229333282,
-      "learning_rate": 0.00019999970404051967,
-      "loss": 46.0,
-      "step": 4819
-    },
-    {
-      "epoch": 0.7761987197552236,
-      "grad_norm": 0.0012417907128110528,
-      "learning_rate": 0.00019999970391742125,
-      "loss": 46.0,
-      "step": 4820
-    },
-    {
-      "epoch": 0.776359756834011,
-      "grad_norm": 0.003090334590524435,
-      "learning_rate": 0.00019999970379429728,
-      "loss": 46.0,
-      "step": 4821
-    },
-    {
-      "epoch": 0.7765207939127984,
-      "grad_norm": 0.0007975895423442125,
-      "learning_rate": 0.00019999970367114766,
-      "loss": 46.0,
-      "step": 4822
-    },
-    {
-      "epoch": 0.7766818309915858,
-      "grad_norm": 0.00222119246609509,
-      "learning_rate": 0.0001999997035479725,
-      "loss": 46.0,
-      "step": 4823
-    },
-    {
-      "epoch": 0.7768428680703732,
-      "grad_norm": 0.000647918728645891,
-      "learning_rate": 0.0001999997034247717,
-      "loss": 46.0,
-      "step": 4824
-    },
-    {
-      "epoch": 0.7770039051491606,
-      "grad_norm": 0.003519634949043393,
-      "learning_rate": 0.00019999970330154533,
-      "loss": 46.0,
-      "step": 4825
-    },
-    {
-      "epoch": 0.777164942227948,
-      "grad_norm": 0.0021936146076768637,
-      "learning_rate": 0.00019999970317829336,
-      "loss": 46.0,
-      "step": 4826
-    },
-    {
-      "epoch": 0.7773259793067354,
-      "grad_norm": 0.0012823280412703753,
-      "learning_rate": 0.0001999997030550158,
-      "loss": 46.0,
-      "step": 4827
-    },
-    {
-      "epoch": 0.7774870163855228,
-      "grad_norm": 0.0012383607681840658,
-      "learning_rate": 0.00019999970293171264,
-      "loss": 46.0,
-      "step": 4828
-    },
-    {
-      "epoch": 0.7776480534643102,
-      "grad_norm": 0.0011226102942600846,
-      "learning_rate": 0.00019999970280838389,
-      "loss": 46.0,
-      "step": 4829
-    },
-    {
-      "epoch": 0.7778090905430975,
-      "grad_norm": 0.001930180354975164,
-      "learning_rate": 0.00019999970268502954,
-      "loss": 46.0,
-      "step": 4830
-    },
-    {
-      "epoch": 0.7779701276218849,
-      "grad_norm": 0.0007477269391529262,
-      "learning_rate": 0.00019999970256164958,
-      "loss": 46.0,
-      "step": 4831
-    },
-    {
-      "epoch": 0.7781311647006723,
-      "grad_norm": 0.0005857248906977475,
-      "learning_rate": 0.00019999970243824407,
-      "loss": 46.0,
-      "step": 4832
-    },
-    {
-      "epoch": 0.7782922017794597,
-      "grad_norm": 0.0009798255050554872,
-      "learning_rate": 0.00019999970231481293,
-      "loss": 46.0,
-      "step": 4833
-    },
-    {
-      "epoch": 0.7784532388582471,
-      "grad_norm": 0.00033370786695741117,
-      "learning_rate": 0.0001999997021913562,
-      "loss": 46.0,
-      "step": 4834
-    },
-    {
-      "epoch": 0.7786142759370345,
-      "grad_norm": 0.00040524365613237023,
-      "learning_rate": 0.00019999970206787388,
-      "loss": 46.0,
-      "step": 4835
-    },
-    {
-      "epoch": 0.7787753130158219,
-      "grad_norm": 0.0015656567411497235,
-      "learning_rate": 0.00019999970194436596,
-      "loss": 46.0,
-      "step": 4836
-    },
-    {
-      "epoch": 0.7789363500946093,
-      "grad_norm": 0.0004574234480969608,
-      "learning_rate": 0.00019999970182083243,
-      "loss": 46.0,
-      "step": 4837
-    },
-    {
-      "epoch": 0.7790973871733967,
-      "grad_norm": 0.0009385333396494389,
-      "learning_rate": 0.00019999970169727333,
-      "loss": 46.0,
-      "step": 4838
-    },
-    {
-      "epoch": 0.7792584242521841,
-      "grad_norm": 0.00122968596406281,
-      "learning_rate": 0.00019999970157368862,
-      "loss": 46.0,
-      "step": 4839
-    },
-    {
-      "epoch": 0.7794194613309715,
-      "grad_norm": 0.00411304971203208,
-      "learning_rate": 0.00019999970145007833,
-      "loss": 46.0,
-      "step": 4840
-    },
-    {
-      "epoch": 0.7795804984097588,
-      "grad_norm": 0.0006165204686112702,
-      "learning_rate": 0.00019999970132644244,
-      "loss": 46.0,
-      "step": 4841
-    },
-    {
-      "epoch": 0.7797415354885462,
-      "grad_norm": 0.0006161246565170586,
-      "learning_rate": 0.00019999970120278095,
-      "loss": 46.0,
-      "step": 4842
-    },
-    {
-      "epoch": 0.7799025725673336,
-      "grad_norm": 0.0030055155511945486,
-      "learning_rate": 0.00019999970107909386,
-      "loss": 46.0,
-      "step": 4843
-    },
-    {
-      "epoch": 0.780063609646121,
-      "grad_norm": 0.0006349720642901957,
-      "learning_rate": 0.00019999970095538122,
-      "loss": 46.0,
-      "step": 4844
-    },
-    {
-      "epoch": 0.7802246467249084,
-      "grad_norm": 0.0028327363543212414,
-      "learning_rate": 0.00019999970083164293,
-      "loss": 46.0,
-      "step": 4845
-    },
-    {
-      "epoch": 0.7803856838036958,
-      "grad_norm": 0.0011354926973581314,
-      "learning_rate": 0.00019999970070787906,
-      "loss": 46.0,
-      "step": 4846
-    },
-    {
-      "epoch": 0.7805467208824832,
-      "grad_norm": 0.0043993135914206505,
-      "learning_rate": 0.0001999997005840896,
-      "loss": 46.0,
-      "step": 4847
-    },
-    {
-      "epoch": 0.7807077579612706,
-      "grad_norm": 0.0015721928793936968,
-      "learning_rate": 0.00019999970046027455,
-      "loss": 46.0,
-      "step": 4848
-    },
-    {
-      "epoch": 0.780868795040058,
-      "grad_norm": 0.0005242746556177735,
-      "learning_rate": 0.00019999970033643389,
-      "loss": 46.0,
-      "step": 4849
-    },
-    {
-      "epoch": 0.7810298321188454,
-      "grad_norm": 0.0011350411223247647,
-      "learning_rate": 0.00019999970021256766,
-      "loss": 46.0,
-      "step": 4850
-    },
-    {
-      "epoch": 0.7811908691976328,
-      "grad_norm": 0.0019493073923513293,
-      "learning_rate": 0.0001999997000886758,
-      "loss": 46.0,
-      "step": 4851
-    },
-    {
-      "epoch": 0.7813519062764201,
-      "grad_norm": 0.0007735356921330094,
-      "learning_rate": 0.00019999969996475838,
-      "loss": 46.0,
-      "step": 4852
-    },
-    {
-      "epoch": 0.7815129433552075,
-      "grad_norm": 0.0006262017996050417,
-      "learning_rate": 0.00019999969984081534,
-      "loss": 46.0,
-      "step": 4853
-    },
-    {
-      "epoch": 0.7816739804339949,
-      "grad_norm": 0.0014904163544997573,
-      "learning_rate": 0.00019999969971684674,
-      "loss": 46.0,
-      "step": 4854
-    },
-    {
-      "epoch": 0.7818350175127823,
-      "grad_norm": 0.0012866721954196692,
-      "learning_rate": 0.0001999996995928525,
-      "loss": 46.0,
-      "step": 4855
-    },
-    {
-      "epoch": 0.7819960545915697,
-      "grad_norm": 0.0006702210521325469,
-      "learning_rate": 0.00019999969946883267,
-      "loss": 46.0,
-      "step": 4856
-    },
-    {
-      "epoch": 0.7821570916703571,
-      "grad_norm": 0.0007764595211483538,
-      "learning_rate": 0.0001999996993447873,
-      "loss": 46.0,
-      "step": 4857
-    },
-    {
-      "epoch": 0.7823181287491445,
-      "grad_norm": 0.0002539651177357882,
-      "learning_rate": 0.00019999969922071626,
-      "loss": 46.0,
-      "step": 4858
-    },
-    {
-      "epoch": 0.7824791658279319,
-      "grad_norm": 0.0004988660803064704,
-      "learning_rate": 0.00019999969909661967,
-      "loss": 46.0,
-      "step": 4859
-    },
-    {
-      "epoch": 0.7826402029067193,
-      "grad_norm": 0.001338389003649354,
-      "learning_rate": 0.0001999996989724975,
-      "loss": 46.0,
-      "step": 4860
-    },
-    {
-      "epoch": 0.7828012399855067,
-      "grad_norm": 0.0012628274271264672,
-      "learning_rate": 0.0001999996988483497,
-      "loss": 46.0,
-      "step": 4861
-    },
-    {
-      "epoch": 0.782962277064294,
-      "grad_norm": 0.0005042063421569765,
-      "learning_rate": 0.00019999969872417633,
-      "loss": 46.0,
-      "step": 4862
-    },
-    {
-      "epoch": 0.7831233141430815,
-      "grad_norm": 0.004220644477754831,
-      "learning_rate": 0.00019999969859997734,
-      "loss": 46.0,
-      "step": 4863
-    },
-    {
-      "epoch": 0.7832843512218688,
-      "grad_norm": 0.00038851742283441126,
-      "learning_rate": 0.00019999969847575276,
-      "loss": 46.0,
-      "step": 4864
-    },
-    {
-      "epoch": 0.7834453883006562,
-      "grad_norm": 0.0008721654303371906,
-      "learning_rate": 0.0001999996983515026,
-      "loss": 46.0,
-      "step": 4865
-    },
-    {
-      "epoch": 0.7836064253794436,
-      "grad_norm": 0.002580776112154126,
-      "learning_rate": 0.00019999969822722684,
-      "loss": 46.0,
-      "step": 4866
-    },
-    {
-      "epoch": 0.783767462458231,
-      "grad_norm": 0.0004379845631774515,
-      "learning_rate": 0.00019999969810292548,
-      "loss": 46.0,
-      "step": 4867
-    },
-    {
-      "epoch": 0.7839284995370184,
-      "grad_norm": 0.0007844308856874704,
-      "learning_rate": 0.00019999969797859852,
-      "loss": 46.0,
-      "step": 4868
-    },
-    {
-      "epoch": 0.7840895366158058,
-      "grad_norm": 0.0023445901460945606,
-      "learning_rate": 0.00019999969785424598,
-      "loss": 46.0,
-      "step": 4869
-    },
-    {
-      "epoch": 0.7842505736945932,
-      "grad_norm": 0.001878255046904087,
-      "learning_rate": 0.00019999969772986785,
-      "loss": 46.0,
-      "step": 4870
-    },
-    {
-      "epoch": 0.7844116107733806,
-      "grad_norm": 0.0017349470872431993,
-      "learning_rate": 0.0001999996976054641,
-      "loss": 46.0,
-      "step": 4871
-    },
-    {
-      "epoch": 0.784572647852168,
-      "grad_norm": 0.0031744639854878187,
-      "learning_rate": 0.00019999969748103478,
-      "loss": 46.0,
-      "step": 4872
-    },
-    {
-      "epoch": 0.7847336849309553,
-      "grad_norm": 0.0037809982895851135,
-      "learning_rate": 0.00019999969735657986,
-      "loss": 46.0,
-      "step": 4873
-    },
-    {
-      "epoch": 0.7848947220097428,
-      "grad_norm": 0.0003046912606805563,
-      "learning_rate": 0.00019999969723209936,
-      "loss": 46.0,
-      "step": 4874
-    },
-    {
-      "epoch": 0.7850557590885301,
-      "grad_norm": 0.0007008084212429821,
-      "learning_rate": 0.00019999969710759322,
-      "loss": 46.0,
-      "step": 4875
-    },
-    {
-      "epoch": 0.7852167961673175,
-      "grad_norm": 0.0029918367508798838,
-      "learning_rate": 0.0001999996969830615,
-      "loss": 46.0,
-      "step": 4876
-    },
-    {
-      "epoch": 0.7853778332461049,
-      "grad_norm": 0.00042898106039501727,
-      "learning_rate": 0.0001999996968585042,
-      "loss": 46.0,
-      "step": 4877
-    },
-    {
-      "epoch": 0.7855388703248923,
-      "grad_norm": 0.0013238437240943313,
-      "learning_rate": 0.0001999996967339213,
-      "loss": 46.0,
-      "step": 4878
-    },
-    {
-      "epoch": 0.7856999074036797,
-      "grad_norm": 0.0015143059426918626,
-      "learning_rate": 0.00019999969660931282,
-      "loss": 46.0,
-      "step": 4879
-    },
-    {
-      "epoch": 0.7858609444824671,
-      "grad_norm": 0.004708718508481979,
-      "learning_rate": 0.00019999969648467874,
-      "loss": 46.0,
-      "step": 4880
-    },
-    {
-      "epoch": 0.7860219815612545,
-      "grad_norm": 0.0009256972698494792,
-      "learning_rate": 0.00019999969636001904,
-      "loss": 46.0,
-      "step": 4881
-    },
-    {
-      "epoch": 0.7861830186400419,
-      "grad_norm": 0.0019786139018833637,
-      "learning_rate": 0.00019999969623533376,
-      "loss": 46.0,
-      "step": 4882
-    },
-    {
-      "epoch": 0.7863440557188293,
-      "grad_norm": 0.0009576291777193546,
-      "learning_rate": 0.0001999996961106229,
-      "loss": 46.0,
-      "step": 4883
-    },
-    {
-      "epoch": 0.7865050927976166,
-      "grad_norm": 0.0004916126490570605,
-      "learning_rate": 0.0001999996959858864,
-      "loss": 46.0,
-      "step": 4884
-    },
-    {
-      "epoch": 0.786666129876404,
-      "grad_norm": 0.002730055944994092,
-      "learning_rate": 0.00019999969586112436,
-      "loss": 46.0,
-      "step": 4885
-    },
-    {
-      "epoch": 0.7868271669551914,
-      "grad_norm": 0.0011625067563727498,
-      "learning_rate": 0.0001999996957363367,
-      "loss": 46.0,
-      "step": 4886
-    },
-    {
-      "epoch": 0.7869882040339788,
-      "grad_norm": 0.002465092809870839,
-      "learning_rate": 0.00019999969561152345,
-      "loss": 46.0,
-      "step": 4887
-    },
-    {
-      "epoch": 0.7871492411127662,
-      "grad_norm": 0.0007205911679193377,
-      "learning_rate": 0.0001999996954866846,
-      "loss": 46.0,
-      "step": 4888
-    },
-    {
-      "epoch": 0.7873102781915536,
-      "grad_norm": 0.0003511893155518919,
-      "learning_rate": 0.00019999969536182017,
-      "loss": 46.0,
-      "step": 4889
-    },
-    {
-      "epoch": 0.787471315270341,
-      "grad_norm": 0.000942239654250443,
-      "learning_rate": 0.0001999996952369301,
-      "loss": 46.0,
-      "step": 4890
-    },
-    {
-      "epoch": 0.7876323523491284,
-      "grad_norm": 0.00026638482813723385,
-      "learning_rate": 0.0001999996951120145,
-      "loss": 46.0,
-      "step": 4891
-    },
-    {
-      "epoch": 0.7877933894279158,
-      "grad_norm": 0.0027480493299663067,
-      "learning_rate": 0.00019999969498707328,
-      "loss": 46.0,
-      "step": 4892
-    },
-    {
-      "epoch": 0.7879544265067032,
-      "grad_norm": 0.0018945022020488977,
-      "learning_rate": 0.00019999969486210643,
-      "loss": 46.0,
-      "step": 4893
-    },
-    {
-      "epoch": 0.7881154635854906,
-      "grad_norm": 0.001131606986746192,
-      "learning_rate": 0.00019999969473711402,
-      "loss": 46.0,
-      "step": 4894
-    },
-    {
-      "epoch": 0.7882765006642779,
-      "grad_norm": 0.0037077227607369423,
-      "learning_rate": 0.00019999969461209603,
-      "loss": 46.0,
-      "step": 4895
-    },
-    {
-      "epoch": 0.7884375377430654,
-      "grad_norm": 0.0014807177940383554,
-      "learning_rate": 0.00019999969448705241,
-      "loss": 46.0,
-      "step": 4896
-    },
-    {
-      "epoch": 0.7885985748218527,
-      "grad_norm": 0.0004393410054035485,
-      "learning_rate": 0.00019999969436198322,
-      "loss": 46.0,
-      "step": 4897
-    },
-    {
-      "epoch": 0.7887596119006401,
-      "grad_norm": 0.0011072284542024136,
-      "learning_rate": 0.00019999969423688843,
-      "loss": 46.0,
-      "step": 4898
-    },
-    {
-      "epoch": 0.7889206489794275,
-      "grad_norm": 0.001103946240618825,
-      "learning_rate": 0.00019999969411176803,
-      "loss": 46.0,
-      "step": 4899
-    },
-    {
-      "epoch": 0.7890816860582149,
-      "grad_norm": 0.0005947441677562892,
-      "learning_rate": 0.00019999969398662204,
-      "loss": 46.0,
-      "step": 4900
-    },
-    {
-      "epoch": 0.7892427231370023,
-      "grad_norm": 0.0017201130976900458,
-      "learning_rate": 0.00019999969386145047,
-      "loss": 46.0,
-      "step": 4901
-    },
-    {
-      "epoch": 0.7894037602157897,
-      "grad_norm": 0.0026060270611196756,
-      "learning_rate": 0.00019999969373625328,
-      "loss": 46.0,
-      "step": 4902
-    },
-    {
-      "epoch": 0.7895647972945771,
-      "grad_norm": 0.0015082237077876925,
-      "learning_rate": 0.00019999969361103053,
-      "loss": 46.0,
-      "step": 4903
-    },
-    {
-      "epoch": 0.7897258343733645,
-      "grad_norm": 0.0013800321612507105,
-      "learning_rate": 0.00019999969348578217,
-      "loss": 46.0,
-      "step": 4904
-    },
-    {
-      "epoch": 0.7898868714521519,
-      "grad_norm": 0.001854727161116898,
-      "learning_rate": 0.00019999969336050822,
-      "loss": 46.0,
-      "step": 4905
-    },
-    {
-      "epoch": 0.7900479085309392,
-      "grad_norm": 0.0003750800678972155,
-      "learning_rate": 0.00019999969323520865,
-      "loss": 46.0,
-      "step": 4906
-    },
-    {
-      "epoch": 0.7902089456097267,
-      "grad_norm": 0.0006828468176536262,
-      "learning_rate": 0.0001999996931098835,
-      "loss": 46.0,
-      "step": 4907
-    },
-    {
-      "epoch": 0.790369982688514,
-      "grad_norm": 0.004550055600702763,
-      "learning_rate": 0.00019999969298453273,
-      "loss": 46.0,
-      "step": 4908
-    },
-    {
-      "epoch": 0.7905310197673014,
-      "grad_norm": 0.0003666391712613404,
-      "learning_rate": 0.0001999996928591564,
-      "loss": 46.0,
-      "step": 4909
-    },
-    {
-      "epoch": 0.7906920568460888,
-      "grad_norm": 0.00104003946762532,
-      "learning_rate": 0.0001999996927337545,
-      "loss": 46.0,
-      "step": 4910
-    },
-    {
-      "epoch": 0.7908530939248762,
-      "grad_norm": 0.0006893830723129213,
-      "learning_rate": 0.00019999969260832697,
-      "loss": 46.0,
-      "step": 4911
-    },
-    {
-      "epoch": 0.7910141310036636,
-      "grad_norm": 0.0005217760917730629,
-      "learning_rate": 0.00019999969248287385,
-      "loss": 46.0,
-      "step": 4912
-    },
-    {
-      "epoch": 0.791175168082451,
-      "grad_norm": 0.004187182988971472,
-      "learning_rate": 0.0001999996923573951,
-      "loss": 46.0,
-      "step": 4913
-    },
-    {
-      "epoch": 0.7913362051612384,
-      "grad_norm": 0.00039599742740392685,
-      "learning_rate": 0.00019999969223189078,
-      "loss": 46.0,
-      "step": 4914
-    },
-    {
-      "epoch": 0.7914972422400257,
-      "grad_norm": 0.002221059752628207,
-      "learning_rate": 0.0001999996921063609,
-      "loss": 46.0,
-      "step": 4915
-    },
-    {
-      "epoch": 0.7916582793188132,
-      "grad_norm": 0.0005677093286067247,
-      "learning_rate": 0.0001999996919808054,
-      "loss": 46.0,
-      "step": 4916
-    },
-    {
-      "epoch": 0.7918193163976005,
-      "grad_norm": 0.002351069124415517,
-      "learning_rate": 0.0001999996918552243,
-      "loss": 46.0,
-      "step": 4917
-    },
-    {
-      "epoch": 0.791980353476388,
-      "grad_norm": 0.0014926479198038578,
-      "learning_rate": 0.00019999969172961762,
-      "loss": 46.0,
-      "step": 4918
-    },
-    {
-      "epoch": 0.7921413905551753,
-      "grad_norm": 0.0008763189543969929,
-      "learning_rate": 0.00019999969160398534,
-      "loss": 46.0,
-      "step": 4919
-    },
-    {
-      "epoch": 0.7923024276339627,
-      "grad_norm": 0.0007234427612274885,
-      "learning_rate": 0.00019999969147832744,
-      "loss": 46.0,
-      "step": 4920
-    },
-    {
-      "epoch": 0.7924634647127501,
-      "grad_norm": 0.0005524869775399566,
-      "learning_rate": 0.000199999691352644,
-      "loss": 46.0,
-      "step": 4921
-    },
-    {
-      "epoch": 0.7926245017915375,
-      "grad_norm": 0.003294931724667549,
-      "learning_rate": 0.0001999996912269349,
-      "loss": 46.0,
-      "step": 4922
-    },
-    {
-      "epoch": 0.7927855388703249,
-      "grad_norm": 0.0017873761244118214,
-      "learning_rate": 0.00019999969110120021,
-      "loss": 46.0,
-      "step": 4923
-    },
-    {
-      "epoch": 0.7929465759491123,
-      "grad_norm": 0.001016058144159615,
-      "learning_rate": 0.00019999969097543997,
-      "loss": 46.0,
-      "step": 4924
-    },
-    {
-      "epoch": 0.7931076130278997,
-      "grad_norm": 0.00024809749447740614,
-      "learning_rate": 0.00019999969084965412,
-      "loss": 46.0,
-      "step": 4925
-    },
-    {
-      "epoch": 0.793268650106687,
-      "grad_norm": 0.0007221294799819589,
-      "learning_rate": 0.00019999969072384265,
-      "loss": 46.0,
-      "step": 4926
-    },
-    {
-      "epoch": 0.7934296871854745,
-      "grad_norm": 0.004974100738763809,
-      "learning_rate": 0.00019999969059800564,
-      "loss": 46.0,
-      "step": 4927
-    },
-    {
-      "epoch": 0.7935907242642618,
-      "grad_norm": 0.0026517470832914114,
-      "learning_rate": 0.00019999969047214297,
-      "loss": 46.0,
-      "step": 4928
-    },
-    {
-      "epoch": 0.7937517613430493,
-      "grad_norm": 0.0012876553228124976,
-      "learning_rate": 0.00019999969034625474,
-      "loss": 46.0,
-      "step": 4929
-    },
-    {
-      "epoch": 0.7939127984218366,
-      "grad_norm": 0.0008008077275007963,
-      "learning_rate": 0.00019999969022034092,
-      "loss": 46.0,
-      "step": 4930
-    },
-    {
-      "epoch": 0.794073835500624,
-      "grad_norm": 0.0033534260001033545,
-      "learning_rate": 0.00019999969009440148,
-      "loss": 46.0,
-      "step": 4931
-    },
-    {
-      "epoch": 0.7942348725794114,
-      "grad_norm": 0.0005811078590340912,
-      "learning_rate": 0.00019999968996843646,
-      "loss": 46.0,
-      "step": 4932
-    },
-    {
-      "epoch": 0.7943959096581988,
-      "grad_norm": 0.0007451678975485265,
-      "learning_rate": 0.00019999968984244588,
-      "loss": 46.0,
-      "step": 4933
-    },
-    {
-      "epoch": 0.7945569467369862,
-      "grad_norm": 0.0032919771037995815,
-      "learning_rate": 0.00019999968971642966,
-      "loss": 46.0,
-      "step": 4934
-    },
-    {
-      "epoch": 0.7947179838157736,
-      "grad_norm": 0.0005804127431474626,
-      "learning_rate": 0.00019999968959038785,
-      "loss": 46.0,
-      "step": 4935
-    },
-    {
-      "epoch": 0.794879020894561,
-      "grad_norm": 0.00047154416097328067,
-      "learning_rate": 0.00019999968946432045,
-      "loss": 46.0,
-      "step": 4936
-    },
-    {
-      "epoch": 0.7950400579733483,
-      "grad_norm": 0.0025603254325687885,
-      "learning_rate": 0.00019999968933822747,
-      "loss": 46.0,
-      "step": 4937
-    },
-    {
-      "epoch": 0.7952010950521358,
-      "grad_norm": 0.0012943173060193658,
-      "learning_rate": 0.0001999996892121089,
-      "loss": 46.0,
-      "step": 4938
-    },
-    {
-      "epoch": 0.7953621321309231,
-      "grad_norm": 0.0006102478946559131,
-      "learning_rate": 0.0001999996890859647,
-      "loss": 46.0,
-      "step": 4939
-    },
-    {
-      "epoch": 0.7955231692097106,
-      "grad_norm": 0.0005684745847247541,
-      "learning_rate": 0.00019999968895979492,
-      "loss": 46.0,
-      "step": 4940
-    },
-    {
-      "epoch": 0.7956842062884979,
-      "grad_norm": 0.002167138736695051,
-      "learning_rate": 0.00019999968883359956,
-      "loss": 46.0,
-      "step": 4941
-    },
-    {
-      "epoch": 0.7958452433672853,
-      "grad_norm": 0.0008825754630379379,
-      "learning_rate": 0.0001999996887073786,
-      "loss": 46.0,
-      "step": 4942
-    },
-    {
-      "epoch": 0.7960062804460727,
-      "grad_norm": 0.0007869398104958236,
-      "learning_rate": 0.00019999968858113202,
-      "loss": 46.0,
-      "step": 4943
-    },
-    {
-      "epoch": 0.7961673175248601,
-      "grad_norm": 0.00489926477894187,
-      "learning_rate": 0.00019999968845485987,
-      "loss": 46.0,
-      "step": 4944
-    },
-    {
-      "epoch": 0.7963283546036475,
-      "grad_norm": 0.0006775857764296234,
-      "learning_rate": 0.00019999968832856214,
-      "loss": 46.0,
-      "step": 4945
-    },
-    {
-      "epoch": 0.7964893916824349,
-      "grad_norm": 0.0008033980848267674,
-      "learning_rate": 0.0001999996882022388,
-      "loss": 46.0,
-      "step": 4946
-    },
-    {
-      "epoch": 0.7966504287612223,
-      "grad_norm": 0.0008058302919380367,
-      "learning_rate": 0.00019999968807588985,
-      "loss": 46.0,
-      "step": 4947
-    },
-    {
-      "epoch": 0.7968114658400096,
-      "grad_norm": 0.00040172855369746685,
-      "learning_rate": 0.0001999996879495153,
-      "loss": 46.0,
-      "step": 4948
-    },
-    {
-      "epoch": 0.7969725029187971,
-      "grad_norm": 0.0006174162263050675,
-      "learning_rate": 0.0001999996878231152,
-      "loss": 46.0,
-      "step": 4949
-    },
-    {
-      "epoch": 0.7971335399975844,
-      "grad_norm": 0.0011549360351637006,
-      "learning_rate": 0.00019999968769668946,
-      "loss": 46.0,
-      "step": 4950
-    },
-    {
-      "epoch": 0.7972945770763719,
-      "grad_norm": 0.0007918311166577041,
-      "learning_rate": 0.00019999968757023812,
-      "loss": 46.0,
-      "step": 4951
-    },
-    {
-      "epoch": 0.7974556141551592,
-      "grad_norm": 0.0003489319933578372,
-      "learning_rate": 0.0001999996874437612,
-      "loss": 46.0,
-      "step": 4952
-    },
-    {
-      "epoch": 0.7976166512339466,
-      "grad_norm": 0.000532814534381032,
-      "learning_rate": 0.0001999996873172587,
-      "loss": 46.0,
-      "step": 4953
-    },
-    {
-      "epoch": 0.797777688312734,
-      "grad_norm": 0.0018908587517216802,
-      "learning_rate": 0.0001999996871907306,
-      "loss": 46.0,
-      "step": 4954
-    },
-    {
-      "epoch": 0.7979387253915214,
-      "grad_norm": 0.002615349367260933,
-      "learning_rate": 0.00019999968706417692,
-      "loss": 46.0,
-      "step": 4955
-    },
-    {
-      "epoch": 0.7980997624703088,
-      "grad_norm": 0.0007406870718114078,
-      "learning_rate": 0.00019999968693759762,
-      "loss": 46.0,
-      "step": 4956
-    },
-    {
-      "epoch": 0.7982607995490962,
-      "grad_norm": 0.0006270628655329347,
-      "learning_rate": 0.00019999968681099276,
-      "loss": 46.0,
-      "step": 4957
-    },
-    {
-      "epoch": 0.7984218366278836,
-      "grad_norm": 0.0002609911607578397,
-      "learning_rate": 0.00019999968668436225,
-      "loss": 46.0,
-      "step": 4958
-    },
-    {
-      "epoch": 0.7985828737066709,
-      "grad_norm": 0.0011190030490979552,
-      "learning_rate": 0.00019999968655770616,
-      "loss": 46.0,
-      "step": 4959
-    },
-    {
-      "epoch": 0.7987439107854584,
-      "grad_norm": 0.0007208367460407317,
-      "learning_rate": 0.0001999996864310245,
-      "loss": 46.0,
-      "step": 4960
-    },
-    {
-      "epoch": 0.7989049478642457,
-      "grad_norm": 0.0008196748676709831,
-      "learning_rate": 0.00019999968630431724,
-      "loss": 46.0,
-      "step": 4961
-    },
-    {
-      "epoch": 0.7990659849430332,
-      "grad_norm": 0.0007227154565043747,
-      "learning_rate": 0.00019999968617758436,
-      "loss": 46.0,
-      "step": 4962
-    },
-    {
-      "epoch": 0.7992270220218205,
-      "grad_norm": 0.0007980055524967611,
-      "learning_rate": 0.00019999968605082592,
-      "loss": 46.0,
-      "step": 4963
-    },
-    {
-      "epoch": 0.799388059100608,
-      "grad_norm": 0.0005577775300480425,
-      "learning_rate": 0.00019999968592404187,
-      "loss": 46.0,
-      "step": 4964
-    },
-    {
-      "epoch": 0.7995490961793953,
-      "grad_norm": 0.0010210535256192088,
-      "learning_rate": 0.0001999996857972322,
-      "loss": 46.0,
-      "step": 4965
-    },
-    {
-      "epoch": 0.7997101332581827,
-      "grad_norm": 0.00203165621496737,
-      "learning_rate": 0.00019999968567039697,
-      "loss": 46.0,
-      "step": 4966
-    },
-    {
-      "epoch": 0.7998711703369701,
-      "grad_norm": 0.0004046688845846802,
-      "learning_rate": 0.00019999968554353613,
-      "loss": 46.0,
-      "step": 4967
-    },
-    {
-      "epoch": 0.8000322074157574,
-      "grad_norm": 0.002894398057833314,
-      "learning_rate": 0.0001999996854166497,
-      "loss": 46.0,
-      "step": 4968
-    },
-    {
-      "epoch": 0.8001932444945449,
-      "grad_norm": 0.0013177957152947783,
-      "learning_rate": 0.00019999968528973768,
-      "loss": 46.0,
-      "step": 4969
-    },
-    {
-      "epoch": 0.8003542815733322,
-      "grad_norm": 0.0026697125285863876,
-      "learning_rate": 0.00019999968516280005,
-      "loss": 46.0,
-      "step": 4970
-    },
-    {
-      "epoch": 0.8005153186521197,
-      "grad_norm": 0.0017101905541494489,
-      "learning_rate": 0.00019999968503583686,
-      "loss": 46.0,
-      "step": 4971
-    },
-    {
-      "epoch": 0.800676355730907,
-      "grad_norm": 0.0006968029774725437,
-      "learning_rate": 0.00019999968490884803,
-      "loss": 46.0,
-      "step": 4972
-    },
-    {
-      "epoch": 0.8008373928096945,
-      "grad_norm": 0.0038087673019617796,
-      "learning_rate": 0.00019999968478183363,
-      "loss": 46.0,
-      "step": 4973
-    },
-    {
-      "epoch": 0.8009984298884818,
-      "grad_norm": 0.0013463938375934958,
-      "learning_rate": 0.00019999968465479363,
-      "loss": 46.0,
-      "step": 4974
-    },
-    {
-      "epoch": 0.8011594669672693,
-      "grad_norm": 0.00038374605355784297,
-      "learning_rate": 0.00019999968452772806,
-      "loss": 46.0,
-      "step": 4975
-    },
-    {
-      "epoch": 0.8013205040460566,
-      "grad_norm": 0.003065139288082719,
-      "learning_rate": 0.00019999968440063685,
-      "loss": 46.0,
-      "step": 4976
-    },
-    {
-      "epoch": 0.801481541124844,
-      "grad_norm": 0.00507735600695014,
-      "learning_rate": 0.00019999968427352008,
-      "loss": 46.0,
-      "step": 4977
-    },
-    {
-      "epoch": 0.8016425782036314,
-      "grad_norm": 0.0011967107420787215,
-      "learning_rate": 0.00019999968414637767,
-      "loss": 46.0,
-      "step": 4978
-    },
-    {
-      "epoch": 0.8018036152824187,
-      "grad_norm": 0.0011453555198386312,
-      "learning_rate": 0.00019999968401920973,
-      "loss": 46.0,
-      "step": 4979
-    },
-    {
-      "epoch": 0.8019646523612062,
-      "grad_norm": 0.0003181898209732026,
-      "learning_rate": 0.00019999968389201615,
-      "loss": 46.0,
-      "step": 4980
-    },
-    {
-      "epoch": 0.8021256894399935,
-      "grad_norm": 0.0033435707446187735,
-      "learning_rate": 0.000199999683764797,
-      "loss": 46.0,
-      "step": 4981
-    },
-    {
-      "epoch": 0.802286726518781,
-      "grad_norm": 0.0011169268982484937,
-      "learning_rate": 0.00019999968363755224,
-      "loss": 46.0,
-      "step": 4982
-    },
-    {
-      "epoch": 0.8024477635975683,
-      "grad_norm": 0.0012636369792744517,
-      "learning_rate": 0.0001999996835102819,
-      "loss": 46.0,
-      "step": 4983
-    },
-    {
-      "epoch": 0.8026088006763558,
-      "grad_norm": 0.0007384041091427207,
-      "learning_rate": 0.00019999968338298594,
-      "loss": 46.0,
-      "step": 4984
-    },
-    {
-      "epoch": 0.8027698377551431,
-      "grad_norm": 0.0012772405752912164,
-      "learning_rate": 0.00019999968325566442,
-      "loss": 46.0,
-      "step": 4985
-    },
-    {
-      "epoch": 0.8029308748339306,
-      "grad_norm": 0.0005906779551878572,
-      "learning_rate": 0.00019999968312831725,
-      "loss": 46.0,
-      "step": 4986
-    },
-    {
-      "epoch": 0.8030919119127179,
-      "grad_norm": 0.000568014511372894,
-      "learning_rate": 0.0001999996830009445,
-      "loss": 46.0,
-      "step": 4987
-    },
-    {
-      "epoch": 0.8032529489915053,
-      "grad_norm": 0.00076268520206213,
-      "learning_rate": 0.0001999996828735462,
-      "loss": 46.0,
-      "step": 4988
-    },
-    {
-      "epoch": 0.8034139860702927,
-      "grad_norm": 0.004757401067763567,
-      "learning_rate": 0.00019999968274612227,
-      "loss": 46.0,
-      "step": 4989
-    },
-    {
-      "epoch": 0.80357502314908,
-      "grad_norm": 0.0024922143202275038,
-      "learning_rate": 0.00019999968261867276,
-      "loss": 46.0,
-      "step": 4990
-    },
-    {
-      "epoch": 0.8037360602278675,
-      "grad_norm": 0.003020449075847864,
-      "learning_rate": 0.00019999968249119764,
-      "loss": 46.0,
-      "step": 4991
-    },
-    {
-      "epoch": 0.8038970973066548,
-      "grad_norm": 0.0004710409848485142,
-      "learning_rate": 0.00019999968236369695,
-      "loss": 46.0,
-      "step": 4992
-    },
-    {
-      "epoch": 0.8040581343854423,
-      "grad_norm": 0.0016607147408649325,
-      "learning_rate": 0.00019999968223617065,
-      "loss": 46.0,
-      "step": 4993
-    },
-    {
-      "epoch": 0.8042191714642296,
-      "grad_norm": 0.0007125571719370782,
-      "learning_rate": 0.00019999968210861874,
-      "loss": 46.0,
-      "step": 4994
-    },
-    {
-      "epoch": 0.8043802085430171,
-      "grad_norm": 0.0012517754221335053,
-      "learning_rate": 0.00019999968198104127,
-      "loss": 46.0,
-      "step": 4995
-    },
-    {
-      "epoch": 0.8045412456218044,
-      "grad_norm": 0.0008217499125748873,
-      "learning_rate": 0.00019999968185343818,
-      "loss": 46.0,
-      "step": 4996
-    },
-    {
-      "epoch": 0.8047022827005919,
-      "grad_norm": 0.00045296180178411305,
-      "learning_rate": 0.0001999996817258095,
-      "loss": 46.0,
-      "step": 4997
-    },
-    {
-      "epoch": 0.8048633197793792,
-      "grad_norm": 0.00023589274496771395,
-      "learning_rate": 0.00019999968159815524,
-      "loss": 46.0,
-      "step": 4998
-    },
-    {
-      "epoch": 0.8050243568581666,
-      "grad_norm": 0.0025104512460529804,
-      "learning_rate": 0.00019999968147047537,
-      "loss": 46.0,
-      "step": 4999
-    },
-    {
-      "epoch": 0.805185393936954,
-      "grad_norm": 0.0012525818310678005,
-      "learning_rate": 0.0001999996813427699,
-      "loss": 46.0,
-      "step": 5000
-    },
-    {
-      "epoch": 0.8053464310157413,
-      "grad_norm": 0.0007998804212547839,
-      "learning_rate": 0.00019999968121503883,
-      "loss": 46.0,
-      "step": 5001
-    },
-    {
-      "epoch": 0.8055074680945288,
-      "grad_norm": 0.007886652834713459,
-      "learning_rate": 0.0001999996810872822,
-      "loss": 46.0,
-      "step": 5002
-    },
-    {
-      "epoch": 0.8056685051733161,
-      "grad_norm": 0.0035920701920986176,
-      "learning_rate": 0.00019999968095949994,
-      "loss": 46.0,
-      "step": 5003
-    },
-    {
-      "epoch": 0.8058295422521036,
-      "grad_norm": 0.0019281983841210604,
-      "learning_rate": 0.0001999996808316921,
-      "loss": 46.0,
-      "step": 5004
-    },
-    {
-      "epoch": 0.8059905793308909,
-      "grad_norm": 0.0008937554666772485,
-      "learning_rate": 0.00019999968070385867,
-      "loss": 46.0,
-      "step": 5005
-    },
-    {
-      "epoch": 0.8061516164096784,
-      "grad_norm": 0.000303289940347895,
-      "learning_rate": 0.00019999968057599963,
-      "loss": 46.0,
-      "step": 5006
-    },
-    {
-      "epoch": 0.8063126534884657,
-      "grad_norm": 0.0005682788323611021,
-      "learning_rate": 0.000199999680448115,
-      "loss": 46.0,
-      "step": 5007
-    },
-    {
-      "epoch": 0.8064736905672532,
-      "grad_norm": 0.0003950601676478982,
-      "learning_rate": 0.0001999996803202048,
-      "loss": 46.0,
-      "step": 5008
-    },
-    {
-      "epoch": 0.8066347276460405,
-      "grad_norm": 0.00045897779637016356,
-      "learning_rate": 0.00019999968019226896,
-      "loss": 46.0,
-      "step": 5009
-    },
-    {
-      "epoch": 0.806795764724828,
-      "grad_norm": 0.0011327541433274746,
-      "learning_rate": 0.00019999968006430754,
-      "loss": 46.0,
-      "step": 5010
-    },
-    {
-      "epoch": 0.8069568018036153,
-      "grad_norm": 0.0008953461074270308,
-      "learning_rate": 0.00019999967993632054,
-      "loss": 46.0,
-      "step": 5011
-    },
-    {
-      "epoch": 0.8071178388824026,
-      "grad_norm": 0.0023495692294090986,
-      "learning_rate": 0.00019999967980830795,
-      "loss": 46.0,
-      "step": 5012
-    },
-    {
-      "epoch": 0.8072788759611901,
-      "grad_norm": 0.0019212618935853243,
-      "learning_rate": 0.00019999967968026977,
-      "loss": 46.0,
-      "step": 5013
-    },
-    {
-      "epoch": 0.8074399130399774,
-      "grad_norm": 0.0010068389819934964,
-      "learning_rate": 0.00019999967955220595,
-      "loss": 46.0,
-      "step": 5014
-    },
-    {
-      "epoch": 0.8076009501187649,
-      "grad_norm": 0.00038365210639312863,
-      "learning_rate": 0.00019999967942411657,
-      "loss": 46.0,
-      "step": 5015
-    },
-    {
-      "epoch": 0.8077619871975522,
-      "grad_norm": 0.0007847102242521942,
-      "learning_rate": 0.00019999967929600158,
-      "loss": 46.0,
-      "step": 5016
-    },
-    {
-      "epoch": 0.8079230242763397,
-      "grad_norm": 0.0008404904510825872,
-      "learning_rate": 0.00019999967916786102,
-      "loss": 46.0,
-      "step": 5017
-    },
-    {
-      "epoch": 0.808084061355127,
-      "grad_norm": 0.0009001229773275554,
-      "learning_rate": 0.00019999967903969483,
-      "loss": 46.0,
-      "step": 5018
-    },
-    {
-      "epoch": 0.8082450984339145,
-      "grad_norm": 0.0018883231095969677,
-      "learning_rate": 0.00019999967891150307,
-      "loss": 46.0,
-      "step": 5019
-    },
-    {
-      "epoch": 0.8084061355127018,
-      "grad_norm": 0.0006504155462607741,
-      "learning_rate": 0.0001999996787832857,
-      "loss": 46.0,
-      "step": 5020
-    },
-    {
-      "epoch": 0.8085671725914891,
-      "grad_norm": 0.0006420365534722805,
-      "learning_rate": 0.00019999967865504277,
-      "loss": 46.0,
-      "step": 5021
-    },
-    {
-      "epoch": 0.8087282096702766,
-      "grad_norm": 0.001215107156895101,
-      "learning_rate": 0.0001999996785267742,
-      "loss": 46.0,
-      "step": 5022
-    },
-    {
-      "epoch": 0.8088892467490639,
-      "grad_norm": 0.0016643713461235166,
-      "learning_rate": 0.00019999967839848007,
-      "loss": 46.0,
-      "step": 5023
-    },
-    {
-      "epoch": 0.8090502838278514,
-      "grad_norm": 0.0009740013047121465,
-      "learning_rate": 0.00019999967827016035,
-      "loss": 46.0,
-      "step": 5024
-    },
-    {
-      "epoch": 0.8092113209066387,
-      "grad_norm": 0.00035582008422352374,
-      "learning_rate": 0.00019999967814181502,
-      "loss": 46.0,
-      "step": 5025
-    },
-    {
-      "epoch": 0.8093723579854262,
-      "grad_norm": 0.0012886858312413096,
-      "learning_rate": 0.00019999967801344408,
-      "loss": 46.0,
-      "step": 5026
-    },
-    {
-      "epoch": 0.8095333950642135,
-      "grad_norm": 0.001792251132428646,
-      "learning_rate": 0.00019999967788504757,
-      "loss": 46.0,
-      "step": 5027
-    },
-    {
-      "epoch": 0.809694432143001,
-      "grad_norm": 0.0004244278825353831,
-      "learning_rate": 0.00019999967775662545,
-      "loss": 46.0,
-      "step": 5028
-    },
-    {
-      "epoch": 0.8098554692217883,
-      "grad_norm": 0.0006020162254571915,
-      "learning_rate": 0.0001999996776281777,
-      "loss": 46.0,
-      "step": 5029
-    },
-    {
-      "epoch": 0.8100165063005758,
-      "grad_norm": 0.0013522617518901825,
-      "learning_rate": 0.00019999967749970442,
-      "loss": 46.0,
-      "step": 5030
-    },
-    {
-      "epoch": 0.8101775433793631,
-      "grad_norm": 0.0005848242435604334,
-      "learning_rate": 0.00019999967737120553,
-      "loss": 46.0,
-      "step": 5031
-    },
-    {
-      "epoch": 0.8103385804581504,
-      "grad_norm": 0.0006331526674330235,
-      "learning_rate": 0.000199999677242681,
-      "loss": 46.0,
-      "step": 5032
-    },
-    {
-      "epoch": 0.8104996175369379,
-      "grad_norm": 0.0007234473014250398,
-      "learning_rate": 0.00019999967711413093,
-      "loss": 46.0,
-      "step": 5033
-    },
-    {
-      "epoch": 0.8106606546157252,
-      "grad_norm": 0.0042146965861320496,
-      "learning_rate": 0.00019999967698555523,
-      "loss": 46.0,
-      "step": 5034
-    },
-    {
-      "epoch": 0.8108216916945127,
-      "grad_norm": 0.0005461159162223339,
-      "learning_rate": 0.00019999967685695394,
-      "loss": 46.0,
-      "step": 5035
-    },
-    {
-      "epoch": 0.8109827287733,
-      "grad_norm": 0.003211408853530884,
-      "learning_rate": 0.00019999967672832707,
-      "loss": 46.0,
-      "step": 5036
-    },
-    {
-      "epoch": 0.8111437658520875,
-      "grad_norm": 0.0005885272985324264,
-      "learning_rate": 0.0001999996765996746,
-      "loss": 46.0,
-      "step": 5037
-    },
-    {
-      "epoch": 0.8113048029308748,
-      "grad_norm": 0.0016029799589887261,
-      "learning_rate": 0.00019999967647099656,
-      "loss": 46.0,
-      "step": 5038
-    },
-    {
-      "epoch": 0.8114658400096623,
-      "grad_norm": 0.0013309215428307652,
-      "learning_rate": 0.00019999967634229287,
-      "loss": 46.0,
-      "step": 5039
-    },
-    {
-      "epoch": 0.8116268770884496,
-      "grad_norm": 0.0015502702444791794,
-      "learning_rate": 0.00019999967621356363,
-      "loss": 46.0,
-      "step": 5040
-    },
-    {
-      "epoch": 0.8117879141672371,
-      "grad_norm": 0.0012823846191167831,
-      "learning_rate": 0.0001999996760848088,
-      "loss": 46.0,
-      "step": 5041
-    },
-    {
-      "epoch": 0.8119489512460244,
-      "grad_norm": 0.0007383690099231899,
-      "learning_rate": 0.00019999967595602834,
-      "loss": 46.0,
-      "step": 5042
-    },
-    {
-      "epoch": 0.8121099883248117,
-      "grad_norm": 0.00037825643084943295,
-      "learning_rate": 0.0001999996758272223,
-      "loss": 46.0,
-      "step": 5043
-    },
-    {
-      "epoch": 0.8122710254035992,
-      "grad_norm": 0.00038448069244623184,
-      "learning_rate": 0.00019999967569839065,
-      "loss": 46.0,
-      "step": 5044
-    },
-    {
-      "epoch": 0.8124320624823865,
-      "grad_norm": 0.0008973998483270407,
-      "learning_rate": 0.00019999967556953344,
-      "loss": 46.0,
-      "step": 5045
-    },
-    {
-      "epoch": 0.812593099561174,
-      "grad_norm": 0.0008637954015284777,
-      "learning_rate": 0.00019999967544065062,
-      "loss": 46.0,
-      "step": 5046
-    },
-    {
-      "epoch": 0.8127541366399613,
-      "grad_norm": 0.0038218258414417505,
-      "learning_rate": 0.0001999996753117422,
-      "loss": 46.0,
-      "step": 5047
-    },
-    {
-      "epoch": 0.8129151737187488,
-      "grad_norm": 0.0007042171200737357,
-      "learning_rate": 0.00019999967518280818,
-      "loss": 46.0,
-      "step": 5048
-    },
-    {
-      "epoch": 0.8130762107975361,
-      "grad_norm": 0.000422991142841056,
-      "learning_rate": 0.00019999967505384856,
-      "loss": 46.0,
-      "step": 5049
-    },
-    {
-      "epoch": 0.8132372478763236,
-      "grad_norm": 0.0008916970691643655,
-      "learning_rate": 0.0001999996749248634,
-      "loss": 46.0,
-      "step": 5050
-    },
-    {
-      "epoch": 0.8133982849551109,
-      "grad_norm": 0.0009910401422530413,
-      "learning_rate": 0.00019999967479585257,
-      "loss": 46.0,
-      "step": 5051
-    },
-    {
-      "epoch": 0.8135593220338984,
-      "grad_norm": 0.0027690459974110126,
-      "learning_rate": 0.0001999996746668162,
-      "loss": 46.0,
-      "step": 5052
-    },
-    {
-      "epoch": 0.8137203591126857,
-      "grad_norm": 0.0004160216776654124,
-      "learning_rate": 0.00019999967453775418,
-      "loss": 46.0,
-      "step": 5053
-    },
-    {
-      "epoch": 0.813881396191473,
-      "grad_norm": 0.004208892118185759,
-      "learning_rate": 0.0001999996744086666,
-      "loss": 46.0,
-      "step": 5054
-    },
-    {
-      "epoch": 0.8140424332702605,
-      "grad_norm": 0.00034192390739917755,
-      "learning_rate": 0.00019999967427955344,
-      "loss": 46.0,
-      "step": 5055
-    },
-    {
-      "epoch": 0.8142034703490478,
-      "grad_norm": 0.001324885874055326,
-      "learning_rate": 0.00019999967415041466,
-      "loss": 46.0,
-      "step": 5056
-    },
-    {
-      "epoch": 0.8143645074278353,
-      "grad_norm": 0.0003919968439731747,
-      "learning_rate": 0.0001999996740212503,
-      "loss": 46.0,
-      "step": 5057
-    },
-    {
-      "epoch": 0.8145255445066226,
-      "grad_norm": 0.00042274646693840623,
-      "learning_rate": 0.00019999967389206034,
-      "loss": 46.0,
-      "step": 5058
-    },
-    {
-      "epoch": 0.8146865815854101,
-      "grad_norm": 0.0005962664727121592,
-      "learning_rate": 0.00019999967376284477,
-      "loss": 46.0,
-      "step": 5059
-    },
-    {
-      "epoch": 0.8148476186641974,
-      "grad_norm": 0.001136332517489791,
-      "learning_rate": 0.00019999967363360362,
-      "loss": 46.0,
-      "step": 5060
-    },
-    {
-      "epoch": 0.8150086557429849,
-      "grad_norm": 0.0004930444993078709,
-      "learning_rate": 0.00019999967350433688,
-      "loss": 46.0,
-      "step": 5061
-    },
-    {
-      "epoch": 0.8151696928217722,
-      "grad_norm": 0.0006116154254414141,
-      "learning_rate": 0.00019999967337504452,
-      "loss": 46.0,
-      "step": 5062
-    },
-    {
-      "epoch": 0.8153307299005597,
-      "grad_norm": 0.0006569311372004449,
-      "learning_rate": 0.0001999996732457266,
-      "loss": 46.0,
-      "step": 5063
-    },
-    {
-      "epoch": 0.815491766979347,
-      "grad_norm": 0.0032229472417384386,
-      "learning_rate": 0.00019999967311638307,
-      "loss": 46.0,
-      "step": 5064
-    },
-    {
-      "epoch": 0.8156528040581343,
-      "grad_norm": 0.0067578526213765144,
-      "learning_rate": 0.00019999967298701393,
-      "loss": 46.0,
-      "step": 5065
-    },
-    {
-      "epoch": 0.8158138411369218,
-      "grad_norm": 0.001125033595599234,
-      "learning_rate": 0.00019999967285761922,
-      "loss": 46.0,
-      "step": 5066
-    },
-    {
-      "epoch": 0.8159748782157091,
-      "grad_norm": 0.0028744186274707317,
-      "learning_rate": 0.00019999967272819893,
-      "loss": 46.0,
-      "step": 5067
-    },
-    {
-      "epoch": 0.8161359152944966,
-      "grad_norm": 0.0007400520844385028,
-      "learning_rate": 0.000199999672598753,
-      "loss": 46.0,
-      "step": 5068
-    },
-    {
-      "epoch": 0.8162969523732839,
-      "grad_norm": 0.0031711275223642588,
-      "learning_rate": 0.0001999996724692815,
-      "loss": 46.0,
-      "step": 5069
-    },
-    {
-      "epoch": 0.8164579894520714,
-      "grad_norm": 0.0010434158612042665,
-      "learning_rate": 0.0001999996723397844,
-      "loss": 46.0,
-      "step": 5070
-    },
-    {
-      "epoch": 0.8166190265308587,
-      "grad_norm": 0.00046424055472016335,
-      "learning_rate": 0.00019999967221026173,
-      "loss": 46.0,
-      "step": 5071
-    },
-    {
-      "epoch": 0.8167800636096462,
-      "grad_norm": 0.0009833801304921508,
-      "learning_rate": 0.00019999967208071345,
-      "loss": 46.0,
-      "step": 5072
-    },
-    {
-      "epoch": 0.8169411006884335,
-      "grad_norm": 0.001199959428049624,
-      "learning_rate": 0.00019999967195113953,
-      "loss": 46.0,
-      "step": 5073
-    },
-    {
-      "epoch": 0.8171021377672208,
-      "grad_norm": 0.002370106056332588,
-      "learning_rate": 0.00019999967182154007,
-      "loss": 46.0,
-      "step": 5074
-    },
-    {
-      "epoch": 0.8172631748460083,
-      "grad_norm": 0.0013635419309139252,
-      "learning_rate": 0.00019999967169191497,
-      "loss": 46.0,
-      "step": 5075
-    },
-    {
-      "epoch": 0.8174242119247956,
-      "grad_norm": 0.0043939026072621346,
-      "learning_rate": 0.00019999967156226434,
-      "loss": 46.0,
-      "step": 5076
-    },
-    {
-      "epoch": 0.8175852490035831,
-      "grad_norm": 0.0003139891487080604,
-      "learning_rate": 0.00019999967143258807,
-      "loss": 46.0,
-      "step": 5077
-    },
-    {
-      "epoch": 0.8177462860823704,
-      "grad_norm": 0.000730452942661941,
-      "learning_rate": 0.0001999996713028862,
-      "loss": 46.0,
-      "step": 5078
-    },
-    {
-      "epoch": 0.8179073231611579,
-      "grad_norm": 0.0018734782934188843,
-      "learning_rate": 0.00019999967117315874,
-      "loss": 46.0,
-      "step": 5079
-    },
-    {
-      "epoch": 0.8180683602399452,
-      "grad_norm": 0.0011661278549581766,
-      "learning_rate": 0.0001999996710434057,
-      "loss": 46.0,
-      "step": 5080
-    },
-    {
-      "epoch": 0.8182293973187327,
-      "grad_norm": 0.005574526730924845,
-      "learning_rate": 0.00019999967091362706,
-      "loss": 46.0,
-      "step": 5081
-    },
-    {
-      "epoch": 0.81839043439752,
-      "grad_norm": 0.0009668830898590386,
-      "learning_rate": 0.00019999967078382285,
-      "loss": 46.0,
-      "step": 5082
-    },
-    {
-      "epoch": 0.8185514714763075,
-      "grad_norm": 0.0014443137915804982,
-      "learning_rate": 0.000199999670653993,
-      "loss": 46.0,
-      "step": 5083
-    },
-    {
-      "epoch": 0.8187125085550948,
-      "grad_norm": 0.0013937398325651884,
-      "learning_rate": 0.00019999967052413756,
-      "loss": 46.0,
-      "step": 5084
-    },
-    {
-      "epoch": 0.8188735456338821,
-      "grad_norm": 0.0007658318500034511,
-      "learning_rate": 0.00019999967039425657,
-      "loss": 46.0,
-      "step": 5085
-    },
-    {
-      "epoch": 0.8190345827126696,
-      "grad_norm": 0.001386961666867137,
-      "learning_rate": 0.00019999967026434993,
-      "loss": 46.0,
-      "step": 5086
-    },
-    {
-      "epoch": 0.8191956197914569,
-      "grad_norm": 0.004695508629083633,
-      "learning_rate": 0.00019999967013441773,
-      "loss": 46.0,
-      "step": 5087
-    },
-    {
-      "epoch": 0.8193566568702444,
-      "grad_norm": 0.0005968232289887965,
-      "learning_rate": 0.00019999967000445992,
-      "loss": 46.0,
-      "step": 5088
-    },
-    {
-      "epoch": 0.8195176939490317,
-      "grad_norm": 0.0008575798710808158,
-      "learning_rate": 0.00019999966987447652,
-      "loss": 46.0,
-      "step": 5089
-    },
-    {
-      "epoch": 0.8196787310278192,
-      "grad_norm": 0.0040957932360470295,
-      "learning_rate": 0.00019999966974446753,
-      "loss": 46.0,
-      "step": 5090
-    },
-    {
-      "epoch": 0.8198397681066065,
-      "grad_norm": 0.001956131774932146,
-      "learning_rate": 0.00019999966961443293,
-      "loss": 46.0,
-      "step": 5091
-    },
-    {
-      "epoch": 0.820000805185394,
-      "grad_norm": 0.0004973598406650126,
-      "learning_rate": 0.00019999966948437277,
-      "loss": 46.0,
-      "step": 5092
-    },
-    {
-      "epoch": 0.8201618422641813,
-      "grad_norm": 0.002371931914240122,
-      "learning_rate": 0.00019999966935428697,
-      "loss": 46.0,
-      "step": 5093
-    },
-    {
-      "epoch": 0.8203228793429688,
-      "grad_norm": 0.0005688421661034226,
-      "learning_rate": 0.0001999996692241756,
-      "loss": 46.0,
-      "step": 5094
-    },
-    {
-      "epoch": 0.8204839164217561,
-      "grad_norm": 0.004293699748814106,
-      "learning_rate": 0.00019999966909403863,
-      "loss": 46.0,
-      "step": 5095
-    },
-    {
-      "epoch": 0.8206449535005435,
-      "grad_norm": 0.0007094823522493243,
-      "learning_rate": 0.00019999966896387607,
-      "loss": 46.0,
-      "step": 5096
-    },
-    {
-      "epoch": 0.8208059905793309,
-      "grad_norm": 0.000648115819785744,
-      "learning_rate": 0.0001999996688336879,
-      "loss": 46.0,
-      "step": 5097
-    },
-    {
-      "epoch": 0.8209670276581182,
-      "grad_norm": 0.000635456875897944,
-      "learning_rate": 0.00019999966870347415,
-      "loss": 46.0,
-      "step": 5098
-    },
-    {
-      "epoch": 0.8211280647369057,
-      "grad_norm": 0.0007707527838647366,
-      "learning_rate": 0.00019999966857323483,
-      "loss": 46.0,
-      "step": 5099
-    },
-    {
-      "epoch": 0.821289101815693,
-      "grad_norm": 0.0008147604530677199,
-      "learning_rate": 0.00019999966844296986,
-      "loss": 46.0,
-      "step": 5100
-    },
-    {
-      "epoch": 0.8214501388944805,
-      "grad_norm": 0.0008703813655301929,
-      "learning_rate": 0.00019999966831267933,
-      "loss": 46.0,
-      "step": 5101
-    },
-    {
-      "epoch": 0.8216111759732678,
-      "grad_norm": 0.0010957014746963978,
-      "learning_rate": 0.00019999966818236322,
-      "loss": 46.0,
-      "step": 5102
-    },
-    {
-      "epoch": 0.8217722130520553,
-      "grad_norm": 0.001453327713534236,
-      "learning_rate": 0.0001999996680520215,
-      "loss": 46.0,
-      "step": 5103
-    },
-    {
-      "epoch": 0.8219332501308426,
-      "grad_norm": 0.0025897640734910965,
-      "learning_rate": 0.00019999966792165418,
-      "loss": 46.0,
-      "step": 5104
-    },
-    {
-      "epoch": 0.8220942872096301,
-      "grad_norm": 0.0007721189176663756,
-      "learning_rate": 0.00019999966779126125,
-      "loss": 46.0,
-      "step": 5105
-    },
-    {
-      "epoch": 0.8222553242884174,
-      "grad_norm": 0.0013284151209518313,
-      "learning_rate": 0.00019999966766084276,
-      "loss": 46.0,
-      "step": 5106
-    },
-    {
-      "epoch": 0.8224163613672048,
-      "grad_norm": 0.000944761501159519,
-      "learning_rate": 0.00019999966753039863,
-      "loss": 46.0,
-      "step": 5107
-    },
-    {
-      "epoch": 0.8225773984459922,
-      "grad_norm": 0.0005585766048170626,
-      "learning_rate": 0.00019999966739992894,
-      "loss": 46.0,
-      "step": 5108
-    },
-    {
-      "epoch": 0.8227384355247795,
-      "grad_norm": 0.0022968715056777,
-      "learning_rate": 0.00019999966726943363,
-      "loss": 46.0,
-      "step": 5109
-    },
-    {
-      "epoch": 0.822899472603567,
-      "grad_norm": 0.0046213772147893906,
-      "learning_rate": 0.00019999966713891274,
-      "loss": 46.0,
-      "step": 5110
-    },
-    {
-      "epoch": 0.8230605096823543,
-      "grad_norm": 0.004788071382790804,
-      "learning_rate": 0.00019999966700836626,
-      "loss": 46.0,
-      "step": 5111
-    },
-    {
-      "epoch": 0.8232215467611418,
-      "grad_norm": 0.000617998477537185,
-      "learning_rate": 0.0001999996668777942,
-      "loss": 46.0,
-      "step": 5112
-    },
-    {
-      "epoch": 0.8233825838399291,
-      "grad_norm": 0.0010753405513241887,
-      "learning_rate": 0.0001999996667471965,
-      "loss": 46.0,
-      "step": 5113
-    },
-    {
-      "epoch": 0.8235436209187166,
-      "grad_norm": 0.0015353562775999308,
-      "learning_rate": 0.00019999966661657324,
-      "loss": 46.0,
-      "step": 5114
-    },
-    {
-      "epoch": 0.8237046579975039,
-      "grad_norm": 0.0008767135441303253,
-      "learning_rate": 0.00019999966648592436,
-      "loss": 46.0,
-      "step": 5115
-    },
-    {
-      "epoch": 0.8238656950762914,
-      "grad_norm": 0.001422939240001142,
-      "learning_rate": 0.00019999966635524992,
-      "loss": 46.0,
-      "step": 5116
-    },
-    {
-      "epoch": 0.8240267321550787,
-      "grad_norm": 0.0006881275912746787,
-      "learning_rate": 0.00019999966622454986,
-      "loss": 46.0,
-      "step": 5117
-    },
-    {
-      "epoch": 0.824187769233866,
-      "grad_norm": 0.0009932867251336575,
-      "learning_rate": 0.00019999966609382422,
-      "loss": 46.0,
-      "step": 5118
-    },
-    {
-      "epoch": 0.8243488063126535,
-      "grad_norm": 0.0010311916703358293,
-      "learning_rate": 0.00019999966596307296,
-      "loss": 46.0,
-      "step": 5119
-    },
-    {
-      "epoch": 0.8245098433914408,
-      "grad_norm": 0.00047003134386613965,
-      "learning_rate": 0.00019999966583229614,
-      "loss": 46.0,
-      "step": 5120
-    },
-    {
-      "epoch": 0.8246708804702283,
-      "grad_norm": 0.003551816800609231,
-      "learning_rate": 0.0001999996657014937,
-      "loss": 46.0,
-      "step": 5121
-    },
-    {
-      "epoch": 0.8248319175490156,
-      "grad_norm": 0.0004784806224051863,
-      "learning_rate": 0.00019999966557066569,
-      "loss": 46.0,
-      "step": 5122
-    },
-    {
-      "epoch": 0.8249929546278031,
-      "grad_norm": 0.001361970091238618,
-      "learning_rate": 0.00019999966543981205,
-      "loss": 46.0,
-      "step": 5123
-    },
-    {
-      "epoch": 0.8251539917065904,
-      "grad_norm": 0.0008016018546186388,
-      "learning_rate": 0.00019999966530893283,
-      "loss": 46.0,
-      "step": 5124
-    },
-    {
-      "epoch": 0.8253150287853779,
-      "grad_norm": 0.0015518533764407039,
-      "learning_rate": 0.00019999966517802802,
-      "loss": 46.0,
-      "step": 5125
-    },
-    {
-      "epoch": 0.8254760658641652,
-      "grad_norm": 0.0009186487877741456,
-      "learning_rate": 0.00019999966504709763,
-      "loss": 46.0,
-      "step": 5126
-    },
-    {
-      "epoch": 0.8256371029429526,
-      "grad_norm": 0.0014622766757383943,
-      "learning_rate": 0.00019999966491614162,
-      "loss": 46.0,
-      "step": 5127
-    },
-    {
-      "epoch": 0.82579814002174,
-      "grad_norm": 0.0023140297271311283,
-      "learning_rate": 0.00019999966478516002,
-      "loss": 46.0,
-      "step": 5128
-    },
-    {
-      "epoch": 0.8259591771005274,
-      "grad_norm": 0.0023945504799485207,
-      "learning_rate": 0.00019999966465415284,
-      "loss": 46.0,
-      "step": 5129
-    },
-    {
-      "epoch": 0.8261202141793148,
-      "grad_norm": 0.00042854438652284443,
-      "learning_rate": 0.00019999966452312004,
-      "loss": 46.0,
-      "step": 5130
-    },
-    {
-      "epoch": 0.8262812512581021,
-      "grad_norm": 0.0013614867348223925,
-      "learning_rate": 0.00019999966439206165,
-      "loss": 46.0,
-      "step": 5131
-    },
-    {
-      "epoch": 0.8264422883368896,
-      "grad_norm": 0.0005885640275664628,
-      "learning_rate": 0.0001999996642609777,
-      "loss": 46.0,
-      "step": 5132
-    },
-    {
-      "epoch": 0.8266033254156769,
-      "grad_norm": 0.0010897950269281864,
-      "learning_rate": 0.00019999966412986812,
-      "loss": 46.0,
-      "step": 5133
-    },
-    {
-      "epoch": 0.8267643624944644,
-      "grad_norm": 0.0013300544815137982,
-      "learning_rate": 0.00019999966399873294,
-      "loss": 46.0,
-      "step": 5134
-    },
-    {
-      "epoch": 0.8269253995732517,
-      "grad_norm": 0.000617846439126879,
-      "learning_rate": 0.0001999996638675722,
-      "loss": 46.0,
-      "step": 5135
-    },
-    {
-      "epoch": 0.8270864366520392,
-      "grad_norm": 0.0006907997885718942,
-      "learning_rate": 0.00019999966373638583,
-      "loss": 46.0,
-      "step": 5136
-    },
-    {
-      "epoch": 0.8272474737308265,
-      "grad_norm": 0.0005879865493625402,
-      "learning_rate": 0.00019999966360517387,
-      "loss": 46.0,
-      "step": 5137
-    },
-    {
-      "epoch": 0.8274085108096139,
-      "grad_norm": 0.0007456826278939843,
-      "learning_rate": 0.00019999966347393632,
-      "loss": 46.0,
-      "step": 5138
-    },
-    {
-      "epoch": 0.8275695478884013,
-      "grad_norm": 0.0006295247003436089,
-      "learning_rate": 0.00019999966334267318,
-      "loss": 46.0,
-      "step": 5139
-    },
-    {
-      "epoch": 0.8277305849671887,
-      "grad_norm": 0.0007951277657411993,
-      "learning_rate": 0.00019999966321138449,
-      "loss": 46.0,
-      "step": 5140
-    },
-    {
-      "epoch": 0.8278916220459761,
-      "grad_norm": 0.0006441830191761255,
-      "learning_rate": 0.00019999966308007015,
-      "loss": 46.0,
-      "step": 5141
-    },
-    {
-      "epoch": 0.8280526591247634,
-      "grad_norm": 0.0011686583748087287,
-      "learning_rate": 0.00019999966294873022,
-      "loss": 46.0,
-      "step": 5142
-    },
-    {
-      "epoch": 0.8282136962035509,
-      "grad_norm": 0.0032313179690390825,
-      "learning_rate": 0.0001999996628173647,
-      "loss": 46.0,
-      "step": 5143
-    },
-    {
-      "epoch": 0.8283747332823382,
-      "grad_norm": 0.0010612113401293755,
-      "learning_rate": 0.00019999966268597358,
-      "loss": 46.0,
-      "step": 5144
-    },
-    {
-      "epoch": 0.8285357703611257,
-      "grad_norm": 0.0068147857673466206,
-      "learning_rate": 0.0001999996625545569,
-      "loss": 46.0,
-      "step": 5145
-    },
-    {
-      "epoch": 0.828696807439913,
-      "grad_norm": 0.0029785921797156334,
-      "learning_rate": 0.0001999996624231146,
-      "loss": 46.0,
-      "step": 5146
-    },
-    {
-      "epoch": 0.8288578445187005,
-      "grad_norm": 0.0004786409845110029,
-      "learning_rate": 0.00019999966229164668,
-      "loss": 46.0,
-      "step": 5147
-    },
-    {
-      "epoch": 0.8290188815974878,
-      "grad_norm": 0.0011323199141770601,
-      "learning_rate": 0.0001999996621601532,
-      "loss": 46.0,
-      "step": 5148
-    },
-    {
-      "epoch": 0.8291799186762752,
-      "grad_norm": 0.005421845242381096,
-      "learning_rate": 0.00019999966202863408,
-      "loss": 46.0,
-      "step": 5149
-    },
-    {
-      "epoch": 0.8293409557550626,
-      "grad_norm": 0.002832339145243168,
-      "learning_rate": 0.00019999966189708943,
-      "loss": 46.0,
-      "step": 5150
-    },
-    {
-      "epoch": 0.82950199283385,
-      "grad_norm": 0.00041840950143523514,
-      "learning_rate": 0.00019999966176551914,
-      "loss": 46.0,
-      "step": 5151
-    },
-    {
-      "epoch": 0.8296630299126374,
-      "grad_norm": 0.0016298888949677348,
-      "learning_rate": 0.00019999966163392326,
-      "loss": 46.0,
-      "step": 5152
-    },
-    {
-      "epoch": 0.8298240669914247,
-      "grad_norm": 0.0020185254979878664,
-      "learning_rate": 0.00019999966150230182,
-      "loss": 46.0,
-      "step": 5153
-    },
-    {
-      "epoch": 0.8299851040702122,
-      "grad_norm": 0.0010778631549328566,
-      "learning_rate": 0.00019999966137065477,
-      "loss": 46.0,
-      "step": 5154
-    },
-    {
-      "epoch": 0.8301461411489995,
-      "grad_norm": 0.0006393252406269312,
-      "learning_rate": 0.0001999996612389821,
-      "loss": 46.0,
-      "step": 5155
-    },
-    {
-      "epoch": 0.830307178227787,
-      "grad_norm": 0.0006912277895025909,
-      "learning_rate": 0.00019999966110728385,
-      "loss": 46.0,
-      "step": 5156
-    },
-    {
-      "epoch": 0.8304682153065743,
-      "grad_norm": 0.00045371902524493635,
-      "learning_rate": 0.00019999966097556,
-      "loss": 46.0,
-      "step": 5157
-    },
-    {
-      "epoch": 0.8306292523853618,
-      "grad_norm": 0.0007965494296513498,
-      "learning_rate": 0.00019999966084381058,
-      "loss": 46.0,
-      "step": 5158
-    },
-    {
-      "epoch": 0.8307902894641491,
-      "grad_norm": 0.001802053302526474,
-      "learning_rate": 0.00019999966071203553,
-      "loss": 46.0,
-      "step": 5159
-    },
-    {
-      "epoch": 0.8309513265429365,
-      "grad_norm": 0.000842803216073662,
-      "learning_rate": 0.00019999966058023493,
-      "loss": 46.0,
-      "step": 5160
-    },
-    {
-      "epoch": 0.8311123636217239,
-      "grad_norm": 0.0008711520349606872,
-      "learning_rate": 0.00019999966044840868,
-      "loss": 46.0,
-      "step": 5161
-    },
-    {
-      "epoch": 0.8312734007005113,
-      "grad_norm": 0.00036608861410059035,
-      "learning_rate": 0.00019999966031655685,
-      "loss": 46.0,
-      "step": 5162
-    },
-    {
-      "epoch": 0.8314344377792987,
-      "grad_norm": 0.0019751228392124176,
-      "learning_rate": 0.00019999966018467943,
-      "loss": 46.0,
-      "step": 5163
-    },
-    {
-      "epoch": 0.831595474858086,
-      "grad_norm": 0.0008423432591371238,
-      "learning_rate": 0.00019999966005277643,
-      "loss": 46.0,
-      "step": 5164
-    },
-    {
-      "epoch": 0.8317565119368735,
-      "grad_norm": 0.0033180860336869955,
-      "learning_rate": 0.00019999965992084783,
-      "loss": 46.0,
-      "step": 5165
-    },
-    {
-      "epoch": 0.8319175490156608,
-      "grad_norm": 0.004071468487381935,
-      "learning_rate": 0.00019999965978889363,
-      "loss": 46.0,
-      "step": 5166
-    },
-    {
-      "epoch": 0.8320785860944483,
-      "grad_norm": 0.0012572259875014424,
-      "learning_rate": 0.00019999965965691386,
-      "loss": 46.0,
-      "step": 5167
-    },
-    {
-      "epoch": 0.8322396231732356,
-      "grad_norm": 0.0006736009963788092,
-      "learning_rate": 0.00019999965952490848,
-      "loss": 46.0,
-      "step": 5168
-    },
-    {
-      "epoch": 0.8324006602520231,
-      "grad_norm": 0.004254922736436129,
-      "learning_rate": 0.00019999965939287745,
-      "loss": 46.0,
-      "step": 5169
-    },
-    {
-      "epoch": 0.8325616973308104,
-      "grad_norm": 0.0007358370930887759,
-      "learning_rate": 0.0001999996592608209,
-      "loss": 46.0,
-      "step": 5170
-    },
-    {
-      "epoch": 0.8327227344095978,
-      "grad_norm": 0.0020096038933843374,
-      "learning_rate": 0.00019999965912873872,
-      "loss": 46.0,
-      "step": 5171
-    },
-    {
-      "epoch": 0.8328837714883852,
-      "grad_norm": 0.0010486580431461334,
-      "learning_rate": 0.00019999965899663097,
-      "loss": 46.0,
-      "step": 5172
-    },
-    {
-      "epoch": 0.8330448085671726,
-      "grad_norm": 0.0010104109533131123,
-      "learning_rate": 0.0001999996588644976,
-      "loss": 46.0,
-      "step": 5173
-    },
-    {
-      "epoch": 0.83320584564596,
-      "grad_norm": 0.0009896950796246529,
-      "learning_rate": 0.00019999965873233864,
-      "loss": 46.0,
-      "step": 5174
-    },
-    {
-      "epoch": 0.8333668827247473,
-      "grad_norm": 0.0012071151286363602,
-      "learning_rate": 0.00019999965860015412,
-      "loss": 46.0,
-      "step": 5175
-    },
-    {
-      "epoch": 0.8335279198035348,
-      "grad_norm": 0.0007486878894269466,
-      "learning_rate": 0.00019999965846794396,
-      "loss": 46.0,
-      "step": 5176
-    },
-    {
-      "epoch": 0.8336889568823221,
-      "grad_norm": 0.0030823503620922565,
-      "learning_rate": 0.0001999996583357082,
-      "loss": 46.0,
-      "step": 5177
-    },
-    {
-      "epoch": 0.8338499939611096,
-      "grad_norm": 0.0012784936698153615,
-      "learning_rate": 0.00019999965820344687,
-      "loss": 46.0,
-      "step": 5178
-    },
-    {
-      "epoch": 0.8340110310398969,
-      "grad_norm": 0.000638270634226501,
-      "learning_rate": 0.00019999965807115992,
-      "loss": 46.0,
-      "step": 5179
-    },
-    {
-      "epoch": 0.8341720681186843,
-      "grad_norm": 0.0019719786942005157,
-      "learning_rate": 0.0001999996579388474,
-      "loss": 46.0,
-      "step": 5180
-    },
-    {
-      "epoch": 0.8343331051974717,
-      "grad_norm": 0.0009430918726138771,
-      "learning_rate": 0.00019999965780650932,
-      "loss": 46.0,
-      "step": 5181
-    },
-    {
-      "epoch": 0.8344941422762591,
-      "grad_norm": 0.000830064935144037,
-      "learning_rate": 0.00019999965767414558,
-      "loss": 46.0,
-      "step": 5182
-    },
-    {
-      "epoch": 0.8346551793550465,
-      "grad_norm": 0.0015963127370923758,
-      "learning_rate": 0.00019999965754175628,
-      "loss": 46.0,
-      "step": 5183
-    },
-    {
-      "epoch": 0.8348162164338339,
-      "grad_norm": 0.0006838154513388872,
-      "learning_rate": 0.00019999965740934137,
-      "loss": 46.0,
-      "step": 5184
-    },
-    {
-      "epoch": 0.8349772535126213,
-      "grad_norm": 0.0011424896074458957,
-      "learning_rate": 0.00019999965727690087,
-      "loss": 46.0,
-      "step": 5185
-    },
-    {
-      "epoch": 0.8351382905914086,
-      "grad_norm": 0.0030657697934657335,
-      "learning_rate": 0.00019999965714443478,
-      "loss": 46.0,
-      "step": 5186
-    },
-    {
-      "epoch": 0.8352993276701961,
-      "grad_norm": 0.0010062780929729342,
-      "learning_rate": 0.00019999965701194308,
-      "loss": 46.0,
-      "step": 5187
-    },
-    {
-      "epoch": 0.8354603647489834,
-      "grad_norm": 0.0032498009968549013,
-      "learning_rate": 0.0001999996568794258,
-      "loss": 46.0,
-      "step": 5188
-    },
-    {
-      "epoch": 0.8356214018277709,
-      "grad_norm": 0.001423538546077907,
-      "learning_rate": 0.00019999965674688292,
-      "loss": 46.0,
-      "step": 5189
-    },
-    {
-      "epoch": 0.8357824389065582,
-      "grad_norm": 0.0016784424660727382,
-      "learning_rate": 0.00019999965661431446,
-      "loss": 46.0,
-      "step": 5190
-    },
-    {
-      "epoch": 0.8359434759853456,
-      "grad_norm": 0.0017954377690330148,
-      "learning_rate": 0.00019999965648172038,
-      "loss": 46.0,
-      "step": 5191
-    },
-    {
-      "epoch": 0.836104513064133,
-      "grad_norm": 0.0004234558145981282,
-      "learning_rate": 0.00019999965634910071,
-      "loss": 46.0,
-      "step": 5192
-    },
-    {
-      "epoch": 0.8362655501429204,
-      "grad_norm": 0.00044127603177912533,
-      "learning_rate": 0.00019999965621645546,
-      "loss": 46.0,
-      "step": 5193
-    },
-    {
-      "epoch": 0.8364265872217078,
-      "grad_norm": 0.0005068947793915868,
-      "learning_rate": 0.00019999965608378462,
-      "loss": 46.0,
-      "step": 5194
-    },
-    {
-      "epoch": 0.8365876243004952,
-      "grad_norm": 0.0034015923738479614,
-      "learning_rate": 0.00019999965595108817,
-      "loss": 46.0,
-      "step": 5195
-    },
-    {
-      "epoch": 0.8367486613792826,
-      "grad_norm": 0.001955636776983738,
-      "learning_rate": 0.00019999965581836613,
-      "loss": 46.0,
-      "step": 5196
-    },
-    {
-      "epoch": 0.83690969845807,
-      "grad_norm": 0.0012738144723698497,
-      "learning_rate": 0.0001999996556856185,
-      "loss": 46.0,
-      "step": 5197
-    },
-    {
-      "epoch": 0.8370707355368574,
-      "grad_norm": 0.0005093156942166388,
-      "learning_rate": 0.00019999965555284526,
-      "loss": 46.0,
-      "step": 5198
-    },
-    {
-      "epoch": 0.8372317726156447,
-      "grad_norm": 0.0020549949258565903,
-      "learning_rate": 0.00019999965542004644,
-      "loss": 46.0,
-      "step": 5199
-    },
-    {
-      "epoch": 0.8373928096944322,
-      "grad_norm": 0.0015045731561258435,
-      "learning_rate": 0.00019999965528722202,
-      "loss": 46.0,
-      "step": 5200
-    },
-    {
-      "epoch": 0.8375538467732195,
-      "grad_norm": 0.0009060234297066927,
-      "learning_rate": 0.000199999655154372,
-      "loss": 46.0,
-      "step": 5201
-    },
-    {
-      "epoch": 0.8377148838520069,
-      "grad_norm": 0.0005925034638494253,
-      "learning_rate": 0.00019999965502149637,
-      "loss": 46.0,
-      "step": 5202
-    },
-    {
-      "epoch": 0.8378759209307943,
-      "grad_norm": 0.0013877975288778543,
-      "learning_rate": 0.00019999965488859517,
-      "loss": 46.0,
-      "step": 5203
-    },
-    {
-      "epoch": 0.8380369580095817,
-      "grad_norm": 0.0007647479069419205,
-      "learning_rate": 0.0001999996547556684,
-      "loss": 46.0,
-      "step": 5204
-    },
-    {
-      "epoch": 0.8381979950883691,
-      "grad_norm": 0.0012101917527616024,
-      "learning_rate": 0.000199999654622716,
-      "loss": 46.0,
-      "step": 5205
-    },
-    {
-      "epoch": 0.8383590321671565,
-      "grad_norm": 0.0013114619068801403,
-      "learning_rate": 0.000199999654489738,
-      "loss": 46.0,
-      "step": 5206
-    },
-    {
-      "epoch": 0.8385200692459439,
-      "grad_norm": 0.0016159799415618181,
-      "learning_rate": 0.00019999965435673443,
-      "loss": 46.0,
-      "step": 5207
-    },
-    {
-      "epoch": 0.8386811063247313,
-      "grad_norm": 0.0010230635525658727,
-      "learning_rate": 0.00019999965422370526,
-      "loss": 46.0,
-      "step": 5208
-    },
-    {
-      "epoch": 0.8388421434035187,
-      "grad_norm": 0.001263906480744481,
-      "learning_rate": 0.00019999965409065048,
-      "loss": 46.0,
-      "step": 5209
-    },
-    {
-      "epoch": 0.839003180482306,
-      "grad_norm": 0.002087854780256748,
-      "learning_rate": 0.0001999996539575701,
-      "loss": 46.0,
-      "step": 5210
-    },
-    {
-      "epoch": 0.8391642175610935,
-      "grad_norm": 0.0005869521992281079,
-      "learning_rate": 0.00019999965382446413,
-      "loss": 46.0,
-      "step": 5211
-    },
-    {
-      "epoch": 0.8393252546398808,
-      "grad_norm": 0.00035122825647704303,
-      "learning_rate": 0.00019999965369133258,
-      "loss": 46.0,
-      "step": 5212
-    },
-    {
-      "epoch": 0.8394862917186682,
-      "grad_norm": 0.0005226124776527286,
-      "learning_rate": 0.00019999965355817543,
-      "loss": 46.0,
-      "step": 5213
-    },
-    {
-      "epoch": 0.8396473287974556,
-      "grad_norm": 0.0019003982888534665,
-      "learning_rate": 0.00019999965342499268,
-      "loss": 46.0,
-      "step": 5214
-    },
-    {
-      "epoch": 0.839808365876243,
-      "grad_norm": 0.0009372249478474259,
-      "learning_rate": 0.00019999965329178435,
-      "loss": 46.0,
-      "step": 5215
-    },
-    {
-      "epoch": 0.8399694029550304,
-      "grad_norm": 0.002607081551104784,
-      "learning_rate": 0.0001999996531585504,
-      "loss": 46.0,
-      "step": 5216
-    },
-    {
-      "epoch": 0.8401304400338178,
-      "grad_norm": 0.0017386317485943437,
-      "learning_rate": 0.00019999965302529087,
-      "loss": 46.0,
-      "step": 5217
-    },
-    {
-      "epoch": 0.8402914771126052,
-      "grad_norm": 0.0016375096747651696,
-      "learning_rate": 0.00019999965289200575,
-      "loss": 46.0,
-      "step": 5218
-    },
-    {
-      "epoch": 0.8404525141913926,
-      "grad_norm": 0.0007120749796740711,
-      "learning_rate": 0.00019999965275869502,
-      "loss": 46.0,
-      "step": 5219
-    },
-    {
-      "epoch": 0.84061355127018,
-      "grad_norm": 0.0006274740444496274,
-      "learning_rate": 0.0001999996526253587,
-      "loss": 46.0,
-      "step": 5220
-    },
-    {
-      "epoch": 0.8407745883489673,
-      "grad_norm": 0.000432023371104151,
-      "learning_rate": 0.0001999996524919968,
-      "loss": 46.0,
-      "step": 5221
-    },
-    {
-      "epoch": 0.8409356254277548,
-      "grad_norm": 0.0038063465617597103,
-      "learning_rate": 0.0001999996523586093,
-      "loss": 46.0,
-      "step": 5222
-    },
-    {
-      "epoch": 0.8410966625065421,
-      "grad_norm": 0.00364326243288815,
-      "learning_rate": 0.00019999965222519618,
-      "loss": 46.0,
-      "step": 5223
-    },
-    {
-      "epoch": 0.8412576995853295,
-      "grad_norm": 0.0008819945505820215,
-      "learning_rate": 0.00019999965209175748,
-      "loss": 46.0,
-      "step": 5224
-    },
-    {
-      "epoch": 0.8414187366641169,
-      "grad_norm": 0.0007538118516094983,
-      "learning_rate": 0.0001999996519582932,
-      "loss": 46.0,
-      "step": 5225
-    },
-    {
-      "epoch": 0.8415797737429043,
-      "grad_norm": 0.0011087398743256927,
-      "learning_rate": 0.00019999965182480333,
-      "loss": 46.0,
-      "step": 5226
-    },
-    {
-      "epoch": 0.8417408108216917,
-      "grad_norm": 0.0004935208708047867,
-      "learning_rate": 0.00019999965169128781,
-      "loss": 46.0,
-      "step": 5227
-    },
-    {
-      "epoch": 0.8419018479004791,
-      "grad_norm": 0.0019165939884260297,
-      "learning_rate": 0.00019999965155774677,
-      "loss": 46.0,
-      "step": 5228
-    },
-    {
-      "epoch": 0.8420628849792665,
-      "grad_norm": 0.0011473592603579164,
-      "learning_rate": 0.0001999996514241801,
-      "loss": 46.0,
-      "step": 5229
-    },
-    {
-      "epoch": 0.8422239220580539,
-      "grad_norm": 0.0012515212874859571,
-      "learning_rate": 0.0001999996512905878,
-      "loss": 46.0,
-      "step": 5230
-    },
-    {
-      "epoch": 0.8423849591368413,
-      "grad_norm": 0.00712304562330246,
-      "learning_rate": 0.00019999965115697,
-      "loss": 46.0,
-      "step": 5231
-    },
-    {
-      "epoch": 0.8425459962156286,
-      "grad_norm": 0.002442289376631379,
-      "learning_rate": 0.00019999965102332652,
-      "loss": 46.0,
-      "step": 5232
-    },
-    {
-      "epoch": 0.842707033294416,
-      "grad_norm": 0.0008963527507148683,
-      "learning_rate": 0.00019999965088965746,
-      "loss": 46.0,
-      "step": 5233
-    },
-    {
-      "epoch": 0.8428680703732034,
-      "grad_norm": 0.001386677147820592,
-      "learning_rate": 0.00019999965075596284,
-      "loss": 46.0,
-      "step": 5234
-    },
-    {
-      "epoch": 0.8430291074519908,
-      "grad_norm": 0.0009650291176512837,
-      "learning_rate": 0.0001999996506222426,
-      "loss": 46.0,
-      "step": 5235
-    },
-    {
-      "epoch": 0.8431901445307782,
-      "grad_norm": 0.0004423398931976408,
-      "learning_rate": 0.00019999965048849675,
-      "loss": 46.0,
-      "step": 5236
-    },
-    {
-      "epoch": 0.8433511816095656,
-      "grad_norm": 0.00396422715857625,
-      "learning_rate": 0.0001999996503547253,
-      "loss": 46.0,
-      "step": 5237
-    },
-    {
-      "epoch": 0.843512218688353,
-      "grad_norm": 0.0008061259286478162,
-      "learning_rate": 0.0001999996502209283,
-      "loss": 46.0,
-      "step": 5238
-    },
-    {
-      "epoch": 0.8436732557671404,
-      "grad_norm": 0.004156382754445076,
-      "learning_rate": 0.0001999996500871057,
-      "loss": 46.0,
-      "step": 5239
-    },
-    {
-      "epoch": 0.8438342928459278,
-      "grad_norm": 0.0015063994796946645,
-      "learning_rate": 0.0001999996499532575,
-      "loss": 46.0,
-      "step": 5240
-    },
-    {
-      "epoch": 0.8439953299247152,
-      "grad_norm": 0.0027145070489495993,
-      "learning_rate": 0.00019999964981938368,
-      "loss": 46.0,
-      "step": 5241
-    },
-    {
-      "epoch": 0.8441563670035026,
-      "grad_norm": 0.0008000176749192178,
-      "learning_rate": 0.00019999964968548428,
-      "loss": 46.0,
-      "step": 5242
-    },
-    {
-      "epoch": 0.84431740408229,
-      "grad_norm": 0.0041970014572143555,
-      "learning_rate": 0.0001999996495515593,
-      "loss": 46.0,
-      "step": 5243
-    },
-    {
-      "epoch": 0.8444784411610773,
-      "grad_norm": 0.0006794112850911915,
-      "learning_rate": 0.00019999964941760869,
-      "loss": 46.0,
-      "step": 5244
-    },
-    {
-      "epoch": 0.8446394782398647,
-      "grad_norm": 0.0006485472549684346,
-      "learning_rate": 0.0001999996492836325,
-      "loss": 46.0,
-      "step": 5245
-    },
-    {
-      "epoch": 0.8448005153186521,
-      "grad_norm": 0.0009412890649400651,
-      "learning_rate": 0.00019999964914963072,
-      "loss": 46.0,
-      "step": 5246
-    },
-    {
-      "epoch": 0.8449615523974395,
-      "grad_norm": 0.0013822100590914488,
-      "learning_rate": 0.00019999964901560335,
-      "loss": 46.0,
-      "step": 5247
-    },
-    {
-      "epoch": 0.8451225894762269,
-      "grad_norm": 0.0021872129291296005,
-      "learning_rate": 0.00019999964888155037,
-      "loss": 46.0,
-      "step": 5248
-    },
-    {
-      "epoch": 0.8452836265550143,
-      "grad_norm": 0.0002928603789769113,
-      "learning_rate": 0.0001999996487474718,
-      "loss": 46.0,
-      "step": 5249
-    },
-    {
-      "epoch": 0.8454446636338017,
-      "grad_norm": 0.0018276500049978495,
-      "learning_rate": 0.00019999964861336765,
-      "loss": 46.0,
-      "step": 5250
-    },
-    {
-      "epoch": 0.8456057007125891,
-      "grad_norm": 0.005268497858196497,
-      "learning_rate": 0.0001999996484792379,
-      "loss": 46.0,
-      "step": 5251
-    },
-    {
-      "epoch": 0.8457667377913765,
-      "grad_norm": 0.00033274025190621614,
-      "learning_rate": 0.00019999964834508256,
-      "loss": 46.0,
-      "step": 5252
-    },
-    {
-      "epoch": 0.8459277748701639,
-      "grad_norm": 0.0003299397649243474,
-      "learning_rate": 0.00019999964821090162,
-      "loss": 46.0,
-      "step": 5253
-    },
-    {
-      "epoch": 0.8460888119489512,
-      "grad_norm": 0.0012615256709977984,
-      "learning_rate": 0.00019999964807669506,
-      "loss": 46.0,
-      "step": 5254
-    },
-    {
-      "epoch": 0.8462498490277386,
-      "grad_norm": 0.0011292611015960574,
-      "learning_rate": 0.00019999964794246294,
-      "loss": 46.0,
-      "step": 5255
-    },
-    {
-      "epoch": 0.846410886106526,
-      "grad_norm": 0.00420538242906332,
-      "learning_rate": 0.0001999996478082052,
-      "loss": 46.0,
-      "step": 5256
-    },
-    {
-      "epoch": 0.8465719231853134,
-      "grad_norm": 0.0014823495876044035,
-      "learning_rate": 0.00019999964767392187,
-      "loss": 46.0,
-      "step": 5257
-    },
-    {
-      "epoch": 0.8467329602641008,
-      "grad_norm": 0.0019098966149613261,
-      "learning_rate": 0.00019999964753961296,
-      "loss": 46.0,
-      "step": 5258
-    },
-    {
-      "epoch": 0.8468939973428882,
-      "grad_norm": 0.0024737513158470392,
-      "learning_rate": 0.00019999964740527847,
-      "loss": 46.0,
-      "step": 5259
-    },
-    {
-      "epoch": 0.8470550344216756,
-      "grad_norm": 0.0009053554385900497,
-      "learning_rate": 0.00019999964727091834,
-      "loss": 46.0,
-      "step": 5260
-    },
-    {
-      "epoch": 0.847216071500463,
-      "grad_norm": 0.0014724271604791284,
-      "learning_rate": 0.00019999964713653264,
-      "loss": 46.0,
-      "step": 5261
-    },
-    {
-      "epoch": 0.8473771085792504,
-      "grad_norm": 0.0046810791827738285,
-      "learning_rate": 0.00019999964700212136,
-      "loss": 46.0,
-      "step": 5262
-    },
-    {
-      "epoch": 0.8475381456580378,
-      "grad_norm": 0.0009719933732412755,
-      "learning_rate": 0.00019999964686768447,
-      "loss": 46.0,
-      "step": 5263
-    },
-    {
-      "epoch": 0.8476991827368252,
-      "grad_norm": 0.0030296803452074528,
-      "learning_rate": 0.00019999964673322196,
-      "loss": 46.0,
-      "step": 5264
-    },
-    {
-      "epoch": 0.8478602198156125,
-      "grad_norm": 0.0014079560060054064,
-      "learning_rate": 0.00019999964659873389,
-      "loss": 46.0,
-      "step": 5265
-    },
-    {
-      "epoch": 0.8480212568943999,
-      "grad_norm": 0.002824049210175872,
-      "learning_rate": 0.0001999996464642202,
-      "loss": 46.0,
-      "step": 5266
-    },
-    {
-      "epoch": 0.8481822939731873,
-      "grad_norm": 0.0005882881814613938,
-      "learning_rate": 0.00019999964632968093,
-      "loss": 46.0,
-      "step": 5267
-    },
-    {
-      "epoch": 0.8483433310519747,
-      "grad_norm": 0.0012592347338795662,
-      "learning_rate": 0.00019999964619511607,
-      "loss": 46.0,
-      "step": 5268
-    },
-    {
-      "epoch": 0.8485043681307621,
-      "grad_norm": 0.0007983875111676753,
-      "learning_rate": 0.0001999996460605256,
-      "loss": 46.0,
-      "step": 5269
-    },
-    {
-      "epoch": 0.8486654052095495,
-      "grad_norm": 0.002135386923328042,
-      "learning_rate": 0.00019999964592590957,
-      "loss": 46.0,
-      "step": 5270
-    },
-    {
-      "epoch": 0.8488264422883369,
-      "grad_norm": 0.0006917768623679876,
-      "learning_rate": 0.00019999964579126792,
-      "loss": 46.0,
-      "step": 5271
-    },
-    {
-      "epoch": 0.8489874793671243,
-      "grad_norm": 0.0004974614130333066,
-      "learning_rate": 0.0001999996456566007,
-      "loss": 46.0,
-      "step": 5272
-    },
-    {
-      "epoch": 0.8491485164459117,
-      "grad_norm": 0.0021558695007115602,
-      "learning_rate": 0.00019999964552190784,
-      "loss": 46.0,
-      "step": 5273
-    },
-    {
-      "epoch": 0.8493095535246991,
-      "grad_norm": 0.0007196618244051933,
-      "learning_rate": 0.00019999964538718938,
-      "loss": 46.0,
-      "step": 5274
-    },
-    {
-      "epoch": 0.8494705906034864,
-      "grad_norm": 0.001629539649002254,
-      "learning_rate": 0.00019999964525244535,
-      "loss": 46.0,
-      "step": 5275
-    },
-    {
-      "epoch": 0.8496316276822738,
-      "grad_norm": 0.0021558755543082952,
-      "learning_rate": 0.00019999964511767574,
-      "loss": 46.0,
-      "step": 5276
-    },
-    {
-      "epoch": 0.8497926647610612,
-      "grad_norm": 0.002615951234474778,
-      "learning_rate": 0.00019999964498288052,
-      "loss": 46.0,
-      "step": 5277
-    },
-    {
-      "epoch": 0.8499537018398486,
-      "grad_norm": 0.0005678863381035626,
-      "learning_rate": 0.0001999996448480597,
-      "loss": 46.0,
-      "step": 5278
-    },
-    {
-      "epoch": 0.850114738918636,
-      "grad_norm": 0.004259361419826746,
-      "learning_rate": 0.0001999996447132133,
-      "loss": 46.0,
-      "step": 5279
-    },
-    {
-      "epoch": 0.8502757759974234,
-      "grad_norm": 0.0012753693154081702,
-      "learning_rate": 0.0001999996445783413,
-      "loss": 46.0,
-      "step": 5280
-    },
-    {
-      "epoch": 0.8504368130762108,
-      "grad_norm": 0.001906612771563232,
-      "learning_rate": 0.0001999996444434437,
-      "loss": 46.0,
-      "step": 5281
-    },
-    {
-      "epoch": 0.8505978501549982,
-      "grad_norm": 0.0013399564195424318,
-      "learning_rate": 0.0001999996443085205,
-      "loss": 46.0,
-      "step": 5282
-    },
-    {
-      "epoch": 0.8507588872337856,
-      "grad_norm": 0.0012392661301419139,
-      "learning_rate": 0.00019999964417357174,
-      "loss": 46.0,
-      "step": 5283
-    },
-    {
-      "epoch": 0.850919924312573,
-      "grad_norm": 0.0004138743388466537,
-      "learning_rate": 0.00019999964403859735,
-      "loss": 46.0,
-      "step": 5284
-    },
-    {
-      "epoch": 0.8510809613913604,
-      "grad_norm": 0.0015947495121508837,
-      "learning_rate": 0.00019999964390359737,
-      "loss": 46.0,
-      "step": 5285
-    },
-    {
-      "epoch": 0.8512419984701477,
-      "grad_norm": 0.0023456921335309744,
-      "learning_rate": 0.00019999964376857178,
-      "loss": 46.0,
-      "step": 5286
-    },
-    {
-      "epoch": 0.8514030355489351,
-      "grad_norm": 0.0006762603879906237,
-      "learning_rate": 0.0001999996436335206,
-      "loss": 46.0,
-      "step": 5287
-    },
-    {
-      "epoch": 0.8515640726277225,
-      "grad_norm": 0.0011161736911162734,
-      "learning_rate": 0.00019999964349844387,
-      "loss": 46.0,
-      "step": 5288
-    },
-    {
-      "epoch": 0.8517251097065099,
-      "grad_norm": 0.0011126425815746188,
-      "learning_rate": 0.00019999964336334152,
-      "loss": 46.0,
-      "step": 5289
-    },
-    {
-      "epoch": 0.8518861467852973,
-      "grad_norm": 0.0007951778243295848,
-      "learning_rate": 0.00019999964322821355,
-      "loss": 46.0,
-      "step": 5290
-    },
-    {
-      "epoch": 0.8520471838640847,
-      "grad_norm": 0.0037739980034530163,
-      "learning_rate": 0.00019999964309306,
-      "loss": 46.0,
-      "step": 5291
-    },
-    {
-      "epoch": 0.8522082209428721,
-      "grad_norm": 0.0006727104191668332,
-      "learning_rate": 0.00019999964295788089,
-      "loss": 46.0,
-      "step": 5292
-    },
-    {
-      "epoch": 0.8523692580216595,
-      "grad_norm": 0.0008476900984533131,
-      "learning_rate": 0.00019999964282267616,
-      "loss": 46.0,
-      "step": 5293
-    },
-    {
-      "epoch": 0.8525302951004469,
-      "grad_norm": 0.0014890005113556981,
-      "learning_rate": 0.0001999996426874458,
-      "loss": 46.0,
-      "step": 5294
-    },
-    {
-      "epoch": 0.8526913321792343,
-      "grad_norm": 0.0024564142804592848,
-      "learning_rate": 0.00019999964255218992,
-      "loss": 46.0,
-      "step": 5295
-    },
-    {
-      "epoch": 0.8528523692580217,
-      "grad_norm": 0.003224824322387576,
-      "learning_rate": 0.00019999964241690837,
-      "loss": 46.0,
-      "step": 5296
-    },
-    {
-      "epoch": 0.853013406336809,
-      "grad_norm": 0.0016472613206133246,
-      "learning_rate": 0.00019999964228160127,
-      "loss": 46.0,
-      "step": 5297
-    },
-    {
-      "epoch": 0.8531744434155965,
-      "grad_norm": 0.001739516039378941,
-      "learning_rate": 0.00019999964214626855,
-      "loss": 46.0,
-      "step": 5298
-    },
-    {
-      "epoch": 0.8533354804943838,
-      "grad_norm": 0.0007754797115921974,
-      "learning_rate": 0.00019999964201091025,
-      "loss": 46.0,
-      "step": 5299
-    },
-    {
-      "epoch": 0.8534965175731712,
-      "grad_norm": 0.0010283966548740864,
-      "learning_rate": 0.00019999964187552633,
-      "loss": 46.0,
-      "step": 5300
-    },
-    {
-      "epoch": 0.8536575546519586,
-      "grad_norm": 0.0010554902255535126,
-      "learning_rate": 0.00019999964174011685,
-      "loss": 46.0,
-      "step": 5301
-    },
-    {
-      "epoch": 0.853818591730746,
-      "grad_norm": 0.0022285226732492447,
-      "learning_rate": 0.00019999964160468176,
-      "loss": 46.0,
-      "step": 5302
-    },
-    {
-      "epoch": 0.8539796288095334,
-      "grad_norm": 0.0018978987354785204,
-      "learning_rate": 0.00019999964146922108,
-      "loss": 46.0,
-      "step": 5303
-    },
-    {
-      "epoch": 0.8541406658883208,
-      "grad_norm": 0.0016012756386771798,
-      "learning_rate": 0.0001999996413337348,
-      "loss": 46.0,
-      "step": 5304
-    },
-    {
-      "epoch": 0.8543017029671082,
-      "grad_norm": 0.0010058789048343897,
-      "learning_rate": 0.00019999964119822293,
-      "loss": 46.0,
-      "step": 5305
-    },
-    {
-      "epoch": 0.8544627400458956,
-      "grad_norm": 0.002617072081193328,
-      "learning_rate": 0.00019999964106268546,
-      "loss": 46.0,
-      "step": 5306
-    },
-    {
-      "epoch": 0.854623777124683,
-      "grad_norm": 0.0012582477647811174,
-      "learning_rate": 0.0001999996409271224,
-      "loss": 46.0,
-      "step": 5307
-    },
-    {
-      "epoch": 0.8547848142034703,
-      "grad_norm": 0.0003376406675670296,
-      "learning_rate": 0.00019999964079153376,
-      "loss": 46.0,
-      "step": 5308
-    },
-    {
-      "epoch": 0.8549458512822578,
-      "grad_norm": 0.006716587580740452,
-      "learning_rate": 0.0001999996406559195,
-      "loss": 46.0,
-      "step": 5309
-    },
-    {
-      "epoch": 0.8551068883610451,
-      "grad_norm": 0.0012386812595650554,
-      "learning_rate": 0.00019999964052027966,
-      "loss": 46.0,
-      "step": 5310
-    },
-    {
-      "epoch": 0.8552679254398325,
-      "grad_norm": 0.0006290592718869448,
-      "learning_rate": 0.0001999996403846142,
-      "loss": 46.0,
-      "step": 5311
-    },
-    {
-      "epoch": 0.8554289625186199,
-      "grad_norm": 0.00058881810400635,
-      "learning_rate": 0.00019999964024892318,
-      "loss": 46.0,
-      "step": 5312
-    },
-    {
-      "epoch": 0.8555899995974073,
-      "grad_norm": 0.0008410885347984731,
-      "learning_rate": 0.00019999964011320652,
-      "loss": 46.0,
-      "step": 5313
-    },
-    {
-      "epoch": 0.8557510366761947,
-      "grad_norm": 0.0017486875876784325,
-      "learning_rate": 0.00019999963997746433,
-      "loss": 46.0,
-      "step": 5314
-    },
-    {
-      "epoch": 0.8559120737549821,
-      "grad_norm": 0.006869988515973091,
-      "learning_rate": 0.0001999996398416965,
-      "loss": 46.0,
-      "step": 5315
-    },
-    {
-      "epoch": 0.8560731108337695,
-      "grad_norm": 0.0015343126142397523,
-      "learning_rate": 0.0001999996397059031,
-      "loss": 46.0,
-      "step": 5316
-    },
-    {
-      "epoch": 0.8562341479125569,
-      "grad_norm": 0.0007613686611875892,
-      "learning_rate": 0.00019999963957008406,
-      "loss": 46.0,
-      "step": 5317
-    },
-    {
-      "epoch": 0.8563951849913443,
-      "grad_norm": 0.0009837878169491887,
-      "learning_rate": 0.00019999963943423947,
-      "loss": 46.0,
-      "step": 5318
-    },
-    {
-      "epoch": 0.8565562220701316,
-      "grad_norm": 0.0031392339151352644,
-      "learning_rate": 0.00019999963929836925,
-      "loss": 46.0,
-      "step": 5319
-    },
-    {
-      "epoch": 0.856717259148919,
-      "grad_norm": 0.0007671216153539717,
-      "learning_rate": 0.00019999963916247348,
-      "loss": 46.0,
-      "step": 5320
-    },
-    {
-      "epoch": 0.8568782962277064,
-      "grad_norm": 0.0007769123767502606,
-      "learning_rate": 0.00019999963902655207,
-      "loss": 46.0,
-      "step": 5321
-    },
-    {
-      "epoch": 0.8570393333064938,
-      "grad_norm": 0.0020357626490294933,
-      "learning_rate": 0.0001999996388906051,
-      "loss": 46.0,
-      "step": 5322
-    },
-    {
-      "epoch": 0.8572003703852812,
-      "grad_norm": 0.0008045024587772787,
-      "learning_rate": 0.00019999963875463251,
-      "loss": 46.0,
-      "step": 5323
-    },
-    {
-      "epoch": 0.8573614074640686,
-      "grad_norm": 0.001799550955183804,
-      "learning_rate": 0.00019999963861863437,
-      "loss": 46.0,
-      "step": 5324
-    },
-    {
-      "epoch": 0.857522444542856,
-      "grad_norm": 0.0007439398323185742,
-      "learning_rate": 0.00019999963848261058,
-      "loss": 46.0,
-      "step": 5325
-    },
-    {
-      "epoch": 0.8576834816216434,
-      "grad_norm": 0.0014336802996695042,
-      "learning_rate": 0.0001999996383465612,
-      "loss": 46.0,
-      "step": 5326
-    },
-    {
-      "epoch": 0.8578445187004308,
-      "grad_norm": 0.001388859935104847,
-      "learning_rate": 0.0001999996382104863,
-      "loss": 46.0,
-      "step": 5327
-    },
-    {
-      "epoch": 0.8580055557792181,
-      "grad_norm": 0.000919417361728847,
-      "learning_rate": 0.00019999963807438572,
-      "loss": 46.0,
-      "step": 5328
-    },
-    {
-      "epoch": 0.8581665928580056,
-      "grad_norm": 0.0009685653494670987,
-      "learning_rate": 0.00019999963793825956,
-      "loss": 46.0,
-      "step": 5329
-    },
-    {
-      "epoch": 0.8583276299367929,
-      "grad_norm": 0.0030787650030106306,
-      "learning_rate": 0.00019999963780210783,
-      "loss": 46.0,
-      "step": 5330
-    },
-    {
-      "epoch": 0.8584886670155804,
-      "grad_norm": 0.0017713705310598016,
-      "learning_rate": 0.0001999996376659305,
-      "loss": 46.0,
-      "step": 5331
-    },
-    {
-      "epoch": 0.8586497040943677,
-      "grad_norm": 0.0009224415989592671,
-      "learning_rate": 0.00019999963752972754,
-      "loss": 46.0,
-      "step": 5332
-    },
-    {
-      "epoch": 0.8588107411731551,
-      "grad_norm": 0.0019316660473123193,
-      "learning_rate": 0.00019999963739349906,
-      "loss": 46.0,
-      "step": 5333
-    },
-    {
-      "epoch": 0.8589717782519425,
-      "grad_norm": 0.011059368029236794,
-      "learning_rate": 0.0001999996372572449,
-      "loss": 46.0,
-      "step": 5334
-    },
-    {
-      "epoch": 0.8591328153307299,
-      "grad_norm": 0.0013488434487953782,
-      "learning_rate": 0.0001999996371209652,
-      "loss": 46.0,
-      "step": 5335
-    },
-    {
-      "epoch": 0.8592938524095173,
-      "grad_norm": 0.0007839535246603191,
-      "learning_rate": 0.0001999996369846599,
-      "loss": 46.0,
-      "step": 5336
-    },
-    {
-      "epoch": 0.8594548894883047,
-      "grad_norm": 0.0004955589538440108,
-      "learning_rate": 0.00019999963684832898,
-      "loss": 46.0,
-      "step": 5337
-    },
-    {
-      "epoch": 0.8596159265670921,
-      "grad_norm": 0.0004249998601153493,
-      "learning_rate": 0.00019999963671197247,
-      "loss": 46.0,
-      "step": 5338
-    },
-    {
-      "epoch": 0.8597769636458794,
-      "grad_norm": 0.0005931945634074509,
-      "learning_rate": 0.00019999963657559038,
-      "loss": 46.0,
-      "step": 5339
-    },
-    {
-      "epoch": 0.8599380007246669,
-      "grad_norm": 0.001101514557376504,
-      "learning_rate": 0.00019999963643918268,
-      "loss": 46.0,
-      "step": 5340
-    },
-    {
-      "epoch": 0.8600990378034542,
-      "grad_norm": 0.0017062551341950893,
-      "learning_rate": 0.00019999963630274942,
-      "loss": 46.0,
-      "step": 5341
-    },
-    {
-      "epoch": 0.8602600748822417,
-      "grad_norm": 0.0007795640267431736,
-      "learning_rate": 0.00019999963616629054,
-      "loss": 46.0,
-      "step": 5342
-    },
-    {
-      "epoch": 0.860421111961029,
-      "grad_norm": 0.0008742253994569182,
-      "learning_rate": 0.00019999963602980607,
-      "loss": 46.0,
-      "step": 5343
-    },
-    {
-      "epoch": 0.8605821490398164,
-      "grad_norm": 0.0009085464989766479,
-      "learning_rate": 0.000199999635893296,
-      "loss": 46.0,
-      "step": 5344
-    },
-    {
-      "epoch": 0.8607431861186038,
-      "grad_norm": 0.0014292245032265782,
-      "learning_rate": 0.00019999963575676033,
-      "loss": 46.0,
-      "step": 5345
-    },
-    {
-      "epoch": 0.8609042231973912,
-      "grad_norm": 0.0007941921357996762,
-      "learning_rate": 0.00019999963562019907,
-      "loss": 46.0,
-      "step": 5346
-    },
-    {
-      "epoch": 0.8610652602761786,
-      "grad_norm": 0.002296466613188386,
-      "learning_rate": 0.00019999963548361223,
-      "loss": 46.0,
-      "step": 5347
-    },
-    {
-      "epoch": 0.861226297354966,
-      "grad_norm": 0.001823195954784751,
-      "learning_rate": 0.00019999963534699978,
-      "loss": 46.0,
-      "step": 5348
-    },
-    {
-      "epoch": 0.8613873344337534,
-      "grad_norm": 0.001987638184800744,
-      "learning_rate": 0.00019999963521036173,
-      "loss": 46.0,
-      "step": 5349
-    },
-    {
-      "epoch": 0.8615483715125407,
-      "grad_norm": 0.0005175631958991289,
-      "learning_rate": 0.0001999996350736981,
-      "loss": 46.0,
-      "step": 5350
-    },
-    {
-      "epoch": 0.8617094085913282,
-      "grad_norm": 0.0008064642897807062,
-      "learning_rate": 0.00019999963493700886,
-      "loss": 46.0,
-      "step": 5351
-    },
-    {
-      "epoch": 0.8618704456701155,
-      "grad_norm": 0.0023549997713416815,
-      "learning_rate": 0.00019999963480029403,
-      "loss": 46.0,
-      "step": 5352
-    },
-    {
-      "epoch": 0.862031482748903,
-      "grad_norm": 0.0006986106163822114,
-      "learning_rate": 0.00019999963466355364,
-      "loss": 46.0,
-      "step": 5353
-    },
-    {
-      "epoch": 0.8621925198276903,
-      "grad_norm": 0.0022354733664542437,
-      "learning_rate": 0.0001999996345267876,
-      "loss": 46.0,
-      "step": 5354
-    },
-    {
-      "epoch": 0.8623535569064777,
-      "grad_norm": 0.0006340537802316248,
-      "learning_rate": 0.00019999963438999601,
-      "loss": 46.0,
-      "step": 5355
-    },
-    {
-      "epoch": 0.8625145939852651,
-      "grad_norm": 0.0009771937038749456,
-      "learning_rate": 0.00019999963425317878,
-      "loss": 46.0,
-      "step": 5356
-    },
-    {
-      "epoch": 0.8626756310640525,
-      "grad_norm": 0.0015052348608151078,
-      "learning_rate": 0.00019999963411633599,
-      "loss": 46.0,
-      "step": 5357
-    },
-    {
-      "epoch": 0.8628366681428399,
-      "grad_norm": 0.0006980052567087114,
-      "learning_rate": 0.0001999996339794676,
-      "loss": 46.0,
-      "step": 5358
-    },
-    {
-      "epoch": 0.8629977052216273,
-      "grad_norm": 0.0009572468115948141,
-      "learning_rate": 0.00019999963384257358,
-      "loss": 46.0,
-      "step": 5359
-    },
-    {
-      "epoch": 0.8631587423004147,
-      "grad_norm": 0.00037086111842654645,
-      "learning_rate": 0.000199999633705654,
-      "loss": 46.0,
-      "step": 5360
-    },
-    {
-      "epoch": 0.863319779379202,
-      "grad_norm": 0.0028319223783910275,
-      "learning_rate": 0.00019999963356870883,
-      "loss": 46.0,
-      "step": 5361
-    },
-    {
-      "epoch": 0.8634808164579895,
-      "grad_norm": 0.0009068941581062973,
-      "learning_rate": 0.00019999963343173805,
-      "loss": 46.0,
-      "step": 5362
-    },
-    {
-      "epoch": 0.8636418535367768,
-      "grad_norm": 0.0013134769396856427,
-      "learning_rate": 0.00019999963329474167,
-      "loss": 46.0,
-      "step": 5363
-    },
-    {
-      "epoch": 0.8638028906155643,
-      "grad_norm": 0.0009307610453106463,
-      "learning_rate": 0.00019999963315771972,
-      "loss": 46.0,
-      "step": 5364
-    },
-    {
-      "epoch": 0.8639639276943516,
-      "grad_norm": 0.0004538406792562455,
-      "learning_rate": 0.00019999963302067214,
-      "loss": 46.0,
-      "step": 5365
-    },
-    {
-      "epoch": 0.864124964773139,
-      "grad_norm": 0.0013267614413052797,
-      "learning_rate": 0.000199999632883599,
-      "loss": 46.0,
-      "step": 5366
-    },
-    {
-      "epoch": 0.8642860018519264,
-      "grad_norm": 0.0005370157305151224,
-      "learning_rate": 0.00019999963274650026,
-      "loss": 46.0,
-      "step": 5367
-    },
-    {
-      "epoch": 0.8644470389307138,
-      "grad_norm": 0.0004512017185334116,
-      "learning_rate": 0.0001999996326093759,
-      "loss": 46.0,
-      "step": 5368
-    },
-    {
-      "epoch": 0.8646080760095012,
-      "grad_norm": 0.0007292244699783623,
-      "learning_rate": 0.00019999963247222598,
-      "loss": 46.0,
-      "step": 5369
-    },
-    {
-      "epoch": 0.8647691130882886,
-      "grad_norm": 0.0006310438620857894,
-      "learning_rate": 0.00019999963233505044,
-      "loss": 46.0,
-      "step": 5370
-    },
-    {
-      "epoch": 0.864930150167076,
-      "grad_norm": 0.0023617143742740154,
-      "learning_rate": 0.00019999963219784932,
-      "loss": 46.0,
-      "step": 5371
-    },
-    {
-      "epoch": 0.8650911872458633,
-      "grad_norm": 0.0006030204240232706,
-      "learning_rate": 0.0001999996320606226,
-      "loss": 46.0,
-      "step": 5372
-    },
-    {
-      "epoch": 0.8652522243246508,
-      "grad_norm": 0.0018976028077304363,
-      "learning_rate": 0.0001999996319233703,
-      "loss": 46.0,
-      "step": 5373
-    },
-    {
-      "epoch": 0.8654132614034381,
-      "grad_norm": 0.002421885496005416,
-      "learning_rate": 0.00019999963178609238,
-      "loss": 46.0,
-      "step": 5374
-    },
-    {
-      "epoch": 0.8655742984822256,
-      "grad_norm": 0.0008818595670163631,
-      "learning_rate": 0.00019999963164878888,
-      "loss": 46.0,
-      "step": 5375
-    },
-    {
-      "epoch": 0.8657353355610129,
-      "grad_norm": 0.0030620465986430645,
-      "learning_rate": 0.00019999963151145976,
-      "loss": 46.0,
-      "step": 5376
-    },
-    {
-      "epoch": 0.8658963726398003,
-      "grad_norm": 0.0010807124199345708,
-      "learning_rate": 0.00019999963137410506,
-      "loss": 46.0,
-      "step": 5377
-    },
-    {
-      "epoch": 0.8660574097185877,
-      "grad_norm": 0.005061842035502195,
-      "learning_rate": 0.00019999963123672478,
-      "loss": 46.0,
-      "step": 5378
-    },
-    {
-      "epoch": 0.8662184467973751,
-      "grad_norm": 0.002218588488176465,
-      "learning_rate": 0.0001999996310993189,
-      "loss": 46.0,
-      "step": 5379
-    },
-    {
-      "epoch": 0.8663794838761625,
-      "grad_norm": 0.0028829635120928288,
-      "learning_rate": 0.00019999963096188741,
-      "loss": 46.0,
-      "step": 5380
-    },
-    {
-      "epoch": 0.8665405209549498,
-      "grad_norm": 0.003809395944699645,
-      "learning_rate": 0.00019999963082443034,
-      "loss": 46.0,
-      "step": 5381
-    },
-    {
-      "epoch": 0.8667015580337373,
-      "grad_norm": 0.0029538862872868776,
-      "learning_rate": 0.00019999963068694767,
-      "loss": 46.0,
-      "step": 5382
-    },
-    {
-      "epoch": 0.8668625951125246,
-      "grad_norm": 0.0009006448090076447,
-      "learning_rate": 0.0001999996305494394,
-      "loss": 46.0,
-      "step": 5383
-    },
-    {
-      "epoch": 0.8670236321913121,
-      "grad_norm": 0.001607679994776845,
-      "learning_rate": 0.00019999963041190556,
-      "loss": 46.0,
-      "step": 5384
-    },
-    {
-      "epoch": 0.8671846692700994,
-      "grad_norm": 0.004454975016415119,
-      "learning_rate": 0.00019999963027434608,
-      "loss": 46.0,
-      "step": 5385
-    },
-    {
-      "epoch": 0.8673457063488869,
-      "grad_norm": 0.0012642252258956432,
-      "learning_rate": 0.00019999963013676104,
-      "loss": 46.0,
-      "step": 5386
-    },
-    {
-      "epoch": 0.8675067434276742,
-      "grad_norm": 0.0029488876461982727,
-      "learning_rate": 0.00019999962999915042,
-      "loss": 46.0,
-      "step": 5387
-    },
-    {
-      "epoch": 0.8676677805064616,
-      "grad_norm": 0.00043018089490942657,
-      "learning_rate": 0.00019999962986151415,
-      "loss": 46.0,
-      "step": 5388
-    },
-    {
-      "epoch": 0.867828817585249,
-      "grad_norm": 0.002301212400197983,
-      "learning_rate": 0.00019999962972385232,
-      "loss": 46.0,
-      "step": 5389
-    },
-    {
-      "epoch": 0.8679898546640364,
-      "grad_norm": 0.0005062994314357638,
-      "learning_rate": 0.0001999996295861649,
-      "loss": 46.0,
-      "step": 5390
-    },
-    {
-      "epoch": 0.8681508917428238,
-      "grad_norm": 0.0008245017379522324,
-      "learning_rate": 0.00019999962944845188,
-      "loss": 46.0,
-      "step": 5391
-    },
-    {
-      "epoch": 0.8683119288216111,
-      "grad_norm": 0.0010551703162491322,
-      "learning_rate": 0.00019999962931071326,
-      "loss": 46.0,
-      "step": 5392
-    },
-    {
-      "epoch": 0.8684729659003986,
-      "grad_norm": 0.0013676321832463145,
-      "learning_rate": 0.00019999962917294903,
-      "loss": 46.0,
-      "step": 5393
-    },
-    {
-      "epoch": 0.8686340029791859,
-      "grad_norm": 0.005799941252917051,
-      "learning_rate": 0.00019999962903515924,
-      "loss": 46.0,
-      "step": 5394
-    },
-    {
-      "epoch": 0.8687950400579734,
-      "grad_norm": 0.0008178535499610007,
-      "learning_rate": 0.00019999962889734384,
-      "loss": 46.0,
-      "step": 5395
-    },
-    {
-      "epoch": 0.8689560771367607,
-      "grad_norm": 0.0011396018089726567,
-      "learning_rate": 0.00019999962875950284,
-      "loss": 46.0,
-      "step": 5396
-    },
-    {
-      "epoch": 0.8691171142155482,
-      "grad_norm": 0.0009169162367470562,
-      "learning_rate": 0.00019999962862163627,
-      "loss": 46.0,
-      "step": 5397
-    },
-    {
-      "epoch": 0.8692781512943355,
-      "grad_norm": 0.0007868690881878138,
-      "learning_rate": 0.00019999962848374407,
-      "loss": 46.0,
-      "step": 5398
-    },
-    {
-      "epoch": 0.869439188373123,
-      "grad_norm": 0.0017590143252164125,
-      "learning_rate": 0.0001999996283458263,
-      "loss": 46.0,
-      "step": 5399
-    },
-    {
-      "epoch": 0.8696002254519103,
-      "grad_norm": 0.0032193479128181934,
-      "learning_rate": 0.00019999962820788292,
-      "loss": 46.0,
-      "step": 5400
-    },
-    {
-      "epoch": 0.8697612625306977,
-      "grad_norm": 0.001666118623688817,
-      "learning_rate": 0.00019999962806991394,
-      "loss": 46.0,
-      "step": 5401
-    },
-    {
-      "epoch": 0.8699222996094851,
-      "grad_norm": 0.0012950766831636429,
-      "learning_rate": 0.00019999962793191935,
-      "loss": 46.0,
-      "step": 5402
-    },
-    {
-      "epoch": 0.8700833366882724,
-      "grad_norm": 0.0007219741819426417,
-      "learning_rate": 0.00019999962779389922,
-      "loss": 46.0,
-      "step": 5403
-    },
-    {
-      "epoch": 0.8702443737670599,
-      "grad_norm": 0.0006782657583244145,
-      "learning_rate": 0.00019999962765585345,
-      "loss": 46.0,
-      "step": 5404
-    },
-    {
-      "epoch": 0.8704054108458472,
-      "grad_norm": 0.0006449512438848615,
-      "learning_rate": 0.0001999996275177821,
-      "loss": 46.0,
-      "step": 5405
-    },
-    {
-      "epoch": 0.8705664479246347,
-      "grad_norm": 0.0008436316275037825,
-      "learning_rate": 0.00019999962737968517,
-      "loss": 46.0,
-      "step": 5406
-    },
-    {
-      "epoch": 0.870727485003422,
-      "grad_norm": 0.00047500242362730205,
-      "learning_rate": 0.0001999996272415626,
-      "loss": 46.0,
-      "step": 5407
-    },
-    {
-      "epoch": 0.8708885220822095,
-      "grad_norm": 0.0009098974987864494,
-      "learning_rate": 0.0001999996271034145,
-      "loss": 46.0,
-      "step": 5408
-    },
-    {
-      "epoch": 0.8710495591609968,
-      "grad_norm": 0.0013777875574305654,
-      "learning_rate": 0.00019999962696524076,
-      "loss": 46.0,
-      "step": 5409
-    },
-    {
-      "epoch": 0.8712105962397843,
-      "grad_norm": 0.0009028459899127483,
-      "learning_rate": 0.00019999962682704144,
-      "loss": 46.0,
-      "step": 5410
-    },
-    {
-      "epoch": 0.8713716333185716,
-      "grad_norm": 0.0035416826140135527,
-      "learning_rate": 0.0001999996266888165,
-      "loss": 46.0,
-      "step": 5411
-    },
-    {
-      "epoch": 0.871532670397359,
-      "grad_norm": 0.00044533328036777675,
-      "learning_rate": 0.000199999626550566,
-      "loss": 46.0,
-      "step": 5412
-    },
-    {
-      "epoch": 0.8716937074761464,
-      "grad_norm": 0.0020277979783713818,
-      "learning_rate": 0.0001999996264122899,
-      "loss": 46.0,
-      "step": 5413
-    },
-    {
-      "epoch": 0.8718547445549337,
-      "grad_norm": 0.0010008830577135086,
-      "learning_rate": 0.00019999962627398818,
-      "loss": 46.0,
-      "step": 5414
-    },
-    {
-      "epoch": 0.8720157816337212,
-      "grad_norm": 0.0005248826346360147,
-      "learning_rate": 0.0001999996261356609,
-      "loss": 46.0,
-      "step": 5415
-    },
-    {
-      "epoch": 0.8721768187125085,
-      "grad_norm": 0.000740030431188643,
-      "learning_rate": 0.000199999625997308,
-      "loss": 46.0,
-      "step": 5416
-    },
-    {
-      "epoch": 0.872337855791296,
-      "grad_norm": 0.0006334640784189105,
-      "learning_rate": 0.00019999962585892954,
-      "loss": 46.0,
-      "step": 5417
-    },
-    {
-      "epoch": 0.8724988928700833,
-      "grad_norm": 0.0008111082715913653,
-      "learning_rate": 0.00019999962572052544,
-      "loss": 46.0,
-      "step": 5418
-    },
-    {
-      "epoch": 0.8726599299488708,
-      "grad_norm": 0.0012994561111554503,
-      "learning_rate": 0.00019999962558209578,
-      "loss": 46.0,
-      "step": 5419
-    },
-    {
-      "epoch": 0.8728209670276581,
-      "grad_norm": 0.0011322868522256613,
-      "learning_rate": 0.00019999962544364048,
-      "loss": 46.0,
-      "step": 5420
-    },
-    {
-      "epoch": 0.8729820041064456,
-      "grad_norm": 0.0005961346323601902,
-      "learning_rate": 0.00019999962530515961,
-      "loss": 46.0,
-      "step": 5421
-    },
-    {
-      "epoch": 0.8731430411852329,
-      "grad_norm": 0.005132320336997509,
-      "learning_rate": 0.00019999962516665317,
-      "loss": 46.0,
-      "step": 5422
-    },
-    {
-      "epoch": 0.8733040782640203,
-      "grad_norm": 0.0015798952663317323,
-      "learning_rate": 0.0001999996250281211,
-      "loss": 46.0,
-      "step": 5423
-    },
-    {
-      "epoch": 0.8734651153428077,
-      "grad_norm": 0.0006656598416157067,
-      "learning_rate": 0.00019999962488956345,
-      "loss": 46.0,
-      "step": 5424
-    },
-    {
-      "epoch": 0.873626152421595,
-      "grad_norm": 0.001495807315222919,
-      "learning_rate": 0.0001999996247509802,
-      "loss": 46.0,
-      "step": 5425
-    },
-    {
-      "epoch": 0.8737871895003825,
-      "grad_norm": 0.0006847336771897972,
-      "learning_rate": 0.0001999996246123714,
-      "loss": 46.0,
-      "step": 5426
-    },
-    {
-      "epoch": 0.8739482265791698,
-      "grad_norm": 0.0017865067347884178,
-      "learning_rate": 0.00019999962447373696,
-      "loss": 46.0,
-      "step": 5427
-    },
-    {
-      "epoch": 0.8741092636579573,
-      "grad_norm": 0.0022510443814098835,
-      "learning_rate": 0.0001999996243350769,
-      "loss": 46.0,
-      "step": 5428
-    },
-    {
-      "epoch": 0.8742703007367446,
-      "grad_norm": 0.002816633088514209,
-      "learning_rate": 0.0001999996241963913,
-      "loss": 46.0,
-      "step": 5429
-    },
-    {
-      "epoch": 0.8744313378155321,
-      "grad_norm": 0.0013194011989980936,
-      "learning_rate": 0.00019999962405768006,
-      "loss": 46.0,
-      "step": 5430
-    },
-    {
-      "epoch": 0.8745923748943194,
-      "grad_norm": 0.0011149300262331963,
-      "learning_rate": 0.00019999962391894325,
-      "loss": 46.0,
-      "step": 5431
-    },
-    {
-      "epoch": 0.8747534119731069,
-      "grad_norm": 0.0005363540258258581,
-      "learning_rate": 0.00019999962378018085,
-      "loss": 46.0,
-      "step": 5432
-    },
-    {
-      "epoch": 0.8749144490518942,
-      "grad_norm": 0.0017888746224343777,
-      "learning_rate": 0.00019999962364139283,
-      "loss": 46.0,
-      "step": 5433
-    },
-    {
-      "epoch": 0.8750754861306815,
-      "grad_norm": 0.002010643482208252,
-      "learning_rate": 0.00019999962350257923,
-      "loss": 46.0,
-      "step": 5434
-    },
-    {
-      "epoch": 0.875236523209469,
-      "grad_norm": 0.00449001369997859,
-      "learning_rate": 0.00019999962336374004,
-      "loss": 46.0,
-      "step": 5435
-    },
-    {
-      "epoch": 0.8753975602882563,
-      "grad_norm": 0.0009560094913467765,
-      "learning_rate": 0.00019999962322487527,
-      "loss": 46.0,
-      "step": 5436
-    },
-    {
-      "epoch": 0.8755585973670438,
-      "grad_norm": 0.0014784683007746935,
-      "learning_rate": 0.00019999962308598487,
-      "loss": 46.0,
-      "step": 5437
-    },
-    {
-      "epoch": 0.8757196344458311,
-      "grad_norm": 0.0007613886846229434,
-      "learning_rate": 0.00019999962294706892,
-      "loss": 46.0,
-      "step": 5438
-    },
-    {
-      "epoch": 0.8758806715246186,
-      "grad_norm": 0.0017266601789742708,
-      "learning_rate": 0.00019999962280812733,
-      "loss": 46.0,
-      "step": 5439
-    },
-    {
-      "epoch": 0.8760417086034059,
-      "grad_norm": 0.002928278874605894,
-      "learning_rate": 0.00019999962266916015,
-      "loss": 46.0,
-      "step": 5440
-    },
-    {
-      "epoch": 0.8762027456821934,
-      "grad_norm": 0.0007385804201476276,
-      "learning_rate": 0.00019999962253016738,
-      "loss": 46.0,
-      "step": 5441
-    },
-    {
-      "epoch": 0.8763637827609807,
-      "grad_norm": 0.0007163072004914284,
-      "learning_rate": 0.00019999962239114906,
-      "loss": 46.0,
-      "step": 5442
-    },
-    {
-      "epoch": 0.8765248198397682,
-      "grad_norm": 0.00032691159867681563,
-      "learning_rate": 0.0001999996222521051,
-      "loss": 46.0,
-      "step": 5443
-    },
-    {
-      "epoch": 0.8766858569185555,
-      "grad_norm": 0.0009719034424051642,
-      "learning_rate": 0.00019999962211303553,
-      "loss": 46.0,
-      "step": 5444
-    },
-    {
-      "epoch": 0.8768468939973428,
-      "grad_norm": 0.00503349956125021,
-      "learning_rate": 0.0001999996219739404,
-      "loss": 46.0,
-      "step": 5445
-    },
-    {
-      "epoch": 0.8770079310761303,
-      "grad_norm": 0.0019561199005693197,
-      "learning_rate": 0.0001999996218348197,
-      "loss": 46.0,
-      "step": 5446
-    },
-    {
-      "epoch": 0.8771689681549176,
-      "grad_norm": 0.001731664757244289,
-      "learning_rate": 0.00019999962169567334,
-      "loss": 46.0,
-      "step": 5447
-    },
-    {
-      "epoch": 0.8773300052337051,
-      "grad_norm": 0.0007576161879114807,
-      "learning_rate": 0.00019999962155650144,
-      "loss": 46.0,
-      "step": 5448
-    },
-    {
-      "epoch": 0.8774910423124924,
-      "grad_norm": 0.0023225368931889534,
-      "learning_rate": 0.0001999996214173039,
-      "loss": 46.0,
-      "step": 5449
-    },
-    {
-      "epoch": 0.8776520793912799,
-      "grad_norm": 0.002116841496899724,
-      "learning_rate": 0.00019999962127808076,
-      "loss": 46.0,
-      "step": 5450
-    },
-    {
-      "epoch": 0.8778131164700672,
-      "grad_norm": 0.000905901484657079,
-      "learning_rate": 0.00019999962113883207,
-      "loss": 46.0,
-      "step": 5451
-    },
-    {
-      "epoch": 0.8779741535488547,
-      "grad_norm": 0.001066633383743465,
-      "learning_rate": 0.0001999996209995578,
-      "loss": 46.0,
-      "step": 5452
-    },
-    {
-      "epoch": 0.878135190627642,
-      "grad_norm": 0.003724220674484968,
-      "learning_rate": 0.00019999962086025787,
-      "loss": 46.0,
-      "step": 5453
-    },
-    {
-      "epoch": 0.8782962277064295,
-      "grad_norm": 0.0016309581696987152,
-      "learning_rate": 0.00019999962072093236,
-      "loss": 46.0,
-      "step": 5454
-    },
-    {
-      "epoch": 0.8784572647852168,
-      "grad_norm": 0.0006372524658218026,
-      "learning_rate": 0.0001999996205815813,
-      "loss": 46.0,
-      "step": 5455
-    },
-    {
-      "epoch": 0.8786183018640041,
-      "grad_norm": 0.004177498165518045,
-      "learning_rate": 0.0001999996204422046,
-      "loss": 46.0,
-      "step": 5456
-    },
-    {
-      "epoch": 0.8787793389427916,
-      "grad_norm": 0.0012889906065538526,
-      "learning_rate": 0.00019999962030280234,
-      "loss": 46.0,
-      "step": 5457
-    },
-    {
-      "epoch": 0.8789403760215789,
-      "grad_norm": 0.0012320299865677953,
-      "learning_rate": 0.00019999962016337442,
-      "loss": 46.0,
-      "step": 5458
-    },
-    {
-      "epoch": 0.8791014131003664,
-      "grad_norm": 0.002860185457393527,
-      "learning_rate": 0.00019999962002392098,
-      "loss": 46.0,
-      "step": 5459
-    },
-    {
-      "epoch": 0.8792624501791537,
-      "grad_norm": 0.0008574715466238558,
-      "learning_rate": 0.00019999961988444192,
-      "loss": 46.0,
-      "step": 5460
-    },
-    {
-      "epoch": 0.8794234872579412,
-      "grad_norm": 0.0009642100776545703,
-      "learning_rate": 0.00019999961974493725,
-      "loss": 46.0,
-      "step": 5461
-    },
-    {
-      "epoch": 0.8795845243367285,
-      "grad_norm": 0.004119659774005413,
-      "learning_rate": 0.00019999961960540704,
-      "loss": 46.0,
-      "step": 5462
-    },
-    {
-      "epoch": 0.879745561415516,
-      "grad_norm": 0.0006561786867678165,
-      "learning_rate": 0.00019999961946585114,
-      "loss": 46.0,
-      "step": 5463
-    },
-    {
-      "epoch": 0.8799065984943033,
-      "grad_norm": 0.007290414068847895,
-      "learning_rate": 0.00019999961932626973,
-      "loss": 46.0,
-      "step": 5464
-    },
-    {
-      "epoch": 0.8800676355730908,
-      "grad_norm": 0.0007360965246334672,
-      "learning_rate": 0.0001999996191866627,
-      "loss": 46.0,
-      "step": 5465
-    },
-    {
-      "epoch": 0.8802286726518781,
-      "grad_norm": 0.005771649070084095,
-      "learning_rate": 0.00019999961904703008,
-      "loss": 46.0,
-      "step": 5466
-    },
-    {
-      "epoch": 0.8803897097306654,
-      "grad_norm": 0.0018337457440793514,
-      "learning_rate": 0.00019999961890737185,
-      "loss": 46.0,
-      "step": 5467
-    },
-    {
-      "epoch": 0.8805507468094529,
-      "grad_norm": 0.0016193913761526346,
-      "learning_rate": 0.00019999961876768801,
-      "loss": 46.0,
-      "step": 5468
-    },
-    {
-      "epoch": 0.8807117838882402,
-      "grad_norm": 0.002049280796200037,
-      "learning_rate": 0.0001999996186279786,
-      "loss": 46.0,
-      "step": 5469
-    },
-    {
-      "epoch": 0.8808728209670277,
-      "grad_norm": 0.00038791593397036195,
-      "learning_rate": 0.0001999996184882436,
-      "loss": 46.0,
-      "step": 5470
-    },
-    {
-      "epoch": 0.881033858045815,
-      "grad_norm": 0.0014635240659117699,
-      "learning_rate": 0.000199999618348483,
-      "loss": 46.0,
-      "step": 5471
-    },
-    {
-      "epoch": 0.8811948951246025,
-      "grad_norm": 0.0008576675318181515,
-      "learning_rate": 0.00019999961820869677,
-      "loss": 46.0,
-      "step": 5472
-    },
-    {
-      "epoch": 0.8813559322033898,
-      "grad_norm": 0.0004542733950074762,
-      "learning_rate": 0.000199999618068885,
-      "loss": 46.0,
-      "step": 5473
-    },
-    {
-      "epoch": 0.8815169692821773,
-      "grad_norm": 0.0005807816050946712,
-      "learning_rate": 0.0001999996179290476,
-      "loss": 46.0,
-      "step": 5474
-    },
-    {
-      "epoch": 0.8816780063609646,
-      "grad_norm": 0.002323566470295191,
-      "learning_rate": 0.0001999996177891846,
-      "loss": 46.0,
-      "step": 5475
-    },
-    {
-      "epoch": 0.8818390434397521,
-      "grad_norm": 0.0022160017397254705,
-      "learning_rate": 0.00019999961764929601,
-      "loss": 46.0,
-      "step": 5476
-    },
-    {
-      "epoch": 0.8820000805185394,
-      "grad_norm": 0.002299319254234433,
-      "learning_rate": 0.00019999961750938187,
-      "loss": 46.0,
-      "step": 5477
-    },
-    {
-      "epoch": 0.8821611175973267,
-      "grad_norm": 0.0012396796373650432,
-      "learning_rate": 0.0001999996173694421,
-      "loss": 46.0,
-      "step": 5478
-    },
-    {
-      "epoch": 0.8823221546761142,
-      "grad_norm": 0.0007765289628878236,
-      "learning_rate": 0.00019999961722947672,
-      "loss": 46.0,
-      "step": 5479
-    },
-    {
-      "epoch": 0.8824831917549015,
-      "grad_norm": 0.0007036970928311348,
-      "learning_rate": 0.0001999996170894858,
-      "loss": 46.0,
-      "step": 5480
-    },
-    {
-      "epoch": 0.882644228833689,
-      "grad_norm": 0.000626065768301487,
-      "learning_rate": 0.00019999961694946923,
-      "loss": 46.0,
-      "step": 5481
-    },
-    {
-      "epoch": 0.8828052659124763,
-      "grad_norm": 0.0008546230383217335,
-      "learning_rate": 0.0001999996168094271,
-      "loss": 46.0,
-      "step": 5482
-    },
-    {
-      "epoch": 0.8829663029912638,
-      "grad_norm": 0.004829421639442444,
-      "learning_rate": 0.00019999961666935934,
-      "loss": 46.0,
-      "step": 5483
-    },
-    {
-      "epoch": 0.8831273400700511,
-      "grad_norm": 0.0008491275948472321,
-      "learning_rate": 0.000199999616529266,
-      "loss": 46.0,
-      "step": 5484
-    },
-    {
-      "epoch": 0.8832883771488386,
-      "grad_norm": 0.003391230246052146,
-      "learning_rate": 0.00019999961638914705,
-      "loss": 46.0,
-      "step": 5485
-    },
-    {
-      "epoch": 0.8834494142276259,
-      "grad_norm": 0.0014533017529174685,
-      "learning_rate": 0.00019999961624900253,
-      "loss": 46.0,
-      "step": 5486
-    },
-    {
-      "epoch": 0.8836104513064132,
-      "grad_norm": 0.0034319807309657335,
-      "learning_rate": 0.00019999961610883243,
-      "loss": 46.0,
-      "step": 5487
-    },
-    {
-      "epoch": 0.8837714883852007,
-      "grad_norm": 0.0006078967708162963,
-      "learning_rate": 0.00019999961596863669,
-      "loss": 46.0,
-      "step": 5488
-    },
-    {
-      "epoch": 0.883932525463988,
-      "grad_norm": 0.00028470889083109796,
-      "learning_rate": 0.0001999996158284154,
-      "loss": 46.0,
-      "step": 5489
-    },
-    {
-      "epoch": 0.8840935625427755,
-      "grad_norm": 0.001391082420013845,
-      "learning_rate": 0.0001999996156881685,
-      "loss": 46.0,
-      "step": 5490
-    },
-    {
-      "epoch": 0.8842545996215628,
-      "grad_norm": 0.0005989590426906943,
-      "learning_rate": 0.000199999615547896,
-      "loss": 46.0,
-      "step": 5491
-    },
-    {
-      "epoch": 0.8844156367003503,
-      "grad_norm": 0.0017267513321712613,
-      "learning_rate": 0.0001999996154075979,
-      "loss": 46.0,
-      "step": 5492
-    },
-    {
-      "epoch": 0.8845766737791376,
-      "grad_norm": 0.0006365755689330399,
-      "learning_rate": 0.0001999996152672742,
-      "loss": 46.0,
-      "step": 5493
-    },
-    {
-      "epoch": 0.8847377108579251,
-      "grad_norm": 0.0016927524702623487,
-      "learning_rate": 0.00019999961512692492,
-      "loss": 46.0,
-      "step": 5494
-    },
-    {
-      "epoch": 0.8848987479367124,
-      "grad_norm": 0.0007248780457302928,
-      "learning_rate": 0.00019999961498655004,
-      "loss": 46.0,
-      "step": 5495
-    },
-    {
-      "epoch": 0.8850597850154999,
-      "grad_norm": 0.0002985633327625692,
-      "learning_rate": 0.00019999961484614955,
-      "loss": 46.0,
-      "step": 5496
-    },
-    {
-      "epoch": 0.8852208220942872,
-      "grad_norm": 0.0013461465714499354,
-      "learning_rate": 0.0001999996147057235,
-      "loss": 46.0,
-      "step": 5497
-    },
-    {
-      "epoch": 0.8853818591730745,
-      "grad_norm": 0.0025419513694941998,
-      "learning_rate": 0.00019999961456527182,
-      "loss": 46.0,
-      "step": 5498
-    },
-    {
-      "epoch": 0.885542896251862,
-      "grad_norm": 0.00048314695595763624,
-      "learning_rate": 0.0001999996144247946,
-      "loss": 46.0,
-      "step": 5499
-    },
-    {
-      "epoch": 0.8857039333306493,
-      "grad_norm": 0.0005405222764238715,
-      "learning_rate": 0.00019999961428429172,
-      "loss": 46.0,
-      "step": 5500
-    },
-    {
-      "epoch": 0.8858649704094368,
-      "grad_norm": 0.0009830312337726355,
-      "learning_rate": 0.00019999961414376326,
-      "loss": 46.0,
-      "step": 5501
-    },
-    {
-      "epoch": 0.8860260074882241,
-      "grad_norm": 0.0010159957455471158,
-      "learning_rate": 0.00019999961400320925,
-      "loss": 46.0,
-      "step": 5502
-    },
-    {
-      "epoch": 0.8861870445670116,
-      "grad_norm": 0.002079539932310581,
-      "learning_rate": 0.00019999961386262959,
-      "loss": 46.0,
-      "step": 5503
-    },
-    {
-      "epoch": 0.8863480816457989,
-      "grad_norm": 0.0027389926835894585,
-      "learning_rate": 0.00019999961372202437,
-      "loss": 46.0,
-      "step": 5504
-    },
-    {
-      "epoch": 0.8865091187245864,
-      "grad_norm": 0.001533121452666819,
-      "learning_rate": 0.00019999961358139353,
-      "loss": 46.0,
-      "step": 5505
-    },
-    {
-      "epoch": 0.8866701558033737,
-      "grad_norm": 0.0008724723011255264,
-      "learning_rate": 0.0001999996134407371,
-      "loss": 46.0,
-      "step": 5506
-    },
-    {
-      "epoch": 0.8868311928821612,
-      "grad_norm": 0.0034404757898300886,
-      "learning_rate": 0.0001999996133000551,
-      "loss": 46.0,
-      "step": 5507
-    },
-    {
-      "epoch": 0.8869922299609485,
-      "grad_norm": 0.0017426405102014542,
-      "learning_rate": 0.00019999961315934748,
-      "loss": 46.0,
-      "step": 5508
-    },
-    {
-      "epoch": 0.8871532670397358,
-      "grad_norm": 0.000947405700571835,
-      "learning_rate": 0.00019999961301861427,
-      "loss": 46.0,
-      "step": 5509
-    },
-    {
-      "epoch": 0.8873143041185233,
-      "grad_norm": 0.0005466010188683867,
-      "learning_rate": 0.0001999996128778555,
-      "loss": 46.0,
-      "step": 5510
-    },
-    {
-      "epoch": 0.8874753411973106,
-      "grad_norm": 0.0017102526035159826,
-      "learning_rate": 0.0001999996127370711,
-      "loss": 46.0,
-      "step": 5511
-    },
-    {
-      "epoch": 0.8876363782760981,
-      "grad_norm": 0.0003407685726415366,
-      "learning_rate": 0.0001999996125962611,
-      "loss": 46.0,
-      "step": 5512
-    },
-    {
-      "epoch": 0.8877974153548854,
-      "grad_norm": 0.0015317793004214764,
-      "learning_rate": 0.0001999996124554255,
-      "loss": 46.0,
-      "step": 5513
-    },
-    {
-      "epoch": 0.8879584524336729,
-      "grad_norm": 0.001786372042261064,
-      "learning_rate": 0.00019999961231456433,
-      "loss": 46.0,
-      "step": 5514
-    },
-    {
-      "epoch": 0.8881194895124602,
-      "grad_norm": 0.0005082154530100524,
-      "learning_rate": 0.00019999961217367755,
-      "loss": 46.0,
-      "step": 5515
-    },
-    {
-      "epoch": 0.8882805265912477,
-      "grad_norm": 0.0008765160455368459,
-      "learning_rate": 0.00019999961203276517,
-      "loss": 46.0,
-      "step": 5516
-    },
-    {
-      "epoch": 0.888441563670035,
-      "grad_norm": 0.0014389300486072898,
-      "learning_rate": 0.0001999996118918272,
-      "loss": 46.0,
-      "step": 5517
-    },
-    {
-      "epoch": 0.8886026007488225,
-      "grad_norm": 0.0007574508199468255,
-      "learning_rate": 0.00019999961175086366,
-      "loss": 46.0,
-      "step": 5518
-    },
-    {
-      "epoch": 0.8887636378276098,
-      "grad_norm": 0.0010004672221839428,
-      "learning_rate": 0.0001999996116098745,
-      "loss": 46.0,
-      "step": 5519
-    },
-    {
-      "epoch": 0.8889246749063971,
-      "grad_norm": 0.0006244581891223788,
-      "learning_rate": 0.00019999961146885978,
-      "loss": 46.0,
-      "step": 5520
-    },
-    {
-      "epoch": 0.8890857119851846,
-      "grad_norm": 0.0005050482577644289,
-      "learning_rate": 0.00019999961132781941,
-      "loss": 46.0,
-      "step": 5521
-    },
-    {
-      "epoch": 0.8892467490639719,
-      "grad_norm": 0.001633846783079207,
-      "learning_rate": 0.00019999961118675346,
-      "loss": 46.0,
-      "step": 5522
-    },
-    {
-      "epoch": 0.8894077861427594,
-      "grad_norm": 0.0005150714423507452,
-      "learning_rate": 0.00019999961104566193,
-      "loss": 46.0,
-      "step": 5523
-    },
-    {
-      "epoch": 0.8895688232215467,
-      "grad_norm": 0.0013295452808961272,
-      "learning_rate": 0.00019999961090454483,
-      "loss": 46.0,
-      "step": 5524
-    },
-    {
-      "epoch": 0.8897298603003342,
-      "grad_norm": 0.0009041702724061906,
-      "learning_rate": 0.00019999961076340211,
-      "loss": 46.0,
-      "step": 5525
-    },
-    {
-      "epoch": 0.8898908973791215,
-      "grad_norm": 0.00155701267067343,
-      "learning_rate": 0.0001999996106222338,
-      "loss": 46.0,
-      "step": 5526
-    },
-    {
-      "epoch": 0.890051934457909,
-      "grad_norm": 0.0006994783179834485,
-      "learning_rate": 0.00019999961048103985,
-      "loss": 46.0,
-      "step": 5527
-    },
-    {
-      "epoch": 0.8902129715366963,
-      "grad_norm": 0.004856567829847336,
-      "learning_rate": 0.00019999961033982037,
-      "loss": 46.0,
-      "step": 5528
-    },
-    {
-      "epoch": 0.8903740086154838,
-      "grad_norm": 0.0012894722167402506,
-      "learning_rate": 0.00019999961019857528,
-      "loss": 46.0,
-      "step": 5529
-    },
-    {
-      "epoch": 0.8905350456942711,
-      "grad_norm": 0.0009183423244394362,
-      "learning_rate": 0.00019999961005730455,
-      "loss": 46.0,
-      "step": 5530
-    },
-    {
-      "epoch": 0.8906960827730585,
-      "grad_norm": 0.0011541893472895026,
-      "learning_rate": 0.0001999996099160083,
-      "loss": 46.0,
-      "step": 5531
-    },
-    {
-      "epoch": 0.8908571198518459,
-      "grad_norm": 0.0033031266648322344,
-      "learning_rate": 0.00019999960977468639,
-      "loss": 46.0,
-      "step": 5532
-    },
-    {
-      "epoch": 0.8910181569306332,
-      "grad_norm": 0.001081187860108912,
-      "learning_rate": 0.0001999996096333389,
-      "loss": 46.0,
-      "step": 5533
-    },
-    {
-      "epoch": 0.8911791940094207,
-      "grad_norm": 0.0005713963182643056,
-      "learning_rate": 0.00019999960949196584,
-      "loss": 46.0,
-      "step": 5534
-    },
-    {
-      "epoch": 0.891340231088208,
-      "grad_norm": 0.001636003260500729,
-      "learning_rate": 0.00019999960935056715,
-      "loss": 46.0,
-      "step": 5535
-    },
-    {
-      "epoch": 0.8915012681669955,
-      "grad_norm": 0.0020119203254580498,
-      "learning_rate": 0.00019999960920914292,
-      "loss": 46.0,
-      "step": 5536
-    },
-    {
-      "epoch": 0.8916623052457828,
-      "grad_norm": 0.001871225074864924,
-      "learning_rate": 0.00019999960906769303,
-      "loss": 46.0,
-      "step": 5537
-    },
-    {
-      "epoch": 0.8918233423245703,
-      "grad_norm": 0.0006526735960505903,
-      "learning_rate": 0.00019999960892621758,
-      "loss": 46.0,
-      "step": 5538
-    },
-    {
-      "epoch": 0.8919843794033576,
-      "grad_norm": 0.0018917681882157922,
-      "learning_rate": 0.00019999960878471653,
-      "loss": 46.0,
-      "step": 5539
-    },
-    {
-      "epoch": 0.892145416482145,
-      "grad_norm": 0.0007902439683675766,
-      "learning_rate": 0.0001999996086431899,
-      "loss": 46.0,
-      "step": 5540
-    },
-    {
-      "epoch": 0.8923064535609324,
-      "grad_norm": 0.002496239962056279,
-      "learning_rate": 0.00019999960850163766,
-      "loss": 46.0,
-      "step": 5541
-    },
-    {
-      "epoch": 0.8924674906397198,
-      "grad_norm": 0.0005020059179514647,
-      "learning_rate": 0.00019999960836005983,
-      "loss": 46.0,
-      "step": 5542
-    },
-    {
-      "epoch": 0.8926285277185072,
-      "grad_norm": 0.001550781773403287,
-      "learning_rate": 0.00019999960821845641,
-      "loss": 46.0,
-      "step": 5543
-    },
-    {
-      "epoch": 0.8927895647972945,
-      "grad_norm": 0.0007806345238350332,
-      "learning_rate": 0.00019999960807682738,
-      "loss": 46.0,
-      "step": 5544
-    },
-    {
-      "epoch": 0.892950601876082,
-      "grad_norm": 0.0017287826631218195,
-      "learning_rate": 0.00019999960793517274,
-      "loss": 46.0,
-      "step": 5545
-    },
-    {
-      "epoch": 0.8931116389548693,
-      "grad_norm": 0.0007485355599783361,
-      "learning_rate": 0.00019999960779349253,
-      "loss": 46.0,
-      "step": 5546
-    },
-    {
-      "epoch": 0.8932726760336568,
-      "grad_norm": 0.0016300444258376956,
-      "learning_rate": 0.00019999960765178674,
-      "loss": 46.0,
-      "step": 5547
-    },
-    {
-      "epoch": 0.8934337131124441,
-      "grad_norm": 0.00109186302870512,
-      "learning_rate": 0.00019999960751005533,
-      "loss": 46.0,
-      "step": 5548
-    },
-    {
-      "epoch": 0.8935947501912316,
-      "grad_norm": 0.0007668111938983202,
-      "learning_rate": 0.00019999960736829834,
-      "loss": 46.0,
-      "step": 5549
-    },
-    {
-      "epoch": 0.8937557872700189,
-      "grad_norm": 0.0016533834859728813,
-      "learning_rate": 0.00019999960722651573,
-      "loss": 46.0,
-      "step": 5550
-    },
-    {
-      "epoch": 0.8939168243488063,
-      "grad_norm": 0.003962287213653326,
-      "learning_rate": 0.00019999960708470756,
-      "loss": 46.0,
-      "step": 5551
-    },
-    {
-      "epoch": 0.8940778614275937,
-      "grad_norm": 0.0022235820069909096,
-      "learning_rate": 0.00019999960694287377,
-      "loss": 46.0,
-      "step": 5552
-    },
-    {
-      "epoch": 0.894238898506381,
-      "grad_norm": 0.0011650846572592854,
-      "learning_rate": 0.0001999996068010144,
-      "loss": 46.0,
-      "step": 5553
-    },
-    {
-      "epoch": 0.8943999355851685,
-      "grad_norm": 0.000557522289454937,
-      "learning_rate": 0.00019999960665912942,
-      "loss": 46.0,
-      "step": 5554
-    },
-    {
-      "epoch": 0.8945609726639558,
-      "grad_norm": 0.0014321466442197561,
-      "learning_rate": 0.00019999960651721885,
-      "loss": 46.0,
-      "step": 5555
-    },
-    {
-      "epoch": 0.8947220097427433,
-      "grad_norm": 0.0008956867386586964,
-      "learning_rate": 0.0001999996063752827,
-      "loss": 46.0,
-      "step": 5556
-    },
-    {
-      "epoch": 0.8948830468215306,
-      "grad_norm": 0.004334675148129463,
-      "learning_rate": 0.00019999960623332094,
-      "loss": 46.0,
-      "step": 5557
-    },
-    {
-      "epoch": 0.8950440839003181,
-      "grad_norm": 0.001942360308021307,
-      "learning_rate": 0.0001999996060913336,
-      "loss": 46.0,
-      "step": 5558
-    },
-    {
-      "epoch": 0.8952051209791054,
-      "grad_norm": 0.0013158803340047598,
-      "learning_rate": 0.00019999960594932066,
-      "loss": 46.0,
-      "step": 5559
-    },
-    {
-      "epoch": 0.8953661580578929,
-      "grad_norm": 0.0010415952419862151,
-      "learning_rate": 0.00019999960580728213,
-      "loss": 46.0,
-      "step": 5560
-    },
-    {
-      "epoch": 0.8955271951366802,
-      "grad_norm": 0.0019848961383104324,
-      "learning_rate": 0.00019999960566521798,
-      "loss": 46.0,
-      "step": 5561
-    },
-    {
-      "epoch": 0.8956882322154676,
-      "grad_norm": 0.005372314713895321,
-      "learning_rate": 0.00019999960552312827,
-      "loss": 46.0,
-      "step": 5562
-    },
-    {
-      "epoch": 0.895849269294255,
-      "grad_norm": 0.0013121406082063913,
-      "learning_rate": 0.00019999960538101292,
-      "loss": 46.0,
-      "step": 5563
-    },
-    {
-      "epoch": 0.8960103063730424,
-      "grad_norm": 0.0017942595295608044,
-      "learning_rate": 0.000199999605238872,
-      "loss": 46.0,
-      "step": 5564
-    },
-    {
-      "epoch": 0.8961713434518298,
-      "grad_norm": 0.0036651925183832645,
-      "learning_rate": 0.0001999996050967055,
-      "loss": 46.0,
-      "step": 5565
-    },
-    {
-      "epoch": 0.8963323805306171,
-      "grad_norm": 0.0016075974563136697,
-      "learning_rate": 0.00019999960495451337,
-      "loss": 46.0,
-      "step": 5566
-    },
-    {
-      "epoch": 0.8964934176094046,
-      "grad_norm": 0.002354175318032503,
-      "learning_rate": 0.00019999960481229567,
-      "loss": 46.0,
-      "step": 5567
-    },
-    {
-      "epoch": 0.8966544546881919,
-      "grad_norm": 0.0004906888934783638,
-      "learning_rate": 0.00019999960467005239,
-      "loss": 46.0,
-      "step": 5568
-    },
-    {
-      "epoch": 0.8968154917669794,
-      "grad_norm": 0.0006193263689056039,
-      "learning_rate": 0.0001999996045277835,
-      "loss": 46.0,
-      "step": 5569
-    },
-    {
-      "epoch": 0.8969765288457667,
-      "grad_norm": 0.004660008475184441,
-      "learning_rate": 0.00019999960438548902,
-      "loss": 46.0,
-      "step": 5570
-    },
-    {
-      "epoch": 0.8971375659245542,
-      "grad_norm": 0.0016145760891959071,
-      "learning_rate": 0.00019999960424316892,
-      "loss": 46.0,
-      "step": 5571
-    },
-    {
-      "epoch": 0.8972986030033415,
-      "grad_norm": 0.0008006156422197819,
-      "learning_rate": 0.00019999960410082326,
-      "loss": 46.0,
-      "step": 5572
-    },
-    {
-      "epoch": 0.8974596400821289,
-      "grad_norm": 0.0028087706305086613,
-      "learning_rate": 0.00019999960395845196,
-      "loss": 46.0,
-      "step": 5573
-    },
-    {
-      "epoch": 0.8976206771609163,
-      "grad_norm": 0.005286147352308035,
-      "learning_rate": 0.00019999960381605512,
-      "loss": 46.0,
-      "step": 5574
-    },
-    {
-      "epoch": 0.8977817142397037,
-      "grad_norm": 0.0018280866788700223,
-      "learning_rate": 0.00019999960367363264,
-      "loss": 46.0,
-      "step": 5575
-    },
-    {
-      "epoch": 0.8979427513184911,
-      "grad_norm": 0.001505769439972937,
-      "learning_rate": 0.00019999960353118458,
-      "loss": 46.0,
-      "step": 5576
-    },
-    {
-      "epoch": 0.8981037883972784,
-      "grad_norm": 0.0006667435518465936,
-      "learning_rate": 0.00019999960338871095,
-      "loss": 46.0,
-      "step": 5577
-    },
-    {
-      "epoch": 0.8982648254760659,
-      "grad_norm": 0.004198502283543348,
-      "learning_rate": 0.00019999960324621168,
-      "loss": 46.0,
-      "step": 5578
-    },
-    {
-      "epoch": 0.8984258625548532,
-      "grad_norm": 0.0030197720043361187,
-      "learning_rate": 0.00019999960310368686,
-      "loss": 46.0,
-      "step": 5579
-    },
-    {
-      "epoch": 0.8985868996336407,
-      "grad_norm": 0.0007644347497262061,
-      "learning_rate": 0.00019999960296113644,
-      "loss": 46.0,
-      "step": 5580
-    },
-    {
-      "epoch": 0.898747936712428,
-      "grad_norm": 0.0010789462830871344,
-      "learning_rate": 0.00019999960281856042,
-      "loss": 46.0,
-      "step": 5581
-    },
-    {
-      "epoch": 0.8989089737912155,
-      "grad_norm": 0.0008686682558618486,
-      "learning_rate": 0.00019999960267595877,
-      "loss": 46.0,
-      "step": 5582
-    },
-    {
-      "epoch": 0.8990700108700028,
-      "grad_norm": 0.0025454119313508272,
-      "learning_rate": 0.00019999960253333154,
-      "loss": 46.0,
-      "step": 5583
-    },
-    {
-      "epoch": 0.8992310479487902,
-      "grad_norm": 0.0009476811392232776,
-      "learning_rate": 0.00019999960239067876,
-      "loss": 46.0,
-      "step": 5584
-    },
-    {
-      "epoch": 0.8993920850275776,
-      "grad_norm": 0.0009671699372120202,
-      "learning_rate": 0.00019999960224800035,
-      "loss": 46.0,
-      "step": 5585
-    },
-    {
-      "epoch": 0.899553122106365,
-      "grad_norm": 0.000562049972359091,
-      "learning_rate": 0.00019999960210529633,
-      "loss": 46.0,
-      "step": 5586
-    },
-    {
-      "epoch": 0.8997141591851524,
-      "grad_norm": 0.0005062755662947893,
-      "learning_rate": 0.00019999960196256673,
-      "loss": 46.0,
-      "step": 5587
-    },
-    {
-      "epoch": 0.8998751962639397,
-      "grad_norm": 0.00045780991786159575,
-      "learning_rate": 0.00019999960181981154,
-      "loss": 46.0,
-      "step": 5588
-    },
-    {
-      "epoch": 0.9000362333427272,
-      "grad_norm": 0.0012315440690144897,
-      "learning_rate": 0.00019999960167703076,
-      "loss": 46.0,
-      "step": 5589
-    },
-    {
-      "epoch": 0.9001972704215145,
-      "grad_norm": 0.00034972140565514565,
-      "learning_rate": 0.00019999960153422436,
-      "loss": 46.0,
-      "step": 5590
-    },
-    {
-      "epoch": 0.900358307500302,
-      "grad_norm": 0.0012553875567391515,
-      "learning_rate": 0.0001999996013913924,
-      "loss": 46.0,
-      "step": 5591
-    },
-    {
-      "epoch": 0.9005193445790893,
-      "grad_norm": 0.0021845693700015545,
-      "learning_rate": 0.00019999960124853484,
-      "loss": 46.0,
-      "step": 5592
-    },
-    {
-      "epoch": 0.9006803816578767,
-      "grad_norm": 0.0016618001973256469,
-      "learning_rate": 0.00019999960110565166,
-      "loss": 46.0,
-      "step": 5593
-    },
-    {
-      "epoch": 0.9008414187366641,
-      "grad_norm": 0.0011843597749248147,
-      "learning_rate": 0.00019999960096274291,
-      "loss": 46.0,
-      "step": 5594
-    },
-    {
-      "epoch": 0.9010024558154515,
-      "grad_norm": 0.000884176348336041,
-      "learning_rate": 0.00019999960081980853,
-      "loss": 46.0,
-      "step": 5595
-    },
-    {
-      "epoch": 0.9011634928942389,
-      "grad_norm": 0.0023520884569734335,
-      "learning_rate": 0.0001999996006768486,
-      "loss": 46.0,
-      "step": 5596
-    },
-    {
-      "epoch": 0.9013245299730263,
-      "grad_norm": 0.003695164807140827,
-      "learning_rate": 0.00019999960053386303,
-      "loss": 46.0,
-      "step": 5597
-    },
-    {
-      "epoch": 0.9014855670518137,
-      "grad_norm": 0.0038136360235512257,
-      "learning_rate": 0.0001999996003908519,
-      "loss": 46.0,
-      "step": 5598
-    },
-    {
-      "epoch": 0.901646604130601,
-      "grad_norm": 0.0005589498905465007,
-      "learning_rate": 0.00019999960024781518,
-      "loss": 46.0,
-      "step": 5599
-    },
-    {
-      "epoch": 0.9018076412093885,
-      "grad_norm": 0.0005771736614406109,
-      "learning_rate": 0.00019999960010475283,
-      "loss": 46.0,
-      "step": 5600
-    },
-    {
-      "epoch": 0.9019686782881758,
-      "grad_norm": 0.0008498323149979115,
-      "learning_rate": 0.00019999959996166492,
-      "loss": 46.0,
-      "step": 5601
-    },
-    {
-      "epoch": 0.9021297153669633,
-      "grad_norm": 0.0008164996979758143,
-      "learning_rate": 0.0001999995998185514,
-      "loss": 46.0,
-      "step": 5602
-    },
-    {
-      "epoch": 0.9022907524457506,
-      "grad_norm": 0.004340397194027901,
-      "learning_rate": 0.00019999959967541227,
-      "loss": 46.0,
-      "step": 5603
-    },
-    {
-      "epoch": 0.902451789524538,
-      "grad_norm": 0.0010171361500397325,
-      "learning_rate": 0.00019999959953224757,
-      "loss": 46.0,
-      "step": 5604
-    },
-    {
-      "epoch": 0.9026128266033254,
-      "grad_norm": 0.004007955547422171,
-      "learning_rate": 0.0001999995993890573,
-      "loss": 46.0,
-      "step": 5605
-    },
-    {
-      "epoch": 0.9027738636821128,
-      "grad_norm": 0.0009484782931394875,
-      "learning_rate": 0.00019999959924584136,
-      "loss": 46.0,
-      "step": 5606
-    },
-    {
-      "epoch": 0.9029349007609002,
-      "grad_norm": 0.0010315830586478114,
-      "learning_rate": 0.00019999959910259988,
-      "loss": 46.0,
-      "step": 5607
-    },
-    {
-      "epoch": 0.9030959378396876,
-      "grad_norm": 0.002179377945140004,
-      "learning_rate": 0.0001999995989593328,
-      "loss": 46.0,
-      "step": 5608
-    },
-    {
-      "epoch": 0.903256974918475,
-      "grad_norm": 0.0018888365011662245,
-      "learning_rate": 0.0001999995988160401,
-      "loss": 46.0,
-      "step": 5609
-    },
-    {
-      "epoch": 0.9034180119972623,
-      "grad_norm": 0.0010518798371776938,
-      "learning_rate": 0.00019999959867272182,
-      "loss": 46.0,
-      "step": 5610
-    },
-    {
-      "epoch": 0.9035790490760498,
-      "grad_norm": 0.0016064199153333902,
-      "learning_rate": 0.00019999959852937796,
-      "loss": 46.0,
-      "step": 5611
-    },
-    {
-      "epoch": 0.9037400861548371,
-      "grad_norm": 0.002031848533079028,
-      "learning_rate": 0.0001999995983860085,
-      "loss": 46.0,
-      "step": 5612
-    },
-    {
-      "epoch": 0.9039011232336246,
-      "grad_norm": 0.0007423665374517441,
-      "learning_rate": 0.00019999959824261343,
-      "loss": 46.0,
-      "step": 5613
-    },
-    {
-      "epoch": 0.9040621603124119,
-      "grad_norm": 0.005528111010789871,
-      "learning_rate": 0.00019999959809919275,
-      "loss": 46.0,
-      "step": 5614
-    },
-    {
-      "epoch": 0.9042231973911993,
-      "grad_norm": 0.0009552434785291553,
-      "learning_rate": 0.00019999959795574652,
-      "loss": 46.0,
-      "step": 5615
-    },
-    {
-      "epoch": 0.9043842344699867,
-      "grad_norm": 0.0010574173647910357,
-      "learning_rate": 0.00019999959781227467,
-      "loss": 46.0,
-      "step": 5616
-    },
-    {
-      "epoch": 0.9045452715487741,
-      "grad_norm": 0.001811443711631,
-      "learning_rate": 0.00019999959766877723,
-      "loss": 46.0,
-      "step": 5617
-    },
-    {
-      "epoch": 0.9047063086275615,
-      "grad_norm": 0.001466817338950932,
-      "learning_rate": 0.00019999959752525418,
-      "loss": 46.0,
-      "step": 5618
-    },
-    {
-      "epoch": 0.9048673457063489,
-      "grad_norm": 0.0010563834803178906,
-      "learning_rate": 0.00019999959738170557,
-      "loss": 46.0,
-      "step": 5619
-    },
-    {
-      "epoch": 0.9050283827851363,
-      "grad_norm": 0.0012623857473954558,
-      "learning_rate": 0.00019999959723813134,
-      "loss": 46.0,
-      "step": 5620
-    },
-    {
-      "epoch": 0.9051894198639236,
-      "grad_norm": 0.0020953472703695297,
-      "learning_rate": 0.0001999995970945315,
-      "loss": 46.0,
-      "step": 5621
-    },
-    {
-      "epoch": 0.9053504569427111,
-      "grad_norm": 0.0009686333360150456,
-      "learning_rate": 0.0001999995969509061,
-      "loss": 46.0,
-      "step": 5622
-    },
-    {
-      "epoch": 0.9055114940214984,
-      "grad_norm": 0.001027517719194293,
-      "learning_rate": 0.00019999959680725509,
-      "loss": 46.0,
-      "step": 5623
-    },
-    {
-      "epoch": 0.9056725311002859,
-      "grad_norm": 0.0004785277706105262,
-      "learning_rate": 0.0001999995966635785,
-      "loss": 46.0,
-      "step": 5624
-    },
-    {
-      "epoch": 0.9058335681790732,
-      "grad_norm": 0.000614838267210871,
-      "learning_rate": 0.0001999995965198763,
-      "loss": 46.0,
-      "step": 5625
-    },
-    {
-      "epoch": 0.9059946052578606,
-      "grad_norm": 0.0012222366640344262,
-      "learning_rate": 0.00019999959637614852,
-      "loss": 46.0,
-      "step": 5626
-    },
-    {
-      "epoch": 0.906155642336648,
-      "grad_norm": 0.0007167730946093798,
-      "learning_rate": 0.0001999995962323951,
-      "loss": 46.0,
-      "step": 5627
-    },
-    {
-      "epoch": 0.9063166794154354,
-      "grad_norm": 0.001726824208162725,
-      "learning_rate": 0.00019999959608861612,
-      "loss": 46.0,
-      "step": 5628
-    },
-    {
-      "epoch": 0.9064777164942228,
-      "grad_norm": 0.0005700625479221344,
-      "learning_rate": 0.00019999959594481156,
-      "loss": 46.0,
-      "step": 5629
-    },
-    {
-      "epoch": 0.9066387535730102,
-      "grad_norm": 0.003727745031937957,
-      "learning_rate": 0.00019999959580098138,
-      "loss": 46.0,
-      "step": 5630
-    },
-    {
-      "epoch": 0.9067997906517976,
-      "grad_norm": 0.000594357552472502,
-      "learning_rate": 0.00019999959565712561,
-      "loss": 46.0,
-      "step": 5631
-    },
-    {
-      "epoch": 0.906960827730585,
-      "grad_norm": 0.003796704113483429,
-      "learning_rate": 0.00019999959551324423,
-      "loss": 46.0,
-      "step": 5632
-    },
-    {
-      "epoch": 0.9071218648093724,
-      "grad_norm": 0.00047868990805000067,
-      "learning_rate": 0.0001999995953693373,
-      "loss": 46.0,
-      "step": 5633
-    },
-    {
-      "epoch": 0.9072829018881597,
-      "grad_norm": 0.002626352244988084,
-      "learning_rate": 0.00019999959522540474,
-      "loss": 46.0,
-      "step": 5634
-    },
-    {
-      "epoch": 0.9074439389669472,
-      "grad_norm": 0.003925809636712074,
-      "learning_rate": 0.0001999995950814466,
-      "loss": 46.0,
-      "step": 5635
-    },
-    {
-      "epoch": 0.9076049760457345,
-      "grad_norm": 0.0020973600912839174,
-      "learning_rate": 0.00019999959493746284,
-      "loss": 46.0,
-      "step": 5636
-    },
-    {
-      "epoch": 0.9077660131245219,
-      "grad_norm": 0.001194565906189382,
-      "learning_rate": 0.00019999959479345352,
-      "loss": 46.0,
-      "step": 5637
-    },
-    {
-      "epoch": 0.9079270502033093,
-      "grad_norm": 0.0007073345477692783,
-      "learning_rate": 0.0001999995946494186,
-      "loss": 46.0,
-      "step": 5638
-    },
-    {
-      "epoch": 0.9080880872820967,
-      "grad_norm": 0.0011025781277567148,
-      "learning_rate": 0.00019999959450535805,
-      "loss": 46.0,
-      "step": 5639
-    },
-    {
-      "epoch": 0.9082491243608841,
-      "grad_norm": 0.003514037001878023,
-      "learning_rate": 0.00019999959436127194,
-      "loss": 46.0,
-      "step": 5640
-    },
-    {
-      "epoch": 0.9084101614396715,
-      "grad_norm": 0.0011931579792872071,
-      "learning_rate": 0.00019999959421716023,
-      "loss": 46.0,
-      "step": 5641
-    },
-    {
-      "epoch": 0.9085711985184589,
-      "grad_norm": 0.0006472069071605802,
-      "learning_rate": 0.00019999959407302292,
-      "loss": 46.0,
-      "step": 5642
-    },
-    {
-      "epoch": 0.9087322355972463,
-      "grad_norm": 0.0013911984860897064,
-      "learning_rate": 0.00019999959392886,
-      "loss": 46.0,
-      "step": 5643
-    },
-    {
-      "epoch": 0.9088932726760337,
-      "grad_norm": 0.0010574093321338296,
-      "learning_rate": 0.0001999995937846715,
-      "loss": 46.0,
-      "step": 5644
-    },
-    {
-      "epoch": 0.909054309754821,
-      "grad_norm": 0.002115755807608366,
-      "learning_rate": 0.00019999959364045742,
-      "loss": 46.0,
-      "step": 5645
-    },
-    {
-      "epoch": 0.9092153468336084,
-      "grad_norm": 0.001006089965812862,
-      "learning_rate": 0.00019999959349621774,
-      "loss": 46.0,
-      "step": 5646
-    },
-    {
-      "epoch": 0.9093763839123958,
-      "grad_norm": 0.0014010483864694834,
-      "learning_rate": 0.00019999959335195244,
-      "loss": 46.0,
-      "step": 5647
-    },
-    {
-      "epoch": 0.9095374209911832,
-      "grad_norm": 0.0005963574512861669,
-      "learning_rate": 0.00019999959320766156,
-      "loss": 46.0,
-      "step": 5648
-    },
-    {
-      "epoch": 0.9096984580699706,
-      "grad_norm": 0.0020547525491565466,
-      "learning_rate": 0.0001999995930633451,
-      "loss": 46.0,
-      "step": 5649
-    },
-    {
-      "epoch": 0.909859495148758,
-      "grad_norm": 0.0019808122888207436,
-      "learning_rate": 0.000199999592919003,
-      "loss": 46.0,
-      "step": 5650
-    },
-    {
-      "epoch": 0.9100205322275454,
-      "grad_norm": 0.0020298794843256474,
-      "learning_rate": 0.00019999959277463536,
-      "loss": 46.0,
-      "step": 5651
-    },
-    {
-      "epoch": 0.9101815693063328,
-      "grad_norm": 0.000573399942368269,
-      "learning_rate": 0.00019999959263024207,
-      "loss": 46.0,
-      "step": 5652
-    },
-    {
-      "epoch": 0.9103426063851202,
-      "grad_norm": 0.005184104200452566,
-      "learning_rate": 0.00019999959248582326,
-      "loss": 46.0,
-      "step": 5653
-    },
-    {
-      "epoch": 0.9105036434639076,
-      "grad_norm": 0.0014675381826236844,
-      "learning_rate": 0.0001999995923413788,
-      "loss": 46.0,
-      "step": 5654
-    },
-    {
-      "epoch": 0.910664680542695,
-      "grad_norm": 0.002199732232838869,
-      "learning_rate": 0.00019999959219690875,
-      "loss": 46.0,
-      "step": 5655
-    },
-    {
-      "epoch": 0.9108257176214823,
-      "grad_norm": 0.0011606800835579634,
-      "learning_rate": 0.0001999995920524131,
-      "loss": 46.0,
-      "step": 5656
-    },
-    {
-      "epoch": 0.9109867547002697,
-      "grad_norm": 0.004471417050808668,
-      "learning_rate": 0.00019999959190789192,
-      "loss": 46.0,
-      "step": 5657
-    },
-    {
-      "epoch": 0.9111477917790571,
-      "grad_norm": 0.0007940044160932302,
-      "learning_rate": 0.00019999959176334505,
-      "loss": 46.0,
-      "step": 5658
-    },
-    {
-      "epoch": 0.9113088288578445,
-      "grad_norm": 0.000527512573171407,
-      "learning_rate": 0.00019999959161877263,
-      "loss": 46.0,
-      "step": 5659
-    },
-    {
-      "epoch": 0.9114698659366319,
-      "grad_norm": 0.0015550669049844146,
-      "learning_rate": 0.00019999959147417462,
-      "loss": 46.0,
-      "step": 5660
-    },
-    {
-      "epoch": 0.9116309030154193,
-      "grad_norm": 0.0014338825130835176,
-      "learning_rate": 0.00019999959132955102,
-      "loss": 46.0,
-      "step": 5661
-    },
-    {
-      "epoch": 0.9117919400942067,
-      "grad_norm": 0.0022002665791660547,
-      "learning_rate": 0.0001999995911849018,
-      "loss": 46.0,
-      "step": 5662
-    },
-    {
-      "epoch": 0.9119529771729941,
-      "grad_norm": 0.0035580561961978674,
-      "learning_rate": 0.000199999591040227,
-      "loss": 46.0,
-      "step": 5663
-    },
-    {
-      "epoch": 0.9121140142517815,
-      "grad_norm": 0.0014668663498014212,
-      "learning_rate": 0.0001999995908955266,
-      "loss": 46.0,
-      "step": 5664
-    },
-    {
-      "epoch": 0.9122750513305689,
-      "grad_norm": 0.005410606972873211,
-      "learning_rate": 0.00019999959075080065,
-      "loss": 46.0,
-      "step": 5665
-    },
-    {
-      "epoch": 0.9124360884093563,
-      "grad_norm": 0.0010199042735621333,
-      "learning_rate": 0.00019999959060604906,
-      "loss": 46.0,
-      "step": 5666
-    },
-    {
-      "epoch": 0.9125971254881436,
-      "grad_norm": 0.0006429336499422789,
-      "learning_rate": 0.00019999959046127186,
-      "loss": 46.0,
-      "step": 5667
-    },
-    {
-      "epoch": 0.912758162566931,
-      "grad_norm": 0.00044360265019349754,
-      "learning_rate": 0.0001999995903164691,
-      "loss": 46.0,
-      "step": 5668
-    },
-    {
-      "epoch": 0.9129191996457184,
-      "grad_norm": 0.003370842197909951,
-      "learning_rate": 0.00019999959017164072,
-      "loss": 46.0,
-      "step": 5669
-    },
-    {
-      "epoch": 0.9130802367245058,
-      "grad_norm": 0.0008861338719725609,
-      "learning_rate": 0.00019999959002678675,
-      "loss": 46.0,
-      "step": 5670
-    },
-    {
-      "epoch": 0.9132412738032932,
-      "grad_norm": 0.0012375286314636469,
-      "learning_rate": 0.0001999995898819072,
-      "loss": 46.0,
-      "step": 5671
-    },
-    {
-      "epoch": 0.9134023108820806,
-      "grad_norm": 0.0008010775782167912,
-      "learning_rate": 0.00019999958973700204,
-      "loss": 46.0,
-      "step": 5672
-    },
-    {
-      "epoch": 0.913563347960868,
-      "grad_norm": 0.0012551102554425597,
-      "learning_rate": 0.0001999995895920713,
-      "loss": 46.0,
-      "step": 5673
-    },
-    {
-      "epoch": 0.9137243850396554,
-      "grad_norm": 0.0006155106239020824,
-      "learning_rate": 0.00019999958944711495,
-      "loss": 46.0,
-      "step": 5674
-    },
-    {
-      "epoch": 0.9138854221184428,
-      "grad_norm": 0.0007423459901474416,
-      "learning_rate": 0.00019999958930213302,
-      "loss": 46.0,
-      "step": 5675
-    },
-    {
-      "epoch": 0.9140464591972302,
-      "grad_norm": 0.0007193029741756618,
-      "learning_rate": 0.00019999958915712548,
-      "loss": 46.0,
-      "step": 5676
-    },
-    {
-      "epoch": 0.9142074962760176,
-      "grad_norm": 0.0012417606776580215,
-      "learning_rate": 0.00019999958901209235,
-      "loss": 46.0,
-      "step": 5677
-    },
-    {
-      "epoch": 0.914368533354805,
-      "grad_norm": 0.0021162887569516897,
-      "learning_rate": 0.00019999958886703363,
-      "loss": 46.0,
-      "step": 5678
-    },
-    {
-      "epoch": 0.9145295704335923,
-      "grad_norm": 0.0016738566337153316,
-      "learning_rate": 0.00019999958872194933,
-      "loss": 46.0,
-      "step": 5679
-    },
-    {
-      "epoch": 0.9146906075123797,
-      "grad_norm": 0.001967610325664282,
-      "learning_rate": 0.00019999958857683941,
-      "loss": 46.0,
-      "step": 5680
-    },
-    {
-      "epoch": 0.9148516445911671,
-      "grad_norm": 0.0006160161574371159,
-      "learning_rate": 0.0001999995884317039,
-      "loss": 46.0,
-      "step": 5681
-    },
-    {
-      "epoch": 0.9150126816699545,
-      "grad_norm": 0.000579670537263155,
-      "learning_rate": 0.00019999958828654282,
-      "loss": 46.0,
-      "step": 5682
-    },
-    {
-      "epoch": 0.9151737187487419,
-      "grad_norm": 0.0010190429165959358,
-      "learning_rate": 0.0001999995881413561,
-      "loss": 46.0,
-      "step": 5683
-    },
-    {
-      "epoch": 0.9153347558275293,
-      "grad_norm": 0.0032319920137524605,
-      "learning_rate": 0.00019999958799614382,
-      "loss": 46.0,
-      "step": 5684
-    },
-    {
-      "epoch": 0.9154957929063167,
-      "grad_norm": 0.001973667647689581,
-      "learning_rate": 0.00019999958785090594,
-      "loss": 46.0,
-      "step": 5685
-    },
-    {
-      "epoch": 0.9156568299851041,
-      "grad_norm": 0.0008971799979917705,
-      "learning_rate": 0.00019999958770564247,
-      "loss": 46.0,
-      "step": 5686
-    },
-    {
-      "epoch": 0.9158178670638915,
-      "grad_norm": 0.0017805973766371608,
-      "learning_rate": 0.00019999958756035337,
-      "loss": 46.0,
-      "step": 5687
-    },
-    {
-      "epoch": 0.9159789041426789,
-      "grad_norm": 0.0023819077759981155,
-      "learning_rate": 0.0001999995874150387,
-      "loss": 46.0,
-      "step": 5688
-    },
-    {
-      "epoch": 0.9161399412214662,
-      "grad_norm": 0.0023476211354136467,
-      "learning_rate": 0.00019999958726969844,
-      "loss": 46.0,
-      "step": 5689
-    },
-    {
-      "epoch": 0.9163009783002536,
-      "grad_norm": 0.0005144411697983742,
-      "learning_rate": 0.00019999958712433257,
-      "loss": 46.0,
-      "step": 5690
-    },
-    {
-      "epoch": 0.916462015379041,
-      "grad_norm": 0.0013660414842888713,
-      "learning_rate": 0.00019999958697894114,
-      "loss": 46.0,
-      "step": 5691
-    },
-    {
-      "epoch": 0.9166230524578284,
-      "grad_norm": 0.003989432007074356,
-      "learning_rate": 0.0001999995868335241,
-      "loss": 46.0,
-      "step": 5692
-    },
-    {
-      "epoch": 0.9167840895366158,
-      "grad_norm": 0.001444809720851481,
-      "learning_rate": 0.00019999958668808147,
-      "loss": 46.0,
-      "step": 5693
-    },
-    {
-      "epoch": 0.9169451266154032,
-      "grad_norm": 0.0006382620777003467,
-      "learning_rate": 0.0001999995865426132,
-      "loss": 46.0,
-      "step": 5694
-    },
-    {
-      "epoch": 0.9171061636941906,
-      "grad_norm": 0.005084271542727947,
-      "learning_rate": 0.00019999958639711936,
-      "loss": 46.0,
-      "step": 5695
-    },
-    {
-      "epoch": 0.917267200772978,
-      "grad_norm": 0.0015088410582393408,
-      "learning_rate": 0.00019999958625159994,
-      "loss": 46.0,
-      "step": 5696
-    },
-    {
-      "epoch": 0.9174282378517654,
-      "grad_norm": 0.0015899842837825418,
-      "learning_rate": 0.00019999958610605493,
-      "loss": 46.0,
-      "step": 5697
-    },
-    {
-      "epoch": 0.9175892749305528,
-      "grad_norm": 0.0006153786671347916,
-      "learning_rate": 0.0001999995859604843,
-      "loss": 46.0,
-      "step": 5698
-    },
-    {
-      "epoch": 0.9177503120093401,
-      "grad_norm": 0.0024292119778692722,
-      "learning_rate": 0.0001999995858148881,
-      "loss": 46.0,
-      "step": 5699
-    },
-    {
-      "epoch": 0.9179113490881275,
-      "grad_norm": 0.0004447857791092247,
-      "learning_rate": 0.00019999958566926628,
-      "loss": 46.0,
-      "step": 5700
-    },
-    {
-      "epoch": 0.9180723861669149,
-      "grad_norm": 0.0008880120585672557,
-      "learning_rate": 0.00019999958552361887,
-      "loss": 46.0,
-      "step": 5701
-    },
-    {
-      "epoch": 0.9182334232457023,
-      "grad_norm": 0.0012786429142579436,
-      "learning_rate": 0.0001999995853779459,
-      "loss": 46.0,
-      "step": 5702
-    },
-    {
-      "epoch": 0.9183944603244897,
-      "grad_norm": 0.0025664877612143755,
-      "learning_rate": 0.0001999995852322473,
-      "loss": 46.0,
-      "step": 5703
-    },
-    {
-      "epoch": 0.9185554974032771,
-      "grad_norm": 0.0007213133503682911,
-      "learning_rate": 0.0001999995850865231,
-      "loss": 46.0,
-      "step": 5704
-    },
-    {
-      "epoch": 0.9187165344820645,
-      "grad_norm": 0.0006415305542759597,
-      "learning_rate": 0.00019999958494077333,
-      "loss": 46.0,
-      "step": 5705
-    },
-    {
-      "epoch": 0.9188775715608519,
-      "grad_norm": 0.0004968970897607505,
-      "learning_rate": 0.00019999958479499796,
-      "loss": 46.0,
-      "step": 5706
-    },
-    {
-      "epoch": 0.9190386086396393,
-      "grad_norm": 0.0013987241545692086,
-      "learning_rate": 0.000199999584649197,
-      "loss": 46.0,
-      "step": 5707
-    },
-    {
-      "epoch": 0.9191996457184267,
-      "grad_norm": 0.0016386124771088362,
-      "learning_rate": 0.00019999958450337042,
-      "loss": 46.0,
-      "step": 5708
-    },
-    {
-      "epoch": 0.9193606827972141,
-      "grad_norm": 0.0006625541718676686,
-      "learning_rate": 0.00019999958435751826,
-      "loss": 46.0,
-      "step": 5709
-    },
-    {
-      "epoch": 0.9195217198760014,
-      "grad_norm": 0.0021546788047999144,
-      "learning_rate": 0.0001999995842116405,
-      "loss": 46.0,
-      "step": 5710
-    },
-    {
-      "epoch": 0.9196827569547888,
-      "grad_norm": 0.0012270796578377485,
-      "learning_rate": 0.00019999958406573715,
-      "loss": 46.0,
-      "step": 5711
-    },
-    {
-      "epoch": 0.9198437940335762,
-      "grad_norm": 0.0015551653923466802,
-      "learning_rate": 0.00019999958391980822,
-      "loss": 46.0,
-      "step": 5712
-    },
-    {
-      "epoch": 0.9200048311123636,
-      "grad_norm": 0.0005300075863488019,
-      "learning_rate": 0.0001999995837738537,
-      "loss": 46.0,
-      "step": 5713
-    },
-    {
-      "epoch": 0.920165868191151,
-      "grad_norm": 0.0013175175990909338,
-      "learning_rate": 0.00019999958362787354,
-      "loss": 46.0,
-      "step": 5714
-    },
-    {
-      "epoch": 0.9203269052699384,
-      "grad_norm": 0.0015574973076581955,
-      "learning_rate": 0.00019999958348186782,
-      "loss": 46.0,
-      "step": 5715
-    },
-    {
-      "epoch": 0.9204879423487258,
-      "grad_norm": 0.0014362430665642023,
-      "learning_rate": 0.00019999958333583647,
-      "loss": 46.0,
-      "step": 5716
-    },
-    {
-      "epoch": 0.9206489794275132,
-      "grad_norm": 0.001071386388503015,
-      "learning_rate": 0.00019999958318977958,
-      "loss": 46.0,
-      "step": 5717
-    },
-    {
-      "epoch": 0.9208100165063006,
-      "grad_norm": 0.0031662776600569487,
-      "learning_rate": 0.00019999958304369706,
-      "loss": 46.0,
-      "step": 5718
-    },
-    {
-      "epoch": 0.920971053585088,
-      "grad_norm": 0.002556638093665242,
-      "learning_rate": 0.00019999958289758897,
-      "loss": 46.0,
-      "step": 5719
-    },
-    {
-      "epoch": 0.9211320906638754,
-      "grad_norm": 0.0006898163119331002,
-      "learning_rate": 0.00019999958275145527,
-      "loss": 46.0,
-      "step": 5720
-    },
-    {
-      "epoch": 0.9212931277426627,
-      "grad_norm": 0.003467016154900193,
-      "learning_rate": 0.00019999958260529598,
-      "loss": 46.0,
-      "step": 5721
-    },
-    {
-      "epoch": 0.9214541648214502,
-      "grad_norm": 0.0011490893084555864,
-      "learning_rate": 0.00019999958245911107,
-      "loss": 46.0,
-      "step": 5722
-    },
-    {
-      "epoch": 0.9216152019002375,
-      "grad_norm": 0.001350061153061688,
-      "learning_rate": 0.00019999958231290058,
-      "loss": 46.0,
-      "step": 5723
-    },
-    {
-      "epoch": 0.9217762389790249,
-      "grad_norm": 0.002056166995316744,
-      "learning_rate": 0.0001999995821666645,
-      "loss": 46.0,
-      "step": 5724
-    },
-    {
-      "epoch": 0.9219372760578123,
-      "grad_norm": 0.004254150204360485,
-      "learning_rate": 0.00019999958202040284,
-      "loss": 46.0,
-      "step": 5725
-    },
-    {
-      "epoch": 0.9220983131365997,
-      "grad_norm": 0.0024846720043569803,
-      "learning_rate": 0.00019999958187411556,
-      "loss": 46.0,
-      "step": 5726
-    },
-    {
-      "epoch": 0.9222593502153871,
-      "grad_norm": 0.0014163716696202755,
-      "learning_rate": 0.0001999995817278027,
-      "loss": 46.0,
-      "step": 5727
-    },
-    {
-      "epoch": 0.9224203872941745,
-      "grad_norm": 0.0003627694386523217,
-      "learning_rate": 0.00019999958158146424,
-      "loss": 46.0,
-      "step": 5728
-    },
-    {
-      "epoch": 0.9225814243729619,
-      "grad_norm": 0.003675812855362892,
-      "learning_rate": 0.00019999958143510017,
-      "loss": 46.0,
-      "step": 5729
-    },
-    {
-      "epoch": 0.9227424614517493,
-      "grad_norm": 0.0007954856264404953,
-      "learning_rate": 0.00019999958128871055,
-      "loss": 46.0,
-      "step": 5730
-    },
-    {
-      "epoch": 0.9229034985305367,
-      "grad_norm": 0.0010724146850407124,
-      "learning_rate": 0.00019999958114229528,
-      "loss": 46.0,
-      "step": 5731
-    },
-    {
-      "epoch": 0.923064535609324,
-      "grad_norm": 0.0024460607673972845,
-      "learning_rate": 0.00019999958099585445,
-      "loss": 46.0,
-      "step": 5732
-    },
-    {
-      "epoch": 0.9232255726881115,
-      "grad_norm": 0.0005035282811149955,
-      "learning_rate": 0.000199999580849388,
-      "loss": 46.0,
-      "step": 5733
-    },
-    {
-      "epoch": 0.9233866097668988,
-      "grad_norm": 0.001239884877577424,
-      "learning_rate": 0.000199999580702896,
-      "loss": 46.0,
-      "step": 5734
-    },
-    {
-      "epoch": 0.9235476468456862,
-      "grad_norm": 0.004846284165978432,
-      "learning_rate": 0.00019999958055637836,
-      "loss": 46.0,
-      "step": 5735
-    },
-    {
-      "epoch": 0.9237086839244736,
-      "grad_norm": 0.0009243261883966625,
-      "learning_rate": 0.00019999958040983512,
-      "loss": 46.0,
-      "step": 5736
-    },
-    {
-      "epoch": 0.923869721003261,
-      "grad_norm": 0.0006513491389341652,
-      "learning_rate": 0.00019999958026326633,
-      "loss": 46.0,
-      "step": 5737
-    },
-    {
-      "epoch": 0.9240307580820484,
-      "grad_norm": 0.0007005998049862683,
-      "learning_rate": 0.00019999958011667192,
-      "loss": 46.0,
-      "step": 5738
-    },
-    {
-      "epoch": 0.9241917951608358,
-      "grad_norm": 0.0010139001533389091,
-      "learning_rate": 0.00019999957997005193,
-      "loss": 46.0,
-      "step": 5739
-    },
-    {
-      "epoch": 0.9243528322396232,
-      "grad_norm": 0.0007072404841892421,
-      "learning_rate": 0.00019999957982340632,
-      "loss": 46.0,
-      "step": 5740
-    },
-    {
-      "epoch": 0.9245138693184106,
-      "grad_norm": 0.0004549682780634612,
-      "learning_rate": 0.00019999957967673515,
-      "loss": 46.0,
-      "step": 5741
-    },
-    {
-      "epoch": 0.924674906397198,
-      "grad_norm": 0.00048193742986768484,
-      "learning_rate": 0.00019999957953003837,
-      "loss": 46.0,
-      "step": 5742
-    },
-    {
-      "epoch": 0.9248359434759853,
-      "grad_norm": 0.0022708552423864603,
-      "learning_rate": 0.00019999957938331598,
-      "loss": 46.0,
-      "step": 5743
-    },
-    {
-      "epoch": 0.9249969805547728,
-      "grad_norm": 0.0005292826099321246,
-      "learning_rate": 0.000199999579236568,
-      "loss": 46.0,
-      "step": 5744
-    },
-    {
-      "epoch": 0.9251580176335601,
-      "grad_norm": 0.0009936413262039423,
-      "learning_rate": 0.00019999957908979442,
-      "loss": 46.0,
-      "step": 5745
-    },
-    {
-      "epoch": 0.9253190547123475,
-      "grad_norm": 0.0023525094147771597,
-      "learning_rate": 0.00019999957894299526,
-      "loss": 46.0,
-      "step": 5746
-    },
-    {
-      "epoch": 0.9254800917911349,
-      "grad_norm": 0.0060401796363294125,
-      "learning_rate": 0.00019999957879617052,
-      "loss": 46.0,
-      "step": 5747
-    },
-    {
-      "epoch": 0.9256411288699223,
-      "grad_norm": 0.0020601237192749977,
-      "learning_rate": 0.00019999957864932016,
-      "loss": 46.0,
-      "step": 5748
-    },
-    {
-      "epoch": 0.9258021659487097,
-      "grad_norm": 0.0006590397679246962,
-      "learning_rate": 0.00019999957850244418,
-      "loss": 46.0,
-      "step": 5749
-    },
-    {
-      "epoch": 0.9259632030274971,
-      "grad_norm": 0.0006065714405849576,
-      "learning_rate": 0.00019999957835554265,
-      "loss": 46.0,
-      "step": 5750
-    },
-    {
-      "epoch": 0.9261242401062845,
-      "grad_norm": 0.000712477311026305,
-      "learning_rate": 0.00019999957820861553,
-      "loss": 46.0,
-      "step": 5751
-    },
-    {
-      "epoch": 0.9262852771850718,
-      "grad_norm": 0.00225071026943624,
-      "learning_rate": 0.00019999957806166277,
-      "loss": 46.0,
-      "step": 5752
-    },
-    {
-      "epoch": 0.9264463142638593,
-      "grad_norm": 0.00215413561090827,
-      "learning_rate": 0.00019999957791468444,
-      "loss": 46.0,
-      "step": 5753
-    },
-    {
-      "epoch": 0.9266073513426466,
-      "grad_norm": 0.0007556015043519437,
-      "learning_rate": 0.00019999957776768053,
-      "loss": 46.0,
-      "step": 5754
-    },
-    {
-      "epoch": 0.926768388421434,
-      "grad_norm": 0.002231378573924303,
-      "learning_rate": 0.000199999577620651,
-      "loss": 46.0,
-      "step": 5755
-    },
-    {
-      "epoch": 0.9269294255002214,
-      "grad_norm": 0.0008061740081757307,
-      "learning_rate": 0.00019999957747359587,
-      "loss": 46.0,
-      "step": 5756
-    },
-    {
-      "epoch": 0.9270904625790088,
-      "grad_norm": 0.00124307069927454,
-      "learning_rate": 0.00019999957732651517,
-      "loss": 46.0,
-      "step": 5757
-    },
-    {
-      "epoch": 0.9272514996577962,
-      "grad_norm": 0.0009388653561472893,
-      "learning_rate": 0.0001999995771794089,
-      "loss": 46.0,
-      "step": 5758
-    },
-    {
-      "epoch": 0.9274125367365836,
-      "grad_norm": 0.00048344730748794973,
-      "learning_rate": 0.000199999577032277,
-      "loss": 46.0,
-      "step": 5759
-    },
-    {
-      "epoch": 0.927573573815371,
-      "grad_norm": 0.0006193193839862943,
-      "learning_rate": 0.0001999995768851195,
-      "loss": 46.0,
-      "step": 5760
-    },
-    {
-      "epoch": 0.9277346108941584,
-      "grad_norm": 0.0007393594132736325,
-      "learning_rate": 0.0001999995767379364,
-      "loss": 46.0,
-      "step": 5761
-    },
-    {
-      "epoch": 0.9278956479729458,
-      "grad_norm": 0.0015328206354752183,
-      "learning_rate": 0.0001999995765907277,
-      "loss": 46.0,
-      "step": 5762
-    },
-    {
-      "epoch": 0.9280566850517331,
-      "grad_norm": 0.00216746237128973,
-      "learning_rate": 0.00019999957644349344,
-      "loss": 46.0,
-      "step": 5763
-    },
-    {
-      "epoch": 0.9282177221305206,
-      "grad_norm": 0.0024387789890170097,
-      "learning_rate": 0.00019999957629623357,
-      "loss": 46.0,
-      "step": 5764
-    },
-    {
-      "epoch": 0.9283787592093079,
-      "grad_norm": 0.00032075357739813626,
-      "learning_rate": 0.0001999995761489481,
-      "loss": 46.0,
-      "step": 5765
-    },
-    {
-      "epoch": 0.9285397962880954,
-      "grad_norm": 0.0038370888214558363,
-      "learning_rate": 0.00019999957600163706,
-      "loss": 46.0,
-      "step": 5766
-    },
-    {
-      "epoch": 0.9287008333668827,
-      "grad_norm": 0.002521386370062828,
-      "learning_rate": 0.00019999957585430038,
-      "loss": 46.0,
-      "step": 5767
-    },
-    {
-      "epoch": 0.9288618704456701,
-      "grad_norm": 0.0012193474685773253,
-      "learning_rate": 0.00019999957570693814,
-      "loss": 46.0,
-      "step": 5768
-    },
-    {
-      "epoch": 0.9290229075244575,
-      "grad_norm": 0.0020336946472525597,
-      "learning_rate": 0.0001999995755595503,
-      "loss": 46.0,
-      "step": 5769
-    },
-    {
-      "epoch": 0.9291839446032449,
-      "grad_norm": 0.0009782103588804603,
-      "learning_rate": 0.00019999957541213685,
-      "loss": 46.0,
-      "step": 5770
-    },
-    {
-      "epoch": 0.9293449816820323,
-      "grad_norm": 0.0008099670521914959,
-      "learning_rate": 0.00019999957526469782,
-      "loss": 46.0,
-      "step": 5771
-    },
-    {
-      "epoch": 0.9295060187608197,
-      "grad_norm": 0.0004197184753138572,
-      "learning_rate": 0.0001999995751172332,
-      "loss": 46.0,
-      "step": 5772
-    },
-    {
-      "epoch": 0.9296670558396071,
-      "grad_norm": 0.0006177496397867799,
-      "learning_rate": 0.00019999957496974298,
-      "loss": 46.0,
-      "step": 5773
-    },
-    {
-      "epoch": 0.9298280929183944,
-      "grad_norm": 0.0006528380909003317,
-      "learning_rate": 0.00019999957482222716,
-      "loss": 46.0,
-      "step": 5774
-    },
-    {
-      "epoch": 0.9299891299971819,
-      "grad_norm": 0.0010686871828511357,
-      "learning_rate": 0.00019999957467468576,
-      "loss": 46.0,
-      "step": 5775
-    },
-    {
-      "epoch": 0.9301501670759692,
-      "grad_norm": 0.0006162556237541139,
-      "learning_rate": 0.00019999957452711872,
-      "loss": 46.0,
-      "step": 5776
-    },
-    {
-      "epoch": 0.9303112041547567,
-      "grad_norm": 0.0004212743660900742,
-      "learning_rate": 0.0001999995743795261,
-      "loss": 46.0,
-      "step": 5777
-    },
-    {
-      "epoch": 0.930472241233544,
-      "grad_norm": 0.0004566374991554767,
-      "learning_rate": 0.00019999957423190795,
-      "loss": 46.0,
-      "step": 5778
-    },
-    {
-      "epoch": 0.9306332783123314,
-      "grad_norm": 0.000821933033876121,
-      "learning_rate": 0.00019999957408426414,
-      "loss": 46.0,
-      "step": 5779
-    },
-    {
-      "epoch": 0.9307943153911188,
-      "grad_norm": 0.0017378123011440039,
-      "learning_rate": 0.00019999957393659475,
-      "loss": 46.0,
-      "step": 5780
-    },
-    {
-      "epoch": 0.9309553524699062,
-      "grad_norm": 0.0002255280560348183,
-      "learning_rate": 0.00019999957378889977,
-      "loss": 46.0,
-      "step": 5781
-    },
-    {
-      "epoch": 0.9311163895486936,
-      "grad_norm": 0.005280837882310152,
-      "learning_rate": 0.00019999957364117918,
-      "loss": 46.0,
-      "step": 5782
-    },
-    {
-      "epoch": 0.931277426627481,
-      "grad_norm": 0.0015987369697540998,
-      "learning_rate": 0.00019999957349343302,
-      "loss": 46.0,
-      "step": 5783
-    },
-    {
-      "epoch": 0.9314384637062684,
-      "grad_norm": 0.002849208889529109,
-      "learning_rate": 0.00019999957334566125,
-      "loss": 46.0,
-      "step": 5784
-    },
-    {
-      "epoch": 0.9315995007850557,
-      "grad_norm": 0.0006868364871479571,
-      "learning_rate": 0.0001999995731978639,
-      "loss": 46.0,
-      "step": 5785
-    },
-    {
-      "epoch": 0.9317605378638432,
-      "grad_norm": 0.003278016345575452,
-      "learning_rate": 0.00019999957305004093,
-      "loss": 46.0,
-      "step": 5786
-    },
-    {
-      "epoch": 0.9319215749426305,
-      "grad_norm": 0.0007232369389384985,
-      "learning_rate": 0.0001999995729021924,
-      "loss": 46.0,
-      "step": 5787
-    },
-    {
-      "epoch": 0.932082612021418,
-      "grad_norm": 0.002182962140068412,
-      "learning_rate": 0.00019999957275431825,
-      "loss": 46.0,
-      "step": 5788
-    },
-    {
-      "epoch": 0.9322436491002053,
-      "grad_norm": 0.002265210961923003,
-      "learning_rate": 0.00019999957260641852,
-      "loss": 46.0,
-      "step": 5789
-    },
-    {
-      "epoch": 0.9324046861789927,
-      "grad_norm": 0.002262046094983816,
-      "learning_rate": 0.00019999957245849317,
-      "loss": 46.0,
-      "step": 5790
-    },
-    {
-      "epoch": 0.9325657232577801,
-      "grad_norm": 0.005041072610765696,
-      "learning_rate": 0.00019999957231054224,
-      "loss": 46.0,
-      "step": 5791
-    },
-    {
-      "epoch": 0.9327267603365675,
-      "grad_norm": 0.0015471072401851416,
-      "learning_rate": 0.00019999957216256572,
-      "loss": 46.0,
-      "step": 5792
-    },
-    {
-      "epoch": 0.9328877974153549,
-      "grad_norm": 0.0007371532265096903,
-      "learning_rate": 0.00019999957201456358,
-      "loss": 46.0,
-      "step": 5793
-    },
-    {
-      "epoch": 0.9330488344941423,
-      "grad_norm": 0.003574784379452467,
-      "learning_rate": 0.00019999957186653586,
-      "loss": 46.0,
-      "step": 5794
-    },
-    {
-      "epoch": 0.9332098715729297,
-      "grad_norm": 0.0003592149878386408,
-      "learning_rate": 0.00019999957171848255,
-      "loss": 46.0,
-      "step": 5795
-    },
-    {
-      "epoch": 0.933370908651717,
-      "grad_norm": 0.0006847637705504894,
-      "learning_rate": 0.00019999957157040369,
-      "loss": 46.0,
-      "step": 5796
-    },
-    {
-      "epoch": 0.9335319457305045,
-      "grad_norm": 0.0038793254643678665,
-      "learning_rate": 0.00019999957142229917,
-      "loss": 46.0,
-      "step": 5797
-    },
-    {
-      "epoch": 0.9336929828092918,
-      "grad_norm": 0.0005826210253871977,
-      "learning_rate": 0.00019999957127416908,
-      "loss": 46.0,
-      "step": 5798
-    },
-    {
-      "epoch": 0.9338540198880793,
-      "grad_norm": 0.001212799339555204,
-      "learning_rate": 0.00019999957112601337,
-      "loss": 46.0,
-      "step": 5799
-    },
-    {
-      "epoch": 0.9340150569668666,
-      "grad_norm": 0.0006134874420240521,
-      "learning_rate": 0.00019999957097783212,
-      "loss": 46.0,
-      "step": 5800
-    },
-    {
-      "epoch": 0.934176094045654,
-      "grad_norm": 0.0017032917821779847,
-      "learning_rate": 0.0001999995708296252,
-      "loss": 46.0,
-      "step": 5801
-    },
-    {
-      "epoch": 0.9343371311244414,
-      "grad_norm": 0.005153973586857319,
-      "learning_rate": 0.00019999957068139276,
-      "loss": 46.0,
-      "step": 5802
-    },
-    {
-      "epoch": 0.9344981682032288,
-      "grad_norm": 0.004429292399436235,
-      "learning_rate": 0.0001999995705331347,
-      "loss": 46.0,
-      "step": 5803
-    },
-    {
-      "epoch": 0.9346592052820162,
-      "grad_norm": 0.0010363658657297492,
-      "learning_rate": 0.00019999957038485105,
-      "loss": 46.0,
-      "step": 5804
-    },
-    {
-      "epoch": 0.9348202423608035,
-      "grad_norm": 0.0014482052065432072,
-      "learning_rate": 0.00019999957023654176,
-      "loss": 46.0,
-      "step": 5805
-    },
-    {
-      "epoch": 0.934981279439591,
-      "grad_norm": 0.008390652947127819,
-      "learning_rate": 0.00019999957008820691,
-      "loss": 46.0,
-      "step": 5806
-    },
-    {
-      "epoch": 0.9351423165183783,
-      "grad_norm": 0.00042950219358317554,
-      "learning_rate": 0.00019999956993984648,
-      "loss": 46.0,
-      "step": 5807
-    },
-    {
-      "epoch": 0.9353033535971658,
-      "grad_norm": 0.0008071645861491561,
-      "learning_rate": 0.00019999956979146043,
-      "loss": 46.0,
-      "step": 5808
-    },
-    {
-      "epoch": 0.9354643906759531,
-      "grad_norm": 0.0005485149449668825,
-      "learning_rate": 0.0001999995696430488,
-      "loss": 46.0,
-      "step": 5809
-    },
-    {
-      "epoch": 0.9356254277547406,
-      "grad_norm": 0.0007950183353386819,
-      "learning_rate": 0.00019999956949461156,
-      "loss": 46.0,
-      "step": 5810
-    },
-    {
-      "epoch": 0.9357864648335279,
-      "grad_norm": 0.0004549167933873832,
-      "learning_rate": 0.00019999956934614875,
-      "loss": 46.0,
-      "step": 5811
-    },
-    {
-      "epoch": 0.9359475019123153,
-      "grad_norm": 0.0013707609614357352,
-      "learning_rate": 0.00019999956919766033,
-      "loss": 46.0,
-      "step": 5812
-    },
-    {
-      "epoch": 0.9361085389911027,
-      "grad_norm": 0.0006515044951811433,
-      "learning_rate": 0.00019999956904914629,
-      "loss": 46.0,
-      "step": 5813
-    },
-    {
-      "epoch": 0.9362695760698901,
-      "grad_norm": 0.002092970535159111,
-      "learning_rate": 0.0001999995689006067,
-      "loss": 46.0,
-      "step": 5814
-    },
-    {
-      "epoch": 0.9364306131486775,
-      "grad_norm": 0.0017171568470075727,
-      "learning_rate": 0.00019999956875204152,
-      "loss": 46.0,
-      "step": 5815
-    },
-    {
-      "epoch": 0.9365916502274648,
-      "grad_norm": 0.0009057503193616867,
-      "learning_rate": 0.0001999995686034507,
-      "loss": 46.0,
-      "step": 5816
-    },
-    {
-      "epoch": 0.9367526873062523,
-      "grad_norm": 0.0009272541501559317,
-      "learning_rate": 0.0001999995684548343,
-      "loss": 46.0,
-      "step": 5817
-    },
-    {
-      "epoch": 0.9369137243850396,
-      "grad_norm": 0.001422026427462697,
-      "learning_rate": 0.00019999956830619233,
-      "loss": 46.0,
-      "step": 5818
-    },
-    {
-      "epoch": 0.9370747614638271,
-      "grad_norm": 0.0013472591526806355,
-      "learning_rate": 0.0001999995681575247,
-      "loss": 46.0,
-      "step": 5819
-    },
-    {
-      "epoch": 0.9372357985426144,
-      "grad_norm": 0.0024569458328187466,
-      "learning_rate": 0.00019999956800883156,
-      "loss": 46.0,
-      "step": 5820
-    },
-    {
-      "epoch": 0.9373968356214019,
-      "grad_norm": 0.0018863404402509332,
-      "learning_rate": 0.00019999956786011277,
-      "loss": 46.0,
-      "step": 5821
-    },
-    {
-      "epoch": 0.9375578727001892,
-      "grad_norm": 0.0006114105344749987,
-      "learning_rate": 0.00019999956771136841,
-      "loss": 46.0,
-      "step": 5822
-    },
-    {
-      "epoch": 0.9377189097789767,
-      "grad_norm": 0.0005984083982184529,
-      "learning_rate": 0.00019999956756259845,
-      "loss": 46.0,
-      "step": 5823
-    },
-    {
-      "epoch": 0.937879946857764,
-      "grad_norm": 0.0016396831488236785,
-      "learning_rate": 0.0001999995674138029,
-      "loss": 46.0,
-      "step": 5824
-    },
-    {
-      "epoch": 0.9380409839365514,
-      "grad_norm": 0.0022460962645709515,
-      "learning_rate": 0.00019999956726498175,
-      "loss": 46.0,
-      "step": 5825
-    },
-    {
-      "epoch": 0.9382020210153388,
-      "grad_norm": 0.001208878238685429,
-      "learning_rate": 0.000199999567116135,
-      "loss": 46.0,
-      "step": 5826
-    },
-    {
-      "epoch": 0.9383630580941261,
-      "grad_norm": 0.00513978349044919,
-      "learning_rate": 0.00019999956696726266,
-      "loss": 46.0,
-      "step": 5827
-    },
-    {
-      "epoch": 0.9385240951729136,
-      "grad_norm": 0.001671300269663334,
-      "learning_rate": 0.00019999956681836473,
-      "loss": 46.0,
-      "step": 5828
-    },
-    {
-      "epoch": 0.9386851322517009,
-      "grad_norm": 0.0008806994301266968,
-      "learning_rate": 0.0001999995666694412,
-      "loss": 46.0,
-      "step": 5829
-    },
-    {
-      "epoch": 0.9388461693304884,
-      "grad_norm": 0.0012135577853769064,
-      "learning_rate": 0.00019999956652049208,
-      "loss": 46.0,
-      "step": 5830
-    },
-    {
-      "epoch": 0.9390072064092757,
-      "grad_norm": 0.006969158537685871,
-      "learning_rate": 0.00019999956637151733,
-      "loss": 46.0,
-      "step": 5831
-    },
-    {
-      "epoch": 0.9391682434880632,
-      "grad_norm": 0.0003466187044978142,
-      "learning_rate": 0.00019999956622251703,
-      "loss": 46.0,
-      "step": 5832
-    },
-    {
-      "epoch": 0.9393292805668505,
-      "grad_norm": 0.0009236810728907585,
-      "learning_rate": 0.00019999956607349114,
-      "loss": 46.0,
-      "step": 5833
-    },
-    {
-      "epoch": 0.939490317645638,
-      "grad_norm": 0.0006928643560968339,
-      "learning_rate": 0.00019999956592443963,
-      "loss": 46.0,
-      "step": 5834
-    },
-    {
-      "epoch": 0.9396513547244253,
-      "grad_norm": 0.0008363567176274955,
-      "learning_rate": 0.0001999995657753625,
-      "loss": 46.0,
-      "step": 5835
-    },
-    {
-      "epoch": 0.9398123918032127,
-      "grad_norm": 0.004356647375971079,
-      "learning_rate": 0.00019999956562625983,
-      "loss": 46.0,
-      "step": 5836
-    },
-    {
-      "epoch": 0.9399734288820001,
-      "grad_norm": 0.0008948463364504278,
-      "learning_rate": 0.00019999956547713153,
-      "loss": 46.0,
-      "step": 5837
-    },
-    {
-      "epoch": 0.9401344659607874,
-      "grad_norm": 0.0028034490533173084,
-      "learning_rate": 0.00019999956532797765,
-      "loss": 46.0,
-      "step": 5838
-    },
-    {
-      "epoch": 0.9402955030395749,
-      "grad_norm": 0.004898726940155029,
-      "learning_rate": 0.00019999956517879818,
-      "loss": 46.0,
-      "step": 5839
-    },
-    {
-      "epoch": 0.9404565401183622,
-      "grad_norm": 0.0006895270198583603,
-      "learning_rate": 0.00019999956502959312,
-      "loss": 46.0,
-      "step": 5840
-    },
-    {
-      "epoch": 0.9406175771971497,
-      "grad_norm": 0.001562860794365406,
-      "learning_rate": 0.00019999956488036245,
-      "loss": 46.0,
-      "step": 5841
-    },
-    {
-      "epoch": 0.940778614275937,
-      "grad_norm": 0.000748677586670965,
-      "learning_rate": 0.00019999956473110616,
-      "loss": 46.0,
-      "step": 5842
-    },
-    {
-      "epoch": 0.9409396513547245,
-      "grad_norm": 0.0008708340465091169,
-      "learning_rate": 0.00019999956458182432,
-      "loss": 46.0,
-      "step": 5843
-    },
-    {
-      "epoch": 0.9411006884335118,
-      "grad_norm": 0.00749856187030673,
-      "learning_rate": 0.00019999956443251686,
-      "loss": 46.0,
-      "step": 5844
-    },
-    {
-      "epoch": 0.9412617255122993,
-      "grad_norm": 0.0004217370005790144,
-      "learning_rate": 0.00019999956428318384,
-      "loss": 46.0,
-      "step": 5845
-    },
-    {
-      "epoch": 0.9414227625910866,
-      "grad_norm": 0.0010512169683352113,
-      "learning_rate": 0.00019999956413382518,
-      "loss": 46.0,
-      "step": 5846
-    },
-    {
-      "epoch": 0.9415837996698739,
-      "grad_norm": 0.0006663579843007028,
-      "learning_rate": 0.00019999956398444093,
-      "loss": 46.0,
-      "step": 5847
-    },
-    {
-      "epoch": 0.9417448367486614,
-      "grad_norm": 0.0011730447877198458,
-      "learning_rate": 0.00019999956383503112,
-      "loss": 46.0,
-      "step": 5848
-    },
-    {
-      "epoch": 0.9419058738274487,
-      "grad_norm": 0.00117689638864249,
-      "learning_rate": 0.00019999956368559567,
-      "loss": 46.0,
-      "step": 5849
-    },
-    {
-      "epoch": 0.9420669109062362,
-      "grad_norm": 0.0011857007630169392,
-      "learning_rate": 0.00019999956353613466,
-      "loss": 46.0,
-      "step": 5850
-    },
-    {
-      "epoch": 0.9422279479850235,
-      "grad_norm": 0.002133960835635662,
-      "learning_rate": 0.00019999956338664804,
-      "loss": 46.0,
-      "step": 5851
-    },
-    {
-      "epoch": 0.942388985063811,
-      "grad_norm": 0.002602097112685442,
-      "learning_rate": 0.00019999956323713583,
-      "loss": 46.0,
-      "step": 5852
-    },
-    {
-      "epoch": 0.9425500221425983,
-      "grad_norm": 0.0007457935716956854,
-      "learning_rate": 0.00019999956308759803,
-      "loss": 46.0,
-      "step": 5853
-    },
-    {
-      "epoch": 0.9427110592213858,
-      "grad_norm": 0.0006414727540686727,
-      "learning_rate": 0.00019999956293803464,
-      "loss": 46.0,
-      "step": 5854
-    },
-    {
-      "epoch": 0.9428720963001731,
-      "grad_norm": 0.0017223377944901586,
-      "learning_rate": 0.00019999956278844561,
-      "loss": 46.0,
-      "step": 5855
-    },
-    {
-      "epoch": 0.9430331333789606,
-      "grad_norm": 0.0011650595115497708,
-      "learning_rate": 0.00019999956263883103,
-      "loss": 46.0,
-      "step": 5856
-    },
-    {
-      "epoch": 0.9431941704577479,
-      "grad_norm": 0.0007957100169733167,
-      "learning_rate": 0.00019999956248919088,
-      "loss": 46.0,
-      "step": 5857
-    },
-    {
-      "epoch": 0.9433552075365352,
-      "grad_norm": 0.0018951845122501254,
-      "learning_rate": 0.0001999995623395251,
-      "loss": 46.0,
-      "step": 5858
-    },
-    {
-      "epoch": 0.9435162446153227,
-      "grad_norm": 0.0031983281951397657,
-      "learning_rate": 0.00019999956218983372,
-      "loss": 46.0,
-      "step": 5859
-    },
-    {
-      "epoch": 0.94367728169411,
-      "grad_norm": 0.0008174518588930368,
-      "learning_rate": 0.00019999956204011675,
-      "loss": 46.0,
-      "step": 5860
-    },
-    {
-      "epoch": 0.9438383187728975,
-      "grad_norm": 0.0013758268905803561,
-      "learning_rate": 0.00019999956189037417,
-      "loss": 46.0,
-      "step": 5861
-    },
-    {
-      "epoch": 0.9439993558516848,
-      "grad_norm": 0.0009675536421127617,
-      "learning_rate": 0.00019999956174060604,
-      "loss": 46.0,
-      "step": 5862
-    },
-    {
-      "epoch": 0.9441603929304723,
-      "grad_norm": 0.0029827342368662357,
-      "learning_rate": 0.00019999956159081229,
-      "loss": 46.0,
-      "step": 5863
-    },
-    {
-      "epoch": 0.9443214300092596,
-      "grad_norm": 0.0030735281761735678,
-      "learning_rate": 0.00019999956144099295,
-      "loss": 46.0,
-      "step": 5864
-    },
-    {
-      "epoch": 0.9444824670880471,
-      "grad_norm": 0.0005248349043540657,
-      "learning_rate": 0.000199999561291148,
-      "loss": 46.0,
-      "step": 5865
-    },
-    {
-      "epoch": 0.9446435041668344,
-      "grad_norm": 0.001475742319598794,
-      "learning_rate": 0.00019999956114127745,
-      "loss": 46.0,
-      "step": 5866
-    },
-    {
-      "epoch": 0.9448045412456219,
-      "grad_norm": 0.0005304484511725605,
-      "learning_rate": 0.0001999995609913813,
-      "loss": 46.0,
-      "step": 5867
-    },
-    {
-      "epoch": 0.9449655783244092,
-      "grad_norm": 0.001460114843212068,
-      "learning_rate": 0.00019999956084145958,
-      "loss": 46.0,
-      "step": 5868
-    },
-    {
-      "epoch": 0.9451266154031965,
-      "grad_norm": 0.004002466797828674,
-      "learning_rate": 0.00019999956069151228,
-      "loss": 46.0,
-      "step": 5869
-    },
-    {
-      "epoch": 0.945287652481984,
-      "grad_norm": 0.0030936822295188904,
-      "learning_rate": 0.00019999956054153937,
-      "loss": 46.0,
-      "step": 5870
-    },
-    {
-      "epoch": 0.9454486895607713,
-      "grad_norm": 0.0013588797301054,
-      "learning_rate": 0.00019999956039154086,
-      "loss": 46.0,
-      "step": 5871
-    },
-    {
-      "epoch": 0.9456097266395588,
-      "grad_norm": 0.0015367834130302072,
-      "learning_rate": 0.00019999956024151675,
-      "loss": 46.0,
-      "step": 5872
-    },
-    {
-      "epoch": 0.9457707637183461,
-      "grad_norm": 0.0035890305880457163,
-      "learning_rate": 0.00019999956009146704,
-      "loss": 46.0,
-      "step": 5873
-    },
-    {
-      "epoch": 0.9459318007971336,
-      "grad_norm": 0.0009281002567149699,
-      "learning_rate": 0.00019999955994139175,
-      "loss": 46.0,
-      "step": 5874
-    },
-    {
-      "epoch": 0.9460928378759209,
-      "grad_norm": 0.0007293022354133427,
-      "learning_rate": 0.00019999955979129087,
-      "loss": 46.0,
-      "step": 5875
-    },
-    {
-      "epoch": 0.9462538749547084,
-      "grad_norm": 0.0013419538736343384,
-      "learning_rate": 0.0001999995596411644,
-      "loss": 46.0,
-      "step": 5876
-    },
-    {
-      "epoch": 0.9464149120334957,
-      "grad_norm": 0.007017700932919979,
-      "learning_rate": 0.0001999995594910123,
-      "loss": 46.0,
-      "step": 5877
-    },
-    {
-      "epoch": 0.9465759491122832,
-      "grad_norm": 0.0006878911517560482,
-      "learning_rate": 0.00019999955934083463,
-      "loss": 46.0,
-      "step": 5878
-    },
-    {
-      "epoch": 0.9467369861910705,
-      "grad_norm": 0.0011896201176568866,
-      "learning_rate": 0.00019999955919063135,
-      "loss": 46.0,
-      "step": 5879
-    },
-    {
-      "epoch": 0.9468980232698578,
-      "grad_norm": 0.0009057950810529292,
-      "learning_rate": 0.0001999995590404025,
-      "loss": 46.0,
-      "step": 5880
-    },
-    {
-      "epoch": 0.9470590603486453,
-      "grad_norm": 0.0023123540449887514,
-      "learning_rate": 0.00019999955889014802,
-      "loss": 46.0,
-      "step": 5881
-    },
-    {
-      "epoch": 0.9472200974274326,
-      "grad_norm": 0.0022262483835220337,
-      "learning_rate": 0.00019999955873986798,
-      "loss": 46.0,
-      "step": 5882
-    },
-    {
-      "epoch": 0.9473811345062201,
-      "grad_norm": 0.002448925981298089,
-      "learning_rate": 0.00019999955858956235,
-      "loss": 46.0,
-      "step": 5883
-    },
-    {
-      "epoch": 0.9475421715850074,
-      "grad_norm": 0.0006434541428461671,
-      "learning_rate": 0.00019999955843923108,
-      "loss": 46.0,
-      "step": 5884
-    },
-    {
-      "epoch": 0.9477032086637949,
-      "grad_norm": 0.0012345166178420186,
-      "learning_rate": 0.00019999955828887424,
-      "loss": 46.0,
-      "step": 5885
-    },
-    {
-      "epoch": 0.9478642457425822,
-      "grad_norm": 0.006125054322183132,
-      "learning_rate": 0.00019999955813849183,
-      "loss": 46.0,
-      "step": 5886
-    },
-    {
-      "epoch": 0.9480252828213697,
-      "grad_norm": 0.001413311343640089,
-      "learning_rate": 0.0001999995579880838,
-      "loss": 46.0,
-      "step": 5887
-    },
-    {
-      "epoch": 0.948186319900157,
-      "grad_norm": 0.0014180360594764352,
-      "learning_rate": 0.00019999955783765017,
-      "loss": 46.0,
-      "step": 5888
-    },
-    {
-      "epoch": 0.9483473569789445,
-      "grad_norm": 0.0007942916126921773,
-      "learning_rate": 0.00019999955768719096,
-      "loss": 46.0,
-      "step": 5889
-    },
-    {
-      "epoch": 0.9485083940577318,
-      "grad_norm": 0.003239106386899948,
-      "learning_rate": 0.00019999955753670614,
-      "loss": 46.0,
-      "step": 5890
-    },
-    {
-      "epoch": 0.9486694311365191,
-      "grad_norm": 0.0017419438809156418,
-      "learning_rate": 0.00019999955738619573,
-      "loss": 46.0,
-      "step": 5891
-    },
-    {
-      "epoch": 0.9488304682153066,
-      "grad_norm": 0.00036202429328113794,
-      "learning_rate": 0.00019999955723565976,
-      "loss": 46.0,
-      "step": 5892
-    },
-    {
-      "epoch": 0.9489915052940939,
-      "grad_norm": 0.0016309432685375214,
-      "learning_rate": 0.00019999955708509815,
-      "loss": 46.0,
-      "step": 5893
-    },
-    {
-      "epoch": 0.9491525423728814,
-      "grad_norm": 0.0016795072006061673,
-      "learning_rate": 0.00019999955693451095,
-      "loss": 46.0,
-      "step": 5894
-    },
-    {
-      "epoch": 0.9493135794516687,
-      "grad_norm": 0.0009731344762258232,
-      "learning_rate": 0.0001999995567838982,
-      "loss": 46.0,
-      "step": 5895
-    },
-    {
-      "epoch": 0.9494746165304562,
-      "grad_norm": 0.0026060317177325487,
-      "learning_rate": 0.0001999995566332598,
-      "loss": 46.0,
-      "step": 5896
-    },
-    {
-      "epoch": 0.9496356536092435,
-      "grad_norm": 0.0031692541670054197,
-      "learning_rate": 0.00019999955648259584,
-      "loss": 46.0,
-      "step": 5897
-    },
-    {
-      "epoch": 0.949796690688031,
-      "grad_norm": 0.0012209743726998568,
-      "learning_rate": 0.00019999955633190626,
-      "loss": 46.0,
-      "step": 5898
-    },
-    {
-      "epoch": 0.9499577277668183,
-      "grad_norm": 0.0004304127360228449,
-      "learning_rate": 0.00019999955618119108,
-      "loss": 46.0,
-      "step": 5899
-    },
-    {
-      "epoch": 0.9501187648456056,
-      "grad_norm": 0.0005454396014101803,
-      "learning_rate": 0.00019999955603045033,
-      "loss": 46.0,
-      "step": 5900
-    },
-    {
-      "epoch": 0.9502798019243931,
-      "grad_norm": 0.001358877052552998,
-      "learning_rate": 0.00019999955587968397,
-      "loss": 46.0,
-      "step": 5901
-    },
-    {
-      "epoch": 0.9504408390031804,
-      "grad_norm": 0.0015809446340426803,
-      "learning_rate": 0.00019999955572889202,
-      "loss": 46.0,
-      "step": 5902
-    },
-    {
-      "epoch": 0.9506018760819679,
-      "grad_norm": 0.0011048831511288881,
-      "learning_rate": 0.00019999955557807448,
-      "loss": 46.0,
-      "step": 5903
-    },
-    {
-      "epoch": 0.9507629131607552,
-      "grad_norm": 0.0015819414984434843,
-      "learning_rate": 0.00019999955542723136,
-      "loss": 46.0,
-      "step": 5904
-    },
-    {
-      "epoch": 0.9509239502395427,
-      "grad_norm": 0.0007353777764365077,
-      "learning_rate": 0.00019999955527636262,
-      "loss": 46.0,
-      "step": 5905
-    },
-    {
-      "epoch": 0.95108498731833,
-      "grad_norm": 0.00422245915979147,
-      "learning_rate": 0.00019999955512546832,
-      "loss": 46.0,
-      "step": 5906
-    },
-    {
-      "epoch": 0.9512460243971175,
-      "grad_norm": 0.0010974772740155458,
-      "learning_rate": 0.00019999955497454835,
-      "loss": 46.0,
-      "step": 5907
-    },
-    {
-      "epoch": 0.9514070614759048,
-      "grad_norm": 0.0035644201561808586,
-      "learning_rate": 0.00019999955482360285,
-      "loss": 46.0,
-      "step": 5908
-    },
-    {
-      "epoch": 0.9515680985546923,
-      "grad_norm": 0.0015335929347202182,
-      "learning_rate": 0.00019999955467263174,
-      "loss": 46.0,
-      "step": 5909
-    },
-    {
-      "epoch": 0.9517291356334796,
-      "grad_norm": 0.0008414181647822261,
-      "learning_rate": 0.00019999955452163504,
-      "loss": 46.0,
-      "step": 5910
-    },
-    {
-      "epoch": 0.951890172712267,
-      "grad_norm": 0.003015967318788171,
-      "learning_rate": 0.00019999955437061275,
-      "loss": 46.0,
-      "step": 5911
-    },
-    {
-      "epoch": 0.9520512097910544,
-      "grad_norm": 0.0017776868771761656,
-      "learning_rate": 0.00019999955421956485,
-      "loss": 46.0,
-      "step": 5912
-    },
-    {
-      "epoch": 0.9522122468698417,
-      "grad_norm": 0.0005025176797062159,
-      "learning_rate": 0.00019999955406849136,
-      "loss": 46.0,
-      "step": 5913
-    },
-    {
-      "epoch": 0.9523732839486292,
-      "grad_norm": 0.0004937931662425399,
-      "learning_rate": 0.00019999955391739228,
-      "loss": 46.0,
-      "step": 5914
-    },
-    {
-      "epoch": 0.9525343210274165,
-      "grad_norm": 0.00099847256205976,
-      "learning_rate": 0.0001999995537662676,
-      "loss": 46.0,
-      "step": 5915
-    },
-    {
-      "epoch": 0.952695358106204,
-      "grad_norm": 0.0013106995029374957,
-      "learning_rate": 0.0001999995536151173,
-      "loss": 46.0,
-      "step": 5916
-    },
-    {
-      "epoch": 0.9528563951849913,
-      "grad_norm": 0.002080469625070691,
-      "learning_rate": 0.00019999955346394145,
-      "loss": 46.0,
-      "step": 5917
-    },
-    {
-      "epoch": 0.9530174322637788,
-      "grad_norm": 0.0019621686078608036,
-      "learning_rate": 0.00019999955331274,
-      "loss": 46.0,
-      "step": 5918
-    },
-    {
-      "epoch": 0.9531784693425661,
-      "grad_norm": 0.0005493960343301296,
-      "learning_rate": 0.0001999995531615129,
-      "loss": 46.0,
-      "step": 5919
-    },
-    {
-      "epoch": 0.9533395064213536,
-      "grad_norm": 0.002767246449366212,
-      "learning_rate": 0.00019999955301026025,
-      "loss": 46.0,
-      "step": 5920
-    },
-    {
-      "epoch": 0.9535005435001409,
-      "grad_norm": 0.0032995629590004683,
-      "learning_rate": 0.00019999955285898203,
-      "loss": 46.0,
-      "step": 5921
-    },
-    {
-      "epoch": 0.9536615805789282,
-      "grad_norm": 0.0009536586003378034,
-      "learning_rate": 0.00019999955270767815,
-      "loss": 46.0,
-      "step": 5922
-    },
-    {
-      "epoch": 0.9538226176577157,
-      "grad_norm": 0.0013269956689327955,
-      "learning_rate": 0.00019999955255634873,
-      "loss": 46.0,
-      "step": 5923
-    },
-    {
-      "epoch": 0.953983654736503,
-      "grad_norm": 0.0009970528772100806,
-      "learning_rate": 0.0001999995524049937,
-      "loss": 46.0,
-      "step": 5924
-    },
-    {
-      "epoch": 0.9541446918152905,
-      "grad_norm": 0.0005479459068737924,
-      "learning_rate": 0.00019999955225361306,
-      "loss": 46.0,
-      "step": 5925
-    },
-    {
-      "epoch": 0.9543057288940778,
-      "grad_norm": 0.0019050290575250983,
-      "learning_rate": 0.00019999955210220686,
-      "loss": 46.0,
-      "step": 5926
-    },
-    {
-      "epoch": 0.9544667659728653,
-      "grad_norm": 0.00706638814881444,
-      "learning_rate": 0.00019999955195077504,
-      "loss": 46.0,
-      "step": 5927
-    },
-    {
-      "epoch": 0.9546278030516526,
-      "grad_norm": 0.0009241214138455689,
-      "learning_rate": 0.00019999955179931763,
-      "loss": 46.0,
-      "step": 5928
-    },
-    {
-      "epoch": 0.9547888401304401,
-      "grad_norm": 0.0005663994234055281,
-      "learning_rate": 0.0001999995516478346,
-      "loss": 46.0,
-      "step": 5929
-    },
-    {
-      "epoch": 0.9549498772092274,
-      "grad_norm": 0.0011111380299553275,
-      "learning_rate": 0.00019999955149632603,
-      "loss": 46.0,
-      "step": 5930
-    },
-    {
-      "epoch": 0.9551109142880149,
-      "grad_norm": 0.0005237525911070406,
-      "learning_rate": 0.00019999955134479184,
-      "loss": 46.0,
-      "step": 5931
-    },
-    {
-      "epoch": 0.9552719513668022,
-      "grad_norm": 0.001158655621111393,
-      "learning_rate": 0.00019999955119323203,
-      "loss": 46.0,
-      "step": 5932
-    },
-    {
-      "epoch": 0.9554329884455895,
-      "grad_norm": 0.0019604817498475313,
-      "learning_rate": 0.00019999955104164666,
-      "loss": 46.0,
-      "step": 5933
-    },
-    {
-      "epoch": 0.955594025524377,
-      "grad_norm": 0.0011811573058366776,
-      "learning_rate": 0.00019999955089003568,
-      "loss": 46.0,
-      "step": 5934
-    },
-    {
-      "epoch": 0.9557550626031643,
-      "grad_norm": 0.0034880167804658413,
-      "learning_rate": 0.0001999995507383991,
-      "loss": 46.0,
-      "step": 5935
-    },
-    {
-      "epoch": 0.9559160996819518,
-      "grad_norm": 0.0013889665715396404,
-      "learning_rate": 0.00019999955058673692,
-      "loss": 46.0,
-      "step": 5936
-    },
-    {
-      "epoch": 0.9560771367607391,
-      "grad_norm": 0.0012758998200297356,
-      "learning_rate": 0.00019999955043504918,
-      "loss": 46.0,
-      "step": 5937
-    },
-    {
-      "epoch": 0.9562381738395266,
-      "grad_norm": 0.0007953454623930156,
-      "learning_rate": 0.0001999995502833358,
-      "loss": 46.0,
-      "step": 5938
-    },
-    {
-      "epoch": 0.9563992109183139,
-      "grad_norm": 0.001341340015642345,
-      "learning_rate": 0.00019999955013159685,
-      "loss": 46.0,
-      "step": 5939
-    },
-    {
-      "epoch": 0.9565602479971014,
-      "grad_norm": 0.000832945283036679,
-      "learning_rate": 0.0001999995499798323,
-      "loss": 46.0,
-      "step": 5940
-    },
-    {
-      "epoch": 0.9567212850758887,
-      "grad_norm": 0.00111189647577703,
-      "learning_rate": 0.00019999954982804217,
-      "loss": 46.0,
-      "step": 5941
-    },
-    {
-      "epoch": 0.9568823221546762,
-      "grad_norm": 0.001660353154875338,
-      "learning_rate": 0.00019999954967622643,
-      "loss": 46.0,
-      "step": 5942
-    },
-    {
-      "epoch": 0.9570433592334635,
-      "grad_norm": 0.0005583629244938493,
-      "learning_rate": 0.00019999954952438508,
-      "loss": 46.0,
-      "step": 5943
-    },
-    {
-      "epoch": 0.9572043963122508,
-      "grad_norm": 0.0011081737466156483,
-      "learning_rate": 0.00019999954937251815,
-      "loss": 46.0,
-      "step": 5944
-    },
-    {
-      "epoch": 0.9573654333910383,
-      "grad_norm": 0.0011991882929578424,
-      "learning_rate": 0.00019999954922062565,
-      "loss": 46.0,
-      "step": 5945
-    },
-    {
-      "epoch": 0.9575264704698256,
-      "grad_norm": 0.005034707952290773,
-      "learning_rate": 0.00019999954906870751,
-      "loss": 46.0,
-      "step": 5946
-    },
-    {
-      "epoch": 0.9576875075486131,
-      "grad_norm": 0.00042172239045612514,
-      "learning_rate": 0.00019999954891676382,
-      "loss": 46.0,
-      "step": 5947
-    },
-    {
-      "epoch": 0.9578485446274004,
-      "grad_norm": 0.0019035013392567635,
-      "learning_rate": 0.0001999995487647945,
-      "loss": 46.0,
-      "step": 5948
-    },
-    {
-      "epoch": 0.9580095817061879,
-      "grad_norm": 0.0025095639284700155,
-      "learning_rate": 0.0001999995486127996,
-      "loss": 46.0,
-      "step": 5949
-    },
-    {
-      "epoch": 0.9581706187849752,
-      "grad_norm": 0.0015437612310051918,
-      "learning_rate": 0.00019999954846077912,
-      "loss": 46.0,
-      "step": 5950
-    },
-    {
-      "epoch": 0.9583316558637627,
-      "grad_norm": 0.002686878899112344,
-      "learning_rate": 0.00019999954830873302,
-      "loss": 46.0,
-      "step": 5951
-    },
-    {
-      "epoch": 0.95849269294255,
-      "grad_norm": 0.001571105094626546,
-      "learning_rate": 0.00019999954815666133,
-      "loss": 46.0,
-      "step": 5952
-    },
-    {
-      "epoch": 0.9586537300213374,
-      "grad_norm": 0.0015382863348349929,
-      "learning_rate": 0.00019999954800456403,
-      "loss": 46.0,
-      "step": 5953
-    },
-    {
-      "epoch": 0.9588147671001248,
-      "grad_norm": 0.0010662860004231334,
-      "learning_rate": 0.0001999995478524412,
-      "loss": 46.0,
-      "step": 5954
-    },
-    {
-      "epoch": 0.9589758041789122,
-      "grad_norm": 0.0037217061035335064,
-      "learning_rate": 0.00019999954770029272,
-      "loss": 46.0,
-      "step": 5955
-    },
-    {
-      "epoch": 0.9591368412576996,
-      "grad_norm": 0.0003007417544722557,
-      "learning_rate": 0.00019999954754811866,
-      "loss": 46.0,
-      "step": 5956
-    },
-    {
-      "epoch": 0.9592978783364869,
-      "grad_norm": 0.00027555067208595574,
-      "learning_rate": 0.000199999547395919,
-      "loss": 46.0,
-      "step": 5957
-    },
-    {
-      "epoch": 0.9594589154152744,
-      "grad_norm": 0.0029099839739501476,
-      "learning_rate": 0.00019999954724369374,
-      "loss": 46.0,
-      "step": 5958
-    },
-    {
-      "epoch": 0.9596199524940617,
-      "grad_norm": 0.0012330202152952552,
-      "learning_rate": 0.0001999995470914429,
-      "loss": 46.0,
-      "step": 5959
-    },
-    {
-      "epoch": 0.9597809895728492,
-      "grad_norm": 0.0026096138171851635,
-      "learning_rate": 0.00019999954693916645,
-      "loss": 46.0,
-      "step": 5960
-    },
-    {
-      "epoch": 0.9599420266516365,
-      "grad_norm": 0.0009886746993288398,
-      "learning_rate": 0.00019999954678686442,
-      "loss": 46.0,
-      "step": 5961
-    },
-    {
-      "epoch": 0.960103063730424,
-      "grad_norm": 0.0012607522075995803,
-      "learning_rate": 0.00019999954663453678,
-      "loss": 46.0,
-      "step": 5962
-    },
-    {
-      "epoch": 0.9602641008092113,
-      "grad_norm": 0.0012485279003158212,
-      "learning_rate": 0.00019999954648218356,
-      "loss": 46.0,
-      "step": 5963
-    },
-    {
-      "epoch": 0.9604251378879987,
-      "grad_norm": 0.0006192713626660407,
-      "learning_rate": 0.00019999954632980474,
-      "loss": 46.0,
-      "step": 5964
-    },
-    {
-      "epoch": 0.9605861749667861,
-      "grad_norm": 0.0023051968310028315,
-      "learning_rate": 0.00019999954617740028,
-      "loss": 46.0,
-      "step": 5965
-    },
-    {
-      "epoch": 0.9607472120455735,
-      "grad_norm": 0.0008143573650158942,
-      "learning_rate": 0.0001999995460249703,
-      "loss": 46.0,
-      "step": 5966
-    },
-    {
-      "epoch": 0.9609082491243609,
-      "grad_norm": 0.0017534703947603703,
-      "learning_rate": 0.0001999995458725147,
-      "loss": 46.0,
-      "step": 5967
-    },
-    {
-      "epoch": 0.9610692862031482,
-      "grad_norm": 0.01042917650192976,
-      "learning_rate": 0.00019999954572003347,
-      "loss": 46.0,
-      "step": 5968
-    },
-    {
-      "epoch": 0.9612303232819357,
-      "grad_norm": 0.0029092624317854643,
-      "learning_rate": 0.00019999954556752667,
-      "loss": 46.0,
-      "step": 5969
-    },
-    {
-      "epoch": 0.961391360360723,
-      "grad_norm": 0.0017549536423757672,
-      "learning_rate": 0.00019999954541499428,
-      "loss": 46.0,
-      "step": 5970
-    },
-    {
-      "epoch": 0.9615523974395105,
-      "grad_norm": 0.0007472428842447698,
-      "learning_rate": 0.00019999954526243632,
-      "loss": 46.0,
-      "step": 5971
-    },
-    {
-      "epoch": 0.9617134345182978,
-      "grad_norm": 0.0005359402857720852,
-      "learning_rate": 0.00019999954510985273,
-      "loss": 46.0,
-      "step": 5972
-    },
-    {
-      "epoch": 0.9618744715970853,
-      "grad_norm": 0.0010029306868091226,
-      "learning_rate": 0.00019999954495724355,
-      "loss": 46.0,
-      "step": 5973
-    },
-    {
-      "epoch": 0.9620355086758726,
-      "grad_norm": 0.0012263440294191241,
-      "learning_rate": 0.00019999954480460878,
-      "loss": 46.0,
-      "step": 5974
-    },
-    {
-      "epoch": 0.96219654575466,
-      "grad_norm": 0.0028613756876438856,
-      "learning_rate": 0.0001999995446519484,
-      "loss": 46.0,
-      "step": 5975
-    },
-    {
-      "epoch": 0.9623575828334474,
-      "grad_norm": 0.002050126902759075,
-      "learning_rate": 0.00019999954449926246,
-      "loss": 46.0,
-      "step": 5976
-    },
-    {
-      "epoch": 0.9625186199122348,
-      "grad_norm": 0.0005310737178660929,
-      "learning_rate": 0.0001999995443465509,
-      "loss": 46.0,
-      "step": 5977
-    },
-    {
-      "epoch": 0.9626796569910222,
-      "grad_norm": 0.0033825808204710484,
-      "learning_rate": 0.00019999954419381376,
-      "loss": 46.0,
-      "step": 5978
-    },
-    {
-      "epoch": 0.9628406940698095,
-      "grad_norm": 0.001322429277934134,
-      "learning_rate": 0.000199999544041051,
-      "loss": 46.0,
-      "step": 5979
-    },
-    {
-      "epoch": 0.963001731148597,
-      "grad_norm": 0.0008859616937115788,
-      "learning_rate": 0.00019999954388826268,
-      "loss": 46.0,
-      "step": 5980
-    },
-    {
-      "epoch": 0.9631627682273843,
-      "grad_norm": 0.0009631182183511555,
-      "learning_rate": 0.00019999954373544872,
-      "loss": 46.0,
-      "step": 5981
-    },
-    {
-      "epoch": 0.9633238053061718,
-      "grad_norm": 0.001079257926903665,
-      "learning_rate": 0.0001999995435826092,
-      "loss": 46.0,
-      "step": 5982
-    },
-    {
-      "epoch": 0.9634848423849591,
-      "grad_norm": 0.000829666038043797,
-      "learning_rate": 0.0001999995434297441,
-      "loss": 46.0,
-      "step": 5983
-    },
-    {
-      "epoch": 0.9636458794637466,
-      "grad_norm": 0.002197215799242258,
-      "learning_rate": 0.00019999954327685338,
-      "loss": 46.0,
-      "step": 5984
-    },
-    {
-      "epoch": 0.9638069165425339,
-      "grad_norm": 0.00848348904401064,
-      "learning_rate": 0.00019999954312393707,
-      "loss": 46.0,
-      "step": 5985
-    },
-    {
-      "epoch": 0.9639679536213213,
-      "grad_norm": 0.0003334719513077289,
-      "learning_rate": 0.00019999954297099514,
-      "loss": 46.0,
-      "step": 5986
-    },
-    {
-      "epoch": 0.9641289907001087,
-      "grad_norm": 0.0023209962528198957,
-      "learning_rate": 0.00019999954281802764,
-      "loss": 46.0,
-      "step": 5987
-    },
-    {
-      "epoch": 0.964290027778896,
-      "grad_norm": 0.0005076867528259754,
-      "learning_rate": 0.00019999954266503457,
-      "loss": 46.0,
-      "step": 5988
-    },
-    {
-      "epoch": 0.9644510648576835,
-      "grad_norm": 0.0031284557189792395,
-      "learning_rate": 0.00019999954251201585,
-      "loss": 46.0,
-      "step": 5989
-    },
-    {
-      "epoch": 0.9646121019364708,
-      "grad_norm": 0.0003518647572491318,
-      "learning_rate": 0.00019999954235897156,
-      "loss": 46.0,
-      "step": 5990
-    },
-    {
-      "epoch": 0.9647731390152583,
-      "grad_norm": 0.0013615258503705263,
-      "learning_rate": 0.0001999995422059017,
-      "loss": 46.0,
-      "step": 5991
-    },
-    {
-      "epoch": 0.9649341760940456,
-      "grad_norm": 0.0015308080473914742,
-      "learning_rate": 0.00019999954205280623,
-      "loss": 46.0,
-      "step": 5992
-    },
-    {
-      "epoch": 0.9650952131728331,
-      "grad_norm": 0.0005138955893926322,
-      "learning_rate": 0.00019999954189968514,
-      "loss": 46.0,
-      "step": 5993
-    },
-    {
-      "epoch": 0.9652562502516204,
-      "grad_norm": 0.0006996492738835514,
-      "learning_rate": 0.00019999954174653847,
-      "loss": 46.0,
-      "step": 5994
-    },
-    {
-      "epoch": 0.9654172873304079,
-      "grad_norm": 0.0010123642859980464,
-      "learning_rate": 0.00019999954159336623,
-      "loss": 46.0,
-      "step": 5995
-    },
-    {
-      "epoch": 0.9655783244091952,
-      "grad_norm": 0.0007726819603703916,
-      "learning_rate": 0.00019999954144016838,
-      "loss": 46.0,
-      "step": 5996
-    },
-    {
-      "epoch": 0.9657393614879826,
-      "grad_norm": 0.0010896530002355576,
-      "learning_rate": 0.00019999954128694492,
-      "loss": 46.0,
-      "step": 5997
-    },
-    {
-      "epoch": 0.96590039856677,
-      "grad_norm": 0.0035233329981565475,
-      "learning_rate": 0.00019999954113369587,
-      "loss": 46.0,
-      "step": 5998
-    },
-    {
-      "epoch": 0.9660614356455574,
-      "grad_norm": 0.0015133344568312168,
-      "learning_rate": 0.00019999954098042123,
-      "loss": 46.0,
-      "step": 5999
-    },
-    {
-      "epoch": 0.9662224727243448,
-      "grad_norm": 0.0007407511002384126,
-      "learning_rate": 0.000199999540827121,
-      "loss": 46.0,
-      "step": 6000
-    },
-    {
-      "epoch": 0.9663835098031321,
-      "grad_norm": 0.0012960054446011782,
-      "learning_rate": 0.00019999954067379517,
-      "loss": 46.0,
-      "step": 6001
-    },
-    {
-      "epoch": 0.9665445468819196,
-      "grad_norm": 0.0016822877805680037,
-      "learning_rate": 0.00019999954052044375,
-      "loss": 46.0,
-      "step": 6002
-    },
-    {
-      "epoch": 0.9667055839607069,
-      "grad_norm": 0.001962418667972088,
-      "learning_rate": 0.00019999954036706673,
-      "loss": 46.0,
-      "step": 6003
-    },
-    {
-      "epoch": 0.9668666210394944,
-      "grad_norm": 0.00287793786264956,
-      "learning_rate": 0.0001999995402136641,
-      "loss": 46.0,
-      "step": 6004
-    },
-    {
-      "epoch": 0.9670276581182817,
-      "grad_norm": 0.003555223112925887,
-      "learning_rate": 0.00019999954006023592,
-      "loss": 46.0,
-      "step": 6005
-    },
-    {
-      "epoch": 0.9671886951970691,
-      "grad_norm": 0.0007314881077036262,
-      "learning_rate": 0.00019999953990678212,
-      "loss": 46.0,
-      "step": 6006
-    },
-    {
-      "epoch": 0.9673497322758565,
-      "grad_norm": 0.0008118348196148872,
-      "learning_rate": 0.00019999953975330273,
-      "loss": 46.0,
-      "step": 6007
-    },
-    {
-      "epoch": 0.9675107693546439,
-      "grad_norm": 0.007767799776047468,
-      "learning_rate": 0.00019999953959979773,
-      "loss": 46.0,
-      "step": 6008
-    },
-    {
-      "epoch": 0.9676718064334313,
-      "grad_norm": 0.0008505704463459551,
-      "learning_rate": 0.00019999953944626714,
-      "loss": 46.0,
-      "step": 6009
-    },
-    {
-      "epoch": 0.9678328435122187,
-      "grad_norm": 0.0016857705777511,
-      "learning_rate": 0.00019999953929271096,
-      "loss": 46.0,
-      "step": 6010
-    },
-    {
-      "epoch": 0.9679938805910061,
-      "grad_norm": 0.0015252127777785063,
-      "learning_rate": 0.00019999953913912917,
-      "loss": 46.0,
-      "step": 6011
-    },
-    {
-      "epoch": 0.9681549176697934,
-      "grad_norm": 0.0004844553186558187,
-      "learning_rate": 0.00019999953898552182,
-      "loss": 46.0,
-      "step": 6012
-    },
-    {
-      "epoch": 0.9683159547485809,
-      "grad_norm": 0.0029004274401813745,
-      "learning_rate": 0.00019999953883188885,
-      "loss": 46.0,
-      "step": 6013
-    },
-    {
-      "epoch": 0.9684769918273682,
-      "grad_norm": 0.0016834705602377653,
-      "learning_rate": 0.0001999995386782303,
-      "loss": 46.0,
-      "step": 6014
-    },
-    {
-      "epoch": 0.9686380289061557,
-      "grad_norm": 0.0020192009396851063,
-      "learning_rate": 0.0001999995385245461,
-      "loss": 46.0,
-      "step": 6015
-    },
-    {
-      "epoch": 0.968799065984943,
-      "grad_norm": 0.0019935774616897106,
-      "learning_rate": 0.00019999953837083638,
-      "loss": 46.0,
-      "step": 6016
-    },
-    {
-      "epoch": 0.9689601030637304,
-      "grad_norm": 0.001451910356990993,
-      "learning_rate": 0.00019999953821710103,
-      "loss": 46.0,
-      "step": 6017
-    },
-    {
-      "epoch": 0.9691211401425178,
-      "grad_norm": 0.0007066584075801075,
-      "learning_rate": 0.0001999995380633401,
-      "loss": 46.0,
-      "step": 6018
-    },
-    {
-      "epoch": 0.9692821772213052,
-      "grad_norm": 0.0014588235644623637,
-      "learning_rate": 0.00019999953790955356,
-      "loss": 46.0,
-      "step": 6019
-    },
-    {
-      "epoch": 0.9694432143000926,
-      "grad_norm": 0.0026323446072638035,
-      "learning_rate": 0.00019999953775574146,
-      "loss": 46.0,
-      "step": 6020
-    },
-    {
-      "epoch": 0.96960425137888,
-      "grad_norm": 0.0007777338032610714,
-      "learning_rate": 0.00019999953760190372,
-      "loss": 46.0,
-      "step": 6021
-    },
-    {
-      "epoch": 0.9697652884576674,
-      "grad_norm": 0.0020895982161164284,
-      "learning_rate": 0.00019999953744804038,
-      "loss": 46.0,
-      "step": 6022
-    },
-    {
-      "epoch": 0.9699263255364547,
-      "grad_norm": 0.0007298138807527721,
-      "learning_rate": 0.0001999995372941515,
-      "loss": 46.0,
-      "step": 6023
-    },
-    {
-      "epoch": 0.9700873626152422,
-      "grad_norm": 0.001311998930759728,
-      "learning_rate": 0.00019999953714023696,
-      "loss": 46.0,
-      "step": 6024
-    },
-    {
-      "epoch": 0.9702483996940295,
-      "grad_norm": 0.0003248398134019226,
-      "learning_rate": 0.00019999953698629687,
-      "loss": 46.0,
-      "step": 6025
-    },
-    {
-      "epoch": 0.970409436772817,
-      "grad_norm": 0.004449520725756884,
-      "learning_rate": 0.00019999953683233116,
-      "loss": 46.0,
-      "step": 6026
-    },
-    {
-      "epoch": 0.9705704738516043,
-      "grad_norm": 0.002645839238539338,
-      "learning_rate": 0.0001999995366783399,
-      "loss": 46.0,
-      "step": 6027
-    },
-    {
-      "epoch": 0.9707315109303917,
-      "grad_norm": 0.002953458344563842,
-      "learning_rate": 0.00019999953652432298,
-      "loss": 46.0,
-      "step": 6028
-    },
-    {
-      "epoch": 0.9708925480091791,
-      "grad_norm": 0.00044393062125891447,
-      "learning_rate": 0.00019999953637028051,
-      "loss": 46.0,
-      "step": 6029
-    },
-    {
-      "epoch": 0.9710535850879665,
-      "grad_norm": 0.0023680441081523895,
-      "learning_rate": 0.00019999953621621243,
-      "loss": 46.0,
-      "step": 6030
-    },
-    {
-      "epoch": 0.9712146221667539,
-      "grad_norm": 0.0012556271394714713,
-      "learning_rate": 0.00019999953606211876,
-      "loss": 46.0,
-      "step": 6031
-    },
-    {
-      "epoch": 0.9713756592455413,
-      "grad_norm": 0.0029499754309654236,
-      "learning_rate": 0.0001999995359079995,
-      "loss": 46.0,
-      "step": 6032
-    },
-    {
-      "epoch": 0.9715366963243287,
-      "grad_norm": 0.003256219904869795,
-      "learning_rate": 0.00019999953575385463,
-      "loss": 46.0,
-      "step": 6033
-    },
-    {
-      "epoch": 0.971697733403116,
-      "grad_norm": 0.0016910432605072856,
-      "learning_rate": 0.00019999953559968417,
-      "loss": 46.0,
-      "step": 6034
-    },
-    {
-      "epoch": 0.9718587704819035,
-      "grad_norm": 0.002183531643822789,
-      "learning_rate": 0.00019999953544548813,
-      "loss": 46.0,
-      "step": 6035
-    },
-    {
-      "epoch": 0.9720198075606908,
-      "grad_norm": 0.0009558479650877416,
-      "learning_rate": 0.0001999995352912665,
-      "loss": 46.0,
-      "step": 6036
-    },
-    {
-      "epoch": 0.9721808446394783,
-      "grad_norm": 0.0009155339212156832,
-      "learning_rate": 0.00019999953513701925,
-      "loss": 46.0,
-      "step": 6037
-    },
-    {
-      "epoch": 0.9723418817182656,
-      "grad_norm": 0.001110175158828497,
-      "learning_rate": 0.0001999995349827464,
-      "loss": 46.0,
-      "step": 6038
-    },
-    {
-      "epoch": 0.972502918797053,
-      "grad_norm": 0.0020170703064650297,
-      "learning_rate": 0.000199999534828448,
-      "loss": 46.0,
-      "step": 6039
-    },
-    {
-      "epoch": 0.9726639558758404,
-      "grad_norm": 0.004659927915781736,
-      "learning_rate": 0.00019999953467412395,
-      "loss": 46.0,
-      "step": 6040
-    },
-    {
-      "epoch": 0.9728249929546278,
-      "grad_norm": 0.0008497110102325678,
-      "learning_rate": 0.00019999953451977436,
-      "loss": 46.0,
-      "step": 6041
-    },
-    {
-      "epoch": 0.9729860300334152,
-      "grad_norm": 0.000985593069344759,
-      "learning_rate": 0.00019999953436539915,
-      "loss": 46.0,
-      "step": 6042
-    },
-    {
-      "epoch": 0.9731470671122026,
-      "grad_norm": 0.0009190439595840871,
-      "learning_rate": 0.00019999953421099832,
-      "loss": 46.0,
-      "step": 6043
-    },
-    {
-      "epoch": 0.97330810419099,
-      "grad_norm": 0.004370598588138819,
-      "learning_rate": 0.00019999953405657194,
-      "loss": 46.0,
-      "step": 6044
-    },
-    {
-      "epoch": 0.9734691412697773,
-      "grad_norm": 0.00344197079539299,
-      "learning_rate": 0.00019999953390211994,
-      "loss": 46.0,
-      "step": 6045
-    },
-    {
-      "epoch": 0.9736301783485648,
-      "grad_norm": 0.0007090232684276998,
-      "learning_rate": 0.00019999953374764232,
-      "loss": 46.0,
-      "step": 6046
-    },
-    {
-      "epoch": 0.9737912154273521,
-      "grad_norm": 0.004273815080523491,
-      "learning_rate": 0.00019999953359313915,
-      "loss": 46.0,
-      "step": 6047
-    },
-    {
-      "epoch": 0.9739522525061396,
-      "grad_norm": 0.0013869086978957057,
-      "learning_rate": 0.00019999953343861036,
-      "loss": 46.0,
-      "step": 6048
-    },
-    {
-      "epoch": 0.9741132895849269,
-      "grad_norm": 0.002385001163929701,
-      "learning_rate": 0.000199999533284056,
-      "loss": 46.0,
-      "step": 6049
-    },
-    {
-      "epoch": 0.9742743266637143,
-      "grad_norm": 0.000777302891947329,
-      "learning_rate": 0.00019999953312947603,
-      "loss": 46.0,
-      "step": 6050
-    },
-    {
-      "epoch": 0.9744353637425017,
-      "grad_norm": 0.003631291911005974,
-      "learning_rate": 0.00019999953297487048,
-      "loss": 46.0,
-      "step": 6051
-    },
-    {
-      "epoch": 0.9745964008212891,
-      "grad_norm": 0.0011221128515899181,
-      "learning_rate": 0.00019999953282023929,
-      "loss": 46.0,
-      "step": 6052
-    },
-    {
-      "epoch": 0.9747574379000765,
-      "grad_norm": 0.002175306435674429,
-      "learning_rate": 0.00019999953266558253,
-      "loss": 46.0,
-      "step": 6053
-    },
-    {
-      "epoch": 0.9749184749788639,
-      "grad_norm": 0.002293642144650221,
-      "learning_rate": 0.00019999953251090017,
-      "loss": 46.0,
-      "step": 6054
-    },
-    {
-      "epoch": 0.9750795120576513,
-      "grad_norm": 0.0025275596417486668,
-      "learning_rate": 0.00019999953235619224,
-      "loss": 46.0,
-      "step": 6055
-    },
-    {
-      "epoch": 0.9752405491364387,
-      "grad_norm": 0.0005897557130083442,
-      "learning_rate": 0.0001999995322014587,
-      "loss": 46.0,
-      "step": 6056
-    },
-    {
-      "epoch": 0.9754015862152261,
-      "grad_norm": 0.0048878611996769905,
-      "learning_rate": 0.00019999953204669958,
-      "loss": 46.0,
-      "step": 6057
-    },
-    {
-      "epoch": 0.9755626232940134,
-      "grad_norm": 0.002488535363227129,
-      "learning_rate": 0.00019999953189191484,
-      "loss": 46.0,
-      "step": 6058
-    },
-    {
-      "epoch": 0.9757236603728008,
-      "grad_norm": 0.0006691949092783034,
-      "learning_rate": 0.00019999953173710454,
-      "loss": 46.0,
-      "step": 6059
-    },
-    {
-      "epoch": 0.9758846974515882,
-      "grad_norm": 0.003292854642495513,
-      "learning_rate": 0.0001999995315822686,
-      "loss": 46.0,
-      "step": 6060
-    },
-    {
-      "epoch": 0.9760457345303756,
-      "grad_norm": 0.0017195430118590593,
-      "learning_rate": 0.0001999995314274071,
-      "loss": 46.0,
-      "step": 6061
-    },
-    {
-      "epoch": 0.976206771609163,
-      "grad_norm": 0.0008291300036944449,
-      "learning_rate": 0.00019999953127251998,
-      "loss": 46.0,
-      "step": 6062
-    },
-    {
-      "epoch": 0.9763678086879504,
-      "grad_norm": 0.001436087884940207,
-      "learning_rate": 0.00019999953111760727,
-      "loss": 46.0,
-      "step": 6063
-    },
-    {
-      "epoch": 0.9765288457667378,
-      "grad_norm": 0.0002890973992180079,
-      "learning_rate": 0.00019999953096266898,
-      "loss": 46.0,
-      "step": 6064
-    },
-    {
-      "epoch": 0.9766898828455252,
-      "grad_norm": 0.0010119887301698327,
-      "learning_rate": 0.00019999953080770508,
-      "loss": 46.0,
-      "step": 6065
-    },
-    {
-      "epoch": 0.9768509199243126,
-      "grad_norm": 0.0005323869991116226,
-      "learning_rate": 0.00019999953065271558,
-      "loss": 46.0,
-      "step": 6066
-    },
-    {
-      "epoch": 0.9770119570031,
-      "grad_norm": 0.007950590923428535,
-      "learning_rate": 0.00019999953049770053,
-      "loss": 46.0,
-      "step": 6067
-    },
-    {
-      "epoch": 0.9771729940818874,
-      "grad_norm": 0.0006231916486285627,
-      "learning_rate": 0.00019999953034265984,
-      "loss": 46.0,
-      "step": 6068
-    },
-    {
-      "epoch": 0.9773340311606747,
-      "grad_norm": 0.002003576373681426,
-      "learning_rate": 0.00019999953018759356,
-      "loss": 46.0,
-      "step": 6069
-    },
-    {
-      "epoch": 0.9774950682394621,
-      "grad_norm": 0.0023399987258017063,
-      "learning_rate": 0.00019999953003250172,
-      "loss": 46.0,
-      "step": 6070
-    },
-    {
-      "epoch": 0.9776561053182495,
-      "grad_norm": 0.0010155565105378628,
-      "learning_rate": 0.00019999952987738423,
-      "loss": 46.0,
-      "step": 6071
-    },
-    {
-      "epoch": 0.9778171423970369,
-      "grad_norm": 0.0037375055253505707,
-      "learning_rate": 0.0001999995297222412,
-      "loss": 46.0,
-      "step": 6072
-    },
-    {
-      "epoch": 0.9779781794758243,
-      "grad_norm": 0.0005912245251238346,
-      "learning_rate": 0.00019999952956707253,
-      "loss": 46.0,
-      "step": 6073
-    },
-    {
-      "epoch": 0.9781392165546117,
-      "grad_norm": 0.0022507691755890846,
-      "learning_rate": 0.0001999995294118783,
-      "loss": 46.0,
-      "step": 6074
-    },
-    {
-      "epoch": 0.9783002536333991,
-      "grad_norm": 0.000521384528838098,
-      "learning_rate": 0.00019999952925665846,
-      "loss": 46.0,
-      "step": 6075
-    },
-    {
-      "epoch": 0.9784612907121865,
-      "grad_norm": 0.001833320944570005,
-      "learning_rate": 0.00019999952910141299,
-      "loss": 46.0,
-      "step": 6076
-    },
-    {
-      "epoch": 0.9786223277909739,
-      "grad_norm": 0.0007558513316325843,
-      "learning_rate": 0.00019999952894614198,
-      "loss": 46.0,
-      "step": 6077
-    },
-    {
-      "epoch": 0.9787833648697613,
-      "grad_norm": 0.0008772094151936471,
-      "learning_rate": 0.00019999952879084536,
-      "loss": 46.0,
-      "step": 6078
-    },
-    {
-      "epoch": 0.9789444019485487,
-      "grad_norm": 0.0007068760460242629,
-      "learning_rate": 0.00019999952863552313,
-      "loss": 46.0,
-      "step": 6079
-    },
-    {
-      "epoch": 0.979105439027336,
-      "grad_norm": 0.00278997584246099,
-      "learning_rate": 0.0001999995284801753,
-      "loss": 46.0,
-      "step": 6080
-    },
-    {
-      "epoch": 0.9792664761061234,
-      "grad_norm": 0.0016853873385116458,
-      "learning_rate": 0.0001999995283248019,
-      "loss": 46.0,
-      "step": 6081
-    },
-    {
-      "epoch": 0.9794275131849108,
-      "grad_norm": 0.00478401780128479,
-      "learning_rate": 0.00019999952816940293,
-      "loss": 46.0,
-      "step": 6082
-    },
-    {
-      "epoch": 0.9795885502636982,
-      "grad_norm": 0.003959829453378916,
-      "learning_rate": 0.00019999952801397832,
-      "loss": 46.0,
-      "step": 6083
-    },
-    {
-      "epoch": 0.9797495873424856,
-      "grad_norm": 0.0004952999297529459,
-      "learning_rate": 0.00019999952785852812,
-      "loss": 46.0,
-      "step": 6084
-    },
-    {
-      "epoch": 0.979910624421273,
-      "grad_norm": 0.0024953284300863743,
-      "learning_rate": 0.00019999952770305234,
-      "loss": 46.0,
-      "step": 6085
-    },
-    {
-      "epoch": 0.9800716615000604,
-      "grad_norm": 0.0019067783141508698,
-      "learning_rate": 0.00019999952754755097,
-      "loss": 46.0,
-      "step": 6086
-    },
-    {
-      "epoch": 0.9802326985788478,
-      "grad_norm": 0.002127291401848197,
-      "learning_rate": 0.00019999952739202395,
-      "loss": 46.0,
-      "step": 6087
-    },
-    {
-      "epoch": 0.9803937356576352,
-      "grad_norm": 0.0008091144845820963,
-      "learning_rate": 0.0001999995272364714,
-      "loss": 46.0,
-      "step": 6088
-    },
-    {
-      "epoch": 0.9805547727364226,
-      "grad_norm": 0.0011850473238155246,
-      "learning_rate": 0.00019999952708089325,
-      "loss": 46.0,
-      "step": 6089
-    },
-    {
-      "epoch": 0.98071580981521,
-      "grad_norm": 0.0023602095898240805,
-      "learning_rate": 0.0001999995269252895,
-      "loss": 46.0,
-      "step": 6090
-    },
-    {
-      "epoch": 0.9808768468939973,
-      "grad_norm": 0.0015418841503560543,
-      "learning_rate": 0.0001999995267696601,
-      "loss": 46.0,
-      "step": 6091
-    },
-    {
-      "epoch": 0.9810378839727847,
-      "grad_norm": 0.0027921677101403475,
-      "learning_rate": 0.00019999952661400516,
-      "loss": 46.0,
-      "step": 6092
-    },
-    {
-      "epoch": 0.9811989210515721,
-      "grad_norm": 0.007058043964207172,
-      "learning_rate": 0.00019999952645832463,
-      "loss": 46.0,
-      "step": 6093
-    },
-    {
-      "epoch": 0.9813599581303595,
-      "grad_norm": 0.002366888104006648,
-      "learning_rate": 0.00019999952630261848,
-      "loss": 46.0,
-      "step": 6094
-    },
-    {
-      "epoch": 0.9815209952091469,
-      "grad_norm": 0.0012653230223804712,
-      "learning_rate": 0.00019999952614688674,
-      "loss": 46.0,
-      "step": 6095
-    },
-    {
-      "epoch": 0.9816820322879343,
-      "grad_norm": 0.003651617793366313,
-      "learning_rate": 0.00019999952599112942,
-      "loss": 46.0,
-      "step": 6096
-    },
-    {
-      "epoch": 0.9818430693667217,
-      "grad_norm": 0.0010562005918473005,
-      "learning_rate": 0.0001999995258353465,
-      "loss": 46.0,
-      "step": 6097
-    },
-    {
-      "epoch": 0.9820041064455091,
-      "grad_norm": 0.001356892753392458,
-      "learning_rate": 0.00019999952567953798,
-      "loss": 46.0,
-      "step": 6098
-    },
-    {
-      "epoch": 0.9821651435242965,
-      "grad_norm": 0.0009286191780120134,
-      "learning_rate": 0.00019999952552370387,
-      "loss": 46.0,
-      "step": 6099
-    },
-    {
-      "epoch": 0.9823261806030839,
-      "grad_norm": 0.002225863514468074,
-      "learning_rate": 0.00019999952536784417,
-      "loss": 46.0,
-      "step": 6100
-    },
-    {
-      "epoch": 0.9824872176818713,
-      "grad_norm": 0.0033421157859265804,
-      "learning_rate": 0.00019999952521195885,
-      "loss": 46.0,
-      "step": 6101
-    },
-    {
-      "epoch": 0.9826482547606586,
-      "grad_norm": 0.0005790418363176286,
-      "learning_rate": 0.00019999952505604795,
-      "loss": 46.0,
-      "step": 6102
-    },
-    {
-      "epoch": 0.982809291839446,
-      "grad_norm": 0.0025347555056214333,
-      "learning_rate": 0.00019999952490011146,
-      "loss": 46.0,
-      "step": 6103
-    },
-    {
-      "epoch": 0.9829703289182334,
-      "grad_norm": 0.0011037011863663793,
-      "learning_rate": 0.00019999952474414939,
-      "loss": 46.0,
-      "step": 6104
-    },
-    {
-      "epoch": 0.9831313659970208,
-      "grad_norm": 0.001047627185471356,
-      "learning_rate": 0.0001999995245881617,
-      "loss": 46.0,
-      "step": 6105
-    },
-    {
-      "epoch": 0.9832924030758082,
-      "grad_norm": 0.0007858876488171518,
-      "learning_rate": 0.00019999952443214842,
-      "loss": 46.0,
-      "step": 6106
-    },
-    {
-      "epoch": 0.9834534401545956,
-      "grad_norm": 0.0009689973667263985,
-      "learning_rate": 0.00019999952427610955,
-      "loss": 46.0,
-      "step": 6107
-    },
-    {
-      "epoch": 0.983614477233383,
-      "grad_norm": 0.001379107590764761,
-      "learning_rate": 0.00019999952412004507,
-      "loss": 46.0,
-      "step": 6108
-    },
-    {
-      "epoch": 0.9837755143121704,
-      "grad_norm": 0.002949258778244257,
-      "learning_rate": 0.00019999952396395504,
-      "loss": 46.0,
-      "step": 6109
-    },
-    {
-      "epoch": 0.9839365513909578,
-      "grad_norm": 0.00038959752419032156,
-      "learning_rate": 0.00019999952380783938,
-      "loss": 46.0,
-      "step": 6110
-    },
-    {
-      "epoch": 0.9840975884697452,
-      "grad_norm": 0.005300094839185476,
-      "learning_rate": 0.00019999952365169811,
-      "loss": 46.0,
-      "step": 6111
-    },
-    {
-      "epoch": 0.9842586255485325,
-      "grad_norm": 0.012284980155527592,
-      "learning_rate": 0.0001999995234955313,
-      "loss": 46.0,
-      "step": 6112
-    },
-    {
-      "epoch": 0.98441966262732,
-      "grad_norm": 0.0005090790218673646,
-      "learning_rate": 0.00019999952333933882,
-      "loss": 46.0,
-      "step": 6113
-    },
-    {
-      "epoch": 0.9845806997061073,
-      "grad_norm": 0.0006898490828461945,
-      "learning_rate": 0.0001999995231831208,
-      "loss": 46.0,
-      "step": 6114
-    },
-    {
-      "epoch": 0.9847417367848947,
-      "grad_norm": 0.0010778669966384768,
-      "learning_rate": 0.00019999952302687717,
-      "loss": 46.0,
-      "step": 6115
-    },
-    {
-      "epoch": 0.9849027738636821,
-      "grad_norm": 0.002038671402260661,
-      "learning_rate": 0.00019999952287060792,
-      "loss": 46.0,
-      "step": 6116
-    },
-    {
-      "epoch": 0.9850638109424695,
-      "grad_norm": 0.0024660706985741854,
-      "learning_rate": 0.00019999952271431313,
-      "loss": 46.0,
-      "step": 6117
-    },
-    {
-      "epoch": 0.9852248480212569,
-      "grad_norm": 0.0007153709884732962,
-      "learning_rate": 0.0001999995225579927,
-      "loss": 46.0,
-      "step": 6118
-    },
-    {
-      "epoch": 0.9853858851000443,
-      "grad_norm": 0.001145454472862184,
-      "learning_rate": 0.0001999995224016467,
-      "loss": 46.0,
-      "step": 6119
-    },
-    {
-      "epoch": 0.9855469221788317,
-      "grad_norm": 0.0011187976924702525,
-      "learning_rate": 0.0001999995222452751,
-      "loss": 46.0,
-      "step": 6120
-    },
-    {
-      "epoch": 0.9857079592576191,
-      "grad_norm": 0.0016687295865267515,
-      "learning_rate": 0.0001999995220888779,
-      "loss": 46.0,
-      "step": 6121
-    },
-    {
-      "epoch": 0.9858689963364065,
-      "grad_norm": 0.0014770758571103215,
-      "learning_rate": 0.0001999995219324551,
-      "loss": 46.0,
-      "step": 6122
-    },
-    {
-      "epoch": 0.9860300334151938,
-      "grad_norm": 0.002323250286281109,
-      "learning_rate": 0.00019999952177600673,
-      "loss": 46.0,
-      "step": 6123
-    },
-    {
-      "epoch": 0.9861910704939812,
-      "grad_norm": 0.00024691029102541506,
-      "learning_rate": 0.00019999952161953272,
-      "loss": 46.0,
-      "step": 6124
-    },
-    {
-      "epoch": 0.9863521075727686,
-      "grad_norm": 0.0010923704830929637,
-      "learning_rate": 0.00019999952146303315,
-      "loss": 46.0,
-      "step": 6125
-    },
-    {
-      "epoch": 0.986513144651556,
-      "grad_norm": 0.0007815223070792854,
-      "learning_rate": 0.000199999521306508,
-      "loss": 46.0,
-      "step": 6126
-    },
-    {
-      "epoch": 0.9866741817303434,
-      "grad_norm": 0.0011243766639381647,
-      "learning_rate": 0.00019999952114995722,
-      "loss": 46.0,
-      "step": 6127
-    },
-    {
-      "epoch": 0.9868352188091308,
-      "grad_norm": 0.0031221131794154644,
-      "learning_rate": 0.00019999952099338084,
-      "loss": 46.0,
-      "step": 6128
-    },
-    {
-      "epoch": 0.9869962558879182,
-      "grad_norm": 0.0011449974263086915,
-      "learning_rate": 0.00019999952083677892,
-      "loss": 46.0,
-      "step": 6129
-    },
-    {
-      "epoch": 0.9871572929667056,
-      "grad_norm": 0.003295786678791046,
-      "learning_rate": 0.00019999952068015136,
-      "loss": 46.0,
-      "step": 6130
-    },
-    {
-      "epoch": 0.987318330045493,
-      "grad_norm": 0.0012135219294577837,
-      "learning_rate": 0.00019999952052349819,
-      "loss": 46.0,
-      "step": 6131
-    },
-    {
-      "epoch": 0.9874793671242804,
-      "grad_norm": 0.0015659809578210115,
-      "learning_rate": 0.00019999952036681945,
-      "loss": 46.0,
-      "step": 6132
-    },
-    {
-      "epoch": 0.9876404042030678,
-      "grad_norm": 0.0018343861447647214,
-      "learning_rate": 0.00019999952021011513,
-      "loss": 46.0,
-      "step": 6133
-    },
-    {
-      "epoch": 0.9878014412818551,
-      "grad_norm": 0.0014180458383634686,
-      "learning_rate": 0.0001999995200533852,
-      "loss": 46.0,
-      "step": 6134
-    },
-    {
-      "epoch": 0.9879624783606425,
-      "grad_norm": 0.0023068648297339678,
-      "learning_rate": 0.00019999951989662967,
-      "loss": 46.0,
-      "step": 6135
-    },
-    {
-      "epoch": 0.9881235154394299,
-      "grad_norm": 0.0016210369067266583,
-      "learning_rate": 0.00019999951973984854,
-      "loss": 46.0,
-      "step": 6136
-    },
-    {
-      "epoch": 0.9882845525182173,
-      "grad_norm": 0.002114700386300683,
-      "learning_rate": 0.00019999951958304184,
-      "loss": 46.0,
-      "step": 6137
-    },
-    {
-      "epoch": 0.9884455895970047,
-      "grad_norm": 0.0036070598289370537,
-      "learning_rate": 0.0001999995194262095,
-      "loss": 46.0,
-      "step": 6138
-    },
-    {
-      "epoch": 0.9886066266757921,
-      "grad_norm": 0.0006896013510413468,
-      "learning_rate": 0.0001999995192693516,
-      "loss": 46.0,
-      "step": 6139
-    },
-    {
-      "epoch": 0.9887676637545795,
-      "grad_norm": 0.000816740095615387,
-      "learning_rate": 0.00019999951911246812,
-      "loss": 46.0,
-      "step": 6140
-    },
-    {
-      "epoch": 0.9889287008333669,
-      "grad_norm": 0.0014266902580857277,
-      "learning_rate": 0.00019999951895555902,
-      "loss": 46.0,
-      "step": 6141
-    },
-    {
-      "epoch": 0.9890897379121543,
-      "grad_norm": 0.0009997619781643152,
-      "learning_rate": 0.00019999951879862433,
-      "loss": 46.0,
-      "step": 6142
-    },
-    {
-      "epoch": 0.9892507749909417,
-      "grad_norm": 0.0007714173989370465,
-      "learning_rate": 0.00019999951864166406,
-      "loss": 46.0,
-      "step": 6143
-    },
-    {
-      "epoch": 0.9894118120697291,
-      "grad_norm": 0.004053396638482809,
-      "learning_rate": 0.00019999951848467817,
-      "loss": 46.0,
-      "step": 6144
-    },
-    {
-      "epoch": 0.9895728491485164,
-      "grad_norm": 0.0006980125326663256,
-      "learning_rate": 0.0001999995183276667,
-      "loss": 46.0,
-      "step": 6145
-    },
-    {
-      "epoch": 0.9897338862273038,
-      "grad_norm": 0.0022529347334057093,
-      "learning_rate": 0.00019999951817062966,
-      "loss": 46.0,
-      "step": 6146
-    },
-    {
-      "epoch": 0.9898949233060912,
-      "grad_norm": 0.00037822689046151936,
-      "learning_rate": 0.00019999951801356699,
-      "loss": 46.0,
-      "step": 6147
-    },
-    {
-      "epoch": 0.9900559603848786,
-      "grad_norm": 0.002851964207366109,
-      "learning_rate": 0.00019999951785647872,
-      "loss": 46.0,
-      "step": 6148
-    },
-    {
-      "epoch": 0.990216997463666,
-      "grad_norm": 0.003097419859841466,
-      "learning_rate": 0.0001999995176993649,
-      "loss": 46.0,
-      "step": 6149
-    },
-    {
-      "epoch": 0.9903780345424534,
-      "grad_norm": 0.001526164822280407,
-      "learning_rate": 0.00019999951754222543,
-      "loss": 46.0,
-      "step": 6150
-    },
-    {
-      "epoch": 0.9905390716212408,
-      "grad_norm": 0.0011916793882846832,
-      "learning_rate": 0.0001999995173850604,
-      "loss": 46.0,
-      "step": 6151
-    },
-    {
-      "epoch": 0.9907001087000282,
-      "grad_norm": 0.0030071926303207874,
-      "learning_rate": 0.00019999951722786977,
-      "loss": 46.0,
-      "step": 6152
-    },
-    {
-      "epoch": 0.9908611457788156,
-      "grad_norm": 0.003600070718675852,
-      "learning_rate": 0.00019999951707065354,
-      "loss": 46.0,
-      "step": 6153
-    },
-    {
-      "epoch": 0.991022182857603,
-      "grad_norm": 0.001538835815154016,
-      "learning_rate": 0.0001999995169134117,
-      "loss": 46.0,
-      "step": 6154
-    },
-    {
-      "epoch": 0.9911832199363904,
-      "grad_norm": 0.001332353800535202,
-      "learning_rate": 0.00019999951675614427,
-      "loss": 46.0,
-      "step": 6155
-    },
-    {
-      "epoch": 0.9913442570151777,
-      "grad_norm": 0.001979838591068983,
-      "learning_rate": 0.00019999951659885126,
-      "loss": 46.0,
-      "step": 6156
-    },
-    {
-      "epoch": 0.9915052940939652,
-      "grad_norm": 0.0010774682741612196,
-      "learning_rate": 0.00019999951644153265,
-      "loss": 46.0,
-      "step": 6157
-    },
-    {
-      "epoch": 0.9916663311727525,
-      "grad_norm": 0.0015265361871570349,
-      "learning_rate": 0.00019999951628418846,
-      "loss": 46.0,
-      "step": 6158
-    },
-    {
-      "epoch": 0.9918273682515399,
-      "grad_norm": 0.0020841562654823065,
-      "learning_rate": 0.00019999951612681866,
-      "loss": 46.0,
-      "step": 6159
-    },
-    {
-      "epoch": 0.9919884053303273,
-      "grad_norm": 0.0016260124975815415,
-      "learning_rate": 0.00019999951596942327,
-      "loss": 46.0,
-      "step": 6160
-    },
-    {
-      "epoch": 0.9921494424091147,
-      "grad_norm": 0.00033598599839024246,
-      "learning_rate": 0.00019999951581200226,
-      "loss": 46.0,
-      "step": 6161
-    },
-    {
-      "epoch": 0.9923104794879021,
-      "grad_norm": 0.001430588774383068,
-      "learning_rate": 0.00019999951565455567,
-      "loss": 46.0,
-      "step": 6162
-    },
-    {
-      "epoch": 0.9924715165666895,
-      "grad_norm": 0.0015464945463463664,
-      "learning_rate": 0.00019999951549708352,
-      "loss": 46.0,
-      "step": 6163
-    },
-    {
-      "epoch": 0.9926325536454769,
-      "grad_norm": 0.0009622170473448932,
-      "learning_rate": 0.00019999951533958575,
-      "loss": 46.0,
-      "step": 6164
-    },
-    {
-      "epoch": 0.9927935907242642,
-      "grad_norm": 0.0022616719361394644,
-      "learning_rate": 0.0001999995151820624,
-      "loss": 46.0,
-      "step": 6165
-    },
-    {
-      "epoch": 0.9929546278030517,
-      "grad_norm": 0.002432753797620535,
-      "learning_rate": 0.00019999951502451343,
-      "loss": 46.0,
-      "step": 6166
-    },
-    {
-      "epoch": 0.993115664881839,
-      "grad_norm": 0.002840930363163352,
-      "learning_rate": 0.00019999951486693887,
-      "loss": 46.0,
-      "step": 6167
-    },
-    {
-      "epoch": 0.9932767019606265,
-      "grad_norm": 0.005911883432418108,
-      "learning_rate": 0.0001999995147093387,
-      "loss": 46.0,
-      "step": 6168
-    },
-    {
-      "epoch": 0.9934377390394138,
-      "grad_norm": 0.0017866501584649086,
-      "learning_rate": 0.00019999951455171297,
-      "loss": 46.0,
-      "step": 6169
-    },
-    {
-      "epoch": 0.9935987761182012,
-      "grad_norm": 0.0006273352773860097,
-      "learning_rate": 0.00019999951439406163,
-      "loss": 46.0,
-      "step": 6170
-    },
-    {
-      "epoch": 0.9937598131969886,
-      "grad_norm": 0.0024944450706243515,
-      "learning_rate": 0.00019999951423638467,
-      "loss": 46.0,
-      "step": 6171
-    },
-    {
-      "epoch": 0.993920850275776,
-      "grad_norm": 0.0007860513869673014,
-      "learning_rate": 0.00019999951407868215,
-      "loss": 46.0,
-      "step": 6172
-    },
-    {
-      "epoch": 0.9940818873545634,
-      "grad_norm": 0.00033432774944230914,
-      "learning_rate": 0.00019999951392095402,
-      "loss": 46.0,
-      "step": 6173
-    },
-    {
-      "epoch": 0.9942429244333508,
-      "grad_norm": 0.00497489282861352,
-      "learning_rate": 0.0001999995137632003,
-      "loss": 46.0,
-      "step": 6174
-    },
-    {
-      "epoch": 0.9944039615121382,
-      "grad_norm": 0.003430631710216403,
-      "learning_rate": 0.000199999513605421,
-      "loss": 46.0,
-      "step": 6175
-    },
-    {
-      "epoch": 0.9945649985909255,
-      "grad_norm": 0.00470144534483552,
-      "learning_rate": 0.00019999951344761607,
-      "loss": 46.0,
-      "step": 6176
-    },
-    {
-      "epoch": 0.994726035669713,
-      "grad_norm": 0.0005960713024251163,
-      "learning_rate": 0.00019999951328978556,
-      "loss": 46.0,
-      "step": 6177
-    },
-    {
-      "epoch": 0.9948870727485003,
-      "grad_norm": 0.00037680185050703585,
-      "learning_rate": 0.0001999995131319295,
-      "loss": 46.0,
-      "step": 6178
-    },
-    {
-      "epoch": 0.9950481098272878,
-      "grad_norm": 0.0030698818154633045,
-      "learning_rate": 0.00019999951297404778,
-      "loss": 46.0,
-      "step": 6179
-    },
-    {
-      "epoch": 0.9952091469060751,
-      "grad_norm": 0.0023581250570714474,
-      "learning_rate": 0.00019999951281614048,
-      "loss": 46.0,
-      "step": 6180
-    },
-    {
-      "epoch": 0.9953701839848625,
-      "grad_norm": 0.0025764857418835163,
-      "learning_rate": 0.00019999951265820763,
-      "loss": 46.0,
-      "step": 6181
-    },
-    {
-      "epoch": 0.9955312210636499,
-      "grad_norm": 0.003658497938886285,
-      "learning_rate": 0.00019999951250024913,
-      "loss": 46.0,
-      "step": 6182
-    },
-    {
-      "epoch": 0.9956922581424373,
-      "grad_norm": 0.005764478351920843,
-      "learning_rate": 0.00019999951234226507,
-      "loss": 46.0,
-      "step": 6183
-    },
-    {
-      "epoch": 0.9958532952212247,
-      "grad_norm": 0.000824173737782985,
-      "learning_rate": 0.00019999951218425542,
-      "loss": 46.0,
-      "step": 6184
-    },
-    {
-      "epoch": 0.9960143323000121,
-      "grad_norm": 0.0009022654267027974,
-      "learning_rate": 0.00019999951202622014,
-      "loss": 46.0,
-      "step": 6185
-    },
-    {
-      "epoch": 0.9961753693787995,
-      "grad_norm": 0.0014750818954780698,
-      "learning_rate": 0.0001999995118681593,
-      "loss": 46.0,
-      "step": 6186
-    },
-    {
-      "epoch": 0.9963364064575868,
-      "grad_norm": 0.0011005938285961747,
-      "learning_rate": 0.00019999951171007283,
-      "loss": 46.0,
-      "step": 6187
-    },
-    {
-      "epoch": 0.9964974435363743,
-      "grad_norm": 0.0008392144227400422,
-      "learning_rate": 0.0001999995115519608,
-      "loss": 46.0,
-      "step": 6188
-    },
-    {
-      "epoch": 0.9966584806151616,
-      "grad_norm": 0.004620526451617479,
-      "learning_rate": 0.00019999951139382317,
-      "loss": 46.0,
-      "step": 6189
-    },
-    {
-      "epoch": 0.996819517693949,
-      "grad_norm": 0.0014713795389980078,
-      "learning_rate": 0.0001999995112356599,
-      "loss": 46.0,
-      "step": 6190
-    },
-    {
-      "epoch": 0.9969805547727364,
-      "grad_norm": 0.006917021702975035,
-      "learning_rate": 0.00019999951107747108,
-      "loss": 46.0,
-      "step": 6191
-    },
-    {
-      "epoch": 0.9971415918515238,
-      "grad_norm": 0.0013163541443645954,
-      "learning_rate": 0.00019999951091925663,
-      "loss": 46.0,
-      "step": 6192
-    },
-    {
-      "epoch": 0.9973026289303112,
-      "grad_norm": 0.001013058121316135,
-      "learning_rate": 0.00019999951076101665,
-      "loss": 46.0,
-      "step": 6193
-    },
-    {
-      "epoch": 0.9974636660090986,
-      "grad_norm": 0.001080639660358429,
-      "learning_rate": 0.00019999951060275102,
-      "loss": 46.0,
-      "step": 6194
-    },
-    {
-      "epoch": 0.997624703087886,
-      "grad_norm": 0.002676317933946848,
-      "learning_rate": 0.0001999995104444598,
-      "loss": 46.0,
-      "step": 6195
-    },
-    {
-      "epoch": 0.9977857401666734,
-      "grad_norm": 0.003335464047268033,
-      "learning_rate": 0.00019999951028614298,
-      "loss": 46.0,
-      "step": 6196
-    },
-    {
-      "epoch": 0.9979467772454608,
-      "grad_norm": 0.0012337196385487914,
-      "learning_rate": 0.0001999995101278006,
-      "loss": 46.0,
-      "step": 6197
-    },
-    {
-      "epoch": 0.9981078143242481,
-      "grad_norm": 0.0004911215510219336,
-      "learning_rate": 0.0001999995099694326,
-      "loss": 46.0,
-      "step": 6198
-    },
-    {
-      "epoch": 0.9982688514030356,
-      "grad_norm": 0.0061832452192902565,
-      "learning_rate": 0.00019999950981103903,
-      "loss": 46.0,
-      "step": 6199
-    },
-    {
-      "epoch": 0.9984298884818229,
-      "grad_norm": 0.000253285194048658,
-      "learning_rate": 0.00019999950965261985,
-      "loss": 46.0,
-      "step": 6200
-    },
-    {
-      "epoch": 0.9985909255606104,
-      "grad_norm": 0.0027165801730006933,
-      "learning_rate": 0.0001999995094941751,
-      "loss": 46.0,
-      "step": 6201
-    },
-    {
-      "epoch": 0.9987519626393977,
-      "grad_norm": 0.006130980793386698,
-      "learning_rate": 0.00019999950933570468,
-      "loss": 46.0,
-      "step": 6202
-    },
-    {
-      "epoch": 0.9989129997181851,
-      "grad_norm": 0.0023346436209976673,
-      "learning_rate": 0.00019999950917720872,
-      "loss": 46.0,
-      "step": 6203
-    },
-    {
-      "epoch": 0.9990740367969725,
-      "grad_norm": 0.001313018728978932,
-      "learning_rate": 0.00019999950901868717,
-      "loss": 46.0,
-      "step": 6204
-    },
-    {
-      "epoch": 0.9992350738757599,
-      "grad_norm": 0.0004894621088169515,
-      "learning_rate": 0.00019999950886014,
-      "loss": 46.0,
-      "step": 6205
-    },
-    {
-      "epoch": 0.9993961109545473,
-      "grad_norm": 0.0008191371452994645,
-      "learning_rate": 0.00019999950870156725,
-      "loss": 46.0,
-      "step": 6206
-    },
-    {
-      "epoch": 0.9995571480333347,
-      "grad_norm": 0.0005024021957069635,
-      "learning_rate": 0.0001999995085429689,
-      "loss": 46.0,
-      "step": 6207
-    },
-    {
-      "epoch": 0.9997181851121221,
-      "grad_norm": 0.0012005992466583848,
-      "learning_rate": 0.00019999950838434498,
-      "loss": 46.0,
-      "step": 6208
-    },
-    {
-      "epoch": 0.9998792221909094,
-      "grad_norm": 0.0037025536876171827,
-      "learning_rate": 0.00019999950822569544,
-      "loss": 46.0,
-      "step": 6209
-    },
-    {
-      "epoch": 0.9998792221909094,
-      "eval_loss": 11.5,
-      "eval_runtime": 14.5868,
-      "eval_samples_per_second": 179.272,
-      "eval_steps_per_second": 89.67,
-      "step": 6209
-    },
-    {
-      "epoch": 1.0000805185393937,
-      "grad_norm": 0.003451248398050666,
-      "learning_rate": 0.0001999995080670203,
-      "loss": 46.0,
-      "step": 6210
-    },
-    {
-      "epoch": 1.0002415556181812,
-      "grad_norm": 0.0024098986759781837,
-      "learning_rate": 0.0001999995079083196,
-      "loss": 46.0,
-      "step": 6211
-    },
-    {
-      "epoch": 1.0004025926969684,
-      "grad_norm": 0.0011229664087295532,
-      "learning_rate": 0.00019999950774959326,
-      "loss": 46.0,
-      "step": 6212
-    },
-    {
-      "epoch": 1.0005636297757559,
-      "grad_norm": 0.0014348207041621208,
-      "learning_rate": 0.00019999950759084134,
-      "loss": 46.0,
-      "step": 6213
-    },
-    {
-      "epoch": 1.0007246668545433,
-      "grad_norm": 0.003249640576541424,
-      "learning_rate": 0.00019999950743206384,
-      "loss": 46.0,
-      "step": 6214
-    },
-    {
-      "epoch": 1.0008857039333305,
-      "grad_norm": 0.0010153213515877724,
-      "learning_rate": 0.00019999950727326074,
-      "loss": 46.0,
-      "step": 6215
-    },
-    {
-      "epoch": 1.001046741012118,
-      "grad_norm": 0.0006885263719595969,
-      "learning_rate": 0.00019999950711443204,
-      "loss": 46.0,
-      "step": 6216
-    },
-    {
-      "epoch": 1.0012077780909054,
-      "grad_norm": 0.0004964792169630527,
-      "learning_rate": 0.00019999950695557774,
-      "loss": 46.0,
-      "step": 6217
-    },
-    {
-      "epoch": 1.001368815169693,
-      "grad_norm": 0.0010101882508024573,
-      "learning_rate": 0.00019999950679669786,
-      "loss": 46.0,
-      "step": 6218
-    },
-    {
-      "epoch": 1.0015298522484801,
-      "grad_norm": 0.0008583376184105873,
-      "learning_rate": 0.00019999950663779237,
-      "loss": 46.0,
-      "step": 6219
-    },
-    {
-      "epoch": 1.0016908893272676,
-      "grad_norm": 0.002517750021070242,
-      "learning_rate": 0.00019999950647886129,
-      "loss": 46.0,
-      "step": 6220
-    },
-    {
-      "epoch": 1.001851926406055,
-      "grad_norm": 0.0023000449873507023,
-      "learning_rate": 0.00019999950631990462,
-      "loss": 46.0,
-      "step": 6221
-    },
-    {
-      "epoch": 1.0020129634848425,
-      "grad_norm": 0.003362311515957117,
-      "learning_rate": 0.00019999950616092233,
-      "loss": 46.0,
-      "step": 6222
-    },
-    {
-      "epoch": 1.0021740005636297,
-      "grad_norm": 0.001914225285872817,
-      "learning_rate": 0.0001999995060019145,
-      "loss": 46.0,
-      "step": 6223
-    },
-    {
-      "epoch": 1.0023350376424172,
-      "grad_norm": 0.001551188761368394,
-      "learning_rate": 0.00019999950584288103,
-      "loss": 46.0,
-      "step": 6224
-    },
-    {
-      "epoch": 1.0024960747212046,
-      "grad_norm": 0.0007930328138172626,
-      "learning_rate": 0.00019999950568382199,
-      "loss": 46.0,
-      "step": 6225
-    },
-    {
-      "epoch": 1.0026571117999918,
-      "grad_norm": 0.0013733393279835582,
-      "learning_rate": 0.00019999950552473733,
-      "loss": 46.0,
-      "step": 6226
-    },
-    {
-      "epoch": 1.0028181488787793,
-      "grad_norm": 0.0002670790418051183,
-      "learning_rate": 0.00019999950536562708,
-      "loss": 46.0,
-      "step": 6227
-    },
-    {
-      "epoch": 1.0029791859575667,
-      "grad_norm": 0.0027424772270023823,
-      "learning_rate": 0.00019999950520649125,
-      "loss": 46.0,
-      "step": 6228
-    },
-    {
-      "epoch": 1.0031402230363542,
-      "grad_norm": 0.0014997286489233375,
-      "learning_rate": 0.00019999950504732982,
-      "loss": 46.0,
-      "step": 6229
-    },
-    {
-      "epoch": 1.0033012601151414,
-      "grad_norm": 0.0002768393896985799,
-      "learning_rate": 0.00019999950488814282,
-      "loss": 46.0,
-      "step": 6230
-    },
-    {
-      "epoch": 1.0034622971939289,
-      "grad_norm": 0.0011887544533237815,
-      "learning_rate": 0.00019999950472893017,
-      "loss": 46.0,
-      "step": 6231
-    },
-    {
-      "epoch": 1.0036233342727163,
-      "grad_norm": 0.0019595029298216105,
-      "learning_rate": 0.00019999950456969193,
-      "loss": 46.0,
-      "step": 6232
-    },
-    {
-      "epoch": 1.0037843713515038,
-      "grad_norm": 0.0012800341937690973,
-      "learning_rate": 0.00019999950441042813,
-      "loss": 46.0,
-      "step": 6233
-    },
-    {
-      "epoch": 1.003945408430291,
-      "grad_norm": 0.0013854592107236385,
-      "learning_rate": 0.00019999950425113875,
-      "loss": 46.0,
-      "step": 6234
-    },
-    {
-      "epoch": 1.0041064455090785,
-      "grad_norm": 0.0010163936531171203,
-      "learning_rate": 0.00019999950409182375,
-      "loss": 46.0,
-      "step": 6235
-    },
-    {
-      "epoch": 1.004267482587866,
-      "grad_norm": 0.000986337661743164,
-      "learning_rate": 0.00019999950393248317,
-      "loss": 46.0,
-      "step": 6236
-    },
-    {
-      "epoch": 1.0044285196666531,
-      "grad_norm": 0.0013223548885434866,
-      "learning_rate": 0.00019999950377311697,
-      "loss": 46.0,
-      "step": 6237
-    },
-    {
-      "epoch": 1.0045895567454406,
-      "grad_norm": 0.005291024222970009,
-      "learning_rate": 0.00019999950361372518,
-      "loss": 46.0,
-      "step": 6238
-    },
-    {
-      "epoch": 1.004750593824228,
-      "grad_norm": 0.001994798658415675,
-      "learning_rate": 0.0001999995034543078,
-      "loss": 46.0,
-      "step": 6239
-    },
-    {
-      "epoch": 1.0049116309030155,
-      "grad_norm": 0.0061939433217048645,
-      "learning_rate": 0.00019999950329486484,
-      "loss": 46.0,
-      "step": 6240
-    },
-    {
-      "epoch": 1.0050726679818027,
-      "grad_norm": 0.0008416906930506229,
-      "learning_rate": 0.00019999950313539627,
-      "loss": 46.0,
-      "step": 6241
-    },
-    {
-      "epoch": 1.0052337050605902,
-      "grad_norm": 0.0013776666019111872,
-      "learning_rate": 0.00019999950297590208,
-      "loss": 46.0,
-      "step": 6242
-    },
-    {
-      "epoch": 1.0053947421393776,
-      "grad_norm": 0.0006404008599929512,
-      "learning_rate": 0.00019999950281638236,
-      "loss": 46.0,
-      "step": 6243
-    },
-    {
-      "epoch": 1.005555779218165,
-      "grad_norm": 0.0013202581321820617,
-      "learning_rate": 0.000199999502656837,
-      "loss": 46.0,
-      "step": 6244
-    },
-    {
-      "epoch": 1.0057168162969523,
-      "grad_norm": 0.0026587299071252346,
-      "learning_rate": 0.00019999950249726604,
-      "loss": 46.0,
-      "step": 6245
-    },
-    {
-      "epoch": 1.0058778533757398,
-      "grad_norm": 0.0008761993958614767,
-      "learning_rate": 0.0001999995023376695,
-      "loss": 46.0,
-      "step": 6246
-    },
-    {
-      "epoch": 1.0060388904545272,
-      "grad_norm": 0.0009400885901413858,
-      "learning_rate": 0.00019999950217804738,
-      "loss": 46.0,
-      "step": 6247
-    },
-    {
-      "epoch": 1.0061999275333144,
-      "grad_norm": 0.0012943897163495421,
-      "learning_rate": 0.00019999950201839964,
-      "loss": 46.0,
-      "step": 6248
-    },
-    {
-      "epoch": 1.006360964612102,
-      "grad_norm": 0.0029224518220871687,
-      "learning_rate": 0.0001999995018587263,
-      "loss": 46.0,
-      "step": 6249
-    },
-    {
-      "epoch": 1.0065220016908893,
-      "grad_norm": 0.003726367373019457,
-      "learning_rate": 0.0001999995016990274,
-      "loss": 46.0,
-      "step": 6250
-    },
-    {
-      "epoch": 1.0066830387696768,
-      "grad_norm": 0.0005758950719609857,
-      "learning_rate": 0.0001999995015393029,
-      "loss": 46.0,
-      "step": 6251
-    },
-    {
-      "epoch": 1.006844075848464,
-      "grad_norm": 0.0006697188946418464,
-      "learning_rate": 0.00019999950137955278,
-      "loss": 46.0,
-      "step": 6252
-    },
-    {
-      "epoch": 1.0070051129272515,
-      "grad_norm": 0.0005896021611988544,
-      "learning_rate": 0.00019999950121977708,
-      "loss": 46.0,
-      "step": 6253
-    },
-    {
-      "epoch": 1.007166150006039,
-      "grad_norm": 0.0008820137591101229,
-      "learning_rate": 0.00019999950105997576,
-      "loss": 46.0,
-      "step": 6254
-    },
-    {
-      "epoch": 1.0073271870848264,
-      "grad_norm": 0.008077237755060196,
-      "learning_rate": 0.00019999950090014888,
-      "loss": 46.0,
-      "step": 6255
-    },
-    {
-      "epoch": 1.0074882241636136,
-      "grad_norm": 0.001512572169303894,
-      "learning_rate": 0.0001999995007402964,
-      "loss": 46.0,
-      "step": 6256
-    },
-    {
-      "epoch": 1.007649261242401,
-      "grad_norm": 0.003671261016279459,
-      "learning_rate": 0.0001999995005804183,
-      "loss": 46.0,
-      "step": 6257
-    },
-    {
-      "epoch": 1.0078102983211885,
-      "grad_norm": 0.0007956259069032967,
-      "learning_rate": 0.00019999950042051462,
-      "loss": 46.0,
-      "step": 6258
-    },
-    {
-      "epoch": 1.0079713353999757,
-      "grad_norm": 0.0009433970553800464,
-      "learning_rate": 0.00019999950026058536,
-      "loss": 46.0,
-      "step": 6259
-    },
-    {
-      "epoch": 1.0081323724787632,
-      "grad_norm": 0.0012183671351522207,
-      "learning_rate": 0.00019999950010063047,
-      "loss": 46.0,
-      "step": 6260
-    },
-    {
-      "epoch": 1.0082934095575506,
-      "grad_norm": 0.0008284273208118975,
-      "learning_rate": 0.00019999949994065002,
-      "loss": 46.0,
-      "step": 6261
-    },
-    {
-      "epoch": 1.008454446636338,
-      "grad_norm": 0.0007555127376690507,
-      "learning_rate": 0.00019999949978064397,
-      "loss": 46.0,
-      "step": 6262
-    },
-    {
-      "epoch": 1.0086154837151253,
-      "grad_norm": 0.000943769293371588,
-      "learning_rate": 0.00019999949962061232,
-      "loss": 46.0,
-      "step": 6263
-    },
-    {
-      "epoch": 1.0087765207939128,
-      "grad_norm": 0.001409639255143702,
-      "learning_rate": 0.00019999949946055505,
-      "loss": 46.0,
-      "step": 6264
-    },
-    {
-      "epoch": 1.0089375578727002,
-      "grad_norm": 0.0004043137887492776,
-      "learning_rate": 0.00019999949930047222,
-      "loss": 46.0,
-      "step": 6265
-    },
-    {
-      "epoch": 1.0090985949514877,
-      "grad_norm": 0.00034933481947518885,
-      "learning_rate": 0.0001999994991403638,
-      "loss": 46.0,
-      "step": 6266
-    },
-    {
-      "epoch": 1.009259632030275,
-      "grad_norm": 0.0020100735127925873,
-      "learning_rate": 0.00019999949898022977,
-      "loss": 46.0,
-      "step": 6267
-    },
-    {
-      "epoch": 1.0094206691090624,
-      "grad_norm": 0.0033263566438108683,
-      "learning_rate": 0.00019999949882007015,
-      "loss": 46.0,
-      "step": 6268
-    },
-    {
-      "epoch": 1.0095817061878498,
-      "grad_norm": 0.001117824693210423,
-      "learning_rate": 0.0001999994986598849,
-      "loss": 46.0,
-      "step": 6269
-    },
-    {
-      "epoch": 1.009742743266637,
-      "grad_norm": 0.0016281973803415895,
-      "learning_rate": 0.00019999949849967407,
-      "loss": 46.0,
-      "step": 6270
-    },
-    {
-      "epoch": 1.0099037803454245,
-      "grad_norm": 0.002180148847401142,
-      "learning_rate": 0.00019999949833943766,
-      "loss": 46.0,
-      "step": 6271
-    },
-    {
-      "epoch": 1.010064817424212,
-      "grad_norm": 0.0015702710952609777,
-      "learning_rate": 0.0001999994981791757,
-      "loss": 46.0,
-      "step": 6272
-    },
-    {
-      "epoch": 1.0102258545029994,
-      "grad_norm": 0.004505687393248081,
-      "learning_rate": 0.00019999949801888809,
-      "loss": 46.0,
-      "step": 6273
-    },
-    {
-      "epoch": 1.0103868915817866,
-      "grad_norm": 0.0006167357787489891,
-      "learning_rate": 0.0001999994978585749,
-      "loss": 46.0,
-      "step": 6274
-    },
-    {
-      "epoch": 1.010547928660574,
-      "grad_norm": 0.0011231355601921678,
-      "learning_rate": 0.0001999994976982361,
-      "loss": 46.0,
-      "step": 6275
-    },
-    {
-      "epoch": 1.0107089657393615,
-      "grad_norm": 0.0024628296960145235,
-      "learning_rate": 0.0001999994975378717,
-      "loss": 46.0,
-      "step": 6276
-    },
-    {
-      "epoch": 1.010870002818149,
-      "grad_norm": 0.0019293796503916383,
-      "learning_rate": 0.00019999949737748172,
-      "loss": 46.0,
-      "step": 6277
-    },
-    {
-      "epoch": 1.0110310398969362,
-      "grad_norm": 0.00152417435310781,
-      "learning_rate": 0.00019999949721706615,
-      "loss": 46.0,
-      "step": 6278
-    },
-    {
-      "epoch": 1.0111920769757237,
-      "grad_norm": 0.0005781169165857136,
-      "learning_rate": 0.000199999497056625,
-      "loss": 46.0,
-      "step": 6279
-    },
-    {
-      "epoch": 1.0113531140545111,
-      "grad_norm": 0.0015225467504933476,
-      "learning_rate": 0.00019999949689615822,
-      "loss": 46.0,
-      "step": 6280
-    },
-    {
-      "epoch": 1.0115141511332983,
-      "grad_norm": 0.006066483445465565,
-      "learning_rate": 0.00019999949673566586,
-      "loss": 46.0,
-      "step": 6281
-    },
-    {
-      "epoch": 1.0116751882120858,
-      "grad_norm": 0.006567680276930332,
-      "learning_rate": 0.00019999949657514794,
-      "loss": 46.0,
-      "step": 6282
-    },
-    {
-      "epoch": 1.0118362252908732,
-      "grad_norm": 0.0009104766068048775,
-      "learning_rate": 0.00019999949641460435,
-      "loss": 46.0,
-      "step": 6283
-    },
-    {
-      "epoch": 1.0119972623696607,
-      "grad_norm": 0.0005526019376702607,
-      "learning_rate": 0.00019999949625403525,
-      "loss": 46.0,
-      "step": 6284
-    },
-    {
-      "epoch": 1.012158299448448,
-      "grad_norm": 0.002163332886993885,
-      "learning_rate": 0.00019999949609344052,
-      "loss": 46.0,
-      "step": 6285
-    },
-    {
-      "epoch": 1.0123193365272354,
-      "grad_norm": 0.00041490071453154087,
-      "learning_rate": 0.00019999949593282017,
-      "loss": 46.0,
-      "step": 6286
-    },
-    {
-      "epoch": 1.0124803736060228,
-      "grad_norm": 0.0006570565747097135,
-      "learning_rate": 0.00019999949577217426,
-      "loss": 46.0,
-      "step": 6287
-    },
-    {
-      "epoch": 1.0126414106848103,
-      "grad_norm": 0.0015555809950456023,
-      "learning_rate": 0.00019999949561150273,
-      "loss": 46.0,
-      "step": 6288
-    },
-    {
-      "epoch": 1.0128024477635975,
-      "grad_norm": 0.0017545541049912572,
-      "learning_rate": 0.00019999949545080562,
-      "loss": 46.0,
-      "step": 6289
-    },
-    {
-      "epoch": 1.012963484842385,
-      "grad_norm": 0.0017391914734616876,
-      "learning_rate": 0.0001999994952900829,
-      "loss": 46.0,
-      "step": 6290
-    },
-    {
-      "epoch": 1.0131245219211724,
-      "grad_norm": 0.003497850615531206,
-      "learning_rate": 0.0001999994951293346,
-      "loss": 46.0,
-      "step": 6291
-    },
-    {
-      "epoch": 1.0132855589999596,
-      "grad_norm": 0.0010639895917847753,
-      "learning_rate": 0.0001999994949685607,
-      "loss": 46.0,
-      "step": 6292
-    },
-    {
-      "epoch": 1.013446596078747,
-      "grad_norm": 0.001168062910437584,
-      "learning_rate": 0.00019999949480776122,
-      "loss": 46.0,
-      "step": 6293
-    },
-    {
-      "epoch": 1.0136076331575345,
-      "grad_norm": 0.002597469836473465,
-      "learning_rate": 0.00019999949464693612,
-      "loss": 46.0,
-      "step": 6294
-    },
-    {
-      "epoch": 1.013768670236322,
-      "grad_norm": 0.0006971464608795941,
-      "learning_rate": 0.00019999949448608543,
-      "loss": 46.0,
-      "step": 6295
-    },
-    {
-      "epoch": 1.0139297073151092,
-      "grad_norm": 0.006824319716542959,
-      "learning_rate": 0.00019999949432520915,
-      "loss": 46.0,
-      "step": 6296
-    },
-    {
-      "epoch": 1.0140907443938967,
-      "grad_norm": 0.004372299648821354,
-      "learning_rate": 0.0001999994941643073,
-      "loss": 46.0,
-      "step": 6297
-    },
-    {
-      "epoch": 1.0142517814726841,
-      "grad_norm": 0.0016161275561898947,
-      "learning_rate": 0.0001999994940033798,
-      "loss": 46.0,
-      "step": 6298
-    },
-    {
-      "epoch": 1.0144128185514716,
-      "grad_norm": 0.0012474050745368004,
-      "learning_rate": 0.00019999949384242674,
-      "loss": 46.0,
-      "step": 6299
-    },
-    {
-      "epoch": 1.0145738556302588,
-      "grad_norm": 0.0009295811760239303,
-      "learning_rate": 0.0001999994936814481,
-      "loss": 46.0,
-      "step": 6300
-    },
-    {
-      "epoch": 1.0147348927090463,
-      "grad_norm": 0.0023029143922030926,
-      "learning_rate": 0.00019999949352044385,
-      "loss": 46.0,
-      "step": 6301
-    },
-    {
-      "epoch": 1.0148959297878337,
-      "grad_norm": 0.003243709448724985,
-      "learning_rate": 0.000199999493359414,
-      "loss": 46.0,
-      "step": 6302
-    },
-    {
-      "epoch": 1.015056966866621,
-      "grad_norm": 0.0025950914714485407,
-      "learning_rate": 0.00019999949319835856,
-      "loss": 46.0,
-      "step": 6303
-    },
-    {
-      "epoch": 1.0152180039454084,
-      "grad_norm": 0.00060755864251405,
-      "learning_rate": 0.0001999994930372775,
-      "loss": 46.0,
-      "step": 6304
-    },
-    {
-      "epoch": 1.0153790410241958,
-      "grad_norm": 0.0015983114717528224,
-      "learning_rate": 0.0001999994928761709,
-      "loss": 46.0,
-      "step": 6305
-    },
-    {
-      "epoch": 1.0155400781029833,
-      "grad_norm": 0.0027972955722361803,
-      "learning_rate": 0.00019999949271503863,
-      "loss": 46.0,
-      "step": 6306
-    },
-    {
-      "epoch": 1.0157011151817705,
-      "grad_norm": 0.0010275689419358969,
-      "learning_rate": 0.00019999949255388084,
-      "loss": 46.0,
-      "step": 6307
-    },
-    {
-      "epoch": 1.015862152260558,
-      "grad_norm": 0.0013191692996770144,
-      "learning_rate": 0.0001999994923926974,
-      "loss": 46.0,
-      "step": 6308
-    },
-    {
-      "epoch": 1.0160231893393454,
-      "grad_norm": 0.0011703158961609006,
-      "learning_rate": 0.00019999949223148842,
-      "loss": 46.0,
-      "step": 6309
-    },
-    {
-      "epoch": 1.0161842264181327,
-      "grad_norm": 0.0005833222530782223,
-      "learning_rate": 0.0001999994920702538,
-      "loss": 46.0,
-      "step": 6310
-    },
-    {
-      "epoch": 1.0163452634969201,
-      "grad_norm": 0.002289395546540618,
-      "learning_rate": 0.00019999949190899362,
-      "loss": 46.0,
-      "step": 6311
-    },
-    {
-      "epoch": 1.0165063005757076,
-      "grad_norm": 0.0058857472613453865,
-      "learning_rate": 0.00019999949174770781,
-      "loss": 46.0,
-      "step": 6312
-    },
-    {
-      "epoch": 1.016667337654495,
-      "grad_norm": 0.004240898415446281,
-      "learning_rate": 0.00019999949158639645,
-      "loss": 46.0,
-      "step": 6313
-    },
-    {
-      "epoch": 1.0168283747332822,
-      "grad_norm": 0.00167121272534132,
-      "learning_rate": 0.00019999949142505944,
-      "loss": 46.0,
-      "step": 6314
-    },
-    {
-      "epoch": 1.0169894118120697,
-      "grad_norm": 0.0036721595097333193,
-      "learning_rate": 0.00019999949126369687,
-      "loss": 46.0,
-      "step": 6315
-    },
-    {
-      "epoch": 1.0171504488908572,
-      "grad_norm": 0.002000790787860751,
-      "learning_rate": 0.0001999994911023087,
-      "loss": 46.0,
-      "step": 6316
-    },
-    {
-      "epoch": 1.0173114859696446,
-      "grad_norm": 0.0015422106953337789,
-      "learning_rate": 0.00019999949094089492,
-      "loss": 46.0,
-      "step": 6317
-    },
-    {
-      "epoch": 1.0174725230484318,
-      "grad_norm": 0.0019068531692028046,
-      "learning_rate": 0.00019999949077945556,
-      "loss": 46.0,
-      "step": 6318
-    },
-    {
-      "epoch": 1.0176335601272193,
-      "grad_norm": 0.0013318938435986638,
-      "learning_rate": 0.0001999994906179906,
-      "loss": 46.0,
-      "step": 6319
-    },
-    {
-      "epoch": 1.0177945972060067,
-      "grad_norm": 0.0036504121962934732,
-      "learning_rate": 0.00019999949045650003,
-      "loss": 46.0,
-      "step": 6320
-    },
-    {
-      "epoch": 1.0179556342847942,
-      "grad_norm": 0.0029523440171033144,
-      "learning_rate": 0.00019999949029498391,
-      "loss": 46.0,
-      "step": 6321
-    },
-    {
-      "epoch": 1.0181166713635814,
-      "grad_norm": 0.003808134701102972,
-      "learning_rate": 0.00019999949013344215,
-      "loss": 46.0,
-      "step": 6322
-    },
-    {
-      "epoch": 1.0182777084423689,
-      "grad_norm": 0.0011177808046340942,
-      "learning_rate": 0.0001999994899718748,
-      "loss": 46.0,
-      "step": 6323
-    },
-    {
-      "epoch": 1.0184387455211563,
-      "grad_norm": 0.0025043736677616835,
-      "learning_rate": 0.0001999994898102819,
-      "loss": 46.0,
-      "step": 6324
-    },
-    {
-      "epoch": 1.0185997825999435,
-      "grad_norm": 0.00926585216075182,
-      "learning_rate": 0.00019999948964866335,
-      "loss": 46.0,
-      "step": 6325
-    },
-    {
-      "epoch": 1.018760819678731,
-      "grad_norm": 0.004121112637221813,
-      "learning_rate": 0.00019999948948701924,
-      "loss": 46.0,
-      "step": 6326
-    },
-    {
-      "epoch": 1.0189218567575185,
-      "grad_norm": 0.0010230483021587133,
-      "learning_rate": 0.00019999948932534952,
-      "loss": 46.0,
-      "step": 6327
-    },
-    {
-      "epoch": 1.019082893836306,
-      "grad_norm": 0.0010295668616890907,
-      "learning_rate": 0.0001999994891636542,
-      "loss": 46.0,
-      "step": 6328
-    },
-    {
-      "epoch": 1.0192439309150931,
-      "grad_norm": 0.00418697576969862,
-      "learning_rate": 0.00019999948900193331,
-      "loss": 46.0,
-      "step": 6329
-    },
-    {
-      "epoch": 1.0194049679938806,
-      "grad_norm": 0.0012874839594587684,
-      "learning_rate": 0.00019999948884018683,
-      "loss": 46.0,
-      "step": 6330
-    },
-    {
-      "epoch": 1.019566005072668,
-      "grad_norm": 0.00042142043821513653,
-      "learning_rate": 0.0001999994886784147,
-      "loss": 46.0,
-      "step": 6331
-    },
-    {
-      "epoch": 1.0197270421514553,
-      "grad_norm": 0.0008691985858604312,
-      "learning_rate": 0.00019999948851661702,
-      "loss": 46.0,
-      "step": 6332
-    },
-    {
-      "epoch": 1.0198880792302427,
-      "grad_norm": 0.004565051756799221,
-      "learning_rate": 0.00019999948835479377,
-      "loss": 46.0,
-      "step": 6333
-    },
-    {
-      "epoch": 1.0200491163090302,
-      "grad_norm": 0.0029316702857613564,
-      "learning_rate": 0.00019999948819294489,
-      "loss": 46.0,
-      "step": 6334
-    },
-    {
-      "epoch": 1.0202101533878176,
-      "grad_norm": 0.0013656109804287553,
-      "learning_rate": 0.0001999994880310704,
-      "loss": 46.0,
-      "step": 6335
-    },
-    {
-      "epoch": 1.0203711904666048,
-      "grad_norm": 0.0027131284587085247,
-      "learning_rate": 0.00019999948786917032,
-      "loss": 46.0,
-      "step": 6336
-    },
-    {
-      "epoch": 1.0205322275453923,
-      "grad_norm": 0.0009259256767109036,
-      "learning_rate": 0.00019999948770724467,
-      "loss": 46.0,
-      "step": 6337
-    },
-    {
-      "epoch": 1.0206932646241798,
-      "grad_norm": 0.0009209402487613261,
-      "learning_rate": 0.0001999994875452934,
-      "loss": 46.0,
-      "step": 6338
-    },
-    {
-      "epoch": 1.0208543017029672,
-      "grad_norm": 0.000972357636783272,
-      "learning_rate": 0.00019999948738331656,
-      "loss": 46.0,
-      "step": 6339
-    },
-    {
-      "epoch": 1.0210153387817544,
-      "grad_norm": 0.0014667310751974583,
-      "learning_rate": 0.0001999994872213141,
-      "loss": 46.0,
-      "step": 6340
-    },
-    {
-      "epoch": 1.0211763758605419,
-      "grad_norm": 0.005109541118144989,
-      "learning_rate": 0.00019999948705928607,
-      "loss": 46.0,
-      "step": 6341
-    },
-    {
-      "epoch": 1.0213374129393293,
-      "grad_norm": 0.0029329066164791584,
-      "learning_rate": 0.00019999948689723246,
-      "loss": 46.0,
-      "step": 6342
-    },
-    {
-      "epoch": 1.0214984500181166,
-      "grad_norm": 0.0017394509632140398,
-      "learning_rate": 0.0001999994867351532,
-      "loss": 46.0,
-      "step": 6343
-    },
-    {
-      "epoch": 1.021659487096904,
-      "grad_norm": 0.000741521071176976,
-      "learning_rate": 0.0001999994865730484,
-      "loss": 46.0,
-      "step": 6344
-    },
-    {
-      "epoch": 1.0218205241756915,
-      "grad_norm": 0.0040016938000917435,
-      "learning_rate": 0.00019999948641091794,
-      "loss": 46.0,
-      "step": 6345
-    },
-    {
-      "epoch": 1.021981561254479,
-      "grad_norm": 0.0010161378886550665,
-      "learning_rate": 0.00019999948624876195,
-      "loss": 46.0,
-      "step": 6346
-    },
-    {
-      "epoch": 1.0221425983332662,
-      "grad_norm": 0.0007833954878151417,
-      "learning_rate": 0.00019999948608658032,
-      "loss": 46.0,
-      "step": 6347
-    },
-    {
-      "epoch": 1.0223036354120536,
-      "grad_norm": 0.00138769019395113,
-      "learning_rate": 0.00019999948592437313,
-      "loss": 46.0,
-      "step": 6348
-    },
-    {
-      "epoch": 1.022464672490841,
-      "grad_norm": 0.0010274132946506143,
-      "learning_rate": 0.00019999948576214033,
-      "loss": 46.0,
-      "step": 6349
-    },
-    {
-      "epoch": 1.0226257095696285,
-      "grad_norm": 0.00833053793758154,
-      "learning_rate": 0.00019999948559988194,
-      "loss": 46.0,
-      "step": 6350
-    },
-    {
-      "epoch": 1.0227867466484157,
-      "grad_norm": 0.0006872293306514621,
-      "learning_rate": 0.00019999948543759794,
-      "loss": 46.0,
-      "step": 6351
-    },
-    {
-      "epoch": 1.0229477837272032,
-      "grad_norm": 0.0022577079944312572,
-      "learning_rate": 0.00019999948527528834,
-      "loss": 46.0,
-      "step": 6352
-    },
-    {
-      "epoch": 1.0231088208059906,
-      "grad_norm": 0.000995549955405295,
-      "learning_rate": 0.0001999994851129532,
-      "loss": 46.0,
-      "step": 6353
-    },
-    {
-      "epoch": 1.0232698578847779,
-      "grad_norm": 0.0005734183941967785,
-      "learning_rate": 0.0001999994849505924,
-      "loss": 46.0,
-      "step": 6354
-    },
-    {
-      "epoch": 1.0234308949635653,
-      "grad_norm": 0.0009836237877607346,
-      "learning_rate": 0.00019999948478820602,
-      "loss": 46.0,
-      "step": 6355
-    },
-    {
-      "epoch": 1.0235919320423528,
-      "grad_norm": 0.001009430387057364,
-      "learning_rate": 0.00019999948462579408,
-      "loss": 46.0,
-      "step": 6356
-    },
-    {
-      "epoch": 1.0237529691211402,
-      "grad_norm": 0.0026525987777858973,
-      "learning_rate": 0.00019999948446335652,
-      "loss": 46.0,
-      "step": 6357
-    },
-    {
-      "epoch": 1.0239140061999275,
-      "grad_norm": 0.0013297498226165771,
-      "learning_rate": 0.00019999948430089338,
-      "loss": 46.0,
-      "step": 6358
-    },
-    {
-      "epoch": 1.024075043278715,
-      "grad_norm": 0.000306474685203284,
-      "learning_rate": 0.00019999948413840463,
-      "loss": 46.0,
-      "step": 6359
-    },
-    {
-      "epoch": 1.0242360803575024,
-      "grad_norm": 0.0015022518346086144,
-      "learning_rate": 0.00019999948397589028,
-      "loss": 46.0,
-      "step": 6360
-    },
-    {
-      "epoch": 1.0243971174362898,
-      "grad_norm": 0.0032586290035396814,
-      "learning_rate": 0.00019999948381335035,
-      "loss": 46.0,
-      "step": 6361
-    },
-    {
-      "epoch": 1.024558154515077,
-      "grad_norm": 0.0025659811217337847,
-      "learning_rate": 0.0001999994836507848,
-      "loss": 46.0,
-      "step": 6362
-    },
-    {
-      "epoch": 1.0247191915938645,
-      "grad_norm": 0.0005430930759757757,
-      "learning_rate": 0.0001999994834881937,
-      "loss": 46.0,
-      "step": 6363
-    },
-    {
-      "epoch": 1.024880228672652,
-      "grad_norm": 0.0007301949081011117,
-      "learning_rate": 0.00019999948332557696,
-      "loss": 46.0,
-      "step": 6364
-    },
-    {
-      "epoch": 1.0250412657514392,
-      "grad_norm": 0.004017373081296682,
-      "learning_rate": 0.00019999948316293465,
-      "loss": 46.0,
-      "step": 6365
-    },
-    {
-      "epoch": 1.0252023028302266,
-      "grad_norm": 0.0013297304976731539,
-      "learning_rate": 0.00019999948300026673,
-      "loss": 46.0,
-      "step": 6366
-    },
-    {
-      "epoch": 1.025363339909014,
-      "grad_norm": 0.00029139453545212746,
-      "learning_rate": 0.00019999948283757325,
-      "loss": 46.0,
-      "step": 6367
-    },
-    {
-      "epoch": 1.0255243769878015,
-      "grad_norm": 0.0018476119730621576,
-      "learning_rate": 0.00019999948267485413,
-      "loss": 46.0,
-      "step": 6368
-    },
-    {
-      "epoch": 1.0256854140665888,
-      "grad_norm": 0.0031040534377098083,
-      "learning_rate": 0.00019999948251210944,
-      "loss": 46.0,
-      "step": 6369
-    },
-    {
-      "epoch": 1.0258464511453762,
-      "grad_norm": 0.0005913755740039051,
-      "learning_rate": 0.00019999948234933918,
-      "loss": 46.0,
-      "step": 6370
-    },
-    {
-      "epoch": 1.0260074882241637,
-      "grad_norm": 0.010243549011647701,
-      "learning_rate": 0.00019999948218654326,
-      "loss": 46.0,
-      "step": 6371
-    },
-    {
-      "epoch": 1.026168525302951,
-      "grad_norm": 0.006522131618112326,
-      "learning_rate": 0.0001999994820237218,
-      "loss": 46.0,
-      "step": 6372
-    },
-    {
-      "epoch": 1.0263295623817383,
-      "grad_norm": 0.0024571893736720085,
-      "learning_rate": 0.0001999994818608747,
-      "loss": 46.0,
-      "step": 6373
-    },
-    {
-      "epoch": 1.0264905994605258,
-      "grad_norm": 0.0019137188792228699,
-      "learning_rate": 0.00019999948169800206,
-      "loss": 46.0,
-      "step": 6374
-    },
-    {
-      "epoch": 1.0266516365393132,
-      "grad_norm": 0.0012045350158587098,
-      "learning_rate": 0.0001999994815351038,
-      "loss": 46.0,
-      "step": 6375
-    },
-    {
-      "epoch": 1.0268126736181005,
-      "grad_norm": 0.002175415400415659,
-      "learning_rate": 0.00019999948137217993,
-      "loss": 46.0,
-      "step": 6376
-    },
-    {
-      "epoch": 1.026973710696888,
-      "grad_norm": 0.0003425965260248631,
-      "learning_rate": 0.00019999948120923047,
-      "loss": 46.0,
-      "step": 6377
-    },
-    {
-      "epoch": 1.0271347477756754,
-      "grad_norm": 0.00037628054269589484,
-      "learning_rate": 0.00019999948104625542,
-      "loss": 46.0,
-      "step": 6378
-    },
-    {
-      "epoch": 1.0272957848544628,
-      "grad_norm": 0.002756637055426836,
-      "learning_rate": 0.0001999994808832548,
-      "loss": 46.0,
-      "step": 6379
-    },
-    {
-      "epoch": 1.02745682193325,
-      "grad_norm": 0.0026547012384980917,
-      "learning_rate": 0.00019999948072022856,
-      "loss": 46.0,
-      "step": 6380
-    },
-    {
-      "epoch": 1.0276178590120375,
-      "grad_norm": 0.0008873838814906776,
-      "learning_rate": 0.00019999948055717673,
-      "loss": 46.0,
-      "step": 6381
-    },
-    {
-      "epoch": 1.027778896090825,
-      "grad_norm": 0.0028432304970920086,
-      "learning_rate": 0.0001999994803940993,
-      "loss": 46.0,
-      "step": 6382
-    },
-    {
-      "epoch": 1.0279399331696124,
-      "grad_norm": 0.003174504265189171,
-      "learning_rate": 0.0001999994802309963,
-      "loss": 46.0,
-      "step": 6383
-    },
-    {
-      "epoch": 1.0281009702483996,
-      "grad_norm": 0.0007440876215696335,
-      "learning_rate": 0.00019999948006786767,
-      "loss": 46.0,
-      "step": 6384
-    },
-    {
-      "epoch": 1.028262007327187,
-      "grad_norm": 0.0019496644381433725,
-      "learning_rate": 0.00019999947990471346,
-      "loss": 46.0,
-      "step": 6385
-    },
-    {
-      "epoch": 1.0284230444059745,
-      "grad_norm": 0.0023042848333716393,
-      "learning_rate": 0.00019999947974153366,
-      "loss": 46.0,
-      "step": 6386
-    },
-    {
-      "epoch": 1.0285840814847618,
-      "grad_norm": 0.0005255836294963956,
-      "learning_rate": 0.00019999947957832824,
-      "loss": 46.0,
-      "step": 6387
-    },
-    {
-      "epoch": 1.0287451185635492,
-      "grad_norm": 0.002195813460275531,
-      "learning_rate": 0.00019999947941509727,
-      "loss": 46.0,
-      "step": 6388
-    },
-    {
-      "epoch": 1.0289061556423367,
-      "grad_norm": 0.0016663081478327513,
-      "learning_rate": 0.00019999947925184065,
-      "loss": 46.0,
-      "step": 6389
-    },
-    {
-      "epoch": 1.0290671927211241,
-      "grad_norm": 0.0007405616925098002,
-      "learning_rate": 0.00019999947908855848,
-      "loss": 46.0,
-      "step": 6390
-    },
-    {
-      "epoch": 1.0292282297999114,
-      "grad_norm": 0.0011065138969570398,
-      "learning_rate": 0.00019999947892525072,
-      "loss": 46.0,
-      "step": 6391
-    },
-    {
-      "epoch": 1.0293892668786988,
-      "grad_norm": 0.001951077370904386,
-      "learning_rate": 0.00019999947876191734,
-      "loss": 46.0,
-      "step": 6392
-    },
-    {
-      "epoch": 1.0295503039574863,
-      "grad_norm": 0.004220605827867985,
-      "learning_rate": 0.00019999947859855837,
-      "loss": 46.0,
-      "step": 6393
-    },
-    {
-      "epoch": 1.0297113410362737,
-      "grad_norm": 0.0007138778455555439,
-      "learning_rate": 0.0001999994784351738,
-      "loss": 46.0,
-      "step": 6394
-    },
-    {
-      "epoch": 1.029872378115061,
-      "grad_norm": 0.0004966930719092488,
-      "learning_rate": 0.00019999947827176363,
-      "loss": 46.0,
-      "step": 6395
-    },
-    {
-      "epoch": 1.0300334151938484,
-      "grad_norm": 0.0006892001256346703,
-      "learning_rate": 0.0001999994781083279,
-      "loss": 46.0,
-      "step": 6396
-    },
-    {
-      "epoch": 1.0301944522726358,
-      "grad_norm": 0.0012400030391290784,
-      "learning_rate": 0.00019999947794486654,
-      "loss": 46.0,
-      "step": 6397
-    },
-    {
-      "epoch": 1.030355489351423,
-      "grad_norm": 0.0011087305610999465,
-      "learning_rate": 0.0001999994777813796,
-      "loss": 46.0,
-      "step": 6398
-    },
-    {
-      "epoch": 1.0305165264302105,
-      "grad_norm": 0.0029914244078099728,
-      "learning_rate": 0.00019999947761786707,
-      "loss": 46.0,
-      "step": 6399
-    },
-    {
-      "epoch": 1.030677563508998,
-      "grad_norm": 0.0016571154119446874,
-      "learning_rate": 0.00019999947745432894,
-      "loss": 46.0,
-      "step": 6400
-    },
-    {
-      "epoch": 1.0308386005877854,
-      "grad_norm": 0.0011288548121228814,
-      "learning_rate": 0.00019999947729076523,
-      "loss": 46.0,
-      "step": 6401
-    },
-    {
-      "epoch": 1.0309996376665727,
-      "grad_norm": 0.0012964156921952963,
-      "learning_rate": 0.0001999994771271759,
-      "loss": 46.0,
-      "step": 6402
-    },
-    {
-      "epoch": 1.03116067474536,
-      "grad_norm": 0.003449299605563283,
-      "learning_rate": 0.00019999947696356095,
-      "loss": 46.0,
-      "step": 6403
-    },
-    {
-      "epoch": 1.0313217118241476,
-      "grad_norm": 0.0014574660453945398,
-      "learning_rate": 0.00019999947679992048,
-      "loss": 46.0,
-      "step": 6404
-    },
-    {
-      "epoch": 1.0314827489029348,
-      "grad_norm": 0.003803375642746687,
-      "learning_rate": 0.00019999947663625438,
-      "loss": 46.0,
-      "step": 6405
-    },
-    {
-      "epoch": 1.0316437859817222,
-      "grad_norm": 0.0012264088727533817,
-      "learning_rate": 0.00019999947647256265,
-      "loss": 46.0,
-      "step": 6406
-    },
-    {
-      "epoch": 1.0318048230605097,
-      "grad_norm": 0.00090862310025841,
-      "learning_rate": 0.00019999947630884536,
-      "loss": 46.0,
-      "step": 6407
-    },
-    {
-      "epoch": 1.0319658601392971,
-      "grad_norm": 0.003105464158579707,
-      "learning_rate": 0.00019999947614510248,
-      "loss": 46.0,
-      "step": 6408
-    },
-    {
-      "epoch": 1.0321268972180844,
-      "grad_norm": 0.0008733617141842842,
-      "learning_rate": 0.00019999947598133399,
-      "loss": 46.0,
-      "step": 6409
-    },
-    {
-      "epoch": 1.0322879342968718,
-      "grad_norm": 0.0036914756055921316,
-      "learning_rate": 0.0001999994758175399,
-      "loss": 46.0,
-      "step": 6410
-    },
-    {
-      "epoch": 1.0324489713756593,
-      "grad_norm": 0.001088437158614397,
-      "learning_rate": 0.00019999947565372024,
-      "loss": 46.0,
-      "step": 6411
-    },
-    {
-      "epoch": 1.0326100084544467,
-      "grad_norm": 0.003957650158554316,
-      "learning_rate": 0.00019999947548987498,
-      "loss": 46.0,
-      "step": 6412
-    },
-    {
-      "epoch": 1.032771045533234,
-      "grad_norm": 0.0009474657708778977,
-      "learning_rate": 0.0001999994753260041,
-      "loss": 46.0,
-      "step": 6413
-    },
-    {
-      "epoch": 1.0329320826120214,
-      "grad_norm": 0.0010248093167319894,
-      "learning_rate": 0.00019999947516210765,
-      "loss": 46.0,
-      "step": 6414
-    },
-    {
-      "epoch": 1.0330931196908089,
-      "grad_norm": 0.002393402624875307,
-      "learning_rate": 0.0001999994749981856,
-      "loss": 46.0,
-      "step": 6415
-    },
-    {
-      "epoch": 1.0332541567695963,
-      "grad_norm": 0.0007063332595862448,
-      "learning_rate": 0.00019999947483423795,
-      "loss": 46.0,
-      "step": 6416
-    },
-    {
-      "epoch": 1.0334151938483835,
-      "grad_norm": 0.000694070418830961,
-      "learning_rate": 0.00019999947467026473,
-      "loss": 46.0,
-      "step": 6417
-    },
-    {
-      "epoch": 1.033576230927171,
-      "grad_norm": 0.003966480027884245,
-      "learning_rate": 0.00019999947450626587,
-      "loss": 46.0,
-      "step": 6418
-    },
-    {
-      "epoch": 1.0337372680059584,
-      "grad_norm": 0.002380999969318509,
-      "learning_rate": 0.00019999947434224145,
-      "loss": 46.0,
-      "step": 6419
-    },
-    {
-      "epoch": 1.0338983050847457,
-      "grad_norm": 0.000830343400593847,
-      "learning_rate": 0.00019999947417819142,
-      "loss": 46.0,
-      "step": 6420
-    },
-    {
-      "epoch": 1.0340593421635331,
-      "grad_norm": 0.008824542164802551,
-      "learning_rate": 0.0001999994740141158,
-      "loss": 46.0,
-      "step": 6421
-    },
-    {
-      "epoch": 1.0342203792423206,
-      "grad_norm": 0.002761711599305272,
-      "learning_rate": 0.00019999947385001459,
-      "loss": 46.0,
-      "step": 6422
-    },
-    {
-      "epoch": 1.034381416321108,
-      "grad_norm": 0.00043466774513944983,
-      "learning_rate": 0.0001999994736858878,
-      "loss": 46.0,
-      "step": 6423
-    },
-    {
-      "epoch": 1.0345424533998953,
-      "grad_norm": 0.0008645984344184399,
-      "learning_rate": 0.00019999947352173538,
-      "loss": 46.0,
-      "step": 6424
-    },
-    {
-      "epoch": 1.0347034904786827,
-      "grad_norm": 0.0027858305256813765,
-      "learning_rate": 0.00019999947335755738,
-      "loss": 46.0,
-      "step": 6425
-    },
-    {
-      "epoch": 1.0348645275574702,
-      "grad_norm": 0.000993607915006578,
-      "learning_rate": 0.0001999994731933538,
-      "loss": 46.0,
-      "step": 6426
-    },
-    {
-      "epoch": 1.0350255646362574,
-      "grad_norm": 0.0009302391554228961,
-      "learning_rate": 0.00019999947302912457,
-      "loss": 46.0,
-      "step": 6427
-    },
-    {
-      "epoch": 1.0351866017150448,
-      "grad_norm": 0.0005664990749210119,
-      "learning_rate": 0.00019999947286486979,
-      "loss": 46.0,
-      "step": 6428
-    },
-    {
-      "epoch": 1.0353476387938323,
-      "grad_norm": 0.003783769905567169,
-      "learning_rate": 0.00019999947270058944,
-      "loss": 46.0,
-      "step": 6429
-    },
-    {
-      "epoch": 1.0355086758726197,
-      "grad_norm": 0.0017254615668207407,
-      "learning_rate": 0.00019999947253628345,
-      "loss": 46.0,
-      "step": 6430
-    },
-    {
-      "epoch": 1.035669712951407,
-      "grad_norm": 0.00509891239926219,
-      "learning_rate": 0.00019999947237195188,
-      "loss": 46.0,
-      "step": 6431
-    },
-    {
-      "epoch": 1.0358307500301944,
-      "grad_norm": 0.0012198666809126735,
-      "learning_rate": 0.00019999947220759472,
-      "loss": 46.0,
-      "step": 6432
-    },
-    {
-      "epoch": 1.0359917871089819,
-      "grad_norm": 0.004192420747131109,
-      "learning_rate": 0.00019999947204321197,
-      "loss": 46.0,
-      "step": 6433
-    },
-    {
-      "epoch": 1.0361528241877693,
-      "grad_norm": 0.0004590976459439844,
-      "learning_rate": 0.00019999947187880363,
-      "loss": 46.0,
-      "step": 6434
-    },
-    {
-      "epoch": 1.0363138612665566,
-      "grad_norm": 0.0010390236275270581,
-      "learning_rate": 0.00019999947171436965,
-      "loss": 46.0,
-      "step": 6435
-    },
-    {
-      "epoch": 1.036474898345344,
-      "grad_norm": 0.00484297564253211,
-      "learning_rate": 0.00019999947154991014,
-      "loss": 46.0,
-      "step": 6436
-    },
-    {
-      "epoch": 1.0366359354241315,
-      "grad_norm": 0.0010238185059279203,
-      "learning_rate": 0.00019999947138542502,
-      "loss": 46.0,
-      "step": 6437
-    },
-    {
-      "epoch": 1.0367969725029187,
-      "grad_norm": 0.00038492766907438636,
-      "learning_rate": 0.00019999947122091425,
-      "loss": 46.0,
-      "step": 6438
-    },
-    {
-      "epoch": 1.0369580095817061,
-      "grad_norm": 0.001970977522432804,
-      "learning_rate": 0.00019999947105637795,
-      "loss": 46.0,
-      "step": 6439
-    },
-    {
-      "epoch": 1.0371190466604936,
-      "grad_norm": 0.0012383307330310345,
-      "learning_rate": 0.00019999947089181604,
-      "loss": 46.0,
-      "step": 6440
-    },
-    {
-      "epoch": 1.037280083739281,
-      "grad_norm": 0.002113117603585124,
-      "learning_rate": 0.00019999947072722848,
-      "loss": 46.0,
-      "step": 6441
-    },
-    {
-      "epoch": 1.0374411208180683,
-      "grad_norm": 0.0011089416220784187,
-      "learning_rate": 0.0001999994705626154,
-      "loss": 46.0,
-      "step": 6442
-    },
-    {
-      "epoch": 1.0376021578968557,
-      "grad_norm": 0.004895965103060007,
-      "learning_rate": 0.0001999994703979767,
-      "loss": 46.0,
-      "step": 6443
-    },
-    {
-      "epoch": 1.0377631949756432,
-      "grad_norm": 0.003429856151342392,
-      "learning_rate": 0.00019999947023331238,
-      "loss": 46.0,
-      "step": 6444
-    },
-    {
-      "epoch": 1.0379242320544306,
-      "grad_norm": 0.0034076583106070757,
-      "learning_rate": 0.0001999994700686225,
-      "loss": 46.0,
-      "step": 6445
-    },
-    {
-      "epoch": 1.0380852691332179,
-      "grad_norm": 0.0024926767218858004,
-      "learning_rate": 0.000199999469903907,
-      "loss": 46.0,
-      "step": 6446
-    },
-    {
-      "epoch": 1.0382463062120053,
-      "grad_norm": 0.0011860548984259367,
-      "learning_rate": 0.00019999946973916593,
-      "loss": 46.0,
-      "step": 6447
-    },
-    {
-      "epoch": 1.0384073432907928,
-      "grad_norm": 0.0075462814420461655,
-      "learning_rate": 0.00019999946957439924,
-      "loss": 46.0,
-      "step": 6448
-    },
-    {
-      "epoch": 1.03856838036958,
-      "grad_norm": 0.0015892903320491314,
-      "learning_rate": 0.00019999946940960696,
-      "loss": 46.0,
-      "step": 6449
-    },
-    {
-      "epoch": 1.0387294174483674,
-      "grad_norm": 0.00043625105172395706,
-      "learning_rate": 0.00019999946924478912,
-      "loss": 46.0,
-      "step": 6450
-    },
-    {
-      "epoch": 1.038890454527155,
-      "grad_norm": 0.0007666315650567412,
-      "learning_rate": 0.00019999946907994564,
-      "loss": 46.0,
-      "step": 6451
-    },
-    {
-      "epoch": 1.0390514916059423,
-      "grad_norm": 0.0015255830949172378,
-      "learning_rate": 0.0001999994689150766,
-      "loss": 46.0,
-      "step": 6452
-    },
-    {
-      "epoch": 1.0392125286847296,
-      "grad_norm": 0.003198272315785289,
-      "learning_rate": 0.00019999946875018192,
-      "loss": 46.0,
-      "step": 6453
-    },
-    {
-      "epoch": 1.039373565763517,
-      "grad_norm": 0.0020768484100699425,
-      "learning_rate": 0.00019999946858526168,
-      "loss": 46.0,
-      "step": 6454
-    },
-    {
-      "epoch": 1.0395346028423045,
-      "grad_norm": 0.0009525313507765532,
-      "learning_rate": 0.00019999946842031585,
-      "loss": 46.0,
-      "step": 6455
-    },
-    {
-      "epoch": 1.039695639921092,
-      "grad_norm": 0.004989515524357557,
-      "learning_rate": 0.0001999994682553444,
-      "loss": 46.0,
-      "step": 6456
-    },
-    {
-      "epoch": 1.0398566769998792,
-      "grad_norm": 0.0013025598600506783,
-      "learning_rate": 0.00019999946809034737,
-      "loss": 46.0,
-      "step": 6457
-    },
-    {
-      "epoch": 1.0400177140786666,
-      "grad_norm": 0.000817545922473073,
-      "learning_rate": 0.00019999946792532476,
-      "loss": 46.0,
-      "step": 6458
-    },
-    {
-      "epoch": 1.040178751157454,
-      "grad_norm": 0.0007511943113058805,
-      "learning_rate": 0.00019999946776027655,
-      "loss": 46.0,
-      "step": 6459
-    },
-    {
-      "epoch": 1.0403397882362413,
-      "grad_norm": 0.00288000819273293,
-      "learning_rate": 0.0001999994675952027,
-      "loss": 46.0,
-      "step": 6460
-    },
-    {
-      "epoch": 1.0405008253150287,
-      "grad_norm": 0.0014183727325871587,
-      "learning_rate": 0.0001999994674301033,
-      "loss": 46.0,
-      "step": 6461
-    },
-    {
-      "epoch": 1.0406618623938162,
-      "grad_norm": 0.0005618046852760017,
-      "learning_rate": 0.0001999994672649783,
-      "loss": 46.0,
-      "step": 6462
-    },
-    {
-      "epoch": 1.0408228994726036,
-      "grad_norm": 0.0005620548618026078,
-      "learning_rate": 0.0001999994670998277,
-      "loss": 46.0,
-      "step": 6463
-    },
-    {
-      "epoch": 1.0409839365513909,
-      "grad_norm": 0.0006562886410392821,
-      "learning_rate": 0.00019999946693465153,
-      "loss": 46.0,
-      "step": 6464
-    },
-    {
-      "epoch": 1.0411449736301783,
-      "grad_norm": 0.0034100054763257504,
-      "learning_rate": 0.00019999946676944972,
-      "loss": 46.0,
-      "step": 6465
-    },
-    {
-      "epoch": 1.0413060107089658,
-      "grad_norm": 0.0016375224804505706,
-      "learning_rate": 0.00019999946660422232,
-      "loss": 46.0,
-      "step": 6466
-    },
-    {
-      "epoch": 1.0414670477877532,
-      "grad_norm": 0.0009322431869804859,
-      "learning_rate": 0.00019999946643896937,
-      "loss": 46.0,
-      "step": 6467
-    },
-    {
-      "epoch": 1.0416280848665405,
-      "grad_norm": 0.0007160952081903815,
-      "learning_rate": 0.00019999946627369077,
-      "loss": 46.0,
-      "step": 6468
-    },
-    {
-      "epoch": 1.041789121945328,
-      "grad_norm": 0.0005553049268200994,
-      "learning_rate": 0.00019999946610838664,
-      "loss": 46.0,
-      "step": 6469
-    },
-    {
-      "epoch": 1.0419501590241154,
-      "grad_norm": 0.0008507245220243931,
-      "learning_rate": 0.00019999946594305686,
-      "loss": 46.0,
-      "step": 6470
-    },
-    {
-      "epoch": 1.0421111961029026,
-      "grad_norm": 0.0011316745076328516,
-      "learning_rate": 0.0001999994657777015,
-      "loss": 46.0,
-      "step": 6471
-    },
-    {
-      "epoch": 1.04227223318169,
-      "grad_norm": 0.0013225803850218654,
-      "learning_rate": 0.00019999946561232056,
-      "loss": 46.0,
-      "step": 6472
-    },
-    {
-      "epoch": 1.0424332702604775,
-      "grad_norm": 0.009100942872464657,
-      "learning_rate": 0.000199999465446914,
-      "loss": 46.0,
-      "step": 6473
-    },
-    {
-      "epoch": 1.042594307339265,
-      "grad_norm": 0.0006715467316098511,
-      "learning_rate": 0.00019999946528148185,
-      "loss": 46.0,
-      "step": 6474
-    },
-    {
-      "epoch": 1.0427553444180522,
-      "grad_norm": 0.008572102524340153,
-      "learning_rate": 0.0001999994651160241,
-      "loss": 46.0,
-      "step": 6475
-    },
-    {
-      "epoch": 1.0429163814968396,
-      "grad_norm": 0.0006895264377817512,
-      "learning_rate": 0.0001999994649505408,
-      "loss": 46.0,
-      "step": 6476
-    },
-    {
-      "epoch": 1.043077418575627,
-      "grad_norm": 0.003065183525905013,
-      "learning_rate": 0.00019999946478503185,
-      "loss": 46.0,
-      "step": 6477
-    },
-    {
-      "epoch": 1.0432384556544145,
-      "grad_norm": 0.0008019735105335712,
-      "learning_rate": 0.00019999946461949736,
-      "loss": 46.0,
-      "step": 6478
-    },
-    {
-      "epoch": 1.0433994927332018,
-      "grad_norm": 0.0020395226310938597,
-      "learning_rate": 0.00019999946445393722,
-      "loss": 46.0,
-      "step": 6479
-    },
-    {
-      "epoch": 1.0435605298119892,
-      "grad_norm": 0.0023664236068725586,
-      "learning_rate": 0.00019999946428835152,
-      "loss": 46.0,
-      "step": 6480
-    },
-    {
-      "epoch": 1.0437215668907767,
-      "grad_norm": 0.0007842747145332396,
-      "learning_rate": 0.00019999946412274023,
-      "loss": 46.0,
-      "step": 6481
-    },
-    {
-      "epoch": 1.043882603969564,
-      "grad_norm": 0.0040185400284826756,
-      "learning_rate": 0.0001999994639571033,
-      "loss": 46.0,
-      "step": 6482
-    },
-    {
-      "epoch": 1.0440436410483513,
-      "grad_norm": 0.0007887196261435747,
-      "learning_rate": 0.00019999946379144082,
-      "loss": 46.0,
-      "step": 6483
-    },
-    {
-      "epoch": 1.0442046781271388,
-      "grad_norm": 0.0006416417891159654,
-      "learning_rate": 0.00019999946362575272,
-      "loss": 46.0,
-      "step": 6484
-    },
-    {
-      "epoch": 1.0443657152059262,
-      "grad_norm": 0.0019829971715807915,
-      "learning_rate": 0.00019999946346003906,
-      "loss": 46.0,
-      "step": 6485
-    },
-    {
-      "epoch": 1.0445267522847135,
-      "grad_norm": 0.0009399994742125273,
-      "learning_rate": 0.00019999946329429978,
-      "loss": 46.0,
-      "step": 6486
-    },
-    {
-      "epoch": 1.044687789363501,
-      "grad_norm": 0.0013273813528940082,
-      "learning_rate": 0.0001999994631285349,
-      "loss": 46.0,
-      "step": 6487
-    },
-    {
-      "epoch": 1.0448488264422884,
-      "grad_norm": 0.0029128033202141523,
-      "learning_rate": 0.00019999946296274444,
-      "loss": 46.0,
-      "step": 6488
-    },
-    {
-      "epoch": 1.0450098635210758,
-      "grad_norm": 0.0012755950447171926,
-      "learning_rate": 0.00019999946279692838,
-      "loss": 46.0,
-      "step": 6489
-    },
-    {
-      "epoch": 1.045170900599863,
-      "grad_norm": 0.0035470547154545784,
-      "learning_rate": 0.00019999946263108673,
-      "loss": 46.0,
-      "step": 6490
-    },
-    {
-      "epoch": 1.0453319376786505,
-      "grad_norm": 0.0008489295141771436,
-      "learning_rate": 0.00019999946246521946,
-      "loss": 46.0,
-      "step": 6491
-    },
-    {
-      "epoch": 1.045492974757438,
-      "grad_norm": 0.0044787791557610035,
-      "learning_rate": 0.00019999946229932664,
-      "loss": 46.0,
-      "step": 6492
-    },
-    {
-      "epoch": 1.0456540118362252,
-      "grad_norm": 0.0030958594288676977,
-      "learning_rate": 0.00019999946213340817,
-      "loss": 46.0,
-      "step": 6493
-    },
-    {
-      "epoch": 1.0458150489150126,
-      "grad_norm": 0.0024783099070191383,
-      "learning_rate": 0.00019999946196746414,
-      "loss": 46.0,
-      "step": 6494
-    },
-    {
-      "epoch": 1.0459760859938,
-      "grad_norm": 0.001907074823975563,
-      "learning_rate": 0.00019999946180149453,
-      "loss": 46.0,
-      "step": 6495
-    },
-    {
-      "epoch": 1.0461371230725875,
-      "grad_norm": 0.001247995998710394,
-      "learning_rate": 0.0001999994616354993,
-      "loss": 46.0,
-      "step": 6496
-    },
-    {
-      "epoch": 1.0462981601513748,
-      "grad_norm": 0.00138873013202101,
-      "learning_rate": 0.00019999946146947845,
-      "loss": 46.0,
-      "step": 6497
-    },
-    {
-      "epoch": 1.0464591972301622,
-      "grad_norm": 0.0014799445634707808,
-      "learning_rate": 0.00019999946130343205,
-      "loss": 46.0,
-      "step": 6498
-    },
-    {
-      "epoch": 1.0466202343089497,
-      "grad_norm": 0.0005235079443082213,
-      "learning_rate": 0.00019999946113736003,
-      "loss": 46.0,
-      "step": 6499
-    },
-    {
-      "epoch": 1.0467812713877371,
-      "grad_norm": 0.0013926014071330428,
-      "learning_rate": 0.00019999946097126246,
-      "loss": 46.0,
-      "step": 6500
-    },
-    {
-      "epoch": 1.0469423084665244,
-      "grad_norm": 0.002305097645148635,
-      "learning_rate": 0.00019999946080513924,
-      "loss": 46.0,
-      "step": 6501
-    },
-    {
-      "epoch": 1.0471033455453118,
-      "grad_norm": 0.0009211329161189497,
-      "learning_rate": 0.00019999946063899043,
-      "loss": 46.0,
-      "step": 6502
-    },
-    {
-      "epoch": 1.0472643826240993,
-      "grad_norm": 0.0018815496005117893,
-      "learning_rate": 0.00019999946047281607,
-      "loss": 46.0,
-      "step": 6503
-    },
-    {
-      "epoch": 1.0474254197028865,
-      "grad_norm": 0.0011471969773992896,
-      "learning_rate": 0.00019999946030661606,
-      "loss": 46.0,
-      "step": 6504
-    },
-    {
-      "epoch": 1.047586456781674,
-      "grad_norm": 0.0037697593215852976,
-      "learning_rate": 0.0001999994601403905,
-      "loss": 46.0,
-      "step": 6505
-    },
-    {
-      "epoch": 1.0477474938604614,
-      "grad_norm": 0.0010831323452293873,
-      "learning_rate": 0.0001999994599741393,
-      "loss": 46.0,
-      "step": 6506
-    },
-    {
-      "epoch": 1.0479085309392489,
-      "grad_norm": 0.0006948586087673903,
-      "learning_rate": 0.00019999945980786254,
-      "loss": 46.0,
-      "step": 6507
-    },
-    {
-      "epoch": 1.048069568018036,
-      "grad_norm": 0.0007108302670530975,
-      "learning_rate": 0.00019999945964156019,
-      "loss": 46.0,
-      "step": 6508
-    },
-    {
-      "epoch": 1.0482306050968235,
-      "grad_norm": 0.001614067587070167,
-      "learning_rate": 0.00019999945947523222,
-      "loss": 46.0,
-      "step": 6509
-    },
-    {
-      "epoch": 1.048391642175611,
-      "grad_norm": 0.001632467843592167,
-      "learning_rate": 0.00019999945930887869,
-      "loss": 46.0,
-      "step": 6510
-    },
-    {
-      "epoch": 1.0485526792543984,
-      "grad_norm": 0.0008568188641220331,
-      "learning_rate": 0.00019999945914249954,
-      "loss": 46.0,
-      "step": 6511
-    },
-    {
-      "epoch": 1.0487137163331857,
-      "grad_norm": 0.0009465326438657939,
-      "learning_rate": 0.0001999994589760948,
-      "loss": 46.0,
-      "step": 6512
-    },
-    {
-      "epoch": 1.0488747534119731,
-      "grad_norm": 0.003011822234839201,
-      "learning_rate": 0.00019999945880966446,
-      "loss": 46.0,
-      "step": 6513
-    },
-    {
-      "epoch": 1.0490357904907606,
-      "grad_norm": 0.0006395489326678216,
-      "learning_rate": 0.00019999945864320853,
-      "loss": 46.0,
-      "step": 6514
-    },
-    {
-      "epoch": 1.0491968275695478,
-      "grad_norm": 0.0012729649897664785,
-      "learning_rate": 0.000199999458476727,
-      "loss": 46.0,
-      "step": 6515
-    },
-    {
-      "epoch": 1.0493578646483352,
-      "grad_norm": 0.0007966042612679303,
-      "learning_rate": 0.0001999994583102199,
-      "loss": 46.0,
-      "step": 6516
-    },
-    {
-      "epoch": 1.0495189017271227,
-      "grad_norm": 0.0023721344769001007,
-      "learning_rate": 0.00019999945814368718,
-      "loss": 46.0,
-      "step": 6517
-    },
-    {
-      "epoch": 1.0496799388059102,
-      "grad_norm": 0.0023809236008673906,
-      "learning_rate": 0.00019999945797712884,
-      "loss": 46.0,
-      "step": 6518
-    },
-    {
-      "epoch": 1.0498409758846974,
-      "grad_norm": 0.002216779161244631,
-      "learning_rate": 0.00019999945781054495,
-      "loss": 46.0,
-      "step": 6519
-    },
-    {
-      "epoch": 1.0500020129634848,
-      "grad_norm": 0.000763561693020165,
-      "learning_rate": 0.00019999945764393546,
-      "loss": 46.0,
-      "step": 6520
-    },
-    {
-      "epoch": 1.0501630500422723,
-      "grad_norm": 0.0009227501577697694,
-      "learning_rate": 0.00019999945747730037,
-      "loss": 46.0,
-      "step": 6521
-    },
-    {
-      "epoch": 1.0503240871210595,
-      "grad_norm": 0.0016767920460551977,
-      "learning_rate": 0.00019999945731063968,
-      "loss": 46.0,
-      "step": 6522
-    },
-    {
-      "epoch": 1.050485124199847,
-      "grad_norm": 0.0015196447493508458,
-      "learning_rate": 0.0001999994571439534,
-      "loss": 46.0,
-      "step": 6523
-    },
-    {
-      "epoch": 1.0506461612786344,
-      "grad_norm": 0.00454733707010746,
-      "learning_rate": 0.00019999945697724152,
-      "loss": 46.0,
-      "step": 6524
-    },
-    {
-      "epoch": 1.0508071983574219,
-      "grad_norm": 0.0025887833908200264,
-      "learning_rate": 0.00019999945681050405,
-      "loss": 46.0,
-      "step": 6525
-    },
-    {
-      "epoch": 1.050968235436209,
-      "grad_norm": 0.0030436452943831682,
-      "learning_rate": 0.000199999456643741,
-      "loss": 46.0,
-      "step": 6526
-    },
-    {
-      "epoch": 1.0511292725149965,
-      "grad_norm": 0.001121029956266284,
-      "learning_rate": 0.00019999945647695234,
-      "loss": 46.0,
-      "step": 6527
-    },
-    {
-      "epoch": 1.051290309593784,
-      "grad_norm": 0.004655222874134779,
-      "learning_rate": 0.00019999945631013805,
-      "loss": 46.0,
-      "step": 6528
-    },
-    {
-      "epoch": 1.0514513466725715,
-      "grad_norm": 0.0009520708117634058,
-      "learning_rate": 0.0001999994561432982,
-      "loss": 46.0,
-      "step": 6529
-    },
-    {
-      "epoch": 1.0516123837513587,
-      "grad_norm": 0.0004847474046982825,
-      "learning_rate": 0.00019999945597643274,
-      "loss": 46.0,
-      "step": 6530
-    },
-    {
-      "epoch": 1.0517734208301461,
-      "grad_norm": 0.003370928578078747,
-      "learning_rate": 0.00019999945580954172,
-      "loss": 46.0,
-      "step": 6531
-    },
-    {
-      "epoch": 1.0519344579089336,
-      "grad_norm": 0.0007878663600422442,
-      "learning_rate": 0.00019999945564262508,
-      "loss": 46.0,
-      "step": 6532
-    },
-    {
-      "epoch": 1.052095494987721,
-      "grad_norm": 0.0005941825802437961,
-      "learning_rate": 0.00019999945547568285,
-      "loss": 46.0,
-      "step": 6533
-    },
-    {
-      "epoch": 1.0522565320665083,
-      "grad_norm": 0.001493097865022719,
-      "learning_rate": 0.00019999945530871501,
-      "loss": 46.0,
-      "step": 6534
-    },
-    {
-      "epoch": 1.0524175691452957,
-      "grad_norm": 0.003453702200204134,
-      "learning_rate": 0.00019999945514172161,
-      "loss": 46.0,
-      "step": 6535
-    },
-    {
-      "epoch": 1.0525786062240832,
-      "grad_norm": 0.0004432977002579719,
-      "learning_rate": 0.00019999945497470257,
-      "loss": 46.0,
-      "step": 6536
-    },
-    {
-      "epoch": 1.0527396433028704,
-      "grad_norm": 0.004447145387530327,
-      "learning_rate": 0.00019999945480765797,
-      "loss": 46.0,
-      "step": 6537
-    },
-    {
-      "epoch": 1.0529006803816578,
-      "grad_norm": 0.004706928972154856,
-      "learning_rate": 0.00019999945464058778,
-      "loss": 46.0,
-      "step": 6538
-    },
-    {
-      "epoch": 1.0530617174604453,
-      "grad_norm": 0.0007073060842230916,
-      "learning_rate": 0.00019999945447349198,
-      "loss": 46.0,
-      "step": 6539
-    },
-    {
-      "epoch": 1.0532227545392328,
-      "grad_norm": 0.001652634353376925,
-      "learning_rate": 0.00019999945430637057,
-      "loss": 46.0,
-      "step": 6540
-    },
-    {
-      "epoch": 1.05338379161802,
-      "grad_norm": 0.0024723096285015345,
-      "learning_rate": 0.00019999945413922362,
-      "loss": 46.0,
-      "step": 6541
-    },
-    {
-      "epoch": 1.0535448286968074,
-      "grad_norm": 0.007637431845068932,
-      "learning_rate": 0.00019999945397205103,
-      "loss": 46.0,
-      "step": 6542
-    },
-    {
-      "epoch": 1.0537058657755949,
-      "grad_norm": 0.005236849654465914,
-      "learning_rate": 0.00019999945380485285,
-      "loss": 46.0,
-      "step": 6543
-    },
-    {
-      "epoch": 1.0538669028543821,
-      "grad_norm": 0.0006668042042292655,
-      "learning_rate": 0.00019999945363762906,
-      "loss": 46.0,
-      "step": 6544
-    },
-    {
-      "epoch": 1.0540279399331696,
-      "grad_norm": 0.0012896248372271657,
-      "learning_rate": 0.0001999994534703797,
-      "loss": 46.0,
-      "step": 6545
-    },
-    {
-      "epoch": 1.054188977011957,
-      "grad_norm": 0.001372764934785664,
-      "learning_rate": 0.00019999945330310474,
-      "loss": 46.0,
-      "step": 6546
-    },
-    {
-      "epoch": 1.0543500140907445,
-      "grad_norm": 0.0012557193404063582,
-      "learning_rate": 0.00019999945313580418,
-      "loss": 46.0,
-      "step": 6547
-    },
-    {
-      "epoch": 1.0545110511695317,
-      "grad_norm": 0.0017768212128430605,
-      "learning_rate": 0.00019999945296847801,
-      "loss": 46.0,
-      "step": 6548
-    },
-    {
-      "epoch": 1.0546720882483192,
-      "grad_norm": 0.002252156613394618,
-      "learning_rate": 0.00019999945280112629,
-      "loss": 46.0,
-      "step": 6549
-    },
-    {
-      "epoch": 1.0548331253271066,
-      "grad_norm": 0.002895511919632554,
-      "learning_rate": 0.00019999945263374892,
-      "loss": 46.0,
-      "step": 6550
-    },
-    {
-      "epoch": 1.054994162405894,
-      "grad_norm": 0.0017416340997442603,
-      "learning_rate": 0.000199999452466346,
-      "loss": 46.0,
-      "step": 6551
-    },
-    {
-      "epoch": 1.0551551994846813,
-      "grad_norm": 0.0004678528930526227,
-      "learning_rate": 0.00019999945229891747,
-      "loss": 46.0,
-      "step": 6552
-    },
-    {
-      "epoch": 1.0553162365634687,
-      "grad_norm": 0.0014010531594976783,
-      "learning_rate": 0.00019999945213146334,
-      "loss": 46.0,
-      "step": 6553
-    },
-    {
-      "epoch": 1.0554772736422562,
-      "grad_norm": 0.00083293387433514,
-      "learning_rate": 0.00019999945196398362,
-      "loss": 46.0,
-      "step": 6554
-    },
-    {
-      "epoch": 1.0556383107210434,
-      "grad_norm": 0.002613247139379382,
-      "learning_rate": 0.00019999945179647832,
-      "loss": 46.0,
-      "step": 6555
-    },
-    {
-      "epoch": 1.0557993477998309,
-      "grad_norm": 0.0029394791927188635,
-      "learning_rate": 0.0001999994516289474,
-      "loss": 46.0,
-      "step": 6556
-    },
-    {
-      "epoch": 1.0559603848786183,
-      "grad_norm": 0.0015253908932209015,
-      "learning_rate": 0.0001999994514613909,
-      "loss": 46.0,
-      "step": 6557
-    },
-    {
-      "epoch": 1.0561214219574058,
-      "grad_norm": 0.001834549126215279,
-      "learning_rate": 0.00019999945129380882,
-      "loss": 46.0,
-      "step": 6558
-    },
-    {
-      "epoch": 1.056282459036193,
-      "grad_norm": 0.003350884886458516,
-      "learning_rate": 0.0001999994511262011,
-      "loss": 46.0,
-      "step": 6559
-    },
-    {
-      "epoch": 1.0564434961149805,
-      "grad_norm": 0.0011456571519374847,
-      "learning_rate": 0.00019999945095856782,
-      "loss": 46.0,
-      "step": 6560
-    },
-    {
-      "epoch": 1.056604533193768,
-      "grad_norm": 0.0008790491847321391,
-      "learning_rate": 0.00019999945079090894,
-      "loss": 46.0,
-      "step": 6561
-    },
-    {
-      "epoch": 1.0567655702725554,
-      "grad_norm": 0.0011808705748990178,
-      "learning_rate": 0.00019999945062322447,
-      "loss": 46.0,
-      "step": 6562
-    },
-    {
-      "epoch": 1.0569266073513426,
-      "grad_norm": 0.0038276398554444313,
-      "learning_rate": 0.0001999994504555144,
-      "loss": 46.0,
-      "step": 6563
-    },
-    {
-      "epoch": 1.05708764443013,
-      "grad_norm": 0.003473035292699933,
-      "learning_rate": 0.0001999994502877787,
-      "loss": 46.0,
-      "step": 6564
-    },
-    {
-      "epoch": 1.0572486815089175,
-      "grad_norm": 0.001024050754494965,
-      "learning_rate": 0.00019999945012001745,
-      "loss": 46.0,
-      "step": 6565
-    },
-    {
-      "epoch": 1.0574097185877047,
-      "grad_norm": 0.0014409383293241262,
-      "learning_rate": 0.0001999994499522306,
-      "loss": 46.0,
-      "step": 6566
-    },
-    {
-      "epoch": 1.0575707556664922,
-      "grad_norm": 0.0012544054770842195,
-      "learning_rate": 0.00019999944978441815,
-      "loss": 46.0,
-      "step": 6567
-    },
-    {
-      "epoch": 1.0577317927452796,
-      "grad_norm": 0.003618619404733181,
-      "learning_rate": 0.0001999994496165801,
-      "loss": 46.0,
-      "step": 6568
-    },
-    {
-      "epoch": 1.057892829824067,
-      "grad_norm": 0.001071578823029995,
-      "learning_rate": 0.00019999944944871646,
-      "loss": 46.0,
-      "step": 6569
-    },
-    {
-      "epoch": 1.0580538669028543,
-      "grad_norm": 0.0038524705450981855,
-      "learning_rate": 0.00019999944928082724,
-      "loss": 46.0,
-      "step": 6570
-    },
-    {
-      "epoch": 1.0582149039816418,
-      "grad_norm": 0.0068861437030136585,
-      "learning_rate": 0.0001999994491129124,
-      "loss": 46.0,
-      "step": 6571
-    },
-    {
-      "epoch": 1.0583759410604292,
-      "grad_norm": 0.007640148047357798,
-      "learning_rate": 0.00019999944894497199,
-      "loss": 46.0,
-      "step": 6572
-    },
-    {
-      "epoch": 1.0585369781392167,
-      "grad_norm": 0.004169008228927851,
-      "learning_rate": 0.00019999944877700595,
-      "loss": 46.0,
-      "step": 6573
-    },
-    {
-      "epoch": 1.0586980152180039,
-      "grad_norm": 0.0015938846627250314,
-      "learning_rate": 0.00019999944860901435,
-      "loss": 46.0,
-      "step": 6574
-    },
-    {
-      "epoch": 1.0588590522967913,
-      "grad_norm": 0.0011657922295853496,
-      "learning_rate": 0.00019999944844099714,
-      "loss": 46.0,
-      "step": 6575
-    },
-    {
-      "epoch": 1.0590200893755788,
-      "grad_norm": 0.0004942793166264892,
-      "learning_rate": 0.00019999944827295434,
-      "loss": 46.0,
-      "step": 6576
-    },
-    {
-      "epoch": 1.059181126454366,
-      "grad_norm": 0.0010235768277198076,
-      "learning_rate": 0.00019999944810488593,
-      "loss": 46.0,
-      "step": 6577
-    },
-    {
-      "epoch": 1.0593421635331535,
-      "grad_norm": 0.000977733638137579,
-      "learning_rate": 0.00019999944793679193,
-      "loss": 46.0,
-      "step": 6578
-    },
-    {
-      "epoch": 1.059503200611941,
-      "grad_norm": 0.0009617151226848364,
-      "learning_rate": 0.00019999944776867237,
-      "loss": 46.0,
-      "step": 6579
-    },
-    {
-      "epoch": 1.0596642376907284,
-      "grad_norm": 0.001249989727512002,
-      "learning_rate": 0.00019999944760052717,
-      "loss": 46.0,
-      "step": 6580
-    },
-    {
-      "epoch": 1.0598252747695156,
-      "grad_norm": 0.005666469223797321,
-      "learning_rate": 0.00019999944743235638,
-      "loss": 46.0,
-      "step": 6581
-    },
-    {
-      "epoch": 1.059986311848303,
-      "grad_norm": 0.001946413074620068,
-      "learning_rate": 0.00019999944726416003,
-      "loss": 46.0,
-      "step": 6582
-    },
-    {
-      "epoch": 1.0601473489270905,
-      "grad_norm": 0.004809623118489981,
-      "learning_rate": 0.00019999944709593807,
-      "loss": 46.0,
-      "step": 6583
-    },
-    {
-      "epoch": 1.060308386005878,
-      "grad_norm": 0.005720277316868305,
-      "learning_rate": 0.00019999944692769052,
-      "loss": 46.0,
-      "step": 6584
-    },
-    {
-      "epoch": 1.0604694230846652,
-      "grad_norm": 0.005022967234253883,
-      "learning_rate": 0.00019999944675941732,
-      "loss": 46.0,
-      "step": 6585
-    },
-    {
-      "epoch": 1.0606304601634526,
-      "grad_norm": 0.004009624943137169,
-      "learning_rate": 0.00019999944659111857,
-      "loss": 46.0,
-      "step": 6586
-    },
-    {
-      "epoch": 1.06079149724224,
-      "grad_norm": 0.0013781647430732846,
-      "learning_rate": 0.00019999944642279423,
-      "loss": 46.0,
-      "step": 6587
-    },
-    {
-      "epoch": 1.0609525343210273,
-      "grad_norm": 0.000837072788272053,
-      "learning_rate": 0.00019999944625444428,
-      "loss": 46.0,
-      "step": 6588
-    },
-    {
-      "epoch": 1.0611135713998148,
-      "grad_norm": 0.001662730472162366,
-      "learning_rate": 0.00019999944608606877,
-      "loss": 46.0,
-      "step": 6589
-    },
-    {
-      "epoch": 1.0612746084786022,
-      "grad_norm": 0.0010261985007673502,
-      "learning_rate": 0.00019999944591766764,
-      "loss": 46.0,
-      "step": 6590
-    },
-    {
-      "epoch": 1.0614356455573897,
-      "grad_norm": 0.0037187645211815834,
-      "learning_rate": 0.0001999994457492409,
-      "loss": 46.0,
-      "step": 6591
-    },
-    {
-      "epoch": 1.061596682636177,
-      "grad_norm": 0.00648907758295536,
-      "learning_rate": 0.00019999944558078857,
-      "loss": 46.0,
-      "step": 6592
-    },
-    {
-      "epoch": 1.0617577197149644,
-      "grad_norm": 0.003581485478207469,
-      "learning_rate": 0.00019999944541231068,
-      "loss": 46.0,
-      "step": 6593
-    },
-    {
-      "epoch": 1.0619187567937518,
-      "grad_norm": 0.0022922155912965536,
-      "learning_rate": 0.00019999944524380715,
-      "loss": 46.0,
-      "step": 6594
-    },
-    {
-      "epoch": 1.0620797938725393,
-      "grad_norm": 0.002288256539031863,
-      "learning_rate": 0.00019999944507527806,
-      "loss": 46.0,
-      "step": 6595
-    },
-    {
-      "epoch": 1.0622408309513265,
-      "grad_norm": 0.002190555213019252,
-      "learning_rate": 0.00019999944490672338,
-      "loss": 46.0,
-      "step": 6596
-    },
-    {
-      "epoch": 1.062401868030114,
-      "grad_norm": 0.0005871456232853234,
-      "learning_rate": 0.00019999944473814306,
-      "loss": 46.0,
-      "step": 6597
-    },
-    {
-      "epoch": 1.0625629051089014,
-      "grad_norm": 0.0016827057115733624,
-      "learning_rate": 0.00019999944456953718,
-      "loss": 46.0,
-      "step": 6598
-    },
-    {
-      "epoch": 1.0627239421876886,
-      "grad_norm": 0.00044768163934350014,
-      "learning_rate": 0.0001999994444009057,
-      "loss": 46.0,
-      "step": 6599
-    },
-    {
-      "epoch": 1.062884979266476,
-      "grad_norm": 0.0016417003935202956,
-      "learning_rate": 0.00019999944423224864,
-      "loss": 46.0,
-      "step": 6600
-    },
-    {
-      "epoch": 1.0630460163452635,
-      "grad_norm": 0.0014568604528903961,
-      "learning_rate": 0.00019999944406356597,
-      "loss": 46.0,
-      "step": 6601
-    },
-    {
-      "epoch": 1.063207053424051,
-      "grad_norm": 0.0007451761630363762,
-      "learning_rate": 0.0001999994438948577,
-      "loss": 46.0,
-      "step": 6602
-    },
-    {
-      "epoch": 1.0633680905028382,
-      "grad_norm": 0.00037166790571063757,
-      "learning_rate": 0.00019999944372612382,
-      "loss": 46.0,
-      "step": 6603
-    },
-    {
-      "epoch": 1.0635291275816257,
-      "grad_norm": 0.007840367034077644,
-      "learning_rate": 0.00019999944355736436,
-      "loss": 46.0,
-      "step": 6604
-    },
-    {
-      "epoch": 1.063690164660413,
-      "grad_norm": 0.0015667991247028112,
-      "learning_rate": 0.00019999944338857932,
-      "loss": 46.0,
-      "step": 6605
-    },
-    {
-      "epoch": 1.0638512017392006,
-      "grad_norm": 0.0016182404942810535,
-      "learning_rate": 0.00019999944321976866,
-      "loss": 46.0,
-      "step": 6606
-    },
-    {
-      "epoch": 1.0640122388179878,
-      "grad_norm": 0.0012526375940069556,
-      "learning_rate": 0.00019999944305093244,
-      "loss": 46.0,
-      "step": 6607
-    },
-    {
-      "epoch": 1.0641732758967752,
-      "grad_norm": 0.003042567288503051,
-      "learning_rate": 0.0001999994428820706,
-      "loss": 46.0,
-      "step": 6608
-    },
-    {
-      "epoch": 1.0643343129755627,
-      "grad_norm": 0.0016685795271769166,
-      "learning_rate": 0.0001999994427131832,
-      "loss": 46.0,
-      "step": 6609
-    },
-    {
-      "epoch": 1.06449535005435,
-      "grad_norm": 0.0008374441531486809,
-      "learning_rate": 0.00019999944254427013,
-      "loss": 46.0,
-      "step": 6610
-    },
-    {
-      "epoch": 1.0646563871331374,
-      "grad_norm": 0.0019776904955506325,
-      "learning_rate": 0.00019999944237533154,
-      "loss": 46.0,
-      "step": 6611
-    },
-    {
-      "epoch": 1.0648174242119248,
-      "grad_norm": 0.001994943479076028,
-      "learning_rate": 0.0001999994422063673,
-      "loss": 46.0,
-      "step": 6612
-    },
-    {
-      "epoch": 1.0649784612907123,
-      "grad_norm": 0.0004205632722005248,
-      "learning_rate": 0.0001999994420373775,
-      "loss": 46.0,
-      "step": 6613
-    },
-    {
-      "epoch": 1.0651394983694995,
-      "grad_norm": 0.004561276640743017,
-      "learning_rate": 0.0001999994418683621,
-      "loss": 46.0,
-      "step": 6614
-    },
-    {
-      "epoch": 1.065300535448287,
-      "grad_norm": 0.0008837788482196629,
-      "learning_rate": 0.0001999994416993211,
-      "loss": 46.0,
-      "step": 6615
-    },
-    {
-      "epoch": 1.0654615725270744,
-      "grad_norm": 0.001033691456541419,
-      "learning_rate": 0.0001999994415302545,
-      "loss": 46.0,
-      "step": 6616
-    },
-    {
-      "epoch": 1.0656226096058616,
-      "grad_norm": 0.004542914219200611,
-      "learning_rate": 0.00019999944136116232,
-      "loss": 46.0,
-      "step": 6617
-    },
-    {
-      "epoch": 1.065783646684649,
-      "grad_norm": 0.001034711254760623,
-      "learning_rate": 0.00019999944119204453,
-      "loss": 46.0,
-      "step": 6618
-    },
-    {
-      "epoch": 1.0659446837634365,
-      "grad_norm": 0.0005485193687491119,
-      "learning_rate": 0.00019999944102290116,
-      "loss": 46.0,
-      "step": 6619
-    },
-    {
-      "epoch": 1.066105720842224,
-      "grad_norm": 0.002726083155721426,
-      "learning_rate": 0.0001999994408537322,
-      "loss": 46.0,
-      "step": 6620
-    },
-    {
-      "epoch": 1.0662667579210112,
-      "grad_norm": 0.002699152799323201,
-      "learning_rate": 0.00019999944068453763,
-      "loss": 46.0,
-      "step": 6621
-    },
-    {
-      "epoch": 1.0664277949997987,
-      "grad_norm": 0.0004922068328596652,
-      "learning_rate": 0.00019999944051531747,
-      "loss": 46.0,
-      "step": 6622
-    },
-    {
-      "epoch": 1.0665888320785861,
-      "grad_norm": 0.0008032437763176858,
-      "learning_rate": 0.0001999994403460717,
-      "loss": 46.0,
-      "step": 6623
-    },
-    {
-      "epoch": 1.0667498691573736,
-      "grad_norm": 0.002333537209779024,
-      "learning_rate": 0.00019999944017680036,
-      "loss": 46.0,
-      "step": 6624
-    },
-    {
-      "epoch": 1.0669109062361608,
-      "grad_norm": 0.0003647351113613695,
-      "learning_rate": 0.0001999994400075034,
-      "loss": 46.0,
-      "step": 6625
-    },
-    {
-      "epoch": 1.0670719433149483,
-      "grad_norm": 0.0071593099273741245,
-      "learning_rate": 0.00019999943983818084,
-      "loss": 46.0,
-      "step": 6626
-    },
-    {
-      "epoch": 1.0672329803937357,
-      "grad_norm": 0.000454216351499781,
-      "learning_rate": 0.00019999943966883272,
-      "loss": 46.0,
-      "step": 6627
-    },
-    {
-      "epoch": 1.0673940174725232,
-      "grad_norm": 0.00444055488333106,
-      "learning_rate": 0.000199999439499459,
-      "loss": 46.0,
-      "step": 6628
-    },
-    {
-      "epoch": 1.0675550545513104,
-      "grad_norm": 0.0045326100662350655,
-      "learning_rate": 0.00019999943933005966,
-      "loss": 46.0,
-      "step": 6629
-    },
-    {
-      "epoch": 1.0677160916300978,
-      "grad_norm": 0.0038118334487080574,
-      "learning_rate": 0.00019999943916063474,
-      "loss": 46.0,
-      "step": 6630
-    },
-    {
-      "epoch": 1.0678771287088853,
-      "grad_norm": 0.0044481027871370316,
-      "learning_rate": 0.00019999943899118425,
-      "loss": 46.0,
-      "step": 6631
-    },
-    {
-      "epoch": 1.0680381657876725,
-      "grad_norm": 0.00202358141541481,
-      "learning_rate": 0.00019999943882170813,
-      "loss": 46.0,
-      "step": 6632
-    },
-    {
-      "epoch": 1.06819920286646,
-      "grad_norm": 0.002532506361603737,
-      "learning_rate": 0.00019999943865220643,
-      "loss": 46.0,
-      "step": 6633
-    },
-    {
-      "epoch": 1.0683602399452474,
-      "grad_norm": 0.0014133837539702654,
-      "learning_rate": 0.00019999943848267914,
-      "loss": 46.0,
-      "step": 6634
-    },
-    {
-      "epoch": 1.0685212770240349,
-      "grad_norm": 0.001062836148776114,
-      "learning_rate": 0.00019999943831312624,
-      "loss": 46.0,
-      "step": 6635
-    },
-    {
-      "epoch": 1.068682314102822,
-      "grad_norm": 0.001685763243585825,
-      "learning_rate": 0.00019999943814354775,
-      "loss": 46.0,
-      "step": 6636
-    },
-    {
-      "epoch": 1.0688433511816096,
-      "grad_norm": 0.0011892104521393776,
-      "learning_rate": 0.0001999994379739437,
-      "loss": 46.0,
-      "step": 6637
-    },
-    {
-      "epoch": 1.069004388260397,
-      "grad_norm": 0.0014539366820827127,
-      "learning_rate": 0.000199999437804314,
-      "loss": 46.0,
-      "step": 6638
-    },
-    {
-      "epoch": 1.0691654253391842,
-      "grad_norm": 0.0028371347580105066,
-      "learning_rate": 0.00019999943763465873,
-      "loss": 46.0,
-      "step": 6639
-    },
-    {
-      "epoch": 1.0693264624179717,
-      "grad_norm": 0.0009564222418703139,
-      "learning_rate": 0.00019999943746497786,
-      "loss": 46.0,
-      "step": 6640
-    },
-    {
-      "epoch": 1.0694874994967591,
-      "grad_norm": 0.002137527335435152,
-      "learning_rate": 0.00019999943729527139,
-      "loss": 46.0,
-      "step": 6641
-    },
-    {
-      "epoch": 1.0696485365755466,
-      "grad_norm": 0.0014893212355673313,
-      "learning_rate": 0.00019999943712553935,
-      "loss": 46.0,
-      "step": 6642
-    },
-    {
-      "epoch": 1.0698095736543338,
-      "grad_norm": 0.0016106904949992895,
-      "learning_rate": 0.0001999994369557817,
-      "loss": 46.0,
-      "step": 6643
-    },
-    {
-      "epoch": 1.0699706107331213,
-      "grad_norm": 0.0011139264097437263,
-      "learning_rate": 0.00019999943678599845,
-      "loss": 46.0,
-      "step": 6644
-    },
-    {
-      "epoch": 1.0701316478119087,
-      "grad_norm": 0.0014995168894529343,
-      "learning_rate": 0.0001999994366161896,
-      "loss": 46.0,
-      "step": 6645
-    },
-    {
-      "epoch": 1.0702926848906962,
-      "grad_norm": 0.0015475673135370016,
-      "learning_rate": 0.00019999943644635518,
-      "loss": 46.0,
-      "step": 6646
-    },
-    {
-      "epoch": 1.0704537219694834,
-      "grad_norm": 0.0013178804656490684,
-      "learning_rate": 0.00019999943627649515,
-      "loss": 46.0,
-      "step": 6647
-    },
-    {
-      "epoch": 1.0706147590482709,
-      "grad_norm": 0.0007449607946909964,
-      "learning_rate": 0.00019999943610660953,
-      "loss": 46.0,
-      "step": 6648
-    },
-    {
-      "epoch": 1.0707757961270583,
-      "grad_norm": 0.0007005247171036899,
-      "learning_rate": 0.0001999994359366983,
-      "loss": 46.0,
-      "step": 6649
-    },
-    {
-      "epoch": 1.0709368332058458,
-      "grad_norm": 0.001908188103698194,
-      "learning_rate": 0.00019999943576676148,
-      "loss": 46.0,
-      "step": 6650
-    },
-    {
-      "epoch": 1.071097870284633,
-      "grad_norm": 0.0014229501830413938,
-      "learning_rate": 0.00019999943559679908,
-      "loss": 46.0,
-      "step": 6651
-    },
-    {
-      "epoch": 1.0712589073634204,
-      "grad_norm": 0.0010961941443383694,
-      "learning_rate": 0.00019999943542681108,
-      "loss": 46.0,
-      "step": 6652
-    },
-    {
-      "epoch": 1.071419944442208,
-      "grad_norm": 0.0024302878882735968,
-      "learning_rate": 0.0001999994352567975,
-      "loss": 46.0,
-      "step": 6653
-    },
-    {
-      "epoch": 1.0715809815209951,
-      "grad_norm": 0.0012020883150398731,
-      "learning_rate": 0.00019999943508675828,
-      "loss": 46.0,
-      "step": 6654
-    },
-    {
-      "epoch": 1.0717420185997826,
-      "grad_norm": 0.003786248154938221,
-      "learning_rate": 0.0001999994349166935,
-      "loss": 46.0,
-      "step": 6655
-    },
-    {
-      "epoch": 1.07190305567857,
-      "grad_norm": 0.0012250854633748531,
-      "learning_rate": 0.00019999943474660313,
-      "loss": 46.0,
-      "step": 6656
-    },
-    {
-      "epoch": 1.0720640927573575,
-      "grad_norm": 0.002798636443912983,
-      "learning_rate": 0.00019999943457648715,
-      "loss": 46.0,
-      "step": 6657
-    },
-    {
-      "epoch": 1.0722251298361447,
-      "grad_norm": 0.0004034322628285736,
-      "learning_rate": 0.00019999943440634555,
-      "loss": 46.0,
-      "step": 6658
-    },
-    {
-      "epoch": 1.0723861669149322,
-      "grad_norm": 0.0023523124400526285,
-      "learning_rate": 0.0001999994342361784,
-      "loss": 46.0,
-      "step": 6659
-    },
-    {
-      "epoch": 1.0725472039937196,
-      "grad_norm": 0.0028253584168851376,
-      "learning_rate": 0.00019999943406598565,
-      "loss": 46.0,
-      "step": 6660
-    },
-    {
-      "epoch": 1.0727082410725068,
-      "grad_norm": 0.0015455724205821753,
-      "learning_rate": 0.00019999943389576726,
-      "loss": 46.0,
-      "step": 6661
-    },
-    {
-      "epoch": 1.0728692781512943,
-      "grad_norm": 0.0009458760032430291,
-      "learning_rate": 0.00019999943372552335,
-      "loss": 46.0,
-      "step": 6662
-    },
-    {
-      "epoch": 1.0730303152300817,
-      "grad_norm": 0.007920492440462112,
-      "learning_rate": 0.00019999943355525379,
-      "loss": 46.0,
-      "step": 6663
-    },
-    {
-      "epoch": 1.0731913523088692,
-      "grad_norm": 0.0004345966153778136,
-      "learning_rate": 0.00019999943338495866,
-      "loss": 46.0,
-      "step": 6664
-    },
-    {
-      "epoch": 1.0733523893876564,
-      "grad_norm": 0.005320028867572546,
-      "learning_rate": 0.00019999943321463793,
-      "loss": 46.0,
-      "step": 6665
-    },
-    {
-      "epoch": 1.0735134264664439,
-      "grad_norm": 0.0019873802084475756,
-      "learning_rate": 0.00019999943304429158,
-      "loss": 46.0,
-      "step": 6666
-    },
-    {
-      "epoch": 1.0736744635452313,
-      "grad_norm": 0.0007979539805091918,
-      "learning_rate": 0.00019999943287391967,
-      "loss": 46.0,
-      "step": 6667
-    },
-    {
-      "epoch": 1.0738355006240188,
-      "grad_norm": 0.0006145013612695038,
-      "learning_rate": 0.00019999943270352215,
-      "loss": 46.0,
-      "step": 6668
-    },
-    {
-      "epoch": 1.073996537702806,
-      "grad_norm": 0.0025720722042024136,
-      "learning_rate": 0.00019999943253309904,
-      "loss": 46.0,
-      "step": 6669
-    },
-    {
-      "epoch": 1.0741575747815935,
-      "grad_norm": 0.0013371184468269348,
-      "learning_rate": 0.00019999943236265034,
-      "loss": 46.0,
-      "step": 6670
-    },
-    {
-      "epoch": 1.074318611860381,
-      "grad_norm": 0.0010885385563597083,
-      "learning_rate": 0.00019999943219217603,
-      "loss": 46.0,
-      "step": 6671
-    },
-    {
-      "epoch": 1.0744796489391681,
-      "grad_norm": 0.003194937715306878,
-      "learning_rate": 0.00019999943202167613,
-      "loss": 46.0,
-      "step": 6672
-    },
-    {
-      "epoch": 1.0746406860179556,
-      "grad_norm": 0.0012660721549764276,
-      "learning_rate": 0.00019999943185115065,
-      "loss": 46.0,
-      "step": 6673
-    },
-    {
-      "epoch": 1.074801723096743,
-      "grad_norm": 0.0026645513717085123,
-      "learning_rate": 0.00019999943168059954,
-      "loss": 46.0,
-      "step": 6674
-    },
-    {
-      "epoch": 1.0749627601755305,
-      "grad_norm": 0.0016193187329918146,
-      "learning_rate": 0.00019999943151002288,
-      "loss": 46.0,
-      "step": 6675
-    },
-    {
-      "epoch": 1.0751237972543177,
-      "grad_norm": 0.007118883077055216,
-      "learning_rate": 0.00019999943133942058,
-      "loss": 46.0,
-      "step": 6676
-    },
-    {
-      "epoch": 1.0752848343331052,
-      "grad_norm": 0.0007365164929069579,
-      "learning_rate": 0.00019999943116879272,
-      "loss": 46.0,
-      "step": 6677
-    },
-    {
-      "epoch": 1.0754458714118926,
-      "grad_norm": 0.003612341359257698,
-      "learning_rate": 0.00019999943099813927,
-      "loss": 46.0,
-      "step": 6678
-    },
-    {
-      "epoch": 1.07560690849068,
-      "grad_norm": 0.0019475065637379885,
-      "learning_rate": 0.0001999994308274602,
-      "loss": 46.0,
-      "step": 6679
-    },
-    {
-      "epoch": 1.0757679455694673,
-      "grad_norm": 0.0010829285020008683,
-      "learning_rate": 0.00019999943065675556,
-      "loss": 46.0,
-      "step": 6680
-    },
-    {
-      "epoch": 1.0759289826482548,
-      "grad_norm": 0.003223517443984747,
-      "learning_rate": 0.00019999943048602527,
-      "loss": 46.0,
-      "step": 6681
-    },
-    {
-      "epoch": 1.0760900197270422,
-      "grad_norm": 0.0016091213328763843,
-      "learning_rate": 0.00019999943031526944,
-      "loss": 46.0,
-      "step": 6682
-    },
-    {
-      "epoch": 1.0762510568058294,
-      "grad_norm": 0.001880658557638526,
-      "learning_rate": 0.000199999430144488,
-      "loss": 46.0,
-      "step": 6683
-    },
-    {
-      "epoch": 1.076412093884617,
-      "grad_norm": 0.0012574403081089258,
-      "learning_rate": 0.00019999942997368098,
-      "loss": 46.0,
-      "step": 6684
-    },
-    {
-      "epoch": 1.0765731309634043,
-      "grad_norm": 0.0010737484553828835,
-      "learning_rate": 0.00019999942980284833,
-      "loss": 46.0,
-      "step": 6685
-    },
-    {
-      "epoch": 1.0767341680421918,
-      "grad_norm": 0.0016168591100722551,
-      "learning_rate": 0.0001999994296319901,
-      "loss": 46.0,
-      "step": 6686
-    },
-    {
-      "epoch": 1.076895205120979,
-      "grad_norm": 0.0037596020847558975,
-      "learning_rate": 0.0001999994294611063,
-      "loss": 46.0,
-      "step": 6687
-    },
-    {
-      "epoch": 1.0770562421997665,
-      "grad_norm": 0.0009737680666148663,
-      "learning_rate": 0.00019999942929019692,
-      "loss": 46.0,
-      "step": 6688
-    },
-    {
-      "epoch": 1.077217279278554,
-      "grad_norm": 0.003411609446629882,
-      "learning_rate": 0.00019999942911926187,
-      "loss": 46.0,
-      "step": 6689
-    },
-    {
-      "epoch": 1.0773783163573414,
-      "grad_norm": 0.0033264923840761185,
-      "learning_rate": 0.00019999942894830127,
-      "loss": 46.0,
-      "step": 6690
-    },
-    {
-      "epoch": 1.0775393534361286,
-      "grad_norm": 0.0007492205477319658,
-      "learning_rate": 0.00019999942877731508,
-      "loss": 46.0,
-      "step": 6691
-    },
-    {
-      "epoch": 1.077700390514916,
-      "grad_norm": 0.0008512393105775118,
-      "learning_rate": 0.0001999994286063033,
-      "loss": 46.0,
-      "step": 6692
-    },
-    {
-      "epoch": 1.0778614275937035,
-      "grad_norm": 0.008641490712761879,
-      "learning_rate": 0.00019999942843526588,
-      "loss": 46.0,
-      "step": 6693
-    },
-    {
-      "epoch": 1.0780224646724907,
-      "grad_norm": 0.002022741362452507,
-      "learning_rate": 0.00019999942826420293,
-      "loss": 46.0,
-      "step": 6694
-    },
-    {
-      "epoch": 1.0781835017512782,
-      "grad_norm": 0.0009082197211682796,
-      "learning_rate": 0.00019999942809311434,
-      "loss": 46.0,
-      "step": 6695
-    },
-    {
-      "epoch": 1.0783445388300656,
-      "grad_norm": 0.0042103007435798645,
-      "learning_rate": 0.00019999942792200018,
-      "loss": 46.0,
-      "step": 6696
-    },
-    {
-      "epoch": 1.078505575908853,
-      "grad_norm": 0.0010721717262640595,
-      "learning_rate": 0.00019999942775086042,
-      "loss": 46.0,
-      "step": 6697
-    },
-    {
-      "epoch": 1.0786666129876403,
-      "grad_norm": 0.002388992113992572,
-      "learning_rate": 0.00019999942757969506,
-      "loss": 46.0,
-      "step": 6698
-    },
-    {
-      "epoch": 1.0788276500664278,
-      "grad_norm": 0.002236271509900689,
-      "learning_rate": 0.0001999994274085041,
-      "loss": 46.0,
-      "step": 6699
-    },
-    {
-      "epoch": 1.0789886871452152,
-      "grad_norm": 0.0028281386476010084,
-      "learning_rate": 0.00019999942723728756,
-      "loss": 46.0,
-      "step": 6700
-    },
-    {
-      "epoch": 1.0791497242240027,
-      "grad_norm": 0.0015305830165743828,
-      "learning_rate": 0.0001999994270660454,
-      "loss": 46.0,
-      "step": 6701
-    },
-    {
-      "epoch": 1.07931076130279,
-      "grad_norm": 0.00285886088386178,
-      "learning_rate": 0.00019999942689477766,
-      "loss": 46.0,
-      "step": 6702
-    },
-    {
-      "epoch": 1.0794717983815774,
-      "grad_norm": 0.00805856753140688,
-      "learning_rate": 0.00019999942672348434,
-      "loss": 46.0,
-      "step": 6703
-    },
-    {
-      "epoch": 1.0796328354603648,
-      "grad_norm": 0.003864812897518277,
-      "learning_rate": 0.0001999994265521654,
-      "loss": 46.0,
-      "step": 6704
-    },
-    {
-      "epoch": 1.079793872539152,
-      "grad_norm": 0.0018149057868868113,
-      "learning_rate": 0.00019999942638082086,
-      "loss": 46.0,
-      "step": 6705
-    },
-    {
-      "epoch": 1.0799549096179395,
-      "grad_norm": 0.0015075807459652424,
-      "learning_rate": 0.00019999942620945076,
-      "loss": 46.0,
-      "step": 6706
-    },
-    {
-      "epoch": 1.080115946696727,
-      "grad_norm": 0.0005201415624469519,
-      "learning_rate": 0.00019999942603805506,
-      "loss": 46.0,
-      "step": 6707
-    },
-    {
-      "epoch": 1.0802769837755144,
-      "grad_norm": 0.0013000390026718378,
-      "learning_rate": 0.00019999942586663373,
-      "loss": 46.0,
-      "step": 6708
-    },
-    {
-      "epoch": 1.0804380208543016,
-      "grad_norm": 0.002200833521783352,
-      "learning_rate": 0.00019999942569518683,
-      "loss": 46.0,
-      "step": 6709
-    },
-    {
-      "epoch": 1.080599057933089,
-      "grad_norm": 0.009349943138659,
-      "learning_rate": 0.00019999942552371432,
-      "loss": 46.0,
-      "step": 6710
-    },
-    {
-      "epoch": 1.0807600950118765,
-      "grad_norm": 0.00045281765051186085,
-      "learning_rate": 0.00019999942535221625,
-      "loss": 46.0,
-      "step": 6711
-    },
-    {
-      "epoch": 1.0809211320906638,
-      "grad_norm": 0.005064214114099741,
-      "learning_rate": 0.00019999942518069257,
-      "loss": 46.0,
-      "step": 6712
-    },
-    {
-      "epoch": 1.0810821691694512,
-      "grad_norm": 0.0003647044359240681,
-      "learning_rate": 0.00019999942500914327,
-      "loss": 46.0,
-      "step": 6713
-    },
-    {
-      "epoch": 1.0812432062482387,
-      "grad_norm": 0.004237434361129999,
-      "learning_rate": 0.0001999994248375684,
-      "loss": 46.0,
-      "step": 6714
-    },
-    {
-      "epoch": 1.0814042433270261,
-      "grad_norm": 0.0019238648237660527,
-      "learning_rate": 0.00019999942466596794,
-      "loss": 46.0,
-      "step": 6715
-    },
-    {
-      "epoch": 1.0815652804058133,
-      "grad_norm": 0.0005576108233071864,
-      "learning_rate": 0.00019999942449434185,
-      "loss": 46.0,
-      "step": 6716
-    },
-    {
-      "epoch": 1.0817263174846008,
-      "grad_norm": 0.0017798951594159007,
-      "learning_rate": 0.0001999994243226902,
-      "loss": 46.0,
-      "step": 6717
-    },
-    {
-      "epoch": 1.0818873545633882,
-      "grad_norm": 0.004515098407864571,
-      "learning_rate": 0.00019999942415101294,
-      "loss": 46.0,
-      "step": 6718
-    },
-    {
-      "epoch": 1.0820483916421757,
-      "grad_norm": 0.0021000646520406008,
-      "learning_rate": 0.0001999994239793101,
-      "loss": 46.0,
-      "step": 6719
-    },
-    {
-      "epoch": 1.082209428720963,
-      "grad_norm": 0.0025087560061365366,
-      "learning_rate": 0.00019999942380758166,
-      "loss": 46.0,
-      "step": 6720
-    },
-    {
-      "epoch": 1.0823704657997504,
-      "grad_norm": 0.006699474528431892,
-      "learning_rate": 0.0001999994236358276,
-      "loss": 46.0,
-      "step": 6721
-    },
-    {
-      "epoch": 1.0825315028785378,
-      "grad_norm": 0.0008362880325876176,
-      "learning_rate": 0.00019999942346404797,
-      "loss": 46.0,
-      "step": 6722
-    },
-    {
-      "epoch": 1.0826925399573253,
-      "grad_norm": 0.0028973568696528673,
-      "learning_rate": 0.00019999942329224278,
-      "loss": 46.0,
-      "step": 6723
-    },
-    {
-      "epoch": 1.0828535770361125,
-      "grad_norm": 0.0021264434326440096,
-      "learning_rate": 0.00019999942312041194,
-      "loss": 46.0,
-      "step": 6724
-    },
-    {
-      "epoch": 1.0830146141149,
-      "grad_norm": 0.0006560378824360669,
-      "learning_rate": 0.0001999994229485555,
-      "loss": 46.0,
-      "step": 6725
-    },
-    {
-      "epoch": 1.0831756511936874,
-      "grad_norm": 0.0004560265224426985,
-      "learning_rate": 0.0001999994227766735,
-      "loss": 46.0,
-      "step": 6726
-    },
-    {
-      "epoch": 1.0833366882724746,
-      "grad_norm": 0.005121591035276651,
-      "learning_rate": 0.0001999994226047659,
-      "loss": 46.0,
-      "step": 6727
-    },
-    {
-      "epoch": 1.083497725351262,
-      "grad_norm": 0.0008057549712248147,
-      "learning_rate": 0.00019999942243283269,
-      "loss": 46.0,
-      "step": 6728
-    },
-    {
-      "epoch": 1.0836587624300495,
-      "grad_norm": 0.0007746375631541014,
-      "learning_rate": 0.0001999994222608739,
-      "loss": 46.0,
-      "step": 6729
-    },
-    {
-      "epoch": 1.083819799508837,
-      "grad_norm": 0.0009633756708353758,
-      "learning_rate": 0.00019999942208888952,
-      "loss": 46.0,
-      "step": 6730
-    },
-    {
-      "epoch": 1.0839808365876242,
-      "grad_norm": 0.001698309788480401,
-      "learning_rate": 0.00019999942191687952,
-      "loss": 46.0,
-      "step": 6731
-    },
-    {
-      "epoch": 1.0841418736664117,
-      "grad_norm": 0.0007028720574453473,
-      "learning_rate": 0.00019999942174484393,
-      "loss": 46.0,
-      "step": 6732
-    },
-    {
-      "epoch": 1.0843029107451991,
-      "grad_norm": 0.0017314193537458777,
-      "learning_rate": 0.00019999942157278278,
-      "loss": 46.0,
-      "step": 6733
-    },
-    {
-      "epoch": 1.0844639478239864,
-      "grad_norm": 0.004393136128783226,
-      "learning_rate": 0.000199999421400696,
-      "loss": 46.0,
-      "step": 6734
-    },
-    {
-      "epoch": 1.0846249849027738,
-      "grad_norm": 0.0031233245972543955,
-      "learning_rate": 0.0001999994212285836,
-      "loss": 46.0,
-      "step": 6735
-    },
-    {
-      "epoch": 1.0847860219815613,
-      "grad_norm": 0.0013053212314844131,
-      "learning_rate": 0.00019999942105644567,
-      "loss": 46.0,
-      "step": 6736
-    },
-    {
-      "epoch": 1.0849470590603487,
-      "grad_norm": 0.001647605444304645,
-      "learning_rate": 0.00019999942088428212,
-      "loss": 46.0,
-      "step": 6737
-    },
-    {
-      "epoch": 1.085108096139136,
-      "grad_norm": 0.0013077345211058855,
-      "learning_rate": 0.00019999942071209298,
-      "loss": 46.0,
-      "step": 6738
-    },
-    {
-      "epoch": 1.0852691332179234,
-      "grad_norm": 0.0014701293548569083,
-      "learning_rate": 0.00019999942053987822,
-      "loss": 46.0,
-      "step": 6739
-    },
-    {
-      "epoch": 1.0854301702967109,
-      "grad_norm": 0.0010987964924424887,
-      "learning_rate": 0.0001999994203676379,
-      "loss": 46.0,
-      "step": 6740
-    },
-    {
-      "epoch": 1.0855912073754983,
-      "grad_norm": 0.0034313222859054804,
-      "learning_rate": 0.00019999942019537195,
-      "loss": 46.0,
-      "step": 6741
-    },
-    {
-      "epoch": 1.0857522444542855,
-      "grad_norm": 0.00329037313349545,
-      "learning_rate": 0.00019999942002308044,
-      "loss": 46.0,
-      "step": 6742
-    },
-    {
-      "epoch": 1.085913281533073,
-      "grad_norm": 0.0015830009942874312,
-      "learning_rate": 0.0001999994198507633,
-      "loss": 46.0,
-      "step": 6743
-    },
-    {
-      "epoch": 1.0860743186118604,
-      "grad_norm": 0.00520638981834054,
-      "learning_rate": 0.0001999994196784206,
-      "loss": 46.0,
-      "step": 6744
-    },
-    {
-      "epoch": 1.0862353556906479,
-      "grad_norm": 0.0029223295859992504,
-      "learning_rate": 0.00019999941950605229,
-      "loss": 46.0,
-      "step": 6745
-    },
-    {
-      "epoch": 1.0863963927694351,
-      "grad_norm": 0.0019706296734511852,
-      "learning_rate": 0.00019999941933365837,
-      "loss": 46.0,
-      "step": 6746
-    },
-    {
-      "epoch": 1.0865574298482226,
-      "grad_norm": 0.0026354717556387186,
-      "learning_rate": 0.0001999994191612389,
-      "loss": 46.0,
-      "step": 6747
-    },
-    {
-      "epoch": 1.08671846692701,
-      "grad_norm": 0.000934434006921947,
-      "learning_rate": 0.00019999941898879377,
-      "loss": 46.0,
-      "step": 6748
-    },
-    {
-      "epoch": 1.0868795040057972,
-      "grad_norm": 0.0021225588861852884,
-      "learning_rate": 0.0001999994188163231,
-      "loss": 46.0,
-      "step": 6749
-    },
-    {
-      "epoch": 1.0870405410845847,
-      "grad_norm": 0.0012658957857638597,
-      "learning_rate": 0.00019999941864382682,
-      "loss": 46.0,
-      "step": 6750
-    },
-    {
-      "epoch": 1.0872015781633722,
-      "grad_norm": 0.0008927016751840711,
-      "learning_rate": 0.0001999994184713049,
-      "loss": 46.0,
-      "step": 6751
-    },
-    {
-      "epoch": 1.0873626152421596,
-      "grad_norm": 0.0006779024843126535,
-      "learning_rate": 0.00019999941829875744,
-      "loss": 46.0,
-      "step": 6752
-    },
-    {
-      "epoch": 1.0875236523209468,
-      "grad_norm": 0.0025173358153551817,
-      "learning_rate": 0.00019999941812618436,
-      "loss": 46.0,
-      "step": 6753
-    },
-    {
-      "epoch": 1.0876846893997343,
-      "grad_norm": 0.006518359761685133,
-      "learning_rate": 0.0001999994179535857,
-      "loss": 46.0,
-      "step": 6754
-    },
-    {
-      "epoch": 1.0878457264785217,
-      "grad_norm": 0.0003246386186219752,
-      "learning_rate": 0.00019999941778096143,
-      "loss": 46.0,
-      "step": 6755
-    },
-    {
-      "epoch": 1.088006763557309,
-      "grad_norm": 0.0023419458884745836,
-      "learning_rate": 0.00019999941760831159,
-      "loss": 46.0,
-      "step": 6756
-    },
-    {
-      "epoch": 1.0881678006360964,
-      "grad_norm": 0.0032568045426160097,
-      "learning_rate": 0.00019999941743563615,
-      "loss": 46.0,
-      "step": 6757
-    },
-    {
-      "epoch": 1.0883288377148839,
-      "grad_norm": 0.0010872224811464548,
-      "learning_rate": 0.0001999994172629351,
-      "loss": 46.0,
-      "step": 6758
-    },
-    {
-      "epoch": 1.0884898747936713,
-      "grad_norm": 0.0010463227517902851,
-      "learning_rate": 0.00019999941709020847,
-      "loss": 46.0,
-      "step": 6759
-    },
-    {
-      "epoch": 1.0886509118724585,
-      "grad_norm": 0.0005634522531181574,
-      "learning_rate": 0.00019999941691745623,
-      "loss": 46.0,
-      "step": 6760
-    },
-    {
-      "epoch": 1.088811948951246,
-      "grad_norm": 0.0021014881785959005,
-      "learning_rate": 0.00019999941674467842,
-      "loss": 46.0,
-      "step": 6761
-    },
-    {
-      "epoch": 1.0889729860300335,
-      "grad_norm": 0.001237224554643035,
-      "learning_rate": 0.00019999941657187497,
-      "loss": 46.0,
-      "step": 6762
-    },
-    {
-      "epoch": 1.089134023108821,
-      "grad_norm": 0.0007352237007580698,
-      "learning_rate": 0.00019999941639904596,
-      "loss": 46.0,
-      "step": 6763
-    },
-    {
-      "epoch": 1.0892950601876081,
-      "grad_norm": 0.0009498083963990211,
-      "learning_rate": 0.00019999941622619134,
-      "loss": 46.0,
-      "step": 6764
-    },
-    {
-      "epoch": 1.0894560972663956,
-      "grad_norm": 0.0015944220358505845,
-      "learning_rate": 0.00019999941605331113,
-      "loss": 46.0,
-      "step": 6765
-    },
-    {
-      "epoch": 1.089617134345183,
-      "grad_norm": 0.0013007746310904622,
-      "learning_rate": 0.00019999941588040533,
-      "loss": 46.0,
-      "step": 6766
-    },
-    {
-      "epoch": 1.0897781714239703,
-      "grad_norm": 0.002352488460019231,
-      "learning_rate": 0.00019999941570747397,
-      "loss": 46.0,
-      "step": 6767
-    },
-    {
-      "epoch": 1.0899392085027577,
-      "grad_norm": 0.0005662386538460851,
-      "learning_rate": 0.00019999941553451695,
-      "loss": 46.0,
-      "step": 6768
-    },
-    {
-      "epoch": 1.0901002455815452,
-      "grad_norm": 0.0014966700691729784,
-      "learning_rate": 0.00019999941536153436,
-      "loss": 46.0,
-      "step": 6769
-    },
-    {
-      "epoch": 1.0902612826603326,
-      "grad_norm": 0.0016662016278132796,
-      "learning_rate": 0.00019999941518852619,
-      "loss": 46.0,
-      "step": 6770
-    },
-    {
-      "epoch": 1.0904223197391198,
-      "grad_norm": 0.0013122496893629432,
-      "learning_rate": 0.00019999941501549242,
-      "loss": 46.0,
-      "step": 6771
-    },
-    {
-      "epoch": 1.0905833568179073,
-      "grad_norm": 0.00036764625110663474,
-      "learning_rate": 0.00019999941484243302,
-      "loss": 46.0,
-      "step": 6772
-    },
-    {
-      "epoch": 1.0907443938966948,
-      "grad_norm": 0.0008052490884438157,
-      "learning_rate": 0.0001999994146693481,
-      "loss": 46.0,
-      "step": 6773
-    },
-    {
-      "epoch": 1.0909054309754822,
-      "grad_norm": 0.0010541316587477922,
-      "learning_rate": 0.0001999994144962375,
-      "loss": 46.0,
-      "step": 6774
-    },
-    {
-      "epoch": 1.0910664680542694,
-      "grad_norm": 0.0012618504697456956,
-      "learning_rate": 0.00019999941432310137,
-      "loss": 46.0,
-      "step": 6775
-    },
-    {
-      "epoch": 1.0912275051330569,
-      "grad_norm": 0.0013316760305315256,
-      "learning_rate": 0.00019999941414993962,
-      "loss": 46.0,
-      "step": 6776
-    },
-    {
-      "epoch": 1.0913885422118443,
-      "grad_norm": 0.0008574627572670579,
-      "learning_rate": 0.00019999941397675226,
-      "loss": 46.0,
-      "step": 6777
-    },
-    {
-      "epoch": 1.0915495792906316,
-      "grad_norm": 0.0014363937079906464,
-      "learning_rate": 0.00019999941380353936,
-      "loss": 46.0,
-      "step": 6778
-    },
-    {
-      "epoch": 1.091710616369419,
-      "grad_norm": 0.008141132071614265,
-      "learning_rate": 0.00019999941363030082,
-      "loss": 46.0,
-      "step": 6779
-    },
-    {
-      "epoch": 1.0918716534482065,
-      "grad_norm": 0.0018811540212482214,
-      "learning_rate": 0.00019999941345703667,
-      "loss": 46.0,
-      "step": 6780
-    },
-    {
-      "epoch": 1.092032690526994,
-      "grad_norm": 0.0008768909028731287,
-      "learning_rate": 0.00019999941328374695,
-      "loss": 46.0,
-      "step": 6781
-    },
-    {
-      "epoch": 1.0921937276057812,
-      "grad_norm": 0.002609496470540762,
-      "learning_rate": 0.00019999941311043165,
-      "loss": 46.0,
-      "step": 6782
-    },
-    {
-      "epoch": 1.0923547646845686,
-      "grad_norm": 0.004338674712926149,
-      "learning_rate": 0.0001999994129370907,
-      "loss": 46.0,
-      "step": 6783
-    },
-    {
-      "epoch": 1.092515801763356,
-      "grad_norm": 0.0009369601611979306,
-      "learning_rate": 0.00019999941276372424,
-      "loss": 46.0,
-      "step": 6784
-    },
-    {
-      "epoch": 1.0926768388421435,
-      "grad_norm": 0.002386624226346612,
-      "learning_rate": 0.00019999941259033212,
-      "loss": 46.0,
-      "step": 6785
-    },
-    {
-      "epoch": 1.0928378759209307,
-      "grad_norm": 0.001545778359286487,
-      "learning_rate": 0.00019999941241691442,
-      "loss": 46.0,
-      "step": 6786
-    },
-    {
-      "epoch": 1.0929989129997182,
-      "grad_norm": 0.0007242248393595219,
-      "learning_rate": 0.00019999941224347115,
-      "loss": 46.0,
-      "step": 6787
-    },
-    {
-      "epoch": 1.0931599500785056,
-      "grad_norm": 0.004175287671387196,
-      "learning_rate": 0.00019999941207000225,
-      "loss": 46.0,
-      "step": 6788
-    },
-    {
-      "epoch": 1.0933209871572929,
-      "grad_norm": 0.004481961019337177,
-      "learning_rate": 0.00019999941189650778,
-      "loss": 46.0,
-      "step": 6789
-    },
-    {
-      "epoch": 1.0934820242360803,
-      "grad_norm": 0.0010569783626124263,
-      "learning_rate": 0.0001999994117229877,
-      "loss": 46.0,
-      "step": 6790
-    },
-    {
-      "epoch": 1.0936430613148678,
-      "grad_norm": 0.0016120291547849774,
-      "learning_rate": 0.00019999941154944203,
-      "loss": 46.0,
-      "step": 6791
-    },
-    {
-      "epoch": 1.0938040983936552,
-      "grad_norm": 0.001475730910897255,
-      "learning_rate": 0.00019999941137587075,
-      "loss": 46.0,
-      "step": 6792
-    },
-    {
-      "epoch": 1.0939651354724425,
-      "grad_norm": 0.0006012814701534808,
-      "learning_rate": 0.0001999994112022739,
-      "loss": 46.0,
-      "step": 6793
-    },
-    {
-      "epoch": 1.09412617255123,
-      "grad_norm": 0.001240721670910716,
-      "learning_rate": 0.00019999941102865146,
-      "loss": 46.0,
-      "step": 6794
-    },
-    {
-      "epoch": 1.0942872096300174,
-      "grad_norm": 0.0008815933251753449,
-      "learning_rate": 0.0001999994108550034,
-      "loss": 46.0,
-      "step": 6795
-    },
-    {
-      "epoch": 1.0944482467088048,
-      "grad_norm": 0.0003538366872817278,
-      "learning_rate": 0.00019999941068132976,
-      "loss": 46.0,
-      "step": 6796
-    },
-    {
-      "epoch": 1.094609283787592,
-      "grad_norm": 0.002052916446700692,
-      "learning_rate": 0.0001999994105076305,
-      "loss": 46.0,
-      "step": 6797
-    },
-    {
-      "epoch": 1.0947703208663795,
-      "grad_norm": 0.0032214089296758175,
-      "learning_rate": 0.0001999994103339057,
-      "loss": 46.0,
-      "step": 6798
-    },
-    {
-      "epoch": 1.094931357945167,
-      "grad_norm": 0.0043851411901414394,
-      "learning_rate": 0.00019999941016015526,
-      "loss": 46.0,
-      "step": 6799
-    },
-    {
-      "epoch": 1.0950923950239542,
-      "grad_norm": 0.0014077771920710802,
-      "learning_rate": 0.00019999940998637926,
-      "loss": 46.0,
-      "step": 6800
-    },
-    {
-      "epoch": 1.0952534321027416,
-      "grad_norm": 0.004380984231829643,
-      "learning_rate": 0.0001999994098125776,
-      "loss": 46.0,
-      "step": 6801
-    },
-    {
-      "epoch": 1.095414469181529,
-      "grad_norm": 0.00826328620314598,
-      "learning_rate": 0.0001999994096387504,
-      "loss": 46.0,
-      "step": 6802
-    },
-    {
-      "epoch": 1.0955755062603165,
-      "grad_norm": 0.0007693120278418064,
-      "learning_rate": 0.00019999940946489758,
-      "loss": 46.0,
-      "step": 6803
-    },
-    {
-      "epoch": 1.0957365433391038,
-      "grad_norm": 0.0010625033173710108,
-      "learning_rate": 0.0001999994092910192,
-      "loss": 46.0,
-      "step": 6804
-    },
-    {
-      "epoch": 1.0958975804178912,
-      "grad_norm": 0.0011092345230281353,
-      "learning_rate": 0.0001999994091171152,
-      "loss": 46.0,
-      "step": 6805
-    },
-    {
-      "epoch": 1.0960586174966787,
-      "grad_norm": 0.0014680151361972094,
-      "learning_rate": 0.00019999940894318562,
-      "loss": 46.0,
-      "step": 6806
-    },
-    {
-      "epoch": 1.0962196545754659,
-      "grad_norm": 0.001505075255408883,
-      "learning_rate": 0.00019999940876923043,
-      "loss": 46.0,
-      "step": 6807
-    },
-    {
-      "epoch": 1.0963806916542533,
-      "grad_norm": 0.0010777050629258156,
-      "learning_rate": 0.00019999940859524967,
-      "loss": 46.0,
-      "step": 6808
-    },
-    {
-      "epoch": 1.0965417287330408,
-      "grad_norm": 0.0018195834709331393,
-      "learning_rate": 0.00019999940842124327,
-      "loss": 46.0,
-      "step": 6809
-    },
-    {
-      "epoch": 1.0967027658118282,
-      "grad_norm": 0.0023834342136979103,
-      "learning_rate": 0.00019999940824721134,
-      "loss": 46.0,
-      "step": 6810
-    },
-    {
-      "epoch": 1.0968638028906155,
-      "grad_norm": 0.0006886071059852839,
-      "learning_rate": 0.00019999940807315377,
-      "loss": 46.0,
-      "step": 6811
-    },
-    {
-      "epoch": 1.097024839969403,
-      "grad_norm": 0.0017052574548870325,
-      "learning_rate": 0.00019999940789907058,
-      "loss": 46.0,
-      "step": 6812
-    },
-    {
-      "epoch": 1.0971858770481904,
-      "grad_norm": 0.0003902724711224437,
-      "learning_rate": 0.00019999940772496183,
-      "loss": 46.0,
-      "step": 6813
-    },
-    {
-      "epoch": 1.0973469141269778,
-      "grad_norm": 0.0005600210279226303,
-      "learning_rate": 0.0001999994075508275,
-      "loss": 46.0,
-      "step": 6814
-    },
-    {
-      "epoch": 1.097507951205765,
-      "grad_norm": 0.0019528912380337715,
-      "learning_rate": 0.00019999940737666755,
-      "loss": 46.0,
-      "step": 6815
-    },
-    {
-      "epoch": 1.0976689882845525,
-      "grad_norm": 0.0006542624323628843,
-      "learning_rate": 0.00019999940720248201,
-      "loss": 46.0,
-      "step": 6816
-    },
-    {
-      "epoch": 1.09783002536334,
-      "grad_norm": 0.0009030955261550844,
-      "learning_rate": 0.00019999940702827086,
-      "loss": 46.0,
-      "step": 6817
-    },
-    {
-      "epoch": 1.0979910624421274,
-      "grad_norm": 0.0014593598898500204,
-      "learning_rate": 0.00019999940685403415,
-      "loss": 46.0,
-      "step": 6818
-    },
-    {
-      "epoch": 1.0981520995209146,
-      "grad_norm": 0.000376840413082391,
-      "learning_rate": 0.00019999940667977183,
-      "loss": 46.0,
-      "step": 6819
-    },
-    {
-      "epoch": 1.098313136599702,
-      "grad_norm": 0.001374902785755694,
-      "learning_rate": 0.00019999940650548392,
-      "loss": 46.0,
-      "step": 6820
-    },
-    {
-      "epoch": 1.0984741736784895,
-      "grad_norm": 0.0014474052004516125,
-      "learning_rate": 0.0001999994063311704,
-      "loss": 46.0,
-      "step": 6821
-    },
-    {
-      "epoch": 1.0986352107572768,
-      "grad_norm": 0.0009384119184687734,
-      "learning_rate": 0.0001999994061568313,
-      "loss": 46.0,
-      "step": 6822
-    },
-    {
-      "epoch": 1.0987962478360642,
-      "grad_norm": 0.006101029459387064,
-      "learning_rate": 0.0001999994059824666,
-      "loss": 46.0,
-      "step": 6823
-    },
-    {
-      "epoch": 1.0989572849148517,
-      "grad_norm": 0.00048114280798472464,
-      "learning_rate": 0.0001999994058080763,
-      "loss": 46.0,
-      "step": 6824
-    },
-    {
-      "epoch": 1.0991183219936391,
-      "grad_norm": 0.0008684517233632505,
-      "learning_rate": 0.00019999940563366042,
-      "loss": 46.0,
-      "step": 6825
-    },
-    {
-      "epoch": 1.0992793590724264,
-      "grad_norm": 0.0012686484260484576,
-      "learning_rate": 0.00019999940545921893,
-      "loss": 46.0,
-      "step": 6826
-    },
-    {
-      "epoch": 1.0994403961512138,
-      "grad_norm": 0.002097642980515957,
-      "learning_rate": 0.00019999940528475185,
-      "loss": 46.0,
-      "step": 6827
-    },
-    {
-      "epoch": 1.0996014332300013,
-      "grad_norm": 0.0007745048496872187,
-      "learning_rate": 0.00019999940511025916,
-      "loss": 46.0,
-      "step": 6828
-    },
-    {
-      "epoch": 1.0997624703087885,
-      "grad_norm": 0.0015851400094106793,
-      "learning_rate": 0.00019999940493574088,
-      "loss": 46.0,
-      "step": 6829
-    },
-    {
-      "epoch": 1.099923507387576,
-      "grad_norm": 0.0006307436851784587,
-      "learning_rate": 0.00019999940476119705,
-      "loss": 46.0,
-      "step": 6830
-    },
-    {
-      "epoch": 1.1000845444663634,
-      "grad_norm": 0.003988225478678942,
-      "learning_rate": 0.00019999940458662757,
-      "loss": 46.0,
-      "step": 6831
-    },
-    {
-      "epoch": 1.1002455815451508,
-      "grad_norm": 0.002111879177391529,
-      "learning_rate": 0.00019999940441203253,
-      "loss": 46.0,
-      "step": 6832
-    },
-    {
-      "epoch": 1.100406618623938,
-      "grad_norm": 0.0018911371007561684,
-      "learning_rate": 0.0001999994042374119,
-      "loss": 46.0,
-      "step": 6833
-    },
-    {
-      "epoch": 1.1005676557027255,
-      "grad_norm": 0.0010132085299119353,
-      "learning_rate": 0.00019999940406276566,
-      "loss": 46.0,
-      "step": 6834
-    },
-    {
-      "epoch": 1.100728692781513,
-      "grad_norm": 0.0006456273840740323,
-      "learning_rate": 0.0001999994038880938,
-      "loss": 46.0,
-      "step": 6835
-    },
-    {
-      "epoch": 1.1008897298603004,
-      "grad_norm": 0.000746123434510082,
-      "learning_rate": 0.0001999994037133964,
-      "loss": 46.0,
-      "step": 6836
-    },
-    {
-      "epoch": 1.1010507669390877,
-      "grad_norm": 0.0006802022689953446,
-      "learning_rate": 0.00019999940353867336,
-      "loss": 46.0,
-      "step": 6837
-    },
-    {
-      "epoch": 1.101211804017875,
-      "grad_norm": 0.0008651096140965819,
-      "learning_rate": 0.00019999940336392472,
-      "loss": 46.0,
-      "step": 6838
-    },
-    {
-      "epoch": 1.1013728410966626,
-      "grad_norm": 0.003311104141175747,
-      "learning_rate": 0.00019999940318915052,
-      "loss": 46.0,
-      "step": 6839
-    },
-    {
-      "epoch": 1.10153387817545,
-      "grad_norm": 0.007221200969070196,
-      "learning_rate": 0.0001999994030143507,
-      "loss": 46.0,
-      "step": 6840
-    },
-    {
-      "epoch": 1.1016949152542372,
-      "grad_norm": 0.003303400706499815,
-      "learning_rate": 0.0001999994028395253,
-      "loss": 46.0,
-      "step": 6841
-    },
-    {
-      "epoch": 1.1018559523330247,
-      "grad_norm": 0.0015620229532942176,
-      "learning_rate": 0.0001999994026646743,
-      "loss": 46.0,
-      "step": 6842
-    },
-    {
-      "epoch": 1.1020169894118121,
-      "grad_norm": 0.001295104855671525,
-      "learning_rate": 0.00019999940248979772,
-      "loss": 46.0,
-      "step": 6843
-    },
-    {
-      "epoch": 1.1021780264905994,
-      "grad_norm": 0.0012015128741040826,
-      "learning_rate": 0.0001999994023148955,
-      "loss": 46.0,
-      "step": 6844
-    },
-    {
-      "epoch": 1.1023390635693868,
-      "grad_norm": 0.003155094338580966,
-      "learning_rate": 0.00019999940213996772,
-      "loss": 46.0,
-      "step": 6845
-    },
-    {
-      "epoch": 1.1025001006481743,
-      "grad_norm": 0.001283306279219687,
-      "learning_rate": 0.00019999940196501435,
-      "loss": 46.0,
-      "step": 6846
-    },
-    {
-      "epoch": 1.1026611377269617,
-      "grad_norm": 0.0008306552772410214,
-      "learning_rate": 0.00019999940179003537,
-      "loss": 46.0,
-      "step": 6847
-    },
-    {
-      "epoch": 1.102822174805749,
-      "grad_norm": 0.004457590635865927,
-      "learning_rate": 0.0001999994016150308,
-      "loss": 46.0,
-      "step": 6848
-    },
-    {
-      "epoch": 1.1029832118845364,
-      "grad_norm": 0.0037536320742219687,
-      "learning_rate": 0.00019999940144000067,
-      "loss": 46.0,
-      "step": 6849
-    },
-    {
-      "epoch": 1.1031442489633239,
-      "grad_norm": 0.0015843348810449243,
-      "learning_rate": 0.0001999994012649449,
-      "loss": 46.0,
-      "step": 6850
-    },
-    {
-      "epoch": 1.103305286042111,
-      "grad_norm": 0.0005735429003834724,
-      "learning_rate": 0.00019999940108986355,
-      "loss": 46.0,
-      "step": 6851
-    },
-    {
-      "epoch": 1.1034663231208985,
-      "grad_norm": 0.0034495049621909857,
-      "learning_rate": 0.0001999994009147566,
-      "loss": 46.0,
-      "step": 6852
-    },
-    {
-      "epoch": 1.103627360199686,
-      "grad_norm": 0.0008431568858213723,
-      "learning_rate": 0.00019999940073962407,
-      "loss": 46.0,
-      "step": 6853
-    },
-    {
-      "epoch": 1.1037883972784734,
-      "grad_norm": 0.0005970880738459527,
-      "learning_rate": 0.00019999940056446592,
-      "loss": 46.0,
-      "step": 6854
-    },
-    {
-      "epoch": 1.1039494343572607,
-      "grad_norm": 0.0007126206764951348,
-      "learning_rate": 0.0001999994003892822,
-      "loss": 46.0,
-      "step": 6855
-    },
-    {
-      "epoch": 1.1041104714360481,
-      "grad_norm": 0.000985718797892332,
-      "learning_rate": 0.00019999940021407287,
-      "loss": 46.0,
-      "step": 6856
-    },
-    {
-      "epoch": 1.1042715085148356,
-      "grad_norm": 0.0038014438468962908,
-      "learning_rate": 0.00019999940003883793,
-      "loss": 46.0,
-      "step": 6857
-    },
-    {
-      "epoch": 1.104432545593623,
-      "grad_norm": 0.0004968709545210004,
-      "learning_rate": 0.0001999993998635774,
-      "loss": 46.0,
-      "step": 6858
-    },
-    {
-      "epoch": 1.1045935826724103,
-      "grad_norm": 0.0015719138318672776,
-      "learning_rate": 0.0001999993996882913,
-      "loss": 46.0,
-      "step": 6859
-    },
-    {
-      "epoch": 1.1047546197511977,
-      "grad_norm": 0.006736638955771923,
-      "learning_rate": 0.0001999993995129796,
-      "loss": 46.0,
-      "step": 6860
-    },
-    {
-      "epoch": 1.1049156568299852,
-      "grad_norm": 0.0026190783828496933,
-      "learning_rate": 0.00019999939933764232,
-      "loss": 46.0,
-      "step": 6861
-    },
-    {
-      "epoch": 1.1050766939087726,
-      "grad_norm": 0.0017144321464002132,
-      "learning_rate": 0.00019999939916227942,
-      "loss": 46.0,
-      "step": 6862
-    },
-    {
-      "epoch": 1.1052377309875598,
-      "grad_norm": 0.0010487454710528255,
-      "learning_rate": 0.00019999939898689094,
-      "loss": 46.0,
-      "step": 6863
-    },
-    {
-      "epoch": 1.1053987680663473,
-      "grad_norm": 0.0012597967870533466,
-      "learning_rate": 0.00019999939881147684,
-      "loss": 46.0,
-      "step": 6864
-    },
-    {
-      "epoch": 1.1055598051451347,
-      "grad_norm": 0.0018494793912395835,
-      "learning_rate": 0.00019999939863603715,
-      "loss": 46.0,
-      "step": 6865
-    },
-    {
-      "epoch": 1.105720842223922,
-      "grad_norm": 0.0009795259684324265,
-      "learning_rate": 0.0001999993984605719,
-      "loss": 46.0,
-      "step": 6866
-    },
-    {
-      "epoch": 1.1058818793027094,
-      "grad_norm": 0.0018543446203693748,
-      "learning_rate": 0.00019999939828508102,
-      "loss": 46.0,
-      "step": 6867
-    },
-    {
-      "epoch": 1.1060429163814969,
-      "grad_norm": 0.0024540466256439686,
-      "learning_rate": 0.00019999939810956457,
-      "loss": 46.0,
-      "step": 6868
-    },
-    {
-      "epoch": 1.1062039534602843,
-      "grad_norm": 0.001838721102103591,
-      "learning_rate": 0.00019999939793402248,
-      "loss": 46.0,
-      "step": 6869
-    },
-    {
-      "epoch": 1.1063649905390716,
-      "grad_norm": 0.0008247812511399388,
-      "learning_rate": 0.00019999939775845486,
-      "loss": 46.0,
-      "step": 6870
-    },
-    {
-      "epoch": 1.106526027617859,
-      "grad_norm": 0.005626663099974394,
-      "learning_rate": 0.00019999939758286162,
-      "loss": 46.0,
-      "step": 6871
-    },
-    {
-      "epoch": 1.1066870646966465,
-      "grad_norm": 0.003019805531948805,
-      "learning_rate": 0.00019999939740724277,
-      "loss": 46.0,
-      "step": 6872
-    },
-    {
-      "epoch": 1.1068481017754337,
-      "grad_norm": 0.0013774039689451456,
-      "learning_rate": 0.00019999939723159833,
-      "loss": 46.0,
-      "step": 6873
-    },
-    {
-      "epoch": 1.1070091388542211,
-      "grad_norm": 0.0008389906142838299,
-      "learning_rate": 0.0001999993970559283,
-      "loss": 46.0,
-      "step": 6874
-    },
-    {
-      "epoch": 1.1071701759330086,
-      "grad_norm": 0.002184239448979497,
-      "learning_rate": 0.0001999993968802327,
-      "loss": 46.0,
-      "step": 6875
-    },
-    {
-      "epoch": 1.107331213011796,
-      "grad_norm": 0.002110743895173073,
-      "learning_rate": 0.00019999939670451147,
-      "loss": 46.0,
-      "step": 6876
-    },
-    {
-      "epoch": 1.1074922500905833,
-      "grad_norm": 0.0015071170637384057,
-      "learning_rate": 0.00019999939652876462,
-      "loss": 46.0,
-      "step": 6877
-    },
-    {
-      "epoch": 1.1076532871693707,
-      "grad_norm": 0.0038081109523773193,
-      "learning_rate": 0.00019999939635299222,
-      "loss": 46.0,
-      "step": 6878
-    },
-    {
-      "epoch": 1.1078143242481582,
-      "grad_norm": 0.002234517829492688,
-      "learning_rate": 0.0001999993961771942,
-      "loss": 46.0,
-      "step": 6879
-    },
-    {
-      "epoch": 1.1079753613269456,
-      "grad_norm": 0.0027877003885805607,
-      "learning_rate": 0.00019999939600137063,
-      "loss": 46.0,
-      "step": 6880
-    },
-    {
-      "epoch": 1.1081363984057329,
-      "grad_norm": 0.0018064118921756744,
-      "learning_rate": 0.00019999939582552144,
-      "loss": 46.0,
-      "step": 6881
-    },
-    {
-      "epoch": 1.1082974354845203,
-      "grad_norm": 0.0010408146772533655,
-      "learning_rate": 0.00019999939564964664,
-      "loss": 46.0,
-      "step": 6882
-    },
-    {
-      "epoch": 1.1084584725633078,
-      "grad_norm": 0.006879028398543596,
-      "learning_rate": 0.00019999939547374627,
-      "loss": 46.0,
-      "step": 6883
-    },
-    {
-      "epoch": 1.108619509642095,
-      "grad_norm": 0.0004829927929677069,
-      "learning_rate": 0.00019999939529782027,
-      "loss": 46.0,
-      "step": 6884
-    },
-    {
-      "epoch": 1.1087805467208824,
-      "grad_norm": 0.0015737295616418123,
-      "learning_rate": 0.0001999993951218687,
-      "loss": 46.0,
-      "step": 6885
-    },
-    {
-      "epoch": 1.10894158379967,
-      "grad_norm": 0.0008924181456677616,
-      "learning_rate": 0.00019999939494589155,
-      "loss": 46.0,
-      "step": 6886
-    },
-    {
-      "epoch": 1.1091026208784573,
-      "grad_norm": 0.0016155537450686097,
-      "learning_rate": 0.00019999939476988878,
-      "loss": 46.0,
-      "step": 6887
-    },
-    {
-      "epoch": 1.1092636579572446,
-      "grad_norm": 0.0012315240455791354,
-      "learning_rate": 0.00019999939459386043,
-      "loss": 46.0,
-      "step": 6888
-    },
-    {
-      "epoch": 1.109424695036032,
-      "grad_norm": 0.0035185578744858503,
-      "learning_rate": 0.00019999939441780646,
-      "loss": 46.0,
-      "step": 6889
-    },
-    {
-      "epoch": 1.1095857321148195,
-      "grad_norm": 0.004599418491125107,
-      "learning_rate": 0.0001999993942417269,
-      "loss": 46.0,
-      "step": 6890
-    },
-    {
-      "epoch": 1.109746769193607,
-      "grad_norm": 0.0008205798803828657,
-      "learning_rate": 0.0001999993940656218,
-      "loss": 46.0,
-      "step": 6891
-    },
-    {
-      "epoch": 1.1099078062723942,
-      "grad_norm": 0.0006457548006437719,
-      "learning_rate": 0.00019999939388949106,
-      "loss": 46.0,
-      "step": 6892
-    },
-    {
-      "epoch": 1.1100688433511816,
-      "grad_norm": 0.0024489983916282654,
-      "learning_rate": 0.00019999939371333471,
-      "loss": 46.0,
-      "step": 6893
-    },
-    {
-      "epoch": 1.110229880429969,
-      "grad_norm": 0.002080045873299241,
-      "learning_rate": 0.00019999939353715278,
-      "loss": 46.0,
-      "step": 6894
-    },
-    {
-      "epoch": 1.1103909175087563,
-      "grad_norm": 0.0023526570294052362,
-      "learning_rate": 0.0001999993933609453,
-      "loss": 46.0,
-      "step": 6895
-    },
-    {
-      "epoch": 1.1105519545875437,
-      "grad_norm": 0.0025399865116924047,
-      "learning_rate": 0.00019999939318471216,
-      "loss": 46.0,
-      "step": 6896
-    },
-    {
-      "epoch": 1.1107129916663312,
-      "grad_norm": 0.002852516481652856,
-      "learning_rate": 0.00019999939300845344,
-      "loss": 46.0,
-      "step": 6897
-    },
-    {
-      "epoch": 1.1108740287451186,
-      "grad_norm": 0.0005394486943259835,
-      "learning_rate": 0.00019999939283216916,
-      "loss": 46.0,
-      "step": 6898
-    },
-    {
-      "epoch": 1.1110350658239059,
-      "grad_norm": 0.0025807165075093508,
-      "learning_rate": 0.00019999939265585926,
-      "loss": 46.0,
-      "step": 6899
-    },
-    {
-      "epoch": 1.1111961029026933,
-      "grad_norm": 0.0040620192885398865,
-      "learning_rate": 0.00019999939247952376,
-      "loss": 46.0,
-      "step": 6900
-    },
-    {
-      "epoch": 1.1113571399814808,
-      "grad_norm": 0.0039823055267333984,
-      "learning_rate": 0.00019999939230316266,
-      "loss": 46.0,
-      "step": 6901
-    },
-    {
-      "epoch": 1.111518177060268,
-      "grad_norm": 0.0021288138814270496,
-      "learning_rate": 0.00019999939212677598,
-      "loss": 46.0,
-      "step": 6902
-    },
-    {
-      "epoch": 1.1116792141390555,
-      "grad_norm": 0.0011207035277038813,
-      "learning_rate": 0.00019999939195036373,
-      "loss": 46.0,
-      "step": 6903
-    },
-    {
-      "epoch": 1.111840251217843,
-      "grad_norm": 0.008520621806383133,
-      "learning_rate": 0.00019999939177392585,
-      "loss": 46.0,
-      "step": 6904
-    },
-    {
-      "epoch": 1.1120012882966304,
-      "grad_norm": 0.001497647725045681,
-      "learning_rate": 0.00019999939159746238,
-      "loss": 46.0,
-      "step": 6905
-    },
-    {
-      "epoch": 1.1121623253754176,
-      "grad_norm": 0.001006730948574841,
-      "learning_rate": 0.0001999993914209733,
-      "loss": 46.0,
-      "step": 6906
-    },
-    {
-      "epoch": 1.112323362454205,
-      "grad_norm": 0.0009130086400546134,
-      "learning_rate": 0.00019999939124445865,
-      "loss": 46.0,
-      "step": 6907
-    },
-    {
-      "epoch": 1.1124843995329925,
-      "grad_norm": 0.001340411719866097,
-      "learning_rate": 0.0001999993910679184,
-      "loss": 46.0,
-      "step": 6908
-    },
-    {
-      "epoch": 1.11264543661178,
-      "grad_norm": 0.0021440873388201,
-      "learning_rate": 0.00019999939089135256,
-      "loss": 46.0,
-      "step": 6909
-    },
-    {
-      "epoch": 1.1128064736905672,
-      "grad_norm": 0.00844606477767229,
-      "learning_rate": 0.00019999939071476113,
-      "loss": 46.0,
-      "step": 6910
-    },
-    {
-      "epoch": 1.1129675107693546,
-      "grad_norm": 0.001317870570346713,
-      "learning_rate": 0.00019999939053814408,
-      "loss": 46.0,
-      "step": 6911
-    },
-    {
-      "epoch": 1.113128547848142,
-      "grad_norm": 0.0011348029365763068,
-      "learning_rate": 0.00019999939036150145,
-      "loss": 46.0,
-      "step": 6912
-    },
-    {
-      "epoch": 1.1132895849269295,
-      "grad_norm": 0.00458654435351491,
-      "learning_rate": 0.00019999939018483322,
-      "loss": 46.0,
-      "step": 6913
-    },
-    {
-      "epoch": 1.1134506220057168,
-      "grad_norm": 0.005441383924335241,
-      "learning_rate": 0.0001999993900081394,
-      "loss": 46.0,
-      "step": 6914
-    },
-    {
-      "epoch": 1.1136116590845042,
-      "grad_norm": 0.0005340529605746269,
-      "learning_rate": 0.00019999938983142,
-      "loss": 46.0,
-      "step": 6915
-    },
-    {
-      "epoch": 1.1137726961632917,
-      "grad_norm": 0.001655799918808043,
-      "learning_rate": 0.00019999938965467497,
-      "loss": 46.0,
-      "step": 6916
-    },
-    {
-      "epoch": 1.113933733242079,
-      "grad_norm": 0.0025167756248265505,
-      "learning_rate": 0.00019999938947790438,
-      "loss": 46.0,
-      "step": 6917
-    },
-    {
-      "epoch": 1.1140947703208663,
-      "grad_norm": 0.0007257751422002912,
-      "learning_rate": 0.0001999993893011082,
-      "loss": 46.0,
-      "step": 6918
-    },
-    {
-      "epoch": 1.1142558073996538,
-      "grad_norm": 0.0004469645209610462,
-      "learning_rate": 0.0001999993891242864,
-      "loss": 46.0,
-      "step": 6919
-    },
-    {
-      "epoch": 1.1144168444784412,
-      "grad_norm": 0.0010013979626819491,
-      "learning_rate": 0.000199999388947439,
-      "loss": 46.0,
-      "step": 6920
-    },
-    {
-      "epoch": 1.1145778815572285,
-      "grad_norm": 0.002204634714871645,
-      "learning_rate": 0.000199999388770566,
-      "loss": 46.0,
-      "step": 6921
-    },
-    {
-      "epoch": 1.114738918636016,
-      "grad_norm": 0.0015005613677203655,
-      "learning_rate": 0.00019999938859366744,
-      "loss": 46.0,
-      "step": 6922
-    },
-    {
-      "epoch": 1.1148999557148034,
-      "grad_norm": 0.006144202779978514,
-      "learning_rate": 0.00019999938841674326,
-      "loss": 46.0,
-      "step": 6923
-    },
-    {
-      "epoch": 1.1150609927935906,
-      "grad_norm": 0.0006316326325759292,
-      "learning_rate": 0.00019999938823979353,
-      "loss": 46.0,
-      "step": 6924
-    },
-    {
-      "epoch": 1.115222029872378,
-      "grad_norm": 0.0026485207490622997,
-      "learning_rate": 0.00019999938806281815,
-      "loss": 46.0,
-      "step": 6925
-    },
-    {
-      "epoch": 1.1153830669511655,
-      "grad_norm": 0.0012750618625432253,
-      "learning_rate": 0.0001999993878858172,
-      "loss": 46.0,
-      "step": 6926
-    },
-    {
-      "epoch": 1.115544104029953,
-      "grad_norm": 0.006311360280960798,
-      "learning_rate": 0.00019999938770879063,
-      "loss": 46.0,
-      "step": 6927
-    },
-    {
-      "epoch": 1.1157051411087402,
-      "grad_norm": 0.012786568142473698,
-      "learning_rate": 0.0001999993875317385,
-      "loss": 46.0,
-      "step": 6928
-    },
-    {
-      "epoch": 1.1158661781875276,
-      "grad_norm": 0.0030675893649458885,
-      "learning_rate": 0.00019999938735466074,
-      "loss": 46.0,
-      "step": 6929
-    },
-    {
-      "epoch": 1.116027215266315,
-      "grad_norm": 0.0030196455772966146,
-      "learning_rate": 0.00019999938717755742,
-      "loss": 46.0,
-      "step": 6930
-    },
-    {
-      "epoch": 1.1161882523451025,
-      "grad_norm": 0.0015942202880978584,
-      "learning_rate": 0.0001999993870004285,
-      "loss": 46.0,
-      "step": 6931
-    },
-    {
-      "epoch": 1.1163492894238898,
-      "grad_norm": 0.006608026567846537,
-      "learning_rate": 0.00019999938682327395,
-      "loss": 46.0,
-      "step": 6932
-    },
-    {
-      "epoch": 1.1165103265026772,
-      "grad_norm": 0.0030436262022703886,
-      "learning_rate": 0.00019999938664609385,
-      "loss": 46.0,
-      "step": 6933
-    },
-    {
-      "epoch": 1.1166713635814647,
-      "grad_norm": 0.002232971368357539,
-      "learning_rate": 0.00019999938646888814,
-      "loss": 46.0,
-      "step": 6934
-    },
-    {
-      "epoch": 1.1168324006602521,
-      "grad_norm": 0.000834618229418993,
-      "learning_rate": 0.00019999938629165683,
-      "loss": 46.0,
-      "step": 6935
-    },
-    {
-      "epoch": 1.1169934377390394,
-      "grad_norm": 0.0016175888013094664,
-      "learning_rate": 0.00019999938611439991,
-      "loss": 46.0,
-      "step": 6936
-    },
-    {
-      "epoch": 1.1171544748178268,
-      "grad_norm": 0.0007204286521300673,
-      "learning_rate": 0.00019999938593711744,
-      "loss": 46.0,
-      "step": 6937
-    },
-    {
-      "epoch": 1.1173155118966143,
-      "grad_norm": 0.0012183092767372727,
-      "learning_rate": 0.00019999938575980934,
-      "loss": 46.0,
-      "step": 6938
-    },
-    {
-      "epoch": 1.1174765489754015,
-      "grad_norm": 0.0014345013769343495,
-      "learning_rate": 0.00019999938558247564,
-      "loss": 46.0,
-      "step": 6939
-    },
-    {
-      "epoch": 1.117637586054189,
-      "grad_norm": 0.0010583841940388083,
-      "learning_rate": 0.00019999938540511637,
-      "loss": 46.0,
-      "step": 6940
-    },
-    {
-      "epoch": 1.1177986231329764,
-      "grad_norm": 0.016156338155269623,
-      "learning_rate": 0.0001999993852277315,
-      "loss": 46.0,
-      "step": 6941
-    },
-    {
-      "epoch": 1.1179596602117639,
-      "grad_norm": 0.0037791484501212835,
-      "learning_rate": 0.00019999938505032102,
-      "loss": 46.0,
-      "step": 6942
-    },
-    {
-      "epoch": 1.118120697290551,
-      "grad_norm": 0.0006212688167579472,
-      "learning_rate": 0.00019999938487288494,
-      "loss": 46.0,
-      "step": 6943
-    },
-    {
-      "epoch": 1.1182817343693385,
-      "grad_norm": 0.0011253413977101445,
-      "learning_rate": 0.0001999993846954233,
-      "loss": 46.0,
-      "step": 6944
-    },
-    {
-      "epoch": 1.118442771448126,
-      "grad_norm": 0.0038602829445153475,
-      "learning_rate": 0.00019999938451793604,
-      "loss": 46.0,
-      "step": 6945
-    },
-    {
-      "epoch": 1.1186038085269132,
-      "grad_norm": 0.0014489177847281098,
-      "learning_rate": 0.00019999938434042317,
-      "loss": 46.0,
-      "step": 6946
-    },
-    {
-      "epoch": 1.1187648456057007,
-      "grad_norm": 0.0029896770138293505,
-      "learning_rate": 0.00019999938416288474,
-      "loss": 46.0,
-      "step": 6947
-    },
-    {
-      "epoch": 1.1189258826844881,
-      "grad_norm": 0.0008623116300441325,
-      "learning_rate": 0.0001999993839853207,
-      "loss": 46.0,
-      "step": 6948
-    },
-    {
-      "epoch": 1.1190869197632756,
-      "grad_norm": 0.0012256938498467207,
-      "learning_rate": 0.00019999938380773106,
-      "loss": 46.0,
-      "step": 6949
-    },
-    {
-      "epoch": 1.1192479568420628,
-      "grad_norm": 0.0012182240607216954,
-      "learning_rate": 0.00019999938363011584,
-      "loss": 46.0,
-      "step": 6950
-    },
-    {
-      "epoch": 1.1194089939208502,
-      "grad_norm": 0.01165693812072277,
-      "learning_rate": 0.000199999383452475,
-      "loss": 46.0,
-      "step": 6951
-    },
-    {
-      "epoch": 1.1195700309996377,
-      "grad_norm": 0.002061246894299984,
-      "learning_rate": 0.0001999993832748086,
-      "loss": 46.0,
-      "step": 6952
-    },
-    {
-      "epoch": 1.1197310680784252,
-      "grad_norm": 0.006831257604062557,
-      "learning_rate": 0.00019999938309711658,
-      "loss": 46.0,
-      "step": 6953
-    },
-    {
-      "epoch": 1.1198921051572124,
-      "grad_norm": 0.0019761386793106794,
-      "learning_rate": 0.00019999938291939898,
-      "loss": 46.0,
-      "step": 6954
-    },
-    {
-      "epoch": 1.1200531422359998,
-      "grad_norm": 0.0024278834462165833,
-      "learning_rate": 0.00019999938274165577,
-      "loss": 46.0,
-      "step": 6955
-    },
-    {
-      "epoch": 1.1202141793147873,
-      "grad_norm": 0.00209654844366014,
-      "learning_rate": 0.00019999938256388698,
-      "loss": 46.0,
-      "step": 6956
-    },
-    {
-      "epoch": 1.1203752163935747,
-      "grad_norm": 0.0024988569784909487,
-      "learning_rate": 0.00019999938238609256,
-      "loss": 46.0,
-      "step": 6957
-    },
-    {
-      "epoch": 1.120536253472362,
-      "grad_norm": 0.0006901687593199313,
-      "learning_rate": 0.0001999993822082726,
-      "loss": 46.0,
-      "step": 6958
-    },
-    {
-      "epoch": 1.1206972905511494,
-      "grad_norm": 0.0005171674420125782,
-      "learning_rate": 0.000199999382030427,
-      "loss": 46.0,
-      "step": 6959
-    },
-    {
-      "epoch": 1.1208583276299369,
-      "grad_norm": 0.0021077613346278667,
-      "learning_rate": 0.00019999938185255584,
-      "loss": 46.0,
-      "step": 6960
-    },
-    {
-      "epoch": 1.121019364708724,
-      "grad_norm": 0.00047611480113118887,
-      "learning_rate": 0.00019999938167465908,
-      "loss": 46.0,
-      "step": 6961
-    },
-    {
-      "epoch": 1.1211804017875115,
-      "grad_norm": 0.0010797953000292182,
-      "learning_rate": 0.0001999993814967367,
-      "loss": 46.0,
-      "step": 6962
-    },
-    {
-      "epoch": 1.121341438866299,
-      "grad_norm": 0.0005051333573646843,
-      "learning_rate": 0.0001999993813187887,
-      "loss": 46.0,
-      "step": 6963
-    },
-    {
-      "epoch": 1.1215024759450865,
-      "grad_norm": 0.0012195693561807275,
-      "learning_rate": 0.0001999993811408152,
-      "loss": 46.0,
-      "step": 6964
-    },
-    {
-      "epoch": 1.1216635130238737,
-      "grad_norm": 0.0005467704613693058,
-      "learning_rate": 0.00019999938096281606,
-      "loss": 46.0,
-      "step": 6965
-    },
-    {
-      "epoch": 1.1218245501026611,
-      "grad_norm": 0.0033836483489722013,
-      "learning_rate": 0.0001999993807847913,
-      "loss": 46.0,
-      "step": 6966
-    },
-    {
-      "epoch": 1.1219855871814486,
-      "grad_norm": 0.001822745194658637,
-      "learning_rate": 0.00019999938060674094,
-      "loss": 46.0,
-      "step": 6967
-    },
-    {
-      "epoch": 1.1221466242602358,
-      "grad_norm": 0.0010443114442750812,
-      "learning_rate": 0.00019999938042866502,
-      "loss": 46.0,
-      "step": 6968
-    },
-    {
-      "epoch": 1.1223076613390233,
-      "grad_norm": 0.0022754084784537554,
-      "learning_rate": 0.00019999938025056348,
-      "loss": 46.0,
-      "step": 6969
-    },
-    {
-      "epoch": 1.1224686984178107,
-      "grad_norm": 0.002585018053650856,
-      "learning_rate": 0.00019999938007243638,
-      "loss": 46.0,
-      "step": 6970
-    },
-    {
-      "epoch": 1.1226297354965982,
-      "grad_norm": 0.002244913950562477,
-      "learning_rate": 0.00019999937989428364,
-      "loss": 46.0,
-      "step": 6971
-    },
-    {
-      "epoch": 1.1227907725753854,
-      "grad_norm": 0.0019119534408673644,
-      "learning_rate": 0.00019999937971610534,
-      "loss": 46.0,
-      "step": 6972
-    },
-    {
-      "epoch": 1.1229518096541729,
-      "grad_norm": 0.0038164029829204082,
-      "learning_rate": 0.00019999937953790143,
-      "loss": 46.0,
-      "step": 6973
-    },
-    {
-      "epoch": 1.1231128467329603,
-      "grad_norm": 0.0006349575123749673,
-      "learning_rate": 0.00019999937935967192,
-      "loss": 46.0,
-      "step": 6974
-    },
-    {
-      "epoch": 1.1232738838117478,
-      "grad_norm": 0.006171821616590023,
-      "learning_rate": 0.00019999937918141684,
-      "loss": 46.0,
-      "step": 6975
-    },
-    {
-      "epoch": 1.123434920890535,
-      "grad_norm": 0.0037995383609086275,
-      "learning_rate": 0.00019999937900313613,
-      "loss": 46.0,
-      "step": 6976
-    },
-    {
-      "epoch": 1.1235959579693224,
-      "grad_norm": 0.0012308359146118164,
-      "learning_rate": 0.00019999937882482987,
-      "loss": 46.0,
-      "step": 6977
-    },
-    {
-      "epoch": 1.1237569950481099,
-      "grad_norm": 0.0010234408546239138,
-      "learning_rate": 0.000199999378646498,
-      "loss": 46.0,
-      "step": 6978
-    },
-    {
-      "epoch": 1.1239180321268971,
-      "grad_norm": 0.0017657250864431262,
-      "learning_rate": 0.0001999993784681405,
-      "loss": 46.0,
-      "step": 6979
-    },
-    {
-      "epoch": 1.1240790692056846,
-      "grad_norm": 0.0010053877485916018,
-      "learning_rate": 0.00019999937828975742,
-      "loss": 46.0,
-      "step": 6980
-    },
-    {
-      "epoch": 1.124240106284472,
-      "grad_norm": 0.0008249666425399482,
-      "learning_rate": 0.00019999937811134876,
-      "loss": 46.0,
-      "step": 6981
-    },
-    {
-      "epoch": 1.1244011433632595,
-      "grad_norm": 0.0029670277144759893,
-      "learning_rate": 0.0001999993779329145,
-      "loss": 46.0,
-      "step": 6982
-    },
-    {
-      "epoch": 1.1245621804420467,
-      "grad_norm": 0.005452977493405342,
-      "learning_rate": 0.00019999937775445466,
-      "loss": 46.0,
-      "step": 6983
-    },
-    {
-      "epoch": 1.1247232175208342,
-      "grad_norm": 0.0006747141596861184,
-      "learning_rate": 0.0001999993775759692,
-      "loss": 46.0,
-      "step": 6984
-    },
-    {
-      "epoch": 1.1248842545996216,
-      "grad_norm": 0.002479668240994215,
-      "learning_rate": 0.00019999937739745817,
-      "loss": 46.0,
-      "step": 6985
-    },
-    {
-      "epoch": 1.125045291678409,
-      "grad_norm": 0.003732510609552264,
-      "learning_rate": 0.0001999993772189215,
-      "loss": 46.0,
-      "step": 6986
-    },
-    {
-      "epoch": 1.1252063287571963,
-      "grad_norm": 0.0010793304536491632,
-      "learning_rate": 0.0001999993770403593,
-      "loss": 46.0,
-      "step": 6987
-    },
-    {
-      "epoch": 1.1253673658359837,
-      "grad_norm": 0.0019333201926201582,
-      "learning_rate": 0.00019999937686177147,
-      "loss": 46.0,
-      "step": 6988
-    },
-    {
-      "epoch": 1.1255284029147712,
-      "grad_norm": 0.0018508994253352284,
-      "learning_rate": 0.00019999937668315805,
-      "loss": 46.0,
-      "step": 6989
-    },
-    {
-      "epoch": 1.1256894399935584,
-      "grad_norm": 0.00103022123221308,
-      "learning_rate": 0.00019999937650451904,
-      "loss": 46.0,
-      "step": 6990
-    },
-    {
-      "epoch": 1.1258504770723459,
-      "grad_norm": 0.004638159181922674,
-      "learning_rate": 0.0001999993763258544,
-      "loss": 46.0,
-      "step": 6991
-    },
-    {
-      "epoch": 1.1260115141511333,
-      "grad_norm": 0.0008935858495533466,
-      "learning_rate": 0.0001999993761471642,
-      "loss": 46.0,
-      "step": 6992
-    },
-    {
-      "epoch": 1.1261725512299208,
-      "grad_norm": 0.0008927615126594901,
-      "learning_rate": 0.0001999993759684484,
-      "loss": 46.0,
-      "step": 6993
-    },
-    {
-      "epoch": 1.126333588308708,
-      "grad_norm": 0.0019390768138691783,
-      "learning_rate": 0.000199999375789707,
-      "loss": 46.0,
-      "step": 6994
-    },
-    {
-      "epoch": 1.1264946253874955,
-      "grad_norm": 0.0005231575923971832,
-      "learning_rate": 0.00019999937561094,
-      "loss": 46.0,
-      "step": 6995
-    },
-    {
-      "epoch": 1.126655662466283,
-      "grad_norm": 0.0029506096616387367,
-      "learning_rate": 0.00019999937543214741,
-      "loss": 46.0,
-      "step": 6996
-    },
-    {
-      "epoch": 1.1268166995450701,
-      "grad_norm": 0.007539860438555479,
-      "learning_rate": 0.00019999937525332924,
-      "loss": 46.0,
-      "step": 6997
-    },
-    {
-      "epoch": 1.1269777366238576,
-      "grad_norm": 0.0010391726391389966,
-      "learning_rate": 0.00019999937507448546,
-      "loss": 46.0,
-      "step": 6998
-    },
-    {
-      "epoch": 1.127138773702645,
-      "grad_norm": 0.0033676750026643276,
-      "learning_rate": 0.00019999937489561612,
-      "loss": 46.0,
-      "step": 6999
-    },
-    {
-      "epoch": 1.1272998107814325,
-      "grad_norm": 0.0012933469843119383,
-      "learning_rate": 0.00019999937471672113,
-      "loss": 46.0,
-      "step": 7000
-    },
-    {
-      "epoch": 1.12746084786022,
-      "grad_norm": 0.0030426557641476393,
-      "learning_rate": 0.0001999993745378006,
-      "loss": 46.0,
-      "step": 7001
-    },
-    {
-      "epoch": 1.1276218849390072,
-      "grad_norm": 0.001418073894456029,
-      "learning_rate": 0.0001999993743588544,
-      "loss": 46.0,
-      "step": 7002
-    },
-    {
-      "epoch": 1.1277829220177946,
-      "grad_norm": 0.0005312525318004191,
-      "learning_rate": 0.00019999937417988268,
-      "loss": 46.0,
-      "step": 7003
-    },
-    {
-      "epoch": 1.127943959096582,
-      "grad_norm": 0.00363905425183475,
-      "learning_rate": 0.00019999937400088532,
-      "loss": 46.0,
-      "step": 7004
-    },
-    {
-      "epoch": 1.1281049961753693,
-      "grad_norm": 0.004000222310423851,
-      "learning_rate": 0.0001999993738218624,
-      "loss": 46.0,
-      "step": 7005
-    },
-    {
-      "epoch": 1.1282660332541568,
-      "grad_norm": 0.0005363834789022803,
-      "learning_rate": 0.00019999937364281384,
-      "loss": 46.0,
-      "step": 7006
-    },
-    {
-      "epoch": 1.1284270703329442,
-      "grad_norm": 0.001692510792054236,
-      "learning_rate": 0.00019999937346373974,
-      "loss": 46.0,
-      "step": 7007
-    },
-    {
-      "epoch": 1.1285881074117317,
-      "grad_norm": 0.0005582827143371105,
-      "learning_rate": 0.00019999937328464,
-      "loss": 46.0,
-      "step": 7008
-    },
-    {
-      "epoch": 1.1287491444905189,
-      "grad_norm": 0.0021540625020861626,
-      "learning_rate": 0.0001999993731055147,
-      "loss": 46.0,
-      "step": 7009
-    },
-    {
-      "epoch": 1.1289101815693063,
-      "grad_norm": 0.00135741604026407,
-      "learning_rate": 0.0001999993729263638,
-      "loss": 46.0,
-      "step": 7010
-    },
-    {
-      "epoch": 1.1290712186480938,
-      "grad_norm": 0.002207091310992837,
-      "learning_rate": 0.00019999937274718728,
-      "loss": 46.0,
-      "step": 7011
-    },
-    {
-      "epoch": 1.129232255726881,
-      "grad_norm": 0.0011262692278251052,
-      "learning_rate": 0.00019999937256798516,
-      "loss": 46.0,
-      "step": 7012
-    },
-    {
-      "epoch": 1.1293932928056685,
-      "grad_norm": 0.004497850779443979,
-      "learning_rate": 0.00019999937238875747,
-      "loss": 46.0,
-      "step": 7013
-    },
-    {
-      "epoch": 1.129554329884456,
-      "grad_norm": 0.0009169720578938723,
-      "learning_rate": 0.0001999993722095042,
-      "loss": 46.0,
-      "step": 7014
-    },
-    {
-      "epoch": 1.1297153669632434,
-      "grad_norm": 0.0029274439439177513,
-      "learning_rate": 0.0001999993720302253,
-      "loss": 46.0,
-      "step": 7015
-    },
-    {
-      "epoch": 1.1298764040420306,
-      "grad_norm": 0.0007946903351694345,
-      "learning_rate": 0.00019999937185092082,
-      "loss": 46.0,
-      "step": 7016
-    },
-    {
-      "epoch": 1.130037441120818,
-      "grad_norm": 0.002871324075385928,
-      "learning_rate": 0.00019999937167159074,
-      "loss": 46.0,
-      "step": 7017
-    },
-    {
-      "epoch": 1.1301984781996055,
-      "grad_norm": 0.0010288202902302146,
-      "learning_rate": 0.00019999937149223505,
-      "loss": 46.0,
-      "step": 7018
-    },
-    {
-      "epoch": 1.1303595152783927,
-      "grad_norm": 0.008500954136252403,
-      "learning_rate": 0.00019999937131285383,
-      "loss": 46.0,
-      "step": 7019
-    },
-    {
-      "epoch": 1.1305205523571802,
-      "grad_norm": 0.006231991108506918,
-      "learning_rate": 0.00019999937113344697,
-      "loss": 46.0,
-      "step": 7020
-    },
-    {
-      "epoch": 1.1306815894359676,
-      "grad_norm": 0.0017418725183233619,
-      "learning_rate": 0.00019999937095401452,
-      "loss": 46.0,
-      "step": 7021
-    },
-    {
-      "epoch": 1.130842626514755,
-      "grad_norm": 0.0019019623287022114,
-      "learning_rate": 0.00019999937077455645,
-      "loss": 46.0,
-      "step": 7022
-    },
-    {
-      "epoch": 1.1310036635935423,
-      "grad_norm": 0.009048969484865665,
-      "learning_rate": 0.0001999993705950728,
-      "loss": 46.0,
-      "step": 7023
-    },
-    {
-      "epoch": 1.1311647006723298,
-      "grad_norm": 0.000814581464510411,
-      "learning_rate": 0.00019999937041556359,
-      "loss": 46.0,
-      "step": 7024
-    },
-    {
-      "epoch": 1.1313257377511172,
-      "grad_norm": 0.0012518613366410136,
-      "learning_rate": 0.00019999937023602876,
-      "loss": 46.0,
-      "step": 7025
-    },
-    {
-      "epoch": 1.1314867748299047,
-      "grad_norm": 0.0014614618849009275,
-      "learning_rate": 0.00019999937005646832,
-      "loss": 46.0,
-      "step": 7026
-    },
-    {
-      "epoch": 1.131647811908692,
-      "grad_norm": 0.007246515713632107,
-      "learning_rate": 0.0001999993698768823,
-      "loss": 46.0,
-      "step": 7027
-    },
-    {
-      "epoch": 1.1318088489874794,
-      "grad_norm": 0.001126777264289558,
-      "learning_rate": 0.00019999936969727067,
-      "loss": 46.0,
-      "step": 7028
-    },
-    {
-      "epoch": 1.1319698860662668,
-      "grad_norm": 0.0008017787477001548,
-      "learning_rate": 0.00019999936951763347,
-      "loss": 46.0,
-      "step": 7029
-    },
-    {
-      "epoch": 1.1321309231450543,
-      "grad_norm": 0.0003722316469065845,
-      "learning_rate": 0.00019999936933797066,
-      "loss": 46.0,
-      "step": 7030
-    },
-    {
-      "epoch": 1.1322919602238415,
-      "grad_norm": 0.0008829387370496988,
-      "learning_rate": 0.00019999936915828225,
-      "loss": 46.0,
-      "step": 7031
-    },
-    {
-      "epoch": 1.132452997302629,
-      "grad_norm": 0.000955459545366466,
-      "learning_rate": 0.0001999993689785683,
-      "loss": 46.0,
-      "step": 7032
-    },
-    {
-      "epoch": 1.1326140343814164,
-      "grad_norm": 0.0025489337276667356,
-      "learning_rate": 0.00019999936879882868,
-      "loss": 46.0,
-      "step": 7033
-    },
-    {
-      "epoch": 1.1327750714602036,
-      "grad_norm": 0.0011475490173324943,
-      "learning_rate": 0.0001999993686190635,
-      "loss": 46.0,
-      "step": 7034
-    },
-    {
-      "epoch": 1.132936108538991,
-      "grad_norm": 0.009986408054828644,
-      "learning_rate": 0.0001999993684392727,
-      "loss": 46.0,
-      "step": 7035
-    },
-    {
-      "epoch": 1.1330971456177785,
-      "grad_norm": 0.0021582748740911484,
-      "learning_rate": 0.00019999936825945634,
-      "loss": 46.0,
-      "step": 7036
-    },
-    {
-      "epoch": 1.133258182696566,
-      "grad_norm": 0.0006016548722982407,
-      "learning_rate": 0.00019999936807961436,
-      "loss": 46.0,
-      "step": 7037
-    },
-    {
-      "epoch": 1.1334192197753532,
-      "grad_norm": 0.0005165237234905362,
-      "learning_rate": 0.0001999993678997468,
-      "loss": 46.0,
-      "step": 7038
-    },
-    {
-      "epoch": 1.1335802568541407,
-      "grad_norm": 0.0011950712651014328,
-      "learning_rate": 0.00019999936771985364,
-      "loss": 46.0,
-      "step": 7039
-    },
-    {
-      "epoch": 1.133741293932928,
-      "grad_norm": 0.005373006220906973,
-      "learning_rate": 0.0001999993675399349,
-      "loss": 46.0,
-      "step": 7040
-    },
-    {
-      "epoch": 1.1339023310117153,
-      "grad_norm": 0.0016636315267533064,
-      "learning_rate": 0.00019999936735999057,
-      "loss": 46.0,
-      "step": 7041
-    },
-    {
-      "epoch": 1.1340633680905028,
-      "grad_norm": 0.002621646737679839,
-      "learning_rate": 0.00019999936718002062,
-      "loss": 46.0,
-      "step": 7042
-    },
-    {
-      "epoch": 1.1342244051692902,
-      "grad_norm": 0.0009070023079402745,
-      "learning_rate": 0.0001999993670000251,
-      "loss": 46.0,
-      "step": 7043
-    },
-    {
-      "epoch": 1.1343854422480777,
-      "grad_norm": 0.0005544809391722083,
-      "learning_rate": 0.00019999936682000392,
-      "loss": 46.0,
-      "step": 7044
-    },
-    {
-      "epoch": 1.134546479326865,
-      "grad_norm": 0.000758998969104141,
-      "learning_rate": 0.0001999993666399572,
-      "loss": 46.0,
-      "step": 7045
-    },
-    {
-      "epoch": 1.1347075164056524,
-      "grad_norm": 0.0022243887651711702,
-      "learning_rate": 0.0001999993664598849,
-      "loss": 46.0,
-      "step": 7046
-    },
-    {
-      "epoch": 1.1348685534844398,
-      "grad_norm": 0.0008686042856425047,
-      "learning_rate": 0.00019999936627978699,
-      "loss": 46.0,
-      "step": 7047
-    },
-    {
-      "epoch": 1.1350295905632273,
-      "grad_norm": 0.0031492654234170914,
-      "learning_rate": 0.00019999936609966346,
-      "loss": 46.0,
-      "step": 7048
-    },
-    {
-      "epoch": 1.1351906276420145,
-      "grad_norm": 0.0041769687086343765,
-      "learning_rate": 0.00019999936591951436,
-      "loss": 46.0,
-      "step": 7049
-    },
-    {
-      "epoch": 1.135351664720802,
-      "grad_norm": 0.0015751132741570473,
-      "learning_rate": 0.00019999936573933966,
-      "loss": 46.0,
-      "step": 7050
-    },
-    {
-      "epoch": 1.1355127017995894,
-      "grad_norm": 0.0004279755230527371,
-      "learning_rate": 0.00019999936555913935,
-      "loss": 46.0,
-      "step": 7051
-    },
-    {
-      "epoch": 1.1356737388783769,
-      "grad_norm": 0.002107194159179926,
-      "learning_rate": 0.00019999936537891348,
-      "loss": 46.0,
-      "step": 7052
-    },
-    {
-      "epoch": 1.135834775957164,
-      "grad_norm": 0.0016004648059606552,
-      "learning_rate": 0.00019999936519866197,
-      "loss": 46.0,
-      "step": 7053
-    },
-    {
-      "epoch": 1.1359958130359515,
-      "grad_norm": 0.0026183025911450386,
-      "learning_rate": 0.00019999936501838492,
-      "loss": 46.0,
-      "step": 7054
-    },
-    {
-      "epoch": 1.136156850114739,
-      "grad_norm": 0.0004843698407057673,
-      "learning_rate": 0.00019999936483808224,
-      "loss": 46.0,
-      "step": 7055
-    },
-    {
-      "epoch": 1.1363178871935262,
-      "grad_norm": 0.0010027765529230237,
-      "learning_rate": 0.00019999936465775397,
-      "loss": 46.0,
-      "step": 7056
-    },
-    {
-      "epoch": 1.1364789242723137,
-      "grad_norm": 0.0016240050317719579,
-      "learning_rate": 0.0001999993644774001,
-      "loss": 46.0,
-      "step": 7057
-    },
-    {
-      "epoch": 1.1366399613511011,
-      "grad_norm": 0.0015592457493767142,
-      "learning_rate": 0.00019999936429702063,
-      "loss": 46.0,
-      "step": 7058
-    },
-    {
-      "epoch": 1.1368009984298886,
-      "grad_norm": 0.0026613944210112095,
-      "learning_rate": 0.00019999936411661562,
-      "loss": 46.0,
-      "step": 7059
-    },
-    {
-      "epoch": 1.1369620355086758,
-      "grad_norm": 0.0009260620572604239,
-      "learning_rate": 0.00019999936393618497,
-      "loss": 46.0,
-      "step": 7060
-    },
-    {
-      "epoch": 1.1371230725874633,
-      "grad_norm": 0.005152855534106493,
-      "learning_rate": 0.0001999993637557287,
-      "loss": 46.0,
-      "step": 7061
-    },
-    {
-      "epoch": 1.1372841096662507,
-      "grad_norm": 0.0038004093803465366,
-      "learning_rate": 0.0001999993635752469,
-      "loss": 46.0,
-      "step": 7062
-    },
-    {
-      "epoch": 1.137445146745038,
-      "grad_norm": 0.0005911357002332807,
-      "learning_rate": 0.00019999936339473945,
-      "loss": 46.0,
-      "step": 7063
-    },
-    {
-      "epoch": 1.1376061838238254,
-      "grad_norm": 0.0011656349524855614,
-      "learning_rate": 0.00019999936321420643,
-      "loss": 46.0,
-      "step": 7064
-    },
-    {
-      "epoch": 1.1377672209026128,
-      "grad_norm": 0.002185005694627762,
-      "learning_rate": 0.00019999936303364781,
-      "loss": 46.0,
-      "step": 7065
-    },
-    {
-      "epoch": 1.1379282579814003,
-      "grad_norm": 0.0018050329526886344,
-      "learning_rate": 0.0001999993628530636,
-      "loss": 46.0,
-      "step": 7066
-    },
-    {
-      "epoch": 1.1380892950601875,
-      "grad_norm": 0.0029668938368558884,
-      "learning_rate": 0.00019999936267245378,
-      "loss": 46.0,
-      "step": 7067
-    },
-    {
-      "epoch": 1.138250332138975,
-      "grad_norm": 0.0016116938786581159,
-      "learning_rate": 0.00019999936249181838,
-      "loss": 46.0,
-      "step": 7068
-    },
-    {
-      "epoch": 1.1384113692177624,
-      "grad_norm": 0.001741871703416109,
-      "learning_rate": 0.0001999993623111574,
-      "loss": 46.0,
-      "step": 7069
-    },
-    {
-      "epoch": 1.1385724062965499,
-      "grad_norm": 0.0011630411026999354,
-      "learning_rate": 0.00019999936213047076,
-      "loss": 46.0,
-      "step": 7070
-    },
-    {
-      "epoch": 1.138733443375337,
-      "grad_norm": 0.0006004149909131229,
-      "learning_rate": 0.0001999993619497586,
-      "loss": 46.0,
-      "step": 7071
-    },
-    {
-      "epoch": 1.1388944804541246,
-      "grad_norm": 0.0016764452448114753,
-      "learning_rate": 0.00019999936176902082,
-      "loss": 46.0,
-      "step": 7072
-    },
-    {
-      "epoch": 1.139055517532912,
-      "grad_norm": 0.0013854493154212832,
-      "learning_rate": 0.00019999936158825743,
-      "loss": 46.0,
-      "step": 7073
-    },
-    {
-      "epoch": 1.1392165546116995,
-      "grad_norm": 0.0014358782209455967,
-      "learning_rate": 0.00019999936140746845,
-      "loss": 46.0,
-      "step": 7074
-    },
-    {
-      "epoch": 1.1393775916904867,
-      "grad_norm": 0.005022861994802952,
-      "learning_rate": 0.0001999993612266539,
-      "loss": 46.0,
-      "step": 7075
-    },
-    {
-      "epoch": 1.1395386287692741,
-      "grad_norm": 0.000553262943867594,
-      "learning_rate": 0.00019999936104581374,
-      "loss": 46.0,
-      "step": 7076
-    },
-    {
-      "epoch": 1.1396996658480616,
-      "grad_norm": 0.005023543257266283,
-      "learning_rate": 0.00019999936086494797,
-      "loss": 46.0,
-      "step": 7077
-    },
-    {
-      "epoch": 1.1398607029268488,
-      "grad_norm": 0.0012056485284119844,
-      "learning_rate": 0.00019999936068405662,
-      "loss": 46.0,
-      "step": 7078
-    },
-    {
-      "epoch": 1.1400217400056363,
-      "grad_norm": 0.0006060278392396867,
-      "learning_rate": 0.00019999936050313968,
-      "loss": 46.0,
-      "step": 7079
-    },
-    {
-      "epoch": 1.1401827770844237,
-      "grad_norm": 0.0006432574591599405,
-      "learning_rate": 0.00019999936032219712,
-      "loss": 46.0,
-      "step": 7080
-    },
-    {
-      "epoch": 1.1403438141632112,
-      "grad_norm": 0.0017074752831831574,
-      "learning_rate": 0.000199999360141229,
-      "loss": 46.0,
-      "step": 7081
-    },
-    {
-      "epoch": 1.1405048512419984,
-      "grad_norm": 0.002766041085124016,
-      "learning_rate": 0.00019999935996023528,
-      "loss": 46.0,
-      "step": 7082
-    },
-    {
-      "epoch": 1.1406658883207859,
-      "grad_norm": 0.0030352489557117224,
-      "learning_rate": 0.00019999935977921593,
-      "loss": 46.0,
-      "step": 7083
-    },
-    {
-      "epoch": 1.1408269253995733,
-      "grad_norm": 0.0029198937118053436,
-      "learning_rate": 0.00019999935959817103,
-      "loss": 46.0,
-      "step": 7084
-    },
-    {
-      "epoch": 1.1409879624783605,
-      "grad_norm": 0.003776973346248269,
-      "learning_rate": 0.0001999993594171005,
-      "loss": 46.0,
-      "step": 7085
-    },
-    {
-      "epoch": 1.141148999557148,
-      "grad_norm": 0.003045264631509781,
-      "learning_rate": 0.0001999993592360044,
-      "loss": 46.0,
-      "step": 7086
-    },
-    {
-      "epoch": 1.1413100366359354,
-      "grad_norm": 0.00043938314775004983,
-      "learning_rate": 0.0001999993590548827,
-      "loss": 46.0,
-      "step": 7087
-    },
-    {
-      "epoch": 1.141471073714723,
-      "grad_norm": 0.0018584945937618613,
-      "learning_rate": 0.0001999993588737354,
-      "loss": 46.0,
-      "step": 7088
-    },
-    {
-      "epoch": 1.1416321107935101,
-      "grad_norm": 0.002093725837767124,
-      "learning_rate": 0.0001999993586925625,
-      "loss": 46.0,
-      "step": 7089
-    },
-    {
-      "epoch": 1.1417931478722976,
-      "grad_norm": 0.0020268792286515236,
-      "learning_rate": 0.000199999358511364,
-      "loss": 46.0,
-      "step": 7090
-    },
-    {
-      "epoch": 1.141954184951085,
-      "grad_norm": 0.0018076930427923799,
-      "learning_rate": 0.00019999935833013991,
-      "loss": 46.0,
-      "step": 7091
-    },
-    {
-      "epoch": 1.1421152220298723,
-      "grad_norm": 0.005321699660271406,
-      "learning_rate": 0.00019999935814889026,
-      "loss": 46.0,
-      "step": 7092
-    },
-    {
-      "epoch": 1.1422762591086597,
-      "grad_norm": 0.0018107375362887979,
-      "learning_rate": 0.00019999935796761496,
-      "loss": 46.0,
-      "step": 7093
-    },
-    {
-      "epoch": 1.1424372961874472,
-      "grad_norm": 0.0012942444300279021,
-      "learning_rate": 0.0001999993577863141,
-      "loss": 46.0,
-      "step": 7094
-    },
-    {
-      "epoch": 1.1425983332662346,
-      "grad_norm": 0.0019107672851532698,
-      "learning_rate": 0.00019999935760498766,
-      "loss": 46.0,
-      "step": 7095
-    },
-    {
-      "epoch": 1.142759370345022,
-      "grad_norm": 0.003010336309671402,
-      "learning_rate": 0.00019999935742363558,
-      "loss": 46.0,
-      "step": 7096
-    },
-    {
-      "epoch": 1.1429204074238093,
-      "grad_norm": 0.0010296377586200833,
-      "learning_rate": 0.00019999935724225796,
-      "loss": 46.0,
-      "step": 7097
-    },
-    {
-      "epoch": 1.1430814445025967,
-      "grad_norm": 0.0009530838578939438,
-      "learning_rate": 0.0001999993570608547,
-      "loss": 46.0,
-      "step": 7098
-    },
-    {
-      "epoch": 1.1432424815813842,
-      "grad_norm": 0.0004894350422546268,
-      "learning_rate": 0.00019999935687942585,
-      "loss": 46.0,
-      "step": 7099
-    },
-    {
-      "epoch": 1.1434035186601714,
-      "grad_norm": 0.0013363274047151208,
-      "learning_rate": 0.00019999935669797142,
-      "loss": 46.0,
-      "step": 7100
-    },
-    {
-      "epoch": 1.1435645557389589,
-      "grad_norm": 0.005797158926725388,
-      "learning_rate": 0.0001999993565164914,
-      "loss": 46.0,
-      "step": 7101
-    },
-    {
-      "epoch": 1.1437255928177463,
-      "grad_norm": 0.0009479834116064012,
-      "learning_rate": 0.00019999935633498576,
-      "loss": 46.0,
-      "step": 7102
-    },
-    {
-      "epoch": 1.1438866298965338,
-      "grad_norm": 0.0013979620998725295,
-      "learning_rate": 0.00019999935615345454,
-      "loss": 46.0,
-      "step": 7103
-    },
-    {
-      "epoch": 1.144047666975321,
-      "grad_norm": 0.0022883708588778973,
-      "learning_rate": 0.00019999935597189773,
-      "loss": 46.0,
-      "step": 7104
-    },
-    {
-      "epoch": 1.1442087040541085,
-      "grad_norm": 0.0005104238516651094,
-      "learning_rate": 0.00019999935579031533,
-      "loss": 46.0,
-      "step": 7105
-    },
-    {
-      "epoch": 1.144369741132896,
-      "grad_norm": 0.003547649597749114,
-      "learning_rate": 0.00019999935560870732,
-      "loss": 46.0,
-      "step": 7106
-    },
-    {
-      "epoch": 1.1445307782116831,
-      "grad_norm": 0.002930952003225684,
-      "learning_rate": 0.00019999935542707373,
-      "loss": 46.0,
-      "step": 7107
-    },
-    {
-      "epoch": 1.1446918152904706,
-      "grad_norm": 0.0011777014005929232,
-      "learning_rate": 0.00019999935524541454,
-      "loss": 46.0,
-      "step": 7108
-    },
-    {
-      "epoch": 1.144852852369258,
-      "grad_norm": 0.0031634350307285786,
-      "learning_rate": 0.00019999935506372974,
-      "loss": 46.0,
-      "step": 7109
-    },
-    {
-      "epoch": 1.1450138894480455,
-      "grad_norm": 0.0040549724362790585,
-      "learning_rate": 0.00019999935488201938,
-      "loss": 46.0,
-      "step": 7110
-    },
-    {
-      "epoch": 1.1451749265268327,
-      "grad_norm": 0.0045827715657651424,
-      "learning_rate": 0.00019999935470028338,
-      "loss": 46.0,
-      "step": 7111
-    },
-    {
-      "epoch": 1.1453359636056202,
-      "grad_norm": 0.003957042470574379,
-      "learning_rate": 0.00019999935451852182,
-      "loss": 46.0,
-      "step": 7112
-    },
-    {
-      "epoch": 1.1454970006844076,
-      "grad_norm": 0.0020988809410482645,
-      "learning_rate": 0.00019999935433673467,
-      "loss": 46.0,
-      "step": 7113
-    },
-    {
-      "epoch": 1.1456580377631949,
-      "grad_norm": 0.0010748854838311672,
-      "learning_rate": 0.00019999935415492188,
-      "loss": 46.0,
-      "step": 7114
-    },
-    {
-      "epoch": 1.1458190748419823,
-      "grad_norm": 0.002551599871367216,
-      "learning_rate": 0.00019999935397308353,
-      "loss": 46.0,
-      "step": 7115
-    },
-    {
-      "epoch": 1.1459801119207698,
-      "grad_norm": 0.0006531869294121861,
-      "learning_rate": 0.00019999935379121957,
-      "loss": 46.0,
-      "step": 7116
-    },
-    {
-      "epoch": 1.1461411489995572,
-      "grad_norm": 0.0010318257845938206,
-      "learning_rate": 0.00019999935360933004,
-      "loss": 46.0,
-      "step": 7117
-    },
-    {
-      "epoch": 1.1463021860783444,
-      "grad_norm": 0.003948194440454245,
-      "learning_rate": 0.0001999993534274149,
-      "loss": 46.0,
-      "step": 7118
-    },
-    {
-      "epoch": 1.146463223157132,
-      "grad_norm": 0.002598825842142105,
-      "learning_rate": 0.00019999935324547415,
-      "loss": 46.0,
-      "step": 7119
-    },
-    {
-      "epoch": 1.1466242602359193,
-      "grad_norm": 0.0009227597038261592,
-      "learning_rate": 0.00019999935306350784,
-      "loss": 46.0,
-      "step": 7120
-    },
-    {
-      "epoch": 1.1467852973147068,
-      "grad_norm": 0.0015254870522767305,
-      "learning_rate": 0.0001999993528815159,
-      "loss": 46.0,
-      "step": 7121
-    },
-    {
-      "epoch": 1.146946334393494,
-      "grad_norm": 0.002448051003739238,
-      "learning_rate": 0.0001999993526994984,
-      "loss": 46.0,
-      "step": 7122
-    },
-    {
-      "epoch": 1.1471073714722815,
-      "grad_norm": 0.0005201072199270129,
-      "learning_rate": 0.00019999935251745527,
-      "loss": 46.0,
-      "step": 7123
-    },
-    {
-      "epoch": 1.147268408551069,
-      "grad_norm": 0.0027441829442977905,
-      "learning_rate": 0.00019999935233538655,
-      "loss": 46.0,
-      "step": 7124
-    },
-    {
-      "epoch": 1.1474294456298564,
-      "grad_norm": 0.0030863042920827866,
-      "learning_rate": 0.00019999935215329225,
-      "loss": 46.0,
-      "step": 7125
-    },
-    {
-      "epoch": 1.1475904827086436,
-      "grad_norm": 0.0007674030493944883,
-      "learning_rate": 0.0001999993519711724,
-      "loss": 46.0,
-      "step": 7126
-    },
-    {
-      "epoch": 1.147751519787431,
-      "grad_norm": 0.0026222362648695707,
-      "learning_rate": 0.00019999935178902688,
-      "loss": 46.0,
-      "step": 7127
-    },
-    {
-      "epoch": 1.1479125568662185,
-      "grad_norm": 0.003919719718396664,
-      "learning_rate": 0.0001999993516068558,
-      "loss": 46.0,
-      "step": 7128
-    },
-    {
-      "epoch": 1.1480735939450057,
-      "grad_norm": 0.00328479940071702,
-      "learning_rate": 0.00019999935142465912,
-      "loss": 46.0,
-      "step": 7129
-    },
-    {
-      "epoch": 1.1482346310237932,
-      "grad_norm": 0.0028343668673187494,
-      "learning_rate": 0.00019999935124243682,
-      "loss": 46.0,
-      "step": 7130
-    },
-    {
-      "epoch": 1.1483956681025806,
-      "grad_norm": 0.0006880885921418667,
-      "learning_rate": 0.00019999935106018897,
-      "loss": 46.0,
-      "step": 7131
-    },
-    {
-      "epoch": 1.148556705181368,
-      "grad_norm": 0.005528413224965334,
-      "learning_rate": 0.0001999993508779155,
-      "loss": 46.0,
-      "step": 7132
-    },
-    {
-      "epoch": 1.1487177422601553,
-      "grad_norm": 0.0032945945858955383,
-      "learning_rate": 0.00019999935069561645,
-      "loss": 46.0,
-      "step": 7133
-    },
-    {
-      "epoch": 1.1488787793389428,
-      "grad_norm": 0.00171490793582052,
-      "learning_rate": 0.00019999935051329178,
-      "loss": 46.0,
-      "step": 7134
-    },
-    {
-      "epoch": 1.1490398164177302,
-      "grad_norm": 0.001397947664372623,
-      "learning_rate": 0.00019999935033094155,
-      "loss": 46.0,
-      "step": 7135
-    },
-    {
-      "epoch": 1.1492008534965175,
-      "grad_norm": 0.0019568291027098894,
-      "learning_rate": 0.0001999993501485657,
-      "loss": 46.0,
-      "step": 7136
-    },
-    {
-      "epoch": 1.149361890575305,
-      "grad_norm": 0.0018681793007999659,
-      "learning_rate": 0.00019999934996616425,
-      "loss": 46.0,
-      "step": 7137
-    },
-    {
-      "epoch": 1.1495229276540924,
-      "grad_norm": 0.002650015288963914,
-      "learning_rate": 0.00019999934978373723,
-      "loss": 46.0,
-      "step": 7138
-    },
-    {
-      "epoch": 1.1496839647328798,
-      "grad_norm": 0.0010102004744112492,
-      "learning_rate": 0.0001999993496012846,
-      "loss": 46.0,
-      "step": 7139
-    },
-    {
-      "epoch": 1.149845001811667,
-      "grad_norm": 0.002710662316530943,
-      "learning_rate": 0.00019999934941880638,
-      "loss": 46.0,
-      "step": 7140
-    },
-    {
-      "epoch": 1.1500060388904545,
-      "grad_norm": 0.0019491378916427493,
-      "learning_rate": 0.00019999934923630255,
-      "loss": 46.0,
-      "step": 7141
-    },
-    {
-      "epoch": 1.150167075969242,
-      "grad_norm": 0.0015951680252328515,
-      "learning_rate": 0.00019999934905377313,
-      "loss": 46.0,
-      "step": 7142
-    },
-    {
-      "epoch": 1.1503281130480294,
-      "grad_norm": 0.0041846781969070435,
-      "learning_rate": 0.00019999934887121815,
-      "loss": 46.0,
-      "step": 7143
-    },
-    {
-      "epoch": 1.1504891501268166,
-      "grad_norm": 0.0010823012562468648,
-      "learning_rate": 0.00019999934868863753,
-      "loss": 46.0,
-      "step": 7144
-    },
-    {
-      "epoch": 1.150650187205604,
-      "grad_norm": 0.001921929302625358,
-      "learning_rate": 0.00019999934850603132,
-      "loss": 46.0,
-      "step": 7145
-    },
-    {
-      "epoch": 1.1508112242843915,
-      "grad_norm": 0.0019734189845621586,
-      "learning_rate": 0.00019999934832339955,
-      "loss": 46.0,
-      "step": 7146
-    },
-    {
-      "epoch": 1.150972261363179,
-      "grad_norm": 0.004689333029091358,
-      "learning_rate": 0.00019999934814074214,
-      "loss": 46.0,
-      "step": 7147
-    },
-    {
-      "epoch": 1.1511332984419662,
-      "grad_norm": 0.006085149943828583,
-      "learning_rate": 0.00019999934795805917,
-      "loss": 46.0,
-      "step": 7148
-    },
-    {
-      "epoch": 1.1512943355207537,
-      "grad_norm": 0.0030720606446266174,
-      "learning_rate": 0.0001999993477753506,
-      "loss": 46.0,
-      "step": 7149
-    },
-    {
-      "epoch": 1.1514553725995411,
-      "grad_norm": 0.0005983941373415291,
-      "learning_rate": 0.0001999993475926164,
-      "loss": 46.0,
-      "step": 7150
-    },
-    {
-      "epoch": 1.1516164096783283,
-      "grad_norm": 0.0017330769915133715,
-      "learning_rate": 0.00019999934740985665,
-      "loss": 46.0,
-      "step": 7151
-    },
-    {
-      "epoch": 1.1517774467571158,
-      "grad_norm": 0.0025837167631834745,
-      "learning_rate": 0.0001999993472270713,
-      "loss": 46.0,
-      "step": 7152
-    },
-    {
-      "epoch": 1.1519384838359032,
-      "grad_norm": 0.0020570303313434124,
-      "learning_rate": 0.00019999934704426035,
-      "loss": 46.0,
-      "step": 7153
-    },
-    {
-      "epoch": 1.1520995209146907,
-      "grad_norm": 0.004520559683442116,
-      "learning_rate": 0.0001999993468614238,
-      "loss": 46.0,
-      "step": 7154
-    },
-    {
-      "epoch": 1.152260557993478,
-      "grad_norm": 0.0005783848464488983,
-      "learning_rate": 0.00019999934667856164,
-      "loss": 46.0,
-      "step": 7155
-    },
-    {
-      "epoch": 1.1524215950722654,
-      "grad_norm": 0.001902230316773057,
-      "learning_rate": 0.00019999934649567392,
-      "loss": 46.0,
-      "step": 7156
-    },
-    {
-      "epoch": 1.1525826321510528,
-      "grad_norm": 0.0016362016322091222,
-      "learning_rate": 0.00019999934631276058,
-      "loss": 46.0,
-      "step": 7157
-    },
-    {
-      "epoch": 1.15274366922984,
-      "grad_norm": 0.0024809706956148148,
-      "learning_rate": 0.00019999934612982166,
-      "loss": 46.0,
-      "step": 7158
-    },
-    {
-      "epoch": 1.1529047063086275,
-      "grad_norm": 0.0056978934444487095,
-      "learning_rate": 0.00019999934594685712,
-      "loss": 46.0,
-      "step": 7159
-    },
-    {
-      "epoch": 1.153065743387415,
-      "grad_norm": 0.0004316475533414632,
-      "learning_rate": 0.000199999345763867,
-      "loss": 46.0,
-      "step": 7160
-    },
-    {
-      "epoch": 1.1532267804662024,
-      "grad_norm": 0.001528278342448175,
-      "learning_rate": 0.0001999993455808513,
-      "loss": 46.0,
-      "step": 7161
-    },
-    {
-      "epoch": 1.1533878175449896,
-      "grad_norm": 0.0010144361294806004,
-      "learning_rate": 0.00019999934539781,
-      "loss": 46.0,
-      "step": 7162
-    },
-    {
-      "epoch": 1.153548854623777,
-      "grad_norm": 0.001706136972643435,
-      "learning_rate": 0.0001999993452147431,
-      "loss": 46.0,
-      "step": 7163
-    },
-    {
-      "epoch": 1.1537098917025645,
-      "grad_norm": 0.0011096884263679385,
-      "learning_rate": 0.0001999993450316506,
-      "loss": 46.0,
-      "step": 7164
-    },
-    {
-      "epoch": 1.153870928781352,
-      "grad_norm": 0.008691283874213696,
-      "learning_rate": 0.0001999993448485325,
-      "loss": 46.0,
-      "step": 7165
-    },
-    {
-      "epoch": 1.1540319658601392,
-      "grad_norm": 0.0010033799335360527,
-      "learning_rate": 0.0001999993446653888,
-      "loss": 46.0,
-      "step": 7166
-    },
-    {
-      "epoch": 1.1541930029389267,
-      "grad_norm": 0.0005838305805809796,
-      "learning_rate": 0.00019999934448221954,
-      "loss": 46.0,
-      "step": 7167
-    },
-    {
-      "epoch": 1.1543540400177141,
-      "grad_norm": 0.001546303159557283,
-      "learning_rate": 0.00019999934429902467,
-      "loss": 46.0,
-      "step": 7168
-    },
-    {
-      "epoch": 1.1545150770965016,
-      "grad_norm": 0.0004625291330739856,
-      "learning_rate": 0.0001999993441158042,
-      "loss": 46.0,
-      "step": 7169
-    },
-    {
-      "epoch": 1.1546761141752888,
-      "grad_norm": 0.0023863117676228285,
-      "learning_rate": 0.00019999934393255813,
-      "loss": 46.0,
-      "step": 7170
-    },
-    {
-      "epoch": 1.1548371512540763,
-      "grad_norm": 0.0021304620895534754,
-      "learning_rate": 0.00019999934374928646,
-      "loss": 46.0,
-      "step": 7171
-    },
-    {
-      "epoch": 1.1549981883328637,
-      "grad_norm": 0.0034188157878816128,
-      "learning_rate": 0.0001999993435659892,
-      "loss": 46.0,
-      "step": 7172
-    },
-    {
-      "epoch": 1.155159225411651,
-      "grad_norm": 0.002695410279557109,
-      "learning_rate": 0.00019999934338266637,
-      "loss": 46.0,
-      "step": 7173
-    },
-    {
-      "epoch": 1.1553202624904384,
-      "grad_norm": 0.004395594354718924,
-      "learning_rate": 0.00019999934319931794,
-      "loss": 46.0,
-      "step": 7174
-    },
-    {
-      "epoch": 1.1554812995692259,
-      "grad_norm": 0.0012124684872105718,
-      "learning_rate": 0.0001999993430159439,
-      "loss": 46.0,
-      "step": 7175
-    },
-    {
-      "epoch": 1.1556423366480133,
-      "grad_norm": 0.005230535753071308,
-      "learning_rate": 0.00019999934283254428,
-      "loss": 46.0,
-      "step": 7176
-    },
-    {
-      "epoch": 1.1558033737268005,
-      "grad_norm": 0.0009849924826994538,
-      "learning_rate": 0.00019999934264911903,
-      "loss": 46.0,
-      "step": 7177
-    },
-    {
-      "epoch": 1.155964410805588,
-      "grad_norm": 0.0028686292935162783,
-      "learning_rate": 0.00019999934246566823,
-      "loss": 46.0,
-      "step": 7178
-    },
-    {
-      "epoch": 1.1561254478843754,
-      "grad_norm": 0.0017365794628858566,
-      "learning_rate": 0.0001999993422821918,
-      "loss": 46.0,
-      "step": 7179
-    },
-    {
-      "epoch": 1.1562864849631627,
-      "grad_norm": 0.0004623793647624552,
-      "learning_rate": 0.00019999934209868978,
-      "loss": 46.0,
-      "step": 7180
-    },
-    {
-      "epoch": 1.1564475220419501,
-      "grad_norm": 0.0009670014842413366,
-      "learning_rate": 0.0001999993419151622,
-      "loss": 46.0,
-      "step": 7181
-    },
-    {
-      "epoch": 1.1566085591207376,
-      "grad_norm": 0.0014378369087353349,
-      "learning_rate": 0.00019999934173160896,
-      "loss": 46.0,
-      "step": 7182
-    },
-    {
-      "epoch": 1.156769596199525,
-      "grad_norm": 0.0014132558135315776,
-      "learning_rate": 0.00019999934154803017,
-      "loss": 46.0,
-      "step": 7183
-    },
-    {
-      "epoch": 1.1569306332783122,
-      "grad_norm": 0.002175344852730632,
-      "learning_rate": 0.0001999993413644258,
-      "loss": 46.0,
-      "step": 7184
-    },
-    {
-      "epoch": 1.1570916703570997,
-      "grad_norm": 0.0045136939734220505,
-      "learning_rate": 0.0001999993411807958,
-      "loss": 46.0,
-      "step": 7185
-    },
-    {
-      "epoch": 1.1572527074358872,
-      "grad_norm": 0.009081362746655941,
-      "learning_rate": 0.00019999934099714021,
-      "loss": 46.0,
-      "step": 7186
-    },
-    {
-      "epoch": 1.1574137445146744,
-      "grad_norm": 0.0009936243295669556,
-      "learning_rate": 0.00019999934081345905,
-      "loss": 46.0,
-      "step": 7187
-    },
-    {
-      "epoch": 1.1575747815934618,
-      "grad_norm": 0.002974365372210741,
-      "learning_rate": 0.00019999934062975226,
-      "loss": 46.0,
-      "step": 7188
-    },
-    {
-      "epoch": 1.1577358186722493,
-      "grad_norm": 0.005104478448629379,
-      "learning_rate": 0.00019999934044601992,
-      "loss": 46.0,
-      "step": 7189
-    },
-    {
-      "epoch": 1.1578968557510367,
-      "grad_norm": 0.0020864761900156736,
-      "learning_rate": 0.00019999934026226196,
-      "loss": 46.0,
-      "step": 7190
-    },
-    {
-      "epoch": 1.1580578928298242,
-      "grad_norm": 0.001229654997587204,
-      "learning_rate": 0.0001999993400784784,
-      "loss": 46.0,
-      "step": 7191
-    },
-    {
-      "epoch": 1.1582189299086114,
-      "grad_norm": 0.001190318027511239,
-      "learning_rate": 0.00019999933989466926,
-      "loss": 46.0,
-      "step": 7192
-    },
-    {
-      "epoch": 1.1583799669873989,
-      "grad_norm": 0.0005623722681775689,
-      "learning_rate": 0.00019999933971083454,
-      "loss": 46.0,
-      "step": 7193
-    },
-    {
-      "epoch": 1.1585410040661863,
-      "grad_norm": 0.002388324122875929,
-      "learning_rate": 0.00019999933952697418,
-      "loss": 46.0,
-      "step": 7194
-    },
-    {
-      "epoch": 1.1587020411449735,
-      "grad_norm": 0.005184921436011791,
-      "learning_rate": 0.00019999933934308826,
-      "loss": 46.0,
-      "step": 7195
-    },
-    {
-      "epoch": 1.158863078223761,
-      "grad_norm": 0.0021305829286575317,
-      "learning_rate": 0.00019999933915917673,
-      "loss": 46.0,
-      "step": 7196
-    },
-    {
-      "epoch": 1.1590241153025485,
-      "grad_norm": 0.003159014508128166,
-      "learning_rate": 0.0001999993389752396,
-      "loss": 46.0,
-      "step": 7197
-    },
-    {
-      "epoch": 1.159185152381336,
-      "grad_norm": 0.0020916624926030636,
-      "learning_rate": 0.0001999993387912769,
-      "loss": 46.0,
-      "step": 7198
-    },
-    {
-      "epoch": 1.1593461894601231,
-      "grad_norm": 0.0012887873454019427,
-      "learning_rate": 0.00019999933860728858,
-      "loss": 46.0,
-      "step": 7199
-    },
-    {
-      "epoch": 1.1595072265389106,
-      "grad_norm": 0.0020860384684056044,
-      "learning_rate": 0.00019999933842327467,
-      "loss": 46.0,
-      "step": 7200
-    },
-    {
-      "epoch": 1.159668263617698,
-      "grad_norm": 0.0009818424005061388,
-      "learning_rate": 0.00019999933823923514,
-      "loss": 46.0,
-      "step": 7201
-    },
-    {
-      "epoch": 1.1598293006964853,
-      "grad_norm": 0.0026219182182103395,
-      "learning_rate": 0.0001999993380551701,
-      "loss": 46.0,
-      "step": 7202
-    },
-    {
-      "epoch": 1.1599903377752727,
-      "grad_norm": 0.0016805967316031456,
-      "learning_rate": 0.0001999993378710794,
-      "loss": 46.0,
-      "step": 7203
-    },
-    {
-      "epoch": 1.1601513748540602,
-      "grad_norm": 0.0015912859234958887,
-      "learning_rate": 0.0001999993376869631,
-      "loss": 46.0,
-      "step": 7204
-    },
-    {
-      "epoch": 1.1603124119328476,
-      "grad_norm": 0.001784482505172491,
-      "learning_rate": 0.00019999933750282123,
-      "loss": 46.0,
-      "step": 7205
-    },
-    {
-      "epoch": 1.1604734490116348,
-      "grad_norm": 0.0005728021496906877,
-      "learning_rate": 0.00019999933731865374,
-      "loss": 46.0,
-      "step": 7206
-    },
-    {
-      "epoch": 1.1606344860904223,
-      "grad_norm": 0.0021490806248039007,
-      "learning_rate": 0.0001999993371344607,
-      "loss": 46.0,
-      "step": 7207
-    },
-    {
-      "epoch": 1.1607955231692098,
-      "grad_norm": 0.0029135493095964193,
-      "learning_rate": 0.00019999933695024203,
-      "loss": 46.0,
-      "step": 7208
-    },
-    {
-      "epoch": 1.160956560247997,
-      "grad_norm": 0.0021653245203197002,
-      "learning_rate": 0.00019999933676599776,
-      "loss": 46.0,
-      "step": 7209
-    },
-    {
-      "epoch": 1.1611175973267844,
-      "grad_norm": 0.0006895381957292557,
-      "learning_rate": 0.00019999933658172792,
-      "loss": 46.0,
-      "step": 7210
-    },
-    {
-      "epoch": 1.1612786344055719,
-      "grad_norm": 0.0009092371328733861,
-      "learning_rate": 0.00019999933639743245,
-      "loss": 46.0,
-      "step": 7211
-    },
-    {
-      "epoch": 1.1614396714843593,
-      "grad_norm": 0.006001030560582876,
-      "learning_rate": 0.0001999993362131114,
-      "loss": 46.0,
-      "step": 7212
-    },
-    {
-      "epoch": 1.1616007085631468,
-      "grad_norm": 0.000743579410482198,
-      "learning_rate": 0.00019999933602876478,
-      "loss": 46.0,
-      "step": 7213
-    },
-    {
-      "epoch": 1.161761745641934,
-      "grad_norm": 0.0032386414241045713,
-      "learning_rate": 0.00019999933584439257,
-      "loss": 46.0,
-      "step": 7214
-    },
-    {
-      "epoch": 1.1619227827207215,
-      "grad_norm": 0.0024503853637725115,
-      "learning_rate": 0.00019999933565999475,
-      "loss": 46.0,
-      "step": 7215
-    },
-    {
-      "epoch": 1.162083819799509,
-      "grad_norm": 0.0004362843174021691,
-      "learning_rate": 0.0001999993354755713,
-      "loss": 46.0,
-      "step": 7216
-    },
-    {
-      "epoch": 1.1622448568782962,
-      "grad_norm": 0.0035724483896046877,
-      "learning_rate": 0.00019999933529112228,
-      "loss": 46.0,
-      "step": 7217
-    },
-    {
-      "epoch": 1.1624058939570836,
-      "grad_norm": 0.0008450191235169768,
-      "learning_rate": 0.00019999933510664767,
-      "loss": 46.0,
-      "step": 7218
-    },
-    {
-      "epoch": 1.162566931035871,
-      "grad_norm": 0.0011962229618802667,
-      "learning_rate": 0.0001999993349221475,
-      "loss": 46.0,
-      "step": 7219
-    },
-    {
-      "epoch": 1.1627279681146585,
-      "grad_norm": 0.0015914092073217034,
-      "learning_rate": 0.00019999933473762168,
-      "loss": 46.0,
-      "step": 7220
-    },
-    {
-      "epoch": 1.1628890051934457,
-      "grad_norm": 0.0011281126644462347,
-      "learning_rate": 0.00019999933455307027,
-      "loss": 46.0,
-      "step": 7221
-    },
-    {
-      "epoch": 1.1630500422722332,
-      "grad_norm": 0.002893853234127164,
-      "learning_rate": 0.0001999993343684933,
-      "loss": 46.0,
-      "step": 7222
-    },
-    {
-      "epoch": 1.1632110793510206,
-      "grad_norm": 0.0015901165315881371,
-      "learning_rate": 0.0001999993341838907,
-      "loss": 46.0,
-      "step": 7223
-    },
-    {
-      "epoch": 1.1633721164298079,
-      "grad_norm": 0.0025290148332715034,
-      "learning_rate": 0.00019999933399926254,
-      "loss": 46.0,
-      "step": 7224
-    },
-    {
-      "epoch": 1.1635331535085953,
-      "grad_norm": 0.0034997931215912104,
-      "learning_rate": 0.00019999933381460876,
-      "loss": 46.0,
-      "step": 7225
-    },
-    {
-      "epoch": 1.1636941905873828,
-      "grad_norm": 0.005680668167769909,
-      "learning_rate": 0.0001999993336299294,
-      "loss": 46.0,
-      "step": 7226
-    },
-    {
-      "epoch": 1.1638552276661702,
-      "grad_norm": 0.0025335336104035378,
-      "learning_rate": 0.00019999933344522444,
-      "loss": 46.0,
-      "step": 7227
-    },
-    {
-      "epoch": 1.1640162647449575,
-      "grad_norm": 0.0023252523969858885,
-      "learning_rate": 0.00019999933326049387,
-      "loss": 46.0,
-      "step": 7228
-    },
-    {
-      "epoch": 1.164177301823745,
-      "grad_norm": 0.0052711511962115765,
-      "learning_rate": 0.00019999933307573772,
-      "loss": 46.0,
-      "step": 7229
-    },
-    {
-      "epoch": 1.1643383389025324,
-      "grad_norm": 0.003125605173408985,
-      "learning_rate": 0.00019999933289095598,
-      "loss": 46.0,
-      "step": 7230
-    },
-    {
-      "epoch": 1.1644993759813196,
-      "grad_norm": 0.0004743531462736428,
-      "learning_rate": 0.00019999933270614862,
-      "loss": 46.0,
-      "step": 7231
-    },
-    {
-      "epoch": 1.164660413060107,
-      "grad_norm": 0.004543558228760958,
-      "learning_rate": 0.00019999933252131568,
-      "loss": 46.0,
-      "step": 7232
-    },
-    {
-      "epoch": 1.1648214501388945,
-      "grad_norm": 0.006632626056671143,
-      "learning_rate": 0.00019999933233645715,
-      "loss": 46.0,
-      "step": 7233
-    },
-    {
-      "epoch": 1.164982487217682,
-      "grad_norm": 0.00105772924143821,
-      "learning_rate": 0.00019999933215157306,
-      "loss": 46.0,
-      "step": 7234
-    },
-    {
-      "epoch": 1.1651435242964692,
-      "grad_norm": 0.0014500149991363287,
-      "learning_rate": 0.00019999933196666333,
-      "loss": 46.0,
-      "step": 7235
-    },
-    {
-      "epoch": 1.1653045613752566,
-      "grad_norm": 0.0018124900525435805,
-      "learning_rate": 0.000199999331781728,
-      "loss": 46.0,
-      "step": 7236
-    },
-    {
-      "epoch": 1.165465598454044,
-      "grad_norm": 0.0007351732929237187,
-      "learning_rate": 0.0001999993315967671,
-      "loss": 46.0,
-      "step": 7237
-    },
-    {
-      "epoch": 1.1656266355328315,
-      "grad_norm": 0.0017234667902812362,
-      "learning_rate": 0.0001999993314117806,
-      "loss": 46.0,
-      "step": 7238
-    },
-    {
-      "epoch": 1.1657876726116188,
-      "grad_norm": 0.0007107029086910188,
-      "learning_rate": 0.0001999993312267685,
-      "loss": 46.0,
-      "step": 7239
-    },
-    {
-      "epoch": 1.1659487096904062,
-      "grad_norm": 0.0015769059536978602,
-      "learning_rate": 0.0001999993310417308,
-      "loss": 46.0,
-      "step": 7240
-    },
-    {
-      "epoch": 1.1661097467691937,
-      "grad_norm": 0.0030417274683713913,
-      "learning_rate": 0.00019999933085666752,
-      "loss": 46.0,
-      "step": 7241
-    },
-    {
-      "epoch": 1.166270783847981,
-      "grad_norm": 0.001748350914567709,
-      "learning_rate": 0.00019999933067157863,
-      "loss": 46.0,
-      "step": 7242
-    },
-    {
-      "epoch": 1.1664318209267683,
-      "grad_norm": 0.0006359772523865104,
-      "learning_rate": 0.00019999933048646414,
-      "loss": 46.0,
-      "step": 7243
-    },
-    {
-      "epoch": 1.1665928580055558,
-      "grad_norm": 0.0017812632722780108,
-      "learning_rate": 0.00019999933030132407,
-      "loss": 46.0,
-      "step": 7244
-    },
-    {
-      "epoch": 1.1667538950843432,
-      "grad_norm": 0.0012820129049941897,
-      "learning_rate": 0.00019999933011615842,
-      "loss": 46.0,
-      "step": 7245
-    },
-    {
-      "epoch": 1.1669149321631305,
-      "grad_norm": 0.004732340574264526,
-      "learning_rate": 0.00019999932993096717,
-      "loss": 46.0,
-      "step": 7246
-    },
-    {
-      "epoch": 1.167075969241918,
-      "grad_norm": 0.0010660496773198247,
-      "learning_rate": 0.0001999993297457503,
-      "loss": 46.0,
-      "step": 7247
-    },
-    {
-      "epoch": 1.1672370063207054,
-      "grad_norm": 0.001341055380180478,
-      "learning_rate": 0.00019999932956050786,
-      "loss": 46.0,
-      "step": 7248
-    },
-    {
-      "epoch": 1.1673980433994928,
-      "grad_norm": 0.0010324451141059399,
-      "learning_rate": 0.0001999993293752398,
-      "loss": 46.0,
-      "step": 7249
-    },
-    {
-      "epoch": 1.16755908047828,
-      "grad_norm": 0.000381359423045069,
-      "learning_rate": 0.00019999932918994618,
-      "loss": 46.0,
-      "step": 7250
-    },
-    {
-      "epoch": 1.1677201175570675,
-      "grad_norm": 0.0008992409566417336,
-      "learning_rate": 0.00019999932900462692,
-      "loss": 46.0,
-      "step": 7251
-    },
-    {
-      "epoch": 1.167881154635855,
-      "grad_norm": 0.001845250721089542,
-      "learning_rate": 0.00019999932881928213,
-      "loss": 46.0,
-      "step": 7252
-    },
-    {
-      "epoch": 1.1680421917146422,
-      "grad_norm": 0.003021238837391138,
-      "learning_rate": 0.0001999993286339117,
-      "loss": 46.0,
-      "step": 7253
-    },
-    {
-      "epoch": 1.1682032287934296,
-      "grad_norm": 0.005578834097832441,
-      "learning_rate": 0.00019999932844851567,
-      "loss": 46.0,
-      "step": 7254
-    },
-    {
-      "epoch": 1.168364265872217,
-      "grad_norm": 0.0012006013421341777,
-      "learning_rate": 0.00019999932826309406,
-      "loss": 46.0,
-      "step": 7255
-    },
-    {
-      "epoch": 1.1685253029510045,
-      "grad_norm": 0.00130279955919832,
-      "learning_rate": 0.00019999932807764683,
-      "loss": 46.0,
-      "step": 7256
-    },
-    {
-      "epoch": 1.1686863400297918,
-      "grad_norm": 0.0006026290939189494,
-      "learning_rate": 0.00019999932789217405,
-      "loss": 46.0,
-      "step": 7257
-    },
-    {
-      "epoch": 1.1688473771085792,
-      "grad_norm": 0.004377065226435661,
-      "learning_rate": 0.00019999932770667562,
-      "loss": 46.0,
-      "step": 7258
-    },
-    {
-      "epoch": 1.1690084141873667,
-      "grad_norm": 0.002834476064890623,
-      "learning_rate": 0.00019999932752115166,
-      "loss": 46.0,
-      "step": 7259
-    },
-    {
-      "epoch": 1.1691694512661541,
-      "grad_norm": 0.004933305084705353,
-      "learning_rate": 0.00019999932733560206,
-      "loss": 46.0,
-      "step": 7260
-    },
-    {
-      "epoch": 1.1693304883449414,
-      "grad_norm": 0.0021575321443378925,
-      "learning_rate": 0.00019999932715002687,
-      "loss": 46.0,
-      "step": 7261
-    },
-    {
-      "epoch": 1.1694915254237288,
-      "grad_norm": 0.001254245056770742,
-      "learning_rate": 0.00019999932696442612,
-      "loss": 46.0,
-      "step": 7262
-    },
-    {
-      "epoch": 1.1696525625025163,
-      "grad_norm": 0.0013567038113251328,
-      "learning_rate": 0.0001999993267787997,
-      "loss": 46.0,
-      "step": 7263
-    },
-    {
-      "epoch": 1.1698135995813037,
-      "grad_norm": 0.0012087170034646988,
-      "learning_rate": 0.00019999932659314778,
-      "loss": 46.0,
-      "step": 7264
-    },
-    {
-      "epoch": 1.169974636660091,
-      "grad_norm": 0.002291240496560931,
-      "learning_rate": 0.00019999932640747022,
-      "loss": 46.0,
-      "step": 7265
-    },
-    {
-      "epoch": 1.1701356737388784,
-      "grad_norm": 0.005932828411459923,
-      "learning_rate": 0.00019999932622176704,
-      "loss": 46.0,
-      "step": 7266
-    },
-    {
-      "epoch": 1.1702967108176658,
-      "grad_norm": 0.0006304119015112519,
-      "learning_rate": 0.0001999993260360383,
-      "loss": 46.0,
-      "step": 7267
-    },
-    {
-      "epoch": 1.170457747896453,
-      "grad_norm": 0.0004513065214268863,
-      "learning_rate": 0.00019999932585028395,
-      "loss": 46.0,
-      "step": 7268
-    },
-    {
-      "epoch": 1.1706187849752405,
-      "grad_norm": 0.0024564098566770554,
-      "learning_rate": 0.000199999325664504,
-      "loss": 46.0,
-      "step": 7269
-    },
-    {
-      "epoch": 1.170779822054028,
-      "grad_norm": 0.0038533704355359077,
-      "learning_rate": 0.00019999932547869848,
-      "loss": 46.0,
-      "step": 7270
-    },
-    {
-      "epoch": 1.1709408591328154,
-      "grad_norm": 0.005282533820718527,
-      "learning_rate": 0.00019999932529286734,
-      "loss": 46.0,
-      "step": 7271
-    },
-    {
-      "epoch": 1.1711018962116027,
-      "grad_norm": 0.0011273275595158339,
-      "learning_rate": 0.00019999932510701062,
-      "loss": 46.0,
-      "step": 7272
-    },
-    {
-      "epoch": 1.17126293329039,
-      "grad_norm": 0.0006987282540649176,
-      "learning_rate": 0.00019999932492112833,
-      "loss": 46.0,
-      "step": 7273
-    },
-    {
-      "epoch": 1.1714239703691776,
-      "grad_norm": 0.0004023484652861953,
-      "learning_rate": 0.0001999993247352204,
-      "loss": 46.0,
-      "step": 7274
-    },
-    {
-      "epoch": 1.1715850074479648,
-      "grad_norm": 0.0016293387161567807,
-      "learning_rate": 0.00019999932454928688,
-      "loss": 46.0,
-      "step": 7275
-    },
-    {
-      "epoch": 1.1717460445267522,
-      "grad_norm": 0.0005998879205435514,
-      "learning_rate": 0.0001999993243633278,
-      "loss": 46.0,
-      "step": 7276
-    },
-    {
-      "epoch": 1.1719070816055397,
-      "grad_norm": 0.0014990271301940084,
-      "learning_rate": 0.00019999932417734309,
-      "loss": 46.0,
-      "step": 7277
-    },
-    {
-      "epoch": 1.1720681186843271,
-      "grad_norm": 0.0009858167031779885,
-      "learning_rate": 0.00019999932399133278,
-      "loss": 46.0,
-      "step": 7278
-    },
-    {
-      "epoch": 1.1722291557631144,
-      "grad_norm": 0.0030931327491998672,
-      "learning_rate": 0.00019999932380529691,
-      "loss": 46.0,
-      "step": 7279
-    },
-    {
-      "epoch": 1.1723901928419018,
-      "grad_norm": 0.0014890697784721851,
-      "learning_rate": 0.0001999993236192354,
-      "loss": 46.0,
-      "step": 7280
-    },
-    {
-      "epoch": 1.1725512299206893,
-      "grad_norm": 0.00184484978672117,
-      "learning_rate": 0.00019999932343314837,
-      "loss": 46.0,
-      "step": 7281
-    },
-    {
-      "epoch": 1.1727122669994765,
-      "grad_norm": 0.0006240683142095804,
-      "learning_rate": 0.00019999932324703569,
-      "loss": 46.0,
-      "step": 7282
-    },
-    {
-      "epoch": 1.172873304078264,
-      "grad_norm": 0.002089295070618391,
-      "learning_rate": 0.00019999932306089742,
-      "loss": 46.0,
-      "step": 7283
-    },
-    {
-      "epoch": 1.1730343411570514,
-      "grad_norm": 0.002121787518262863,
-      "learning_rate": 0.00019999932287473356,
-      "loss": 46.0,
-      "step": 7284
-    },
-    {
-      "epoch": 1.1731953782358389,
-      "grad_norm": 0.008654619567096233,
-      "learning_rate": 0.00019999932268854412,
-      "loss": 46.0,
-      "step": 7285
-    },
-    {
-      "epoch": 1.1733564153146263,
-      "grad_norm": 0.0018087526550516486,
-      "learning_rate": 0.00019999932250232906,
-      "loss": 46.0,
-      "step": 7286
-    },
-    {
-      "epoch": 1.1735174523934135,
-      "grad_norm": 0.0009565085638314486,
-      "learning_rate": 0.0001999993223160884,
-      "loss": 46.0,
-      "step": 7287
-    },
-    {
-      "epoch": 1.173678489472201,
-      "grad_norm": 0.0021637678146362305,
-      "learning_rate": 0.00019999932212982216,
-      "loss": 46.0,
-      "step": 7288
-    },
-    {
-      "epoch": 1.1738395265509884,
-      "grad_norm": 0.0012099015293642879,
-      "learning_rate": 0.00019999932194353031,
-      "loss": 46.0,
-      "step": 7289
-    },
-    {
-      "epoch": 1.1740005636297757,
-      "grad_norm": 0.002240993082523346,
-      "learning_rate": 0.0001999993217572129,
-      "loss": 46.0,
-      "step": 7290
-    },
-    {
-      "epoch": 1.1741616007085631,
-      "grad_norm": 0.0015738093061372638,
-      "learning_rate": 0.0001999993215708699,
-      "loss": 46.0,
-      "step": 7291
-    },
-    {
-      "epoch": 1.1743226377873506,
-      "grad_norm": 0.0020281930919736624,
-      "learning_rate": 0.00019999932138450128,
-      "loss": 46.0,
-      "step": 7292
-    },
-    {
-      "epoch": 1.174483674866138,
-      "grad_norm": 0.000978005351498723,
-      "learning_rate": 0.00019999932119810706,
-      "loss": 46.0,
-      "step": 7293
-    },
-    {
-      "epoch": 1.1746447119449253,
-      "grad_norm": 0.0012880610302090645,
-      "learning_rate": 0.00019999932101168722,
-      "loss": 46.0,
-      "step": 7294
-    },
-    {
-      "epoch": 1.1748057490237127,
-      "grad_norm": 0.0026429451536387205,
-      "learning_rate": 0.00019999932082524186,
-      "loss": 46.0,
-      "step": 7295
-    },
-    {
-      "epoch": 1.1749667861025002,
-      "grad_norm": 0.0023867327254265547,
-      "learning_rate": 0.00019999932063877084,
-      "loss": 46.0,
-      "step": 7296
-    },
-    {
-      "epoch": 1.1751278231812874,
-      "grad_norm": 0.0013893722789362073,
-      "learning_rate": 0.00019999932045227427,
-      "loss": 46.0,
-      "step": 7297
-    },
-    {
-      "epoch": 1.1752888602600748,
-      "grad_norm": 0.002012244425714016,
-      "learning_rate": 0.0001999993202657521,
-      "loss": 46.0,
-      "step": 7298
-    },
-    {
-      "epoch": 1.1754498973388623,
-      "grad_norm": 0.0033517684787511826,
-      "learning_rate": 0.0001999993200792043,
-      "loss": 46.0,
-      "step": 7299
-    },
-    {
-      "epoch": 1.1756109344176497,
-      "grad_norm": 0.0014964157016947865,
-      "learning_rate": 0.00019999931989263093,
-      "loss": 46.0,
-      "step": 7300
-    },
-    {
-      "epoch": 1.175771971496437,
-      "grad_norm": 0.0009307727450504899,
-      "learning_rate": 0.00019999931970603196,
-      "loss": 46.0,
-      "step": 7301
-    },
-    {
-      "epoch": 1.1759330085752244,
-      "grad_norm": 0.0026985756121575832,
-      "learning_rate": 0.00019999931951940737,
-      "loss": 46.0,
-      "step": 7302
-    },
-    {
-      "epoch": 1.1760940456540119,
-      "grad_norm": 0.005229425150901079,
-      "learning_rate": 0.00019999931933275723,
-      "loss": 46.0,
-      "step": 7303
-    },
-    {
-      "epoch": 1.176255082732799,
-      "grad_norm": 0.005166003946214914,
-      "learning_rate": 0.00019999931914608146,
-      "loss": 46.0,
-      "step": 7304
-    },
-    {
-      "epoch": 1.1764161198115866,
-      "grad_norm": 0.0017078024102374911,
-      "learning_rate": 0.00019999931895938011,
-      "loss": 46.0,
-      "step": 7305
-    },
-    {
-      "epoch": 1.176577156890374,
-      "grad_norm": 0.0006303245900198817,
-      "learning_rate": 0.00019999931877265318,
-      "loss": 46.0,
-      "step": 7306
-    },
-    {
-      "epoch": 1.1767381939691615,
-      "grad_norm": 0.00028924617799930274,
-      "learning_rate": 0.00019999931858590063,
-      "loss": 46.0,
-      "step": 7307
-    },
-    {
-      "epoch": 1.176899231047949,
-      "grad_norm": 0.002286311471834779,
-      "learning_rate": 0.00019999931839912252,
-      "loss": 46.0,
-      "step": 7308
-    },
-    {
-      "epoch": 1.1770602681267361,
-      "grad_norm": 0.0020275702700018883,
-      "learning_rate": 0.0001999993182123188,
-      "loss": 46.0,
-      "step": 7309
-    },
-    {
-      "epoch": 1.1772213052055236,
-      "grad_norm": 0.0018592944834381342,
-      "learning_rate": 0.00019999931802548945,
-      "loss": 46.0,
-      "step": 7310
-    },
-    {
-      "epoch": 1.177382342284311,
-      "grad_norm": 0.0011300251353532076,
-      "learning_rate": 0.00019999931783863456,
-      "loss": 46.0,
-      "step": 7311
-    },
-    {
-      "epoch": 1.1775433793630983,
-      "grad_norm": 0.0021666372194886208,
-      "learning_rate": 0.00019999931765175404,
-      "loss": 46.0,
-      "step": 7312
-    },
-    {
-      "epoch": 1.1777044164418857,
-      "grad_norm": 0.0016806612256914377,
-      "learning_rate": 0.00019999931746484791,
-      "loss": 46.0,
-      "step": 7313
-    },
-    {
-      "epoch": 1.1778654535206732,
-      "grad_norm": 0.001958418870344758,
-      "learning_rate": 0.00019999931727791623,
-      "loss": 46.0,
-      "step": 7314
-    },
-    {
-      "epoch": 1.1780264905994606,
-      "grad_norm": 0.0034716171212494373,
-      "learning_rate": 0.00019999931709095893,
-      "loss": 46.0,
-      "step": 7315
-    },
-    {
-      "epoch": 1.1781875276782479,
-      "grad_norm": 0.0005778723862022161,
-      "learning_rate": 0.00019999931690397604,
-      "loss": 46.0,
-      "step": 7316
-    },
-    {
-      "epoch": 1.1783485647570353,
-      "grad_norm": 0.005233352072536945,
-      "learning_rate": 0.00019999931671696753,
-      "loss": 46.0,
-      "step": 7317
-    },
-    {
-      "epoch": 1.1785096018358228,
-      "grad_norm": 0.002933028619736433,
-      "learning_rate": 0.00019999931652993347,
-      "loss": 46.0,
-      "step": 7318
-    },
-    {
-      "epoch": 1.17867063891461,
-      "grad_norm": 0.0032525649294257164,
-      "learning_rate": 0.00019999931634287382,
-      "loss": 46.0,
-      "step": 7319
-    },
-    {
-      "epoch": 1.1788316759933974,
-      "grad_norm": 0.0003374084772076458,
-      "learning_rate": 0.00019999931615578853,
-      "loss": 46.0,
-      "step": 7320
-    },
-    {
-      "epoch": 1.178992713072185,
-      "grad_norm": 0.002870365511626005,
-      "learning_rate": 0.00019999931596867768,
-      "loss": 46.0,
-      "step": 7321
-    },
-    {
-      "epoch": 1.1791537501509723,
-      "grad_norm": 0.0010719697456806898,
-      "learning_rate": 0.0001999993157815412,
-      "loss": 46.0,
-      "step": 7322
-    },
-    {
-      "epoch": 1.1793147872297596,
-      "grad_norm": 0.001004177494905889,
-      "learning_rate": 0.00019999931559437916,
-      "loss": 46.0,
-      "step": 7323
-    },
-    {
-      "epoch": 1.179475824308547,
-      "grad_norm": 0.0014371376018971205,
-      "learning_rate": 0.00019999931540719152,
-      "loss": 46.0,
-      "step": 7324
-    },
-    {
-      "epoch": 1.1796368613873345,
-      "grad_norm": 0.0006791423074901104,
-      "learning_rate": 0.00019999931521997826,
-      "loss": 46.0,
-      "step": 7325
-    },
-    {
-      "epoch": 1.1797978984661217,
-      "grad_norm": 0.0012079504085704684,
-      "learning_rate": 0.00019999931503273942,
-      "loss": 46.0,
-      "step": 7326
-    },
-    {
-      "epoch": 1.1799589355449092,
-      "grad_norm": 0.005674366839230061,
-      "learning_rate": 0.00019999931484547502,
-      "loss": 46.0,
-      "step": 7327
-    },
-    {
-      "epoch": 1.1801199726236966,
-      "grad_norm": 0.0011005571577697992,
-      "learning_rate": 0.00019999931465818497,
-      "loss": 46.0,
-      "step": 7328
-    },
-    {
-      "epoch": 1.180281009702484,
-      "grad_norm": 0.007835574448108673,
-      "learning_rate": 0.00019999931447086934,
-      "loss": 46.0,
-      "step": 7329
-    },
-    {
-      "epoch": 1.1804420467812713,
-      "grad_norm": 0.0007141850655898452,
-      "learning_rate": 0.00019999931428352815,
-      "loss": 46.0,
-      "step": 7330
-    },
-    {
-      "epoch": 1.1806030838600587,
-      "grad_norm": 0.0016291745705530047,
-      "learning_rate": 0.00019999931409616132,
-      "loss": 46.0,
-      "step": 7331
-    },
-    {
-      "epoch": 1.1807641209388462,
-      "grad_norm": 0.000596243015024811,
-      "learning_rate": 0.00019999931390876893,
-      "loss": 46.0,
-      "step": 7332
-    },
-    {
-      "epoch": 1.1809251580176336,
-      "grad_norm": 0.0007783645414747298,
-      "learning_rate": 0.00019999931372135092,
-      "loss": 46.0,
-      "step": 7333
-    },
-    {
-      "epoch": 1.1810861950964209,
-      "grad_norm": 0.0017039704835042357,
-      "learning_rate": 0.00019999931353390733,
-      "loss": 46.0,
-      "step": 7334
-    },
-    {
-      "epoch": 1.1812472321752083,
-      "grad_norm": 0.006421853322535753,
-      "learning_rate": 0.00019999931334643815,
-      "loss": 46.0,
-      "step": 7335
-    },
-    {
-      "epoch": 1.1814082692539958,
-      "grad_norm": 0.0016386555507779121,
-      "learning_rate": 0.00019999931315894335,
-      "loss": 46.0,
-      "step": 7336
-    },
-    {
-      "epoch": 1.1815693063327832,
-      "grad_norm": 0.004061900544911623,
-      "learning_rate": 0.00019999931297142297,
-      "loss": 46.0,
-      "step": 7337
-    },
-    {
-      "epoch": 1.1817303434115705,
-      "grad_norm": 0.001283218851312995,
-      "learning_rate": 0.000199999312783877,
-      "loss": 46.0,
-      "step": 7338
-    },
-    {
-      "epoch": 1.181891380490358,
-      "grad_norm": 0.0006205645040608943,
-      "learning_rate": 0.00019999931259630542,
-      "loss": 46.0,
-      "step": 7339
-    },
-    {
-      "epoch": 1.1820524175691454,
-      "grad_norm": 0.0011197656858712435,
-      "learning_rate": 0.00019999931240870827,
-      "loss": 46.0,
-      "step": 7340
-    },
-    {
-      "epoch": 1.1822134546479326,
-      "grad_norm": 0.004171561449766159,
-      "learning_rate": 0.00019999931222108551,
-      "loss": 46.0,
-      "step": 7341
-    },
-    {
-      "epoch": 1.18237449172672,
-      "grad_norm": 0.0030213803984224796,
-      "learning_rate": 0.00019999931203343717,
-      "loss": 46.0,
-      "step": 7342
-    },
-    {
-      "epoch": 1.1825355288055075,
-      "grad_norm": 0.002057740930467844,
-      "learning_rate": 0.0001999993118457632,
-      "loss": 46.0,
-      "step": 7343
-    },
-    {
-      "epoch": 1.182696565884295,
-      "grad_norm": 0.0021582413464784622,
-      "learning_rate": 0.00019999931165806366,
-      "loss": 46.0,
-      "step": 7344
-    },
-    {
-      "epoch": 1.1828576029630822,
-      "grad_norm": 0.004215786699205637,
-      "learning_rate": 0.00019999931147033856,
-      "loss": 46.0,
-      "step": 7345
-    },
-    {
-      "epoch": 1.1830186400418696,
-      "grad_norm": 0.0018000471172854304,
-      "learning_rate": 0.0001999993112825878,
-      "loss": 46.0,
-      "step": 7346
-    },
-    {
-      "epoch": 1.183179677120657,
-      "grad_norm": 0.0007439761538989842,
-      "learning_rate": 0.00019999931109481147,
-      "loss": 46.0,
-      "step": 7347
-    },
-    {
-      "epoch": 1.1833407141994443,
-      "grad_norm": 0.004136794712394476,
-      "learning_rate": 0.00019999931090700955,
-      "loss": 46.0,
-      "step": 7348
-    },
-    {
-      "epoch": 1.1835017512782318,
-      "grad_norm": 0.002294244011864066,
-      "learning_rate": 0.00019999931071918204,
-      "loss": 46.0,
-      "step": 7349
-    },
-    {
-      "epoch": 1.1836627883570192,
-      "grad_norm": 0.0035066576674580574,
-      "learning_rate": 0.00019999931053132892,
-      "loss": 46.0,
-      "step": 7350
-    },
-    {
-      "epoch": 1.1838238254358067,
-      "grad_norm": 0.000524053000845015,
-      "learning_rate": 0.00019999931034345023,
-      "loss": 46.0,
-      "step": 7351
-    },
-    {
-      "epoch": 1.183984862514594,
-      "grad_norm": 0.004031005315482616,
-      "learning_rate": 0.0001999993101555459,
-      "loss": 46.0,
-      "step": 7352
-    },
-    {
-      "epoch": 1.1841458995933813,
-      "grad_norm": 0.0007816303404979408,
-      "learning_rate": 0.00019999930996761602,
-      "loss": 46.0,
-      "step": 7353
-    },
-    {
-      "epoch": 1.1843069366721688,
-      "grad_norm": 0.0023815478198230267,
-      "learning_rate": 0.00019999930977966055,
-      "loss": 46.0,
-      "step": 7354
-    },
-    {
-      "epoch": 1.1844679737509562,
-      "grad_norm": 0.00274737854488194,
-      "learning_rate": 0.00019999930959167946,
-      "loss": 46.0,
-      "step": 7355
-    },
-    {
-      "epoch": 1.1846290108297435,
-      "grad_norm": 0.0018352791666984558,
-      "learning_rate": 0.00019999930940367276,
-      "loss": 46.0,
-      "step": 7356
-    },
-    {
-      "epoch": 1.184790047908531,
-      "grad_norm": 0.004598895087838173,
-      "learning_rate": 0.00019999930921564053,
-      "loss": 46.0,
-      "step": 7357
-    },
-    {
-      "epoch": 1.1849510849873184,
-      "grad_norm": 0.0040381369180977345,
-      "learning_rate": 0.00019999930902758262,
-      "loss": 46.0,
-      "step": 7358
-    },
-    {
-      "epoch": 1.1851121220661058,
-      "grad_norm": 0.0026830604765564203,
-      "learning_rate": 0.0001999993088394992,
-      "loss": 46.0,
-      "step": 7359
-    },
-    {
-      "epoch": 1.185273159144893,
-      "grad_norm": 0.004803386516869068,
-      "learning_rate": 0.00019999930865139014,
-      "loss": 46.0,
-      "step": 7360
-    },
-    {
-      "epoch": 1.1854341962236805,
-      "grad_norm": 0.002542213536798954,
-      "learning_rate": 0.00019999930846325547,
-      "loss": 46.0,
-      "step": 7361
-    },
-    {
-      "epoch": 1.185595233302468,
-      "grad_norm": 0.0015223032096400857,
-      "learning_rate": 0.00019999930827509522,
-      "loss": 46.0,
-      "step": 7362
-    },
-    {
-      "epoch": 1.1857562703812552,
-      "grad_norm": 0.004237472545355558,
-      "learning_rate": 0.00019999930808690939,
-      "loss": 46.0,
-      "step": 7363
-    },
-    {
-      "epoch": 1.1859173074600426,
-      "grad_norm": 0.0011035024654120207,
-      "learning_rate": 0.00019999930789869796,
-      "loss": 46.0,
-      "step": 7364
-    },
-    {
-      "epoch": 1.18607834453883,
-      "grad_norm": 0.001071646111086011,
-      "learning_rate": 0.00019999930771046095,
-      "loss": 46.0,
-      "step": 7365
-    },
-    {
-      "epoch": 1.1862393816176175,
-      "grad_norm": 0.0038553241174668074,
-      "learning_rate": 0.0001999993075221983,
-      "loss": 46.0,
-      "step": 7366
-    },
-    {
-      "epoch": 1.1864004186964048,
-      "grad_norm": 0.005944760516285896,
-      "learning_rate": 0.00019999930733391008,
-      "loss": 46.0,
-      "step": 7367
-    },
-    {
-      "epoch": 1.1865614557751922,
-      "grad_norm": 0.0004066654946655035,
-      "learning_rate": 0.00019999930714559628,
-      "loss": 46.0,
-      "step": 7368
-    },
-    {
-      "epoch": 1.1867224928539797,
-      "grad_norm": 0.0027252298314124346,
-      "learning_rate": 0.00019999930695725686,
-      "loss": 46.0,
-      "step": 7369
-    },
-    {
-      "epoch": 1.186883529932767,
-      "grad_norm": 0.0009907943895086646,
-      "learning_rate": 0.00019999930676889186,
-      "loss": 46.0,
-      "step": 7370
-    },
-    {
-      "epoch": 1.1870445670115544,
-      "grad_norm": 0.006962188985198736,
-      "learning_rate": 0.00019999930658050127,
-      "loss": 46.0,
-      "step": 7371
-    },
-    {
-      "epoch": 1.1872056040903418,
-      "grad_norm": 0.0016975963953882456,
-      "learning_rate": 0.00019999930639208506,
-      "loss": 46.0,
-      "step": 7372
-    },
-    {
-      "epoch": 1.1873666411691293,
-      "grad_norm": 0.00544242188334465,
-      "learning_rate": 0.0001999993062036433,
-      "loss": 46.0,
-      "step": 7373
-    },
-    {
-      "epoch": 1.1875276782479165,
-      "grad_norm": 0.0008141516009345651,
-      "learning_rate": 0.0001999993060151759,
-      "loss": 46.0,
-      "step": 7374
-    },
-    {
-      "epoch": 1.187688715326704,
-      "grad_norm": 0.0009726479183882475,
-      "learning_rate": 0.0001999993058266829,
-      "loss": 46.0,
-      "step": 7375
-    },
-    {
-      "epoch": 1.1878497524054914,
-      "grad_norm": 0.002043060725554824,
-      "learning_rate": 0.00019999930563816435,
-      "loss": 46.0,
-      "step": 7376
-    },
-    {
-      "epoch": 1.1880107894842789,
-      "grad_norm": 0.0012031319784000516,
-      "learning_rate": 0.00019999930544962018,
-      "loss": 46.0,
-      "step": 7377
-    },
-    {
-      "epoch": 1.188171826563066,
-      "grad_norm": 0.002612357959151268,
-      "learning_rate": 0.00019999930526105043,
-      "loss": 46.0,
-      "step": 7378
-    },
-    {
-      "epoch": 1.1883328636418535,
-      "grad_norm": 0.0018331792671233416,
-      "learning_rate": 0.00019999930507245508,
-      "loss": 46.0,
-      "step": 7379
-    },
-    {
-      "epoch": 1.188493900720641,
-      "grad_norm": 0.00260573229752481,
-      "learning_rate": 0.00019999930488383413,
-      "loss": 46.0,
-      "step": 7380
-    },
-    {
-      "epoch": 1.1886549377994284,
-      "grad_norm": 0.0032797090243548155,
-      "learning_rate": 0.00019999930469518759,
-      "loss": 46.0,
-      "step": 7381
-    },
-    {
-      "epoch": 1.1888159748782157,
-      "grad_norm": 0.0003878821444232017,
-      "learning_rate": 0.00019999930450651543,
-      "loss": 46.0,
-      "step": 7382
-    },
-    {
-      "epoch": 1.1889770119570031,
-      "grad_norm": 0.005170104093849659,
-      "learning_rate": 0.0001999993043178177,
-      "loss": 46.0,
-      "step": 7383
-    },
-    {
-      "epoch": 1.1891380490357906,
-      "grad_norm": 0.003059866139665246,
-      "learning_rate": 0.00019999930412909435,
-      "loss": 46.0,
-      "step": 7384
-    },
-    {
-      "epoch": 1.1892990861145778,
-      "grad_norm": 0.0032380837947130203,
-      "learning_rate": 0.00019999930394034543,
-      "loss": 46.0,
-      "step": 7385
-    },
-    {
-      "epoch": 1.1894601231933652,
-      "grad_norm": 0.000769248406868428,
-      "learning_rate": 0.00019999930375157093,
-      "loss": 46.0,
-      "step": 7386
-    },
-    {
-      "epoch": 1.1896211602721527,
-      "grad_norm": 0.0011289390968158841,
-      "learning_rate": 0.0001999993035627708,
-      "loss": 46.0,
-      "step": 7387
-    },
-    {
-      "epoch": 1.1897821973509402,
-      "grad_norm": 0.0006276924978010356,
-      "learning_rate": 0.0001999993033739451,
-      "loss": 46.0,
-      "step": 7388
-    },
-    {
-      "epoch": 1.1899432344297274,
-      "grad_norm": 0.0010052825091406703,
-      "learning_rate": 0.00019999930318509378,
-      "loss": 46.0,
-      "step": 7389
-    },
-    {
-      "epoch": 1.1901042715085148,
-      "grad_norm": 0.004562370013445616,
-      "learning_rate": 0.0001999993029962169,
-      "loss": 46.0,
-      "step": 7390
-    },
-    {
-      "epoch": 1.1902653085873023,
-      "grad_norm": 0.0004512219747994095,
-      "learning_rate": 0.0001999993028073144,
-      "loss": 46.0,
-      "step": 7391
-    },
-    {
-      "epoch": 1.1904263456660895,
-      "grad_norm": 0.0041924635879695415,
-      "learning_rate": 0.00019999930261838632,
-      "loss": 46.0,
-      "step": 7392
-    },
-    {
-      "epoch": 1.190587382744877,
-      "grad_norm": 0.000982915167696774,
-      "learning_rate": 0.00019999930242943265,
-      "loss": 46.0,
-      "step": 7393
-    },
-    {
-      "epoch": 1.1907484198236644,
-      "grad_norm": 0.0014111737255007029,
-      "learning_rate": 0.0001999993022404534,
-      "loss": 46.0,
-      "step": 7394
-    },
-    {
-      "epoch": 1.1909094569024519,
-      "grad_norm": 0.004768189042806625,
-      "learning_rate": 0.0001999993020514485,
-      "loss": 46.0,
-      "step": 7395
-    },
-    {
-      "epoch": 1.191070493981239,
-      "grad_norm": 0.002425622660666704,
-      "learning_rate": 0.00019999930186241803,
-      "loss": 46.0,
-      "step": 7396
-    },
-    {
-      "epoch": 1.1912315310600265,
-      "grad_norm": 0.0008161734440363944,
-      "learning_rate": 0.00019999930167336198,
-      "loss": 46.0,
-      "step": 7397
-    },
-    {
-      "epoch": 1.191392568138814,
-      "grad_norm": 0.0012945233611389995,
-      "learning_rate": 0.0001999993014842803,
-      "loss": 46.0,
-      "step": 7398
-    },
-    {
-      "epoch": 1.1915536052176012,
-      "grad_norm": 0.001209920272231102,
-      "learning_rate": 0.00019999930129517305,
-      "loss": 46.0,
-      "step": 7399
-    },
-    {
-      "epoch": 1.1917146422963887,
-      "grad_norm": 0.0005322758224792778,
-      "learning_rate": 0.00019999930110604021,
-      "loss": 46.0,
-      "step": 7400
-    },
-    {
-      "epoch": 1.1918756793751761,
-      "grad_norm": 0.00047942172386683524,
-      "learning_rate": 0.00019999930091688177,
-      "loss": 46.0,
-      "step": 7401
-    },
-    {
-      "epoch": 1.1920367164539636,
-      "grad_norm": 0.0017445423873141408,
-      "learning_rate": 0.00019999930072769776,
-      "loss": 46.0,
-      "step": 7402
-    },
-    {
-      "epoch": 1.192197753532751,
-      "grad_norm": 0.0020476365461945534,
-      "learning_rate": 0.00019999930053848813,
-      "loss": 46.0,
-      "step": 7403
-    },
-    {
-      "epoch": 1.1923587906115383,
-      "grad_norm": 0.0036957187112420797,
-      "learning_rate": 0.0001999993003492529,
-      "loss": 46.0,
-      "step": 7404
-    },
-    {
-      "epoch": 1.1925198276903257,
-      "grad_norm": 0.00220651482231915,
-      "learning_rate": 0.00019999930015999207,
-      "loss": 46.0,
-      "step": 7405
-    },
-    {
-      "epoch": 1.1926808647691132,
-      "grad_norm": 0.0008695185533724725,
-      "learning_rate": 0.00019999929997070566,
-      "loss": 46.0,
-      "step": 7406
-    },
-    {
-      "epoch": 1.1928419018479004,
-      "grad_norm": 0.0010108186397701502,
-      "learning_rate": 0.00019999929978139366,
-      "loss": 46.0,
-      "step": 7407
-    },
-    {
-      "epoch": 1.1930029389266879,
-      "grad_norm": 0.0006435916875489056,
-      "learning_rate": 0.00019999929959205605,
-      "loss": 46.0,
-      "step": 7408
-    },
-    {
-      "epoch": 1.1931639760054753,
-      "grad_norm": 0.0014818136114627123,
-      "learning_rate": 0.00019999929940269287,
-      "loss": 46.0,
-      "step": 7409
-    },
-    {
-      "epoch": 1.1933250130842628,
-      "grad_norm": 0.0004252234648447484,
-      "learning_rate": 0.00019999929921330406,
-      "loss": 46.0,
-      "step": 7410
-    },
-    {
-      "epoch": 1.19348605016305,
-      "grad_norm": 0.0009287706343457103,
-      "learning_rate": 0.0001999992990238897,
-      "loss": 46.0,
-      "step": 7411
-    },
-    {
-      "epoch": 1.1936470872418374,
-      "grad_norm": 0.000723625416867435,
-      "learning_rate": 0.0001999992988344497,
-      "loss": 46.0,
-      "step": 7412
-    },
-    {
-      "epoch": 1.1938081243206249,
-      "grad_norm": 0.0014388399431481957,
-      "learning_rate": 0.00019999929864498412,
-      "loss": 46.0,
-      "step": 7413
-    },
-    {
-      "epoch": 1.1939691613994121,
-      "grad_norm": 0.00482554268091917,
-      "learning_rate": 0.00019999929845549295,
-      "loss": 46.0,
-      "step": 7414
-    },
-    {
-      "epoch": 1.1941301984781996,
-      "grad_norm": 0.0002463750133756548,
-      "learning_rate": 0.00019999929826597617,
-      "loss": 46.0,
-      "step": 7415
-    },
-    {
-      "epoch": 1.194291235556987,
-      "grad_norm": 0.0011618523858487606,
-      "learning_rate": 0.00019999929807643384,
-      "loss": 46.0,
-      "step": 7416
-    },
-    {
-      "epoch": 1.1944522726357745,
-      "grad_norm": 0.002354543888941407,
-      "learning_rate": 0.00019999929788686588,
-      "loss": 46.0,
-      "step": 7417
-    },
-    {
-      "epoch": 1.1946133097145617,
-      "grad_norm": 0.000962183577939868,
-      "learning_rate": 0.00019999929769727232,
-      "loss": 46.0,
-      "step": 7418
-    },
-    {
-      "epoch": 1.1947743467933492,
-      "grad_norm": 0.001150438329204917,
-      "learning_rate": 0.0001999992975076532,
-      "loss": 46.0,
-      "step": 7419
-    },
-    {
-      "epoch": 1.1949353838721366,
-      "grad_norm": 0.001018354669213295,
-      "learning_rate": 0.00019999929731800845,
-      "loss": 46.0,
-      "step": 7420
-    },
-    {
-      "epoch": 1.1950964209509238,
-      "grad_norm": 0.0009027085034176707,
-      "learning_rate": 0.00019999929712833812,
-      "loss": 46.0,
-      "step": 7421
-    },
-    {
-      "epoch": 1.1952574580297113,
-      "grad_norm": 0.006388049107044935,
-      "learning_rate": 0.00019999929693864218,
-      "loss": 46.0,
-      "step": 7422
-    },
-    {
-      "epoch": 1.1954184951084987,
-      "grad_norm": 0.0009071163949556649,
-      "learning_rate": 0.00019999929674892065,
-      "loss": 46.0,
-      "step": 7423
-    },
-    {
-      "epoch": 1.1955795321872862,
-      "grad_norm": 0.0011035699862986803,
-      "learning_rate": 0.00019999929655917356,
-      "loss": 46.0,
-      "step": 7424
-    },
-    {
-      "epoch": 1.1957405692660734,
-      "grad_norm": 0.0010314702522009611,
-      "learning_rate": 0.00019999929636940086,
-      "loss": 46.0,
-      "step": 7425
-    },
-    {
-      "epoch": 1.1959016063448609,
-      "grad_norm": 0.0004092316667083651,
-      "learning_rate": 0.00019999929617960254,
-      "loss": 46.0,
-      "step": 7426
-    },
-    {
-      "epoch": 1.1960626434236483,
-      "grad_norm": 0.0012313921470195055,
-      "learning_rate": 0.00019999929598977863,
-      "loss": 46.0,
-      "step": 7427
-    },
-    {
-      "epoch": 1.1962236805024358,
-      "grad_norm": 0.0015070174122229218,
-      "learning_rate": 0.00019999929579992914,
-      "loss": 46.0,
-      "step": 7428
-    },
-    {
-      "epoch": 1.196384717581223,
-      "grad_norm": 0.0010475274175405502,
-      "learning_rate": 0.00019999929561005403,
-      "loss": 46.0,
-      "step": 7429
-    },
-    {
-      "epoch": 1.1965457546600105,
-      "grad_norm": 0.004691485781222582,
-      "learning_rate": 0.00019999929542015337,
-      "loss": 46.0,
-      "step": 7430
-    },
-    {
-      "epoch": 1.196706791738798,
-      "grad_norm": 0.0021805621217936277,
-      "learning_rate": 0.00019999929523022706,
-      "loss": 46.0,
-      "step": 7431
-    },
-    {
-      "epoch": 1.1968678288175854,
-      "grad_norm": 0.0036309207789599895,
-      "learning_rate": 0.0001999992950402752,
-      "loss": 46.0,
-      "step": 7432
-    },
-    {
-      "epoch": 1.1970288658963726,
-      "grad_norm": 0.0011829808354377747,
-      "learning_rate": 0.00019999929485029774,
-      "loss": 46.0,
-      "step": 7433
-    },
-    {
-      "epoch": 1.19718990297516,
-      "grad_norm": 0.0009272071765735745,
-      "learning_rate": 0.00019999929466029467,
-      "loss": 46.0,
-      "step": 7434
-    },
-    {
-      "epoch": 1.1973509400539475,
-      "grad_norm": 0.0074990191496908665,
-      "learning_rate": 0.000199999294470266,
-      "loss": 46.0,
-      "step": 7435
-    },
-    {
-      "epoch": 1.1975119771327347,
-      "grad_norm": 0.0015586653025820851,
-      "learning_rate": 0.00019999929428021177,
-      "loss": 46.0,
-      "step": 7436
-    },
-    {
-      "epoch": 1.1976730142115222,
-      "grad_norm": 0.0021267388947308064,
-      "learning_rate": 0.0001999992940901319,
-      "loss": 46.0,
-      "step": 7437
-    },
-    {
-      "epoch": 1.1978340512903096,
-      "grad_norm": 0.0005111548234708607,
-      "learning_rate": 0.00019999929390002646,
-      "loss": 46.0,
-      "step": 7438
-    },
-    {
-      "epoch": 1.197995088369097,
-      "grad_norm": 0.0037622665986418724,
-      "learning_rate": 0.00019999929370989543,
-      "loss": 46.0,
-      "step": 7439
-    },
-    {
-      "epoch": 1.1981561254478843,
-      "grad_norm": 0.0017674292903393507,
-      "learning_rate": 0.0001999992935197388,
-      "loss": 46.0,
-      "step": 7440
-    },
-    {
-      "epoch": 1.1983171625266718,
-      "grad_norm": 0.001636980683542788,
-      "learning_rate": 0.00019999929332955657,
-      "loss": 46.0,
-      "step": 7441
-    },
-    {
-      "epoch": 1.1984781996054592,
-      "grad_norm": 0.0010594683699309826,
-      "learning_rate": 0.00019999929313934875,
-      "loss": 46.0,
-      "step": 7442
-    },
-    {
-      "epoch": 1.1986392366842464,
-      "grad_norm": 0.0013347334461286664,
-      "learning_rate": 0.00019999929294911532,
-      "loss": 46.0,
-      "step": 7443
-    },
-    {
-      "epoch": 1.1988002737630339,
-      "grad_norm": 0.004444236867129803,
-      "learning_rate": 0.00019999929275885632,
-      "loss": 46.0,
-      "step": 7444
-    },
-    {
-      "epoch": 1.1989613108418213,
-      "grad_norm": 0.0015611716080456972,
-      "learning_rate": 0.00019999929256857174,
-      "loss": 46.0,
-      "step": 7445
-    },
-    {
-      "epoch": 1.1991223479206088,
-      "grad_norm": 0.0004436145245563239,
-      "learning_rate": 0.00019999929237826152,
-      "loss": 46.0,
-      "step": 7446
-    },
-    {
-      "epoch": 1.199283384999396,
-      "grad_norm": 0.004913871642202139,
-      "learning_rate": 0.00019999929218792573,
-      "loss": 46.0,
-      "step": 7447
-    },
-    {
-      "epoch": 1.1994444220781835,
-      "grad_norm": 0.0017446422716602683,
-      "learning_rate": 0.00019999929199756436,
-      "loss": 46.0,
-      "step": 7448
-    },
-    {
-      "epoch": 1.199605459156971,
-      "grad_norm": 0.0016749318456277251,
-      "learning_rate": 0.00019999929180717735,
-      "loss": 46.0,
-      "step": 7449
-    },
-    {
-      "epoch": 1.1997664962357584,
-      "grad_norm": 0.0005602969904430211,
-      "learning_rate": 0.00019999929161676477,
-      "loss": 46.0,
-      "step": 7450
-    },
-    {
-      "epoch": 1.1999275333145456,
-      "grad_norm": 0.0004954435862600803,
-      "learning_rate": 0.00019999929142632661,
-      "loss": 46.0,
-      "step": 7451
-    },
-    {
-      "epoch": 1.200088570393333,
-      "grad_norm": 0.0007596751675009727,
-      "learning_rate": 0.00019999929123586284,
-      "loss": 46.0,
-      "step": 7452
-    },
-    {
-      "epoch": 1.2002496074721205,
-      "grad_norm": 0.001637642621062696,
-      "learning_rate": 0.00019999929104537348,
-      "loss": 46.0,
-      "step": 7453
-    },
-    {
-      "epoch": 1.200410644550908,
-      "grad_norm": 0.0042334916070103645,
-      "learning_rate": 0.0001999992908548585,
-      "loss": 46.0,
-      "step": 7454
-    },
-    {
-      "epoch": 1.2005716816296952,
-      "grad_norm": 0.0008722651982679963,
-      "learning_rate": 0.00019999929066431794,
-      "loss": 46.0,
-      "step": 7455
-    },
-    {
-      "epoch": 1.2007327187084826,
-      "grad_norm": 0.0017842648085206747,
-      "learning_rate": 0.00019999929047375182,
-      "loss": 46.0,
-      "step": 7456
-    },
-    {
-      "epoch": 1.20089375578727,
-      "grad_norm": 0.002939926227554679,
-      "learning_rate": 0.00019999929028316008,
-      "loss": 46.0,
-      "step": 7457
-    },
-    {
-      "epoch": 1.2010547928660573,
-      "grad_norm": 0.001383136142976582,
-      "learning_rate": 0.00019999929009254275,
-      "loss": 46.0,
-      "step": 7458
-    },
-    {
-      "epoch": 1.2012158299448448,
-      "grad_norm": 0.0012861231807619333,
-      "learning_rate": 0.00019999928990189982,
-      "loss": 46.0,
-      "step": 7459
-    },
-    {
-      "epoch": 1.2013768670236322,
-      "grad_norm": 0.0017660737503319979,
-      "learning_rate": 0.00019999928971123126,
-      "loss": 46.0,
-      "step": 7460
-    },
-    {
-      "epoch": 1.2015379041024197,
-      "grad_norm": 0.0038430211134254932,
-      "learning_rate": 0.00019999928952053715,
-      "loss": 46.0,
-      "step": 7461
-    },
-    {
-      "epoch": 1.201698941181207,
-      "grad_norm": 0.001807059976272285,
-      "learning_rate": 0.00019999928932981745,
-      "loss": 46.0,
-      "step": 7462
-    },
-    {
-      "epoch": 1.2018599782599944,
-      "grad_norm": 0.0013017336605116725,
-      "learning_rate": 0.00019999928913907214,
-      "loss": 46.0,
-      "step": 7463
-    },
-    {
-      "epoch": 1.2020210153387818,
-      "grad_norm": 0.0018817775417119265,
-      "learning_rate": 0.00019999928894830123,
-      "loss": 46.0,
-      "step": 7464
-    },
-    {
-      "epoch": 1.202182052417569,
-      "grad_norm": 0.0006017991690896451,
-      "learning_rate": 0.00019999928875750474,
-      "loss": 46.0,
-      "step": 7465
-    },
-    {
-      "epoch": 1.2023430894963565,
-      "grad_norm": 0.0017661134479567409,
-      "learning_rate": 0.00019999928856668264,
-      "loss": 46.0,
-      "step": 7466
-    },
-    {
-      "epoch": 1.202504126575144,
-      "grad_norm": 0.0006214601453393698,
-      "learning_rate": 0.00019999928837583495,
-      "loss": 46.0,
-      "step": 7467
-    },
-    {
-      "epoch": 1.2026651636539314,
-      "grad_norm": 0.0011267195222899318,
-      "learning_rate": 0.00019999928818496167,
-      "loss": 46.0,
-      "step": 7468
-    },
-    {
-      "epoch": 1.2028262007327186,
-      "grad_norm": 0.0017872502794489264,
-      "learning_rate": 0.00019999928799406278,
-      "loss": 46.0,
-      "step": 7469
-    },
-    {
-      "epoch": 1.202987237811506,
-      "grad_norm": 0.0035053330939263105,
-      "learning_rate": 0.00019999928780313833,
-      "loss": 46.0,
-      "step": 7470
-    },
-    {
-      "epoch": 1.2031482748902935,
-      "grad_norm": 0.0019433449488133192,
-      "learning_rate": 0.00019999928761218824,
-      "loss": 46.0,
-      "step": 7471
-    },
-    {
-      "epoch": 1.203309311969081,
-      "grad_norm": 0.012096031568944454,
-      "learning_rate": 0.00019999928742121258,
-      "loss": 46.0,
-      "step": 7472
-    },
-    {
-      "epoch": 1.2034703490478682,
-      "grad_norm": 0.002600267995148897,
-      "learning_rate": 0.00019999928723021132,
-      "loss": 46.0,
-      "step": 7473
-    },
-    {
-      "epoch": 1.2036313861266557,
-      "grad_norm": 0.003844389459118247,
-      "learning_rate": 0.00019999928703918446,
-      "loss": 46.0,
-      "step": 7474
-    },
-    {
-      "epoch": 1.203792423205443,
-      "grad_norm": 0.0008677352452650666,
-      "learning_rate": 0.00019999928684813205,
-      "loss": 46.0,
-      "step": 7475
-    },
-    {
-      "epoch": 1.2039534602842306,
-      "grad_norm": 0.0035268752835690975,
-      "learning_rate": 0.00019999928665705402,
-      "loss": 46.0,
-      "step": 7476
-    },
-    {
-      "epoch": 1.2041144973630178,
-      "grad_norm": 0.0038812810089439154,
-      "learning_rate": 0.00019999928646595037,
-      "loss": 46.0,
-      "step": 7477
-    },
-    {
-      "epoch": 1.2042755344418052,
-      "grad_norm": 0.004013581667095423,
-      "learning_rate": 0.00019999928627482114,
-      "loss": 46.0,
-      "step": 7478
-    },
-    {
-      "epoch": 1.2044365715205927,
-      "grad_norm": 0.0026490387972444296,
-      "learning_rate": 0.00019999928608366633,
-      "loss": 46.0,
-      "step": 7479
-    },
-    {
-      "epoch": 1.20459760859938,
-      "grad_norm": 0.003590575885027647,
-      "learning_rate": 0.0001999992858924859,
-      "loss": 46.0,
-      "step": 7480
-    },
-    {
-      "epoch": 1.2047586456781674,
-      "grad_norm": 0.0010156849166378379,
-      "learning_rate": 0.00019999928570127988,
-      "loss": 46.0,
-      "step": 7481
-    },
-    {
-      "epoch": 1.2049196827569548,
-      "grad_norm": 0.0028317461255937815,
-      "learning_rate": 0.00019999928551004827,
-      "loss": 46.0,
-      "step": 7482
-    },
-    {
-      "epoch": 1.2050807198357423,
-      "grad_norm": 0.001909225364215672,
-      "learning_rate": 0.00019999928531879108,
-      "loss": 46.0,
-      "step": 7483
-    },
-    {
-      "epoch": 1.2052417569145295,
-      "grad_norm": 0.0010485837701708078,
-      "learning_rate": 0.00019999928512750827,
-      "loss": 46.0,
-      "step": 7484
-    },
-    {
-      "epoch": 1.205402793993317,
-      "grad_norm": 0.0007255824166350067,
-      "learning_rate": 0.00019999928493619987,
-      "loss": 46.0,
-      "step": 7485
-    },
-    {
-      "epoch": 1.2055638310721044,
-      "grad_norm": 0.0006895649712532759,
-      "learning_rate": 0.00019999928474486586,
-      "loss": 46.0,
-      "step": 7486
-    },
-    {
-      "epoch": 1.2057248681508916,
-      "grad_norm": 0.000973891990724951,
-      "learning_rate": 0.00019999928455350632,
-      "loss": 46.0,
-      "step": 7487
-    },
-    {
-      "epoch": 1.205885905229679,
-      "grad_norm": 0.0014355878811329603,
-      "learning_rate": 0.0001999992843621211,
-      "loss": 46.0,
-      "step": 7488
-    },
-    {
-      "epoch": 1.2060469423084665,
-      "grad_norm": 0.0020541890989989042,
-      "learning_rate": 0.00019999928417071034,
-      "loss": 46.0,
-      "step": 7489
-    },
-    {
-      "epoch": 1.206207979387254,
-      "grad_norm": 0.0005340231582522392,
-      "learning_rate": 0.00019999928397927398,
-      "loss": 46.0,
-      "step": 7490
-    },
-    {
-      "epoch": 1.2063690164660412,
-      "grad_norm": 0.005234366282820702,
-      "learning_rate": 0.000199999283787812,
-      "loss": 46.0,
-      "step": 7491
-    },
-    {
-      "epoch": 1.2065300535448287,
-      "grad_norm": 0.0008825716795399785,
-      "learning_rate": 0.00019999928359632448,
-      "loss": 46.0,
-      "step": 7492
-    },
-    {
-      "epoch": 1.2066910906236161,
-      "grad_norm": 0.0014141829451546073,
-      "learning_rate": 0.0001999992834048113,
-      "loss": 46.0,
-      "step": 7493
-    },
-    {
-      "epoch": 1.2068521277024034,
-      "grad_norm": 0.0009230491705238819,
-      "learning_rate": 0.00019999928321327257,
-      "loss": 46.0,
-      "step": 7494
-    },
-    {
-      "epoch": 1.2070131647811908,
-      "grad_norm": 0.0006315942155197263,
-      "learning_rate": 0.00019999928302170825,
-      "loss": 46.0,
-      "step": 7495
-    },
-    {
-      "epoch": 1.2071742018599783,
-      "grad_norm": 0.0007494970341213048,
-      "learning_rate": 0.00019999928283011831,
-      "loss": 46.0,
-      "step": 7496
-    },
-    {
-      "epoch": 1.2073352389387657,
-      "grad_norm": 0.009130037389695644,
-      "learning_rate": 0.0001999992826385028,
-      "loss": 46.0,
-      "step": 7497
-    },
-    {
-      "epoch": 1.2074962760175532,
-      "grad_norm": 0.0030808583833277225,
-      "learning_rate": 0.00019999928244686163,
-      "loss": 46.0,
-      "step": 7498
-    },
-    {
-      "epoch": 1.2076573130963404,
-      "grad_norm": 0.0030400161631405354,
-      "learning_rate": 0.00019999928225519493,
-      "loss": 46.0,
-      "step": 7499
-    },
-    {
-      "epoch": 1.2078183501751278,
-      "grad_norm": 0.000541490619070828,
-      "learning_rate": 0.0001999992820635026,
-      "loss": 46.0,
-      "step": 7500
-    },
-    {
-      "epoch": 1.2079793872539153,
-      "grad_norm": 0.002253288635984063,
-      "learning_rate": 0.00019999928187178472,
-      "loss": 46.0,
-      "step": 7501
-    },
-    {
-      "epoch": 1.2081404243327025,
-      "grad_norm": 0.0013256544480100274,
-      "learning_rate": 0.0001999992816800412,
-      "loss": 46.0,
-      "step": 7502
-    },
-    {
-      "epoch": 1.20830146141149,
-      "grad_norm": 0.0010246033780276775,
-      "learning_rate": 0.0001999992814882721,
-      "loss": 46.0,
-      "step": 7503
-    },
-    {
-      "epoch": 1.2084624984902774,
-      "grad_norm": 0.005492016673088074,
-      "learning_rate": 0.0001999992812964774,
-      "loss": 46.0,
-      "step": 7504
-    },
-    {
-      "epoch": 1.2086235355690649,
-      "grad_norm": 0.006982710212469101,
-      "learning_rate": 0.00019999928110465712,
-      "loss": 46.0,
-      "step": 7505
-    },
-    {
-      "epoch": 1.208784572647852,
-      "grad_norm": 0.0042646704241633415,
-      "learning_rate": 0.00019999928091281124,
-      "loss": 46.0,
-      "step": 7506
-    },
-    {
-      "epoch": 1.2089456097266396,
-      "grad_norm": 0.0019437269074842334,
-      "learning_rate": 0.0001999992807209398,
-      "loss": 46.0,
-      "step": 7507
-    },
-    {
-      "epoch": 1.209106646805427,
-      "grad_norm": 0.003257865086197853,
-      "learning_rate": 0.0001999992805290427,
-      "loss": 46.0,
-      "step": 7508
-    },
-    {
-      "epoch": 1.2092676838842142,
-      "grad_norm": 0.008707267232239246,
-      "learning_rate": 0.00019999928033712005,
-      "loss": 46.0,
-      "step": 7509
-    },
-    {
-      "epoch": 1.2094287209630017,
-      "grad_norm": 0.0007214149227365851,
-      "learning_rate": 0.00019999928014517178,
-      "loss": 46.0,
-      "step": 7510
-    },
-    {
-      "epoch": 1.2095897580417891,
-      "grad_norm": 0.004151150584220886,
-      "learning_rate": 0.0001999992799531979,
-      "loss": 46.0,
-      "step": 7511
-    },
-    {
-      "epoch": 1.2097507951205766,
-      "grad_norm": 0.0015474956016987562,
-      "learning_rate": 0.00019999927976119847,
-      "loss": 46.0,
-      "step": 7512
-    },
-    {
-      "epoch": 1.2099118321993638,
-      "grad_norm": 0.0008899067179299891,
-      "learning_rate": 0.00019999927956917342,
-      "loss": 46.0,
-      "step": 7513
-    },
-    {
-      "epoch": 1.2100728692781513,
-      "grad_norm": 0.0036538224667310715,
-      "learning_rate": 0.0001999992793771228,
-      "loss": 46.0,
-      "step": 7514
-    },
-    {
-      "epoch": 1.2102339063569387,
-      "grad_norm": 0.0029800725169479847,
-      "learning_rate": 0.00019999927918504655,
-      "loss": 46.0,
-      "step": 7515
-    },
-    {
-      "epoch": 1.210394943435726,
-      "grad_norm": 0.002335120690986514,
-      "learning_rate": 0.00019999927899294473,
-      "loss": 46.0,
-      "step": 7516
-    },
-    {
-      "epoch": 1.2105559805145134,
-      "grad_norm": 0.0017036811914294958,
-      "learning_rate": 0.00019999927880081733,
-      "loss": 46.0,
-      "step": 7517
-    },
-    {
-      "epoch": 1.2107170175933009,
-      "grad_norm": 0.0012495525879785419,
-      "learning_rate": 0.00019999927860866426,
-      "loss": 46.0,
-      "step": 7518
-    },
-    {
-      "epoch": 1.2108780546720883,
-      "grad_norm": 0.0018980697495862842,
-      "learning_rate": 0.0001999992784164857,
-      "loss": 46.0,
-      "step": 7519
-    },
-    {
-      "epoch": 1.2110390917508758,
-      "grad_norm": 0.0017072854097932577,
-      "learning_rate": 0.00019999927822428147,
-      "loss": 46.0,
-      "step": 7520
-    },
-    {
-      "epoch": 1.211200128829663,
-      "grad_norm": 0.0014214892871677876,
-      "learning_rate": 0.00019999927803205167,
-      "loss": 46.0,
-      "step": 7521
-    },
-    {
-      "epoch": 1.2113611659084504,
-      "grad_norm": 0.011226499453186989,
-      "learning_rate": 0.00019999927783979628,
-      "loss": 46.0,
-      "step": 7522
-    },
-    {
-      "epoch": 1.211522202987238,
-      "grad_norm": 0.00080280308611691,
-      "learning_rate": 0.0001999992776475153,
-      "loss": 46.0,
-      "step": 7523
-    },
-    {
-      "epoch": 1.2116832400660251,
-      "grad_norm": 0.0008106543682515621,
-      "learning_rate": 0.0001999992774552087,
-      "loss": 46.0,
-      "step": 7524
-    },
-    {
-      "epoch": 1.2118442771448126,
-      "grad_norm": 0.0005572764202952385,
-      "learning_rate": 0.00019999927726287652,
-      "loss": 46.0,
-      "step": 7525
-    },
-    {
-      "epoch": 1.2120053142236,
-      "grad_norm": 0.0012713458854705095,
-      "learning_rate": 0.00019999927707051876,
-      "loss": 46.0,
-      "step": 7526
-    },
-    {
-      "epoch": 1.2121663513023875,
-      "grad_norm": 0.0009874645620584488,
-      "learning_rate": 0.00019999927687813538,
-      "loss": 46.0,
-      "step": 7527
-    },
-    {
-      "epoch": 1.2123273883811747,
-      "grad_norm": 0.0040589929558336735,
-      "learning_rate": 0.00019999927668572643,
-      "loss": 46.0,
-      "step": 7528
-    },
-    {
-      "epoch": 1.2124884254599622,
-      "grad_norm": 0.0011651463573798537,
-      "learning_rate": 0.00019999927649329185,
-      "loss": 46.0,
-      "step": 7529
-    },
-    {
-      "epoch": 1.2126494625387496,
-      "grad_norm": 0.0027236214373260736,
-      "learning_rate": 0.00019999927630083168,
-      "loss": 46.0,
-      "step": 7530
-    },
-    {
-      "epoch": 1.2128104996175368,
-      "grad_norm": 0.0005804926622658968,
-      "learning_rate": 0.00019999927610834595,
-      "loss": 46.0,
-      "step": 7531
-    },
-    {
-      "epoch": 1.2129715366963243,
-      "grad_norm": 0.0013138545909896493,
-      "learning_rate": 0.0001999992759158346,
-      "loss": 46.0,
-      "step": 7532
-    },
-    {
-      "epoch": 1.2131325737751117,
-      "grad_norm": 0.0020083794370293617,
-      "learning_rate": 0.00019999927572329768,
-      "loss": 46.0,
-      "step": 7533
-    },
-    {
-      "epoch": 1.2132936108538992,
-      "grad_norm": 0.001530150417238474,
-      "learning_rate": 0.00019999927553073513,
-      "loss": 46.0,
-      "step": 7534
-    },
-    {
-      "epoch": 1.2134546479326864,
-      "grad_norm": 0.0006744688143953681,
-      "learning_rate": 0.000199999275338147,
-      "loss": 46.0,
-      "step": 7535
-    },
-    {
-      "epoch": 1.2136156850114739,
-      "grad_norm": 0.0009230764699168503,
-      "learning_rate": 0.0001999992751455333,
-      "loss": 46.0,
-      "step": 7536
-    },
-    {
-      "epoch": 1.2137767220902613,
-      "grad_norm": 0.005693175829946995,
-      "learning_rate": 0.00019999927495289397,
-      "loss": 46.0,
-      "step": 7537
-    },
-    {
-      "epoch": 1.2139377591690486,
-      "grad_norm": 0.004383837804198265,
-      "learning_rate": 0.00019999927476022905,
-      "loss": 46.0,
-      "step": 7538
-    },
-    {
-      "epoch": 1.214098796247836,
-      "grad_norm": 0.001445006811991334,
-      "learning_rate": 0.00019999927456753854,
-      "loss": 46.0,
-      "step": 7539
-    },
-    {
-      "epoch": 1.2142598333266235,
-      "grad_norm": 0.0013964000390842557,
-      "learning_rate": 0.00019999927437482247,
-      "loss": 46.0,
-      "step": 7540
-    },
-    {
-      "epoch": 1.214420870405411,
-      "grad_norm": 0.0019224357092753053,
-      "learning_rate": 0.00019999927418208076,
-      "loss": 46.0,
-      "step": 7541
-    },
-    {
-      "epoch": 1.2145819074841981,
-      "grad_norm": 0.0012270859442651272,
-      "learning_rate": 0.00019999927398931346,
-      "loss": 46.0,
-      "step": 7542
-    },
-    {
-      "epoch": 1.2147429445629856,
-      "grad_norm": 0.0035568969324231148,
-      "learning_rate": 0.00019999927379652058,
-      "loss": 46.0,
-      "step": 7543
-    },
-    {
-      "epoch": 1.214903981641773,
-      "grad_norm": 0.004302465356886387,
-      "learning_rate": 0.0001999992736037021,
-      "loss": 46.0,
-      "step": 7544
-    },
-    {
-      "epoch": 1.2150650187205605,
-      "grad_norm": 0.0010684158187359571,
-      "learning_rate": 0.00019999927341085805,
-      "loss": 46.0,
-      "step": 7545
-    },
-    {
-      "epoch": 1.2152260557993477,
-      "grad_norm": 0.009326261468231678,
-      "learning_rate": 0.00019999927321798837,
-      "loss": 46.0,
-      "step": 7546
-    },
-    {
-      "epoch": 1.2153870928781352,
-      "grad_norm": 0.0024467259645462036,
-      "learning_rate": 0.00019999927302509311,
-      "loss": 46.0,
-      "step": 7547
-    },
-    {
-      "epoch": 1.2155481299569226,
-      "grad_norm": 0.0045942021533846855,
-      "learning_rate": 0.00019999927283217224,
-      "loss": 46.0,
-      "step": 7548
-    },
-    {
-      "epoch": 1.21570916703571,
-      "grad_norm": 0.004820413421839476,
-      "learning_rate": 0.00019999927263922578,
-      "loss": 46.0,
-      "step": 7549
-    },
-    {
-      "epoch": 1.2158702041144973,
-      "grad_norm": 0.001074828440323472,
-      "learning_rate": 0.00019999927244625373,
-      "loss": 46.0,
-      "step": 7550
-    },
-    {
-      "epoch": 1.2160312411932848,
-      "grad_norm": 0.003381298389285803,
-      "learning_rate": 0.0001999992722532561,
-      "loss": 46.0,
-      "step": 7551
-    },
-    {
-      "epoch": 1.2161922782720722,
-      "grad_norm": 0.010321678593754768,
-      "learning_rate": 0.00019999927206023287,
-      "loss": 46.0,
-      "step": 7552
-    },
-    {
-      "epoch": 1.2163533153508594,
-      "grad_norm": 0.0007929296116344631,
-      "learning_rate": 0.00019999927186718403,
-      "loss": 46.0,
-      "step": 7553
-    },
-    {
-      "epoch": 1.216514352429647,
-      "grad_norm": 0.006196086760610342,
-      "learning_rate": 0.0001999992716741096,
-      "loss": 46.0,
-      "step": 7554
-    },
-    {
-      "epoch": 1.2166753895084343,
-      "grad_norm": 0.0011012848699465394,
-      "learning_rate": 0.0001999992714810096,
-      "loss": 46.0,
-      "step": 7555
-    },
-    {
-      "epoch": 1.2168364265872218,
-      "grad_norm": 0.0007859840989112854,
-      "learning_rate": 0.00019999927128788397,
-      "loss": 46.0,
-      "step": 7556
-    },
-    {
-      "epoch": 1.216997463666009,
-      "grad_norm": 0.005725369323045015,
-      "learning_rate": 0.00019999927109473276,
-      "loss": 46.0,
-      "step": 7557
-    },
-    {
-      "epoch": 1.2171585007447965,
-      "grad_norm": 0.0012831080239266157,
-      "learning_rate": 0.00019999927090155596,
-      "loss": 46.0,
-      "step": 7558
-    },
-    {
-      "epoch": 1.217319537823584,
-      "grad_norm": 0.0032234350219368935,
-      "learning_rate": 0.00019999927070835357,
-      "loss": 46.0,
-      "step": 7559
-    },
-    {
-      "epoch": 1.2174805749023712,
-      "grad_norm": 0.0037518665194511414,
-      "learning_rate": 0.00019999927051512554,
-      "loss": 46.0,
-      "step": 7560
-    },
-    {
-      "epoch": 1.2176416119811586,
-      "grad_norm": 0.00048648836673237383,
-      "learning_rate": 0.00019999927032187198,
-      "loss": 46.0,
-      "step": 7561
-    },
-    {
-      "epoch": 1.217802649059946,
-      "grad_norm": 0.008244657889008522,
-      "learning_rate": 0.00019999927012859277,
-      "loss": 46.0,
-      "step": 7562
-    },
-    {
-      "epoch": 1.2179636861387335,
-      "grad_norm": 0.002935724565759301,
-      "learning_rate": 0.000199999269935288,
-      "loss": 46.0,
-      "step": 7563
-    },
-    {
-      "epoch": 1.2181247232175207,
-      "grad_norm": 0.0009815345983952284,
-      "learning_rate": 0.00019999926974195763,
-      "loss": 46.0,
-      "step": 7564
-    },
-    {
-      "epoch": 1.2182857602963082,
-      "grad_norm": 0.0011572600342333317,
-      "learning_rate": 0.00019999926954860167,
-      "loss": 46.0,
-      "step": 7565
-    },
-    {
-      "epoch": 1.2184467973750956,
-      "grad_norm": 0.0006085620843805373,
-      "learning_rate": 0.00019999926935522012,
-      "loss": 46.0,
-      "step": 7566
-    },
-    {
-      "epoch": 1.218607834453883,
-      "grad_norm": 0.010725236497819424,
-      "learning_rate": 0.00019999926916181295,
-      "loss": 46.0,
-      "step": 7567
-    },
-    {
-      "epoch": 1.2187688715326703,
-      "grad_norm": 0.0013549436116591096,
-      "learning_rate": 0.0001999992689683802,
-      "loss": 46.0,
-      "step": 7568
-    },
-    {
-      "epoch": 1.2189299086114578,
-      "grad_norm": 0.0008937516831792891,
-      "learning_rate": 0.00019999926877492185,
-      "loss": 46.0,
-      "step": 7569
-    },
-    {
-      "epoch": 1.2190909456902452,
-      "grad_norm": 0.0013728990452364087,
-      "learning_rate": 0.00019999926858143793,
-      "loss": 46.0,
-      "step": 7570
-    },
-    {
-      "epoch": 1.2192519827690327,
-      "grad_norm": 0.0009747871663421392,
-      "learning_rate": 0.00019999926838792838,
-      "loss": 46.0,
-      "step": 7571
-    },
-    {
-      "epoch": 1.21941301984782,
-      "grad_norm": 0.0011147890472784638,
-      "learning_rate": 0.00019999926819439326,
-      "loss": 46.0,
-      "step": 7572
-    },
-    {
-      "epoch": 1.2195740569266074,
-      "grad_norm": 0.0007510284194722772,
-      "learning_rate": 0.0001999992680008325,
-      "loss": 46.0,
-      "step": 7573
-    },
-    {
-      "epoch": 1.2197350940053948,
-      "grad_norm": 0.0020506775472313166,
-      "learning_rate": 0.0001999992678072462,
-      "loss": 46.0,
-      "step": 7574
-    },
-    {
-      "epoch": 1.219896131084182,
-      "grad_norm": 0.0007354197441600263,
-      "learning_rate": 0.0001999992676136343,
-      "loss": 46.0,
-      "step": 7575
-    },
-    {
-      "epoch": 1.2200571681629695,
-      "grad_norm": 0.0015805386938154697,
-      "learning_rate": 0.00019999926741999679,
-      "loss": 46.0,
-      "step": 7576
-    },
-    {
-      "epoch": 1.220218205241757,
-      "grad_norm": 0.002647318411618471,
-      "learning_rate": 0.0001999992672263337,
-      "loss": 46.0,
-      "step": 7577
-    },
-    {
-      "epoch": 1.2203792423205444,
-      "grad_norm": 0.0003503689949866384,
-      "learning_rate": 0.00019999926703264499,
-      "loss": 46.0,
-      "step": 7578
-    },
-    {
-      "epoch": 1.2205402793993316,
-      "grad_norm": 0.0007300485740415752,
-      "learning_rate": 0.0001999992668389307,
-      "loss": 46.0,
-      "step": 7579
-    },
-    {
-      "epoch": 1.220701316478119,
-      "grad_norm": 0.009703694842755795,
-      "learning_rate": 0.00019999926664519078,
-      "loss": 46.0,
-      "step": 7580
-    },
-    {
-      "epoch": 1.2208623535569065,
-      "grad_norm": 0.0012521852040663362,
-      "learning_rate": 0.00019999926645142532,
-      "loss": 46.0,
-      "step": 7581
-    },
-    {
-      "epoch": 1.2210233906356938,
-      "grad_norm": 0.00207143509760499,
-      "learning_rate": 0.00019999926625763426,
-      "loss": 46.0,
-      "step": 7582
-    },
-    {
-      "epoch": 1.2211844277144812,
-      "grad_norm": 0.00037611162406392395,
-      "learning_rate": 0.00019999926606381756,
-      "loss": 46.0,
-      "step": 7583
-    },
-    {
-      "epoch": 1.2213454647932687,
-      "grad_norm": 0.0014601427828893065,
-      "learning_rate": 0.0001999992658699753,
-      "loss": 46.0,
-      "step": 7584
-    },
-    {
-      "epoch": 1.2215065018720561,
-      "grad_norm": 0.0020650255028158426,
-      "learning_rate": 0.00019999926567610744,
-      "loss": 46.0,
-      "step": 7585
-    },
-    {
-      "epoch": 1.2216675389508433,
-      "grad_norm": 0.0004501556686591357,
-      "learning_rate": 0.000199999265482214,
-      "loss": 46.0,
-      "step": 7586
-    },
-    {
-      "epoch": 1.2218285760296308,
-      "grad_norm": 0.0039031270425766706,
-      "learning_rate": 0.00019999926528829493,
-      "loss": 46.0,
-      "step": 7587
-    },
-    {
-      "epoch": 1.2219896131084182,
-      "grad_norm": 0.006299312692135572,
-      "learning_rate": 0.0001999992650943503,
-      "loss": 46.0,
-      "step": 7588
-    },
-    {
-      "epoch": 1.2221506501872057,
-      "grad_norm": 0.0011198463616892695,
-      "learning_rate": 0.00019999926490038003,
-      "loss": 46.0,
-      "step": 7589
-    },
-    {
-      "epoch": 1.222311687265993,
-      "grad_norm": 0.0013702830765396357,
-      "learning_rate": 0.00019999926470638422,
-      "loss": 46.0,
-      "step": 7590
-    },
-    {
-      "epoch": 1.2224727243447804,
-      "grad_norm": 0.0017298564780503511,
-      "learning_rate": 0.00019999926451236277,
-      "loss": 46.0,
-      "step": 7591
-    },
-    {
-      "epoch": 1.2226337614235678,
-      "grad_norm": 0.0037376952823251486,
-      "learning_rate": 0.00019999926431831576,
-      "loss": 46.0,
-      "step": 7592
-    },
-    {
-      "epoch": 1.2227947985023553,
-      "grad_norm": 0.001607323531061411,
-      "learning_rate": 0.00019999926412424314,
-      "loss": 46.0,
-      "step": 7593
-    },
-    {
-      "epoch": 1.2229558355811425,
-      "grad_norm": 0.0032464477699249983,
-      "learning_rate": 0.00019999926393014493,
-      "loss": 46.0,
-      "step": 7594
-    },
-    {
-      "epoch": 1.22311687265993,
-      "grad_norm": 0.007269986905157566,
-      "learning_rate": 0.0001999992637360211,
-      "loss": 46.0,
-      "step": 7595
-    },
-    {
-      "epoch": 1.2232779097387174,
-      "grad_norm": 0.0024941288866102695,
-      "learning_rate": 0.0001999992635418717,
-      "loss": 46.0,
-      "step": 7596
-    },
-    {
-      "epoch": 1.2234389468175046,
-      "grad_norm": 0.0009921793825924397,
-      "learning_rate": 0.00019999926334769672,
-      "loss": 46.0,
-      "step": 7597
-    },
-    {
-      "epoch": 1.223599983896292,
-      "grad_norm": 0.0012252703309059143,
-      "learning_rate": 0.00019999926315349614,
-      "loss": 46.0,
-      "step": 7598
-    },
-    {
-      "epoch": 1.2237610209750795,
-      "grad_norm": 0.0026946994476020336,
-      "learning_rate": 0.00019999926295926994,
-      "loss": 46.0,
-      "step": 7599
-    },
-    {
-      "epoch": 1.223922058053867,
-      "grad_norm": 0.001200696686282754,
-      "learning_rate": 0.00019999926276501815,
-      "loss": 46.0,
-      "step": 7600
-    },
-    {
-      "epoch": 1.2240830951326542,
-      "grad_norm": 0.0002305166271980852,
-      "learning_rate": 0.0001999992625707408,
-      "loss": 46.0,
-      "step": 7601
-    },
-    {
-      "epoch": 1.2242441322114417,
-      "grad_norm": 0.0006880192668177187,
-      "learning_rate": 0.00019999926237643782,
-      "loss": 46.0,
-      "step": 7602
-    },
-    {
-      "epoch": 1.2244051692902291,
-      "grad_norm": 0.0009748409502208233,
-      "learning_rate": 0.00019999926218210924,
-      "loss": 46.0,
-      "step": 7603
-    },
-    {
-      "epoch": 1.2245662063690164,
-      "grad_norm": 0.005277007352560759,
-      "learning_rate": 0.0001999992619877551,
-      "loss": 46.0,
-      "step": 7604
-    },
-    {
-      "epoch": 1.2247272434478038,
-      "grad_norm": 0.0005417760694399476,
-      "learning_rate": 0.00019999926179337533,
-      "loss": 46.0,
-      "step": 7605
-    },
-    {
-      "epoch": 1.2248882805265913,
-      "grad_norm": 0.0014984805602580309,
-      "learning_rate": 0.00019999926159897,
-      "loss": 46.0,
-      "step": 7606
-    },
-    {
-      "epoch": 1.2250493176053787,
-      "grad_norm": 0.000900235609151423,
-      "learning_rate": 0.00019999926140453904,
-      "loss": 46.0,
-      "step": 7607
-    },
-    {
-      "epoch": 1.225210354684166,
-      "grad_norm": 0.0006910812226124108,
-      "learning_rate": 0.0001999992612100825,
-      "loss": 46.0,
-      "step": 7608
-    },
-    {
-      "epoch": 1.2253713917629534,
-      "grad_norm": 0.0021148917730897665,
-      "learning_rate": 0.00019999926101560038,
-      "loss": 46.0,
-      "step": 7609
-    },
-    {
-      "epoch": 1.2255324288417409,
-      "grad_norm": 0.000786135729867965,
-      "learning_rate": 0.00019999926082109264,
-      "loss": 46.0,
-      "step": 7610
-    },
-    {
-      "epoch": 1.225693465920528,
-      "grad_norm": 0.0008698116871528327,
-      "learning_rate": 0.0001999992606265593,
-      "loss": 46.0,
-      "step": 7611
-    },
-    {
-      "epoch": 1.2258545029993155,
-      "grad_norm": 0.0007346338243223727,
-      "learning_rate": 0.00019999926043200042,
-      "loss": 46.0,
-      "step": 7612
-    },
-    {
-      "epoch": 1.226015540078103,
-      "grad_norm": 0.001855868031270802,
-      "learning_rate": 0.0001999992602374159,
-      "loss": 46.0,
-      "step": 7613
-    },
-    {
-      "epoch": 1.2261765771568904,
-      "grad_norm": 0.0010466381208971143,
-      "learning_rate": 0.00019999926004280578,
-      "loss": 46.0,
-      "step": 7614
-    },
-    {
-      "epoch": 1.2263376142356779,
-      "grad_norm": 0.0024692926090210676,
-      "learning_rate": 0.0001999992598481701,
-      "loss": 46.0,
-      "step": 7615
-    },
-    {
-      "epoch": 1.2264986513144651,
-      "grad_norm": 0.002758033573627472,
-      "learning_rate": 0.00019999925965350878,
-      "loss": 46.0,
-      "step": 7616
-    },
-    {
-      "epoch": 1.2266596883932526,
-      "grad_norm": 0.004525685682892799,
-      "learning_rate": 0.0001999992594588219,
-      "loss": 46.0,
-      "step": 7617
-    },
-    {
-      "epoch": 1.22682072547204,
-      "grad_norm": 0.003646411467343569,
-      "learning_rate": 0.00019999925926410942,
-      "loss": 46.0,
-      "step": 7618
-    },
-    {
-      "epoch": 1.2269817625508272,
-      "grad_norm": 0.0031161773949861526,
-      "learning_rate": 0.00019999925906937134,
-      "loss": 46.0,
-      "step": 7619
-    },
-    {
-      "epoch": 1.2271427996296147,
-      "grad_norm": 0.006743249017745256,
-      "learning_rate": 0.00019999925887460767,
-      "loss": 46.0,
-      "step": 7620
-    },
-    {
-      "epoch": 1.2273038367084022,
-      "grad_norm": 0.0008316087769344449,
-      "learning_rate": 0.00019999925867981836,
-      "loss": 46.0,
-      "step": 7621
-    },
-    {
-      "epoch": 1.2274648737871896,
-      "grad_norm": 0.001978185959160328,
-      "learning_rate": 0.00019999925848500352,
-      "loss": 46.0,
-      "step": 7622
-    },
-    {
-      "epoch": 1.2276259108659768,
-      "grad_norm": 0.006658573169261217,
-      "learning_rate": 0.00019999925829016307,
-      "loss": 46.0,
-      "step": 7623
-    },
-    {
-      "epoch": 1.2277869479447643,
-      "grad_norm": 0.0013616629876196384,
-      "learning_rate": 0.000199999258095297,
-      "loss": 46.0,
-      "step": 7624
-    },
-    {
-      "epoch": 1.2279479850235517,
-      "grad_norm": 0.0017589150229468942,
-      "learning_rate": 0.00019999925790040534,
-      "loss": 46.0,
-      "step": 7625
-    },
-    {
-      "epoch": 1.228109022102339,
-      "grad_norm": 0.0010458810720592737,
-      "learning_rate": 0.0001999992577054881,
-      "loss": 46.0,
-      "step": 7626
-    },
-    {
-      "epoch": 1.2282700591811264,
-      "grad_norm": 0.0017738519236445427,
-      "learning_rate": 0.0001999992575105453,
-      "loss": 46.0,
-      "step": 7627
-    },
-    {
-      "epoch": 1.2284310962599139,
-      "grad_norm": 0.002235678257420659,
-      "learning_rate": 0.00019999925731557685,
-      "loss": 46.0,
-      "step": 7628
-    },
-    {
-      "epoch": 1.2285921333387013,
-      "grad_norm": 0.005524209700524807,
-      "learning_rate": 0.00019999925712058282,
-      "loss": 46.0,
-      "step": 7629
-    },
-    {
-      "epoch": 1.2287531704174885,
-      "grad_norm": 0.0012723833788186312,
-      "learning_rate": 0.00019999925692556318,
-      "loss": 46.0,
-      "step": 7630
-    },
-    {
-      "epoch": 1.228914207496276,
-      "grad_norm": 0.0008022763067856431,
-      "learning_rate": 0.00019999925673051797,
-      "loss": 46.0,
-      "step": 7631
-    },
-    {
-      "epoch": 1.2290752445750635,
-      "grad_norm": 0.001587162259966135,
-      "learning_rate": 0.00019999925653544715,
-      "loss": 46.0,
-      "step": 7632
-    },
-    {
-      "epoch": 1.2292362816538507,
-      "grad_norm": 0.0008148058550432324,
-      "learning_rate": 0.00019999925634035077,
-      "loss": 46.0,
-      "step": 7633
-    },
-    {
-      "epoch": 1.2293973187326381,
-      "grad_norm": 0.0032398866023868322,
-      "learning_rate": 0.00019999925614522877,
-      "loss": 46.0,
-      "step": 7634
-    },
-    {
-      "epoch": 1.2295583558114256,
-      "grad_norm": 0.0021040227729827166,
-      "learning_rate": 0.00019999925595008116,
-      "loss": 46.0,
-      "step": 7635
-    },
-    {
-      "epoch": 1.229719392890213,
-      "grad_norm": 0.0009233879973180592,
-      "learning_rate": 0.00019999925575490797,
-      "loss": 46.0,
-      "step": 7636
-    },
-    {
-      "epoch": 1.2298804299690003,
-      "grad_norm": 0.0011741952039301395,
-      "learning_rate": 0.00019999925555970918,
-      "loss": 46.0,
-      "step": 7637
-    },
-    {
-      "epoch": 1.2300414670477877,
-      "grad_norm": 0.0007919566705822945,
-      "learning_rate": 0.00019999925536448481,
-      "loss": 46.0,
-      "step": 7638
-    },
-    {
-      "epoch": 1.2302025041265752,
-      "grad_norm": 0.0016446413937956095,
-      "learning_rate": 0.00019999925516923486,
-      "loss": 46.0,
-      "step": 7639
-    },
-    {
-      "epoch": 1.2303635412053626,
-      "grad_norm": 0.004901389125734568,
-      "learning_rate": 0.00019999925497395928,
-      "loss": 46.0,
-      "step": 7640
-    },
-    {
-      "epoch": 1.2305245782841499,
-      "grad_norm": 0.0046926806680858135,
-      "learning_rate": 0.0001999992547786581,
-      "loss": 46.0,
-      "step": 7641
-    },
-    {
-      "epoch": 1.2306856153629373,
-      "grad_norm": 0.010192851535975933,
-      "learning_rate": 0.00019999925458333135,
-      "loss": 46.0,
-      "step": 7642
-    },
-    {
-      "epoch": 1.2308466524417248,
-      "grad_norm": 0.0009865125175565481,
-      "learning_rate": 0.000199999254387979,
-      "loss": 46.0,
-      "step": 7643
-    },
-    {
-      "epoch": 1.2310076895205122,
-      "grad_norm": 0.0004271470825187862,
-      "learning_rate": 0.00019999925419260107,
-      "loss": 46.0,
-      "step": 7644
-    },
-    {
-      "epoch": 1.2311687265992994,
-      "grad_norm": 0.0004858318134211004,
-      "learning_rate": 0.0001999992539971975,
-      "loss": 46.0,
-      "step": 7645
-    },
-    {
-      "epoch": 1.2313297636780869,
-      "grad_norm": 0.00045591805246658623,
-      "learning_rate": 0.00019999925380176836,
-      "loss": 46.0,
-      "step": 7646
-    },
-    {
-      "epoch": 1.2314908007568743,
-      "grad_norm": 0.001879748422652483,
-      "learning_rate": 0.00019999925360631365,
-      "loss": 46.0,
-      "step": 7647
-    },
-    {
-      "epoch": 1.2316518378356616,
-      "grad_norm": 0.002072389703243971,
-      "learning_rate": 0.0001999992534108333,
-      "loss": 46.0,
-      "step": 7648
-    },
-    {
-      "epoch": 1.231812874914449,
-      "grad_norm": 0.001408843556419015,
-      "learning_rate": 0.0001999992532153274,
-      "loss": 46.0,
-      "step": 7649
-    },
-    {
-      "epoch": 1.2319739119932365,
-      "grad_norm": 0.0006739210220985115,
-      "learning_rate": 0.00019999925301979586,
-      "loss": 46.0,
-      "step": 7650
-    },
-    {
-      "epoch": 1.232134949072024,
-      "grad_norm": 0.001930776285007596,
-      "learning_rate": 0.00019999925282423878,
-      "loss": 46.0,
-      "step": 7651
-    },
-    {
-      "epoch": 1.2322959861508112,
-      "grad_norm": 0.005610702559351921,
-      "learning_rate": 0.00019999925262865605,
-      "loss": 46.0,
-      "step": 7652
-    },
-    {
-      "epoch": 1.2324570232295986,
-      "grad_norm": 0.001402527210302651,
-      "learning_rate": 0.00019999925243304777,
-      "loss": 46.0,
-      "step": 7653
-    },
-    {
-      "epoch": 1.232618060308386,
-      "grad_norm": 0.0014332960126921535,
-      "learning_rate": 0.00019999925223741387,
-      "loss": 46.0,
-      "step": 7654
-    },
-    {
-      "epoch": 1.2327790973871733,
-      "grad_norm": 0.0015186292584985495,
-      "learning_rate": 0.00019999925204175438,
-      "loss": 46.0,
-      "step": 7655
-    },
-    {
-      "epoch": 1.2329401344659607,
-      "grad_norm": 0.0013117969501763582,
-      "learning_rate": 0.0001999992518460693,
-      "loss": 46.0,
-      "step": 7656
-    },
-    {
-      "epoch": 1.2331011715447482,
-      "grad_norm": 0.001633496955037117,
-      "learning_rate": 0.0001999992516503586,
-      "loss": 46.0,
-      "step": 7657
-    },
-    {
-      "epoch": 1.2332622086235356,
-      "grad_norm": 0.005964313633739948,
-      "learning_rate": 0.00019999925145462234,
-      "loss": 46.0,
-      "step": 7658
-    },
-    {
-      "epoch": 1.2334232457023229,
-      "grad_norm": 0.0020560745615512133,
-      "learning_rate": 0.00019999925125886047,
-      "loss": 46.0,
-      "step": 7659
-    },
-    {
-      "epoch": 1.2335842827811103,
-      "grad_norm": 0.0005800784565508366,
-      "learning_rate": 0.00019999925106307302,
-      "loss": 46.0,
-      "step": 7660
-    },
-    {
-      "epoch": 1.2337453198598978,
-      "grad_norm": 0.008018608205020428,
-      "learning_rate": 0.00019999925086725996,
-      "loss": 46.0,
-      "step": 7661
-    },
-    {
-      "epoch": 1.2339063569386852,
-      "grad_norm": 0.011703826487064362,
-      "learning_rate": 0.0001999992506714213,
-      "loss": 46.0,
-      "step": 7662
-    },
-    {
-      "epoch": 1.2340673940174725,
-      "grad_norm": 0.0022539463825523853,
-      "learning_rate": 0.00019999925047555706,
-      "loss": 46.0,
-      "step": 7663
-    },
-    {
-      "epoch": 1.23422843109626,
-      "grad_norm": 0.0010529665742069483,
-      "learning_rate": 0.0001999992502796672,
-      "loss": 46.0,
-      "step": 7664
-    },
-    {
-      "epoch": 1.2343894681750474,
-      "grad_norm": 0.0015553730772808194,
-      "learning_rate": 0.0001999992500837518,
-      "loss": 46.0,
-      "step": 7665
-    },
-    {
-      "epoch": 1.2345505052538348,
-      "grad_norm": 0.0028824489563703537,
-      "learning_rate": 0.00019999924988781077,
-      "loss": 46.0,
-      "step": 7666
-    },
-    {
-      "epoch": 1.234711542332622,
-      "grad_norm": 0.00441374396905303,
-      "learning_rate": 0.00019999924969184413,
-      "loss": 46.0,
-      "step": 7667
-    },
-    {
-      "epoch": 1.2348725794114095,
-      "grad_norm": 0.0005688901874236763,
-      "learning_rate": 0.00019999924949585192,
-      "loss": 46.0,
-      "step": 7668
-    },
-    {
-      "epoch": 1.235033616490197,
-      "grad_norm": 0.0012712272582575679,
-      "learning_rate": 0.0001999992492998341,
-      "loss": 46.0,
-      "step": 7669
-    },
-    {
-      "epoch": 1.2351946535689842,
-      "grad_norm": 0.003068245016038418,
-      "learning_rate": 0.00019999924910379068,
-      "loss": 46.0,
-      "step": 7670
-    },
-    {
-      "epoch": 1.2353556906477716,
-      "grad_norm": 0.0014743818901479244,
-      "learning_rate": 0.00019999924890772168,
-      "loss": 46.0,
-      "step": 7671
-    },
-    {
-      "epoch": 1.235516727726559,
-      "grad_norm": 0.002503872150555253,
-      "learning_rate": 0.00019999924871162708,
-      "loss": 46.0,
-      "step": 7672
-    },
-    {
-      "epoch": 1.2356777648053465,
-      "grad_norm": 0.0005188011564314365,
-      "learning_rate": 0.0001999992485155069,
-      "loss": 46.0,
-      "step": 7673
-    },
-    {
-      "epoch": 1.2358388018841338,
-      "grad_norm": 0.0010696490062400699,
-      "learning_rate": 0.00019999924831936113,
-      "loss": 46.0,
-      "step": 7674
-    },
-    {
-      "epoch": 1.2359998389629212,
-      "grad_norm": 0.004267524462193251,
-      "learning_rate": 0.00019999924812318974,
-      "loss": 46.0,
-      "step": 7675
-    },
-    {
-      "epoch": 1.2361608760417087,
-      "grad_norm": 0.003083720337599516,
-      "learning_rate": 0.00019999924792699276,
-      "loss": 46.0,
-      "step": 7676
-    },
-    {
-      "epoch": 1.2363219131204959,
-      "grad_norm": 0.0018491089576855302,
-      "learning_rate": 0.00019999924773077016,
-      "loss": 46.0,
-      "step": 7677
-    },
-    {
-      "epoch": 1.2364829501992833,
-      "grad_norm": 0.0014480621321126819,
-      "learning_rate": 0.000199999247534522,
-      "loss": 46.0,
-      "step": 7678
-    },
-    {
-      "epoch": 1.2366439872780708,
-      "grad_norm": 0.000995188718661666,
-      "learning_rate": 0.00019999924733824824,
-      "loss": 46.0,
-      "step": 7679
-    },
-    {
-      "epoch": 1.2368050243568582,
-      "grad_norm": 0.0004588558222167194,
-      "learning_rate": 0.0001999992471419489,
-      "loss": 46.0,
-      "step": 7680
-    },
-    {
-      "epoch": 1.2369660614356455,
-      "grad_norm": 0.0009994530119001865,
-      "learning_rate": 0.00019999924694562396,
-      "loss": 46.0,
-      "step": 7681
-    },
-    {
-      "epoch": 1.237127098514433,
-      "grad_norm": 0.004038609564304352,
-      "learning_rate": 0.0001999992467492734,
-      "loss": 46.0,
-      "step": 7682
-    },
-    {
-      "epoch": 1.2372881355932204,
-      "grad_norm": 0.0009063747129403055,
-      "learning_rate": 0.00019999924655289726,
-      "loss": 46.0,
-      "step": 7683
-    },
-    {
-      "epoch": 1.2374491726720078,
-      "grad_norm": 0.0032376996241509914,
-      "learning_rate": 0.00019999924635649553,
-      "loss": 46.0,
-      "step": 7684
-    },
-    {
-      "epoch": 1.237610209750795,
-      "grad_norm": 0.004285745322704315,
-      "learning_rate": 0.0001999992461600682,
-      "loss": 46.0,
-      "step": 7685
-    },
-    {
-      "epoch": 1.2377712468295825,
-      "grad_norm": 0.0007059372728690505,
-      "learning_rate": 0.00019999924596361527,
-      "loss": 46.0,
-      "step": 7686
-    },
-    {
-      "epoch": 1.23793228390837,
-      "grad_norm": 0.0046587600372731686,
-      "learning_rate": 0.00019999924576713675,
-      "loss": 46.0,
-      "step": 7687
-    },
-    {
-      "epoch": 1.2380933209871574,
-      "grad_norm": 0.0014678449369966984,
-      "learning_rate": 0.00019999924557063264,
-      "loss": 46.0,
-      "step": 7688
-    },
-    {
-      "epoch": 1.2382543580659446,
-      "grad_norm": 0.002469314495101571,
-      "learning_rate": 0.00019999924537410295,
-      "loss": 46.0,
-      "step": 7689
-    },
-    {
-      "epoch": 1.238415395144732,
-      "grad_norm": 0.006369494367390871,
-      "learning_rate": 0.00019999924517754764,
-      "loss": 46.0,
-      "step": 7690
-    },
-    {
-      "epoch": 1.2385764322235195,
-      "grad_norm": 0.0013271247735247016,
-      "learning_rate": 0.00019999924498096674,
-      "loss": 46.0,
-      "step": 7691
-    },
-    {
-      "epoch": 1.2387374693023068,
-      "grad_norm": 0.0009874668903648853,
-      "learning_rate": 0.00019999924478436026,
-      "loss": 46.0,
-      "step": 7692
-    },
-    {
-      "epoch": 1.2388985063810942,
-      "grad_norm": 0.0019541261717677116,
-      "learning_rate": 0.00019999924458772816,
-      "loss": 46.0,
-      "step": 7693
-    },
-    {
-      "epoch": 1.2390595434598817,
-      "grad_norm": 0.0023790907580405474,
-      "learning_rate": 0.00019999924439107047,
-      "loss": 46.0,
-      "step": 7694
-    },
-    {
-      "epoch": 1.2392205805386691,
-      "grad_norm": 0.0010876428568735719,
-      "learning_rate": 0.0001999992441943872,
-      "loss": 46.0,
-      "step": 7695
-    },
-    {
-      "epoch": 1.2393816176174564,
-      "grad_norm": 0.001680009183473885,
-      "learning_rate": 0.00019999924399767834,
-      "loss": 46.0,
-      "step": 7696
-    },
-    {
-      "epoch": 1.2395426546962438,
-      "grad_norm": 0.0030936039984226227,
-      "learning_rate": 0.0001999992438009439,
-      "loss": 46.0,
-      "step": 7697
-    },
-    {
-      "epoch": 1.2397036917750313,
-      "grad_norm": 0.00127962336409837,
-      "learning_rate": 0.0001999992436041838,
-      "loss": 46.0,
-      "step": 7698
-    },
-    {
-      "epoch": 1.2398647288538185,
-      "grad_norm": 0.007616166956722736,
-      "learning_rate": 0.00019999924340739815,
-      "loss": 46.0,
-      "step": 7699
-    },
-    {
-      "epoch": 1.240025765932606,
-      "grad_norm": 0.0006716623902320862,
-      "learning_rate": 0.00019999924321058692,
-      "loss": 46.0,
-      "step": 7700
-    },
-    {
-      "epoch": 1.2401868030113934,
-      "grad_norm": 0.0009732271428219974,
-      "learning_rate": 0.00019999924301375007,
-      "loss": 46.0,
-      "step": 7701
-    },
-    {
-      "epoch": 1.2403478400901808,
-      "grad_norm": 0.007000810466706753,
-      "learning_rate": 0.00019999924281688763,
-      "loss": 46.0,
-      "step": 7702
-    },
-    {
-      "epoch": 1.240508877168968,
-      "grad_norm": 0.0018774871714413166,
-      "learning_rate": 0.00019999924261999958,
-      "loss": 46.0,
-      "step": 7703
-    },
-    {
-      "epoch": 1.2406699142477555,
-      "grad_norm": 0.0026521796826273203,
-      "learning_rate": 0.00019999924242308597,
-      "loss": 46.0,
-      "step": 7704
-    },
-    {
-      "epoch": 1.240830951326543,
-      "grad_norm": 0.0009369513718411326,
-      "learning_rate": 0.00019999924222614674,
-      "loss": 46.0,
-      "step": 7705
-    },
-    {
-      "epoch": 1.2409919884053302,
-      "grad_norm": 0.0013953230809420347,
-      "learning_rate": 0.00019999924202918193,
-      "loss": 46.0,
-      "step": 7706
-    },
-    {
-      "epoch": 1.2411530254841177,
-      "grad_norm": 0.0031619705259799957,
-      "learning_rate": 0.00019999924183219153,
-      "loss": 46.0,
-      "step": 7707
-    },
-    {
-      "epoch": 1.241314062562905,
-      "grad_norm": 0.0009811873314902186,
-      "learning_rate": 0.0001999992416351755,
-      "loss": 46.0,
-      "step": 7708
-    },
-    {
-      "epoch": 1.2414750996416926,
-      "grad_norm": 0.0016565483529120684,
-      "learning_rate": 0.00019999924143813388,
-      "loss": 46.0,
-      "step": 7709
-    },
-    {
-      "epoch": 1.24163613672048,
-      "grad_norm": 0.0018206652021035552,
-      "learning_rate": 0.0001999992412410667,
-      "loss": 46.0,
-      "step": 7710
-    },
-    {
-      "epoch": 1.2417971737992672,
-      "grad_norm": 0.001449157833121717,
-      "learning_rate": 0.00019999924104397392,
-      "loss": 46.0,
-      "step": 7711
-    },
-    {
-      "epoch": 1.2419582108780547,
-      "grad_norm": 0.005289965309202671,
-      "learning_rate": 0.00019999924084685553,
-      "loss": 46.0,
-      "step": 7712
-    },
-    {
-      "epoch": 1.2421192479568421,
-      "grad_norm": 0.0035281176678836346,
-      "learning_rate": 0.00019999924064971155,
-      "loss": 46.0,
-      "step": 7713
-    },
-    {
-      "epoch": 1.2422802850356294,
-      "grad_norm": 0.0022060268092900515,
-      "learning_rate": 0.00019999924045254196,
-      "loss": 46.0,
-      "step": 7714
-    },
-    {
-      "epoch": 1.2424413221144168,
-      "grad_norm": 0.0017881705425679684,
-      "learning_rate": 0.0001999992402553468,
-      "loss": 46.0,
-      "step": 7715
-    },
-    {
-      "epoch": 1.2426023591932043,
-      "grad_norm": 0.0008391349110752344,
-      "learning_rate": 0.00019999924005812604,
-      "loss": 46.0,
-      "step": 7716
-    },
-    {
-      "epoch": 1.2427633962719917,
-      "grad_norm": 0.002548717427998781,
-      "learning_rate": 0.0001999992398608797,
-      "loss": 46.0,
-      "step": 7717
-    },
-    {
-      "epoch": 1.242924433350779,
-      "grad_norm": 0.0028777257539331913,
-      "learning_rate": 0.00019999923966360772,
-      "loss": 46.0,
-      "step": 7718
-    },
-    {
-      "epoch": 1.2430854704295664,
-      "grad_norm": 0.001732136937789619,
-      "learning_rate": 0.0001999992394663102,
-      "loss": 46.0,
-      "step": 7719
-    },
-    {
-      "epoch": 1.2432465075083539,
-      "grad_norm": 0.004271198995411396,
-      "learning_rate": 0.00019999923926898702,
-      "loss": 46.0,
-      "step": 7720
-    },
-    {
-      "epoch": 1.243407544587141,
-      "grad_norm": 0.0016677387757226825,
-      "learning_rate": 0.00019999923907163832,
-      "loss": 46.0,
-      "step": 7721
-    },
-    {
-      "epoch": 1.2435685816659285,
-      "grad_norm": 0.0005603987374342978,
-      "learning_rate": 0.00019999923887426398,
-      "loss": 46.0,
-      "step": 7722
-    },
-    {
-      "epoch": 1.243729618744716,
-      "grad_norm": 0.0015048218192532659,
-      "learning_rate": 0.00019999923867686405,
-      "loss": 46.0,
-      "step": 7723
-    },
-    {
-      "epoch": 1.2438906558235034,
-      "grad_norm": 0.0017956284573301673,
-      "learning_rate": 0.0001999992384794385,
-      "loss": 46.0,
-      "step": 7724
-    },
-    {
-      "epoch": 1.2440516929022907,
-      "grad_norm": 0.0007143549155443907,
-      "learning_rate": 0.0001999992382819874,
-      "loss": 46.0,
-      "step": 7725
-    },
-    {
-      "epoch": 1.2442127299810781,
-      "grad_norm": 0.00385107658803463,
-      "learning_rate": 0.00019999923808451068,
-      "loss": 46.0,
-      "step": 7726
-    },
-    {
-      "epoch": 1.2443737670598656,
-      "grad_norm": 0.0009276755154132843,
-      "learning_rate": 0.0001999992378870084,
-      "loss": 46.0,
-      "step": 7727
-    },
-    {
-      "epoch": 1.2445348041386528,
-      "grad_norm": 0.0008219244773499668,
-      "learning_rate": 0.0001999992376894805,
-      "loss": 46.0,
-      "step": 7728
-    },
-    {
-      "epoch": 1.2446958412174403,
-      "grad_norm": 0.002346830442547798,
-      "learning_rate": 0.000199999237491927,
-      "loss": 46.0,
-      "step": 7729
-    },
-    {
-      "epoch": 1.2448568782962277,
-      "grad_norm": 0.003625713288784027,
-      "learning_rate": 0.0001999992372943479,
-      "loss": 46.0,
-      "step": 7730
-    },
-    {
-      "epoch": 1.2450179153750152,
-      "grad_norm": 0.0018733638571575284,
-      "learning_rate": 0.00019999923709674322,
-      "loss": 46.0,
-      "step": 7731
-    },
-    {
-      "epoch": 1.2451789524538026,
-      "grad_norm": 0.0011365804821252823,
-      "learning_rate": 0.00019999923689911295,
-      "loss": 46.0,
-      "step": 7732
-    },
-    {
-      "epoch": 1.2453399895325898,
-      "grad_norm": 0.0005309239495545626,
-      "learning_rate": 0.00019999923670145707,
-      "loss": 46.0,
-      "step": 7733
-    },
-    {
-      "epoch": 1.2455010266113773,
-      "grad_norm": 0.0011068431194871664,
-      "learning_rate": 0.0001999992365037756,
-      "loss": 46.0,
-      "step": 7734
-    },
-    {
-      "epoch": 1.2456620636901647,
-      "grad_norm": 0.006984272971749306,
-      "learning_rate": 0.00019999923630606854,
-      "loss": 46.0,
-      "step": 7735
-    },
-    {
-      "epoch": 1.245823100768952,
-      "grad_norm": 0.007078150287270546,
-      "learning_rate": 0.0001999992361083359,
-      "loss": 46.0,
-      "step": 7736
-    },
-    {
-      "epoch": 1.2459841378477394,
-      "grad_norm": 0.0004573630867525935,
-      "learning_rate": 0.00019999923591057763,
-      "loss": 46.0,
-      "step": 7737
-    },
-    {
-      "epoch": 1.2461451749265269,
-      "grad_norm": 0.0017036707140505314,
-      "learning_rate": 0.0001999992357127938,
-      "loss": 46.0,
-      "step": 7738
-    },
-    {
-      "epoch": 1.2463062120053143,
-      "grad_norm": 0.0037747807800769806,
-      "learning_rate": 0.00019999923551498435,
-      "loss": 46.0,
-      "step": 7739
-    },
-    {
-      "epoch": 1.2464672490841016,
-      "grad_norm": 0.0032288769725710154,
-      "learning_rate": 0.0001999992353171493,
-      "loss": 46.0,
-      "step": 7740
-    },
-    {
-      "epoch": 1.246628286162889,
-      "grad_norm": 0.005686294753104448,
-      "learning_rate": 0.00019999923511928867,
-      "loss": 46.0,
-      "step": 7741
-    },
-    {
-      "epoch": 1.2467893232416765,
-      "grad_norm": 0.001286122016608715,
-      "learning_rate": 0.00019999923492140245,
-      "loss": 46.0,
-      "step": 7742
-    },
-    {
-      "epoch": 1.2469503603204637,
-      "grad_norm": 0.006018731743097305,
-      "learning_rate": 0.00019999923472349064,
-      "loss": 46.0,
-      "step": 7743
-    },
-    {
-      "epoch": 1.2471113973992511,
-      "grad_norm": 0.002859055530279875,
-      "learning_rate": 0.00019999923452555321,
-      "loss": 46.0,
-      "step": 7744
-    },
-    {
-      "epoch": 1.2472724344780386,
-      "grad_norm": 0.002178067108616233,
-      "learning_rate": 0.0001999992343275902,
-      "loss": 46.0,
-      "step": 7745
-    },
-    {
-      "epoch": 1.247433471556826,
-      "grad_norm": 0.003421215573325753,
-      "learning_rate": 0.0001999992341296016,
-      "loss": 46.0,
-      "step": 7746
-    },
-    {
-      "epoch": 1.2475945086356133,
-      "grad_norm": 0.0030373604968190193,
-      "learning_rate": 0.0001999992339315874,
-      "loss": 46.0,
-      "step": 7747
-    },
-    {
-      "epoch": 1.2477555457144007,
-      "grad_norm": 0.0007116520428098738,
-      "learning_rate": 0.00019999923373354762,
-      "loss": 46.0,
-      "step": 7748
-    },
-    {
-      "epoch": 1.2479165827931882,
-      "grad_norm": 0.0012223977828398347,
-      "learning_rate": 0.00019999923353548223,
-      "loss": 46.0,
-      "step": 7749
-    },
-    {
-      "epoch": 1.2480776198719754,
-      "grad_norm": 0.0027469934429973364,
-      "learning_rate": 0.00019999923333739123,
-      "loss": 46.0,
-      "step": 7750
-    },
-    {
-      "epoch": 1.2482386569507629,
-      "grad_norm": 0.0073843407444655895,
-      "learning_rate": 0.00019999923313927464,
-      "loss": 46.0,
-      "step": 7751
-    },
-    {
-      "epoch": 1.2483996940295503,
-      "grad_norm": 0.0005689335521310568,
-      "learning_rate": 0.00019999923294113247,
-      "loss": 46.0,
-      "step": 7752
-    },
-    {
-      "epoch": 1.2485607311083378,
-      "grad_norm": 0.0016763632884249091,
-      "learning_rate": 0.00019999923274296473,
-      "loss": 46.0,
-      "step": 7753
-    },
-    {
-      "epoch": 1.248721768187125,
-      "grad_norm": 0.0015108821680769324,
-      "learning_rate": 0.00019999923254477136,
-      "loss": 46.0,
-      "step": 7754
-    },
-    {
-      "epoch": 1.2488828052659124,
-      "grad_norm": 0.004024966154247522,
-      "learning_rate": 0.0001999992323465524,
-      "loss": 46.0,
-      "step": 7755
-    },
-    {
-      "epoch": 1.2490438423447,
-      "grad_norm": 0.001047232304699719,
-      "learning_rate": 0.00019999923214830784,
-      "loss": 46.0,
-      "step": 7756
-    },
-    {
-      "epoch": 1.2492048794234873,
-      "grad_norm": 0.002008634153753519,
-      "learning_rate": 0.0001999992319500377,
-      "loss": 46.0,
-      "step": 7757
-    },
-    {
-      "epoch": 1.2493659165022746,
-      "grad_norm": 0.003829686203971505,
-      "learning_rate": 0.00019999923175174198,
-      "loss": 46.0,
-      "step": 7758
-    },
-    {
-      "epoch": 1.249526953581062,
-      "grad_norm": 0.001497588586062193,
-      "learning_rate": 0.0001999992315534206,
-      "loss": 46.0,
-      "step": 7759
-    },
-    {
-      "epoch": 1.2496879906598495,
-      "grad_norm": 0.0035700860898941755,
-      "learning_rate": 0.0001999992313550737,
-      "loss": 46.0,
-      "step": 7760
-    },
-    {
-      "epoch": 1.249849027738637,
-      "grad_norm": 0.005076493602246046,
-      "learning_rate": 0.0001999992311567012,
-      "loss": 46.0,
-      "step": 7761
-    },
-    {
-      "epoch": 1.2500100648174242,
-      "grad_norm": 0.0015731723979115486,
-      "learning_rate": 0.00019999923095830307,
-      "loss": 46.0,
-      "step": 7762
-    },
-    {
-      "epoch": 1.2501711018962116,
-      "grad_norm": 0.005798127502202988,
-      "learning_rate": 0.00019999923075987935,
-      "loss": 46.0,
-      "step": 7763
-    },
-    {
-      "epoch": 1.250332138974999,
-      "grad_norm": 0.0013734509702771902,
-      "learning_rate": 0.00019999923056143005,
-      "loss": 46.0,
-      "step": 7764
-    },
-    {
-      "epoch": 1.2504931760537863,
-      "grad_norm": 0.004089319612830877,
-      "learning_rate": 0.00019999923036295513,
-      "loss": 46.0,
-      "step": 7765
-    },
-    {
-      "epoch": 1.2506542131325737,
-      "grad_norm": 0.0013517532497644424,
-      "learning_rate": 0.00019999923016445463,
-      "loss": 46.0,
-      "step": 7766
-    },
-    {
-      "epoch": 1.2508152502113612,
-      "grad_norm": 0.004541502799838781,
-      "learning_rate": 0.00019999922996592854,
-      "loss": 46.0,
-      "step": 7767
-    },
-    {
-      "epoch": 1.2509762872901486,
-      "grad_norm": 0.0010161533718928695,
-      "learning_rate": 0.00019999922976737686,
-      "loss": 46.0,
-      "step": 7768
-    },
-    {
-      "epoch": 1.2511373243689359,
-      "grad_norm": 0.0017163571901619434,
-      "learning_rate": 0.00019999922956879957,
-      "loss": 46.0,
-      "step": 7769
-    },
-    {
-      "epoch": 1.2512983614477233,
-      "grad_norm": 0.00045029327156953514,
-      "learning_rate": 0.0001999992293701967,
-      "loss": 46.0,
-      "step": 7770
-    },
-    {
-      "epoch": 1.2514593985265108,
-      "grad_norm": 0.0012799405958503485,
-      "learning_rate": 0.00019999922917156822,
-      "loss": 46.0,
-      "step": 7771
-    },
-    {
-      "epoch": 1.251620435605298,
-      "grad_norm": 0.0010593110928311944,
-      "learning_rate": 0.00019999922897291416,
-      "loss": 46.0,
-      "step": 7772
-    },
-    {
-      "epoch": 1.2517814726840855,
-      "grad_norm": 0.002537592314183712,
-      "learning_rate": 0.0001999992287742345,
-      "loss": 46.0,
-      "step": 7773
-    },
-    {
-      "epoch": 1.251942509762873,
-      "grad_norm": 0.0019495715387165546,
-      "learning_rate": 0.00019999922857552927,
-      "loss": 46.0,
-      "step": 7774
-    },
-    {
-      "epoch": 1.2521035468416604,
-      "grad_norm": 0.0004583829140756279,
-      "learning_rate": 0.0001999992283767984,
-      "loss": 46.0,
-      "step": 7775
-    },
-    {
-      "epoch": 1.2522645839204478,
-      "grad_norm": 0.001894280663691461,
-      "learning_rate": 0.00019999922817804197,
-      "loss": 46.0,
-      "step": 7776
-    },
-    {
-      "epoch": 1.252425620999235,
-      "grad_norm": 0.0008542780415154994,
-      "learning_rate": 0.00019999922797925992,
-      "loss": 46.0,
-      "step": 7777
-    },
-    {
-      "epoch": 1.2525866580780225,
-      "grad_norm": 0.006037456914782524,
-      "learning_rate": 0.0001999992277804523,
-      "loss": 46.0,
-      "step": 7778
-    },
-    {
-      "epoch": 1.2527476951568097,
-      "grad_norm": 0.002517266198992729,
-      "learning_rate": 0.00019999922758161907,
-      "loss": 46.0,
-      "step": 7779
-    },
-    {
-      "epoch": 1.2529087322355972,
-      "grad_norm": 0.0006100016180425882,
-      "learning_rate": 0.0001999992273827602,
-      "loss": 46.0,
-      "step": 7780
-    },
-    {
-      "epoch": 1.2530697693143846,
-      "grad_norm": 0.0007259164121933281,
-      "learning_rate": 0.0001999992271838758,
-      "loss": 46.0,
-      "step": 7781
-    },
-    {
-      "epoch": 1.253230806393172,
-      "grad_norm": 0.0017611620714887977,
-      "learning_rate": 0.0001999992269849658,
-      "loss": 46.0,
-      "step": 7782
-    },
-    {
-      "epoch": 1.2533918434719595,
-      "grad_norm": 0.0013355906121432781,
-      "learning_rate": 0.0001999992267860302,
-      "loss": 46.0,
-      "step": 7783
-    },
-    {
-      "epoch": 1.2535528805507468,
-      "grad_norm": 0.0031003172043710947,
-      "learning_rate": 0.00019999922658706898,
-      "loss": 46.0,
-      "step": 7784
-    },
-    {
-      "epoch": 1.2537139176295342,
-      "grad_norm": 0.0006265430711209774,
-      "learning_rate": 0.00019999922638808218,
-      "loss": 46.0,
-      "step": 7785
-    },
-    {
-      "epoch": 1.2538749547083217,
-      "grad_norm": 0.0015952771063894033,
-      "learning_rate": 0.0001999992261890698,
-      "loss": 46.0,
-      "step": 7786
-    },
-    {
-      "epoch": 1.254035991787109,
-      "grad_norm": 0.0006227141129784286,
-      "learning_rate": 0.00019999922599003183,
-      "loss": 46.0,
-      "step": 7787
-    },
-    {
-      "epoch": 1.2541970288658963,
-      "grad_norm": 0.0023414224851876497,
-      "learning_rate": 0.00019999922579096825,
-      "loss": 46.0,
-      "step": 7788
-    },
-    {
-      "epoch": 1.2543580659446838,
-      "grad_norm": 0.0013503084192052484,
-      "learning_rate": 0.00019999922559187905,
-      "loss": 46.0,
-      "step": 7789
-    },
-    {
-      "epoch": 1.2545191030234712,
-      "grad_norm": 0.0003863623715005815,
-      "learning_rate": 0.0001999992253927643,
-      "loss": 46.0,
-      "step": 7790
-    },
-    {
-      "epoch": 1.2546801401022585,
-      "grad_norm": 0.0008104327134788036,
-      "learning_rate": 0.00019999922519362394,
-      "loss": 46.0,
-      "step": 7791
-    },
-    {
-      "epoch": 1.254841177181046,
-      "grad_norm": 0.0016009743558242917,
-      "learning_rate": 0.00019999922499445795,
-      "loss": 46.0,
-      "step": 7792
-    },
-    {
-      "epoch": 1.2550022142598334,
-      "grad_norm": 0.0007865499937906861,
-      "learning_rate": 0.0001999992247952664,
-      "loss": 46.0,
-      "step": 7793
-    },
-    {
-      "epoch": 1.2551632513386206,
-      "grad_norm": 0.000501057889778167,
-      "learning_rate": 0.00019999922459604927,
-      "loss": 46.0,
-      "step": 7794
-    },
-    {
-      "epoch": 1.255324288417408,
-      "grad_norm": 0.0023141943383961916,
-      "learning_rate": 0.00019999922439680652,
-      "loss": 46.0,
-      "step": 7795
-    },
-    {
-      "epoch": 1.2554853254961955,
-      "grad_norm": 0.002178687136620283,
-      "learning_rate": 0.0001999992241975382,
-      "loss": 46.0,
-      "step": 7796
-    },
-    {
-      "epoch": 1.255646362574983,
-      "grad_norm": 0.0018516256241127849,
-      "learning_rate": 0.00019999922399824426,
-      "loss": 46.0,
-      "step": 7797
-    },
-    {
-      "epoch": 1.2558073996537704,
-      "grad_norm": 0.0013372964458540082,
-      "learning_rate": 0.00019999922379892473,
-      "loss": 46.0,
-      "step": 7798
-    },
-    {
-      "epoch": 1.2559684367325576,
-      "grad_norm": 0.0020784952212125063,
-      "learning_rate": 0.0001999992235995796,
-      "loss": 46.0,
-      "step": 7799
-    },
-    {
-      "epoch": 1.256129473811345,
-      "grad_norm": 0.0026113821659237146,
-      "learning_rate": 0.0001999992234002089,
-      "loss": 46.0,
-      "step": 7800
-    },
-    {
-      "epoch": 1.2562905108901323,
-      "grad_norm": 0.0039160423912107944,
-      "learning_rate": 0.0001999992232008126,
-      "loss": 46.0,
-      "step": 7801
-    },
-    {
-      "epoch": 1.2564515479689198,
-      "grad_norm": 0.0022273152135312557,
-      "learning_rate": 0.00019999922300139068,
-      "loss": 46.0,
-      "step": 7802
-    },
-    {
-      "epoch": 1.2566125850477072,
-      "grad_norm": 0.001879971125163138,
-      "learning_rate": 0.00019999922280194318,
-      "loss": 46.0,
-      "step": 7803
-    },
-    {
-      "epoch": 1.2567736221264947,
-      "grad_norm": 0.00146833760663867,
-      "learning_rate": 0.0001999992226024701,
-      "loss": 46.0,
-      "step": 7804
-    },
-    {
-      "epoch": 1.2569346592052821,
-      "grad_norm": 0.0015800370601937175,
-      "learning_rate": 0.00019999922240297139,
-      "loss": 46.0,
-      "step": 7805
-    },
-    {
-      "epoch": 1.2570956962840694,
-      "grad_norm": 0.0011895514326170087,
-      "learning_rate": 0.00019999922220344712,
-      "loss": 46.0,
-      "step": 7806
-    },
-    {
-      "epoch": 1.2572567333628568,
-      "grad_norm": 0.006521036382764578,
-      "learning_rate": 0.00019999922200389722,
-      "loss": 46.0,
-      "step": 7807
-    },
-    {
-      "epoch": 1.2574177704416443,
-      "grad_norm": 0.0016586346318945289,
-      "learning_rate": 0.00019999922180432176,
-      "loss": 46.0,
-      "step": 7808
-    },
-    {
-      "epoch": 1.2575788075204315,
-      "grad_norm": 0.00036867480957880616,
-      "learning_rate": 0.00019999922160472068,
-      "loss": 46.0,
-      "step": 7809
-    },
-    {
-      "epoch": 1.257739844599219,
-      "grad_norm": 0.004525313153862953,
-      "learning_rate": 0.00019999922140509401,
-      "loss": 46.0,
-      "step": 7810
-    },
-    {
-      "epoch": 1.2579008816780064,
-      "grad_norm": 0.0016664706636220217,
-      "learning_rate": 0.00019999922120544176,
-      "loss": 46.0,
-      "step": 7811
-    },
-    {
-      "epoch": 1.2580619187567939,
-      "grad_norm": 0.0030280391220003366,
-      "learning_rate": 0.00019999922100576392,
-      "loss": 46.0,
-      "step": 7812
-    },
-    {
-      "epoch": 1.258222955835581,
-      "grad_norm": 0.001684386981651187,
-      "learning_rate": 0.00019999922080606047,
-      "loss": 46.0,
-      "step": 7813
-    },
-    {
-      "epoch": 1.2583839929143685,
-      "grad_norm": 0.0018927217461168766,
-      "learning_rate": 0.00019999922060633143,
-      "loss": 46.0,
-      "step": 7814
-    },
-    {
-      "epoch": 1.258545029993156,
-      "grad_norm": 0.0013537174090743065,
-      "learning_rate": 0.00019999922040657677,
-      "loss": 46.0,
-      "step": 7815
-    },
-    {
-      "epoch": 1.2587060670719432,
-      "grad_norm": 0.0008411871967837214,
-      "learning_rate": 0.00019999922020679656,
-      "loss": 46.0,
-      "step": 7816
-    },
-    {
-      "epoch": 1.2588671041507307,
-      "grad_norm": 0.005712592042982578,
-      "learning_rate": 0.00019999922000699073,
-      "loss": 46.0,
-      "step": 7817
-    },
-    {
-      "epoch": 1.2590281412295181,
-      "grad_norm": 0.00286250957287848,
-      "learning_rate": 0.0001999992198071593,
-      "loss": 46.0,
-      "step": 7818
-    },
-    {
-      "epoch": 1.2591891783083056,
-      "grad_norm": 0.0030506982002407312,
-      "learning_rate": 0.0001999992196073023,
-      "loss": 46.0,
-      "step": 7819
-    },
-    {
-      "epoch": 1.2593502153870928,
-      "grad_norm": 0.0009191773133352399,
-      "learning_rate": 0.0001999992194074197,
-      "loss": 46.0,
-      "step": 7820
-    },
-    {
-      "epoch": 1.2595112524658802,
-      "grad_norm": 0.0027808849699795246,
-      "learning_rate": 0.00019999921920751148,
-      "loss": 46.0,
-      "step": 7821
-    },
-    {
-      "epoch": 1.2596722895446677,
-      "grad_norm": 0.0005924653960391879,
-      "learning_rate": 0.00019999921900757766,
-      "loss": 46.0,
-      "step": 7822
-    },
-    {
-      "epoch": 1.259833326623455,
-      "grad_norm": 0.001545565901324153,
-      "learning_rate": 0.00019999921880761829,
-      "loss": 46.0,
-      "step": 7823
-    },
-    {
-      "epoch": 1.2599943637022424,
-      "grad_norm": 0.0006674687028862536,
-      "learning_rate": 0.0001999992186076333,
-      "loss": 46.0,
-      "step": 7824
-    },
-    {
-      "epoch": 1.2601554007810298,
-      "grad_norm": 0.0017125386511906981,
-      "learning_rate": 0.0001999992184076227,
-      "loss": 46.0,
-      "step": 7825
-    },
-    {
-      "epoch": 1.2603164378598173,
-      "grad_norm": 0.009428374469280243,
-      "learning_rate": 0.00019999921820758654,
-      "loss": 46.0,
-      "step": 7826
-    },
-    {
-      "epoch": 1.2604774749386047,
-      "grad_norm": 0.0012851067585870624,
-      "learning_rate": 0.00019999921800752476,
-      "loss": 46.0,
-      "step": 7827
-    },
-    {
-      "epoch": 1.260638512017392,
-      "grad_norm": 0.0033618267625570297,
-      "learning_rate": 0.0001999992178074374,
-      "loss": 46.0,
-      "step": 7828
-    },
-    {
-      "epoch": 1.2607995490961794,
-      "grad_norm": 0.0018210412235930562,
-      "learning_rate": 0.00019999921760732443,
-      "loss": 46.0,
-      "step": 7829
-    },
-    {
-      "epoch": 1.2609605861749669,
-      "grad_norm": 0.0026901294477283955,
-      "learning_rate": 0.0001999992174071859,
-      "loss": 46.0,
-      "step": 7830
-    },
-    {
-      "epoch": 1.261121623253754,
-      "grad_norm": 0.004382093902677298,
-      "learning_rate": 0.00019999921720702173,
-      "loss": 46.0,
-      "step": 7831
-    },
-    {
-      "epoch": 1.2612826603325415,
-      "grad_norm": 0.0008639134466648102,
-      "learning_rate": 0.00019999921700683199,
-      "loss": 46.0,
-      "step": 7832
-    },
-    {
-      "epoch": 1.261443697411329,
-      "grad_norm": 0.0022201361134648323,
-      "learning_rate": 0.00019999921680661665,
-      "loss": 46.0,
-      "step": 7833
-    },
-    {
-      "epoch": 1.2616047344901165,
-      "grad_norm": 0.0019140188815072179,
-      "learning_rate": 0.0001999992166063757,
-      "loss": 46.0,
-      "step": 7834
-    },
-    {
-      "epoch": 1.2617657715689037,
-      "grad_norm": 0.0018385075964033604,
-      "learning_rate": 0.00019999921640610917,
-      "loss": 46.0,
-      "step": 7835
-    },
-    {
-      "epoch": 1.2619268086476911,
-      "grad_norm": 0.0075474693439900875,
-      "learning_rate": 0.00019999921620581705,
-      "loss": 46.0,
-      "step": 7836
-    },
-    {
-      "epoch": 1.2620878457264786,
-      "grad_norm": 0.0004926960682496428,
-      "learning_rate": 0.00019999921600549934,
-      "loss": 46.0,
-      "step": 7837
-    },
-    {
-      "epoch": 1.2622488828052658,
-      "grad_norm": 0.00046431299415417016,
-      "learning_rate": 0.00019999921580515602,
-      "loss": 46.0,
-      "step": 7838
-    },
-    {
-      "epoch": 1.2624099198840533,
-      "grad_norm": 0.0012626488460227847,
-      "learning_rate": 0.00019999921560478708,
-      "loss": 46.0,
-      "step": 7839
-    },
-    {
-      "epoch": 1.2625709569628407,
-      "grad_norm": 0.001826937892474234,
-      "learning_rate": 0.00019999921540439259,
-      "loss": 46.0,
-      "step": 7840
-    },
-    {
-      "epoch": 1.2627319940416282,
-      "grad_norm": 0.004643454682081938,
-      "learning_rate": 0.0001999992152039725,
-      "loss": 46.0,
-      "step": 7841
-    },
-    {
-      "epoch": 1.2628930311204154,
-      "grad_norm": 0.0019279230618849397,
-      "learning_rate": 0.0001999992150035268,
-      "loss": 46.0,
-      "step": 7842
-    },
-    {
-      "epoch": 1.2630540681992029,
-      "grad_norm": 0.0012744168052449822,
-      "learning_rate": 0.00019999921480305552,
-      "loss": 46.0,
-      "step": 7843
-    },
-    {
-      "epoch": 1.2632151052779903,
-      "grad_norm": 0.005128585267812014,
-      "learning_rate": 0.00019999921460255867,
-      "loss": 46.0,
-      "step": 7844
-    },
-    {
-      "epoch": 1.2633761423567775,
-      "grad_norm": 0.006046860944479704,
-      "learning_rate": 0.00019999921440203618,
-      "loss": 46.0,
-      "step": 7845
-    },
-    {
-      "epoch": 1.263537179435565,
-      "grad_norm": 0.003778246697038412,
-      "learning_rate": 0.0001999992142014881,
-      "loss": 46.0,
-      "step": 7846
-    },
-    {
-      "epoch": 1.2636982165143524,
-      "grad_norm": 0.0005131994839757681,
-      "learning_rate": 0.00019999921400091445,
-      "loss": 46.0,
-      "step": 7847
-    },
-    {
-      "epoch": 1.2638592535931399,
-      "grad_norm": 0.003927554935216904,
-      "learning_rate": 0.0001999992138003152,
-      "loss": 46.0,
-      "step": 7848
-    },
-    {
-      "epoch": 1.2640202906719273,
-      "grad_norm": 0.0013735465472564101,
-      "learning_rate": 0.0001999992135996903,
-      "loss": 46.0,
-      "step": 7849
-    },
-    {
-      "epoch": 1.2641813277507146,
-      "grad_norm": 0.0036202704068273306,
-      "learning_rate": 0.00019999921339903986,
-      "loss": 46.0,
-      "step": 7850
-    },
-    {
-      "epoch": 1.264342364829502,
-      "grad_norm": 0.001460652332752943,
-      "learning_rate": 0.00019999921319836382,
-      "loss": 46.0,
-      "step": 7851
-    },
-    {
-      "epoch": 1.2645034019082892,
-      "grad_norm": 0.0021991070825606585,
-      "learning_rate": 0.00019999921299766217,
-      "loss": 46.0,
-      "step": 7852
-    },
-    {
-      "epoch": 1.2646644389870767,
-      "grad_norm": 0.0044478788040578365,
-      "learning_rate": 0.00019999921279693496,
-      "loss": 46.0,
-      "step": 7853
-    },
-    {
-      "epoch": 1.2648254760658642,
-      "grad_norm": 0.0053705861791968346,
-      "learning_rate": 0.00019999921259618213,
-      "loss": 46.0,
-      "step": 7854
-    },
-    {
-      "epoch": 1.2649865131446516,
-      "grad_norm": 0.001818041317164898,
-      "learning_rate": 0.00019999921239540372,
-      "loss": 46.0,
-      "step": 7855
-    },
-    {
-      "epoch": 1.265147550223439,
-      "grad_norm": 0.0007710724021308124,
-      "learning_rate": 0.00019999921219459967,
-      "loss": 46.0,
-      "step": 7856
-    },
-    {
-      "epoch": 1.2653085873022263,
-      "grad_norm": 0.0025311822537332773,
-      "learning_rate": 0.00019999921199377005,
-      "loss": 46.0,
-      "step": 7857
-    },
-    {
-      "epoch": 1.2654696243810137,
-      "grad_norm": 0.004060943610966206,
-      "learning_rate": 0.00019999921179291485,
-      "loss": 46.0,
-      "step": 7858
-    },
-    {
-      "epoch": 1.2656306614598012,
-      "grad_norm": 0.0022596365306526423,
-      "learning_rate": 0.00019999921159203406,
-      "loss": 46.0,
-      "step": 7859
-    },
-    {
-      "epoch": 1.2657916985385884,
-      "grad_norm": 0.0014797047479078174,
-      "learning_rate": 0.00019999921139112766,
-      "loss": 46.0,
-      "step": 7860
-    },
-    {
-      "epoch": 1.2659527356173759,
-      "grad_norm": 0.005822623148560524,
-      "learning_rate": 0.00019999921119019567,
-      "loss": 46.0,
-      "step": 7861
-    },
-    {
-      "epoch": 1.2661137726961633,
-      "grad_norm": 0.006442841142416,
-      "learning_rate": 0.0001999992109892381,
-      "loss": 46.0,
-      "step": 7862
-    },
-    {
-      "epoch": 1.2662748097749508,
-      "grad_norm": 0.0025016714353114367,
-      "learning_rate": 0.0001999992107882549,
-      "loss": 46.0,
-      "step": 7863
-    },
-    {
-      "epoch": 1.266435846853738,
-      "grad_norm": 0.004256992135196924,
-      "learning_rate": 0.00019999921058724612,
-      "loss": 46.0,
-      "step": 7864
-    },
-    {
-      "epoch": 1.2665968839325255,
-      "grad_norm": 0.0008148694760166109,
-      "learning_rate": 0.00019999921038621176,
-      "loss": 46.0,
-      "step": 7865
-    },
-    {
-      "epoch": 1.266757921011313,
-      "grad_norm": 0.002305702306330204,
-      "learning_rate": 0.0001999992101851518,
-      "loss": 46.0,
-      "step": 7866
-    },
-    {
-      "epoch": 1.2669189580901001,
-      "grad_norm": 0.002403831807896495,
-      "learning_rate": 0.00019999920998406623,
-      "loss": 46.0,
-      "step": 7867
-    },
-    {
-      "epoch": 1.2670799951688876,
-      "grad_norm": 0.004615579731762409,
-      "learning_rate": 0.00019999920978295508,
-      "loss": 46.0,
-      "step": 7868
-    },
-    {
-      "epoch": 1.267241032247675,
-      "grad_norm": 0.002948813373222947,
-      "learning_rate": 0.00019999920958181834,
-      "loss": 46.0,
-      "step": 7869
-    },
-    {
-      "epoch": 1.2674020693264625,
-      "grad_norm": 0.000665841274894774,
-      "learning_rate": 0.00019999920938065598,
-      "loss": 46.0,
-      "step": 7870
-    },
-    {
-      "epoch": 1.26756310640525,
-      "grad_norm": 0.0013801068998873234,
-      "learning_rate": 0.00019999920917946807,
-      "loss": 46.0,
-      "step": 7871
-    },
-    {
-      "epoch": 1.2677241434840372,
-      "grad_norm": 0.00414777547121048,
-      "learning_rate": 0.00019999920897825454,
-      "loss": 46.0,
-      "step": 7872
-    },
-    {
-      "epoch": 1.2678851805628246,
-      "grad_norm": 0.0009371436899527907,
-      "learning_rate": 0.0001999992087770154,
-      "loss": 46.0,
-      "step": 7873
-    },
-    {
-      "epoch": 1.2680462176416119,
-      "grad_norm": 0.0004641699488274753,
-      "learning_rate": 0.00019999920857575069,
-      "loss": 46.0,
-      "step": 7874
-    },
-    {
-      "epoch": 1.2682072547203993,
-      "grad_norm": 0.0019806360360234976,
-      "learning_rate": 0.00019999920837446037,
-      "loss": 46.0,
-      "step": 7875
-    },
-    {
-      "epoch": 1.2683682917991868,
-      "grad_norm": 0.0008508330211043358,
-      "learning_rate": 0.00019999920817314446,
-      "loss": 46.0,
-      "step": 7876
-    },
-    {
-      "epoch": 1.2685293288779742,
-      "grad_norm": 0.0032252187374979258,
-      "learning_rate": 0.00019999920797180294,
-      "loss": 46.0,
-      "step": 7877
-    },
-    {
-      "epoch": 1.2686903659567617,
-      "grad_norm": 0.0025532187428325415,
-      "learning_rate": 0.00019999920777043583,
-      "loss": 46.0,
-      "step": 7878
-    },
-    {
-      "epoch": 1.2688514030355489,
-      "grad_norm": 0.0009707274148240685,
-      "learning_rate": 0.00019999920756904316,
-      "loss": 46.0,
-      "step": 7879
-    },
-    {
-      "epoch": 1.2690124401143363,
-      "grad_norm": 0.00444596353918314,
-      "learning_rate": 0.00019999920736762486,
-      "loss": 46.0,
-      "step": 7880
-    },
-    {
-      "epoch": 1.2691734771931238,
-      "grad_norm": 0.0009638170013204217,
-      "learning_rate": 0.00019999920716618096,
-      "loss": 46.0,
-      "step": 7881
-    },
-    {
-      "epoch": 1.269334514271911,
-      "grad_norm": 0.00812914501875639,
-      "learning_rate": 0.00019999920696471148,
-      "loss": 46.0,
-      "step": 7882
-    },
-    {
-      "epoch": 1.2694955513506985,
-      "grad_norm": 0.0049952557310462,
-      "learning_rate": 0.0001999992067632164,
-      "loss": 46.0,
-      "step": 7883
-    },
-    {
-      "epoch": 1.269656588429486,
-      "grad_norm": 0.006431289482861757,
-      "learning_rate": 0.00019999920656169575,
-      "loss": 46.0,
-      "step": 7884
-    },
-    {
-      "epoch": 1.2698176255082734,
-      "grad_norm": 0.0091160973533988,
-      "learning_rate": 0.00019999920636014948,
-      "loss": 46.0,
-      "step": 7885
-    },
-    {
-      "epoch": 1.2699786625870606,
-      "grad_norm": 0.0008121522841975093,
-      "learning_rate": 0.00019999920615857764,
-      "loss": 46.0,
-      "step": 7886
-    },
-    {
-      "epoch": 1.270139699665848,
-      "grad_norm": 0.001287944265641272,
-      "learning_rate": 0.0001999992059569802,
-      "loss": 46.0,
-      "step": 7887
-    },
-    {
-      "epoch": 1.2703007367446355,
-      "grad_norm": 0.0017001189989969134,
-      "learning_rate": 0.0001999992057553571,
-      "loss": 46.0,
-      "step": 7888
-    },
-    {
-      "epoch": 1.2704617738234227,
-      "grad_norm": 0.0034251012839376926,
-      "learning_rate": 0.0001999992055537085,
-      "loss": 46.0,
-      "step": 7889
-    },
-    {
-      "epoch": 1.2706228109022102,
-      "grad_norm": 0.0019064397783949971,
-      "learning_rate": 0.00019999920535203425,
-      "loss": 46.0,
-      "step": 7890
-    },
-    {
-      "epoch": 1.2707838479809976,
-      "grad_norm": 0.0015276261838153005,
-      "learning_rate": 0.0001999992051503344,
-      "loss": 46.0,
-      "step": 7891
-    },
-    {
-      "epoch": 1.270944885059785,
-      "grad_norm": 0.0007089645368978381,
-      "learning_rate": 0.00019999920494860897,
-      "loss": 46.0,
-      "step": 7892
-    },
-    {
-      "epoch": 1.2711059221385725,
-      "grad_norm": 0.0020035470370203257,
-      "learning_rate": 0.00019999920474685797,
-      "loss": 46.0,
-      "step": 7893
-    },
-    {
-      "epoch": 1.2712669592173598,
-      "grad_norm": 0.0019431532127782702,
-      "learning_rate": 0.00019999920454508136,
-      "loss": 46.0,
-      "step": 7894
-    },
-    {
-      "epoch": 1.2714279962961472,
-      "grad_norm": 0.0028469907119870186,
-      "learning_rate": 0.00019999920434327914,
-      "loss": 46.0,
-      "step": 7895
-    },
-    {
-      "epoch": 1.2715890333749345,
-      "grad_norm": 0.00600505480542779,
-      "learning_rate": 0.00019999920414145135,
-      "loss": 46.0,
-      "step": 7896
-    },
-    {
-      "epoch": 1.271750070453722,
-      "grad_norm": 0.0009883149759843946,
-      "learning_rate": 0.00019999920393959795,
-      "loss": 46.0,
-      "step": 7897
-    },
-    {
-      "epoch": 1.2719111075325094,
-      "grad_norm": 0.004567850846797228,
-      "learning_rate": 0.00019999920373771894,
-      "loss": 46.0,
-      "step": 7898
-    },
-    {
-      "epoch": 1.2720721446112968,
-      "grad_norm": 0.0021470014471560717,
-      "learning_rate": 0.00019999920353581436,
-      "loss": 46.0,
-      "step": 7899
-    },
-    {
-      "epoch": 1.2722331816900843,
-      "grad_norm": 0.006687674205750227,
-      "learning_rate": 0.00019999920333388417,
-      "loss": 46.0,
-      "step": 7900
-    },
-    {
-      "epoch": 1.2723942187688715,
-      "grad_norm": 0.004898585379123688,
-      "learning_rate": 0.0001999992031319284,
-      "loss": 46.0,
-      "step": 7901
-    },
-    {
-      "epoch": 1.272555255847659,
-      "grad_norm": 0.0005736006423830986,
-      "learning_rate": 0.00019999920292994704,
-      "loss": 46.0,
-      "step": 7902
-    },
-    {
-      "epoch": 1.2727162929264464,
-      "grad_norm": 0.0014147100737318397,
-      "learning_rate": 0.00019999920272794006,
-      "loss": 46.0,
-      "step": 7903
-    },
-    {
-      "epoch": 1.2728773300052336,
-      "grad_norm": 0.0024097587447613478,
-      "learning_rate": 0.0001999992025259075,
-      "loss": 46.0,
-      "step": 7904
-    },
-    {
-      "epoch": 1.273038367084021,
-      "grad_norm": 0.0018156928708776832,
-      "learning_rate": 0.00019999920232384935,
-      "loss": 46.0,
-      "step": 7905
-    },
-    {
-      "epoch": 1.2731994041628085,
-      "grad_norm": 0.0015606241067871451,
-      "learning_rate": 0.0001999992021217656,
-      "loss": 46.0,
-      "step": 7906
-    },
-    {
-      "epoch": 1.273360441241596,
-      "grad_norm": 0.006038035731762648,
-      "learning_rate": 0.00019999920191965625,
-      "loss": 46.0,
-      "step": 7907
-    },
-    {
-      "epoch": 1.2735214783203832,
-      "grad_norm": 0.0036547789350152016,
-      "learning_rate": 0.00019999920171752132,
-      "loss": 46.0,
-      "step": 7908
-    },
-    {
-      "epoch": 1.2736825153991707,
-      "grad_norm": 0.0023034082259982824,
-      "learning_rate": 0.00019999920151536076,
-      "loss": 46.0,
-      "step": 7909
-    },
-    {
-      "epoch": 1.273843552477958,
-      "grad_norm": 0.0011867902940139174,
-      "learning_rate": 0.00019999920131317462,
-      "loss": 46.0,
-      "step": 7910
-    },
-    {
-      "epoch": 1.2740045895567453,
-      "grad_norm": 0.0023318377789109945,
-      "learning_rate": 0.00019999920111096292,
-      "loss": 46.0,
-      "step": 7911
-    },
-    {
-      "epoch": 1.2741656266355328,
-      "grad_norm": 0.0019217395456507802,
-      "learning_rate": 0.0001999992009087256,
-      "loss": 46.0,
-      "step": 7912
-    },
-    {
-      "epoch": 1.2743266637143202,
-      "grad_norm": 0.0010135539341717958,
-      "learning_rate": 0.00019999920070646267,
-      "loss": 46.0,
-      "step": 7913
-    },
-    {
-      "epoch": 1.2744877007931077,
-      "grad_norm": 0.0016933500301092863,
-      "learning_rate": 0.00019999920050417416,
-      "loss": 46.0,
-      "step": 7914
-    },
-    {
-      "epoch": 1.274648737871895,
-      "grad_norm": 0.0009722772520035505,
-      "learning_rate": 0.00019999920030186008,
-      "loss": 46.0,
-      "step": 7915
-    },
-    {
-      "epoch": 1.2748097749506824,
-      "grad_norm": 0.0015697190538048744,
-      "learning_rate": 0.0001999992000995204,
-      "loss": 46.0,
-      "step": 7916
-    },
-    {
-      "epoch": 1.2749708120294698,
-      "grad_norm": 0.0006854765233583748,
-      "learning_rate": 0.00019999919989715508,
-      "loss": 46.0,
-      "step": 7917
-    },
-    {
-      "epoch": 1.275131849108257,
-      "grad_norm": 0.0028528235852718353,
-      "learning_rate": 0.0001999991996947642,
-      "loss": 46.0,
-      "step": 7918
-    },
-    {
-      "epoch": 1.2752928861870445,
-      "grad_norm": 0.003435129765421152,
-      "learning_rate": 0.0001999991994923477,
-      "loss": 46.0,
-      "step": 7919
-    },
-    {
-      "epoch": 1.275453923265832,
-      "grad_norm": 0.0017934262286871672,
-      "learning_rate": 0.00019999919928990564,
-      "loss": 46.0,
-      "step": 7920
-    },
-    {
-      "epoch": 1.2756149603446194,
-      "grad_norm": 0.0005804290412925184,
-      "learning_rate": 0.00019999919908743796,
-      "loss": 46.0,
-      "step": 7921
-    },
-    {
-      "epoch": 1.2757759974234069,
-      "grad_norm": 0.004956151824444532,
-      "learning_rate": 0.0001999991988849447,
-      "loss": 46.0,
-      "step": 7922
-    },
-    {
-      "epoch": 1.275937034502194,
-      "grad_norm": 0.0014738744357600808,
-      "learning_rate": 0.00019999919868242584,
-      "loss": 46.0,
-      "step": 7923
-    },
-    {
-      "epoch": 1.2760980715809815,
-      "grad_norm": 0.002577925566583872,
-      "learning_rate": 0.00019999919847988137,
-      "loss": 46.0,
-      "step": 7924
-    },
-    {
-      "epoch": 1.276259108659769,
-      "grad_norm": 0.0007445004303008318,
-      "learning_rate": 0.00019999919827731134,
-      "loss": 46.0,
-      "step": 7925
-    },
-    {
-      "epoch": 1.2764201457385562,
-      "grad_norm": 0.0010513279121369123,
-      "learning_rate": 0.0001999991980747157,
-      "loss": 46.0,
-      "step": 7926
-    },
-    {
-      "epoch": 1.2765811828173437,
-      "grad_norm": 0.0005647443467751145,
-      "learning_rate": 0.00019999919787209444,
-      "loss": 46.0,
-      "step": 7927
-    },
-    {
-      "epoch": 1.2767422198961311,
-      "grad_norm": 0.00068237679079175,
-      "learning_rate": 0.00019999919766944762,
-      "loss": 46.0,
-      "step": 7928
-    },
-    {
-      "epoch": 1.2769032569749186,
-      "grad_norm": 0.011096027679741383,
-      "learning_rate": 0.0001999991974667752,
-      "loss": 46.0,
-      "step": 7929
-    },
-    {
-      "epoch": 1.2770642940537058,
-      "grad_norm": 0.0011830878211185336,
-      "learning_rate": 0.00019999919726407716,
-      "loss": 46.0,
-      "step": 7930
-    },
-    {
-      "epoch": 1.2772253311324933,
-      "grad_norm": 0.0021713164169341326,
-      "learning_rate": 0.00019999919706135356,
-      "loss": 46.0,
-      "step": 7931
-    },
-    {
-      "epoch": 1.2773863682112807,
-      "grad_norm": 0.0003846340987365693,
-      "learning_rate": 0.00019999919685860434,
-      "loss": 46.0,
-      "step": 7932
-    },
-    {
-      "epoch": 1.277547405290068,
-      "grad_norm": 0.0022516839671880007,
-      "learning_rate": 0.00019999919665582953,
-      "loss": 46.0,
-      "step": 7933
-    },
-    {
-      "epoch": 1.2777084423688554,
-      "grad_norm": 0.0006441550212912261,
-      "learning_rate": 0.00019999919645302913,
-      "loss": 46.0,
-      "step": 7934
-    },
-    {
-      "epoch": 1.2778694794476428,
-      "grad_norm": 0.00512974988669157,
-      "learning_rate": 0.00019999919625020312,
-      "loss": 46.0,
-      "step": 7935
-    },
-    {
-      "epoch": 1.2780305165264303,
-      "grad_norm": 0.001224461360834539,
-      "learning_rate": 0.00019999919604735152,
-      "loss": 46.0,
-      "step": 7936
-    },
-    {
-      "epoch": 1.2781915536052175,
-      "grad_norm": 0.0009842407889664173,
-      "learning_rate": 0.00019999919584447434,
-      "loss": 46.0,
-      "step": 7937
-    },
-    {
-      "epoch": 1.278352590684005,
-      "grad_norm": 0.0005291182897053659,
-      "learning_rate": 0.00019999919564157157,
-      "loss": 46.0,
-      "step": 7938
-    },
-    {
-      "epoch": 1.2785136277627924,
-      "grad_norm": 0.007361816707998514,
-      "learning_rate": 0.00019999919543864318,
-      "loss": 46.0,
-      "step": 7939
-    },
-    {
-      "epoch": 1.2786746648415797,
-      "grad_norm": 0.004945671651512384,
-      "learning_rate": 0.0001999991952356892,
-      "loss": 46.0,
-      "step": 7940
-    },
-    {
-      "epoch": 1.278835701920367,
-      "grad_norm": 0.004143549595028162,
-      "learning_rate": 0.00019999919503270968,
-      "loss": 46.0,
-      "step": 7941
-    },
-    {
-      "epoch": 1.2789967389991546,
-      "grad_norm": 0.002231571124866605,
-      "learning_rate": 0.0001999991948297045,
-      "loss": 46.0,
-      "step": 7942
-    },
-    {
-      "epoch": 1.279157776077942,
-      "grad_norm": 0.0015816737432032824,
-      "learning_rate": 0.00019999919462667377,
-      "loss": 46.0,
-      "step": 7943
-    },
-    {
-      "epoch": 1.2793188131567295,
-      "grad_norm": 0.0026737286243587732,
-      "learning_rate": 0.0001999991944236174,
-      "loss": 46.0,
-      "step": 7944
-    },
-    {
-      "epoch": 1.2794798502355167,
-      "grad_norm": 0.003992780111730099,
-      "learning_rate": 0.00019999919422053545,
-      "loss": 46.0,
-      "step": 7945
-    },
-    {
-      "epoch": 1.2796408873143041,
-      "grad_norm": 0.002354080555960536,
-      "learning_rate": 0.0001999991940174279,
-      "loss": 46.0,
-      "step": 7946
-    },
-    {
-      "epoch": 1.2798019243930914,
-      "grad_norm": 0.0004585007845889777,
-      "learning_rate": 0.0001999991938142948,
-      "loss": 46.0,
-      "step": 7947
-    },
-    {
-      "epoch": 1.2799629614718788,
-      "grad_norm": 0.006793987471610308,
-      "learning_rate": 0.00019999919361113607,
-      "loss": 46.0,
-      "step": 7948
-    },
-    {
-      "epoch": 1.2801239985506663,
-      "grad_norm": 0.003035103902220726,
-      "learning_rate": 0.00019999919340795176,
-      "loss": 46.0,
-      "step": 7949
-    },
-    {
-      "epoch": 1.2802850356294537,
-      "grad_norm": 0.003445405513048172,
-      "learning_rate": 0.0001999991932047418,
-      "loss": 46.0,
-      "step": 7950
-    },
-    {
-      "epoch": 1.2804460727082412,
-      "grad_norm": 0.0009236861951649189,
-      "learning_rate": 0.00019999919300150632,
-      "loss": 46.0,
-      "step": 7951
-    },
-    {
-      "epoch": 1.2806071097870284,
-      "grad_norm": 0.004041364882141352,
-      "learning_rate": 0.0001999991927982452,
-      "loss": 46.0,
-      "step": 7952
-    },
-    {
-      "epoch": 1.2807681468658159,
-      "grad_norm": 0.0007511607836931944,
-      "learning_rate": 0.00019999919259495853,
-      "loss": 46.0,
-      "step": 7953
-    },
-    {
-      "epoch": 1.2809291839446033,
-      "grad_norm": 0.0024063477758318186,
-      "learning_rate": 0.00019999919239164623,
-      "loss": 46.0,
-      "step": 7954
-    },
-    {
-      "epoch": 1.2810902210233905,
-      "grad_norm": 0.000695219321642071,
-      "learning_rate": 0.0001999991921883083,
-      "loss": 46.0,
-      "step": 7955
-    },
-    {
-      "epoch": 1.281251258102178,
-      "grad_norm": 0.002616358920931816,
-      "learning_rate": 0.00019999919198494484,
-      "loss": 46.0,
-      "step": 7956
-    },
-    {
-      "epoch": 1.2814122951809654,
-      "grad_norm": 0.0017433814937248826,
-      "learning_rate": 0.00019999919178155575,
-      "loss": 46.0,
-      "step": 7957
-    },
-    {
-      "epoch": 1.281573332259753,
-      "grad_norm": 0.0014205455081537366,
-      "learning_rate": 0.0001999991915781411,
-      "loss": 46.0,
-      "step": 7958
-    },
-    {
-      "epoch": 1.2817343693385401,
-      "grad_norm": 0.000655974552500993,
-      "learning_rate": 0.0001999991913747008,
-      "loss": 46.0,
-      "step": 7959
-    },
-    {
-      "epoch": 1.2818954064173276,
-      "grad_norm": 0.0025297480169683695,
-      "learning_rate": 0.00019999919117123492,
-      "loss": 46.0,
-      "step": 7960
-    },
-    {
-      "epoch": 1.282056443496115,
-      "grad_norm": 0.0021449653431773186,
-      "learning_rate": 0.00019999919096774348,
-      "loss": 46.0,
-      "step": 7961
-    },
-    {
-      "epoch": 1.2822174805749023,
-      "grad_norm": 0.002024073153734207,
-      "learning_rate": 0.00019999919076422643,
-      "loss": 46.0,
-      "step": 7962
-    },
-    {
-      "epoch": 1.2823785176536897,
-      "grad_norm": 0.0012765323044732213,
-      "learning_rate": 0.0001999991905606838,
-      "loss": 46.0,
-      "step": 7963
-    },
-    {
-      "epoch": 1.2825395547324772,
-      "grad_norm": 0.0009103547781705856,
-      "learning_rate": 0.00019999919035711554,
-      "loss": 46.0,
-      "step": 7964
-    },
-    {
-      "epoch": 1.2827005918112646,
-      "grad_norm": 0.0007491717697121203,
-      "learning_rate": 0.00019999919015352172,
-      "loss": 46.0,
-      "step": 7965
-    },
-    {
-      "epoch": 1.282861628890052,
-      "grad_norm": 0.0026717265136539936,
-      "learning_rate": 0.0001999991899499023,
-      "loss": 46.0,
-      "step": 7966
-    },
-    {
-      "epoch": 1.2830226659688393,
-      "grad_norm": 0.0019305183086544275,
-      "learning_rate": 0.00019999918974625725,
-      "loss": 46.0,
-      "step": 7967
-    },
-    {
-      "epoch": 1.2831837030476267,
-      "grad_norm": 0.0021675850730389357,
-      "learning_rate": 0.00019999918954258664,
-      "loss": 46.0,
-      "step": 7968
-    },
-    {
-      "epoch": 1.283344740126414,
-      "grad_norm": 0.0006487573264166713,
-      "learning_rate": 0.00019999918933889043,
-      "loss": 46.0,
-      "step": 7969
-    },
-    {
-      "epoch": 1.2835057772052014,
-      "grad_norm": 0.001013029832392931,
-      "learning_rate": 0.00019999918913516862,
-      "loss": 46.0,
-      "step": 7970
-    },
-    {
-      "epoch": 1.2836668142839889,
-      "grad_norm": 0.001681437948718667,
-      "learning_rate": 0.00019999918893142123,
-      "loss": 46.0,
-      "step": 7971
-    },
-    {
-      "epoch": 1.2838278513627763,
-      "grad_norm": 0.0013573182513937354,
-      "learning_rate": 0.00019999918872764822,
-      "loss": 46.0,
-      "step": 7972
-    },
-    {
-      "epoch": 1.2839888884415638,
-      "grad_norm": 0.00044439014163799584,
-      "learning_rate": 0.0001999991885238496,
-      "loss": 46.0,
-      "step": 7973
-    },
-    {
-      "epoch": 1.284149925520351,
-      "grad_norm": 0.001315410016104579,
-      "learning_rate": 0.00019999918832002545,
-      "loss": 46.0,
-      "step": 7974
-    },
-    {
-      "epoch": 1.2843109625991385,
-      "grad_norm": 0.0022111060097813606,
-      "learning_rate": 0.00019999918811617565,
-      "loss": 46.0,
-      "step": 7975
-    },
-    {
-      "epoch": 1.284471999677926,
-      "grad_norm": 0.0005336496396921575,
-      "learning_rate": 0.00019999918791230024,
-      "loss": 46.0,
-      "step": 7976
-    },
-    {
-      "epoch": 1.2846330367567131,
-      "grad_norm": 0.0023534928914159536,
-      "learning_rate": 0.0001999991877083993,
-      "loss": 46.0,
-      "step": 7977
-    },
-    {
-      "epoch": 1.2847940738355006,
-      "grad_norm": 0.001318909227848053,
-      "learning_rate": 0.00019999918750447272,
-      "loss": 46.0,
-      "step": 7978
-    },
-    {
-      "epoch": 1.284955110914288,
-      "grad_norm": 0.001514313742518425,
-      "learning_rate": 0.00019999918730052057,
-      "loss": 46.0,
-      "step": 7979
-    },
-    {
-      "epoch": 1.2851161479930755,
-      "grad_norm": 0.0018094531260430813,
-      "learning_rate": 0.00019999918709654282,
-      "loss": 46.0,
-      "step": 7980
-    },
-    {
-      "epoch": 1.2852771850718627,
-      "grad_norm": 0.004572336096316576,
-      "learning_rate": 0.00019999918689253944,
-      "loss": 46.0,
-      "step": 7981
-    },
-    {
-      "epoch": 1.2854382221506502,
-      "grad_norm": 0.004093192517757416,
-      "learning_rate": 0.0001999991866885105,
-      "loss": 46.0,
-      "step": 7982
-    },
-    {
-      "epoch": 1.2855992592294376,
-      "grad_norm": 0.0007442169589921832,
-      "learning_rate": 0.00019999918648445596,
-      "loss": 46.0,
-      "step": 7983
-    },
-    {
-      "epoch": 1.2857602963082249,
-      "grad_norm": 0.0020866876002401114,
-      "learning_rate": 0.00019999918628037583,
-      "loss": 46.0,
-      "step": 7984
-    },
-    {
-      "epoch": 1.2859213333870123,
-      "grad_norm": 0.0008816205663606524,
-      "learning_rate": 0.0001999991860762701,
-      "loss": 46.0,
-      "step": 7985
-    },
-    {
-      "epoch": 1.2860823704657998,
-      "grad_norm": 0.001046821940690279,
-      "learning_rate": 0.0001999991858721388,
-      "loss": 46.0,
-      "step": 7986
-    },
-    {
-      "epoch": 1.2862434075445872,
-      "grad_norm": 0.004963348153978586,
-      "learning_rate": 0.00019999918566798188,
-      "loss": 46.0,
-      "step": 7987
-    },
-    {
-      "epoch": 1.2864044446233747,
-      "grad_norm": 0.0036255496088415384,
-      "learning_rate": 0.00019999918546379934,
-      "loss": 46.0,
-      "step": 7988
-    },
-    {
-      "epoch": 1.286565481702162,
-      "grad_norm": 0.0006340309628285468,
-      "learning_rate": 0.00019999918525959124,
-      "loss": 46.0,
-      "step": 7989
-    },
-    {
-      "epoch": 1.2867265187809493,
-      "grad_norm": 0.0018060792936012149,
-      "learning_rate": 0.00019999918505535753,
-      "loss": 46.0,
-      "step": 7990
-    },
-    {
-      "epoch": 1.2868875558597366,
-      "grad_norm": 0.0011644319165498018,
-      "learning_rate": 0.00019999918485109826,
-      "loss": 46.0,
-      "step": 7991
-    },
-    {
-      "epoch": 1.287048592938524,
-      "grad_norm": 0.004299150314182043,
-      "learning_rate": 0.00019999918464681335,
-      "loss": 46.0,
-      "step": 7992
-    },
-    {
-      "epoch": 1.2872096300173115,
-      "grad_norm": 0.0017348442925140262,
-      "learning_rate": 0.00019999918444250288,
-      "loss": 46.0,
-      "step": 7993
-    },
-    {
-      "epoch": 1.287370667096099,
-      "grad_norm": 0.001312283449806273,
-      "learning_rate": 0.00019999918423816676,
-      "loss": 46.0,
-      "step": 7994
-    },
-    {
-      "epoch": 1.2875317041748864,
-      "grad_norm": 0.004243955947458744,
-      "learning_rate": 0.00019999918403380512,
-      "loss": 46.0,
-      "step": 7995
-    },
-    {
-      "epoch": 1.2876927412536736,
-      "grad_norm": 0.0038759559392929077,
-      "learning_rate": 0.00019999918382941783,
-      "loss": 46.0,
-      "step": 7996
-    },
-    {
-      "epoch": 1.287853778332461,
-      "grad_norm": 0.000978336320258677,
-      "learning_rate": 0.00019999918362500498,
-      "loss": 46.0,
-      "step": 7997
-    },
-    {
-      "epoch": 1.2880148154112485,
-      "grad_norm": 0.0017428981373086572,
-      "learning_rate": 0.00019999918342056652,
-      "loss": 46.0,
-      "step": 7998
-    },
-    {
-      "epoch": 1.2881758524900357,
-      "grad_norm": 0.0021736298222094774,
-      "learning_rate": 0.00019999918321610247,
-      "loss": 46.0,
-      "step": 7999
-    },
-    {
-      "epoch": 1.2883368895688232,
-      "grad_norm": 0.0005048715975135565,
-      "learning_rate": 0.0001999991830116128,
-      "loss": 46.0,
-      "step": 8000
-    },
-    {
-      "epoch": 1.2884979266476106,
-      "grad_norm": 0.0011691147228702903,
-      "learning_rate": 0.00019999918280709755,
-      "loss": 46.0,
-      "step": 8001
-    },
-    {
-      "epoch": 1.288658963726398,
-      "grad_norm": 0.003715418977662921,
-      "learning_rate": 0.0001999991826025567,
-      "loss": 46.0,
-      "step": 8002
-    },
-    {
-      "epoch": 1.2888200008051853,
-      "grad_norm": 0.0007326299091801047,
-      "learning_rate": 0.0001999991823979903,
-      "loss": 46.0,
-      "step": 8003
-    },
-    {
-      "epoch": 1.2889810378839728,
-      "grad_norm": 0.0018270571017637849,
-      "learning_rate": 0.00019999918219339825,
-      "loss": 46.0,
-      "step": 8004
-    },
-    {
-      "epoch": 1.2891420749627602,
-      "grad_norm": 0.004157285206019878,
-      "learning_rate": 0.00019999918198878062,
-      "loss": 46.0,
-      "step": 8005
-    },
-    {
-      "epoch": 1.2893031120415475,
-      "grad_norm": 0.0014705565990880132,
-      "learning_rate": 0.00019999918178413743,
-      "loss": 46.0,
-      "step": 8006
-    },
-    {
-      "epoch": 1.289464149120335,
-      "grad_norm": 0.0027261364739388227,
-      "learning_rate": 0.00019999918157946863,
-      "loss": 46.0,
-      "step": 8007
-    },
-    {
-      "epoch": 1.2896251861991224,
-      "grad_norm": 0.002807356882840395,
-      "learning_rate": 0.00019999918137477422,
-      "loss": 46.0,
-      "step": 8008
-    },
-    {
-      "epoch": 1.2897862232779098,
-      "grad_norm": 0.0049120294861495495,
-      "learning_rate": 0.0001999991811700542,
-      "loss": 46.0,
-      "step": 8009
-    },
-    {
-      "epoch": 1.2899472603566973,
-      "grad_norm": 0.00293182791210711,
-      "learning_rate": 0.0001999991809653086,
-      "loss": 46.0,
-      "step": 8010
-    },
-    {
-      "epoch": 1.2901082974354845,
-      "grad_norm": 0.0010758258868008852,
-      "learning_rate": 0.00019999918076053742,
-      "loss": 46.0,
-      "step": 8011
-    },
-    {
-      "epoch": 1.290269334514272,
-      "grad_norm": 0.0014885530108585954,
-      "learning_rate": 0.00019999918055574063,
-      "loss": 46.0,
-      "step": 8012
-    },
-    {
-      "epoch": 1.2904303715930592,
-      "grad_norm": 0.0022727432660758495,
-      "learning_rate": 0.00019999918035091825,
-      "loss": 46.0,
-      "step": 8013
-    },
-    {
-      "epoch": 1.2905914086718466,
-      "grad_norm": 0.000856993836350739,
-      "learning_rate": 0.00019999918014607028,
-      "loss": 46.0,
-      "step": 8014
-    },
-    {
-      "epoch": 1.290752445750634,
-      "grad_norm": 0.0037907164078205824,
-      "learning_rate": 0.0001999991799411967,
-      "loss": 46.0,
-      "step": 8015
-    },
-    {
-      "epoch": 1.2909134828294215,
-      "grad_norm": 0.0027179638855159283,
-      "learning_rate": 0.00019999917973629754,
-      "loss": 46.0,
-      "step": 8016
-    },
-    {
-      "epoch": 1.291074519908209,
-      "grad_norm": 0.0005476229125633836,
-      "learning_rate": 0.00019999917953137278,
-      "loss": 46.0,
-      "step": 8017
-    },
-    {
-      "epoch": 1.2912355569869962,
-      "grad_norm": 0.013994633220136166,
-      "learning_rate": 0.00019999917932642244,
-      "loss": 46.0,
-      "step": 8018
-    },
-    {
-      "epoch": 1.2913965940657837,
-      "grad_norm": 0.0033847116865217686,
-      "learning_rate": 0.00019999917912144648,
-      "loss": 46.0,
-      "step": 8019
-    },
-    {
-      "epoch": 1.2915576311445711,
-      "grad_norm": 0.0005505895824171603,
-      "learning_rate": 0.00019999917891644494,
-      "loss": 46.0,
-      "step": 8020
-    },
-    {
-      "epoch": 1.2917186682233583,
-      "grad_norm": 0.0038216772954910994,
-      "learning_rate": 0.0001999991787114178,
-      "loss": 46.0,
-      "step": 8021
-    },
-    {
-      "epoch": 1.2918797053021458,
-      "grad_norm": 0.002420647768303752,
-      "learning_rate": 0.00019999917850636507,
-      "loss": 46.0,
-      "step": 8022
-    },
-    {
-      "epoch": 1.2920407423809332,
-      "grad_norm": 0.001554326037876308,
-      "learning_rate": 0.00019999917830128676,
-      "loss": 46.0,
-      "step": 8023
-    },
-    {
-      "epoch": 1.2922017794597207,
-      "grad_norm": 0.0006883572787046432,
-      "learning_rate": 0.00019999917809618281,
-      "loss": 46.0,
-      "step": 8024
-    },
-    {
-      "epoch": 1.292362816538508,
-      "grad_norm": 0.008520892821252346,
-      "learning_rate": 0.0001999991778910533,
-      "loss": 46.0,
-      "step": 8025
-    },
-    {
-      "epoch": 1.2925238536172954,
-      "grad_norm": 0.0022111323196440935,
-      "learning_rate": 0.0001999991776858982,
-      "loss": 46.0,
-      "step": 8026
-    },
-    {
-      "epoch": 1.2926848906960828,
-      "grad_norm": 0.0007049944251775742,
-      "learning_rate": 0.0001999991774807175,
-      "loss": 46.0,
-      "step": 8027
-    },
-    {
-      "epoch": 1.29284592777487,
-      "grad_norm": 0.001421093475073576,
-      "learning_rate": 0.0001999991772755112,
-      "loss": 46.0,
-      "step": 8028
-    },
-    {
-      "epoch": 1.2930069648536575,
-      "grad_norm": 0.004672470968216658,
-      "learning_rate": 0.00019999917707027928,
-      "loss": 46.0,
-      "step": 8029
-    },
-    {
-      "epoch": 1.293168001932445,
-      "grad_norm": 0.0018201512284576893,
-      "learning_rate": 0.0001999991768650218,
-      "loss": 46.0,
-      "step": 8030
-    },
-    {
-      "epoch": 1.2933290390112324,
-      "grad_norm": 0.0022638379596173763,
-      "learning_rate": 0.00019999917665973872,
-      "loss": 46.0,
-      "step": 8031
-    },
-    {
-      "epoch": 1.2934900760900196,
-      "grad_norm": 0.00047650974011048675,
-      "learning_rate": 0.00019999917645443005,
-      "loss": 46.0,
-      "step": 8032
-    },
-    {
-      "epoch": 1.293651113168807,
-      "grad_norm": 0.0015986618818715215,
-      "learning_rate": 0.00019999917624909577,
-      "loss": 46.0,
-      "step": 8033
-    },
-    {
-      "epoch": 1.2938121502475946,
-      "grad_norm": 0.0008542933501303196,
-      "learning_rate": 0.0001999991760437359,
-      "loss": 46.0,
-      "step": 8034
-    },
-    {
-      "epoch": 1.2939731873263818,
-      "grad_norm": 0.0021304574329406023,
-      "learning_rate": 0.00019999917583835044,
-      "loss": 46.0,
-      "step": 8035
-    },
-    {
-      "epoch": 1.2941342244051692,
-      "grad_norm": 0.0023620950523763895,
-      "learning_rate": 0.0001999991756329394,
-      "loss": 46.0,
-      "step": 8036
-    },
-    {
-      "epoch": 1.2942952614839567,
-      "grad_norm": 0.004709760192781687,
-      "learning_rate": 0.00019999917542750276,
-      "loss": 46.0,
-      "step": 8037
-    },
-    {
-      "epoch": 1.2944562985627441,
-      "grad_norm": 0.0010166389402002096,
-      "learning_rate": 0.00019999917522204048,
-      "loss": 46.0,
-      "step": 8038
-    },
-    {
-      "epoch": 1.2946173356415316,
-      "grad_norm": 0.0028152933809906244,
-      "learning_rate": 0.00019999917501655262,
-      "loss": 46.0,
-      "step": 8039
-    },
-    {
-      "epoch": 1.2947783727203188,
-      "grad_norm": 0.0018348614685237408,
-      "learning_rate": 0.0001999991748110392,
-      "loss": 46.0,
-      "step": 8040
-    },
-    {
-      "epoch": 1.2949394097991063,
-      "grad_norm": 0.0011196538107469678,
-      "learning_rate": 0.0001999991746055002,
-      "loss": 46.0,
-      "step": 8041
-    },
-    {
-      "epoch": 1.2951004468778935,
-      "grad_norm": 0.0016232288908213377,
-      "learning_rate": 0.00019999917439993557,
-      "loss": 46.0,
-      "step": 8042
-    },
-    {
-      "epoch": 1.295261483956681,
-      "grad_norm": 0.004233572166413069,
-      "learning_rate": 0.00019999917419434536,
-      "loss": 46.0,
-      "step": 8043
-    },
-    {
-      "epoch": 1.2954225210354684,
-      "grad_norm": 0.0019040943589061499,
-      "learning_rate": 0.0001999991739887295,
-      "loss": 46.0,
-      "step": 8044
-    },
-    {
-      "epoch": 1.2955835581142559,
-      "grad_norm": 0.003112334990873933,
-      "learning_rate": 0.0001999991737830881,
-      "loss": 46.0,
-      "step": 8045
-    },
-    {
-      "epoch": 1.2957445951930433,
-      "grad_norm": 0.0008003904949873686,
-      "learning_rate": 0.0001999991735774211,
-      "loss": 46.0,
-      "step": 8046
-    },
-    {
-      "epoch": 1.2959056322718305,
-      "grad_norm": 0.0007273473311215639,
-      "learning_rate": 0.0001999991733717285,
-      "loss": 46.0,
-      "step": 8047
-    },
-    {
-      "epoch": 1.296066669350618,
-      "grad_norm": 0.0004309994983486831,
-      "learning_rate": 0.00019999917316601033,
-      "loss": 46.0,
-      "step": 8048
-    },
-    {
-      "epoch": 1.2962277064294054,
-      "grad_norm": 0.003821717808023095,
-      "learning_rate": 0.00019999917296026655,
-      "loss": 46.0,
-      "step": 8049
-    },
-    {
-      "epoch": 1.2963887435081927,
-      "grad_norm": 0.0004929801216349006,
-      "learning_rate": 0.00019999917275449714,
-      "loss": 46.0,
-      "step": 8050
-    },
-    {
-      "epoch": 1.2965497805869801,
-      "grad_norm": 0.0012608777033165097,
-      "learning_rate": 0.00019999917254870218,
-      "loss": 46.0,
-      "step": 8051
-    },
-    {
-      "epoch": 1.2967108176657676,
-      "grad_norm": 0.0006955444696359336,
-      "learning_rate": 0.0001999991723428816,
-      "loss": 46.0,
-      "step": 8052
-    },
-    {
-      "epoch": 1.296871854744555,
-      "grad_norm": 0.0009961365722119808,
-      "learning_rate": 0.00019999917213703544,
-      "loss": 46.0,
-      "step": 8053
-    },
-    {
-      "epoch": 1.2970328918233422,
-      "grad_norm": 0.0014007980935275555,
-      "learning_rate": 0.00019999917193116366,
-      "loss": 46.0,
-      "step": 8054
-    },
-    {
-      "epoch": 1.2971939289021297,
-      "grad_norm": 0.0021253363229334354,
-      "learning_rate": 0.00019999917172526632,
-      "loss": 46.0,
-      "step": 8055
-    },
-    {
-      "epoch": 1.2973549659809172,
-      "grad_norm": 0.001655158819630742,
-      "learning_rate": 0.00019999917151934335,
-      "loss": 46.0,
-      "step": 8056
-    },
-    {
-      "epoch": 1.2975160030597044,
-      "grad_norm": 0.0010083351517096162,
-      "learning_rate": 0.0001999991713133948,
-      "loss": 46.0,
-      "step": 8057
-    },
-    {
-      "epoch": 1.2976770401384918,
-      "grad_norm": 0.000745421159081161,
-      "learning_rate": 0.00019999917110742068,
-      "loss": 46.0,
-      "step": 8058
-    },
-    {
-      "epoch": 1.2978380772172793,
-      "grad_norm": 0.008971952833235264,
-      "learning_rate": 0.00019999917090142094,
-      "loss": 46.0,
-      "step": 8059
-    },
-    {
-      "epoch": 1.2979991142960667,
-      "grad_norm": 0.006118796765804291,
-      "learning_rate": 0.0001999991706953956,
-      "loss": 46.0,
-      "step": 8060
-    },
-    {
-      "epoch": 1.2981601513748542,
-      "grad_norm": 0.004308641888201237,
-      "learning_rate": 0.0001999991704893447,
-      "loss": 46.0,
-      "step": 8061
-    },
-    {
-      "epoch": 1.2983211884536414,
-      "grad_norm": 0.005008167587220669,
-      "learning_rate": 0.00019999917028326817,
-      "loss": 46.0,
-      "step": 8062
-    },
-    {
-      "epoch": 1.2984822255324289,
-      "grad_norm": 0.0022040249314159155,
-      "learning_rate": 0.00019999917007716608,
-      "loss": 46.0,
-      "step": 8063
-    },
-    {
-      "epoch": 1.298643262611216,
-      "grad_norm": 0.0010592591715976596,
-      "learning_rate": 0.00019999916987103835,
-      "loss": 46.0,
-      "step": 8064
-    },
-    {
-      "epoch": 1.2988042996900035,
-      "grad_norm": 0.00405563460662961,
-      "learning_rate": 0.00019999916966488505,
-      "loss": 46.0,
-      "step": 8065
-    },
-    {
-      "epoch": 1.298965336768791,
-      "grad_norm": 0.0020846552215516567,
-      "learning_rate": 0.00019999916945870615,
-      "loss": 46.0,
-      "step": 8066
-    },
-    {
-      "epoch": 1.2991263738475785,
-      "grad_norm": 0.00427275849506259,
-      "learning_rate": 0.00019999916925250166,
-      "loss": 46.0,
-      "step": 8067
-    },
-    {
-      "epoch": 1.299287410926366,
-      "grad_norm": 0.0016410695388913155,
-      "learning_rate": 0.00019999916904627158,
-      "loss": 46.0,
-      "step": 8068
-    },
-    {
-      "epoch": 1.2994484480051531,
-      "grad_norm": 0.004746242426335812,
-      "learning_rate": 0.00019999916884001588,
-      "loss": 46.0,
-      "step": 8069
-    },
-    {
-      "epoch": 1.2996094850839406,
-      "grad_norm": 0.002526788739487529,
-      "learning_rate": 0.00019999916863373463,
-      "loss": 46.0,
-      "step": 8070
-    },
-    {
-      "epoch": 1.299770522162728,
-      "grad_norm": 0.002641509985551238,
-      "learning_rate": 0.00019999916842742773,
-      "loss": 46.0,
-      "step": 8071
-    },
-    {
-      "epoch": 1.2999315592415153,
-      "grad_norm": 0.0013726758770644665,
-      "learning_rate": 0.00019999916822109528,
-      "loss": 46.0,
-      "step": 8072
-    },
-    {
-      "epoch": 1.3000925963203027,
-      "grad_norm": 0.0021210613194853067,
-      "learning_rate": 0.00019999916801473724,
-      "loss": 46.0,
-      "step": 8073
-    },
-    {
-      "epoch": 1.3002536333990902,
-      "grad_norm": 0.0007135680061765015,
-      "learning_rate": 0.00019999916780835358,
-      "loss": 46.0,
-      "step": 8074
-    },
-    {
-      "epoch": 1.3004146704778776,
-      "grad_norm": 0.0015486666234210134,
-      "learning_rate": 0.00019999916760194433,
-      "loss": 46.0,
-      "step": 8075
-    },
-    {
-      "epoch": 1.3005757075566649,
-      "grad_norm": 0.0015272445743903518,
-      "learning_rate": 0.00019999916739550948,
-      "loss": 46.0,
-      "step": 8076
-    },
-    {
-      "epoch": 1.3007367446354523,
-      "grad_norm": 0.0007526118424721062,
-      "learning_rate": 0.00019999916718904906,
-      "loss": 46.0,
-      "step": 8077
-    },
-    {
-      "epoch": 1.3008977817142398,
-      "grad_norm": 0.006024092901498079,
-      "learning_rate": 0.00019999916698256302,
-      "loss": 46.0,
-      "step": 8078
-    },
-    {
-      "epoch": 1.301058818793027,
-      "grad_norm": 0.0013033177237957716,
-      "learning_rate": 0.00019999916677605138,
-      "loss": 46.0,
-      "step": 8079
-    },
-    {
-      "epoch": 1.3012198558718144,
-      "grad_norm": 0.000533826241735369,
-      "learning_rate": 0.00019999916656951417,
-      "loss": 46.0,
-      "step": 8080
-    },
-    {
-      "epoch": 1.3013808929506019,
-      "grad_norm": 0.0009822201682254672,
-      "learning_rate": 0.00019999916636295132,
-      "loss": 46.0,
-      "step": 8081
-    },
-    {
-      "epoch": 1.3015419300293893,
-      "grad_norm": 0.0017239375738427043,
-      "learning_rate": 0.00019999916615636294,
-      "loss": 46.0,
-      "step": 8082
-    },
-    {
-      "epoch": 1.3017029671081768,
-      "grad_norm": 0.0006431580986827612,
-      "learning_rate": 0.00019999916594974892,
-      "loss": 46.0,
-      "step": 8083
-    },
-    {
-      "epoch": 1.301864004186964,
-      "grad_norm": 0.0009649729472585022,
-      "learning_rate": 0.00019999916574310934,
-      "loss": 46.0,
-      "step": 8084
-    },
-    {
-      "epoch": 1.3020250412657515,
-      "grad_norm": 0.0007157270447351038,
-      "learning_rate": 0.00019999916553644414,
-      "loss": 46.0,
-      "step": 8085
-    },
-    {
-      "epoch": 1.3021860783445387,
-      "grad_norm": 0.002580914180725813,
-      "learning_rate": 0.00019999916532975335,
-      "loss": 46.0,
-      "step": 8086
-    },
-    {
-      "epoch": 1.3023471154233262,
-      "grad_norm": 0.002775158267468214,
-      "learning_rate": 0.00019999916512303696,
-      "loss": 46.0,
-      "step": 8087
-    },
-    {
-      "epoch": 1.3025081525021136,
-      "grad_norm": 0.002985249971970916,
-      "learning_rate": 0.00019999916491629497,
-      "loss": 46.0,
-      "step": 8088
-    },
-    {
-      "epoch": 1.302669189580901,
-      "grad_norm": 0.0054305437952280045,
-      "learning_rate": 0.0001999991647095274,
-      "loss": 46.0,
-      "step": 8089
-    },
-    {
-      "epoch": 1.3028302266596885,
-      "grad_norm": 0.0016031991690397263,
-      "learning_rate": 0.00019999916450273424,
-      "loss": 46.0,
-      "step": 8090
-    },
-    {
-      "epoch": 1.3029912637384757,
-      "grad_norm": 0.00307249603793025,
-      "learning_rate": 0.00019999916429591546,
-      "loss": 46.0,
-      "step": 8091
-    },
-    {
-      "epoch": 1.3031523008172632,
-      "grad_norm": 0.0022639601957052946,
-      "learning_rate": 0.0001999991640890711,
-      "loss": 46.0,
-      "step": 8092
-    },
-    {
-      "epoch": 1.3033133378960506,
-      "grad_norm": 0.0016816904535517097,
-      "learning_rate": 0.00019999916388220115,
-      "loss": 46.0,
-      "step": 8093
-    },
-    {
-      "epoch": 1.3034743749748379,
-      "grad_norm": 0.0011698901653289795,
-      "learning_rate": 0.00019999916367530562,
-      "loss": 46.0,
-      "step": 8094
-    },
-    {
-      "epoch": 1.3036354120536253,
-      "grad_norm": 0.0005653820117004216,
-      "learning_rate": 0.00019999916346838447,
-      "loss": 46.0,
-      "step": 8095
-    },
-    {
-      "epoch": 1.3037964491324128,
-      "grad_norm": 0.0018024810124188662,
-      "learning_rate": 0.00019999916326143776,
-      "loss": 46.0,
-      "step": 8096
-    },
-    {
-      "epoch": 1.3039574862112002,
-      "grad_norm": 0.0014141090214252472,
-      "learning_rate": 0.0001999991630544654,
-      "loss": 46.0,
-      "step": 8097
-    },
-    {
-      "epoch": 1.3041185232899875,
-      "grad_norm": 0.0006488273502327502,
-      "learning_rate": 0.00019999916284746747,
-      "loss": 46.0,
-      "step": 8098
-    },
-    {
-      "epoch": 1.304279560368775,
-      "grad_norm": 0.0029810857959091663,
-      "learning_rate": 0.00019999916264044397,
-      "loss": 46.0,
-      "step": 8099
-    },
-    {
-      "epoch": 1.3044405974475624,
-      "grad_norm": 0.002607325091958046,
-      "learning_rate": 0.00019999916243339485,
-      "loss": 46.0,
-      "step": 8100
-    },
-    {
-      "epoch": 1.3046016345263496,
-      "grad_norm": 0.0005921588744968176,
-      "learning_rate": 0.00019999916222632013,
-      "loss": 46.0,
-      "step": 8101
-    },
-    {
-      "epoch": 1.304762671605137,
-      "grad_norm": 0.0008126826141960919,
-      "learning_rate": 0.0001999991620192198,
-      "loss": 46.0,
-      "step": 8102
-    },
-    {
-      "epoch": 1.3049237086839245,
-      "grad_norm": 0.001323246629908681,
-      "learning_rate": 0.00019999916181209394,
-      "loss": 46.0,
-      "step": 8103
-    },
-    {
-      "epoch": 1.305084745762712,
-      "grad_norm": 0.006293023470789194,
-      "learning_rate": 0.00019999916160494242,
-      "loss": 46.0,
-      "step": 8104
-    },
-    {
-      "epoch": 1.3052457828414994,
-      "grad_norm": 0.0030664147343486547,
-      "learning_rate": 0.00019999916139776534,
-      "loss": 46.0,
-      "step": 8105
-    },
-    {
-      "epoch": 1.3054068199202866,
-      "grad_norm": 0.0019583632238209248,
-      "learning_rate": 0.00019999916119056268,
-      "loss": 46.0,
-      "step": 8106
-    },
-    {
-      "epoch": 1.305567856999074,
-      "grad_norm": 0.0006180562195368111,
-      "learning_rate": 0.0001999991609833344,
-      "loss": 46.0,
-      "step": 8107
-    },
-    {
-      "epoch": 1.3057288940778613,
-      "grad_norm": 0.001685330760665238,
-      "learning_rate": 0.0001999991607760805,
-      "loss": 46.0,
-      "step": 8108
-    },
-    {
-      "epoch": 1.3058899311566488,
-      "grad_norm": 0.0012558940798044205,
-      "learning_rate": 0.00019999916056880106,
-      "loss": 46.0,
-      "step": 8109
-    },
-    {
-      "epoch": 1.3060509682354362,
-      "grad_norm": 0.0027943886816501617,
-      "learning_rate": 0.000199999160361496,
-      "loss": 46.0,
-      "step": 8110
-    },
-    {
-      "epoch": 1.3062120053142237,
-      "grad_norm": 0.005581276956945658,
-      "learning_rate": 0.00019999916015416534,
-      "loss": 46.0,
-      "step": 8111
-    },
-    {
-      "epoch": 1.306373042393011,
-      "grad_norm": 0.0030407344456762075,
-      "learning_rate": 0.0001999991599468091,
-      "loss": 46.0,
-      "step": 8112
-    },
-    {
-      "epoch": 1.3065340794717983,
-      "grad_norm": 0.0007889422122389078,
-      "learning_rate": 0.00019999915973942725,
-      "loss": 46.0,
-      "step": 8113
-    },
-    {
-      "epoch": 1.3066951165505858,
-      "grad_norm": 0.007997584529221058,
-      "learning_rate": 0.0001999991595320198,
-      "loss": 46.0,
-      "step": 8114
-    },
-    {
-      "epoch": 1.3068561536293732,
-      "grad_norm": 0.0024954320397228003,
-      "learning_rate": 0.00019999915932458677,
-      "loss": 46.0,
-      "step": 8115
-    },
-    {
-      "epoch": 1.3070171907081605,
-      "grad_norm": 0.0025826154742389917,
-      "learning_rate": 0.00019999915911712813,
-      "loss": 46.0,
-      "step": 8116
-    },
-    {
-      "epoch": 1.307178227786948,
-      "grad_norm": 0.000843958870973438,
-      "learning_rate": 0.0001999991589096439,
-      "loss": 46.0,
-      "step": 8117
-    },
-    {
-      "epoch": 1.3073392648657354,
-      "grad_norm": 0.002617934951558709,
-      "learning_rate": 0.00019999915870213408,
-      "loss": 46.0,
-      "step": 8118
-    },
-    {
-      "epoch": 1.3075003019445228,
-      "grad_norm": 0.0013686921447515488,
-      "learning_rate": 0.00019999915849459868,
-      "loss": 46.0,
-      "step": 8119
-    },
-    {
-      "epoch": 1.30766133902331,
-      "grad_norm": 0.0015199461486190557,
-      "learning_rate": 0.00019999915828703766,
-      "loss": 46.0,
-      "step": 8120
-    },
-    {
-      "epoch": 1.3078223761020975,
-      "grad_norm": 0.00356725649908185,
-      "learning_rate": 0.00019999915807945105,
-      "loss": 46.0,
-      "step": 8121
-    },
-    {
-      "epoch": 1.307983413180885,
-      "grad_norm": 0.002082472899928689,
-      "learning_rate": 0.00019999915787183886,
-      "loss": 46.0,
-      "step": 8122
-    },
-    {
-      "epoch": 1.3081444502596722,
-      "grad_norm": 0.0005661632167175412,
-      "learning_rate": 0.00019999915766420108,
-      "loss": 46.0,
-      "step": 8123
-    },
-    {
-      "epoch": 1.3083054873384596,
-      "grad_norm": 0.0008750281413085759,
-      "learning_rate": 0.00019999915745653768,
-      "loss": 46.0,
-      "step": 8124
-    },
-    {
-      "epoch": 1.308466524417247,
-      "grad_norm": 0.002260432578623295,
-      "learning_rate": 0.0001999991572488487,
-      "loss": 46.0,
-      "step": 8125
-    },
-    {
-      "epoch": 1.3086275614960345,
-      "grad_norm": 0.0023060941603034735,
-      "learning_rate": 0.00019999915704113413,
-      "loss": 46.0,
-      "step": 8126
-    },
-    {
-      "epoch": 1.3087885985748218,
-      "grad_norm": 0.0011326707899570465,
-      "learning_rate": 0.00019999915683339398,
-      "loss": 46.0,
-      "step": 8127
-    },
-    {
-      "epoch": 1.3089496356536092,
-      "grad_norm": 0.006058217026293278,
-      "learning_rate": 0.0001999991566256282,
-      "loss": 46.0,
-      "step": 8128
-    },
-    {
-      "epoch": 1.3091106727323967,
-      "grad_norm": 0.002363781677559018,
-      "learning_rate": 0.00019999915641783682,
-      "loss": 46.0,
-      "step": 8129
-    },
-    {
-      "epoch": 1.309271709811184,
-      "grad_norm": 0.004301850683987141,
-      "learning_rate": 0.00019999915621001988,
-      "loss": 46.0,
-      "step": 8130
-    },
-    {
-      "epoch": 1.3094327468899714,
-      "grad_norm": 0.001516456832177937,
-      "learning_rate": 0.00019999915600217732,
-      "loss": 46.0,
-      "step": 8131
-    },
-    {
-      "epoch": 1.3095937839687588,
-      "grad_norm": 0.005582747980952263,
-      "learning_rate": 0.0001999991557943092,
-      "loss": 46.0,
-      "step": 8132
-    },
-    {
-      "epoch": 1.3097548210475463,
-      "grad_norm": 0.008259088732302189,
-      "learning_rate": 0.00019999915558641544,
-      "loss": 46.0,
-      "step": 8133
-    },
-    {
-      "epoch": 1.3099158581263337,
-      "grad_norm": 0.0004960817750543356,
-      "learning_rate": 0.00019999915537849611,
-      "loss": 46.0,
-      "step": 8134
-    },
-    {
-      "epoch": 1.310076895205121,
-      "grad_norm": 0.0035664236638695,
-      "learning_rate": 0.00019999915517055118,
-      "loss": 46.0,
-      "step": 8135
-    },
-    {
-      "epoch": 1.3102379322839084,
-      "grad_norm": 0.002415742026641965,
-      "learning_rate": 0.00019999915496258066,
-      "loss": 46.0,
-      "step": 8136
-    },
-    {
-      "epoch": 1.3103989693626958,
-      "grad_norm": 0.0005769507260993123,
-      "learning_rate": 0.00019999915475458452,
-      "loss": 46.0,
-      "step": 8137
-    },
-    {
-      "epoch": 1.310560006441483,
-      "grad_norm": 0.00028646091232076287,
-      "learning_rate": 0.00019999915454656285,
-      "loss": 46.0,
-      "step": 8138
-    },
-    {
-      "epoch": 1.3107210435202705,
-      "grad_norm": 0.0006671464652754366,
-      "learning_rate": 0.0001999991543385155,
-      "loss": 46.0,
-      "step": 8139
-    },
-    {
-      "epoch": 1.310882080599058,
-      "grad_norm": 0.0017367142718285322,
-      "learning_rate": 0.00019999915413044264,
-      "loss": 46.0,
-      "step": 8140
-    },
-    {
-      "epoch": 1.3110431176778454,
-      "grad_norm": 0.01110994815826416,
-      "learning_rate": 0.00019999915392234413,
-      "loss": 46.0,
-      "step": 8141
-    },
-    {
-      "epoch": 1.3112041547566327,
-      "grad_norm": 0.0017920555546879768,
-      "learning_rate": 0.00019999915371422003,
-      "loss": 46.0,
-      "step": 8142
-    },
-    {
-      "epoch": 1.31136519183542,
-      "grad_norm": 0.0013599625090137124,
-      "learning_rate": 0.00019999915350607034,
-      "loss": 46.0,
-      "step": 8143
-    },
-    {
-      "epoch": 1.3115262289142076,
-      "grad_norm": 0.003930533304810524,
-      "learning_rate": 0.00019999915329789507,
-      "loss": 46.0,
-      "step": 8144
-    },
-    {
-      "epoch": 1.3116872659929948,
-      "grad_norm": 0.002713278168812394,
-      "learning_rate": 0.00019999915308969418,
-      "loss": 46.0,
-      "step": 8145
-    },
-    {
-      "epoch": 1.3118483030717822,
-      "grad_norm": 0.0009364794241264462,
-      "learning_rate": 0.0001999991528814677,
-      "loss": 46.0,
-      "step": 8146
-    },
-    {
-      "epoch": 1.3120093401505697,
-      "grad_norm": 0.002139805117622018,
-      "learning_rate": 0.00019999915267321567,
-      "loss": 46.0,
-      "step": 8147
-    },
-    {
-      "epoch": 1.3121703772293571,
-      "grad_norm": 0.004375578835606575,
-      "learning_rate": 0.00019999915246493802,
-      "loss": 46.0,
-      "step": 8148
-    },
-    {
-      "epoch": 1.3123314143081444,
-      "grad_norm": 0.0029165996238589287,
-      "learning_rate": 0.00019999915225663476,
-      "loss": 46.0,
-      "step": 8149
-    },
-    {
-      "epoch": 1.3124924513869318,
-      "grad_norm": 0.0011333934962749481,
-      "learning_rate": 0.0001999991520483059,
-      "loss": 46.0,
-      "step": 8150
-    },
-    {
-      "epoch": 1.3126534884657193,
-      "grad_norm": 0.0010142400860786438,
-      "learning_rate": 0.00019999915183995147,
-      "loss": 46.0,
-      "step": 8151
-    },
-    {
-      "epoch": 1.3128145255445065,
-      "grad_norm": 0.0018940945155918598,
-      "learning_rate": 0.00019999915163157144,
-      "loss": 46.0,
-      "step": 8152
-    },
-    {
-      "epoch": 1.312975562623294,
-      "grad_norm": 0.0011889304732903838,
-      "learning_rate": 0.00019999915142316578,
-      "loss": 46.0,
-      "step": 8153
-    },
-    {
-      "epoch": 1.3131365997020814,
-      "grad_norm": 0.001042266609147191,
-      "learning_rate": 0.00019999915121473455,
-      "loss": 46.0,
-      "step": 8154
-    },
-    {
-      "epoch": 1.3132976367808689,
-      "grad_norm": 0.002718703355640173,
-      "learning_rate": 0.00019999915100627776,
-      "loss": 46.0,
-      "step": 8155
-    },
-    {
-      "epoch": 1.3134586738596563,
-      "grad_norm": 0.0007215013029053807,
-      "learning_rate": 0.00019999915079779533,
-      "loss": 46.0,
-      "step": 8156
-    },
-    {
-      "epoch": 1.3136197109384435,
-      "grad_norm": 0.0013306067558005452,
-      "learning_rate": 0.00019999915058928732,
-      "loss": 46.0,
-      "step": 8157
-    },
-    {
-      "epoch": 1.313780748017231,
-      "grad_norm": 0.0013448548270389438,
-      "learning_rate": 0.00019999915038075371,
-      "loss": 46.0,
-      "step": 8158
-    },
-    {
-      "epoch": 1.3139417850960182,
-      "grad_norm": 0.002382321050390601,
-      "learning_rate": 0.0001999991501721945,
-      "loss": 46.0,
-      "step": 8159
-    },
-    {
-      "epoch": 1.3141028221748057,
-      "grad_norm": 0.000424124562414363,
-      "learning_rate": 0.00019999914996360972,
-      "loss": 46.0,
-      "step": 8160
-    },
-    {
-      "epoch": 1.3142638592535931,
-      "grad_norm": 0.0051336102187633514,
-      "learning_rate": 0.00019999914975499933,
-      "loss": 46.0,
-      "step": 8161
-    },
-    {
-      "epoch": 1.3144248963323806,
-      "grad_norm": 0.0005932336207479239,
-      "learning_rate": 0.00019999914954636335,
-      "loss": 46.0,
-      "step": 8162
-    },
-    {
-      "epoch": 1.314585933411168,
-      "grad_norm": 0.0005597893614321947,
-      "learning_rate": 0.00019999914933770176,
-      "loss": 46.0,
-      "step": 8163
-    },
-    {
-      "epoch": 1.3147469704899553,
-      "grad_norm": 0.0035444789100438356,
-      "learning_rate": 0.0001999991491290146,
-      "loss": 46.0,
-      "step": 8164
-    },
-    {
-      "epoch": 1.3149080075687427,
-      "grad_norm": 0.00033637703745625913,
-      "learning_rate": 0.00019999914892030184,
-      "loss": 46.0,
-      "step": 8165
-    },
-    {
-      "epoch": 1.3150690446475302,
-      "grad_norm": 0.0006662870291620493,
-      "learning_rate": 0.00019999914871156346,
-      "loss": 46.0,
-      "step": 8166
-    },
-    {
-      "epoch": 1.3152300817263174,
-      "grad_norm": 0.0007983579998835921,
-      "learning_rate": 0.00019999914850279952,
-      "loss": 46.0,
-      "step": 8167
-    },
-    {
-      "epoch": 1.3153911188051048,
-      "grad_norm": 0.004298283252865076,
-      "learning_rate": 0.00019999914829400996,
-      "loss": 46.0,
-      "step": 8168
-    },
-    {
-      "epoch": 1.3155521558838923,
-      "grad_norm": 0.002112639369443059,
-      "learning_rate": 0.00019999914808519482,
-      "loss": 46.0,
-      "step": 8169
-    },
-    {
-      "epoch": 1.3157131929626797,
-      "grad_norm": 0.005377735011279583,
-      "learning_rate": 0.00019999914787635406,
-      "loss": 46.0,
-      "step": 8170
-    },
-    {
-      "epoch": 1.315874230041467,
-      "grad_norm": 0.0009230033028870821,
-      "learning_rate": 0.00019999914766748774,
-      "loss": 46.0,
-      "step": 8171
-    },
-    {
-      "epoch": 1.3160352671202544,
-      "grad_norm": 0.0005012241890653968,
-      "learning_rate": 0.0001999991474585958,
-      "loss": 46.0,
-      "step": 8172
-    },
-    {
-      "epoch": 1.3161963041990419,
-      "grad_norm": 0.004119296092540026,
-      "learning_rate": 0.00019999914724967826,
-      "loss": 46.0,
-      "step": 8173
-    },
-    {
-      "epoch": 1.316357341277829,
-      "grad_norm": 0.0025279750116169453,
-      "learning_rate": 0.00019999914704073516,
-      "loss": 46.0,
-      "step": 8174
-    },
-    {
-      "epoch": 1.3165183783566166,
-      "grad_norm": 0.0009732013568282127,
-      "learning_rate": 0.00019999914683176644,
-      "loss": 46.0,
-      "step": 8175
-    },
-    {
-      "epoch": 1.316679415435404,
-      "grad_norm": 0.0019259214168414474,
-      "learning_rate": 0.00019999914662277216,
-      "loss": 46.0,
-      "step": 8176
-    },
-    {
-      "epoch": 1.3168404525141915,
-      "grad_norm": 0.0021715043112635612,
-      "learning_rate": 0.00019999914641375224,
-      "loss": 46.0,
-      "step": 8177
-    },
-    {
-      "epoch": 1.317001489592979,
-      "grad_norm": 0.003505889791995287,
-      "learning_rate": 0.00019999914620470676,
-      "loss": 46.0,
-      "step": 8178
-    },
-    {
-      "epoch": 1.3171625266717661,
-      "grad_norm": 0.0014351534191519022,
-      "learning_rate": 0.00019999914599563563,
-      "loss": 46.0,
-      "step": 8179
-    },
-    {
-      "epoch": 1.3173235637505536,
-      "grad_norm": 0.0009196783066727221,
-      "learning_rate": 0.00019999914578653895,
-      "loss": 46.0,
-      "step": 8180
-    },
-    {
-      "epoch": 1.3174846008293408,
-      "grad_norm": 0.0009344897698611021,
-      "learning_rate": 0.00019999914557741665,
-      "loss": 46.0,
-      "step": 8181
-    },
-    {
-      "epoch": 1.3176456379081283,
-      "grad_norm": 0.0017691469984129071,
-      "learning_rate": 0.0001999991453682688,
-      "loss": 46.0,
-      "step": 8182
-    },
-    {
-      "epoch": 1.3178066749869157,
-      "grad_norm": 0.001179224462248385,
-      "learning_rate": 0.00019999914515909532,
-      "loss": 46.0,
-      "step": 8183
-    },
-    {
-      "epoch": 1.3179677120657032,
-      "grad_norm": 0.0020149152260273695,
-      "learning_rate": 0.00019999914494989626,
-      "loss": 46.0,
-      "step": 8184
-    },
-    {
-      "epoch": 1.3181287491444906,
-      "grad_norm": 0.002091505564749241,
-      "learning_rate": 0.0001999991447406716,
-      "loss": 46.0,
-      "step": 8185
-    },
-    {
-      "epoch": 1.3182897862232779,
-      "grad_norm": 0.0004325432819314301,
-      "learning_rate": 0.00019999914453142136,
-      "loss": 46.0,
-      "step": 8186
-    },
-    {
-      "epoch": 1.3184508233020653,
-      "grad_norm": 0.0008748378604650497,
-      "learning_rate": 0.00019999914432214548,
-      "loss": 46.0,
-      "step": 8187
-    },
-    {
-      "epoch": 1.3186118603808528,
-      "grad_norm": 0.0022790180519223213,
-      "learning_rate": 0.00019999914411284405,
-      "loss": 46.0,
-      "step": 8188
-    },
-    {
-      "epoch": 1.31877289745964,
-      "grad_norm": 0.0016114182071760297,
-      "learning_rate": 0.00019999914390351703,
-      "loss": 46.0,
-      "step": 8189
-    },
-    {
-      "epoch": 1.3189339345384274,
-      "grad_norm": 0.0009810769697651267,
-      "learning_rate": 0.00019999914369416436,
-      "loss": 46.0,
-      "step": 8190
-    },
-    {
-      "epoch": 1.319094971617215,
-      "grad_norm": 0.0019941620994359255,
-      "learning_rate": 0.00019999914348478614,
-      "loss": 46.0,
-      "step": 8191
-    },
-    {
-      "epoch": 1.3192560086960023,
-      "grad_norm": 0.0012314112391322851,
-      "learning_rate": 0.00019999914327538233,
-      "loss": 46.0,
-      "step": 8192
-    },
-    {
-      "epoch": 1.3194170457747896,
-      "grad_norm": 0.0009301109821535647,
-      "learning_rate": 0.0001999991430659529,
-      "loss": 46.0,
-      "step": 8193
-    },
-    {
-      "epoch": 1.319578082853577,
-      "grad_norm": 0.005204427987337112,
-      "learning_rate": 0.00019999914285649787,
-      "loss": 46.0,
-      "step": 8194
-    },
-    {
-      "epoch": 1.3197391199323645,
-      "grad_norm": 0.002175047993659973,
-      "learning_rate": 0.0001999991426470173,
-      "loss": 46.0,
-      "step": 8195
-    },
-    {
-      "epoch": 1.3199001570111517,
-      "grad_norm": 0.004512629471719265,
-      "learning_rate": 0.00019999914243751108,
-      "loss": 46.0,
-      "step": 8196
-    },
-    {
-      "epoch": 1.3200611940899392,
-      "grad_norm": 0.00025794017710722983,
-      "learning_rate": 0.00019999914222797928,
-      "loss": 46.0,
-      "step": 8197
-    },
-    {
-      "epoch": 1.3202222311687266,
-      "grad_norm": 0.00371658056974411,
-      "learning_rate": 0.0001999991420184219,
-      "loss": 46.0,
-      "step": 8198
-    },
-    {
-      "epoch": 1.320383268247514,
-      "grad_norm": 0.0019222635310143232,
-      "learning_rate": 0.0001999991418088389,
-      "loss": 46.0,
-      "step": 8199
-    },
-    {
-      "epoch": 1.3205443053263015,
-      "grad_norm": 0.0017025705892592669,
-      "learning_rate": 0.00019999914159923035,
-      "loss": 46.0,
-      "step": 8200
-    },
-    {
-      "epoch": 1.3207053424050887,
-      "grad_norm": 0.002271057805046439,
-      "learning_rate": 0.00019999914138959618,
-      "loss": 46.0,
-      "step": 8201
-    },
-    {
-      "epoch": 1.3208663794838762,
-      "grad_norm": 0.001219138503074646,
-      "learning_rate": 0.0001999991411799364,
-      "loss": 46.0,
-      "step": 8202
-    },
-    {
-      "epoch": 1.3210274165626634,
-      "grad_norm": 0.004323808941990137,
-      "learning_rate": 0.00019999914097025104,
-      "loss": 46.0,
-      "step": 8203
-    },
-    {
-      "epoch": 1.3211884536414509,
-      "grad_norm": 0.0010273369261994958,
-      "learning_rate": 0.00019999914076054007,
-      "loss": 46.0,
-      "step": 8204
-    },
-    {
-      "epoch": 1.3213494907202383,
-      "grad_norm": 0.002005660906434059,
-      "learning_rate": 0.00019999914055080352,
-      "loss": 46.0,
-      "step": 8205
-    },
-    {
-      "epoch": 1.3215105277990258,
-      "grad_norm": 0.001105654053390026,
-      "learning_rate": 0.00019999914034104138,
-      "loss": 46.0,
-      "step": 8206
-    },
-    {
-      "epoch": 1.3216715648778132,
-      "grad_norm": 0.001803745049983263,
-      "learning_rate": 0.00019999914013125363,
-      "loss": 46.0,
-      "step": 8207
-    },
-    {
-      "epoch": 1.3218326019566005,
-      "grad_norm": 0.0011665953788906336,
-      "learning_rate": 0.00019999913992144029,
-      "loss": 46.0,
-      "step": 8208
-    },
-    {
-      "epoch": 1.321993639035388,
-      "grad_norm": 0.0028275202494114637,
-      "learning_rate": 0.00019999913971160136,
-      "loss": 46.0,
-      "step": 8209
-    },
-    {
-      "epoch": 1.3221546761141754,
-      "grad_norm": 0.0011850474402308464,
-      "learning_rate": 0.00019999913950173684,
-      "loss": 46.0,
-      "step": 8210
-    },
-    {
-      "epoch": 1.3223157131929626,
-      "grad_norm": 0.0030700850766152143,
-      "learning_rate": 0.00019999913929184671,
-      "loss": 46.0,
-      "step": 8211
-    },
-    {
-      "epoch": 1.32247675027175,
-      "grad_norm": 0.00034545682137832046,
-      "learning_rate": 0.000199999139081931,
-      "loss": 46.0,
-      "step": 8212
-    },
-    {
-      "epoch": 1.3226377873505375,
-      "grad_norm": 0.0015467837220057845,
-      "learning_rate": 0.0001999991388719897,
-      "loss": 46.0,
-      "step": 8213
-    },
-    {
-      "epoch": 1.322798824429325,
-      "grad_norm": 0.001453143428079784,
-      "learning_rate": 0.00019999913866202275,
-      "loss": 46.0,
-      "step": 8214
-    },
-    {
-      "epoch": 1.3229598615081122,
-      "grad_norm": 0.004985189996659756,
-      "learning_rate": 0.00019999913845203027,
-      "loss": 46.0,
-      "step": 8215
-    },
-    {
-      "epoch": 1.3231208985868996,
-      "grad_norm": 0.001013303641229868,
-      "learning_rate": 0.00019999913824201218,
-      "loss": 46.0,
-      "step": 8216
-    },
-    {
-      "epoch": 1.323281935665687,
-      "grad_norm": 0.0010608448646962643,
-      "learning_rate": 0.00019999913803196847,
-      "loss": 46.0,
-      "step": 8217
-    },
-    {
-      "epoch": 1.3234429727444743,
-      "grad_norm": 0.0008629285730421543,
-      "learning_rate": 0.00019999913782189918,
-      "loss": 46.0,
-      "step": 8218
-    },
-    {
-      "epoch": 1.3236040098232618,
-      "grad_norm": 0.0007541061495430768,
-      "learning_rate": 0.0001999991376118043,
-      "loss": 46.0,
-      "step": 8219
-    },
-    {
-      "epoch": 1.3237650469020492,
-      "grad_norm": 0.0009547242661938071,
-      "learning_rate": 0.00019999913740168386,
-      "loss": 46.0,
-      "step": 8220
-    },
-    {
-      "epoch": 1.3239260839808367,
-      "grad_norm": 0.001845221733674407,
-      "learning_rate": 0.00019999913719153777,
-      "loss": 46.0,
-      "step": 8221
-    },
-    {
-      "epoch": 1.3240871210596241,
-      "grad_norm": 0.0059294309467077255,
-      "learning_rate": 0.00019999913698136613,
-      "loss": 46.0,
-      "step": 8222
-    },
-    {
-      "epoch": 1.3242481581384113,
-      "grad_norm": 0.004400711972266436,
-      "learning_rate": 0.00019999913677116885,
-      "loss": 46.0,
-      "step": 8223
-    },
-    {
-      "epoch": 1.3244091952171988,
-      "grad_norm": 0.01056316401809454,
-      "learning_rate": 0.00019999913656094603,
-      "loss": 46.0,
-      "step": 8224
-    },
-    {
-      "epoch": 1.324570232295986,
-      "grad_norm": 0.0015021535800769925,
-      "learning_rate": 0.00019999913635069755,
-      "loss": 46.0,
-      "step": 8225
-    },
-    {
-      "epoch": 1.3247312693747735,
-      "grad_norm": 0.0036306639667600393,
-      "learning_rate": 0.00019999913614042353,
-      "loss": 46.0,
-      "step": 8226
-    },
-    {
-      "epoch": 1.324892306453561,
-      "grad_norm": 0.001111185410991311,
-      "learning_rate": 0.00019999913593012387,
-      "loss": 46.0,
-      "step": 8227
-    },
-    {
-      "epoch": 1.3250533435323484,
-      "grad_norm": 0.0039011219050735235,
-      "learning_rate": 0.00019999913571979865,
-      "loss": 46.0,
-      "step": 8228
-    },
-    {
-      "epoch": 1.3252143806111358,
-      "grad_norm": 0.0017056837677955627,
-      "learning_rate": 0.00019999913550944784,
-      "loss": 46.0,
-      "step": 8229
-    },
-    {
-      "epoch": 1.325375417689923,
-      "grad_norm": 0.0009715649066492915,
-      "learning_rate": 0.00019999913529907142,
-      "loss": 46.0,
-      "step": 8230
-    },
-    {
-      "epoch": 1.3255364547687105,
-      "grad_norm": 0.005172718316316605,
-      "learning_rate": 0.0001999991350886694,
-      "loss": 46.0,
-      "step": 8231
-    },
-    {
-      "epoch": 1.325697491847498,
-      "grad_norm": 0.0013129191938787699,
-      "learning_rate": 0.0001999991348782418,
-      "loss": 46.0,
-      "step": 8232
-    },
-    {
-      "epoch": 1.3258585289262852,
-      "grad_norm": 0.0009031934314407408,
-      "learning_rate": 0.00019999913466778858,
-      "loss": 46.0,
-      "step": 8233
-    },
-    {
-      "epoch": 1.3260195660050726,
-      "grad_norm": 0.001726498012430966,
-      "learning_rate": 0.00019999913445730978,
-      "loss": 46.0,
-      "step": 8234
-    },
-    {
-      "epoch": 1.32618060308386,
-      "grad_norm": 0.00969169195741415,
-      "learning_rate": 0.0001999991342468054,
-      "loss": 46.0,
-      "step": 8235
-    },
-    {
-      "epoch": 1.3263416401626476,
-      "grad_norm": 0.002460531424731016,
-      "learning_rate": 0.00019999913403627543,
-      "loss": 46.0,
-      "step": 8236
-    },
-    {
-      "epoch": 1.3265026772414348,
-      "grad_norm": 0.0015143678756430745,
-      "learning_rate": 0.00019999913382571981,
-      "loss": 46.0,
-      "step": 8237
-    },
-    {
-      "epoch": 1.3266637143202222,
-      "grad_norm": 0.00221410789526999,
-      "learning_rate": 0.00019999913361513864,
-      "loss": 46.0,
-      "step": 8238
-    },
-    {
-      "epoch": 1.3268247513990097,
-      "grad_norm": 0.0012020812137052417,
-      "learning_rate": 0.00019999913340453188,
-      "loss": 46.0,
-      "step": 8239
-    },
-    {
-      "epoch": 1.326985788477797,
-      "grad_norm": 0.000679194345138967,
-      "learning_rate": 0.0001999991331938995,
-      "loss": 46.0,
-      "step": 8240
-    },
-    {
-      "epoch": 1.3271468255565844,
-      "grad_norm": 0.0010646703885868192,
-      "learning_rate": 0.00019999913298324155,
-      "loss": 46.0,
-      "step": 8241
-    },
-    {
-      "epoch": 1.3273078626353718,
-      "grad_norm": 0.00038416351890191436,
-      "learning_rate": 0.00019999913277255797,
-      "loss": 46.0,
-      "step": 8242
-    },
-    {
-      "epoch": 1.3274688997141593,
-      "grad_norm": 0.0044890837743878365,
-      "learning_rate": 0.00019999913256184883,
-      "loss": 46.0,
-      "step": 8243
-    },
-    {
-      "epoch": 1.3276299367929465,
-      "grad_norm": 0.0014554171357303858,
-      "learning_rate": 0.00019999913235111408,
-      "loss": 46.0,
-      "step": 8244
-    },
-    {
-      "epoch": 1.327790973871734,
-      "grad_norm": 0.012298003770411015,
-      "learning_rate": 0.00019999913214035375,
-      "loss": 46.0,
-      "step": 8245
-    },
-    {
-      "epoch": 1.3279520109505214,
-      "grad_norm": 0.0005114679806865752,
-      "learning_rate": 0.0001999991319295678,
-      "loss": 46.0,
-      "step": 8246
-    },
-    {
-      "epoch": 1.3281130480293086,
-      "grad_norm": 0.003929303027689457,
-      "learning_rate": 0.0001999991317187563,
-      "loss": 46.0,
-      "step": 8247
-    },
-    {
-      "epoch": 1.328274085108096,
-      "grad_norm": 0.00125032605137676,
-      "learning_rate": 0.00019999913150791919,
-      "loss": 46.0,
-      "step": 8248
-    },
-    {
-      "epoch": 1.3284351221868835,
-      "grad_norm": 0.00529604172334075,
-      "learning_rate": 0.00019999913129705645,
-      "loss": 46.0,
-      "step": 8249
-    },
-    {
-      "epoch": 1.328596159265671,
-      "grad_norm": 0.0008090316550806165,
-      "learning_rate": 0.00019999913108616815,
-      "loss": 46.0,
-      "step": 8250
-    },
-    {
-      "epoch": 1.3287571963444584,
-      "grad_norm": 0.0021193642169237137,
-      "learning_rate": 0.00019999913087525423,
-      "loss": 46.0,
-      "step": 8251
-    },
-    {
-      "epoch": 1.3289182334232457,
-      "grad_norm": 0.0016279650153592229,
-      "learning_rate": 0.00019999913066431473,
-      "loss": 46.0,
-      "step": 8252
-    },
-    {
-      "epoch": 1.3290792705020331,
-      "grad_norm": 0.0006489729275926948,
-      "learning_rate": 0.00019999913045334964,
-      "loss": 46.0,
-      "step": 8253
-    },
-    {
-      "epoch": 1.3292403075808203,
-      "grad_norm": 0.005904868245124817,
-      "learning_rate": 0.00019999913024235894,
-      "loss": 46.0,
-      "step": 8254
-    },
-    {
-      "epoch": 1.3294013446596078,
-      "grad_norm": 0.0029582222923636436,
-      "learning_rate": 0.00019999913003134265,
-      "loss": 46.0,
-      "step": 8255
-    },
-    {
-      "epoch": 1.3295623817383952,
-      "grad_norm": 0.0005188260111026466,
-      "learning_rate": 0.00019999912982030077,
-      "loss": 46.0,
-      "step": 8256
-    },
-    {
-      "epoch": 1.3297234188171827,
-      "grad_norm": 0.002158254152163863,
-      "learning_rate": 0.0001999991296092333,
-      "loss": 46.0,
-      "step": 8257
-    },
-    {
-      "epoch": 1.3298844558959702,
-      "grad_norm": 0.0006617026519961655,
-      "learning_rate": 0.00019999912939814023,
-      "loss": 46.0,
-      "step": 8258
-    },
-    {
-      "epoch": 1.3300454929747574,
-      "grad_norm": 0.0007311085937544703,
-      "learning_rate": 0.00019999912918702156,
-      "loss": 46.0,
-      "step": 8259
-    },
-    {
-      "epoch": 1.3302065300535448,
-      "grad_norm": 0.00880738440901041,
-      "learning_rate": 0.0001999991289758773,
-      "loss": 46.0,
-      "step": 8260
-    },
-    {
-      "epoch": 1.3303675671323323,
-      "grad_norm": 0.0009422603179700673,
-      "learning_rate": 0.00019999912876470747,
-      "loss": 46.0,
-      "step": 8261
-    },
-    {
-      "epoch": 1.3305286042111195,
-      "grad_norm": 0.0026198592968285084,
-      "learning_rate": 0.000199999128553512,
-      "loss": 46.0,
-      "step": 8262
-    },
-    {
-      "epoch": 1.330689641289907,
-      "grad_norm": 0.0023824789095669985,
-      "learning_rate": 0.00019999912834229094,
-      "loss": 46.0,
-      "step": 8263
-    },
-    {
-      "epoch": 1.3308506783686944,
-      "grad_norm": 0.0024635002482682467,
-      "learning_rate": 0.00019999912813104432,
-      "loss": 46.0,
-      "step": 8264
-    },
-    {
-      "epoch": 1.3310117154474819,
-      "grad_norm": 0.0005961424903944135,
-      "learning_rate": 0.0001999991279197721,
-      "loss": 46.0,
-      "step": 8265
-    },
-    {
-      "epoch": 1.331172752526269,
-      "grad_norm": 0.0017014414770528674,
-      "learning_rate": 0.00019999912770847427,
-      "loss": 46.0,
-      "step": 8266
-    },
-    {
-      "epoch": 1.3313337896050566,
-      "grad_norm": 0.003466135123744607,
-      "learning_rate": 0.00019999912749715085,
-      "loss": 46.0,
-      "step": 8267
-    },
-    {
-      "epoch": 1.331494826683844,
-      "grad_norm": 0.0069140056148171425,
-      "learning_rate": 0.00019999912728580182,
-      "loss": 46.0,
-      "step": 8268
-    },
-    {
-      "epoch": 1.3316558637626312,
-      "grad_norm": 0.0010820548050105572,
-      "learning_rate": 0.0001999991270744272,
-      "loss": 46.0,
-      "step": 8269
-    },
-    {
-      "epoch": 1.3318169008414187,
-      "grad_norm": 0.0007110327132977545,
-      "learning_rate": 0.00019999912686302702,
-      "loss": 46.0,
-      "step": 8270
-    },
-    {
-      "epoch": 1.3319779379202061,
-      "grad_norm": 0.002185997786000371,
-      "learning_rate": 0.0001999991266516012,
-      "loss": 46.0,
-      "step": 8271
-    },
-    {
-      "epoch": 1.3321389749989936,
-      "grad_norm": 0.00239480915479362,
-      "learning_rate": 0.00019999912644014982,
-      "loss": 46.0,
-      "step": 8272
-    },
-    {
-      "epoch": 1.332300012077781,
-      "grad_norm": 0.001562080578878522,
-      "learning_rate": 0.0001999991262286728,
-      "loss": 46.0,
-      "step": 8273
-    },
-    {
-      "epoch": 1.3324610491565683,
-      "grad_norm": 0.001237689284607768,
-      "learning_rate": 0.00019999912601717024,
-      "loss": 46.0,
-      "step": 8274
-    },
-    {
-      "epoch": 1.3326220862353557,
-      "grad_norm": 0.0019330834038555622,
-      "learning_rate": 0.00019999912580564208,
-      "loss": 46.0,
-      "step": 8275
-    },
-    {
-      "epoch": 1.332783123314143,
-      "grad_norm": 0.004047016613185406,
-      "learning_rate": 0.0001999991255940883,
-      "loss": 46.0,
-      "step": 8276
-    },
-    {
-      "epoch": 1.3329441603929304,
-      "grad_norm": 0.0012751294998452067,
-      "learning_rate": 0.00019999912538250892,
-      "loss": 46.0,
-      "step": 8277
-    },
-    {
-      "epoch": 1.3331051974717179,
-      "grad_norm": 0.0012406246969476342,
-      "learning_rate": 0.00019999912517090396,
-      "loss": 46.0,
-      "step": 8278
-    },
-    {
-      "epoch": 1.3332662345505053,
-      "grad_norm": 0.0010301086585968733,
-      "learning_rate": 0.0001999991249592734,
-      "loss": 46.0,
-      "step": 8279
-    },
-    {
-      "epoch": 1.3334272716292928,
-      "grad_norm": 0.0007084830431267619,
-      "learning_rate": 0.00019999912474761723,
-      "loss": 46.0,
-      "step": 8280
-    },
-    {
-      "epoch": 1.33358830870808,
-      "grad_norm": 0.0009050874505192041,
-      "learning_rate": 0.0001999991245359355,
-      "loss": 46.0,
-      "step": 8281
-    },
-    {
-      "epoch": 1.3337493457868674,
-      "grad_norm": 0.005876012146472931,
-      "learning_rate": 0.00019999912432422815,
-      "loss": 46.0,
-      "step": 8282
-    },
-    {
-      "epoch": 1.3339103828656549,
-      "grad_norm": 0.0011896018404513597,
-      "learning_rate": 0.0001999991241124952,
-      "loss": 46.0,
-      "step": 8283
-    },
-    {
-      "epoch": 1.3340714199444421,
-      "grad_norm": 0.0024625216610729694,
-      "learning_rate": 0.0001999991239007367,
-      "loss": 46.0,
-      "step": 8284
-    },
-    {
-      "epoch": 1.3342324570232296,
-      "grad_norm": 0.0014154164819046855,
-      "learning_rate": 0.00019999912368895255,
-      "loss": 46.0,
-      "step": 8285
-    },
-    {
-      "epoch": 1.334393494102017,
-      "grad_norm": 0.002725822851061821,
-      "learning_rate": 0.00019999912347714284,
-      "loss": 46.0,
-      "step": 8286
-    },
-    {
-      "epoch": 1.3345545311808045,
-      "grad_norm": 0.0017862452659755945,
-      "learning_rate": 0.00019999912326530751,
-      "loss": 46.0,
-      "step": 8287
-    },
-    {
-      "epoch": 1.3347155682595917,
-      "grad_norm": 0.0027259038761258125,
-      "learning_rate": 0.00019999912305344663,
-      "loss": 46.0,
-      "step": 8288
-    },
-    {
-      "epoch": 1.3348766053383792,
-      "grad_norm": 0.0010943528031930327,
-      "learning_rate": 0.00019999912284156013,
-      "loss": 46.0,
-      "step": 8289
-    },
-    {
-      "epoch": 1.3350376424171666,
-      "grad_norm": 0.001049652462825179,
-      "learning_rate": 0.00019999912262964802,
-      "loss": 46.0,
-      "step": 8290
-    },
-    {
-      "epoch": 1.3351986794959538,
-      "grad_norm": 0.0014066135045140982,
-      "learning_rate": 0.00019999912241771032,
-      "loss": 46.0,
-      "step": 8291
-    },
-    {
-      "epoch": 1.3353597165747413,
-      "grad_norm": 0.002039481420069933,
-      "learning_rate": 0.00019999912220574703,
-      "loss": 46.0,
-      "step": 8292
-    },
-    {
-      "epoch": 1.3355207536535287,
-      "grad_norm": 0.002038148930296302,
-      "learning_rate": 0.00019999912199375816,
-      "loss": 46.0,
-      "step": 8293
-    },
-    {
-      "epoch": 1.3356817907323162,
-      "grad_norm": 0.00338871986605227,
-      "learning_rate": 0.00019999912178174367,
-      "loss": 46.0,
-      "step": 8294
-    },
-    {
-      "epoch": 1.3358428278111036,
-      "grad_norm": 0.001932501676492393,
-      "learning_rate": 0.0001999991215697036,
-      "loss": 46.0,
-      "step": 8295
-    },
-    {
-      "epoch": 1.3360038648898909,
-      "grad_norm": 0.0007987224380485713,
-      "learning_rate": 0.00019999912135763794,
-      "loss": 46.0,
-      "step": 8296
-    },
-    {
-      "epoch": 1.3361649019686783,
-      "grad_norm": 0.0014676302671432495,
-      "learning_rate": 0.00019999912114554666,
-      "loss": 46.0,
-      "step": 8297
-    },
-    {
-      "epoch": 1.3363259390474655,
-      "grad_norm": 0.0007820677710697055,
-      "learning_rate": 0.0001999991209334298,
-      "loss": 46.0,
-      "step": 8298
-    },
-    {
-      "epoch": 1.336486976126253,
-      "grad_norm": 0.0009669442079029977,
-      "learning_rate": 0.00019999912072128737,
-      "loss": 46.0,
-      "step": 8299
-    },
-    {
-      "epoch": 1.3366480132050405,
-      "grad_norm": 0.0017099491087719798,
-      "learning_rate": 0.00019999912050911933,
-      "loss": 46.0,
-      "step": 8300
-    },
-    {
-      "epoch": 1.336809050283828,
-      "grad_norm": 0.003365869401022792,
-      "learning_rate": 0.00019999912029692568,
-      "loss": 46.0,
-      "step": 8301
-    },
-    {
-      "epoch": 1.3369700873626154,
-      "grad_norm": 0.0011575553799048066,
-      "learning_rate": 0.00019999912008470644,
-      "loss": 46.0,
-      "step": 8302
-    },
-    {
-      "epoch": 1.3371311244414026,
-      "grad_norm": 0.0024347794242203236,
-      "learning_rate": 0.00019999911987246161,
-      "loss": 46.0,
-      "step": 8303
-    },
-    {
-      "epoch": 1.33729216152019,
-      "grad_norm": 0.002584053436294198,
-      "learning_rate": 0.0001999991196601912,
-      "loss": 46.0,
-      "step": 8304
-    },
-    {
-      "epoch": 1.3374531985989775,
-      "grad_norm": 0.001957808155566454,
-      "learning_rate": 0.00019999911944789517,
-      "loss": 46.0,
-      "step": 8305
-    },
-    {
-      "epoch": 1.3376142356777647,
-      "grad_norm": 0.0006251740269362926,
-      "learning_rate": 0.00019999911923557358,
-      "loss": 46.0,
-      "step": 8306
-    },
-    {
-      "epoch": 1.3377752727565522,
-      "grad_norm": 0.0016109747812151909,
-      "learning_rate": 0.00019999911902322635,
-      "loss": 46.0,
-      "step": 8307
-    },
-    {
-      "epoch": 1.3379363098353396,
-      "grad_norm": 0.001177363214083016,
-      "learning_rate": 0.00019999911881085356,
-      "loss": 46.0,
-      "step": 8308
-    },
-    {
-      "epoch": 1.338097346914127,
-      "grad_norm": 0.001695385668426752,
-      "learning_rate": 0.00019999911859845516,
-      "loss": 46.0,
-      "step": 8309
-    },
-    {
-      "epoch": 1.3382583839929143,
-      "grad_norm": 0.0008704941719770432,
-      "learning_rate": 0.00019999911838603117,
-      "loss": 46.0,
-      "step": 8310
-    },
-    {
-      "epoch": 1.3384194210717018,
-      "grad_norm": 0.0013970899162814021,
-      "learning_rate": 0.0001999991181735816,
-      "loss": 46.0,
-      "step": 8311
-    },
-    {
-      "epoch": 1.3385804581504892,
-      "grad_norm": 0.005044432822614908,
-      "learning_rate": 0.0001999991179611064,
-      "loss": 46.0,
-      "step": 8312
-    },
-    {
-      "epoch": 1.3387414952292764,
-      "grad_norm": 0.001802628394216299,
-      "learning_rate": 0.00019999911774860562,
-      "loss": 46.0,
-      "step": 8313
-    },
-    {
-      "epoch": 1.3389025323080639,
-      "grad_norm": 0.005088148172944784,
-      "learning_rate": 0.00019999911753607928,
-      "loss": 46.0,
-      "step": 8314
-    },
-    {
-      "epoch": 1.3390635693868513,
-      "grad_norm": 0.0018335465574637055,
-      "learning_rate": 0.0001999991173235273,
-      "loss": 46.0,
-      "step": 8315
-    },
-    {
-      "epoch": 1.3392246064656388,
-      "grad_norm": 0.0007183181587606668,
-      "learning_rate": 0.00019999911711094973,
-      "loss": 46.0,
-      "step": 8316
-    },
-    {
-      "epoch": 1.3393856435444262,
-      "grad_norm": 0.0007244597654789686,
-      "learning_rate": 0.00019999911689834657,
-      "loss": 46.0,
-      "step": 8317
-    },
-    {
-      "epoch": 1.3395466806232135,
-      "grad_norm": 0.009339780546724796,
-      "learning_rate": 0.00019999911668571786,
-      "loss": 46.0,
-      "step": 8318
-    },
-    {
-      "epoch": 1.339707717702001,
-      "grad_norm": 0.0034740956034511328,
-      "learning_rate": 0.00019999911647306347,
-      "loss": 46.0,
-      "step": 8319
-    },
-    {
-      "epoch": 1.3398687547807882,
-      "grad_norm": 0.0004981555393896997,
-      "learning_rate": 0.00019999911626038356,
-      "loss": 46.0,
-      "step": 8320
-    },
-    {
-      "epoch": 1.3400297918595756,
-      "grad_norm": 0.001770513248629868,
-      "learning_rate": 0.000199999116047678,
-      "loss": 46.0,
-      "step": 8321
-    },
-    {
-      "epoch": 1.340190828938363,
-      "grad_norm": 0.0013742547016590834,
-      "learning_rate": 0.00019999911583494688,
-      "loss": 46.0,
-      "step": 8322
-    },
-    {
-      "epoch": 1.3403518660171505,
-      "grad_norm": 0.0006776098161935806,
-      "learning_rate": 0.00019999911562219017,
-      "loss": 46.0,
-      "step": 8323
-    },
-    {
-      "epoch": 1.340512903095938,
-      "grad_norm": 0.0008843560353852808,
-      "learning_rate": 0.00019999911540940785,
-      "loss": 46.0,
-      "step": 8324
-    },
-    {
-      "epoch": 1.3406739401747252,
-      "grad_norm": 0.003426025388762355,
-      "learning_rate": 0.00019999911519659995,
-      "loss": 46.0,
-      "step": 8325
-    },
-    {
-      "epoch": 1.3408349772535126,
-      "grad_norm": 0.004045134410262108,
-      "learning_rate": 0.00019999911498376643,
-      "loss": 46.0,
-      "step": 8326
-    },
-    {
-      "epoch": 1.3409960143323,
-      "grad_norm": 0.005195715464651585,
-      "learning_rate": 0.00019999911477090734,
-      "loss": 46.0,
-      "step": 8327
-    },
-    {
-      "epoch": 1.3411570514110873,
-      "grad_norm": 0.00196321913972497,
-      "learning_rate": 0.00019999911455802262,
-      "loss": 46.0,
-      "step": 8328
-    },
-    {
-      "epoch": 1.3413180884898748,
-      "grad_norm": 0.001250079134479165,
-      "learning_rate": 0.00019999911434511237,
-      "loss": 46.0,
-      "step": 8329
-    },
-    {
-      "epoch": 1.3414791255686622,
-      "grad_norm": 0.0014327957760542631,
-      "learning_rate": 0.00019999911413217647,
-      "loss": 46.0,
-      "step": 8330
-    },
-    {
-      "epoch": 1.3416401626474497,
-      "grad_norm": 0.0031670050229877234,
-      "learning_rate": 0.000199999113919215,
-      "loss": 46.0,
-      "step": 8331
-    },
-    {
-      "epoch": 1.341801199726237,
-      "grad_norm": 0.0003307102888356894,
-      "learning_rate": 0.0001999991137062279,
-      "loss": 46.0,
-      "step": 8332
-    },
-    {
-      "epoch": 1.3419622368050244,
-      "grad_norm": 0.0006395092932507396,
-      "learning_rate": 0.00019999911349321523,
-      "loss": 46.0,
-      "step": 8333
-    },
-    {
-      "epoch": 1.3421232738838118,
-      "grad_norm": 0.0009109979146160185,
-      "learning_rate": 0.00019999911328017698,
-      "loss": 46.0,
-      "step": 8334
-    },
-    {
-      "epoch": 1.342284310962599,
-      "grad_norm": 0.0006914462428539991,
-      "learning_rate": 0.00019999911306711312,
-      "loss": 46.0,
-      "step": 8335
-    },
-    {
-      "epoch": 1.3424453480413865,
-      "grad_norm": 0.0029380943160504103,
-      "learning_rate": 0.00019999911285402367,
-      "loss": 46.0,
-      "step": 8336
-    },
-    {
-      "epoch": 1.342606385120174,
-      "grad_norm": 0.003370831022039056,
-      "learning_rate": 0.00019999911264090864,
-      "loss": 46.0,
-      "step": 8337
-    },
-    {
-      "epoch": 1.3427674221989614,
-      "grad_norm": 0.004577951040118933,
-      "learning_rate": 0.000199999112427768,
-      "loss": 46.0,
-      "step": 8338
-    },
-    {
-      "epoch": 1.3429284592777486,
-      "grad_norm": 0.0004588786978274584,
-      "learning_rate": 0.00019999911221460175,
-      "loss": 46.0,
-      "step": 8339
-    },
-    {
-      "epoch": 1.343089496356536,
-      "grad_norm": 0.0007934690802358091,
-      "learning_rate": 0.00019999911200140993,
-      "loss": 46.0,
-      "step": 8340
-    },
-    {
-      "epoch": 1.3432505334353235,
-      "grad_norm": 0.010320227593183517,
-      "learning_rate": 0.0001999991117881925,
-      "loss": 46.0,
-      "step": 8341
-    },
-    {
-      "epoch": 1.3434115705141108,
-      "grad_norm": 0.005865951534360647,
-      "learning_rate": 0.00019999911157494947,
-      "loss": 46.0,
-      "step": 8342
-    },
-    {
-      "epoch": 1.3435726075928982,
-      "grad_norm": 0.0008823686512187123,
-      "learning_rate": 0.00019999911136168086,
-      "loss": 46.0,
-      "step": 8343
-    },
-    {
-      "epoch": 1.3437336446716857,
-      "grad_norm": 0.0012530016247183084,
-      "learning_rate": 0.00019999911114838666,
-      "loss": 46.0,
-      "step": 8344
-    },
-    {
-      "epoch": 1.343894681750473,
-      "grad_norm": 0.0032118402887135744,
-      "learning_rate": 0.00019999911093506684,
-      "loss": 46.0,
-      "step": 8345
-    },
-    {
-      "epoch": 1.3440557188292606,
-      "grad_norm": 0.0004974527982994914,
-      "learning_rate": 0.00019999911072172144,
-      "loss": 46.0,
-      "step": 8346
-    },
-    {
-      "epoch": 1.3442167559080478,
-      "grad_norm": 0.0007516128825955093,
-      "learning_rate": 0.00019999911050835046,
-      "loss": 46.0,
-      "step": 8347
-    },
-    {
-      "epoch": 1.3443777929868352,
-      "grad_norm": 0.002259405329823494,
-      "learning_rate": 0.00019999911029495386,
-      "loss": 46.0,
-      "step": 8348
-    },
-    {
-      "epoch": 1.3445388300656227,
-      "grad_norm": 0.0014529795153066516,
-      "learning_rate": 0.00019999911008153167,
-      "loss": 46.0,
-      "step": 8349
-    },
-    {
-      "epoch": 1.34469986714441,
-      "grad_norm": 0.0017328370595350862,
-      "learning_rate": 0.0001999991098680839,
-      "loss": 46.0,
-      "step": 8350
-    },
-    {
-      "epoch": 1.3448609042231974,
-      "grad_norm": 0.0014722829218953848,
-      "learning_rate": 0.00019999910965461053,
-      "loss": 46.0,
-      "step": 8351
-    },
-    {
-      "epoch": 1.3450219413019848,
-      "grad_norm": 0.0019518237095326185,
-      "learning_rate": 0.00019999910944111155,
-      "loss": 46.0,
-      "step": 8352
-    },
-    {
-      "epoch": 1.3451829783807723,
-      "grad_norm": 0.0008531832718290389,
-      "learning_rate": 0.000199999109227587,
-      "loss": 46.0,
-      "step": 8353
-    },
-    {
-      "epoch": 1.3453440154595595,
-      "grad_norm": 0.0020764674991369247,
-      "learning_rate": 0.00019999910901403683,
-      "loss": 46.0,
-      "step": 8354
-    },
-    {
-      "epoch": 1.345505052538347,
-      "grad_norm": 0.0012880354188382626,
-      "learning_rate": 0.0001999991088004611,
-      "loss": 46.0,
-      "step": 8355
-    },
-    {
-      "epoch": 1.3456660896171344,
-      "grad_norm": 0.0020424723625183105,
-      "learning_rate": 0.00019999910858685974,
-      "loss": 46.0,
-      "step": 8356
-    },
-    {
-      "epoch": 1.3458271266959216,
-      "grad_norm": 0.004828402306884527,
-      "learning_rate": 0.00019999910837323283,
-      "loss": 46.0,
-      "step": 8357
-    },
-    {
-      "epoch": 1.345988163774709,
-      "grad_norm": 0.0019889031536877155,
-      "learning_rate": 0.00019999910815958027,
-      "loss": 46.0,
-      "step": 8358
-    },
-    {
-      "epoch": 1.3461492008534965,
-      "grad_norm": 0.0031129762064665556,
-      "learning_rate": 0.00019999910794590216,
-      "loss": 46.0,
-      "step": 8359
-    },
-    {
-      "epoch": 1.346310237932284,
-      "grad_norm": 0.00445446465164423,
-      "learning_rate": 0.00019999910773219843,
-      "loss": 46.0,
-      "step": 8360
-    },
-    {
-      "epoch": 1.3464712750110712,
-      "grad_norm": 0.001002848264761269,
-      "learning_rate": 0.0001999991075184691,
-      "loss": 46.0,
-      "step": 8361
-    },
-    {
-      "epoch": 1.3466323120898587,
-      "grad_norm": 0.003077493514865637,
-      "learning_rate": 0.00019999910730471418,
-      "loss": 46.0,
-      "step": 8362
-    },
-    {
-      "epoch": 1.3467933491686461,
-      "grad_norm": 0.000879055995028466,
-      "learning_rate": 0.0001999991070909337,
-      "loss": 46.0,
-      "step": 8363
-    },
-    {
-      "epoch": 1.3469543862474334,
-      "grad_norm": 0.00426897220313549,
-      "learning_rate": 0.00019999910687712758,
-      "loss": 46.0,
-      "step": 8364
-    },
-    {
-      "epoch": 1.3471154233262208,
-      "grad_norm": 0.002269066171720624,
-      "learning_rate": 0.0001999991066632959,
-      "loss": 46.0,
-      "step": 8365
-    },
-    {
-      "epoch": 1.3472764604050083,
-      "grad_norm": 0.0005826735286973417,
-      "learning_rate": 0.0001999991064494386,
-      "loss": 46.0,
-      "step": 8366
-    },
-    {
-      "epoch": 1.3474374974837957,
-      "grad_norm": 0.0007702274597249925,
-      "learning_rate": 0.00019999910623555572,
-      "loss": 46.0,
-      "step": 8367
-    },
-    {
-      "epoch": 1.3475985345625832,
-      "grad_norm": 0.009286720305681229,
-      "learning_rate": 0.00019999910602164724,
-      "loss": 46.0,
-      "step": 8368
-    },
-    {
-      "epoch": 1.3477595716413704,
-      "grad_norm": 0.005011996254324913,
-      "learning_rate": 0.00019999910580771317,
-      "loss": 46.0,
-      "step": 8369
-    },
-    {
-      "epoch": 1.3479206087201578,
-      "grad_norm": 0.0004813605628442019,
-      "learning_rate": 0.0001999991055937535,
-      "loss": 46.0,
-      "step": 8370
-    },
-    {
-      "epoch": 1.348081645798945,
-      "grad_norm": 0.001700040651485324,
-      "learning_rate": 0.00019999910537976822,
-      "loss": 46.0,
-      "step": 8371
-    },
-    {
-      "epoch": 1.3482426828777325,
-      "grad_norm": 0.001677106600254774,
-      "learning_rate": 0.00019999910516575736,
-      "loss": 46.0,
-      "step": 8372
-    },
-    {
-      "epoch": 1.34840371995652,
-      "grad_norm": 0.000986463506706059,
-      "learning_rate": 0.00019999910495172092,
-      "loss": 46.0,
-      "step": 8373
-    },
-    {
-      "epoch": 1.3485647570353074,
-      "grad_norm": 0.0046012806706130505,
-      "learning_rate": 0.00019999910473765886,
-      "loss": 46.0,
-      "step": 8374
-    },
-    {
-      "epoch": 1.3487257941140949,
-      "grad_norm": 0.0017061902908608317,
-      "learning_rate": 0.00019999910452357122,
-      "loss": 46.0,
-      "step": 8375
-    },
-    {
-      "epoch": 1.348886831192882,
-      "grad_norm": 0.0032872790470719337,
-      "learning_rate": 0.00019999910430945798,
-      "loss": 46.0,
-      "step": 8376
-    },
-    {
-      "epoch": 1.3490478682716696,
-      "grad_norm": 0.0020111144986003637,
-      "learning_rate": 0.00019999910409531914,
-      "loss": 46.0,
-      "step": 8377
-    },
-    {
-      "epoch": 1.349208905350457,
-      "grad_norm": 0.0009973772102966905,
-      "learning_rate": 0.00019999910388115473,
-      "loss": 46.0,
-      "step": 8378
-    },
-    {
-      "epoch": 1.3493699424292442,
-      "grad_norm": 0.0016136944759637117,
-      "learning_rate": 0.0001999991036669647,
-      "loss": 46.0,
-      "step": 8379
-    },
-    {
-      "epoch": 1.3495309795080317,
-      "grad_norm": 0.00300107360817492,
-      "learning_rate": 0.00019999910345274907,
-      "loss": 46.0,
-      "step": 8380
-    },
-    {
-      "epoch": 1.3496920165868191,
-      "grad_norm": 0.0011417981004342437,
-      "learning_rate": 0.00019999910323850788,
-      "loss": 46.0,
-      "step": 8381
-    },
-    {
-      "epoch": 1.3498530536656066,
-      "grad_norm": 0.0031487373635172844,
-      "learning_rate": 0.00019999910302424104,
-      "loss": 46.0,
-      "step": 8382
-    },
-    {
-      "epoch": 1.3500140907443938,
-      "grad_norm": 0.001366683398373425,
-      "learning_rate": 0.00019999910280994867,
-      "loss": 46.0,
-      "step": 8383
-    },
-    {
-      "epoch": 1.3501751278231813,
-      "grad_norm": 0.0005385983968153596,
-      "learning_rate": 0.00019999910259563066,
-      "loss": 46.0,
-      "step": 8384
-    },
-    {
-      "epoch": 1.3503361649019687,
-      "grad_norm": 0.002977861789986491,
-      "learning_rate": 0.0001999991023812871,
-      "loss": 46.0,
-      "step": 8385
-    },
-    {
-      "epoch": 1.350497201980756,
-      "grad_norm": 0.004040272906422615,
-      "learning_rate": 0.0001999991021669179,
-      "loss": 46.0,
-      "step": 8386
-    },
-    {
-      "epoch": 1.3506582390595434,
-      "grad_norm": 0.0012413655640557408,
-      "learning_rate": 0.0001999991019525231,
-      "loss": 46.0,
-      "step": 8387
-    },
-    {
-      "epoch": 1.3508192761383309,
-      "grad_norm": 0.0014118441613391042,
-      "learning_rate": 0.00019999910173810275,
-      "loss": 46.0,
-      "step": 8388
-    },
-    {
-      "epoch": 1.3509803132171183,
-      "grad_norm": 0.0009189636330120265,
-      "learning_rate": 0.00019999910152365677,
-      "loss": 46.0,
-      "step": 8389
-    },
-    {
-      "epoch": 1.3511413502959058,
-      "grad_norm": 0.0013718331465497613,
-      "learning_rate": 0.0001999991013091852,
-      "loss": 46.0,
-      "step": 8390
-    },
-    {
-      "epoch": 1.351302387374693,
-      "grad_norm": 0.004593018442392349,
-      "learning_rate": 0.00019999910109468803,
-      "loss": 46.0,
-      "step": 8391
-    },
-    {
-      "epoch": 1.3514634244534804,
-      "grad_norm": 0.008490891195833683,
-      "learning_rate": 0.0001999991008801653,
-      "loss": 46.0,
-      "step": 8392
-    },
-    {
-      "epoch": 1.3516244615322677,
-      "grad_norm": 0.0019652703776955605,
-      "learning_rate": 0.00019999910066561695,
-      "loss": 46.0,
-      "step": 8393
-    },
-    {
-      "epoch": 1.3517854986110551,
-      "grad_norm": 0.0015580151230096817,
-      "learning_rate": 0.000199999100451043,
-      "loss": 46.0,
-      "step": 8394
-    },
-    {
-      "epoch": 1.3519465356898426,
-      "grad_norm": 0.0029250483494251966,
-      "learning_rate": 0.0001999991002364435,
-      "loss": 46.0,
-      "step": 8395
-    },
-    {
-      "epoch": 1.35210757276863,
-      "grad_norm": 0.002641297411173582,
-      "learning_rate": 0.00019999910002181835,
-      "loss": 46.0,
-      "step": 8396
-    },
-    {
-      "epoch": 1.3522686098474175,
-      "grad_norm": 0.0012804210418835282,
-      "learning_rate": 0.00019999909980716762,
-      "loss": 46.0,
-      "step": 8397
-    },
-    {
-      "epoch": 1.3524296469262047,
-      "grad_norm": 0.012449518777430058,
-      "learning_rate": 0.0001999990995924913,
-      "loss": 46.0,
-      "step": 8398
-    },
-    {
-      "epoch": 1.3525906840049922,
-      "grad_norm": 0.00150947249494493,
-      "learning_rate": 0.00019999909937778938,
-      "loss": 46.0,
-      "step": 8399
-    },
-    {
-      "epoch": 1.3527517210837796,
-      "grad_norm": 0.001049962593242526,
-      "learning_rate": 0.00019999909916306187,
-      "loss": 46.0,
-      "step": 8400
-    },
-    {
-      "epoch": 1.3529127581625668,
-      "grad_norm": 0.0005767635302618146,
-      "learning_rate": 0.0001999990989483088,
-      "loss": 46.0,
-      "step": 8401
-    },
-    {
-      "epoch": 1.3530737952413543,
-      "grad_norm": 0.0065422276966273785,
-      "learning_rate": 0.00019999909873353005,
-      "loss": 46.0,
-      "step": 8402
-    },
-    {
-      "epoch": 1.3532348323201417,
-      "grad_norm": 0.0030258605256676674,
-      "learning_rate": 0.0001999990985187258,
-      "loss": 46.0,
-      "step": 8403
-    },
-    {
-      "epoch": 1.3533958693989292,
-      "grad_norm": 0.01033786404877901,
-      "learning_rate": 0.0001999990983038959,
-      "loss": 46.0,
-      "step": 8404
-    },
-    {
-      "epoch": 1.3535569064777164,
-      "grad_norm": 0.0029609203338623047,
-      "learning_rate": 0.0001999990980890404,
-      "loss": 46.0,
-      "step": 8405
-    },
-    {
-      "epoch": 1.3537179435565039,
-      "grad_norm": 0.007678866386413574,
-      "learning_rate": 0.00019999909787415935,
-      "loss": 46.0,
-      "step": 8406
-    },
-    {
-      "epoch": 1.3538789806352913,
-      "grad_norm": 0.004396035801619291,
-      "learning_rate": 0.00019999909765925267,
-      "loss": 46.0,
-      "step": 8407
-    },
-    {
-      "epoch": 1.3540400177140786,
-      "grad_norm": 0.0047828382812440395,
-      "learning_rate": 0.0001999990974443204,
-      "loss": 46.0,
-      "step": 8408
-    },
-    {
-      "epoch": 1.354201054792866,
-      "grad_norm": 0.001196342404000461,
-      "learning_rate": 0.00019999909722936255,
-      "loss": 46.0,
-      "step": 8409
-    },
-    {
-      "epoch": 1.3543620918716535,
-      "grad_norm": 0.0012605487136170268,
-      "learning_rate": 0.00019999909701437908,
-      "loss": 46.0,
-      "step": 8410
-    },
-    {
-      "epoch": 1.354523128950441,
-      "grad_norm": 0.0005816929624415934,
-      "learning_rate": 0.00019999909679937003,
-      "loss": 46.0,
-      "step": 8411
-    },
-    {
-      "epoch": 1.3546841660292284,
-      "grad_norm": 0.0007248731562867761,
-      "learning_rate": 0.00019999909658433539,
-      "loss": 46.0,
-      "step": 8412
-    },
-    {
-      "epoch": 1.3548452031080156,
-      "grad_norm": 0.000497568107675761,
-      "learning_rate": 0.00019999909636927513,
-      "loss": 46.0,
-      "step": 8413
-    },
-    {
-      "epoch": 1.355006240186803,
-      "grad_norm": 0.0009381204145029187,
-      "learning_rate": 0.00019999909615418931,
-      "loss": 46.0,
-      "step": 8414
-    },
-    {
-      "epoch": 1.3551672772655903,
-      "grad_norm": 0.0036037079989910126,
-      "learning_rate": 0.00019999909593907788,
-      "loss": 46.0,
-      "step": 8415
-    },
-    {
-      "epoch": 1.3553283143443777,
-      "grad_norm": 0.0032943258993327618,
-      "learning_rate": 0.00019999909572394087,
-      "loss": 46.0,
-      "step": 8416
-    },
-    {
-      "epoch": 1.3554893514231652,
-      "grad_norm": 0.0037067849189043045,
-      "learning_rate": 0.00019999909550877823,
-      "loss": 46.0,
-      "step": 8417
-    },
-    {
-      "epoch": 1.3556503885019526,
-      "grad_norm": 0.007158919703215361,
-      "learning_rate": 0.00019999909529359004,
-      "loss": 46.0,
-      "step": 8418
-    },
-    {
-      "epoch": 1.35581142558074,
-      "grad_norm": 0.0005020997487008572,
-      "learning_rate": 0.00019999909507837624,
-      "loss": 46.0,
-      "step": 8419
-    },
-    {
-      "epoch": 1.3559724626595273,
-      "grad_norm": 0.001645746175199747,
-      "learning_rate": 0.00019999909486313682,
-      "loss": 46.0,
-      "step": 8420
-    },
-    {
-      "epoch": 1.3561334997383148,
-      "grad_norm": 0.0009111354011110961,
-      "learning_rate": 0.00019999909464787184,
-      "loss": 46.0,
-      "step": 8421
-    },
-    {
-      "epoch": 1.3562945368171022,
-      "grad_norm": 0.004183598328381777,
-      "learning_rate": 0.00019999909443258121,
-      "loss": 46.0,
-      "step": 8422
-    },
-    {
-      "epoch": 1.3564555738958894,
-      "grad_norm": 0.002702893689274788,
-      "learning_rate": 0.00019999909421726506,
-      "loss": 46.0,
-      "step": 8423
-    },
-    {
-      "epoch": 1.356616610974677,
-      "grad_norm": 0.0006687360582873225,
-      "learning_rate": 0.00019999909400192326,
-      "loss": 46.0,
-      "step": 8424
-    },
-    {
-      "epoch": 1.3567776480534643,
-      "grad_norm": 0.011225173249840736,
-      "learning_rate": 0.0001999990937865559,
-      "loss": 46.0,
-      "step": 8425
-    },
-    {
-      "epoch": 1.3569386851322518,
-      "grad_norm": 0.004288196563720703,
-      "learning_rate": 0.0001999990935711629,
-      "loss": 46.0,
-      "step": 8426
-    },
-    {
-      "epoch": 1.357099722211039,
-      "grad_norm": 0.002006173599511385,
-      "learning_rate": 0.00019999909335574435,
-      "loss": 46.0,
-      "step": 8427
-    },
-    {
-      "epoch": 1.3572607592898265,
-      "grad_norm": 0.0019895550794899464,
-      "learning_rate": 0.00019999909314030018,
-      "loss": 46.0,
-      "step": 8428
-    },
-    {
-      "epoch": 1.357421796368614,
-      "grad_norm": 0.0011818077182397246,
-      "learning_rate": 0.00019999909292483045,
-      "loss": 46.0,
-      "step": 8429
-    },
-    {
-      "epoch": 1.3575828334474012,
-      "grad_norm": 0.0022285941522568464,
-      "learning_rate": 0.0001999990927093351,
-      "loss": 46.0,
-      "step": 8430
-    },
-    {
-      "epoch": 1.3577438705261886,
-      "grad_norm": 0.0030746273696422577,
-      "learning_rate": 0.00019999909249381417,
-      "loss": 46.0,
-      "step": 8431
-    },
-    {
-      "epoch": 1.357904907604976,
-      "grad_norm": 0.0015490445075556636,
-      "learning_rate": 0.0001999990922782676,
-      "loss": 46.0,
-      "step": 8432
-    },
-    {
-      "epoch": 1.3580659446837635,
-      "grad_norm": 0.0040275962091982365,
-      "learning_rate": 0.0001999990920626955,
-      "loss": 46.0,
-      "step": 8433
-    },
-    {
-      "epoch": 1.3582269817625507,
-      "grad_norm": 0.0028057959862053394,
-      "learning_rate": 0.00019999909184709777,
-      "loss": 46.0,
-      "step": 8434
-    },
-    {
-      "epoch": 1.3583880188413382,
-      "grad_norm": 0.00052833795780316,
-      "learning_rate": 0.00019999909163147446,
-      "loss": 46.0,
-      "step": 8435
-    },
-    {
-      "epoch": 1.3585490559201256,
-      "grad_norm": 0.0020677833817899227,
-      "learning_rate": 0.00019999909141582553,
-      "loss": 46.0,
-      "step": 8436
-    },
-    {
-      "epoch": 1.3587100929989129,
-      "grad_norm": 0.008361995220184326,
-      "learning_rate": 0.00019999909120015102,
-      "loss": 46.0,
-      "step": 8437
-    },
-    {
-      "epoch": 1.3588711300777003,
-      "grad_norm": 0.0047791688703000546,
-      "learning_rate": 0.0001999990909844509,
-      "loss": 46.0,
-      "step": 8438
-    },
-    {
-      "epoch": 1.3590321671564878,
-      "grad_norm": 0.0005320021882653236,
-      "learning_rate": 0.0001999990907687252,
-      "loss": 46.0,
-      "step": 8439
-    },
-    {
-      "epoch": 1.3591932042352752,
-      "grad_norm": 0.005229340400546789,
-      "learning_rate": 0.0001999990905529739,
-      "loss": 46.0,
-      "step": 8440
-    },
-    {
-      "epoch": 1.3593542413140627,
-      "grad_norm": 0.0016253853682428598,
-      "learning_rate": 0.00019999909033719703,
-      "loss": 46.0,
-      "step": 8441
-    },
-    {
-      "epoch": 1.35951527839285,
-      "grad_norm": 0.0013752507511526346,
-      "learning_rate": 0.00019999909012139453,
-      "loss": 46.0,
-      "step": 8442
-    },
-    {
-      "epoch": 1.3596763154716374,
-      "grad_norm": 0.004130436107516289,
-      "learning_rate": 0.00019999908990556647,
-      "loss": 46.0,
-      "step": 8443
-    },
-    {
-      "epoch": 1.3598373525504248,
-      "grad_norm": 0.0005620232550427318,
-      "learning_rate": 0.0001999990896897128,
-      "loss": 46.0,
-      "step": 8444
-    },
-    {
-      "epoch": 1.359998389629212,
-      "grad_norm": 0.003313005669042468,
-      "learning_rate": 0.00019999908947383353,
-      "loss": 46.0,
-      "step": 8445
-    },
-    {
-      "epoch": 1.3601594267079995,
-      "grad_norm": 0.004215719643980265,
-      "learning_rate": 0.00019999908925792865,
-      "loss": 46.0,
-      "step": 8446
-    },
-    {
-      "epoch": 1.360320463786787,
-      "grad_norm": 0.005193840712308884,
-      "learning_rate": 0.0001999990890419982,
-      "loss": 46.0,
-      "step": 8447
-    },
-    {
-      "epoch": 1.3604815008655744,
-      "grad_norm": 0.009163262322545052,
-      "learning_rate": 0.00019999908882604214,
-      "loss": 46.0,
-      "step": 8448
-    },
-    {
-      "epoch": 1.3606425379443616,
-      "grad_norm": 0.006165005266666412,
-      "learning_rate": 0.0001999990886100605,
-      "loss": 46.0,
-      "step": 8449
-    },
-    {
-      "epoch": 1.360803575023149,
-      "grad_norm": 0.003316365648061037,
-      "learning_rate": 0.00019999908839405325,
-      "loss": 46.0,
-      "step": 8450
-    },
-    {
-      "epoch": 1.3609646121019365,
-      "grad_norm": 0.001776957418769598,
-      "learning_rate": 0.00019999908817802043,
-      "loss": 46.0,
-      "step": 8451
-    },
-    {
-      "epoch": 1.3611256491807238,
-      "grad_norm": 0.003165928181260824,
-      "learning_rate": 0.00019999908796196198,
-      "loss": 46.0,
-      "step": 8452
-    },
-    {
-      "epoch": 1.3612866862595112,
-      "grad_norm": 0.0013354551047086716,
-      "learning_rate": 0.00019999908774587797,
-      "loss": 46.0,
-      "step": 8453
-    },
-    {
-      "epoch": 1.3614477233382987,
-      "grad_norm": 0.00647937273606658,
-      "learning_rate": 0.00019999908752976834,
-      "loss": 46.0,
-      "step": 8454
-    },
-    {
-      "epoch": 1.3616087604170861,
-      "grad_norm": 0.00040617745253257453,
-      "learning_rate": 0.00019999908731363312,
-      "loss": 46.0,
-      "step": 8455
-    },
-    {
-      "epoch": 1.3617697974958733,
-      "grad_norm": 0.0016904743388295174,
-      "learning_rate": 0.00019999908709747232,
-      "loss": 46.0,
-      "step": 8456
-    },
-    {
-      "epoch": 1.3619308345746608,
-      "grad_norm": 0.0007965763215906918,
-      "learning_rate": 0.00019999908688128593,
-      "loss": 46.0,
-      "step": 8457
-    },
-    {
-      "epoch": 1.3620918716534482,
-      "grad_norm": 0.0035071561578661203,
-      "learning_rate": 0.00019999908666507393,
-      "loss": 46.0,
-      "step": 8458
-    },
-    {
-      "epoch": 1.3622529087322355,
-      "grad_norm": 0.0011240806197747588,
-      "learning_rate": 0.00019999908644883634,
-      "loss": 46.0,
-      "step": 8459
-    },
-    {
-      "epoch": 1.362413945811023,
-      "grad_norm": 0.0010247306199744344,
-      "learning_rate": 0.00019999908623257313,
-      "loss": 46.0,
-      "step": 8460
-    },
-    {
-      "epoch": 1.3625749828898104,
-      "grad_norm": 0.002692255424335599,
-      "learning_rate": 0.00019999908601628437,
-      "loss": 46.0,
-      "step": 8461
-    },
-    {
-      "epoch": 1.3627360199685978,
-      "grad_norm": 0.005207649897783995,
-      "learning_rate": 0.00019999908579997,
-      "loss": 46.0,
-      "step": 8462
-    },
-    {
-      "epoch": 1.3628970570473853,
-      "grad_norm": 0.0031039121095091105,
-      "learning_rate": 0.00019999908558363002,
-      "loss": 46.0,
-      "step": 8463
-    },
-    {
-      "epoch": 1.3630580941261725,
-      "grad_norm": 0.0008966705063357949,
-      "learning_rate": 0.00019999908536726447,
-      "loss": 46.0,
-      "step": 8464
-    },
-    {
-      "epoch": 1.36321913120496,
-      "grad_norm": 0.002622065832838416,
-      "learning_rate": 0.0001999990851508733,
-      "loss": 46.0,
-      "step": 8465
-    },
-    {
-      "epoch": 1.3633801682837472,
-      "grad_norm": 0.005339410621672869,
-      "learning_rate": 0.00019999908493445654,
-      "loss": 46.0,
-      "step": 8466
-    },
-    {
-      "epoch": 1.3635412053625346,
-      "grad_norm": 0.006051977165043354,
-      "learning_rate": 0.00019999908471801417,
-      "loss": 46.0,
-      "step": 8467
-    },
-    {
-      "epoch": 1.363702242441322,
-      "grad_norm": 0.001430383650586009,
-      "learning_rate": 0.00019999908450154624,
-      "loss": 46.0,
-      "step": 8468
-    },
-    {
-      "epoch": 1.3638632795201096,
-      "grad_norm": 0.0021215640008449554,
-      "learning_rate": 0.00019999908428505273,
-      "loss": 46.0,
-      "step": 8469
-    },
-    {
-      "epoch": 1.364024316598897,
-      "grad_norm": 0.002879285253584385,
-      "learning_rate": 0.00019999908406853357,
-      "loss": 46.0,
-      "step": 8470
-    },
-    {
-      "epoch": 1.3641853536776842,
-      "grad_norm": 0.0015226696850731969,
-      "learning_rate": 0.00019999908385198885,
-      "loss": 46.0,
-      "step": 8471
-    },
-    {
-      "epoch": 1.3643463907564717,
-      "grad_norm": 0.003810349851846695,
-      "learning_rate": 0.00019999908363541852,
-      "loss": 46.0,
-      "step": 8472
-    },
-    {
-      "epoch": 1.3645074278352591,
-      "grad_norm": 0.0025322316214442253,
-      "learning_rate": 0.0001999990834188226,
-      "loss": 46.0,
-      "step": 8473
-    },
-    {
-      "epoch": 1.3646684649140464,
-      "grad_norm": 0.0008757595205679536,
-      "learning_rate": 0.0001999990832022011,
-      "loss": 46.0,
-      "step": 8474
-    },
-    {
-      "epoch": 1.3648295019928338,
-      "grad_norm": 0.0014992282958701253,
-      "learning_rate": 0.000199999082985554,
-      "loss": 46.0,
-      "step": 8475
-    },
-    {
-      "epoch": 1.3649905390716213,
-      "grad_norm": 0.0012428911868482828,
-      "learning_rate": 0.00019999908276888126,
-      "loss": 46.0,
-      "step": 8476
-    },
-    {
-      "epoch": 1.3651515761504087,
-      "grad_norm": 0.001803399994969368,
-      "learning_rate": 0.000199999082552183,
-      "loss": 46.0,
-      "step": 8477
-    },
-    {
-      "epoch": 1.365312613229196,
-      "grad_norm": 0.00048699433682486415,
-      "learning_rate": 0.00019999908233545908,
-      "loss": 46.0,
-      "step": 8478
-    },
-    {
-      "epoch": 1.3654736503079834,
-      "grad_norm": 0.0044006346724927425,
-      "learning_rate": 0.0001999990821187096,
-      "loss": 46.0,
-      "step": 8479
-    },
-    {
-      "epoch": 1.3656346873867709,
-      "grad_norm": 0.0015177465975284576,
-      "learning_rate": 0.00019999908190193453,
-      "loss": 46.0,
-      "step": 8480
-    },
-    {
-      "epoch": 1.365795724465558,
-      "grad_norm": 0.0033245442900806665,
-      "learning_rate": 0.00019999908168513386,
-      "loss": 46.0,
-      "step": 8481
-    },
-    {
-      "epoch": 1.3659567615443455,
-      "grad_norm": 0.0016388514777645469,
-      "learning_rate": 0.00019999908146830757,
-      "loss": 46.0,
-      "step": 8482
-    },
-    {
-      "epoch": 1.366117798623133,
-      "grad_norm": 0.0009756018407642841,
-      "learning_rate": 0.00019999908125145573,
-      "loss": 46.0,
-      "step": 8483
-    },
-    {
-      "epoch": 1.3662788357019204,
-      "grad_norm": 0.0023046028800308704,
-      "learning_rate": 0.00019999908103457824,
-      "loss": 46.0,
-      "step": 8484
-    },
-    {
-      "epoch": 1.3664398727807079,
-      "grad_norm": 0.0067170062102377415,
-      "learning_rate": 0.00019999908081767522,
-      "loss": 46.0,
-      "step": 8485
-    },
-    {
-      "epoch": 1.3666009098594951,
-      "grad_norm": 0.002000573556870222,
-      "learning_rate": 0.00019999908060074656,
-      "loss": 46.0,
-      "step": 8486
-    },
-    {
-      "epoch": 1.3667619469382826,
-      "grad_norm": 0.0027586210053414106,
-      "learning_rate": 0.0001999990803837923,
-      "loss": 46.0,
-      "step": 8487
-    },
-    {
-      "epoch": 1.3669229840170698,
-      "grad_norm": 0.0005638182628899813,
-      "learning_rate": 0.00019999908016681247,
-      "loss": 46.0,
-      "step": 8488
-    },
-    {
-      "epoch": 1.3670840210958572,
-      "grad_norm": 0.00064820860279724,
-      "learning_rate": 0.00019999907994980705,
-      "loss": 46.0,
-      "step": 8489
-    },
-    {
-      "epoch": 1.3672450581746447,
-      "grad_norm": 0.0018096237909048796,
-      "learning_rate": 0.00019999907973277604,
-      "loss": 46.0,
-      "step": 8490
-    },
-    {
-      "epoch": 1.3674060952534322,
-      "grad_norm": 0.0019132404122501612,
-      "learning_rate": 0.0001999990795157194,
-      "loss": 46.0,
-      "step": 8491
-    },
-    {
-      "epoch": 1.3675671323322196,
-      "grad_norm": 0.003457923186942935,
-      "learning_rate": 0.00019999907929863718,
-      "loss": 46.0,
-      "step": 8492
-    },
-    {
-      "epoch": 1.3677281694110068,
-      "grad_norm": 0.0004081083170603961,
-      "learning_rate": 0.00019999907908152938,
-      "loss": 46.0,
-      "step": 8493
-    },
-    {
-      "epoch": 1.3678892064897943,
-      "grad_norm": 0.0013288760092109442,
-      "learning_rate": 0.00019999907886439596,
-      "loss": 46.0,
-      "step": 8494
-    },
-    {
-      "epoch": 1.3680502435685817,
-      "grad_norm": 0.0012940262677147985,
-      "learning_rate": 0.00019999907864723696,
-      "loss": 46.0,
-      "step": 8495
-    },
-    {
-      "epoch": 1.368211280647369,
-      "grad_norm": 0.003123154165223241,
-      "learning_rate": 0.00019999907843005238,
-      "loss": 46.0,
-      "step": 8496
-    },
-    {
-      "epoch": 1.3683723177261564,
-      "grad_norm": 0.0014393990859389305,
-      "learning_rate": 0.00019999907821284218,
-      "loss": 46.0,
-      "step": 8497
-    },
-    {
-      "epoch": 1.3685333548049439,
-      "grad_norm": 0.0007467317045666277,
-      "learning_rate": 0.0001999990779956064,
-      "loss": 46.0,
-      "step": 8498
-    },
-    {
-      "epoch": 1.3686943918837313,
-      "grad_norm": 0.003697833977639675,
-      "learning_rate": 0.000199999077778345,
-      "loss": 46.0,
-      "step": 8499
-    },
-    {
-      "epoch": 1.3688554289625186,
-      "grad_norm": 0.0012142921332269907,
-      "learning_rate": 0.00019999907756105807,
-      "loss": 46.0,
-      "step": 8500
-    },
-    {
-      "epoch": 1.369016466041306,
-      "grad_norm": 0.0026814532466232777,
-      "learning_rate": 0.0001999990773437455,
-      "loss": 46.0,
-      "step": 8501
-    },
-    {
-      "epoch": 1.3691775031200935,
-      "grad_norm": 0.0012107130605727434,
-      "learning_rate": 0.0001999990771264073,
-      "loss": 46.0,
-      "step": 8502
-    },
-    {
-      "epoch": 1.3693385401988807,
-      "grad_norm": 0.0011837449856102467,
-      "learning_rate": 0.00019999907690904358,
-      "loss": 46.0,
-      "step": 8503
-    },
-    {
-      "epoch": 1.3694995772776681,
-      "grad_norm": 0.011731158941984177,
-      "learning_rate": 0.00019999907669165424,
-      "loss": 46.0,
-      "step": 8504
-    },
-    {
-      "epoch": 1.3696606143564556,
-      "grad_norm": 0.0006542187184095383,
-      "learning_rate": 0.00019999907647423926,
-      "loss": 46.0,
-      "step": 8505
-    },
-    {
-      "epoch": 1.369821651435243,
-      "grad_norm": 0.004485903307795525,
-      "learning_rate": 0.00019999907625679872,
-      "loss": 46.0,
-      "step": 8506
-    },
-    {
-      "epoch": 1.3699826885140305,
-      "grad_norm": 0.004687386564910412,
-      "learning_rate": 0.0001999990760393326,
-      "loss": 46.0,
-      "step": 8507
-    },
-    {
-      "epoch": 1.3701437255928177,
-      "grad_norm": 0.003480333834886551,
-      "learning_rate": 0.00019999907582184085,
-      "loss": 46.0,
-      "step": 8508
-    },
-    {
-      "epoch": 1.3703047626716052,
-      "grad_norm": 0.0032353985588997602,
-      "learning_rate": 0.00019999907560432355,
-      "loss": 46.0,
-      "step": 8509
-    },
-    {
-      "epoch": 1.3704657997503924,
-      "grad_norm": 0.0030441845301538706,
-      "learning_rate": 0.00019999907538678063,
-      "loss": 46.0,
-      "step": 8510
-    },
-    {
-      "epoch": 1.3706268368291799,
-      "grad_norm": 0.009550451301038265,
-      "learning_rate": 0.0001999990751692121,
-      "loss": 46.0,
-      "step": 8511
-    },
-    {
-      "epoch": 1.3707878739079673,
-      "grad_norm": 0.002966528292745352,
-      "learning_rate": 0.00019999907495161799,
-      "loss": 46.0,
-      "step": 8512
-    },
-    {
-      "epoch": 1.3709489109867548,
-      "grad_norm": 0.007207896094769239,
-      "learning_rate": 0.00019999907473399828,
-      "loss": 46.0,
-      "step": 8513
-    },
-    {
-      "epoch": 1.3711099480655422,
-      "grad_norm": 0.001371773425489664,
-      "learning_rate": 0.00019999907451635302,
-      "loss": 46.0,
-      "step": 8514
-    },
-    {
-      "epoch": 1.3712709851443294,
-      "grad_norm": 0.004324706271290779,
-      "learning_rate": 0.0001999990742986821,
-      "loss": 46.0,
-      "step": 8515
-    },
-    {
-      "epoch": 1.3714320222231169,
-      "grad_norm": 0.004359810147434473,
-      "learning_rate": 0.00019999907408098562,
-      "loss": 46.0,
-      "step": 8516
-    },
-    {
-      "epoch": 1.3715930593019043,
-      "grad_norm": 0.008214239962399006,
-      "learning_rate": 0.00019999907386326356,
-      "loss": 46.0,
-      "step": 8517
-    },
-    {
-      "epoch": 1.3717540963806916,
-      "grad_norm": 0.0034329467453062534,
-      "learning_rate": 0.00019999907364551587,
-      "loss": 46.0,
-      "step": 8518
-    },
-    {
-      "epoch": 1.371915133459479,
-      "grad_norm": 0.003290883032605052,
-      "learning_rate": 0.0001999990734277426,
-      "loss": 46.0,
-      "step": 8519
-    },
-    {
-      "epoch": 1.3720761705382665,
-      "grad_norm": 0.002334638498723507,
-      "learning_rate": 0.00019999907320994372,
-      "loss": 46.0,
-      "step": 8520
-    },
-    {
-      "epoch": 1.372237207617054,
-      "grad_norm": 0.00335302553139627,
-      "learning_rate": 0.00019999907299211926,
-      "loss": 46.0,
-      "step": 8521
-    },
-    {
-      "epoch": 1.3723982446958412,
-      "grad_norm": 0.002353750402107835,
-      "learning_rate": 0.00019999907277426922,
-      "loss": 46.0,
-      "step": 8522
-    },
-    {
-      "epoch": 1.3725592817746286,
-      "grad_norm": 0.0003986102237831801,
-      "learning_rate": 0.00019999907255639356,
-      "loss": 46.0,
-      "step": 8523
-    },
-    {
-      "epoch": 1.372720318853416,
-      "grad_norm": 0.0026938191149383783,
-      "learning_rate": 0.00019999907233849231,
-      "loss": 46.0,
-      "step": 8524
-    },
-    {
-      "epoch": 1.3728813559322033,
-      "grad_norm": 0.006727281026542187,
-      "learning_rate": 0.00019999907212056548,
-      "loss": 46.0,
-      "step": 8525
-    },
-    {
-      "epoch": 1.3730423930109907,
-      "grad_norm": 0.0052147917449474335,
-      "learning_rate": 0.00019999907190261303,
-      "loss": 46.0,
-      "step": 8526
-    },
-    {
-      "epoch": 1.3732034300897782,
-      "grad_norm": 0.0011838451027870178,
-      "learning_rate": 0.00019999907168463503,
-      "loss": 46.0,
-      "step": 8527
-    },
-    {
-      "epoch": 1.3733644671685656,
-      "grad_norm": 0.0006170262349769473,
-      "learning_rate": 0.00019999907146663138,
-      "loss": 46.0,
-      "step": 8528
-    },
-    {
-      "epoch": 1.373525504247353,
-      "grad_norm": 0.0008709498215466738,
-      "learning_rate": 0.00019999907124860217,
-      "loss": 46.0,
-      "step": 8529
-    },
-    {
-      "epoch": 1.3736865413261403,
-      "grad_norm": 0.0012861124705523252,
-      "learning_rate": 0.00019999907103054735,
-      "loss": 46.0,
-      "step": 8530
-    },
-    {
-      "epoch": 1.3738475784049278,
-      "grad_norm": 0.002076564123854041,
-      "learning_rate": 0.00019999907081246697,
-      "loss": 46.0,
-      "step": 8531
-    },
-    {
-      "epoch": 1.374008615483715,
-      "grad_norm": 0.0015019255224615335,
-      "learning_rate": 0.00019999907059436097,
-      "loss": 46.0,
-      "step": 8532
-    },
-    {
-      "epoch": 1.3741696525625025,
-      "grad_norm": 0.0017158534610643983,
-      "learning_rate": 0.00019999907037622936,
-      "loss": 46.0,
-      "step": 8533
-    },
-    {
-      "epoch": 1.37433068964129,
-      "grad_norm": 0.002254312625154853,
-      "learning_rate": 0.0001999990701580722,
-      "loss": 46.0,
-      "step": 8534
-    },
-    {
-      "epoch": 1.3744917267200774,
-      "grad_norm": 0.0016968031413853168,
-      "learning_rate": 0.0001999990699398894,
-      "loss": 46.0,
-      "step": 8535
-    },
-    {
-      "epoch": 1.3746527637988648,
-      "grad_norm": 0.0005931690684519708,
-      "learning_rate": 0.000199999069721681,
-      "loss": 46.0,
-      "step": 8536
-    },
-    {
-      "epoch": 1.374813800877652,
-      "grad_norm": 0.002256153617054224,
-      "learning_rate": 0.00019999906950344705,
-      "loss": 46.0,
-      "step": 8537
-    },
-    {
-      "epoch": 1.3749748379564395,
-      "grad_norm": 0.0028346905019134283,
-      "learning_rate": 0.00019999906928518744,
-      "loss": 46.0,
-      "step": 8538
-    },
-    {
-      "epoch": 1.375135875035227,
-      "grad_norm": 0.0025116789620369673,
-      "learning_rate": 0.00019999906906690228,
-      "loss": 46.0,
-      "step": 8539
-    },
-    {
-      "epoch": 1.3752969121140142,
-      "grad_norm": 0.0016416899161413312,
-      "learning_rate": 0.00019999906884859154,
-      "loss": 46.0,
-      "step": 8540
-    },
-    {
-      "epoch": 1.3754579491928016,
-      "grad_norm": 0.0009125908836722374,
-      "learning_rate": 0.0001999990686302552,
-      "loss": 46.0,
-      "step": 8541
-    },
-    {
-      "epoch": 1.375618986271589,
-      "grad_norm": 0.002481288742274046,
-      "learning_rate": 0.00019999906841189325,
-      "loss": 46.0,
-      "step": 8542
-    },
-    {
-      "epoch": 1.3757800233503765,
-      "grad_norm": 0.0016046366654336452,
-      "learning_rate": 0.00019999906819350569,
-      "loss": 46.0,
-      "step": 8543
-    },
-    {
-      "epoch": 1.3759410604291638,
-      "grad_norm": 0.0023816947359591722,
-      "learning_rate": 0.0001999990679750926,
-      "loss": 46.0,
-      "step": 8544
-    },
-    {
-      "epoch": 1.3761020975079512,
-      "grad_norm": 0.0014378972118720412,
-      "learning_rate": 0.00019999906775665385,
-      "loss": 46.0,
-      "step": 8545
-    },
-    {
-      "epoch": 1.3762631345867387,
-      "grad_norm": 0.0006608270341530442,
-      "learning_rate": 0.0001999990675381895,
-      "loss": 46.0,
-      "step": 8546
-    },
-    {
-      "epoch": 1.3764241716655259,
-      "grad_norm": 0.0014959038235247135,
-      "learning_rate": 0.00019999906731969959,
-      "loss": 46.0,
-      "step": 8547
-    },
-    {
-      "epoch": 1.3765852087443133,
-      "grad_norm": 0.0011902515543624759,
-      "learning_rate": 0.0001999990671011841,
-      "loss": 46.0,
-      "step": 8548
-    },
-    {
-      "epoch": 1.3767462458231008,
-      "grad_norm": 0.009871224872767925,
-      "learning_rate": 0.00019999906688264297,
-      "loss": 46.0,
-      "step": 8549
-    },
-    {
-      "epoch": 1.3769072829018882,
-      "grad_norm": 0.0010699726408347487,
-      "learning_rate": 0.00019999906666407627,
-      "loss": 46.0,
-      "step": 8550
-    },
-    {
-      "epoch": 1.3770683199806755,
-      "grad_norm": 0.008970269002020359,
-      "learning_rate": 0.00019999906644548396,
-      "loss": 46.0,
-      "step": 8551
-    },
-    {
-      "epoch": 1.377229357059463,
-      "grad_norm": 0.0017508044838905334,
-      "learning_rate": 0.00019999906622686608,
-      "loss": 46.0,
-      "step": 8552
-    },
-    {
-      "epoch": 1.3773903941382504,
-      "grad_norm": 0.001487939734943211,
-      "learning_rate": 0.0001999990660082226,
-      "loss": 46.0,
-      "step": 8553
-    },
-    {
-      "epoch": 1.3775514312170376,
-      "grad_norm": 0.0007740810397081077,
-      "learning_rate": 0.00019999906578955349,
-      "loss": 46.0,
-      "step": 8554
-    },
-    {
-      "epoch": 1.377712468295825,
-      "grad_norm": 0.004181104712188244,
-      "learning_rate": 0.00019999906557085882,
-      "loss": 46.0,
-      "step": 8555
-    },
-    {
-      "epoch": 1.3778735053746125,
-      "grad_norm": 0.0027874098159372807,
-      "learning_rate": 0.00019999906535213854,
-      "loss": 46.0,
-      "step": 8556
-    },
-    {
-      "epoch": 1.3780345424534,
-      "grad_norm": 0.008951113559305668,
-      "learning_rate": 0.00019999906513339268,
-      "loss": 46.0,
-      "step": 8557
-    },
-    {
-      "epoch": 1.3781955795321874,
-      "grad_norm": 0.0015791063196957111,
-      "learning_rate": 0.00019999906491462123,
-      "loss": 46.0,
-      "step": 8558
-    },
-    {
-      "epoch": 1.3783566166109746,
-      "grad_norm": 0.001212267903611064,
-      "learning_rate": 0.00019999906469582416,
-      "loss": 46.0,
-      "step": 8559
-    },
-    {
-      "epoch": 1.378517653689762,
-      "grad_norm": 0.0009738767403177917,
-      "learning_rate": 0.00019999906447700153,
-      "loss": 46.0,
-      "step": 8560
-    },
-    {
-      "epoch": 1.3786786907685493,
-      "grad_norm": 0.0019429456442594528,
-      "learning_rate": 0.00019999906425815326,
-      "loss": 46.0,
-      "step": 8561
-    },
-    {
-      "epoch": 1.3788397278473368,
-      "grad_norm": 0.005254549905657768,
-      "learning_rate": 0.00019999906403927943,
-      "loss": 46.0,
-      "step": 8562
-    },
-    {
-      "epoch": 1.3790007649261242,
-      "grad_norm": 0.00786639004945755,
-      "learning_rate": 0.00019999906382038002,
-      "loss": 46.0,
-      "step": 8563
-    },
-    {
-      "epoch": 1.3791618020049117,
-      "grad_norm": 0.0016958412015810609,
-      "learning_rate": 0.00019999906360145496,
-      "loss": 46.0,
-      "step": 8564
-    },
-    {
-      "epoch": 1.3793228390836991,
-      "grad_norm": 0.004124409053474665,
-      "learning_rate": 0.00019999906338250434,
-      "loss": 46.0,
-      "step": 8565
-    },
-    {
-      "epoch": 1.3794838761624864,
-      "grad_norm": 0.0012453460367396474,
-      "learning_rate": 0.0001999990631635281,
-      "loss": 46.0,
-      "step": 8566
-    },
-    {
-      "epoch": 1.3796449132412738,
-      "grad_norm": 0.004079313017427921,
-      "learning_rate": 0.0001999990629445263,
-      "loss": 46.0,
-      "step": 8567
-    },
-    {
-      "epoch": 1.3798059503200613,
-      "grad_norm": 0.0018956586718559265,
-      "learning_rate": 0.0001999990627254989,
-      "loss": 46.0,
-      "step": 8568
-    },
-    {
-      "epoch": 1.3799669873988485,
-      "grad_norm": 0.005786104127764702,
-      "learning_rate": 0.0001999990625064459,
-      "loss": 46.0,
-      "step": 8569
-    },
-    {
-      "epoch": 1.380128024477636,
-      "grad_norm": 0.0016839215531945229,
-      "learning_rate": 0.00019999906228736732,
-      "loss": 46.0,
-      "step": 8570
-    },
-    {
-      "epoch": 1.3802890615564234,
-      "grad_norm": 0.0018276222981512547,
-      "learning_rate": 0.00019999906206826312,
-      "loss": 46.0,
-      "step": 8571
-    },
-    {
-      "epoch": 1.3804500986352108,
-      "grad_norm": 0.0021915456745773554,
-      "learning_rate": 0.00019999906184913334,
-      "loss": 46.0,
-      "step": 8572
-    },
-    {
-      "epoch": 1.380611135713998,
-      "grad_norm": 0.001955562736839056,
-      "learning_rate": 0.00019999906162997794,
-      "loss": 46.0,
-      "step": 8573
-    },
-    {
-      "epoch": 1.3807721727927855,
-      "grad_norm": 0.003083518473431468,
-      "learning_rate": 0.000199999061410797,
-      "loss": 46.0,
-      "step": 8574
-    },
-    {
-      "epoch": 1.380933209871573,
-      "grad_norm": 0.0024714069440960884,
-      "learning_rate": 0.0001999990611915904,
-      "loss": 46.0,
-      "step": 8575
-    },
-    {
-      "epoch": 1.3810942469503602,
-      "grad_norm": 0.0015121960313990712,
-      "learning_rate": 0.00019999906097235823,
-      "loss": 46.0,
-      "step": 8576
-    },
-    {
-      "epoch": 1.3812552840291477,
-      "grad_norm": 0.001893801148980856,
-      "learning_rate": 0.00019999906075310046,
-      "loss": 46.0,
-      "step": 8577
-    },
-    {
-      "epoch": 1.381416321107935,
-      "grad_norm": 0.0029527561273425817,
-      "learning_rate": 0.00019999906053381713,
-      "loss": 46.0,
-      "step": 8578
-    },
-    {
-      "epoch": 1.3815773581867226,
-      "grad_norm": 0.002330329967662692,
-      "learning_rate": 0.00019999906031450816,
-      "loss": 46.0,
-      "step": 8579
-    },
-    {
-      "epoch": 1.38173839526551,
-      "grad_norm": 0.0015530623495578766,
-      "learning_rate": 0.00019999906009517365,
-      "loss": 46.0,
-      "step": 8580
-    },
-    {
-      "epoch": 1.3818994323442972,
-      "grad_norm": 0.00039809252484701574,
-      "learning_rate": 0.00019999905987581348,
-      "loss": 46.0,
-      "step": 8581
-    },
-    {
-      "epoch": 1.3820604694230847,
-      "grad_norm": 0.0032321594189852476,
-      "learning_rate": 0.00019999905965642777,
-      "loss": 46.0,
-      "step": 8582
-    },
-    {
-      "epoch": 1.382221506501872,
-      "grad_norm": 0.0016787732020020485,
-      "learning_rate": 0.00019999905943701642,
-      "loss": 46.0,
-      "step": 8583
-    },
-    {
-      "epoch": 1.3823825435806594,
-      "grad_norm": 0.0013373702531680465,
-      "learning_rate": 0.0001999990592175795,
-      "loss": 46.0,
-      "step": 8584
-    },
-    {
-      "epoch": 1.3825435806594468,
-      "grad_norm": 0.0046799397096037865,
-      "learning_rate": 0.000199999058998117,
-      "loss": 46.0,
-      "step": 8585
-    },
-    {
-      "epoch": 1.3827046177382343,
-      "grad_norm": 0.007765064015984535,
-      "learning_rate": 0.00019999905877862888,
-      "loss": 46.0,
-      "step": 8586
-    },
-    {
-      "epoch": 1.3828656548170217,
-      "grad_norm": 0.005820256657898426,
-      "learning_rate": 0.00019999905855911515,
-      "loss": 46.0,
-      "step": 8587
-    },
-    {
-      "epoch": 1.383026691895809,
-      "grad_norm": 0.002248318400233984,
-      "learning_rate": 0.0001999990583395759,
-      "loss": 46.0,
-      "step": 8588
-    },
-    {
-      "epoch": 1.3831877289745964,
-      "grad_norm": 0.0011756689054891467,
-      "learning_rate": 0.000199999058120011,
-      "loss": 46.0,
-      "step": 8589
-    },
-    {
-      "epoch": 1.3833487660533839,
-      "grad_norm": 0.0028867514338344336,
-      "learning_rate": 0.0001999990579004205,
-      "loss": 46.0,
-      "step": 8590
-    },
-    {
-      "epoch": 1.383509803132171,
-      "grad_norm": 0.0005549927009269595,
-      "learning_rate": 0.0001999990576808044,
-      "loss": 46.0,
-      "step": 8591
-    },
-    {
-      "epoch": 1.3836708402109585,
-      "grad_norm": 0.0016280576819553971,
-      "learning_rate": 0.00019999905746116272,
-      "loss": 46.0,
-      "step": 8592
-    },
-    {
-      "epoch": 1.383831877289746,
-      "grad_norm": 0.003444229019805789,
-      "learning_rate": 0.00019999905724149545,
-      "loss": 46.0,
-      "step": 8593
-    },
-    {
-      "epoch": 1.3839929143685334,
-      "grad_norm": 0.0005017066723667085,
-      "learning_rate": 0.00019999905702180258,
-      "loss": 46.0,
-      "step": 8594
-    },
-    {
-      "epoch": 1.3841539514473207,
-      "grad_norm": 0.003034519497305155,
-      "learning_rate": 0.00019999905680208413,
-      "loss": 46.0,
-      "step": 8595
-    },
-    {
-      "epoch": 1.3843149885261081,
-      "grad_norm": 0.0020030911546200514,
-      "learning_rate": 0.00019999905658234007,
-      "loss": 46.0,
-      "step": 8596
-    },
-    {
-      "epoch": 1.3844760256048956,
-      "grad_norm": 0.003164656227454543,
-      "learning_rate": 0.00019999905636257042,
-      "loss": 46.0,
-      "step": 8597
-    },
-    {
-      "epoch": 1.3846370626836828,
-      "grad_norm": 0.0007547682034783065,
-      "learning_rate": 0.00019999905614277516,
-      "loss": 46.0,
-      "step": 8598
-    },
-    {
-      "epoch": 1.3847980997624703,
-      "grad_norm": 0.004079625476151705,
-      "learning_rate": 0.0001999990559229543,
-      "loss": 46.0,
-      "step": 8599
-    },
-    {
-      "epoch": 1.3849591368412577,
-      "grad_norm": 0.00556177319958806,
-      "learning_rate": 0.00019999905570310792,
-      "loss": 46.0,
-      "step": 8600
-    },
-    {
-      "epoch": 1.3851201739200452,
-      "grad_norm": 0.002690773457288742,
-      "learning_rate": 0.00019999905548323587,
-      "loss": 46.0,
-      "step": 8601
-    },
-    {
-      "epoch": 1.3852812109988326,
-      "grad_norm": 0.0009012288646772504,
-      "learning_rate": 0.00019999905526333825,
-      "loss": 46.0,
-      "step": 8602
-    },
-    {
-      "epoch": 1.3854422480776198,
-      "grad_norm": 0.0029388947878032923,
-      "learning_rate": 0.00019999905504341502,
-      "loss": 46.0,
-      "step": 8603
-    },
-    {
-      "epoch": 1.3856032851564073,
-      "grad_norm": 0.0008120383718051016,
-      "learning_rate": 0.0001999990548234662,
-      "loss": 46.0,
-      "step": 8604
-    },
-    {
-      "epoch": 1.3857643222351945,
-      "grad_norm": 0.0033238877076655626,
-      "learning_rate": 0.0001999990546034918,
-      "loss": 46.0,
-      "step": 8605
-    },
-    {
-      "epoch": 1.385925359313982,
-      "grad_norm": 0.00048069297918118536,
-      "learning_rate": 0.0001999990543834918,
-      "loss": 46.0,
-      "step": 8606
-    },
-    {
-      "epoch": 1.3860863963927694,
-      "grad_norm": 0.0004616714140865952,
-      "learning_rate": 0.0001999990541634662,
-      "loss": 46.0,
-      "step": 8607
-    },
-    {
-      "epoch": 1.3862474334715569,
-      "grad_norm": 0.0020958795212209225,
-      "learning_rate": 0.000199999053943415,
-      "loss": 46.0,
-      "step": 8608
-    },
-    {
-      "epoch": 1.3864084705503443,
-      "grad_norm": 0.005875249858945608,
-      "learning_rate": 0.00019999905372333822,
-      "loss": 46.0,
-      "step": 8609
-    },
-    {
-      "epoch": 1.3865695076291316,
-      "grad_norm": 0.00039585522608831525,
-      "learning_rate": 0.00019999905350323585,
-      "loss": 46.0,
-      "step": 8610
-    },
-    {
-      "epoch": 1.386730544707919,
-      "grad_norm": 0.002378098201006651,
-      "learning_rate": 0.00019999905328310785,
-      "loss": 46.0,
-      "step": 8611
-    },
-    {
-      "epoch": 1.3868915817867065,
-      "grad_norm": 0.009137958288192749,
-      "learning_rate": 0.00019999905306295428,
-      "loss": 46.0,
-      "step": 8612
-    },
-    {
-      "epoch": 1.3870526188654937,
-      "grad_norm": 0.0042574042454361916,
-      "learning_rate": 0.00019999905284277513,
-      "loss": 46.0,
-      "step": 8613
-    },
-    {
-      "epoch": 1.3872136559442811,
-      "grad_norm": 0.0010369190713390708,
-      "learning_rate": 0.00019999905262257036,
-      "loss": 46.0,
-      "step": 8614
-    },
-    {
-      "epoch": 1.3873746930230686,
-      "grad_norm": 0.005388530436903238,
-      "learning_rate": 0.00019999905240234,
-      "loss": 46.0,
-      "step": 8615
-    },
-    {
-      "epoch": 1.387535730101856,
-      "grad_norm": 0.005703634582459927,
-      "learning_rate": 0.00019999905218208406,
-      "loss": 46.0,
-      "step": 8616
-    },
-    {
-      "epoch": 1.3876967671806433,
-      "grad_norm": 0.004314677324146032,
-      "learning_rate": 0.0001999990519618025,
-      "loss": 46.0,
-      "step": 8617
-    },
-    {
-      "epoch": 1.3878578042594307,
-      "grad_norm": 0.0030692103318870068,
-      "learning_rate": 0.00019999905174149536,
-      "loss": 46.0,
-      "step": 8618
-    },
-    {
-      "epoch": 1.3880188413382182,
-      "grad_norm": 0.0033350626472383738,
-      "learning_rate": 0.00019999905152116263,
-      "loss": 46.0,
-      "step": 8619
-    },
-    {
-      "epoch": 1.3881798784170054,
-      "grad_norm": 0.001150695257820189,
-      "learning_rate": 0.0001999990513008043,
-      "loss": 46.0,
-      "step": 8620
-    },
-    {
-      "epoch": 1.3883409154957929,
-      "grad_norm": 0.0031372515950351954,
-      "learning_rate": 0.00019999905108042038,
-      "loss": 46.0,
-      "step": 8621
-    },
-    {
-      "epoch": 1.3885019525745803,
-      "grad_norm": 0.0019240840338170528,
-      "learning_rate": 0.00019999905086001089,
-      "loss": 46.0,
-      "step": 8622
-    },
-    {
-      "epoch": 1.3886629896533678,
-      "grad_norm": 0.0004467913240659982,
-      "learning_rate": 0.00019999905063957575,
-      "loss": 46.0,
-      "step": 8623
-    },
-    {
-      "epoch": 1.3888240267321552,
-      "grad_norm": 0.0076491315849125385,
-      "learning_rate": 0.00019999905041911503,
-      "loss": 46.0,
-      "step": 8624
-    },
-    {
-      "epoch": 1.3889850638109424,
-      "grad_norm": 0.0015910729998722672,
-      "learning_rate": 0.00019999905019862875,
-      "loss": 46.0,
-      "step": 8625
-    },
-    {
-      "epoch": 1.38914610088973,
-      "grad_norm": 0.00404138071462512,
-      "learning_rate": 0.00019999904997811685,
-      "loss": 46.0,
-      "step": 8626
-    },
-    {
-      "epoch": 1.3893071379685171,
-      "grad_norm": 0.0019203597912564874,
-      "learning_rate": 0.00019999904975757934,
-      "loss": 46.0,
-      "step": 8627
-    },
-    {
-      "epoch": 1.3894681750473046,
-      "grad_norm": 0.0036678712349385023,
-      "learning_rate": 0.00019999904953701627,
-      "loss": 46.0,
-      "step": 8628
-    },
-    {
-      "epoch": 1.389629212126092,
-      "grad_norm": 0.004156417679041624,
-      "learning_rate": 0.0001999990493164276,
-      "loss": 46.0,
-      "step": 8629
-    },
-    {
-      "epoch": 1.3897902492048795,
-      "grad_norm": 0.000452744570793584,
-      "learning_rate": 0.00019999904909581332,
-      "loss": 46.0,
-      "step": 8630
-    },
-    {
-      "epoch": 1.389951286283667,
-      "grad_norm": 0.0016183399129658937,
-      "learning_rate": 0.00019999904887517343,
-      "loss": 46.0,
-      "step": 8631
-    },
-    {
-      "epoch": 1.3901123233624542,
-      "grad_norm": 0.0013192167971283197,
-      "learning_rate": 0.000199999048654508,
-      "loss": 46.0,
-      "step": 8632
-    },
-    {
-      "epoch": 1.3902733604412416,
-      "grad_norm": 0.004599005915224552,
-      "learning_rate": 0.00019999904843381693,
-      "loss": 46.0,
-      "step": 8633
-    },
-    {
-      "epoch": 1.390434397520029,
-      "grad_norm": 0.0008950078627094626,
-      "learning_rate": 0.00019999904821310028,
-      "loss": 46.0,
-      "step": 8634
-    },
-    {
-      "epoch": 1.3905954345988163,
-      "grad_norm": 0.001665567047894001,
-      "learning_rate": 0.00019999904799235802,
-      "loss": 46.0,
-      "step": 8635
-    },
-    {
-      "epoch": 1.3907564716776037,
-      "grad_norm": 0.003530068090185523,
-      "learning_rate": 0.0001999990477715902,
-      "loss": 46.0,
-      "step": 8636
-    },
-    {
-      "epoch": 1.3909175087563912,
-      "grad_norm": 0.0019249017350375652,
-      "learning_rate": 0.00019999904755079674,
-      "loss": 46.0,
-      "step": 8637
-    },
-    {
-      "epoch": 1.3910785458351786,
-      "grad_norm": 0.0029693576507270336,
-      "learning_rate": 0.0001999990473299777,
-      "loss": 46.0,
-      "step": 8638
-    },
-    {
-      "epoch": 1.3912395829139659,
-      "grad_norm": 0.001407565432600677,
-      "learning_rate": 0.00019999904710913308,
-      "loss": 46.0,
-      "step": 8639
-    },
-    {
-      "epoch": 1.3914006199927533,
-      "grad_norm": 0.007015765644609928,
-      "learning_rate": 0.00019999904688826288,
-      "loss": 46.0,
-      "step": 8640
-    },
-    {
-      "epoch": 1.3915616570715408,
-      "grad_norm": 0.003454859834164381,
-      "learning_rate": 0.00019999904666736704,
-      "loss": 46.0,
-      "step": 8641
-    },
-    {
-      "epoch": 1.391722694150328,
-      "grad_norm": 0.0015796166844666004,
-      "learning_rate": 0.00019999904644644564,
-      "loss": 46.0,
-      "step": 8642
-    },
-    {
-      "epoch": 1.3918837312291155,
-      "grad_norm": 0.005556751508265734,
-      "learning_rate": 0.00019999904622549863,
-      "loss": 46.0,
-      "step": 8643
-    },
-    {
-      "epoch": 1.392044768307903,
-      "grad_norm": 0.0014212372479960322,
-      "learning_rate": 0.00019999904600452603,
-      "loss": 46.0,
-      "step": 8644
-    },
-    {
-      "epoch": 1.3922058053866904,
-      "grad_norm": 0.0007226198213174939,
-      "learning_rate": 0.00019999904578352787,
-      "loss": 46.0,
-      "step": 8645
-    },
-    {
-      "epoch": 1.3923668424654776,
-      "grad_norm": 0.007025664672255516,
-      "learning_rate": 0.00019999904556250407,
-      "loss": 46.0,
-      "step": 8646
-    },
-    {
-      "epoch": 1.392527879544265,
-      "grad_norm": 0.0009287431603297591,
-      "learning_rate": 0.00019999904534145468,
-      "loss": 46.0,
-      "step": 8647
-    },
-    {
-      "epoch": 1.3926889166230525,
-      "grad_norm": 0.004047059919685125,
-      "learning_rate": 0.0001999990451203797,
-      "loss": 46.0,
-      "step": 8648
-    },
-    {
-      "epoch": 1.3928499537018397,
-      "grad_norm": 0.003496703924611211,
-      "learning_rate": 0.0001999990448992791,
-      "loss": 46.0,
-      "step": 8649
-    },
-    {
-      "epoch": 1.3930109907806272,
-      "grad_norm": 0.0007651507621631026,
-      "learning_rate": 0.00019999904467815296,
-      "loss": 46.0,
-      "step": 8650
-    },
-    {
-      "epoch": 1.3931720278594146,
-      "grad_norm": 0.003980850335210562,
-      "learning_rate": 0.0001999990444570012,
-      "loss": 46.0,
-      "step": 8651
-    },
-    {
-      "epoch": 1.393333064938202,
-      "grad_norm": 0.007391641847789288,
-      "learning_rate": 0.00019999904423582385,
-      "loss": 46.0,
-      "step": 8652
-    },
-    {
-      "epoch": 1.3934941020169895,
-      "grad_norm": 0.008265133015811443,
-      "learning_rate": 0.00019999904401462088,
-      "loss": 46.0,
-      "step": 8653
-    },
-    {
-      "epoch": 1.3936551390957768,
-      "grad_norm": 0.00351297901943326,
-      "learning_rate": 0.00019999904379339233,
-      "loss": 46.0,
-      "step": 8654
-    },
-    {
-      "epoch": 1.3938161761745642,
-      "grad_norm": 0.0011872978648170829,
-      "learning_rate": 0.00019999904357213821,
-      "loss": 46.0,
-      "step": 8655
-    },
-    {
-      "epoch": 1.3939772132533517,
-      "grad_norm": 0.0021320045925676823,
-      "learning_rate": 0.00019999904335085846,
-      "loss": 46.0,
-      "step": 8656
-    },
-    {
-      "epoch": 1.394138250332139,
-      "grad_norm": 0.0029541810508817434,
-      "learning_rate": 0.00019999904312955315,
-      "loss": 46.0,
-      "step": 8657
-    },
-    {
-      "epoch": 1.3942992874109263,
-      "grad_norm": 0.004228669218719006,
-      "learning_rate": 0.00019999904290822222,
-      "loss": 46.0,
-      "step": 8658
-    },
-    {
-      "epoch": 1.3944603244897138,
-      "grad_norm": 0.0008973954827524722,
-      "learning_rate": 0.0001999990426868657,
-      "loss": 46.0,
-      "step": 8659
-    },
-    {
-      "epoch": 1.3946213615685012,
-      "grad_norm": 0.00026754557620733976,
-      "learning_rate": 0.00019999904246548357,
-      "loss": 46.0,
-      "step": 8660
-    },
-    {
-      "epoch": 1.3947823986472885,
-      "grad_norm": 0.0018369912868365645,
-      "learning_rate": 0.00019999904224407588,
-      "loss": 46.0,
-      "step": 8661
-    },
-    {
-      "epoch": 1.394943435726076,
-      "grad_norm": 0.0009069963707588613,
-      "learning_rate": 0.00019999904202264258,
-      "loss": 46.0,
-      "step": 8662
-    },
-    {
-      "epoch": 1.3951044728048634,
-      "grad_norm": 0.0014931592158973217,
-      "learning_rate": 0.00019999904180118368,
-      "loss": 46.0,
-      "step": 8663
-    },
-    {
-      "epoch": 1.3952655098836506,
-      "grad_norm": 0.0027464760933071375,
-      "learning_rate": 0.0001999990415796992,
-      "loss": 46.0,
-      "step": 8664
-    },
-    {
-      "epoch": 1.395426546962438,
-      "grad_norm": 0.0018627849640324712,
-      "learning_rate": 0.0001999990413581891,
-      "loss": 46.0,
-      "step": 8665
-    },
-    {
-      "epoch": 1.3955875840412255,
-      "grad_norm": 0.0030594351701438427,
-      "learning_rate": 0.00019999904113665343,
-      "loss": 46.0,
-      "step": 8666
-    },
-    {
-      "epoch": 1.395748621120013,
-      "grad_norm": 0.0012963516637682915,
-      "learning_rate": 0.00019999904091509214,
-      "loss": 46.0,
-      "step": 8667
-    },
-    {
-      "epoch": 1.3959096581988002,
-      "grad_norm": 0.004459632560610771,
-      "learning_rate": 0.00019999904069350528,
-      "loss": 46.0,
-      "step": 8668
-    },
-    {
-      "epoch": 1.3960706952775876,
-      "grad_norm": 0.00065183563856408,
-      "learning_rate": 0.0001999990404718928,
-      "loss": 46.0,
-      "step": 8669
-    },
-    {
-      "epoch": 1.396231732356375,
-      "grad_norm": 0.001076934626325965,
-      "learning_rate": 0.00019999904025025473,
-      "loss": 46.0,
-      "step": 8670
-    },
-    {
-      "epoch": 1.3963927694351623,
-      "grad_norm": 0.0009930869564414024,
-      "learning_rate": 0.00019999904002859109,
-      "loss": 46.0,
-      "step": 8671
-    },
-    {
-      "epoch": 1.3965538065139498,
-      "grad_norm": 0.002097767312079668,
-      "learning_rate": 0.00019999903980690185,
-      "loss": 46.0,
-      "step": 8672
-    },
-    {
-      "epoch": 1.3967148435927372,
-      "grad_norm": 0.004299834370613098,
-      "learning_rate": 0.000199999039585187,
-      "loss": 46.0,
-      "step": 8673
-    },
-    {
-      "epoch": 1.3968758806715247,
-      "grad_norm": 0.001463218592107296,
-      "learning_rate": 0.00019999903936344655,
-      "loss": 46.0,
-      "step": 8674
-    },
-    {
-      "epoch": 1.3970369177503121,
-      "grad_norm": 0.002759312279522419,
-      "learning_rate": 0.00019999903914168053,
-      "loss": 46.0,
-      "step": 8675
-    },
-    {
-      "epoch": 1.3971979548290994,
-      "grad_norm": 0.00230770418420434,
-      "learning_rate": 0.00019999903891988893,
-      "loss": 46.0,
-      "step": 8676
-    },
-    {
-      "epoch": 1.3973589919078868,
-      "grad_norm": 0.0014615310356020927,
-      "learning_rate": 0.00019999903869807168,
-      "loss": 46.0,
-      "step": 8677
-    },
-    {
-      "epoch": 1.397520028986674,
-      "grad_norm": 0.014054329134523869,
-      "learning_rate": 0.00019999903847622887,
-      "loss": 46.0,
-      "step": 8678
-    },
-    {
-      "epoch": 1.3976810660654615,
-      "grad_norm": 0.0041089532896876335,
-      "learning_rate": 0.00019999903825436045,
-      "loss": 46.0,
-      "step": 8679
-    },
-    {
-      "epoch": 1.397842103144249,
-      "grad_norm": 0.001594877801835537,
-      "learning_rate": 0.00019999903803246644,
-      "loss": 46.0,
-      "step": 8680
-    },
-    {
-      "epoch": 1.3980031402230364,
-      "grad_norm": 0.0009906088234856725,
-      "learning_rate": 0.00019999903781054684,
-      "loss": 46.0,
-      "step": 8681
-    },
-    {
-      "epoch": 1.3981641773018239,
-      "grad_norm": 0.004922871477901936,
-      "learning_rate": 0.00019999903758860166,
-      "loss": 46.0,
-      "step": 8682
-    },
-    {
-      "epoch": 1.398325214380611,
-      "grad_norm": 0.0011025634594261646,
-      "learning_rate": 0.00019999903736663086,
-      "loss": 46.0,
-      "step": 8683
-    },
-    {
-      "epoch": 1.3984862514593985,
-      "grad_norm": 0.0009672152809798717,
-      "learning_rate": 0.00019999903714463448,
-      "loss": 46.0,
-      "step": 8684
-    },
-    {
-      "epoch": 1.398647288538186,
-      "grad_norm": 0.00533898314461112,
-      "learning_rate": 0.00019999903692261248,
-      "loss": 46.0,
-      "step": 8685
-    },
-    {
-      "epoch": 1.3988083256169732,
-      "grad_norm": 0.0032699699513614178,
-      "learning_rate": 0.00019999903670056492,
-      "loss": 46.0,
-      "step": 8686
-    },
-    {
-      "epoch": 1.3989693626957607,
-      "grad_norm": 0.0008260086760856211,
-      "learning_rate": 0.00019999903647849174,
-      "loss": 46.0,
-      "step": 8687
-    },
-    {
-      "epoch": 1.3991303997745481,
-      "grad_norm": 0.0016012934502214193,
-      "learning_rate": 0.000199999036256393,
-      "loss": 46.0,
-      "step": 8688
-    },
-    {
-      "epoch": 1.3992914368533356,
-      "grad_norm": 0.0007186186849139631,
-      "learning_rate": 0.00019999903603426863,
-      "loss": 46.0,
-      "step": 8689
-    },
-    {
-      "epoch": 1.3994524739321228,
-      "grad_norm": 0.0034034280106425285,
-      "learning_rate": 0.0001999990358121187,
-      "loss": 46.0,
-      "step": 8690
-    },
-    {
-      "epoch": 1.3996135110109102,
-      "grad_norm": 0.002227758290246129,
-      "learning_rate": 0.00019999903558994312,
-      "loss": 46.0,
-      "step": 8691
-    },
-    {
-      "epoch": 1.3997745480896977,
-      "grad_norm": 0.004060195758938789,
-      "learning_rate": 0.00019999903536774199,
-      "loss": 46.0,
-      "step": 8692
-    },
-    {
-      "epoch": 1.399935585168485,
-      "grad_norm": 0.0006188918487168849,
-      "learning_rate": 0.00019999903514551526,
-      "loss": 46.0,
-      "step": 8693
-    },
-    {
-      "epoch": 1.4000966222472724,
-      "grad_norm": 0.0016157592181116343,
-      "learning_rate": 0.00019999903492326292,
-      "loss": 46.0,
-      "step": 8694
-    },
-    {
-      "epoch": 1.4002576593260598,
-      "grad_norm": 0.003066223580390215,
-      "learning_rate": 0.000199999034700985,
-      "loss": 46.0,
-      "step": 8695
-    },
-    {
-      "epoch": 1.4004186964048473,
-      "grad_norm": 0.0017708541126921773,
-      "learning_rate": 0.0001999990344786815,
-      "loss": 46.0,
-      "step": 8696
-    },
-    {
-      "epoch": 1.4005797334836347,
-      "grad_norm": 0.0019677416421473026,
-      "learning_rate": 0.0001999990342563524,
-      "loss": 46.0,
-      "step": 8697
-    },
-    {
-      "epoch": 1.400740770562422,
-      "grad_norm": 0.0005356779438443482,
-      "learning_rate": 0.00019999903403399767,
-      "loss": 46.0,
-      "step": 8698
-    },
-    {
-      "epoch": 1.4009018076412094,
-      "grad_norm": 0.0047472030855715275,
-      "learning_rate": 0.00019999903381161735,
-      "loss": 46.0,
-      "step": 8699
-    },
-    {
-      "epoch": 1.4010628447199966,
-      "grad_norm": 0.0035157930105924606,
-      "learning_rate": 0.00019999903358921146,
-      "loss": 46.0,
-      "step": 8700
-    },
-    {
-      "epoch": 1.401223881798784,
-      "grad_norm": 0.0008859345107339323,
-      "learning_rate": 0.00019999903336677995,
-      "loss": 46.0,
-      "step": 8701
-    },
-    {
-      "epoch": 1.4013849188775716,
-      "grad_norm": 0.001499322708696127,
-      "learning_rate": 0.0001999990331443229,
-      "loss": 46.0,
-      "step": 8702
-    },
-    {
-      "epoch": 1.401545955956359,
-      "grad_norm": 0.012457621283829212,
-      "learning_rate": 0.0001999990329218402,
-      "loss": 46.0,
-      "step": 8703
-    },
-    {
-      "epoch": 1.4017069930351465,
-      "grad_norm": 0.0012489799410104752,
-      "learning_rate": 0.00019999903269933192,
-      "loss": 46.0,
-      "step": 8704
-    },
-    {
-      "epoch": 1.4018680301139337,
-      "grad_norm": 0.0008658130536787212,
-      "learning_rate": 0.00019999903247679805,
-      "loss": 46.0,
-      "step": 8705
-    },
-    {
-      "epoch": 1.4020290671927211,
-      "grad_norm": 0.00033852504566311836,
-      "learning_rate": 0.00019999903225423858,
-      "loss": 46.0,
-      "step": 8706
-    },
-    {
-      "epoch": 1.4021901042715086,
-      "grad_norm": 0.0016483375802636147,
-      "learning_rate": 0.0001999990320316535,
-      "loss": 46.0,
-      "step": 8707
-    },
-    {
-      "epoch": 1.4023511413502958,
-      "grad_norm": 0.0017330390401184559,
-      "learning_rate": 0.00019999903180904286,
-      "loss": 46.0,
-      "step": 8708
-    },
-    {
-      "epoch": 1.4025121784290833,
-      "grad_norm": 0.0009625107049942017,
-      "learning_rate": 0.0001999990315864066,
-      "loss": 46.0,
-      "step": 8709
-    },
-    {
-      "epoch": 1.4026732155078707,
-      "grad_norm": 0.0009603157523088157,
-      "learning_rate": 0.00019999903136374477,
-      "loss": 46.0,
-      "step": 8710
-    },
-    {
-      "epoch": 1.4028342525866582,
-      "grad_norm": 0.0016231354093179107,
-      "learning_rate": 0.00019999903114105734,
-      "loss": 46.0,
-      "step": 8711
-    },
-    {
-      "epoch": 1.4029952896654454,
-      "grad_norm": 0.002856927691027522,
-      "learning_rate": 0.0001999990309183443,
-      "loss": 46.0,
-      "step": 8712
-    },
-    {
-      "epoch": 1.4031563267442329,
-      "grad_norm": 0.003310267347842455,
-      "learning_rate": 0.00019999903069560566,
-      "loss": 46.0,
-      "step": 8713
-    },
-    {
-      "epoch": 1.4033173638230203,
-      "grad_norm": 0.0003559023898560554,
-      "learning_rate": 0.00019999903047284145,
-      "loss": 46.0,
-      "step": 8714
-    },
-    {
-      "epoch": 1.4034784009018075,
-      "grad_norm": 0.0037952205166220665,
-      "learning_rate": 0.00019999903025005162,
-      "loss": 46.0,
-      "step": 8715
-    },
-    {
-      "epoch": 1.403639437980595,
-      "grad_norm": 0.004152096342295408,
-      "learning_rate": 0.00019999903002723623,
-      "loss": 46.0,
-      "step": 8716
-    },
-    {
-      "epoch": 1.4038004750593824,
-      "grad_norm": 0.00553463539108634,
-      "learning_rate": 0.00019999902980439522,
-      "loss": 46.0,
-      "step": 8717
-    },
-    {
-      "epoch": 1.4039615121381699,
-      "grad_norm": 0.001983468420803547,
-      "learning_rate": 0.0001999990295815286,
-      "loss": 46.0,
-      "step": 8718
-    },
-    {
-      "epoch": 1.4041225492169573,
-      "grad_norm": 0.0009469576179981232,
-      "learning_rate": 0.00019999902935863642,
-      "loss": 46.0,
-      "step": 8719
-    },
-    {
-      "epoch": 1.4042835862957446,
-      "grad_norm": 0.0015824816655367613,
-      "learning_rate": 0.00019999902913571865,
-      "loss": 46.0,
-      "step": 8720
-    },
-    {
-      "epoch": 1.404444623374532,
-      "grad_norm": 0.004357549361884594,
-      "learning_rate": 0.00019999902891277524,
-      "loss": 46.0,
-      "step": 8721
-    },
-    {
-      "epoch": 1.4046056604533192,
-      "grad_norm": 0.002079347614198923,
-      "learning_rate": 0.00019999902868980628,
-      "loss": 46.0,
-      "step": 8722
-    },
-    {
-      "epoch": 1.4047666975321067,
-      "grad_norm": 0.0003655439068097621,
-      "learning_rate": 0.0001999990284668117,
-      "loss": 46.0,
-      "step": 8723
-    },
-    {
-      "epoch": 1.4049277346108942,
-      "grad_norm": 0.000651561887934804,
-      "learning_rate": 0.00019999902824379152,
-      "loss": 46.0,
-      "step": 8724
-    },
-    {
-      "epoch": 1.4050887716896816,
-      "grad_norm": 0.0011075891088694334,
-      "learning_rate": 0.00019999902802074574,
-      "loss": 46.0,
-      "step": 8725
-    },
-    {
-      "epoch": 1.405249808768469,
-      "grad_norm": 0.0027016333770006895,
-      "learning_rate": 0.0001999990277976744,
-      "loss": 46.0,
-      "step": 8726
-    },
-    {
-      "epoch": 1.4054108458472563,
-      "grad_norm": 0.0016776265110820532,
-      "learning_rate": 0.00019999902757457744,
-      "loss": 46.0,
-      "step": 8727
-    },
-    {
-      "epoch": 1.4055718829260437,
-      "grad_norm": 0.0007533309981226921,
-      "learning_rate": 0.0001999990273514549,
-      "loss": 46.0,
-      "step": 8728
-    },
-    {
-      "epoch": 1.4057329200048312,
-      "grad_norm": 0.0012430923525243998,
-      "learning_rate": 0.00019999902712830676,
-      "loss": 46.0,
-      "step": 8729
-    },
-    {
-      "epoch": 1.4058939570836184,
-      "grad_norm": 0.0019619695376604795,
-      "learning_rate": 0.000199999026905133,
-      "loss": 46.0,
-      "step": 8730
-    },
-    {
-      "epoch": 1.4060549941624059,
-      "grad_norm": 0.0006122136837802827,
-      "learning_rate": 0.00019999902668193367,
-      "loss": 46.0,
-      "step": 8731
-    },
-    {
-      "epoch": 1.4062160312411933,
-      "grad_norm": 0.0037623592652380466,
-      "learning_rate": 0.00019999902645870875,
-      "loss": 46.0,
-      "step": 8732
-    },
-    {
-      "epoch": 1.4063770683199808,
-      "grad_norm": 0.0016203204868361354,
-      "learning_rate": 0.00019999902623545824,
-      "loss": 46.0,
-      "step": 8733
-    },
-    {
-      "epoch": 1.406538105398768,
-      "grad_norm": 0.004895881284028292,
-      "learning_rate": 0.00019999902601218212,
-      "loss": 46.0,
-      "step": 8734
-    },
-    {
-      "epoch": 1.4066991424775555,
-      "grad_norm": 0.0008878657827153802,
-      "learning_rate": 0.0001999990257888804,
-      "loss": 46.0,
-      "step": 8735
-    },
-    {
-      "epoch": 1.406860179556343,
-      "grad_norm": 0.0007570901652798057,
-      "learning_rate": 0.00019999902556555309,
-      "loss": 46.0,
-      "step": 8736
-    },
-    {
-      "epoch": 1.4070212166351301,
-      "grad_norm": 0.006074225064367056,
-      "learning_rate": 0.0001999990253422002,
-      "loss": 46.0,
-      "step": 8737
-    },
-    {
-      "epoch": 1.4071822537139176,
-      "grad_norm": 0.009877574630081654,
-      "learning_rate": 0.0001999990251188217,
-      "loss": 46.0,
-      "step": 8738
-    },
-    {
-      "epoch": 1.407343290792705,
-      "grad_norm": 0.0013487032847478986,
-      "learning_rate": 0.00019999902489541764,
-      "loss": 46.0,
-      "step": 8739
-    },
-    {
-      "epoch": 1.4075043278714925,
-      "grad_norm": 0.0007247634348459542,
-      "learning_rate": 0.00019999902467198794,
-      "loss": 46.0,
-      "step": 8740
-    },
-    {
-      "epoch": 1.40766536495028,
-      "grad_norm": 0.002311364747583866,
-      "learning_rate": 0.00019999902444853266,
-      "loss": 46.0,
-      "step": 8741
-    },
-    {
-      "epoch": 1.4078264020290672,
-      "grad_norm": 0.001140328124165535,
-      "learning_rate": 0.00019999902422505178,
-      "loss": 46.0,
-      "step": 8742
-    },
-    {
-      "epoch": 1.4079874391078546,
-      "grad_norm": 0.0017533078789710999,
-      "learning_rate": 0.00019999902400154532,
-      "loss": 46.0,
-      "step": 8743
-    },
-    {
-      "epoch": 1.4081484761866419,
-      "grad_norm": 0.0011203738395124674,
-      "learning_rate": 0.00019999902377801324,
-      "loss": 46.0,
-      "step": 8744
-    },
-    {
-      "epoch": 1.4083095132654293,
-      "grad_norm": 0.0050362516194581985,
-      "learning_rate": 0.0001999990235544556,
-      "loss": 46.0,
-      "step": 8745
-    },
-    {
-      "epoch": 1.4084705503442168,
-      "grad_norm": 0.008753200061619282,
-      "learning_rate": 0.00019999902333087236,
-      "loss": 46.0,
-      "step": 8746
-    },
-    {
-      "epoch": 1.4086315874230042,
-      "grad_norm": 0.0038523238617926836,
-      "learning_rate": 0.00019999902310726352,
-      "loss": 46.0,
-      "step": 8747
-    },
-    {
-      "epoch": 1.4087926245017917,
-      "grad_norm": 0.010478362441062927,
-      "learning_rate": 0.0001999990228836291,
-      "loss": 46.0,
-      "step": 8748
-    },
-    {
-      "epoch": 1.4089536615805789,
-      "grad_norm": 0.002176894573494792,
-      "learning_rate": 0.00019999902265996903,
-      "loss": 46.0,
-      "step": 8749
-    },
-    {
-      "epoch": 1.4091146986593663,
-      "grad_norm": 0.002317194826900959,
-      "learning_rate": 0.0001999990224362834,
-      "loss": 46.0,
-      "step": 8750
-    },
-    {
-      "epoch": 1.4092757357381538,
-      "grad_norm": 0.0024861914571374655,
-      "learning_rate": 0.0001999990222125722,
-      "loss": 46.0,
-      "step": 8751
-    },
-    {
-      "epoch": 1.409436772816941,
-      "grad_norm": 0.0015292122261598706,
-      "learning_rate": 0.0001999990219888354,
-      "loss": 46.0,
-      "step": 8752
-    },
-    {
-      "epoch": 1.4095978098957285,
-      "grad_norm": 0.0011055519571527839,
-      "learning_rate": 0.00019999902176507295,
-      "loss": 46.0,
-      "step": 8753
-    },
-    {
-      "epoch": 1.409758846974516,
-      "grad_norm": 0.0009253533207811415,
-      "learning_rate": 0.00019999902154128494,
-      "loss": 46.0,
-      "step": 8754
-    },
-    {
-      "epoch": 1.4099198840533034,
-      "grad_norm": 0.0015908654313534498,
-      "learning_rate": 0.00019999902131747136,
-      "loss": 46.0,
-      "step": 8755
-    },
-    {
-      "epoch": 1.4100809211320906,
-      "grad_norm": 0.004116092808544636,
-      "learning_rate": 0.00019999902109363212,
-      "loss": 46.0,
-      "step": 8756
-    },
-    {
-      "epoch": 1.410241958210878,
-      "grad_norm": 0.001024664961732924,
-      "learning_rate": 0.00019999902086976733,
-      "loss": 46.0,
-      "step": 8757
-    },
-    {
-      "epoch": 1.4104029952896655,
-      "grad_norm": 0.0037672175094485283,
-      "learning_rate": 0.00019999902064587696,
-      "loss": 46.0,
-      "step": 8758
-    },
-    {
-      "epoch": 1.4105640323684527,
-      "grad_norm": 0.0013191432226449251,
-      "learning_rate": 0.000199999020421961,
-      "loss": 46.0,
-      "step": 8759
-    },
-    {
-      "epoch": 1.4107250694472402,
-      "grad_norm": 0.0021485602483153343,
-      "learning_rate": 0.0001999990201980194,
-      "loss": 46.0,
-      "step": 8760
-    },
-    {
-      "epoch": 1.4108861065260276,
-      "grad_norm": 0.0012577869929373264,
-      "learning_rate": 0.00019999901997405225,
-      "loss": 46.0,
-      "step": 8761
-    },
-    {
-      "epoch": 1.411047143604815,
-      "grad_norm": 0.0018599767936393619,
-      "learning_rate": 0.0001999990197500595,
-      "loss": 46.0,
-      "step": 8762
-    },
-    {
-      "epoch": 1.4112081806836023,
-      "grad_norm": 0.0014334272127598524,
-      "learning_rate": 0.00019999901952604112,
-      "loss": 46.0,
-      "step": 8763
-    },
-    {
-      "epoch": 1.4113692177623898,
-      "grad_norm": 0.001310729538090527,
-      "learning_rate": 0.00019999901930199717,
-      "loss": 46.0,
-      "step": 8764
-    },
-    {
-      "epoch": 1.4115302548411772,
-      "grad_norm": 0.0015597993042320013,
-      "learning_rate": 0.00019999901907792763,
-      "loss": 46.0,
-      "step": 8765
-    },
-    {
-      "epoch": 1.4116912919199645,
-      "grad_norm": 0.000865362526383251,
-      "learning_rate": 0.00019999901885383247,
-      "loss": 46.0,
-      "step": 8766
-    },
-    {
-      "epoch": 1.411852328998752,
-      "grad_norm": 0.0014334465377032757,
-      "learning_rate": 0.00019999901862971173,
-      "loss": 46.0,
-      "step": 8767
-    },
-    {
-      "epoch": 1.4120133660775394,
-      "grad_norm": 0.0009138078312389553,
-      "learning_rate": 0.0001999990184055654,
-      "loss": 46.0,
-      "step": 8768
-    },
-    {
-      "epoch": 1.4121744031563268,
-      "grad_norm": 0.0018936212873086333,
-      "learning_rate": 0.00019999901818139345,
-      "loss": 46.0,
-      "step": 8769
-    },
-    {
-      "epoch": 1.4123354402351143,
-      "grad_norm": 0.0014860385563224554,
-      "learning_rate": 0.00019999901795719595,
-      "loss": 46.0,
-      "step": 8770
-    },
-    {
-      "epoch": 1.4124964773139015,
-      "grad_norm": 0.0005086000310257077,
-      "learning_rate": 0.00019999901773297283,
-      "loss": 46.0,
-      "step": 8771
-    },
-    {
-      "epoch": 1.412657514392689,
-      "grad_norm": 0.002495589666068554,
-      "learning_rate": 0.00019999901750872412,
-      "loss": 46.0,
-      "step": 8772
-    },
-    {
-      "epoch": 1.4128185514714762,
-      "grad_norm": 0.005120695102959871,
-      "learning_rate": 0.00019999901728444983,
-      "loss": 46.0,
-      "step": 8773
-    },
-    {
-      "epoch": 1.4129795885502636,
-      "grad_norm": 0.0004170707252342254,
-      "learning_rate": 0.00019999901706014992,
-      "loss": 46.0,
-      "step": 8774
-    },
-    {
-      "epoch": 1.413140625629051,
-      "grad_norm": 0.0021690381690859795,
-      "learning_rate": 0.00019999901683582442,
-      "loss": 46.0,
-      "step": 8775
-    },
-    {
-      "epoch": 1.4133016627078385,
-      "grad_norm": 0.003178894752636552,
-      "learning_rate": 0.00019999901661147334,
-      "loss": 46.0,
-      "step": 8776
-    },
-    {
-      "epoch": 1.413462699786626,
-      "grad_norm": 0.005027687177062035,
-      "learning_rate": 0.00019999901638709667,
-      "loss": 46.0,
-      "step": 8777
-    },
-    {
-      "epoch": 1.4136237368654132,
-      "grad_norm": 0.0009337655501440167,
-      "learning_rate": 0.00019999901616269439,
-      "loss": 46.0,
-      "step": 8778
-    },
-    {
-      "epoch": 1.4137847739442007,
-      "grad_norm": 0.0008403321262449026,
-      "learning_rate": 0.0001999990159382665,
-      "loss": 46.0,
-      "step": 8779
-    },
-    {
-      "epoch": 1.413945811022988,
-      "grad_norm": 0.0018781916005536914,
-      "learning_rate": 0.00019999901571381303,
-      "loss": 46.0,
-      "step": 8780
-    },
-    {
-      "epoch": 1.4141068481017753,
-      "grad_norm": 0.004008853808045387,
-      "learning_rate": 0.00019999901548933398,
-      "loss": 46.0,
-      "step": 8781
-    },
-    {
-      "epoch": 1.4142678851805628,
-      "grad_norm": 0.0013632572954520583,
-      "learning_rate": 0.00019999901526482932,
-      "loss": 46.0,
-      "step": 8782
-    },
-    {
-      "epoch": 1.4144289222593502,
-      "grad_norm": 0.0004339565057307482,
-      "learning_rate": 0.00019999901504029905,
-      "loss": 46.0,
-      "step": 8783
-    },
-    {
-      "epoch": 1.4145899593381377,
-      "grad_norm": 0.0027160339523106813,
-      "learning_rate": 0.00019999901481574322,
-      "loss": 46.0,
-      "step": 8784
-    },
-    {
-      "epoch": 1.414750996416925,
-      "grad_norm": 0.003067605197429657,
-      "learning_rate": 0.0001999990145911618,
-      "loss": 46.0,
-      "step": 8785
-    },
-    {
-      "epoch": 1.4149120334957124,
-      "grad_norm": 0.0035614839289337397,
-      "learning_rate": 0.00019999901436655476,
-      "loss": 46.0,
-      "step": 8786
-    },
-    {
-      "epoch": 1.4150730705744998,
-      "grad_norm": 0.0018429290503263474,
-      "learning_rate": 0.00019999901414192214,
-      "loss": 46.0,
-      "step": 8787
-    },
-    {
-      "epoch": 1.415234107653287,
-      "grad_norm": 0.000458260445157066,
-      "learning_rate": 0.00019999901391726387,
-      "loss": 46.0,
-      "step": 8788
-    },
-    {
-      "epoch": 1.4153951447320745,
-      "grad_norm": 0.006405094172805548,
-      "learning_rate": 0.00019999901369258008,
-      "loss": 46.0,
-      "step": 8789
-    },
-    {
-      "epoch": 1.415556181810862,
-      "grad_norm": 0.0011929180473089218,
-      "learning_rate": 0.00019999901346787064,
-      "loss": 46.0,
-      "step": 8790
-    },
-    {
-      "epoch": 1.4157172188896494,
-      "grad_norm": 0.0028130868449807167,
-      "learning_rate": 0.00019999901324313567,
-      "loss": 46.0,
-      "step": 8791
-    },
-    {
-      "epoch": 1.4158782559684369,
-      "grad_norm": 0.0011518935207277536,
-      "learning_rate": 0.00019999901301837505,
-      "loss": 46.0,
-      "step": 8792
-    },
-    {
-      "epoch": 1.416039293047224,
-      "grad_norm": 0.0036750249564647675,
-      "learning_rate": 0.00019999901279358885,
-      "loss": 46.0,
-      "step": 8793
-    },
-    {
-      "epoch": 1.4162003301260115,
-      "grad_norm": 0.0013662953861057758,
-      "learning_rate": 0.00019999901256877704,
-      "loss": 46.0,
-      "step": 8794
-    },
-    {
-      "epoch": 1.4163613672047988,
-      "grad_norm": 0.0008359162020497024,
-      "learning_rate": 0.00019999901234393964,
-      "loss": 46.0,
-      "step": 8795
-    },
-    {
-      "epoch": 1.4165224042835862,
-      "grad_norm": 0.0008566214819438756,
-      "learning_rate": 0.00019999901211907668,
-      "loss": 46.0,
-      "step": 8796
-    },
-    {
-      "epoch": 1.4166834413623737,
-      "grad_norm": 0.0021488158963620663,
-      "learning_rate": 0.0001999990118941881,
-      "loss": 46.0,
-      "step": 8797
-    },
-    {
-      "epoch": 1.4168444784411611,
-      "grad_norm": 0.002395689720287919,
-      "learning_rate": 0.00019999901166927394,
-      "loss": 46.0,
-      "step": 8798
-    },
-    {
-      "epoch": 1.4170055155199486,
-      "grad_norm": 0.006017365492880344,
-      "learning_rate": 0.00019999901144433416,
-      "loss": 46.0,
-      "step": 8799
-    },
-    {
-      "epoch": 1.4171665525987358,
-      "grad_norm": 0.005037990864366293,
-      "learning_rate": 0.00019999901121936883,
-      "loss": 46.0,
-      "step": 8800
-    },
-    {
-      "epoch": 1.4173275896775233,
-      "grad_norm": 0.0011998401023447514,
-      "learning_rate": 0.00019999901099437785,
-      "loss": 46.0,
-      "step": 8801
-    },
-    {
-      "epoch": 1.4174886267563107,
-      "grad_norm": 0.0010431613773107529,
-      "learning_rate": 0.0001999990107693613,
-      "loss": 46.0,
-      "step": 8802
-    },
-    {
-      "epoch": 1.417649663835098,
-      "grad_norm": 0.0011488142190501094,
-      "learning_rate": 0.00019999901054431916,
-      "loss": 46.0,
-      "step": 8803
-    },
-    {
-      "epoch": 1.4178107009138854,
-      "grad_norm": 0.0015358793316408992,
-      "learning_rate": 0.00019999901031925142,
-      "loss": 46.0,
-      "step": 8804
-    },
-    {
-      "epoch": 1.4179717379926728,
-      "grad_norm": 0.00228817667812109,
-      "learning_rate": 0.0001999990100941581,
-      "loss": 46.0,
-      "step": 8805
-    },
-    {
-      "epoch": 1.4181327750714603,
-      "grad_norm": 0.002047525718808174,
-      "learning_rate": 0.00019999900986903917,
-      "loss": 46.0,
-      "step": 8806
-    },
-    {
-      "epoch": 1.4182938121502475,
-      "grad_norm": 0.0016591690946370363,
-      "learning_rate": 0.00019999900964389464,
-      "loss": 46.0,
-      "step": 8807
-    },
-    {
-      "epoch": 1.418454849229035,
-      "grad_norm": 0.0011893196497112513,
-      "learning_rate": 0.0001999990094187245,
-      "loss": 46.0,
-      "step": 8808
-    },
-    {
-      "epoch": 1.4186158863078224,
-      "grad_norm": 0.0021540282759815454,
-      "learning_rate": 0.0001999990091935288,
-      "loss": 46.0,
-      "step": 8809
-    },
-    {
-      "epoch": 1.4187769233866097,
-      "grad_norm": 0.0015994729474186897,
-      "learning_rate": 0.0001999990089683075,
-      "loss": 46.0,
-      "step": 8810
-    },
-    {
-      "epoch": 1.418937960465397,
-      "grad_norm": 0.0015133037231862545,
-      "learning_rate": 0.00019999900874306058,
-      "loss": 46.0,
-      "step": 8811
-    },
-    {
-      "epoch": 1.4190989975441846,
-      "grad_norm": 0.000605659675784409,
-      "learning_rate": 0.0001999990085177881,
-      "loss": 46.0,
-      "step": 8812
-    },
-    {
-      "epoch": 1.419260034622972,
-      "grad_norm": 0.005168214440345764,
-      "learning_rate": 0.00019999900829249003,
-      "loss": 46.0,
-      "step": 8813
-    },
-    {
-      "epoch": 1.4194210717017595,
-      "grad_norm": 0.0012356200022622943,
-      "learning_rate": 0.00019999900806716634,
-      "loss": 46.0,
-      "step": 8814
-    },
-    {
-      "epoch": 1.4195821087805467,
-      "grad_norm": 0.002502727322280407,
-      "learning_rate": 0.00019999900784181706,
-      "loss": 46.0,
-      "step": 8815
-    },
-    {
-      "epoch": 1.4197431458593341,
-      "grad_norm": 0.005924689583480358,
-      "learning_rate": 0.0001999990076164422,
-      "loss": 46.0,
-      "step": 8816
-    },
-    {
-      "epoch": 1.4199041829381214,
-      "grad_norm": 0.0018171198898926377,
-      "learning_rate": 0.0001999990073910417,
-      "loss": 46.0,
-      "step": 8817
-    },
-    {
-      "epoch": 1.4200652200169088,
-      "grad_norm": 0.003842373611405492,
-      "learning_rate": 0.00019999900716561567,
-      "loss": 46.0,
-      "step": 8818
-    },
-    {
-      "epoch": 1.4202262570956963,
-      "grad_norm": 0.0007104586693458259,
-      "learning_rate": 0.000199999006940164,
-      "loss": 46.0,
-      "step": 8819
-    },
-    {
-      "epoch": 1.4203872941744837,
-      "grad_norm": 0.004088105633854866,
-      "learning_rate": 0.00019999900671468674,
-      "loss": 46.0,
-      "step": 8820
-    },
-    {
-      "epoch": 1.4205483312532712,
-      "grad_norm": 0.0031499583274126053,
-      "learning_rate": 0.0001999990064891839,
-      "loss": 46.0,
-      "step": 8821
-    },
-    {
-      "epoch": 1.4207093683320584,
-      "grad_norm": 0.001896240166388452,
-      "learning_rate": 0.00019999900626365547,
-      "loss": 46.0,
-      "step": 8822
-    },
-    {
-      "epoch": 1.4208704054108459,
-      "grad_norm": 0.0025624840054661036,
-      "learning_rate": 0.0001999990060381014,
-      "loss": 46.0,
-      "step": 8823
-    },
-    {
-      "epoch": 1.4210314424896333,
-      "grad_norm": 0.002257821848616004,
-      "learning_rate": 0.0001999990058125218,
-      "loss": 46.0,
-      "step": 8824
-    },
-    {
-      "epoch": 1.4211924795684205,
-      "grad_norm": 0.0005156651604920626,
-      "learning_rate": 0.00019999900558691656,
-      "loss": 46.0,
-      "step": 8825
-    },
-    {
-      "epoch": 1.421353516647208,
-      "grad_norm": 0.0013510119169950485,
-      "learning_rate": 0.00019999900536128574,
-      "loss": 46.0,
-      "step": 8826
-    },
-    {
-      "epoch": 1.4215145537259954,
-      "grad_norm": 0.004422096535563469,
-      "learning_rate": 0.00019999900513562933,
-      "loss": 46.0,
-      "step": 8827
-    },
-    {
-      "epoch": 1.421675590804783,
-      "grad_norm": 0.0021187118254601955,
-      "learning_rate": 0.0001999990049099473,
-      "loss": 46.0,
-      "step": 8828
-    },
-    {
-      "epoch": 1.4218366278835701,
-      "grad_norm": 0.0037883142940700054,
-      "learning_rate": 0.0001999990046842397,
-      "loss": 46.0,
-      "step": 8829
-    },
-    {
-      "epoch": 1.4219976649623576,
-      "grad_norm": 0.0020474051125347614,
-      "learning_rate": 0.0001999990044585065,
-      "loss": 46.0,
-      "step": 8830
-    },
-    {
-      "epoch": 1.422158702041145,
-      "grad_norm": 0.0012470899382606149,
-      "learning_rate": 0.00019999900423274772,
-      "loss": 46.0,
-      "step": 8831
-    },
-    {
-      "epoch": 1.4223197391199323,
-      "grad_norm": 0.0031712038908153772,
-      "learning_rate": 0.00019999900400696335,
-      "loss": 46.0,
-      "step": 8832
-    },
-    {
-      "epoch": 1.4224807761987197,
-      "grad_norm": 0.007882462814450264,
-      "learning_rate": 0.00019999900378115334,
-      "loss": 46.0,
-      "step": 8833
-    },
-    {
-      "epoch": 1.4226418132775072,
-      "grad_norm": 0.0020494707860052586,
-      "learning_rate": 0.00019999900355531776,
-      "loss": 46.0,
-      "step": 8834
-    },
-    {
-      "epoch": 1.4228028503562946,
-      "grad_norm": 0.0005583789898082614,
-      "learning_rate": 0.0001999990033294566,
-      "loss": 46.0,
-      "step": 8835
-    },
-    {
-      "epoch": 1.422963887435082,
-      "grad_norm": 0.0033763465471565723,
-      "learning_rate": 0.00019999900310356986,
-      "loss": 46.0,
-      "step": 8836
-    },
-    {
-      "epoch": 1.4231249245138693,
-      "grad_norm": 0.0016085447277873755,
-      "learning_rate": 0.00019999900287765747,
-      "loss": 46.0,
-      "step": 8837
-    },
-    {
-      "epoch": 1.4232859615926567,
-      "grad_norm": 0.0004522383096627891,
-      "learning_rate": 0.00019999900265171952,
-      "loss": 46.0,
-      "step": 8838
-    },
-    {
-      "epoch": 1.423446998671444,
-      "grad_norm": 0.0052393963560462,
-      "learning_rate": 0.00019999900242575596,
-      "loss": 46.0,
-      "step": 8839
-    },
-    {
-      "epoch": 1.4236080357502314,
-      "grad_norm": 0.001791846239939332,
-      "learning_rate": 0.0001999990021997668,
-      "loss": 46.0,
-      "step": 8840
-    },
-    {
-      "epoch": 1.4237690728290189,
-      "grad_norm": 0.0007898471667431295,
-      "learning_rate": 0.0001999990019737521,
-      "loss": 46.0,
-      "step": 8841
-    },
-    {
-      "epoch": 1.4239301099078063,
-      "grad_norm": 0.0008404402178712189,
-      "learning_rate": 0.00019999900174771175,
-      "loss": 46.0,
-      "step": 8842
-    },
-    {
-      "epoch": 1.4240911469865938,
-      "grad_norm": 0.000669444736558944,
-      "learning_rate": 0.00019999900152164584,
-      "loss": 46.0,
-      "step": 8843
-    },
-    {
-      "epoch": 1.424252184065381,
-      "grad_norm": 0.0038019234780222178,
-      "learning_rate": 0.0001999990012955543,
-      "loss": 46.0,
-      "step": 8844
-    },
-    {
-      "epoch": 1.4244132211441685,
-      "grad_norm": 0.001707833376713097,
-      "learning_rate": 0.00019999900106943718,
-      "loss": 46.0,
-      "step": 8845
-    },
-    {
-      "epoch": 1.424574258222956,
-      "grad_norm": 0.0011256089201197028,
-      "learning_rate": 0.00019999900084329445,
-      "loss": 46.0,
-      "step": 8846
-    },
-    {
-      "epoch": 1.4247352953017431,
-      "grad_norm": 0.0015628539258614182,
-      "learning_rate": 0.00019999900061712616,
-      "loss": 46.0,
-      "step": 8847
-    },
-    {
-      "epoch": 1.4248963323805306,
-      "grad_norm": 0.0013905201340094209,
-      "learning_rate": 0.00019999900039093226,
-      "loss": 46.0,
-      "step": 8848
-    },
-    {
-      "epoch": 1.425057369459318,
-      "grad_norm": 0.0003331666230224073,
-      "learning_rate": 0.00019999900016471277,
-      "loss": 46.0,
-      "step": 8849
-    },
-    {
-      "epoch": 1.4252184065381055,
-      "grad_norm": 0.0056498427875339985,
-      "learning_rate": 0.00019999899993846767,
-      "loss": 46.0,
-      "step": 8850
-    },
-    {
-      "epoch": 1.4253794436168927,
-      "grad_norm": 0.0026023618411272764,
-      "learning_rate": 0.00019999899971219698,
-      "loss": 46.0,
-      "step": 8851
-    },
-    {
-      "epoch": 1.4255404806956802,
-      "grad_norm": 0.0017910264432430267,
-      "learning_rate": 0.0001999989994859007,
-      "loss": 46.0,
-      "step": 8852
-    },
-    {
-      "epoch": 1.4257015177744676,
-      "grad_norm": 0.004014239646494389,
-      "learning_rate": 0.00019999899925957884,
-      "loss": 46.0,
-      "step": 8853
-    },
-    {
-      "epoch": 1.4258625548532549,
-      "grad_norm": 0.0017814973834902048,
-      "learning_rate": 0.00019999899903323136,
-      "loss": 46.0,
-      "step": 8854
-    },
-    {
-      "epoch": 1.4260235919320423,
-      "grad_norm": 0.0007097472553141415,
-      "learning_rate": 0.00019999899880685827,
-      "loss": 46.0,
-      "step": 8855
-    },
-    {
-      "epoch": 1.4261846290108298,
-      "grad_norm": 0.006036133039742708,
-      "learning_rate": 0.00019999899858045965,
-      "loss": 46.0,
-      "step": 8856
-    },
-    {
-      "epoch": 1.4263456660896172,
-      "grad_norm": 0.0005275346338748932,
-      "learning_rate": 0.0001999989983540354,
-      "loss": 46.0,
-      "step": 8857
-    },
-    {
-      "epoch": 1.4265067031684044,
-      "grad_norm": 0.0011749006807804108,
-      "learning_rate": 0.00019999899812758553,
-      "loss": 46.0,
-      "step": 8858
-    },
-    {
-      "epoch": 1.426667740247192,
-      "grad_norm": 0.0012769057648256421,
-      "learning_rate": 0.00019999899790111008,
-      "loss": 46.0,
-      "step": 8859
-    },
-    {
-      "epoch": 1.4268287773259793,
-      "grad_norm": 0.001751145115122199,
-      "learning_rate": 0.00019999899767460906,
-      "loss": 46.0,
-      "step": 8860
-    },
-    {
-      "epoch": 1.4269898144047666,
-      "grad_norm": 0.0005043031414970756,
-      "learning_rate": 0.00019999899744808244,
-      "loss": 46.0,
-      "step": 8861
-    },
-    {
-      "epoch": 1.427150851483554,
-      "grad_norm": 0.007926051504909992,
-      "learning_rate": 0.00019999899722153018,
-      "loss": 46.0,
-      "step": 8862
-    },
-    {
-      "epoch": 1.4273118885623415,
-      "grad_norm": 0.0040412782691419125,
-      "learning_rate": 0.00019999899699495237,
-      "loss": 46.0,
-      "step": 8863
-    },
-    {
-      "epoch": 1.427472925641129,
-      "grad_norm": 0.0006264648400247097,
-      "learning_rate": 0.00019999899676834896,
-      "loss": 46.0,
-      "step": 8864
-    },
-    {
-      "epoch": 1.4276339627199164,
-      "grad_norm": 0.0008083584834821522,
-      "learning_rate": 0.00019999899654171994,
-      "loss": 46.0,
-      "step": 8865
-    },
-    {
-      "epoch": 1.4277949997987036,
-      "grad_norm": 0.006049438379704952,
-      "learning_rate": 0.00019999899631506537,
-      "loss": 46.0,
-      "step": 8866
-    },
-    {
-      "epoch": 1.427956036877491,
-      "grad_norm": 0.0026091355830430984,
-      "learning_rate": 0.00019999899608838515,
-      "loss": 46.0,
-      "step": 8867
-    },
-    {
-      "epoch": 1.4281170739562785,
-      "grad_norm": 0.0013568110298365355,
-      "learning_rate": 0.00019999899586167937,
-      "loss": 46.0,
-      "step": 8868
-    },
-    {
-      "epoch": 1.4282781110350657,
-      "grad_norm": 0.001467092544771731,
-      "learning_rate": 0.00019999899563494797,
-      "loss": 46.0,
-      "step": 8869
-    },
-    {
-      "epoch": 1.4284391481138532,
-      "grad_norm": 0.00386932329274714,
-      "learning_rate": 0.000199998995408191,
-      "loss": 46.0,
-      "step": 8870
-    },
-    {
-      "epoch": 1.4286001851926406,
-      "grad_norm": 0.008279883302748203,
-      "learning_rate": 0.00019999899518140842,
-      "loss": 46.0,
-      "step": 8871
-    },
-    {
-      "epoch": 1.428761222271428,
-      "grad_norm": 0.0006841786671429873,
-      "learning_rate": 0.00019999899495460024,
-      "loss": 46.0,
-      "step": 8872
-    },
-    {
-      "epoch": 1.4289222593502153,
-      "grad_norm": 0.0033430345356464386,
-      "learning_rate": 0.00019999899472776647,
-      "loss": 46.0,
-      "step": 8873
-    },
-    {
-      "epoch": 1.4290832964290028,
-      "grad_norm": 0.0027362238615751266,
-      "learning_rate": 0.00019999899450090711,
-      "loss": 46.0,
-      "step": 8874
-    },
-    {
-      "epoch": 1.4292443335077902,
-      "grad_norm": 0.001019414165057242,
-      "learning_rate": 0.00019999899427402214,
-      "loss": 46.0,
-      "step": 8875
-    },
-    {
-      "epoch": 1.4294053705865775,
-      "grad_norm": 0.0004085192340426147,
-      "learning_rate": 0.0001999989940471116,
-      "loss": 46.0,
-      "step": 8876
-    },
-    {
-      "epoch": 1.429566407665365,
-      "grad_norm": 0.002970003057271242,
-      "learning_rate": 0.00019999899382017547,
-      "loss": 46.0,
-      "step": 8877
-    },
-    {
-      "epoch": 1.4297274447441524,
-      "grad_norm": 0.0005938848480582237,
-      "learning_rate": 0.0001999989935932137,
-      "loss": 46.0,
-      "step": 8878
-    },
-    {
-      "epoch": 1.4298884818229398,
-      "grad_norm": 0.0013657067902386189,
-      "learning_rate": 0.00019999899336622639,
-      "loss": 46.0,
-      "step": 8879
-    },
-    {
-      "epoch": 1.430049518901727,
-      "grad_norm": 0.001114029437303543,
-      "learning_rate": 0.00019999899313921345,
-      "loss": 46.0,
-      "step": 8880
-    },
-    {
-      "epoch": 1.4302105559805145,
-      "grad_norm": 0.002482883632183075,
-      "learning_rate": 0.00019999899291217493,
-      "loss": 46.0,
-      "step": 8881
-    },
-    {
-      "epoch": 1.430371593059302,
-      "grad_norm": 0.0020997305400669575,
-      "learning_rate": 0.00019999899268511082,
-      "loss": 46.0,
-      "step": 8882
-    },
-    {
-      "epoch": 1.4305326301380892,
-      "grad_norm": 0.00156143878120929,
-      "learning_rate": 0.0001999989924580211,
-      "loss": 46.0,
-      "step": 8883
-    },
-    {
-      "epoch": 1.4306936672168766,
-      "grad_norm": 0.005047987215220928,
-      "learning_rate": 0.0001999989922309058,
-      "loss": 46.0,
-      "step": 8884
-    },
-    {
-      "epoch": 1.430854704295664,
-      "grad_norm": 0.0011467499425634742,
-      "learning_rate": 0.0001999989920037649,
-      "loss": 46.0,
-      "step": 8885
-    },
-    {
-      "epoch": 1.4310157413744515,
-      "grad_norm": 0.004054028075188398,
-      "learning_rate": 0.0001999989917765984,
-      "loss": 46.0,
-      "step": 8886
-    },
-    {
-      "epoch": 1.431176778453239,
-      "grad_norm": 0.00480400025844574,
-      "learning_rate": 0.0001999989915494063,
-      "loss": 46.0,
-      "step": 8887
-    },
-    {
-      "epoch": 1.4313378155320262,
-      "grad_norm": 0.0008795669418759644,
-      "learning_rate": 0.0001999989913221886,
-      "loss": 46.0,
-      "step": 8888
-    },
-    {
-      "epoch": 1.4314988526108137,
-      "grad_norm": 0.0016269002808257937,
-      "learning_rate": 0.00019999899109494532,
-      "loss": 46.0,
-      "step": 8889
-    },
-    {
-      "epoch": 1.431659889689601,
-      "grad_norm": 0.0030724999960511923,
-      "learning_rate": 0.00019999899086767646,
-      "loss": 46.0,
-      "step": 8890
-    },
-    {
-      "epoch": 1.4318209267683883,
-      "grad_norm": 0.0027794321067631245,
-      "learning_rate": 0.000199998990640382,
-      "loss": 46.0,
-      "step": 8891
-    },
-    {
-      "epoch": 1.4319819638471758,
-      "grad_norm": 0.002924382919445634,
-      "learning_rate": 0.00019999899041306193,
-      "loss": 46.0,
-      "step": 8892
-    },
-    {
-      "epoch": 1.4321430009259632,
-      "grad_norm": 0.006947075482457876,
-      "learning_rate": 0.00019999899018571625,
-      "loss": 46.0,
-      "step": 8893
-    },
-    {
-      "epoch": 1.4323040380047507,
-      "grad_norm": 0.001057026325725019,
-      "learning_rate": 0.000199998989958345,
-      "loss": 46.0,
-      "step": 8894
-    },
-    {
-      "epoch": 1.432465075083538,
-      "grad_norm": 0.004213789477944374,
-      "learning_rate": 0.00019999898973094816,
-      "loss": 46.0,
-      "step": 8895
-    },
-    {
-      "epoch": 1.4326261121623254,
-      "grad_norm": 0.0013789647491648793,
-      "learning_rate": 0.00019999898950352573,
-      "loss": 46.0,
-      "step": 8896
-    },
-    {
-      "epoch": 1.4327871492411128,
-      "grad_norm": 0.007213205099105835,
-      "learning_rate": 0.00019999898927607767,
-      "loss": 46.0,
-      "step": 8897
-    },
-    {
-      "epoch": 1.4329481863199,
-      "grad_norm": 0.001731844968162477,
-      "learning_rate": 0.00019999898904860404,
-      "loss": 46.0,
-      "step": 8898
-    },
-    {
-      "epoch": 1.4331092233986875,
-      "grad_norm": 0.0005468594608828425,
-      "learning_rate": 0.0001999989888211048,
-      "loss": 46.0,
-      "step": 8899
-    },
-    {
-      "epoch": 1.433270260477475,
-      "grad_norm": 0.001397643587552011,
-      "learning_rate": 0.00019999898859358002,
-      "loss": 46.0,
-      "step": 8900
-    },
-    {
-      "epoch": 1.4334312975562624,
-      "grad_norm": 0.001007004757411778,
-      "learning_rate": 0.0001999989883660296,
-      "loss": 46.0,
-      "step": 8901
-    },
-    {
-      "epoch": 1.4335923346350496,
-      "grad_norm": 0.0024804940912872553,
-      "learning_rate": 0.00019999898813845358,
-      "loss": 46.0,
-      "step": 8902
-    },
-    {
-      "epoch": 1.433753371713837,
-      "grad_norm": 0.002691722009330988,
-      "learning_rate": 0.00019999898791085195,
-      "loss": 46.0,
-      "step": 8903
-    },
-    {
-      "epoch": 1.4339144087926246,
-      "grad_norm": 0.003183201886713505,
-      "learning_rate": 0.00019999898768322476,
-      "loss": 46.0,
-      "step": 8904
-    },
-    {
-      "epoch": 1.4340754458714118,
-      "grad_norm": 0.0037826804909855127,
-      "learning_rate": 0.000199998987455572,
-      "loss": 46.0,
-      "step": 8905
-    },
-    {
-      "epoch": 1.4342364829501992,
-      "grad_norm": 0.0005913475761190057,
-      "learning_rate": 0.0001999989872278936,
-      "loss": 46.0,
-      "step": 8906
-    },
-    {
-      "epoch": 1.4343975200289867,
-      "grad_norm": 0.000808125187177211,
-      "learning_rate": 0.00019999898700018962,
-      "loss": 46.0,
-      "step": 8907
-    },
-    {
-      "epoch": 1.4345585571077741,
-      "grad_norm": 0.0026960677932947874,
-      "learning_rate": 0.00019999898677246003,
-      "loss": 46.0,
-      "step": 8908
-    },
-    {
-      "epoch": 1.4347195941865616,
-      "grad_norm": 0.0020128260366618633,
-      "learning_rate": 0.00019999898654470485,
-      "loss": 46.0,
-      "step": 8909
-    },
-    {
-      "epoch": 1.4348806312653488,
-      "grad_norm": 0.0008650345844216645,
-      "learning_rate": 0.00019999898631692408,
-      "loss": 46.0,
-      "step": 8910
-    },
-    {
-      "epoch": 1.4350416683441363,
-      "grad_norm": 0.0032119606621563435,
-      "learning_rate": 0.00019999898608911773,
-      "loss": 46.0,
-      "step": 8911
-    },
-    {
-      "epoch": 1.4352027054229235,
-      "grad_norm": 0.004212197847664356,
-      "learning_rate": 0.00019999898586128576,
-      "loss": 46.0,
-      "step": 8912
-    },
-    {
-      "epoch": 1.435363742501711,
-      "grad_norm": 0.008080984465777874,
-      "learning_rate": 0.0001999989856334282,
-      "loss": 46.0,
-      "step": 8913
-    },
-    {
-      "epoch": 1.4355247795804984,
-      "grad_norm": 0.0026922726538032293,
-      "learning_rate": 0.00019999898540554506,
-      "loss": 46.0,
-      "step": 8914
-    },
-    {
-      "epoch": 1.4356858166592859,
-      "grad_norm": 0.00260039116255939,
-      "learning_rate": 0.00019999898517763633,
-      "loss": 46.0,
-      "step": 8915
-    },
-    {
-      "epoch": 1.4358468537380733,
-      "grad_norm": 0.004634491633623838,
-      "learning_rate": 0.000199998984949702,
-      "loss": 46.0,
-      "step": 8916
-    },
-    {
-      "epoch": 1.4360078908168605,
-      "grad_norm": 0.00044389566755853593,
-      "learning_rate": 0.00019999898472174206,
-      "loss": 46.0,
-      "step": 8917
-    },
-    {
-      "epoch": 1.436168927895648,
-      "grad_norm": 0.002899541286751628,
-      "learning_rate": 0.00019999898449375654,
-      "loss": 46.0,
-      "step": 8918
-    },
-    {
-      "epoch": 1.4363299649744354,
-      "grad_norm": 0.004273874685168266,
-      "learning_rate": 0.0001999989842657454,
-      "loss": 46.0,
-      "step": 8919
-    },
-    {
-      "epoch": 1.4364910020532227,
-      "grad_norm": 0.0012014367384836078,
-      "learning_rate": 0.0001999989840377087,
-      "loss": 46.0,
-      "step": 8920
-    },
-    {
-      "epoch": 1.4366520391320101,
-      "grad_norm": 0.002292466117069125,
-      "learning_rate": 0.0001999989838096464,
-      "loss": 46.0,
-      "step": 8921
-    },
-    {
-      "epoch": 1.4368130762107976,
-      "grad_norm": 0.0026189687196165323,
-      "learning_rate": 0.00019999898358155852,
-      "loss": 46.0,
-      "step": 8922
-    },
-    {
-      "epoch": 1.436974113289585,
-      "grad_norm": 0.0011711049592122436,
-      "learning_rate": 0.000199998983353445,
-      "loss": 46.0,
-      "step": 8923
-    },
-    {
-      "epoch": 1.4371351503683722,
-      "grad_norm": 0.0024189141113311052,
-      "learning_rate": 0.00019999898312530591,
-      "loss": 46.0,
-      "step": 8924
-    },
-    {
-      "epoch": 1.4372961874471597,
-      "grad_norm": 0.0033526746556162834,
-      "learning_rate": 0.00019999898289714123,
-      "loss": 46.0,
-      "step": 8925
-    },
-    {
-      "epoch": 1.4374572245259472,
-      "grad_norm": 0.0005484105204232037,
-      "learning_rate": 0.00019999898266895094,
-      "loss": 46.0,
-      "step": 8926
-    },
-    {
-      "epoch": 1.4376182616047344,
-      "grad_norm": 0.004912239499390125,
-      "learning_rate": 0.00019999898244073508,
-      "loss": 46.0,
-      "step": 8927
-    },
-    {
-      "epoch": 1.4377792986835218,
-      "grad_norm": 0.0005740529741160572,
-      "learning_rate": 0.0001999989822124936,
-      "loss": 46.0,
-      "step": 8928
-    },
-    {
-      "epoch": 1.4379403357623093,
-      "grad_norm": 0.002250917488709092,
-      "learning_rate": 0.00019999898198422652,
-      "loss": 46.0,
-      "step": 8929
-    },
-    {
-      "epoch": 1.4381013728410967,
-      "grad_norm": 0.007793302647769451,
-      "learning_rate": 0.00019999898175593388,
-      "loss": 46.0,
-      "step": 8930
-    },
-    {
-      "epoch": 1.4382624099198842,
-      "grad_norm": 0.0038926429115235806,
-      "learning_rate": 0.00019999898152761562,
-      "loss": 46.0,
-      "step": 8931
-    },
-    {
-      "epoch": 1.4384234469986714,
-      "grad_norm": 0.00298431608825922,
-      "learning_rate": 0.00019999898129927174,
-      "loss": 46.0,
-      "step": 8932
-    },
-    {
-      "epoch": 1.4385844840774589,
-      "grad_norm": 0.000561332970391959,
-      "learning_rate": 0.00019999898107090234,
-      "loss": 46.0,
-      "step": 8933
-    },
-    {
-      "epoch": 1.438745521156246,
-      "grad_norm": 0.004708392079919577,
-      "learning_rate": 0.0001999989808425073,
-      "loss": 46.0,
-      "step": 8934
-    },
-    {
-      "epoch": 1.4389065582350336,
-      "grad_norm": 0.003812340786680579,
-      "learning_rate": 0.00019999898061408663,
-      "loss": 46.0,
-      "step": 8935
-    },
-    {
-      "epoch": 1.439067595313821,
-      "grad_norm": 0.0015855850651860237,
-      "learning_rate": 0.00019999898038564043,
-      "loss": 46.0,
-      "step": 8936
-    },
-    {
-      "epoch": 1.4392286323926085,
-      "grad_norm": 0.005309516564011574,
-      "learning_rate": 0.0001999989801571686,
-      "loss": 46.0,
-      "step": 8937
-    },
-    {
-      "epoch": 1.439389669471396,
-      "grad_norm": 0.004938654135912657,
-      "learning_rate": 0.00019999897992867117,
-      "loss": 46.0,
-      "step": 8938
-    },
-    {
-      "epoch": 1.4395507065501831,
-      "grad_norm": 0.0016658473759889603,
-      "learning_rate": 0.00019999897970014816,
-      "loss": 46.0,
-      "step": 8939
-    },
-    {
-      "epoch": 1.4397117436289706,
-      "grad_norm": 0.0013597691431641579,
-      "learning_rate": 0.00019999897947159956,
-      "loss": 46.0,
-      "step": 8940
-    },
-    {
-      "epoch": 1.439872780707758,
-      "grad_norm": 0.0016742486041039228,
-      "learning_rate": 0.00019999897924302538,
-      "loss": 46.0,
-      "step": 8941
-    },
-    {
-      "epoch": 1.4400338177865453,
-      "grad_norm": 0.003886726452037692,
-      "learning_rate": 0.00019999897901442558,
-      "loss": 46.0,
-      "step": 8942
-    },
-    {
-      "epoch": 1.4401948548653327,
-      "grad_norm": 0.0016749916831031442,
-      "learning_rate": 0.0001999989787858002,
-      "loss": 46.0,
-      "step": 8943
-    },
-    {
-      "epoch": 1.4403558919441202,
-      "grad_norm": 0.0005074695800431073,
-      "learning_rate": 0.0001999989785571492,
-      "loss": 46.0,
-      "step": 8944
-    },
-    {
-      "epoch": 1.4405169290229076,
-      "grad_norm": 0.0006953999982215464,
-      "learning_rate": 0.00019999897832847263,
-      "loss": 46.0,
-      "step": 8945
-    },
-    {
-      "epoch": 1.4406779661016949,
-      "grad_norm": 0.0015014850068837404,
-      "learning_rate": 0.00019999897809977045,
-      "loss": 46.0,
-      "step": 8946
-    },
-    {
-      "epoch": 1.4408390031804823,
-      "grad_norm": 0.00080797856207937,
-      "learning_rate": 0.00019999897787104266,
-      "loss": 46.0,
-      "step": 8947
-    },
-    {
-      "epoch": 1.4410000402592698,
-      "grad_norm": 0.003238941775634885,
-      "learning_rate": 0.0001999989776422893,
-      "loss": 46.0,
-      "step": 8948
-    },
-    {
-      "epoch": 1.441161077338057,
-      "grad_norm": 0.0038205208256840706,
-      "learning_rate": 0.00019999897741351035,
-      "loss": 46.0,
-      "step": 8949
-    },
-    {
-      "epoch": 1.4413221144168444,
-      "grad_norm": 0.001656422158703208,
-      "learning_rate": 0.00019999897718470582,
-      "loss": 46.0,
-      "step": 8950
-    },
-    {
-      "epoch": 1.4414831514956319,
-      "grad_norm": 0.0015996687579900026,
-      "learning_rate": 0.00019999897695587566,
-      "loss": 46.0,
-      "step": 8951
-    },
-    {
-      "epoch": 1.4416441885744193,
-      "grad_norm": 0.002445490099489689,
-      "learning_rate": 0.00019999897672701993,
-      "loss": 46.0,
-      "step": 8952
-    },
-    {
-      "epoch": 1.4418052256532066,
-      "grad_norm": 0.0065863230265676975,
-      "learning_rate": 0.00019999897649813857,
-      "loss": 46.0,
-      "step": 8953
-    },
-    {
-      "epoch": 1.441966262731994,
-      "grad_norm": 0.002228823024779558,
-      "learning_rate": 0.00019999897626923164,
-      "loss": 46.0,
-      "step": 8954
-    },
-    {
-      "epoch": 1.4421272998107815,
-      "grad_norm": 0.0026095169596374035,
-      "learning_rate": 0.00019999897604029913,
-      "loss": 46.0,
-      "step": 8955
-    },
-    {
-      "epoch": 1.4422883368895687,
-      "grad_norm": 0.0015870523639023304,
-      "learning_rate": 0.00019999897581134102,
-      "loss": 46.0,
-      "step": 8956
-    },
-    {
-      "epoch": 1.4424493739683562,
-      "grad_norm": 0.002546041738241911,
-      "learning_rate": 0.0001999989755823573,
-      "loss": 46.0,
-      "step": 8957
-    },
-    {
-      "epoch": 1.4426104110471436,
-      "grad_norm": 0.0006798851536586881,
-      "learning_rate": 0.000199998975353348,
-      "loss": 46.0,
-      "step": 8958
-    },
-    {
-      "epoch": 1.442771448125931,
-      "grad_norm": 0.0013138747308403254,
-      "learning_rate": 0.00019999897512431306,
-      "loss": 46.0,
-      "step": 8959
-    },
-    {
-      "epoch": 1.4429324852047185,
-      "grad_norm": 0.0011445775162428617,
-      "learning_rate": 0.00019999897489525258,
-      "loss": 46.0,
-      "step": 8960
-    },
-    {
-      "epoch": 1.4430935222835057,
-      "grad_norm": 0.0018516422715038061,
-      "learning_rate": 0.0001999989746661665,
-      "loss": 46.0,
-      "step": 8961
-    },
-    {
-      "epoch": 1.4432545593622932,
-      "grad_norm": 0.00044858240289613605,
-      "learning_rate": 0.00019999897443705481,
-      "loss": 46.0,
-      "step": 8962
-    },
-    {
-      "epoch": 1.4434155964410806,
-      "grad_norm": 0.002639931160956621,
-      "learning_rate": 0.00019999897420791752,
-      "loss": 46.0,
-      "step": 8963
-    },
-    {
-      "epoch": 1.4435766335198679,
-      "grad_norm": 0.009227258153259754,
-      "learning_rate": 0.00019999897397875464,
-      "loss": 46.0,
-      "step": 8964
-    },
-    {
-      "epoch": 1.4437376705986553,
-      "grad_norm": 0.0008425979758612812,
-      "learning_rate": 0.00019999897374956617,
-      "loss": 46.0,
-      "step": 8965
-    },
-    {
-      "epoch": 1.4438987076774428,
-      "grad_norm": 0.0006759099778719246,
-      "learning_rate": 0.00019999897352035212,
-      "loss": 46.0,
-      "step": 8966
-    },
-    {
-      "epoch": 1.4440597447562302,
-      "grad_norm": 0.00044971969327889383,
-      "learning_rate": 0.00019999897329111245,
-      "loss": 46.0,
-      "step": 8967
-    },
-    {
-      "epoch": 1.4442207818350175,
-      "grad_norm": 0.0006275823689065874,
-      "learning_rate": 0.0001999989730618472,
-      "loss": 46.0,
-      "step": 8968
-    },
-    {
-      "epoch": 1.444381818913805,
-      "grad_norm": 0.0018190392293035984,
-      "learning_rate": 0.00019999897283255632,
-      "loss": 46.0,
-      "step": 8969
-    },
-    {
-      "epoch": 1.4445428559925924,
-      "grad_norm": 0.0032785404473543167,
-      "learning_rate": 0.0001999989726032399,
-      "loss": 46.0,
-      "step": 8970
-    },
-    {
-      "epoch": 1.4447038930713796,
-      "grad_norm": 0.0007270055939443409,
-      "learning_rate": 0.00019999897237389788,
-      "loss": 46.0,
-      "step": 8971
-    },
-    {
-      "epoch": 1.444864930150167,
-      "grad_norm": 0.0024905952159315348,
-      "learning_rate": 0.00019999897214453022,
-      "loss": 46.0,
-      "step": 8972
-    },
-    {
-      "epoch": 1.4450259672289545,
-      "grad_norm": 0.00047327953507192433,
-      "learning_rate": 0.000199998971915137,
-      "loss": 46.0,
-      "step": 8973
-    },
-    {
-      "epoch": 1.445187004307742,
-      "grad_norm": 0.0021170086693018675,
-      "learning_rate": 0.0001999989716857182,
-      "loss": 46.0,
-      "step": 8974
-    },
-    {
-      "epoch": 1.4453480413865292,
-      "grad_norm": 0.0029951727483421564,
-      "learning_rate": 0.00019999897145627377,
-      "loss": 46.0,
-      "step": 8975
-    },
-    {
-      "epoch": 1.4455090784653166,
-      "grad_norm": 0.0024707107804715633,
-      "learning_rate": 0.00019999897122680377,
-      "loss": 46.0,
-      "step": 8976
-    },
-    {
-      "epoch": 1.445670115544104,
-      "grad_norm": 0.0008560841088183224,
-      "learning_rate": 0.00019999897099730814,
-      "loss": 46.0,
-      "step": 8977
-    },
-    {
-      "epoch": 1.4458311526228913,
-      "grad_norm": 0.001000311691313982,
-      "learning_rate": 0.00019999897076778696,
-      "loss": 46.0,
-      "step": 8978
-    },
-    {
-      "epoch": 1.4459921897016788,
-      "grad_norm": 0.0013898588949814439,
-      "learning_rate": 0.00019999897053824014,
-      "loss": 46.0,
-      "step": 8979
-    },
-    {
-      "epoch": 1.4461532267804662,
-      "grad_norm": 0.002270257333293557,
-      "learning_rate": 0.00019999897030866776,
-      "loss": 46.0,
-      "step": 8980
-    },
-    {
-      "epoch": 1.4463142638592537,
-      "grad_norm": 0.0007034658337943256,
-      "learning_rate": 0.00019999897007906979,
-      "loss": 46.0,
-      "step": 8981
-    },
-    {
-      "epoch": 1.446475300938041,
-      "grad_norm": 0.003080200869590044,
-      "learning_rate": 0.0001999989698494462,
-      "loss": 46.0,
-      "step": 8982
-    },
-    {
-      "epoch": 1.4466363380168283,
-      "grad_norm": 0.005392127670347691,
-      "learning_rate": 0.00019999896961979703,
-      "loss": 46.0,
-      "step": 8983
-    },
-    {
-      "epoch": 1.4467973750956158,
-      "grad_norm": 0.0011138945119455457,
-      "learning_rate": 0.00019999896939012227,
-      "loss": 46.0,
-      "step": 8984
-    },
-    {
-      "epoch": 1.446958412174403,
-      "grad_norm": 0.002085252432152629,
-      "learning_rate": 0.0001999989691604219,
-      "loss": 46.0,
-      "step": 8985
-    },
-    {
-      "epoch": 1.4471194492531905,
-      "grad_norm": 0.0038350208196789026,
-      "learning_rate": 0.00019999896893069594,
-      "loss": 46.0,
-      "step": 8986
-    },
-    {
-      "epoch": 1.447280486331978,
-      "grad_norm": 0.0015229257987812161,
-      "learning_rate": 0.0001999989687009444,
-      "loss": 46.0,
-      "step": 8987
-    },
-    {
-      "epoch": 1.4474415234107654,
-      "grad_norm": 0.002044251887127757,
-      "learning_rate": 0.00019999896847116723,
-      "loss": 46.0,
-      "step": 8988
-    },
-    {
-      "epoch": 1.4476025604895528,
-      "grad_norm": 0.0011432894971221685,
-      "learning_rate": 0.0001999989682413645,
-      "loss": 46.0,
-      "step": 8989
-    },
-    {
-      "epoch": 1.44776359756834,
-      "grad_norm": 0.0019160597585141659,
-      "learning_rate": 0.00019999896801153617,
-      "loss": 46.0,
-      "step": 8990
-    },
-    {
-      "epoch": 1.4479246346471275,
-      "grad_norm": 0.0007443545036949217,
-      "learning_rate": 0.00019999896778168222,
-      "loss": 46.0,
-      "step": 8991
-    },
-    {
-      "epoch": 1.448085671725915,
-      "grad_norm": 0.0023979567922651768,
-      "learning_rate": 0.0001999989675518027,
-      "loss": 46.0,
-      "step": 8992
-    },
-    {
-      "epoch": 1.4482467088047022,
-      "grad_norm": 0.0017052312614396214,
-      "learning_rate": 0.00019999896732189758,
-      "loss": 46.0,
-      "step": 8993
-    },
-    {
-      "epoch": 1.4484077458834896,
-      "grad_norm": 0.002735251560807228,
-      "learning_rate": 0.00019999896709196687,
-      "loss": 46.0,
-      "step": 8994
-    },
-    {
-      "epoch": 1.448568782962277,
-      "grad_norm": 0.0045250714756548405,
-      "learning_rate": 0.00019999896686201054,
-      "loss": 46.0,
-      "step": 8995
-    },
-    {
-      "epoch": 1.4487298200410645,
-      "grad_norm": 0.0044335597194731236,
-      "learning_rate": 0.00019999896663202863,
-      "loss": 46.0,
-      "step": 8996
-    },
-    {
-      "epoch": 1.4488908571198518,
-      "grad_norm": 0.0023370222188532352,
-      "learning_rate": 0.00019999896640202116,
-      "loss": 46.0,
-      "step": 8997
-    },
-    {
-      "epoch": 1.4490518941986392,
-      "grad_norm": 0.003713067388162017,
-      "learning_rate": 0.00019999896617198807,
-      "loss": 46.0,
-      "step": 8998
-    },
-    {
-      "epoch": 1.4492129312774267,
-      "grad_norm": 0.0035601495765149593,
-      "learning_rate": 0.00019999896594192937,
-      "loss": 46.0,
-      "step": 8999
-    },
-    {
-      "epoch": 1.449373968356214,
-      "grad_norm": 0.0014270787360146642,
-      "learning_rate": 0.00019999896571184508,
-      "loss": 46.0,
-      "step": 9000
-    },
-    {
-      "epoch": 1.4495350054350014,
-      "grad_norm": 0.007551154121756554,
-      "learning_rate": 0.00019999896548173523,
-      "loss": 46.0,
-      "step": 9001
-    },
-    {
-      "epoch": 1.4496960425137888,
-      "grad_norm": 0.005864028353244066,
-      "learning_rate": 0.00019999896525159976,
-      "loss": 46.0,
-      "step": 9002
-    },
-    {
-      "epoch": 1.4498570795925763,
-      "grad_norm": 0.0005610091029666364,
-      "learning_rate": 0.00019999896502143866,
-      "loss": 46.0,
-      "step": 9003
-    },
-    {
-      "epoch": 1.4500181166713637,
-      "grad_norm": 0.0017337421886622906,
-      "learning_rate": 0.00019999896479125202,
-      "loss": 46.0,
-      "step": 9004
-    },
-    {
-      "epoch": 1.450179153750151,
-      "grad_norm": 0.0008135696989484131,
-      "learning_rate": 0.00019999896456103974,
-      "loss": 46.0,
-      "step": 9005
-    },
-    {
-      "epoch": 1.4503401908289384,
-      "grad_norm": 0.006821257993578911,
-      "learning_rate": 0.00019999896433080193,
-      "loss": 46.0,
-      "step": 9006
-    },
-    {
-      "epoch": 1.4505012279077256,
-      "grad_norm": 0.005144826602190733,
-      "learning_rate": 0.00019999896410053848,
-      "loss": 46.0,
-      "step": 9007
-    },
-    {
-      "epoch": 1.450662264986513,
-      "grad_norm": 0.0009511394891887903,
-      "learning_rate": 0.00019999896387024944,
-      "loss": 46.0,
-      "step": 9008
-    },
-    {
-      "epoch": 1.4508233020653005,
-      "grad_norm": 0.0011892416514456272,
-      "learning_rate": 0.0001999989636399348,
-      "loss": 46.0,
-      "step": 9009
-    },
-    {
-      "epoch": 1.450984339144088,
-      "grad_norm": 0.009082089178264141,
-      "learning_rate": 0.00019999896340959456,
-      "loss": 46.0,
-      "step": 9010
-    },
-    {
-      "epoch": 1.4511453762228754,
-      "grad_norm": 0.0015414797235280275,
-      "learning_rate": 0.00019999896317922874,
-      "loss": 46.0,
-      "step": 9011
-    },
-    {
-      "epoch": 1.4513064133016627,
-      "grad_norm": 0.0003873731184285134,
-      "learning_rate": 0.00019999896294883732,
-      "loss": 46.0,
-      "step": 9012
-    },
-    {
-      "epoch": 1.45146745038045,
-      "grad_norm": 0.0009792075725272298,
-      "learning_rate": 0.00019999896271842031,
-      "loss": 46.0,
-      "step": 9013
-    },
-    {
-      "epoch": 1.4516284874592376,
-      "grad_norm": 0.0015056653646752238,
-      "learning_rate": 0.00019999896248797772,
-      "loss": 46.0,
-      "step": 9014
-    },
-    {
-      "epoch": 1.4517895245380248,
-      "grad_norm": 0.0022380282171070576,
-      "learning_rate": 0.0001999989622575095,
-      "loss": 46.0,
-      "step": 9015
-    },
-    {
-      "epoch": 1.4519505616168122,
-      "grad_norm": 0.0013196611544117332,
-      "learning_rate": 0.00019999896202701573,
-      "loss": 46.0,
-      "step": 9016
-    },
-    {
-      "epoch": 1.4521115986955997,
-      "grad_norm": 0.0054290881380438805,
-      "learning_rate": 0.00019999896179649632,
-      "loss": 46.0,
-      "step": 9017
-    },
-    {
-      "epoch": 1.4522726357743871,
-      "grad_norm": 0.0009963031625375152,
-      "learning_rate": 0.00019999896156595135,
-      "loss": 46.0,
-      "step": 9018
-    },
-    {
-      "epoch": 1.4524336728531744,
-      "grad_norm": 0.00265924958512187,
-      "learning_rate": 0.00019999896133538077,
-      "loss": 46.0,
-      "step": 9019
-    },
-    {
-      "epoch": 1.4525947099319618,
-      "grad_norm": 0.0035344508942216635,
-      "learning_rate": 0.00019999896110478458,
-      "loss": 46.0,
-      "step": 9020
-    },
-    {
-      "epoch": 1.4527557470107493,
-      "grad_norm": 0.002570083364844322,
-      "learning_rate": 0.00019999896087416282,
-      "loss": 46.0,
-      "step": 9021
-    },
-    {
-      "epoch": 1.4529167840895365,
-      "grad_norm": 0.0011035854695364833,
-      "learning_rate": 0.00019999896064351545,
-      "loss": 46.0,
-      "step": 9022
-    },
-    {
-      "epoch": 1.453077821168324,
-      "grad_norm": 0.002907379297539592,
-      "learning_rate": 0.0001999989604128425,
-      "loss": 46.0,
-      "step": 9023
-    },
-    {
-      "epoch": 1.4532388582471114,
-      "grad_norm": 0.0038665556348860264,
-      "learning_rate": 0.00019999896018214392,
-      "loss": 46.0,
-      "step": 9024
-    },
-    {
-      "epoch": 1.4533998953258989,
-      "grad_norm": 0.002441034186631441,
-      "learning_rate": 0.0001999989599514198,
-      "loss": 46.0,
-      "step": 9025
-    },
-    {
-      "epoch": 1.4535609324046863,
-      "grad_norm": 0.005121355410665274,
-      "learning_rate": 0.00019999895972067005,
-      "loss": 46.0,
-      "step": 9026
-    },
-    {
-      "epoch": 1.4537219694834735,
-      "grad_norm": 0.0029127185698598623,
-      "learning_rate": 0.00019999895948989471,
-      "loss": 46.0,
-      "step": 9027
-    },
-    {
-      "epoch": 1.453883006562261,
-      "grad_norm": 0.006232391111552715,
-      "learning_rate": 0.0001999989592590938,
-      "loss": 46.0,
-      "step": 9028
-    },
-    {
-      "epoch": 1.4540440436410482,
-      "grad_norm": 0.001079633249901235,
-      "learning_rate": 0.00019999895902826723,
-      "loss": 46.0,
-      "step": 9029
-    },
-    {
-      "epoch": 1.4542050807198357,
-      "grad_norm": 0.0013632382033392787,
-      "learning_rate": 0.00019999895879741514,
-      "loss": 46.0,
-      "step": 9030
-    },
-    {
-      "epoch": 1.4543661177986231,
-      "grad_norm": 0.0015014581149443984,
-      "learning_rate": 0.00019999895856653743,
-      "loss": 46.0,
-      "step": 9031
-    },
-    {
-      "epoch": 1.4545271548774106,
-      "grad_norm": 0.0028064846992492676,
-      "learning_rate": 0.0001999989583356341,
-      "loss": 46.0,
-      "step": 9032
-    },
-    {
-      "epoch": 1.454688191956198,
-      "grad_norm": 0.0010828478261828423,
-      "learning_rate": 0.00019999895810470522,
-      "loss": 46.0,
-      "step": 9033
-    },
-    {
-      "epoch": 1.4548492290349853,
-      "grad_norm": 0.0010927043622359633,
-      "learning_rate": 0.00019999895787375073,
-      "loss": 46.0,
-      "step": 9034
-    },
-    {
-      "epoch": 1.4550102661137727,
-      "grad_norm": 0.0018936865963041782,
-      "learning_rate": 0.00019999895764277062,
-      "loss": 46.0,
-      "step": 9035
-    },
-    {
-      "epoch": 1.4551713031925602,
-      "grad_norm": 0.003322634380310774,
-      "learning_rate": 0.00019999895741176492,
-      "loss": 46.0,
-      "step": 9036
-    },
-    {
-      "epoch": 1.4553323402713474,
-      "grad_norm": 0.0016357441199943423,
-      "learning_rate": 0.00019999895718073366,
-      "loss": 46.0,
-      "step": 9037
-    },
-    {
-      "epoch": 1.4554933773501348,
-      "grad_norm": 0.005125960800796747,
-      "learning_rate": 0.00019999895694967679,
-      "loss": 46.0,
-      "step": 9038
-    },
-    {
-      "epoch": 1.4556544144289223,
-      "grad_norm": 0.0005522160790860653,
-      "learning_rate": 0.0001999989567185943,
-      "loss": 46.0,
-      "step": 9039
-    },
-    {
-      "epoch": 1.4558154515077097,
-      "grad_norm": 0.0020277206785976887,
-      "learning_rate": 0.00019999895648748622,
-      "loss": 46.0,
-      "step": 9040
-    },
-    {
-      "epoch": 1.455976488586497,
-      "grad_norm": 0.0015582212945446372,
-      "learning_rate": 0.00019999895625635256,
-      "loss": 46.0,
-      "step": 9041
-    },
-    {
-      "epoch": 1.4561375256652844,
-      "grad_norm": 0.010356155224144459,
-      "learning_rate": 0.00019999895602519331,
-      "loss": 46.0,
-      "step": 9042
-    },
-    {
-      "epoch": 1.4562985627440719,
-      "grad_norm": 0.0014032598119229078,
-      "learning_rate": 0.00019999895579400848,
-      "loss": 46.0,
-      "step": 9043
-    },
-    {
-      "epoch": 1.456459599822859,
-      "grad_norm": 0.00033606356009840965,
-      "learning_rate": 0.00019999895556279803,
-      "loss": 46.0,
-      "step": 9044
-    },
-    {
-      "epoch": 1.4566206369016466,
-      "grad_norm": 0.0014868822181597352,
-      "learning_rate": 0.000199998955331562,
-      "loss": 46.0,
-      "step": 9045
-    },
-    {
-      "epoch": 1.456781673980434,
-      "grad_norm": 0.0010604661656543612,
-      "learning_rate": 0.00019999895510030037,
-      "loss": 46.0,
-      "step": 9046
-    },
-    {
-      "epoch": 1.4569427110592215,
-      "grad_norm": 0.0005255446303635836,
-      "learning_rate": 0.00019999895486901313,
-      "loss": 46.0,
-      "step": 9047
-    },
-    {
-      "epoch": 1.457103748138009,
-      "grad_norm": 0.0013461486669257283,
-      "learning_rate": 0.0001999989546377003,
-      "loss": 46.0,
-      "step": 9048
-    },
-    {
-      "epoch": 1.4572647852167961,
-      "grad_norm": 0.0026272833347320557,
-      "learning_rate": 0.0001999989544063619,
-      "loss": 46.0,
-      "step": 9049
-    },
-    {
-      "epoch": 1.4574258222955836,
-      "grad_norm": 0.0005306096281856298,
-      "learning_rate": 0.00019999895417499786,
-      "loss": 46.0,
-      "step": 9050
-    },
-    {
-      "epoch": 1.4575868593743708,
-      "grad_norm": 0.0021950858645141125,
-      "learning_rate": 0.00019999895394360825,
-      "loss": 46.0,
-      "step": 9051
-    },
-    {
-      "epoch": 1.4577478964531583,
-      "grad_norm": 0.005761492531746626,
-      "learning_rate": 0.00019999895371219307,
-      "loss": 46.0,
-      "step": 9052
-    },
-    {
-      "epoch": 1.4579089335319457,
-      "grad_norm": 0.0008243282209150493,
-      "learning_rate": 0.00019999895348075226,
-      "loss": 46.0,
-      "step": 9053
-    },
-    {
-      "epoch": 1.4580699706107332,
-      "grad_norm": 0.0006277612410485744,
-      "learning_rate": 0.00019999895324928585,
-      "loss": 46.0,
-      "step": 9054
-    },
-    {
-      "epoch": 1.4582310076895206,
-      "grad_norm": 0.000492796883918345,
-      "learning_rate": 0.0001999989530177939,
-      "loss": 46.0,
-      "step": 9055
-    },
-    {
-      "epoch": 1.4583920447683079,
-      "grad_norm": 0.003343243384733796,
-      "learning_rate": 0.0001999989527862763,
-      "loss": 46.0,
-      "step": 9056
-    },
-    {
-      "epoch": 1.4585530818470953,
-      "grad_norm": 0.0018881288124248385,
-      "learning_rate": 0.00019999895255473312,
-      "loss": 46.0,
-      "step": 9057
-    },
-    {
-      "epoch": 1.4587141189258828,
-      "grad_norm": 0.002505961339920759,
-      "learning_rate": 0.00019999895232316437,
-      "loss": 46.0,
-      "step": 9058
-    },
-    {
-      "epoch": 1.45887515600467,
-      "grad_norm": 0.007228973787277937,
-      "learning_rate": 0.00019999895209157,
-      "loss": 46.0,
-      "step": 9059
-    },
-    {
-      "epoch": 1.4590361930834574,
-      "grad_norm": 0.0009150710538960993,
-      "learning_rate": 0.00019999895185995005,
-      "loss": 46.0,
-      "step": 9060
-    },
-    {
-      "epoch": 1.459197230162245,
-      "grad_norm": 0.0005061214906163514,
-      "learning_rate": 0.00019999895162830448,
-      "loss": 46.0,
-      "step": 9061
-    },
-    {
-      "epoch": 1.4593582672410323,
-      "grad_norm": 0.0010553308529779315,
-      "learning_rate": 0.00019999895139663333,
-      "loss": 46.0,
-      "step": 9062
-    },
-    {
-      "epoch": 1.4595193043198196,
-      "grad_norm": 0.0007964483229443431,
-      "learning_rate": 0.00019999895116493659,
-      "loss": 46.0,
-      "step": 9063
-    },
-    {
-      "epoch": 1.459680341398607,
-      "grad_norm": 0.0025870187673717737,
-      "learning_rate": 0.00019999895093321426,
-      "loss": 46.0,
-      "step": 9064
-    },
-    {
-      "epoch": 1.4598413784773945,
-      "grad_norm": 0.0016578648937866092,
-      "learning_rate": 0.00019999895070146631,
-      "loss": 46.0,
-      "step": 9065
-    },
-    {
-      "epoch": 1.4600024155561817,
-      "grad_norm": 0.007649368140846491,
-      "learning_rate": 0.0001999989504696928,
-      "loss": 46.0,
-      "step": 9066
-    },
-    {
-      "epoch": 1.4601634526349692,
-      "grad_norm": 0.0017081158002838492,
-      "learning_rate": 0.00019999895023789367,
-      "loss": 46.0,
-      "step": 9067
-    },
-    {
-      "epoch": 1.4603244897137566,
-      "grad_norm": 0.0014930607285350561,
-      "learning_rate": 0.00019999895000606896,
-      "loss": 46.0,
-      "step": 9068
-    },
-    {
-      "epoch": 1.460485526792544,
-      "grad_norm": 0.001350192236714065,
-      "learning_rate": 0.00019999894977421864,
-      "loss": 46.0,
-      "step": 9069
-    },
-    {
-      "epoch": 1.4606465638713313,
-      "grad_norm": 0.0026856905315071344,
-      "learning_rate": 0.00019999894954234276,
-      "loss": 46.0,
-      "step": 9070
-    },
-    {
-      "epoch": 1.4608076009501187,
-      "grad_norm": 0.0008866675198078156,
-      "learning_rate": 0.00019999894931044124,
-      "loss": 46.0,
-      "step": 9071
-    },
-    {
-      "epoch": 1.4609686380289062,
-      "grad_norm": 0.0015195850282907486,
-      "learning_rate": 0.00019999894907851416,
-      "loss": 46.0,
-      "step": 9072
-    },
-    {
-      "epoch": 1.4611296751076934,
-      "grad_norm": 0.006534232757985592,
-      "learning_rate": 0.00019999894884656147,
-      "loss": 46.0,
-      "step": 9073
-    },
-    {
-      "epoch": 1.4612907121864809,
-      "grad_norm": 0.0015818667598068714,
-      "learning_rate": 0.00019999894861458318,
-      "loss": 46.0,
-      "step": 9074
-    },
-    {
-      "epoch": 1.4614517492652683,
-      "grad_norm": 0.00365554541349411,
-      "learning_rate": 0.00019999894838257932,
-      "loss": 46.0,
-      "step": 9075
-    },
-    {
-      "epoch": 1.4616127863440558,
-      "grad_norm": 0.0020009437575936317,
-      "learning_rate": 0.00019999894815054986,
-      "loss": 46.0,
-      "step": 9076
-    },
-    {
-      "epoch": 1.4617738234228432,
-      "grad_norm": 0.004016316030174494,
-      "learning_rate": 0.00019999894791849476,
-      "loss": 46.0,
-      "step": 9077
-    },
-    {
-      "epoch": 1.4619348605016305,
-      "grad_norm": 0.003942732699215412,
-      "learning_rate": 0.00019999894768641413,
-      "loss": 46.0,
-      "step": 9078
-    },
-    {
-      "epoch": 1.462095897580418,
-      "grad_norm": 0.003082046052441001,
-      "learning_rate": 0.00019999894745430786,
-      "loss": 46.0,
-      "step": 9079
-    },
-    {
-      "epoch": 1.4622569346592054,
-      "grad_norm": 0.0013199597597122192,
-      "learning_rate": 0.00019999894722217603,
-      "loss": 46.0,
-      "step": 9080
-    },
-    {
-      "epoch": 1.4624179717379926,
-      "grad_norm": 0.0016363190952688456,
-      "learning_rate": 0.00019999894699001858,
-      "loss": 46.0,
-      "step": 9081
-    },
-    {
-      "epoch": 1.46257900881678,
-      "grad_norm": 0.00294755888171494,
-      "learning_rate": 0.00019999894675783552,
-      "loss": 46.0,
-      "step": 9082
-    },
-    {
-      "epoch": 1.4627400458955675,
-      "grad_norm": 0.002998506650328636,
-      "learning_rate": 0.0001999989465256269,
-      "loss": 46.0,
-      "step": 9083
-    },
-    {
-      "epoch": 1.462901082974355,
-      "grad_norm": 0.0039437864907085896,
-      "learning_rate": 0.00019999894629339266,
-      "loss": 46.0,
-      "step": 9084
-    },
-    {
-      "epoch": 1.4630621200531422,
-      "grad_norm": 0.001787786721251905,
-      "learning_rate": 0.00019999894606113284,
-      "loss": 46.0,
-      "step": 9085
-    },
-    {
-      "epoch": 1.4632231571319296,
-      "grad_norm": 0.005266864784061909,
-      "learning_rate": 0.0001999989458288474,
-      "loss": 46.0,
-      "step": 9086
-    },
-    {
-      "epoch": 1.463384194210717,
-      "grad_norm": 0.004001102410256863,
-      "learning_rate": 0.0001999989455965364,
-      "loss": 46.0,
-      "step": 9087
-    },
-    {
-      "epoch": 1.4635452312895043,
-      "grad_norm": 0.004251067992299795,
-      "learning_rate": 0.0001999989453641998,
-      "loss": 46.0,
-      "step": 9088
-    },
-    {
-      "epoch": 1.4637062683682918,
-      "grad_norm": 0.00332429981790483,
-      "learning_rate": 0.0001999989451318376,
-      "loss": 46.0,
-      "step": 9089
-    },
-    {
-      "epoch": 1.4638673054470792,
-      "grad_norm": 0.0018108172807842493,
-      "learning_rate": 0.00019999894489944979,
-      "loss": 46.0,
-      "step": 9090
-    },
-    {
-      "epoch": 1.4640283425258667,
-      "grad_norm": 0.005174580030143261,
-      "learning_rate": 0.00019999894466703641,
-      "loss": 46.0,
-      "step": 9091
-    },
-    {
-      "epoch": 1.464189379604654,
-      "grad_norm": 0.001297090551815927,
-      "learning_rate": 0.0001999989444345974,
-      "loss": 46.0,
-      "step": 9092
-    },
-    {
-      "epoch": 1.4643504166834413,
-      "grad_norm": 0.0008506930316798389,
-      "learning_rate": 0.00019999894420213283,
-      "loss": 46.0,
-      "step": 9093
-    },
-    {
-      "epoch": 1.4645114537622288,
-      "grad_norm": 0.00564588513225317,
-      "learning_rate": 0.00019999894396964267,
-      "loss": 46.0,
-      "step": 9094
-    },
-    {
-      "epoch": 1.464672490841016,
-      "grad_norm": 0.002092865528538823,
-      "learning_rate": 0.0001999989437371269,
-      "loss": 46.0,
-      "step": 9095
-    },
-    {
-      "epoch": 1.4648335279198035,
-      "grad_norm": 0.0006037953426130116,
-      "learning_rate": 0.00019999894350458553,
-      "loss": 46.0,
-      "step": 9096
-    },
-    {
-      "epoch": 1.464994564998591,
-      "grad_norm": 0.0027539010625332594,
-      "learning_rate": 0.00019999894327201858,
-      "loss": 46.0,
-      "step": 9097
-    },
-    {
-      "epoch": 1.4651556020773784,
-      "grad_norm": 0.0013463215436786413,
-      "learning_rate": 0.00019999894303942601,
-      "loss": 46.0,
-      "step": 9098
-    },
-    {
-      "epoch": 1.4653166391561658,
-      "grad_norm": 0.0013604918494820595,
-      "learning_rate": 0.00019999894280680786,
-      "loss": 46.0,
-      "step": 9099
-    },
-    {
-      "epoch": 1.465477676234953,
-      "grad_norm": 0.0030413393396884203,
-      "learning_rate": 0.00019999894257416415,
-      "loss": 46.0,
-      "step": 9100
-    },
-    {
-      "epoch": 1.4656387133137405,
-      "grad_norm": 0.0007645843434147537,
-      "learning_rate": 0.0001999989423414948,
-      "loss": 46.0,
-      "step": 9101
-    },
-    {
-      "epoch": 1.4657997503925277,
-      "grad_norm": 0.0010827609803527594,
-      "learning_rate": 0.00019999894210879986,
-      "loss": 46.0,
-      "step": 9102
-    },
-    {
-      "epoch": 1.4659607874713152,
-      "grad_norm": 0.0022910968400537968,
-      "learning_rate": 0.00019999894187607936,
-      "loss": 46.0,
-      "step": 9103
-    },
-    {
-      "epoch": 1.4661218245501026,
-      "grad_norm": 0.0005529465852305293,
-      "learning_rate": 0.00019999894164333322,
-      "loss": 46.0,
-      "step": 9104
-    },
-    {
-      "epoch": 1.46628286162889,
-      "grad_norm": 0.004184665624052286,
-      "learning_rate": 0.0001999989414105615,
-      "loss": 46.0,
-      "step": 9105
-    },
-    {
-      "epoch": 1.4664438987076776,
-      "grad_norm": 0.0013064953964203596,
-      "learning_rate": 0.0001999989411777642,
-      "loss": 46.0,
-      "step": 9106
-    },
-    {
-      "epoch": 1.4666049357864648,
-      "grad_norm": 0.0020443282555788755,
-      "learning_rate": 0.0001999989409449413,
-      "loss": 46.0,
-      "step": 9107
-    },
-    {
-      "epoch": 1.4667659728652522,
-      "grad_norm": 0.0005454795900732279,
-      "learning_rate": 0.0001999989407120928,
-      "loss": 46.0,
-      "step": 9108
-    },
-    {
-      "epoch": 1.4669270099440397,
-      "grad_norm": 0.0059854923747479916,
-      "learning_rate": 0.0001999989404792187,
-      "loss": 46.0,
-      "step": 9109
-    },
-    {
-      "epoch": 1.467088047022827,
-      "grad_norm": 0.0008014341583475471,
-      "learning_rate": 0.00019999894024631901,
-      "loss": 46.0,
-      "step": 9110
-    },
-    {
-      "epoch": 1.4672490841016144,
-      "grad_norm": 0.001022852840833366,
-      "learning_rate": 0.00019999894001339374,
-      "loss": 46.0,
-      "step": 9111
-    },
-    {
-      "epoch": 1.4674101211804018,
-      "grad_norm": 0.00381990778259933,
-      "learning_rate": 0.00019999893978044284,
-      "loss": 46.0,
-      "step": 9112
-    },
-    {
-      "epoch": 1.4675711582591893,
-      "grad_norm": 0.0009041810408234596,
-      "learning_rate": 0.0001999989395474664,
-      "loss": 46.0,
-      "step": 9113
-    },
-    {
-      "epoch": 1.4677321953379765,
-      "grad_norm": 0.0007273399969562888,
-      "learning_rate": 0.00019999893931446432,
-      "loss": 46.0,
-      "step": 9114
-    },
-    {
-      "epoch": 1.467893232416764,
-      "grad_norm": 0.0011411912273615599,
-      "learning_rate": 0.00019999893908143664,
-      "loss": 46.0,
-      "step": 9115
-    },
-    {
-      "epoch": 1.4680542694955514,
-      "grad_norm": 0.0024391021579504013,
-      "learning_rate": 0.0001999989388483834,
-      "loss": 46.0,
-      "step": 9116
-    },
-    {
-      "epoch": 1.4682153065743386,
-      "grad_norm": 0.0018117907457053661,
-      "learning_rate": 0.00019999893861530454,
-      "loss": 46.0,
-      "step": 9117
-    },
-    {
-      "epoch": 1.468376343653126,
-      "grad_norm": 0.003756811609491706,
-      "learning_rate": 0.0001999989383822001,
-      "loss": 46.0,
-      "step": 9118
-    },
-    {
-      "epoch": 1.4685373807319135,
-      "grad_norm": 0.0021754552144557238,
-      "learning_rate": 0.00019999893814907007,
-      "loss": 46.0,
-      "step": 9119
-    },
-    {
-      "epoch": 1.468698417810701,
-      "grad_norm": 0.0025537365581840277,
-      "learning_rate": 0.00019999893791591445,
-      "loss": 46.0,
-      "step": 9120
-    },
-    {
-      "epoch": 1.4688594548894884,
-      "grad_norm": 0.0020423149690032005,
-      "learning_rate": 0.0001999989376827332,
-      "loss": 46.0,
-      "step": 9121
-    },
-    {
-      "epoch": 1.4690204919682757,
-      "grad_norm": 0.0010749796638265252,
-      "learning_rate": 0.00019999893744952638,
-      "loss": 46.0,
-      "step": 9122
-    },
-    {
-      "epoch": 1.4691815290470631,
-      "grad_norm": 0.0017775411251932383,
-      "learning_rate": 0.00019999893721629394,
-      "loss": 46.0,
-      "step": 9123
-    },
-    {
-      "epoch": 1.4693425661258503,
-      "grad_norm": 0.0027228447142988443,
-      "learning_rate": 0.00019999893698303595,
-      "loss": 46.0,
-      "step": 9124
-    },
-    {
-      "epoch": 1.4695036032046378,
-      "grad_norm": 0.003028589766472578,
-      "learning_rate": 0.00019999893674975234,
-      "loss": 46.0,
-      "step": 9125
-    },
-    {
-      "epoch": 1.4696646402834252,
-      "grad_norm": 0.0011869377922266722,
-      "learning_rate": 0.00019999893651644315,
-      "loss": 46.0,
-      "step": 9126
-    },
-    {
-      "epoch": 1.4698256773622127,
-      "grad_norm": 0.0008890478638932109,
-      "learning_rate": 0.00019999893628310834,
-      "loss": 46.0,
-      "step": 9127
-    },
-    {
-      "epoch": 1.4699867144410002,
-      "grad_norm": 0.005459483712911606,
-      "learning_rate": 0.00019999893604974794,
-      "loss": 46.0,
-      "step": 9128
-    },
-    {
-      "epoch": 1.4701477515197874,
-      "grad_norm": 0.0008363460656255484,
-      "learning_rate": 0.00019999893581636196,
-      "loss": 46.0,
-      "step": 9129
-    },
-    {
-      "epoch": 1.4703087885985748,
-      "grad_norm": 0.0013150371378287673,
-      "learning_rate": 0.0001999989355829504,
-      "loss": 46.0,
-      "step": 9130
-    },
-    {
-      "epoch": 1.4704698256773623,
-      "grad_norm": 0.0022796099074184895,
-      "learning_rate": 0.0001999989353495132,
-      "loss": 46.0,
-      "step": 9131
-    },
-    {
-      "epoch": 1.4706308627561495,
-      "grad_norm": 0.00303072901442647,
-      "learning_rate": 0.00019999893511605044,
-      "loss": 46.0,
-      "step": 9132
-    },
-    {
-      "epoch": 1.470791899834937,
-      "grad_norm": 0.00559916952624917,
-      "learning_rate": 0.00019999893488256208,
-      "loss": 46.0,
-      "step": 9133
-    },
-    {
-      "epoch": 1.4709529369137244,
-      "grad_norm": 0.011432289145886898,
-      "learning_rate": 0.0001999989346490481,
-      "loss": 46.0,
-      "step": 9134
-    },
-    {
-      "epoch": 1.4711139739925119,
-      "grad_norm": 0.003210758324712515,
-      "learning_rate": 0.00019999893441550857,
-      "loss": 46.0,
-      "step": 9135
-    },
-    {
-      "epoch": 1.471275011071299,
-      "grad_norm": 0.012489925138652325,
-      "learning_rate": 0.0001999989341819434,
-      "loss": 46.0,
-      "step": 9136
-    },
-    {
-      "epoch": 1.4714360481500866,
-      "grad_norm": 0.0010342586319893599,
-      "learning_rate": 0.00019999893394835266,
-      "loss": 46.0,
-      "step": 9137
-    },
-    {
-      "epoch": 1.471597085228874,
-      "grad_norm": 0.0007231213967315853,
-      "learning_rate": 0.0001999989337147363,
-      "loss": 46.0,
-      "step": 9138
-    },
-    {
-      "epoch": 1.4717581223076612,
-      "grad_norm": 0.0007497454644180834,
-      "learning_rate": 0.00019999893348109443,
-      "loss": 46.0,
-      "step": 9139
-    },
-    {
-      "epoch": 1.4719191593864487,
-      "grad_norm": 0.0008579182904213667,
-      "learning_rate": 0.00019999893324742688,
-      "loss": 46.0,
-      "step": 9140
-    },
-    {
-      "epoch": 1.4720801964652361,
-      "grad_norm": 0.0013228532625362277,
-      "learning_rate": 0.00019999893301373377,
-      "loss": 46.0,
-      "step": 9141
-    },
-    {
-      "epoch": 1.4722412335440236,
-      "grad_norm": 0.000514134531840682,
-      "learning_rate": 0.00019999893278001504,
-      "loss": 46.0,
-      "step": 9142
-    },
-    {
-      "epoch": 1.472402270622811,
-      "grad_norm": 0.0011756445746868849,
-      "learning_rate": 0.00019999893254627073,
-      "loss": 46.0,
-      "step": 9143
-    },
-    {
-      "epoch": 1.4725633077015983,
-      "grad_norm": 0.0028722293209284544,
-      "learning_rate": 0.00019999893231250083,
-      "loss": 46.0,
-      "step": 9144
-    },
-    {
-      "epoch": 1.4727243447803857,
-      "grad_norm": 0.00224276608787477,
-      "learning_rate": 0.00019999893207870532,
-      "loss": 46.0,
-      "step": 9145
-    },
-    {
-      "epoch": 1.472885381859173,
-      "grad_norm": 0.0013697216054424644,
-      "learning_rate": 0.00019999893184488422,
-      "loss": 46.0,
-      "step": 9146
-    },
-    {
-      "epoch": 1.4730464189379604,
-      "grad_norm": 0.000897496473044157,
-      "learning_rate": 0.00019999893161103756,
-      "loss": 46.0,
-      "step": 9147
-    },
-    {
-      "epoch": 1.4732074560167479,
-      "grad_norm": 0.002812425373122096,
-      "learning_rate": 0.00019999893137716526,
-      "loss": 46.0,
-      "step": 9148
-    },
-    {
-      "epoch": 1.4733684930955353,
-      "grad_norm": 0.0022975506726652384,
-      "learning_rate": 0.00019999893114326737,
-      "loss": 46.0,
-      "step": 9149
-    },
-    {
-      "epoch": 1.4735295301743228,
-      "grad_norm": 0.0005632441025227308,
-      "learning_rate": 0.00019999893090934392,
-      "loss": 46.0,
-      "step": 9150
-    },
-    {
-      "epoch": 1.47369056725311,
-      "grad_norm": 0.0009142677881754935,
-      "learning_rate": 0.00019999893067539483,
-      "loss": 46.0,
-      "step": 9151
-    },
-    {
-      "epoch": 1.4738516043318974,
-      "grad_norm": 0.008548218756914139,
-      "learning_rate": 0.00019999893044142018,
-      "loss": 46.0,
-      "step": 9152
-    },
-    {
-      "epoch": 1.4740126414106849,
-      "grad_norm": 0.0013078803895041347,
-      "learning_rate": 0.00019999893020741994,
-      "loss": 46.0,
-      "step": 9153
-    },
-    {
-      "epoch": 1.4741736784894721,
-      "grad_norm": 0.0006266293930821121,
-      "learning_rate": 0.00019999892997339406,
-      "loss": 46.0,
-      "step": 9154
-    },
-    {
-      "epoch": 1.4743347155682596,
-      "grad_norm": 0.0020333644933998585,
-      "learning_rate": 0.00019999892973934262,
-      "loss": 46.0,
-      "step": 9155
-    },
-    {
-      "epoch": 1.474495752647047,
-      "grad_norm": 0.0019565399270504713,
-      "learning_rate": 0.0001999989295052656,
-      "loss": 46.0,
-      "step": 9156
-    },
-    {
-      "epoch": 1.4746567897258345,
-      "grad_norm": 0.0007945430115796626,
-      "learning_rate": 0.00019999892927116295,
-      "loss": 46.0,
-      "step": 9157
-    },
-    {
-      "epoch": 1.4748178268046217,
-      "grad_norm": 0.0004254174418747425,
-      "learning_rate": 0.00019999892903703473,
-      "loss": 46.0,
-      "step": 9158
-    },
-    {
-      "epoch": 1.4749788638834092,
-      "grad_norm": 0.000837479077745229,
-      "learning_rate": 0.0001999989288028809,
-      "loss": 46.0,
-      "step": 9159
-    },
-    {
-      "epoch": 1.4751399009621966,
-      "grad_norm": 0.0035503138788044453,
-      "learning_rate": 0.00019999892856870148,
-      "loss": 46.0,
-      "step": 9160
-    },
-    {
-      "epoch": 1.4753009380409838,
-      "grad_norm": 0.0038676303811371326,
-      "learning_rate": 0.00019999892833449647,
-      "loss": 46.0,
-      "step": 9161
-    },
-    {
-      "epoch": 1.4754619751197713,
-      "grad_norm": 0.0017247156938537955,
-      "learning_rate": 0.0001999989281002659,
-      "loss": 46.0,
-      "step": 9162
-    },
-    {
-      "epoch": 1.4756230121985587,
-      "grad_norm": 0.008360270410776138,
-      "learning_rate": 0.00019999892786600967,
-      "loss": 46.0,
-      "step": 9163
-    },
-    {
-      "epoch": 1.4757840492773462,
-      "grad_norm": 0.0021021724678575993,
-      "learning_rate": 0.00019999892763172787,
-      "loss": 46.0,
-      "step": 9164
-    },
-    {
-      "epoch": 1.4759450863561334,
-      "grad_norm": 0.00189265760127455,
-      "learning_rate": 0.00019999892739742047,
-      "loss": 46.0,
-      "step": 9165
-    },
-    {
-      "epoch": 1.4761061234349209,
-      "grad_norm": 0.0018198195612058043,
-      "learning_rate": 0.0001999989271630875,
-      "loss": 46.0,
-      "step": 9166
-    },
-    {
-      "epoch": 1.4762671605137083,
-      "grad_norm": 0.0013389589730650187,
-      "learning_rate": 0.00019999892692872893,
-      "loss": 46.0,
-      "step": 9167
-    },
-    {
-      "epoch": 1.4764281975924956,
-      "grad_norm": 0.0029416116885840893,
-      "learning_rate": 0.00019999892669434472,
-      "loss": 46.0,
-      "step": 9168
-    },
-    {
-      "epoch": 1.476589234671283,
-      "grad_norm": 0.002891369629651308,
-      "learning_rate": 0.00019999892645993498,
-      "loss": 46.0,
-      "step": 9169
-    },
-    {
-      "epoch": 1.4767502717500705,
-      "grad_norm": 0.000511622114572674,
-      "learning_rate": 0.0001999989262254996,
-      "loss": 46.0,
-      "step": 9170
-    },
-    {
-      "epoch": 1.476911308828858,
-      "grad_norm": 0.0022698314860463142,
-      "learning_rate": 0.00019999892599103865,
-      "loss": 46.0,
-      "step": 9171
-    },
-    {
-      "epoch": 1.4770723459076454,
-      "grad_norm": 0.0025655021890997887,
-      "learning_rate": 0.00019999892575655207,
-      "loss": 46.0,
-      "step": 9172
-    },
-    {
-      "epoch": 1.4772333829864326,
-      "grad_norm": 0.0019411066314205527,
-      "learning_rate": 0.00019999892552203993,
-      "loss": 46.0,
-      "step": 9173
-    },
-    {
-      "epoch": 1.47739442006522,
-      "grad_norm": 0.001997189363464713,
-      "learning_rate": 0.00019999892528750222,
-      "loss": 46.0,
-      "step": 9174
-    },
-    {
-      "epoch": 1.4775554571440075,
-      "grad_norm": 0.006019538268446922,
-      "learning_rate": 0.00019999892505293888,
-      "loss": 46.0,
-      "step": 9175
-    },
-    {
-      "epoch": 1.4777164942227947,
-      "grad_norm": 0.0023416599724441767,
-      "learning_rate": 0.00019999892481834994,
-      "loss": 46.0,
-      "step": 9176
-    },
-    {
-      "epoch": 1.4778775313015822,
-      "grad_norm": 0.0007697054534219205,
-      "learning_rate": 0.00019999892458373542,
-      "loss": 46.0,
-      "step": 9177
-    },
-    {
-      "epoch": 1.4780385683803696,
-      "grad_norm": 0.0018224541563540697,
-      "learning_rate": 0.0001999989243490953,
-      "loss": 46.0,
-      "step": 9178
-    },
-    {
-      "epoch": 1.478199605459157,
-      "grad_norm": 0.006273406092077494,
-      "learning_rate": 0.00019999892411442957,
-      "loss": 46.0,
-      "step": 9179
-    },
-    {
-      "epoch": 1.4783606425379443,
-      "grad_norm": 0.0025578478816896677,
-      "learning_rate": 0.00019999892387973826,
-      "loss": 46.0,
-      "step": 9180
-    },
-    {
-      "epoch": 1.4785216796167318,
-      "grad_norm": 0.0023115873336791992,
-      "learning_rate": 0.00019999892364502134,
-      "loss": 46.0,
-      "step": 9181
-    },
-    {
-      "epoch": 1.4786827166955192,
-      "grad_norm": 0.012280327267944813,
-      "learning_rate": 0.00019999892341027886,
-      "loss": 46.0,
-      "step": 9182
-    },
-    {
-      "epoch": 1.4788437537743064,
-      "grad_norm": 0.0037991516292095184,
-      "learning_rate": 0.00019999892317551076,
-      "loss": 46.0,
-      "step": 9183
-    },
-    {
-      "epoch": 1.4790047908530939,
-      "grad_norm": 0.0010416858131065965,
-      "learning_rate": 0.0001999989229407171,
-      "loss": 46.0,
-      "step": 9184
-    },
-    {
-      "epoch": 1.4791658279318813,
-      "grad_norm": 0.0033041746355593204,
-      "learning_rate": 0.0001999989227058978,
-      "loss": 46.0,
-      "step": 9185
-    },
-    {
-      "epoch": 1.4793268650106688,
-      "grad_norm": 0.0004504225216805935,
-      "learning_rate": 0.00019999892247105292,
-      "loss": 46.0,
-      "step": 9186
-    },
-    {
-      "epoch": 1.479487902089456,
-      "grad_norm": 0.004607070237398148,
-      "learning_rate": 0.00019999892223618245,
-      "loss": 46.0,
-      "step": 9187
-    },
-    {
-      "epoch": 1.4796489391682435,
-      "grad_norm": 0.011801824904978275,
-      "learning_rate": 0.00019999892200128639,
-      "loss": 46.0,
-      "step": 9188
-    },
-    {
-      "epoch": 1.479809976247031,
-      "grad_norm": 0.0008670894894748926,
-      "learning_rate": 0.0001999989217663647,
-      "loss": 46.0,
-      "step": 9189
-    },
-    {
-      "epoch": 1.4799710133258182,
-      "grad_norm": 0.004203254822641611,
-      "learning_rate": 0.00019999892153141745,
-      "loss": 46.0,
-      "step": 9190
-    },
-    {
-      "epoch": 1.4801320504046056,
-      "grad_norm": 0.0028932723216712475,
-      "learning_rate": 0.0001999989212964446,
-      "loss": 46.0,
-      "step": 9191
-    },
-    {
-      "epoch": 1.480293087483393,
-      "grad_norm": 0.001833435264416039,
-      "learning_rate": 0.00019999892106144617,
-      "loss": 46.0,
-      "step": 9192
-    },
-    {
-      "epoch": 1.4804541245621805,
-      "grad_norm": 0.0006322352564893663,
-      "learning_rate": 0.00019999892082642214,
-      "loss": 46.0,
-      "step": 9193
-    },
-    {
-      "epoch": 1.480615161640968,
-      "grad_norm": 0.006942402571439743,
-      "learning_rate": 0.0001999989205913725,
-      "loss": 46.0,
-      "step": 9194
-    },
-    {
-      "epoch": 1.4807761987197552,
-      "grad_norm": 0.00653650239109993,
-      "learning_rate": 0.00019999892035629726,
-      "loss": 46.0,
-      "step": 9195
-    },
-    {
-      "epoch": 1.4809372357985426,
-      "grad_norm": 0.0007525699911639094,
-      "learning_rate": 0.00019999892012119644,
-      "loss": 46.0,
-      "step": 9196
-    },
-    {
-      "epoch": 1.4810982728773299,
-      "grad_norm": 0.003730009775608778,
-      "learning_rate": 0.00019999891988607002,
-      "loss": 46.0,
-      "step": 9197
-    },
-    {
-      "epoch": 1.4812593099561173,
-      "grad_norm": 0.0032291864044964314,
-      "learning_rate": 0.000199998919650918,
-      "loss": 46.0,
-      "step": 9198
-    },
-    {
-      "epoch": 1.4814203470349048,
-      "grad_norm": 0.001430146861821413,
-      "learning_rate": 0.0001999989194157404,
-      "loss": 46.0,
-      "step": 9199
-    },
-    {
-      "epoch": 1.4815813841136922,
-      "grad_norm": 0.0004250580968800932,
-      "learning_rate": 0.0001999989191805372,
-      "loss": 46.0,
-      "step": 9200
-    },
-    {
-      "epoch": 1.4817424211924797,
-      "grad_norm": 0.0018537724390625954,
-      "learning_rate": 0.0001999989189453084,
-      "loss": 46.0,
-      "step": 9201
-    },
-    {
-      "epoch": 1.481903458271267,
-      "grad_norm": 0.0027568424120545387,
-      "learning_rate": 0.00019999891871005403,
-      "loss": 46.0,
-      "step": 9202
-    },
-    {
-      "epoch": 1.4820644953500544,
-      "grad_norm": 0.00115341623313725,
-      "learning_rate": 0.000199998918474774,
-      "loss": 46.0,
-      "step": 9203
-    },
-    {
-      "epoch": 1.4822255324288418,
-      "grad_norm": 0.0020698762964457273,
-      "learning_rate": 0.00019999891823946846,
-      "loss": 46.0,
-      "step": 9204
-    },
-    {
-      "epoch": 1.482386569507629,
-      "grad_norm": 0.000808591372333467,
-      "learning_rate": 0.00019999891800413728,
-      "loss": 46.0,
-      "step": 9205
-    },
-    {
-      "epoch": 1.4825476065864165,
-      "grad_norm": 0.004555118270218372,
-      "learning_rate": 0.0001999989177687805,
-      "loss": 46.0,
-      "step": 9206
-    },
-    {
-      "epoch": 1.482708643665204,
-      "grad_norm": 0.0008105576853267848,
-      "learning_rate": 0.00019999891753339814,
-      "loss": 46.0,
-      "step": 9207
-    },
-    {
-      "epoch": 1.4828696807439914,
-      "grad_norm": 0.0014980818377807736,
-      "learning_rate": 0.00019999891729799018,
-      "loss": 46.0,
-      "step": 9208
-    },
-    {
-      "epoch": 1.4830307178227786,
-      "grad_norm": 0.005529528018087149,
-      "learning_rate": 0.00019999891706255662,
-      "loss": 46.0,
-      "step": 9209
-    },
-    {
-      "epoch": 1.483191754901566,
-      "grad_norm": 0.003451485652476549,
-      "learning_rate": 0.00019999891682709748,
-      "loss": 46.0,
-      "step": 9210
-    },
-    {
-      "epoch": 1.4833527919803535,
-      "grad_norm": 0.0024196526501327753,
-      "learning_rate": 0.00019999891659161273,
-      "loss": 46.0,
-      "step": 9211
-    },
-    {
-      "epoch": 1.4835138290591408,
-      "grad_norm": 0.001738803111948073,
-      "learning_rate": 0.00019999891635610239,
-      "loss": 46.0,
-      "step": 9212
-    },
-    {
-      "epoch": 1.4836748661379282,
-      "grad_norm": 0.00106618064455688,
-      "learning_rate": 0.00019999891612056648,
-      "loss": 46.0,
-      "step": 9213
-    },
-    {
-      "epoch": 1.4838359032167157,
-      "grad_norm": 0.002034460660070181,
-      "learning_rate": 0.00019999891588500494,
-      "loss": 46.0,
-      "step": 9214
-    },
-    {
-      "epoch": 1.483996940295503,
-      "grad_norm": 0.004390411544591188,
-      "learning_rate": 0.0001999989156494178,
-      "loss": 46.0,
-      "step": 9215
-    },
-    {
-      "epoch": 1.4841579773742906,
-      "grad_norm": 0.0005137246917001903,
-      "learning_rate": 0.00019999891541380512,
-      "loss": 46.0,
-      "step": 9216
-    },
-    {
-      "epoch": 1.4843190144530778,
-      "grad_norm": 0.0015641130739822984,
-      "learning_rate": 0.0001999989151781668,
-      "loss": 46.0,
-      "step": 9217
-    },
-    {
-      "epoch": 1.4844800515318652,
-      "grad_norm": 0.005606240127235651,
-      "learning_rate": 0.0001999989149425029,
-      "loss": 46.0,
-      "step": 9218
-    },
-    {
-      "epoch": 1.4846410886106525,
-      "grad_norm": 0.0023736802395433187,
-      "learning_rate": 0.0001999989147068134,
-      "loss": 46.0,
-      "step": 9219
-    },
-    {
-      "epoch": 1.48480212568944,
-      "grad_norm": 0.0007125459378585219,
-      "learning_rate": 0.0001999989144710983,
-      "loss": 46.0,
-      "step": 9220
-    },
-    {
-      "epoch": 1.4849631627682274,
-      "grad_norm": 0.0022514360025525093,
-      "learning_rate": 0.00019999891423535762,
-      "loss": 46.0,
-      "step": 9221
-    },
-    {
-      "epoch": 1.4851241998470148,
-      "grad_norm": 0.003857055213302374,
-      "learning_rate": 0.00019999891399959135,
-      "loss": 46.0,
-      "step": 9222
-    },
-    {
-      "epoch": 1.4852852369258023,
-      "grad_norm": 0.005299419164657593,
-      "learning_rate": 0.00019999891376379947,
-      "loss": 46.0,
-      "step": 9223
-    },
-    {
-      "epoch": 1.4854462740045895,
-      "grad_norm": 0.005835393443703651,
-      "learning_rate": 0.00019999891352798197,
-      "loss": 46.0,
-      "step": 9224
-    },
-    {
-      "epoch": 1.485607311083377,
-      "grad_norm": 0.0017973246285691857,
-      "learning_rate": 0.00019999891329213892,
-      "loss": 46.0,
-      "step": 9225
-    },
-    {
-      "epoch": 1.4857683481621644,
-      "grad_norm": 0.0010201316326856613,
-      "learning_rate": 0.00019999891305627024,
-      "loss": 46.0,
-      "step": 9226
-    },
-    {
-      "epoch": 1.4859293852409516,
-      "grad_norm": 0.003794406307861209,
-      "learning_rate": 0.000199998912820376,
-      "loss": 46.0,
-      "step": 9227
-    },
-    {
-      "epoch": 1.486090422319739,
-      "grad_norm": 0.002419550670310855,
-      "learning_rate": 0.00019999891258445614,
-      "loss": 46.0,
-      "step": 9228
-    },
-    {
-      "epoch": 1.4862514593985265,
-      "grad_norm": 0.004375604912638664,
-      "learning_rate": 0.00019999891234851068,
-      "loss": 46.0,
-      "step": 9229
-    },
-    {
-      "epoch": 1.486412496477314,
-      "grad_norm": 0.0053855255246162415,
-      "learning_rate": 0.00019999891211253966,
-      "loss": 46.0,
-      "step": 9230
-    },
-    {
-      "epoch": 1.4865735335561012,
-      "grad_norm": 0.0018320202361792326,
-      "learning_rate": 0.00019999891187654303,
-      "loss": 46.0,
-      "step": 9231
-    },
-    {
-      "epoch": 1.4867345706348887,
-      "grad_norm": 0.0006352829514071345,
-      "learning_rate": 0.0001999989116405208,
-      "loss": 46.0,
-      "step": 9232
-    },
-    {
-      "epoch": 1.4868956077136761,
-      "grad_norm": 0.0026084398850798607,
-      "learning_rate": 0.00019999891140447297,
-      "loss": 46.0,
-      "step": 9233
-    },
-    {
-      "epoch": 1.4870566447924634,
-      "grad_norm": 0.0022584833204746246,
-      "learning_rate": 0.00019999891116839958,
-      "loss": 46.0,
-      "step": 9234
-    },
-    {
-      "epoch": 1.4872176818712508,
-      "grad_norm": 0.0017240295419469476,
-      "learning_rate": 0.00019999891093230054,
-      "loss": 46.0,
-      "step": 9235
-    },
-    {
-      "epoch": 1.4873787189500383,
-      "grad_norm": 0.006388253066688776,
-      "learning_rate": 0.00019999891069617595,
-      "loss": 46.0,
-      "step": 9236
-    },
-    {
-      "epoch": 1.4875397560288257,
-      "grad_norm": 0.0011752115096896887,
-      "learning_rate": 0.00019999891046002574,
-      "loss": 46.0,
-      "step": 9237
-    },
-    {
-      "epoch": 1.4877007931076132,
-      "grad_norm": 0.0032273733522742987,
-      "learning_rate": 0.00019999891022384994,
-      "loss": 46.0,
-      "step": 9238
-    },
-    {
-      "epoch": 1.4878618301864004,
-      "grad_norm": 0.0007667818572372198,
-      "learning_rate": 0.00019999890998764855,
-      "loss": 46.0,
-      "step": 9239
-    },
-    {
-      "epoch": 1.4880228672651878,
-      "grad_norm": 0.006122061051428318,
-      "learning_rate": 0.00019999890975142158,
-      "loss": 46.0,
-      "step": 9240
-    },
-    {
-      "epoch": 1.488183904343975,
-      "grad_norm": 0.0007807639776729047,
-      "learning_rate": 0.000199998909515169,
-      "loss": 46.0,
-      "step": 9241
-    },
-    {
-      "epoch": 1.4883449414227625,
-      "grad_norm": 0.0017634417163208127,
-      "learning_rate": 0.0001999989092788908,
-      "loss": 46.0,
-      "step": 9242
-    },
-    {
-      "epoch": 1.48850597850155,
-      "grad_norm": 0.0021173693239688873,
-      "learning_rate": 0.00019999890904258703,
-      "loss": 46.0,
-      "step": 9243
-    },
-    {
-      "epoch": 1.4886670155803374,
-      "grad_norm": 0.0014326149830594659,
-      "learning_rate": 0.00019999890880625766,
-      "loss": 46.0,
-      "step": 9244
-    },
-    {
-      "epoch": 1.4888280526591249,
-      "grad_norm": 0.002000546781346202,
-      "learning_rate": 0.00019999890856990272,
-      "loss": 46.0,
-      "step": 9245
-    },
-    {
-      "epoch": 1.488989089737912,
-      "grad_norm": 0.004835196305066347,
-      "learning_rate": 0.00019999890833352217,
-      "loss": 46.0,
-      "step": 9246
-    },
-    {
-      "epoch": 1.4891501268166996,
-      "grad_norm": 0.0020897649228572845,
-      "learning_rate": 0.000199998908097116,
-      "loss": 46.0,
-      "step": 9247
-    },
-    {
-      "epoch": 1.489311163895487,
-      "grad_norm": 0.0033091893419623375,
-      "learning_rate": 0.00019999890786068426,
-      "loss": 46.0,
-      "step": 9248
-    },
-    {
-      "epoch": 1.4894722009742742,
-      "grad_norm": 0.0021526850759983063,
-      "learning_rate": 0.00019999890762422692,
-      "loss": 46.0,
-      "step": 9249
-    },
-    {
-      "epoch": 1.4896332380530617,
-      "grad_norm": 0.002141405828297138,
-      "learning_rate": 0.000199998907387744,
-      "loss": 46.0,
-      "step": 9250
-    },
-    {
-      "epoch": 1.4897942751318491,
-      "grad_norm": 0.0016003063647076488,
-      "learning_rate": 0.00019999890715123548,
-      "loss": 46.0,
-      "step": 9251
-    },
-    {
-      "epoch": 1.4899553122106366,
-      "grad_norm": 0.004432353191077709,
-      "learning_rate": 0.00019999890691470136,
-      "loss": 46.0,
-      "step": 9252
-    },
-    {
-      "epoch": 1.4901163492894238,
-      "grad_norm": 0.0035880128853023052,
-      "learning_rate": 0.00019999890667814164,
-      "loss": 46.0,
-      "step": 9253
-    },
-    {
-      "epoch": 1.4902773863682113,
-      "grad_norm": 0.004421964753419161,
-      "learning_rate": 0.00019999890644155634,
-      "loss": 46.0,
-      "step": 9254
-    },
-    {
-      "epoch": 1.4904384234469987,
-      "grad_norm": 0.0006389126647263765,
-      "learning_rate": 0.00019999890620494543,
-      "loss": 46.0,
-      "step": 9255
-    },
-    {
-      "epoch": 1.490599460525786,
-      "grad_norm": 0.009319878183305264,
-      "learning_rate": 0.00019999890596830895,
-      "loss": 46.0,
-      "step": 9256
-    },
-    {
-      "epoch": 1.4907604976045734,
-      "grad_norm": 0.002830260433256626,
-      "learning_rate": 0.00019999890573164683,
-      "loss": 46.0,
-      "step": 9257
-    },
-    {
-      "epoch": 1.4909215346833609,
-      "grad_norm": 0.006291674915701151,
-      "learning_rate": 0.00019999890549495916,
-      "loss": 46.0,
-      "step": 9258
-    },
-    {
-      "epoch": 1.4910825717621483,
-      "grad_norm": 0.003054615342989564,
-      "learning_rate": 0.00019999890525824587,
-      "loss": 46.0,
-      "step": 9259
-    },
-    {
-      "epoch": 1.4912436088409358,
-      "grad_norm": 0.0017148348269984126,
-      "learning_rate": 0.000199998905021507,
-      "loss": 46.0,
-      "step": 9260
-    },
-    {
-      "epoch": 1.491404645919723,
-      "grad_norm": 0.0015248499112203717,
-      "learning_rate": 0.0001999989047847425,
-      "loss": 46.0,
-      "step": 9261
-    },
-    {
-      "epoch": 1.4915656829985104,
-      "grad_norm": 0.0018423888832330704,
-      "learning_rate": 0.00019999890454795244,
-      "loss": 46.0,
-      "step": 9262
-    },
-    {
-      "epoch": 1.4917267200772977,
-      "grad_norm": 0.0065639144740998745,
-      "learning_rate": 0.0001999989043111368,
-      "loss": 46.0,
-      "step": 9263
-    },
-    {
-      "epoch": 1.4918877571560851,
-      "grad_norm": 0.0037312640342861414,
-      "learning_rate": 0.00019999890407429552,
-      "loss": 46.0,
-      "step": 9264
-    },
-    {
-      "epoch": 1.4920487942348726,
-      "grad_norm": 0.003073142608627677,
-      "learning_rate": 0.00019999890383742868,
-      "loss": 46.0,
-      "step": 9265
-    },
-    {
-      "epoch": 1.49220983131366,
-      "grad_norm": 0.001604669727385044,
-      "learning_rate": 0.00019999890360053622,
-      "loss": 46.0,
-      "step": 9266
-    },
-    {
-      "epoch": 1.4923708683924475,
-      "grad_norm": 0.003724227659404278,
-      "learning_rate": 0.0001999989033636182,
-      "loss": 46.0,
-      "step": 9267
-    },
-    {
-      "epoch": 1.4925319054712347,
-      "grad_norm": 0.00258837197907269,
-      "learning_rate": 0.00019999890312667455,
-      "loss": 46.0,
-      "step": 9268
-    },
-    {
-      "epoch": 1.4926929425500222,
-      "grad_norm": 0.0008089937036857009,
-      "learning_rate": 0.00019999890288970536,
-      "loss": 46.0,
-      "step": 9269
-    },
-    {
-      "epoch": 1.4928539796288096,
-      "grad_norm": 0.0025259137619286776,
-      "learning_rate": 0.00019999890265271053,
-      "loss": 46.0,
-      "step": 9270
-    },
-    {
-      "epoch": 1.4930150167075968,
-      "grad_norm": 0.0022156857885420322,
-      "learning_rate": 0.00019999890241569008,
-      "loss": 46.0,
-      "step": 9271
-    },
-    {
-      "epoch": 1.4931760537863843,
-      "grad_norm": 0.0041998522356152534,
-      "learning_rate": 0.0001999989021786441,
-      "loss": 46.0,
-      "step": 9272
-    },
-    {
-      "epoch": 1.4933370908651717,
-      "grad_norm": 0.001053009764291346,
-      "learning_rate": 0.00019999890194157249,
-      "loss": 46.0,
-      "step": 9273
-    },
-    {
-      "epoch": 1.4934981279439592,
-      "grad_norm": 0.0009771655313670635,
-      "learning_rate": 0.00019999890170447525,
-      "loss": 46.0,
-      "step": 9274
-    },
-    {
-      "epoch": 1.4936591650227464,
-      "grad_norm": 0.00158273137640208,
-      "learning_rate": 0.00019999890146735249,
-      "loss": 46.0,
-      "step": 9275
-    },
-    {
-      "epoch": 1.4938202021015339,
-      "grad_norm": 0.0026314619462937117,
-      "learning_rate": 0.0001999989012302041,
-      "loss": 46.0,
-      "step": 9276
-    },
-    {
-      "epoch": 1.4939812391803213,
-      "grad_norm": 0.001456606201827526,
-      "learning_rate": 0.00019999890099303008,
-      "loss": 46.0,
-      "step": 9277
-    },
-    {
-      "epoch": 1.4941422762591086,
-      "grad_norm": 0.0006253541796468198,
-      "learning_rate": 0.00019999890075583052,
-      "loss": 46.0,
-      "step": 9278
-    },
-    {
-      "epoch": 1.494303313337896,
-      "grad_norm": 0.0005209855735301971,
-      "learning_rate": 0.00019999890051860535,
-      "loss": 46.0,
-      "step": 9279
-    },
-    {
-      "epoch": 1.4944643504166835,
-      "grad_norm": 0.005019493866711855,
-      "learning_rate": 0.00019999890028135457,
-      "loss": 46.0,
-      "step": 9280
-    },
-    {
-      "epoch": 1.494625387495471,
-      "grad_norm": 0.0013266053283587098,
-      "learning_rate": 0.00019999890004407823,
-      "loss": 46.0,
-      "step": 9281
-    },
-    {
-      "epoch": 1.4947864245742581,
-      "grad_norm": 0.002484564669430256,
-      "learning_rate": 0.00019999889980677627,
-      "loss": 46.0,
-      "step": 9282
-    },
-    {
-      "epoch": 1.4949474616530456,
-      "grad_norm": 0.0008026795694604516,
-      "learning_rate": 0.0001999988995694487,
-      "loss": 46.0,
-      "step": 9283
-    },
-    {
-      "epoch": 1.495108498731833,
-      "grad_norm": 0.002022720407694578,
-      "learning_rate": 0.00019999889933209553,
-      "loss": 46.0,
-      "step": 9284
-    },
-    {
-      "epoch": 1.4952695358106203,
-      "grad_norm": 0.0013786606723442674,
-      "learning_rate": 0.00019999889909471681,
-      "loss": 46.0,
-      "step": 9285
-    },
-    {
-      "epoch": 1.4954305728894077,
-      "grad_norm": 0.001019939430989325,
-      "learning_rate": 0.00019999889885731248,
-      "loss": 46.0,
-      "step": 9286
-    },
-    {
-      "epoch": 1.4955916099681952,
-      "grad_norm": 0.007120917085558176,
-      "learning_rate": 0.00019999889861988256,
-      "loss": 46.0,
-      "step": 9287
-    },
-    {
-      "epoch": 1.4957526470469826,
-      "grad_norm": 0.0011648134095594287,
-      "learning_rate": 0.00019999889838242702,
-      "loss": 46.0,
-      "step": 9288
-    },
-    {
-      "epoch": 1.49591368412577,
-      "grad_norm": 0.0012876932742074132,
-      "learning_rate": 0.0001999988981449459,
-      "loss": 46.0,
-      "step": 9289
-    },
-    {
-      "epoch": 1.4960747212045573,
-      "grad_norm": 0.0029910027515143156,
-      "learning_rate": 0.0001999988979074392,
-      "loss": 46.0,
-      "step": 9290
-    },
-    {
-      "epoch": 1.4962357582833448,
-      "grad_norm": 0.004170332103967667,
-      "learning_rate": 0.0001999988976699069,
-      "loss": 46.0,
-      "step": 9291
-    },
-    {
-      "epoch": 1.496396795362132,
-      "grad_norm": 0.0058249616995453835,
-      "learning_rate": 0.00019999889743234895,
-      "loss": 46.0,
-      "step": 9292
-    },
-    {
-      "epoch": 1.4965578324409194,
-      "grad_norm": 0.002981281839311123,
-      "learning_rate": 0.00019999889719476548,
-      "loss": 46.0,
-      "step": 9293
-    },
-    {
-      "epoch": 1.496718869519707,
-      "grad_norm": 0.0033790236338973045,
-      "learning_rate": 0.00019999889695715637,
-      "loss": 46.0,
-      "step": 9294
-    },
-    {
-      "epoch": 1.4968799065984943,
-      "grad_norm": 0.0013915090821683407,
-      "learning_rate": 0.0001999988967195217,
-      "loss": 46.0,
-      "step": 9295
-    },
-    {
-      "epoch": 1.4970409436772818,
-      "grad_norm": 0.0018759649246931076,
-      "learning_rate": 0.0001999988964818614,
-      "loss": 46.0,
-      "step": 9296
-    },
-    {
-      "epoch": 1.497201980756069,
-      "grad_norm": 0.0022674081847071648,
-      "learning_rate": 0.00019999889624417553,
-      "loss": 46.0,
-      "step": 9297
-    },
-    {
-      "epoch": 1.4973630178348565,
-      "grad_norm": 0.002209196100011468,
-      "learning_rate": 0.00019999889600646407,
-      "loss": 46.0,
-      "step": 9298
-    },
-    {
-      "epoch": 1.497524054913644,
-      "grad_norm": 0.0026589909102767706,
-      "learning_rate": 0.00019999889576872702,
-      "loss": 46.0,
-      "step": 9299
-    },
-    {
-      "epoch": 1.4976850919924312,
-      "grad_norm": 0.006607708986848593,
-      "learning_rate": 0.00019999889553096433,
-      "loss": 46.0,
-      "step": 9300
-    },
-    {
-      "epoch": 1.4978461290712186,
-      "grad_norm": 0.0015110252425074577,
-      "learning_rate": 0.00019999889529317608,
-      "loss": 46.0,
-      "step": 9301
-    },
-    {
-      "epoch": 1.498007166150006,
-      "grad_norm": 0.00298205460421741,
-      "learning_rate": 0.00019999889505536222,
-      "loss": 46.0,
-      "step": 9302
-    },
-    {
-      "epoch": 1.4981682032287935,
-      "grad_norm": 0.003964473493397236,
-      "learning_rate": 0.0001999988948175228,
-      "loss": 46.0,
-      "step": 9303
-    },
-    {
-      "epoch": 1.4983292403075807,
-      "grad_norm": 0.002038866514340043,
-      "learning_rate": 0.00019999889457965775,
-      "loss": 46.0,
-      "step": 9304
-    },
-    {
-      "epoch": 1.4984902773863682,
-      "grad_norm": 0.0035502715036273003,
-      "learning_rate": 0.00019999889434176713,
-      "loss": 46.0,
-      "step": 9305
-    },
-    {
-      "epoch": 1.4986513144651556,
-      "grad_norm": 0.0007741302833892405,
-      "learning_rate": 0.00019999889410385089,
-      "loss": 46.0,
-      "step": 9306
-    },
-    {
-      "epoch": 1.4988123515439429,
-      "grad_norm": 0.0036379615776240826,
-      "learning_rate": 0.00019999889386590906,
-      "loss": 46.0,
-      "step": 9307
-    },
-    {
-      "epoch": 1.4989733886227303,
-      "grad_norm": 0.001970453653484583,
-      "learning_rate": 0.00019999889362794164,
-      "loss": 46.0,
-      "step": 9308
-    },
-    {
-      "epoch": 1.4991344257015178,
-      "grad_norm": 0.004272512160241604,
-      "learning_rate": 0.00019999889338994864,
-      "loss": 46.0,
-      "step": 9309
-    },
-    {
-      "epoch": 1.4992954627803052,
-      "grad_norm": 0.0005225231288932264,
-      "learning_rate": 0.00019999889315193002,
-      "loss": 46.0,
-      "step": 9310
-    },
-    {
-      "epoch": 1.4994564998590927,
-      "grad_norm": 0.003577888011932373,
-      "learning_rate": 0.00019999889291388582,
-      "loss": 46.0,
-      "step": 9311
-    },
-    {
-      "epoch": 1.49961753693788,
-      "grad_norm": 0.0007586209685541689,
-      "learning_rate": 0.00019999889267581603,
-      "loss": 46.0,
-      "step": 9312
-    },
-    {
-      "epoch": 1.4997785740166674,
-      "grad_norm": 0.0017913278425112367,
-      "learning_rate": 0.00019999889243772065,
-      "loss": 46.0,
-      "step": 9313
-    },
-    {
-      "epoch": 1.4999396110954546,
-      "grad_norm": 0.0008608452044427395,
-      "learning_rate": 0.00019999889219959966,
-      "loss": 46.0,
-      "step": 9314
-    },
-    {
-      "epoch": 1.500100648174242,
-      "grad_norm": 0.0011282166233286262,
-      "learning_rate": 0.00019999889196145308,
-      "loss": 46.0,
-      "step": 9315
-    },
-    {
-      "epoch": 1.5002616852530295,
-      "grad_norm": 0.003429776756092906,
-      "learning_rate": 0.00019999889172328091,
-      "loss": 46.0,
-      "step": 9316
-    },
-    {
-      "epoch": 1.500422722331817,
-      "grad_norm": 0.0012224511010572314,
-      "learning_rate": 0.00019999889148508313,
-      "loss": 46.0,
-      "step": 9317
-    },
-    {
-      "epoch": 1.5005837594106044,
-      "grad_norm": 0.0031226149294525385,
-      "learning_rate": 0.00019999889124685977,
-      "loss": 46.0,
-      "step": 9318
-    },
-    {
-      "epoch": 1.5007447964893916,
-      "grad_norm": 0.0020698532462120056,
-      "learning_rate": 0.0001999988910086108,
-      "loss": 46.0,
-      "step": 9319
-    },
-    {
-      "epoch": 1.500905833568179,
-      "grad_norm": 0.013390767388045788,
-      "learning_rate": 0.00019999889077033624,
-      "loss": 46.0,
-      "step": 9320
-    },
-    {
-      "epoch": 1.5010668706469663,
-      "grad_norm": 0.002201109193265438,
-      "learning_rate": 0.0001999988905320361,
-      "loss": 46.0,
-      "step": 9321
-    },
-    {
-      "epoch": 1.5012279077257538,
-      "grad_norm": 0.0023974617943167686,
-      "learning_rate": 0.00019999889029371037,
-      "loss": 46.0,
-      "step": 9322
-    },
-    {
-      "epoch": 1.5013889448045412,
-      "grad_norm": 0.004037914797663689,
-      "learning_rate": 0.00019999889005535904,
-      "loss": 46.0,
-      "step": 9323
-    },
-    {
-      "epoch": 1.5015499818833287,
-      "grad_norm": 0.00031237577786669135,
-      "learning_rate": 0.0001999988898169821,
-      "loss": 46.0,
-      "step": 9324
-    },
-    {
-      "epoch": 1.5017110189621161,
-      "grad_norm": 0.0017866770504042506,
-      "learning_rate": 0.0001999988895785796,
-      "loss": 46.0,
-      "step": 9325
-    },
-    {
-      "epoch": 1.5018720560409036,
-      "grad_norm": 0.0011356729082763195,
-      "learning_rate": 0.00019999888934015147,
-      "loss": 46.0,
-      "step": 9326
-    },
-    {
-      "epoch": 1.5020330931196908,
-      "grad_norm": 0.004935161676257849,
-      "learning_rate": 0.00019999888910169773,
-      "loss": 46.0,
-      "step": 9327
-    },
-    {
-      "epoch": 1.5021941301984783,
-      "grad_norm": 0.0003700161469168961,
-      "learning_rate": 0.00019999888886321844,
-      "loss": 46.0,
-      "step": 9328
-    },
-    {
-      "epoch": 1.5023551672772655,
-      "grad_norm": 0.000601876643486321,
-      "learning_rate": 0.00019999888862471353,
-      "loss": 46.0,
-      "step": 9329
-    },
-    {
-      "epoch": 1.502516204356053,
-      "grad_norm": 0.001284827827475965,
-      "learning_rate": 0.00019999888838618304,
-      "loss": 46.0,
-      "step": 9330
-    },
-    {
-      "epoch": 1.5026772414348404,
-      "grad_norm": 0.0016494819428771734,
-      "learning_rate": 0.00019999888814762695,
-      "loss": 46.0,
-      "step": 9331
-    },
-    {
-      "epoch": 1.5028382785136278,
-      "grad_norm": 0.004624811466783285,
-      "learning_rate": 0.00019999888790904526,
-      "loss": 46.0,
-      "step": 9332
-    },
-    {
-      "epoch": 1.5029993155924153,
-      "grad_norm": 0.0026612093206495047,
-      "learning_rate": 0.00019999888767043797,
-      "loss": 46.0,
-      "step": 9333
-    },
-    {
-      "epoch": 1.5031603526712025,
-      "grad_norm": 0.0004504411481320858,
-      "learning_rate": 0.0001999988874318051,
-      "loss": 46.0,
-      "step": 9334
-    },
-    {
-      "epoch": 1.50332138974999,
-      "grad_norm": 0.0011903643608093262,
-      "learning_rate": 0.00019999888719314664,
-      "loss": 46.0,
-      "step": 9335
-    },
-    {
-      "epoch": 1.5034824268287772,
-      "grad_norm": 0.0015241664368659258,
-      "learning_rate": 0.00019999888695446257,
-      "loss": 46.0,
-      "step": 9336
-    },
-    {
-      "epoch": 1.5036434639075646,
-      "grad_norm": 0.00607029115781188,
-      "learning_rate": 0.0001999988867157529,
-      "loss": 46.0,
-      "step": 9337
-    },
-    {
-      "epoch": 1.503804500986352,
-      "grad_norm": 0.0006795867811888456,
-      "learning_rate": 0.00019999888647701767,
-      "loss": 46.0,
-      "step": 9338
-    },
-    {
-      "epoch": 1.5039655380651396,
-      "grad_norm": 0.001204039203003049,
-      "learning_rate": 0.00019999888623825678,
-      "loss": 46.0,
-      "step": 9339
-    },
-    {
-      "epoch": 1.504126575143927,
-      "grad_norm": 0.005060479044914246,
-      "learning_rate": 0.00019999888599947036,
-      "loss": 46.0,
-      "step": 9340
-    },
-    {
-      "epoch": 1.5042876122227142,
-      "grad_norm": 0.002776928711682558,
-      "learning_rate": 0.0001999988857606583,
-      "loss": 46.0,
-      "step": 9341
-    },
-    {
-      "epoch": 1.5044486493015017,
-      "grad_norm": 0.0007243006257340312,
-      "learning_rate": 0.00019999888552182067,
-      "loss": 46.0,
-      "step": 9342
-    },
-    {
-      "epoch": 1.504609686380289,
-      "grad_norm": 0.0010960230138152838,
-      "learning_rate": 0.00019999888528295746,
-      "loss": 46.0,
-      "step": 9343
-    },
-    {
-      "epoch": 1.5047707234590764,
-      "grad_norm": 0.005405620206147432,
-      "learning_rate": 0.00019999888504406864,
-      "loss": 46.0,
-      "step": 9344
-    },
-    {
-      "epoch": 1.5049317605378638,
-      "grad_norm": 0.003555117640644312,
-      "learning_rate": 0.0001999988848051542,
-      "loss": 46.0,
-      "step": 9345
-    },
-    {
-      "epoch": 1.5050927976166513,
-      "grad_norm": 0.0032495323102921247,
-      "learning_rate": 0.00019999888456621417,
-      "loss": 46.0,
-      "step": 9346
-    },
-    {
-      "epoch": 1.5052538346954387,
-      "grad_norm": 0.002382584847509861,
-      "learning_rate": 0.0001999988843272486,
-      "loss": 46.0,
-      "step": 9347
-    },
-    {
-      "epoch": 1.5054148717742262,
-      "grad_norm": 0.000834115722682327,
-      "learning_rate": 0.0001999988840882574,
-      "loss": 46.0,
-      "step": 9348
-    },
-    {
-      "epoch": 1.5055759088530134,
-      "grad_norm": 0.002534291008487344,
-      "learning_rate": 0.0001999988838492406,
-      "loss": 46.0,
-      "step": 9349
-    },
-    {
-      "epoch": 1.5057369459318009,
-      "grad_norm": 0.003397679189220071,
-      "learning_rate": 0.0001999988836101982,
-      "loss": 46.0,
-      "step": 9350
-    },
-    {
-      "epoch": 1.505897983010588,
-      "grad_norm": 0.0018755896016955376,
-      "learning_rate": 0.0001999988833711302,
-      "loss": 46.0,
-      "step": 9351
-    },
-    {
-      "epoch": 1.5060590200893755,
-      "grad_norm": 0.0029869726859033108,
-      "learning_rate": 0.00019999888313203666,
-      "loss": 46.0,
-      "step": 9352
-    },
-    {
-      "epoch": 1.506220057168163,
-      "grad_norm": 0.002450810046866536,
-      "learning_rate": 0.00019999888289291747,
-      "loss": 46.0,
-      "step": 9353
-    },
-    {
-      "epoch": 1.5063810942469504,
-      "grad_norm": 0.0018947371281683445,
-      "learning_rate": 0.0001999988826537727,
-      "loss": 46.0,
-      "step": 9354
-    },
-    {
-      "epoch": 1.5065421313257379,
-      "grad_norm": 0.009498589672148228,
-      "learning_rate": 0.00019999888241460233,
-      "loss": 46.0,
-      "step": 9355
-    },
-    {
-      "epoch": 1.5067031684045251,
-      "grad_norm": 0.002005457179620862,
-      "learning_rate": 0.00019999888217540638,
-      "loss": 46.0,
-      "step": 9356
-    },
-    {
-      "epoch": 1.5068642054833126,
-      "grad_norm": 0.008147171698510647,
-      "learning_rate": 0.00019999888193618484,
-      "loss": 46.0,
-      "step": 9357
-    },
-    {
-      "epoch": 1.5070252425620998,
-      "grad_norm": 0.0025351904332637787,
-      "learning_rate": 0.0001999988816969377,
-      "loss": 46.0,
-      "step": 9358
-    },
-    {
-      "epoch": 1.5071862796408872,
-      "grad_norm": 0.003502566833049059,
-      "learning_rate": 0.00019999888145766495,
-      "loss": 46.0,
-      "step": 9359
-    },
-    {
-      "epoch": 1.5073473167196747,
-      "grad_norm": 0.0014478863449767232,
-      "learning_rate": 0.00019999888121836662,
-      "loss": 46.0,
-      "step": 9360
-    },
-    {
-      "epoch": 1.5075083537984622,
-      "grad_norm": 0.00041848496766760945,
-      "learning_rate": 0.00019999888097904268,
-      "loss": 46.0,
-      "step": 9361
-    },
-    {
-      "epoch": 1.5076693908772496,
-      "grad_norm": 0.004982709884643555,
-      "learning_rate": 0.00019999888073969315,
-      "loss": 46.0,
-      "step": 9362
-    },
-    {
-      "epoch": 1.5078304279560368,
-      "grad_norm": 0.0007134603220038116,
-      "learning_rate": 0.00019999888050031803,
-      "loss": 46.0,
-      "step": 9363
-    },
-    {
-      "epoch": 1.5079914650348243,
-      "grad_norm": 0.005868175532668829,
-      "learning_rate": 0.0001999988802609173,
-      "loss": 46.0,
-      "step": 9364
-    },
-    {
-      "epoch": 1.5081525021136115,
-      "grad_norm": 0.0010554182808846235,
-      "learning_rate": 0.00019999888002149101,
-      "loss": 46.0,
-      "step": 9365
-    },
-    {
-      "epoch": 1.508313539192399,
-      "grad_norm": 0.003978913649916649,
-      "learning_rate": 0.0001999988797820391,
-      "loss": 46.0,
-      "step": 9366
-    },
-    {
-      "epoch": 1.5084745762711864,
-      "grad_norm": 0.0032237533014267683,
-      "learning_rate": 0.0001999988795425616,
-      "loss": 46.0,
-      "step": 9367
-    },
-    {
-      "epoch": 1.5086356133499739,
-      "grad_norm": 0.004461077507585287,
-      "learning_rate": 0.00019999887930305848,
-      "loss": 46.0,
-      "step": 9368
-    },
-    {
-      "epoch": 1.5087966504287613,
-      "grad_norm": 0.0006379224942065775,
-      "learning_rate": 0.00019999887906352982,
-      "loss": 46.0,
-      "step": 9369
-    },
-    {
-      "epoch": 1.5089576875075488,
-      "grad_norm": 0.0012650849530473351,
-      "learning_rate": 0.00019999887882397554,
-      "loss": 46.0,
-      "step": 9370
-    },
-    {
-      "epoch": 1.509118724586336,
-      "grad_norm": 0.0005488916649483144,
-      "learning_rate": 0.00019999887858439567,
-      "loss": 46.0,
-      "step": 9371
-    },
-    {
-      "epoch": 1.5092797616651232,
-      "grad_norm": 0.001251340378075838,
-      "learning_rate": 0.0001999988783447902,
-      "loss": 46.0,
-      "step": 9372
-    },
-    {
-      "epoch": 1.5094407987439107,
-      "grad_norm": 0.0017519919201731682,
-      "learning_rate": 0.00019999887810515912,
-      "loss": 46.0,
-      "step": 9373
-    },
-    {
-      "epoch": 1.5096018358226981,
-      "grad_norm": 0.004199097398668528,
-      "learning_rate": 0.00019999887786550246,
-      "loss": 46.0,
-      "step": 9374
-    },
-    {
-      "epoch": 1.5097628729014856,
-      "grad_norm": 0.00195657298900187,
-      "learning_rate": 0.00019999887762582022,
-      "loss": 46.0,
-      "step": 9375
-    },
-    {
-      "epoch": 1.509923909980273,
-      "grad_norm": 0.0005312208668328822,
-      "learning_rate": 0.0001999988773861124,
-      "loss": 46.0,
-      "step": 9376
-    },
-    {
-      "epoch": 1.5100849470590605,
-      "grad_norm": 0.0004765833728015423,
-      "learning_rate": 0.00019999887714637892,
-      "loss": 46.0,
-      "step": 9377
-    },
-    {
-      "epoch": 1.5102459841378477,
-      "grad_norm": 0.0007699292618781328,
-      "learning_rate": 0.0001999988769066199,
-      "loss": 46.0,
-      "step": 9378
-    },
-    {
-      "epoch": 1.5104070212166352,
-      "grad_norm": 0.0006723885308019817,
-      "learning_rate": 0.00019999887666683527,
-      "loss": 46.0,
-      "step": 9379
-    },
-    {
-      "epoch": 1.5105680582954224,
-      "grad_norm": 0.0016152807511389256,
-      "learning_rate": 0.00019999887642702503,
-      "loss": 46.0,
-      "step": 9380
-    },
-    {
-      "epoch": 1.5107290953742099,
-      "grad_norm": 0.003390349680557847,
-      "learning_rate": 0.0001999988761871892,
-      "loss": 46.0,
-      "step": 9381
-    },
-    {
-      "epoch": 1.5108901324529973,
-      "grad_norm": 0.0036467250902205706,
-      "learning_rate": 0.00019999887594732778,
-      "loss": 46.0,
-      "step": 9382
-    },
-    {
-      "epoch": 1.5110511695317848,
-      "grad_norm": 0.0031857339199632406,
-      "learning_rate": 0.00019999887570744078,
-      "loss": 46.0,
-      "step": 9383
-    },
-    {
-      "epoch": 1.5112122066105722,
-      "grad_norm": 0.0025473847053945065,
-      "learning_rate": 0.00019999887546752815,
-      "loss": 46.0,
-      "step": 9384
-    },
-    {
-      "epoch": 1.5113732436893594,
-      "grad_norm": 0.0004909922135993838,
-      "learning_rate": 0.00019999887522758995,
-      "loss": 46.0,
-      "step": 9385
-    },
-    {
-      "epoch": 1.5115342807681469,
-      "grad_norm": 0.0008730872650630772,
-      "learning_rate": 0.00019999887498762617,
-      "loss": 46.0,
-      "step": 9386
-    },
-    {
-      "epoch": 1.5116953178469341,
-      "grad_norm": 0.0008864441188052297,
-      "learning_rate": 0.0001999988747476368,
-      "loss": 46.0,
-      "step": 9387
-    },
-    {
-      "epoch": 1.5118563549257216,
-      "grad_norm": 0.0012393741635605693,
-      "learning_rate": 0.0001999988745076218,
-      "loss": 46.0,
-      "step": 9388
-    },
-    {
-      "epoch": 1.512017392004509,
-      "grad_norm": 0.0033559727016836405,
-      "learning_rate": 0.00019999887426758121,
-      "loss": 46.0,
-      "step": 9389
-    },
-    {
-      "epoch": 1.5121784290832965,
-      "grad_norm": 0.004136817064136267,
-      "learning_rate": 0.00019999887402751505,
-      "loss": 46.0,
-      "step": 9390
-    },
-    {
-      "epoch": 1.512339466162084,
-      "grad_norm": 0.006455062422901392,
-      "learning_rate": 0.00019999887378742328,
-      "loss": 46.0,
-      "step": 9391
-    },
-    {
-      "epoch": 1.5125005032408714,
-      "grad_norm": 0.0031629258301109076,
-      "learning_rate": 0.00019999887354730592,
-      "loss": 46.0,
-      "step": 9392
-    },
-    {
-      "epoch": 1.5126615403196586,
-      "grad_norm": 0.0013771563535556197,
-      "learning_rate": 0.00019999887330716297,
-      "loss": 46.0,
-      "step": 9393
-    },
-    {
-      "epoch": 1.5128225773984458,
-      "grad_norm": 0.004468581173568964,
-      "learning_rate": 0.0001999988730669944,
-      "loss": 46.0,
-      "step": 9394
-    },
-    {
-      "epoch": 1.5129836144772333,
-      "grad_norm": 0.001272951834835112,
-      "learning_rate": 0.00019999887282680026,
-      "loss": 46.0,
-      "step": 9395
-    },
-    {
-      "epoch": 1.5131446515560207,
-      "grad_norm": 0.0011829917784780264,
-      "learning_rate": 0.00019999887258658052,
-      "loss": 46.0,
-      "step": 9396
-    },
-    {
-      "epoch": 1.5133056886348082,
-      "grad_norm": 0.0007782801403664052,
-      "learning_rate": 0.00019999887234633517,
-      "loss": 46.0,
-      "step": 9397
-    },
-    {
-      "epoch": 1.5134667257135956,
-      "grad_norm": 0.004144345875829458,
-      "learning_rate": 0.00019999887210606426,
-      "loss": 46.0,
-      "step": 9398
-    },
-    {
-      "epoch": 1.513627762792383,
-      "grad_norm": 0.0007989995647221804,
-      "learning_rate": 0.00019999887186576774,
-      "loss": 46.0,
-      "step": 9399
-    },
-    {
-      "epoch": 1.5137887998711703,
-      "grad_norm": 0.0019902114290744066,
-      "learning_rate": 0.00019999887162544563,
-      "loss": 46.0,
-      "step": 9400
-    },
-    {
-      "epoch": 1.5139498369499578,
-      "grad_norm": 0.006263981107622385,
-      "learning_rate": 0.0001999988713850979,
-      "loss": 46.0,
-      "step": 9401
-    },
-    {
-      "epoch": 1.514110874028745,
-      "grad_norm": 0.0006455009570345283,
-      "learning_rate": 0.00019999887114472459,
-      "loss": 46.0,
-      "step": 9402
-    },
-    {
-      "epoch": 1.5142719111075325,
-      "grad_norm": 0.002476086374372244,
-      "learning_rate": 0.00019999887090432568,
-      "loss": 46.0,
-      "step": 9403
-    },
-    {
-      "epoch": 1.51443294818632,
-      "grad_norm": 0.0046978420577943325,
-      "learning_rate": 0.0001999988706639012,
-      "loss": 46.0,
-      "step": 9404
-    },
-    {
-      "epoch": 1.5145939852651074,
-      "grad_norm": 0.005500440485775471,
-      "learning_rate": 0.0001999988704234511,
-      "loss": 46.0,
-      "step": 9405
-    },
-    {
-      "epoch": 1.5147550223438948,
-      "grad_norm": 0.00466714845970273,
-      "learning_rate": 0.0001999988701829754,
-      "loss": 46.0,
-      "step": 9406
-    },
-    {
-      "epoch": 1.514916059422682,
-      "grad_norm": 0.004333703778684139,
-      "learning_rate": 0.00019999886994247416,
-      "loss": 46.0,
-      "step": 9407
-    },
-    {
-      "epoch": 1.5150770965014695,
-      "grad_norm": 0.0029469789005815983,
-      "learning_rate": 0.00019999886970194726,
-      "loss": 46.0,
-      "step": 9408
-    },
-    {
-      "epoch": 1.5152381335802567,
-      "grad_norm": 0.000881087442394346,
-      "learning_rate": 0.00019999886946139479,
-      "loss": 46.0,
-      "step": 9409
-    },
-    {
-      "epoch": 1.5153991706590442,
-      "grad_norm": 0.0007487268303520977,
-      "learning_rate": 0.00019999886922081672,
-      "loss": 46.0,
-      "step": 9410
-    },
-    {
-      "epoch": 1.5155602077378316,
-      "grad_norm": 0.001845317892730236,
-      "learning_rate": 0.00019999886898021307,
-      "loss": 46.0,
-      "step": 9411
-    },
-    {
-      "epoch": 1.515721244816619,
-      "grad_norm": 0.0044486126862466335,
-      "learning_rate": 0.00019999886873958383,
-      "loss": 46.0,
-      "step": 9412
-    },
-    {
-      "epoch": 1.5158822818954065,
-      "grad_norm": 0.0032403534278273582,
-      "learning_rate": 0.00019999886849892895,
-      "loss": 46.0,
-      "step": 9413
-    },
-    {
-      "epoch": 1.5160433189741938,
-      "grad_norm": 0.001502752536907792,
-      "learning_rate": 0.0001999988682582485,
-      "loss": 46.0,
-      "step": 9414
-    },
-    {
-      "epoch": 1.5162043560529812,
-      "grad_norm": 0.0024749189615249634,
-      "learning_rate": 0.00019999886801754245,
-      "loss": 46.0,
-      "step": 9415
-    },
-    {
-      "epoch": 1.5163653931317684,
-      "grad_norm": 0.004080994985997677,
-      "learning_rate": 0.00019999886777681086,
-      "loss": 46.0,
-      "step": 9416
-    },
-    {
-      "epoch": 1.5165264302105559,
-      "grad_norm": 0.0015440452843904495,
-      "learning_rate": 0.0001999988675360536,
-      "loss": 46.0,
-      "step": 9417
-    },
-    {
-      "epoch": 1.5166874672893433,
-      "grad_norm": 0.003377762855961919,
-      "learning_rate": 0.0001999988672952708,
-      "loss": 46.0,
-      "step": 9418
-    },
-    {
-      "epoch": 1.5168485043681308,
-      "grad_norm": 0.00448443740606308,
-      "learning_rate": 0.00019999886705446239,
-      "loss": 46.0,
-      "step": 9419
-    },
-    {
-      "epoch": 1.5170095414469182,
-      "grad_norm": 0.0027451671194285154,
-      "learning_rate": 0.0001999988668136284,
-      "loss": 46.0,
-      "step": 9420
-    },
-    {
-      "epoch": 1.5171705785257057,
-      "grad_norm": 0.00677466532215476,
-      "learning_rate": 0.00019999886657276876,
-      "loss": 46.0,
-      "step": 9421
-    },
-    {
-      "epoch": 1.517331615604493,
-      "grad_norm": 0.0010044913506135345,
-      "learning_rate": 0.00019999886633188357,
-      "loss": 46.0,
-      "step": 9422
-    },
-    {
-      "epoch": 1.5174926526832804,
-      "grad_norm": 0.0032073038164526224,
-      "learning_rate": 0.00019999886609097276,
-      "loss": 46.0,
-      "step": 9423
-    },
-    {
-      "epoch": 1.5176536897620676,
-      "grad_norm": 0.003198109334334731,
-      "learning_rate": 0.00019999886585003637,
-      "loss": 46.0,
-      "step": 9424
-    },
-    {
-      "epoch": 1.517814726840855,
-      "grad_norm": 0.010036427527666092,
-      "learning_rate": 0.0001999988656090744,
-      "loss": 46.0,
-      "step": 9425
-    },
-    {
-      "epoch": 1.5179757639196425,
-      "grad_norm": 0.0008063007844612002,
-      "learning_rate": 0.00019999886536808682,
-      "loss": 46.0,
-      "step": 9426
-    },
-    {
-      "epoch": 1.51813680099843,
-      "grad_norm": 0.0009237265912815928,
-      "learning_rate": 0.00019999886512707364,
-      "loss": 46.0,
-      "step": 9427
-    },
-    {
-      "epoch": 1.5182978380772174,
-      "grad_norm": 0.0017372327856719494,
-      "learning_rate": 0.00019999886488603487,
-      "loss": 46.0,
-      "step": 9428
-    },
-    {
-      "epoch": 1.5184588751560046,
-      "grad_norm": 0.0023281259927898645,
-      "learning_rate": 0.0001999988646449705,
-      "loss": 46.0,
-      "step": 9429
-    },
-    {
-      "epoch": 1.518619912234792,
-      "grad_norm": 0.0008105894667096436,
-      "learning_rate": 0.00019999886440388056,
-      "loss": 46.0,
-      "step": 9430
-    },
-    {
-      "epoch": 1.5187809493135793,
-      "grad_norm": 0.002628910355269909,
-      "learning_rate": 0.000199998864162765,
-      "loss": 46.0,
-      "step": 9431
-    },
-    {
-      "epoch": 1.5189419863923668,
-      "grad_norm": 0.004149747081100941,
-      "learning_rate": 0.0001999988639216239,
-      "loss": 46.0,
-      "step": 9432
-    },
-    {
-      "epoch": 1.5191030234711542,
-      "grad_norm": 0.0019812181126326323,
-      "learning_rate": 0.00019999886368045713,
-      "loss": 46.0,
-      "step": 9433
-    },
-    {
-      "epoch": 1.5192640605499417,
-      "grad_norm": 0.001852197339758277,
-      "learning_rate": 0.0001999988634392648,
-      "loss": 46.0,
-      "step": 9434
-    },
-    {
-      "epoch": 1.5194250976287291,
-      "grad_norm": 0.001097636646591127,
-      "learning_rate": 0.00019999886319804687,
-      "loss": 46.0,
-      "step": 9435
-    },
-    {
-      "epoch": 1.5195861347075164,
-      "grad_norm": 0.002315138466656208,
-      "learning_rate": 0.00019999886295680335,
-      "loss": 46.0,
-      "step": 9436
-    },
-    {
-      "epoch": 1.5197471717863038,
-      "grad_norm": 0.0005515816737897694,
-      "learning_rate": 0.00019999886271553422,
-      "loss": 46.0,
-      "step": 9437
-    },
-    {
-      "epoch": 1.519908208865091,
-      "grad_norm": 0.013057180680334568,
-      "learning_rate": 0.0001999988624742395,
-      "loss": 46.0,
-      "step": 9438
-    },
-    {
-      "epoch": 1.5200692459438785,
-      "grad_norm": 0.005942794028669596,
-      "learning_rate": 0.00019999886223291918,
-      "loss": 46.0,
-      "step": 9439
-    },
-    {
-      "epoch": 1.520230283022666,
-      "grad_norm": 0.0067812916822731495,
-      "learning_rate": 0.00019999886199157329,
-      "loss": 46.0,
-      "step": 9440
-    },
-    {
-      "epoch": 1.5203913201014534,
-      "grad_norm": 0.0021425816230475903,
-      "learning_rate": 0.0001999988617502018,
-      "loss": 46.0,
-      "step": 9441
-    },
-    {
-      "epoch": 1.5205523571802408,
-      "grad_norm": 0.0017270178068429232,
-      "learning_rate": 0.0001999988615088047,
-      "loss": 46.0,
-      "step": 9442
-    },
-    {
-      "epoch": 1.5207133942590283,
-      "grad_norm": 0.0011038320371881127,
-      "learning_rate": 0.00019999886126738204,
-      "loss": 46.0,
-      "step": 9443
-    },
-    {
-      "epoch": 1.5208744313378155,
-      "grad_norm": 0.0018948365468531847,
-      "learning_rate": 0.00019999886102593374,
-      "loss": 46.0,
-      "step": 9444
-    },
-    {
-      "epoch": 1.521035468416603,
-      "grad_norm": 0.0007690870552323759,
-      "learning_rate": 0.00019999886078445986,
-      "loss": 46.0,
-      "step": 9445
-    },
-    {
-      "epoch": 1.5211965054953902,
-      "grad_norm": 0.0009242234518751502,
-      "learning_rate": 0.0001999988605429604,
-      "loss": 46.0,
-      "step": 9446
-    },
-    {
-      "epoch": 1.5213575425741777,
-      "grad_norm": 0.0010537394555285573,
-      "learning_rate": 0.00019999886030143532,
-      "loss": 46.0,
-      "step": 9447
-    },
-    {
-      "epoch": 1.521518579652965,
-      "grad_norm": 0.0008042344707064331,
-      "learning_rate": 0.0001999988600598847,
-      "loss": 46.0,
-      "step": 9448
-    },
-    {
-      "epoch": 1.5216796167317526,
-      "grad_norm": 0.003296369919553399,
-      "learning_rate": 0.0001999988598183084,
-      "loss": 46.0,
-      "step": 9449
-    },
-    {
-      "epoch": 1.52184065381054,
-      "grad_norm": 0.006282202899456024,
-      "learning_rate": 0.00019999885957670656,
-      "loss": 46.0,
-      "step": 9450
-    },
-    {
-      "epoch": 1.5220016908893272,
-      "grad_norm": 0.002256965497508645,
-      "learning_rate": 0.00019999885933507915,
-      "loss": 46.0,
-      "step": 9451
-    },
-    {
-      "epoch": 1.5221627279681147,
-      "grad_norm": 0.003114009741693735,
-      "learning_rate": 0.0001999988590934261,
-      "loss": 46.0,
-      "step": 9452
-    },
-    {
-      "epoch": 1.522323765046902,
-      "grad_norm": 0.0042876312509179115,
-      "learning_rate": 0.00019999885885174746,
-      "loss": 46.0,
-      "step": 9453
-    },
-    {
-      "epoch": 1.5224848021256894,
-      "grad_norm": 0.0022926165256649256,
-      "learning_rate": 0.00019999885861004323,
-      "loss": 46.0,
-      "step": 9454
-    },
-    {
-      "epoch": 1.5226458392044768,
-      "grad_norm": 0.0018849418265745044,
-      "learning_rate": 0.0001999988583683134,
-      "loss": 46.0,
-      "step": 9455
-    },
-    {
-      "epoch": 1.5228068762832643,
-      "grad_norm": 0.005244969390332699,
-      "learning_rate": 0.000199998858126558,
-      "loss": 46.0,
-      "step": 9456
-    },
-    {
-      "epoch": 1.5229679133620517,
-      "grad_norm": 0.004823376424610615,
-      "learning_rate": 0.000199998857884777,
-      "loss": 46.0,
-      "step": 9457
-    },
-    {
-      "epoch": 1.523128950440839,
-      "grad_norm": 0.001401046640239656,
-      "learning_rate": 0.00019999885764297037,
-      "loss": 46.0,
-      "step": 9458
-    },
-    {
-      "epoch": 1.5232899875196264,
-      "grad_norm": 0.004227713216096163,
-      "learning_rate": 0.00019999885740113818,
-      "loss": 46.0,
-      "step": 9459
-    },
-    {
-      "epoch": 1.5234510245984136,
-      "grad_norm": 0.0009940174641087651,
-      "learning_rate": 0.0001999988571592804,
-      "loss": 46.0,
-      "step": 9460
-    },
-    {
-      "epoch": 1.523612061677201,
-      "grad_norm": 0.0020566503517329693,
-      "learning_rate": 0.000199998856917397,
-      "loss": 46.0,
-      "step": 9461
-    },
-    {
-      "epoch": 1.5237730987559885,
-      "grad_norm": 0.0027434672228991985,
-      "learning_rate": 0.000199998856675488,
-      "loss": 46.0,
-      "step": 9462
-    },
-    {
-      "epoch": 1.523934135834776,
-      "grad_norm": 0.000408309162594378,
-      "learning_rate": 0.00019999885643355345,
-      "loss": 46.0,
-      "step": 9463
-    },
-    {
-      "epoch": 1.5240951729135634,
-      "grad_norm": 0.0031712001655250788,
-      "learning_rate": 0.00019999885619159327,
-      "loss": 46.0,
-      "step": 9464
-    },
-    {
-      "epoch": 1.524256209992351,
-      "grad_norm": 0.003166772425174713,
-      "learning_rate": 0.0001999988559496075,
-      "loss": 46.0,
-      "step": 9465
-    },
-    {
-      "epoch": 1.5244172470711381,
-      "grad_norm": 0.0014062365517020226,
-      "learning_rate": 0.00019999885570759615,
-      "loss": 46.0,
-      "step": 9466
-    },
-    {
-      "epoch": 1.5245782841499254,
-      "grad_norm": 0.002360724611207843,
-      "learning_rate": 0.00019999885546555918,
-      "loss": 46.0,
-      "step": 9467
-    },
-    {
-      "epoch": 1.5247393212287128,
-      "grad_norm": 0.0029434501193463802,
-      "learning_rate": 0.00019999885522349663,
-      "loss": 46.0,
-      "step": 9468
-    },
-    {
-      "epoch": 1.5249003583075003,
-      "grad_norm": 0.0009128785459324718,
-      "learning_rate": 0.00019999885498140848,
-      "loss": 46.0,
-      "step": 9469
-    },
-    {
-      "epoch": 1.5250613953862877,
-      "grad_norm": 0.0022576386108994484,
-      "learning_rate": 0.00019999885473929476,
-      "loss": 46.0,
-      "step": 9470
-    },
-    {
-      "epoch": 1.5252224324650752,
-      "grad_norm": 0.005614512134343386,
-      "learning_rate": 0.00019999885449715544,
-      "loss": 46.0,
-      "step": 9471
-    },
-    {
-      "epoch": 1.5253834695438626,
-      "grad_norm": 0.0014592206571251154,
-      "learning_rate": 0.00019999885425499048,
-      "loss": 46.0,
-      "step": 9472
-    },
-    {
-      "epoch": 1.5255445066226498,
-      "grad_norm": 0.00301287230104208,
-      "learning_rate": 0.00019999885401279996,
-      "loss": 46.0,
-      "step": 9473
-    },
-    {
-      "epoch": 1.5257055437014373,
-      "grad_norm": 0.00235157017596066,
-      "learning_rate": 0.00019999885377058386,
-      "loss": 46.0,
-      "step": 9474
-    },
-    {
-      "epoch": 1.5258665807802245,
-      "grad_norm": 0.0006280430243350565,
-      "learning_rate": 0.00019999885352834214,
-      "loss": 46.0,
-      "step": 9475
-    },
-    {
-      "epoch": 1.526027617859012,
-      "grad_norm": 0.0027024049777537584,
-      "learning_rate": 0.0001999988532860748,
-      "loss": 46.0,
-      "step": 9476
-    },
-    {
-      "epoch": 1.5261886549377994,
-      "grad_norm": 0.002077752025797963,
-      "learning_rate": 0.0001999988530437819,
-      "loss": 46.0,
-      "step": 9477
-    },
-    {
-      "epoch": 1.5263496920165869,
-      "grad_norm": 0.0024714572355151176,
-      "learning_rate": 0.00019999885280146343,
-      "loss": 46.0,
-      "step": 9478
-    },
-    {
-      "epoch": 1.5265107290953743,
-      "grad_norm": 0.0029144366271793842,
-      "learning_rate": 0.00019999885255911934,
-      "loss": 46.0,
-      "step": 9479
-    },
-    {
-      "epoch": 1.5266717661741616,
-      "grad_norm": 0.0006407216424122453,
-      "learning_rate": 0.00019999885231674965,
-      "loss": 46.0,
-      "step": 9480
-    },
-    {
-      "epoch": 1.526832803252949,
-      "grad_norm": 0.002878450322896242,
-      "learning_rate": 0.00019999885207435436,
-      "loss": 46.0,
-      "step": 9481
-    },
-    {
-      "epoch": 1.5269938403317362,
-      "grad_norm": 0.0013184280833229423,
-      "learning_rate": 0.0001999988518319335,
-      "loss": 46.0,
-      "step": 9482
-    },
-    {
-      "epoch": 1.5271548774105237,
-      "grad_norm": 0.0007456211023963988,
-      "learning_rate": 0.000199998851589487,
-      "loss": 46.0,
-      "step": 9483
-    },
-    {
-      "epoch": 1.5273159144893111,
-      "grad_norm": 0.001577394432388246,
-      "learning_rate": 0.00019999885134701497,
-      "loss": 46.0,
-      "step": 9484
-    },
-    {
-      "epoch": 1.5274769515680986,
-      "grad_norm": 0.004111071117222309,
-      "learning_rate": 0.0001999988511045173,
-      "loss": 46.0,
-      "step": 9485
-    },
-    {
-      "epoch": 1.527637988646886,
-      "grad_norm": 0.0010693949880078435,
-      "learning_rate": 0.00019999885086199407,
-      "loss": 46.0,
-      "step": 9486
-    },
-    {
-      "epoch": 1.5277990257256735,
-      "grad_norm": 0.0005185760091990232,
-      "learning_rate": 0.00019999885061944522,
-      "loss": 46.0,
-      "step": 9487
-    },
-    {
-      "epoch": 1.5279600628044607,
-      "grad_norm": 0.0025632947217673063,
-      "learning_rate": 0.00019999885037687076,
-      "loss": 46.0,
-      "step": 9488
-    },
-    {
-      "epoch": 1.528121099883248,
-      "grad_norm": 0.0009062702883966267,
-      "learning_rate": 0.00019999885013427074,
-      "loss": 46.0,
-      "step": 9489
-    },
-    {
-      "epoch": 1.5282821369620354,
-      "grad_norm": 0.0017577996477484703,
-      "learning_rate": 0.0001999988498916451,
-      "loss": 46.0,
-      "step": 9490
-    },
-    {
-      "epoch": 1.5284431740408229,
-      "grad_norm": 0.0030216483864933252,
-      "learning_rate": 0.00019999884964899388,
-      "loss": 46.0,
-      "step": 9491
-    },
-    {
-      "epoch": 1.5286042111196103,
-      "grad_norm": 0.0011699993629008532,
-      "learning_rate": 0.00019999884940631704,
-      "loss": 46.0,
-      "step": 9492
-    },
-    {
-      "epoch": 1.5287652481983978,
-      "grad_norm": 0.0007909003761596978,
-      "learning_rate": 0.00019999884916361462,
-      "loss": 46.0,
-      "step": 9493
-    },
-    {
-      "epoch": 1.5289262852771852,
-      "grad_norm": 0.009436006657779217,
-      "learning_rate": 0.00019999884892088664,
-      "loss": 46.0,
-      "step": 9494
-    },
-    {
-      "epoch": 1.5290873223559724,
-      "grad_norm": 0.0005345451063476503,
-      "learning_rate": 0.000199998848678133,
-      "loss": 46.0,
-      "step": 9495
-    },
-    {
-      "epoch": 1.52924835943476,
-      "grad_norm": 0.0019343576859682798,
-      "learning_rate": 0.00019999884843535382,
-      "loss": 46.0,
-      "step": 9496
-    },
-    {
-      "epoch": 1.5294093965135471,
-      "grad_norm": 0.0012227301485836506,
-      "learning_rate": 0.00019999884819254903,
-      "loss": 46.0,
-      "step": 9497
-    },
-    {
-      "epoch": 1.5295704335923346,
-      "grad_norm": 0.0024370127357542515,
-      "learning_rate": 0.00019999884794971867,
-      "loss": 46.0,
-      "step": 9498
-    },
-    {
-      "epoch": 1.529731470671122,
-      "grad_norm": 0.005993218161165714,
-      "learning_rate": 0.00019999884770686266,
-      "loss": 46.0,
-      "step": 9499
-    },
-    {
-      "epoch": 1.5298925077499095,
-      "grad_norm": 0.0016420215833932161,
-      "learning_rate": 0.00019999884746398108,
-      "loss": 46.0,
-      "step": 9500
-    },
-    {
-      "epoch": 1.530053544828697,
-      "grad_norm": 0.0007157622021622956,
-      "learning_rate": 0.00019999884722107393,
-      "loss": 46.0,
-      "step": 9501
-    },
-    {
-      "epoch": 1.5302145819074842,
-      "grad_norm": 0.002407926134765148,
-      "learning_rate": 0.00019999884697814116,
-      "loss": 46.0,
-      "step": 9502
-    },
-    {
-      "epoch": 1.5303756189862716,
-      "grad_norm": 0.002988303080201149,
-      "learning_rate": 0.0001999988467351828,
-      "loss": 46.0,
-      "step": 9503
-    },
-    {
-      "epoch": 1.5305366560650588,
-      "grad_norm": 0.006468224339187145,
-      "learning_rate": 0.00019999884649219882,
-      "loss": 46.0,
-      "step": 9504
-    },
-    {
-      "epoch": 1.5306976931438463,
-      "grad_norm": 0.0013437657617032528,
-      "learning_rate": 0.0001999988462491893,
-      "loss": 46.0,
-      "step": 9505
-    },
-    {
-      "epoch": 1.5308587302226337,
-      "grad_norm": 0.0013888109242543578,
-      "learning_rate": 0.00019999884600615416,
-      "loss": 46.0,
-      "step": 9506
-    },
-    {
-      "epoch": 1.5310197673014212,
-      "grad_norm": 0.0010514655150473118,
-      "learning_rate": 0.0001999988457630934,
-      "loss": 46.0,
-      "step": 9507
-    },
-    {
-      "epoch": 1.5311808043802086,
-      "grad_norm": 0.00038760315510444343,
-      "learning_rate": 0.00019999884552000707,
-      "loss": 46.0,
-      "step": 9508
-    },
-    {
-      "epoch": 1.531341841458996,
-      "grad_norm": 0.005011890549212694,
-      "learning_rate": 0.00019999884527689514,
-      "loss": 46.0,
-      "step": 9509
-    },
-    {
-      "epoch": 1.5315028785377833,
-      "grad_norm": 0.002129156840965152,
-      "learning_rate": 0.00019999884503375763,
-      "loss": 46.0,
-      "step": 9510
-    },
-    {
-      "epoch": 1.5316639156165706,
-      "grad_norm": 0.0015337747754529119,
-      "learning_rate": 0.0001999988447905945,
-      "loss": 46.0,
-      "step": 9511
-    },
-    {
-      "epoch": 1.531824952695358,
-      "grad_norm": 0.007746582850813866,
-      "learning_rate": 0.00019999884454740578,
-      "loss": 46.0,
-      "step": 9512
-    },
-    {
-      "epoch": 1.5319859897741455,
-      "grad_norm": 0.008065969683229923,
-      "learning_rate": 0.00019999884430419148,
-      "loss": 46.0,
-      "step": 9513
-    },
-    {
-      "epoch": 1.532147026852933,
-      "grad_norm": 0.0006683963001705706,
-      "learning_rate": 0.00019999884406095156,
-      "loss": 46.0,
-      "step": 9514
-    },
-    {
-      "epoch": 1.5323080639317204,
-      "grad_norm": 0.0038896435871720314,
-      "learning_rate": 0.00019999884381768609,
-      "loss": 46.0,
-      "step": 9515
-    },
-    {
-      "epoch": 1.5324691010105078,
-      "grad_norm": 0.011683257296681404,
-      "learning_rate": 0.000199998843574395,
-      "loss": 46.0,
-      "step": 9516
-    },
-    {
-      "epoch": 1.532630138089295,
-      "grad_norm": 0.0010450986446812749,
-      "learning_rate": 0.0001999988433310783,
-      "loss": 46.0,
-      "step": 9517
-    },
-    {
-      "epoch": 1.5327911751680825,
-      "grad_norm": 0.000758933019824326,
-      "learning_rate": 0.00019999884308773602,
-      "loss": 46.0,
-      "step": 9518
-    },
-    {
-      "epoch": 1.5329522122468697,
-      "grad_norm": 0.0019322397420182824,
-      "learning_rate": 0.00019999884284436817,
-      "loss": 46.0,
-      "step": 9519
-    },
-    {
-      "epoch": 1.5331132493256572,
-      "grad_norm": 0.00167573313228786,
-      "learning_rate": 0.00019999884260097468,
-      "loss": 46.0,
-      "step": 9520
-    },
-    {
-      "epoch": 1.5332742864044446,
-      "grad_norm": 0.0021413343492895365,
-      "learning_rate": 0.00019999884235755562,
-      "loss": 46.0,
-      "step": 9521
-    },
-    {
-      "epoch": 1.533435323483232,
-      "grad_norm": 0.0012244421523064375,
-      "learning_rate": 0.00019999884211411098,
-      "loss": 46.0,
-      "step": 9522
-    },
-    {
-      "epoch": 1.5335963605620195,
-      "grad_norm": 0.0019074471201747656,
-      "learning_rate": 0.0001999988418706407,
-      "loss": 46.0,
-      "step": 9523
-    },
-    {
-      "epoch": 1.5337573976408068,
-      "grad_norm": 0.001719399238936603,
-      "learning_rate": 0.00019999884162714485,
-      "loss": 46.0,
-      "step": 9524
-    },
-    {
-      "epoch": 1.5339184347195942,
-      "grad_norm": 0.005042642820626497,
-      "learning_rate": 0.00019999884138362342,
-      "loss": 46.0,
-      "step": 9525
-    },
-    {
-      "epoch": 1.5340794717983814,
-      "grad_norm": 0.0006284856353886425,
-      "learning_rate": 0.00019999884114007638,
-      "loss": 46.0,
-      "step": 9526
-    },
-    {
-      "epoch": 1.534240508877169,
-      "grad_norm": 0.004295391030609608,
-      "learning_rate": 0.00019999884089650375,
-      "loss": 46.0,
-      "step": 9527
-    },
-    {
-      "epoch": 1.5344015459559563,
-      "grad_norm": 0.0026151149068027735,
-      "learning_rate": 0.00019999884065290556,
-      "loss": 46.0,
-      "step": 9528
-    },
-    {
-      "epoch": 1.5345625830347438,
-      "grad_norm": 0.002352813957259059,
-      "learning_rate": 0.00019999884040928172,
-      "loss": 46.0,
-      "step": 9529
-    },
-    {
-      "epoch": 1.5347236201135313,
-      "grad_norm": 0.001912680920213461,
-      "learning_rate": 0.0001999988401656323,
-      "loss": 46.0,
-      "step": 9530
-    },
-    {
-      "epoch": 1.5348846571923185,
-      "grad_norm": 0.0016547852428629994,
-      "learning_rate": 0.0001999988399219573,
-      "loss": 46.0,
-      "step": 9531
-    },
-    {
-      "epoch": 1.535045694271106,
-      "grad_norm": 0.003520425409078598,
-      "learning_rate": 0.0001999988396782567,
-      "loss": 46.0,
-      "step": 9532
-    },
-    {
-      "epoch": 1.5352067313498932,
-      "grad_norm": 0.0017291458789259195,
-      "learning_rate": 0.00019999883943453052,
-      "loss": 46.0,
-      "step": 9533
-    },
-    {
-      "epoch": 1.5353677684286806,
-      "grad_norm": 0.0028484223876148462,
-      "learning_rate": 0.00019999883919077872,
-      "loss": 46.0,
-      "step": 9534
-    },
-    {
-      "epoch": 1.535528805507468,
-      "grad_norm": 0.0013635712675750256,
-      "learning_rate": 0.00019999883894700134,
-      "loss": 46.0,
-      "step": 9535
-    },
-    {
-      "epoch": 1.5356898425862555,
-      "grad_norm": 0.0014535625232383609,
-      "learning_rate": 0.00019999883870319834,
-      "loss": 46.0,
-      "step": 9536
-    },
-    {
-      "epoch": 1.535850879665043,
-      "grad_norm": 0.0004540627123787999,
-      "learning_rate": 0.00019999883845936976,
-      "loss": 46.0,
-      "step": 9537
-    },
-    {
-      "epoch": 1.5360119167438304,
-      "grad_norm": 0.01284743007272482,
-      "learning_rate": 0.00019999883821551558,
-      "loss": 46.0,
-      "step": 9538
-    },
-    {
-      "epoch": 1.5361729538226176,
-      "grad_norm": 0.007350571919232607,
-      "learning_rate": 0.0001999988379716358,
-      "loss": 46.0,
-      "step": 9539
-    },
-    {
-      "epoch": 1.536333990901405,
-      "grad_norm": 0.008967571891844273,
-      "learning_rate": 0.00019999883772773048,
-      "loss": 46.0,
-      "step": 9540
-    },
-    {
-      "epoch": 1.5364950279801923,
-      "grad_norm": 0.0023639057762920856,
-      "learning_rate": 0.00019999883748379952,
-      "loss": 46.0,
-      "step": 9541
-    },
-    {
-      "epoch": 1.5366560650589798,
-      "grad_norm": 0.0007659141556359828,
-      "learning_rate": 0.00019999883723984297,
-      "loss": 46.0,
-      "step": 9542
-    },
-    {
-      "epoch": 1.5368171021377672,
-      "grad_norm": 0.00257940124720335,
-      "learning_rate": 0.0001999988369958608,
-      "loss": 46.0,
-      "step": 9543
-    },
-    {
-      "epoch": 1.5369781392165547,
-      "grad_norm": 0.003213014453649521,
-      "learning_rate": 0.0001999988367518531,
-      "loss": 46.0,
-      "step": 9544
-    },
-    {
-      "epoch": 1.5371391762953421,
-      "grad_norm": 0.000548528681974858,
-      "learning_rate": 0.00019999883650781975,
-      "loss": 46.0,
-      "step": 9545
-    },
-    {
-      "epoch": 1.5373002133741294,
-      "grad_norm": 0.005622244440019131,
-      "learning_rate": 0.00019999883626376083,
-      "loss": 46.0,
-      "step": 9546
-    },
-    {
-      "epoch": 1.5374612504529168,
-      "grad_norm": 0.001243532751686871,
-      "learning_rate": 0.0001999988360196763,
-      "loss": 46.0,
-      "step": 9547
-    },
-    {
-      "epoch": 1.537622287531704,
-      "grad_norm": 0.0014249305240809917,
-      "learning_rate": 0.00019999883577556616,
-      "loss": 46.0,
-      "step": 9548
-    },
-    {
-      "epoch": 1.5377833246104915,
-      "grad_norm": 0.0013568312861025333,
-      "learning_rate": 0.00019999883553143048,
-      "loss": 46.0,
-      "step": 9549
-    },
-    {
-      "epoch": 1.537944361689279,
-      "grad_norm": 0.003219452453777194,
-      "learning_rate": 0.00019999883528726918,
-      "loss": 46.0,
-      "step": 9550
-    },
-    {
-      "epoch": 1.5381053987680664,
-      "grad_norm": 0.0025497402530163527,
-      "learning_rate": 0.00019999883504308227,
-      "loss": 46.0,
-      "step": 9551
-    },
-    {
-      "epoch": 1.5382664358468539,
-      "grad_norm": 0.0031076620798557997,
-      "learning_rate": 0.00019999883479886977,
-      "loss": 46.0,
-      "step": 9552
-    },
-    {
-      "epoch": 1.538427472925641,
-      "grad_norm": 0.0052823834121227264,
-      "learning_rate": 0.00019999883455463168,
-      "loss": 46.0,
-      "step": 9553
-    },
-    {
-      "epoch": 1.5385885100044285,
-      "grad_norm": 0.004734147805720568,
-      "learning_rate": 0.00019999883431036798,
-      "loss": 46.0,
-      "step": 9554
-    },
-    {
-      "epoch": 1.5387495470832158,
-      "grad_norm": 0.002411802299320698,
-      "learning_rate": 0.00019999883406607874,
-      "loss": 46.0,
-      "step": 9555
-    },
-    {
-      "epoch": 1.5389105841620032,
-      "grad_norm": 0.0026178339030593634,
-      "learning_rate": 0.00019999883382176386,
-      "loss": 46.0,
-      "step": 9556
-    },
-    {
-      "epoch": 1.5390716212407907,
-      "grad_norm": 0.0033464953303337097,
-      "learning_rate": 0.0001999988335774234,
-      "loss": 46.0,
-      "step": 9557
-    },
-    {
-      "epoch": 1.5392326583195781,
-      "grad_norm": 0.002232455415651202,
-      "learning_rate": 0.00019999883333305732,
-      "loss": 46.0,
-      "step": 9558
-    },
-    {
-      "epoch": 1.5393936953983656,
-      "grad_norm": 0.0008804902900010347,
-      "learning_rate": 0.00019999883308866566,
-      "loss": 46.0,
-      "step": 9559
-    },
-    {
-      "epoch": 1.539554732477153,
-      "grad_norm": 0.003191632218658924,
-      "learning_rate": 0.0001999988328442484,
-      "loss": 46.0,
-      "step": 9560
-    },
-    {
-      "epoch": 1.5397157695559403,
-      "grad_norm": 0.0005228768568485975,
-      "learning_rate": 0.00019999883259980557,
-      "loss": 46.0,
-      "step": 9561
-    },
-    {
-      "epoch": 1.5398768066347275,
-      "grad_norm": 0.00464447308331728,
-      "learning_rate": 0.0001999988323553371,
-      "loss": 46.0,
-      "step": 9562
-    },
-    {
-      "epoch": 1.540037843713515,
-      "grad_norm": 0.000984556507319212,
-      "learning_rate": 0.0001999988321108431,
-      "loss": 46.0,
-      "step": 9563
-    },
-    {
-      "epoch": 1.5401988807923024,
-      "grad_norm": 0.007177683990448713,
-      "learning_rate": 0.00019999883186632347,
-      "loss": 46.0,
-      "step": 9564
-    },
-    {
-      "epoch": 1.5403599178710898,
-      "grad_norm": 0.0015082447789609432,
-      "learning_rate": 0.00019999883162177823,
-      "loss": 46.0,
-      "step": 9565
-    },
-    {
-      "epoch": 1.5405209549498773,
-      "grad_norm": 0.002229714998975396,
-      "learning_rate": 0.00019999883137720743,
-      "loss": 46.0,
-      "step": 9566
-    },
-    {
-      "epoch": 1.5406819920286647,
-      "grad_norm": 0.002325141802430153,
-      "learning_rate": 0.000199998831132611,
-      "loss": 46.0,
-      "step": 9567
-    },
-    {
-      "epoch": 1.540843029107452,
-      "grad_norm": 0.004455997608602047,
-      "learning_rate": 0.00019999883088798898,
-      "loss": 46.0,
-      "step": 9568
-    },
-    {
-      "epoch": 1.5410040661862394,
-      "grad_norm": 0.008108694106340408,
-      "learning_rate": 0.0001999988306433414,
-      "loss": 46.0,
-      "step": 9569
-    },
-    {
-      "epoch": 1.5411651032650266,
-      "grad_norm": 0.012046033516526222,
-      "learning_rate": 0.0001999988303986682,
-      "loss": 46.0,
-      "step": 9570
-    },
-    {
-      "epoch": 1.541326140343814,
-      "grad_norm": 0.0017901815008372068,
-      "learning_rate": 0.00019999883015396942,
-      "loss": 46.0,
-      "step": 9571
-    },
-    {
-      "epoch": 1.5414871774226016,
-      "grad_norm": 0.0019001057371497154,
-      "learning_rate": 0.000199998829909245,
-      "loss": 46.0,
-      "step": 9572
-    },
-    {
-      "epoch": 1.541648214501389,
-      "grad_norm": 0.011951982975006104,
-      "learning_rate": 0.00019999882966449504,
-      "loss": 46.0,
-      "step": 9573
-    },
-    {
-      "epoch": 1.5418092515801765,
-      "grad_norm": 0.003459338331595063,
-      "learning_rate": 0.0001999988294197195,
-      "loss": 46.0,
-      "step": 9574
-    },
-    {
-      "epoch": 1.5419702886589637,
-      "grad_norm": 0.005500278435647488,
-      "learning_rate": 0.00019999882917491827,
-      "loss": 46.0,
-      "step": 9575
-    },
-    {
-      "epoch": 1.5421313257377511,
-      "grad_norm": 0.0008652671240270138,
-      "learning_rate": 0.00019999882893009154,
-      "loss": 46.0,
-      "step": 9576
-    },
-    {
-      "epoch": 1.5422923628165384,
-      "grad_norm": 0.0032363201025873423,
-      "learning_rate": 0.00019999882868523917,
-      "loss": 46.0,
-      "step": 9577
-    },
-    {
-      "epoch": 1.5424533998953258,
-      "grad_norm": 0.0017805654788389802,
-      "learning_rate": 0.0001999988284403612,
-      "loss": 46.0,
-      "step": 9578
-    },
-    {
-      "epoch": 1.5426144369741133,
-      "grad_norm": 0.00463515892624855,
-      "learning_rate": 0.00019999882819545767,
-      "loss": 46.0,
-      "step": 9579
-    },
-    {
-      "epoch": 1.5427754740529007,
-      "grad_norm": 0.0010887961834669113,
-      "learning_rate": 0.00019999882795052854,
-      "loss": 46.0,
-      "step": 9580
-    },
-    {
-      "epoch": 1.5429365111316882,
-      "grad_norm": 0.0014403262175619602,
-      "learning_rate": 0.0001999988277055738,
-      "loss": 46.0,
-      "step": 9581
-    },
-    {
-      "epoch": 1.5430975482104756,
-      "grad_norm": 0.002026051515713334,
-      "learning_rate": 0.00019999882746059346,
-      "loss": 46.0,
-      "step": 9582
-    },
-    {
-      "epoch": 1.5432585852892629,
-      "grad_norm": 0.0019245789153501391,
-      "learning_rate": 0.00019999882721558754,
-      "loss": 46.0,
-      "step": 9583
-    },
-    {
-      "epoch": 1.54341962236805,
-      "grad_norm": 0.007637613452970982,
-      "learning_rate": 0.000199998826970556,
-      "loss": 46.0,
-      "step": 9584
-    },
-    {
-      "epoch": 1.5435806594468375,
-      "grad_norm": 0.0010679109254851937,
-      "learning_rate": 0.0001999988267254989,
-      "loss": 46.0,
-      "step": 9585
-    },
-    {
-      "epoch": 1.543741696525625,
-      "grad_norm": 0.0013544211396947503,
-      "learning_rate": 0.0001999988264804162,
-      "loss": 46.0,
-      "step": 9586
-    },
-    {
-      "epoch": 1.5439027336044124,
-      "grad_norm": 0.0016832927940413356,
-      "learning_rate": 0.00019999882623530788,
-      "loss": 46.0,
-      "step": 9587
-    },
-    {
-      "epoch": 1.5440637706831999,
-      "grad_norm": 0.0006674944306723773,
-      "learning_rate": 0.000199998825990174,
-      "loss": 46.0,
-      "step": 9588
-    },
-    {
-      "epoch": 1.5442248077619873,
-      "grad_norm": 0.0005326775135472417,
-      "learning_rate": 0.0001999988257450145,
-      "loss": 46.0,
-      "step": 9589
-    },
-    {
-      "epoch": 1.5443858448407746,
-      "grad_norm": 0.0010344121837988496,
-      "learning_rate": 0.00019999882549982941,
-      "loss": 46.0,
-      "step": 9590
-    },
-    {
-      "epoch": 1.544546881919562,
-      "grad_norm": 0.0017514858627691865,
-      "learning_rate": 0.00019999882525461872,
-      "loss": 46.0,
-      "step": 9591
-    },
-    {
-      "epoch": 1.5447079189983492,
-      "grad_norm": 0.0007412286940962076,
-      "learning_rate": 0.00019999882500938246,
-      "loss": 46.0,
-      "step": 9592
-    },
-    {
-      "epoch": 1.5448689560771367,
-      "grad_norm": 0.004125567153096199,
-      "learning_rate": 0.00019999882476412058,
-      "loss": 46.0,
-      "step": 9593
-    },
-    {
-      "epoch": 1.5450299931559242,
-      "grad_norm": 0.0019025934161618352,
-      "learning_rate": 0.00019999882451883312,
-      "loss": 46.0,
-      "step": 9594
-    },
-    {
-      "epoch": 1.5451910302347116,
-      "grad_norm": 0.0009014655952341855,
-      "learning_rate": 0.00019999882427352002,
-      "loss": 46.0,
-      "step": 9595
-    },
-    {
-      "epoch": 1.545352067313499,
-      "grad_norm": 0.008022756315767765,
-      "learning_rate": 0.0001999988240281814,
-      "loss": 46.0,
-      "step": 9596
-    },
-    {
-      "epoch": 1.5455131043922863,
-      "grad_norm": 0.0005562002188526094,
-      "learning_rate": 0.00019999882378281714,
-      "loss": 46.0,
-      "step": 9597
-    },
-    {
-      "epoch": 1.5456741414710737,
-      "grad_norm": 0.002666367683559656,
-      "learning_rate": 0.0001999988235374273,
-      "loss": 46.0,
-      "step": 9598
-    },
-    {
-      "epoch": 1.545835178549861,
-      "grad_norm": 0.0017994075315073133,
-      "learning_rate": 0.00019999882329201185,
-      "loss": 46.0,
-      "step": 9599
-    },
-    {
-      "epoch": 1.5459962156286484,
-      "grad_norm": 0.0017747669480741024,
-      "learning_rate": 0.00019999882304657082,
-      "loss": 46.0,
-      "step": 9600
-    },
-    {
-      "epoch": 1.5461572527074359,
-      "grad_norm": 0.002589545911177993,
-      "learning_rate": 0.0001999988228011042,
-      "loss": 46.0,
-      "step": 9601
-    },
-    {
-      "epoch": 1.5463182897862233,
-      "grad_norm": 0.006421798840165138,
-      "learning_rate": 0.00019999882255561198,
-      "loss": 46.0,
-      "step": 9602
-    },
-    {
-      "epoch": 1.5464793268650108,
-      "grad_norm": 0.00528734689578414,
-      "learning_rate": 0.00019999882231009415,
-      "loss": 46.0,
-      "step": 9603
-    },
-    {
-      "epoch": 1.5466403639437982,
-      "grad_norm": 0.0019034610595554113,
-      "learning_rate": 0.00019999882206455074,
-      "loss": 46.0,
-      "step": 9604
-    },
-    {
-      "epoch": 1.5468014010225855,
-      "grad_norm": 0.004726993851363659,
-      "learning_rate": 0.00019999882181898174,
-      "loss": 46.0,
-      "step": 9605
-    },
-    {
-      "epoch": 1.5469624381013727,
-      "grad_norm": 0.0008454433409497142,
-      "learning_rate": 0.00019999882157338715,
-      "loss": 46.0,
-      "step": 9606
-    },
-    {
-      "epoch": 1.5471234751801601,
-      "grad_norm": 0.009155957959592342,
-      "learning_rate": 0.00019999882132776692,
-      "loss": 46.0,
-      "step": 9607
-    },
-    {
-      "epoch": 1.5472845122589476,
-      "grad_norm": 0.003743691137060523,
-      "learning_rate": 0.00019999882108212114,
-      "loss": 46.0,
-      "step": 9608
-    },
-    {
-      "epoch": 1.547445549337735,
-      "grad_norm": 0.0019337142584845424,
-      "learning_rate": 0.00019999882083644973,
-      "loss": 46.0,
-      "step": 9609
-    },
-    {
-      "epoch": 1.5476065864165225,
-      "grad_norm": 0.0013208661694079638,
-      "learning_rate": 0.00019999882059075277,
-      "loss": 46.0,
-      "step": 9610
-    },
-    {
-      "epoch": 1.54776762349531,
-      "grad_norm": 0.0032328011002391577,
-      "learning_rate": 0.00019999882034503022,
-      "loss": 46.0,
-      "step": 9611
-    },
-    {
-      "epoch": 1.5479286605740972,
-      "grad_norm": 0.011548959650099277,
-      "learning_rate": 0.00019999882009928203,
-      "loss": 46.0,
-      "step": 9612
-    },
-    {
-      "epoch": 1.5480896976528846,
-      "grad_norm": 0.0015373054193332791,
-      "learning_rate": 0.00019999881985350828,
-      "loss": 46.0,
-      "step": 9613
-    },
-    {
-      "epoch": 1.5482507347316719,
-      "grad_norm": 0.001842934056185186,
-      "learning_rate": 0.0001999988196077089,
-      "loss": 46.0,
-      "step": 9614
-    },
-    {
-      "epoch": 1.5484117718104593,
-      "grad_norm": 0.00206050300039351,
-      "learning_rate": 0.00019999881936188396,
-      "loss": 46.0,
-      "step": 9615
-    },
-    {
-      "epoch": 1.5485728088892468,
-      "grad_norm": 0.0033201349433511496,
-      "learning_rate": 0.0001999988191160334,
-      "loss": 46.0,
-      "step": 9616
-    },
-    {
-      "epoch": 1.5487338459680342,
-      "grad_norm": 0.0005659116432070732,
-      "learning_rate": 0.0001999988188701573,
-      "loss": 46.0,
-      "step": 9617
-    },
-    {
-      "epoch": 1.5488948830468217,
-      "grad_norm": 0.0024309554137289524,
-      "learning_rate": 0.00019999881862425555,
-      "loss": 46.0,
-      "step": 9618
-    },
-    {
-      "epoch": 1.5490559201256089,
-      "grad_norm": 0.0032668341882526875,
-      "learning_rate": 0.00019999881837832822,
-      "loss": 46.0,
-      "step": 9619
-    },
-    {
-      "epoch": 1.5492169572043963,
-      "grad_norm": 0.0016206871950998902,
-      "learning_rate": 0.00019999881813237527,
-      "loss": 46.0,
-      "step": 9620
-    },
-    {
-      "epoch": 1.5493779942831836,
-      "grad_norm": 0.006536832544952631,
-      "learning_rate": 0.00019999881788639677,
-      "loss": 46.0,
-      "step": 9621
-    },
-    {
-      "epoch": 1.549539031361971,
-      "grad_norm": 0.0012820035917684436,
-      "learning_rate": 0.00019999881764039263,
-      "loss": 46.0,
-      "step": 9622
-    },
-    {
-      "epoch": 1.5497000684407585,
-      "grad_norm": 0.002206459641456604,
-      "learning_rate": 0.00019999881739436295,
-      "loss": 46.0,
-      "step": 9623
-    },
-    {
-      "epoch": 1.549861105519546,
-      "grad_norm": 0.003899957751855254,
-      "learning_rate": 0.00019999881714830763,
-      "loss": 46.0,
-      "step": 9624
-    },
-    {
-      "epoch": 1.5500221425983334,
-      "grad_norm": 0.0033586008939892054,
-      "learning_rate": 0.00019999881690222675,
-      "loss": 46.0,
-      "step": 9625
-    },
-    {
-      "epoch": 1.5501831796771206,
-      "grad_norm": 0.00246395543217659,
-      "learning_rate": 0.00019999881665612023,
-      "loss": 46.0,
-      "step": 9626
-    },
-    {
-      "epoch": 1.550344216755908,
-      "grad_norm": 0.003294206690043211,
-      "learning_rate": 0.00019999881640998815,
-      "loss": 46.0,
-      "step": 9627
-    },
-    {
-      "epoch": 1.5505052538346953,
-      "grad_norm": 0.0024017507676035166,
-      "learning_rate": 0.00019999881616383048,
-      "loss": 46.0,
-      "step": 9628
-    },
-    {
-      "epoch": 1.5506662909134827,
-      "grad_norm": 0.006382984574884176,
-      "learning_rate": 0.0001999988159176472,
-      "loss": 46.0,
-      "step": 9629
-    },
-    {
-      "epoch": 1.5508273279922702,
-      "grad_norm": 0.0005004876293241978,
-      "learning_rate": 0.00019999881567143833,
-      "loss": 46.0,
-      "step": 9630
-    },
-    {
-      "epoch": 1.5509883650710576,
-      "grad_norm": 0.0020565080922096968,
-      "learning_rate": 0.00019999881542520385,
-      "loss": 46.0,
-      "step": 9631
-    },
-    {
-      "epoch": 1.551149402149845,
-      "grad_norm": 0.002605983754619956,
-      "learning_rate": 0.0001999988151789438,
-      "loss": 46.0,
-      "step": 9632
-    },
-    {
-      "epoch": 1.5513104392286325,
-      "grad_norm": 0.0009897048585116863,
-      "learning_rate": 0.00019999881493265814,
-      "loss": 46.0,
-      "step": 9633
-    },
-    {
-      "epoch": 1.5514714763074198,
-      "grad_norm": 0.0012678259517997503,
-      "learning_rate": 0.00019999881468634693,
-      "loss": 46.0,
-      "step": 9634
-    },
-    {
-      "epoch": 1.5516325133862072,
-      "grad_norm": 0.0029944973066449165,
-      "learning_rate": 0.00019999881444001004,
-      "loss": 46.0,
-      "step": 9635
-    },
-    {
-      "epoch": 1.5517935504649945,
-      "grad_norm": 0.0017019726801663637,
-      "learning_rate": 0.00019999881419364762,
-      "loss": 46.0,
-      "step": 9636
-    },
-    {
-      "epoch": 1.551954587543782,
-      "grad_norm": 0.001605618861503899,
-      "learning_rate": 0.0001999988139472596,
-      "loss": 46.0,
-      "step": 9637
-    },
-    {
-      "epoch": 1.5521156246225694,
-      "grad_norm": 0.0018793237395584583,
-      "learning_rate": 0.00019999881370084594,
-      "loss": 46.0,
-      "step": 9638
-    },
-    {
-      "epoch": 1.5522766617013568,
-      "grad_norm": 0.001324295881204307,
-      "learning_rate": 0.00019999881345440673,
-      "loss": 46.0,
-      "step": 9639
-    },
-    {
-      "epoch": 1.5524376987801443,
-      "grad_norm": 0.0011678725713863969,
-      "learning_rate": 0.0001999988132079419,
-      "loss": 46.0,
-      "step": 9640
-    },
-    {
-      "epoch": 1.5525987358589315,
-      "grad_norm": 0.002261652611196041,
-      "learning_rate": 0.0001999988129614515,
-      "loss": 46.0,
-      "step": 9641
-    },
-    {
-      "epoch": 1.552759772937719,
-      "grad_norm": 0.0017068805173039436,
-      "learning_rate": 0.0001999988127149355,
-      "loss": 46.0,
-      "step": 9642
-    },
-    {
-      "epoch": 1.5529208100165062,
-      "grad_norm": 0.003284177742898464,
-      "learning_rate": 0.0001999988124683939,
-      "loss": 46.0,
-      "step": 9643
-    },
-    {
-      "epoch": 1.5530818470952936,
-      "grad_norm": 0.0024580410681664944,
-      "learning_rate": 0.0001999988122218267,
-      "loss": 46.0,
-      "step": 9644
-    },
-    {
-      "epoch": 1.553242884174081,
-      "grad_norm": 0.0008174455724656582,
-      "learning_rate": 0.0001999988119752339,
-      "loss": 46.0,
-      "step": 9645
-    },
-    {
-      "epoch": 1.5534039212528685,
-      "grad_norm": 0.0011248092632740736,
-      "learning_rate": 0.0001999988117286155,
-      "loss": 46.0,
-      "step": 9646
-    },
-    {
-      "epoch": 1.553564958331656,
-      "grad_norm": 0.00452102767303586,
-      "learning_rate": 0.00019999881148197155,
-      "loss": 46.0,
-      "step": 9647
-    },
-    {
-      "epoch": 1.5537259954104432,
-      "grad_norm": 0.0009618006879463792,
-      "learning_rate": 0.00019999881123530198,
-      "loss": 46.0,
-      "step": 9648
-    },
-    {
-      "epoch": 1.5538870324892307,
-      "grad_norm": 0.003927244339138269,
-      "learning_rate": 0.0001999988109886068,
-      "loss": 46.0,
-      "step": 9649
-    },
-    {
-      "epoch": 1.5540480695680179,
-      "grad_norm": 0.0007460684282705188,
-      "learning_rate": 0.00019999881074188604,
-      "loss": 46.0,
-      "step": 9650
-    },
-    {
-      "epoch": 1.5542091066468053,
-      "grad_norm": 0.005137612111866474,
-      "learning_rate": 0.00019999881049513968,
-      "loss": 46.0,
-      "step": 9651
-    },
-    {
-      "epoch": 1.5543701437255928,
-      "grad_norm": 0.0016791753005236387,
-      "learning_rate": 0.00019999881024836773,
-      "loss": 46.0,
-      "step": 9652
-    },
-    {
-      "epoch": 1.5545311808043802,
-      "grad_norm": 0.0033985706977546215,
-      "learning_rate": 0.0001999988100015702,
-      "loss": 46.0,
-      "step": 9653
-    },
-    {
-      "epoch": 1.5546922178831677,
-      "grad_norm": 0.004292998928576708,
-      "learning_rate": 0.00019999880975474704,
-      "loss": 46.0,
-      "step": 9654
-    },
-    {
-      "epoch": 1.5548532549619551,
-      "grad_norm": 0.0011493455385789275,
-      "learning_rate": 0.0001999988095078983,
-      "loss": 46.0,
-      "step": 9655
-    },
-    {
-      "epoch": 1.5550142920407424,
-      "grad_norm": 0.0006883305031806231,
-      "learning_rate": 0.000199998809261024,
-      "loss": 46.0,
-      "step": 9656
-    },
-    {
-      "epoch": 1.5551753291195298,
-      "grad_norm": 0.0016404492780566216,
-      "learning_rate": 0.00019999880901412406,
-      "loss": 46.0,
-      "step": 9657
-    },
-    {
-      "epoch": 1.555336366198317,
-      "grad_norm": 0.0007315294933505356,
-      "learning_rate": 0.00019999880876719854,
-      "loss": 46.0,
-      "step": 9658
-    },
-    {
-      "epoch": 1.5554974032771045,
-      "grad_norm": 0.0032106267753988504,
-      "learning_rate": 0.00019999880852024745,
-      "loss": 46.0,
-      "step": 9659
-    },
-    {
-      "epoch": 1.555658440355892,
-      "grad_norm": 0.002301374450325966,
-      "learning_rate": 0.00019999880827327072,
-      "loss": 46.0,
-      "step": 9660
-    },
-    {
-      "epoch": 1.5558194774346794,
-      "grad_norm": 0.0032942492980509996,
-      "learning_rate": 0.00019999880802626843,
-      "loss": 46.0,
-      "step": 9661
-    },
-    {
-      "epoch": 1.5559805145134669,
-      "grad_norm": 0.001327955280430615,
-      "learning_rate": 0.00019999880777924053,
-      "loss": 46.0,
-      "step": 9662
-    },
-    {
-      "epoch": 1.556141551592254,
-      "grad_norm": 0.0009405059390701354,
-      "learning_rate": 0.00019999880753218704,
-      "loss": 46.0,
-      "step": 9663
-    },
-    {
-      "epoch": 1.5563025886710415,
-      "grad_norm": 0.0016414924757555127,
-      "learning_rate": 0.00019999880728510796,
-      "loss": 46.0,
-      "step": 9664
-    },
-    {
-      "epoch": 1.5564636257498288,
-      "grad_norm": 0.0015177715104073286,
-      "learning_rate": 0.00019999880703800327,
-      "loss": 46.0,
-      "step": 9665
-    },
-    {
-      "epoch": 1.5566246628286162,
-      "grad_norm": 0.005273567046970129,
-      "learning_rate": 0.000199998806790873,
-      "loss": 46.0,
-      "step": 9666
-    },
-    {
-      "epoch": 1.5567856999074037,
-      "grad_norm": 0.0012911235680803657,
-      "learning_rate": 0.00019999880654371713,
-      "loss": 46.0,
-      "step": 9667
-    },
-    {
-      "epoch": 1.5569467369861911,
-      "grad_norm": 0.00324858445674181,
-      "learning_rate": 0.00019999880629653567,
-      "loss": 46.0,
-      "step": 9668
-    },
-    {
-      "epoch": 1.5571077740649786,
-      "grad_norm": 0.0009071313543245196,
-      "learning_rate": 0.0001999988060493286,
-      "loss": 46.0,
-      "step": 9669
-    },
-    {
-      "epoch": 1.5572688111437658,
-      "grad_norm": 0.0013265524758026004,
-      "learning_rate": 0.00019999880580209593,
-      "loss": 46.0,
-      "step": 9670
-    },
-    {
-      "epoch": 1.5574298482225533,
-      "grad_norm": 0.0018497625133022666,
-      "learning_rate": 0.0001999988055548377,
-      "loss": 46.0,
-      "step": 9671
-    },
-    {
-      "epoch": 1.5575908853013405,
-      "grad_norm": 0.0012562641641125083,
-      "learning_rate": 0.00019999880530755386,
-      "loss": 46.0,
-      "step": 9672
-    },
-    {
-      "epoch": 1.557751922380128,
-      "grad_norm": 0.0012631615391001105,
-      "learning_rate": 0.0001999988050602444,
-      "loss": 46.0,
-      "step": 9673
-    },
-    {
-      "epoch": 1.5579129594589154,
-      "grad_norm": 0.0006363785942085087,
-      "learning_rate": 0.00019999880481290938,
-      "loss": 46.0,
-      "step": 9674
-    },
-    {
-      "epoch": 1.5580739965377028,
-      "grad_norm": 0.004513740073889494,
-      "learning_rate": 0.00019999880456554877,
-      "loss": 46.0,
-      "step": 9675
-    },
-    {
-      "epoch": 1.5582350336164903,
-      "grad_norm": 0.0013083830708637834,
-      "learning_rate": 0.00019999880431816253,
-      "loss": 46.0,
-      "step": 9676
-    },
-    {
-      "epoch": 1.5583960706952777,
-      "grad_norm": 0.01600608602166176,
-      "learning_rate": 0.00019999880407075074,
-      "loss": 46.0,
-      "step": 9677
-    },
-    {
-      "epoch": 1.558557107774065,
-      "grad_norm": 0.002192708197981119,
-      "learning_rate": 0.00019999880382331334,
-      "loss": 46.0,
-      "step": 9678
-    },
-    {
-      "epoch": 1.5587181448528522,
-      "grad_norm": 0.005927729420363903,
-      "learning_rate": 0.0001999988035758503,
-      "loss": 46.0,
-      "step": 9679
-    },
-    {
-      "epoch": 1.5588791819316397,
-      "grad_norm": 0.0029181777499616146,
-      "learning_rate": 0.0001999988033283617,
-      "loss": 46.0,
-      "step": 9680
-    },
-    {
-      "epoch": 1.559040219010427,
-      "grad_norm": 0.011076760478317738,
-      "learning_rate": 0.00019999880308084751,
-      "loss": 46.0,
-      "step": 9681
-    },
-    {
-      "epoch": 1.5592012560892146,
-      "grad_norm": 0.003319684648886323,
-      "learning_rate": 0.00019999880283330773,
-      "loss": 46.0,
-      "step": 9682
-    },
-    {
-      "epoch": 1.559362293168002,
-      "grad_norm": 0.0007327769417315722,
-      "learning_rate": 0.00019999880258574236,
-      "loss": 46.0,
-      "step": 9683
-    },
-    {
-      "epoch": 1.5595233302467895,
-      "grad_norm": 0.0017283943016082048,
-      "learning_rate": 0.00019999880233815138,
-      "loss": 46.0,
-      "step": 9684
-    },
-    {
-      "epoch": 1.5596843673255767,
-      "grad_norm": 0.003965584561228752,
-      "learning_rate": 0.00019999880209053478,
-      "loss": 46.0,
-      "step": 9685
-    },
-    {
-      "epoch": 1.5598454044043641,
-      "grad_norm": 0.000998963718302548,
-      "learning_rate": 0.00019999880184289263,
-      "loss": 46.0,
-      "step": 9686
-    },
-    {
-      "epoch": 1.5600064414831514,
-      "grad_norm": 0.0010566264390945435,
-      "learning_rate": 0.00019999880159522485,
-      "loss": 46.0,
-      "step": 9687
-    },
-    {
-      "epoch": 1.5601674785619388,
-      "grad_norm": 0.004201866220682859,
-      "learning_rate": 0.00019999880134753152,
-      "loss": 46.0,
-      "step": 9688
-    },
-    {
-      "epoch": 1.5603285156407263,
-      "grad_norm": 0.006156921852380037,
-      "learning_rate": 0.00019999880109981258,
-      "loss": 46.0,
-      "step": 9689
-    },
-    {
-      "epoch": 1.5604895527195137,
-      "grad_norm": 0.003583103884011507,
-      "learning_rate": 0.00019999880085206804,
-      "loss": 46.0,
-      "step": 9690
-    },
-    {
-      "epoch": 1.5606505897983012,
-      "grad_norm": 0.002841728273779154,
-      "learning_rate": 0.00019999880060429787,
-      "loss": 46.0,
-      "step": 9691
-    },
-    {
-      "epoch": 1.5608116268770884,
-      "grad_norm": 0.0031647998839616776,
-      "learning_rate": 0.00019999880035650213,
-      "loss": 46.0,
-      "step": 9692
-    },
-    {
-      "epoch": 1.5609726639558759,
-      "grad_norm": 0.006603090558201075,
-      "learning_rate": 0.0001999988001086808,
-      "loss": 46.0,
-      "step": 9693
-    },
-    {
-      "epoch": 1.561133701034663,
-      "grad_norm": 0.0007188520394265652,
-      "learning_rate": 0.0001999987998608339,
-      "loss": 46.0,
-      "step": 9694
-    },
-    {
-      "epoch": 1.5612947381134505,
-      "grad_norm": 0.0038769152015447617,
-      "learning_rate": 0.00019999879961296138,
-      "loss": 46.0,
-      "step": 9695
-    },
-    {
-      "epoch": 1.561455775192238,
-      "grad_norm": 0.006570107769221067,
-      "learning_rate": 0.00019999879936506324,
-      "loss": 46.0,
-      "step": 9696
-    },
-    {
-      "epoch": 1.5616168122710254,
-      "grad_norm": 0.0012102453038096428,
-      "learning_rate": 0.00019999879911713957,
-      "loss": 46.0,
-      "step": 9697
-    },
-    {
-      "epoch": 1.561777849349813,
-      "grad_norm": 0.0004778045986313373,
-      "learning_rate": 0.00019999879886919026,
-      "loss": 46.0,
-      "step": 9698
-    },
-    {
-      "epoch": 1.5619388864286003,
-      "grad_norm": 0.0005573790404014289,
-      "learning_rate": 0.0001999987986212154,
-      "loss": 46.0,
-      "step": 9699
-    },
-    {
-      "epoch": 1.5620999235073876,
-      "grad_norm": 0.0006665913970209658,
-      "learning_rate": 0.00019999879837321487,
-      "loss": 46.0,
-      "step": 9700
-    },
-    {
-      "epoch": 1.5622609605861748,
-      "grad_norm": 0.0038049796130508184,
-      "learning_rate": 0.00019999879812518877,
-      "loss": 46.0,
-      "step": 9701
-    },
-    {
-      "epoch": 1.5624219976649623,
-      "grad_norm": 0.007736711297184229,
-      "learning_rate": 0.0001999987978771371,
-      "loss": 46.0,
-      "step": 9702
-    },
-    {
-      "epoch": 1.5625830347437497,
-      "grad_norm": 0.0007463959627784789,
-      "learning_rate": 0.0001999987976290598,
-      "loss": 46.0,
-      "step": 9703
-    },
-    {
-      "epoch": 1.5627440718225372,
-      "grad_norm": 0.0018275681650266051,
-      "learning_rate": 0.00019999879738095695,
-      "loss": 46.0,
-      "step": 9704
-    },
-    {
-      "epoch": 1.5629051089013246,
-      "grad_norm": 0.0004914948367513716,
-      "learning_rate": 0.0001999987971328285,
-      "loss": 46.0,
-      "step": 9705
-    },
-    {
-      "epoch": 1.563066145980112,
-      "grad_norm": 0.003845184575766325,
-      "learning_rate": 0.00019999879688467444,
-      "loss": 46.0,
-      "step": 9706
-    },
-    {
-      "epoch": 1.5632271830588993,
-      "grad_norm": 0.0013122610980644822,
-      "learning_rate": 0.00019999879663649479,
-      "loss": 46.0,
-      "step": 9707
-    },
-    {
-      "epoch": 1.5633882201376867,
-      "grad_norm": 0.0017416197806596756,
-      "learning_rate": 0.00019999879638828952,
-      "loss": 46.0,
-      "step": 9708
-    },
-    {
-      "epoch": 1.563549257216474,
-      "grad_norm": 0.003517310367897153,
-      "learning_rate": 0.0001999987961400587,
-      "loss": 46.0,
-      "step": 9709
-    },
-    {
-      "epoch": 1.5637102942952614,
-      "grad_norm": 0.0009275232441723347,
-      "learning_rate": 0.00019999879589180228,
-      "loss": 46.0,
-      "step": 9710
-    },
-    {
-      "epoch": 1.5638713313740489,
-      "grad_norm": 0.0020001984667032957,
-      "learning_rate": 0.00019999879564352023,
-      "loss": 46.0,
-      "step": 9711
-    },
-    {
-      "epoch": 1.5640323684528363,
-      "grad_norm": 0.0015724546974524856,
-      "learning_rate": 0.00019999879539521262,
-      "loss": 46.0,
-      "step": 9712
-    },
-    {
-      "epoch": 1.5641934055316238,
-      "grad_norm": 0.0006613938021473587,
-      "learning_rate": 0.00019999879514687942,
-      "loss": 46.0,
-      "step": 9713
-    },
-    {
-      "epoch": 1.564354442610411,
-      "grad_norm": 0.0020729731768369675,
-      "learning_rate": 0.00019999879489852057,
-      "loss": 46.0,
-      "step": 9714
-    },
-    {
-      "epoch": 1.5645154796891985,
-      "grad_norm": 0.000949891924392432,
-      "learning_rate": 0.00019999879465013617,
-      "loss": 46.0,
-      "step": 9715
-    },
-    {
-      "epoch": 1.5646765167679857,
-      "grad_norm": 0.000744186865631491,
-      "learning_rate": 0.00019999879440172618,
-      "loss": 46.0,
-      "step": 9716
-    },
-    {
-      "epoch": 1.5648375538467731,
-      "grad_norm": 0.0011828161077573895,
-      "learning_rate": 0.00019999879415329058,
-      "loss": 46.0,
-      "step": 9717
-    },
-    {
-      "epoch": 1.5649985909255606,
-      "grad_norm": 0.0024074027314782143,
-      "learning_rate": 0.0001999987939048294,
-      "loss": 46.0,
-      "step": 9718
-    },
-    {
-      "epoch": 1.565159628004348,
-      "grad_norm": 0.003987540956586599,
-      "learning_rate": 0.0001999987936563426,
-      "loss": 46.0,
-      "step": 9719
-    },
-    {
-      "epoch": 1.5653206650831355,
-      "grad_norm": 0.004735024180263281,
-      "learning_rate": 0.00019999879340783022,
-      "loss": 46.0,
-      "step": 9720
-    },
-    {
-      "epoch": 1.5654817021619227,
-      "grad_norm": 0.0016514173476025462,
-      "learning_rate": 0.00019999879315929224,
-      "loss": 46.0,
-      "step": 9721
-    },
-    {
-      "epoch": 1.5656427392407102,
-      "grad_norm": 0.002796730725094676,
-      "learning_rate": 0.00019999879291072867,
-      "loss": 46.0,
-      "step": 9722
-    },
-    {
-      "epoch": 1.5658037763194974,
-      "grad_norm": 0.0004002567729912698,
-      "learning_rate": 0.0001999987926621395,
-      "loss": 46.0,
-      "step": 9723
-    },
-    {
-      "epoch": 1.5659648133982849,
-      "grad_norm": 0.0028831178788095713,
-      "learning_rate": 0.00019999879241352475,
-      "loss": 46.0,
-      "step": 9724
-    },
-    {
-      "epoch": 1.5661258504770723,
-      "grad_norm": 0.0030676613096147776,
-      "learning_rate": 0.00019999879216488437,
-      "loss": 46.0,
-      "step": 9725
-    },
-    {
-      "epoch": 1.5662868875558598,
-      "grad_norm": 0.0012627679388970137,
-      "learning_rate": 0.00019999879191621842,
-      "loss": 46.0,
-      "step": 9726
-    },
-    {
-      "epoch": 1.5664479246346472,
-      "grad_norm": 0.0024922697339206934,
-      "learning_rate": 0.0001999987916675269,
-      "loss": 46.0,
-      "step": 9727
-    },
-    {
-      "epoch": 1.5666089617134347,
-      "grad_norm": 0.006966681219637394,
-      "learning_rate": 0.00019999879141880975,
-      "loss": 46.0,
-      "step": 9728
-    },
-    {
-      "epoch": 1.566769998792222,
-      "grad_norm": 0.001992269651964307,
-      "learning_rate": 0.00019999879117006702,
-      "loss": 46.0,
-      "step": 9729
-    },
-    {
-      "epoch": 1.5669310358710093,
-      "grad_norm": 0.0008069348405115306,
-      "learning_rate": 0.00019999879092129867,
-      "loss": 46.0,
-      "step": 9730
-    },
-    {
-      "epoch": 1.5670920729497966,
-      "grad_norm": 0.0012568749953061342,
-      "learning_rate": 0.00019999879067250474,
-      "loss": 46.0,
-      "step": 9731
-    },
-    {
-      "epoch": 1.567253110028584,
-      "grad_norm": 0.0009445696487091482,
-      "learning_rate": 0.00019999879042368525,
-      "loss": 46.0,
-      "step": 9732
-    },
-    {
-      "epoch": 1.5674141471073715,
-      "grad_norm": 0.003530106507241726,
-      "learning_rate": 0.00019999879017484014,
-      "loss": 46.0,
-      "step": 9733
-    },
-    {
-      "epoch": 1.567575184186159,
-      "grad_norm": 0.0036179046146571636,
-      "learning_rate": 0.00019999878992596942,
-      "loss": 46.0,
-      "step": 9734
-    },
-    {
-      "epoch": 1.5677362212649464,
-      "grad_norm": 0.0049944715574383736,
-      "learning_rate": 0.00019999878967707314,
-      "loss": 46.0,
-      "step": 9735
-    },
-    {
-      "epoch": 1.5678972583437336,
-      "grad_norm": 0.0012217058101668954,
-      "learning_rate": 0.00019999878942815124,
-      "loss": 46.0,
-      "step": 9736
-    },
-    {
-      "epoch": 1.568058295422521,
-      "grad_norm": 0.010000333189964294,
-      "learning_rate": 0.00019999878917920376,
-      "loss": 46.0,
-      "step": 9737
-    },
-    {
-      "epoch": 1.5682193325013083,
-      "grad_norm": 0.002759035909548402,
-      "learning_rate": 0.00019999878893023064,
-      "loss": 46.0,
-      "step": 9738
-    },
-    {
-      "epoch": 1.5683803695800957,
-      "grad_norm": 0.0054483008570969105,
-      "learning_rate": 0.00019999878868123198,
-      "loss": 46.0,
-      "step": 9739
-    },
-    {
-      "epoch": 1.5685414066588832,
-      "grad_norm": 0.0006330199539661407,
-      "learning_rate": 0.0001999987884322077,
-      "loss": 46.0,
-      "step": 9740
-    },
-    {
-      "epoch": 1.5687024437376706,
-      "grad_norm": 0.006719228811562061,
-      "learning_rate": 0.00019999878818315785,
-      "loss": 46.0,
-      "step": 9741
-    },
-    {
-      "epoch": 1.568863480816458,
-      "grad_norm": 0.0036212641280144453,
-      "learning_rate": 0.00019999878793408238,
-      "loss": 46.0,
-      "step": 9742
-    },
-    {
-      "epoch": 1.5690245178952453,
-      "grad_norm": 0.005796676501631737,
-      "learning_rate": 0.00019999878768498132,
-      "loss": 46.0,
-      "step": 9743
-    },
-    {
-      "epoch": 1.5691855549740328,
-      "grad_norm": 0.010220407508313656,
-      "learning_rate": 0.0001999987874358547,
-      "loss": 46.0,
-      "step": 9744
-    },
-    {
-      "epoch": 1.56934659205282,
-      "grad_norm": 0.011911019682884216,
-      "learning_rate": 0.00019999878718670244,
-      "loss": 46.0,
-      "step": 9745
-    },
-    {
-      "epoch": 1.5695076291316075,
-      "grad_norm": 0.0013834377750754356,
-      "learning_rate": 0.00019999878693752462,
-      "loss": 46.0,
-      "step": 9746
-    },
-    {
-      "epoch": 1.569668666210395,
-      "grad_norm": 0.003457516198977828,
-      "learning_rate": 0.00019999878668832115,
-      "loss": 46.0,
-      "step": 9747
-    },
-    {
-      "epoch": 1.5698297032891824,
-      "grad_norm": 0.001415478065609932,
-      "learning_rate": 0.00019999878643909213,
-      "loss": 46.0,
-      "step": 9748
-    },
-    {
-      "epoch": 1.5699907403679698,
-      "grad_norm": 0.001333848456852138,
-      "learning_rate": 0.0001999987861898375,
-      "loss": 46.0,
-      "step": 9749
-    },
-    {
-      "epoch": 1.5701517774467573,
-      "grad_norm": 0.004237016197293997,
-      "learning_rate": 0.0001999987859405573,
-      "loss": 46.0,
-      "step": 9750
-    },
-    {
-      "epoch": 1.5703128145255445,
-      "grad_norm": 0.0006624659872613847,
-      "learning_rate": 0.00019999878569125148,
-      "loss": 46.0,
-      "step": 9751
-    },
-    {
-      "epoch": 1.570473851604332,
-      "grad_norm": 0.0010238528484478593,
-      "learning_rate": 0.00019999878544192009,
-      "loss": 46.0,
-      "step": 9752
-    },
-    {
-      "epoch": 1.5706348886831192,
-      "grad_norm": 0.0013221821282058954,
-      "learning_rate": 0.00019999878519256305,
-      "loss": 46.0,
-      "step": 9753
-    },
-    {
-      "epoch": 1.5707959257619066,
-      "grad_norm": 0.0014520084951072931,
-      "learning_rate": 0.00019999878494318047,
-      "loss": 46.0,
-      "step": 9754
-    },
-    {
-      "epoch": 1.570956962840694,
-      "grad_norm": 0.00646643340587616,
-      "learning_rate": 0.00019999878469377226,
-      "loss": 46.0,
-      "step": 9755
-    },
-    {
-      "epoch": 1.5711179999194815,
-      "grad_norm": 0.00045624279300682247,
-      "learning_rate": 0.0001999987844443385,
-      "loss": 46.0,
-      "step": 9756
-    },
-    {
-      "epoch": 1.571279036998269,
-      "grad_norm": 0.0010332841193303466,
-      "learning_rate": 0.00019999878419487912,
-      "loss": 46.0,
-      "step": 9757
-    },
-    {
-      "epoch": 1.5714400740770562,
-      "grad_norm": 0.0020471017342060804,
-      "learning_rate": 0.00019999878394539415,
-      "loss": 46.0,
-      "step": 9758
-    },
-    {
-      "epoch": 1.5716011111558437,
-      "grad_norm": 0.002583115128800273,
-      "learning_rate": 0.00019999878369588356,
-      "loss": 46.0,
-      "step": 9759
-    },
-    {
-      "epoch": 1.571762148234631,
-      "grad_norm": 0.0017519983230158687,
-      "learning_rate": 0.00019999878344634738,
-      "loss": 46.0,
-      "step": 9760
-    },
-    {
-      "epoch": 1.5719231853134183,
-      "grad_norm": 0.005189927294850349,
-      "learning_rate": 0.00019999878319678564,
-      "loss": 46.0,
-      "step": 9761
-    },
-    {
-      "epoch": 1.5720842223922058,
-      "grad_norm": 0.0008023937698453665,
-      "learning_rate": 0.0001999987829471983,
-      "loss": 46.0,
-      "step": 9762
-    },
-    {
-      "epoch": 1.5722452594709933,
-      "grad_norm": 0.002099481411278248,
-      "learning_rate": 0.00019999878269758535,
-      "loss": 46.0,
-      "step": 9763
-    },
-    {
-      "epoch": 1.5724062965497807,
-      "grad_norm": 0.0011411972809582949,
-      "learning_rate": 0.0001999987824479468,
-      "loss": 46.0,
-      "step": 9764
-    },
-    {
-      "epoch": 1.572567333628568,
-      "grad_norm": 0.0011199337895959616,
-      "learning_rate": 0.00019999878219828269,
-      "loss": 46.0,
-      "step": 9765
-    },
-    {
-      "epoch": 1.5727283707073554,
-      "grad_norm": 0.00040268953307531774,
-      "learning_rate": 0.00019999878194859293,
-      "loss": 46.0,
-      "step": 9766
-    },
-    {
-      "epoch": 1.5728894077861426,
-      "grad_norm": 0.0030387945007532835,
-      "learning_rate": 0.00019999878169887762,
-      "loss": 46.0,
-      "step": 9767
-    },
-    {
-      "epoch": 1.57305044486493,
-      "grad_norm": 0.0015354889910668135,
-      "learning_rate": 0.0001999987814491367,
-      "loss": 46.0,
-      "step": 9768
-    },
-    {
-      "epoch": 1.5732114819437175,
-      "grad_norm": 0.007252680603414774,
-      "learning_rate": 0.00019999878119937017,
-      "loss": 46.0,
-      "step": 9769
-    },
-    {
-      "epoch": 1.573372519022505,
-      "grad_norm": 0.0009251062292605639,
-      "learning_rate": 0.00019999878094957807,
-      "loss": 46.0,
-      "step": 9770
-    },
-    {
-      "epoch": 1.5735335561012924,
-      "grad_norm": 0.0029202389996498823,
-      "learning_rate": 0.00019999878069976035,
-      "loss": 46.0,
-      "step": 9771
-    },
-    {
-      "epoch": 1.5736945931800799,
-      "grad_norm": 0.0019986468832939863,
-      "learning_rate": 0.00019999878044991707,
-      "loss": 46.0,
-      "step": 9772
-    },
-    {
-      "epoch": 1.573855630258867,
-      "grad_norm": 0.0016948707634583116,
-      "learning_rate": 0.00019999878020004815,
-      "loss": 46.0,
-      "step": 9773
-    },
-    {
-      "epoch": 1.5740166673376543,
-      "grad_norm": 0.0024093349929898977,
-      "learning_rate": 0.00019999877995015368,
-      "loss": 46.0,
-      "step": 9774
-    },
-    {
-      "epoch": 1.5741777044164418,
-      "grad_norm": 0.0013166608987376094,
-      "learning_rate": 0.0001999987797002336,
-      "loss": 46.0,
-      "step": 9775
-    },
-    {
-      "epoch": 1.5743387414952292,
-      "grad_norm": 0.0019574237521737814,
-      "learning_rate": 0.0001999987794502879,
-      "loss": 46.0,
-      "step": 9776
-    },
-    {
-      "epoch": 1.5744997785740167,
-      "grad_norm": 0.01026909425854683,
-      "learning_rate": 0.00019999877920031666,
-      "loss": 46.0,
-      "step": 9777
-    },
-    {
-      "epoch": 1.5746608156528041,
-      "grad_norm": 0.007019911427050829,
-      "learning_rate": 0.0001999987789503198,
-      "loss": 46.0,
-      "step": 9778
-    },
-    {
-      "epoch": 1.5748218527315916,
-      "grad_norm": 0.0005509941256605089,
-      "learning_rate": 0.00019999877870029734,
-      "loss": 46.0,
-      "step": 9779
-    },
-    {
-      "epoch": 1.5749828898103788,
-      "grad_norm": 0.0015382322017103434,
-      "learning_rate": 0.00019999877845024928,
-      "loss": 46.0,
-      "step": 9780
-    },
-    {
-      "epoch": 1.5751439268891663,
-      "grad_norm": 0.0008280586916953325,
-      "learning_rate": 0.00019999877820017564,
-      "loss": 46.0,
-      "step": 9781
-    },
-    {
-      "epoch": 1.5753049639679535,
-      "grad_norm": 0.0020458255894482136,
-      "learning_rate": 0.00019999877795007638,
-      "loss": 46.0,
-      "step": 9782
-    },
-    {
-      "epoch": 1.575466001046741,
-      "grad_norm": 0.0025958814658224583,
-      "learning_rate": 0.00019999877769995153,
-      "loss": 46.0,
-      "step": 9783
-    },
-    {
-      "epoch": 1.5756270381255284,
-      "grad_norm": 0.002383159939199686,
-      "learning_rate": 0.0001999987774498011,
-      "loss": 46.0,
-      "step": 9784
-    },
-    {
-      "epoch": 1.5757880752043159,
-      "grad_norm": 0.002497895387932658,
-      "learning_rate": 0.00019999877719962508,
-      "loss": 46.0,
-      "step": 9785
-    },
-    {
-      "epoch": 1.5759491122831033,
-      "grad_norm": 0.0009028394124470651,
-      "learning_rate": 0.00019999877694942345,
-      "loss": 46.0,
-      "step": 9786
-    },
-    {
-      "epoch": 1.5761101493618905,
-      "grad_norm": 0.002269017044454813,
-      "learning_rate": 0.00019999877669919626,
-      "loss": 46.0,
-      "step": 9787
-    },
-    {
-      "epoch": 1.576271186440678,
-      "grad_norm": 0.0007879043114371598,
-      "learning_rate": 0.00019999877644894342,
-      "loss": 46.0,
-      "step": 9788
-    },
-    {
-      "epoch": 1.5764322235194652,
-      "grad_norm": 0.0005099507397972047,
-      "learning_rate": 0.00019999877619866503,
-      "loss": 46.0,
-      "step": 9789
-    },
-    {
-      "epoch": 1.5765932605982527,
-      "grad_norm": 0.0065828729420900345,
-      "learning_rate": 0.00019999877594836102,
-      "loss": 46.0,
-      "step": 9790
-    },
-    {
-      "epoch": 1.5767542976770401,
-      "grad_norm": 0.0021500701550394297,
-      "learning_rate": 0.00019999877569803142,
-      "loss": 46.0,
-      "step": 9791
-    },
-    {
-      "epoch": 1.5769153347558276,
-      "grad_norm": 0.0006644924287684262,
-      "learning_rate": 0.00019999877544767624,
-      "loss": 46.0,
-      "step": 9792
-    },
-    {
-      "epoch": 1.577076371834615,
-      "grad_norm": 0.008343099616467953,
-      "learning_rate": 0.00019999877519729547,
-      "loss": 46.0,
-      "step": 9793
-    },
-    {
-      "epoch": 1.5772374089134025,
-      "grad_norm": 0.0004977814387530088,
-      "learning_rate": 0.00019999877494688905,
-      "loss": 46.0,
-      "step": 9794
-    },
-    {
-      "epoch": 1.5773984459921897,
-      "grad_norm": 0.0005223865737207234,
-      "learning_rate": 0.00019999877469645708,
-      "loss": 46.0,
-      "step": 9795
-    },
-    {
-      "epoch": 1.577559483070977,
-      "grad_norm": 0.004047384485602379,
-      "learning_rate": 0.00019999877444599952,
-      "loss": 46.0,
-      "step": 9796
-    },
-    {
-      "epoch": 1.5777205201497644,
-      "grad_norm": 0.0023688289802521467,
-      "learning_rate": 0.00019999877419551635,
-      "loss": 46.0,
-      "step": 9797
-    },
-    {
-      "epoch": 1.5778815572285518,
-      "grad_norm": 0.00135643663816154,
-      "learning_rate": 0.00019999877394500761,
-      "loss": 46.0,
-      "step": 9798
-    },
-    {
-      "epoch": 1.5780425943073393,
-      "grad_norm": 0.003825664985924959,
-      "learning_rate": 0.00019999877369447324,
-      "loss": 46.0,
-      "step": 9799
-    },
-    {
-      "epoch": 1.5782036313861267,
-      "grad_norm": 0.001462132902815938,
-      "learning_rate": 0.0001999987734439133,
-      "loss": 46.0,
-      "step": 9800
-    },
-    {
-      "epoch": 1.5783646684649142,
-      "grad_norm": 0.0008735788869671524,
-      "learning_rate": 0.00019999877319332775,
-      "loss": 46.0,
-      "step": 9801
-    },
-    {
-      "epoch": 1.5785257055437014,
-      "grad_norm": 0.0017034619813784957,
-      "learning_rate": 0.00019999877294271662,
-      "loss": 46.0,
-      "step": 9802
-    },
-    {
-      "epoch": 1.5786867426224889,
-      "grad_norm": 0.0009953982662409544,
-      "learning_rate": 0.00019999877269207992,
-      "loss": 46.0,
-      "step": 9803
-    },
-    {
-      "epoch": 1.578847779701276,
-      "grad_norm": 0.00047335383715108037,
-      "learning_rate": 0.00019999877244141758,
-      "loss": 46.0,
-      "step": 9804
-    },
-    {
-      "epoch": 1.5790088167800636,
-      "grad_norm": 0.0025080873165279627,
-      "learning_rate": 0.00019999877219072966,
-      "loss": 46.0,
-      "step": 9805
-    },
-    {
-      "epoch": 1.579169853858851,
-      "grad_norm": 0.0016478647012263536,
-      "learning_rate": 0.00019999877194001615,
-      "loss": 46.0,
-      "step": 9806
-    },
-    {
-      "epoch": 1.5793308909376385,
-      "grad_norm": 0.010133449919521809,
-      "learning_rate": 0.00019999877168927705,
-      "loss": 46.0,
-      "step": 9807
-    },
-    {
-      "epoch": 1.579491928016426,
-      "grad_norm": 0.0004886648966930807,
-      "learning_rate": 0.00019999877143851233,
-      "loss": 46.0,
-      "step": 9808
-    },
-    {
-      "epoch": 1.5796529650952131,
-      "grad_norm": 0.0027399531099945307,
-      "learning_rate": 0.00019999877118772203,
-      "loss": 46.0,
-      "step": 9809
-    },
-    {
-      "epoch": 1.5798140021740006,
-      "grad_norm": 0.0018911331426352262,
-      "learning_rate": 0.00019999877093690614,
-      "loss": 46.0,
-      "step": 9810
-    },
-    {
-      "epoch": 1.5799750392527878,
-      "grad_norm": 0.000624898646492511,
-      "learning_rate": 0.00019999877068606464,
-      "loss": 46.0,
-      "step": 9811
-    },
-    {
-      "epoch": 1.5801360763315753,
-      "grad_norm": 0.0026830527931451797,
-      "learning_rate": 0.00019999877043519758,
-      "loss": 46.0,
-      "step": 9812
-    },
-    {
-      "epoch": 1.5802971134103627,
-      "grad_norm": 0.0003536730364430696,
-      "learning_rate": 0.0001999987701843049,
-      "loss": 46.0,
-      "step": 9813
-    },
-    {
-      "epoch": 1.5804581504891502,
-      "grad_norm": 0.0011075574439018965,
-      "learning_rate": 0.00019999876993338664,
-      "loss": 46.0,
-      "step": 9814
-    },
-    {
-      "epoch": 1.5806191875679376,
-      "grad_norm": 0.0031218486838042736,
-      "learning_rate": 0.00019999876968244276,
-      "loss": 46.0,
-      "step": 9815
-    },
-    {
-      "epoch": 1.580780224646725,
-      "grad_norm": 0.0012519750744104385,
-      "learning_rate": 0.00019999876943147332,
-      "loss": 46.0,
-      "step": 9816
-    },
-    {
-      "epoch": 1.5809412617255123,
-      "grad_norm": 0.001903881086036563,
-      "learning_rate": 0.00019999876918047827,
-      "loss": 46.0,
-      "step": 9817
-    },
-    {
-      "epoch": 1.5811022988042995,
-      "grad_norm": 0.004829671699553728,
-      "learning_rate": 0.0001999987689294576,
-      "loss": 46.0,
-      "step": 9818
-    },
-    {
-      "epoch": 1.581263335883087,
-      "grad_norm": 0.0009480537264607847,
-      "learning_rate": 0.00019999876867841137,
-      "loss": 46.0,
-      "step": 9819
-    },
-    {
-      "epoch": 1.5814243729618744,
-      "grad_norm": 0.000697803741786629,
-      "learning_rate": 0.00019999876842733953,
-      "loss": 46.0,
-      "step": 9820
-    },
-    {
-      "epoch": 1.5815854100406619,
-      "grad_norm": 0.0010910104028880596,
-      "learning_rate": 0.0001999987681762421,
-      "loss": 46.0,
-      "step": 9821
-    },
-    {
-      "epoch": 1.5817464471194493,
-      "grad_norm": 0.002191560808569193,
-      "learning_rate": 0.00019999876792511906,
-      "loss": 46.0,
-      "step": 9822
-    },
-    {
-      "epoch": 1.5819074841982368,
-      "grad_norm": 0.0026845159009099007,
-      "learning_rate": 0.00019999876767397046,
-      "loss": 46.0,
-      "step": 9823
-    },
-    {
-      "epoch": 1.582068521277024,
-      "grad_norm": 0.0026525049470365047,
-      "learning_rate": 0.00019999876742279624,
-      "loss": 46.0,
-      "step": 9824
-    },
-    {
-      "epoch": 1.5822295583558115,
-      "grad_norm": 0.0050405478104949,
-      "learning_rate": 0.00019999876717159644,
-      "loss": 46.0,
-      "step": 9825
-    },
-    {
-      "epoch": 1.5823905954345987,
-      "grad_norm": 0.0020936548244208097,
-      "learning_rate": 0.00019999876692037102,
-      "loss": 46.0,
-      "step": 9826
-    },
-    {
-      "epoch": 1.5825516325133862,
-      "grad_norm": 0.0021683350205421448,
-      "learning_rate": 0.00019999876666912,
-      "loss": 46.0,
-      "step": 9827
-    },
-    {
-      "epoch": 1.5827126695921736,
-      "grad_norm": 0.00042703395592980087,
-      "learning_rate": 0.00019999876641784342,
-      "loss": 46.0,
-      "step": 9828
-    },
-    {
-      "epoch": 1.582873706670961,
-      "grad_norm": 0.0012084603076800704,
-      "learning_rate": 0.00019999876616654124,
-      "loss": 46.0,
-      "step": 9829
-    },
-    {
-      "epoch": 1.5830347437497485,
-      "grad_norm": 0.001400305307470262,
-      "learning_rate": 0.00019999876591521347,
-      "loss": 46.0,
-      "step": 9830
-    },
-    {
-      "epoch": 1.5831957808285357,
-      "grad_norm": 0.001301633776165545,
-      "learning_rate": 0.0001999987656638601,
-      "loss": 46.0,
-      "step": 9831
-    },
-    {
-      "epoch": 1.5833568179073232,
-      "grad_norm": 0.012440150603652,
-      "learning_rate": 0.0001999987654124811,
-      "loss": 46.0,
-      "step": 9832
-    },
-    {
-      "epoch": 1.5835178549861104,
-      "grad_norm": 0.001563405618071556,
-      "learning_rate": 0.00019999876516107654,
-      "loss": 46.0,
-      "step": 9833
-    },
-    {
-      "epoch": 1.5836788920648979,
-      "grad_norm": 0.0009780620457604527,
-      "learning_rate": 0.00019999876490964637,
-      "loss": 46.0,
-      "step": 9834
-    },
-    {
-      "epoch": 1.5838399291436853,
-      "grad_norm": 0.006246987264603376,
-      "learning_rate": 0.00019999876465819064,
-      "loss": 46.0,
-      "step": 9835
-    },
-    {
-      "epoch": 1.5840009662224728,
-      "grad_norm": 0.001995635684579611,
-      "learning_rate": 0.00019999876440670927,
-      "loss": 46.0,
-      "step": 9836
-    },
-    {
-      "epoch": 1.5841620033012602,
-      "grad_norm": 0.0004036130558233708,
-      "learning_rate": 0.00019999876415520234,
-      "loss": 46.0,
-      "step": 9837
-    },
-    {
-      "epoch": 1.5843230403800475,
-      "grad_norm": 0.0013316834811121225,
-      "learning_rate": 0.0001999987639036698,
-      "loss": 46.0,
-      "step": 9838
-    },
-    {
-      "epoch": 1.584484077458835,
-      "grad_norm": 0.0018691255245357752,
-      "learning_rate": 0.00019999876365211166,
-      "loss": 46.0,
-      "step": 9839
-    },
-    {
-      "epoch": 1.5846451145376221,
-      "grad_norm": 0.003369741840288043,
-      "learning_rate": 0.00019999876340052794,
-      "loss": 46.0,
-      "step": 9840
-    },
-    {
-      "epoch": 1.5848061516164096,
-      "grad_norm": 0.0050245048478245735,
-      "learning_rate": 0.00019999876314891858,
-      "loss": 46.0,
-      "step": 9841
-    },
-    {
-      "epoch": 1.584967188695197,
-      "grad_norm": 0.0013695969246327877,
-      "learning_rate": 0.0001999987628972837,
-      "loss": 46.0,
-      "step": 9842
-    },
-    {
-      "epoch": 1.5851282257739845,
-      "grad_norm": 0.0031018375884741545,
-      "learning_rate": 0.00019999876264562317,
-      "loss": 46.0,
-      "step": 9843
-    },
-    {
-      "epoch": 1.585289262852772,
-      "grad_norm": 0.008670821785926819,
-      "learning_rate": 0.00019999876239393708,
-      "loss": 46.0,
-      "step": 9844
-    },
-    {
-      "epoch": 1.5854502999315594,
-      "grad_norm": 0.001297230483032763,
-      "learning_rate": 0.00019999876214222537,
-      "loss": 46.0,
-      "step": 9845
-    },
-    {
-      "epoch": 1.5856113370103466,
-      "grad_norm": 0.005896645598113537,
-      "learning_rate": 0.00019999876189048807,
-      "loss": 46.0,
-      "step": 9846
-    },
-    {
-      "epoch": 1.585772374089134,
-      "grad_norm": 0.0035727997310459614,
-      "learning_rate": 0.00019999876163872518,
-      "loss": 46.0,
-      "step": 9847
-    },
-    {
-      "epoch": 1.5859334111679213,
-      "grad_norm": 0.000984722631983459,
-      "learning_rate": 0.00019999876138693669,
-      "loss": 46.0,
-      "step": 9848
-    },
-    {
-      "epoch": 1.5860944482467088,
-      "grad_norm": 0.005690367426723242,
-      "learning_rate": 0.00019999876113512263,
-      "loss": 46.0,
-      "step": 9849
-    },
-    {
-      "epoch": 1.5862554853254962,
-      "grad_norm": 0.0014734865399077535,
-      "learning_rate": 0.00019999876088328295,
-      "loss": 46.0,
-      "step": 9850
-    },
-    {
-      "epoch": 1.5864165224042837,
-      "grad_norm": 0.0023802241776138544,
-      "learning_rate": 0.00019999876063141767,
-      "loss": 46.0,
-      "step": 9851
-    },
-    {
-      "epoch": 1.586577559483071,
-      "grad_norm": 0.0004882747307419777,
-      "learning_rate": 0.00019999876037952682,
-      "loss": 46.0,
-      "step": 9852
-    },
-    {
-      "epoch": 1.5867385965618583,
-      "grad_norm": 0.0012015477987006307,
-      "learning_rate": 0.00019999876012761036,
-      "loss": 46.0,
-      "step": 9853
-    },
-    {
-      "epoch": 1.5868996336406458,
-      "grad_norm": 0.003879109164699912,
-      "learning_rate": 0.0001999987598756683,
-      "loss": 46.0,
-      "step": 9854
-    },
-    {
-      "epoch": 1.587060670719433,
-      "grad_norm": 0.011290845461189747,
-      "learning_rate": 0.00019999875962370064,
-      "loss": 46.0,
-      "step": 9855
-    },
-    {
-      "epoch": 1.5872217077982205,
-      "grad_norm": 0.000992082292214036,
-      "learning_rate": 0.00019999875937170742,
-      "loss": 46.0,
-      "step": 9856
-    },
-    {
-      "epoch": 1.587382744877008,
-      "grad_norm": 0.0017628453206270933,
-      "learning_rate": 0.00019999875911968858,
-      "loss": 46.0,
-      "step": 9857
-    },
-    {
-      "epoch": 1.5875437819557954,
-      "grad_norm": 0.0018435249803587794,
-      "learning_rate": 0.00019999875886764413,
-      "loss": 46.0,
-      "step": 9858
-    },
-    {
-      "epoch": 1.5877048190345828,
-      "grad_norm": 0.006634776014834642,
-      "learning_rate": 0.00019999875861557415,
-      "loss": 46.0,
-      "step": 9859
-    },
-    {
-      "epoch": 1.58786585611337,
-      "grad_norm": 0.0014659801963716745,
-      "learning_rate": 0.0001999987583634785,
-      "loss": 46.0,
-      "step": 9860
-    },
-    {
-      "epoch": 1.5880268931921575,
-      "grad_norm": 0.00515702273696661,
-      "learning_rate": 0.00019999875811135728,
-      "loss": 46.0,
-      "step": 9861
-    },
-    {
-      "epoch": 1.5881879302709447,
-      "grad_norm": 0.007606877014040947,
-      "learning_rate": 0.0001999987578592105,
-      "loss": 46.0,
-      "step": 9862
-    },
-    {
-      "epoch": 1.5883489673497322,
-      "grad_norm": 0.0006977077573537827,
-      "learning_rate": 0.0001999987576070381,
-      "loss": 46.0,
-      "step": 9863
-    },
-    {
-      "epoch": 1.5885100044285196,
-      "grad_norm": 0.006941099185496569,
-      "learning_rate": 0.0001999987573548401,
-      "loss": 46.0,
-      "step": 9864
-    },
-    {
-      "epoch": 1.588671041507307,
-      "grad_norm": 0.0014213366666808724,
-      "learning_rate": 0.00019999875710261647,
-      "loss": 46.0,
-      "step": 9865
-    },
-    {
-      "epoch": 1.5888320785860945,
-      "grad_norm": 0.0023262694012373686,
-      "learning_rate": 0.0001999987568503673,
-      "loss": 46.0,
-      "step": 9866
-    },
-    {
-      "epoch": 1.588993115664882,
-      "grad_norm": 0.0031400443986058235,
-      "learning_rate": 0.00019999875659809253,
-      "loss": 46.0,
-      "step": 9867
-    },
-    {
-      "epoch": 1.5891541527436692,
-      "grad_norm": 0.0010237874230369925,
-      "learning_rate": 0.00019999875634579215,
-      "loss": 46.0,
-      "step": 9868
-    },
-    {
-      "epoch": 1.5893151898224567,
-      "grad_norm": 0.0031135983299463987,
-      "learning_rate": 0.0001999987560934662,
-      "loss": 46.0,
-      "step": 9869
-    },
-    {
-      "epoch": 1.589476226901244,
-      "grad_norm": 0.007396283093839884,
-      "learning_rate": 0.00019999875584111464,
-      "loss": 46.0,
-      "step": 9870
-    },
-    {
-      "epoch": 1.5896372639800314,
-      "grad_norm": 0.0020296175498515368,
-      "learning_rate": 0.00019999875558873744,
-      "loss": 46.0,
-      "step": 9871
-    },
-    {
-      "epoch": 1.5897983010588188,
-      "grad_norm": 0.0007125104311853647,
-      "learning_rate": 0.0001999987553363347,
-      "loss": 46.0,
-      "step": 9872
-    },
-    {
-      "epoch": 1.5899593381376063,
-      "grad_norm": 0.0018272243905812502,
-      "learning_rate": 0.00019999875508390635,
-      "loss": 46.0,
-      "step": 9873
-    },
-    {
-      "epoch": 1.5901203752163937,
-      "grad_norm": 0.0014895439380779862,
-      "learning_rate": 0.0001999987548314524,
-      "loss": 46.0,
-      "step": 9874
-    },
-    {
-      "epoch": 1.590281412295181,
-      "grad_norm": 0.001105980365537107,
-      "learning_rate": 0.00019999875457897285,
-      "loss": 46.0,
-      "step": 9875
-    },
-    {
-      "epoch": 1.5904424493739684,
-      "grad_norm": 0.0046920799650251865,
-      "learning_rate": 0.00019999875432646772,
-      "loss": 46.0,
-      "step": 9876
-    },
-    {
-      "epoch": 1.5906034864527556,
-      "grad_norm": 0.0036386162973940372,
-      "learning_rate": 0.00019999875407393703,
-      "loss": 46.0,
-      "step": 9877
-    },
-    {
-      "epoch": 1.590764523531543,
-      "grad_norm": 0.002501716371625662,
-      "learning_rate": 0.0001999987538213807,
-      "loss": 46.0,
-      "step": 9878
-    },
-    {
-      "epoch": 1.5909255606103305,
-      "grad_norm": 0.006380944512784481,
-      "learning_rate": 0.00019999875356879878,
-      "loss": 46.0,
-      "step": 9879
-    },
-    {
-      "epoch": 1.591086597689118,
-      "grad_norm": 0.00047685258323326707,
-      "learning_rate": 0.00019999875331619125,
-      "loss": 46.0,
-      "step": 9880
-    },
-    {
-      "epoch": 1.5912476347679054,
-      "grad_norm": 0.0011004656553268433,
-      "learning_rate": 0.00019999875306355816,
-      "loss": 46.0,
-      "step": 9881
-    },
-    {
-      "epoch": 1.5914086718466927,
-      "grad_norm": 0.0031154248863458633,
-      "learning_rate": 0.00019999875281089945,
-      "loss": 46.0,
-      "step": 9882
-    },
-    {
-      "epoch": 1.59156970892548,
-      "grad_norm": 0.002049099886789918,
-      "learning_rate": 0.00019999875255821516,
-      "loss": 46.0,
-      "step": 9883
-    },
-    {
-      "epoch": 1.5917307460042673,
-      "grad_norm": 0.0016456427983939648,
-      "learning_rate": 0.00019999875230550528,
-      "loss": 46.0,
-      "step": 9884
-    },
-    {
-      "epoch": 1.5918917830830548,
-      "grad_norm": 0.00288618472404778,
-      "learning_rate": 0.00019999875205276978,
-      "loss": 46.0,
-      "step": 9885
-    },
-    {
-      "epoch": 1.5920528201618422,
-      "grad_norm": 0.0011149515630677342,
-      "learning_rate": 0.0001999987518000087,
-      "loss": 46.0,
-      "step": 9886
-    },
-    {
-      "epoch": 1.5922138572406297,
-      "grad_norm": 0.0018803334096446633,
-      "learning_rate": 0.00019999875154722206,
-      "loss": 46.0,
-      "step": 9887
-    },
-    {
-      "epoch": 1.5923748943194171,
-      "grad_norm": 0.0047470806166529655,
-      "learning_rate": 0.00019999875129440977,
-      "loss": 46.0,
-      "step": 9888
-    },
-    {
-      "epoch": 1.5925359313982046,
-      "grad_norm": 0.0011762600624933839,
-      "learning_rate": 0.0001999987510415719,
-      "loss": 46.0,
-      "step": 9889
-    },
-    {
-      "epoch": 1.5926969684769918,
-      "grad_norm": 0.0022575596813112497,
-      "learning_rate": 0.00019999875078870844,
-      "loss": 46.0,
-      "step": 9890
-    },
-    {
-      "epoch": 1.592858005555779,
-      "grad_norm": 0.0043528713285923,
-      "learning_rate": 0.0001999987505358194,
-      "loss": 46.0,
-      "step": 9891
-    },
-    {
-      "epoch": 1.5930190426345665,
-      "grad_norm": 0.00403574900701642,
-      "learning_rate": 0.00019999875028290477,
-      "loss": 46.0,
-      "step": 9892
-    },
-    {
-      "epoch": 1.593180079713354,
-      "grad_norm": 0.0010333933169022202,
-      "learning_rate": 0.00019999875002996452,
-      "loss": 46.0,
-      "step": 9893
-    },
-    {
-      "epoch": 1.5933411167921414,
-      "grad_norm": 0.00641175452619791,
-      "learning_rate": 0.00019999874977699868,
-      "loss": 46.0,
-      "step": 9894
-    },
-    {
-      "epoch": 1.5935021538709289,
-      "grad_norm": 0.0007846063235774636,
-      "learning_rate": 0.00019999874952400726,
-      "loss": 46.0,
-      "step": 9895
-    },
-    {
-      "epoch": 1.5936631909497163,
-      "grad_norm": 0.003928713966161013,
-      "learning_rate": 0.00019999874927099023,
-      "loss": 46.0,
-      "step": 9896
-    },
-    {
-      "epoch": 1.5938242280285035,
-      "grad_norm": 0.0083048976957798,
-      "learning_rate": 0.0001999987490179476,
-      "loss": 46.0,
-      "step": 9897
-    },
-    {
-      "epoch": 1.593985265107291,
-      "grad_norm": 0.003947540186345577,
-      "learning_rate": 0.0001999987487648794,
-      "loss": 46.0,
-      "step": 9898
-    },
-    {
-      "epoch": 1.5941463021860782,
-      "grad_norm": 0.0018085397314280272,
-      "learning_rate": 0.0001999987485117856,
-      "loss": 46.0,
-      "step": 9899
-    },
-    {
-      "epoch": 1.5943073392648657,
-      "grad_norm": 0.0025369005743414164,
-      "learning_rate": 0.00019999874825866621,
-      "loss": 46.0,
-      "step": 9900
-    },
-    {
-      "epoch": 1.5944683763436531,
-      "grad_norm": 0.0011926913866773248,
-      "learning_rate": 0.00019999874800552122,
-      "loss": 46.0,
-      "step": 9901
-    },
-    {
-      "epoch": 1.5946294134224406,
-      "grad_norm": 0.0006960754981264472,
-      "learning_rate": 0.0001999987477523506,
-      "loss": 46.0,
-      "step": 9902
-    },
-    {
-      "epoch": 1.594790450501228,
-      "grad_norm": 0.0019860018510371447,
-      "learning_rate": 0.00019999874749915443,
-      "loss": 46.0,
-      "step": 9903
-    },
-    {
-      "epoch": 1.5949514875800153,
-      "grad_norm": 0.0046670278534293175,
-      "learning_rate": 0.00019999874724593264,
-      "loss": 46.0,
-      "step": 9904
-    },
-    {
-      "epoch": 1.5951125246588027,
-      "grad_norm": 0.004139245953410864,
-      "learning_rate": 0.00019999874699268527,
-      "loss": 46.0,
-      "step": 9905
-    },
-    {
-      "epoch": 1.59527356173759,
-      "grad_norm": 0.01147120725363493,
-      "learning_rate": 0.0001999987467394123,
-      "loss": 46.0,
-      "step": 9906
-    },
-    {
-      "epoch": 1.5954345988163774,
-      "grad_norm": 0.0006726644351147115,
-      "learning_rate": 0.00019999874648611373,
-      "loss": 46.0,
-      "step": 9907
-    },
-    {
-      "epoch": 1.5955956358951648,
-      "grad_norm": 0.0037660591769963503,
-      "learning_rate": 0.00019999874623278957,
-      "loss": 46.0,
-      "step": 9908
-    },
-    {
-      "epoch": 1.5957566729739523,
-      "grad_norm": 0.004735119640827179,
-      "learning_rate": 0.00019999874597943982,
-      "loss": 46.0,
-      "step": 9909
-    },
-    {
-      "epoch": 1.5959177100527397,
-      "grad_norm": 0.0009075236739590764,
-      "learning_rate": 0.0001999987457260645,
-      "loss": 46.0,
-      "step": 9910
-    },
-    {
-      "epoch": 1.5960787471315272,
-      "grad_norm": 0.004769476596266031,
-      "learning_rate": 0.00019999874547266356,
-      "loss": 46.0,
-      "step": 9911
-    },
-    {
-      "epoch": 1.5962397842103144,
-      "grad_norm": 0.0024767101276665926,
-      "learning_rate": 0.000199998745219237,
-      "loss": 46.0,
-      "step": 9912
-    },
-    {
-      "epoch": 1.5964008212891017,
-      "grad_norm": 0.00627156812697649,
-      "learning_rate": 0.00019999874496578487,
-      "loss": 46.0,
-      "step": 9913
-    },
-    {
-      "epoch": 1.596561858367889,
-      "grad_norm": 0.0011349557898938656,
-      "learning_rate": 0.00019999874471230715,
-      "loss": 46.0,
-      "step": 9914
-    },
-    {
-      "epoch": 1.5967228954466766,
-      "grad_norm": 0.0034325974993407726,
-      "learning_rate": 0.00019999874445880385,
-      "loss": 46.0,
-      "step": 9915
-    },
-    {
-      "epoch": 1.596883932525464,
-      "grad_norm": 0.003224226413294673,
-      "learning_rate": 0.0001999987442052749,
-      "loss": 46.0,
-      "step": 9916
-    },
-    {
-      "epoch": 1.5970449696042515,
-      "grad_norm": 0.00788784772157669,
-      "learning_rate": 0.0001999987439517204,
-      "loss": 46.0,
-      "step": 9917
-    },
-    {
-      "epoch": 1.597206006683039,
-      "grad_norm": 0.0029907876159995794,
-      "learning_rate": 0.00019999874369814032,
-      "loss": 46.0,
-      "step": 9918
-    },
-    {
-      "epoch": 1.5973670437618261,
-      "grad_norm": 0.0005910294130444527,
-      "learning_rate": 0.0001999987434445346,
-      "loss": 46.0,
-      "step": 9919
-    },
-    {
-      "epoch": 1.5975280808406136,
-      "grad_norm": 0.002190010854974389,
-      "learning_rate": 0.0001999987431909033,
-      "loss": 46.0,
-      "step": 9920
-    },
-    {
-      "epoch": 1.5976891179194008,
-      "grad_norm": 0.0011697777081280947,
-      "learning_rate": 0.00019999874293724642,
-      "loss": 46.0,
-      "step": 9921
-    },
-    {
-      "epoch": 1.5978501549981883,
-      "grad_norm": 0.004586322233080864,
-      "learning_rate": 0.00019999874268356393,
-      "loss": 46.0,
-      "step": 9922
-    },
-    {
-      "epoch": 1.5980111920769757,
-      "grad_norm": 0.0015874715754762292,
-      "learning_rate": 0.00019999874242985585,
-      "loss": 46.0,
-      "step": 9923
-    },
-    {
-      "epoch": 1.5981722291557632,
-      "grad_norm": 0.0033024463336914778,
-      "learning_rate": 0.0001999987421761222,
-      "loss": 46.0,
-      "step": 9924
-    },
-    {
-      "epoch": 1.5983332662345506,
-      "grad_norm": 0.0003507027868181467,
-      "learning_rate": 0.00019999874192236293,
-      "loss": 46.0,
-      "step": 9925
-    },
-    {
-      "epoch": 1.5984943033133379,
-      "grad_norm": 0.002473590662702918,
-      "learning_rate": 0.00019999874166857807,
-      "loss": 46.0,
-      "step": 9926
-    },
-    {
-      "epoch": 1.5986553403921253,
-      "grad_norm": 0.008915458805859089,
-      "learning_rate": 0.0001999987414147676,
-      "loss": 46.0,
-      "step": 9927
-    },
-    {
-      "epoch": 1.5988163774709125,
-      "grad_norm": 0.003729333635419607,
-      "learning_rate": 0.00019999874116093157,
-      "loss": 46.0,
-      "step": 9928
-    },
-    {
-      "epoch": 1.5989774145497,
-      "grad_norm": 0.001410358352586627,
-      "learning_rate": 0.0001999987409070699,
-      "loss": 46.0,
-      "step": 9929
-    },
-    {
-      "epoch": 1.5991384516284874,
-      "grad_norm": 0.002386793028563261,
-      "learning_rate": 0.0001999987406531827,
-      "loss": 46.0,
-      "step": 9930
-    },
-    {
-      "epoch": 1.599299488707275,
-      "grad_norm": 0.003954610321670771,
-      "learning_rate": 0.00019999874039926984,
-      "loss": 46.0,
-      "step": 9931
-    },
-    {
-      "epoch": 1.5994605257860623,
-      "grad_norm": 0.013110685162246227,
-      "learning_rate": 0.00019999874014533142,
-      "loss": 46.0,
-      "step": 9932
-    },
-    {
-      "epoch": 1.5996215628648496,
-      "grad_norm": 0.006606991868466139,
-      "learning_rate": 0.00019999873989136742,
-      "loss": 46.0,
-      "step": 9933
-    },
-    {
-      "epoch": 1.599782599943637,
-      "grad_norm": 0.005766377784311771,
-      "learning_rate": 0.0001999987396373778,
-      "loss": 46.0,
-      "step": 9934
-    },
-    {
-      "epoch": 1.5999436370224243,
-      "grad_norm": 0.0013292814837768674,
-      "learning_rate": 0.00019999873938336257,
-      "loss": 46.0,
-      "step": 9935
-    },
-    {
-      "epoch": 1.6001046741012117,
-      "grad_norm": 0.0008554612868465483,
-      "learning_rate": 0.0001999987391293218,
-      "loss": 46.0,
-      "step": 9936
-    },
-    {
-      "epoch": 1.6002657111799992,
-      "grad_norm": 0.0012830928899347782,
-      "learning_rate": 0.00019999873887525536,
-      "loss": 46.0,
-      "step": 9937
-    },
-    {
-      "epoch": 1.6004267482587866,
-      "grad_norm": 0.004417533054947853,
-      "learning_rate": 0.0001999987386211634,
-      "loss": 46.0,
-      "step": 9938
-    },
-    {
-      "epoch": 1.600587785337574,
-      "grad_norm": 0.00232595088891685,
-      "learning_rate": 0.00019999873836704581,
-      "loss": 46.0,
-      "step": 9939
-    },
-    {
-      "epoch": 1.6007488224163615,
-      "grad_norm": 0.004983017686754465,
-      "learning_rate": 0.00019999873811290262,
-      "loss": 46.0,
-      "step": 9940
-    },
-    {
-      "epoch": 1.6009098594951487,
-      "grad_norm": 0.0014203991740942001,
-      "learning_rate": 0.00019999873785873383,
-      "loss": 46.0,
-      "step": 9941
-    },
-    {
-      "epoch": 1.6010708965739362,
-      "grad_norm": 0.001022123615257442,
-      "learning_rate": 0.00019999873760453946,
-      "loss": 46.0,
-      "step": 9942
-    },
-    {
-      "epoch": 1.6012319336527234,
-      "grad_norm": 0.0029021401423960924,
-      "learning_rate": 0.0001999987373503195,
-      "loss": 46.0,
-      "step": 9943
-    },
-    {
-      "epoch": 1.6013929707315109,
-      "grad_norm": 0.0009786970913410187,
-      "learning_rate": 0.00019999873709607393,
-      "loss": 46.0,
-      "step": 9944
-    },
-    {
-      "epoch": 1.6015540078102983,
-      "grad_norm": 0.000947914901189506,
-      "learning_rate": 0.00019999873684180277,
-      "loss": 46.0,
-      "step": 9945
-    },
-    {
-      "epoch": 1.6017150448890858,
-      "grad_norm": 0.0036358211655169725,
-      "learning_rate": 0.00019999873658750603,
-      "loss": 46.0,
-      "step": 9946
-    },
-    {
-      "epoch": 1.6018760819678732,
-      "grad_norm": 0.004785642959177494,
-      "learning_rate": 0.00019999873633318367,
-      "loss": 46.0,
-      "step": 9947
-    },
-    {
-      "epoch": 1.6020371190466605,
-      "grad_norm": 0.0014921730617061257,
-      "learning_rate": 0.00019999873607883575,
-      "loss": 46.0,
-      "step": 9948
-    },
-    {
-      "epoch": 1.602198156125448,
-      "grad_norm": 0.004147863946855068,
-      "learning_rate": 0.0001999987358244622,
-      "loss": 46.0,
-      "step": 9949
-    },
-    {
-      "epoch": 1.6023591932042351,
-      "grad_norm": 0.001713430043309927,
-      "learning_rate": 0.0001999987355700631,
-      "loss": 46.0,
-      "step": 9950
-    },
-    {
-      "epoch": 1.6025202302830226,
-      "grad_norm": 0.0037970435805618763,
-      "learning_rate": 0.00019999873531563835,
-      "loss": 46.0,
-      "step": 9951
-    },
-    {
-      "epoch": 1.60268126736181,
-      "grad_norm": 0.00548350578173995,
-      "learning_rate": 0.00019999873506118806,
-      "loss": 46.0,
-      "step": 9952
-    },
-    {
-      "epoch": 1.6028423044405975,
-      "grad_norm": 0.0020618895068764687,
-      "learning_rate": 0.00019999873480671215,
-      "loss": 46.0,
-      "step": 9953
-    },
-    {
-      "epoch": 1.603003341519385,
-      "grad_norm": 0.0013701700372621417,
-      "learning_rate": 0.00019999873455221062,
-      "loss": 46.0,
-      "step": 9954
-    },
-    {
-      "epoch": 1.6031643785981722,
-      "grad_norm": 0.0019201809773221612,
-      "learning_rate": 0.0001999987342976835,
-      "loss": 46.0,
-      "step": 9955
-    },
-    {
-      "epoch": 1.6033254156769596,
-      "grad_norm": 0.0008332083816640079,
-      "learning_rate": 0.0001999987340431308,
-      "loss": 46.0,
-      "step": 9956
-    },
-    {
-      "epoch": 1.6034864527557469,
-      "grad_norm": 0.0005230630631558597,
-      "learning_rate": 0.00019999873378855255,
-      "loss": 46.0,
-      "step": 9957
-    },
-    {
-      "epoch": 1.6036474898345343,
-      "grad_norm": 0.0019583487883210182,
-      "learning_rate": 0.00019999873353394865,
-      "loss": 46.0,
-      "step": 9958
-    },
-    {
-      "epoch": 1.6038085269133218,
-      "grad_norm": 0.001861340133473277,
-      "learning_rate": 0.00019999873327931916,
-      "loss": 46.0,
-      "step": 9959
-    },
-    {
-      "epoch": 1.6039695639921092,
-      "grad_norm": 0.00693413196131587,
-      "learning_rate": 0.00019999873302466409,
-      "loss": 46.0,
-      "step": 9960
-    },
-    {
-      "epoch": 1.6041306010708967,
-      "grad_norm": 0.0014955116203054786,
-      "learning_rate": 0.00019999873276998342,
-      "loss": 46.0,
-      "step": 9961
-    },
-    {
-      "epoch": 1.6042916381496841,
-      "grad_norm": 0.0011993676889687777,
-      "learning_rate": 0.00019999873251527717,
-      "loss": 46.0,
-      "step": 9962
-    },
-    {
-      "epoch": 1.6044526752284713,
-      "grad_norm": 0.0018362307455390692,
-      "learning_rate": 0.0001999987322605453,
-      "loss": 46.0,
-      "step": 9963
-    },
-    {
-      "epoch": 1.6046137123072588,
-      "grad_norm": 0.0010144029511138797,
-      "learning_rate": 0.00019999873200578786,
-      "loss": 46.0,
-      "step": 9964
-    },
-    {
-      "epoch": 1.604774749386046,
-      "grad_norm": 0.0015136083820834756,
-      "learning_rate": 0.00019999873175100482,
-      "loss": 46.0,
-      "step": 9965
-    },
-    {
-      "epoch": 1.6049357864648335,
-      "grad_norm": 0.0010237840469926596,
-      "learning_rate": 0.00019999873149619617,
-      "loss": 46.0,
-      "step": 9966
-    },
-    {
-      "epoch": 1.605096823543621,
-      "grad_norm": 0.0012998317833989859,
-      "learning_rate": 0.00019999873124136193,
-      "loss": 46.0,
-      "step": 9967
-    },
-    {
-      "epoch": 1.6052578606224084,
-      "grad_norm": 0.0008958845282904804,
-      "learning_rate": 0.00019999873098650207,
-      "loss": 46.0,
-      "step": 9968
-    },
-    {
-      "epoch": 1.6054188977011958,
-      "grad_norm": 0.005237857345491648,
-      "learning_rate": 0.0001999987307316167,
-      "loss": 46.0,
-      "step": 9969
-    },
-    {
-      "epoch": 1.605579934779983,
-      "grad_norm": 0.0029390875715762377,
-      "learning_rate": 0.00019999873047670566,
-      "loss": 46.0,
-      "step": 9970
-    },
-    {
-      "epoch": 1.6057409718587705,
-      "grad_norm": 0.00178729637991637,
-      "learning_rate": 0.00019999873022176904,
-      "loss": 46.0,
-      "step": 9971
-    },
-    {
-      "epoch": 1.6059020089375577,
-      "grad_norm": 0.0023640634026378393,
-      "learning_rate": 0.00019999872996680681,
-      "loss": 46.0,
-      "step": 9972
-    },
-    {
-      "epoch": 1.6060630460163452,
-      "grad_norm": 0.0033357671927660704,
-      "learning_rate": 0.00019999872971181902,
-      "loss": 46.0,
-      "step": 9973
-    },
-    {
-      "epoch": 1.6062240830951326,
-      "grad_norm": 0.0024236023891717196,
-      "learning_rate": 0.00019999872945680562,
-      "loss": 46.0,
-      "step": 9974
-    },
-    {
-      "epoch": 1.60638512017392,
-      "grad_norm": 0.0005014078342355788,
-      "learning_rate": 0.00019999872920176663,
-      "loss": 46.0,
-      "step": 9975
-    },
-    {
-      "epoch": 1.6065461572527076,
-      "grad_norm": 0.007713499013334513,
-      "learning_rate": 0.00019999872894670205,
-      "loss": 46.0,
-      "step": 9976
-    },
-    {
-      "epoch": 1.6067071943314948,
-      "grad_norm": 0.0033008288592100143,
-      "learning_rate": 0.00019999872869161186,
-      "loss": 46.0,
-      "step": 9977
-    },
-    {
-      "epoch": 1.6068682314102822,
-      "grad_norm": 0.005757942795753479,
-      "learning_rate": 0.0001999987284364961,
-      "loss": 46.0,
-      "step": 9978
-    },
-    {
-      "epoch": 1.6070292684890695,
-      "grad_norm": 0.003667326644062996,
-      "learning_rate": 0.0001999987281813547,
-      "loss": 46.0,
-      "step": 9979
-    },
-    {
-      "epoch": 1.607190305567857,
-      "grad_norm": 0.002007820876315236,
-      "learning_rate": 0.00019999872792618776,
-      "loss": 46.0,
-      "step": 9980
-    },
-    {
-      "epoch": 1.6073513426466444,
-      "grad_norm": 0.006188342813402414,
-      "learning_rate": 0.0001999987276709952,
-      "loss": 46.0,
-      "step": 9981
-    },
-    {
-      "epoch": 1.6075123797254318,
-      "grad_norm": 0.003902670694515109,
-      "learning_rate": 0.00019999872741577703,
-      "loss": 46.0,
-      "step": 9982
-    },
-    {
-      "epoch": 1.6076734168042193,
-      "grad_norm": 0.001671084901317954,
-      "learning_rate": 0.0001999987271605333,
-      "loss": 46.0,
-      "step": 9983
-    },
-    {
-      "epoch": 1.6078344538830067,
-      "grad_norm": 0.004128395114094019,
-      "learning_rate": 0.00019999872690526393,
-      "loss": 46.0,
-      "step": 9984
-    },
-    {
-      "epoch": 1.607995490961794,
-      "grad_norm": 0.005472517106682062,
-      "learning_rate": 0.00019999872664996902,
-      "loss": 46.0,
-      "step": 9985
-    },
-    {
-      "epoch": 1.6081565280405812,
-      "grad_norm": 0.0011782434303313494,
-      "learning_rate": 0.00019999872639464848,
-      "loss": 46.0,
-      "step": 9986
-    },
-    {
-      "epoch": 1.6083175651193686,
-      "grad_norm": 0.0008484771242365241,
-      "learning_rate": 0.00019999872613930234,
-      "loss": 46.0,
-      "step": 9987
-    },
-    {
-      "epoch": 1.608478602198156,
-      "grad_norm": 0.001118054729886353,
-      "learning_rate": 0.00019999872588393063,
-      "loss": 46.0,
-      "step": 9988
-    },
-    {
-      "epoch": 1.6086396392769435,
-      "grad_norm": 0.0008306379313580692,
-      "learning_rate": 0.0001999987256285333,
-      "loss": 46.0,
-      "step": 9989
-    },
-    {
-      "epoch": 1.608800676355731,
-      "grad_norm": 0.000769666163250804,
-      "learning_rate": 0.0001999987253731104,
-      "loss": 46.0,
-      "step": 9990
-    },
-    {
-      "epoch": 1.6089617134345184,
-      "grad_norm": 0.002219146117568016,
-      "learning_rate": 0.0001999987251176619,
-      "loss": 46.0,
-      "step": 9991
-    },
-    {
-      "epoch": 1.6091227505133057,
-      "grad_norm": 0.003068450838327408,
-      "learning_rate": 0.0001999987248621878,
-      "loss": 46.0,
-      "step": 9992
-    },
-    {
-      "epoch": 1.6092837875920931,
-      "grad_norm": 0.0009180534398183227,
-      "learning_rate": 0.0001999987246066881,
-      "loss": 46.0,
-      "step": 9993
-    },
-    {
-      "epoch": 1.6094448246708803,
-      "grad_norm": 0.0008969181799329817,
-      "learning_rate": 0.00019999872435116282,
-      "loss": 46.0,
-      "step": 9994
-    },
-    {
-      "epoch": 1.6096058617496678,
-      "grad_norm": 0.0014274612767621875,
-      "learning_rate": 0.00019999872409561193,
-      "loss": 46.0,
-      "step": 9995
-    },
-    {
-      "epoch": 1.6097668988284553,
-      "grad_norm": 0.002303382847458124,
-      "learning_rate": 0.00019999872384003547,
-      "loss": 46.0,
-      "step": 9996
-    },
-    {
-      "epoch": 1.6099279359072427,
-      "grad_norm": 0.0036043638829141855,
-      "learning_rate": 0.0001999987235844334,
-      "loss": 46.0,
-      "step": 9997
-    },
-    {
-      "epoch": 1.6100889729860302,
-      "grad_norm": 0.002218181500211358,
-      "learning_rate": 0.0001999987233288057,
-      "loss": 46.0,
-      "step": 9998
-    },
-    {
-      "epoch": 1.6102500100648174,
-      "grad_norm": 0.0015294477343559265,
-      "learning_rate": 0.00019999872307315247,
-      "loss": 46.0,
-      "step": 9999
-    },
-    {
-      "epoch": 1.6104110471436048,
-      "grad_norm": 0.0011496507795527577,
-      "learning_rate": 0.0001999987228174736,
-      "loss": 46.0,
-      "step": 10000
-    },
-    {
-      "epoch": 1.610572084222392,
-      "grad_norm": 0.003749253461137414,
-      "learning_rate": 0.00019999872256176916,
-      "loss": 46.0,
-      "step": 10001
-    },
-    {
-      "epoch": 1.6107331213011795,
-      "grad_norm": 0.006967192515730858,
-      "learning_rate": 0.00019999872230603912,
-      "loss": 46.0,
-      "step": 10002
-    },
-    {
-      "epoch": 1.610894158379967,
-      "grad_norm": 0.003556340467184782,
-      "learning_rate": 0.00019999872205028347,
-      "loss": 46.0,
-      "step": 10003
-    },
-    {
-      "epoch": 1.6110551954587544,
-      "grad_norm": 0.002457259688526392,
-      "learning_rate": 0.00019999872179450224,
-      "loss": 46.0,
-      "step": 10004
-    },
-    {
-      "epoch": 1.6112162325375419,
-      "grad_norm": 0.0020998804830014706,
-      "learning_rate": 0.00019999872153869541,
-      "loss": 46.0,
-      "step": 10005
-    },
-    {
-      "epoch": 1.6113772696163293,
-      "grad_norm": 0.0010032132267951965,
-      "learning_rate": 0.00019999872128286298,
-      "loss": 46.0,
-      "step": 10006
-    },
-    {
-      "epoch": 1.6115383066951166,
-      "grad_norm": 0.0012266855919733644,
-      "learning_rate": 0.00019999872102700495,
-      "loss": 46.0,
-      "step": 10007
-    },
-    {
-      "epoch": 1.6116993437739038,
-      "grad_norm": 0.0028242762200534344,
-      "learning_rate": 0.00019999872077112137,
-      "loss": 46.0,
-      "step": 10008
-    },
-    {
-      "epoch": 1.6118603808526912,
-      "grad_norm": 0.0016357668209820986,
-      "learning_rate": 0.00019999872051521217,
-      "loss": 46.0,
-      "step": 10009
-    },
-    {
-      "epoch": 1.6120214179314787,
-      "grad_norm": 0.0021341571118682623,
-      "learning_rate": 0.00019999872025927733,
-      "loss": 46.0,
-      "step": 10010
-    },
-    {
-      "epoch": 1.6121824550102661,
-      "grad_norm": 0.002838290296494961,
-      "learning_rate": 0.00019999872000331693,
-      "loss": 46.0,
-      "step": 10011
-    },
-    {
-      "epoch": 1.6123434920890536,
-      "grad_norm": 0.0021008311305195093,
-      "learning_rate": 0.00019999871974733094,
-      "loss": 46.0,
-      "step": 10012
-    },
-    {
-      "epoch": 1.612504529167841,
-      "grad_norm": 0.0010792015818879008,
-      "learning_rate": 0.00019999871949131937,
-      "loss": 46.0,
-      "step": 10013
-    },
-    {
-      "epoch": 1.6126655662466283,
-      "grad_norm": 0.008334732614457607,
-      "learning_rate": 0.0001999987192352822,
-      "loss": 46.0,
-      "step": 10014
-    },
-    {
-      "epoch": 1.6128266033254157,
-      "grad_norm": 0.017220793291926384,
-      "learning_rate": 0.0001999987189792194,
-      "loss": 46.0,
-      "step": 10015
-    },
-    {
-      "epoch": 1.612987640404203,
-      "grad_norm": 0.0006704721599817276,
-      "learning_rate": 0.00019999871872313106,
-      "loss": 46.0,
-      "step": 10016
-    },
-    {
-      "epoch": 1.6131486774829904,
-      "grad_norm": 0.0012759979581460357,
-      "learning_rate": 0.0001999987184670171,
-      "loss": 46.0,
-      "step": 10017
-    },
-    {
-      "epoch": 1.6133097145617779,
-      "grad_norm": 0.001640747650526464,
-      "learning_rate": 0.00019999871821087752,
-      "loss": 46.0,
-      "step": 10018
-    },
-    {
-      "epoch": 1.6134707516405653,
-      "grad_norm": 0.001990459393709898,
-      "learning_rate": 0.0001999987179547124,
-      "loss": 46.0,
-      "step": 10019
-    },
-    {
-      "epoch": 1.6136317887193528,
-      "grad_norm": 0.004355066921561956,
-      "learning_rate": 0.00019999871769852163,
-      "loss": 46.0,
-      "step": 10020
-    },
-    {
-      "epoch": 1.61379282579814,
-      "grad_norm": 0.006216387264430523,
-      "learning_rate": 0.0001999987174423053,
-      "loss": 46.0,
-      "step": 10021
-    },
-    {
-      "epoch": 1.6139538628769274,
-      "grad_norm": 0.0019049688708037138,
-      "learning_rate": 0.00019999871718606334,
-      "loss": 46.0,
-      "step": 10022
-    },
-    {
-      "epoch": 1.6141148999557147,
-      "grad_norm": 0.0012330982135608792,
-      "learning_rate": 0.0001999987169297958,
-      "loss": 46.0,
-      "step": 10023
-    },
-    {
-      "epoch": 1.6142759370345021,
-      "grad_norm": 0.004897539969533682,
-      "learning_rate": 0.0001999987166735027,
-      "loss": 46.0,
-      "step": 10024
-    },
-    {
-      "epoch": 1.6144369741132896,
-      "grad_norm": 0.0013105187099426985,
-      "learning_rate": 0.00019999871641718396,
-      "loss": 46.0,
-      "step": 10025
-    },
-    {
-      "epoch": 1.614598011192077,
-      "grad_norm": 0.002436260925605893,
-      "learning_rate": 0.00019999871616083965,
-      "loss": 46.0,
-      "step": 10026
-    },
-    {
-      "epoch": 1.6147590482708645,
-      "grad_norm": 0.0014236794086173177,
-      "learning_rate": 0.00019999871590446974,
-      "loss": 46.0,
-      "step": 10027
-    },
-    {
-      "epoch": 1.614920085349652,
-      "grad_norm": 0.002821186790242791,
-      "learning_rate": 0.00019999871564807425,
-      "loss": 46.0,
-      "step": 10028
-    },
-    {
-      "epoch": 1.6150811224284392,
-      "grad_norm": 0.00046754733193665743,
-      "learning_rate": 0.00019999871539165315,
-      "loss": 46.0,
-      "step": 10029
-    },
-    {
-      "epoch": 1.6152421595072264,
-      "grad_norm": 0.0026387618854641914,
-      "learning_rate": 0.00019999871513520648,
-      "loss": 46.0,
-      "step": 10030
-    },
-    {
-      "epoch": 1.6154031965860138,
-      "grad_norm": 0.0010120307561010122,
-      "learning_rate": 0.00019999871487873415,
-      "loss": 46.0,
-      "step": 10031
-    },
-    {
-      "epoch": 1.6155642336648013,
-      "grad_norm": 0.004986585583537817,
-      "learning_rate": 0.00019999871462223628,
-      "loss": 46.0,
-      "step": 10032
-    },
-    {
-      "epoch": 1.6157252707435887,
-      "grad_norm": 0.006543920375406742,
-      "learning_rate": 0.0001999987143657128,
-      "loss": 46.0,
-      "step": 10033
-    },
-    {
-      "epoch": 1.6158863078223762,
-      "grad_norm": 0.0034259026870131493,
-      "learning_rate": 0.00019999871410916376,
-      "loss": 46.0,
-      "step": 10034
-    },
-    {
-      "epoch": 1.6160473449011636,
-      "grad_norm": 0.000730679021216929,
-      "learning_rate": 0.00019999871385258908,
-      "loss": 46.0,
-      "step": 10035
-    },
-    {
-      "epoch": 1.6162083819799509,
-      "grad_norm": 0.0012241887161508203,
-      "learning_rate": 0.00019999871359598884,
-      "loss": 46.0,
-      "step": 10036
-    },
-    {
-      "epoch": 1.6163694190587383,
-      "grad_norm": 0.0061356136575341225,
-      "learning_rate": 0.00019999871333936295,
-      "loss": 46.0,
-      "step": 10037
-    },
-    {
-      "epoch": 1.6165304561375256,
-      "grad_norm": 0.0018628608668223023,
-      "learning_rate": 0.0001999987130827115,
-      "loss": 46.0,
-      "step": 10038
-    },
-    {
-      "epoch": 1.616691493216313,
-      "grad_norm": 0.0005156780825927854,
-      "learning_rate": 0.00019999871282603445,
-      "loss": 46.0,
-      "step": 10039
-    },
-    {
-      "epoch": 1.6168525302951005,
-      "grad_norm": 0.0021635459270328283,
-      "learning_rate": 0.00019999871256933186,
-      "loss": 46.0,
-      "step": 10040
-    },
-    {
-      "epoch": 1.617013567373888,
-      "grad_norm": 0.0024970651138573885,
-      "learning_rate": 0.0001999987123126036,
-      "loss": 46.0,
-      "step": 10041
-    },
-    {
-      "epoch": 1.6171746044526754,
-      "grad_norm": 0.0034847150091081858,
-      "learning_rate": 0.00019999871205584978,
-      "loss": 46.0,
-      "step": 10042
-    },
-    {
-      "epoch": 1.6173356415314626,
-      "grad_norm": 0.009560378268361092,
-      "learning_rate": 0.00019999871179907035,
-      "loss": 46.0,
-      "step": 10043
-    },
-    {
-      "epoch": 1.61749667861025,
-      "grad_norm": 0.000757599831558764,
-      "learning_rate": 0.00019999871154226533,
-      "loss": 46.0,
-      "step": 10044
-    },
-    {
-      "epoch": 1.6176577156890373,
-      "grad_norm": 0.0008367106202058494,
-      "learning_rate": 0.00019999871128543472,
-      "loss": 46.0,
-      "step": 10045
-    },
-    {
-      "epoch": 1.6178187527678247,
-      "grad_norm": 0.0029868390411138535,
-      "learning_rate": 0.00019999871102857852,
-      "loss": 46.0,
-      "step": 10046
-    },
-    {
-      "epoch": 1.6179797898466122,
-      "grad_norm": 0.0024790638126432896,
-      "learning_rate": 0.00019999871077169674,
-      "loss": 46.0,
-      "step": 10047
-    },
-    {
-      "epoch": 1.6181408269253996,
-      "grad_norm": 0.003104433650150895,
-      "learning_rate": 0.00019999871051478934,
-      "loss": 46.0,
-      "step": 10048
-    },
-    {
-      "epoch": 1.618301864004187,
-      "grad_norm": 0.0014766334788873792,
-      "learning_rate": 0.00019999871025785636,
-      "loss": 46.0,
-      "step": 10049
-    },
-    {
-      "epoch": 1.6184629010829743,
-      "grad_norm": 0.0014610042562708259,
-      "learning_rate": 0.00019999871000089776,
-      "loss": 46.0,
-      "step": 10050
-    },
-    {
-      "epoch": 1.6186239381617618,
-      "grad_norm": 0.0005733121070079505,
-      "learning_rate": 0.0001999987097439136,
-      "loss": 46.0,
-      "step": 10051
-    },
-    {
-      "epoch": 1.618784975240549,
-      "grad_norm": 0.013445157557725906,
-      "learning_rate": 0.0001999987094869038,
-      "loss": 46.0,
-      "step": 10052
-    },
-    {
-      "epoch": 1.6189460123193364,
-      "grad_norm": 0.0010347983334213495,
-      "learning_rate": 0.00019999870922986844,
-      "loss": 46.0,
-      "step": 10053
-    },
-    {
-      "epoch": 1.6191070493981239,
-      "grad_norm": 0.0031378509011119604,
-      "learning_rate": 0.00019999870897280747,
-      "loss": 46.0,
-      "step": 10054
-    },
-    {
-      "epoch": 1.6192680864769113,
-      "grad_norm": 0.0025368474889546633,
-      "learning_rate": 0.00019999870871572093,
-      "loss": 46.0,
-      "step": 10055
-    },
-    {
-      "epoch": 1.6194291235556988,
-      "grad_norm": 0.001810396439395845,
-      "learning_rate": 0.00019999870845860879,
-      "loss": 46.0,
-      "step": 10056
-    },
-    {
-      "epoch": 1.6195901606344862,
-      "grad_norm": 0.004318727646023035,
-      "learning_rate": 0.00019999870820147102,
-      "loss": 46.0,
-      "step": 10057
-    },
-    {
-      "epoch": 1.6197511977132735,
-      "grad_norm": 0.004343224223703146,
-      "learning_rate": 0.0001999987079443077,
-      "loss": 46.0,
-      "step": 10058
-    },
-    {
-      "epoch": 1.619912234792061,
-      "grad_norm": 0.010591312311589718,
-      "learning_rate": 0.00019999870768711876,
-      "loss": 46.0,
-      "step": 10059
-    },
-    {
-      "epoch": 1.6200732718708482,
-      "grad_norm": 0.002126262290403247,
-      "learning_rate": 0.00019999870742990424,
-      "loss": 46.0,
-      "step": 10060
-    },
-    {
-      "epoch": 1.6202343089496356,
-      "grad_norm": 0.002966416534036398,
-      "learning_rate": 0.0001999987071726641,
-      "loss": 46.0,
-      "step": 10061
-    },
-    {
-      "epoch": 1.620395346028423,
-      "grad_norm": 0.0011163183953613043,
-      "learning_rate": 0.0001999987069153984,
-      "loss": 46.0,
-      "step": 10062
-    },
-    {
-      "epoch": 1.6205563831072105,
-      "grad_norm": 0.00035459158243611455,
-      "learning_rate": 0.0001999987066581071,
-      "loss": 46.0,
-      "step": 10063
-    },
-    {
-      "epoch": 1.620717420185998,
-      "grad_norm": 0.0022138794884085655,
-      "learning_rate": 0.00019999870640079016,
-      "loss": 46.0,
-      "step": 10064
-    },
-    {
-      "epoch": 1.6208784572647852,
-      "grad_norm": 0.00615911278873682,
-      "learning_rate": 0.00019999870614344768,
-      "loss": 46.0,
-      "step": 10065
-    },
-    {
-      "epoch": 1.6210394943435726,
-      "grad_norm": 0.0027898841071873903,
-      "learning_rate": 0.00019999870588607957,
-      "loss": 46.0,
-      "step": 10066
-    },
-    {
-      "epoch": 1.6212005314223599,
-      "grad_norm": 0.004338650498539209,
-      "learning_rate": 0.00019999870562868589,
-      "loss": 46.0,
-      "step": 10067
-    },
-    {
-      "epoch": 1.6213615685011473,
-      "grad_norm": 0.001677638734690845,
-      "learning_rate": 0.00019999870537126658,
-      "loss": 46.0,
-      "step": 10068
-    },
-    {
-      "epoch": 1.6215226055799348,
-      "grad_norm": 0.0007829164969734848,
-      "learning_rate": 0.00019999870511382172,
-      "loss": 46.0,
-      "step": 10069
-    },
-    {
-      "epoch": 1.6216836426587222,
-      "grad_norm": 0.002520563080906868,
-      "learning_rate": 0.00019999870485635124,
-      "loss": 46.0,
-      "step": 10070
-    },
-    {
-      "epoch": 1.6218446797375097,
-      "grad_norm": 0.004305894486606121,
-      "learning_rate": 0.00019999870459885518,
-      "loss": 46.0,
-      "step": 10071
-    },
-    {
-      "epoch": 1.622005716816297,
-      "grad_norm": 0.0006340468535199761,
-      "learning_rate": 0.00019999870434133353,
-      "loss": 46.0,
-      "step": 10072
-    },
-    {
-      "epoch": 1.6221667538950844,
-      "grad_norm": 0.0066359075717628,
-      "learning_rate": 0.00019999870408378626,
-      "loss": 46.0,
-      "step": 10073
-    },
-    {
-      "epoch": 1.6223277909738716,
-      "grad_norm": 0.0027352499309927225,
-      "learning_rate": 0.0001999987038262134,
-      "loss": 46.0,
-      "step": 10074
-    },
-    {
-      "epoch": 1.622488828052659,
-      "grad_norm": 0.0009624132653698325,
-      "learning_rate": 0.00019999870356861497,
-      "loss": 46.0,
-      "step": 10075
-    },
-    {
-      "epoch": 1.6226498651314465,
-      "grad_norm": 0.0006998457829467952,
-      "learning_rate": 0.0001999987033109909,
-      "loss": 46.0,
-      "step": 10076
-    },
-    {
-      "epoch": 1.622810902210234,
-      "grad_norm": 0.0012244939571246505,
-      "learning_rate": 0.00019999870305334127,
-      "loss": 46.0,
-      "step": 10077
-    },
-    {
-      "epoch": 1.6229719392890214,
-      "grad_norm": 0.005225862842053175,
-      "learning_rate": 0.00019999870279566607,
-      "loss": 46.0,
-      "step": 10078
-    },
-    {
-      "epoch": 1.6231329763678088,
-      "grad_norm": 0.0021019408013671637,
-      "learning_rate": 0.00019999870253796523,
-      "loss": 46.0,
-      "step": 10079
-    },
-    {
-      "epoch": 1.623294013446596,
-      "grad_norm": 0.0016421145992353559,
-      "learning_rate": 0.0001999987022802388,
-      "loss": 46.0,
-      "step": 10080
-    },
-    {
-      "epoch": 1.6234550505253833,
-      "grad_norm": 0.004970487207174301,
-      "learning_rate": 0.0001999987020224868,
-      "loss": 46.0,
-      "step": 10081
-    },
-    {
-      "epoch": 1.6236160876041708,
-      "grad_norm": 0.0025287049356848,
-      "learning_rate": 0.00019999870176470917,
-      "loss": 46.0,
-      "step": 10082
-    },
-    {
-      "epoch": 1.6237771246829582,
-      "grad_norm": 0.0009500497253611684,
-      "learning_rate": 0.00019999870150690598,
-      "loss": 46.0,
-      "step": 10083
-    },
-    {
-      "epoch": 1.6239381617617457,
-      "grad_norm": 0.004572180565446615,
-      "learning_rate": 0.0001999987012490772,
-      "loss": 46.0,
-      "step": 10084
-    },
-    {
-      "epoch": 1.624099198840533,
-      "grad_norm": 0.0036276737228035927,
-      "learning_rate": 0.0001999987009912228,
-      "loss": 46.0,
-      "step": 10085
-    },
-    {
-      "epoch": 1.6242602359193206,
-      "grad_norm": 0.003646471770480275,
-      "learning_rate": 0.0001999987007333428,
-      "loss": 46.0,
-      "step": 10086
-    },
-    {
-      "epoch": 1.6244212729981078,
-      "grad_norm": 0.003856019349768758,
-      "learning_rate": 0.00019999870047543724,
-      "loss": 46.0,
-      "step": 10087
-    },
-    {
-      "epoch": 1.6245823100768952,
-      "grad_norm": 0.001449517090804875,
-      "learning_rate": 0.00019999870021750605,
-      "loss": 46.0,
-      "step": 10088
-    },
-    {
-      "epoch": 1.6247433471556825,
-      "grad_norm": 0.002414804883301258,
-      "learning_rate": 0.00019999869995954928,
-      "loss": 46.0,
-      "step": 10089
-    },
-    {
-      "epoch": 1.62490438423447,
-      "grad_norm": 0.0013962793163955212,
-      "learning_rate": 0.00019999869970156693,
-      "loss": 46.0,
-      "step": 10090
-    },
-    {
-      "epoch": 1.6250654213132574,
-      "grad_norm": 0.001952785300090909,
-      "learning_rate": 0.00019999869944355898,
-      "loss": 46.0,
-      "step": 10091
-    },
-    {
-      "epoch": 1.6252264583920448,
-      "grad_norm": 0.0030772138852626085,
-      "learning_rate": 0.0001999986991855254,
-      "loss": 46.0,
-      "step": 10092
-    },
-    {
-      "epoch": 1.6253874954708323,
-      "grad_norm": 0.009740648791193962,
-      "learning_rate": 0.00019999869892746626,
-      "loss": 46.0,
-      "step": 10093
-    },
-    {
-      "epoch": 1.6255485325496195,
-      "grad_norm": 0.004067290108650923,
-      "learning_rate": 0.00019999869866938152,
-      "loss": 46.0,
-      "step": 10094
-    },
-    {
-      "epoch": 1.625709569628407,
-      "grad_norm": 0.0013128085993230343,
-      "learning_rate": 0.00019999869841127118,
-      "loss": 46.0,
-      "step": 10095
-    },
-    {
-      "epoch": 1.6258706067071942,
-      "grad_norm": 0.005055727902799845,
-      "learning_rate": 0.00019999869815313524,
-      "loss": 46.0,
-      "step": 10096
-    },
-    {
-      "epoch": 1.6260316437859816,
-      "grad_norm": 0.00602863309904933,
-      "learning_rate": 0.00019999869789497372,
-      "loss": 46.0,
-      "step": 10097
-    },
-    {
-      "epoch": 1.626192680864769,
-      "grad_norm": 0.004003756679594517,
-      "learning_rate": 0.00019999869763678662,
-      "loss": 46.0,
-      "step": 10098
-    },
-    {
-      "epoch": 1.6263537179435565,
-      "grad_norm": 0.001058023888617754,
-      "learning_rate": 0.0001999986973785739,
-      "loss": 46.0,
-      "step": 10099
-    },
-    {
-      "epoch": 1.626514755022344,
-      "grad_norm": 0.0008028603042475879,
-      "learning_rate": 0.00019999869712033558,
-      "loss": 46.0,
-      "step": 10100
-    },
-    {
-      "epoch": 1.6266757921011314,
-      "grad_norm": 0.00355736305937171,
-      "learning_rate": 0.0001999986968620717,
-      "loss": 46.0,
-      "step": 10101
-    },
-    {
-      "epoch": 1.6268368291799187,
-      "grad_norm": 0.0012655669124796987,
-      "learning_rate": 0.00019999869660378218,
-      "loss": 46.0,
-      "step": 10102
-    },
-    {
-      "epoch": 1.626997866258706,
-      "grad_norm": 0.0006586768431589007,
-      "learning_rate": 0.00019999869634546708,
-      "loss": 46.0,
-      "step": 10103
-    },
-    {
-      "epoch": 1.6271589033374934,
-      "grad_norm": 0.001410821219906211,
-      "learning_rate": 0.00019999869608712637,
-      "loss": 46.0,
-      "step": 10104
-    },
-    {
-      "epoch": 1.6273199404162808,
-      "grad_norm": 0.0027455328963696957,
-      "learning_rate": 0.00019999869582876012,
-      "loss": 46.0,
-      "step": 10105
-    },
-    {
-      "epoch": 1.6274809774950683,
-      "grad_norm": 0.004590548574924469,
-      "learning_rate": 0.00019999869557036824,
-      "loss": 46.0,
-      "step": 10106
-    },
-    {
-      "epoch": 1.6276420145738557,
-      "grad_norm": 0.0038277723360806704,
-      "learning_rate": 0.00019999869531195076,
-      "loss": 46.0,
-      "step": 10107
-    },
-    {
-      "epoch": 1.6278030516526432,
-      "grad_norm": 0.006106822285801172,
-      "learning_rate": 0.0001999986950535077,
-      "loss": 46.0,
-      "step": 10108
-    },
-    {
-      "epoch": 1.6279640887314304,
-      "grad_norm": 0.0021642930805683136,
-      "learning_rate": 0.00019999869479503903,
-      "loss": 46.0,
-      "step": 10109
-    },
-    {
-      "epoch": 1.6281251258102178,
-      "grad_norm": 0.004288241732865572,
-      "learning_rate": 0.0001999986945365448,
-      "loss": 46.0,
-      "step": 10110
-    },
-    {
-      "epoch": 1.628286162889005,
-      "grad_norm": 0.0019926796667277813,
-      "learning_rate": 0.00019999869427802494,
-      "loss": 46.0,
-      "step": 10111
-    },
-    {
-      "epoch": 1.6284471999677925,
-      "grad_norm": 0.001437365310266614,
-      "learning_rate": 0.00019999869401947948,
-      "loss": 46.0,
-      "step": 10112
-    },
-    {
-      "epoch": 1.62860823704658,
-      "grad_norm": 0.0014885510317981243,
-      "learning_rate": 0.00019999869376090845,
-      "loss": 46.0,
-      "step": 10113
-    },
-    {
-      "epoch": 1.6287692741253674,
-      "grad_norm": 0.0024821716360747814,
-      "learning_rate": 0.00019999869350231182,
-      "loss": 46.0,
-      "step": 10114
-    },
-    {
-      "epoch": 1.6289303112041549,
-      "grad_norm": 0.001589068560861051,
-      "learning_rate": 0.0001999986932436896,
-      "loss": 46.0,
-      "step": 10115
-    },
-    {
-      "epoch": 1.629091348282942,
-      "grad_norm": 0.0037374033126980066,
-      "learning_rate": 0.00019999869298504178,
-      "loss": 46.0,
-      "step": 10116
-    },
-    {
-      "epoch": 1.6292523853617296,
-      "grad_norm": 0.000879802624695003,
-      "learning_rate": 0.00019999869272636835,
-      "loss": 46.0,
-      "step": 10117
-    },
-    {
-      "epoch": 1.6294134224405168,
-      "grad_norm": 0.004092895891517401,
-      "learning_rate": 0.00019999869246766934,
-      "loss": 46.0,
-      "step": 10118
-    },
-    {
-      "epoch": 1.6295744595193042,
-      "grad_norm": 0.002839423716068268,
-      "learning_rate": 0.00019999869220894474,
-      "loss": 46.0,
-      "step": 10119
-    },
-    {
-      "epoch": 1.6297354965980917,
-      "grad_norm": 0.0038170686457306147,
-      "learning_rate": 0.00019999869195019452,
-      "loss": 46.0,
-      "step": 10120
-    },
-    {
-      "epoch": 1.6298965336768791,
-      "grad_norm": 0.0006704450352117419,
-      "learning_rate": 0.00019999869169141875,
-      "loss": 46.0,
-      "step": 10121
-    },
-    {
-      "epoch": 1.6300575707556666,
-      "grad_norm": 0.004688435699790716,
-      "learning_rate": 0.00019999869143261733,
-      "loss": 46.0,
-      "step": 10122
-    },
-    {
-      "epoch": 1.630218607834454,
-      "grad_norm": 0.0006784223951399326,
-      "learning_rate": 0.00019999869117379038,
-      "loss": 46.0,
-      "step": 10123
-    },
-    {
-      "epoch": 1.6303796449132413,
-      "grad_norm": 0.0009879088029265404,
-      "learning_rate": 0.0001999986909149378,
-      "loss": 46.0,
-      "step": 10124
-    },
-    {
-      "epoch": 1.6305406819920285,
-      "grad_norm": 0.0014195271069183946,
-      "learning_rate": 0.0001999986906560596,
-      "loss": 46.0,
-      "step": 10125
-    },
-    {
-      "epoch": 1.630701719070816,
-      "grad_norm": 0.002923595020547509,
-      "learning_rate": 0.00019999869039715585,
-      "loss": 46.0,
-      "step": 10126
-    },
-    {
-      "epoch": 1.6308627561496034,
-      "grad_norm": 0.0018372135236859322,
-      "learning_rate": 0.00019999869013822647,
-      "loss": 46.0,
-      "step": 10127
-    },
-    {
-      "epoch": 1.6310237932283909,
-      "grad_norm": 0.003369334153831005,
-      "learning_rate": 0.00019999868987927153,
-      "loss": 46.0,
-      "step": 10128
-    },
-    {
-      "epoch": 1.6311848303071783,
-      "grad_norm": 0.0015605166554450989,
-      "learning_rate": 0.00019999868962029094,
-      "loss": 46.0,
-      "step": 10129
-    },
-    {
-      "epoch": 1.6313458673859658,
-      "grad_norm": 0.011558981612324715,
-      "learning_rate": 0.00019999868936128483,
-      "loss": 46.0,
-      "step": 10130
-    },
-    {
-      "epoch": 1.631506904464753,
-      "grad_norm": 0.003711790544912219,
-      "learning_rate": 0.00019999868910225308,
-      "loss": 46.0,
-      "step": 10131
-    },
-    {
-      "epoch": 1.6316679415435404,
-      "grad_norm": 0.008824239484965801,
-      "learning_rate": 0.00019999868884319576,
-      "loss": 46.0,
-      "step": 10132
-    },
-    {
-      "epoch": 1.6318289786223277,
-      "grad_norm": 0.001055867993272841,
-      "learning_rate": 0.00019999868858411283,
-      "loss": 46.0,
-      "step": 10133
-    },
-    {
-      "epoch": 1.6319900157011151,
-      "grad_norm": 0.008462686091661453,
-      "learning_rate": 0.0001999986883250043,
-      "loss": 46.0,
-      "step": 10134
-    },
-    {
-      "epoch": 1.6321510527799026,
-      "grad_norm": 0.0019918810576200485,
-      "learning_rate": 0.00019999868806587018,
-      "loss": 46.0,
-      "step": 10135
-    },
-    {
-      "epoch": 1.63231208985869,
-      "grad_norm": 0.0006822144496254623,
-      "learning_rate": 0.00019999868780671046,
-      "loss": 46.0,
-      "step": 10136
-    },
-    {
-      "epoch": 1.6324731269374775,
-      "grad_norm": 0.0006239225040189922,
-      "learning_rate": 0.00019999868754752516,
-      "loss": 46.0,
-      "step": 10137
-    },
-    {
-      "epoch": 1.6326341640162647,
-      "grad_norm": 0.0004670976777561009,
-      "learning_rate": 0.00019999868728831426,
-      "loss": 46.0,
-      "step": 10138
-    },
-    {
-      "epoch": 1.6327952010950522,
-      "grad_norm": 0.002166206017136574,
-      "learning_rate": 0.00019999868702907776,
-      "loss": 46.0,
-      "step": 10139
-    },
-    {
-      "epoch": 1.6329562381738394,
-      "grad_norm": 0.0006342603010125458,
-      "learning_rate": 0.0001999986867698157,
-      "loss": 46.0,
-      "step": 10140
-    },
-    {
-      "epoch": 1.6331172752526268,
-      "grad_norm": 0.000780830392614007,
-      "learning_rate": 0.00019999868651052798,
-      "loss": 46.0,
-      "step": 10141
-    },
-    {
-      "epoch": 1.6332783123314143,
-      "grad_norm": 0.005827606189996004,
-      "learning_rate": 0.0001999986862512147,
-      "loss": 46.0,
-      "step": 10142
-    },
-    {
-      "epoch": 1.6334393494102017,
-      "grad_norm": 0.0009667090489529073,
-      "learning_rate": 0.00019999868599187583,
-      "loss": 46.0,
-      "step": 10143
-    },
-    {
-      "epoch": 1.6336003864889892,
-      "grad_norm": 0.0012722114333882928,
-      "learning_rate": 0.00019999868573251136,
-      "loss": 46.0,
-      "step": 10144
-    },
-    {
-      "epoch": 1.6337614235677764,
-      "grad_norm": 0.0021969599183648825,
-      "learning_rate": 0.00019999868547312128,
-      "loss": 46.0,
-      "step": 10145
-    },
-    {
-      "epoch": 1.6339224606465639,
-      "grad_norm": 0.00491515826433897,
-      "learning_rate": 0.00019999868521370563,
-      "loss": 46.0,
-      "step": 10146
-    },
-    {
-      "epoch": 1.634083497725351,
-      "grad_norm": 0.0008893043850548565,
-      "learning_rate": 0.00019999868495426437,
-      "loss": 46.0,
-      "step": 10147
-    },
-    {
-      "epoch": 1.6342445348041386,
-      "grad_norm": 0.0010812173131853342,
-      "learning_rate": 0.00019999868469479753,
-      "loss": 46.0,
-      "step": 10148
-    },
-    {
-      "epoch": 1.634405571882926,
-      "grad_norm": 0.008252987638115883,
-      "learning_rate": 0.0001999986844353051,
-      "loss": 46.0,
-      "step": 10149
-    },
-    {
-      "epoch": 1.6345666089617135,
-      "grad_norm": 0.0012162404600530863,
-      "learning_rate": 0.00019999868417578705,
-      "loss": 46.0,
-      "step": 10150
-    },
-    {
-      "epoch": 1.634727646040501,
-      "grad_norm": 0.0004576060746330768,
-      "learning_rate": 0.0001999986839162434,
-      "loss": 46.0,
-      "step": 10151
-    },
-    {
-      "epoch": 1.6348886831192884,
-      "grad_norm": 0.0017532332567498088,
-      "learning_rate": 0.0001999986836566742,
-      "loss": 46.0,
-      "step": 10152
-    },
-    {
-      "epoch": 1.6350497201980756,
-      "grad_norm": 0.0017380168428644538,
-      "learning_rate": 0.00019999868339707936,
-      "loss": 46.0,
-      "step": 10153
-    },
-    {
-      "epoch": 1.635210757276863,
-      "grad_norm": 0.0017242757603526115,
-      "learning_rate": 0.00019999868313745896,
-      "loss": 46.0,
-      "step": 10154
-    },
-    {
-      "epoch": 1.6353717943556503,
-      "grad_norm": 0.004668163601309061,
-      "learning_rate": 0.00019999868287781295,
-      "loss": 46.0,
-      "step": 10155
-    },
-    {
-      "epoch": 1.6355328314344377,
-      "grad_norm": 0.003891122993081808,
-      "learning_rate": 0.00019999868261814135,
-      "loss": 46.0,
-      "step": 10156
-    },
-    {
-      "epoch": 1.6356938685132252,
-      "grad_norm": 0.000941638310905546,
-      "learning_rate": 0.00019999868235844414,
-      "loss": 46.0,
-      "step": 10157
-    },
-    {
-      "epoch": 1.6358549055920126,
-      "grad_norm": 0.0007921447395347059,
-      "learning_rate": 0.00019999868209872137,
-      "loss": 46.0,
-      "step": 10158
-    },
-    {
-      "epoch": 1.6360159426708,
-      "grad_norm": 0.003627803875133395,
-      "learning_rate": 0.00019999868183897296,
-      "loss": 46.0,
-      "step": 10159
-    },
-    {
-      "epoch": 1.6361769797495873,
-      "grad_norm": 0.004973144736140966,
-      "learning_rate": 0.00019999868157919898,
-      "loss": 46.0,
-      "step": 10160
-    },
-    {
-      "epoch": 1.6363380168283748,
-      "grad_norm": 0.0053692711517214775,
-      "learning_rate": 0.00019999868131939942,
-      "loss": 46.0,
-      "step": 10161
-    },
-    {
-      "epoch": 1.636499053907162,
-      "grad_norm": 0.0034927669912576675,
-      "learning_rate": 0.00019999868105957425,
-      "loss": 46.0,
-      "step": 10162
-    },
-    {
-      "epoch": 1.6366600909859494,
-      "grad_norm": 0.001765651279129088,
-      "learning_rate": 0.00019999868079972349,
-      "loss": 46.0,
-      "step": 10163
-    },
-    {
-      "epoch": 1.636821128064737,
-      "grad_norm": 0.001500483718700707,
-      "learning_rate": 0.0001999986805398471,
-      "loss": 46.0,
-      "step": 10164
-    },
-    {
-      "epoch": 1.6369821651435243,
-      "grad_norm": 0.0013322671875357628,
-      "learning_rate": 0.00019999868027994517,
-      "loss": 46.0,
-      "step": 10165
-    },
-    {
-      "epoch": 1.6371432022223118,
-      "grad_norm": 0.0007708729826845229,
-      "learning_rate": 0.00019999868002001762,
-      "loss": 46.0,
-      "step": 10166
-    },
-    {
-      "epoch": 1.637304239301099,
-      "grad_norm": 0.005682768300175667,
-      "learning_rate": 0.00019999867976006449,
-      "loss": 46.0,
-      "step": 10167
-    },
-    {
-      "epoch": 1.6374652763798865,
-      "grad_norm": 0.0011884463019669056,
-      "learning_rate": 0.00019999867950008576,
-      "loss": 46.0,
-      "step": 10168
-    },
-    {
-      "epoch": 1.6376263134586737,
-      "grad_norm": 0.002665251726284623,
-      "learning_rate": 0.0001999986792400814,
-      "loss": 46.0,
-      "step": 10169
-    },
-    {
-      "epoch": 1.6377873505374612,
-      "grad_norm": 0.0014309600228443742,
-      "learning_rate": 0.00019999867898005147,
-      "loss": 46.0,
-      "step": 10170
-    },
-    {
-      "epoch": 1.6379483876162486,
-      "grad_norm": 0.0068282983265817165,
-      "learning_rate": 0.00019999867871999596,
-      "loss": 46.0,
-      "step": 10171
-    },
-    {
-      "epoch": 1.638109424695036,
-      "grad_norm": 0.0024610208347439766,
-      "learning_rate": 0.00019999867845991483,
-      "loss": 46.0,
-      "step": 10172
-    },
-    {
-      "epoch": 1.6382704617738235,
-      "grad_norm": 0.0019890815019607544,
-      "learning_rate": 0.00019999867819980814,
-      "loss": 46.0,
-      "step": 10173
-    },
-    {
-      "epoch": 1.638431498852611,
-      "grad_norm": 0.005978357046842575,
-      "learning_rate": 0.0001999986779396758,
-      "loss": 46.0,
-      "step": 10174
-    },
-    {
-      "epoch": 1.6385925359313982,
-      "grad_norm": 0.000870869611389935,
-      "learning_rate": 0.00019999867767951792,
-      "loss": 46.0,
-      "step": 10175
-    },
-    {
-      "epoch": 1.6387535730101856,
-      "grad_norm": 0.0015010355273261666,
-      "learning_rate": 0.00019999867741933442,
-      "loss": 46.0,
-      "step": 10176
-    },
-    {
-      "epoch": 1.6389146100889729,
-      "grad_norm": 0.0017040945822373033,
-      "learning_rate": 0.00019999867715912535,
-      "loss": 46.0,
-      "step": 10177
-    },
-    {
-      "epoch": 1.6390756471677603,
-      "grad_norm": 0.0031766793690621853,
-      "learning_rate": 0.00019999867689889068,
-      "loss": 46.0,
-      "step": 10178
-    },
-    {
-      "epoch": 1.6392366842465478,
-      "grad_norm": 0.006567544303834438,
-      "learning_rate": 0.00019999867663863036,
-      "loss": 46.0,
-      "step": 10179
-    },
-    {
-      "epoch": 1.6393977213253352,
-      "grad_norm": 0.0015725432895123959,
-      "learning_rate": 0.0001999986763783445,
-      "loss": 46.0,
-      "step": 10180
-    },
-    {
-      "epoch": 1.6395587584041227,
-      "grad_norm": 0.005981299560517073,
-      "learning_rate": 0.000199998676118033,
-      "loss": 46.0,
-      "step": 10181
-    },
-    {
-      "epoch": 1.63971979548291,
-      "grad_norm": 0.0015052786329761147,
-      "learning_rate": 0.00019999867585769599,
-      "loss": 46.0,
-      "step": 10182
-    },
-    {
-      "epoch": 1.6398808325616974,
-      "grad_norm": 0.004097385797649622,
-      "learning_rate": 0.00019999867559733332,
-      "loss": 46.0,
-      "step": 10183
-    },
-    {
-      "epoch": 1.6400418696404846,
-      "grad_norm": 0.006585290189832449,
-      "learning_rate": 0.00019999867533694506,
-      "loss": 46.0,
-      "step": 10184
-    },
-    {
-      "epoch": 1.640202906719272,
-      "grad_norm": 0.0008066860609687865,
-      "learning_rate": 0.00019999867507653122,
-      "loss": 46.0,
-      "step": 10185
-    },
-    {
-      "epoch": 1.6403639437980595,
-      "grad_norm": 0.0019000923493877053,
-      "learning_rate": 0.0001999986748160918,
-      "loss": 46.0,
-      "step": 10186
-    },
-    {
-      "epoch": 1.640524980876847,
-      "grad_norm": 0.0015420460840687156,
-      "learning_rate": 0.00019999867455562675,
-      "loss": 46.0,
-      "step": 10187
-    },
-    {
-      "epoch": 1.6406860179556344,
-      "grad_norm": 0.006298865657299757,
-      "learning_rate": 0.00019999867429513612,
-      "loss": 46.0,
-      "step": 10188
-    },
-    {
-      "epoch": 1.6408470550344216,
-      "grad_norm": 0.0009785126894712448,
-      "learning_rate": 0.0001999986740346199,
-      "loss": 46.0,
-      "step": 10189
-    },
-    {
-      "epoch": 1.641008092113209,
-      "grad_norm": 0.007423033472150564,
-      "learning_rate": 0.0001999986737740781,
-      "loss": 46.0,
-      "step": 10190
-    },
-    {
-      "epoch": 1.6411691291919963,
-      "grad_norm": 0.0024120002053678036,
-      "learning_rate": 0.00019999867351351067,
-      "loss": 46.0,
-      "step": 10191
-    },
-    {
-      "epoch": 1.6413301662707838,
-      "grad_norm": 0.005326361861079931,
-      "learning_rate": 0.00019999867325291767,
-      "loss": 46.0,
-      "step": 10192
-    },
-    {
-      "epoch": 1.6414912033495712,
-      "grad_norm": 0.0005744045483879745,
-      "learning_rate": 0.00019999867299229907,
-      "loss": 46.0,
-      "step": 10193
-    },
-    {
-      "epoch": 1.6416522404283587,
-      "grad_norm": 0.0020154910162091255,
-      "learning_rate": 0.00019999867273165487,
-      "loss": 46.0,
-      "step": 10194
-    },
-    {
-      "epoch": 1.6418132775071461,
-      "grad_norm": 0.0012288293801248074,
-      "learning_rate": 0.00019999867247098507,
-      "loss": 46.0,
-      "step": 10195
-    },
-    {
-      "epoch": 1.6419743145859336,
-      "grad_norm": 0.0016620673704892397,
-      "learning_rate": 0.0001999986722102897,
-      "loss": 46.0,
-      "step": 10196
-    },
-    {
-      "epoch": 1.6421353516647208,
-      "grad_norm": 0.00709503935649991,
-      "learning_rate": 0.0001999986719495687,
-      "loss": 46.0,
-      "step": 10197
-    },
-    {
-      "epoch": 1.642296388743508,
-      "grad_norm": 0.010401827283203602,
-      "learning_rate": 0.00019999867168882214,
-      "loss": 46.0,
-      "step": 10198
-    },
-    {
-      "epoch": 1.6424574258222955,
-      "grad_norm": 0.0028349391650408506,
-      "learning_rate": 0.00019999867142804996,
-      "loss": 46.0,
-      "step": 10199
-    },
-    {
-      "epoch": 1.642618462901083,
-      "grad_norm": 0.00155677052680403,
-      "learning_rate": 0.0001999986711672522,
-      "loss": 46.0,
-      "step": 10200
-    },
-    {
-      "epoch": 1.6427794999798704,
-      "grad_norm": 0.001592310145497322,
-      "learning_rate": 0.00019999867090642886,
-      "loss": 46.0,
-      "step": 10201
-    },
-    {
-      "epoch": 1.6429405370586578,
-      "grad_norm": 0.0013410663232207298,
-      "learning_rate": 0.0001999986706455799,
-      "loss": 46.0,
-      "step": 10202
-    },
-    {
-      "epoch": 1.6431015741374453,
-      "grad_norm": 0.004570495802909136,
-      "learning_rate": 0.00019999867038470535,
-      "loss": 46.0,
-      "step": 10203
-    },
-    {
-      "epoch": 1.6432626112162325,
-      "grad_norm": 0.0069716391153633595,
-      "learning_rate": 0.0001999986701238052,
-      "loss": 46.0,
-      "step": 10204
-    },
-    {
-      "epoch": 1.64342364829502,
-      "grad_norm": 0.0009891677182167768,
-      "learning_rate": 0.00019999866986287947,
-      "loss": 46.0,
-      "step": 10205
-    },
-    {
-      "epoch": 1.6435846853738072,
-      "grad_norm": 0.001915937289595604,
-      "learning_rate": 0.00019999866960192814,
-      "loss": 46.0,
-      "step": 10206
-    },
-    {
-      "epoch": 1.6437457224525946,
-      "grad_norm": 0.004006378818303347,
-      "learning_rate": 0.00019999866934095121,
-      "loss": 46.0,
-      "step": 10207
-    },
-    {
-      "epoch": 1.643906759531382,
-      "grad_norm": 0.001221159240230918,
-      "learning_rate": 0.00019999866907994868,
-      "loss": 46.0,
-      "step": 10208
-    },
-    {
-      "epoch": 1.6440677966101696,
-      "grad_norm": 0.0025894588325172663,
-      "learning_rate": 0.0001999986688189206,
-      "loss": 46.0,
-      "step": 10209
-    },
-    {
-      "epoch": 1.644228833688957,
-      "grad_norm": 0.0018897461704909801,
-      "learning_rate": 0.0001999986685578669,
-      "loss": 46.0,
-      "step": 10210
-    },
-    {
-      "epoch": 1.6443898707677442,
-      "grad_norm": 0.0026194225065410137,
-      "learning_rate": 0.00019999866829678757,
-      "loss": 46.0,
-      "step": 10211
-    },
-    {
-      "epoch": 1.6445509078465317,
-      "grad_norm": 0.0016701591666787863,
-      "learning_rate": 0.00019999866803568268,
-      "loss": 46.0,
-      "step": 10212
-    },
-    {
-      "epoch": 1.644711944925319,
-      "grad_norm": 0.007778188679367304,
-      "learning_rate": 0.00019999866777455219,
-      "loss": 46.0,
-      "step": 10213
-    },
-    {
-      "epoch": 1.6448729820041064,
-      "grad_norm": 0.009426028467714787,
-      "learning_rate": 0.0001999986675133961,
-      "loss": 46.0,
-      "step": 10214
-    },
-    {
-      "epoch": 1.6450340190828938,
-      "grad_norm": 0.002500662812963128,
-      "learning_rate": 0.00019999866725221442,
-      "loss": 46.0,
-      "step": 10215
-    },
-    {
-      "epoch": 1.6451950561616813,
-      "grad_norm": 0.0014767794637009501,
-      "learning_rate": 0.00019999866699100714,
-      "loss": 46.0,
-      "step": 10216
-    },
-    {
-      "epoch": 1.6453560932404687,
-      "grad_norm": 0.0012752891052514315,
-      "learning_rate": 0.0001999986667297743,
-      "loss": 46.0,
-      "step": 10217
-    },
-    {
-      "epoch": 1.6455171303192562,
-      "grad_norm": 0.005201061721891165,
-      "learning_rate": 0.0001999986664685158,
-      "loss": 46.0,
-      "step": 10218
-    },
-    {
-      "epoch": 1.6456781673980434,
-      "grad_norm": 0.0018691556761041284,
-      "learning_rate": 0.00019999866620723172,
-      "loss": 46.0,
-      "step": 10219
-    },
-    {
-      "epoch": 1.6458392044768306,
-      "grad_norm": 0.0005572540685534477,
-      "learning_rate": 0.00019999866594592208,
-      "loss": 46.0,
-      "step": 10220
-    },
-    {
-      "epoch": 1.646000241555618,
-      "grad_norm": 0.0027833315543830395,
-      "learning_rate": 0.00019999866568458683,
-      "loss": 46.0,
-      "step": 10221
-    },
-    {
-      "epoch": 1.6461612786344055,
-      "grad_norm": 0.00598552543669939,
-      "learning_rate": 0.000199998665423226,
-      "loss": 46.0,
-      "step": 10222
-    },
-    {
-      "epoch": 1.646322315713193,
-      "grad_norm": 0.00397255877032876,
-      "learning_rate": 0.00019999866516183957,
-      "loss": 46.0,
-      "step": 10223
-    },
-    {
-      "epoch": 1.6464833527919804,
-      "grad_norm": 0.0010870033875107765,
-      "learning_rate": 0.00019999866490042753,
-      "loss": 46.0,
-      "step": 10224
-    },
-    {
-      "epoch": 1.6466443898707679,
-      "grad_norm": 0.0025970637798309326,
-      "learning_rate": 0.0001999986646389899,
-      "loss": 46.0,
-      "step": 10225
-    },
-    {
-      "epoch": 1.6468054269495551,
-      "grad_norm": 0.0015767483273521066,
-      "learning_rate": 0.00019999866437752669,
-      "loss": 46.0,
-      "step": 10226
-    },
-    {
-      "epoch": 1.6469664640283426,
-      "grad_norm": 0.001275165006518364,
-      "learning_rate": 0.00019999866411603786,
-      "loss": 46.0,
-      "step": 10227
-    },
-    {
-      "epoch": 1.6471275011071298,
-      "grad_norm": 0.0018989950185641646,
-      "learning_rate": 0.00019999866385452347,
-      "loss": 46.0,
-      "step": 10228
-    },
-    {
-      "epoch": 1.6472885381859173,
-      "grad_norm": 0.003920397255569696,
-      "learning_rate": 0.00019999866359298344,
-      "loss": 46.0,
-      "step": 10229
-    },
-    {
-      "epoch": 1.6474495752647047,
-      "grad_norm": 0.01568520814180374,
-      "learning_rate": 0.00019999866333141785,
-      "loss": 46.0,
-      "step": 10230
-    },
-    {
-      "epoch": 1.6476106123434922,
-      "grad_norm": 0.0010440874611958861,
-      "learning_rate": 0.00019999866306982667,
-      "loss": 46.0,
-      "step": 10231
-    },
-    {
-      "epoch": 1.6477716494222796,
-      "grad_norm": 0.0030033974908292294,
-      "learning_rate": 0.00019999866280820988,
-      "loss": 46.0,
-      "step": 10232
-    },
-    {
-      "epoch": 1.6479326865010668,
-      "grad_norm": 0.0017947761807590723,
-      "learning_rate": 0.0001999986625465675,
-      "loss": 46.0,
-      "step": 10233
-    },
-    {
-      "epoch": 1.6480937235798543,
-      "grad_norm": 0.0037991649005562067,
-      "learning_rate": 0.0001999986622848995,
-      "loss": 46.0,
-      "step": 10234
-    },
-    {
-      "epoch": 1.6482547606586415,
-      "grad_norm": 0.0007412537233904004,
-      "learning_rate": 0.00019999866202320593,
-      "loss": 46.0,
-      "step": 10235
-    },
-    {
-      "epoch": 1.648415797737429,
-      "grad_norm": 0.0010383485350757837,
-      "learning_rate": 0.0001999986617614868,
-      "loss": 46.0,
-      "step": 10236
-    },
-    {
-      "epoch": 1.6485768348162164,
-      "grad_norm": 0.003825170686468482,
-      "learning_rate": 0.000199998661499742,
-      "loss": 46.0,
-      "step": 10237
-    },
-    {
-      "epoch": 1.6487378718950039,
-      "grad_norm": 0.0006131341215223074,
-      "learning_rate": 0.00019999866123797166,
-      "loss": 46.0,
-      "step": 10238
-    },
-    {
-      "epoch": 1.6488989089737913,
-      "grad_norm": 0.0015460059512406588,
-      "learning_rate": 0.0001999986609761757,
-      "loss": 46.0,
-      "step": 10239
-    },
-    {
-      "epoch": 1.6490599460525786,
-      "grad_norm": 0.0004370794340502471,
-      "learning_rate": 0.00019999866071435416,
-      "loss": 46.0,
-      "step": 10240
-    },
-    {
-      "epoch": 1.649220983131366,
-      "grad_norm": 0.0009183295769616961,
-      "learning_rate": 0.00019999866045250703,
-      "loss": 46.0,
-      "step": 10241
-    },
-    {
-      "epoch": 1.6493820202101532,
-      "grad_norm": 0.0026633599773049355,
-      "learning_rate": 0.0001999986601906343,
-      "loss": 46.0,
-      "step": 10242
-    },
-    {
-      "epoch": 1.6495430572889407,
-      "grad_norm": 0.0006786429439671338,
-      "learning_rate": 0.00019999865992873596,
-      "loss": 46.0,
-      "step": 10243
-    },
-    {
-      "epoch": 1.6497040943677281,
-      "grad_norm": 0.0035209462512284517,
-      "learning_rate": 0.00019999865966681207,
-      "loss": 46.0,
-      "step": 10244
-    },
-    {
-      "epoch": 1.6498651314465156,
-      "grad_norm": 0.0015742137329652905,
-      "learning_rate": 0.00019999865940486253,
-      "loss": 46.0,
-      "step": 10245
-    },
-    {
-      "epoch": 1.650026168525303,
-      "grad_norm": 0.0005557678523473442,
-      "learning_rate": 0.0001999986591428874,
-      "loss": 46.0,
-      "step": 10246
-    },
-    {
-      "epoch": 1.6501872056040905,
-      "grad_norm": 0.0043942974880337715,
-      "learning_rate": 0.0001999986588808867,
-      "loss": 46.0,
-      "step": 10247
-    },
-    {
-      "epoch": 1.6503482426828777,
-      "grad_norm": 0.001118987798690796,
-      "learning_rate": 0.00019999865861886044,
-      "loss": 46.0,
-      "step": 10248
-    },
-    {
-      "epoch": 1.6505092797616652,
-      "grad_norm": 0.002809450961649418,
-      "learning_rate": 0.00019999865835680853,
-      "loss": 46.0,
-      "step": 10249
-    },
-    {
-      "epoch": 1.6506703168404524,
-      "grad_norm": 0.002697139512747526,
-      "learning_rate": 0.000199998658094731,
-      "loss": 46.0,
-      "step": 10250
-    },
-    {
-      "epoch": 1.6508313539192399,
-      "grad_norm": 0.007467437069863081,
-      "learning_rate": 0.00019999865783262795,
-      "loss": 46.0,
-      "step": 10251
-    },
-    {
-      "epoch": 1.6509923909980273,
-      "grad_norm": 0.002739156596362591,
-      "learning_rate": 0.00019999865757049928,
-      "loss": 46.0,
-      "step": 10252
-    },
-    {
-      "epoch": 1.6511534280768148,
-      "grad_norm": 0.005171090364456177,
-      "learning_rate": 0.000199998657308345,
-      "loss": 46.0,
-      "step": 10253
-    },
-    {
-      "epoch": 1.6513144651556022,
-      "grad_norm": 0.0052896603010594845,
-      "learning_rate": 0.00019999865704616512,
-      "loss": 46.0,
-      "step": 10254
-    },
-    {
-      "epoch": 1.6514755022343894,
-      "grad_norm": 0.0020279455929994583,
-      "learning_rate": 0.00019999865678395966,
-      "loss": 46.0,
-      "step": 10255
-    },
-    {
-      "epoch": 1.6516365393131769,
-      "grad_norm": 0.0008441900718025863,
-      "learning_rate": 0.00019999865652172862,
-      "loss": 46.0,
-      "step": 10256
-    },
-    {
-      "epoch": 1.6517975763919641,
-      "grad_norm": 0.0030460739508271217,
-      "learning_rate": 0.00019999865625947195,
-      "loss": 46.0,
-      "step": 10257
-    },
-    {
-      "epoch": 1.6519586134707516,
-      "grad_norm": 0.0014858118956908584,
-      "learning_rate": 0.00019999865599718973,
-      "loss": 46.0,
-      "step": 10258
-    },
-    {
-      "epoch": 1.652119650549539,
-      "grad_norm": 0.0009504615445621312,
-      "learning_rate": 0.00019999865573488184,
-      "loss": 46.0,
-      "step": 10259
-    },
-    {
-      "epoch": 1.6522806876283265,
-      "grad_norm": 0.0033596823923289776,
-      "learning_rate": 0.00019999865547254842,
-      "loss": 46.0,
-      "step": 10260
-    },
-    {
-      "epoch": 1.652441724707114,
-      "grad_norm": 0.0036205481737852097,
-      "learning_rate": 0.00019999865521018939,
-      "loss": 46.0,
-      "step": 10261
-    },
-    {
-      "epoch": 1.6526027617859012,
-      "grad_norm": 0.00516926683485508,
-      "learning_rate": 0.0001999986549478048,
-      "loss": 46.0,
-      "step": 10262
-    },
-    {
-      "epoch": 1.6527637988646886,
-      "grad_norm": 0.0014452082104980946,
-      "learning_rate": 0.00019999865468539455,
-      "loss": 46.0,
-      "step": 10263
-    },
-    {
-      "epoch": 1.6529248359434758,
-      "grad_norm": 0.0010142697719857097,
-      "learning_rate": 0.00019999865442295875,
-      "loss": 46.0,
-      "step": 10264
-    },
-    {
-      "epoch": 1.6530858730222633,
-      "grad_norm": 0.0022039192263036966,
-      "learning_rate": 0.00019999865416049734,
-      "loss": 46.0,
-      "step": 10265
-    },
-    {
-      "epoch": 1.6532469101010507,
-      "grad_norm": 0.00158783420920372,
-      "learning_rate": 0.00019999865389801037,
-      "loss": 46.0,
-      "step": 10266
-    },
-    {
-      "epoch": 1.6534079471798382,
-      "grad_norm": 0.006714479066431522,
-      "learning_rate": 0.00019999865363549776,
-      "loss": 46.0,
-      "step": 10267
-    },
-    {
-      "epoch": 1.6535689842586256,
-      "grad_norm": 0.001299915136769414,
-      "learning_rate": 0.00019999865337295956,
-      "loss": 46.0,
-      "step": 10268
-    },
-    {
-      "epoch": 1.653730021337413,
-      "grad_norm": 0.0018057352863252163,
-      "learning_rate": 0.00019999865311039577,
-      "loss": 46.0,
-      "step": 10269
-    },
-    {
-      "epoch": 1.6538910584162003,
-      "grad_norm": 0.002297317376360297,
-      "learning_rate": 0.0001999986528478064,
-      "loss": 46.0,
-      "step": 10270
-    },
-    {
-      "epoch": 1.6540520954949878,
-      "grad_norm": 0.004488641861826181,
-      "learning_rate": 0.0001999986525851914,
-      "loss": 46.0,
-      "step": 10271
-    },
-    {
-      "epoch": 1.654213132573775,
-      "grad_norm": 0.00581676559522748,
-      "learning_rate": 0.00019999865232255085,
-      "loss": 46.0,
-      "step": 10272
-    },
-    {
-      "epoch": 1.6543741696525625,
-      "grad_norm": 0.0016572366002947092,
-      "learning_rate": 0.0001999986520598847,
-      "loss": 46.0,
-      "step": 10273
-    },
-    {
-      "epoch": 1.65453520673135,
-      "grad_norm": 0.002199936658143997,
-      "learning_rate": 0.00019999865179719294,
-      "loss": 46.0,
-      "step": 10274
-    },
-    {
-      "epoch": 1.6546962438101374,
-      "grad_norm": 0.0015104350168257952,
-      "learning_rate": 0.00019999865153447557,
-      "loss": 46.0,
-      "step": 10275
-    },
-    {
-      "epoch": 1.6548572808889248,
-      "grad_norm": 0.0017150352941825986,
-      "learning_rate": 0.00019999865127173262,
-      "loss": 46.0,
-      "step": 10276
-    },
-    {
-      "epoch": 1.655018317967712,
-      "grad_norm": 0.001269400236196816,
-      "learning_rate": 0.00019999865100896408,
-      "loss": 46.0,
-      "step": 10277
-    },
-    {
-      "epoch": 1.6551793550464995,
-      "grad_norm": 0.0008260617032647133,
-      "learning_rate": 0.00019999865074616995,
-      "loss": 46.0,
-      "step": 10278
-    },
-    {
-      "epoch": 1.6553403921252867,
-      "grad_norm": 0.0034680834505707026,
-      "learning_rate": 0.0001999986504833502,
-      "loss": 46.0,
-      "step": 10279
-    },
-    {
-      "epoch": 1.6555014292040742,
-      "grad_norm": 0.002518394961953163,
-      "learning_rate": 0.0001999986502205049,
-      "loss": 46.0,
-      "step": 10280
-    },
-    {
-      "epoch": 1.6556624662828616,
-      "grad_norm": 0.0015852086944505572,
-      "learning_rate": 0.00019999864995763397,
-      "loss": 46.0,
-      "step": 10281
-    },
-    {
-      "epoch": 1.655823503361649,
-      "grad_norm": 0.0008194456459023058,
-      "learning_rate": 0.00019999864969473747,
-      "loss": 46.0,
-      "step": 10282
-    },
-    {
-      "epoch": 1.6559845404404365,
-      "grad_norm": 0.0013457053573802114,
-      "learning_rate": 0.00019999864943181535,
-      "loss": 46.0,
-      "step": 10283
-    },
-    {
-      "epoch": 1.6561455775192238,
-      "grad_norm": 0.0010190659668296576,
-      "learning_rate": 0.00019999864916886765,
-      "loss": 46.0,
-      "step": 10284
-    },
-    {
-      "epoch": 1.6563066145980112,
-      "grad_norm": 0.00438708858564496,
-      "learning_rate": 0.00019999864890589436,
-      "loss": 46.0,
-      "step": 10285
-    },
-    {
-      "epoch": 1.6564676516767984,
-      "grad_norm": 0.0016959699569270015,
-      "learning_rate": 0.00019999864864289545,
-      "loss": 46.0,
-      "step": 10286
-    },
-    {
-      "epoch": 1.6566286887555859,
-      "grad_norm": 0.0015905431937426329,
-      "learning_rate": 0.00019999864837987096,
-      "loss": 46.0,
-      "step": 10287
-    },
-    {
-      "epoch": 1.6567897258343733,
-      "grad_norm": 0.00187137839384377,
-      "learning_rate": 0.00019999864811682088,
-      "loss": 46.0,
-      "step": 10288
-    },
-    {
-      "epoch": 1.6569507629131608,
-      "grad_norm": 0.0037138783372938633,
-      "learning_rate": 0.00019999864785374524,
-      "loss": 46.0,
-      "step": 10289
-    },
-    {
-      "epoch": 1.6571117999919482,
-      "grad_norm": 0.0038526931311935186,
-      "learning_rate": 0.00019999864759064396,
-      "loss": 46.0,
-      "step": 10290
-    },
-    {
-      "epoch": 1.6572728370707357,
-      "grad_norm": 0.004028433468192816,
-      "learning_rate": 0.0001999986473275171,
-      "loss": 46.0,
-      "step": 10291
-    },
-    {
-      "epoch": 1.657433874149523,
-      "grad_norm": 0.0028323526494205,
-      "learning_rate": 0.00019999864706436463,
-      "loss": 46.0,
-      "step": 10292
-    },
-    {
-      "epoch": 1.6575949112283102,
-      "grad_norm": 0.002596818143501878,
-      "learning_rate": 0.00019999864680118656,
-      "loss": 46.0,
-      "step": 10293
-    },
-    {
-      "epoch": 1.6577559483070976,
-      "grad_norm": 0.0006851371726952493,
-      "learning_rate": 0.00019999864653798293,
-      "loss": 46.0,
-      "step": 10294
-    },
-    {
-      "epoch": 1.657916985385885,
-      "grad_norm": 0.0004906481481157243,
-      "learning_rate": 0.00019999864627475366,
-      "loss": 46.0,
-      "step": 10295
-    },
-    {
-      "epoch": 1.6580780224646725,
-      "grad_norm": 0.001583976554684341,
-      "learning_rate": 0.00019999864601149883,
-      "loss": 46.0,
-      "step": 10296
-    },
-    {
-      "epoch": 1.65823905954346,
-      "grad_norm": 0.0004942187806591392,
-      "learning_rate": 0.0001999986457482184,
-      "loss": 46.0,
-      "step": 10297
-    },
-    {
-      "epoch": 1.6584000966222474,
-      "grad_norm": 0.001061291666701436,
-      "learning_rate": 0.00019999864548491235,
-      "loss": 46.0,
-      "step": 10298
-    },
-    {
-      "epoch": 1.6585611337010346,
-      "grad_norm": 0.006979250814765692,
-      "learning_rate": 0.00019999864522158073,
-      "loss": 46.0,
-      "step": 10299
-    },
-    {
-      "epoch": 1.658722170779822,
-      "grad_norm": 0.002171772997826338,
-      "learning_rate": 0.00019999864495822352,
-      "loss": 46.0,
-      "step": 10300
-    },
-    {
-      "epoch": 1.6588832078586093,
-      "grad_norm": 0.0016173458425328135,
-      "learning_rate": 0.00019999864469484073,
-      "loss": 46.0,
-      "step": 10301
-    },
-    {
-      "epoch": 1.6590442449373968,
-      "grad_norm": 0.0008664227789267898,
-      "learning_rate": 0.00019999864443143232,
-      "loss": 46.0,
-      "step": 10302
-    },
-    {
-      "epoch": 1.6592052820161842,
-      "grad_norm": 0.008758303709328175,
-      "learning_rate": 0.00019999864416799833,
-      "loss": 46.0,
-      "step": 10303
-    },
-    {
-      "epoch": 1.6593663190949717,
-      "grad_norm": 0.002159829018637538,
-      "learning_rate": 0.00019999864390453872,
-      "loss": 46.0,
-      "step": 10304
-    },
-    {
-      "epoch": 1.6595273561737591,
-      "grad_norm": 0.0011274589924141765,
-      "learning_rate": 0.00019999864364105355,
-      "loss": 46.0,
-      "step": 10305
-    },
-    {
-      "epoch": 1.6596883932525464,
-      "grad_norm": 0.002077885903418064,
-      "learning_rate": 0.00019999864337754276,
-      "loss": 46.0,
-      "step": 10306
-    },
-    {
-      "epoch": 1.6598494303313338,
-      "grad_norm": 0.0023729309905320406,
-      "learning_rate": 0.0001999986431140064,
-      "loss": 46.0,
-      "step": 10307
-    },
-    {
-      "epoch": 1.660010467410121,
-      "grad_norm": 0.0013342833844944835,
-      "learning_rate": 0.0001999986428504444,
-      "loss": 46.0,
-      "step": 10308
-    },
-    {
-      "epoch": 1.6601715044889085,
-      "grad_norm": 0.0010401697363704443,
-      "learning_rate": 0.00019999864258685683,
-      "loss": 46.0,
-      "step": 10309
-    },
-    {
-      "epoch": 1.660332541567696,
-      "grad_norm": 0.0008103396976366639,
-      "learning_rate": 0.00019999864232324367,
-      "loss": 46.0,
-      "step": 10310
-    },
-    {
-      "epoch": 1.6604935786464834,
-      "grad_norm": 0.00264181406237185,
-      "learning_rate": 0.00019999864205960492,
-      "loss": 46.0,
-      "step": 10311
-    },
-    {
-      "epoch": 1.6606546157252708,
-      "grad_norm": 0.0016249459004029632,
-      "learning_rate": 0.00019999864179594056,
-      "loss": 46.0,
-      "step": 10312
-    },
-    {
-      "epoch": 1.6608156528040583,
-      "grad_norm": 0.0004028141556773335,
-      "learning_rate": 0.00019999864153225064,
-      "loss": 46.0,
-      "step": 10313
-    },
-    {
-      "epoch": 1.6609766898828455,
-      "grad_norm": 0.002409035572782159,
-      "learning_rate": 0.0001999986412685351,
-      "loss": 46.0,
-      "step": 10314
-    },
-    {
-      "epoch": 1.6611377269616328,
-      "grad_norm": 0.002102794824168086,
-      "learning_rate": 0.00019999864100479395,
-      "loss": 46.0,
-      "step": 10315
-    },
-    {
-      "epoch": 1.6612987640404202,
-      "grad_norm": 0.002698926953598857,
-      "learning_rate": 0.00019999864074102724,
-      "loss": 46.0,
-      "step": 10316
-    },
-    {
-      "epoch": 1.6614598011192077,
-      "grad_norm": 0.006986147724092007,
-      "learning_rate": 0.00019999864047723492,
-      "loss": 46.0,
-      "step": 10317
-    },
-    {
-      "epoch": 1.661620838197995,
-      "grad_norm": 0.0012697026832029223,
-      "learning_rate": 0.00019999864021341698,
-      "loss": 46.0,
-      "step": 10318
-    },
-    {
-      "epoch": 1.6617818752767826,
-      "grad_norm": 0.0062049999833106995,
-      "learning_rate": 0.00019999863994957348,
-      "loss": 46.0,
-      "step": 10319
-    },
-    {
-      "epoch": 1.66194291235557,
-      "grad_norm": 0.008781444281339645,
-      "learning_rate": 0.00019999863968570437,
-      "loss": 46.0,
-      "step": 10320
-    },
-    {
-      "epoch": 1.6621039494343572,
-      "grad_norm": 0.0037513836286962032,
-      "learning_rate": 0.00019999863942180967,
-      "loss": 46.0,
-      "step": 10321
-    },
-    {
-      "epoch": 1.6622649865131447,
-      "grad_norm": 0.007861552760004997,
-      "learning_rate": 0.00019999863915788935,
-      "loss": 46.0,
-      "step": 10322
-    },
-    {
-      "epoch": 1.662426023591932,
-      "grad_norm": 0.004364061634987593,
-      "learning_rate": 0.00019999863889394348,
-      "loss": 46.0,
-      "step": 10323
-    },
-    {
-      "epoch": 1.6625870606707194,
-      "grad_norm": 0.0011777520412579179,
-      "learning_rate": 0.00019999863862997202,
-      "loss": 46.0,
-      "step": 10324
-    },
-    {
-      "epoch": 1.6627480977495068,
-      "grad_norm": 0.0009405204909853637,
-      "learning_rate": 0.00019999863836597491,
-      "loss": 46.0,
-      "step": 10325
-    },
-    {
-      "epoch": 1.6629091348282943,
-      "grad_norm": 0.0006122518097981811,
-      "learning_rate": 0.00019999863810195225,
-      "loss": 46.0,
-      "step": 10326
-    },
-    {
-      "epoch": 1.6630701719070817,
-      "grad_norm": 0.005554147996008396,
-      "learning_rate": 0.000199998637837904,
-      "loss": 46.0,
-      "step": 10327
-    },
-    {
-      "epoch": 1.663231208985869,
-      "grad_norm": 0.001578698051162064,
-      "learning_rate": 0.00019999863757383014,
-      "loss": 46.0,
-      "step": 10328
-    },
-    {
-      "epoch": 1.6633922460646564,
-      "grad_norm": 0.0014352636644616723,
-      "learning_rate": 0.00019999863730973068,
-      "loss": 46.0,
-      "step": 10329
-    },
-    {
-      "epoch": 1.6635532831434436,
-      "grad_norm": 0.0026615578681230545,
-      "learning_rate": 0.00019999863704560562,
-      "loss": 46.0,
-      "step": 10330
-    },
-    {
-      "epoch": 1.663714320222231,
-      "grad_norm": 0.0030296051409095526,
-      "learning_rate": 0.000199998636781455,
-      "loss": 46.0,
-      "step": 10331
-    },
-    {
-      "epoch": 1.6638753573010185,
-      "grad_norm": 0.0016279906267300248,
-      "learning_rate": 0.00019999863651727875,
-      "loss": 46.0,
-      "step": 10332
-    },
-    {
-      "epoch": 1.664036394379806,
-      "grad_norm": 0.003992682788521051,
-      "learning_rate": 0.0001999986362530769,
-      "loss": 46.0,
-      "step": 10333
-    },
-    {
-      "epoch": 1.6641974314585934,
-      "grad_norm": 0.000827592913992703,
-      "learning_rate": 0.00019999863598884948,
-      "loss": 46.0,
-      "step": 10334
-    },
-    {
-      "epoch": 1.664358468537381,
-      "grad_norm": 0.0023400168865919113,
-      "learning_rate": 0.00019999863572459642,
-      "loss": 46.0,
-      "step": 10335
-    },
-    {
-      "epoch": 1.6645195056161681,
-      "grad_norm": 0.0009639761410653591,
-      "learning_rate": 0.00019999863546031783,
-      "loss": 46.0,
-      "step": 10336
-    },
-    {
-      "epoch": 1.6646805426949554,
-      "grad_norm": 0.005953082349151373,
-      "learning_rate": 0.0001999986351960136,
-      "loss": 46.0,
-      "step": 10337
-    },
-    {
-      "epoch": 1.6648415797737428,
-      "grad_norm": 0.0012322901748120785,
-      "learning_rate": 0.0001999986349316838,
-      "loss": 46.0,
-      "step": 10338
-    },
-    {
-      "epoch": 1.6650026168525303,
-      "grad_norm": 0.0005517352255992591,
-      "learning_rate": 0.0001999986346673284,
-      "loss": 46.0,
-      "step": 10339
-    },
-    {
-      "epoch": 1.6651636539313177,
-      "grad_norm": 0.0006666937842965126,
-      "learning_rate": 0.00019999863440294742,
-      "loss": 46.0,
-      "step": 10340
-    },
-    {
-      "epoch": 1.6653246910101052,
-      "grad_norm": 0.0027998704463243484,
-      "learning_rate": 0.00019999863413854084,
-      "loss": 46.0,
-      "step": 10341
-    },
-    {
-      "epoch": 1.6654857280888926,
-      "grad_norm": 0.008945444598793983,
-      "learning_rate": 0.00019999863387410862,
-      "loss": 46.0,
-      "step": 10342
-    },
-    {
-      "epoch": 1.6656467651676798,
-      "grad_norm": 0.00030072429217398167,
-      "learning_rate": 0.00019999863360965086,
-      "loss": 46.0,
-      "step": 10343
-    },
-    {
-      "epoch": 1.6658078022464673,
-      "grad_norm": 0.002474376931786537,
-      "learning_rate": 0.00019999863334516747,
-      "loss": 46.0,
-      "step": 10344
-    },
-    {
-      "epoch": 1.6659688393252545,
-      "grad_norm": 0.0035807553213089705,
-      "learning_rate": 0.0001999986330806585,
-      "loss": 46.0,
-      "step": 10345
-    },
-    {
-      "epoch": 1.666129876404042,
-      "grad_norm": 0.0034403803292661905,
-      "learning_rate": 0.00019999863281612394,
-      "loss": 46.0,
-      "step": 10346
-    },
-    {
-      "epoch": 1.6662909134828294,
-      "grad_norm": 0.001939290203154087,
-      "learning_rate": 0.0001999986325515638,
-      "loss": 46.0,
-      "step": 10347
-    },
-    {
-      "epoch": 1.6664519505616169,
-      "grad_norm": 0.0046937232837080956,
-      "learning_rate": 0.00019999863228697802,
-      "loss": 46.0,
-      "step": 10348
-    },
-    {
-      "epoch": 1.6666129876404043,
-      "grad_norm": 0.002770220162346959,
-      "learning_rate": 0.00019999863202236668,
-      "loss": 46.0,
-      "step": 10349
-    },
-    {
-      "epoch": 1.6667740247191916,
-      "grad_norm": 0.0027611434925347567,
-      "learning_rate": 0.00019999863175772974,
-      "loss": 46.0,
-      "step": 10350
-    },
-    {
-      "epoch": 1.666935061797979,
-      "grad_norm": 0.0010709353955462575,
-      "learning_rate": 0.0001999986314930672,
-      "loss": 46.0,
-      "step": 10351
-    },
-    {
-      "epoch": 1.6670960988767662,
-      "grad_norm": 0.0037821882870048285,
-      "learning_rate": 0.00019999863122837906,
-      "loss": 46.0,
-      "step": 10352
-    },
-    {
-      "epoch": 1.6672571359555537,
-      "grad_norm": 0.005217619240283966,
-      "learning_rate": 0.00019999863096366535,
-      "loss": 46.0,
-      "step": 10353
-    },
-    {
-      "epoch": 1.6674181730343411,
-      "grad_norm": 0.004672898445278406,
-      "learning_rate": 0.00019999863069892603,
-      "loss": 46.0,
-      "step": 10354
-    },
-    {
-      "epoch": 1.6675792101131286,
-      "grad_norm": 0.0007431456469930708,
-      "learning_rate": 0.00019999863043416112,
-      "loss": 46.0,
-      "step": 10355
-    },
-    {
-      "epoch": 1.667740247191916,
-      "grad_norm": 0.003903965000063181,
-      "learning_rate": 0.0001999986301693706,
-      "loss": 46.0,
-      "step": 10356
-    },
-    {
-      "epoch": 1.6679012842707033,
-      "grad_norm": 0.002124632941558957,
-      "learning_rate": 0.00019999862990455452,
-      "loss": 46.0,
-      "step": 10357
-    },
-    {
-      "epoch": 1.6680623213494907,
-      "grad_norm": 0.0014969630865380168,
-      "learning_rate": 0.0001999986296397128,
-      "loss": 46.0,
-      "step": 10358
-    },
-    {
-      "epoch": 1.668223358428278,
-      "grad_norm": 0.0021574695128947496,
-      "learning_rate": 0.0001999986293748455,
-      "loss": 46.0,
-      "step": 10359
-    },
-    {
-      "epoch": 1.6683843955070654,
-      "grad_norm": 0.0008957255049608648,
-      "learning_rate": 0.00019999862910995261,
-      "loss": 46.0,
-      "step": 10360
-    },
-    {
-      "epoch": 1.6685454325858529,
-      "grad_norm": 0.001666371594183147,
-      "learning_rate": 0.00019999862884503416,
-      "loss": 46.0,
-      "step": 10361
-    },
-    {
-      "epoch": 1.6687064696646403,
-      "grad_norm": 0.0006663691601715982,
-      "learning_rate": 0.00019999862858009006,
-      "loss": 46.0,
-      "step": 10362
-    },
-    {
-      "epoch": 1.6688675067434278,
-      "grad_norm": 0.0009327695588581264,
-      "learning_rate": 0.00019999862831512042,
-      "loss": 46.0,
-      "step": 10363
-    },
-    {
-      "epoch": 1.6690285438222152,
-      "grad_norm": 0.002431015018373728,
-      "learning_rate": 0.00019999862805012515,
-      "loss": 46.0,
-      "step": 10364
-    },
-    {
-      "epoch": 1.6691895809010024,
-      "grad_norm": 0.0015025166794657707,
-      "learning_rate": 0.0001999986277851043,
-      "loss": 46.0,
-      "step": 10365
-    },
-    {
-      "epoch": 1.66935061797979,
-      "grad_norm": 0.001371413585729897,
-      "learning_rate": 0.0001999986275200578,
-      "loss": 46.0,
-      "step": 10366
-    },
-    {
-      "epoch": 1.6695116550585771,
-      "grad_norm": 0.003519990248605609,
-      "learning_rate": 0.00019999862725498578,
-      "loss": 46.0,
-      "step": 10367
-    },
-    {
-      "epoch": 1.6696726921373646,
-      "grad_norm": 0.006053016055375338,
-      "learning_rate": 0.00019999862698988813,
-      "loss": 46.0,
-      "step": 10368
-    },
-    {
-      "epoch": 1.669833729216152,
-      "grad_norm": 0.005175014957785606,
-      "learning_rate": 0.0001999986267247649,
-      "loss": 46.0,
-      "step": 10369
-    },
-    {
-      "epoch": 1.6699947662949395,
-      "grad_norm": 0.0027856845408678055,
-      "learning_rate": 0.00019999862645961607,
-      "loss": 46.0,
-      "step": 10370
-    },
-    {
-      "epoch": 1.670155803373727,
-      "grad_norm": 0.003092190483585,
-      "learning_rate": 0.00019999862619444163,
-      "loss": 46.0,
-      "step": 10371
-    },
-    {
-      "epoch": 1.6703168404525142,
-      "grad_norm": 0.001366854296065867,
-      "learning_rate": 0.0001999986259292416,
-      "loss": 46.0,
-      "step": 10372
-    },
-    {
-      "epoch": 1.6704778775313016,
-      "grad_norm": 0.002011434640735388,
-      "learning_rate": 0.00019999862566401602,
-      "loss": 46.0,
-      "step": 10373
-    },
-    {
-      "epoch": 1.6706389146100888,
-      "grad_norm": 0.0028051792178303003,
-      "learning_rate": 0.0001999986253987648,
-      "loss": 46.0,
-      "step": 10374
-    },
-    {
-      "epoch": 1.6707999516888763,
-      "grad_norm": 0.003266475861892104,
-      "learning_rate": 0.000199998625133488,
-      "loss": 46.0,
-      "step": 10375
-    },
-    {
-      "epoch": 1.6709609887676637,
-      "grad_norm": 0.0028007877990603447,
-      "learning_rate": 0.0001999986248681856,
-      "loss": 46.0,
-      "step": 10376
-    },
-    {
-      "epoch": 1.6711220258464512,
-      "grad_norm": 0.00100128713529557,
-      "learning_rate": 0.00019999862460285759,
-      "loss": 46.0,
-      "step": 10377
-    },
-    {
-      "epoch": 1.6712830629252386,
-      "grad_norm": 0.0011375052854418755,
-      "learning_rate": 0.000199998624337504,
-      "loss": 46.0,
-      "step": 10378
-    },
-    {
-      "epoch": 1.6714441000040259,
-      "grad_norm": 0.0004539019719231874,
-      "learning_rate": 0.00019999862407212482,
-      "loss": 46.0,
-      "step": 10379
-    },
-    {
-      "epoch": 1.6716051370828133,
-      "grad_norm": 0.0016809777589514852,
-      "learning_rate": 0.00019999862380672004,
-      "loss": 46.0,
-      "step": 10380
-    },
-    {
-      "epoch": 1.6717661741616006,
-      "grad_norm": 0.0006973856943659484,
-      "learning_rate": 0.00019999862354128965,
-      "loss": 46.0,
-      "step": 10381
-    },
-    {
-      "epoch": 1.671927211240388,
-      "grad_norm": 0.0005322683718986809,
-      "learning_rate": 0.0001999986232758337,
-      "loss": 46.0,
-      "step": 10382
-    },
-    {
-      "epoch": 1.6720882483191755,
-      "grad_norm": 0.00047642423305660486,
-      "learning_rate": 0.00019999862301035216,
-      "loss": 46.0,
-      "step": 10383
-    },
-    {
-      "epoch": 1.672249285397963,
-      "grad_norm": 0.004444230813533068,
-      "learning_rate": 0.00019999862274484498,
-      "loss": 46.0,
-      "step": 10384
-    },
-    {
-      "epoch": 1.6724103224767504,
-      "grad_norm": 0.001323769916780293,
-      "learning_rate": 0.00019999862247931221,
-      "loss": 46.0,
-      "step": 10385
-    },
-    {
-      "epoch": 1.6725713595555378,
-      "grad_norm": 0.0016513635637238622,
-      "learning_rate": 0.0001999986222137539,
-      "loss": 46.0,
-      "step": 10386
-    },
-    {
-      "epoch": 1.672732396634325,
-      "grad_norm": 0.001027752528898418,
-      "learning_rate": 0.00019999862194816997,
-      "loss": 46.0,
-      "step": 10387
-    },
-    {
-      "epoch": 1.6728934337131125,
-      "grad_norm": 0.005808922462165356,
-      "learning_rate": 0.00019999862168256042,
-      "loss": 46.0,
-      "step": 10388
-    },
-    {
-      "epoch": 1.6730544707918997,
-      "grad_norm": 0.001681674038991332,
-      "learning_rate": 0.0001999986214169253,
-      "loss": 46.0,
-      "step": 10389
-    },
-    {
-      "epoch": 1.6732155078706872,
-      "grad_norm": 0.001717392704449594,
-      "learning_rate": 0.00019999862115126457,
-      "loss": 46.0,
-      "step": 10390
-    },
-    {
-      "epoch": 1.6733765449494746,
-      "grad_norm": 0.0018831808120012283,
-      "learning_rate": 0.00019999862088557825,
-      "loss": 46.0,
-      "step": 10391
-    },
-    {
-      "epoch": 1.673537582028262,
-      "grad_norm": 0.0006793508655391634,
-      "learning_rate": 0.00019999862061986635,
-      "loss": 46.0,
-      "step": 10392
-    },
-    {
-      "epoch": 1.6736986191070495,
-      "grad_norm": 0.001540188561193645,
-      "learning_rate": 0.00019999862035412883,
-      "loss": 46.0,
-      "step": 10393
-    },
-    {
-      "epoch": 1.6738596561858368,
-      "grad_norm": 0.002044814173132181,
-      "learning_rate": 0.00019999862008836575,
-      "loss": 46.0,
-      "step": 10394
-    },
-    {
-      "epoch": 1.6740206932646242,
-      "grad_norm": 0.002085526240989566,
-      "learning_rate": 0.00019999861982257703,
-      "loss": 46.0,
-      "step": 10395
-    },
-    {
-      "epoch": 1.6741817303434114,
-      "grad_norm": 0.005129432305693626,
-      "learning_rate": 0.00019999861955676275,
-      "loss": 46.0,
-      "step": 10396
-    },
-    {
-      "epoch": 1.674342767422199,
-      "grad_norm": 0.001831444795243442,
-      "learning_rate": 0.00019999861929092288,
-      "loss": 46.0,
-      "step": 10397
-    },
-    {
-      "epoch": 1.6745038045009863,
-      "grad_norm": 0.0021805427968502045,
-      "learning_rate": 0.00019999861902505737,
-      "loss": 46.0,
-      "step": 10398
-    },
-    {
-      "epoch": 1.6746648415797738,
-      "grad_norm": 0.008991080336272717,
-      "learning_rate": 0.00019999861875916633,
-      "loss": 46.0,
-      "step": 10399
-    },
-    {
-      "epoch": 1.6748258786585613,
-      "grad_norm": 0.0030124543700367212,
-      "learning_rate": 0.00019999861849324965,
-      "loss": 46.0,
-      "step": 10400
-    },
-    {
-      "epoch": 1.6749869157373485,
-      "grad_norm": 0.0007078531198203564,
-      "learning_rate": 0.00019999861822730738,
-      "loss": 46.0,
-      "step": 10401
-    },
-    {
-      "epoch": 1.675147952816136,
-      "grad_norm": 0.0069088684394955635,
-      "learning_rate": 0.00019999861796133952,
-      "loss": 46.0,
-      "step": 10402
-    },
-    {
-      "epoch": 1.6753089898949232,
-      "grad_norm": 0.0019473673310130835,
-      "learning_rate": 0.00019999861769534608,
-      "loss": 46.0,
-      "step": 10403
-    },
-    {
-      "epoch": 1.6754700269737106,
-      "grad_norm": 0.0026252041570842266,
-      "learning_rate": 0.00019999861742932704,
-      "loss": 46.0,
-      "step": 10404
-    },
-    {
-      "epoch": 1.675631064052498,
-      "grad_norm": 0.000608101487159729,
-      "learning_rate": 0.0001999986171632824,
-      "loss": 46.0,
-      "step": 10405
-    },
-    {
-      "epoch": 1.6757921011312855,
-      "grad_norm": 0.001595841720700264,
-      "learning_rate": 0.00019999861689721217,
-      "loss": 46.0,
-      "step": 10406
-    },
-    {
-      "epoch": 1.675953138210073,
-      "grad_norm": 0.0007078467751853168,
-      "learning_rate": 0.00019999861663111634,
-      "loss": 46.0,
-      "step": 10407
-    },
-    {
-      "epoch": 1.6761141752888604,
-      "grad_norm": 0.006640944164246321,
-      "learning_rate": 0.0001999986163649949,
-      "loss": 46.0,
-      "step": 10408
-    },
-    {
-      "epoch": 1.6762752123676476,
-      "grad_norm": 0.007485799957066774,
-      "learning_rate": 0.0001999986160988479,
-      "loss": 46.0,
-      "step": 10409
-    },
-    {
-      "epoch": 1.6764362494464349,
-      "grad_norm": 0.0008861521491780877,
-      "learning_rate": 0.00019999861583267528,
-      "loss": 46.0,
-      "step": 10410
-    },
-    {
-      "epoch": 1.6765972865252223,
-      "grad_norm": 0.004999096039682627,
-      "learning_rate": 0.00019999861556647708,
-      "loss": 46.0,
-      "step": 10411
-    },
-    {
-      "epoch": 1.6767583236040098,
-      "grad_norm": 0.0010776834096759558,
-      "learning_rate": 0.00019999861530025327,
-      "loss": 46.0,
-      "step": 10412
-    },
-    {
-      "epoch": 1.6769193606827972,
-      "grad_norm": 0.00090724544133991,
-      "learning_rate": 0.00019999861503400387,
-      "loss": 46.0,
-      "step": 10413
-    },
-    {
-      "epoch": 1.6770803977615847,
-      "grad_norm": 0.0009070896776393056,
-      "learning_rate": 0.0001999986147677289,
-      "loss": 46.0,
-      "step": 10414
-    },
-    {
-      "epoch": 1.6772414348403721,
-      "grad_norm": 0.00188609235920012,
-      "learning_rate": 0.0001999986145014283,
-      "loss": 46.0,
-      "step": 10415
-    },
-    {
-      "epoch": 1.6774024719191594,
-      "grad_norm": 0.002111490350216627,
-      "learning_rate": 0.0001999986142351021,
-      "loss": 46.0,
-      "step": 10416
-    },
-    {
-      "epoch": 1.6775635089979468,
-      "grad_norm": 0.0020061295945197344,
-      "learning_rate": 0.00019999861396875036,
-      "loss": 46.0,
-      "step": 10417
-    },
-    {
-      "epoch": 1.677724546076734,
-      "grad_norm": 0.005179078318178654,
-      "learning_rate": 0.00019999861370237297,
-      "loss": 46.0,
-      "step": 10418
-    },
-    {
-      "epoch": 1.6778855831555215,
-      "grad_norm": 0.0013346009654924273,
-      "learning_rate": 0.00019999861343597002,
-      "loss": 46.0,
-      "step": 10419
-    },
-    {
-      "epoch": 1.678046620234309,
-      "grad_norm": 0.001554424874484539,
-      "learning_rate": 0.00019999861316954146,
-      "loss": 46.0,
-      "step": 10420
-    },
-    {
-      "epoch": 1.6782076573130964,
-      "grad_norm": 0.007755435537546873,
-      "learning_rate": 0.0001999986129030873,
-      "loss": 46.0,
-      "step": 10421
-    },
-    {
-      "epoch": 1.6783686943918839,
-      "grad_norm": 0.0023411880247294903,
-      "learning_rate": 0.00019999861263660755,
-      "loss": 46.0,
-      "step": 10422
-    },
-    {
-      "epoch": 1.678529731470671,
-      "grad_norm": 0.0031238319352269173,
-      "learning_rate": 0.00019999861237010222,
-      "loss": 46.0,
-      "step": 10423
-    },
-    {
-      "epoch": 1.6786907685494585,
-      "grad_norm": 0.0005404056864790618,
-      "learning_rate": 0.00019999861210357128,
-      "loss": 46.0,
-      "step": 10424
-    },
-    {
-      "epoch": 1.6788518056282458,
-      "grad_norm": 0.0013121241936460137,
-      "learning_rate": 0.00019999861183701473,
-      "loss": 46.0,
-      "step": 10425
-    },
-    {
-      "epoch": 1.6790128427070332,
-      "grad_norm": 0.002822204725816846,
-      "learning_rate": 0.00019999861157043262,
-      "loss": 46.0,
-      "step": 10426
-    },
-    {
-      "epoch": 1.6791738797858207,
-      "grad_norm": 0.009026975370943546,
-      "learning_rate": 0.0001999986113038249,
-      "loss": 46.0,
-      "step": 10427
-    },
-    {
-      "epoch": 1.6793349168646081,
-      "grad_norm": 0.002976997522637248,
-      "learning_rate": 0.0001999986110371916,
-      "loss": 46.0,
-      "step": 10428
-    },
-    {
-      "epoch": 1.6794959539433956,
-      "grad_norm": 0.0013816198334097862,
-      "learning_rate": 0.00019999861077053268,
-      "loss": 46.0,
-      "step": 10429
-    },
-    {
-      "epoch": 1.679656991022183,
-      "grad_norm": 0.002703386591747403,
-      "learning_rate": 0.0001999986105038482,
-      "loss": 46.0,
-      "step": 10430
-    },
-    {
-      "epoch": 1.6798180281009703,
-      "grad_norm": 0.001805583480745554,
-      "learning_rate": 0.0001999986102371381,
-      "loss": 46.0,
-      "step": 10431
-    },
-    {
-      "epoch": 1.6799790651797575,
-      "grad_norm": 0.0015338781522586942,
-      "learning_rate": 0.00019999860997040237,
-      "loss": 46.0,
-      "step": 10432
-    },
-    {
-      "epoch": 1.680140102258545,
-      "grad_norm": 0.0004922823281958699,
-      "learning_rate": 0.0001999986097036411,
-      "loss": 46.0,
-      "step": 10433
-    },
-    {
-      "epoch": 1.6803011393373324,
-      "grad_norm": 0.00591637659817934,
-      "learning_rate": 0.00019999860943685423,
-      "loss": 46.0,
-      "step": 10434
-    },
-    {
-      "epoch": 1.6804621764161198,
-      "grad_norm": 0.0025271170306950808,
-      "learning_rate": 0.00019999860917004173,
-      "loss": 46.0,
-      "step": 10435
-    },
-    {
-      "epoch": 1.6806232134949073,
-      "grad_norm": 0.0028867062646895647,
-      "learning_rate": 0.0001999986089032037,
-      "loss": 46.0,
-      "step": 10436
-    },
-    {
-      "epoch": 1.6807842505736947,
-      "grad_norm": 0.002870681695640087,
-      "learning_rate": 0.00019999860863634,
-      "loss": 46.0,
-      "step": 10437
-    },
-    {
-      "epoch": 1.680945287652482,
-      "grad_norm": 0.0017795058665797114,
-      "learning_rate": 0.00019999860836945074,
-      "loss": 46.0,
-      "step": 10438
-    },
-    {
-      "epoch": 1.6811063247312694,
-      "grad_norm": 0.004960164427757263,
-      "learning_rate": 0.0001999986081025359,
-      "loss": 46.0,
-      "step": 10439
-    },
-    {
-      "epoch": 1.6812673618100566,
-      "grad_norm": 0.005027100909501314,
-      "learning_rate": 0.00019999860783559545,
-      "loss": 46.0,
-      "step": 10440
-    },
-    {
-      "epoch": 1.681428398888844,
-      "grad_norm": 0.0020708946976810694,
-      "learning_rate": 0.0001999986075686294,
-      "loss": 46.0,
-      "step": 10441
-    },
-    {
-      "epoch": 1.6815894359676316,
-      "grad_norm": 0.002186340279877186,
-      "learning_rate": 0.00019999860730163778,
-      "loss": 46.0,
-      "step": 10442
-    },
-    {
-      "epoch": 1.681750473046419,
-      "grad_norm": 0.0005775339668616652,
-      "learning_rate": 0.00019999860703462052,
-      "loss": 46.0,
-      "step": 10443
-    },
-    {
-      "epoch": 1.6819115101252065,
-      "grad_norm": 0.010404218919575214,
-      "learning_rate": 0.0001999986067675777,
-      "loss": 46.0,
-      "step": 10444
-    },
-    {
-      "epoch": 1.6820725472039937,
-      "grad_norm": 0.0047836159355938435,
-      "learning_rate": 0.00019999860650050927,
-      "loss": 46.0,
-      "step": 10445
-    },
-    {
-      "epoch": 1.6822335842827811,
-      "grad_norm": 0.0027169575914740562,
-      "learning_rate": 0.00019999860623341528,
-      "loss": 46.0,
-      "step": 10446
-    },
-    {
-      "epoch": 1.6823946213615684,
-      "grad_norm": 0.003834177041426301,
-      "learning_rate": 0.00019999860596629567,
-      "loss": 46.0,
-      "step": 10447
-    },
-    {
-      "epoch": 1.6825556584403558,
-      "grad_norm": 0.0018635675078257918,
-      "learning_rate": 0.00019999860569915048,
-      "loss": 46.0,
-      "step": 10448
-    },
-    {
-      "epoch": 1.6827166955191433,
-      "grad_norm": 0.003225313499569893,
-      "learning_rate": 0.00019999860543197965,
-      "loss": 46.0,
-      "step": 10449
-    },
-    {
-      "epoch": 1.6828777325979307,
-      "grad_norm": 0.005864087026566267,
-      "learning_rate": 0.00019999860516478325,
-      "loss": 46.0,
-      "step": 10450
-    },
-    {
-      "epoch": 1.6830387696767182,
-      "grad_norm": 0.0051914844661951065,
-      "learning_rate": 0.00019999860489756127,
-      "loss": 46.0,
-      "step": 10451
-    },
-    {
-      "epoch": 1.6831998067555054,
-      "grad_norm": 0.003226950066164136,
-      "learning_rate": 0.0001999986046303137,
-      "loss": 46.0,
-      "step": 10452
-    },
-    {
-      "epoch": 1.6833608438342929,
-      "grad_norm": 0.002338156569749117,
-      "learning_rate": 0.00019999860436304052,
-      "loss": 46.0,
-      "step": 10453
-    },
-    {
-      "epoch": 1.68352188091308,
-      "grad_norm": 0.0019909224938601255,
-      "learning_rate": 0.00019999860409574172,
-      "loss": 46.0,
-      "step": 10454
-    },
-    {
-      "epoch": 1.6836829179918675,
-      "grad_norm": 0.001400559674948454,
-      "learning_rate": 0.0001999986038284174,
-      "loss": 46.0,
-      "step": 10455
-    },
-    {
-      "epoch": 1.683843955070655,
-      "grad_norm": 0.0006413241499103606,
-      "learning_rate": 0.00019999860356106742,
-      "loss": 46.0,
-      "step": 10456
-    },
-    {
-      "epoch": 1.6840049921494424,
-      "grad_norm": 0.0011588013730943203,
-      "learning_rate": 0.0001999986032936919,
-      "loss": 46.0,
-      "step": 10457
-    },
-    {
-      "epoch": 1.6841660292282299,
-      "grad_norm": 0.0029584101866930723,
-      "learning_rate": 0.00019999860302629072,
-      "loss": 46.0,
-      "step": 10458
-    },
-    {
-      "epoch": 1.6843270663070173,
-      "grad_norm": 0.0014233908150345087,
-      "learning_rate": 0.00019999860275886399,
-      "loss": 46.0,
-      "step": 10459
-    },
-    {
-      "epoch": 1.6844881033858046,
-      "grad_norm": 0.0007274019299075007,
-      "learning_rate": 0.00019999860249141164,
-      "loss": 46.0,
-      "step": 10460
-    },
-    {
-      "epoch": 1.684649140464592,
-      "grad_norm": 0.002970034023746848,
-      "learning_rate": 0.0001999986022239337,
-      "loss": 46.0,
-      "step": 10461
-    },
-    {
-      "epoch": 1.6848101775433793,
-      "grad_norm": 0.0008836840861476958,
-      "learning_rate": 0.00019999860195643016,
-      "loss": 46.0,
-      "step": 10462
-    },
-    {
-      "epoch": 1.6849712146221667,
-      "grad_norm": 0.00044969565351493657,
-      "learning_rate": 0.00019999860168890105,
-      "loss": 46.0,
-      "step": 10463
-    },
-    {
-      "epoch": 1.6851322517009542,
-      "grad_norm": 0.010037108324468136,
-      "learning_rate": 0.00019999860142134635,
-      "loss": 46.0,
-      "step": 10464
-    },
-    {
-      "epoch": 1.6852932887797416,
-      "grad_norm": 0.0020440032240003347,
-      "learning_rate": 0.00019999860115376604,
-      "loss": 46.0,
-      "step": 10465
-    },
-    {
-      "epoch": 1.685454325858529,
-      "grad_norm": 0.005088374018669128,
-      "learning_rate": 0.00019999860088616012,
-      "loss": 46.0,
-      "step": 10466
-    },
-    {
-      "epoch": 1.6856153629373163,
-      "grad_norm": 0.0041874367743730545,
-      "learning_rate": 0.00019999860061852863,
-      "loss": 46.0,
-      "step": 10467
-    },
-    {
-      "epoch": 1.6857764000161037,
-      "grad_norm": 0.0022942230571061373,
-      "learning_rate": 0.00019999860035087153,
-      "loss": 46.0,
-      "step": 10468
-    },
-    {
-      "epoch": 1.685937437094891,
-      "grad_norm": 0.005675806198269129,
-      "learning_rate": 0.00019999860008318885,
-      "loss": 46.0,
-      "step": 10469
-    },
-    {
-      "epoch": 1.6860984741736784,
-      "grad_norm": 0.0020505287684500217,
-      "learning_rate": 0.00019999859981548057,
-      "loss": 46.0,
-      "step": 10470
-    },
-    {
-      "epoch": 1.6862595112524659,
-      "grad_norm": 0.0015413566725328565,
-      "learning_rate": 0.0001999985995477467,
-      "loss": 46.0,
-      "step": 10471
-    },
-    {
-      "epoch": 1.6864205483312533,
-      "grad_norm": 0.0045183924958109856,
-      "learning_rate": 0.0001999985992799872,
-      "loss": 46.0,
-      "step": 10472
-    },
-    {
-      "epoch": 1.6865815854100408,
-      "grad_norm": 0.009042619727551937,
-      "learning_rate": 0.00019999859901220215,
-      "loss": 46.0,
-      "step": 10473
-    },
-    {
-      "epoch": 1.686742622488828,
-      "grad_norm": 0.004377848003059626,
-      "learning_rate": 0.00019999859874439147,
-      "loss": 46.0,
-      "step": 10474
-    },
-    {
-      "epoch": 1.6869036595676155,
-      "grad_norm": 0.0050247712060809135,
-      "learning_rate": 0.0001999985984765552,
-      "loss": 46.0,
-      "step": 10475
-    },
-    {
-      "epoch": 1.6870646966464027,
-      "grad_norm": 0.006816830951720476,
-      "learning_rate": 0.0001999985982086934,
-      "loss": 46.0,
-      "step": 10476
-    },
-    {
-      "epoch": 1.6872257337251901,
-      "grad_norm": 0.0011554453521966934,
-      "learning_rate": 0.00019999859794080592,
-      "loss": 46.0,
-      "step": 10477
-    },
-    {
-      "epoch": 1.6873867708039776,
-      "grad_norm": 0.001298525370657444,
-      "learning_rate": 0.00019999859767289287,
-      "loss": 46.0,
-      "step": 10478
-    },
-    {
-      "epoch": 1.687547807882765,
-      "grad_norm": 0.007545560132712126,
-      "learning_rate": 0.00019999859740495426,
-      "loss": 46.0,
-      "step": 10479
-    },
-    {
-      "epoch": 1.6877088449615525,
-      "grad_norm": 0.0017515876097604632,
-      "learning_rate": 0.00019999859713699003,
-      "loss": 46.0,
-      "step": 10480
-    },
-    {
-      "epoch": 1.68786988204034,
-      "grad_norm": 0.005050517152994871,
-      "learning_rate": 0.00019999859686900022,
-      "loss": 46.0,
-      "step": 10481
-    },
-    {
-      "epoch": 1.6880309191191272,
-      "grad_norm": 0.0007013885769993067,
-      "learning_rate": 0.0001999985966009848,
-      "loss": 46.0,
-      "step": 10482
-    },
-    {
-      "epoch": 1.6881919561979146,
-      "grad_norm": 0.007724381051957607,
-      "learning_rate": 0.00019999859633294378,
-      "loss": 46.0,
-      "step": 10483
-    },
-    {
-      "epoch": 1.6883529932767019,
-      "grad_norm": 0.0010990517912432551,
-      "learning_rate": 0.00019999859606487718,
-      "loss": 46.0,
-      "step": 10484
-    },
-    {
-      "epoch": 1.6885140303554893,
-      "grad_norm": 0.0032372481655329466,
-      "learning_rate": 0.00019999859579678496,
-      "loss": 46.0,
-      "step": 10485
-    },
-    {
-      "epoch": 1.6886750674342768,
-      "grad_norm": 0.0037654205225408077,
-      "learning_rate": 0.00019999859552866718,
-      "loss": 46.0,
-      "step": 10486
-    },
-    {
-      "epoch": 1.6888361045130642,
-      "grad_norm": 0.004126657731831074,
-      "learning_rate": 0.0001999985952605238,
-      "loss": 46.0,
-      "step": 10487
-    },
-    {
-      "epoch": 1.6889971415918517,
-      "grad_norm": 0.005669938400387764,
-      "learning_rate": 0.00019999859499235482,
-      "loss": 46.0,
-      "step": 10488
-    },
-    {
-      "epoch": 1.6891581786706389,
-      "grad_norm": 0.007425690069794655,
-      "learning_rate": 0.00019999859472416022,
-      "loss": 46.0,
-      "step": 10489
-    },
-    {
-      "epoch": 1.6893192157494263,
-      "grad_norm": 0.0020629935897886753,
-      "learning_rate": 0.00019999859445594004,
-      "loss": 46.0,
-      "step": 10490
-    },
-    {
-      "epoch": 1.6894802528282136,
-      "grad_norm": 0.005646908655762672,
-      "learning_rate": 0.00019999859418769428,
-      "loss": 46.0,
-      "step": 10491
-    },
-    {
-      "epoch": 1.689641289907001,
-      "grad_norm": 0.004508323967456818,
-      "learning_rate": 0.00019999859391942292,
-      "loss": 46.0,
-      "step": 10492
-    },
-    {
-      "epoch": 1.6898023269857885,
-      "grad_norm": 0.0023845683317631483,
-      "learning_rate": 0.00019999859365112593,
-      "loss": 46.0,
-      "step": 10493
-    },
-    {
-      "epoch": 1.689963364064576,
-      "grad_norm": 0.003487972542643547,
-      "learning_rate": 0.0001999985933828034,
-      "loss": 46.0,
-      "step": 10494
-    },
-    {
-      "epoch": 1.6901244011433634,
-      "grad_norm": 0.0010903405491262674,
-      "learning_rate": 0.00019999859311445526,
-      "loss": 46.0,
-      "step": 10495
-    },
-    {
-      "epoch": 1.6902854382221506,
-      "grad_norm": 0.0012227409752085805,
-      "learning_rate": 0.0001999985928460815,
-      "loss": 46.0,
-      "step": 10496
-    },
-    {
-      "epoch": 1.690446475300938,
-      "grad_norm": 0.008045082911849022,
-      "learning_rate": 0.0001999985925776822,
-      "loss": 46.0,
-      "step": 10497
-    },
-    {
-      "epoch": 1.6906075123797253,
-      "grad_norm": 0.004485879093408585,
-      "learning_rate": 0.00019999859230925723,
-      "loss": 46.0,
-      "step": 10498
-    },
-    {
-      "epoch": 1.6907685494585127,
-      "grad_norm": 0.009232739917933941,
-      "learning_rate": 0.00019999859204080674,
-      "loss": 46.0,
-      "step": 10499
-    },
-    {
-      "epoch": 1.6909295865373002,
-      "grad_norm": 0.0043303947895765305,
-      "learning_rate": 0.0001999985917723306,
-      "loss": 46.0,
-      "step": 10500
-    },
-    {
-      "epoch": 1.6910906236160876,
-      "grad_norm": 0.003724894020706415,
-      "learning_rate": 0.0001999985915038289,
-      "loss": 46.0,
-      "step": 10501
-    },
-    {
-      "epoch": 1.691251660694875,
-      "grad_norm": 0.0005904401186853647,
-      "learning_rate": 0.0001999985912353016,
-      "loss": 46.0,
-      "step": 10502
-    },
-    {
-      "epoch": 1.6914126977736625,
-      "grad_norm": 0.004251156002283096,
-      "learning_rate": 0.0001999985909667487,
-      "loss": 46.0,
-      "step": 10503
-    },
-    {
-      "epoch": 1.6915737348524498,
-      "grad_norm": 0.001525015919469297,
-      "learning_rate": 0.0001999985906981702,
-      "loss": 46.0,
-      "step": 10504
-    },
-    {
-      "epoch": 1.691734771931237,
-      "grad_norm": 0.0038381016347557306,
-      "learning_rate": 0.0001999985904295661,
-      "loss": 46.0,
-      "step": 10505
-    },
-    {
-      "epoch": 1.6918958090100245,
-      "grad_norm": 0.0017822892405092716,
-      "learning_rate": 0.00019999859016093643,
-      "loss": 46.0,
-      "step": 10506
-    },
-    {
-      "epoch": 1.692056846088812,
-      "grad_norm": 0.004787055775523186,
-      "learning_rate": 0.00019999858989228113,
-      "loss": 46.0,
-      "step": 10507
-    },
-    {
-      "epoch": 1.6922178831675994,
-      "grad_norm": 0.0019879417959600687,
-      "learning_rate": 0.00019999858962360025,
-      "loss": 46.0,
-      "step": 10508
-    },
-    {
-      "epoch": 1.6923789202463868,
-      "grad_norm": 0.0044783358462154865,
-      "learning_rate": 0.00019999858935489378,
-      "loss": 46.0,
-      "step": 10509
-    },
-    {
-      "epoch": 1.6925399573251743,
-      "grad_norm": 0.003443249501287937,
-      "learning_rate": 0.00019999858908616172,
-      "loss": 46.0,
-      "step": 10510
-    },
-    {
-      "epoch": 1.6927009944039615,
-      "grad_norm": 0.0019783645402640104,
-      "learning_rate": 0.00019999858881740407,
-      "loss": 46.0,
-      "step": 10511
-    },
-    {
-      "epoch": 1.692862031482749,
-      "grad_norm": 0.0028253672644495964,
-      "learning_rate": 0.00019999858854862081,
-      "loss": 46.0,
-      "step": 10512
-    },
-    {
-      "epoch": 1.6930230685615362,
-      "grad_norm": 0.0017712815897539258,
-      "learning_rate": 0.00019999858827981197,
-      "loss": 46.0,
-      "step": 10513
-    },
-    {
-      "epoch": 1.6931841056403236,
-      "grad_norm": 0.0016477230237796903,
-      "learning_rate": 0.00019999858801097753,
-      "loss": 46.0,
-      "step": 10514
-    },
-    {
-      "epoch": 1.693345142719111,
-      "grad_norm": 0.0048409923911094666,
-      "learning_rate": 0.0001999985877421175,
-      "loss": 46.0,
-      "step": 10515
-    },
-    {
-      "epoch": 1.6935061797978985,
-      "grad_norm": 0.005189493298530579,
-      "learning_rate": 0.00019999858747323185,
-      "loss": 46.0,
-      "step": 10516
-    },
-    {
-      "epoch": 1.693667216876686,
-      "grad_norm": 0.0038858221378177404,
-      "learning_rate": 0.00019999858720432063,
-      "loss": 46.0,
-      "step": 10517
-    },
-    {
-      "epoch": 1.6938282539554732,
-      "grad_norm": 0.008748218417167664,
-      "learning_rate": 0.0001999985869353838,
-      "loss": 46.0,
-      "step": 10518
-    },
-    {
-      "epoch": 1.6939892910342607,
-      "grad_norm": 0.0016101914225146174,
-      "learning_rate": 0.0001999985866664214,
-      "loss": 46.0,
-      "step": 10519
-    },
-    {
-      "epoch": 1.6941503281130479,
-      "grad_norm": 0.006393904332071543,
-      "learning_rate": 0.00019999858639743338,
-      "loss": 46.0,
-      "step": 10520
-    },
-    {
-      "epoch": 1.6943113651918353,
-      "grad_norm": 0.003465829649940133,
-      "learning_rate": 0.00019999858612841978,
-      "loss": 46.0,
-      "step": 10521
-    },
-    {
-      "epoch": 1.6944724022706228,
-      "grad_norm": 0.0013447541277855635,
-      "learning_rate": 0.00019999858585938057,
-      "loss": 46.0,
-      "step": 10522
-    },
-    {
-      "epoch": 1.6946334393494102,
-      "grad_norm": 0.014294285327196121,
-      "learning_rate": 0.0001999985855903158,
-      "loss": 46.0,
-      "step": 10523
-    },
-    {
-      "epoch": 1.6947944764281977,
-      "grad_norm": 0.0008617123239673674,
-      "learning_rate": 0.00019999858532122539,
-      "loss": 46.0,
-      "step": 10524
-    },
-    {
-      "epoch": 1.6949555135069851,
-      "grad_norm": 0.001656767213717103,
-      "learning_rate": 0.0001999985850521094,
-      "loss": 46.0,
-      "step": 10525
-    },
-    {
-      "epoch": 1.6951165505857724,
-      "grad_norm": 0.0006299621309153736,
-      "learning_rate": 0.00019999858478296785,
-      "loss": 46.0,
-      "step": 10526
-    },
-    {
-      "epoch": 1.6952775876645596,
-      "grad_norm": 0.00292469235137105,
-      "learning_rate": 0.0001999985845138007,
-      "loss": 46.0,
-      "step": 10527
-    },
-    {
-      "epoch": 1.695438624743347,
-      "grad_norm": 0.00276189879514277,
-      "learning_rate": 0.0001999985842446079,
-      "loss": 46.0,
-      "step": 10528
-    },
-    {
-      "epoch": 1.6955996618221345,
-      "grad_norm": 0.0007011765846982598,
-      "learning_rate": 0.00019999858397538956,
-      "loss": 46.0,
-      "step": 10529
-    },
-    {
-      "epoch": 1.695760698900922,
-      "grad_norm": 0.003778912825509906,
-      "learning_rate": 0.0001999985837061456,
-      "loss": 46.0,
-      "step": 10530
-    },
-    {
-      "epoch": 1.6959217359797094,
-      "grad_norm": 0.0007657675887458026,
-      "learning_rate": 0.00019999858343687605,
-      "loss": 46.0,
-      "step": 10531
-    },
-    {
-      "epoch": 1.6960827730584969,
-      "grad_norm": 0.002568767871707678,
-      "learning_rate": 0.0001999985831675809,
-      "loss": 46.0,
-      "step": 10532
-    },
-    {
-      "epoch": 1.696243810137284,
-      "grad_norm": 0.0007921484066173434,
-      "learning_rate": 0.00019999858289826018,
-      "loss": 46.0,
-      "step": 10533
-    },
-    {
-      "epoch": 1.6964048472160715,
-      "grad_norm": 0.001671319711022079,
-      "learning_rate": 0.00019999858262891384,
-      "loss": 46.0,
-      "step": 10534
-    },
-    {
-      "epoch": 1.6965658842948588,
-      "grad_norm": 0.0014859768562018871,
-      "learning_rate": 0.00019999858235954192,
-      "loss": 46.0,
-      "step": 10535
-    },
-    {
-      "epoch": 1.6967269213736462,
-      "grad_norm": 0.0006414958625100553,
-      "learning_rate": 0.0001999985820901444,
-      "loss": 46.0,
-      "step": 10536
-    },
-    {
-      "epoch": 1.6968879584524337,
-      "grad_norm": 0.013287168927490711,
-      "learning_rate": 0.0001999985818207213,
-      "loss": 46.0,
-      "step": 10537
-    },
-    {
-      "epoch": 1.6970489955312211,
-      "grad_norm": 0.0016671044286340475,
-      "learning_rate": 0.00019999858155127256,
-      "loss": 46.0,
-      "step": 10538
-    },
-    {
-      "epoch": 1.6972100326100086,
-      "grad_norm": 0.005191594362258911,
-      "learning_rate": 0.00019999858128179828,
-      "loss": 46.0,
-      "step": 10539
-    },
-    {
-      "epoch": 1.6973710696887958,
-      "grad_norm": 0.0026636270340532064,
-      "learning_rate": 0.00019999858101229836,
-      "loss": 46.0,
-      "step": 10540
-    },
-    {
-      "epoch": 1.6975321067675833,
-      "grad_norm": 0.006815551780164242,
-      "learning_rate": 0.00019999858074277286,
-      "loss": 46.0,
-      "step": 10541
-    },
-    {
-      "epoch": 1.6976931438463705,
-      "grad_norm": 0.0030364496633410454,
-      "learning_rate": 0.0001999985804732218,
-      "loss": 46.0,
-      "step": 10542
-    },
-    {
-      "epoch": 1.697854180925158,
-      "grad_norm": 0.004536924418061972,
-      "learning_rate": 0.0001999985802036451,
-      "loss": 46.0,
-      "step": 10543
-    },
-    {
-      "epoch": 1.6980152180039454,
-      "grad_norm": 0.0014559825649484992,
-      "learning_rate": 0.00019999857993404282,
-      "loss": 46.0,
-      "step": 10544
-    },
-    {
-      "epoch": 1.6981762550827328,
-      "grad_norm": 0.0029906032141298056,
-      "learning_rate": 0.00019999857966441494,
-      "loss": 46.0,
-      "step": 10545
-    },
-    {
-      "epoch": 1.6983372921615203,
-      "grad_norm": 0.0011763200163841248,
-      "learning_rate": 0.0001999985793947615,
-      "loss": 46.0,
-      "step": 10546
-    },
-    {
-      "epoch": 1.6984983292403077,
-      "grad_norm": 0.0030931145884096622,
-      "learning_rate": 0.00019999857912508242,
-      "loss": 46.0,
-      "step": 10547
-    },
-    {
-      "epoch": 1.698659366319095,
-      "grad_norm": 0.0013452148996293545,
-      "learning_rate": 0.00019999857885537775,
-      "loss": 46.0,
-      "step": 10548
-    },
-    {
-      "epoch": 1.6988204033978822,
-      "grad_norm": 0.003104675794020295,
-      "learning_rate": 0.00019999857858564752,
-      "loss": 46.0,
-      "step": 10549
-    },
-    {
-      "epoch": 1.6989814404766697,
-      "grad_norm": 0.002120691817253828,
-      "learning_rate": 0.00019999857831589168,
-      "loss": 46.0,
-      "step": 10550
-    },
-    {
-      "epoch": 1.699142477555457,
-      "grad_norm": 0.0011800124775618315,
-      "learning_rate": 0.00019999857804611025,
-      "loss": 46.0,
-      "step": 10551
-    },
-    {
-      "epoch": 1.6993035146342446,
-      "grad_norm": 0.001063268049620092,
-      "learning_rate": 0.0001999985777763032,
-      "loss": 46.0,
-      "step": 10552
-    },
-    {
-      "epoch": 1.699464551713032,
-      "grad_norm": 0.0006542173214256763,
-      "learning_rate": 0.00019999857750647057,
-      "loss": 46.0,
-      "step": 10553
-    },
-    {
-      "epoch": 1.6996255887918195,
-      "grad_norm": 0.0008364635286852717,
-      "learning_rate": 0.00019999857723661238,
-      "loss": 46.0,
-      "step": 10554
-    },
-    {
-      "epoch": 1.6997866258706067,
-      "grad_norm": 0.0070970249362289906,
-      "learning_rate": 0.00019999857696672852,
-      "loss": 46.0,
-      "step": 10555
-    },
-    {
-      "epoch": 1.6999476629493941,
-      "grad_norm": 0.002407848834991455,
-      "learning_rate": 0.00019999857669681913,
-      "loss": 46.0,
-      "step": 10556
-    },
-    {
-      "epoch": 1.7001087000281814,
-      "grad_norm": 0.0021727916318923235,
-      "learning_rate": 0.00019999857642688412,
-      "loss": 46.0,
-      "step": 10557
-    },
-    {
-      "epoch": 1.7002697371069688,
-      "grad_norm": 0.001900134957395494,
-      "learning_rate": 0.00019999857615692353,
-      "loss": 46.0,
-      "step": 10558
-    },
-    {
-      "epoch": 1.7004307741857563,
-      "grad_norm": 0.0013372318353503942,
-      "learning_rate": 0.00019999857588693735,
-      "loss": 46.0,
-      "step": 10559
-    },
-    {
-      "epoch": 1.7005918112645437,
-      "grad_norm": 0.0026397083420306444,
-      "learning_rate": 0.00019999857561692555,
-      "loss": 46.0,
-      "step": 10560
-    },
-    {
-      "epoch": 1.7007528483433312,
-      "grad_norm": 0.003049772698432207,
-      "learning_rate": 0.00019999857534688814,
-      "loss": 46.0,
-      "step": 10561
-    },
-    {
-      "epoch": 1.7009138854221184,
-      "grad_norm": 0.0014704653294757009,
-      "learning_rate": 0.00019999857507682517,
-      "loss": 46.0,
-      "step": 10562
-    },
-    {
-      "epoch": 1.7010749225009059,
-      "grad_norm": 0.0013092717854306102,
-      "learning_rate": 0.0001999985748067366,
-      "loss": 46.0,
-      "step": 10563
-    },
-    {
-      "epoch": 1.701235959579693,
-      "grad_norm": 0.0011484537972137332,
-      "learning_rate": 0.00019999857453662244,
-      "loss": 46.0,
-      "step": 10564
-    },
-    {
-      "epoch": 1.7013969966584805,
-      "grad_norm": 0.006104714702814817,
-      "learning_rate": 0.00019999857426648268,
-      "loss": 46.0,
-      "step": 10565
-    },
-    {
-      "epoch": 1.701558033737268,
-      "grad_norm": 0.0030514909885823727,
-      "learning_rate": 0.0001999985739963173,
-      "loss": 46.0,
-      "step": 10566
-    },
-    {
-      "epoch": 1.7017190708160554,
-      "grad_norm": 0.0008539797854609787,
-      "learning_rate": 0.00019999857372612637,
-      "loss": 46.0,
-      "step": 10567
-    },
-    {
-      "epoch": 1.701880107894843,
-      "grad_norm": 0.002890487667173147,
-      "learning_rate": 0.00019999857345590982,
-      "loss": 46.0,
-      "step": 10568
-    },
-    {
-      "epoch": 1.7020411449736301,
-      "grad_norm": 0.0019331686198711395,
-      "learning_rate": 0.0001999985731856677,
-      "loss": 46.0,
-      "step": 10569
-    },
-    {
-      "epoch": 1.7022021820524176,
-      "grad_norm": 0.0006012069643475115,
-      "learning_rate": 0.00019999857291539997,
-      "loss": 46.0,
-      "step": 10570
-    },
-    {
-      "epoch": 1.7023632191312048,
-      "grad_norm": 0.005402646958827972,
-      "learning_rate": 0.00019999857264510663,
-      "loss": 46.0,
-      "step": 10571
-    },
-    {
-      "epoch": 1.7025242562099923,
-      "grad_norm": 0.0021092521492391825,
-      "learning_rate": 0.0001999985723747877,
-      "loss": 46.0,
-      "step": 10572
-    },
-    {
-      "epoch": 1.7026852932887797,
-      "grad_norm": 0.001945003285072744,
-      "learning_rate": 0.0001999985721044432,
-      "loss": 46.0,
-      "step": 10573
-    },
-    {
-      "epoch": 1.7028463303675672,
-      "grad_norm": 0.002139392076060176,
-      "learning_rate": 0.00019999857183407307,
-      "loss": 46.0,
-      "step": 10574
-    },
-    {
-      "epoch": 1.7030073674463546,
-      "grad_norm": 0.000649197434540838,
-      "learning_rate": 0.00019999857156367736,
-      "loss": 46.0,
-      "step": 10575
-    },
-    {
-      "epoch": 1.703168404525142,
-      "grad_norm": 0.002186570083722472,
-      "learning_rate": 0.0001999985712932561,
-      "loss": 46.0,
-      "step": 10576
-    },
-    {
-      "epoch": 1.7033294416039293,
-      "grad_norm": 0.0016852536937221885,
-      "learning_rate": 0.00019999857102280917,
-      "loss": 46.0,
-      "step": 10577
-    },
-    {
-      "epoch": 1.7034904786827167,
-      "grad_norm": 0.0052175563760101795,
-      "learning_rate": 0.00019999857075233667,
-      "loss": 46.0,
-      "step": 10578
-    },
-    {
-      "epoch": 1.703651515761504,
-      "grad_norm": 0.004924371372908354,
-      "learning_rate": 0.00019999857048183859,
-      "loss": 46.0,
-      "step": 10579
-    },
-    {
-      "epoch": 1.7038125528402914,
-      "grad_norm": 0.002078639343380928,
-      "learning_rate": 0.0001999985702113149,
-      "loss": 46.0,
-      "step": 10580
-    },
-    {
-      "epoch": 1.7039735899190789,
-      "grad_norm": 0.0028434270061552525,
-      "learning_rate": 0.00019999856994076565,
-      "loss": 46.0,
-      "step": 10581
-    },
-    {
-      "epoch": 1.7041346269978663,
-      "grad_norm": 0.003825762076303363,
-      "learning_rate": 0.00019999856967019077,
-      "loss": 46.0,
-      "step": 10582
-    },
-    {
-      "epoch": 1.7042956640766538,
-      "grad_norm": 0.0025177013594657183,
-      "learning_rate": 0.0001999985693995903,
-      "loss": 46.0,
-      "step": 10583
-    },
-    {
-      "epoch": 1.704456701155441,
-      "grad_norm": 0.00609858613461256,
-      "learning_rate": 0.00019999856912896426,
-      "loss": 46.0,
-      "step": 10584
-    },
-    {
-      "epoch": 1.7046177382342285,
-      "grad_norm": 0.0009502783650532365,
-      "learning_rate": 0.0001999985688583126,
-      "loss": 46.0,
-      "step": 10585
-    },
-    {
-      "epoch": 1.7047787753130157,
-      "grad_norm": 0.003186178160831332,
-      "learning_rate": 0.00019999856858763534,
-      "loss": 46.0,
-      "step": 10586
-    },
-    {
-      "epoch": 1.7049398123918031,
-      "grad_norm": 0.0022999586071819067,
-      "learning_rate": 0.0001999985683169325,
-      "loss": 46.0,
-      "step": 10587
-    },
-    {
-      "epoch": 1.7051008494705906,
-      "grad_norm": 0.0005596989649347961,
-      "learning_rate": 0.00019999856804620407,
-      "loss": 46.0,
-      "step": 10588
-    },
-    {
-      "epoch": 1.705261886549378,
-      "grad_norm": 0.0008181995945051312,
-      "learning_rate": 0.00019999856777545003,
-      "loss": 46.0,
-      "step": 10589
-    },
-    {
-      "epoch": 1.7054229236281655,
-      "grad_norm": 0.0013957306509837508,
-      "learning_rate": 0.0001999985675046704,
-      "loss": 46.0,
-      "step": 10590
-    },
-    {
-      "epoch": 1.7055839607069527,
-      "grad_norm": 0.002393920673057437,
-      "learning_rate": 0.00019999856723386522,
-      "loss": 46.0,
-      "step": 10591
-    },
-    {
-      "epoch": 1.7057449977857402,
-      "grad_norm": 0.0023064424749463797,
-      "learning_rate": 0.0001999985669630344,
-      "loss": 46.0,
-      "step": 10592
-    },
-    {
-      "epoch": 1.7059060348645274,
-      "grad_norm": 0.0005696122534573078,
-      "learning_rate": 0.00019999856669217797,
-      "loss": 46.0,
-      "step": 10593
-    },
-    {
-      "epoch": 1.7060670719433149,
-      "grad_norm": 0.0012736358912661672,
-      "learning_rate": 0.000199998566421296,
-      "loss": 46.0,
-      "step": 10594
-    },
-    {
-      "epoch": 1.7062281090221023,
-      "grad_norm": 0.0030530868098139763,
-      "learning_rate": 0.00019999856615038838,
-      "loss": 46.0,
-      "step": 10595
-    },
-    {
-      "epoch": 1.7063891461008898,
-      "grad_norm": 0.005021310411393642,
-      "learning_rate": 0.0001999985658794552,
-      "loss": 46.0,
-      "step": 10596
-    },
-    {
-      "epoch": 1.7065501831796772,
-      "grad_norm": 0.002605863381177187,
-      "learning_rate": 0.00019999856560849638,
-      "loss": 46.0,
-      "step": 10597
-    },
-    {
-      "epoch": 1.7067112202584647,
-      "grad_norm": 0.0021888704504817724,
-      "learning_rate": 0.00019999856533751203,
-      "loss": 46.0,
-      "step": 10598
-    },
-    {
-      "epoch": 1.706872257337252,
-      "grad_norm": 0.0046259211376309395,
-      "learning_rate": 0.00019999856506650206,
-      "loss": 46.0,
-      "step": 10599
-    },
-    {
-      "epoch": 1.7070332944160393,
-      "grad_norm": 0.0022350698709487915,
-      "learning_rate": 0.00019999856479546648,
-      "loss": 46.0,
-      "step": 10600
-    },
-    {
-      "epoch": 1.7071943314948266,
-      "grad_norm": 0.002283544046804309,
-      "learning_rate": 0.0001999985645244053,
-      "loss": 46.0,
-      "step": 10601
-    },
-    {
-      "epoch": 1.707355368573614,
-      "grad_norm": 0.007390520069748163,
-      "learning_rate": 0.00019999856425331855,
-      "loss": 46.0,
-      "step": 10602
-    },
-    {
-      "epoch": 1.7075164056524015,
-      "grad_norm": 0.0028412521351128817,
-      "learning_rate": 0.0001999985639822062,
-      "loss": 46.0,
-      "step": 10603
-    },
-    {
-      "epoch": 1.707677442731189,
-      "grad_norm": 0.0016974268946796656,
-      "learning_rate": 0.00019999856371106825,
-      "loss": 46.0,
-      "step": 10604
-    },
-    {
-      "epoch": 1.7078384798099764,
-      "grad_norm": 0.0005231903051026165,
-      "learning_rate": 0.00019999856343990474,
-      "loss": 46.0,
-      "step": 10605
-    },
-    {
-      "epoch": 1.7079995168887636,
-      "grad_norm": 0.0020515720825642347,
-      "learning_rate": 0.00019999856316871558,
-      "loss": 46.0,
-      "step": 10606
-    },
-    {
-      "epoch": 1.708160553967551,
-      "grad_norm": 0.0014094403013586998,
-      "learning_rate": 0.00019999856289750086,
-      "loss": 46.0,
-      "step": 10607
-    },
-    {
-      "epoch": 1.7083215910463383,
-      "grad_norm": 0.0012496472336351871,
-      "learning_rate": 0.0001999985626262605,
-      "loss": 46.0,
-      "step": 10608
-    },
-    {
-      "epoch": 1.7084826281251257,
-      "grad_norm": 0.0014657059218734503,
-      "learning_rate": 0.0001999985623549946,
-      "loss": 46.0,
-      "step": 10609
-    },
-    {
-      "epoch": 1.7086436652039132,
-      "grad_norm": 0.0004291498626116663,
-      "learning_rate": 0.0001999985620837031,
-      "loss": 46.0,
-      "step": 10610
-    },
-    {
-      "epoch": 1.7088047022827006,
-      "grad_norm": 0.006984542589634657,
-      "learning_rate": 0.00019999856181238598,
-      "loss": 46.0,
-      "step": 10611
-    },
-    {
-      "epoch": 1.708965739361488,
-      "grad_norm": 0.0013984456891193986,
-      "learning_rate": 0.00019999856154104327,
-      "loss": 46.0,
-      "step": 10612
-    },
-    {
-      "epoch": 1.7091267764402753,
-      "grad_norm": 0.0006348855094984174,
-      "learning_rate": 0.00019999856126967497,
-      "loss": 46.0,
-      "step": 10613
-    },
-    {
-      "epoch": 1.7092878135190628,
-      "grad_norm": 0.00048029652680270374,
-      "learning_rate": 0.0001999985609982811,
-      "loss": 46.0,
-      "step": 10614
-    },
-    {
-      "epoch": 1.70944885059785,
-      "grad_norm": 0.001243694918230176,
-      "learning_rate": 0.0001999985607268616,
-      "loss": 46.0,
-      "step": 10615
-    },
-    {
-      "epoch": 1.7096098876766375,
-      "grad_norm": 0.0006153068970888853,
-      "learning_rate": 0.0001999985604554165,
-      "loss": 46.0,
-      "step": 10616
-    },
-    {
-      "epoch": 1.709770924755425,
-      "grad_norm": 0.010551068931818008,
-      "learning_rate": 0.00019999856018394584,
-      "loss": 46.0,
-      "step": 10617
-    },
-    {
-      "epoch": 1.7099319618342124,
-      "grad_norm": 0.003209968563169241,
-      "learning_rate": 0.00019999855991244958,
-      "loss": 46.0,
-      "step": 10618
-    },
-    {
-      "epoch": 1.7100929989129998,
-      "grad_norm": 0.0038249429780989885,
-      "learning_rate": 0.0001999985596409277,
-      "loss": 46.0,
-      "step": 10619
-    },
-    {
-      "epoch": 1.7102540359917873,
-      "grad_norm": 0.0031154875177890062,
-      "learning_rate": 0.00019999855936938025,
-      "loss": 46.0,
-      "step": 10620
-    },
-    {
-      "epoch": 1.7104150730705745,
-      "grad_norm": 0.004832983948290348,
-      "learning_rate": 0.0001999985590978072,
-      "loss": 46.0,
-      "step": 10621
-    },
-    {
-      "epoch": 1.7105761101493617,
-      "grad_norm": 0.0009924547048285604,
-      "learning_rate": 0.00019999855882620854,
-      "loss": 46.0,
-      "step": 10622
-    },
-    {
-      "epoch": 1.7107371472281492,
-      "grad_norm": 0.0009576721931807697,
-      "learning_rate": 0.00019999855855458432,
-      "loss": 46.0,
-      "step": 10623
-    },
-    {
-      "epoch": 1.7108981843069366,
-      "grad_norm": 0.0005204388871788979,
-      "learning_rate": 0.00019999855828293445,
-      "loss": 46.0,
-      "step": 10624
-    },
-    {
-      "epoch": 1.711059221385724,
-      "grad_norm": 0.001394303166307509,
-      "learning_rate": 0.00019999855801125903,
-      "loss": 46.0,
-      "step": 10625
-    },
-    {
-      "epoch": 1.7112202584645115,
-      "grad_norm": 0.004019347485154867,
-      "learning_rate": 0.000199998557739558,
-      "loss": 46.0,
-      "step": 10626
-    },
-    {
-      "epoch": 1.711381295543299,
-      "grad_norm": 0.00600590743124485,
-      "learning_rate": 0.0001999985574678314,
-      "loss": 46.0,
-      "step": 10627
-    },
-    {
-      "epoch": 1.7115423326220862,
-      "grad_norm": 0.004105504136532545,
-      "learning_rate": 0.00019999855719607918,
-      "loss": 46.0,
-      "step": 10628
-    },
-    {
-      "epoch": 1.7117033697008737,
-      "grad_norm": 0.0021854748483747244,
-      "learning_rate": 0.00019999855692430136,
-      "loss": 46.0,
-      "step": 10629
-    },
-    {
-      "epoch": 1.711864406779661,
-      "grad_norm": 0.004823281895369291,
-      "learning_rate": 0.00019999855665249795,
-      "loss": 46.0,
-      "step": 10630
-    },
-    {
-      "epoch": 1.7120254438584483,
-      "grad_norm": 0.003068856429308653,
-      "learning_rate": 0.00019999855638066895,
-      "loss": 46.0,
-      "step": 10631
-    },
-    {
-      "epoch": 1.7121864809372358,
-      "grad_norm": 0.0004929940914735198,
-      "learning_rate": 0.00019999855610881439,
-      "loss": 46.0,
-      "step": 10632
-    },
-    {
-      "epoch": 1.7123475180160233,
-      "grad_norm": 0.0009507111390121281,
-      "learning_rate": 0.00019999855583693418,
-      "loss": 46.0,
-      "step": 10633
-    },
-    {
-      "epoch": 1.7125085550948107,
-      "grad_norm": 0.0011723112547770143,
-      "learning_rate": 0.0001999985555650284,
-      "loss": 46.0,
-      "step": 10634
-    },
-    {
-      "epoch": 1.712669592173598,
-      "grad_norm": 0.0032061573583632708,
-      "learning_rate": 0.00019999855529309702,
-      "loss": 46.0,
-      "step": 10635
-    },
-    {
-      "epoch": 1.7128306292523854,
-      "grad_norm": 0.003046361729502678,
-      "learning_rate": 0.00019999855502114006,
-      "loss": 46.0,
-      "step": 10636
-    },
-    {
-      "epoch": 1.7129916663311726,
-      "grad_norm": 0.00501189474016428,
-      "learning_rate": 0.0001999985547491575,
-      "loss": 46.0,
-      "step": 10637
-    },
-    {
-      "epoch": 1.71315270340996,
-      "grad_norm": 0.004618125967681408,
-      "learning_rate": 0.0001999985544771493,
-      "loss": 46.0,
-      "step": 10638
-    },
-    {
-      "epoch": 1.7133137404887475,
-      "grad_norm": 0.0027390005998313427,
-      "learning_rate": 0.00019999855420511556,
-      "loss": 46.0,
-      "step": 10639
-    },
-    {
-      "epoch": 1.713474777567535,
-      "grad_norm": 0.007732787169516087,
-      "learning_rate": 0.00019999855393305622,
-      "loss": 46.0,
-      "step": 10640
-    },
-    {
-      "epoch": 1.7136358146463224,
-      "grad_norm": 0.002800698159262538,
-      "learning_rate": 0.0001999985536609713,
-      "loss": 46.0,
-      "step": 10641
-    },
-    {
-      "epoch": 1.7137968517251099,
-      "grad_norm": 0.0013721325667575002,
-      "learning_rate": 0.00019999855338886073,
-      "loss": 46.0,
-      "step": 10642
-    },
-    {
-      "epoch": 1.713957888803897,
-      "grad_norm": 0.0018444042652845383,
-      "learning_rate": 0.0001999985531167246,
-      "loss": 46.0,
-      "step": 10643
-    },
-    {
-      "epoch": 1.7141189258826843,
-      "grad_norm": 0.0048058126121759415,
-      "learning_rate": 0.00019999855284456289,
-      "loss": 46.0,
-      "step": 10644
-    },
-    {
-      "epoch": 1.7142799629614718,
-      "grad_norm": 0.0027314776089042425,
-      "learning_rate": 0.00019999855257237556,
-      "loss": 46.0,
-      "step": 10645
-    },
-    {
-      "epoch": 1.7144410000402592,
-      "grad_norm": 0.00786896888166666,
-      "learning_rate": 0.00019999855230016264,
-      "loss": 46.0,
-      "step": 10646
-    },
-    {
-      "epoch": 1.7146020371190467,
-      "grad_norm": 0.005011396948248148,
-      "learning_rate": 0.00019999855202792414,
-      "loss": 46.0,
-      "step": 10647
-    },
-    {
-      "epoch": 1.7147630741978341,
-      "grad_norm": 0.003180519910529256,
-      "learning_rate": 0.00019999855175566002,
-      "loss": 46.0,
-      "step": 10648
-    },
-    {
-      "epoch": 1.7149241112766216,
-      "grad_norm": 0.015887390822172165,
-      "learning_rate": 0.00019999855148337031,
-      "loss": 46.0,
-      "step": 10649
-    },
-    {
-      "epoch": 1.7150851483554088,
-      "grad_norm": 0.002253420650959015,
-      "learning_rate": 0.00019999855121105502,
-      "loss": 46.0,
-      "step": 10650
-    },
-    {
-      "epoch": 1.7152461854341963,
-      "grad_norm": 0.0015703889075666666,
-      "learning_rate": 0.00019999855093871417,
-      "loss": 46.0,
-      "step": 10651
-    },
-    {
-      "epoch": 1.7154072225129835,
-      "grad_norm": 0.002088170498609543,
-      "learning_rate": 0.00019999855066634768,
-      "loss": 46.0,
-      "step": 10652
-    },
-    {
-      "epoch": 1.715568259591771,
-      "grad_norm": 0.004208439029753208,
-      "learning_rate": 0.0001999985503939556,
-      "loss": 46.0,
-      "step": 10653
-    },
-    {
-      "epoch": 1.7157292966705584,
-      "grad_norm": 0.0012495253467932343,
-      "learning_rate": 0.00019999855012153793,
-      "loss": 46.0,
-      "step": 10654
-    },
-    {
-      "epoch": 1.7158903337493459,
-      "grad_norm": 0.014561749994754791,
-      "learning_rate": 0.00019999854984909465,
-      "loss": 46.0,
-      "step": 10655
-    },
-    {
-      "epoch": 1.7160513708281333,
-      "grad_norm": 0.0022735148668289185,
-      "learning_rate": 0.00019999854957662578,
-      "loss": 46.0,
-      "step": 10656
-    },
-    {
-      "epoch": 1.7162124079069205,
-      "grad_norm": 0.0007662269636057317,
-      "learning_rate": 0.00019999854930413132,
-      "loss": 46.0,
-      "step": 10657
-    },
-    {
-      "epoch": 1.716373444985708,
-      "grad_norm": 0.010263331234455109,
-      "learning_rate": 0.00019999854903161128,
-      "loss": 46.0,
-      "step": 10658
-    },
-    {
-      "epoch": 1.7165344820644952,
-      "grad_norm": 0.0019299895502626896,
-      "learning_rate": 0.00019999854875906562,
-      "loss": 46.0,
-      "step": 10659
-    },
-    {
-      "epoch": 1.7166955191432827,
-      "grad_norm": 0.006850752513855696,
-      "learning_rate": 0.0001999985484864944,
-      "loss": 46.0,
-      "step": 10660
-    },
-    {
-      "epoch": 1.7168565562220701,
-      "grad_norm": 0.00429815286770463,
-      "learning_rate": 0.00019999854821389757,
-      "loss": 46.0,
-      "step": 10661
-    },
-    {
-      "epoch": 1.7170175933008576,
-      "grad_norm": 0.003795732744038105,
-      "learning_rate": 0.00019999854794127512,
-      "loss": 46.0,
-      "step": 10662
-    },
-    {
-      "epoch": 1.717178630379645,
-      "grad_norm": 0.0024037258699536324,
-      "learning_rate": 0.00019999854766862712,
-      "loss": 46.0,
-      "step": 10663
-    },
-    {
-      "epoch": 1.7173396674584323,
-      "grad_norm": 0.0025079180486500263,
-      "learning_rate": 0.0001999985473959535,
-      "loss": 46.0,
-      "step": 10664
-    },
-    {
-      "epoch": 1.7175007045372197,
-      "grad_norm": 0.005466458387672901,
-      "learning_rate": 0.00019999854712325429,
-      "loss": 46.0,
-      "step": 10665
-    },
-    {
-      "epoch": 1.717661741616007,
-      "grad_norm": 0.001966609386727214,
-      "learning_rate": 0.0001999985468505295,
-      "loss": 46.0,
-      "step": 10666
-    },
-    {
-      "epoch": 1.7178227786947944,
-      "grad_norm": 0.0025278415996581316,
-      "learning_rate": 0.00019999854657777908,
-      "loss": 46.0,
-      "step": 10667
-    },
-    {
-      "epoch": 1.7179838157735818,
-      "grad_norm": 0.002257103566080332,
-      "learning_rate": 0.00019999854630500308,
-      "loss": 46.0,
-      "step": 10668
-    },
-    {
-      "epoch": 1.7181448528523693,
-      "grad_norm": 0.0006267820135690272,
-      "learning_rate": 0.0001999985460322015,
-      "loss": 46.0,
-      "step": 10669
-    },
-    {
-      "epoch": 1.7183058899311567,
-      "grad_norm": 0.0011026536813005805,
-      "learning_rate": 0.0001999985457593743,
-      "loss": 46.0,
-      "step": 10670
-    },
-    {
-      "epoch": 1.7184669270099442,
-      "grad_norm": 0.0006493310211226344,
-      "learning_rate": 0.00019999854548652152,
-      "loss": 46.0,
-      "step": 10671
-    },
-    {
-      "epoch": 1.7186279640887314,
-      "grad_norm": 0.0011368111008778214,
-      "learning_rate": 0.00019999854521364317,
-      "loss": 46.0,
-      "step": 10672
-    },
-    {
-      "epoch": 1.7187890011675189,
-      "grad_norm": 0.003450807649642229,
-      "learning_rate": 0.00019999854494073919,
-      "loss": 46.0,
-      "step": 10673
-    },
-    {
-      "epoch": 1.718950038246306,
-      "grad_norm": 0.0024938916321843863,
-      "learning_rate": 0.0001999985446678096,
-      "loss": 46.0,
-      "step": 10674
-    },
-    {
-      "epoch": 1.7191110753250936,
-      "grad_norm": 0.0014378870837390423,
-      "learning_rate": 0.00019999854439485448,
-      "loss": 46.0,
-      "step": 10675
-    },
-    {
-      "epoch": 1.719272112403881,
-      "grad_norm": 0.005106337834149599,
-      "learning_rate": 0.0001999985441218737,
-      "loss": 46.0,
-      "step": 10676
-    },
-    {
-      "epoch": 1.7194331494826685,
-      "grad_norm": 0.001010688953101635,
-      "learning_rate": 0.00019999854384886737,
-      "loss": 46.0,
-      "step": 10677
-    },
-    {
-      "epoch": 1.719594186561456,
-      "grad_norm": 0.005435951519757509,
-      "learning_rate": 0.00019999854357583542,
-      "loss": 46.0,
-      "step": 10678
-    },
-    {
-      "epoch": 1.7197552236402431,
-      "grad_norm": 0.0010438169119879603,
-      "learning_rate": 0.0001999985433027779,
-      "loss": 46.0,
-      "step": 10679
-    },
-    {
-      "epoch": 1.7199162607190306,
-      "grad_norm": 0.006695493124425411,
-      "learning_rate": 0.00019999854302969476,
-      "loss": 46.0,
-      "step": 10680
-    },
-    {
-      "epoch": 1.7200772977978178,
-      "grad_norm": 0.0007165994611568749,
-      "learning_rate": 0.00019999854275658602,
-      "loss": 46.0,
-      "step": 10681
-    },
-    {
-      "epoch": 1.7202383348766053,
-      "grad_norm": 0.0049116904847323895,
-      "learning_rate": 0.00019999854248345172,
-      "loss": 46.0,
-      "step": 10682
-    },
-    {
-      "epoch": 1.7203993719553927,
-      "grad_norm": 0.0026281452737748623,
-      "learning_rate": 0.0001999985422102918,
-      "loss": 46.0,
-      "step": 10683
-    },
-    {
-      "epoch": 1.7205604090341802,
-      "grad_norm": 0.0010164750274270773,
-      "learning_rate": 0.0001999985419371063,
-      "loss": 46.0,
-      "step": 10684
-    },
-    {
-      "epoch": 1.7207214461129676,
-      "grad_norm": 0.0014132176293060184,
-      "learning_rate": 0.0001999985416638952,
-      "loss": 46.0,
-      "step": 10685
-    },
-    {
-      "epoch": 1.7208824831917549,
-      "grad_norm": 0.0012437221594154835,
-      "learning_rate": 0.0001999985413906585,
-      "loss": 46.0,
-      "step": 10686
-    },
-    {
-      "epoch": 1.7210435202705423,
-      "grad_norm": 0.0027496113907545805,
-      "learning_rate": 0.00019999854111739623,
-      "loss": 46.0,
-      "step": 10687
-    },
-    {
-      "epoch": 1.7212045573493295,
-      "grad_norm": 0.004544076509773731,
-      "learning_rate": 0.00019999854084410833,
-      "loss": 46.0,
-      "step": 10688
-    },
-    {
-      "epoch": 1.721365594428117,
-      "grad_norm": 0.004530207719653845,
-      "learning_rate": 0.00019999854057079484,
-      "loss": 46.0,
-      "step": 10689
-    },
-    {
-      "epoch": 1.7215266315069044,
-      "grad_norm": 0.0020138320978730917,
-      "learning_rate": 0.00019999854029745576,
-      "loss": 46.0,
-      "step": 10690
-    },
-    {
-      "epoch": 1.7216876685856919,
-      "grad_norm": 0.006559364497661591,
-      "learning_rate": 0.0001999985400240911,
-      "loss": 46.0,
-      "step": 10691
-    },
-    {
-      "epoch": 1.7218487056644793,
-      "grad_norm": 0.0025726009625941515,
-      "learning_rate": 0.00019999853975070085,
-      "loss": 46.0,
-      "step": 10692
-    },
-    {
-      "epoch": 1.7220097427432668,
-      "grad_norm": 0.0027490509673953056,
-      "learning_rate": 0.00019999853947728498,
-      "loss": 46.0,
-      "step": 10693
-    },
-    {
-      "epoch": 1.722170779822054,
-      "grad_norm": 0.002343369647860527,
-      "learning_rate": 0.00019999853920384353,
-      "loss": 46.0,
-      "step": 10694
-    },
-    {
-      "epoch": 1.7223318169008415,
-      "grad_norm": 0.002913028234615922,
-      "learning_rate": 0.0001999985389303765,
-      "loss": 46.0,
-      "step": 10695
-    },
-    {
-      "epoch": 1.7224928539796287,
-      "grad_norm": 0.0008392426534555852,
-      "learning_rate": 0.00019999853865688386,
-      "loss": 46.0,
-      "step": 10696
-    },
-    {
-      "epoch": 1.7226538910584162,
-      "grad_norm": 0.0016698619583621621,
-      "learning_rate": 0.00019999853838336562,
-      "loss": 46.0,
-      "step": 10697
-    },
-    {
-      "epoch": 1.7228149281372036,
-      "grad_norm": 0.0011312097776681185,
-      "learning_rate": 0.00019999853810982176,
-      "loss": 46.0,
-      "step": 10698
-    },
-    {
-      "epoch": 1.722975965215991,
-      "grad_norm": 0.0017419848591089249,
-      "learning_rate": 0.00019999853783625235,
-      "loss": 46.0,
-      "step": 10699
-    },
-    {
-      "epoch": 1.7231370022947785,
-      "grad_norm": 0.001562235876917839,
-      "learning_rate": 0.00019999853756265734,
-      "loss": 46.0,
-      "step": 10700
-    },
-    {
-      "epoch": 1.7232980393735657,
-      "grad_norm": 0.0017675141571089625,
-      "learning_rate": 0.00019999853728903673,
-      "loss": 46.0,
-      "step": 10701
-    },
-    {
-      "epoch": 1.7234590764523532,
-      "grad_norm": 0.006568699609488249,
-      "learning_rate": 0.0001999985370153905,
-      "loss": 46.0,
-      "step": 10702
-    },
-    {
-      "epoch": 1.7236201135311404,
-      "grad_norm": 0.0073175267316401005,
-      "learning_rate": 0.0001999985367417187,
-      "loss": 46.0,
-      "step": 10703
-    },
-    {
-      "epoch": 1.7237811506099279,
-      "grad_norm": 0.0012807713355869055,
-      "learning_rate": 0.00019999853646802132,
-      "loss": 46.0,
-      "step": 10704
-    },
-    {
-      "epoch": 1.7239421876887153,
-      "grad_norm": 0.003323088865727186,
-      "learning_rate": 0.00019999853619429833,
-      "loss": 46.0,
-      "step": 10705
-    },
-    {
-      "epoch": 1.7241032247675028,
-      "grad_norm": 0.0092667480930686,
-      "learning_rate": 0.00019999853592054972,
-      "loss": 46.0,
-      "step": 10706
-    },
-    {
-      "epoch": 1.7242642618462902,
-      "grad_norm": 0.0019082697108387947,
-      "learning_rate": 0.00019999853564677556,
-      "loss": 46.0,
-      "step": 10707
-    },
-    {
-      "epoch": 1.7244252989250775,
-      "grad_norm": 0.0026916007045656443,
-      "learning_rate": 0.00019999853537297577,
-      "loss": 46.0,
-      "step": 10708
-    },
-    {
-      "epoch": 1.724586336003865,
-      "grad_norm": 0.0018295894842594862,
-      "learning_rate": 0.0001999985350991504,
-      "loss": 46.0,
-      "step": 10709
-    },
-    {
-      "epoch": 1.7247473730826521,
-      "grad_norm": 0.0038157927338033915,
-      "learning_rate": 0.00019999853482529948,
-      "loss": 46.0,
-      "step": 10710
-    },
-    {
-      "epoch": 1.7249084101614396,
-      "grad_norm": 0.0014430490555241704,
-      "learning_rate": 0.0001999985345514229,
-      "loss": 46.0,
-      "step": 10711
-    },
-    {
-      "epoch": 1.725069447240227,
-      "grad_norm": 0.0020202582236379385,
-      "learning_rate": 0.00019999853427752075,
-      "loss": 46.0,
-      "step": 10712
-    },
-    {
-      "epoch": 1.7252304843190145,
-      "grad_norm": 0.0012450660578906536,
-      "learning_rate": 0.00019999853400359298,
-      "loss": 46.0,
-      "step": 10713
-    },
-    {
-      "epoch": 1.725391521397802,
-      "grad_norm": 0.00481125945225358,
-      "learning_rate": 0.00019999853372963965,
-      "loss": 46.0,
-      "step": 10714
-    },
-    {
-      "epoch": 1.7255525584765894,
-      "grad_norm": 0.002589769195765257,
-      "learning_rate": 0.0001999985334556607,
-      "loss": 46.0,
-      "step": 10715
-    },
-    {
-      "epoch": 1.7257135955553766,
-      "grad_norm": 0.0039285593666136265,
-      "learning_rate": 0.00019999853318165617,
-      "loss": 46.0,
-      "step": 10716
-    },
-    {
-      "epoch": 1.7258746326341639,
-      "grad_norm": 0.004839100409299135,
-      "learning_rate": 0.00019999853290762607,
-      "loss": 46.0,
-      "step": 10717
-    },
-    {
-      "epoch": 1.7260356697129513,
-      "grad_norm": 0.001381782116368413,
-      "learning_rate": 0.00019999853263357034,
-      "loss": 46.0,
-      "step": 10718
-    },
-    {
-      "epoch": 1.7261967067917388,
-      "grad_norm": 0.0013794592814520001,
-      "learning_rate": 0.00019999853235948902,
-      "loss": 46.0,
-      "step": 10719
-    },
-    {
-      "epoch": 1.7263577438705262,
-      "grad_norm": 0.0008513790089637041,
-      "learning_rate": 0.0001999985320853821,
-      "loss": 46.0,
-      "step": 10720
-    },
-    {
-      "epoch": 1.7265187809493137,
-      "grad_norm": 0.0006822136929258704,
-      "learning_rate": 0.0001999985318112496,
-      "loss": 46.0,
-      "step": 10721
-    },
-    {
-      "epoch": 1.726679818028101,
-      "grad_norm": 0.006813152693212032,
-      "learning_rate": 0.00019999853153709153,
-      "loss": 46.0,
-      "step": 10722
-    },
-    {
-      "epoch": 1.7268408551068883,
-      "grad_norm": 0.0026363907381892204,
-      "learning_rate": 0.00019999853126290783,
-      "loss": 46.0,
-      "step": 10723
-    },
-    {
-      "epoch": 1.7270018921856758,
-      "grad_norm": 0.002446510596200824,
-      "learning_rate": 0.00019999853098869852,
-      "loss": 46.0,
-      "step": 10724
-    },
-    {
-      "epoch": 1.727162929264463,
-      "grad_norm": 0.0021651689894497395,
-      "learning_rate": 0.00019999853071446367,
-      "loss": 46.0,
-      "step": 10725
-    },
-    {
-      "epoch": 1.7273239663432505,
-      "grad_norm": 0.004140754695981741,
-      "learning_rate": 0.0001999985304402032,
-      "loss": 46.0,
-      "step": 10726
-    },
-    {
-      "epoch": 1.727485003422038,
-      "grad_norm": 0.006260409951210022,
-      "learning_rate": 0.0001999985301659171,
-      "loss": 46.0,
-      "step": 10727
-    },
-    {
-      "epoch": 1.7276460405008254,
-      "grad_norm": 0.0015275570331141353,
-      "learning_rate": 0.00019999852989160543,
-      "loss": 46.0,
-      "step": 10728
-    },
-    {
-      "epoch": 1.7278070775796128,
-      "grad_norm": 0.0005883629783056676,
-      "learning_rate": 0.00019999852961726818,
-      "loss": 46.0,
-      "step": 10729
-    },
-    {
-      "epoch": 1.7279681146584,
-      "grad_norm": 0.005475928541272879,
-      "learning_rate": 0.00019999852934290532,
-      "loss": 46.0,
-      "step": 10730
-    },
-    {
-      "epoch": 1.7281291517371875,
-      "grad_norm": 0.0029752531554549932,
-      "learning_rate": 0.0001999985290685169,
-      "loss": 46.0,
-      "step": 10731
-    },
-    {
-      "epoch": 1.7282901888159747,
-      "grad_norm": 0.0016609854064881802,
-      "learning_rate": 0.00019999852879410283,
-      "loss": 46.0,
-      "step": 10732
-    },
-    {
-      "epoch": 1.7284512258947622,
-      "grad_norm": 0.003141415538266301,
-      "learning_rate": 0.00019999852851966318,
-      "loss": 46.0,
-      "step": 10733
-    },
-    {
-      "epoch": 1.7286122629735496,
-      "grad_norm": 0.0017880770610645413,
-      "learning_rate": 0.00019999852824519797,
-      "loss": 46.0,
-      "step": 10734
-    },
-    {
-      "epoch": 1.728773300052337,
-      "grad_norm": 0.0016692924546077847,
-      "learning_rate": 0.00019999852797070714,
-      "loss": 46.0,
-      "step": 10735
-    },
-    {
-      "epoch": 1.7289343371311245,
-      "grad_norm": 0.0047713397070765495,
-      "learning_rate": 0.00019999852769619073,
-      "loss": 46.0,
-      "step": 10736
-    },
-    {
-      "epoch": 1.729095374209912,
-      "grad_norm": 0.007905526086688042,
-      "learning_rate": 0.0001999985274216487,
-      "loss": 46.0,
-      "step": 10737
-    },
-    {
-      "epoch": 1.7292564112886992,
-      "grad_norm": 0.004555199295282364,
-      "learning_rate": 0.0001999985271470811,
-      "loss": 46.0,
-      "step": 10738
-    },
-    {
-      "epoch": 1.7294174483674865,
-      "grad_norm": 0.0021585929207503796,
-      "learning_rate": 0.0001999985268724879,
-      "loss": 46.0,
-      "step": 10739
-    },
-    {
-      "epoch": 1.729578485446274,
-      "grad_norm": 0.0022551454603672028,
-      "learning_rate": 0.00019999852659786908,
-      "loss": 46.0,
-      "step": 10740
-    },
-    {
-      "epoch": 1.7297395225250614,
-      "grad_norm": 0.005573060363531113,
-      "learning_rate": 0.0001999985263232247,
-      "loss": 46.0,
-      "step": 10741
-    },
-    {
-      "epoch": 1.7299005596038488,
-      "grad_norm": 0.001217327662743628,
-      "learning_rate": 0.0001999985260485547,
-      "loss": 46.0,
-      "step": 10742
-    },
-    {
-      "epoch": 1.7300615966826363,
-      "grad_norm": 0.0032988889142870903,
-      "learning_rate": 0.0001999985257738591,
-      "loss": 46.0,
-      "step": 10743
-    },
-    {
-      "epoch": 1.7302226337614237,
-      "grad_norm": 0.0018557668663561344,
-      "learning_rate": 0.00019999852549913794,
-      "loss": 46.0,
-      "step": 10744
-    },
-    {
-      "epoch": 1.730383670840211,
-      "grad_norm": 0.0011874701594933867,
-      "learning_rate": 0.00019999852522439117,
-      "loss": 46.0,
-      "step": 10745
-    },
-    {
-      "epoch": 1.7305447079189984,
-      "grad_norm": 0.003213569289073348,
-      "learning_rate": 0.00019999852494961883,
-      "loss": 46.0,
-      "step": 10746
-    },
-    {
-      "epoch": 1.7307057449977856,
-      "grad_norm": 0.0032414908055216074,
-      "learning_rate": 0.00019999852467482085,
-      "loss": 46.0,
-      "step": 10747
-    },
-    {
-      "epoch": 1.730866782076573,
-      "grad_norm": 0.001104968716390431,
-      "learning_rate": 0.00019999852439999728,
-      "loss": 46.0,
-      "step": 10748
-    },
-    {
-      "epoch": 1.7310278191553605,
-      "grad_norm": 0.0014070621691644192,
-      "learning_rate": 0.00019999852412514815,
-      "loss": 46.0,
-      "step": 10749
-    },
-    {
-      "epoch": 1.731188856234148,
-      "grad_norm": 0.0006176167516969144,
-      "learning_rate": 0.0001999985238502734,
-      "loss": 46.0,
-      "step": 10750
-    },
-    {
-      "epoch": 1.7313498933129354,
-      "grad_norm": 0.002404885832220316,
-      "learning_rate": 0.00019999852357537306,
-      "loss": 46.0,
-      "step": 10751
-    },
-    {
-      "epoch": 1.7315109303917227,
-      "grad_norm": 0.0027448267210274935,
-      "learning_rate": 0.00019999852330044714,
-      "loss": 46.0,
-      "step": 10752
-    },
-    {
-      "epoch": 1.73167196747051,
-      "grad_norm": 0.0029412193689495325,
-      "learning_rate": 0.0001999985230254956,
-      "loss": 46.0,
-      "step": 10753
-    },
-    {
-      "epoch": 1.7318330045492973,
-      "grad_norm": 0.006089497357606888,
-      "learning_rate": 0.0001999985227505185,
-      "loss": 46.0,
-      "step": 10754
-    },
-    {
-      "epoch": 1.7319940416280848,
-      "grad_norm": 0.0033259836491197348,
-      "learning_rate": 0.0001999985224755158,
-      "loss": 46.0,
-      "step": 10755
-    },
-    {
-      "epoch": 1.7321550787068722,
-      "grad_norm": 0.004637986421585083,
-      "learning_rate": 0.00019999852220048747,
-      "loss": 46.0,
-      "step": 10756
-    },
-    {
-      "epoch": 1.7323161157856597,
-      "grad_norm": 0.0025347708724439144,
-      "learning_rate": 0.00019999852192543357,
-      "loss": 46.0,
-      "step": 10757
-    },
-    {
-      "epoch": 1.7324771528644471,
-      "grad_norm": 0.0022497184108942747,
-      "learning_rate": 0.00019999852165035408,
-      "loss": 46.0,
-      "step": 10758
-    },
-    {
-      "epoch": 1.7326381899432344,
-      "grad_norm": 0.00305104348808527,
-      "learning_rate": 0.00019999852137524897,
-      "loss": 46.0,
-      "step": 10759
-    },
-    {
-      "epoch": 1.7327992270220218,
-      "grad_norm": 0.0047032530419528484,
-      "learning_rate": 0.00019999852110011827,
-      "loss": 46.0,
-      "step": 10760
-    },
-    {
-      "epoch": 1.732960264100809,
-      "grad_norm": 0.003722802270203829,
-      "learning_rate": 0.000199998520824962,
-      "loss": 46.0,
-      "step": 10761
-    },
-    {
-      "epoch": 1.7331213011795965,
-      "grad_norm": 0.006887431256473064,
-      "learning_rate": 0.00019999852054978013,
-      "loss": 46.0,
-      "step": 10762
-    },
-    {
-      "epoch": 1.733282338258384,
-      "grad_norm": 0.0011110524646937847,
-      "learning_rate": 0.00019999852027457264,
-      "loss": 46.0,
-      "step": 10763
-    },
-    {
-      "epoch": 1.7334433753371714,
-      "grad_norm": 0.0016281026182696223,
-      "learning_rate": 0.00019999851999933957,
-      "loss": 46.0,
-      "step": 10764
-    },
-    {
-      "epoch": 1.7336044124159589,
-      "grad_norm": 0.0025362223386764526,
-      "learning_rate": 0.00019999851972408092,
-      "loss": 46.0,
-      "step": 10765
-    },
-    {
-      "epoch": 1.7337654494947463,
-      "grad_norm": 0.0022491959389299154,
-      "learning_rate": 0.00019999851944879664,
-      "loss": 46.0,
-      "step": 10766
-    },
-    {
-      "epoch": 1.7339264865735335,
-      "grad_norm": 0.001708249794319272,
-      "learning_rate": 0.0001999985191734868,
-      "loss": 46.0,
-      "step": 10767
-    },
-    {
-      "epoch": 1.734087523652321,
-      "grad_norm": 0.0020848913118243217,
-      "learning_rate": 0.00019999851889815137,
-      "loss": 46.0,
-      "step": 10768
-    },
-    {
-      "epoch": 1.7342485607311082,
-      "grad_norm": 0.01726636476814747,
-      "learning_rate": 0.0001999985186227903,
-      "loss": 46.0,
-      "step": 10769
-    },
-    {
-      "epoch": 1.7344095978098957,
-      "grad_norm": 0.0009378970135003328,
-      "learning_rate": 0.0001999985183474037,
-      "loss": 46.0,
-      "step": 10770
-    },
-    {
-      "epoch": 1.7345706348886831,
-      "grad_norm": 0.0011219424195587635,
-      "learning_rate": 0.00019999851807199145,
-      "loss": 46.0,
-      "step": 10771
-    },
-    {
-      "epoch": 1.7347316719674706,
-      "grad_norm": 0.0044549452140927315,
-      "learning_rate": 0.00019999851779655363,
-      "loss": 46.0,
-      "step": 10772
-    },
-    {
-      "epoch": 1.734892709046258,
-      "grad_norm": 0.010219248943030834,
-      "learning_rate": 0.00019999851752109022,
-      "loss": 46.0,
-      "step": 10773
-    },
-    {
-      "epoch": 1.7350537461250453,
-      "grad_norm": 0.0013460562331601977,
-      "learning_rate": 0.0001999985172456012,
-      "loss": 46.0,
-      "step": 10774
-    },
-    {
-      "epoch": 1.7352147832038327,
-      "grad_norm": 0.0018596885493025184,
-      "learning_rate": 0.0001999985169700866,
-      "loss": 46.0,
-      "step": 10775
-    },
-    {
-      "epoch": 1.73537582028262,
-      "grad_norm": 0.0012383024441078305,
-      "learning_rate": 0.0001999985166945464,
-      "loss": 46.0,
-      "step": 10776
-    },
-    {
-      "epoch": 1.7355368573614074,
-      "grad_norm": 0.005149056203663349,
-      "learning_rate": 0.0001999985164189806,
-      "loss": 46.0,
-      "step": 10777
-    },
-    {
-      "epoch": 1.7356978944401948,
-      "grad_norm": 0.0019533196464180946,
-      "learning_rate": 0.00019999851614338924,
-      "loss": 46.0,
-      "step": 10778
-    },
-    {
-      "epoch": 1.7358589315189823,
-      "grad_norm": 0.0037077427841722965,
-      "learning_rate": 0.00019999851586777222,
-      "loss": 46.0,
-      "step": 10779
-    },
-    {
-      "epoch": 1.7360199685977697,
-      "grad_norm": 0.0005321336211636662,
-      "learning_rate": 0.00019999851559212965,
-      "loss": 46.0,
-      "step": 10780
-    },
-    {
-      "epoch": 1.736181005676557,
-      "grad_norm": 0.001478381222113967,
-      "learning_rate": 0.0001999985153164615,
-      "loss": 46.0,
-      "step": 10781
-    },
-    {
-      "epoch": 1.7363420427553444,
-      "grad_norm": 0.0036501639988273382,
-      "learning_rate": 0.00019999851504076772,
-      "loss": 46.0,
-      "step": 10782
-    },
-    {
-      "epoch": 1.7365030798341317,
-      "grad_norm": 0.00226794695481658,
-      "learning_rate": 0.00019999851476504836,
-      "loss": 46.0,
-      "step": 10783
-    },
-    {
-      "epoch": 1.736664116912919,
-      "grad_norm": 0.0012095397105440497,
-      "learning_rate": 0.0001999985144893034,
-      "loss": 46.0,
-      "step": 10784
-    },
-    {
-      "epoch": 1.7368251539917066,
-      "grad_norm": 0.0053879451006650925,
-      "learning_rate": 0.00019999851421353287,
-      "loss": 46.0,
-      "step": 10785
-    },
-    {
-      "epoch": 1.736986191070494,
-      "grad_norm": 0.0015686068218201399,
-      "learning_rate": 0.0001999985139377367,
-      "loss": 46.0,
-      "step": 10786
-    },
-    {
-      "epoch": 1.7371472281492815,
-      "grad_norm": 0.00041315285488963127,
-      "learning_rate": 0.00019999851366191496,
-      "loss": 46.0,
-      "step": 10787
-    },
-    {
-      "epoch": 1.737308265228069,
-      "grad_norm": 0.0007580140954814851,
-      "learning_rate": 0.00019999851338606763,
-      "loss": 46.0,
-      "step": 10788
-    },
-    {
-      "epoch": 1.7374693023068561,
-      "grad_norm": 0.004159773234277964,
-      "learning_rate": 0.0001999985131101947,
-      "loss": 46.0,
-      "step": 10789
-    },
-    {
-      "epoch": 1.7376303393856436,
-      "grad_norm": 0.00813358649611473,
-      "learning_rate": 0.0001999985128342962,
-      "loss": 46.0,
-      "step": 10790
-    },
-    {
-      "epoch": 1.7377913764644308,
-      "grad_norm": 0.0037237240467220545,
-      "learning_rate": 0.00019999851255837205,
-      "loss": 46.0,
-      "step": 10791
-    },
-    {
-      "epoch": 1.7379524135432183,
-      "grad_norm": 0.0012467398773878813,
-      "learning_rate": 0.00019999851228242235,
-      "loss": 46.0,
-      "step": 10792
-    },
-    {
-      "epoch": 1.7381134506220057,
-      "grad_norm": 0.01693449728190899,
-      "learning_rate": 0.00019999851200644704,
-      "loss": 46.0,
-      "step": 10793
-    },
-    {
-      "epoch": 1.7382744877007932,
-      "grad_norm": 0.0006624520756304264,
-      "learning_rate": 0.00019999851173044614,
-      "loss": 46.0,
-      "step": 10794
-    },
-    {
-      "epoch": 1.7384355247795806,
-      "grad_norm": 0.0037024826742708683,
-      "learning_rate": 0.00019999851145441965,
-      "loss": 46.0,
-      "step": 10795
-    },
-    {
-      "epoch": 1.7385965618583679,
-      "grad_norm": 0.0020319244358688593,
-      "learning_rate": 0.00019999851117836754,
-      "loss": 46.0,
-      "step": 10796
-    },
-    {
-      "epoch": 1.7387575989371553,
-      "grad_norm": 0.0014004985569044948,
-      "learning_rate": 0.00019999851090228988,
-      "loss": 46.0,
-      "step": 10797
-    },
-    {
-      "epoch": 1.7389186360159425,
-      "grad_norm": 0.006149820983409882,
-      "learning_rate": 0.00019999851062618658,
-      "loss": 46.0,
-      "step": 10798
-    },
-    {
-      "epoch": 1.73907967309473,
-      "grad_norm": 0.001070016180165112,
-      "learning_rate": 0.0001999985103500577,
-      "loss": 46.0,
-      "step": 10799
-    },
-    {
-      "epoch": 1.7392407101735174,
-      "grad_norm": 0.0007530535804107785,
-      "learning_rate": 0.00019999851007390323,
-      "loss": 46.0,
-      "step": 10800
-    },
-    {
-      "epoch": 1.739401747252305,
-      "grad_norm": 0.003215379547327757,
-      "learning_rate": 0.0001999985097977232,
-      "loss": 46.0,
-      "step": 10801
-    },
-    {
-      "epoch": 1.7395627843310923,
-      "grad_norm": 0.004696669988334179,
-      "learning_rate": 0.00019999850952151754,
-      "loss": 46.0,
-      "step": 10802
-    },
-    {
-      "epoch": 1.7397238214098796,
-      "grad_norm": 0.004864111077040434,
-      "learning_rate": 0.00019999850924528627,
-      "loss": 46.0,
-      "step": 10803
-    },
-    {
-      "epoch": 1.739884858488667,
-      "grad_norm": 0.00274386047385633,
-      "learning_rate": 0.00019999850896902942,
-      "loss": 46.0,
-      "step": 10804
-    },
-    {
-      "epoch": 1.7400458955674543,
-      "grad_norm": 0.0027007407043129206,
-      "learning_rate": 0.000199998508692747,
-      "loss": 46.0,
-      "step": 10805
-    },
-    {
-      "epoch": 1.7402069326462417,
-      "grad_norm": 0.0010712225921452045,
-      "learning_rate": 0.00019999850841643895,
-      "loss": 46.0,
-      "step": 10806
-    },
-    {
-      "epoch": 1.7403679697250292,
-      "grad_norm": 0.001988667994737625,
-      "learning_rate": 0.00019999850814010533,
-      "loss": 46.0,
-      "step": 10807
-    },
-    {
-      "epoch": 1.7405290068038166,
-      "grad_norm": 0.0022766145411878824,
-      "learning_rate": 0.0001999985078637461,
-      "loss": 46.0,
-      "step": 10808
-    },
-    {
-      "epoch": 1.740690043882604,
-      "grad_norm": 0.0021920788567513227,
-      "learning_rate": 0.0001999985075873613,
-      "loss": 46.0,
-      "step": 10809
-    },
-    {
-      "epoch": 1.7408510809613915,
-      "grad_norm": 0.006652799900621176,
-      "learning_rate": 0.00019999850731095088,
-      "loss": 46.0,
-      "step": 10810
-    },
-    {
-      "epoch": 1.7410121180401787,
-      "grad_norm": 0.002119033597409725,
-      "learning_rate": 0.00019999850703451486,
-      "loss": 46.0,
-      "step": 10811
-    },
-    {
-      "epoch": 1.741173155118966,
-      "grad_norm": 0.0011126274475827813,
-      "learning_rate": 0.00019999850675805325,
-      "loss": 46.0,
-      "step": 10812
-    },
-    {
-      "epoch": 1.7413341921977534,
-      "grad_norm": 0.0020380797795951366,
-      "learning_rate": 0.00019999850648156606,
-      "loss": 46.0,
-      "step": 10813
-    },
-    {
-      "epoch": 1.7414952292765409,
-      "grad_norm": 0.006282332818955183,
-      "learning_rate": 0.00019999850620505328,
-      "loss": 46.0,
-      "step": 10814
-    },
-    {
-      "epoch": 1.7416562663553283,
-      "grad_norm": 0.002152771456167102,
-      "learning_rate": 0.0001999985059285149,
-      "loss": 46.0,
-      "step": 10815
-    },
-    {
-      "epoch": 1.7418173034341158,
-      "grad_norm": 0.006459617521613836,
-      "learning_rate": 0.0001999985056519509,
-      "loss": 46.0,
-      "step": 10816
-    },
-    {
-      "epoch": 1.7419783405129032,
-      "grad_norm": 0.002866796450689435,
-      "learning_rate": 0.00019999850537536133,
-      "loss": 46.0,
-      "step": 10817
-    },
-    {
-      "epoch": 1.7421393775916905,
-      "grad_norm": 0.005442943423986435,
-      "learning_rate": 0.00019999850509874615,
-      "loss": 46.0,
-      "step": 10818
-    },
-    {
-      "epoch": 1.742300414670478,
-      "grad_norm": 0.001381971058435738,
-      "learning_rate": 0.0001999985048221054,
-      "loss": 46.0,
-      "step": 10819
-    },
-    {
-      "epoch": 1.7424614517492651,
-      "grad_norm": 0.002078956924378872,
-      "learning_rate": 0.00019999850454543902,
-      "loss": 46.0,
-      "step": 10820
-    },
-    {
-      "epoch": 1.7426224888280526,
-      "grad_norm": 0.0022242553532123566,
-      "learning_rate": 0.0001999985042687471,
-      "loss": 46.0,
-      "step": 10821
-    },
-    {
-      "epoch": 1.74278352590684,
-      "grad_norm": 0.0018119171727448702,
-      "learning_rate": 0.00019999850399202954,
-      "loss": 46.0,
-      "step": 10822
-    },
-    {
-      "epoch": 1.7429445629856275,
-      "grad_norm": 0.0018037857953459024,
-      "learning_rate": 0.00019999850371528642,
-      "loss": 46.0,
-      "step": 10823
-    },
-    {
-      "epoch": 1.743105600064415,
-      "grad_norm": 0.002616539131850004,
-      "learning_rate": 0.00019999850343851766,
-      "loss": 46.0,
-      "step": 10824
-    },
-    {
-      "epoch": 1.7432666371432022,
-      "grad_norm": 0.0026036249473690987,
-      "learning_rate": 0.00019999850316172334,
-      "loss": 46.0,
-      "step": 10825
-    },
-    {
-      "epoch": 1.7434276742219896,
-      "grad_norm": 0.004549722652882338,
-      "learning_rate": 0.0001999985028849034,
-      "loss": 46.0,
-      "step": 10826
-    },
-    {
-      "epoch": 1.7435887113007769,
-      "grad_norm": 0.007684149779379368,
-      "learning_rate": 0.00019999850260805788,
-      "loss": 46.0,
-      "step": 10827
-    },
-    {
-      "epoch": 1.7437497483795643,
-      "grad_norm": 0.0133134126663208,
-      "learning_rate": 0.00019999850233118677,
-      "loss": 46.0,
-      "step": 10828
-    },
-    {
-      "epoch": 1.7439107854583518,
-      "grad_norm": 0.0036709101404994726,
-      "learning_rate": 0.00019999850205429007,
-      "loss": 46.0,
-      "step": 10829
-    },
-    {
-      "epoch": 1.7440718225371392,
-      "grad_norm": 0.001921869465149939,
-      "learning_rate": 0.00019999850177736776,
-      "loss": 46.0,
-      "step": 10830
-    },
-    {
-      "epoch": 1.7442328596159267,
-      "grad_norm": 0.001875140005722642,
-      "learning_rate": 0.00019999850150041986,
-      "loss": 46.0,
-      "step": 10831
-    },
-    {
-      "epoch": 1.7443938966947141,
-      "grad_norm": 0.002384265186265111,
-      "learning_rate": 0.00019999850122344638,
-      "loss": 46.0,
-      "step": 10832
-    },
-    {
-      "epoch": 1.7445549337735013,
-      "grad_norm": 0.008249553851783276,
-      "learning_rate": 0.0001999985009464473,
-      "loss": 46.0,
-      "step": 10833
-    },
-    {
-      "epoch": 1.7447159708522886,
-      "grad_norm": 0.00331481103785336,
-      "learning_rate": 0.0001999985006694226,
-      "loss": 46.0,
-      "step": 10834
-    },
-    {
-      "epoch": 1.744877007931076,
-      "grad_norm": 0.0043636648915708065,
-      "learning_rate": 0.00019999850039237235,
-      "loss": 46.0,
-      "step": 10835
-    },
-    {
-      "epoch": 1.7450380450098635,
-      "grad_norm": 0.0020867707207798958,
-      "learning_rate": 0.00019999850011529646,
-      "loss": 46.0,
-      "step": 10836
-    },
-    {
-      "epoch": 1.745199082088651,
-      "grad_norm": 0.008378506638109684,
-      "learning_rate": 0.000199998499838195,
-      "loss": 46.0,
-      "step": 10837
-    },
-    {
-      "epoch": 1.7453601191674384,
-      "grad_norm": 0.0014831061707809567,
-      "learning_rate": 0.00019999849956106792,
-      "loss": 46.0,
-      "step": 10838
-    },
-    {
-      "epoch": 1.7455211562462258,
-      "grad_norm": 0.000990537810139358,
-      "learning_rate": 0.00019999849928391527,
-      "loss": 46.0,
-      "step": 10839
-    },
-    {
-      "epoch": 1.745682193325013,
-      "grad_norm": 0.006173607427626848,
-      "learning_rate": 0.00019999849900673703,
-      "loss": 46.0,
-      "step": 10840
-    },
-    {
-      "epoch": 1.7458432304038005,
-      "grad_norm": 0.004493341781198978,
-      "learning_rate": 0.0001999984987295332,
-      "loss": 46.0,
-      "step": 10841
-    },
-    {
-      "epoch": 1.7460042674825877,
-      "grad_norm": 0.003403579583391547,
-      "learning_rate": 0.00019999849845230377,
-      "loss": 46.0,
-      "step": 10842
-    },
-    {
-      "epoch": 1.7461653045613752,
-      "grad_norm": 0.004680973943322897,
-      "learning_rate": 0.00019999849817504872,
-      "loss": 46.0,
-      "step": 10843
-    },
-    {
-      "epoch": 1.7463263416401626,
-      "grad_norm": 0.002826865529641509,
-      "learning_rate": 0.0001999984978977681,
-      "loss": 46.0,
-      "step": 10844
-    },
-    {
-      "epoch": 1.74648737871895,
-      "grad_norm": 0.0021834252402186394,
-      "learning_rate": 0.0001999984976204619,
-      "loss": 46.0,
-      "step": 10845
-    },
-    {
-      "epoch": 1.7466484157977376,
-      "grad_norm": 0.0014757632743567228,
-      "learning_rate": 0.00019999849734313006,
-      "loss": 46.0,
-      "step": 10846
-    },
-    {
-      "epoch": 1.7468094528765248,
-      "grad_norm": 0.002079938305541873,
-      "learning_rate": 0.00019999849706577266,
-      "loss": 46.0,
-      "step": 10847
-    },
-    {
-      "epoch": 1.7469704899553122,
-      "grad_norm": 0.0010253607761114836,
-      "learning_rate": 0.00019999849678838967,
-      "loss": 46.0,
-      "step": 10848
-    },
-    {
-      "epoch": 1.7471315270340995,
-      "grad_norm": 0.005101647228002548,
-      "learning_rate": 0.00019999849651098107,
-      "loss": 46.0,
-      "step": 10849
-    },
-    {
-      "epoch": 1.747292564112887,
-      "grad_norm": 0.0032603980507701635,
-      "learning_rate": 0.00019999849623354688,
-      "loss": 46.0,
-      "step": 10850
-    },
-    {
-      "epoch": 1.7474536011916744,
-      "grad_norm": 0.013747045770287514,
-      "learning_rate": 0.00019999849595608708,
-      "loss": 46.0,
-      "step": 10851
-    },
-    {
-      "epoch": 1.7476146382704618,
-      "grad_norm": 0.00044209155021235347,
-      "learning_rate": 0.00019999849567860168,
-      "loss": 46.0,
-      "step": 10852
-    },
-    {
-      "epoch": 1.7477756753492493,
-      "grad_norm": 0.0018534163245931268,
-      "learning_rate": 0.00019999849540109073,
-      "loss": 46.0,
-      "step": 10853
-    },
-    {
-      "epoch": 1.7479367124280367,
-      "grad_norm": 0.004616633988916874,
-      "learning_rate": 0.00019999849512355414,
-      "loss": 46.0,
-      "step": 10854
-    },
-    {
-      "epoch": 1.748097749506824,
-      "grad_norm": 0.0012869354104623199,
-      "learning_rate": 0.00019999849484599201,
-      "loss": 46.0,
-      "step": 10855
-    },
-    {
-      "epoch": 1.7482587865856112,
-      "grad_norm": 0.0005482708802446723,
-      "learning_rate": 0.00019999849456840422,
-      "loss": 46.0,
-      "step": 10856
-    },
-    {
-      "epoch": 1.7484198236643986,
-      "grad_norm": 0.0009960244642570615,
-      "learning_rate": 0.00019999849429079087,
-      "loss": 46.0,
-      "step": 10857
-    },
-    {
-      "epoch": 1.748580860743186,
-      "grad_norm": 0.0022940030321478844,
-      "learning_rate": 0.00019999849401315195,
-      "loss": 46.0,
-      "step": 10858
-    },
-    {
-      "epoch": 1.7487418978219735,
-      "grad_norm": 0.0032357554882764816,
-      "learning_rate": 0.00019999849373548737,
-      "loss": 46.0,
-      "step": 10859
-    },
-    {
-      "epoch": 1.748902934900761,
-      "grad_norm": 0.0041608610190451145,
-      "learning_rate": 0.00019999849345779725,
-      "loss": 46.0,
-      "step": 10860
-    },
-    {
-      "epoch": 1.7490639719795484,
-      "grad_norm": 0.0016584380064159632,
-      "learning_rate": 0.00019999849318008152,
-      "loss": 46.0,
-      "step": 10861
-    },
-    {
-      "epoch": 1.7492250090583357,
-      "grad_norm": 0.0013155200285837054,
-      "learning_rate": 0.00019999849290234018,
-      "loss": 46.0,
-      "step": 10862
-    },
-    {
-      "epoch": 1.7493860461371231,
-      "grad_norm": 0.0018878270639106631,
-      "learning_rate": 0.00019999849262457327,
-      "loss": 46.0,
-      "step": 10863
-    },
-    {
-      "epoch": 1.7495470832159103,
-      "grad_norm": 0.0009368465980514884,
-      "learning_rate": 0.00019999849234678075,
-      "loss": 46.0,
-      "step": 10864
-    },
-    {
-      "epoch": 1.7497081202946978,
-      "grad_norm": 0.0011551823699846864,
-      "learning_rate": 0.00019999849206896265,
-      "loss": 46.0,
-      "step": 10865
-    },
-    {
-      "epoch": 1.7498691573734853,
-      "grad_norm": 0.0009309992310591042,
-      "learning_rate": 0.00019999849179111893,
-      "loss": 46.0,
-      "step": 10866
-    },
-    {
-      "epoch": 1.7500301944522727,
-      "grad_norm": 0.0018212506547570229,
-      "learning_rate": 0.00019999849151324965,
-      "loss": 46.0,
-      "step": 10867
-    },
-    {
-      "epoch": 1.7501912315310602,
-      "grad_norm": 0.004179096780717373,
-      "learning_rate": 0.00019999849123535475,
-      "loss": 46.0,
-      "step": 10868
-    },
-    {
-      "epoch": 1.7503522686098474,
-      "grad_norm": 0.003813309594988823,
-      "learning_rate": 0.00019999849095743427,
-      "loss": 46.0,
-      "step": 10869
-    },
-    {
-      "epoch": 1.7505133056886348,
-      "grad_norm": 0.004842016380280256,
-      "learning_rate": 0.00019999849067948818,
-      "loss": 46.0,
-      "step": 10870
-    },
-    {
-      "epoch": 1.750674342767422,
-      "grad_norm": 0.0022694820072501898,
-      "learning_rate": 0.0001999984904015165,
-      "loss": 46.0,
-      "step": 10871
-    },
-    {
-      "epoch": 1.7508353798462095,
-      "grad_norm": 0.00515178544446826,
-      "learning_rate": 0.00019999849012351922,
-      "loss": 46.0,
-      "step": 10872
-    },
-    {
-      "epoch": 1.750996416924997,
-      "grad_norm": 0.001147416653111577,
-      "learning_rate": 0.00019999848984549636,
-      "loss": 46.0,
-      "step": 10873
-    },
-    {
-      "epoch": 1.7511574540037844,
-      "grad_norm": 0.0023850752040743828,
-      "learning_rate": 0.00019999848956744792,
-      "loss": 46.0,
-      "step": 10874
-    },
-    {
-      "epoch": 1.7513184910825719,
-      "grad_norm": 0.002134798327460885,
-      "learning_rate": 0.00019999848928937386,
-      "loss": 46.0,
-      "step": 10875
-    },
-    {
-      "epoch": 1.751479528161359,
-      "grad_norm": 0.001114926184527576,
-      "learning_rate": 0.00019999848901127421,
-      "loss": 46.0,
-      "step": 10876
-    },
-    {
-      "epoch": 1.7516405652401466,
-      "grad_norm": 0.0016794002149254084,
-      "learning_rate": 0.00019999848873314898,
-      "loss": 46.0,
-      "step": 10877
-    },
-    {
-      "epoch": 1.7518016023189338,
-      "grad_norm": 0.0038413633592426777,
-      "learning_rate": 0.0001999984884549981,
-      "loss": 46.0,
-      "step": 10878
-    },
-    {
-      "epoch": 1.7519626393977212,
-      "grad_norm": 0.0028941063210368156,
-      "learning_rate": 0.00019999848817682167,
-      "loss": 46.0,
-      "step": 10879
-    },
-    {
-      "epoch": 1.7521236764765087,
-      "grad_norm": 0.0024616015143692493,
-      "learning_rate": 0.00019999848789861965,
-      "loss": 46.0,
-      "step": 10880
-    },
-    {
-      "epoch": 1.7522847135552961,
-      "grad_norm": 0.0035471355076879263,
-      "learning_rate": 0.00019999848762039204,
-      "loss": 46.0,
-      "step": 10881
-    },
-    {
-      "epoch": 1.7524457506340836,
-      "grad_norm": 0.0022235140204429626,
-      "learning_rate": 0.00019999848734213882,
-      "loss": 46.0,
-      "step": 10882
-    },
-    {
-      "epoch": 1.752606787712871,
-      "grad_norm": 0.0005679349415004253,
-      "learning_rate": 0.00019999848706386,
-      "loss": 46.0,
-      "step": 10883
-    },
-    {
-      "epoch": 1.7527678247916583,
-      "grad_norm": 0.0012403653236106038,
-      "learning_rate": 0.00019999848678555558,
-      "loss": 46.0,
-      "step": 10884
-    },
-    {
-      "epoch": 1.7529288618704457,
-      "grad_norm": 0.005488630384206772,
-      "learning_rate": 0.00019999848650722562,
-      "loss": 46.0,
-      "step": 10885
-    },
-    {
-      "epoch": 1.753089898949233,
-      "grad_norm": 0.006263222545385361,
-      "learning_rate": 0.00019999848622887,
-      "loss": 46.0,
-      "step": 10886
-    },
-    {
-      "epoch": 1.7532509360280204,
-      "grad_norm": 0.0037260418757796288,
-      "learning_rate": 0.0001999984859504888,
-      "loss": 46.0,
-      "step": 10887
-    },
-    {
-      "epoch": 1.7534119731068079,
-      "grad_norm": 0.0018899147398769855,
-      "learning_rate": 0.00019999848567208204,
-      "loss": 46.0,
-      "step": 10888
-    },
-    {
-      "epoch": 1.7535730101855953,
-      "grad_norm": 0.006363023538142443,
-      "learning_rate": 0.00019999848539364968,
-      "loss": 46.0,
-      "step": 10889
-    },
-    {
-      "epoch": 1.7537340472643828,
-      "grad_norm": 0.0022546444088220596,
-      "learning_rate": 0.0001999984851151917,
-      "loss": 46.0,
-      "step": 10890
-    },
-    {
-      "epoch": 1.75389508434317,
-      "grad_norm": 0.003685698611661792,
-      "learning_rate": 0.0001999984848367081,
-      "loss": 46.0,
-      "step": 10891
-    },
-    {
-      "epoch": 1.7540561214219574,
-      "grad_norm": 0.0036330418661236763,
-      "learning_rate": 0.00019999848455819896,
-      "loss": 46.0,
-      "step": 10892
-    },
-    {
-      "epoch": 1.7542171585007447,
-      "grad_norm": 0.007977779023349285,
-      "learning_rate": 0.0001999984842796642,
-      "loss": 46.0,
-      "step": 10893
-    },
-    {
-      "epoch": 1.7543781955795321,
-      "grad_norm": 0.004041213542222977,
-      "learning_rate": 0.00019999848400110385,
-      "loss": 46.0,
-      "step": 10894
-    },
-    {
-      "epoch": 1.7545392326583196,
-      "grad_norm": 0.006915631704032421,
-      "learning_rate": 0.0001999984837225179,
-      "loss": 46.0,
-      "step": 10895
-    },
-    {
-      "epoch": 1.754700269737107,
-      "grad_norm": 0.002794482745230198,
-      "learning_rate": 0.00019999848344390636,
-      "loss": 46.0,
-      "step": 10896
-    },
-    {
-      "epoch": 1.7548613068158945,
-      "grad_norm": 0.0004204817523714155,
-      "learning_rate": 0.00019999848316526925,
-      "loss": 46.0,
-      "step": 10897
-    },
-    {
-      "epoch": 1.7550223438946817,
-      "grad_norm": 0.0068509881384670734,
-      "learning_rate": 0.00019999848288660652,
-      "loss": 46.0,
-      "step": 10898
-    },
-    {
-      "epoch": 1.7551833809734692,
-      "grad_norm": 0.003944136202335358,
-      "learning_rate": 0.0001999984826079182,
-      "loss": 46.0,
-      "step": 10899
-    },
-    {
-      "epoch": 1.7553444180522564,
-      "grad_norm": 0.0008018723456189036,
-      "learning_rate": 0.00019999848232920425,
-      "loss": 46.0,
-      "step": 10900
-    },
-    {
-      "epoch": 1.7555054551310438,
-      "grad_norm": 0.0015046069165691733,
-      "learning_rate": 0.00019999848205046476,
-      "loss": 46.0,
-      "step": 10901
-    },
-    {
-      "epoch": 1.7556664922098313,
-      "grad_norm": 0.00173277803696692,
-      "learning_rate": 0.00019999848177169966,
-      "loss": 46.0,
-      "step": 10902
-    },
-    {
-      "epoch": 1.7558275292886187,
-      "grad_norm": 0.002506011165678501,
-      "learning_rate": 0.00019999848149290897,
-      "loss": 46.0,
-      "step": 10903
-    },
-    {
-      "epoch": 1.7559885663674062,
-      "grad_norm": 0.0020869409199804068,
-      "learning_rate": 0.00019999848121409267,
-      "loss": 46.0,
-      "step": 10904
-    },
-    {
-      "epoch": 1.7561496034461936,
-      "grad_norm": 0.001737023238092661,
-      "learning_rate": 0.00019999848093525078,
-      "loss": 46.0,
-      "step": 10905
-    },
-    {
-      "epoch": 1.7563106405249809,
-      "grad_norm": 0.002367757260799408,
-      "learning_rate": 0.0001999984806563833,
-      "loss": 46.0,
-      "step": 10906
-    },
-    {
-      "epoch": 1.7564716776037683,
-      "grad_norm": 0.0016208909219130874,
-      "learning_rate": 0.00019999848037749023,
-      "loss": 46.0,
-      "step": 10907
-    },
-    {
-      "epoch": 1.7566327146825556,
-      "grad_norm": 0.0015691753942519426,
-      "learning_rate": 0.00019999848009857153,
-      "loss": 46.0,
-      "step": 10908
-    },
-    {
-      "epoch": 1.756793751761343,
-      "grad_norm": 0.007605371531099081,
-      "learning_rate": 0.00019999847981962726,
-      "loss": 46.0,
-      "step": 10909
-    },
-    {
-      "epoch": 1.7569547888401305,
-      "grad_norm": 0.0015439819544553757,
-      "learning_rate": 0.0001999984795406574,
-      "loss": 46.0,
-      "step": 10910
-    },
-    {
-      "epoch": 1.757115825918918,
-      "grad_norm": 0.0035237285774201155,
-      "learning_rate": 0.00019999847926166196,
-      "loss": 46.0,
-      "step": 10911
-    },
-    {
-      "epoch": 1.7572768629977054,
-      "grad_norm": 0.0011734850704669952,
-      "learning_rate": 0.00019999847898264094,
-      "loss": 46.0,
-      "step": 10912
-    },
-    {
-      "epoch": 1.7574379000764926,
-      "grad_norm": 0.002724883845075965,
-      "learning_rate": 0.00019999847870359427,
-      "loss": 46.0,
-      "step": 10913
-    },
-    {
-      "epoch": 1.75759893715528,
-      "grad_norm": 0.00047419476322829723,
-      "learning_rate": 0.00019999847842452204,
-      "loss": 46.0,
-      "step": 10914
-    },
-    {
-      "epoch": 1.7577599742340673,
-      "grad_norm": 0.0021352467592805624,
-      "learning_rate": 0.0001999984781454242,
-      "loss": 46.0,
-      "step": 10915
-    },
-    {
-      "epoch": 1.7579210113128547,
-      "grad_norm": 0.005455997306853533,
-      "learning_rate": 0.00019999847786630076,
-      "loss": 46.0,
-      "step": 10916
-    },
-    {
-      "epoch": 1.7580820483916422,
-      "grad_norm": 0.005487693939357996,
-      "learning_rate": 0.00019999847758715174,
-      "loss": 46.0,
-      "step": 10917
-    },
-    {
-      "epoch": 1.7582430854704296,
-      "grad_norm": 0.0016837469302117825,
-      "learning_rate": 0.0001999984773079771,
-      "loss": 46.0,
-      "step": 10918
-    },
-    {
-      "epoch": 1.758404122549217,
-      "grad_norm": 0.0015634101582691073,
-      "learning_rate": 0.00019999847702877692,
-      "loss": 46.0,
-      "step": 10919
-    },
-    {
-      "epoch": 1.7585651596280043,
-      "grad_norm": 0.002548001706600189,
-      "learning_rate": 0.0001999984767495511,
-      "loss": 46.0,
-      "step": 10920
-    },
-    {
-      "epoch": 1.7587261967067918,
-      "grad_norm": 0.0010101523948833346,
-      "learning_rate": 0.0001999984764702997,
-      "loss": 46.0,
-      "step": 10921
-    },
-    {
-      "epoch": 1.758887233785579,
-      "grad_norm": 0.007910455577075481,
-      "learning_rate": 0.0001999984761910227,
-      "loss": 46.0,
-      "step": 10922
-    },
-    {
-      "epoch": 1.7590482708643664,
-      "grad_norm": 0.0021781001705676317,
-      "learning_rate": 0.0001999984759117201,
-      "loss": 46.0,
-      "step": 10923
-    },
-    {
-      "epoch": 1.7592093079431539,
-      "grad_norm": 0.0010348019422963262,
-      "learning_rate": 0.00019999847563239193,
-      "loss": 46.0,
-      "step": 10924
-    },
-    {
-      "epoch": 1.7593703450219413,
-      "grad_norm": 0.0014855475164949894,
-      "learning_rate": 0.00019999847535303816,
-      "loss": 46.0,
-      "step": 10925
-    },
-    {
-      "epoch": 1.7595313821007288,
-      "grad_norm": 0.004134501330554485,
-      "learning_rate": 0.00019999847507365878,
-      "loss": 46.0,
-      "step": 10926
-    },
-    {
-      "epoch": 1.7596924191795162,
-      "grad_norm": 0.0018700124928727746,
-      "learning_rate": 0.00019999847479425383,
-      "loss": 46.0,
-      "step": 10927
-    },
-    {
-      "epoch": 1.7598534562583035,
-      "grad_norm": 0.0019415926653891802,
-      "learning_rate": 0.00019999847451482325,
-      "loss": 46.0,
-      "step": 10928
-    },
-    {
-      "epoch": 1.7600144933370907,
-      "grad_norm": 0.0006833779625594616,
-      "learning_rate": 0.0001999984742353671,
-      "loss": 46.0,
-      "step": 10929
-    },
-    {
-      "epoch": 1.7601755304158782,
-      "grad_norm": 0.0006792580243200064,
-      "learning_rate": 0.00019999847395588534,
-      "loss": 46.0,
-      "step": 10930
-    },
-    {
-      "epoch": 1.7603365674946656,
-      "grad_norm": 0.0008282547350972891,
-      "learning_rate": 0.000199998473676378,
-      "loss": 46.0,
-      "step": 10931
-    },
-    {
-      "epoch": 1.760497604573453,
-      "grad_norm": 0.0020121545530855656,
-      "learning_rate": 0.00019999847339684506,
-      "loss": 46.0,
-      "step": 10932
-    },
-    {
-      "epoch": 1.7606586416522405,
-      "grad_norm": 0.0007697369437664747,
-      "learning_rate": 0.0001999984731172865,
-      "loss": 46.0,
-      "step": 10933
-    },
-    {
-      "epoch": 1.760819678731028,
-      "grad_norm": 0.0019026916706934571,
-      "learning_rate": 0.0001999984728377024,
-      "loss": 46.0,
-      "step": 10934
-    },
-    {
-      "epoch": 1.7609807158098152,
-      "grad_norm": 0.002064588014036417,
-      "learning_rate": 0.00019999847255809268,
-      "loss": 46.0,
-      "step": 10935
-    },
-    {
-      "epoch": 1.7611417528886026,
-      "grad_norm": 0.004713888745754957,
-      "learning_rate": 0.00019999847227845737,
-      "loss": 46.0,
-      "step": 10936
-    },
-    {
-      "epoch": 1.7613027899673899,
-      "grad_norm": 0.001684645307250321,
-      "learning_rate": 0.00019999847199879644,
-      "loss": 46.0,
-      "step": 10937
-    },
-    {
-      "epoch": 1.7614638270461773,
-      "grad_norm": 0.004242990631610155,
-      "learning_rate": 0.00019999847171910996,
-      "loss": 46.0,
-      "step": 10938
-    },
-    {
-      "epoch": 1.7616248641249648,
-      "grad_norm": 0.007821135222911835,
-      "learning_rate": 0.00019999847143939783,
-      "loss": 46.0,
-      "step": 10939
-    },
-    {
-      "epoch": 1.7617859012037522,
-      "grad_norm": 0.0021978523582220078,
-      "learning_rate": 0.00019999847115966012,
-      "loss": 46.0,
-      "step": 10940
-    },
-    {
-      "epoch": 1.7619469382825397,
-      "grad_norm": 0.0012232176959514618,
-      "learning_rate": 0.00019999847087989685,
-      "loss": 46.0,
-      "step": 10941
-    },
-    {
-      "epoch": 1.762107975361327,
-      "grad_norm": 0.005942869931459427,
-      "learning_rate": 0.00019999847060010796,
-      "loss": 46.0,
-      "step": 10942
-    },
-    {
-      "epoch": 1.7622690124401144,
-      "grad_norm": 0.0005504640284925699,
-      "learning_rate": 0.00019999847032029349,
-      "loss": 46.0,
-      "step": 10943
-    },
-    {
-      "epoch": 1.7624300495189016,
-      "grad_norm": 0.001276353606954217,
-      "learning_rate": 0.0001999984700404534,
-      "loss": 46.0,
-      "step": 10944
-    },
-    {
-      "epoch": 1.762591086597689,
-      "grad_norm": 0.0026126888114959,
-      "learning_rate": 0.00019999846976058775,
-      "loss": 46.0,
-      "step": 10945
-    },
-    {
-      "epoch": 1.7627521236764765,
-      "grad_norm": 0.002981804311275482,
-      "learning_rate": 0.00019999846948069648,
-      "loss": 46.0,
-      "step": 10946
-    },
-    {
-      "epoch": 1.762913160755264,
-      "grad_norm": 0.0033719688653945923,
-      "learning_rate": 0.00019999846920077963,
-      "loss": 46.0,
-      "step": 10947
-    },
-    {
-      "epoch": 1.7630741978340514,
-      "grad_norm": 0.000553334888536483,
-      "learning_rate": 0.00019999846892083717,
-      "loss": 46.0,
-      "step": 10948
-    },
-    {
-      "epoch": 1.7632352349128388,
-      "grad_norm": 0.002527834614738822,
-      "learning_rate": 0.00019999846864086912,
-      "loss": 46.0,
-      "step": 10949
-    },
-    {
-      "epoch": 1.763396271991626,
-      "grad_norm": 0.0035288031212985516,
-      "learning_rate": 0.0001999984683608755,
-      "loss": 46.0,
-      "step": 10950
-    },
-    {
-      "epoch": 1.7635573090704133,
-      "grad_norm": 0.000837587402202189,
-      "learning_rate": 0.00019999846808085625,
-      "loss": 46.0,
-      "step": 10951
-    },
-    {
-      "epoch": 1.7637183461492008,
-      "grad_norm": 0.0011087177554145455,
-      "learning_rate": 0.0001999984678008114,
-      "loss": 46.0,
-      "step": 10952
-    },
-    {
-      "epoch": 1.7638793832279882,
-      "grad_norm": 0.0007402384071610868,
-      "learning_rate": 0.000199998467520741,
-      "loss": 46.0,
-      "step": 10953
-    },
-    {
-      "epoch": 1.7640404203067757,
-      "grad_norm": 0.0013717241818085313,
-      "learning_rate": 0.00019999846724064497,
-      "loss": 46.0,
-      "step": 10954
-    },
-    {
-      "epoch": 1.764201457385563,
-      "grad_norm": 0.002512422390282154,
-      "learning_rate": 0.00019999846696052337,
-      "loss": 46.0,
-      "step": 10955
-    },
-    {
-      "epoch": 1.7643624944643506,
-      "grad_norm": 0.0009923626203089952,
-      "learning_rate": 0.00019999846668037615,
-      "loss": 46.0,
-      "step": 10956
-    },
-    {
-      "epoch": 1.7645235315431378,
-      "grad_norm": 0.0004795429122168571,
-      "learning_rate": 0.00019999846640020335,
-      "loss": 46.0,
-      "step": 10957
-    },
-    {
-      "epoch": 1.7646845686219252,
-      "grad_norm": 0.002552557270973921,
-      "learning_rate": 0.00019999846612000496,
-      "loss": 46.0,
-      "step": 10958
-    },
-    {
-      "epoch": 1.7648456057007125,
-      "grad_norm": 0.004905923269689083,
-      "learning_rate": 0.00019999846583978098,
-      "loss": 46.0,
-      "step": 10959
-    },
-    {
-      "epoch": 1.7650066427795,
-      "grad_norm": 0.0016231703339144588,
-      "learning_rate": 0.0001999984655595314,
-      "loss": 46.0,
-      "step": 10960
-    },
-    {
-      "epoch": 1.7651676798582874,
-      "grad_norm": 0.004946983885020018,
-      "learning_rate": 0.0001999984652792562,
-      "loss": 46.0,
-      "step": 10961
-    },
-    {
-      "epoch": 1.7653287169370748,
-      "grad_norm": 0.0017494576750323176,
-      "learning_rate": 0.00019999846499895542,
-      "loss": 46.0,
-      "step": 10962
-    },
-    {
-      "epoch": 1.7654897540158623,
-      "grad_norm": 0.003942117094993591,
-      "learning_rate": 0.00019999846471862906,
-      "loss": 46.0,
-      "step": 10963
-    },
-    {
-      "epoch": 1.7656507910946495,
-      "grad_norm": 0.002034541219472885,
-      "learning_rate": 0.0001999984644382771,
-      "loss": 46.0,
-      "step": 10964
-    },
-    {
-      "epoch": 1.765811828173437,
-      "grad_norm": 0.0023761694319546223,
-      "learning_rate": 0.00019999846415789954,
-      "loss": 46.0,
-      "step": 10965
-    },
-    {
-      "epoch": 1.7659728652522242,
-      "grad_norm": 0.0012503170873969793,
-      "learning_rate": 0.0001999984638774964,
-      "loss": 46.0,
-      "step": 10966
-    },
-    {
-      "epoch": 1.7661339023310116,
-      "grad_norm": 0.002663065679371357,
-      "learning_rate": 0.00019999846359706764,
-      "loss": 46.0,
-      "step": 10967
-    },
-    {
-      "epoch": 1.766294939409799,
-      "grad_norm": 0.0020416222978383303,
-      "learning_rate": 0.0001999984633166133,
-      "loss": 46.0,
-      "step": 10968
-    },
-    {
-      "epoch": 1.7664559764885865,
-      "grad_norm": 0.003008096944540739,
-      "learning_rate": 0.00019999846303613337,
-      "loss": 46.0,
-      "step": 10969
-    },
-    {
-      "epoch": 1.766617013567374,
-      "grad_norm": 0.0011060424149036407,
-      "learning_rate": 0.00019999846275562785,
-      "loss": 46.0,
-      "step": 10970
-    },
-    {
-      "epoch": 1.7667780506461612,
-      "grad_norm": 0.0020178980194032192,
-      "learning_rate": 0.00019999846247509672,
-      "loss": 46.0,
-      "step": 10971
-    },
-    {
-      "epoch": 1.7669390877249487,
-      "grad_norm": 0.0030181705951690674,
-      "learning_rate": 0.00019999846219454,
-      "loss": 46.0,
-      "step": 10972
-    },
-    {
-      "epoch": 1.767100124803736,
-      "grad_norm": 0.0021134752314537764,
-      "learning_rate": 0.0001999984619139577,
-      "loss": 46.0,
-      "step": 10973
-    },
-    {
-      "epoch": 1.7672611618825234,
-      "grad_norm": 0.0006657113553956151,
-      "learning_rate": 0.00019999846163334977,
-      "loss": 46.0,
-      "step": 10974
-    },
-    {
-      "epoch": 1.7674221989613108,
-      "grad_norm": 0.000494988402351737,
-      "learning_rate": 0.00019999846135271626,
-      "loss": 46.0,
-      "step": 10975
-    },
-    {
-      "epoch": 1.7675832360400983,
-      "grad_norm": 0.0023772960994392633,
-      "learning_rate": 0.00019999846107205717,
-      "loss": 46.0,
-      "step": 10976
-    },
-    {
-      "epoch": 1.7677442731188857,
-      "grad_norm": 0.0017018452053889632,
-      "learning_rate": 0.00019999846079137248,
-      "loss": 46.0,
-      "step": 10977
-    },
-    {
-      "epoch": 1.7679053101976732,
-      "grad_norm": 0.0007068027043715119,
-      "learning_rate": 0.0001999984605106622,
-      "loss": 46.0,
-      "step": 10978
-    },
-    {
-      "epoch": 1.7680663472764604,
-      "grad_norm": 0.0027879702392965555,
-      "learning_rate": 0.0001999984602299263,
-      "loss": 46.0,
-      "step": 10979
-    },
-    {
-      "epoch": 1.7682273843552478,
-      "grad_norm": 0.006277864333242178,
-      "learning_rate": 0.00019999845994916483,
-      "loss": 46.0,
-      "step": 10980
-    },
-    {
-      "epoch": 1.768388421434035,
-      "grad_norm": 0.003597882343456149,
-      "learning_rate": 0.00019999845966837775,
-      "loss": 46.0,
-      "step": 10981
-    },
-    {
-      "epoch": 1.7685494585128225,
-      "grad_norm": 0.0007268081535585225,
-      "learning_rate": 0.0001999984593875651,
-      "loss": 46.0,
-      "step": 10982
-    },
-    {
-      "epoch": 1.76871049559161,
-      "grad_norm": 0.0027878275141119957,
-      "learning_rate": 0.00019999845910672684,
-      "loss": 46.0,
-      "step": 10983
-    },
-    {
-      "epoch": 1.7688715326703974,
-      "grad_norm": 0.0010864431969821453,
-      "learning_rate": 0.000199998458825863,
-      "loss": 46.0,
-      "step": 10984
-    },
-    {
-      "epoch": 1.7690325697491849,
-      "grad_norm": 0.0018990386743098497,
-      "learning_rate": 0.00019999845854497354,
-      "loss": 46.0,
-      "step": 10985
-    },
-    {
-      "epoch": 1.769193606827972,
-      "grad_norm": 0.0013602948747575283,
-      "learning_rate": 0.00019999845826405851,
-      "loss": 46.0,
-      "step": 10986
-    },
-    {
-      "epoch": 1.7693546439067596,
-      "grad_norm": 0.0064102960750460625,
-      "learning_rate": 0.00019999845798311788,
-      "loss": 46.0,
-      "step": 10987
-    },
-    {
-      "epoch": 1.7695156809855468,
-      "grad_norm": 0.0028815045952796936,
-      "learning_rate": 0.00019999845770215163,
-      "loss": 46.0,
-      "step": 10988
-    },
-    {
-      "epoch": 1.7696767180643342,
-      "grad_norm": 0.0006191595457494259,
-      "learning_rate": 0.00019999845742115982,
-      "loss": 46.0,
-      "step": 10989
-    },
-    {
-      "epoch": 1.7698377551431217,
-      "grad_norm": 0.0019128312123939395,
-      "learning_rate": 0.0001999984571401424,
-      "loss": 46.0,
-      "step": 10990
-    },
-    {
-      "epoch": 1.7699987922219091,
-      "grad_norm": 0.0007664788281545043,
-      "learning_rate": 0.00019999845685909938,
-      "loss": 46.0,
-      "step": 10991
-    },
-    {
-      "epoch": 1.7701598293006966,
-      "grad_norm": 0.0071024163626134396,
-      "learning_rate": 0.00019999845657803076,
-      "loss": 46.0,
-      "step": 10992
-    },
-    {
-      "epoch": 1.7703208663794838,
-      "grad_norm": 0.001521787140518427,
-      "learning_rate": 0.00019999845629693657,
-      "loss": 46.0,
-      "step": 10993
-    },
-    {
-      "epoch": 1.7704819034582713,
-      "grad_norm": 0.0025173607282340527,
-      "learning_rate": 0.00019999845601581677,
-      "loss": 46.0,
-      "step": 10994
-    },
-    {
-      "epoch": 1.7706429405370585,
-      "grad_norm": 0.001532764988951385,
-      "learning_rate": 0.0001999984557346714,
-      "loss": 46.0,
-      "step": 10995
-    },
-    {
-      "epoch": 1.770803977615846,
-      "grad_norm": 0.007769228424876928,
-      "learning_rate": 0.00019999845545350039,
-      "loss": 46.0,
-      "step": 10996
-    },
-    {
-      "epoch": 1.7709650146946334,
-      "grad_norm": 0.0013940640492364764,
-      "learning_rate": 0.0001999984551723038,
-      "loss": 46.0,
-      "step": 10997
-    },
-    {
-      "epoch": 1.7711260517734209,
-      "grad_norm": 0.007811498362571001,
-      "learning_rate": 0.00019999845489108165,
-      "loss": 46.0,
-      "step": 10998
-    },
-    {
-      "epoch": 1.7712870888522083,
-      "grad_norm": 0.00822006817907095,
-      "learning_rate": 0.00019999845460983386,
-      "loss": 46.0,
-      "step": 10999
-    },
-    {
-      "epoch": 1.7714481259309958,
-      "grad_norm": 0.007976683788001537,
-      "learning_rate": 0.0001999984543285605,
-      "loss": 46.0,
-      "step": 11000
-    },
-    {
-      "epoch": 1.771609163009783,
-      "grad_norm": 0.005001617129892111,
-      "learning_rate": 0.00019999845404726155,
-      "loss": 46.0,
-      "step": 11001
-    },
-    {
-      "epoch": 1.7717702000885704,
-      "grad_norm": 0.001979105407372117,
-      "learning_rate": 0.000199998453765937,
-      "loss": 46.0,
-      "step": 11002
-    },
-    {
-      "epoch": 1.7719312371673577,
-      "grad_norm": 0.0013469480909407139,
-      "learning_rate": 0.00019999845348458686,
-      "loss": 46.0,
-      "step": 11003
-    },
-    {
-      "epoch": 1.7720922742461451,
-      "grad_norm": 0.0015179608017206192,
-      "learning_rate": 0.00019999845320321108,
-      "loss": 46.0,
-      "step": 11004
-    },
-    {
-      "epoch": 1.7722533113249326,
-      "grad_norm": 0.0017213443061336875,
-      "learning_rate": 0.00019999845292180976,
-      "loss": 46.0,
-      "step": 11005
-    },
-    {
-      "epoch": 1.77241434840372,
-      "grad_norm": 0.0049031986854970455,
-      "learning_rate": 0.0001999984526403828,
-      "loss": 46.0,
-      "step": 11006
-    },
-    {
-      "epoch": 1.7725753854825075,
-      "grad_norm": 0.008329499512910843,
-      "learning_rate": 0.00019999845235893032,
-      "loss": 46.0,
-      "step": 11007
-    },
-    {
-      "epoch": 1.7727364225612947,
-      "grad_norm": 0.0007381970062851906,
-      "learning_rate": 0.0001999984520774522,
-      "loss": 46.0,
-      "step": 11008
-    },
-    {
-      "epoch": 1.7728974596400822,
-      "grad_norm": 0.01034065056592226,
-      "learning_rate": 0.00019999845179594848,
-      "loss": 46.0,
-      "step": 11009
-    },
-    {
-      "epoch": 1.7730584967188694,
-      "grad_norm": 0.007265487220138311,
-      "learning_rate": 0.00019999845151441915,
-      "loss": 46.0,
-      "step": 11010
-    },
-    {
-      "epoch": 1.7732195337976568,
-      "grad_norm": 0.0008545235614292324,
-      "learning_rate": 0.00019999845123286426,
-      "loss": 46.0,
-      "step": 11011
-    },
-    {
-      "epoch": 1.7733805708764443,
-      "grad_norm": 0.0013570995070040226,
-      "learning_rate": 0.00019999845095128378,
-      "loss": 46.0,
-      "step": 11012
-    },
-    {
-      "epoch": 1.7735416079552317,
-      "grad_norm": 0.0018267452251166105,
-      "learning_rate": 0.00019999845066967766,
-      "loss": 46.0,
-      "step": 11013
-    },
-    {
-      "epoch": 1.7737026450340192,
-      "grad_norm": 0.006035581696778536,
-      "learning_rate": 0.00019999845038804596,
-      "loss": 46.0,
-      "step": 11014
-    },
-    {
-      "epoch": 1.7738636821128064,
-      "grad_norm": 0.0011682414915412664,
-      "learning_rate": 0.0001999984501063887,
-      "loss": 46.0,
-      "step": 11015
-    },
-    {
-      "epoch": 1.7740247191915939,
-      "grad_norm": 0.0011177061824128032,
-      "learning_rate": 0.00019999844982470584,
-      "loss": 46.0,
-      "step": 11016
-    },
-    {
-      "epoch": 1.774185756270381,
-      "grad_norm": 0.004796769469976425,
-      "learning_rate": 0.00019999844954299737,
-      "loss": 46.0,
-      "step": 11017
-    },
-    {
-      "epoch": 1.7743467933491686,
-      "grad_norm": 0.0036517970729619265,
-      "learning_rate": 0.0001999984492612633,
-      "loss": 46.0,
-      "step": 11018
-    },
-    {
-      "epoch": 1.774507830427956,
-      "grad_norm": 0.0033792583271861076,
-      "learning_rate": 0.00019999844897950362,
-      "loss": 46.0,
-      "step": 11019
-    },
-    {
-      "epoch": 1.7746688675067435,
-      "grad_norm": 0.002855784958228469,
-      "learning_rate": 0.0001999984486977184,
-      "loss": 46.0,
-      "step": 11020
-    },
-    {
-      "epoch": 1.774829904585531,
-      "grad_norm": 0.008407426066696644,
-      "learning_rate": 0.00019999844841590752,
-      "loss": 46.0,
-      "step": 11021
-    },
-    {
-      "epoch": 1.7749909416643184,
-      "grad_norm": 0.0003833010559901595,
-      "learning_rate": 0.0001999984481340711,
-      "loss": 46.0,
-      "step": 11022
-    },
-    {
-      "epoch": 1.7751519787431056,
-      "grad_norm": 0.0029628907795995474,
-      "learning_rate": 0.00019999844785220904,
-      "loss": 46.0,
-      "step": 11023
-    },
-    {
-      "epoch": 1.7753130158218928,
-      "grad_norm": 0.002565759001299739,
-      "learning_rate": 0.0001999984475703214,
-      "loss": 46.0,
-      "step": 11024
-    },
-    {
-      "epoch": 1.7754740529006803,
-      "grad_norm": 0.0013392409309744835,
-      "learning_rate": 0.00019999844728840817,
-      "loss": 46.0,
-      "step": 11025
-    },
-    {
-      "epoch": 1.7756350899794677,
-      "grad_norm": 0.0029403886292129755,
-      "learning_rate": 0.00019999844700646936,
-      "loss": 46.0,
-      "step": 11026
-    },
-    {
-      "epoch": 1.7757961270582552,
-      "grad_norm": 0.0050983186811208725,
-      "learning_rate": 0.00019999844672450494,
-      "loss": 46.0,
-      "step": 11027
-    },
-    {
-      "epoch": 1.7759571641370426,
-      "grad_norm": 0.0010983977699652314,
-      "learning_rate": 0.00019999844644251496,
-      "loss": 46.0,
-      "step": 11028
-    },
-    {
-      "epoch": 1.77611820121583,
-      "grad_norm": 0.0008620453882031143,
-      "learning_rate": 0.00019999844616049934,
-      "loss": 46.0,
-      "step": 11029
-    },
-    {
-      "epoch": 1.7762792382946173,
-      "grad_norm": 0.004579078871756792,
-      "learning_rate": 0.00019999844587845816,
-      "loss": 46.0,
-      "step": 11030
-    },
-    {
-      "epoch": 1.7764402753734048,
-      "grad_norm": 0.0017242144094780087,
-      "learning_rate": 0.00019999844559639133,
-      "loss": 46.0,
-      "step": 11031
-    },
-    {
-      "epoch": 1.776601312452192,
-      "grad_norm": 0.008659831248223782,
-      "learning_rate": 0.00019999844531429898,
-      "loss": 46.0,
-      "step": 11032
-    },
-    {
-      "epoch": 1.7767623495309794,
-      "grad_norm": 0.0005555757670663297,
-      "learning_rate": 0.00019999844503218098,
-      "loss": 46.0,
-      "step": 11033
-    },
-    {
-      "epoch": 1.776923386609767,
-      "grad_norm": 0.0010951929725706577,
-      "learning_rate": 0.00019999844475003742,
-      "loss": 46.0,
-      "step": 11034
-    },
-    {
-      "epoch": 1.7770844236885543,
-      "grad_norm": 0.0007955901091918349,
-      "learning_rate": 0.00019999844446786825,
-      "loss": 46.0,
-      "step": 11035
-    },
-    {
-      "epoch": 1.7772454607673418,
-      "grad_norm": 0.001042502699419856,
-      "learning_rate": 0.0001999984441856735,
-      "loss": 46.0,
-      "step": 11036
-    },
-    {
-      "epoch": 1.777406497846129,
-      "grad_norm": 0.0028987498953938484,
-      "learning_rate": 0.0001999984439034531,
-      "loss": 46.0,
-      "step": 11037
-    },
-    {
-      "epoch": 1.7775675349249165,
-      "grad_norm": 0.002884478308260441,
-      "learning_rate": 0.00019999844362120713,
-      "loss": 46.0,
-      "step": 11038
-    },
-    {
-      "epoch": 1.7777285720037037,
-      "grad_norm": 0.002744531724601984,
-      "learning_rate": 0.0001999984433389356,
-      "loss": 46.0,
-      "step": 11039
-    },
-    {
-      "epoch": 1.7778896090824912,
-      "grad_norm": 0.000942451530136168,
-      "learning_rate": 0.00019999844305663847,
-      "loss": 46.0,
-      "step": 11040
-    },
-    {
-      "epoch": 1.7780506461612786,
-      "grad_norm": 0.0004730231303256005,
-      "learning_rate": 0.00019999844277431573,
-      "loss": 46.0,
-      "step": 11041
-    },
-    {
-      "epoch": 1.778211683240066,
-      "grad_norm": 0.0009635805035941303,
-      "learning_rate": 0.0001999984424919674,
-      "loss": 46.0,
-      "step": 11042
-    },
-    {
-      "epoch": 1.7783727203188535,
-      "grad_norm": 0.000491833605337888,
-      "learning_rate": 0.00019999844220959347,
-      "loss": 46.0,
-      "step": 11043
-    },
-    {
-      "epoch": 1.778533757397641,
-      "grad_norm": 0.0017154552042484283,
-      "learning_rate": 0.00019999844192719395,
-      "loss": 46.0,
-      "step": 11044
-    },
-    {
-      "epoch": 1.7786947944764282,
-      "grad_norm": 0.0026958847884088755,
-      "learning_rate": 0.00019999844164476883,
-      "loss": 46.0,
-      "step": 11045
-    },
-    {
-      "epoch": 1.7788558315552154,
-      "grad_norm": 0.0055764829739928246,
-      "learning_rate": 0.00019999844136231812,
-      "loss": 46.0,
-      "step": 11046
-    },
-    {
-      "epoch": 1.7790168686340029,
-      "grad_norm": 0.0014100803527981043,
-      "learning_rate": 0.0001999984410798418,
-      "loss": 46.0,
-      "step": 11047
-    },
-    {
-      "epoch": 1.7791779057127903,
-      "grad_norm": 0.0010405626380816102,
-      "learning_rate": 0.0001999984407973399,
-      "loss": 46.0,
-      "step": 11048
-    },
-    {
-      "epoch": 1.7793389427915778,
-      "grad_norm": 0.0006019024294801056,
-      "learning_rate": 0.00019999844051481243,
-      "loss": 46.0,
-      "step": 11049
-    },
-    {
-      "epoch": 1.7794999798703652,
-      "grad_norm": 0.004517868161201477,
-      "learning_rate": 0.00019999844023225934,
-      "loss": 46.0,
-      "step": 11050
-    },
-    {
-      "epoch": 1.7796610169491527,
-      "grad_norm": 0.00223662331700325,
-      "learning_rate": 0.00019999843994968064,
-      "loss": 46.0,
-      "step": 11051
-    },
-    {
-      "epoch": 1.77982205402794,
-      "grad_norm": 0.0016908921534195542,
-      "learning_rate": 0.00019999843966707638,
-      "loss": 46.0,
-      "step": 11052
-    },
-    {
-      "epoch": 1.7799830911067274,
-      "grad_norm": 0.001135583152063191,
-      "learning_rate": 0.0001999984393844465,
-      "loss": 46.0,
-      "step": 11053
-    },
-    {
-      "epoch": 1.7801441281855146,
-      "grad_norm": 0.0024618771858513355,
-      "learning_rate": 0.00019999843910179104,
-      "loss": 46.0,
-      "step": 11054
-    },
-    {
-      "epoch": 1.780305165264302,
-      "grad_norm": 0.001972004771232605,
-      "learning_rate": 0.00019999843881911,
-      "loss": 46.0,
-      "step": 11055
-    },
-    {
-      "epoch": 1.7804662023430895,
-      "grad_norm": 0.0011904244311153889,
-      "learning_rate": 0.00019999843853640332,
-      "loss": 46.0,
-      "step": 11056
-    },
-    {
-      "epoch": 1.780627239421877,
-      "grad_norm": 0.004262834321707487,
-      "learning_rate": 0.00019999843825367107,
-      "loss": 46.0,
-      "step": 11057
-    },
-    {
-      "epoch": 1.7807882765006644,
-      "grad_norm": 0.0008145908359438181,
-      "learning_rate": 0.00019999843797091323,
-      "loss": 46.0,
-      "step": 11058
-    },
-    {
-      "epoch": 1.7809493135794516,
-      "grad_norm": 0.004048207774758339,
-      "learning_rate": 0.00019999843768812978,
-      "loss": 46.0,
-      "step": 11059
-    },
-    {
-      "epoch": 1.781110350658239,
-      "grad_norm": 0.003517892211675644,
-      "learning_rate": 0.00019999843740532076,
-      "loss": 46.0,
-      "step": 11060
-    },
-    {
-      "epoch": 1.7812713877370263,
-      "grad_norm": 0.000693217443767935,
-      "learning_rate": 0.00019999843712248613,
-      "loss": 46.0,
-      "step": 11061
-    },
-    {
-      "epoch": 1.7814324248158138,
-      "grad_norm": 0.0027239397168159485,
-      "learning_rate": 0.00019999843683962592,
-      "loss": 46.0,
-      "step": 11062
-    },
-    {
-      "epoch": 1.7815934618946012,
-      "grad_norm": 0.00394403887912631,
-      "learning_rate": 0.0001999984365567401,
-      "loss": 46.0,
-      "step": 11063
-    },
-    {
-      "epoch": 1.7817544989733887,
-      "grad_norm": 0.0013011100236326456,
-      "learning_rate": 0.00019999843627382867,
-      "loss": 46.0,
-      "step": 11064
-    },
-    {
-      "epoch": 1.7819155360521761,
-      "grad_norm": 0.0030322198290377855,
-      "learning_rate": 0.00019999843599089167,
-      "loss": 46.0,
-      "step": 11065
-    },
-    {
-      "epoch": 1.7820765731309636,
-      "grad_norm": 0.003137225518003106,
-      "learning_rate": 0.00019999843570792908,
-      "loss": 46.0,
-      "step": 11066
-    },
-    {
-      "epoch": 1.7822376102097508,
-      "grad_norm": 0.005587464664131403,
-      "learning_rate": 0.0001999984354249409,
-      "loss": 46.0,
-      "step": 11067
-    },
-    {
-      "epoch": 1.782398647288538,
-      "grad_norm": 0.009867976419627666,
-      "learning_rate": 0.0001999984351419271,
-      "loss": 46.0,
-      "step": 11068
-    },
-    {
-      "epoch": 1.7825596843673255,
-      "grad_norm": 0.002365303225815296,
-      "learning_rate": 0.00019999843485888773,
-      "loss": 46.0,
-      "step": 11069
-    },
-    {
-      "epoch": 1.782720721446113,
-      "grad_norm": 0.005722887814044952,
-      "learning_rate": 0.00019999843457582273,
-      "loss": 46.0,
-      "step": 11070
-    },
-    {
-      "epoch": 1.7828817585249004,
-      "grad_norm": 0.0009947649668902159,
-      "learning_rate": 0.00019999843429273218,
-      "loss": 46.0,
-      "step": 11071
-    },
-    {
-      "epoch": 1.7830427956036878,
-      "grad_norm": 0.004566850606352091,
-      "learning_rate": 0.00019999843400961598,
-      "loss": 46.0,
-      "step": 11072
-    },
-    {
-      "epoch": 1.7832038326824753,
-      "grad_norm": 0.000928681343793869,
-      "learning_rate": 0.00019999843372647426,
-      "loss": 46.0,
-      "step": 11073
-    },
-    {
-      "epoch": 1.7833648697612625,
-      "grad_norm": 0.0011853943578898907,
-      "learning_rate": 0.00019999843344330689,
-      "loss": 46.0,
-      "step": 11074
-    },
-    {
-      "epoch": 1.78352590684005,
-      "grad_norm": 0.0008863420225679874,
-      "learning_rate": 0.00019999843316011393,
-      "loss": 46.0,
-      "step": 11075
-    },
-    {
-      "epoch": 1.7836869439188372,
-      "grad_norm": 0.003711366793140769,
-      "learning_rate": 0.00019999843287689539,
-      "loss": 46.0,
-      "step": 11076
-    },
-    {
-      "epoch": 1.7838479809976246,
-      "grad_norm": 0.0014373952290043235,
-      "learning_rate": 0.00019999843259365125,
-      "loss": 46.0,
-      "step": 11077
-    },
-    {
-      "epoch": 1.784009018076412,
-      "grad_norm": 0.0012223607627674937,
-      "learning_rate": 0.00019999843231038154,
-      "loss": 46.0,
-      "step": 11078
-    },
-    {
-      "epoch": 1.7841700551551996,
-      "grad_norm": 0.0006321249529719353,
-      "learning_rate": 0.0001999984320270862,
-      "loss": 46.0,
-      "step": 11079
-    },
-    {
-      "epoch": 1.784331092233987,
-      "grad_norm": 0.003856287570670247,
-      "learning_rate": 0.00019999843174376526,
-      "loss": 46.0,
-      "step": 11080
-    },
-    {
-      "epoch": 1.7844921293127742,
-      "grad_norm": 0.004373412579298019,
-      "learning_rate": 0.00019999843146041878,
-      "loss": 46.0,
-      "step": 11081
-    },
-    {
-      "epoch": 1.7846531663915617,
-      "grad_norm": 0.001382243586704135,
-      "learning_rate": 0.00019999843117704665,
-      "loss": 46.0,
-      "step": 11082
-    },
-    {
-      "epoch": 1.784814203470349,
-      "grad_norm": 0.0038277197163552046,
-      "learning_rate": 0.00019999843089364895,
-      "loss": 46.0,
-      "step": 11083
-    },
-    {
-      "epoch": 1.7849752405491364,
-      "grad_norm": 0.005997433792799711,
-      "learning_rate": 0.00019999843061022565,
-      "loss": 46.0,
-      "step": 11084
-    },
-    {
-      "epoch": 1.7851362776279238,
-      "grad_norm": 0.001283555757254362,
-      "learning_rate": 0.00019999843032677674,
-      "loss": 46.0,
-      "step": 11085
-    },
-    {
-      "epoch": 1.7852973147067113,
-      "grad_norm": 0.0015188554534688592,
-      "learning_rate": 0.00019999843004330224,
-      "loss": 46.0,
-      "step": 11086
-    },
-    {
-      "epoch": 1.7854583517854987,
-      "grad_norm": 0.0017779256450012326,
-      "learning_rate": 0.00019999842975980219,
-      "loss": 46.0,
-      "step": 11087
-    },
-    {
-      "epoch": 1.785619388864286,
-      "grad_norm": 0.001990334363654256,
-      "learning_rate": 0.0001999984294762765,
-      "loss": 46.0,
-      "step": 11088
-    },
-    {
-      "epoch": 1.7857804259430734,
-      "grad_norm": 0.002235524356365204,
-      "learning_rate": 0.00019999842919272523,
-      "loss": 46.0,
-      "step": 11089
-    },
-    {
-      "epoch": 1.7859414630218606,
-      "grad_norm": 0.004958270117640495,
-      "learning_rate": 0.00019999842890914836,
-      "loss": 46.0,
-      "step": 11090
-    },
-    {
-      "epoch": 1.786102500100648,
-      "grad_norm": 0.007544436492025852,
-      "learning_rate": 0.0001999984286255459,
-      "loss": 46.0,
-      "step": 11091
-    },
-    {
-      "epoch": 1.7862635371794355,
-      "grad_norm": 0.0014472788898274302,
-      "learning_rate": 0.00019999842834191785,
-      "loss": 46.0,
-      "step": 11092
-    },
-    {
-      "epoch": 1.786424574258223,
-      "grad_norm": 0.0021815740037709475,
-      "learning_rate": 0.0001999984280582642,
-      "loss": 46.0,
-      "step": 11093
-    },
-    {
-      "epoch": 1.7865856113370104,
-      "grad_norm": 0.002626087050884962,
-      "learning_rate": 0.00019999842777458494,
-      "loss": 46.0,
-      "step": 11094
-    },
-    {
-      "epoch": 1.786746648415798,
-      "grad_norm": 0.0033277596812695265,
-      "learning_rate": 0.0001999984274908801,
-      "loss": 46.0,
-      "step": 11095
-    },
-    {
-      "epoch": 1.7869076854945851,
-      "grad_norm": 0.0010635490762069821,
-      "learning_rate": 0.00019999842720714968,
-      "loss": 46.0,
-      "step": 11096
-    },
-    {
-      "epoch": 1.7870687225733726,
-      "grad_norm": 0.0014558301772922277,
-      "learning_rate": 0.00019999842692339367,
-      "loss": 46.0,
-      "step": 11097
-    },
-    {
-      "epoch": 1.7872297596521598,
-      "grad_norm": 0.002081095241010189,
-      "learning_rate": 0.00019999842663961204,
-      "loss": 46.0,
-      "step": 11098
-    },
-    {
-      "epoch": 1.7873907967309473,
-      "grad_norm": 0.0071091377176344395,
-      "learning_rate": 0.00019999842635580478,
-      "loss": 46.0,
-      "step": 11099
-    },
-    {
-      "epoch": 1.7875518338097347,
-      "grad_norm": 0.0022333275992423296,
-      "learning_rate": 0.00019999842607197198,
-      "loss": 46.0,
-      "step": 11100
-    },
-    {
-      "epoch": 1.7877128708885222,
-      "grad_norm": 0.0010902582434937358,
-      "learning_rate": 0.0001999984257881136,
-      "loss": 46.0,
-      "step": 11101
-    },
-    {
-      "epoch": 1.7878739079673096,
-      "grad_norm": 0.0030951553490012884,
-      "learning_rate": 0.0001999984255042296,
-      "loss": 46.0,
-      "step": 11102
-    },
-    {
-      "epoch": 1.7880349450460968,
-      "grad_norm": 0.005409586243331432,
-      "learning_rate": 0.00019999842522032,
-      "loss": 46.0,
-      "step": 11103
-    },
-    {
-      "epoch": 1.7881959821248843,
-      "grad_norm": 0.0015288225840777159,
-      "learning_rate": 0.00019999842493638478,
-      "loss": 46.0,
-      "step": 11104
-    },
-    {
-      "epoch": 1.7883570192036715,
-      "grad_norm": 0.0022097965702414513,
-      "learning_rate": 0.00019999842465242402,
-      "loss": 46.0,
-      "step": 11105
-    },
-    {
-      "epoch": 1.788518056282459,
-      "grad_norm": 0.001397671177983284,
-      "learning_rate": 0.00019999842436843764,
-      "loss": 46.0,
-      "step": 11106
-    },
-    {
-      "epoch": 1.7886790933612464,
-      "grad_norm": 0.0022432473488152027,
-      "learning_rate": 0.00019999842408442565,
-      "loss": 46.0,
-      "step": 11107
-    },
-    {
-      "epoch": 1.7888401304400339,
-      "grad_norm": 0.0027394231874495745,
-      "learning_rate": 0.00019999842380038813,
-      "loss": 46.0,
-      "step": 11108
-    },
-    {
-      "epoch": 1.7890011675188213,
-      "grad_norm": 0.0031087223906069994,
-      "learning_rate": 0.00019999842351632493,
-      "loss": 46.0,
-      "step": 11109
-    },
-    {
-      "epoch": 1.7891622045976086,
-      "grad_norm": 0.0008448674925602973,
-      "learning_rate": 0.00019999842323223618,
-      "loss": 46.0,
-      "step": 11110
-    },
-    {
-      "epoch": 1.789323241676396,
-      "grad_norm": 0.0008297382737509906,
-      "learning_rate": 0.00019999842294812184,
-      "loss": 46.0,
-      "step": 11111
-    },
-    {
-      "epoch": 1.7894842787551832,
-      "grad_norm": 0.0034324219450354576,
-      "learning_rate": 0.0001999984226639819,
-      "loss": 46.0,
-      "step": 11112
-    },
-    {
-      "epoch": 1.7896453158339707,
-      "grad_norm": 0.0009411865612491965,
-      "learning_rate": 0.00019999842237981638,
-      "loss": 46.0,
-      "step": 11113
-    },
-    {
-      "epoch": 1.7898063529127581,
-      "grad_norm": 0.005342352204024792,
-      "learning_rate": 0.0001999984220956252,
-      "loss": 46.0,
-      "step": 11114
-    },
-    {
-      "epoch": 1.7899673899915456,
-      "grad_norm": 0.003260920522734523,
-      "learning_rate": 0.00019999842181140848,
-      "loss": 46.0,
-      "step": 11115
-    },
-    {
-      "epoch": 1.790128427070333,
-      "grad_norm": 0.000732087588403374,
-      "learning_rate": 0.00019999842152716618,
-      "loss": 46.0,
-      "step": 11116
-    },
-    {
-      "epoch": 1.7902894641491205,
-      "grad_norm": 0.0010504265083000064,
-      "learning_rate": 0.00019999842124289826,
-      "loss": 46.0,
-      "step": 11117
-    },
-    {
-      "epoch": 1.7904505012279077,
-      "grad_norm": 0.0037848721258342266,
-      "learning_rate": 0.00019999842095860473,
-      "loss": 46.0,
-      "step": 11118
-    },
-    {
-      "epoch": 1.7906115383066952,
-      "grad_norm": 0.011950554326176643,
-      "learning_rate": 0.0001999984206742856,
-      "loss": 46.0,
-      "step": 11119
-    },
-    {
-      "epoch": 1.7907725753854824,
-      "grad_norm": 0.00043025804916396737,
-      "learning_rate": 0.00019999842038994094,
-      "loss": 46.0,
-      "step": 11120
-    },
-    {
-      "epoch": 1.7909336124642699,
-      "grad_norm": 0.00383752747438848,
-      "learning_rate": 0.00019999842010557064,
-      "loss": 46.0,
-      "step": 11121
-    },
-    {
-      "epoch": 1.7910946495430573,
-      "grad_norm": 0.002125471830368042,
-      "learning_rate": 0.00019999841982117476,
-      "loss": 46.0,
-      "step": 11122
-    },
-    {
-      "epoch": 1.7912556866218448,
-      "grad_norm": 0.010294078849256039,
-      "learning_rate": 0.00019999841953675324,
-      "loss": 46.0,
-      "step": 11123
-    },
-    {
-      "epoch": 1.7914167237006322,
-      "grad_norm": 0.0018050142098218203,
-      "learning_rate": 0.0001999984192523062,
-      "loss": 46.0,
-      "step": 11124
-    },
-    {
-      "epoch": 1.7915777607794194,
-      "grad_norm": 0.014061805792152882,
-      "learning_rate": 0.00019999841896783352,
-      "loss": 46.0,
-      "step": 11125
-    },
-    {
-      "epoch": 1.7917387978582069,
-      "grad_norm": 0.004811234772205353,
-      "learning_rate": 0.00019999841868333524,
-      "loss": 46.0,
-      "step": 11126
-    },
-    {
-      "epoch": 1.7918998349369941,
-      "grad_norm": 0.016827654093503952,
-      "learning_rate": 0.00019999841839881137,
-      "loss": 46.0,
-      "step": 11127
-    },
-    {
-      "epoch": 1.7920608720157816,
-      "grad_norm": 0.001457533217035234,
-      "learning_rate": 0.00019999841811426191,
-      "loss": 46.0,
-      "step": 11128
-    },
-    {
-      "epoch": 1.792221909094569,
-      "grad_norm": 0.004937728866934776,
-      "learning_rate": 0.00019999841782968687,
-      "loss": 46.0,
-      "step": 11129
-    },
-    {
-      "epoch": 1.7923829461733565,
-      "grad_norm": 0.0046903169713914394,
-      "learning_rate": 0.00019999841754508624,
-      "loss": 46.0,
-      "step": 11130
-    },
-    {
-      "epoch": 1.792543983252144,
-      "grad_norm": 0.0005806509288959205,
-      "learning_rate": 0.00019999841726046,
-      "loss": 46.0,
-      "step": 11131
-    },
-    {
-      "epoch": 1.7927050203309312,
-      "grad_norm": 0.0009940201416611671,
-      "learning_rate": 0.00019999841697580814,
-      "loss": 46.0,
-      "step": 11132
-    },
-    {
-      "epoch": 1.7928660574097186,
-      "grad_norm": 0.004427929874509573,
-      "learning_rate": 0.00019999841669113074,
-      "loss": 46.0,
-      "step": 11133
-    },
-    {
-      "epoch": 1.7930270944885058,
-      "grad_norm": 0.010929770767688751,
-      "learning_rate": 0.00019999841640642768,
-      "loss": 46.0,
-      "step": 11134
-    },
-    {
-      "epoch": 1.7931881315672933,
-      "grad_norm": 0.000417500821640715,
-      "learning_rate": 0.0001999984161216991,
-      "loss": 46.0,
-      "step": 11135
-    },
-    {
-      "epoch": 1.7933491686460807,
-      "grad_norm": 0.0029296651482582092,
-      "learning_rate": 0.00019999841583694488,
-      "loss": 46.0,
-      "step": 11136
-    },
-    {
-      "epoch": 1.7935102057248682,
-      "grad_norm": 0.006515244022011757,
-      "learning_rate": 0.00019999841555216508,
-      "loss": 46.0,
-      "step": 11137
-    },
-    {
-      "epoch": 1.7936712428036556,
-      "grad_norm": 0.004864226095378399,
-      "learning_rate": 0.00019999841526735965,
-      "loss": 46.0,
-      "step": 11138
-    },
-    {
-      "epoch": 1.793832279882443,
-      "grad_norm": 0.0009025073959492147,
-      "learning_rate": 0.00019999841498252868,
-      "loss": 46.0,
-      "step": 11139
-    },
-    {
-      "epoch": 1.7939933169612303,
-      "grad_norm": 0.0016181219834834337,
-      "learning_rate": 0.00019999841469767207,
-      "loss": 46.0,
-      "step": 11140
-    },
-    {
-      "epoch": 1.7941543540400176,
-      "grad_norm": 0.0027067612390965223,
-      "learning_rate": 0.0001999984144127899,
-      "loss": 46.0,
-      "step": 11141
-    },
-    {
-      "epoch": 1.794315391118805,
-      "grad_norm": 0.0003472942335065454,
-      "learning_rate": 0.0001999984141278821,
-      "loss": 46.0,
-      "step": 11142
-    },
-    {
-      "epoch": 1.7944764281975925,
-      "grad_norm": 0.0009301080135628581,
-      "learning_rate": 0.00019999841384294874,
-      "loss": 46.0,
-      "step": 11143
-    },
-    {
-      "epoch": 1.79463746527638,
-      "grad_norm": 0.0020068504381924868,
-      "learning_rate": 0.00019999841355798975,
-      "loss": 46.0,
-      "step": 11144
-    },
-    {
-      "epoch": 1.7947985023551674,
-      "grad_norm": 0.0008990314672701061,
-      "learning_rate": 0.0001999984132730052,
-      "loss": 46.0,
-      "step": 11145
-    },
-    {
-      "epoch": 1.7949595394339548,
-      "grad_norm": 0.005422293208539486,
-      "learning_rate": 0.00019999841298799504,
-      "loss": 46.0,
-      "step": 11146
-    },
-    {
-      "epoch": 1.795120576512742,
-      "grad_norm": 0.0032308408990502357,
-      "learning_rate": 0.00019999841270295932,
-      "loss": 46.0,
-      "step": 11147
-    },
-    {
-      "epoch": 1.7952816135915295,
-      "grad_norm": 0.002076471922919154,
-      "learning_rate": 0.00019999841241789796,
-      "loss": 46.0,
-      "step": 11148
-    },
-    {
-      "epoch": 1.7954426506703167,
-      "grad_norm": 0.012175084091722965,
-      "learning_rate": 0.00019999841213281104,
-      "loss": 46.0,
-      "step": 11149
-    },
-    {
-      "epoch": 1.7956036877491042,
-      "grad_norm": 0.0008831815794110298,
-      "learning_rate": 0.00019999841184769847,
-      "loss": 46.0,
-      "step": 11150
-    },
-    {
-      "epoch": 1.7957647248278916,
-      "grad_norm": 0.0016825733473524451,
-      "learning_rate": 0.00019999841156256035,
-      "loss": 46.0,
-      "step": 11151
-    },
-    {
-      "epoch": 1.795925761906679,
-      "grad_norm": 0.001760553102940321,
-      "learning_rate": 0.0001999984112773966,
-      "loss": 46.0,
-      "step": 11152
-    },
-    {
-      "epoch": 1.7960867989854665,
-      "grad_norm": 0.0017452806932851672,
-      "learning_rate": 0.00019999841099220734,
-      "loss": 46.0,
-      "step": 11153
-    },
-    {
-      "epoch": 1.7962478360642538,
-      "grad_norm": 0.0019896673038601875,
-      "learning_rate": 0.0001999984107069924,
-      "loss": 46.0,
-      "step": 11154
-    },
-    {
-      "epoch": 1.7964088731430412,
-      "grad_norm": 0.008706822991371155,
-      "learning_rate": 0.0001999984104217519,
-      "loss": 46.0,
-      "step": 11155
-    },
-    {
-      "epoch": 1.7965699102218284,
-      "grad_norm": 0.0017248884541913867,
-      "learning_rate": 0.0001999984101364858,
-      "loss": 46.0,
-      "step": 11156
-    },
-    {
-      "epoch": 1.7967309473006159,
-      "grad_norm": 0.001512069022282958,
-      "learning_rate": 0.00019999840985119409,
-      "loss": 46.0,
-      "step": 11157
-    },
-    {
-      "epoch": 1.7968919843794033,
-      "grad_norm": 0.002633160911500454,
-      "learning_rate": 0.0001999984095658768,
-      "loss": 46.0,
-      "step": 11158
-    },
-    {
-      "epoch": 1.7970530214581908,
-      "grad_norm": 0.004081765189766884,
-      "learning_rate": 0.00019999840928053392,
-      "loss": 46.0,
-      "step": 11159
-    },
-    {
-      "epoch": 1.7972140585369782,
-      "grad_norm": 0.0039664460346102715,
-      "learning_rate": 0.00019999840899516543,
-      "loss": 46.0,
-      "step": 11160
-    },
-    {
-      "epoch": 1.7973750956157657,
-      "grad_norm": 0.003466918133199215,
-      "learning_rate": 0.00019999840870977135,
-      "loss": 46.0,
-      "step": 11161
-    },
-    {
-      "epoch": 1.797536132694553,
-      "grad_norm": 0.0006607816321775317,
-      "learning_rate": 0.00019999840842435172,
-      "loss": 46.0,
-      "step": 11162
-    },
-    {
-      "epoch": 1.7976971697733402,
-      "grad_norm": 0.00045749059063382447,
-      "learning_rate": 0.00019999840813890644,
-      "loss": 46.0,
-      "step": 11163
-    },
-    {
-      "epoch": 1.7978582068521276,
-      "grad_norm": 0.0013399996096268296,
-      "learning_rate": 0.0001999984078534356,
-      "loss": 46.0,
-      "step": 11164
-    },
-    {
-      "epoch": 1.798019243930915,
-      "grad_norm": 0.0010656245285645127,
-      "learning_rate": 0.00019999840756793912,
-      "loss": 46.0,
-      "step": 11165
-    },
-    {
-      "epoch": 1.7981802810097025,
-      "grad_norm": 0.0008989977068267763,
-      "learning_rate": 0.00019999840728241708,
-      "loss": 46.0,
-      "step": 11166
-    },
-    {
-      "epoch": 1.79834131808849,
-      "grad_norm": 0.003475752891972661,
-      "learning_rate": 0.00019999840699686945,
-      "loss": 46.0,
-      "step": 11167
-    },
-    {
-      "epoch": 1.7985023551672774,
-      "grad_norm": 0.006849256344139576,
-      "learning_rate": 0.0001999984067112962,
-      "loss": 46.0,
-      "step": 11168
-    },
-    {
-      "epoch": 1.7986633922460646,
-      "grad_norm": 0.003347514197230339,
-      "learning_rate": 0.00019999840642569738,
-      "loss": 46.0,
-      "step": 11169
-    },
-    {
-      "epoch": 1.798824429324852,
-      "grad_norm": 0.0012060721637681127,
-      "learning_rate": 0.00019999840614007294,
-      "loss": 46.0,
-      "step": 11170
-    },
-    {
-      "epoch": 1.7989854664036393,
-      "grad_norm": 0.003787968773394823,
-      "learning_rate": 0.00019999840585442296,
-      "loss": 46.0,
-      "step": 11171
-    },
-    {
-      "epoch": 1.7991465034824268,
-      "grad_norm": 0.0014672133838757873,
-      "learning_rate": 0.00019999840556874735,
-      "loss": 46.0,
-      "step": 11172
-    },
-    {
-      "epoch": 1.7993075405612142,
-      "grad_norm": 0.00232651992700994,
-      "learning_rate": 0.00019999840528304611,
-      "loss": 46.0,
-      "step": 11173
-    },
-    {
-      "epoch": 1.7994685776400017,
-      "grad_norm": 0.0010064038215205073,
-      "learning_rate": 0.00019999840499731932,
-      "loss": 46.0,
-      "step": 11174
-    },
-    {
-      "epoch": 1.7996296147187891,
-      "grad_norm": 0.0007844401989132166,
-      "learning_rate": 0.00019999840471156692,
-      "loss": 46.0,
-      "step": 11175
-    },
-    {
-      "epoch": 1.7997906517975764,
-      "grad_norm": 0.00077760562999174,
-      "learning_rate": 0.00019999840442578895,
-      "loss": 46.0,
-      "step": 11176
-    },
-    {
-      "epoch": 1.7999516888763638,
-      "grad_norm": 0.0030735102482140064,
-      "learning_rate": 0.00019999840413998534,
-      "loss": 46.0,
-      "step": 11177
-    },
-    {
-      "epoch": 1.800112725955151,
-      "grad_norm": 0.0036428668536245823,
-      "learning_rate": 0.00019999840385415615,
-      "loss": 46.0,
-      "step": 11178
-    },
-    {
-      "epoch": 1.8002737630339385,
-      "grad_norm": 0.0007202904671430588,
-      "learning_rate": 0.0001999984035683014,
-      "loss": 46.0,
-      "step": 11179
-    },
-    {
-      "epoch": 1.800434800112726,
-      "grad_norm": 0.008539890870451927,
-      "learning_rate": 0.000199998403282421,
-      "loss": 46.0,
-      "step": 11180
-    },
-    {
-      "epoch": 1.8005958371915134,
-      "grad_norm": 0.008461201563477516,
-      "learning_rate": 0.00019999840299651507,
-      "loss": 46.0,
-      "step": 11181
-    },
-    {
-      "epoch": 1.8007568742703008,
-      "grad_norm": 0.004049209877848625,
-      "learning_rate": 0.0001999984027105835,
-      "loss": 46.0,
-      "step": 11182
-    },
-    {
-      "epoch": 1.800917911349088,
-      "grad_norm": 0.0008900587563402951,
-      "learning_rate": 0.00019999840242462634,
-      "loss": 46.0,
-      "step": 11183
-    },
-    {
-      "epoch": 1.8010789484278755,
-      "grad_norm": 0.0012589116813614964,
-      "learning_rate": 0.0001999984021386436,
-      "loss": 46.0,
-      "step": 11184
-    },
-    {
-      "epoch": 1.8012399855066628,
-      "grad_norm": 0.001192400697618723,
-      "learning_rate": 0.00019999840185263527,
-      "loss": 46.0,
-      "step": 11185
-    },
-    {
-      "epoch": 1.8014010225854502,
-      "grad_norm": 0.004492586944252253,
-      "learning_rate": 0.00019999840156660135,
-      "loss": 46.0,
-      "step": 11186
-    },
-    {
-      "epoch": 1.8015620596642377,
-      "grad_norm": 0.0010713412193581462,
-      "learning_rate": 0.00019999840128054181,
-      "loss": 46.0,
-      "step": 11187
-    },
-    {
-      "epoch": 1.801723096743025,
-      "grad_norm": 0.004886297509074211,
-      "learning_rate": 0.0001999984009944567,
-      "loss": 46.0,
-      "step": 11188
-    },
-    {
-      "epoch": 1.8018841338218126,
-      "grad_norm": 0.00322319520637393,
-      "learning_rate": 0.00019999840070834596,
-      "loss": 46.0,
-      "step": 11189
-    },
-    {
-      "epoch": 1.8020451709006,
-      "grad_norm": 0.0016112119192257524,
-      "learning_rate": 0.00019999840042220966,
-      "loss": 46.0,
-      "step": 11190
-    },
-    {
-      "epoch": 1.8022062079793872,
-      "grad_norm": 0.0011197234271094203,
-      "learning_rate": 0.00019999840013604773,
-      "loss": 46.0,
-      "step": 11191
-    },
-    {
-      "epoch": 1.8023672450581747,
-      "grad_norm": 0.0025336844846606255,
-      "learning_rate": 0.00019999839984986023,
-      "loss": 46.0,
-      "step": 11192
-    },
-    {
-      "epoch": 1.802528282136962,
-      "grad_norm": 0.0019435323774814606,
-      "learning_rate": 0.00019999839956364715,
-      "loss": 46.0,
-      "step": 11193
-    },
-    {
-      "epoch": 1.8026893192157494,
-      "grad_norm": 0.005620239768177271,
-      "learning_rate": 0.00019999839927740848,
-      "loss": 46.0,
-      "step": 11194
-    },
-    {
-      "epoch": 1.8028503562945368,
-      "grad_norm": 0.006061875261366367,
-      "learning_rate": 0.00019999839899114417,
-      "loss": 46.0,
-      "step": 11195
-    },
-    {
-      "epoch": 1.8030113933733243,
-      "grad_norm": 0.0011227671056985855,
-      "learning_rate": 0.0001999983987048543,
-      "loss": 46.0,
-      "step": 11196
-    },
-    {
-      "epoch": 1.8031724304521117,
-      "grad_norm": 0.00596654461696744,
-      "learning_rate": 0.00019999839841853883,
-      "loss": 46.0,
-      "step": 11197
-    },
-    {
-      "epoch": 1.803333467530899,
-      "grad_norm": 0.0009404318989254534,
-      "learning_rate": 0.00019999839813219773,
-      "loss": 46.0,
-      "step": 11198
-    },
-    {
-      "epoch": 1.8034945046096864,
-      "grad_norm": 0.0018894915701821446,
-      "learning_rate": 0.00019999839784583107,
-      "loss": 46.0,
-      "step": 11199
-    },
-    {
-      "epoch": 1.8036555416884736,
-      "grad_norm": 0.0013029336696490645,
-      "learning_rate": 0.00019999839755943883,
-      "loss": 46.0,
-      "step": 11200
-    },
-    {
-      "epoch": 1.803816578767261,
-      "grad_norm": 0.006391707342118025,
-      "learning_rate": 0.00019999839727302096,
-      "loss": 46.0,
-      "step": 11201
-    },
-    {
-      "epoch": 1.8039776158460485,
-      "grad_norm": 0.001937276916578412,
-      "learning_rate": 0.00019999839698657751,
-      "loss": 46.0,
-      "step": 11202
-    },
-    {
-      "epoch": 1.804138652924836,
-      "grad_norm": 0.004002357367426157,
-      "learning_rate": 0.0001999983967001085,
-      "loss": 46.0,
-      "step": 11203
-    },
-    {
-      "epoch": 1.8042996900036234,
-      "grad_norm": 0.0018983158515766263,
-      "learning_rate": 0.00019999839641361385,
-      "loss": 46.0,
-      "step": 11204
-    },
-    {
-      "epoch": 1.8044607270824107,
-      "grad_norm": 0.0010678655235096812,
-      "learning_rate": 0.00019999839612709362,
-      "loss": 46.0,
-      "step": 11205
-    },
-    {
-      "epoch": 1.8046217641611981,
-      "grad_norm": 0.002290706615895033,
-      "learning_rate": 0.0001999983958405478,
-      "loss": 46.0,
-      "step": 11206
-    },
-    {
-      "epoch": 1.8047828012399854,
-      "grad_norm": 0.0022672186605632305,
-      "learning_rate": 0.00019999839555397638,
-      "loss": 46.0,
-      "step": 11207
-    },
-    {
-      "epoch": 1.8049438383187728,
-      "grad_norm": 0.007677840534597635,
-      "learning_rate": 0.00019999839526737935,
-      "loss": 46.0,
-      "step": 11208
-    },
-    {
-      "epoch": 1.8051048753975603,
-      "grad_norm": 0.003321537747979164,
-      "learning_rate": 0.00019999839498075677,
-      "loss": 46.0,
-      "step": 11209
-    },
-    {
-      "epoch": 1.8052659124763477,
-      "grad_norm": 0.0005885874852538109,
-      "learning_rate": 0.00019999839469410854,
-      "loss": 46.0,
-      "step": 11210
-    },
-    {
-      "epoch": 1.8054269495551352,
-      "grad_norm": 0.0008507912280037999,
-      "learning_rate": 0.00019999839440743475,
-      "loss": 46.0,
-      "step": 11211
-    },
-    {
-      "epoch": 1.8055879866339226,
-      "grad_norm": 0.009996253065764904,
-      "learning_rate": 0.00019999839412073537,
-      "loss": 46.0,
-      "step": 11212
-    },
-    {
-      "epoch": 1.8057490237127098,
-      "grad_norm": 0.0023658068384975195,
-      "learning_rate": 0.00019999839383401039,
-      "loss": 46.0,
-      "step": 11213
-    },
-    {
-      "epoch": 1.8059100607914973,
-      "grad_norm": 0.0008899774984456599,
-      "learning_rate": 0.0001999983935472598,
-      "loss": 46.0,
-      "step": 11214
-    },
-    {
-      "epoch": 1.8060710978702845,
-      "grad_norm": 0.0015827525639906526,
-      "learning_rate": 0.00019999839326048362,
-      "loss": 46.0,
-      "step": 11215
-    },
-    {
-      "epoch": 1.806232134949072,
-      "grad_norm": 0.0017011427553370595,
-      "learning_rate": 0.00019999839297368187,
-      "loss": 46.0,
-      "step": 11216
-    },
-    {
-      "epoch": 1.8063931720278594,
-      "grad_norm": 0.0013393504777923226,
-      "learning_rate": 0.0001999983926868545,
-      "loss": 46.0,
-      "step": 11217
-    },
-    {
-      "epoch": 1.8065542091066469,
-      "grad_norm": 0.000996196991764009,
-      "learning_rate": 0.00019999839240000152,
-      "loss": 46.0,
-      "step": 11218
-    },
-    {
-      "epoch": 1.8067152461854343,
-      "grad_norm": 0.004124980419874191,
-      "learning_rate": 0.00019999839211312298,
-      "loss": 46.0,
-      "step": 11219
-    },
-    {
-      "epoch": 1.8068762832642216,
-      "grad_norm": 0.0042527890764176846,
-      "learning_rate": 0.00019999839182621886,
-      "loss": 46.0,
-      "step": 11220
-    },
-    {
-      "epoch": 1.807037320343009,
-      "grad_norm": 0.001644301344640553,
-      "learning_rate": 0.0001999983915392891,
-      "loss": 46.0,
-      "step": 11221
-    },
-    {
-      "epoch": 1.8071983574217962,
-      "grad_norm": 0.006192200351506472,
-      "learning_rate": 0.00019999839125233376,
-      "loss": 46.0,
-      "step": 11222
-    },
-    {
-      "epoch": 1.8073593945005837,
-      "grad_norm": 0.010850329883396626,
-      "learning_rate": 0.00019999839096535282,
-      "loss": 46.0,
-      "step": 11223
-    },
-    {
-      "epoch": 1.8075204315793711,
-      "grad_norm": 0.0019111887086182833,
-      "learning_rate": 0.0001999983906783463,
-      "loss": 46.0,
-      "step": 11224
-    },
-    {
-      "epoch": 1.8076814686581586,
-      "grad_norm": 0.0009883996099233627,
-      "learning_rate": 0.00019999839039131417,
-      "loss": 46.0,
-      "step": 11225
-    },
-    {
-      "epoch": 1.807842505736946,
-      "grad_norm": 0.001829432207159698,
-      "learning_rate": 0.00019999839010425647,
-      "loss": 46.0,
-      "step": 11226
-    },
-    {
-      "epoch": 1.8080035428157333,
-      "grad_norm": 0.0009486796916462481,
-      "learning_rate": 0.00019999838981717315,
-      "loss": 46.0,
-      "step": 11227
-    },
-    {
-      "epoch": 1.8081645798945207,
-      "grad_norm": 0.007451027166098356,
-      "learning_rate": 0.00019999838953006424,
-      "loss": 46.0,
-      "step": 11228
-    },
-    {
-      "epoch": 1.808325616973308,
-      "grad_norm": 0.0038440402131527662,
-      "learning_rate": 0.00019999838924292975,
-      "loss": 46.0,
-      "step": 11229
-    },
-    {
-      "epoch": 1.8084866540520954,
-      "grad_norm": 0.0021127837244421244,
-      "learning_rate": 0.00019999838895576967,
-      "loss": 46.0,
-      "step": 11230
-    },
-    {
-      "epoch": 1.8086476911308829,
-      "grad_norm": 0.0008059654501266778,
-      "learning_rate": 0.00019999838866858398,
-      "loss": 46.0,
-      "step": 11231
-    },
-    {
-      "epoch": 1.8088087282096703,
-      "grad_norm": 0.0007508113048970699,
-      "learning_rate": 0.0001999983883813727,
-      "loss": 46.0,
-      "step": 11232
-    },
-    {
-      "epoch": 1.8089697652884578,
-      "grad_norm": 0.0016701332060620189,
-      "learning_rate": 0.00019999838809413585,
-      "loss": 46.0,
-      "step": 11233
-    },
-    {
-      "epoch": 1.8091308023672452,
-      "grad_norm": 0.0005689797108061612,
-      "learning_rate": 0.00019999838780687337,
-      "loss": 46.0,
-      "step": 11234
-    },
-    {
-      "epoch": 1.8092918394460324,
-      "grad_norm": 0.0019886535592377186,
-      "learning_rate": 0.0001999983875195853,
-      "loss": 46.0,
-      "step": 11235
-    },
-    {
-      "epoch": 1.8094528765248197,
-      "grad_norm": 0.003964284434914589,
-      "learning_rate": 0.00019999838723227164,
-      "loss": 46.0,
-      "step": 11236
-    },
-    {
-      "epoch": 1.8096139136036071,
-      "grad_norm": 0.002529874909669161,
-      "learning_rate": 0.0001999983869449324,
-      "loss": 46.0,
-      "step": 11237
-    },
-    {
-      "epoch": 1.8097749506823946,
-      "grad_norm": 0.0005866981227882206,
-      "learning_rate": 0.00019999838665756754,
-      "loss": 46.0,
-      "step": 11238
-    },
-    {
-      "epoch": 1.809935987761182,
-      "grad_norm": 0.0005656818393617868,
-      "learning_rate": 0.0001999983863701771,
-      "loss": 46.0,
-      "step": 11239
-    },
-    {
-      "epoch": 1.8100970248399695,
-      "grad_norm": 0.0052567641250789165,
-      "learning_rate": 0.00019999838608276106,
-      "loss": 46.0,
-      "step": 11240
-    },
-    {
-      "epoch": 1.810258061918757,
-      "grad_norm": 0.0016020999755710363,
-      "learning_rate": 0.00019999838579531942,
-      "loss": 46.0,
-      "step": 11241
-    },
-    {
-      "epoch": 1.8104190989975442,
-      "grad_norm": 0.012540334835648537,
-      "learning_rate": 0.0001999983855078522,
-      "loss": 46.0,
-      "step": 11242
-    },
-    {
-      "epoch": 1.8105801360763316,
-      "grad_norm": 0.005992401856929064,
-      "learning_rate": 0.0001999983852203594,
-      "loss": 46.0,
-      "step": 11243
-    },
-    {
-      "epoch": 1.8107411731551188,
-      "grad_norm": 0.0042040590196847916,
-      "learning_rate": 0.00019999838493284098,
-      "loss": 46.0,
-      "step": 11244
-    },
-    {
-      "epoch": 1.8109022102339063,
-      "grad_norm": 0.0014169916976243258,
-      "learning_rate": 0.00019999838464529696,
-      "loss": 46.0,
-      "step": 11245
-    },
-    {
-      "epoch": 1.8110632473126937,
-      "grad_norm": 0.00492845568805933,
-      "learning_rate": 0.00019999838435772737,
-      "loss": 46.0,
-      "step": 11246
-    },
-    {
-      "epoch": 1.8112242843914812,
-      "grad_norm": 0.0024931305088102818,
-      "learning_rate": 0.00019999838407013218,
-      "loss": 46.0,
-      "step": 11247
-    },
-    {
-      "epoch": 1.8113853214702687,
-      "grad_norm": 0.0008138782577589154,
-      "learning_rate": 0.00019999838378251136,
-      "loss": 46.0,
-      "step": 11248
-    },
-    {
-      "epoch": 1.8115463585490559,
-      "grad_norm": 0.0016083779046311975,
-      "learning_rate": 0.000199998383494865,
-      "loss": 46.0,
-      "step": 11249
-    },
-    {
-      "epoch": 1.8117073956278433,
-      "grad_norm": 0.0009699150687083602,
-      "learning_rate": 0.000199998383207193,
-      "loss": 46.0,
-      "step": 11250
-    },
-    {
-      "epoch": 1.8118684327066306,
-      "grad_norm": 0.0062147765420377254,
-      "learning_rate": 0.00019999838291949543,
-      "loss": 46.0,
-      "step": 11251
-    },
-    {
-      "epoch": 1.812029469785418,
-      "grad_norm": 0.0003888392820954323,
-      "learning_rate": 0.00019999838263177227,
-      "loss": 46.0,
-      "step": 11252
-    },
-    {
-      "epoch": 1.8121905068642055,
-      "grad_norm": 0.0010161014506593347,
-      "learning_rate": 0.0001999983823440235,
-      "loss": 46.0,
-      "step": 11253
-    },
-    {
-      "epoch": 1.812351543942993,
-      "grad_norm": 0.0005654977867379785,
-      "learning_rate": 0.00019999838205624914,
-      "loss": 46.0,
-      "step": 11254
-    },
-    {
-      "epoch": 1.8125125810217804,
-      "grad_norm": 0.00197664275765419,
-      "learning_rate": 0.00019999838176844922,
-      "loss": 46.0,
-      "step": 11255
-    },
-    {
-      "epoch": 1.8126736181005678,
-      "grad_norm": 0.001464974251575768,
-      "learning_rate": 0.00019999838148062365,
-      "loss": 46.0,
-      "step": 11256
-    },
-    {
-      "epoch": 1.812834655179355,
-      "grad_norm": 0.0038541234098374844,
-      "learning_rate": 0.0001999983811927725,
-      "loss": 46.0,
-      "step": 11257
-    },
-    {
-      "epoch": 1.8129956922581423,
-      "grad_norm": 0.0004624953435268253,
-      "learning_rate": 0.00019999838090489576,
-      "loss": 46.0,
-      "step": 11258
-    },
-    {
-      "epoch": 1.8131567293369297,
-      "grad_norm": 0.005003497004508972,
-      "learning_rate": 0.00019999838061699344,
-      "loss": 46.0,
-      "step": 11259
-    },
-    {
-      "epoch": 1.8133177664157172,
-      "grad_norm": 0.0018303110264241695,
-      "learning_rate": 0.00019999838032906553,
-      "loss": 46.0,
-      "step": 11260
-    },
-    {
-      "epoch": 1.8134788034945046,
-      "grad_norm": 0.000560486689209938,
-      "learning_rate": 0.000199998380041112,
-      "loss": 46.0,
-      "step": 11261
-    },
-    {
-      "epoch": 1.813639840573292,
-      "grad_norm": 0.0006303409463725984,
-      "learning_rate": 0.0001999983797531329,
-      "loss": 46.0,
-      "step": 11262
-    },
-    {
-      "epoch": 1.8138008776520795,
-      "grad_norm": 0.000938627403229475,
-      "learning_rate": 0.00019999837946512816,
-      "loss": 46.0,
-      "step": 11263
-    },
-    {
-      "epoch": 1.8139619147308668,
-      "grad_norm": 0.0007530644070357084,
-      "learning_rate": 0.00019999837917709787,
-      "loss": 46.0,
-      "step": 11264
-    },
-    {
-      "epoch": 1.8141229518096542,
-      "grad_norm": 0.00154742575250566,
-      "learning_rate": 0.00019999837888904197,
-      "loss": 46.0,
-      "step": 11265
-    },
-    {
-      "epoch": 1.8142839888884414,
-      "grad_norm": 0.0017158687114715576,
-      "learning_rate": 0.00019999837860096048,
-      "loss": 46.0,
-      "step": 11266
-    },
-    {
-      "epoch": 1.814445025967229,
-      "grad_norm": 0.0024674772284924984,
-      "learning_rate": 0.0001999983783128534,
-      "loss": 46.0,
-      "step": 11267
-    },
-    {
-      "epoch": 1.8146060630460163,
-      "grad_norm": 0.0018584816716611385,
-      "learning_rate": 0.00019999837802472072,
-      "loss": 46.0,
-      "step": 11268
-    },
-    {
-      "epoch": 1.8147671001248038,
-      "grad_norm": 0.002340343315154314,
-      "learning_rate": 0.00019999837773656244,
-      "loss": 46.0,
-      "step": 11269
-    },
-    {
-      "epoch": 1.8149281372035913,
-      "grad_norm": 0.001429402851499617,
-      "learning_rate": 0.00019999837744837857,
-      "loss": 46.0,
-      "step": 11270
-    },
-    {
-      "epoch": 1.8150891742823785,
-      "grad_norm": 0.0009373221546411514,
-      "learning_rate": 0.0001999983771601691,
-      "loss": 46.0,
-      "step": 11271
-    },
-    {
-      "epoch": 1.815250211361166,
-      "grad_norm": 0.00586380809545517,
-      "learning_rate": 0.00019999837687193406,
-      "loss": 46.0,
-      "step": 11272
-    },
-    {
-      "epoch": 1.8154112484399532,
-      "grad_norm": 0.0020879872608929873,
-      "learning_rate": 0.0001999983765836734,
-      "loss": 46.0,
-      "step": 11273
-    },
-    {
-      "epoch": 1.8155722855187406,
-      "grad_norm": 0.0012953775003552437,
-      "learning_rate": 0.00019999837629538716,
-      "loss": 46.0,
-      "step": 11274
-    },
-    {
-      "epoch": 1.815733322597528,
-      "grad_norm": 0.0011106071760877967,
-      "learning_rate": 0.0001999983760070753,
-      "loss": 46.0,
-      "step": 11275
-    },
-    {
-      "epoch": 1.8158943596763155,
-      "grad_norm": 0.0014667804352939129,
-      "learning_rate": 0.00019999837571873784,
-      "loss": 46.0,
-      "step": 11276
-    },
-    {
-      "epoch": 1.816055396755103,
-      "grad_norm": 0.000782331801019609,
-      "learning_rate": 0.00019999837543037484,
-      "loss": 46.0,
-      "step": 11277
-    },
-    {
-      "epoch": 1.8162164338338902,
-      "grad_norm": 0.001211573020555079,
-      "learning_rate": 0.00019999837514198622,
-      "loss": 46.0,
-      "step": 11278
-    },
-    {
-      "epoch": 1.8163774709126776,
-      "grad_norm": 0.02746388502418995,
-      "learning_rate": 0.000199998374853572,
-      "loss": 46.0,
-      "step": 11279
-    },
-    {
-      "epoch": 1.8165385079914649,
-      "grad_norm": 0.0011048989836126566,
-      "learning_rate": 0.00019999837456513217,
-      "loss": 46.0,
-      "step": 11280
-    },
-    {
-      "epoch": 1.8166995450702523,
-      "grad_norm": 0.0009253582684323192,
-      "learning_rate": 0.00019999837427666677,
-      "loss": 46.0,
-      "step": 11281
-    },
-    {
-      "epoch": 1.8168605821490398,
-      "grad_norm": 0.0002986912149935961,
-      "learning_rate": 0.00019999837398817575,
-      "loss": 46.0,
-      "step": 11282
-    },
-    {
-      "epoch": 1.8170216192278272,
-      "grad_norm": 0.0059528895653784275,
-      "learning_rate": 0.00019999837369965917,
-      "loss": 46.0,
-      "step": 11283
-    },
-    {
-      "epoch": 1.8171826563066147,
-      "grad_norm": 0.0021651028655469418,
-      "learning_rate": 0.00019999837341111697,
-      "loss": 46.0,
-      "step": 11284
-    },
-    {
-      "epoch": 1.8173436933854021,
-      "grad_norm": 0.002069145906716585,
-      "learning_rate": 0.0001999983731225492,
-      "loss": 46.0,
-      "step": 11285
-    },
-    {
-      "epoch": 1.8175047304641894,
-      "grad_norm": 0.0011010889429599047,
-      "learning_rate": 0.0001999983728339558,
-      "loss": 46.0,
-      "step": 11286
-    },
-    {
-      "epoch": 1.8176657675429768,
-      "grad_norm": 0.0021088651847094297,
-      "learning_rate": 0.00019999837254533684,
-      "loss": 46.0,
-      "step": 11287
-    },
-    {
-      "epoch": 1.817826804621764,
-      "grad_norm": 0.0018376075895503163,
-      "learning_rate": 0.00019999837225669227,
-      "loss": 46.0,
-      "step": 11288
-    },
-    {
-      "epoch": 1.8179878417005515,
-      "grad_norm": 0.003436165861785412,
-      "learning_rate": 0.00019999837196802212,
-      "loss": 46.0,
-      "step": 11289
-    },
-    {
-      "epoch": 1.818148878779339,
-      "grad_norm": 0.001971396617591381,
-      "learning_rate": 0.00019999837167932635,
-      "loss": 46.0,
-      "step": 11290
-    },
-    {
-      "epoch": 1.8183099158581264,
-      "grad_norm": 0.0022066179662942886,
-      "learning_rate": 0.000199998371390605,
-      "loss": 46.0,
-      "step": 11291
-    },
-    {
-      "epoch": 1.8184709529369139,
-      "grad_norm": 0.0029349513351917267,
-      "learning_rate": 0.00019999837110185804,
-      "loss": 46.0,
-      "step": 11292
-    },
-    {
-      "epoch": 1.818631990015701,
-      "grad_norm": 0.0008590839570388198,
-      "learning_rate": 0.0001999983708130855,
-      "loss": 46.0,
-      "step": 11293
-    },
-    {
-      "epoch": 1.8187930270944885,
-      "grad_norm": 0.0011101976269856095,
-      "learning_rate": 0.00019999837052428736,
-      "loss": 46.0,
-      "step": 11294
-    },
-    {
-      "epoch": 1.8189540641732758,
-      "grad_norm": 0.0019354196265339851,
-      "learning_rate": 0.00019999837023546363,
-      "loss": 46.0,
-      "step": 11295
-    },
-    {
-      "epoch": 1.8191151012520632,
-      "grad_norm": 0.0041565727442502975,
-      "learning_rate": 0.0001999983699466143,
-      "loss": 46.0,
-      "step": 11296
-    },
-    {
-      "epoch": 1.8192761383308507,
-      "grad_norm": 0.0008438201039098203,
-      "learning_rate": 0.0001999983696577394,
-      "loss": 46.0,
-      "step": 11297
-    },
-    {
-      "epoch": 1.8194371754096381,
-      "grad_norm": 0.000578771170694381,
-      "learning_rate": 0.00019999836936883888,
-      "loss": 46.0,
-      "step": 11298
-    },
-    {
-      "epoch": 1.8195982124884256,
-      "grad_norm": 0.006573663093149662,
-      "learning_rate": 0.00019999836907991277,
-      "loss": 46.0,
-      "step": 11299
-    },
-    {
-      "epoch": 1.8197592495672128,
-      "grad_norm": 0.0014425547560676932,
-      "learning_rate": 0.00019999836879096105,
-      "loss": 46.0,
-      "step": 11300
-    },
-    {
-      "epoch": 1.8199202866460003,
-      "grad_norm": 0.002111180918291211,
-      "learning_rate": 0.00019999836850198376,
-      "loss": 46.0,
-      "step": 11301
-    },
-    {
-      "epoch": 1.8200813237247875,
-      "grad_norm": 0.000962501741014421,
-      "learning_rate": 0.00019999836821298086,
-      "loss": 46.0,
-      "step": 11302
-    },
-    {
-      "epoch": 1.820242360803575,
-      "grad_norm": 0.0017520018154755235,
-      "learning_rate": 0.00019999836792395238,
-      "loss": 46.0,
-      "step": 11303
-    },
-    {
-      "epoch": 1.8204033978823624,
-      "grad_norm": 0.003518523881211877,
-      "learning_rate": 0.0001999983676348983,
-      "loss": 46.0,
-      "step": 11304
-    },
-    {
-      "epoch": 1.8205644349611498,
-      "grad_norm": 0.005110064055770636,
-      "learning_rate": 0.00019999836734581862,
-      "loss": 46.0,
-      "step": 11305
-    },
-    {
-      "epoch": 1.8207254720399373,
-      "grad_norm": 0.0010770312510430813,
-      "learning_rate": 0.00019999836705671337,
-      "loss": 46.0,
-      "step": 11306
-    },
-    {
-      "epoch": 1.8208865091187247,
-      "grad_norm": 0.003938900772482157,
-      "learning_rate": 0.00019999836676758248,
-      "loss": 46.0,
-      "step": 11307
-    },
-    {
-      "epoch": 1.821047546197512,
-      "grad_norm": 0.004902741406112909,
-      "learning_rate": 0.00019999836647842603,
-      "loss": 46.0,
-      "step": 11308
-    },
-    {
-      "epoch": 1.8212085832762994,
-      "grad_norm": 0.001318879658356309,
-      "learning_rate": 0.000199998366189244,
-      "loss": 46.0,
-      "step": 11309
-    },
-    {
-      "epoch": 1.8213696203550866,
-      "grad_norm": 0.0043305763974785805,
-      "learning_rate": 0.00019999836590003632,
-      "loss": 46.0,
-      "step": 11310
-    },
-    {
-      "epoch": 1.821530657433874,
-      "grad_norm": 0.008512002415955067,
-      "learning_rate": 0.00019999836561080309,
-      "loss": 46.0,
-      "step": 11311
-    },
-    {
-      "epoch": 1.8216916945126616,
-      "grad_norm": 0.0012935369741171598,
-      "learning_rate": 0.00019999836532154426,
-      "loss": 46.0,
-      "step": 11312
-    },
-    {
-      "epoch": 1.821852731591449,
-      "grad_norm": 0.0015662142541259527,
-      "learning_rate": 0.00019999836503225985,
-      "loss": 46.0,
-      "step": 11313
-    },
-    {
-      "epoch": 1.8220137686702365,
-      "grad_norm": 0.005370223894715309,
-      "learning_rate": 0.0001999983647429498,
-      "loss": 46.0,
-      "step": 11314
-    },
-    {
-      "epoch": 1.8221748057490237,
-      "grad_norm": 0.003937610890716314,
-      "learning_rate": 0.00019999836445361418,
-      "loss": 46.0,
-      "step": 11315
-    },
-    {
-      "epoch": 1.8223358428278111,
-      "grad_norm": 0.000888905196916312,
-      "learning_rate": 0.00019999836416425298,
-      "loss": 46.0,
-      "step": 11316
-    },
-    {
-      "epoch": 1.8224968799065984,
-      "grad_norm": 0.0008690909016877413,
-      "learning_rate": 0.00019999836387486614,
-      "loss": 46.0,
-      "step": 11317
-    },
-    {
-      "epoch": 1.8226579169853858,
-      "grad_norm": 0.005930494517087936,
-      "learning_rate": 0.00019999836358545374,
-      "loss": 46.0,
-      "step": 11318
-    },
-    {
-      "epoch": 1.8228189540641733,
-      "grad_norm": 0.0005782339721918106,
-      "learning_rate": 0.00019999836329601575,
-      "loss": 46.0,
-      "step": 11319
-    },
-    {
-      "epoch": 1.8229799911429607,
-      "grad_norm": 0.006285279057919979,
-      "learning_rate": 0.00019999836300655218,
-      "loss": 46.0,
-      "step": 11320
-    },
-    {
-      "epoch": 1.8231410282217482,
-      "grad_norm": 0.0019795040134340525,
-      "learning_rate": 0.000199998362717063,
-      "loss": 46.0,
-      "step": 11321
-    },
-    {
-      "epoch": 1.8233020653005354,
-      "grad_norm": 0.0017771889688447118,
-      "learning_rate": 0.00019999836242754818,
-      "loss": 46.0,
-      "step": 11322
-    },
-    {
-      "epoch": 1.8234631023793229,
-      "grad_norm": 0.0016590761952102184,
-      "learning_rate": 0.00019999836213800782,
-      "loss": 46.0,
-      "step": 11323
-    },
-    {
-      "epoch": 1.82362413945811,
-      "grad_norm": 0.0014575679088011384,
-      "learning_rate": 0.00019999836184844184,
-      "loss": 46.0,
-      "step": 11324
-    },
-    {
-      "epoch": 1.8237851765368975,
-      "grad_norm": 0.0008977229590527713,
-      "learning_rate": 0.00019999836155885027,
-      "loss": 46.0,
-      "step": 11325
-    },
-    {
-      "epoch": 1.823946213615685,
-      "grad_norm": 0.010617556981742382,
-      "learning_rate": 0.00019999836126923312,
-      "loss": 46.0,
-      "step": 11326
-    },
-    {
-      "epoch": 1.8241072506944724,
-      "grad_norm": 0.0016906767850741744,
-      "learning_rate": 0.00019999836097959038,
-      "loss": 46.0,
-      "step": 11327
-    },
-    {
-      "epoch": 1.8242682877732599,
-      "grad_norm": 0.0035667498596012592,
-      "learning_rate": 0.00019999836068992203,
-      "loss": 46.0,
-      "step": 11328
-    },
-    {
-      "epoch": 1.8244293248520473,
-      "grad_norm": 0.009043061174452305,
-      "learning_rate": 0.0001999983604002281,
-      "loss": 46.0,
-      "step": 11329
-    },
-    {
-      "epoch": 1.8245903619308346,
-      "grad_norm": 0.007218494080007076,
-      "learning_rate": 0.00019999836011050856,
-      "loss": 46.0,
-      "step": 11330
-    },
-    {
-      "epoch": 1.8247513990096218,
-      "grad_norm": 0.0045862761326134205,
-      "learning_rate": 0.0001999983598207634,
-      "loss": 46.0,
-      "step": 11331
-    },
-    {
-      "epoch": 1.8249124360884093,
-      "grad_norm": 0.0011916913790628314,
-      "learning_rate": 0.0001999983595309927,
-      "loss": 46.0,
-      "step": 11332
-    },
-    {
-      "epoch": 1.8250734731671967,
-      "grad_norm": 0.0031784556340426207,
-      "learning_rate": 0.0001999983592411964,
-      "loss": 46.0,
-      "step": 11333
-    },
-    {
-      "epoch": 1.8252345102459842,
-      "grad_norm": 0.000653648457955569,
-      "learning_rate": 0.00019999835895137443,
-      "loss": 46.0,
-      "step": 11334
-    },
-    {
-      "epoch": 1.8253955473247716,
-      "grad_norm": 0.0008914176723919809,
-      "learning_rate": 0.00019999835866152697,
-      "loss": 46.0,
-      "step": 11335
-    },
-    {
-      "epoch": 1.825556584403559,
-      "grad_norm": 0.000840010994579643,
-      "learning_rate": 0.00019999835837165386,
-      "loss": 46.0,
-      "step": 11336
-    },
-    {
-      "epoch": 1.8257176214823463,
-      "grad_norm": 0.004974375944584608,
-      "learning_rate": 0.00019999835808175514,
-      "loss": 46.0,
-      "step": 11337
-    },
-    {
-      "epoch": 1.8258786585611337,
-      "grad_norm": 0.0030716145411133766,
-      "learning_rate": 0.00019999835779183086,
-      "loss": 46.0,
-      "step": 11338
-    },
-    {
-      "epoch": 1.826039695639921,
-      "grad_norm": 0.0022781628649681807,
-      "learning_rate": 0.000199998357501881,
-      "loss": 46.0,
-      "step": 11339
-    },
-    {
-      "epoch": 1.8262007327187084,
-      "grad_norm": 0.0011669025989249349,
-      "learning_rate": 0.0001999983572119055,
-      "loss": 46.0,
-      "step": 11340
-    },
-    {
-      "epoch": 1.8263617697974959,
-      "grad_norm": 0.003255236893892288,
-      "learning_rate": 0.00019999835692190442,
-      "loss": 46.0,
-      "step": 11341
-    },
-    {
-      "epoch": 1.8265228068762833,
-      "grad_norm": 0.0025436037685722113,
-      "learning_rate": 0.00019999835663187776,
-      "loss": 46.0,
-      "step": 11342
-    },
-    {
-      "epoch": 1.8266838439550708,
-      "grad_norm": 0.0008401311351917684,
-      "learning_rate": 0.00019999835634182546,
-      "loss": 46.0,
-      "step": 11343
-    },
-    {
-      "epoch": 1.826844881033858,
-      "grad_norm": 0.02161703258752823,
-      "learning_rate": 0.0001999983560517476,
-      "loss": 46.0,
-      "step": 11344
-    },
-    {
-      "epoch": 1.8270059181126455,
-      "grad_norm": 0.0020726663060486317,
-      "learning_rate": 0.00019999835576164416,
-      "loss": 46.0,
-      "step": 11345
-    },
-    {
-      "epoch": 1.8271669551914327,
-      "grad_norm": 0.004923627711832523,
-      "learning_rate": 0.0001999983554715151,
-      "loss": 46.0,
-      "step": 11346
-    },
-    {
-      "epoch": 1.8273279922702201,
-      "grad_norm": 0.0007343343459069729,
-      "learning_rate": 0.00019999835518136048,
-      "loss": 46.0,
-      "step": 11347
-    },
-    {
-      "epoch": 1.8274890293490076,
-      "grad_norm": 0.000806807482149452,
-      "learning_rate": 0.00019999835489118025,
-      "loss": 46.0,
-      "step": 11348
-    },
-    {
-      "epoch": 1.827650066427795,
-      "grad_norm": 0.002343647414818406,
-      "learning_rate": 0.0001999983546009744,
-      "loss": 46.0,
-      "step": 11349
-    },
-    {
-      "epoch": 1.8278111035065825,
-      "grad_norm": 0.0006349949981085956,
-      "learning_rate": 0.000199998354310743,
-      "loss": 46.0,
-      "step": 11350
-    },
-    {
-      "epoch": 1.82797214058537,
-      "grad_norm": 0.002084990032017231,
-      "learning_rate": 0.00019999835402048597,
-      "loss": 46.0,
-      "step": 11351
-    },
-    {
-      "epoch": 1.8281331776641572,
-      "grad_norm": 0.0035368534736335278,
-      "learning_rate": 0.00019999835373020334,
-      "loss": 46.0,
-      "step": 11352
-    },
-    {
-      "epoch": 1.8282942147429444,
-      "grad_norm": 0.005981109105050564,
-      "learning_rate": 0.0001999983534398951,
-      "loss": 46.0,
-      "step": 11353
-    },
-    {
-      "epoch": 1.8284552518217319,
-      "grad_norm": 0.0016671504126861691,
-      "learning_rate": 0.00019999835314956133,
-      "loss": 46.0,
-      "step": 11354
-    },
-    {
-      "epoch": 1.8286162889005193,
-      "grad_norm": 0.004126428160816431,
-      "learning_rate": 0.00019999835285920193,
-      "loss": 46.0,
-      "step": 11355
-    },
-    {
-      "epoch": 1.8287773259793068,
-      "grad_norm": 0.001696393359452486,
-      "learning_rate": 0.00019999835256881695,
-      "loss": 46.0,
-      "step": 11356
-    },
-    {
-      "epoch": 1.8289383630580942,
-      "grad_norm": 0.0052888840436935425,
-      "learning_rate": 0.00019999835227840635,
-      "loss": 46.0,
-      "step": 11357
-    },
-    {
-      "epoch": 1.8290994001368817,
-      "grad_norm": 0.002679096534848213,
-      "learning_rate": 0.0001999983519879702,
-      "loss": 46.0,
-      "step": 11358
-    },
-    {
-      "epoch": 1.8292604372156689,
-      "grad_norm": 0.0017173058586195111,
-      "learning_rate": 0.0001999983516975084,
-      "loss": 46.0,
-      "step": 11359
-    },
-    {
-      "epoch": 1.8294214742944563,
-      "grad_norm": 0.00480238813906908,
-      "learning_rate": 0.00019999835140702103,
-      "loss": 46.0,
-      "step": 11360
-    },
-    {
-      "epoch": 1.8295825113732436,
-      "grad_norm": 0.0010833420092239976,
-      "learning_rate": 0.00019999835111650805,
-      "loss": 46.0,
-      "step": 11361
-    },
-    {
-      "epoch": 1.829743548452031,
-      "grad_norm": 0.0047287773340940475,
-      "learning_rate": 0.00019999835082596952,
-      "loss": 46.0,
-      "step": 11362
-    },
-    {
-      "epoch": 1.8299045855308185,
-      "grad_norm": 0.0011713794665411115,
-      "learning_rate": 0.00019999835053540537,
-      "loss": 46.0,
-      "step": 11363
-    },
-    {
-      "epoch": 1.830065622609606,
-      "grad_norm": 0.0070471977815032005,
-      "learning_rate": 0.0001999983502448156,
-      "loss": 46.0,
-      "step": 11364
-    },
-    {
-      "epoch": 1.8302266596883934,
-      "grad_norm": 0.0023480458185076714,
-      "learning_rate": 0.00019999834995420028,
-      "loss": 46.0,
-      "step": 11365
-    },
-    {
-      "epoch": 1.8303876967671806,
-      "grad_norm": 0.0011858188081532717,
-      "learning_rate": 0.00019999834966355932,
-      "loss": 46.0,
-      "step": 11366
-    },
-    {
-      "epoch": 1.830548733845968,
-      "grad_norm": 0.0026653336826711893,
-      "learning_rate": 0.0001999983493728928,
-      "loss": 46.0,
-      "step": 11367
-    },
-    {
-      "epoch": 1.8307097709247553,
-      "grad_norm": 0.002672742586582899,
-      "learning_rate": 0.00019999834908220068,
-      "loss": 46.0,
-      "step": 11368
-    },
-    {
-      "epoch": 1.8308708080035427,
-      "grad_norm": 0.0010761015582829714,
-      "learning_rate": 0.00019999834879148292,
-      "loss": 46.0,
-      "step": 11369
-    },
-    {
-      "epoch": 1.8310318450823302,
-      "grad_norm": 0.011883526109158993,
-      "learning_rate": 0.00019999834850073964,
-      "loss": 46.0,
-      "step": 11370
-    },
-    {
-      "epoch": 1.8311928821611176,
-      "grad_norm": 0.0021962234750390053,
-      "learning_rate": 0.0001999983482099707,
-      "loss": 46.0,
-      "step": 11371
-    },
-    {
-      "epoch": 1.831353919239905,
-      "grad_norm": 0.0008276525768451393,
-      "learning_rate": 0.00019999834791917622,
-      "loss": 46.0,
-      "step": 11372
-    },
-    {
-      "epoch": 1.8315149563186925,
-      "grad_norm": 0.0004925379762426019,
-      "learning_rate": 0.00019999834762835615,
-      "loss": 46.0,
-      "step": 11373
-    },
-    {
-      "epoch": 1.8316759933974798,
-      "grad_norm": 0.0026018882635980844,
-      "learning_rate": 0.00019999834733751043,
-      "loss": 46.0,
-      "step": 11374
-    },
-    {
-      "epoch": 1.831837030476267,
-      "grad_norm": 0.0019435338908806443,
-      "learning_rate": 0.00019999834704663915,
-      "loss": 46.0,
-      "step": 11375
-    },
-    {
-      "epoch": 1.8319980675550545,
-      "grad_norm": 0.0010475931921973825,
-      "learning_rate": 0.00019999834675574226,
-      "loss": 46.0,
-      "step": 11376
-    },
-    {
-      "epoch": 1.832159104633842,
-      "grad_norm": 0.0008468086016364396,
-      "learning_rate": 0.0001999983464648198,
-      "loss": 46.0,
-      "step": 11377
-    },
-    {
-      "epoch": 1.8323201417126294,
-      "grad_norm": 0.004870264325290918,
-      "learning_rate": 0.00019999834617387172,
-      "loss": 46.0,
-      "step": 11378
-    },
-    {
-      "epoch": 1.8324811787914168,
-      "grad_norm": 0.0036125630140304565,
-      "learning_rate": 0.00019999834588289807,
-      "loss": 46.0,
-      "step": 11379
-    },
-    {
-      "epoch": 1.8326422158702043,
-      "grad_norm": 0.0034376876428723335,
-      "learning_rate": 0.00019999834559189877,
-      "loss": 46.0,
-      "step": 11380
-    },
-    {
-      "epoch": 1.8328032529489915,
-      "grad_norm": 0.0023218696005642414,
-      "learning_rate": 0.00019999834530087395,
-      "loss": 46.0,
-      "step": 11381
-    },
-    {
-      "epoch": 1.832964290027779,
-      "grad_norm": 0.0018498097779229283,
-      "learning_rate": 0.0001999983450098235,
-      "loss": 46.0,
-      "step": 11382
-    },
-    {
-      "epoch": 1.8331253271065662,
-      "grad_norm": 0.007345468737185001,
-      "learning_rate": 0.00019999834471874745,
-      "loss": 46.0,
-      "step": 11383
-    },
-    {
-      "epoch": 1.8332863641853536,
-      "grad_norm": 0.002498486777767539,
-      "learning_rate": 0.00019999834442764583,
-      "loss": 46.0,
-      "step": 11384
-    },
-    {
-      "epoch": 1.833447401264141,
-      "grad_norm": 0.003603970631957054,
-      "learning_rate": 0.00019999834413651858,
-      "loss": 46.0,
-      "step": 11385
-    },
-    {
-      "epoch": 1.8336084383429285,
-      "grad_norm": 0.0016164917033165693,
-      "learning_rate": 0.0001999983438453658,
-      "loss": 46.0,
-      "step": 11386
-    },
-    {
-      "epoch": 1.833769475421716,
-      "grad_norm": 0.002884289715439081,
-      "learning_rate": 0.00019999834355418733,
-      "loss": 46.0,
-      "step": 11387
-    },
-    {
-      "epoch": 1.8339305125005032,
-      "grad_norm": 0.004006147384643555,
-      "learning_rate": 0.0001999983432629833,
-      "loss": 46.0,
-      "step": 11388
-    },
-    {
-      "epoch": 1.8340915495792907,
-      "grad_norm": 0.0023818037007004023,
-      "learning_rate": 0.0001999983429717537,
-      "loss": 46.0,
-      "step": 11389
-    },
-    {
-      "epoch": 1.8342525866580779,
-      "grad_norm": 0.0025085010565817356,
-      "learning_rate": 0.0001999983426804985,
-      "loss": 46.0,
-      "step": 11390
-    },
-    {
-      "epoch": 1.8344136237368653,
-      "grad_norm": 0.005747461691498756,
-      "learning_rate": 0.00019999834238921773,
-      "loss": 46.0,
-      "step": 11391
-    },
-    {
-      "epoch": 1.8345746608156528,
-      "grad_norm": 0.0006192016880959272,
-      "learning_rate": 0.00019999834209791134,
-      "loss": 46.0,
-      "step": 11392
-    },
-    {
-      "epoch": 1.8347356978944402,
-      "grad_norm": 0.00795027893036604,
-      "learning_rate": 0.00019999834180657936,
-      "loss": 46.0,
-      "step": 11393
-    },
-    {
-      "epoch": 1.8348967349732277,
-      "grad_norm": 0.003281216137111187,
-      "learning_rate": 0.00019999834151522173,
-      "loss": 46.0,
-      "step": 11394
-    },
-    {
-      "epoch": 1.835057772052015,
-      "grad_norm": 0.0007629744941368699,
-      "learning_rate": 0.00019999834122383858,
-      "loss": 46.0,
-      "step": 11395
-    },
-    {
-      "epoch": 1.8352188091308024,
-      "grad_norm": 0.0009485826594755054,
-      "learning_rate": 0.0001999983409324298,
-      "loss": 46.0,
-      "step": 11396
-    },
-    {
-      "epoch": 1.8353798462095896,
-      "grad_norm": 0.006807301193475723,
-      "learning_rate": 0.00019999834064099545,
-      "loss": 46.0,
-      "step": 11397
-    },
-    {
-      "epoch": 1.835540883288377,
-      "grad_norm": 0.0008100178674794734,
-      "learning_rate": 0.00019999834034953548,
-      "loss": 46.0,
-      "step": 11398
-    },
-    {
-      "epoch": 1.8357019203671645,
-      "grad_norm": 0.0005330259446054697,
-      "learning_rate": 0.00019999834005804995,
-      "loss": 46.0,
-      "step": 11399
-    },
-    {
-      "epoch": 1.835862957445952,
-      "grad_norm": 0.004904255736619234,
-      "learning_rate": 0.0001999983397665388,
-      "loss": 46.0,
-      "step": 11400
-    },
-    {
-      "epoch": 1.8360239945247394,
-      "grad_norm": 0.003417830914258957,
-      "learning_rate": 0.00019999833947500204,
-      "loss": 46.0,
-      "step": 11401
-    },
-    {
-      "epoch": 1.8361850316035269,
-      "grad_norm": 0.0037887864746153355,
-      "learning_rate": 0.00019999833918343972,
-      "loss": 46.0,
-      "step": 11402
-    },
-    {
-      "epoch": 1.836346068682314,
-      "grad_norm": 0.003927609417587519,
-      "learning_rate": 0.0001999983388918518,
-      "loss": 46.0,
-      "step": 11403
-    },
-    {
-      "epoch": 1.8365071057611015,
-      "grad_norm": 0.0014977715909481049,
-      "learning_rate": 0.00019999833860023824,
-      "loss": 46.0,
-      "step": 11404
-    },
-    {
-      "epoch": 1.8366681428398888,
-      "grad_norm": 0.002865336136892438,
-      "learning_rate": 0.00019999833830859913,
-      "loss": 46.0,
-      "step": 11405
-    },
-    {
-      "epoch": 1.8368291799186762,
-      "grad_norm": 0.0008764376980252564,
-      "learning_rate": 0.0001999983380169344,
-      "loss": 46.0,
-      "step": 11406
-    },
-    {
-      "epoch": 1.8369902169974637,
-      "grad_norm": 0.0014379406347870827,
-      "learning_rate": 0.00019999833772524412,
-      "loss": 46.0,
-      "step": 11407
-    },
-    {
-      "epoch": 1.8371512540762511,
-      "grad_norm": 0.0012700404040515423,
-      "learning_rate": 0.0001999983374335282,
-      "loss": 46.0,
-      "step": 11408
-    },
-    {
-      "epoch": 1.8373122911550386,
-      "grad_norm": 0.0008513263310305774,
-      "learning_rate": 0.00019999833714178672,
-      "loss": 46.0,
-      "step": 11409
-    },
-    {
-      "epoch": 1.8374733282338258,
-      "grad_norm": 0.0018022852018475533,
-      "learning_rate": 0.00019999833685001962,
-      "loss": 46.0,
-      "step": 11410
-    },
-    {
-      "epoch": 1.8376343653126133,
-      "grad_norm": 0.0019342032028362155,
-      "learning_rate": 0.00019999833655822693,
-      "loss": 46.0,
-      "step": 11411
-    },
-    {
-      "epoch": 1.8377954023914005,
-      "grad_norm": 0.0008623074973002076,
-      "learning_rate": 0.00019999833626640866,
-      "loss": 46.0,
-      "step": 11412
-    },
-    {
-      "epoch": 1.837956439470188,
-      "grad_norm": 0.005737262777984142,
-      "learning_rate": 0.00019999833597456477,
-      "loss": 46.0,
-      "step": 11413
-    },
-    {
-      "epoch": 1.8381174765489754,
-      "grad_norm": 0.002845275914296508,
-      "learning_rate": 0.0001999983356826953,
-      "loss": 46.0,
-      "step": 11414
-    },
-    {
-      "epoch": 1.8382785136277628,
-      "grad_norm": 0.0012129410170018673,
-      "learning_rate": 0.00019999833539080023,
-      "loss": 46.0,
-      "step": 11415
-    },
-    {
-      "epoch": 1.8384395507065503,
-      "grad_norm": 0.006648307200521231,
-      "learning_rate": 0.00019999833509887959,
-      "loss": 46.0,
-      "step": 11416
-    },
-    {
-      "epoch": 1.8386005877853375,
-      "grad_norm": 0.002591558964923024,
-      "learning_rate": 0.00019999833480693332,
-      "loss": 46.0,
-      "step": 11417
-    },
-    {
-      "epoch": 1.838761624864125,
-      "grad_norm": 0.0012823701836168766,
-      "learning_rate": 0.00019999833451496147,
-      "loss": 46.0,
-      "step": 11418
-    },
-    {
-      "epoch": 1.8389226619429122,
-      "grad_norm": 0.004582727327942848,
-      "learning_rate": 0.00019999833422296403,
-      "loss": 46.0,
-      "step": 11419
-    },
-    {
-      "epoch": 1.8390836990216997,
-      "grad_norm": 0.0009938521543517709,
-      "learning_rate": 0.00019999833393094098,
-      "loss": 46.0,
-      "step": 11420
-    },
-    {
-      "epoch": 1.839244736100487,
-      "grad_norm": 0.007162528112530708,
-      "learning_rate": 0.00019999833363889237,
-      "loss": 46.0,
-      "step": 11421
-    },
-    {
-      "epoch": 1.8394057731792746,
-      "grad_norm": 0.0027463152073323727,
-      "learning_rate": 0.00019999833334681814,
-      "loss": 46.0,
-      "step": 11422
-    },
-    {
-      "epoch": 1.839566810258062,
-      "grad_norm": 0.0008964014705270529,
-      "learning_rate": 0.00019999833305471833,
-      "loss": 46.0,
-      "step": 11423
-    },
-    {
-      "epoch": 1.8397278473368495,
-      "grad_norm": 0.0054879747331142426,
-      "learning_rate": 0.0001999983327625929,
-      "loss": 46.0,
-      "step": 11424
-    },
-    {
-      "epoch": 1.8398888844156367,
-      "grad_norm": 0.0012115861754864454,
-      "learning_rate": 0.0001999983324704419,
-      "loss": 46.0,
-      "step": 11425
-    },
-    {
-      "epoch": 1.8400499214944241,
-      "grad_norm": 0.0011201546294614673,
-      "learning_rate": 0.0001999983321782653,
-      "loss": 46.0,
-      "step": 11426
-    },
-    {
-      "epoch": 1.8402109585732114,
-      "grad_norm": 0.0012666896218433976,
-      "learning_rate": 0.0001999983318860631,
-      "loss": 46.0,
-      "step": 11427
-    },
-    {
-      "epoch": 1.8403719956519988,
-      "grad_norm": 0.0021752959582954645,
-      "learning_rate": 0.0001999983315938353,
-      "loss": 46.0,
-      "step": 11428
-    },
-    {
-      "epoch": 1.8405330327307863,
-      "grad_norm": 0.002495809458196163,
-      "learning_rate": 0.00019999833130158193,
-      "loss": 46.0,
-      "step": 11429
-    },
-    {
-      "epoch": 1.8406940698095737,
-      "grad_norm": 0.0005240380996838212,
-      "learning_rate": 0.00019999833100930295,
-      "loss": 46.0,
-      "step": 11430
-    },
-    {
-      "epoch": 1.8408551068883612,
-      "grad_norm": 0.0048749265260994434,
-      "learning_rate": 0.00019999833071699836,
-      "loss": 46.0,
-      "step": 11431
-    },
-    {
-      "epoch": 1.8410161439671484,
-      "grad_norm": 0.009719732217490673,
-      "learning_rate": 0.0001999983304246682,
-      "loss": 46.0,
-      "step": 11432
-    },
-    {
-      "epoch": 1.8411771810459359,
-      "grad_norm": 0.0030871760100126266,
-      "learning_rate": 0.00019999833013231242,
-      "loss": 46.0,
-      "step": 11433
-    },
-    {
-      "epoch": 1.841338218124723,
-      "grad_norm": 0.0014130567433312535,
-      "learning_rate": 0.00019999832983993106,
-      "loss": 46.0,
-      "step": 11434
-    },
-    {
-      "epoch": 1.8414992552035105,
-      "grad_norm": 0.0034021991305053234,
-      "learning_rate": 0.00019999832954752412,
-      "loss": 46.0,
-      "step": 11435
-    },
-    {
-      "epoch": 1.841660292282298,
-      "grad_norm": 0.0028927046805620193,
-      "learning_rate": 0.00019999832925509157,
-      "loss": 46.0,
-      "step": 11436
-    },
-    {
-      "epoch": 1.8418213293610854,
-      "grad_norm": 0.0019391439855098724,
-      "learning_rate": 0.00019999832896263342,
-      "loss": 46.0,
-      "step": 11437
-    },
-    {
-      "epoch": 1.841982366439873,
-      "grad_norm": 0.004207367543131113,
-      "learning_rate": 0.00019999832867014967,
-      "loss": 46.0,
-      "step": 11438
-    },
-    {
-      "epoch": 1.8421434035186601,
-      "grad_norm": 0.004299505148082972,
-      "learning_rate": 0.00019999832837764035,
-      "loss": 46.0,
-      "step": 11439
-    },
-    {
-      "epoch": 1.8423044405974476,
-      "grad_norm": 0.0037303371354937553,
-      "learning_rate": 0.00019999832808510545,
-      "loss": 46.0,
-      "step": 11440
-    },
-    {
-      "epoch": 1.8424654776762348,
-      "grad_norm": 0.0047846343368291855,
-      "learning_rate": 0.0001999983277925449,
-      "loss": 46.0,
-      "step": 11441
-    },
-    {
-      "epoch": 1.8426265147550223,
-      "grad_norm": 0.0018996148137375712,
-      "learning_rate": 0.0001999983274999588,
-      "loss": 46.0,
-      "step": 11442
-    },
-    {
-      "epoch": 1.8427875518338097,
-      "grad_norm": 0.0009257634519599378,
-      "learning_rate": 0.00019999832720734708,
-      "loss": 46.0,
-      "step": 11443
-    },
-    {
-      "epoch": 1.8429485889125972,
-      "grad_norm": 0.009656638838350773,
-      "learning_rate": 0.0001999983269147098,
-      "loss": 46.0,
-      "step": 11444
-    },
-    {
-      "epoch": 1.8431096259913846,
-      "grad_norm": 0.002508472418412566,
-      "learning_rate": 0.00019999832662204688,
-      "loss": 46.0,
-      "step": 11445
-    },
-    {
-      "epoch": 1.843270663070172,
-      "grad_norm": 0.0007254515076056123,
-      "learning_rate": 0.00019999832632935837,
-      "loss": 46.0,
-      "step": 11446
-    },
-    {
-      "epoch": 1.8434317001489593,
-      "grad_norm": 0.001405279035679996,
-      "learning_rate": 0.0001999983260366443,
-      "loss": 46.0,
-      "step": 11447
-    },
-    {
-      "epoch": 1.8435927372277465,
-      "grad_norm": 0.0017582313157618046,
-      "learning_rate": 0.00019999832574390462,
-      "loss": 46.0,
-      "step": 11448
-    },
-    {
-      "epoch": 1.843753774306534,
-      "grad_norm": 0.0026010593865066767,
-      "learning_rate": 0.00019999832545113935,
-      "loss": 46.0,
-      "step": 11449
-    },
-    {
-      "epoch": 1.8439148113853214,
-      "grad_norm": 0.001315655536018312,
-      "learning_rate": 0.00019999832515834847,
-      "loss": 46.0,
-      "step": 11450
-    },
-    {
-      "epoch": 1.8440758484641089,
-      "grad_norm": 0.0004457289178390056,
-      "learning_rate": 0.00019999832486553202,
-      "loss": 46.0,
-      "step": 11451
-    },
-    {
-      "epoch": 1.8442368855428963,
-      "grad_norm": 0.0066457754001021385,
-      "learning_rate": 0.00019999832457268996,
-      "loss": 46.0,
-      "step": 11452
-    },
-    {
-      "epoch": 1.8443979226216838,
-      "grad_norm": 0.0007113823085092008,
-      "learning_rate": 0.0001999983242798223,
-      "loss": 46.0,
-      "step": 11453
-    },
-    {
-      "epoch": 1.844558959700471,
-      "grad_norm": 0.002563699847087264,
-      "learning_rate": 0.00019999832398692906,
-      "loss": 46.0,
-      "step": 11454
-    },
-    {
-      "epoch": 1.8447199967792585,
-      "grad_norm": 0.009163310751318932,
-      "learning_rate": 0.0001999983236940102,
-      "loss": 46.0,
-      "step": 11455
-    },
-    {
-      "epoch": 1.8448810338580457,
-      "grad_norm": 0.001349274767562747,
-      "learning_rate": 0.00019999832340106578,
-      "loss": 46.0,
-      "step": 11456
-    },
-    {
-      "epoch": 1.8450420709368331,
-      "grad_norm": 0.0012398994294926524,
-      "learning_rate": 0.00019999832310809573,
-      "loss": 46.0,
-      "step": 11457
-    },
-    {
-      "epoch": 1.8452031080156206,
-      "grad_norm": 0.0011812186567112803,
-      "learning_rate": 0.00019999832281510012,
-      "loss": 46.0,
-      "step": 11458
-    },
-    {
-      "epoch": 1.845364145094408,
-      "grad_norm": 0.007003973703831434,
-      "learning_rate": 0.0001999983225220789,
-      "loss": 46.0,
-      "step": 11459
-    },
-    {
-      "epoch": 1.8455251821731955,
-      "grad_norm": 0.0018861133139580488,
-      "learning_rate": 0.00019999832222903206,
-      "loss": 46.0,
-      "step": 11460
-    },
-    {
-      "epoch": 1.8456862192519827,
-      "grad_norm": 0.0014199883444234729,
-      "learning_rate": 0.00019999832193595967,
-      "loss": 46.0,
-      "step": 11461
-    },
-    {
-      "epoch": 1.8458472563307702,
-      "grad_norm": 0.0026348743122071028,
-      "learning_rate": 0.00019999832164286168,
-      "loss": 46.0,
-      "step": 11462
-    },
-    {
-      "epoch": 1.8460082934095574,
-      "grad_norm": 0.0029695951379835606,
-      "learning_rate": 0.00019999832134973806,
-      "loss": 46.0,
-      "step": 11463
-    },
-    {
-      "epoch": 1.8461693304883449,
-      "grad_norm": 0.0029272474348545074,
-      "learning_rate": 0.00019999832105658887,
-      "loss": 46.0,
-      "step": 11464
-    },
-    {
-      "epoch": 1.8463303675671323,
-      "grad_norm": 0.00149630312807858,
-      "learning_rate": 0.0001999983207634141,
-      "loss": 46.0,
-      "step": 11465
-    },
-    {
-      "epoch": 1.8464914046459198,
-      "grad_norm": 0.003263527061790228,
-      "learning_rate": 0.0001999983204702137,
-      "loss": 46.0,
-      "step": 11466
-    },
-    {
-      "epoch": 1.8466524417247072,
-      "grad_norm": 0.0008884789422154427,
-      "learning_rate": 0.00019999832017698774,
-      "loss": 46.0,
-      "step": 11467
-    },
-    {
-      "epoch": 1.8468134788034947,
-      "grad_norm": 0.0009438117849640548,
-      "learning_rate": 0.00019999831988373615,
-      "loss": 46.0,
-      "step": 11468
-    },
-    {
-      "epoch": 1.846974515882282,
-      "grad_norm": 0.0008721146732568741,
-      "learning_rate": 0.000199998319590459,
-      "loss": 46.0,
-      "step": 11469
-    },
-    {
-      "epoch": 1.8471355529610691,
-      "grad_norm": 0.004317587707191706,
-      "learning_rate": 0.00019999831929715624,
-      "loss": 46.0,
-      "step": 11470
-    },
-    {
-      "epoch": 1.8472965900398566,
-      "grad_norm": 0.003280248725786805,
-      "learning_rate": 0.0001999983190038279,
-      "loss": 46.0,
-      "step": 11471
-    },
-    {
-      "epoch": 1.847457627118644,
-      "grad_norm": 0.008520183153450489,
-      "learning_rate": 0.00019999831871047395,
-      "loss": 46.0,
-      "step": 11472
-    },
-    {
-      "epoch": 1.8476186641974315,
-      "grad_norm": 0.0033995364792644978,
-      "learning_rate": 0.0001999983184170944,
-      "loss": 46.0,
-      "step": 11473
-    },
-    {
-      "epoch": 1.847779701276219,
-      "grad_norm": 0.009308294393122196,
-      "learning_rate": 0.00019999831812368926,
-      "loss": 46.0,
-      "step": 11474
-    },
-    {
-      "epoch": 1.8479407383550064,
-      "grad_norm": 0.002007026458159089,
-      "learning_rate": 0.00019999831783025856,
-      "loss": 46.0,
-      "step": 11475
-    },
-    {
-      "epoch": 1.8481017754337936,
-      "grad_norm": 0.00215074117295444,
-      "learning_rate": 0.00019999831753680222,
-      "loss": 46.0,
-      "step": 11476
-    },
-    {
-      "epoch": 1.848262812512581,
-      "grad_norm": 0.0014058294473215938,
-      "learning_rate": 0.00019999831724332027,
-      "loss": 46.0,
-      "step": 11477
-    },
-    {
-      "epoch": 1.8484238495913683,
-      "grad_norm": 0.0022512900177389383,
-      "learning_rate": 0.00019999831694981278,
-      "loss": 46.0,
-      "step": 11478
-    },
-    {
-      "epoch": 1.8485848866701557,
-      "grad_norm": 0.009319338016211987,
-      "learning_rate": 0.00019999831665627968,
-      "loss": 46.0,
-      "step": 11479
-    },
-    {
-      "epoch": 1.8487459237489432,
-      "grad_norm": 0.004338906146585941,
-      "learning_rate": 0.00019999831636272094,
-      "loss": 46.0,
-      "step": 11480
-    },
-    {
-      "epoch": 1.8489069608277307,
-      "grad_norm": 0.00343674561008811,
-      "learning_rate": 0.0001999983160691367,
-      "loss": 46.0,
-      "step": 11481
-    },
-    {
-      "epoch": 1.849067997906518,
-      "grad_norm": 0.000801559304818511,
-      "learning_rate": 0.0001999983157755268,
-      "loss": 46.0,
-      "step": 11482
-    },
-    {
-      "epoch": 1.8492290349853053,
-      "grad_norm": 0.0018976490246132016,
-      "learning_rate": 0.00019999831548189127,
-      "loss": 46.0,
-      "step": 11483
-    },
-    {
-      "epoch": 1.8493900720640928,
-      "grad_norm": 0.001177231315523386,
-      "learning_rate": 0.00019999831518823023,
-      "loss": 46.0,
-      "step": 11484
-    },
-    {
-      "epoch": 1.84955110914288,
-      "grad_norm": 0.0013412476982921362,
-      "learning_rate": 0.00019999831489454355,
-      "loss": 46.0,
-      "step": 11485
-    },
-    {
-      "epoch": 1.8497121462216675,
-      "grad_norm": 0.006405207794159651,
-      "learning_rate": 0.00019999831460083126,
-      "loss": 46.0,
-      "step": 11486
-    },
-    {
-      "epoch": 1.849873183300455,
-      "grad_norm": 0.008552424609661102,
-      "learning_rate": 0.0001999983143070934,
-      "loss": 46.0,
-      "step": 11487
-    },
-    {
-      "epoch": 1.8500342203792424,
-      "grad_norm": 0.0073061492294073105,
-      "learning_rate": 0.00019999831401332994,
-      "loss": 46.0,
-      "step": 11488
-    },
-    {
-      "epoch": 1.8501952574580298,
-      "grad_norm": 0.0010532313026487827,
-      "learning_rate": 0.00019999831371954088,
-      "loss": 46.0,
-      "step": 11489
-    },
-    {
-      "epoch": 1.850356294536817,
-      "grad_norm": 0.0008252942352555692,
-      "learning_rate": 0.00019999831342572624,
-      "loss": 46.0,
-      "step": 11490
-    },
-    {
-      "epoch": 1.8505173316156045,
-      "grad_norm": 0.0019892535638064146,
-      "learning_rate": 0.00019999831313188599,
-      "loss": 46.0,
-      "step": 11491
-    },
-    {
-      "epoch": 1.8506783686943917,
-      "grad_norm": 0.006403793580830097,
-      "learning_rate": 0.00019999831283802014,
-      "loss": 46.0,
-      "step": 11492
-    },
-    {
-      "epoch": 1.8508394057731792,
-      "grad_norm": 0.00657274853438139,
-      "learning_rate": 0.00019999831254412874,
-      "loss": 46.0,
-      "step": 11493
-    },
-    {
-      "epoch": 1.8510004428519666,
-      "grad_norm": 0.001463776221498847,
-      "learning_rate": 0.0001999983122502117,
-      "loss": 46.0,
-      "step": 11494
-    },
-    {
-      "epoch": 1.851161479930754,
-      "grad_norm": 0.0025666444562375546,
-      "learning_rate": 0.0001999983119562691,
-      "loss": 46.0,
-      "step": 11495
-    },
-    {
-      "epoch": 1.8513225170095415,
-      "grad_norm": 0.0010501445503905416,
-      "learning_rate": 0.00019999831166230087,
-      "loss": 46.0,
-      "step": 11496
-    },
-    {
-      "epoch": 1.851483554088329,
-      "grad_norm": 0.001359405112452805,
-      "learning_rate": 0.00019999831136830707,
-      "loss": 46.0,
-      "step": 11497
-    },
-    {
-      "epoch": 1.8516445911671162,
-      "grad_norm": 0.002873091259971261,
-      "learning_rate": 0.00019999831107428767,
-      "loss": 46.0,
-      "step": 11498
-    },
-    {
-      "epoch": 1.8518056282459037,
-      "grad_norm": 0.00535928551107645,
-      "learning_rate": 0.00019999831078024264,
-      "loss": 46.0,
-      "step": 11499
-    },
-    {
-      "epoch": 1.851966665324691,
-      "grad_norm": 0.000866666785441339,
-      "learning_rate": 0.00019999831048617207,
-      "loss": 46.0,
-      "step": 11500
-    },
-    {
-      "epoch": 1.8521277024034783,
-      "grad_norm": 0.002204623306170106,
-      "learning_rate": 0.0001999983101920759,
-      "loss": 46.0,
-      "step": 11501
-    },
-    {
-      "epoch": 1.8522887394822658,
-      "grad_norm": 0.0007331451633945107,
-      "learning_rate": 0.0001999983098979541,
-      "loss": 46.0,
-      "step": 11502
-    },
-    {
-      "epoch": 1.8524497765610533,
-      "grad_norm": 0.0011486758012324572,
-      "learning_rate": 0.0001999983096038067,
-      "loss": 46.0,
-      "step": 11503
-    },
-    {
-      "epoch": 1.8526108136398407,
-      "grad_norm": 0.005773249547928572,
-      "learning_rate": 0.00019999830930963377,
-      "loss": 46.0,
-      "step": 11504
-    },
-    {
-      "epoch": 1.852771850718628,
-      "grad_norm": 0.003237711265683174,
-      "learning_rate": 0.00019999830901543518,
-      "loss": 46.0,
-      "step": 11505
-    },
-    {
-      "epoch": 1.8529328877974154,
-      "grad_norm": 0.0017491430044174194,
-      "learning_rate": 0.00019999830872121104,
-      "loss": 46.0,
-      "step": 11506
-    },
-    {
-      "epoch": 1.8530939248762026,
-      "grad_norm": 0.0005698419408872724,
-      "learning_rate": 0.00019999830842696128,
-      "loss": 46.0,
-      "step": 11507
-    },
-    {
-      "epoch": 1.85325496195499,
-      "grad_norm": 0.005310474429279566,
-      "learning_rate": 0.00019999830813268593,
-      "loss": 46.0,
-      "step": 11508
-    },
-    {
-      "epoch": 1.8534159990337775,
-      "grad_norm": 0.003777619218453765,
-      "learning_rate": 0.000199998307838385,
-      "loss": 46.0,
-      "step": 11509
-    },
-    {
-      "epoch": 1.853577036112565,
-      "grad_norm": 0.003426079172641039,
-      "learning_rate": 0.00019999830754405845,
-      "loss": 46.0,
-      "step": 11510
-    },
-    {
-      "epoch": 1.8537380731913524,
-      "grad_norm": 0.0011892326874658465,
-      "learning_rate": 0.00019999830724970634,
-      "loss": 46.0,
-      "step": 11511
-    },
-    {
-      "epoch": 1.8538991102701396,
-      "grad_norm": 0.0034862798638641834,
-      "learning_rate": 0.0001999983069553286,
-      "loss": 46.0,
-      "step": 11512
-    },
-    {
-      "epoch": 1.854060147348927,
-      "grad_norm": 0.003071578685194254,
-      "learning_rate": 0.00019999830666092528,
-      "loss": 46.0,
-      "step": 11513
-    },
-    {
-      "epoch": 1.8542211844277143,
-      "grad_norm": 0.0022352172527462244,
-      "learning_rate": 0.00019999830636649636,
-      "loss": 46.0,
-      "step": 11514
-    },
-    {
-      "epoch": 1.8543822215065018,
-      "grad_norm": 0.0010983470128849149,
-      "learning_rate": 0.00019999830607204188,
-      "loss": 46.0,
-      "step": 11515
-    },
-    {
-      "epoch": 1.8545432585852892,
-      "grad_norm": 0.012094882316887379,
-      "learning_rate": 0.00019999830577756175,
-      "loss": 46.0,
-      "step": 11516
-    },
-    {
-      "epoch": 1.8547042956640767,
-      "grad_norm": 0.0008304381044581532,
-      "learning_rate": 0.00019999830548305604,
-      "loss": 46.0,
-      "step": 11517
-    },
-    {
-      "epoch": 1.8548653327428641,
-      "grad_norm": 0.0022238020319491625,
-      "learning_rate": 0.00019999830518852477,
-      "loss": 46.0,
-      "step": 11518
-    },
-    {
-      "epoch": 1.8550263698216516,
-      "grad_norm": 0.007429752964526415,
-      "learning_rate": 0.00019999830489396788,
-      "loss": 46.0,
-      "step": 11519
-    },
-    {
-      "epoch": 1.8551874069004388,
-      "grad_norm": 0.005457818508148193,
-      "learning_rate": 0.0001999983045993854,
-      "loss": 46.0,
-      "step": 11520
-    },
-    {
-      "epoch": 1.8553484439792263,
-      "grad_norm": 0.0009096584981307387,
-      "learning_rate": 0.00019999830430477732,
-      "loss": 46.0,
-      "step": 11521
-    },
-    {
-      "epoch": 1.8555094810580135,
-      "grad_norm": 0.0026716883294284344,
-      "learning_rate": 0.00019999830401014364,
-      "loss": 46.0,
-      "step": 11522
-    },
-    {
-      "epoch": 1.855670518136801,
-      "grad_norm": 0.0003413644735701382,
-      "learning_rate": 0.0001999983037154844,
-      "loss": 46.0,
-      "step": 11523
-    },
-    {
-      "epoch": 1.8558315552155884,
-      "grad_norm": 0.0016049012774601579,
-      "learning_rate": 0.00019999830342079956,
-      "loss": 46.0,
-      "step": 11524
-    },
-    {
-      "epoch": 1.8559925922943759,
-      "grad_norm": 0.006110606715083122,
-      "learning_rate": 0.00019999830312608907,
-      "loss": 46.0,
-      "step": 11525
-    },
-    {
-      "epoch": 1.8561536293731633,
-      "grad_norm": 0.002021973952651024,
-      "learning_rate": 0.00019999830283135304,
-      "loss": 46.0,
-      "step": 11526
-    },
-    {
-      "epoch": 1.8563146664519505,
-      "grad_norm": 0.008725442923605442,
-      "learning_rate": 0.00019999830253659138,
-      "loss": 46.0,
-      "step": 11527
-    },
-    {
-      "epoch": 1.856475703530738,
-      "grad_norm": 0.0022730696946382523,
-      "learning_rate": 0.00019999830224180415,
-      "loss": 46.0,
-      "step": 11528
-    },
-    {
-      "epoch": 1.8566367406095252,
-      "grad_norm": 0.00481440918520093,
-      "learning_rate": 0.00019999830194699131,
-      "loss": 46.0,
-      "step": 11529
-    },
-    {
-      "epoch": 1.8567977776883127,
-      "grad_norm": 0.00156695069745183,
-      "learning_rate": 0.0001999983016521529,
-      "loss": 46.0,
-      "step": 11530
-    },
-    {
-      "epoch": 1.8569588147671001,
-      "grad_norm": 0.009470343589782715,
-      "learning_rate": 0.00019999830135728887,
-      "loss": 46.0,
-      "step": 11531
-    },
-    {
-      "epoch": 1.8571198518458876,
-      "grad_norm": 0.0007126287673600018,
-      "learning_rate": 0.00019999830106239927,
-      "loss": 46.0,
-      "step": 11532
-    },
-    {
-      "epoch": 1.857280888924675,
-      "grad_norm": 0.0016426616348326206,
-      "learning_rate": 0.00019999830076748406,
-      "loss": 46.0,
-      "step": 11533
-    },
-    {
-      "epoch": 1.8574419260034623,
-      "grad_norm": 0.0019434961723163724,
-      "learning_rate": 0.00019999830047254325,
-      "loss": 46.0,
-      "step": 11534
-    },
-    {
-      "epoch": 1.8576029630822497,
-      "grad_norm": 0.0013755474938079715,
-      "learning_rate": 0.00019999830017757687,
-      "loss": 46.0,
-      "step": 11535
-    },
-    {
-      "epoch": 1.857764000161037,
-      "grad_norm": 0.0008873820188455284,
-      "learning_rate": 0.00019999829988258486,
-      "loss": 46.0,
-      "step": 11536
-    },
-    {
-      "epoch": 1.8579250372398244,
-      "grad_norm": 0.004061360843479633,
-      "learning_rate": 0.0001999982995875673,
-      "loss": 46.0,
-      "step": 11537
-    },
-    {
-      "epoch": 1.8580860743186118,
-      "grad_norm": 0.0008830491569824517,
-      "learning_rate": 0.0001999982992925241,
-      "loss": 46.0,
-      "step": 11538
-    },
-    {
-      "epoch": 1.8582471113973993,
-      "grad_norm": 0.0017756131710484624,
-      "learning_rate": 0.00019999829899745533,
-      "loss": 46.0,
-      "step": 11539
-    },
-    {
-      "epoch": 1.8584081484761867,
-      "grad_norm": 0.00431841891258955,
-      "learning_rate": 0.00019999829870236095,
-      "loss": 46.0,
-      "step": 11540
-    },
-    {
-      "epoch": 1.8585691855549742,
-      "grad_norm": 0.004266228061169386,
-      "learning_rate": 0.000199998298407241,
-      "loss": 46.0,
-      "step": 11541
-    },
-    {
-      "epoch": 1.8587302226337614,
-      "grad_norm": 0.0016785580664873123,
-      "learning_rate": 0.00019999829811209545,
-      "loss": 46.0,
-      "step": 11542
-    },
-    {
-      "epoch": 1.8588912597125486,
-      "grad_norm": 0.004245802294462919,
-      "learning_rate": 0.00019999829781692428,
-      "loss": 46.0,
-      "step": 11543
-    },
-    {
-      "epoch": 1.859052296791336,
-      "grad_norm": 0.006059711799025536,
-      "learning_rate": 0.00019999829752172753,
-      "loss": 46.0,
-      "step": 11544
-    },
-    {
-      "epoch": 1.8592133338701236,
-      "grad_norm": 0.002805279800668359,
-      "learning_rate": 0.00019999829722650519,
-      "loss": 46.0,
-      "step": 11545
-    },
-    {
-      "epoch": 1.859374370948911,
-      "grad_norm": 0.003590754931792617,
-      "learning_rate": 0.00019999829693125726,
-      "loss": 46.0,
-      "step": 11546
-    },
-    {
-      "epoch": 1.8595354080276985,
-      "grad_norm": 0.0025306125171482563,
-      "learning_rate": 0.00019999829663598374,
-      "loss": 46.0,
-      "step": 11547
-    },
-    {
-      "epoch": 1.859696445106486,
-      "grad_norm": 0.0012398662511259317,
-      "learning_rate": 0.0001999982963406846,
-      "loss": 46.0,
-      "step": 11548
-    },
-    {
-      "epoch": 1.8598574821852731,
-      "grad_norm": 0.0016159494407474995,
-      "learning_rate": 0.00019999829604535986,
-      "loss": 46.0,
-      "step": 11549
-    },
-    {
-      "epoch": 1.8600185192640606,
-      "grad_norm": 0.002433615271002054,
-      "learning_rate": 0.00019999829575000956,
-      "loss": 46.0,
-      "step": 11550
-    },
-    {
-      "epoch": 1.8601795563428478,
-      "grad_norm": 0.000999464187771082,
-      "learning_rate": 0.00019999829545463366,
-      "loss": 46.0,
-      "step": 11551
-    },
-    {
-      "epoch": 1.8603405934216353,
-      "grad_norm": 0.002432487206533551,
-      "learning_rate": 0.00019999829515923216,
-      "loss": 46.0,
-      "step": 11552
-    },
-    {
-      "epoch": 1.8605016305004227,
-      "grad_norm": 0.0009705192060209811,
-      "learning_rate": 0.00019999829486380506,
-      "loss": 46.0,
-      "step": 11553
-    },
-    {
-      "epoch": 1.8606626675792102,
-      "grad_norm": 0.004051779862493277,
-      "learning_rate": 0.00019999829456835238,
-      "loss": 46.0,
-      "step": 11554
-    },
-    {
-      "epoch": 1.8608237046579976,
-      "grad_norm": 0.001016879454255104,
-      "learning_rate": 0.00019999829427287409,
-      "loss": 46.0,
-      "step": 11555
-    },
-    {
-      "epoch": 1.8609847417367849,
-      "grad_norm": 0.0016925642266869545,
-      "learning_rate": 0.00019999829397737023,
-      "loss": 46.0,
-      "step": 11556
-    },
-    {
-      "epoch": 1.8611457788155723,
-      "grad_norm": 0.004090378526598215,
-      "learning_rate": 0.00019999829368184073,
-      "loss": 46.0,
-      "step": 11557
-    },
-    {
-      "epoch": 1.8613068158943595,
-      "grad_norm": 0.005295174196362495,
-      "learning_rate": 0.00019999829338628568,
-      "loss": 46.0,
-      "step": 11558
-    },
-    {
-      "epoch": 1.861467852973147,
-      "grad_norm": 0.004252488724887371,
-      "learning_rate": 0.000199998293090705,
-      "loss": 46.0,
-      "step": 11559
-    },
-    {
-      "epoch": 1.8616288900519344,
-      "grad_norm": 0.0010820492170751095,
-      "learning_rate": 0.00019999829279509875,
-      "loss": 46.0,
-      "step": 11560
-    },
-    {
-      "epoch": 1.8617899271307219,
-      "grad_norm": 0.0027612620033323765,
-      "learning_rate": 0.0001999982924994669,
-      "loss": 46.0,
-      "step": 11561
-    },
-    {
-      "epoch": 1.8619509642095093,
-      "grad_norm": 0.002777194371446967,
-      "learning_rate": 0.00019999829220380947,
-      "loss": 46.0,
-      "step": 11562
-    },
-    {
-      "epoch": 1.8621120012882968,
-      "grad_norm": 0.0005786261754110456,
-      "learning_rate": 0.0001999982919081264,
-      "loss": 46.0,
-      "step": 11563
-    },
-    {
-      "epoch": 1.862273038367084,
-      "grad_norm": 0.0017317441524937749,
-      "learning_rate": 0.00019999829161241776,
-      "loss": 46.0,
-      "step": 11564
-    },
-    {
-      "epoch": 1.8624340754458713,
-      "grad_norm": 0.0030908421613276005,
-      "learning_rate": 0.00019999829131668354,
-      "loss": 46.0,
-      "step": 11565
-    },
-    {
-      "epoch": 1.8625951125246587,
-      "grad_norm": 0.0004570278979372233,
-      "learning_rate": 0.0001999982910209237,
-      "loss": 46.0,
-      "step": 11566
-    },
-    {
-      "epoch": 1.8627561496034462,
-      "grad_norm": 0.0009962570620700717,
-      "learning_rate": 0.0001999982907251383,
-      "loss": 46.0,
-      "step": 11567
-    },
-    {
-      "epoch": 1.8629171866822336,
-      "grad_norm": 0.0015381964622065425,
-      "learning_rate": 0.00019999829042932727,
-      "loss": 46.0,
-      "step": 11568
-    },
-    {
-      "epoch": 1.863078223761021,
-      "grad_norm": 0.0021539456211030483,
-      "learning_rate": 0.00019999829013349067,
-      "loss": 46.0,
-      "step": 11569
-    },
-    {
-      "epoch": 1.8632392608398085,
-      "grad_norm": 0.002047761343419552,
-      "learning_rate": 0.00019999828983762849,
-      "loss": 46.0,
-      "step": 11570
-    },
-    {
-      "epoch": 1.8634002979185957,
-      "grad_norm": 0.003137151710689068,
-      "learning_rate": 0.00019999828954174066,
-      "loss": 46.0,
-      "step": 11571
-    },
-    {
-      "epoch": 1.8635613349973832,
-      "grad_norm": 0.01071912981569767,
-      "learning_rate": 0.00019999828924582727,
-      "loss": 46.0,
-      "step": 11572
-    },
-    {
-      "epoch": 1.8637223720761704,
-      "grad_norm": 0.0056599159725010395,
-      "learning_rate": 0.00019999828894988827,
-      "loss": 46.0,
-      "step": 11573
-    },
-    {
-      "epoch": 1.8638834091549579,
-      "grad_norm": 0.004475565627217293,
-      "learning_rate": 0.0001999982886539237,
-      "loss": 46.0,
-      "step": 11574
-    },
-    {
-      "epoch": 1.8640444462337453,
-      "grad_norm": 0.005175105761736631,
-      "learning_rate": 0.00019999828835793354,
-      "loss": 46.0,
-      "step": 11575
-    },
-    {
-      "epoch": 1.8642054833125328,
-      "grad_norm": 0.0025757038965821266,
-      "learning_rate": 0.00019999828806191778,
-      "loss": 46.0,
-      "step": 11576
-    },
-    {
-      "epoch": 1.8643665203913202,
-      "grad_norm": 0.0015190429985523224,
-      "learning_rate": 0.0001999982877658764,
-      "loss": 46.0,
-      "step": 11577
-    },
-    {
-      "epoch": 1.8645275574701075,
-      "grad_norm": 0.0010950901778414845,
-      "learning_rate": 0.00019999828746980946,
-      "loss": 46.0,
-      "step": 11578
-    },
-    {
-      "epoch": 1.864688594548895,
-      "grad_norm": 0.0014548160834237933,
-      "learning_rate": 0.00019999828717371689,
-      "loss": 46.0,
-      "step": 11579
-    },
-    {
-      "epoch": 1.8648496316276821,
-      "grad_norm": 0.0021911978255957365,
-      "learning_rate": 0.00019999828687759875,
-      "loss": 46.0,
-      "step": 11580
-    },
-    {
-      "epoch": 1.8650106687064696,
-      "grad_norm": 0.0030266642570495605,
-      "learning_rate": 0.000199998286581455,
-      "loss": 46.0,
-      "step": 11581
-    },
-    {
-      "epoch": 1.865171705785257,
-      "grad_norm": 0.002875728067010641,
-      "learning_rate": 0.00019999828628528568,
-      "loss": 46.0,
-      "step": 11582
-    },
-    {
-      "epoch": 1.8653327428640445,
-      "grad_norm": 0.0014987209578976035,
-      "learning_rate": 0.00019999828598909073,
-      "loss": 46.0,
-      "step": 11583
-    },
-    {
-      "epoch": 1.865493779942832,
-      "grad_norm": 0.0017213376704603434,
-      "learning_rate": 0.00019999828569287022,
-      "loss": 46.0,
-      "step": 11584
-    },
-    {
-      "epoch": 1.8656548170216194,
-      "grad_norm": 0.0033472541254013777,
-      "learning_rate": 0.00019999828539662412,
-      "loss": 46.0,
-      "step": 11585
-    },
-    {
-      "epoch": 1.8658158541004066,
-      "grad_norm": 0.0066199894063174725,
-      "learning_rate": 0.0001999982851003524,
-      "loss": 46.0,
-      "step": 11586
-    },
-    {
-      "epoch": 1.8659768911791939,
-      "grad_norm": 0.0005458767409436405,
-      "learning_rate": 0.0001999982848040551,
-      "loss": 46.0,
-      "step": 11587
-    },
-    {
-      "epoch": 1.8661379282579813,
-      "grad_norm": 0.001661198097281158,
-      "learning_rate": 0.00019999828450773216,
-      "loss": 46.0,
-      "step": 11588
-    },
-    {
-      "epoch": 1.8662989653367688,
-      "grad_norm": 0.0021142465993762016,
-      "learning_rate": 0.00019999828421138368,
-      "loss": 46.0,
-      "step": 11589
-    },
-    {
-      "epoch": 1.8664600024155562,
-      "grad_norm": 0.008855815045535564,
-      "learning_rate": 0.0001999982839150096,
-      "loss": 46.0,
-      "step": 11590
-    },
-    {
-      "epoch": 1.8666210394943437,
-      "grad_norm": 0.012997549958527088,
-      "learning_rate": 0.0001999982836186099,
-      "loss": 46.0,
-      "step": 11591
-    },
-    {
-      "epoch": 1.866782076573131,
-      "grad_norm": 0.00476828683167696,
-      "learning_rate": 0.00019999828332218462,
-      "loss": 46.0,
-      "step": 11592
-    },
-    {
-      "epoch": 1.8669431136519183,
-      "grad_norm": 0.0006384350708685815,
-      "learning_rate": 0.00019999828302573374,
-      "loss": 46.0,
-      "step": 11593
-    },
-    {
-      "epoch": 1.8671041507307058,
-      "grad_norm": 0.00263122352771461,
-      "learning_rate": 0.0001999982827292573,
-      "loss": 46.0,
-      "step": 11594
-    },
-    {
-      "epoch": 1.867265187809493,
-      "grad_norm": 0.00040063809137791395,
-      "learning_rate": 0.00019999828243275522,
-      "loss": 46.0,
-      "step": 11595
-    },
-    {
-      "epoch": 1.8674262248882805,
-      "grad_norm": 0.001571177621372044,
-      "learning_rate": 0.00019999828213622758,
-      "loss": 46.0,
-      "step": 11596
-    },
-    {
-      "epoch": 1.867587261967068,
-      "grad_norm": 0.0012726279674097896,
-      "learning_rate": 0.0001999982818396743,
-      "loss": 46.0,
-      "step": 11597
-    },
-    {
-      "epoch": 1.8677482990458554,
-      "grad_norm": 0.006881889421492815,
-      "learning_rate": 0.00019999828154309546,
-      "loss": 46.0,
-      "step": 11598
-    },
-    {
-      "epoch": 1.8679093361246428,
-      "grad_norm": 0.0048666601069271564,
-      "learning_rate": 0.00019999828124649103,
-      "loss": 46.0,
-      "step": 11599
-    },
-    {
-      "epoch": 1.86807037320343,
-      "grad_norm": 0.0014073617057874799,
-      "learning_rate": 0.00019999828094986098,
-      "loss": 46.0,
-      "step": 11600
-    },
-    {
-      "epoch": 1.8682314102822175,
-      "grad_norm": 0.0013188141165301204,
-      "learning_rate": 0.00019999828065320535,
-      "loss": 46.0,
-      "step": 11601
-    },
-    {
-      "epoch": 1.8683924473610047,
-      "grad_norm": 0.005279910285025835,
-      "learning_rate": 0.00019999828035652413,
-      "loss": 46.0,
-      "step": 11602
-    },
-    {
-      "epoch": 1.8685534844397922,
-      "grad_norm": 0.0031094886362552643,
-      "learning_rate": 0.0001999982800598173,
-      "loss": 46.0,
-      "step": 11603
-    },
-    {
-      "epoch": 1.8687145215185796,
-      "grad_norm": 0.0006005127215757966,
-      "learning_rate": 0.0001999982797630849,
-      "loss": 46.0,
-      "step": 11604
-    },
-    {
-      "epoch": 1.868875558597367,
-      "grad_norm": 0.002337319077923894,
-      "learning_rate": 0.0001999982794663269,
-      "loss": 46.0,
-      "step": 11605
-    },
-    {
-      "epoch": 1.8690365956761545,
-      "grad_norm": 0.005711226258426905,
-      "learning_rate": 0.0001999982791695433,
-      "loss": 46.0,
-      "step": 11606
-    },
-    {
-      "epoch": 1.8691976327549418,
-      "grad_norm": 0.0006825447198934853,
-      "learning_rate": 0.0001999982788727341,
-      "loss": 46.0,
-      "step": 11607
-    },
-    {
-      "epoch": 1.8693586698337292,
-      "grad_norm": 0.002837394131347537,
-      "learning_rate": 0.00019999827857589933,
-      "loss": 46.0,
-      "step": 11608
-    },
-    {
-      "epoch": 1.8695197069125165,
-      "grad_norm": 0.00829993560910225,
-      "learning_rate": 0.00019999827827903892,
-      "loss": 46.0,
-      "step": 11609
-    },
-    {
-      "epoch": 1.869680743991304,
-      "grad_norm": 0.00921242218464613,
-      "learning_rate": 0.00019999827798215293,
-      "loss": 46.0,
-      "step": 11610
-    },
-    {
-      "epoch": 1.8698417810700914,
-      "grad_norm": 0.0067711821757256985,
-      "learning_rate": 0.00019999827768524137,
-      "loss": 46.0,
-      "step": 11611
-    },
-    {
-      "epoch": 1.8700028181488788,
-      "grad_norm": 0.0025815435219556093,
-      "learning_rate": 0.0001999982773883042,
-      "loss": 46.0,
-      "step": 11612
-    },
-    {
-      "epoch": 1.8701638552276663,
-      "grad_norm": 0.0024665442761033773,
-      "learning_rate": 0.0001999982770913414,
-      "loss": 46.0,
-      "step": 11613
-    },
-    {
-      "epoch": 1.8703248923064537,
-      "grad_norm": 0.0028704360593110323,
-      "learning_rate": 0.00019999827679435307,
-      "loss": 46.0,
-      "step": 11614
-    },
-    {
-      "epoch": 1.870485929385241,
-      "grad_norm": 0.0065760137513279915,
-      "learning_rate": 0.00019999827649733914,
-      "loss": 46.0,
-      "step": 11615
-    },
-    {
-      "epoch": 1.8706469664640284,
-      "grad_norm": 0.004264004994183779,
-      "learning_rate": 0.0001999982762002996,
-      "loss": 46.0,
-      "step": 11616
-    },
-    {
-      "epoch": 1.8708080035428156,
-      "grad_norm": 0.002643846906721592,
-      "learning_rate": 0.00019999827590323443,
-      "loss": 46.0,
-      "step": 11617
-    },
-    {
-      "epoch": 1.870969040621603,
-      "grad_norm": 0.0017271623946726322,
-      "learning_rate": 0.0001999982756061437,
-      "loss": 46.0,
-      "step": 11618
-    },
-    {
-      "epoch": 1.8711300777003905,
-      "grad_norm": 0.0018560613971203566,
-      "learning_rate": 0.00019999827530902737,
-      "loss": 46.0,
-      "step": 11619
-    },
-    {
-      "epoch": 1.871291114779178,
-      "grad_norm": 0.0025461779441684484,
-      "learning_rate": 0.00019999827501188545,
-      "loss": 46.0,
-      "step": 11620
-    },
-    {
-      "epoch": 1.8714521518579654,
-      "grad_norm": 0.001947728917002678,
-      "learning_rate": 0.00019999827471471794,
-      "loss": 46.0,
-      "step": 11621
-    },
-    {
-      "epoch": 1.8716131889367527,
-      "grad_norm": 0.0035788097884505987,
-      "learning_rate": 0.00019999827441752484,
-      "loss": 46.0,
-      "step": 11622
-    },
-    {
-      "epoch": 1.87177422601554,
-      "grad_norm": 0.002290220931172371,
-      "learning_rate": 0.00019999827412030613,
-      "loss": 46.0,
-      "step": 11623
-    },
-    {
-      "epoch": 1.8719352630943273,
-      "grad_norm": 0.0038459710776805878,
-      "learning_rate": 0.00019999827382306183,
-      "loss": 46.0,
-      "step": 11624
-    },
-    {
-      "epoch": 1.8720963001731148,
-      "grad_norm": 0.002096254611387849,
-      "learning_rate": 0.00019999827352579192,
-      "loss": 46.0,
-      "step": 11625
-    },
-    {
-      "epoch": 1.8722573372519022,
-      "grad_norm": 0.004677208140492439,
-      "learning_rate": 0.00019999827322849642,
-      "loss": 46.0,
-      "step": 11626
-    },
-    {
-      "epoch": 1.8724183743306897,
-      "grad_norm": 0.001063927891664207,
-      "learning_rate": 0.00019999827293117533,
-      "loss": 46.0,
-      "step": 11627
-    },
-    {
-      "epoch": 1.8725794114094771,
-      "grad_norm": 0.01260798703879118,
-      "learning_rate": 0.00019999827263382866,
-      "loss": 46.0,
-      "step": 11628
-    },
-    {
-      "epoch": 1.8727404484882644,
-      "grad_norm": 0.0033484536688774824,
-      "learning_rate": 0.0001999982723364564,
-      "loss": 46.0,
-      "step": 11629
-    },
-    {
-      "epoch": 1.8729014855670518,
-      "grad_norm": 0.005107702221721411,
-      "learning_rate": 0.00019999827203905852,
-      "loss": 46.0,
-      "step": 11630
-    },
-    {
-      "epoch": 1.873062522645839,
-      "grad_norm": 0.0005591654335148633,
-      "learning_rate": 0.00019999827174163506,
-      "loss": 46.0,
-      "step": 11631
-    },
-    {
-      "epoch": 1.8732235597246265,
-      "grad_norm": 0.002750370418652892,
-      "learning_rate": 0.000199998271444186,
-      "loss": 46.0,
-      "step": 11632
-    },
-    {
-      "epoch": 1.873384596803414,
-      "grad_norm": 0.0036735234316438437,
-      "learning_rate": 0.00019999827114671134,
-      "loss": 46.0,
-      "step": 11633
-    },
-    {
-      "epoch": 1.8735456338822014,
-      "grad_norm": 0.001277828123420477,
-      "learning_rate": 0.0001999982708492111,
-      "loss": 46.0,
-      "step": 11634
-    },
-    {
-      "epoch": 1.8737066709609889,
-      "grad_norm": 0.0029443674720823765,
-      "learning_rate": 0.00019999827055168528,
-      "loss": 46.0,
-      "step": 11635
-    },
-    {
-      "epoch": 1.8738677080397763,
-      "grad_norm": 0.0015158854657784104,
-      "learning_rate": 0.00019999827025413383,
-      "loss": 46.0,
-      "step": 11636
-    },
-    {
-      "epoch": 1.8740287451185635,
-      "grad_norm": 0.005589430220425129,
-      "learning_rate": 0.0001999982699565568,
-      "loss": 46.0,
-      "step": 11637
-    },
-    {
-      "epoch": 1.874189782197351,
-      "grad_norm": 0.001601747702807188,
-      "learning_rate": 0.0001999982696589542,
-      "loss": 46.0,
-      "step": 11638
-    },
-    {
-      "epoch": 1.8743508192761382,
-      "grad_norm": 0.002457372611388564,
-      "learning_rate": 0.00019999826936132595,
-      "loss": 46.0,
-      "step": 11639
-    },
-    {
-      "epoch": 1.8745118563549257,
-      "grad_norm": 0.0007880728226155043,
-      "learning_rate": 0.00019999826906367215,
-      "loss": 46.0,
-      "step": 11640
-    },
-    {
-      "epoch": 1.8746728934337131,
-      "grad_norm": 0.002802757080644369,
-      "learning_rate": 0.00019999826876599276,
-      "loss": 46.0,
-      "step": 11641
-    },
-    {
-      "epoch": 1.8748339305125006,
-      "grad_norm": 0.0026106261648237705,
-      "learning_rate": 0.00019999826846828773,
-      "loss": 46.0,
-      "step": 11642
-    },
-    {
-      "epoch": 1.874994967591288,
-      "grad_norm": 0.0028694793581962585,
-      "learning_rate": 0.00019999826817055714,
-      "loss": 46.0,
-      "step": 11643
-    },
-    {
-      "epoch": 1.8751560046700753,
-      "grad_norm": 0.0012493396643549204,
-      "learning_rate": 0.00019999826787280096,
-      "loss": 46.0,
-      "step": 11644
-    },
-    {
-      "epoch": 1.8753170417488627,
-      "grad_norm": 0.003053989727050066,
-      "learning_rate": 0.00019999826757501917,
-      "loss": 46.0,
-      "step": 11645
-    },
-    {
-      "epoch": 1.87547807882765,
-      "grad_norm": 0.0018995861755684018,
-      "learning_rate": 0.0001999982672772118,
-      "loss": 46.0,
-      "step": 11646
-    },
-    {
-      "epoch": 1.8756391159064374,
-      "grad_norm": 0.0034354848321527243,
-      "learning_rate": 0.00019999826697937882,
-      "loss": 46.0,
-      "step": 11647
-    },
-    {
-      "epoch": 1.8758001529852248,
-      "grad_norm": 0.002067354740574956,
-      "learning_rate": 0.00019999826668152024,
-      "loss": 46.0,
-      "step": 11648
-    },
-    {
-      "epoch": 1.8759611900640123,
-      "grad_norm": 0.002321695676073432,
-      "learning_rate": 0.00019999826638363607,
-      "loss": 46.0,
-      "step": 11649
-    },
-    {
-      "epoch": 1.8761222271427997,
-      "grad_norm": 0.0025709739420562983,
-      "learning_rate": 0.00019999826608572635,
-      "loss": 46.0,
-      "step": 11650
-    },
-    {
-      "epoch": 1.876283264221587,
-      "grad_norm": 0.001130796386860311,
-      "learning_rate": 0.00019999826578779098,
-      "loss": 46.0,
-      "step": 11651
-    },
-    {
-      "epoch": 1.8764443013003744,
-      "grad_norm": 0.0021283223759382963,
-      "learning_rate": 0.00019999826548983002,
-      "loss": 46.0,
-      "step": 11652
-    },
-    {
-      "epoch": 1.8766053383791617,
-      "grad_norm": 0.0015117021976038814,
-      "learning_rate": 0.0001999982651918435,
-      "loss": 46.0,
-      "step": 11653
-    },
-    {
-      "epoch": 1.876766375457949,
-      "grad_norm": 0.0017427293350920081,
-      "learning_rate": 0.00019999826489383135,
-      "loss": 46.0,
-      "step": 11654
-    },
-    {
-      "epoch": 1.8769274125367366,
-      "grad_norm": 0.0008957125246524811,
-      "learning_rate": 0.00019999826459579363,
-      "loss": 46.0,
-      "step": 11655
-    },
-    {
-      "epoch": 1.877088449615524,
-      "grad_norm": 0.0009859104175120592,
-      "learning_rate": 0.00019999826429773033,
-      "loss": 46.0,
-      "step": 11656
-    },
-    {
-      "epoch": 1.8772494866943115,
-      "grad_norm": 0.0010851643746718764,
-      "learning_rate": 0.00019999826399964138,
-      "loss": 46.0,
-      "step": 11657
-    },
-    {
-      "epoch": 1.877410523773099,
-      "grad_norm": 0.0011887041619047523,
-      "learning_rate": 0.00019999826370152687,
-      "loss": 46.0,
-      "step": 11658
-    },
-    {
-      "epoch": 1.8775715608518861,
-      "grad_norm": 0.0006630821735598147,
-      "learning_rate": 0.00019999826340338678,
-      "loss": 46.0,
-      "step": 11659
-    },
-    {
-      "epoch": 1.8777325979306734,
-      "grad_norm": 0.002834790386259556,
-      "learning_rate": 0.00019999826310522107,
-      "loss": 46.0,
-      "step": 11660
-    },
-    {
-      "epoch": 1.8778936350094608,
-      "grad_norm": 0.0010537327034398913,
-      "learning_rate": 0.00019999826280702978,
-      "loss": 46.0,
-      "step": 11661
-    },
-    {
-      "epoch": 1.8780546720882483,
-      "grad_norm": 0.0024477799888700247,
-      "learning_rate": 0.0001999982625088129,
-      "loss": 46.0,
-      "step": 11662
-    },
-    {
-      "epoch": 1.8782157091670357,
-      "grad_norm": 0.0026908584404736757,
-      "learning_rate": 0.0001999982622105704,
-      "loss": 46.0,
-      "step": 11663
-    },
-    {
-      "epoch": 1.8783767462458232,
-      "grad_norm": 0.001839268021285534,
-      "learning_rate": 0.00019999826191230234,
-      "loss": 46.0,
-      "step": 11664
-    },
-    {
-      "epoch": 1.8785377833246106,
-      "grad_norm": 0.0004310556105338037,
-      "learning_rate": 0.00019999826161400865,
-      "loss": 46.0,
-      "step": 11665
-    },
-    {
-      "epoch": 1.8786988204033979,
-      "grad_norm": 0.0031633267644792795,
-      "learning_rate": 0.0001999982613156894,
-      "loss": 46.0,
-      "step": 11666
-    },
-    {
-      "epoch": 1.8788598574821853,
-      "grad_norm": 0.0012949950760230422,
-      "learning_rate": 0.00019999826101734454,
-      "loss": 46.0,
-      "step": 11667
-    },
-    {
-      "epoch": 1.8790208945609725,
-      "grad_norm": 0.01011162344366312,
-      "learning_rate": 0.00019999826071897408,
-      "loss": 46.0,
-      "step": 11668
-    },
-    {
-      "epoch": 1.87918193163976,
-      "grad_norm": 0.0018819993129000068,
-      "learning_rate": 0.00019999826042057804,
-      "loss": 46.0,
-      "step": 11669
-    },
-    {
-      "epoch": 1.8793429687185474,
-      "grad_norm": 0.0006986952503211796,
-      "learning_rate": 0.00019999826012215638,
-      "loss": 46.0,
-      "step": 11670
-    },
-    {
-      "epoch": 1.879504005797335,
-      "grad_norm": 0.002481229603290558,
-      "learning_rate": 0.00019999825982370913,
-      "loss": 46.0,
-      "step": 11671
-    },
-    {
-      "epoch": 1.8796650428761223,
-      "grad_norm": 0.0022390899248421192,
-      "learning_rate": 0.0001999982595252363,
-      "loss": 46.0,
-      "step": 11672
-    },
-    {
-      "epoch": 1.8798260799549096,
-      "grad_norm": 0.0016635936917737126,
-      "learning_rate": 0.00019999825922673787,
-      "loss": 46.0,
-      "step": 11673
-    },
-    {
-      "epoch": 1.879987117033697,
-      "grad_norm": 0.0009636238100938499,
-      "learning_rate": 0.00019999825892821384,
-      "loss": 46.0,
-      "step": 11674
-    },
-    {
-      "epoch": 1.8801481541124843,
-      "grad_norm": 0.0032580525148659945,
-      "learning_rate": 0.0001999982586296642,
-      "loss": 46.0,
-      "step": 11675
-    },
-    {
-      "epoch": 1.8803091911912717,
-      "grad_norm": 0.003457412589341402,
-      "learning_rate": 0.000199998258331089,
-      "loss": 46.0,
-      "step": 11676
-    },
-    {
-      "epoch": 1.8804702282700592,
-      "grad_norm": 0.0006288232980296016,
-      "learning_rate": 0.0001999982580324882,
-      "loss": 46.0,
-      "step": 11677
-    },
-    {
-      "epoch": 1.8806312653488466,
-      "grad_norm": 0.0021243446972221136,
-      "learning_rate": 0.0001999982577338618,
-      "loss": 46.0,
-      "step": 11678
-    },
-    {
-      "epoch": 1.880792302427634,
-      "grad_norm": 0.003309336258098483,
-      "learning_rate": 0.00019999825743520982,
-      "loss": 46.0,
-      "step": 11679
-    },
-    {
-      "epoch": 1.8809533395064215,
-      "grad_norm": 0.0008861838723532856,
-      "learning_rate": 0.00019999825713653223,
-      "loss": 46.0,
-      "step": 11680
-    },
-    {
-      "epoch": 1.8811143765852087,
-      "grad_norm": 0.005014640279114246,
-      "learning_rate": 0.00019999825683782906,
-      "loss": 46.0,
-      "step": 11681
-    },
-    {
-      "epoch": 1.881275413663996,
-      "grad_norm": 0.003561594756320119,
-      "learning_rate": 0.00019999825653910027,
-      "loss": 46.0,
-      "step": 11682
-    },
-    {
-      "epoch": 1.8814364507427834,
-      "grad_norm": 0.0036646199878305197,
-      "learning_rate": 0.0001999982562403459,
-      "loss": 46.0,
-      "step": 11683
-    },
-    {
-      "epoch": 1.8815974878215709,
-      "grad_norm": 0.003349104430526495,
-      "learning_rate": 0.00019999825594156593,
-      "loss": 46.0,
-      "step": 11684
-    },
-    {
-      "epoch": 1.8817585249003583,
-      "grad_norm": 0.010918417945504189,
-      "learning_rate": 0.00019999825564276038,
-      "loss": 46.0,
-      "step": 11685
-    },
-    {
-      "epoch": 1.8819195619791458,
-      "grad_norm": 0.0031295963563024998,
-      "learning_rate": 0.0001999982553439292,
-      "loss": 46.0,
-      "step": 11686
-    },
-    {
-      "epoch": 1.8820805990579332,
-      "grad_norm": 0.0008877825457602739,
-      "learning_rate": 0.00019999825504507247,
-      "loss": 46.0,
-      "step": 11687
-    },
-    {
-      "epoch": 1.8822416361367205,
-      "grad_norm": 0.0033893361687660217,
-      "learning_rate": 0.0001999982547461901,
-      "loss": 46.0,
-      "step": 11688
-    },
-    {
-      "epoch": 1.882402673215508,
-      "grad_norm": 0.0012382828863337636,
-      "learning_rate": 0.00019999825444728218,
-      "loss": 46.0,
-      "step": 11689
-    },
-    {
-      "epoch": 1.8825637102942951,
-      "grad_norm": 0.0024679130874574184,
-      "learning_rate": 0.00019999825414834864,
-      "loss": 46.0,
-      "step": 11690
-    },
-    {
-      "epoch": 1.8827247473730826,
-      "grad_norm": 0.0005847827414982021,
-      "learning_rate": 0.0001999982538493895,
-      "loss": 46.0,
-      "step": 11691
-    },
-    {
-      "epoch": 1.88288578445187,
-      "grad_norm": 0.0009010388166643679,
-      "learning_rate": 0.0001999982535504048,
-      "loss": 46.0,
-      "step": 11692
-    },
-    {
-      "epoch": 1.8830468215306575,
-      "grad_norm": 0.003370491787791252,
-      "learning_rate": 0.00019999825325139447,
-      "loss": 46.0,
-      "step": 11693
-    },
-    {
-      "epoch": 1.883207858609445,
-      "grad_norm": 0.0011616488918662071,
-      "learning_rate": 0.00019999825295235855,
-      "loss": 46.0,
-      "step": 11694
-    },
-    {
-      "epoch": 1.8833688956882322,
-      "grad_norm": 0.006525881588459015,
-      "learning_rate": 0.00019999825265329705,
-      "loss": 46.0,
-      "step": 11695
-    },
-    {
-      "epoch": 1.8835299327670196,
-      "grad_norm": 0.001878933166153729,
-      "learning_rate": 0.00019999825235420996,
-      "loss": 46.0,
-      "step": 11696
-    },
-    {
-      "epoch": 1.8836909698458069,
-      "grad_norm": 0.0008436068892478943,
-      "learning_rate": 0.00019999825205509729,
-      "loss": 46.0,
-      "step": 11697
-    },
-    {
-      "epoch": 1.8838520069245943,
-      "grad_norm": 0.0012475269613787532,
-      "learning_rate": 0.00019999825175595897,
-      "loss": 46.0,
-      "step": 11698
-    },
-    {
-      "epoch": 1.8840130440033818,
-      "grad_norm": 0.005057751666754484,
-      "learning_rate": 0.0001999982514567951,
-      "loss": 46.0,
-      "step": 11699
-    },
-    {
-      "epoch": 1.8841740810821692,
-      "grad_norm": 0.001804113737307489,
-      "learning_rate": 0.0001999982511576056,
-      "loss": 46.0,
-      "step": 11700
-    },
-    {
-      "epoch": 1.8843351181609567,
-      "grad_norm": 0.0028276785742491484,
-      "learning_rate": 0.00019999825085839054,
-      "loss": 46.0,
-      "step": 11701
-    },
-    {
-      "epoch": 1.884496155239744,
-      "grad_norm": 0.004614020697772503,
-      "learning_rate": 0.00019999825055914988,
-      "loss": 46.0,
-      "step": 11702
-    },
-    {
-      "epoch": 1.8846571923185313,
-      "grad_norm": 0.0017730029067024589,
-      "learning_rate": 0.00019999825025988362,
-      "loss": 46.0,
-      "step": 11703
-    },
-    {
-      "epoch": 1.8848182293973186,
-      "grad_norm": 0.008341995067894459,
-      "learning_rate": 0.00019999824996059175,
-      "loss": 46.0,
-      "step": 11704
-    },
-    {
-      "epoch": 1.884979266476106,
-      "grad_norm": 0.0023817610926926136,
-      "learning_rate": 0.00019999824966127433,
-      "loss": 46.0,
-      "step": 11705
-    },
-    {
-      "epoch": 1.8851403035548935,
-      "grad_norm": 0.002262350171804428,
-      "learning_rate": 0.00019999824936193128,
-      "loss": 46.0,
-      "step": 11706
-    },
-    {
-      "epoch": 1.885301340633681,
-      "grad_norm": 0.0019864633213728666,
-      "learning_rate": 0.00019999824906256263,
-      "loss": 46.0,
-      "step": 11707
-    },
-    {
-      "epoch": 1.8854623777124684,
-      "grad_norm": 0.0007944966200739145,
-      "learning_rate": 0.00019999824876316838,
-      "loss": 46.0,
-      "step": 11708
-    },
-    {
-      "epoch": 1.8856234147912558,
-      "grad_norm": 0.0011223673354834318,
-      "learning_rate": 0.00019999824846374855,
-      "loss": 46.0,
-      "step": 11709
-    },
-    {
-      "epoch": 1.885784451870043,
-      "grad_norm": 0.002139553427696228,
-      "learning_rate": 0.00019999824816430316,
-      "loss": 46.0,
-      "step": 11710
-    },
-    {
-      "epoch": 1.8859454889488305,
-      "grad_norm": 0.003116552485153079,
-      "learning_rate": 0.00019999824786483213,
-      "loss": 46.0,
-      "step": 11711
-    },
-    {
-      "epoch": 1.8861065260276177,
-      "grad_norm": 0.0035305460914969444,
-      "learning_rate": 0.0001999982475653355,
-      "loss": 46.0,
-      "step": 11712
-    },
-    {
-      "epoch": 1.8862675631064052,
-      "grad_norm": 0.00039185149944387376,
-      "learning_rate": 0.0001999982472658133,
-      "loss": 46.0,
-      "step": 11713
-    },
-    {
-      "epoch": 1.8864286001851926,
-      "grad_norm": 0.0018278735224157572,
-      "learning_rate": 0.00019999824696626553,
-      "loss": 46.0,
-      "step": 11714
-    },
-    {
-      "epoch": 1.88658963726398,
-      "grad_norm": 0.002578949322924018,
-      "learning_rate": 0.00019999824666669212,
-      "loss": 46.0,
-      "step": 11715
-    },
-    {
-      "epoch": 1.8867506743427676,
-      "grad_norm": 0.005684688687324524,
-      "learning_rate": 0.00019999824636709313,
-      "loss": 46.0,
-      "step": 11716
-    },
-    {
-      "epoch": 1.8869117114215548,
-      "grad_norm": 0.0036906027235090733,
-      "learning_rate": 0.00019999824606746854,
-      "loss": 46.0,
-      "step": 11717
-    },
-    {
-      "epoch": 1.8870727485003422,
-      "grad_norm": 0.001963048242032528,
-      "learning_rate": 0.00019999824576781835,
-      "loss": 46.0,
-      "step": 11718
-    },
-    {
-      "epoch": 1.8872337855791295,
-      "grad_norm": 0.0016451362753286958,
-      "learning_rate": 0.00019999824546814262,
-      "loss": 46.0,
-      "step": 11719
-    },
-    {
-      "epoch": 1.887394822657917,
-      "grad_norm": 0.0032822638750076294,
-      "learning_rate": 0.00019999824516844124,
-      "loss": 46.0,
-      "step": 11720
-    },
-    {
-      "epoch": 1.8875558597367044,
-      "grad_norm": 0.0008854749030433595,
-      "learning_rate": 0.00019999824486871429,
-      "loss": 46.0,
-      "step": 11721
-    },
-    {
-      "epoch": 1.8877168968154918,
-      "grad_norm": 0.002443799050524831,
-      "learning_rate": 0.0001999982445689617,
-      "loss": 46.0,
-      "step": 11722
-    },
-    {
-      "epoch": 1.8878779338942793,
-      "grad_norm": 0.00199572229757905,
-      "learning_rate": 0.00019999824426918355,
-      "loss": 46.0,
-      "step": 11723
-    },
-    {
-      "epoch": 1.8880389709730665,
-      "grad_norm": 0.0017114938236773014,
-      "learning_rate": 0.00019999824396937983,
-      "loss": 46.0,
-      "step": 11724
-    },
-    {
-      "epoch": 1.888200008051854,
-      "grad_norm": 0.0031553099397569895,
-      "learning_rate": 0.0001999982436695505,
-      "loss": 46.0,
-      "step": 11725
-    },
-    {
-      "epoch": 1.8883610451306412,
-      "grad_norm": 0.0050039710476994514,
-      "learning_rate": 0.00019999824336969555,
-      "loss": 46.0,
-      "step": 11726
-    },
-    {
-      "epoch": 1.8885220822094286,
-      "grad_norm": 0.006236762274056673,
-      "learning_rate": 0.00019999824306981504,
-      "loss": 46.0,
-      "step": 11727
-    },
-    {
-      "epoch": 1.888683119288216,
-      "grad_norm": 0.002946816384792328,
-      "learning_rate": 0.00019999824276990892,
-      "loss": 46.0,
-      "step": 11728
-    },
-    {
-      "epoch": 1.8888441563670035,
-      "grad_norm": 0.000903732783626765,
-      "learning_rate": 0.0001999982424699772,
-      "loss": 46.0,
-      "step": 11729
-    },
-    {
-      "epoch": 1.889005193445791,
-      "grad_norm": 0.006186710204929113,
-      "learning_rate": 0.00019999824217001988,
-      "loss": 46.0,
-      "step": 11730
-    },
-    {
-      "epoch": 1.8891662305245784,
-      "grad_norm": 0.0011079182149842381,
-      "learning_rate": 0.00019999824187003697,
-      "loss": 46.0,
-      "step": 11731
-    },
-    {
-      "epoch": 1.8893272676033657,
-      "grad_norm": 0.006587904412299395,
-      "learning_rate": 0.00019999824157002847,
-      "loss": 46.0,
-      "step": 11732
-    },
-    {
-      "epoch": 1.8894883046821531,
-      "grad_norm": 0.013779995031654835,
-      "learning_rate": 0.00019999824126999439,
-      "loss": 46.0,
-      "step": 11733
-    },
-    {
-      "epoch": 1.8896493417609403,
-      "grad_norm": 0.001720408326946199,
-      "learning_rate": 0.00019999824096993469,
-      "loss": 46.0,
-      "step": 11734
-    },
-    {
-      "epoch": 1.8898103788397278,
-      "grad_norm": 0.001307257916778326,
-      "learning_rate": 0.00019999824066984942,
-      "loss": 46.0,
-      "step": 11735
-    },
-    {
-      "epoch": 1.8899714159185153,
-      "grad_norm": 0.0035038997884839773,
-      "learning_rate": 0.00019999824036973852,
-      "loss": 46.0,
-      "step": 11736
-    },
-    {
-      "epoch": 1.8901324529973027,
-      "grad_norm": 0.0012819096446037292,
-      "learning_rate": 0.00019999824006960206,
-      "loss": 46.0,
-      "step": 11737
-    },
-    {
-      "epoch": 1.8902934900760902,
-      "grad_norm": 0.0017669338267296553,
-      "learning_rate": 0.00019999823976943998,
-      "loss": 46.0,
-      "step": 11738
-    },
-    {
-      "epoch": 1.8904545271548774,
-      "grad_norm": 0.0025631729513406754,
-      "learning_rate": 0.00019999823946925235,
-      "loss": 46.0,
-      "step": 11739
-    },
-    {
-      "epoch": 1.8906155642336648,
-      "grad_norm": 0.005793641321361065,
-      "learning_rate": 0.00019999823916903907,
-      "loss": 46.0,
-      "step": 11740
-    },
-    {
-      "epoch": 1.890776601312452,
-      "grad_norm": 0.002448096638545394,
-      "learning_rate": 0.00019999823886880026,
-      "loss": 46.0,
-      "step": 11741
-    },
-    {
-      "epoch": 1.8909376383912395,
-      "grad_norm": 0.0005394391482695937,
-      "learning_rate": 0.00019999823856853578,
-      "loss": 46.0,
-      "step": 11742
-    },
-    {
-      "epoch": 1.891098675470027,
-      "grad_norm": 0.002117827069014311,
-      "learning_rate": 0.00019999823826824574,
-      "loss": 46.0,
-      "step": 11743
-    },
-    {
-      "epoch": 1.8912597125488144,
-      "grad_norm": 0.0027311628218740225,
-      "learning_rate": 0.00019999823796793012,
-      "loss": 46.0,
-      "step": 11744
-    },
-    {
-      "epoch": 1.8914207496276019,
-      "grad_norm": 0.001291807508096099,
-      "learning_rate": 0.0001999982376675889,
-      "loss": 46.0,
-      "step": 11745
-    },
-    {
-      "epoch": 1.891581786706389,
-      "grad_norm": 0.0012104653287678957,
-      "learning_rate": 0.00019999823736722207,
-      "loss": 46.0,
-      "step": 11746
-    },
-    {
-      "epoch": 1.8917428237851766,
-      "grad_norm": 0.0023579923436045647,
-      "learning_rate": 0.00019999823706682966,
-      "loss": 46.0,
-      "step": 11747
-    },
-    {
-      "epoch": 1.8919038608639638,
-      "grad_norm": 0.000674968643579632,
-      "learning_rate": 0.00019999823676641166,
-      "loss": 46.0,
-      "step": 11748
-    },
-    {
-      "epoch": 1.8920648979427512,
-      "grad_norm": 0.007668052334338427,
-      "learning_rate": 0.00019999823646596804,
-      "loss": 46.0,
-      "step": 11749
-    },
-    {
-      "epoch": 1.8922259350215387,
-      "grad_norm": 0.004024151246994734,
-      "learning_rate": 0.00019999823616549884,
-      "loss": 46.0,
-      "step": 11750
-    },
-    {
-      "epoch": 1.8923869721003261,
-      "grad_norm": 0.0013241079868748784,
-      "learning_rate": 0.00019999823586500405,
-      "loss": 46.0,
-      "step": 11751
-    },
-    {
-      "epoch": 1.8925480091791136,
-      "grad_norm": 0.0012380306143313646,
-      "learning_rate": 0.00019999823556448364,
-      "loss": 46.0,
-      "step": 11752
-    },
-    {
-      "epoch": 1.892709046257901,
-      "grad_norm": 0.0003440381260588765,
-      "learning_rate": 0.00019999823526393768,
-      "loss": 46.0,
-      "step": 11753
-    },
-    {
-      "epoch": 1.8928700833366883,
-      "grad_norm": 0.002382012316957116,
-      "learning_rate": 0.0001999982349633661,
-      "loss": 46.0,
-      "step": 11754
-    },
-    {
-      "epoch": 1.8930311204154755,
-      "grad_norm": 0.0012605522060766816,
-      "learning_rate": 0.00019999823466276896,
-      "loss": 46.0,
-      "step": 11755
-    },
-    {
-      "epoch": 1.893192157494263,
-      "grad_norm": 0.008916929364204407,
-      "learning_rate": 0.00019999823436214618,
-      "loss": 46.0,
-      "step": 11756
-    },
-    {
-      "epoch": 1.8933531945730504,
-      "grad_norm": 0.0017264783382415771,
-      "learning_rate": 0.00019999823406149784,
-      "loss": 46.0,
-      "step": 11757
-    },
-    {
-      "epoch": 1.8935142316518379,
-      "grad_norm": 0.0005386592820286751,
-      "learning_rate": 0.00019999823376082385,
-      "loss": 46.0,
-      "step": 11758
-    },
-    {
-      "epoch": 1.8936752687306253,
-      "grad_norm": 0.004780413582921028,
-      "learning_rate": 0.00019999823346012428,
-      "loss": 46.0,
-      "step": 11759
-    },
-    {
-      "epoch": 1.8938363058094128,
-      "grad_norm": 0.002630789065733552,
-      "learning_rate": 0.00019999823315939916,
-      "loss": 46.0,
-      "step": 11760
-    },
-    {
-      "epoch": 1.8939973428882,
-      "grad_norm": 0.004380307625979185,
-      "learning_rate": 0.00019999823285864844,
-      "loss": 46.0,
-      "step": 11761
-    },
-    {
-      "epoch": 1.8941583799669874,
-      "grad_norm": 0.004099314566701651,
-      "learning_rate": 0.00019999823255787208,
-      "loss": 46.0,
-      "step": 11762
-    },
-    {
-      "epoch": 1.8943194170457747,
-      "grad_norm": 0.0047810859978199005,
-      "learning_rate": 0.00019999823225707016,
-      "loss": 46.0,
-      "step": 11763
-    },
-    {
-      "epoch": 1.8944804541245621,
-      "grad_norm": 0.004029067233204842,
-      "learning_rate": 0.00019999823195624266,
-      "loss": 46.0,
-      "step": 11764
-    },
-    {
-      "epoch": 1.8946414912033496,
-      "grad_norm": 0.000969112734310329,
-      "learning_rate": 0.00019999823165538954,
-      "loss": 46.0,
-      "step": 11765
-    },
-    {
-      "epoch": 1.894802528282137,
-      "grad_norm": 0.0019878216553479433,
-      "learning_rate": 0.00019999823135451083,
-      "loss": 46.0,
-      "step": 11766
-    },
-    {
-      "epoch": 1.8949635653609245,
-      "grad_norm": 0.00527028227224946,
-      "learning_rate": 0.0001999982310536065,
-      "loss": 46.0,
-      "step": 11767
-    },
-    {
-      "epoch": 1.8951246024397117,
-      "grad_norm": 0.001325234305113554,
-      "learning_rate": 0.0001999982307526766,
-      "loss": 46.0,
-      "step": 11768
-    },
-    {
-      "epoch": 1.8952856395184992,
-      "grad_norm": 0.0025457125157117844,
-      "learning_rate": 0.00019999823045172113,
-      "loss": 46.0,
-      "step": 11769
-    },
-    {
-      "epoch": 1.8954466765972864,
-      "grad_norm": 0.0024882894940674305,
-      "learning_rate": 0.00019999823015074005,
-      "loss": 46.0,
-      "step": 11770
-    },
-    {
-      "epoch": 1.8956077136760738,
-      "grad_norm": 0.0008611671510152519,
-      "learning_rate": 0.00019999822984973338,
-      "loss": 46.0,
-      "step": 11771
-    },
-    {
-      "epoch": 1.8957687507548613,
-      "grad_norm": 0.0015332571929320693,
-      "learning_rate": 0.0001999982295487011,
-      "loss": 46.0,
-      "step": 11772
-    },
-    {
-      "epoch": 1.8959297878336487,
-      "grad_norm": 0.0038195359520614147,
-      "learning_rate": 0.00019999822924764323,
-      "loss": 46.0,
-      "step": 11773
-    },
-    {
-      "epoch": 1.8960908249124362,
-      "grad_norm": 0.0017652831738814712,
-      "learning_rate": 0.00019999822894655974,
-      "loss": 46.0,
-      "step": 11774
-    },
-    {
-      "epoch": 1.8962518619912236,
-      "grad_norm": 0.0010880673071369529,
-      "learning_rate": 0.0001999982286454507,
-      "loss": 46.0,
-      "step": 11775
-    },
-    {
-      "epoch": 1.8964128990700109,
-      "grad_norm": 0.0014009218430146575,
-      "learning_rate": 0.00019999822834431604,
-      "loss": 46.0,
-      "step": 11776
-    },
-    {
-      "epoch": 1.896573936148798,
-      "grad_norm": 0.0028121969662606716,
-      "learning_rate": 0.0001999982280431558,
-      "loss": 46.0,
-      "step": 11777
-    },
-    {
-      "epoch": 1.8967349732275856,
-      "grad_norm": 0.002201802795752883,
-      "learning_rate": 0.00019999822774196995,
-      "loss": 46.0,
-      "step": 11778
-    },
-    {
-      "epoch": 1.896896010306373,
-      "grad_norm": 0.0013814257690683007,
-      "learning_rate": 0.0001999982274407585,
-      "loss": 46.0,
-      "step": 11779
-    },
-    {
-      "epoch": 1.8970570473851605,
-      "grad_norm": 0.0011241371976211667,
-      "learning_rate": 0.00019999822713952147,
-      "loss": 46.0,
-      "step": 11780
-    },
-    {
-      "epoch": 1.897218084463948,
-      "grad_norm": 0.00313038588501513,
-      "learning_rate": 0.00019999822683825885,
-      "loss": 46.0,
-      "step": 11781
-    },
-    {
-      "epoch": 1.8973791215427354,
-      "grad_norm": 0.0045711323618888855,
-      "learning_rate": 0.00019999822653697064,
-      "loss": 46.0,
-      "step": 11782
-    },
-    {
-      "epoch": 1.8975401586215226,
-      "grad_norm": 0.0038199895061552525,
-      "learning_rate": 0.00019999822623565681,
-      "loss": 46.0,
-      "step": 11783
-    },
-    {
-      "epoch": 1.89770119570031,
-      "grad_norm": 0.0032299107406288385,
-      "learning_rate": 0.0001999982259343174,
-      "loss": 46.0,
-      "step": 11784
-    },
-    {
-      "epoch": 1.8978622327790973,
-      "grad_norm": 0.00629310542717576,
-      "learning_rate": 0.0001999982256329524,
-      "loss": 46.0,
-      "step": 11785
-    },
-    {
-      "epoch": 1.8980232698578847,
-      "grad_norm": 0.0045644245110452175,
-      "learning_rate": 0.00019999822533156182,
-      "loss": 46.0,
-      "step": 11786
-    },
-    {
-      "epoch": 1.8981843069366722,
-      "grad_norm": 0.0014680320164188743,
-      "learning_rate": 0.00019999822503014562,
-      "loss": 46.0,
-      "step": 11787
-    },
-    {
-      "epoch": 1.8983453440154596,
-      "grad_norm": 0.0023908084258437157,
-      "learning_rate": 0.00019999822472870383,
-      "loss": 46.0,
-      "step": 11788
-    },
-    {
-      "epoch": 1.898506381094247,
-      "grad_norm": 0.001071223639883101,
-      "learning_rate": 0.00019999822442723643,
-      "loss": 46.0,
-      "step": 11789
-    },
-    {
-      "epoch": 1.8986674181730343,
-      "grad_norm": 0.0025115911848843098,
-      "learning_rate": 0.00019999822412574347,
-      "loss": 46.0,
-      "step": 11790
-    },
-    {
-      "epoch": 1.8988284552518218,
-      "grad_norm": 0.0007221614359878004,
-      "learning_rate": 0.0001999982238242249,
-      "loss": 46.0,
-      "step": 11791
-    },
-    {
-      "epoch": 1.898989492330609,
-      "grad_norm": 0.0032353224232792854,
-      "learning_rate": 0.00019999822352268073,
-      "loss": 46.0,
-      "step": 11792
-    },
-    {
-      "epoch": 1.8991505294093964,
-      "grad_norm": 0.000811191217508167,
-      "learning_rate": 0.00019999822322111098,
-      "loss": 46.0,
-      "step": 11793
-    },
-    {
-      "epoch": 1.8993115664881839,
-      "grad_norm": 0.0025089846458286047,
-      "learning_rate": 0.00019999822291951564,
-      "loss": 46.0,
-      "step": 11794
-    },
-    {
-      "epoch": 1.8994726035669713,
-      "grad_norm": 0.0027704325038939714,
-      "learning_rate": 0.0001999982226178947,
-      "loss": 46.0,
-      "step": 11795
-    },
-    {
-      "epoch": 1.8996336406457588,
-      "grad_norm": 0.008282319642603397,
-      "learning_rate": 0.00019999822231624815,
-      "loss": 46.0,
-      "step": 11796
-    },
-    {
-      "epoch": 1.899794677724546,
-      "grad_norm": 0.0018473841482773423,
-      "learning_rate": 0.000199998222014576,
-      "loss": 46.0,
-      "step": 11797
-    },
-    {
-      "epoch": 1.8999557148033335,
-      "grad_norm": 0.001742801396176219,
-      "learning_rate": 0.0001999982217128783,
-      "loss": 46.0,
-      "step": 11798
-    },
-    {
-      "epoch": 1.9001167518821207,
-      "grad_norm": 0.001645389012992382,
-      "learning_rate": 0.00019999822141115496,
-      "loss": 46.0,
-      "step": 11799
-    },
-    {
-      "epoch": 1.9002777889609082,
-      "grad_norm": 0.0004979412187822163,
-      "learning_rate": 0.00019999822110940602,
-      "loss": 46.0,
-      "step": 11800
-    },
-    {
-      "epoch": 1.9004388260396956,
-      "grad_norm": 0.002033951925113797,
-      "learning_rate": 0.00019999822080763152,
-      "loss": 46.0,
-      "step": 11801
-    },
-    {
-      "epoch": 1.900599863118483,
-      "grad_norm": 0.0012474007671698928,
-      "learning_rate": 0.0001999982205058314,
-      "loss": 46.0,
-      "step": 11802
-    },
-    {
-      "epoch": 1.9007609001972705,
-      "grad_norm": 0.002050091279670596,
-      "learning_rate": 0.0001999982202040057,
-      "loss": 46.0,
-      "step": 11803
-    },
-    {
-      "epoch": 1.900921937276058,
-      "grad_norm": 0.0013502688379958272,
-      "learning_rate": 0.0001999982199021544,
-      "loss": 46.0,
-      "step": 11804
-    },
-    {
-      "epoch": 1.9010829743548452,
-      "grad_norm": 0.0049523357301950455,
-      "learning_rate": 0.0001999982196002775,
-      "loss": 46.0,
-      "step": 11805
-    },
-    {
-      "epoch": 1.9012440114336326,
-      "grad_norm": 0.0007400332833640277,
-      "learning_rate": 0.000199998219298375,
-      "loss": 46.0,
-      "step": 11806
-    },
-    {
-      "epoch": 1.9014050485124199,
-      "grad_norm": 0.0011523393914103508,
-      "learning_rate": 0.00019999821899644696,
-      "loss": 46.0,
-      "step": 11807
-    },
-    {
-      "epoch": 1.9015660855912073,
-      "grad_norm": 0.0010133797768503428,
-      "learning_rate": 0.00019999821869449327,
-      "loss": 46.0,
-      "step": 11808
-    },
-    {
-      "epoch": 1.9017271226699948,
-      "grad_norm": 0.010040556080639362,
-      "learning_rate": 0.000199998218392514,
-      "loss": 46.0,
-      "step": 11809
-    },
-    {
-      "epoch": 1.9018881597487822,
-      "grad_norm": 0.0024087845813483,
-      "learning_rate": 0.00019999821809050914,
-      "loss": 46.0,
-      "step": 11810
-    },
-    {
-      "epoch": 1.9020491968275697,
-      "grad_norm": 0.006123644299805164,
-      "learning_rate": 0.00019999821778847866,
-      "loss": 46.0,
-      "step": 11811
-    },
-    {
-      "epoch": 1.902210233906357,
-      "grad_norm": 0.006247910670936108,
-      "learning_rate": 0.00019999821748642265,
-      "loss": 46.0,
-      "step": 11812
-    },
-    {
-      "epoch": 1.9023712709851444,
-      "grad_norm": 0.0018093495164066553,
-      "learning_rate": 0.000199998217184341,
-      "loss": 46.0,
-      "step": 11813
-    },
-    {
-      "epoch": 1.9025323080639316,
-      "grad_norm": 0.0012550739338621497,
-      "learning_rate": 0.00019999821688223372,
-      "loss": 46.0,
-      "step": 11814
-    },
-    {
-      "epoch": 1.902693345142719,
-      "grad_norm": 0.0013865187065675855,
-      "learning_rate": 0.00019999821658010092,
-      "loss": 46.0,
-      "step": 11815
-    },
-    {
-      "epoch": 1.9028543822215065,
-      "grad_norm": 0.002715684939175844,
-      "learning_rate": 0.00019999821627794247,
-      "loss": 46.0,
-      "step": 11816
-    },
-    {
-      "epoch": 1.903015419300294,
-      "grad_norm": 0.001148345647379756,
-      "learning_rate": 0.00019999821597575844,
-      "loss": 46.0,
-      "step": 11817
-    },
-    {
-      "epoch": 1.9031764563790814,
-      "grad_norm": 0.003584346268326044,
-      "learning_rate": 0.00019999821567354885,
-      "loss": 46.0,
-      "step": 11818
-    },
-    {
-      "epoch": 1.9033374934578686,
-      "grad_norm": 0.004240552894771099,
-      "learning_rate": 0.00019999821537131362,
-      "loss": 46.0,
-      "step": 11819
-    },
-    {
-      "epoch": 1.903498530536656,
-      "grad_norm": 0.0011552739888429642,
-      "learning_rate": 0.0001999982150690528,
-      "loss": 46.0,
-      "step": 11820
-    },
-    {
-      "epoch": 1.9036595676154433,
-      "grad_norm": 0.0034946631640195847,
-      "learning_rate": 0.00019999821476676642,
-      "loss": 46.0,
-      "step": 11821
-    },
-    {
-      "epoch": 1.9038206046942308,
-      "grad_norm": 0.0022506811656057835,
-      "learning_rate": 0.00019999821446445442,
-      "loss": 46.0,
-      "step": 11822
-    },
-    {
-      "epoch": 1.9039816417730182,
-      "grad_norm": 0.0007530305301770568,
-      "learning_rate": 0.00019999821416211684,
-      "loss": 46.0,
-      "step": 11823
-    },
-    {
-      "epoch": 1.9041426788518057,
-      "grad_norm": 0.0020015668123960495,
-      "learning_rate": 0.00019999821385975364,
-      "loss": 46.0,
-      "step": 11824
-    },
-    {
-      "epoch": 1.904303715930593,
-      "grad_norm": 0.0006485689664259553,
-      "learning_rate": 0.00019999821355736486,
-      "loss": 46.0,
-      "step": 11825
-    },
-    {
-      "epoch": 1.9044647530093806,
-      "grad_norm": 0.00039832803304307163,
-      "learning_rate": 0.0001999982132549505,
-      "loss": 46.0,
-      "step": 11826
-    },
-    {
-      "epoch": 1.9046257900881678,
-      "grad_norm": 0.0047916872426867485,
-      "learning_rate": 0.0001999982129525105,
-      "loss": 46.0,
-      "step": 11827
-    },
-    {
-      "epoch": 1.9047868271669552,
-      "grad_norm": 0.00035810013650916517,
-      "learning_rate": 0.00019999821265004496,
-      "loss": 46.0,
-      "step": 11828
-    },
-    {
-      "epoch": 1.9049478642457425,
-      "grad_norm": 0.0009615529561415315,
-      "learning_rate": 0.0001999982123475538,
-      "loss": 46.0,
-      "step": 11829
-    },
-    {
-      "epoch": 1.90510890132453,
-      "grad_norm": 0.001052516046911478,
-      "learning_rate": 0.00019999821204503702,
-      "loss": 46.0,
-      "step": 11830
-    },
-    {
-      "epoch": 1.9052699384033174,
-      "grad_norm": 0.0010156475473195314,
-      "learning_rate": 0.0001999982117424947,
-      "loss": 46.0,
-      "step": 11831
-    },
-    {
-      "epoch": 1.9054309754821048,
-      "grad_norm": 0.0048035853542387486,
-      "learning_rate": 0.00019999821143992677,
-      "loss": 46.0,
-      "step": 11832
-    },
-    {
-      "epoch": 1.9055920125608923,
-      "grad_norm": 0.007428744807839394,
-      "learning_rate": 0.0001999982111373332,
-      "loss": 46.0,
-      "step": 11833
-    },
-    {
-      "epoch": 1.9057530496396795,
-      "grad_norm": 0.0004149885498918593,
-      "learning_rate": 0.00019999821083471409,
-      "loss": 46.0,
-      "step": 11834
-    },
-    {
-      "epoch": 1.905914086718467,
-      "grad_norm": 0.0016499670455232263,
-      "learning_rate": 0.00019999821053206938,
-      "loss": 46.0,
-      "step": 11835
-    },
-    {
-      "epoch": 1.9060751237972542,
-      "grad_norm": 0.0011334343580529094,
-      "learning_rate": 0.00019999821022939905,
-      "loss": 46.0,
-      "step": 11836
-    },
-    {
-      "epoch": 1.9062361608760416,
-      "grad_norm": 0.0021435662638396025,
-      "learning_rate": 0.00019999820992670314,
-      "loss": 46.0,
-      "step": 11837
-    },
-    {
-      "epoch": 1.906397197954829,
-      "grad_norm": 0.0036536231637001038,
-      "learning_rate": 0.00019999820962398164,
-      "loss": 46.0,
-      "step": 11838
-    },
-    {
-      "epoch": 1.9065582350336165,
-      "grad_norm": 0.005240217316895723,
-      "learning_rate": 0.00019999820932123453,
-      "loss": 46.0,
-      "step": 11839
-    },
-    {
-      "epoch": 1.906719272112404,
-      "grad_norm": 0.004845750518143177,
-      "learning_rate": 0.00019999820901846183,
-      "loss": 46.0,
-      "step": 11840
-    },
-    {
-      "epoch": 1.9068803091911912,
-      "grad_norm": 0.008263902738690376,
-      "learning_rate": 0.00019999820871566355,
-      "loss": 46.0,
-      "step": 11841
-    },
-    {
-      "epoch": 1.9070413462699787,
-      "grad_norm": 0.0008414392941631377,
-      "learning_rate": 0.00019999820841283964,
-      "loss": 46.0,
-      "step": 11842
-    },
-    {
-      "epoch": 1.907202383348766,
-      "grad_norm": 0.002106528962031007,
-      "learning_rate": 0.00019999820810999018,
-      "loss": 46.0,
-      "step": 11843
-    },
-    {
-      "epoch": 1.9073634204275534,
-      "grad_norm": 0.000593257718719542,
-      "learning_rate": 0.0001999982078071151,
-      "loss": 46.0,
-      "step": 11844
-    },
-    {
-      "epoch": 1.9075244575063408,
-      "grad_norm": 0.0013861947227269411,
-      "learning_rate": 0.00019999820750421442,
-      "loss": 46.0,
-      "step": 11845
-    },
-    {
-      "epoch": 1.9076854945851283,
-      "grad_norm": 0.0005206508794799447,
-      "learning_rate": 0.00019999820720128817,
-      "loss": 46.0,
-      "step": 11846
-    },
-    {
-      "epoch": 1.9078465316639157,
-      "grad_norm": 0.00434001674875617,
-      "learning_rate": 0.00019999820689833633,
-      "loss": 46.0,
-      "step": 11847
-    },
-    {
-      "epoch": 1.9080075687427032,
-      "grad_norm": 0.007575648836791515,
-      "learning_rate": 0.00019999820659535885,
-      "loss": 46.0,
-      "step": 11848
-    },
-    {
-      "epoch": 1.9081686058214904,
-      "grad_norm": 0.002601374639198184,
-      "learning_rate": 0.00019999820629235582,
-      "loss": 46.0,
-      "step": 11849
-    },
-    {
-      "epoch": 1.9083296429002776,
-      "grad_norm": 0.00039806030690670013,
-      "learning_rate": 0.0001999982059893272,
-      "loss": 46.0,
-      "step": 11850
-    },
-    {
-      "epoch": 1.908490679979065,
-      "grad_norm": 0.0015142896445468068,
-      "learning_rate": 0.00019999820568627295,
-      "loss": 46.0,
-      "step": 11851
-    },
-    {
-      "epoch": 1.9086517170578525,
-      "grad_norm": 0.003050045343115926,
-      "learning_rate": 0.00019999820538319312,
-      "loss": 46.0,
-      "step": 11852
-    },
-    {
-      "epoch": 1.90881275413664,
-      "grad_norm": 0.0016080980421975255,
-      "learning_rate": 0.00019999820508008768,
-      "loss": 46.0,
-      "step": 11853
-    },
-    {
-      "epoch": 1.9089737912154274,
-      "grad_norm": 0.005588393658399582,
-      "learning_rate": 0.00019999820477695668,
-      "loss": 46.0,
-      "step": 11854
-    },
-    {
-      "epoch": 1.9091348282942149,
-      "grad_norm": 0.0015004201559349895,
-      "learning_rate": 0.00019999820447380007,
-      "loss": 46.0,
-      "step": 11855
-    },
-    {
-      "epoch": 1.909295865373002,
-      "grad_norm": 0.0038349470123648643,
-      "learning_rate": 0.00019999820417061786,
-      "loss": 46.0,
-      "step": 11856
-    },
-    {
-      "epoch": 1.9094569024517896,
-      "grad_norm": 0.0018715556943789124,
-      "learning_rate": 0.00019999820386741007,
-      "loss": 46.0,
-      "step": 11857
-    },
-    {
-      "epoch": 1.9096179395305768,
-      "grad_norm": 0.0017635863041505218,
-      "learning_rate": 0.00019999820356417667,
-      "loss": 46.0,
-      "step": 11858
-    },
-    {
-      "epoch": 1.9097789766093642,
-      "grad_norm": 0.0016591364983469248,
-      "learning_rate": 0.00019999820326091768,
-      "loss": 46.0,
-      "step": 11859
-    },
-    {
-      "epoch": 1.9099400136881517,
-      "grad_norm": 0.00836408045142889,
-      "learning_rate": 0.0001999982029576331,
-      "loss": 46.0,
-      "step": 11860
-    },
-    {
-      "epoch": 1.9101010507669391,
-      "grad_norm": 0.0060034324415028095,
-      "learning_rate": 0.0001999982026543229,
-      "loss": 46.0,
-      "step": 11861
-    },
-    {
-      "epoch": 1.9102620878457266,
-      "grad_norm": 0.0016696983948349953,
-      "learning_rate": 0.00019999820235098713,
-      "loss": 46.0,
-      "step": 11862
-    },
-    {
-      "epoch": 1.9104231249245138,
-      "grad_norm": 0.009441199712455273,
-      "learning_rate": 0.00019999820204762576,
-      "loss": 46.0,
-      "step": 11863
-    },
-    {
-      "epoch": 1.9105841620033013,
-      "grad_norm": 0.0008062032284215093,
-      "learning_rate": 0.0001999982017442388,
-      "loss": 46.0,
-      "step": 11864
-    },
-    {
-      "epoch": 1.9107451990820885,
-      "grad_norm": 0.0006342515698634088,
-      "learning_rate": 0.00019999820144082624,
-      "loss": 46.0,
-      "step": 11865
-    },
-    {
-      "epoch": 1.910906236160876,
-      "grad_norm": 0.0013712665531784296,
-      "learning_rate": 0.0001999982011373881,
-      "loss": 46.0,
-      "step": 11866
-    },
-    {
-      "epoch": 1.9110672732396634,
-      "grad_norm": 0.007080291397869587,
-      "learning_rate": 0.00019999820083392437,
-      "loss": 46.0,
-      "step": 11867
-    },
-    {
-      "epoch": 1.9112283103184509,
-      "grad_norm": 0.002430324675515294,
-      "learning_rate": 0.000199998200530435,
-      "loss": 46.0,
-      "step": 11868
-    },
-    {
-      "epoch": 1.9113893473972383,
-      "grad_norm": 0.000615751720033586,
-      "learning_rate": 0.00019999820022692006,
-      "loss": 46.0,
-      "step": 11869
-    },
-    {
-      "epoch": 1.9115503844760258,
-      "grad_norm": 0.005030791740864515,
-      "learning_rate": 0.00019999819992337953,
-      "loss": 46.0,
-      "step": 11870
-    },
-    {
-      "epoch": 1.911711421554813,
-      "grad_norm": 0.0007704586023464799,
-      "learning_rate": 0.00019999819961981344,
-      "loss": 46.0,
-      "step": 11871
-    },
-    {
-      "epoch": 1.9118724586336002,
-      "grad_norm": 0.0034569334238767624,
-      "learning_rate": 0.0001999981993162217,
-      "loss": 46.0,
-      "step": 11872
-    },
-    {
-      "epoch": 1.9120334957123877,
-      "grad_norm": 0.006942248437553644,
-      "learning_rate": 0.0001999981990126044,
-      "loss": 46.0,
-      "step": 11873
-    },
-    {
-      "epoch": 1.9121945327911751,
-      "grad_norm": 0.001468125730752945,
-      "learning_rate": 0.0001999981987089615,
-      "loss": 46.0,
-      "step": 11874
-    },
-    {
-      "epoch": 1.9123555698699626,
-      "grad_norm": 0.0004956931807100773,
-      "learning_rate": 0.000199998198405293,
-      "loss": 46.0,
-      "step": 11875
-    },
-    {
-      "epoch": 1.91251660694875,
-      "grad_norm": 0.000587780843488872,
-      "learning_rate": 0.0001999981981015989,
-      "loss": 46.0,
-      "step": 11876
-    },
-    {
-      "epoch": 1.9126776440275375,
-      "grad_norm": 0.0024365379940718412,
-      "learning_rate": 0.0001999981977978792,
-      "loss": 46.0,
-      "step": 11877
-    },
-    {
-      "epoch": 1.9128386811063247,
-      "grad_norm": 0.0019610903691500425,
-      "learning_rate": 0.00019999819749413393,
-      "loss": 46.0,
-      "step": 11878
-    },
-    {
-      "epoch": 1.9129997181851122,
-      "grad_norm": 0.0007827310473658144,
-      "learning_rate": 0.00019999819719036306,
-      "loss": 46.0,
-      "step": 11879
-    },
-    {
-      "epoch": 1.9131607552638994,
-      "grad_norm": 0.0012719838414341211,
-      "learning_rate": 0.00019999819688656657,
-      "loss": 46.0,
-      "step": 11880
-    },
-    {
-      "epoch": 1.9133217923426868,
-      "grad_norm": 0.0010255719535052776,
-      "learning_rate": 0.00019999819658274453,
-      "loss": 46.0,
-      "step": 11881
-    },
-    {
-      "epoch": 1.9134828294214743,
-      "grad_norm": 0.0011179463472217321,
-      "learning_rate": 0.00019999819627889684,
-      "loss": 46.0,
-      "step": 11882
-    },
-    {
-      "epoch": 1.9136438665002617,
-      "grad_norm": 0.0008510363404639065,
-      "learning_rate": 0.0001999981959750236,
-      "loss": 46.0,
-      "step": 11883
-    },
-    {
-      "epoch": 1.9138049035790492,
-      "grad_norm": 0.0035883383825421333,
-      "learning_rate": 0.00019999819567112473,
-      "loss": 46.0,
-      "step": 11884
-    },
-    {
-      "epoch": 1.9139659406578364,
-      "grad_norm": 0.0015765272546559572,
-      "learning_rate": 0.0001999981953672003,
-      "loss": 46.0,
-      "step": 11885
-    },
-    {
-      "epoch": 1.9141269777366239,
-      "grad_norm": 0.006658013444393873,
-      "learning_rate": 0.00019999819506325025,
-      "loss": 46.0,
-      "step": 11886
-    },
-    {
-      "epoch": 1.914288014815411,
-      "grad_norm": 0.0028836934361606836,
-      "learning_rate": 0.0001999981947592746,
-      "loss": 46.0,
-      "step": 11887
-    },
-    {
-      "epoch": 1.9144490518941986,
-      "grad_norm": 0.008031723089516163,
-      "learning_rate": 0.0001999981944552734,
-      "loss": 46.0,
-      "step": 11888
-    },
-    {
-      "epoch": 1.914610088972986,
-      "grad_norm": 0.0014731878181919456,
-      "learning_rate": 0.0001999981941512466,
-      "loss": 46.0,
-      "step": 11889
-    },
-    {
-      "epoch": 1.9147711260517735,
-      "grad_norm": 0.006518073845654726,
-      "learning_rate": 0.00019999819384719416,
-      "loss": 46.0,
-      "step": 11890
-    },
-    {
-      "epoch": 1.914932163130561,
-      "grad_norm": 0.0006891588564030826,
-      "learning_rate": 0.00019999819354311616,
-      "loss": 46.0,
-      "step": 11891
-    },
-    {
-      "epoch": 1.9150932002093484,
-      "grad_norm": 0.004762472119182348,
-      "learning_rate": 0.00019999819323901255,
-      "loss": 46.0,
-      "step": 11892
-    },
-    {
-      "epoch": 1.9152542372881356,
-      "grad_norm": 0.007304450962692499,
-      "learning_rate": 0.00019999819293488335,
-      "loss": 46.0,
-      "step": 11893
-    },
-    {
-      "epoch": 1.9154152743669228,
-      "grad_norm": 0.0039252061396837234,
-      "learning_rate": 0.00019999819263072857,
-      "loss": 46.0,
-      "step": 11894
-    },
-    {
-      "epoch": 1.9155763114457103,
-      "grad_norm": 0.0012136398581787944,
-      "learning_rate": 0.00019999819232654814,
-      "loss": 46.0,
-      "step": 11895
-    },
-    {
-      "epoch": 1.9157373485244977,
-      "grad_norm": 0.0037531491834670305,
-      "learning_rate": 0.00019999819202234218,
-      "loss": 46.0,
-      "step": 11896
-    },
-    {
-      "epoch": 1.9158983856032852,
-      "grad_norm": 0.0005400461377575994,
-      "learning_rate": 0.0001999981917181106,
-      "loss": 46.0,
-      "step": 11897
-    },
-    {
-      "epoch": 1.9160594226820726,
-      "grad_norm": 0.017477765679359436,
-      "learning_rate": 0.00019999819141385344,
-      "loss": 46.0,
-      "step": 11898
-    },
-    {
-      "epoch": 1.91622045976086,
-      "grad_norm": 0.002094994531944394,
-      "learning_rate": 0.00019999819110957067,
-      "loss": 46.0,
-      "step": 11899
-    },
-    {
-      "epoch": 1.9163814968396473,
-      "grad_norm": 0.002632502233609557,
-      "learning_rate": 0.0001999981908052623,
-      "loss": 46.0,
-      "step": 11900
-    },
-    {
-      "epoch": 1.9165425339184348,
-      "grad_norm": 0.0007712795049883425,
-      "learning_rate": 0.00019999819050092835,
-      "loss": 46.0,
-      "step": 11901
-    },
-    {
-      "epoch": 1.916703570997222,
-      "grad_norm": 0.002002400578930974,
-      "learning_rate": 0.0001999981901965688,
-      "loss": 46.0,
-      "step": 11902
-    },
-    {
-      "epoch": 1.9168646080760094,
-      "grad_norm": 0.006843820679932833,
-      "learning_rate": 0.00019999818989218366,
-      "loss": 46.0,
-      "step": 11903
-    },
-    {
-      "epoch": 1.917025645154797,
-      "grad_norm": 0.0019006015500053763,
-      "learning_rate": 0.00019999818958777292,
-      "loss": 46.0,
-      "step": 11904
-    },
-    {
-      "epoch": 1.9171866822335843,
-      "grad_norm": 0.0035611672792583704,
-      "learning_rate": 0.00019999818928333657,
-      "loss": 46.0,
-      "step": 11905
-    },
-    {
-      "epoch": 1.9173477193123718,
-      "grad_norm": 0.0006191295105963945,
-      "learning_rate": 0.00019999818897887465,
-      "loss": 46.0,
-      "step": 11906
-    },
-    {
-      "epoch": 1.917508756391159,
-      "grad_norm": 0.004043434746563435,
-      "learning_rate": 0.00019999818867438713,
-      "loss": 46.0,
-      "step": 11907
-    },
-    {
-      "epoch": 1.9176697934699465,
-      "grad_norm": 0.0005644129705615342,
-      "learning_rate": 0.000199998188369874,
-      "loss": 46.0,
-      "step": 11908
-    },
-    {
-      "epoch": 1.9178308305487337,
-      "grad_norm": 0.0009777158265933394,
-      "learning_rate": 0.0001999981880653353,
-      "loss": 46.0,
-      "step": 11909
-    },
-    {
-      "epoch": 1.9179918676275212,
-      "grad_norm": 0.0033065034076571465,
-      "learning_rate": 0.00019999818776077102,
-      "loss": 46.0,
-      "step": 11910
-    },
-    {
-      "epoch": 1.9181529047063086,
-      "grad_norm": 0.003878120332956314,
-      "learning_rate": 0.00019999818745618112,
-      "loss": 46.0,
-      "step": 11911
-    },
-    {
-      "epoch": 1.918313941785096,
-      "grad_norm": 0.0016693917568773031,
-      "learning_rate": 0.00019999818715156563,
-      "loss": 46.0,
-      "step": 11912
-    },
-    {
-      "epoch": 1.9184749788638835,
-      "grad_norm": 0.0008196932612918317,
-      "learning_rate": 0.00019999818684692452,
-      "loss": 46.0,
-      "step": 11913
-    },
-    {
-      "epoch": 1.9186360159426707,
-      "grad_norm": 0.00625289510935545,
-      "learning_rate": 0.00019999818654225786,
-      "loss": 46.0,
-      "step": 11914
-    },
-    {
-      "epoch": 1.9187970530214582,
-      "grad_norm": 0.0007819284801371396,
-      "learning_rate": 0.00019999818623756558,
-      "loss": 46.0,
-      "step": 11915
-    },
-    {
-      "epoch": 1.9189580901002454,
-      "grad_norm": 0.005089349113404751,
-      "learning_rate": 0.00019999818593284768,
-      "loss": 46.0,
-      "step": 11916
-    },
-    {
-      "epoch": 1.9191191271790329,
-      "grad_norm": 0.0016359409783035517,
-      "learning_rate": 0.00019999818562810423,
-      "loss": 46.0,
-      "step": 11917
-    },
-    {
-      "epoch": 1.9192801642578203,
-      "grad_norm": 0.007776940241456032,
-      "learning_rate": 0.00019999818532333516,
-      "loss": 46.0,
-      "step": 11918
-    },
-    {
-      "epoch": 1.9194412013366078,
-      "grad_norm": 0.0012445615138858557,
-      "learning_rate": 0.00019999818501854053,
-      "loss": 46.0,
-      "step": 11919
-    },
-    {
-      "epoch": 1.9196022384153952,
-      "grad_norm": 0.012229708023369312,
-      "learning_rate": 0.00019999818471372026,
-      "loss": 46.0,
-      "step": 11920
-    },
-    {
-      "epoch": 1.9197632754941827,
-      "grad_norm": 0.0050226133316755295,
-      "learning_rate": 0.00019999818440887443,
-      "loss": 46.0,
-      "step": 11921
-    },
-    {
-      "epoch": 1.91992431257297,
-      "grad_norm": 0.0026193547528237104,
-      "learning_rate": 0.00019999818410400296,
-      "loss": 46.0,
-      "step": 11922
-    },
-    {
-      "epoch": 1.9200853496517574,
-      "grad_norm": 0.0010318810818716884,
-      "learning_rate": 0.00019999818379910596,
-      "loss": 46.0,
-      "step": 11923
-    },
-    {
-      "epoch": 1.9202463867305446,
-      "grad_norm": 0.0027994862757623196,
-      "learning_rate": 0.00019999818349418334,
-      "loss": 46.0,
-      "step": 11924
-    },
-    {
-      "epoch": 1.920407423809332,
-      "grad_norm": 0.0026863550301641226,
-      "learning_rate": 0.0001999981831892351,
-      "loss": 46.0,
-      "step": 11925
-    },
-    {
-      "epoch": 1.9205684608881195,
-      "grad_norm": 0.0015136533183977008,
-      "learning_rate": 0.0001999981828842613,
-      "loss": 46.0,
-      "step": 11926
-    },
-    {
-      "epoch": 1.920729497966907,
-      "grad_norm": 0.005652355030179024,
-      "learning_rate": 0.00019999818257926188,
-      "loss": 46.0,
-      "step": 11927
-    },
-    {
-      "epoch": 1.9208905350456944,
-      "grad_norm": 0.00524381035938859,
-      "learning_rate": 0.0001999981822742369,
-      "loss": 46.0,
-      "step": 11928
-    },
-    {
-      "epoch": 1.9210515721244816,
-      "grad_norm": 0.0036435918882489204,
-      "learning_rate": 0.0001999981819691863,
-      "loss": 46.0,
-      "step": 11929
-    },
-    {
-      "epoch": 1.921212609203269,
-      "grad_norm": 0.0010698448168113828,
-      "learning_rate": 0.00019999818166411008,
-      "loss": 46.0,
-      "step": 11930
-    },
-    {
-      "epoch": 1.9213736462820563,
-      "grad_norm": 0.003018536139279604,
-      "learning_rate": 0.0001999981813590083,
-      "loss": 46.0,
-      "step": 11931
-    },
-    {
-      "epoch": 1.9215346833608438,
-      "grad_norm": 0.0017038368387147784,
-      "learning_rate": 0.0001999981810538809,
-      "loss": 46.0,
-      "step": 11932
-    },
-    {
-      "epoch": 1.9216957204396312,
-      "grad_norm": 0.0009454243700020015,
-      "learning_rate": 0.00019999818074872795,
-      "loss": 46.0,
-      "step": 11933
-    },
-    {
-      "epoch": 1.9218567575184187,
-      "grad_norm": 0.004435803275555372,
-      "learning_rate": 0.00019999818044354938,
-      "loss": 46.0,
-      "step": 11934
-    },
-    {
-      "epoch": 1.9220177945972061,
-      "grad_norm": 0.0007561726961284876,
-      "learning_rate": 0.00019999818013834522,
-      "loss": 46.0,
-      "step": 11935
-    },
-    {
-      "epoch": 1.9221788316759933,
-      "grad_norm": 0.0016916776075959206,
-      "learning_rate": 0.00019999817983311547,
-      "loss": 46.0,
-      "step": 11936
-    },
-    {
-      "epoch": 1.9223398687547808,
-      "grad_norm": 0.0016145188128575683,
-      "learning_rate": 0.00019999817952786009,
-      "loss": 46.0,
-      "step": 11937
-    },
-    {
-      "epoch": 1.922500905833568,
-      "grad_norm": 0.0007525762193836272,
-      "learning_rate": 0.00019999817922257914,
-      "loss": 46.0,
-      "step": 11938
-    },
-    {
-      "epoch": 1.9226619429123555,
-      "grad_norm": 0.001972717000171542,
-      "learning_rate": 0.0001999981789172726,
-      "loss": 46.0,
-      "step": 11939
-    },
-    {
-      "epoch": 1.922822979991143,
-      "grad_norm": 0.001384893897920847,
-      "learning_rate": 0.00019999817861194048,
-      "loss": 46.0,
-      "step": 11940
-    },
-    {
-      "epoch": 1.9229840170699304,
-      "grad_norm": 0.0035784346982836723,
-      "learning_rate": 0.00019999817830658275,
-      "loss": 46.0,
-      "step": 11941
-    },
-    {
-      "epoch": 1.9231450541487178,
-      "grad_norm": 0.005776277277618647,
-      "learning_rate": 0.00019999817800119942,
-      "loss": 46.0,
-      "step": 11942
-    },
-    {
-      "epoch": 1.9233060912275053,
-      "grad_norm": 0.0026387604884803295,
-      "learning_rate": 0.0001999981776957905,
-      "loss": 46.0,
-      "step": 11943
-    },
-    {
-      "epoch": 1.9234671283062925,
-      "grad_norm": 0.0018318184884265065,
-      "learning_rate": 0.000199998177390356,
-      "loss": 46.0,
-      "step": 11944
-    },
-    {
-      "epoch": 1.92362816538508,
-      "grad_norm": 0.0019589755684137344,
-      "learning_rate": 0.00019999817708489588,
-      "loss": 46.0,
-      "step": 11945
-    },
-    {
-      "epoch": 1.9237892024638672,
-      "grad_norm": 0.0005852350150234997,
-      "learning_rate": 0.00019999817677941018,
-      "loss": 46.0,
-      "step": 11946
-    },
-    {
-      "epoch": 1.9239502395426546,
-      "grad_norm": 0.0018827073508873582,
-      "learning_rate": 0.00019999817647389887,
-      "loss": 46.0,
-      "step": 11947
-    },
-    {
-      "epoch": 1.924111276621442,
-      "grad_norm": 0.0016252314671874046,
-      "learning_rate": 0.000199998176168362,
-      "loss": 46.0,
-      "step": 11948
-    },
-    {
-      "epoch": 1.9242723137002296,
-      "grad_norm": 0.00324031594209373,
-      "learning_rate": 0.0001999981758627995,
-      "loss": 46.0,
-      "step": 11949
-    },
-    {
-      "epoch": 1.924433350779017,
-      "grad_norm": 0.0012846144381910563,
-      "learning_rate": 0.0001999981755572114,
-      "loss": 46.0,
-      "step": 11950
-    },
-    {
-      "epoch": 1.9245943878578042,
-      "grad_norm": 0.0031126427929848433,
-      "learning_rate": 0.00019999817525159774,
-      "loss": 46.0,
-      "step": 11951
-    },
-    {
-      "epoch": 1.9247554249365917,
-      "grad_norm": 0.0015743878902867436,
-      "learning_rate": 0.00019999817494595847,
-      "loss": 46.0,
-      "step": 11952
-    },
-    {
-      "epoch": 1.924916462015379,
-      "grad_norm": 0.0008324464433826506,
-      "learning_rate": 0.0001999981746402936,
-      "loss": 46.0,
-      "step": 11953
-    },
-    {
-      "epoch": 1.9250774990941664,
-      "grad_norm": 0.0020896082278341055,
-      "learning_rate": 0.00019999817433460316,
-      "loss": 46.0,
-      "step": 11954
-    },
-    {
-      "epoch": 1.9252385361729538,
-      "grad_norm": 0.004017847124487162,
-      "learning_rate": 0.0001999981740288871,
-      "loss": 46.0,
-      "step": 11955
-    },
-    {
-      "epoch": 1.9253995732517413,
-      "grad_norm": 0.002839105436578393,
-      "learning_rate": 0.00019999817372314544,
-      "loss": 46.0,
-      "step": 11956
-    },
-    {
-      "epoch": 1.9255606103305287,
-      "grad_norm": 0.0010540698422119021,
-      "learning_rate": 0.0001999981734173782,
-      "loss": 46.0,
-      "step": 11957
-    },
-    {
-      "epoch": 1.925721647409316,
-      "grad_norm": 0.0012023458257317543,
-      "learning_rate": 0.00019999817311158537,
-      "loss": 46.0,
-      "step": 11958
-    },
-    {
-      "epoch": 1.9258826844881034,
-      "grad_norm": 0.00403686985373497,
-      "learning_rate": 0.00019999817280576693,
-      "loss": 46.0,
-      "step": 11959
-    },
-    {
-      "epoch": 1.9260437215668906,
-      "grad_norm": 0.001765172346495092,
-      "learning_rate": 0.0001999981724999229,
-      "loss": 46.0,
-      "step": 11960
-    },
-    {
-      "epoch": 1.926204758645678,
-      "grad_norm": 0.004205659963190556,
-      "learning_rate": 0.0001999981721940533,
-      "loss": 46.0,
-      "step": 11961
-    },
-    {
-      "epoch": 1.9263657957244655,
-      "grad_norm": 0.0014108573086559772,
-      "learning_rate": 0.0001999981718881581,
-      "loss": 46.0,
-      "step": 11962
-    },
-    {
-      "epoch": 1.926526832803253,
-      "grad_norm": 0.0016446582740172744,
-      "learning_rate": 0.00019999817158223725,
-      "loss": 46.0,
-      "step": 11963
-    },
-    {
-      "epoch": 1.9266878698820404,
-      "grad_norm": 0.008473742753267288,
-      "learning_rate": 0.00019999817127629087,
-      "loss": 46.0,
-      "step": 11964
-    },
-    {
-      "epoch": 1.926848906960828,
-      "grad_norm": 0.0026354636065661907,
-      "learning_rate": 0.00019999817097031888,
-      "loss": 46.0,
-      "step": 11965
-    },
-    {
-      "epoch": 1.9270099440396151,
-      "grad_norm": 0.0021241875365376472,
-      "learning_rate": 0.0001999981706643213,
-      "loss": 46.0,
-      "step": 11966
-    },
-    {
-      "epoch": 1.9271709811184023,
-      "grad_norm": 0.0019948366098105907,
-      "learning_rate": 0.00019999817035829808,
-      "loss": 46.0,
-      "step": 11967
-    },
-    {
-      "epoch": 1.9273320181971898,
-      "grad_norm": 0.0019483129726722836,
-      "learning_rate": 0.00019999817005224933,
-      "loss": 46.0,
-      "step": 11968
-    },
-    {
-      "epoch": 1.9274930552759773,
-      "grad_norm": 0.001787984510883689,
-      "learning_rate": 0.00019999816974617494,
-      "loss": 46.0,
-      "step": 11969
-    },
-    {
-      "epoch": 1.9276540923547647,
-      "grad_norm": 0.001117595355026424,
-      "learning_rate": 0.00019999816944007499,
-      "loss": 46.0,
-      "step": 11970
-    },
-    {
-      "epoch": 1.9278151294335522,
-      "grad_norm": 0.003120437962934375,
-      "learning_rate": 0.00019999816913394942,
-      "loss": 46.0,
-      "step": 11971
-    },
-    {
-      "epoch": 1.9279761665123396,
-      "grad_norm": 0.003077798755839467,
-      "learning_rate": 0.00019999816882779824,
-      "loss": 46.0,
-      "step": 11972
-    },
-    {
-      "epoch": 1.9281372035911268,
-      "grad_norm": 0.001689390861429274,
-      "learning_rate": 0.0001999981685216215,
-      "loss": 46.0,
-      "step": 11973
-    },
-    {
-      "epoch": 1.9282982406699143,
-      "grad_norm": 0.0013024532236158848,
-      "learning_rate": 0.00019999816821541914,
-      "loss": 46.0,
-      "step": 11974
-    },
-    {
-      "epoch": 1.9284592777487015,
-      "grad_norm": 0.0056304638274014,
-      "learning_rate": 0.00019999816790919122,
-      "loss": 46.0,
-      "step": 11975
-    },
-    {
-      "epoch": 1.928620314827489,
-      "grad_norm": 0.007379127666354179,
-      "learning_rate": 0.0001999981676029377,
-      "loss": 46.0,
-      "step": 11976
-    },
-    {
-      "epoch": 1.9287813519062764,
-      "grad_norm": 0.00048339113709516823,
-      "learning_rate": 0.00019999816729665857,
-      "loss": 46.0,
-      "step": 11977
-    },
-    {
-      "epoch": 1.9289423889850639,
-      "grad_norm": 0.007068888284265995,
-      "learning_rate": 0.00019999816699035384,
-      "loss": 46.0,
-      "step": 11978
-    },
-    {
-      "epoch": 1.9291034260638513,
-      "grad_norm": 0.0009395984816364944,
-      "learning_rate": 0.00019999816668402352,
-      "loss": 46.0,
-      "step": 11979
-    },
-    {
-      "epoch": 1.9292644631426386,
-      "grad_norm": 0.0007488145492970943,
-      "learning_rate": 0.0001999981663776676,
-      "loss": 46.0,
-      "step": 11980
-    },
-    {
-      "epoch": 1.929425500221426,
-      "grad_norm": 0.0009382746065966785,
-      "learning_rate": 0.0001999981660712861,
-      "loss": 46.0,
-      "step": 11981
-    },
-    {
-      "epoch": 1.9295865373002132,
-      "grad_norm": 0.002310621552169323,
-      "learning_rate": 0.000199998165764879,
-      "loss": 46.0,
-      "step": 11982
-    },
-    {
-      "epoch": 1.9297475743790007,
-      "grad_norm": 0.0011846248526126146,
-      "learning_rate": 0.0001999981654584463,
-      "loss": 46.0,
-      "step": 11983
-    },
-    {
-      "epoch": 1.9299086114577881,
-      "grad_norm": 0.0026515538338571787,
-      "learning_rate": 0.00019999816515198798,
-      "loss": 46.0,
-      "step": 11984
-    },
-    {
-      "epoch": 1.9300696485365756,
-      "grad_norm": 0.003187221474945545,
-      "learning_rate": 0.00019999816484550414,
-      "loss": 46.0,
-      "step": 11985
-    },
-    {
-      "epoch": 1.930230685615363,
-      "grad_norm": 0.0017586504109203815,
-      "learning_rate": 0.00019999816453899466,
-      "loss": 46.0,
-      "step": 11986
-    },
-    {
-      "epoch": 1.9303917226941505,
-      "grad_norm": 0.0070524695329368114,
-      "learning_rate": 0.00019999816423245956,
-      "loss": 46.0,
-      "step": 11987
-    },
-    {
-      "epoch": 1.9305527597729377,
-      "grad_norm": 0.0031549998093396425,
-      "learning_rate": 0.0001999981639258989,
-      "loss": 46.0,
-      "step": 11988
-    },
-    {
-      "epoch": 1.930713796851725,
-      "grad_norm": 0.00520916236564517,
-      "learning_rate": 0.00019999816361931263,
-      "loss": 46.0,
-      "step": 11989
-    },
-    {
-      "epoch": 1.9308748339305124,
-      "grad_norm": 0.0011408912250772119,
-      "learning_rate": 0.00019999816331270077,
-      "loss": 46.0,
-      "step": 11990
-    },
-    {
-      "epoch": 1.9310358710092999,
-      "grad_norm": 0.001446875394321978,
-      "learning_rate": 0.00019999816300606332,
-      "loss": 46.0,
-      "step": 11991
-    },
-    {
-      "epoch": 1.9311969080880873,
-      "grad_norm": 0.0007984877447597682,
-      "learning_rate": 0.0001999981626994003,
-      "loss": 46.0,
-      "step": 11992
-    },
-    {
-      "epoch": 1.9313579451668748,
-      "grad_norm": 0.000950752233620733,
-      "learning_rate": 0.00019999816239271164,
-      "loss": 46.0,
-      "step": 11993
-    },
-    {
-      "epoch": 1.9315189822456622,
-      "grad_norm": 0.0015661190263926983,
-      "learning_rate": 0.0001999981620859974,
-      "loss": 46.0,
-      "step": 11994
-    },
-    {
-      "epoch": 1.9316800193244494,
-      "grad_norm": 0.010314003564417362,
-      "learning_rate": 0.00019999816177925758,
-      "loss": 46.0,
-      "step": 11995
-    },
-    {
-      "epoch": 1.931841056403237,
-      "grad_norm": 0.0005025442806072533,
-      "learning_rate": 0.00019999816147249215,
-      "loss": 46.0,
-      "step": 11996
-    },
-    {
-      "epoch": 1.9320020934820241,
-      "grad_norm": 0.0019186304416507483,
-      "learning_rate": 0.00019999816116570112,
-      "loss": 46.0,
-      "step": 11997
-    },
-    {
-      "epoch": 1.9321631305608116,
-      "grad_norm": 0.006390773691236973,
-      "learning_rate": 0.0001999981608588845,
-      "loss": 46.0,
-      "step": 11998
-    },
-    {
-      "epoch": 1.932324167639599,
-      "grad_norm": 0.0012256887275725603,
-      "learning_rate": 0.00019999816055204231,
-      "loss": 46.0,
-      "step": 11999
-    },
-    {
-      "epoch": 1.9324852047183865,
-      "grad_norm": 0.0008525012526661158,
-      "learning_rate": 0.0001999981602451745,
-      "loss": 46.0,
-      "step": 12000
-    },
-    {
-      "epoch": 1.932646241797174,
-      "grad_norm": 0.0026415903121232986,
-      "learning_rate": 0.00019999815993828113,
-      "loss": 46.0,
-      "step": 12001
-    },
-    {
-      "epoch": 1.9328072788759612,
-      "grad_norm": 0.0022421609610319138,
-      "learning_rate": 0.00019999815963136212,
-      "loss": 46.0,
-      "step": 12002
-    },
-    {
-      "epoch": 1.9329683159547486,
-      "grad_norm": 0.0009753041085787117,
-      "learning_rate": 0.00019999815932441754,
-      "loss": 46.0,
-      "step": 12003
-    },
-    {
-      "epoch": 1.9331293530335358,
-      "grad_norm": 0.007640308700501919,
-      "learning_rate": 0.00019999815901744738,
-      "loss": 46.0,
-      "step": 12004
-    },
-    {
-      "epoch": 1.9332903901123233,
-      "grad_norm": 0.0008999984711408615,
-      "learning_rate": 0.0001999981587104516,
-      "loss": 46.0,
-      "step": 12005
-    },
-    {
-      "epoch": 1.9334514271911107,
-      "grad_norm": 0.005782170686870813,
-      "learning_rate": 0.00019999815840343024,
-      "loss": 46.0,
-      "step": 12006
-    },
-    {
-      "epoch": 1.9336124642698982,
-      "grad_norm": 0.0008568445919081569,
-      "learning_rate": 0.00019999815809638327,
-      "loss": 46.0,
-      "step": 12007
-    },
-    {
-      "epoch": 1.9337735013486856,
-      "grad_norm": 0.002353344112634659,
-      "learning_rate": 0.00019999815778931073,
-      "loss": 46.0,
-      "step": 12008
-    },
-    {
-      "epoch": 1.9339345384274729,
-      "grad_norm": 0.00510431220754981,
-      "learning_rate": 0.00019999815748221255,
-      "loss": 46.0,
-      "step": 12009
-    },
-    {
-      "epoch": 1.9340955755062603,
-      "grad_norm": 0.006691760383546352,
-      "learning_rate": 0.0001999981571750888,
-      "loss": 46.0,
-      "step": 12010
-    },
-    {
-      "epoch": 1.9342566125850476,
-      "grad_norm": 0.0024249027483165264,
-      "learning_rate": 0.0001999981568679395,
-      "loss": 46.0,
-      "step": 12011
-    },
-    {
-      "epoch": 1.934417649663835,
-      "grad_norm": 0.0032153818756341934,
-      "learning_rate": 0.00019999815656076455,
-      "loss": 46.0,
-      "step": 12012
-    },
-    {
-      "epoch": 1.9345786867426225,
-      "grad_norm": 0.007203936576843262,
-      "learning_rate": 0.00019999815625356402,
-      "loss": 46.0,
-      "step": 12013
-    },
-    {
-      "epoch": 1.93473972382141,
-      "grad_norm": 0.0010269597405567765,
-      "learning_rate": 0.00019999815594633788,
-      "loss": 46.0,
-      "step": 12014
-    },
-    {
-      "epoch": 1.9349007609001974,
-      "grad_norm": 0.001805074862204492,
-      "learning_rate": 0.0001999981556390862,
-      "loss": 46.0,
-      "step": 12015
-    },
-    {
-      "epoch": 1.9350617979789848,
-      "grad_norm": 0.004218493588268757,
-      "learning_rate": 0.0001999981553318089,
-      "loss": 46.0,
-      "step": 12016
-    },
-    {
-      "epoch": 1.935222835057772,
-      "grad_norm": 0.0006299649248830974,
-      "learning_rate": 0.00019999815502450596,
-      "loss": 46.0,
-      "step": 12017
-    },
-    {
-      "epoch": 1.9353838721365595,
-      "grad_norm": 0.009320312179625034,
-      "learning_rate": 0.00019999815471717747,
-      "loss": 46.0,
-      "step": 12018
-    },
-    {
-      "epoch": 1.9355449092153467,
-      "grad_norm": 0.009105801582336426,
-      "learning_rate": 0.00019999815440982337,
-      "loss": 46.0,
-      "step": 12019
-    },
-    {
-      "epoch": 1.9357059462941342,
-      "grad_norm": 0.0045250155963003635,
-      "learning_rate": 0.0001999981541024437,
-      "loss": 46.0,
-      "step": 12020
-    },
-    {
-      "epoch": 1.9358669833729216,
-      "grad_norm": 0.0034765484742820263,
-      "learning_rate": 0.0001999981537950384,
-      "loss": 46.0,
-      "step": 12021
-    },
-    {
-      "epoch": 1.936028020451709,
-      "grad_norm": 0.0009283124818466604,
-      "learning_rate": 0.0001999981534876075,
-      "loss": 46.0,
-      "step": 12022
-    },
-    {
-      "epoch": 1.9361890575304965,
-      "grad_norm": 0.0021041701547801495,
-      "learning_rate": 0.00019999815318015105,
-      "loss": 46.0,
-      "step": 12023
-    },
-    {
-      "epoch": 1.9363500946092838,
-      "grad_norm": 0.001083271112293005,
-      "learning_rate": 0.00019999815287266898,
-      "loss": 46.0,
-      "step": 12024
-    },
-    {
-      "epoch": 1.9365111316880712,
-      "grad_norm": 0.0017777506727725267,
-      "learning_rate": 0.00019999815256516133,
-      "loss": 46.0,
-      "step": 12025
-    },
-    {
-      "epoch": 1.9366721687668584,
-      "grad_norm": 0.002739750314503908,
-      "learning_rate": 0.00019999815225762806,
-      "loss": 46.0,
-      "step": 12026
-    },
-    {
-      "epoch": 1.9368332058456459,
-      "grad_norm": 0.0025823721662163734,
-      "learning_rate": 0.0001999981519500692,
-      "loss": 46.0,
-      "step": 12027
-    },
-    {
-      "epoch": 1.9369942429244333,
-      "grad_norm": 0.0014158979756757617,
-      "learning_rate": 0.0001999981516424848,
-      "loss": 46.0,
-      "step": 12028
-    },
-    {
-      "epoch": 1.9371552800032208,
-      "grad_norm": 0.004590214695781469,
-      "learning_rate": 0.00019999815133487473,
-      "loss": 46.0,
-      "step": 12029
-    },
-    {
-      "epoch": 1.9373163170820082,
-      "grad_norm": 0.00962040200829506,
-      "learning_rate": 0.00019999815102723911,
-      "loss": 46.0,
-      "step": 12030
-    },
-    {
-      "epoch": 1.9374773541607955,
-      "grad_norm": 0.0015417892718687654,
-      "learning_rate": 0.00019999815071957788,
-      "loss": 46.0,
-      "step": 12031
-    },
-    {
-      "epoch": 1.937638391239583,
-      "grad_norm": 0.0013389629311859608,
-      "learning_rate": 0.00019999815041189104,
-      "loss": 46.0,
-      "step": 12032
-    },
-    {
-      "epoch": 1.9377994283183702,
-      "grad_norm": 0.0007798003498464823,
-      "learning_rate": 0.00019999815010417863,
-      "loss": 46.0,
-      "step": 12033
-    },
-    {
-      "epoch": 1.9379604653971576,
-      "grad_norm": 0.00914925429970026,
-      "learning_rate": 0.00019999814979644064,
-      "loss": 46.0,
-      "step": 12034
-    },
-    {
-      "epoch": 1.938121502475945,
-      "grad_norm": 0.001808097935281694,
-      "learning_rate": 0.00019999814948867703,
-      "loss": 46.0,
-      "step": 12035
-    },
-    {
-      "epoch": 1.9382825395547325,
-      "grad_norm": 0.0011173714883625507,
-      "learning_rate": 0.0001999981491808878,
-      "loss": 46.0,
-      "step": 12036
-    },
-    {
-      "epoch": 1.93844357663352,
-      "grad_norm": 0.001478369114920497,
-      "learning_rate": 0.00019999814887307303,
-      "loss": 46.0,
-      "step": 12037
-    },
-    {
-      "epoch": 1.9386046137123074,
-      "grad_norm": 0.005412651225924492,
-      "learning_rate": 0.00019999814856523266,
-      "loss": 46.0,
-      "step": 12038
-    },
-    {
-      "epoch": 1.9387656507910946,
-      "grad_norm": 0.005451761186122894,
-      "learning_rate": 0.00019999814825736665,
-      "loss": 46.0,
-      "step": 12039
-    },
-    {
-      "epoch": 1.938926687869882,
-      "grad_norm": 0.0007236241362988949,
-      "learning_rate": 0.00019999814794947508,
-      "loss": 46.0,
-      "step": 12040
-    },
-    {
-      "epoch": 1.9390877249486693,
-      "grad_norm": 0.00831711757928133,
-      "learning_rate": 0.00019999814764155792,
-      "loss": 46.0,
-      "step": 12041
-    },
-    {
-      "epoch": 1.9392487620274568,
-      "grad_norm": 0.01236477680504322,
-      "learning_rate": 0.00019999814733361515,
-      "loss": 46.0,
-      "step": 12042
-    },
-    {
-      "epoch": 1.9394097991062442,
-      "grad_norm": 0.005869931075721979,
-      "learning_rate": 0.0001999981470256468,
-      "loss": 46.0,
-      "step": 12043
-    },
-    {
-      "epoch": 1.9395708361850317,
-      "grad_norm": 0.001580044045113027,
-      "learning_rate": 0.00019999814671765282,
-      "loss": 46.0,
-      "step": 12044
-    },
-    {
-      "epoch": 1.9397318732638191,
-      "grad_norm": 0.003241234226152301,
-      "learning_rate": 0.0001999981464096333,
-      "loss": 46.0,
-      "step": 12045
-    },
-    {
-      "epoch": 1.9398929103426064,
-      "grad_norm": 0.0021619251929223537,
-      "learning_rate": 0.00019999814610158816,
-      "loss": 46.0,
-      "step": 12046
-    },
-    {
-      "epoch": 1.9400539474213938,
-      "grad_norm": 0.0015432409709319472,
-      "learning_rate": 0.0001999981457935174,
-      "loss": 46.0,
-      "step": 12047
-    },
-    {
-      "epoch": 1.940214984500181,
-      "grad_norm": 0.0023786805104464293,
-      "learning_rate": 0.00019999814548542108,
-      "loss": 46.0,
-      "step": 12048
-    },
-    {
-      "epoch": 1.9403760215789685,
-      "grad_norm": 0.0018266643164679408,
-      "learning_rate": 0.00019999814517729917,
-      "loss": 46.0,
-      "step": 12049
-    },
-    {
-      "epoch": 1.940537058657756,
-      "grad_norm": 0.0009219483472406864,
-      "learning_rate": 0.00019999814486915162,
-      "loss": 46.0,
-      "step": 12050
-    },
-    {
-      "epoch": 1.9406980957365434,
-      "grad_norm": 0.011202193796634674,
-      "learning_rate": 0.0001999981445609785,
-      "loss": 46.0,
-      "step": 12051
-    },
-    {
-      "epoch": 1.9408591328153308,
-      "grad_norm": 0.002483846852555871,
-      "learning_rate": 0.0001999981442527798,
-      "loss": 46.0,
-      "step": 12052
-    },
-    {
-      "epoch": 1.941020169894118,
-      "grad_norm": 0.0009753602207638323,
-      "learning_rate": 0.00019999814394455553,
-      "loss": 46.0,
-      "step": 12053
-    },
-    {
-      "epoch": 1.9411812069729055,
-      "grad_norm": 0.0028262832202017307,
-      "learning_rate": 0.0001999981436363056,
-      "loss": 46.0,
-      "step": 12054
-    },
-    {
-      "epoch": 1.9413422440516928,
-      "grad_norm": 0.0019340101862326264,
-      "learning_rate": 0.00019999814332803011,
-      "loss": 46.0,
-      "step": 12055
-    },
-    {
-      "epoch": 1.9415032811304802,
-      "grad_norm": 0.0034789543133229017,
-      "learning_rate": 0.00019999814301972904,
-      "loss": 46.0,
-      "step": 12056
-    },
-    {
-      "epoch": 1.9416643182092677,
-      "grad_norm": 0.004788337275385857,
-      "learning_rate": 0.00019999814271140233,
-      "loss": 46.0,
-      "step": 12057
-    },
-    {
-      "epoch": 1.941825355288055,
-      "grad_norm": 0.002175217494368553,
-      "learning_rate": 0.00019999814240305005,
-      "loss": 46.0,
-      "step": 12058
-    },
-    {
-      "epoch": 1.9419863923668426,
-      "grad_norm": 0.004098773468285799,
-      "learning_rate": 0.00019999814209467222,
-      "loss": 46.0,
-      "step": 12059
-    },
-    {
-      "epoch": 1.94214742944563,
-      "grad_norm": 0.006668771151453257,
-      "learning_rate": 0.0001999981417862687,
-      "loss": 46.0,
-      "step": 12060
-    },
-    {
-      "epoch": 1.9423084665244172,
-      "grad_norm": 0.0009941081516444683,
-      "learning_rate": 0.00019999814147783968,
-      "loss": 46.0,
-      "step": 12061
-    },
-    {
-      "epoch": 1.9424695036032045,
-      "grad_norm": 0.008171027526259422,
-      "learning_rate": 0.00019999814116938502,
-      "loss": 46.0,
-      "step": 12062
-    },
-    {
-      "epoch": 1.942630540681992,
-      "grad_norm": 0.0019051500130444765,
-      "learning_rate": 0.00019999814086090476,
-      "loss": 46.0,
-      "step": 12063
-    },
-    {
-      "epoch": 1.9427915777607794,
-      "grad_norm": 0.0026300756726413965,
-      "learning_rate": 0.0001999981405523989,
-      "loss": 46.0,
-      "step": 12064
-    },
-    {
-      "epoch": 1.9429526148395668,
-      "grad_norm": 0.001015787827782333,
-      "learning_rate": 0.0001999981402438675,
-      "loss": 46.0,
-      "step": 12065
-    },
-    {
-      "epoch": 1.9431136519183543,
-      "grad_norm": 0.0007785811321809888,
-      "learning_rate": 0.00019999813993531044,
-      "loss": 46.0,
-      "step": 12066
-    },
-    {
-      "epoch": 1.9432746889971417,
-      "grad_norm": 0.001544971950352192,
-      "learning_rate": 0.00019999813962672783,
-      "loss": 46.0,
-      "step": 12067
-    },
-    {
-      "epoch": 1.943435726075929,
-      "grad_norm": 0.006294172257184982,
-      "learning_rate": 0.00019999813931811963,
-      "loss": 46.0,
-      "step": 12068
-    },
-    {
-      "epoch": 1.9435967631547164,
-      "grad_norm": 0.00624111806973815,
-      "learning_rate": 0.0001999981390094858,
-      "loss": 46.0,
-      "step": 12069
-    },
-    {
-      "epoch": 1.9437578002335036,
-      "grad_norm": 0.007776613347232342,
-      "learning_rate": 0.0001999981387008264,
-      "loss": 46.0,
-      "step": 12070
-    },
-    {
-      "epoch": 1.943918837312291,
-      "grad_norm": 0.004170961678028107,
-      "learning_rate": 0.0001999981383921414,
-      "loss": 46.0,
-      "step": 12071
-    },
-    {
-      "epoch": 1.9440798743910785,
-      "grad_norm": 0.0032628020271658897,
-      "learning_rate": 0.0001999981380834308,
-      "loss": 46.0,
-      "step": 12072
-    },
-    {
-      "epoch": 1.944240911469866,
-      "grad_norm": 0.004319639410823584,
-      "learning_rate": 0.0001999981377746946,
-      "loss": 46.0,
-      "step": 12073
-    },
-    {
-      "epoch": 1.9444019485486534,
-      "grad_norm": 0.00107934873085469,
-      "learning_rate": 0.00019999813746593282,
-      "loss": 46.0,
-      "step": 12074
-    },
-    {
-      "epoch": 1.9445629856274407,
-      "grad_norm": 0.0034845848567783833,
-      "learning_rate": 0.00019999813715714543,
-      "loss": 46.0,
-      "step": 12075
-    },
-    {
-      "epoch": 1.9447240227062281,
-      "grad_norm": 0.008552225306630135,
-      "learning_rate": 0.00019999813684833248,
-      "loss": 46.0,
-      "step": 12076
-    },
-    {
-      "epoch": 1.9448850597850154,
-      "grad_norm": 0.0025991920847445726,
-      "learning_rate": 0.00019999813653949388,
-      "loss": 46.0,
-      "step": 12077
-    },
-    {
-      "epoch": 1.9450460968638028,
-      "grad_norm": 0.001660397625528276,
-      "learning_rate": 0.00019999813623062973,
-      "loss": 46.0,
-      "step": 12078
-    },
-    {
-      "epoch": 1.9452071339425903,
-      "grad_norm": 0.0020927749574184418,
-      "learning_rate": 0.00019999813592173996,
-      "loss": 46.0,
-      "step": 12079
-    },
-    {
-      "epoch": 1.9453681710213777,
-      "grad_norm": 0.001581791671924293,
-      "learning_rate": 0.0001999981356128246,
-      "loss": 46.0,
-      "step": 12080
-    },
-    {
-      "epoch": 1.9455292081001652,
-      "grad_norm": 0.0007431380799971521,
-      "learning_rate": 0.00019999813530388366,
-      "loss": 46.0,
-      "step": 12081
-    },
-    {
-      "epoch": 1.9456902451789526,
-      "grad_norm": 0.0017823704984039068,
-      "learning_rate": 0.00019999813499491713,
-      "loss": 46.0,
-      "step": 12082
-    },
-    {
-      "epoch": 1.9458512822577398,
-      "grad_norm": 0.0059769004583358765,
-      "learning_rate": 0.000199998134685925,
-      "loss": 46.0,
-      "step": 12083
-    },
-    {
-      "epoch": 1.946012319336527,
-      "grad_norm": 0.0010595706989988685,
-      "learning_rate": 0.00019999813437690726,
-      "loss": 46.0,
-      "step": 12084
-    },
-    {
-      "epoch": 1.9461733564153145,
-      "grad_norm": 0.0004557165375445038,
-      "learning_rate": 0.00019999813406786394,
-      "loss": 46.0,
-      "step": 12085
-    },
-    {
-      "epoch": 1.946334393494102,
-      "grad_norm": 0.002186733530834317,
-      "learning_rate": 0.000199998133758795,
-      "loss": 46.0,
-      "step": 12086
-    },
-    {
-      "epoch": 1.9464954305728894,
-      "grad_norm": 0.0007412717095576227,
-      "learning_rate": 0.00019999813344970051,
-      "loss": 46.0,
-      "step": 12087
-    },
-    {
-      "epoch": 1.9466564676516769,
-      "grad_norm": 0.005248753819614649,
-      "learning_rate": 0.0001999981331405804,
-      "loss": 46.0,
-      "step": 12088
-    },
-    {
-      "epoch": 1.9468175047304643,
-      "grad_norm": 0.002423160243779421,
-      "learning_rate": 0.00019999813283143469,
-      "loss": 46.0,
-      "step": 12089
-    },
-    {
-      "epoch": 1.9469785418092516,
-      "grad_norm": 0.0014704273780807853,
-      "learning_rate": 0.0001999981325222634,
-      "loss": 46.0,
-      "step": 12090
-    },
-    {
-      "epoch": 1.947139578888039,
-      "grad_norm": 0.0005859193624928594,
-      "learning_rate": 0.0001999981322130665,
-      "loss": 46.0,
-      "step": 12091
-    },
-    {
-      "epoch": 1.9473006159668262,
-      "grad_norm": 0.00194712751545012,
-      "learning_rate": 0.00019999813190384403,
-      "loss": 46.0,
-      "step": 12092
-    },
-    {
-      "epoch": 1.9474616530456137,
-      "grad_norm": 0.0008329588454216719,
-      "learning_rate": 0.00019999813159459596,
-      "loss": 46.0,
-      "step": 12093
-    },
-    {
-      "epoch": 1.9476226901244011,
-      "grad_norm": 0.001748097944073379,
-      "learning_rate": 0.00019999813128532227,
-      "loss": 46.0,
-      "step": 12094
-    },
-    {
-      "epoch": 1.9477837272031886,
-      "grad_norm": 0.001053764601238072,
-      "learning_rate": 0.000199998130976023,
-      "loss": 46.0,
-      "step": 12095
-    },
-    {
-      "epoch": 1.947944764281976,
-      "grad_norm": 0.0012390092015266418,
-      "learning_rate": 0.00019999813066669814,
-      "loss": 46.0,
-      "step": 12096
-    },
-    {
-      "epoch": 1.9481058013607633,
-      "grad_norm": 0.008858664892613888,
-      "learning_rate": 0.0001999981303573477,
-      "loss": 46.0,
-      "step": 12097
-    },
-    {
-      "epoch": 1.9482668384395507,
-      "grad_norm": 0.0015460547292605042,
-      "learning_rate": 0.00019999813004797164,
-      "loss": 46.0,
-      "step": 12098
-    },
-    {
-      "epoch": 1.948427875518338,
-      "grad_norm": 0.001218714052811265,
-      "learning_rate": 0.00019999812973857,
-      "loss": 46.0,
-      "step": 12099
-    },
-    {
-      "epoch": 1.9485889125971254,
-      "grad_norm": 0.011520106345415115,
-      "learning_rate": 0.00019999812942914276,
-      "loss": 46.0,
-      "step": 12100
-    },
-    {
-      "epoch": 1.9487499496759129,
-      "grad_norm": 0.005516184028238058,
-      "learning_rate": 0.0001999981291196899,
-      "loss": 46.0,
-      "step": 12101
-    },
-    {
-      "epoch": 1.9489109867547003,
-      "grad_norm": 0.0005748493713326752,
-      "learning_rate": 0.0001999981288102115,
-      "loss": 46.0,
-      "step": 12102
-    },
-    {
-      "epoch": 1.9490720238334878,
-      "grad_norm": 0.0038536195643246174,
-      "learning_rate": 0.00019999812850070745,
-      "loss": 46.0,
-      "step": 12103
-    },
-    {
-      "epoch": 1.9492330609122752,
-      "grad_norm": 0.008117640390992165,
-      "learning_rate": 0.00019999812819117784,
-      "loss": 46.0,
-      "step": 12104
-    },
-    {
-      "epoch": 1.9493940979910624,
-      "grad_norm": 0.0025573784951120615,
-      "learning_rate": 0.00019999812788162261,
-      "loss": 46.0,
-      "step": 12105
-    },
-    {
-      "epoch": 1.9495551350698497,
-      "grad_norm": 0.003354765707626939,
-      "learning_rate": 0.0001999981275720418,
-      "loss": 46.0,
-      "step": 12106
-    },
-    {
-      "epoch": 1.9497161721486371,
-      "grad_norm": 0.012561849318444729,
-      "learning_rate": 0.00019999812726243543,
-      "loss": 46.0,
-      "step": 12107
-    },
-    {
-      "epoch": 1.9498772092274246,
-      "grad_norm": 0.0015255310572683811,
-      "learning_rate": 0.00019999812695280342,
-      "loss": 46.0,
-      "step": 12108
-    },
-    {
-      "epoch": 1.950038246306212,
-      "grad_norm": 0.0014020310482010245,
-      "learning_rate": 0.00019999812664314585,
-      "loss": 46.0,
-      "step": 12109
-    },
-    {
-      "epoch": 1.9501992833849995,
-      "grad_norm": 0.0013271570205688477,
-      "learning_rate": 0.00019999812633346266,
-      "loss": 46.0,
-      "step": 12110
-    },
-    {
-      "epoch": 1.950360320463787,
-      "grad_norm": 0.0008281288901343942,
-      "learning_rate": 0.00019999812602375388,
-      "loss": 46.0,
-      "step": 12111
-    },
-    {
-      "epoch": 1.9505213575425742,
-      "grad_norm": 0.0006146468804217875,
-      "learning_rate": 0.0001999981257140195,
-      "loss": 46.0,
-      "step": 12112
-    },
-    {
-      "epoch": 1.9506823946213616,
-      "grad_norm": 0.001222972059622407,
-      "learning_rate": 0.00019999812540425955,
-      "loss": 46.0,
-      "step": 12113
-    },
-    {
-      "epoch": 1.9508434317001488,
-      "grad_norm": 0.0031757354736328125,
-      "learning_rate": 0.00019999812509447398,
-      "loss": 46.0,
-      "step": 12114
-    },
-    {
-      "epoch": 1.9510044687789363,
-      "grad_norm": 0.009279381483793259,
-      "learning_rate": 0.00019999812478466283,
-      "loss": 46.0,
-      "step": 12115
-    },
-    {
-      "epoch": 1.9511655058577237,
-      "grad_norm": 0.0017381719080731273,
-      "learning_rate": 0.0001999981244748261,
-      "loss": 46.0,
-      "step": 12116
-    },
-    {
-      "epoch": 1.9513265429365112,
-      "grad_norm": 0.002244340954348445,
-      "learning_rate": 0.00019999812416496372,
-      "loss": 46.0,
-      "step": 12117
-    },
-    {
-      "epoch": 1.9514875800152987,
-      "grad_norm": 0.003751095151528716,
-      "learning_rate": 0.0001999981238550758,
-      "loss": 46.0,
-      "step": 12118
-    },
-    {
-      "epoch": 1.9516486170940859,
-      "grad_norm": 0.0019959721248596907,
-      "learning_rate": 0.00019999812354516225,
-      "loss": 46.0,
-      "step": 12119
-    },
-    {
-      "epoch": 1.9518096541728733,
-      "grad_norm": 0.003004837315529585,
-      "learning_rate": 0.00019999812323522314,
-      "loss": 46.0,
-      "step": 12120
-    },
-    {
-      "epoch": 1.9519706912516606,
-      "grad_norm": 0.0034260652028024197,
-      "learning_rate": 0.0001999981229252584,
-      "loss": 46.0,
-      "step": 12121
-    },
-    {
-      "epoch": 1.952131728330448,
-      "grad_norm": 0.00870165042579174,
-      "learning_rate": 0.0001999981226152681,
-      "loss": 46.0,
-      "step": 12122
-    },
-    {
-      "epoch": 1.9522927654092355,
-      "grad_norm": 0.002273171441629529,
-      "learning_rate": 0.0001999981223052522,
-      "loss": 46.0,
-      "step": 12123
-    },
-    {
-      "epoch": 1.952453802488023,
-      "grad_norm": 0.002056877827271819,
-      "learning_rate": 0.00019999812199521068,
-      "loss": 46.0,
-      "step": 12124
-    },
-    {
-      "epoch": 1.9526148395668104,
-      "grad_norm": 0.0037193584721535444,
-      "learning_rate": 0.00019999812168514355,
-      "loss": 46.0,
-      "step": 12125
-    },
-    {
-      "epoch": 1.9527758766455976,
-      "grad_norm": 0.0007669219048693776,
-      "learning_rate": 0.00019999812137505088,
-      "loss": 46.0,
-      "step": 12126
-    },
-    {
-      "epoch": 1.952936913724385,
-      "grad_norm": 0.0026137731038033962,
-      "learning_rate": 0.00019999812106493258,
-      "loss": 46.0,
-      "step": 12127
-    },
-    {
-      "epoch": 1.9530979508031723,
-      "grad_norm": 0.0008542754221707582,
-      "learning_rate": 0.0001999981207547887,
-      "loss": 46.0,
-      "step": 12128
-    },
-    {
-      "epoch": 1.9532589878819597,
-      "grad_norm": 0.0013925344683229923,
-      "learning_rate": 0.00019999812044461924,
-      "loss": 46.0,
-      "step": 12129
-    },
-    {
-      "epoch": 1.9534200249607472,
-      "grad_norm": 0.0016747916815802455,
-      "learning_rate": 0.00019999812013442417,
-      "loss": 46.0,
-      "step": 12130
-    },
-    {
-      "epoch": 1.9535810620395346,
-      "grad_norm": 0.0005838229553773999,
-      "learning_rate": 0.0001999981198242035,
-      "loss": 46.0,
-      "step": 12131
-    },
-    {
-      "epoch": 1.953742099118322,
-      "grad_norm": 0.0108532989397645,
-      "learning_rate": 0.00019999811951395722,
-      "loss": 46.0,
-      "step": 12132
-    },
-    {
-      "epoch": 1.9539031361971095,
-      "grad_norm": 0.006761074066162109,
-      "learning_rate": 0.00019999811920368537,
-      "loss": 46.0,
-      "step": 12133
-    },
-    {
-      "epoch": 1.9540641732758968,
-      "grad_norm": 0.0003684191033244133,
-      "learning_rate": 0.00019999811889338792,
-      "loss": 46.0,
-      "step": 12134
-    },
-    {
-      "epoch": 1.9542252103546842,
-      "grad_norm": 0.007435506675392389,
-      "learning_rate": 0.00019999811858306487,
-      "loss": 46.0,
-      "step": 12135
-    },
-    {
-      "epoch": 1.9543862474334714,
-      "grad_norm": 0.0034958310425281525,
-      "learning_rate": 0.00019999811827271625,
-      "loss": 46.0,
-      "step": 12136
-    },
-    {
-      "epoch": 1.954547284512259,
-      "grad_norm": 0.0012527265353128314,
-      "learning_rate": 0.00019999811796234202,
-      "loss": 46.0,
-      "step": 12137
-    },
-    {
-      "epoch": 1.9547083215910463,
-      "grad_norm": 0.0012181499041616917,
-      "learning_rate": 0.00019999811765194218,
-      "loss": 46.0,
-      "step": 12138
-    },
-    {
-      "epoch": 1.9548693586698338,
-      "grad_norm": 0.0006066956557333469,
-      "learning_rate": 0.00019999811734151677,
-      "loss": 46.0,
-      "step": 12139
-    },
-    {
-      "epoch": 1.9550303957486213,
-      "grad_norm": 0.0031212479807436466,
-      "learning_rate": 0.00019999811703106575,
-      "loss": 46.0,
-      "step": 12140
-    },
-    {
-      "epoch": 1.9551914328274085,
-      "grad_norm": 0.000998818315565586,
-      "learning_rate": 0.00019999811672058915,
-      "loss": 46.0,
-      "step": 12141
-    },
-    {
-      "epoch": 1.955352469906196,
-      "grad_norm": 0.0007590508321300149,
-      "learning_rate": 0.00019999811641008695,
-      "loss": 46.0,
-      "step": 12142
-    },
-    {
-      "epoch": 1.9555135069849832,
-      "grad_norm": 0.00371168227866292,
-      "learning_rate": 0.00019999811609955912,
-      "loss": 46.0,
-      "step": 12143
-    },
-    {
-      "epoch": 1.9556745440637706,
-      "grad_norm": 0.0026444531977176666,
-      "learning_rate": 0.00019999811578900575,
-      "loss": 46.0,
-      "step": 12144
-    },
-    {
-      "epoch": 1.955835581142558,
-      "grad_norm": 0.0006410221103578806,
-      "learning_rate": 0.00019999811547842674,
-      "loss": 46.0,
-      "step": 12145
-    },
-    {
-      "epoch": 1.9559966182213455,
-      "grad_norm": 0.011332300491631031,
-      "learning_rate": 0.00019999811516782217,
-      "loss": 46.0,
-      "step": 12146
-    },
-    {
-      "epoch": 1.956157655300133,
-      "grad_norm": 0.0016425964422523975,
-      "learning_rate": 0.00019999811485719199,
-      "loss": 46.0,
-      "step": 12147
-    },
-    {
-      "epoch": 1.9563186923789202,
-      "grad_norm": 0.0014186163898557425,
-      "learning_rate": 0.00019999811454653624,
-      "loss": 46.0,
-      "step": 12148
-    },
-    {
-      "epoch": 1.9564797294577077,
-      "grad_norm": 0.0037449817173182964,
-      "learning_rate": 0.00019999811423585486,
-      "loss": 46.0,
-      "step": 12149
-    },
-    {
-      "epoch": 1.9566407665364949,
-      "grad_norm": 0.0006519765593111515,
-      "learning_rate": 0.00019999811392514789,
-      "loss": 46.0,
-      "step": 12150
-    },
-    {
-      "epoch": 1.9568018036152823,
-      "grad_norm": 0.0019342554733157158,
-      "learning_rate": 0.00019999811361441533,
-      "loss": 46.0,
-      "step": 12151
-    },
-    {
-      "epoch": 1.9569628406940698,
-      "grad_norm": 0.006986892782151699,
-      "learning_rate": 0.0001999981133036572,
-      "loss": 46.0,
-      "step": 12152
-    },
-    {
-      "epoch": 1.9571238777728572,
-      "grad_norm": 0.00025443563936278224,
-      "learning_rate": 0.00019999811299287345,
-      "loss": 46.0,
-      "step": 12153
-    },
-    {
-      "epoch": 1.9572849148516447,
-      "grad_norm": 0.007195987738668919,
-      "learning_rate": 0.00019999811268206412,
-      "loss": 46.0,
-      "step": 12154
-    },
-    {
-      "epoch": 1.9574459519304321,
-      "grad_norm": 0.003083311254158616,
-      "learning_rate": 0.0001999981123712292,
-      "loss": 46.0,
-      "step": 12155
-    },
-    {
-      "epoch": 1.9576069890092194,
-      "grad_norm": 0.009662473574280739,
-      "learning_rate": 0.00019999811206036867,
-      "loss": 46.0,
-      "step": 12156
-    },
-    {
-      "epoch": 1.9577680260880068,
-      "grad_norm": 0.002636327873915434,
-      "learning_rate": 0.00019999811174948256,
-      "loss": 46.0,
-      "step": 12157
-    },
-    {
-      "epoch": 1.957929063166794,
-      "grad_norm": 0.0024721070658415556,
-      "learning_rate": 0.0001999981114385708,
-      "loss": 46.0,
-      "step": 12158
-    },
-    {
-      "epoch": 1.9580901002455815,
-      "grad_norm": 0.0004560021625366062,
-      "learning_rate": 0.00019999811112763352,
-      "loss": 46.0,
-      "step": 12159
-    },
-    {
-      "epoch": 1.958251137324369,
-      "grad_norm": 0.0020225245971232653,
-      "learning_rate": 0.00019999811081667062,
-      "loss": 46.0,
-      "step": 12160
-    },
-    {
-      "epoch": 1.9584121744031564,
-      "grad_norm": 0.0010688375914469361,
-      "learning_rate": 0.0001999981105056821,
-      "loss": 46.0,
-      "step": 12161
-    },
-    {
-      "epoch": 1.9585732114819439,
-      "grad_norm": 0.0017607322661206126,
-      "learning_rate": 0.000199998110194668,
-      "loss": 46.0,
-      "step": 12162
-    },
-    {
-      "epoch": 1.958734248560731,
-      "grad_norm": 0.0012470840010792017,
-      "learning_rate": 0.00019999810988362832,
-      "loss": 46.0,
-      "step": 12163
-    },
-    {
-      "epoch": 1.9588952856395185,
-      "grad_norm": 0.0035028427373617887,
-      "learning_rate": 0.00019999810957256305,
-      "loss": 46.0,
-      "step": 12164
-    },
-    {
-      "epoch": 1.9590563227183058,
-      "grad_norm": 0.0007340355659835041,
-      "learning_rate": 0.0001999981092614722,
-      "loss": 46.0,
-      "step": 12165
-    },
-    {
-      "epoch": 1.9592173597970932,
-      "grad_norm": 0.003109299810603261,
-      "learning_rate": 0.00019999810895035571,
-      "loss": 46.0,
-      "step": 12166
-    },
-    {
-      "epoch": 1.9593783968758807,
-      "grad_norm": 0.0033639948815107346,
-      "learning_rate": 0.00019999810863921365,
-      "loss": 46.0,
-      "step": 12167
-    },
-    {
-      "epoch": 1.9595394339546681,
-      "grad_norm": 0.002353239804506302,
-      "learning_rate": 0.000199998108328046,
-      "loss": 46.0,
-      "step": 12168
-    },
-    {
-      "epoch": 1.9597004710334556,
-      "grad_norm": 0.000553739897441119,
-      "learning_rate": 0.00019999810801685274,
-      "loss": 46.0,
-      "step": 12169
-    },
-    {
-      "epoch": 1.9598615081122428,
-      "grad_norm": 0.0019046416273340583,
-      "learning_rate": 0.00019999810770563389,
-      "loss": 46.0,
-      "step": 12170
-    },
-    {
-      "epoch": 1.9600225451910303,
-      "grad_norm": 0.01497845072299242,
-      "learning_rate": 0.00019999810739438945,
-      "loss": 46.0,
-      "step": 12171
-    },
-    {
-      "epoch": 1.9601835822698175,
-      "grad_norm": 0.0008863619877956808,
-      "learning_rate": 0.00019999810708311942,
-      "loss": 46.0,
-      "step": 12172
-    },
-    {
-      "epoch": 1.960344619348605,
-      "grad_norm": 0.0016912948340177536,
-      "learning_rate": 0.00019999810677182378,
-      "loss": 46.0,
-      "step": 12173
-    },
-    {
-      "epoch": 1.9605056564273924,
-      "grad_norm": 0.001350126345641911,
-      "learning_rate": 0.00019999810646050256,
-      "loss": 46.0,
-      "step": 12174
-    },
-    {
-      "epoch": 1.9606666935061798,
-      "grad_norm": 0.0021432330831885338,
-      "learning_rate": 0.00019999810614915574,
-      "loss": 46.0,
-      "step": 12175
-    },
-    {
-      "epoch": 1.9608277305849673,
-      "grad_norm": 0.0010685812449082732,
-      "learning_rate": 0.00019999810583778334,
-      "loss": 46.0,
-      "step": 12176
-    },
-    {
-      "epoch": 1.9609887676637547,
-      "grad_norm": 0.002427290426567197,
-      "learning_rate": 0.00019999810552638532,
-      "loss": 46.0,
-      "step": 12177
-    },
-    {
-      "epoch": 1.961149804742542,
-      "grad_norm": 0.003014141460880637,
-      "learning_rate": 0.00019999810521496172,
-      "loss": 46.0,
-      "step": 12178
-    },
-    {
-      "epoch": 1.9613108418213292,
-      "grad_norm": 0.0024816745426505804,
-      "learning_rate": 0.0001999981049035125,
-      "loss": 46.0,
-      "step": 12179
-    },
-    {
-      "epoch": 1.9614718789001166,
-      "grad_norm": 0.004163929261267185,
-      "learning_rate": 0.00019999810459203773,
-      "loss": 46.0,
-      "step": 12180
-    },
-    {
-      "epoch": 1.961632915978904,
-      "grad_norm": 0.003371578874066472,
-      "learning_rate": 0.00019999810428053734,
-      "loss": 46.0,
-      "step": 12181
-    },
-    {
-      "epoch": 1.9617939530576916,
-      "grad_norm": 0.004334550350904465,
-      "learning_rate": 0.00019999810396901136,
-      "loss": 46.0,
-      "step": 12182
-    },
-    {
-      "epoch": 1.961954990136479,
-      "grad_norm": 0.0031515981536358595,
-      "learning_rate": 0.0001999981036574598,
-      "loss": 46.0,
-      "step": 12183
-    },
-    {
-      "epoch": 1.9621160272152665,
-      "grad_norm": 0.002309974981471896,
-      "learning_rate": 0.0001999981033458826,
-      "loss": 46.0,
-      "step": 12184
-    },
-    {
-      "epoch": 1.9622770642940537,
-      "grad_norm": 0.0008523603319190443,
-      "learning_rate": 0.00019999810303427987,
-      "loss": 46.0,
-      "step": 12185
-    },
-    {
-      "epoch": 1.9624381013728411,
-      "grad_norm": 0.002786514349281788,
-      "learning_rate": 0.0001999981027226515,
-      "loss": 46.0,
-      "step": 12186
-    },
-    {
-      "epoch": 1.9625991384516284,
-      "grad_norm": 0.006536405999213457,
-      "learning_rate": 0.00019999810241099755,
-      "loss": 46.0,
-      "step": 12187
-    },
-    {
-      "epoch": 1.9627601755304158,
-      "grad_norm": 0.0012258117785677314,
-      "learning_rate": 0.000199998102099318,
-      "loss": 46.0,
-      "step": 12188
-    },
-    {
-      "epoch": 1.9629212126092033,
-      "grad_norm": 0.0020775615703314543,
-      "learning_rate": 0.00019999810178761288,
-      "loss": 46.0,
-      "step": 12189
-    },
-    {
-      "epoch": 1.9630822496879907,
-      "grad_norm": 0.0036292194854468107,
-      "learning_rate": 0.00019999810147588215,
-      "loss": 46.0,
-      "step": 12190
-    },
-    {
-      "epoch": 1.9632432867667782,
-      "grad_norm": 0.0007245729211717844,
-      "learning_rate": 0.00019999810116412577,
-      "loss": 46.0,
-      "step": 12191
-    },
-    {
-      "epoch": 1.9634043238455654,
-      "grad_norm": 0.010219019837677479,
-      "learning_rate": 0.00019999810085234387,
-      "loss": 46.0,
-      "step": 12192
-    },
-    {
-      "epoch": 1.9635653609243529,
-      "grad_norm": 0.0030370333697646856,
-      "learning_rate": 0.00019999810054053635,
-      "loss": 46.0,
-      "step": 12193
-    },
-    {
-      "epoch": 1.96372639800314,
-      "grad_norm": 0.00286252424120903,
-      "learning_rate": 0.00019999810022870324,
-      "loss": 46.0,
-      "step": 12194
-    },
-    {
-      "epoch": 1.9638874350819275,
-      "grad_norm": 0.0015990732936188579,
-      "learning_rate": 0.00019999809991684452,
-      "loss": 46.0,
-      "step": 12195
-    },
-    {
-      "epoch": 1.964048472160715,
-      "grad_norm": 0.002753674518316984,
-      "learning_rate": 0.00019999809960496024,
-      "loss": 46.0,
-      "step": 12196
-    },
-    {
-      "epoch": 1.9642095092395024,
-      "grad_norm": 0.0009453110978938639,
-      "learning_rate": 0.00019999809929305032,
-      "loss": 46.0,
-      "step": 12197
-    },
-    {
-      "epoch": 1.96437054631829,
-      "grad_norm": 0.001847909647040069,
-      "learning_rate": 0.00019999809898111487,
-      "loss": 46.0,
-      "step": 12198
-    },
-    {
-      "epoch": 1.9645315833970773,
-      "grad_norm": 0.011321339756250381,
-      "learning_rate": 0.00019999809866915377,
-      "loss": 46.0,
-      "step": 12199
-    },
-    {
-      "epoch": 1.9646926204758646,
-      "grad_norm": 0.00615813909098506,
-      "learning_rate": 0.00019999809835716706,
-      "loss": 46.0,
-      "step": 12200
-    },
-    {
-      "epoch": 1.9648536575546518,
-      "grad_norm": 0.0003013201057910919,
-      "learning_rate": 0.00019999809804515482,
-      "loss": 46.0,
-      "step": 12201
-    },
-    {
-      "epoch": 1.9650146946334393,
-      "grad_norm": 0.003417704487219453,
-      "learning_rate": 0.00019999809773311693,
-      "loss": 46.0,
-      "step": 12202
-    },
-    {
-      "epoch": 1.9651757317122267,
-      "grad_norm": 0.0013419464230537415,
-      "learning_rate": 0.0001999980974210535,
-      "loss": 46.0,
-      "step": 12203
-    },
-    {
-      "epoch": 1.9653367687910142,
-      "grad_norm": 0.0008844972471706569,
-      "learning_rate": 0.00019999809710896443,
-      "loss": 46.0,
-      "step": 12204
-    },
-    {
-      "epoch": 1.9654978058698016,
-      "grad_norm": 0.0007280951831489801,
-      "learning_rate": 0.00019999809679684978,
-      "loss": 46.0,
-      "step": 12205
-    },
-    {
-      "epoch": 1.965658842948589,
-      "grad_norm": 0.0015828980831429362,
-      "learning_rate": 0.00019999809648470952,
-      "loss": 46.0,
-      "step": 12206
-    },
-    {
-      "epoch": 1.9658198800273763,
-      "grad_norm": 0.0012401886051520705,
-      "learning_rate": 0.0001999980961725437,
-      "loss": 46.0,
-      "step": 12207
-    },
-    {
-      "epoch": 1.9659809171061637,
-      "grad_norm": 0.0008402077364735305,
-      "learning_rate": 0.00019999809586035224,
-      "loss": 46.0,
-      "step": 12208
-    },
-    {
-      "epoch": 1.966141954184951,
-      "grad_norm": 0.005274999886751175,
-      "learning_rate": 0.00019999809554813522,
-      "loss": 46.0,
-      "step": 12209
-    },
-    {
-      "epoch": 1.9663029912637384,
-      "grad_norm": 0.004728623665869236,
-      "learning_rate": 0.00019999809523589258,
-      "loss": 46.0,
-      "step": 12210
-    },
-    {
-      "epoch": 1.9664640283425259,
-      "grad_norm": 0.001169433817267418,
-      "learning_rate": 0.00019999809492362439,
-      "loss": 46.0,
-      "step": 12211
-    },
-    {
-      "epoch": 1.9666250654213133,
-      "grad_norm": 0.0015337723307311535,
-      "learning_rate": 0.00019999809461133058,
-      "loss": 46.0,
-      "step": 12212
-    },
-    {
-      "epoch": 1.9667861025001008,
-      "grad_norm": 0.00455063208937645,
-      "learning_rate": 0.00019999809429901118,
-      "loss": 46.0,
-      "step": 12213
-    },
-    {
-      "epoch": 1.966947139578888,
-      "grad_norm": 0.002513535087928176,
-      "learning_rate": 0.0001999980939866662,
-      "loss": 46.0,
-      "step": 12214
-    },
-    {
-      "epoch": 1.9671081766576755,
-      "grad_norm": 0.0019140273798257113,
-      "learning_rate": 0.00019999809367429557,
-      "loss": 46.0,
-      "step": 12215
-    },
-    {
-      "epoch": 1.9672692137364627,
-      "grad_norm": 0.0010622062254697084,
-      "learning_rate": 0.00019999809336189938,
-      "loss": 46.0,
-      "step": 12216
-    },
-    {
-      "epoch": 1.9674302508152501,
-      "grad_norm": 0.0009454112150706351,
-      "learning_rate": 0.0001999980930494776,
-      "loss": 46.0,
-      "step": 12217
-    },
-    {
-      "epoch": 1.9675912878940376,
-      "grad_norm": 0.0007914511952549219,
-      "learning_rate": 0.00019999809273703022,
-      "loss": 46.0,
-      "step": 12218
-    },
-    {
-      "epoch": 1.967752324972825,
-      "grad_norm": 0.0033709832932800055,
-      "learning_rate": 0.00019999809242455724,
-      "loss": 46.0,
-      "step": 12219
-    },
-    {
-      "epoch": 1.9679133620516125,
-      "grad_norm": 0.0011022750986739993,
-      "learning_rate": 0.00019999809211205868,
-      "loss": 46.0,
-      "step": 12220
-    },
-    {
-      "epoch": 1.9680743991303997,
-      "grad_norm": 0.0025576017796993256,
-      "learning_rate": 0.00019999809179953453,
-      "loss": 46.0,
-      "step": 12221
-    },
-    {
-      "epoch": 1.9682354362091872,
-      "grad_norm": 0.0008091646595858037,
-      "learning_rate": 0.0001999980914869848,
-      "loss": 46.0,
-      "step": 12222
-    },
-    {
-      "epoch": 1.9683964732879744,
-      "grad_norm": 0.0037050878163427114,
-      "learning_rate": 0.00019999809117440942,
-      "loss": 46.0,
-      "step": 12223
-    },
-    {
-      "epoch": 1.9685575103667619,
-      "grad_norm": 0.0006190096610225737,
-      "learning_rate": 0.00019999809086180848,
-      "loss": 46.0,
-      "step": 12224
-    },
-    {
-      "epoch": 1.9687185474455493,
-      "grad_norm": 0.00044239015551283956,
-      "learning_rate": 0.00019999809054918193,
-      "loss": 46.0,
-      "step": 12225
-    },
-    {
-      "epoch": 1.9688795845243368,
-      "grad_norm": 0.0003960916365031153,
-      "learning_rate": 0.0001999980902365298,
-      "loss": 46.0,
-      "step": 12226
-    },
-    {
-      "epoch": 1.9690406216031242,
-      "grad_norm": 0.0038154316134750843,
-      "learning_rate": 0.0001999980899238521,
-      "loss": 46.0,
-      "step": 12227
-    },
-    {
-      "epoch": 1.9692016586819117,
-      "grad_norm": 0.005771647207438946,
-      "learning_rate": 0.00019999808961114877,
-      "loss": 46.0,
-      "step": 12228
-    },
-    {
-      "epoch": 1.969362695760699,
-      "grad_norm": 0.0010224670404568315,
-      "learning_rate": 0.00019999808929841985,
-      "loss": 46.0,
-      "step": 12229
-    },
-    {
-      "epoch": 1.9695237328394863,
-      "grad_norm": 0.0019001205218955874,
-      "learning_rate": 0.00019999808898566533,
-      "loss": 46.0,
-      "step": 12230
-    },
-    {
-      "epoch": 1.9696847699182736,
-      "grad_norm": 0.000559313572011888,
-      "learning_rate": 0.00019999808867288523,
-      "loss": 46.0,
-      "step": 12231
-    },
-    {
-      "epoch": 1.969845806997061,
-      "grad_norm": 0.002748290542513132,
-      "learning_rate": 0.0001999980883600795,
-      "loss": 46.0,
-      "step": 12232
-    },
-    {
-      "epoch": 1.9700068440758485,
-      "grad_norm": 0.005919093266129494,
-      "learning_rate": 0.00019999808804724824,
-      "loss": 46.0,
-      "step": 12233
-    },
-    {
-      "epoch": 1.970167881154636,
-      "grad_norm": 0.00418158108368516,
-      "learning_rate": 0.00019999808773439134,
-      "loss": 46.0,
-      "step": 12234
-    },
-    {
-      "epoch": 1.9703289182334234,
-      "grad_norm": 0.0008404531399719417,
-      "learning_rate": 0.00019999808742150887,
-      "loss": 46.0,
-      "step": 12235
-    },
-    {
-      "epoch": 1.9704899553122106,
-      "grad_norm": 0.0033515128307044506,
-      "learning_rate": 0.0001999980871086008,
-      "loss": 46.0,
-      "step": 12236
-    },
-    {
-      "epoch": 1.970650992390998,
-      "grad_norm": 0.0004960817750543356,
-      "learning_rate": 0.00019999808679566712,
-      "loss": 46.0,
-      "step": 12237
-    },
-    {
-      "epoch": 1.9708120294697853,
-      "grad_norm": 0.006968372967094183,
-      "learning_rate": 0.00019999808648270785,
-      "loss": 46.0,
-      "step": 12238
-    },
-    {
-      "epoch": 1.9709730665485727,
-      "grad_norm": 0.0009137294837273657,
-      "learning_rate": 0.000199998086169723,
-      "loss": 46.0,
-      "step": 12239
-    },
-    {
-      "epoch": 1.9711341036273602,
-      "grad_norm": 0.0008460060344077647,
-      "learning_rate": 0.00019999808585671253,
-      "loss": 46.0,
-      "step": 12240
-    },
-    {
-      "epoch": 1.9712951407061476,
-      "grad_norm": 0.010460236109793186,
-      "learning_rate": 0.0001999980855436765,
-      "loss": 46.0,
-      "step": 12241
-    },
-    {
-      "epoch": 1.971456177784935,
-      "grad_norm": 0.00291981128975749,
-      "learning_rate": 0.00019999808523061486,
-      "loss": 46.0,
-      "step": 12242
-    },
-    {
-      "epoch": 1.9716172148637223,
-      "grad_norm": 0.0007473889854736626,
-      "learning_rate": 0.00019999808491752763,
-      "loss": 46.0,
-      "step": 12243
-    },
-    {
-      "epoch": 1.9717782519425098,
-      "grad_norm": 0.0021796675864607096,
-      "learning_rate": 0.0001999980846044148,
-      "loss": 46.0,
-      "step": 12244
-    },
-    {
-      "epoch": 1.971939289021297,
-      "grad_norm": 0.005767948925495148,
-      "learning_rate": 0.00019999808429127636,
-      "loss": 46.0,
-      "step": 12245
-    },
-    {
-      "epoch": 1.9721003261000845,
-      "grad_norm": 0.0014900406822562218,
-      "learning_rate": 0.00019999808397811234,
-      "loss": 46.0,
-      "step": 12246
-    },
-    {
-      "epoch": 1.972261363178872,
-      "grad_norm": 0.0019085450330749154,
-      "learning_rate": 0.00019999808366492273,
-      "loss": 46.0,
-      "step": 12247
-    },
-    {
-      "epoch": 1.9724224002576594,
-      "grad_norm": 0.003345359116792679,
-      "learning_rate": 0.0001999980833517075,
-      "loss": 46.0,
-      "step": 12248
-    },
-    {
-      "epoch": 1.9725834373364468,
-      "grad_norm": 0.0007762960158288479,
-      "learning_rate": 0.00019999808303846673,
-      "loss": 46.0,
-      "step": 12249
-    },
-    {
-      "epoch": 1.9727444744152343,
-      "grad_norm": 0.0006640477222390473,
-      "learning_rate": 0.0001999980827252003,
-      "loss": 46.0,
-      "step": 12250
-    },
-    {
-      "epoch": 1.9729055114940215,
-      "grad_norm": 0.0009199056075885892,
-      "learning_rate": 0.0001999980824119083,
-      "loss": 46.0,
-      "step": 12251
-    },
-    {
-      "epoch": 1.973066548572809,
-      "grad_norm": 0.009267180226743221,
-      "learning_rate": 0.00019999808209859073,
-      "loss": 46.0,
-      "step": 12252
-    },
-    {
-      "epoch": 1.9732275856515962,
-      "grad_norm": 0.003825358347967267,
-      "learning_rate": 0.00019999808178524755,
-      "loss": 46.0,
-      "step": 12253
-    },
-    {
-      "epoch": 1.9733886227303836,
-      "grad_norm": 0.004644161555916071,
-      "learning_rate": 0.00019999808147187878,
-      "loss": 46.0,
-      "step": 12254
-    },
-    {
-      "epoch": 1.973549659809171,
-      "grad_norm": 0.004506648983806372,
-      "learning_rate": 0.00019999808115848442,
-      "loss": 46.0,
-      "step": 12255
-    },
-    {
-      "epoch": 1.9737106968879585,
-      "grad_norm": 0.0012973000993952155,
-      "learning_rate": 0.00019999808084506443,
-      "loss": 46.0,
-      "step": 12256
-    },
-    {
-      "epoch": 1.973871733966746,
-      "grad_norm": 0.000974331283941865,
-      "learning_rate": 0.00019999808053161887,
-      "loss": 46.0,
-      "step": 12257
-    },
-    {
-      "epoch": 1.9740327710455332,
-      "grad_norm": 0.0029749576933681965,
-      "learning_rate": 0.00019999808021814772,
-      "loss": 46.0,
-      "step": 12258
-    },
-    {
-      "epoch": 1.9741938081243207,
-      "grad_norm": 0.0019794791005551815,
-      "learning_rate": 0.000199998079904651,
-      "loss": 46.0,
-      "step": 12259
-    },
-    {
-      "epoch": 1.9743548452031079,
-      "grad_norm": 0.007509741000831127,
-      "learning_rate": 0.00019999807959112864,
-      "loss": 46.0,
-      "step": 12260
-    },
-    {
-      "epoch": 1.9745158822818953,
-      "grad_norm": 0.0055227261036634445,
-      "learning_rate": 0.0001999980792775807,
-      "loss": 46.0,
-      "step": 12261
-    },
-    {
-      "epoch": 1.9746769193606828,
-      "grad_norm": 0.0009662671363912523,
-      "learning_rate": 0.00019999807896400719,
-      "loss": 46.0,
-      "step": 12262
-    },
-    {
-      "epoch": 1.9748379564394702,
-      "grad_norm": 0.023432206362485886,
-      "learning_rate": 0.00019999807865040805,
-      "loss": 46.0,
-      "step": 12263
-    },
-    {
-      "epoch": 1.9749989935182577,
-      "grad_norm": 0.0019245572621002793,
-      "learning_rate": 0.00019999807833678333,
-      "loss": 46.0,
-      "step": 12264
-    },
-    {
-      "epoch": 1.975160030597045,
-      "grad_norm": 0.0009568097302690148,
-      "learning_rate": 0.00019999807802313302,
-      "loss": 46.0,
-      "step": 12265
-    },
-    {
-      "epoch": 1.9753210676758324,
-      "grad_norm": 0.0018147414084523916,
-      "learning_rate": 0.00019999807770945712,
-      "loss": 46.0,
-      "step": 12266
-    },
-    {
-      "epoch": 1.9754821047546196,
-      "grad_norm": 0.0030487615149468184,
-      "learning_rate": 0.00019999807739575564,
-      "loss": 46.0,
-      "step": 12267
-    },
-    {
-      "epoch": 1.975643141833407,
-      "grad_norm": 0.0027401866391301155,
-      "learning_rate": 0.0001999980770820285,
-      "loss": 46.0,
-      "step": 12268
-    },
-    {
-      "epoch": 1.9758041789121945,
-      "grad_norm": 0.0011288942769169807,
-      "learning_rate": 0.00019999807676827582,
-      "loss": 46.0,
-      "step": 12269
-    },
-    {
-      "epoch": 1.975965215990982,
-      "grad_norm": 0.0037102545611560345,
-      "learning_rate": 0.00019999807645449752,
-      "loss": 46.0,
-      "step": 12270
-    },
-    {
-      "epoch": 1.9761262530697694,
-      "grad_norm": 0.0009827904868870974,
-      "learning_rate": 0.00019999807614069366,
-      "loss": 46.0,
-      "step": 12271
-    },
-    {
-      "epoch": 1.9762872901485569,
-      "grad_norm": 0.0012922310270369053,
-      "learning_rate": 0.0001999980758268642,
-      "loss": 46.0,
-      "step": 12272
-    },
-    {
-      "epoch": 1.976448327227344,
-      "grad_norm": 0.002203173702582717,
-      "learning_rate": 0.0001999980755130091,
-      "loss": 46.0,
-      "step": 12273
-    },
-    {
-      "epoch": 1.9766093643061313,
-      "grad_norm": 0.001265895669348538,
-      "learning_rate": 0.00019999807519912845,
-      "loss": 46.0,
-      "step": 12274
-    },
-    {
-      "epoch": 1.9767704013849188,
-      "grad_norm": 0.0048488047905266285,
-      "learning_rate": 0.0001999980748852222,
-      "loss": 46.0,
-      "step": 12275
-    },
-    {
-      "epoch": 1.9769314384637062,
-      "grad_norm": 0.005234608892351389,
-      "learning_rate": 0.00019999807457129034,
-      "loss": 46.0,
-      "step": 12276
-    },
-    {
-      "epoch": 1.9770924755424937,
-      "grad_norm": 0.009345931932330132,
-      "learning_rate": 0.0001999980742573329,
-      "loss": 46.0,
-      "step": 12277
-    },
-    {
-      "epoch": 1.9772535126212811,
-      "grad_norm": 0.0040670582093298435,
-      "learning_rate": 0.00019999807394334985,
-      "loss": 46.0,
-      "step": 12278
-    },
-    {
-      "epoch": 1.9774145497000686,
-      "grad_norm": 0.0071364580653607845,
-      "learning_rate": 0.0001999980736293412,
-      "loss": 46.0,
-      "step": 12279
-    },
-    {
-      "epoch": 1.9775755867788558,
-      "grad_norm": 0.0061156777665019035,
-      "learning_rate": 0.00019999807331530698,
-      "loss": 46.0,
-      "step": 12280
-    },
-    {
-      "epoch": 1.9777366238576433,
-      "grad_norm": 0.0024092956446111202,
-      "learning_rate": 0.00019999807300124717,
-      "loss": 46.0,
-      "step": 12281
-    },
-    {
-      "epoch": 1.9778976609364305,
-      "grad_norm": 0.00545324943959713,
-      "learning_rate": 0.00019999807268716174,
-      "loss": 46.0,
-      "step": 12282
-    },
-    {
-      "epoch": 1.978058698015218,
-      "grad_norm": 0.0022237629164010286,
-      "learning_rate": 0.00019999807237305075,
-      "loss": 46.0,
-      "step": 12283
-    },
-    {
-      "epoch": 1.9782197350940054,
-      "grad_norm": 0.005282882135361433,
-      "learning_rate": 0.00019999807205891413,
-      "loss": 46.0,
-      "step": 12284
-    },
-    {
-      "epoch": 1.9783807721727928,
-      "grad_norm": 0.0026031166780740023,
-      "learning_rate": 0.00019999807174475194,
-      "loss": 46.0,
-      "step": 12285
-    },
-    {
-      "epoch": 1.9785418092515803,
-      "grad_norm": 0.002004999201744795,
-      "learning_rate": 0.00019999807143056413,
-      "loss": 46.0,
-      "step": 12286
-    },
-    {
-      "epoch": 1.9787028463303675,
-      "grad_norm": 0.0025790894869714975,
-      "learning_rate": 0.00019999807111635077,
-      "loss": 46.0,
-      "step": 12287
-    },
-    {
-      "epoch": 1.978863883409155,
-      "grad_norm": 0.0008057039813138545,
-      "learning_rate": 0.00019999807080211176,
-      "loss": 46.0,
-      "step": 12288
-    },
-    {
-      "epoch": 1.9790249204879422,
-      "grad_norm": 0.007503815460950136,
-      "learning_rate": 0.00019999807048784717,
-      "loss": 46.0,
-      "step": 12289
-    },
-    {
-      "epoch": 1.9791859575667297,
-      "grad_norm": 0.00282031437382102,
-      "learning_rate": 0.000199998070173557,
-      "loss": 46.0,
-      "step": 12290
-    },
-    {
-      "epoch": 1.979346994645517,
-      "grad_norm": 0.0022822634782642126,
-      "learning_rate": 0.00019999806985924123,
-      "loss": 46.0,
-      "step": 12291
-    },
-    {
-      "epoch": 1.9795080317243046,
-      "grad_norm": 0.008359306491911411,
-      "learning_rate": 0.00019999806954489987,
-      "loss": 46.0,
-      "step": 12292
-    },
-    {
-      "epoch": 1.979669068803092,
-      "grad_norm": 0.003279345342889428,
-      "learning_rate": 0.00019999806923053293,
-      "loss": 46.0,
-      "step": 12293
-    },
-    {
-      "epoch": 1.9798301058818795,
-      "grad_norm": 0.0025783029850572348,
-      "learning_rate": 0.00019999806891614038,
-      "loss": 46.0,
-      "step": 12294
-    },
-    {
-      "epoch": 1.9799911429606667,
-      "grad_norm": 0.0023204863537102938,
-      "learning_rate": 0.00019999806860172224,
-      "loss": 46.0,
-      "step": 12295
-    },
-    {
-      "epoch": 1.980152180039454,
-      "grad_norm": 0.0014210662338882685,
-      "learning_rate": 0.0001999980682872785,
-      "loss": 46.0,
-      "step": 12296
-    },
-    {
-      "epoch": 1.9803132171182414,
-      "grad_norm": 0.0030528847128152847,
-      "learning_rate": 0.00019999806797280916,
-      "loss": 46.0,
-      "step": 12297
-    },
-    {
-      "epoch": 1.9804742541970288,
-      "grad_norm": 0.0021437364630401134,
-      "learning_rate": 0.00019999806765831423,
-      "loss": 46.0,
-      "step": 12298
-    },
-    {
-      "epoch": 1.9806352912758163,
-      "grad_norm": 0.0025230783503502607,
-      "learning_rate": 0.0001999980673437937,
-      "loss": 46.0,
-      "step": 12299
-    },
-    {
-      "epoch": 1.9807963283546037,
-      "grad_norm": 0.0017503659473732114,
-      "learning_rate": 0.00019999806702924758,
-      "loss": 46.0,
-      "step": 12300
-    },
-    {
-      "epoch": 1.9809573654333912,
-      "grad_norm": 0.005607995670288801,
-      "learning_rate": 0.00019999806671467586,
-      "loss": 46.0,
-      "step": 12301
-    },
-    {
-      "epoch": 1.9811184025121784,
-      "grad_norm": 0.0011990922503173351,
-      "learning_rate": 0.00019999806640007858,
-      "loss": 46.0,
-      "step": 12302
-    },
-    {
-      "epoch": 1.9812794395909659,
-      "grad_norm": 0.0012647804105654359,
-      "learning_rate": 0.0001999980660854557,
-      "loss": 46.0,
-      "step": 12303
-    },
-    {
-      "epoch": 1.981440476669753,
-      "grad_norm": 0.002947992878034711,
-      "learning_rate": 0.0001999980657708072,
-      "loss": 46.0,
-      "step": 12304
-    },
-    {
-      "epoch": 1.9816015137485405,
-      "grad_norm": 0.001517716096714139,
-      "learning_rate": 0.00019999806545613309,
-      "loss": 46.0,
-      "step": 12305
-    },
-    {
-      "epoch": 1.981762550827328,
-      "grad_norm": 0.002484687604010105,
-      "learning_rate": 0.0001999980651414334,
-      "loss": 46.0,
-      "step": 12306
-    },
-    {
-      "epoch": 1.9819235879061154,
-      "grad_norm": 0.0037229699082672596,
-      "learning_rate": 0.00019999806482670813,
-      "loss": 46.0,
-      "step": 12307
-    },
-    {
-      "epoch": 1.982084624984903,
-      "grad_norm": 0.0006061809835955501,
-      "learning_rate": 0.00019999806451195725,
-      "loss": 46.0,
-      "step": 12308
-    },
-    {
-      "epoch": 1.9822456620636901,
-      "grad_norm": 0.007558826357126236,
-      "learning_rate": 0.00019999806419718078,
-      "loss": 46.0,
-      "step": 12309
-    },
-    {
-      "epoch": 1.9824066991424776,
-      "grad_norm": 0.004070258233696222,
-      "learning_rate": 0.00019999806388237875,
-      "loss": 46.0,
-      "step": 12310
-    },
-    {
-      "epoch": 1.9825677362212648,
-      "grad_norm": 0.0019721612334251404,
-      "learning_rate": 0.00019999806356755108,
-      "loss": 46.0,
-      "step": 12311
-    },
-    {
-      "epoch": 1.9827287733000523,
-      "grad_norm": 0.006729131564497948,
-      "learning_rate": 0.00019999806325269784,
-      "loss": 46.0,
-      "step": 12312
-    },
-    {
-      "epoch": 1.9828898103788397,
-      "grad_norm": 0.009025485254824162,
-      "learning_rate": 0.000199998062937819,
-      "loss": 46.0,
-      "step": 12313
-    },
-    {
-      "epoch": 1.9830508474576272,
-      "grad_norm": 0.0011525398585945368,
-      "learning_rate": 0.00019999806262291454,
-      "loss": 46.0,
-      "step": 12314
-    },
-    {
-      "epoch": 1.9832118845364146,
-      "grad_norm": 0.0012309578014537692,
-      "learning_rate": 0.00019999806230798451,
-      "loss": 46.0,
-      "step": 12315
-    },
-    {
-      "epoch": 1.9833729216152018,
-      "grad_norm": 0.004987578373402357,
-      "learning_rate": 0.00019999806199302888,
-      "loss": 46.0,
-      "step": 12316
-    },
-    {
-      "epoch": 1.9835339586939893,
-      "grad_norm": 0.0017838859930634499,
-      "learning_rate": 0.00019999806167804766,
-      "loss": 46.0,
-      "step": 12317
-    },
-    {
-      "epoch": 1.9836949957727765,
-      "grad_norm": 0.0008300473564304411,
-      "learning_rate": 0.00019999806136304087,
-      "loss": 46.0,
-      "step": 12318
-    },
-    {
-      "epoch": 1.983856032851564,
-      "grad_norm": 0.0036320278886705637,
-      "learning_rate": 0.00019999806104800845,
-      "loss": 46.0,
-      "step": 12319
-    },
-    {
-      "epoch": 1.9840170699303514,
-      "grad_norm": 0.002548553515225649,
-      "learning_rate": 0.00019999806073295044,
-      "loss": 46.0,
-      "step": 12320
-    },
-    {
-      "epoch": 1.9841781070091389,
-      "grad_norm": 0.002902851440012455,
-      "learning_rate": 0.00019999806041786684,
-      "loss": 46.0,
-      "step": 12321
-    },
-    {
-      "epoch": 1.9843391440879263,
-      "grad_norm": 0.004910650663077831,
-      "learning_rate": 0.00019999806010275766,
-      "loss": 46.0,
-      "step": 12322
-    },
-    {
-      "epoch": 1.9845001811667138,
-      "grad_norm": 0.0009098303853534162,
-      "learning_rate": 0.00019999805978762286,
-      "loss": 46.0,
-      "step": 12323
-    },
-    {
-      "epoch": 1.984661218245501,
-      "grad_norm": 0.002984803169965744,
-      "learning_rate": 0.0001999980594724625,
-      "loss": 46.0,
-      "step": 12324
-    },
-    {
-      "epoch": 1.9848222553242885,
-      "grad_norm": 0.012350688688457012,
-      "learning_rate": 0.0001999980591572765,
-      "loss": 46.0,
-      "step": 12325
-    },
-    {
-      "epoch": 1.9849832924030757,
-      "grad_norm": 0.00577581487596035,
-      "learning_rate": 0.00019999805884206496,
-      "loss": 46.0,
-      "step": 12326
-    },
-    {
-      "epoch": 1.9851443294818631,
-      "grad_norm": 0.0006859779241494834,
-      "learning_rate": 0.00019999805852682778,
-      "loss": 46.0,
-      "step": 12327
-    },
-    {
-      "epoch": 1.9853053665606506,
-      "grad_norm": 0.0006017529522068799,
-      "learning_rate": 0.00019999805821156502,
-      "loss": 46.0,
-      "step": 12328
-    },
-    {
-      "epoch": 1.985466403639438,
-      "grad_norm": 0.00808424036949873,
-      "learning_rate": 0.00019999805789627665,
-      "loss": 46.0,
-      "step": 12329
-    },
-    {
-      "epoch": 1.9856274407182255,
-      "grad_norm": 0.012796023860573769,
-      "learning_rate": 0.0001999980575809627,
-      "loss": 46.0,
-      "step": 12330
-    },
-    {
-      "epoch": 1.9857884777970127,
-      "grad_norm": 0.004385095555335283,
-      "learning_rate": 0.00019999805726562318,
-      "loss": 46.0,
-      "step": 12331
-    },
-    {
-      "epoch": 1.9859495148758002,
-      "grad_norm": 0.0036203661002218723,
-      "learning_rate": 0.00019999805695025802,
-      "loss": 46.0,
-      "step": 12332
-    },
-    {
-      "epoch": 1.9861105519545874,
-      "grad_norm": 0.008089341223239899,
-      "learning_rate": 0.0001999980566348673,
-      "loss": 46.0,
-      "step": 12333
-    },
-    {
-      "epoch": 1.9862715890333749,
-      "grad_norm": 0.00829344056546688,
-      "learning_rate": 0.00019999805631945095,
-      "loss": 46.0,
-      "step": 12334
-    },
-    {
-      "epoch": 1.9864326261121623,
-      "grad_norm": 0.0019083835650235415,
-      "learning_rate": 0.00019999805600400908,
-      "loss": 46.0,
-      "step": 12335
-    },
-    {
-      "epoch": 1.9865936631909498,
-      "grad_norm": 0.005012488458305597,
-      "learning_rate": 0.00019999805568854157,
-      "loss": 46.0,
-      "step": 12336
-    },
-    {
-      "epoch": 1.9867547002697372,
-      "grad_norm": 0.009535104967653751,
-      "learning_rate": 0.00019999805537304844,
-      "loss": 46.0,
-      "step": 12337
-    },
-    {
-      "epoch": 1.9869157373485244,
-      "grad_norm": 0.0005328900297172368,
-      "learning_rate": 0.00019999805505752972,
-      "loss": 46.0,
-      "step": 12338
-    },
-    {
-      "epoch": 1.987076774427312,
-      "grad_norm": 0.00236369576305151,
-      "learning_rate": 0.00019999805474198542,
-      "loss": 46.0,
-      "step": 12339
-    },
-    {
-      "epoch": 1.9872378115060991,
-      "grad_norm": 0.001165008987300098,
-      "learning_rate": 0.00019999805442641556,
-      "loss": 46.0,
-      "step": 12340
-    },
-    {
-      "epoch": 1.9873988485848866,
-      "grad_norm": 0.0036361468955874443,
-      "learning_rate": 0.00019999805411082005,
-      "loss": 46.0,
-      "step": 12341
-    },
-    {
-      "epoch": 1.987559885663674,
-      "grad_norm": 0.0006017803680151701,
-      "learning_rate": 0.00019999805379519899,
-      "loss": 46.0,
-      "step": 12342
-    },
-    {
-      "epoch": 1.9877209227424615,
-      "grad_norm": 0.0038714283145964146,
-      "learning_rate": 0.0001999980534795523,
-      "loss": 46.0,
-      "step": 12343
-    },
-    {
-      "epoch": 1.987881959821249,
-      "grad_norm": 0.009130291640758514,
-      "learning_rate": 0.00019999805316388001,
-      "loss": 46.0,
-      "step": 12344
-    },
-    {
-      "epoch": 1.9880429969000364,
-      "grad_norm": 0.001152730779722333,
-      "learning_rate": 0.0001999980528481822,
-      "loss": 46.0,
-      "step": 12345
-    },
-    {
-      "epoch": 1.9882040339788236,
-      "grad_norm": 0.0018681924557313323,
-      "learning_rate": 0.00019999805253245872,
-      "loss": 46.0,
-      "step": 12346
-    },
-    {
-      "epoch": 1.988365071057611,
-      "grad_norm": 0.005925884936004877,
-      "learning_rate": 0.00019999805221670967,
-      "loss": 46.0,
-      "step": 12347
-    },
-    {
-      "epoch": 1.9885261081363983,
-      "grad_norm": 0.0007004171493463218,
-      "learning_rate": 0.00019999805190093503,
-      "loss": 46.0,
-      "step": 12348
-    },
-    {
-      "epoch": 1.9886871452151857,
-      "grad_norm": 0.0013518970226868987,
-      "learning_rate": 0.00019999805158513477,
-      "loss": 46.0,
-      "step": 12349
-    },
-    {
-      "epoch": 1.9888481822939732,
-      "grad_norm": 0.0009766726288944483,
-      "learning_rate": 0.00019999805126930893,
-      "loss": 46.0,
-      "step": 12350
-    },
-    {
-      "epoch": 1.9890092193727607,
-      "grad_norm": 0.004946235101670027,
-      "learning_rate": 0.0001999980509534575,
-      "loss": 46.0,
-      "step": 12351
-    },
-    {
-      "epoch": 1.989170256451548,
-      "grad_norm": 0.0025586055126041174,
-      "learning_rate": 0.00019999805063758048,
-      "loss": 46.0,
-      "step": 12352
-    },
-    {
-      "epoch": 1.9893312935303353,
-      "grad_norm": 0.0016643865965306759,
-      "learning_rate": 0.00019999805032167787,
-      "loss": 46.0,
-      "step": 12353
-    },
-    {
-      "epoch": 1.9894923306091228,
-      "grad_norm": 0.008810807019472122,
-      "learning_rate": 0.00019999805000574965,
-      "loss": 46.0,
-      "step": 12354
-    },
-    {
-      "epoch": 1.98965336768791,
-      "grad_norm": 0.001997843850404024,
-      "learning_rate": 0.00019999804968979582,
-      "loss": 46.0,
-      "step": 12355
-    },
-    {
-      "epoch": 1.9898144047666975,
-      "grad_norm": 0.003036949783563614,
-      "learning_rate": 0.00019999804937381646,
-      "loss": 46.0,
-      "step": 12356
-    },
-    {
-      "epoch": 1.989975441845485,
-      "grad_norm": 0.0014212050009518862,
-      "learning_rate": 0.00019999804905781145,
-      "loss": 46.0,
-      "step": 12357
-    },
-    {
-      "epoch": 1.9901364789242724,
-      "grad_norm": 0.007264718413352966,
-      "learning_rate": 0.00019999804874178088,
-      "loss": 46.0,
-      "step": 12358
-    },
-    {
-      "epoch": 1.9902975160030598,
-      "grad_norm": 0.007655519060790539,
-      "learning_rate": 0.00019999804842572467,
-      "loss": 46.0,
-      "step": 12359
-    },
-    {
-      "epoch": 1.990458553081847,
-      "grad_norm": 0.00354858860373497,
-      "learning_rate": 0.00019999804810964288,
-      "loss": 46.0,
-      "step": 12360
-    },
-    {
-      "epoch": 1.9906195901606345,
-      "grad_norm": 0.001752532203681767,
-      "learning_rate": 0.00019999804779353552,
-      "loss": 46.0,
-      "step": 12361
-    },
-    {
-      "epoch": 1.9907806272394217,
-      "grad_norm": 0.01585347205400467,
-      "learning_rate": 0.00019999804747740255,
-      "loss": 46.0,
-      "step": 12362
-    },
-    {
-      "epoch": 1.9909416643182092,
-      "grad_norm": 0.0026188462506979704,
-      "learning_rate": 0.000199998047161244,
-      "loss": 46.0,
-      "step": 12363
-    },
-    {
-      "epoch": 1.9911027013969966,
-      "grad_norm": 0.0007879911572672427,
-      "learning_rate": 0.00019999804684505982,
-      "loss": 46.0,
-      "step": 12364
-    },
-    {
-      "epoch": 1.991263738475784,
-      "grad_norm": 0.011935262940824032,
-      "learning_rate": 0.0001999980465288501,
-      "loss": 46.0,
-      "step": 12365
-    },
-    {
-      "epoch": 1.9914247755545715,
-      "grad_norm": 0.00202621566131711,
-      "learning_rate": 0.00019999804621261474,
-      "loss": 46.0,
-      "step": 12366
-    },
-    {
-      "epoch": 1.991585812633359,
-      "grad_norm": 0.0012810584157705307,
-      "learning_rate": 0.0001999980458963538,
-      "loss": 46.0,
-      "step": 12367
-    },
-    {
-      "epoch": 1.9917468497121462,
-      "grad_norm": 0.011172051541507244,
-      "learning_rate": 0.00019999804558006726,
-      "loss": 46.0,
-      "step": 12368
-    },
-    {
-      "epoch": 1.9919078867909334,
-      "grad_norm": 0.0009969781385734677,
-      "learning_rate": 0.00019999804526375515,
-      "loss": 46.0,
-      "step": 12369
-    },
-    {
-      "epoch": 1.992068923869721,
-      "grad_norm": 0.014885064214468002,
-      "learning_rate": 0.00019999804494741743,
-      "loss": 46.0,
-      "step": 12370
-    },
-    {
-      "epoch": 1.9922299609485083,
-      "grad_norm": 0.007742216344922781,
-      "learning_rate": 0.0001999980446310541,
-      "loss": 46.0,
-      "step": 12371
-    },
-    {
-      "epoch": 1.9923909980272958,
-      "grad_norm": 0.004301742650568485,
-      "learning_rate": 0.00019999804431466517,
-      "loss": 46.0,
-      "step": 12372
-    },
-    {
-      "epoch": 1.9925520351060833,
-      "grad_norm": 0.0026088471058756113,
-      "learning_rate": 0.00019999804399825069,
-      "loss": 46.0,
-      "step": 12373
-    },
-    {
-      "epoch": 1.9927130721848707,
-      "grad_norm": 0.0007938767084851861,
-      "learning_rate": 0.00019999804368181056,
-      "loss": 46.0,
-      "step": 12374
-    },
-    {
-      "epoch": 1.992874109263658,
-      "grad_norm": 0.002270581666380167,
-      "learning_rate": 0.00019999804336534488,
-      "loss": 46.0,
-      "step": 12375
-    },
-    {
-      "epoch": 1.9930351463424454,
-      "grad_norm": 0.0047074127942323685,
-      "learning_rate": 0.00019999804304885358,
-      "loss": 46.0,
-      "step": 12376
-    },
-    {
-      "epoch": 1.9931961834212326,
-      "grad_norm": 0.0019312455551698804,
-      "learning_rate": 0.00019999804273233672,
-      "loss": 46.0,
-      "step": 12377
-    },
-    {
-      "epoch": 1.99335722050002,
-      "grad_norm": 0.003867735853418708,
-      "learning_rate": 0.00019999804241579424,
-      "loss": 46.0,
-      "step": 12378
-    },
-    {
-      "epoch": 1.9935182575788075,
-      "grad_norm": 0.0008616733830422163,
-      "learning_rate": 0.00019999804209922616,
-      "loss": 46.0,
-      "step": 12379
-    },
-    {
-      "epoch": 1.993679294657595,
-      "grad_norm": 0.0006523603806272149,
-      "learning_rate": 0.0001999980417826325,
-      "loss": 46.0,
-      "step": 12380
-    },
-    {
-      "epoch": 1.9938403317363824,
-      "grad_norm": 0.004674069117754698,
-      "learning_rate": 0.00019999804146601325,
-      "loss": 46.0,
-      "step": 12381
-    },
-    {
-      "epoch": 1.9940013688151697,
-      "grad_norm": 0.001432825461961329,
-      "learning_rate": 0.0001999980411493684,
-      "loss": 46.0,
-      "step": 12382
-    },
-    {
-      "epoch": 1.994162405893957,
-      "grad_norm": 0.003109590383246541,
-      "learning_rate": 0.00019999804083269793,
-      "loss": 46.0,
-      "step": 12383
-    },
-    {
-      "epoch": 1.9943234429727443,
-      "grad_norm": 0.003053095657378435,
-      "learning_rate": 0.00019999804051600188,
-      "loss": 46.0,
-      "step": 12384
-    },
-    {
-      "epoch": 1.9944844800515318,
-      "grad_norm": 0.0062919482588768005,
-      "learning_rate": 0.00019999804019928024,
-      "loss": 46.0,
-      "step": 12385
-    },
-    {
-      "epoch": 1.9946455171303192,
-      "grad_norm": 0.002199358306825161,
-      "learning_rate": 0.00019999803988253302,
-      "loss": 46.0,
-      "step": 12386
-    },
-    {
-      "epoch": 1.9948065542091067,
-      "grad_norm": 0.0048120892606675625,
-      "learning_rate": 0.0001999980395657602,
-      "loss": 46.0,
-      "step": 12387
-    },
-    {
-      "epoch": 1.9949675912878941,
-      "grad_norm": 0.0067680599167943,
-      "learning_rate": 0.00019999803924896178,
-      "loss": 46.0,
-      "step": 12388
-    },
-    {
-      "epoch": 1.9951286283666816,
-      "grad_norm": 0.0009305099956691265,
-      "learning_rate": 0.00019999803893213777,
-      "loss": 46.0,
-      "step": 12389
-    },
-    {
-      "epoch": 1.9952896654454688,
-      "grad_norm": 0.0011082387063652277,
-      "learning_rate": 0.00019999803861528814,
-      "loss": 46.0,
-      "step": 12390
-    },
-    {
-      "epoch": 1.995450702524256,
-      "grad_norm": 0.005581849720329046,
-      "learning_rate": 0.00019999803829841292,
-      "loss": 46.0,
-      "step": 12391
-    },
-    {
-      "epoch": 1.9956117396030435,
-      "grad_norm": 0.0034909637179225683,
-      "learning_rate": 0.00019999803798151215,
-      "loss": 46.0,
-      "step": 12392
-    },
-    {
-      "epoch": 1.995772776681831,
-      "grad_norm": 0.0030998988077044487,
-      "learning_rate": 0.00019999803766458576,
-      "loss": 46.0,
-      "step": 12393
-    },
-    {
-      "epoch": 1.9959338137606184,
-      "grad_norm": 0.0020426588598638773,
-      "learning_rate": 0.00019999803734763375,
-      "loss": 46.0,
-      "step": 12394
-    },
-    {
-      "epoch": 1.9960948508394059,
-      "grad_norm": 0.0010780710726976395,
-      "learning_rate": 0.0001999980370306562,
-      "loss": 46.0,
-      "step": 12395
-    },
-    {
-      "epoch": 1.9962558879181933,
-      "grad_norm": 0.0020355668384581804,
-      "learning_rate": 0.000199998036713653,
-      "loss": 46.0,
-      "step": 12396
-    },
-    {
-      "epoch": 1.9964169249969805,
-      "grad_norm": 0.00808663759380579,
-      "learning_rate": 0.00019999803639662425,
-      "loss": 46.0,
-      "step": 12397
-    },
-    {
-      "epoch": 1.996577962075768,
-      "grad_norm": 0.001027940772473812,
-      "learning_rate": 0.00019999803607956987,
-      "loss": 46.0,
-      "step": 12398
-    },
-    {
-      "epoch": 1.9967389991545552,
-      "grad_norm": 0.0017057600198313594,
-      "learning_rate": 0.0001999980357624899,
-      "loss": 46.0,
-      "step": 12399
-    },
-    {
-      "epoch": 1.9969000362333427,
-      "grad_norm": 0.0031915432773530483,
-      "learning_rate": 0.00019999803544538434,
-      "loss": 46.0,
-      "step": 12400
-    },
-    {
-      "epoch": 1.9970610733121301,
-      "grad_norm": 0.005933266133069992,
-      "learning_rate": 0.0001999980351282532,
-      "loss": 46.0,
-      "step": 12401
-    },
-    {
-      "epoch": 1.9972221103909176,
-      "grad_norm": 0.00589543953537941,
-      "learning_rate": 0.00019999803481109647,
-      "loss": 46.0,
-      "step": 12402
-    },
-    {
-      "epoch": 1.997383147469705,
-      "grad_norm": 0.003595212008804083,
-      "learning_rate": 0.00019999803449391413,
-      "loss": 46.0,
-      "step": 12403
-    },
-    {
-      "epoch": 1.9975441845484923,
-      "grad_norm": 0.005101496819406748,
-      "learning_rate": 0.0001999980341767062,
-      "loss": 46.0,
-      "step": 12404
-    },
-    {
-      "epoch": 1.9977052216272797,
-      "grad_norm": 0.0034306419547647238,
-      "learning_rate": 0.00019999803385947269,
-      "loss": 46.0,
-      "step": 12405
-    },
-    {
-      "epoch": 1.997866258706067,
-      "grad_norm": 0.003043942851945758,
-      "learning_rate": 0.00019999803354221355,
-      "loss": 46.0,
-      "step": 12406
-    },
-    {
-      "epoch": 1.9980272957848544,
-      "grad_norm": 0.0005994051462039351,
-      "learning_rate": 0.00019999803322492886,
-      "loss": 46.0,
-      "step": 12407
-    },
-    {
-      "epoch": 1.9981883328636418,
-      "grad_norm": 0.0021799085661768913,
-      "learning_rate": 0.00019999803290761853,
-      "loss": 46.0,
-      "step": 12408
-    },
-    {
-      "epoch": 1.9983493699424293,
-      "grad_norm": 0.0038078760262578726,
-      "learning_rate": 0.00019999803259028264,
-      "loss": 46.0,
-      "step": 12409
-    },
-    {
-      "epoch": 1.9985104070212167,
-      "grad_norm": 0.0029125921428203583,
-      "learning_rate": 0.00019999803227292113,
-      "loss": 46.0,
-      "step": 12410
-    },
-    {
-      "epoch": 1.9986714441000042,
-      "grad_norm": 0.011689925566315651,
-      "learning_rate": 0.00019999803195553404,
-      "loss": 46.0,
-      "step": 12411
-    },
-    {
-      "epoch": 1.9988324811787914,
-      "grad_norm": 0.00571091752499342,
-      "learning_rate": 0.00019999803163812135,
-      "loss": 46.0,
-      "step": 12412
-    },
-    {
-      "epoch": 1.9989935182575786,
-      "grad_norm": 0.00724672619253397,
-      "learning_rate": 0.00019999803132068306,
-      "loss": 46.0,
-      "step": 12413
-    },
-    {
-      "epoch": 1.999154555336366,
-      "grad_norm": 0.0006026331102475524,
-      "learning_rate": 0.00019999803100321918,
-      "loss": 46.0,
-      "step": 12414
-    },
-    {
-      "epoch": 1.9993155924151536,
-      "grad_norm": 0.008923816494643688,
-      "learning_rate": 0.00019999803068572973,
-      "loss": 46.0,
-      "step": 12415
-    },
-    {
-      "epoch": 1.999476629493941,
-      "grad_norm": 0.0019138951320201159,
-      "learning_rate": 0.00019999803036821465,
-      "loss": 46.0,
-      "step": 12416
-    },
-    {
-      "epoch": 1.9996376665727285,
-      "grad_norm": 0.004096074495464563,
-      "learning_rate": 0.000199998030050674,
-      "loss": 46.0,
-      "step": 12417
-    },
-    {
-      "epoch": 1.999798703651516,
-      "grad_norm": 0.002738706301897764,
-      "learning_rate": 0.00019999802973310775,
-      "loss": 46.0,
-      "step": 12418
-    },
-    {
-      "epoch": 1.9999597407303031,
-      "grad_norm": 0.0036281358916312456,
-      "learning_rate": 0.0001999980294155159,
-      "loss": 46.0,
-      "step": 12419
-    },
-    {
-      "epoch": 1.9999597407303031,
-      "eval_loss": 11.5,
-      "eval_runtime": 14.8838,
-      "eval_samples_per_second": 175.695,
-      "eval_steps_per_second": 87.881,
-      "step": 12419
-    },
-    {
-      "epoch": 2.0001610370787875,
-      "grad_norm": 0.008145570755004883,
-      "learning_rate": 0.00019999802909789847,
-      "loss": 46.0,
-      "step": 12420
-    },
-    {
-      "epoch": 2.000322074157575,
-      "grad_norm": 0.005223439075052738,
-      "learning_rate": 0.00019999802878025542,
-      "loss": 46.0,
-      "step": 12421
-    },
-    {
-      "epoch": 2.0004831112363624,
-      "grad_norm": 0.0007635240908712149,
-      "learning_rate": 0.0001999980284625868,
-      "loss": 46.0,
-      "step": 12422
-    },
-    {
-      "epoch": 2.0006441483151494,
-      "grad_norm": 0.003887373022735119,
-      "learning_rate": 0.0001999980281448926,
-      "loss": 46.0,
-      "step": 12423
-    },
-    {
-      "epoch": 2.000805185393937,
-      "grad_norm": 0.002868556184694171,
-      "learning_rate": 0.00019999802782717275,
-      "loss": 46.0,
-      "step": 12424
-    },
-    {
-      "epoch": 2.0009662224727243,
-      "grad_norm": 0.0013190106255933642,
-      "learning_rate": 0.00019999802750942736,
-      "loss": 46.0,
-      "step": 12425
-    },
-    {
-      "epoch": 2.0011272595515117,
-      "grad_norm": 0.003386542433872819,
-      "learning_rate": 0.00019999802719165635,
-      "loss": 46.0,
-      "step": 12426
-    },
-    {
-      "epoch": 2.001288296630299,
-      "grad_norm": 0.0027957663405686617,
-      "learning_rate": 0.00019999802687385972,
-      "loss": 46.0,
-      "step": 12427
-    },
-    {
-      "epoch": 2.0014493337090866,
-      "grad_norm": 0.005035643000155687,
-      "learning_rate": 0.00019999802655603754,
-      "loss": 46.0,
-      "step": 12428
-    },
-    {
-      "epoch": 2.001610370787874,
-      "grad_norm": 0.00128930585924536,
-      "learning_rate": 0.00019999802623818974,
-      "loss": 46.0,
-      "step": 12429
-    },
-    {
-      "epoch": 2.001771407866661,
-      "grad_norm": 0.002517638262361288,
-      "learning_rate": 0.00019999802592031638,
-      "loss": 46.0,
-      "step": 12430
-    },
-    {
-      "epoch": 2.0019324449454485,
-      "grad_norm": 0.0030535205733031034,
-      "learning_rate": 0.00019999802560241738,
-      "loss": 46.0,
-      "step": 12431
-    },
-    {
-      "epoch": 2.002093482024236,
-      "grad_norm": 0.003312586573883891,
-      "learning_rate": 0.0001999980252844928,
-      "loss": 46.0,
-      "step": 12432
-    },
-    {
-      "epoch": 2.0022545191030234,
-      "grad_norm": 0.002608202863484621,
-      "learning_rate": 0.00019999802496654265,
-      "loss": 46.0,
-      "step": 12433
-    },
-    {
-      "epoch": 2.002415556181811,
-      "grad_norm": 0.003110248129814863,
-      "learning_rate": 0.00019999802464856688,
-      "loss": 46.0,
-      "step": 12434
-    },
-    {
-      "epoch": 2.0025765932605983,
-      "grad_norm": 0.00047402121708728373,
-      "learning_rate": 0.0001999980243305655,
-      "loss": 46.0,
-      "step": 12435
-    },
-    {
-      "epoch": 2.002737630339386,
-      "grad_norm": 0.0006869040662422776,
-      "learning_rate": 0.00019999802401253857,
-      "loss": 46.0,
-      "step": 12436
-    },
-    {
-      "epoch": 2.0028986674181732,
-      "grad_norm": 0.0009354244684800506,
-      "learning_rate": 0.00019999802369448605,
-      "loss": 46.0,
-      "step": 12437
-    },
-    {
-      "epoch": 2.0030597044969602,
-      "grad_norm": 0.00657620606943965,
-      "learning_rate": 0.00019999802337640788,
-      "loss": 46.0,
-      "step": 12438
-    },
-    {
-      "epoch": 2.0032207415757477,
-      "grad_norm": 0.0007929457351565361,
-      "learning_rate": 0.00019999802305830416,
-      "loss": 46.0,
-      "step": 12439
-    },
-    {
-      "epoch": 2.003381778654535,
-      "grad_norm": 0.006124570500105619,
-      "learning_rate": 0.00019999802274017482,
-      "loss": 46.0,
-      "step": 12440
-    },
-    {
-      "epoch": 2.0035428157333226,
-      "grad_norm": 0.0017109365435317159,
-      "learning_rate": 0.00019999802242201992,
-      "loss": 46.0,
-      "step": 12441
-    },
-    {
-      "epoch": 2.00370385281211,
-      "grad_norm": 0.0011208091164007783,
-      "learning_rate": 0.00019999802210383938,
-      "loss": 46.0,
-      "step": 12442
-    },
-    {
-      "epoch": 2.0038648898908975,
-      "grad_norm": 0.0011709654936566949,
-      "learning_rate": 0.0001999980217856333,
-      "loss": 46.0,
-      "step": 12443
-    },
-    {
-      "epoch": 2.004025926969685,
-      "grad_norm": 0.0027863120194524527,
-      "learning_rate": 0.0001999980214674016,
-      "loss": 46.0,
-      "step": 12444
-    },
-    {
-      "epoch": 2.004186964048472,
-      "grad_norm": 0.0015842863358557224,
-      "learning_rate": 0.0001999980211491443,
-      "loss": 46.0,
-      "step": 12445
-    },
-    {
-      "epoch": 2.0043480011272594,
-      "grad_norm": 0.004461794160306454,
-      "learning_rate": 0.0001999980208308614,
-      "loss": 46.0,
-      "step": 12446
-    },
-    {
-      "epoch": 2.004509038206047,
-      "grad_norm": 0.0010177140356972814,
-      "learning_rate": 0.0001999980205125529,
-      "loss": 46.0,
-      "step": 12447
-    },
-    {
-      "epoch": 2.0046700752848343,
-      "grad_norm": 0.0020138367544859648,
-      "learning_rate": 0.0001999980201942188,
-      "loss": 46.0,
-      "step": 12448
-    },
-    {
-      "epoch": 2.0048311123636218,
-      "grad_norm": 0.003289173124358058,
-      "learning_rate": 0.00019999801987585913,
-      "loss": 46.0,
-      "step": 12449
-    },
-    {
-      "epoch": 2.004992149442409,
-      "grad_norm": 0.0035305796191096306,
-      "learning_rate": 0.00019999801955747386,
-      "loss": 46.0,
-      "step": 12450
-    },
-    {
-      "epoch": 2.0051531865211967,
-      "grad_norm": 0.0022860136814415455,
-      "learning_rate": 0.000199998019239063,
-      "loss": 46.0,
-      "step": 12451
-    },
-    {
-      "epoch": 2.0053142235999837,
-      "grad_norm": 0.0015331333270296454,
-      "learning_rate": 0.00019999801892062652,
-      "loss": 46.0,
-      "step": 12452
-    },
-    {
-      "epoch": 2.005475260678771,
-      "grad_norm": 0.008465247228741646,
-      "learning_rate": 0.00019999801860216446,
-      "loss": 46.0,
-      "step": 12453
-    },
-    {
-      "epoch": 2.0056362977575586,
-      "grad_norm": 0.0015940681332722306,
-      "learning_rate": 0.00019999801828367682,
-      "loss": 46.0,
-      "step": 12454
-    },
-    {
-      "epoch": 2.005797334836346,
-      "grad_norm": 0.009646641090512276,
-      "learning_rate": 0.00019999801796516357,
-      "loss": 46.0,
-      "step": 12455
-    },
-    {
-      "epoch": 2.0059583719151335,
-      "grad_norm": 0.006479586940258741,
-      "learning_rate": 0.00019999801764662473,
-      "loss": 46.0,
-      "step": 12456
-    },
-    {
-      "epoch": 2.006119408993921,
-      "grad_norm": 0.013124020770192146,
-      "learning_rate": 0.00019999801732806032,
-      "loss": 46.0,
-      "step": 12457
-    },
-    {
-      "epoch": 2.0062804460727084,
-      "grad_norm": 0.001967004267498851,
-      "learning_rate": 0.00019999801700947028,
-      "loss": 46.0,
-      "step": 12458
-    },
-    {
-      "epoch": 2.006441483151496,
-      "grad_norm": 0.008753619156777859,
-      "learning_rate": 0.00019999801669085468,
-      "loss": 46.0,
-      "step": 12459
-    },
-    {
-      "epoch": 2.006602520230283,
-      "grad_norm": 0.002026032656431198,
-      "learning_rate": 0.00019999801637221346,
-      "loss": 46.0,
-      "step": 12460
-    },
-    {
-      "epoch": 2.0067635573090703,
-      "grad_norm": 0.0008684775093570352,
-      "learning_rate": 0.00019999801605354665,
-      "loss": 46.0,
-      "step": 12461
-    },
-    {
-      "epoch": 2.0069245943878578,
-      "grad_norm": 0.0017957454547286034,
-      "learning_rate": 0.00019999801573485423,
-      "loss": 46.0,
-      "step": 12462
-    },
-    {
-      "epoch": 2.007085631466645,
-      "grad_norm": 0.010439875535666943,
-      "learning_rate": 0.00019999801541613623,
-      "loss": 46.0,
-      "step": 12463
-    },
-    {
-      "epoch": 2.0072466685454327,
-      "grad_norm": 0.00128399976529181,
-      "learning_rate": 0.00019999801509739263,
-      "loss": 46.0,
-      "step": 12464
-    },
-    {
-      "epoch": 2.00740770562422,
-      "grad_norm": 0.0012941404711455107,
-      "learning_rate": 0.00019999801477862345,
-      "loss": 46.0,
-      "step": 12465
-    },
-    {
-      "epoch": 2.0075687427030076,
-      "grad_norm": 0.0016802349127829075,
-      "learning_rate": 0.00019999801445982866,
-      "loss": 46.0,
-      "step": 12466
-    },
-    {
-      "epoch": 2.0077297797817946,
-      "grad_norm": 0.003150808857753873,
-      "learning_rate": 0.00019999801414100833,
-      "loss": 46.0,
-      "step": 12467
-    },
-    {
-      "epoch": 2.007890816860582,
-      "grad_norm": 0.0138546796515584,
-      "learning_rate": 0.00019999801382216233,
-      "loss": 46.0,
-      "step": 12468
-    },
-    {
-      "epoch": 2.0080518539393695,
-      "grad_norm": 0.0014055476058274508,
-      "learning_rate": 0.00019999801350329078,
-      "loss": 46.0,
-      "step": 12469
-    },
-    {
-      "epoch": 2.008212891018157,
-      "grad_norm": 0.001947572105564177,
-      "learning_rate": 0.0001999980131843936,
-      "loss": 46.0,
-      "step": 12470
-    },
-    {
-      "epoch": 2.0083739280969444,
-      "grad_norm": 0.004537053406238556,
-      "learning_rate": 0.00019999801286547087,
-      "loss": 46.0,
-      "step": 12471
-    },
-    {
-      "epoch": 2.008534965175732,
-      "grad_norm": 0.0028735287487506866,
-      "learning_rate": 0.0001999980125465225,
-      "loss": 46.0,
-      "step": 12472
-    },
-    {
-      "epoch": 2.0086960022545193,
-      "grad_norm": 0.0015545415226370096,
-      "learning_rate": 0.00019999801222754857,
-      "loss": 46.0,
-      "step": 12473
-    },
-    {
-      "epoch": 2.0088570393333063,
-      "grad_norm": 0.004813038744032383,
-      "learning_rate": 0.00019999801190854902,
-      "loss": 46.0,
-      "step": 12474
-    },
-    {
-      "epoch": 2.0090180764120937,
-      "grad_norm": 0.0013712627114728093,
-      "learning_rate": 0.0001999980115895239,
-      "loss": 46.0,
-      "step": 12475
-    },
-    {
-      "epoch": 2.009179113490881,
-      "grad_norm": 0.004010814242064953,
-      "learning_rate": 0.00019999801127047317,
-      "loss": 46.0,
-      "step": 12476
-    },
-    {
-      "epoch": 2.0093401505696686,
-      "grad_norm": 0.0028936523012816906,
-      "learning_rate": 0.00019999801095139686,
-      "loss": 46.0,
-      "step": 12477
-    },
-    {
-      "epoch": 2.009501187648456,
-      "grad_norm": 0.004158404655754566,
-      "learning_rate": 0.00019999801063229496,
-      "loss": 46.0,
-      "step": 12478
-    },
-    {
-      "epoch": 2.0096622247272435,
-      "grad_norm": 0.001904657226987183,
-      "learning_rate": 0.00019999801031316743,
-      "loss": 46.0,
-      "step": 12479
-    },
-    {
-      "epoch": 2.009823261806031,
-      "grad_norm": 0.002171450527384877,
-      "learning_rate": 0.00019999800999401433,
-      "loss": 46.0,
-      "step": 12480
-    },
-    {
-      "epoch": 2.0099842988848184,
-      "grad_norm": 0.005121269263327122,
-      "learning_rate": 0.00019999800967483564,
-      "loss": 46.0,
-      "step": 12481
-    },
-    {
-      "epoch": 2.0101453359636055,
-      "grad_norm": 0.0019119092030450702,
-      "learning_rate": 0.00019999800935563132,
-      "loss": 46.0,
-      "step": 12482
-    },
-    {
-      "epoch": 2.010306373042393,
-      "grad_norm": 0.0029260199517011642,
-      "learning_rate": 0.00019999800903640143,
-      "loss": 46.0,
-      "step": 12483
-    },
-    {
-      "epoch": 2.0104674101211804,
-      "grad_norm": 0.0010152553441002965,
-      "learning_rate": 0.000199998008717146,
-      "loss": 46.0,
-      "step": 12484
-    },
-    {
-      "epoch": 2.010628447199968,
-      "grad_norm": 0.004634276498109102,
-      "learning_rate": 0.0001999980083978649,
-      "loss": 46.0,
-      "step": 12485
-    },
-    {
-      "epoch": 2.0107894842787553,
-      "grad_norm": 0.004788288380950689,
-      "learning_rate": 0.00019999800807855823,
-      "loss": 46.0,
-      "step": 12486
-    },
-    {
-      "epoch": 2.0109505213575427,
-      "grad_norm": 0.0010063032386824489,
-      "learning_rate": 0.00019999800775922597,
-      "loss": 46.0,
-      "step": 12487
-    },
-    {
-      "epoch": 2.01111155843633,
-      "grad_norm": 0.004236471839249134,
-      "learning_rate": 0.0001999980074398681,
-      "loss": 46.0,
-      "step": 12488
-    },
-    {
-      "epoch": 2.011272595515117,
-      "grad_norm": 0.0013123058015480638,
-      "learning_rate": 0.00019999800712048465,
-      "loss": 46.0,
-      "step": 12489
-    },
-    {
-      "epoch": 2.0114336325939046,
-      "grad_norm": 0.0034236281644552946,
-      "learning_rate": 0.0001999980068010756,
-      "loss": 46.0,
-      "step": 12490
-    },
-    {
-      "epoch": 2.011594669672692,
-      "grad_norm": 0.001787322573363781,
-      "learning_rate": 0.00019999800648164097,
-      "loss": 46.0,
-      "step": 12491
-    },
-    {
-      "epoch": 2.0117557067514795,
-      "grad_norm": 0.001953398110345006,
-      "learning_rate": 0.00019999800616218074,
-      "loss": 46.0,
-      "step": 12492
-    },
-    {
-      "epoch": 2.011916743830267,
-      "grad_norm": 0.0016378588043153286,
-      "learning_rate": 0.0001999980058426949,
-      "loss": 46.0,
-      "step": 12493
-    },
-    {
-      "epoch": 2.0120777809090544,
-      "grad_norm": 0.010391559451818466,
-      "learning_rate": 0.00019999800552318348,
-      "loss": 46.0,
-      "step": 12494
-    },
-    {
-      "epoch": 2.012238817987842,
-      "grad_norm": 0.0006751961773261428,
-      "learning_rate": 0.00019999800520364647,
-      "loss": 46.0,
-      "step": 12495
-    },
-    {
-      "epoch": 2.012399855066629,
-      "grad_norm": 0.0050230189226567745,
-      "learning_rate": 0.00019999800488408384,
-      "loss": 46.0,
-      "step": 12496
-    },
-    {
-      "epoch": 2.0125608921454163,
-      "grad_norm": 0.0012957988074049354,
-      "learning_rate": 0.00019999800456449565,
-      "loss": 46.0,
-      "step": 12497
-    },
-    {
-      "epoch": 2.012721929224204,
-      "grad_norm": 0.0011957393726333976,
-      "learning_rate": 0.00019999800424488185,
-      "loss": 46.0,
-      "step": 12498
-    },
-    {
-      "epoch": 2.0128829663029912,
-      "grad_norm": 0.00101442018058151,
-      "learning_rate": 0.00019999800392524243,
-      "loss": 46.0,
-      "step": 12499
-    },
-    {
-      "epoch": 2.0130440033817787,
-      "grad_norm": 0.002004617126658559,
-      "learning_rate": 0.00019999800360557746,
-      "loss": 46.0,
-      "step": 12500
-    },
-    {
-      "epoch": 2.013205040460566,
-      "grad_norm": 0.0009082549950107932,
-      "learning_rate": 0.00019999800328588687,
-      "loss": 46.0,
-      "step": 12501
-    },
-    {
-      "epoch": 2.0133660775393536,
-      "grad_norm": 0.0003384132869541645,
-      "learning_rate": 0.0001999980029661707,
-      "loss": 46.0,
-      "step": 12502
-    },
-    {
-      "epoch": 2.013527114618141,
-      "grad_norm": 0.0033907992765307426,
-      "learning_rate": 0.00019999800264642895,
-      "loss": 46.0,
-      "step": 12503
-    },
-    {
-      "epoch": 2.013688151696928,
-      "grad_norm": 0.003555042902007699,
-      "learning_rate": 0.00019999800232666157,
-      "loss": 46.0,
-      "step": 12504
-    },
-    {
-      "epoch": 2.0138491887757155,
-      "grad_norm": 0.002011653268709779,
-      "learning_rate": 0.00019999800200686858,
-      "loss": 46.0,
-      "step": 12505
-    },
-    {
-      "epoch": 2.014010225854503,
-      "grad_norm": 0.0020098022650927305,
-      "learning_rate": 0.00019999800168705005,
-      "loss": 46.0,
-      "step": 12506
-    },
-    {
-      "epoch": 2.0141712629332904,
-      "grad_norm": 0.002362739061936736,
-      "learning_rate": 0.0001999980013672059,
-      "loss": 46.0,
-      "step": 12507
-    },
-    {
-      "epoch": 2.014332300012078,
-      "grad_norm": 0.001455823890864849,
-      "learning_rate": 0.00019999800104733616,
-      "loss": 46.0,
-      "step": 12508
-    },
-    {
-      "epoch": 2.0144933370908653,
-      "grad_norm": 0.005797697696834803,
-      "learning_rate": 0.00019999800072744082,
-      "loss": 46.0,
-      "step": 12509
-    },
-    {
-      "epoch": 2.0146543741696528,
-      "grad_norm": 0.004650041926652193,
-      "learning_rate": 0.0001999980004075199,
-      "loss": 46.0,
-      "step": 12510
-    },
-    {
-      "epoch": 2.0148154112484398,
-      "grad_norm": 0.0028827732894569635,
-      "learning_rate": 0.00019999800008757335,
-      "loss": 46.0,
-      "step": 12511
-    },
-    {
-      "epoch": 2.014976448327227,
-      "grad_norm": 0.0007172313635237515,
-      "learning_rate": 0.00019999799976760124,
-      "loss": 46.0,
-      "step": 12512
-    },
-    {
-      "epoch": 2.0151374854060147,
-      "grad_norm": 0.0020117510575801134,
-      "learning_rate": 0.00019999799944760352,
-      "loss": 46.0,
-      "step": 12513
-    },
-    {
-      "epoch": 2.015298522484802,
-      "grad_norm": 0.004131100606173277,
-      "learning_rate": 0.0001999979991275802,
-      "loss": 46.0,
-      "step": 12514
-    },
-    {
-      "epoch": 2.0154595595635896,
-      "grad_norm": 0.0020162505097687244,
-      "learning_rate": 0.0001999979988075313,
-      "loss": 46.0,
-      "step": 12515
-    },
-    {
-      "epoch": 2.015620596642377,
-      "grad_norm": 0.0005785388639196754,
-      "learning_rate": 0.0001999979984874568,
-      "loss": 46.0,
-      "step": 12516
-    },
-    {
-      "epoch": 2.0157816337211645,
-      "grad_norm": 0.0015340599929913878,
-      "learning_rate": 0.0001999979981673567,
-      "loss": 46.0,
-      "step": 12517
-    },
-    {
-      "epoch": 2.0159426707999515,
-      "grad_norm": 0.009182674810290337,
-      "learning_rate": 0.00019999799784723102,
-      "loss": 46.0,
-      "step": 12518
-    },
-    {
-      "epoch": 2.016103707878739,
-      "grad_norm": 0.002607476431876421,
-      "learning_rate": 0.00019999799752707973,
-      "loss": 46.0,
-      "step": 12519
-    },
-    {
-      "epoch": 2.0162647449575264,
-      "grad_norm": 0.008819465525448322,
-      "learning_rate": 0.00019999799720690284,
-      "loss": 46.0,
-      "step": 12520
-    },
-    {
-      "epoch": 2.016425782036314,
-      "grad_norm": 0.0014921671245247126,
-      "learning_rate": 0.00019999799688670038,
-      "loss": 46.0,
-      "step": 12521
-    },
-    {
-      "epoch": 2.0165868191151013,
-      "grad_norm": 0.0022067397367209196,
-      "learning_rate": 0.00019999799656647232,
-      "loss": 46.0,
-      "step": 12522
-    },
-    {
-      "epoch": 2.0167478561938887,
-      "grad_norm": 0.004075291100889444,
-      "learning_rate": 0.00019999799624621868,
-      "loss": 46.0,
-      "step": 12523
-    },
-    {
-      "epoch": 2.016908893272676,
-      "grad_norm": 0.005843945313245058,
-      "learning_rate": 0.0001999979959259394,
-      "loss": 46.0,
-      "step": 12524
-    },
-    {
-      "epoch": 2.0170699303514636,
-      "grad_norm": 0.004273079801350832,
-      "learning_rate": 0.00019999799560563455,
-      "loss": 46.0,
-      "step": 12525
-    },
-    {
-      "epoch": 2.0172309674302507,
-      "grad_norm": 0.003784353844821453,
-      "learning_rate": 0.0001999979952853041,
-      "loss": 46.0,
-      "step": 12526
-    },
-    {
-      "epoch": 2.017392004509038,
-      "grad_norm": 0.004242415074259043,
-      "learning_rate": 0.00019999799496494807,
-      "loss": 46.0,
-      "step": 12527
-    },
-    {
-      "epoch": 2.0175530415878256,
-      "grad_norm": 0.0064863828010857105,
-      "learning_rate": 0.00019999799464456643,
-      "loss": 46.0,
-      "step": 12528
-    },
-    {
-      "epoch": 2.017714078666613,
-      "grad_norm": 0.002360024256631732,
-      "learning_rate": 0.0001999979943241592,
-      "loss": 46.0,
-      "step": 12529
-    },
-    {
-      "epoch": 2.0178751157454005,
-      "grad_norm": 0.01183509360998869,
-      "learning_rate": 0.00019999799400372638,
-      "loss": 46.0,
-      "step": 12530
-    },
-    {
-      "epoch": 2.018036152824188,
-      "grad_norm": 0.001177316065877676,
-      "learning_rate": 0.00019999799368326798,
-      "loss": 46.0,
-      "step": 12531
-    },
-    {
-      "epoch": 2.0181971899029754,
-      "grad_norm": 0.0018052023369818926,
-      "learning_rate": 0.00019999799336278397,
-      "loss": 46.0,
-      "step": 12532
-    },
-    {
-      "epoch": 2.0183582269817624,
-      "grad_norm": 0.0010589802404865623,
-      "learning_rate": 0.00019999799304227435,
-      "loss": 46.0,
-      "step": 12533
-    },
-    {
-      "epoch": 2.01851926406055,
-      "grad_norm": 0.0011090995976701379,
-      "learning_rate": 0.00019999799272173917,
-      "loss": 46.0,
-      "step": 12534
-    },
-    {
-      "epoch": 2.0186803011393373,
-      "grad_norm": 0.0019332461524754763,
-      "learning_rate": 0.00019999799240117837,
-      "loss": 46.0,
-      "step": 12535
-    },
-    {
-      "epoch": 2.0188413382181247,
-      "grad_norm": 0.00811974797397852,
-      "learning_rate": 0.00019999799208059198,
-      "loss": 46.0,
-      "step": 12536
-    },
-    {
-      "epoch": 2.019002375296912,
-      "grad_norm": 0.002995526883751154,
-      "learning_rate": 0.00019999799175997998,
-      "loss": 46.0,
-      "step": 12537
-    },
-    {
-      "epoch": 2.0191634123756996,
-      "grad_norm": 0.0013452901039272547,
-      "learning_rate": 0.00019999799143934242,
-      "loss": 46.0,
-      "step": 12538
-    },
-    {
-      "epoch": 2.019324449454487,
-      "grad_norm": 0.0016195857897400856,
-      "learning_rate": 0.00019999799111867925,
-      "loss": 46.0,
-      "step": 12539
-    },
-    {
-      "epoch": 2.019485486533274,
-      "grad_norm": 0.016419364139437675,
-      "learning_rate": 0.0001999979907979905,
-      "loss": 46.0,
-      "step": 12540
-    },
-    {
-      "epoch": 2.0196465236120615,
-      "grad_norm": 0.005423232913017273,
-      "learning_rate": 0.00019999799047727614,
-      "loss": 46.0,
-      "step": 12541
-    },
-    {
-      "epoch": 2.019807560690849,
-      "grad_norm": 0.003274044021964073,
-      "learning_rate": 0.00019999799015653618,
-      "loss": 46.0,
-      "step": 12542
-    },
-    {
-      "epoch": 2.0199685977696364,
-      "grad_norm": 0.002777092158794403,
-      "learning_rate": 0.00019999798983577063,
-      "loss": 46.0,
-      "step": 12543
-    },
-    {
-      "epoch": 2.020129634848424,
-      "grad_norm": 0.005474648904055357,
-      "learning_rate": 0.0001999979895149795,
-      "loss": 46.0,
-      "step": 12544
-    },
-    {
-      "epoch": 2.0202906719272113,
-      "grad_norm": 0.0010304292663931847,
-      "learning_rate": 0.00019999798919416274,
-      "loss": 46.0,
-      "step": 12545
-    },
-    {
-      "epoch": 2.020451709005999,
-      "grad_norm": 0.0009154936415143311,
-      "learning_rate": 0.00019999798887332043,
-      "loss": 46.0,
-      "step": 12546
-    },
-    {
-      "epoch": 2.020612746084786,
-      "grad_norm": 0.0007544083637185395,
-      "learning_rate": 0.0001999979885524525,
-      "loss": 46.0,
-      "step": 12547
-    },
-    {
-      "epoch": 2.0207737831635733,
-      "grad_norm": 0.00824920367449522,
-      "learning_rate": 0.00019999798823155897,
-      "loss": 46.0,
-      "step": 12548
-    },
-    {
-      "epoch": 2.0209348202423607,
-      "grad_norm": 0.0035660217981785536,
-      "learning_rate": 0.00019999798791063987,
-      "loss": 46.0,
-      "step": 12549
-    },
-    {
-      "epoch": 2.021095857321148,
-      "grad_norm": 0.003505911212414503,
-      "learning_rate": 0.00019999798758969518,
-      "loss": 46.0,
-      "step": 12550
-    },
-    {
-      "epoch": 2.0212568943999356,
-      "grad_norm": 0.0022898083552718163,
-      "learning_rate": 0.00019999798726872485,
-      "loss": 46.0,
-      "step": 12551
-    },
-    {
-      "epoch": 2.021417931478723,
-      "grad_norm": 0.0015310329617932439,
-      "learning_rate": 0.00019999798694772896,
-      "loss": 46.0,
-      "step": 12552
-    },
-    {
-      "epoch": 2.0215789685575105,
-      "grad_norm": 0.0007999847293831408,
-      "learning_rate": 0.0001999979866267075,
-      "loss": 46.0,
-      "step": 12553
-    },
-    {
-      "epoch": 2.021740005636298,
-      "grad_norm": 0.0013681076234206557,
-      "learning_rate": 0.00019999798630566037,
-      "loss": 46.0,
-      "step": 12554
-    },
-    {
-      "epoch": 2.021901042715085,
-      "grad_norm": 0.010000420734286308,
-      "learning_rate": 0.0001999979859845877,
-      "loss": 46.0,
-      "step": 12555
-    },
-    {
-      "epoch": 2.0220620797938724,
-      "grad_norm": 0.00205078162252903,
-      "learning_rate": 0.00019999798566348943,
-      "loss": 46.0,
-      "step": 12556
-    },
-    {
-      "epoch": 2.02222311687266,
-      "grad_norm": 0.0012896863045170903,
-      "learning_rate": 0.00019999798534236555,
-      "loss": 46.0,
-      "step": 12557
-    },
-    {
-      "epoch": 2.0223841539514473,
-      "grad_norm": 0.0004371644463390112,
-      "learning_rate": 0.00019999798502121609,
-      "loss": 46.0,
-      "step": 12558
-    },
-    {
-      "epoch": 2.022545191030235,
-      "grad_norm": 0.005417472217231989,
-      "learning_rate": 0.00019999798470004106,
-      "loss": 46.0,
-      "step": 12559
-    },
-    {
-      "epoch": 2.0227062281090222,
-      "grad_norm": 0.001667987322434783,
-      "learning_rate": 0.0001999979843788404,
-      "loss": 46.0,
-      "step": 12560
-    },
-    {
-      "epoch": 2.0228672651878097,
-      "grad_norm": 0.004480540752410889,
-      "learning_rate": 0.00019999798405761414,
-      "loss": 46.0,
-      "step": 12561
-    },
-    {
-      "epoch": 2.0230283022665967,
-      "grad_norm": 0.008831013925373554,
-      "learning_rate": 0.00019999798373636232,
-      "loss": 46.0,
-      "step": 12562
-    },
-    {
-      "epoch": 2.023189339345384,
-      "grad_norm": 0.0016567183192819357,
-      "learning_rate": 0.00019999798341508487,
-      "loss": 46.0,
-      "step": 12563
-    },
-    {
-      "epoch": 2.0233503764241716,
-      "grad_norm": 0.0003811770584434271,
-      "learning_rate": 0.00019999798309378183,
-      "loss": 46.0,
-      "step": 12564
-    },
-    {
-      "epoch": 2.023511413502959,
-      "grad_norm": 0.0006745964055880904,
-      "learning_rate": 0.00019999798277245322,
-      "loss": 46.0,
-      "step": 12565
-    },
-    {
-      "epoch": 2.0236724505817465,
-      "grad_norm": 0.0057113138027489185,
-      "learning_rate": 0.000199997982451099,
-      "loss": 46.0,
-      "step": 12566
-    },
-    {
-      "epoch": 2.023833487660534,
-      "grad_norm": 0.011550800874829292,
-      "learning_rate": 0.00019999798212971917,
-      "loss": 46.0,
-      "step": 12567
-    },
-    {
-      "epoch": 2.0239945247393214,
-      "grad_norm": 0.0019337967969477177,
-      "learning_rate": 0.00019999798180831378,
-      "loss": 46.0,
-      "step": 12568
-    },
-    {
-      "epoch": 2.0241555618181084,
-      "grad_norm": 0.0020467578433454037,
-      "learning_rate": 0.00019999798148688278,
-      "loss": 46.0,
-      "step": 12569
-    },
-    {
-      "epoch": 2.024316598896896,
-      "grad_norm": 0.0032627317123115063,
-      "learning_rate": 0.00019999798116542618,
-      "loss": 46.0,
-      "step": 12570
-    },
-    {
-      "epoch": 2.0244776359756833,
-      "grad_norm": 0.0009862061124294996,
-      "learning_rate": 0.000199997980843944,
-      "loss": 46.0,
-      "step": 12571
-    },
-    {
-      "epoch": 2.0246386730544708,
-      "grad_norm": 0.0033507738262414932,
-      "learning_rate": 0.0001999979805224362,
-      "loss": 46.0,
-      "step": 12572
-    },
-    {
-      "epoch": 2.024799710133258,
-      "grad_norm": 0.000880859384778887,
-      "learning_rate": 0.00019999798020090283,
-      "loss": 46.0,
-      "step": 12573
-    },
-    {
-      "epoch": 2.0249607472120457,
-      "grad_norm": 0.0016476516611874104,
-      "learning_rate": 0.0001999979798793439,
-      "loss": 46.0,
-      "step": 12574
-    },
-    {
-      "epoch": 2.025121784290833,
-      "grad_norm": 0.0018749985611066222,
-      "learning_rate": 0.0001999979795577593,
-      "loss": 46.0,
-      "step": 12575
-    },
-    {
-      "epoch": 2.0252828213696206,
-      "grad_norm": 0.0035091356839984655,
-      "learning_rate": 0.00019999797923614913,
-      "loss": 46.0,
-      "step": 12576
-    },
-    {
-      "epoch": 2.0254438584484076,
-      "grad_norm": 0.002460634335875511,
-      "learning_rate": 0.0001999979789145134,
-      "loss": 46.0,
-      "step": 12577
-    },
-    {
-      "epoch": 2.025604895527195,
-      "grad_norm": 0.001522272708825767,
-      "learning_rate": 0.00019999797859285203,
-      "loss": 46.0,
-      "step": 12578
-    },
-    {
-      "epoch": 2.0257659326059825,
-      "grad_norm": 0.0009102607145905495,
-      "learning_rate": 0.0001999979782711651,
-      "loss": 46.0,
-      "step": 12579
-    },
-    {
-      "epoch": 2.02592696968477,
-      "grad_norm": 0.004572173114866018,
-      "learning_rate": 0.00019999797794945255,
-      "loss": 46.0,
-      "step": 12580
-    },
-    {
-      "epoch": 2.0260880067635574,
-      "grad_norm": 0.0006295728962868452,
-      "learning_rate": 0.00019999797762771442,
-      "loss": 46.0,
-      "step": 12581
-    },
-    {
-      "epoch": 2.026249043842345,
-      "grad_norm": 0.00104395707603544,
-      "learning_rate": 0.0001999979773059507,
-      "loss": 46.0,
-      "step": 12582
-    },
-    {
-      "epoch": 2.0264100809211323,
-      "grad_norm": 0.005980762653052807,
-      "learning_rate": 0.0001999979769841614,
-      "loss": 46.0,
-      "step": 12583
-    },
-    {
-      "epoch": 2.0265711179999193,
-      "grad_norm": 0.001454058918170631,
-      "learning_rate": 0.00019999797666234645,
-      "loss": 46.0,
-      "step": 12584
-    },
-    {
-      "epoch": 2.0267321550787067,
-      "grad_norm": 0.004987990017980337,
-      "learning_rate": 0.00019999797634050594,
-      "loss": 46.0,
-      "step": 12585
-    },
-    {
-      "epoch": 2.026893192157494,
-      "grad_norm": 0.004416910465806723,
-      "learning_rate": 0.00019999797601863987,
-      "loss": 46.0,
-      "step": 12586
-    },
-    {
-      "epoch": 2.0270542292362816,
-      "grad_norm": 0.0011555826058611274,
-      "learning_rate": 0.00019999797569674813,
-      "loss": 46.0,
-      "step": 12587
-    },
-    {
-      "epoch": 2.027215266315069,
-      "grad_norm": 0.005862301215529442,
-      "learning_rate": 0.00019999797537483086,
-      "loss": 46.0,
-      "step": 12588
-    },
-    {
-      "epoch": 2.0273763033938565,
-      "grad_norm": 0.008833087980747223,
-      "learning_rate": 0.00019999797505288797,
-      "loss": 46.0,
-      "step": 12589
-    },
-    {
-      "epoch": 2.027537340472644,
-      "grad_norm": 0.0021030204370617867,
-      "learning_rate": 0.00019999797473091948,
-      "loss": 46.0,
-      "step": 12590
-    },
-    {
-      "epoch": 2.027698377551431,
-      "grad_norm": 0.0026271375827491283,
-      "learning_rate": 0.00019999797440892542,
-      "loss": 46.0,
-      "step": 12591
-    },
-    {
-      "epoch": 2.0278594146302185,
-      "grad_norm": 0.008646892383694649,
-      "learning_rate": 0.00019999797408690574,
-      "loss": 46.0,
-      "step": 12592
-    },
-    {
-      "epoch": 2.028020451709006,
-      "grad_norm": 0.003314882516860962,
-      "learning_rate": 0.00019999797376486046,
-      "loss": 46.0,
-      "step": 12593
-    },
-    {
-      "epoch": 2.0281814887877934,
-      "grad_norm": 0.0013615135103464127,
-      "learning_rate": 0.0001999979734427896,
-      "loss": 46.0,
-      "step": 12594
-    },
-    {
-      "epoch": 2.028342525866581,
-      "grad_norm": 0.0067373900674283504,
-      "learning_rate": 0.00019999797312069317,
-      "loss": 46.0,
-      "step": 12595
-    },
-    {
-      "epoch": 2.0285035629453683,
-      "grad_norm": 0.0036978526040911674,
-      "learning_rate": 0.0001999979727985711,
-      "loss": 46.0,
-      "step": 12596
-    },
-    {
-      "epoch": 2.0286646000241557,
-      "grad_norm": 0.0008455890347249806,
-      "learning_rate": 0.00019999797247642346,
-      "loss": 46.0,
-      "step": 12597
-    },
-    {
-      "epoch": 2.028825637102943,
-      "grad_norm": 0.0017260325839743018,
-      "learning_rate": 0.00019999797215425024,
-      "loss": 46.0,
-      "step": 12598
-    },
-    {
-      "epoch": 2.02898667418173,
-      "grad_norm": 0.005904557649046183,
-      "learning_rate": 0.0001999979718320514,
-      "loss": 46.0,
-      "step": 12599
-    },
-    {
-      "epoch": 2.0291477112605176,
-      "grad_norm": 0.001210798160172999,
-      "learning_rate": 0.00019999797150982698,
-      "loss": 46.0,
-      "step": 12600
-    },
-    {
-      "epoch": 2.029308748339305,
-      "grad_norm": 0.0010602562688291073,
-      "learning_rate": 0.00019999797118757696,
-      "loss": 46.0,
-      "step": 12601
-    },
-    {
-      "epoch": 2.0294697854180925,
-      "grad_norm": 0.001755236298777163,
-      "learning_rate": 0.00019999797086530134,
-      "loss": 46.0,
-      "step": 12602
-    },
-    {
-      "epoch": 2.02963082249688,
-      "grad_norm": 0.004284292459487915,
-      "learning_rate": 0.00019999797054300012,
-      "loss": 46.0,
-      "step": 12603
-    },
-    {
-      "epoch": 2.0297918595756674,
-      "grad_norm": 0.0039684101939201355,
-      "learning_rate": 0.00019999797022067335,
-      "loss": 46.0,
-      "step": 12604
-    },
-    {
-      "epoch": 2.029952896654455,
-      "grad_norm": 0.0030487023759633303,
-      "learning_rate": 0.00019999796989832093,
-      "loss": 46.0,
-      "step": 12605
-    },
-    {
-      "epoch": 2.030113933733242,
-      "grad_norm": 0.0030690382700413465,
-      "learning_rate": 0.00019999796957594293,
-      "loss": 46.0,
-      "step": 12606
-    },
-    {
-      "epoch": 2.0302749708120293,
-      "grad_norm": 0.0019898023456335068,
-      "learning_rate": 0.00019999796925353937,
-      "loss": 46.0,
-      "step": 12607
-    },
-    {
-      "epoch": 2.030436007890817,
-      "grad_norm": 0.0011768710101023316,
-      "learning_rate": 0.00019999796893111017,
-      "loss": 46.0,
-      "step": 12608
-    },
-    {
-      "epoch": 2.0305970449696042,
-      "grad_norm": 0.006985583808273077,
-      "learning_rate": 0.0001999979686086554,
-      "loss": 46.0,
-      "step": 12609
-    },
-    {
-      "epoch": 2.0307580820483917,
-      "grad_norm": 0.011857746168971062,
-      "learning_rate": 0.00019999796828617505,
-      "loss": 46.0,
-      "step": 12610
-    },
-    {
-      "epoch": 2.030919119127179,
-      "grad_norm": 0.001817281125113368,
-      "learning_rate": 0.0001999979679636691,
-      "loss": 46.0,
-      "step": 12611
-    },
-    {
-      "epoch": 2.0310801562059666,
-      "grad_norm": 0.0022048638202250004,
-      "learning_rate": 0.0001999979676411375,
-      "loss": 46.0,
-      "step": 12612
-    },
-    {
-      "epoch": 2.0312411932847536,
-      "grad_norm": 0.0007136227213777602,
-      "learning_rate": 0.00019999796731858037,
-      "loss": 46.0,
-      "step": 12613
-    },
-    {
-      "epoch": 2.031402230363541,
-      "grad_norm": 0.002254886319860816,
-      "learning_rate": 0.00019999796699599762,
-      "loss": 46.0,
-      "step": 12614
-    },
-    {
-      "epoch": 2.0315632674423285,
-      "grad_norm": 0.0008421302190981805,
-      "learning_rate": 0.00019999796667338927,
-      "loss": 46.0,
-      "step": 12615
-    },
-    {
-      "epoch": 2.031724304521116,
-      "grad_norm": 0.0008954409859143198,
-      "learning_rate": 0.00019999796635075537,
-      "loss": 46.0,
-      "step": 12616
-    },
-    {
-      "epoch": 2.0318853415999034,
-      "grad_norm": 0.0011271878611296415,
-      "learning_rate": 0.00019999796602809583,
-      "loss": 46.0,
-      "step": 12617
-    },
-    {
-      "epoch": 2.032046378678691,
-      "grad_norm": 0.0015551706310361624,
-      "learning_rate": 0.0001999979657054107,
-      "loss": 46.0,
-      "step": 12618
-    },
-    {
-      "epoch": 2.0322074157574783,
-      "grad_norm": 0.0010714060626924038,
-      "learning_rate": 0.00019999796538269998,
-      "loss": 46.0,
-      "step": 12619
-    },
-    {
-      "epoch": 2.0323684528362653,
-      "grad_norm": 0.0012057216372340918,
-      "learning_rate": 0.00019999796505996365,
-      "loss": 46.0,
-      "step": 12620
-    },
-    {
-      "epoch": 2.032529489915053,
-      "grad_norm": 0.002392561873421073,
-      "learning_rate": 0.00019999796473720176,
-      "loss": 46.0,
-      "step": 12621
-    },
-    {
-      "epoch": 2.0326905269938402,
-      "grad_norm": 0.003465986577793956,
-      "learning_rate": 0.00019999796441441426,
-      "loss": 46.0,
-      "step": 12622
-    },
-    {
-      "epoch": 2.0328515640726277,
-      "grad_norm": 0.0007431176491081715,
-      "learning_rate": 0.00019999796409160117,
-      "loss": 46.0,
-      "step": 12623
-    },
-    {
-      "epoch": 2.033012601151415,
-      "grad_norm": 0.003488802118226886,
-      "learning_rate": 0.00019999796376876249,
-      "loss": 46.0,
-      "step": 12624
-    },
-    {
-      "epoch": 2.0331736382302026,
-      "grad_norm": 0.001669726800173521,
-      "learning_rate": 0.0001999979634458982,
-      "loss": 46.0,
-      "step": 12625
-    },
-    {
-      "epoch": 2.03333467530899,
-      "grad_norm": 0.004264045041054487,
-      "learning_rate": 0.0001999979631230083,
-      "loss": 46.0,
-      "step": 12626
-    },
-    {
-      "epoch": 2.0334957123877775,
-      "grad_norm": 0.0014957647072151303,
-      "learning_rate": 0.00019999796280009284,
-      "loss": 46.0,
-      "step": 12627
-    },
-    {
-      "epoch": 2.0336567494665645,
-      "grad_norm": 0.003543445374816656,
-      "learning_rate": 0.00019999796247715176,
-      "loss": 46.0,
-      "step": 12628
-    },
-    {
-      "epoch": 2.033817786545352,
-      "grad_norm": 0.008305152878165245,
-      "learning_rate": 0.00019999796215418512,
-      "loss": 46.0,
-      "step": 12629
-    },
-    {
-      "epoch": 2.0339788236241394,
-      "grad_norm": 0.0015745005803182721,
-      "learning_rate": 0.00019999796183119286,
-      "loss": 46.0,
-      "step": 12630
-    },
-    {
-      "epoch": 2.034139860702927,
-      "grad_norm": 0.004378624726086855,
-      "learning_rate": 0.000199997961508175,
-      "loss": 46.0,
-      "step": 12631
-    },
-    {
-      "epoch": 2.0343008977817143,
-      "grad_norm": 0.0025321580469608307,
-      "learning_rate": 0.00019999796118513156,
-      "loss": 46.0,
-      "step": 12632
-    },
-    {
-      "epoch": 2.0344619348605018,
-      "grad_norm": 0.0020093375351279974,
-      "learning_rate": 0.00019999796086206254,
-      "loss": 46.0,
-      "step": 12633
-    },
-    {
-      "epoch": 2.034622971939289,
-      "grad_norm": 0.005774980410933495,
-      "learning_rate": 0.00019999796053896788,
-      "loss": 46.0,
-      "step": 12634
-    },
-    {
-      "epoch": 2.034784009018076,
-      "grad_norm": 0.0015911576338112354,
-      "learning_rate": 0.00019999796021584766,
-      "loss": 46.0,
-      "step": 12635
-    },
-    {
-      "epoch": 2.0349450460968637,
-      "grad_norm": 0.0012758868979290128,
-      "learning_rate": 0.00019999795989270185,
-      "loss": 46.0,
-      "step": 12636
-    },
-    {
-      "epoch": 2.035106083175651,
-      "grad_norm": 0.006038719322532415,
-      "learning_rate": 0.00019999795956953043,
-      "loss": 46.0,
-      "step": 12637
-    },
-    {
-      "epoch": 2.0352671202544386,
-      "grad_norm": 0.0008050496107898653,
-      "learning_rate": 0.00019999795924633342,
-      "loss": 46.0,
-      "step": 12638
-    },
-    {
-      "epoch": 2.035428157333226,
-      "grad_norm": 0.0037582528311759233,
-      "learning_rate": 0.00019999795892311083,
-      "loss": 46.0,
-      "step": 12639
-    },
-    {
-      "epoch": 2.0355891944120135,
-      "grad_norm": 0.006221001967787743,
-      "learning_rate": 0.00019999795859986262,
-      "loss": 46.0,
-      "step": 12640
-    },
-    {
-      "epoch": 2.035750231490801,
-      "grad_norm": 0.0018074963008984923,
-      "learning_rate": 0.00019999795827658882,
-      "loss": 46.0,
-      "step": 12641
-    },
-    {
-      "epoch": 2.0359112685695884,
-      "grad_norm": 0.0023985167499631643,
-      "learning_rate": 0.00019999795795328944,
-      "loss": 46.0,
-      "step": 12642
-    },
-    {
-      "epoch": 2.0360723056483754,
-      "grad_norm": 0.0016935777384787798,
-      "learning_rate": 0.00019999795762996444,
-      "loss": 46.0,
-      "step": 12643
-    },
-    {
-      "epoch": 2.036233342727163,
-      "grad_norm": 0.0029211658984422684,
-      "learning_rate": 0.00019999795730661388,
-      "loss": 46.0,
-      "step": 12644
-    },
-    {
-      "epoch": 2.0363943798059503,
-      "grad_norm": 0.0013775217812508345,
-      "learning_rate": 0.0001999979569832377,
-      "loss": 46.0,
-      "step": 12645
-    },
-    {
-      "epoch": 2.0365554168847377,
-      "grad_norm": 0.0031842817552387714,
-      "learning_rate": 0.00019999795665983592,
-      "loss": 46.0,
-      "step": 12646
-    },
-    {
-      "epoch": 2.036716453963525,
-      "grad_norm": 0.0036360002122819424,
-      "learning_rate": 0.0001999979563364086,
-      "loss": 46.0,
-      "step": 12647
-    },
-    {
-      "epoch": 2.0368774910423126,
-      "grad_norm": 0.007040873169898987,
-      "learning_rate": 0.0001999979560129556,
-      "loss": 46.0,
-      "step": 12648
-    },
-    {
-      "epoch": 2.0370385281211,
-      "grad_norm": 0.0015644961968064308,
-      "learning_rate": 0.00019999795568947706,
-      "loss": 46.0,
-      "step": 12649
-    },
-    {
-      "epoch": 2.037199565199887,
-      "grad_norm": 0.0035487052518874407,
-      "learning_rate": 0.00019999795536597293,
-      "loss": 46.0,
-      "step": 12650
-    },
-    {
-      "epoch": 2.0373606022786745,
-      "grad_norm": 0.0020636108238250017,
-      "learning_rate": 0.00019999795504244318,
-      "loss": 46.0,
-      "step": 12651
-    },
-    {
-      "epoch": 2.037521639357462,
-      "grad_norm": 0.0022487486712634563,
-      "learning_rate": 0.00019999795471888784,
-      "loss": 46.0,
-      "step": 12652
-    },
-    {
-      "epoch": 2.0376826764362495,
-      "grad_norm": 0.0012746353168040514,
-      "learning_rate": 0.00019999795439530694,
-      "loss": 46.0,
-      "step": 12653
-    },
-    {
-      "epoch": 2.037843713515037,
-      "grad_norm": 0.0033677066676318645,
-      "learning_rate": 0.0001999979540717004,
-      "loss": 46.0,
-      "step": 12654
-    },
-    {
-      "epoch": 2.0380047505938244,
-      "grad_norm": 0.0022001334000378847,
-      "learning_rate": 0.00019999795374806828,
-      "loss": 46.0,
-      "step": 12655
-    },
-    {
-      "epoch": 2.038165787672612,
-      "grad_norm": 0.0033290269784629345,
-      "learning_rate": 0.0001999979534244106,
-      "loss": 46.0,
-      "step": 12656
-    },
-    {
-      "epoch": 2.038326824751399,
-      "grad_norm": 0.0032610977068543434,
-      "learning_rate": 0.0001999979531007273,
-      "loss": 46.0,
-      "step": 12657
-    },
-    {
-      "epoch": 2.0384878618301863,
-      "grad_norm": 0.001511602196842432,
-      "learning_rate": 0.00019999795277701838,
-      "loss": 46.0,
-      "step": 12658
-    },
-    {
-      "epoch": 2.0386488989089737,
-      "grad_norm": 0.0030205200891941786,
-      "learning_rate": 0.0001999979524532839,
-      "loss": 46.0,
-      "step": 12659
-    },
-    {
-      "epoch": 2.038809935987761,
-      "grad_norm": 0.005918892100453377,
-      "learning_rate": 0.0001999979521295238,
-      "loss": 46.0,
-      "step": 12660
-    },
-    {
-      "epoch": 2.0389709730665486,
-      "grad_norm": 0.001982569694519043,
-      "learning_rate": 0.00019999795180573814,
-      "loss": 46.0,
-      "step": 12661
-    },
-    {
-      "epoch": 2.039132010145336,
-      "grad_norm": 0.0024182030465453863,
-      "learning_rate": 0.00019999795148192685,
-      "loss": 46.0,
-      "step": 12662
-    },
-    {
-      "epoch": 2.0392930472241235,
-      "grad_norm": 0.0017946631414815784,
-      "learning_rate": 0.00019999795115809,
-      "loss": 46.0,
-      "step": 12663
-    },
-    {
-      "epoch": 2.0394540843029105,
-      "grad_norm": 0.0018103251932188869,
-      "learning_rate": 0.00019999795083422753,
-      "loss": 46.0,
-      "step": 12664
-    },
-    {
-      "epoch": 2.039615121381698,
-      "grad_norm": 0.0017367289401590824,
-      "learning_rate": 0.00019999795051033945,
-      "loss": 46.0,
-      "step": 12665
-    },
-    {
-      "epoch": 2.0397761584604854,
-      "grad_norm": 0.004122640006244183,
-      "learning_rate": 0.0001999979501864258,
-      "loss": 46.0,
-      "step": 12666
-    },
-    {
-      "epoch": 2.039937195539273,
-      "grad_norm": 0.0018341062823310494,
-      "learning_rate": 0.00019999794986248656,
-      "loss": 46.0,
-      "step": 12667
-    },
-    {
-      "epoch": 2.0400982326180603,
-      "grad_norm": 0.0010154114570468664,
-      "learning_rate": 0.00019999794953852172,
-      "loss": 46.0,
-      "step": 12668
-    },
-    {
-      "epoch": 2.040259269696848,
-      "grad_norm": 0.004763561766594648,
-      "learning_rate": 0.00019999794921453127,
-      "loss": 46.0,
-      "step": 12669
-    },
-    {
-      "epoch": 2.0404203067756352,
-      "grad_norm": 0.0014819520292803645,
-      "learning_rate": 0.00019999794889051525,
-      "loss": 46.0,
-      "step": 12670
-    },
-    {
-      "epoch": 2.0405813438544227,
-      "grad_norm": 0.002452469663694501,
-      "learning_rate": 0.00019999794856647362,
-      "loss": 46.0,
-      "step": 12671
-    },
-    {
-      "epoch": 2.0407423809332097,
-      "grad_norm": 0.0009148407843895257,
-      "learning_rate": 0.0001999979482424064,
-      "loss": 46.0,
-      "step": 12672
-    },
-    {
-      "epoch": 2.040903418011997,
-      "grad_norm": 0.0014726783847436309,
-      "learning_rate": 0.0001999979479183136,
-      "loss": 46.0,
-      "step": 12673
-    },
-    {
-      "epoch": 2.0410644550907846,
-      "grad_norm": 0.0015294309705495834,
-      "learning_rate": 0.00019999794759419518,
-      "loss": 46.0,
-      "step": 12674
-    },
-    {
-      "epoch": 2.041225492169572,
-      "grad_norm": 0.0019074780866503716,
-      "learning_rate": 0.00019999794727005118,
-      "loss": 46.0,
-      "step": 12675
-    },
-    {
-      "epoch": 2.0413865292483595,
-      "grad_norm": 0.00789949856698513,
-      "learning_rate": 0.00019999794694588162,
-      "loss": 46.0,
-      "step": 12676
-    },
-    {
-      "epoch": 2.041547566327147,
-      "grad_norm": 0.0014568453188985586,
-      "learning_rate": 0.0001999979466216864,
-      "loss": 46.0,
-      "step": 12677
-    },
-    {
-      "epoch": 2.0417086034059344,
-      "grad_norm": 0.0010552306193858385,
-      "learning_rate": 0.00019999794629746562,
-      "loss": 46.0,
-      "step": 12678
-    },
-    {
-      "epoch": 2.0418696404847214,
-      "grad_norm": 0.0004922501975670457,
-      "learning_rate": 0.00019999794597321926,
-      "loss": 46.0,
-      "step": 12679
-    },
-    {
-      "epoch": 2.042030677563509,
-      "grad_norm": 0.0026808297261595726,
-      "learning_rate": 0.00019999794564894727,
-      "loss": 46.0,
-      "step": 12680
-    },
-    {
-      "epoch": 2.0421917146422963,
-      "grad_norm": 0.0010481438366696239,
-      "learning_rate": 0.0001999979453246497,
-      "loss": 46.0,
-      "step": 12681
-    },
-    {
-      "epoch": 2.0423527517210838,
-      "grad_norm": 0.014141246676445007,
-      "learning_rate": 0.00019999794500032657,
-      "loss": 46.0,
-      "step": 12682
-    },
-    {
-      "epoch": 2.042513788799871,
-      "grad_norm": 0.001846129773184657,
-      "learning_rate": 0.00019999794467597779,
-      "loss": 46.0,
-      "step": 12683
-    },
-    {
-      "epoch": 2.0426748258786587,
-      "grad_norm": 0.002073851181194186,
-      "learning_rate": 0.00019999794435160344,
-      "loss": 46.0,
-      "step": 12684
-    },
-    {
-      "epoch": 2.042835862957446,
-      "grad_norm": 0.0026352526620030403,
-      "learning_rate": 0.0001999979440272035,
-      "loss": 46.0,
-      "step": 12685
-    },
-    {
-      "epoch": 2.042996900036233,
-      "grad_norm": 0.003275652416050434,
-      "learning_rate": 0.00019999794370277797,
-      "loss": 46.0,
-      "step": 12686
-    },
-    {
-      "epoch": 2.0431579371150206,
-      "grad_norm": 0.00578341307118535,
-      "learning_rate": 0.00019999794337832683,
-      "loss": 46.0,
-      "step": 12687
-    },
-    {
-      "epoch": 2.043318974193808,
-      "grad_norm": 0.0017546183662489057,
-      "learning_rate": 0.00019999794305385012,
-      "loss": 46.0,
-      "step": 12688
-    },
-    {
-      "epoch": 2.0434800112725955,
-      "grad_norm": 0.003102260874584317,
-      "learning_rate": 0.00019999794272934778,
-      "loss": 46.0,
-      "step": 12689
-    },
-    {
-      "epoch": 2.043641048351383,
-      "grad_norm": 0.0030546931084245443,
-      "learning_rate": 0.00019999794240481986,
-      "loss": 46.0,
-      "step": 12690
-    },
-    {
-      "epoch": 2.0438020854301704,
-      "grad_norm": 0.006423891056329012,
-      "learning_rate": 0.00019999794208026635,
-      "loss": 46.0,
-      "step": 12691
-    },
-    {
-      "epoch": 2.043963122508958,
-      "grad_norm": 0.0012006739852949977,
-      "learning_rate": 0.00019999794175568726,
-      "loss": 46.0,
-      "step": 12692
-    },
-    {
-      "epoch": 2.0441241595877453,
-      "grad_norm": 0.0014467312721535563,
-      "learning_rate": 0.00019999794143108255,
-      "loss": 46.0,
-      "step": 12693
-    },
-    {
-      "epoch": 2.0442851966665323,
-      "grad_norm": 0.0004815906868316233,
-      "learning_rate": 0.00019999794110645225,
-      "loss": 46.0,
-      "step": 12694
-    },
-    {
-      "epoch": 2.0444462337453198,
-      "grad_norm": 0.002055687829852104,
-      "learning_rate": 0.00019999794078179637,
-      "loss": 46.0,
-      "step": 12695
-    },
-    {
-      "epoch": 2.044607270824107,
-      "grad_norm": 0.002334614284336567,
-      "learning_rate": 0.0001999979404571149,
-      "loss": 46.0,
-      "step": 12696
-    },
-    {
-      "epoch": 2.0447683079028947,
-      "grad_norm": 0.0013030642876401544,
-      "learning_rate": 0.0001999979401324078,
-      "loss": 46.0,
-      "step": 12697
-    },
-    {
-      "epoch": 2.044929344981682,
-      "grad_norm": 0.0007385505014099181,
-      "learning_rate": 0.00019999793980767514,
-      "loss": 46.0,
-      "step": 12698
-    },
-    {
-      "epoch": 2.0450903820604696,
-      "grad_norm": 0.0025091059505939484,
-      "learning_rate": 0.00019999793948291688,
-      "loss": 46.0,
-      "step": 12699
-    },
-    {
-      "epoch": 2.045251419139257,
-      "grad_norm": 0.000989076099358499,
-      "learning_rate": 0.000199997939158133,
-      "loss": 46.0,
-      "step": 12700
-    },
-    {
-      "epoch": 2.045412456218044,
-      "grad_norm": 0.004465559963136911,
-      "learning_rate": 0.00019999793883332357,
-      "loss": 46.0,
-      "step": 12701
-    },
-    {
-      "epoch": 2.0455734932968315,
-      "grad_norm": 0.0021706765983253717,
-      "learning_rate": 0.00019999793850848852,
-      "loss": 46.0,
-      "step": 12702
-    },
-    {
-      "epoch": 2.045734530375619,
-      "grad_norm": 0.002952922135591507,
-      "learning_rate": 0.00019999793818362786,
-      "loss": 46.0,
-      "step": 12703
-    },
-    {
-      "epoch": 2.0458955674544064,
-      "grad_norm": 0.002846744144335389,
-      "learning_rate": 0.00019999793785874164,
-      "loss": 46.0,
-      "step": 12704
-    },
-    {
-      "epoch": 2.046056604533194,
-      "grad_norm": 0.003411015262827277,
-      "learning_rate": 0.0001999979375338298,
-      "loss": 46.0,
-      "step": 12705
-    },
-    {
-      "epoch": 2.0462176416119813,
-      "grad_norm": 0.002308523515239358,
-      "learning_rate": 0.00019999793720889235,
-      "loss": 46.0,
-      "step": 12706
-    },
-    {
-      "epoch": 2.0463786786907687,
-      "grad_norm": 0.0029919177759438753,
-      "learning_rate": 0.00019999793688392937,
-      "loss": 46.0,
-      "step": 12707
-    },
-    {
-      "epoch": 2.0465397157695557,
-      "grad_norm": 0.002041302155703306,
-      "learning_rate": 0.00019999793655894074,
-      "loss": 46.0,
-      "step": 12708
-    },
-    {
-      "epoch": 2.046700752848343,
-      "grad_norm": 0.004689004737883806,
-      "learning_rate": 0.0001999979362339265,
-      "loss": 46.0,
-      "step": 12709
-    },
-    {
-      "epoch": 2.0468617899271306,
-      "grad_norm": 0.0023718990851193666,
-      "learning_rate": 0.00019999793590888673,
-      "loss": 46.0,
-      "step": 12710
-    },
-    {
-      "epoch": 2.047022827005918,
-      "grad_norm": 0.0010333546670153737,
-      "learning_rate": 0.00019999793558382132,
-      "loss": 46.0,
-      "step": 12711
-    },
-    {
-      "epoch": 2.0471838640847055,
-      "grad_norm": 0.0007368649821728468,
-      "learning_rate": 0.00019999793525873031,
-      "loss": 46.0,
-      "step": 12712
-    },
-    {
-      "epoch": 2.047344901163493,
-      "grad_norm": 0.003788518253713846,
-      "learning_rate": 0.00019999793493361375,
-      "loss": 46.0,
-      "step": 12713
-    },
-    {
-      "epoch": 2.0475059382422804,
-      "grad_norm": 0.0016053363215178251,
-      "learning_rate": 0.00019999793460847155,
-      "loss": 46.0,
-      "step": 12714
-    },
-    {
-      "epoch": 2.047666975321068,
-      "grad_norm": 0.003955433145165443,
-      "learning_rate": 0.0001999979342833038,
-      "loss": 46.0,
-      "step": 12715
-    },
-    {
-      "epoch": 2.047828012399855,
-      "grad_norm": 0.005060320254415274,
-      "learning_rate": 0.0001999979339581104,
-      "loss": 46.0,
-      "step": 12716
-    },
-    {
-      "epoch": 2.0479890494786424,
-      "grad_norm": 0.002011056523770094,
-      "learning_rate": 0.00019999793363289145,
-      "loss": 46.0,
-      "step": 12717
-    },
-    {
-      "epoch": 2.04815008655743,
-      "grad_norm": 0.001769879600033164,
-      "learning_rate": 0.0001999979333076469,
-      "loss": 46.0,
-      "step": 12718
-    },
-    {
-      "epoch": 2.0483111236362173,
-      "grad_norm": 0.003078615292906761,
-      "learning_rate": 0.00019999793298237676,
-      "loss": 46.0,
-      "step": 12719
-    },
-    {
-      "epoch": 2.0484721607150047,
-      "grad_norm": 0.0004540045338217169,
-      "learning_rate": 0.000199997932657081,
-      "loss": 46.0,
-      "step": 12720
-    },
-    {
-      "epoch": 2.048633197793792,
-      "grad_norm": 0.0016792655223980546,
-      "learning_rate": 0.00019999793233175966,
-      "loss": 46.0,
-      "step": 12721
-    },
-    {
-      "epoch": 2.0487942348725796,
-      "grad_norm": 0.0018158906605094671,
-      "learning_rate": 0.00019999793200641274,
-      "loss": 46.0,
-      "step": 12722
-    },
-    {
-      "epoch": 2.0489552719513666,
-      "grad_norm": 0.001656759181059897,
-      "learning_rate": 0.0001999979316810402,
-      "loss": 46.0,
-      "step": 12723
-    },
-    {
-      "epoch": 2.049116309030154,
-      "grad_norm": 0.0052232141606509686,
-      "learning_rate": 0.00019999793135564207,
-      "loss": 46.0,
-      "step": 12724
-    },
-    {
-      "epoch": 2.0492773461089415,
-      "grad_norm": 0.006812802981585264,
-      "learning_rate": 0.00019999793103021835,
-      "loss": 46.0,
-      "step": 12725
-    },
-    {
-      "epoch": 2.049438383187729,
-      "grad_norm": 0.0046859378926455975,
-      "learning_rate": 0.00019999793070476902,
-      "loss": 46.0,
-      "step": 12726
-    },
-    {
-      "epoch": 2.0495994202665164,
-      "grad_norm": 0.0004233905056025833,
-      "learning_rate": 0.00019999793037929413,
-      "loss": 46.0,
-      "step": 12727
-    },
-    {
-      "epoch": 2.049760457345304,
-      "grad_norm": 0.002785461489111185,
-      "learning_rate": 0.00019999793005379363,
-      "loss": 46.0,
-      "step": 12728
-    },
-    {
-      "epoch": 2.0499214944240913,
-      "grad_norm": 0.0018482727464288473,
-      "learning_rate": 0.0001999979297282675,
-      "loss": 46.0,
-      "step": 12729
-    },
-    {
-      "epoch": 2.0500825315028783,
-      "grad_norm": 0.0068340180441737175,
-      "learning_rate": 0.00019999792940271586,
-      "loss": 46.0,
-      "step": 12730
-    },
-    {
-      "epoch": 2.050243568581666,
-      "grad_norm": 0.00059436698211357,
-      "learning_rate": 0.00019999792907713854,
-      "loss": 46.0,
-      "step": 12731
-    },
-    {
-      "epoch": 2.0504046056604532,
-      "grad_norm": 0.003365225624293089,
-      "learning_rate": 0.00019999792875153566,
-      "loss": 46.0,
-      "step": 12732
-    },
-    {
-      "epoch": 2.0505656427392407,
-      "grad_norm": 0.002422794932499528,
-      "learning_rate": 0.00019999792842590722,
-      "loss": 46.0,
-      "step": 12733
-    },
-    {
-      "epoch": 2.050726679818028,
-      "grad_norm": 0.0009200755739584565,
-      "learning_rate": 0.00019999792810025313,
-      "loss": 46.0,
-      "step": 12734
-    },
-    {
-      "epoch": 2.0508877168968156,
-      "grad_norm": 0.004491645842790604,
-      "learning_rate": 0.00019999792777457347,
-      "loss": 46.0,
-      "step": 12735
-    },
-    {
-      "epoch": 2.051048753975603,
-      "grad_norm": 0.0031578242778778076,
-      "learning_rate": 0.00019999792744886824,
-      "loss": 46.0,
-      "step": 12736
-    },
-    {
-      "epoch": 2.05120979105439,
-      "grad_norm": 0.001434703473933041,
-      "learning_rate": 0.00019999792712313737,
-      "loss": 46.0,
-      "step": 12737
-    },
-    {
-      "epoch": 2.0513708281331775,
-      "grad_norm": 0.004946847911924124,
-      "learning_rate": 0.0001999979267973809,
-      "loss": 46.0,
-      "step": 12738
-    },
-    {
-      "epoch": 2.051531865211965,
-      "grad_norm": 0.002262961817905307,
-      "learning_rate": 0.0001999979264715989,
-      "loss": 46.0,
-      "step": 12739
-    },
-    {
-      "epoch": 2.0516929022907524,
-      "grad_norm": 0.001229085261002183,
-      "learning_rate": 0.00019999792614579126,
-      "loss": 46.0,
-      "step": 12740
-    },
-    {
-      "epoch": 2.05185393936954,
-      "grad_norm": 0.0016119007486850023,
-      "learning_rate": 0.000199997925819958,
-      "loss": 46.0,
-      "step": 12741
-    },
-    {
-      "epoch": 2.0520149764483273,
-      "grad_norm": 0.002097120974212885,
-      "learning_rate": 0.0001999979254940992,
-      "loss": 46.0,
-      "step": 12742
-    },
-    {
-      "epoch": 2.0521760135271148,
-      "grad_norm": 0.006228986196219921,
-      "learning_rate": 0.0001999979251682148,
-      "loss": 46.0,
-      "step": 12743
-    },
-    {
-      "epoch": 2.052337050605902,
-      "grad_norm": 0.0009796464582905173,
-      "learning_rate": 0.00019999792484230478,
-      "loss": 46.0,
-      "step": 12744
-    },
-    {
-      "epoch": 2.052498087684689,
-      "grad_norm": 0.0006819817936047912,
-      "learning_rate": 0.00019999792451636918,
-      "loss": 46.0,
-      "step": 12745
-    },
-    {
-      "epoch": 2.0526591247634767,
-      "grad_norm": 0.0014942156849429011,
-      "learning_rate": 0.00019999792419040798,
-      "loss": 46.0,
-      "step": 12746
-    },
-    {
-      "epoch": 2.052820161842264,
-      "grad_norm": 0.001344421529211104,
-      "learning_rate": 0.00019999792386442118,
-      "loss": 46.0,
-      "step": 12747
-    },
-    {
-      "epoch": 2.0529811989210516,
-      "grad_norm": 0.0029171700589358807,
-      "learning_rate": 0.0001999979235384088,
-      "loss": 46.0,
-      "step": 12748
-    },
-    {
-      "epoch": 2.053142235999839,
-      "grad_norm": 0.017736775800585747,
-      "learning_rate": 0.0001999979232123708,
-      "loss": 46.0,
-      "step": 12749
-    },
-    {
-      "epoch": 2.0533032730786265,
-      "grad_norm": 0.0020546342711895704,
-      "learning_rate": 0.00019999792288630724,
-      "loss": 46.0,
-      "step": 12750
-    },
-    {
-      "epoch": 2.053464310157414,
-      "grad_norm": 0.0008654060657136142,
-      "learning_rate": 0.00019999792256021807,
-      "loss": 46.0,
-      "step": 12751
-    },
-    {
-      "epoch": 2.053625347236201,
-      "grad_norm": 0.02104876935482025,
-      "learning_rate": 0.0001999979222341033,
-      "loss": 46.0,
-      "step": 12752
-    },
-    {
-      "epoch": 2.0537863843149884,
-      "grad_norm": 0.002866534749045968,
-      "learning_rate": 0.00019999792190796294,
-      "loss": 46.0,
-      "step": 12753
-    },
-    {
-      "epoch": 2.053947421393776,
-      "grad_norm": 0.0024026569444686174,
-      "learning_rate": 0.00019999792158179698,
-      "loss": 46.0,
-      "step": 12754
-    },
-    {
-      "epoch": 2.0541084584725633,
-      "grad_norm": 0.0010798060102388263,
-      "learning_rate": 0.00019999792125560543,
-      "loss": 46.0,
-      "step": 12755
-    },
-    {
-      "epoch": 2.0542694955513507,
-      "grad_norm": 0.0018468784401193261,
-      "learning_rate": 0.0001999979209293883,
-      "loss": 46.0,
-      "step": 12756
-    },
-    {
-      "epoch": 2.054430532630138,
-      "grad_norm": 0.00404136348515749,
-      "learning_rate": 0.00019999792060314554,
-      "loss": 46.0,
-      "step": 12757
-    },
-    {
-      "epoch": 2.0545915697089256,
-      "grad_norm": 0.001251041772775352,
-      "learning_rate": 0.00019999792027687723,
-      "loss": 46.0,
-      "step": 12758
-    },
-    {
-      "epoch": 2.0547526067877127,
-      "grad_norm": 0.0028351852670311928,
-      "learning_rate": 0.0001999979199505833,
-      "loss": 46.0,
-      "step": 12759
-    },
-    {
-      "epoch": 2.0549136438665,
-      "grad_norm": 0.002025420079007745,
-      "learning_rate": 0.00019999791962426377,
-      "loss": 46.0,
-      "step": 12760
-    },
-    {
-      "epoch": 2.0550746809452876,
-      "grad_norm": 0.002291368320584297,
-      "learning_rate": 0.00019999791929791868,
-      "loss": 46.0,
-      "step": 12761
-    },
-    {
-      "epoch": 2.055235718024075,
-      "grad_norm": 0.004026977811008692,
-      "learning_rate": 0.00019999791897154796,
-      "loss": 46.0,
-      "step": 12762
-    },
-    {
-      "epoch": 2.0553967551028625,
-      "grad_norm": 0.0031712024938315153,
-      "learning_rate": 0.00019999791864515164,
-      "loss": 46.0,
-      "step": 12763
-    },
-    {
-      "epoch": 2.05555779218165,
-      "grad_norm": 0.0044436086900532246,
-      "learning_rate": 0.00019999791831872975,
-      "loss": 46.0,
-      "step": 12764
-    },
-    {
-      "epoch": 2.0557188292604374,
-      "grad_norm": 0.008992526680231094,
-      "learning_rate": 0.00019999791799228228,
-      "loss": 46.0,
-      "step": 12765
-    },
-    {
-      "epoch": 2.055879866339225,
-      "grad_norm": 0.0018407624447718263,
-      "learning_rate": 0.0001999979176658092,
-      "loss": 46.0,
-      "step": 12766
-    },
-    {
-      "epoch": 2.056040903418012,
-      "grad_norm": 0.003313424065709114,
-      "learning_rate": 0.00019999791733931051,
-      "loss": 46.0,
-      "step": 12767
-    },
-    {
-      "epoch": 2.0562019404967993,
-      "grad_norm": 0.0008020804962143302,
-      "learning_rate": 0.00019999791701278622,
-      "loss": 46.0,
-      "step": 12768
-    },
-    {
-      "epoch": 2.0563629775755867,
-      "grad_norm": 0.007835103198885918,
-      "learning_rate": 0.00019999791668623635,
-      "loss": 46.0,
-      "step": 12769
-    },
-    {
-      "epoch": 2.056524014654374,
-      "grad_norm": 0.0018140114843845367,
-      "learning_rate": 0.00019999791635966088,
-      "loss": 46.0,
-      "step": 12770
-    },
-    {
-      "epoch": 2.0566850517331616,
-      "grad_norm": 0.0009696898050606251,
-      "learning_rate": 0.00019999791603305983,
-      "loss": 46.0,
-      "step": 12771
-    },
-    {
-      "epoch": 2.056846088811949,
-      "grad_norm": 0.0019198352238163352,
-      "learning_rate": 0.00019999791570643317,
-      "loss": 46.0,
-      "step": 12772
-    },
-    {
-      "epoch": 2.0570071258907365,
-      "grad_norm": 0.0011088484898209572,
-      "learning_rate": 0.00019999791537978094,
-      "loss": 46.0,
-      "step": 12773
-    },
-    {
-      "epoch": 2.0571681629695235,
-      "grad_norm": 0.0034949174150824547,
-      "learning_rate": 0.0001999979150531031,
-      "loss": 46.0,
-      "step": 12774
-    },
-    {
-      "epoch": 2.057329200048311,
-      "grad_norm": 0.0019022391643375158,
-      "learning_rate": 0.00019999791472639965,
-      "loss": 46.0,
-      "step": 12775
-    },
-    {
-      "epoch": 2.0574902371270984,
-      "grad_norm": 0.006527475547045469,
-      "learning_rate": 0.00019999791439967064,
-      "loss": 46.0,
-      "step": 12776
-    },
-    {
-      "epoch": 2.057651274205886,
-      "grad_norm": 0.0025081345811486244,
-      "learning_rate": 0.000199997914072916,
-      "loss": 46.0,
-      "step": 12777
-    },
-    {
-      "epoch": 2.0578123112846733,
-      "grad_norm": 0.001696562161669135,
-      "learning_rate": 0.00019999791374613577,
-      "loss": 46.0,
-      "step": 12778
-    },
-    {
-      "epoch": 2.057973348363461,
-      "grad_norm": 0.0009385391604155302,
-      "learning_rate": 0.00019999791341933,
-      "loss": 46.0,
-      "step": 12779
-    },
-    {
-      "epoch": 2.0581343854422482,
-      "grad_norm": 0.0030935273971408606,
-      "learning_rate": 0.00019999791309249855,
-      "loss": 46.0,
-      "step": 12780
-    },
-    {
-      "epoch": 2.0582954225210353,
-      "grad_norm": 0.001058962196111679,
-      "learning_rate": 0.00019999791276564154,
-      "loss": 46.0,
-      "step": 12781
-    },
-    {
-      "epoch": 2.0584564595998227,
-      "grad_norm": 0.0014233856927603483,
-      "learning_rate": 0.00019999791243875895,
-      "loss": 46.0,
-      "step": 12782
-    },
-    {
-      "epoch": 2.05861749667861,
-      "grad_norm": 0.003509201342239976,
-      "learning_rate": 0.00019999791211185077,
-      "loss": 46.0,
-      "step": 12783
-    },
-    {
-      "epoch": 2.0587785337573976,
-      "grad_norm": 0.004560582805424929,
-      "learning_rate": 0.00019999791178491698,
-      "loss": 46.0,
-      "step": 12784
-    },
-    {
-      "epoch": 2.058939570836185,
-      "grad_norm": 0.0026095740031450987,
-      "learning_rate": 0.0001999979114579576,
-      "loss": 46.0,
-      "step": 12785
-    },
-    {
-      "epoch": 2.0591006079149725,
-      "grad_norm": 0.0011534725781530142,
-      "learning_rate": 0.00019999791113097263,
-      "loss": 46.0,
-      "step": 12786
-    },
-    {
-      "epoch": 2.05926164499376,
-      "grad_norm": 0.0037338025867938995,
-      "learning_rate": 0.00019999791080396205,
-      "loss": 46.0,
-      "step": 12787
-    },
-    {
-      "epoch": 2.0594226820725474,
-      "grad_norm": 0.0014899610541760921,
-      "learning_rate": 0.0001999979104769259,
-      "loss": 46.0,
-      "step": 12788
-    },
-    {
-      "epoch": 2.0595837191513344,
-      "grad_norm": 0.002470711711794138,
-      "learning_rate": 0.00019999791014986416,
-      "loss": 46.0,
-      "step": 12789
-    },
-    {
-      "epoch": 2.059744756230122,
-      "grad_norm": 0.0021710586734116077,
-      "learning_rate": 0.00019999790982277679,
-      "loss": 46.0,
-      "step": 12790
-    },
-    {
-      "epoch": 2.0599057933089093,
-      "grad_norm": 0.01125306822359562,
-      "learning_rate": 0.00019999790949566386,
-      "loss": 46.0,
-      "step": 12791
-    },
-    {
-      "epoch": 2.060066830387697,
-      "grad_norm": 0.001020616153255105,
-      "learning_rate": 0.0001999979091685253,
-      "loss": 46.0,
-      "step": 12792
-    },
-    {
-      "epoch": 2.0602278674664842,
-      "grad_norm": 0.0013511452125385404,
-      "learning_rate": 0.00019999790884136118,
-      "loss": 46.0,
-      "step": 12793
-    },
-    {
-      "epoch": 2.0603889045452717,
-      "grad_norm": 0.0037048342637717724,
-      "learning_rate": 0.00019999790851417146,
-      "loss": 46.0,
-      "step": 12794
-    },
-    {
-      "epoch": 2.060549941624059,
-      "grad_norm": 0.001499238540418446,
-      "learning_rate": 0.00019999790818695613,
-      "loss": 46.0,
-      "step": 12795
-    },
-    {
-      "epoch": 2.060710978702846,
-      "grad_norm": 0.0019282250432297587,
-      "learning_rate": 0.0001999979078597152,
-      "loss": 46.0,
-      "step": 12796
-    },
-    {
-      "epoch": 2.0608720157816336,
-      "grad_norm": 0.0005515927914530039,
-      "learning_rate": 0.0001999979075324487,
-      "loss": 46.0,
-      "step": 12797
-    },
-    {
-      "epoch": 2.061033052860421,
-      "grad_norm": 0.002887873910367489,
-      "learning_rate": 0.0001999979072051566,
-      "loss": 46.0,
-      "step": 12798
-    },
-    {
-      "epoch": 2.0611940899392085,
-      "grad_norm": 0.008013742975890636,
-      "learning_rate": 0.0001999979068778389,
-      "loss": 46.0,
-      "step": 12799
-    },
-    {
-      "epoch": 2.061355127017996,
-      "grad_norm": 0.011362212710082531,
-      "learning_rate": 0.0001999979065504956,
-      "loss": 46.0,
-      "step": 12800
-    },
-    {
-      "epoch": 2.0615161640967834,
-      "grad_norm": 0.004647592082619667,
-      "learning_rate": 0.00019999790622312672,
-      "loss": 46.0,
-      "step": 12801
-    },
-    {
-      "epoch": 2.061677201175571,
-      "grad_norm": 0.0007746084011159837,
-      "learning_rate": 0.00019999790589573225,
-      "loss": 46.0,
-      "step": 12802
-    },
-    {
-      "epoch": 2.061838238254358,
-      "grad_norm": 0.002557308180257678,
-      "learning_rate": 0.00019999790556831217,
-      "loss": 46.0,
-      "step": 12803
-    },
-    {
-      "epoch": 2.0619992753331453,
-      "grad_norm": 0.0033991560339927673,
-      "learning_rate": 0.0001999979052408665,
-      "loss": 46.0,
-      "step": 12804
-    },
-    {
-      "epoch": 2.0621603124119328,
-      "grad_norm": 0.0017291069962084293,
-      "learning_rate": 0.00019999790491339524,
-      "loss": 46.0,
-      "step": 12805
-    },
-    {
-      "epoch": 2.06232134949072,
-      "grad_norm": 0.0011777442414313555,
-      "learning_rate": 0.00019999790458589836,
-      "loss": 46.0,
-      "step": 12806
-    },
-    {
-      "epoch": 2.0624823865695077,
-      "grad_norm": 0.0020870326552540064,
-      "learning_rate": 0.00019999790425837593,
-      "loss": 46.0,
-      "step": 12807
-    },
-    {
-      "epoch": 2.062643423648295,
-      "grad_norm": 0.0026128143072128296,
-      "learning_rate": 0.00019999790393082786,
-      "loss": 46.0,
-      "step": 12808
-    },
-    {
-      "epoch": 2.0628044607270826,
-      "grad_norm": 0.00748836062848568,
-      "learning_rate": 0.00019999790360325422,
-      "loss": 46.0,
-      "step": 12809
-    },
-    {
-      "epoch": 2.0629654978058696,
-      "grad_norm": 0.004799081943929195,
-      "learning_rate": 0.00019999790327565497,
-      "loss": 46.0,
-      "step": 12810
-    },
-    {
-      "epoch": 2.063126534884657,
-      "grad_norm": 0.003735225647687912,
-      "learning_rate": 0.00019999790294803016,
-      "loss": 46.0,
-      "step": 12811
-    },
-    {
-      "epoch": 2.0632875719634445,
-      "grad_norm": 0.0010754746617749333,
-      "learning_rate": 0.00019999790262037971,
-      "loss": 46.0,
-      "step": 12812
-    },
-    {
-      "epoch": 2.063448609042232,
-      "grad_norm": 0.006830889265984297,
-      "learning_rate": 0.0001999979022927037,
-      "loss": 46.0,
-      "step": 12813
-    },
-    {
-      "epoch": 2.0636096461210194,
-      "grad_norm": 0.005894643720239401,
-      "learning_rate": 0.0001999979019650021,
-      "loss": 46.0,
-      "step": 12814
-    },
-    {
-      "epoch": 2.063770683199807,
-      "grad_norm": 0.002159679774194956,
-      "learning_rate": 0.00019999790163727487,
-      "loss": 46.0,
-      "step": 12815
-    },
-    {
-      "epoch": 2.0639317202785943,
-      "grad_norm": 0.004094042349606752,
-      "learning_rate": 0.00019999790130952207,
-      "loss": 46.0,
-      "step": 12816
-    },
-    {
-      "epoch": 2.0640927573573817,
-      "grad_norm": 0.0006622025393880904,
-      "learning_rate": 0.00019999790098174368,
-      "loss": 46.0,
-      "step": 12817
-    },
-    {
-      "epoch": 2.0642537944361687,
-      "grad_norm": 0.0011210956145077944,
-      "learning_rate": 0.00019999790065393968,
-      "loss": 46.0,
-      "step": 12818
-    },
-    {
-      "epoch": 2.064414831514956,
-      "grad_norm": 0.0006348789320327342,
-      "learning_rate": 0.0001999979003261101,
-      "loss": 46.0,
-      "step": 12819
-    },
-    {
-      "epoch": 2.0645758685937436,
-      "grad_norm": 0.0014930064789950848,
-      "learning_rate": 0.00019999789999825492,
-      "loss": 46.0,
-      "step": 12820
-    },
-    {
-      "epoch": 2.064736905672531,
-      "grad_norm": 0.0006207611877471209,
-      "learning_rate": 0.00019999789967037413,
-      "loss": 46.0,
-      "step": 12821
-    },
-    {
-      "epoch": 2.0648979427513185,
-      "grad_norm": 0.0024403254501521587,
-      "learning_rate": 0.00019999789934246778,
-      "loss": 46.0,
-      "step": 12822
-    },
-    {
-      "epoch": 2.065058979830106,
-      "grad_norm": 0.0029341739136725664,
-      "learning_rate": 0.00019999789901453582,
-      "loss": 46.0,
-      "step": 12823
-    },
-    {
-      "epoch": 2.0652200169088935,
-      "grad_norm": 0.006413840688765049,
-      "learning_rate": 0.00019999789868657827,
-      "loss": 46.0,
-      "step": 12824
-    },
-    {
-      "epoch": 2.0653810539876805,
-      "grad_norm": 0.0017412956804037094,
-      "learning_rate": 0.0001999978983585951,
-      "loss": 46.0,
-      "step": 12825
-    },
-    {
-      "epoch": 2.065542091066468,
-      "grad_norm": 0.0032224729657173157,
-      "learning_rate": 0.00019999789803058638,
-      "loss": 46.0,
-      "step": 12826
-    },
-    {
-      "epoch": 2.0657031281452554,
-      "grad_norm": 0.000601353996898979,
-      "learning_rate": 0.000199997897702552,
-      "loss": 46.0,
-      "step": 12827
-    },
-    {
-      "epoch": 2.065864165224043,
-      "grad_norm": 0.0022773239761590958,
-      "learning_rate": 0.0001999978973744921,
-      "loss": 46.0,
-      "step": 12828
-    },
-    {
-      "epoch": 2.0660252023028303,
-      "grad_norm": 0.0016295919194817543,
-      "learning_rate": 0.00019999789704640657,
-      "loss": 46.0,
-      "step": 12829
-    },
-    {
-      "epoch": 2.0661862393816177,
-      "grad_norm": 0.008981181308627129,
-      "learning_rate": 0.00019999789671829542,
-      "loss": 46.0,
-      "step": 12830
-    },
-    {
-      "epoch": 2.066347276460405,
-      "grad_norm": 0.000781839422415942,
-      "learning_rate": 0.0001999978963901587,
-      "loss": 46.0,
-      "step": 12831
-    },
-    {
-      "epoch": 2.0665083135391926,
-      "grad_norm": 0.0004905819660052657,
-      "learning_rate": 0.0001999978960619964,
-      "loss": 46.0,
-      "step": 12832
-    },
-    {
-      "epoch": 2.0666693506179796,
-      "grad_norm": 0.004294846672564745,
-      "learning_rate": 0.00019999789573380849,
-      "loss": 46.0,
-      "step": 12833
-    },
-    {
-      "epoch": 2.066830387696767,
-      "grad_norm": 0.002000445267185569,
-      "learning_rate": 0.00019999789540559498,
-      "loss": 46.0,
-      "step": 12834
-    },
-    {
-      "epoch": 2.0669914247755545,
-      "grad_norm": 0.00406459579244256,
-      "learning_rate": 0.00019999789507735592,
-      "loss": 46.0,
-      "step": 12835
-    },
-    {
-      "epoch": 2.067152461854342,
-      "grad_norm": 0.0018631048733368516,
-      "learning_rate": 0.00019999789474909121,
-      "loss": 46.0,
-      "step": 12836
-    },
-    {
-      "epoch": 2.0673134989331294,
-      "grad_norm": 0.0061733657494187355,
-      "learning_rate": 0.00019999789442080092,
-      "loss": 46.0,
-      "step": 12837
-    },
-    {
-      "epoch": 2.067474536011917,
-      "grad_norm": 0.0037168734706938267,
-      "learning_rate": 0.00019999789409248507,
-      "loss": 46.0,
-      "step": 12838
-    },
-    {
-      "epoch": 2.0676355730907043,
-      "grad_norm": 0.0013421680778265,
-      "learning_rate": 0.00019999789376414358,
-      "loss": 46.0,
-      "step": 12839
-    },
-    {
-      "epoch": 2.0677966101694913,
-      "grad_norm": 0.0020350811537355185,
-      "learning_rate": 0.0001999978934357765,
-      "loss": 46.0,
-      "step": 12840
-    },
-    {
-      "epoch": 2.067957647248279,
-      "grad_norm": 0.0029729909729212523,
-      "learning_rate": 0.00019999789310738385,
-      "loss": 46.0,
-      "step": 12841
-    },
-    {
-      "epoch": 2.0681186843270662,
-      "grad_norm": 0.001266554114408791,
-      "learning_rate": 0.0001999978927789656,
-      "loss": 46.0,
-      "step": 12842
-    },
-    {
-      "epoch": 2.0682797214058537,
-      "grad_norm": 0.0020449126604944468,
-      "learning_rate": 0.00019999789245052173,
-      "loss": 46.0,
-      "step": 12843
-    },
-    {
-      "epoch": 2.068440758484641,
-      "grad_norm": 0.0028160871006548405,
-      "learning_rate": 0.0001999978921220523,
-      "loss": 46.0,
-      "step": 12844
-    },
-    {
-      "epoch": 2.0686017955634286,
-      "grad_norm": 0.0015975756105035543,
-      "learning_rate": 0.00019999789179355726,
-      "loss": 46.0,
-      "step": 12845
-    },
-    {
-      "epoch": 2.068762832642216,
-      "grad_norm": 0.0023738001473248005,
-      "learning_rate": 0.00019999789146503663,
-      "loss": 46.0,
-      "step": 12846
-    },
-    {
-      "epoch": 2.068923869721003,
-      "grad_norm": 0.0012726528802886605,
-      "learning_rate": 0.0001999978911364904,
-      "loss": 46.0,
-      "step": 12847
-    },
-    {
-      "epoch": 2.0690849067997905,
-      "grad_norm": 0.0018322101095691323,
-      "learning_rate": 0.00019999789080791858,
-      "loss": 46.0,
-      "step": 12848
-    },
-    {
-      "epoch": 2.069245943878578,
-      "grad_norm": 0.0034243674017488956,
-      "learning_rate": 0.00019999789047932116,
-      "loss": 46.0,
-      "step": 12849
-    },
-    {
-      "epoch": 2.0694069809573654,
-      "grad_norm": 0.0015597471501678228,
-      "learning_rate": 0.00019999789015069818,
-      "loss": 46.0,
-      "step": 12850
-    },
-    {
-      "epoch": 2.069568018036153,
-      "grad_norm": 0.002911241492256522,
-      "learning_rate": 0.00019999788982204956,
-      "loss": 46.0,
-      "step": 12851
-    },
-    {
-      "epoch": 2.0697290551149403,
-      "grad_norm": 0.0012249392457306385,
-      "learning_rate": 0.00019999788949337535,
-      "loss": 46.0,
-      "step": 12852
-    },
-    {
-      "epoch": 2.0698900921937278,
-      "grad_norm": 0.0021083063911646605,
-      "learning_rate": 0.00019999788916467558,
-      "loss": 46.0,
-      "step": 12853
-    },
-    {
-      "epoch": 2.070051129272515,
-      "grad_norm": 0.004616187885403633,
-      "learning_rate": 0.00019999788883595017,
-      "loss": 46.0,
-      "step": 12854
-    },
-    {
-      "epoch": 2.0702121663513022,
-      "grad_norm": 0.0022661376278847456,
-      "learning_rate": 0.00019999788850719917,
-      "loss": 46.0,
-      "step": 12855
-    },
-    {
-      "epoch": 2.0703732034300897,
-      "grad_norm": 0.007422221824526787,
-      "learning_rate": 0.00019999788817842262,
-      "loss": 46.0,
-      "step": 12856
-    },
-    {
-      "epoch": 2.070534240508877,
-      "grad_norm": 0.005455732811242342,
-      "learning_rate": 0.00019999788784962045,
-      "loss": 46.0,
-      "step": 12857
-    },
-    {
-      "epoch": 2.0706952775876646,
-      "grad_norm": 0.004719725344330072,
-      "learning_rate": 0.00019999788752079266,
-      "loss": 46.0,
-      "step": 12858
-    },
-    {
-      "epoch": 2.070856314666452,
-      "grad_norm": 0.0017437150236219168,
-      "learning_rate": 0.00019999788719193932,
-      "loss": 46.0,
-      "step": 12859
-    },
-    {
-      "epoch": 2.0710173517452395,
-      "grad_norm": 0.002943633357062936,
-      "learning_rate": 0.00019999788686306038,
-      "loss": 46.0,
-      "step": 12860
-    },
-    {
-      "epoch": 2.071178388824027,
-      "grad_norm": 0.007313722278922796,
-      "learning_rate": 0.00019999788653415584,
-      "loss": 46.0,
-      "step": 12861
-    },
-    {
-      "epoch": 2.071339425902814,
-      "grad_norm": 0.00593662541359663,
-      "learning_rate": 0.00019999788620522567,
-      "loss": 46.0,
-      "step": 12862
-    },
-    {
-      "epoch": 2.0715004629816014,
-      "grad_norm": 0.003066117875277996,
-      "learning_rate": 0.00019999788587626995,
-      "loss": 46.0,
-      "step": 12863
-    },
-    {
-      "epoch": 2.071661500060389,
-      "grad_norm": 0.003927651327103376,
-      "learning_rate": 0.00019999788554728862,
-      "loss": 46.0,
-      "step": 12864
-    },
-    {
-      "epoch": 2.0718225371391763,
-      "grad_norm": 0.0005517390090972185,
-      "learning_rate": 0.0001999978852182817,
-      "loss": 46.0,
-      "step": 12865
-    },
-    {
-      "epoch": 2.0719835742179638,
-      "grad_norm": 0.0039609442465007305,
-      "learning_rate": 0.00019999788488924919,
-      "loss": 46.0,
-      "step": 12866
-    },
-    {
-      "epoch": 2.072144611296751,
-      "grad_norm": 0.0011377392802387476,
-      "learning_rate": 0.00019999788456019106,
-      "loss": 46.0,
-      "step": 12867
-    },
-    {
-      "epoch": 2.0723056483755387,
-      "grad_norm": 0.0016544148093089461,
-      "learning_rate": 0.00019999788423110735,
-      "loss": 46.0,
-      "step": 12868
-    },
-    {
-      "epoch": 2.0724666854543257,
-      "grad_norm": 0.001906959223560989,
-      "learning_rate": 0.00019999788390199805,
-      "loss": 46.0,
-      "step": 12869
-    },
-    {
-      "epoch": 2.072627722533113,
-      "grad_norm": 0.0015103211626410484,
-      "learning_rate": 0.00019999788357286317,
-      "loss": 46.0,
-      "step": 12870
-    },
-    {
-      "epoch": 2.0727887596119006,
-      "grad_norm": 0.0020772633142769337,
-      "learning_rate": 0.00019999788324370267,
-      "loss": 46.0,
-      "step": 12871
-    },
-    {
-      "epoch": 2.072949796690688,
-      "grad_norm": 0.002363672247156501,
-      "learning_rate": 0.00019999788291451658,
-      "loss": 46.0,
-      "step": 12872
-    },
-    {
-      "epoch": 2.0731108337694755,
-      "grad_norm": 0.00032840410131029785,
-      "learning_rate": 0.0001999978825853049,
-      "loss": 46.0,
-      "step": 12873
-    },
-    {
-      "epoch": 2.073271870848263,
-      "grad_norm": 0.0035052697639912367,
-      "learning_rate": 0.00019999788225606762,
-      "loss": 46.0,
-      "step": 12874
-    },
-    {
-      "epoch": 2.0734329079270504,
-      "grad_norm": 0.003118172287940979,
-      "learning_rate": 0.00019999788192680474,
-      "loss": 46.0,
-      "step": 12875
-    },
-    {
-      "epoch": 2.0735939450058374,
-      "grad_norm": 0.000833704776596278,
-      "learning_rate": 0.00019999788159751628,
-      "loss": 46.0,
-      "step": 12876
-    },
-    {
-      "epoch": 2.073754982084625,
-      "grad_norm": 0.0006109802052378654,
-      "learning_rate": 0.00019999788126820223,
-      "loss": 46.0,
-      "step": 12877
-    },
-    {
-      "epoch": 2.0739160191634123,
-      "grad_norm": 0.010422511957585812,
-      "learning_rate": 0.0001999978809388626,
-      "loss": 46.0,
-      "step": 12878
-    },
-    {
-      "epoch": 2.0740770562421997,
-      "grad_norm": 0.0008158034179359674,
-      "learning_rate": 0.00019999788060949734,
-      "loss": 46.0,
-      "step": 12879
-    },
-    {
-      "epoch": 2.074238093320987,
-      "grad_norm": 0.0028415508568286896,
-      "learning_rate": 0.00019999788028010647,
-      "loss": 46.0,
-      "step": 12880
-    },
-    {
-      "epoch": 2.0743991303997746,
-      "grad_norm": 0.0031324869487434626,
-      "learning_rate": 0.00019999787995069005,
-      "loss": 46.0,
-      "step": 12881
-    },
-    {
-      "epoch": 2.074560167478562,
-      "grad_norm": 0.0007474896847270429,
-      "learning_rate": 0.00019999787962124804,
-      "loss": 46.0,
-      "step": 12882
-    },
-    {
-      "epoch": 2.0747212045573495,
-      "grad_norm": 0.006062533240765333,
-      "learning_rate": 0.0001999978792917804,
-      "loss": 46.0,
-      "step": 12883
-    },
-    {
-      "epoch": 2.0748822416361365,
-      "grad_norm": 0.004317732993513346,
-      "learning_rate": 0.00019999787896228722,
-      "loss": 46.0,
-      "step": 12884
-    },
-    {
-      "epoch": 2.075043278714924,
-      "grad_norm": 0.004423650912940502,
-      "learning_rate": 0.0001999978786327684,
-      "loss": 46.0,
-      "step": 12885
-    },
-    {
-      "epoch": 2.0752043157937115,
-      "grad_norm": 0.0064974562264978886,
-      "learning_rate": 0.00019999787830322395,
-      "loss": 46.0,
-      "step": 12886
-    },
-    {
-      "epoch": 2.075365352872499,
-      "grad_norm": 0.0017117992974817753,
-      "learning_rate": 0.000199997877973654,
-      "loss": 46.0,
-      "step": 12887
-    },
-    {
-      "epoch": 2.0755263899512864,
-      "grad_norm": 0.004673875868320465,
-      "learning_rate": 0.00019999787764405838,
-      "loss": 46.0,
-      "step": 12888
-    },
-    {
-      "epoch": 2.075687427030074,
-      "grad_norm": 0.010830063372850418,
-      "learning_rate": 0.0001999978773144372,
-      "loss": 46.0,
-      "step": 12889
-    },
-    {
-      "epoch": 2.0758484641088613,
-      "grad_norm": 0.00277833198197186,
-      "learning_rate": 0.00019999787698479041,
-      "loss": 46.0,
-      "step": 12890
-    },
-    {
-      "epoch": 2.0760095011876483,
-      "grad_norm": 0.002786587690934539,
-      "learning_rate": 0.00019999787665511804,
-      "loss": 46.0,
-      "step": 12891
-    },
-    {
-      "epoch": 2.0761705382664357,
-      "grad_norm": 0.0015934628900140524,
-      "learning_rate": 0.00019999787632542007,
-      "loss": 46.0,
-      "step": 12892
-    },
-    {
-      "epoch": 2.076331575345223,
-      "grad_norm": 0.005051181651651859,
-      "learning_rate": 0.00019999787599569652,
-      "loss": 46.0,
-      "step": 12893
-    },
-    {
-      "epoch": 2.0764926124240106,
-      "grad_norm": 0.0005999958957545459,
-      "learning_rate": 0.00019999787566594735,
-      "loss": 46.0,
-      "step": 12894
-    },
-    {
-      "epoch": 2.076653649502798,
-      "grad_norm": 0.0016861752374097705,
-      "learning_rate": 0.00019999787533617257,
-      "loss": 46.0,
-      "step": 12895
-    },
-    {
-      "epoch": 2.0768146865815855,
-      "grad_norm": 0.0026603303849697113,
-      "learning_rate": 0.00019999787500637223,
-      "loss": 46.0,
-      "step": 12896
-    },
-    {
-      "epoch": 2.076975723660373,
-      "grad_norm": 0.005328537430614233,
-      "learning_rate": 0.0001999978746765463,
-      "loss": 46.0,
-      "step": 12897
-    },
-    {
-      "epoch": 2.07713676073916,
-      "grad_norm": 0.012002729810774326,
-      "learning_rate": 0.00019999787434669475,
-      "loss": 46.0,
-      "step": 12898
-    },
-    {
-      "epoch": 2.0772977978179474,
-      "grad_norm": 0.012205596081912518,
-      "learning_rate": 0.00019999787401681762,
-      "loss": 46.0,
-      "step": 12899
-    },
-    {
-      "epoch": 2.077458834896735,
-      "grad_norm": 0.00956760160624981,
-      "learning_rate": 0.0001999978736869149,
-      "loss": 46.0,
-      "step": 12900
-    },
-    {
-      "epoch": 2.0776198719755223,
-      "grad_norm": 0.003218648489564657,
-      "learning_rate": 0.00019999787335698657,
-      "loss": 46.0,
-      "step": 12901
-    },
-    {
-      "epoch": 2.07778090905431,
-      "grad_norm": 0.0009184836526401341,
-      "learning_rate": 0.00019999787302703265,
-      "loss": 46.0,
-      "step": 12902
-    },
-    {
-      "epoch": 2.0779419461330972,
-      "grad_norm": 0.001078438712283969,
-      "learning_rate": 0.00019999787269705315,
-      "loss": 46.0,
-      "step": 12903
-    },
-    {
-      "epoch": 2.0781029832118847,
-      "grad_norm": 0.0052854218520224094,
-      "learning_rate": 0.00019999787236704805,
-      "loss": 46.0,
-      "step": 12904
-    },
-    {
-      "epoch": 2.078264020290672,
-      "grad_norm": 0.0005770517163909972,
-      "learning_rate": 0.00019999787203701735,
-      "loss": 46.0,
-      "step": 12905
-    },
-    {
-      "epoch": 2.078425057369459,
-      "grad_norm": 0.0018130955286324024,
-      "learning_rate": 0.00019999787170696105,
-      "loss": 46.0,
-      "step": 12906
-    },
-    {
-      "epoch": 2.0785860944482466,
-      "grad_norm": 0.0015770385507494211,
-      "learning_rate": 0.00019999787137687917,
-      "loss": 46.0,
-      "step": 12907
-    },
-    {
-      "epoch": 2.078747131527034,
-      "grad_norm": 0.007847209461033344,
-      "learning_rate": 0.00019999787104677167,
-      "loss": 46.0,
-      "step": 12908
-    },
-    {
-      "epoch": 2.0789081686058215,
-      "grad_norm": 0.008562360890209675,
-      "learning_rate": 0.0001999978707166386,
-      "loss": 46.0,
-      "step": 12909
-    },
-    {
-      "epoch": 2.079069205684609,
-      "grad_norm": 0.0010221872944384813,
-      "learning_rate": 0.00019999787038647995,
-      "loss": 46.0,
-      "step": 12910
-    },
-    {
-      "epoch": 2.0792302427633964,
-      "grad_norm": 0.000811954028904438,
-      "learning_rate": 0.00019999787005629566,
-      "loss": 46.0,
-      "step": 12911
-    },
-    {
-      "epoch": 2.079391279842184,
-      "grad_norm": 0.0024959570728242397,
-      "learning_rate": 0.00019999786972608582,
-      "loss": 46.0,
-      "step": 12912
-    },
-    {
-      "epoch": 2.079552316920971,
-      "grad_norm": 0.0021960786543786526,
-      "learning_rate": 0.00019999786939585036,
-      "loss": 46.0,
-      "step": 12913
-    },
-    {
-      "epoch": 2.0797133539997583,
-      "grad_norm": 0.004885167349129915,
-      "learning_rate": 0.0001999978690655893,
-      "loss": 46.0,
-      "step": 12914
-    },
-    {
-      "epoch": 2.0798743910785458,
-      "grad_norm": 0.004590335302054882,
-      "learning_rate": 0.00019999786873530265,
-      "loss": 46.0,
-      "step": 12915
-    },
-    {
-      "epoch": 2.080035428157333,
-      "grad_norm": 0.001211219234392047,
-      "learning_rate": 0.00019999786840499043,
-      "loss": 46.0,
-      "step": 12916
-    },
-    {
-      "epoch": 2.0801964652361207,
-      "grad_norm": 0.00057620630832389,
-      "learning_rate": 0.0001999978680746526,
-      "loss": 46.0,
-      "step": 12917
-    },
-    {
-      "epoch": 2.080357502314908,
-      "grad_norm": 0.0022125651594251394,
-      "learning_rate": 0.00019999786774428917,
-      "loss": 46.0,
-      "step": 12918
-    },
-    {
-      "epoch": 2.0805185393936956,
-      "grad_norm": 0.006694595795124769,
-      "learning_rate": 0.00019999786741390016,
-      "loss": 46.0,
-      "step": 12919
-    },
-    {
-      "epoch": 2.0806795764724826,
-      "grad_norm": 0.0012308426667004824,
-      "learning_rate": 0.00019999786708348557,
-      "loss": 46.0,
-      "step": 12920
-    },
-    {
-      "epoch": 2.08084061355127,
-      "grad_norm": 0.0026756240986287594,
-      "learning_rate": 0.00019999786675304536,
-      "loss": 46.0,
-      "step": 12921
-    },
-    {
-      "epoch": 2.0810016506300575,
-      "grad_norm": 0.003326231613755226,
-      "learning_rate": 0.00019999786642257953,
-      "loss": 46.0,
-      "step": 12922
-    },
-    {
-      "epoch": 2.081162687708845,
-      "grad_norm": 0.001794333104044199,
-      "learning_rate": 0.00019999786609208815,
-      "loss": 46.0,
-      "step": 12923
-    },
-    {
-      "epoch": 2.0813237247876324,
-      "grad_norm": 0.0016420054016634822,
-      "learning_rate": 0.00019999786576157115,
-      "loss": 46.0,
-      "step": 12924
-    },
-    {
-      "epoch": 2.08148476186642,
-      "grad_norm": 0.0035400076303631067,
-      "learning_rate": 0.00019999786543102856,
-      "loss": 46.0,
-      "step": 12925
-    },
-    {
-      "epoch": 2.0816457989452073,
-      "grad_norm": 0.006951572373509407,
-      "learning_rate": 0.0001999978651004604,
-      "loss": 46.0,
-      "step": 12926
-    },
-    {
-      "epoch": 2.0818068360239943,
-      "grad_norm": 0.005445762071758509,
-      "learning_rate": 0.00019999786476986663,
-      "loss": 46.0,
-      "step": 12927
-    },
-    {
-      "epoch": 2.0819678731027818,
-      "grad_norm": 0.005064631812274456,
-      "learning_rate": 0.00019999786443924725,
-      "loss": 46.0,
-      "step": 12928
-    },
-    {
-      "epoch": 2.082128910181569,
-      "grad_norm": 0.003887486644089222,
-      "learning_rate": 0.00019999786410860232,
-      "loss": 46.0,
-      "step": 12929
-    },
-    {
-      "epoch": 2.0822899472603567,
-      "grad_norm": 0.0003751896438188851,
-      "learning_rate": 0.00019999786377793174,
-      "loss": 46.0,
-      "step": 12930
-    },
-    {
-      "epoch": 2.082450984339144,
-      "grad_norm": 0.001582261174917221,
-      "learning_rate": 0.0001999978634472356,
-      "loss": 46.0,
-      "step": 12931
-    },
-    {
-      "epoch": 2.0826120214179316,
-      "grad_norm": 0.0010145421838387847,
-      "learning_rate": 0.00019999786311651385,
-      "loss": 46.0,
-      "step": 12932
-    },
-    {
-      "epoch": 2.082773058496719,
-      "grad_norm": 0.007230785675346851,
-      "learning_rate": 0.0001999978627857665,
-      "loss": 46.0,
-      "step": 12933
-    },
-    {
-      "epoch": 2.0829340955755065,
-      "grad_norm": 0.004808077588677406,
-      "learning_rate": 0.00019999786245499356,
-      "loss": 46.0,
-      "step": 12934
-    },
-    {
-      "epoch": 2.0830951326542935,
-      "grad_norm": 0.001964322756975889,
-      "learning_rate": 0.00019999786212419505,
-      "loss": 46.0,
-      "step": 12935
-    },
-    {
-      "epoch": 2.083256169733081,
-      "grad_norm": 0.005786800291389227,
-      "learning_rate": 0.00019999786179337092,
-      "loss": 46.0,
-      "step": 12936
-    },
-    {
-      "epoch": 2.0834172068118684,
-      "grad_norm": 0.006222918163985014,
-      "learning_rate": 0.0001999978614625212,
-      "loss": 46.0,
-      "step": 12937
-    },
-    {
-      "epoch": 2.083578243890656,
-      "grad_norm": 0.003976618405431509,
-      "learning_rate": 0.00019999786113164588,
-      "loss": 46.0,
-      "step": 12938
-    },
-    {
-      "epoch": 2.0837392809694433,
-      "grad_norm": 0.009521611966192722,
-      "learning_rate": 0.000199997860800745,
-      "loss": 46.0,
-      "step": 12939
-    },
-    {
-      "epoch": 2.0839003180482307,
-      "grad_norm": 0.005143940448760986,
-      "learning_rate": 0.0001999978604698185,
-      "loss": 46.0,
-      "step": 12940
-    },
-    {
-      "epoch": 2.084061355127018,
-      "grad_norm": 0.0019475278677418828,
-      "learning_rate": 0.0001999978601388664,
-      "loss": 46.0,
-      "step": 12941
-    },
-    {
-      "epoch": 2.084222392205805,
-      "grad_norm": 0.001794849755242467,
-      "learning_rate": 0.0001999978598078887,
-      "loss": 46.0,
-      "step": 12942
-    },
-    {
-      "epoch": 2.0843834292845926,
-      "grad_norm": 0.017196884378790855,
-      "learning_rate": 0.00019999785947688543,
-      "loss": 46.0,
-      "step": 12943
-    },
-    {
-      "epoch": 2.08454446636338,
-      "grad_norm": 0.0007669737678952515,
-      "learning_rate": 0.00019999785914585653,
-      "loss": 46.0,
-      "step": 12944
-    },
-    {
-      "epoch": 2.0847055034421675,
-      "grad_norm": 0.0021896001417189837,
-      "learning_rate": 0.00019999785881480206,
-      "loss": 46.0,
-      "step": 12945
-    },
-    {
-      "epoch": 2.084866540520955,
-      "grad_norm": 0.010119922459125519,
-      "learning_rate": 0.000199997858483722,
-      "loss": 46.0,
-      "step": 12946
-    },
-    {
-      "epoch": 2.0850275775997424,
-      "grad_norm": 0.004330337047576904,
-      "learning_rate": 0.00019999785815261634,
-      "loss": 46.0,
-      "step": 12947
-    },
-    {
-      "epoch": 2.08518861467853,
-      "grad_norm": 0.0010863668285310268,
-      "learning_rate": 0.00019999785782148509,
-      "loss": 46.0,
-      "step": 12948
-    },
-    {
-      "epoch": 2.0853496517573173,
-      "grad_norm": 0.0032476340420544147,
-      "learning_rate": 0.00019999785749032824,
-      "loss": 46.0,
-      "step": 12949
-    },
-    {
-      "epoch": 2.0855106888361044,
-      "grad_norm": 0.0007361643365584314,
-      "learning_rate": 0.00019999785715914582,
-      "loss": 46.0,
-      "step": 12950
-    },
-    {
-      "epoch": 2.085671725914892,
-      "grad_norm": 0.01038648933172226,
-      "learning_rate": 0.00019999785682793777,
-      "loss": 46.0,
-      "step": 12951
-    },
-    {
-      "epoch": 2.0858327629936793,
-      "grad_norm": 0.0009936820715665817,
-      "learning_rate": 0.00019999785649670412,
-      "loss": 46.0,
-      "step": 12952
-    },
-    {
-      "epoch": 2.0859938000724667,
-      "grad_norm": 0.0013045460218563676,
-      "learning_rate": 0.0001999978561654449,
-      "loss": 46.0,
-      "step": 12953
-    },
-    {
-      "epoch": 2.086154837151254,
-      "grad_norm": 0.002221314935013652,
-      "learning_rate": 0.0001999978558341601,
-      "loss": 46.0,
-      "step": 12954
-    },
-    {
-      "epoch": 2.0863158742300416,
-      "grad_norm": 0.0003671857703011483,
-      "learning_rate": 0.00019999785550284965,
-      "loss": 46.0,
-      "step": 12955
-    },
-    {
-      "epoch": 2.086476911308829,
-      "grad_norm": 0.0013719316339120269,
-      "learning_rate": 0.00019999785517151367,
-      "loss": 46.0,
-      "step": 12956
-    },
-    {
-      "epoch": 2.086637948387616,
-      "grad_norm": 0.008065132424235344,
-      "learning_rate": 0.00019999785484015208,
-      "loss": 46.0,
-      "step": 12957
-    },
-    {
-      "epoch": 2.0867989854664035,
-      "grad_norm": 0.006490033119916916,
-      "learning_rate": 0.00019999785450876487,
-      "loss": 46.0,
-      "step": 12958
-    },
-    {
-      "epoch": 2.086960022545191,
-      "grad_norm": 0.0018406733870506287,
-      "learning_rate": 0.00019999785417735208,
-      "loss": 46.0,
-      "step": 12959
-    },
-    {
-      "epoch": 2.0871210596239784,
-      "grad_norm": 0.001462517655454576,
-      "learning_rate": 0.0001999978538459137,
-      "loss": 46.0,
-      "step": 12960
-    },
-    {
-      "epoch": 2.087282096702766,
-      "grad_norm": 0.0026362589560449123,
-      "learning_rate": 0.0001999978535144497,
-      "loss": 46.0,
-      "step": 12961
-    },
-    {
-      "epoch": 2.0874431337815533,
-      "grad_norm": 0.0017384827369824052,
-      "learning_rate": 0.00019999785318296014,
-      "loss": 46.0,
-      "step": 12962
-    },
-    {
-      "epoch": 2.087604170860341,
-      "grad_norm": 0.0013901966158300638,
-      "learning_rate": 0.00019999785285144497,
-      "loss": 46.0,
-      "step": 12963
-    },
-    {
-      "epoch": 2.087765207939128,
-      "grad_norm": 0.0026942843105643988,
-      "learning_rate": 0.00019999785251990422,
-      "loss": 46.0,
-      "step": 12964
-    },
-    {
-      "epoch": 2.0879262450179152,
-      "grad_norm": 0.002685442566871643,
-      "learning_rate": 0.00019999785218833784,
-      "loss": 46.0,
-      "step": 12965
-    },
-    {
-      "epoch": 2.0880872820967027,
-      "grad_norm": 0.004451130982488394,
-      "learning_rate": 0.00019999785185674589,
-      "loss": 46.0,
-      "step": 12966
-    },
-    {
-      "epoch": 2.08824831917549,
-      "grad_norm": 0.0017370268469676375,
-      "learning_rate": 0.00019999785152512837,
-      "loss": 46.0,
-      "step": 12967
-    },
-    {
-      "epoch": 2.0884093562542776,
-      "grad_norm": 0.0009535567369312048,
-      "learning_rate": 0.0001999978511934852,
-      "loss": 46.0,
-      "step": 12968
-    },
-    {
-      "epoch": 2.088570393333065,
-      "grad_norm": 0.00570228835567832,
-      "learning_rate": 0.00019999785086181646,
-      "loss": 46.0,
-      "step": 12969
-    },
-    {
-      "epoch": 2.0887314304118525,
-      "grad_norm": 0.0011478904634714127,
-      "learning_rate": 0.00019999785053012215,
-      "loss": 46.0,
-      "step": 12970
-    },
-    {
-      "epoch": 2.0888924674906395,
-      "grad_norm": 0.00048434018390253186,
-      "learning_rate": 0.00019999785019840223,
-      "loss": 46.0,
-      "step": 12971
-    },
-    {
-      "epoch": 2.089053504569427,
-      "grad_norm": 0.011068599298596382,
-      "learning_rate": 0.0001999978498666567,
-      "loss": 46.0,
-      "step": 12972
-    },
-    {
-      "epoch": 2.0892145416482144,
-      "grad_norm": 0.0014579817652702332,
-      "learning_rate": 0.0001999978495348856,
-      "loss": 46.0,
-      "step": 12973
-    },
-    {
-      "epoch": 2.089375578727002,
-      "grad_norm": 0.0009975058492273092,
-      "learning_rate": 0.0001999978492030889,
-      "loss": 46.0,
-      "step": 12974
-    },
-    {
-      "epoch": 2.0895366158057893,
-      "grad_norm": 0.0036850841715931892,
-      "learning_rate": 0.0001999978488712666,
-      "loss": 46.0,
-      "step": 12975
-    },
-    {
-      "epoch": 2.0896976528845768,
-      "grad_norm": 0.005624666344374418,
-      "learning_rate": 0.00019999784853941868,
-      "loss": 46.0,
-      "step": 12976
-    },
-    {
-      "epoch": 2.089858689963364,
-      "grad_norm": 0.0011068080784752965,
-      "learning_rate": 0.0001999978482075452,
-      "loss": 46.0,
-      "step": 12977
-    },
-    {
-      "epoch": 2.0900197270421517,
-      "grad_norm": 0.0018457322148606181,
-      "learning_rate": 0.00019999784787564612,
-      "loss": 46.0,
-      "step": 12978
-    },
-    {
-      "epoch": 2.0901807641209387,
-      "grad_norm": 0.003644770011305809,
-      "learning_rate": 0.00019999784754372145,
-      "loss": 46.0,
-      "step": 12979
-    },
-    {
-      "epoch": 2.090341801199726,
-      "grad_norm": 0.0008228995138779283,
-      "learning_rate": 0.00019999784721177116,
-      "loss": 46.0,
-      "step": 12980
-    },
-    {
-      "epoch": 2.0905028382785136,
-      "grad_norm": 0.00238042208366096,
-      "learning_rate": 0.0001999978468797953,
-      "loss": 46.0,
-      "step": 12981
-    },
-    {
-      "epoch": 2.090663875357301,
-      "grad_norm": 0.002125946804881096,
-      "learning_rate": 0.00019999784654779385,
-      "loss": 46.0,
-      "step": 12982
-    },
-    {
-      "epoch": 2.0908249124360885,
-      "grad_norm": 0.0018968727672472596,
-      "learning_rate": 0.00019999784621576677,
-      "loss": 46.0,
-      "step": 12983
-    },
-    {
-      "epoch": 2.090985949514876,
-      "grad_norm": 0.001954280072823167,
-      "learning_rate": 0.00019999784588371414,
-      "loss": 46.0,
-      "step": 12984
-    },
-    {
-      "epoch": 2.0911469865936634,
-      "grad_norm": 0.0034703118726611137,
-      "learning_rate": 0.0001999978455516359,
-      "loss": 46.0,
-      "step": 12985
-    },
-    {
-      "epoch": 2.0913080236724504,
-      "grad_norm": 0.0006526318611577153,
-      "learning_rate": 0.00019999784521953205,
-      "loss": 46.0,
-      "step": 12986
-    },
-    {
-      "epoch": 2.091469060751238,
-      "grad_norm": 0.0031117235776036978,
-      "learning_rate": 0.0001999978448874026,
-      "loss": 46.0,
-      "step": 12987
-    },
-    {
-      "epoch": 2.0916300978300253,
-      "grad_norm": 0.006501418072730303,
-      "learning_rate": 0.00019999784455524758,
-      "loss": 46.0,
-      "step": 12988
-    },
-    {
-      "epoch": 2.0917911349088127,
-      "grad_norm": 0.004377106670290232,
-      "learning_rate": 0.00019999784422306699,
-      "loss": 46.0,
-      "step": 12989
-    },
-    {
-      "epoch": 2.0919521719876,
-      "grad_norm": 0.0044606709852814674,
-      "learning_rate": 0.00019999784389086077,
-      "loss": 46.0,
-      "step": 12990
-    },
-    {
-      "epoch": 2.0921132090663876,
-      "grad_norm": 0.003383436007425189,
-      "learning_rate": 0.00019999784355862894,
-      "loss": 46.0,
-      "step": 12991
-    },
-    {
-      "epoch": 2.092274246145175,
-      "grad_norm": 0.0005474574281834066,
-      "learning_rate": 0.00019999784322637156,
-      "loss": 46.0,
-      "step": 12992
-    },
-    {
-      "epoch": 2.092435283223962,
-      "grad_norm": 0.00288202497176826,
-      "learning_rate": 0.00019999784289408855,
-      "loss": 46.0,
-      "step": 12993
-    },
-    {
-      "epoch": 2.0925963203027496,
-      "grad_norm": 0.004795857239514589,
-      "learning_rate": 0.00019999784256177997,
-      "loss": 46.0,
-      "step": 12994
-    },
-    {
-      "epoch": 2.092757357381537,
-      "grad_norm": 0.0028374146204441786,
-      "learning_rate": 0.0001999978422294458,
-      "loss": 46.0,
-      "step": 12995
-    },
-    {
-      "epoch": 2.0929183944603245,
-      "grad_norm": 0.0005240382743068039,
-      "learning_rate": 0.000199997841897086,
-      "loss": 46.0,
-      "step": 12996
-    },
-    {
-      "epoch": 2.093079431539112,
-      "grad_norm": 0.00736638531088829,
-      "learning_rate": 0.0001999978415647006,
-      "loss": 46.0,
-      "step": 12997
-    },
-    {
-      "epoch": 2.0932404686178994,
-      "grad_norm": 0.0016489975387230515,
-      "learning_rate": 0.00019999784123228963,
-      "loss": 46.0,
-      "step": 12998
-    },
-    {
-      "epoch": 2.093401505696687,
-      "grad_norm": 0.0005634868866764009,
-      "learning_rate": 0.00019999784089985308,
-      "loss": 46.0,
-      "step": 12999
-    },
-    {
-      "epoch": 2.0935625427754743,
-      "grad_norm": 0.0010365262860432267,
-      "learning_rate": 0.00019999784056739094,
-      "loss": 46.0,
-      "step": 13000
-    },
-    {
-      "epoch": 2.0937235798542613,
-      "grad_norm": 0.0011834435863420367,
-      "learning_rate": 0.00019999784023490316,
-      "loss": 46.0,
-      "step": 13001
-    },
-    {
-      "epoch": 2.0938846169330487,
-      "grad_norm": 0.001617851434275508,
-      "learning_rate": 0.00019999783990238981,
-      "loss": 46.0,
-      "step": 13002
-    },
-    {
-      "epoch": 2.094045654011836,
-      "grad_norm": 0.0016286453464999795,
-      "learning_rate": 0.0001999978395698509,
-      "loss": 46.0,
-      "step": 13003
-    },
-    {
-      "epoch": 2.0942066910906236,
-      "grad_norm": 0.005443901289254427,
-      "learning_rate": 0.00019999783923728634,
-      "loss": 46.0,
-      "step": 13004
-    },
-    {
-      "epoch": 2.094367728169411,
-      "grad_norm": 0.0020517813973128796,
-      "learning_rate": 0.00019999783890469621,
-      "loss": 46.0,
-      "step": 13005
-    },
-    {
-      "epoch": 2.0945287652481985,
-      "grad_norm": 0.0007304801838472486,
-      "learning_rate": 0.00019999783857208047,
-      "loss": 46.0,
-      "step": 13006
-    },
-    {
-      "epoch": 2.094689802326986,
-      "grad_norm": 0.002236368600279093,
-      "learning_rate": 0.0001999978382394392,
-      "loss": 46.0,
-      "step": 13007
-    },
-    {
-      "epoch": 2.094850839405773,
-      "grad_norm": 0.002079067286103964,
-      "learning_rate": 0.00019999783790677228,
-      "loss": 46.0,
-      "step": 13008
-    },
-    {
-      "epoch": 2.0950118764845604,
-      "grad_norm": 0.01510265190154314,
-      "learning_rate": 0.00019999783757407974,
-      "loss": 46.0,
-      "step": 13009
-    },
-    {
-      "epoch": 2.095172913563348,
-      "grad_norm": 0.004322418011724949,
-      "learning_rate": 0.00019999783724136165,
-      "loss": 46.0,
-      "step": 13010
-    },
-    {
-      "epoch": 2.0953339506421353,
-      "grad_norm": 0.0032783453352749348,
-      "learning_rate": 0.00019999783690861794,
-      "loss": 46.0,
-      "step": 13011
-    },
-    {
-      "epoch": 2.095494987720923,
-      "grad_norm": 0.003865213831886649,
-      "learning_rate": 0.00019999783657584868,
-      "loss": 46.0,
-      "step": 13012
-    },
-    {
-      "epoch": 2.0956560247997102,
-      "grad_norm": 0.0034887182991951704,
-      "learning_rate": 0.00019999783624305377,
-      "loss": 46.0,
-      "step": 13013
-    },
-    {
-      "epoch": 2.0958170618784977,
-      "grad_norm": 0.016281094402074814,
-      "learning_rate": 0.0001999978359102333,
-      "loss": 46.0,
-      "step": 13014
-    },
-    {
-      "epoch": 2.0959780989572847,
-      "grad_norm": 0.0019259456312283874,
-      "learning_rate": 0.00019999783557738722,
-      "loss": 46.0,
-      "step": 13015
-    },
-    {
-      "epoch": 2.096139136036072,
-      "grad_norm": 0.004159122239798307,
-      "learning_rate": 0.00019999783524451555,
-      "loss": 46.0,
-      "step": 13016
-    },
-    {
-      "epoch": 2.0963001731148596,
-      "grad_norm": 0.001633423031307757,
-      "learning_rate": 0.0001999978349116183,
-      "loss": 46.0,
-      "step": 13017
-    },
-    {
-      "epoch": 2.096461210193647,
-      "grad_norm": 0.005753161385655403,
-      "learning_rate": 0.00019999783457869542,
-      "loss": 46.0,
-      "step": 13018
-    },
-    {
-      "epoch": 2.0966222472724345,
-      "grad_norm": 0.025512289255857468,
-      "learning_rate": 0.000199997834245747,
-      "loss": 46.0,
-      "step": 13019
-    },
-    {
-      "epoch": 2.096783284351222,
-      "grad_norm": 0.003740867367014289,
-      "learning_rate": 0.00019999783391277294,
-      "loss": 46.0,
-      "step": 13020
-    },
-    {
-      "epoch": 2.0969443214300094,
-      "grad_norm": 0.005378149449825287,
-      "learning_rate": 0.00019999783357977328,
-      "loss": 46.0,
-      "step": 13021
-    },
-    {
-      "epoch": 2.097105358508797,
-      "grad_norm": 0.0030334487091749907,
-      "learning_rate": 0.0001999978332467481,
-      "loss": 46.0,
-      "step": 13022
-    },
-    {
-      "epoch": 2.097266395587584,
-      "grad_norm": 0.00285663572140038,
-      "learning_rate": 0.00019999783291369726,
-      "loss": 46.0,
-      "step": 13023
-    },
-    {
-      "epoch": 2.0974274326663713,
-      "grad_norm": 0.004079996608197689,
-      "learning_rate": 0.0001999978325806208,
-      "loss": 46.0,
-      "step": 13024
-    },
-    {
-      "epoch": 2.097588469745159,
-      "grad_norm": 0.0036319908685982227,
-      "learning_rate": 0.0001999978322475188,
-      "loss": 46.0,
-      "step": 13025
-    },
-    {
-      "epoch": 2.0977495068239462,
-      "grad_norm": 0.002361035207286477,
-      "learning_rate": 0.00019999783191439118,
-      "loss": 46.0,
-      "step": 13026
-    },
-    {
-      "epoch": 2.0979105439027337,
-      "grad_norm": 0.003262534271925688,
-      "learning_rate": 0.000199997831581238,
-      "loss": 46.0,
-      "step": 13027
-    },
-    {
-      "epoch": 2.098071580981521,
-      "grad_norm": 0.0017374531598761678,
-      "learning_rate": 0.00019999783124805917,
-      "loss": 46.0,
-      "step": 13028
-    },
-    {
-      "epoch": 2.0982326180603086,
-      "grad_norm": 0.009705738164484501,
-      "learning_rate": 0.0001999978309148548,
-      "loss": 46.0,
-      "step": 13029
-    },
-    {
-      "epoch": 2.0983936551390956,
-      "grad_norm": 0.006310978904366493,
-      "learning_rate": 0.0001999978305816248,
-      "loss": 46.0,
-      "step": 13030
-    },
-    {
-      "epoch": 2.098554692217883,
-      "grad_norm": 0.007194725796580315,
-      "learning_rate": 0.00019999783024836923,
-      "loss": 46.0,
-      "step": 13031
-    },
-    {
-      "epoch": 2.0987157292966705,
-      "grad_norm": 0.0006201033247634768,
-      "learning_rate": 0.00019999782991508803,
-      "loss": 46.0,
-      "step": 13032
-    },
-    {
-      "epoch": 2.098876766375458,
-      "grad_norm": 0.003160185646265745,
-      "learning_rate": 0.00019999782958178127,
-      "loss": 46.0,
-      "step": 13033
-    },
-    {
-      "epoch": 2.0990378034542454,
-      "grad_norm": 0.003502171253785491,
-      "learning_rate": 0.00019999782924844892,
-      "loss": 46.0,
-      "step": 13034
-    },
-    {
-      "epoch": 2.099198840533033,
-      "grad_norm": 0.005906878039240837,
-      "learning_rate": 0.00019999782891509096,
-      "loss": 46.0,
-      "step": 13035
-    },
-    {
-      "epoch": 2.0993598776118203,
-      "grad_norm": 0.004086571279913187,
-      "learning_rate": 0.0001999978285817074,
-      "loss": 46.0,
-      "step": 13036
-    },
-    {
-      "epoch": 2.0995209146906073,
-      "grad_norm": 0.0012891882797703147,
-      "learning_rate": 0.00019999782824829825,
-      "loss": 46.0,
-      "step": 13037
-    },
-    {
-      "epoch": 2.0996819517693948,
-      "grad_norm": 0.0017189068021252751,
-      "learning_rate": 0.0001999978279148635,
-      "loss": 46.0,
-      "step": 13038
-    },
-    {
-      "epoch": 2.099842988848182,
-      "grad_norm": 0.0016231165500357747,
-      "learning_rate": 0.00019999782758140316,
-      "loss": 46.0,
-      "step": 13039
-    },
-    {
-      "epoch": 2.1000040259269697,
-      "grad_norm": 0.01039725448936224,
-      "learning_rate": 0.0001999978272479172,
-      "loss": 46.0,
-      "step": 13040
-    },
-    {
-      "epoch": 2.100165063005757,
-      "grad_norm": 0.0015360346296802163,
-      "learning_rate": 0.00019999782691440573,
-      "loss": 46.0,
-      "step": 13041
-    },
-    {
-      "epoch": 2.1003261000845446,
-      "grad_norm": 0.001290195039473474,
-      "learning_rate": 0.00019999782658086857,
-      "loss": 46.0,
-      "step": 13042
-    },
-    {
-      "epoch": 2.100487137163332,
-      "grad_norm": 0.0018424431327730417,
-      "learning_rate": 0.0001999978262473059,
-      "loss": 46.0,
-      "step": 13043
-    },
-    {
-      "epoch": 2.100648174242119,
-      "grad_norm": 0.0008400485967285931,
-      "learning_rate": 0.0001999978259137176,
-      "loss": 46.0,
-      "step": 13044
-    },
-    {
-      "epoch": 2.1008092113209065,
-      "grad_norm": 0.002809921745210886,
-      "learning_rate": 0.00019999782558010367,
-      "loss": 46.0,
-      "step": 13045
-    },
-    {
-      "epoch": 2.100970248399694,
-      "grad_norm": 0.0026175386738032103,
-      "learning_rate": 0.00019999782524646417,
-      "loss": 46.0,
-      "step": 13046
-    },
-    {
-      "epoch": 2.1011312854784814,
-      "grad_norm": 0.00609249621629715,
-      "learning_rate": 0.00019999782491279908,
-      "loss": 46.0,
-      "step": 13047
-    },
-    {
-      "epoch": 2.101292322557269,
-      "grad_norm": 0.0023897849023342133,
-      "learning_rate": 0.0001999978245791084,
-      "loss": 46.0,
-      "step": 13048
-    },
-    {
-      "epoch": 2.1014533596360563,
-      "grad_norm": 0.0007847766391932964,
-      "learning_rate": 0.00019999782424539212,
-      "loss": 46.0,
-      "step": 13049
-    },
-    {
-      "epoch": 2.1016143967148437,
-      "grad_norm": 0.004051079507917166,
-      "learning_rate": 0.00019999782391165024,
-      "loss": 46.0,
-      "step": 13050
-    },
-    {
-      "epoch": 2.101775433793631,
-      "grad_norm": 0.004867277108132839,
-      "learning_rate": 0.00019999782357788277,
-      "loss": 46.0,
-      "step": 13051
-    },
-    {
-      "epoch": 2.101936470872418,
-      "grad_norm": 0.0021208066027611494,
-      "learning_rate": 0.00019999782324408972,
-      "loss": 46.0,
-      "step": 13052
-    },
-    {
-      "epoch": 2.1020975079512056,
-      "grad_norm": 0.001158708124421537,
-      "learning_rate": 0.00019999782291027106,
-      "loss": 46.0,
-      "step": 13053
-    },
-    {
-      "epoch": 2.102258545029993,
-      "grad_norm": 0.00791366770863533,
-      "learning_rate": 0.00019999782257642678,
-      "loss": 46.0,
-      "step": 13054
-    },
-    {
-      "epoch": 2.1024195821087805,
-      "grad_norm": 0.00036076310789212584,
-      "learning_rate": 0.00019999782224255696,
-      "loss": 46.0,
-      "step": 13055
-    },
-    {
-      "epoch": 2.102580619187568,
-      "grad_norm": 0.003293624147772789,
-      "learning_rate": 0.0001999978219086615,
-      "loss": 46.0,
-      "step": 13056
-    },
-    {
-      "epoch": 2.1027416562663555,
-      "grad_norm": 0.0014254057314246893,
-      "learning_rate": 0.0001999978215747405,
-      "loss": 46.0,
-      "step": 13057
-    },
-    {
-      "epoch": 2.102902693345143,
-      "grad_norm": 0.0019138320349156857,
-      "learning_rate": 0.00019999782124079384,
-      "loss": 46.0,
-      "step": 13058
-    },
-    {
-      "epoch": 2.10306373042393,
-      "grad_norm": 0.001688587130047381,
-      "learning_rate": 0.00019999782090682162,
-      "loss": 46.0,
-      "step": 13059
-    },
-    {
-      "epoch": 2.1032247675027174,
-      "grad_norm": 0.001946649863384664,
-      "learning_rate": 0.00019999782057282382,
-      "loss": 46.0,
-      "step": 13060
-    },
-    {
-      "epoch": 2.103385804581505,
-      "grad_norm": 0.013654623180627823,
-      "learning_rate": 0.0001999978202388004,
-      "loss": 46.0,
-      "step": 13061
-    },
-    {
-      "epoch": 2.1035468416602923,
-      "grad_norm": 0.0007878972101025283,
-      "learning_rate": 0.0001999978199047514,
-      "loss": 46.0,
-      "step": 13062
-    },
-    {
-      "epoch": 2.1037078787390797,
-      "grad_norm": 0.0008884586859494448,
-      "learning_rate": 0.00019999781957067678,
-      "loss": 46.0,
-      "step": 13063
-    },
-    {
-      "epoch": 2.103868915817867,
-      "grad_norm": 0.0005142706213518977,
-      "learning_rate": 0.00019999781923657662,
-      "loss": 46.0,
-      "step": 13064
-    },
-    {
-      "epoch": 2.1040299528966546,
-      "grad_norm": 0.001775235403329134,
-      "learning_rate": 0.00019999781890245083,
-      "loss": 46.0,
-      "step": 13065
-    },
-    {
-      "epoch": 2.104190989975442,
-      "grad_norm": 0.0008990365313366055,
-      "learning_rate": 0.00019999781856829942,
-      "loss": 46.0,
-      "step": 13066
-    },
-    {
-      "epoch": 2.104352027054229,
-      "grad_norm": 0.0032157900277525187,
-      "learning_rate": 0.00019999781823412246,
-      "loss": 46.0,
-      "step": 13067
-    },
-    {
-      "epoch": 2.1045130641330165,
-      "grad_norm": 0.0034256966318935156,
-      "learning_rate": 0.00019999781789991987,
-      "loss": 46.0,
-      "step": 13068
-    },
-    {
-      "epoch": 2.104674101211804,
-      "grad_norm": 0.002097911899909377,
-      "learning_rate": 0.0001999978175656917,
-      "loss": 46.0,
-      "step": 13069
-    },
-    {
-      "epoch": 2.1048351382905914,
-      "grad_norm": 0.0034216258209198713,
-      "learning_rate": 0.00019999781723143795,
-      "loss": 46.0,
-      "step": 13070
-    },
-    {
-      "epoch": 2.104996175369379,
-      "grad_norm": 0.0008702049381099641,
-      "learning_rate": 0.0001999978168971586,
-      "loss": 46.0,
-      "step": 13071
-    },
-    {
-      "epoch": 2.1051572124481663,
-      "grad_norm": 0.003039182862266898,
-      "learning_rate": 0.00019999781656285365,
-      "loss": 46.0,
-      "step": 13072
-    },
-    {
-      "epoch": 2.105318249526954,
-      "grad_norm": 0.005514364689588547,
-      "learning_rate": 0.00019999781622852313,
-      "loss": 46.0,
-      "step": 13073
-    },
-    {
-      "epoch": 2.105479286605741,
-      "grad_norm": 0.0034514337312430143,
-      "learning_rate": 0.00019999781589416697,
-      "loss": 46.0,
-      "step": 13074
-    },
-    {
-      "epoch": 2.1056403236845282,
-      "grad_norm": 0.0012727950233966112,
-      "learning_rate": 0.00019999781555978522,
-      "loss": 46.0,
-      "step": 13075
-    },
-    {
-      "epoch": 2.1058013607633157,
-      "grad_norm": 0.002618958940729499,
-      "learning_rate": 0.00019999781522537792,
-      "loss": 46.0,
-      "step": 13076
-    },
-    {
-      "epoch": 2.105962397842103,
-      "grad_norm": 0.0012678103521466255,
-      "learning_rate": 0.00019999781489094497,
-      "loss": 46.0,
-      "step": 13077
-    },
-    {
-      "epoch": 2.1061234349208906,
-      "grad_norm": 0.0037736110389232635,
-      "learning_rate": 0.0001999978145564865,
-      "loss": 46.0,
-      "step": 13078
-    },
-    {
-      "epoch": 2.106284471999678,
-      "grad_norm": 0.003596524940803647,
-      "learning_rate": 0.00019999781422200236,
-      "loss": 46.0,
-      "step": 13079
-    },
-    {
-      "epoch": 2.1064455090784655,
-      "grad_norm": 0.0021964493207633495,
-      "learning_rate": 0.00019999781388749266,
-      "loss": 46.0,
-      "step": 13080
-    },
-    {
-      "epoch": 2.1066065461572525,
-      "grad_norm": 0.003467605682089925,
-      "learning_rate": 0.00019999781355295736,
-      "loss": 46.0,
-      "step": 13081
-    },
-    {
-      "epoch": 2.10676758323604,
-      "grad_norm": 0.005575900431722403,
-      "learning_rate": 0.00019999781321839647,
-      "loss": 46.0,
-      "step": 13082
-    },
-    {
-      "epoch": 2.1069286203148274,
-      "grad_norm": 0.002288502175360918,
-      "learning_rate": 0.00019999781288380998,
-      "loss": 46.0,
-      "step": 13083
-    },
-    {
-      "epoch": 2.107089657393615,
-      "grad_norm": 0.0015003661392256618,
-      "learning_rate": 0.00019999781254919792,
-      "loss": 46.0,
-      "step": 13084
-    },
-    {
-      "epoch": 2.1072506944724023,
-      "grad_norm": 0.0071339975111186504,
-      "learning_rate": 0.00019999781221456022,
-      "loss": 46.0,
-      "step": 13085
-    },
-    {
-      "epoch": 2.1074117315511898,
-      "grad_norm": 0.002094605704769492,
-      "learning_rate": 0.00019999781187989693,
-      "loss": 46.0,
-      "step": 13086
-    },
-    {
-      "epoch": 2.1075727686299772,
-      "grad_norm": 0.0029553426429629326,
-      "learning_rate": 0.00019999781154520809,
-      "loss": 46.0,
-      "step": 13087
-    },
-    {
-      "epoch": 2.1077338057087642,
-      "grad_norm": 0.0033231982961297035,
-      "learning_rate": 0.00019999781121049362,
-      "loss": 46.0,
-      "step": 13088
-    },
-    {
-      "epoch": 2.1078948427875517,
-      "grad_norm": 0.0011876259231939912,
-      "learning_rate": 0.00019999781087575358,
-      "loss": 46.0,
-      "step": 13089
-    },
-    {
-      "epoch": 2.108055879866339,
-      "grad_norm": 0.004289860837161541,
-      "learning_rate": 0.00019999781054098794,
-      "loss": 46.0,
-      "step": 13090
-    },
-    {
-      "epoch": 2.1082169169451266,
-      "grad_norm": 0.0029466752894222736,
-      "learning_rate": 0.0001999978102061967,
-      "loss": 46.0,
-      "step": 13091
-    },
-    {
-      "epoch": 2.108377954023914,
-      "grad_norm": 0.0024188857059925795,
-      "learning_rate": 0.00019999780987137985,
-      "loss": 46.0,
-      "step": 13092
-    },
-    {
-      "epoch": 2.1085389911027015,
-      "grad_norm": 0.004038097336888313,
-      "learning_rate": 0.00019999780953653743,
-      "loss": 46.0,
-      "step": 13093
-    },
-    {
-      "epoch": 2.108700028181489,
-      "grad_norm": 0.0012877986300736666,
-      "learning_rate": 0.0001999978092016694,
-      "loss": 46.0,
-      "step": 13094
-    },
-    {
-      "epoch": 2.1088610652602764,
-      "grad_norm": 0.00403616763651371,
-      "learning_rate": 0.0001999978088667758,
-      "loss": 46.0,
-      "step": 13095
-    },
-    {
-      "epoch": 2.1090221023390634,
-      "grad_norm": 0.006237053778022528,
-      "learning_rate": 0.00019999780853185655,
-      "loss": 46.0,
-      "step": 13096
-    },
-    {
-      "epoch": 2.109183139417851,
-      "grad_norm": 0.0037090268451720476,
-      "learning_rate": 0.00019999780819691178,
-      "loss": 46.0,
-      "step": 13097
-    },
-    {
-      "epoch": 2.1093441764966383,
-      "grad_norm": 0.002308735391125083,
-      "learning_rate": 0.00019999780786194137,
-      "loss": 46.0,
-      "step": 13098
-    },
-    {
-      "epoch": 2.1095052135754258,
-      "grad_norm": 0.0072404914535582066,
-      "learning_rate": 0.00019999780752694537,
-      "loss": 46.0,
-      "step": 13099
-    },
-    {
-      "epoch": 2.109666250654213,
-      "grad_norm": 0.0009635112946853042,
-      "learning_rate": 0.00019999780719192378,
-      "loss": 46.0,
-      "step": 13100
-    },
-    {
-      "epoch": 2.1098272877330007,
-      "grad_norm": 0.007864450104534626,
-      "learning_rate": 0.0001999978068568766,
-      "loss": 46.0,
-      "step": 13101
-    },
-    {
-      "epoch": 2.109988324811788,
-      "grad_norm": 0.002165738958865404,
-      "learning_rate": 0.0001999978065218038,
-      "loss": 46.0,
-      "step": 13102
-    },
-    {
-      "epoch": 2.110149361890575,
-      "grad_norm": 0.003186293877661228,
-      "learning_rate": 0.00019999780618670546,
-      "loss": 46.0,
-      "step": 13103
-    },
-    {
-      "epoch": 2.1103103989693626,
-      "grad_norm": 0.0014151227660477161,
-      "learning_rate": 0.00019999780585158147,
-      "loss": 46.0,
-      "step": 13104
-    },
-    {
-      "epoch": 2.11047143604815,
-      "grad_norm": 0.0019516373286023736,
-      "learning_rate": 0.00019999780551643192,
-      "loss": 46.0,
-      "step": 13105
-    },
-    {
-      "epoch": 2.1106324731269375,
-      "grad_norm": 0.00819750688970089,
-      "learning_rate": 0.00019999780518125678,
-      "loss": 46.0,
-      "step": 13106
-    },
-    {
-      "epoch": 2.110793510205725,
-      "grad_norm": 0.002146652899682522,
-      "learning_rate": 0.000199997804846056,
-      "loss": 46.0,
-      "step": 13107
-    },
-    {
-      "epoch": 2.1109545472845124,
-      "grad_norm": 0.003578031435608864,
-      "learning_rate": 0.0001999978045108297,
-      "loss": 46.0,
-      "step": 13108
-    },
-    {
-      "epoch": 2.1111155843633,
-      "grad_norm": 0.00167667877394706,
-      "learning_rate": 0.0001999978041755777,
-      "loss": 46.0,
-      "step": 13109
-    },
-    {
-      "epoch": 2.111276621442087,
-      "grad_norm": 0.002079801866784692,
-      "learning_rate": 0.0001999978038403002,
-      "loss": 46.0,
-      "step": 13110
-    },
-    {
-      "epoch": 2.1114376585208743,
-      "grad_norm": 0.0006874401587992907,
-      "learning_rate": 0.00019999780350499707,
-      "loss": 46.0,
-      "step": 13111
-    },
-    {
-      "epoch": 2.1115986955996617,
-      "grad_norm": 0.0007555847405456007,
-      "learning_rate": 0.00019999780316966838,
-      "loss": 46.0,
-      "step": 13112
-    },
-    {
-      "epoch": 2.111759732678449,
-      "grad_norm": 0.0034825955517590046,
-      "learning_rate": 0.00019999780283431405,
-      "loss": 46.0,
-      "step": 13113
-    },
-    {
-      "epoch": 2.1119207697572366,
-      "grad_norm": 0.0012161614140495658,
-      "learning_rate": 0.00019999780249893413,
-      "loss": 46.0,
-      "step": 13114
-    },
-    {
-      "epoch": 2.112081806836024,
-      "grad_norm": 0.0020795147866010666,
-      "learning_rate": 0.00019999780216352862,
-      "loss": 46.0,
-      "step": 13115
-    },
-    {
-      "epoch": 2.1122428439148115,
-      "grad_norm": 0.0027438029646873474,
-      "learning_rate": 0.0001999978018280975,
-      "loss": 46.0,
-      "step": 13116
-    },
-    {
-      "epoch": 2.1124038809935985,
-      "grad_norm": 0.001997697865590453,
-      "learning_rate": 0.00019999780149264083,
-      "loss": 46.0,
-      "step": 13117
-    },
-    {
-      "epoch": 2.112564918072386,
-      "grad_norm": 0.0015676217153668404,
-      "learning_rate": 0.00019999780115715853,
-      "loss": 46.0,
-      "step": 13118
-    },
-    {
-      "epoch": 2.1127259551511735,
-      "grad_norm": 0.001409698510542512,
-      "learning_rate": 0.00019999780082165068,
-      "loss": 46.0,
-      "step": 13119
-    },
-    {
-      "epoch": 2.112886992229961,
-      "grad_norm": 0.0010853136191144586,
-      "learning_rate": 0.0001999978004861172,
-      "loss": 46.0,
-      "step": 13120
-    },
-    {
-      "epoch": 2.1130480293087484,
-      "grad_norm": 0.0007200825493782759,
-      "learning_rate": 0.0001999978001505581,
-      "loss": 46.0,
-      "step": 13121
-    },
-    {
-      "epoch": 2.113209066387536,
-      "grad_norm": 0.0033002144191414118,
-      "learning_rate": 0.00019999779981497344,
-      "loss": 46.0,
-      "step": 13122
-    },
-    {
-      "epoch": 2.1133701034663233,
-      "grad_norm": 0.0008897663210518658,
-      "learning_rate": 0.00019999779947936318,
-      "loss": 46.0,
-      "step": 13123
-    },
-    {
-      "epoch": 2.1135311405451107,
-      "grad_norm": 0.00624080840498209,
-      "learning_rate": 0.00019999779914372734,
-      "loss": 46.0,
-      "step": 13124
-    },
-    {
-      "epoch": 2.1136921776238977,
-      "grad_norm": 0.0023192770313471556,
-      "learning_rate": 0.0001999977988080659,
-      "loss": 46.0,
-      "step": 13125
-    },
-    {
-      "epoch": 2.113853214702685,
-      "grad_norm": 0.0022517463658005,
-      "learning_rate": 0.00019999779847237886,
-      "loss": 46.0,
-      "step": 13126
-    },
-    {
-      "epoch": 2.1140142517814726,
-      "grad_norm": 0.0012670793803408742,
-      "learning_rate": 0.0001999977981366662,
-      "loss": 46.0,
-      "step": 13127
-    },
-    {
-      "epoch": 2.11417528886026,
-      "grad_norm": 0.0024502556771039963,
-      "learning_rate": 0.00019999779780092798,
-      "loss": 46.0,
-      "step": 13128
-    },
-    {
-      "epoch": 2.1143363259390475,
-      "grad_norm": 0.0022671790793538094,
-      "learning_rate": 0.00019999779746516415,
-      "loss": 46.0,
-      "step": 13129
-    },
-    {
-      "epoch": 2.114497363017835,
-      "grad_norm": 0.0013940547360107303,
-      "learning_rate": 0.00019999779712937473,
-      "loss": 46.0,
-      "step": 13130
-    },
-    {
-      "epoch": 2.1146584000966224,
-      "grad_norm": 0.0017259205924347043,
-      "learning_rate": 0.00019999779679355972,
-      "loss": 46.0,
-      "step": 13131
-    },
-    {
-      "epoch": 2.1148194371754094,
-      "grad_norm": 0.0031034525018185377,
-      "learning_rate": 0.00019999779645771913,
-      "loss": 46.0,
-      "step": 13132
-    },
-    {
-      "epoch": 2.114980474254197,
-      "grad_norm": 0.001265834434889257,
-      "learning_rate": 0.00019999779612185292,
-      "loss": 46.0,
-      "step": 13133
-    },
-    {
-      "epoch": 2.1151415113329843,
-      "grad_norm": 0.004677072167396545,
-      "learning_rate": 0.00019999779578596112,
-      "loss": 46.0,
-      "step": 13134
-    },
-    {
-      "epoch": 2.115302548411772,
-      "grad_norm": 0.004497617483139038,
-      "learning_rate": 0.00019999779545004374,
-      "loss": 46.0,
-      "step": 13135
-    },
-    {
-      "epoch": 2.1154635854905592,
-      "grad_norm": 0.003626951714977622,
-      "learning_rate": 0.00019999779511410072,
-      "loss": 46.0,
-      "step": 13136
-    },
-    {
-      "epoch": 2.1156246225693467,
-      "grad_norm": 0.0022104203235358,
-      "learning_rate": 0.00019999779477813216,
-      "loss": 46.0,
-      "step": 13137
-    },
-    {
-      "epoch": 2.115785659648134,
-      "grad_norm": 0.0009194849408231676,
-      "learning_rate": 0.00019999779444213799,
-      "loss": 46.0,
-      "step": 13138
-    },
-    {
-      "epoch": 2.1159466967269216,
-      "grad_norm": 0.008031724952161312,
-      "learning_rate": 0.00019999779410611823,
-      "loss": 46.0,
-      "step": 13139
-    },
-    {
-      "epoch": 2.1161077338057086,
-      "grad_norm": 0.003028605366125703,
-      "learning_rate": 0.00019999779377007285,
-      "loss": 46.0,
-      "step": 13140
-    },
-    {
-      "epoch": 2.116268770884496,
-      "grad_norm": 0.00220558256842196,
-      "learning_rate": 0.00019999779343400192,
-      "loss": 46.0,
-      "step": 13141
-    },
-    {
-      "epoch": 2.1164298079632835,
-      "grad_norm": 0.001776234246790409,
-      "learning_rate": 0.00019999779309790535,
-      "loss": 46.0,
-      "step": 13142
-    },
-    {
-      "epoch": 2.116590845042071,
-      "grad_norm": 0.0032892825547605753,
-      "learning_rate": 0.0001999977927617832,
-      "loss": 46.0,
-      "step": 13143
-    },
-    {
-      "epoch": 2.1167518821208584,
-      "grad_norm": 0.003959726542234421,
-      "learning_rate": 0.00019999779242563546,
-      "loss": 46.0,
-      "step": 13144
-    },
-    {
-      "epoch": 2.116912919199646,
-      "grad_norm": 0.006606176495552063,
-      "learning_rate": 0.00019999779208946215,
-      "loss": 46.0,
-      "step": 13145
-    },
-    {
-      "epoch": 2.1170739562784333,
-      "grad_norm": 0.0038326343055814505,
-      "learning_rate": 0.0001999977917532632,
-      "loss": 46.0,
-      "step": 13146
-    },
-    {
-      "epoch": 2.1172349933572203,
-      "grad_norm": 0.006243380717933178,
-      "learning_rate": 0.0001999977914170387,
-      "loss": 46.0,
-      "step": 13147
-    },
-    {
-      "epoch": 2.1173960304360078,
-      "grad_norm": 0.0035989389289170504,
-      "learning_rate": 0.00019999779108078857,
-      "loss": 46.0,
-      "step": 13148
-    },
-    {
-      "epoch": 2.117557067514795,
-      "grad_norm": 0.0010204442078247666,
-      "learning_rate": 0.00019999779074451285,
-      "loss": 46.0,
-      "step": 13149
-    },
-    {
-      "epoch": 2.1177181045935827,
-      "grad_norm": 0.0005883739213459194,
-      "learning_rate": 0.00019999779040821155,
-      "loss": 46.0,
-      "step": 13150
-    },
-    {
-      "epoch": 2.11787914167237,
-      "grad_norm": 0.005654613021761179,
-      "learning_rate": 0.00019999779007188467,
-      "loss": 46.0,
-      "step": 13151
-    },
-    {
-      "epoch": 2.1180401787511576,
-      "grad_norm": 0.0012487141648307443,
-      "learning_rate": 0.00019999778973553217,
-      "loss": 46.0,
-      "step": 13152
-    },
-    {
-      "epoch": 2.118201215829945,
-      "grad_norm": 0.0007987206918187439,
-      "learning_rate": 0.00019999778939915408,
-      "loss": 46.0,
-      "step": 13153
-    },
-    {
-      "epoch": 2.118362252908732,
-      "grad_norm": 0.007598782889544964,
-      "learning_rate": 0.0001999977890627504,
-      "loss": 46.0,
-      "step": 13154
-    },
-    {
-      "epoch": 2.1185232899875195,
-      "grad_norm": 0.0019899557810276747,
-      "learning_rate": 0.00019999778872632111,
-      "loss": 46.0,
-      "step": 13155
-    },
-    {
-      "epoch": 2.118684327066307,
-      "grad_norm": 0.00354297598823905,
-      "learning_rate": 0.00019999778838986624,
-      "loss": 46.0,
-      "step": 13156
-    },
-    {
-      "epoch": 2.1188453641450944,
-      "grad_norm": 0.001969203818589449,
-      "learning_rate": 0.00019999778805338577,
-      "loss": 46.0,
-      "step": 13157
-    },
-    {
-      "epoch": 2.119006401223882,
-      "grad_norm": 0.001812206581234932,
-      "learning_rate": 0.00019999778771687972,
-      "loss": 46.0,
-      "step": 13158
-    },
-    {
-      "epoch": 2.1191674383026693,
-      "grad_norm": 0.0026534204371273518,
-      "learning_rate": 0.00019999778738034806,
-      "loss": 46.0,
-      "step": 13159
-    },
-    {
-      "epoch": 2.1193284753814567,
-      "grad_norm": 0.002966444008052349,
-      "learning_rate": 0.0001999977870437908,
-      "loss": 46.0,
-      "step": 13160
-    },
-    {
-      "epoch": 2.1194895124602438,
-      "grad_norm": 0.0016665224684402347,
-      "learning_rate": 0.000199997786707208,
-      "loss": 46.0,
-      "step": 13161
-    },
-    {
-      "epoch": 2.119650549539031,
-      "grad_norm": 0.0010220034746453166,
-      "learning_rate": 0.00019999778637059954,
-      "loss": 46.0,
-      "step": 13162
-    },
-    {
-      "epoch": 2.1198115866178187,
-      "grad_norm": 0.003974624909460545,
-      "learning_rate": 0.0001999977860339655,
-      "loss": 46.0,
-      "step": 13163
-    },
-    {
-      "epoch": 2.119972623696606,
-      "grad_norm": 0.001191667397506535,
-      "learning_rate": 0.00019999778569730587,
-      "loss": 46.0,
-      "step": 13164
-    },
-    {
-      "epoch": 2.1201336607753936,
-      "grad_norm": 0.008775805123150349,
-      "learning_rate": 0.00019999778536062066,
-      "loss": 46.0,
-      "step": 13165
-    },
-    {
-      "epoch": 2.120294697854181,
-      "grad_norm": 0.004642212297767401,
-      "learning_rate": 0.00019999778502390985,
-      "loss": 46.0,
-      "step": 13166
-    },
-    {
-      "epoch": 2.1204557349329685,
-      "grad_norm": 0.003119399771094322,
-      "learning_rate": 0.00019999778468717346,
-      "loss": 46.0,
-      "step": 13167
-    },
-    {
-      "epoch": 2.120616772011756,
-      "grad_norm": 0.0015318223740905523,
-      "learning_rate": 0.00019999778435041146,
-      "loss": 46.0,
-      "step": 13168
-    },
-    {
-      "epoch": 2.120777809090543,
-      "grad_norm": 0.0025191716849803925,
-      "learning_rate": 0.00019999778401362384,
-      "loss": 46.0,
-      "step": 13169
-    },
-    {
-      "epoch": 2.1209388461693304,
-      "grad_norm": 0.0011357973562553525,
-      "learning_rate": 0.00019999778367681064,
-      "loss": 46.0,
-      "step": 13170
-    },
-    {
-      "epoch": 2.121099883248118,
-      "grad_norm": 0.003429251490160823,
-      "learning_rate": 0.00019999778333997184,
-      "loss": 46.0,
-      "step": 13171
-    },
-    {
-      "epoch": 2.1212609203269053,
-      "grad_norm": 0.004361074883490801,
-      "learning_rate": 0.0001999977830031075,
-      "loss": 46.0,
-      "step": 13172
-    },
-    {
-      "epoch": 2.1214219574056927,
-      "grad_norm": 0.0060005937702953815,
-      "learning_rate": 0.00019999778266621752,
-      "loss": 46.0,
-      "step": 13173
-    },
-    {
-      "epoch": 2.12158299448448,
-      "grad_norm": 0.015361596830189228,
-      "learning_rate": 0.00019999778232930194,
-      "loss": 46.0,
-      "step": 13174
-    },
-    {
-      "epoch": 2.1217440315632676,
-      "grad_norm": 0.003407333977520466,
-      "learning_rate": 0.00019999778199236077,
-      "loss": 46.0,
-      "step": 13175
-    },
-    {
-      "epoch": 2.1219050686420546,
-      "grad_norm": 0.002047986490651965,
-      "learning_rate": 0.00019999778165539402,
-      "loss": 46.0,
-      "step": 13176
-    },
-    {
-      "epoch": 2.122066105720842,
-      "grad_norm": 0.0006436305120587349,
-      "learning_rate": 0.00019999778131840168,
-      "loss": 46.0,
-      "step": 13177
-    },
-    {
-      "epoch": 2.1222271427996295,
-      "grad_norm": 0.0034564253874123096,
-      "learning_rate": 0.00019999778098138372,
-      "loss": 46.0,
-      "step": 13178
-    },
-    {
-      "epoch": 2.122388179878417,
-      "grad_norm": 0.0026176944375038147,
-      "learning_rate": 0.00019999778064434018,
-      "loss": 46.0,
-      "step": 13179
-    },
-    {
-      "epoch": 2.1225492169572044,
-      "grad_norm": 0.004664492327719927,
-      "learning_rate": 0.00019999778030727104,
-      "loss": 46.0,
-      "step": 13180
-    },
-    {
-      "epoch": 2.122710254035992,
-      "grad_norm": 0.014643349684774876,
-      "learning_rate": 0.00019999777997017633,
-      "loss": 46.0,
-      "step": 13181
-    },
-    {
-      "epoch": 2.1228712911147793,
-      "grad_norm": 0.001551997964270413,
-      "learning_rate": 0.00019999777963305602,
-      "loss": 46.0,
-      "step": 13182
-    },
-    {
-      "epoch": 2.1230323281935664,
-      "grad_norm": 0.003320495132356882,
-      "learning_rate": 0.00019999777929591007,
-      "loss": 46.0,
-      "step": 13183
-    },
-    {
-      "epoch": 2.123193365272354,
-      "grad_norm": 0.00405589584261179,
-      "learning_rate": 0.00019999777895873857,
-      "loss": 46.0,
-      "step": 13184
-    },
-    {
-      "epoch": 2.1233544023511413,
-      "grad_norm": 0.0006907575880177319,
-      "learning_rate": 0.00019999777862154144,
-      "loss": 46.0,
-      "step": 13185
-    },
-    {
-      "epoch": 2.1235154394299287,
-      "grad_norm": 0.002527642762288451,
-      "learning_rate": 0.00019999777828431876,
-      "loss": 46.0,
-      "step": 13186
-    },
-    {
-      "epoch": 2.123676476508716,
-      "grad_norm": 0.006069401744753122,
-      "learning_rate": 0.0001999977779470705,
-      "loss": 46.0,
-      "step": 13187
-    },
-    {
-      "epoch": 2.1238375135875036,
-      "grad_norm": 0.0019359297584742308,
-      "learning_rate": 0.00019999777760979658,
-      "loss": 46.0,
-      "step": 13188
-    },
-    {
-      "epoch": 2.123998550666291,
-      "grad_norm": 0.0017593411030247808,
-      "learning_rate": 0.00019999777727249711,
-      "loss": 46.0,
-      "step": 13189
-    },
-    {
-      "epoch": 2.1241595877450785,
-      "grad_norm": 0.004458352457731962,
-      "learning_rate": 0.000199997776935172,
-      "loss": 46.0,
-      "step": 13190
-    },
-    {
-      "epoch": 2.1243206248238655,
-      "grad_norm": 0.003931291401386261,
-      "learning_rate": 0.00019999777659782136,
-      "loss": 46.0,
-      "step": 13191
-    },
-    {
-      "epoch": 2.124481661902653,
-      "grad_norm": 0.004808352794498205,
-      "learning_rate": 0.0001999977762604451,
-      "loss": 46.0,
-      "step": 13192
-    },
-    {
-      "epoch": 2.1246426989814404,
-      "grad_norm": 0.0015759035013616085,
-      "learning_rate": 0.00019999777592304323,
-      "loss": 46.0,
-      "step": 13193
-    },
-    {
-      "epoch": 2.124803736060228,
-      "grad_norm": 0.0029267179779708385,
-      "learning_rate": 0.00019999777558561577,
-      "loss": 46.0,
-      "step": 13194
-    },
-    {
-      "epoch": 2.1249647731390153,
-      "grad_norm": 0.000670771871227771,
-      "learning_rate": 0.00019999777524816272,
-      "loss": 46.0,
-      "step": 13195
-    },
-    {
-      "epoch": 2.125125810217803,
-      "grad_norm": 0.0008510480402037501,
-      "learning_rate": 0.0001999977749106841,
-      "loss": 46.0,
-      "step": 13196
-    },
-    {
-      "epoch": 2.1252868472965902,
-      "grad_norm": 0.0035457415506243706,
-      "learning_rate": 0.00019999777457317986,
-      "loss": 46.0,
-      "step": 13197
-    },
-    {
-      "epoch": 2.1254478843753772,
-      "grad_norm": 0.002282337285578251,
-      "learning_rate": 0.00019999777423565003,
-      "loss": 46.0,
-      "step": 13198
-    },
-    {
-      "epoch": 2.1256089214541647,
-      "grad_norm": 0.0036603291518986225,
-      "learning_rate": 0.00019999777389809458,
-      "loss": 46.0,
-      "step": 13199
-    },
-    {
-      "epoch": 2.125769958532952,
-      "grad_norm": 0.003987677861005068,
-      "learning_rate": 0.00019999777356051357,
-      "loss": 46.0,
-      "step": 13200
-    },
-    {
-      "epoch": 2.1259309956117396,
-      "grad_norm": 0.007842597551643848,
-      "learning_rate": 0.00019999777322290697,
-      "loss": 46.0,
-      "step": 13201
-    },
-    {
-      "epoch": 2.126092032690527,
-      "grad_norm": 0.001646373770199716,
-      "learning_rate": 0.00019999777288527476,
-      "loss": 46.0,
-      "step": 13202
-    },
-    {
-      "epoch": 2.1262530697693145,
-      "grad_norm": 0.0014534646179527044,
-      "learning_rate": 0.00019999777254761693,
-      "loss": 46.0,
-      "step": 13203
-    },
-    {
-      "epoch": 2.126414106848102,
-      "grad_norm": 0.004115855786949396,
-      "learning_rate": 0.00019999777220993355,
-      "loss": 46.0,
-      "step": 13204
-    },
-    {
-      "epoch": 2.126575143926889,
-      "grad_norm": 0.001109655131585896,
-      "learning_rate": 0.00019999777187222457,
-      "loss": 46.0,
-      "step": 13205
-    },
-    {
-      "epoch": 2.1267361810056764,
-      "grad_norm": 0.0011397922644391656,
-      "learning_rate": 0.00019999777153448999,
-      "loss": 46.0,
-      "step": 13206
-    },
-    {
-      "epoch": 2.126897218084464,
-      "grad_norm": 0.0014358978951349854,
-      "learning_rate": 0.0001999977711967298,
-      "loss": 46.0,
-      "step": 13207
-    },
-    {
-      "epoch": 2.1270582551632513,
-      "grad_norm": 0.00215542851947248,
-      "learning_rate": 0.000199997770858944,
-      "loss": 46.0,
-      "step": 13208
-    },
-    {
-      "epoch": 2.1272192922420388,
-      "grad_norm": 0.0012454851530492306,
-      "learning_rate": 0.00019999777052113265,
-      "loss": 46.0,
-      "step": 13209
-    },
-    {
-      "epoch": 2.127380329320826,
-      "grad_norm": 0.006422761827707291,
-      "learning_rate": 0.00019999777018329568,
-      "loss": 46.0,
-      "step": 13210
-    },
-    {
-      "epoch": 2.1275413663996137,
-      "grad_norm": 0.0028827968053519726,
-      "learning_rate": 0.00019999776984543313,
-      "loss": 46.0,
-      "step": 13211
-    },
-    {
-      "epoch": 2.127702403478401,
-      "grad_norm": 0.0038349488750100136,
-      "learning_rate": 0.00019999776950754497,
-      "loss": 46.0,
-      "step": 13212
-    },
-    {
-      "epoch": 2.127863440557188,
-      "grad_norm": 0.004856688901782036,
-      "learning_rate": 0.00019999776916963122,
-      "loss": 46.0,
-      "step": 13213
-    },
-    {
-      "epoch": 2.1280244776359756,
-      "grad_norm": 0.0012035154504701495,
-      "learning_rate": 0.0001999977688316919,
-      "loss": 46.0,
-      "step": 13214
-    },
-    {
-      "epoch": 2.128185514714763,
-      "grad_norm": 0.000896472716704011,
-      "learning_rate": 0.00019999776849372695,
-      "loss": 46.0,
-      "step": 13215
-    },
-    {
-      "epoch": 2.1283465517935505,
-      "grad_norm": 0.003361812559887767,
-      "learning_rate": 0.00019999776815573644,
-      "loss": 46.0,
-      "step": 13216
-    },
-    {
-      "epoch": 2.128507588872338,
-      "grad_norm": 0.0012828216422349215,
-      "learning_rate": 0.0001999977678177203,
-      "loss": 46.0,
-      "step": 13217
-    },
-    {
-      "epoch": 2.1286686259511254,
-      "grad_norm": 0.0016373510006815195,
-      "learning_rate": 0.00019999776747967857,
-      "loss": 46.0,
-      "step": 13218
-    },
-    {
-      "epoch": 2.128829663029913,
-      "grad_norm": 0.0030524181202054024,
-      "learning_rate": 0.00019999776714161127,
-      "loss": 46.0,
-      "step": 13219
-    },
-    {
-      "epoch": 2.1289907001087,
-      "grad_norm": 0.0026899180375039577,
-      "learning_rate": 0.00019999776680351835,
-      "loss": 46.0,
-      "step": 13220
-    },
-    {
-      "epoch": 2.1291517371874873,
-      "grad_norm": 0.0019747407641261816,
-      "learning_rate": 0.00019999776646539985,
-      "loss": 46.0,
-      "step": 13221
-    },
-    {
-      "epoch": 2.1293127742662747,
-      "grad_norm": 0.0024062597658485174,
-      "learning_rate": 0.00019999776612725575,
-      "loss": 46.0,
-      "step": 13222
-    },
-    {
-      "epoch": 2.129473811345062,
-      "grad_norm": 0.001246427884325385,
-      "learning_rate": 0.00019999776578908608,
-      "loss": 46.0,
-      "step": 13223
-    },
-    {
-      "epoch": 2.1296348484238496,
-      "grad_norm": 0.0037448867224156857,
-      "learning_rate": 0.00019999776545089079,
-      "loss": 46.0,
-      "step": 13224
-    },
-    {
-      "epoch": 2.129795885502637,
-      "grad_norm": 0.002583857625722885,
-      "learning_rate": 0.00019999776511266993,
-      "loss": 46.0,
-      "step": 13225
-    },
-    {
-      "epoch": 2.1299569225814245,
-      "grad_norm": 0.0011997292749583721,
-      "learning_rate": 0.00019999776477442344,
-      "loss": 46.0,
-      "step": 13226
-    },
-    {
-      "epoch": 2.1301179596602116,
-      "grad_norm": 0.002700342796742916,
-      "learning_rate": 0.00019999776443615139,
-      "loss": 46.0,
-      "step": 13227
-    },
-    {
-      "epoch": 2.130278996738999,
-      "grad_norm": 0.0008182794554159045,
-      "learning_rate": 0.00019999776409785375,
-      "loss": 46.0,
-      "step": 13228
-    },
-    {
-      "epoch": 2.1304400338177865,
-      "grad_norm": 0.003988698124885559,
-      "learning_rate": 0.00019999776375953046,
-      "loss": 46.0,
-      "step": 13229
-    },
-    {
-      "epoch": 2.130601070896574,
-      "grad_norm": 0.002513501327484846,
-      "learning_rate": 0.00019999776342118162,
-      "loss": 46.0,
-      "step": 13230
-    },
-    {
-      "epoch": 2.1307621079753614,
-      "grad_norm": 0.004659769590944052,
-      "learning_rate": 0.00019999776308280716,
-      "loss": 46.0,
-      "step": 13231
-    },
-    {
-      "epoch": 2.130923145054149,
-      "grad_norm": 0.0014631019439548254,
-      "learning_rate": 0.00019999776274440715,
-      "loss": 46.0,
-      "step": 13232
-    },
-    {
-      "epoch": 2.1310841821329363,
-      "grad_norm": 0.0012769061140716076,
-      "learning_rate": 0.0001999977624059815,
-      "loss": 46.0,
-      "step": 13233
-    },
-    {
-      "epoch": 2.1312452192117233,
-      "grad_norm": 0.003122994676232338,
-      "learning_rate": 0.00019999776206753027,
-      "loss": 46.0,
-      "step": 13234
-    },
-    {
-      "epoch": 2.1314062562905107,
-      "grad_norm": 0.0034654391929507256,
-      "learning_rate": 0.00019999776172905347,
-      "loss": 46.0,
-      "step": 13235
-    },
-    {
-      "epoch": 2.131567293369298,
-      "grad_norm": 0.008263527415692806,
-      "learning_rate": 0.00019999776139055105,
-      "loss": 46.0,
-      "step": 13236
-    },
-    {
-      "epoch": 2.1317283304480856,
-      "grad_norm": 0.0006701046950183809,
-      "learning_rate": 0.00019999776105202304,
-      "loss": 46.0,
-      "step": 13237
-    },
-    {
-      "epoch": 2.131889367526873,
-      "grad_norm": 0.003449707990512252,
-      "learning_rate": 0.00019999776071346942,
-      "loss": 46.0,
-      "step": 13238
-    },
-    {
-      "epoch": 2.1320504046056605,
-      "grad_norm": 0.0008667586953379214,
-      "learning_rate": 0.00019999776037489024,
-      "loss": 46.0,
-      "step": 13239
-    },
-    {
-      "epoch": 2.132211441684448,
-      "grad_norm": 0.0030587271321564913,
-      "learning_rate": 0.00019999776003628544,
-      "loss": 46.0,
-      "step": 13240
-    },
-    {
-      "epoch": 2.1323724787632354,
-      "grad_norm": 0.0013762717135250568,
-      "learning_rate": 0.00019999775969765503,
-      "loss": 46.0,
-      "step": 13241
-    },
-    {
-      "epoch": 2.1325335158420224,
-      "grad_norm": 0.0029092037584632635,
-      "learning_rate": 0.00019999775935899906,
-      "loss": 46.0,
-      "step": 13242
-    },
-    {
-      "epoch": 2.13269455292081,
-      "grad_norm": 0.003711071563884616,
-      "learning_rate": 0.0001999977590203175,
-      "loss": 46.0,
-      "step": 13243
-    },
-    {
-      "epoch": 2.1328555899995973,
-      "grad_norm": 0.004583151079714298,
-      "learning_rate": 0.00019999775868161034,
-      "loss": 46.0,
-      "step": 13244
-    },
-    {
-      "epoch": 2.133016627078385,
-      "grad_norm": 0.0033804697450250387,
-      "learning_rate": 0.00019999775834287755,
-      "loss": 46.0,
-      "step": 13245
-    },
-    {
-      "epoch": 2.1331776641571722,
-      "grad_norm": 0.0007176435901783407,
-      "learning_rate": 0.0001999977580041192,
-      "loss": 46.0,
-      "step": 13246
-    },
-    {
-      "epoch": 2.1333387012359597,
-      "grad_norm": 0.008193590678274632,
-      "learning_rate": 0.00019999775766533527,
-      "loss": 46.0,
-      "step": 13247
-    },
-    {
-      "epoch": 2.133499738314747,
-      "grad_norm": 0.0021499425638467073,
-      "learning_rate": 0.0001999977573265257,
-      "loss": 46.0,
-      "step": 13248
-    },
-    {
-      "epoch": 2.133660775393534,
-      "grad_norm": 0.001611798768863082,
-      "learning_rate": 0.00019999775698769057,
-      "loss": 46.0,
-      "step": 13249
-    },
-    {
-      "epoch": 2.1338218124723216,
-      "grad_norm": 0.003761806059628725,
-      "learning_rate": 0.00019999775664882984,
-      "loss": 46.0,
-      "step": 13250
-    },
-    {
-      "epoch": 2.133982849551109,
-      "grad_norm": 0.0024567090440541506,
-      "learning_rate": 0.0001999977563099435,
-      "loss": 46.0,
-      "step": 13251
-    },
-    {
-      "epoch": 2.1341438866298965,
-      "grad_norm": 0.006173844914883375,
-      "learning_rate": 0.00019999775597103159,
-      "loss": 46.0,
-      "step": 13252
-    },
-    {
-      "epoch": 2.134304923708684,
-      "grad_norm": 0.0030623492784798145,
-      "learning_rate": 0.00019999775563209408,
-      "loss": 46.0,
-      "step": 13253
-    },
-    {
-      "epoch": 2.1344659607874714,
-      "grad_norm": 0.0032281617168337107,
-      "learning_rate": 0.00019999775529313095,
-      "loss": 46.0,
-      "step": 13254
-    },
-    {
-      "epoch": 2.134626997866259,
-      "grad_norm": 0.00297993840649724,
-      "learning_rate": 0.00019999775495414227,
-      "loss": 46.0,
-      "step": 13255
-    },
-    {
-      "epoch": 2.1347880349450463,
-      "grad_norm": 0.0069718980230391026,
-      "learning_rate": 0.00019999775461512794,
-      "loss": 46.0,
-      "step": 13256
-    },
-    {
-      "epoch": 2.1349490720238333,
-      "grad_norm": 0.0014099756954237819,
-      "learning_rate": 0.00019999775427608806,
-      "loss": 46.0,
-      "step": 13257
-    },
-    {
-      "epoch": 2.135110109102621,
-      "grad_norm": 0.0018751703901216388,
-      "learning_rate": 0.00019999775393702256,
-      "loss": 46.0,
-      "step": 13258
-    },
-    {
-      "epoch": 2.1352711461814082,
-      "grad_norm": 0.008302632719278336,
-      "learning_rate": 0.00019999775359793147,
-      "loss": 46.0,
-      "step": 13259
-    },
-    {
-      "epoch": 2.1354321832601957,
-      "grad_norm": 0.000918448029551655,
-      "learning_rate": 0.0001999977532588148,
-      "loss": 46.0,
-      "step": 13260
-    },
-    {
-      "epoch": 2.135593220338983,
-      "grad_norm": 0.0009239473729394376,
-      "learning_rate": 0.00019999775291967253,
-      "loss": 46.0,
-      "step": 13261
-    },
-    {
-      "epoch": 2.1357542574177706,
-      "grad_norm": 0.012068535201251507,
-      "learning_rate": 0.00019999775258050466,
-      "loss": 46.0,
-      "step": 13262
-    },
-    {
-      "epoch": 2.135915294496558,
-      "grad_norm": 0.003447233932092786,
-      "learning_rate": 0.0001999977522413112,
-      "loss": 46.0,
-      "step": 13263
-    },
-    {
-      "epoch": 2.136076331575345,
-      "grad_norm": 0.001390350516885519,
-      "learning_rate": 0.00019999775190209215,
-      "loss": 46.0,
-      "step": 13264
-    },
-    {
-      "epoch": 2.1362373686541325,
-      "grad_norm": 0.009289112873375416,
-      "learning_rate": 0.0001999977515628475,
-      "loss": 46.0,
-      "step": 13265
-    },
-    {
-      "epoch": 2.13639840573292,
-      "grad_norm": 0.0019515432650223374,
-      "learning_rate": 0.00019999775122357726,
-      "loss": 46.0,
-      "step": 13266
-    },
-    {
-      "epoch": 2.1365594428117074,
-      "grad_norm": 0.005900282878428698,
-      "learning_rate": 0.00019999775088428142,
-      "loss": 46.0,
-      "step": 13267
-    },
-    {
-      "epoch": 2.136720479890495,
-      "grad_norm": 0.0018689382122829556,
-      "learning_rate": 0.00019999775054496,
-      "loss": 46.0,
-      "step": 13268
-    },
-    {
-      "epoch": 2.1368815169692823,
-      "grad_norm": 0.0006710558081977069,
-      "learning_rate": 0.00019999775020561295,
-      "loss": 46.0,
-      "step": 13269
-    },
-    {
-      "epoch": 2.1370425540480698,
-      "grad_norm": 0.007788967806845903,
-      "learning_rate": 0.00019999774986624033,
-      "loss": 46.0,
-      "step": 13270
-    },
-    {
-      "epoch": 2.1372035911268568,
-      "grad_norm": 0.0011311451671645045,
-      "learning_rate": 0.0001999977495268421,
-      "loss": 46.0,
-      "step": 13271
-    },
-    {
-      "epoch": 2.137364628205644,
-      "grad_norm": 0.003137932624667883,
-      "learning_rate": 0.0001999977491874183,
-      "loss": 46.0,
-      "step": 13272
-    },
-    {
-      "epoch": 2.1375256652844317,
-      "grad_norm": 0.004380833823233843,
-      "learning_rate": 0.00019999774884796892,
-      "loss": 46.0,
-      "step": 13273
-    },
-    {
-      "epoch": 2.137686702363219,
-      "grad_norm": 0.008016890846192837,
-      "learning_rate": 0.00019999774850849392,
-      "loss": 46.0,
-      "step": 13274
-    },
-    {
-      "epoch": 2.1378477394420066,
-      "grad_norm": 0.0038034694734960794,
-      "learning_rate": 0.00019999774816899333,
-      "loss": 46.0,
-      "step": 13275
-    },
-    {
-      "epoch": 2.138008776520794,
-      "grad_norm": 0.003832681104540825,
-      "learning_rate": 0.00019999774782946715,
-      "loss": 46.0,
-      "step": 13276
-    },
-    {
-      "epoch": 2.1381698135995815,
-      "grad_norm": 0.0004931480507366359,
-      "learning_rate": 0.00019999774748991533,
-      "loss": 46.0,
-      "step": 13277
-    },
-    {
-      "epoch": 2.1383308506783685,
-      "grad_norm": 0.0018774536438286304,
-      "learning_rate": 0.00019999774715033798,
-      "loss": 46.0,
-      "step": 13278
-    },
-    {
-      "epoch": 2.138491887757156,
-      "grad_norm": 0.007036134134978056,
-      "learning_rate": 0.000199997746810735,
-      "loss": 46.0,
-      "step": 13279
-    },
-    {
-      "epoch": 2.1386529248359434,
-      "grad_norm": 0.0035310224629938602,
-      "learning_rate": 0.00019999774647110643,
-      "loss": 46.0,
-      "step": 13280
-    },
-    {
-      "epoch": 2.138813961914731,
-      "grad_norm": 0.005686099175363779,
-      "learning_rate": 0.00019999774613145227,
-      "loss": 46.0,
-      "step": 13281
-    },
-    {
-      "epoch": 2.1389749989935183,
-      "grad_norm": 0.01061179954558611,
-      "learning_rate": 0.00019999774579177254,
-      "loss": 46.0,
-      "step": 13282
-    },
-    {
-      "epoch": 2.1391360360723057,
-      "grad_norm": 0.005625057499855757,
-      "learning_rate": 0.00019999774545206717,
-      "loss": 46.0,
-      "step": 13283
-    },
-    {
-      "epoch": 2.139297073151093,
-      "grad_norm": 0.005064083728939295,
-      "learning_rate": 0.00019999774511233624,
-      "loss": 46.0,
-      "step": 13284
-    },
-    {
-      "epoch": 2.1394581102298806,
-      "grad_norm": 0.0027127470821142197,
-      "learning_rate": 0.0001999977447725797,
-      "loss": 46.0,
-      "step": 13285
-    },
-    {
-      "epoch": 2.1396191473086676,
-      "grad_norm": 0.0013896109303459525,
-      "learning_rate": 0.00019999774443279757,
-      "loss": 46.0,
-      "step": 13286
-    },
-    {
-      "epoch": 2.139780184387455,
-      "grad_norm": 0.001895049586892128,
-      "learning_rate": 0.00019999774409298985,
-      "loss": 46.0,
-      "step": 13287
-    },
-    {
-      "epoch": 2.1399412214662425,
-      "grad_norm": 0.0008827557321637869,
-      "learning_rate": 0.00019999774375315652,
-      "loss": 46.0,
-      "step": 13288
-    },
-    {
-      "epoch": 2.14010225854503,
-      "grad_norm": 0.0012584917712956667,
-      "learning_rate": 0.00019999774341329763,
-      "loss": 46.0,
-      "step": 13289
-    },
-    {
-      "epoch": 2.1402632956238175,
-      "grad_norm": 0.0017334287986159325,
-      "learning_rate": 0.0001999977430734131,
-      "loss": 46.0,
-      "step": 13290
-    },
-    {
-      "epoch": 2.140424332702605,
-      "grad_norm": 0.009314744733273983,
-      "learning_rate": 0.000199997742733503,
-      "loss": 46.0,
-      "step": 13291
-    },
-    {
-      "epoch": 2.1405853697813924,
-      "grad_norm": 0.0022184152621775866,
-      "learning_rate": 0.00019999774239356732,
-      "loss": 46.0,
-      "step": 13292
-    },
-    {
-      "epoch": 2.1407464068601794,
-      "grad_norm": 0.011970209889113903,
-      "learning_rate": 0.00019999774205360602,
-      "loss": 46.0,
-      "step": 13293
-    },
-    {
-      "epoch": 2.140907443938967,
-      "grad_norm": 0.0012183836661279202,
-      "learning_rate": 0.00019999774171361914,
-      "loss": 46.0,
-      "step": 13294
-    },
-    {
-      "epoch": 2.1410684810177543,
-      "grad_norm": 0.0015864853048697114,
-      "learning_rate": 0.00019999774137360667,
-      "loss": 46.0,
-      "step": 13295
-    },
-    {
-      "epoch": 2.1412295180965417,
-      "grad_norm": 0.0019399285083636642,
-      "learning_rate": 0.0001999977410335686,
-      "loss": 46.0,
-      "step": 13296
-    },
-    {
-      "epoch": 2.141390555175329,
-      "grad_norm": 0.0019598123617470264,
-      "learning_rate": 0.00019999774069350492,
-      "loss": 46.0,
-      "step": 13297
-    },
-    {
-      "epoch": 2.1415515922541166,
-      "grad_norm": 0.0034518218599259853,
-      "learning_rate": 0.00019999774035341566,
-      "loss": 46.0,
-      "step": 13298
-    },
-    {
-      "epoch": 2.141712629332904,
-      "grad_norm": 0.005137486383318901,
-      "learning_rate": 0.00019999774001330082,
-      "loss": 46.0,
-      "step": 13299
-    },
-    {
-      "epoch": 2.1418736664116915,
-      "grad_norm": 0.0024593959096819162,
-      "learning_rate": 0.00019999773967316036,
-      "loss": 46.0,
-      "step": 13300
-    },
-    {
-      "epoch": 2.1420347034904785,
-      "grad_norm": 0.002304528374224901,
-      "learning_rate": 0.00019999773933299434,
-      "loss": 46.0,
-      "step": 13301
-    },
-    {
-      "epoch": 2.142195740569266,
-      "grad_norm": 0.0008896937943063676,
-      "learning_rate": 0.00019999773899280268,
-      "loss": 46.0,
-      "step": 13302
-    },
-    {
-      "epoch": 2.1423567776480534,
-      "grad_norm": 0.0025536175817251205,
-      "learning_rate": 0.00019999773865258543,
-      "loss": 46.0,
-      "step": 13303
-    },
-    {
-      "epoch": 2.142517814726841,
-      "grad_norm": 0.0036763439420610666,
-      "learning_rate": 0.00019999773831234262,
-      "loss": 46.0,
-      "step": 13304
-    },
-    {
-      "epoch": 2.1426788518056283,
-      "grad_norm": 0.007506121415644884,
-      "learning_rate": 0.0001999977379720742,
-      "loss": 46.0,
-      "step": 13305
-    },
-    {
-      "epoch": 2.142839888884416,
-      "grad_norm": 0.0028328776825219393,
-      "learning_rate": 0.0001999977376317802,
-      "loss": 46.0,
-      "step": 13306
-    },
-    {
-      "epoch": 2.143000925963203,
-      "grad_norm": 0.0006790681509301066,
-      "learning_rate": 0.0001999977372914606,
-      "loss": 46.0,
-      "step": 13307
-    },
-    {
-      "epoch": 2.1431619630419902,
-      "grad_norm": 0.0014149787602946162,
-      "learning_rate": 0.00019999773695111536,
-      "loss": 46.0,
-      "step": 13308
-    },
-    {
-      "epoch": 2.1433230001207777,
-      "grad_norm": 0.0022059523034840822,
-      "learning_rate": 0.00019999773661074456,
-      "loss": 46.0,
-      "step": 13309
-    },
-    {
-      "epoch": 2.143484037199565,
-      "grad_norm": 0.001987648429349065,
-      "learning_rate": 0.0001999977362703482,
-      "loss": 46.0,
-      "step": 13310
-    },
-    {
-      "epoch": 2.1436450742783526,
-      "grad_norm": 0.00417117727920413,
-      "learning_rate": 0.0001999977359299262,
-      "loss": 46.0,
-      "step": 13311
-    },
-    {
-      "epoch": 2.14380611135714,
-      "grad_norm": 0.0023212896194308996,
-      "learning_rate": 0.00019999773558947861,
-      "loss": 46.0,
-      "step": 13312
-    },
-    {
-      "epoch": 2.1439671484359275,
-      "grad_norm": 0.0011835447512567043,
-      "learning_rate": 0.00019999773524900544,
-      "loss": 46.0,
-      "step": 13313
-    },
-    {
-      "epoch": 2.144128185514715,
-      "grad_norm": 0.001484884531237185,
-      "learning_rate": 0.00019999773490850665,
-      "loss": 46.0,
-      "step": 13314
-    },
-    {
-      "epoch": 2.144289222593502,
-      "grad_norm": 0.0009144832729361951,
-      "learning_rate": 0.0001999977345679823,
-      "loss": 46.0,
-      "step": 13315
-    },
-    {
-      "epoch": 2.1444502596722894,
-      "grad_norm": 0.002882937900722027,
-      "learning_rate": 0.00019999773422743234,
-      "loss": 46.0,
-      "step": 13316
-    },
-    {
-      "epoch": 2.144611296751077,
-      "grad_norm": 0.005968336947262287,
-      "learning_rate": 0.0001999977338868568,
-      "loss": 46.0,
-      "step": 13317
-    },
-    {
-      "epoch": 2.1447723338298643,
-      "grad_norm": 0.0036911708302795887,
-      "learning_rate": 0.00019999773354625563,
-      "loss": 46.0,
-      "step": 13318
-    },
-    {
-      "epoch": 2.1449333709086518,
-      "grad_norm": 0.0036230271216481924,
-      "learning_rate": 0.0001999977332056289,
-      "loss": 46.0,
-      "step": 13319
-    },
-    {
-      "epoch": 2.1450944079874392,
-      "grad_norm": 0.006205302197486162,
-      "learning_rate": 0.00019999773286497657,
-      "loss": 46.0,
-      "step": 13320
-    },
-    {
-      "epoch": 2.1452554450662267,
-      "grad_norm": 0.0023907211143523455,
-      "learning_rate": 0.00019999773252429864,
-      "loss": 46.0,
-      "step": 13321
-    },
-    {
-      "epoch": 2.1454164821450137,
-      "grad_norm": 0.0029476932249963284,
-      "learning_rate": 0.00019999773218359513,
-      "loss": 46.0,
-      "step": 13322
-    },
-    {
-      "epoch": 2.145577519223801,
-      "grad_norm": 0.006936867721378803,
-      "learning_rate": 0.00019999773184286597,
-      "loss": 46.0,
-      "step": 13323
-    },
-    {
-      "epoch": 2.1457385563025886,
-      "grad_norm": 0.008803711272776127,
-      "learning_rate": 0.00019999773150211126,
-      "loss": 46.0,
-      "step": 13324
-    },
-    {
-      "epoch": 2.145899593381376,
-      "grad_norm": 0.0010045906528830528,
-      "learning_rate": 0.000199997731161331,
-      "loss": 46.0,
-      "step": 13325
-    },
-    {
-      "epoch": 2.1460606304601635,
-      "grad_norm": 0.0005345331155695021,
-      "learning_rate": 0.00019999773082052507,
-      "loss": 46.0,
-      "step": 13326
-    },
-    {
-      "epoch": 2.146221667538951,
-      "grad_norm": 0.002014959929510951,
-      "learning_rate": 0.00019999773047969357,
-      "loss": 46.0,
-      "step": 13327
-    },
-    {
-      "epoch": 2.1463827046177384,
-      "grad_norm": 0.0044423057697713375,
-      "learning_rate": 0.00019999773013883648,
-      "loss": 46.0,
-      "step": 13328
-    },
-    {
-      "epoch": 2.146543741696526,
-      "grad_norm": 0.003911557141691446,
-      "learning_rate": 0.0001999977297979538,
-      "loss": 46.0,
-      "step": 13329
-    },
-    {
-      "epoch": 2.146704778775313,
-      "grad_norm": 0.0013346080668270588,
-      "learning_rate": 0.00019999772945704554,
-      "loss": 46.0,
-      "step": 13330
-    },
-    {
-      "epoch": 2.1468658158541003,
-      "grad_norm": 0.003525221487507224,
-      "learning_rate": 0.00019999772911611166,
-      "loss": 46.0,
-      "step": 13331
-    },
-    {
-      "epoch": 2.1470268529328878,
-      "grad_norm": 0.00832782220095396,
-      "learning_rate": 0.0001999977287751522,
-      "loss": 46.0,
-      "step": 13332
-    },
-    {
-      "epoch": 2.147187890011675,
-      "grad_norm": 0.0008926745504140854,
-      "learning_rate": 0.00019999772843416714,
-      "loss": 46.0,
-      "step": 13333
-    },
-    {
-      "epoch": 2.1473489270904627,
-      "grad_norm": 0.002892832038924098,
-      "learning_rate": 0.00019999772809315647,
-      "loss": 46.0,
-      "step": 13334
-    },
-    {
-      "epoch": 2.14750996416925,
-      "grad_norm": 0.003158863401040435,
-      "learning_rate": 0.00019999772775212022,
-      "loss": 46.0,
-      "step": 13335
-    },
-    {
-      "epoch": 2.1476710012480376,
-      "grad_norm": 0.0020852971356362104,
-      "learning_rate": 0.00019999772741105838,
-      "loss": 46.0,
-      "step": 13336
-    },
-    {
-      "epoch": 2.1478320383268246,
-      "grad_norm": 0.0033024419099092484,
-      "learning_rate": 0.00019999772706997092,
-      "loss": 46.0,
-      "step": 13337
-    },
-    {
-      "epoch": 2.147993075405612,
-      "grad_norm": 0.005663240794092417,
-      "learning_rate": 0.00019999772672885793,
-      "loss": 46.0,
-      "step": 13338
-    },
-    {
-      "epoch": 2.1481541124843995,
-      "grad_norm": 0.0026012801099568605,
-      "learning_rate": 0.0001999977263877193,
-      "loss": 46.0,
-      "step": 13339
-    },
-    {
-      "epoch": 2.148315149563187,
-      "grad_norm": 0.002319332677870989,
-      "learning_rate": 0.00019999772604655506,
-      "loss": 46.0,
-      "step": 13340
-    },
-    {
-      "epoch": 2.1484761866419744,
-      "grad_norm": 0.005728776566684246,
-      "learning_rate": 0.00019999772570536526,
-      "loss": 46.0,
-      "step": 13341
-    },
-    {
-      "epoch": 2.148637223720762,
-      "grad_norm": 0.0020345617085695267,
-      "learning_rate": 0.00019999772536414984,
-      "loss": 46.0,
-      "step": 13342
-    },
-    {
-      "epoch": 2.1487982607995493,
-      "grad_norm": 0.0026169035118073225,
-      "learning_rate": 0.00019999772502290883,
-      "loss": 46.0,
-      "step": 13343
-    },
-    {
-      "epoch": 2.1489592978783363,
-      "grad_norm": 0.003275137161836028,
-      "learning_rate": 0.00019999772468164224,
-      "loss": 46.0,
-      "step": 13344
-    },
-    {
-      "epoch": 2.1491203349571237,
-      "grad_norm": 0.004151905421167612,
-      "learning_rate": 0.00019999772434035003,
-      "loss": 46.0,
-      "step": 13345
-    },
-    {
-      "epoch": 2.149281372035911,
-      "grad_norm": 0.006651798263192177,
-      "learning_rate": 0.00019999772399903226,
-      "loss": 46.0,
-      "step": 13346
-    },
-    {
-      "epoch": 2.1494424091146986,
-      "grad_norm": 0.003913423046469688,
-      "learning_rate": 0.00019999772365768888,
-      "loss": 46.0,
-      "step": 13347
-    },
-    {
-      "epoch": 2.149603446193486,
-      "grad_norm": 0.0009208098636008799,
-      "learning_rate": 0.0001999977233163199,
-      "loss": 46.0,
-      "step": 13348
-    },
-    {
-      "epoch": 2.1497644832722735,
-      "grad_norm": 0.003083003219217062,
-      "learning_rate": 0.00019999772297492533,
-      "loss": 46.0,
-      "step": 13349
-    },
-    {
-      "epoch": 2.149925520351061,
-      "grad_norm": 0.0050685168243944645,
-      "learning_rate": 0.00019999772263350516,
-      "loss": 46.0,
-      "step": 13350
-    },
-    {
-      "epoch": 2.150086557429848,
-      "grad_norm": 0.014121463522315025,
-      "learning_rate": 0.0001999977222920594,
-      "loss": 46.0,
-      "step": 13351
-    },
-    {
-      "epoch": 2.1502475945086355,
-      "grad_norm": 0.0011735630687326193,
-      "learning_rate": 0.00019999772195058806,
-      "loss": 46.0,
-      "step": 13352
-    },
-    {
-      "epoch": 2.150408631587423,
-      "grad_norm": 0.005555521231144667,
-      "learning_rate": 0.0001999977216090911,
-      "loss": 46.0,
-      "step": 13353
-    },
-    {
-      "epoch": 2.1505696686662104,
-      "grad_norm": 0.001185104250907898,
-      "learning_rate": 0.00019999772126756858,
-      "loss": 46.0,
-      "step": 13354
-    },
-    {
-      "epoch": 2.150730705744998,
-      "grad_norm": 0.0030374207999557257,
-      "learning_rate": 0.00019999772092602042,
-      "loss": 46.0,
-      "step": 13355
-    },
-    {
-      "epoch": 2.1508917428237853,
-      "grad_norm": 0.002993746427819133,
-      "learning_rate": 0.0001999977205844467,
-      "loss": 46.0,
-      "step": 13356
-    },
-    {
-      "epoch": 2.1510527799025727,
-      "grad_norm": 0.0034053742419928312,
-      "learning_rate": 0.00019999772024284737,
-      "loss": 46.0,
-      "step": 13357
-    },
-    {
-      "epoch": 2.15121381698136,
-      "grad_norm": 0.003543669590726495,
-      "learning_rate": 0.00019999771990122245,
-      "loss": 46.0,
-      "step": 13358
-    },
-    {
-      "epoch": 2.151374854060147,
-      "grad_norm": 0.0017898465739563107,
-      "learning_rate": 0.00019999771955957196,
-      "loss": 46.0,
-      "step": 13359
-    },
-    {
-      "epoch": 2.1515358911389346,
-      "grad_norm": 0.0029814443551003933,
-      "learning_rate": 0.00019999771921789584,
-      "loss": 46.0,
-      "step": 13360
-    },
-    {
-      "epoch": 2.151696928217722,
-      "grad_norm": 0.0021095196716487408,
-      "learning_rate": 0.00019999771887619413,
-      "loss": 46.0,
-      "step": 13361
-    },
-    {
-      "epoch": 2.1518579652965095,
-      "grad_norm": 0.0015045434702187777,
-      "learning_rate": 0.00019999771853446686,
-      "loss": 46.0,
-      "step": 13362
-    },
-    {
-      "epoch": 2.152019002375297,
-      "grad_norm": 0.00288623059168458,
-      "learning_rate": 0.00019999771819271395,
-      "loss": 46.0,
-      "step": 13363
-    },
-    {
-      "epoch": 2.1521800394540844,
-      "grad_norm": 0.0014310912229120731,
-      "learning_rate": 0.00019999771785093548,
-      "loss": 46.0,
-      "step": 13364
-    },
-    {
-      "epoch": 2.152341076532872,
-      "grad_norm": 0.0014442679239436984,
-      "learning_rate": 0.0001999977175091314,
-      "loss": 46.0,
-      "step": 13365
-    },
-    {
-      "epoch": 2.152502113611659,
-      "grad_norm": 0.004888609517365694,
-      "learning_rate": 0.00019999771716730172,
-      "loss": 46.0,
-      "step": 13366
-    },
-    {
-      "epoch": 2.1526631506904463,
-      "grad_norm": 0.005491163115948439,
-      "learning_rate": 0.00019999771682544646,
-      "loss": 46.0,
-      "step": 13367
-    },
-    {
-      "epoch": 2.152824187769234,
-      "grad_norm": 0.0019303741864860058,
-      "learning_rate": 0.00019999771648356558,
-      "loss": 46.0,
-      "step": 13368
-    },
-    {
-      "epoch": 2.1529852248480212,
-      "grad_norm": 0.015973171219229698,
-      "learning_rate": 0.00019999771614165912,
-      "loss": 46.0,
-      "step": 13369
-    },
-    {
-      "epoch": 2.1531462619268087,
-      "grad_norm": 0.003926645498722792,
-      "learning_rate": 0.00019999771579972707,
-      "loss": 46.0,
-      "step": 13370
-    },
-    {
-      "epoch": 2.153307299005596,
-      "grad_norm": 0.0033199922181665897,
-      "learning_rate": 0.00019999771545776943,
-      "loss": 46.0,
-      "step": 13371
-    },
-    {
-      "epoch": 2.1534683360843836,
-      "grad_norm": 0.013160881586372852,
-      "learning_rate": 0.0001999977151157862,
-      "loss": 46.0,
-      "step": 13372
-    },
-    {
-      "epoch": 2.153629373163171,
-      "grad_norm": 0.0035904760006815195,
-      "learning_rate": 0.00019999771477377735,
-      "loss": 46.0,
-      "step": 13373
-    },
-    {
-      "epoch": 2.153790410241958,
-      "grad_norm": 0.007611855398863554,
-      "learning_rate": 0.00019999771443174295,
-      "loss": 46.0,
-      "step": 13374
-    },
-    {
-      "epoch": 2.1539514473207455,
-      "grad_norm": 0.004493820481002331,
-      "learning_rate": 0.00019999771408968294,
-      "loss": 46.0,
-      "step": 13375
-    },
-    {
-      "epoch": 2.154112484399533,
-      "grad_norm": 0.0066480860114097595,
-      "learning_rate": 0.0001999977137475973,
-      "loss": 46.0,
-      "step": 13376
-    },
-    {
-      "epoch": 2.1542735214783204,
-      "grad_norm": 0.0009913266403600574,
-      "learning_rate": 0.0001999977134054861,
-      "loss": 46.0,
-      "step": 13377
-    },
-    {
-      "epoch": 2.154434558557108,
-      "grad_norm": 0.0014127322938293219,
-      "learning_rate": 0.0001999977130633493,
-      "loss": 46.0,
-      "step": 13378
-    },
-    {
-      "epoch": 2.1545955956358953,
-      "grad_norm": 0.001301866490393877,
-      "learning_rate": 0.0001999977127211869,
-      "loss": 46.0,
-      "step": 13379
-    },
-    {
-      "epoch": 2.1547566327146828,
-      "grad_norm": 0.008188326843082905,
-      "learning_rate": 0.00019999771237899888,
-      "loss": 46.0,
-      "step": 13380
-    },
-    {
-      "epoch": 2.1549176697934698,
-      "grad_norm": 0.0006416022079065442,
-      "learning_rate": 0.0001999977120367853,
-      "loss": 46.0,
-      "step": 13381
-    },
-    {
-      "epoch": 2.155078706872257,
-      "grad_norm": 0.000580929743591696,
-      "learning_rate": 0.00019999771169454614,
-      "loss": 46.0,
-      "step": 13382
-    },
-    {
-      "epoch": 2.1552397439510447,
-      "grad_norm": 0.0005832845345139503,
-      "learning_rate": 0.00019999771135228135,
-      "loss": 46.0,
-      "step": 13383
-    },
-    {
-      "epoch": 2.155400781029832,
-      "grad_norm": 0.0024417205713689327,
-      "learning_rate": 0.000199997711009991,
-      "loss": 46.0,
-      "step": 13384
-    },
-    {
-      "epoch": 2.1555618181086196,
-      "grad_norm": 0.008255396038293839,
-      "learning_rate": 0.00019999771066767503,
-      "loss": 46.0,
-      "step": 13385
-    },
-    {
-      "epoch": 2.155722855187407,
-      "grad_norm": 0.0010429261019453406,
-      "learning_rate": 0.00019999771032533348,
-      "loss": 46.0,
-      "step": 13386
-    },
-    {
-      "epoch": 2.1558838922661945,
-      "grad_norm": 0.005211732350289822,
-      "learning_rate": 0.00019999770998296631,
-      "loss": 46.0,
-      "step": 13387
-    },
-    {
-      "epoch": 2.1560449293449815,
-      "grad_norm": 0.0020214738324284554,
-      "learning_rate": 0.0001999977096405736,
-      "loss": 46.0,
-      "step": 13388
-    },
-    {
-      "epoch": 2.156205966423769,
-      "grad_norm": 0.001470905146561563,
-      "learning_rate": 0.00019999770929815522,
-      "loss": 46.0,
-      "step": 13389
-    },
-    {
-      "epoch": 2.1563670035025564,
-      "grad_norm": 0.001472577452659607,
-      "learning_rate": 0.0001999977089557113,
-      "loss": 46.0,
-      "step": 13390
-    },
-    {
-      "epoch": 2.156528040581344,
-      "grad_norm": 0.005086067132651806,
-      "learning_rate": 0.00019999770861324175,
-      "loss": 46.0,
-      "step": 13391
-    },
-    {
-      "epoch": 2.1566890776601313,
-      "grad_norm": 0.0018468507332727313,
-      "learning_rate": 0.00019999770827074665,
-      "loss": 46.0,
-      "step": 13392
-    },
-    {
-      "epoch": 2.1568501147389187,
-      "grad_norm": 0.002536970656365156,
-      "learning_rate": 0.00019999770792822593,
-      "loss": 46.0,
-      "step": 13393
-    },
-    {
-      "epoch": 2.157011151817706,
-      "grad_norm": 0.006078988313674927,
-      "learning_rate": 0.0001999977075856796,
-      "loss": 46.0,
-      "step": 13394
-    },
-    {
-      "epoch": 2.157172188896493,
-      "grad_norm": 0.0013384041376411915,
-      "learning_rate": 0.0001999977072431077,
-      "loss": 46.0,
-      "step": 13395
-    },
-    {
-      "epoch": 2.1573332259752807,
-      "grad_norm": 0.0009293656330555677,
-      "learning_rate": 0.0001999977069005102,
-      "loss": 46.0,
-      "step": 13396
-    },
-    {
-      "epoch": 2.157494263054068,
-      "grad_norm": 0.0025627962313592434,
-      "learning_rate": 0.0001999977065578871,
-      "loss": 46.0,
-      "step": 13397
-    },
-    {
-      "epoch": 2.1576553001328556,
-      "grad_norm": 0.005264953710138798,
-      "learning_rate": 0.0001999977062152384,
-      "loss": 46.0,
-      "step": 13398
-    },
-    {
-      "epoch": 2.157816337211643,
-      "grad_norm": 0.0017437953501939774,
-      "learning_rate": 0.0001999977058725641,
-      "loss": 46.0,
-      "step": 13399
-    },
-    {
-      "epoch": 2.1579773742904305,
-      "grad_norm": 0.0018247002735733986,
-      "learning_rate": 0.00019999770552986426,
-      "loss": 46.0,
-      "step": 13400
-    },
-    {
-      "epoch": 2.158138411369218,
-      "grad_norm": 0.0029213817324489355,
-      "learning_rate": 0.0001999977051871388,
-      "loss": 46.0,
-      "step": 13401
-    },
-    {
-      "epoch": 2.1582994484480054,
-      "grad_norm": 0.002560011576861143,
-      "learning_rate": 0.0001999977048443877,
-      "loss": 46.0,
-      "step": 13402
-    },
-    {
-      "epoch": 2.1584604855267924,
-      "grad_norm": 0.0017181907314807177,
-      "learning_rate": 0.00019999770450161104,
-      "loss": 46.0,
-      "step": 13403
-    },
-    {
-      "epoch": 2.15862152260558,
-      "grad_norm": 0.0032093869522213936,
-      "learning_rate": 0.0001999977041588088,
-      "loss": 46.0,
-      "step": 13404
-    },
-    {
-      "epoch": 2.1587825596843673,
-      "grad_norm": 0.0010084158275276423,
-      "learning_rate": 0.00019999770381598096,
-      "loss": 46.0,
-      "step": 13405
-    },
-    {
-      "epoch": 2.1589435967631547,
-      "grad_norm": 0.00312529387883842,
-      "learning_rate": 0.0001999977034731275,
-      "loss": 46.0,
-      "step": 13406
-    },
-    {
-      "epoch": 2.159104633841942,
-      "grad_norm": 0.0033588879741728306,
-      "learning_rate": 0.00019999770313024846,
-      "loss": 46.0,
-      "step": 13407
-    },
-    {
-      "epoch": 2.1592656709207296,
-      "grad_norm": 0.00045731355203315616,
-      "learning_rate": 0.00019999770278734383,
-      "loss": 46.0,
-      "step": 13408
-    },
-    {
-      "epoch": 2.159426707999517,
-      "grad_norm": 0.0008029004675336182,
-      "learning_rate": 0.0001999977024444136,
-      "loss": 46.0,
-      "step": 13409
-    },
-    {
-      "epoch": 2.159587745078304,
-      "grad_norm": 0.0038672753144055605,
-      "learning_rate": 0.00019999770210145777,
-      "loss": 46.0,
-      "step": 13410
-    },
-    {
-      "epoch": 2.1597487821570915,
-      "grad_norm": 0.0004022162174805999,
-      "learning_rate": 0.00019999770175847638,
-      "loss": 46.0,
-      "step": 13411
-    },
-    {
-      "epoch": 2.159909819235879,
-      "grad_norm": 0.0070533487014472485,
-      "learning_rate": 0.00019999770141546937,
-      "loss": 46.0,
-      "step": 13412
-    },
-    {
-      "epoch": 2.1600708563146664,
-      "grad_norm": 0.000979344709776342,
-      "learning_rate": 0.00019999770107243675,
-      "loss": 46.0,
-      "step": 13413
-    },
-    {
-      "epoch": 2.160231893393454,
-      "grad_norm": 0.0010132818715646863,
-      "learning_rate": 0.00019999770072937856,
-      "loss": 46.0,
-      "step": 13414
-    },
-    {
-      "epoch": 2.1603929304722413,
-      "grad_norm": 0.005230887793004513,
-      "learning_rate": 0.00019999770038629477,
-      "loss": 46.0,
-      "step": 13415
-    },
-    {
-      "epoch": 2.160553967551029,
-      "grad_norm": 0.0013651868794113398,
-      "learning_rate": 0.00019999770004318536,
-      "loss": 46.0,
-      "step": 13416
-    },
-    {
-      "epoch": 2.1607150046298162,
-      "grad_norm": 0.001582533004693687,
-      "learning_rate": 0.0001999976997000504,
-      "loss": 46.0,
-      "step": 13417
-    },
-    {
-      "epoch": 2.1608760417086033,
-      "grad_norm": 0.004795411601662636,
-      "learning_rate": 0.0001999976993568898,
-      "loss": 46.0,
-      "step": 13418
-    },
-    {
-      "epoch": 2.1610370787873907,
-      "grad_norm": 0.0022396952845156193,
-      "learning_rate": 0.00019999769901370365,
-      "loss": 46.0,
-      "step": 13419
-    },
-    {
-      "epoch": 2.161198115866178,
-      "grad_norm": 0.0018817392410710454,
-      "learning_rate": 0.0001999976986704919,
-      "loss": 46.0,
-      "step": 13420
-    },
-    {
-      "epoch": 2.1613591529449656,
-      "grad_norm": 0.0008986614411696792,
-      "learning_rate": 0.00019999769832725455,
-      "loss": 46.0,
-      "step": 13421
-    },
-    {
-      "epoch": 2.161520190023753,
-      "grad_norm": 0.0010536988265812397,
-      "learning_rate": 0.00019999769798399158,
-      "loss": 46.0,
-      "step": 13422
-    },
-    {
-      "epoch": 2.1616812271025405,
-      "grad_norm": 0.003245594911277294,
-      "learning_rate": 0.000199997697640703,
-      "loss": 46.0,
-      "step": 13423
-    },
-    {
-      "epoch": 2.1618422641813275,
-      "grad_norm": 0.0024706011172384024,
-      "learning_rate": 0.00019999769729738887,
-      "loss": 46.0,
-      "step": 13424
-    },
-    {
-      "epoch": 2.162003301260115,
-      "grad_norm": 0.0007533066673204303,
-      "learning_rate": 0.00019999769695404915,
-      "loss": 46.0,
-      "step": 13425
-    },
-    {
-      "epoch": 2.1621643383389024,
-      "grad_norm": 0.004144506994634867,
-      "learning_rate": 0.00019999769661068384,
-      "loss": 46.0,
-      "step": 13426
-    },
-    {
-      "epoch": 2.16232537541769,
-      "grad_norm": 0.00324800331145525,
-      "learning_rate": 0.0001999976962672929,
-      "loss": 46.0,
-      "step": 13427
-    },
-    {
-      "epoch": 2.1624864124964773,
-      "grad_norm": 0.013405922800302505,
-      "learning_rate": 0.00019999769592387638,
-      "loss": 46.0,
-      "step": 13428
-    },
-    {
-      "epoch": 2.162647449575265,
-      "grad_norm": 0.0038236540276557207,
-      "learning_rate": 0.00019999769558043425,
-      "loss": 46.0,
-      "step": 13429
-    },
-    {
-      "epoch": 2.1628084866540522,
-      "grad_norm": 0.006123604718595743,
-      "learning_rate": 0.00019999769523696656,
-      "loss": 46.0,
-      "step": 13430
-    },
-    {
-      "epoch": 2.1629695237328397,
-      "grad_norm": 0.0027882412541657686,
-      "learning_rate": 0.00019999769489347326,
-      "loss": 46.0,
-      "step": 13431
-    },
-    {
-      "epoch": 2.1631305608116267,
-      "grad_norm": 0.0015664227539673448,
-      "learning_rate": 0.00019999769454995438,
-      "loss": 46.0,
-      "step": 13432
-    },
-    {
-      "epoch": 2.163291597890414,
-      "grad_norm": 0.010436464101076126,
-      "learning_rate": 0.00019999769420640988,
-      "loss": 46.0,
-      "step": 13433
-    },
-    {
-      "epoch": 2.1634526349692016,
-      "grad_norm": 0.0018199981423094869,
-      "learning_rate": 0.0001999976938628398,
-      "loss": 46.0,
-      "step": 13434
-    },
-    {
-      "epoch": 2.163613672047989,
-      "grad_norm": 0.00535723427310586,
-      "learning_rate": 0.0001999976935192441,
-      "loss": 46.0,
-      "step": 13435
-    },
-    {
-      "epoch": 2.1637747091267765,
-      "grad_norm": 0.0023336452431976795,
-      "learning_rate": 0.00019999769317562285,
-      "loss": 46.0,
-      "step": 13436
-    },
-    {
-      "epoch": 2.163935746205564,
-      "grad_norm": 0.0016138700302690268,
-      "learning_rate": 0.00019999769283197597,
-      "loss": 46.0,
-      "step": 13437
-    },
-    {
-      "epoch": 2.1640967832843514,
-      "grad_norm": 0.0008041071705520153,
-      "learning_rate": 0.00019999769248830353,
-      "loss": 46.0,
-      "step": 13438
-    },
-    {
-      "epoch": 2.1642578203631384,
-      "grad_norm": 0.001124918577261269,
-      "learning_rate": 0.00019999769214460545,
-      "loss": 46.0,
-      "step": 13439
-    },
-    {
-      "epoch": 2.164418857441926,
-      "grad_norm": 0.003743311855942011,
-      "learning_rate": 0.00019999769180088181,
-      "loss": 46.0,
-      "step": 13440
-    },
-    {
-      "epoch": 2.1645798945207133,
-      "grad_norm": 0.0021203504875302315,
-      "learning_rate": 0.00019999769145713256,
-      "loss": 46.0,
-      "step": 13441
-    },
-    {
-      "epoch": 2.1647409315995008,
-      "grad_norm": 0.0012205262901261449,
-      "learning_rate": 0.00019999769111335772,
-      "loss": 46.0,
-      "step": 13442
-    },
-    {
-      "epoch": 2.164901968678288,
-      "grad_norm": 0.011249815113842487,
-      "learning_rate": 0.0001999976907695573,
-      "loss": 46.0,
-      "step": 13443
-    },
-    {
-      "epoch": 2.1650630057570757,
-      "grad_norm": 0.002064640400931239,
-      "learning_rate": 0.00019999769042573128,
-      "loss": 46.0,
-      "step": 13444
-    },
-    {
-      "epoch": 2.165224042835863,
-      "grad_norm": 0.004764914512634277,
-      "learning_rate": 0.00019999769008187965,
-      "loss": 46.0,
-      "step": 13445
-    },
-    {
-      "epoch": 2.1653850799146506,
-      "grad_norm": 0.0018297492060810328,
-      "learning_rate": 0.00019999768973800243,
-      "loss": 46.0,
-      "step": 13446
-    },
-    {
-      "epoch": 2.1655461169934376,
-      "grad_norm": 0.004330933094024658,
-      "learning_rate": 0.00019999768939409963,
-      "loss": 46.0,
-      "step": 13447
-    },
-    {
-      "epoch": 2.165707154072225,
-      "grad_norm": 0.002149335341528058,
-      "learning_rate": 0.0001999976890501712,
-      "loss": 46.0,
-      "step": 13448
-    },
-    {
-      "epoch": 2.1658681911510125,
-      "grad_norm": 0.0019518863409757614,
-      "learning_rate": 0.00019999768870621723,
-      "loss": 46.0,
-      "step": 13449
-    },
-    {
-      "epoch": 2.1660292282298,
-      "grad_norm": 0.001130544114857912,
-      "learning_rate": 0.00019999768836223764,
-      "loss": 46.0,
-      "step": 13450
-    },
-    {
-      "epoch": 2.1661902653085874,
-      "grad_norm": 0.0018277255585417151,
-      "learning_rate": 0.00019999768801823246,
-      "loss": 46.0,
-      "step": 13451
-    },
-    {
-      "epoch": 2.166351302387375,
-      "grad_norm": 0.006490216590464115,
-      "learning_rate": 0.00019999768767420167,
-      "loss": 46.0,
-      "step": 13452
-    },
-    {
-      "epoch": 2.1665123394661623,
-      "grad_norm": 0.0028237171936780214,
-      "learning_rate": 0.0001999976873301453,
-      "loss": 46.0,
-      "step": 13453
-    },
-    {
-      "epoch": 2.1666733765449493,
-      "grad_norm": 0.0005406445707194507,
-      "learning_rate": 0.00019999768698606332,
-      "loss": 46.0,
-      "step": 13454
-    },
-    {
-      "epoch": 2.1668344136237367,
-      "grad_norm": 0.00112696154974401,
-      "learning_rate": 0.00019999768664195577,
-      "loss": 46.0,
-      "step": 13455
-    },
-    {
-      "epoch": 2.166995450702524,
-      "grad_norm": 0.003206856083124876,
-      "learning_rate": 0.0001999976862978226,
-      "loss": 46.0,
-      "step": 13456
-    },
-    {
-      "epoch": 2.1671564877813116,
-      "grad_norm": 0.0022405830677598715,
-      "learning_rate": 0.00019999768595366387,
-      "loss": 46.0,
-      "step": 13457
-    },
-    {
-      "epoch": 2.167317524860099,
-      "grad_norm": 0.0033505712635815144,
-      "learning_rate": 0.0001999976856094795,
-      "loss": 46.0,
-      "step": 13458
-    },
-    {
-      "epoch": 2.1674785619388865,
-      "grad_norm": 0.0035899514332413673,
-      "learning_rate": 0.00019999768526526957,
-      "loss": 46.0,
-      "step": 13459
-    },
-    {
-      "epoch": 2.167639599017674,
-      "grad_norm": 0.00091147655621171,
-      "learning_rate": 0.00019999768492103402,
-      "loss": 46.0,
-      "step": 13460
-    },
-    {
-      "epoch": 2.167800636096461,
-      "grad_norm": 0.0022993178572505713,
-      "learning_rate": 0.00019999768457677292,
-      "loss": 46.0,
-      "step": 13461
-    },
-    {
-      "epoch": 2.1679616731752485,
-      "grad_norm": 0.001613577944226563,
-      "learning_rate": 0.00019999768423248617,
-      "loss": 46.0,
-      "step": 13462
-    },
-    {
-      "epoch": 2.168122710254036,
-      "grad_norm": 0.0008811029838398099,
-      "learning_rate": 0.00019999768388817386,
-      "loss": 46.0,
-      "step": 13463
-    },
-    {
-      "epoch": 2.1682837473328234,
-      "grad_norm": 0.0015586374793201685,
-      "learning_rate": 0.00019999768354383594,
-      "loss": 46.0,
-      "step": 13464
-    },
-    {
-      "epoch": 2.168444784411611,
-      "grad_norm": 0.003854429814964533,
-      "learning_rate": 0.00019999768319947246,
-      "loss": 46.0,
-      "step": 13465
-    },
-    {
-      "epoch": 2.1686058214903983,
-      "grad_norm": 0.00105179357342422,
-      "learning_rate": 0.00019999768285508334,
-      "loss": 46.0,
-      "step": 13466
-    },
-    {
-      "epoch": 2.1687668585691857,
-      "grad_norm": 0.002585239941254258,
-      "learning_rate": 0.00019999768251066863,
-      "loss": 46.0,
-      "step": 13467
-    },
-    {
-      "epoch": 2.1689278956479727,
-      "grad_norm": 0.0019705623853951693,
-      "learning_rate": 0.00019999768216622833,
-      "loss": 46.0,
-      "step": 13468
-    },
-    {
-      "epoch": 2.16908893272676,
-      "grad_norm": 0.0028651857282966375,
-      "learning_rate": 0.00019999768182176245,
-      "loss": 46.0,
-      "step": 13469
-    },
-    {
-      "epoch": 2.1692499698055476,
-      "grad_norm": 0.003413299098610878,
-      "learning_rate": 0.000199997681477271,
-      "loss": 46.0,
-      "step": 13470
-    },
-    {
-      "epoch": 2.169411006884335,
-      "grad_norm": 0.001498452853411436,
-      "learning_rate": 0.00019999768113275392,
-      "loss": 46.0,
-      "step": 13471
-    },
-    {
-      "epoch": 2.1695720439631225,
-      "grad_norm": 0.004488243255764246,
-      "learning_rate": 0.00019999768078821125,
-      "loss": 46.0,
-      "step": 13472
-    },
-    {
-      "epoch": 2.16973308104191,
-      "grad_norm": 0.005841054953634739,
-      "learning_rate": 0.000199997680443643,
-      "loss": 46.0,
-      "step": 13473
-    },
-    {
-      "epoch": 2.1698941181206974,
-      "grad_norm": 0.0042350334115326405,
-      "learning_rate": 0.00019999768009904917,
-      "loss": 46.0,
-      "step": 13474
-    },
-    {
-      "epoch": 2.170055155199485,
-      "grad_norm": 0.007741840090602636,
-      "learning_rate": 0.00019999767975442968,
-      "loss": 46.0,
-      "step": 13475
-    },
-    {
-      "epoch": 2.170216192278272,
-      "grad_norm": 0.008840867318212986,
-      "learning_rate": 0.00019999767940978463,
-      "loss": 46.0,
-      "step": 13476
-    },
-    {
-      "epoch": 2.1703772293570593,
-      "grad_norm": 0.00532876281067729,
-      "learning_rate": 0.000199997679065114,
-      "loss": 46.0,
-      "step": 13477
-    },
-    {
-      "epoch": 2.170538266435847,
-      "grad_norm": 0.004787585698068142,
-      "learning_rate": 0.00019999767872041778,
-      "loss": 46.0,
-      "step": 13478
-    },
-    {
-      "epoch": 2.1706993035146342,
-      "grad_norm": 0.004419929347932339,
-      "learning_rate": 0.00019999767837569597,
-      "loss": 46.0,
-      "step": 13479
-    },
-    {
-      "epoch": 2.1708603405934217,
-      "grad_norm": 0.003897394286468625,
-      "learning_rate": 0.00019999767803094854,
-      "loss": 46.0,
-      "step": 13480
-    },
-    {
-      "epoch": 2.171021377672209,
-      "grad_norm": 0.0010128198191523552,
-      "learning_rate": 0.00019999767768617553,
-      "loss": 46.0,
-      "step": 13481
-    },
-    {
-      "epoch": 2.1711824147509966,
-      "grad_norm": 0.0014552520588040352,
-      "learning_rate": 0.0001999976773413769,
-      "loss": 46.0,
-      "step": 13482
-    },
-    {
-      "epoch": 2.1713434518297836,
-      "grad_norm": 0.0016089060809463263,
-      "learning_rate": 0.00019999767699655272,
-      "loss": 46.0,
-      "step": 13483
-    },
-    {
-      "epoch": 2.171504488908571,
-      "grad_norm": 0.005142230540513992,
-      "learning_rate": 0.0001999976766517029,
-      "loss": 46.0,
-      "step": 13484
-    },
-    {
-      "epoch": 2.1716655259873585,
-      "grad_norm": 0.005287645384669304,
-      "learning_rate": 0.0001999976763068275,
-      "loss": 46.0,
-      "step": 13485
-    },
-    {
-      "epoch": 2.171826563066146,
-      "grad_norm": 0.006501209922134876,
-      "learning_rate": 0.00019999767596192653,
-      "loss": 46.0,
-      "step": 13486
-    },
-    {
-      "epoch": 2.1719876001449334,
-      "grad_norm": 0.006598499603569508,
-      "learning_rate": 0.00019999767561699994,
-      "loss": 46.0,
-      "step": 13487
-    },
-    {
-      "epoch": 2.172148637223721,
-      "grad_norm": 0.0012409052578732371,
-      "learning_rate": 0.00019999767527204777,
-      "loss": 46.0,
-      "step": 13488
-    },
-    {
-      "epoch": 2.1723096743025083,
-      "grad_norm": 0.0018534971168264747,
-      "learning_rate": 0.00019999767492707,
-      "loss": 46.0,
-      "step": 13489
-    },
-    {
-      "epoch": 2.1724707113812958,
-      "grad_norm": 0.002854087157174945,
-      "learning_rate": 0.00019999767458206663,
-      "loss": 46.0,
-      "step": 13490
-    },
-    {
-      "epoch": 2.172631748460083,
-      "grad_norm": 0.008386479690670967,
-      "learning_rate": 0.0001999976742370377,
-      "loss": 46.0,
-      "step": 13491
-    },
-    {
-      "epoch": 2.1727927855388702,
-      "grad_norm": 0.0009295197669416666,
-      "learning_rate": 0.00019999767389198314,
-      "loss": 46.0,
-      "step": 13492
-    },
-    {
-      "epoch": 2.1729538226176577,
-      "grad_norm": 0.0010711585637181997,
-      "learning_rate": 0.000199997673546903,
-      "loss": 46.0,
-      "step": 13493
-    },
-    {
-      "epoch": 2.173114859696445,
-      "grad_norm": 0.007884289138019085,
-      "learning_rate": 0.00019999767320179722,
-      "loss": 46.0,
-      "step": 13494
-    },
-    {
-      "epoch": 2.1732758967752326,
-      "grad_norm": 0.0021540382876992226,
-      "learning_rate": 0.0001999976728566659,
-      "loss": 46.0,
-      "step": 13495
-    },
-    {
-      "epoch": 2.17343693385402,
-      "grad_norm": 0.004816845059394836,
-      "learning_rate": 0.00019999767251150898,
-      "loss": 46.0,
-      "step": 13496
-    },
-    {
-      "epoch": 2.173597970932807,
-      "grad_norm": 0.002338269492611289,
-      "learning_rate": 0.00019999767216632647,
-      "loss": 46.0,
-      "step": 13497
-    },
-    {
-      "epoch": 2.1737590080115945,
-      "grad_norm": 0.0019003247143700719,
-      "learning_rate": 0.00019999767182111834,
-      "loss": 46.0,
-      "step": 13498
-    },
-    {
-      "epoch": 2.173920045090382,
-      "grad_norm": 0.005153437610715628,
-      "learning_rate": 0.00019999767147588462,
-      "loss": 46.0,
-      "step": 13499
-    },
-    {
-      "epoch": 2.1740810821691694,
-      "grad_norm": 0.011673007160425186,
-      "learning_rate": 0.00019999767113062532,
-      "loss": 46.0,
-      "step": 13500
-    },
-    {
-      "epoch": 2.174242119247957,
-      "grad_norm": 0.0049699158407747746,
-      "learning_rate": 0.00019999767078534043,
-      "loss": 46.0,
-      "step": 13501
-    },
-    {
-      "epoch": 2.1744031563267443,
-      "grad_norm": 0.0030734918545931578,
-      "learning_rate": 0.00019999767044002993,
-      "loss": 46.0,
-      "step": 13502
-    },
-    {
-      "epoch": 2.1745641934055318,
-      "grad_norm": 0.0014631218509748578,
-      "learning_rate": 0.00019999767009469386,
-      "loss": 46.0,
-      "step": 13503
-    },
-    {
-      "epoch": 2.174725230484319,
-      "grad_norm": 0.0022403006441891193,
-      "learning_rate": 0.00019999766974933216,
-      "loss": 46.0,
-      "step": 13504
-    },
-    {
-      "epoch": 2.174886267563106,
-      "grad_norm": 0.0016182791441679,
-      "learning_rate": 0.0001999976694039449,
-      "loss": 46.0,
-      "step": 13505
-    },
-    {
-      "epoch": 2.1750473046418937,
-      "grad_norm": 0.002961906138807535,
-      "learning_rate": 0.000199997669058532,
-      "loss": 46.0,
-      "step": 13506
-    },
-    {
-      "epoch": 2.175208341720681,
-      "grad_norm": 0.0019697381649166346,
-      "learning_rate": 0.00019999766871309357,
-      "loss": 46.0,
-      "step": 13507
-    },
-    {
-      "epoch": 2.1753693787994686,
-      "grad_norm": 0.0013614243362098932,
-      "learning_rate": 0.0001999976683676295,
-      "loss": 46.0,
-      "step": 13508
-    },
-    {
-      "epoch": 2.175530415878256,
-      "grad_norm": 0.006952998694032431,
-      "learning_rate": 0.00019999766802213982,
-      "loss": 46.0,
-      "step": 13509
-    },
-    {
-      "epoch": 2.1756914529570435,
-      "grad_norm": 0.002162138931453228,
-      "learning_rate": 0.0001999976676766246,
-      "loss": 46.0,
-      "step": 13510
-    },
-    {
-      "epoch": 2.175852490035831,
-      "grad_norm": 0.0007870601839385927,
-      "learning_rate": 0.00019999766733108375,
-      "loss": 46.0,
-      "step": 13511
-    },
-    {
-      "epoch": 2.176013527114618,
-      "grad_norm": 0.001855554641224444,
-      "learning_rate": 0.0001999976669855173,
-      "loss": 46.0,
-      "step": 13512
-    },
-    {
-      "epoch": 2.1761745641934054,
-      "grad_norm": 0.004577834624797106,
-      "learning_rate": 0.00019999766663992527,
-      "loss": 46.0,
-      "step": 13513
-    },
-    {
-      "epoch": 2.176335601272193,
-      "grad_norm": 0.0012973316479474306,
-      "learning_rate": 0.00019999766629430764,
-      "loss": 46.0,
-      "step": 13514
-    },
-    {
-      "epoch": 2.1764966383509803,
-      "grad_norm": 0.002179705537855625,
-      "learning_rate": 0.00019999766594866442,
-      "loss": 46.0,
-      "step": 13515
-    },
-    {
-      "epoch": 2.1766576754297677,
-      "grad_norm": 0.0010821259347721934,
-      "learning_rate": 0.0001999976656029956,
-      "loss": 46.0,
-      "step": 13516
-    },
-    {
-      "epoch": 2.176818712508555,
-      "grad_norm": 0.002393249422311783,
-      "learning_rate": 0.0001999976652573012,
-      "loss": 46.0,
-      "step": 13517
-    },
-    {
-      "epoch": 2.1769797495873426,
-      "grad_norm": 0.005909727420657873,
-      "learning_rate": 0.00019999766491158121,
-      "loss": 46.0,
-      "step": 13518
-    },
-    {
-      "epoch": 2.17714078666613,
-      "grad_norm": 0.004526886157691479,
-      "learning_rate": 0.0001999976645658356,
-      "loss": 46.0,
-      "step": 13519
-    },
-    {
-      "epoch": 2.177301823744917,
-      "grad_norm": 0.007912407629191875,
-      "learning_rate": 0.0001999976642200644,
-      "loss": 46.0,
-      "step": 13520
-    },
-    {
-      "epoch": 2.1774628608237045,
-      "grad_norm": 0.006462841760367155,
-      "learning_rate": 0.00019999766387426764,
-      "loss": 46.0,
-      "step": 13521
-    },
-    {
-      "epoch": 2.177623897902492,
-      "grad_norm": 0.0014400375075638294,
-      "learning_rate": 0.00019999766352844526,
-      "loss": 46.0,
-      "step": 13522
-    },
-    {
-      "epoch": 2.1777849349812795,
-      "grad_norm": 0.0027416078373789787,
-      "learning_rate": 0.00019999766318259726,
-      "loss": 46.0,
-      "step": 13523
-    },
-    {
-      "epoch": 2.177945972060067,
-      "grad_norm": 0.0034235159400850534,
-      "learning_rate": 0.0001999976628367237,
-      "loss": 46.0,
-      "step": 13524
-    },
-    {
-      "epoch": 2.1781070091388544,
-      "grad_norm": 0.0032882671803236008,
-      "learning_rate": 0.00019999766249082453,
-      "loss": 46.0,
-      "step": 13525
-    },
-    {
-      "epoch": 2.178268046217642,
-      "grad_norm": 0.006466683000326157,
-      "learning_rate": 0.0001999976621448998,
-      "loss": 46.0,
-      "step": 13526
-    },
-    {
-      "epoch": 2.178429083296429,
-      "grad_norm": 0.0036410633474588394,
-      "learning_rate": 0.00019999766179894943,
-      "loss": 46.0,
-      "step": 13527
-    },
-    {
-      "epoch": 2.1785901203752163,
-      "grad_norm": 0.002103250240907073,
-      "learning_rate": 0.0001999976614529735,
-      "loss": 46.0,
-      "step": 13528
-    },
-    {
-      "epoch": 2.1787511574540037,
-      "grad_norm": 0.001565914019010961,
-      "learning_rate": 0.00019999766110697194,
-      "loss": 46.0,
-      "step": 13529
-    },
-    {
-      "epoch": 2.178912194532791,
-      "grad_norm": 0.0023175831884145737,
-      "learning_rate": 0.00019999766076094478,
-      "loss": 46.0,
-      "step": 13530
-    },
-    {
-      "epoch": 2.1790732316115786,
-      "grad_norm": 0.0011863374384120107,
-      "learning_rate": 0.0001999976604148921,
-      "loss": 46.0,
-      "step": 13531
-    },
-    {
-      "epoch": 2.179234268690366,
-      "grad_norm": 0.0015350491739809513,
-      "learning_rate": 0.00019999766006881378,
-      "loss": 46.0,
-      "step": 13532
-    },
-    {
-      "epoch": 2.1793953057691535,
-      "grad_norm": 0.0012241523945704103,
-      "learning_rate": 0.00019999765972270983,
-      "loss": 46.0,
-      "step": 13533
-    },
-    {
-      "epoch": 2.1795563428479405,
-      "grad_norm": 0.0008566062897443771,
-      "learning_rate": 0.00019999765937658032,
-      "loss": 46.0,
-      "step": 13534
-    },
-    {
-      "epoch": 2.179717379926728,
-      "grad_norm": 0.003918860107660294,
-      "learning_rate": 0.00019999765903042522,
-      "loss": 46.0,
-      "step": 13535
-    },
-    {
-      "epoch": 2.1798784170055154,
-      "grad_norm": 0.005466017406433821,
-      "learning_rate": 0.0001999976586842445,
-      "loss": 46.0,
-      "step": 13536
-    },
-    {
-      "epoch": 2.180039454084303,
-      "grad_norm": 0.0029734678100794554,
-      "learning_rate": 0.00019999765833803823,
-      "loss": 46.0,
-      "step": 13537
-    },
-    {
-      "epoch": 2.1802004911630903,
-      "grad_norm": 0.0028642357792705297,
-      "learning_rate": 0.00019999765799180635,
-      "loss": 46.0,
-      "step": 13538
-    },
-    {
-      "epoch": 2.180361528241878,
-      "grad_norm": 0.006614120211452246,
-      "learning_rate": 0.00019999765764554885,
-      "loss": 46.0,
-      "step": 13539
-    },
-    {
-      "epoch": 2.1805225653206652,
-      "grad_norm": 0.0020899188239127398,
-      "learning_rate": 0.0001999976572992658,
-      "loss": 46.0,
-      "step": 13540
-    },
-    {
-      "epoch": 2.1806836023994522,
-      "grad_norm": 0.0011575022945180535,
-      "learning_rate": 0.0001999976569529571,
-      "loss": 46.0,
-      "step": 13541
-    },
-    {
-      "epoch": 2.1808446394782397,
-      "grad_norm": 0.004462036769837141,
-      "learning_rate": 0.00019999765660662285,
-      "loss": 46.0,
-      "step": 13542
-    },
-    {
-      "epoch": 2.181005676557027,
-      "grad_norm": 0.0042463792487978935,
-      "learning_rate": 0.000199997656260263,
-      "loss": 46.0,
-      "step": 13543
-    },
-    {
-      "epoch": 2.1811667136358146,
-      "grad_norm": 0.0012500300072133541,
-      "learning_rate": 0.00019999765591387754,
-      "loss": 46.0,
-      "step": 13544
-    },
-    {
-      "epoch": 2.181327750714602,
-      "grad_norm": 0.0015369309112429619,
-      "learning_rate": 0.00019999765556746649,
-      "loss": 46.0,
-      "step": 13545
-    },
-    {
-      "epoch": 2.1814887877933895,
-      "grad_norm": 0.002138678450137377,
-      "learning_rate": 0.00019999765522102985,
-      "loss": 46.0,
-      "step": 13546
-    },
-    {
-      "epoch": 2.181649824872177,
-      "grad_norm": 0.0021681070793420076,
-      "learning_rate": 0.0001999976548745676,
-      "loss": 46.0,
-      "step": 13547
-    },
-    {
-      "epoch": 2.1818108619509644,
-      "grad_norm": 0.0012066204799339175,
-      "learning_rate": 0.00019999765452807978,
-      "loss": 46.0,
-      "step": 13548
-    },
-    {
-      "epoch": 2.1819718990297514,
-      "grad_norm": 0.00296192429959774,
-      "learning_rate": 0.00019999765418156636,
-      "loss": 46.0,
-      "step": 13549
-    },
-    {
-      "epoch": 2.182132936108539,
-      "grad_norm": 0.002858041552826762,
-      "learning_rate": 0.00019999765383502734,
-      "loss": 46.0,
-      "step": 13550
-    },
-    {
-      "epoch": 2.1822939731873263,
-      "grad_norm": 0.0004258046974427998,
-      "learning_rate": 0.00019999765348846271,
-      "loss": 46.0,
-      "step": 13551
-    },
-    {
-      "epoch": 2.1824550102661138,
-      "grad_norm": 0.006359814200550318,
-      "learning_rate": 0.0001999976531418725,
-      "loss": 46.0,
-      "step": 13552
-    },
-    {
-      "epoch": 2.1826160473449012,
-      "grad_norm": 0.0016410560347139835,
-      "learning_rate": 0.0001999976527952567,
-      "loss": 46.0,
-      "step": 13553
-    },
-    {
-      "epoch": 2.1827770844236887,
-      "grad_norm": 0.006681407801806927,
-      "learning_rate": 0.0001999976524486153,
-      "loss": 46.0,
-      "step": 13554
-    },
-    {
-      "epoch": 2.182938121502476,
-      "grad_norm": 0.0008605445618741214,
-      "learning_rate": 0.00019999765210194833,
-      "loss": 46.0,
-      "step": 13555
-    },
-    {
-      "epoch": 2.183099158581263,
-      "grad_norm": 0.00569927878677845,
-      "learning_rate": 0.00019999765175525574,
-      "loss": 46.0,
-      "step": 13556
-    },
-    {
-      "epoch": 2.1832601956600506,
-      "grad_norm": 0.005860517732799053,
-      "learning_rate": 0.00019999765140853753,
-      "loss": 46.0,
-      "step": 13557
-    },
-    {
-      "epoch": 2.183421232738838,
-      "grad_norm": 0.0004395033756736666,
-      "learning_rate": 0.00019999765106179377,
-      "loss": 46.0,
-      "step": 13558
-    },
-    {
-      "epoch": 2.1835822698176255,
-      "grad_norm": 0.000844571040943265,
-      "learning_rate": 0.00019999765071502442,
-      "loss": 46.0,
-      "step": 13559
-    },
-    {
-      "epoch": 2.183743306896413,
-      "grad_norm": 0.004070487339049578,
-      "learning_rate": 0.00019999765036822945,
-      "loss": 46.0,
-      "step": 13560
-    },
-    {
-      "epoch": 2.1839043439752004,
-      "grad_norm": 0.009051503613591194,
-      "learning_rate": 0.00019999765002140892,
-      "loss": 46.0,
-      "step": 13561
-    },
-    {
-      "epoch": 2.184065381053988,
-      "grad_norm": 0.001517138327471912,
-      "learning_rate": 0.00019999764967456275,
-      "loss": 46.0,
-      "step": 13562
-    },
-    {
-      "epoch": 2.1842264181327753,
-      "grad_norm": 0.0025216175708919764,
-      "learning_rate": 0.000199997649327691,
-      "loss": 46.0,
-      "step": 13563
-    },
-    {
-      "epoch": 2.1843874552115623,
-      "grad_norm": 0.0007327153580263257,
-      "learning_rate": 0.00019999764898079368,
-      "loss": 46.0,
-      "step": 13564
-    },
-    {
-      "epoch": 2.1845484922903498,
-      "grad_norm": 0.0018117144936695695,
-      "learning_rate": 0.00019999764863387073,
-      "loss": 46.0,
-      "step": 13565
-    },
-    {
-      "epoch": 2.184709529369137,
-      "grad_norm": 0.0005709845572710037,
-      "learning_rate": 0.0001999976482869222,
-      "loss": 46.0,
-      "step": 13566
-    },
-    {
-      "epoch": 2.1848705664479247,
-      "grad_norm": 0.0023238679859787226,
-      "learning_rate": 0.00019999764793994808,
-      "loss": 46.0,
-      "step": 13567
-    },
-    {
-      "epoch": 2.185031603526712,
-      "grad_norm": 0.002491126535460353,
-      "learning_rate": 0.00019999764759294836,
-      "loss": 46.0,
-      "step": 13568
-    },
-    {
-      "epoch": 2.1851926406054996,
-      "grad_norm": 0.0031552943401038647,
-      "learning_rate": 0.00019999764724592305,
-      "loss": 46.0,
-      "step": 13569
-    },
-    {
-      "epoch": 2.185353677684287,
-      "grad_norm": 0.00076824682764709,
-      "learning_rate": 0.00019999764689887216,
-      "loss": 46.0,
-      "step": 13570
-    },
-    {
-      "epoch": 2.185514714763074,
-      "grad_norm": 0.0005623137694783509,
-      "learning_rate": 0.00019999764655179565,
-      "loss": 46.0,
-      "step": 13571
-    },
-    {
-      "epoch": 2.1856757518418615,
-      "grad_norm": 0.0022073371801525354,
-      "learning_rate": 0.00019999764620469356,
-      "loss": 46.0,
-      "step": 13572
-    },
-    {
-      "epoch": 2.185836788920649,
-      "grad_norm": 0.0006765663274563849,
-      "learning_rate": 0.00019999764585756585,
-      "loss": 46.0,
-      "step": 13573
-    },
-    {
-      "epoch": 2.1859978259994364,
-      "grad_norm": 0.0013320172438398004,
-      "learning_rate": 0.00019999764551041258,
-      "loss": 46.0,
-      "step": 13574
-    },
-    {
-      "epoch": 2.186158863078224,
-      "grad_norm": 0.0021768000442534685,
-      "learning_rate": 0.00019999764516323373,
-      "loss": 46.0,
-      "step": 13575
-    },
-    {
-      "epoch": 2.1863199001570113,
-      "grad_norm": 0.0003684274561237544,
-      "learning_rate": 0.00019999764481602923,
-      "loss": 46.0,
-      "step": 13576
-    },
-    {
-      "epoch": 2.1864809372357987,
-      "grad_norm": 0.001275241025723517,
-      "learning_rate": 0.00019999764446879917,
-      "loss": 46.0,
-      "step": 13577
-    },
-    {
-      "epoch": 2.1866419743145857,
-      "grad_norm": 0.0036178433801978827,
-      "learning_rate": 0.00019999764412154353,
-      "loss": 46.0,
-      "step": 13578
-    },
-    {
-      "epoch": 2.186803011393373,
-      "grad_norm": 0.002656420459970832,
-      "learning_rate": 0.00019999764377426224,
-      "loss": 46.0,
-      "step": 13579
-    },
-    {
-      "epoch": 2.1869640484721606,
-      "grad_norm": 0.009571787901222706,
-      "learning_rate": 0.00019999764342695542,
-      "loss": 46.0,
-      "step": 13580
-    },
-    {
-      "epoch": 2.187125085550948,
-      "grad_norm": 0.001973456237465143,
-      "learning_rate": 0.00019999764307962296,
-      "loss": 46.0,
-      "step": 13581
-    },
-    {
-      "epoch": 2.1872861226297355,
-      "grad_norm": 0.005300706718116999,
-      "learning_rate": 0.00019999764273226494,
-      "loss": 46.0,
-      "step": 13582
-    },
-    {
-      "epoch": 2.187447159708523,
-      "grad_norm": 0.0007981974049471319,
-      "learning_rate": 0.0001999976423848813,
-      "loss": 46.0,
-      "step": 13583
-    },
-    {
-      "epoch": 2.1876081967873104,
-      "grad_norm": 0.0029203679878264666,
-      "learning_rate": 0.00019999764203747206,
-      "loss": 46.0,
-      "step": 13584
-    },
-    {
-      "epoch": 2.1877692338660975,
-      "grad_norm": 0.0031833008397370577,
-      "learning_rate": 0.00019999764169003728,
-      "loss": 46.0,
-      "step": 13585
-    },
-    {
-      "epoch": 2.187930270944885,
-      "grad_norm": 0.0007549699512310326,
-      "learning_rate": 0.00019999764134257685,
-      "loss": 46.0,
-      "step": 13586
-    },
-    {
-      "epoch": 2.1880913080236724,
-      "grad_norm": 0.001534285955131054,
-      "learning_rate": 0.00019999764099509084,
-      "loss": 46.0,
-      "step": 13587
-    },
-    {
-      "epoch": 2.18825234510246,
-      "grad_norm": 0.002749880775809288,
-      "learning_rate": 0.00019999764064757925,
-      "loss": 46.0,
-      "step": 13588
-    },
-    {
-      "epoch": 2.1884133821812473,
-      "grad_norm": 0.004389625508338213,
-      "learning_rate": 0.00019999764030004204,
-      "loss": 46.0,
-      "step": 13589
-    },
-    {
-      "epoch": 2.1885744192600347,
-      "grad_norm": 0.0018506953492760658,
-      "learning_rate": 0.00019999763995247926,
-      "loss": 46.0,
-      "step": 13590
-    },
-    {
-      "epoch": 2.188735456338822,
-      "grad_norm": 0.0014044672716408968,
-      "learning_rate": 0.00019999763960489085,
-      "loss": 46.0,
-      "step": 13591
-    },
-    {
-      "epoch": 2.1888964934176096,
-      "grad_norm": 0.0017050118185579777,
-      "learning_rate": 0.00019999763925727688,
-      "loss": 46.0,
-      "step": 13592
-    },
-    {
-      "epoch": 2.1890575304963966,
-      "grad_norm": 0.0045059421099722385,
-      "learning_rate": 0.00019999763890963732,
-      "loss": 46.0,
-      "step": 13593
-    },
-    {
-      "epoch": 2.189218567575184,
-      "grad_norm": 0.0016245106235146523,
-      "learning_rate": 0.00019999763856197214,
-      "loss": 46.0,
-      "step": 13594
-    },
-    {
-      "epoch": 2.1893796046539715,
-      "grad_norm": 0.0026045513805001974,
-      "learning_rate": 0.00019999763821428138,
-      "loss": 46.0,
-      "step": 13595
-    },
-    {
-      "epoch": 2.189540641732759,
-      "grad_norm": 0.0014010741142556071,
-      "learning_rate": 0.00019999763786656503,
-      "loss": 46.0,
-      "step": 13596
-    },
-    {
-      "epoch": 2.1897016788115464,
-      "grad_norm": 0.0023435086477547884,
-      "learning_rate": 0.00019999763751882307,
-      "loss": 46.0,
-      "step": 13597
-    },
-    {
-      "epoch": 2.189862715890334,
-      "grad_norm": 0.0005747080431319773,
-      "learning_rate": 0.00019999763717105554,
-      "loss": 46.0,
-      "step": 13598
-    },
-    {
-      "epoch": 2.1900237529691213,
-      "grad_norm": 0.0029229726642370224,
-      "learning_rate": 0.0001999976368232624,
-      "loss": 46.0,
-      "step": 13599
-    },
-    {
-      "epoch": 2.1901847900479083,
-      "grad_norm": 0.002026109956204891,
-      "learning_rate": 0.00019999763647544368,
-      "loss": 46.0,
-      "step": 13600
-    },
-    {
-      "epoch": 2.190345827126696,
-      "grad_norm": 0.008457713760435581,
-      "learning_rate": 0.00019999763612759934,
-      "loss": 46.0,
-      "step": 13601
-    },
-    {
-      "epoch": 2.1905068642054832,
-      "grad_norm": 0.0016848701052367687,
-      "learning_rate": 0.00019999763577972944,
-      "loss": 46.0,
-      "step": 13602
-    },
-    {
-      "epoch": 2.1906679012842707,
-      "grad_norm": 0.0011079522082582116,
-      "learning_rate": 0.0001999976354318339,
-      "loss": 46.0,
-      "step": 13603
-    },
-    {
-      "epoch": 2.190828938363058,
-      "grad_norm": 0.00237249955534935,
-      "learning_rate": 0.00019999763508391278,
-      "loss": 46.0,
-      "step": 13604
-    },
-    {
-      "epoch": 2.1909899754418456,
-      "grad_norm": 0.003441508859395981,
-      "learning_rate": 0.0001999976347359661,
-      "loss": 46.0,
-      "step": 13605
-    },
-    {
-      "epoch": 2.191151012520633,
-      "grad_norm": 0.004961010534316301,
-      "learning_rate": 0.00019999763438799379,
-      "loss": 46.0,
-      "step": 13606
-    },
-    {
-      "epoch": 2.1913120495994205,
-      "grad_norm": 0.0010814645793288946,
-      "learning_rate": 0.0001999976340399959,
-      "loss": 46.0,
-      "step": 13607
-    },
-    {
-      "epoch": 2.1914730866782075,
-      "grad_norm": 0.010299154557287693,
-      "learning_rate": 0.0001999976336919724,
-      "loss": 46.0,
-      "step": 13608
-    },
-    {
-      "epoch": 2.191634123756995,
-      "grad_norm": 0.002826516982167959,
-      "learning_rate": 0.00019999763334392333,
-      "loss": 46.0,
-      "step": 13609
-    },
-    {
-      "epoch": 2.1917951608357824,
-      "grad_norm": 0.008937855251133442,
-      "learning_rate": 0.00019999763299584865,
-      "loss": 46.0,
-      "step": 13610
-    },
-    {
-      "epoch": 2.19195619791457,
-      "grad_norm": 0.011181384325027466,
-      "learning_rate": 0.00019999763264774841,
-      "loss": 46.0,
-      "step": 13611
-    },
-    {
-      "epoch": 2.1921172349933573,
-      "grad_norm": 0.0022061506751924753,
-      "learning_rate": 0.0001999976322996225,
-      "loss": 46.0,
-      "step": 13612
-    },
-    {
-      "epoch": 2.1922782720721448,
-      "grad_norm": 0.0011596990516409278,
-      "learning_rate": 0.00019999763195147104,
-      "loss": 46.0,
-      "step": 13613
-    },
-    {
-      "epoch": 2.1924393091509318,
-      "grad_norm": 0.0029557885136455297,
-      "learning_rate": 0.00019999763160329401,
-      "loss": 46.0,
-      "step": 13614
-    },
-    {
-      "epoch": 2.192600346229719,
-      "grad_norm": 0.0026413993909955025,
-      "learning_rate": 0.00019999763125509137,
-      "loss": 46.0,
-      "step": 13615
-    },
-    {
-      "epoch": 2.1927613833085067,
-      "grad_norm": 0.0010780281154438853,
-      "learning_rate": 0.00019999763090686312,
-      "loss": 46.0,
-      "step": 13616
-    },
-    {
-      "epoch": 2.192922420387294,
-      "grad_norm": 0.003673151833936572,
-      "learning_rate": 0.00019999763055860928,
-      "loss": 46.0,
-      "step": 13617
-    },
-    {
-      "epoch": 2.1930834574660816,
-      "grad_norm": 0.0017527603777125478,
-      "learning_rate": 0.00019999763021032987,
-      "loss": 46.0,
-      "step": 13618
-    },
-    {
-      "epoch": 2.193244494544869,
-      "grad_norm": 0.000748249760363251,
-      "learning_rate": 0.00019999762986202483,
-      "loss": 46.0,
-      "step": 13619
-    },
-    {
-      "epoch": 2.1934055316236565,
-      "grad_norm": 0.0035505208652466536,
-      "learning_rate": 0.00019999762951369423,
-      "loss": 46.0,
-      "step": 13620
-    },
-    {
-      "epoch": 2.193566568702444,
-      "grad_norm": 0.01069399993866682,
-      "learning_rate": 0.000199997629165338,
-      "loss": 46.0,
-      "step": 13621
-    },
-    {
-      "epoch": 2.193727605781231,
-      "grad_norm": 0.00884457677602768,
-      "learning_rate": 0.00019999762881695618,
-      "loss": 46.0,
-      "step": 13622
-    },
-    {
-      "epoch": 2.1938886428600184,
-      "grad_norm": 0.009702994488179684,
-      "learning_rate": 0.00019999762846854878,
-      "loss": 46.0,
-      "step": 13623
-    },
-    {
-      "epoch": 2.194049679938806,
-      "grad_norm": 0.002473380183801055,
-      "learning_rate": 0.0001999976281201158,
-      "loss": 46.0,
-      "step": 13624
-    },
-    {
-      "epoch": 2.1942107170175933,
-      "grad_norm": 0.0032200992573052645,
-      "learning_rate": 0.00019999762777165724,
-      "loss": 46.0,
-      "step": 13625
-    },
-    {
-      "epoch": 2.1943717540963807,
-      "grad_norm": 0.0026594691444188356,
-      "learning_rate": 0.00019999762742317303,
-      "loss": 46.0,
-      "step": 13626
-    },
-    {
-      "epoch": 2.194532791175168,
-      "grad_norm": 0.003416158026084304,
-      "learning_rate": 0.00019999762707466326,
-      "loss": 46.0,
-      "step": 13627
-    },
-    {
-      "epoch": 2.1946938282539556,
-      "grad_norm": 0.008864456787705421,
-      "learning_rate": 0.00019999762672612788,
-      "loss": 46.0,
-      "step": 13628
-    },
-    {
-      "epoch": 2.1948548653327427,
-      "grad_norm": 0.001293074106797576,
-      "learning_rate": 0.0001999976263775669,
-      "loss": 46.0,
-      "step": 13629
-    },
-    {
-      "epoch": 2.19501590241153,
-      "grad_norm": 0.0007040550117380917,
-      "learning_rate": 0.00019999762602898035,
-      "loss": 46.0,
-      "step": 13630
-    },
-    {
-      "epoch": 2.1951769394903176,
-      "grad_norm": 0.007560614496469498,
-      "learning_rate": 0.00019999762568036818,
-      "loss": 46.0,
-      "step": 13631
-    },
-    {
-      "epoch": 2.195337976569105,
-      "grad_norm": 0.004180553834885359,
-      "learning_rate": 0.00019999762533173042,
-      "loss": 46.0,
-      "step": 13632
-    },
-    {
-      "epoch": 2.1954990136478925,
-      "grad_norm": 0.0016429858515039086,
-      "learning_rate": 0.0001999976249830671,
-      "loss": 46.0,
-      "step": 13633
-    },
-    {
-      "epoch": 2.19566005072668,
-      "grad_norm": 0.004915562923997641,
-      "learning_rate": 0.00019999762463437817,
-      "loss": 46.0,
-      "step": 13634
-    },
-    {
-      "epoch": 2.1958210878054674,
-      "grad_norm": 0.0035727759823203087,
-      "learning_rate": 0.00019999762428566365,
-      "loss": 46.0,
-      "step": 13635
-    },
-    {
-      "epoch": 2.195982124884255,
-      "grad_norm": 0.002737649716436863,
-      "learning_rate": 0.0001999976239369235,
-      "loss": 46.0,
-      "step": 13636
-    },
-    {
-      "epoch": 2.196143161963042,
-      "grad_norm": 0.0015738006914034486,
-      "learning_rate": 0.0001999976235881578,
-      "loss": 46.0,
-      "step": 13637
-    },
-    {
-      "epoch": 2.1963041990418293,
-      "grad_norm": 0.008683986030519009,
-      "learning_rate": 0.00019999762323936646,
-      "loss": 46.0,
-      "step": 13638
-    },
-    {
-      "epoch": 2.1964652361206167,
-      "grad_norm": 0.0032735627610236406,
-      "learning_rate": 0.00019999762289054957,
-      "loss": 46.0,
-      "step": 13639
-    },
-    {
-      "epoch": 2.196626273199404,
-      "grad_norm": 0.003026894759386778,
-      "learning_rate": 0.00019999762254170706,
-      "loss": 46.0,
-      "step": 13640
-    },
-    {
-      "epoch": 2.1967873102781916,
-      "grad_norm": 0.000801481248345226,
-      "learning_rate": 0.00019999762219283896,
-      "loss": 46.0,
-      "step": 13641
-    },
-    {
-      "epoch": 2.196948347356979,
-      "grad_norm": 0.0017323765205219388,
-      "learning_rate": 0.00019999762184394525,
-      "loss": 46.0,
-      "step": 13642
-    },
-    {
-      "epoch": 2.1971093844357665,
-      "grad_norm": 0.001625302480533719,
-      "learning_rate": 0.00019999762149502598,
-      "loss": 46.0,
-      "step": 13643
-    },
-    {
-      "epoch": 2.1972704215145535,
-      "grad_norm": 0.0010175653733313084,
-      "learning_rate": 0.0001999976211460811,
-      "loss": 46.0,
-      "step": 13644
-    },
-    {
-      "epoch": 2.197431458593341,
-      "grad_norm": 0.002600245177745819,
-      "learning_rate": 0.00019999762079711065,
-      "loss": 46.0,
-      "step": 13645
-    },
-    {
-      "epoch": 2.1975924956721284,
-      "grad_norm": 0.0017785308882594109,
-      "learning_rate": 0.00019999762044811456,
-      "loss": 46.0,
-      "step": 13646
-    },
-    {
-      "epoch": 2.197753532750916,
-      "grad_norm": 0.0011726110242307186,
-      "learning_rate": 0.0001999976200990929,
-      "loss": 46.0,
-      "step": 13647
-    },
-    {
-      "epoch": 2.1979145698297033,
-      "grad_norm": 0.0029361031483858824,
-      "learning_rate": 0.00019999761975004562,
-      "loss": 46.0,
-      "step": 13648
-    },
-    {
-      "epoch": 2.198075606908491,
-      "grad_norm": 0.0029444019310176373,
-      "learning_rate": 0.0001999976194009728,
-      "loss": 46.0,
-      "step": 13649
-    },
-    {
-      "epoch": 2.1982366439872782,
-      "grad_norm": 0.001622056239284575,
-      "learning_rate": 0.00019999761905187434,
-      "loss": 46.0,
-      "step": 13650
-    },
-    {
-      "epoch": 2.1983976810660653,
-      "grad_norm": 0.001456732745282352,
-      "learning_rate": 0.00019999761870275032,
-      "loss": 46.0,
-      "step": 13651
-    },
-    {
-      "epoch": 2.1985587181448527,
-      "grad_norm": 0.008921913802623749,
-      "learning_rate": 0.00019999761835360068,
-      "loss": 46.0,
-      "step": 13652
-    },
-    {
-      "epoch": 2.19871975522364,
-      "grad_norm": 0.0012266503181308508,
-      "learning_rate": 0.00019999761800442546,
-      "loss": 46.0,
-      "step": 13653
-    },
-    {
-      "epoch": 2.1988807923024276,
-      "grad_norm": 0.0022826273925602436,
-      "learning_rate": 0.0001999976176552246,
-      "loss": 46.0,
-      "step": 13654
-    },
-    {
-      "epoch": 2.199041829381215,
-      "grad_norm": 0.0012355137150734663,
-      "learning_rate": 0.0001999976173059982,
-      "loss": 46.0,
-      "step": 13655
-    },
-    {
-      "epoch": 2.1992028664600025,
-      "grad_norm": 0.0031980867497622967,
-      "learning_rate": 0.00019999761695674618,
-      "loss": 46.0,
-      "step": 13656
-    },
-    {
-      "epoch": 2.19936390353879,
-      "grad_norm": 0.0019669425673782825,
-      "learning_rate": 0.00019999761660746858,
-      "loss": 46.0,
-      "step": 13657
-    },
-    {
-      "epoch": 2.199524940617577,
-      "grad_norm": 0.0008439529337920249,
-      "learning_rate": 0.00019999761625816536,
-      "loss": 46.0,
-      "step": 13658
-    },
-    {
-      "epoch": 2.1996859776963644,
-      "grad_norm": 0.0023680536542087793,
-      "learning_rate": 0.0001999976159088366,
-      "loss": 46.0,
-      "step": 13659
-    },
-    {
-      "epoch": 2.199847014775152,
-      "grad_norm": 0.0019078385084867477,
-      "learning_rate": 0.0001999976155594822,
-      "loss": 46.0,
-      "step": 13660
-    },
-    {
-      "epoch": 2.2000080518539393,
-      "grad_norm": 0.0012144406791776419,
-      "learning_rate": 0.00019999761521010223,
-      "loss": 46.0,
-      "step": 13661
-    },
-    {
-      "epoch": 2.200169088932727,
-      "grad_norm": 0.0021109990775585175,
-      "learning_rate": 0.00019999761486069663,
-      "loss": 46.0,
-      "step": 13662
-    },
-    {
-      "epoch": 2.2003301260115142,
-      "grad_norm": 0.008120409213006496,
-      "learning_rate": 0.00019999761451126546,
-      "loss": 46.0,
-      "step": 13663
-    },
-    {
-      "epoch": 2.2004911630903017,
-      "grad_norm": 0.0020527790766209364,
-      "learning_rate": 0.0001999976141618087,
-      "loss": 46.0,
-      "step": 13664
-    },
-    {
-      "epoch": 2.200652200169089,
-      "grad_norm": 0.0007342428434640169,
-      "learning_rate": 0.00019999761381232634,
-      "loss": 46.0,
-      "step": 13665
-    },
-    {
-      "epoch": 2.200813237247876,
-      "grad_norm": 0.006230971310287714,
-      "learning_rate": 0.0001999976134628184,
-      "loss": 46.0,
-      "step": 13666
-    },
-    {
-      "epoch": 2.2009742743266636,
-      "grad_norm": 0.0014894644264131784,
-      "learning_rate": 0.00019999761311328485,
-      "loss": 46.0,
-      "step": 13667
-    },
-    {
-      "epoch": 2.201135311405451,
-      "grad_norm": 0.010603736154735088,
-      "learning_rate": 0.0001999976127637257,
-      "loss": 46.0,
-      "step": 13668
-    },
-    {
-      "epoch": 2.2012963484842385,
-      "grad_norm": 0.0042411889880895615,
-      "learning_rate": 0.00019999761241414095,
-      "loss": 46.0,
-      "step": 13669
-    },
-    {
-      "epoch": 2.201457385563026,
-      "grad_norm": 0.008183501660823822,
-      "learning_rate": 0.00019999761206453064,
-      "loss": 46.0,
-      "step": 13670
-    },
-    {
-      "epoch": 2.2016184226418134,
-      "grad_norm": 0.004855779930949211,
-      "learning_rate": 0.0001999976117148947,
-      "loss": 46.0,
-      "step": 13671
-    },
-    {
-      "epoch": 2.201779459720601,
-      "grad_norm": 0.0013469979166984558,
-      "learning_rate": 0.0001999976113652332,
-      "loss": 46.0,
-      "step": 13672
-    },
-    {
-      "epoch": 2.201940496799388,
-      "grad_norm": 0.0020328834652900696,
-      "learning_rate": 0.00019999761101554606,
-      "loss": 46.0,
-      "step": 13673
-    },
-    {
-      "epoch": 2.2021015338781753,
-      "grad_norm": 0.005405766889452934,
-      "learning_rate": 0.00019999761066583335,
-      "loss": 46.0,
-      "step": 13674
-    },
-    {
-      "epoch": 2.2022625709569628,
-      "grad_norm": 0.004016388673335314,
-      "learning_rate": 0.00019999761031609507,
-      "loss": 46.0,
-      "step": 13675
-    },
-    {
-      "epoch": 2.20242360803575,
-      "grad_norm": 0.0033229391556233168,
-      "learning_rate": 0.00019999760996633118,
-      "loss": 46.0,
-      "step": 13676
-    },
-    {
-      "epoch": 2.2025846451145377,
-      "grad_norm": 0.0011385190300643444,
-      "learning_rate": 0.00019999760961654167,
-      "loss": 46.0,
-      "step": 13677
-    },
-    {
-      "epoch": 2.202745682193325,
-      "grad_norm": 0.004269816912710667,
-      "learning_rate": 0.0001999976092667266,
-      "loss": 46.0,
-      "step": 13678
-    },
-    {
-      "epoch": 2.2029067192721126,
-      "grad_norm": 0.0019309286726638675,
-      "learning_rate": 0.0001999976089168859,
-      "loss": 46.0,
-      "step": 13679
-    },
-    {
-      "epoch": 2.2030677563509,
-      "grad_norm": 0.0038594515062868595,
-      "learning_rate": 0.00019999760856701963,
-      "loss": 46.0,
-      "step": 13680
-    },
-    {
-      "epoch": 2.203228793429687,
-      "grad_norm": 0.002060219179838896,
-      "learning_rate": 0.00019999760821712777,
-      "loss": 46.0,
-      "step": 13681
-    },
-    {
-      "epoch": 2.2033898305084745,
-      "grad_norm": 0.0009372822241857648,
-      "learning_rate": 0.00019999760786721028,
-      "loss": 46.0,
-      "step": 13682
-    },
-    {
-      "epoch": 2.203550867587262,
-      "grad_norm": 0.006065111141651869,
-      "learning_rate": 0.00019999760751726725,
-      "loss": 46.0,
-      "step": 13683
-    },
-    {
-      "epoch": 2.2037119046660494,
-      "grad_norm": 0.002058411715552211,
-      "learning_rate": 0.0001999976071672986,
-      "loss": 46.0,
-      "step": 13684
-    },
-    {
-      "epoch": 2.203872941744837,
-      "grad_norm": 0.00735425716266036,
-      "learning_rate": 0.00019999760681730435,
-      "loss": 46.0,
-      "step": 13685
-    },
-    {
-      "epoch": 2.2040339788236243,
-      "grad_norm": 0.0031345027964562178,
-      "learning_rate": 0.0001999976064672845,
-      "loss": 46.0,
-      "step": 13686
-    },
-    {
-      "epoch": 2.2041950159024117,
-      "grad_norm": 0.0033000244293361902,
-      "learning_rate": 0.00019999760611723907,
-      "loss": 46.0,
-      "step": 13687
-    },
-    {
-      "epoch": 2.2043560529811987,
-      "grad_norm": 0.0011922132689505816,
-      "learning_rate": 0.00019999760576716805,
-      "loss": 46.0,
-      "step": 13688
-    },
-    {
-      "epoch": 2.204517090059986,
-      "grad_norm": 0.007995802909135818,
-      "learning_rate": 0.00019999760541707145,
-      "loss": 46.0,
-      "step": 13689
-    },
-    {
-      "epoch": 2.2046781271387736,
-      "grad_norm": 0.001073703053407371,
-      "learning_rate": 0.0001999976050669492,
-      "loss": 46.0,
-      "step": 13690
-    },
-    {
-      "epoch": 2.204839164217561,
-      "grad_norm": 0.0011595672694966197,
-      "learning_rate": 0.0001999976047168014,
-      "loss": 46.0,
-      "step": 13691
-    },
-    {
-      "epoch": 2.2050002012963485,
-      "grad_norm": 0.006388626527041197,
-      "learning_rate": 0.000199997604366628,
-      "loss": 46.0,
-      "step": 13692
-    },
-    {
-      "epoch": 2.205161238375136,
-      "grad_norm": 0.0010825737845152617,
-      "learning_rate": 0.00019999760401642899,
-      "loss": 46.0,
-      "step": 13693
-    },
-    {
-      "epoch": 2.2053222754539235,
-      "grad_norm": 0.003770017297938466,
-      "learning_rate": 0.0001999976036662044,
-      "loss": 46.0,
-      "step": 13694
-    },
-    {
-      "epoch": 2.2054833125327105,
-      "grad_norm": 0.007412966340780258,
-      "learning_rate": 0.0001999976033159542,
-      "loss": 46.0,
-      "step": 13695
-    },
-    {
-      "epoch": 2.205644349611498,
-      "grad_norm": 0.0010894156293943524,
-      "learning_rate": 0.00019999760296567843,
-      "loss": 46.0,
-      "step": 13696
-    },
-    {
-      "epoch": 2.2058053866902854,
-      "grad_norm": 0.0007538596983067691,
-      "learning_rate": 0.00019999760261537705,
-      "loss": 46.0,
-      "step": 13697
-    },
-    {
-      "epoch": 2.205966423769073,
-      "grad_norm": 0.0003239456273149699,
-      "learning_rate": 0.00019999760226505008,
-      "loss": 46.0,
-      "step": 13698
-    },
-    {
-      "epoch": 2.2061274608478603,
-      "grad_norm": 0.0018409915501251817,
-      "learning_rate": 0.00019999760191469754,
-      "loss": 46.0,
-      "step": 13699
-    },
-    {
-      "epoch": 2.2062884979266477,
-      "grad_norm": 0.003168245777487755,
-      "learning_rate": 0.00019999760156431937,
-      "loss": 46.0,
-      "step": 13700
-    },
-    {
-      "epoch": 2.206449535005435,
-      "grad_norm": 0.002283796202391386,
-      "learning_rate": 0.0001999976012139156,
-      "loss": 46.0,
-      "step": 13701
-    },
-    {
-      "epoch": 2.206610572084222,
-      "grad_norm": 0.0011646875645965338,
-      "learning_rate": 0.00019999760086348626,
-      "loss": 46.0,
-      "step": 13702
-    },
-    {
-      "epoch": 2.2067716091630096,
-      "grad_norm": 0.009998772293329239,
-      "learning_rate": 0.00019999760051303133,
-      "loss": 46.0,
-      "step": 13703
-    },
-    {
-      "epoch": 2.206932646241797,
-      "grad_norm": 0.002271430566906929,
-      "learning_rate": 0.0001999976001625508,
-      "loss": 46.0,
-      "step": 13704
-    },
-    {
-      "epoch": 2.2070936833205845,
-      "grad_norm": 0.007020572666078806,
-      "learning_rate": 0.00019999759981204464,
-      "loss": 46.0,
-      "step": 13705
-    },
-    {
-      "epoch": 2.207254720399372,
-      "grad_norm": 0.003235981334000826,
-      "learning_rate": 0.00019999759946151294,
-      "loss": 46.0,
-      "step": 13706
-    },
-    {
-      "epoch": 2.2074157574781594,
-      "grad_norm": 0.002830745652318001,
-      "learning_rate": 0.0001999975991109556,
-      "loss": 46.0,
-      "step": 13707
-    },
-    {
-      "epoch": 2.207576794556947,
-      "grad_norm": 0.0016132837627083063,
-      "learning_rate": 0.0001999975987603727,
-      "loss": 46.0,
-      "step": 13708
-    },
-    {
-      "epoch": 2.2077378316357343,
-      "grad_norm": 0.0038657637778669596,
-      "learning_rate": 0.0001999975984097642,
-      "loss": 46.0,
-      "step": 13709
-    },
-    {
-      "epoch": 2.2078988687145213,
-      "grad_norm": 0.0012947127688676119,
-      "learning_rate": 0.0001999975980591301,
-      "loss": 46.0,
-      "step": 13710
-    },
-    {
-      "epoch": 2.208059905793309,
-      "grad_norm": 0.0009376327507197857,
-      "learning_rate": 0.0001999975977084704,
-      "loss": 46.0,
-      "step": 13711
-    },
-    {
-      "epoch": 2.2082209428720962,
-      "grad_norm": 0.0003879389551002532,
-      "learning_rate": 0.0001999975973577851,
-      "loss": 46.0,
-      "step": 13712
-    },
-    {
-      "epoch": 2.2083819799508837,
-      "grad_norm": 0.004862251225858927,
-      "learning_rate": 0.00019999759700707422,
-      "loss": 46.0,
-      "step": 13713
-    },
-    {
-      "epoch": 2.208543017029671,
-      "grad_norm": 0.006597419735044241,
-      "learning_rate": 0.00019999759665633774,
-      "loss": 46.0,
-      "step": 13714
-    },
-    {
-      "epoch": 2.2087040541084586,
-      "grad_norm": 0.00931819062680006,
-      "learning_rate": 0.00019999759630557565,
-      "loss": 46.0,
-      "step": 13715
-    },
-    {
-      "epoch": 2.208865091187246,
-      "grad_norm": 0.0049352184869349,
-      "learning_rate": 0.000199997595954788,
-      "loss": 46.0,
-      "step": 13716
-    },
-    {
-      "epoch": 2.209026128266033,
-      "grad_norm": 0.0030654005240648985,
-      "learning_rate": 0.00019999759560397474,
-      "loss": 46.0,
-      "step": 13717
-    },
-    {
-      "epoch": 2.2091871653448205,
-      "grad_norm": 0.008191955275833607,
-      "learning_rate": 0.0001999975952531359,
-      "loss": 46.0,
-      "step": 13718
-    },
-    {
-      "epoch": 2.209348202423608,
-      "grad_norm": 0.0031054536812007427,
-      "learning_rate": 0.00019999759490227142,
-      "loss": 46.0,
-      "step": 13719
-    },
-    {
-      "epoch": 2.2095092395023954,
-      "grad_norm": 0.005711531266570091,
-      "learning_rate": 0.0001999975945513814,
-      "loss": 46.0,
-      "step": 13720
-    },
-    {
-      "epoch": 2.209670276581183,
-      "grad_norm": 0.002103405073285103,
-      "learning_rate": 0.00019999759420046576,
-      "loss": 46.0,
-      "step": 13721
-    },
-    {
-      "epoch": 2.2098313136599703,
-      "grad_norm": 0.0025062395725399256,
-      "learning_rate": 0.00019999759384952453,
-      "loss": 46.0,
-      "step": 13722
-    },
-    {
-      "epoch": 2.2099923507387578,
-      "grad_norm": 0.0048634884878993034,
-      "learning_rate": 0.00019999759349855772,
-      "loss": 46.0,
-      "step": 13723
-    },
-    {
-      "epoch": 2.2101533878175452,
-      "grad_norm": 0.0032078830990940332,
-      "learning_rate": 0.0001999975931475653,
-      "loss": 46.0,
-      "step": 13724
-    },
-    {
-      "epoch": 2.2103144248963322,
-      "grad_norm": 0.003515372984111309,
-      "learning_rate": 0.00019999759279654727,
-      "loss": 46.0,
-      "step": 13725
-    },
-    {
-      "epoch": 2.2104754619751197,
-      "grad_norm": 0.003074072999879718,
-      "learning_rate": 0.00019999759244550367,
-      "loss": 46.0,
-      "step": 13726
-    },
-    {
-      "epoch": 2.210636499053907,
-      "grad_norm": 0.002210519975051284,
-      "learning_rate": 0.00019999759209443445,
-      "loss": 46.0,
-      "step": 13727
-    },
-    {
-      "epoch": 2.2107975361326946,
-      "grad_norm": 0.005355529952794313,
-      "learning_rate": 0.00019999759174333965,
-      "loss": 46.0,
-      "step": 13728
-    },
-    {
-      "epoch": 2.210958573211482,
-      "grad_norm": 0.0009055686532519758,
-      "learning_rate": 0.00019999759139221928,
-      "loss": 46.0,
-      "step": 13729
-    },
-    {
-      "epoch": 2.2111196102902695,
-      "grad_norm": 0.008600863628089428,
-      "learning_rate": 0.00019999759104107328,
-      "loss": 46.0,
-      "step": 13730
-    },
-    {
-      "epoch": 2.2112806473690565,
-      "grad_norm": 0.0024756542406976223,
-      "learning_rate": 0.00019999759068990168,
-      "loss": 46.0,
-      "step": 13731
-    },
-    {
-      "epoch": 2.211441684447844,
-      "grad_norm": 0.0011623684549704194,
-      "learning_rate": 0.00019999759033870453,
-      "loss": 46.0,
-      "step": 13732
-    },
-    {
-      "epoch": 2.2116027215266314,
-      "grad_norm": 0.003117821179330349,
-      "learning_rate": 0.00019999758998748174,
-      "loss": 46.0,
-      "step": 13733
-    },
-    {
-      "epoch": 2.211763758605419,
-      "grad_norm": 0.006619364954531193,
-      "learning_rate": 0.00019999758963623338,
-      "loss": 46.0,
-      "step": 13734
-    },
-    {
-      "epoch": 2.2119247956842063,
-      "grad_norm": 0.004035099409520626,
-      "learning_rate": 0.00019999758928495944,
-      "loss": 46.0,
-      "step": 13735
-    },
-    {
-      "epoch": 2.2120858327629938,
-      "grad_norm": 0.004001006484031677,
-      "learning_rate": 0.00019999758893365989,
-      "loss": 46.0,
-      "step": 13736
-    },
-    {
-      "epoch": 2.212246869841781,
-      "grad_norm": 0.0008899013046175241,
-      "learning_rate": 0.00019999758858233472,
-      "loss": 46.0,
-      "step": 13737
-    },
-    {
-      "epoch": 2.2124079069205687,
-      "grad_norm": 0.00424217851832509,
-      "learning_rate": 0.000199997588230984,
-      "loss": 46.0,
-      "step": 13738
-    },
-    {
-      "epoch": 2.2125689439993557,
-      "grad_norm": 0.0032466743141412735,
-      "learning_rate": 0.00019999758787960764,
-      "loss": 46.0,
-      "step": 13739
-    },
-    {
-      "epoch": 2.212729981078143,
-      "grad_norm": 0.0024136712308973074,
-      "learning_rate": 0.0001999975875282057,
-      "loss": 46.0,
-      "step": 13740
-    },
-    {
-      "epoch": 2.2128910181569306,
-      "grad_norm": 0.005446125287562609,
-      "learning_rate": 0.00019999758717677822,
-      "loss": 46.0,
-      "step": 13741
-    },
-    {
-      "epoch": 2.213052055235718,
-      "grad_norm": 0.0022979495115578175,
-      "learning_rate": 0.00019999758682532508,
-      "loss": 46.0,
-      "step": 13742
-    },
-    {
-      "epoch": 2.2132130923145055,
-      "grad_norm": 0.0006464969483204186,
-      "learning_rate": 0.00019999758647384636,
-      "loss": 46.0,
-      "step": 13743
-    },
-    {
-      "epoch": 2.213374129393293,
-      "grad_norm": 0.00036187004297971725,
-      "learning_rate": 0.00019999758612234208,
-      "loss": 46.0,
-      "step": 13744
-    },
-    {
-      "epoch": 2.2135351664720804,
-      "grad_norm": 0.0017232020618394017,
-      "learning_rate": 0.00019999758577081216,
-      "loss": 46.0,
-      "step": 13745
-    },
-    {
-      "epoch": 2.2136962035508674,
-      "grad_norm": 0.0010562578681856394,
-      "learning_rate": 0.00019999758541925665,
-      "loss": 46.0,
-      "step": 13746
-    },
-    {
-      "epoch": 2.213857240629655,
-      "grad_norm": 0.0016176251228898764,
-      "learning_rate": 0.00019999758506767558,
-      "loss": 46.0,
-      "step": 13747
-    },
-    {
-      "epoch": 2.2140182777084423,
-      "grad_norm": 0.008211284875869751,
-      "learning_rate": 0.0001999975847160689,
-      "loss": 46.0,
-      "step": 13748
-    },
-    {
-      "epoch": 2.2141793147872297,
-      "grad_norm": 0.0013719237176701427,
-      "learning_rate": 0.00019999758436443663,
-      "loss": 46.0,
-      "step": 13749
-    },
-    {
-      "epoch": 2.214340351866017,
-      "grad_norm": 0.0012339329114183784,
-      "learning_rate": 0.00019999758401277875,
-      "loss": 46.0,
-      "step": 13750
-    },
-    {
-      "epoch": 2.2145013889448046,
-      "grad_norm": 0.0028864461928606033,
-      "learning_rate": 0.0001999975836610953,
-      "loss": 46.0,
-      "step": 13751
-    },
-    {
-      "epoch": 2.214662426023592,
-      "grad_norm": 0.0033955059479922056,
-      "learning_rate": 0.00019999758330938622,
-      "loss": 46.0,
-      "step": 13752
-    },
-    {
-      "epoch": 2.2148234631023795,
-      "grad_norm": 0.0017166660400107503,
-      "learning_rate": 0.00019999758295765157,
-      "loss": 46.0,
-      "step": 13753
-    },
-    {
-      "epoch": 2.2149845001811665,
-      "grad_norm": 0.0011917110532522202,
-      "learning_rate": 0.00019999758260589134,
-      "loss": 46.0,
-      "step": 13754
-    },
-    {
-      "epoch": 2.215145537259954,
-      "grad_norm": 0.0034384748432785273,
-      "learning_rate": 0.0001999975822541055,
-      "loss": 46.0,
-      "step": 13755
-    },
-    {
-      "epoch": 2.2153065743387415,
-      "grad_norm": 0.003914912696927786,
-      "learning_rate": 0.00019999758190229405,
-      "loss": 46.0,
-      "step": 13756
-    },
-    {
-      "epoch": 2.215467611417529,
-      "grad_norm": 0.0019371293019503355,
-      "learning_rate": 0.00019999758155045703,
-      "loss": 46.0,
-      "step": 13757
-    },
-    {
-      "epoch": 2.2156286484963164,
-      "grad_norm": 0.0005503292195498943,
-      "learning_rate": 0.0001999975811985944,
-      "loss": 46.0,
-      "step": 13758
-    },
-    {
-      "epoch": 2.215789685575104,
-      "grad_norm": 0.005088392645120621,
-      "learning_rate": 0.00019999758084670617,
-      "loss": 46.0,
-      "step": 13759
-    },
-    {
-      "epoch": 2.2159507226538913,
-      "grad_norm": 0.011506959795951843,
-      "learning_rate": 0.00019999758049479236,
-      "loss": 46.0,
-      "step": 13760
-    },
-    {
-      "epoch": 2.2161117597326783,
-      "grad_norm": 0.001494453172199428,
-      "learning_rate": 0.00019999758014285294,
-      "loss": 46.0,
-      "step": 13761
-    },
-    {
-      "epoch": 2.2162727968114657,
-      "grad_norm": 0.0008600427536293864,
-      "learning_rate": 0.00019999757979088795,
-      "loss": 46.0,
-      "step": 13762
-    },
-    {
-      "epoch": 2.216433833890253,
-      "grad_norm": 0.0018290492007508874,
-      "learning_rate": 0.00019999757943889735,
-      "loss": 46.0,
-      "step": 13763
-    },
-    {
-      "epoch": 2.2165948709690406,
-      "grad_norm": 0.004759151488542557,
-      "learning_rate": 0.00019999757908688114,
-      "loss": 46.0,
-      "step": 13764
-    },
-    {
-      "epoch": 2.216755908047828,
-      "grad_norm": 0.007206445559859276,
-      "learning_rate": 0.00019999757873483936,
-      "loss": 46.0,
-      "step": 13765
-    },
-    {
-      "epoch": 2.2169169451266155,
-      "grad_norm": 0.001072881743311882,
-      "learning_rate": 0.000199997578382772,
-      "loss": 46.0,
-      "step": 13766
-    },
-    {
-      "epoch": 2.217077982205403,
-      "grad_norm": 0.002913131844252348,
-      "learning_rate": 0.00019999757803067903,
-      "loss": 46.0,
-      "step": 13767
-    },
-    {
-      "epoch": 2.21723901928419,
-      "grad_norm": 0.0028124507516622543,
-      "learning_rate": 0.00019999757767856044,
-      "loss": 46.0,
-      "step": 13768
-    },
-    {
-      "epoch": 2.2174000563629774,
-      "grad_norm": 0.0023997551761567593,
-      "learning_rate": 0.00019999757732641632,
-      "loss": 46.0,
-      "step": 13769
-    },
-    {
-      "epoch": 2.217561093441765,
-      "grad_norm": 0.004580811131745577,
-      "learning_rate": 0.00019999757697424655,
-      "loss": 46.0,
-      "step": 13770
-    },
-    {
-      "epoch": 2.2177221305205523,
-      "grad_norm": 0.003870446467772126,
-      "learning_rate": 0.00019999757662205118,
-      "loss": 46.0,
-      "step": 13771
-    },
-    {
-      "epoch": 2.21788316759934,
-      "grad_norm": 0.005958774592727423,
-      "learning_rate": 0.00019999757626983024,
-      "loss": 46.0,
-      "step": 13772
-    },
-    {
-      "epoch": 2.2180442046781272,
-      "grad_norm": 0.0019876330625265837,
-      "learning_rate": 0.0001999975759175837,
-      "loss": 46.0,
-      "step": 13773
-    },
-    {
-      "epoch": 2.2182052417569147,
-      "grad_norm": 0.0012807289604097605,
-      "learning_rate": 0.00019999757556531157,
-      "loss": 46.0,
-      "step": 13774
-    },
-    {
-      "epoch": 2.2183662788357017,
-      "grad_norm": 0.010696006007492542,
-      "learning_rate": 0.00019999757521301382,
-      "loss": 46.0,
-      "step": 13775
-    },
-    {
-      "epoch": 2.218527315914489,
-      "grad_norm": 0.001628201687708497,
-      "learning_rate": 0.0001999975748606905,
-      "loss": 46.0,
-      "step": 13776
-    },
-    {
-      "epoch": 2.2186883529932766,
-      "grad_norm": 0.009343422017991543,
-      "learning_rate": 0.0001999975745083416,
-      "loss": 46.0,
-      "step": 13777
-    },
-    {
-      "epoch": 2.218849390072064,
-      "grad_norm": 0.002921203849837184,
-      "learning_rate": 0.0001999975741559671,
-      "loss": 46.0,
-      "step": 13778
-    },
-    {
-      "epoch": 2.2190104271508515,
-      "grad_norm": 0.005728908348828554,
-      "learning_rate": 0.000199997573803567,
-      "loss": 46.0,
-      "step": 13779
-    },
-    {
-      "epoch": 2.219171464229639,
-      "grad_norm": 0.00424401368945837,
-      "learning_rate": 0.00019999757345114127,
-      "loss": 46.0,
-      "step": 13780
-    },
-    {
-      "epoch": 2.2193325013084264,
-      "grad_norm": 0.001296310918405652,
-      "learning_rate": 0.00019999757309869,
-      "loss": 46.0,
-      "step": 13781
-    },
-    {
-      "epoch": 2.219493538387214,
-      "grad_norm": 0.005168564151972532,
-      "learning_rate": 0.0001999975727462131,
-      "loss": 46.0,
-      "step": 13782
-    },
-    {
-      "epoch": 2.219654575466001,
-      "grad_norm": 0.0007294094539247453,
-      "learning_rate": 0.00019999757239371065,
-      "loss": 46.0,
-      "step": 13783
-    },
-    {
-      "epoch": 2.2198156125447883,
-      "grad_norm": 0.010267596691846848,
-      "learning_rate": 0.00019999757204118253,
-      "loss": 46.0,
-      "step": 13784
-    },
-    {
-      "epoch": 2.2199766496235758,
-      "grad_norm": 0.002379964804276824,
-      "learning_rate": 0.00019999757168862888,
-      "loss": 46.0,
-      "step": 13785
-    },
-    {
-      "epoch": 2.2201376867023632,
-      "grad_norm": 0.0021109574008733034,
-      "learning_rate": 0.00019999757133604961,
-      "loss": 46.0,
-      "step": 13786
-    },
-    {
-      "epoch": 2.2202987237811507,
-      "grad_norm": 0.0028596543706953526,
-      "learning_rate": 0.00019999757098344476,
-      "loss": 46.0,
-      "step": 13787
-    },
-    {
-      "epoch": 2.220459760859938,
-      "grad_norm": 0.001897316542454064,
-      "learning_rate": 0.0001999975706308143,
-      "loss": 46.0,
-      "step": 13788
-    },
-    {
-      "epoch": 2.2206207979387256,
-      "grad_norm": 0.002938944846391678,
-      "learning_rate": 0.00019999757027815824,
-      "loss": 46.0,
-      "step": 13789
-    },
-    {
-      "epoch": 2.2207818350175126,
-      "grad_norm": 0.0007160701206885278,
-      "learning_rate": 0.00019999756992547662,
-      "loss": 46.0,
-      "step": 13790
-    },
-    {
-      "epoch": 2.2209428720963,
-      "grad_norm": 0.0088806739076972,
-      "learning_rate": 0.0001999975695727694,
-      "loss": 46.0,
-      "step": 13791
-    },
-    {
-      "epoch": 2.2211039091750875,
-      "grad_norm": 0.005651654209941626,
-      "learning_rate": 0.00019999756922003657,
-      "loss": 46.0,
-      "step": 13792
-    },
-    {
-      "epoch": 2.221264946253875,
-      "grad_norm": 0.0036610146053135395,
-      "learning_rate": 0.00019999756886727812,
-      "loss": 46.0,
-      "step": 13793
-    },
-    {
-      "epoch": 2.2214259833326624,
-      "grad_norm": 0.0004612719058059156,
-      "learning_rate": 0.0001999975685144941,
-      "loss": 46.0,
-      "step": 13794
-    },
-    {
-      "epoch": 2.22158702041145,
-      "grad_norm": 0.004077673889696598,
-      "learning_rate": 0.0001999975681616845,
-      "loss": 46.0,
-      "step": 13795
-    },
-    {
-      "epoch": 2.2217480574902373,
-      "grad_norm": 0.004129249136894941,
-      "learning_rate": 0.00019999756780884927,
-      "loss": 46.0,
-      "step": 13796
-    },
-    {
-      "epoch": 2.2219090945690247,
-      "grad_norm": 0.0007824993226677179,
-      "learning_rate": 0.0001999975674559885,
-      "loss": 46.0,
-      "step": 13797
-    },
-    {
-      "epoch": 2.2220701316478118,
-      "grad_norm": 0.009709149599075317,
-      "learning_rate": 0.0001999975671031021,
-      "loss": 46.0,
-      "step": 13798
-    },
-    {
-      "epoch": 2.222231168726599,
-      "grad_norm": 0.0019398038275539875,
-      "learning_rate": 0.0001999975667501901,
-      "loss": 46.0,
-      "step": 13799
-    },
-    {
-      "epoch": 2.2223922058053867,
-      "grad_norm": 0.0011810872238129377,
-      "learning_rate": 0.00019999756639725252,
-      "loss": 46.0,
-      "step": 13800
-    },
-    {
-      "epoch": 2.222553242884174,
-      "grad_norm": 0.009734337218105793,
-      "learning_rate": 0.00019999756604428934,
-      "loss": 46.0,
-      "step": 13801
-    },
-    {
-      "epoch": 2.2227142799629616,
-      "grad_norm": 0.004123416729271412,
-      "learning_rate": 0.00019999756569130057,
-      "loss": 46.0,
-      "step": 13802
-    },
-    {
-      "epoch": 2.222875317041749,
-      "grad_norm": 0.0038641542196273804,
-      "learning_rate": 0.00019999756533828618,
-      "loss": 46.0,
-      "step": 13803
-    },
-    {
-      "epoch": 2.223036354120536,
-      "grad_norm": 0.005792050156742334,
-      "learning_rate": 0.00019999756498524624,
-      "loss": 46.0,
-      "step": 13804
-    },
-    {
-      "epoch": 2.2231973911993235,
-      "grad_norm": 0.002649559173732996,
-      "learning_rate": 0.00019999756463218068,
-      "loss": 46.0,
-      "step": 13805
-    },
-    {
-      "epoch": 2.223358428278111,
-      "grad_norm": 0.002285517519339919,
-      "learning_rate": 0.00019999756427908954,
-      "loss": 46.0,
-      "step": 13806
-    },
-    {
-      "epoch": 2.2235194653568984,
-      "grad_norm": 0.004329164978116751,
-      "learning_rate": 0.0001999975639259728,
-      "loss": 46.0,
-      "step": 13807
-    },
-    {
-      "epoch": 2.223680502435686,
-      "grad_norm": 0.006754350382834673,
-      "learning_rate": 0.00019999756357283046,
-      "loss": 46.0,
-      "step": 13808
-    },
-    {
-      "epoch": 2.2238415395144733,
-      "grad_norm": 0.00586355896666646,
-      "learning_rate": 0.00019999756321966252,
-      "loss": 46.0,
-      "step": 13809
-    },
-    {
-      "epoch": 2.2240025765932607,
-      "grad_norm": 0.0006915184203535318,
-      "learning_rate": 0.000199997562866469,
-      "loss": 46.0,
-      "step": 13810
-    },
-    {
-      "epoch": 2.224163613672048,
-      "grad_norm": 0.0025954199954867363,
-      "learning_rate": 0.0001999975625132499,
-      "loss": 46.0,
-      "step": 13811
-    },
-    {
-      "epoch": 2.224324650750835,
-      "grad_norm": 0.004700150340795517,
-      "learning_rate": 0.00019999756216000517,
-      "loss": 46.0,
-      "step": 13812
-    },
-    {
-      "epoch": 2.2244856878296226,
-      "grad_norm": 0.002075010910630226,
-      "learning_rate": 0.00019999756180673486,
-      "loss": 46.0,
-      "step": 13813
-    },
-    {
-      "epoch": 2.22464672490841,
-      "grad_norm": 0.001411769655533135,
-      "learning_rate": 0.00019999756145343896,
-      "loss": 46.0,
-      "step": 13814
-    },
-    {
-      "epoch": 2.2248077619871975,
-      "grad_norm": 0.0015281918458640575,
-      "learning_rate": 0.00019999756110011745,
-      "loss": 46.0,
-      "step": 13815
-    },
-    {
-      "epoch": 2.224968799065985,
-      "grad_norm": 0.0006174591835588217,
-      "learning_rate": 0.00019999756074677035,
-      "loss": 46.0,
-      "step": 13816
-    },
-    {
-      "epoch": 2.2251298361447724,
-      "grad_norm": 0.011503061279654503,
-      "learning_rate": 0.00019999756039339766,
-      "loss": 46.0,
-      "step": 13817
-    },
-    {
-      "epoch": 2.22529087322356,
-      "grad_norm": 0.0024040464777499437,
-      "learning_rate": 0.0001999975600399994,
-      "loss": 46.0,
-      "step": 13818
-    },
-    {
-      "epoch": 2.225451910302347,
-      "grad_norm": 0.006287943106144667,
-      "learning_rate": 0.00019999755968657553,
-      "loss": 46.0,
-      "step": 13819
-    },
-    {
-      "epoch": 2.2256129473811344,
-      "grad_norm": 0.001936339191161096,
-      "learning_rate": 0.00019999755933312606,
-      "loss": 46.0,
-      "step": 13820
-    },
-    {
-      "epoch": 2.225773984459922,
-      "grad_norm": 0.0013389616506174207,
-      "learning_rate": 0.000199997558979651,
-      "loss": 46.0,
-      "step": 13821
-    },
-    {
-      "epoch": 2.2259350215387093,
-      "grad_norm": 0.000814276107121259,
-      "learning_rate": 0.00019999755862615032,
-      "loss": 46.0,
-      "step": 13822
-    },
-    {
-      "epoch": 2.2260960586174967,
-      "grad_norm": 0.0010869288817048073,
-      "learning_rate": 0.00019999755827262408,
-      "loss": 46.0,
-      "step": 13823
-    },
-    {
-      "epoch": 2.226257095696284,
-      "grad_norm": 0.0013230699114501476,
-      "learning_rate": 0.00019999755791907223,
-      "loss": 46.0,
-      "step": 13824
-    },
-    {
-      "epoch": 2.2264181327750716,
-      "grad_norm": 0.014168528839945793,
-      "learning_rate": 0.0001999975575654948,
-      "loss": 46.0,
-      "step": 13825
-    },
-    {
-      "epoch": 2.226579169853859,
-      "grad_norm": 0.007537615951150656,
-      "learning_rate": 0.00019999755721189177,
-      "loss": 46.0,
-      "step": 13826
-    },
-    {
-      "epoch": 2.226740206932646,
-      "grad_norm": 0.012373830191791058,
-      "learning_rate": 0.00019999755685826313,
-      "loss": 46.0,
-      "step": 13827
-    },
-    {
-      "epoch": 2.2269012440114335,
-      "grad_norm": 0.00253257411532104,
-      "learning_rate": 0.00019999755650460893,
-      "loss": 46.0,
-      "step": 13828
-    },
-    {
-      "epoch": 2.227062281090221,
-      "grad_norm": 0.0018066565971821547,
-      "learning_rate": 0.00019999755615092906,
-      "loss": 46.0,
-      "step": 13829
-    },
-    {
-      "epoch": 2.2272233181690084,
-      "grad_norm": 0.000954370538238436,
-      "learning_rate": 0.0001999975557972237,
-      "loss": 46.0,
-      "step": 13830
-    },
-    {
-      "epoch": 2.227384355247796,
-      "grad_norm": 0.007515036500990391,
-      "learning_rate": 0.00019999755544349268,
-      "loss": 46.0,
-      "step": 13831
-    },
-    {
-      "epoch": 2.2275453923265833,
-      "grad_norm": 0.0008438150398433208,
-      "learning_rate": 0.00019999755508973607,
-      "loss": 46.0,
-      "step": 13832
-    },
-    {
-      "epoch": 2.227706429405371,
-      "grad_norm": 0.0064139775931835175,
-      "learning_rate": 0.00019999755473595389,
-      "loss": 46.0,
-      "step": 13833
-    },
-    {
-      "epoch": 2.227867466484158,
-      "grad_norm": 0.006094703450798988,
-      "learning_rate": 0.00019999755438214608,
-      "loss": 46.0,
-      "step": 13834
-    },
-    {
-      "epoch": 2.2280285035629452,
-      "grad_norm": 0.0032856641337275505,
-      "learning_rate": 0.00019999755402831272,
-      "loss": 46.0,
-      "step": 13835
-    },
-    {
-      "epoch": 2.2281895406417327,
-      "grad_norm": 0.003365366719663143,
-      "learning_rate": 0.00019999755367445374,
-      "loss": 46.0,
-      "step": 13836
-    },
-    {
-      "epoch": 2.22835057772052,
-      "grad_norm": 0.0020414208993315697,
-      "learning_rate": 0.00019999755332056918,
-      "loss": 46.0,
-      "step": 13837
-    },
-    {
-      "epoch": 2.2285116147993076,
-      "grad_norm": 0.0013070415006950498,
-      "learning_rate": 0.000199997552966659,
-      "loss": 46.0,
-      "step": 13838
-    },
-    {
-      "epoch": 2.228672651878095,
-      "grad_norm": 0.0014498542295768857,
-      "learning_rate": 0.00019999755261272323,
-      "loss": 46.0,
-      "step": 13839
-    },
-    {
-      "epoch": 2.2288336889568825,
-      "grad_norm": 0.004660328384488821,
-      "learning_rate": 0.00019999755225876188,
-      "loss": 46.0,
-      "step": 13840
-    },
-    {
-      "epoch": 2.22899472603567,
-      "grad_norm": 0.002556582447141409,
-      "learning_rate": 0.00019999755190477494,
-      "loss": 46.0,
-      "step": 13841
-    },
-    {
-      "epoch": 2.229155763114457,
-      "grad_norm": 0.0025003976188600063,
-      "learning_rate": 0.00019999755155076238,
-      "loss": 46.0,
-      "step": 13842
-    },
-    {
-      "epoch": 2.2293168001932444,
-      "grad_norm": 0.005243140272796154,
-      "learning_rate": 0.00019999755119672427,
-      "loss": 46.0,
-      "step": 13843
-    },
-    {
-      "epoch": 2.229477837272032,
-      "grad_norm": 0.002213727915659547,
-      "learning_rate": 0.00019999755084266054,
-      "loss": 46.0,
-      "step": 13844
-    },
-    {
-      "epoch": 2.2296388743508193,
-      "grad_norm": 0.0006454632384702563,
-      "learning_rate": 0.0001999975504885712,
-      "loss": 46.0,
-      "step": 13845
-    },
-    {
-      "epoch": 2.2297999114296068,
-      "grad_norm": 0.0009064690093509853,
-      "learning_rate": 0.0001999975501344563,
-      "loss": 46.0,
-      "step": 13846
-    },
-    {
-      "epoch": 2.229960948508394,
-      "grad_norm": 0.0014630145160481334,
-      "learning_rate": 0.00019999754978031577,
-      "loss": 46.0,
-      "step": 13847
-    },
-    {
-      "epoch": 2.230121985587181,
-      "grad_norm": 0.003800722537562251,
-      "learning_rate": 0.00019999754942614967,
-      "loss": 46.0,
-      "step": 13848
-    },
-    {
-      "epoch": 2.2302830226659687,
-      "grad_norm": 0.001235337695106864,
-      "learning_rate": 0.00019999754907195795,
-      "loss": 46.0,
-      "step": 13849
-    },
-    {
-      "epoch": 2.230444059744756,
-      "grad_norm": 0.0037756022065877914,
-      "learning_rate": 0.00019999754871774067,
-      "loss": 46.0,
-      "step": 13850
-    },
-    {
-      "epoch": 2.2306050968235436,
-      "grad_norm": 0.002481248462572694,
-      "learning_rate": 0.00019999754836349777,
-      "loss": 46.0,
-      "step": 13851
-    },
-    {
-      "epoch": 2.230766133902331,
-      "grad_norm": 0.004169543739408255,
-      "learning_rate": 0.00019999754800922927,
-      "loss": 46.0,
-      "step": 13852
-    },
-    {
-      "epoch": 2.2309271709811185,
-      "grad_norm": 0.0013243324356153607,
-      "learning_rate": 0.00019999754765493523,
-      "loss": 46.0,
-      "step": 13853
-    },
-    {
-      "epoch": 2.231088208059906,
-      "grad_norm": 0.0020127661991864443,
-      "learning_rate": 0.00019999754730061552,
-      "loss": 46.0,
-      "step": 13854
-    },
-    {
-      "epoch": 2.2312492451386934,
-      "grad_norm": 0.0019206737633794546,
-      "learning_rate": 0.00019999754694627027,
-      "loss": 46.0,
-      "step": 13855
-    },
-    {
-      "epoch": 2.2314102822174804,
-      "grad_norm": 0.0027539818547666073,
-      "learning_rate": 0.00019999754659189942,
-      "loss": 46.0,
-      "step": 13856
-    },
-    {
-      "epoch": 2.231571319296268,
-      "grad_norm": 0.010542550124228,
-      "learning_rate": 0.00019999754623750295,
-      "loss": 46.0,
-      "step": 13857
-    },
-    {
-      "epoch": 2.2317323563750553,
-      "grad_norm": 0.0016768169589340687,
-      "learning_rate": 0.00019999754588308091,
-      "loss": 46.0,
-      "step": 13858
-    },
-    {
-      "epoch": 2.2318933934538427,
-      "grad_norm": 0.0043511539697647095,
-      "learning_rate": 0.00019999754552863327,
-      "loss": 46.0,
-      "step": 13859
-    },
-    {
-      "epoch": 2.23205443053263,
-      "grad_norm": 0.000767124816775322,
-      "learning_rate": 0.00019999754517416004,
-      "loss": 46.0,
-      "step": 13860
-    },
-    {
-      "epoch": 2.2322154676114176,
-      "grad_norm": 0.006702875252813101,
-      "learning_rate": 0.0001999975448196612,
-      "loss": 46.0,
-      "step": 13861
-    },
-    {
-      "epoch": 2.232376504690205,
-      "grad_norm": 0.0034737042151391506,
-      "learning_rate": 0.00019999754446513676,
-      "loss": 46.0,
-      "step": 13862
-    },
-    {
-      "epoch": 2.232537541768992,
-      "grad_norm": 0.001093236729502678,
-      "learning_rate": 0.00019999754411058676,
-      "loss": 46.0,
-      "step": 13863
-    },
-    {
-      "epoch": 2.2326985788477796,
-      "grad_norm": 0.008258485235273838,
-      "learning_rate": 0.00019999754375601115,
-      "loss": 46.0,
-      "step": 13864
-    },
-    {
-      "epoch": 2.232859615926567,
-      "grad_norm": 0.00031749665504321456,
-      "learning_rate": 0.0001999975434014099,
-      "loss": 46.0,
-      "step": 13865
-    },
-    {
-      "epoch": 2.2330206530053545,
-      "grad_norm": 0.0024656273890286684,
-      "learning_rate": 0.00019999754304678312,
-      "loss": 46.0,
-      "step": 13866
-    },
-    {
-      "epoch": 2.233181690084142,
-      "grad_norm": 0.0011689583770930767,
-      "learning_rate": 0.00019999754269213073,
-      "loss": 46.0,
-      "step": 13867
-    },
-    {
-      "epoch": 2.2333427271629294,
-      "grad_norm": 0.005402153357863426,
-      "learning_rate": 0.00019999754233745271,
-      "loss": 46.0,
-      "step": 13868
-    },
-    {
-      "epoch": 2.233503764241717,
-      "grad_norm": 0.0008777216426096857,
-      "learning_rate": 0.00019999754198274914,
-      "loss": 46.0,
-      "step": 13869
-    },
-    {
-      "epoch": 2.2336648013205043,
-      "grad_norm": 0.0011804127134382725,
-      "learning_rate": 0.00019999754162801996,
-      "loss": 46.0,
-      "step": 13870
-    },
-    {
-      "epoch": 2.2338258383992913,
-      "grad_norm": 0.0028473285492509604,
-      "learning_rate": 0.00019999754127326518,
-      "loss": 46.0,
-      "step": 13871
-    },
-    {
-      "epoch": 2.2339868754780787,
-      "grad_norm": 0.0008439497323706746,
-      "learning_rate": 0.0001999975409184848,
-      "loss": 46.0,
-      "step": 13872
-    },
-    {
-      "epoch": 2.234147912556866,
-      "grad_norm": 0.002870025811716914,
-      "learning_rate": 0.00019999754056367882,
-      "loss": 46.0,
-      "step": 13873
-    },
-    {
-      "epoch": 2.2343089496356536,
-      "grad_norm": 0.00207132613286376,
-      "learning_rate": 0.0001999975402088473,
-      "loss": 46.0,
-      "step": 13874
-    },
-    {
-      "epoch": 2.234469986714441,
-      "grad_norm": 0.0046815224923193455,
-      "learning_rate": 0.00019999753985399011,
-      "loss": 46.0,
-      "step": 13875
-    },
-    {
-      "epoch": 2.2346310237932285,
-      "grad_norm": 0.002843681490048766,
-      "learning_rate": 0.00019999753949910738,
-      "loss": 46.0,
-      "step": 13876
-    },
-    {
-      "epoch": 2.234792060872016,
-      "grad_norm": 0.001463265623897314,
-      "learning_rate": 0.00019999753914419906,
-      "loss": 46.0,
-      "step": 13877
-    },
-    {
-      "epoch": 2.234953097950803,
-      "grad_norm": 0.008958492428064346,
-      "learning_rate": 0.0001999975387892651,
-      "loss": 46.0,
-      "step": 13878
-    },
-    {
-      "epoch": 2.2351141350295904,
-      "grad_norm": 0.00046426264452748,
-      "learning_rate": 0.00019999753843430557,
-      "loss": 46.0,
-      "step": 13879
-    },
-    {
-      "epoch": 2.235275172108378,
-      "grad_norm": 0.0009298831573687494,
-      "learning_rate": 0.00019999753807932046,
-      "loss": 46.0,
-      "step": 13880
-    },
-    {
-      "epoch": 2.2354362091871653,
-      "grad_norm": 0.0017463818658143282,
-      "learning_rate": 0.00019999753772430973,
-      "loss": 46.0,
-      "step": 13881
-    },
-    {
-      "epoch": 2.235597246265953,
-      "grad_norm": 0.0019432863919064403,
-      "learning_rate": 0.00019999753736927342,
-      "loss": 46.0,
-      "step": 13882
-    },
-    {
-      "epoch": 2.2357582833447402,
-      "grad_norm": 0.0020308897364884615,
-      "learning_rate": 0.00019999753701421154,
-      "loss": 46.0,
-      "step": 13883
-    },
-    {
-      "epoch": 2.2359193204235277,
-      "grad_norm": 0.008548577316105366,
-      "learning_rate": 0.00019999753665912403,
-      "loss": 46.0,
-      "step": 13884
-    },
-    {
-      "epoch": 2.2360803575023147,
-      "grad_norm": 0.0007212921627797186,
-      "learning_rate": 0.00019999753630401093,
-      "loss": 46.0,
-      "step": 13885
-    },
-    {
-      "epoch": 2.236241394581102,
-      "grad_norm": 0.0028578033670783043,
-      "learning_rate": 0.00019999753594887224,
-      "loss": 46.0,
-      "step": 13886
-    },
-    {
-      "epoch": 2.2364024316598896,
-      "grad_norm": 0.004118448123335838,
-      "learning_rate": 0.00019999753559370796,
-      "loss": 46.0,
-      "step": 13887
-    },
-    {
-      "epoch": 2.236563468738677,
-      "grad_norm": 0.0014290903927758336,
-      "learning_rate": 0.00019999753523851807,
-      "loss": 46.0,
-      "step": 13888
-    },
-    {
-      "epoch": 2.2367245058174645,
-      "grad_norm": 0.005826373118907213,
-      "learning_rate": 0.00019999753488330262,
-      "loss": 46.0,
-      "step": 13889
-    },
-    {
-      "epoch": 2.236885542896252,
-      "grad_norm": 0.004704639315605164,
-      "learning_rate": 0.00019999753452806153,
-      "loss": 46.0,
-      "step": 13890
-    },
-    {
-      "epoch": 2.2370465799750394,
-      "grad_norm": 0.0041178148239851,
-      "learning_rate": 0.00019999753417279488,
-      "loss": 46.0,
-      "step": 13891
-    },
-    {
-      "epoch": 2.2372076170538264,
-      "grad_norm": 0.008068711496889591,
-      "learning_rate": 0.00019999753381750264,
-      "loss": 46.0,
-      "step": 13892
-    },
-    {
-      "epoch": 2.237368654132614,
-      "grad_norm": 0.001038464717566967,
-      "learning_rate": 0.00019999753346218476,
-      "loss": 46.0,
-      "step": 13893
-    },
-    {
-      "epoch": 2.2375296912114013,
-      "grad_norm": 0.0023721521720290184,
-      "learning_rate": 0.00019999753310684132,
-      "loss": 46.0,
-      "step": 13894
-    },
-    {
-      "epoch": 2.237690728290189,
-      "grad_norm": 0.004016389138996601,
-      "learning_rate": 0.0001999975327514723,
-      "loss": 46.0,
-      "step": 13895
-    },
-    {
-      "epoch": 2.2378517653689762,
-      "grad_norm": 0.0006583682843483984,
-      "learning_rate": 0.00019999753239607765,
-      "loss": 46.0,
-      "step": 13896
-    },
-    {
-      "epoch": 2.2380128024477637,
-      "grad_norm": 0.001595151494257152,
-      "learning_rate": 0.00019999753204065742,
-      "loss": 46.0,
-      "step": 13897
-    },
-    {
-      "epoch": 2.238173839526551,
-      "grad_norm": 0.0033382533583790064,
-      "learning_rate": 0.00019999753168521163,
-      "loss": 46.0,
-      "step": 13898
-    },
-    {
-      "epoch": 2.2383348766053386,
-      "grad_norm": 0.003177135717123747,
-      "learning_rate": 0.0001999975313297402,
-      "loss": 46.0,
-      "step": 13899
-    },
-    {
-      "epoch": 2.2384959136841256,
-      "grad_norm": 0.003992767538875341,
-      "learning_rate": 0.00019999753097424321,
-      "loss": 46.0,
-      "step": 13900
-    },
-    {
-      "epoch": 2.238656950762913,
-      "grad_norm": 0.0021767315920442343,
-      "learning_rate": 0.0001999975306187206,
-      "loss": 46.0,
-      "step": 13901
-    },
-    {
-      "epoch": 2.2388179878417005,
-      "grad_norm": 0.00391074363142252,
-      "learning_rate": 0.00019999753026317242,
-      "loss": 46.0,
-      "step": 13902
-    },
-    {
-      "epoch": 2.238979024920488,
-      "grad_norm": 0.004004878457635641,
-      "learning_rate": 0.0001999975299075986,
-      "loss": 46.0,
-      "step": 13903
-    },
-    {
-      "epoch": 2.2391400619992754,
-      "grad_norm": 0.002291835378855467,
-      "learning_rate": 0.00019999752955199922,
-      "loss": 46.0,
-      "step": 13904
-    },
-    {
-      "epoch": 2.239301099078063,
-      "grad_norm": 0.0015953409019857645,
-      "learning_rate": 0.00019999752919637427,
-      "loss": 46.0,
-      "step": 13905
-    },
-    {
-      "epoch": 2.2394621361568503,
-      "grad_norm": 0.004095470067113638,
-      "learning_rate": 0.0001999975288407237,
-      "loss": 46.0,
-      "step": 13906
-    },
-    {
-      "epoch": 2.2396231732356373,
-      "grad_norm": 0.0016459259204566479,
-      "learning_rate": 0.00019999752848504752,
-      "loss": 46.0,
-      "step": 13907
-    },
-    {
-      "epoch": 2.2397842103144248,
-      "grad_norm": 0.0006697041681036353,
-      "learning_rate": 0.00019999752812934575,
-      "loss": 46.0,
-      "step": 13908
-    },
-    {
-      "epoch": 2.239945247393212,
-      "grad_norm": 0.0038877055048942566,
-      "learning_rate": 0.0001999975277736184,
-      "loss": 46.0,
-      "step": 13909
-    },
-    {
-      "epoch": 2.2401062844719997,
-      "grad_norm": 0.0023673148825764656,
-      "learning_rate": 0.00019999752741786545,
-      "loss": 46.0,
-      "step": 13910
-    },
-    {
-      "epoch": 2.240267321550787,
-      "grad_norm": 0.0030799394007772207,
-      "learning_rate": 0.0001999975270620869,
-      "loss": 46.0,
-      "step": 13911
-    },
-    {
-      "epoch": 2.2404283586295746,
-      "grad_norm": 0.00277404161170125,
-      "learning_rate": 0.00019999752670628274,
-      "loss": 46.0,
-      "step": 13912
-    },
-    {
-      "epoch": 2.240589395708362,
-      "grad_norm": 0.008009926415979862,
-      "learning_rate": 0.00019999752635045304,
-      "loss": 46.0,
-      "step": 13913
-    },
-    {
-      "epoch": 2.2407504327871495,
-      "grad_norm": 0.004241722635924816,
-      "learning_rate": 0.0001999975259945977,
-      "loss": 46.0,
-      "step": 13914
-    },
-    {
-      "epoch": 2.2409114698659365,
-      "grad_norm": 0.001049460144713521,
-      "learning_rate": 0.00019999752563871676,
-      "loss": 46.0,
-      "step": 13915
-    },
-    {
-      "epoch": 2.241072506944724,
-      "grad_norm": 0.0017421930097043514,
-      "learning_rate": 0.00019999752528281027,
-      "loss": 46.0,
-      "step": 13916
-    },
-    {
-      "epoch": 2.2412335440235114,
-      "grad_norm": 0.0014313283609226346,
-      "learning_rate": 0.00019999752492687816,
-      "loss": 46.0,
-      "step": 13917
-    },
-    {
-      "epoch": 2.241394581102299,
-      "grad_norm": 0.0008132740622386336,
-      "learning_rate": 0.00019999752457092044,
-      "loss": 46.0,
-      "step": 13918
-    },
-    {
-      "epoch": 2.2415556181810863,
-      "grad_norm": 0.007433802355080843,
-      "learning_rate": 0.00019999752421493713,
-      "loss": 46.0,
-      "step": 13919
-    },
-    {
-      "epoch": 2.2417166552598737,
-      "grad_norm": 0.0022407250944525003,
-      "learning_rate": 0.00019999752385892826,
-      "loss": 46.0,
-      "step": 13920
-    },
-    {
-      "epoch": 2.2418776923386607,
-      "grad_norm": 0.004558785352855921,
-      "learning_rate": 0.00019999752350289377,
-      "loss": 46.0,
-      "step": 13921
-    },
-    {
-      "epoch": 2.242038729417448,
-      "grad_norm": 0.001668662647716701,
-      "learning_rate": 0.00019999752314683368,
-      "loss": 46.0,
-      "step": 13922
-    },
-    {
-      "epoch": 2.2421997664962356,
-      "grad_norm": 0.0007237204699777067,
-      "learning_rate": 0.00019999752279074802,
-      "loss": 46.0,
-      "step": 13923
-    },
-    {
-      "epoch": 2.242360803575023,
-      "grad_norm": 0.00036969332722947,
-      "learning_rate": 0.00019999752243463675,
-      "loss": 46.0,
-      "step": 13924
-    },
-    {
-      "epoch": 2.2425218406538105,
-      "grad_norm": 0.0026159719564020634,
-      "learning_rate": 0.0001999975220784999,
-      "loss": 46.0,
-      "step": 13925
-    },
-    {
-      "epoch": 2.242682877732598,
-      "grad_norm": 0.008634045720100403,
-      "learning_rate": 0.00019999752172233741,
-      "loss": 46.0,
-      "step": 13926
-    },
-    {
-      "epoch": 2.2428439148113855,
-      "grad_norm": 0.007141964975744486,
-      "learning_rate": 0.00019999752136614935,
-      "loss": 46.0,
-      "step": 13927
-    },
-    {
-      "epoch": 2.243004951890173,
-      "grad_norm": 0.00138210563454777,
-      "learning_rate": 0.0001999975210099357,
-      "loss": 46.0,
-      "step": 13928
-    },
-    {
-      "epoch": 2.24316598896896,
-      "grad_norm": 0.0020694248378276825,
-      "learning_rate": 0.00019999752065369647,
-      "loss": 46.0,
-      "step": 13929
-    },
-    {
-      "epoch": 2.2433270260477474,
-      "grad_norm": 0.005013769958168268,
-      "learning_rate": 0.00019999752029743167,
-      "loss": 46.0,
-      "step": 13930
-    },
-    {
-      "epoch": 2.243488063126535,
-      "grad_norm": 0.006472241133451462,
-      "learning_rate": 0.0001999975199411412,
-      "loss": 46.0,
-      "step": 13931
-    },
-    {
-      "epoch": 2.2436491002053223,
-      "grad_norm": 0.0019514893647283316,
-      "learning_rate": 0.0001999975195848252,
-      "loss": 46.0,
-      "step": 13932
-    },
-    {
-      "epoch": 2.2438101372841097,
-      "grad_norm": 0.0032446354161947966,
-      "learning_rate": 0.00019999751922848358,
-      "loss": 46.0,
-      "step": 13933
-    },
-    {
-      "epoch": 2.243971174362897,
-      "grad_norm": 0.0012381155975162983,
-      "learning_rate": 0.00019999751887211638,
-      "loss": 46.0,
-      "step": 13934
-    },
-    {
-      "epoch": 2.2441322114416846,
-      "grad_norm": 0.0012549014063552022,
-      "learning_rate": 0.00019999751851572354,
-      "loss": 46.0,
-      "step": 13935
-    },
-    {
-      "epoch": 2.2442932485204716,
-      "grad_norm": 0.0014573488151654601,
-      "learning_rate": 0.00019999751815930514,
-      "loss": 46.0,
-      "step": 13936
-    },
-    {
-      "epoch": 2.244454285599259,
-      "grad_norm": 0.005086123943328857,
-      "learning_rate": 0.00019999751780286115,
-      "loss": 46.0,
-      "step": 13937
-    },
-    {
-      "epoch": 2.2446153226780465,
-      "grad_norm": 0.0057265725918114185,
-      "learning_rate": 0.00019999751744639158,
-      "loss": 46.0,
-      "step": 13938
-    },
-    {
-      "epoch": 2.244776359756834,
-      "grad_norm": 0.002571030519902706,
-      "learning_rate": 0.0001999975170898964,
-      "loss": 46.0,
-      "step": 13939
-    },
-    {
-      "epoch": 2.2449373968356214,
-      "grad_norm": 0.00338509283028543,
-      "learning_rate": 0.00019999751673337561,
-      "loss": 46.0,
-      "step": 13940
-    },
-    {
-      "epoch": 2.245098433914409,
-      "grad_norm": 0.0008936020894907415,
-      "learning_rate": 0.00019999751637682925,
-      "loss": 46.0,
-      "step": 13941
-    },
-    {
-      "epoch": 2.2452594709931963,
-      "grad_norm": 0.002808912890031934,
-      "learning_rate": 0.00019999751602025728,
-      "loss": 46.0,
-      "step": 13942
-    },
-    {
-      "epoch": 2.245420508071984,
-      "grad_norm": 0.00510428985580802,
-      "learning_rate": 0.0001999975156636597,
-      "loss": 46.0,
-      "step": 13943
-    },
-    {
-      "epoch": 2.245581545150771,
-      "grad_norm": 0.00565666938200593,
-      "learning_rate": 0.00019999751530703656,
-      "loss": 46.0,
-      "step": 13944
-    },
-    {
-      "epoch": 2.2457425822295582,
-      "grad_norm": 0.0033489130437374115,
-      "learning_rate": 0.00019999751495038782,
-      "loss": 46.0,
-      "step": 13945
-    },
-    {
-      "epoch": 2.2459036193083457,
-      "grad_norm": 0.0011091098422184587,
-      "learning_rate": 0.00019999751459371344,
-      "loss": 46.0,
-      "step": 13946
-    },
-    {
-      "epoch": 2.246064656387133,
-      "grad_norm": 0.0026009439025074244,
-      "learning_rate": 0.00019999751423701353,
-      "loss": 46.0,
-      "step": 13947
-    },
-    {
-      "epoch": 2.2462256934659206,
-      "grad_norm": 0.005592728033661842,
-      "learning_rate": 0.000199997513880288,
-      "loss": 46.0,
-      "step": 13948
-    },
-    {
-      "epoch": 2.246386730544708,
-      "grad_norm": 0.0036806659772992134,
-      "learning_rate": 0.0001999975135235369,
-      "loss": 46.0,
-      "step": 13949
-    },
-    {
-      "epoch": 2.2465477676234955,
-      "grad_norm": 0.0007729671779088676,
-      "learning_rate": 0.00019999751316676013,
-      "loss": 46.0,
-      "step": 13950
-    },
-    {
-      "epoch": 2.2467088047022825,
-      "grad_norm": 0.010362662374973297,
-      "learning_rate": 0.00019999751280995784,
-      "loss": 46.0,
-      "step": 13951
-    },
-    {
-      "epoch": 2.24686984178107,
-      "grad_norm": 0.0007634959765709937,
-      "learning_rate": 0.00019999751245312991,
-      "loss": 46.0,
-      "step": 13952
-    },
-    {
-      "epoch": 2.2470308788598574,
-      "grad_norm": 0.0015676041366532445,
-      "learning_rate": 0.00019999751209627645,
-      "loss": 46.0,
-      "step": 13953
-    },
-    {
-      "epoch": 2.247191915938645,
-      "grad_norm": 0.007791996467858553,
-      "learning_rate": 0.00019999751173939735,
-      "loss": 46.0,
-      "step": 13954
-    },
-    {
-      "epoch": 2.2473529530174323,
-      "grad_norm": 0.003171280026435852,
-      "learning_rate": 0.00019999751138249266,
-      "loss": 46.0,
-      "step": 13955
-    },
-    {
-      "epoch": 2.2475139900962198,
-      "grad_norm": 0.0008149652276188135,
-      "learning_rate": 0.00019999751102556235,
-      "loss": 46.0,
-      "step": 13956
-    },
-    {
-      "epoch": 2.2476750271750072,
-      "grad_norm": 0.0016248066676780581,
-      "learning_rate": 0.00019999751066860648,
-      "loss": 46.0,
-      "step": 13957
-    },
-    {
-      "epoch": 2.2478360642537942,
-      "grad_norm": 0.0009147682576440275,
-      "learning_rate": 0.000199997510311625,
-      "loss": 46.0,
-      "step": 13958
-    },
-    {
-      "epoch": 2.2479971013325817,
-      "grad_norm": 0.004151404835283756,
-      "learning_rate": 0.00019999750995461794,
-      "loss": 46.0,
-      "step": 13959
-    },
-    {
-      "epoch": 2.248158138411369,
-      "grad_norm": 0.0007496776524931192,
-      "learning_rate": 0.00019999750959758528,
-      "loss": 46.0,
-      "step": 13960
-    },
-    {
-      "epoch": 2.2483191754901566,
-      "grad_norm": 0.002037890488281846,
-      "learning_rate": 0.00019999750924052701,
-      "loss": 46.0,
-      "step": 13961
-    },
-    {
-      "epoch": 2.248480212568944,
-      "grad_norm": 0.0019364447798579931,
-      "learning_rate": 0.00019999750888344316,
-      "loss": 46.0,
-      "step": 13962
-    },
-    {
-      "epoch": 2.2486412496477315,
-      "grad_norm": 0.005926021374762058,
-      "learning_rate": 0.00019999750852633372,
-      "loss": 46.0,
-      "step": 13963
-    },
-    {
-      "epoch": 2.248802286726519,
-      "grad_norm": 0.0032404763624072075,
-      "learning_rate": 0.00019999750816919869,
-      "loss": 46.0,
-      "step": 13964
-    },
-    {
-      "epoch": 2.248963323805306,
-      "grad_norm": 0.009493254125118256,
-      "learning_rate": 0.00019999750781203804,
-      "loss": 46.0,
-      "step": 13965
-    },
-    {
-      "epoch": 2.2491243608840934,
-      "grad_norm": 0.004816398024559021,
-      "learning_rate": 0.0001999975074548518,
-      "loss": 46.0,
-      "step": 13966
-    },
-    {
-      "epoch": 2.249285397962881,
-      "grad_norm": 0.001781512750312686,
-      "learning_rate": 0.00019999750709764,
-      "loss": 46.0,
-      "step": 13967
-    },
-    {
-      "epoch": 2.2494464350416683,
-      "grad_norm": 0.005384115967899561,
-      "learning_rate": 0.00019999750674040259,
-      "loss": 46.0,
-      "step": 13968
-    },
-    {
-      "epoch": 2.2496074721204558,
-      "grad_norm": 0.0015986048383638263,
-      "learning_rate": 0.0001999975063831396,
-      "loss": 46.0,
-      "step": 13969
-    },
-    {
-      "epoch": 2.249768509199243,
-      "grad_norm": 0.0026110601611435413,
-      "learning_rate": 0.00019999750602585096,
-      "loss": 46.0,
-      "step": 13970
-    },
-    {
-      "epoch": 2.2499295462780307,
-      "grad_norm": 0.01227173488587141,
-      "learning_rate": 0.00019999750566853676,
-      "loss": 46.0,
-      "step": 13971
-    },
-    {
-      "epoch": 2.250090583356818,
-      "grad_norm": 0.0045133428648114204,
-      "learning_rate": 0.00019999750531119695,
-      "loss": 46.0,
-      "step": 13972
-    },
-    {
-      "epoch": 2.250251620435605,
-      "grad_norm": 0.0007369336090050638,
-      "learning_rate": 0.00019999750495383159,
-      "loss": 46.0,
-      "step": 13973
-    },
-    {
-      "epoch": 2.2504126575143926,
-      "grad_norm": 0.0014093368081375957,
-      "learning_rate": 0.00019999750459644063,
-      "loss": 46.0,
-      "step": 13974
-    },
-    {
-      "epoch": 2.25057369459318,
-      "grad_norm": 0.0006778715178370476,
-      "learning_rate": 0.00019999750423902403,
-      "loss": 46.0,
-      "step": 13975
-    },
-    {
-      "epoch": 2.2507347316719675,
-      "grad_norm": 0.003913197200745344,
-      "learning_rate": 0.00019999750388158188,
-      "loss": 46.0,
-      "step": 13976
-    },
-    {
-      "epoch": 2.250895768750755,
-      "grad_norm": 0.0008797182817943394,
-      "learning_rate": 0.0001999975035241141,
-      "loss": 46.0,
-      "step": 13977
-    },
-    {
-      "epoch": 2.2510568058295424,
-      "grad_norm": 0.0010605764109641314,
-      "learning_rate": 0.00019999750316662077,
-      "loss": 46.0,
-      "step": 13978
-    },
-    {
-      "epoch": 2.25121784290833,
-      "grad_norm": 0.0032169094774872065,
-      "learning_rate": 0.0001999975028091018,
-      "loss": 46.0,
-      "step": 13979
-    },
-    {
-      "epoch": 2.251378879987117,
-      "grad_norm": 0.0009241517982445657,
-      "learning_rate": 0.00019999750245155724,
-      "loss": 46.0,
-      "step": 13980
-    },
-    {
-      "epoch": 2.2515399170659043,
-      "grad_norm": 0.0006561768823303282,
-      "learning_rate": 0.0001999975020939871,
-      "loss": 46.0,
-      "step": 13981
-    },
-    {
-      "epoch": 2.2517009541446917,
-      "grad_norm": 0.0007065307581797242,
-      "learning_rate": 0.00019999750173639136,
-      "loss": 46.0,
-      "step": 13982
-    },
-    {
-      "epoch": 2.251861991223479,
-      "grad_norm": 0.0031118488404899836,
-      "learning_rate": 0.00019999750137877003,
-      "loss": 46.0,
-      "step": 13983
-    },
-    {
-      "epoch": 2.2520230283022666,
-      "grad_norm": 0.0019888062961399555,
-      "learning_rate": 0.00019999750102112312,
-      "loss": 46.0,
-      "step": 13984
-    },
-    {
-      "epoch": 2.252184065381054,
-      "grad_norm": 0.0024370860774070024,
-      "learning_rate": 0.0001999975006634506,
-      "loss": 46.0,
-      "step": 13985
-    },
-    {
-      "epoch": 2.2523451024598415,
-      "grad_norm": 0.008042311295866966,
-      "learning_rate": 0.0001999975003057525,
-      "loss": 46.0,
-      "step": 13986
-    },
-    {
-      "epoch": 2.252506139538629,
-      "grad_norm": 0.006295335944741964,
-      "learning_rate": 0.0001999974999480288,
-      "loss": 46.0,
-      "step": 13987
-    },
-    {
-      "epoch": 2.252667176617416,
-      "grad_norm": 0.0051359799690544605,
-      "learning_rate": 0.00019999749959027948,
-      "loss": 46.0,
-      "step": 13988
-    },
-    {
-      "epoch": 2.2528282136962035,
-      "grad_norm": 0.007540623191744089,
-      "learning_rate": 0.00019999749923250458,
-      "loss": 46.0,
-      "step": 13989
-    },
-    {
-      "epoch": 2.252989250774991,
-      "grad_norm": 0.004492628388106823,
-      "learning_rate": 0.0001999974988747041,
-      "loss": 46.0,
-      "step": 13990
-    },
-    {
-      "epoch": 2.2531502878537784,
-      "grad_norm": 0.0016589396400377154,
-      "learning_rate": 0.00019999749851687802,
-      "loss": 46.0,
-      "step": 13991
-    },
-    {
-      "epoch": 2.253311324932566,
-      "grad_norm": 0.004892733413726091,
-      "learning_rate": 0.00019999749815902633,
-      "loss": 46.0,
-      "step": 13992
-    },
-    {
-      "epoch": 2.2534723620113533,
-      "grad_norm": 0.005860128439962864,
-      "learning_rate": 0.00019999749780114906,
-      "loss": 46.0,
-      "step": 13993
-    },
-    {
-      "epoch": 2.2536333990901403,
-      "grad_norm": 0.003079865127801895,
-      "learning_rate": 0.0001999974974432462,
-      "loss": 46.0,
-      "step": 13994
-    },
-    {
-      "epoch": 2.2537944361689277,
-      "grad_norm": 0.007694889325648546,
-      "learning_rate": 0.00019999749708531775,
-      "loss": 46.0,
-      "step": 13995
-    },
-    {
-      "epoch": 2.253955473247715,
-      "grad_norm": 0.005706564988940954,
-      "learning_rate": 0.00019999749672736368,
-      "loss": 46.0,
-      "step": 13996
-    },
-    {
-      "epoch": 2.2541165103265026,
-      "grad_norm": 0.0013106977567076683,
-      "learning_rate": 0.00019999749636938403,
-      "loss": 46.0,
-      "step": 13997
-    },
-    {
-      "epoch": 2.25427754740529,
-      "grad_norm": 0.005500475410372019,
-      "learning_rate": 0.0001999974960113788,
-      "loss": 46.0,
-      "step": 13998
-    },
-    {
-      "epoch": 2.2544385844840775,
-      "grad_norm": 0.0030034456867724657,
-      "learning_rate": 0.00019999749565334794,
-      "loss": 46.0,
-      "step": 13999
-    },
-    {
-      "epoch": 2.254599621562865,
-      "grad_norm": 0.0043139080516994,
-      "learning_rate": 0.00019999749529529153,
-      "loss": 46.0,
-      "step": 14000
-    },
-    {
-      "epoch": 2.2547606586416524,
-      "grad_norm": 0.004762556403875351,
-      "learning_rate": 0.0001999974949372095,
-      "loss": 46.0,
-      "step": 14001
-    },
-    {
-      "epoch": 2.25492169572044,
-      "grad_norm": 0.002331068739295006,
-      "learning_rate": 0.0001999974945791019,
-      "loss": 46.0,
-      "step": 14002
-    },
-    {
-      "epoch": 2.255082732799227,
-      "grad_norm": 0.0008562738075852394,
-      "learning_rate": 0.00019999749422096866,
-      "loss": 46.0,
-      "step": 14003
-    },
-    {
-      "epoch": 2.2552437698780143,
-      "grad_norm": 0.003933022264391184,
-      "learning_rate": 0.00019999749386280985,
-      "loss": 46.0,
-      "step": 14004
-    },
-    {
-      "epoch": 2.255404806956802,
-      "grad_norm": 0.0024339770898222923,
-      "learning_rate": 0.00019999749350462547,
-      "loss": 46.0,
-      "step": 14005
-    },
-    {
-      "epoch": 2.2555658440355892,
-      "grad_norm": 0.003784996923059225,
-      "learning_rate": 0.00019999749314641548,
-      "loss": 46.0,
-      "step": 14006
-    },
-    {
-      "epoch": 2.2557268811143767,
-      "grad_norm": 0.001622296986170113,
-      "learning_rate": 0.00019999749278817985,
-      "loss": 46.0,
-      "step": 14007
-    },
-    {
-      "epoch": 2.255887918193164,
-      "grad_norm": 0.010320202447474003,
-      "learning_rate": 0.00019999749242991866,
-      "loss": 46.0,
-      "step": 14008
-    },
-    {
-      "epoch": 2.256048955271951,
-      "grad_norm": 0.0008709838148206472,
-      "learning_rate": 0.00019999749207163188,
-      "loss": 46.0,
-      "step": 14009
-    },
-    {
-      "epoch": 2.2562099923507386,
-      "grad_norm": 0.0019778672140091658,
-      "learning_rate": 0.00019999749171331952,
-      "loss": 46.0,
-      "step": 14010
-    },
-    {
-      "epoch": 2.256371029429526,
-      "grad_norm": 0.004650791175663471,
-      "learning_rate": 0.00019999749135498157,
-      "loss": 46.0,
-      "step": 14011
-    },
-    {
-      "epoch": 2.2565320665083135,
-      "grad_norm": 0.0013059972552582622,
-      "learning_rate": 0.000199997490996618,
-      "loss": 46.0,
-      "step": 14012
-    },
-    {
-      "epoch": 2.256693103587101,
-      "grad_norm": 0.0027204507496207952,
-      "learning_rate": 0.00019999749063822885,
-      "loss": 46.0,
-      "step": 14013
-    },
-    {
-      "epoch": 2.2568541406658884,
-      "grad_norm": 0.004203621298074722,
-      "learning_rate": 0.0001999974902798141,
-      "loss": 46.0,
-      "step": 14014
-    },
-    {
-      "epoch": 2.257015177744676,
-      "grad_norm": 0.005804077722132206,
-      "learning_rate": 0.00019999748992137375,
-      "loss": 46.0,
-      "step": 14015
-    },
-    {
-      "epoch": 2.2571762148234633,
-      "grad_norm": 0.0010115739423781633,
-      "learning_rate": 0.0001999974895629078,
-      "loss": 46.0,
-      "step": 14016
-    },
-    {
-      "epoch": 2.2573372519022503,
-      "grad_norm": 0.0033738361671566963,
-      "learning_rate": 0.00019999748920441628,
-      "loss": 46.0,
-      "step": 14017
-    },
-    {
-      "epoch": 2.2574982889810378,
-      "grad_norm": 0.0015907844062894583,
-      "learning_rate": 0.00019999748884589916,
-      "loss": 46.0,
-      "step": 14018
-    },
-    {
-      "epoch": 2.2576593260598252,
-      "grad_norm": 0.0015422507422044873,
-      "learning_rate": 0.00019999748848735643,
-      "loss": 46.0,
-      "step": 14019
-    },
-    {
-      "epoch": 2.2578203631386127,
-      "grad_norm": 0.0022961099166423082,
-      "learning_rate": 0.0001999974881287881,
-      "loss": 46.0,
-      "step": 14020
-    },
-    {
-      "epoch": 2.2579814002174,
-      "grad_norm": 0.0034369323402643204,
-      "learning_rate": 0.0001999974877701942,
-      "loss": 46.0,
-      "step": 14021
-    },
-    {
-      "epoch": 2.2581424372961876,
-      "grad_norm": 0.0065235658548772335,
-      "learning_rate": 0.0001999974874115747,
-      "loss": 46.0,
-      "step": 14022
-    },
-    {
-      "epoch": 2.258303474374975,
-      "grad_norm": 0.0041592237539589405,
-      "learning_rate": 0.0001999974870529296,
-      "loss": 46.0,
-      "step": 14023
-    },
-    {
-      "epoch": 2.258464511453762,
-      "grad_norm": 0.002702559344470501,
-      "learning_rate": 0.0001999974866942589,
-      "loss": 46.0,
-      "step": 14024
-    },
-    {
-      "epoch": 2.2586255485325495,
-      "grad_norm": 0.003903142875060439,
-      "learning_rate": 0.0001999974863355626,
-      "loss": 46.0,
-      "step": 14025
-    },
-    {
-      "epoch": 2.258786585611337,
-      "grad_norm": 0.0040299296379089355,
-      "learning_rate": 0.00019999748597684074,
-      "loss": 46.0,
-      "step": 14026
-    },
-    {
-      "epoch": 2.2589476226901244,
-      "grad_norm": 0.0017029212322086096,
-      "learning_rate": 0.00019999748561809325,
-      "loss": 46.0,
-      "step": 14027
-    },
-    {
-      "epoch": 2.259108659768912,
-      "grad_norm": 0.00410170154646039,
-      "learning_rate": 0.00019999748525932018,
-      "loss": 46.0,
-      "step": 14028
-    },
-    {
-      "epoch": 2.2592696968476993,
-      "grad_norm": 0.013370919041335583,
-      "learning_rate": 0.00019999748490052153,
-      "loss": 46.0,
-      "step": 14029
-    },
-    {
-      "epoch": 2.2594307339264867,
-      "grad_norm": 0.0016519188648089767,
-      "learning_rate": 0.00019999748454169726,
-      "loss": 46.0,
-      "step": 14030
-    },
-    {
-      "epoch": 2.259591771005274,
-      "grad_norm": 0.005586672108620405,
-      "learning_rate": 0.0001999974841828474,
-      "loss": 46.0,
-      "step": 14031
-    },
-    {
-      "epoch": 2.259752808084061,
-      "grad_norm": 0.0018956048879772425,
-      "learning_rate": 0.00019999748382397198,
-      "loss": 46.0,
-      "step": 14032
-    },
-    {
-      "epoch": 2.2599138451628487,
-      "grad_norm": 0.0038738581351935863,
-      "learning_rate": 0.00019999748346507095,
-      "loss": 46.0,
-      "step": 14033
-    },
-    {
-      "epoch": 2.260074882241636,
-      "grad_norm": 0.001372222090139985,
-      "learning_rate": 0.00019999748310614427,
-      "loss": 46.0,
-      "step": 14034
-    },
-    {
-      "epoch": 2.2602359193204236,
-      "grad_norm": 0.00560833141207695,
-      "learning_rate": 0.0001999974827471921,
-      "loss": 46.0,
-      "step": 14035
-    },
-    {
-      "epoch": 2.260396956399211,
-      "grad_norm": 0.001876492751762271,
-      "learning_rate": 0.00019999748238821424,
-      "loss": 46.0,
-      "step": 14036
-    },
-    {
-      "epoch": 2.2605579934779985,
-      "grad_norm": 0.004113512579351664,
-      "learning_rate": 0.00019999748202921083,
-      "loss": 46.0,
-      "step": 14037
-    },
-    {
-      "epoch": 2.2607190305567855,
-      "grad_norm": 0.002034259494394064,
-      "learning_rate": 0.0001999974816701818,
-      "loss": 46.0,
-      "step": 14038
-    },
-    {
-      "epoch": 2.260880067635573,
-      "grad_norm": 0.0014255945570766926,
-      "learning_rate": 0.0001999974813111272,
-      "loss": 46.0,
-      "step": 14039
-    },
-    {
-      "epoch": 2.2610411047143604,
-      "grad_norm": 0.008380401879549026,
-      "learning_rate": 0.00019999748095204703,
-      "loss": 46.0,
-      "step": 14040
-    },
-    {
-      "epoch": 2.261202141793148,
-      "grad_norm": 0.001174133736640215,
-      "learning_rate": 0.00019999748059294122,
-      "loss": 46.0,
-      "step": 14041
-    },
-    {
-      "epoch": 2.2613631788719353,
-      "grad_norm": 0.0016927948454394937,
-      "learning_rate": 0.00019999748023380982,
-      "loss": 46.0,
-      "step": 14042
-    },
-    {
-      "epoch": 2.2615242159507227,
-      "grad_norm": 0.008112607523798943,
-      "learning_rate": 0.00019999747987465286,
-      "loss": 46.0,
-      "step": 14043
-    },
-    {
-      "epoch": 2.26168525302951,
-      "grad_norm": 0.0027933986857533455,
-      "learning_rate": 0.00019999747951547026,
-      "loss": 46.0,
-      "step": 14044
-    },
-    {
-      "epoch": 2.2618462901082976,
-      "grad_norm": 0.005972488783299923,
-      "learning_rate": 0.00019999747915626207,
-      "loss": 46.0,
-      "step": 14045
-    },
-    {
-      "epoch": 2.2620073271870846,
-      "grad_norm": 0.0016154962358996272,
-      "learning_rate": 0.00019999747879702833,
-      "loss": 46.0,
-      "step": 14046
-    },
-    {
-      "epoch": 2.262168364265872,
-      "grad_norm": 0.000661564408801496,
-      "learning_rate": 0.00019999747843776897,
-      "loss": 46.0,
-      "step": 14047
-    },
-    {
-      "epoch": 2.2623294013446595,
-      "grad_norm": 0.0013685725862160325,
-      "learning_rate": 0.00019999747807848402,
-      "loss": 46.0,
-      "step": 14048
-    },
-    {
-      "epoch": 2.262490438423447,
-      "grad_norm": 0.0008987082983367145,
-      "learning_rate": 0.00019999747771917345,
-      "loss": 46.0,
-      "step": 14049
-    },
-    {
-      "epoch": 2.2626514755022344,
-      "grad_norm": 0.002367316512390971,
-      "learning_rate": 0.00019999747735983733,
-      "loss": 46.0,
-      "step": 14050
-    },
-    {
-      "epoch": 2.262812512581022,
-      "grad_norm": 0.0014323643408715725,
-      "learning_rate": 0.0001999974770004756,
-      "loss": 46.0,
-      "step": 14051
-    },
-    {
-      "epoch": 2.2629735496598093,
-      "grad_norm": 0.0011208319338038564,
-      "learning_rate": 0.00019999747664108824,
-      "loss": 46.0,
-      "step": 14052
-    },
-    {
-      "epoch": 2.2631345867385964,
-      "grad_norm": 0.0026527922600507736,
-      "learning_rate": 0.00019999747628167533,
-      "loss": 46.0,
-      "step": 14053
-    },
-    {
-      "epoch": 2.263295623817384,
-      "grad_norm": 0.0012756938813254237,
-      "learning_rate": 0.0001999974759222368,
-      "loss": 46.0,
-      "step": 14054
-    },
-    {
-      "epoch": 2.2634566608961713,
-      "grad_norm": 0.002167481230571866,
-      "learning_rate": 0.00019999747556277272,
-      "loss": 46.0,
-      "step": 14055
-    },
-    {
-      "epoch": 2.2636176979749587,
-      "grad_norm": 0.0022501309867948294,
-      "learning_rate": 0.000199997475203283,
-      "loss": 46.0,
-      "step": 14056
-    },
-    {
-      "epoch": 2.263778735053746,
-      "grad_norm": 0.0005933298380114138,
-      "learning_rate": 0.00019999747484376768,
-      "loss": 46.0,
-      "step": 14057
-    },
-    {
-      "epoch": 2.2639397721325336,
-      "grad_norm": 0.0018701039953157306,
-      "learning_rate": 0.0001999974744842268,
-      "loss": 46.0,
-      "step": 14058
-    },
-    {
-      "epoch": 2.264100809211321,
-      "grad_norm": 0.0044218674302101135,
-      "learning_rate": 0.0001999974741246603,
-      "loss": 46.0,
-      "step": 14059
-    },
-    {
-      "epoch": 2.2642618462901085,
-      "grad_norm": 0.009935093112289906,
-      "learning_rate": 0.0001999974737650682,
-      "loss": 46.0,
-      "step": 14060
-    },
-    {
-      "epoch": 2.2644228833688955,
-      "grad_norm": 0.004034010693430901,
-      "learning_rate": 0.00019999747340545052,
-      "loss": 46.0,
-      "step": 14061
-    },
-    {
-      "epoch": 2.264583920447683,
-      "grad_norm": 0.003645299468189478,
-      "learning_rate": 0.00019999747304580727,
-      "loss": 46.0,
-      "step": 14062
-    },
-    {
-      "epoch": 2.2647449575264704,
-      "grad_norm": 0.0011655972339212894,
-      "learning_rate": 0.0001999974726861384,
-      "loss": 46.0,
-      "step": 14063
-    },
-    {
-      "epoch": 2.264905994605258,
-      "grad_norm": 0.0012588875833898783,
-      "learning_rate": 0.00019999747232644395,
-      "loss": 46.0,
-      "step": 14064
-    },
-    {
-      "epoch": 2.2650670316840453,
-      "grad_norm": 0.002597309648990631,
-      "learning_rate": 0.00019999747196672386,
-      "loss": 46.0,
-      "step": 14065
-    },
-    {
-      "epoch": 2.265228068762833,
-      "grad_norm": 0.0021933913230895996,
-      "learning_rate": 0.0001999974716069782,
-      "loss": 46.0,
-      "step": 14066
-    },
-    {
-      "epoch": 2.26538910584162,
-      "grad_norm": 0.007438479457050562,
-      "learning_rate": 0.000199997471247207,
-      "loss": 46.0,
-      "step": 14067
-    },
-    {
-      "epoch": 2.2655501429204072,
-      "grad_norm": 0.004280315712094307,
-      "learning_rate": 0.00019999747088741014,
-      "loss": 46.0,
-      "step": 14068
-    },
-    {
-      "epoch": 2.2657111799991947,
-      "grad_norm": 0.0029569321777671576,
-      "learning_rate": 0.0001999974705275877,
-      "loss": 46.0,
-      "step": 14069
-    },
-    {
-      "epoch": 2.265872217077982,
-      "grad_norm": 0.006991604343056679,
-      "learning_rate": 0.0001999974701677397,
-      "loss": 46.0,
-      "step": 14070
-    },
-    {
-      "epoch": 2.2660332541567696,
-      "grad_norm": 0.0019413722911849618,
-      "learning_rate": 0.00019999746980786608,
-      "loss": 46.0,
-      "step": 14071
-    },
-    {
-      "epoch": 2.266194291235557,
-      "grad_norm": 0.0034805142786353827,
-      "learning_rate": 0.00019999746944796684,
-      "loss": 46.0,
-      "step": 14072
-    },
-    {
-      "epoch": 2.2663553283143445,
-      "grad_norm": 0.0010933381272479892,
-      "learning_rate": 0.00019999746908804205,
-      "loss": 46.0,
-      "step": 14073
-    },
-    {
-      "epoch": 2.266516365393132,
-      "grad_norm": 0.006724651902914047,
-      "learning_rate": 0.00019999746872809162,
-      "loss": 46.0,
-      "step": 14074
-    },
-    {
-      "epoch": 2.2666774024719194,
-      "grad_norm": 0.0058417245745658875,
-      "learning_rate": 0.00019999746836811563,
-      "loss": 46.0,
-      "step": 14075
-    },
-    {
-      "epoch": 2.2668384395507064,
-      "grad_norm": 0.001392334932461381,
-      "learning_rate": 0.00019999746800811402,
-      "loss": 46.0,
-      "step": 14076
-    },
-    {
-      "epoch": 2.266999476629494,
-      "grad_norm": 0.0026276155840605497,
-      "learning_rate": 0.00019999746764808686,
-      "loss": 46.0,
-      "step": 14077
-    },
-    {
-      "epoch": 2.2671605137082813,
-      "grad_norm": 0.0057683903723955154,
-      "learning_rate": 0.00019999746728803408,
-      "loss": 46.0,
-      "step": 14078
-    },
-    {
-      "epoch": 2.2673215507870688,
-      "grad_norm": 0.0021580192260444164,
-      "learning_rate": 0.0001999974669279557,
-      "loss": 46.0,
-      "step": 14079
-    },
-    {
-      "epoch": 2.267482587865856,
-      "grad_norm": 0.005027227569371462,
-      "learning_rate": 0.00019999746656785173,
-      "loss": 46.0,
-      "step": 14080
-    },
-    {
-      "epoch": 2.2676436249446437,
-      "grad_norm": 0.004958228208124638,
-      "learning_rate": 0.00019999746620772218,
-      "loss": 46.0,
-      "step": 14081
-    },
-    {
-      "epoch": 2.2678046620234307,
-      "grad_norm": 0.0020200826693326235,
-      "learning_rate": 0.000199997465847567,
-      "loss": 46.0,
-      "step": 14082
-    },
-    {
-      "epoch": 2.267965699102218,
-      "grad_norm": 0.004566613584756851,
-      "learning_rate": 0.00019999746548738628,
-      "loss": 46.0,
-      "step": 14083
-    },
-    {
-      "epoch": 2.2681267361810056,
-      "grad_norm": 0.003017387818545103,
-      "learning_rate": 0.00019999746512717993,
-      "loss": 46.0,
-      "step": 14084
-    },
-    {
-      "epoch": 2.268287773259793,
-      "grad_norm": 0.0005770005518570542,
-      "learning_rate": 0.00019999746476694798,
-      "loss": 46.0,
-      "step": 14085
-    },
-    {
-      "epoch": 2.2684488103385805,
-      "grad_norm": 0.009023269638419151,
-      "learning_rate": 0.00019999746440669042,
-      "loss": 46.0,
-      "step": 14086
-    },
-    {
-      "epoch": 2.268609847417368,
-      "grad_norm": 0.0013379602460190654,
-      "learning_rate": 0.0001999974640464073,
-      "loss": 46.0,
-      "step": 14087
-    },
-    {
-      "epoch": 2.2687708844961554,
-      "grad_norm": 0.001787109998986125,
-      "learning_rate": 0.00019999746368609857,
-      "loss": 46.0,
-      "step": 14088
-    },
-    {
-      "epoch": 2.268931921574943,
-      "grad_norm": 0.0006844739546068013,
-      "learning_rate": 0.00019999746332576427,
-      "loss": 46.0,
-      "step": 14089
-    },
-    {
-      "epoch": 2.26909295865373,
-      "grad_norm": 0.0012267547426745296,
-      "learning_rate": 0.00019999746296540437,
-      "loss": 46.0,
-      "step": 14090
-    },
-    {
-      "epoch": 2.2692539957325173,
-      "grad_norm": 0.0031221327371895313,
-      "learning_rate": 0.00019999746260501887,
-      "loss": 46.0,
-      "step": 14091
-    },
-    {
-      "epoch": 2.2694150328113047,
-      "grad_norm": 0.0006667615962214768,
-      "learning_rate": 0.00019999746224460773,
-      "loss": 46.0,
-      "step": 14092
-    },
-    {
-      "epoch": 2.269576069890092,
-      "grad_norm": 0.0006900188745930791,
-      "learning_rate": 0.00019999746188417106,
-      "loss": 46.0,
-      "step": 14093
-    },
-    {
-      "epoch": 2.2697371069688796,
-      "grad_norm": 0.00814956147223711,
-      "learning_rate": 0.00019999746152370878,
-      "loss": 46.0,
-      "step": 14094
-    },
-    {
-      "epoch": 2.269898144047667,
-      "grad_norm": 0.0011044967686757445,
-      "learning_rate": 0.00019999746116322088,
-      "loss": 46.0,
-      "step": 14095
-    },
-    {
-      "epoch": 2.2700591811264546,
-      "grad_norm": 0.0026550323236733675,
-      "learning_rate": 0.0001999974608027074,
-      "loss": 46.0,
-      "step": 14096
-    },
-    {
-      "epoch": 2.2702202182052416,
-      "grad_norm": 0.0033718652557581663,
-      "learning_rate": 0.00019999746044216835,
-      "loss": 46.0,
-      "step": 14097
-    },
-    {
-      "epoch": 2.270381255284029,
-      "grad_norm": 0.008355478756129742,
-      "learning_rate": 0.00019999746008160366,
-      "loss": 46.0,
-      "step": 14098
-    },
-    {
-      "epoch": 2.2705422923628165,
-      "grad_norm": 0.005124479066580534,
-      "learning_rate": 0.00019999745972101342,
-      "loss": 46.0,
-      "step": 14099
-    },
-    {
-      "epoch": 2.270703329441604,
-      "grad_norm": 0.0027128334622830153,
-      "learning_rate": 0.00019999745936039758,
-      "loss": 46.0,
-      "step": 14100
-    },
-    {
-      "epoch": 2.2708643665203914,
-      "grad_norm": 0.001338453497737646,
-      "learning_rate": 0.0001999974589997561,
-      "loss": 46.0,
-      "step": 14101
-    },
-    {
-      "epoch": 2.271025403599179,
-      "grad_norm": 0.0014710178365930915,
-      "learning_rate": 0.00019999745863908907,
-      "loss": 46.0,
-      "step": 14102
-    },
-    {
-      "epoch": 2.2711864406779663,
-      "grad_norm": 0.0020965919829905033,
-      "learning_rate": 0.00019999745827839642,
-      "loss": 46.0,
-      "step": 14103
-    },
-    {
-      "epoch": 2.2713474777567537,
-      "grad_norm": 0.00815815944224596,
-      "learning_rate": 0.0001999974579176782,
-      "loss": 46.0,
-      "step": 14104
-    },
-    {
-      "epoch": 2.2715085148355407,
-      "grad_norm": 0.001573242712765932,
-      "learning_rate": 0.0001999974575569344,
-      "loss": 46.0,
-      "step": 14105
-    },
-    {
-      "epoch": 2.271669551914328,
-      "grad_norm": 0.002166392281651497,
-      "learning_rate": 0.00019999745719616495,
-      "loss": 46.0,
-      "step": 14106
-    },
-    {
-      "epoch": 2.2718305889931156,
-      "grad_norm": 0.004180111922323704,
-      "learning_rate": 0.00019999745683536995,
-      "loss": 46.0,
-      "step": 14107
-    },
-    {
-      "epoch": 2.271991626071903,
-      "grad_norm": 0.0008303820504806936,
-      "learning_rate": 0.00019999745647454934,
-      "loss": 46.0,
-      "step": 14108
-    },
-    {
-      "epoch": 2.2721526631506905,
-      "grad_norm": 0.00083970429841429,
-      "learning_rate": 0.00019999745611370317,
-      "loss": 46.0,
-      "step": 14109
-    },
-    {
-      "epoch": 2.272313700229478,
-      "grad_norm": 0.0021275547333061695,
-      "learning_rate": 0.00019999745575283132,
-      "loss": 46.0,
-      "step": 14110
-    },
-    {
-      "epoch": 2.272474737308265,
-      "grad_norm": 0.0011850871378555894,
-      "learning_rate": 0.00019999745539193395,
-      "loss": 46.0,
-      "step": 14111
-    },
-    {
-      "epoch": 2.2726357743870524,
-      "grad_norm": 0.0021773711778223515,
-      "learning_rate": 0.00019999745503101096,
-      "loss": 46.0,
-      "step": 14112
-    },
-    {
-      "epoch": 2.27279681146584,
-      "grad_norm": 0.005824453197419643,
-      "learning_rate": 0.00019999745467006239,
-      "loss": 46.0,
-      "step": 14113
-    },
-    {
-      "epoch": 2.2729578485446273,
-      "grad_norm": 0.001174977165646851,
-      "learning_rate": 0.0001999974543090882,
-      "loss": 46.0,
-      "step": 14114
-    },
-    {
-      "epoch": 2.273118885623415,
-      "grad_norm": 0.0017567890463396907,
-      "learning_rate": 0.00019999745394808845,
-      "loss": 46.0,
-      "step": 14115
-    },
-    {
-      "epoch": 2.2732799227022022,
-      "grad_norm": 0.002168092178180814,
-      "learning_rate": 0.0001999974535870631,
-      "loss": 46.0,
-      "step": 14116
-    },
-    {
-      "epoch": 2.2734409597809897,
-      "grad_norm": 0.001511600916273892,
-      "learning_rate": 0.00019999745322601213,
-      "loss": 46.0,
-      "step": 14117
-    },
-    {
-      "epoch": 2.273601996859777,
-      "grad_norm": 0.005391054321080446,
-      "learning_rate": 0.0001999974528649356,
-      "loss": 46.0,
-      "step": 14118
-    },
-    {
-      "epoch": 2.2737630339385646,
-      "grad_norm": 0.0030814053025096655,
-      "learning_rate": 0.00019999745250383344,
-      "loss": 46.0,
-      "step": 14119
-    },
-    {
-      "epoch": 2.2739240710173516,
-      "grad_norm": 0.008579681627452374,
-      "learning_rate": 0.0001999974521427057,
-      "loss": 46.0,
-      "step": 14120
-    },
-    {
-      "epoch": 2.274085108096139,
-      "grad_norm": 0.0012616177555173635,
-      "learning_rate": 0.00019999745178155237,
-      "loss": 46.0,
-      "step": 14121
-    },
-    {
-      "epoch": 2.2742461451749265,
-      "grad_norm": 0.0008829036960378289,
-      "learning_rate": 0.00019999745142037343,
-      "loss": 46.0,
-      "step": 14122
-    },
-    {
-      "epoch": 2.274407182253714,
-      "grad_norm": 0.010570581071078777,
-      "learning_rate": 0.00019999745105916893,
-      "loss": 46.0,
-      "step": 14123
-    },
-    {
-      "epoch": 2.2745682193325014,
-      "grad_norm": 0.0018389852484688163,
-      "learning_rate": 0.0001999974506979388,
-      "loss": 46.0,
-      "step": 14124
-    },
-    {
-      "epoch": 2.274729256411289,
-      "grad_norm": 0.0007157388608902693,
-      "learning_rate": 0.00019999745033668308,
-      "loss": 46.0,
-      "step": 14125
-    },
-    {
-      "epoch": 2.274890293490076,
-      "grad_norm": 0.0015591455157846212,
-      "learning_rate": 0.0001999974499754018,
-      "loss": 46.0,
-      "step": 14126
-    },
-    {
-      "epoch": 2.2750513305688633,
-      "grad_norm": 0.0007260916172526777,
-      "learning_rate": 0.00019999744961409491,
-      "loss": 46.0,
-      "step": 14127
-    },
-    {
-      "epoch": 2.275212367647651,
-      "grad_norm": 0.0007884514052420855,
-      "learning_rate": 0.0001999974492527624,
-      "loss": 46.0,
-      "step": 14128
-    },
-    {
-      "epoch": 2.2753734047264382,
-      "grad_norm": 0.0014991557691246271,
-      "learning_rate": 0.00019999744889140432,
-      "loss": 46.0,
-      "step": 14129
-    },
-    {
-      "epoch": 2.2755344418052257,
-      "grad_norm": 0.0013217049418017268,
-      "learning_rate": 0.00019999744853002062,
-      "loss": 46.0,
-      "step": 14130
-    },
-    {
-      "epoch": 2.275695478884013,
-      "grad_norm": 0.0012469340581446886,
-      "learning_rate": 0.00019999744816861137,
-      "loss": 46.0,
-      "step": 14131
-    },
-    {
-      "epoch": 2.2758565159628006,
-      "grad_norm": 0.00056753761600703,
-      "learning_rate": 0.00019999744780717648,
-      "loss": 46.0,
-      "step": 14132
-    },
-    {
-      "epoch": 2.276017553041588,
-      "grad_norm": 0.0014820970827713609,
-      "learning_rate": 0.00019999744744571602,
-      "loss": 46.0,
-      "step": 14133
-    },
-    {
-      "epoch": 2.276178590120375,
-      "grad_norm": 0.0015191090060397983,
-      "learning_rate": 0.00019999744708422998,
-      "loss": 46.0,
-      "step": 14134
-    },
-    {
-      "epoch": 2.2763396271991625,
-      "grad_norm": 0.004156601149588823,
-      "learning_rate": 0.00019999744672271832,
-      "loss": 46.0,
-      "step": 14135
-    },
-    {
-      "epoch": 2.27650066427795,
-      "grad_norm": 0.0034528549294918776,
-      "learning_rate": 0.00019999744636118108,
-      "loss": 46.0,
-      "step": 14136
-    },
-    {
-      "epoch": 2.2766617013567374,
-      "grad_norm": 0.004388425964862108,
-      "learning_rate": 0.00019999744599961825,
-      "loss": 46.0,
-      "step": 14137
-    },
-    {
-      "epoch": 2.276822738435525,
-      "grad_norm": 0.008133472874760628,
-      "learning_rate": 0.0001999974456380298,
-      "loss": 46.0,
-      "step": 14138
-    },
-    {
-      "epoch": 2.2769837755143123,
-      "grad_norm": 0.010087854228913784,
-      "learning_rate": 0.00019999744527641577,
-      "loss": 46.0,
-      "step": 14139
-    },
-    {
-      "epoch": 2.2771448125930998,
-      "grad_norm": 0.003923302982002497,
-      "learning_rate": 0.00019999744491477615,
-      "loss": 46.0,
-      "step": 14140
-    },
-    {
-      "epoch": 2.2773058496718868,
-      "grad_norm": 0.0033097814302891493,
-      "learning_rate": 0.00019999744455311095,
-      "loss": 46.0,
-      "step": 14141
-    },
-    {
-      "epoch": 2.277466886750674,
-      "grad_norm": 0.006234543398022652,
-      "learning_rate": 0.00019999744419142013,
-      "loss": 46.0,
-      "step": 14142
-    },
-    {
-      "epoch": 2.2776279238294617,
-      "grad_norm": 0.0005176379927434027,
-      "learning_rate": 0.00019999744382970375,
-      "loss": 46.0,
-      "step": 14143
-    },
-    {
-      "epoch": 2.277788960908249,
-      "grad_norm": 0.0010656907688826323,
-      "learning_rate": 0.00019999744346796172,
-      "loss": 46.0,
-      "step": 14144
-    },
-    {
-      "epoch": 2.2779499979870366,
-      "grad_norm": 0.00569300027564168,
-      "learning_rate": 0.00019999744310619411,
-      "loss": 46.0,
-      "step": 14145
-    },
-    {
-      "epoch": 2.278111035065824,
-      "grad_norm": 0.0027295351028442383,
-      "learning_rate": 0.00019999744274440095,
-      "loss": 46.0,
-      "step": 14146
-    },
-    {
-      "epoch": 2.2782720721446115,
-      "grad_norm": 0.0013617512304335833,
-      "learning_rate": 0.00019999744238258216,
-      "loss": 46.0,
-      "step": 14147
-    },
-    {
-      "epoch": 2.278433109223399,
-      "grad_norm": 0.0008217511349357665,
-      "learning_rate": 0.0001999974420207378,
-      "loss": 46.0,
-      "step": 14148
-    },
-    {
-      "epoch": 2.278594146302186,
-      "grad_norm": 0.005439859349280596,
-      "learning_rate": 0.00019999744165886783,
-      "loss": 46.0,
-      "step": 14149
-    },
-    {
-      "epoch": 2.2787551833809734,
-      "grad_norm": 0.0035284413024783134,
-      "learning_rate": 0.00019999744129697223,
-      "loss": 46.0,
-      "step": 14150
-    },
-    {
-      "epoch": 2.278916220459761,
-      "grad_norm": 0.0019542109221220016,
-      "learning_rate": 0.00019999744093505107,
-      "loss": 46.0,
-      "step": 14151
-    },
-    {
-      "epoch": 2.2790772575385483,
-      "grad_norm": 0.0008412288152612746,
-      "learning_rate": 0.00019999744057310436,
-      "loss": 46.0,
-      "step": 14152
-    },
-    {
-      "epoch": 2.2792382946173357,
-      "grad_norm": 0.0017408864805474877,
-      "learning_rate": 0.00019999744021113197,
-      "loss": 46.0,
-      "step": 14153
-    },
-    {
-      "epoch": 2.279399331696123,
-      "grad_norm": 0.007856925949454308,
-      "learning_rate": 0.00019999743984913405,
-      "loss": 46.0,
-      "step": 14154
-    },
-    {
-      "epoch": 2.27956036877491,
-      "grad_norm": 0.002513745566830039,
-      "learning_rate": 0.0001999974394871105,
-      "loss": 46.0,
-      "step": 14155
-    },
-    {
-      "epoch": 2.2797214058536976,
-      "grad_norm": 0.0028763373848050833,
-      "learning_rate": 0.0001999974391250614,
-      "loss": 46.0,
-      "step": 14156
-    },
-    {
-      "epoch": 2.279882442932485,
-      "grad_norm": 0.0010043465299531817,
-      "learning_rate": 0.00019999743876298665,
-      "loss": 46.0,
-      "step": 14157
-    },
-    {
-      "epoch": 2.2800434800112725,
-      "grad_norm": 0.0013878229074180126,
-      "learning_rate": 0.00019999743840088633,
-      "loss": 46.0,
-      "step": 14158
-    },
-    {
-      "epoch": 2.28020451709006,
-      "grad_norm": 0.0012308438308537006,
-      "learning_rate": 0.00019999743803876042,
-      "loss": 46.0,
-      "step": 14159
-    },
-    {
-      "epoch": 2.2803655541688475,
-      "grad_norm": 0.0009700900409370661,
-      "learning_rate": 0.00019999743767660892,
-      "loss": 46.0,
-      "step": 14160
-    },
-    {
-      "epoch": 2.280526591247635,
-      "grad_norm": 0.007130511105060577,
-      "learning_rate": 0.0001999974373144318,
-      "loss": 46.0,
-      "step": 14161
-    },
-    {
-      "epoch": 2.2806876283264224,
-      "grad_norm": 0.0023319628089666367,
-      "learning_rate": 0.00019999743695222914,
-      "loss": 46.0,
-      "step": 14162
-    },
-    {
-      "epoch": 2.2808486654052094,
-      "grad_norm": 0.0016403687186539173,
-      "learning_rate": 0.00019999743659000085,
-      "loss": 46.0,
-      "step": 14163
-    },
-    {
-      "epoch": 2.281009702483997,
-      "grad_norm": 0.001171128824353218,
-      "learning_rate": 0.00019999743622774695,
-      "loss": 46.0,
-      "step": 14164
-    },
-    {
-      "epoch": 2.2811707395627843,
-      "grad_norm": 0.002000638982281089,
-      "learning_rate": 0.0001999974358654675,
-      "loss": 46.0,
-      "step": 14165
-    },
-    {
-      "epoch": 2.2813317766415717,
-      "grad_norm": 0.0017989149782806635,
-      "learning_rate": 0.0001999974355031624,
-      "loss": 46.0,
-      "step": 14166
-    },
-    {
-      "epoch": 2.281492813720359,
-      "grad_norm": 0.003598693758249283,
-      "learning_rate": 0.00019999743514083175,
-      "loss": 46.0,
-      "step": 14167
-    },
-    {
-      "epoch": 2.2816538507991466,
-      "grad_norm": 0.004784971009939909,
-      "learning_rate": 0.00019999743477847547,
-      "loss": 46.0,
-      "step": 14168
-    },
-    {
-      "epoch": 2.281814887877934,
-      "grad_norm": 0.0012154612923040986,
-      "learning_rate": 0.0001999974344160936,
-      "loss": 46.0,
-      "step": 14169
-    },
-    {
-      "epoch": 2.281975924956721,
-      "grad_norm": 0.005106545519083738,
-      "learning_rate": 0.00019999743405368619,
-      "loss": 46.0,
-      "step": 14170
-    },
-    {
-      "epoch": 2.2821369620355085,
-      "grad_norm": 0.0068846410140395164,
-      "learning_rate": 0.00019999743369125312,
-      "loss": 46.0,
-      "step": 14171
-    },
-    {
-      "epoch": 2.282297999114296,
-      "grad_norm": 0.001367128104902804,
-      "learning_rate": 0.0001999974333287945,
-      "loss": 46.0,
-      "step": 14172
-    },
-    {
-      "epoch": 2.2824590361930834,
-      "grad_norm": 0.0030000554397702217,
-      "learning_rate": 0.00019999743296631025,
-      "loss": 46.0,
-      "step": 14173
-    },
-    {
-      "epoch": 2.282620073271871,
-      "grad_norm": 0.0023538365494459867,
-      "learning_rate": 0.00019999743260380043,
-      "loss": 46.0,
-      "step": 14174
-    },
-    {
-      "epoch": 2.2827811103506583,
-      "grad_norm": 0.0004020679334644228,
-      "learning_rate": 0.00019999743224126504,
-      "loss": 46.0,
-      "step": 14175
-    },
-    {
-      "epoch": 2.282942147429446,
-      "grad_norm": 0.0010587330907583237,
-      "learning_rate": 0.000199997431878704,
-      "loss": 46.0,
-      "step": 14176
-    },
-    {
-      "epoch": 2.2831031845082332,
-      "grad_norm": 0.0009267361019738019,
-      "learning_rate": 0.0001999974315161174,
-      "loss": 46.0,
-      "step": 14177
-    },
-    {
-      "epoch": 2.2832642215870202,
-      "grad_norm": 0.002004109090194106,
-      "learning_rate": 0.0001999974311535052,
-      "loss": 46.0,
-      "step": 14178
-    },
-    {
-      "epoch": 2.2834252586658077,
-      "grad_norm": 0.007367661688476801,
-      "learning_rate": 0.0001999974307908674,
-      "loss": 46.0,
-      "step": 14179
-    },
-    {
-      "epoch": 2.283586295744595,
-      "grad_norm": 0.0013515958562493324,
-      "learning_rate": 0.000199997430428204,
-      "loss": 46.0,
-      "step": 14180
-    },
-    {
-      "epoch": 2.2837473328233826,
-      "grad_norm": 0.0038866482209414244,
-      "learning_rate": 0.000199997430065515,
-      "loss": 46.0,
-      "step": 14181
-    },
-    {
-      "epoch": 2.28390836990217,
-      "grad_norm": 0.002496507717296481,
-      "learning_rate": 0.00019999742970280043,
-      "loss": 46.0,
-      "step": 14182
-    },
-    {
-      "epoch": 2.2840694069809575,
-      "grad_norm": 0.006878688465803862,
-      "learning_rate": 0.00019999742934006027,
-      "loss": 46.0,
-      "step": 14183
-    },
-    {
-      "epoch": 2.2842304440597445,
-      "grad_norm": 0.003757967846468091,
-      "learning_rate": 0.0001999974289772945,
-      "loss": 46.0,
-      "step": 14184
-    },
-    {
-      "epoch": 2.284391481138532,
-      "grad_norm": 0.001205588225275278,
-      "learning_rate": 0.00019999742861450315,
-      "loss": 46.0,
-      "step": 14185
-    },
-    {
-      "epoch": 2.2845525182173194,
-      "grad_norm": 0.0037523917853832245,
-      "learning_rate": 0.0001999974282516862,
-      "loss": 46.0,
-      "step": 14186
-    },
-    {
-      "epoch": 2.284713555296107,
-      "grad_norm": 0.008161641657352448,
-      "learning_rate": 0.00019999742788884362,
-      "loss": 46.0,
-      "step": 14187
-    },
-    {
-      "epoch": 2.2848745923748943,
-      "grad_norm": 0.003397412830963731,
-      "learning_rate": 0.0001999974275259755,
-      "loss": 46.0,
-      "step": 14188
-    },
-    {
-      "epoch": 2.2850356294536818,
-      "grad_norm": 0.006318398751318455,
-      "learning_rate": 0.00019999742716308175,
-      "loss": 46.0,
-      "step": 14189
-    },
-    {
-      "epoch": 2.2851966665324692,
-      "grad_norm": 0.0005041666445322335,
-      "learning_rate": 0.00019999742680016242,
-      "loss": 46.0,
-      "step": 14190
-    },
-    {
-      "epoch": 2.2853577036112567,
-      "grad_norm": 0.0012668590061366558,
-      "learning_rate": 0.00019999742643721748,
-      "loss": 46.0,
-      "step": 14191
-    },
-    {
-      "epoch": 2.285518740690044,
-      "grad_norm": 0.0006019673892296851,
-      "learning_rate": 0.00019999742607424697,
-      "loss": 46.0,
-      "step": 14192
-    },
-    {
-      "epoch": 2.285679777768831,
-      "grad_norm": 0.006719973403960466,
-      "learning_rate": 0.00019999742571125085,
-      "loss": 46.0,
-      "step": 14193
-    },
-    {
-      "epoch": 2.2858408148476186,
-      "grad_norm": 0.00344136287458241,
-      "learning_rate": 0.00019999742534822915,
-      "loss": 46.0,
-      "step": 14194
-    },
-    {
-      "epoch": 2.286001851926406,
-      "grad_norm": 0.0031660243403166533,
-      "learning_rate": 0.00019999742498518183,
-      "loss": 46.0,
-      "step": 14195
-    },
-    {
-      "epoch": 2.2861628890051935,
-      "grad_norm": 0.005834201816469431,
-      "learning_rate": 0.00019999742462210892,
-      "loss": 46.0,
-      "step": 14196
-    },
-    {
-      "epoch": 2.286323926083981,
-      "grad_norm": 0.0006354416836984456,
-      "learning_rate": 0.00019999742425901045,
-      "loss": 46.0,
-      "step": 14197
-    },
-    {
-      "epoch": 2.2864849631627684,
-      "grad_norm": 0.003480158746242523,
-      "learning_rate": 0.00019999742389588637,
-      "loss": 46.0,
-      "step": 14198
-    },
-    {
-      "epoch": 2.2866460002415554,
-      "grad_norm": 0.003493017051368952,
-      "learning_rate": 0.0001999974235327367,
-      "loss": 46.0,
-      "step": 14199
-    },
-    {
-      "epoch": 2.286807037320343,
-      "grad_norm": 0.0016648083692416549,
-      "learning_rate": 0.00019999742316956142,
-      "loss": 46.0,
-      "step": 14200
-    },
-    {
-      "epoch": 2.2869680743991303,
-      "grad_norm": 0.0014448920264840126,
-      "learning_rate": 0.00019999742280636052,
-      "loss": 46.0,
-      "step": 14201
-    },
-    {
-      "epoch": 2.2871291114779178,
-      "grad_norm": 0.0028257097583264112,
-      "learning_rate": 0.00019999742244313407,
-      "loss": 46.0,
-      "step": 14202
-    },
-    {
-      "epoch": 2.287290148556705,
-      "grad_norm": 0.0020331856794655323,
-      "learning_rate": 0.00019999742207988202,
-      "loss": 46.0,
-      "step": 14203
-    },
-    {
-      "epoch": 2.2874511856354927,
-      "grad_norm": 0.003894432447850704,
-      "learning_rate": 0.0001999974217166044,
-      "loss": 46.0,
-      "step": 14204
-    },
-    {
-      "epoch": 2.28761222271428,
-      "grad_norm": 0.004158115014433861,
-      "learning_rate": 0.00019999742135330112,
-      "loss": 46.0,
-      "step": 14205
-    },
-    {
-      "epoch": 2.2877732597930676,
-      "grad_norm": 0.005285014398396015,
-      "learning_rate": 0.00019999742098997228,
-      "loss": 46.0,
-      "step": 14206
-    },
-    {
-      "epoch": 2.2879342968718546,
-      "grad_norm": 0.002789412159472704,
-      "learning_rate": 0.00019999742062661784,
-      "loss": 46.0,
-      "step": 14207
-    },
-    {
-      "epoch": 2.288095333950642,
-      "grad_norm": 0.0015902383020147681,
-      "learning_rate": 0.00019999742026323783,
-      "loss": 46.0,
-      "step": 14208
-    },
-    {
-      "epoch": 2.2882563710294295,
-      "grad_norm": 0.0014236696297302842,
-      "learning_rate": 0.00019999741989983218,
-      "loss": 46.0,
-      "step": 14209
-    },
-    {
-      "epoch": 2.288417408108217,
-      "grad_norm": 0.003085595788434148,
-      "learning_rate": 0.000199997419536401,
-      "loss": 46.0,
-      "step": 14210
-    },
-    {
-      "epoch": 2.2885784451870044,
-      "grad_norm": 0.0011553569929674268,
-      "learning_rate": 0.00019999741917294417,
-      "loss": 46.0,
-      "step": 14211
-    },
-    {
-      "epoch": 2.288739482265792,
-      "grad_norm": 0.002870664931833744,
-      "learning_rate": 0.00019999741880946176,
-      "loss": 46.0,
-      "step": 14212
-    },
-    {
-      "epoch": 2.2889005193445793,
-      "grad_norm": 0.0032424384262412786,
-      "learning_rate": 0.00019999741844595377,
-      "loss": 46.0,
-      "step": 14213
-    },
-    {
-      "epoch": 2.2890615564233663,
-      "grad_norm": 0.0015928858192637563,
-      "learning_rate": 0.00019999741808242015,
-      "loss": 46.0,
-      "step": 14214
-    },
-    {
-      "epoch": 2.2892225935021537,
-      "grad_norm": 0.0037057893350720406,
-      "learning_rate": 0.00019999741771886098,
-      "loss": 46.0,
-      "step": 14215
-    },
-    {
-      "epoch": 2.289383630580941,
-      "grad_norm": 0.0011992177460342646,
-      "learning_rate": 0.0001999974173552762,
-      "loss": 46.0,
-      "step": 14216
-    },
-    {
-      "epoch": 2.2895446676597286,
-      "grad_norm": 0.0025299210101366043,
-      "learning_rate": 0.0001999974169916658,
-      "loss": 46.0,
-      "step": 14217
-    },
-    {
-      "epoch": 2.289705704738516,
-      "grad_norm": 0.0021669615525752306,
-      "learning_rate": 0.00019999741662802984,
-      "loss": 46.0,
-      "step": 14218
-    },
-    {
-      "epoch": 2.2898667418173035,
-      "grad_norm": 0.0013629866298288107,
-      "learning_rate": 0.0001999974162643683,
-      "loss": 46.0,
-      "step": 14219
-    },
-    {
-      "epoch": 2.290027778896091,
-      "grad_norm": 0.000873226614203304,
-      "learning_rate": 0.00019999741590068113,
-      "loss": 46.0,
-      "step": 14220
-    },
-    {
-      "epoch": 2.2901888159748784,
-      "grad_norm": 0.012677308171987534,
-      "learning_rate": 0.00019999741553696838,
-      "loss": 46.0,
-      "step": 14221
-    },
-    {
-      "epoch": 2.2903498530536655,
-      "grad_norm": 0.0020746940281242132,
-      "learning_rate": 0.00019999741517323004,
-      "loss": 46.0,
-      "step": 14222
-    },
-    {
-      "epoch": 2.290510890132453,
-      "grad_norm": 0.0023188902996480465,
-      "learning_rate": 0.0001999974148094661,
-      "loss": 46.0,
-      "step": 14223
-    },
-    {
-      "epoch": 2.2906719272112404,
-      "grad_norm": 0.0024838189128786325,
-      "learning_rate": 0.00019999741444567655,
-      "loss": 46.0,
-      "step": 14224
-    },
-    {
-      "epoch": 2.290832964290028,
-      "grad_norm": 0.004235392436385155,
-      "learning_rate": 0.0001999974140818614,
-      "loss": 46.0,
-      "step": 14225
-    },
-    {
-      "epoch": 2.2909940013688153,
-      "grad_norm": 0.0020327134989202023,
-      "learning_rate": 0.0001999974137180207,
-      "loss": 46.0,
-      "step": 14226
-    },
-    {
-      "epoch": 2.2911550384476027,
-      "grad_norm": 0.004046470392495394,
-      "learning_rate": 0.00019999741335415436,
-      "loss": 46.0,
-      "step": 14227
-    },
-    {
-      "epoch": 2.2913160755263897,
-      "grad_norm": 0.000421986827859655,
-      "learning_rate": 0.00019999741299026247,
-      "loss": 46.0,
-      "step": 14228
-    },
-    {
-      "epoch": 2.291477112605177,
-      "grad_norm": 0.004184163175523281,
-      "learning_rate": 0.00019999741262634495,
-      "loss": 46.0,
-      "step": 14229
-    },
-    {
-      "epoch": 2.2916381496839646,
-      "grad_norm": 0.0027920235879719257,
-      "learning_rate": 0.00019999741226240186,
-      "loss": 46.0,
-      "step": 14230
-    },
-    {
-      "epoch": 2.291799186762752,
-      "grad_norm": 0.004182478412985802,
-      "learning_rate": 0.00019999741189843318,
-      "loss": 46.0,
-      "step": 14231
-    },
-    {
-      "epoch": 2.2919602238415395,
-      "grad_norm": 0.0017277334118261933,
-      "learning_rate": 0.00019999741153443887,
-      "loss": 46.0,
-      "step": 14232
-    },
-    {
-      "epoch": 2.292121260920327,
-      "grad_norm": 0.0011476981453597546,
-      "learning_rate": 0.000199997411170419,
-      "loss": 46.0,
-      "step": 14233
-    },
-    {
-      "epoch": 2.2922822979991144,
-      "grad_norm": 0.002320182742550969,
-      "learning_rate": 0.0001999974108063735,
-      "loss": 46.0,
-      "step": 14234
-    },
-    {
-      "epoch": 2.292443335077902,
-      "grad_norm": 0.006419925019145012,
-      "learning_rate": 0.00019999741044230245,
-      "loss": 46.0,
-      "step": 14235
-    },
-    {
-      "epoch": 2.292604372156689,
-      "grad_norm": 0.004862943664193153,
-      "learning_rate": 0.00019999741007820578,
-      "loss": 46.0,
-      "step": 14236
-    },
-    {
-      "epoch": 2.2927654092354763,
-      "grad_norm": 0.005383777432143688,
-      "learning_rate": 0.0001999974097140835,
-      "loss": 46.0,
-      "step": 14237
-    },
-    {
-      "epoch": 2.292926446314264,
-      "grad_norm": 0.0025240981485694647,
-      "learning_rate": 0.00019999740934993564,
-      "loss": 46.0,
-      "step": 14238
-    },
-    {
-      "epoch": 2.2930874833930512,
-      "grad_norm": 0.0017017071368172765,
-      "learning_rate": 0.0001999974089857622,
-      "loss": 46.0,
-      "step": 14239
-    },
-    {
-      "epoch": 2.2932485204718387,
-      "grad_norm": 0.01041108462959528,
-      "learning_rate": 0.00019999740862156314,
-      "loss": 46.0,
-      "step": 14240
-    },
-    {
-      "epoch": 2.293409557550626,
-      "grad_norm": 0.010626653209328651,
-      "learning_rate": 0.00019999740825733852,
-      "loss": 46.0,
-      "step": 14241
-    },
-    {
-      "epoch": 2.2935705946294136,
-      "grad_norm": 0.0021872264333069324,
-      "learning_rate": 0.00019999740789308827,
-      "loss": 46.0,
-      "step": 14242
-    },
-    {
-      "epoch": 2.2937316317082006,
-      "grad_norm": 0.0034235031343996525,
-      "learning_rate": 0.00019999740752881247,
-      "loss": 46.0,
-      "step": 14243
-    },
-    {
-      "epoch": 2.293892668786988,
-      "grad_norm": 0.0019473993452265859,
-      "learning_rate": 0.00019999740716451103,
-      "loss": 46.0,
-      "step": 14244
-    },
-    {
-      "epoch": 2.2940537058657755,
-      "grad_norm": 0.0050170328468084335,
-      "learning_rate": 0.00019999740680018402,
-      "loss": 46.0,
-      "step": 14245
-    },
-    {
-      "epoch": 2.294214742944563,
-      "grad_norm": 0.0022957122419029474,
-      "learning_rate": 0.00019999740643583143,
-      "loss": 46.0,
-      "step": 14246
-    },
-    {
-      "epoch": 2.2943757800233504,
-      "grad_norm": 0.0026920344680547714,
-      "learning_rate": 0.0001999974060714532,
-      "loss": 46.0,
-      "step": 14247
-    },
-    {
-      "epoch": 2.294536817102138,
-      "grad_norm": 0.0025211991742253304,
-      "learning_rate": 0.0001999974057070494,
-      "loss": 46.0,
-      "step": 14248
-    },
-    {
-      "epoch": 2.2946978541809253,
-      "grad_norm": 0.004697122145444155,
-      "learning_rate": 0.00019999740534262003,
-      "loss": 46.0,
-      "step": 14249
-    },
-    {
-      "epoch": 2.2948588912597128,
-      "grad_norm": 0.0004291724180802703,
-      "learning_rate": 0.00019999740497816503,
-      "loss": 46.0,
-      "step": 14250
-    },
-    {
-      "epoch": 2.2950199283384998,
-      "grad_norm": 0.0021702591329813004,
-      "learning_rate": 0.00019999740461368445,
-      "loss": 46.0,
-      "step": 14251
-    },
-    {
-      "epoch": 2.2951809654172872,
-      "grad_norm": 0.004615240730345249,
-      "learning_rate": 0.0001999974042491783,
-      "loss": 46.0,
-      "step": 14252
-    },
-    {
-      "epoch": 2.2953420024960747,
-      "grad_norm": 0.011581462807953358,
-      "learning_rate": 0.00019999740388464653,
-      "loss": 46.0,
-      "step": 14253
-    },
-    {
-      "epoch": 2.295503039574862,
-      "grad_norm": 0.0035150221083313227,
-      "learning_rate": 0.00019999740352008916,
-      "loss": 46.0,
-      "step": 14254
-    },
-    {
-      "epoch": 2.2956640766536496,
-      "grad_norm": 0.005080073140561581,
-      "learning_rate": 0.0001999974031555062,
-      "loss": 46.0,
-      "step": 14255
-    },
-    {
-      "epoch": 2.295825113732437,
-      "grad_norm": 0.001563195139169693,
-      "learning_rate": 0.00019999740279089766,
-      "loss": 46.0,
-      "step": 14256
-    },
-    {
-      "epoch": 2.295986150811224,
-      "grad_norm": 0.004271410871297121,
-      "learning_rate": 0.0001999974024262635,
-      "loss": 46.0,
-      "step": 14257
-    },
-    {
-      "epoch": 2.2961471878900115,
-      "grad_norm": 0.00366918439976871,
-      "learning_rate": 0.00019999740206160378,
-      "loss": 46.0,
-      "step": 14258
-    },
-    {
-      "epoch": 2.296308224968799,
-      "grad_norm": 0.0017850485164672136,
-      "learning_rate": 0.00019999740169691842,
-      "loss": 46.0,
-      "step": 14259
-    },
-    {
-      "epoch": 2.2964692620475864,
-      "grad_norm": 0.0018911348888650537,
-      "learning_rate": 0.0001999974013322075,
-      "loss": 46.0,
-      "step": 14260
-    },
-    {
-      "epoch": 2.296630299126374,
-      "grad_norm": 0.00279340916313231,
-      "learning_rate": 0.000199997400967471,
-      "loss": 46.0,
-      "step": 14261
-    },
-    {
-      "epoch": 2.2967913362051613,
-      "grad_norm": 0.006460262928158045,
-      "learning_rate": 0.00019999740060270887,
-      "loss": 46.0,
-      "step": 14262
-    },
-    {
-      "epoch": 2.2969523732839487,
-      "grad_norm": 0.0020022308453917503,
-      "learning_rate": 0.00019999740023792116,
-      "loss": 46.0,
-      "step": 14263
-    },
-    {
-      "epoch": 2.297113410362736,
-      "grad_norm": 0.0008887032745406032,
-      "learning_rate": 0.00019999739987310787,
-      "loss": 46.0,
-      "step": 14264
-    },
-    {
-      "epoch": 2.2972744474415236,
-      "grad_norm": 0.0008106484892778099,
-      "learning_rate": 0.00019999739950826896,
-      "loss": 46.0,
-      "step": 14265
-    },
-    {
-      "epoch": 2.2974354845203107,
-      "grad_norm": 0.0006976912845857441,
-      "learning_rate": 0.00019999739914340446,
-      "loss": 46.0,
-      "step": 14266
-    },
-    {
-      "epoch": 2.297596521599098,
-      "grad_norm": 0.0012451495276764035,
-      "learning_rate": 0.00019999739877851438,
-      "loss": 46.0,
-      "step": 14267
-    },
-    {
-      "epoch": 2.2977575586778856,
-      "grad_norm": 0.002070675604045391,
-      "learning_rate": 0.0001999973984135987,
-      "loss": 46.0,
-      "step": 14268
-    },
-    {
-      "epoch": 2.297918595756673,
-      "grad_norm": 0.000902310770470649,
-      "learning_rate": 0.00019999739804865742,
-      "loss": 46.0,
-      "step": 14269
-    },
-    {
-      "epoch": 2.2980796328354605,
-      "grad_norm": 0.001746731111779809,
-      "learning_rate": 0.00019999739768369055,
-      "loss": 46.0,
-      "step": 14270
-    },
-    {
-      "epoch": 2.298240669914248,
-      "grad_norm": 0.002341133076697588,
-      "learning_rate": 0.0001999973973186981,
-      "loss": 46.0,
-      "step": 14271
-    },
-    {
-      "epoch": 2.298401706993035,
-      "grad_norm": 0.008440453559160233,
-      "learning_rate": 0.00019999739695368004,
-      "loss": 46.0,
-      "step": 14272
-    },
-    {
-      "epoch": 2.2985627440718224,
-      "grad_norm": 0.0009725481504574418,
-      "learning_rate": 0.00019999739658863638,
-      "loss": 46.0,
-      "step": 14273
-    },
-    {
-      "epoch": 2.29872378115061,
-      "grad_norm": 0.002638152800500393,
-      "learning_rate": 0.00019999739622356716,
-      "loss": 46.0,
-      "step": 14274
-    },
-    {
-      "epoch": 2.2988848182293973,
-      "grad_norm": 0.0019563420210033655,
-      "learning_rate": 0.0001999973958584723,
-      "loss": 46.0,
-      "step": 14275
-    },
-    {
-      "epoch": 2.2990458553081847,
-      "grad_norm": 0.004565945826470852,
-      "learning_rate": 0.00019999739549335187,
-      "loss": 46.0,
-      "step": 14276
-    },
-    {
-      "epoch": 2.299206892386972,
-      "grad_norm": 0.0025682856794446707,
-      "learning_rate": 0.00019999739512820583,
-      "loss": 46.0,
-      "step": 14277
-    },
-    {
-      "epoch": 2.2993679294657596,
-      "grad_norm": 0.0042879958637058735,
-      "learning_rate": 0.0001999973947630342,
-      "loss": 46.0,
-      "step": 14278
-    },
-    {
-      "epoch": 2.299528966544547,
-      "grad_norm": 0.0011807095725089312,
-      "learning_rate": 0.000199997394397837,
-      "loss": 46.0,
-      "step": 14279
-    },
-    {
-      "epoch": 2.299690003623334,
-      "grad_norm": 0.0043216426856815815,
-      "learning_rate": 0.00019999739403261417,
-      "loss": 46.0,
-      "step": 14280
-    },
-    {
-      "epoch": 2.2998510407021215,
-      "grad_norm": 0.0039243851788342,
-      "learning_rate": 0.00019999739366736576,
-      "loss": 46.0,
-      "step": 14281
-    },
-    {
-      "epoch": 2.300012077780909,
-      "grad_norm": 0.01166997104883194,
-      "learning_rate": 0.00019999739330209178,
-      "loss": 46.0,
-      "step": 14282
-    },
-    {
-      "epoch": 2.3001731148596964,
-      "grad_norm": 0.0018461483996361494,
-      "learning_rate": 0.00019999739293679217,
-      "loss": 46.0,
-      "step": 14283
-    },
-    {
-      "epoch": 2.300334151938484,
-      "grad_norm": 0.0016292390646412969,
-      "learning_rate": 0.000199997392571467,
-      "loss": 46.0,
-      "step": 14284
-    },
-    {
-      "epoch": 2.3004951890172713,
-      "grad_norm": 0.00411981251090765,
-      "learning_rate": 0.0001999973922061162,
-      "loss": 46.0,
-      "step": 14285
-    },
-    {
-      "epoch": 2.300656226096059,
-      "grad_norm": 0.00183815136551857,
-      "learning_rate": 0.00019999739184073986,
-      "loss": 46.0,
-      "step": 14286
-    },
-    {
-      "epoch": 2.300817263174846,
-      "grad_norm": 0.0005163150490261614,
-      "learning_rate": 0.00019999739147533787,
-      "loss": 46.0,
-      "step": 14287
-    },
-    {
-      "epoch": 2.3009783002536333,
-      "grad_norm": 0.0014224780024960637,
-      "learning_rate": 0.0001999973911099103,
-      "loss": 46.0,
-      "step": 14288
-    },
-    {
-      "epoch": 2.3011393373324207,
-      "grad_norm": 0.002101413905620575,
-      "learning_rate": 0.00019999739074445713,
-      "loss": 46.0,
-      "step": 14289
-    },
-    {
-      "epoch": 2.301300374411208,
-      "grad_norm": 0.004940839949995279,
-      "learning_rate": 0.00019999739037897837,
-      "loss": 46.0,
-      "step": 14290
-    },
-    {
-      "epoch": 2.3014614114899956,
-      "grad_norm": 0.008614122867584229,
-      "learning_rate": 0.00019999739001347403,
-      "loss": 46.0,
-      "step": 14291
-    },
-    {
-      "epoch": 2.301622448568783,
-      "grad_norm": 0.0016401347238570452,
-      "learning_rate": 0.00019999738964794408,
-      "loss": 46.0,
-      "step": 14292
-    },
-    {
-      "epoch": 2.3017834856475705,
-      "grad_norm": 0.0012262946693226695,
-      "learning_rate": 0.00019999738928238857,
-      "loss": 46.0,
-      "step": 14293
-    },
-    {
-      "epoch": 2.301944522726358,
-      "grad_norm": 0.0008248728699982166,
-      "learning_rate": 0.0001999973889168074,
-      "loss": 46.0,
-      "step": 14294
-    },
-    {
-      "epoch": 2.302105559805145,
-      "grad_norm": 0.009803717024624348,
-      "learning_rate": 0.00019999738855120072,
-      "loss": 46.0,
-      "step": 14295
-    },
-    {
-      "epoch": 2.3022665968839324,
-      "grad_norm": 0.0015913811512291431,
-      "learning_rate": 0.0001999973881855684,
-      "loss": 46.0,
-      "step": 14296
-    },
-    {
-      "epoch": 2.30242763396272,
-      "grad_norm": 0.0033310051076114178,
-      "learning_rate": 0.00019999738781991045,
-      "loss": 46.0,
-      "step": 14297
-    },
-    {
-      "epoch": 2.3025886710415073,
-      "grad_norm": 0.0027968536596745253,
-      "learning_rate": 0.00019999738745422695,
-      "loss": 46.0,
-      "step": 14298
-    },
-    {
-      "epoch": 2.302749708120295,
-      "grad_norm": 0.0011340482160449028,
-      "learning_rate": 0.00019999738708851786,
-      "loss": 46.0,
-      "step": 14299
-    },
-    {
-      "epoch": 2.3029107451990822,
-      "grad_norm": 0.0032437255140393972,
-      "learning_rate": 0.00019999738672278315,
-      "loss": 46.0,
-      "step": 14300
-    },
-    {
-      "epoch": 2.3030717822778692,
-      "grad_norm": 0.001587683567777276,
-      "learning_rate": 0.0001999973863570229,
-      "loss": 46.0,
-      "step": 14301
-    },
-    {
-      "epoch": 2.3032328193566567,
-      "grad_norm": 0.003964809235185385,
-      "learning_rate": 0.000199997385991237,
-      "loss": 46.0,
-      "step": 14302
-    },
-    {
-      "epoch": 2.303393856435444,
-      "grad_norm": 0.0011232803808525205,
-      "learning_rate": 0.00019999738562542552,
-      "loss": 46.0,
-      "step": 14303
-    },
-    {
-      "epoch": 2.3035548935142316,
-      "grad_norm": 0.005081545561552048,
-      "learning_rate": 0.00019999738525958844,
-      "loss": 46.0,
-      "step": 14304
-    },
-    {
-      "epoch": 2.303715930593019,
-      "grad_norm": 0.003260744037106633,
-      "learning_rate": 0.00019999738489372577,
-      "loss": 46.0,
-      "step": 14305
-    },
-    {
-      "epoch": 2.3038769676718065,
-      "grad_norm": 0.004290606826543808,
-      "learning_rate": 0.0001999973845278375,
-      "loss": 46.0,
-      "step": 14306
-    },
-    {
-      "epoch": 2.304038004750594,
-      "grad_norm": 0.002330171875655651,
-      "learning_rate": 0.00019999738416192367,
-      "loss": 46.0,
-      "step": 14307
-    },
-    {
-      "epoch": 2.3041990418293814,
-      "grad_norm": 0.002846022602170706,
-      "learning_rate": 0.00019999738379598421,
-      "loss": 46.0,
-      "step": 14308
-    },
-    {
-      "epoch": 2.304360078908169,
-      "grad_norm": 0.002025551162660122,
-      "learning_rate": 0.00019999738343001917,
-      "loss": 46.0,
-      "step": 14309
-    },
-    {
-      "epoch": 2.304521115986956,
-      "grad_norm": 0.0012415667297318578,
-      "learning_rate": 0.00019999738306402854,
-      "loss": 46.0,
-      "step": 14310
-    },
-    {
-      "epoch": 2.3046821530657433,
-      "grad_norm": 0.0018311068415641785,
-      "learning_rate": 0.0001999973826980123,
-      "loss": 46.0,
-      "step": 14311
-    },
-    {
-      "epoch": 2.3048431901445308,
-      "grad_norm": 0.002649351954460144,
-      "learning_rate": 0.00019999738233197046,
-      "loss": 46.0,
-      "step": 14312
-    },
-    {
-      "epoch": 2.305004227223318,
-      "grad_norm": 0.0074116941541433334,
-      "learning_rate": 0.00019999738196590304,
-      "loss": 46.0,
-      "step": 14313
-    },
-    {
-      "epoch": 2.3051652643021057,
-      "grad_norm": 0.0034662620164453983,
-      "learning_rate": 0.00019999738159981006,
-      "loss": 46.0,
-      "step": 14314
-    },
-    {
-      "epoch": 2.305326301380893,
-      "grad_norm": 0.0014256008435040712,
-      "learning_rate": 0.00019999738123369141,
-      "loss": 46.0,
-      "step": 14315
-    },
-    {
-      "epoch": 2.30548733845968,
-      "grad_norm": 0.005318333860486746,
-      "learning_rate": 0.00019999738086754723,
-      "loss": 46.0,
-      "step": 14316
-    },
-    {
-      "epoch": 2.3056483755384676,
-      "grad_norm": 0.0006057993159629405,
-      "learning_rate": 0.0001999973805013774,
-      "loss": 46.0,
-      "step": 14317
-    },
-    {
-      "epoch": 2.305809412617255,
-      "grad_norm": 0.003133472753688693,
-      "learning_rate": 0.00019999738013518203,
-      "loss": 46.0,
-      "step": 14318
-    },
-    {
-      "epoch": 2.3059704496960425,
-      "grad_norm": 0.001851937035098672,
-      "learning_rate": 0.00019999737976896106,
-      "loss": 46.0,
-      "step": 14319
-    },
-    {
-      "epoch": 2.30613148677483,
-      "grad_norm": 0.003944995813071728,
-      "learning_rate": 0.00019999737940271448,
-      "loss": 46.0,
-      "step": 14320
-    },
-    {
-      "epoch": 2.3062925238536174,
-      "grad_norm": 0.0003884835459757596,
-      "learning_rate": 0.0001999973790364423,
-      "loss": 46.0,
-      "step": 14321
-    },
-    {
-      "epoch": 2.306453560932405,
-      "grad_norm": 0.004032814409583807,
-      "learning_rate": 0.00019999737867014452,
-      "loss": 46.0,
-      "step": 14322
-    },
-    {
-      "epoch": 2.3066145980111923,
-      "grad_norm": 0.0020193925593048334,
-      "learning_rate": 0.00019999737830382117,
-      "loss": 46.0,
-      "step": 14323
-    },
-    {
-      "epoch": 2.3067756350899793,
-      "grad_norm": 0.002641970757395029,
-      "learning_rate": 0.0001999973779374722,
-      "loss": 46.0,
-      "step": 14324
-    },
-    {
-      "epoch": 2.3069366721687667,
-      "grad_norm": 0.0020113217178732157,
-      "learning_rate": 0.00019999737757109767,
-      "loss": 46.0,
-      "step": 14325
-    },
-    {
-      "epoch": 2.307097709247554,
-      "grad_norm": 0.002422425663098693,
-      "learning_rate": 0.0001999973772046975,
-      "loss": 46.0,
-      "step": 14326
-    },
-    {
-      "epoch": 2.3072587463263416,
-      "grad_norm": 0.0014323890209197998,
-      "learning_rate": 0.00019999737683827179,
-      "loss": 46.0,
-      "step": 14327
-    },
-    {
-      "epoch": 2.307419783405129,
-      "grad_norm": 0.0009370298939757049,
-      "learning_rate": 0.00019999737647182045,
-      "loss": 46.0,
-      "step": 14328
-    },
-    {
-      "epoch": 2.3075808204839166,
-      "grad_norm": 0.006395894102752209,
-      "learning_rate": 0.00019999737610534353,
-      "loss": 46.0,
-      "step": 14329
-    },
-    {
-      "epoch": 2.307741857562704,
-      "grad_norm": 0.0007703409646637738,
-      "learning_rate": 0.00019999737573884102,
-      "loss": 46.0,
-      "step": 14330
-    },
-    {
-      "epoch": 2.307902894641491,
-      "grad_norm": 0.0033213100396096706,
-      "learning_rate": 0.00019999737537231287,
-      "loss": 46.0,
-      "step": 14331
-    },
-    {
-      "epoch": 2.3080639317202785,
-      "grad_norm": 0.0036874287761747837,
-      "learning_rate": 0.00019999737500575916,
-      "loss": 46.0,
-      "step": 14332
-    },
-    {
-      "epoch": 2.308224968799066,
-      "grad_norm": 0.0031320613343268633,
-      "learning_rate": 0.00019999737463917986,
-      "loss": 46.0,
-      "step": 14333
-    },
-    {
-      "epoch": 2.3083860058778534,
-      "grad_norm": 0.003985895775258541,
-      "learning_rate": 0.00019999737427257497,
-      "loss": 46.0,
-      "step": 14334
-    },
-    {
-      "epoch": 2.308547042956641,
-      "grad_norm": 0.003459648694843054,
-      "learning_rate": 0.00019999737390594447,
-      "loss": 46.0,
-      "step": 14335
-    },
-    {
-      "epoch": 2.3087080800354283,
-      "grad_norm": 0.0037921795155853033,
-      "learning_rate": 0.00019999737353928838,
-      "loss": 46.0,
-      "step": 14336
-    },
-    {
-      "epoch": 2.3088691171142157,
-      "grad_norm": 0.0005934066139161587,
-      "learning_rate": 0.0001999973731726067,
-      "loss": 46.0,
-      "step": 14337
-    },
-    {
-      "epoch": 2.309030154193003,
-      "grad_norm": 0.0029144128784537315,
-      "learning_rate": 0.00019999737280589942,
-      "loss": 46.0,
-      "step": 14338
-    },
-    {
-      "epoch": 2.30919119127179,
-      "grad_norm": 0.003398422384634614,
-      "learning_rate": 0.00019999737243916657,
-      "loss": 46.0,
-      "step": 14339
-    },
-    {
-      "epoch": 2.3093522283505776,
-      "grad_norm": 0.0014305945951491594,
-      "learning_rate": 0.00019999737207240808,
-      "loss": 46.0,
-      "step": 14340
-    },
-    {
-      "epoch": 2.309513265429365,
-      "grad_norm": 0.0023654219694435596,
-      "learning_rate": 0.00019999737170562406,
-      "loss": 46.0,
-      "step": 14341
-    },
-    {
-      "epoch": 2.3096743025081525,
-      "grad_norm": 0.0015124857891350985,
-      "learning_rate": 0.0001999973713388144,
-      "loss": 46.0,
-      "step": 14342
-    },
-    {
-      "epoch": 2.30983533958694,
-      "grad_norm": 0.0005553358932957053,
-      "learning_rate": 0.00019999737097197915,
-      "loss": 46.0,
-      "step": 14343
-    },
-    {
-      "epoch": 2.3099963766657274,
-      "grad_norm": 0.004616947844624519,
-      "learning_rate": 0.0001999973706051183,
-      "loss": 46.0,
-      "step": 14344
-    },
-    {
-      "epoch": 2.3101574137445144,
-      "grad_norm": 0.0018007889157161117,
-      "learning_rate": 0.00019999737023823188,
-      "loss": 46.0,
-      "step": 14345
-    },
-    {
-      "epoch": 2.310318450823302,
-      "grad_norm": 0.0023752308916300535,
-      "learning_rate": 0.00019999736987131984,
-      "loss": 46.0,
-      "step": 14346
-    },
-    {
-      "epoch": 2.3104794879020893,
-      "grad_norm": 0.0038026210386306047,
-      "learning_rate": 0.0001999973695043822,
-      "loss": 46.0,
-      "step": 14347
-    },
-    {
-      "epoch": 2.310640524980877,
-      "grad_norm": 0.004556956235319376,
-      "learning_rate": 0.000199997369137419,
-      "loss": 46.0,
-      "step": 14348
-    },
-    {
-      "epoch": 2.3108015620596642,
-      "grad_norm": 0.010329912416636944,
-      "learning_rate": 0.00019999736877043017,
-      "loss": 46.0,
-      "step": 14349
-    },
-    {
-      "epoch": 2.3109625991384517,
-      "grad_norm": 0.0013289510970935225,
-      "learning_rate": 0.00019999736840341578,
-      "loss": 46.0,
-      "step": 14350
-    },
-    {
-      "epoch": 2.311123636217239,
-      "grad_norm": 0.0009634209563955665,
-      "learning_rate": 0.00019999736803637578,
-      "loss": 46.0,
-      "step": 14351
-    },
-    {
-      "epoch": 2.3112846732960266,
-      "grad_norm": 0.004501132760196924,
-      "learning_rate": 0.0001999973676693102,
-      "loss": 46.0,
-      "step": 14352
-    },
-    {
-      "epoch": 2.3114457103748136,
-      "grad_norm": 0.006060115061700344,
-      "learning_rate": 0.000199997367302219,
-      "loss": 46.0,
-      "step": 14353
-    },
-    {
-      "epoch": 2.311606747453601,
-      "grad_norm": 0.00291638495400548,
-      "learning_rate": 0.00019999736693510222,
-      "loss": 46.0,
-      "step": 14354
-    },
-    {
-      "epoch": 2.3117677845323885,
-      "grad_norm": 0.0017816550098359585,
-      "learning_rate": 0.00019999736656795984,
-      "loss": 46.0,
-      "step": 14355
-    },
-    {
-      "epoch": 2.311928821611176,
-      "grad_norm": 0.003616332309320569,
-      "learning_rate": 0.00019999736620079187,
-      "loss": 46.0,
-      "step": 14356
-    },
-    {
-      "epoch": 2.3120898586899634,
-      "grad_norm": 0.001433882280252874,
-      "learning_rate": 0.00019999736583359832,
-      "loss": 46.0,
-      "step": 14357
-    },
-    {
-      "epoch": 2.312250895768751,
-      "grad_norm": 0.0012286021374166012,
-      "learning_rate": 0.00019999736546637915,
-      "loss": 46.0,
-      "step": 14358
-    },
-    {
-      "epoch": 2.3124119328475383,
-      "grad_norm": 0.0017214068211615086,
-      "learning_rate": 0.00019999736509913437,
-      "loss": 46.0,
-      "step": 14359
-    },
-    {
-      "epoch": 2.3125729699263253,
-      "grad_norm": 0.00138735375367105,
-      "learning_rate": 0.00019999736473186403,
-      "loss": 46.0,
-      "step": 14360
-    },
-    {
-      "epoch": 2.312734007005113,
-      "grad_norm": 0.006167803891003132,
-      "learning_rate": 0.0001999973643645681,
-      "loss": 46.0,
-      "step": 14361
-    },
-    {
-      "epoch": 2.3128950440839002,
-      "grad_norm": 0.001048176665790379,
-      "learning_rate": 0.00019999736399724659,
-      "loss": 46.0,
-      "step": 14362
-    },
-    {
-      "epoch": 2.3130560811626877,
-      "grad_norm": 0.0004285548347979784,
-      "learning_rate": 0.00019999736362989946,
-      "loss": 46.0,
-      "step": 14363
-    },
-    {
-      "epoch": 2.313217118241475,
-      "grad_norm": 0.0031015214044600725,
-      "learning_rate": 0.0001999973632625267,
-      "loss": 46.0,
-      "step": 14364
-    },
-    {
-      "epoch": 2.3133781553202626,
-      "grad_norm": 0.005603897850960493,
-      "learning_rate": 0.0001999973628951284,
-      "loss": 46.0,
-      "step": 14365
-    },
-    {
-      "epoch": 2.31353919239905,
-      "grad_norm": 0.0017382488586008549,
-      "learning_rate": 0.0001999973625277045,
-      "loss": 46.0,
-      "step": 14366
-    },
-    {
-      "epoch": 2.3137002294778375,
-      "grad_norm": 0.001150375115685165,
-      "learning_rate": 0.00019999736216025498,
-      "loss": 46.0,
-      "step": 14367
-    },
-    {
-      "epoch": 2.3138612665566245,
-      "grad_norm": 0.00736632663756609,
-      "learning_rate": 0.0001999973617927799,
-      "loss": 46.0,
-      "step": 14368
-    },
-    {
-      "epoch": 2.314022303635412,
-      "grad_norm": 0.0058825197629630566,
-      "learning_rate": 0.00019999736142527918,
-      "loss": 46.0,
-      "step": 14369
-    },
-    {
-      "epoch": 2.3141833407141994,
-      "grad_norm": 0.00473377900198102,
-      "learning_rate": 0.00019999736105775289,
-      "loss": 46.0,
-      "step": 14370
-    },
-    {
-      "epoch": 2.314344377792987,
-      "grad_norm": 0.0009924237383529544,
-      "learning_rate": 0.000199997360690201,
-      "loss": 46.0,
-      "step": 14371
-    },
-    {
-      "epoch": 2.3145054148717743,
-      "grad_norm": 0.0013253850629553199,
-      "learning_rate": 0.00019999736032262356,
-      "loss": 46.0,
-      "step": 14372
-    },
-    {
-      "epoch": 2.3146664519505618,
-      "grad_norm": 0.002478858921676874,
-      "learning_rate": 0.00019999735995502045,
-      "loss": 46.0,
-      "step": 14373
-    },
-    {
-      "epoch": 2.3148274890293488,
-      "grad_norm": 0.009007279761135578,
-      "learning_rate": 0.0001999973595873918,
-      "loss": 46.0,
-      "step": 14374
-    },
-    {
-      "epoch": 2.314988526108136,
-      "grad_norm": 0.0030016270466148853,
-      "learning_rate": 0.00019999735921973755,
-      "loss": 46.0,
-      "step": 14375
-    },
-    {
-      "epoch": 2.3151495631869237,
-      "grad_norm": 0.003416825085878372,
-      "learning_rate": 0.00019999735885205768,
-      "loss": 46.0,
-      "step": 14376
-    },
-    {
-      "epoch": 2.315310600265711,
-      "grad_norm": 0.0080626355484128,
-      "learning_rate": 0.00019999735848435225,
-      "loss": 46.0,
-      "step": 14377
-    },
-    {
-      "epoch": 2.3154716373444986,
-      "grad_norm": 0.003937988542020321,
-      "learning_rate": 0.0001999973581166212,
-      "loss": 46.0,
-      "step": 14378
-    },
-    {
-      "epoch": 2.315632674423286,
-      "grad_norm": 0.006850492209196091,
-      "learning_rate": 0.00019999735774886457,
-      "loss": 46.0,
-      "step": 14379
-    },
-    {
-      "epoch": 2.3157937115020735,
-      "grad_norm": 0.0020145776215940714,
-      "learning_rate": 0.00019999735738108232,
-      "loss": 46.0,
-      "step": 14380
-    },
-    {
-      "epoch": 2.315954748580861,
-      "grad_norm": 0.0019912244752049446,
-      "learning_rate": 0.00019999735701327448,
-      "loss": 46.0,
-      "step": 14381
-    },
-    {
-      "epoch": 2.3161157856596484,
-      "grad_norm": 0.002376051153987646,
-      "learning_rate": 0.0001999973566454411,
-      "loss": 46.0,
-      "step": 14382
-    },
-    {
-      "epoch": 2.3162768227384354,
-      "grad_norm": 0.0009518018341623247,
-      "learning_rate": 0.00019999735627758205,
-      "loss": 46.0,
-      "step": 14383
-    },
-    {
-      "epoch": 2.316437859817223,
-      "grad_norm": 0.007644172292202711,
-      "learning_rate": 0.00019999735590969746,
-      "loss": 46.0,
-      "step": 14384
-    },
-    {
-      "epoch": 2.3165988968960103,
-      "grad_norm": 0.00538233108818531,
-      "learning_rate": 0.00019999735554178724,
-      "loss": 46.0,
-      "step": 14385
-    },
-    {
-      "epoch": 2.3167599339747977,
-      "grad_norm": 0.005443088710308075,
-      "learning_rate": 0.00019999735517385145,
-      "loss": 46.0,
-      "step": 14386
-    },
-    {
-      "epoch": 2.316920971053585,
-      "grad_norm": 0.0018549426458775997,
-      "learning_rate": 0.0001999973548058901,
-      "loss": 46.0,
-      "step": 14387
-    },
-    {
-      "epoch": 2.3170820081323726,
-      "grad_norm": 0.0059515186585485935,
-      "learning_rate": 0.0001999973544379031,
-      "loss": 46.0,
-      "step": 14388
-    },
-    {
-      "epoch": 2.3172430452111596,
-      "grad_norm": 0.0036792298778891563,
-      "learning_rate": 0.0001999973540698905,
-      "loss": 46.0,
-      "step": 14389
-    },
-    {
-      "epoch": 2.317404082289947,
-      "grad_norm": 0.0018364182906225324,
-      "learning_rate": 0.00019999735370185233,
-      "loss": 46.0,
-      "step": 14390
-    },
-    {
-      "epoch": 2.3175651193687345,
-      "grad_norm": 0.0043747033923864365,
-      "learning_rate": 0.00019999735333378857,
-      "loss": 46.0,
-      "step": 14391
-    },
-    {
-      "epoch": 2.317726156447522,
-      "grad_norm": 0.0015284968540072441,
-      "learning_rate": 0.0001999973529656992,
-      "loss": 46.0,
-      "step": 14392
-    },
-    {
-      "epoch": 2.3178871935263095,
-      "grad_norm": 0.0009432507795281708,
-      "learning_rate": 0.00019999735259758423,
-      "loss": 46.0,
-      "step": 14393
-    },
-    {
-      "epoch": 2.318048230605097,
-      "grad_norm": 0.002124601975083351,
-      "learning_rate": 0.00019999735222944368,
-      "loss": 46.0,
-      "step": 14394
-    },
-    {
-      "epoch": 2.3182092676838844,
-      "grad_norm": 0.004489487037062645,
-      "learning_rate": 0.00019999735186127754,
-      "loss": 46.0,
-      "step": 14395
-    },
-    {
-      "epoch": 2.318370304762672,
-      "grad_norm": 0.006924944929778576,
-      "learning_rate": 0.00019999735149308582,
-      "loss": 46.0,
-      "step": 14396
-    },
-    {
-      "epoch": 2.318531341841459,
-      "grad_norm": 0.0012586968950927258,
-      "learning_rate": 0.00019999735112486848,
-      "loss": 46.0,
-      "step": 14397
-    },
-    {
-      "epoch": 2.3186923789202463,
-      "grad_norm": 0.0008771170396357775,
-      "learning_rate": 0.00019999735075662556,
-      "loss": 46.0,
-      "step": 14398
-    },
-    {
-      "epoch": 2.3188534159990337,
-      "grad_norm": 0.0014779956545680761,
-      "learning_rate": 0.00019999735038835702,
-      "loss": 46.0,
-      "step": 14399
-    },
-    {
-      "epoch": 2.319014453077821,
-      "grad_norm": 0.0020623544696718454,
-      "learning_rate": 0.00019999735002006292,
-      "loss": 46.0,
-      "step": 14400
-    },
-    {
-      "epoch": 2.3191754901566086,
-      "grad_norm": 0.0032336018048226833,
-      "learning_rate": 0.0001999973496517432,
-      "loss": 46.0,
-      "step": 14401
-    },
-    {
-      "epoch": 2.319336527235396,
-      "grad_norm": 0.005756643135100603,
-      "learning_rate": 0.0001999973492833979,
-      "loss": 46.0,
-      "step": 14402
-    },
-    {
-      "epoch": 2.3194975643141835,
-      "grad_norm": 0.0010476246243342757,
-      "learning_rate": 0.000199997348915027,
-      "loss": 46.0,
-      "step": 14403
-    },
-    {
-      "epoch": 2.3196586013929705,
-      "grad_norm": 0.006958004552870989,
-      "learning_rate": 0.0001999973485466305,
-      "loss": 46.0,
-      "step": 14404
-    },
-    {
-      "epoch": 2.319819638471758,
-      "grad_norm": 0.0006009237840771675,
-      "learning_rate": 0.00019999734817820842,
-      "loss": 46.0,
-      "step": 14405
-    },
-    {
-      "epoch": 2.3199806755505454,
-      "grad_norm": 0.0033234297297894955,
-      "learning_rate": 0.00019999734780976077,
-      "loss": 46.0,
-      "step": 14406
-    },
-    {
-      "epoch": 2.320141712629333,
-      "grad_norm": 0.004243780858814716,
-      "learning_rate": 0.00019999734744128745,
-      "loss": 46.0,
-      "step": 14407
-    },
-    {
-      "epoch": 2.3203027497081203,
-      "grad_norm": 0.0007655392400920391,
-      "learning_rate": 0.0001999973470727886,
-      "loss": 46.0,
-      "step": 14408
-    },
-    {
-      "epoch": 2.320463786786908,
-      "grad_norm": 0.005025248508900404,
-      "learning_rate": 0.00019999734670426414,
-      "loss": 46.0,
-      "step": 14409
-    },
-    {
-      "epoch": 2.3206248238656952,
-      "grad_norm": 0.0018490883521735668,
-      "learning_rate": 0.00019999734633571408,
-      "loss": 46.0,
-      "step": 14410
-    },
-    {
-      "epoch": 2.3207858609444827,
-      "grad_norm": 0.0029667378403246403,
-      "learning_rate": 0.00019999734596713844,
-      "loss": 46.0,
-      "step": 14411
-    },
-    {
-      "epoch": 2.3209468980232697,
-      "grad_norm": 0.0011171108344569802,
-      "learning_rate": 0.0001999973455985372,
-      "loss": 46.0,
-      "step": 14412
-    },
-    {
-      "epoch": 2.321107935102057,
-      "grad_norm": 0.0028639500960707664,
-      "learning_rate": 0.00019999734522991035,
-      "loss": 46.0,
-      "step": 14413
-    },
-    {
-      "epoch": 2.3212689721808446,
-      "grad_norm": 0.004016688093543053,
-      "learning_rate": 0.00019999734486125792,
-      "loss": 46.0,
-      "step": 14414
-    },
-    {
-      "epoch": 2.321430009259632,
-      "grad_norm": 0.002466451143845916,
-      "learning_rate": 0.00019999734449257988,
-      "loss": 46.0,
-      "step": 14415
-    },
-    {
-      "epoch": 2.3215910463384195,
-      "grad_norm": 0.0021750994492322206,
-      "learning_rate": 0.00019999734412387628,
-      "loss": 46.0,
-      "step": 14416
-    },
-    {
-      "epoch": 2.321752083417207,
-      "grad_norm": 0.0008831341983750463,
-      "learning_rate": 0.00019999734375514706,
-      "loss": 46.0,
-      "step": 14417
-    },
-    {
-      "epoch": 2.321913120495994,
-      "grad_norm": 0.006661385763436556,
-      "learning_rate": 0.00019999734338639225,
-      "loss": 46.0,
-      "step": 14418
-    },
-    {
-      "epoch": 2.3220741575747814,
-      "grad_norm": 0.005850530695170164,
-      "learning_rate": 0.00019999734301761184,
-      "loss": 46.0,
-      "step": 14419
-    },
-    {
-      "epoch": 2.322235194653569,
-      "grad_norm": 0.0030837017111480236,
-      "learning_rate": 0.00019999734264880586,
-      "loss": 46.0,
-      "step": 14420
-    },
-    {
-      "epoch": 2.3223962317323563,
-      "grad_norm": 0.008410700596868992,
-      "learning_rate": 0.00019999734227997424,
-      "loss": 46.0,
-      "step": 14421
-    },
-    {
-      "epoch": 2.3225572688111438,
-      "grad_norm": 0.004040918778628111,
-      "learning_rate": 0.00019999734191111706,
-      "loss": 46.0,
-      "step": 14422
-    },
-    {
-      "epoch": 2.3227183058899312,
-      "grad_norm": 0.004590666387230158,
-      "learning_rate": 0.00019999734154223426,
-      "loss": 46.0,
-      "step": 14423
-    },
-    {
-      "epoch": 2.3228793429687187,
-      "grad_norm": 0.010090491734445095,
-      "learning_rate": 0.00019999734117332588,
-      "loss": 46.0,
-      "step": 14424
-    },
-    {
-      "epoch": 2.323040380047506,
-      "grad_norm": 0.0024465101305395365,
-      "learning_rate": 0.00019999734080439194,
-      "loss": 46.0,
-      "step": 14425
-    },
-    {
-      "epoch": 2.3232014171262936,
-      "grad_norm": 0.0039410837925970554,
-      "learning_rate": 0.00019999734043543236,
-      "loss": 46.0,
-      "step": 14426
-    },
-    {
-      "epoch": 2.3233624542050806,
-      "grad_norm": 0.013590541668236256,
-      "learning_rate": 0.00019999734006644721,
-      "loss": 46.0,
-      "step": 14427
-    },
-    {
-      "epoch": 2.323523491283868,
-      "grad_norm": 0.006381265819072723,
-      "learning_rate": 0.00019999733969743646,
-      "loss": 46.0,
-      "step": 14428
-    },
-    {
-      "epoch": 2.3236845283626555,
-      "grad_norm": 0.002500042086467147,
-      "learning_rate": 0.0001999973393284001,
-      "loss": 46.0,
-      "step": 14429
-    },
-    {
-      "epoch": 2.323845565441443,
-      "grad_norm": 0.001969830598682165,
-      "learning_rate": 0.00019999733895933815,
-      "loss": 46.0,
-      "step": 14430
-    },
-    {
-      "epoch": 2.3240066025202304,
-      "grad_norm": 0.0016278802650049329,
-      "learning_rate": 0.00019999733859025063,
-      "loss": 46.0,
-      "step": 14431
-    },
-    {
-      "epoch": 2.324167639599018,
-      "grad_norm": 0.00324880238622427,
-      "learning_rate": 0.00019999733822113747,
-      "loss": 46.0,
-      "step": 14432
-    },
-    {
-      "epoch": 2.324328676677805,
-      "grad_norm": 0.0015159628819674253,
-      "learning_rate": 0.00019999733785199875,
-      "loss": 46.0,
-      "step": 14433
-    },
-    {
-      "epoch": 2.3244897137565923,
-      "grad_norm": 0.0071920352056622505,
-      "learning_rate": 0.00019999733748283447,
-      "loss": 46.0,
-      "step": 14434
-    },
-    {
-      "epoch": 2.3246507508353798,
-      "grad_norm": 0.0006405320600606501,
-      "learning_rate": 0.00019999733711364455,
-      "loss": 46.0,
-      "step": 14435
-    },
-    {
-      "epoch": 2.324811787914167,
-      "grad_norm": 0.001085036899894476,
-      "learning_rate": 0.00019999733674442904,
-      "loss": 46.0,
-      "step": 14436
-    },
-    {
-      "epoch": 2.3249728249929547,
-      "grad_norm": 0.005076251458376646,
-      "learning_rate": 0.00019999733637518795,
-      "loss": 46.0,
-      "step": 14437
-    },
-    {
-      "epoch": 2.325133862071742,
-      "grad_norm": 0.0016791843809187412,
-      "learning_rate": 0.00019999733600592126,
-      "loss": 46.0,
-      "step": 14438
-    },
-    {
-      "epoch": 2.3252948991505296,
-      "grad_norm": 0.0012755368370562792,
-      "learning_rate": 0.00019999733563662897,
-      "loss": 46.0,
-      "step": 14439
-    },
-    {
-      "epoch": 2.325455936229317,
-      "grad_norm": 0.001654770807363093,
-      "learning_rate": 0.00019999733526731108,
-      "loss": 46.0,
-      "step": 14440
-    },
-    {
-      "epoch": 2.325616973308104,
-      "grad_norm": 0.002825975650921464,
-      "learning_rate": 0.0001999973348979676,
-      "loss": 46.0,
-      "step": 14441
-    },
-    {
-      "epoch": 2.3257780103868915,
-      "grad_norm": 0.0038898123893886805,
-      "learning_rate": 0.00019999733452859855,
-      "loss": 46.0,
-      "step": 14442
-    },
-    {
-      "epoch": 2.325939047465679,
-      "grad_norm": 0.0050439839251339436,
-      "learning_rate": 0.00019999733415920385,
-      "loss": 46.0,
-      "step": 14443
-    },
-    {
-      "epoch": 2.3261000845444664,
-      "grad_norm": 0.0011504930444061756,
-      "learning_rate": 0.0001999973337897836,
-      "loss": 46.0,
-      "step": 14444
-    },
-    {
-      "epoch": 2.326261121623254,
-      "grad_norm": 0.0024441902060061693,
-      "learning_rate": 0.00019999733342033774,
-      "loss": 46.0,
-      "step": 14445
-    },
-    {
-      "epoch": 2.3264221587020413,
-      "grad_norm": 0.0015909976791590452,
-      "learning_rate": 0.0001999973330508663,
-      "loss": 46.0,
-      "step": 14446
-    },
-    {
-      "epoch": 2.3265831957808287,
-      "grad_norm": 0.005187246948480606,
-      "learning_rate": 0.00019999733268136926,
-      "loss": 46.0,
-      "step": 14447
-    },
-    {
-      "epoch": 2.3267442328596157,
-      "grad_norm": 0.0075108944438397884,
-      "learning_rate": 0.00019999733231184662,
-      "loss": 46.0,
-      "step": 14448
-    },
-    {
-      "epoch": 2.326905269938403,
-      "grad_norm": 0.000731126288883388,
-      "learning_rate": 0.0001999973319422984,
-      "loss": 46.0,
-      "step": 14449
-    },
-    {
-      "epoch": 2.3270663070171906,
-      "grad_norm": 0.0032302439212799072,
-      "learning_rate": 0.0001999973315727246,
-      "loss": 46.0,
-      "step": 14450
-    },
-    {
-      "epoch": 2.327227344095978,
-      "grad_norm": 0.0005128925549797714,
-      "learning_rate": 0.00019999733120312514,
-      "loss": 46.0,
-      "step": 14451
-    },
-    {
-      "epoch": 2.3273883811747655,
-      "grad_norm": 0.003352082334458828,
-      "learning_rate": 0.00019999733083350015,
-      "loss": 46.0,
-      "step": 14452
-    },
-    {
-      "epoch": 2.327549418253553,
-      "grad_norm": 0.001001317403279245,
-      "learning_rate": 0.00019999733046384953,
-      "loss": 46.0,
-      "step": 14453
-    },
-    {
-      "epoch": 2.3277104553323404,
-      "grad_norm": 0.004955696407705545,
-      "learning_rate": 0.0001999973300941733,
-      "loss": 46.0,
-      "step": 14454
-    },
-    {
-      "epoch": 2.327871492411128,
-      "grad_norm": 0.001593914465047419,
-      "learning_rate": 0.00019999732972447154,
-      "loss": 46.0,
-      "step": 14455
-    },
-    {
-      "epoch": 2.328032529489915,
-      "grad_norm": 0.001249672845005989,
-      "learning_rate": 0.00019999732935474412,
-      "loss": 46.0,
-      "step": 14456
-    },
-    {
-      "epoch": 2.3281935665687024,
-      "grad_norm": 0.0013835192658007145,
-      "learning_rate": 0.00019999732898499115,
-      "loss": 46.0,
-      "step": 14457
-    },
-    {
-      "epoch": 2.32835460364749,
-      "grad_norm": 0.0015039057470858097,
-      "learning_rate": 0.00019999732861521259,
-      "loss": 46.0,
-      "step": 14458
-    },
-    {
-      "epoch": 2.3285156407262773,
-      "grad_norm": 0.0008849624427966774,
-      "learning_rate": 0.0001999973282454084,
-      "loss": 46.0,
-      "step": 14459
-    },
-    {
-      "epoch": 2.3286766778050647,
-      "grad_norm": 0.001315623288974166,
-      "learning_rate": 0.00019999732787557865,
-      "loss": 46.0,
-      "step": 14460
-    },
-    {
-      "epoch": 2.328837714883852,
-      "grad_norm": 0.0036567416973412037,
-      "learning_rate": 0.0001999973275057233,
-      "loss": 46.0,
-      "step": 14461
-    },
-    {
-      "epoch": 2.328998751962639,
-      "grad_norm": 0.003333244239911437,
-      "learning_rate": 0.0001999973271358423,
-      "loss": 46.0,
-      "step": 14462
-    },
-    {
-      "epoch": 2.3291597890414266,
-      "grad_norm": 0.0010374191915616393,
-      "learning_rate": 0.00019999732676593578,
-      "loss": 46.0,
-      "step": 14463
-    },
-    {
-      "epoch": 2.329320826120214,
-      "grad_norm": 0.004147120285779238,
-      "learning_rate": 0.00019999732639600364,
-      "loss": 46.0,
-      "step": 14464
-    },
-    {
-      "epoch": 2.3294818631990015,
-      "grad_norm": 0.004270139615982771,
-      "learning_rate": 0.0001999973260260459,
-      "loss": 46.0,
-      "step": 14465
-    },
-    {
-      "epoch": 2.329642900277789,
-      "grad_norm": 0.003310170490294695,
-      "learning_rate": 0.00019999732565606257,
-      "loss": 46.0,
-      "step": 14466
-    },
-    {
-      "epoch": 2.3298039373565764,
-      "grad_norm": 0.0004039326449856162,
-      "learning_rate": 0.00019999732528605364,
-      "loss": 46.0,
-      "step": 14467
-    },
-    {
-      "epoch": 2.329964974435364,
-      "grad_norm": 0.006733214017003775,
-      "learning_rate": 0.00019999732491601913,
-      "loss": 46.0,
-      "step": 14468
-    },
-    {
-      "epoch": 2.3301260115141513,
-      "grad_norm": 0.004155580885708332,
-      "learning_rate": 0.000199997324545959,
-      "loss": 46.0,
-      "step": 14469
-    },
-    {
-      "epoch": 2.3302870485929383,
-      "grad_norm": 0.011322176083922386,
-      "learning_rate": 0.00019999732417587328,
-      "loss": 46.0,
-      "step": 14470
-    },
-    {
-      "epoch": 2.330448085671726,
-      "grad_norm": 0.0024532803799957037,
-      "learning_rate": 0.000199997323805762,
-      "loss": 46.0,
-      "step": 14471
-    },
-    {
-      "epoch": 2.3306091227505132,
-      "grad_norm": 0.0013457033783197403,
-      "learning_rate": 0.00019999732343562511,
-      "loss": 46.0,
-      "step": 14472
-    },
-    {
-      "epoch": 2.3307701598293007,
-      "grad_norm": 0.003234673058614135,
-      "learning_rate": 0.0001999973230654626,
-      "loss": 46.0,
-      "step": 14473
-    },
-    {
-      "epoch": 2.330931196908088,
-      "grad_norm": 0.008821835741400719,
-      "learning_rate": 0.00019999732269527454,
-      "loss": 46.0,
-      "step": 14474
-    },
-    {
-      "epoch": 2.3310922339868756,
-      "grad_norm": 0.004558505956083536,
-      "learning_rate": 0.00019999732232506086,
-      "loss": 46.0,
-      "step": 14475
-    },
-    {
-      "epoch": 2.331253271065663,
-      "grad_norm": 0.004502071533352137,
-      "learning_rate": 0.0001999973219548216,
-      "loss": 46.0,
-      "step": 14476
-    },
-    {
-      "epoch": 2.33141430814445,
-      "grad_norm": 0.0018074504332616925,
-      "learning_rate": 0.00019999732158455672,
-      "loss": 46.0,
-      "step": 14477
-    },
-    {
-      "epoch": 2.3315753452232375,
-      "grad_norm": 0.004936224315315485,
-      "learning_rate": 0.00019999732121426625,
-      "loss": 46.0,
-      "step": 14478
-    },
-    {
-      "epoch": 2.331736382302025,
-      "grad_norm": 0.0023984541185200214,
-      "learning_rate": 0.0001999973208439502,
-      "loss": 46.0,
-      "step": 14479
-    },
-    {
-      "epoch": 2.3318974193808124,
-      "grad_norm": 0.002477796282619238,
-      "learning_rate": 0.00019999732047360852,
-      "loss": 46.0,
-      "step": 14480
-    },
-    {
-      "epoch": 2.3320584564596,
-      "grad_norm": 0.000671895919367671,
-      "learning_rate": 0.0001999973201032413,
-      "loss": 46.0,
-      "step": 14481
-    },
-    {
-      "epoch": 2.3322194935383873,
-      "grad_norm": 0.011366534046828747,
-      "learning_rate": 0.00019999731973284845,
-      "loss": 46.0,
-      "step": 14482
-    },
-    {
-      "epoch": 2.3323805306171748,
-      "grad_norm": 0.005910404492169619,
-      "learning_rate": 0.00019999731936243002,
-      "loss": 46.0,
-      "step": 14483
-    },
-    {
-      "epoch": 2.332541567695962,
-      "grad_norm": 0.0013881241902709007,
-      "learning_rate": 0.00019999731899198603,
-      "loss": 46.0,
-      "step": 14484
-    },
-    {
-      "epoch": 2.3327026047747492,
-      "grad_norm": 0.004842846654355526,
-      "learning_rate": 0.0001999973186215164,
-      "loss": 46.0,
-      "step": 14485
-    },
-    {
-      "epoch": 2.3328636418535367,
-      "grad_norm": 0.0016857337905094028,
-      "learning_rate": 0.00019999731825102117,
-      "loss": 46.0,
-      "step": 14486
-    },
-    {
-      "epoch": 2.333024678932324,
-      "grad_norm": 0.003317498601973057,
-      "learning_rate": 0.00019999731788050037,
-      "loss": 46.0,
-      "step": 14487
-    },
-    {
-      "epoch": 2.3331857160111116,
-      "grad_norm": 0.0031158719211816788,
-      "learning_rate": 0.00019999731750995397,
-      "loss": 46.0,
-      "step": 14488
-    },
-    {
-      "epoch": 2.333346753089899,
-      "grad_norm": 0.0009414818487130105,
-      "learning_rate": 0.00019999731713938194,
-      "loss": 46.0,
-      "step": 14489
-    },
-    {
-      "epoch": 2.3335077901686865,
-      "grad_norm": 0.0009883511811494827,
-      "learning_rate": 0.00019999731676878434,
-      "loss": 46.0,
-      "step": 14490
-    },
-    {
-      "epoch": 2.3336688272474735,
-      "grad_norm": 0.005314912647008896,
-      "learning_rate": 0.00019999731639816116,
-      "loss": 46.0,
-      "step": 14491
-    },
-    {
-      "epoch": 2.333829864326261,
-      "grad_norm": 0.0005769595736637712,
-      "learning_rate": 0.0001999973160275124,
-      "loss": 46.0,
-      "step": 14492
-    },
-    {
-      "epoch": 2.3339909014050484,
-      "grad_norm": 0.0016682537971064448,
-      "learning_rate": 0.00019999731565683803,
-      "loss": 46.0,
-      "step": 14493
-    },
-    {
-      "epoch": 2.334151938483836,
-      "grad_norm": 0.0011705616489052773,
-      "learning_rate": 0.00019999731528613806,
-      "loss": 46.0,
-      "step": 14494
-    },
-    {
-      "epoch": 2.3343129755626233,
-      "grad_norm": 0.0011205258779227734,
-      "learning_rate": 0.0001999973149154125,
-      "loss": 46.0,
-      "step": 14495
-    },
-    {
-      "epoch": 2.3344740126414107,
-      "grad_norm": 0.0009598342003300786,
-      "learning_rate": 0.00019999731454466133,
-      "loss": 46.0,
-      "step": 14496
-    },
-    {
-      "epoch": 2.334635049720198,
-      "grad_norm": 0.004641356877982616,
-      "learning_rate": 0.0001999973141738846,
-      "loss": 46.0,
-      "step": 14497
-    },
-    {
-      "epoch": 2.3347960867989856,
-      "grad_norm": 0.005388359073549509,
-      "learning_rate": 0.00019999731380308222,
-      "loss": 46.0,
-      "step": 14498
-    },
-    {
-      "epoch": 2.334957123877773,
-      "grad_norm": 0.0012352181365713477,
-      "learning_rate": 0.0001999973134322543,
-      "loss": 46.0,
-      "step": 14499
-    },
-    {
-      "epoch": 2.33511816095656,
-      "grad_norm": 0.003375965403392911,
-      "learning_rate": 0.00019999731306140074,
-      "loss": 46.0,
-      "step": 14500
-    },
-    {
-      "epoch": 2.3352791980353476,
-      "grad_norm": 0.008423916064202785,
-      "learning_rate": 0.00019999731269052163,
-      "loss": 46.0,
-      "step": 14501
-    },
-    {
-      "epoch": 2.335440235114135,
-      "grad_norm": 0.005135141778737307,
-      "learning_rate": 0.0001999973123196169,
-      "loss": 46.0,
-      "step": 14502
-    },
-    {
-      "epoch": 2.3356012721929225,
-      "grad_norm": 0.0012736058561131358,
-      "learning_rate": 0.0001999973119486866,
-      "loss": 46.0,
-      "step": 14503
-    },
-    {
-      "epoch": 2.33576230927171,
-      "grad_norm": 0.004501240327954292,
-      "learning_rate": 0.00019999731157773067,
-      "loss": 46.0,
-      "step": 14504
-    },
-    {
-      "epoch": 2.3359233463504974,
-      "grad_norm": 0.0011809362331405282,
-      "learning_rate": 0.0001999973112067492,
-      "loss": 46.0,
-      "step": 14505
-    },
-    {
-      "epoch": 2.3360843834292844,
-      "grad_norm": 0.005849318578839302,
-      "learning_rate": 0.00019999731083574206,
-      "loss": 46.0,
-      "step": 14506
-    },
-    {
-      "epoch": 2.336245420508072,
-      "grad_norm": 0.007024803198873997,
-      "learning_rate": 0.0001999973104647094,
-      "loss": 46.0,
-      "step": 14507
-    },
-    {
-      "epoch": 2.3364064575868593,
-      "grad_norm": 0.0015148492529988289,
-      "learning_rate": 0.0001999973100936511,
-      "loss": 46.0,
-      "step": 14508
-    },
-    {
-      "epoch": 2.3365674946656467,
-      "grad_norm": 0.0012120538158342242,
-      "learning_rate": 0.00019999730972256722,
-      "loss": 46.0,
-      "step": 14509
-    },
-    {
-      "epoch": 2.336728531744434,
-      "grad_norm": 0.0033014232758432627,
-      "learning_rate": 0.00019999730935145774,
-      "loss": 46.0,
-      "step": 14510
-    },
-    {
-      "epoch": 2.3368895688232216,
-      "grad_norm": 0.006597817875444889,
-      "learning_rate": 0.00019999730898032268,
-      "loss": 46.0,
-      "step": 14511
-    },
-    {
-      "epoch": 2.337050605902009,
-      "grad_norm": 0.0015597755555063486,
-      "learning_rate": 0.000199997308609162,
-      "loss": 46.0,
-      "step": 14512
-    },
-    {
-      "epoch": 2.3372116429807965,
-      "grad_norm": 0.005910459440201521,
-      "learning_rate": 0.00019999730823797577,
-      "loss": 46.0,
-      "step": 14513
-    },
-    {
-      "epoch": 2.3373726800595835,
-      "grad_norm": 0.006780663505196571,
-      "learning_rate": 0.0001999973078667639,
-      "loss": 46.0,
-      "step": 14514
-    },
-    {
-      "epoch": 2.337533717138371,
-      "grad_norm": 0.004193446133285761,
-      "learning_rate": 0.00019999730749552645,
-      "loss": 46.0,
-      "step": 14515
-    },
-    {
-      "epoch": 2.3376947542171584,
-      "grad_norm": 0.0010009112302213907,
-      "learning_rate": 0.0001999973071242634,
-      "loss": 46.0,
-      "step": 14516
-    },
-    {
-      "epoch": 2.337855791295946,
-      "grad_norm": 0.003624504432082176,
-      "learning_rate": 0.0001999973067529748,
-      "loss": 46.0,
-      "step": 14517
-    },
-    {
-      "epoch": 2.3380168283747333,
-      "grad_norm": 0.004376953933387995,
-      "learning_rate": 0.00019999730638166056,
-      "loss": 46.0,
-      "step": 14518
-    },
-    {
-      "epoch": 2.338177865453521,
-      "grad_norm": 0.005358573514968157,
-      "learning_rate": 0.00019999730601032072,
-      "loss": 46.0,
-      "step": 14519
-    },
-    {
-      "epoch": 2.3383389025323082,
-      "grad_norm": 0.006792285013943911,
-      "learning_rate": 0.00019999730563895532,
-      "loss": 46.0,
-      "step": 14520
-    },
-    {
-      "epoch": 2.3384999396110953,
-      "grad_norm": 0.004689349792897701,
-      "learning_rate": 0.00019999730526756428,
-      "loss": 46.0,
-      "step": 14521
-    },
-    {
-      "epoch": 2.3386609766898827,
-      "grad_norm": 0.003819323144853115,
-      "learning_rate": 0.0001999973048961477,
-      "loss": 46.0,
-      "step": 14522
-    },
-    {
-      "epoch": 2.33882201376867,
-      "grad_norm": 0.004058188758790493,
-      "learning_rate": 0.0001999973045247055,
-      "loss": 46.0,
-      "step": 14523
-    },
-    {
-      "epoch": 2.3389830508474576,
-      "grad_norm": 0.001914954511448741,
-      "learning_rate": 0.0001999973041532377,
-      "loss": 46.0,
-      "step": 14524
-    },
-    {
-      "epoch": 2.339144087926245,
-      "grad_norm": 0.005341460462659597,
-      "learning_rate": 0.00019999730378174432,
-      "loss": 46.0,
-      "step": 14525
-    },
-    {
-      "epoch": 2.3393051250050325,
-      "grad_norm": 0.0023206814657896757,
-      "learning_rate": 0.00019999730341022535,
-      "loss": 46.0,
-      "step": 14526
-    },
-    {
-      "epoch": 2.33946616208382,
-      "grad_norm": 0.0015874041710048914,
-      "learning_rate": 0.00019999730303868075,
-      "loss": 46.0,
-      "step": 14527
-    },
-    {
-      "epoch": 2.3396271991626074,
-      "grad_norm": 0.002369649475440383,
-      "learning_rate": 0.00019999730266711057,
-      "loss": 46.0,
-      "step": 14528
-    },
-    {
-      "epoch": 2.3397882362413944,
-      "grad_norm": 0.001215968863107264,
-      "learning_rate": 0.00019999730229551484,
-      "loss": 46.0,
-      "step": 14529
-    },
-    {
-      "epoch": 2.339949273320182,
-      "grad_norm": 0.005948281846940517,
-      "learning_rate": 0.00019999730192389346,
-      "loss": 46.0,
-      "step": 14530
-    },
-    {
-      "epoch": 2.3401103103989693,
-      "grad_norm": 0.0033018679823726416,
-      "learning_rate": 0.0001999973015522465,
-      "loss": 46.0,
-      "step": 14531
-    },
-    {
-      "epoch": 2.340271347477757,
-      "grad_norm": 0.0013913020957261324,
-      "learning_rate": 0.00019999730118057396,
-      "loss": 46.0,
-      "step": 14532
-    },
-    {
-      "epoch": 2.3404323845565442,
-      "grad_norm": 0.013141454197466373,
-      "learning_rate": 0.00019999730080887582,
-      "loss": 46.0,
-      "step": 14533
-    },
-    {
-      "epoch": 2.3405934216353317,
-      "grad_norm": 0.010674758814275265,
-      "learning_rate": 0.0001999973004371521,
-      "loss": 46.0,
-      "step": 14534
-    },
-    {
-      "epoch": 2.3407544587141187,
-      "grad_norm": 0.0015034181997179985,
-      "learning_rate": 0.00019999730006540274,
-      "loss": 46.0,
-      "step": 14535
-    },
-    {
-      "epoch": 2.340915495792906,
-      "grad_norm": 0.0012631191639229655,
-      "learning_rate": 0.00019999729969362781,
-      "loss": 46.0,
-      "step": 14536
-    },
-    {
-      "epoch": 2.3410765328716936,
-      "grad_norm": 0.002949714893475175,
-      "learning_rate": 0.00019999729932182732,
-      "loss": 46.0,
-      "step": 14537
-    },
-    {
-      "epoch": 2.341237569950481,
-      "grad_norm": 0.00282492209225893,
-      "learning_rate": 0.0001999972989500012,
-      "loss": 46.0,
-      "step": 14538
-    },
-    {
-      "epoch": 2.3413986070292685,
-      "grad_norm": 0.0036208292003721,
-      "learning_rate": 0.00019999729857814947,
-      "loss": 46.0,
-      "step": 14539
-    },
-    {
-      "epoch": 2.341559644108056,
-      "grad_norm": 0.003570919157937169,
-      "learning_rate": 0.0001999972982062722,
-      "loss": 46.0,
-      "step": 14540
-    },
-    {
-      "epoch": 2.3417206811868434,
-      "grad_norm": 0.0020160884596407413,
-      "learning_rate": 0.0001999972978343693,
-      "loss": 46.0,
-      "step": 14541
-    },
-    {
-      "epoch": 2.341881718265631,
-      "grad_norm": 0.004158018156886101,
-      "learning_rate": 0.0001999972974624408,
-      "loss": 46.0,
-      "step": 14542
-    },
-    {
-      "epoch": 2.342042755344418,
-      "grad_norm": 0.005943570286035538,
-      "learning_rate": 0.00019999729709048672,
-      "loss": 46.0,
-      "step": 14543
-    },
-    {
-      "epoch": 2.3422037924232053,
-      "grad_norm": 0.0007683020667172968,
-      "learning_rate": 0.00019999729671850704,
-      "loss": 46.0,
-      "step": 14544
-    },
-    {
-      "epoch": 2.3423648295019928,
-      "grad_norm": 0.010151375085115433,
-      "learning_rate": 0.00019999729634650177,
-      "loss": 46.0,
-      "step": 14545
-    },
-    {
-      "epoch": 2.34252586658078,
-      "grad_norm": 0.0011483295820653439,
-      "learning_rate": 0.0001999972959744709,
-      "loss": 46.0,
-      "step": 14546
-    },
-    {
-      "epoch": 2.3426869036595677,
-      "grad_norm": 0.0015884640160948038,
-      "learning_rate": 0.00019999729560241444,
-      "loss": 46.0,
-      "step": 14547
-    },
-    {
-      "epoch": 2.342847940738355,
-      "grad_norm": 0.0023586833849549294,
-      "learning_rate": 0.00019999729523033239,
-      "loss": 46.0,
-      "step": 14548
-    },
-    {
-      "epoch": 2.3430089778171426,
-      "grad_norm": 0.002251328667625785,
-      "learning_rate": 0.00019999729485822474,
-      "loss": 46.0,
-      "step": 14549
-    },
-    {
-      "epoch": 2.3431700148959296,
-      "grad_norm": 0.002140614204108715,
-      "learning_rate": 0.0001999972944860915,
-      "loss": 46.0,
-      "step": 14550
-    },
-    {
-      "epoch": 2.343331051974717,
-      "grad_norm": 0.0035234992392361164,
-      "learning_rate": 0.00019999729411393266,
-      "loss": 46.0,
-      "step": 14551
-    },
-    {
-      "epoch": 2.3434920890535045,
-      "grad_norm": 0.0005595273105427623,
-      "learning_rate": 0.00019999729374174825,
-      "loss": 46.0,
-      "step": 14552
-    },
-    {
-      "epoch": 2.343653126132292,
-      "grad_norm": 0.0020130749326199293,
-      "learning_rate": 0.0001999972933695382,
-      "loss": 46.0,
-      "step": 14553
-    },
-    {
-      "epoch": 2.3438141632110794,
-      "grad_norm": 0.004177992232143879,
-      "learning_rate": 0.0001999972929973026,
-      "loss": 46.0,
-      "step": 14554
-    },
-    {
-      "epoch": 2.343975200289867,
-      "grad_norm": 0.0009268194553442299,
-      "learning_rate": 0.00019999729262504138,
-      "loss": 46.0,
-      "step": 14555
-    },
-    {
-      "epoch": 2.3441362373686543,
-      "grad_norm": 0.004611593205481768,
-      "learning_rate": 0.00019999729225275457,
-      "loss": 46.0,
-      "step": 14556
-    },
-    {
-      "epoch": 2.3442972744474417,
-      "grad_norm": 0.0008703362545929849,
-      "learning_rate": 0.00019999729188044217,
-      "loss": 46.0,
-      "step": 14557
-    },
-    {
-      "epoch": 2.3444583115262287,
-      "grad_norm": 0.00476370844990015,
-      "learning_rate": 0.00019999729150810418,
-      "loss": 46.0,
-      "step": 14558
-    },
-    {
-      "epoch": 2.344619348605016,
-      "grad_norm": 0.0023048759903758764,
-      "learning_rate": 0.00019999729113574056,
-      "loss": 46.0,
-      "step": 14559
-    },
-    {
-      "epoch": 2.3447803856838036,
-      "grad_norm": 0.005057384259998798,
-      "learning_rate": 0.00019999729076335137,
-      "loss": 46.0,
-      "step": 14560
-    },
-    {
-      "epoch": 2.344941422762591,
-      "grad_norm": 0.007750918157398701,
-      "learning_rate": 0.00019999729039093663,
-      "loss": 46.0,
-      "step": 14561
-    },
-    {
-      "epoch": 2.3451024598413786,
-      "grad_norm": 0.0019412693800404668,
-      "learning_rate": 0.00019999729001849624,
-      "loss": 46.0,
-      "step": 14562
-    },
-    {
-      "epoch": 2.345263496920166,
-      "grad_norm": 0.0013990812003612518,
-      "learning_rate": 0.00019999728964603027,
-      "loss": 46.0,
-      "step": 14563
-    },
-    {
-      "epoch": 2.345424533998953,
-      "grad_norm": 0.003175352932885289,
-      "learning_rate": 0.0001999972892735387,
-      "loss": 46.0,
-      "step": 14564
-    },
-    {
-      "epoch": 2.3455855710777405,
-      "grad_norm": 0.0075635043904185295,
-      "learning_rate": 0.00019999728890102156,
-      "loss": 46.0,
-      "step": 14565
-    },
-    {
-      "epoch": 2.345746608156528,
-      "grad_norm": 0.008029410615563393,
-      "learning_rate": 0.00019999728852847882,
-      "loss": 46.0,
-      "step": 14566
-    },
-    {
-      "epoch": 2.3459076452353154,
-      "grad_norm": 0.0016418209997937083,
-      "learning_rate": 0.00019999728815591044,
-      "loss": 46.0,
-      "step": 14567
-    },
-    {
-      "epoch": 2.346068682314103,
-      "grad_norm": 0.003333429340273142,
-      "learning_rate": 0.00019999728778331654,
-      "loss": 46.0,
-      "step": 14568
-    },
-    {
-      "epoch": 2.3462297193928903,
-      "grad_norm": 0.004290881101042032,
-      "learning_rate": 0.000199997287410697,
-      "loss": 46.0,
-      "step": 14569
-    },
-    {
-      "epoch": 2.3463907564716777,
-      "grad_norm": 0.0023911602329462767,
-      "learning_rate": 0.00019999728703805187,
-      "loss": 46.0,
-      "step": 14570
-    },
-    {
-      "epoch": 2.346551793550465,
-      "grad_norm": 0.007665710058063269,
-      "learning_rate": 0.00019999728666538115,
-      "loss": 46.0,
-      "step": 14571
-    },
-    {
-      "epoch": 2.3467128306292526,
-      "grad_norm": 0.0028941675554960966,
-      "learning_rate": 0.00019999728629268483,
-      "loss": 46.0,
-      "step": 14572
-    },
-    {
-      "epoch": 2.3468738677080396,
-      "grad_norm": 0.0006980894249863923,
-      "learning_rate": 0.00019999728591996293,
-      "loss": 46.0,
-      "step": 14573
-    },
-    {
-      "epoch": 2.347034904786827,
-      "grad_norm": 0.0026659141294658184,
-      "learning_rate": 0.00019999728554721542,
-      "loss": 46.0,
-      "step": 14574
-    },
-    {
-      "epoch": 2.3471959418656145,
-      "grad_norm": 0.006842970848083496,
-      "learning_rate": 0.00019999728517444232,
-      "loss": 46.0,
-      "step": 14575
-    },
-    {
-      "epoch": 2.347356978944402,
-      "grad_norm": 0.007536585908383131,
-      "learning_rate": 0.00019999728480164363,
-      "loss": 46.0,
-      "step": 14576
-    },
-    {
-      "epoch": 2.3475180160231894,
-      "grad_norm": 0.0008965461165644228,
-      "learning_rate": 0.00019999728442881935,
-      "loss": 46.0,
-      "step": 14577
-    },
-    {
-      "epoch": 2.347679053101977,
-      "grad_norm": 0.003250017063692212,
-      "learning_rate": 0.00019999728405596946,
-      "loss": 46.0,
-      "step": 14578
-    },
-    {
-      "epoch": 2.347840090180764,
-      "grad_norm": 0.00532780634239316,
-      "learning_rate": 0.00019999728368309399,
-      "loss": 46.0,
-      "step": 14579
-    },
-    {
-      "epoch": 2.3480011272595513,
-      "grad_norm": 0.0025089976843446493,
-      "learning_rate": 0.00019999728331019295,
-      "loss": 46.0,
-      "step": 14580
-    },
-    {
-      "epoch": 2.348162164338339,
-      "grad_norm": 0.0033368461299687624,
-      "learning_rate": 0.00019999728293726627,
-      "loss": 46.0,
-      "step": 14581
-    },
-    {
-      "epoch": 2.3483232014171262,
-      "grad_norm": 0.000839138578157872,
-      "learning_rate": 0.00019999728256431403,
-      "loss": 46.0,
-      "step": 14582
-    },
-    {
-      "epoch": 2.3484842384959137,
-      "grad_norm": 0.001168245100416243,
-      "learning_rate": 0.00019999728219133615,
-      "loss": 46.0,
-      "step": 14583
-    },
-    {
-      "epoch": 2.348645275574701,
-      "grad_norm": 0.003142967354506254,
-      "learning_rate": 0.0001999972818183327,
-      "loss": 46.0,
-      "step": 14584
-    },
-    {
-      "epoch": 2.3488063126534886,
-      "grad_norm": 0.005092882085591555,
-      "learning_rate": 0.00019999728144530368,
-      "loss": 46.0,
-      "step": 14585
-    },
-    {
-      "epoch": 2.348967349732276,
-      "grad_norm": 0.010230480693280697,
-      "learning_rate": 0.00019999728107224904,
-      "loss": 46.0,
-      "step": 14586
-    },
-    {
-      "epoch": 2.349128386811063,
-      "grad_norm": 0.0034412737004458904,
-      "learning_rate": 0.00019999728069916884,
-      "loss": 46.0,
-      "step": 14587
-    },
-    {
-      "epoch": 2.3492894238898505,
-      "grad_norm": 0.012240048497915268,
-      "learning_rate": 0.000199997280326063,
-      "loss": 46.0,
-      "step": 14588
-    },
-    {
-      "epoch": 2.349450460968638,
-      "grad_norm": 0.009447953663766384,
-      "learning_rate": 0.0001999972799529316,
-      "loss": 46.0,
-      "step": 14589
-    },
-    {
-      "epoch": 2.3496114980474254,
-      "grad_norm": 0.0072211576625704765,
-      "learning_rate": 0.00019999727957977458,
-      "loss": 46.0,
-      "step": 14590
-    },
-    {
-      "epoch": 2.349772535126213,
-      "grad_norm": 0.0038550281897187233,
-      "learning_rate": 0.00019999727920659197,
-      "loss": 46.0,
-      "step": 14591
-    },
-    {
-      "epoch": 2.3499335722050003,
-      "grad_norm": 0.005188861396163702,
-      "learning_rate": 0.00019999727883338378,
-      "loss": 46.0,
-      "step": 14592
-    },
-    {
-      "epoch": 2.3500946092837878,
-      "grad_norm": 0.0036283619701862335,
-      "learning_rate": 0.00019999727846014997,
-      "loss": 46.0,
-      "step": 14593
-    },
-    {
-      "epoch": 2.350255646362575,
-      "grad_norm": 0.0006508153164759278,
-      "learning_rate": 0.00019999727808689058,
-      "loss": 46.0,
-      "step": 14594
-    },
-    {
-      "epoch": 2.3504166834413622,
-      "grad_norm": 0.009944497607648373,
-      "learning_rate": 0.0001999972777136056,
-      "loss": 46.0,
-      "step": 14595
-    },
-    {
-      "epoch": 2.3505777205201497,
-      "grad_norm": 0.009078366681933403,
-      "learning_rate": 0.00019999727734029503,
-      "loss": 46.0,
-      "step": 14596
-    },
-    {
-      "epoch": 2.350738757598937,
-      "grad_norm": 0.0038890123832970858,
-      "learning_rate": 0.00019999727696695888,
-      "loss": 46.0,
-      "step": 14597
-    },
-    {
-      "epoch": 2.3508997946777246,
-      "grad_norm": 0.0009394356166012585,
-      "learning_rate": 0.0001999972765935971,
-      "loss": 46.0,
-      "step": 14598
-    },
-    {
-      "epoch": 2.351060831756512,
-      "grad_norm": 0.002508430043235421,
-      "learning_rate": 0.00019999727622020975,
-      "loss": 46.0,
-      "step": 14599
-    },
-    {
-      "epoch": 2.3512218688352995,
-      "grad_norm": 0.010747738182544708,
-      "learning_rate": 0.00019999727584679678,
-      "loss": 46.0,
-      "step": 14600
-    },
-    {
-      "epoch": 2.351382905914087,
-      "grad_norm": 0.0017628934001550078,
-      "learning_rate": 0.00019999727547335825,
-      "loss": 46.0,
-      "step": 14601
-    },
-    {
-      "epoch": 2.351543942992874,
-      "grad_norm": 0.0013982991222292185,
-      "learning_rate": 0.0001999972750998941,
-      "loss": 46.0,
-      "step": 14602
-    },
-    {
-      "epoch": 2.3517049800716614,
-      "grad_norm": 0.006652480456978083,
-      "learning_rate": 0.00019999727472640438,
-      "loss": 46.0,
-      "step": 14603
-    },
-    {
-      "epoch": 2.351866017150449,
-      "grad_norm": 0.010210984386503696,
-      "learning_rate": 0.00019999727435288906,
-      "loss": 46.0,
-      "step": 14604
-    },
-    {
-      "epoch": 2.3520270542292363,
-      "grad_norm": 0.002986464649438858,
-      "learning_rate": 0.00019999727397934812,
-      "loss": 46.0,
-      "step": 14605
-    },
-    {
-      "epoch": 2.3521880913080238,
-      "grad_norm": 0.004176314454525709,
-      "learning_rate": 0.00019999727360578163,
-      "loss": 46.0,
-      "step": 14606
-    },
-    {
-      "epoch": 2.352349128386811,
-      "grad_norm": 0.006032838951796293,
-      "learning_rate": 0.0001999972732321895,
-      "loss": 46.0,
-      "step": 14607
-    },
-    {
-      "epoch": 2.352510165465598,
-      "grad_norm": 0.002583032939583063,
-      "learning_rate": 0.0001999972728585718,
-      "loss": 46.0,
-      "step": 14608
-    },
-    {
-      "epoch": 2.3526712025443857,
-      "grad_norm": 0.0016644189599901438,
-      "learning_rate": 0.0001999972724849285,
-      "loss": 46.0,
-      "step": 14609
-    },
-    {
-      "epoch": 2.352832239623173,
-      "grad_norm": 0.0020572547800838947,
-      "learning_rate": 0.0001999972721112596,
-      "loss": 46.0,
-      "step": 14610
-    },
-    {
-      "epoch": 2.3529932767019606,
-      "grad_norm": 0.005446291994303465,
-      "learning_rate": 0.0001999972717375651,
-      "loss": 46.0,
-      "step": 14611
-    },
-    {
-      "epoch": 2.353154313780748,
-      "grad_norm": 0.003693035803735256,
-      "learning_rate": 0.00019999727136384504,
-      "loss": 46.0,
-      "step": 14612
-    },
-    {
-      "epoch": 2.3533153508595355,
-      "grad_norm": 0.0006589462282136083,
-      "learning_rate": 0.00019999727099009938,
-      "loss": 46.0,
-      "step": 14613
-    },
-    {
-      "epoch": 2.353476387938323,
-      "grad_norm": 0.0004005154187325388,
-      "learning_rate": 0.00019999727061632808,
-      "loss": 46.0,
-      "step": 14614
-    },
-    {
-      "epoch": 2.3536374250171104,
-      "grad_norm": 0.004574046935886145,
-      "learning_rate": 0.00019999727024253122,
-      "loss": 46.0,
-      "step": 14615
-    },
-    {
-      "epoch": 2.353798462095898,
-      "grad_norm": 0.003939672838896513,
-      "learning_rate": 0.00019999726986870878,
-      "loss": 46.0,
-      "step": 14616
-    },
-    {
-      "epoch": 2.353959499174685,
-      "grad_norm": 0.0012294218176975846,
-      "learning_rate": 0.00019999726949486072,
-      "loss": 46.0,
-      "step": 14617
-    },
-    {
-      "epoch": 2.3541205362534723,
-      "grad_norm": 0.005164221860468388,
-      "learning_rate": 0.00019999726912098707,
-      "loss": 46.0,
-      "step": 14618
-    },
-    {
-      "epoch": 2.3542815733322597,
-      "grad_norm": 0.0019406527280807495,
-      "learning_rate": 0.0001999972687470878,
-      "loss": 46.0,
-      "step": 14619
-    },
-    {
-      "epoch": 2.354442610411047,
-      "grad_norm": 0.0016347814816981554,
-      "learning_rate": 0.00019999726837316298,
-      "loss": 46.0,
-      "step": 14620
-    },
-    {
-      "epoch": 2.3546036474898346,
-      "grad_norm": 0.0034431861713528633,
-      "learning_rate": 0.00019999726799921257,
-      "loss": 46.0,
-      "step": 14621
-    },
-    {
-      "epoch": 2.354764684568622,
-      "grad_norm": 0.008472124114632607,
-      "learning_rate": 0.00019999726762523655,
-      "loss": 46.0,
-      "step": 14622
-    },
-    {
-      "epoch": 2.354925721647409,
-      "grad_norm": 0.003613944398239255,
-      "learning_rate": 0.0001999972672512349,
-      "loss": 46.0,
-      "step": 14623
-    },
-    {
-      "epoch": 2.3550867587261965,
-      "grad_norm": 0.007264977321028709,
-      "learning_rate": 0.00019999726687720771,
-      "loss": 46.0,
-      "step": 14624
-    },
-    {
-      "epoch": 2.355247795804984,
-      "grad_norm": 0.0028498023748397827,
-      "learning_rate": 0.0001999972665031549,
-      "loss": 46.0,
-      "step": 14625
-    },
-    {
-      "epoch": 2.3554088328837715,
-      "grad_norm": 0.002462163334712386,
-      "learning_rate": 0.0001999972661290765,
-      "loss": 46.0,
-      "step": 14626
-    },
-    {
-      "epoch": 2.355569869962559,
-      "grad_norm": 0.0006279117078520358,
-      "learning_rate": 0.00019999726575497252,
-      "loss": 46.0,
-      "step": 14627
-    },
-    {
-      "epoch": 2.3557309070413464,
-      "grad_norm": 0.004870972596108913,
-      "learning_rate": 0.00019999726538084294,
-      "loss": 46.0,
-      "step": 14628
-    },
-    {
-      "epoch": 2.355891944120134,
-      "grad_norm": 0.0027601835317909718,
-      "learning_rate": 0.00019999726500668773,
-      "loss": 46.0,
-      "step": 14629
-    },
-    {
-      "epoch": 2.3560529811989213,
-      "grad_norm": 0.004281362984329462,
-      "learning_rate": 0.00019999726463250695,
-      "loss": 46.0,
-      "step": 14630
-    },
-    {
-      "epoch": 2.3562140182777083,
-      "grad_norm": 0.0011230497620999813,
-      "learning_rate": 0.00019999726425830062,
-      "loss": 46.0,
-      "step": 14631
-    },
-    {
-      "epoch": 2.3563750553564957,
-      "grad_norm": 0.01176318246871233,
-      "learning_rate": 0.00019999726388406864,
-      "loss": 46.0,
-      "step": 14632
-    },
-    {
-      "epoch": 2.356536092435283,
-      "grad_norm": 0.011590281501412392,
-      "learning_rate": 0.00019999726350981107,
-      "loss": 46.0,
-      "step": 14633
-    },
-    {
-      "epoch": 2.3566971295140706,
-      "grad_norm": 0.0013203065609559417,
-      "learning_rate": 0.00019999726313552792,
-      "loss": 46.0,
-      "step": 14634
-    },
-    {
-      "epoch": 2.356858166592858,
-      "grad_norm": 0.0014492750633507967,
-      "learning_rate": 0.00019999726276121919,
-      "loss": 46.0,
-      "step": 14635
-    },
-    {
-      "epoch": 2.3570192036716455,
-      "grad_norm": 0.001195266260765493,
-      "learning_rate": 0.00019999726238688483,
-      "loss": 46.0,
-      "step": 14636
-    },
-    {
-      "epoch": 2.357180240750433,
-      "grad_norm": 0.007008660584688187,
-      "learning_rate": 0.0001999972620125249,
-      "loss": 46.0,
-      "step": 14637
-    },
-    {
-      "epoch": 2.35734127782922,
-      "grad_norm": 0.002571654040366411,
-      "learning_rate": 0.00019999726163813937,
-      "loss": 46.0,
-      "step": 14638
-    },
-    {
-      "epoch": 2.3575023149080074,
-      "grad_norm": 0.00040334215736947954,
-      "learning_rate": 0.00019999726126372825,
-      "loss": 46.0,
-      "step": 14639
-    },
-    {
-      "epoch": 2.357663351986795,
-      "grad_norm": 0.007106542121618986,
-      "learning_rate": 0.00019999726088929155,
-      "loss": 46.0,
-      "step": 14640
-    },
-    {
-      "epoch": 2.3578243890655823,
-      "grad_norm": 0.0012591092381626368,
-      "learning_rate": 0.0001999972605148292,
-      "loss": 46.0,
-      "step": 14641
-    },
-    {
-      "epoch": 2.35798542614437,
-      "grad_norm": 0.002261930610984564,
-      "learning_rate": 0.0001999972601403413,
-      "loss": 46.0,
-      "step": 14642
-    },
-    {
-      "epoch": 2.3581464632231572,
-      "grad_norm": 0.002048380207270384,
-      "learning_rate": 0.00019999725976582782,
-      "loss": 46.0,
-      "step": 14643
-    },
-    {
-      "epoch": 2.3583075003019447,
-      "grad_norm": 0.00038474483881145716,
-      "learning_rate": 0.0001999972593912887,
-      "loss": 46.0,
-      "step": 14644
-    },
-    {
-      "epoch": 2.358468537380732,
-      "grad_norm": 0.003740274580195546,
-      "learning_rate": 0.00019999725901672402,
-      "loss": 46.0,
-      "step": 14645
-    },
-    {
-      "epoch": 2.358629574459519,
-      "grad_norm": 0.0005207539070397615,
-      "learning_rate": 0.00019999725864213374,
-      "loss": 46.0,
-      "step": 14646
-    },
-    {
-      "epoch": 2.3587906115383066,
-      "grad_norm": 0.003268243046477437,
-      "learning_rate": 0.00019999725826751785,
-      "loss": 46.0,
-      "step": 14647
-    },
-    {
-      "epoch": 2.358951648617094,
-      "grad_norm": 0.0018632604042068124,
-      "learning_rate": 0.00019999725789287637,
-      "loss": 46.0,
-      "step": 14648
-    },
-    {
-      "epoch": 2.3591126856958815,
-      "grad_norm": 0.0018150507239624858,
-      "learning_rate": 0.00019999725751820933,
-      "loss": 46.0,
-      "step": 14649
-    },
-    {
-      "epoch": 2.359273722774669,
-      "grad_norm": 0.0014653692487627268,
-      "learning_rate": 0.00019999725714351665,
-      "loss": 46.0,
-      "step": 14650
-    },
-    {
-      "epoch": 2.3594347598534564,
-      "grad_norm": 0.005365099757909775,
-      "learning_rate": 0.00019999725676879838,
-      "loss": 46.0,
-      "step": 14651
-    },
-    {
-      "epoch": 2.3595957969322434,
-      "grad_norm": 0.0016364929033443332,
-      "learning_rate": 0.00019999725639405455,
-      "loss": 46.0,
-      "step": 14652
-    },
-    {
-      "epoch": 2.359756834011031,
-      "grad_norm": 0.0016520697390660644,
-      "learning_rate": 0.0001999972560192851,
-      "loss": 46.0,
-      "step": 14653
-    },
-    {
-      "epoch": 2.3599178710898183,
-      "grad_norm": 0.002825340023264289,
-      "learning_rate": 0.00019999725564449008,
-      "loss": 46.0,
-      "step": 14654
-    },
-    {
-      "epoch": 2.3600789081686058,
-      "grad_norm": 0.003544642822816968,
-      "learning_rate": 0.00019999725526966943,
-      "loss": 46.0,
-      "step": 14655
-    },
-    {
-      "epoch": 2.3602399452473932,
-      "grad_norm": 0.0015102059114724398,
-      "learning_rate": 0.00019999725489482323,
-      "loss": 46.0,
-      "step": 14656
-    },
-    {
-      "epoch": 2.3604009823261807,
-      "grad_norm": 0.01047484204173088,
-      "learning_rate": 0.0001999972545199514,
-      "loss": 46.0,
-      "step": 14657
-    },
-    {
-      "epoch": 2.360562019404968,
-      "grad_norm": 0.0022438461892306805,
-      "learning_rate": 0.000199997254145054,
-      "loss": 46.0,
-      "step": 14658
-    },
-    {
-      "epoch": 2.3607230564837556,
-      "grad_norm": 0.0013933316804468632,
-      "learning_rate": 0.00019999725377013098,
-      "loss": 46.0,
-      "step": 14659
-    },
-    {
-      "epoch": 2.3608840935625426,
-      "grad_norm": 0.0035694336984306574,
-      "learning_rate": 0.00019999725339518238,
-      "loss": 46.0,
-      "step": 14660
-    },
-    {
-      "epoch": 2.36104513064133,
-      "grad_norm": 0.005351288244128227,
-      "learning_rate": 0.0001999972530202082,
-      "loss": 46.0,
-      "step": 14661
-    },
-    {
-      "epoch": 2.3612061677201175,
-      "grad_norm": 0.0035978180821985006,
-      "learning_rate": 0.0001999972526452084,
-      "loss": 46.0,
-      "step": 14662
-    },
-    {
-      "epoch": 2.361367204798905,
-      "grad_norm": 0.006602867506444454,
-      "learning_rate": 0.000199997252270183,
-      "loss": 46.0,
-      "step": 14663
-    },
-    {
-      "epoch": 2.3615282418776924,
-      "grad_norm": 0.006617933511734009,
-      "learning_rate": 0.00019999725189513202,
-      "loss": 46.0,
-      "step": 14664
-    },
-    {
-      "epoch": 2.36168927895648,
-      "grad_norm": 0.001885879086330533,
-      "learning_rate": 0.00019999725152005545,
-      "loss": 46.0,
-      "step": 14665
-    },
-    {
-      "epoch": 2.3618503160352673,
-      "grad_norm": 0.0033384154085069895,
-      "learning_rate": 0.00019999725114495332,
-      "loss": 46.0,
-      "step": 14666
-    },
-    {
-      "epoch": 2.3620113531140543,
-      "grad_norm": 0.0019791023805737495,
-      "learning_rate": 0.00019999725076982555,
-      "loss": 46.0,
-      "step": 14667
-    },
-    {
-      "epoch": 2.3621723901928418,
-      "grad_norm": 0.005918820854276419,
-      "learning_rate": 0.0001999972503946722,
-      "loss": 46.0,
-      "step": 14668
-    },
-    {
-      "epoch": 2.362333427271629,
-      "grad_norm": 0.006328626070171595,
-      "learning_rate": 0.00019999725001949324,
-      "loss": 46.0,
-      "step": 14669
-    },
-    {
-      "epoch": 2.3624944643504167,
-      "grad_norm": 0.004716383758932352,
-      "learning_rate": 0.0001999972496442887,
-      "loss": 46.0,
-      "step": 14670
-    },
-    {
-      "epoch": 2.362655501429204,
-      "grad_norm": 0.007791764102876186,
-      "learning_rate": 0.00019999724926905854,
-      "loss": 46.0,
-      "step": 14671
-    },
-    {
-      "epoch": 2.3628165385079916,
-      "grad_norm": 0.0012216089526191354,
-      "learning_rate": 0.00019999724889380283,
-      "loss": 46.0,
-      "step": 14672
-    },
-    {
-      "epoch": 2.362977575586779,
-      "grad_norm": 0.010274586267769337,
-      "learning_rate": 0.00019999724851852148,
-      "loss": 46.0,
-      "step": 14673
-    },
-    {
-      "epoch": 2.3631386126655665,
-      "grad_norm": 0.002135322894901037,
-      "learning_rate": 0.0001999972481432146,
-      "loss": 46.0,
-      "step": 14674
-    },
-    {
-      "epoch": 2.3632996497443535,
-      "grad_norm": 0.001198525307700038,
-      "learning_rate": 0.00019999724776788207,
-      "loss": 46.0,
-      "step": 14675
-    },
-    {
-      "epoch": 2.363460686823141,
-      "grad_norm": 0.0005366932600736618,
-      "learning_rate": 0.00019999724739252396,
-      "loss": 46.0,
-      "step": 14676
-    },
-    {
-      "epoch": 2.3636217239019284,
-      "grad_norm": 0.004019653890281916,
-      "learning_rate": 0.00019999724701714024,
-      "loss": 46.0,
-      "step": 14677
-    },
-    {
-      "epoch": 2.363782760980716,
-      "grad_norm": 0.002869992284104228,
-      "learning_rate": 0.00019999724664173093,
-      "loss": 46.0,
-      "step": 14678
-    },
-    {
-      "epoch": 2.3639437980595033,
-      "grad_norm": 0.003484451211988926,
-      "learning_rate": 0.00019999724626629605,
-      "loss": 46.0,
-      "step": 14679
-    },
-    {
-      "epoch": 2.3641048351382907,
-      "grad_norm": 0.0034715859219431877,
-      "learning_rate": 0.0001999972458908356,
-      "loss": 46.0,
-      "step": 14680
-    },
-    {
-      "epoch": 2.3642658722170777,
-      "grad_norm": 0.0035185804590582848,
-      "learning_rate": 0.0001999972455153495,
-      "loss": 46.0,
-      "step": 14681
-    },
-    {
-      "epoch": 2.364426909295865,
-      "grad_norm": 0.0008727594395168126,
-      "learning_rate": 0.00019999724513983783,
-      "loss": 46.0,
-      "step": 14682
-    },
-    {
-      "epoch": 2.3645879463746526,
-      "grad_norm": 0.0005920006078667939,
-      "learning_rate": 0.00019999724476430056,
-      "loss": 46.0,
-      "step": 14683
-    },
-    {
-      "epoch": 2.36474898345344,
-      "grad_norm": 0.000762270123232156,
-      "learning_rate": 0.0001999972443887377,
-      "loss": 46.0,
-      "step": 14684
-    },
-    {
-      "epoch": 2.3649100205322275,
-      "grad_norm": 0.004592764191329479,
-      "learning_rate": 0.00019999724401314922,
-      "loss": 46.0,
-      "step": 14685
-    },
-    {
-      "epoch": 2.365071057611015,
-      "grad_norm": 0.005814696196466684,
-      "learning_rate": 0.0001999972436375352,
-      "loss": 46.0,
-      "step": 14686
-    },
-    {
-      "epoch": 2.3652320946898024,
-      "grad_norm": 0.01590869575738907,
-      "learning_rate": 0.00019999724326189556,
-      "loss": 46.0,
-      "step": 14687
-    },
-    {
-      "epoch": 2.36539313176859,
-      "grad_norm": 0.0025805397890508175,
-      "learning_rate": 0.00019999724288623032,
-      "loss": 46.0,
-      "step": 14688
-    },
-    {
-      "epoch": 2.3655541688473773,
-      "grad_norm": 0.0008912792545743287,
-      "learning_rate": 0.00019999724251053947,
-      "loss": 46.0,
-      "step": 14689
-    },
-    {
-      "epoch": 2.3657152059261644,
-      "grad_norm": 0.0018187996465712786,
-      "learning_rate": 0.00019999724213482306,
-      "loss": 46.0,
-      "step": 14690
-    },
-    {
-      "epoch": 2.365876243004952,
-      "grad_norm": 0.007388916332274675,
-      "learning_rate": 0.00019999724175908103,
-      "loss": 46.0,
-      "step": 14691
-    },
-    {
-      "epoch": 2.3660372800837393,
-      "grad_norm": 0.0017756903544068336,
-      "learning_rate": 0.00019999724138331342,
-      "loss": 46.0,
-      "step": 14692
-    },
-    {
-      "epoch": 2.3661983171625267,
-      "grad_norm": 0.0015834859805181623,
-      "learning_rate": 0.0001999972410075202,
-      "loss": 46.0,
-      "step": 14693
-    },
-    {
-      "epoch": 2.366359354241314,
-      "grad_norm": 0.00608128122985363,
-      "learning_rate": 0.0001999972406317014,
-      "loss": 46.0,
-      "step": 14694
-    },
-    {
-      "epoch": 2.3665203913201016,
-      "grad_norm": 0.008327258750796318,
-      "learning_rate": 0.00019999724025585703,
-      "loss": 46.0,
-      "step": 14695
-    },
-    {
-      "epoch": 2.3666814283988886,
-      "grad_norm": 0.0007105885888449848,
-      "learning_rate": 0.000199997239879987,
-      "loss": 46.0,
-      "step": 14696
-    },
-    {
-      "epoch": 2.366842465477676,
-      "grad_norm": 0.0014984381850808859,
-      "learning_rate": 0.00019999723950409143,
-      "loss": 46.0,
-      "step": 14697
-    },
-    {
-      "epoch": 2.3670035025564635,
-      "grad_norm": 0.0008859752560965717,
-      "learning_rate": 0.00019999723912817024,
-      "loss": 46.0,
-      "step": 14698
-    },
-    {
-      "epoch": 2.367164539635251,
-      "grad_norm": 0.002013867488130927,
-      "learning_rate": 0.0001999972387522235,
-      "loss": 46.0,
-      "step": 14699
-    },
-    {
-      "epoch": 2.3673255767140384,
-      "grad_norm": 0.0023132688365876675,
-      "learning_rate": 0.0001999972383762511,
-      "loss": 46.0,
-      "step": 14700
-    },
-    {
-      "epoch": 2.367486613792826,
-      "grad_norm": 0.0020256282296031713,
-      "learning_rate": 0.00019999723800025315,
-      "loss": 46.0,
-      "step": 14701
-    },
-    {
-      "epoch": 2.3676476508716133,
-      "grad_norm": 0.001999534899368882,
-      "learning_rate": 0.00019999723762422958,
-      "loss": 46.0,
-      "step": 14702
-    },
-    {
-      "epoch": 2.367808687950401,
-      "grad_norm": 0.00644769286736846,
-      "learning_rate": 0.00019999723724818043,
-      "loss": 46.0,
-      "step": 14703
-    },
-    {
-      "epoch": 2.367969725029188,
-      "grad_norm": 0.00218023918569088,
-      "learning_rate": 0.0001999972368721057,
-      "loss": 46.0,
-      "step": 14704
-    },
-    {
-      "epoch": 2.3681307621079752,
-      "grad_norm": 0.006280168425291777,
-      "learning_rate": 0.00019999723649600536,
-      "loss": 46.0,
-      "step": 14705
-    },
-    {
-      "epoch": 2.3682917991867627,
-      "grad_norm": 0.0030314770992845297,
-      "learning_rate": 0.00019999723611987942,
-      "loss": 46.0,
-      "step": 14706
-    },
-    {
-      "epoch": 2.36845283626555,
-      "grad_norm": 0.001299206051044166,
-      "learning_rate": 0.00019999723574372791,
-      "loss": 46.0,
-      "step": 14707
-    },
-    {
-      "epoch": 2.3686138733443376,
-      "grad_norm": 0.002936696633696556,
-      "learning_rate": 0.00019999723536755077,
-      "loss": 46.0,
-      "step": 14708
-    },
-    {
-      "epoch": 2.368774910423125,
-      "grad_norm": 0.0010150355519726872,
-      "learning_rate": 0.00019999723499134807,
-      "loss": 46.0,
-      "step": 14709
-    },
-    {
-      "epoch": 2.3689359475019125,
-      "grad_norm": 0.0006543753552250564,
-      "learning_rate": 0.00019999723461511975,
-      "loss": 46.0,
-      "step": 14710
-    },
-    {
-      "epoch": 2.3690969845806995,
-      "grad_norm": 0.0041079167276620865,
-      "learning_rate": 0.00019999723423886584,
-      "loss": 46.0,
-      "step": 14711
-    },
-    {
-      "epoch": 2.369258021659487,
-      "grad_norm": 0.0016417294973507524,
-      "learning_rate": 0.00019999723386258635,
-      "loss": 46.0,
-      "step": 14712
-    },
-    {
-      "epoch": 2.3694190587382744,
-      "grad_norm": 0.001662685303017497,
-      "learning_rate": 0.00019999723348628124,
-      "loss": 46.0,
-      "step": 14713
-    },
-    {
-      "epoch": 2.369580095817062,
-      "grad_norm": 0.0009432067163288593,
-      "learning_rate": 0.00019999723310995058,
-      "loss": 46.0,
-      "step": 14714
-    },
-    {
-      "epoch": 2.3697411328958493,
-      "grad_norm": 0.003060807939618826,
-      "learning_rate": 0.0001999972327335943,
-      "loss": 46.0,
-      "step": 14715
-    },
-    {
-      "epoch": 2.3699021699746368,
-      "grad_norm": 0.002123053651303053,
-      "learning_rate": 0.00019999723235721243,
-      "loss": 46.0,
-      "step": 14716
-    },
-    {
-      "epoch": 2.370063207053424,
-      "grad_norm": 0.007227302063256502,
-      "learning_rate": 0.00019999723198080494,
-      "loss": 46.0,
-      "step": 14717
-    },
-    {
-      "epoch": 2.3702242441322117,
-      "grad_norm": 0.005050660111010075,
-      "learning_rate": 0.0001999972316043719,
-      "loss": 46.0,
-      "step": 14718
-    },
-    {
-      "epoch": 2.3703852812109987,
-      "grad_norm": 0.0077496604062616825,
-      "learning_rate": 0.00019999723122791322,
-      "loss": 46.0,
-      "step": 14719
-    },
-    {
-      "epoch": 2.370546318289786,
-      "grad_norm": 0.004484896548092365,
-      "learning_rate": 0.00019999723085142897,
-      "loss": 46.0,
-      "step": 14720
-    },
-    {
-      "epoch": 2.3707073553685736,
-      "grad_norm": 0.003823069855570793,
-      "learning_rate": 0.00019999723047491914,
-      "loss": 46.0,
-      "step": 14721
-    },
-    {
-      "epoch": 2.370868392447361,
-      "grad_norm": 0.0010556193301454186,
-      "learning_rate": 0.0001999972300983837,
-      "loss": 46.0,
-      "step": 14722
-    },
-    {
-      "epoch": 2.3710294295261485,
-      "grad_norm": 0.000592644268181175,
-      "learning_rate": 0.00019999722972182266,
-      "loss": 46.0,
-      "step": 14723
-    },
-    {
-      "epoch": 2.371190466604936,
-      "grad_norm": 0.009649808518588543,
-      "learning_rate": 0.00019999722934523604,
-      "loss": 46.0,
-      "step": 14724
-    },
-    {
-      "epoch": 2.371351503683723,
-      "grad_norm": 0.003973050974309444,
-      "learning_rate": 0.0001999972289686238,
-      "loss": 46.0,
-      "step": 14725
-    },
-    {
-      "epoch": 2.3715125407625104,
-      "grad_norm": 0.0018054497195407748,
-      "learning_rate": 0.000199997228591986,
-      "loss": 46.0,
-      "step": 14726
-    },
-    {
-      "epoch": 2.371673577841298,
-      "grad_norm": 0.001413486199453473,
-      "learning_rate": 0.00019999722821532258,
-      "loss": 46.0,
-      "step": 14727
-    },
-    {
-      "epoch": 2.3718346149200853,
-      "grad_norm": 0.0015482193557545543,
-      "learning_rate": 0.00019999722783863358,
-      "loss": 46.0,
-      "step": 14728
-    },
-    {
-      "epoch": 2.3719956519988727,
-      "grad_norm": 0.0022801414597779512,
-      "learning_rate": 0.00019999722746191897,
-      "loss": 46.0,
-      "step": 14729
-    },
-    {
-      "epoch": 2.37215668907766,
-      "grad_norm": 0.0022812786046415567,
-      "learning_rate": 0.00019999722708517878,
-      "loss": 46.0,
-      "step": 14730
-    },
-    {
-      "epoch": 2.3723177261564476,
-      "grad_norm": 0.0009693917236290872,
-      "learning_rate": 0.00019999722670841302,
-      "loss": 46.0,
-      "step": 14731
-    },
-    {
-      "epoch": 2.372478763235235,
-      "grad_norm": 0.0022720452398061752,
-      "learning_rate": 0.00019999722633162162,
-      "loss": 46.0,
-      "step": 14732
-    },
-    {
-      "epoch": 2.3726398003140226,
-      "grad_norm": 0.0035725245252251625,
-      "learning_rate": 0.00019999722595480463,
-      "loss": 46.0,
-      "step": 14733
-    },
-    {
-      "epoch": 2.3728008373928096,
-      "grad_norm": 0.0024852852802723646,
-      "learning_rate": 0.0001999972255779621,
-      "loss": 46.0,
-      "step": 14734
-    },
-    {
-      "epoch": 2.372961874471597,
-      "grad_norm": 0.0014489275636151433,
-      "learning_rate": 0.0001999972252010939,
-      "loss": 46.0,
-      "step": 14735
-    },
-    {
-      "epoch": 2.3731229115503845,
-      "grad_norm": 0.003032973036170006,
-      "learning_rate": 0.00019999722482420013,
-      "loss": 46.0,
-      "step": 14736
-    },
-    {
-      "epoch": 2.373283948629172,
-      "grad_norm": 0.0027989703230559826,
-      "learning_rate": 0.0001999972244472808,
-      "loss": 46.0,
-      "step": 14737
-    },
-    {
-      "epoch": 2.3734449857079594,
-      "grad_norm": 0.0029548613820225,
-      "learning_rate": 0.00019999722407033584,
-      "loss": 46.0,
-      "step": 14738
-    },
-    {
-      "epoch": 2.373606022786747,
-      "grad_norm": 0.002398983808234334,
-      "learning_rate": 0.0001999972236933653,
-      "loss": 46.0,
-      "step": 14739
-    },
-    {
-      "epoch": 2.373767059865534,
-      "grad_norm": 0.003854044945910573,
-      "learning_rate": 0.00019999722331636916,
-      "loss": 46.0,
-      "step": 14740
-    },
-    {
-      "epoch": 2.3739280969443213,
-      "grad_norm": 0.001955259358510375,
-      "learning_rate": 0.00019999722293934747,
-      "loss": 46.0,
-      "step": 14741
-    },
-    {
-      "epoch": 2.3740891340231087,
-      "grad_norm": 0.002410275163128972,
-      "learning_rate": 0.00019999722256230012,
-      "loss": 46.0,
-      "step": 14742
-    },
-    {
-      "epoch": 2.374250171101896,
-      "grad_norm": 0.0031933600548654795,
-      "learning_rate": 0.0001999972221852272,
-      "loss": 46.0,
-      "step": 14743
-    },
-    {
-      "epoch": 2.3744112081806836,
-      "grad_norm": 0.005795541685074568,
-      "learning_rate": 0.0001999972218081287,
-      "loss": 46.0,
-      "step": 14744
-    },
-    {
-      "epoch": 2.374572245259471,
-      "grad_norm": 0.007466630078852177,
-      "learning_rate": 0.0001999972214310046,
-      "loss": 46.0,
-      "step": 14745
-    },
-    {
-      "epoch": 2.3747332823382585,
-      "grad_norm": 0.003981498070061207,
-      "learning_rate": 0.0001999972210538549,
-      "loss": 46.0,
-      "step": 14746
-    },
-    {
-      "epoch": 2.374894319417046,
-      "grad_norm": 0.0017899404047057033,
-      "learning_rate": 0.00019999722067667958,
-      "loss": 46.0,
-      "step": 14747
-    },
-    {
-      "epoch": 2.375055356495833,
-      "grad_norm": 0.007729220669716597,
-      "learning_rate": 0.00019999722029947868,
-      "loss": 46.0,
-      "step": 14748
-    },
-    {
-      "epoch": 2.3752163935746204,
-      "grad_norm": 0.0016655372455716133,
-      "learning_rate": 0.00019999721992225222,
-      "loss": 46.0,
-      "step": 14749
-    },
-    {
-      "epoch": 2.375377430653408,
-      "grad_norm": 0.0016331017250195146,
-      "learning_rate": 0.00019999721954500014,
-      "loss": 46.0,
-      "step": 14750
-    },
-    {
-      "epoch": 2.3755384677321953,
-      "grad_norm": 0.00180652248673141,
-      "learning_rate": 0.0001999972191677225,
-      "loss": 46.0,
-      "step": 14751
-    },
-    {
-      "epoch": 2.375699504810983,
-      "grad_norm": 0.0006934988196007907,
-      "learning_rate": 0.00019999721879041923,
-      "loss": 46.0,
-      "step": 14752
-    },
-    {
-      "epoch": 2.3758605418897702,
-      "grad_norm": 0.003972357604652643,
-      "learning_rate": 0.00019999721841309036,
-      "loss": 46.0,
-      "step": 14753
-    },
-    {
-      "epoch": 2.3760215789685577,
-      "grad_norm": 0.004908293019980192,
-      "learning_rate": 0.0001999972180357359,
-      "loss": 46.0,
-      "step": 14754
-    },
-    {
-      "epoch": 2.3761826160473447,
-      "grad_norm": 0.01182826142758131,
-      "learning_rate": 0.00019999721765835587,
-      "loss": 46.0,
-      "step": 14755
-    },
-    {
-      "epoch": 2.376343653126132,
-      "grad_norm": 0.00035638068220578134,
-      "learning_rate": 0.00019999721728095022,
-      "loss": 46.0,
-      "step": 14756
-    },
-    {
-      "epoch": 2.3765046902049196,
-      "grad_norm": 0.0019747004844248295,
-      "learning_rate": 0.00019999721690351898,
-      "loss": 46.0,
-      "step": 14757
-    },
-    {
-      "epoch": 2.376665727283707,
-      "grad_norm": 0.0015068651409819722,
-      "learning_rate": 0.00019999721652606215,
-      "loss": 46.0,
-      "step": 14758
-    },
-    {
-      "epoch": 2.3768267643624945,
-      "grad_norm": 0.000996322720311582,
-      "learning_rate": 0.00019999721614857974,
-      "loss": 46.0,
-      "step": 14759
-    },
-    {
-      "epoch": 2.376987801441282,
-      "grad_norm": 0.004778548143804073,
-      "learning_rate": 0.0001999972157710717,
-      "loss": 46.0,
-      "step": 14760
-    },
-    {
-      "epoch": 2.3771488385200694,
-      "grad_norm": 0.0013410389656201005,
-      "learning_rate": 0.00019999721539353812,
-      "loss": 46.0,
-      "step": 14761
-    },
-    {
-      "epoch": 2.377309875598857,
-      "grad_norm": 0.0065715196542441845,
-      "learning_rate": 0.0001999972150159789,
-      "loss": 46.0,
-      "step": 14762
-    },
-    {
-      "epoch": 2.377470912677644,
-      "grad_norm": 0.0010227446909993887,
-      "learning_rate": 0.0001999972146383941,
-      "loss": 46.0,
-      "step": 14763
-    },
-    {
-      "epoch": 2.3776319497564313,
-      "grad_norm": 0.001720933592878282,
-      "learning_rate": 0.0001999972142607837,
-      "loss": 46.0,
-      "step": 14764
-    },
-    {
-      "epoch": 2.377792986835219,
-      "grad_norm": 0.0023274989798665047,
-      "learning_rate": 0.00019999721388314773,
-      "loss": 46.0,
-      "step": 14765
-    },
-    {
-      "epoch": 2.3779540239140062,
-      "grad_norm": 0.002022327622398734,
-      "learning_rate": 0.00019999721350548612,
-      "loss": 46.0,
-      "step": 14766
-    },
-    {
-      "epoch": 2.3781150609927937,
-      "grad_norm": 0.0007186635630205274,
-      "learning_rate": 0.00019999721312779895,
-      "loss": 46.0,
-      "step": 14767
-    },
-    {
-      "epoch": 2.378276098071581,
-      "grad_norm": 0.0006771271000616252,
-      "learning_rate": 0.00019999721275008617,
-      "loss": 46.0,
-      "step": 14768
-    },
-    {
-      "epoch": 2.378437135150368,
-      "grad_norm": 0.007208125665783882,
-      "learning_rate": 0.00019999721237234783,
-      "loss": 46.0,
-      "step": 14769
-    },
-    {
-      "epoch": 2.3785981722291556,
-      "grad_norm": 0.001271755900233984,
-      "learning_rate": 0.00019999721199458385,
-      "loss": 46.0,
-      "step": 14770
-    },
-    {
-      "epoch": 2.378759209307943,
-      "grad_norm": 0.003842888167127967,
-      "learning_rate": 0.0001999972116167943,
-      "loss": 46.0,
-      "step": 14771
-    },
-    {
-      "epoch": 2.3789202463867305,
-      "grad_norm": 0.0008683003834448755,
-      "learning_rate": 0.00019999721123897915,
-      "loss": 46.0,
-      "step": 14772
-    },
-    {
-      "epoch": 2.379081283465518,
-      "grad_norm": 0.0010391527321189642,
-      "learning_rate": 0.0001999972108611384,
-      "loss": 46.0,
-      "step": 14773
-    },
-    {
-      "epoch": 2.3792423205443054,
-      "grad_norm": 0.0033191770780831575,
-      "learning_rate": 0.00019999721048327208,
-      "loss": 46.0,
-      "step": 14774
-    },
-    {
-      "epoch": 2.379403357623093,
-      "grad_norm": 0.0008084566215984523,
-      "learning_rate": 0.00019999721010538016,
-      "loss": 46.0,
-      "step": 14775
-    },
-    {
-      "epoch": 2.3795643947018803,
-      "grad_norm": 0.0020298969466239214,
-      "learning_rate": 0.00019999720972746263,
-      "loss": 46.0,
-      "step": 14776
-    },
-    {
-      "epoch": 2.3797254317806673,
-      "grad_norm": 0.0011275119613856077,
-      "learning_rate": 0.0001999972093495195,
-      "loss": 46.0,
-      "step": 14777
-    },
-    {
-      "epoch": 2.3798864688594548,
-      "grad_norm": 0.00478333467617631,
-      "learning_rate": 0.0001999972089715508,
-      "loss": 46.0,
-      "step": 14778
-    },
-    {
-      "epoch": 2.380047505938242,
-      "grad_norm": 0.002023993758484721,
-      "learning_rate": 0.00019999720859355648,
-      "loss": 46.0,
-      "step": 14779
-    },
-    {
-      "epoch": 2.3802085430170297,
-      "grad_norm": 0.008766070939600468,
-      "learning_rate": 0.00019999720821553657,
-      "loss": 46.0,
-      "step": 14780
-    },
-    {
-      "epoch": 2.380369580095817,
-      "grad_norm": 0.00048753657029010355,
-      "learning_rate": 0.0001999972078374911,
-      "loss": 46.0,
-      "step": 14781
-    },
-    {
-      "epoch": 2.3805306171746046,
-      "grad_norm": 0.016918152570724487,
-      "learning_rate": 0.00019999720745942,
-      "loss": 46.0,
-      "step": 14782
-    },
-    {
-      "epoch": 2.380691654253392,
-      "grad_norm": 0.009116408415138721,
-      "learning_rate": 0.00019999720708132332,
-      "loss": 46.0,
-      "step": 14783
-    },
-    {
-      "epoch": 2.380852691332179,
-      "grad_norm": 0.0025388484355062246,
-      "learning_rate": 0.00019999720670320104,
-      "loss": 46.0,
-      "step": 14784
-    },
-    {
-      "epoch": 2.3810137284109665,
-      "grad_norm": 0.0019006110960617661,
-      "learning_rate": 0.00019999720632505317,
-      "loss": 46.0,
-      "step": 14785
-    },
-    {
-      "epoch": 2.381174765489754,
-      "grad_norm": 0.002911211922764778,
-      "learning_rate": 0.0001999972059468797,
-      "loss": 46.0,
-      "step": 14786
-    },
-    {
-      "epoch": 2.3813358025685414,
-      "grad_norm": 0.001458571059629321,
-      "learning_rate": 0.00019999720556868064,
-      "loss": 46.0,
-      "step": 14787
-    },
-    {
-      "epoch": 2.381496839647329,
-      "grad_norm": 0.0031664129346609116,
-      "learning_rate": 0.000199997205190456,
-      "loss": 46.0,
-      "step": 14788
-    },
-    {
-      "epoch": 2.3816578767261163,
-      "grad_norm": 0.002005769405514002,
-      "learning_rate": 0.00019999720481220573,
-      "loss": 46.0,
-      "step": 14789
-    },
-    {
-      "epoch": 2.3818189138049037,
-      "grad_norm": 0.0038936263881623745,
-      "learning_rate": 0.0001999972044339299,
-      "loss": 46.0,
-      "step": 14790
-    },
-    {
-      "epoch": 2.381979950883691,
-      "grad_norm": 0.0027838812675327063,
-      "learning_rate": 0.00019999720405562848,
-      "loss": 46.0,
-      "step": 14791
-    },
-    {
-      "epoch": 2.382140987962478,
-      "grad_norm": 0.0007916048052720726,
-      "learning_rate": 0.00019999720367730144,
-      "loss": 46.0,
-      "step": 14792
-    },
-    {
-      "epoch": 2.3823020250412656,
-      "grad_norm": 0.0008116143872030079,
-      "learning_rate": 0.00019999720329894882,
-      "loss": 46.0,
-      "step": 14793
-    },
-    {
-      "epoch": 2.382463062120053,
-      "grad_norm": 0.0021078018471598625,
-      "learning_rate": 0.0001999972029205706,
-      "loss": 46.0,
-      "step": 14794
-    },
-    {
-      "epoch": 2.3826240991988406,
-      "grad_norm": 0.0019509702688083053,
-      "learning_rate": 0.00019999720254216676,
-      "loss": 46.0,
-      "step": 14795
-    },
-    {
-      "epoch": 2.382785136277628,
-      "grad_norm": 0.01486461702734232,
-      "learning_rate": 0.00019999720216373737,
-      "loss": 46.0,
-      "step": 14796
-    },
-    {
-      "epoch": 2.3829461733564155,
-      "grad_norm": 0.004105749074369669,
-      "learning_rate": 0.00019999720178528237,
-      "loss": 46.0,
-      "step": 14797
-    },
-    {
-      "epoch": 2.3831072104352025,
-      "grad_norm": 0.0022464694920927286,
-      "learning_rate": 0.00019999720140680176,
-      "loss": 46.0,
-      "step": 14798
-    },
-    {
-      "epoch": 2.38326824751399,
-      "grad_norm": 0.014298107475042343,
-      "learning_rate": 0.0001999972010282956,
-      "loss": 46.0,
-      "step": 14799
-    },
-    {
-      "epoch": 2.3834292845927774,
-      "grad_norm": 0.0015294221229851246,
-      "learning_rate": 0.00019999720064976383,
-      "loss": 46.0,
-      "step": 14800
-    },
-    {
-      "epoch": 2.383590321671565,
-      "grad_norm": 0.011498850770294666,
-      "learning_rate": 0.0001999972002712064,
-      "loss": 46.0,
-      "step": 14801
-    },
-    {
-      "epoch": 2.3837513587503523,
-      "grad_norm": 0.01088939979672432,
-      "learning_rate": 0.00019999719989262347,
-      "loss": 46.0,
-      "step": 14802
-    },
-    {
-      "epoch": 2.3839123958291397,
-      "grad_norm": 0.006420196499675512,
-      "learning_rate": 0.0001999971995140149,
-      "loss": 46.0,
-      "step": 14803
-    },
-    {
-      "epoch": 2.384073432907927,
-      "grad_norm": 0.002970593748614192,
-      "learning_rate": 0.0001999971991353807,
-      "loss": 46.0,
-      "step": 14804
-    },
-    {
-      "epoch": 2.3842344699867146,
-      "grad_norm": 0.0008491531480103731,
-      "learning_rate": 0.00019999719875672098,
-      "loss": 46.0,
-      "step": 14805
-    },
-    {
-      "epoch": 2.384395507065502,
-      "grad_norm": 0.0015757104847580194,
-      "learning_rate": 0.00019999719837803564,
-      "loss": 46.0,
-      "step": 14806
-    },
-    {
-      "epoch": 2.384556544144289,
-      "grad_norm": 0.0011876648059114814,
-      "learning_rate": 0.0001999971979993247,
-      "loss": 46.0,
-      "step": 14807
-    },
-    {
-      "epoch": 2.3847175812230765,
-      "grad_norm": 0.0011319088516756892,
-      "learning_rate": 0.00019999719762058815,
-      "loss": 46.0,
-      "step": 14808
-    },
-    {
-      "epoch": 2.384878618301864,
-      "grad_norm": 0.0015093715628609061,
-      "learning_rate": 0.00019999719724182602,
-      "loss": 46.0,
-      "step": 14809
-    },
-    {
-      "epoch": 2.3850396553806514,
-      "grad_norm": 0.0016423582565039396,
-      "learning_rate": 0.00019999719686303828,
-      "loss": 46.0,
-      "step": 14810
-    },
-    {
-      "epoch": 2.385200692459439,
-      "grad_norm": 0.0007229059119708836,
-      "learning_rate": 0.00019999719648422496,
-      "loss": 46.0,
-      "step": 14811
-    },
-    {
-      "epoch": 2.3853617295382263,
-      "grad_norm": 0.0011481448309496045,
-      "learning_rate": 0.00019999719610538604,
-      "loss": 46.0,
-      "step": 14812
-    },
-    {
-      "epoch": 2.3855227666170133,
-      "grad_norm": 0.006847640499472618,
-      "learning_rate": 0.00019999719572652157,
-      "loss": 46.0,
-      "step": 14813
-    },
-    {
-      "epoch": 2.385683803695801,
-      "grad_norm": 0.0017043565167114139,
-      "learning_rate": 0.00019999719534763143,
-      "loss": 46.0,
-      "step": 14814
-    },
-    {
-      "epoch": 2.3858448407745882,
-      "grad_norm": 0.002039613900706172,
-      "learning_rate": 0.00019999719496871575,
-      "loss": 46.0,
-      "step": 14815
-    },
-    {
-      "epoch": 2.3860058778533757,
-      "grad_norm": 0.002016628859564662,
-      "learning_rate": 0.00019999719458977446,
-      "loss": 46.0,
-      "step": 14816
-    },
-    {
-      "epoch": 2.386166914932163,
-      "grad_norm": 0.0033311331644654274,
-      "learning_rate": 0.00019999719421080755,
-      "loss": 46.0,
-      "step": 14817
-    },
-    {
-      "epoch": 2.3863279520109506,
-      "grad_norm": 0.0027217241004109383,
-      "learning_rate": 0.0001999971938318151,
-      "loss": 46.0,
-      "step": 14818
-    },
-    {
-      "epoch": 2.386488989089738,
-      "grad_norm": 0.0003538668970577419,
-      "learning_rate": 0.000199997193452797,
-      "loss": 46.0,
-      "step": 14819
-    },
-    {
-      "epoch": 2.3866500261685255,
-      "grad_norm": 0.007403582334518433,
-      "learning_rate": 0.00019999719307375334,
-      "loss": 46.0,
-      "step": 14820
-    },
-    {
-      "epoch": 2.3868110632473125,
-      "grad_norm": 0.0077627371065318584,
-      "learning_rate": 0.0001999971926946841,
-      "loss": 46.0,
-      "step": 14821
-    },
-    {
-      "epoch": 2.3869721003261,
-      "grad_norm": 0.006375632248818874,
-      "learning_rate": 0.00019999719231558922,
-      "loss": 46.0,
-      "step": 14822
-    },
-    {
-      "epoch": 2.3871331374048874,
-      "grad_norm": 0.009840652346611023,
-      "learning_rate": 0.0001999971919364688,
-      "loss": 46.0,
-      "step": 14823
-    },
-    {
-      "epoch": 2.387294174483675,
-      "grad_norm": 0.002215093933045864,
-      "learning_rate": 0.00019999719155732275,
-      "loss": 46.0,
-      "step": 14824
-    },
-    {
-      "epoch": 2.3874552115624623,
-      "grad_norm": 0.0007720871944911778,
-      "learning_rate": 0.00019999719117815113,
-      "loss": 46.0,
-      "step": 14825
-    },
-    {
-      "epoch": 2.3876162486412498,
-      "grad_norm": 0.00482658576220274,
-      "learning_rate": 0.00019999719079895388,
-      "loss": 46.0,
-      "step": 14826
-    },
-    {
-      "epoch": 2.3877772857200372,
-      "grad_norm": 0.001992464531213045,
-      "learning_rate": 0.00019999719041973105,
-      "loss": 46.0,
-      "step": 14827
-    },
-    {
-      "epoch": 2.3879383227988242,
-      "grad_norm": 0.002385077066719532,
-      "learning_rate": 0.00019999719004048263,
-      "loss": 46.0,
-      "step": 14828
-    },
-    {
-      "epoch": 2.3880993598776117,
-      "grad_norm": 0.005993297789245844,
-      "learning_rate": 0.00019999718966120863,
-      "loss": 46.0,
-      "step": 14829
-    },
-    {
-      "epoch": 2.388260396956399,
-      "grad_norm": 0.0031910741236060858,
-      "learning_rate": 0.000199997189281909,
-      "loss": 46.0,
-      "step": 14830
-    },
-    {
-      "epoch": 2.3884214340351866,
-      "grad_norm": 0.0024330129381269217,
-      "learning_rate": 0.00019999718890258378,
-      "loss": 46.0,
-      "step": 14831
-    },
-    {
-      "epoch": 2.388582471113974,
-      "grad_norm": 0.0025851314421743155,
-      "learning_rate": 0.000199997188523233,
-      "loss": 46.0,
-      "step": 14832
-    },
-    {
-      "epoch": 2.3887435081927615,
-      "grad_norm": 0.001485941931605339,
-      "learning_rate": 0.00019999718814385663,
-      "loss": 46.0,
-      "step": 14833
-    },
-    {
-      "epoch": 2.388904545271549,
-      "grad_norm": 0.004651450552046299,
-      "learning_rate": 0.00019999718776445464,
-      "loss": 46.0,
-      "step": 14834
-    },
-    {
-      "epoch": 2.3890655823503364,
-      "grad_norm": 0.00271437456831336,
-      "learning_rate": 0.00019999718738502705,
-      "loss": 46.0,
-      "step": 14835
-    },
-    {
-      "epoch": 2.3892266194291234,
-      "grad_norm": 0.0004945089458487928,
-      "learning_rate": 0.00019999718700557388,
-      "loss": 46.0,
-      "step": 14836
-    },
-    {
-      "epoch": 2.389387656507911,
-      "grad_norm": 0.001572782639414072,
-      "learning_rate": 0.0001999971866260951,
-      "loss": 46.0,
-      "step": 14837
-    },
-    {
-      "epoch": 2.3895486935866983,
-      "grad_norm": 0.0014373004669323564,
-      "learning_rate": 0.00019999718624659076,
-      "loss": 46.0,
-      "step": 14838
-    },
-    {
-      "epoch": 2.3897097306654858,
-      "grad_norm": 0.0012333723716437817,
-      "learning_rate": 0.0001999971858670608,
-      "loss": 46.0,
-      "step": 14839
-    },
-    {
-      "epoch": 2.389870767744273,
-      "grad_norm": 0.006921854335814714,
-      "learning_rate": 0.00019999718548750525,
-      "loss": 46.0,
-      "step": 14840
-    },
-    {
-      "epoch": 2.3900318048230607,
-      "grad_norm": 0.0035812475252896547,
-      "learning_rate": 0.0001999971851079241,
-      "loss": 46.0,
-      "step": 14841
-    },
-    {
-      "epoch": 2.3901928419018477,
-      "grad_norm": 0.0031608252320438623,
-      "learning_rate": 0.00019999718472831738,
-      "loss": 46.0,
-      "step": 14842
-    },
-    {
-      "epoch": 2.390353878980635,
-      "grad_norm": 0.0011296499287709594,
-      "learning_rate": 0.00019999718434868502,
-      "loss": 46.0,
-      "step": 14843
-    },
-    {
-      "epoch": 2.3905149160594226,
-      "grad_norm": 0.003385860240086913,
-      "learning_rate": 0.00019999718396902712,
-      "loss": 46.0,
-      "step": 14844
-    },
-    {
-      "epoch": 2.39067595313821,
-      "grad_norm": 0.0009854146046563983,
-      "learning_rate": 0.0001999971835893436,
-      "loss": 46.0,
-      "step": 14845
-    },
-    {
-      "epoch": 2.3908369902169975,
-      "grad_norm": 0.0012606142554432154,
-      "learning_rate": 0.00019999718320963446,
-      "loss": 46.0,
-      "step": 14846
-    },
-    {
-      "epoch": 2.390998027295785,
-      "grad_norm": 0.005932171829044819,
-      "learning_rate": 0.00019999718282989976,
-      "loss": 46.0,
-      "step": 14847
-    },
-    {
-      "epoch": 2.3911590643745724,
-      "grad_norm": 0.01149502582848072,
-      "learning_rate": 0.00019999718245013946,
-      "loss": 46.0,
-      "step": 14848
-    },
-    {
-      "epoch": 2.39132010145336,
-      "grad_norm": 0.0016206150176003575,
-      "learning_rate": 0.00019999718207035357,
-      "loss": 46.0,
-      "step": 14849
-    },
-    {
-      "epoch": 2.391481138532147,
-      "grad_norm": 0.010081893764436245,
-      "learning_rate": 0.00019999718169054208,
-      "loss": 46.0,
-      "step": 14850
-    },
-    {
-      "epoch": 2.3916421756109343,
-      "grad_norm": 0.004157517571002245,
-      "learning_rate": 0.00019999718131070496,
-      "loss": 46.0,
-      "step": 14851
-    },
-    {
-      "epoch": 2.3918032126897217,
-      "grad_norm": 0.00331402150914073,
-      "learning_rate": 0.0001999971809308423,
-      "loss": 46.0,
-      "step": 14852
-    },
-    {
-      "epoch": 2.391964249768509,
-      "grad_norm": 0.005962159018963575,
-      "learning_rate": 0.00019999718055095403,
-      "loss": 46.0,
-      "step": 14853
-    },
-    {
-      "epoch": 2.3921252868472966,
-      "grad_norm": 0.0069003477692604065,
-      "learning_rate": 0.00019999718017104016,
-      "loss": 46.0,
-      "step": 14854
-    },
-    {
-      "epoch": 2.392286323926084,
-      "grad_norm": 0.0042977905832231045,
-      "learning_rate": 0.0001999971797911007,
-      "loss": 46.0,
-      "step": 14855
-    },
-    {
-      "epoch": 2.3924473610048715,
-      "grad_norm": 0.002373104216530919,
-      "learning_rate": 0.00019999717941113565,
-      "loss": 46.0,
-      "step": 14856
-    },
-    {
-      "epoch": 2.3926083980836585,
-      "grad_norm": 0.0027774544432759285,
-      "learning_rate": 0.00019999717903114501,
-      "loss": 46.0,
-      "step": 14857
-    },
-    {
-      "epoch": 2.392769435162446,
-      "grad_norm": 0.004388848785310984,
-      "learning_rate": 0.00019999717865112874,
-      "loss": 46.0,
-      "step": 14858
-    },
-    {
-      "epoch": 2.3929304722412335,
-      "grad_norm": 0.007151155732572079,
-      "learning_rate": 0.0001999971782710869,
-      "loss": 46.0,
-      "step": 14859
-    },
-    {
-      "epoch": 2.393091509320021,
-      "grad_norm": 0.004671855829656124,
-      "learning_rate": 0.00019999717789101948,
-      "loss": 46.0,
-      "step": 14860
-    },
-    {
-      "epoch": 2.3932525463988084,
-      "grad_norm": 0.005775718484073877,
-      "learning_rate": 0.00019999717751092644,
-      "loss": 46.0,
-      "step": 14861
-    },
-    {
-      "epoch": 2.393413583477596,
-      "grad_norm": 0.003141038352623582,
-      "learning_rate": 0.00019999717713080784,
-      "loss": 46.0,
-      "step": 14862
-    },
-    {
-      "epoch": 2.3935746205563833,
-      "grad_norm": 0.00044997193617746234,
-      "learning_rate": 0.0001999971767506636,
-      "loss": 46.0,
-      "step": 14863
-    },
-    {
-      "epoch": 2.3937356576351707,
-      "grad_norm": 0.008000251837074757,
-      "learning_rate": 0.0001999971763704938,
-      "loss": 46.0,
-      "step": 14864
-    },
-    {
-      "epoch": 2.3938966947139577,
-      "grad_norm": 0.001150397234596312,
-      "learning_rate": 0.00019999717599029841,
-      "loss": 46.0,
-      "step": 14865
-    },
-    {
-      "epoch": 2.394057731792745,
-      "grad_norm": 0.005656852386891842,
-      "learning_rate": 0.0001999971756100774,
-      "loss": 46.0,
-      "step": 14866
-    },
-    {
-      "epoch": 2.3942187688715326,
-      "grad_norm": 0.0030843340791761875,
-      "learning_rate": 0.00019999717522983082,
-      "loss": 46.0,
-      "step": 14867
-    },
-    {
-      "epoch": 2.39437980595032,
-      "grad_norm": 0.000997645198367536,
-      "learning_rate": 0.00019999717484955862,
-      "loss": 46.0,
-      "step": 14868
-    },
-    {
-      "epoch": 2.3945408430291075,
-      "grad_norm": 0.007723720744252205,
-      "learning_rate": 0.00019999717446926086,
-      "loss": 46.0,
-      "step": 14869
-    },
-    {
-      "epoch": 2.394701880107895,
-      "grad_norm": 0.004024618770927191,
-      "learning_rate": 0.00019999717408893748,
-      "loss": 46.0,
-      "step": 14870
-    },
-    {
-      "epoch": 2.3948629171866824,
-      "grad_norm": 0.0029589401092380285,
-      "learning_rate": 0.0001999971737085885,
-      "loss": 46.0,
-      "step": 14871
-    },
-    {
-      "epoch": 2.3950239542654694,
-      "grad_norm": 0.006197699345648289,
-      "learning_rate": 0.00019999717332821394,
-      "loss": 46.0,
-      "step": 14872
-    },
-    {
-      "epoch": 2.395184991344257,
-      "grad_norm": 0.00799486506730318,
-      "learning_rate": 0.0001999971729478138,
-      "loss": 46.0,
-      "step": 14873
-    },
-    {
-      "epoch": 2.3953460284230443,
-      "grad_norm": 0.005815913435071707,
-      "learning_rate": 0.00019999717256738805,
-      "loss": 46.0,
-      "step": 14874
-    },
-    {
-      "epoch": 2.395507065501832,
-      "grad_norm": 0.0010205275611951947,
-      "learning_rate": 0.0001999971721869367,
-      "loss": 46.0,
-      "step": 14875
-    },
-    {
-      "epoch": 2.3956681025806192,
-      "grad_norm": 0.003628675825893879,
-      "learning_rate": 0.00019999717180645975,
-      "loss": 46.0,
-      "step": 14876
-    },
-    {
-      "epoch": 2.3958291396594067,
-      "grad_norm": 0.0016113735036924481,
-      "learning_rate": 0.0001999971714259572,
-      "loss": 46.0,
-      "step": 14877
-    },
-    {
-      "epoch": 2.395990176738194,
-      "grad_norm": 0.00639236718416214,
-      "learning_rate": 0.0001999971710454291,
-      "loss": 46.0,
-      "step": 14878
-    },
-    {
-      "epoch": 2.3961512138169816,
-      "grad_norm": 0.007282912265509367,
-      "learning_rate": 0.00019999717066487537,
-      "loss": 46.0,
-      "step": 14879
-    },
-    {
-      "epoch": 2.3963122508957686,
-      "grad_norm": 0.002596805104985833,
-      "learning_rate": 0.00019999717028429607,
-      "loss": 46.0,
-      "step": 14880
-    },
-    {
-      "epoch": 2.396473287974556,
-      "grad_norm": 0.003063717857003212,
-      "learning_rate": 0.00019999716990369115,
-      "loss": 46.0,
-      "step": 14881
-    },
-    {
-      "epoch": 2.3966343250533435,
-      "grad_norm": 0.0022629154846072197,
-      "learning_rate": 0.00019999716952306064,
-      "loss": 46.0,
-      "step": 14882
-    },
-    {
-      "epoch": 2.396795362132131,
-      "grad_norm": 0.00248700357042253,
-      "learning_rate": 0.00019999716914240453,
-      "loss": 46.0,
-      "step": 14883
-    },
-    {
-      "epoch": 2.3969563992109184,
-      "grad_norm": 0.002188142156228423,
-      "learning_rate": 0.00019999716876172285,
-      "loss": 46.0,
-      "step": 14884
-    },
-    {
-      "epoch": 2.397117436289706,
-      "grad_norm": 0.004589318763464689,
-      "learning_rate": 0.00019999716838101555,
-      "loss": 46.0,
-      "step": 14885
-    },
-    {
-      "epoch": 2.397278473368493,
-      "grad_norm": 0.001246978179551661,
-      "learning_rate": 0.00019999716800028267,
-      "loss": 46.0,
-      "step": 14886
-    },
-    {
-      "epoch": 2.3974395104472803,
-      "grad_norm": 0.00201410916633904,
-      "learning_rate": 0.0001999971676195242,
-      "loss": 46.0,
-      "step": 14887
-    },
-    {
-      "epoch": 2.3976005475260678,
-      "grad_norm": 0.0018756977515295148,
-      "learning_rate": 0.00019999716723874013,
-      "loss": 46.0,
-      "step": 14888
-    },
-    {
-      "epoch": 2.3977615846048552,
-      "grad_norm": 0.001583478064276278,
-      "learning_rate": 0.00019999716685793048,
-      "loss": 46.0,
-      "step": 14889
-    },
-    {
-      "epoch": 2.3979226216836427,
-      "grad_norm": 0.0005613099783658981,
-      "learning_rate": 0.0001999971664770952,
-      "loss": 46.0,
-      "step": 14890
-    },
-    {
-      "epoch": 2.39808365876243,
-      "grad_norm": 0.0013335056137293577,
-      "learning_rate": 0.00019999716609623436,
-      "loss": 46.0,
-      "step": 14891
-    },
-    {
-      "epoch": 2.3982446958412176,
-      "grad_norm": 0.0022021501790732145,
-      "learning_rate": 0.00019999716571534793,
-      "loss": 46.0,
-      "step": 14892
-    },
-    {
-      "epoch": 2.398405732920005,
-      "grad_norm": 0.004060426261276007,
-      "learning_rate": 0.00019999716533443586,
-      "loss": 46.0,
-      "step": 14893
-    },
-    {
-      "epoch": 2.398566769998792,
-      "grad_norm": 0.0009838812984526157,
-      "learning_rate": 0.00019999716495349825,
-      "loss": 46.0,
-      "step": 14894
-    },
-    {
-      "epoch": 2.3987278070775795,
-      "grad_norm": 0.0049540214240550995,
-      "learning_rate": 0.000199997164572535,
-      "loss": 46.0,
-      "step": 14895
-    },
-    {
-      "epoch": 2.398888844156367,
-      "grad_norm": 0.005037226714193821,
-      "learning_rate": 0.0001999971641915462,
-      "loss": 46.0,
-      "step": 14896
-    },
-    {
-      "epoch": 2.3990498812351544,
-      "grad_norm": 0.001430519507266581,
-      "learning_rate": 0.0001999971638105318,
-      "loss": 46.0,
-      "step": 14897
-    },
-    {
-      "epoch": 2.399210918313942,
-      "grad_norm": 0.0011447274591773748,
-      "learning_rate": 0.00019999716342949177,
-      "loss": 46.0,
-      "step": 14898
-    },
-    {
-      "epoch": 2.3993719553927293,
-      "grad_norm": 0.0023657106794416904,
-      "learning_rate": 0.00019999716304842618,
-      "loss": 46.0,
-      "step": 14899
-    },
-    {
-      "epoch": 2.3995329924715167,
-      "grad_norm": 0.0013266261667013168,
-      "learning_rate": 0.00019999716266733497,
-      "loss": 46.0,
-      "step": 14900
-    },
-    {
-      "epoch": 2.3996940295503038,
-      "grad_norm": 0.006027529016137123,
-      "learning_rate": 0.0001999971622862182,
-      "loss": 46.0,
-      "step": 14901
-    },
-    {
-      "epoch": 2.399855066629091,
-      "grad_norm": 0.0024401552509516478,
-      "learning_rate": 0.0001999971619050758,
-      "loss": 46.0,
-      "step": 14902
-    },
-    {
-      "epoch": 2.4000161037078787,
-      "grad_norm": 0.004408691078424454,
-      "learning_rate": 0.0001999971615239078,
-      "loss": 46.0,
-      "step": 14903
-    },
-    {
-      "epoch": 2.400177140786666,
-      "grad_norm": 0.007748245261609554,
-      "learning_rate": 0.00019999716114271423,
-      "loss": 46.0,
-      "step": 14904
-    },
-    {
-      "epoch": 2.4003381778654536,
-      "grad_norm": 0.0010851210681721568,
-      "learning_rate": 0.00019999716076149506,
-      "loss": 46.0,
-      "step": 14905
-    },
-    {
-      "epoch": 2.400499214944241,
-      "grad_norm": 0.0025137283373624086,
-      "learning_rate": 0.0001999971603802503,
-      "loss": 46.0,
-      "step": 14906
-    },
-    {
-      "epoch": 2.4006602520230285,
-      "grad_norm": 0.0047279237769544125,
-      "learning_rate": 0.00019999715999897996,
-      "loss": 46.0,
-      "step": 14907
-    },
-    {
-      "epoch": 2.400821289101816,
-      "grad_norm": 0.004317928571254015,
-      "learning_rate": 0.000199997159617684,
-      "loss": 46.0,
-      "step": 14908
-    },
-    {
-      "epoch": 2.400982326180603,
-      "grad_norm": 0.001495165517553687,
-      "learning_rate": 0.00019999715923636245,
-      "loss": 46.0,
-      "step": 14909
-    },
-    {
-      "epoch": 2.4011433632593904,
-      "grad_norm": 0.0009965263307094574,
-      "learning_rate": 0.0001999971588550153,
-      "loss": 46.0,
-      "step": 14910
-    },
-    {
-      "epoch": 2.401304400338178,
-      "grad_norm": 0.0025162403471767902,
-      "learning_rate": 0.00019999715847364256,
-      "loss": 46.0,
-      "step": 14911
-    },
-    {
-      "epoch": 2.4014654374169653,
-      "grad_norm": 0.002974242437630892,
-      "learning_rate": 0.00019999715809224423,
-      "loss": 46.0,
-      "step": 14912
-    },
-    {
-      "epoch": 2.4016264744957527,
-      "grad_norm": 0.002061853650957346,
-      "learning_rate": 0.0001999971577108203,
-      "loss": 46.0,
-      "step": 14913
-    },
-    {
-      "epoch": 2.40178751157454,
-      "grad_norm": 0.0032915319316089153,
-      "learning_rate": 0.00019999715732937082,
-      "loss": 46.0,
-      "step": 14914
-    },
-    {
-      "epoch": 2.401948548653327,
-      "grad_norm": 0.0010055049788206816,
-      "learning_rate": 0.0001999971569478957,
-      "loss": 46.0,
-      "step": 14915
-    },
-    {
-      "epoch": 2.4021095857321146,
-      "grad_norm": 0.0016115966718643904,
-      "learning_rate": 0.00019999715656639498,
-      "loss": 46.0,
-      "step": 14916
-    },
-    {
-      "epoch": 2.402270622810902,
-      "grad_norm": 0.0027726360131055117,
-      "learning_rate": 0.00019999715618486868,
-      "loss": 46.0,
-      "step": 14917
-    },
-    {
-      "epoch": 2.4024316598896895,
-      "grad_norm": 0.0062407683581113815,
-      "learning_rate": 0.0001999971558033168,
-      "loss": 46.0,
-      "step": 14918
-    },
-    {
-      "epoch": 2.402592696968477,
-      "grad_norm": 0.00368437054567039,
-      "learning_rate": 0.00019999715542173932,
-      "loss": 46.0,
-      "step": 14919
-    },
-    {
-      "epoch": 2.4027537340472644,
-      "grad_norm": 0.0019279866246506572,
-      "learning_rate": 0.00019999715504013624,
-      "loss": 46.0,
-      "step": 14920
-    },
-    {
-      "epoch": 2.402914771126052,
-      "grad_norm": 0.006922426633536816,
-      "learning_rate": 0.00019999715465850753,
-      "loss": 46.0,
-      "step": 14921
-    },
-    {
-      "epoch": 2.4030758082048393,
-      "grad_norm": 0.0036330455914139748,
-      "learning_rate": 0.00019999715427685327,
-      "loss": 46.0,
-      "step": 14922
-    },
-    {
-      "epoch": 2.403236845283627,
-      "grad_norm": 0.004200490657240152,
-      "learning_rate": 0.00019999715389517342,
-      "loss": 46.0,
-      "step": 14923
-    },
-    {
-      "epoch": 2.403397882362414,
-      "grad_norm": 0.00803413987159729,
-      "learning_rate": 0.00019999715351346796,
-      "loss": 46.0,
-      "step": 14924
-    },
-    {
-      "epoch": 2.4035589194412013,
-      "grad_norm": 0.003241396276280284,
-      "learning_rate": 0.0001999971531317369,
-      "loss": 46.0,
-      "step": 14925
-    },
-    {
-      "epoch": 2.4037199565199887,
-      "grad_norm": 0.0018149223178625107,
-      "learning_rate": 0.00019999715274998024,
-      "loss": 46.0,
-      "step": 14926
-    },
-    {
-      "epoch": 2.403880993598776,
-      "grad_norm": 0.012932018376886845,
-      "learning_rate": 0.000199997152368198,
-      "loss": 46.0,
-      "step": 14927
-    },
-    {
-      "epoch": 2.4040420306775636,
-      "grad_norm": 0.001356132561340928,
-      "learning_rate": 0.00019999715198639018,
-      "loss": 46.0,
-      "step": 14928
-    },
-    {
-      "epoch": 2.404203067756351,
-      "grad_norm": 0.001434671925380826,
-      "learning_rate": 0.00019999715160455675,
-      "loss": 46.0,
-      "step": 14929
-    },
-    {
-      "epoch": 2.404364104835138,
-      "grad_norm": 0.00057082693092525,
-      "learning_rate": 0.00019999715122269774,
-      "loss": 46.0,
-      "step": 14930
-    },
-    {
-      "epoch": 2.4045251419139255,
-      "grad_norm": 0.001431204262189567,
-      "learning_rate": 0.0001999971508408131,
-      "loss": 46.0,
-      "step": 14931
-    },
-    {
-      "epoch": 2.404686178992713,
-      "grad_norm": 0.002392623107880354,
-      "learning_rate": 0.0001999971504589029,
-      "loss": 46.0,
-      "step": 14932
-    },
-    {
-      "epoch": 2.4048472160715004,
-      "grad_norm": 0.0021177795715630054,
-      "learning_rate": 0.00019999715007696712,
-      "loss": 46.0,
-      "step": 14933
-    },
-    {
-      "epoch": 2.405008253150288,
-      "grad_norm": 0.0009942085016518831,
-      "learning_rate": 0.0001999971496950057,
-      "loss": 46.0,
-      "step": 14934
-    },
-    {
-      "epoch": 2.4051692902290753,
-      "grad_norm": 0.0022986209951341152,
-      "learning_rate": 0.0001999971493130187,
-      "loss": 46.0,
-      "step": 14935
-    },
-    {
-      "epoch": 2.405330327307863,
-      "grad_norm": 0.0008761899662204087,
-      "learning_rate": 0.0001999971489310061,
-      "loss": 46.0,
-      "step": 14936
-    },
-    {
-      "epoch": 2.4054913643866502,
-      "grad_norm": 0.0021304022520780563,
-      "learning_rate": 0.00019999714854896793,
-      "loss": 46.0,
-      "step": 14937
-    },
-    {
-      "epoch": 2.4056524014654372,
-      "grad_norm": 0.0006061995518393815,
-      "learning_rate": 0.00019999714816690416,
-      "loss": 46.0,
-      "step": 14938
-    },
-    {
-      "epoch": 2.4058134385442247,
-      "grad_norm": 0.0013981732772663236,
-      "learning_rate": 0.0001999971477848148,
-      "loss": 46.0,
-      "step": 14939
-    },
-    {
-      "epoch": 2.405974475623012,
-      "grad_norm": 0.0017136174719780684,
-      "learning_rate": 0.00019999714740269982,
-      "loss": 46.0,
-      "step": 14940
-    },
-    {
-      "epoch": 2.4061355127017996,
-      "grad_norm": 0.00294609391130507,
-      "learning_rate": 0.00019999714702055926,
-      "loss": 46.0,
-      "step": 14941
-    },
-    {
-      "epoch": 2.406296549780587,
-      "grad_norm": 0.009054875001311302,
-      "learning_rate": 0.0001999971466383931,
-      "loss": 46.0,
-      "step": 14942
-    },
-    {
-      "epoch": 2.4064575868593745,
-      "grad_norm": 0.0005580727593041956,
-      "learning_rate": 0.00019999714625620137,
-      "loss": 46.0,
-      "step": 14943
-    },
-    {
-      "epoch": 2.406618623938162,
-      "grad_norm": 0.004075147211551666,
-      "learning_rate": 0.00019999714587398402,
-      "loss": 46.0,
-      "step": 14944
-    },
-    {
-      "epoch": 2.406779661016949,
-      "grad_norm": 0.0031745126470923424,
-      "learning_rate": 0.0001999971454917411,
-      "loss": 46.0,
-      "step": 14945
-    },
-    {
-      "epoch": 2.4069406980957364,
-      "grad_norm": 0.00141071155667305,
-      "learning_rate": 0.00019999714510947255,
-      "loss": 46.0,
-      "step": 14946
-    },
-    {
-      "epoch": 2.407101735174524,
-      "grad_norm": 0.0023850006982684135,
-      "learning_rate": 0.00019999714472717845,
-      "loss": 46.0,
-      "step": 14947
-    },
-    {
-      "epoch": 2.4072627722533113,
-      "grad_norm": 0.002219144022092223,
-      "learning_rate": 0.00019999714434485873,
-      "loss": 46.0,
-      "step": 14948
-    },
-    {
-      "epoch": 2.4074238093320988,
-      "grad_norm": 0.0008797465707175434,
-      "learning_rate": 0.00019999714396251342,
-      "loss": 46.0,
-      "step": 14949
-    },
-    {
-      "epoch": 2.407584846410886,
-      "grad_norm": 0.006275119259953499,
-      "learning_rate": 0.00019999714358014253,
-      "loss": 46.0,
-      "step": 14950
-    },
-    {
-      "epoch": 2.4077458834896737,
-      "grad_norm": 0.0007379829185083508,
-      "learning_rate": 0.000199997143197746,
-      "loss": 46.0,
-      "step": 14951
-    },
-    {
-      "epoch": 2.407906920568461,
-      "grad_norm": 0.0011120568960905075,
-      "learning_rate": 0.0001999971428153239,
-      "loss": 46.0,
-      "step": 14952
-    },
-    {
-      "epoch": 2.408067957647248,
-      "grad_norm": 0.003335511777549982,
-      "learning_rate": 0.00019999714243287625,
-      "loss": 46.0,
-      "step": 14953
-    },
-    {
-      "epoch": 2.4082289947260356,
-      "grad_norm": 0.002802102128043771,
-      "learning_rate": 0.00019999714205040295,
-      "loss": 46.0,
-      "step": 14954
-    },
-    {
-      "epoch": 2.408390031804823,
-      "grad_norm": 0.002060298342257738,
-      "learning_rate": 0.00019999714166790407,
-      "loss": 46.0,
-      "step": 14955
-    },
-    {
-      "epoch": 2.4085510688836105,
-      "grad_norm": 0.0016143816756084561,
-      "learning_rate": 0.0001999971412853796,
-      "loss": 46.0,
-      "step": 14956
-    },
-    {
-      "epoch": 2.408712105962398,
-      "grad_norm": 0.0033419255632907152,
-      "learning_rate": 0.00019999714090282951,
-      "loss": 46.0,
-      "step": 14957
-    },
-    {
-      "epoch": 2.4088731430411854,
-      "grad_norm": 0.004869893193244934,
-      "learning_rate": 0.00019999714052025387,
-      "loss": 46.0,
-      "step": 14958
-    },
-    {
-      "epoch": 2.4090341801199724,
-      "grad_norm": 0.00765066035091877,
-      "learning_rate": 0.00019999714013765264,
-      "loss": 46.0,
-      "step": 14959
-    },
-    {
-      "epoch": 2.40919521719876,
-      "grad_norm": 0.009709512814879417,
-      "learning_rate": 0.00019999713975502577,
-      "loss": 46.0,
-      "step": 14960
-    },
-    {
-      "epoch": 2.4093562542775473,
-      "grad_norm": 0.005278721451759338,
-      "learning_rate": 0.00019999713937237333,
-      "loss": 46.0,
-      "step": 14961
-    },
-    {
-      "epoch": 2.4095172913563347,
-      "grad_norm": 0.008919495157897472,
-      "learning_rate": 0.00019999713898969529,
-      "loss": 46.0,
-      "step": 14962
-    },
-    {
-      "epoch": 2.409678328435122,
-      "grad_norm": 0.004585885908454657,
-      "learning_rate": 0.00019999713860699168,
-      "loss": 46.0,
-      "step": 14963
-    },
-    {
-      "epoch": 2.4098393655139096,
-      "grad_norm": 0.0020918655209243298,
-      "learning_rate": 0.00019999713822426246,
-      "loss": 46.0,
-      "step": 14964
-    },
-    {
-      "epoch": 2.410000402592697,
-      "grad_norm": 0.0037259364034980536,
-      "learning_rate": 0.00019999713784150762,
-      "loss": 46.0,
-      "step": 14965
-    },
-    {
-      "epoch": 2.4101614396714846,
-      "grad_norm": 0.002379780635237694,
-      "learning_rate": 0.00019999713745872722,
-      "loss": 46.0,
-      "step": 14966
-    },
-    {
-      "epoch": 2.4103224767502716,
-      "grad_norm": 0.0013787498464807868,
-      "learning_rate": 0.00019999713707592121,
-      "loss": 46.0,
-      "step": 14967
-    },
-    {
-      "epoch": 2.410483513829059,
-      "grad_norm": 0.003032082924619317,
-      "learning_rate": 0.00019999713669308962,
-      "loss": 46.0,
-      "step": 14968
-    },
-    {
-      "epoch": 2.4106445509078465,
-      "grad_norm": 0.0030817643273621798,
-      "learning_rate": 0.00019999713631023243,
-      "loss": 46.0,
-      "step": 14969
-    },
-    {
-      "epoch": 2.410805587986634,
-      "grad_norm": 0.0008349365671165287,
-      "learning_rate": 0.00019999713592734963,
-      "loss": 46.0,
-      "step": 14970
-    },
-    {
-      "epoch": 2.4109666250654214,
-      "grad_norm": 0.0023388417903333902,
-      "learning_rate": 0.00019999713554444125,
-      "loss": 46.0,
-      "step": 14971
-    },
-    {
-      "epoch": 2.411127662144209,
-      "grad_norm": 0.0008196062408387661,
-      "learning_rate": 0.00019999713516150727,
-      "loss": 46.0,
-      "step": 14972
-    },
-    {
-      "epoch": 2.4112886992229963,
-      "grad_norm": 0.0014353410806506872,
-      "learning_rate": 0.0001999971347785477,
-      "loss": 46.0,
-      "step": 14973
-    },
-    {
-      "epoch": 2.4114497363017833,
-      "grad_norm": 0.0014467175351455808,
-      "learning_rate": 0.00019999713439556254,
-      "loss": 46.0,
-      "step": 14974
-    },
-    {
-      "epoch": 2.4116107733805707,
-      "grad_norm": 0.0007559700752608478,
-      "learning_rate": 0.00019999713401255178,
-      "loss": 46.0,
-      "step": 14975
-    },
-    {
-      "epoch": 2.411771810459358,
-      "grad_norm": 0.0010023186914622784,
-      "learning_rate": 0.00019999713362951543,
-      "loss": 46.0,
-      "step": 14976
-    },
-    {
-      "epoch": 2.4119328475381456,
-      "grad_norm": 0.0034047209192067385,
-      "learning_rate": 0.00019999713324645347,
-      "loss": 46.0,
-      "step": 14977
-    },
-    {
-      "epoch": 2.412093884616933,
-      "grad_norm": 0.009829115122556686,
-      "learning_rate": 0.00019999713286336594,
-      "loss": 46.0,
-      "step": 14978
-    },
-    {
-      "epoch": 2.4122549216957205,
-      "grad_norm": 0.0035766204819083214,
-      "learning_rate": 0.0001999971324802528,
-      "loss": 46.0,
-      "step": 14979
-    },
-    {
-      "epoch": 2.412415958774508,
-      "grad_norm": 0.003498433856293559,
-      "learning_rate": 0.00019999713209711408,
-      "loss": 46.0,
-      "step": 14980
-    },
-    {
-      "epoch": 2.4125769958532954,
-      "grad_norm": 0.008506706915795803,
-      "learning_rate": 0.00019999713171394974,
-      "loss": 46.0,
-      "step": 14981
-    },
-    {
-      "epoch": 2.4127380329320824,
-      "grad_norm": 0.0007884965161792934,
-      "learning_rate": 0.00019999713133075981,
-      "loss": 46.0,
-      "step": 14982
-    },
-    {
-      "epoch": 2.41289907001087,
-      "grad_norm": 0.014257362112402916,
-      "learning_rate": 0.0001999971309475443,
-      "loss": 46.0,
-      "step": 14983
-    },
-    {
-      "epoch": 2.4130601070896573,
-      "grad_norm": 0.0014589891070500016,
-      "learning_rate": 0.0001999971305643032,
-      "loss": 46.0,
-      "step": 14984
-    },
-    {
-      "epoch": 2.413221144168445,
-      "grad_norm": 0.0008877819636836648,
-      "learning_rate": 0.0001999971301810365,
-      "loss": 46.0,
-      "step": 14985
-    },
-    {
-      "epoch": 2.4133821812472322,
-      "grad_norm": 0.00781265553086996,
-      "learning_rate": 0.0001999971297977442,
-      "loss": 46.0,
-      "step": 14986
-    },
-    {
-      "epoch": 2.4135432183260197,
-      "grad_norm": 0.007220007479190826,
-      "learning_rate": 0.00019999712941442632,
-      "loss": 46.0,
-      "step": 14987
-    },
-    {
-      "epoch": 2.4137042554048067,
-      "grad_norm": 0.0015027987537905574,
-      "learning_rate": 0.00019999712903108282,
-      "loss": 46.0,
-      "step": 14988
-    },
-    {
-      "epoch": 2.413865292483594,
-      "grad_norm": 0.003292992478236556,
-      "learning_rate": 0.00019999712864771375,
-      "loss": 46.0,
-      "step": 14989
-    },
-    {
-      "epoch": 2.4140263295623816,
-      "grad_norm": 0.003830128815025091,
-      "learning_rate": 0.00019999712826431908,
-      "loss": 46.0,
-      "step": 14990
-    },
-    {
-      "epoch": 2.414187366641169,
-      "grad_norm": 0.001486469409428537,
-      "learning_rate": 0.00019999712788089884,
-      "loss": 46.0,
-      "step": 14991
-    },
-    {
-      "epoch": 2.4143484037199565,
-      "grad_norm": 0.0014831129228696227,
-      "learning_rate": 0.00019999712749745296,
-      "loss": 46.0,
-      "step": 14992
-    },
-    {
-      "epoch": 2.414509440798744,
-      "grad_norm": 0.007500668056309223,
-      "learning_rate": 0.00019999712711398152,
-      "loss": 46.0,
-      "step": 14993
-    },
-    {
-      "epoch": 2.4146704778775314,
-      "grad_norm": 0.001474690856412053,
-      "learning_rate": 0.00019999712673048444,
-      "loss": 46.0,
-      "step": 14994
-    },
-    {
-      "epoch": 2.414831514956319,
-      "grad_norm": 0.0032330448739230633,
-      "learning_rate": 0.00019999712634696183,
-      "loss": 46.0,
-      "step": 14995
-    },
-    {
-      "epoch": 2.4149925520351063,
-      "grad_norm": 0.0046609858982264996,
-      "learning_rate": 0.0001999971259634136,
-      "loss": 46.0,
-      "step": 14996
-    },
-    {
-      "epoch": 2.4151535891138933,
-      "grad_norm": 0.004924978129565716,
-      "learning_rate": 0.00019999712557983976,
-      "loss": 46.0,
-      "step": 14997
-    },
-    {
-      "epoch": 2.415314626192681,
-      "grad_norm": 0.004856908228248358,
-      "learning_rate": 0.00019999712519624035,
-      "loss": 46.0,
-      "step": 14998
-    },
-    {
-      "epoch": 2.4154756632714682,
-      "grad_norm": 0.004884161055088043,
-      "learning_rate": 0.0001999971248126153,
-      "loss": 46.0,
-      "step": 14999
-    },
-    {
-      "epoch": 2.4156367003502557,
-      "grad_norm": 0.002176631474867463,
-      "learning_rate": 0.00019999712442896473,
-      "loss": 46.0,
-      "step": 15000
-    },
-    {
-      "epoch": 2.415797737429043,
-      "grad_norm": 0.002556507708504796,
-      "learning_rate": 0.0001999971240452885,
-      "loss": 46.0,
-      "step": 15001
-    },
-    {
-      "epoch": 2.4159587745078306,
-      "grad_norm": 0.002298701787367463,
-      "learning_rate": 0.00019999712366158668,
-      "loss": 46.0,
-      "step": 15002
-    },
-    {
-      "epoch": 2.4161198115866176,
-      "grad_norm": 0.0019604400731623173,
-      "learning_rate": 0.0001999971232778593,
-      "loss": 46.0,
-      "step": 15003
-    },
-    {
-      "epoch": 2.416280848665405,
-      "grad_norm": 0.0022047865204513073,
-      "learning_rate": 0.0001999971228941063,
-      "loss": 46.0,
-      "step": 15004
-    },
-    {
-      "epoch": 2.4164418857441925,
-      "grad_norm": 0.0036231528501957655,
-      "learning_rate": 0.00019999712251032775,
-      "loss": 46.0,
-      "step": 15005
-    },
-    {
-      "epoch": 2.41660292282298,
-      "grad_norm": 0.0019326030742377043,
-      "learning_rate": 0.00019999712212652354,
-      "loss": 46.0,
-      "step": 15006
-    },
-    {
-      "epoch": 2.4167639599017674,
-      "grad_norm": 0.004777549300342798,
-      "learning_rate": 0.00019999712174269377,
-      "loss": 46.0,
-      "step": 15007
-    },
-    {
-      "epoch": 2.416924996980555,
-      "grad_norm": 0.0011353356530889869,
-      "learning_rate": 0.00019999712135883841,
-      "loss": 46.0,
-      "step": 15008
-    },
-    {
-      "epoch": 2.4170860340593423,
-      "grad_norm": 0.00082369614392519,
-      "learning_rate": 0.00019999712097495747,
-      "loss": 46.0,
-      "step": 15009
-    },
-    {
-      "epoch": 2.4172470711381298,
-      "grad_norm": 0.0021870385389775038,
-      "learning_rate": 0.00019999712059105091,
-      "loss": 46.0,
-      "step": 15010
-    },
-    {
-      "epoch": 2.4174081082169168,
-      "grad_norm": 0.0009749503806233406,
-      "learning_rate": 0.00019999712020711877,
-      "loss": 46.0,
-      "step": 15011
-    },
-    {
-      "epoch": 2.417569145295704,
-      "grad_norm": 0.004610496107488871,
-      "learning_rate": 0.00019999711982316104,
-      "loss": 46.0,
-      "step": 15012
-    },
-    {
-      "epoch": 2.4177301823744917,
-      "grad_norm": 0.0026237955316901207,
-      "learning_rate": 0.0001999971194391777,
-      "loss": 46.0,
-      "step": 15013
-    },
-    {
-      "epoch": 2.417891219453279,
-      "grad_norm": 0.0005824874388054013,
-      "learning_rate": 0.00019999711905516873,
-      "loss": 46.0,
-      "step": 15014
-    },
-    {
-      "epoch": 2.4180522565320666,
-      "grad_norm": 0.002638162113726139,
-      "learning_rate": 0.00019999711867113422,
-      "loss": 46.0,
-      "step": 15015
-    },
-    {
-      "epoch": 2.418213293610854,
-      "grad_norm": 0.004464765544980764,
-      "learning_rate": 0.0001999971182870741,
-      "loss": 46.0,
-      "step": 15016
-    },
-    {
-      "epoch": 2.4183743306896415,
-      "grad_norm": 0.00509841600432992,
-      "learning_rate": 0.0001999971179029884,
-      "loss": 46.0,
-      "step": 15017
-    },
-    {
-      "epoch": 2.4185353677684285,
-      "grad_norm": 0.0038666953332722187,
-      "learning_rate": 0.00019999711751887708,
-      "loss": 46.0,
-      "step": 15018
-    },
-    {
-      "epoch": 2.418696404847216,
-      "grad_norm": 0.0007318682037293911,
-      "learning_rate": 0.00019999711713474018,
-      "loss": 46.0,
-      "step": 15019
-    },
-    {
-      "epoch": 2.4188574419260034,
-      "grad_norm": 0.0016925224335864186,
-      "learning_rate": 0.00019999711675057767,
-      "loss": 46.0,
-      "step": 15020
-    },
-    {
-      "epoch": 2.419018479004791,
-      "grad_norm": 0.004939534701406956,
-      "learning_rate": 0.0001999971163663896,
-      "loss": 46.0,
-      "step": 15021
-    },
-    {
-      "epoch": 2.4191795160835783,
-      "grad_norm": 0.00682695722207427,
-      "learning_rate": 0.00019999711598217592,
-      "loss": 46.0,
-      "step": 15022
-    },
-    {
-      "epoch": 2.4193405531623657,
-      "grad_norm": 0.001709630829282105,
-      "learning_rate": 0.00019999711559793665,
-      "loss": 46.0,
-      "step": 15023
-    },
-    {
-      "epoch": 2.419501590241153,
-      "grad_norm": 0.003777744248509407,
-      "learning_rate": 0.00019999711521367176,
-      "loss": 46.0,
-      "step": 15024
-    },
-    {
-      "epoch": 2.4196626273199406,
-      "grad_norm": 0.000905441353097558,
-      "learning_rate": 0.00019999711482938132,
-      "loss": 46.0,
-      "step": 15025
-    },
-    {
-      "epoch": 2.4198236643987276,
-      "grad_norm": 0.0019152357708662748,
-      "learning_rate": 0.00019999711444506523,
-      "loss": 46.0,
-      "step": 15026
-    },
-    {
-      "epoch": 2.419984701477515,
-      "grad_norm": 0.012113110162317753,
-      "learning_rate": 0.00019999711406072358,
-      "loss": 46.0,
-      "step": 15027
-    },
-    {
-      "epoch": 2.4201457385563026,
-      "grad_norm": 0.002100149402394891,
-      "learning_rate": 0.00019999711367635635,
-      "loss": 46.0,
-      "step": 15028
-    },
-    {
-      "epoch": 2.42030677563509,
-      "grad_norm": 0.006856370717287064,
-      "learning_rate": 0.0001999971132919635,
-      "loss": 46.0,
-      "step": 15029
-    },
-    {
-      "epoch": 2.4204678127138775,
-      "grad_norm": 0.0014345949748530984,
-      "learning_rate": 0.00019999711290754507,
-      "loss": 46.0,
-      "step": 15030
-    },
-    {
-      "epoch": 2.420628849792665,
-      "grad_norm": 0.004308851435780525,
-      "learning_rate": 0.00019999711252310104,
-      "loss": 46.0,
-      "step": 15031
-    },
-    {
-      "epoch": 2.420789886871452,
-      "grad_norm": 0.0012742943363264203,
-      "learning_rate": 0.0001999971121386314,
-      "loss": 46.0,
-      "step": 15032
-    },
-    {
-      "epoch": 2.4209509239502394,
-      "grad_norm": 0.005597689189016819,
-      "learning_rate": 0.0001999971117541362,
-      "loss": 46.0,
-      "step": 15033
-    },
-    {
-      "epoch": 2.421111961029027,
-      "grad_norm": 0.001218323246575892,
-      "learning_rate": 0.0001999971113696154,
-      "loss": 46.0,
-      "step": 15034
-    },
-    {
-      "epoch": 2.4212729981078143,
-      "grad_norm": 0.006999565754085779,
-      "learning_rate": 0.00019999711098506897,
-      "loss": 46.0,
-      "step": 15035
-    },
-    {
-      "epoch": 2.4214340351866017,
-      "grad_norm": 0.002158647170290351,
-      "learning_rate": 0.00019999711060049698,
-      "loss": 46.0,
-      "step": 15036
-    },
-    {
-      "epoch": 2.421595072265389,
-      "grad_norm": 0.0011474784696474671,
-      "learning_rate": 0.00019999711021589938,
-      "loss": 46.0,
-      "step": 15037
-    },
-    {
-      "epoch": 2.4217561093441766,
-      "grad_norm": 0.006823455449193716,
-      "learning_rate": 0.0001999971098312762,
-      "loss": 46.0,
-      "step": 15038
-    },
-    {
-      "epoch": 2.421917146422964,
-      "grad_norm": 0.0015572105767205358,
-      "learning_rate": 0.0001999971094466274,
-      "loss": 46.0,
-      "step": 15039
-    },
-    {
-      "epoch": 2.4220781835017515,
-      "grad_norm": 0.002025666180998087,
-      "learning_rate": 0.000199997109061953,
-      "loss": 46.0,
-      "step": 15040
-    },
-    {
-      "epoch": 2.4222392205805385,
-      "grad_norm": 0.006478309165686369,
-      "learning_rate": 0.00019999710867725306,
-      "loss": 46.0,
-      "step": 15041
-    },
-    {
-      "epoch": 2.422400257659326,
-      "grad_norm": 0.0015047671040520072,
-      "learning_rate": 0.00019999710829252747,
-      "loss": 46.0,
-      "step": 15042
-    },
-    {
-      "epoch": 2.4225612947381134,
-      "grad_norm": 0.00624757120385766,
-      "learning_rate": 0.00019999710790777632,
-      "loss": 46.0,
-      "step": 15043
-    },
-    {
-      "epoch": 2.422722331816901,
-      "grad_norm": 0.0006859186105430126,
-      "learning_rate": 0.00019999710752299955,
-      "loss": 46.0,
-      "step": 15044
-    },
-    {
-      "epoch": 2.4228833688956883,
-      "grad_norm": 0.0015119699528440833,
-      "learning_rate": 0.00019999710713819723,
-      "loss": 46.0,
-      "step": 15045
-    },
-    {
-      "epoch": 2.423044405974476,
-      "grad_norm": 0.0013596673961728811,
-      "learning_rate": 0.0001999971067533693,
-      "loss": 46.0,
-      "step": 15046
-    },
-    {
-      "epoch": 2.423205443053263,
-      "grad_norm": 0.006602685898542404,
-      "learning_rate": 0.00019999710636851574,
-      "loss": 46.0,
-      "step": 15047
-    },
-    {
-      "epoch": 2.4233664801320502,
-      "grad_norm": 0.0016174034681171179,
-      "learning_rate": 0.00019999710598363662,
-      "loss": 46.0,
-      "step": 15048
-    },
-    {
-      "epoch": 2.4235275172108377,
-      "grad_norm": 0.001133667421527207,
-      "learning_rate": 0.00019999710559873187,
-      "loss": 46.0,
-      "step": 15049
-    },
-    {
-      "epoch": 2.423688554289625,
-      "grad_norm": 0.0011858440702781081,
-      "learning_rate": 0.00019999710521380158,
-      "loss": 46.0,
-      "step": 15050
-    },
-    {
-      "epoch": 2.4238495913684126,
-      "grad_norm": 0.0017343696672469378,
-      "learning_rate": 0.00019999710482884565,
-      "loss": 46.0,
-      "step": 15051
-    },
-    {
-      "epoch": 2.4240106284472,
-      "grad_norm": 0.0026560635305941105,
-      "learning_rate": 0.00019999710444386414,
-      "loss": 46.0,
-      "step": 15052
-    },
-    {
-      "epoch": 2.4241716655259875,
-      "grad_norm": 0.005059913732111454,
-      "learning_rate": 0.00019999710405885703,
-      "loss": 46.0,
-      "step": 15053
-    },
-    {
-      "epoch": 2.424332702604775,
-      "grad_norm": 0.005734195001423359,
-      "learning_rate": 0.00019999710367382437,
-      "loss": 46.0,
-      "step": 15054
-    },
-    {
-      "epoch": 2.424493739683562,
-      "grad_norm": 0.004790952894836664,
-      "learning_rate": 0.00019999710328876607,
-      "loss": 46.0,
-      "step": 15055
-    },
-    {
-      "epoch": 2.4246547767623494,
-      "grad_norm": 0.001556164352223277,
-      "learning_rate": 0.00019999710290368217,
-      "loss": 46.0,
-      "step": 15056
-    },
-    {
-      "epoch": 2.424815813841137,
-      "grad_norm": 0.0006882189773023129,
-      "learning_rate": 0.0001999971025185727,
-      "loss": 46.0,
-      "step": 15057
-    },
-    {
-      "epoch": 2.4249768509199243,
-      "grad_norm": 0.000901675783097744,
-      "learning_rate": 0.00019999710213343763,
-      "loss": 46.0,
-      "step": 15058
-    },
-    {
-      "epoch": 2.4251378879987118,
-      "grad_norm": 0.002039017155766487,
-      "learning_rate": 0.00019999710174827695,
-      "loss": 46.0,
-      "step": 15059
-    },
-    {
-      "epoch": 2.4252989250774992,
-      "grad_norm": 0.010212313383817673,
-      "learning_rate": 0.0001999971013630907,
-      "loss": 46.0,
-      "step": 15060
-    },
-    {
-      "epoch": 2.4254599621562867,
-      "grad_norm": 0.00674798060208559,
-      "learning_rate": 0.00019999710097787885,
-      "loss": 46.0,
-      "step": 15061
-    },
-    {
-      "epoch": 2.4256209992350737,
-      "grad_norm": 0.003238980658352375,
-      "learning_rate": 0.0001999971005926414,
-      "loss": 46.0,
-      "step": 15062
-    },
-    {
-      "epoch": 2.425782036313861,
-      "grad_norm": 0.007025308907032013,
-      "learning_rate": 0.00019999710020737835,
-      "loss": 46.0,
-      "step": 15063
-    },
-    {
-      "epoch": 2.4259430733926486,
-      "grad_norm": 0.006400213576853275,
-      "learning_rate": 0.0001999970998220897,
-      "loss": 46.0,
-      "step": 15064
-    },
-    {
-      "epoch": 2.426104110471436,
-      "grad_norm": 0.003598931012675166,
-      "learning_rate": 0.00019999709943677548,
-      "loss": 46.0,
-      "step": 15065
-    },
-    {
-      "epoch": 2.4262651475502235,
-      "grad_norm": 0.002956765005365014,
-      "learning_rate": 0.00019999709905143566,
-      "loss": 46.0,
-      "step": 15066
-    },
-    {
-      "epoch": 2.426426184629011,
-      "grad_norm": 0.003289261134341359,
-      "learning_rate": 0.00019999709866607023,
-      "loss": 46.0,
-      "step": 15067
-    },
-    {
-      "epoch": 2.4265872217077984,
-      "grad_norm": 0.00262340996414423,
-      "learning_rate": 0.00019999709828067924,
-      "loss": 46.0,
-      "step": 15068
-    },
-    {
-      "epoch": 2.426748258786586,
-      "grad_norm": 0.002898283302783966,
-      "learning_rate": 0.00019999709789526263,
-      "loss": 46.0,
-      "step": 15069
-    },
-    {
-      "epoch": 2.426909295865373,
-      "grad_norm": 0.005563546437770128,
-      "learning_rate": 0.00019999709750982044,
-      "loss": 46.0,
-      "step": 15070
-    },
-    {
-      "epoch": 2.4270703329441603,
-      "grad_norm": 0.0016081866342574358,
-      "learning_rate": 0.00019999709712435263,
-      "loss": 46.0,
-      "step": 15071
-    },
-    {
-      "epoch": 2.4272313700229478,
-      "grad_norm": 0.001116144354455173,
-      "learning_rate": 0.00019999709673885924,
-      "loss": 46.0,
-      "step": 15072
-    },
-    {
-      "epoch": 2.427392407101735,
-      "grad_norm": 0.0029423756059259176,
-      "learning_rate": 0.00019999709635334025,
-      "loss": 46.0,
-      "step": 15073
-    },
-    {
-      "epoch": 2.4275534441805227,
-      "grad_norm": 0.002192451385781169,
-      "learning_rate": 0.00019999709596779568,
-      "loss": 46.0,
-      "step": 15074
-    },
-    {
-      "epoch": 2.42771448125931,
-      "grad_norm": 0.0009762877598404884,
-      "learning_rate": 0.0001999970955822255,
-      "loss": 46.0,
-      "step": 15075
-    },
-    {
-      "epoch": 2.427875518338097,
-      "grad_norm": 0.002556285820901394,
-      "learning_rate": 0.00019999709519662973,
-      "loss": 46.0,
-      "step": 15076
-    },
-    {
-      "epoch": 2.4280365554168846,
-      "grad_norm": 0.008240649476647377,
-      "learning_rate": 0.00019999709481100837,
-      "loss": 46.0,
-      "step": 15077
-    },
-    {
-      "epoch": 2.428197592495672,
-      "grad_norm": 0.0010634667705744505,
-      "learning_rate": 0.00019999709442536143,
-      "loss": 46.0,
-      "step": 15078
-    },
-    {
-      "epoch": 2.4283586295744595,
-      "grad_norm": 0.0036353811156004667,
-      "learning_rate": 0.00019999709403968887,
-      "loss": 46.0,
-      "step": 15079
-    },
-    {
-      "epoch": 2.428519666653247,
-      "grad_norm": 0.0009891962399706244,
-      "learning_rate": 0.00019999709365399072,
-      "loss": 46.0,
-      "step": 15080
-    },
-    {
-      "epoch": 2.4286807037320344,
-      "grad_norm": 0.002965726889669895,
-      "learning_rate": 0.00019999709326826701,
-      "loss": 46.0,
-      "step": 15081
-    },
-    {
-      "epoch": 2.428841740810822,
-      "grad_norm": 0.008746610954403877,
-      "learning_rate": 0.00019999709288251767,
-      "loss": 46.0,
-      "step": 15082
-    },
-    {
-      "epoch": 2.4290027778896093,
-      "grad_norm": 0.002273347694426775,
-      "learning_rate": 0.00019999709249674273,
-      "loss": 46.0,
-      "step": 15083
-    },
-    {
-      "epoch": 2.4291638149683963,
-      "grad_norm": 0.0005529571790248156,
-      "learning_rate": 0.0001999970921109422,
-      "loss": 46.0,
-      "step": 15084
-    },
-    {
-      "epoch": 2.4293248520471837,
-      "grad_norm": 0.005225272383540869,
-      "learning_rate": 0.00019999709172511613,
-      "loss": 46.0,
-      "step": 15085
-    },
-    {
-      "epoch": 2.429485889125971,
-      "grad_norm": 0.0026787114329636097,
-      "learning_rate": 0.0001999970913392644,
-      "loss": 46.0,
-      "step": 15086
-    },
-    {
-      "epoch": 2.4296469262047586,
-      "grad_norm": 0.0020417559426277876,
-      "learning_rate": 0.0001999970909533871,
-      "loss": 46.0,
-      "step": 15087
-    },
-    {
-      "epoch": 2.429807963283546,
-      "grad_norm": 0.0030723819509148598,
-      "learning_rate": 0.00019999709056748422,
-      "loss": 46.0,
-      "step": 15088
-    },
-    {
-      "epoch": 2.4299690003623335,
-      "grad_norm": 0.002557441359385848,
-      "learning_rate": 0.00019999709018155573,
-      "loss": 46.0,
-      "step": 15089
-    },
-    {
-      "epoch": 2.430130037441121,
-      "grad_norm": 0.0005858919466845691,
-      "learning_rate": 0.00019999708979560163,
-      "loss": 46.0,
-      "step": 15090
-    },
-    {
-      "epoch": 2.430291074519908,
-      "grad_norm": 0.0015735187334939837,
-      "learning_rate": 0.00019999708940962198,
-      "loss": 46.0,
-      "step": 15091
-    },
-    {
-      "epoch": 2.4304521115986955,
-      "grad_norm": 0.00510045513510704,
-      "learning_rate": 0.0001999970890236167,
-      "loss": 46.0,
-      "step": 15092
-    },
-    {
-      "epoch": 2.430613148677483,
-      "grad_norm": 0.0010346854105591774,
-      "learning_rate": 0.0001999970886375858,
-      "loss": 46.0,
-      "step": 15093
-    },
-    {
-      "epoch": 2.4307741857562704,
-      "grad_norm": 0.0014213686808943748,
-      "learning_rate": 0.00019999708825152936,
-      "loss": 46.0,
-      "step": 15094
-    },
-    {
-      "epoch": 2.430935222835058,
-      "grad_norm": 0.0025495279114693403,
-      "learning_rate": 0.0001999970878654473,
-      "loss": 46.0,
-      "step": 15095
-    },
-    {
-      "epoch": 2.4310962599138453,
-      "grad_norm": 0.004010245203971863,
-      "learning_rate": 0.00019999708747933965,
-      "loss": 46.0,
-      "step": 15096
-    },
-    {
-      "epoch": 2.4312572969926327,
-      "grad_norm": 0.002637492259964347,
-      "learning_rate": 0.00019999708709320642,
-      "loss": 46.0,
-      "step": 15097
-    },
-    {
-      "epoch": 2.43141833407142,
-      "grad_norm": 0.0017560409614816308,
-      "learning_rate": 0.00019999708670704756,
-      "loss": 46.0,
-      "step": 15098
-    },
-    {
-      "epoch": 2.431579371150207,
-      "grad_norm": 0.0018749248702079058,
-      "learning_rate": 0.00019999708632086313,
-      "loss": 46.0,
-      "step": 15099
-    },
-    {
-      "epoch": 2.4317404082289946,
-      "grad_norm": 0.0029514131601899862,
-      "learning_rate": 0.00019999708593465313,
-      "loss": 46.0,
-      "step": 15100
-    },
-    {
-      "epoch": 2.431901445307782,
-      "grad_norm": 0.0014235855778679252,
-      "learning_rate": 0.0001999970855484175,
-      "loss": 46.0,
-      "step": 15101
-    },
-    {
-      "epoch": 2.4320624823865695,
-      "grad_norm": 0.0014500432880595326,
-      "learning_rate": 0.0001999970851621563,
-      "loss": 46.0,
-      "step": 15102
-    },
-    {
-      "epoch": 2.432223519465357,
-      "grad_norm": 0.003533906303346157,
-      "learning_rate": 0.00019999708477586947,
-      "loss": 46.0,
-      "step": 15103
-    },
-    {
-      "epoch": 2.4323845565441444,
-      "grad_norm": 0.005040640011429787,
-      "learning_rate": 0.00019999708438955707,
-      "loss": 46.0,
-      "step": 15104
-    },
-    {
-      "epoch": 2.4325455936229314,
-      "grad_norm": 0.003788456553593278,
-      "learning_rate": 0.00019999708400321908,
-      "loss": 46.0,
-      "step": 15105
-    },
-    {
-      "epoch": 2.432706630701719,
-      "grad_norm": 0.002008729847148061,
-      "learning_rate": 0.00019999708361685548,
-      "loss": 46.0,
-      "step": 15106
-    },
-    {
-      "epoch": 2.4328676677805063,
-      "grad_norm": 0.0017324623186141253,
-      "learning_rate": 0.0001999970832304663,
-      "loss": 46.0,
-      "step": 15107
-    },
-    {
-      "epoch": 2.433028704859294,
-      "grad_norm": 0.0033673131838440895,
-      "learning_rate": 0.00019999708284405151,
-      "loss": 46.0,
-      "step": 15108
-    },
-    {
-      "epoch": 2.4331897419380812,
-      "grad_norm": 0.0046812789514660835,
-      "learning_rate": 0.00019999708245761115,
-      "loss": 46.0,
-      "step": 15109
-    },
-    {
-      "epoch": 2.4333507790168687,
-      "grad_norm": 0.00128064991440624,
-      "learning_rate": 0.00019999708207114517,
-      "loss": 46.0,
-      "step": 15110
-    },
-    {
-      "epoch": 2.433511816095656,
-      "grad_norm": 0.00835944153368473,
-      "learning_rate": 0.0001999970816846536,
-      "loss": 46.0,
-      "step": 15111
-    },
-    {
-      "epoch": 2.4336728531744436,
-      "grad_norm": 0.003728966461494565,
-      "learning_rate": 0.00019999708129813648,
-      "loss": 46.0,
-      "step": 15112
-    },
-    {
-      "epoch": 2.433833890253231,
-      "grad_norm": 0.001890342915430665,
-      "learning_rate": 0.00019999708091159374,
-      "loss": 46.0,
-      "step": 15113
-    },
-    {
-      "epoch": 2.433994927332018,
-      "grad_norm": 0.012927904725074768,
-      "learning_rate": 0.00019999708052502538,
-      "loss": 46.0,
-      "step": 15114
-    },
-    {
-      "epoch": 2.4341559644108055,
-      "grad_norm": 0.001587510691024363,
-      "learning_rate": 0.00019999708013843144,
-      "loss": 46.0,
-      "step": 15115
-    },
-    {
-      "epoch": 2.434317001489593,
-      "grad_norm": 0.0012859930284321308,
-      "learning_rate": 0.0001999970797518119,
-      "loss": 46.0,
-      "step": 15116
-    },
-    {
-      "epoch": 2.4344780385683804,
-      "grad_norm": 0.0025028849486261606,
-      "learning_rate": 0.00019999707936516677,
-      "loss": 46.0,
-      "step": 15117
-    },
-    {
-      "epoch": 2.434639075647168,
-      "grad_norm": 0.005002887919545174,
-      "learning_rate": 0.00019999707897849607,
-      "loss": 46.0,
-      "step": 15118
-    },
-    {
-      "epoch": 2.4348001127259553,
-      "grad_norm": 0.0005653257830999792,
-      "learning_rate": 0.00019999707859179975,
-      "loss": 46.0,
-      "step": 15119
-    },
-    {
-      "epoch": 2.4349611498047423,
-      "grad_norm": 0.001642723334953189,
-      "learning_rate": 0.00019999707820507785,
-      "loss": 46.0,
-      "step": 15120
-    },
-    {
-      "epoch": 2.4351221868835298,
-      "grad_norm": 0.002714628353714943,
-      "learning_rate": 0.00019999707781833036,
-      "loss": 46.0,
-      "step": 15121
-    },
-    {
-      "epoch": 2.4352832239623172,
-      "grad_norm": 0.001865481142885983,
-      "learning_rate": 0.00019999707743155725,
-      "loss": 46.0,
-      "step": 15122
-    },
-    {
-      "epoch": 2.4354442610411047,
-      "grad_norm": 0.009342843666672707,
-      "learning_rate": 0.00019999707704475856,
-      "loss": 46.0,
-      "step": 15123
-    },
-    {
-      "epoch": 2.435605298119892,
-      "grad_norm": 0.0034045828506350517,
-      "learning_rate": 0.00019999707665793425,
-      "loss": 46.0,
-      "step": 15124
-    },
-    {
-      "epoch": 2.4357663351986796,
-      "grad_norm": 0.0012691464507952332,
-      "learning_rate": 0.00019999707627108438,
-      "loss": 46.0,
-      "step": 15125
-    },
-    {
-      "epoch": 2.435927372277467,
-      "grad_norm": 0.0018481611041352153,
-      "learning_rate": 0.0001999970758842089,
-      "loss": 46.0,
-      "step": 15126
-    },
-    {
-      "epoch": 2.4360884093562545,
-      "grad_norm": 0.0030695737805217505,
-      "learning_rate": 0.00019999707549730786,
-      "loss": 46.0,
-      "step": 15127
-    },
-    {
-      "epoch": 2.4362494464350415,
-      "grad_norm": 0.0012932880781590939,
-      "learning_rate": 0.0001999970751103812,
-      "loss": 46.0,
-      "step": 15128
-    },
-    {
-      "epoch": 2.436410483513829,
-      "grad_norm": 0.0030933471862226725,
-      "learning_rate": 0.00019999707472342896,
-      "loss": 46.0,
-      "step": 15129
-    },
-    {
-      "epoch": 2.4365715205926164,
-      "grad_norm": 0.001895736437290907,
-      "learning_rate": 0.0001999970743364511,
-      "loss": 46.0,
-      "step": 15130
-    },
-    {
-      "epoch": 2.436732557671404,
-      "grad_norm": 0.0023412881419062614,
-      "learning_rate": 0.00019999707394944766,
-      "loss": 46.0,
-      "step": 15131
-    },
-    {
-      "epoch": 2.4368935947501913,
-      "grad_norm": 0.0022735181264579296,
-      "learning_rate": 0.00019999707356241862,
-      "loss": 46.0,
-      "step": 15132
-    },
-    {
-      "epoch": 2.4370546318289787,
-      "grad_norm": 0.0024098760914057493,
-      "learning_rate": 0.00019999707317536398,
-      "loss": 46.0,
-      "step": 15133
-    },
-    {
-      "epoch": 2.437215668907766,
-      "grad_norm": 0.0007730990764684975,
-      "learning_rate": 0.00019999707278828374,
-      "loss": 46.0,
-      "step": 15134
-    },
-    {
-      "epoch": 2.437376705986553,
-      "grad_norm": 0.004460522439330816,
-      "learning_rate": 0.00019999707240117792,
-      "loss": 46.0,
-      "step": 15135
-    },
-    {
-      "epoch": 2.4375377430653407,
-      "grad_norm": 0.005358945578336716,
-      "learning_rate": 0.00019999707201404652,
-      "loss": 46.0,
-      "step": 15136
-    },
-    {
-      "epoch": 2.437698780144128,
-      "grad_norm": 0.003004533238708973,
-      "learning_rate": 0.00019999707162688952,
-      "loss": 46.0,
-      "step": 15137
-    },
-    {
-      "epoch": 2.4378598172229156,
-      "grad_norm": 0.0032510908786207438,
-      "learning_rate": 0.0001999970712397069,
-      "loss": 46.0,
-      "step": 15138
-    },
-    {
-      "epoch": 2.438020854301703,
-      "grad_norm": 0.002162961522117257,
-      "learning_rate": 0.0001999970708524987,
-      "loss": 46.0,
-      "step": 15139
-    },
-    {
-      "epoch": 2.4381818913804905,
-      "grad_norm": 0.0015356499934569001,
-      "learning_rate": 0.00019999707046526493,
-      "loss": 46.0,
-      "step": 15140
-    },
-    {
-      "epoch": 2.438342928459278,
-      "grad_norm": 0.0042129456996917725,
-      "learning_rate": 0.00019999707007800554,
-      "loss": 46.0,
-      "step": 15141
-    },
-    {
-      "epoch": 2.4385039655380654,
-      "grad_norm": 0.0011561081046238542,
-      "learning_rate": 0.00019999706969072055,
-      "loss": 46.0,
-      "step": 15142
-    },
-    {
-      "epoch": 2.4386650026168524,
-      "grad_norm": 0.00693810498341918,
-      "learning_rate": 0.00019999706930340998,
-      "loss": 46.0,
-      "step": 15143
-    },
-    {
-      "epoch": 2.43882603969564,
-      "grad_norm": 0.0008224930497817695,
-      "learning_rate": 0.00019999706891607382,
-      "loss": 46.0,
-      "step": 15144
-    },
-    {
-      "epoch": 2.4389870767744273,
-      "grad_norm": 0.000820188841316849,
-      "learning_rate": 0.00019999706852871207,
-      "loss": 46.0,
-      "step": 15145
-    },
-    {
-      "epoch": 2.4391481138532147,
-      "grad_norm": 0.0014454848133027554,
-      "learning_rate": 0.0001999970681413247,
-      "loss": 46.0,
-      "step": 15146
-    },
-    {
-      "epoch": 2.439309150932002,
-      "grad_norm": 0.006936587393283844,
-      "learning_rate": 0.00019999706775391176,
-      "loss": 46.0,
-      "step": 15147
-    },
-    {
-      "epoch": 2.4394701880107896,
-      "grad_norm": 0.0022375667467713356,
-      "learning_rate": 0.0001999970673664732,
-      "loss": 46.0,
-      "step": 15148
-    },
-    {
-      "epoch": 2.4396312250895766,
-      "grad_norm": 0.005586666986346245,
-      "learning_rate": 0.00019999706697900908,
-      "loss": 46.0,
-      "step": 15149
-    },
-    {
-      "epoch": 2.439792262168364,
-      "grad_norm": 0.0030356997158378363,
-      "learning_rate": 0.00019999706659151934,
-      "loss": 46.0,
-      "step": 15150
-    },
-    {
-      "epoch": 2.4399532992471515,
-      "grad_norm": 0.006025013048201799,
-      "learning_rate": 0.00019999706620400402,
-      "loss": 46.0,
-      "step": 15151
-    },
-    {
-      "epoch": 2.440114336325939,
-      "grad_norm": 0.0006395922973752022,
-      "learning_rate": 0.0001999970658164631,
-      "loss": 46.0,
-      "step": 15152
-    },
-    {
-      "epoch": 2.4402753734047264,
-      "grad_norm": 0.016943635419011116,
-      "learning_rate": 0.0001999970654288966,
-      "loss": 46.0,
-      "step": 15153
-    },
-    {
-      "epoch": 2.440436410483514,
-      "grad_norm": 0.0020490989554673433,
-      "learning_rate": 0.0001999970650413045,
-      "loss": 46.0,
-      "step": 15154
-    },
-    {
-      "epoch": 2.4405974475623013,
-      "grad_norm": 0.003953774459660053,
-      "learning_rate": 0.0001999970646536868,
-      "loss": 46.0,
-      "step": 15155
-    },
-    {
-      "epoch": 2.440758484641089,
-      "grad_norm": 0.003514370182529092,
-      "learning_rate": 0.0001999970642660435,
-      "loss": 46.0,
-      "step": 15156
-    },
-    {
-      "epoch": 2.4409195217198763,
-      "grad_norm": 0.0014731710543856025,
-      "learning_rate": 0.00019999706387837458,
-      "loss": 46.0,
-      "step": 15157
-    },
-    {
-      "epoch": 2.4410805587986633,
-      "grad_norm": 0.002514501567929983,
-      "learning_rate": 0.00019999706349068012,
-      "loss": 46.0,
-      "step": 15158
-    },
-    {
-      "epoch": 2.4412415958774507,
-      "grad_norm": 0.0017521616537123919,
-      "learning_rate": 0.00019999706310296005,
-      "loss": 46.0,
-      "step": 15159
-    },
-    {
-      "epoch": 2.441402632956238,
-      "grad_norm": 0.007570066023617983,
-      "learning_rate": 0.00019999706271521433,
-      "loss": 46.0,
-      "step": 15160
-    },
-    {
-      "epoch": 2.4415636700350256,
-      "grad_norm": 0.007405099458992481,
-      "learning_rate": 0.0001999970623274431,
-      "loss": 46.0,
-      "step": 15161
-    },
-    {
-      "epoch": 2.441724707113813,
-      "grad_norm": 0.004569333977997303,
-      "learning_rate": 0.00019999706193964625,
-      "loss": 46.0,
-      "step": 15162
-    },
-    {
-      "epoch": 2.4418857441926005,
-      "grad_norm": 0.0010243856813758612,
-      "learning_rate": 0.0001999970615518238,
-      "loss": 46.0,
-      "step": 15163
-    },
-    {
-      "epoch": 2.4420467812713875,
-      "grad_norm": 0.0019582172390073538,
-      "learning_rate": 0.00019999706116397573,
-      "loss": 46.0,
-      "step": 15164
-    },
-    {
-      "epoch": 2.442207818350175,
-      "grad_norm": 0.002119868528097868,
-      "learning_rate": 0.0001999970607761021,
-      "loss": 46.0,
-      "step": 15165
-    },
-    {
-      "epoch": 2.4423688554289624,
-      "grad_norm": 0.0009905437473207712,
-      "learning_rate": 0.00019999706038820284,
-      "loss": 46.0,
-      "step": 15166
-    },
-    {
-      "epoch": 2.44252989250775,
-      "grad_norm": 0.0014513899805024266,
-      "learning_rate": 0.00019999706000027804,
-      "loss": 46.0,
-      "step": 15167
-    },
-    {
-      "epoch": 2.4426909295865373,
-      "grad_norm": 0.0010807986836880445,
-      "learning_rate": 0.0001999970596123276,
-      "loss": 46.0,
-      "step": 15168
-    },
-    {
-      "epoch": 2.442851966665325,
-      "grad_norm": 0.002748708939179778,
-      "learning_rate": 0.00019999705922435157,
-      "loss": 46.0,
-      "step": 15169
-    },
-    {
-      "epoch": 2.4430130037441122,
-      "grad_norm": 0.0014928893651813269,
-      "learning_rate": 0.00019999705883634995,
-      "loss": 46.0,
-      "step": 15170
-    },
-    {
-      "epoch": 2.4431740408228997,
-      "grad_norm": 0.0013083310332149267,
-      "learning_rate": 0.00019999705844832275,
-      "loss": 46.0,
-      "step": 15171
-    },
-    {
-      "epoch": 2.4433350779016867,
-      "grad_norm": 0.006076828110963106,
-      "learning_rate": 0.00019999705806026996,
-      "loss": 46.0,
-      "step": 15172
-    },
-    {
-      "epoch": 2.443496114980474,
-      "grad_norm": 0.003243883140385151,
-      "learning_rate": 0.00019999705767219156,
-      "loss": 46.0,
-      "step": 15173
-    },
-    {
-      "epoch": 2.4436571520592616,
-      "grad_norm": 0.0012554647400975227,
-      "learning_rate": 0.00019999705728408756,
-      "loss": 46.0,
-      "step": 15174
-    },
-    {
-      "epoch": 2.443818189138049,
-      "grad_norm": 0.0049966671504080296,
-      "learning_rate": 0.000199997056895958,
-      "loss": 46.0,
-      "step": 15175
-    },
-    {
-      "epoch": 2.4439792262168365,
-      "grad_norm": 0.009989946149289608,
-      "learning_rate": 0.0001999970565078028,
-      "loss": 46.0,
-      "step": 15176
-    },
-    {
-      "epoch": 2.444140263295624,
-      "grad_norm": 0.0031812230590730906,
-      "learning_rate": 0.00019999705611962204,
-      "loss": 46.0,
-      "step": 15177
-    },
-    {
-      "epoch": 2.4443013003744114,
-      "grad_norm": 0.0013804750051349401,
-      "learning_rate": 0.00019999705573141565,
-      "loss": 46.0,
-      "step": 15178
-    },
-    {
-      "epoch": 2.4444623374531984,
-      "grad_norm": 0.0021921817678958178,
-      "learning_rate": 0.00019999705534318372,
-      "loss": 46.0,
-      "step": 15179
-    },
-    {
-      "epoch": 2.444623374531986,
-      "grad_norm": 0.00106932467315346,
-      "learning_rate": 0.00019999705495492615,
-      "loss": 46.0,
-      "step": 15180
-    },
-    {
-      "epoch": 2.4447844116107733,
-      "grad_norm": 0.004719895776361227,
-      "learning_rate": 0.000199997054566643,
-      "loss": 46.0,
-      "step": 15181
-    },
-    {
-      "epoch": 2.4449454486895608,
-      "grad_norm": 0.0025353101082146168,
-      "learning_rate": 0.00019999705417833425,
-      "loss": 46.0,
-      "step": 15182
-    },
-    {
-      "epoch": 2.445106485768348,
-      "grad_norm": 0.0012668155832216144,
-      "learning_rate": 0.00019999705378999992,
-      "loss": 46.0,
-      "step": 15183
-    },
-    {
-      "epoch": 2.4452675228471357,
-      "grad_norm": 0.002183428732678294,
-      "learning_rate": 0.00019999705340163998,
-      "loss": 46.0,
-      "step": 15184
-    },
-    {
-      "epoch": 2.445428559925923,
-      "grad_norm": 0.0007727840566076338,
-      "learning_rate": 0.00019999705301325447,
-      "loss": 46.0,
-      "step": 15185
-    },
-    {
-      "epoch": 2.4455895970047106,
-      "grad_norm": 0.002086078282445669,
-      "learning_rate": 0.00019999705262484333,
-      "loss": 46.0,
-      "step": 15186
-    },
-    {
-      "epoch": 2.4457506340834976,
-      "grad_norm": 0.006999995559453964,
-      "learning_rate": 0.00019999705223640662,
-      "loss": 46.0,
-      "step": 15187
-    },
-    {
-      "epoch": 2.445911671162285,
-      "grad_norm": 0.004946023691445589,
-      "learning_rate": 0.00019999705184794433,
-      "loss": 46.0,
-      "step": 15188
-    },
-    {
-      "epoch": 2.4460727082410725,
-      "grad_norm": 0.0027686201501637697,
-      "learning_rate": 0.00019999705145945642,
-      "loss": 46.0,
-      "step": 15189
-    },
-    {
-      "epoch": 2.44623374531986,
-      "grad_norm": 0.002385171363130212,
-      "learning_rate": 0.00019999705107094293,
-      "loss": 46.0,
-      "step": 15190
-    },
-    {
-      "epoch": 2.4463947823986474,
-      "grad_norm": 0.009861724451184273,
-      "learning_rate": 0.00019999705068240382,
-      "loss": 46.0,
-      "step": 15191
-    },
-    {
-      "epoch": 2.446555819477435,
-      "grad_norm": 0.00755418511107564,
-      "learning_rate": 0.00019999705029383915,
-      "loss": 46.0,
-      "step": 15192
-    },
-    {
-      "epoch": 2.446716856556222,
-      "grad_norm": 0.0023931374307721853,
-      "learning_rate": 0.00019999704990524887,
-      "loss": 46.0,
-      "step": 15193
-    },
-    {
-      "epoch": 2.4468778936350093,
-      "grad_norm": 0.007331613451242447,
-      "learning_rate": 0.000199997049516633,
-      "loss": 46.0,
-      "step": 15194
-    },
-    {
-      "epoch": 2.4470389307137967,
-      "grad_norm": 0.00670094508677721,
-      "learning_rate": 0.00019999704912799154,
-      "loss": 46.0,
-      "step": 15195
-    },
-    {
-      "epoch": 2.447199967792584,
-      "grad_norm": 0.004451238550245762,
-      "learning_rate": 0.00019999704873932447,
-      "loss": 46.0,
-      "step": 15196
-    },
-    {
-      "epoch": 2.4473610048713716,
-      "grad_norm": 0.006701333448290825,
-      "learning_rate": 0.0001999970483506318,
-      "loss": 46.0,
-      "step": 15197
-    },
-    {
-      "epoch": 2.447522041950159,
-      "grad_norm": 0.003331623040139675,
-      "learning_rate": 0.00019999704796191356,
-      "loss": 46.0,
-      "step": 15198
-    },
-    {
-      "epoch": 2.4476830790289466,
-      "grad_norm": 0.009729933924973011,
-      "learning_rate": 0.00019999704757316973,
-      "loss": 46.0,
-      "step": 15199
-    },
-    {
-      "epoch": 2.447844116107734,
-      "grad_norm": 0.007817571051418781,
-      "learning_rate": 0.00019999704718440028,
-      "loss": 46.0,
-      "step": 15200
-    },
-    {
-      "epoch": 2.448005153186521,
-      "grad_norm": 0.0011042491532862186,
-      "learning_rate": 0.00019999704679560525,
-      "loss": 46.0,
-      "step": 15201
-    },
-    {
-      "epoch": 2.4481661902653085,
-      "grad_norm": 0.008335348218679428,
-      "learning_rate": 0.00019999704640678463,
-      "loss": 46.0,
-      "step": 15202
-    },
-    {
-      "epoch": 2.448327227344096,
-      "grad_norm": 0.0063887545838952065,
-      "learning_rate": 0.00019999704601793842,
-      "loss": 46.0,
-      "step": 15203
-    },
-    {
-      "epoch": 2.4484882644228834,
-      "grad_norm": 0.001595465000718832,
-      "learning_rate": 0.0001999970456290666,
-      "loss": 46.0,
-      "step": 15204
-    },
-    {
-      "epoch": 2.448649301501671,
-      "grad_norm": 0.002345220185816288,
-      "learning_rate": 0.0001999970452401692,
-      "loss": 46.0,
-      "step": 15205
-    },
-    {
-      "epoch": 2.4488103385804583,
-      "grad_norm": 0.0016180694801732898,
-      "learning_rate": 0.0001999970448512462,
-      "loss": 46.0,
-      "step": 15206
-    },
-    {
-      "epoch": 2.4489713756592457,
-      "grad_norm": 0.002247192431241274,
-      "learning_rate": 0.0001999970444622976,
-      "loss": 46.0,
-      "step": 15207
-    },
-    {
-      "epoch": 2.4491324127380327,
-      "grad_norm": 0.000935447053052485,
-      "learning_rate": 0.0001999970440733234,
-      "loss": 46.0,
-      "step": 15208
-    },
-    {
-      "epoch": 2.44929344981682,
-      "grad_norm": 0.002954952185973525,
-      "learning_rate": 0.00019999704368432362,
-      "loss": 46.0,
-      "step": 15209
-    },
-    {
-      "epoch": 2.4494544868956076,
-      "grad_norm": 0.004415550269186497,
-      "learning_rate": 0.00019999704329529824,
-      "loss": 46.0,
-      "step": 15210
-    },
-    {
-      "epoch": 2.449615523974395,
-      "grad_norm": 0.0007914695888757706,
-      "learning_rate": 0.00019999704290624726,
-      "loss": 46.0,
-      "step": 15211
-    },
-    {
-      "epoch": 2.4497765610531825,
-      "grad_norm": 0.001555940369144082,
-      "learning_rate": 0.0001999970425171707,
-      "loss": 46.0,
-      "step": 15212
-    },
-    {
-      "epoch": 2.44993759813197,
-      "grad_norm": 0.0003372641222085804,
-      "learning_rate": 0.00019999704212806855,
-      "loss": 46.0,
-      "step": 15213
-    },
-    {
-      "epoch": 2.4500986352107574,
-      "grad_norm": 0.009569228626787663,
-      "learning_rate": 0.00019999704173894077,
-      "loss": 46.0,
-      "step": 15214
-    },
-    {
-      "epoch": 2.450259672289545,
-      "grad_norm": 0.003237672382965684,
-      "learning_rate": 0.00019999704134978746,
-      "loss": 46.0,
-      "step": 15215
-    },
-    {
-      "epoch": 2.450420709368332,
-      "grad_norm": 0.003576381830498576,
-      "learning_rate": 0.00019999704096060848,
-      "loss": 46.0,
-      "step": 15216
-    },
-    {
-      "epoch": 2.4505817464471193,
-      "grad_norm": 0.005659961141645908,
-      "learning_rate": 0.00019999704057140395,
-      "loss": 46.0,
-      "step": 15217
-    },
-    {
-      "epoch": 2.450742783525907,
-      "grad_norm": 0.004003557842224836,
-      "learning_rate": 0.00019999704018217385,
-      "loss": 46.0,
-      "step": 15218
-    },
-    {
-      "epoch": 2.4509038206046942,
-      "grad_norm": 0.006936076562851667,
-      "learning_rate": 0.0001999970397929181,
-      "loss": 46.0,
-      "step": 15219
-    },
-    {
-      "epoch": 2.4510648576834817,
-      "grad_norm": 0.0014235333073884249,
-      "learning_rate": 0.00019999703940363678,
-      "loss": 46.0,
-      "step": 15220
-    },
-    {
-      "epoch": 2.451225894762269,
-      "grad_norm": 0.010424616746604443,
-      "learning_rate": 0.00019999703901432987,
-      "loss": 46.0,
-      "step": 15221
-    },
-    {
-      "epoch": 2.451386931841056,
-      "grad_norm": 0.0011427258141338825,
-      "learning_rate": 0.00019999703862499737,
-      "loss": 46.0,
-      "step": 15222
-    },
-    {
-      "epoch": 2.4515479689198436,
-      "grad_norm": 0.002132270485162735,
-      "learning_rate": 0.00019999703823563925,
-      "loss": 46.0,
-      "step": 15223
-    },
-    {
-      "epoch": 2.451709005998631,
-      "grad_norm": 0.0016713673248887062,
-      "learning_rate": 0.00019999703784625555,
-      "loss": 46.0,
-      "step": 15224
-    },
-    {
-      "epoch": 2.4518700430774185,
-      "grad_norm": 0.005765865091234446,
-      "learning_rate": 0.0001999970374568463,
-      "loss": 46.0,
-      "step": 15225
-    },
-    {
-      "epoch": 2.452031080156206,
-      "grad_norm": 0.0059107705019414425,
-      "learning_rate": 0.00019999703706741139,
-      "loss": 46.0,
-      "step": 15226
-    },
-    {
-      "epoch": 2.4521921172349934,
-      "grad_norm": 0.0017577632097527385,
-      "learning_rate": 0.00019999703667795092,
-      "loss": 46.0,
-      "step": 15227
-    },
-    {
-      "epoch": 2.452353154313781,
-      "grad_norm": 0.0013335597468540072,
-      "learning_rate": 0.00019999703628846484,
-      "loss": 46.0,
-      "step": 15228
-    },
-    {
-      "epoch": 2.4525141913925683,
-      "grad_norm": 0.0015751039609313011,
-      "learning_rate": 0.00019999703589895318,
-      "loss": 46.0,
-      "step": 15229
-    },
-    {
-      "epoch": 2.4526752284713558,
-      "grad_norm": 0.004719140939414501,
-      "learning_rate": 0.00019999703550941593,
-      "loss": 46.0,
-      "step": 15230
-    },
-    {
-      "epoch": 2.452836265550143,
-      "grad_norm": 0.0012358779786154628,
-      "learning_rate": 0.00019999703511985306,
-      "loss": 46.0,
-      "step": 15231
-    },
-    {
-      "epoch": 2.4529973026289302,
-      "grad_norm": 0.0019273769576102495,
-      "learning_rate": 0.0001999970347302646,
-      "loss": 46.0,
-      "step": 15232
-    },
-    {
-      "epoch": 2.4531583397077177,
-      "grad_norm": 0.0045170956291258335,
-      "learning_rate": 0.0001999970343406506,
-      "loss": 46.0,
-      "step": 15233
-    },
-    {
-      "epoch": 2.453319376786505,
-      "grad_norm": 0.0035083426628261805,
-      "learning_rate": 0.00019999703395101094,
-      "loss": 46.0,
-      "step": 15234
-    },
-    {
-      "epoch": 2.4534804138652926,
-      "grad_norm": 0.0018365541473031044,
-      "learning_rate": 0.0001999970335613457,
-      "loss": 46.0,
-      "step": 15235
-    },
-    {
-      "epoch": 2.45364145094408,
-      "grad_norm": 0.003354813437908888,
-      "learning_rate": 0.0001999970331716549,
-      "loss": 46.0,
-      "step": 15236
-    },
-    {
-      "epoch": 2.453802488022867,
-      "grad_norm": 0.002219960791990161,
-      "learning_rate": 0.00019999703278193845,
-      "loss": 46.0,
-      "step": 15237
-    },
-    {
-      "epoch": 2.4539635251016545,
-      "grad_norm": 0.0006025100010447204,
-      "learning_rate": 0.00019999703239219644,
-      "loss": 46.0,
-      "step": 15238
-    },
-    {
-      "epoch": 2.454124562180442,
-      "grad_norm": 0.0017555751837790012,
-      "learning_rate": 0.00019999703200242883,
-      "loss": 46.0,
-      "step": 15239
-    },
-    {
-      "epoch": 2.4542855992592294,
-      "grad_norm": 0.0005677935550920665,
-      "learning_rate": 0.00019999703161263562,
-      "loss": 46.0,
-      "step": 15240
-    },
-    {
-      "epoch": 2.454446636338017,
-      "grad_norm": 0.0015791229670867324,
-      "learning_rate": 0.00019999703122281686,
-      "loss": 46.0,
-      "step": 15241
-    },
-    {
-      "epoch": 2.4546076734168043,
-      "grad_norm": 0.00508342869579792,
-      "learning_rate": 0.00019999703083297245,
-      "loss": 46.0,
-      "step": 15242
-    },
-    {
-      "epoch": 2.4547687104955918,
-      "grad_norm": 0.0024866480380296707,
-      "learning_rate": 0.00019999703044310248,
-      "loss": 46.0,
-      "step": 15243
-    },
-    {
-      "epoch": 2.454929747574379,
-      "grad_norm": 0.0010701797436922789,
-      "learning_rate": 0.0001999970300532069,
-      "loss": 46.0,
-      "step": 15244
-    },
-    {
-      "epoch": 2.455090784653166,
-      "grad_norm": 0.001672895741648972,
-      "learning_rate": 0.00019999702966328573,
-      "loss": 46.0,
-      "step": 15245
-    },
-    {
-      "epoch": 2.4552518217319537,
-      "grad_norm": 0.0010022710775956511,
-      "learning_rate": 0.00019999702927333895,
-      "loss": 46.0,
-      "step": 15246
-    },
-    {
-      "epoch": 2.455412858810741,
-      "grad_norm": 0.0030301359947770834,
-      "learning_rate": 0.00019999702888336658,
-      "loss": 46.0,
-      "step": 15247
-    },
-    {
-      "epoch": 2.4555738958895286,
-      "grad_norm": 0.0033945185132324696,
-      "learning_rate": 0.00019999702849336862,
-      "loss": 46.0,
-      "step": 15248
-    },
-    {
-      "epoch": 2.455734932968316,
-      "grad_norm": 0.0018690304132178426,
-      "learning_rate": 0.00019999702810334508,
-      "loss": 46.0,
-      "step": 15249
-    },
-    {
-      "epoch": 2.4558959700471035,
-      "grad_norm": 0.006364199798554182,
-      "learning_rate": 0.00019999702771329595,
-      "loss": 46.0,
-      "step": 15250
-    },
-    {
-      "epoch": 2.456057007125891,
-      "grad_norm": 0.002885845024138689,
-      "learning_rate": 0.0001999970273232212,
-      "loss": 46.0,
-      "step": 15251
-    },
-    {
-      "epoch": 2.456218044204678,
-      "grad_norm": 0.0018506854539737105,
-      "learning_rate": 0.00019999702693312087,
-      "loss": 46.0,
-      "step": 15252
-    },
-    {
-      "epoch": 2.4563790812834654,
-      "grad_norm": 0.002193527529016137,
-      "learning_rate": 0.00019999702654299495,
-      "loss": 46.0,
-      "step": 15253
-    },
-    {
-      "epoch": 2.456540118362253,
-      "grad_norm": 0.008521744050085545,
-      "learning_rate": 0.00019999702615284344,
-      "loss": 46.0,
-      "step": 15254
-    },
-    {
-      "epoch": 2.4567011554410403,
-      "grad_norm": 0.0013411276740953326,
-      "learning_rate": 0.00019999702576266632,
-      "loss": 46.0,
-      "step": 15255
-    },
-    {
-      "epoch": 2.4568621925198277,
-      "grad_norm": 0.0075626433826982975,
-      "learning_rate": 0.00019999702537246361,
-      "loss": 46.0,
-      "step": 15256
-    },
-    {
-      "epoch": 2.457023229598615,
-      "grad_norm": 0.002869025571271777,
-      "learning_rate": 0.00019999702498223532,
-      "loss": 46.0,
-      "step": 15257
-    },
-    {
-      "epoch": 2.4571842666774026,
-      "grad_norm": 0.00453701987862587,
-      "learning_rate": 0.0001999970245919814,
-      "loss": 46.0,
-      "step": 15258
-    },
-    {
-      "epoch": 2.45734530375619,
-      "grad_norm": 0.0061545055359601974,
-      "learning_rate": 0.0001999970242017019,
-      "loss": 46.0,
-      "step": 15259
-    },
-    {
-      "epoch": 2.457506340834977,
-      "grad_norm": 0.0009747164440341294,
-      "learning_rate": 0.00019999702381139685,
-      "loss": 46.0,
-      "step": 15260
-    },
-    {
-      "epoch": 2.4576673779137646,
-      "grad_norm": 0.0009364444995298982,
-      "learning_rate": 0.00019999702342106616,
-      "loss": 46.0,
-      "step": 15261
-    },
-    {
-      "epoch": 2.457828414992552,
-      "grad_norm": 0.0016260718693956733,
-      "learning_rate": 0.0001999970230307099,
-      "loss": 46.0,
-      "step": 15262
-    },
-    {
-      "epoch": 2.4579894520713395,
-      "grad_norm": 0.0007474543526768684,
-      "learning_rate": 0.00019999702264032802,
-      "loss": 46.0,
-      "step": 15263
-    },
-    {
-      "epoch": 2.458150489150127,
-      "grad_norm": 0.00042786996345967054,
-      "learning_rate": 0.00019999702224992056,
-      "loss": 46.0,
-      "step": 15264
-    },
-    {
-      "epoch": 2.4583115262289144,
-      "grad_norm": 0.0020948343444615602,
-      "learning_rate": 0.0001999970218594875,
-      "loss": 46.0,
-      "step": 15265
-    },
-    {
-      "epoch": 2.4584725633077014,
-      "grad_norm": 0.004851227160543203,
-      "learning_rate": 0.00019999702146902886,
-      "loss": 46.0,
-      "step": 15266
-    },
-    {
-      "epoch": 2.458633600386489,
-      "grad_norm": 0.005663125775754452,
-      "learning_rate": 0.00019999702107854463,
-      "loss": 46.0,
-      "step": 15267
-    },
-    {
-      "epoch": 2.4587946374652763,
-      "grad_norm": 0.002608599839732051,
-      "learning_rate": 0.0001999970206880348,
-      "loss": 46.0,
-      "step": 15268
-    },
-    {
-      "epoch": 2.4589556745440637,
-      "grad_norm": 0.005585620645433664,
-      "learning_rate": 0.00019999702029749935,
-      "loss": 46.0,
-      "step": 15269
-    },
-    {
-      "epoch": 2.459116711622851,
-      "grad_norm": 0.001136680832132697,
-      "learning_rate": 0.00019999701990693834,
-      "loss": 46.0,
-      "step": 15270
-    },
-    {
-      "epoch": 2.4592777487016386,
-      "grad_norm": 0.0014970415504649282,
-      "learning_rate": 0.0001999970195163517,
-      "loss": 46.0,
-      "step": 15271
-    },
-    {
-      "epoch": 2.459438785780426,
-      "grad_norm": 0.025296634063124657,
-      "learning_rate": 0.00019999701912573953,
-      "loss": 46.0,
-      "step": 15272
-    },
-    {
-      "epoch": 2.4595998228592135,
-      "grad_norm": 0.0009249933646060526,
-      "learning_rate": 0.00019999701873510168,
-      "loss": 46.0,
-      "step": 15273
-    },
-    {
-      "epoch": 2.4597608599380005,
-      "grad_norm": 0.0022794988472014666,
-      "learning_rate": 0.0001999970183444383,
-      "loss": 46.0,
-      "step": 15274
-    },
-    {
-      "epoch": 2.459921897016788,
-      "grad_norm": 0.0019057078752666712,
-      "learning_rate": 0.0001999970179537493,
-      "loss": 46.0,
-      "step": 15275
-    },
-    {
-      "epoch": 2.4600829340955754,
-      "grad_norm": 0.0028241551481187344,
-      "learning_rate": 0.0001999970175630347,
-      "loss": 46.0,
-      "step": 15276
-    },
-    {
-      "epoch": 2.460243971174363,
-      "grad_norm": 0.000636660901363939,
-      "learning_rate": 0.0001999970171722945,
-      "loss": 46.0,
-      "step": 15277
-    },
-    {
-      "epoch": 2.4604050082531503,
-      "grad_norm": 0.001617730362340808,
-      "learning_rate": 0.00019999701678152874,
-      "loss": 46.0,
-      "step": 15278
-    },
-    {
-      "epoch": 2.460566045331938,
-      "grad_norm": 0.0014377145562320948,
-      "learning_rate": 0.0001999970163907374,
-      "loss": 46.0,
-      "step": 15279
-    },
-    {
-      "epoch": 2.4607270824107252,
-      "grad_norm": 0.00459147896617651,
-      "learning_rate": 0.0001999970159999204,
-      "loss": 46.0,
-      "step": 15280
-    },
-    {
-      "epoch": 2.4608881194895122,
-      "grad_norm": 0.0019778336863964796,
-      "learning_rate": 0.00019999701560907785,
-      "loss": 46.0,
-      "step": 15281
-    },
-    {
-      "epoch": 2.4610491565682997,
-      "grad_norm": 0.002988354302942753,
-      "learning_rate": 0.0001999970152182097,
-      "loss": 46.0,
-      "step": 15282
-    },
-    {
-      "epoch": 2.461210193647087,
-      "grad_norm": 0.0014327128883451223,
-      "learning_rate": 0.00019999701482731593,
-      "loss": 46.0,
-      "step": 15283
-    },
-    {
-      "epoch": 2.4613712307258746,
-      "grad_norm": 0.00252849911339581,
-      "learning_rate": 0.0001999970144363966,
-      "loss": 46.0,
-      "step": 15284
-    },
-    {
-      "epoch": 2.461532267804662,
-      "grad_norm": 0.0063990941271185875,
-      "learning_rate": 0.00019999701404545167,
-      "loss": 46.0,
-      "step": 15285
-    },
-    {
-      "epoch": 2.4616933048834495,
-      "grad_norm": 0.0021387904416769743,
-      "learning_rate": 0.00019999701365448113,
-      "loss": 46.0,
-      "step": 15286
-    },
-    {
-      "epoch": 2.461854341962237,
-      "grad_norm": 0.0062814331613481045,
-      "learning_rate": 0.000199997013263485,
-      "loss": 46.0,
-      "step": 15287
-    },
-    {
-      "epoch": 2.4620153790410244,
-      "grad_norm": 0.0012795914663001895,
-      "learning_rate": 0.0001999970128724633,
-      "loss": 46.0,
-      "step": 15288
-    },
-    {
-      "epoch": 2.4621764161198114,
-      "grad_norm": 0.00279766833409667,
-      "learning_rate": 0.00019999701248141598,
-      "loss": 46.0,
-      "step": 15289
-    },
-    {
-      "epoch": 2.462337453198599,
-      "grad_norm": 0.00167724653147161,
-      "learning_rate": 0.00019999701209034307,
-      "loss": 46.0,
-      "step": 15290
-    },
-    {
-      "epoch": 2.4624984902773863,
-      "grad_norm": 0.0009418915724381804,
-      "learning_rate": 0.0001999970116992446,
-      "loss": 46.0,
-      "step": 15291
-    },
-    {
-      "epoch": 2.4626595273561738,
-      "grad_norm": 0.0020624124445021152,
-      "learning_rate": 0.00019999701130812047,
-      "loss": 46.0,
-      "step": 15292
-    },
-    {
-      "epoch": 2.4628205644349612,
-      "grad_norm": 0.0005897238734178245,
-      "learning_rate": 0.0001999970109169708,
-      "loss": 46.0,
-      "step": 15293
-    },
-    {
-      "epoch": 2.4629816015137487,
-      "grad_norm": 0.0012851186329498887,
-      "learning_rate": 0.0001999970105257955,
-      "loss": 46.0,
-      "step": 15294
-    },
-    {
-      "epoch": 2.4631426385925357,
-      "grad_norm": 0.005485603120177984,
-      "learning_rate": 0.00019999701013459462,
-      "loss": 46.0,
-      "step": 15295
-    },
-    {
-      "epoch": 2.463303675671323,
-      "grad_norm": 0.008505761623382568,
-      "learning_rate": 0.00019999700974336816,
-      "loss": 46.0,
-      "step": 15296
-    },
-    {
-      "epoch": 2.4634647127501106,
-      "grad_norm": 0.0012100688181817532,
-      "learning_rate": 0.0001999970093521161,
-      "loss": 46.0,
-      "step": 15297
-    },
-    {
-      "epoch": 2.463625749828898,
-      "grad_norm": 0.0034214206971228123,
-      "learning_rate": 0.00019999700896083844,
-      "loss": 46.0,
-      "step": 15298
-    },
-    {
-      "epoch": 2.4637867869076855,
-      "grad_norm": 0.0004850381810683757,
-      "learning_rate": 0.00019999700856953518,
-      "loss": 46.0,
-      "step": 15299
-    },
-    {
-      "epoch": 2.463947823986473,
-      "grad_norm": 0.004832991864532232,
-      "learning_rate": 0.0001999970081782063,
-      "loss": 46.0,
-      "step": 15300
-    },
-    {
-      "epoch": 2.4641088610652604,
-      "grad_norm": 0.0006791831692680717,
-      "learning_rate": 0.00019999700778685188,
-      "loss": 46.0,
-      "step": 15301
-    },
-    {
-      "epoch": 2.464269898144048,
-      "grad_norm": 0.005785138346254826,
-      "learning_rate": 0.00019999700739547187,
-      "loss": 46.0,
-      "step": 15302
-    },
-    {
-      "epoch": 2.4644309352228353,
-      "grad_norm": 0.0032062644604593515,
-      "learning_rate": 0.00019999700700406623,
-      "loss": 46.0,
-      "step": 15303
-    },
-    {
-      "epoch": 2.4645919723016223,
-      "grad_norm": 0.002940522972494364,
-      "learning_rate": 0.00019999700661263502,
-      "loss": 46.0,
-      "step": 15304
-    },
-    {
-      "epoch": 2.4647530093804098,
-      "grad_norm": 0.004605089779943228,
-      "learning_rate": 0.00019999700622117818,
-      "loss": 46.0,
-      "step": 15305
-    },
-    {
-      "epoch": 2.464914046459197,
-      "grad_norm": 0.001077226479537785,
-      "learning_rate": 0.0001999970058296958,
-      "loss": 46.0,
-      "step": 15306
-    },
-    {
-      "epoch": 2.4650750835379847,
-      "grad_norm": 0.011306436732411385,
-      "learning_rate": 0.00019999700543818776,
-      "loss": 46.0,
-      "step": 15307
-    },
-    {
-      "epoch": 2.465236120616772,
-      "grad_norm": 0.012863697484135628,
-      "learning_rate": 0.00019999700504665414,
-      "loss": 46.0,
-      "step": 15308
-    },
-    {
-      "epoch": 2.4653971576955596,
-      "grad_norm": 0.0012285165721550584,
-      "learning_rate": 0.00019999700465509496,
-      "loss": 46.0,
-      "step": 15309
-    },
-    {
-      "epoch": 2.4655581947743466,
-      "grad_norm": 0.003248112043365836,
-      "learning_rate": 0.0001999970042635102,
-      "loss": 46.0,
-      "step": 15310
-    },
-    {
-      "epoch": 2.465719231853134,
-      "grad_norm": 0.0025695119984447956,
-      "learning_rate": 0.00019999700387189978,
-      "loss": 46.0,
-      "step": 15311
-    },
-    {
-      "epoch": 2.4658802689319215,
-      "grad_norm": 0.003024042584002018,
-      "learning_rate": 0.0001999970034802638,
-      "loss": 46.0,
-      "step": 15312
-    },
-    {
-      "epoch": 2.466041306010709,
-      "grad_norm": 0.0055190748535096645,
-      "learning_rate": 0.00019999700308860225,
-      "loss": 46.0,
-      "step": 15313
-    },
-    {
-      "epoch": 2.4662023430894964,
-      "grad_norm": 0.001415694016031921,
-      "learning_rate": 0.00019999700269691508,
-      "loss": 46.0,
-      "step": 15314
-    },
-    {
-      "epoch": 2.466363380168284,
-      "grad_norm": 0.004782662261277437,
-      "learning_rate": 0.00019999700230520232,
-      "loss": 46.0,
-      "step": 15315
-    },
-    {
-      "epoch": 2.4665244172470713,
-      "grad_norm": 0.0019428686937317252,
-      "learning_rate": 0.00019999700191346395,
-      "loss": 46.0,
-      "step": 15316
-    },
-    {
-      "epoch": 2.4666854543258587,
-      "grad_norm": 0.0031769087072461843,
-      "learning_rate": 0.00019999700152170002,
-      "loss": 46.0,
-      "step": 15317
-    },
-    {
-      "epoch": 2.4668464914046457,
-      "grad_norm": 0.003012104192748666,
-      "learning_rate": 0.00019999700112991047,
-      "loss": 46.0,
-      "step": 15318
-    },
-    {
-      "epoch": 2.467007528483433,
-      "grad_norm": 0.002719815354794264,
-      "learning_rate": 0.00019999700073809537,
-      "loss": 46.0,
-      "step": 15319
-    },
-    {
-      "epoch": 2.4671685655622206,
-      "grad_norm": 0.0027671062853187323,
-      "learning_rate": 0.00019999700034625462,
-      "loss": 46.0,
-      "step": 15320
-    },
-    {
-      "epoch": 2.467329602641008,
-      "grad_norm": 0.0009664149256423116,
-      "learning_rate": 0.0001999969999543883,
-      "loss": 46.0,
-      "step": 15321
-    },
-    {
-      "epoch": 2.4674906397197955,
-      "grad_norm": 0.0020425315015017986,
-      "learning_rate": 0.00019999699956249636,
-      "loss": 46.0,
-      "step": 15322
-    },
-    {
-      "epoch": 2.467651676798583,
-      "grad_norm": 0.008751987479627132,
-      "learning_rate": 0.00019999699917057885,
-      "loss": 46.0,
-      "step": 15323
-    },
-    {
-      "epoch": 2.4678127138773704,
-      "grad_norm": 0.0017249657539650798,
-      "learning_rate": 0.00019999699877863575,
-      "loss": 46.0,
-      "step": 15324
-    },
-    {
-      "epoch": 2.4679737509561575,
-      "grad_norm": 0.0011153070954605937,
-      "learning_rate": 0.00019999699838666707,
-      "loss": 46.0,
-      "step": 15325
-    },
-    {
-      "epoch": 2.468134788034945,
-      "grad_norm": 0.0008131980430334806,
-      "learning_rate": 0.00019999699799467274,
-      "loss": 46.0,
-      "step": 15326
-    },
-    {
-      "epoch": 2.4682958251137324,
-      "grad_norm": 0.0010892102727666497,
-      "learning_rate": 0.00019999699760265286,
-      "loss": 46.0,
-      "step": 15327
-    },
-    {
-      "epoch": 2.46845686219252,
-      "grad_norm": 0.0010309504577890038,
-      "learning_rate": 0.00019999699721060738,
-      "loss": 46.0,
-      "step": 15328
-    },
-    {
-      "epoch": 2.4686178992713073,
-      "grad_norm": 0.003638485912233591,
-      "learning_rate": 0.00019999699681853632,
-      "loss": 46.0,
-      "step": 15329
-    },
-    {
-      "epoch": 2.4687789363500947,
-      "grad_norm": 0.0011200020089745522,
-      "learning_rate": 0.00019999699642643965,
-      "loss": 46.0,
-      "step": 15330
-    },
-    {
-      "epoch": 2.468939973428882,
-      "grad_norm": 0.007710890378803015,
-      "learning_rate": 0.00019999699603431736,
-      "loss": 46.0,
-      "step": 15331
-    },
-    {
-      "epoch": 2.4691010105076696,
-      "grad_norm": 0.007329565938562155,
-      "learning_rate": 0.00019999699564216951,
-      "loss": 46.0,
-      "step": 15332
-    },
-    {
-      "epoch": 2.4692620475864566,
-      "grad_norm": 0.0017656265990808606,
-      "learning_rate": 0.00019999699524999608,
-      "loss": 46.0,
-      "step": 15333
-    },
-    {
-      "epoch": 2.469423084665244,
-      "grad_norm": 0.0009448722121305764,
-      "learning_rate": 0.00019999699485779703,
-      "loss": 46.0,
-      "step": 15334
-    },
-    {
-      "epoch": 2.4695841217440315,
-      "grad_norm": 0.0010442073689773679,
-      "learning_rate": 0.00019999699446557236,
-      "loss": 46.0,
-      "step": 15335
-    },
-    {
-      "epoch": 2.469745158822819,
-      "grad_norm": 0.009211316704750061,
-      "learning_rate": 0.00019999699407332214,
-      "loss": 46.0,
-      "step": 15336
-    },
-    {
-      "epoch": 2.4699061959016064,
-      "grad_norm": 0.007101106457412243,
-      "learning_rate": 0.00019999699368104633,
-      "loss": 46.0,
-      "step": 15337
-    },
-    {
-      "epoch": 2.470067232980394,
-      "grad_norm": 0.005491578485816717,
-      "learning_rate": 0.00019999699328874488,
-      "loss": 46.0,
-      "step": 15338
-    },
-    {
-      "epoch": 2.470228270059181,
-      "grad_norm": 0.007363234180957079,
-      "learning_rate": 0.00019999699289641786,
-      "loss": 46.0,
-      "step": 15339
-    },
-    {
-      "epoch": 2.4703893071379683,
-      "grad_norm": 0.001275977585464716,
-      "learning_rate": 0.00019999699250406526,
-      "loss": 46.0,
-      "step": 15340
-    },
-    {
-      "epoch": 2.470550344216756,
-      "grad_norm": 0.005832497496157885,
-      "learning_rate": 0.00019999699211168705,
-      "loss": 46.0,
-      "step": 15341
-    },
-    {
-      "epoch": 2.4707113812955432,
-      "grad_norm": 0.0006780567346140742,
-      "learning_rate": 0.00019999699171928325,
-      "loss": 46.0,
-      "step": 15342
-    },
-    {
-      "epoch": 2.4708724183743307,
-      "grad_norm": 0.0016721902647987008,
-      "learning_rate": 0.00019999699132685386,
-      "loss": 46.0,
-      "step": 15343
-    },
-    {
-      "epoch": 2.471033455453118,
-      "grad_norm": 0.0041099428199231625,
-      "learning_rate": 0.00019999699093439888,
-      "loss": 46.0,
-      "step": 15344
-    },
-    {
-      "epoch": 2.4711944925319056,
-      "grad_norm": 0.011762848123908043,
-      "learning_rate": 0.0001999969905419183,
-      "loss": 46.0,
-      "step": 15345
-    },
-    {
-      "epoch": 2.471355529610693,
-      "grad_norm": 0.0023676606360822916,
-      "learning_rate": 0.00019999699014941211,
-      "loss": 46.0,
-      "step": 15346
-    },
-    {
-      "epoch": 2.4715165666894805,
-      "grad_norm": 0.000821719819214195,
-      "learning_rate": 0.00019999698975688035,
-      "loss": 46.0,
-      "step": 15347
-    },
-    {
-      "epoch": 2.4716776037682675,
-      "grad_norm": 0.0027383260894566774,
-      "learning_rate": 0.00019999698936432297,
-      "loss": 46.0,
-      "step": 15348
-    },
-    {
-      "epoch": 2.471838640847055,
-      "grad_norm": 0.005373249761760235,
-      "learning_rate": 0.00019999698897174,
-      "loss": 46.0,
-      "step": 15349
-    },
-    {
-      "epoch": 2.4719996779258424,
-      "grad_norm": 0.0031262037809938192,
-      "learning_rate": 0.00019999698857913148,
-      "loss": 46.0,
-      "step": 15350
-    },
-    {
-      "epoch": 2.47216071500463,
-      "grad_norm": 0.002600323176011443,
-      "learning_rate": 0.0001999969881864973,
-      "loss": 46.0,
-      "step": 15351
-    },
-    {
-      "epoch": 2.4723217520834173,
-      "grad_norm": 0.0008210457162931561,
-      "learning_rate": 0.00019999698779383758,
-      "loss": 46.0,
-      "step": 15352
-    },
-    {
-      "epoch": 2.4724827891622048,
-      "grad_norm": 0.0038396476302295923,
-      "learning_rate": 0.00019999698740115224,
-      "loss": 46.0,
-      "step": 15353
-    },
-    {
-      "epoch": 2.4726438262409918,
-      "grad_norm": 0.004013812635093927,
-      "learning_rate": 0.00019999698700844131,
-      "loss": 46.0,
-      "step": 15354
-    },
-    {
-      "epoch": 2.4728048633197792,
-      "grad_norm": 0.0030101195443421602,
-      "learning_rate": 0.00019999698661570477,
-      "loss": 46.0,
-      "step": 15355
-    },
-    {
-      "epoch": 2.4729659003985667,
-      "grad_norm": 0.003551741596311331,
-      "learning_rate": 0.00019999698622294267,
-      "loss": 46.0,
-      "step": 15356
-    },
-    {
-      "epoch": 2.473126937477354,
-      "grad_norm": 0.006579568609595299,
-      "learning_rate": 0.00019999698583015495,
-      "loss": 46.0,
-      "step": 15357
-    },
-    {
-      "epoch": 2.4732879745561416,
-      "grad_norm": 0.0028752542566508055,
-      "learning_rate": 0.00019999698543734165,
-      "loss": 46.0,
-      "step": 15358
-    },
-    {
-      "epoch": 2.473449011634929,
-      "grad_norm": 0.003337211674079299,
-      "learning_rate": 0.00019999698504450275,
-      "loss": 46.0,
-      "step": 15359
-    },
-    {
-      "epoch": 2.4736100487137165,
-      "grad_norm": 0.004037004429847002,
-      "learning_rate": 0.00019999698465163825,
-      "loss": 46.0,
-      "step": 15360
-    },
-    {
-      "epoch": 2.473771085792504,
-      "grad_norm": 0.0017564335139468312,
-      "learning_rate": 0.00019999698425874818,
-      "loss": 46.0,
-      "step": 15361
-    },
-    {
-      "epoch": 2.473932122871291,
-      "grad_norm": 0.0026432075537741184,
-      "learning_rate": 0.00019999698386583247,
-      "loss": 46.0,
-      "step": 15362
-    },
-    {
-      "epoch": 2.4740931599500784,
-      "grad_norm": 0.006666726898401976,
-      "learning_rate": 0.0001999969834728912,
-      "loss": 46.0,
-      "step": 15363
-    },
-    {
-      "epoch": 2.474254197028866,
-      "grad_norm": 0.0017813860904425383,
-      "learning_rate": 0.00019999698307992435,
-      "loss": 46.0,
-      "step": 15364
-    },
-    {
-      "epoch": 2.4744152341076533,
-      "grad_norm": 0.0011548922630026937,
-      "learning_rate": 0.00019999698268693188,
-      "loss": 46.0,
-      "step": 15365
-    },
-    {
-      "epoch": 2.4745762711864407,
-      "grad_norm": 0.01014232449233532,
-      "learning_rate": 0.00019999698229391383,
-      "loss": 46.0,
-      "step": 15366
-    },
-    {
-      "epoch": 2.474737308265228,
-      "grad_norm": 0.0008472424815408885,
-      "learning_rate": 0.00019999698190087016,
-      "loss": 46.0,
-      "step": 15367
-    },
-    {
-      "epoch": 2.4748983453440156,
-      "grad_norm": 0.0013406119542196393,
-      "learning_rate": 0.00019999698150780093,
-      "loss": 46.0,
-      "step": 15368
-    },
-    {
-      "epoch": 2.4750593824228027,
-      "grad_norm": 0.002712155692279339,
-      "learning_rate": 0.00019999698111470608,
-      "loss": 46.0,
-      "step": 15369
-    },
-    {
-      "epoch": 2.47522041950159,
-      "grad_norm": 0.0007146503776311874,
-      "learning_rate": 0.00019999698072158565,
-      "loss": 46.0,
-      "step": 15370
-    },
-    {
-      "epoch": 2.4753814565803776,
-      "grad_norm": 0.0066480678506195545,
-      "learning_rate": 0.0001999969803284396,
-      "loss": 46.0,
-      "step": 15371
-    },
-    {
-      "epoch": 2.475542493659165,
-      "grad_norm": 0.0040450431406497955,
-      "learning_rate": 0.00019999697993526797,
-      "loss": 46.0,
-      "step": 15372
-    },
-    {
-      "epoch": 2.4757035307379525,
-      "grad_norm": 0.0003057619323953986,
-      "learning_rate": 0.00019999697954207077,
-      "loss": 46.0,
-      "step": 15373
-    },
-    {
-      "epoch": 2.47586456781674,
-      "grad_norm": 0.0017980325501412153,
-      "learning_rate": 0.00019999697914884797,
-      "loss": 46.0,
-      "step": 15374
-    },
-    {
-      "epoch": 2.4760256048955274,
-      "grad_norm": 0.0051462966948747635,
-      "learning_rate": 0.00019999697875559954,
-      "loss": 46.0,
-      "step": 15375
-    },
-    {
-      "epoch": 2.476186641974315,
-      "grad_norm": 0.0020662748720496893,
-      "learning_rate": 0.00019999697836232556,
-      "loss": 46.0,
-      "step": 15376
-    },
-    {
-      "epoch": 2.476347679053102,
-      "grad_norm": 0.0019423920894041657,
-      "learning_rate": 0.00019999697796902597,
-      "loss": 46.0,
-      "step": 15377
-    },
-    {
-      "epoch": 2.4765087161318893,
-      "grad_norm": 0.002050557406619191,
-      "learning_rate": 0.00019999697757570078,
-      "loss": 46.0,
-      "step": 15378
-    },
-    {
-      "epoch": 2.4766697532106767,
-      "grad_norm": 0.008099086582660675,
-      "learning_rate": 0.00019999697718234998,
-      "loss": 46.0,
-      "step": 15379
-    },
-    {
-      "epoch": 2.476830790289464,
-      "grad_norm": 0.011153250932693481,
-      "learning_rate": 0.00019999697678897363,
-      "loss": 46.0,
-      "step": 15380
-    },
-    {
-      "epoch": 2.4769918273682516,
-      "grad_norm": 0.004111459478735924,
-      "learning_rate": 0.00019999697639557165,
-      "loss": 46.0,
-      "step": 15381
-    },
-    {
-      "epoch": 2.477152864447039,
-      "grad_norm": 0.002271935110911727,
-      "learning_rate": 0.0001999969760021441,
-      "loss": 46.0,
-      "step": 15382
-    },
-    {
-      "epoch": 2.477313901525826,
-      "grad_norm": 0.0054304893128573895,
-      "learning_rate": 0.00019999697560869092,
-      "loss": 46.0,
-      "step": 15383
-    },
-    {
-      "epoch": 2.4774749386046135,
-      "grad_norm": 0.007937449030578136,
-      "learning_rate": 0.00019999697521521216,
-      "loss": 46.0,
-      "step": 15384
-    },
-    {
-      "epoch": 2.477635975683401,
-      "grad_norm": 0.0027499040588736534,
-      "learning_rate": 0.00019999697482170784,
-      "loss": 46.0,
-      "step": 15385
-    },
-    {
-      "epoch": 2.4777970127621884,
-      "grad_norm": 0.00043278661905787885,
-      "learning_rate": 0.00019999697442817788,
-      "loss": 46.0,
-      "step": 15386
-    },
-    {
-      "epoch": 2.477958049840976,
-      "grad_norm": 0.0015744095435366035,
-      "learning_rate": 0.00019999697403462235,
-      "loss": 46.0,
-      "step": 15387
-    },
-    {
-      "epoch": 2.4781190869197633,
-      "grad_norm": 0.0023089037276804447,
-      "learning_rate": 0.00019999697364104125,
-      "loss": 46.0,
-      "step": 15388
-    },
-    {
-      "epoch": 2.478280123998551,
-      "grad_norm": 0.004893219098448753,
-      "learning_rate": 0.0001999969732474345,
-      "loss": 46.0,
-      "step": 15389
-    },
-    {
-      "epoch": 2.4784411610773383,
-      "grad_norm": 0.002226802986115217,
-      "learning_rate": 0.00019999697285380218,
-      "loss": 46.0,
-      "step": 15390
-    },
-    {
-      "epoch": 2.4786021981561253,
-      "grad_norm": 0.0015342776896432042,
-      "learning_rate": 0.00019999697246014426,
-      "loss": 46.0,
-      "step": 15391
-    },
-    {
-      "epoch": 2.4787632352349127,
-      "grad_norm": 0.0013666960876435041,
-      "learning_rate": 0.00019999697206646077,
-      "loss": 46.0,
-      "step": 15392
-    },
-    {
-      "epoch": 2.4789242723137,
-      "grad_norm": 0.006661290302872658,
-      "learning_rate": 0.00019999697167275167,
-      "loss": 46.0,
-      "step": 15393
-    },
-    {
-      "epoch": 2.4790853093924876,
-      "grad_norm": 0.003398258937522769,
-      "learning_rate": 0.000199996971279017,
-      "loss": 46.0,
-      "step": 15394
-    },
-    {
-      "epoch": 2.479246346471275,
-      "grad_norm": 0.003983153495937586,
-      "learning_rate": 0.00019999697088525669,
-      "loss": 46.0,
-      "step": 15395
-    },
-    {
-      "epoch": 2.4794073835500625,
-      "grad_norm": 0.0010558993089944124,
-      "learning_rate": 0.00019999697049147082,
-      "loss": 46.0,
-      "step": 15396
-    },
-    {
-      "epoch": 2.47956842062885,
-      "grad_norm": 0.001031601568683982,
-      "learning_rate": 0.00019999697009765935,
-      "loss": 46.0,
-      "step": 15397
-    },
-    {
-      "epoch": 2.479729457707637,
-      "grad_norm": 0.0014107549795880914,
-      "learning_rate": 0.00019999696970382226,
-      "loss": 46.0,
-      "step": 15398
-    },
-    {
-      "epoch": 2.4798904947864244,
-      "grad_norm": 0.0007549440488219261,
-      "learning_rate": 0.0001999969693099596,
-      "loss": 46.0,
-      "step": 15399
-    },
-    {
-      "epoch": 2.480051531865212,
-      "grad_norm": 0.005329488310962915,
-      "learning_rate": 0.00019999696891607135,
-      "loss": 46.0,
-      "step": 15400
-    },
-    {
-      "epoch": 2.4802125689439993,
-      "grad_norm": 0.00309107918292284,
-      "learning_rate": 0.00019999696852215747,
-      "loss": 46.0,
-      "step": 15401
-    },
-    {
-      "epoch": 2.480373606022787,
-      "grad_norm": 0.0018073205137625337,
-      "learning_rate": 0.00019999696812821806,
-      "loss": 46.0,
-      "step": 15402
-    },
-    {
-      "epoch": 2.4805346431015742,
-      "grad_norm": 0.0013899158220738173,
-      "learning_rate": 0.000199996967734253,
-      "loss": 46.0,
-      "step": 15403
-    },
-    {
-      "epoch": 2.4806956801803617,
-      "grad_norm": 0.0008333500591106713,
-      "learning_rate": 0.00019999696734026236,
-      "loss": 46.0,
-      "step": 15404
-    },
-    {
-      "epoch": 2.480856717259149,
-      "grad_norm": 0.004558071959763765,
-      "learning_rate": 0.00019999696694624614,
-      "loss": 46.0,
-      "step": 15405
-    },
-    {
-      "epoch": 2.481017754337936,
-      "grad_norm": 0.007042016834020615,
-      "learning_rate": 0.00019999696655220432,
-      "loss": 46.0,
-      "step": 15406
-    },
-    {
-      "epoch": 2.4811787914167236,
-      "grad_norm": 0.0023991644848138094,
-      "learning_rate": 0.0001999969661581369,
-      "loss": 46.0,
-      "step": 15407
-    },
-    {
-      "epoch": 2.481339828495511,
-      "grad_norm": 0.01589331403374672,
-      "learning_rate": 0.0001999969657640439,
-      "loss": 46.0,
-      "step": 15408
-    },
-    {
-      "epoch": 2.4815008655742985,
-      "grad_norm": 0.004673053044825792,
-      "learning_rate": 0.0001999969653699253,
-      "loss": 46.0,
-      "step": 15409
-    },
-    {
-      "epoch": 2.481661902653086,
-      "grad_norm": 0.004645536653697491,
-      "learning_rate": 0.00019999696497578109,
-      "loss": 46.0,
-      "step": 15410
-    },
-    {
-      "epoch": 2.4818229397318734,
-      "grad_norm": 0.0024915996473282576,
-      "learning_rate": 0.0001999969645816113,
-      "loss": 46.0,
-      "step": 15411
-    },
-    {
-      "epoch": 2.4819839768106604,
-      "grad_norm": 0.0017545539885759354,
-      "learning_rate": 0.00019999696418741592,
-      "loss": 46.0,
-      "step": 15412
-    },
-    {
-      "epoch": 2.482145013889448,
-      "grad_norm": 0.002297217957675457,
-      "learning_rate": 0.0001999969637931949,
-      "loss": 46.0,
-      "step": 15413
-    },
-    {
-      "epoch": 2.4823060509682353,
-      "grad_norm": 0.007391363848000765,
-      "learning_rate": 0.00019999696339894835,
-      "loss": 46.0,
-      "step": 15414
-    },
-    {
-      "epoch": 2.4824670880470228,
-      "grad_norm": 0.0032937314826995134,
-      "learning_rate": 0.0001999969630046762,
-      "loss": 46.0,
-      "step": 15415
-    },
-    {
-      "epoch": 2.48262812512581,
-      "grad_norm": 0.0007750993245281279,
-      "learning_rate": 0.00019999696261037843,
-      "loss": 46.0,
-      "step": 15416
-    },
-    {
-      "epoch": 2.4827891622045977,
-      "grad_norm": 0.00166051066480577,
-      "learning_rate": 0.00019999696221605507,
-      "loss": 46.0,
-      "step": 15417
-    },
-    {
-      "epoch": 2.482950199283385,
-      "grad_norm": 0.006945026572793722,
-      "learning_rate": 0.00019999696182170613,
-      "loss": 46.0,
-      "step": 15418
-    },
-    {
-      "epoch": 2.4831112363621726,
-      "grad_norm": 0.002210700884461403,
-      "learning_rate": 0.00019999696142733155,
-      "loss": 46.0,
-      "step": 15419
-    },
-    {
-      "epoch": 2.48327227344096,
-      "grad_norm": 0.005010916851460934,
-      "learning_rate": 0.00019999696103293143,
-      "loss": 46.0,
-      "step": 15420
-    },
-    {
-      "epoch": 2.483433310519747,
-      "grad_norm": 0.0005126605392433703,
-      "learning_rate": 0.0001999969606385057,
-      "loss": 46.0,
-      "step": 15421
-    },
-    {
-      "epoch": 2.4835943475985345,
-      "grad_norm": 0.0022305878810584545,
-      "learning_rate": 0.00019999696024405436,
-      "loss": 46.0,
-      "step": 15422
-    },
-    {
-      "epoch": 2.483755384677322,
-      "grad_norm": 0.0021804654970765114,
-      "learning_rate": 0.00019999695984957745,
-      "loss": 46.0,
-      "step": 15423
-    },
-    {
-      "epoch": 2.4839164217561094,
-      "grad_norm": 0.0023431049194186926,
-      "learning_rate": 0.00019999695945507493,
-      "loss": 46.0,
-      "step": 15424
-    },
-    {
-      "epoch": 2.484077458834897,
-      "grad_norm": 0.0016466027591377497,
-      "learning_rate": 0.00019999695906054683,
-      "loss": 46.0,
-      "step": 15425
-    },
-    {
-      "epoch": 2.4842384959136843,
-      "grad_norm": 0.0035464242100715637,
-      "learning_rate": 0.0001999969586659931,
-      "loss": 46.0,
-      "step": 15426
-    },
-    {
-      "epoch": 2.4843995329924713,
-      "grad_norm": 0.002741391770541668,
-      "learning_rate": 0.0001999969582714138,
-      "loss": 46.0,
-      "step": 15427
-    },
-    {
-      "epoch": 2.4845605700712587,
-      "grad_norm": 0.001966150477528572,
-      "learning_rate": 0.00019999695787680893,
-      "loss": 46.0,
-      "step": 15428
-    },
-    {
-      "epoch": 2.484721607150046,
-      "grad_norm": 0.00907079130411148,
-      "learning_rate": 0.00019999695748217845,
-      "loss": 46.0,
-      "step": 15429
-    },
-    {
-      "epoch": 2.4848826442288336,
-      "grad_norm": 0.000851382443215698,
-      "learning_rate": 0.00019999695708752238,
-      "loss": 46.0,
-      "step": 15430
-    },
-    {
-      "epoch": 2.485043681307621,
-      "grad_norm": 0.004795973654836416,
-      "learning_rate": 0.0001999969566928407,
-      "loss": 46.0,
-      "step": 15431
-    },
-    {
-      "epoch": 2.4852047183864086,
-      "grad_norm": 0.000985900405794382,
-      "learning_rate": 0.00019999695629813343,
-      "loss": 46.0,
-      "step": 15432
-    },
-    {
-      "epoch": 2.485365755465196,
-      "grad_norm": 0.006527276244014502,
-      "learning_rate": 0.00019999695590340054,
-      "loss": 46.0,
-      "step": 15433
-    },
-    {
-      "epoch": 2.4855267925439835,
-      "grad_norm": 0.006105063483119011,
-      "learning_rate": 0.0001999969555086421,
-      "loss": 46.0,
-      "step": 15434
-    },
-    {
-      "epoch": 2.4856878296227705,
-      "grad_norm": 0.0022340097930282354,
-      "learning_rate": 0.00019999695511385804,
-      "loss": 46.0,
-      "step": 15435
-    },
-    {
-      "epoch": 2.485848866701558,
-      "grad_norm": 0.001533149159513414,
-      "learning_rate": 0.0001999969547190484,
-      "loss": 46.0,
-      "step": 15436
-    },
-    {
-      "epoch": 2.4860099037803454,
-      "grad_norm": 0.008326280862092972,
-      "learning_rate": 0.00019999695432421316,
-      "loss": 46.0,
-      "step": 15437
-    },
-    {
-      "epoch": 2.486170940859133,
-      "grad_norm": 0.00044093994074501097,
-      "learning_rate": 0.00019999695392935234,
-      "loss": 46.0,
-      "step": 15438
-    },
-    {
-      "epoch": 2.4863319779379203,
-      "grad_norm": 0.0013709098566323519,
-      "learning_rate": 0.0001999969535344659,
-      "loss": 46.0,
-      "step": 15439
-    },
-    {
-      "epoch": 2.4864930150167077,
-      "grad_norm": 0.0010089291026815772,
-      "learning_rate": 0.00019999695313955388,
-      "loss": 46.0,
-      "step": 15440
-    },
-    {
-      "epoch": 2.486654052095495,
-      "grad_norm": 0.00929345190525055,
-      "learning_rate": 0.00019999695274461625,
-      "loss": 46.0,
-      "step": 15441
-    },
-    {
-      "epoch": 2.486815089174282,
-      "grad_norm": 0.005143813788890839,
-      "learning_rate": 0.00019999695234965305,
-      "loss": 46.0,
-      "step": 15442
-    },
-    {
-      "epoch": 2.4869761262530696,
-      "grad_norm": 0.001325281453318894,
-      "learning_rate": 0.00019999695195466424,
-      "loss": 46.0,
-      "step": 15443
-    },
-    {
-      "epoch": 2.487137163331857,
-      "grad_norm": 0.0017885920824483037,
-      "learning_rate": 0.00019999695155964987,
-      "loss": 46.0,
-      "step": 15444
-    },
-    {
-      "epoch": 2.4872982004106445,
-      "grad_norm": 0.0013921266654506326,
-      "learning_rate": 0.00019999695116460986,
-      "loss": 46.0,
-      "step": 15445
-    },
-    {
-      "epoch": 2.487459237489432,
-      "grad_norm": 0.004191992804408073,
-      "learning_rate": 0.00019999695076954429,
-      "loss": 46.0,
-      "step": 15446
-    },
-    {
-      "epoch": 2.4876202745682194,
-      "grad_norm": 0.002598048886284232,
-      "learning_rate": 0.0001999969503744531,
-      "loss": 46.0,
-      "step": 15447
-    },
-    {
-      "epoch": 2.487781311647007,
-      "grad_norm": 0.002194049069657922,
-      "learning_rate": 0.00019999694997933633,
-      "loss": 46.0,
-      "step": 15448
-    },
-    {
-      "epoch": 2.4879423487257943,
-      "grad_norm": 0.002046563196927309,
-      "learning_rate": 0.00019999694958419394,
-      "loss": 46.0,
-      "step": 15449
-    },
-    {
-      "epoch": 2.4881033858045813,
-      "grad_norm": 0.0022202348336577415,
-      "learning_rate": 0.000199996949189026,
-      "loss": 46.0,
-      "step": 15450
-    },
-    {
-      "epoch": 2.488264422883369,
-      "grad_norm": 0.000771569088101387,
-      "learning_rate": 0.00019999694879383243,
-      "loss": 46.0,
-      "step": 15451
-    },
-    {
-      "epoch": 2.4884254599621562,
-      "grad_norm": 0.0006596861639991403,
-      "learning_rate": 0.00019999694839861328,
-      "loss": 46.0,
-      "step": 15452
-    },
-    {
-      "epoch": 2.4885864970409437,
-      "grad_norm": 0.001168744289316237,
-      "learning_rate": 0.00019999694800336854,
-      "loss": 46.0,
-      "step": 15453
-    },
-    {
-      "epoch": 2.488747534119731,
-      "grad_norm": 0.0031377053819596767,
-      "learning_rate": 0.0001999969476080982,
-      "loss": 46.0,
-      "step": 15454
-    },
-    {
-      "epoch": 2.4889085711985186,
-      "grad_norm": 0.0023955118376761675,
-      "learning_rate": 0.00019999694721280225,
-      "loss": 46.0,
-      "step": 15455
-    },
-    {
-      "epoch": 2.4890696082773056,
-      "grad_norm": 0.0022035203874111176,
-      "learning_rate": 0.00019999694681748076,
-      "loss": 46.0,
-      "step": 15456
-    },
-    {
-      "epoch": 2.489230645356093,
-      "grad_norm": 0.004684357903897762,
-      "learning_rate": 0.00019999694642213362,
-      "loss": 46.0,
-      "step": 15457
-    },
-    {
-      "epoch": 2.4893916824348805,
-      "grad_norm": 0.004034834448248148,
-      "learning_rate": 0.00019999694602676092,
-      "loss": 46.0,
-      "step": 15458
-    },
-    {
-      "epoch": 2.489552719513668,
-      "grad_norm": 0.0030605599749833345,
-      "learning_rate": 0.0001999969456313626,
-      "loss": 46.0,
-      "step": 15459
-    },
-    {
-      "epoch": 2.4897137565924554,
-      "grad_norm": 0.00829601101577282,
-      "learning_rate": 0.0001999969452359387,
-      "loss": 46.0,
-      "step": 15460
-    },
-    {
-      "epoch": 2.489874793671243,
-      "grad_norm": 0.00346462894231081,
-      "learning_rate": 0.00019999694484048921,
-      "loss": 46.0,
-      "step": 15461
-    },
-    {
-      "epoch": 2.4900358307500303,
-      "grad_norm": 0.00260978564620018,
-      "learning_rate": 0.00019999694444501408,
-      "loss": 46.0,
-      "step": 15462
-    },
-    {
-      "epoch": 2.4901968678288178,
-      "grad_norm": 0.0013203290291130543,
-      "learning_rate": 0.00019999694404951342,
-      "loss": 46.0,
-      "step": 15463
-    },
-    {
-      "epoch": 2.4903579049076052,
-      "grad_norm": 0.007363055367022753,
-      "learning_rate": 0.00019999694365398715,
-      "loss": 46.0,
-      "step": 15464
-    },
-    {
-      "epoch": 2.4905189419863922,
-      "grad_norm": 0.004174639470875263,
-      "learning_rate": 0.00019999694325843525,
-      "loss": 46.0,
-      "step": 15465
-    },
-    {
-      "epoch": 2.4906799790651797,
-      "grad_norm": 0.0006472582463175058,
-      "learning_rate": 0.0001999969428628578,
-      "loss": 46.0,
-      "step": 15466
-    },
-    {
-      "epoch": 2.490841016143967,
-      "grad_norm": 0.001878245733678341,
-      "learning_rate": 0.00019999694246725474,
-      "loss": 46.0,
-      "step": 15467
-    },
-    {
-      "epoch": 2.4910020532227546,
-      "grad_norm": 0.0013203345006331801,
-      "learning_rate": 0.00019999694207162606,
-      "loss": 46.0,
-      "step": 15468
-    },
-    {
-      "epoch": 2.491163090301542,
-      "grad_norm": 0.00864538736641407,
-      "learning_rate": 0.00019999694167597182,
-      "loss": 46.0,
-      "step": 15469
-    },
-    {
-      "epoch": 2.4913241273803295,
-      "grad_norm": 0.0008616847335360944,
-      "learning_rate": 0.000199996941280292,
-      "loss": 46.0,
-      "step": 15470
-    },
-    {
-      "epoch": 2.4914851644591165,
-      "grad_norm": 0.011509637348353863,
-      "learning_rate": 0.00019999694088458652,
-      "loss": 46.0,
-      "step": 15471
-    },
-    {
-      "epoch": 2.491646201537904,
-      "grad_norm": 0.006392920855432749,
-      "learning_rate": 0.00019999694048885552,
-      "loss": 46.0,
-      "step": 15472
-    },
-    {
-      "epoch": 2.4918072386166914,
-      "grad_norm": 0.0009942905744537711,
-      "learning_rate": 0.0001999969400930989,
-      "loss": 46.0,
-      "step": 15473
-    },
-    {
-      "epoch": 2.491968275695479,
-      "grad_norm": 0.0076873814687132835,
-      "learning_rate": 0.00019999693969731668,
-      "loss": 46.0,
-      "step": 15474
-    },
-    {
-      "epoch": 2.4921293127742663,
-      "grad_norm": 0.0010989124421030283,
-      "learning_rate": 0.00019999693930150886,
-      "loss": 46.0,
-      "step": 15475
-    },
-    {
-      "epoch": 2.4922903498530538,
-      "grad_norm": 0.001140486216172576,
-      "learning_rate": 0.00019999693890567546,
-      "loss": 46.0,
-      "step": 15476
-    },
-    {
-      "epoch": 2.492451386931841,
-      "grad_norm": 0.002440092386677861,
-      "learning_rate": 0.00019999693850981644,
-      "loss": 46.0,
-      "step": 15477
-    },
-    {
-      "epoch": 2.4926124240106287,
-      "grad_norm": 0.0034286475274711847,
-      "learning_rate": 0.00019999693811393186,
-      "loss": 46.0,
-      "step": 15478
-    },
-    {
-      "epoch": 2.4927734610894157,
-      "grad_norm": 0.0026829675771296024,
-      "learning_rate": 0.00019999693771802167,
-      "loss": 46.0,
-      "step": 15479
-    },
-    {
-      "epoch": 2.492934498168203,
-      "grad_norm": 0.009589627385139465,
-      "learning_rate": 0.0001999969373220859,
-      "loss": 46.0,
-      "step": 15480
-    },
-    {
-      "epoch": 2.4930955352469906,
-      "grad_norm": 0.0019724280573427677,
-      "learning_rate": 0.0001999969369261245,
-      "loss": 46.0,
-      "step": 15481
-    },
-    {
-      "epoch": 2.493256572325778,
-      "grad_norm": 0.001608406426385045,
-      "learning_rate": 0.0001999969365301375,
-      "loss": 46.0,
-      "step": 15482
-    },
-    {
-      "epoch": 2.4934176094045655,
-      "grad_norm": 0.0018170453840866685,
-      "learning_rate": 0.00019999693613412494,
-      "loss": 46.0,
-      "step": 15483
-    },
-    {
-      "epoch": 2.493578646483353,
-      "grad_norm": 0.0027995998971164227,
-      "learning_rate": 0.0001999969357380868,
-      "loss": 46.0,
-      "step": 15484
-    },
-    {
-      "epoch": 2.4937396835621404,
-      "grad_norm": 0.0012566003715619445,
-      "learning_rate": 0.00019999693534202305,
-      "loss": 46.0,
-      "step": 15485
-    },
-    {
-      "epoch": 2.4939007206409274,
-      "grad_norm": 0.0020886787679046392,
-      "learning_rate": 0.0001999969349459337,
-      "loss": 46.0,
-      "step": 15486
-    },
-    {
-      "epoch": 2.494061757719715,
-      "grad_norm": 0.0009640268399380147,
-      "learning_rate": 0.00019999693454981877,
-      "loss": 46.0,
-      "step": 15487
-    },
-    {
-      "epoch": 2.4942227947985023,
-      "grad_norm": 0.00232880562543869,
-      "learning_rate": 0.0001999969341536782,
-      "loss": 46.0,
-      "step": 15488
-    },
-    {
-      "epoch": 2.4943838318772897,
-      "grad_norm": 0.0019670519977808,
-      "learning_rate": 0.00019999693375751207,
-      "loss": 46.0,
-      "step": 15489
-    },
-    {
-      "epoch": 2.494544868956077,
-      "grad_norm": 0.0031430241651833057,
-      "learning_rate": 0.00019999693336132036,
-      "loss": 46.0,
-      "step": 15490
-    },
-    {
-      "epoch": 2.4947059060348646,
-      "grad_norm": 0.0027613763231784105,
-      "learning_rate": 0.00019999693296510301,
-      "loss": 46.0,
-      "step": 15491
-    },
-    {
-      "epoch": 2.494866943113652,
-      "grad_norm": 0.0004651047638617456,
-      "learning_rate": 0.0001999969325688601,
-      "loss": 46.0,
-      "step": 15492
-    },
-    {
-      "epoch": 2.4950279801924395,
-      "grad_norm": 0.0029555028304457664,
-      "learning_rate": 0.0001999969321725916,
-      "loss": 46.0,
-      "step": 15493
-    },
-    {
-      "epoch": 2.4951890172712266,
-      "grad_norm": 0.0048257531598210335,
-      "learning_rate": 0.0001999969317762975,
-      "loss": 46.0,
-      "step": 15494
-    },
-    {
-      "epoch": 2.495350054350014,
-      "grad_norm": 0.0019671109039336443,
-      "learning_rate": 0.0001999969313799778,
-      "loss": 46.0,
-      "step": 15495
-    },
-    {
-      "epoch": 2.4955110914288015,
-      "grad_norm": 0.0012454113457351923,
-      "learning_rate": 0.00019999693098363253,
-      "loss": 46.0,
-      "step": 15496
-    },
-    {
-      "epoch": 2.495672128507589,
-      "grad_norm": 0.004872001241892576,
-      "learning_rate": 0.00019999693058726163,
-      "loss": 46.0,
-      "step": 15497
-    },
-    {
-      "epoch": 2.4958331655863764,
-      "grad_norm": 0.0027476674877107143,
-      "learning_rate": 0.00019999693019086517,
-      "loss": 46.0,
-      "step": 15498
-    },
-    {
-      "epoch": 2.495994202665164,
-      "grad_norm": 0.0028232336044311523,
-      "learning_rate": 0.0001999969297944431,
-      "loss": 46.0,
-      "step": 15499
-    },
-    {
-      "epoch": 2.496155239743951,
-      "grad_norm": 0.002674079267308116,
-      "learning_rate": 0.00019999692939799544,
-      "loss": 46.0,
-      "step": 15500
-    },
-    {
-      "epoch": 2.4963162768227383,
-      "grad_norm": 0.001613573869690299,
-      "learning_rate": 0.00019999692900152217,
-      "loss": 46.0,
-      "step": 15501
-    },
-    {
-      "epoch": 2.4964773139015257,
-      "grad_norm": 0.002364043379202485,
-      "learning_rate": 0.0001999969286050233,
-      "loss": 46.0,
-      "step": 15502
-    },
-    {
-      "epoch": 2.496638350980313,
-      "grad_norm": 0.0010843501659110188,
-      "learning_rate": 0.00019999692820849883,
-      "loss": 46.0,
-      "step": 15503
-    },
-    {
-      "epoch": 2.4967993880591006,
-      "grad_norm": 0.00433829240500927,
-      "learning_rate": 0.0001999969278119488,
-      "loss": 46.0,
-      "step": 15504
-    },
-    {
-      "epoch": 2.496960425137888,
-      "grad_norm": 0.0006053737597540021,
-      "learning_rate": 0.0001999969274153732,
-      "loss": 46.0,
-      "step": 15505
-    },
-    {
-      "epoch": 2.4971214622166755,
-      "grad_norm": 0.004616125021129847,
-      "learning_rate": 0.00019999692701877194,
-      "loss": 46.0,
-      "step": 15506
-    },
-    {
-      "epoch": 2.497282499295463,
-      "grad_norm": 0.001427347306162119,
-      "learning_rate": 0.00019999692662214512,
-      "loss": 46.0,
-      "step": 15507
-    },
-    {
-      "epoch": 2.49744353637425,
-      "grad_norm": 0.0005079785478301346,
-      "learning_rate": 0.00019999692622549268,
-      "loss": 46.0,
-      "step": 15508
-    },
-    {
-      "epoch": 2.4976045734530374,
-      "grad_norm": 0.006529994774609804,
-      "learning_rate": 0.00019999692582881468,
-      "loss": 46.0,
-      "step": 15509
-    },
-    {
-      "epoch": 2.497765610531825,
-      "grad_norm": 0.003117122920230031,
-      "learning_rate": 0.00019999692543211107,
-      "loss": 46.0,
-      "step": 15510
-    },
-    {
-      "epoch": 2.4979266476106123,
-      "grad_norm": 0.010971128940582275,
-      "learning_rate": 0.00019999692503538187,
-      "loss": 46.0,
-      "step": 15511
-    },
-    {
-      "epoch": 2.4980876846894,
-      "grad_norm": 0.003298766678199172,
-      "learning_rate": 0.00019999692463862706,
-      "loss": 46.0,
-      "step": 15512
-    },
-    {
-      "epoch": 2.4982487217681872,
-      "grad_norm": 0.008274292573332787,
-      "learning_rate": 0.00019999692424184669,
-      "loss": 46.0,
-      "step": 15513
-    },
-    {
-      "epoch": 2.4984097588469747,
-      "grad_norm": 0.0023968422319740057,
-      "learning_rate": 0.00019999692384504072,
-      "loss": 46.0,
-      "step": 15514
-    },
-    {
-      "epoch": 2.4985707959257617,
-      "grad_norm": 0.0019090829882770777,
-      "learning_rate": 0.00019999692344820912,
-      "loss": 46.0,
-      "step": 15515
-    },
-    {
-      "epoch": 2.498731833004549,
-      "grad_norm": 0.0037231524474918842,
-      "learning_rate": 0.00019999692305135196,
-      "loss": 46.0,
-      "step": 15516
-    },
-    {
-      "epoch": 2.4988928700833366,
-      "grad_norm": 0.006756780203431845,
-      "learning_rate": 0.0001999969226544692,
-      "loss": 46.0,
-      "step": 15517
-    },
-    {
-      "epoch": 2.499053907162124,
-      "grad_norm": 0.005277234595268965,
-      "learning_rate": 0.00019999692225756085,
-      "loss": 46.0,
-      "step": 15518
-    },
-    {
-      "epoch": 2.4992149442409115,
-      "grad_norm": 0.001203195541165769,
-      "learning_rate": 0.0001999969218606269,
-      "loss": 46.0,
-      "step": 15519
-    },
-    {
-      "epoch": 2.499375981319699,
-      "grad_norm": 0.002336875768378377,
-      "learning_rate": 0.00019999692146366733,
-      "loss": 46.0,
-      "step": 15520
-    },
-    {
-      "epoch": 2.4995370183984864,
-      "grad_norm": 0.009139231406152248,
-      "learning_rate": 0.0001999969210666822,
-      "loss": 46.0,
-      "step": 15521
-    },
-    {
-      "epoch": 2.499698055477274,
-      "grad_norm": 0.0034921576734632254,
-      "learning_rate": 0.00019999692066967147,
-      "loss": 46.0,
-      "step": 15522
-    },
-    {
-      "epoch": 2.499859092556061,
-      "grad_norm": 0.0009065789636224508,
-      "learning_rate": 0.00019999692027263512,
-      "loss": 46.0,
-      "step": 15523
-    },
-    {
-      "epoch": 2.5000201296348483,
-      "grad_norm": 0.0011126770405098796,
-      "learning_rate": 0.0001999969198755732,
-      "loss": 46.0,
-      "step": 15524
-    },
-    {
-      "epoch": 2.5001811667136358,
-      "grad_norm": 0.0009970295941457152,
-      "learning_rate": 0.0001999969194784857,
-      "loss": 46.0,
-      "step": 15525
-    },
-    {
-      "epoch": 2.5003422037924232,
-      "grad_norm": 0.006593253929167986,
-      "learning_rate": 0.00019999691908137256,
-      "loss": 46.0,
-      "step": 15526
-    },
-    {
-      "epoch": 2.5005032408712107,
-      "grad_norm": 0.0012017093831673265,
-      "learning_rate": 0.00019999691868423386,
-      "loss": 46.0,
-      "step": 15527
-    },
-    {
-      "epoch": 2.500664277949998,
-      "grad_norm": 0.00404240470379591,
-      "learning_rate": 0.00019999691828706957,
-      "loss": 46.0,
-      "step": 15528
-    },
-    {
-      "epoch": 2.500825315028785,
-      "grad_norm": 0.001371907303109765,
-      "learning_rate": 0.00019999691788987966,
-      "loss": 46.0,
-      "step": 15529
-    },
-    {
-      "epoch": 2.5009863521075726,
-      "grad_norm": 0.002074048388749361,
-      "learning_rate": 0.00019999691749266417,
-      "loss": 46.0,
-      "step": 15530
-    },
-    {
-      "epoch": 2.50114738918636,
-      "grad_norm": 0.007534390315413475,
-      "learning_rate": 0.0001999969170954231,
-      "loss": 46.0,
-      "step": 15531
-    },
-    {
-      "epoch": 2.5013084262651475,
-      "grad_norm": 0.002561690052971244,
-      "learning_rate": 0.0001999969166981564,
-      "loss": 46.0,
-      "step": 15532
-    },
-    {
-      "epoch": 2.501469463343935,
-      "grad_norm": 0.0016198353841900826,
-      "learning_rate": 0.00019999691630086415,
-      "loss": 46.0,
-      "step": 15533
-    },
-    {
-      "epoch": 2.5016305004227224,
-      "grad_norm": 0.004757654387503862,
-      "learning_rate": 0.00019999691590354628,
-      "loss": 46.0,
-      "step": 15534
-    },
-    {
-      "epoch": 2.50179153750151,
-      "grad_norm": 0.0013877920573577285,
-      "learning_rate": 0.00019999691550620283,
-      "loss": 46.0,
-      "step": 15535
-    },
-    {
-      "epoch": 2.5019525745802973,
-      "grad_norm": 0.0051576984114944935,
-      "learning_rate": 0.00019999691510883376,
-      "loss": 46.0,
-      "step": 15536
-    },
-    {
-      "epoch": 2.5021136116590847,
-      "grad_norm": 0.00174257205799222,
-      "learning_rate": 0.0001999969147114391,
-      "loss": 46.0,
-      "step": 15537
-    },
-    {
-      "epoch": 2.5022746487378718,
-      "grad_norm": 0.0060266535729169846,
-      "learning_rate": 0.0001999969143140189,
-      "loss": 46.0,
-      "step": 15538
-    },
-    {
-      "epoch": 2.502435685816659,
-      "grad_norm": 0.008352944627404213,
-      "learning_rate": 0.00019999691391657304,
-      "loss": 46.0,
-      "step": 15539
-    },
-    {
-      "epoch": 2.5025967228954467,
-      "grad_norm": 0.0027539320290088654,
-      "learning_rate": 0.00019999691351910162,
-      "loss": 46.0,
-      "step": 15540
-    },
-    {
-      "epoch": 2.502757759974234,
-      "grad_norm": 0.002372451825067401,
-      "learning_rate": 0.0001999969131216046,
-      "loss": 46.0,
-      "step": 15541
-    },
-    {
-      "epoch": 2.5029187970530216,
-      "grad_norm": 0.00531651871278882,
-      "learning_rate": 0.000199996912724082,
-      "loss": 46.0,
-      "step": 15542
-    },
-    {
-      "epoch": 2.503079834131809,
-      "grad_norm": 0.0063546826131641865,
-      "learning_rate": 0.00019999691232653376,
-      "loss": 46.0,
-      "step": 15543
-    },
-    {
-      "epoch": 2.503240871210596,
-      "grad_norm": 0.0009585191146470606,
-      "learning_rate": 0.00019999691192895995,
-      "loss": 46.0,
-      "step": 15544
-    },
-    {
-      "epoch": 2.5034019082893835,
-      "grad_norm": 0.0018841332057490945,
-      "learning_rate": 0.00019999691153136057,
-      "loss": 46.0,
-      "step": 15545
-    },
-    {
-      "epoch": 2.503562945368171,
-      "grad_norm": 0.003862903220579028,
-      "learning_rate": 0.00019999691113373554,
-      "loss": 46.0,
-      "step": 15546
-    },
-    {
-      "epoch": 2.5037239824469584,
-      "grad_norm": 0.0038681388832628727,
-      "learning_rate": 0.00019999691073608496,
-      "loss": 46.0,
-      "step": 15547
-    },
-    {
-      "epoch": 2.503885019525746,
-      "grad_norm": 0.001058459747582674,
-      "learning_rate": 0.00019999691033840877,
-      "loss": 46.0,
-      "step": 15548
-    },
-    {
-      "epoch": 2.5040460566045333,
-      "grad_norm": 0.0011678003938868642,
-      "learning_rate": 0.000199996909940707,
-      "loss": 46.0,
-      "step": 15549
-    },
-    {
-      "epoch": 2.5042070936833207,
-      "grad_norm": 0.0016315385000780225,
-      "learning_rate": 0.00019999690954297962,
-      "loss": 46.0,
-      "step": 15550
-    },
-    {
-      "epoch": 2.504368130762108,
-      "grad_norm": 0.0024730025324970484,
-      "learning_rate": 0.00019999690914522669,
-      "loss": 46.0,
-      "step": 15551
-    },
-    {
-      "epoch": 2.5045291678408956,
-      "grad_norm": 0.0024479362182319164,
-      "learning_rate": 0.00019999690874744812,
-      "loss": 46.0,
-      "step": 15552
-    },
-    {
-      "epoch": 2.5046902049196826,
-      "grad_norm": 0.0019860391039401293,
-      "learning_rate": 0.00019999690834964398,
-      "loss": 46.0,
-      "step": 15553
-    },
-    {
-      "epoch": 2.50485124199847,
-      "grad_norm": 0.0030104625038802624,
-      "learning_rate": 0.0001999969079518142,
-      "loss": 46.0,
-      "step": 15554
-    },
-    {
-      "epoch": 2.5050122790772575,
-      "grad_norm": 0.0040182070806622505,
-      "learning_rate": 0.00019999690755395888,
-      "loss": 46.0,
-      "step": 15555
-    },
-    {
-      "epoch": 2.505173316156045,
-      "grad_norm": 0.0012150746770203114,
-      "learning_rate": 0.00019999690715607793,
-      "loss": 46.0,
-      "step": 15556
-    },
-    {
-      "epoch": 2.5053343532348324,
-      "grad_norm": 0.010076258331537247,
-      "learning_rate": 0.00019999690675817142,
-      "loss": 46.0,
-      "step": 15557
-    },
-    {
-      "epoch": 2.5054953903136195,
-      "grad_norm": 0.0036031913477927446,
-      "learning_rate": 0.00019999690636023928,
-      "loss": 46.0,
-      "step": 15558
-    },
-    {
-      "epoch": 2.505656427392407,
-      "grad_norm": 0.0033157074358314276,
-      "learning_rate": 0.00019999690596228157,
-      "loss": 46.0,
-      "step": 15559
-    },
-    {
-      "epoch": 2.5058174644711944,
-      "grad_norm": 0.0029015454929322004,
-      "learning_rate": 0.00019999690556429827,
-      "loss": 46.0,
-      "step": 15560
-    },
-    {
-      "epoch": 2.505978501549982,
-      "grad_norm": 0.0010072861332446337,
-      "learning_rate": 0.00019999690516628936,
-      "loss": 46.0,
-      "step": 15561
-    },
-    {
-      "epoch": 2.5061395386287693,
-      "grad_norm": 0.003890155116096139,
-      "learning_rate": 0.00019999690476825487,
-      "loss": 46.0,
-      "step": 15562
-    },
-    {
-      "epoch": 2.5063005757075567,
-      "grad_norm": 0.001681747380644083,
-      "learning_rate": 0.00019999690437019475,
-      "loss": 46.0,
-      "step": 15563
-    },
-    {
-      "epoch": 2.506461612786344,
-      "grad_norm": 0.0047065881080925465,
-      "learning_rate": 0.00019999690397210906,
-      "loss": 46.0,
-      "step": 15564
-    },
-    {
-      "epoch": 2.5066226498651316,
-      "grad_norm": 0.002220695838332176,
-      "learning_rate": 0.0001999969035739978,
-      "loss": 46.0,
-      "step": 15565
-    },
-    {
-      "epoch": 2.506783686943919,
-      "grad_norm": 0.0048478953540325165,
-      "learning_rate": 0.00019999690317586092,
-      "loss": 46.0,
-      "step": 15566
-    },
-    {
-      "epoch": 2.506944724022706,
-      "grad_norm": 0.004355294164270163,
-      "learning_rate": 0.00019999690277769844,
-      "loss": 46.0,
-      "step": 15567
-    },
-    {
-      "epoch": 2.5071057611014935,
-      "grad_norm": 0.004388112109154463,
-      "learning_rate": 0.0001999969023795104,
-      "loss": 46.0,
-      "step": 15568
-    },
-    {
-      "epoch": 2.507266798180281,
-      "grad_norm": 0.01185726746916771,
-      "learning_rate": 0.00019999690198129673,
-      "loss": 46.0,
-      "step": 15569
-    },
-    {
-      "epoch": 2.5074278352590684,
-      "grad_norm": 0.001428323914296925,
-      "learning_rate": 0.00019999690158305748,
-      "loss": 46.0,
-      "step": 15570
-    },
-    {
-      "epoch": 2.507588872337856,
-      "grad_norm": 0.01658983901143074,
-      "learning_rate": 0.00019999690118479264,
-      "loss": 46.0,
-      "step": 15571
-    },
-    {
-      "epoch": 2.5077499094166433,
-      "grad_norm": 0.0010460596531629562,
-      "learning_rate": 0.00019999690078650217,
-      "loss": 46.0,
-      "step": 15572
-    },
-    {
-      "epoch": 2.5079109464954303,
-      "grad_norm": 0.0032196880783885717,
-      "learning_rate": 0.00019999690038818616,
-      "loss": 46.0,
-      "step": 15573
-    },
-    {
-      "epoch": 2.508071983574218,
-      "grad_norm": 0.001326313940808177,
-      "learning_rate": 0.00019999689998984453,
-      "loss": 46.0,
-      "step": 15574
-    },
-    {
-      "epoch": 2.5082330206530052,
-      "grad_norm": 0.0025898038875311613,
-      "learning_rate": 0.0001999968995914773,
-      "loss": 46.0,
-      "step": 15575
-    },
-    {
-      "epoch": 2.5083940577317927,
-      "grad_norm": 0.007924860343337059,
-      "learning_rate": 0.0001999968991930845,
-      "loss": 46.0,
-      "step": 15576
-    },
-    {
-      "epoch": 2.50855509481058,
-      "grad_norm": 0.0020931025501340628,
-      "learning_rate": 0.00019999689879466608,
-      "loss": 46.0,
-      "step": 15577
-    },
-    {
-      "epoch": 2.5087161318893676,
-      "grad_norm": 0.0003746419388335198,
-      "learning_rate": 0.00019999689839622205,
-      "loss": 46.0,
-      "step": 15578
-    },
-    {
-      "epoch": 2.508877168968155,
-      "grad_norm": 0.009165162220597267,
-      "learning_rate": 0.00019999689799775247,
-      "loss": 46.0,
-      "step": 15579
-    },
-    {
-      "epoch": 2.5090382060469425,
-      "grad_norm": 0.001062337658368051,
-      "learning_rate": 0.0001999968975992573,
-      "loss": 46.0,
-      "step": 15580
-    },
-    {
-      "epoch": 2.50919924312573,
-      "grad_norm": 0.006670957896858454,
-      "learning_rate": 0.00019999689720073648,
-      "loss": 46.0,
-      "step": 15581
-    },
-    {
-      "epoch": 2.509360280204517,
-      "grad_norm": 0.0018045203760266304,
-      "learning_rate": 0.00019999689680219013,
-      "loss": 46.0,
-      "step": 15582
-    },
-    {
-      "epoch": 2.5095213172833044,
-      "grad_norm": 0.0033146946225315332,
-      "learning_rate": 0.00019999689640361814,
-      "loss": 46.0,
-      "step": 15583
-    },
-    {
-      "epoch": 2.509682354362092,
-      "grad_norm": 0.008988477289676666,
-      "learning_rate": 0.00019999689600502056,
-      "loss": 46.0,
-      "step": 15584
-    },
-    {
-      "epoch": 2.5098433914408793,
-      "grad_norm": 0.0010040607303380966,
-      "learning_rate": 0.0001999968956063974,
-      "loss": 46.0,
-      "step": 15585
-    },
-    {
-      "epoch": 2.5100044285196668,
-      "grad_norm": 0.0007079103961586952,
-      "learning_rate": 0.00019999689520774864,
-      "loss": 46.0,
-      "step": 15586
-    },
-    {
-      "epoch": 2.5101654655984538,
-      "grad_norm": 0.0005958284600637853,
-      "learning_rate": 0.00019999689480907428,
-      "loss": 46.0,
-      "step": 15587
-    },
-    {
-      "epoch": 2.5103265026772412,
-      "grad_norm": 0.007474029902368784,
-      "learning_rate": 0.00019999689441037435,
-      "loss": 46.0,
-      "step": 15588
-    },
-    {
-      "epoch": 2.5104875397560287,
-      "grad_norm": 0.0017049798043444753,
-      "learning_rate": 0.0001999968940116488,
-      "loss": 46.0,
-      "step": 15589
-    },
-    {
-      "epoch": 2.510648576834816,
-      "grad_norm": 0.0053446246311068535,
-      "learning_rate": 0.00019999689361289766,
-      "loss": 46.0,
-      "step": 15590
-    },
-    {
-      "epoch": 2.5108096139136036,
-      "grad_norm": 0.007703241426497698,
-      "learning_rate": 0.00019999689321412097,
-      "loss": 46.0,
-      "step": 15591
-    },
-    {
-      "epoch": 2.510970650992391,
-      "grad_norm": 0.000802951748482883,
-      "learning_rate": 0.00019999689281531864,
-      "loss": 46.0,
-      "step": 15592
-    },
-    {
-      "epoch": 2.5111316880711785,
-      "grad_norm": 0.002584241796284914,
-      "learning_rate": 0.00019999689241649072,
-      "loss": 46.0,
-      "step": 15593
-    },
-    {
-      "epoch": 2.511292725149966,
-      "grad_norm": 0.001001784228719771,
-      "learning_rate": 0.0001999968920176372,
-      "loss": 46.0,
-      "step": 15594
-    },
-    {
-      "epoch": 2.5114537622287534,
-      "grad_norm": 0.0016463365172967315,
-      "learning_rate": 0.0001999968916187581,
-      "loss": 46.0,
-      "step": 15595
-    },
-    {
-      "epoch": 2.511614799307541,
-      "grad_norm": 0.0009321488323621452,
-      "learning_rate": 0.00019999689121985342,
-      "loss": 46.0,
-      "step": 15596
-    },
-    {
-      "epoch": 2.511775836386328,
-      "grad_norm": 0.00227958126924932,
-      "learning_rate": 0.0001999968908209231,
-      "loss": 46.0,
-      "step": 15597
-    },
-    {
-      "epoch": 2.5119368734651153,
-      "grad_norm": 0.0020615821704268456,
-      "learning_rate": 0.0001999968904219672,
-      "loss": 46.0,
-      "step": 15598
-    },
-    {
-      "epoch": 2.5120979105439027,
-      "grad_norm": 0.0005884967977181077,
-      "learning_rate": 0.00019999689002298573,
-      "loss": 46.0,
-      "step": 15599
-    },
-    {
-      "epoch": 2.51225894762269,
-      "grad_norm": 0.0010356927523389459,
-      "learning_rate": 0.00019999688962397868,
-      "loss": 46.0,
-      "step": 15600
-    },
-    {
-      "epoch": 2.5124199847014776,
-      "grad_norm": 0.010368126444518566,
-      "learning_rate": 0.00019999688922494598,
-      "loss": 46.0,
-      "step": 15601
-    },
-    {
-      "epoch": 2.5125810217802647,
-      "grad_norm": 0.0043180398643016815,
-      "learning_rate": 0.00019999688882588773,
-      "loss": 46.0,
-      "step": 15602
-    },
-    {
-      "epoch": 2.512742058859052,
-      "grad_norm": 0.0012822233838960528,
-      "learning_rate": 0.00019999688842680389,
-      "loss": 46.0,
-      "step": 15603
-    },
-    {
-      "epoch": 2.5129030959378396,
-      "grad_norm": 0.0037582844961434603,
-      "learning_rate": 0.00019999688802769443,
-      "loss": 46.0,
-      "step": 15604
-    },
-    {
-      "epoch": 2.513064133016627,
-      "grad_norm": 0.006926591973751783,
-      "learning_rate": 0.00019999688762855939,
-      "loss": 46.0,
-      "step": 15605
-    },
-    {
-      "epoch": 2.5132251700954145,
-      "grad_norm": 0.002861557062715292,
-      "learning_rate": 0.00019999688722939873,
-      "loss": 46.0,
-      "step": 15606
-    },
-    {
-      "epoch": 2.513386207174202,
-      "grad_norm": 0.0034386483021080494,
-      "learning_rate": 0.0001999968868302125,
-      "loss": 46.0,
-      "step": 15607
-    },
-    {
-      "epoch": 2.5135472442529894,
-      "grad_norm": 0.007441123481839895,
-      "learning_rate": 0.00019999688643100068,
-      "loss": 46.0,
-      "step": 15608
-    },
-    {
-      "epoch": 2.513708281331777,
-      "grad_norm": 0.0021146193612366915,
-      "learning_rate": 0.00019999688603176329,
-      "loss": 46.0,
-      "step": 15609
-    },
-    {
-      "epoch": 2.5138693184105643,
-      "grad_norm": 0.00563927972689271,
-      "learning_rate": 0.00019999688563250025,
-      "loss": 46.0,
-      "step": 15610
-    },
-    {
-      "epoch": 2.5140303554893513,
-      "grad_norm": 0.002676117466762662,
-      "learning_rate": 0.00019999688523321166,
-      "loss": 46.0,
-      "step": 15611
-    },
-    {
-      "epoch": 2.5141913925681387,
-      "grad_norm": 0.009635621681809425,
-      "learning_rate": 0.00019999688483389742,
-      "loss": 46.0,
-      "step": 15612
-    },
-    {
-      "epoch": 2.514352429646926,
-      "grad_norm": 0.005568181164562702,
-      "learning_rate": 0.00019999688443455763,
-      "loss": 46.0,
-      "step": 15613
-    },
-    {
-      "epoch": 2.5145134667257136,
-      "grad_norm": 0.0030140401795506477,
-      "learning_rate": 0.00019999688403519225,
-      "loss": 46.0,
-      "step": 15614
-    },
-    {
-      "epoch": 2.514674503804501,
-      "grad_norm": 0.003594675799831748,
-      "learning_rate": 0.00019999688363580128,
-      "loss": 46.0,
-      "step": 15615
-    },
-    {
-      "epoch": 2.5148355408832885,
-      "grad_norm": 0.004549028351902962,
-      "learning_rate": 0.0001999968832363847,
-      "loss": 46.0,
-      "step": 15616
-    },
-    {
-      "epoch": 2.5149965779620755,
-      "grad_norm": 0.001601845258846879,
-      "learning_rate": 0.00019999688283694252,
-      "loss": 46.0,
-      "step": 15617
-    },
-    {
-      "epoch": 2.515157615040863,
-      "grad_norm": 0.0016188089502975345,
-      "learning_rate": 0.00019999688243747474,
-      "loss": 46.0,
-      "step": 15618
-    },
-    {
-      "epoch": 2.5153186521196504,
-      "grad_norm": 0.009368407540023327,
-      "learning_rate": 0.0001999968820379814,
-      "loss": 46.0,
-      "step": 15619
-    },
-    {
-      "epoch": 2.515479689198438,
-      "grad_norm": 0.0025999932549893856,
-      "learning_rate": 0.00019999688163846243,
-      "loss": 46.0,
-      "step": 15620
-    },
-    {
-      "epoch": 2.5156407262772253,
-      "grad_norm": 0.0028097087051719427,
-      "learning_rate": 0.00019999688123891786,
-      "loss": 46.0,
-      "step": 15621
-    },
-    {
-      "epoch": 2.515801763356013,
-      "grad_norm": 0.005572411231696606,
-      "learning_rate": 0.00019999688083934772,
-      "loss": 46.0,
-      "step": 15622
-    },
-    {
-      "epoch": 2.5159628004348003,
-      "grad_norm": 0.0024771352764219046,
-      "learning_rate": 0.000199996880439752,
-      "loss": 46.0,
-      "step": 15623
-    },
-    {
-      "epoch": 2.5161238375135877,
-      "grad_norm": 0.0062670293264091015,
-      "learning_rate": 0.00019999688004013067,
-      "loss": 46.0,
-      "step": 15624
-    },
-    {
-      "epoch": 2.516284874592375,
-      "grad_norm": 0.0016047862591221929,
-      "learning_rate": 0.00019999687964048372,
-      "loss": 46.0,
-      "step": 15625
-    },
-    {
-      "epoch": 2.516445911671162,
-      "grad_norm": 0.0021048770286142826,
-      "learning_rate": 0.0001999968792408112,
-      "loss": 46.0,
-      "step": 15626
-    },
-    {
-      "epoch": 2.5166069487499496,
-      "grad_norm": 0.006055477075278759,
-      "learning_rate": 0.0001999968788411131,
-      "loss": 46.0,
-      "step": 15627
-    },
-    {
-      "epoch": 2.516767985828737,
-      "grad_norm": 0.0007138599758036435,
-      "learning_rate": 0.0001999968784413894,
-      "loss": 46.0,
-      "step": 15628
-    },
-    {
-      "epoch": 2.5169290229075245,
-      "grad_norm": 0.0019480041228234768,
-      "learning_rate": 0.0001999968780416401,
-      "loss": 46.0,
-      "step": 15629
-    },
-    {
-      "epoch": 2.517090059986312,
-      "grad_norm": 0.004484706092625856,
-      "learning_rate": 0.00019999687764186522,
-      "loss": 46.0,
-      "step": 15630
-    },
-    {
-      "epoch": 2.517251097065099,
-      "grad_norm": 0.002487529069185257,
-      "learning_rate": 0.00019999687724206472,
-      "loss": 46.0,
-      "step": 15631
-    },
-    {
-      "epoch": 2.5174121341438864,
-      "grad_norm": 0.00363441975787282,
-      "learning_rate": 0.00019999687684223863,
-      "loss": 46.0,
-      "step": 15632
-    },
-    {
-      "epoch": 2.517573171222674,
-      "grad_norm": 0.0015664249658584595,
-      "learning_rate": 0.00019999687644238693,
-      "loss": 46.0,
-      "step": 15633
-    },
-    {
-      "epoch": 2.5177342083014613,
-      "grad_norm": 0.0047117420472204685,
-      "learning_rate": 0.00019999687604250967,
-      "loss": 46.0,
-      "step": 15634
-    },
-    {
-      "epoch": 2.517895245380249,
-      "grad_norm": 0.0007667237659916282,
-      "learning_rate": 0.00019999687564260682,
-      "loss": 46.0,
-      "step": 15635
-    },
-    {
-      "epoch": 2.5180562824590362,
-      "grad_norm": 0.0014587810728698969,
-      "learning_rate": 0.00019999687524267833,
-      "loss": 46.0,
-      "step": 15636
-    },
-    {
-      "epoch": 2.5182173195378237,
-      "grad_norm": 0.0013479279587045312,
-      "learning_rate": 0.00019999687484272428,
-      "loss": 46.0,
-      "step": 15637
-    },
-    {
-      "epoch": 2.518378356616611,
-      "grad_norm": 0.0033141302410513163,
-      "learning_rate": 0.00019999687444274464,
-      "loss": 46.0,
-      "step": 15638
-    },
-    {
-      "epoch": 2.5185393936953986,
-      "grad_norm": 0.006300832610577345,
-      "learning_rate": 0.0001999968740427394,
-      "loss": 46.0,
-      "step": 15639
-    },
-    {
-      "epoch": 2.5187004307741856,
-      "grad_norm": 0.0022819561418145895,
-      "learning_rate": 0.00019999687364270855,
-      "loss": 46.0,
-      "step": 15640
-    },
-    {
-      "epoch": 2.518861467852973,
-      "grad_norm": 0.003823867067694664,
-      "learning_rate": 0.00019999687324265213,
-      "loss": 46.0,
-      "step": 15641
-    },
-    {
-      "epoch": 2.5190225049317605,
-      "grad_norm": 0.0007260690326802433,
-      "learning_rate": 0.0001999968728425701,
-      "loss": 46.0,
-      "step": 15642
-    },
-    {
-      "epoch": 2.519183542010548,
-      "grad_norm": 0.0003675887128338218,
-      "learning_rate": 0.00019999687244246249,
-      "loss": 46.0,
-      "step": 15643
-    },
-    {
-      "epoch": 2.5193445790893354,
-      "grad_norm": 0.003494833828881383,
-      "learning_rate": 0.00019999687204232927,
-      "loss": 46.0,
-      "step": 15644
-    },
-    {
-      "epoch": 2.519505616168123,
-      "grad_norm": 0.002476646564900875,
-      "learning_rate": 0.00019999687164217047,
-      "loss": 46.0,
-      "step": 15645
-    },
-    {
-      "epoch": 2.51966665324691,
-      "grad_norm": 0.0030917986296117306,
-      "learning_rate": 0.00019999687124198605,
-      "loss": 46.0,
-      "step": 15646
-    },
-    {
-      "epoch": 2.5198276903256973,
-      "grad_norm": 0.001155301695689559,
-      "learning_rate": 0.00019999687084177605,
-      "loss": 46.0,
-      "step": 15647
-    },
-    {
-      "epoch": 2.5199887274044848,
-      "grad_norm": 0.0011219088919460773,
-      "learning_rate": 0.00019999687044154046,
-      "loss": 46.0,
-      "step": 15648
-    },
-    {
-      "epoch": 2.520149764483272,
-      "grad_norm": 0.005977956112474203,
-      "learning_rate": 0.00019999687004127928,
-      "loss": 46.0,
-      "step": 15649
-    },
-    {
-      "epoch": 2.5203108015620597,
-      "grad_norm": 0.002195816021412611,
-      "learning_rate": 0.00019999686964099252,
-      "loss": 46.0,
-      "step": 15650
-    },
-    {
-      "epoch": 2.520471838640847,
-      "grad_norm": 0.006962589919567108,
-      "learning_rate": 0.00019999686924068014,
-      "loss": 46.0,
-      "step": 15651
-    },
-    {
-      "epoch": 2.5206328757196346,
-      "grad_norm": 0.0008813330787234008,
-      "learning_rate": 0.0001999968688403422,
-      "loss": 46.0,
-      "step": 15652
-    },
-    {
-      "epoch": 2.520793912798422,
-      "grad_norm": 0.004322392400354147,
-      "learning_rate": 0.00019999686843997862,
-      "loss": 46.0,
-      "step": 15653
-    },
-    {
-      "epoch": 2.5209549498772095,
-      "grad_norm": 0.005127814132720232,
-      "learning_rate": 0.00019999686803958945,
-      "loss": 46.0,
-      "step": 15654
-    },
-    {
-      "epoch": 2.5211159869559965,
-      "grad_norm": 0.0008404377149417996,
-      "learning_rate": 0.0001999968676391747,
-      "loss": 46.0,
-      "step": 15655
-    },
-    {
-      "epoch": 2.521277024034784,
-      "grad_norm": 0.003995373845100403,
-      "learning_rate": 0.00019999686723873436,
-      "loss": 46.0,
-      "step": 15656
-    },
-    {
-      "epoch": 2.5214380611135714,
-      "grad_norm": 0.006385750137269497,
-      "learning_rate": 0.00019999686683826843,
-      "loss": 46.0,
-      "step": 15657
-    },
-    {
-      "epoch": 2.521599098192359,
-      "grad_norm": 0.005439152475446463,
-      "learning_rate": 0.00019999686643777689,
-      "loss": 46.0,
-      "step": 15658
-    },
-    {
-      "epoch": 2.5217601352711463,
-      "grad_norm": 0.0017037151847034693,
-      "learning_rate": 0.00019999686603725978,
-      "loss": 46.0,
-      "step": 15659
-    },
-    {
-      "epoch": 2.5219211723499337,
-      "grad_norm": 0.0009728716686367989,
-      "learning_rate": 0.00019999686563671704,
-      "loss": 46.0,
-      "step": 15660
-    },
-    {
-      "epoch": 2.5220822094287207,
-      "grad_norm": 0.009522090665996075,
-      "learning_rate": 0.00019999686523614876,
-      "loss": 46.0,
-      "step": 15661
-    },
-    {
-      "epoch": 2.522243246507508,
-      "grad_norm": 0.0010149332229048014,
-      "learning_rate": 0.00019999686483555484,
-      "loss": 46.0,
-      "step": 15662
-    },
-    {
-      "epoch": 2.5224042835862956,
-      "grad_norm": 0.0019164497498422861,
-      "learning_rate": 0.00019999686443493534,
-      "loss": 46.0,
-      "step": 15663
-    },
-    {
-      "epoch": 2.522565320665083,
-      "grad_norm": 0.00240589608438313,
-      "learning_rate": 0.00019999686403429024,
-      "loss": 46.0,
-      "step": 15664
-    },
-    {
-      "epoch": 2.5227263577438706,
-      "grad_norm": 0.002015412785112858,
-      "learning_rate": 0.00019999686363361956,
-      "loss": 46.0,
-      "step": 15665
-    },
-    {
-      "epoch": 2.522887394822658,
-      "grad_norm": 0.00035457449848763645,
-      "learning_rate": 0.00019999686323292327,
-      "loss": 46.0,
-      "step": 15666
-    },
-    {
-      "epoch": 2.5230484319014455,
-      "grad_norm": 0.011164935305714607,
-      "learning_rate": 0.0001999968628322014,
-      "loss": 46.0,
-      "step": 15667
-    },
-    {
-      "epoch": 2.523209468980233,
-      "grad_norm": 0.004561011679470539,
-      "learning_rate": 0.00019999686243145392,
-      "loss": 46.0,
-      "step": 15668
-    },
-    {
-      "epoch": 2.5233705060590204,
-      "grad_norm": 0.003374389372766018,
-      "learning_rate": 0.00019999686203068086,
-      "loss": 46.0,
-      "step": 15669
-    },
-    {
-      "epoch": 2.5235315431378074,
-      "grad_norm": 0.0009957189904525876,
-      "learning_rate": 0.00019999686162988222,
-      "loss": 46.0,
-      "step": 15670
-    },
-    {
-      "epoch": 2.523692580216595,
-      "grad_norm": 0.004869405180215836,
-      "learning_rate": 0.00019999686122905793,
-      "loss": 46.0,
-      "step": 15671
-    },
-    {
-      "epoch": 2.5238536172953823,
-      "grad_norm": 0.009459338150918484,
-      "learning_rate": 0.0001999968608282081,
-      "loss": 46.0,
-      "step": 15672
-    },
-    {
-      "epoch": 2.5240146543741697,
-      "grad_norm": 0.0032983017154037952,
-      "learning_rate": 0.00019999686042733266,
-      "loss": 46.0,
-      "step": 15673
-    },
-    {
-      "epoch": 2.524175691452957,
-      "grad_norm": 0.00269155390560627,
-      "learning_rate": 0.0001999968600264316,
-      "loss": 46.0,
-      "step": 15674
-    },
-    {
-      "epoch": 2.524336728531744,
-      "grad_norm": 0.008832383900880814,
-      "learning_rate": 0.00019999685962550498,
-      "loss": 46.0,
-      "step": 15675
-    },
-    {
-      "epoch": 2.5244977656105316,
-      "grad_norm": 0.002459681360051036,
-      "learning_rate": 0.00019999685922455278,
-      "loss": 46.0,
-      "step": 15676
-    },
-    {
-      "epoch": 2.524658802689319,
-      "grad_norm": 0.008503999561071396,
-      "learning_rate": 0.00019999685882357495,
-      "loss": 46.0,
-      "step": 15677
-    },
-    {
-      "epoch": 2.5248198397681065,
-      "grad_norm": 0.001488016452640295,
-      "learning_rate": 0.00019999685842257155,
-      "loss": 46.0,
-      "step": 15678
-    },
-    {
-      "epoch": 2.524980876846894,
-      "grad_norm": 0.001180895371362567,
-      "learning_rate": 0.00019999685802154254,
-      "loss": 46.0,
-      "step": 15679
-    },
-    {
-      "epoch": 2.5251419139256814,
-      "grad_norm": 0.0034441722091287374,
-      "learning_rate": 0.00019999685762048795,
-      "loss": 46.0,
-      "step": 15680
-    },
-    {
-      "epoch": 2.525302951004469,
-      "grad_norm": 0.0035903742536902428,
-      "learning_rate": 0.00019999685721940776,
-      "loss": 46.0,
-      "step": 15681
-    },
-    {
-      "epoch": 2.5254639880832563,
-      "grad_norm": 0.0016748111229389906,
-      "learning_rate": 0.00019999685681830194,
-      "loss": 46.0,
-      "step": 15682
-    },
-    {
-      "epoch": 2.525625025162044,
-      "grad_norm": 0.0018760936800390482,
-      "learning_rate": 0.00019999685641717058,
-      "loss": 46.0,
-      "step": 15683
-    },
-    {
-      "epoch": 2.525786062240831,
-      "grad_norm": 0.00208725081756711,
-      "learning_rate": 0.00019999685601601358,
-      "loss": 46.0,
-      "step": 15684
-    },
-    {
-      "epoch": 2.5259470993196182,
-      "grad_norm": 0.0019400343298912048,
-      "learning_rate": 0.00019999685561483102,
-      "loss": 46.0,
-      "step": 15685
-    },
-    {
-      "epoch": 2.5261081363984057,
-      "grad_norm": 0.0032016015611588955,
-      "learning_rate": 0.00019999685521362285,
-      "loss": 46.0,
-      "step": 15686
-    },
-    {
-      "epoch": 2.526269173477193,
-      "grad_norm": 0.0012566620716825128,
-      "learning_rate": 0.00019999685481238911,
-      "loss": 46.0,
-      "step": 15687
-    },
-    {
-      "epoch": 2.5264302105559806,
-      "grad_norm": 0.002221318194642663,
-      "learning_rate": 0.00019999685441112974,
-      "loss": 46.0,
-      "step": 15688
-    },
-    {
-      "epoch": 2.526591247634768,
-      "grad_norm": 0.0016371557721868157,
-      "learning_rate": 0.0001999968540098448,
-      "loss": 46.0,
-      "step": 15689
-    },
-    {
-      "epoch": 2.526752284713555,
-      "grad_norm": 0.0012239270145073533,
-      "learning_rate": 0.00019999685360853425,
-      "loss": 46.0,
-      "step": 15690
-    },
-    {
-      "epoch": 2.5269133217923425,
-      "grad_norm": 0.001530595705844462,
-      "learning_rate": 0.00019999685320719812,
-      "loss": 46.0,
-      "step": 15691
-    },
-    {
-      "epoch": 2.52707435887113,
-      "grad_norm": 0.0008028285228647292,
-      "learning_rate": 0.0001999968528058364,
-      "loss": 46.0,
-      "step": 15692
-    },
-    {
-      "epoch": 2.5272353959499174,
-      "grad_norm": 0.007375949993729591,
-      "learning_rate": 0.00019999685240444908,
-      "loss": 46.0,
-      "step": 15693
-    },
-    {
-      "epoch": 2.527396433028705,
-      "grad_norm": 0.002284626942127943,
-      "learning_rate": 0.00019999685200303616,
-      "loss": 46.0,
-      "step": 15694
-    },
-    {
-      "epoch": 2.5275574701074923,
-      "grad_norm": 0.0019130881410092115,
-      "learning_rate": 0.00019999685160159764,
-      "loss": 46.0,
-      "step": 15695
-    },
-    {
-      "epoch": 2.5277185071862798,
-      "grad_norm": 0.0006606450770050287,
-      "learning_rate": 0.00019999685120013355,
-      "loss": 46.0,
-      "step": 15696
-    },
-    {
-      "epoch": 2.5278795442650672,
-      "grad_norm": 0.0025943140499293804,
-      "learning_rate": 0.00019999685079864383,
-      "loss": 46.0,
-      "step": 15697
-    },
-    {
-      "epoch": 2.5280405813438547,
-      "grad_norm": 0.0006648980197496712,
-      "learning_rate": 0.00019999685039712856,
-      "loss": 46.0,
-      "step": 15698
-    },
-    {
-      "epoch": 2.5282016184226417,
-      "grad_norm": 0.0018283718964084983,
-      "learning_rate": 0.00019999684999558767,
-      "loss": 46.0,
-      "step": 15699
-    },
-    {
-      "epoch": 2.528362655501429,
-      "grad_norm": 0.0010951616568490863,
-      "learning_rate": 0.0001999968495940212,
-      "loss": 46.0,
-      "step": 15700
-    },
-    {
-      "epoch": 2.5285236925802166,
-      "grad_norm": 0.005087762605398893,
-      "learning_rate": 0.0001999968491924291,
-      "loss": 46.0,
-      "step": 15701
-    },
-    {
-      "epoch": 2.528684729659004,
-      "grad_norm": 0.004206470213830471,
-      "learning_rate": 0.00019999684879081145,
-      "loss": 46.0,
-      "step": 15702
-    },
-    {
-      "epoch": 2.5288457667377915,
-      "grad_norm": 0.0051929946057498455,
-      "learning_rate": 0.0001999968483891682,
-      "loss": 46.0,
-      "step": 15703
-    },
-    {
-      "epoch": 2.5290068038165785,
-      "grad_norm": 0.0040488215163350105,
-      "learning_rate": 0.0001999968479874993,
-      "loss": 46.0,
-      "step": 15704
-    },
-    {
-      "epoch": 2.529167840895366,
-      "grad_norm": 0.007053784094750881,
-      "learning_rate": 0.00019999684758580485,
-      "loss": 46.0,
-      "step": 15705
-    },
-    {
-      "epoch": 2.5293288779741534,
-      "grad_norm": 0.008661018684506416,
-      "learning_rate": 0.00019999684718408482,
-      "loss": 46.0,
-      "step": 15706
-    },
-    {
-      "epoch": 2.529489915052941,
-      "grad_norm": 0.001179200946353376,
-      "learning_rate": 0.00019999684678233918,
-      "loss": 46.0,
-      "step": 15707
-    },
-    {
-      "epoch": 2.5296509521317283,
-      "grad_norm": 0.0025416710413992405,
-      "learning_rate": 0.00019999684638056795,
-      "loss": 46.0,
-      "step": 15708
-    },
-    {
-      "epoch": 2.5298119892105158,
-      "grad_norm": 0.003737241495400667,
-      "learning_rate": 0.0001999968459787711,
-      "loss": 46.0,
-      "step": 15709
-    },
-    {
-      "epoch": 2.529973026289303,
-      "grad_norm": 0.007871951907873154,
-      "learning_rate": 0.0001999968455769487,
-      "loss": 46.0,
-      "step": 15710
-    },
-    {
-      "epoch": 2.5301340633680907,
-      "grad_norm": 0.0009385569137521088,
-      "learning_rate": 0.0001999968451751007,
-      "loss": 46.0,
-      "step": 15711
-    },
-    {
-      "epoch": 2.530295100446878,
-      "grad_norm": 0.0017049235757440329,
-      "learning_rate": 0.0001999968447732271,
-      "loss": 46.0,
-      "step": 15712
-    },
-    {
-      "epoch": 2.5304561375256656,
-      "grad_norm": 0.002088160952553153,
-      "learning_rate": 0.00019999684437132785,
-      "loss": 46.0,
-      "step": 15713
-    },
-    {
-      "epoch": 2.5306171746044526,
-      "grad_norm": 0.0007619318785145879,
-      "learning_rate": 0.00019999684396940305,
-      "loss": 46.0,
-      "step": 15714
-    },
-    {
-      "epoch": 2.53077821168324,
-      "grad_norm": 0.0006408804329112172,
-      "learning_rate": 0.00019999684356745268,
-      "loss": 46.0,
-      "step": 15715
-    },
-    {
-      "epoch": 2.5309392487620275,
-      "grad_norm": 0.004502313677221537,
-      "learning_rate": 0.0001999968431654767,
-      "loss": 46.0,
-      "step": 15716
-    },
-    {
-      "epoch": 2.531100285840815,
-      "grad_norm": 0.0038554640486836433,
-      "learning_rate": 0.0001999968427634751,
-      "loss": 46.0,
-      "step": 15717
-    },
-    {
-      "epoch": 2.5312613229196024,
-      "grad_norm": 0.0018654678715392947,
-      "learning_rate": 0.00019999684236144796,
-      "loss": 46.0,
-      "step": 15718
-    },
-    {
-      "epoch": 2.5314223599983894,
-      "grad_norm": 0.004255733452737331,
-      "learning_rate": 0.00019999684195939516,
-      "loss": 46.0,
-      "step": 15719
-    },
-    {
-      "epoch": 2.531583397077177,
-      "grad_norm": 0.0025084849912673235,
-      "learning_rate": 0.00019999684155731678,
-      "loss": 46.0,
-      "step": 15720
-    },
-    {
-      "epoch": 2.5317444341559643,
-      "grad_norm": 0.004545204341411591,
-      "learning_rate": 0.00019999684115521284,
-      "loss": 46.0,
-      "step": 15721
-    },
-    {
-      "epoch": 2.5319054712347517,
-      "grad_norm": 0.0012935440754517913,
-      "learning_rate": 0.0001999968407530833,
-      "loss": 46.0,
-      "step": 15722
-    },
-    {
-      "epoch": 2.532066508313539,
-      "grad_norm": 0.0109461210668087,
-      "learning_rate": 0.00019999684035092817,
-      "loss": 46.0,
-      "step": 15723
-    },
-    {
-      "epoch": 2.5322275453923266,
-      "grad_norm": 0.0017023389227688313,
-      "learning_rate": 0.00019999683994874744,
-      "loss": 46.0,
-      "step": 15724
-    },
-    {
-      "epoch": 2.532388582471114,
-      "grad_norm": 0.0012963255867362022,
-      "learning_rate": 0.00019999683954654107,
-      "loss": 46.0,
-      "step": 15725
-    },
-    {
-      "epoch": 2.5325496195499015,
-      "grad_norm": 0.004191033076494932,
-      "learning_rate": 0.00019999683914430916,
-      "loss": 46.0,
-      "step": 15726
-    },
-    {
-      "epoch": 2.532710656628689,
-      "grad_norm": 0.0024496472906321287,
-      "learning_rate": 0.00019999683874205164,
-      "loss": 46.0,
-      "step": 15727
-    },
-    {
-      "epoch": 2.532871693707476,
-      "grad_norm": 0.0008307733805850148,
-      "learning_rate": 0.0001999968383397685,
-      "loss": 46.0,
-      "step": 15728
-    },
-    {
-      "epoch": 2.5330327307862635,
-      "grad_norm": 0.0008192848763428628,
-      "learning_rate": 0.0001999968379374598,
-      "loss": 46.0,
-      "step": 15729
-    },
-    {
-      "epoch": 2.533193767865051,
-      "grad_norm": 0.003604383207857609,
-      "learning_rate": 0.0001999968375351255,
-      "loss": 46.0,
-      "step": 15730
-    },
-    {
-      "epoch": 2.5333548049438384,
-      "grad_norm": 0.005528864450752735,
-      "learning_rate": 0.00019999683713276561,
-      "loss": 46.0,
-      "step": 15731
-    },
-    {
-      "epoch": 2.533515842022626,
-      "grad_norm": 0.005030097905546427,
-      "learning_rate": 0.00019999683673038013,
-      "loss": 46.0,
-      "step": 15732
-    },
-    {
-      "epoch": 2.5336768791014133,
-      "grad_norm": 0.0011510084150359035,
-      "learning_rate": 0.00019999683632796904,
-      "loss": 46.0,
-      "step": 15733
-    },
-    {
-      "epoch": 2.5338379161802003,
-      "grad_norm": 0.0024968741927295923,
-      "learning_rate": 0.00019999683592553235,
-      "loss": 46.0,
-      "step": 15734
-    },
-    {
-      "epoch": 2.5339989532589877,
-      "grad_norm": 0.0025567039847373962,
-      "learning_rate": 0.00019999683552307008,
-      "loss": 46.0,
-      "step": 15735
-    },
-    {
-      "epoch": 2.534159990337775,
-      "grad_norm": 0.008249803446233273,
-      "learning_rate": 0.00019999683512058223,
-      "loss": 46.0,
-      "step": 15736
-    },
-    {
-      "epoch": 2.5343210274165626,
-      "grad_norm": 0.006332262884825468,
-      "learning_rate": 0.00019999683471806875,
-      "loss": 46.0,
-      "step": 15737
-    },
-    {
-      "epoch": 2.53448206449535,
-      "grad_norm": 0.0017382610822096467,
-      "learning_rate": 0.00019999683431552972,
-      "loss": 46.0,
-      "step": 15738
-    },
-    {
-      "epoch": 2.5346431015741375,
-      "grad_norm": 0.0012596466112881899,
-      "learning_rate": 0.00019999683391296508,
-      "loss": 46.0,
-      "step": 15739
-    },
-    {
-      "epoch": 2.534804138652925,
-      "grad_norm": 0.0007674636435694993,
-      "learning_rate": 0.00019999683351037482,
-      "loss": 46.0,
-      "step": 15740
-    },
-    {
-      "epoch": 2.5349651757317124,
-      "grad_norm": 0.0037965932860970497,
-      "learning_rate": 0.000199996833107759,
-      "loss": 46.0,
-      "step": 15741
-    },
-    {
-      "epoch": 2.5351262128105,
-      "grad_norm": 0.0009536274010315537,
-      "learning_rate": 0.00019999683270511756,
-      "loss": 46.0,
-      "step": 15742
-    },
-    {
-      "epoch": 2.535287249889287,
-      "grad_norm": 0.004552791360765696,
-      "learning_rate": 0.0001999968323024505,
-      "loss": 46.0,
-      "step": 15743
-    },
-    {
-      "epoch": 2.5354482869680743,
-      "grad_norm": 0.0035578443203121424,
-      "learning_rate": 0.0001999968318997579,
-      "loss": 46.0,
-      "step": 15744
-    },
-    {
-      "epoch": 2.535609324046862,
-      "grad_norm": 0.0018478657584637403,
-      "learning_rate": 0.0001999968314970397,
-      "loss": 46.0,
-      "step": 15745
-    },
-    {
-      "epoch": 2.5357703611256492,
-      "grad_norm": 0.0032003531232476234,
-      "learning_rate": 0.00019999683109429587,
-      "loss": 46.0,
-      "step": 15746
-    },
-    {
-      "epoch": 2.5359313982044367,
-      "grad_norm": 0.0011525305453687906,
-      "learning_rate": 0.00019999683069152647,
-      "loss": 46.0,
-      "step": 15747
-    },
-    {
-      "epoch": 2.5360924352832237,
-      "grad_norm": 0.0024513343814760447,
-      "learning_rate": 0.00019999683028873149,
-      "loss": 46.0,
-      "step": 15748
-    },
-    {
-      "epoch": 2.536253472362011,
-      "grad_norm": 0.0009150298428721726,
-      "learning_rate": 0.0001999968298859109,
-      "loss": 46.0,
-      "step": 15749
-    },
-    {
-      "epoch": 2.5364145094407986,
-      "grad_norm": 0.008602021262049675,
-      "learning_rate": 0.0001999968294830647,
-      "loss": 46.0,
-      "step": 15750
-    },
-    {
-      "epoch": 2.536575546519586,
-      "grad_norm": 0.0025046514347195625,
-      "learning_rate": 0.00019999682908019293,
-      "loss": 46.0,
-      "step": 15751
-    },
-    {
-      "epoch": 2.5367365835983735,
-      "grad_norm": 0.004044221714138985,
-      "learning_rate": 0.00019999682867729557,
-      "loss": 46.0,
-      "step": 15752
-    },
-    {
-      "epoch": 2.536897620677161,
-      "grad_norm": 0.0028303838334977627,
-      "learning_rate": 0.0001999968282743726,
-      "loss": 46.0,
-      "step": 15753
-    },
-    {
-      "epoch": 2.5370586577559484,
-      "grad_norm": 0.003672609804198146,
-      "learning_rate": 0.00019999682787142403,
-      "loss": 46.0,
-      "step": 15754
-    },
-    {
-      "epoch": 2.537219694834736,
-      "grad_norm": 0.0014963069697842002,
-      "learning_rate": 0.00019999682746844988,
-      "loss": 46.0,
-      "step": 15755
-    },
-    {
-      "epoch": 2.5373807319135233,
-      "grad_norm": 0.0022372633684426546,
-      "learning_rate": 0.00019999682706545012,
-      "loss": 46.0,
-      "step": 15756
-    },
-    {
-      "epoch": 2.5375417689923103,
-      "grad_norm": 0.012474932707846165,
-      "learning_rate": 0.00019999682666242477,
-      "loss": 46.0,
-      "step": 15757
-    },
-    {
-      "epoch": 2.5377028060710978,
-      "grad_norm": 0.006394247990101576,
-      "learning_rate": 0.00019999682625937383,
-      "loss": 46.0,
-      "step": 15758
-    },
-    {
-      "epoch": 2.5378638431498852,
-      "grad_norm": 0.0035678413696587086,
-      "learning_rate": 0.0001999968258562973,
-      "loss": 46.0,
-      "step": 15759
-    },
-    {
-      "epoch": 2.5380248802286727,
-      "grad_norm": 0.0031321633141487837,
-      "learning_rate": 0.0001999968254531952,
-      "loss": 46.0,
-      "step": 15760
-    },
-    {
-      "epoch": 2.53818591730746,
-      "grad_norm": 0.0026392501313239336,
-      "learning_rate": 0.00019999682505006746,
-      "loss": 46.0,
-      "step": 15761
-    },
-    {
-      "epoch": 2.5383469543862476,
-      "grad_norm": 0.0024764128029346466,
-      "learning_rate": 0.00019999682464691415,
-      "loss": 46.0,
-      "step": 15762
-    },
-    {
-      "epoch": 2.5385079914650346,
-      "grad_norm": 0.0013382750330492854,
-      "learning_rate": 0.00019999682424373525,
-      "loss": 46.0,
-      "step": 15763
-    },
-    {
-      "epoch": 2.538669028543822,
-      "grad_norm": 0.004467046353965998,
-      "learning_rate": 0.00019999682384053076,
-      "loss": 46.0,
-      "step": 15764
-    },
-    {
-      "epoch": 2.5388300656226095,
-      "grad_norm": 0.009162072092294693,
-      "learning_rate": 0.00019999682343730063,
-      "loss": 46.0,
-      "step": 15765
-    },
-    {
-      "epoch": 2.538991102701397,
-      "grad_norm": 0.00311396480537951,
-      "learning_rate": 0.00019999682303404497,
-      "loss": 46.0,
-      "step": 15766
-    },
-    {
-      "epoch": 2.5391521397801844,
-      "grad_norm": 0.005254401359707117,
-      "learning_rate": 0.00019999682263076367,
-      "loss": 46.0,
-      "step": 15767
-    },
-    {
-      "epoch": 2.539313176858972,
-      "grad_norm": 0.0014897024957463145,
-      "learning_rate": 0.0001999968222274568,
-      "loss": 46.0,
-      "step": 15768
-    },
-    {
-      "epoch": 2.5394742139377593,
-      "grad_norm": 0.004809871781617403,
-      "learning_rate": 0.00019999682182412432,
-      "loss": 46.0,
-      "step": 15769
-    },
-    {
-      "epoch": 2.5396352510165467,
-      "grad_norm": 0.002005638089030981,
-      "learning_rate": 0.00019999682142076629,
-      "loss": 46.0,
-      "step": 15770
-    },
-    {
-      "epoch": 2.539796288095334,
-      "grad_norm": 0.016677599400281906,
-      "learning_rate": 0.0001999968210173826,
-      "loss": 46.0,
-      "step": 15771
-    },
-    {
-      "epoch": 2.539957325174121,
-      "grad_norm": 0.0007019637268967927,
-      "learning_rate": 0.00019999682061397337,
-      "loss": 46.0,
-      "step": 15772
-    },
-    {
-      "epoch": 2.5401183622529087,
-      "grad_norm": 0.0022845263592898846,
-      "learning_rate": 0.00019999682021053849,
-      "loss": 46.0,
-      "step": 15773
-    },
-    {
-      "epoch": 2.540279399331696,
-      "grad_norm": 0.0008481305558234453,
-      "learning_rate": 0.00019999681980707807,
-      "loss": 46.0,
-      "step": 15774
-    },
-    {
-      "epoch": 2.5404404364104836,
-      "grad_norm": 0.0017113655339926481,
-      "learning_rate": 0.00019999681940359202,
-      "loss": 46.0,
-      "step": 15775
-    },
-    {
-      "epoch": 2.540601473489271,
-      "grad_norm": 0.006946255918592215,
-      "learning_rate": 0.0001999968190000804,
-      "loss": 46.0,
-      "step": 15776
-    },
-    {
-      "epoch": 2.540762510568058,
-      "grad_norm": 0.0022909340914338827,
-      "learning_rate": 0.00019999681859654317,
-      "loss": 46.0,
-      "step": 15777
-    },
-    {
-      "epoch": 2.5409235476468455,
-      "grad_norm": 0.0024680462665855885,
-      "learning_rate": 0.00019999681819298038,
-      "loss": 46.0,
-      "step": 15778
-    },
-    {
-      "epoch": 2.541084584725633,
-      "grad_norm": 0.003536262083798647,
-      "learning_rate": 0.00019999681778939195,
-      "loss": 46.0,
-      "step": 15779
-    },
-    {
-      "epoch": 2.5412456218044204,
-      "grad_norm": 0.0143395084887743,
-      "learning_rate": 0.00019999681738577793,
-      "loss": 46.0,
-      "step": 15780
-    },
-    {
-      "epoch": 2.541406658883208,
-      "grad_norm": 0.0016347103519365191,
-      "learning_rate": 0.00019999681698213835,
-      "loss": 46.0,
-      "step": 15781
-    },
-    {
-      "epoch": 2.5415676959619953,
-      "grad_norm": 0.001784398453310132,
-      "learning_rate": 0.00019999681657847313,
-      "loss": 46.0,
-      "step": 15782
-    },
-    {
-      "epoch": 2.5417287330407827,
-      "grad_norm": 0.003143638838082552,
-      "learning_rate": 0.00019999681617478235,
-      "loss": 46.0,
-      "step": 15783
-    },
-    {
-      "epoch": 2.54188977011957,
-      "grad_norm": 0.001713858568109572,
-      "learning_rate": 0.00019999681577106596,
-      "loss": 46.0,
-      "step": 15784
-    },
-    {
-      "epoch": 2.5420508071983576,
-      "grad_norm": 0.000525288050994277,
-      "learning_rate": 0.000199996815367324,
-      "loss": 46.0,
-      "step": 15785
-    },
-    {
-      "epoch": 2.542211844277145,
-      "grad_norm": 0.0009199513588100672,
-      "learning_rate": 0.00019999681496355644,
-      "loss": 46.0,
-      "step": 15786
-    },
-    {
-      "epoch": 2.542372881355932,
-      "grad_norm": 0.0026275611016899347,
-      "learning_rate": 0.00019999681455976326,
-      "loss": 46.0,
-      "step": 15787
-    },
-    {
-      "epoch": 2.5425339184347195,
-      "grad_norm": 0.0056701297871768475,
-      "learning_rate": 0.0001999968141559445,
-      "loss": 46.0,
-      "step": 15788
-    },
-    {
-      "epoch": 2.542694955513507,
-      "grad_norm": 0.004300985485315323,
-      "learning_rate": 0.00019999681375210016,
-      "loss": 46.0,
-      "step": 15789
-    },
-    {
-      "epoch": 2.5428559925922944,
-      "grad_norm": 0.0016123721143230796,
-      "learning_rate": 0.0001999968133482302,
-      "loss": 46.0,
-      "step": 15790
-    },
-    {
-      "epoch": 2.543017029671082,
-      "grad_norm": 0.002945336978882551,
-      "learning_rate": 0.00019999681294433468,
-      "loss": 46.0,
-      "step": 15791
-    },
-    {
-      "epoch": 2.543178066749869,
-      "grad_norm": 0.006586259696632624,
-      "learning_rate": 0.00019999681254041353,
-      "loss": 46.0,
-      "step": 15792
-    },
-    {
-      "epoch": 2.5433391038286564,
-      "grad_norm": 0.005842157639563084,
-      "learning_rate": 0.00019999681213646683,
-      "loss": 46.0,
-      "step": 15793
-    },
-    {
-      "epoch": 2.543500140907444,
-      "grad_norm": 0.0006493720575235784,
-      "learning_rate": 0.00019999681173249448,
-      "loss": 46.0,
-      "step": 15794
-    },
-    {
-      "epoch": 2.5436611779862313,
-      "grad_norm": 0.005443824455142021,
-      "learning_rate": 0.00019999681132849657,
-      "loss": 46.0,
-      "step": 15795
-    },
-    {
-      "epoch": 2.5438222150650187,
-      "grad_norm": 0.016953447833657265,
-      "learning_rate": 0.00019999681092447305,
-      "loss": 46.0,
-      "step": 15796
-    },
-    {
-      "epoch": 2.543983252143806,
-      "grad_norm": 0.0009105742210522294,
-      "learning_rate": 0.00019999681052042397,
-      "loss": 46.0,
-      "step": 15797
-    },
-    {
-      "epoch": 2.5441442892225936,
-      "grad_norm": 0.005569086410105228,
-      "learning_rate": 0.00019999681011634925,
-      "loss": 46.0,
-      "step": 15798
-    },
-    {
-      "epoch": 2.544305326301381,
-      "grad_norm": 0.0008822057861834764,
-      "learning_rate": 0.00019999680971224897,
-      "loss": 46.0,
-      "step": 15799
-    },
-    {
-      "epoch": 2.5444663633801685,
-      "grad_norm": 0.0061337389051914215,
-      "learning_rate": 0.00019999680930812307,
-      "loss": 46.0,
-      "step": 15800
-    },
-    {
-      "epoch": 2.5446274004589555,
-      "grad_norm": 0.005552330985665321,
-      "learning_rate": 0.00019999680890397158,
-      "loss": 46.0,
-      "step": 15801
-    },
-    {
-      "epoch": 2.544788437537743,
-      "grad_norm": 0.005401782225817442,
-      "learning_rate": 0.0001999968084997945,
-      "loss": 46.0,
-      "step": 15802
-    },
-    {
-      "epoch": 2.5449494746165304,
-      "grad_norm": 0.0020015074405819178,
-      "learning_rate": 0.00019999680809559185,
-      "loss": 46.0,
-      "step": 15803
-    },
-    {
-      "epoch": 2.545110511695318,
-      "grad_norm": 0.00097870163153857,
-      "learning_rate": 0.0001999968076913636,
-      "loss": 46.0,
-      "step": 15804
-    },
-    {
-      "epoch": 2.5452715487741053,
-      "grad_norm": 0.003985776100307703,
-      "learning_rate": 0.00019999680728710972,
-      "loss": 46.0,
-      "step": 15805
-    },
-    {
-      "epoch": 2.545432585852893,
-      "grad_norm": 0.0039864699356257915,
-      "learning_rate": 0.00019999680688283027,
-      "loss": 46.0,
-      "step": 15806
-    },
-    {
-      "epoch": 2.54559362293168,
-      "grad_norm": 0.010370858013629913,
-      "learning_rate": 0.00019999680647852524,
-      "loss": 46.0,
-      "step": 15807
-    },
-    {
-      "epoch": 2.5457546600104672,
-      "grad_norm": 0.0010867168894037604,
-      "learning_rate": 0.0001999968060741946,
-      "loss": 46.0,
-      "step": 15808
-    },
-    {
-      "epoch": 2.5459156970892547,
-      "grad_norm": 0.006409656722098589,
-      "learning_rate": 0.00019999680566983835,
-      "loss": 46.0,
-      "step": 15809
-    },
-    {
-      "epoch": 2.546076734168042,
-      "grad_norm": 0.007434573955833912,
-      "learning_rate": 0.00019999680526545653,
-      "loss": 46.0,
-      "step": 15810
-    },
-    {
-      "epoch": 2.5462377712468296,
-      "grad_norm": 0.0016469808761030436,
-      "learning_rate": 0.00019999680486104912,
-      "loss": 46.0,
-      "step": 15811
-    },
-    {
-      "epoch": 2.546398808325617,
-      "grad_norm": 0.00819756742566824,
-      "learning_rate": 0.0001999968044566161,
-      "loss": 46.0,
-      "step": 15812
-    },
-    {
-      "epoch": 2.5465598454044045,
-      "grad_norm": 0.0026401225477457047,
-      "learning_rate": 0.00019999680405215748,
-      "loss": 46.0,
-      "step": 15813
-    },
-    {
-      "epoch": 2.546720882483192,
-      "grad_norm": 0.001077825902029872,
-      "learning_rate": 0.00019999680364767328,
-      "loss": 46.0,
-      "step": 15814
-    },
-    {
-      "epoch": 2.5468819195619794,
-      "grad_norm": 0.001470989198423922,
-      "learning_rate": 0.00019999680324316347,
-      "loss": 46.0,
-      "step": 15815
-    },
-    {
-      "epoch": 2.5470429566407664,
-      "grad_norm": 0.005357936955988407,
-      "learning_rate": 0.0001999968028386281,
-      "loss": 46.0,
-      "step": 15816
-    },
-    {
-      "epoch": 2.547203993719554,
-      "grad_norm": 0.0006705376436002553,
-      "learning_rate": 0.0001999968024340671,
-      "loss": 46.0,
-      "step": 15817
-    },
-    {
-      "epoch": 2.5473650307983413,
-      "grad_norm": 0.010283122770488262,
-      "learning_rate": 0.00019999680202948054,
-      "loss": 46.0,
-      "step": 15818
-    },
-    {
-      "epoch": 2.5475260678771288,
-      "grad_norm": 0.002999626798555255,
-      "learning_rate": 0.00019999680162486835,
-      "loss": 46.0,
-      "step": 15819
-    },
-    {
-      "epoch": 2.547687104955916,
-      "grad_norm": 0.0015617627650499344,
-      "learning_rate": 0.0001999968012202306,
-      "loss": 46.0,
-      "step": 15820
-    },
-    {
-      "epoch": 2.5478481420347032,
-      "grad_norm": 0.003764509689062834,
-      "learning_rate": 0.0001999968008155672,
-      "loss": 46.0,
-      "step": 15821
-    },
-    {
-      "epoch": 2.5480091791134907,
-      "grad_norm": 0.0030546344351023436,
-      "learning_rate": 0.00019999680041087828,
-      "loss": 46.0,
-      "step": 15822
-    },
-    {
-      "epoch": 2.548170216192278,
-      "grad_norm": 0.0035476309712976217,
-      "learning_rate": 0.00019999680000616372,
-      "loss": 46.0,
-      "step": 15823
-    },
-    {
-      "epoch": 2.5483312532710656,
-      "grad_norm": 0.004039666149765253,
-      "learning_rate": 0.0001999967996014236,
-      "loss": 46.0,
-      "step": 15824
-    },
-    {
-      "epoch": 2.548492290349853,
-      "grad_norm": 0.002095070667564869,
-      "learning_rate": 0.00019999679919665783,
-      "loss": 46.0,
-      "step": 15825
-    },
-    {
-      "epoch": 2.5486533274286405,
-      "grad_norm": 0.009739681147038937,
-      "learning_rate": 0.0001999967987918665,
-      "loss": 46.0,
-      "step": 15826
-    },
-    {
-      "epoch": 2.548814364507428,
-      "grad_norm": 0.0004628387396223843,
-      "learning_rate": 0.00019999679838704956,
-      "loss": 46.0,
-      "step": 15827
-    },
-    {
-      "epoch": 2.5489754015862154,
-      "grad_norm": 0.0025958239566534758,
-      "learning_rate": 0.00019999679798220706,
-      "loss": 46.0,
-      "step": 15828
-    },
-    {
-      "epoch": 2.549136438665003,
-      "grad_norm": 0.0018363907001912594,
-      "learning_rate": 0.00019999679757733895,
-      "loss": 46.0,
-      "step": 15829
-    },
-    {
-      "epoch": 2.54929747574379,
-      "grad_norm": 0.007587160915136337,
-      "learning_rate": 0.00019999679717244524,
-      "loss": 46.0,
-      "step": 15830
-    },
-    {
-      "epoch": 2.5494585128225773,
-      "grad_norm": 0.0028079207986593246,
-      "learning_rate": 0.00019999679676752593,
-      "loss": 46.0,
-      "step": 15831
-    },
-    {
-      "epoch": 2.5496195499013647,
-      "grad_norm": 0.0011105615412816405,
-      "learning_rate": 0.00019999679636258105,
-      "loss": 46.0,
-      "step": 15832
-    },
-    {
-      "epoch": 2.549780586980152,
-      "grad_norm": 0.005107423290610313,
-      "learning_rate": 0.00019999679595761056,
-      "loss": 46.0,
-      "step": 15833
-    },
-    {
-      "epoch": 2.5499416240589396,
-      "grad_norm": 0.0034990699496120214,
-      "learning_rate": 0.00019999679555261445,
-      "loss": 46.0,
-      "step": 15834
-    },
-    {
-      "epoch": 2.550102661137727,
-      "grad_norm": 0.01129177026450634,
-      "learning_rate": 0.00019999679514759276,
-      "loss": 46.0,
-      "step": 15835
-    },
-    {
-      "epoch": 2.550263698216514,
-      "grad_norm": 0.005279368255287409,
-      "learning_rate": 0.0001999967947425455,
-      "loss": 46.0,
-      "step": 15836
-    },
-    {
-      "epoch": 2.5504247352953016,
-      "grad_norm": 0.0017063929699361324,
-      "learning_rate": 0.00019999679433747262,
-      "loss": 46.0,
-      "step": 15837
-    },
-    {
-      "epoch": 2.550585772374089,
-      "grad_norm": 0.0009304551058448851,
-      "learning_rate": 0.0001999967939323742,
-      "loss": 46.0,
-      "step": 15838
-    },
-    {
-      "epoch": 2.5507468094528765,
-      "grad_norm": 0.01194256916642189,
-      "learning_rate": 0.00019999679352725012,
-      "loss": 46.0,
-      "step": 15839
-    },
-    {
-      "epoch": 2.550907846531664,
-      "grad_norm": 0.001702172332443297,
-      "learning_rate": 0.00019999679312210047,
-      "loss": 46.0,
-      "step": 15840
-    },
-    {
-      "epoch": 2.5510688836104514,
-      "grad_norm": 0.0012780199758708477,
-      "learning_rate": 0.00019999679271692525,
-      "loss": 46.0,
-      "step": 15841
-    },
-    {
-      "epoch": 2.551229920689239,
-      "grad_norm": 0.0014256832655519247,
-      "learning_rate": 0.0001999967923117244,
-      "loss": 46.0,
-      "step": 15842
-    },
-    {
-      "epoch": 2.5513909577680263,
-      "grad_norm": 0.005153740756213665,
-      "learning_rate": 0.00019999679190649795,
-      "loss": 46.0,
-      "step": 15843
-    },
-    {
-      "epoch": 2.5515519948468137,
-      "grad_norm": 0.003195536555722356,
-      "learning_rate": 0.00019999679150124595,
-      "loss": 46.0,
-      "step": 15844
-    },
-    {
-      "epoch": 2.5517130319256007,
-      "grad_norm": 0.007489427458494902,
-      "learning_rate": 0.00019999679109596833,
-      "loss": 46.0,
-      "step": 15845
-    },
-    {
-      "epoch": 2.551874069004388,
-      "grad_norm": 0.001013877335935831,
-      "learning_rate": 0.0001999967906906651,
-      "loss": 46.0,
-      "step": 15846
-    },
-    {
-      "epoch": 2.5520351060831756,
-      "grad_norm": 0.0009140408365055919,
-      "learning_rate": 0.0001999967902853363,
-      "loss": 46.0,
-      "step": 15847
-    },
-    {
-      "epoch": 2.552196143161963,
-      "grad_norm": 0.0032715012785047293,
-      "learning_rate": 0.0001999967898799819,
-      "loss": 46.0,
-      "step": 15848
-    },
-    {
-      "epoch": 2.5523571802407505,
-      "grad_norm": 0.004841975402086973,
-      "learning_rate": 0.0001999967894746019,
-      "loss": 46.0,
-      "step": 15849
-    },
-    {
-      "epoch": 2.552518217319538,
-      "grad_norm": 0.004513719584792852,
-      "learning_rate": 0.00019999678906919632,
-      "loss": 46.0,
-      "step": 15850
-    },
-    {
-      "epoch": 2.552679254398325,
-      "grad_norm": 0.0011710169492289424,
-      "learning_rate": 0.00019999678866376513,
-      "loss": 46.0,
-      "step": 15851
-    },
-    {
-      "epoch": 2.5528402914771124,
-      "grad_norm": 0.004223324358463287,
-      "learning_rate": 0.00019999678825830834,
-      "loss": 46.0,
-      "step": 15852
-    },
-    {
-      "epoch": 2.5530013285559,
-      "grad_norm": 0.013558804988861084,
-      "learning_rate": 0.00019999678785282598,
-      "loss": 46.0,
-      "step": 15853
-    },
-    {
-      "epoch": 2.5531623656346873,
-      "grad_norm": 0.01734413020312786,
-      "learning_rate": 0.00019999678744731802,
-      "loss": 46.0,
-      "step": 15854
-    },
-    {
-      "epoch": 2.553323402713475,
-      "grad_norm": 0.0027816323563456535,
-      "learning_rate": 0.00019999678704178447,
-      "loss": 46.0,
-      "step": 15855
-    },
-    {
-      "epoch": 2.5534844397922623,
-      "grad_norm": 0.010898136533796787,
-      "learning_rate": 0.0001999967866362253,
-      "loss": 46.0,
-      "step": 15856
-    },
-    {
-      "epoch": 2.5536454768710497,
-      "grad_norm": 0.009072975255548954,
-      "learning_rate": 0.00019999678623064054,
-      "loss": 46.0,
-      "step": 15857
-    },
-    {
-      "epoch": 2.553806513949837,
-      "grad_norm": 0.0017958837561309338,
-      "learning_rate": 0.00019999678582503024,
-      "loss": 46.0,
-      "step": 15858
-    },
-    {
-      "epoch": 2.5539675510286246,
-      "grad_norm": 0.0006768766324967146,
-      "learning_rate": 0.0001999967854193943,
-      "loss": 46.0,
-      "step": 15859
-    },
-    {
-      "epoch": 2.5541285881074116,
-      "grad_norm": 0.006000916939228773,
-      "learning_rate": 0.00019999678501373276,
-      "loss": 46.0,
-      "step": 15860
-    },
-    {
-      "epoch": 2.554289625186199,
-      "grad_norm": 0.005242292769253254,
-      "learning_rate": 0.00019999678460804564,
-      "loss": 46.0,
-      "step": 15861
-    },
-    {
-      "epoch": 2.5544506622649865,
-      "grad_norm": 0.009133460000157356,
-      "learning_rate": 0.00019999678420233293,
-      "loss": 46.0,
-      "step": 15862
-    },
-    {
-      "epoch": 2.554611699343774,
-      "grad_norm": 0.0031896091531962156,
-      "learning_rate": 0.00019999678379659458,
-      "loss": 46.0,
-      "step": 15863
-    },
-    {
-      "epoch": 2.5547727364225614,
-      "grad_norm": 0.0012896254193037748,
-      "learning_rate": 0.00019999678339083067,
-      "loss": 46.0,
-      "step": 15864
-    },
-    {
-      "epoch": 2.5549337735013484,
-      "grad_norm": 0.001649995450861752,
-      "learning_rate": 0.0001999967829850412,
-      "loss": 46.0,
-      "step": 15865
-    },
-    {
-      "epoch": 2.555094810580136,
-      "grad_norm": 0.004237490706145763,
-      "learning_rate": 0.00019999678257922609,
-      "loss": 46.0,
-      "step": 15866
-    },
-    {
-      "epoch": 2.5552558476589233,
-      "grad_norm": 0.001801868318580091,
-      "learning_rate": 0.00019999678217338541,
-      "loss": 46.0,
-      "step": 15867
-    },
-    {
-      "epoch": 2.555416884737711,
-      "grad_norm": 0.0009800817351788282,
-      "learning_rate": 0.00019999678176751913,
-      "loss": 46.0,
-      "step": 15868
-    },
-    {
-      "epoch": 2.5555779218164982,
-      "grad_norm": 0.0007903212099336088,
-      "learning_rate": 0.00019999678136162726,
-      "loss": 46.0,
-      "step": 15869
-    },
-    {
-      "epoch": 2.5557389588952857,
-      "grad_norm": 0.0025581642985343933,
-      "learning_rate": 0.00019999678095570977,
-      "loss": 46.0,
-      "step": 15870
-    },
-    {
-      "epoch": 2.555899995974073,
-      "grad_norm": 0.01888599619269371,
-      "learning_rate": 0.00019999678054976672,
-      "loss": 46.0,
-      "step": 15871
-    },
-    {
-      "epoch": 2.5560610330528606,
-      "grad_norm": 0.000634224561508745,
-      "learning_rate": 0.00019999678014379806,
-      "loss": 46.0,
-      "step": 15872
-    },
-    {
-      "epoch": 2.556222070131648,
-      "grad_norm": 0.004928269423544407,
-      "learning_rate": 0.0001999967797378038,
-      "loss": 46.0,
-      "step": 15873
-    },
-    {
-      "epoch": 2.556383107210435,
-      "grad_norm": 0.009454751387238503,
-      "learning_rate": 0.00019999677933178395,
-      "loss": 46.0,
-      "step": 15874
-    },
-    {
-      "epoch": 2.5565441442892225,
-      "grad_norm": 0.005087968427687883,
-      "learning_rate": 0.00019999677892573853,
-      "loss": 46.0,
-      "step": 15875
-    },
-    {
-      "epoch": 2.55670518136801,
-      "grad_norm": 0.0016252817586064339,
-      "learning_rate": 0.00019999677851966746,
-      "loss": 46.0,
-      "step": 15876
-    },
-    {
-      "epoch": 2.5568662184467974,
-      "grad_norm": 0.0022396978456526995,
-      "learning_rate": 0.00019999677811357084,
-      "loss": 46.0,
-      "step": 15877
-    },
-    {
-      "epoch": 2.557027255525585,
-      "grad_norm": 0.001236640033312142,
-      "learning_rate": 0.00019999677770744865,
-      "loss": 46.0,
-      "step": 15878
-    },
-    {
-      "epoch": 2.5571882926043723,
-      "grad_norm": 0.0020077975932508707,
-      "learning_rate": 0.00019999677730130083,
-      "loss": 46.0,
-      "step": 15879
-    },
-    {
-      "epoch": 2.5573493296831593,
-      "grad_norm": 0.0029608351178467274,
-      "learning_rate": 0.00019999677689512741,
-      "loss": 46.0,
-      "step": 15880
-    },
-    {
-      "epoch": 2.5575103667619468,
-      "grad_norm": 0.0064052557572722435,
-      "learning_rate": 0.0001999967764889284,
-      "loss": 46.0,
-      "step": 15881
-    },
-    {
-      "epoch": 2.557671403840734,
-      "grad_norm": 0.0032616094686090946,
-      "learning_rate": 0.0001999967760827038,
-      "loss": 46.0,
-      "step": 15882
-    },
-    {
-      "epoch": 2.5578324409195217,
-      "grad_norm": 0.0058501046150922775,
-      "learning_rate": 0.0001999967756764536,
-      "loss": 46.0,
-      "step": 15883
-    },
-    {
-      "epoch": 2.557993477998309,
-      "grad_norm": 0.004450731445103884,
-      "learning_rate": 0.0001999967752701778,
-      "loss": 46.0,
-      "step": 15884
-    },
-    {
-      "epoch": 2.5581545150770966,
-      "grad_norm": 0.002734450390562415,
-      "learning_rate": 0.00019999677486387643,
-      "loss": 46.0,
-      "step": 15885
-    },
-    {
-      "epoch": 2.558315552155884,
-      "grad_norm": 0.004494440741837025,
-      "learning_rate": 0.00019999677445754944,
-      "loss": 46.0,
-      "step": 15886
-    },
-    {
-      "epoch": 2.5584765892346715,
-      "grad_norm": 0.0019422023324295878,
-      "learning_rate": 0.0001999967740511969,
-      "loss": 46.0,
-      "step": 15887
-    },
-    {
-      "epoch": 2.558637626313459,
-      "grad_norm": 0.0033686214592307806,
-      "learning_rate": 0.00019999677364481873,
-      "loss": 46.0,
-      "step": 15888
-    },
-    {
-      "epoch": 2.558798663392246,
-      "grad_norm": 0.00145612598862499,
-      "learning_rate": 0.00019999677323841498,
-      "loss": 46.0,
-      "step": 15889
-    },
-    {
-      "epoch": 2.5589597004710334,
-      "grad_norm": 0.0010881536873057485,
-      "learning_rate": 0.00019999677283198564,
-      "loss": 46.0,
-      "step": 15890
-    },
-    {
-      "epoch": 2.559120737549821,
-      "grad_norm": 0.0022914018481969833,
-      "learning_rate": 0.00019999677242553068,
-      "loss": 46.0,
-      "step": 15891
-    },
-    {
-      "epoch": 2.5592817746286083,
-      "grad_norm": 0.0010935964528471231,
-      "learning_rate": 0.00019999677201905014,
-      "loss": 46.0,
-      "step": 15892
-    },
-    {
-      "epoch": 2.5594428117073957,
-      "grad_norm": 0.00904909148812294,
-      "learning_rate": 0.00019999677161254402,
-      "loss": 46.0,
-      "step": 15893
-    },
-    {
-      "epoch": 2.5596038487861827,
-      "grad_norm": 0.00311499391682446,
-      "learning_rate": 0.00019999677120601227,
-      "loss": 46.0,
-      "step": 15894
-    },
-    {
-      "epoch": 2.55976488586497,
-      "grad_norm": 0.0005413953913375735,
-      "learning_rate": 0.00019999677079945494,
-      "loss": 46.0,
-      "step": 15895
-    },
-    {
-      "epoch": 2.5599259229437576,
-      "grad_norm": 0.00992573518306017,
-      "learning_rate": 0.00019999677039287206,
-      "loss": 46.0,
-      "step": 15896
-    },
-    {
-      "epoch": 2.560086960022545,
-      "grad_norm": 0.0009334718342870474,
-      "learning_rate": 0.00019999676998626352,
-      "loss": 46.0,
-      "step": 15897
-    },
-    {
-      "epoch": 2.5602479971013326,
-      "grad_norm": 0.0042174989357590675,
-      "learning_rate": 0.00019999676957962943,
-      "loss": 46.0,
-      "step": 15898
-    },
-    {
-      "epoch": 2.56040903418012,
-      "grad_norm": 0.002316203899681568,
-      "learning_rate": 0.00019999676917296976,
-      "loss": 46.0,
-      "step": 15899
-    },
-    {
-      "epoch": 2.5605700712589075,
-      "grad_norm": 0.0018187230452895164,
-      "learning_rate": 0.00019999676876628444,
-      "loss": 46.0,
-      "step": 15900
-    },
-    {
-      "epoch": 2.560731108337695,
-      "grad_norm": 0.0043976749293506145,
-      "learning_rate": 0.00019999676835957356,
-      "loss": 46.0,
-      "step": 15901
-    },
-    {
-      "epoch": 2.5608921454164824,
-      "grad_norm": 0.0006605296512134373,
-      "learning_rate": 0.0001999967679528371,
-      "loss": 46.0,
-      "step": 15902
-    },
-    {
-      "epoch": 2.56105318249527,
-      "grad_norm": 0.0021460733842104673,
-      "learning_rate": 0.00019999676754607498,
-      "loss": 46.0,
-      "step": 15903
-    },
-    {
-      "epoch": 2.561214219574057,
-      "grad_norm": 0.004835005849599838,
-      "learning_rate": 0.00019999676713928732,
-      "loss": 46.0,
-      "step": 15904
-    },
-    {
-      "epoch": 2.5613752566528443,
-      "grad_norm": 0.000948254601098597,
-      "learning_rate": 0.00019999676673247406,
-      "loss": 46.0,
-      "step": 15905
-    },
-    {
-      "epoch": 2.5615362937316317,
-      "grad_norm": 0.0014655952109023929,
-      "learning_rate": 0.00019999676632563522,
-      "loss": 46.0,
-      "step": 15906
-    },
-    {
-      "epoch": 2.561697330810419,
-      "grad_norm": 0.0012304936535656452,
-      "learning_rate": 0.00019999676591877076,
-      "loss": 46.0,
-      "step": 15907
-    },
-    {
-      "epoch": 2.5618583678892066,
-      "grad_norm": 0.0010085111716762185,
-      "learning_rate": 0.00019999676551188072,
-      "loss": 46.0,
-      "step": 15908
-    },
-    {
-      "epoch": 2.5620194049679936,
-      "grad_norm": 0.0017989878542721272,
-      "learning_rate": 0.0001999967651049651,
-      "loss": 46.0,
-      "step": 15909
-    },
-    {
-      "epoch": 2.562180442046781,
-      "grad_norm": 0.0007498011691495776,
-      "learning_rate": 0.00019999676469802387,
-      "loss": 46.0,
-      "step": 15910
-    },
-    {
-      "epoch": 2.5623414791255685,
-      "grad_norm": 0.00799358170479536,
-      "learning_rate": 0.00019999676429105704,
-      "loss": 46.0,
-      "step": 15911
-    },
-    {
-      "epoch": 2.562502516204356,
-      "grad_norm": 0.0011987709440290928,
-      "learning_rate": 0.00019999676388406462,
-      "loss": 46.0,
-      "step": 15912
-    },
-    {
-      "epoch": 2.5626635532831434,
-      "grad_norm": 0.0016713483491912484,
-      "learning_rate": 0.00019999676347704659,
-      "loss": 46.0,
-      "step": 15913
-    },
-    {
-      "epoch": 2.562824590361931,
-      "grad_norm": 0.0010146618587896228,
-      "learning_rate": 0.000199996763070003,
-      "loss": 46.0,
-      "step": 15914
-    },
-    {
-      "epoch": 2.5629856274407183,
-      "grad_norm": 0.0006958627491258085,
-      "learning_rate": 0.00019999676266293376,
-      "loss": 46.0,
-      "step": 15915
-    },
-    {
-      "epoch": 2.563146664519506,
-      "grad_norm": 0.002969681750983,
-      "learning_rate": 0.000199996762255839,
-      "loss": 46.0,
-      "step": 15916
-    },
-    {
-      "epoch": 2.5633077015982932,
-      "grad_norm": 0.00719532510265708,
-      "learning_rate": 0.00019999676184871858,
-      "loss": 46.0,
-      "step": 15917
-    },
-    {
-      "epoch": 2.5634687386770802,
-      "grad_norm": 0.00809550005942583,
-      "learning_rate": 0.0001999967614415726,
-      "loss": 46.0,
-      "step": 15918
-    },
-    {
-      "epoch": 2.5636297757558677,
-      "grad_norm": 0.0008598658023402095,
-      "learning_rate": 0.00019999676103440103,
-      "loss": 46.0,
-      "step": 15919
-    },
-    {
-      "epoch": 2.563790812834655,
-      "grad_norm": 0.0015236702747642994,
-      "learning_rate": 0.00019999676062720388,
-      "loss": 46.0,
-      "step": 15920
-    },
-    {
-      "epoch": 2.5639518499134426,
-      "grad_norm": 0.007354318164288998,
-      "learning_rate": 0.00019999676021998107,
-      "loss": 46.0,
-      "step": 15921
-    },
-    {
-      "epoch": 2.56411288699223,
-      "grad_norm": 0.004132983740419149,
-      "learning_rate": 0.00019999675981273272,
-      "loss": 46.0,
-      "step": 15922
-    },
-    {
-      "epoch": 2.5642739240710175,
-      "grad_norm": 0.0035002650693058968,
-      "learning_rate": 0.00019999675940545876,
-      "loss": 46.0,
-      "step": 15923
-    },
-    {
-      "epoch": 2.5644349611498045,
-      "grad_norm": 0.007502646651118994,
-      "learning_rate": 0.00019999675899815925,
-      "loss": 46.0,
-      "step": 15924
-    },
-    {
-      "epoch": 2.564595998228592,
-      "grad_norm": 0.007813737727701664,
-      "learning_rate": 0.00019999675859083408,
-      "loss": 46.0,
-      "step": 15925
-    },
-    {
-      "epoch": 2.5647570353073794,
-      "grad_norm": 0.00215246737934649,
-      "learning_rate": 0.00019999675818348336,
-      "loss": 46.0,
-      "step": 15926
-    },
-    {
-      "epoch": 2.564918072386167,
-      "grad_norm": 0.002319880062714219,
-      "learning_rate": 0.00019999675777610703,
-      "loss": 46.0,
-      "step": 15927
-    },
-    {
-      "epoch": 2.5650791094649543,
-      "grad_norm": 0.0009606433450244367,
-      "learning_rate": 0.0001999967573687051,
-      "loss": 46.0,
-      "step": 15928
-    },
-    {
-      "epoch": 2.5652401465437418,
-      "grad_norm": 0.0005641624447889626,
-      "learning_rate": 0.00019999675696127757,
-      "loss": 46.0,
-      "step": 15929
-    },
-    {
-      "epoch": 2.5654011836225292,
-      "grad_norm": 0.0009018388227559626,
-      "learning_rate": 0.00019999675655382447,
-      "loss": 46.0,
-      "step": 15930
-    },
-    {
-      "epoch": 2.5655622207013167,
-      "grad_norm": 0.004454219713807106,
-      "learning_rate": 0.00019999675614634573,
-      "loss": 46.0,
-      "step": 15931
-    },
-    {
-      "epoch": 2.565723257780104,
-      "grad_norm": 0.0028883586637675762,
-      "learning_rate": 0.00019999675573884144,
-      "loss": 46.0,
-      "step": 15932
-    },
-    {
-      "epoch": 2.565884294858891,
-      "grad_norm": 0.009811880998313427,
-      "learning_rate": 0.00019999675533131158,
-      "loss": 46.0,
-      "step": 15933
-    },
-    {
-      "epoch": 2.5660453319376786,
-      "grad_norm": 0.0022590751759707928,
-      "learning_rate": 0.00019999675492375608,
-      "loss": 46.0,
-      "step": 15934
-    },
-    {
-      "epoch": 2.566206369016466,
-      "grad_norm": 0.002512830076739192,
-      "learning_rate": 0.000199996754516175,
-      "loss": 46.0,
-      "step": 15935
-    },
-    {
-      "epoch": 2.5663674060952535,
-      "grad_norm": 0.0011828721035271883,
-      "learning_rate": 0.00019999675410856832,
-      "loss": 46.0,
-      "step": 15936
-    },
-    {
-      "epoch": 2.566528443174041,
-      "grad_norm": 0.004234505817294121,
-      "learning_rate": 0.00019999675370093605,
-      "loss": 46.0,
-      "step": 15937
-    },
-    {
-      "epoch": 2.566689480252828,
-      "grad_norm": 0.0024975440464913845,
-      "learning_rate": 0.00019999675329327818,
-      "loss": 46.0,
-      "step": 15938
-    },
-    {
-      "epoch": 2.5668505173316154,
-      "grad_norm": 0.006123023573309183,
-      "learning_rate": 0.00019999675288559472,
-      "loss": 46.0,
-      "step": 15939
-    },
-    {
-      "epoch": 2.567011554410403,
-      "grad_norm": 0.0021886578761041164,
-      "learning_rate": 0.00019999675247788567,
-      "loss": 46.0,
-      "step": 15940
-    },
-    {
-      "epoch": 2.5671725914891903,
-      "grad_norm": 0.0011832137824967504,
-      "learning_rate": 0.00019999675207015103,
-      "loss": 46.0,
-      "step": 15941
-    },
-    {
-      "epoch": 2.5673336285679778,
-      "grad_norm": 0.0157995093613863,
-      "learning_rate": 0.00019999675166239078,
-      "loss": 46.0,
-      "step": 15942
-    },
-    {
-      "epoch": 2.567494665646765,
-      "grad_norm": 0.001965870847925544,
-      "learning_rate": 0.00019999675125460494,
-      "loss": 46.0,
-      "step": 15943
-    },
-    {
-      "epoch": 2.5676557027255527,
-      "grad_norm": 0.0033829710446298122,
-      "learning_rate": 0.00019999675084679351,
-      "loss": 46.0,
-      "step": 15944
-    },
-    {
-      "epoch": 2.56781673980434,
-      "grad_norm": 0.0032487178687006235,
-      "learning_rate": 0.0001999967504389565,
-      "loss": 46.0,
-      "step": 15945
-    },
-    {
-      "epoch": 2.5679777768831276,
-      "grad_norm": 0.0016249909531325102,
-      "learning_rate": 0.00019999675003109387,
-      "loss": 46.0,
-      "step": 15946
-    },
-    {
-      "epoch": 2.5681388139619146,
-      "grad_norm": 0.004730153363198042,
-      "learning_rate": 0.00019999674962320566,
-      "loss": 46.0,
-      "step": 15947
-    },
-    {
-      "epoch": 2.568299851040702,
-      "grad_norm": 0.0010158108780160546,
-      "learning_rate": 0.00019999674921529186,
-      "loss": 46.0,
-      "step": 15948
-    },
-    {
-      "epoch": 2.5684608881194895,
-      "grad_norm": 0.002735650632530451,
-      "learning_rate": 0.00019999674880735247,
-      "loss": 46.0,
-      "step": 15949
-    },
-    {
-      "epoch": 2.568621925198277,
-      "grad_norm": 0.008639516308903694,
-      "learning_rate": 0.00019999674839938747,
-      "loss": 46.0,
-      "step": 15950
-    },
-    {
-      "epoch": 2.5687829622770644,
-      "grad_norm": 0.0016998723149299622,
-      "learning_rate": 0.00019999674799139688,
-      "loss": 46.0,
-      "step": 15951
-    },
-    {
-      "epoch": 2.568943999355852,
-      "grad_norm": 0.0037123789079487324,
-      "learning_rate": 0.0001999967475833807,
-      "loss": 46.0,
-      "step": 15952
-    },
-    {
-      "epoch": 2.569105036434639,
-      "grad_norm": 0.0032129765022546053,
-      "learning_rate": 0.0001999967471753389,
-      "loss": 46.0,
-      "step": 15953
-    },
-    {
-      "epoch": 2.5692660735134263,
-      "grad_norm": 0.002516761887818575,
-      "learning_rate": 0.00019999674676727156,
-      "loss": 46.0,
-      "step": 15954
-    },
-    {
-      "epoch": 2.5694271105922137,
-      "grad_norm": 0.0021904422901570797,
-      "learning_rate": 0.0001999967463591786,
-      "loss": 46.0,
-      "step": 15955
-    },
-    {
-      "epoch": 2.569588147671001,
-      "grad_norm": 0.0009876337135210633,
-      "learning_rate": 0.00019999674595106,
-      "loss": 46.0,
-      "step": 15956
-    },
-    {
-      "epoch": 2.5697491847497886,
-      "grad_norm": 0.005452492740005255,
-      "learning_rate": 0.00019999674554291587,
-      "loss": 46.0,
-      "step": 15957
-    },
-    {
-      "epoch": 2.569910221828576,
-      "grad_norm": 0.0014200947480276227,
-      "learning_rate": 0.00019999674513474612,
-      "loss": 46.0,
-      "step": 15958
-    },
-    {
-      "epoch": 2.5700712589073635,
-      "grad_norm": 0.0021983920596539974,
-      "learning_rate": 0.0001999967447265508,
-      "loss": 46.0,
-      "step": 15959
-    },
-    {
-      "epoch": 2.570232295986151,
-      "grad_norm": 0.002464075107127428,
-      "learning_rate": 0.00019999674431832984,
-      "loss": 46.0,
-      "step": 15960
-    },
-    {
-      "epoch": 2.5703933330649384,
-      "grad_norm": 0.0007721377769485116,
-      "learning_rate": 0.00019999674391008333,
-      "loss": 46.0,
-      "step": 15961
-    },
-    {
-      "epoch": 2.5705543701437255,
-      "grad_norm": 0.002863567089661956,
-      "learning_rate": 0.00019999674350181123,
-      "loss": 46.0,
-      "step": 15962
-    },
-    {
-      "epoch": 2.570715407222513,
-      "grad_norm": 0.002807021141052246,
-      "learning_rate": 0.00019999674309351348,
-      "loss": 46.0,
-      "step": 15963
-    },
-    {
-      "epoch": 2.5708764443013004,
-      "grad_norm": 0.002399713033810258,
-      "learning_rate": 0.00019999674268519018,
-      "loss": 46.0,
-      "step": 15964
-    },
-    {
-      "epoch": 2.571037481380088,
-      "grad_norm": 0.006840342655777931,
-      "learning_rate": 0.00019999674227684126,
-      "loss": 46.0,
-      "step": 15965
-    },
-    {
-      "epoch": 2.5711985184588753,
-      "grad_norm": 0.0045151119120419025,
-      "learning_rate": 0.00019999674186846678,
-      "loss": 46.0,
-      "step": 15966
-    },
-    {
-      "epoch": 2.5713595555376627,
-      "grad_norm": 0.0021470310166478157,
-      "learning_rate": 0.00019999674146006669,
-      "loss": 46.0,
-      "step": 15967
-    },
-    {
-      "epoch": 2.5715205926164497,
-      "grad_norm": 0.0029068486765027046,
-      "learning_rate": 0.000199996741051641,
-      "loss": 46.0,
-      "step": 15968
-    },
-    {
-      "epoch": 2.571681629695237,
-      "grad_norm": 0.002300978172570467,
-      "learning_rate": 0.0001999967406431897,
-      "loss": 46.0,
-      "step": 15969
-    },
-    {
-      "epoch": 2.5718426667740246,
-      "grad_norm": 0.006745205260813236,
-      "learning_rate": 0.00019999674023471286,
-      "loss": 46.0,
-      "step": 15970
-    },
-    {
-      "epoch": 2.572003703852812,
-      "grad_norm": 0.0007168450392782688,
-      "learning_rate": 0.00019999673982621036,
-      "loss": 46.0,
-      "step": 15971
-    },
-    {
-      "epoch": 2.5721647409315995,
-      "grad_norm": 0.001623555552214384,
-      "learning_rate": 0.0001999967394176823,
-      "loss": 46.0,
-      "step": 15972
-    },
-    {
-      "epoch": 2.572325778010387,
-      "grad_norm": 0.0015095233684405684,
-      "learning_rate": 0.00019999673900912866,
-      "loss": 46.0,
-      "step": 15973
-    },
-    {
-      "epoch": 2.5724868150891744,
-      "grad_norm": 0.003773897420614958,
-      "learning_rate": 0.00019999673860054943,
-      "loss": 46.0,
-      "step": 15974
-    },
-    {
-      "epoch": 2.572647852167962,
-      "grad_norm": 0.0010073219891637564,
-      "learning_rate": 0.00019999673819194456,
-      "loss": 46.0,
-      "step": 15975
-    },
-    {
-      "epoch": 2.5728088892467493,
-      "grad_norm": 0.004301685839891434,
-      "learning_rate": 0.00019999673778331412,
-      "loss": 46.0,
-      "step": 15976
-    },
-    {
-      "epoch": 2.5729699263255363,
-      "grad_norm": 0.001037120702676475,
-      "learning_rate": 0.00019999673737465808,
-      "loss": 46.0,
-      "step": 15977
-    },
-    {
-      "epoch": 2.573130963404324,
-      "grad_norm": 0.0012596957385540009,
-      "learning_rate": 0.00019999673696597645,
-      "loss": 46.0,
-      "step": 15978
-    },
-    {
-      "epoch": 2.5732920004831112,
-      "grad_norm": 0.002911132061854005,
-      "learning_rate": 0.00019999673655726925,
-      "loss": 46.0,
-      "step": 15979
-    },
-    {
-      "epoch": 2.5734530375618987,
-      "grad_norm": 0.003323172451928258,
-      "learning_rate": 0.00019999673614853642,
-      "loss": 46.0,
-      "step": 15980
-    },
-    {
-      "epoch": 2.573614074640686,
-      "grad_norm": 0.004930560011416674,
-      "learning_rate": 0.00019999673573977802,
-      "loss": 46.0,
-      "step": 15981
-    },
-    {
-      "epoch": 2.573775111719473,
-      "grad_norm": 0.0009155586594715714,
-      "learning_rate": 0.000199996735330994,
-      "loss": 46.0,
-      "step": 15982
-    },
-    {
-      "epoch": 2.5739361487982606,
-      "grad_norm": 0.0032252734526991844,
-      "learning_rate": 0.00019999673492218442,
-      "loss": 46.0,
-      "step": 15983
-    },
-    {
-      "epoch": 2.574097185877048,
-      "grad_norm": 0.0023324915673583746,
-      "learning_rate": 0.00019999673451334923,
-      "loss": 46.0,
-      "step": 15984
-    },
-    {
-      "epoch": 2.5742582229558355,
-      "grad_norm": 0.002216553781181574,
-      "learning_rate": 0.00019999673410448843,
-      "loss": 46.0,
-      "step": 15985
-    },
-    {
-      "epoch": 2.574419260034623,
-      "grad_norm": 0.0008827780256979167,
-      "learning_rate": 0.00019999673369560208,
-      "loss": 46.0,
-      "step": 15986
-    },
-    {
-      "epoch": 2.5745802971134104,
-      "grad_norm": 0.002947204513475299,
-      "learning_rate": 0.0001999967332866901,
-      "loss": 46.0,
-      "step": 15987
-    },
-    {
-      "epoch": 2.574741334192198,
-      "grad_norm": 0.0026153482031077147,
-      "learning_rate": 0.00019999673287775252,
-      "loss": 46.0,
-      "step": 15988
-    },
-    {
-      "epoch": 2.5749023712709853,
-      "grad_norm": 0.0019219820387661457,
-      "learning_rate": 0.00019999673246878937,
-      "loss": 46.0,
-      "step": 15989
-    },
-    {
-      "epoch": 2.5750634083497728,
-      "grad_norm": 0.002843607449904084,
-      "learning_rate": 0.0001999967320598006,
-      "loss": 46.0,
-      "step": 15990
-    },
-    {
-      "epoch": 2.5752244454285598,
-      "grad_norm": 0.000634106807410717,
-      "learning_rate": 0.00019999673165078626,
-      "loss": 46.0,
-      "step": 15991
-    },
-    {
-      "epoch": 2.5753854825073472,
-      "grad_norm": 0.0032939857337623835,
-      "learning_rate": 0.00019999673124174633,
-      "loss": 46.0,
-      "step": 15992
-    },
-    {
-      "epoch": 2.5755465195861347,
-      "grad_norm": 0.0035462365485727787,
-      "learning_rate": 0.00019999673083268078,
-      "loss": 46.0,
-      "step": 15993
-    },
-    {
-      "epoch": 2.575707556664922,
-      "grad_norm": 0.001677834428846836,
-      "learning_rate": 0.00019999673042358964,
-      "loss": 46.0,
-      "step": 15994
-    },
-    {
-      "epoch": 2.5758685937437096,
-      "grad_norm": 0.0054684896022081375,
-      "learning_rate": 0.00019999673001447292,
-      "loss": 46.0,
-      "step": 15995
-    },
-    {
-      "epoch": 2.576029630822497,
-      "grad_norm": 0.003958237823098898,
-      "learning_rate": 0.0001999967296053306,
-      "loss": 46.0,
-      "step": 15996
-    },
-    {
-      "epoch": 2.576190667901284,
-      "grad_norm": 0.00340511417016387,
-      "learning_rate": 0.00019999672919616268,
-      "loss": 46.0,
-      "step": 15997
-    },
-    {
-      "epoch": 2.5763517049800715,
-      "grad_norm": 0.004183250013738871,
-      "learning_rate": 0.00019999672878696917,
-      "loss": 46.0,
-      "step": 15998
-    },
-    {
-      "epoch": 2.576512742058859,
-      "grad_norm": 0.0023353854194283485,
-      "learning_rate": 0.0001999967283777501,
-      "loss": 46.0,
-      "step": 15999
-    },
-    {
-      "epoch": 2.5766737791376464,
-      "grad_norm": 0.0015198108740150928,
-      "learning_rate": 0.00019999672796850538,
-      "loss": 46.0,
-      "step": 16000
-    },
-    {
-      "epoch": 2.576834816216434,
-      "grad_norm": 0.0014344934606924653,
-      "learning_rate": 0.0001999967275592351,
-      "loss": 46.0,
-      "step": 16001
-    },
-    {
-      "epoch": 2.5769958532952213,
-      "grad_norm": 0.0011377371847629547,
-      "learning_rate": 0.00019999672714993922,
-      "loss": 46.0,
-      "step": 16002
-    },
-    {
-      "epoch": 2.5771568903740087,
-      "grad_norm": 0.0012556801084429026,
-      "learning_rate": 0.00019999672674061775,
-      "loss": 46.0,
-      "step": 16003
-    },
-    {
-      "epoch": 2.577317927452796,
-      "grad_norm": 0.0008868277072906494,
-      "learning_rate": 0.00019999672633127066,
-      "loss": 46.0,
-      "step": 16004
-    },
-    {
-      "epoch": 2.5774789645315836,
-      "grad_norm": 0.0026483850087970495,
-      "learning_rate": 0.000199996725921898,
-      "loss": 46.0,
-      "step": 16005
-    },
-    {
-      "epoch": 2.5776400016103707,
-      "grad_norm": 0.0015649127308279276,
-      "learning_rate": 0.00019999672551249974,
-      "loss": 46.0,
-      "step": 16006
-    },
-    {
-      "epoch": 2.577801038689158,
-      "grad_norm": 0.000816620362456888,
-      "learning_rate": 0.00019999672510307586,
-      "loss": 46.0,
-      "step": 16007
-    },
-    {
-      "epoch": 2.5779620757679456,
-      "grad_norm": 0.0012617866741493344,
-      "learning_rate": 0.00019999672469362643,
-      "loss": 46.0,
-      "step": 16008
-    },
-    {
-      "epoch": 2.578123112846733,
-      "grad_norm": 0.00266861985437572,
-      "learning_rate": 0.0001999967242841514,
-      "loss": 46.0,
-      "step": 16009
-    },
-    {
-      "epoch": 2.5782841499255205,
-      "grad_norm": 0.005883903242647648,
-      "learning_rate": 0.00019999672387465076,
-      "loss": 46.0,
-      "step": 16010
-    },
-    {
-      "epoch": 2.5784451870043075,
-      "grad_norm": 0.001797378994524479,
-      "learning_rate": 0.00019999672346512456,
-      "loss": 46.0,
-      "step": 16011
-    },
-    {
-      "epoch": 2.578606224083095,
-      "grad_norm": 0.0016678535612300038,
-      "learning_rate": 0.00019999672305557272,
-      "loss": 46.0,
-      "step": 16012
-    },
-    {
-      "epoch": 2.5787672611618824,
-      "grad_norm": 0.002299239858984947,
-      "learning_rate": 0.0001999967226459953,
-      "loss": 46.0,
-      "step": 16013
-    },
-    {
-      "epoch": 2.57892829824067,
-      "grad_norm": 0.0014426435809582472,
-      "learning_rate": 0.00019999672223639227,
-      "loss": 46.0,
-      "step": 16014
-    },
-    {
-      "epoch": 2.5790893353194573,
-      "grad_norm": 0.0012001210125163198,
-      "learning_rate": 0.00019999672182676367,
-      "loss": 46.0,
-      "step": 16015
-    },
-    {
-      "epoch": 2.5792503723982447,
-      "grad_norm": 0.0030750101432204247,
-      "learning_rate": 0.00019999672141710948,
-      "loss": 46.0,
-      "step": 16016
-    },
-    {
-      "epoch": 2.579411409477032,
-      "grad_norm": 0.005097745917737484,
-      "learning_rate": 0.00019999672100742968,
-      "loss": 46.0,
-      "step": 16017
-    },
-    {
-      "epoch": 2.5795724465558196,
-      "grad_norm": 0.0013655796647071838,
-      "learning_rate": 0.0001999967205977243,
-      "loss": 46.0,
-      "step": 16018
-    },
-    {
-      "epoch": 2.579733483634607,
-      "grad_norm": 0.002719012089073658,
-      "learning_rate": 0.0001999967201879933,
-      "loss": 46.0,
-      "step": 16019
-    },
-    {
-      "epoch": 2.5798945207133945,
-      "grad_norm": 0.0022950624115765095,
-      "learning_rate": 0.00019999671977823674,
-      "loss": 46.0,
-      "step": 16020
-    },
-    {
-      "epoch": 2.5800555577921815,
-      "grad_norm": 0.0007586430292576551,
-      "learning_rate": 0.00019999671936845456,
-      "loss": 46.0,
-      "step": 16021
-    },
-    {
-      "epoch": 2.580216594870969,
-      "grad_norm": 0.0025612623430788517,
-      "learning_rate": 0.0001999967189586468,
-      "loss": 46.0,
-      "step": 16022
-    },
-    {
-      "epoch": 2.5803776319497564,
-      "grad_norm": 0.0009127302910201252,
-      "learning_rate": 0.00019999671854881347,
-      "loss": 46.0,
-      "step": 16023
-    },
-    {
-      "epoch": 2.580538669028544,
-      "grad_norm": 0.0006393424118869007,
-      "learning_rate": 0.0001999967181389545,
-      "loss": 46.0,
-      "step": 16024
-    },
-    {
-      "epoch": 2.5806997061073313,
-      "grad_norm": 0.011125031858682632,
-      "learning_rate": 0.00019999671772906994,
-      "loss": 46.0,
-      "step": 16025
-    },
-    {
-      "epoch": 2.5808607431861184,
-      "grad_norm": 0.000548688170965761,
-      "learning_rate": 0.0001999967173191598,
-      "loss": 46.0,
-      "step": 16026
-    },
-    {
-      "epoch": 2.581021780264906,
-      "grad_norm": 0.001740701962262392,
-      "learning_rate": 0.00019999671690922407,
-      "loss": 46.0,
-      "step": 16027
-    },
-    {
-      "epoch": 2.5811828173436933,
-      "grad_norm": 0.0031184684485197067,
-      "learning_rate": 0.00019999671649926275,
-      "loss": 46.0,
-      "step": 16028
-    },
-    {
-      "epoch": 2.5813438544224807,
-      "grad_norm": 0.002004853216931224,
-      "learning_rate": 0.00019999671608927582,
-      "loss": 46.0,
-      "step": 16029
-    },
-    {
-      "epoch": 2.581504891501268,
-      "grad_norm": 0.002701597986742854,
-      "learning_rate": 0.0001999967156792633,
-      "loss": 46.0,
-      "step": 16030
-    },
-    {
-      "epoch": 2.5816659285800556,
-      "grad_norm": 0.0029638635460287333,
-      "learning_rate": 0.0001999967152692252,
-      "loss": 46.0,
-      "step": 16031
-    },
-    {
-      "epoch": 2.581826965658843,
-      "grad_norm": 0.00609456654638052,
-      "learning_rate": 0.0001999967148591615,
-      "loss": 46.0,
-      "step": 16032
-    },
-    {
-      "epoch": 2.5819880027376305,
-      "grad_norm": 0.0033189135137945414,
-      "learning_rate": 0.0001999967144490722,
-      "loss": 46.0,
-      "step": 16033
-    },
-    {
-      "epoch": 2.582149039816418,
-      "grad_norm": 0.004438711795955896,
-      "learning_rate": 0.0001999967140389573,
-      "loss": 46.0,
-      "step": 16034
-    },
-    {
-      "epoch": 2.582310076895205,
-      "grad_norm": 0.002540210960432887,
-      "learning_rate": 0.00019999671362881682,
-      "loss": 46.0,
-      "step": 16035
-    },
-    {
-      "epoch": 2.5824711139739924,
-      "grad_norm": 0.0055688670836389065,
-      "learning_rate": 0.00019999671321865075,
-      "loss": 46.0,
-      "step": 16036
-    },
-    {
-      "epoch": 2.58263215105278,
-      "grad_norm": 0.0027148216031491756,
-      "learning_rate": 0.00019999671280845907,
-      "loss": 46.0,
-      "step": 16037
-    },
-    {
-      "epoch": 2.5827931881315673,
-      "grad_norm": 0.002419590251520276,
-      "learning_rate": 0.0001999967123982418,
-      "loss": 46.0,
-      "step": 16038
-    },
-    {
-      "epoch": 2.582954225210355,
-      "grad_norm": 0.0013533795718103647,
-      "learning_rate": 0.00019999671198799894,
-      "loss": 46.0,
-      "step": 16039
-    },
-    {
-      "epoch": 2.5831152622891422,
-      "grad_norm": 0.0010592100443318486,
-      "learning_rate": 0.00019999671157773047,
-      "loss": 46.0,
-      "step": 16040
-    },
-    {
-      "epoch": 2.5832762993679292,
-      "grad_norm": 0.0006331202457658947,
-      "learning_rate": 0.00019999671116743644,
-      "loss": 46.0,
-      "step": 16041
-    },
-    {
-      "epoch": 2.5834373364467167,
-      "grad_norm": 0.000930843991227448,
-      "learning_rate": 0.0001999967107571168,
-      "loss": 46.0,
-      "step": 16042
-    },
-    {
-      "epoch": 2.583598373525504,
-      "grad_norm": 0.003928042482584715,
-      "learning_rate": 0.00019999671034677153,
-      "loss": 46.0,
-      "step": 16043
-    },
-    {
-      "epoch": 2.5837594106042916,
-      "grad_norm": 0.003953359555453062,
-      "learning_rate": 0.0001999967099364007,
-      "loss": 46.0,
-      "step": 16044
-    },
-    {
-      "epoch": 2.583920447683079,
-      "grad_norm": 0.0010099763749167323,
-      "learning_rate": 0.00019999670952600428,
-      "loss": 46.0,
-      "step": 16045
-    },
-    {
-      "epoch": 2.5840814847618665,
-      "grad_norm": 0.011977887712419033,
-      "learning_rate": 0.00019999670911558225,
-      "loss": 46.0,
-      "step": 16046
-    },
-    {
-      "epoch": 2.584242521840654,
-      "grad_norm": 0.0047042942605912685,
-      "learning_rate": 0.00019999670870513464,
-      "loss": 46.0,
-      "step": 16047
-    },
-    {
-      "epoch": 2.5844035589194414,
-      "grad_norm": 0.004303669091314077,
-      "learning_rate": 0.00019999670829466145,
-      "loss": 46.0,
-      "step": 16048
-    },
-    {
-      "epoch": 2.584564595998229,
-      "grad_norm": 0.0009295951458625495,
-      "learning_rate": 0.00019999670788416264,
-      "loss": 46.0,
-      "step": 16049
-    },
-    {
-      "epoch": 2.584725633077016,
-      "grad_norm": 0.0008656475692987442,
-      "learning_rate": 0.00019999670747363824,
-      "loss": 46.0,
-      "step": 16050
-    },
-    {
-      "epoch": 2.5848866701558033,
-      "grad_norm": 0.0015673545422032475,
-      "learning_rate": 0.00019999670706308823,
-      "loss": 46.0,
-      "step": 16051
-    },
-    {
-      "epoch": 2.5850477072345908,
-      "grad_norm": 0.00130812160205096,
-      "learning_rate": 0.00019999670665251266,
-      "loss": 46.0,
-      "step": 16052
-    },
-    {
-      "epoch": 2.585208744313378,
-      "grad_norm": 0.0018158942693844438,
-      "learning_rate": 0.00019999670624191147,
-      "loss": 46.0,
-      "step": 16053
-    },
-    {
-      "epoch": 2.5853697813921657,
-      "grad_norm": 0.0024436796084046364,
-      "learning_rate": 0.0001999967058312847,
-      "loss": 46.0,
-      "step": 16054
-    },
-    {
-      "epoch": 2.5855308184709527,
-      "grad_norm": 0.0010908261174336076,
-      "learning_rate": 0.00019999670542063234,
-      "loss": 46.0,
-      "step": 16055
-    },
-    {
-      "epoch": 2.58569185554974,
-      "grad_norm": 0.001932608662173152,
-      "learning_rate": 0.0001999967050099544,
-      "loss": 46.0,
-      "step": 16056
-    },
-    {
-      "epoch": 2.5858528926285276,
-      "grad_norm": 0.0035636143293231726,
-      "learning_rate": 0.00019999670459925083,
-      "loss": 46.0,
-      "step": 16057
-    },
-    {
-      "epoch": 2.586013929707315,
-      "grad_norm": 0.008837430737912655,
-      "learning_rate": 0.00019999670418852168,
-      "loss": 46.0,
-      "step": 16058
-    },
-    {
-      "epoch": 2.5861749667861025,
-      "grad_norm": 0.012465517036616802,
-      "learning_rate": 0.00019999670377776694,
-      "loss": 46.0,
-      "step": 16059
-    },
-    {
-      "epoch": 2.58633600386489,
-      "grad_norm": 0.009264561347663403,
-      "learning_rate": 0.0001999967033669866,
-      "loss": 46.0,
-      "step": 16060
-    },
-    {
-      "epoch": 2.5864970409436774,
-      "grad_norm": 0.0013143039541319013,
-      "learning_rate": 0.00019999670295618068,
-      "loss": 46.0,
-      "step": 16061
-    },
-    {
-      "epoch": 2.586658078022465,
-      "grad_norm": 0.0039000045508146286,
-      "learning_rate": 0.00019999670254534915,
-      "loss": 46.0,
-      "step": 16062
-    },
-    {
-      "epoch": 2.5868191151012523,
-      "grad_norm": 0.00105371349491179,
-      "learning_rate": 0.00019999670213449201,
-      "loss": 46.0,
-      "step": 16063
-    },
-    {
-      "epoch": 2.5869801521800393,
-      "grad_norm": 0.005493657663464546,
-      "learning_rate": 0.0001999967017236093,
-      "loss": 46.0,
-      "step": 16064
-    },
-    {
-      "epoch": 2.5871411892588267,
-      "grad_norm": 0.0008403283427469432,
-      "learning_rate": 0.000199996701312701,
-      "loss": 46.0,
-      "step": 16065
-    },
-    {
-      "epoch": 2.587302226337614,
-      "grad_norm": 0.002454224741086364,
-      "learning_rate": 0.0001999967009017671,
-      "loss": 46.0,
-      "step": 16066
-    },
-    {
-      "epoch": 2.5874632634164016,
-      "grad_norm": 0.001540362136438489,
-      "learning_rate": 0.0001999967004908076,
-      "loss": 46.0,
-      "step": 16067
-    },
-    {
-      "epoch": 2.587624300495189,
-      "grad_norm": 0.005297347903251648,
-      "learning_rate": 0.0001999967000798225,
-      "loss": 46.0,
-      "step": 16068
-    },
-    {
-      "epoch": 2.5877853375739766,
-      "grad_norm": 0.0014679948799312115,
-      "learning_rate": 0.00019999669966881185,
-      "loss": 46.0,
-      "step": 16069
-    },
-    {
-      "epoch": 2.5879463746527636,
-      "grad_norm": 0.0024977836292237043,
-      "learning_rate": 0.00019999669925777554,
-      "loss": 46.0,
-      "step": 16070
-    },
-    {
-      "epoch": 2.588107411731551,
-      "grad_norm": 0.0032964071724563837,
-      "learning_rate": 0.00019999669884671368,
-      "loss": 46.0,
-      "step": 16071
-    },
-    {
-      "epoch": 2.5882684488103385,
-      "grad_norm": 0.004492423962801695,
-      "learning_rate": 0.0001999966984356262,
-      "loss": 46.0,
-      "step": 16072
-    },
-    {
-      "epoch": 2.588429485889126,
-      "grad_norm": 0.0006257380591705441,
-      "learning_rate": 0.00019999669802451316,
-      "loss": 46.0,
-      "step": 16073
-    },
-    {
-      "epoch": 2.5885905229679134,
-      "grad_norm": 0.0009773288620635867,
-      "learning_rate": 0.0001999966976133745,
-      "loss": 46.0,
-      "step": 16074
-    },
-    {
-      "epoch": 2.588751560046701,
-      "grad_norm": 0.002816564403474331,
-      "learning_rate": 0.00019999669720221027,
-      "loss": 46.0,
-      "step": 16075
-    },
-    {
-      "epoch": 2.5889125971254883,
-      "grad_norm": 0.0033466166350990534,
-      "learning_rate": 0.00019999669679102042,
-      "loss": 46.0,
-      "step": 16076
-    },
-    {
-      "epoch": 2.5890736342042757,
-      "grad_norm": 0.0023996103554964066,
-      "learning_rate": 0.00019999669637980498,
-      "loss": 46.0,
-      "step": 16077
-    },
-    {
-      "epoch": 2.589234671283063,
-      "grad_norm": 0.0008122390718199313,
-      "learning_rate": 0.00019999669596856398,
-      "loss": 46.0,
-      "step": 16078
-    },
-    {
-      "epoch": 2.58939570836185,
-      "grad_norm": 0.0015022241277620196,
-      "learning_rate": 0.00019999669555729733,
-      "loss": 46.0,
-      "step": 16079
-    },
-    {
-      "epoch": 2.5895567454406376,
-      "grad_norm": 0.0010299028363078833,
-      "learning_rate": 0.0001999966951460051,
-      "loss": 46.0,
-      "step": 16080
-    },
-    {
-      "epoch": 2.589717782519425,
-      "grad_norm": 0.006779077462852001,
-      "learning_rate": 0.0001999966947346873,
-      "loss": 46.0,
-      "step": 16081
-    },
-    {
-      "epoch": 2.5898788195982125,
-      "grad_norm": 0.0018228732515126467,
-      "learning_rate": 0.00019999669432334388,
-      "loss": 46.0,
-      "step": 16082
-    },
-    {
-      "epoch": 2.590039856677,
-      "grad_norm": 0.005500158295035362,
-      "learning_rate": 0.0001999966939119749,
-      "loss": 46.0,
-      "step": 16083
-    },
-    {
-      "epoch": 2.590200893755787,
-      "grad_norm": 0.004686194006353617,
-      "learning_rate": 0.00019999669350058031,
-      "loss": 46.0,
-      "step": 16084
-    },
-    {
-      "epoch": 2.5903619308345744,
-      "grad_norm": 0.0014073998900130391,
-      "learning_rate": 0.00019999669308916012,
-      "loss": 46.0,
-      "step": 16085
-    },
-    {
-      "epoch": 2.590522967913362,
-      "grad_norm": 0.0015975123969838023,
-      "learning_rate": 0.00019999669267771431,
-      "loss": 46.0,
-      "step": 16086
-    },
-    {
-      "epoch": 2.5906840049921493,
-      "grad_norm": 0.0017667760839685798,
-      "learning_rate": 0.00019999669226624295,
-      "loss": 46.0,
-      "step": 16087
-    },
-    {
-      "epoch": 2.590845042070937,
-      "grad_norm": 0.0007539974758401513,
-      "learning_rate": 0.000199996691854746,
-      "loss": 46.0,
-      "step": 16088
-    },
-    {
-      "epoch": 2.5910060791497243,
-      "grad_norm": 0.0030309869907796383,
-      "learning_rate": 0.00019999669144322342,
-      "loss": 46.0,
-      "step": 16089
-    },
-    {
-      "epoch": 2.5911671162285117,
-      "grad_norm": 0.0014217736897990108,
-      "learning_rate": 0.00019999669103167527,
-      "loss": 46.0,
-      "step": 16090
-    },
-    {
-      "epoch": 2.591328153307299,
-      "grad_norm": 0.0010116341290995479,
-      "learning_rate": 0.0001999966906201015,
-      "loss": 46.0,
-      "step": 16091
-    },
-    {
-      "epoch": 2.5914891903860866,
-      "grad_norm": 0.0036425767466425896,
-      "learning_rate": 0.00019999669020850217,
-      "loss": 46.0,
-      "step": 16092
-    },
-    {
-      "epoch": 2.591650227464874,
-      "grad_norm": 0.0017039704835042357,
-      "learning_rate": 0.00019999668979687723,
-      "loss": 46.0,
-      "step": 16093
-    },
-    {
-      "epoch": 2.591811264543661,
-      "grad_norm": 0.0012954877456650138,
-      "learning_rate": 0.00019999668938522672,
-      "loss": 46.0,
-      "step": 16094
-    },
-    {
-      "epoch": 2.5919723016224485,
-      "grad_norm": 0.001290460117161274,
-      "learning_rate": 0.00019999668897355058,
-      "loss": 46.0,
-      "step": 16095
-    },
-    {
-      "epoch": 2.592133338701236,
-      "grad_norm": 0.0036992921959608793,
-      "learning_rate": 0.00019999668856184884,
-      "loss": 46.0,
-      "step": 16096
-    },
-    {
-      "epoch": 2.5922943757800234,
-      "grad_norm": 0.0026420990470796824,
-      "learning_rate": 0.00019999668815012155,
-      "loss": 46.0,
-      "step": 16097
-    },
-    {
-      "epoch": 2.592455412858811,
-      "grad_norm": 0.0012626516399905086,
-      "learning_rate": 0.00019999668773836862,
-      "loss": 46.0,
-      "step": 16098
-    },
-    {
-      "epoch": 2.592616449937598,
-      "grad_norm": 0.0025810569059103727,
-      "learning_rate": 0.00019999668732659012,
-      "loss": 46.0,
-      "step": 16099
-    },
-    {
-      "epoch": 2.5927774870163853,
-      "grad_norm": 0.0027801140677183867,
-      "learning_rate": 0.000199996686914786,
-      "loss": 46.0,
-      "step": 16100
-    },
-    {
-      "epoch": 2.592938524095173,
-      "grad_norm": 0.0021034367382526398,
-      "learning_rate": 0.00019999668650295634,
-      "loss": 46.0,
-      "step": 16101
-    },
-    {
-      "epoch": 2.5930995611739602,
-      "grad_norm": 0.007078695576637983,
-      "learning_rate": 0.00019999668609110103,
-      "loss": 46.0,
-      "step": 16102
-    },
-    {
-      "epoch": 2.5932605982527477,
-      "grad_norm": 0.0016240933910012245,
-      "learning_rate": 0.00019999668567922016,
-      "loss": 46.0,
-      "step": 16103
-    },
-    {
-      "epoch": 2.593421635331535,
-      "grad_norm": 0.0008488434832543135,
-      "learning_rate": 0.00019999668526731368,
-      "loss": 46.0,
-      "step": 16104
-    },
-    {
-      "epoch": 2.5935826724103226,
-      "grad_norm": 0.0046164835803210735,
-      "learning_rate": 0.00019999668485538163,
-      "loss": 46.0,
-      "step": 16105
-    },
-    {
-      "epoch": 2.59374370948911,
-      "grad_norm": 0.0018604089273139834,
-      "learning_rate": 0.00019999668444342398,
-      "loss": 46.0,
-      "step": 16106
-    },
-    {
-      "epoch": 2.5939047465678975,
-      "grad_norm": 0.0017046486027538776,
-      "learning_rate": 0.0001999966840314407,
-      "loss": 46.0,
-      "step": 16107
-    },
-    {
-      "epoch": 2.5940657836466845,
-      "grad_norm": 0.007591297850012779,
-      "learning_rate": 0.00019999668361943184,
-      "loss": 46.0,
-      "step": 16108
-    },
-    {
-      "epoch": 2.594226820725472,
-      "grad_norm": 0.002911948598921299,
-      "learning_rate": 0.0001999966832073974,
-      "loss": 46.0,
-      "step": 16109
-    },
-    {
-      "epoch": 2.5943878578042594,
-      "grad_norm": 0.0032055643387138844,
-      "learning_rate": 0.00019999668279533736,
-      "loss": 46.0,
-      "step": 16110
-    },
-    {
-      "epoch": 2.594548894883047,
-      "grad_norm": 0.00234204507432878,
-      "learning_rate": 0.00019999668238325174,
-      "loss": 46.0,
-      "step": 16111
-    },
-    {
-      "epoch": 2.5947099319618343,
-      "grad_norm": 0.004169377498328686,
-      "learning_rate": 0.00019999668197114053,
-      "loss": 46.0,
-      "step": 16112
-    },
-    {
-      "epoch": 2.5948709690406218,
-      "grad_norm": 0.00201976066455245,
-      "learning_rate": 0.00019999668155900368,
-      "loss": 46.0,
-      "step": 16113
-    },
-    {
-      "epoch": 2.5950320061194088,
-      "grad_norm": 0.00225282390601933,
-      "learning_rate": 0.00019999668114684127,
-      "loss": 46.0,
-      "step": 16114
-    },
-    {
-      "epoch": 2.595193043198196,
-      "grad_norm": 0.001989091048017144,
-      "learning_rate": 0.00019999668073465324,
-      "loss": 46.0,
-      "step": 16115
-    },
-    {
-      "epoch": 2.5953540802769837,
-      "grad_norm": 0.00479503720998764,
-      "learning_rate": 0.00019999668032243966,
-      "loss": 46.0,
-      "step": 16116
-    },
-    {
-      "epoch": 2.595515117355771,
-      "grad_norm": 0.003390047699213028,
-      "learning_rate": 0.0001999966799102005,
-      "loss": 46.0,
-      "step": 16117
-    },
-    {
-      "epoch": 2.5956761544345586,
-      "grad_norm": 0.0019227584125474095,
-      "learning_rate": 0.00019999667949793568,
-      "loss": 46.0,
-      "step": 16118
-    },
-    {
-      "epoch": 2.595837191513346,
-      "grad_norm": 0.0043724942952394485,
-      "learning_rate": 0.0001999966790856453,
-      "loss": 46.0,
-      "step": 16119
-    },
-    {
-      "epoch": 2.5959982285921335,
-      "grad_norm": 0.0021974837873131037,
-      "learning_rate": 0.00019999667867332931,
-      "loss": 46.0,
-      "step": 16120
-    },
-    {
-      "epoch": 2.596159265670921,
-      "grad_norm": 0.011448969133198261,
-      "learning_rate": 0.00019999667826098777,
-      "loss": 46.0,
-      "step": 16121
-    },
-    {
-      "epoch": 2.5963203027497084,
-      "grad_norm": 0.005732130259275436,
-      "learning_rate": 0.00019999667784862058,
-      "loss": 46.0,
-      "step": 16122
-    },
-    {
-      "epoch": 2.5964813398284954,
-      "grad_norm": 0.0028530533891171217,
-      "learning_rate": 0.0001999966774362278,
-      "loss": 46.0,
-      "step": 16123
-    },
-    {
-      "epoch": 2.596642376907283,
-      "grad_norm": 0.007159261964261532,
-      "learning_rate": 0.00019999667702380947,
-      "loss": 46.0,
-      "step": 16124
-    },
-    {
-      "epoch": 2.5968034139860703,
-      "grad_norm": 0.009948200546205044,
-      "learning_rate": 0.00019999667661136551,
-      "loss": 46.0,
-      "step": 16125
-    },
-    {
-      "epoch": 2.5969644510648577,
-      "grad_norm": 0.00519247492775321,
-      "learning_rate": 0.00019999667619889598,
-      "loss": 46.0,
-      "step": 16126
-    },
-    {
-      "epoch": 2.597125488143645,
-      "grad_norm": 0.015372388996183872,
-      "learning_rate": 0.00019999667578640083,
-      "loss": 46.0,
-      "step": 16127
-    },
-    {
-      "epoch": 2.597286525222432,
-      "grad_norm": 0.003131683450192213,
-      "learning_rate": 0.00019999667537388009,
-      "loss": 46.0,
-      "step": 16128
-    },
-    {
-      "epoch": 2.5974475623012196,
-      "grad_norm": 0.0027298214845359325,
-      "learning_rate": 0.0001999966749613338,
-      "loss": 46.0,
-      "step": 16129
-    },
-    {
-      "epoch": 2.597608599380007,
-      "grad_norm": 0.0033878039103001356,
-      "learning_rate": 0.00019999667454876185,
-      "loss": 46.0,
-      "step": 16130
-    },
-    {
-      "epoch": 2.5977696364587946,
-      "grad_norm": 0.0021357000805437565,
-      "learning_rate": 0.00019999667413616435,
-      "loss": 46.0,
-      "step": 16131
-    },
-    {
-      "epoch": 2.597930673537582,
-      "grad_norm": 0.003726204624399543,
-      "learning_rate": 0.00019999667372354126,
-      "loss": 46.0,
-      "step": 16132
-    },
-    {
-      "epoch": 2.5980917106163695,
-      "grad_norm": 0.003822133643552661,
-      "learning_rate": 0.00019999667331089253,
-      "loss": 46.0,
-      "step": 16133
-    },
-    {
-      "epoch": 2.598252747695157,
-      "grad_norm": 0.0016124286921694875,
-      "learning_rate": 0.00019999667289821824,
-      "loss": 46.0,
-      "step": 16134
-    },
-    {
-      "epoch": 2.5984137847739444,
-      "grad_norm": 0.0036133346147835255,
-      "learning_rate": 0.00019999667248551836,
-      "loss": 46.0,
-      "step": 16135
-    },
-    {
-      "epoch": 2.598574821852732,
-      "grad_norm": 0.0014677965082228184,
-      "learning_rate": 0.00019999667207279287,
-      "loss": 46.0,
-      "step": 16136
-    },
-    {
-      "epoch": 2.598735858931519,
-      "grad_norm": 0.0017298039747402072,
-      "learning_rate": 0.0001999966716600418,
-      "loss": 46.0,
-      "step": 16137
-    },
-    {
-      "epoch": 2.5988968960103063,
-      "grad_norm": 0.0009745733113959432,
-      "learning_rate": 0.00019999667124726513,
-      "loss": 46.0,
-      "step": 16138
-    },
-    {
-      "epoch": 2.5990579330890937,
-      "grad_norm": 0.0032255914993584156,
-      "learning_rate": 0.00019999667083446285,
-      "loss": 46.0,
-      "step": 16139
-    },
-    {
-      "epoch": 2.599218970167881,
-      "grad_norm": 0.007755223195999861,
-      "learning_rate": 0.00019999667042163498,
-      "loss": 46.0,
-      "step": 16140
-    },
-    {
-      "epoch": 2.5993800072466686,
-      "grad_norm": 0.003692959202453494,
-      "learning_rate": 0.00019999667000878156,
-      "loss": 46.0,
-      "step": 16141
-    },
-    {
-      "epoch": 2.599541044325456,
-      "grad_norm": 0.0010905322851613164,
-      "learning_rate": 0.00019999666959590252,
-      "loss": 46.0,
-      "step": 16142
-    },
-    {
-      "epoch": 2.599702081404243,
-      "grad_norm": 0.00252927141264081,
-      "learning_rate": 0.00019999666918299786,
-      "loss": 46.0,
-      "step": 16143
-    },
-    {
-      "epoch": 2.5998631184830305,
-      "grad_norm": 0.0017952235648408532,
-      "learning_rate": 0.00019999666877006762,
-      "loss": 46.0,
-      "step": 16144
-    },
-    {
-      "epoch": 2.600024155561818,
-      "grad_norm": 0.002521948888897896,
-      "learning_rate": 0.00019999666835711182,
-      "loss": 46.0,
-      "step": 16145
-    },
-    {
-      "epoch": 2.6001851926406054,
-      "grad_norm": 0.0033577398862689734,
-      "learning_rate": 0.0001999966679441304,
-      "loss": 46.0,
-      "step": 16146
-    },
-    {
-      "epoch": 2.600346229719393,
-      "grad_norm": 0.0005824739346280694,
-      "learning_rate": 0.00019999666753112337,
-      "loss": 46.0,
-      "step": 16147
-    },
-    {
-      "epoch": 2.6005072667981803,
-      "grad_norm": 0.0018328281585127115,
-      "learning_rate": 0.00019999666711809075,
-      "loss": 46.0,
-      "step": 16148
-    },
-    {
-      "epoch": 2.600668303876968,
-      "grad_norm": 0.0020289828535169363,
-      "learning_rate": 0.00019999666670503255,
-      "loss": 46.0,
-      "step": 16149
-    },
-    {
-      "epoch": 2.6008293409557552,
-      "grad_norm": 0.010007748380303383,
-      "learning_rate": 0.00019999666629194875,
-      "loss": 46.0,
-      "step": 16150
-    },
-    {
-      "epoch": 2.6009903780345427,
-      "grad_norm": 0.0024312378372997046,
-      "learning_rate": 0.00019999666587883935,
-      "loss": 46.0,
-      "step": 16151
-    },
-    {
-      "epoch": 2.6011514151133297,
-      "grad_norm": 0.00319654680788517,
-      "learning_rate": 0.00019999666546570438,
-      "loss": 46.0,
-      "step": 16152
-    },
-    {
-      "epoch": 2.601312452192117,
-      "grad_norm": 0.005473264493048191,
-      "learning_rate": 0.00019999666505254377,
-      "loss": 46.0,
-      "step": 16153
-    },
-    {
-      "epoch": 2.6014734892709046,
-      "grad_norm": 0.0007003513746894896,
-      "learning_rate": 0.00019999666463935763,
-      "loss": 46.0,
-      "step": 16154
-    },
-    {
-      "epoch": 2.601634526349692,
-      "grad_norm": 0.003339738119393587,
-      "learning_rate": 0.00019999666422614585,
-      "loss": 46.0,
-      "step": 16155
-    },
-    {
-      "epoch": 2.6017955634284795,
-      "grad_norm": 0.0023215902037918568,
-      "learning_rate": 0.00019999666381290848,
-      "loss": 46.0,
-      "step": 16156
-    },
-    {
-      "epoch": 2.601956600507267,
-      "grad_norm": 0.001549035543575883,
-      "learning_rate": 0.00019999666339964552,
-      "loss": 46.0,
-      "step": 16157
-    },
-    {
-      "epoch": 2.602117637586054,
-      "grad_norm": 0.0012019840069115162,
-      "learning_rate": 0.00019999666298635698,
-      "loss": 46.0,
-      "step": 16158
-    },
-    {
-      "epoch": 2.6022786746648414,
-      "grad_norm": 0.00277071725577116,
-      "learning_rate": 0.00019999666257304282,
-      "loss": 46.0,
-      "step": 16159
-    },
-    {
-      "epoch": 2.602439711743629,
-      "grad_norm": 0.0015820463886484504,
-      "learning_rate": 0.0001999966621597031,
-      "loss": 46.0,
-      "step": 16160
-    },
-    {
-      "epoch": 2.6026007488224163,
-      "grad_norm": 0.006370109040290117,
-      "learning_rate": 0.00019999666174633777,
-      "loss": 46.0,
-      "step": 16161
-    },
-    {
-      "epoch": 2.6027617859012038,
-      "grad_norm": 0.0070269908756017685,
-      "learning_rate": 0.00019999666133294682,
-      "loss": 46.0,
-      "step": 16162
-    },
-    {
-      "epoch": 2.6029228229799912,
-      "grad_norm": 0.0020513904746621847,
-      "learning_rate": 0.00019999666091953031,
-      "loss": 46.0,
-      "step": 16163
-    },
-    {
-      "epoch": 2.6030838600587787,
-      "grad_norm": 0.006313891615718603,
-      "learning_rate": 0.00019999666050608822,
-      "loss": 46.0,
-      "step": 16164
-    },
-    {
-      "epoch": 2.603244897137566,
-      "grad_norm": 0.0008577312109991908,
-      "learning_rate": 0.00019999666009262048,
-      "loss": 46.0,
-      "step": 16165
-    },
-    {
-      "epoch": 2.6034059342163536,
-      "grad_norm": 0.007441112305969,
-      "learning_rate": 0.00019999665967912716,
-      "loss": 46.0,
-      "step": 16166
-    },
-    {
-      "epoch": 2.6035669712951406,
-      "grad_norm": 0.0008037133375182748,
-      "learning_rate": 0.00019999665926560828,
-      "loss": 46.0,
-      "step": 16167
-    },
-    {
-      "epoch": 2.603728008373928,
-      "grad_norm": 0.0006772925844416022,
-      "learning_rate": 0.00019999665885206378,
-      "loss": 46.0,
-      "step": 16168
-    },
-    {
-      "epoch": 2.6038890454527155,
-      "grad_norm": 0.007704761810600758,
-      "learning_rate": 0.00019999665843849372,
-      "loss": 46.0,
-      "step": 16169
-    },
-    {
-      "epoch": 2.604050082531503,
-      "grad_norm": 0.001972059952095151,
-      "learning_rate": 0.00019999665802489803,
-      "loss": 46.0,
-      "step": 16170
-    },
-    {
-      "epoch": 2.6042111196102904,
-      "grad_norm": 0.0006038892897777259,
-      "learning_rate": 0.00019999665761127674,
-      "loss": 46.0,
-      "step": 16171
-    },
-    {
-      "epoch": 2.6043721566890774,
-      "grad_norm": 0.0017083912389352918,
-      "learning_rate": 0.0001999966571976299,
-      "loss": 46.0,
-      "step": 16172
-    },
-    {
-      "epoch": 2.604533193767865,
-      "grad_norm": 0.0024328369181603193,
-      "learning_rate": 0.00019999665678395743,
-      "loss": 46.0,
-      "step": 16173
-    },
-    {
-      "epoch": 2.6046942308466523,
-      "grad_norm": 0.002566979033872485,
-      "learning_rate": 0.00019999665637025936,
-      "loss": 46.0,
-      "step": 16174
-    },
-    {
-      "epoch": 2.6048552679254398,
-      "grad_norm": 0.0035113622434437275,
-      "learning_rate": 0.00019999665595653573,
-      "loss": 46.0,
-      "step": 16175
-    },
-    {
-      "epoch": 2.605016305004227,
-      "grad_norm": 0.0013542110100388527,
-      "learning_rate": 0.00019999665554278648,
-      "loss": 46.0,
-      "step": 16176
-    },
-    {
-      "epoch": 2.6051773420830147,
-      "grad_norm": 0.002967189298942685,
-      "learning_rate": 0.00019999665512901164,
-      "loss": 46.0,
-      "step": 16177
-    },
-    {
-      "epoch": 2.605338379161802,
-      "grad_norm": 0.0010982576059177518,
-      "learning_rate": 0.00019999665471521122,
-      "loss": 46.0,
-      "step": 16178
-    },
-    {
-      "epoch": 2.6054994162405896,
-      "grad_norm": 0.003811403177678585,
-      "learning_rate": 0.00019999665430138518,
-      "loss": 46.0,
-      "step": 16179
-    },
-    {
-      "epoch": 2.605660453319377,
-      "grad_norm": 0.0016570906154811382,
-      "learning_rate": 0.00019999665388753356,
-      "loss": 46.0,
-      "step": 16180
-    },
-    {
-      "epoch": 2.605821490398164,
-      "grad_norm": 0.0015925776679068804,
-      "learning_rate": 0.00019999665347365635,
-      "loss": 46.0,
-      "step": 16181
-    },
-    {
-      "epoch": 2.6059825274769515,
-      "grad_norm": 0.006657804828137159,
-      "learning_rate": 0.00019999665305975355,
-      "loss": 46.0,
-      "step": 16182
-    },
-    {
-      "epoch": 2.606143564555739,
-      "grad_norm": 0.005006028339266777,
-      "learning_rate": 0.00019999665264582513,
-      "loss": 46.0,
-      "step": 16183
-    },
-    {
-      "epoch": 2.6063046016345264,
-      "grad_norm": 0.00213949684984982,
-      "learning_rate": 0.00019999665223187116,
-      "loss": 46.0,
-      "step": 16184
-    },
-    {
-      "epoch": 2.606465638713314,
-      "grad_norm": 0.0017056141514331102,
-      "learning_rate": 0.00019999665181789155,
-      "loss": 46.0,
-      "step": 16185
-    },
-    {
-      "epoch": 2.6066266757921013,
-      "grad_norm": 0.01463579386472702,
-      "learning_rate": 0.00019999665140388637,
-      "loss": 46.0,
-      "step": 16186
-    },
-    {
-      "epoch": 2.6067877128708883,
-      "grad_norm": 0.005354403052479029,
-      "learning_rate": 0.0001999966509898556,
-      "loss": 46.0,
-      "step": 16187
-    },
-    {
-      "epoch": 2.6069487499496757,
-      "grad_norm": 0.0022061849012970924,
-      "learning_rate": 0.0001999966505757992,
-      "loss": 46.0,
-      "step": 16188
-    },
-    {
-      "epoch": 2.607109787028463,
-      "grad_norm": 0.024766871705651283,
-      "learning_rate": 0.00019999665016171724,
-      "loss": 46.0,
-      "step": 16189
-    },
-    {
-      "epoch": 2.6072708241072506,
-      "grad_norm": 0.0014470957685261965,
-      "learning_rate": 0.0001999966497476097,
-      "loss": 46.0,
-      "step": 16190
-    },
-    {
-      "epoch": 2.607431861186038,
-      "grad_norm": 0.009810199029743671,
-      "learning_rate": 0.00019999664933347656,
-      "loss": 46.0,
-      "step": 16191
-    },
-    {
-      "epoch": 2.6075928982648255,
-      "grad_norm": 0.008456065319478512,
-      "learning_rate": 0.0001999966489193178,
-      "loss": 46.0,
-      "step": 16192
-    },
-    {
-      "epoch": 2.607753935343613,
-      "grad_norm": 0.0027164542116224766,
-      "learning_rate": 0.00019999664850513344,
-      "loss": 46.0,
-      "step": 16193
-    },
-    {
-      "epoch": 2.6079149724224004,
-      "grad_norm": 0.0014124843291938305,
-      "learning_rate": 0.00019999664809092354,
-      "loss": 46.0,
-      "step": 16194
-    },
-    {
-      "epoch": 2.608076009501188,
-      "grad_norm": 0.005745426286011934,
-      "learning_rate": 0.000199996647676688,
-      "loss": 46.0,
-      "step": 16195
-    },
-    {
-      "epoch": 2.608237046579975,
-      "grad_norm": 0.0013142259558662772,
-      "learning_rate": 0.00019999664726242687,
-      "loss": 46.0,
-      "step": 16196
-    },
-    {
-      "epoch": 2.6083980836587624,
-      "grad_norm": 0.003737054066732526,
-      "learning_rate": 0.00019999664684814016,
-      "loss": 46.0,
-      "step": 16197
-    },
-    {
-      "epoch": 2.60855912073755,
-      "grad_norm": 0.0020233402028679848,
-      "learning_rate": 0.00019999664643382783,
-      "loss": 46.0,
-      "step": 16198
-    },
-    {
-      "epoch": 2.6087201578163373,
-      "grad_norm": 0.002590385265648365,
-      "learning_rate": 0.00019999664601948994,
-      "loss": 46.0,
-      "step": 16199
-    },
-    {
-      "epoch": 2.6088811948951247,
-      "grad_norm": 0.004802698735147715,
-      "learning_rate": 0.00019999664560512646,
-      "loss": 46.0,
-      "step": 16200
-    },
-    {
-      "epoch": 2.6090422319739117,
-      "grad_norm": 0.0017088550375774503,
-      "learning_rate": 0.00019999664519073734,
-      "loss": 46.0,
-      "step": 16201
-    },
-    {
-      "epoch": 2.609203269052699,
-      "grad_norm": 0.0033691904973238707,
-      "learning_rate": 0.00019999664477632267,
-      "loss": 46.0,
-      "step": 16202
-    },
-    {
-      "epoch": 2.6093643061314866,
-      "grad_norm": 0.007423325441777706,
-      "learning_rate": 0.00019999664436188237,
-      "loss": 46.0,
-      "step": 16203
-    },
-    {
-      "epoch": 2.609525343210274,
-      "grad_norm": 0.0019671330228447914,
-      "learning_rate": 0.0001999966439474165,
-      "loss": 46.0,
-      "step": 16204
-    },
-    {
-      "epoch": 2.6096863802890615,
-      "grad_norm": 0.0012009054189547896,
-      "learning_rate": 0.00019999664353292503,
-      "loss": 46.0,
-      "step": 16205
-    },
-    {
-      "epoch": 2.609847417367849,
-      "grad_norm": 0.0010096444748342037,
-      "learning_rate": 0.00019999664311840798,
-      "loss": 46.0,
-      "step": 16206
-    },
-    {
-      "epoch": 2.6100084544466364,
-      "grad_norm": 0.0012120584724470973,
-      "learning_rate": 0.0001999966427038653,
-      "loss": 46.0,
-      "step": 16207
-    },
-    {
-      "epoch": 2.610169491525424,
-      "grad_norm": 0.0019064630614593625,
-      "learning_rate": 0.00019999664228929705,
-      "loss": 46.0,
-      "step": 16208
-    },
-    {
-      "epoch": 2.6103305286042113,
-      "grad_norm": 0.008776336908340454,
-      "learning_rate": 0.0001999966418747032,
-      "loss": 46.0,
-      "step": 16209
-    },
-    {
-      "epoch": 2.610491565682999,
-      "grad_norm": 0.003920337185263634,
-      "learning_rate": 0.00019999664146008378,
-      "loss": 46.0,
-      "step": 16210
-    },
-    {
-      "epoch": 2.610652602761786,
-      "grad_norm": 0.0019731582142412663,
-      "learning_rate": 0.00019999664104543874,
-      "loss": 46.0,
-      "step": 16211
-    },
-    {
-      "epoch": 2.6108136398405732,
-      "grad_norm": 0.0006241137161850929,
-      "learning_rate": 0.0001999966406307681,
-      "loss": 46.0,
-      "step": 16212
-    },
-    {
-      "epoch": 2.6109746769193607,
-      "grad_norm": 0.02755666896700859,
-      "learning_rate": 0.00019999664021607186,
-      "loss": 46.0,
-      "step": 16213
-    },
-    {
-      "epoch": 2.611135713998148,
-      "grad_norm": 0.0020803448278456926,
-      "learning_rate": 0.00019999663980135006,
-      "loss": 46.0,
-      "step": 16214
-    },
-    {
-      "epoch": 2.6112967510769356,
-      "grad_norm": 0.0009348891908302903,
-      "learning_rate": 0.00019999663938660266,
-      "loss": 46.0,
-      "step": 16215
-    },
-    {
-      "epoch": 2.6114577881557226,
-      "grad_norm": 0.002950292080640793,
-      "learning_rate": 0.00019999663897182966,
-      "loss": 46.0,
-      "step": 16216
-    },
-    {
-      "epoch": 2.61161882523451,
-      "grad_norm": 0.0013186151627451181,
-      "learning_rate": 0.00019999663855703106,
-      "loss": 46.0,
-      "step": 16217
-    },
-    {
-      "epoch": 2.6117798623132975,
-      "grad_norm": 0.000810387427918613,
-      "learning_rate": 0.00019999663814220688,
-      "loss": 46.0,
-      "step": 16218
-    },
-    {
-      "epoch": 2.611940899392085,
-      "grad_norm": 0.000788987148553133,
-      "learning_rate": 0.0001999966377273571,
-      "loss": 46.0,
-      "step": 16219
-    },
-    {
-      "epoch": 2.6121019364708724,
-      "grad_norm": 0.00318744033575058,
-      "learning_rate": 0.0001999966373124817,
-      "loss": 46.0,
-      "step": 16220
-    },
-    {
-      "epoch": 2.61226297354966,
-      "grad_norm": 0.0044906553812325,
-      "learning_rate": 0.00019999663689758074,
-      "loss": 46.0,
-      "step": 16221
-    },
-    {
-      "epoch": 2.6124240106284473,
-      "grad_norm": 0.00794381182640791,
-      "learning_rate": 0.00019999663648265415,
-      "loss": 46.0,
-      "step": 16222
-    },
-    {
-      "epoch": 2.6125850477072348,
-      "grad_norm": 0.001446620561182499,
-      "learning_rate": 0.000199996636067702,
-      "loss": 46.0,
-      "step": 16223
-    },
-    {
-      "epoch": 2.612746084786022,
-      "grad_norm": 0.004321327432990074,
-      "learning_rate": 0.00019999663565272425,
-      "loss": 46.0,
-      "step": 16224
-    },
-    {
-      "epoch": 2.6129071218648092,
-      "grad_norm": 0.0009887315100058913,
-      "learning_rate": 0.0001999966352377209,
-      "loss": 46.0,
-      "step": 16225
-    },
-    {
-      "epoch": 2.6130681589435967,
-      "grad_norm": 0.0034912792034447193,
-      "learning_rate": 0.00019999663482269195,
-      "loss": 46.0,
-      "step": 16226
-    },
-    {
-      "epoch": 2.613229196022384,
-      "grad_norm": 0.0031000266317278147,
-      "learning_rate": 0.00019999663440763743,
-      "loss": 46.0,
-      "step": 16227
-    },
-    {
-      "epoch": 2.6133902331011716,
-      "grad_norm": 0.00028321624267846346,
-      "learning_rate": 0.0001999966339925573,
-      "loss": 46.0,
-      "step": 16228
-    },
-    {
-      "epoch": 2.613551270179959,
-      "grad_norm": 0.0017997113754972816,
-      "learning_rate": 0.00019999663357745154,
-      "loss": 46.0,
-      "step": 16229
-    },
-    {
-      "epoch": 2.6137123072587465,
-      "grad_norm": 0.002274578670039773,
-      "learning_rate": 0.00019999663316232024,
-      "loss": 46.0,
-      "step": 16230
-    },
-    {
-      "epoch": 2.6138733443375335,
-      "grad_norm": 0.0009919337462633848,
-      "learning_rate": 0.00019999663274716331,
-      "loss": 46.0,
-      "step": 16231
-    },
-    {
-      "epoch": 2.614034381416321,
-      "grad_norm": 0.0007810642127878964,
-      "learning_rate": 0.0001999966323319808,
-      "loss": 46.0,
-      "step": 16232
-    },
-    {
-      "epoch": 2.6141954184951084,
-      "grad_norm": 0.003255092306062579,
-      "learning_rate": 0.0001999966319167727,
-      "loss": 46.0,
-      "step": 16233
-    },
-    {
-      "epoch": 2.614356455573896,
-      "grad_norm": 0.0023576554376631975,
-      "learning_rate": 0.00019999663150153902,
-      "loss": 46.0,
-      "step": 16234
-    },
-    {
-      "epoch": 2.6145174926526833,
-      "grad_norm": 0.0025528348051011562,
-      "learning_rate": 0.00019999663108627972,
-      "loss": 46.0,
-      "step": 16235
-    },
-    {
-      "epoch": 2.6146785297314707,
-      "grad_norm": 0.0012402617139741778,
-      "learning_rate": 0.00019999663067099484,
-      "loss": 46.0,
-      "step": 16236
-    },
-    {
-      "epoch": 2.614839566810258,
-      "grad_norm": 0.0059888362884521484,
-      "learning_rate": 0.0001999966302556844,
-      "loss": 46.0,
-      "step": 16237
-    },
-    {
-      "epoch": 2.6150006038890456,
-      "grad_norm": 0.0010131146991625428,
-      "learning_rate": 0.0001999966298403483,
-      "loss": 46.0,
-      "step": 16238
-    },
-    {
-      "epoch": 2.615161640967833,
-      "grad_norm": 0.007483948487788439,
-      "learning_rate": 0.00019999662942498663,
-      "loss": 46.0,
-      "step": 16239
-    },
-    {
-      "epoch": 2.61532267804662,
-      "grad_norm": 0.0021092721726745367,
-      "learning_rate": 0.00019999662900959937,
-      "loss": 46.0,
-      "step": 16240
-    },
-    {
-      "epoch": 2.6154837151254076,
-      "grad_norm": 0.007061942480504513,
-      "learning_rate": 0.0001999966285941865,
-      "loss": 46.0,
-      "step": 16241
-    },
-    {
-      "epoch": 2.615644752204195,
-      "grad_norm": 0.0026891794987022877,
-      "learning_rate": 0.00019999662817874806,
-      "loss": 46.0,
-      "step": 16242
-    },
-    {
-      "epoch": 2.6158057892829825,
-      "grad_norm": 0.005143879912793636,
-      "learning_rate": 0.00019999662776328404,
-      "loss": 46.0,
-      "step": 16243
-    },
-    {
-      "epoch": 2.61596682636177,
-      "grad_norm": 0.002124219899997115,
-      "learning_rate": 0.0001999966273477944,
-      "loss": 46.0,
-      "step": 16244
-    },
-    {
-      "epoch": 2.616127863440557,
-      "grad_norm": 0.0013617079239338636,
-      "learning_rate": 0.00019999662693227918,
-      "loss": 46.0,
-      "step": 16245
-    },
-    {
-      "epoch": 2.6162889005193444,
-      "grad_norm": 0.0014032250037416816,
-      "learning_rate": 0.00019999662651673834,
-      "loss": 46.0,
-      "step": 16246
-    },
-    {
-      "epoch": 2.616449937598132,
-      "grad_norm": 0.0035202796570956707,
-      "learning_rate": 0.0001999966261011719,
-      "loss": 46.0,
-      "step": 16247
-    },
-    {
-      "epoch": 2.6166109746769193,
-      "grad_norm": 0.000518196087796241,
-      "learning_rate": 0.0001999966256855799,
-      "loss": 46.0,
-      "step": 16248
-    },
-    {
-      "epoch": 2.6167720117557067,
-      "grad_norm": 0.001013385714031756,
-      "learning_rate": 0.0001999966252699623,
-      "loss": 46.0,
-      "step": 16249
-    },
-    {
-      "epoch": 2.616933048834494,
-      "grad_norm": 0.0013176481006667018,
-      "learning_rate": 0.0001999966248543191,
-      "loss": 46.0,
-      "step": 16250
-    },
-    {
-      "epoch": 2.6170940859132816,
-      "grad_norm": 0.001168896909803152,
-      "learning_rate": 0.0001999966244386503,
-      "loss": 46.0,
-      "step": 16251
-    },
-    {
-      "epoch": 2.617255122992069,
-      "grad_norm": 0.011545871384441853,
-      "learning_rate": 0.00019999662402295592,
-      "loss": 46.0,
-      "step": 16252
-    },
-    {
-      "epoch": 2.6174161600708565,
-      "grad_norm": 0.007152176462113857,
-      "learning_rate": 0.00019999662360723592,
-      "loss": 46.0,
-      "step": 16253
-    },
-    {
-      "epoch": 2.6175771971496435,
-      "grad_norm": 0.001512520364485681,
-      "learning_rate": 0.00019999662319149036,
-      "loss": 46.0,
-      "step": 16254
-    },
-    {
-      "epoch": 2.617738234228431,
-      "grad_norm": 0.0027331719174981117,
-      "learning_rate": 0.0001999966227757192,
-      "loss": 46.0,
-      "step": 16255
-    },
-    {
-      "epoch": 2.6178992713072184,
-      "grad_norm": 0.0008448728476651013,
-      "learning_rate": 0.00019999662235992241,
-      "loss": 46.0,
-      "step": 16256
-    },
-    {
-      "epoch": 2.618060308386006,
-      "grad_norm": 0.0049787405878305435,
-      "learning_rate": 0.00019999662194410006,
-      "loss": 46.0,
-      "step": 16257
-    },
-    {
-      "epoch": 2.6182213454647933,
-      "grad_norm": 0.004492932464927435,
-      "learning_rate": 0.00019999662152825212,
-      "loss": 46.0,
-      "step": 16258
-    },
-    {
-      "epoch": 2.618382382543581,
-      "grad_norm": 0.0031222079414874315,
-      "learning_rate": 0.00019999662111237857,
-      "loss": 46.0,
-      "step": 16259
-    },
-    {
-      "epoch": 2.618543419622368,
-      "grad_norm": 0.010371342301368713,
-      "learning_rate": 0.00019999662069647943,
-      "loss": 46.0,
-      "step": 16260
-    },
-    {
-      "epoch": 2.6187044567011553,
-      "grad_norm": 0.002968019340187311,
-      "learning_rate": 0.0001999966202805547,
-      "loss": 46.0,
-      "step": 16261
-    },
-    {
-      "epoch": 2.6188654937799427,
-      "grad_norm": 0.011174728162586689,
-      "learning_rate": 0.0001999966198646044,
-      "loss": 46.0,
-      "step": 16262
-    },
-    {
-      "epoch": 2.61902653085873,
-      "grad_norm": 0.001504142303019762,
-      "learning_rate": 0.00019999661944862846,
-      "loss": 46.0,
-      "step": 16263
-    },
-    {
-      "epoch": 2.6191875679375176,
-      "grad_norm": 0.0010710597271099687,
-      "learning_rate": 0.00019999661903262692,
-      "loss": 46.0,
-      "step": 16264
-    },
-    {
-      "epoch": 2.619348605016305,
-      "grad_norm": 0.0027598801534622908,
-      "learning_rate": 0.00019999661861659981,
-      "loss": 46.0,
-      "step": 16265
-    },
-    {
-      "epoch": 2.6195096420950925,
-      "grad_norm": 0.002100490964949131,
-      "learning_rate": 0.00019999661820054712,
-      "loss": 46.0,
-      "step": 16266
-    },
-    {
-      "epoch": 2.61967067917388,
-      "grad_norm": 0.00361101096495986,
-      "learning_rate": 0.00019999661778446882,
-      "loss": 46.0,
-      "step": 16267
-    },
-    {
-      "epoch": 2.6198317162526674,
-      "grad_norm": 0.0012040912406519055,
-      "learning_rate": 0.00019999661736836493,
-      "loss": 46.0,
-      "step": 16268
-    },
-    {
-      "epoch": 2.6199927533314544,
-      "grad_norm": 0.0016230870969593525,
-      "learning_rate": 0.00019999661695223545,
-      "loss": 46.0,
-      "step": 16269
-    },
-    {
-      "epoch": 2.620153790410242,
-      "grad_norm": 0.006094573996961117,
-      "learning_rate": 0.00019999661653608035,
-      "loss": 46.0,
-      "step": 16270
-    },
-    {
-      "epoch": 2.6203148274890293,
-      "grad_norm": 0.0020876533817499876,
-      "learning_rate": 0.0001999966161198997,
-      "loss": 46.0,
-      "step": 16271
-    },
-    {
-      "epoch": 2.620475864567817,
-      "grad_norm": 0.014123592525720596,
-      "learning_rate": 0.00019999661570369343,
-      "loss": 46.0,
-      "step": 16272
-    },
-    {
-      "epoch": 2.6206369016466042,
-      "grad_norm": 0.0008789273560978472,
-      "learning_rate": 0.00019999661528746155,
-      "loss": 46.0,
-      "step": 16273
-    },
-    {
-      "epoch": 2.6207979387253917,
-      "grad_norm": 0.0037456608843058348,
-      "learning_rate": 0.0001999966148712041,
-      "loss": 46.0,
-      "step": 16274
-    },
-    {
-      "epoch": 2.6209589758041787,
-      "grad_norm": 0.004488199949264526,
-      "learning_rate": 0.00019999661445492105,
-      "loss": 46.0,
-      "step": 16275
-    },
-    {
-      "epoch": 2.621120012882966,
-      "grad_norm": 0.005014956928789616,
-      "learning_rate": 0.0001999966140386124,
-      "loss": 46.0,
-      "step": 16276
-    },
-    {
-      "epoch": 2.6212810499617536,
-      "grad_norm": 0.003245734144002199,
-      "learning_rate": 0.00019999661362227818,
-      "loss": 46.0,
-      "step": 16277
-    },
-    {
-      "epoch": 2.621442087040541,
-      "grad_norm": 0.004849734716117382,
-      "learning_rate": 0.00019999661320591834,
-      "loss": 46.0,
-      "step": 16278
-    },
-    {
-      "epoch": 2.6216031241193285,
-      "grad_norm": 0.0035920669324696064,
-      "learning_rate": 0.00019999661278953293,
-      "loss": 46.0,
-      "step": 16279
-    },
-    {
-      "epoch": 2.621764161198116,
-      "grad_norm": 0.0021898269187659025,
-      "learning_rate": 0.00019999661237312188,
-      "loss": 46.0,
-      "step": 16280
-    },
-    {
-      "epoch": 2.6219251982769034,
-      "grad_norm": 0.001706786802969873,
-      "learning_rate": 0.00019999661195668528,
-      "loss": 46.0,
-      "step": 16281
-    },
-    {
-      "epoch": 2.622086235355691,
-      "grad_norm": 0.002933080308139324,
-      "learning_rate": 0.00019999661154022306,
-      "loss": 46.0,
-      "step": 16282
-    },
-    {
-      "epoch": 2.6222472724344783,
-      "grad_norm": 0.002068144269287586,
-      "learning_rate": 0.00019999661112373528,
-      "loss": 46.0,
-      "step": 16283
-    },
-    {
-      "epoch": 2.6224083095132653,
-      "grad_norm": 0.007859934121370316,
-      "learning_rate": 0.00019999661070722188,
-      "loss": 46.0,
-      "step": 16284
-    },
-    {
-      "epoch": 2.6225693465920528,
-      "grad_norm": 0.00048201545723713934,
-      "learning_rate": 0.00019999661029068287,
-      "loss": 46.0,
-      "step": 16285
-    },
-    {
-      "epoch": 2.62273038367084,
-      "grad_norm": 0.010907069779932499,
-      "learning_rate": 0.0001999966098741183,
-      "loss": 46.0,
-      "step": 16286
-    },
-    {
-      "epoch": 2.6228914207496277,
-      "grad_norm": 0.004363309592008591,
-      "learning_rate": 0.00019999660945752812,
-      "loss": 46.0,
-      "step": 16287
-    },
-    {
-      "epoch": 2.623052457828415,
-      "grad_norm": 0.009198774583637714,
-      "learning_rate": 0.00019999660904091235,
-      "loss": 46.0,
-      "step": 16288
-    },
-    {
-      "epoch": 2.623213494907202,
-      "grad_norm": 0.0019503688672557473,
-      "learning_rate": 0.000199996608624271,
-      "loss": 46.0,
-      "step": 16289
-    },
-    {
-      "epoch": 2.6233745319859896,
-      "grad_norm": 0.0022886747028678656,
-      "learning_rate": 0.00019999660820760402,
-      "loss": 46.0,
-      "step": 16290
-    },
-    {
-      "epoch": 2.623535569064777,
-      "grad_norm": 0.0052734678611159325,
-      "learning_rate": 0.00019999660779091146,
-      "loss": 46.0,
-      "step": 16291
-    },
-    {
-      "epoch": 2.6236966061435645,
-      "grad_norm": 0.0016602105461061,
-      "learning_rate": 0.0001999966073741933,
-      "loss": 46.0,
-      "step": 16292
-    },
-    {
-      "epoch": 2.623857643222352,
-      "grad_norm": 0.0042768982239067554,
-      "learning_rate": 0.00019999660695744958,
-      "loss": 46.0,
-      "step": 16293
-    },
-    {
-      "epoch": 2.6240186803011394,
-      "grad_norm": 0.007833343930542469,
-      "learning_rate": 0.00019999660654068023,
-      "loss": 46.0,
-      "step": 16294
-    },
-    {
-      "epoch": 2.624179717379927,
-      "grad_norm": 0.0031322180293500423,
-      "learning_rate": 0.00019999660612388532,
-      "loss": 46.0,
-      "step": 16295
-    },
-    {
-      "epoch": 2.6243407544587143,
-      "grad_norm": 0.0011031373869627714,
-      "learning_rate": 0.0001999966057070648,
-      "loss": 46.0,
-      "step": 16296
-    },
-    {
-      "epoch": 2.6245017915375017,
-      "grad_norm": 0.003968015778809786,
-      "learning_rate": 0.0001999966052902187,
-      "loss": 46.0,
-      "step": 16297
-    },
-    {
-      "epoch": 2.6246628286162887,
-      "grad_norm": 0.0004365532658994198,
-      "learning_rate": 0.00019999660487334694,
-      "loss": 46.0,
-      "step": 16298
-    },
-    {
-      "epoch": 2.624823865695076,
-      "grad_norm": 0.0008367628324776888,
-      "learning_rate": 0.00019999660445644966,
-      "loss": 46.0,
-      "step": 16299
-    },
-    {
-      "epoch": 2.6249849027738636,
-      "grad_norm": 0.007303046062588692,
-      "learning_rate": 0.00019999660403952676,
-      "loss": 46.0,
-      "step": 16300
-    },
-    {
-      "epoch": 2.625145939852651,
-      "grad_norm": 0.0011429721489548683,
-      "learning_rate": 0.00019999660362257827,
-      "loss": 46.0,
-      "step": 16301
-    },
-    {
-      "epoch": 2.6253069769314386,
-      "grad_norm": 0.004899218678474426,
-      "learning_rate": 0.00019999660320560417,
-      "loss": 46.0,
-      "step": 16302
-    },
-    {
-      "epoch": 2.625468014010226,
-      "grad_norm": 0.00478737335652113,
-      "learning_rate": 0.0001999966027886045,
-      "loss": 46.0,
-      "step": 16303
-    },
-    {
-      "epoch": 2.625629051089013,
-      "grad_norm": 0.001814377843402326,
-      "learning_rate": 0.0001999966023715792,
-      "loss": 46.0,
-      "step": 16304
-    },
-    {
-      "epoch": 2.6257900881678005,
-      "grad_norm": 0.0023631909862160683,
-      "learning_rate": 0.00019999660195452835,
-      "loss": 46.0,
-      "step": 16305
-    },
-    {
-      "epoch": 2.625951125246588,
-      "grad_norm": 0.010905549861490726,
-      "learning_rate": 0.00019999660153745188,
-      "loss": 46.0,
-      "step": 16306
-    },
-    {
-      "epoch": 2.6261121623253754,
-      "grad_norm": 0.019005555659532547,
-      "learning_rate": 0.00019999660112034984,
-      "loss": 46.0,
-      "step": 16307
-    },
-    {
-      "epoch": 2.626273199404163,
-      "grad_norm": 0.0016651147743687034,
-      "learning_rate": 0.00019999660070322216,
-      "loss": 46.0,
-      "step": 16308
-    },
-    {
-      "epoch": 2.6264342364829503,
-      "grad_norm": 0.003206402761861682,
-      "learning_rate": 0.00019999660028606893,
-      "loss": 46.0,
-      "step": 16309
-    },
-    {
-      "epoch": 2.6265952735617377,
-      "grad_norm": 0.013352095149457455,
-      "learning_rate": 0.00019999659986889008,
-      "loss": 46.0,
-      "step": 16310
-    },
-    {
-      "epoch": 2.626756310640525,
-      "grad_norm": 0.0025833584368228912,
-      "learning_rate": 0.00019999659945168564,
-      "loss": 46.0,
-      "step": 16311
-    },
-    {
-      "epoch": 2.6269173477193126,
-      "grad_norm": 0.003954403102397919,
-      "learning_rate": 0.0001999965990344556,
-      "loss": 46.0,
-      "step": 16312
-    },
-    {
-      "epoch": 2.6270783847980996,
-      "grad_norm": 0.006242019589990377,
-      "learning_rate": 0.00019999659861720003,
-      "loss": 46.0,
-      "step": 16313
-    },
-    {
-      "epoch": 2.627239421876887,
-      "grad_norm": 0.005097391549497843,
-      "learning_rate": 0.00019999659819991877,
-      "loss": 46.0,
-      "step": 16314
-    },
-    {
-      "epoch": 2.6274004589556745,
-      "grad_norm": 0.0010899916524067521,
-      "learning_rate": 0.00019999659778261196,
-      "loss": 46.0,
-      "step": 16315
-    },
-    {
-      "epoch": 2.627561496034462,
-      "grad_norm": 0.006583353504538536,
-      "learning_rate": 0.00019999659736527958,
-      "loss": 46.0,
-      "step": 16316
-    },
-    {
-      "epoch": 2.6277225331132494,
-      "grad_norm": 0.0010600329842418432,
-      "learning_rate": 0.00019999659694792157,
-      "loss": 46.0,
-      "step": 16317
-    },
-    {
-      "epoch": 2.6278835701920364,
-      "grad_norm": 0.001082659000530839,
-      "learning_rate": 0.00019999659653053797,
-      "loss": 46.0,
-      "step": 16318
-    },
-    {
-      "epoch": 2.628044607270824,
-      "grad_norm": 0.004644326400011778,
-      "learning_rate": 0.0001999965961131288,
-      "loss": 46.0,
-      "step": 16319
-    },
-    {
-      "epoch": 2.6282056443496113,
-      "grad_norm": 0.003597764065489173,
-      "learning_rate": 0.000199996595695694,
-      "loss": 46.0,
-      "step": 16320
-    },
-    {
-      "epoch": 2.628366681428399,
-      "grad_norm": 0.0030136071145534515,
-      "learning_rate": 0.00019999659527823364,
-      "loss": 46.0,
-      "step": 16321
-    },
-    {
-      "epoch": 2.6285277185071863,
-      "grad_norm": 0.0018900932045653462,
-      "learning_rate": 0.00019999659486074766,
-      "loss": 46.0,
-      "step": 16322
-    },
-    {
-      "epoch": 2.6286887555859737,
-      "grad_norm": 0.0007270133937709033,
-      "learning_rate": 0.00019999659444323612,
-      "loss": 46.0,
-      "step": 16323
-    },
-    {
-      "epoch": 2.628849792664761,
-      "grad_norm": 0.0011344680096954107,
-      "learning_rate": 0.00019999659402569897,
-      "loss": 46.0,
-      "step": 16324
-    },
-    {
-      "epoch": 2.6290108297435486,
-      "grad_norm": 0.005970783997327089,
-      "learning_rate": 0.0001999965936081362,
-      "loss": 46.0,
-      "step": 16325
-    },
-    {
-      "epoch": 2.629171866822336,
-      "grad_norm": 0.001079970272257924,
-      "learning_rate": 0.00019999659319054787,
-      "loss": 46.0,
-      "step": 16326
-    },
-    {
-      "epoch": 2.6293329039011235,
-      "grad_norm": 0.001130154705606401,
-      "learning_rate": 0.00019999659277293393,
-      "loss": 46.0,
-      "step": 16327
-    },
-    {
-      "epoch": 2.6294939409799105,
-      "grad_norm": 0.003891661064699292,
-      "learning_rate": 0.0001999965923552944,
-      "loss": 46.0,
-      "step": 16328
-    },
-    {
-      "epoch": 2.629654978058698,
-      "grad_norm": 0.0006424893508665264,
-      "learning_rate": 0.00019999659193762926,
-      "loss": 46.0,
-      "step": 16329
-    },
-    {
-      "epoch": 2.6298160151374854,
-      "grad_norm": 0.003793942043557763,
-      "learning_rate": 0.00019999659151993855,
-      "loss": 46.0,
-      "step": 16330
-    },
-    {
-      "epoch": 2.629977052216273,
-      "grad_norm": 0.004247124306857586,
-      "learning_rate": 0.0001999965911022222,
-      "loss": 46.0,
-      "step": 16331
-    },
-    {
-      "epoch": 2.6301380892950603,
-      "grad_norm": 0.010247626341879368,
-      "learning_rate": 0.00019999659068448033,
-      "loss": 46.0,
-      "step": 16332
-    },
-    {
-      "epoch": 2.6302991263738473,
-      "grad_norm": 0.012627347372472286,
-      "learning_rate": 0.00019999659026671281,
-      "loss": 46.0,
-      "step": 16333
-    },
-    {
-      "epoch": 2.630460163452635,
-      "grad_norm": 0.00166188960429281,
-      "learning_rate": 0.00019999658984891974,
-      "loss": 46.0,
-      "step": 16334
-    },
-    {
-      "epoch": 2.6306212005314222,
-      "grad_norm": 0.0021874040830880404,
-      "learning_rate": 0.00019999658943110102,
-      "loss": 46.0,
-      "step": 16335
-    },
-    {
-      "epoch": 2.6307822376102097,
-      "grad_norm": 0.006883494555950165,
-      "learning_rate": 0.00019999658901325676,
-      "loss": 46.0,
-      "step": 16336
-    },
-    {
-      "epoch": 2.630943274688997,
-      "grad_norm": 0.003918439615517855,
-      "learning_rate": 0.00019999658859538687,
-      "loss": 46.0,
-      "step": 16337
-    },
-    {
-      "epoch": 2.6311043117677846,
-      "grad_norm": 0.0025312795769423246,
-      "learning_rate": 0.0001999965881774914,
-      "loss": 46.0,
-      "step": 16338
-    },
-    {
-      "epoch": 2.631265348846572,
-      "grad_norm": 0.01011947263032198,
-      "learning_rate": 0.00019999658775957034,
-      "loss": 46.0,
-      "step": 16339
-    },
-    {
-      "epoch": 2.6314263859253595,
-      "grad_norm": 0.00772851100191474,
-      "learning_rate": 0.0001999965873416237,
-      "loss": 46.0,
-      "step": 16340
-    },
-    {
-      "epoch": 2.631587423004147,
-      "grad_norm": 0.0011113417567685246,
-      "learning_rate": 0.00019999658692365142,
-      "loss": 46.0,
-      "step": 16341
-    },
-    {
-      "epoch": 2.631748460082934,
-      "grad_norm": 0.0014731999253854156,
-      "learning_rate": 0.0001999965865056536,
-      "loss": 46.0,
-      "step": 16342
-    },
-    {
-      "epoch": 2.6319094971617214,
-      "grad_norm": 0.0020360820926725864,
-      "learning_rate": 0.00019999658608763014,
-      "loss": 46.0,
-      "step": 16343
-    },
-    {
-      "epoch": 2.632070534240509,
-      "grad_norm": 0.003389952937141061,
-      "learning_rate": 0.00019999658566958108,
-      "loss": 46.0,
-      "step": 16344
-    },
-    {
-      "epoch": 2.6322315713192963,
-      "grad_norm": 0.007669869810342789,
-      "learning_rate": 0.00019999658525150647,
-      "loss": 46.0,
-      "step": 16345
-    },
-    {
-      "epoch": 2.6323926083980838,
-      "grad_norm": 0.0030061539728194475,
-      "learning_rate": 0.00019999658483340623,
-      "loss": 46.0,
-      "step": 16346
-    },
-    {
-      "epoch": 2.632553645476871,
-      "grad_norm": 0.011876963078975677,
-      "learning_rate": 0.0001999965844152804,
-      "loss": 46.0,
-      "step": 16347
-    },
-    {
-      "epoch": 2.632714682555658,
-      "grad_norm": 0.0028468468226492405,
-      "learning_rate": 0.000199996583997129,
-      "loss": 46.0,
-      "step": 16348
-    },
-    {
-      "epoch": 2.6328757196344457,
-      "grad_norm": 0.003951325546950102,
-      "learning_rate": 0.00019999658357895198,
-      "loss": 46.0,
-      "step": 16349
-    },
-    {
-      "epoch": 2.633036756713233,
-      "grad_norm": 0.0010181076359003782,
-      "learning_rate": 0.0001999965831607494,
-      "loss": 46.0,
-      "step": 16350
-    },
-    {
-      "epoch": 2.6331977937920206,
-      "grad_norm": 0.005227404646575451,
-      "learning_rate": 0.0001999965827425212,
-      "loss": 46.0,
-      "step": 16351
-    },
-    {
-      "epoch": 2.633358830870808,
-      "grad_norm": 0.0031755631789565086,
-      "learning_rate": 0.00019999658232426742,
-      "loss": 46.0,
-      "step": 16352
-    },
-    {
-      "epoch": 2.6335198679495955,
-      "grad_norm": 0.008981127291917801,
-      "learning_rate": 0.00019999658190598805,
-      "loss": 46.0,
-      "step": 16353
-    },
-    {
-      "epoch": 2.633680905028383,
-      "grad_norm": 0.0023199168499559164,
-      "learning_rate": 0.00019999658148768307,
-      "loss": 46.0,
-      "step": 16354
-    },
-    {
-      "epoch": 2.6338419421071704,
-      "grad_norm": 0.0010527439881116152,
-      "learning_rate": 0.0001999965810693525,
-      "loss": 46.0,
-      "step": 16355
-    },
-    {
-      "epoch": 2.634002979185958,
-      "grad_norm": 0.0025388584472239017,
-      "learning_rate": 0.00019999658065099633,
-      "loss": 46.0,
-      "step": 16356
-    },
-    {
-      "epoch": 2.634164016264745,
-      "grad_norm": 0.0014624694595113397,
-      "learning_rate": 0.00019999658023261456,
-      "loss": 46.0,
-      "step": 16357
-    },
-    {
-      "epoch": 2.6343250533435323,
-      "grad_norm": 0.0018244596431031823,
-      "learning_rate": 0.0001999965798142072,
-      "loss": 46.0,
-      "step": 16358
-    },
-    {
-      "epoch": 2.6344860904223197,
-      "grad_norm": 0.0009498837753199041,
-      "learning_rate": 0.00019999657939577428,
-      "loss": 46.0,
-      "step": 16359
-    },
-    {
-      "epoch": 2.634647127501107,
-      "grad_norm": 0.0043435897678136826,
-      "learning_rate": 0.00019999657897731574,
-      "loss": 46.0,
-      "step": 16360
-    },
-    {
-      "epoch": 2.6348081645798946,
-      "grad_norm": 0.0012757699005305767,
-      "learning_rate": 0.0001999965785588316,
-      "loss": 46.0,
-      "step": 16361
-    },
-    {
-      "epoch": 2.6349692016586816,
-      "grad_norm": 0.004824684467166662,
-      "learning_rate": 0.00019999657814032186,
-      "loss": 46.0,
-      "step": 16362
-    },
-    {
-      "epoch": 2.635130238737469,
-      "grad_norm": 0.006517571397125721,
-      "learning_rate": 0.00019999657772178656,
-      "loss": 46.0,
-      "step": 16363
-    },
-    {
-      "epoch": 2.6352912758162566,
-      "grad_norm": 0.0072199320420622826,
-      "learning_rate": 0.00019999657730322565,
-      "loss": 46.0,
-      "step": 16364
-    },
-    {
-      "epoch": 2.635452312895044,
-      "grad_norm": 0.0005466794827952981,
-      "learning_rate": 0.00019999657688463913,
-      "loss": 46.0,
-      "step": 16365
-    },
-    {
-      "epoch": 2.6356133499738315,
-      "grad_norm": 0.002083239611238241,
-      "learning_rate": 0.00019999657646602704,
-      "loss": 46.0,
-      "step": 16366
-    },
-    {
-      "epoch": 2.635774387052619,
-      "grad_norm": 0.0070607662200927734,
-      "learning_rate": 0.00019999657604738931,
-      "loss": 46.0,
-      "step": 16367
-    },
-    {
-      "epoch": 2.6359354241314064,
-      "grad_norm": 0.00488640321418643,
-      "learning_rate": 0.00019999657562872606,
-      "loss": 46.0,
-      "step": 16368
-    },
-    {
-      "epoch": 2.636096461210194,
-      "grad_norm": 0.0016258973628282547,
-      "learning_rate": 0.00019999657521003713,
-      "loss": 46.0,
-      "step": 16369
-    },
-    {
-      "epoch": 2.6362574982889813,
-      "grad_norm": 0.0017399077769368887,
-      "learning_rate": 0.00019999657479132267,
-      "loss": 46.0,
-      "step": 16370
-    },
-    {
-      "epoch": 2.6364185353677683,
-      "grad_norm": 0.003619509283453226,
-      "learning_rate": 0.0001999965743725826,
-      "loss": 46.0,
-      "step": 16371
-    },
-    {
-      "epoch": 2.6365795724465557,
-      "grad_norm": 0.00186413642950356,
-      "learning_rate": 0.00019999657395381693,
-      "loss": 46.0,
-      "step": 16372
-    },
-    {
-      "epoch": 2.636740609525343,
-      "grad_norm": 0.0023832248989492655,
-      "learning_rate": 0.00019999657353502568,
-      "loss": 46.0,
-      "step": 16373
-    },
-    {
-      "epoch": 2.6369016466041306,
-      "grad_norm": 0.0005428649019449949,
-      "learning_rate": 0.00019999657311620882,
-      "loss": 46.0,
-      "step": 16374
-    },
-    {
-      "epoch": 2.637062683682918,
-      "grad_norm": 0.000989394960924983,
-      "learning_rate": 0.00019999657269736636,
-      "loss": 46.0,
-      "step": 16375
-    },
-    {
-      "epoch": 2.6372237207617055,
-      "grad_norm": 0.0031070520635694265,
-      "learning_rate": 0.0001999965722784983,
-      "loss": 46.0,
-      "step": 16376
-    },
-    {
-      "epoch": 2.6373847578404925,
-      "grad_norm": 0.0014868798898532987,
-      "learning_rate": 0.00019999657185960467,
-      "loss": 46.0,
-      "step": 16377
-    },
-    {
-      "epoch": 2.63754579491928,
-      "grad_norm": 0.005715236533433199,
-      "learning_rate": 0.00019999657144068543,
-      "loss": 46.0,
-      "step": 16378
-    },
-    {
-      "epoch": 2.6377068319980674,
-      "grad_norm": 0.0018072109669446945,
-      "learning_rate": 0.00019999657102174063,
-      "loss": 46.0,
-      "step": 16379
-    },
-    {
-      "epoch": 2.637867869076855,
-      "grad_norm": 0.0011455584317445755,
-      "learning_rate": 0.00019999657060277022,
-      "loss": 46.0,
-      "step": 16380
-    },
-    {
-      "epoch": 2.6380289061556423,
-      "grad_norm": 0.0022247028537094593,
-      "learning_rate": 0.00019999657018377417,
-      "loss": 46.0,
-      "step": 16381
-    },
-    {
-      "epoch": 2.63818994323443,
-      "grad_norm": 0.0022691092453897,
-      "learning_rate": 0.00019999656976475258,
-      "loss": 46.0,
-      "step": 16382
-    },
-    {
-      "epoch": 2.6383509803132172,
-      "grad_norm": 0.0019194961059838533,
-      "learning_rate": 0.00019999656934570537,
-      "loss": 46.0,
-      "step": 16383
-    },
-    {
-      "epoch": 2.6385120173920047,
-      "grad_norm": 0.0031690141186118126,
-      "learning_rate": 0.00019999656892663259,
-      "loss": 46.0,
-      "step": 16384
-    },
-    {
-      "epoch": 2.638673054470792,
-      "grad_norm": 0.0012817782117053866,
-      "learning_rate": 0.00019999656850753418,
-      "loss": 46.0,
-      "step": 16385
-    },
-    {
-      "epoch": 2.638834091549579,
-      "grad_norm": 0.01577017828822136,
-      "learning_rate": 0.00019999656808841022,
-      "loss": 46.0,
-      "step": 16386
-    },
-    {
-      "epoch": 2.6389951286283666,
-      "grad_norm": 0.0030287159606814384,
-      "learning_rate": 0.00019999656766926064,
-      "loss": 46.0,
-      "step": 16387
-    },
-    {
-      "epoch": 2.639156165707154,
-      "grad_norm": 0.009831307455897331,
-      "learning_rate": 0.00019999656725008547,
-      "loss": 46.0,
-      "step": 16388
-    },
-    {
-      "epoch": 2.6393172027859415,
-      "grad_norm": 0.005242469254881144,
-      "learning_rate": 0.0001999965668308847,
-      "loss": 46.0,
-      "step": 16389
-    },
-    {
-      "epoch": 2.639478239864729,
-      "grad_norm": 0.0026832332368940115,
-      "learning_rate": 0.00019999656641165833,
-      "loss": 46.0,
-      "step": 16390
-    },
-    {
-      "epoch": 2.6396392769435164,
-      "grad_norm": 0.003587692975997925,
-      "learning_rate": 0.0001999965659924064,
-      "loss": 46.0,
-      "step": 16391
-    },
-    {
-      "epoch": 2.6398003140223034,
-      "grad_norm": 0.005455745849758387,
-      "learning_rate": 0.00019999656557312886,
-      "loss": 46.0,
-      "step": 16392
-    },
-    {
-      "epoch": 2.639961351101091,
-      "grad_norm": 0.0031906720250844955,
-      "learning_rate": 0.00019999656515382568,
-      "loss": 46.0,
-      "step": 16393
-    },
-    {
-      "epoch": 2.6401223881798783,
-      "grad_norm": 0.00945283193141222,
-      "learning_rate": 0.00019999656473449696,
-      "loss": 46.0,
-      "step": 16394
-    },
-    {
-      "epoch": 2.6402834252586658,
-      "grad_norm": 0.0037267126608639956,
-      "learning_rate": 0.00019999656431514263,
-      "loss": 46.0,
-      "step": 16395
-    },
-    {
-      "epoch": 2.6404444623374532,
-      "grad_norm": 0.0008340401109308004,
-      "learning_rate": 0.00019999656389576269,
-      "loss": 46.0,
-      "step": 16396
-    },
-    {
-      "epoch": 2.6406054994162407,
-      "grad_norm": 0.0025725211016833782,
-      "learning_rate": 0.0001999965634763572,
-      "loss": 46.0,
-      "step": 16397
-    },
-    {
-      "epoch": 2.640766536495028,
-      "grad_norm": 0.0006869092467240989,
-      "learning_rate": 0.0001999965630569261,
-      "loss": 46.0,
-      "step": 16398
-    },
-    {
-      "epoch": 2.6409275735738156,
-      "grad_norm": 0.0019361295271664858,
-      "learning_rate": 0.00019999656263746936,
-      "loss": 46.0,
-      "step": 16399
-    },
-    {
-      "epoch": 2.641088610652603,
-      "grad_norm": 0.008035583421587944,
-      "learning_rate": 0.0001999965622179871,
-      "loss": 46.0,
-      "step": 16400
-    },
-    {
-      "epoch": 2.64124964773139,
-      "grad_norm": 0.003356541972607374,
-      "learning_rate": 0.00019999656179847918,
-      "loss": 46.0,
-      "step": 16401
-    },
-    {
-      "epoch": 2.6414106848101775,
-      "grad_norm": 0.003267604624852538,
-      "learning_rate": 0.0001999965613789457,
-      "loss": 46.0,
-      "step": 16402
-    },
-    {
-      "epoch": 2.641571721888965,
-      "grad_norm": 0.0013382801553234458,
-      "learning_rate": 0.0001999965609593866,
-      "loss": 46.0,
-      "step": 16403
-    },
-    {
-      "epoch": 2.6417327589677524,
-      "grad_norm": 0.010065632872283459,
-      "learning_rate": 0.00019999656053980194,
-      "loss": 46.0,
-      "step": 16404
-    },
-    {
-      "epoch": 2.64189379604654,
-      "grad_norm": 0.0026141307316720486,
-      "learning_rate": 0.00019999656012019168,
-      "loss": 46.0,
-      "step": 16405
-    },
-    {
-      "epoch": 2.642054833125327,
-      "grad_norm": 0.0028353743255138397,
-      "learning_rate": 0.0001999965597005558,
-      "loss": 46.0,
-      "step": 16406
-    },
-    {
-      "epoch": 2.6422158702041143,
-      "grad_norm": 0.009485107846558094,
-      "learning_rate": 0.00019999655928089433,
-      "loss": 46.0,
-      "step": 16407
-    },
-    {
-      "epoch": 2.6423769072829018,
-      "grad_norm": 0.0012441881699487567,
-      "learning_rate": 0.00019999655886120729,
-      "loss": 46.0,
-      "step": 16408
-    },
-    {
-      "epoch": 2.642537944361689,
-      "grad_norm": 0.0043039084412157536,
-      "learning_rate": 0.00019999655844149463,
-      "loss": 46.0,
-      "step": 16409
-    },
-    {
-      "epoch": 2.6426989814404767,
-      "grad_norm": 0.004510168451815844,
-      "learning_rate": 0.0001999965580217564,
-      "loss": 46.0,
-      "step": 16410
-    },
-    {
-      "epoch": 2.642860018519264,
-      "grad_norm": 0.0011374372988939285,
-      "learning_rate": 0.00019999655760199257,
-      "loss": 46.0,
-      "step": 16411
-    },
-    {
-      "epoch": 2.6430210555980516,
-      "grad_norm": 0.0026707665529102087,
-      "learning_rate": 0.00019999655718220315,
-      "loss": 46.0,
-      "step": 16412
-    },
-    {
-      "epoch": 2.643182092676839,
-      "grad_norm": 0.0004974910407327116,
-      "learning_rate": 0.00019999655676238812,
-      "loss": 46.0,
-      "step": 16413
-    },
-    {
-      "epoch": 2.6433431297556265,
-      "grad_norm": 0.0034076026640832424,
-      "learning_rate": 0.0001999965563425475,
-      "loss": 46.0,
-      "step": 16414
-    },
-    {
-      "epoch": 2.6435041668344135,
-      "grad_norm": 0.0019066615495830774,
-      "learning_rate": 0.0001999965559226813,
-      "loss": 46.0,
-      "step": 16415
-    },
-    {
-      "epoch": 2.643665203913201,
-      "grad_norm": 0.005310556851327419,
-      "learning_rate": 0.0001999965555027895,
-      "loss": 46.0,
-      "step": 16416
-    },
-    {
-      "epoch": 2.6438262409919884,
-      "grad_norm": 0.003847218118607998,
-      "learning_rate": 0.00019999655508287208,
-      "loss": 46.0,
-      "step": 16417
-    },
-    {
-      "epoch": 2.643987278070776,
-      "grad_norm": 0.015537664294242859,
-      "learning_rate": 0.0001999965546629291,
-      "loss": 46.0,
-      "step": 16418
-    },
-    {
-      "epoch": 2.6441483151495633,
-      "grad_norm": 0.0014112257631495595,
-      "learning_rate": 0.00019999655424296053,
-      "loss": 46.0,
-      "step": 16419
-    },
-    {
-      "epoch": 2.6443093522283507,
-      "grad_norm": 0.0019733665976673365,
-      "learning_rate": 0.00019999655382296636,
-      "loss": 46.0,
-      "step": 16420
-    },
-    {
-      "epoch": 2.6444703893071377,
-      "grad_norm": 0.0028667664155364037,
-      "learning_rate": 0.00019999655340294657,
-      "loss": 46.0,
-      "step": 16421
-    },
-    {
-      "epoch": 2.644631426385925,
-      "grad_norm": 0.0013341999147087336,
-      "learning_rate": 0.0001999965529829012,
-      "loss": 46.0,
-      "step": 16422
-    },
-    {
-      "epoch": 2.6447924634647126,
-      "grad_norm": 0.0019776078406721354,
-      "learning_rate": 0.00019999655256283024,
-      "loss": 46.0,
-      "step": 16423
-    },
-    {
-      "epoch": 2.6449535005435,
-      "grad_norm": 0.0022960947826504707,
-      "learning_rate": 0.0001999965521427337,
-      "loss": 46.0,
-      "step": 16424
-    },
-    {
-      "epoch": 2.6451145376222875,
-      "grad_norm": 0.0014121850254014134,
-      "learning_rate": 0.00019999655172261153,
-      "loss": 46.0,
-      "step": 16425
-    },
-    {
-      "epoch": 2.645275574701075,
-      "grad_norm": 0.0008594037499278784,
-      "learning_rate": 0.0001999965513024638,
-      "loss": 46.0,
-      "step": 16426
-    },
-    {
-      "epoch": 2.6454366117798624,
-      "grad_norm": 0.0009809122420847416,
-      "learning_rate": 0.00019999655088229047,
-      "loss": 46.0,
-      "step": 16427
-    },
-    {
-      "epoch": 2.64559764885865,
-      "grad_norm": 0.0013061140198260546,
-      "learning_rate": 0.00019999655046209155,
-      "loss": 46.0,
-      "step": 16428
-    },
-    {
-      "epoch": 2.6457586859374373,
-      "grad_norm": 0.006480553187429905,
-      "learning_rate": 0.000199996550041867,
-      "loss": 46.0,
-      "step": 16429
-    },
-    {
-      "epoch": 2.6459197230162244,
-      "grad_norm": 0.0007608724990859628,
-      "learning_rate": 0.00019999654962161686,
-      "loss": 46.0,
-      "step": 16430
-    },
-    {
-      "epoch": 2.646080760095012,
-      "grad_norm": 0.0014058458618819714,
-      "learning_rate": 0.00019999654920134117,
-      "loss": 46.0,
-      "step": 16431
-    },
-    {
-      "epoch": 2.6462417971737993,
-      "grad_norm": 0.004803846590220928,
-      "learning_rate": 0.00019999654878103987,
-      "loss": 46.0,
-      "step": 16432
-    },
-    {
-      "epoch": 2.6464028342525867,
-      "grad_norm": 0.0008249848615378141,
-      "learning_rate": 0.00019999654836071296,
-      "loss": 46.0,
-      "step": 16433
-    },
-    {
-      "epoch": 2.646563871331374,
-      "grad_norm": 0.0007507161935791373,
-      "learning_rate": 0.00019999654794036046,
-      "loss": 46.0,
-      "step": 16434
-    },
-    {
-      "epoch": 2.646724908410161,
-      "grad_norm": 0.0018223902443423867,
-      "learning_rate": 0.00019999654751998237,
-      "loss": 46.0,
-      "step": 16435
-    },
-    {
-      "epoch": 2.6468859454889486,
-      "grad_norm": 0.0004306924529373646,
-      "learning_rate": 0.0001999965470995787,
-      "loss": 46.0,
-      "step": 16436
-    },
-    {
-      "epoch": 2.647046982567736,
-      "grad_norm": 0.004863760434091091,
-      "learning_rate": 0.00019999654667914943,
-      "loss": 46.0,
-      "step": 16437
-    },
-    {
-      "epoch": 2.6472080196465235,
-      "grad_norm": 0.005212968215346336,
-      "learning_rate": 0.00019999654625869453,
-      "loss": 46.0,
-      "step": 16438
-    },
-    {
-      "epoch": 2.647369056725311,
-      "grad_norm": 0.0017275518039241433,
-      "learning_rate": 0.00019999654583821407,
-      "loss": 46.0,
-      "step": 16439
-    },
-    {
-      "epoch": 2.6475300938040984,
-      "grad_norm": 0.004870988428592682,
-      "learning_rate": 0.00019999654541770801,
-      "loss": 46.0,
-      "step": 16440
-    },
-    {
-      "epoch": 2.647691130882886,
-      "grad_norm": 0.0032627710606902838,
-      "learning_rate": 0.00019999654499717638,
-      "loss": 46.0,
-      "step": 16441
-    },
-    {
-      "epoch": 2.6478521679616733,
-      "grad_norm": 0.001680032815784216,
-      "learning_rate": 0.00019999654457661912,
-      "loss": 46.0,
-      "step": 16442
-    },
-    {
-      "epoch": 2.648013205040461,
-      "grad_norm": 0.0036580024752765894,
-      "learning_rate": 0.00019999654415603629,
-      "loss": 46.0,
-      "step": 16443
-    },
-    {
-      "epoch": 2.6481742421192482,
-      "grad_norm": 0.002199007896706462,
-      "learning_rate": 0.00019999654373542783,
-      "loss": 46.0,
-      "step": 16444
-    },
-    {
-      "epoch": 2.6483352791980352,
-      "grad_norm": 0.001317737391218543,
-      "learning_rate": 0.00019999654331479382,
-      "loss": 46.0,
-      "step": 16445
-    },
-    {
-      "epoch": 2.6484963162768227,
-      "grad_norm": 0.007964441552758217,
-      "learning_rate": 0.00019999654289413422,
-      "loss": 46.0,
-      "step": 16446
-    },
-    {
-      "epoch": 2.64865735335561,
-      "grad_norm": 0.0014704192290082574,
-      "learning_rate": 0.000199996542473449,
-      "loss": 46.0,
-      "step": 16447
-    },
-    {
-      "epoch": 2.6488183904343976,
-      "grad_norm": 0.0015728285070508718,
-      "learning_rate": 0.00019999654205273817,
-      "loss": 46.0,
-      "step": 16448
-    },
-    {
-      "epoch": 2.648979427513185,
-      "grad_norm": 0.004670578520745039,
-      "learning_rate": 0.00019999654163200176,
-      "loss": 46.0,
-      "step": 16449
-    },
-    {
-      "epoch": 2.649140464591972,
-      "grad_norm": 0.0008197281276807189,
-      "learning_rate": 0.00019999654121123975,
-      "loss": 46.0,
-      "step": 16450
-    },
-    {
-      "epoch": 2.6493015016707595,
-      "grad_norm": 0.0018704469548538327,
-      "learning_rate": 0.0001999965407904522,
-      "loss": 46.0,
-      "step": 16451
-    },
-    {
-      "epoch": 2.649462538749547,
-      "grad_norm": 0.004465532023459673,
-      "learning_rate": 0.00019999654036963898,
-      "loss": 46.0,
-      "step": 16452
-    },
-    {
-      "epoch": 2.6496235758283344,
-      "grad_norm": 0.0020445717964321375,
-      "learning_rate": 0.0001999965399488002,
-      "loss": 46.0,
-      "step": 16453
-    },
-    {
-      "epoch": 2.649784612907122,
-      "grad_norm": 0.006952350027859211,
-      "learning_rate": 0.0001999965395279358,
-      "loss": 46.0,
-      "step": 16454
-    },
-    {
-      "epoch": 2.6499456499859093,
-      "grad_norm": 0.0012763794511556625,
-      "learning_rate": 0.00019999653910704584,
-      "loss": 46.0,
-      "step": 16455
-    },
-    {
-      "epoch": 2.6501066870646968,
-      "grad_norm": 0.0048883287236094475,
-      "learning_rate": 0.00019999653868613032,
-      "loss": 46.0,
-      "step": 16456
-    },
-    {
-      "epoch": 2.650267724143484,
-      "grad_norm": 0.005108126439154148,
-      "learning_rate": 0.00019999653826518915,
-      "loss": 46.0,
-      "step": 16457
-    },
-    {
-      "epoch": 2.6504287612222717,
-      "grad_norm": 0.0032781974878162146,
-      "learning_rate": 0.0001999965378442224,
-      "loss": 46.0,
-      "step": 16458
-    },
-    {
-      "epoch": 2.6505897983010587,
-      "grad_norm": 0.002586958697065711,
-      "learning_rate": 0.00019999653742323002,
-      "loss": 46.0,
-      "step": 16459
-    },
-    {
-      "epoch": 2.650750835379846,
-      "grad_norm": 0.0010242596035823226,
-      "learning_rate": 0.0001999965370022121,
-      "loss": 46.0,
-      "step": 16460
-    },
-    {
-      "epoch": 2.6509118724586336,
-      "grad_norm": 0.0031969924457371235,
-      "learning_rate": 0.00019999653658116858,
-      "loss": 46.0,
-      "step": 16461
-    },
-    {
-      "epoch": 2.651072909537421,
-      "grad_norm": 0.003769185161218047,
-      "learning_rate": 0.00019999653616009945,
-      "loss": 46.0,
-      "step": 16462
-    },
-    {
-      "epoch": 2.6512339466162085,
-      "grad_norm": 0.009780793450772762,
-      "learning_rate": 0.00019999653573900473,
-      "loss": 46.0,
-      "step": 16463
-    },
-    {
-      "epoch": 2.651394983694996,
-      "grad_norm": 0.0017458347138017416,
-      "learning_rate": 0.0001999965353178844,
-      "loss": 46.0,
-      "step": 16464
-    },
-    {
-      "epoch": 2.651556020773783,
-      "grad_norm": 0.0035560824908316135,
-      "learning_rate": 0.0001999965348967385,
-      "loss": 46.0,
-      "step": 16465
-    },
-    {
-      "epoch": 2.6517170578525704,
-      "grad_norm": 0.007063970435410738,
-      "learning_rate": 0.000199996534475567,
-      "loss": 46.0,
-      "step": 16466
-    },
-    {
-      "epoch": 2.651878094931358,
-      "grad_norm": 0.0006652420852333307,
-      "learning_rate": 0.0001999965340543699,
-      "loss": 46.0,
-      "step": 16467
-    },
-    {
-      "epoch": 2.6520391320101453,
-      "grad_norm": 0.0009053508983924985,
-      "learning_rate": 0.00019999653363314722,
-      "loss": 46.0,
-      "step": 16468
-    },
-    {
-      "epoch": 2.6522001690889327,
-      "grad_norm": 0.004392145201563835,
-      "learning_rate": 0.00019999653321189892,
-      "loss": 46.0,
-      "step": 16469
-    },
-    {
-      "epoch": 2.65236120616772,
-      "grad_norm": 0.006357663776725531,
-      "learning_rate": 0.00019999653279062504,
-      "loss": 46.0,
-      "step": 16470
-    },
-    {
-      "epoch": 2.6525222432465076,
-      "grad_norm": 0.0048867687582969666,
-      "learning_rate": 0.00019999653236932557,
-      "loss": 46.0,
-      "step": 16471
-    },
-    {
-      "epoch": 2.652683280325295,
-      "grad_norm": 0.006181687116622925,
-      "learning_rate": 0.00019999653194800051,
-      "loss": 46.0,
-      "step": 16472
-    },
-    {
-      "epoch": 2.6528443174040826,
-      "grad_norm": 0.0037122254725545645,
-      "learning_rate": 0.00019999653152664984,
-      "loss": 46.0,
-      "step": 16473
-    },
-    {
-      "epoch": 2.6530053544828696,
-      "grad_norm": 0.006071495357900858,
-      "learning_rate": 0.0001999965311052736,
-      "loss": 46.0,
-      "step": 16474
-    },
-    {
-      "epoch": 2.653166391561657,
-      "grad_norm": 0.0028920916374772787,
-      "learning_rate": 0.00019999653068387177,
-      "loss": 46.0,
-      "step": 16475
-    },
-    {
-      "epoch": 2.6533274286404445,
-      "grad_norm": 0.0030148280784487724,
-      "learning_rate": 0.0001999965302624443,
-      "loss": 46.0,
-      "step": 16476
-    },
-    {
-      "epoch": 2.653488465719232,
-      "grad_norm": 0.004583471920341253,
-      "learning_rate": 0.00019999652984099126,
-      "loss": 46.0,
-      "step": 16477
-    },
-    {
-      "epoch": 2.6536495027980194,
-      "grad_norm": 0.004693299997597933,
-      "learning_rate": 0.00019999652941951262,
-      "loss": 46.0,
-      "step": 16478
-    },
-    {
-      "epoch": 2.6538105398768064,
-      "grad_norm": 0.0014970963820815086,
-      "learning_rate": 0.0001999965289980084,
-      "loss": 46.0,
-      "step": 16479
-    },
-    {
-      "epoch": 2.653971576955594,
-      "grad_norm": 0.009805742651224136,
-      "learning_rate": 0.0001999965285764786,
-      "loss": 46.0,
-      "step": 16480
-    },
-    {
-      "epoch": 2.6541326140343813,
-      "grad_norm": 0.0033442846033722162,
-      "learning_rate": 0.00019999652815492317,
-      "loss": 46.0,
-      "step": 16481
-    },
-    {
-      "epoch": 2.6542936511131687,
-      "grad_norm": 0.0008747574174776673,
-      "learning_rate": 0.00019999652773334216,
-      "loss": 46.0,
-      "step": 16482
-    },
-    {
-      "epoch": 2.654454688191956,
-      "grad_norm": 0.0018716448685154319,
-      "learning_rate": 0.0001999965273117356,
-      "loss": 46.0,
-      "step": 16483
-    },
-    {
-      "epoch": 2.6546157252707436,
-      "grad_norm": 0.000537240644916892,
-      "learning_rate": 0.0001999965268901034,
-      "loss": 46.0,
-      "step": 16484
-    },
-    {
-      "epoch": 2.654776762349531,
-      "grad_norm": 0.0037817684933543205,
-      "learning_rate": 0.0001999965264684456,
-      "loss": 46.0,
-      "step": 16485
-    },
-    {
-      "epoch": 2.6549377994283185,
-      "grad_norm": 0.005018272902816534,
-      "learning_rate": 0.00019999652604676222,
-      "loss": 46.0,
-      "step": 16486
-    },
-    {
-      "epoch": 2.655098836507106,
-      "grad_norm": 0.0030136287678033113,
-      "learning_rate": 0.00019999652562505325,
-      "loss": 46.0,
-      "step": 16487
-    },
-    {
-      "epoch": 2.655259873585893,
-      "grad_norm": 0.0029723059851676226,
-      "learning_rate": 0.00019999652520331866,
-      "loss": 46.0,
-      "step": 16488
-    },
-    {
-      "epoch": 2.6554209106646804,
-      "grad_norm": 0.00226042652502656,
-      "learning_rate": 0.00019999652478155851,
-      "loss": 46.0,
-      "step": 16489
-    },
-    {
-      "epoch": 2.655581947743468,
-      "grad_norm": 0.0018430107738822699,
-      "learning_rate": 0.00019999652435977275,
-      "loss": 46.0,
-      "step": 16490
-    },
-    {
-      "epoch": 2.6557429848222553,
-      "grad_norm": 0.0017959438264369965,
-      "learning_rate": 0.0001999965239379614,
-      "loss": 46.0,
-      "step": 16491
-    },
-    {
-      "epoch": 2.655904021901043,
-      "grad_norm": 0.0033439442049711943,
-      "learning_rate": 0.00019999652351612444,
-      "loss": 46.0,
-      "step": 16492
-    },
-    {
-      "epoch": 2.6560650589798303,
-      "grad_norm": 0.0006606042152270675,
-      "learning_rate": 0.00019999652309426192,
-      "loss": 46.0,
-      "step": 16493
-    },
-    {
-      "epoch": 2.6562260960586173,
-      "grad_norm": 0.001083419076167047,
-      "learning_rate": 0.00019999652267237375,
-      "loss": 46.0,
-      "step": 16494
-    },
-    {
-      "epoch": 2.6563871331374047,
-      "grad_norm": 0.0024354367051273584,
-      "learning_rate": 0.00019999652225046006,
-      "loss": 46.0,
-      "step": 16495
-    },
-    {
-      "epoch": 2.656548170216192,
-      "grad_norm": 0.008430845104157925,
-      "learning_rate": 0.00019999652182852074,
-      "loss": 46.0,
-      "step": 16496
-    },
-    {
-      "epoch": 2.6567092072949796,
-      "grad_norm": 0.0030144881457090378,
-      "learning_rate": 0.0001999965214065558,
-      "loss": 46.0,
-      "step": 16497
-    },
-    {
-      "epoch": 2.656870244373767,
-      "grad_norm": 0.0049331337213516235,
-      "learning_rate": 0.0001999965209845653,
-      "loss": 46.0,
-      "step": 16498
-    },
-    {
-      "epoch": 2.6570312814525545,
-      "grad_norm": 0.0020491608884185553,
-      "learning_rate": 0.00019999652056254918,
-      "loss": 46.0,
-      "step": 16499
-    },
-    {
-      "epoch": 2.657192318531342,
-      "grad_norm": 0.006340573076158762,
-      "learning_rate": 0.0001999965201405075,
-      "loss": 46.0,
-      "step": 16500
-    },
-    {
-      "epoch": 2.6573533556101294,
-      "grad_norm": 0.014400620944797993,
-      "learning_rate": 0.00019999651971844021,
-      "loss": 46.0,
-      "step": 16501
-    },
-    {
-      "epoch": 2.657514392688917,
-      "grad_norm": 0.0008878426742739975,
-      "learning_rate": 0.00019999651929634732,
-      "loss": 46.0,
-      "step": 16502
-    },
-    {
-      "epoch": 2.657675429767704,
-      "grad_norm": 0.0006870725192129612,
-      "learning_rate": 0.00019999651887422882,
-      "loss": 46.0,
-      "step": 16503
-    },
-    {
-      "epoch": 2.6578364668464913,
-      "grad_norm": 0.011201038025319576,
-      "learning_rate": 0.00019999651845208476,
-      "loss": 46.0,
-      "step": 16504
-    },
-    {
-      "epoch": 2.657997503925279,
-      "grad_norm": 0.0021067147608846426,
-      "learning_rate": 0.00019999651802991508,
-      "loss": 46.0,
-      "step": 16505
-    },
-    {
-      "epoch": 2.6581585410040662,
-      "grad_norm": 0.002912551397457719,
-      "learning_rate": 0.00019999651760771982,
-      "loss": 46.0,
-      "step": 16506
-    },
-    {
-      "epoch": 2.6583195780828537,
-      "grad_norm": 0.002470936393365264,
-      "learning_rate": 0.000199996517185499,
-      "loss": 46.0,
-      "step": 16507
-    },
-    {
-      "epoch": 2.6584806151616407,
-      "grad_norm": 0.0007000712212175131,
-      "learning_rate": 0.00019999651676325252,
-      "loss": 46.0,
-      "step": 16508
-    },
-    {
-      "epoch": 2.658641652240428,
-      "grad_norm": 0.001906967838294804,
-      "learning_rate": 0.0001999965163409805,
-      "loss": 46.0,
-      "step": 16509
-    },
-    {
-      "epoch": 2.6588026893192156,
-      "grad_norm": 0.010152967646718025,
-      "learning_rate": 0.00019999651591868286,
-      "loss": 46.0,
-      "step": 16510
-    },
-    {
-      "epoch": 2.658963726398003,
-      "grad_norm": 0.004215776454657316,
-      "learning_rate": 0.00019999651549635963,
-      "loss": 46.0,
-      "step": 16511
-    },
-    {
-      "epoch": 2.6591247634767905,
-      "grad_norm": 0.0021116817370057106,
-      "learning_rate": 0.00019999651507401081,
-      "loss": 46.0,
-      "step": 16512
-    },
-    {
-      "epoch": 2.659285800555578,
-      "grad_norm": 0.005937662906944752,
-      "learning_rate": 0.00019999651465163636,
-      "loss": 46.0,
-      "step": 16513
-    },
-    {
-      "epoch": 2.6594468376343654,
-      "grad_norm": 0.0008011190802790225,
-      "learning_rate": 0.00019999651422923634,
-      "loss": 46.0,
-      "step": 16514
-    },
-    {
-      "epoch": 2.659607874713153,
-      "grad_norm": 0.0013191973557695746,
-      "learning_rate": 0.00019999651380681074,
-      "loss": 46.0,
-      "step": 16515
-    },
-    {
-      "epoch": 2.6597689117919403,
-      "grad_norm": 0.003918023779988289,
-      "learning_rate": 0.00019999651338435955,
-      "loss": 46.0,
-      "step": 16516
-    },
-    {
-      "epoch": 2.6599299488707278,
-      "grad_norm": 0.004984579514712095,
-      "learning_rate": 0.00019999651296188274,
-      "loss": 46.0,
-      "step": 16517
-    },
-    {
-      "epoch": 2.6600909859495148,
-      "grad_norm": 0.0013943187659606338,
-      "learning_rate": 0.00019999651253938038,
-      "loss": 46.0,
-      "step": 16518
-    },
-    {
-      "epoch": 2.660252023028302,
-      "grad_norm": 0.010174227878451347,
-      "learning_rate": 0.00019999651211685237,
-      "loss": 46.0,
-      "step": 16519
-    },
-    {
-      "epoch": 2.6604130601070897,
-      "grad_norm": 0.0063035693019628525,
-      "learning_rate": 0.0001999965116942988,
-      "loss": 46.0,
-      "step": 16520
-    },
-    {
-      "epoch": 2.660574097185877,
-      "grad_norm": 0.0035327086225152016,
-      "learning_rate": 0.00019999651127171962,
-      "loss": 46.0,
-      "step": 16521
-    },
-    {
-      "epoch": 2.6607351342646646,
-      "grad_norm": 0.0023248244542628527,
-      "learning_rate": 0.00019999651084911486,
-      "loss": 46.0,
-      "step": 16522
-    },
-    {
-      "epoch": 2.6608961713434516,
-      "grad_norm": 0.000654800096526742,
-      "learning_rate": 0.0001999965104264845,
-      "loss": 46.0,
-      "step": 16523
-    },
-    {
-      "epoch": 2.661057208422239,
-      "grad_norm": 0.0014816172188147902,
-      "learning_rate": 0.00019999651000382856,
-      "loss": 46.0,
-      "step": 16524
-    },
-    {
-      "epoch": 2.6612182455010265,
-      "grad_norm": 0.00244570872746408,
-      "learning_rate": 0.000199996509581147,
-      "loss": 46.0,
-      "step": 16525
-    },
-    {
-      "epoch": 2.661379282579814,
-      "grad_norm": 0.0023470923770219088,
-      "learning_rate": 0.00019999650915843986,
-      "loss": 46.0,
-      "step": 16526
-    },
-    {
-      "epoch": 2.6615403196586014,
-      "grad_norm": 0.004165975842624903,
-      "learning_rate": 0.00019999650873570715,
-      "loss": 46.0,
-      "step": 16527
-    },
-    {
-      "epoch": 2.661701356737389,
-      "grad_norm": 0.017686238512396812,
-      "learning_rate": 0.0001999965083129488,
-      "loss": 46.0,
-      "step": 16528
-    },
-    {
-      "epoch": 2.6618623938161763,
-      "grad_norm": 0.003436204744502902,
-      "learning_rate": 0.00019999650789016488,
-      "loss": 46.0,
-      "step": 16529
-    },
-    {
-      "epoch": 2.6620234308949637,
-      "grad_norm": 0.0025715362280607224,
-      "learning_rate": 0.00019999650746735536,
-      "loss": 46.0,
-      "step": 16530
-    },
-    {
-      "epoch": 2.662184467973751,
-      "grad_norm": 0.0018459368729963899,
-      "learning_rate": 0.00019999650704452028,
-      "loss": 46.0,
-      "step": 16531
-    },
-    {
-      "epoch": 2.662345505052538,
-      "grad_norm": 0.0055804126895964146,
-      "learning_rate": 0.00019999650662165953,
-      "loss": 46.0,
-      "step": 16532
-    },
-    {
-      "epoch": 2.6625065421313256,
-      "grad_norm": 0.0012605715310201049,
-      "learning_rate": 0.00019999650619877325,
-      "loss": 46.0,
-      "step": 16533
-    },
-    {
-      "epoch": 2.662667579210113,
-      "grad_norm": 0.009583881124854088,
-      "learning_rate": 0.00019999650577586135,
-      "loss": 46.0,
-      "step": 16534
-    },
-    {
-      "epoch": 2.6628286162889006,
-      "grad_norm": 0.0037427470088005066,
-      "learning_rate": 0.00019999650535292387,
-      "loss": 46.0,
-      "step": 16535
-    },
-    {
-      "epoch": 2.662989653367688,
-      "grad_norm": 0.004453995730727911,
-      "learning_rate": 0.00019999650492996077,
-      "loss": 46.0,
-      "step": 16536
-    },
-    {
-      "epoch": 2.6631506904464755,
-      "grad_norm": 0.0017059422098100185,
-      "learning_rate": 0.0001999965045069721,
-      "loss": 46.0,
-      "step": 16537
-    },
-    {
-      "epoch": 2.6633117275252625,
-      "grad_norm": 0.0028158461209386587,
-      "learning_rate": 0.00019999650408395784,
-      "loss": 46.0,
-      "step": 16538
-    },
-    {
-      "epoch": 2.66347276460405,
-      "grad_norm": 0.004828350618481636,
-      "learning_rate": 0.00019999650366091799,
-      "loss": 46.0,
-      "step": 16539
-    },
-    {
-      "epoch": 2.6636338016828374,
-      "grad_norm": 0.0024651936255395412,
-      "learning_rate": 0.00019999650323785254,
-      "loss": 46.0,
-      "step": 16540
-    },
-    {
-      "epoch": 2.663794838761625,
-      "grad_norm": 0.004665489308536053,
-      "learning_rate": 0.00019999650281476145,
-      "loss": 46.0,
-      "step": 16541
-    },
-    {
-      "epoch": 2.6639558758404123,
-      "grad_norm": 0.0008253231062553823,
-      "learning_rate": 0.00019999650239164483,
-      "loss": 46.0,
-      "step": 16542
-    },
-    {
-      "epoch": 2.6641169129191997,
-      "grad_norm": 0.0009690801962278783,
-      "learning_rate": 0.00019999650196850257,
-      "loss": 46.0,
-      "step": 16543
-    },
-    {
-      "epoch": 2.664277949997987,
-      "grad_norm": 0.0010703481966629624,
-      "learning_rate": 0.00019999650154533475,
-      "loss": 46.0,
-      "step": 16544
-    },
-    {
-      "epoch": 2.6644389870767746,
-      "grad_norm": 0.006409147754311562,
-      "learning_rate": 0.00019999650112214132,
-      "loss": 46.0,
-      "step": 16545
-    },
-    {
-      "epoch": 2.664600024155562,
-      "grad_norm": 0.0008583225426264107,
-      "learning_rate": 0.00019999650069892232,
-      "loss": 46.0,
-      "step": 16546
-    },
-    {
-      "epoch": 2.664761061234349,
-      "grad_norm": 0.005577814765274525,
-      "learning_rate": 0.00019999650027567768,
-      "loss": 46.0,
-      "step": 16547
-    },
-    {
-      "epoch": 2.6649220983131365,
-      "grad_norm": 0.002359146950766444,
-      "learning_rate": 0.00019999649985240749,
-      "loss": 46.0,
-      "step": 16548
-    },
-    {
-      "epoch": 2.665083135391924,
-      "grad_norm": 0.0012180627090856433,
-      "learning_rate": 0.0001999964994291117,
-      "loss": 46.0,
-      "step": 16549
-    },
-    {
-      "epoch": 2.6652441724707114,
-      "grad_norm": 0.0014827927807345986,
-      "learning_rate": 0.00019999649900579028,
-      "loss": 46.0,
-      "step": 16550
-    },
-    {
-      "epoch": 2.665405209549499,
-      "grad_norm": 0.006505758035928011,
-      "learning_rate": 0.0001999964985824433,
-      "loss": 46.0,
-      "step": 16551
-    },
-    {
-      "epoch": 2.665566246628286,
-      "grad_norm": 0.0057903132401406765,
-      "learning_rate": 0.00019999649815907072,
-      "loss": 46.0,
-      "step": 16552
-    },
-    {
-      "epoch": 2.6657272837070733,
-      "grad_norm": 0.006674161646515131,
-      "learning_rate": 0.00019999649773567253,
-      "loss": 46.0,
-      "step": 16553
-    },
-    {
-      "epoch": 2.665888320785861,
-      "grad_norm": 0.009513991884887218,
-      "learning_rate": 0.00019999649731224876,
-      "loss": 46.0,
-      "step": 16554
-    },
-    {
-      "epoch": 2.6660493578646483,
-      "grad_norm": 0.008672289550304413,
-      "learning_rate": 0.00019999649688879937,
-      "loss": 46.0,
-      "step": 16555
-    },
-    {
-      "epoch": 2.6662103949434357,
-      "grad_norm": 0.0022527240216732025,
-      "learning_rate": 0.00019999649646532445,
-      "loss": 46.0,
-      "step": 16556
-    },
-    {
-      "epoch": 2.666371432022223,
-      "grad_norm": 0.005996780004352331,
-      "learning_rate": 0.00019999649604182386,
-      "loss": 46.0,
-      "step": 16557
-    },
-    {
-      "epoch": 2.6665324691010106,
-      "grad_norm": 0.001205631298944354,
-      "learning_rate": 0.00019999649561829773,
-      "loss": 46.0,
-      "step": 16558
-    },
-    {
-      "epoch": 2.666693506179798,
-      "grad_norm": 0.002092850860208273,
-      "learning_rate": 0.000199996495194746,
-      "loss": 46.0,
-      "step": 16559
-    },
-    {
-      "epoch": 2.6668545432585855,
-      "grad_norm": 0.002493825741112232,
-      "learning_rate": 0.00019999649477116864,
-      "loss": 46.0,
-      "step": 16560
-    },
-    {
-      "epoch": 2.6670155803373725,
-      "grad_norm": 0.0016343793831765652,
-      "learning_rate": 0.0001999964943475657,
-      "loss": 46.0,
-      "step": 16561
-    },
-    {
-      "epoch": 2.66717661741616,
-      "grad_norm": 0.001397967105731368,
-      "learning_rate": 0.0001999964939239372,
-      "loss": 46.0,
-      "step": 16562
-    },
-    {
-      "epoch": 2.6673376544949474,
-      "grad_norm": 0.003354318905621767,
-      "learning_rate": 0.00019999649350028307,
-      "loss": 46.0,
-      "step": 16563
-    },
-    {
-      "epoch": 2.667498691573735,
-      "grad_norm": 0.0009036913397721946,
-      "learning_rate": 0.00019999649307660337,
-      "loss": 46.0,
-      "step": 16564
-    },
-    {
-      "epoch": 2.6676597286525223,
-      "grad_norm": 0.0015470064245164394,
-      "learning_rate": 0.00019999649265289805,
-      "loss": 46.0,
-      "step": 16565
-    },
-    {
-      "epoch": 2.6678207657313098,
-      "grad_norm": 0.005684650503098965,
-      "learning_rate": 0.00019999649222916715,
-      "loss": 46.0,
-      "step": 16566
-    },
-    {
-      "epoch": 2.667981802810097,
-      "grad_norm": 0.0037616060581058264,
-      "learning_rate": 0.00019999649180541063,
-      "loss": 46.0,
-      "step": 16567
-    },
-    {
-      "epoch": 2.6681428398888842,
-      "grad_norm": 0.005537806544452906,
-      "learning_rate": 0.00019999649138162856,
-      "loss": 46.0,
-      "step": 16568
-    },
-    {
-      "epoch": 2.6683038769676717,
-      "grad_norm": 0.000822942063678056,
-      "learning_rate": 0.0001999964909578209,
-      "loss": 46.0,
-      "step": 16569
-    },
-    {
-      "epoch": 2.668464914046459,
-      "grad_norm": 0.004411452449858189,
-      "learning_rate": 0.00019999649053398762,
-      "loss": 46.0,
-      "step": 16570
-    },
-    {
-      "epoch": 2.6686259511252466,
-      "grad_norm": 0.005157234612852335,
-      "learning_rate": 0.00019999649011012872,
-      "loss": 46.0,
-      "step": 16571
-    },
-    {
-      "epoch": 2.668786988204034,
-      "grad_norm": 0.0033159011509269476,
-      "learning_rate": 0.0001999964896862443,
-      "loss": 46.0,
-      "step": 16572
-    },
-    {
-      "epoch": 2.6689480252828215,
-      "grad_norm": 0.009530625306069851,
-      "learning_rate": 0.00019999648926233423,
-      "loss": 46.0,
-      "step": 16573
-    },
-    {
-      "epoch": 2.669109062361609,
-      "grad_norm": 0.0070307631976902485,
-      "learning_rate": 0.00019999648883839855,
-      "loss": 46.0,
-      "step": 16574
-    },
-    {
-      "epoch": 2.6692700994403964,
-      "grad_norm": 0.0067693330347537994,
-      "learning_rate": 0.0001999964884144373,
-      "loss": 46.0,
-      "step": 16575
-    },
-    {
-      "epoch": 2.6694311365191834,
-      "grad_norm": 0.008922696113586426,
-      "learning_rate": 0.00019999648799045049,
-      "loss": 46.0,
-      "step": 16576
-    },
-    {
-      "epoch": 2.669592173597971,
-      "grad_norm": 0.003183950437232852,
-      "learning_rate": 0.00019999648756643804,
-      "loss": 46.0,
-      "step": 16577
-    },
-    {
-      "epoch": 2.6697532106767583,
-      "grad_norm": 0.008305241353809834,
-      "learning_rate": 0.00019999648714240001,
-      "loss": 46.0,
-      "step": 16578
-    },
-    {
-      "epoch": 2.6699142477555458,
-      "grad_norm": 0.0006374673102982342,
-      "learning_rate": 0.00019999648671833643,
-      "loss": 46.0,
-      "step": 16579
-    },
-    {
-      "epoch": 2.670075284834333,
-      "grad_norm": 0.003601290052756667,
-      "learning_rate": 0.0001999964862942472,
-      "loss": 46.0,
-      "step": 16580
-    },
-    {
-      "epoch": 2.6702363219131207,
-      "grad_norm": 0.002714934991672635,
-      "learning_rate": 0.00019999648587013238,
-      "loss": 46.0,
-      "step": 16581
-    },
-    {
-      "epoch": 2.6703973589919077,
-      "grad_norm": 0.00787340011447668,
-      "learning_rate": 0.00019999648544599197,
-      "loss": 46.0,
-      "step": 16582
-    },
-    {
-      "epoch": 2.670558396070695,
-      "grad_norm": 0.003337181406095624,
-      "learning_rate": 0.00019999648502182598,
-      "loss": 46.0,
-      "step": 16583
-    },
-    {
-      "epoch": 2.6707194331494826,
-      "grad_norm": 0.0009643948287703097,
-      "learning_rate": 0.00019999648459763437,
-      "loss": 46.0,
-      "step": 16584
-    },
-    {
-      "epoch": 2.67088047022827,
-      "grad_norm": 0.0015101605094969273,
-      "learning_rate": 0.0001999964841734172,
-      "loss": 46.0,
-      "step": 16585
-    },
-    {
-      "epoch": 2.6710415073070575,
-      "grad_norm": 0.002747853985056281,
-      "learning_rate": 0.0001999964837491744,
-      "loss": 46.0,
-      "step": 16586
-    },
-    {
-      "epoch": 2.671202544385845,
-      "grad_norm": 0.005050466395914555,
-      "learning_rate": 0.00019999648332490603,
-      "loss": 46.0,
-      "step": 16587
-    },
-    {
-      "epoch": 2.6713635814646324,
-      "grad_norm": 0.000775398570112884,
-      "learning_rate": 0.00019999648290061208,
-      "loss": 46.0,
-      "step": 16588
-    },
-    {
-      "epoch": 2.67152461854342,
-      "grad_norm": 0.003925311379134655,
-      "learning_rate": 0.0001999964824762925,
-      "loss": 46.0,
-      "step": 16589
-    },
-    {
-      "epoch": 2.6716856556222073,
-      "grad_norm": 0.0026923860423266888,
-      "learning_rate": 0.00019999648205194738,
-      "loss": 46.0,
-      "step": 16590
-    },
-    {
-      "epoch": 2.6718466927009943,
-      "grad_norm": 0.0015187134267762303,
-      "learning_rate": 0.0001999964816275766,
-      "loss": 46.0,
-      "step": 16591
-    },
-    {
-      "epoch": 2.6720077297797817,
-      "grad_norm": 0.0023581660352647305,
-      "learning_rate": 0.00019999648120318028,
-      "loss": 46.0,
-      "step": 16592
-    },
-    {
-      "epoch": 2.672168766858569,
-      "grad_norm": 0.002250324934720993,
-      "learning_rate": 0.00019999648077875833,
-      "loss": 46.0,
-      "step": 16593
-    },
-    {
-      "epoch": 2.6723298039373566,
-      "grad_norm": 0.002318785758689046,
-      "learning_rate": 0.0001999964803543108,
-      "loss": 46.0,
-      "step": 16594
-    },
-    {
-      "epoch": 2.672490841016144,
-      "grad_norm": 0.0013619036180898547,
-      "learning_rate": 0.00019999647992983768,
-      "loss": 46.0,
-      "step": 16595
-    },
-    {
-      "epoch": 2.672651878094931,
-      "grad_norm": 0.0015416962560266256,
-      "learning_rate": 0.00019999647950533895,
-      "loss": 46.0,
-      "step": 16596
-    },
-    {
-      "epoch": 2.6728129151737186,
-      "grad_norm": 0.0010195387294515967,
-      "learning_rate": 0.00019999647908081465,
-      "loss": 46.0,
-      "step": 16597
-    },
-    {
-      "epoch": 2.672973952252506,
-      "grad_norm": 0.00118343741632998,
-      "learning_rate": 0.00019999647865626474,
-      "loss": 46.0,
-      "step": 16598
-    },
-    {
-      "epoch": 2.6731349893312935,
-      "grad_norm": 0.001540732104331255,
-      "learning_rate": 0.00019999647823168922,
-      "loss": 46.0,
-      "step": 16599
-    },
-    {
-      "epoch": 2.673296026410081,
-      "grad_norm": 0.0022478620521724224,
-      "learning_rate": 0.00019999647780708814,
-      "loss": 46.0,
-      "step": 16600
-    },
-    {
-      "epoch": 2.6734570634888684,
-      "grad_norm": 0.0011191079393029213,
-      "learning_rate": 0.00019999647738246144,
-      "loss": 46.0,
-      "step": 16601
-    },
-    {
-      "epoch": 2.673618100567656,
-      "grad_norm": 0.0008199792937375605,
-      "learning_rate": 0.00019999647695780916,
-      "loss": 46.0,
-      "step": 16602
-    },
-    {
-      "epoch": 2.6737791376464433,
-      "grad_norm": 0.0017260703025385737,
-      "learning_rate": 0.0001999964765331313,
-      "loss": 46.0,
-      "step": 16603
-    },
-    {
-      "epoch": 2.6739401747252307,
-      "grad_norm": 0.007604490499943495,
-      "learning_rate": 0.00019999647610842783,
-      "loss": 46.0,
-      "step": 16604
-    },
-    {
-      "epoch": 2.6741012118040177,
-      "grad_norm": 0.0042333463206887245,
-      "learning_rate": 0.00019999647568369876,
-      "loss": 46.0,
-      "step": 16605
-    },
-    {
-      "epoch": 2.674262248882805,
-      "grad_norm": 0.002328177448362112,
-      "learning_rate": 0.0001999964752589441,
-      "loss": 46.0,
-      "step": 16606
-    },
-    {
-      "epoch": 2.6744232859615926,
-      "grad_norm": 0.0010306008625775576,
-      "learning_rate": 0.00019999647483416385,
-      "loss": 46.0,
-      "step": 16607
-    },
-    {
-      "epoch": 2.67458432304038,
-      "grad_norm": 0.0008662262698635459,
-      "learning_rate": 0.000199996474409358,
-      "loss": 46.0,
-      "step": 16608
-    },
-    {
-      "epoch": 2.6747453601191675,
-      "grad_norm": 0.0031616119667887688,
-      "learning_rate": 0.00019999647398452654,
-      "loss": 46.0,
-      "step": 16609
-    },
-    {
-      "epoch": 2.674906397197955,
-      "grad_norm": 0.012205038219690323,
-      "learning_rate": 0.0001999964735596695,
-      "loss": 46.0,
-      "step": 16610
-    },
-    {
-      "epoch": 2.675067434276742,
-      "grad_norm": 0.0018624645890668035,
-      "learning_rate": 0.00019999647313478688,
-      "loss": 46.0,
-      "step": 16611
-    },
-    {
-      "epoch": 2.6752284713555294,
-      "grad_norm": 0.001965672941878438,
-      "learning_rate": 0.00019999647270987867,
-      "loss": 46.0,
-      "step": 16612
-    },
-    {
-      "epoch": 2.675389508434317,
-      "grad_norm": 0.0015672518638893962,
-      "learning_rate": 0.00019999647228494485,
-      "loss": 46.0,
-      "step": 16613
-    },
-    {
-      "epoch": 2.6755505455131043,
-      "grad_norm": 0.003995127975940704,
-      "learning_rate": 0.0001999964718599854,
-      "loss": 46.0,
-      "step": 16614
-    },
-    {
-      "epoch": 2.675711582591892,
-      "grad_norm": 0.002233319915831089,
-      "learning_rate": 0.00019999647143500044,
-      "loss": 46.0,
-      "step": 16615
-    },
-    {
-      "epoch": 2.6758726196706792,
-      "grad_norm": 0.0017141100252047181,
-      "learning_rate": 0.00019999647100998983,
-      "loss": 46.0,
-      "step": 16616
-    },
-    {
-      "epoch": 2.6760336567494667,
-      "grad_norm": 0.002074493793770671,
-      "learning_rate": 0.00019999647058495366,
-      "loss": 46.0,
-      "step": 16617
-    },
-    {
-      "epoch": 2.676194693828254,
-      "grad_norm": 0.002965449821203947,
-      "learning_rate": 0.00019999647015989184,
-      "loss": 46.0,
-      "step": 16618
-    },
-    {
-      "epoch": 2.6763557309070416,
-      "grad_norm": 0.004293539095669985,
-      "learning_rate": 0.00019999646973480447,
-      "loss": 46.0,
-      "step": 16619
-    },
-    {
-      "epoch": 2.6765167679858286,
-      "grad_norm": 0.0017037526704370975,
-      "learning_rate": 0.0001999964693096915,
-      "loss": 46.0,
-      "step": 16620
-    },
-    {
-      "epoch": 2.676677805064616,
-      "grad_norm": 0.0015041976002976298,
-      "learning_rate": 0.00019999646888455296,
-      "loss": 46.0,
-      "step": 16621
-    },
-    {
-      "epoch": 2.6768388421434035,
-      "grad_norm": 0.004690492991358042,
-      "learning_rate": 0.00019999646845938877,
-      "loss": 46.0,
-      "step": 16622
-    },
-    {
-      "epoch": 2.676999879222191,
-      "grad_norm": 0.0028543814551085234,
-      "learning_rate": 0.00019999646803419902,
-      "loss": 46.0,
-      "step": 16623
-    },
-    {
-      "epoch": 2.6771609163009784,
-      "grad_norm": 0.00493650184944272,
-      "learning_rate": 0.00019999646760898365,
-      "loss": 46.0,
-      "step": 16624
-    },
-    {
-      "epoch": 2.6773219533797654,
-      "grad_norm": 0.019306128844618797,
-      "learning_rate": 0.00019999646718374273,
-      "loss": 46.0,
-      "step": 16625
-    },
-    {
-      "epoch": 2.677482990458553,
-      "grad_norm": 0.00233146152459085,
-      "learning_rate": 0.00019999646675847616,
-      "loss": 46.0,
-      "step": 16626
-    },
-    {
-      "epoch": 2.6776440275373403,
-      "grad_norm": 0.0014080959372222424,
-      "learning_rate": 0.00019999646633318406,
-      "loss": 46.0,
-      "step": 16627
-    },
-    {
-      "epoch": 2.6778050646161278,
-      "grad_norm": 0.0031507282983511686,
-      "learning_rate": 0.00019999646590786632,
-      "loss": 46.0,
-      "step": 16628
-    },
-    {
-      "epoch": 2.6779661016949152,
-      "grad_norm": 0.0016994421603158116,
-      "learning_rate": 0.000199996465482523,
-      "loss": 46.0,
-      "step": 16629
-    },
-    {
-      "epoch": 2.6781271387737027,
-      "grad_norm": 0.0038471859879791737,
-      "learning_rate": 0.0001999964650571541,
-      "loss": 46.0,
-      "step": 16630
-    },
-    {
-      "epoch": 2.67828817585249,
-      "grad_norm": 0.0016469808761030436,
-      "learning_rate": 0.00019999646463175958,
-      "loss": 46.0,
-      "step": 16631
-    },
-    {
-      "epoch": 2.6784492129312776,
-      "grad_norm": 0.0036466640885919333,
-      "learning_rate": 0.0001999964642063395,
-      "loss": 46.0,
-      "step": 16632
-    },
-    {
-      "epoch": 2.678610250010065,
-      "grad_norm": 0.00043725993600673974,
-      "learning_rate": 0.0001999964637808938,
-      "loss": 46.0,
-      "step": 16633
-    },
-    {
-      "epoch": 2.6787712870888525,
-      "grad_norm": 0.0007817916921339929,
-      "learning_rate": 0.0001999964633554225,
-      "loss": 46.0,
-      "step": 16634
-    },
-    {
-      "epoch": 2.6789323241676395,
-      "grad_norm": 0.0016522860387340188,
-      "learning_rate": 0.00019999646292992562,
-      "loss": 46.0,
-      "step": 16635
-    },
-    {
-      "epoch": 2.679093361246427,
-      "grad_norm": 0.0026875450275838375,
-      "learning_rate": 0.0001999964625044031,
-      "loss": 46.0,
-      "step": 16636
-    },
-    {
-      "epoch": 2.6792543983252144,
-      "grad_norm": 0.0026240721344947815,
-      "learning_rate": 0.00019999646207885508,
-      "loss": 46.0,
-      "step": 16637
-    },
-    {
-      "epoch": 2.679415435404002,
-      "grad_norm": 0.002033310243859887,
-      "learning_rate": 0.00019999646165328139,
-      "loss": 46.0,
-      "step": 16638
-    },
-    {
-      "epoch": 2.6795764724827893,
-      "grad_norm": 0.006061685737222433,
-      "learning_rate": 0.0001999964612276821,
-      "loss": 46.0,
-      "step": 16639
-    },
-    {
-      "epoch": 2.6797375095615763,
-      "grad_norm": 0.0009897294221445918,
-      "learning_rate": 0.00019999646080205727,
-      "loss": 46.0,
-      "step": 16640
-    },
-    {
-      "epoch": 2.6798985466403638,
-      "grad_norm": 0.011613779701292515,
-      "learning_rate": 0.00019999646037640684,
-      "loss": 46.0,
-      "step": 16641
-    },
-    {
-      "epoch": 2.680059583719151,
-      "grad_norm": 0.0005371918086893857,
-      "learning_rate": 0.00019999645995073077,
-      "loss": 46.0,
-      "step": 16642
-    },
-    {
-      "epoch": 2.6802206207979387,
-      "grad_norm": 0.004039292223751545,
-      "learning_rate": 0.00019999645952502917,
-      "loss": 46.0,
-      "step": 16643
-    },
-    {
-      "epoch": 2.680381657876726,
-      "grad_norm": 0.008642755448818207,
-      "learning_rate": 0.00019999645909930192,
-      "loss": 46.0,
-      "step": 16644
-    },
-    {
-      "epoch": 2.6805426949555136,
-      "grad_norm": 0.0037548034451901913,
-      "learning_rate": 0.0001999964586735491,
-      "loss": 46.0,
-      "step": 16645
-    },
-    {
-      "epoch": 2.680703732034301,
-      "grad_norm": 0.0025954842567443848,
-      "learning_rate": 0.0001999964582477707,
-      "loss": 46.0,
-      "step": 16646
-    },
-    {
-      "epoch": 2.6808647691130885,
-      "grad_norm": 0.0008959742845036089,
-      "learning_rate": 0.00019999645782196667,
-      "loss": 46.0,
-      "step": 16647
-    },
-    {
-      "epoch": 2.681025806191876,
-      "grad_norm": 0.003832322545349598,
-      "learning_rate": 0.00019999645739613705,
-      "loss": 46.0,
-      "step": 16648
-    },
-    {
-      "epoch": 2.681186843270663,
-      "grad_norm": 0.000573773228097707,
-      "learning_rate": 0.00019999645697028187,
-      "loss": 46.0,
-      "step": 16649
-    },
-    {
-      "epoch": 2.6813478803494504,
-      "grad_norm": 0.001437184982933104,
-      "learning_rate": 0.00019999645654440108,
-      "loss": 46.0,
-      "step": 16650
-    },
-    {
-      "epoch": 2.681508917428238,
-      "grad_norm": 0.002907233312726021,
-      "learning_rate": 0.0001999964561184947,
-      "loss": 46.0,
-      "step": 16651
-    },
-    {
-      "epoch": 2.6816699545070253,
-      "grad_norm": 0.0009279531077481806,
-      "learning_rate": 0.00019999645569256268,
-      "loss": 46.0,
-      "step": 16652
-    },
-    {
-      "epoch": 2.6818309915858127,
-      "grad_norm": 0.0007110547740012407,
-      "learning_rate": 0.00019999645526660512,
-      "loss": 46.0,
-      "step": 16653
-    },
-    {
-      "epoch": 2.6819920286646,
-      "grad_norm": 0.0028687051963061094,
-      "learning_rate": 0.00019999645484062195,
-      "loss": 46.0,
-      "step": 16654
-    },
-    {
-      "epoch": 2.682153065743387,
-      "grad_norm": 0.0021604718640446663,
-      "learning_rate": 0.0001999964544146132,
-      "loss": 46.0,
-      "step": 16655
-    },
-    {
-      "epoch": 2.6823141028221746,
-      "grad_norm": 0.0013892232673242688,
-      "learning_rate": 0.00019999645398857885,
-      "loss": 46.0,
-      "step": 16656
-    },
-    {
-      "epoch": 2.682475139900962,
-      "grad_norm": 0.0012530168751254678,
-      "learning_rate": 0.0001999964535625189,
-      "loss": 46.0,
-      "step": 16657
-    },
-    {
-      "epoch": 2.6826361769797495,
-      "grad_norm": 0.0016381407622247934,
-      "learning_rate": 0.00019999645313643332,
-      "loss": 46.0,
-      "step": 16658
-    },
-    {
-      "epoch": 2.682797214058537,
-      "grad_norm": 0.0024876834359019995,
-      "learning_rate": 0.0001999964527103222,
-      "loss": 46.0,
-      "step": 16659
-    },
-    {
-      "epoch": 2.6829582511373244,
-      "grad_norm": 0.0011124046286568046,
-      "learning_rate": 0.00019999645228418547,
-      "loss": 46.0,
-      "step": 16660
-    },
-    {
-      "epoch": 2.683119288216112,
-      "grad_norm": 0.002676828298717737,
-      "learning_rate": 0.00019999645185802317,
-      "loss": 46.0,
-      "step": 16661
-    },
-    {
-      "epoch": 2.6832803252948993,
-      "grad_norm": 0.0034460797905921936,
-      "learning_rate": 0.00019999645143183522,
-      "loss": 46.0,
-      "step": 16662
-    },
-    {
-      "epoch": 2.683441362373687,
-      "grad_norm": 0.002822249662131071,
-      "learning_rate": 0.0001999964510056217,
-      "loss": 46.0,
-      "step": 16663
-    },
-    {
-      "epoch": 2.683602399452474,
-      "grad_norm": 0.0006364263244904578,
-      "learning_rate": 0.0001999964505793826,
-      "loss": 46.0,
-      "step": 16664
-    },
-    {
-      "epoch": 2.6837634365312613,
-      "grad_norm": 0.0015507631469517946,
-      "learning_rate": 0.0001999964501531179,
-      "loss": 46.0,
-      "step": 16665
-    },
-    {
-      "epoch": 2.6839244736100487,
-      "grad_norm": 0.0025115031749010086,
-      "learning_rate": 0.00019999644972682758,
-      "loss": 46.0,
-      "step": 16666
-    },
-    {
-      "epoch": 2.684085510688836,
-      "grad_norm": 0.005299859680235386,
-      "learning_rate": 0.0001999964493005117,
-      "loss": 46.0,
-      "step": 16667
-    },
-    {
-      "epoch": 2.6842465477676236,
-      "grad_norm": 0.0016116708284243941,
-      "learning_rate": 0.00019999644887417023,
-      "loss": 46.0,
-      "step": 16668
-    },
-    {
-      "epoch": 2.6844075848464106,
-      "grad_norm": 0.0013413728447631001,
-      "learning_rate": 0.00019999644844780314,
-      "loss": 46.0,
-      "step": 16669
-    },
-    {
-      "epoch": 2.684568621925198,
-      "grad_norm": 0.0014985156012699008,
-      "learning_rate": 0.0001999964480214105,
-      "loss": 46.0,
-      "step": 16670
-    },
-    {
-      "epoch": 2.6847296590039855,
-      "grad_norm": 0.0019531450234353542,
-      "learning_rate": 0.0001999964475949922,
-      "loss": 46.0,
-      "step": 16671
-    },
-    {
-      "epoch": 2.684890696082773,
-      "grad_norm": 0.0011201798915863037,
-      "learning_rate": 0.00019999644716854834,
-      "loss": 46.0,
-      "step": 16672
-    },
-    {
-      "epoch": 2.6850517331615604,
-      "grad_norm": 0.0028693575877696276,
-      "learning_rate": 0.00019999644674207888,
-      "loss": 46.0,
-      "step": 16673
-    },
-    {
-      "epoch": 2.685212770240348,
-      "grad_norm": 0.0028689310420304537,
-      "learning_rate": 0.00019999644631558386,
-      "loss": 46.0,
-      "step": 16674
-    },
-    {
-      "epoch": 2.6853738073191353,
-      "grad_norm": 0.011154502630233765,
-      "learning_rate": 0.00019999644588906322,
-      "loss": 46.0,
-      "step": 16675
-    },
-    {
-      "epoch": 2.685534844397923,
-      "grad_norm": 0.001409060787409544,
-      "learning_rate": 0.00019999644546251697,
-      "loss": 46.0,
-      "step": 16676
-    },
-    {
-      "epoch": 2.6856958814767102,
-      "grad_norm": 0.0055886381305754185,
-      "learning_rate": 0.00019999644503594514,
-      "loss": 46.0,
-      "step": 16677
-    },
-    {
-      "epoch": 2.6858569185554972,
-      "grad_norm": 0.001857449417002499,
-      "learning_rate": 0.0001999964446093477,
-      "loss": 46.0,
-      "step": 16678
-    },
-    {
-      "epoch": 2.6860179556342847,
-      "grad_norm": 0.003105325624346733,
-      "learning_rate": 0.0001999964441827247,
-      "loss": 46.0,
-      "step": 16679
-    },
-    {
-      "epoch": 2.686178992713072,
-      "grad_norm": 0.005410938523709774,
-      "learning_rate": 0.0001999964437560761,
-      "loss": 46.0,
-      "step": 16680
-    },
-    {
-      "epoch": 2.6863400297918596,
-      "grad_norm": 0.009219738654792309,
-      "learning_rate": 0.0001999964433294019,
-      "loss": 46.0,
-      "step": 16681
-    },
-    {
-      "epoch": 2.686501066870647,
-      "grad_norm": 0.0007834638236090541,
-      "learning_rate": 0.00019999644290270206,
-      "loss": 46.0,
-      "step": 16682
-    },
-    {
-      "epoch": 2.6866621039494345,
-      "grad_norm": 0.0034031688701361418,
-      "learning_rate": 0.00019999644247597668,
-      "loss": 46.0,
-      "step": 16683
-    },
-    {
-      "epoch": 2.6868231410282215,
-      "grad_norm": 0.002383209764957428,
-      "learning_rate": 0.0001999964420492257,
-      "loss": 46.0,
-      "step": 16684
-    },
-    {
-      "epoch": 2.686984178107009,
-      "grad_norm": 0.003262204583734274,
-      "learning_rate": 0.00019999644162244911,
-      "loss": 46.0,
-      "step": 16685
-    },
-    {
-      "epoch": 2.6871452151857964,
-      "grad_norm": 0.0019388990476727486,
-      "learning_rate": 0.00019999644119564694,
-      "loss": 46.0,
-      "step": 16686
-    },
-    {
-      "epoch": 2.687306252264584,
-      "grad_norm": 0.0013207612792029977,
-      "learning_rate": 0.00019999644076881915,
-      "loss": 46.0,
-      "step": 16687
-    },
-    {
-      "epoch": 2.6874672893433713,
-      "grad_norm": 0.0009569731773808599,
-      "learning_rate": 0.0001999964403419658,
-      "loss": 46.0,
-      "step": 16688
-    },
-    {
-      "epoch": 2.6876283264221588,
-      "grad_norm": 0.004633996170014143,
-      "learning_rate": 0.00019999643991508684,
-      "loss": 46.0,
-      "step": 16689
-    },
-    {
-      "epoch": 2.687789363500946,
-      "grad_norm": 0.004391348920762539,
-      "learning_rate": 0.00019999643948818228,
-      "loss": 46.0,
-      "step": 16690
-    },
-    {
-      "epoch": 2.6879504005797337,
-      "grad_norm": 0.005670290905982256,
-      "learning_rate": 0.00019999643906125215,
-      "loss": 46.0,
-      "step": 16691
-    },
-    {
-      "epoch": 2.688111437658521,
-      "grad_norm": 0.001460038824006915,
-      "learning_rate": 0.0001999964386342964,
-      "loss": 46.0,
-      "step": 16692
-    },
-    {
-      "epoch": 2.688272474737308,
-      "grad_norm": 0.005185308866202831,
-      "learning_rate": 0.00019999643820731508,
-      "loss": 46.0,
-      "step": 16693
-    },
-    {
-      "epoch": 2.6884335118160956,
-      "grad_norm": 0.003059952985495329,
-      "learning_rate": 0.00019999643778030815,
-      "loss": 46.0,
-      "step": 16694
-    },
-    {
-      "epoch": 2.688594548894883,
-      "grad_norm": 0.0008635777630843222,
-      "learning_rate": 0.0001999964373532756,
-      "loss": 46.0,
-      "step": 16695
-    },
-    {
-      "epoch": 2.6887555859736705,
-      "grad_norm": 0.005537484306842089,
-      "learning_rate": 0.0001999964369262175,
-      "loss": 46.0,
-      "step": 16696
-    },
-    {
-      "epoch": 2.688916623052458,
-      "grad_norm": 0.003225943073630333,
-      "learning_rate": 0.0001999964364991338,
-      "loss": 46.0,
-      "step": 16697
-    },
-    {
-      "epoch": 2.6890776601312454,
-      "grad_norm": 0.0032401406206190586,
-      "learning_rate": 0.0001999964360720245,
-      "loss": 46.0,
-      "step": 16698
-    },
-    {
-      "epoch": 2.6892386972100324,
-      "grad_norm": 0.0031407340429723263,
-      "learning_rate": 0.0001999964356448896,
-      "loss": 46.0,
-      "step": 16699
-    },
-    {
-      "epoch": 2.68939973428882,
-      "grad_norm": 0.001973002217710018,
-      "learning_rate": 0.0001999964352177291,
-      "loss": 46.0,
-      "step": 16700
-    },
-    {
-      "epoch": 2.6895607713676073,
-      "grad_norm": 0.005462112836539745,
-      "learning_rate": 0.000199996434790543,
-      "loss": 46.0,
-      "step": 16701
-    },
-    {
-      "epoch": 2.6897218084463947,
-      "grad_norm": 0.006081890780478716,
-      "learning_rate": 0.00019999643436333133,
-      "loss": 46.0,
-      "step": 16702
-    },
-    {
-      "epoch": 2.689882845525182,
-      "grad_norm": 0.003943631891161203,
-      "learning_rate": 0.00019999643393609406,
-      "loss": 46.0,
-      "step": 16703
-    },
-    {
-      "epoch": 2.6900438826039696,
-      "grad_norm": 0.006720670498907566,
-      "learning_rate": 0.00019999643350883118,
-      "loss": 46.0,
-      "step": 16704
-    },
-    {
-      "epoch": 2.690204919682757,
-      "grad_norm": 0.008004121482372284,
-      "learning_rate": 0.00019999643308154274,
-      "loss": 46.0,
-      "step": 16705
-    },
-    {
-      "epoch": 2.6903659567615446,
-      "grad_norm": 0.0008740870398469269,
-      "learning_rate": 0.0001999964326542287,
-      "loss": 46.0,
-      "step": 16706
-    },
-    {
-      "epoch": 2.690526993840332,
-      "grad_norm": 0.0026336382143199444,
-      "learning_rate": 0.00019999643222688902,
-      "loss": 46.0,
-      "step": 16707
-    },
-    {
-      "epoch": 2.690688030919119,
-      "grad_norm": 0.00536775030195713,
-      "learning_rate": 0.0001999964317995238,
-      "loss": 46.0,
-      "step": 16708
-    },
-    {
-      "epoch": 2.6908490679979065,
-      "grad_norm": 0.0021431914065033197,
-      "learning_rate": 0.00019999643137213295,
-      "loss": 46.0,
-      "step": 16709
-    },
-    {
-      "epoch": 2.691010105076694,
-      "grad_norm": 0.0003728725132532418,
-      "learning_rate": 0.00019999643094471652,
-      "loss": 46.0,
-      "step": 16710
-    },
-    {
-      "epoch": 2.6911711421554814,
-      "grad_norm": 0.0013024089857935905,
-      "learning_rate": 0.00019999643051727447,
-      "loss": 46.0,
-      "step": 16711
-    },
-    {
-      "epoch": 2.691332179234269,
-      "grad_norm": 0.0015731453895568848,
-      "learning_rate": 0.00019999643008980687,
-      "loss": 46.0,
-      "step": 16712
-    },
-    {
-      "epoch": 2.691493216313056,
-      "grad_norm": 0.0014754356816411018,
-      "learning_rate": 0.00019999642966231368,
-      "loss": 46.0,
-      "step": 16713
-    },
-    {
-      "epoch": 2.6916542533918433,
-      "grad_norm": 0.00304122781381011,
-      "learning_rate": 0.00019999642923479484,
-      "loss": 46.0,
-      "step": 16714
-    },
-    {
-      "epoch": 2.6918152904706307,
-      "grad_norm": 0.0015769213205203414,
-      "learning_rate": 0.00019999642880725045,
-      "loss": 46.0,
-      "step": 16715
-    },
-    {
-      "epoch": 2.691976327549418,
-      "grad_norm": 0.012477298267185688,
-      "learning_rate": 0.00019999642837968047,
-      "loss": 46.0,
-      "step": 16716
-    },
-    {
-      "epoch": 2.6921373646282056,
-      "grad_norm": 0.0024488577619194984,
-      "learning_rate": 0.00019999642795208488,
-      "loss": 46.0,
-      "step": 16717
-    },
-    {
-      "epoch": 2.692298401706993,
-      "grad_norm": 0.009438673965632915,
-      "learning_rate": 0.0001999964275244637,
-      "loss": 46.0,
-      "step": 16718
-    },
-    {
-      "epoch": 2.6924594387857805,
-      "grad_norm": 0.008150266483426094,
-      "learning_rate": 0.00019999642709681693,
-      "loss": 46.0,
-      "step": 16719
-    },
-    {
-      "epoch": 2.692620475864568,
-      "grad_norm": 0.0016168324509635568,
-      "learning_rate": 0.00019999642666914454,
-      "loss": 46.0,
-      "step": 16720
-    },
-    {
-      "epoch": 2.6927815129433554,
-      "grad_norm": 0.005180926527827978,
-      "learning_rate": 0.00019999642624144657,
-      "loss": 46.0,
-      "step": 16721
-    },
-    {
-      "epoch": 2.6929425500221424,
-      "grad_norm": 0.00736199552193284,
-      "learning_rate": 0.00019999642581372301,
-      "loss": 46.0,
-      "step": 16722
-    },
-    {
-      "epoch": 2.69310358710093,
-      "grad_norm": 0.0028334404341876507,
-      "learning_rate": 0.00019999642538597387,
-      "loss": 46.0,
-      "step": 16723
-    },
-    {
-      "epoch": 2.6932646241797173,
-      "grad_norm": 0.0014165194006636739,
-      "learning_rate": 0.00019999642495819914,
-      "loss": 46.0,
-      "step": 16724
-    },
-    {
-      "epoch": 2.693425661258505,
-      "grad_norm": 0.001712548197247088,
-      "learning_rate": 0.00019999642453039876,
-      "loss": 46.0,
-      "step": 16725
-    },
-    {
-      "epoch": 2.6935866983372923,
-      "grad_norm": 0.0011719324393197894,
-      "learning_rate": 0.00019999642410257283,
-      "loss": 46.0,
-      "step": 16726
-    },
-    {
-      "epoch": 2.6937477354160797,
-      "grad_norm": 0.0020610301289707422,
-      "learning_rate": 0.00019999642367472134,
-      "loss": 46.0,
-      "step": 16727
-    },
-    {
-      "epoch": 2.6939087724948667,
-      "grad_norm": 0.0011872589820995927,
-      "learning_rate": 0.0001999964232468442,
-      "loss": 46.0,
-      "step": 16728
-    },
-    {
-      "epoch": 2.694069809573654,
-      "grad_norm": 0.0010340037988498807,
-      "learning_rate": 0.00019999642281894148,
-      "loss": 46.0,
-      "step": 16729
-    },
-    {
-      "epoch": 2.6942308466524416,
-      "grad_norm": 0.0005314347217790782,
-      "learning_rate": 0.00019999642239101317,
-      "loss": 46.0,
-      "step": 16730
-    },
-    {
-      "epoch": 2.694391883731229,
-      "grad_norm": 0.001003214973025024,
-      "learning_rate": 0.00019999642196305928,
-      "loss": 46.0,
-      "step": 16731
-    },
-    {
-      "epoch": 2.6945529208100165,
-      "grad_norm": 0.005966759752482176,
-      "learning_rate": 0.00019999642153507976,
-      "loss": 46.0,
-      "step": 16732
-    },
-    {
-      "epoch": 2.694713957888804,
-      "grad_norm": 0.005754299461841583,
-      "learning_rate": 0.0001999964211070747,
-      "loss": 46.0,
-      "step": 16733
-    },
-    {
-      "epoch": 2.6948749949675914,
-      "grad_norm": 0.006493737455457449,
-      "learning_rate": 0.00019999642067904398,
-      "loss": 46.0,
-      "step": 16734
-    },
-    {
-      "epoch": 2.695036032046379,
-      "grad_norm": 0.0005120303831063211,
-      "learning_rate": 0.00019999642025098774,
-      "loss": 46.0,
-      "step": 16735
-    },
-    {
-      "epoch": 2.6951970691251663,
-      "grad_norm": 0.0069949328899383545,
-      "learning_rate": 0.00019999641982290585,
-      "loss": 46.0,
-      "step": 16736
-    },
-    {
-      "epoch": 2.6953581062039533,
-      "grad_norm": 0.0009330203756690025,
-      "learning_rate": 0.00019999641939479838,
-      "loss": 46.0,
-      "step": 16737
-    },
-    {
-      "epoch": 2.695519143282741,
-      "grad_norm": 0.0012559612514451146,
-      "learning_rate": 0.0001999964189666653,
-      "loss": 46.0,
-      "step": 16738
-    },
-    {
-      "epoch": 2.6956801803615282,
-      "grad_norm": 0.013318131677806377,
-      "learning_rate": 0.00019999641853850667,
-      "loss": 46.0,
-      "step": 16739
-    },
-    {
-      "epoch": 2.6958412174403157,
-      "grad_norm": 0.001347652287222445,
-      "learning_rate": 0.00019999641811032238,
-      "loss": 46.0,
-      "step": 16740
-    },
-    {
-      "epoch": 2.696002254519103,
-      "grad_norm": 0.015256439335644245,
-      "learning_rate": 0.00019999641768211256,
-      "loss": 46.0,
-      "step": 16741
-    },
-    {
-      "epoch": 2.69616329159789,
-      "grad_norm": 0.0045393602922558784,
-      "learning_rate": 0.00019999641725387712,
-      "loss": 46.0,
-      "step": 16742
-    },
-    {
-      "epoch": 2.6963243286766776,
-      "grad_norm": 0.013451792299747467,
-      "learning_rate": 0.0001999964168256161,
-      "loss": 46.0,
-      "step": 16743
-    },
-    {
-      "epoch": 2.696485365755465,
-      "grad_norm": 0.0029337380547076464,
-      "learning_rate": 0.00019999641639732946,
-      "loss": 46.0,
-      "step": 16744
-    },
-    {
-      "epoch": 2.6966464028342525,
-      "grad_norm": 0.0027077605482190847,
-      "learning_rate": 0.00019999641596901726,
-      "loss": 46.0,
-      "step": 16745
-    },
-    {
-      "epoch": 2.69680743991304,
-      "grad_norm": 0.010373102501034737,
-      "learning_rate": 0.00019999641554067945,
-      "loss": 46.0,
-      "step": 16746
-    },
-    {
-      "epoch": 2.6969684769918274,
-      "grad_norm": 0.0015750667080283165,
-      "learning_rate": 0.00019999641511231602,
-      "loss": 46.0,
-      "step": 16747
-    },
-    {
-      "epoch": 2.697129514070615,
-      "grad_norm": 0.00912637822329998,
-      "learning_rate": 0.00019999641468392704,
-      "loss": 46.0,
-      "step": 16748
-    },
-    {
-      "epoch": 2.6972905511494023,
-      "grad_norm": 0.0009323619306087494,
-      "learning_rate": 0.00019999641425551244,
-      "loss": 46.0,
-      "step": 16749
-    },
-    {
-      "epoch": 2.6974515882281898,
-      "grad_norm": 0.007917189970612526,
-      "learning_rate": 0.00019999641382707225,
-      "loss": 46.0,
-      "step": 16750
-    },
-    {
-      "epoch": 2.697612625306977,
-      "grad_norm": 0.006665208842605352,
-      "learning_rate": 0.00019999641339860647,
-      "loss": 46.0,
-      "step": 16751
-    },
-    {
-      "epoch": 2.697773662385764,
-      "grad_norm": 0.0041270493529737,
-      "learning_rate": 0.00019999641297011506,
-      "loss": 46.0,
-      "step": 16752
-    },
-    {
-      "epoch": 2.6979346994645517,
-      "grad_norm": 0.0009448397904634476,
-      "learning_rate": 0.0001999964125415981,
-      "loss": 46.0,
-      "step": 16753
-    },
-    {
-      "epoch": 2.698095736543339,
-      "grad_norm": 0.0011741402558982372,
-      "learning_rate": 0.00019999641211305554,
-      "loss": 46.0,
-      "step": 16754
-    },
-    {
-      "epoch": 2.6982567736221266,
-      "grad_norm": 0.0037936896551400423,
-      "learning_rate": 0.00019999641168448736,
-      "loss": 46.0,
-      "step": 16755
-    },
-    {
-      "epoch": 2.698417810700914,
-      "grad_norm": 0.003401179565116763,
-      "learning_rate": 0.00019999641125589363,
-      "loss": 46.0,
-      "step": 16756
-    },
-    {
-      "epoch": 2.698578847779701,
-      "grad_norm": 0.009221816435456276,
-      "learning_rate": 0.00019999641082727427,
-      "loss": 46.0,
-      "step": 16757
-    },
-    {
-      "epoch": 2.6987398848584885,
-      "grad_norm": 0.004651486407965422,
-      "learning_rate": 0.00019999641039862933,
-      "loss": 46.0,
-      "step": 16758
-    },
-    {
-      "epoch": 2.698900921937276,
-      "grad_norm": 0.0019382634200155735,
-      "learning_rate": 0.0001999964099699588,
-      "loss": 46.0,
-      "step": 16759
-    },
-    {
-      "epoch": 2.6990619590160634,
-      "grad_norm": 0.000692696135956794,
-      "learning_rate": 0.0001999964095412627,
-      "loss": 46.0,
-      "step": 16760
-    },
-    {
-      "epoch": 2.699222996094851,
-      "grad_norm": 0.001394860097207129,
-      "learning_rate": 0.00019999640911254096,
-      "loss": 46.0,
-      "step": 16761
-    },
-    {
-      "epoch": 2.6993840331736383,
-      "grad_norm": 0.0010718988487496972,
-      "learning_rate": 0.00019999640868379362,
-      "loss": 46.0,
-      "step": 16762
-    },
-    {
-      "epoch": 2.6995450702524257,
-      "grad_norm": 0.005175519268959761,
-      "learning_rate": 0.00019999640825502075,
-      "loss": 46.0,
-      "step": 16763
-    },
-    {
-      "epoch": 2.699706107331213,
-      "grad_norm": 0.002591567113995552,
-      "learning_rate": 0.00019999640782622223,
-      "loss": 46.0,
-      "step": 16764
-    },
-    {
-      "epoch": 2.6998671444100006,
-      "grad_norm": 0.002293346682563424,
-      "learning_rate": 0.00019999640739739813,
-      "loss": 46.0,
-      "step": 16765
-    },
-    {
-      "epoch": 2.7000281814887876,
-      "grad_norm": 0.0018217446049675345,
-      "learning_rate": 0.00019999640696854843,
-      "loss": 46.0,
-      "step": 16766
-    },
-    {
-      "epoch": 2.700189218567575,
-      "grad_norm": 0.004994076676666737,
-      "learning_rate": 0.00019999640653967316,
-      "loss": 46.0,
-      "step": 16767
-    },
-    {
-      "epoch": 2.7003502556463626,
-      "grad_norm": 0.0022861426696181297,
-      "learning_rate": 0.0001999964061107723,
-      "loss": 46.0,
-      "step": 16768
-    },
-    {
-      "epoch": 2.70051129272515,
-      "grad_norm": 0.0044463323429226875,
-      "learning_rate": 0.00019999640568184578,
-      "loss": 46.0,
-      "step": 16769
-    },
-    {
-      "epoch": 2.7006723298039375,
-      "grad_norm": 0.002989982021972537,
-      "learning_rate": 0.00019999640525289372,
-      "loss": 46.0,
-      "step": 16770
-    },
-    {
-      "epoch": 2.700833366882725,
-      "grad_norm": 0.0012834639055654407,
-      "learning_rate": 0.00019999640482391606,
-      "loss": 46.0,
-      "step": 16771
-    },
-    {
-      "epoch": 2.700994403961512,
-      "grad_norm": 0.005140871275216341,
-      "learning_rate": 0.00019999640439491282,
-      "loss": 46.0,
-      "step": 16772
-    },
-    {
-      "epoch": 2.7011554410402994,
-      "grad_norm": 0.0017137883696705103,
-      "learning_rate": 0.00019999640396588397,
-      "loss": 46.0,
-      "step": 16773
-    },
-    {
-      "epoch": 2.701316478119087,
-      "grad_norm": 0.006277171429246664,
-      "learning_rate": 0.00019999640353682952,
-      "loss": 46.0,
-      "step": 16774
-    },
-    {
-      "epoch": 2.7014775151978743,
-      "grad_norm": 0.007688910700380802,
-      "learning_rate": 0.00019999640310774947,
-      "loss": 46.0,
-      "step": 16775
-    },
-    {
-      "epoch": 2.7016385522766617,
-      "grad_norm": 0.0008759669144637883,
-      "learning_rate": 0.00019999640267864385,
-      "loss": 46.0,
-      "step": 16776
-    },
-    {
-      "epoch": 2.701799589355449,
-      "grad_norm": 0.001542112324386835,
-      "learning_rate": 0.00019999640224951262,
-      "loss": 46.0,
-      "step": 16777
-    },
-    {
-      "epoch": 2.7019606264342366,
-      "grad_norm": 0.0038784516509622335,
-      "learning_rate": 0.00019999640182035583,
-      "loss": 46.0,
-      "step": 16778
-    },
-    {
-      "epoch": 2.702121663513024,
-      "grad_norm": 0.000979863223619759,
-      "learning_rate": 0.0001999964013911734,
-      "loss": 46.0,
-      "step": 16779
-    },
-    {
-      "epoch": 2.7022827005918115,
-      "grad_norm": 0.010109331458806992,
-      "learning_rate": 0.0001999964009619654,
-      "loss": 46.0,
-      "step": 16780
-    },
-    {
-      "epoch": 2.7024437376705985,
-      "grad_norm": 0.0008103515137918293,
-      "learning_rate": 0.00019999640053273177,
-      "loss": 46.0,
-      "step": 16781
-    },
-    {
-      "epoch": 2.702604774749386,
-      "grad_norm": 0.004866925999522209,
-      "learning_rate": 0.00019999640010347257,
-      "loss": 46.0,
-      "step": 16782
-    },
-    {
-      "epoch": 2.7027658118281734,
-      "grad_norm": 0.0006917114951647818,
-      "learning_rate": 0.0001999963996741878,
-      "loss": 46.0,
-      "step": 16783
-    },
-    {
-      "epoch": 2.702926848906961,
-      "grad_norm": 0.0007846925291232765,
-      "learning_rate": 0.00019999639924487742,
-      "loss": 46.0,
-      "step": 16784
-    },
-    {
-      "epoch": 2.7030878859857483,
-      "grad_norm": 0.004074486438184977,
-      "learning_rate": 0.0001999963988155414,
-      "loss": 46.0,
-      "step": 16785
-    },
-    {
-      "epoch": 2.7032489230645353,
-      "grad_norm": 0.006787537597119808,
-      "learning_rate": 0.00019999639838617984,
-      "loss": 46.0,
-      "step": 16786
-    },
-    {
-      "epoch": 2.703409960143323,
-      "grad_norm": 0.004249176941812038,
-      "learning_rate": 0.0001999963979567927,
-      "loss": 46.0,
-      "step": 16787
-    },
-    {
-      "epoch": 2.7035709972221103,
-      "grad_norm": 0.001418575062416494,
-      "learning_rate": 0.00019999639752737994,
-      "loss": 46.0,
-      "step": 16788
-    },
-    {
-      "epoch": 2.7037320343008977,
-      "grad_norm": 0.01058497466146946,
-      "learning_rate": 0.00019999639709794158,
-      "loss": 46.0,
-      "step": 16789
-    },
-    {
-      "epoch": 2.703893071379685,
-      "grad_norm": 0.0012655415339395404,
-      "learning_rate": 0.00019999639666847766,
-      "loss": 46.0,
-      "step": 16790
-    },
-    {
-      "epoch": 2.7040541084584726,
-      "grad_norm": 0.003899478819221258,
-      "learning_rate": 0.0001999963962389881,
-      "loss": 46.0,
-      "step": 16791
-    },
-    {
-      "epoch": 2.70421514553726,
-      "grad_norm": 0.008875411935150623,
-      "learning_rate": 0.00019999639580947295,
-      "loss": 46.0,
-      "step": 16792
-    },
-    {
-      "epoch": 2.7043761826160475,
-      "grad_norm": 0.0038673330564051867,
-      "learning_rate": 0.00019999639537993225,
-      "loss": 46.0,
-      "step": 16793
-    },
-    {
-      "epoch": 2.704537219694835,
-      "grad_norm": 0.005630007945001125,
-      "learning_rate": 0.0001999963949503659,
-      "loss": 46.0,
-      "step": 16794
-    },
-    {
-      "epoch": 2.704698256773622,
-      "grad_norm": 0.001027962425723672,
-      "learning_rate": 0.000199996394520774,
-      "loss": 46.0,
-      "step": 16795
-    },
-    {
-      "epoch": 2.7048592938524094,
-      "grad_norm": 0.0008963792934082448,
-      "learning_rate": 0.0001999963940911565,
-      "loss": 46.0,
-      "step": 16796
-    },
-    {
-      "epoch": 2.705020330931197,
-      "grad_norm": 0.007647103630006313,
-      "learning_rate": 0.00019999639366151338,
-      "loss": 46.0,
-      "step": 16797
-    },
-    {
-      "epoch": 2.7051813680099843,
-      "grad_norm": 0.002506924793124199,
-      "learning_rate": 0.00019999639323184466,
-      "loss": 46.0,
-      "step": 16798
-    },
-    {
-      "epoch": 2.7053424050887718,
-      "grad_norm": 0.004357772413641214,
-      "learning_rate": 0.0001999963928021504,
-      "loss": 46.0,
-      "step": 16799
-    },
-    {
-      "epoch": 2.7055034421675592,
-      "grad_norm": 0.006169793661683798,
-      "learning_rate": 0.0001999963923724305,
-      "loss": 46.0,
-      "step": 16800
-    },
-    {
-      "epoch": 2.7056644792463462,
-      "grad_norm": 0.0024810058530420065,
-      "learning_rate": 0.00019999639194268501,
-      "loss": 46.0,
-      "step": 16801
-    },
-    {
-      "epoch": 2.7058255163251337,
-      "grad_norm": 0.0013228394091129303,
-      "learning_rate": 0.00019999639151291394,
-      "loss": 46.0,
-      "step": 16802
-    },
-    {
-      "epoch": 2.705986553403921,
-      "grad_norm": 0.0034430488012731075,
-      "learning_rate": 0.0001999963910831173,
-      "loss": 46.0,
-      "step": 16803
-    },
-    {
-      "epoch": 2.7061475904827086,
-      "grad_norm": 0.004819844849407673,
-      "learning_rate": 0.00019999639065329503,
-      "loss": 46.0,
-      "step": 16804
-    },
-    {
-      "epoch": 2.706308627561496,
-      "grad_norm": 0.0024054015520960093,
-      "learning_rate": 0.00019999639022344717,
-      "loss": 46.0,
-      "step": 16805
-    },
-    {
-      "epoch": 2.7064696646402835,
-      "grad_norm": 0.0015642436919733882,
-      "learning_rate": 0.00019999638979357372,
-      "loss": 46.0,
-      "step": 16806
-    },
-    {
-      "epoch": 2.706630701719071,
-      "grad_norm": 0.0018935829866677523,
-      "learning_rate": 0.00019999638936367466,
-      "loss": 46.0,
-      "step": 16807
-    },
-    {
-      "epoch": 2.7067917387978584,
-      "grad_norm": 0.0017554400255903602,
-      "learning_rate": 0.00019999638893375003,
-      "loss": 46.0,
-      "step": 16808
-    },
-    {
-      "epoch": 2.706952775876646,
-      "grad_norm": 0.0012979459716007113,
-      "learning_rate": 0.00019999638850379977,
-      "loss": 46.0,
-      "step": 16809
-    },
-    {
-      "epoch": 2.707113812955433,
-      "grad_norm": 0.005419580731540918,
-      "learning_rate": 0.00019999638807382397,
-      "loss": 46.0,
-      "step": 16810
-    },
-    {
-      "epoch": 2.7072748500342203,
-      "grad_norm": 0.0005155244725756347,
-      "learning_rate": 0.00019999638764382253,
-      "loss": 46.0,
-      "step": 16811
-    },
-    {
-      "epoch": 2.7074358871130078,
-      "grad_norm": 0.0033099977299571037,
-      "learning_rate": 0.00019999638721379553,
-      "loss": 46.0,
-      "step": 16812
-    },
-    {
-      "epoch": 2.707596924191795,
-      "grad_norm": 0.005884178448468447,
-      "learning_rate": 0.00019999638678374292,
-      "loss": 46.0,
-      "step": 16813
-    },
-    {
-      "epoch": 2.7077579612705827,
-      "grad_norm": 0.004039480816572905,
-      "learning_rate": 0.00019999638635366475,
-      "loss": 46.0,
-      "step": 16814
-    },
-    {
-      "epoch": 2.7079189983493697,
-      "grad_norm": 0.008092648349702358,
-      "learning_rate": 0.00019999638592356093,
-      "loss": 46.0,
-      "step": 16815
-    },
-    {
-      "epoch": 2.708080035428157,
-      "grad_norm": 0.006574242375791073,
-      "learning_rate": 0.00019999638549343153,
-      "loss": 46.0,
-      "step": 16816
-    },
-    {
-      "epoch": 2.7082410725069446,
-      "grad_norm": 0.0010245295707136393,
-      "learning_rate": 0.00019999638506327657,
-      "loss": 46.0,
-      "step": 16817
-    },
-    {
-      "epoch": 2.708402109585732,
-      "grad_norm": 0.00125090591609478,
-      "learning_rate": 0.000199996384633096,
-      "loss": 46.0,
-      "step": 16818
-    },
-    {
-      "epoch": 2.7085631466645195,
-      "grad_norm": 0.0008006748394109309,
-      "learning_rate": 0.0001999963842028898,
-      "loss": 46.0,
-      "step": 16819
-    },
-    {
-      "epoch": 2.708724183743307,
-      "grad_norm": 0.006743420381098986,
-      "learning_rate": 0.00019999638377265805,
-      "loss": 46.0,
-      "step": 16820
-    },
-    {
-      "epoch": 2.7088852208220944,
-      "grad_norm": 0.0009717715438455343,
-      "learning_rate": 0.00019999638334240068,
-      "loss": 46.0,
-      "step": 16821
-    },
-    {
-      "epoch": 2.709046257900882,
-      "grad_norm": 0.0025457418523728848,
-      "learning_rate": 0.00019999638291211773,
-      "loss": 46.0,
-      "step": 16822
-    },
-    {
-      "epoch": 2.7092072949796693,
-      "grad_norm": 0.00406483793631196,
-      "learning_rate": 0.0001999963824818092,
-      "loss": 46.0,
-      "step": 16823
-    },
-    {
-      "epoch": 2.7093683320584567,
-      "grad_norm": 0.005037358496338129,
-      "learning_rate": 0.00019999638205147504,
-      "loss": 46.0,
-      "step": 16824
-    },
-    {
-      "epoch": 2.7095293691372437,
-      "grad_norm": 0.0051717557944357395,
-      "learning_rate": 0.00019999638162111532,
-      "loss": 46.0,
-      "step": 16825
-    },
-    {
-      "epoch": 2.709690406216031,
-      "grad_norm": 0.018700547516345978,
-      "learning_rate": 0.00019999638119073,
-      "loss": 46.0,
-      "step": 16826
-    },
-    {
-      "epoch": 2.7098514432948186,
-      "grad_norm": 0.0010653989156708121,
-      "learning_rate": 0.00019999638076031905,
-      "loss": 46.0,
-      "step": 16827
-    },
-    {
-      "epoch": 2.710012480373606,
-      "grad_norm": 0.0027403435669839382,
-      "learning_rate": 0.00019999638032988255,
-      "loss": 46.0,
-      "step": 16828
-    },
-    {
-      "epoch": 2.7101735174523935,
-      "grad_norm": 0.004282352048903704,
-      "learning_rate": 0.00019999637989942043,
-      "loss": 46.0,
-      "step": 16829
-    },
-    {
-      "epoch": 2.7103345545311806,
-      "grad_norm": 0.0018186519155278802,
-      "learning_rate": 0.00019999637946893273,
-      "loss": 46.0,
-      "step": 16830
-    },
-    {
-      "epoch": 2.710495591609968,
-      "grad_norm": 0.00401414604857564,
-      "learning_rate": 0.00019999637903841944,
-      "loss": 46.0,
-      "step": 16831
-    },
-    {
-      "epoch": 2.7106566286887555,
-      "grad_norm": 0.0056733377277851105,
-      "learning_rate": 0.00019999637860788053,
-      "loss": 46.0,
-      "step": 16832
-    },
-    {
-      "epoch": 2.710817665767543,
-      "grad_norm": 0.007255855482071638,
-      "learning_rate": 0.00019999637817731604,
-      "loss": 46.0,
-      "step": 16833
-    },
-    {
-      "epoch": 2.7109787028463304,
-      "grad_norm": 0.002388984663411975,
-      "learning_rate": 0.00019999637774672596,
-      "loss": 46.0,
-      "step": 16834
-    },
-    {
-      "epoch": 2.711139739925118,
-      "grad_norm": 0.0017402877565473318,
-      "learning_rate": 0.0001999963773161103,
-      "loss": 46.0,
-      "step": 16835
-    },
-    {
-      "epoch": 2.7113007770039053,
-      "grad_norm": 0.0021618285682052374,
-      "learning_rate": 0.00019999637688546901,
-      "loss": 46.0,
-      "step": 16836
-    },
-    {
-      "epoch": 2.7114618140826927,
-      "grad_norm": 0.0017112528439611197,
-      "learning_rate": 0.00019999637645480215,
-      "loss": 46.0,
-      "step": 16837
-    },
-    {
-      "epoch": 2.71162285116148,
-      "grad_norm": 0.0033222208730876446,
-      "learning_rate": 0.0001999963760241097,
-      "loss": 46.0,
-      "step": 16838
-    },
-    {
-      "epoch": 2.711783888240267,
-      "grad_norm": 0.0048472392372787,
-      "learning_rate": 0.00019999637559339165,
-      "loss": 46.0,
-      "step": 16839
-    },
-    {
-      "epoch": 2.7119449253190546,
-      "grad_norm": 0.0017181599978357553,
-      "learning_rate": 0.000199996375162648,
-      "loss": 46.0,
-      "step": 16840
-    },
-    {
-      "epoch": 2.712105962397842,
-      "grad_norm": 0.004129703156650066,
-      "learning_rate": 0.00019999637473187875,
-      "loss": 46.0,
-      "step": 16841
-    },
-    {
-      "epoch": 2.7122669994766295,
-      "grad_norm": 0.000810042314697057,
-      "learning_rate": 0.00019999637430108394,
-      "loss": 46.0,
-      "step": 16842
-    },
-    {
-      "epoch": 2.712428036555417,
-      "grad_norm": 0.00144431262742728,
-      "learning_rate": 0.0001999963738702635,
-      "loss": 46.0,
-      "step": 16843
-    },
-    {
-      "epoch": 2.7125890736342044,
-      "grad_norm": 0.007915292866528034,
-      "learning_rate": 0.00019999637343941752,
-      "loss": 46.0,
-      "step": 16844
-    },
-    {
-      "epoch": 2.7127501107129914,
-      "grad_norm": 0.002444526879116893,
-      "learning_rate": 0.00019999637300854587,
-      "loss": 46.0,
-      "step": 16845
-    },
-    {
-      "epoch": 2.712911147791779,
-      "grad_norm": 0.0017184007447212934,
-      "learning_rate": 0.00019999637257764867,
-      "loss": 46.0,
-      "step": 16846
-    },
-    {
-      "epoch": 2.7130721848705663,
-      "grad_norm": 0.004984188359230757,
-      "learning_rate": 0.00019999637214672587,
-      "loss": 46.0,
-      "step": 16847
-    },
-    {
-      "epoch": 2.713233221949354,
-      "grad_norm": 0.003552086651325226,
-      "learning_rate": 0.00019999637171577746,
-      "loss": 46.0,
-      "step": 16848
-    },
-    {
-      "epoch": 2.7133942590281412,
-      "grad_norm": 0.0052740732207894325,
-      "learning_rate": 0.0001999963712848035,
-      "loss": 46.0,
-      "step": 16849
-    },
-    {
-      "epoch": 2.7135552961069287,
-      "grad_norm": 0.0037684133276343346,
-      "learning_rate": 0.0001999963708538039,
-      "loss": 46.0,
-      "step": 16850
-    },
-    {
-      "epoch": 2.713716333185716,
-      "grad_norm": 0.0009654142195358872,
-      "learning_rate": 0.00019999637042277871,
-      "loss": 46.0,
-      "step": 16851
-    },
-    {
-      "epoch": 2.7138773702645036,
-      "grad_norm": 0.0017676504794508219,
-      "learning_rate": 0.00019999636999172796,
-      "loss": 46.0,
-      "step": 16852
-    },
-    {
-      "epoch": 2.714038407343291,
-      "grad_norm": 0.003571895882487297,
-      "learning_rate": 0.00019999636956065159,
-      "loss": 46.0,
-      "step": 16853
-    },
-    {
-      "epoch": 2.714199444422078,
-      "grad_norm": 0.0020821099169552326,
-      "learning_rate": 0.00019999636912954963,
-      "loss": 46.0,
-      "step": 16854
-    },
-    {
-      "epoch": 2.7143604815008655,
-      "grad_norm": 0.002076129661872983,
-      "learning_rate": 0.00019999636869842208,
-      "loss": 46.0,
-      "step": 16855
-    },
-    {
-      "epoch": 2.714521518579653,
-      "grad_norm": 0.004761275369673967,
-      "learning_rate": 0.00019999636826726892,
-      "loss": 46.0,
-      "step": 16856
-    },
-    {
-      "epoch": 2.7146825556584404,
-      "grad_norm": 0.0009966637007892132,
-      "learning_rate": 0.0001999963678360902,
-      "loss": 46.0,
-      "step": 16857
-    },
-    {
-      "epoch": 2.714843592737228,
-      "grad_norm": 0.0015429897466674447,
-      "learning_rate": 0.00019999636740488587,
-      "loss": 46.0,
-      "step": 16858
-    },
-    {
-      "epoch": 2.715004629816015,
-      "grad_norm": 0.00360603304579854,
-      "learning_rate": 0.00019999636697365595,
-      "loss": 46.0,
-      "step": 16859
-    },
-    {
-      "epoch": 2.7151656668948023,
-      "grad_norm": 0.0017117158276960254,
-      "learning_rate": 0.00019999636654240038,
-      "loss": 46.0,
-      "step": 16860
-    },
-    {
-      "epoch": 2.7153267039735898,
-      "grad_norm": 0.002073318464681506,
-      "learning_rate": 0.00019999636611111926,
-      "loss": 46.0,
-      "step": 16861
-    },
-    {
-      "epoch": 2.7154877410523772,
-      "grad_norm": 0.001088133081793785,
-      "learning_rate": 0.00019999636567981258,
-      "loss": 46.0,
-      "step": 16862
-    },
-    {
-      "epoch": 2.7156487781311647,
-      "grad_norm": 0.003052320796996355,
-      "learning_rate": 0.00019999636524848028,
-      "loss": 46.0,
-      "step": 16863
-    },
-    {
-      "epoch": 2.715809815209952,
-      "grad_norm": 0.0020853327587246895,
-      "learning_rate": 0.00019999636481712237,
-      "loss": 46.0,
-      "step": 16864
-    },
-    {
-      "epoch": 2.7159708522887396,
-      "grad_norm": 0.015134351328015327,
-      "learning_rate": 0.00019999636438573887,
-      "loss": 46.0,
-      "step": 16865
-    },
-    {
-      "epoch": 2.716131889367527,
-      "grad_norm": 0.001310597755946219,
-      "learning_rate": 0.00019999636395432978,
-      "loss": 46.0,
-      "step": 16866
-    },
-    {
-      "epoch": 2.7162929264463145,
-      "grad_norm": 0.0006142769125290215,
-      "learning_rate": 0.0001999963635228951,
-      "loss": 46.0,
-      "step": 16867
-    },
-    {
-      "epoch": 2.7164539635251015,
-      "grad_norm": 0.0016236824449151754,
-      "learning_rate": 0.00019999636309143482,
-      "loss": 46.0,
-      "step": 16868
-    },
-    {
-      "epoch": 2.716615000603889,
-      "grad_norm": 0.004155067726969719,
-      "learning_rate": 0.00019999636265994897,
-      "loss": 46.0,
-      "step": 16869
-    },
-    {
-      "epoch": 2.7167760376826764,
-      "grad_norm": 0.0015501020243391395,
-      "learning_rate": 0.00019999636222843749,
-      "loss": 46.0,
-      "step": 16870
-    },
-    {
-      "epoch": 2.716937074761464,
-      "grad_norm": 0.0024400365073233843,
-      "learning_rate": 0.00019999636179690044,
-      "loss": 46.0,
-      "step": 16871
-    },
-    {
-      "epoch": 2.7170981118402513,
-      "grad_norm": 0.0010739787248894572,
-      "learning_rate": 0.0001999963613653378,
-      "loss": 46.0,
-      "step": 16872
-    },
-    {
-      "epoch": 2.7172591489190387,
-      "grad_norm": 0.0010813574772328138,
-      "learning_rate": 0.00019999636093374955,
-      "loss": 46.0,
-      "step": 16873
-    },
-    {
-      "epoch": 2.7174201859978258,
-      "grad_norm": 0.0032039438374340534,
-      "learning_rate": 0.00019999636050213574,
-      "loss": 46.0,
-      "step": 16874
-    },
-    {
-      "epoch": 2.717581223076613,
-      "grad_norm": 0.01437385380268097,
-      "learning_rate": 0.0001999963600704963,
-      "loss": 46.0,
-      "step": 16875
-    },
-    {
-      "epoch": 2.7177422601554007,
-      "grad_norm": 0.001206106855534017,
-      "learning_rate": 0.00019999635963883125,
-      "loss": 46.0,
-      "step": 16876
-    },
-    {
-      "epoch": 2.717903297234188,
-      "grad_norm": 0.0036650982219725847,
-      "learning_rate": 0.00019999635920714065,
-      "loss": 46.0,
-      "step": 16877
-    },
-    {
-      "epoch": 2.7180643343129756,
-      "grad_norm": 0.0030384594574570656,
-      "learning_rate": 0.0001999963587754244,
-      "loss": 46.0,
-      "step": 16878
-    },
-    {
-      "epoch": 2.718225371391763,
-      "grad_norm": 0.007299521472305059,
-      "learning_rate": 0.0001999963583436826,
-      "loss": 46.0,
-      "step": 16879
-    },
-    {
-      "epoch": 2.7183864084705505,
-      "grad_norm": 0.0033372414764016867,
-      "learning_rate": 0.0001999963579119152,
-      "loss": 46.0,
-      "step": 16880
-    },
-    {
-      "epoch": 2.718547445549338,
-      "grad_norm": 0.005126938223838806,
-      "learning_rate": 0.0001999963574801222,
-      "loss": 46.0,
-      "step": 16881
-    },
-    {
-      "epoch": 2.7187084826281254,
-      "grad_norm": 0.0023445086553692818,
-      "learning_rate": 0.00019999635704830363,
-      "loss": 46.0,
-      "step": 16882
-    },
-    {
-      "epoch": 2.7188695197069124,
-      "grad_norm": 0.0013366233324632049,
-      "learning_rate": 0.00019999635661645943,
-      "loss": 46.0,
-      "step": 16883
-    },
-    {
-      "epoch": 2.7190305567857,
-      "grad_norm": 0.0010768286883831024,
-      "learning_rate": 0.00019999635618458966,
-      "loss": 46.0,
-      "step": 16884
-    },
-    {
-      "epoch": 2.7191915938644873,
-      "grad_norm": 0.005694425664842129,
-      "learning_rate": 0.00019999635575269429,
-      "loss": 46.0,
-      "step": 16885
-    },
-    {
-      "epoch": 2.7193526309432747,
-      "grad_norm": 0.0034576845355331898,
-      "learning_rate": 0.00019999635532077332,
-      "loss": 46.0,
-      "step": 16886
-    },
-    {
-      "epoch": 2.719513668022062,
-      "grad_norm": 0.003664106596261263,
-      "learning_rate": 0.00019999635488882674,
-      "loss": 46.0,
-      "step": 16887
-    },
-    {
-      "epoch": 2.7196747051008496,
-      "grad_norm": 0.0022734389640390873,
-      "learning_rate": 0.0001999963544568546,
-      "loss": 46.0,
-      "step": 16888
-    },
-    {
-      "epoch": 2.7198357421796366,
-      "grad_norm": 0.0016641359543427825,
-      "learning_rate": 0.00019999635402485688,
-      "loss": 46.0,
-      "step": 16889
-    },
-    {
-      "epoch": 2.719996779258424,
-      "grad_norm": 0.00746966153383255,
-      "learning_rate": 0.00019999635359283353,
-      "loss": 46.0,
-      "step": 16890
-    },
-    {
-      "epoch": 2.7201578163372115,
-      "grad_norm": 0.0013379884185269475,
-      "learning_rate": 0.0001999963531607846,
-      "loss": 46.0,
-      "step": 16891
-    },
-    {
-      "epoch": 2.720318853415999,
-      "grad_norm": 0.005165590904653072,
-      "learning_rate": 0.00019999635272871004,
-      "loss": 46.0,
-      "step": 16892
-    },
-    {
-      "epoch": 2.7204798904947864,
-      "grad_norm": 0.004248811397701502,
-      "learning_rate": 0.00019999635229660994,
-      "loss": 46.0,
-      "step": 16893
-    },
-    {
-      "epoch": 2.720640927573574,
-      "grad_norm": 0.00433771125972271,
-      "learning_rate": 0.00019999635186448422,
-      "loss": 46.0,
-      "step": 16894
-    },
-    {
-      "epoch": 2.7208019646523613,
-      "grad_norm": 0.0009444046299904585,
-      "learning_rate": 0.00019999635143233292,
-      "loss": 46.0,
-      "step": 16895
-    },
-    {
-      "epoch": 2.720963001731149,
-      "grad_norm": 0.00263758422806859,
-      "learning_rate": 0.000199996351000156,
-      "loss": 46.0,
-      "step": 16896
-    },
-    {
-      "epoch": 2.7211240388099363,
-      "grad_norm": 0.0028068292886018753,
-      "learning_rate": 0.0001999963505679535,
-      "loss": 46.0,
-      "step": 16897
-    },
-    {
-      "epoch": 2.7212850758887233,
-      "grad_norm": 0.0009865900501608849,
-      "learning_rate": 0.00019999635013572543,
-      "loss": 46.0,
-      "step": 16898
-    },
-    {
-      "epoch": 2.7214461129675107,
-      "grad_norm": 0.0031842507887631655,
-      "learning_rate": 0.00019999634970347172,
-      "loss": 46.0,
-      "step": 16899
-    },
-    {
-      "epoch": 2.721607150046298,
-      "grad_norm": 0.006077928002923727,
-      "learning_rate": 0.00019999634927119243,
-      "loss": 46.0,
-      "step": 16900
-    },
-    {
-      "epoch": 2.7217681871250856,
-      "grad_norm": 0.0009807192254811525,
-      "learning_rate": 0.00019999634883888757,
-      "loss": 46.0,
-      "step": 16901
-    },
-    {
-      "epoch": 2.721929224203873,
-      "grad_norm": 0.0038750700186938047,
-      "learning_rate": 0.0001999963484065571,
-      "loss": 46.0,
-      "step": 16902
-    },
-    {
-      "epoch": 2.72209026128266,
-      "grad_norm": 0.0015949290245771408,
-      "learning_rate": 0.00019999634797420105,
-      "loss": 46.0,
-      "step": 16903
-    },
-    {
-      "epoch": 2.7222512983614475,
-      "grad_norm": 0.005834200419485569,
-      "learning_rate": 0.0001999963475418194,
-      "loss": 46.0,
-      "step": 16904
-    },
-    {
-      "epoch": 2.722412335440235,
-      "grad_norm": 0.004745587706565857,
-      "learning_rate": 0.00019999634710941212,
-      "loss": 46.0,
-      "step": 16905
-    },
-    {
-      "epoch": 2.7225733725190224,
-      "grad_norm": 0.002906380221247673,
-      "learning_rate": 0.00019999634667697928,
-      "loss": 46.0,
-      "step": 16906
-    },
-    {
-      "epoch": 2.72273440959781,
-      "grad_norm": 0.0029213724192231894,
-      "learning_rate": 0.00019999634624452085,
-      "loss": 46.0,
-      "step": 16907
-    },
-    {
-      "epoch": 2.7228954466765973,
-      "grad_norm": 0.0010784838814288378,
-      "learning_rate": 0.0001999963458120368,
-      "loss": 46.0,
-      "step": 16908
-    },
-    {
-      "epoch": 2.723056483755385,
-      "grad_norm": 0.008651919662952423,
-      "learning_rate": 0.00019999634537952717,
-      "loss": 46.0,
-      "step": 16909
-    },
-    {
-      "epoch": 2.7232175208341722,
-      "grad_norm": 0.0023041178938001394,
-      "learning_rate": 0.00019999634494699198,
-      "loss": 46.0,
-      "step": 16910
-    },
-    {
-      "epoch": 2.7233785579129597,
-      "grad_norm": 0.00258329208008945,
-      "learning_rate": 0.00019999634451443114,
-      "loss": 46.0,
-      "step": 16911
-    },
-    {
-      "epoch": 2.7235395949917467,
-      "grad_norm": 0.0018589961109682918,
-      "learning_rate": 0.00019999634408184475,
-      "loss": 46.0,
-      "step": 16912
-    },
-    {
-      "epoch": 2.723700632070534,
-      "grad_norm": 0.002541129244491458,
-      "learning_rate": 0.00019999634364923274,
-      "loss": 46.0,
-      "step": 16913
-    },
-    {
-      "epoch": 2.7238616691493216,
-      "grad_norm": 0.00421768007799983,
-      "learning_rate": 0.00019999634321659515,
-      "loss": 46.0,
-      "step": 16914
-    },
-    {
-      "epoch": 2.724022706228109,
-      "grad_norm": 0.0016531634610146284,
-      "learning_rate": 0.000199996342783932,
-      "loss": 46.0,
-      "step": 16915
-    },
-    {
-      "epoch": 2.7241837433068965,
-      "grad_norm": 0.0028021091129630804,
-      "learning_rate": 0.0001999963423512432,
-      "loss": 46.0,
-      "step": 16916
-    },
-    {
-      "epoch": 2.724344780385684,
-      "grad_norm": 0.0019305588211864233,
-      "learning_rate": 0.0001999963419185288,
-      "loss": 46.0,
-      "step": 16917
-    },
-    {
-      "epoch": 2.724505817464471,
-      "grad_norm": 0.0008829992148093879,
-      "learning_rate": 0.00019999634148578884,
-      "loss": 46.0,
-      "step": 16918
-    },
-    {
-      "epoch": 2.7246668545432584,
-      "grad_norm": 0.0004995696363039315,
-      "learning_rate": 0.00019999634105302328,
-      "loss": 46.0,
-      "step": 16919
-    },
-    {
-      "epoch": 2.724827891622046,
-      "grad_norm": 0.003270841436460614,
-      "learning_rate": 0.0001999963406202321,
-      "loss": 46.0,
-      "step": 16920
-    },
-    {
-      "epoch": 2.7249889287008333,
-      "grad_norm": 0.005590683314949274,
-      "learning_rate": 0.00019999634018741535,
-      "loss": 46.0,
-      "step": 16921
-    },
-    {
-      "epoch": 2.7251499657796208,
-      "grad_norm": 0.005640998482704163,
-      "learning_rate": 0.00019999633975457303,
-      "loss": 46.0,
-      "step": 16922
-    },
-    {
-      "epoch": 2.725311002858408,
-      "grad_norm": 0.003803322557359934,
-      "learning_rate": 0.00019999633932170507,
-      "loss": 46.0,
-      "step": 16923
-    },
-    {
-      "epoch": 2.7254720399371957,
-      "grad_norm": 0.002323080552741885,
-      "learning_rate": 0.00019999633888881152,
-      "loss": 46.0,
-      "step": 16924
-    },
-    {
-      "epoch": 2.725633077015983,
-      "grad_norm": 0.002194333355873823,
-      "learning_rate": 0.0001999963384558924,
-      "loss": 46.0,
-      "step": 16925
-    },
-    {
-      "epoch": 2.7257941140947706,
-      "grad_norm": 0.0053588952869176865,
-      "learning_rate": 0.0001999963380229477,
-      "loss": 46.0,
-      "step": 16926
-    },
-    {
-      "epoch": 2.7259551511735576,
-      "grad_norm": 0.00917847827076912,
-      "learning_rate": 0.00019999633758997738,
-      "loss": 46.0,
-      "step": 16927
-    },
-    {
-      "epoch": 2.726116188252345,
-      "grad_norm": 0.004154964815825224,
-      "learning_rate": 0.00019999633715698145,
-      "loss": 46.0,
-      "step": 16928
-    },
-    {
-      "epoch": 2.7262772253311325,
-      "grad_norm": 0.0015438022091984749,
-      "learning_rate": 0.00019999633672395994,
-      "loss": 46.0,
-      "step": 16929
-    },
-    {
-      "epoch": 2.72643826240992,
-      "grad_norm": 0.009665518067777157,
-      "learning_rate": 0.00019999633629091284,
-      "loss": 46.0,
-      "step": 16930
-    },
-    {
-      "epoch": 2.7265992994887074,
-      "grad_norm": 0.0051448955200612545,
-      "learning_rate": 0.00019999633585784016,
-      "loss": 46.0,
-      "step": 16931
-    },
-    {
-      "epoch": 2.7267603365674944,
-      "grad_norm": 0.0009643063531257212,
-      "learning_rate": 0.00019999633542474188,
-      "loss": 46.0,
-      "step": 16932
-    },
-    {
-      "epoch": 2.726921373646282,
-      "grad_norm": 0.009452329017221928,
-      "learning_rate": 0.000199996334991618,
-      "loss": 46.0,
-      "step": 16933
-    },
-    {
-      "epoch": 2.7270824107250693,
-      "grad_norm": 0.00345798977650702,
-      "learning_rate": 0.00019999633455846852,
-      "loss": 46.0,
-      "step": 16934
-    },
-    {
-      "epoch": 2.7272434478038567,
-      "grad_norm": 0.002070992486551404,
-      "learning_rate": 0.00019999633412529346,
-      "loss": 46.0,
-      "step": 16935
-    },
-    {
-      "epoch": 2.727404484882644,
-      "grad_norm": 0.008523769676685333,
-      "learning_rate": 0.00019999633369209278,
-      "loss": 46.0,
-      "step": 16936
-    },
-    {
-      "epoch": 2.7275655219614316,
-      "grad_norm": 0.001795751741155982,
-      "learning_rate": 0.00019999633325886652,
-      "loss": 46.0,
-      "step": 16937
-    },
-    {
-      "epoch": 2.727726559040219,
-      "grad_norm": 0.00038119166856631637,
-      "learning_rate": 0.0001999963328256147,
-      "loss": 46.0,
-      "step": 16938
-    },
-    {
-      "epoch": 2.7278875961190066,
-      "grad_norm": 0.0009577765013091266,
-      "learning_rate": 0.00019999633239233723,
-      "loss": 46.0,
-      "step": 16939
-    },
-    {
-      "epoch": 2.728048633197794,
-      "grad_norm": 0.005162945948541164,
-      "learning_rate": 0.0001999963319590342,
-      "loss": 46.0,
-      "step": 16940
-    },
-    {
-      "epoch": 2.7282096702765815,
-      "grad_norm": 0.005504915490746498,
-      "learning_rate": 0.00019999633152570557,
-      "loss": 46.0,
-      "step": 16941
-    },
-    {
-      "epoch": 2.7283707073553685,
-      "grad_norm": 0.0016659394605085254,
-      "learning_rate": 0.00019999633109235134,
-      "loss": 46.0,
-      "step": 16942
-    },
-    {
-      "epoch": 2.728531744434156,
-      "grad_norm": 0.0030824693385511637,
-      "learning_rate": 0.00019999633065897153,
-      "loss": 46.0,
-      "step": 16943
-    },
-    {
-      "epoch": 2.7286927815129434,
-      "grad_norm": 0.004563783295452595,
-      "learning_rate": 0.00019999633022556613,
-      "loss": 46.0,
-      "step": 16944
-    },
-    {
-      "epoch": 2.728853818591731,
-      "grad_norm": 0.011392472311854362,
-      "learning_rate": 0.0001999963297921351,
-      "loss": 46.0,
-      "step": 16945
-    },
-    {
-      "epoch": 2.7290148556705183,
-      "grad_norm": 0.00310645648278296,
-      "learning_rate": 0.0001999963293586785,
-      "loss": 46.0,
-      "step": 16946
-    },
-    {
-      "epoch": 2.7291758927493053,
-      "grad_norm": 0.0014629033394157887,
-      "learning_rate": 0.0001999963289251963,
-      "loss": 46.0,
-      "step": 16947
-    },
-    {
-      "epoch": 2.7293369298280927,
-      "grad_norm": 0.005917146801948547,
-      "learning_rate": 0.0001999963284916885,
-      "loss": 46.0,
-      "step": 16948
-    },
-    {
-      "epoch": 2.72949796690688,
-      "grad_norm": 0.0015914080431684852,
-      "learning_rate": 0.00019999632805815513,
-      "loss": 46.0,
-      "step": 16949
-    },
-    {
-      "epoch": 2.7296590039856676,
-      "grad_norm": 0.002734109992161393,
-      "learning_rate": 0.00019999632762459615,
-      "loss": 46.0,
-      "step": 16950
-    },
-    {
-      "epoch": 2.729820041064455,
-      "grad_norm": 0.00970515701919794,
-      "learning_rate": 0.0001999963271910116,
-      "loss": 46.0,
-      "step": 16951
-    },
-    {
-      "epoch": 2.7299810781432425,
-      "grad_norm": 0.006925088819116354,
-      "learning_rate": 0.00019999632675740144,
-      "loss": 46.0,
-      "step": 16952
-    },
-    {
-      "epoch": 2.73014211522203,
-      "grad_norm": 0.002014497760683298,
-      "learning_rate": 0.00019999632632376567,
-      "loss": 46.0,
-      "step": 16953
-    },
-    {
-      "epoch": 2.7303031523008174,
-      "grad_norm": 0.014175421558320522,
-      "learning_rate": 0.00019999632589010432,
-      "loss": 46.0,
-      "step": 16954
-    },
-    {
-      "epoch": 2.730464189379605,
-      "grad_norm": 0.0023810931015759706,
-      "learning_rate": 0.00019999632545641735,
-      "loss": 46.0,
-      "step": 16955
-    },
-    {
-      "epoch": 2.730625226458392,
-      "grad_norm": 0.0023386916145682335,
-      "learning_rate": 0.00019999632502270482,
-      "loss": 46.0,
-      "step": 16956
-    },
-    {
-      "epoch": 2.7307862635371793,
-      "grad_norm": 0.0014831279404461384,
-      "learning_rate": 0.00019999632458896668,
-      "loss": 46.0,
-      "step": 16957
-    },
-    {
-      "epoch": 2.730947300615967,
-      "grad_norm": 0.004147290717810392,
-      "learning_rate": 0.00019999632415520295,
-      "loss": 46.0,
-      "step": 16958
-    },
-    {
-      "epoch": 2.7311083376947543,
-      "grad_norm": 0.004907776601612568,
-      "learning_rate": 0.00019999632372141364,
-      "loss": 46.0,
-      "step": 16959
-    },
-    {
-      "epoch": 2.7312693747735417,
-      "grad_norm": 0.005365044344216585,
-      "learning_rate": 0.00019999632328759873,
-      "loss": 46.0,
-      "step": 16960
-    },
-    {
-      "epoch": 2.731430411852329,
-      "grad_norm": 0.00871161837130785,
-      "learning_rate": 0.0001999963228537582,
-      "loss": 46.0,
-      "step": 16961
-    },
-    {
-      "epoch": 2.731591448931116,
-      "grad_norm": 0.003037975635379553,
-      "learning_rate": 0.0001999963224198921,
-      "loss": 46.0,
-      "step": 16962
-    },
-    {
-      "epoch": 2.7317524860099036,
-      "grad_norm": 0.0018639479530975223,
-      "learning_rate": 0.0001999963219860004,
-      "loss": 46.0,
-      "step": 16963
-    },
-    {
-      "epoch": 2.731913523088691,
-      "grad_norm": 0.013850276358425617,
-      "learning_rate": 0.0001999963215520831,
-      "loss": 46.0,
-      "step": 16964
-    },
-    {
-      "epoch": 2.7320745601674785,
-      "grad_norm": 0.0006425424944609404,
-      "learning_rate": 0.00019999632111814024,
-      "loss": 46.0,
-      "step": 16965
-    },
-    {
-      "epoch": 2.732235597246266,
-      "grad_norm": 0.0076021356508135796,
-      "learning_rate": 0.00019999632068417174,
-      "loss": 46.0,
-      "step": 16966
-    },
-    {
-      "epoch": 2.7323966343250534,
-      "grad_norm": 0.002191032748669386,
-      "learning_rate": 0.00019999632025017767,
-      "loss": 46.0,
-      "step": 16967
-    },
-    {
-      "epoch": 2.732557671403841,
-      "grad_norm": 0.0030889525078237057,
-      "learning_rate": 0.00019999631981615801,
-      "loss": 46.0,
-      "step": 16968
-    },
-    {
-      "epoch": 2.7327187084826283,
-      "grad_norm": 0.0007897743489593267,
-      "learning_rate": 0.00019999631938211274,
-      "loss": 46.0,
-      "step": 16969
-    },
-    {
-      "epoch": 2.7328797455614158,
-      "grad_norm": 0.010255039669573307,
-      "learning_rate": 0.00019999631894804186,
-      "loss": 46.0,
-      "step": 16970
-    },
-    {
-      "epoch": 2.733040782640203,
-      "grad_norm": 0.0007939495262689888,
-      "learning_rate": 0.00019999631851394542,
-      "loss": 46.0,
-      "step": 16971
-    },
-    {
-      "epoch": 2.7332018197189902,
-      "grad_norm": 0.0021262671798467636,
-      "learning_rate": 0.00019999631807982339,
-      "loss": 46.0,
-      "step": 16972
-    },
-    {
-      "epoch": 2.7333628567977777,
-      "grad_norm": 0.0038738714065402746,
-      "learning_rate": 0.00019999631764567574,
-      "loss": 46.0,
-      "step": 16973
-    },
-    {
-      "epoch": 2.733523893876565,
-      "grad_norm": 0.009010964073240757,
-      "learning_rate": 0.0001999963172115025,
-      "loss": 46.0,
-      "step": 16974
-    },
-    {
-      "epoch": 2.7336849309553526,
-      "grad_norm": 0.0008244227501563728,
-      "learning_rate": 0.00019999631677730372,
-      "loss": 46.0,
-      "step": 16975
-    },
-    {
-      "epoch": 2.7338459680341396,
-      "grad_norm": 0.005699793342500925,
-      "learning_rate": 0.00019999631634307928,
-      "loss": 46.0,
-      "step": 16976
-    },
-    {
-      "epoch": 2.734007005112927,
-      "grad_norm": 0.011394317261874676,
-      "learning_rate": 0.00019999631590882926,
-      "loss": 46.0,
-      "step": 16977
-    },
-    {
-      "epoch": 2.7341680421917145,
-      "grad_norm": 0.0070154257118701935,
-      "learning_rate": 0.00019999631547455365,
-      "loss": 46.0,
-      "step": 16978
-    },
-    {
-      "epoch": 2.734329079270502,
-      "grad_norm": 0.0019240652909502387,
-      "learning_rate": 0.00019999631504025246,
-      "loss": 46.0,
-      "step": 16979
-    },
-    {
-      "epoch": 2.7344901163492894,
-      "grad_norm": 0.018269382417201996,
-      "learning_rate": 0.00019999631460592565,
-      "loss": 46.0,
-      "step": 16980
-    },
-    {
-      "epoch": 2.734651153428077,
-      "grad_norm": 0.0011777982581406832,
-      "learning_rate": 0.00019999631417157328,
-      "loss": 46.0,
-      "step": 16981
-    },
-    {
-      "epoch": 2.7348121905068643,
-      "grad_norm": 0.006356328725814819,
-      "learning_rate": 0.0001999963137371953,
-      "loss": 46.0,
-      "step": 16982
-    },
-    {
-      "epoch": 2.7349732275856518,
-      "grad_norm": 0.0010452274000272155,
-      "learning_rate": 0.00019999631330279172,
-      "loss": 46.0,
-      "step": 16983
-    },
-    {
-      "epoch": 2.735134264664439,
-      "grad_norm": 0.002229926409199834,
-      "learning_rate": 0.00019999631286836254,
-      "loss": 46.0,
-      "step": 16984
-    },
-    {
-      "epoch": 2.735295301743226,
-      "grad_norm": 0.0020133969374001026,
-      "learning_rate": 0.0001999963124339078,
-      "loss": 46.0,
-      "step": 16985
-    },
-    {
-      "epoch": 2.7354563388220137,
-      "grad_norm": 0.0017598966369405389,
-      "learning_rate": 0.00019999631199942743,
-      "loss": 46.0,
-      "step": 16986
-    },
-    {
-      "epoch": 2.735617375900801,
-      "grad_norm": 0.008132735267281532,
-      "learning_rate": 0.00019999631156492146,
-      "loss": 46.0,
-      "step": 16987
-    },
-    {
-      "epoch": 2.7357784129795886,
-      "grad_norm": 0.006711173802614212,
-      "learning_rate": 0.00019999631113038993,
-      "loss": 46.0,
-      "step": 16988
-    },
-    {
-      "epoch": 2.735939450058376,
-      "grad_norm": 0.0009318398078903556,
-      "learning_rate": 0.00019999631069583278,
-      "loss": 46.0,
-      "step": 16989
-    },
-    {
-      "epoch": 2.7361004871371635,
-      "grad_norm": 0.0020174412056803703,
-      "learning_rate": 0.00019999631026125004,
-      "loss": 46.0,
-      "step": 16990
-    },
-    {
-      "epoch": 2.7362615242159505,
-      "grad_norm": 0.007003118749707937,
-      "learning_rate": 0.00019999630982664172,
-      "loss": 46.0,
-      "step": 16991
-    },
-    {
-      "epoch": 2.736422561294738,
-      "grad_norm": 0.0023433587048202753,
-      "learning_rate": 0.0001999963093920078,
-      "loss": 46.0,
-      "step": 16992
-    },
-    {
-      "epoch": 2.7365835983735254,
-      "grad_norm": 0.0013160071102902293,
-      "learning_rate": 0.00019999630895734829,
-      "loss": 46.0,
-      "step": 16993
-    },
-    {
-      "epoch": 2.736744635452313,
-      "grad_norm": 0.0037288167513906956,
-      "learning_rate": 0.00019999630852266317,
-      "loss": 46.0,
-      "step": 16994
-    },
-    {
-      "epoch": 2.7369056725311003,
-      "grad_norm": 0.003628902602940798,
-      "learning_rate": 0.00019999630808795248,
-      "loss": 46.0,
-      "step": 16995
-    },
-    {
-      "epoch": 2.7370667096098877,
-      "grad_norm": 0.00916481763124466,
-      "learning_rate": 0.00019999630765321616,
-      "loss": 46.0,
-      "step": 16996
-    },
-    {
-      "epoch": 2.737227746688675,
-      "grad_norm": 0.0007951876032166183,
-      "learning_rate": 0.0001999963072184543,
-      "loss": 46.0,
-      "step": 16997
-    },
-    {
-      "epoch": 2.7373887837674626,
-      "grad_norm": 0.0014403134118765593,
-      "learning_rate": 0.0001999963067836668,
-      "loss": 46.0,
-      "step": 16998
-    },
-    {
-      "epoch": 2.73754982084625,
-      "grad_norm": 0.007754639256745577,
-      "learning_rate": 0.0001999963063488537,
-      "loss": 46.0,
-      "step": 16999
-    },
-    {
-      "epoch": 2.737710857925037,
-      "grad_norm": 0.004751814529299736,
-      "learning_rate": 0.00019999630591401504,
-      "loss": 46.0,
-      "step": 17000
-    },
-    {
-      "epoch": 2.7378718950038246,
-      "grad_norm": 0.019630469381809235,
-      "learning_rate": 0.00019999630547915076,
-      "loss": 46.0,
-      "step": 17001
-    },
-    {
-      "epoch": 2.738032932082612,
-      "grad_norm": 0.0025465500075370073,
-      "learning_rate": 0.0001999963050442609,
-      "loss": 46.0,
-      "step": 17002
-    },
-    {
-      "epoch": 2.7381939691613995,
-      "grad_norm": 0.00048430985771119595,
-      "learning_rate": 0.00019999630460934545,
-      "loss": 46.0,
-      "step": 17003
-    },
-    {
-      "epoch": 2.738355006240187,
-      "grad_norm": 0.0027142332401126623,
-      "learning_rate": 0.00019999630417440441,
-      "loss": 46.0,
-      "step": 17004
-    },
-    {
-      "epoch": 2.7385160433189744,
-      "grad_norm": 0.007888139225542545,
-      "learning_rate": 0.00019999630373943776,
-      "loss": 46.0,
-      "step": 17005
-    },
-    {
-      "epoch": 2.7386770803977614,
-      "grad_norm": 0.000852732453495264,
-      "learning_rate": 0.00019999630330444555,
-      "loss": 46.0,
-      "step": 17006
-    },
-    {
-      "epoch": 2.738838117476549,
-      "grad_norm": 0.000506145937833935,
-      "learning_rate": 0.0001999963028694277,
-      "loss": 46.0,
-      "step": 17007
-    },
-    {
-      "epoch": 2.7389991545553363,
-      "grad_norm": 0.0009367187740281224,
-      "learning_rate": 0.00019999630243438429,
-      "loss": 46.0,
-      "step": 17008
-    },
-    {
-      "epoch": 2.7391601916341237,
-      "grad_norm": 0.00575898727402091,
-      "learning_rate": 0.00019999630199931526,
-      "loss": 46.0,
-      "step": 17009
-    },
-    {
-      "epoch": 2.739321228712911,
-      "grad_norm": 0.0005599577561952174,
-      "learning_rate": 0.00019999630156422064,
-      "loss": 46.0,
-      "step": 17010
-    },
-    {
-      "epoch": 2.7394822657916986,
-      "grad_norm": 0.0029721814207732677,
-      "learning_rate": 0.00019999630112910044,
-      "loss": 46.0,
-      "step": 17011
-    },
-    {
-      "epoch": 2.739643302870486,
-      "grad_norm": 0.007054353132843971,
-      "learning_rate": 0.00019999630069395463,
-      "loss": 46.0,
-      "step": 17012
-    },
-    {
-      "epoch": 2.7398043399492735,
-      "grad_norm": 0.0020257323049008846,
-      "learning_rate": 0.00019999630025878325,
-      "loss": 46.0,
-      "step": 17013
-    },
-    {
-      "epoch": 2.739965377028061,
-      "grad_norm": 0.004529378842562437,
-      "learning_rate": 0.00019999629982358626,
-      "loss": 46.0,
-      "step": 17014
-    },
-    {
-      "epoch": 2.740126414106848,
-      "grad_norm": 0.0008891024626791477,
-      "learning_rate": 0.00019999629938836368,
-      "loss": 46.0,
-      "step": 17015
-    },
-    {
-      "epoch": 2.7402874511856354,
-      "grad_norm": 0.0015078552532941103,
-      "learning_rate": 0.0001999962989531155,
-      "loss": 46.0,
-      "step": 17016
-    },
-    {
-      "epoch": 2.740448488264423,
-      "grad_norm": 0.0031485690269619226,
-      "learning_rate": 0.0001999962985178417,
-      "loss": 46.0,
-      "step": 17017
-    },
-    {
-      "epoch": 2.7406095253432103,
-      "grad_norm": 0.0015434781089425087,
-      "learning_rate": 0.00019999629808254237,
-      "loss": 46.0,
-      "step": 17018
-    },
-    {
-      "epoch": 2.740770562421998,
-      "grad_norm": 0.0034655025228857994,
-      "learning_rate": 0.0001999962976472174,
-      "loss": 46.0,
-      "step": 17019
-    },
-    {
-      "epoch": 2.740931599500785,
-      "grad_norm": 0.0029644996393471956,
-      "learning_rate": 0.00019999629721186683,
-      "loss": 46.0,
-      "step": 17020
-    },
-    {
-      "epoch": 2.7410926365795723,
-      "grad_norm": 0.005331673193722963,
-      "learning_rate": 0.0001999962967764907,
-      "loss": 46.0,
-      "step": 17021
-    },
-    {
-      "epoch": 2.7412536736583597,
-      "grad_norm": 0.0036207479424774647,
-      "learning_rate": 0.00019999629634108896,
-      "loss": 46.0,
-      "step": 17022
-    },
-    {
-      "epoch": 2.741414710737147,
-      "grad_norm": 0.0013629754539579153,
-      "learning_rate": 0.00019999629590566166,
-      "loss": 46.0,
-      "step": 17023
-    },
-    {
-      "epoch": 2.7415757478159346,
-      "grad_norm": 0.004165869206190109,
-      "learning_rate": 0.0001999962954702087,
-      "loss": 46.0,
-      "step": 17024
-    },
-    {
-      "epoch": 2.741736784894722,
-      "grad_norm": 0.0029115048237144947,
-      "learning_rate": 0.00019999629503473018,
-      "loss": 46.0,
-      "step": 17025
-    },
-    {
-      "epoch": 2.7418978219735095,
-      "grad_norm": 0.0013682834105566144,
-      "learning_rate": 0.00019999629459922606,
-      "loss": 46.0,
-      "step": 17026
-    },
-    {
-      "epoch": 2.742058859052297,
-      "grad_norm": 0.0060378173366189,
-      "learning_rate": 0.00019999629416369636,
-      "loss": 46.0,
-      "step": 17027
-    },
-    {
-      "epoch": 2.7422198961310844,
-      "grad_norm": 0.002196847926825285,
-      "learning_rate": 0.00019999629372814104,
-      "loss": 46.0,
-      "step": 17028
-    },
-    {
-      "epoch": 2.7423809332098714,
-      "grad_norm": 0.0008671838440932333,
-      "learning_rate": 0.00019999629329256016,
-      "loss": 46.0,
-      "step": 17029
-    },
-    {
-      "epoch": 2.742541970288659,
-      "grad_norm": 0.0036622064653784037,
-      "learning_rate": 0.00019999629285695367,
-      "loss": 46.0,
-      "step": 17030
-    },
-    {
-      "epoch": 2.7427030073674463,
-      "grad_norm": 0.0011343695223331451,
-      "learning_rate": 0.0001999962924213216,
-      "loss": 46.0,
-      "step": 17031
-    },
-    {
-      "epoch": 2.7428640444462338,
-      "grad_norm": 0.0010958018247038126,
-      "learning_rate": 0.0001999962919856639,
-      "loss": 46.0,
-      "step": 17032
-    },
-    {
-      "epoch": 2.7430250815250212,
-      "grad_norm": 0.001842381781898439,
-      "learning_rate": 0.00019999629154998064,
-      "loss": 46.0,
-      "step": 17033
-    },
-    {
-      "epoch": 2.7431861186038087,
-      "grad_norm": 0.0020009654108434916,
-      "learning_rate": 0.00019999629111427177,
-      "loss": 46.0,
-      "step": 17034
-    },
-    {
-      "epoch": 2.7433471556825957,
-      "grad_norm": 0.0005218570586293936,
-      "learning_rate": 0.0001999962906785373,
-      "loss": 46.0,
-      "step": 17035
-    },
-    {
-      "epoch": 2.743508192761383,
-      "grad_norm": 0.002518236171454191,
-      "learning_rate": 0.00019999629024277727,
-      "loss": 46.0,
-      "step": 17036
-    },
-    {
-      "epoch": 2.7436692298401706,
-      "grad_norm": 0.007329927291721106,
-      "learning_rate": 0.00019999628980699158,
-      "loss": 46.0,
-      "step": 17037
-    },
-    {
-      "epoch": 2.743830266918958,
-      "grad_norm": 0.001647655968554318,
-      "learning_rate": 0.00019999628937118037,
-      "loss": 46.0,
-      "step": 17038
-    },
-    {
-      "epoch": 2.7439913039977455,
-      "grad_norm": 0.00754706934094429,
-      "learning_rate": 0.00019999628893534353,
-      "loss": 46.0,
-      "step": 17039
-    },
-    {
-      "epoch": 2.744152341076533,
-      "grad_norm": 0.002074816031381488,
-      "learning_rate": 0.0001999962884994811,
-      "loss": 46.0,
-      "step": 17040
-    },
-    {
-      "epoch": 2.7443133781553204,
-      "grad_norm": 0.006157160736620426,
-      "learning_rate": 0.00019999628806359305,
-      "loss": 46.0,
-      "step": 17041
-    },
-    {
-      "epoch": 2.744474415234108,
-      "grad_norm": 0.0017099973047152162,
-      "learning_rate": 0.00019999628762767943,
-      "loss": 46.0,
-      "step": 17042
-    },
-    {
-      "epoch": 2.7446354523128953,
-      "grad_norm": 0.002026587026193738,
-      "learning_rate": 0.00019999628719174025,
-      "loss": 46.0,
-      "step": 17043
-    },
-    {
-      "epoch": 2.7447964893916823,
-      "grad_norm": 0.0036466491874307394,
-      "learning_rate": 0.00019999628675577543,
-      "loss": 46.0,
-      "step": 17044
-    },
-    {
-      "epoch": 2.7449575264704698,
-      "grad_norm": 0.0018727057613432407,
-      "learning_rate": 0.00019999628631978502,
-      "loss": 46.0,
-      "step": 17045
-    },
-    {
-      "epoch": 2.745118563549257,
-      "grad_norm": 0.0014251800021156669,
-      "learning_rate": 0.00019999628588376905,
-      "loss": 46.0,
-      "step": 17046
-    },
-    {
-      "epoch": 2.7452796006280447,
-      "grad_norm": 0.0037118534091860056,
-      "learning_rate": 0.00019999628544772744,
-      "loss": 46.0,
-      "step": 17047
-    },
-    {
-      "epoch": 2.745440637706832,
-      "grad_norm": 0.007601283490657806,
-      "learning_rate": 0.00019999628501166027,
-      "loss": 46.0,
-      "step": 17048
-    },
-    {
-      "epoch": 2.745601674785619,
-      "grad_norm": 0.004464562050998211,
-      "learning_rate": 0.00019999628457556748,
-      "loss": 46.0,
-      "step": 17049
-    },
-    {
-      "epoch": 2.7457627118644066,
-      "grad_norm": 0.008064158260822296,
-      "learning_rate": 0.0001999962841394491,
-      "loss": 46.0,
-      "step": 17050
-    },
-    {
-      "epoch": 2.745923748943194,
-      "grad_norm": 0.0014315111329779029,
-      "learning_rate": 0.00019999628370330515,
-      "loss": 46.0,
-      "step": 17051
-    },
-    {
-      "epoch": 2.7460847860219815,
-      "grad_norm": 0.0015736041823402047,
-      "learning_rate": 0.0001999962832671356,
-      "loss": 46.0,
-      "step": 17052
-    },
-    {
-      "epoch": 2.746245823100769,
-      "grad_norm": 0.005847330670803785,
-      "learning_rate": 0.00019999628283094044,
-      "loss": 46.0,
-      "step": 17053
-    },
-    {
-      "epoch": 2.7464068601795564,
-      "grad_norm": 0.004249951336532831,
-      "learning_rate": 0.0001999962823947197,
-      "loss": 46.0,
-      "step": 17054
-    },
-    {
-      "epoch": 2.746567897258344,
-      "grad_norm": 0.0013741877628490329,
-      "learning_rate": 0.00019999628195847336,
-      "loss": 46.0,
-      "step": 17055
-    },
-    {
-      "epoch": 2.7467289343371313,
-      "grad_norm": 0.0016543520614504814,
-      "learning_rate": 0.00019999628152220144,
-      "loss": 46.0,
-      "step": 17056
-    },
-    {
-      "epoch": 2.7468899714159187,
-      "grad_norm": 0.005567203741520643,
-      "learning_rate": 0.0001999962810859039,
-      "loss": 46.0,
-      "step": 17057
-    },
-    {
-      "epoch": 2.747051008494706,
-      "grad_norm": 0.0024511455558240414,
-      "learning_rate": 0.00019999628064958078,
-      "loss": 46.0,
-      "step": 17058
-    },
-    {
-      "epoch": 2.747212045573493,
-      "grad_norm": 0.0012431987561285496,
-      "learning_rate": 0.00019999628021323207,
-      "loss": 46.0,
-      "step": 17059
-    },
-    {
-      "epoch": 2.7473730826522806,
-      "grad_norm": 0.001957609783858061,
-      "learning_rate": 0.00019999627977685774,
-      "loss": 46.0,
-      "step": 17060
-    },
-    {
-      "epoch": 2.747534119731068,
-      "grad_norm": 0.00042120725265704095,
-      "learning_rate": 0.00019999627934045786,
-      "loss": 46.0,
-      "step": 17061
-    },
-    {
-      "epoch": 2.7476951568098555,
-      "grad_norm": 0.007067305967211723,
-      "learning_rate": 0.00019999627890403236,
-      "loss": 46.0,
-      "step": 17062
-    },
-    {
-      "epoch": 2.747856193888643,
-      "grad_norm": 0.0051729436963796616,
-      "learning_rate": 0.00019999627846758127,
-      "loss": 46.0,
-      "step": 17063
-    },
-    {
-      "epoch": 2.74801723096743,
-      "grad_norm": 0.007195282261818647,
-      "learning_rate": 0.0001999962780311046,
-      "loss": 46.0,
-      "step": 17064
-    },
-    {
-      "epoch": 2.7481782680462175,
-      "grad_norm": 0.0033652957063168287,
-      "learning_rate": 0.0001999962775946023,
-      "loss": 46.0,
-      "step": 17065
-    },
-    {
-      "epoch": 2.748339305125005,
-      "grad_norm": 0.004628072027117014,
-      "learning_rate": 0.00019999627715807443,
-      "loss": 46.0,
-      "step": 17066
-    },
-    {
-      "epoch": 2.7485003422037924,
-      "grad_norm": 0.009264047257602215,
-      "learning_rate": 0.00019999627672152097,
-      "loss": 46.0,
-      "step": 17067
-    },
-    {
-      "epoch": 2.74866137928258,
-      "grad_norm": 0.005872600711882114,
-      "learning_rate": 0.00019999627628494192,
-      "loss": 46.0,
-      "step": 17068
-    },
-    {
-      "epoch": 2.7488224163613673,
-      "grad_norm": 0.0036728999111801386,
-      "learning_rate": 0.00019999627584833726,
-      "loss": 46.0,
-      "step": 17069
-    },
-    {
-      "epoch": 2.7489834534401547,
-      "grad_norm": 0.0038058655336499214,
-      "learning_rate": 0.00019999627541170703,
-      "loss": 46.0,
-      "step": 17070
-    },
-    {
-      "epoch": 2.749144490518942,
-      "grad_norm": 0.008635470643639565,
-      "learning_rate": 0.00019999627497505117,
-      "loss": 46.0,
-      "step": 17071
-    },
-    {
-      "epoch": 2.7493055275977296,
-      "grad_norm": 0.0016123350942507386,
-      "learning_rate": 0.00019999627453836974,
-      "loss": 46.0,
-      "step": 17072
-    },
-    {
-      "epoch": 2.7494665646765166,
-      "grad_norm": 0.0011007452849298716,
-      "learning_rate": 0.0001999962741016627,
-      "loss": 46.0,
-      "step": 17073
-    },
-    {
-      "epoch": 2.749627601755304,
-      "grad_norm": 0.0030108599457889795,
-      "learning_rate": 0.00019999627366493007,
-      "loss": 46.0,
-      "step": 17074
-    },
-    {
-      "epoch": 2.7497886388340915,
-      "grad_norm": 0.005482657812535763,
-      "learning_rate": 0.00019999627322817186,
-      "loss": 46.0,
-      "step": 17075
-    },
-    {
-      "epoch": 2.749949675912879,
-      "grad_norm": 0.00085295713506639,
-      "learning_rate": 0.00019999627279138803,
-      "loss": 46.0,
-      "step": 17076
-    },
-    {
-      "epoch": 2.7501107129916664,
-      "grad_norm": 0.01183785405009985,
-      "learning_rate": 0.00019999627235457864,
-      "loss": 46.0,
-      "step": 17077
-    },
-    {
-      "epoch": 2.750271750070454,
-      "grad_norm": 0.004167368169873953,
-      "learning_rate": 0.00019999627191774364,
-      "loss": 46.0,
-      "step": 17078
-    },
-    {
-      "epoch": 2.750432787149241,
-      "grad_norm": 0.0018456553807482123,
-      "learning_rate": 0.00019999627148088305,
-      "loss": 46.0,
-      "step": 17079
-    },
-    {
-      "epoch": 2.7505938242280283,
-      "grad_norm": 0.00039835498319007456,
-      "learning_rate": 0.00019999627104399687,
-      "loss": 46.0,
-      "step": 17080
-    },
-    {
-      "epoch": 2.750754861306816,
-      "grad_norm": 0.0021245498210191727,
-      "learning_rate": 0.00019999627060708508,
-      "loss": 46.0,
-      "step": 17081
-    },
-    {
-      "epoch": 2.7509158983856032,
-      "grad_norm": 0.0012092370307072997,
-      "learning_rate": 0.0001999962701701477,
-      "loss": 46.0,
-      "step": 17082
-    },
-    {
-      "epoch": 2.7510769354643907,
-      "grad_norm": 0.00638191681355238,
-      "learning_rate": 0.00019999626973318474,
-      "loss": 46.0,
-      "step": 17083
-    },
-    {
-      "epoch": 2.751237972543178,
-      "grad_norm": 0.005046867299824953,
-      "learning_rate": 0.00019999626929619618,
-      "loss": 46.0,
-      "step": 17084
-    },
-    {
-      "epoch": 2.7513990096219656,
-      "grad_norm": 0.003371955594047904,
-      "learning_rate": 0.00019999626885918202,
-      "loss": 46.0,
-      "step": 17085
-    },
-    {
-      "epoch": 2.751560046700753,
-      "grad_norm": 0.006602564360946417,
-      "learning_rate": 0.00019999626842214226,
-      "loss": 46.0,
-      "step": 17086
-    },
-    {
-      "epoch": 2.7517210837795405,
-      "grad_norm": 0.002260745968669653,
-      "learning_rate": 0.00019999626798507692,
-      "loss": 46.0,
-      "step": 17087
-    },
-    {
-      "epoch": 2.7518821208583275,
-      "grad_norm": 0.0013032500864937901,
-      "learning_rate": 0.000199996267547986,
-      "loss": 46.0,
-      "step": 17088
-    },
-    {
-      "epoch": 2.752043157937115,
-      "grad_norm": 0.0014441004022955894,
-      "learning_rate": 0.00019999626711086945,
-      "loss": 46.0,
-      "step": 17089
-    },
-    {
-      "epoch": 2.7522041950159024,
-      "grad_norm": 0.008795741014182568,
-      "learning_rate": 0.00019999626667372732,
-      "loss": 46.0,
-      "step": 17090
-    },
-    {
-      "epoch": 2.75236523209469,
-      "grad_norm": 0.0006345787551254034,
-      "learning_rate": 0.00019999626623655963,
-      "loss": 46.0,
-      "step": 17091
-    },
-    {
-      "epoch": 2.7525262691734773,
-      "grad_norm": 0.0006841893773525953,
-      "learning_rate": 0.0001999962657993663,
-      "loss": 46.0,
-      "step": 17092
-    },
-    {
-      "epoch": 2.7526873062522643,
-      "grad_norm": 0.00434846943244338,
-      "learning_rate": 0.0001999962653621474,
-      "loss": 46.0,
-      "step": 17093
-    },
-    {
-      "epoch": 2.7528483433310518,
-      "grad_norm": 0.001952580758370459,
-      "learning_rate": 0.0001999962649249029,
-      "loss": 46.0,
-      "step": 17094
-    },
-    {
-      "epoch": 2.7530093804098392,
-      "grad_norm": 0.0019157506758347154,
-      "learning_rate": 0.0001999962644876328,
-      "loss": 46.0,
-      "step": 17095
-    },
-    {
-      "epoch": 2.7531704174886267,
-      "grad_norm": 0.0026203191373497248,
-      "learning_rate": 0.0001999962640503371,
-      "loss": 46.0,
-      "step": 17096
-    },
-    {
-      "epoch": 2.753331454567414,
-      "grad_norm": 0.007479417137801647,
-      "learning_rate": 0.00019999626361301583,
-      "loss": 46.0,
-      "step": 17097
-    },
-    {
-      "epoch": 2.7534924916462016,
-      "grad_norm": 0.002465332392603159,
-      "learning_rate": 0.00019999626317566892,
-      "loss": 46.0,
-      "step": 17098
-    },
-    {
-      "epoch": 2.753653528724989,
-      "grad_norm": 0.0036086381878703833,
-      "learning_rate": 0.00019999626273829648,
-      "loss": 46.0,
-      "step": 17099
-    },
-    {
-      "epoch": 2.7538145658037765,
-      "grad_norm": 0.0011274574790149927,
-      "learning_rate": 0.0001999962623008984,
-      "loss": 46.0,
-      "step": 17100
-    },
-    {
-      "epoch": 2.753975602882564,
-      "grad_norm": 0.00920192338526249,
-      "learning_rate": 0.00019999626186347475,
-      "loss": 46.0,
-      "step": 17101
-    },
-    {
-      "epoch": 2.754136639961351,
-      "grad_norm": 0.0008029105956666172,
-      "learning_rate": 0.00019999626142602547,
-      "loss": 46.0,
-      "step": 17102
-    },
-    {
-      "epoch": 2.7542976770401384,
-      "grad_norm": 0.0016800423618406057,
-      "learning_rate": 0.00019999626098855065,
-      "loss": 46.0,
-      "step": 17103
-    },
-    {
-      "epoch": 2.754458714118926,
-      "grad_norm": 0.0028716272208839655,
-      "learning_rate": 0.00019999626055105022,
-      "loss": 46.0,
-      "step": 17104
-    },
-    {
-      "epoch": 2.7546197511977133,
-      "grad_norm": 0.0016534305177628994,
-      "learning_rate": 0.00019999626011352417,
-      "loss": 46.0,
-      "step": 17105
-    },
-    {
-      "epoch": 2.7547807882765007,
-      "grad_norm": 0.002045708242803812,
-      "learning_rate": 0.00019999625967597254,
-      "loss": 46.0,
-      "step": 17106
-    },
-    {
-      "epoch": 2.754941825355288,
-      "grad_norm": 0.008509145118296146,
-      "learning_rate": 0.00019999625923839532,
-      "loss": 46.0,
-      "step": 17107
-    },
-    {
-      "epoch": 2.755102862434075,
-      "grad_norm": 0.0014312396524474025,
-      "learning_rate": 0.0001999962588007925,
-      "loss": 46.0,
-      "step": 17108
-    },
-    {
-      "epoch": 2.7552638995128627,
-      "grad_norm": 0.0030597897712141275,
-      "learning_rate": 0.0001999962583631641,
-      "loss": 46.0,
-      "step": 17109
-    },
-    {
-      "epoch": 2.75542493659165,
-      "grad_norm": 0.0068954406306147575,
-      "learning_rate": 0.00019999625792551008,
-      "loss": 46.0,
-      "step": 17110
-    },
-    {
-      "epoch": 2.7555859736704376,
-      "grad_norm": 0.003485912922769785,
-      "learning_rate": 0.00019999625748783048,
-      "loss": 46.0,
-      "step": 17111
-    },
-    {
-      "epoch": 2.755747010749225,
-      "grad_norm": 0.011335739865899086,
-      "learning_rate": 0.0001999962570501253,
-      "loss": 46.0,
-      "step": 17112
-    },
-    {
-      "epoch": 2.7559080478280125,
-      "grad_norm": 0.004586663097143173,
-      "learning_rate": 0.00019999625661239453,
-      "loss": 46.0,
-      "step": 17113
-    },
-    {
-      "epoch": 2.7560690849068,
-      "grad_norm": 0.011939934454858303,
-      "learning_rate": 0.00019999625617463812,
-      "loss": 46.0,
-      "step": 17114
-    },
-    {
-      "epoch": 2.7562301219855874,
-      "grad_norm": 0.003280117642134428,
-      "learning_rate": 0.00019999625573685617,
-      "loss": 46.0,
-      "step": 17115
-    },
-    {
-      "epoch": 2.756391159064375,
-      "grad_norm": 0.02143050730228424,
-      "learning_rate": 0.00019999625529904859,
-      "loss": 46.0,
-      "step": 17116
-    },
-    {
-      "epoch": 2.756552196143162,
-      "grad_norm": 0.0033806515857577324,
-      "learning_rate": 0.00019999625486121544,
-      "loss": 46.0,
-      "step": 17117
-    },
-    {
-      "epoch": 2.7567132332219493,
-      "grad_norm": 0.002159483265131712,
-      "learning_rate": 0.00019999625442335668,
-      "loss": 46.0,
-      "step": 17118
-    },
-    {
-      "epoch": 2.7568742703007367,
-      "grad_norm": 0.0027408525347709656,
-      "learning_rate": 0.00019999625398547233,
-      "loss": 46.0,
-      "step": 17119
-    },
-    {
-      "epoch": 2.757035307379524,
-      "grad_norm": 0.005222344771027565,
-      "learning_rate": 0.00019999625354756237,
-      "loss": 46.0,
-      "step": 17120
-    },
-    {
-      "epoch": 2.7571963444583116,
-      "grad_norm": 0.0006948627415113151,
-      "learning_rate": 0.00019999625310962685,
-      "loss": 46.0,
-      "step": 17121
-    },
-    {
-      "epoch": 2.7573573815370986,
-      "grad_norm": 0.0007239735568873584,
-      "learning_rate": 0.0001999962526716657,
-      "loss": 46.0,
-      "step": 17122
-    },
-    {
-      "epoch": 2.757518418615886,
-      "grad_norm": 0.005654027685523033,
-      "learning_rate": 0.00019999625223367899,
-      "loss": 46.0,
-      "step": 17123
-    },
-    {
-      "epoch": 2.7576794556946735,
-      "grad_norm": 0.005968195386230946,
-      "learning_rate": 0.00019999625179566665,
-      "loss": 46.0,
-      "step": 17124
-    },
-    {
-      "epoch": 2.757840492773461,
-      "grad_norm": 0.0015129344537854195,
-      "learning_rate": 0.00019999625135762875,
-      "loss": 46.0,
-      "step": 17125
-    },
-    {
-      "epoch": 2.7580015298522484,
-      "grad_norm": 0.007854820229113102,
-      "learning_rate": 0.00019999625091956524,
-      "loss": 46.0,
-      "step": 17126
-    },
-    {
-      "epoch": 2.758162566931036,
-      "grad_norm": 0.0008547230972908437,
-      "learning_rate": 0.00019999625048147614,
-      "loss": 46.0,
-      "step": 17127
-    },
-    {
-      "epoch": 2.7583236040098233,
-      "grad_norm": 0.0034407672937959433,
-      "learning_rate": 0.00019999625004336145,
-      "loss": 46.0,
-      "step": 17128
-    },
-    {
-      "epoch": 2.758484641088611,
-      "grad_norm": 0.0030282388906925917,
-      "learning_rate": 0.00019999624960522118,
-      "loss": 46.0,
-      "step": 17129
-    },
-    {
-      "epoch": 2.7586456781673983,
-      "grad_norm": 0.006146104075014591,
-      "learning_rate": 0.0001999962491670553,
-      "loss": 46.0,
-      "step": 17130
-    },
-    {
-      "epoch": 2.7588067152461857,
-      "grad_norm": 0.0004937821649946272,
-      "learning_rate": 0.0001999962487288638,
-      "loss": 46.0,
-      "step": 17131
-    },
-    {
-      "epoch": 2.7589677523249727,
-      "grad_norm": 0.007552339229732752,
-      "learning_rate": 0.00019999624829064672,
-      "loss": 46.0,
-      "step": 17132
-    },
-    {
-      "epoch": 2.75912878940376,
-      "grad_norm": 0.005795442499220371,
-      "learning_rate": 0.00019999624785240405,
-      "loss": 46.0,
-      "step": 17133
-    },
-    {
-      "epoch": 2.7592898264825476,
-      "grad_norm": 0.00309941079467535,
-      "learning_rate": 0.0001999962474141358,
-      "loss": 46.0,
-      "step": 17134
-    },
-    {
-      "epoch": 2.759450863561335,
-      "grad_norm": 0.007975741289556026,
-      "learning_rate": 0.00019999624697584196,
-      "loss": 46.0,
-      "step": 17135
-    },
-    {
-      "epoch": 2.7596119006401225,
-      "grad_norm": 0.001728990231640637,
-      "learning_rate": 0.00019999624653752252,
-      "loss": 46.0,
-      "step": 17136
-    },
-    {
-      "epoch": 2.7597729377189095,
-      "grad_norm": 0.004385133273899555,
-      "learning_rate": 0.00019999624609917747,
-      "loss": 46.0,
-      "step": 17137
-    },
-    {
-      "epoch": 2.759933974797697,
-      "grad_norm": 0.0034701130352914333,
-      "learning_rate": 0.00019999624566080683,
-      "loss": 46.0,
-      "step": 17138
-    },
-    {
-      "epoch": 2.7600950118764844,
-      "grad_norm": 0.0009467012714594603,
-      "learning_rate": 0.0001999962452224106,
-      "loss": 46.0,
-      "step": 17139
-    },
-    {
-      "epoch": 2.760256048955272,
-      "grad_norm": 0.008176835253834724,
-      "learning_rate": 0.0001999962447839888,
-      "loss": 46.0,
-      "step": 17140
-    },
-    {
-      "epoch": 2.7604170860340593,
-      "grad_norm": 0.009744575247168541,
-      "learning_rate": 0.0001999962443455414,
-      "loss": 46.0,
-      "step": 17141
-    },
-    {
-      "epoch": 2.760578123112847,
-      "grad_norm": 0.0012876407708972692,
-      "learning_rate": 0.00019999624390706834,
-      "loss": 46.0,
-      "step": 17142
-    },
-    {
-      "epoch": 2.7607391601916342,
-      "grad_norm": 0.0009183231159113348,
-      "learning_rate": 0.00019999624346856974,
-      "loss": 46.0,
-      "step": 17143
-    },
-    {
-      "epoch": 2.7609001972704217,
-      "grad_norm": 0.002574438927695155,
-      "learning_rate": 0.00019999624303004555,
-      "loss": 46.0,
-      "step": 17144
-    },
-    {
-      "epoch": 2.761061234349209,
-      "grad_norm": 0.0023398450575768948,
-      "learning_rate": 0.00019999624259149577,
-      "loss": 46.0,
-      "step": 17145
-    },
-    {
-      "epoch": 2.761222271427996,
-      "grad_norm": 0.0059340642765164375,
-      "learning_rate": 0.00019999624215292038,
-      "loss": 46.0,
-      "step": 17146
-    },
-    {
-      "epoch": 2.7613833085067836,
-      "grad_norm": 0.0037918237503618,
-      "learning_rate": 0.0001999962417143194,
-      "loss": 46.0,
-      "step": 17147
-    },
-    {
-      "epoch": 2.761544345585571,
-      "grad_norm": 0.006784059107303619,
-      "learning_rate": 0.00019999624127569284,
-      "loss": 46.0,
-      "step": 17148
-    },
-    {
-      "epoch": 2.7617053826643585,
-      "grad_norm": 0.005706030875444412,
-      "learning_rate": 0.00019999624083704066,
-      "loss": 46.0,
-      "step": 17149
-    },
-    {
-      "epoch": 2.761866419743146,
-      "grad_norm": 0.0014201728627085686,
-      "learning_rate": 0.00019999624039836292,
-      "loss": 46.0,
-      "step": 17150
-    },
-    {
-      "epoch": 2.7620274568219334,
-      "grad_norm": 0.002819629618898034,
-      "learning_rate": 0.00019999623995965954,
-      "loss": 46.0,
-      "step": 17151
-    },
-    {
-      "epoch": 2.7621884939007204,
-      "grad_norm": 0.0010610180906951427,
-      "learning_rate": 0.0001999962395209306,
-      "loss": 46.0,
-      "step": 17152
-    },
-    {
-      "epoch": 2.762349530979508,
-      "grad_norm": 0.006625819485634565,
-      "learning_rate": 0.00019999623908217607,
-      "loss": 46.0,
-      "step": 17153
-    },
-    {
-      "epoch": 2.7625105680582953,
-      "grad_norm": 0.008228259161114693,
-      "learning_rate": 0.00019999623864339593,
-      "loss": 46.0,
-      "step": 17154
-    },
-    {
-      "epoch": 2.7626716051370828,
-      "grad_norm": 0.009319491684436798,
-      "learning_rate": 0.0001999962382045902,
-      "loss": 46.0,
-      "step": 17155
-    },
-    {
-      "epoch": 2.76283264221587,
-      "grad_norm": 0.0008523929864168167,
-      "learning_rate": 0.00019999623776575888,
-      "loss": 46.0,
-      "step": 17156
-    },
-    {
-      "epoch": 2.7629936792946577,
-      "grad_norm": 0.00168919179122895,
-      "learning_rate": 0.00019999623732690195,
-      "loss": 46.0,
-      "step": 17157
-    },
-    {
-      "epoch": 2.763154716373445,
-      "grad_norm": 0.010237525217235088,
-      "learning_rate": 0.00019999623688801946,
-      "loss": 46.0,
-      "step": 17158
-    },
-    {
-      "epoch": 2.7633157534522326,
-      "grad_norm": 0.008353172801434994,
-      "learning_rate": 0.00019999623644911135,
-      "loss": 46.0,
-      "step": 17159
-    },
-    {
-      "epoch": 2.76347679053102,
-      "grad_norm": 0.007083939854055643,
-      "learning_rate": 0.00019999623601017763,
-      "loss": 46.0,
-      "step": 17160
-    },
-    {
-      "epoch": 2.763637827609807,
-      "grad_norm": 0.012969098053872585,
-      "learning_rate": 0.00019999623557121835,
-      "loss": 46.0,
-      "step": 17161
-    },
-    {
-      "epoch": 2.7637988646885945,
-      "grad_norm": 0.00047954244655556977,
-      "learning_rate": 0.00019999623513223345,
-      "loss": 46.0,
-      "step": 17162
-    },
-    {
-      "epoch": 2.763959901767382,
-      "grad_norm": 0.007116248365491629,
-      "learning_rate": 0.00019999623469322297,
-      "loss": 46.0,
-      "step": 17163
-    },
-    {
-      "epoch": 2.7641209388461694,
-      "grad_norm": 0.003631382016465068,
-      "learning_rate": 0.0001999962342541869,
-      "loss": 46.0,
-      "step": 17164
-    },
-    {
-      "epoch": 2.764281975924957,
-      "grad_norm": 0.0019030936527997255,
-      "learning_rate": 0.00019999623381512522,
-      "loss": 46.0,
-      "step": 17165
-    },
-    {
-      "epoch": 2.764443013003744,
-      "grad_norm": 0.0013142320094630122,
-      "learning_rate": 0.00019999623337603798,
-      "loss": 46.0,
-      "step": 17166
-    },
-    {
-      "epoch": 2.7646040500825313,
-      "grad_norm": 0.0013567854184657335,
-      "learning_rate": 0.00019999623293692512,
-      "loss": 46.0,
-      "step": 17167
-    },
-    {
-      "epoch": 2.7647650871613187,
-      "grad_norm": 0.002399209188297391,
-      "learning_rate": 0.00019999623249778665,
-      "loss": 46.0,
-      "step": 17168
-    },
-    {
-      "epoch": 2.764926124240106,
-      "grad_norm": 0.007906840182840824,
-      "learning_rate": 0.00019999623205862262,
-      "loss": 46.0,
-      "step": 17169
-    },
-    {
-      "epoch": 2.7650871613188936,
-      "grad_norm": 0.002173451706767082,
-      "learning_rate": 0.00019999623161943297,
-      "loss": 46.0,
-      "step": 17170
-    },
-    {
-      "epoch": 2.765248198397681,
-      "grad_norm": 0.006524259690195322,
-      "learning_rate": 0.00019999623118021776,
-      "loss": 46.0,
-      "step": 17171
-    },
-    {
-      "epoch": 2.7654092354764686,
-      "grad_norm": 0.005930430721491575,
-      "learning_rate": 0.00019999623074097692,
-      "loss": 46.0,
-      "step": 17172
-    },
-    {
-      "epoch": 2.765570272555256,
-      "grad_norm": 0.010853881016373634,
-      "learning_rate": 0.00019999623030171048,
-      "loss": 46.0,
-      "step": 17173
-    },
-    {
-      "epoch": 2.7657313096340435,
-      "grad_norm": 0.0033220115583389997,
-      "learning_rate": 0.00019999622986241846,
-      "loss": 46.0,
-      "step": 17174
-    },
-    {
-      "epoch": 2.7658923467128305,
-      "grad_norm": 0.005506960209459066,
-      "learning_rate": 0.00019999622942310085,
-      "loss": 46.0,
-      "step": 17175
-    },
-    {
-      "epoch": 2.766053383791618,
-      "grad_norm": 0.004830505233258009,
-      "learning_rate": 0.00019999622898375768,
-      "loss": 46.0,
-      "step": 17176
-    },
-    {
-      "epoch": 2.7662144208704054,
-      "grad_norm": 0.004754712339490652,
-      "learning_rate": 0.00019999622854438887,
-      "loss": 46.0,
-      "step": 17177
-    },
-    {
-      "epoch": 2.766375457949193,
-      "grad_norm": 0.0052373409271240234,
-      "learning_rate": 0.00019999622810499447,
-      "loss": 46.0,
-      "step": 17178
-    },
-    {
-      "epoch": 2.7665364950279803,
-      "grad_norm": 0.011678526178002357,
-      "learning_rate": 0.00019999622766557452,
-      "loss": 46.0,
-      "step": 17179
-    },
-    {
-      "epoch": 2.7666975321067677,
-      "grad_norm": 0.002471431391313672,
-      "learning_rate": 0.00019999622722612892,
-      "loss": 46.0,
-      "step": 17180
-    },
-    {
-      "epoch": 2.7668585691855547,
-      "grad_norm": 0.0028403140604496002,
-      "learning_rate": 0.00019999622678665776,
-      "loss": 46.0,
-      "step": 17181
-    },
-    {
-      "epoch": 2.767019606264342,
-      "grad_norm": 0.0040451339446008205,
-      "learning_rate": 0.00019999622634716099,
-      "loss": 46.0,
-      "step": 17182
-    },
-    {
-      "epoch": 2.7671806433431296,
-      "grad_norm": 0.006811333354562521,
-      "learning_rate": 0.00019999622590763863,
-      "loss": 46.0,
-      "step": 17183
-    },
-    {
-      "epoch": 2.767341680421917,
-      "grad_norm": 0.0028553283773362637,
-      "learning_rate": 0.00019999622546809068,
-      "loss": 46.0,
-      "step": 17184
-    },
-    {
-      "epoch": 2.7675027175007045,
-      "grad_norm": 0.0010586761636659503,
-      "learning_rate": 0.00019999622502851712,
-      "loss": 46.0,
-      "step": 17185
-    },
-    {
-      "epoch": 2.767663754579492,
-      "grad_norm": 0.0017935532378032804,
-      "learning_rate": 0.00019999622458891797,
-      "loss": 46.0,
-      "step": 17186
-    },
-    {
-      "epoch": 2.7678247916582794,
-      "grad_norm": 0.001102588721551001,
-      "learning_rate": 0.00019999622414929323,
-      "loss": 46.0,
-      "step": 17187
-    },
-    {
-      "epoch": 2.767985828737067,
-      "grad_norm": 0.0004795800778083503,
-      "learning_rate": 0.00019999622370964293,
-      "loss": 46.0,
-      "step": 17188
-    },
-    {
-      "epoch": 2.7681468658158543,
-      "grad_norm": 0.005188523791730404,
-      "learning_rate": 0.000199996223269967,
-      "loss": 46.0,
-      "step": 17189
-    },
-    {
-      "epoch": 2.7683079028946413,
-      "grad_norm": 0.0007705677999183536,
-      "learning_rate": 0.00019999622283026547,
-      "loss": 46.0,
-      "step": 17190
-    },
-    {
-      "epoch": 2.768468939973429,
-      "grad_norm": 0.003192085074260831,
-      "learning_rate": 0.00019999622239053836,
-      "loss": 46.0,
-      "step": 17191
-    },
-    {
-      "epoch": 2.7686299770522163,
-      "grad_norm": 0.008505075238645077,
-      "learning_rate": 0.00019999622195078563,
-      "loss": 46.0,
-      "step": 17192
-    },
-    {
-      "epoch": 2.7687910141310037,
-      "grad_norm": 0.0007946674013510346,
-      "learning_rate": 0.00019999622151100737,
-      "loss": 46.0,
-      "step": 17193
-    },
-    {
-      "epoch": 2.768952051209791,
-      "grad_norm": 0.004044679459184408,
-      "learning_rate": 0.00019999622107120347,
-      "loss": 46.0,
-      "step": 17194
-    },
-    {
-      "epoch": 2.7691130882885786,
-      "grad_norm": 0.0008073249482549727,
-      "learning_rate": 0.00019999622063137398,
-      "loss": 46.0,
-      "step": 17195
-    },
-    {
-      "epoch": 2.7692741253673656,
-      "grad_norm": 0.0007710932986810803,
-      "learning_rate": 0.0001999962201915189,
-      "loss": 46.0,
-      "step": 17196
-    },
-    {
-      "epoch": 2.769435162446153,
-      "grad_norm": 0.002095641568303108,
-      "learning_rate": 0.00019999621975163822,
-      "loss": 46.0,
-      "step": 17197
-    },
-    {
-      "epoch": 2.7695961995249405,
-      "grad_norm": 0.006098724901676178,
-      "learning_rate": 0.00019999621931173197,
-      "loss": 46.0,
-      "step": 17198
-    },
-    {
-      "epoch": 2.769757236603728,
-      "grad_norm": 0.007688706740736961,
-      "learning_rate": 0.00019999621887180008,
-      "loss": 46.0,
-      "step": 17199
-    },
-    {
-      "epoch": 2.7699182736825154,
-      "grad_norm": 0.003622662043198943,
-      "learning_rate": 0.00019999621843184263,
-      "loss": 46.0,
-      "step": 17200
-    },
-    {
-      "epoch": 2.770079310761303,
-      "grad_norm": 0.0020627097692340612,
-      "learning_rate": 0.0001999962179918596,
-      "loss": 46.0,
-      "step": 17201
-    },
-    {
-      "epoch": 2.7702403478400903,
-      "grad_norm": 0.003132890211418271,
-      "learning_rate": 0.00019999621755185097,
-      "loss": 46.0,
-      "step": 17202
-    },
-    {
-      "epoch": 2.7704013849188778,
-      "grad_norm": 0.009673915803432465,
-      "learning_rate": 0.0001999962171118167,
-      "loss": 46.0,
-      "step": 17203
-    },
-    {
-      "epoch": 2.7705624219976652,
-      "grad_norm": 0.002972107380628586,
-      "learning_rate": 0.00019999621667175687,
-      "loss": 46.0,
-      "step": 17204
-    },
-    {
-      "epoch": 2.7707234590764522,
-      "grad_norm": 0.0029015387408435345,
-      "learning_rate": 0.00019999621623167143,
-      "loss": 46.0,
-      "step": 17205
-    },
-    {
-      "epoch": 2.7708844961552397,
-      "grad_norm": 0.005969700403511524,
-      "learning_rate": 0.00019999621579156043,
-      "loss": 46.0,
-      "step": 17206
-    },
-    {
-      "epoch": 2.771045533234027,
-      "grad_norm": 0.007375736255198717,
-      "learning_rate": 0.00019999621535142384,
-      "loss": 46.0,
-      "step": 17207
-    },
-    {
-      "epoch": 2.7712065703128146,
-      "grad_norm": 0.00248416094109416,
-      "learning_rate": 0.0001999962149112616,
-      "loss": 46.0,
-      "step": 17208
-    },
-    {
-      "epoch": 2.771367607391602,
-      "grad_norm": 0.0014966547023504972,
-      "learning_rate": 0.00019999621447107382,
-      "loss": 46.0,
-      "step": 17209
-    },
-    {
-      "epoch": 2.771528644470389,
-      "grad_norm": 0.009774049744009972,
-      "learning_rate": 0.00019999621403086042,
-      "loss": 46.0,
-      "step": 17210
-    },
-    {
-      "epoch": 2.7716896815491765,
-      "grad_norm": 0.00196552905254066,
-      "learning_rate": 0.00019999621359062143,
-      "loss": 46.0,
-      "step": 17211
-    },
-    {
-      "epoch": 2.771850718627964,
-      "grad_norm": 0.0007924105157144368,
-      "learning_rate": 0.00019999621315035685,
-      "loss": 46.0,
-      "step": 17212
-    },
-    {
-      "epoch": 2.7720117557067514,
-      "grad_norm": 0.002094122814014554,
-      "learning_rate": 0.00019999621271006666,
-      "loss": 46.0,
-      "step": 17213
-    },
-    {
-      "epoch": 2.772172792785539,
-      "grad_norm": 0.009498411789536476,
-      "learning_rate": 0.0001999962122697509,
-      "loss": 46.0,
-      "step": 17214
-    },
-    {
-      "epoch": 2.7723338298643263,
-      "grad_norm": 0.0010483688674867153,
-      "learning_rate": 0.00019999621182940954,
-      "loss": 46.0,
-      "step": 17215
-    },
-    {
-      "epoch": 2.7724948669431138,
-      "grad_norm": 0.0011623470345512033,
-      "learning_rate": 0.00019999621138904256,
-      "loss": 46.0,
-      "step": 17216
-    },
-    {
-      "epoch": 2.772655904021901,
-      "grad_norm": 0.01052230317145586,
-      "learning_rate": 0.00019999621094865001,
-      "loss": 46.0,
-      "step": 17217
-    },
-    {
-      "epoch": 2.7728169411006887,
-      "grad_norm": 0.00584650132805109,
-      "learning_rate": 0.00019999621050823189,
-      "loss": 46.0,
-      "step": 17218
-    },
-    {
-      "epoch": 2.7729779781794757,
-      "grad_norm": 0.0006701999227516353,
-      "learning_rate": 0.00019999621006778812,
-      "loss": 46.0,
-      "step": 17219
-    },
-    {
-      "epoch": 2.773139015258263,
-      "grad_norm": 0.00454955268651247,
-      "learning_rate": 0.00019999620962731881,
-      "loss": 46.0,
-      "step": 17220
-    },
-    {
-      "epoch": 2.7733000523370506,
-      "grad_norm": 0.0047552818432450294,
-      "learning_rate": 0.0001999962091868239,
-      "loss": 46.0,
-      "step": 17221
-    },
-    {
-      "epoch": 2.773461089415838,
-      "grad_norm": 0.004514632746577263,
-      "learning_rate": 0.00019999620874630334,
-      "loss": 46.0,
-      "step": 17222
-    },
-    {
-      "epoch": 2.7736221264946255,
-      "grad_norm": 0.005523323081433773,
-      "learning_rate": 0.00019999620830575725,
-      "loss": 46.0,
-      "step": 17223
-    },
-    {
-      "epoch": 2.773783163573413,
-      "grad_norm": 0.0011826539412140846,
-      "learning_rate": 0.00019999620786518554,
-      "loss": 46.0,
-      "step": 17224
-    },
-    {
-      "epoch": 2.7739442006522,
-      "grad_norm": 0.0007408465025946498,
-      "learning_rate": 0.00019999620742458822,
-      "loss": 46.0,
-      "step": 17225
-    },
-    {
-      "epoch": 2.7741052377309874,
-      "grad_norm": 0.0014174769166857004,
-      "learning_rate": 0.00019999620698396534,
-      "loss": 46.0,
-      "step": 17226
-    },
-    {
-      "epoch": 2.774266274809775,
-      "grad_norm": 0.0009421113645657897,
-      "learning_rate": 0.00019999620654331682,
-      "loss": 46.0,
-      "step": 17227
-    },
-    {
-      "epoch": 2.7744273118885623,
-      "grad_norm": 0.001157391001470387,
-      "learning_rate": 0.00019999620610264277,
-      "loss": 46.0,
-      "step": 17228
-    },
-    {
-      "epoch": 2.7745883489673497,
-      "grad_norm": 0.0023011311423033476,
-      "learning_rate": 0.00019999620566194307,
-      "loss": 46.0,
-      "step": 17229
-    },
-    {
-      "epoch": 2.774749386046137,
-      "grad_norm": 0.001446246518753469,
-      "learning_rate": 0.00019999620522121781,
-      "loss": 46.0,
-      "step": 17230
-    },
-    {
-      "epoch": 2.7749104231249246,
-      "grad_norm": 0.0018936104606837034,
-      "learning_rate": 0.00019999620478046692,
-      "loss": 46.0,
-      "step": 17231
-    },
-    {
-      "epoch": 2.775071460203712,
-      "grad_norm": 0.0025069648399949074,
-      "learning_rate": 0.0001999962043396905,
-      "loss": 46.0,
-      "step": 17232
-    },
-    {
-      "epoch": 2.7752324972824995,
-      "grad_norm": 0.003439794760197401,
-      "learning_rate": 0.00019999620389888842,
-      "loss": 46.0,
-      "step": 17233
-    },
-    {
-      "epoch": 2.7753935343612866,
-      "grad_norm": 0.0008250723476521671,
-      "learning_rate": 0.00019999620345806078,
-      "loss": 46.0,
-      "step": 17234
-    },
-    {
-      "epoch": 2.775554571440074,
-      "grad_norm": 0.003856923431158066,
-      "learning_rate": 0.0001999962030172075,
-      "loss": 46.0,
-      "step": 17235
-    },
-    {
-      "epoch": 2.7757156085188615,
-      "grad_norm": 0.004507602658122778,
-      "learning_rate": 0.0001999962025763287,
-      "loss": 46.0,
-      "step": 17236
-    },
-    {
-      "epoch": 2.775876645597649,
-      "grad_norm": 0.005155741702765226,
-      "learning_rate": 0.00019999620213542426,
-      "loss": 46.0,
-      "step": 17237
-    },
-    {
-      "epoch": 2.7760376826764364,
-      "grad_norm": 0.009643965400755405,
-      "learning_rate": 0.00019999620169449425,
-      "loss": 46.0,
-      "step": 17238
-    },
-    {
-      "epoch": 2.7761987197552234,
-      "grad_norm": 0.0009915520204231143,
-      "learning_rate": 0.00019999620125353863,
-      "loss": 46.0,
-      "step": 17239
-    },
-    {
-      "epoch": 2.776359756834011,
-      "grad_norm": 0.0011607661144807935,
-      "learning_rate": 0.00019999620081255742,
-      "loss": 46.0,
-      "step": 17240
-    },
-    {
-      "epoch": 2.7765207939127983,
-      "grad_norm": 0.001892865402624011,
-      "learning_rate": 0.0001999962003715506,
-      "loss": 46.0,
-      "step": 17241
-    },
-    {
-      "epoch": 2.7766818309915857,
-      "grad_norm": 0.012929723598062992,
-      "learning_rate": 0.00019999619993051821,
-      "loss": 46.0,
-      "step": 17242
-    },
-    {
-      "epoch": 2.776842868070373,
-      "grad_norm": 0.005594770889729261,
-      "learning_rate": 0.00019999619948946022,
-      "loss": 46.0,
-      "step": 17243
-    },
-    {
-      "epoch": 2.7770039051491606,
-      "grad_norm": 0.0034008945804089308,
-      "learning_rate": 0.0001999961990483766,
-      "loss": 46.0,
-      "step": 17244
-    },
-    {
-      "epoch": 2.777164942227948,
-      "grad_norm": 0.004892023745924234,
-      "learning_rate": 0.00019999619860726743,
-      "loss": 46.0,
-      "step": 17245
-    },
-    {
-      "epoch": 2.7773259793067355,
-      "grad_norm": 0.001985863083973527,
-      "learning_rate": 0.00019999619816613267,
-      "loss": 46.0,
-      "step": 17246
-    },
-    {
-      "epoch": 2.777487016385523,
-      "grad_norm": 0.0025620614178478718,
-      "learning_rate": 0.0001999961977249723,
-      "loss": 46.0,
-      "step": 17247
-    },
-    {
-      "epoch": 2.7776480534643104,
-      "grad_norm": 0.0015864807646721601,
-      "learning_rate": 0.00019999619728378634,
-      "loss": 46.0,
-      "step": 17248
-    },
-    {
-      "epoch": 2.7778090905430974,
-      "grad_norm": 0.0024884187150746584,
-      "learning_rate": 0.00019999619684257477,
-      "loss": 46.0,
-      "step": 17249
-    },
-    {
-      "epoch": 2.777970127621885,
-      "grad_norm": 0.0026782918721437454,
-      "learning_rate": 0.00019999619640133763,
-      "loss": 46.0,
-      "step": 17250
-    },
-    {
-      "epoch": 2.7781311647006723,
-      "grad_norm": 0.009245862253010273,
-      "learning_rate": 0.00019999619596007488,
-      "loss": 46.0,
-      "step": 17251
-    },
-    {
-      "epoch": 2.77829220177946,
-      "grad_norm": 0.0026006782427430153,
-      "learning_rate": 0.00019999619551878655,
-      "loss": 46.0,
-      "step": 17252
-    },
-    {
-      "epoch": 2.7784532388582472,
-      "grad_norm": 0.0018891041399911046,
-      "learning_rate": 0.0001999961950774726,
-      "loss": 46.0,
-      "step": 17253
-    },
-    {
-      "epoch": 2.7786142759370343,
-      "grad_norm": 0.001433408702723682,
-      "learning_rate": 0.00019999619463613308,
-      "loss": 46.0,
-      "step": 17254
-    },
-    {
-      "epoch": 2.7787753130158217,
-      "grad_norm": 0.0011654626578092575,
-      "learning_rate": 0.00019999619419476796,
-      "loss": 46.0,
-      "step": 17255
-    },
-    {
-      "epoch": 2.778936350094609,
-      "grad_norm": 0.00508570671081543,
-      "learning_rate": 0.00019999619375337725,
-      "loss": 46.0,
-      "step": 17256
-    },
-    {
-      "epoch": 2.7790973871733966,
-      "grad_norm": 0.0018112065736204386,
-      "learning_rate": 0.00019999619331196092,
-      "loss": 46.0,
-      "step": 17257
-    },
-    {
-      "epoch": 2.779258424252184,
-      "grad_norm": 0.0017384226666763425,
-      "learning_rate": 0.00019999619287051906,
-      "loss": 46.0,
-      "step": 17258
-    },
-    {
-      "epoch": 2.7794194613309715,
-      "grad_norm": 0.004295541439205408,
-      "learning_rate": 0.00019999619242905154,
-      "loss": 46.0,
-      "step": 17259
-    },
-    {
-      "epoch": 2.779580498409759,
-      "grad_norm": 0.000518948188982904,
-      "learning_rate": 0.00019999619198755845,
-      "loss": 46.0,
-      "step": 17260
-    },
-    {
-      "epoch": 2.7797415354885464,
-      "grad_norm": 0.005840987432748079,
-      "learning_rate": 0.00019999619154603977,
-      "loss": 46.0,
-      "step": 17261
-    },
-    {
-      "epoch": 2.779902572567334,
-      "grad_norm": 0.003383237635716796,
-      "learning_rate": 0.0001999961911044955,
-      "loss": 46.0,
-      "step": 17262
-    },
-    {
-      "epoch": 2.780063609646121,
-      "grad_norm": 0.012057020328938961,
-      "learning_rate": 0.0001999961906629256,
-      "loss": 46.0,
-      "step": 17263
-    },
-    {
-      "epoch": 2.7802246467249083,
-      "grad_norm": 0.017308978363871574,
-      "learning_rate": 0.00019999619022133014,
-      "loss": 46.0,
-      "step": 17264
-    },
-    {
-      "epoch": 2.7803856838036958,
-      "grad_norm": 0.006352408789098263,
-      "learning_rate": 0.0001999961897797091,
-      "loss": 46.0,
-      "step": 17265
-    },
-    {
-      "epoch": 2.7805467208824832,
-      "grad_norm": 0.0016968533163890243,
-      "learning_rate": 0.00019999618933806245,
-      "loss": 46.0,
-      "step": 17266
-    },
-    {
-      "epoch": 2.7807077579612707,
-      "grad_norm": 0.0031767133623361588,
-      "learning_rate": 0.0001999961888963902,
-      "loss": 46.0,
-      "step": 17267
-    },
-    {
-      "epoch": 2.780868795040058,
-      "grad_norm": 0.0020461815875023603,
-      "learning_rate": 0.00019999618845469236,
-      "loss": 46.0,
-      "step": 17268
-    },
-    {
-      "epoch": 2.781029832118845,
-      "grad_norm": 0.005839423276484013,
-      "learning_rate": 0.0001999961880129689,
-      "loss": 46.0,
-      "step": 17269
-    },
-    {
-      "epoch": 2.7811908691976326,
-      "grad_norm": 0.0022688517346978188,
-      "learning_rate": 0.00019999618757121987,
-      "loss": 46.0,
-      "step": 17270
-    },
-    {
-      "epoch": 2.78135190627642,
-      "grad_norm": 0.0011125602759420872,
-      "learning_rate": 0.00019999618712944527,
-      "loss": 46.0,
-      "step": 17271
-    },
-    {
-      "epoch": 2.7815129433552075,
-      "grad_norm": 0.0046051060780882835,
-      "learning_rate": 0.00019999618668764505,
-      "loss": 46.0,
-      "step": 17272
-    },
-    {
-      "epoch": 2.781673980433995,
-      "grad_norm": 0.0004647821479011327,
-      "learning_rate": 0.00019999618624581925,
-      "loss": 46.0,
-      "step": 17273
-    },
-    {
-      "epoch": 2.7818350175127824,
-      "grad_norm": 0.0032389399129897356,
-      "learning_rate": 0.00019999618580396783,
-      "loss": 46.0,
-      "step": 17274
-    },
-    {
-      "epoch": 2.78199605459157,
-      "grad_norm": 0.015031995251774788,
-      "learning_rate": 0.00019999618536209083,
-      "loss": 46.0,
-      "step": 17275
-    },
-    {
-      "epoch": 2.7821570916703573,
-      "grad_norm": 0.004603092558681965,
-      "learning_rate": 0.00019999618492018827,
-      "loss": 46.0,
-      "step": 17276
-    },
-    {
-      "epoch": 2.7823181287491447,
-      "grad_norm": 0.0019079707562923431,
-      "learning_rate": 0.00019999618447826006,
-      "loss": 46.0,
-      "step": 17277
-    },
-    {
-      "epoch": 2.7824791658279318,
-      "grad_norm": 0.002006322843953967,
-      "learning_rate": 0.00019999618403630627,
-      "loss": 46.0,
-      "step": 17278
-    },
-    {
-      "epoch": 2.782640202906719,
-      "grad_norm": 0.0019319868879392743,
-      "learning_rate": 0.0001999961835943269,
-      "loss": 46.0,
-      "step": 17279
-    },
-    {
-      "epoch": 2.7828012399855067,
-      "grad_norm": 0.0018949775258079171,
-      "learning_rate": 0.00019999618315232195,
-      "loss": 46.0,
-      "step": 17280
-    },
-    {
-      "epoch": 2.782962277064294,
-      "grad_norm": 0.005325635429471731,
-      "learning_rate": 0.00019999618271029137,
-      "loss": 46.0,
-      "step": 17281
-    },
-    {
-      "epoch": 2.7831233141430816,
-      "grad_norm": 0.0014188267523422837,
-      "learning_rate": 0.00019999618226823523,
-      "loss": 46.0,
-      "step": 17282
-    },
-    {
-      "epoch": 2.7832843512218686,
-      "grad_norm": 0.006055557169020176,
-      "learning_rate": 0.00019999618182615347,
-      "loss": 46.0,
-      "step": 17283
-    },
-    {
-      "epoch": 2.783445388300656,
-      "grad_norm": 0.0017244759947061539,
-      "learning_rate": 0.00019999618138404613,
-      "loss": 46.0,
-      "step": 17284
-    },
-    {
-      "epoch": 2.7836064253794435,
-      "grad_norm": 0.0049844346940517426,
-      "learning_rate": 0.0001999961809419132,
-      "loss": 46.0,
-      "step": 17285
-    },
-    {
-      "epoch": 2.783767462458231,
-      "grad_norm": 0.006922903936356306,
-      "learning_rate": 0.00019999618049975466,
-      "loss": 46.0,
-      "step": 17286
-    },
-    {
-      "epoch": 2.7839284995370184,
-      "grad_norm": 0.004829640034586191,
-      "learning_rate": 0.00019999618005757053,
-      "loss": 46.0,
-      "step": 17287
-    },
-    {
-      "epoch": 2.784089536615806,
-      "grad_norm": 0.0011554949451237917,
-      "learning_rate": 0.0001999961796153608,
-      "loss": 46.0,
-      "step": 17288
-    },
-    {
-      "epoch": 2.7842505736945933,
-      "grad_norm": 0.01126040518283844,
-      "learning_rate": 0.00019999617917312553,
-      "loss": 46.0,
-      "step": 17289
-    },
-    {
-      "epoch": 2.7844116107733807,
-      "grad_norm": 0.0038276687264442444,
-      "learning_rate": 0.00019999617873086458,
-      "loss": 46.0,
-      "step": 17290
-    },
-    {
-      "epoch": 2.784572647852168,
-      "grad_norm": 0.0012934472179040313,
-      "learning_rate": 0.0001999961782885781,
-      "loss": 46.0,
-      "step": 17291
-    },
-    {
-      "epoch": 2.784733684930955,
-      "grad_norm": 0.002570344367995858,
-      "learning_rate": 0.000199996177846266,
-      "loss": 46.0,
-      "step": 17292
-    },
-    {
-      "epoch": 2.7848947220097426,
-      "grad_norm": 0.0021267388947308064,
-      "learning_rate": 0.0001999961774039283,
-      "loss": 46.0,
-      "step": 17293
-    },
-    {
-      "epoch": 2.78505575908853,
-      "grad_norm": 0.0008051579934544861,
-      "learning_rate": 0.00019999617696156503,
-      "loss": 46.0,
-      "step": 17294
-    },
-    {
-      "epoch": 2.7852167961673175,
-      "grad_norm": 0.0015287662390619516,
-      "learning_rate": 0.00019999617651917615,
-      "loss": 46.0,
-      "step": 17295
-    },
-    {
-      "epoch": 2.785377833246105,
-      "grad_norm": 0.00878703873604536,
-      "learning_rate": 0.00019999617607676168,
-      "loss": 46.0,
-      "step": 17296
-    },
-    {
-      "epoch": 2.7855388703248924,
-      "grad_norm": 0.006900724023580551,
-      "learning_rate": 0.00019999617563432162,
-      "loss": 46.0,
-      "step": 17297
-    },
-    {
-      "epoch": 2.7856999074036795,
-      "grad_norm": 0.003436419880017638,
-      "learning_rate": 0.00019999617519185595,
-      "loss": 46.0,
-      "step": 17298
-    },
-    {
-      "epoch": 2.785860944482467,
-      "grad_norm": 0.002675537019968033,
-      "learning_rate": 0.00019999617474936472,
-      "loss": 46.0,
-      "step": 17299
-    },
-    {
-      "epoch": 2.7860219815612544,
-      "grad_norm": 0.0018416759558022022,
-      "learning_rate": 0.00019999617430684787,
-      "loss": 46.0,
-      "step": 17300
-    },
-    {
-      "epoch": 2.786183018640042,
-      "grad_norm": 0.005380876827985048,
-      "learning_rate": 0.0001999961738643054,
-      "loss": 46.0,
-      "step": 17301
-    },
-    {
-      "epoch": 2.7863440557188293,
-      "grad_norm": 0.0022907995153218508,
-      "learning_rate": 0.0001999961734217374,
-      "loss": 46.0,
-      "step": 17302
-    },
-    {
-      "epoch": 2.7865050927976167,
-      "grad_norm": 0.0019048057729378343,
-      "learning_rate": 0.00019999617297914376,
-      "loss": 46.0,
-      "step": 17303
-    },
-    {
-      "epoch": 2.786666129876404,
-      "grad_norm": 0.006211045663803816,
-      "learning_rate": 0.00019999617253652456,
-      "loss": 46.0,
-      "step": 17304
-    },
-    {
-      "epoch": 2.7868271669551916,
-      "grad_norm": 0.003978512715548277,
-      "learning_rate": 0.00019999617209387973,
-      "loss": 46.0,
-      "step": 17305
-    },
-    {
-      "epoch": 2.786988204033979,
-      "grad_norm": 0.010241970419883728,
-      "learning_rate": 0.0001999961716512093,
-      "loss": 46.0,
-      "step": 17306
-    },
-    {
-      "epoch": 2.787149241112766,
-      "grad_norm": 0.0029114349745213985,
-      "learning_rate": 0.0001999961712085133,
-      "loss": 46.0,
-      "step": 17307
-    },
-    {
-      "epoch": 2.7873102781915535,
-      "grad_norm": 0.0019352325471118093,
-      "learning_rate": 0.00019999617076579173,
-      "loss": 46.0,
-      "step": 17308
-    },
-    {
-      "epoch": 2.787471315270341,
-      "grad_norm": 0.005099796690046787,
-      "learning_rate": 0.00019999617032304454,
-      "loss": 46.0,
-      "step": 17309
-    },
-    {
-      "epoch": 2.7876323523491284,
-      "grad_norm": 0.0015000685816630721,
-      "learning_rate": 0.00019999616988027174,
-      "loss": 46.0,
-      "step": 17310
-    },
-    {
-      "epoch": 2.787793389427916,
-      "grad_norm": 0.00374378333799541,
-      "learning_rate": 0.00019999616943747336,
-      "loss": 46.0,
-      "step": 17311
-    },
-    {
-      "epoch": 2.7879544265067033,
-      "grad_norm": 0.005975484382361174,
-      "learning_rate": 0.0001999961689946494,
-      "loss": 46.0,
-      "step": 17312
-    },
-    {
-      "epoch": 2.7881154635854903,
-      "grad_norm": 0.0068528796546161175,
-      "learning_rate": 0.00019999616855179982,
-      "loss": 46.0,
-      "step": 17313
-    },
-    {
-      "epoch": 2.788276500664278,
-      "grad_norm": 0.0008332461002282798,
-      "learning_rate": 0.00019999616810892468,
-      "loss": 46.0,
-      "step": 17314
-    },
-    {
-      "epoch": 2.7884375377430652,
-      "grad_norm": 0.002808322198688984,
-      "learning_rate": 0.00019999616766602391,
-      "loss": 46.0,
-      "step": 17315
-    },
-    {
-      "epoch": 2.7885985748218527,
-      "grad_norm": 0.012756533920764923,
-      "learning_rate": 0.00019999616722309757,
-      "loss": 46.0,
-      "step": 17316
-    },
-    {
-      "epoch": 2.78875961190064,
-      "grad_norm": 0.0030364638660103083,
-      "learning_rate": 0.00019999616678014563,
-      "loss": 46.0,
-      "step": 17317
-    },
-    {
-      "epoch": 2.7889206489794276,
-      "grad_norm": 0.0013004435459151864,
-      "learning_rate": 0.00019999616633716808,
-      "loss": 46.0,
-      "step": 17318
-    },
-    {
-      "epoch": 2.789081686058215,
-      "grad_norm": 0.007769137620925903,
-      "learning_rate": 0.00019999616589416497,
-      "loss": 46.0,
-      "step": 17319
-    },
-    {
-      "epoch": 2.7892427231370025,
-      "grad_norm": 0.004966042470186949,
-      "learning_rate": 0.00019999616545113624,
-      "loss": 46.0,
-      "step": 17320
-    },
-    {
-      "epoch": 2.78940376021579,
-      "grad_norm": 0.003941311500966549,
-      "learning_rate": 0.00019999616500808193,
-      "loss": 46.0,
-      "step": 17321
-    },
-    {
-      "epoch": 2.789564797294577,
-      "grad_norm": 0.003876606235280633,
-      "learning_rate": 0.000199996164565002,
-      "loss": 46.0,
-      "step": 17322
-    },
-    {
-      "epoch": 2.7897258343733644,
-      "grad_norm": 0.0012116816360503435,
-      "learning_rate": 0.0001999961641218965,
-      "loss": 46.0,
-      "step": 17323
-    },
-    {
-      "epoch": 2.789886871452152,
-      "grad_norm": 0.0012563696363940835,
-      "learning_rate": 0.00019999616367876542,
-      "loss": 46.0,
-      "step": 17324
-    },
-    {
-      "epoch": 2.7900479085309393,
-      "grad_norm": 0.002336804987862706,
-      "learning_rate": 0.0001999961632356087,
-      "loss": 46.0,
-      "step": 17325
-    },
-    {
-      "epoch": 2.7902089456097268,
-      "grad_norm": 0.0017057409277185798,
-      "learning_rate": 0.00019999616279242643,
-      "loss": 46.0,
-      "step": 17326
-    },
-    {
-      "epoch": 2.7903699826885138,
-      "grad_norm": 0.0013661442790180445,
-      "learning_rate": 0.00019999616234921854,
-      "loss": 46.0,
-      "step": 17327
-    },
-    {
-      "epoch": 2.7905310197673012,
-      "grad_norm": 0.0021454968955367804,
-      "learning_rate": 0.0001999961619059851,
-      "loss": 46.0,
-      "step": 17328
-    },
-    {
-      "epoch": 2.7906920568460887,
-      "grad_norm": 0.002131877699866891,
-      "learning_rate": 0.00019999616146272603,
-      "loss": 46.0,
-      "step": 17329
-    },
-    {
-      "epoch": 2.790853093924876,
-      "grad_norm": 0.00490427203476429,
-      "learning_rate": 0.00019999616101944135,
-      "loss": 46.0,
-      "step": 17330
-    },
-    {
-      "epoch": 2.7910141310036636,
-      "grad_norm": 0.005995448678731918,
-      "learning_rate": 0.0001999961605761311,
-      "loss": 46.0,
-      "step": 17331
-    },
-    {
-      "epoch": 2.791175168082451,
-      "grad_norm": 0.013836621306836605,
-      "learning_rate": 0.00019999616013279524,
-      "loss": 46.0,
-      "step": 17332
-    },
-    {
-      "epoch": 2.7913362051612385,
-      "grad_norm": 0.0013461961643770337,
-      "learning_rate": 0.0001999961596894338,
-      "loss": 46.0,
-      "step": 17333
-    },
-    {
-      "epoch": 2.791497242240026,
-      "grad_norm": 0.004238417837768793,
-      "learning_rate": 0.00019999615924604674,
-      "loss": 46.0,
-      "step": 17334
-    },
-    {
-      "epoch": 2.7916582793188134,
-      "grad_norm": 0.008564460091292858,
-      "learning_rate": 0.00019999615880263413,
-      "loss": 46.0,
-      "step": 17335
-    },
-    {
-      "epoch": 2.7918193163976004,
-      "grad_norm": 0.0028342341538518667,
-      "learning_rate": 0.0001999961583591959,
-      "loss": 46.0,
-      "step": 17336
-    },
-    {
-      "epoch": 2.791980353476388,
-      "grad_norm": 0.001389666460454464,
-      "learning_rate": 0.0001999961579157321,
-      "loss": 46.0,
-      "step": 17337
-    },
-    {
-      "epoch": 2.7921413905551753,
-      "grad_norm": 0.003649804275482893,
-      "learning_rate": 0.00019999615747224266,
-      "loss": 46.0,
-      "step": 17338
-    },
-    {
-      "epoch": 2.7923024276339627,
-      "grad_norm": 0.02480003610253334,
-      "learning_rate": 0.00019999615702872764,
-      "loss": 46.0,
-      "step": 17339
-    },
-    {
-      "epoch": 2.79246346471275,
-      "grad_norm": 0.004762317985296249,
-      "learning_rate": 0.00019999615658518706,
-      "loss": 46.0,
-      "step": 17340
-    },
-    {
-      "epoch": 2.7926245017915377,
-      "grad_norm": 0.00366000784561038,
-      "learning_rate": 0.00019999615614162087,
-      "loss": 46.0,
-      "step": 17341
-    },
-    {
-      "epoch": 2.7927855388703247,
-      "grad_norm": 0.0016108189010992646,
-      "learning_rate": 0.00019999615569802907,
-      "loss": 46.0,
-      "step": 17342
-    },
-    {
-      "epoch": 2.792946575949112,
-      "grad_norm": 0.01248826552182436,
-      "learning_rate": 0.00019999615525441168,
-      "loss": 46.0,
-      "step": 17343
-    },
-    {
-      "epoch": 2.7931076130278996,
-      "grad_norm": 0.006517420988529921,
-      "learning_rate": 0.00019999615481076872,
-      "loss": 46.0,
-      "step": 17344
-    },
-    {
-      "epoch": 2.793268650106687,
-      "grad_norm": 0.0022465353831648827,
-      "learning_rate": 0.00019999615436710016,
-      "loss": 46.0,
-      "step": 17345
-    },
-    {
-      "epoch": 2.7934296871854745,
-      "grad_norm": 0.002218261593952775,
-      "learning_rate": 0.00019999615392340598,
-      "loss": 46.0,
-      "step": 17346
-    },
-    {
-      "epoch": 2.793590724264262,
-      "grad_norm": 0.006983265746384859,
-      "learning_rate": 0.00019999615347968624,
-      "loss": 46.0,
-      "step": 17347
-    },
-    {
-      "epoch": 2.7937517613430494,
-      "grad_norm": 0.0011270599206909537,
-      "learning_rate": 0.00019999615303594088,
-      "loss": 46.0,
-      "step": 17348
-    },
-    {
-      "epoch": 2.793912798421837,
-      "grad_norm": 0.0007337352144531906,
-      "learning_rate": 0.00019999615259216994,
-      "loss": 46.0,
-      "step": 17349
-    },
-    {
-      "epoch": 2.7940738355006243,
-      "grad_norm": 0.0015230899443849921,
-      "learning_rate": 0.00019999615214837338,
-      "loss": 46.0,
-      "step": 17350
-    },
-    {
-      "epoch": 2.7942348725794113,
-      "grad_norm": 0.003638229798525572,
-      "learning_rate": 0.00019999615170455127,
-      "loss": 46.0,
-      "step": 17351
-    },
-    {
-      "epoch": 2.7943959096581987,
-      "grad_norm": 0.00909519661217928,
-      "learning_rate": 0.00019999615126070353,
-      "loss": 46.0,
-      "step": 17352
-    },
-    {
-      "epoch": 2.794556946736986,
-      "grad_norm": 0.00896409247070551,
-      "learning_rate": 0.00019999615081683022,
-      "loss": 46.0,
-      "step": 17353
-    },
-    {
-      "epoch": 2.7947179838157736,
-      "grad_norm": 0.007621936034411192,
-      "learning_rate": 0.00019999615037293128,
-      "loss": 46.0,
-      "step": 17354
-    },
-    {
-      "epoch": 2.794879020894561,
-      "grad_norm": 0.00622302433475852,
-      "learning_rate": 0.0001999961499290068,
-      "loss": 46.0,
-      "step": 17355
-    },
-    {
-      "epoch": 2.795040057973348,
-      "grad_norm": 0.0024968821089714766,
-      "learning_rate": 0.00019999614948505669,
-      "loss": 46.0,
-      "step": 17356
-    },
-    {
-      "epoch": 2.7952010950521355,
-      "grad_norm": 0.0006029472569935024,
-      "learning_rate": 0.00019999614904108096,
-      "loss": 46.0,
-      "step": 17357
-    },
-    {
-      "epoch": 2.795362132130923,
-      "grad_norm": 0.005191711708903313,
-      "learning_rate": 0.00019999614859707968,
-      "loss": 46.0,
-      "step": 17358
-    },
-    {
-      "epoch": 2.7955231692097104,
-      "grad_norm": 0.0016551089938730001,
-      "learning_rate": 0.0001999961481530528,
-      "loss": 46.0,
-      "step": 17359
-    },
-    {
-      "epoch": 2.795684206288498,
-      "grad_norm": 0.004970055539160967,
-      "learning_rate": 0.00019999614770900033,
-      "loss": 46.0,
-      "step": 17360
-    },
-    {
-      "epoch": 2.7958452433672853,
-      "grad_norm": 0.0071669877506792545,
-      "learning_rate": 0.00019999614726492224,
-      "loss": 46.0,
-      "step": 17361
-    },
-    {
-      "epoch": 2.796006280446073,
-      "grad_norm": 0.017554905265569687,
-      "learning_rate": 0.0001999961468208186,
-      "loss": 46.0,
-      "step": 17362
-    },
-    {
-      "epoch": 2.7961673175248603,
-      "grad_norm": 0.007663843221962452,
-      "learning_rate": 0.00019999614637668933,
-      "loss": 46.0,
-      "step": 17363
-    },
-    {
-      "epoch": 2.7963283546036477,
-      "grad_norm": 0.0031698443926870823,
-      "learning_rate": 0.00019999614593253448,
-      "loss": 46.0,
-      "step": 17364
-    },
-    {
-      "epoch": 2.796489391682435,
-      "grad_norm": 0.004886985756456852,
-      "learning_rate": 0.000199996145488354,
-      "loss": 46.0,
-      "step": 17365
-    },
-    {
-      "epoch": 2.796650428761222,
-      "grad_norm": 0.0017400650540366769,
-      "learning_rate": 0.00019999614504414797,
-      "loss": 46.0,
-      "step": 17366
-    },
-    {
-      "epoch": 2.7968114658400096,
-      "grad_norm": 0.0028208396397531033,
-      "learning_rate": 0.00019999614459991632,
-      "loss": 46.0,
-      "step": 17367
-    },
-    {
-      "epoch": 2.796972502918797,
-      "grad_norm": 0.0038511583115905523,
-      "learning_rate": 0.00019999614415565912,
-      "loss": 46.0,
-      "step": 17368
-    },
-    {
-      "epoch": 2.7971335399975845,
-      "grad_norm": 0.003789086826145649,
-      "learning_rate": 0.0001999961437113763,
-      "loss": 46.0,
-      "step": 17369
-    },
-    {
-      "epoch": 2.797294577076372,
-      "grad_norm": 0.0020728325471282005,
-      "learning_rate": 0.0001999961432670679,
-      "loss": 46.0,
-      "step": 17370
-    },
-    {
-      "epoch": 2.797455614155159,
-      "grad_norm": 0.0005276558804325759,
-      "learning_rate": 0.00019999614282273386,
-      "loss": 46.0,
-      "step": 17371
-    },
-    {
-      "epoch": 2.7976166512339464,
-      "grad_norm": 0.0015684147365391254,
-      "learning_rate": 0.00019999614237837426,
-      "loss": 46.0,
-      "step": 17372
-    },
-    {
-      "epoch": 2.797777688312734,
-      "grad_norm": 0.001548346015624702,
-      "learning_rate": 0.00019999614193398906,
-      "loss": 46.0,
-      "step": 17373
-    },
-    {
-      "epoch": 2.7979387253915213,
-      "grad_norm": 0.014401966705918312,
-      "learning_rate": 0.00019999614148957825,
-      "loss": 46.0,
-      "step": 17374
-    },
-    {
-      "epoch": 2.798099762470309,
-      "grad_norm": 0.004531376529484987,
-      "learning_rate": 0.00019999614104514188,
-      "loss": 46.0,
-      "step": 17375
-    },
-    {
-      "epoch": 2.7982607995490962,
-      "grad_norm": 0.001251931069418788,
-      "learning_rate": 0.0001999961406006799,
-      "loss": 46.0,
-      "step": 17376
-    },
-    {
-      "epoch": 2.7984218366278837,
-      "grad_norm": 0.008024302311241627,
-      "learning_rate": 0.00019999614015619232,
-      "loss": 46.0,
-      "step": 17377
-    },
-    {
-      "epoch": 2.798582873706671,
-      "grad_norm": 0.0005931921768933535,
-      "learning_rate": 0.00019999613971167913,
-      "loss": 46.0,
-      "step": 17378
-    },
-    {
-      "epoch": 2.7987439107854586,
-      "grad_norm": 0.0016716797836124897,
-      "learning_rate": 0.00019999613926714038,
-      "loss": 46.0,
-      "step": 17379
-    },
-    {
-      "epoch": 2.7989049478642456,
-      "grad_norm": 0.0008711446425877512,
-      "learning_rate": 0.00019999613882257605,
-      "loss": 46.0,
-      "step": 17380
-    },
-    {
-      "epoch": 2.799065984943033,
-      "grad_norm": 0.00184560043271631,
-      "learning_rate": 0.00019999613837798607,
-      "loss": 46.0,
-      "step": 17381
-    },
-    {
-      "epoch": 2.7992270220218205,
-      "grad_norm": 0.008441537618637085,
-      "learning_rate": 0.00019999613793337054,
-      "loss": 46.0,
-      "step": 17382
-    },
-    {
-      "epoch": 2.799388059100608,
-      "grad_norm": 0.004199116490781307,
-      "learning_rate": 0.00019999613748872941,
-      "loss": 46.0,
-      "step": 17383
-    },
-    {
-      "epoch": 2.7995490961793954,
-      "grad_norm": 0.005086650140583515,
-      "learning_rate": 0.00019999613704406265,
-      "loss": 46.0,
-      "step": 17384
-    },
-    {
-      "epoch": 2.799710133258183,
-      "grad_norm": 0.0013988835271447897,
-      "learning_rate": 0.00019999613659937032,
-      "loss": 46.0,
-      "step": 17385
-    },
-    {
-      "epoch": 2.79987117033697,
-      "grad_norm": 0.0007828865782357752,
-      "learning_rate": 0.0001999961361546524,
-      "loss": 46.0,
-      "step": 17386
-    },
-    {
-      "epoch": 2.8000322074157573,
-      "grad_norm": 0.006819195579737425,
-      "learning_rate": 0.00019999613570990891,
-      "loss": 46.0,
-      "step": 17387
-    },
-    {
-      "epoch": 2.8001932444945448,
-      "grad_norm": 0.0010173274204134941,
-      "learning_rate": 0.0001999961352651398,
-      "loss": 46.0,
-      "step": 17388
-    },
-    {
-      "epoch": 2.800354281573332,
-      "grad_norm": 0.00251241447404027,
-      "learning_rate": 0.0001999961348203451,
-      "loss": 46.0,
-      "step": 17389
-    },
-    {
-      "epoch": 2.8005153186521197,
-      "grad_norm": 0.0016244055004790425,
-      "learning_rate": 0.00019999613437552481,
-      "loss": 46.0,
-      "step": 17390
-    },
-    {
-      "epoch": 2.800676355730907,
-      "grad_norm": 0.0047190189361572266,
-      "learning_rate": 0.00019999613393067894,
-      "loss": 46.0,
-      "step": 17391
-    },
-    {
-      "epoch": 2.8008373928096946,
-      "grad_norm": 0.0026683418545871973,
-      "learning_rate": 0.00019999613348580742,
-      "loss": 46.0,
-      "step": 17392
-    },
-    {
-      "epoch": 2.800998429888482,
-      "grad_norm": 0.005497898440808058,
-      "learning_rate": 0.00019999613304091035,
-      "loss": 46.0,
-      "step": 17393
-    },
-    {
-      "epoch": 2.8011594669672695,
-      "grad_norm": 0.0013796283164992929,
-      "learning_rate": 0.00019999613259598768,
-      "loss": 46.0,
-      "step": 17394
-    },
-    {
-      "epoch": 2.8013205040460565,
-      "grad_norm": 0.001371831982396543,
-      "learning_rate": 0.00019999613215103943,
-      "loss": 46.0,
-      "step": 17395
-    },
-    {
-      "epoch": 2.801481541124844,
-      "grad_norm": 0.002239825204014778,
-      "learning_rate": 0.00019999613170606557,
-      "loss": 46.0,
-      "step": 17396
-    },
-    {
-      "epoch": 2.8016425782036314,
-      "grad_norm": 0.0028003628831356764,
-      "learning_rate": 0.00019999613126106612,
-      "loss": 46.0,
-      "step": 17397
-    },
-    {
-      "epoch": 2.801803615282419,
-      "grad_norm": 0.0028887924272567034,
-      "learning_rate": 0.00019999613081604108,
-      "loss": 46.0,
-      "step": 17398
-    },
-    {
-      "epoch": 2.8019646523612063,
-      "grad_norm": 0.007571985945105553,
-      "learning_rate": 0.00019999613037099043,
-      "loss": 46.0,
-      "step": 17399
-    },
-    {
-      "epoch": 2.8021256894399933,
-      "grad_norm": 0.007642506621778011,
-      "learning_rate": 0.00019999612992591419,
-      "loss": 46.0,
-      "step": 17400
-    },
-    {
-      "epoch": 2.8022867265187807,
-      "grad_norm": 0.014479024335741997,
-      "learning_rate": 0.00019999612948081236,
-      "loss": 46.0,
-      "step": 17401
-    },
-    {
-      "epoch": 2.802447763597568,
-      "grad_norm": 0.0010544572724029422,
-      "learning_rate": 0.00019999612903568494,
-      "loss": 46.0,
-      "step": 17402
-    },
-    {
-      "epoch": 2.8026088006763556,
-      "grad_norm": 0.0021329568699002266,
-      "learning_rate": 0.00019999612859053194,
-      "loss": 46.0,
-      "step": 17403
-    },
-    {
-      "epoch": 2.802769837755143,
-      "grad_norm": 0.0023929246235638857,
-      "learning_rate": 0.00019999612814535332,
-      "loss": 46.0,
-      "step": 17404
-    },
-    {
-      "epoch": 2.8029308748339306,
-      "grad_norm": 0.0013512979494407773,
-      "learning_rate": 0.0001999961277001491,
-      "loss": 46.0,
-      "step": 17405
-    },
-    {
-      "epoch": 2.803091911912718,
-      "grad_norm": 0.003351834136992693,
-      "learning_rate": 0.00019999612725491933,
-      "loss": 46.0,
-      "step": 17406
-    },
-    {
-      "epoch": 2.8032529489915055,
-      "grad_norm": 0.00213625468313694,
-      "learning_rate": 0.00019999612680966393,
-      "loss": 46.0,
-      "step": 17407
-    },
-    {
-      "epoch": 2.803413986070293,
-      "grad_norm": 0.0011837321799248457,
-      "learning_rate": 0.00019999612636438293,
-      "loss": 46.0,
-      "step": 17408
-    },
-    {
-      "epoch": 2.80357502314908,
-      "grad_norm": 0.0022808159701526165,
-      "learning_rate": 0.00019999612591907635,
-      "loss": 46.0,
-      "step": 17409
-    },
-    {
-      "epoch": 2.8037360602278674,
-      "grad_norm": 0.006732139736413956,
-      "learning_rate": 0.0001999961254737442,
-      "loss": 46.0,
-      "step": 17410
-    },
-    {
-      "epoch": 2.803897097306655,
-      "grad_norm": 0.000943644205108285,
-      "learning_rate": 0.00019999612502838643,
-      "loss": 46.0,
-      "step": 17411
-    },
-    {
-      "epoch": 2.8040581343854423,
-      "grad_norm": 0.0026921082753688097,
-      "learning_rate": 0.00019999612458300304,
-      "loss": 46.0,
-      "step": 17412
-    },
-    {
-      "epoch": 2.8042191714642297,
-      "grad_norm": 0.0054645235650241375,
-      "learning_rate": 0.00019999612413759409,
-      "loss": 46.0,
-      "step": 17413
-    },
-    {
-      "epoch": 2.804380208543017,
-      "grad_norm": 0.009223084896802902,
-      "learning_rate": 0.00019999612369215957,
-      "loss": 46.0,
-      "step": 17414
-    },
-    {
-      "epoch": 2.804541245621804,
-      "grad_norm": 0.0013247141614556313,
-      "learning_rate": 0.00019999612324669941,
-      "loss": 46.0,
-      "step": 17415
-    },
-    {
-      "epoch": 2.8047022827005916,
-      "grad_norm": 0.0039679259061813354,
-      "learning_rate": 0.00019999612280121367,
-      "loss": 46.0,
-      "step": 17416
-    },
-    {
-      "epoch": 2.804863319779379,
-      "grad_norm": 0.010341739282011986,
-      "learning_rate": 0.00019999612235570237,
-      "loss": 46.0,
-      "step": 17417
-    },
-    {
-      "epoch": 2.8050243568581665,
-      "grad_norm": 0.00178498774766922,
-      "learning_rate": 0.00019999612191016542,
-      "loss": 46.0,
-      "step": 17418
-    },
-    {
-      "epoch": 2.805185393936954,
-      "grad_norm": 0.002988457912579179,
-      "learning_rate": 0.00019999612146460292,
-      "loss": 46.0,
-      "step": 17419
-    },
-    {
-      "epoch": 2.8053464310157414,
-      "grad_norm": 0.014675927348434925,
-      "learning_rate": 0.0001999961210190148,
-      "loss": 46.0,
-      "step": 17420
-    },
-    {
-      "epoch": 2.805507468094529,
-      "grad_norm": 0.0009310987661592662,
-      "learning_rate": 0.0001999961205734011,
-      "loss": 46.0,
-      "step": 17421
-    },
-    {
-      "epoch": 2.8056685051733163,
-      "grad_norm": 0.0076582725159823895,
-      "learning_rate": 0.0001999961201277618,
-      "loss": 46.0,
-      "step": 17422
-    },
-    {
-      "epoch": 2.805829542252104,
-      "grad_norm": 0.0009604257065802813,
-      "learning_rate": 0.00019999611968209692,
-      "loss": 46.0,
-      "step": 17423
-    },
-    {
-      "epoch": 2.805990579330891,
-      "grad_norm": 0.0016388790681958199,
-      "learning_rate": 0.00019999611923640642,
-      "loss": 46.0,
-      "step": 17424
-    },
-    {
-      "epoch": 2.8061516164096783,
-      "grad_norm": 0.009259761311113834,
-      "learning_rate": 0.00019999611879069034,
-      "loss": 46.0,
-      "step": 17425
-    },
-    {
-      "epoch": 2.8063126534884657,
-      "grad_norm": 0.009581991471350193,
-      "learning_rate": 0.00019999611834494867,
-      "loss": 46.0,
-      "step": 17426
-    },
-    {
-      "epoch": 2.806473690567253,
-      "grad_norm": 0.0010605021379888058,
-      "learning_rate": 0.00019999611789918139,
-      "loss": 46.0,
-      "step": 17427
-    },
-    {
-      "epoch": 2.8066347276460406,
-      "grad_norm": 0.01231616735458374,
-      "learning_rate": 0.00019999611745338854,
-      "loss": 46.0,
-      "step": 17428
-    },
-    {
-      "epoch": 2.806795764724828,
-      "grad_norm": 0.004481412470340729,
-      "learning_rate": 0.00019999611700757006,
-      "loss": 46.0,
-      "step": 17429
-    },
-    {
-      "epoch": 2.806956801803615,
-      "grad_norm": 0.001579686300829053,
-      "learning_rate": 0.00019999611656172604,
-      "loss": 46.0,
-      "step": 17430
-    },
-    {
-      "epoch": 2.8071178388824025,
-      "grad_norm": 0.0026307052467018366,
-      "learning_rate": 0.00019999611611585638,
-      "loss": 46.0,
-      "step": 17431
-    },
-    {
-      "epoch": 2.80727887596119,
-      "grad_norm": 0.0011595413088798523,
-      "learning_rate": 0.00019999611566996116,
-      "loss": 46.0,
-      "step": 17432
-    },
-    {
-      "epoch": 2.8074399130399774,
-      "grad_norm": 0.004979744087904692,
-      "learning_rate": 0.00019999611522404033,
-      "loss": 46.0,
-      "step": 17433
-    },
-    {
-      "epoch": 2.807600950118765,
-      "grad_norm": 0.0010374952107667923,
-      "learning_rate": 0.00019999611477809388,
-      "loss": 46.0,
-      "step": 17434
-    },
-    {
-      "epoch": 2.8077619871975523,
-      "grad_norm": 0.0011462147813290358,
-      "learning_rate": 0.00019999611433212187,
-      "loss": 46.0,
-      "step": 17435
-    },
-    {
-      "epoch": 2.8079230242763398,
-      "grad_norm": 0.0030980510637164116,
-      "learning_rate": 0.00019999611388612425,
-      "loss": 46.0,
-      "step": 17436
-    },
-    {
-      "epoch": 2.8080840613551272,
-      "grad_norm": 0.0029788825195282698,
-      "learning_rate": 0.00019999611344010104,
-      "loss": 46.0,
-      "step": 17437
-    },
-    {
-      "epoch": 2.8082450984339147,
-      "grad_norm": 0.0008766724495217204,
-      "learning_rate": 0.00019999611299405224,
-      "loss": 46.0,
-      "step": 17438
-    },
-    {
-      "epoch": 2.8084061355127017,
-      "grad_norm": 0.006660845596343279,
-      "learning_rate": 0.00019999611254797786,
-      "loss": 46.0,
-      "step": 17439
-    },
-    {
-      "epoch": 2.808567172591489,
-      "grad_norm": 0.002069348469376564,
-      "learning_rate": 0.00019999611210187786,
-      "loss": 46.0,
-      "step": 17440
-    },
-    {
-      "epoch": 2.8087282096702766,
-      "grad_norm": 0.0055584535002708435,
-      "learning_rate": 0.00019999611165575227,
-      "loss": 46.0,
-      "step": 17441
-    },
-    {
-      "epoch": 2.808889246749064,
-      "grad_norm": 0.0033744191750884056,
-      "learning_rate": 0.0001999961112096011,
-      "loss": 46.0,
-      "step": 17442
-    },
-    {
-      "epoch": 2.8090502838278515,
-      "grad_norm": 0.002229581121355295,
-      "learning_rate": 0.00019999611076342434,
-      "loss": 46.0,
-      "step": 17443
-    },
-    {
-      "epoch": 2.8092113209066385,
-      "grad_norm": 0.0027567625511437654,
-      "learning_rate": 0.00019999611031722197,
-      "loss": 46.0,
-      "step": 17444
-    },
-    {
-      "epoch": 2.809372357985426,
-      "grad_norm": 0.007182674016803503,
-      "learning_rate": 0.00019999610987099403,
-      "loss": 46.0,
-      "step": 17445
-    },
-    {
-      "epoch": 2.8095333950642134,
-      "grad_norm": 0.0012006674660369754,
-      "learning_rate": 0.00019999610942474046,
-      "loss": 46.0,
-      "step": 17446
-    },
-    {
-      "epoch": 2.809694432143001,
-      "grad_norm": 0.000705803744494915,
-      "learning_rate": 0.0001999961089784613,
-      "loss": 46.0,
-      "step": 17447
-    },
-    {
-      "epoch": 2.8098554692217883,
-      "grad_norm": 0.0032836601603776217,
-      "learning_rate": 0.00019999610853215657,
-      "loss": 46.0,
-      "step": 17448
-    },
-    {
-      "epoch": 2.8100165063005758,
-      "grad_norm": 0.0024060728028416634,
-      "learning_rate": 0.00019999610808582623,
-      "loss": 46.0,
-      "step": 17449
-    },
-    {
-      "epoch": 2.810177543379363,
-      "grad_norm": 0.0015321644023060799,
-      "learning_rate": 0.0001999961076394703,
-      "loss": 46.0,
-      "step": 17450
-    },
-    {
-      "epoch": 2.8103385804581507,
-      "grad_norm": 0.0024709589779376984,
-      "learning_rate": 0.0001999961071930888,
-      "loss": 46.0,
-      "step": 17451
-    },
-    {
-      "epoch": 2.810499617536938,
-      "grad_norm": 0.0026140264235436916,
-      "learning_rate": 0.00019999610674668164,
-      "loss": 46.0,
-      "step": 17452
-    },
-    {
-      "epoch": 2.810660654615725,
-      "grad_norm": 0.0013854526914656162,
-      "learning_rate": 0.00019999610630024896,
-      "loss": 46.0,
-      "step": 17453
-    },
-    {
-      "epoch": 2.8108216916945126,
-      "grad_norm": 0.002409999957308173,
-      "learning_rate": 0.00019999610585379066,
-      "loss": 46.0,
-      "step": 17454
-    },
-    {
-      "epoch": 2.8109827287733,
-      "grad_norm": 0.000760366499889642,
-      "learning_rate": 0.00019999610540730674,
-      "loss": 46.0,
-      "step": 17455
-    },
-    {
-      "epoch": 2.8111437658520875,
-      "grad_norm": 0.003204125678166747,
-      "learning_rate": 0.00019999610496079727,
-      "loss": 46.0,
-      "step": 17456
-    },
-    {
-      "epoch": 2.811304802930875,
-      "grad_norm": 0.002333524404093623,
-      "learning_rate": 0.00019999610451426218,
-      "loss": 46.0,
-      "step": 17457
-    },
-    {
-      "epoch": 2.8114658400096624,
-      "grad_norm": 0.006015934515744448,
-      "learning_rate": 0.0001999961040677015,
-      "loss": 46.0,
-      "step": 17458
-    },
-    {
-      "epoch": 2.8116268770884494,
-      "grad_norm": 0.005155366845428944,
-      "learning_rate": 0.00019999610362111522,
-      "loss": 46.0,
-      "step": 17459
-    },
-    {
-      "epoch": 2.811787914167237,
-      "grad_norm": 0.0031221392564475536,
-      "learning_rate": 0.00019999610317450337,
-      "loss": 46.0,
-      "step": 17460
-    },
-    {
-      "epoch": 2.8119489512460243,
-      "grad_norm": 0.008233534172177315,
-      "learning_rate": 0.0001999961027278659,
-      "loss": 46.0,
-      "step": 17461
-    },
-    {
-      "epoch": 2.8121099883248117,
-      "grad_norm": 0.0010821737814694643,
-      "learning_rate": 0.00019999610228120288,
-      "loss": 46.0,
-      "step": 17462
-    },
-    {
-      "epoch": 2.812271025403599,
-      "grad_norm": 0.003293958492577076,
-      "learning_rate": 0.0001999961018345142,
-      "loss": 46.0,
-      "step": 17463
-    },
-    {
-      "epoch": 2.8124320624823866,
-      "grad_norm": 0.0023541077971458435,
-      "learning_rate": 0.00019999610138779996,
-      "loss": 46.0,
-      "step": 17464
-    },
-    {
-      "epoch": 2.812593099561174,
-      "grad_norm": 0.0018221113132312894,
-      "learning_rate": 0.00019999610094106012,
-      "loss": 46.0,
-      "step": 17465
-    },
-    {
-      "epoch": 2.8127541366399615,
-      "grad_norm": 0.0005643051699735224,
-      "learning_rate": 0.0001999961004942947,
-      "loss": 46.0,
-      "step": 17466
-    },
-    {
-      "epoch": 2.812915173718749,
-      "grad_norm": 0.0006361603154800832,
-      "learning_rate": 0.00019999610004750368,
-      "loss": 46.0,
-      "step": 17467
-    },
-    {
-      "epoch": 2.813076210797536,
-      "grad_norm": 0.012806505896151066,
-      "learning_rate": 0.00019999609960068705,
-      "loss": 46.0,
-      "step": 17468
-    },
-    {
-      "epoch": 2.8132372478763235,
-      "grad_norm": 0.005844034720212221,
-      "learning_rate": 0.00019999609915384486,
-      "loss": 46.0,
-      "step": 17469
-    },
-    {
-      "epoch": 2.813398284955111,
-      "grad_norm": 0.0005421608802862465,
-      "learning_rate": 0.00019999609870697705,
-      "loss": 46.0,
-      "step": 17470
-    },
-    {
-      "epoch": 2.8135593220338984,
-      "grad_norm": 0.004232186358422041,
-      "learning_rate": 0.00019999609826008366,
-      "loss": 46.0,
-      "step": 17471
-    },
-    {
-      "epoch": 2.813720359112686,
-      "grad_norm": 0.0008606420014984906,
-      "learning_rate": 0.00019999609781316466,
-      "loss": 46.0,
-      "step": 17472
-    },
-    {
-      "epoch": 2.813881396191473,
-      "grad_norm": 0.0009822994470596313,
-      "learning_rate": 0.00019999609736622004,
-      "loss": 46.0,
-      "step": 17473
-    },
-    {
-      "epoch": 2.8140424332702603,
-      "grad_norm": 0.002020971616730094,
-      "learning_rate": 0.0001999960969192499,
-      "loss": 46.0,
-      "step": 17474
-    },
-    {
-      "epoch": 2.8142034703490477,
-      "grad_norm": 0.0010751087684184313,
-      "learning_rate": 0.0001999960964722541,
-      "loss": 46.0,
-      "step": 17475
-    },
-    {
-      "epoch": 2.814364507427835,
-      "grad_norm": 0.006463815923780203,
-      "learning_rate": 0.00019999609602523272,
-      "loss": 46.0,
-      "step": 17476
-    },
-    {
-      "epoch": 2.8145255445066226,
-      "grad_norm": 0.0031545457895845175,
-      "learning_rate": 0.00019999609557818577,
-      "loss": 46.0,
-      "step": 17477
-    },
-    {
-      "epoch": 2.81468658158541,
-      "grad_norm": 0.001236637239344418,
-      "learning_rate": 0.00019999609513111322,
-      "loss": 46.0,
-      "step": 17478
-    },
-    {
-      "epoch": 2.8148476186641975,
-      "grad_norm": 0.004620371852070093,
-      "learning_rate": 0.00019999609468401505,
-      "loss": 46.0,
-      "step": 17479
-    },
-    {
-      "epoch": 2.815008655742985,
-      "grad_norm": 0.00397595064714551,
-      "learning_rate": 0.00019999609423689135,
-      "loss": 46.0,
-      "step": 17480
-    },
-    {
-      "epoch": 2.8151696928217724,
-      "grad_norm": 0.004133298061788082,
-      "learning_rate": 0.000199996093789742,
-      "loss": 46.0,
-      "step": 17481
-    },
-    {
-      "epoch": 2.81533072990056,
-      "grad_norm": 0.0071642352268099785,
-      "learning_rate": 0.00019999609334256705,
-      "loss": 46.0,
-      "step": 17482
-    },
-    {
-      "epoch": 2.815491766979347,
-      "grad_norm": 0.0074384515173733234,
-      "learning_rate": 0.00019999609289536653,
-      "loss": 46.0,
-      "step": 17483
-    },
-    {
-      "epoch": 2.8156528040581343,
-      "grad_norm": 0.0032049394212663174,
-      "learning_rate": 0.00019999609244814043,
-      "loss": 46.0,
-      "step": 17484
-    },
-    {
-      "epoch": 2.815813841136922,
-      "grad_norm": 0.0021838336251676083,
-      "learning_rate": 0.0001999960920008887,
-      "loss": 46.0,
-      "step": 17485
-    },
-    {
-      "epoch": 2.8159748782157092,
-      "grad_norm": 0.008074485696852207,
-      "learning_rate": 0.0001999960915536114,
-      "loss": 46.0,
-      "step": 17486
-    },
-    {
-      "epoch": 2.8161359152944967,
-      "grad_norm": 0.006901143118739128,
-      "learning_rate": 0.00019999609110630848,
-      "loss": 46.0,
-      "step": 17487
-    },
-    {
-      "epoch": 2.8162969523732837,
-      "grad_norm": 0.0017682092729955912,
-      "learning_rate": 0.00019999609065898,
-      "loss": 46.0,
-      "step": 17488
-    },
-    {
-      "epoch": 2.816457989452071,
-      "grad_norm": 0.00814884528517723,
-      "learning_rate": 0.0001999960902116259,
-      "loss": 46.0,
-      "step": 17489
-    },
-    {
-      "epoch": 2.8166190265308586,
-      "grad_norm": 0.0044253915548324585,
-      "learning_rate": 0.00019999608976424622,
-      "loss": 46.0,
-      "step": 17490
-    },
-    {
-      "epoch": 2.816780063609646,
-      "grad_norm": 0.006440346594899893,
-      "learning_rate": 0.00019999608931684095,
-      "loss": 46.0,
-      "step": 17491
-    },
-    {
-      "epoch": 2.8169411006884335,
-      "grad_norm": 0.0031027065124362707,
-      "learning_rate": 0.00019999608886941007,
-      "loss": 46.0,
-      "step": 17492
-    },
-    {
-      "epoch": 2.817102137767221,
-      "grad_norm": 0.00102567533031106,
-      "learning_rate": 0.0001999960884219536,
-      "loss": 46.0,
-      "step": 17493
-    },
-    {
-      "epoch": 2.8172631748460084,
-      "grad_norm": 0.012003621086478233,
-      "learning_rate": 0.00019999608797447157,
-      "loss": 46.0,
-      "step": 17494
-    },
-    {
-      "epoch": 2.817424211924796,
-      "grad_norm": 0.003172435099259019,
-      "learning_rate": 0.0001999960875269639,
-      "loss": 46.0,
-      "step": 17495
-    },
-    {
-      "epoch": 2.8175852490035833,
-      "grad_norm": 0.0009293970069848001,
-      "learning_rate": 0.00019999608707943066,
-      "loss": 46.0,
-      "step": 17496
-    },
-    {
-      "epoch": 2.8177462860823703,
-      "grad_norm": 0.001602035015821457,
-      "learning_rate": 0.00019999608663187182,
-      "loss": 46.0,
-      "step": 17497
-    },
-    {
-      "epoch": 2.8179073231611578,
-      "grad_norm": 0.0015408919425681233,
-      "learning_rate": 0.00019999608618428738,
-      "loss": 46.0,
-      "step": 17498
-    },
-    {
-      "epoch": 2.8180683602399452,
-      "grad_norm": 0.0017288177041336894,
-      "learning_rate": 0.00019999608573667736,
-      "loss": 46.0,
-      "step": 17499
-    },
-    {
-      "epoch": 2.8182293973187327,
-      "grad_norm": 0.006491424515843391,
-      "learning_rate": 0.00019999608528904173,
-      "loss": 46.0,
-      "step": 17500
-    },
-    {
-      "epoch": 2.81839043439752,
-      "grad_norm": 0.005350694991648197,
-      "learning_rate": 0.00019999608484138053,
-      "loss": 46.0,
-      "step": 17501
-    },
-    {
-      "epoch": 2.8185514714763076,
-      "grad_norm": 0.0028305589221417904,
-      "learning_rate": 0.0001999960843936937,
-      "loss": 46.0,
-      "step": 17502
-    },
-    {
-      "epoch": 2.8187125085550946,
-      "grad_norm": 0.0009118483867496252,
-      "learning_rate": 0.0001999960839459813,
-      "loss": 46.0,
-      "step": 17503
-    },
-    {
-      "epoch": 2.818873545633882,
-      "grad_norm": 0.0023907534778118134,
-      "learning_rate": 0.00019999608349824331,
-      "loss": 46.0,
-      "step": 17504
-    },
-    {
-      "epoch": 2.8190345827126695,
-      "grad_norm": 0.002552842488512397,
-      "learning_rate": 0.00019999608305047972,
-      "loss": 46.0,
-      "step": 17505
-    },
-    {
-      "epoch": 2.819195619791457,
-      "grad_norm": 0.011367013677954674,
-      "learning_rate": 0.00019999608260269053,
-      "loss": 46.0,
-      "step": 17506
-    },
-    {
-      "epoch": 2.8193566568702444,
-      "grad_norm": 0.000869512849021703,
-      "learning_rate": 0.00019999608215487579,
-      "loss": 46.0,
-      "step": 17507
-    },
-    {
-      "epoch": 2.819517693949032,
-      "grad_norm": 0.01703316532075405,
-      "learning_rate": 0.0001999960817070354,
-      "loss": 46.0,
-      "step": 17508
-    },
-    {
-      "epoch": 2.8196787310278193,
-      "grad_norm": 0.0046666469424963,
-      "learning_rate": 0.00019999608125916942,
-      "loss": 46.0,
-      "step": 17509
-    },
-    {
-      "epoch": 2.8198397681066067,
-      "grad_norm": 0.0012241367949172854,
-      "learning_rate": 0.00019999608081127786,
-      "loss": 46.0,
-      "step": 17510
-    },
-    {
-      "epoch": 2.820000805185394,
-      "grad_norm": 0.004727122373878956,
-      "learning_rate": 0.00019999608036336072,
-      "loss": 46.0,
-      "step": 17511
-    },
-    {
-      "epoch": 2.820161842264181,
-      "grad_norm": 0.0017181276343762875,
-      "learning_rate": 0.00019999607991541795,
-      "loss": 46.0,
-      "step": 17512
-    },
-    {
-      "epoch": 2.8203228793429687,
-      "grad_norm": 0.0038001309148967266,
-      "learning_rate": 0.00019999607946744963,
-      "loss": 46.0,
-      "step": 17513
-    },
-    {
-      "epoch": 2.820483916421756,
-      "grad_norm": 0.0056593157351017,
-      "learning_rate": 0.00019999607901945567,
-      "loss": 46.0,
-      "step": 17514
-    },
-    {
-      "epoch": 2.8206449535005436,
-      "grad_norm": 0.005102424882352352,
-      "learning_rate": 0.00019999607857143614,
-      "loss": 46.0,
-      "step": 17515
-    },
-    {
-      "epoch": 2.820805990579331,
-      "grad_norm": 0.00345412059687078,
-      "learning_rate": 0.00019999607812339103,
-      "loss": 46.0,
-      "step": 17516
-    },
-    {
-      "epoch": 2.820967027658118,
-      "grad_norm": 0.0009473130339756608,
-      "learning_rate": 0.0001999960776753203,
-      "loss": 46.0,
-      "step": 17517
-    },
-    {
-      "epoch": 2.8211280647369055,
-      "grad_norm": 0.0008056455408222973,
-      "learning_rate": 0.000199996077227224,
-      "loss": 46.0,
-      "step": 17518
-    },
-    {
-      "epoch": 2.821289101815693,
-      "grad_norm": 0.0019042566418647766,
-      "learning_rate": 0.0001999960767791021,
-      "loss": 46.0,
-      "step": 17519
-    },
-    {
-      "epoch": 2.8214501388944804,
-      "grad_norm": 0.0017730704275891185,
-      "learning_rate": 0.0001999960763309546,
-      "loss": 46.0,
-      "step": 17520
-    },
-    {
-      "epoch": 2.821611175973268,
-      "grad_norm": 0.001036213943734765,
-      "learning_rate": 0.0001999960758827815,
-      "loss": 46.0,
-      "step": 17521
-    },
-    {
-      "epoch": 2.8217722130520553,
-      "grad_norm": 0.016822606325149536,
-      "learning_rate": 0.00019999607543458284,
-      "loss": 46.0,
-      "step": 17522
-    },
-    {
-      "epoch": 2.8219332501308427,
-      "grad_norm": 0.0023057369980961084,
-      "learning_rate": 0.00019999607498635854,
-      "loss": 46.0,
-      "step": 17523
-    },
-    {
-      "epoch": 2.82209428720963,
-      "grad_norm": 0.00485507445409894,
-      "learning_rate": 0.00019999607453810868,
-      "loss": 46.0,
-      "step": 17524
-    },
-    {
-      "epoch": 2.8222553242884176,
-      "grad_norm": 0.004225747659802437,
-      "learning_rate": 0.00019999607408983317,
-      "loss": 46.0,
-      "step": 17525
-    },
-    {
-      "epoch": 2.8224163613672046,
-      "grad_norm": 0.010278395377099514,
-      "learning_rate": 0.00019999607364153214,
-      "loss": 46.0,
-      "step": 17526
-    },
-    {
-      "epoch": 2.822577398445992,
-      "grad_norm": 0.0015109482919797301,
-      "learning_rate": 0.00019999607319320548,
-      "loss": 46.0,
-      "step": 17527
-    },
-    {
-      "epoch": 2.8227384355247795,
-      "grad_norm": 0.007660284638404846,
-      "learning_rate": 0.00019999607274485325,
-      "loss": 46.0,
-      "step": 17528
-    },
-    {
-      "epoch": 2.822899472603567,
-      "grad_norm": 0.00685298815369606,
-      "learning_rate": 0.0001999960722964754,
-      "loss": 46.0,
-      "step": 17529
-    },
-    {
-      "epoch": 2.8230605096823544,
-      "grad_norm": 0.0014923186972737312,
-      "learning_rate": 0.00019999607184807195,
-      "loss": 46.0,
-      "step": 17530
-    },
-    {
-      "epoch": 2.823221546761142,
-      "grad_norm": 0.0020321037154644728,
-      "learning_rate": 0.00019999607139964293,
-      "loss": 46.0,
-      "step": 17531
-    },
-    {
-      "epoch": 2.823382583839929,
-      "grad_norm": 0.0030750269070267677,
-      "learning_rate": 0.00019999607095118828,
-      "loss": 46.0,
-      "step": 17532
-    },
-    {
-      "epoch": 2.8235436209187164,
-      "grad_norm": 0.006575407925993204,
-      "learning_rate": 0.00019999607050270808,
-      "loss": 46.0,
-      "step": 17533
-    },
-    {
-      "epoch": 2.823704657997504,
-      "grad_norm": 0.0012792469933629036,
-      "learning_rate": 0.00019999607005420227,
-      "loss": 46.0,
-      "step": 17534
-    },
-    {
-      "epoch": 2.8238656950762913,
-      "grad_norm": 0.0046748509630560875,
-      "learning_rate": 0.00019999606960567086,
-      "loss": 46.0,
-      "step": 17535
-    },
-    {
-      "epoch": 2.8240267321550787,
-      "grad_norm": 0.0030210623517632484,
-      "learning_rate": 0.00019999606915711387,
-      "loss": 46.0,
-      "step": 17536
-    },
-    {
-      "epoch": 2.824187769233866,
-      "grad_norm": 0.003818291937932372,
-      "learning_rate": 0.00019999606870853127,
-      "loss": 46.0,
-      "step": 17537
-    },
-    {
-      "epoch": 2.8243488063126536,
-      "grad_norm": 0.0031532743014395237,
-      "learning_rate": 0.00019999606825992308,
-      "loss": 46.0,
-      "step": 17538
-    },
-    {
-      "epoch": 2.824509843391441,
-      "grad_norm": 0.0021734449546784163,
-      "learning_rate": 0.0001999960678112893,
-      "loss": 46.0,
-      "step": 17539
-    },
-    {
-      "epoch": 2.8246708804702285,
-      "grad_norm": 0.0036230776458978653,
-      "learning_rate": 0.00019999606736262993,
-      "loss": 46.0,
-      "step": 17540
-    },
-    {
-      "epoch": 2.8248319175490155,
-      "grad_norm": 0.0009043588652275503,
-      "learning_rate": 0.00019999606691394495,
-      "loss": 46.0,
-      "step": 17541
-    },
-    {
-      "epoch": 2.824992954627803,
-      "grad_norm": 0.0036086265463382006,
-      "learning_rate": 0.00019999606646523439,
-      "loss": 46.0,
-      "step": 17542
-    },
-    {
-      "epoch": 2.8251539917065904,
-      "grad_norm": 0.0015189432306215167,
-      "learning_rate": 0.00019999606601649823,
-      "loss": 46.0,
-      "step": 17543
-    },
-    {
-      "epoch": 2.825315028785378,
-      "grad_norm": 0.0013452733401209116,
-      "learning_rate": 0.00019999606556773646,
-      "loss": 46.0,
-      "step": 17544
-    },
-    {
-      "epoch": 2.8254760658641653,
-      "grad_norm": 0.011437585577368736,
-      "learning_rate": 0.00019999606511894913,
-      "loss": 46.0,
-      "step": 17545
-    },
-    {
-      "epoch": 2.8256371029429523,
-      "grad_norm": 0.005635588895529509,
-      "learning_rate": 0.00019999606467013622,
-      "loss": 46.0,
-      "step": 17546
-    },
-    {
-      "epoch": 2.82579814002174,
-      "grad_norm": 0.004393447656184435,
-      "learning_rate": 0.00019999606422129766,
-      "loss": 46.0,
-      "step": 17547
-    },
-    {
-      "epoch": 2.8259591771005272,
-      "grad_norm": 0.00277325720526278,
-      "learning_rate": 0.00019999606377243354,
-      "loss": 46.0,
-      "step": 17548
-    },
-    {
-      "epoch": 2.8261202141793147,
-      "grad_norm": 0.004375135991722345,
-      "learning_rate": 0.0001999960633235438,
-      "loss": 46.0,
-      "step": 17549
-    },
-    {
-      "epoch": 2.826281251258102,
-      "grad_norm": 0.006475480273365974,
-      "learning_rate": 0.00019999606287462852,
-      "loss": 46.0,
-      "step": 17550
-    },
-    {
-      "epoch": 2.8264422883368896,
-      "grad_norm": 0.011471494100987911,
-      "learning_rate": 0.00019999606242568759,
-      "loss": 46.0,
-      "step": 17551
-    },
-    {
-      "epoch": 2.826603325415677,
-      "grad_norm": 0.0025683369021862745,
-      "learning_rate": 0.0001999960619767211,
-      "loss": 46.0,
-      "step": 17552
-    },
-    {
-      "epoch": 2.8267643624944645,
-      "grad_norm": 0.0005862105754204094,
-      "learning_rate": 0.00019999606152772898,
-      "loss": 46.0,
-      "step": 17553
-    },
-    {
-      "epoch": 2.826925399573252,
-      "grad_norm": 0.0005172690143808722,
-      "learning_rate": 0.00019999606107871132,
-      "loss": 46.0,
-      "step": 17554
-    },
-    {
-      "epoch": 2.8270864366520394,
-      "grad_norm": 0.007394160609692335,
-      "learning_rate": 0.00019999606062966803,
-      "loss": 46.0,
-      "step": 17555
-    },
-    {
-      "epoch": 2.8272474737308264,
-      "grad_norm": 0.0019491255516186357,
-      "learning_rate": 0.00019999606018059917,
-      "loss": 46.0,
-      "step": 17556
-    },
-    {
-      "epoch": 2.827408510809614,
-      "grad_norm": 0.001105102594010532,
-      "learning_rate": 0.00019999605973150466,
-      "loss": 46.0,
-      "step": 17557
-    },
-    {
-      "epoch": 2.8275695478884013,
-      "grad_norm": 0.001430126721970737,
-      "learning_rate": 0.0001999960592823846,
-      "loss": 46.0,
-      "step": 17558
-    },
-    {
-      "epoch": 2.8277305849671888,
-      "grad_norm": 0.0014700109604746103,
-      "learning_rate": 0.00019999605883323893,
-      "loss": 46.0,
-      "step": 17559
-    },
-    {
-      "epoch": 2.827891622045976,
-      "grad_norm": 0.004617826547473669,
-      "learning_rate": 0.0001999960583840677,
-      "loss": 46.0,
-      "step": 17560
-    },
-    {
-      "epoch": 2.8280526591247632,
-      "grad_norm": 0.001922737923450768,
-      "learning_rate": 0.00019999605793487085,
-      "loss": 46.0,
-      "step": 17561
-    },
-    {
-      "epoch": 2.8282136962035507,
-      "grad_norm": 0.0019020611653104424,
-      "learning_rate": 0.0001999960574856484,
-      "loss": 46.0,
-      "step": 17562
-    },
-    {
-      "epoch": 2.828374733282338,
-      "grad_norm": 0.0010919132037088275,
-      "learning_rate": 0.0001999960570364004,
-      "loss": 46.0,
-      "step": 17563
-    },
-    {
-      "epoch": 2.8285357703611256,
-      "grad_norm": 0.0024385685101151466,
-      "learning_rate": 0.00019999605658712675,
-      "loss": 46.0,
-      "step": 17564
-    },
-    {
-      "epoch": 2.828696807439913,
-      "grad_norm": 0.0008681168546900153,
-      "learning_rate": 0.00019999605613782751,
-      "loss": 46.0,
-      "step": 17565
-    },
-    {
-      "epoch": 2.8288578445187005,
-      "grad_norm": 0.00507500721141696,
-      "learning_rate": 0.0001999960556885027,
-      "loss": 46.0,
-      "step": 17566
-    },
-    {
-      "epoch": 2.829018881597488,
-      "grad_norm": 0.0032564294524490833,
-      "learning_rate": 0.0001999960552391523,
-      "loss": 46.0,
-      "step": 17567
-    },
-    {
-      "epoch": 2.8291799186762754,
-      "grad_norm": 0.001755494624376297,
-      "learning_rate": 0.0001999960547897763,
-      "loss": 46.0,
-      "step": 17568
-    },
-    {
-      "epoch": 2.829340955755063,
-      "grad_norm": 0.007341737858951092,
-      "learning_rate": 0.0001999960543403747,
-      "loss": 46.0,
-      "step": 17569
-    },
-    {
-      "epoch": 2.82950199283385,
-      "grad_norm": 0.014459128491580486,
-      "learning_rate": 0.0001999960538909475,
-      "loss": 46.0,
-      "step": 17570
-    },
-    {
-      "epoch": 2.8296630299126373,
-      "grad_norm": 0.005415915511548519,
-      "learning_rate": 0.00019999605344149472,
-      "loss": 46.0,
-      "step": 17571
-    },
-    {
-      "epoch": 2.8298240669914247,
-      "grad_norm": 0.0010363005567342043,
-      "learning_rate": 0.00019999605299201635,
-      "loss": 46.0,
-      "step": 17572
-    },
-    {
-      "epoch": 2.829985104070212,
-      "grad_norm": 0.0017835585167631507,
-      "learning_rate": 0.00019999605254251236,
-      "loss": 46.0,
-      "step": 17573
-    },
-    {
-      "epoch": 2.8301461411489997,
-      "grad_norm": 0.004828247707337141,
-      "learning_rate": 0.0001999960520929828,
-      "loss": 46.0,
-      "step": 17574
-    },
-    {
-      "epoch": 2.830307178227787,
-      "grad_norm": 0.005219201557338238,
-      "learning_rate": 0.00019999605164342765,
-      "loss": 46.0,
-      "step": 17575
-    },
-    {
-      "epoch": 2.830468215306574,
-      "grad_norm": 0.005991618148982525,
-      "learning_rate": 0.00019999605119384688,
-      "loss": 46.0,
-      "step": 17576
-    },
-    {
-      "epoch": 2.8306292523853616,
-      "grad_norm": 0.005289238411933184,
-      "learning_rate": 0.00019999605074424052,
-      "loss": 46.0,
-      "step": 17577
-    },
-    {
-      "epoch": 2.830790289464149,
-      "grad_norm": 0.0008138579432852566,
-      "learning_rate": 0.0001999960502946086,
-      "loss": 46.0,
-      "step": 17578
-    },
-    {
-      "epoch": 2.8309513265429365,
-      "grad_norm": 0.005250999238342047,
-      "learning_rate": 0.00019999604984495103,
-      "loss": 46.0,
-      "step": 17579
-    },
-    {
-      "epoch": 2.831112363621724,
-      "grad_norm": 0.0007484708330594003,
-      "learning_rate": 0.00019999604939526793,
-      "loss": 46.0,
-      "step": 17580
-    },
-    {
-      "epoch": 2.8312734007005114,
-      "grad_norm": 0.01723730005323887,
-      "learning_rate": 0.0001999960489455592,
-      "loss": 46.0,
-      "step": 17581
-    },
-    {
-      "epoch": 2.831434437779299,
-      "grad_norm": 0.0007146120187826455,
-      "learning_rate": 0.0001999960484958249,
-      "loss": 46.0,
-      "step": 17582
-    },
-    {
-      "epoch": 2.8315954748580863,
-      "grad_norm": 0.009761675260961056,
-      "learning_rate": 0.00019999604804606498,
-      "loss": 46.0,
-      "step": 17583
-    },
-    {
-      "epoch": 2.8317565119368737,
-      "grad_norm": 0.000902310770470649,
-      "learning_rate": 0.00019999604759627948,
-      "loss": 46.0,
-      "step": 17584
-    },
-    {
-      "epoch": 2.8319175490156607,
-      "grad_norm": 0.0028070963453501463,
-      "learning_rate": 0.00019999604714646834,
-      "loss": 46.0,
-      "step": 17585
-    },
-    {
-      "epoch": 2.832078586094448,
-      "grad_norm": 0.0011297531891614199,
-      "learning_rate": 0.00019999604669663164,
-      "loss": 46.0,
-      "step": 17586
-    },
-    {
-      "epoch": 2.8322396231732356,
-      "grad_norm": 0.0006591191049665213,
-      "learning_rate": 0.00019999604624676938,
-      "loss": 46.0,
-      "step": 17587
-    },
-    {
-      "epoch": 2.832400660252023,
-      "grad_norm": 0.005738638807088137,
-      "learning_rate": 0.0001999960457968815,
-      "loss": 46.0,
-      "step": 17588
-    },
-    {
-      "epoch": 2.8325616973308105,
-      "grad_norm": 0.0013909791596233845,
-      "learning_rate": 0.00019999604534696804,
-      "loss": 46.0,
-      "step": 17589
-    },
-    {
-      "epoch": 2.8327227344095975,
-      "grad_norm": 0.005785842891782522,
-      "learning_rate": 0.00019999604489702897,
-      "loss": 46.0,
-      "step": 17590
-    },
-    {
-      "epoch": 2.832883771488385,
-      "grad_norm": 0.0012799163814634085,
-      "learning_rate": 0.00019999604444706428,
-      "loss": 46.0,
-      "step": 17591
-    },
-    {
-      "epoch": 2.8330448085671724,
-      "grad_norm": 0.0015748272417113185,
-      "learning_rate": 0.00019999604399707402,
-      "loss": 46.0,
-      "step": 17592
-    },
-    {
-      "epoch": 2.83320584564596,
-      "grad_norm": 0.01842391863465309,
-      "learning_rate": 0.00019999604354705819,
-      "loss": 46.0,
-      "step": 17593
-    },
-    {
-      "epoch": 2.8333668827247473,
-      "grad_norm": 0.005611111875623465,
-      "learning_rate": 0.00019999604309701673,
-      "loss": 46.0,
-      "step": 17594
-    },
-    {
-      "epoch": 2.833527919803535,
-      "grad_norm": 0.003330002073198557,
-      "learning_rate": 0.00019999604264694972,
-      "loss": 46.0,
-      "step": 17595
-    },
-    {
-      "epoch": 2.8336889568823223,
-      "grad_norm": 0.0019941949285566807,
-      "learning_rate": 0.00019999604219685707,
-      "loss": 46.0,
-      "step": 17596
-    },
-    {
-      "epoch": 2.8338499939611097,
-      "grad_norm": 0.0060850828886032104,
-      "learning_rate": 0.00019999604174673885,
-      "loss": 46.0,
-      "step": 17597
-    },
-    {
-      "epoch": 2.834011031039897,
-      "grad_norm": 0.0014871172606945038,
-      "learning_rate": 0.00019999604129659502,
-      "loss": 46.0,
-      "step": 17598
-    },
-    {
-      "epoch": 2.834172068118684,
-      "grad_norm": 0.0020752577111124992,
-      "learning_rate": 0.0001999960408464256,
-      "loss": 46.0,
-      "step": 17599
-    },
-    {
-      "epoch": 2.8343331051974716,
-      "grad_norm": 0.0017564443405717611,
-      "learning_rate": 0.0001999960403962306,
-      "loss": 46.0,
-      "step": 17600
-    },
-    {
-      "epoch": 2.834494142276259,
-      "grad_norm": 0.002166620222851634,
-      "learning_rate": 0.00019999603994601,
-      "loss": 46.0,
-      "step": 17601
-    },
-    {
-      "epoch": 2.8346551793550465,
-      "grad_norm": 0.002695689210668206,
-      "learning_rate": 0.0001999960394957638,
-      "loss": 46.0,
-      "step": 17602
-    },
-    {
-      "epoch": 2.834816216433834,
-      "grad_norm": 0.0049211750738322735,
-      "learning_rate": 0.00019999603904549202,
-      "loss": 46.0,
-      "step": 17603
-    },
-    {
-      "epoch": 2.8349772535126214,
-      "grad_norm": 0.001928769052028656,
-      "learning_rate": 0.00019999603859519464,
-      "loss": 46.0,
-      "step": 17604
-    },
-    {
-      "epoch": 2.8351382905914084,
-      "grad_norm": 0.0035950620658695698,
-      "learning_rate": 0.00019999603814487168,
-      "loss": 46.0,
-      "step": 17605
-    },
-    {
-      "epoch": 2.835299327670196,
-      "grad_norm": 0.002129684202373028,
-      "learning_rate": 0.0001999960376945231,
-      "loss": 46.0,
-      "step": 17606
-    },
-    {
-      "epoch": 2.8354603647489833,
-      "grad_norm": 0.0011342416983097792,
-      "learning_rate": 0.00019999603724414893,
-      "loss": 46.0,
-      "step": 17607
-    },
-    {
-      "epoch": 2.835621401827771,
-      "grad_norm": 0.00642255088314414,
-      "learning_rate": 0.00019999603679374917,
-      "loss": 46.0,
-      "step": 17608
-    },
-    {
-      "epoch": 2.8357824389065582,
-      "grad_norm": 0.001449377159588039,
-      "learning_rate": 0.00019999603634332383,
-      "loss": 46.0,
-      "step": 17609
-    },
-    {
-      "epoch": 2.8359434759853457,
-      "grad_norm": 0.0029043208342045546,
-      "learning_rate": 0.00019999603589287288,
-      "loss": 46.0,
-      "step": 17610
-    },
-    {
-      "epoch": 2.836104513064133,
-      "grad_norm": 0.0011060646502301097,
-      "learning_rate": 0.00019999603544239633,
-      "loss": 46.0,
-      "step": 17611
-    },
-    {
-      "epoch": 2.8362655501429206,
-      "grad_norm": 0.0008838083012960851,
-      "learning_rate": 0.0001999960349918942,
-      "loss": 46.0,
-      "step": 17612
-    },
-    {
-      "epoch": 2.836426587221708,
-      "grad_norm": 0.013734808191657066,
-      "learning_rate": 0.00019999603454136646,
-      "loss": 46.0,
-      "step": 17613
-    },
-    {
-      "epoch": 2.836587624300495,
-      "grad_norm": 0.0023814975284039974,
-      "learning_rate": 0.00019999603409081315,
-      "loss": 46.0,
-      "step": 17614
-    },
-    {
-      "epoch": 2.8367486613792825,
-      "grad_norm": 0.0037989169359207153,
-      "learning_rate": 0.00019999603364023424,
-      "loss": 46.0,
-      "step": 17615
-    },
-    {
-      "epoch": 2.83690969845807,
-      "grad_norm": 0.010734109207987785,
-      "learning_rate": 0.00019999603318962973,
-      "loss": 46.0,
-      "step": 17616
-    },
-    {
-      "epoch": 2.8370707355368574,
-      "grad_norm": 0.003938655834645033,
-      "learning_rate": 0.00019999603273899964,
-      "loss": 46.0,
-      "step": 17617
-    },
-    {
-      "epoch": 2.837231772615645,
-      "grad_norm": 0.00327532016672194,
-      "learning_rate": 0.00019999603228834393,
-      "loss": 46.0,
-      "step": 17618
-    },
-    {
-      "epoch": 2.8373928096944323,
-      "grad_norm": 0.004189345054328442,
-      "learning_rate": 0.00019999603183766263,
-      "loss": 46.0,
-      "step": 17619
-    },
-    {
-      "epoch": 2.8375538467732193,
-      "grad_norm": 0.001494970521889627,
-      "learning_rate": 0.00019999603138695575,
-      "loss": 46.0,
-      "step": 17620
-    },
-    {
-      "epoch": 2.8377148838520068,
-      "grad_norm": 0.002382122678682208,
-      "learning_rate": 0.00019999603093622328,
-      "loss": 46.0,
-      "step": 17621
-    },
-    {
-      "epoch": 2.837875920930794,
-      "grad_norm": 0.0014835756737738848,
-      "learning_rate": 0.0001999960304854652,
-      "loss": 46.0,
-      "step": 17622
-    },
-    {
-      "epoch": 2.8380369580095817,
-      "grad_norm": 0.007294792681932449,
-      "learning_rate": 0.00019999603003468153,
-      "loss": 46.0,
-      "step": 17623
-    },
-    {
-      "epoch": 2.838197995088369,
-      "grad_norm": 0.0007438446045853198,
-      "learning_rate": 0.00019999602958387227,
-      "loss": 46.0,
-      "step": 17624
-    },
-    {
-      "epoch": 2.8383590321671566,
-      "grad_norm": 0.0023958845995366573,
-      "learning_rate": 0.00019999602913303743,
-      "loss": 46.0,
-      "step": 17625
-    },
-    {
-      "epoch": 2.838520069245944,
-      "grad_norm": 0.0011832868913188577,
-      "learning_rate": 0.00019999602868217697,
-      "loss": 46.0,
-      "step": 17626
-    },
-    {
-      "epoch": 2.8386811063247315,
-      "grad_norm": 0.002492837607860565,
-      "learning_rate": 0.00019999602823129092,
-      "loss": 46.0,
-      "step": 17627
-    },
-    {
-      "epoch": 2.838842143403519,
-      "grad_norm": 0.0073603675700724125,
-      "learning_rate": 0.00019999602778037932,
-      "loss": 46.0,
-      "step": 17628
-    },
-    {
-      "epoch": 2.839003180482306,
-      "grad_norm": 0.0014479657402262092,
-      "learning_rate": 0.00019999602732944207,
-      "loss": 46.0,
-      "step": 17629
-    },
-    {
-      "epoch": 2.8391642175610934,
-      "grad_norm": 0.0027166493237018585,
-      "learning_rate": 0.00019999602687847926,
-      "loss": 46.0,
-      "step": 17630
-    },
-    {
-      "epoch": 2.839325254639881,
-      "grad_norm": 0.001018046517856419,
-      "learning_rate": 0.00019999602642749084,
-      "loss": 46.0,
-      "step": 17631
-    },
-    {
-      "epoch": 2.8394862917186683,
-      "grad_norm": 0.0034496565349400043,
-      "learning_rate": 0.00019999602597647683,
-      "loss": 46.0,
-      "step": 17632
-    },
-    {
-      "epoch": 2.8396473287974557,
-      "grad_norm": 0.0017235990380868316,
-      "learning_rate": 0.0001999960255254372,
-      "loss": 46.0,
-      "step": 17633
-    },
-    {
-      "epoch": 2.8398083658762427,
-      "grad_norm": 0.005632879212498665,
-      "learning_rate": 0.00019999602507437202,
-      "loss": 46.0,
-      "step": 17634
-    },
-    {
-      "epoch": 2.83996940295503,
-      "grad_norm": 0.00149669055826962,
-      "learning_rate": 0.00019999602462328123,
-      "loss": 46.0,
-      "step": 17635
-    },
-    {
-      "epoch": 2.8401304400338176,
-      "grad_norm": 0.005096688400954008,
-      "learning_rate": 0.00019999602417216484,
-      "loss": 46.0,
-      "step": 17636
-    },
-    {
-      "epoch": 2.840291477112605,
-      "grad_norm": 0.005132357124239206,
-      "learning_rate": 0.00019999602372102287,
-      "loss": 46.0,
-      "step": 17637
-    },
-    {
-      "epoch": 2.8404525141913926,
-      "grad_norm": 0.006723702419549227,
-      "learning_rate": 0.00019999602326985528,
-      "loss": 46.0,
-      "step": 17638
-    },
-    {
-      "epoch": 2.84061355127018,
-      "grad_norm": 0.0012959045125171542,
-      "learning_rate": 0.0001999960228186621,
-      "loss": 46.0,
-      "step": 17639
-    },
-    {
-      "epoch": 2.8407745883489675,
-      "grad_norm": 0.001360506284981966,
-      "learning_rate": 0.00019999602236744335,
-      "loss": 46.0,
-      "step": 17640
-    },
-    {
-      "epoch": 2.840935625427755,
-      "grad_norm": 0.006803812924772501,
-      "learning_rate": 0.000199996021916199,
-      "loss": 46.0,
-      "step": 17641
-    },
-    {
-      "epoch": 2.8410966625065424,
-      "grad_norm": 0.004713117145001888,
-      "learning_rate": 0.00019999602146492906,
-      "loss": 46.0,
-      "step": 17642
-    },
-    {
-      "epoch": 2.8412576995853294,
-      "grad_norm": 0.00247765495441854,
-      "learning_rate": 0.00019999602101363352,
-      "loss": 46.0,
-      "step": 17643
-    },
-    {
-      "epoch": 2.841418736664117,
-      "grad_norm": 0.0069712125696241856,
-      "learning_rate": 0.00019999602056231238,
-      "loss": 46.0,
-      "step": 17644
-    },
-    {
-      "epoch": 2.8415797737429043,
-      "grad_norm": 0.003584443125873804,
-      "learning_rate": 0.00019999602011096563,
-      "loss": 46.0,
-      "step": 17645
-    },
-    {
-      "epoch": 2.8417408108216917,
-      "grad_norm": 0.0021740689408034086,
-      "learning_rate": 0.00019999601965959332,
-      "loss": 46.0,
-      "step": 17646
-    },
-    {
-      "epoch": 2.841901847900479,
-      "grad_norm": 0.0029067890718579292,
-      "learning_rate": 0.0001999960192081954,
-      "loss": 46.0,
-      "step": 17647
-    },
-    {
-      "epoch": 2.8420628849792666,
-      "grad_norm": 0.0024205760564655066,
-      "learning_rate": 0.0001999960187567719,
-      "loss": 46.0,
-      "step": 17648
-    },
-    {
-      "epoch": 2.8422239220580536,
-      "grad_norm": 0.005348925478756428,
-      "learning_rate": 0.00019999601830532278,
-      "loss": 46.0,
-      "step": 17649
-    },
-    {
-      "epoch": 2.842384959136841,
-      "grad_norm": 0.0009694330510683358,
-      "learning_rate": 0.00019999601785384807,
-      "loss": 46.0,
-      "step": 17650
-    },
-    {
-      "epoch": 2.8425459962156285,
-      "grad_norm": 0.0021761669777333736,
-      "learning_rate": 0.00019999601740234777,
-      "loss": 46.0,
-      "step": 17651
-    },
-    {
-      "epoch": 2.842707033294416,
-      "grad_norm": 0.00867735780775547,
-      "learning_rate": 0.00019999601695082188,
-      "loss": 46.0,
-      "step": 17652
-    },
-    {
-      "epoch": 2.8428680703732034,
-      "grad_norm": 0.004964365158230066,
-      "learning_rate": 0.00019999601649927043,
-      "loss": 46.0,
-      "step": 17653
-    },
-    {
-      "epoch": 2.843029107451991,
-      "grad_norm": 0.0013917895266786218,
-      "learning_rate": 0.00019999601604769334,
-      "loss": 46.0,
-      "step": 17654
-    },
-    {
-      "epoch": 2.8431901445307783,
-      "grad_norm": 0.006648889742791653,
-      "learning_rate": 0.00019999601559609066,
-      "loss": 46.0,
-      "step": 17655
-    },
-    {
-      "epoch": 2.843351181609566,
-      "grad_norm": 0.0009410177008248866,
-      "learning_rate": 0.00019999601514446243,
-      "loss": 46.0,
-      "step": 17656
-    },
-    {
-      "epoch": 2.8435122186883532,
-      "grad_norm": 0.00480504659935832,
-      "learning_rate": 0.00019999601469280855,
-      "loss": 46.0,
-      "step": 17657
-    },
-    {
-      "epoch": 2.8436732557671403,
-      "grad_norm": 0.002117637312039733,
-      "learning_rate": 0.0001999960142411291,
-      "loss": 46.0,
-      "step": 17658
-    },
-    {
-      "epoch": 2.8438342928459277,
-      "grad_norm": 0.005389770492911339,
-      "learning_rate": 0.00019999601378942406,
-      "loss": 46.0,
-      "step": 17659
-    },
-    {
-      "epoch": 2.843995329924715,
-      "grad_norm": 0.003491622395813465,
-      "learning_rate": 0.00019999601333769342,
-      "loss": 46.0,
-      "step": 17660
-    },
-    {
-      "epoch": 2.8441563670035026,
-      "grad_norm": 0.004554687067866325,
-      "learning_rate": 0.0001999960128859372,
-      "loss": 46.0,
-      "step": 17661
-    },
-    {
-      "epoch": 2.84431740408229,
-      "grad_norm": 0.004166826140135527,
-      "learning_rate": 0.00019999601243415535,
-      "loss": 46.0,
-      "step": 17662
-    },
-    {
-      "epoch": 2.844478441161077,
-      "grad_norm": 0.0027111927047371864,
-      "learning_rate": 0.00019999601198234795,
-      "loss": 46.0,
-      "step": 17663
-    },
-    {
-      "epoch": 2.8446394782398645,
-      "grad_norm": 0.005219809710979462,
-      "learning_rate": 0.00019999601153051493,
-      "loss": 46.0,
-      "step": 17664
-    },
-    {
-      "epoch": 2.844800515318652,
-      "grad_norm": 0.0071383328177034855,
-      "learning_rate": 0.00019999601107865633,
-      "loss": 46.0,
-      "step": 17665
-    },
-    {
-      "epoch": 2.8449615523974394,
-      "grad_norm": 0.002584136324003339,
-      "learning_rate": 0.00019999601062677211,
-      "loss": 46.0,
-      "step": 17666
-    },
-    {
-      "epoch": 2.845122589476227,
-      "grad_norm": 0.004258283879607916,
-      "learning_rate": 0.0001999960101748623,
-      "loss": 46.0,
-      "step": 17667
-    },
-    {
-      "epoch": 2.8452836265550143,
-      "grad_norm": 0.001980163622647524,
-      "learning_rate": 0.00019999600972292695,
-      "loss": 46.0,
-      "step": 17668
-    },
-    {
-      "epoch": 2.8454446636338018,
-      "grad_norm": 0.003112790873274207,
-      "learning_rate": 0.00019999600927096594,
-      "loss": 46.0,
-      "step": 17669
-    },
-    {
-      "epoch": 2.8456057007125892,
-      "grad_norm": 0.0022935864981263876,
-      "learning_rate": 0.00019999600881897937,
-      "loss": 46.0,
-      "step": 17670
-    },
-    {
-      "epoch": 2.8457667377913767,
-      "grad_norm": 0.004193443804979324,
-      "learning_rate": 0.0001999960083669672,
-      "loss": 46.0,
-      "step": 17671
-    },
-    {
-      "epoch": 2.845927774870164,
-      "grad_norm": 0.00351294269785285,
-      "learning_rate": 0.00019999600791492945,
-      "loss": 46.0,
-      "step": 17672
-    },
-    {
-      "epoch": 2.846088811948951,
-      "grad_norm": 0.008579579181969166,
-      "learning_rate": 0.00019999600746286607,
-      "loss": 46.0,
-      "step": 17673
-    },
-    {
-      "epoch": 2.8462498490277386,
-      "grad_norm": 0.003046198282390833,
-      "learning_rate": 0.00019999600701077713,
-      "loss": 46.0,
-      "step": 17674
-    },
-    {
-      "epoch": 2.846410886106526,
-      "grad_norm": 0.0005978193948976696,
-      "learning_rate": 0.00019999600655866258,
-      "loss": 46.0,
-      "step": 17675
-    },
-    {
-      "epoch": 2.8465719231853135,
-      "grad_norm": 0.001720809144899249,
-      "learning_rate": 0.00019999600610652243,
-      "loss": 46.0,
-      "step": 17676
-    },
-    {
-      "epoch": 2.846732960264101,
-      "grad_norm": 0.005646899808198214,
-      "learning_rate": 0.0001999960056543567,
-      "loss": 46.0,
-      "step": 17677
-    },
-    {
-      "epoch": 2.846893997342888,
-      "grad_norm": 0.014304318465292454,
-      "learning_rate": 0.0001999960052021654,
-      "loss": 46.0,
-      "step": 17678
-    },
-    {
-      "epoch": 2.8470550344216754,
-      "grad_norm": 0.0007857764721848071,
-      "learning_rate": 0.00019999600474994846,
-      "loss": 46.0,
-      "step": 17679
-    },
-    {
-      "epoch": 2.847216071500463,
-      "grad_norm": 0.0018906394252553582,
-      "learning_rate": 0.00019999600429770596,
-      "loss": 46.0,
-      "step": 17680
-    },
-    {
-      "epoch": 2.8473771085792503,
-      "grad_norm": 0.0013708010083064437,
-      "learning_rate": 0.00019999600384543783,
-      "loss": 46.0,
-      "step": 17681
-    },
-    {
-      "epoch": 2.8475381456580378,
-      "grad_norm": 0.001256629591807723,
-      "learning_rate": 0.00019999600339314414,
-      "loss": 46.0,
-      "step": 17682
-    },
-    {
-      "epoch": 2.847699182736825,
-      "grad_norm": 0.0033032239880412817,
-      "learning_rate": 0.00019999600294082483,
-      "loss": 46.0,
-      "step": 17683
-    },
-    {
-      "epoch": 2.8478602198156127,
-      "grad_norm": 0.0008098266553133726,
-      "learning_rate": 0.00019999600248847994,
-      "loss": 46.0,
-      "step": 17684
-    },
-    {
-      "epoch": 2.8480212568944,
-      "grad_norm": 0.0014712949050590396,
-      "learning_rate": 0.00019999600203610945,
-      "loss": 46.0,
-      "step": 17685
-    },
-    {
-      "epoch": 2.8481822939731876,
-      "grad_norm": 0.01069496851414442,
-      "learning_rate": 0.00019999600158371336,
-      "loss": 46.0,
-      "step": 17686
-    },
-    {
-      "epoch": 2.8483433310519746,
-      "grad_norm": 0.001078604138456285,
-      "learning_rate": 0.0001999960011312917,
-      "loss": 46.0,
-      "step": 17687
-    },
-    {
-      "epoch": 2.848504368130762,
-      "grad_norm": 0.0028587363194674253,
-      "learning_rate": 0.00019999600067884443,
-      "loss": 46.0,
-      "step": 17688
-    },
-    {
-      "epoch": 2.8486654052095495,
-      "grad_norm": 0.004137709736824036,
-      "learning_rate": 0.00019999600022637158,
-      "loss": 46.0,
-      "step": 17689
-    },
-    {
-      "epoch": 2.848826442288337,
-      "grad_norm": 0.0066100540570914745,
-      "learning_rate": 0.00019999599977387313,
-      "loss": 46.0,
-      "step": 17690
-    },
-    {
-      "epoch": 2.8489874793671244,
-      "grad_norm": 0.003406495088711381,
-      "learning_rate": 0.00019999599932134907,
-      "loss": 46.0,
-      "step": 17691
-    },
-    {
-      "epoch": 2.849148516445912,
-      "grad_norm": 0.0011324897641316056,
-      "learning_rate": 0.00019999599886879943,
-      "loss": 46.0,
-      "step": 17692
-    },
-    {
-      "epoch": 2.849309553524699,
-      "grad_norm": 0.0024638455361127853,
-      "learning_rate": 0.0001999959984162242,
-      "loss": 46.0,
-      "step": 17693
-    },
-    {
-      "epoch": 2.8494705906034863,
-      "grad_norm": 0.006360653322190046,
-      "learning_rate": 0.00019999599796362337,
-      "loss": 46.0,
-      "step": 17694
-    },
-    {
-      "epoch": 2.8496316276822737,
-      "grad_norm": 0.0026383153162896633,
-      "learning_rate": 0.00019999599751099694,
-      "loss": 46.0,
-      "step": 17695
-    },
-    {
-      "epoch": 2.849792664761061,
-      "grad_norm": 0.0013061851495876908,
-      "learning_rate": 0.00019999599705834494,
-      "loss": 46.0,
-      "step": 17696
-    },
-    {
-      "epoch": 2.8499537018398486,
-      "grad_norm": 0.0036800026427954435,
-      "learning_rate": 0.00019999599660566733,
-      "loss": 46.0,
-      "step": 17697
-    },
-    {
-      "epoch": 2.850114738918636,
-      "grad_norm": 0.0013708402402698994,
-      "learning_rate": 0.0001999959961529641,
-      "loss": 46.0,
-      "step": 17698
-    },
-    {
-      "epoch": 2.8502757759974235,
-      "grad_norm": 0.0006732487236149609,
-      "learning_rate": 0.0001999959957002353,
-      "loss": 46.0,
-      "step": 17699
-    },
-    {
-      "epoch": 2.850436813076211,
-      "grad_norm": 0.0023737475275993347,
-      "learning_rate": 0.00019999599524748093,
-      "loss": 46.0,
-      "step": 17700
-    },
-    {
-      "epoch": 2.8505978501549984,
-      "grad_norm": 0.008403660729527473,
-      "learning_rate": 0.00019999599479470095,
-      "loss": 46.0,
-      "step": 17701
-    },
-    {
-      "epoch": 2.8507588872337855,
-      "grad_norm": 0.0007045620586723089,
-      "learning_rate": 0.00019999599434189535,
-      "loss": 46.0,
-      "step": 17702
-    },
-    {
-      "epoch": 2.850919924312573,
-      "grad_norm": 0.0028128752019256353,
-      "learning_rate": 0.0001999959938890642,
-      "loss": 46.0,
-      "step": 17703
-    },
-    {
-      "epoch": 2.8510809613913604,
-      "grad_norm": 0.001996441511437297,
-      "learning_rate": 0.00019999599343620741,
-      "loss": 46.0,
-      "step": 17704
-    },
-    {
-      "epoch": 2.851241998470148,
-      "grad_norm": 0.010239379480481148,
-      "learning_rate": 0.00019999599298332505,
-      "loss": 46.0,
-      "step": 17705
-    },
-    {
-      "epoch": 2.8514030355489353,
-      "grad_norm": 0.0018866312457248569,
-      "learning_rate": 0.0001999959925304171,
-      "loss": 46.0,
-      "step": 17706
-    },
-    {
-      "epoch": 2.8515640726277223,
-      "grad_norm": 0.003044862300157547,
-      "learning_rate": 0.00019999599207748357,
-      "loss": 46.0,
-      "step": 17707
-    },
-    {
-      "epoch": 2.8517251097065097,
-      "grad_norm": 0.0015488950302824378,
-      "learning_rate": 0.0001999959916245244,
-      "loss": 46.0,
-      "step": 17708
-    },
-    {
-      "epoch": 2.851886146785297,
-      "grad_norm": 0.0015393532812595367,
-      "learning_rate": 0.00019999599117153966,
-      "loss": 46.0,
-      "step": 17709
-    },
-    {
-      "epoch": 2.8520471838640846,
-      "grad_norm": 0.0030128757935017347,
-      "learning_rate": 0.00019999599071852933,
-      "loss": 46.0,
-      "step": 17710
-    },
-    {
-      "epoch": 2.852208220942872,
-      "grad_norm": 0.004676430951803923,
-      "learning_rate": 0.0001999959902654934,
-      "loss": 46.0,
-      "step": 17711
-    },
-    {
-      "epoch": 2.8523692580216595,
-      "grad_norm": 0.0019623960833996534,
-      "learning_rate": 0.0001999959898124319,
-      "loss": 46.0,
-      "step": 17712
-    },
-    {
-      "epoch": 2.852530295100447,
-      "grad_norm": 0.013692514970898628,
-      "learning_rate": 0.0001999959893593448,
-      "loss": 46.0,
-      "step": 17713
-    },
-    {
-      "epoch": 2.8526913321792344,
-      "grad_norm": 0.0005461315158754587,
-      "learning_rate": 0.0001999959889062321,
-      "loss": 46.0,
-      "step": 17714
-    },
-    {
-      "epoch": 2.852852369258022,
-      "grad_norm": 0.0011717305751517415,
-      "learning_rate": 0.00019999598845309377,
-      "loss": 46.0,
-      "step": 17715
-    },
-    {
-      "epoch": 2.853013406336809,
-      "grad_norm": 0.008190694265067577,
-      "learning_rate": 0.0001999959879999299,
-      "loss": 46.0,
-      "step": 17716
-    },
-    {
-      "epoch": 2.8531744434155963,
-      "grad_norm": 0.00229228800162673,
-      "learning_rate": 0.0001999959875467404,
-      "loss": 46.0,
-      "step": 17717
-    },
-    {
-      "epoch": 2.853335480494384,
-      "grad_norm": 0.0023601411376148462,
-      "learning_rate": 0.0001999959870935253,
-      "loss": 46.0,
-      "step": 17718
-    },
-    {
-      "epoch": 2.8534965175731712,
-      "grad_norm": 0.0032715650741010904,
-      "learning_rate": 0.00019999598664028465,
-      "loss": 46.0,
-      "step": 17719
-    },
-    {
-      "epoch": 2.8536575546519587,
-      "grad_norm": 0.001448446186259389,
-      "learning_rate": 0.00019999598618701837,
-      "loss": 46.0,
-      "step": 17720
-    },
-    {
-      "epoch": 2.853818591730746,
-      "grad_norm": 0.0018138595623895526,
-      "learning_rate": 0.0001999959857337265,
-      "loss": 46.0,
-      "step": 17721
-    },
-    {
-      "epoch": 2.853979628809533,
-      "grad_norm": 0.0012017472181469202,
-      "learning_rate": 0.00019999598528040906,
-      "loss": 46.0,
-      "step": 17722
-    },
-    {
-      "epoch": 2.8541406658883206,
-      "grad_norm": 0.0039217215962708,
-      "learning_rate": 0.00019999598482706602,
-      "loss": 46.0,
-      "step": 17723
-    },
-    {
-      "epoch": 2.854301702967108,
-      "grad_norm": 0.0018078441498801112,
-      "learning_rate": 0.00019999598437369737,
-      "loss": 46.0,
-      "step": 17724
-    },
-    {
-      "epoch": 2.8544627400458955,
-      "grad_norm": 0.006035876926034689,
-      "learning_rate": 0.00019999598392030313,
-      "loss": 46.0,
-      "step": 17725
-    },
-    {
-      "epoch": 2.854623777124683,
-      "grad_norm": 0.0025969508569687605,
-      "learning_rate": 0.00019999598346688327,
-      "loss": 46.0,
-      "step": 17726
-    },
-    {
-      "epoch": 2.8547848142034704,
-      "grad_norm": 0.0037794699892401695,
-      "learning_rate": 0.00019999598301343786,
-      "loss": 46.0,
-      "step": 17727
-    },
-    {
-      "epoch": 2.854945851282258,
-      "grad_norm": 0.00448583485558629,
-      "learning_rate": 0.00019999598255996683,
-      "loss": 46.0,
-      "step": 17728
-    },
-    {
-      "epoch": 2.8551068883610453,
-      "grad_norm": 0.020436882972717285,
-      "learning_rate": 0.00019999598210647022,
-      "loss": 46.0,
-      "step": 17729
-    },
-    {
-      "epoch": 2.8552679254398328,
-      "grad_norm": 0.0007145946146920323,
-      "learning_rate": 0.00019999598165294801,
-      "loss": 46.0,
-      "step": 17730
-    },
-    {
-      "epoch": 2.8554289625186198,
-      "grad_norm": 0.004013553261756897,
-      "learning_rate": 0.00019999598119940022,
-      "loss": 46.0,
-      "step": 17731
-    },
-    {
-      "epoch": 2.8555899995974072,
-      "grad_norm": 0.0005296072922646999,
-      "learning_rate": 0.00019999598074582682,
-      "loss": 46.0,
-      "step": 17732
-    },
-    {
-      "epoch": 2.8557510366761947,
-      "grad_norm": 0.0019515667809173465,
-      "learning_rate": 0.00019999598029222783,
-      "loss": 46.0,
-      "step": 17733
-    },
-    {
-      "epoch": 2.855912073754982,
-      "grad_norm": 0.001788418390788138,
-      "learning_rate": 0.00019999597983860325,
-      "loss": 46.0,
-      "step": 17734
-    },
-    {
-      "epoch": 2.8560731108337696,
-      "grad_norm": 0.0023962599225342274,
-      "learning_rate": 0.00019999597938495306,
-      "loss": 46.0,
-      "step": 17735
-    },
-    {
-      "epoch": 2.856234147912557,
-      "grad_norm": 0.0008774466696195304,
-      "learning_rate": 0.0001999959789312773,
-      "loss": 46.0,
-      "step": 17736
-    },
-    {
-      "epoch": 2.856395184991344,
-      "grad_norm": 0.00515154330059886,
-      "learning_rate": 0.00019999597847757594,
-      "loss": 46.0,
-      "step": 17737
-    },
-    {
-      "epoch": 2.8565562220701315,
-      "grad_norm": 0.004023326560854912,
-      "learning_rate": 0.00019999597802384899,
-      "loss": 46.0,
-      "step": 17738
-    },
-    {
-      "epoch": 2.856717259148919,
-      "grad_norm": 0.008450892753899097,
-      "learning_rate": 0.00019999597757009642,
-      "loss": 46.0,
-      "step": 17739
-    },
-    {
-      "epoch": 2.8568782962277064,
-      "grad_norm": 0.00282103568315506,
-      "learning_rate": 0.0001999959771163183,
-      "loss": 46.0,
-      "step": 17740
-    },
-    {
-      "epoch": 2.857039333306494,
-      "grad_norm": 0.0014612689847126603,
-      "learning_rate": 0.00019999597666251452,
-      "loss": 46.0,
-      "step": 17741
-    },
-    {
-      "epoch": 2.8572003703852813,
-      "grad_norm": 0.009119594469666481,
-      "learning_rate": 0.0001999959762086852,
-      "loss": 46.0,
-      "step": 17742
-    },
-    {
-      "epoch": 2.8573614074640687,
-      "grad_norm": 0.007086241617798805,
-      "learning_rate": 0.0001999959757548303,
-      "loss": 46.0,
-      "step": 17743
-    },
-    {
-      "epoch": 2.857522444542856,
-      "grad_norm": 0.006481421645730734,
-      "learning_rate": 0.00019999597530094977,
-      "loss": 46.0,
-      "step": 17744
-    },
-    {
-      "epoch": 2.8576834816216437,
-      "grad_norm": 0.004874011967331171,
-      "learning_rate": 0.00019999597484704365,
-      "loss": 46.0,
-      "step": 17745
-    },
-    {
-      "epoch": 2.8578445187004307,
-      "grad_norm": 0.0023945835418999195,
-      "learning_rate": 0.00019999597439311192,
-      "loss": 46.0,
-      "step": 17746
-    },
-    {
-      "epoch": 2.858005555779218,
-      "grad_norm": 0.001118845073506236,
-      "learning_rate": 0.00019999597393915462,
-      "loss": 46.0,
-      "step": 17747
-    },
-    {
-      "epoch": 2.8581665928580056,
-      "grad_norm": 0.0018878093687817454,
-      "learning_rate": 0.00019999597348517174,
-      "loss": 46.0,
-      "step": 17748
-    },
-    {
-      "epoch": 2.858327629936793,
-      "grad_norm": 0.0008037827792577446,
-      "learning_rate": 0.00019999597303116325,
-      "loss": 46.0,
-      "step": 17749
-    },
-    {
-      "epoch": 2.8584886670155805,
-      "grad_norm": 0.0031560324132442474,
-      "learning_rate": 0.00019999597257712917,
-      "loss": 46.0,
-      "step": 17750
-    },
-    {
-      "epoch": 2.8586497040943675,
-      "grad_norm": 0.0038127838633954525,
-      "learning_rate": 0.00019999597212306947,
-      "loss": 46.0,
-      "step": 17751
-    },
-    {
-      "epoch": 2.858810741173155,
-      "grad_norm": 0.002726686652749777,
-      "learning_rate": 0.00019999597166898422,
-      "loss": 46.0,
-      "step": 17752
-    },
-    {
-      "epoch": 2.8589717782519424,
-      "grad_norm": 0.0007126725977286696,
-      "learning_rate": 0.00019999597121487335,
-      "loss": 46.0,
-      "step": 17753
-    },
-    {
-      "epoch": 2.85913281533073,
-      "grad_norm": 0.0019691763445734978,
-      "learning_rate": 0.0001999959707607369,
-      "loss": 46.0,
-      "step": 17754
-    },
-    {
-      "epoch": 2.8592938524095173,
-      "grad_norm": 0.0029765560757368803,
-      "learning_rate": 0.00019999597030657485,
-      "loss": 46.0,
-      "step": 17755
-    },
-    {
-      "epoch": 2.8594548894883047,
-      "grad_norm": 0.005691281985491514,
-      "learning_rate": 0.0001999959698523872,
-      "loss": 46.0,
-      "step": 17756
-    },
-    {
-      "epoch": 2.859615926567092,
-      "grad_norm": 0.00326905376277864,
-      "learning_rate": 0.00019999596939817394,
-      "loss": 46.0,
-      "step": 17757
-    },
-    {
-      "epoch": 2.8597769636458796,
-      "grad_norm": 0.0003961396578233689,
-      "learning_rate": 0.0001999959689439351,
-      "loss": 46.0,
-      "step": 17758
-    },
-    {
-      "epoch": 2.859938000724667,
-      "grad_norm": 0.001628469442948699,
-      "learning_rate": 0.0001999959684896707,
-      "loss": 46.0,
-      "step": 17759
-    },
-    {
-      "epoch": 2.860099037803454,
-      "grad_norm": 0.0012904424220323563,
-      "learning_rate": 0.00019999596803538065,
-      "loss": 46.0,
-      "step": 17760
-    },
-    {
-      "epoch": 2.8602600748822415,
-      "grad_norm": 0.0023224635515362024,
-      "learning_rate": 0.00019999596758106503,
-      "loss": 46.0,
-      "step": 17761
-    },
-    {
-      "epoch": 2.860421111961029,
-      "grad_norm": 0.0023614896927028894,
-      "learning_rate": 0.00019999596712672385,
-      "loss": 46.0,
-      "step": 17762
-    },
-    {
-      "epoch": 2.8605821490398164,
-      "grad_norm": 0.0032101981341838837,
-      "learning_rate": 0.00019999596667235703,
-      "loss": 46.0,
-      "step": 17763
-    },
-    {
-      "epoch": 2.860743186118604,
-      "grad_norm": 0.0038493045140057802,
-      "learning_rate": 0.00019999596621796464,
-      "loss": 46.0,
-      "step": 17764
-    },
-    {
-      "epoch": 2.8609042231973913,
-      "grad_norm": 0.002698685275390744,
-      "learning_rate": 0.00019999596576354662,
-      "loss": 46.0,
-      "step": 17765
-    },
-    {
-      "epoch": 2.8610652602761784,
-      "grad_norm": 0.0008765386301092803,
-      "learning_rate": 0.00019999596530910306,
-      "loss": 46.0,
-      "step": 17766
-    },
-    {
-      "epoch": 2.861226297354966,
-      "grad_norm": 0.0070189377292990685,
-      "learning_rate": 0.00019999596485463386,
-      "loss": 46.0,
-      "step": 17767
-    },
-    {
-      "epoch": 2.8613873344337533,
-      "grad_norm": 0.0061255996115505695,
-      "learning_rate": 0.0001999959644001391,
-      "loss": 46.0,
-      "step": 17768
-    },
-    {
-      "epoch": 2.8615483715125407,
-      "grad_norm": 0.0007536065531894565,
-      "learning_rate": 0.00019999596394561873,
-      "loss": 46.0,
-      "step": 17769
-    },
-    {
-      "epoch": 2.861709408591328,
-      "grad_norm": 0.002301283646374941,
-      "learning_rate": 0.00019999596349107277,
-      "loss": 46.0,
-      "step": 17770
-    },
-    {
-      "epoch": 2.8618704456701156,
-      "grad_norm": 0.002608307171612978,
-      "learning_rate": 0.0001999959630365012,
-      "loss": 46.0,
-      "step": 17771
-    },
-    {
-      "epoch": 2.862031482748903,
-      "grad_norm": 0.0010824125492945313,
-      "learning_rate": 0.00019999596258190406,
-      "loss": 46.0,
-      "step": 17772
-    },
-    {
-      "epoch": 2.8621925198276905,
-      "grad_norm": 0.0028726982418447733,
-      "learning_rate": 0.00019999596212728134,
-      "loss": 46.0,
-      "step": 17773
-    },
-    {
-      "epoch": 2.862353556906478,
-      "grad_norm": 0.0016900365008041263,
-      "learning_rate": 0.00019999596167263298,
-      "loss": 46.0,
-      "step": 17774
-    },
-    {
-      "epoch": 2.862514593985265,
-      "grad_norm": 0.004488042555749416,
-      "learning_rate": 0.00019999596121795905,
-      "loss": 46.0,
-      "step": 17775
-    },
-    {
-      "epoch": 2.8626756310640524,
-      "grad_norm": 0.006179673131555319,
-      "learning_rate": 0.00019999596076325952,
-      "loss": 46.0,
-      "step": 17776
-    },
-    {
-      "epoch": 2.86283666814284,
-      "grad_norm": 0.00732179032638669,
-      "learning_rate": 0.0001999959603085344,
-      "loss": 46.0,
-      "step": 17777
-    },
-    {
-      "epoch": 2.8629977052216273,
-      "grad_norm": 0.0023710872046649456,
-      "learning_rate": 0.0001999959598537837,
-      "loss": 46.0,
-      "step": 17778
-    },
-    {
-      "epoch": 2.863158742300415,
-      "grad_norm": 0.004059616010636091,
-      "learning_rate": 0.00019999595939900738,
-      "loss": 46.0,
-      "step": 17779
-    },
-    {
-      "epoch": 2.863319779379202,
-      "grad_norm": 0.0026830388233065605,
-      "learning_rate": 0.0001999959589442055,
-      "loss": 46.0,
-      "step": 17780
-    },
-    {
-      "epoch": 2.8634808164579892,
-      "grad_norm": 0.011139930225908756,
-      "learning_rate": 0.000199995958489378,
-      "loss": 46.0,
-      "step": 17781
-    },
-    {
-      "epoch": 2.8636418535367767,
-      "grad_norm": 0.00361109827645123,
-      "learning_rate": 0.0001999959580345249,
-      "loss": 46.0,
-      "step": 17782
-    },
-    {
-      "epoch": 2.863802890615564,
-      "grad_norm": 0.0017445327248424292,
-      "learning_rate": 0.00019999595757964623,
-      "loss": 46.0,
-      "step": 17783
-    },
-    {
-      "epoch": 2.8639639276943516,
-      "grad_norm": 0.0037039124872535467,
-      "learning_rate": 0.00019999595712474197,
-      "loss": 46.0,
-      "step": 17784
-    },
-    {
-      "epoch": 2.864124964773139,
-      "grad_norm": 0.0007199071696959436,
-      "learning_rate": 0.0001999959566698121,
-      "loss": 46.0,
-      "step": 17785
-    },
-    {
-      "epoch": 2.8642860018519265,
-      "grad_norm": 0.0014654486440122128,
-      "learning_rate": 0.00019999595621485663,
-      "loss": 46.0,
-      "step": 17786
-    },
-    {
-      "epoch": 2.864447038930714,
-      "grad_norm": 0.0020696097053587437,
-      "learning_rate": 0.00019999595575987558,
-      "loss": 46.0,
-      "step": 17787
-    },
-    {
-      "epoch": 2.8646080760095014,
-      "grad_norm": 0.0011353419395163655,
-      "learning_rate": 0.00019999595530486891,
-      "loss": 46.0,
-      "step": 17788
-    },
-    {
-      "epoch": 2.864769113088289,
-      "grad_norm": 0.0013259720290079713,
-      "learning_rate": 0.0001999959548498367,
-      "loss": 46.0,
-      "step": 17789
-    },
-    {
-      "epoch": 2.864930150167076,
-      "grad_norm": 0.0018845100421458483,
-      "learning_rate": 0.00019999595439477882,
-      "loss": 46.0,
-      "step": 17790
-    },
-    {
-      "epoch": 2.8650911872458633,
-      "grad_norm": 0.0032751886174082756,
-      "learning_rate": 0.0001999959539396954,
-      "loss": 46.0,
-      "step": 17791
-    },
-    {
-      "epoch": 2.8652522243246508,
-      "grad_norm": 0.002355837030336261,
-      "learning_rate": 0.00019999595348458638,
-      "loss": 46.0,
-      "step": 17792
-    },
-    {
-      "epoch": 2.865413261403438,
-      "grad_norm": 0.0010306062176823616,
-      "learning_rate": 0.00019999595302945178,
-      "loss": 46.0,
-      "step": 17793
-    },
-    {
-      "epoch": 2.8655742984822257,
-      "grad_norm": 0.005004832521080971,
-      "learning_rate": 0.00019999595257429157,
-      "loss": 46.0,
-      "step": 17794
-    },
-    {
-      "epoch": 2.8657353355610127,
-      "grad_norm": 0.004597513936460018,
-      "learning_rate": 0.00019999595211910576,
-      "loss": 46.0,
-      "step": 17795
-    },
-    {
-      "epoch": 2.8658963726398,
-      "grad_norm": 0.001008680323138833,
-      "learning_rate": 0.00019999595166389432,
-      "loss": 46.0,
-      "step": 17796
-    },
-    {
-      "epoch": 2.8660574097185876,
-      "grad_norm": 0.0038172886706888676,
-      "learning_rate": 0.00019999595120865734,
-      "loss": 46.0,
-      "step": 17797
-    },
-    {
-      "epoch": 2.866218446797375,
-      "grad_norm": 0.005964105483144522,
-      "learning_rate": 0.00019999595075339475,
-      "loss": 46.0,
-      "step": 17798
-    },
-    {
-      "epoch": 2.8663794838761625,
-      "grad_norm": 0.0015238994965329766,
-      "learning_rate": 0.00019999595029810658,
-      "loss": 46.0,
-      "step": 17799
-    },
-    {
-      "epoch": 2.86654052095495,
-      "grad_norm": 0.004664433654397726,
-      "learning_rate": 0.00019999594984279278,
-      "loss": 46.0,
-      "step": 17800
-    },
-    {
-      "epoch": 2.8667015580337374,
-      "grad_norm": 0.003416199702769518,
-      "learning_rate": 0.00019999594938745343,
-      "loss": 46.0,
-      "step": 17801
-    },
-    {
-      "epoch": 2.866862595112525,
-      "grad_norm": 0.00044466060353443027,
-      "learning_rate": 0.00019999594893208846,
-      "loss": 46.0,
-      "step": 17802
-    },
-    {
-      "epoch": 2.8670236321913123,
-      "grad_norm": 0.003705657320097089,
-      "learning_rate": 0.0001999959484766979,
-      "loss": 46.0,
-      "step": 17803
-    },
-    {
-      "epoch": 2.8671846692700993,
-      "grad_norm": 0.004809023346751928,
-      "learning_rate": 0.00019999594802128177,
-      "loss": 46.0,
-      "step": 17804
-    },
-    {
-      "epoch": 2.8673457063488867,
-      "grad_norm": 0.0016219865065068007,
-      "learning_rate": 0.00019999594756584001,
-      "loss": 46.0,
-      "step": 17805
-    },
-    {
-      "epoch": 2.867506743427674,
-      "grad_norm": 0.009811586700379848,
-      "learning_rate": 0.00019999594711037267,
-      "loss": 46.0,
-      "step": 17806
-    },
-    {
-      "epoch": 2.8676677805064616,
-      "grad_norm": 0.0015374321956187487,
-      "learning_rate": 0.00019999594665487974,
-      "loss": 46.0,
-      "step": 17807
-    },
-    {
-      "epoch": 2.867828817585249,
-      "grad_norm": 0.0043531060218811035,
-      "learning_rate": 0.00019999594619936122,
-      "loss": 46.0,
-      "step": 17808
-    },
-    {
-      "epoch": 2.8679898546640366,
-      "grad_norm": 0.001166231231763959,
-      "learning_rate": 0.0001999959457438171,
-      "loss": 46.0,
-      "step": 17809
-    },
-    {
-      "epoch": 2.8681508917428236,
-      "grad_norm": 0.0009849559282884002,
-      "learning_rate": 0.00019999594528824738,
-      "loss": 46.0,
-      "step": 17810
-    },
-    {
-      "epoch": 2.868311928821611,
-      "grad_norm": 0.010321375913918018,
-      "learning_rate": 0.00019999594483265207,
-      "loss": 46.0,
-      "step": 17811
-    },
-    {
-      "epoch": 2.8684729659003985,
-      "grad_norm": 0.001669671619310975,
-      "learning_rate": 0.00019999594437703118,
-      "loss": 46.0,
-      "step": 17812
-    },
-    {
-      "epoch": 2.868634002979186,
-      "grad_norm": 0.004847492557018995,
-      "learning_rate": 0.00019999594392138467,
-      "loss": 46.0,
-      "step": 17813
-    },
-    {
-      "epoch": 2.8687950400579734,
-      "grad_norm": 0.004640201572328806,
-      "learning_rate": 0.00019999594346571258,
-      "loss": 46.0,
-      "step": 17814
-    },
-    {
-      "epoch": 2.868956077136761,
-      "grad_norm": 0.005793978925794363,
-      "learning_rate": 0.0001999959430100149,
-      "loss": 46.0,
-      "step": 17815
-    },
-    {
-      "epoch": 2.8691171142155483,
-      "grad_norm": 0.003426800947636366,
-      "learning_rate": 0.00019999594255429163,
-      "loss": 46.0,
-      "step": 17816
-    },
-    {
-      "epoch": 2.8692781512943357,
-      "grad_norm": 0.0007502498920075595,
-      "learning_rate": 0.00019999594209854274,
-      "loss": 46.0,
-      "step": 17817
-    },
-    {
-      "epoch": 2.869439188373123,
-      "grad_norm": 0.0014112351927906275,
-      "learning_rate": 0.00019999594164276827,
-      "loss": 46.0,
-      "step": 17818
-    },
-    {
-      "epoch": 2.86960022545191,
-      "grad_norm": 0.0016646521398797631,
-      "learning_rate": 0.00019999594118696822,
-      "loss": 46.0,
-      "step": 17819
-    },
-    {
-      "epoch": 2.8697612625306976,
-      "grad_norm": 0.0032315412536263466,
-      "learning_rate": 0.00019999594073114257,
-      "loss": 46.0,
-      "step": 17820
-    },
-    {
-      "epoch": 2.869922299609485,
-      "grad_norm": 0.0027037032414227724,
-      "learning_rate": 0.00019999594027529134,
-      "loss": 46.0,
-      "step": 17821
-    },
-    {
-      "epoch": 2.8700833366882725,
-      "grad_norm": 0.007093028165400028,
-      "learning_rate": 0.0001999959398194145,
-      "loss": 46.0,
-      "step": 17822
-    },
-    {
-      "epoch": 2.87024437376706,
-      "grad_norm": 0.004408692009747028,
-      "learning_rate": 0.00019999593936351206,
-      "loss": 46.0,
-      "step": 17823
-    },
-    {
-      "epoch": 2.870405410845847,
-      "grad_norm": 0.0010976019548252225,
-      "learning_rate": 0.00019999593890758402,
-      "loss": 46.0,
-      "step": 17824
-    },
-    {
-      "epoch": 2.8705664479246344,
-      "grad_norm": 0.006968142930418253,
-      "learning_rate": 0.0001999959384516304,
-      "loss": 46.0,
-      "step": 17825
-    },
-    {
-      "epoch": 2.870727485003422,
-      "grad_norm": 0.0027809094171971083,
-      "learning_rate": 0.00019999593799565116,
-      "loss": 46.0,
-      "step": 17826
-    },
-    {
-      "epoch": 2.8708885220822093,
-      "grad_norm": 0.0034778385888785124,
-      "learning_rate": 0.00019999593753964638,
-      "loss": 46.0,
-      "step": 17827
-    },
-    {
-      "epoch": 2.871049559160997,
-      "grad_norm": 0.0013363156467676163,
-      "learning_rate": 0.00019999593708361596,
-      "loss": 46.0,
-      "step": 17828
-    },
-    {
-      "epoch": 2.8712105962397843,
-      "grad_norm": 0.002102689351886511,
-      "learning_rate": 0.00019999593662755997,
-      "loss": 46.0,
-      "step": 17829
-    },
-    {
-      "epoch": 2.8713716333185717,
-      "grad_norm": 0.009751196019351482,
-      "learning_rate": 0.00019999593617147838,
-      "loss": 46.0,
-      "step": 17830
-    },
-    {
-      "epoch": 2.871532670397359,
-      "grad_norm": 0.0028450442478060722,
-      "learning_rate": 0.00019999593571537122,
-      "loss": 46.0,
-      "step": 17831
-    },
-    {
-      "epoch": 2.8716937074761466,
-      "grad_norm": 0.004424213897436857,
-      "learning_rate": 0.00019999593525923842,
-      "loss": 46.0,
-      "step": 17832
-    },
-    {
-      "epoch": 2.8718547445549336,
-      "grad_norm": 0.0027277308981865644,
-      "learning_rate": 0.00019999593480308004,
-      "loss": 46.0,
-      "step": 17833
-    },
-    {
-      "epoch": 2.872015781633721,
-      "grad_norm": 0.0025158394128084183,
-      "learning_rate": 0.0001999959343468961,
-      "loss": 46.0,
-      "step": 17834
-    },
-    {
-      "epoch": 2.8721768187125085,
-      "grad_norm": 0.004669091664254665,
-      "learning_rate": 0.0001999959338906865,
-      "loss": 46.0,
-      "step": 17835
-    },
-    {
-      "epoch": 2.872337855791296,
-      "grad_norm": 0.003903195494785905,
-      "learning_rate": 0.00019999593343445135,
-      "loss": 46.0,
-      "step": 17836
-    },
-    {
-      "epoch": 2.8724988928700834,
-      "grad_norm": 0.00392779428511858,
-      "learning_rate": 0.00019999593297819062,
-      "loss": 46.0,
-      "step": 17837
-    },
-    {
-      "epoch": 2.872659929948871,
-      "grad_norm": 0.0007379152229987085,
-      "learning_rate": 0.00019999593252190427,
-      "loss": 46.0,
-      "step": 17838
-    },
-    {
-      "epoch": 2.872820967027658,
-      "grad_norm": 0.0028645016718655825,
-      "learning_rate": 0.00019999593206559234,
-      "loss": 46.0,
-      "step": 17839
-    },
-    {
-      "epoch": 2.8729820041064453,
-      "grad_norm": 0.0012807894963771105,
-      "learning_rate": 0.00019999593160925484,
-      "loss": 46.0,
-      "step": 17840
-    },
-    {
-      "epoch": 2.873143041185233,
-      "grad_norm": 0.01279291883111,
-      "learning_rate": 0.0001999959311528917,
-      "loss": 46.0,
-      "step": 17841
-    },
-    {
-      "epoch": 2.8733040782640202,
-      "grad_norm": 0.0027771827299147844,
-      "learning_rate": 0.00019999593069650298,
-      "loss": 46.0,
-      "step": 17842
-    },
-    {
-      "epoch": 2.8734651153428077,
-      "grad_norm": 0.001404869370162487,
-      "learning_rate": 0.00019999593024008867,
-      "loss": 46.0,
-      "step": 17843
-    },
-    {
-      "epoch": 2.873626152421595,
-      "grad_norm": 0.0039648148231208324,
-      "learning_rate": 0.00019999592978364877,
-      "loss": 46.0,
-      "step": 17844
-    },
-    {
-      "epoch": 2.8737871895003826,
-      "grad_norm": 0.004427431151270866,
-      "learning_rate": 0.00019999592932718328,
-      "loss": 46.0,
-      "step": 17845
-    },
-    {
-      "epoch": 2.87394822657917,
-      "grad_norm": 0.0030001287814229727,
-      "learning_rate": 0.00019999592887069215,
-      "loss": 46.0,
-      "step": 17846
-    },
-    {
-      "epoch": 2.8741092636579575,
-      "grad_norm": 0.0028888951055705547,
-      "learning_rate": 0.00019999592841417547,
-      "loss": 46.0,
-      "step": 17847
-    },
-    {
-      "epoch": 2.8742703007367445,
-      "grad_norm": 0.007748141419142485,
-      "learning_rate": 0.0001999959279576332,
-      "loss": 46.0,
-      "step": 17848
-    },
-    {
-      "epoch": 2.874431337815532,
-      "grad_norm": 0.003926098812371492,
-      "learning_rate": 0.00019999592750106533,
-      "loss": 46.0,
-      "step": 17849
-    },
-    {
-      "epoch": 2.8745923748943194,
-      "grad_norm": 0.010174036026000977,
-      "learning_rate": 0.00019999592704447185,
-      "loss": 46.0,
-      "step": 17850
-    },
-    {
-      "epoch": 2.874753411973107,
-      "grad_norm": 0.007152326870709658,
-      "learning_rate": 0.0001999959265878528,
-      "loss": 46.0,
-      "step": 17851
-    },
-    {
-      "epoch": 2.8749144490518943,
-      "grad_norm": 0.0013236486120149493,
-      "learning_rate": 0.00019999592613120814,
-      "loss": 46.0,
-      "step": 17852
-    },
-    {
-      "epoch": 2.8750754861306813,
-      "grad_norm": 0.00904405303299427,
-      "learning_rate": 0.00019999592567453787,
-      "loss": 46.0,
-      "step": 17853
-    },
-    {
-      "epoch": 2.8752365232094688,
-      "grad_norm": 0.0025345038156956434,
-      "learning_rate": 0.00019999592521784205,
-      "loss": 46.0,
-      "step": 17854
-    },
-    {
-      "epoch": 2.875397560288256,
-      "grad_norm": 0.00425531854853034,
-      "learning_rate": 0.0001999959247611206,
-      "loss": 46.0,
-      "step": 17855
-    },
-    {
-      "epoch": 2.8755585973670437,
-      "grad_norm": 0.0020722730550915003,
-      "learning_rate": 0.00019999592430437358,
-      "loss": 46.0,
-      "step": 17856
-    },
-    {
-      "epoch": 2.875719634445831,
-      "grad_norm": 0.005723000969737768,
-      "learning_rate": 0.00019999592384760097,
-      "loss": 46.0,
-      "step": 17857
-    },
-    {
-      "epoch": 2.8758806715246186,
-      "grad_norm": 0.0029709741938859224,
-      "learning_rate": 0.00019999592339080274,
-      "loss": 46.0,
-      "step": 17858
-    },
-    {
-      "epoch": 2.876041708603406,
-      "grad_norm": 0.003711389610543847,
-      "learning_rate": 0.00019999592293397893,
-      "loss": 46.0,
-      "step": 17859
-    },
-    {
-      "epoch": 2.8762027456821935,
-      "grad_norm": 0.003403382608667016,
-      "learning_rate": 0.00019999592247712953,
-      "loss": 46.0,
-      "step": 17860
-    },
-    {
-      "epoch": 2.876363782760981,
-      "grad_norm": 0.004919823259115219,
-      "learning_rate": 0.0001999959220202545,
-      "loss": 46.0,
-      "step": 17861
-    },
-    {
-      "epoch": 2.8765248198397684,
-      "grad_norm": 0.010195001028478146,
-      "learning_rate": 0.0001999959215633539,
-      "loss": 46.0,
-      "step": 17862
-    },
-    {
-      "epoch": 2.8766858569185554,
-      "grad_norm": 0.028367798775434494,
-      "learning_rate": 0.00019999592110642774,
-      "loss": 46.0,
-      "step": 17863
-    },
-    {
-      "epoch": 2.876846893997343,
-      "grad_norm": 0.0027255869936197996,
-      "learning_rate": 0.00019999592064947597,
-      "loss": 46.0,
-      "step": 17864
-    },
-    {
-      "epoch": 2.8770079310761303,
-      "grad_norm": 0.0018845719750970602,
-      "learning_rate": 0.00019999592019249857,
-      "loss": 46.0,
-      "step": 17865
-    },
-    {
-      "epoch": 2.8771689681549177,
-      "grad_norm": 0.00484988559037447,
-      "learning_rate": 0.0001999959197354956,
-      "loss": 46.0,
-      "step": 17866
-    },
-    {
-      "epoch": 2.877330005233705,
-      "grad_norm": 0.004739455413073301,
-      "learning_rate": 0.00019999591927846705,
-      "loss": 46.0,
-      "step": 17867
-    },
-    {
-      "epoch": 2.877491042312492,
-      "grad_norm": 0.003980754408985376,
-      "learning_rate": 0.00019999591882141287,
-      "loss": 46.0,
-      "step": 17868
-    },
-    {
-      "epoch": 2.8776520793912796,
-      "grad_norm": 0.003942277282476425,
-      "learning_rate": 0.00019999591836433313,
-      "loss": 46.0,
-      "step": 17869
-    },
-    {
-      "epoch": 2.877813116470067,
-      "grad_norm": 0.0025070379488170147,
-      "learning_rate": 0.0001999959179072278,
-      "loss": 46.0,
-      "step": 17870
-    },
-    {
-      "epoch": 2.8779741535488546,
-      "grad_norm": 0.0045952266082167625,
-      "learning_rate": 0.00019999591745009684,
-      "loss": 46.0,
-      "step": 17871
-    },
-    {
-      "epoch": 2.878135190627642,
-      "grad_norm": 0.0027326405979692936,
-      "learning_rate": 0.00019999591699294033,
-      "loss": 46.0,
-      "step": 17872
-    },
-    {
-      "epoch": 2.8782962277064295,
-      "grad_norm": 0.0009066511411219835,
-      "learning_rate": 0.0001999959165357582,
-      "loss": 46.0,
-      "step": 17873
-    },
-    {
-      "epoch": 2.878457264785217,
-      "grad_norm": 0.01229983102530241,
-      "learning_rate": 0.00019999591607855049,
-      "loss": 46.0,
-      "step": 17874
-    },
-    {
-      "epoch": 2.8786183018640044,
-      "grad_norm": 0.0006595427985303104,
-      "learning_rate": 0.0001999959156213172,
-      "loss": 46.0,
-      "step": 17875
-    },
-    {
-      "epoch": 2.878779338942792,
-      "grad_norm": 0.006917824503034353,
-      "learning_rate": 0.00019999591516405826,
-      "loss": 46.0,
-      "step": 17876
-    },
-    {
-      "epoch": 2.878940376021579,
-      "grad_norm": 0.0011825596448034048,
-      "learning_rate": 0.00019999591470677377,
-      "loss": 46.0,
-      "step": 17877
-    },
-    {
-      "epoch": 2.8791014131003663,
-      "grad_norm": 0.0011919609969481826,
-      "learning_rate": 0.00019999591424946366,
-      "loss": 46.0,
-      "step": 17878
-    },
-    {
-      "epoch": 2.8792624501791537,
-      "grad_norm": 0.003728542709723115,
-      "learning_rate": 0.000199995913792128,
-      "loss": 46.0,
-      "step": 17879
-    },
-    {
-      "epoch": 2.879423487257941,
-      "grad_norm": 0.005177161656320095,
-      "learning_rate": 0.00019999591333476672,
-      "loss": 46.0,
-      "step": 17880
-    },
-    {
-      "epoch": 2.8795845243367286,
-      "grad_norm": 0.00561678409576416,
-      "learning_rate": 0.00019999591287737982,
-      "loss": 46.0,
-      "step": 17881
-    },
-    {
-      "epoch": 2.879745561415516,
-      "grad_norm": 0.005106620490550995,
-      "learning_rate": 0.00019999591241996736,
-      "loss": 46.0,
-      "step": 17882
-    },
-    {
-      "epoch": 2.879906598494303,
-      "grad_norm": 0.015724681317806244,
-      "learning_rate": 0.0001999959119625293,
-      "loss": 46.0,
-      "step": 17883
-    },
-    {
-      "epoch": 2.8800676355730905,
-      "grad_norm": 0.00477852625772357,
-      "learning_rate": 0.00019999591150506564,
-      "loss": 46.0,
-      "step": 17884
-    },
-    {
-      "epoch": 2.880228672651878,
-      "grad_norm": 0.0018896848196163774,
-      "learning_rate": 0.00019999591104757637,
-      "loss": 46.0,
-      "step": 17885
-    },
-    {
-      "epoch": 2.8803897097306654,
-      "grad_norm": 0.002053465461358428,
-      "learning_rate": 0.00019999591059006156,
-      "loss": 46.0,
-      "step": 17886
-    },
-    {
-      "epoch": 2.880550746809453,
-      "grad_norm": 0.0014558566035702825,
-      "learning_rate": 0.0001999959101325211,
-      "loss": 46.0,
-      "step": 17887
-    },
-    {
-      "epoch": 2.8807117838882403,
-      "grad_norm": 0.008170108310878277,
-      "learning_rate": 0.00019999590967495509,
-      "loss": 46.0,
-      "step": 17888
-    },
-    {
-      "epoch": 2.880872820967028,
-      "grad_norm": 0.003906745929270983,
-      "learning_rate": 0.00019999590921736344,
-      "loss": 46.0,
-      "step": 17889
-    },
-    {
-      "epoch": 2.8810338580458152,
-      "grad_norm": 0.003990508615970612,
-      "learning_rate": 0.00019999590875974623,
-      "loss": 46.0,
-      "step": 17890
-    },
-    {
-      "epoch": 2.8811948951246027,
-      "grad_norm": 0.001963198184967041,
-      "learning_rate": 0.0001999959083021034,
-      "loss": 46.0,
-      "step": 17891
-    },
-    {
-      "epoch": 2.8813559322033897,
-      "grad_norm": 0.003254386829212308,
-      "learning_rate": 0.00019999590784443497,
-      "loss": 46.0,
-      "step": 17892
-    },
-    {
-      "epoch": 2.881516969282177,
-      "grad_norm": 0.0008436123025603592,
-      "learning_rate": 0.00019999590738674098,
-      "loss": 46.0,
-      "step": 17893
-    },
-    {
-      "epoch": 2.8816780063609646,
-      "grad_norm": 0.0027702623046934605,
-      "learning_rate": 0.0001999959069290214,
-      "loss": 46.0,
-      "step": 17894
-    },
-    {
-      "epoch": 2.881839043439752,
-      "grad_norm": 0.002942032413557172,
-      "learning_rate": 0.0001999959064712762,
-      "loss": 46.0,
-      "step": 17895
-    },
-    {
-      "epoch": 2.8820000805185395,
-      "grad_norm": 0.0005536332610063255,
-      "learning_rate": 0.00019999590601350542,
-      "loss": 46.0,
-      "step": 17896
-    },
-    {
-      "epoch": 2.8821611175973265,
-      "grad_norm": 0.0010056201135739684,
-      "learning_rate": 0.00019999590555570907,
-      "loss": 46.0,
-      "step": 17897
-    },
-    {
-      "epoch": 2.882322154676114,
-      "grad_norm": 0.007605133578181267,
-      "learning_rate": 0.00019999590509788706,
-      "loss": 46.0,
-      "step": 17898
-    },
-    {
-      "epoch": 2.8824831917549014,
-      "grad_norm": 0.010303414426743984,
-      "learning_rate": 0.0001999959046400395,
-      "loss": 46.0,
-      "step": 17899
-    },
-    {
-      "epoch": 2.882644228833689,
-      "grad_norm": 0.00073988240910694,
-      "learning_rate": 0.00019999590418216635,
-      "loss": 46.0,
-      "step": 17900
-    },
-    {
-      "epoch": 2.8828052659124763,
-      "grad_norm": 0.0015515421982854605,
-      "learning_rate": 0.0001999959037242676,
-      "loss": 46.0,
-      "step": 17901
-    },
-    {
-      "epoch": 2.8829663029912638,
-      "grad_norm": 0.0033378845546394587,
-      "learning_rate": 0.00019999590326634324,
-      "loss": 46.0,
-      "step": 17902
-    },
-    {
-      "epoch": 2.8831273400700512,
-      "grad_norm": 0.002852408913895488,
-      "learning_rate": 0.00019999590280839332,
-      "loss": 46.0,
-      "step": 17903
-    },
-    {
-      "epoch": 2.8832883771488387,
-      "grad_norm": 0.0057228063233196735,
-      "learning_rate": 0.0001999959023504178,
-      "loss": 46.0,
-      "step": 17904
-    },
-    {
-      "epoch": 2.883449414227626,
-      "grad_norm": 0.0023409484419971704,
-      "learning_rate": 0.00019999590189241664,
-      "loss": 46.0,
-      "step": 17905
-    },
-    {
-      "epoch": 2.883610451306413,
-      "grad_norm": 0.00564181013032794,
-      "learning_rate": 0.00019999590143438993,
-      "loss": 46.0,
-      "step": 17906
-    },
-    {
-      "epoch": 2.8837714883852006,
-      "grad_norm": 0.009571834467351437,
-      "learning_rate": 0.0001999959009763376,
-      "loss": 46.0,
-      "step": 17907
-    },
-    {
-      "epoch": 2.883932525463988,
-      "grad_norm": 0.002726753009483218,
-      "learning_rate": 0.00019999590051825972,
-      "loss": 46.0,
-      "step": 17908
-    },
-    {
-      "epoch": 2.8840935625427755,
-      "grad_norm": 0.003085640724748373,
-      "learning_rate": 0.0001999959000601562,
-      "loss": 46.0,
-      "step": 17909
-    },
-    {
-      "epoch": 2.884254599621563,
-      "grad_norm": 0.007829601876437664,
-      "learning_rate": 0.0001999958996020271,
-      "loss": 46.0,
-      "step": 17910
-    },
-    {
-      "epoch": 2.8844156367003504,
-      "grad_norm": 0.0019419188611209393,
-      "learning_rate": 0.0001999958991438724,
-      "loss": 46.0,
-      "step": 17911
-    },
-    {
-      "epoch": 2.8845766737791374,
-      "grad_norm": 0.015280799008905888,
-      "learning_rate": 0.00019999589868569212,
-      "loss": 46.0,
-      "step": 17912
-    },
-    {
-      "epoch": 2.884737710857925,
-      "grad_norm": 0.009347494691610336,
-      "learning_rate": 0.00019999589822748624,
-      "loss": 46.0,
-      "step": 17913
-    },
-    {
-      "epoch": 2.8848987479367123,
-      "grad_norm": 0.004690299276262522,
-      "learning_rate": 0.00019999589776925478,
-      "loss": 46.0,
-      "step": 17914
-    },
-    {
-      "epoch": 2.8850597850154998,
-      "grad_norm": 0.00586359528824687,
-      "learning_rate": 0.0001999958973109977,
-      "loss": 46.0,
-      "step": 17915
-    },
-    {
-      "epoch": 2.885220822094287,
-      "grad_norm": 0.004095832817256451,
-      "learning_rate": 0.00019999589685271504,
-      "loss": 46.0,
-      "step": 17916
-    },
-    {
-      "epoch": 2.8853818591730747,
-      "grad_norm": 0.018502598628401756,
-      "learning_rate": 0.00019999589639440682,
-      "loss": 46.0,
-      "step": 17917
-    },
-    {
-      "epoch": 2.885542896251862,
-      "grad_norm": 0.001328574144281447,
-      "learning_rate": 0.00019999589593607296,
-      "loss": 46.0,
-      "step": 17918
-    },
-    {
-      "epoch": 2.8857039333306496,
-      "grad_norm": 0.01616380736231804,
-      "learning_rate": 0.0001999958954777135,
-      "loss": 46.0,
-      "step": 17919
-    },
-    {
-      "epoch": 2.885864970409437,
-      "grad_norm": 0.0033234127331525087,
-      "learning_rate": 0.00019999589501932847,
-      "loss": 46.0,
-      "step": 17920
-    },
-    {
-      "epoch": 2.886026007488224,
-      "grad_norm": 0.0006904685287736356,
-      "learning_rate": 0.00019999589456091784,
-      "loss": 46.0,
-      "step": 17921
-    },
-    {
-      "epoch": 2.8861870445670115,
-      "grad_norm": 0.0013475097948685288,
-      "learning_rate": 0.00019999589410248163,
-      "loss": 46.0,
-      "step": 17922
-    },
-    {
-      "epoch": 2.886348081645799,
-      "grad_norm": 0.003214400727301836,
-      "learning_rate": 0.0001999958936440198,
-      "loss": 46.0,
-      "step": 17923
-    },
-    {
-      "epoch": 2.8865091187245864,
-      "grad_norm": 0.009359680116176605,
-      "learning_rate": 0.0001999958931855324,
-      "loss": 46.0,
-      "step": 17924
-    },
-    {
-      "epoch": 2.886670155803374,
-      "grad_norm": 0.0005583578022196889,
-      "learning_rate": 0.0001999958927270194,
-      "loss": 46.0,
-      "step": 17925
-    },
-    {
-      "epoch": 2.8868311928821613,
-      "grad_norm": 0.0026180287823081017,
-      "learning_rate": 0.0001999958922684808,
-      "loss": 46.0,
-      "step": 17926
-    },
-    {
-      "epoch": 2.8869922299609483,
-      "grad_norm": 0.001509267371147871,
-      "learning_rate": 0.0001999958918099166,
-      "loss": 46.0,
-      "step": 17927
-    },
-    {
-      "epoch": 2.8871532670397357,
-      "grad_norm": 0.0015041035367175937,
-      "learning_rate": 0.00019999589135132683,
-      "loss": 46.0,
-      "step": 17928
-    },
-    {
-      "epoch": 2.887314304118523,
-      "grad_norm": 0.004807292949408293,
-      "learning_rate": 0.00019999589089271145,
-      "loss": 46.0,
-      "step": 17929
-    },
-    {
-      "epoch": 2.8874753411973106,
-      "grad_norm": 0.006815982982516289,
-      "learning_rate": 0.00019999589043407046,
-      "loss": 46.0,
-      "step": 17930
-    },
-    {
-      "epoch": 2.887636378276098,
-      "grad_norm": 0.0037054154090583324,
-      "learning_rate": 0.00019999588997540388,
-      "loss": 46.0,
-      "step": 17931
-    },
-    {
-      "epoch": 2.8877974153548855,
-      "grad_norm": 0.002724493620917201,
-      "learning_rate": 0.00019999588951671174,
-      "loss": 46.0,
-      "step": 17932
-    },
-    {
-      "epoch": 2.887958452433673,
-      "grad_norm": 0.0028292504139244556,
-      "learning_rate": 0.00019999588905799398,
-      "loss": 46.0,
-      "step": 17933
-    },
-    {
-      "epoch": 2.8881194895124604,
-      "grad_norm": 0.002641974249854684,
-      "learning_rate": 0.00019999588859925062,
-      "loss": 46.0,
-      "step": 17934
-    },
-    {
-      "epoch": 2.888280526591248,
-      "grad_norm": 0.0010813666740432382,
-      "learning_rate": 0.0001999958881404817,
-      "loss": 46.0,
-      "step": 17935
-    },
-    {
-      "epoch": 2.888441563670035,
-      "grad_norm": 0.003969601821154356,
-      "learning_rate": 0.00019999588768168715,
-      "loss": 46.0,
-      "step": 17936
-    },
-    {
-      "epoch": 2.8886026007488224,
-      "grad_norm": 0.0051483758725225925,
-      "learning_rate": 0.00019999588722286702,
-      "loss": 46.0,
-      "step": 17937
-    },
-    {
-      "epoch": 2.88876363782761,
-      "grad_norm": 0.0006969957030378282,
-      "learning_rate": 0.0001999958867640213,
-      "loss": 46.0,
-      "step": 17938
-    },
-    {
-      "epoch": 2.8889246749063973,
-      "grad_norm": 0.0020348557736724615,
-      "learning_rate": 0.00019999588630514997,
-      "loss": 46.0,
-      "step": 17939
-    },
-    {
-      "epoch": 2.8890857119851847,
-      "grad_norm": 0.005832456983625889,
-      "learning_rate": 0.00019999588584625305,
-      "loss": 46.0,
-      "step": 17940
-    },
-    {
-      "epoch": 2.8892467490639717,
-      "grad_norm": 0.008833184838294983,
-      "learning_rate": 0.00019999588538733055,
-      "loss": 46.0,
-      "step": 17941
-    },
-    {
-      "epoch": 2.889407786142759,
-      "grad_norm": 0.003829443361610174,
-      "learning_rate": 0.00019999588492838246,
-      "loss": 46.0,
-      "step": 17942
-    },
-    {
-      "epoch": 2.8895688232215466,
-      "grad_norm": 0.0023214484099298716,
-      "learning_rate": 0.00019999588446940875,
-      "loss": 46.0,
-      "step": 17943
-    },
-    {
-      "epoch": 2.889729860300334,
-      "grad_norm": 0.006406680215150118,
-      "learning_rate": 0.00019999588401040948,
-      "loss": 46.0,
-      "step": 17944
-    },
-    {
-      "epoch": 2.8898908973791215,
-      "grad_norm": 0.0021372276823967695,
-      "learning_rate": 0.0001999958835513846,
-      "loss": 46.0,
-      "step": 17945
-    },
-    {
-      "epoch": 2.890051934457909,
-      "grad_norm": 0.004949562717229128,
-      "learning_rate": 0.00019999588309233413,
-      "loss": 46.0,
-      "step": 17946
-    },
-    {
-      "epoch": 2.8902129715366964,
-      "grad_norm": 0.005079805850982666,
-      "learning_rate": 0.00019999588263325805,
-      "loss": 46.0,
-      "step": 17947
-    },
-    {
-      "epoch": 2.890374008615484,
-      "grad_norm": 0.0028896895237267017,
-      "learning_rate": 0.0001999958821741564,
-      "loss": 46.0,
-      "step": 17948
-    },
-    {
-      "epoch": 2.8905350456942713,
-      "grad_norm": 0.0009584093349985778,
-      "learning_rate": 0.00019999588171502915,
-      "loss": 46.0,
-      "step": 17949
-    },
-    {
-      "epoch": 2.8906960827730583,
-      "grad_norm": 0.0053268177434802055,
-      "learning_rate": 0.00019999588125587628,
-      "loss": 46.0,
-      "step": 17950
-    },
-    {
-      "epoch": 2.890857119851846,
-      "grad_norm": 0.001871272106654942,
-      "learning_rate": 0.00019999588079669785,
-      "loss": 46.0,
-      "step": 17951
-    },
-    {
-      "epoch": 2.8910181569306332,
-      "grad_norm": 0.005613320041447878,
-      "learning_rate": 0.0001999958803374938,
-      "loss": 46.0,
-      "step": 17952
-    },
-    {
-      "epoch": 2.8911791940094207,
-      "grad_norm": 0.006312674377113581,
-      "learning_rate": 0.00019999587987826417,
-      "loss": 46.0,
-      "step": 17953
-    },
-    {
-      "epoch": 2.891340231088208,
-      "grad_norm": 0.008954005315899849,
-      "learning_rate": 0.00019999587941900898,
-      "loss": 46.0,
-      "step": 17954
-    },
-    {
-      "epoch": 2.8915012681669956,
-      "grad_norm": 0.0011943741701543331,
-      "learning_rate": 0.00019999587895972815,
-      "loss": 46.0,
-      "step": 17955
-    },
-    {
-      "epoch": 2.8916623052457826,
-      "grad_norm": 0.008451980538666248,
-      "learning_rate": 0.00019999587850042173,
-      "loss": 46.0,
-      "step": 17956
-    },
-    {
-      "epoch": 2.89182334232457,
-      "grad_norm": 0.0007649046019650996,
-      "learning_rate": 0.00019999587804108972,
-      "loss": 46.0,
-      "step": 17957
-    },
-    {
-      "epoch": 2.8919843794033575,
-      "grad_norm": 0.0014025458367541432,
-      "learning_rate": 0.0001999958775817321,
-      "loss": 46.0,
-      "step": 17958
-    },
-    {
-      "epoch": 2.892145416482145,
-      "grad_norm": 0.0024657694157212973,
-      "learning_rate": 0.0001999958771223489,
-      "loss": 46.0,
-      "step": 17959
-    },
-    {
-      "epoch": 2.8923064535609324,
-      "grad_norm": 0.002833641367033124,
-      "learning_rate": 0.00019999587666294012,
-      "loss": 46.0,
-      "step": 17960
-    },
-    {
-      "epoch": 2.89246749063972,
-      "grad_norm": 0.002567199757322669,
-      "learning_rate": 0.00019999587620350576,
-      "loss": 46.0,
-      "step": 17961
-    },
-    {
-      "epoch": 2.8926285277185073,
-      "grad_norm": 0.004578740336000919,
-      "learning_rate": 0.0001999958757440458,
-      "loss": 46.0,
-      "step": 17962
-    },
-    {
-      "epoch": 2.8927895647972948,
-      "grad_norm": 0.0018670371500775218,
-      "learning_rate": 0.0001999958752845602,
-      "loss": 46.0,
-      "step": 17963
-    },
-    {
-      "epoch": 2.892950601876082,
-      "grad_norm": 0.001153697376139462,
-      "learning_rate": 0.00019999587482504906,
-      "loss": 46.0,
-      "step": 17964
-    },
-    {
-      "epoch": 2.8931116389548692,
-      "grad_norm": 0.001426900620572269,
-      "learning_rate": 0.00019999587436551227,
-      "loss": 46.0,
-      "step": 17965
-    },
-    {
-      "epoch": 2.8932726760336567,
-      "grad_norm": 0.008064872585237026,
-      "learning_rate": 0.00019999587390594995,
-      "loss": 46.0,
-      "step": 17966
-    },
-    {
-      "epoch": 2.893433713112444,
-      "grad_norm": 0.0007149272714741528,
-      "learning_rate": 0.000199995873446362,
-      "loss": 46.0,
-      "step": 17967
-    },
-    {
-      "epoch": 2.8935947501912316,
-      "grad_norm": 0.007296116556972265,
-      "learning_rate": 0.00019999587298674847,
-      "loss": 46.0,
-      "step": 17968
-    },
-    {
-      "epoch": 2.893755787270019,
-      "grad_norm": 0.009523754008114338,
-      "learning_rate": 0.0001999958725271093,
-      "loss": 46.0,
-      "step": 17969
-    },
-    {
-      "epoch": 2.893916824348806,
-      "grad_norm": 0.0011613063979893923,
-      "learning_rate": 0.00019999587206744458,
-      "loss": 46.0,
-      "step": 17970
-    },
-    {
-      "epoch": 2.8940778614275935,
-      "grad_norm": 0.004892035387456417,
-      "learning_rate": 0.00019999587160775427,
-      "loss": 46.0,
-      "step": 17971
-    },
-    {
-      "epoch": 2.894238898506381,
-      "grad_norm": 0.001188953290693462,
-      "learning_rate": 0.00019999587114803835,
-      "loss": 46.0,
-      "step": 17972
-    },
-    {
-      "epoch": 2.8943999355851684,
-      "grad_norm": 0.006213276647031307,
-      "learning_rate": 0.00019999587068829686,
-      "loss": 46.0,
-      "step": 17973
-    },
-    {
-      "epoch": 2.894560972663956,
-      "grad_norm": 0.005128875840455294,
-      "learning_rate": 0.00019999587022852974,
-      "loss": 46.0,
-      "step": 17974
-    },
-    {
-      "epoch": 2.8947220097427433,
-      "grad_norm": 0.0022443195339292288,
-      "learning_rate": 0.00019999586976873705,
-      "loss": 46.0,
-      "step": 17975
-    },
-    {
-      "epoch": 2.8948830468215307,
-      "grad_norm": 0.005779961124062538,
-      "learning_rate": 0.00019999586930891878,
-      "loss": 46.0,
-      "step": 17976
-    },
-    {
-      "epoch": 2.895044083900318,
-      "grad_norm": 0.002377884928137064,
-      "learning_rate": 0.0001999958688490749,
-      "loss": 46.0,
-      "step": 17977
-    },
-    {
-      "epoch": 2.8952051209791057,
-      "grad_norm": 0.0014512325869873166,
-      "learning_rate": 0.00019999586838920542,
-      "loss": 46.0,
-      "step": 17978
-    },
-    {
-      "epoch": 2.895366158057893,
-      "grad_norm": 0.0014586142497137189,
-      "learning_rate": 0.00019999586792931036,
-      "loss": 46.0,
-      "step": 17979
-    },
-    {
-      "epoch": 2.89552719513668,
-      "grad_norm": 0.002845063805580139,
-      "learning_rate": 0.00019999586746938968,
-      "loss": 46.0,
-      "step": 17980
-    },
-    {
-      "epoch": 2.8956882322154676,
-      "grad_norm": 0.005690551362931728,
-      "learning_rate": 0.00019999586700944342,
-      "loss": 46.0,
-      "step": 17981
-    },
-    {
-      "epoch": 2.895849269294255,
-      "grad_norm": 0.004403562750667334,
-      "learning_rate": 0.00019999586654947157,
-      "loss": 46.0,
-      "step": 17982
-    },
-    {
-      "epoch": 2.8960103063730425,
-      "grad_norm": 0.00412313686683774,
-      "learning_rate": 0.0001999958660894741,
-      "loss": 46.0,
-      "step": 17983
-    },
-    {
-      "epoch": 2.89617134345183,
-      "grad_norm": 0.0013536461628973484,
-      "learning_rate": 0.00019999586562945108,
-      "loss": 46.0,
-      "step": 17984
-    },
-    {
-      "epoch": 2.896332380530617,
-      "grad_norm": 0.003979323897510767,
-      "learning_rate": 0.00019999586516940244,
-      "loss": 46.0,
-      "step": 17985
-    },
-    {
-      "epoch": 2.8964934176094044,
-      "grad_norm": 0.002008540090173483,
-      "learning_rate": 0.00019999586470932821,
-      "loss": 46.0,
-      "step": 17986
-    },
-    {
-      "epoch": 2.896654454688192,
-      "grad_norm": 0.0048518432304263115,
-      "learning_rate": 0.0001999958642492284,
-      "loss": 46.0,
-      "step": 17987
-    },
-    {
-      "epoch": 2.8968154917669793,
-      "grad_norm": 0.003518218407407403,
-      "learning_rate": 0.00019999586378910297,
-      "loss": 46.0,
-      "step": 17988
-    },
-    {
-      "epoch": 2.8969765288457667,
-      "grad_norm": 0.0016916838940232992,
-      "learning_rate": 0.000199995863328952,
-      "loss": 46.0,
-      "step": 17989
-    },
-    {
-      "epoch": 2.897137565924554,
-      "grad_norm": 0.002047501038759947,
-      "learning_rate": 0.00019999586286877539,
-      "loss": 46.0,
-      "step": 17990
-    },
-    {
-      "epoch": 2.8972986030033416,
-      "grad_norm": 0.012943160720169544,
-      "learning_rate": 0.00019999586240857317,
-      "loss": 46.0,
-      "step": 17991
-    },
-    {
-      "epoch": 2.897459640082129,
-      "grad_norm": 0.013662423938512802,
-      "learning_rate": 0.0001999958619483454,
-      "loss": 46.0,
-      "step": 17992
-    },
-    {
-      "epoch": 2.8976206771609165,
-      "grad_norm": 0.0072461869567632675,
-      "learning_rate": 0.000199995861488092,
-      "loss": 46.0,
-      "step": 17993
-    },
-    {
-      "epoch": 2.8977817142397035,
-      "grad_norm": 0.002726959530264139,
-      "learning_rate": 0.00019999586102781303,
-      "loss": 46.0,
-      "step": 17994
-    },
-    {
-      "epoch": 2.897942751318491,
-      "grad_norm": 0.0014070110628381371,
-      "learning_rate": 0.0001999958605675085,
-      "loss": 46.0,
-      "step": 17995
-    },
-    {
-      "epoch": 2.8981037883972784,
-      "grad_norm": 0.006584594491869211,
-      "learning_rate": 0.0001999958601071783,
-      "loss": 46.0,
-      "step": 17996
-    },
-    {
-      "epoch": 2.898264825476066,
-      "grad_norm": 0.004790824837982655,
-      "learning_rate": 0.00019999585964682254,
-      "loss": 46.0,
-      "step": 17997
-    },
-    {
-      "epoch": 2.8984258625548533,
-      "grad_norm": 0.0011857866775244474,
-      "learning_rate": 0.00019999585918644116,
-      "loss": 46.0,
-      "step": 17998
-    },
-    {
-      "epoch": 2.898586899633641,
-      "grad_norm": 0.002769236220046878,
-      "learning_rate": 0.00019999585872603422,
-      "loss": 46.0,
-      "step": 17999
-    },
-    {
-      "epoch": 2.898747936712428,
-      "grad_norm": 0.0023509215097874403,
-      "learning_rate": 0.0001999958582656017,
-      "loss": 46.0,
-      "step": 18000
-    },
-    {
-      "epoch": 2.8989089737912153,
-      "grad_norm": 0.008861015550792217,
-      "learning_rate": 0.00019999585780514355,
-      "loss": 46.0,
-      "step": 18001
-    },
-    {
-      "epoch": 2.8990700108700027,
-      "grad_norm": 0.002614819910377264,
-      "learning_rate": 0.00019999585734465982,
-      "loss": 46.0,
-      "step": 18002
-    },
-    {
-      "epoch": 2.89923104794879,
-      "grad_norm": 0.0011969548650085926,
-      "learning_rate": 0.0001999958568841505,
-      "loss": 46.0,
-      "step": 18003
-    },
-    {
-      "epoch": 2.8993920850275776,
-      "grad_norm": 0.0031611144077032804,
-      "learning_rate": 0.0001999958564236156,
-      "loss": 46.0,
-      "step": 18004
-    },
-    {
-      "epoch": 2.899553122106365,
-      "grad_norm": 0.003020546166226268,
-      "learning_rate": 0.00019999585596305509,
-      "loss": 46.0,
-      "step": 18005
-    },
-    {
-      "epoch": 2.8997141591851525,
-      "grad_norm": 0.0036904688458889723,
-      "learning_rate": 0.00019999585550246898,
-      "loss": 46.0,
-      "step": 18006
-    },
-    {
-      "epoch": 2.89987519626394,
-      "grad_norm": 0.001113614533096552,
-      "learning_rate": 0.0001999958550418573,
-      "loss": 46.0,
-      "step": 18007
-    },
-    {
-      "epoch": 2.9000362333427274,
-      "grad_norm": 0.003312904154881835,
-      "learning_rate": 0.00019999585458121998,
-      "loss": 46.0,
-      "step": 18008
-    },
-    {
-      "epoch": 2.9001972704215144,
-      "grad_norm": 0.004726111423224211,
-      "learning_rate": 0.0001999958541205571,
-      "loss": 46.0,
-      "step": 18009
-    },
-    {
-      "epoch": 2.900358307500302,
-      "grad_norm": 0.0022232099436223507,
-      "learning_rate": 0.00019999585365986864,
-      "loss": 46.0,
-      "step": 18010
-    },
-    {
-      "epoch": 2.9005193445790893,
-      "grad_norm": 0.008369456976652145,
-      "learning_rate": 0.00019999585319915454,
-      "loss": 46.0,
-      "step": 18011
-    },
-    {
-      "epoch": 2.900680381657877,
-      "grad_norm": 0.006913413759320974,
-      "learning_rate": 0.00019999585273841489,
-      "loss": 46.0,
-      "step": 18012
-    },
-    {
-      "epoch": 2.9008414187366642,
-      "grad_norm": 0.005468557123094797,
-      "learning_rate": 0.00019999585227764964,
-      "loss": 46.0,
-      "step": 18013
-    },
-    {
-      "epoch": 2.9010024558154512,
-      "grad_norm": 0.000931234797462821,
-      "learning_rate": 0.0001999958518168588,
-      "loss": 46.0,
-      "step": 18014
-    },
-    {
-      "epoch": 2.9011634928942387,
-      "grad_norm": 0.0005673962878063321,
-      "learning_rate": 0.00019999585135604232,
-      "loss": 46.0,
-      "step": 18015
-    },
-    {
-      "epoch": 2.901324529973026,
-      "grad_norm": 0.0006478444556705654,
-      "learning_rate": 0.00019999585089520029,
-      "loss": 46.0,
-      "step": 18016
-    },
-    {
-      "epoch": 2.9014855670518136,
-      "grad_norm": 0.0009890447836369276,
-      "learning_rate": 0.00019999585043433267,
-      "loss": 46.0,
-      "step": 18017
-    },
-    {
-      "epoch": 2.901646604130601,
-      "grad_norm": 0.004521854687482119,
-      "learning_rate": 0.00019999584997343944,
-      "loss": 46.0,
-      "step": 18018
-    },
-    {
-      "epoch": 2.9018076412093885,
-      "grad_norm": 0.0035858110059052706,
-      "learning_rate": 0.00019999584951252062,
-      "loss": 46.0,
-      "step": 18019
-    },
-    {
-      "epoch": 2.901968678288176,
-      "grad_norm": 0.005323626101016998,
-      "learning_rate": 0.00019999584905157618,
-      "loss": 46.0,
-      "step": 18020
-    },
-    {
-      "epoch": 2.9021297153669634,
-      "grad_norm": 0.003042892087250948,
-      "learning_rate": 0.0001999958485906062,
-      "loss": 46.0,
-      "step": 18021
-    },
-    {
-      "epoch": 2.902290752445751,
-      "grad_norm": 0.001452945638448,
-      "learning_rate": 0.00019999584812961055,
-      "loss": 46.0,
-      "step": 18022
-    },
-    {
-      "epoch": 2.902451789524538,
-      "grad_norm": 0.0007284888415597379,
-      "learning_rate": 0.00019999584766858939,
-      "loss": 46.0,
-      "step": 18023
-    },
-    {
-      "epoch": 2.9026128266033253,
-      "grad_norm": 0.007866005413234234,
-      "learning_rate": 0.00019999584720754258,
-      "loss": 46.0,
-      "step": 18024
-    },
-    {
-      "epoch": 2.9027738636821128,
-      "grad_norm": 0.0077062686905264854,
-      "learning_rate": 0.0001999958467464702,
-      "loss": 46.0,
-      "step": 18025
-    },
-    {
-      "epoch": 2.9029349007609,
-      "grad_norm": 0.0011744254734367132,
-      "learning_rate": 0.00019999584628537222,
-      "loss": 46.0,
-      "step": 18026
-    },
-    {
-      "epoch": 2.9030959378396877,
-      "grad_norm": 0.002215759828686714,
-      "learning_rate": 0.00019999584582424865,
-      "loss": 46.0,
-      "step": 18027
-    },
-    {
-      "epoch": 2.903256974918475,
-      "grad_norm": 0.004121643025428057,
-      "learning_rate": 0.00019999584536309947,
-      "loss": 46.0,
-      "step": 18028
-    },
-    {
-      "epoch": 2.903418011997262,
-      "grad_norm": 0.0020589560735970736,
-      "learning_rate": 0.00019999584490192472,
-      "loss": 46.0,
-      "step": 18029
-    },
-    {
-      "epoch": 2.9035790490760496,
-      "grad_norm": 0.002641829429194331,
-      "learning_rate": 0.00019999584444072436,
-      "loss": 46.0,
-      "step": 18030
-    },
-    {
-      "epoch": 2.903740086154837,
-      "grad_norm": 0.0031198933720588684,
-      "learning_rate": 0.00019999584397949842,
-      "loss": 46.0,
-      "step": 18031
-    },
-    {
-      "epoch": 2.9039011232336245,
-      "grad_norm": 0.00571768032386899,
-      "learning_rate": 0.00019999584351824688,
-      "loss": 46.0,
-      "step": 18032
-    },
-    {
-      "epoch": 2.904062160312412,
-      "grad_norm": 0.004765548277646303,
-      "learning_rate": 0.00019999584305696973,
-      "loss": 46.0,
-      "step": 18033
-    },
-    {
-      "epoch": 2.9042231973911994,
-      "grad_norm": 0.0016701009590178728,
-      "learning_rate": 0.000199995842595667,
-      "loss": 46.0,
-      "step": 18034
-    },
-    {
-      "epoch": 2.904384234469987,
-      "grad_norm": 0.0026695849373936653,
-      "learning_rate": 0.00019999584213433868,
-      "loss": 46.0,
-      "step": 18035
-    },
-    {
-      "epoch": 2.9045452715487743,
-      "grad_norm": 0.0013127813581377268,
-      "learning_rate": 0.00019999584167298477,
-      "loss": 46.0,
-      "step": 18036
-    },
-    {
-      "epoch": 2.9047063086275617,
-      "grad_norm": 0.008485066704452038,
-      "learning_rate": 0.00019999584121160527,
-      "loss": 46.0,
-      "step": 18037
-    },
-    {
-      "epoch": 2.9048673457063487,
-      "grad_norm": 0.0013082843506708741,
-      "learning_rate": 0.00019999584075020016,
-      "loss": 46.0,
-      "step": 18038
-    },
-    {
-      "epoch": 2.905028382785136,
-      "grad_norm": 0.0007086473633535206,
-      "learning_rate": 0.00019999584028876946,
-      "loss": 46.0,
-      "step": 18039
-    },
-    {
-      "epoch": 2.9051894198639236,
-      "grad_norm": 0.0026978799141943455,
-      "learning_rate": 0.00019999583982731315,
-      "loss": 46.0,
-      "step": 18040
-    },
-    {
-      "epoch": 2.905350456942711,
-      "grad_norm": 0.0029192203655838966,
-      "learning_rate": 0.00019999583936583127,
-      "loss": 46.0,
-      "step": 18041
-    },
-    {
-      "epoch": 2.9055114940214986,
-      "grad_norm": 0.0014519410906359553,
-      "learning_rate": 0.00019999583890432381,
-      "loss": 46.0,
-      "step": 18042
-    },
-    {
-      "epoch": 2.905672531100286,
-      "grad_norm": 0.0009246626868844032,
-      "learning_rate": 0.00019999583844279074,
-      "loss": 46.0,
-      "step": 18043
-    },
-    {
-      "epoch": 2.905833568179073,
-      "grad_norm": 0.002123977057635784,
-      "learning_rate": 0.00019999583798123205,
-      "loss": 46.0,
-      "step": 18044
-    },
-    {
-      "epoch": 2.9059946052578605,
-      "grad_norm": 0.014459672383964062,
-      "learning_rate": 0.00019999583751964778,
-      "loss": 46.0,
-      "step": 18045
-    },
-    {
-      "epoch": 2.906155642336648,
-      "grad_norm": 0.0011551082134246826,
-      "learning_rate": 0.00019999583705803794,
-      "loss": 46.0,
-      "step": 18046
-    },
-    {
-      "epoch": 2.9063166794154354,
-      "grad_norm": 0.0029360789339989424,
-      "learning_rate": 0.0001999958365964025,
-      "loss": 46.0,
-      "step": 18047
-    },
-    {
-      "epoch": 2.906477716494223,
-      "grad_norm": 0.0042358869686722755,
-      "learning_rate": 0.00019999583613474145,
-      "loss": 46.0,
-      "step": 18048
-    },
-    {
-      "epoch": 2.9066387535730103,
-      "grad_norm": 0.001116848667152226,
-      "learning_rate": 0.00019999583567305483,
-      "loss": 46.0,
-      "step": 18049
-    },
-    {
-      "epoch": 2.9067997906517977,
-      "grad_norm": 0.001232419046573341,
-      "learning_rate": 0.00019999583521134256,
-      "loss": 46.0,
-      "step": 18050
-    },
-    {
-      "epoch": 2.906960827730585,
-      "grad_norm": 0.010546743869781494,
-      "learning_rate": 0.00019999583474960474,
-      "loss": 46.0,
-      "step": 18051
-    },
-    {
-      "epoch": 2.9071218648093726,
-      "grad_norm": 0.0016749532660469413,
-      "learning_rate": 0.00019999583428784132,
-      "loss": 46.0,
-      "step": 18052
-    },
-    {
-      "epoch": 2.9072829018881596,
-      "grad_norm": 0.003740352811291814,
-      "learning_rate": 0.00019999583382605232,
-      "loss": 46.0,
-      "step": 18053
-    },
-    {
-      "epoch": 2.907443938966947,
-      "grad_norm": 0.0039942823350429535,
-      "learning_rate": 0.00019999583336423774,
-      "loss": 46.0,
-      "step": 18054
-    },
-    {
-      "epoch": 2.9076049760457345,
-      "grad_norm": 0.0011449663434177637,
-      "learning_rate": 0.0001999958329023975,
-      "loss": 46.0,
-      "step": 18055
-    },
-    {
-      "epoch": 2.907766013124522,
-      "grad_norm": 0.0022023271303623915,
-      "learning_rate": 0.00019999583244053172,
-      "loss": 46.0,
-      "step": 18056
-    },
-    {
-      "epoch": 2.9079270502033094,
-      "grad_norm": 0.0025111560244113207,
-      "learning_rate": 0.00019999583197864034,
-      "loss": 46.0,
-      "step": 18057
-    },
-    {
-      "epoch": 2.9080880872820964,
-      "grad_norm": 0.0006832068902440369,
-      "learning_rate": 0.00019999583151672335,
-      "loss": 46.0,
-      "step": 18058
-    },
-    {
-      "epoch": 2.908249124360884,
-      "grad_norm": 0.004509684629738331,
-      "learning_rate": 0.0001999958310547808,
-      "loss": 46.0,
-      "step": 18059
-    },
-    {
-      "epoch": 2.9084101614396713,
-      "grad_norm": 0.004035685211420059,
-      "learning_rate": 0.0001999958305928126,
-      "loss": 46.0,
-      "step": 18060
-    },
-    {
-      "epoch": 2.908571198518459,
-      "grad_norm": 0.0012917224084958434,
-      "learning_rate": 0.00019999583013081886,
-      "loss": 46.0,
-      "step": 18061
-    },
-    {
-      "epoch": 2.9087322355972463,
-      "grad_norm": 0.012897004373371601,
-      "learning_rate": 0.0001999958296687995,
-      "loss": 46.0,
-      "step": 18062
-    },
-    {
-      "epoch": 2.9088932726760337,
-      "grad_norm": 0.0030188795644789934,
-      "learning_rate": 0.00019999582920675451,
-      "loss": 46.0,
-      "step": 18063
-    },
-    {
-      "epoch": 2.909054309754821,
-      "grad_norm": 0.0007855913718231022,
-      "learning_rate": 0.000199995828744684,
-      "loss": 46.0,
-      "step": 18064
-    },
-    {
-      "epoch": 2.9092153468336086,
-      "grad_norm": 0.005138868000358343,
-      "learning_rate": 0.00019999582828258787,
-      "loss": 46.0,
-      "step": 18065
-    },
-    {
-      "epoch": 2.909376383912396,
-      "grad_norm": 0.00433430215343833,
-      "learning_rate": 0.00019999582782046613,
-      "loss": 46.0,
-      "step": 18066
-    },
-    {
-      "epoch": 2.909537420991183,
-      "grad_norm": 0.005203371401876211,
-      "learning_rate": 0.0001999958273583188,
-      "loss": 46.0,
-      "step": 18067
-    },
-    {
-      "epoch": 2.9096984580699705,
-      "grad_norm": 0.005903835408389568,
-      "learning_rate": 0.0001999958268961459,
-      "loss": 46.0,
-      "step": 18068
-    },
-    {
-      "epoch": 2.909859495148758,
-      "grad_norm": 0.0006122100166976452,
-      "learning_rate": 0.00019999582643394738,
-      "loss": 46.0,
-      "step": 18069
-    },
-    {
-      "epoch": 2.9100205322275454,
-      "grad_norm": 0.003275372087955475,
-      "learning_rate": 0.00019999582597172327,
-      "loss": 46.0,
-      "step": 18070
-    },
-    {
-      "epoch": 2.910181569306333,
-      "grad_norm": 0.0025616525672376156,
-      "learning_rate": 0.00019999582550947356,
-      "loss": 46.0,
-      "step": 18071
-    },
-    {
-      "epoch": 2.9103426063851203,
-      "grad_norm": 0.0010053514270111918,
-      "learning_rate": 0.00019999582504719827,
-      "loss": 46.0,
-      "step": 18072
-    },
-    {
-      "epoch": 2.9105036434639073,
-      "grad_norm": 0.006247455719858408,
-      "learning_rate": 0.0001999958245848974,
-      "loss": 46.0,
-      "step": 18073
-    },
-    {
-      "epoch": 2.910664680542695,
-      "grad_norm": 0.01034959964454174,
-      "learning_rate": 0.0001999958241225709,
-      "loss": 46.0,
-      "step": 18074
-    },
-    {
-      "epoch": 2.9108257176214822,
-      "grad_norm": 0.0024148093070834875,
-      "learning_rate": 0.00019999582366021882,
-      "loss": 46.0,
-      "step": 18075
-    },
-    {
-      "epoch": 2.9109867547002697,
-      "grad_norm": 0.0012868557823821902,
-      "learning_rate": 0.00019999582319784115,
-      "loss": 46.0,
-      "step": 18076
-    },
-    {
-      "epoch": 2.911147791779057,
-      "grad_norm": 0.009551554918289185,
-      "learning_rate": 0.0001999958227354379,
-      "loss": 46.0,
-      "step": 18077
-    },
-    {
-      "epoch": 2.9113088288578446,
-      "grad_norm": 0.012105186469852924,
-      "learning_rate": 0.00019999582227300903,
-      "loss": 46.0,
-      "step": 18078
-    },
-    {
-      "epoch": 2.911469865936632,
-      "grad_norm": 0.0015006529865786433,
-      "learning_rate": 0.0001999958218105546,
-      "loss": 46.0,
-      "step": 18079
-    },
-    {
-      "epoch": 2.9116309030154195,
-      "grad_norm": 0.0014158220728859305,
-      "learning_rate": 0.00019999582134807455,
-      "loss": 46.0,
-      "step": 18080
-    },
-    {
-      "epoch": 2.911791940094207,
-      "grad_norm": 0.0015673128655180335,
-      "learning_rate": 0.0001999958208855689,
-      "loss": 46.0,
-      "step": 18081
-    },
-    {
-      "epoch": 2.911952977172994,
-      "grad_norm": 0.002433421555906534,
-      "learning_rate": 0.00019999582042303765,
-      "loss": 46.0,
-      "step": 18082
-    },
-    {
-      "epoch": 2.9121140142517814,
-      "grad_norm": 0.00192752736620605,
-      "learning_rate": 0.00019999581996048085,
-      "loss": 46.0,
-      "step": 18083
-    },
-    {
-      "epoch": 2.912275051330569,
-      "grad_norm": 0.0029742789920419455,
-      "learning_rate": 0.00019999581949789843,
-      "loss": 46.0,
-      "step": 18084
-    },
-    {
-      "epoch": 2.9124360884093563,
-      "grad_norm": 0.006985408719629049,
-      "learning_rate": 0.00019999581903529042,
-      "loss": 46.0,
-      "step": 18085
-    },
-    {
-      "epoch": 2.9125971254881438,
-      "grad_norm": 0.0010164888808503747,
-      "learning_rate": 0.0001999958185726568,
-      "loss": 46.0,
-      "step": 18086
-    },
-    {
-      "epoch": 2.9127581625669308,
-      "grad_norm": 0.009255086071789265,
-      "learning_rate": 0.0001999958181099976,
-      "loss": 46.0,
-      "step": 18087
-    },
-    {
-      "epoch": 2.912919199645718,
-      "grad_norm": 0.004060061648488045,
-      "learning_rate": 0.00019999581764731282,
-      "loss": 46.0,
-      "step": 18088
-    },
-    {
-      "epoch": 2.9130802367245057,
-      "grad_norm": 0.0021101615857332945,
-      "learning_rate": 0.00019999581718460241,
-      "loss": 46.0,
-      "step": 18089
-    },
-    {
-      "epoch": 2.913241273803293,
-      "grad_norm": 0.0030538870487362146,
-      "learning_rate": 0.00019999581672186642,
-      "loss": 46.0,
-      "step": 18090
-    },
-    {
-      "epoch": 2.9134023108820806,
-      "grad_norm": 0.002641197992488742,
-      "learning_rate": 0.00019999581625910486,
-      "loss": 46.0,
-      "step": 18091
-    },
-    {
-      "epoch": 2.913563347960868,
-      "grad_norm": 0.002696691546589136,
-      "learning_rate": 0.00019999581579631766,
-      "loss": 46.0,
-      "step": 18092
-    },
-    {
-      "epoch": 2.9137243850396555,
-      "grad_norm": 0.0015544277848675847,
-      "learning_rate": 0.0001999958153335049,
-      "loss": 46.0,
-      "step": 18093
-    },
-    {
-      "epoch": 2.913885422118443,
-      "grad_norm": 0.0009993682615458965,
-      "learning_rate": 0.00019999581487066656,
-      "loss": 46.0,
-      "step": 18094
-    },
-    {
-      "epoch": 2.9140464591972304,
-      "grad_norm": 0.005122416187077761,
-      "learning_rate": 0.00019999581440780263,
-      "loss": 46.0,
-      "step": 18095
-    },
-    {
-      "epoch": 2.914207496276018,
-      "grad_norm": 0.0014448229921981692,
-      "learning_rate": 0.00019999581394491306,
-      "loss": 46.0,
-      "step": 18096
-    },
-    {
-      "epoch": 2.914368533354805,
-      "grad_norm": 0.0014858562499284744,
-      "learning_rate": 0.00019999581348199792,
-      "loss": 46.0,
-      "step": 18097
-    },
-    {
-      "epoch": 2.9145295704335923,
-      "grad_norm": 0.002398568671196699,
-      "learning_rate": 0.0001999958130190572,
-      "loss": 46.0,
-      "step": 18098
-    },
-    {
-      "epoch": 2.9146906075123797,
-      "grad_norm": 0.0005177139537408948,
-      "learning_rate": 0.00019999581255609087,
-      "loss": 46.0,
-      "step": 18099
-    },
-    {
-      "epoch": 2.914851644591167,
-      "grad_norm": 0.010388285852968693,
-      "learning_rate": 0.00019999581209309894,
-      "loss": 46.0,
-      "step": 18100
-    },
-    {
-      "epoch": 2.9150126816699546,
-      "grad_norm": 0.012333924882113934,
-      "learning_rate": 0.00019999581163008143,
-      "loss": 46.0,
-      "step": 18101
-    },
-    {
-      "epoch": 2.9151737187487416,
-      "grad_norm": 0.005881312303245068,
-      "learning_rate": 0.0001999958111670383,
-      "loss": 46.0,
-      "step": 18102
-    },
-    {
-      "epoch": 2.915334755827529,
-      "grad_norm": 0.004993600770831108,
-      "learning_rate": 0.00019999581070396963,
-      "loss": 46.0,
-      "step": 18103
-    },
-    {
-      "epoch": 2.9154957929063166,
-      "grad_norm": 0.0024717652704566717,
-      "learning_rate": 0.00019999581024087533,
-      "loss": 46.0,
-      "step": 18104
-    },
-    {
-      "epoch": 2.915656829985104,
-      "grad_norm": 0.0033058105036616325,
-      "learning_rate": 0.00019999580977775544,
-      "loss": 46.0,
-      "step": 18105
-    },
-    {
-      "epoch": 2.9158178670638915,
-      "grad_norm": 0.002527159173041582,
-      "learning_rate": 0.00019999580931460997,
-      "loss": 46.0,
-      "step": 18106
-    },
-    {
-      "epoch": 2.915978904142679,
-      "grad_norm": 0.006367997732013464,
-      "learning_rate": 0.00019999580885143888,
-      "loss": 46.0,
-      "step": 18107
-    },
-    {
-      "epoch": 2.9161399412214664,
-      "grad_norm": 0.00848577730357647,
-      "learning_rate": 0.0001999958083882422,
-      "loss": 46.0,
-      "step": 18108
-    },
-    {
-      "epoch": 2.916300978300254,
-      "grad_norm": 0.0037274009082466364,
-      "learning_rate": 0.00019999580792501992,
-      "loss": 46.0,
-      "step": 18109
-    },
-    {
-      "epoch": 2.9164620153790413,
-      "grad_norm": 0.003142056055366993,
-      "learning_rate": 0.0001999958074617721,
-      "loss": 46.0,
-      "step": 18110
-    },
-    {
-      "epoch": 2.9166230524578283,
-      "grad_norm": 0.0006899714935570955,
-      "learning_rate": 0.00019999580699849864,
-      "loss": 46.0,
-      "step": 18111
-    },
-    {
-      "epoch": 2.9167840895366157,
-      "grad_norm": 0.0008837699424475431,
-      "learning_rate": 0.0001999958065351996,
-      "loss": 46.0,
-      "step": 18112
-    },
-    {
-      "epoch": 2.916945126615403,
-      "grad_norm": 0.009626165963709354,
-      "learning_rate": 0.00019999580607187495,
-      "loss": 46.0,
-      "step": 18113
-    },
-    {
-      "epoch": 2.9171061636941906,
-      "grad_norm": 0.014297491870820522,
-      "learning_rate": 0.0001999958056085247,
-      "loss": 46.0,
-      "step": 18114
-    },
-    {
-      "epoch": 2.917267200772978,
-      "grad_norm": 0.0028548925183713436,
-      "learning_rate": 0.0001999958051451489,
-      "loss": 46.0,
-      "step": 18115
-    },
-    {
-      "epoch": 2.9174282378517655,
-      "grad_norm": 0.006810382008552551,
-      "learning_rate": 0.0001999958046817475,
-      "loss": 46.0,
-      "step": 18116
-    },
-    {
-      "epoch": 2.9175892749305525,
-      "grad_norm": 0.005385892000049353,
-      "learning_rate": 0.00019999580421832045,
-      "loss": 46.0,
-      "step": 18117
-    },
-    {
-      "epoch": 2.91775031200934,
-      "grad_norm": 0.0019443599740043283,
-      "learning_rate": 0.00019999580375486785,
-      "loss": 46.0,
-      "step": 18118
-    },
-    {
-      "epoch": 2.9179113490881274,
-      "grad_norm": 0.003477794351056218,
-      "learning_rate": 0.00019999580329138966,
-      "loss": 46.0,
-      "step": 18119
-    },
-    {
-      "epoch": 2.918072386166915,
-      "grad_norm": 0.0009519299492239952,
-      "learning_rate": 0.00019999580282788583,
-      "loss": 46.0,
-      "step": 18120
-    },
-    {
-      "epoch": 2.9182334232457023,
-      "grad_norm": 0.005061762407422066,
-      "learning_rate": 0.00019999580236435645,
-      "loss": 46.0,
-      "step": 18121
-    },
-    {
-      "epoch": 2.91839446032449,
-      "grad_norm": 0.001759837963618338,
-      "learning_rate": 0.00019999580190080147,
-      "loss": 46.0,
-      "step": 18122
-    },
-    {
-      "epoch": 2.9185554974032772,
-      "grad_norm": 0.003119898959994316,
-      "learning_rate": 0.00019999580143722088,
-      "loss": 46.0,
-      "step": 18123
-    },
-    {
-      "epoch": 2.9187165344820647,
-      "grad_norm": 0.0008854255429469049,
-      "learning_rate": 0.0001999958009736147,
-      "loss": 46.0,
-      "step": 18124
-    },
-    {
-      "epoch": 2.918877571560852,
-      "grad_norm": 0.0008090348565019667,
-      "learning_rate": 0.00019999580050998294,
-      "loss": 46.0,
-      "step": 18125
-    },
-    {
-      "epoch": 2.919038608639639,
-      "grad_norm": 0.001469541690312326,
-      "learning_rate": 0.0001999958000463256,
-      "loss": 46.0,
-      "step": 18126
-    },
-    {
-      "epoch": 2.9191996457184266,
-      "grad_norm": 0.0010692086070775986,
-      "learning_rate": 0.00019999579958264265,
-      "loss": 46.0,
-      "step": 18127
-    },
-    {
-      "epoch": 2.919360682797214,
-      "grad_norm": 0.0030428345780819654,
-      "learning_rate": 0.0001999957991189341,
-      "loss": 46.0,
-      "step": 18128
-    },
-    {
-      "epoch": 2.9195217198760015,
-      "grad_norm": 0.0013786237686872482,
-      "learning_rate": 0.00019999579865519996,
-      "loss": 46.0,
-      "step": 18129
-    },
-    {
-      "epoch": 2.919682756954789,
-      "grad_norm": 0.00037617204361595213,
-      "learning_rate": 0.0001999957981914402,
-      "loss": 46.0,
-      "step": 18130
-    },
-    {
-      "epoch": 2.919843794033576,
-      "grad_norm": 0.0031854051630944014,
-      "learning_rate": 0.0001999957977276549,
-      "loss": 46.0,
-      "step": 18131
-    },
-    {
-      "epoch": 2.9200048311123634,
-      "grad_norm": 0.005881123710423708,
-      "learning_rate": 0.00019999579726384396,
-      "loss": 46.0,
-      "step": 18132
-    },
-    {
-      "epoch": 2.920165868191151,
-      "grad_norm": 0.010901257395744324,
-      "learning_rate": 0.00019999579680000747,
-      "loss": 46.0,
-      "step": 18133
-    },
-    {
-      "epoch": 2.9203269052699383,
-      "grad_norm": 0.0049719843082129955,
-      "learning_rate": 0.00019999579633614534,
-      "loss": 46.0,
-      "step": 18134
-    },
-    {
-      "epoch": 2.9204879423487258,
-      "grad_norm": 0.0030031991191208363,
-      "learning_rate": 0.00019999579587225762,
-      "loss": 46.0,
-      "step": 18135
-    },
-    {
-      "epoch": 2.9206489794275132,
-      "grad_norm": 0.0007713107625022531,
-      "learning_rate": 0.00019999579540834434,
-      "loss": 46.0,
-      "step": 18136
-    },
-    {
-      "epoch": 2.9208100165063007,
-      "grad_norm": 0.0035574648063629866,
-      "learning_rate": 0.00019999579494440543,
-      "loss": 46.0,
-      "step": 18137
-    },
-    {
-      "epoch": 2.920971053585088,
-      "grad_norm": 0.0011207188945263624,
-      "learning_rate": 0.00019999579448044097,
-      "loss": 46.0,
-      "step": 18138
-    },
-    {
-      "epoch": 2.9211320906638756,
-      "grad_norm": 0.004560600034892559,
-      "learning_rate": 0.00019999579401645088,
-      "loss": 46.0,
-      "step": 18139
-    },
-    {
-      "epoch": 2.9212931277426626,
-      "grad_norm": 0.001656700624153018,
-      "learning_rate": 0.0001999957935524352,
-      "loss": 46.0,
-      "step": 18140
-    },
-    {
-      "epoch": 2.92145416482145,
-      "grad_norm": 0.0032139443792402744,
-      "learning_rate": 0.00019999579308839396,
-      "loss": 46.0,
-      "step": 18141
-    },
-    {
-      "epoch": 2.9216152019002375,
-      "grad_norm": 0.001980012282729149,
-      "learning_rate": 0.00019999579262432708,
-      "loss": 46.0,
-      "step": 18142
-    },
-    {
-      "epoch": 2.921776238979025,
-      "grad_norm": 0.014282112009823322,
-      "learning_rate": 0.00019999579216023464,
-      "loss": 46.0,
-      "step": 18143
-    },
-    {
-      "epoch": 2.9219372760578124,
-      "grad_norm": 0.00427097175270319,
-      "learning_rate": 0.0001999957916961166,
-      "loss": 46.0,
-      "step": 18144
-    },
-    {
-      "epoch": 2.9220983131366,
-      "grad_norm": 0.006991812027990818,
-      "learning_rate": 0.00019999579123197294,
-      "loss": 46.0,
-      "step": 18145
-    },
-    {
-      "epoch": 2.922259350215387,
-      "grad_norm": 0.004101167898625135,
-      "learning_rate": 0.0001999957907678037,
-      "loss": 46.0,
-      "step": 18146
-    },
-    {
-      "epoch": 2.9224203872941743,
-      "grad_norm": 0.005737594328820705,
-      "learning_rate": 0.0001999957903036089,
-      "loss": 46.0,
-      "step": 18147
-    },
-    {
-      "epoch": 2.9225814243729618,
-      "grad_norm": 0.002010551281273365,
-      "learning_rate": 0.00019999578983938846,
-      "loss": 46.0,
-      "step": 18148
-    },
-    {
-      "epoch": 2.922742461451749,
-      "grad_norm": 0.0009590897825546563,
-      "learning_rate": 0.00019999578937514244,
-      "loss": 46.0,
-      "step": 18149
-    },
-    {
-      "epoch": 2.9229034985305367,
-      "grad_norm": 0.004255289677530527,
-      "learning_rate": 0.00019999578891087083,
-      "loss": 46.0,
-      "step": 18150
-    },
-    {
-      "epoch": 2.923064535609324,
-      "grad_norm": 0.005566461477428675,
-      "learning_rate": 0.00019999578844657364,
-      "loss": 46.0,
-      "step": 18151
-    },
-    {
-      "epoch": 2.9232255726881116,
-      "grad_norm": 0.00228495872579515,
-      "learning_rate": 0.00019999578798225083,
-      "loss": 46.0,
-      "step": 18152
-    },
-    {
-      "epoch": 2.923386609766899,
-      "grad_norm": 0.0022519587073475122,
-      "learning_rate": 0.0001999957875179024,
-      "loss": 46.0,
-      "step": 18153
-    },
-    {
-      "epoch": 2.9235476468456865,
-      "grad_norm": 0.0016503919614478946,
-      "learning_rate": 0.00019999578705352846,
-      "loss": 46.0,
-      "step": 18154
-    },
-    {
-      "epoch": 2.9237086839244735,
-      "grad_norm": 0.00110530573874712,
-      "learning_rate": 0.00019999578658912889,
-      "loss": 46.0,
-      "step": 18155
-    },
-    {
-      "epoch": 2.923869721003261,
-      "grad_norm": 0.002533811377361417,
-      "learning_rate": 0.0001999957861247037,
-      "loss": 46.0,
-      "step": 18156
-    },
-    {
-      "epoch": 2.9240307580820484,
-      "grad_norm": 0.0009799996623769403,
-      "learning_rate": 0.00019999578566025293,
-      "loss": 46.0,
-      "step": 18157
-    },
-    {
-      "epoch": 2.924191795160836,
-      "grad_norm": 0.001082068309187889,
-      "learning_rate": 0.00019999578519577657,
-      "loss": 46.0,
-      "step": 18158
-    },
-    {
-      "epoch": 2.9243528322396233,
-      "grad_norm": 0.005331976804882288,
-      "learning_rate": 0.0001999957847312746,
-      "loss": 46.0,
-      "step": 18159
-    },
-    {
-      "epoch": 2.9245138693184107,
-      "grad_norm": 0.002563070273026824,
-      "learning_rate": 0.00019999578426674707,
-      "loss": 46.0,
-      "step": 18160
-    },
-    {
-      "epoch": 2.9246749063971977,
-      "grad_norm": 0.0009754052152857184,
-      "learning_rate": 0.00019999578380219392,
-      "loss": 46.0,
-      "step": 18161
-    },
-    {
-      "epoch": 2.924835943475985,
-      "grad_norm": 0.002539311069995165,
-      "learning_rate": 0.0001999957833376152,
-      "loss": 46.0,
-      "step": 18162
-    },
-    {
-      "epoch": 2.9249969805547726,
-      "grad_norm": 0.0021239614579826593,
-      "learning_rate": 0.00019999578287301087,
-      "loss": 46.0,
-      "step": 18163
-    },
-    {
-      "epoch": 2.92515801763356,
-      "grad_norm": 0.0005640267627313733,
-      "learning_rate": 0.00019999578240838096,
-      "loss": 46.0,
-      "step": 18164
-    },
-    {
-      "epoch": 2.9253190547123475,
-      "grad_norm": 0.0017377551412209868,
-      "learning_rate": 0.00019999578194372544,
-      "loss": 46.0,
-      "step": 18165
-    },
-    {
-      "epoch": 2.925480091791135,
-      "grad_norm": 0.0019512108992785215,
-      "learning_rate": 0.0001999957814790443,
-      "loss": 46.0,
-      "step": 18166
-    },
-    {
-      "epoch": 2.9256411288699224,
-      "grad_norm": 0.00739320321008563,
-      "learning_rate": 0.0001999957810143376,
-      "loss": 46.0,
-      "step": 18167
-    },
-    {
-      "epoch": 2.92580216594871,
-      "grad_norm": 0.0009076702990569174,
-      "learning_rate": 0.0001999957805496053,
-      "loss": 46.0,
-      "step": 18168
-    },
-    {
-      "epoch": 2.9259632030274974,
-      "grad_norm": 0.002244069706648588,
-      "learning_rate": 0.0001999957800848474,
-      "loss": 46.0,
-      "step": 18169
-    },
-    {
-      "epoch": 2.9261242401062844,
-      "grad_norm": 0.0006175077287480235,
-      "learning_rate": 0.00019999577962006394,
-      "loss": 46.0,
-      "step": 18170
-    },
-    {
-      "epoch": 2.926285277185072,
-      "grad_norm": 0.0016929727280512452,
-      "learning_rate": 0.00019999577915525484,
-      "loss": 46.0,
-      "step": 18171
-    },
-    {
-      "epoch": 2.9264463142638593,
-      "grad_norm": 0.00583892734721303,
-      "learning_rate": 0.00019999577869042015,
-      "loss": 46.0,
-      "step": 18172
-    },
-    {
-      "epoch": 2.9266073513426467,
-      "grad_norm": 0.005278992000967264,
-      "learning_rate": 0.0001999957782255599,
-      "loss": 46.0,
-      "step": 18173
-    },
-    {
-      "epoch": 2.926768388421434,
-      "grad_norm": 0.00039456624654121697,
-      "learning_rate": 0.00019999577776067405,
-      "loss": 46.0,
-      "step": 18174
-    },
-    {
-      "epoch": 2.926929425500221,
-      "grad_norm": 0.0060783615335822105,
-      "learning_rate": 0.00019999577729576257,
-      "loss": 46.0,
-      "step": 18175
-    },
-    {
-      "epoch": 2.9270904625790086,
-      "grad_norm": 0.002739834599196911,
-      "learning_rate": 0.00019999577683082556,
-      "loss": 46.0,
-      "step": 18176
-    },
-    {
-      "epoch": 2.927251499657796,
-      "grad_norm": 0.0010865024523809552,
-      "learning_rate": 0.0001999957763658629,
-      "loss": 46.0,
-      "step": 18177
-    },
-    {
-      "epoch": 2.9274125367365835,
-      "grad_norm": 0.012886730954051018,
-      "learning_rate": 0.00019999577590087465,
-      "loss": 46.0,
-      "step": 18178
-    },
-    {
-      "epoch": 2.927573573815371,
-      "grad_norm": 0.0036216166336089373,
-      "learning_rate": 0.00019999577543586083,
-      "loss": 46.0,
-      "step": 18179
-    },
-    {
-      "epoch": 2.9277346108941584,
-      "grad_norm": 0.0008295559673570096,
-      "learning_rate": 0.00019999577497082141,
-      "loss": 46.0,
-      "step": 18180
-    },
-    {
-      "epoch": 2.927895647972946,
-      "grad_norm": 0.0019190424354746938,
-      "learning_rate": 0.0001999957745057564,
-      "loss": 46.0,
-      "step": 18181
-    },
-    {
-      "epoch": 2.9280566850517333,
-      "grad_norm": 0.00409006979316473,
-      "learning_rate": 0.00019999577404066578,
-      "loss": 46.0,
-      "step": 18182
-    },
-    {
-      "epoch": 2.928217722130521,
-      "grad_norm": 0.0017831475706771016,
-      "learning_rate": 0.00019999577357554958,
-      "loss": 46.0,
-      "step": 18183
-    },
-    {
-      "epoch": 2.928378759209308,
-      "grad_norm": 0.0017056305659934878,
-      "learning_rate": 0.00019999577311040776,
-      "loss": 46.0,
-      "step": 18184
-    },
-    {
-      "epoch": 2.9285397962880952,
-      "grad_norm": 0.001114071230404079,
-      "learning_rate": 0.0001999957726452404,
-      "loss": 46.0,
-      "step": 18185
-    },
-    {
-      "epoch": 2.9287008333668827,
-      "grad_norm": 0.002630940405651927,
-      "learning_rate": 0.0001999957721800474,
-      "loss": 46.0,
-      "step": 18186
-    },
-    {
-      "epoch": 2.92886187044567,
-      "grad_norm": 0.0017845238326117396,
-      "learning_rate": 0.00019999577171482883,
-      "loss": 46.0,
-      "step": 18187
-    },
-    {
-      "epoch": 2.9290229075244576,
-      "grad_norm": 0.0072834608145058155,
-      "learning_rate": 0.00019999577124958464,
-      "loss": 46.0,
-      "step": 18188
-    },
-    {
-      "epoch": 2.929183944603245,
-      "grad_norm": 0.0021217649336904287,
-      "learning_rate": 0.00019999577078431489,
-      "loss": 46.0,
-      "step": 18189
-    },
-    {
-      "epoch": 2.929344981682032,
-      "grad_norm": 0.0017602955922484398,
-      "learning_rate": 0.0001999957703190195,
-      "loss": 46.0,
-      "step": 18190
-    },
-    {
-      "epoch": 2.9295060187608195,
-      "grad_norm": 0.008133293129503727,
-      "learning_rate": 0.00019999576985369854,
-      "loss": 46.0,
-      "step": 18191
-    },
-    {
-      "epoch": 2.929667055839607,
-      "grad_norm": 0.0015090502565726638,
-      "learning_rate": 0.000199995769388352,
-      "loss": 46.0,
-      "step": 18192
-    },
-    {
-      "epoch": 2.9298280929183944,
-      "grad_norm": 0.006169312167912722,
-      "learning_rate": 0.00019999576892297985,
-      "loss": 46.0,
-      "step": 18193
-    },
-    {
-      "epoch": 2.929989129997182,
-      "grad_norm": 0.0006887398776598275,
-      "learning_rate": 0.0001999957684575821,
-      "loss": 46.0,
-      "step": 18194
-    },
-    {
-      "epoch": 2.9301501670759693,
-      "grad_norm": 0.007141907699406147,
-      "learning_rate": 0.00019999576799215879,
-      "loss": 46.0,
-      "step": 18195
-    },
-    {
-      "epoch": 2.9303112041547568,
-      "grad_norm": 0.002672359114512801,
-      "learning_rate": 0.00019999576752670984,
-      "loss": 46.0,
-      "step": 18196
-    },
-    {
-      "epoch": 2.930472241233544,
-      "grad_norm": 0.004060429520905018,
-      "learning_rate": 0.00019999576706123534,
-      "loss": 46.0,
-      "step": 18197
-    },
-    {
-      "epoch": 2.9306332783123317,
-      "grad_norm": 0.0029106049332767725,
-      "learning_rate": 0.00019999576659573523,
-      "loss": 46.0,
-      "step": 18198
-    },
-    {
-      "epoch": 2.9307943153911187,
-      "grad_norm": 0.0019907609093934298,
-      "learning_rate": 0.00019999576613020952,
-      "loss": 46.0,
-      "step": 18199
-    },
-    {
-      "epoch": 2.930955352469906,
-      "grad_norm": 0.0011841485975310206,
-      "learning_rate": 0.00019999576566465823,
-      "loss": 46.0,
-      "step": 18200
-    },
-    {
-      "epoch": 2.9311163895486936,
-      "grad_norm": 0.0009769797325134277,
-      "learning_rate": 0.00019999576519908133,
-      "loss": 46.0,
-      "step": 18201
-    },
-    {
-      "epoch": 2.931277426627481,
-      "grad_norm": 0.00409746216610074,
-      "learning_rate": 0.00019999576473347887,
-      "loss": 46.0,
-      "step": 18202
-    },
-    {
-      "epoch": 2.9314384637062685,
-      "grad_norm": 0.004400520119816065,
-      "learning_rate": 0.00019999576426785076,
-      "loss": 46.0,
-      "step": 18203
-    },
-    {
-      "epoch": 2.9315995007850555,
-      "grad_norm": 0.0016638454981148243,
-      "learning_rate": 0.00019999576380219712,
-      "loss": 46.0,
-      "step": 18204
-    },
-    {
-      "epoch": 2.931760537863843,
-      "grad_norm": 0.0017848439747467637,
-      "learning_rate": 0.00019999576333651784,
-      "loss": 46.0,
-      "step": 18205
-    },
-    {
-      "epoch": 2.9319215749426304,
-      "grad_norm": 0.001913099898956716,
-      "learning_rate": 0.00019999576287081297,
-      "loss": 46.0,
-      "step": 18206
-    },
-    {
-      "epoch": 2.932082612021418,
-      "grad_norm": 0.001097127329558134,
-      "learning_rate": 0.00019999576240508252,
-      "loss": 46.0,
-      "step": 18207
-    },
-    {
-      "epoch": 2.9322436491002053,
-      "grad_norm": 0.0012283776886761189,
-      "learning_rate": 0.00019999576193932645,
-      "loss": 46.0,
-      "step": 18208
-    },
-    {
-      "epoch": 2.9324046861789927,
-      "grad_norm": 0.0013275261735543609,
-      "learning_rate": 0.00019999576147354482,
-      "loss": 46.0,
-      "step": 18209
-    },
-    {
-      "epoch": 2.93256572325778,
-      "grad_norm": 0.0025266394950449467,
-      "learning_rate": 0.00019999576100773758,
-      "loss": 46.0,
-      "step": 18210
-    },
-    {
-      "epoch": 2.9327267603365677,
-      "grad_norm": 0.0018194651929661632,
-      "learning_rate": 0.00019999576054190475,
-      "loss": 46.0,
-      "step": 18211
-    },
-    {
-      "epoch": 2.932887797415355,
-      "grad_norm": 0.00701428996399045,
-      "learning_rate": 0.00019999576007604633,
-      "loss": 46.0,
-      "step": 18212
-    },
-    {
-      "epoch": 2.9330488344941426,
-      "grad_norm": 0.0025370344519615173,
-      "learning_rate": 0.0001999957596101623,
-      "loss": 46.0,
-      "step": 18213
-    },
-    {
-      "epoch": 2.9332098715729296,
-      "grad_norm": 0.004439545329660177,
-      "learning_rate": 0.00019999575914425268,
-      "loss": 46.0,
-      "step": 18214
-    },
-    {
-      "epoch": 2.933370908651717,
-      "grad_norm": 0.006848150864243507,
-      "learning_rate": 0.00019999575867831747,
-      "loss": 46.0,
-      "step": 18215
-    },
-    {
-      "epoch": 2.9335319457305045,
-      "grad_norm": 0.0038663074374198914,
-      "learning_rate": 0.0001999957582123567,
-      "loss": 46.0,
-      "step": 18216
-    },
-    {
-      "epoch": 2.933692982809292,
-      "grad_norm": 0.002190664177760482,
-      "learning_rate": 0.0001999957577463703,
-      "loss": 46.0,
-      "step": 18217
-    },
-    {
-      "epoch": 2.9338540198880794,
-      "grad_norm": 0.0013622419210150838,
-      "learning_rate": 0.0001999957572803583,
-      "loss": 46.0,
-      "step": 18218
-    },
-    {
-      "epoch": 2.9340150569668664,
-      "grad_norm": 0.0006758912932127714,
-      "learning_rate": 0.00019999575681432072,
-      "loss": 46.0,
-      "step": 18219
-    },
-    {
-      "epoch": 2.934176094045654,
-      "grad_norm": 0.0020648129284381866,
-      "learning_rate": 0.00019999575634825756,
-      "loss": 46.0,
-      "step": 18220
-    },
-    {
-      "epoch": 2.9343371311244413,
-      "grad_norm": 0.0015850800555199385,
-      "learning_rate": 0.0001999957558821688,
-      "loss": 46.0,
-      "step": 18221
-    },
-    {
-      "epoch": 2.9344981682032287,
-      "grad_norm": 0.006224527955055237,
-      "learning_rate": 0.0001999957554160544,
-      "loss": 46.0,
-      "step": 18222
-    },
-    {
-      "epoch": 2.934659205282016,
-      "grad_norm": 0.003197029232978821,
-      "learning_rate": 0.00019999575494991444,
-      "loss": 46.0,
-      "step": 18223
-    },
-    {
-      "epoch": 2.9348202423608036,
-      "grad_norm": 0.0011069177417084575,
-      "learning_rate": 0.0001999957544837489,
-      "loss": 46.0,
-      "step": 18224
-    },
-    {
-      "epoch": 2.934981279439591,
-      "grad_norm": 0.004962557926774025,
-      "learning_rate": 0.00019999575401755774,
-      "loss": 46.0,
-      "step": 18225
-    },
-    {
-      "epoch": 2.9351423165183785,
-      "grad_norm": 0.0017906212015077472,
-      "learning_rate": 0.00019999575355134105,
-      "loss": 46.0,
-      "step": 18226
-    },
-    {
-      "epoch": 2.935303353597166,
-      "grad_norm": 0.005940783303231001,
-      "learning_rate": 0.00019999575308509872,
-      "loss": 46.0,
-      "step": 18227
-    },
-    {
-      "epoch": 2.935464390675953,
-      "grad_norm": 0.0029781844932585955,
-      "learning_rate": 0.00019999575261883077,
-      "loss": 46.0,
-      "step": 18228
-    },
-    {
-      "epoch": 2.9356254277547404,
-      "grad_norm": 0.0019347805064171553,
-      "learning_rate": 0.00019999575215253726,
-      "loss": 46.0,
-      "step": 18229
-    },
-    {
-      "epoch": 2.935786464833528,
-      "grad_norm": 0.0028998963534832,
-      "learning_rate": 0.00019999575168621814,
-      "loss": 46.0,
-      "step": 18230
-    },
-    {
-      "epoch": 2.9359475019123153,
-      "grad_norm": 0.0012944411719217896,
-      "learning_rate": 0.00019999575121987343,
-      "loss": 46.0,
-      "step": 18231
-    },
-    {
-      "epoch": 2.936108538991103,
-      "grad_norm": 0.0021356791257858276,
-      "learning_rate": 0.00019999575075350313,
-      "loss": 46.0,
-      "step": 18232
-    },
-    {
-      "epoch": 2.9362695760698903,
-      "grad_norm": 0.001533346250653267,
-      "learning_rate": 0.00019999575028710722,
-      "loss": 46.0,
-      "step": 18233
-    },
-    {
-      "epoch": 2.9364306131486773,
-      "grad_norm": 0.0021803805138915777,
-      "learning_rate": 0.00019999574982068575,
-      "loss": 46.0,
-      "step": 18234
-    },
-    {
-      "epoch": 2.9365916502274647,
-      "grad_norm": 0.004516003653407097,
-      "learning_rate": 0.00019999574935423867,
-      "loss": 46.0,
-      "step": 18235
-    },
-    {
-      "epoch": 2.936752687306252,
-      "grad_norm": 0.0015579540049657226,
-      "learning_rate": 0.00019999574888776597,
-      "loss": 46.0,
-      "step": 18236
-    },
-    {
-      "epoch": 2.9369137243850396,
-      "grad_norm": 0.005141932051628828,
-      "learning_rate": 0.0001999957484212677,
-      "loss": 46.0,
-      "step": 18237
-    },
-    {
-      "epoch": 2.937074761463827,
-      "grad_norm": 0.0009286055574193597,
-      "learning_rate": 0.00019999574795474386,
-      "loss": 46.0,
-      "step": 18238
-    },
-    {
-      "epoch": 2.9372357985426145,
-      "grad_norm": 0.00913744792342186,
-      "learning_rate": 0.00019999574748819437,
-      "loss": 46.0,
-      "step": 18239
-    },
-    {
-      "epoch": 2.937396835621402,
-      "grad_norm": 0.0016055910382419825,
-      "learning_rate": 0.00019999574702161935,
-      "loss": 46.0,
-      "step": 18240
-    },
-    {
-      "epoch": 2.9375578727001894,
-      "grad_norm": 0.004685543943196535,
-      "learning_rate": 0.0001999957465550187,
-      "loss": 46.0,
-      "step": 18241
-    },
-    {
-      "epoch": 2.937718909778977,
-      "grad_norm": 0.004667813424021006,
-      "learning_rate": 0.00019999574608839247,
-      "loss": 46.0,
-      "step": 18242
-    },
-    {
-      "epoch": 2.937879946857764,
-      "grad_norm": 0.0025856574065983295,
-      "learning_rate": 0.00019999574562174063,
-      "loss": 46.0,
-      "step": 18243
-    },
-    {
-      "epoch": 2.9380409839365513,
-      "grad_norm": 0.005429683718830347,
-      "learning_rate": 0.00019999574515506318,
-      "loss": 46.0,
-      "step": 18244
-    },
-    {
-      "epoch": 2.938202021015339,
-      "grad_norm": 0.004247988574206829,
-      "learning_rate": 0.00019999574468836017,
-      "loss": 46.0,
-      "step": 18245
-    },
-    {
-      "epoch": 2.9383630580941262,
-      "grad_norm": 0.0015145476208999753,
-      "learning_rate": 0.00019999574422163157,
-      "loss": 46.0,
-      "step": 18246
-    },
-    {
-      "epoch": 2.9385240951729137,
-      "grad_norm": 0.004039260093122721,
-      "learning_rate": 0.00019999574375487736,
-      "loss": 46.0,
-      "step": 18247
-    },
-    {
-      "epoch": 2.9386851322517007,
-      "grad_norm": 0.012637422420084476,
-      "learning_rate": 0.00019999574328809756,
-      "loss": 46.0,
-      "step": 18248
-    },
-    {
-      "epoch": 2.938846169330488,
-      "grad_norm": 0.0026153794024139643,
-      "learning_rate": 0.00019999574282129217,
-      "loss": 46.0,
-      "step": 18249
-    },
-    {
-      "epoch": 2.9390072064092756,
-      "grad_norm": 0.0010796021670103073,
-      "learning_rate": 0.00019999574235446117,
-      "loss": 46.0,
-      "step": 18250
-    },
-    {
-      "epoch": 2.939168243488063,
-      "grad_norm": 0.0024613006971776485,
-      "learning_rate": 0.00019999574188760458,
-      "loss": 46.0,
-      "step": 18251
-    },
-    {
-      "epoch": 2.9393292805668505,
-      "grad_norm": 0.003761656815186143,
-      "learning_rate": 0.0001999957414207224,
-      "loss": 46.0,
-      "step": 18252
-    },
-    {
-      "epoch": 2.939490317645638,
-      "grad_norm": 0.0019356246339157224,
-      "learning_rate": 0.00019999574095381465,
-      "loss": 46.0,
-      "step": 18253
-    },
-    {
-      "epoch": 2.9396513547244254,
-      "grad_norm": 0.0005578165291808546,
-      "learning_rate": 0.00019999574048688127,
-      "loss": 46.0,
-      "step": 18254
-    },
-    {
-      "epoch": 2.939812391803213,
-      "grad_norm": 0.0009123950148932636,
-      "learning_rate": 0.0001999957400199223,
-      "loss": 46.0,
-      "step": 18255
-    },
-    {
-      "epoch": 2.9399734288820003,
-      "grad_norm": 0.0008915271027944982,
-      "learning_rate": 0.00019999573955293775,
-      "loss": 46.0,
-      "step": 18256
-    },
-    {
-      "epoch": 2.9401344659607873,
-      "grad_norm": 0.0008778031333349645,
-      "learning_rate": 0.00019999573908592764,
-      "loss": 46.0,
-      "step": 18257
-    },
-    {
-      "epoch": 2.9402955030395748,
-      "grad_norm": 0.004145601764321327,
-      "learning_rate": 0.0001999957386188919,
-      "loss": 46.0,
-      "step": 18258
-    },
-    {
-      "epoch": 2.940456540118362,
-      "grad_norm": 0.003352200146764517,
-      "learning_rate": 0.00019999573815183055,
-      "loss": 46.0,
-      "step": 18259
-    },
-    {
-      "epoch": 2.9406175771971497,
-      "grad_norm": 0.004106913693249226,
-      "learning_rate": 0.00019999573768474362,
-      "loss": 46.0,
-      "step": 18260
-    },
-    {
-      "epoch": 2.940778614275937,
-      "grad_norm": 0.00277924002148211,
-      "learning_rate": 0.00019999573721763108,
-      "loss": 46.0,
-      "step": 18261
-    },
-    {
-      "epoch": 2.9409396513547246,
-      "grad_norm": 0.003632959211245179,
-      "learning_rate": 0.00019999573675049298,
-      "loss": 46.0,
-      "step": 18262
-    },
-    {
-      "epoch": 2.9411006884335116,
-      "grad_norm": 0.011196434497833252,
-      "learning_rate": 0.0001999957362833293,
-      "loss": 46.0,
-      "step": 18263
-    },
-    {
-      "epoch": 2.941261725512299,
-      "grad_norm": 0.005271113943308592,
-      "learning_rate": 0.00019999573581614,
-      "loss": 46.0,
-      "step": 18264
-    },
-    {
-      "epoch": 2.9414227625910865,
-      "grad_norm": 0.002945388900116086,
-      "learning_rate": 0.00019999573534892507,
-      "loss": 46.0,
-      "step": 18265
-    },
-    {
-      "epoch": 2.941583799669874,
-      "grad_norm": 0.003602365031838417,
-      "learning_rate": 0.0001999957348816846,
-      "loss": 46.0,
-      "step": 18266
-    },
-    {
-      "epoch": 2.9417448367486614,
-      "grad_norm": 0.0059127239510416985,
-      "learning_rate": 0.0001999957344144185,
-      "loss": 46.0,
-      "step": 18267
-    },
-    {
-      "epoch": 2.941905873827449,
-      "grad_norm": 0.0019192382460460067,
-      "learning_rate": 0.00019999573394712683,
-      "loss": 46.0,
-      "step": 18268
-    },
-    {
-      "epoch": 2.9420669109062363,
-      "grad_norm": 0.010361652821302414,
-      "learning_rate": 0.0001999957334798096,
-      "loss": 46.0,
-      "step": 18269
-    },
-    {
-      "epoch": 2.9422279479850237,
-      "grad_norm": 0.0008517851820215583,
-      "learning_rate": 0.0001999957330124667,
-      "loss": 46.0,
-      "step": 18270
-    },
-    {
-      "epoch": 2.942388985063811,
-      "grad_norm": 0.003357992274686694,
-      "learning_rate": 0.00019999573254509824,
-      "loss": 46.0,
-      "step": 18271
-    },
-    {
-      "epoch": 2.942550022142598,
-      "grad_norm": 0.003572401124984026,
-      "learning_rate": 0.0001999957320777042,
-      "loss": 46.0,
-      "step": 18272
-    },
-    {
-      "epoch": 2.9427110592213856,
-      "grad_norm": 0.0007625145954079926,
-      "learning_rate": 0.00019999573161028455,
-      "loss": 46.0,
-      "step": 18273
-    },
-    {
-      "epoch": 2.942872096300173,
-      "grad_norm": 0.0054258918389678,
-      "learning_rate": 0.00019999573114283932,
-      "loss": 46.0,
-      "step": 18274
-    },
-    {
-      "epoch": 2.9430331333789606,
-      "grad_norm": 0.0019069387344643474,
-      "learning_rate": 0.00019999573067536848,
-      "loss": 46.0,
-      "step": 18275
-    },
-    {
-      "epoch": 2.943194170457748,
-      "grad_norm": 0.0009239456849172711,
-      "learning_rate": 0.00019999573020787205,
-      "loss": 46.0,
-      "step": 18276
-    },
-    {
-      "epoch": 2.943355207536535,
-      "grad_norm": 0.0055352505296468735,
-      "learning_rate": 0.00019999572974035003,
-      "loss": 46.0,
-      "step": 18277
-    },
-    {
-      "epoch": 2.9435162446153225,
-      "grad_norm": 0.004443612415343523,
-      "learning_rate": 0.00019999572927280242,
-      "loss": 46.0,
-      "step": 18278
-    },
-    {
-      "epoch": 2.94367728169411,
-      "grad_norm": 0.0023634915705770254,
-      "learning_rate": 0.0001999957288052292,
-      "loss": 46.0,
-      "step": 18279
-    },
-    {
-      "epoch": 2.9438383187728974,
-      "grad_norm": 0.0011116719106212258,
-      "learning_rate": 0.0001999957283376304,
-      "loss": 46.0,
-      "step": 18280
-    },
-    {
-      "epoch": 2.943999355851685,
-      "grad_norm": 0.00572171388193965,
-      "learning_rate": 0.000199995727870006,
-      "loss": 46.0,
-      "step": 18281
-    },
-    {
-      "epoch": 2.9441603929304723,
-      "grad_norm": 0.005229401867836714,
-      "learning_rate": 0.00019999572740235603,
-      "loss": 46.0,
-      "step": 18282
-    },
-    {
-      "epoch": 2.9443214300092597,
-      "grad_norm": 0.001199902268126607,
-      "learning_rate": 0.00019999572693468043,
-      "loss": 46.0,
-      "step": 18283
-    },
-    {
-      "epoch": 2.944482467088047,
-      "grad_norm": 0.004327108152210712,
-      "learning_rate": 0.00019999572646697925,
-      "loss": 46.0,
-      "step": 18284
-    },
-    {
-      "epoch": 2.9446435041668346,
-      "grad_norm": 0.0029401604551821947,
-      "learning_rate": 0.00019999572599925248,
-      "loss": 46.0,
-      "step": 18285
-    },
-    {
-      "epoch": 2.944804541245622,
-      "grad_norm": 0.0028358884155750275,
-      "learning_rate": 0.00019999572553150013,
-      "loss": 46.0,
-      "step": 18286
-    },
-    {
-      "epoch": 2.944965578324409,
-      "grad_norm": 0.0019006096990779042,
-      "learning_rate": 0.00019999572506372216,
-      "loss": 46.0,
-      "step": 18287
-    },
-    {
-      "epoch": 2.9451266154031965,
-      "grad_norm": 0.001222077989950776,
-      "learning_rate": 0.00019999572459591862,
-      "loss": 46.0,
-      "step": 18288
-    },
-    {
-      "epoch": 2.945287652481984,
-      "grad_norm": 0.008274531923234463,
-      "learning_rate": 0.00019999572412808948,
-      "loss": 46.0,
-      "step": 18289
-    },
-    {
-      "epoch": 2.9454486895607714,
-      "grad_norm": 0.0011886481661349535,
-      "learning_rate": 0.00019999572366023475,
-      "loss": 46.0,
-      "step": 18290
-    },
-    {
-      "epoch": 2.945609726639559,
-      "grad_norm": 0.00539096724241972,
-      "learning_rate": 0.00019999572319235438,
-      "loss": 46.0,
-      "step": 18291
-    },
-    {
-      "epoch": 2.945770763718346,
-      "grad_norm": 0.0031763585284352303,
-      "learning_rate": 0.00019999572272444847,
-      "loss": 46.0,
-      "step": 18292
-    },
-    {
-      "epoch": 2.9459318007971333,
-      "grad_norm": 0.0030268875416368246,
-      "learning_rate": 0.00019999572225651692,
-      "loss": 46.0,
-      "step": 18293
-    },
-    {
-      "epoch": 2.946092837875921,
-      "grad_norm": 0.006517102010548115,
-      "learning_rate": 0.00019999572178855984,
-      "loss": 46.0,
-      "step": 18294
-    },
-    {
-      "epoch": 2.9462538749547083,
-      "grad_norm": 0.004996228031814098,
-      "learning_rate": 0.00019999572132057712,
-      "loss": 46.0,
-      "step": 18295
-    },
-    {
-      "epoch": 2.9464149120334957,
-      "grad_norm": 0.0035531383473426104,
-      "learning_rate": 0.0001999957208525688,
-      "loss": 46.0,
-      "step": 18296
-    },
-    {
-      "epoch": 2.946575949112283,
-      "grad_norm": 0.008447499945759773,
-      "learning_rate": 0.00019999572038453491,
-      "loss": 46.0,
-      "step": 18297
-    },
-    {
-      "epoch": 2.9467369861910706,
-      "grad_norm": 0.0010228337487205863,
-      "learning_rate": 0.00019999571991647543,
-      "loss": 46.0,
-      "step": 18298
-    },
-    {
-      "epoch": 2.946898023269858,
-      "grad_norm": 0.0023939877282828093,
-      "learning_rate": 0.00019999571944839036,
-      "loss": 46.0,
-      "step": 18299
-    },
-    {
-      "epoch": 2.9470590603486455,
-      "grad_norm": 0.0032466899137943983,
-      "learning_rate": 0.00019999571898027968,
-      "loss": 46.0,
-      "step": 18300
-    },
-    {
-      "epoch": 2.9472200974274325,
-      "grad_norm": 0.002225608332082629,
-      "learning_rate": 0.0001999957185121434,
-      "loss": 46.0,
-      "step": 18301
-    },
-    {
-      "epoch": 2.94738113450622,
-      "grad_norm": 0.005308377090841532,
-      "learning_rate": 0.00019999571804398154,
-      "loss": 46.0,
-      "step": 18302
-    },
-    {
-      "epoch": 2.9475421715850074,
-      "grad_norm": 0.0020556533709168434,
-      "learning_rate": 0.00019999571757579407,
-      "loss": 46.0,
-      "step": 18303
-    },
-    {
-      "epoch": 2.947703208663795,
-      "grad_norm": 0.004077361896634102,
-      "learning_rate": 0.00019999571710758104,
-      "loss": 46.0,
-      "step": 18304
-    },
-    {
-      "epoch": 2.9478642457425823,
-      "grad_norm": 0.002882294822484255,
-      "learning_rate": 0.0001999957166393424,
-      "loss": 46.0,
-      "step": 18305
-    },
-    {
-      "epoch": 2.9480252828213698,
-      "grad_norm": 0.004241040907800198,
-      "learning_rate": 0.00019999571617107813,
-      "loss": 46.0,
-      "step": 18306
-    },
-    {
-      "epoch": 2.948186319900157,
-      "grad_norm": 0.0017040788661688566,
-      "learning_rate": 0.0001999957157027883,
-      "loss": 46.0,
-      "step": 18307
-    },
-    {
-      "epoch": 2.9483473569789442,
-      "grad_norm": 0.0016345313051715493,
-      "learning_rate": 0.00019999571523447287,
-      "loss": 46.0,
-      "step": 18308
-    },
-    {
-      "epoch": 2.9485083940577317,
-      "grad_norm": 0.0007096792687661946,
-      "learning_rate": 0.00019999571476613185,
-      "loss": 46.0,
-      "step": 18309
-    },
-    {
-      "epoch": 2.948669431136519,
-      "grad_norm": 0.0031859949231147766,
-      "learning_rate": 0.00019999571429776524,
-      "loss": 46.0,
-      "step": 18310
-    },
-    {
-      "epoch": 2.9488304682153066,
-      "grad_norm": 0.0014084330759942532,
-      "learning_rate": 0.000199995713829373,
-      "loss": 46.0,
-      "step": 18311
-    },
-    {
-      "epoch": 2.948991505294094,
-      "grad_norm": 0.006162330042570829,
-      "learning_rate": 0.00019999571336095523,
-      "loss": 46.0,
-      "step": 18312
-    },
-    {
-      "epoch": 2.9491525423728815,
-      "grad_norm": 0.003467687638476491,
-      "learning_rate": 0.00019999571289251183,
-      "loss": 46.0,
-      "step": 18313
-    },
-    {
-      "epoch": 2.949313579451669,
-      "grad_norm": 0.004696844145655632,
-      "learning_rate": 0.0001999957124240428,
-      "loss": 46.0,
-      "step": 18314
-    },
-    {
-      "epoch": 2.9494746165304564,
-      "grad_norm": 0.0028436086140573025,
-      "learning_rate": 0.00019999571195554824,
-      "loss": 46.0,
-      "step": 18315
-    },
-    {
-      "epoch": 2.9496356536092434,
-      "grad_norm": 0.0010815279092639685,
-      "learning_rate": 0.00019999571148702808,
-      "loss": 46.0,
-      "step": 18316
-    },
-    {
-      "epoch": 2.949796690688031,
-      "grad_norm": 0.0009222500957548618,
-      "learning_rate": 0.0001999957110184823,
-      "loss": 46.0,
-      "step": 18317
-    },
-    {
-      "epoch": 2.9499577277668183,
-      "grad_norm": 0.0007939033093862236,
-      "learning_rate": 0.00019999571054991094,
-      "loss": 46.0,
-      "step": 18318
-    },
-    {
-      "epoch": 2.9501187648456058,
-      "grad_norm": 0.0018598157912492752,
-      "learning_rate": 0.00019999571008131396,
-      "loss": 46.0,
-      "step": 18319
-    },
-    {
-      "epoch": 2.950279801924393,
-      "grad_norm": 0.0037508851382881403,
-      "learning_rate": 0.00019999570961269143,
-      "loss": 46.0,
-      "step": 18320
-    },
-    {
-      "epoch": 2.95044083900318,
-      "grad_norm": 0.0066314502619206905,
-      "learning_rate": 0.00019999570914404328,
-      "loss": 46.0,
-      "step": 18321
-    },
-    {
-      "epoch": 2.9506018760819677,
-      "grad_norm": 0.005467196926474571,
-      "learning_rate": 0.0001999957086753695,
-      "loss": 46.0,
-      "step": 18322
-    },
-    {
-      "epoch": 2.950762913160755,
-      "grad_norm": 0.0018607608508318663,
-      "learning_rate": 0.0001999957082066702,
-      "loss": 46.0,
-      "step": 18323
-    },
-    {
-      "epoch": 2.9509239502395426,
-      "grad_norm": 0.003991422709077597,
-      "learning_rate": 0.00019999570773794527,
-      "loss": 46.0,
-      "step": 18324
-    },
-    {
-      "epoch": 2.95108498731833,
-      "grad_norm": 0.0021336579229682684,
-      "learning_rate": 0.00019999570726919475,
-      "loss": 46.0,
-      "step": 18325
-    },
-    {
-      "epoch": 2.9512460243971175,
-      "grad_norm": 0.004980088211596012,
-      "learning_rate": 0.0001999957068004186,
-      "loss": 46.0,
-      "step": 18326
-    },
-    {
-      "epoch": 2.951407061475905,
-      "grad_norm": 0.01494060829281807,
-      "learning_rate": 0.0001999957063316169,
-      "loss": 46.0,
-      "step": 18327
-    },
-    {
-      "epoch": 2.9515680985546924,
-      "grad_norm": 0.003776193130761385,
-      "learning_rate": 0.00019999570586278962,
-      "loss": 46.0,
-      "step": 18328
-    },
-    {
-      "epoch": 2.95172913563348,
-      "grad_norm": 0.0026578884571790695,
-      "learning_rate": 0.00019999570539393671,
-      "loss": 46.0,
-      "step": 18329
-    },
-    {
-      "epoch": 2.951890172712267,
-      "grad_norm": 0.003074975684285164,
-      "learning_rate": 0.00019999570492505823,
-      "loss": 46.0,
-      "step": 18330
-    },
-    {
-      "epoch": 2.9520512097910543,
-      "grad_norm": 0.0071158758364617825,
-      "learning_rate": 0.00019999570445615415,
-      "loss": 46.0,
-      "step": 18331
-    },
-    {
-      "epoch": 2.9522122468698417,
-      "grad_norm": 0.004280238412320614,
-      "learning_rate": 0.00019999570398722446,
-      "loss": 46.0,
-      "step": 18332
-    },
-    {
-      "epoch": 2.952373283948629,
-      "grad_norm": 0.0025755814276635647,
-      "learning_rate": 0.0001999957035182692,
-      "loss": 46.0,
-      "step": 18333
-    },
-    {
-      "epoch": 2.9525343210274166,
-      "grad_norm": 0.00290506100282073,
-      "learning_rate": 0.00019999570304928831,
-      "loss": 46.0,
-      "step": 18334
-    },
-    {
-      "epoch": 2.952695358106204,
-      "grad_norm": 0.006528573576360941,
-      "learning_rate": 0.00019999570258028186,
-      "loss": 46.0,
-      "step": 18335
-    },
-    {
-      "epoch": 2.952856395184991,
-      "grad_norm": 0.0032045214902609587,
-      "learning_rate": 0.00019999570211124982,
-      "loss": 46.0,
-      "step": 18336
-    },
-    {
-      "epoch": 2.9530174322637786,
-      "grad_norm": 0.001571933040395379,
-      "learning_rate": 0.00019999570164219217,
-      "loss": 46.0,
-      "step": 18337
-    },
-    {
-      "epoch": 2.953178469342566,
-      "grad_norm": 0.001303008641116321,
-      "learning_rate": 0.00019999570117310893,
-      "loss": 46.0,
-      "step": 18338
-    },
-    {
-      "epoch": 2.9533395064213535,
-      "grad_norm": 0.0017148804618045688,
-      "learning_rate": 0.0001999957007040001,
-      "loss": 46.0,
-      "step": 18339
-    },
-    {
-      "epoch": 2.953500543500141,
-      "grad_norm": 0.0008733664290048182,
-      "learning_rate": 0.00019999570023486565,
-      "loss": 46.0,
-      "step": 18340
-    },
-    {
-      "epoch": 2.9536615805789284,
-      "grad_norm": 0.001065896824002266,
-      "learning_rate": 0.00019999569976570565,
-      "loss": 46.0,
-      "step": 18341
-    },
-    {
-      "epoch": 2.953822617657716,
-      "grad_norm": 0.0012946195201948285,
-      "learning_rate": 0.00019999569929652004,
-      "loss": 46.0,
-      "step": 18342
-    },
-    {
-      "epoch": 2.9539836547365033,
-      "grad_norm": 0.009168010205030441,
-      "learning_rate": 0.00019999569882730883,
-      "loss": 46.0,
-      "step": 18343
-    },
-    {
-      "epoch": 2.9541446918152907,
-      "grad_norm": 0.008732241578400135,
-      "learning_rate": 0.000199995698358072,
-      "loss": 46.0,
-      "step": 18344
-    },
-    {
-      "epoch": 2.9543057288940777,
-      "grad_norm": 0.0012411995558068156,
-      "learning_rate": 0.0001999956978888096,
-      "loss": 46.0,
-      "step": 18345
-    },
-    {
-      "epoch": 2.954466765972865,
-      "grad_norm": 0.0030234933365136385,
-      "learning_rate": 0.00019999569741952161,
-      "loss": 46.0,
-      "step": 18346
-    },
-    {
-      "epoch": 2.9546278030516526,
-      "grad_norm": 0.0009080435265786946,
-      "learning_rate": 0.00019999569695020803,
-      "loss": 46.0,
-      "step": 18347
-    },
-    {
-      "epoch": 2.95478884013044,
-      "grad_norm": 0.001072784187272191,
-      "learning_rate": 0.00019999569648086887,
-      "loss": 46.0,
-      "step": 18348
-    },
-    {
-      "epoch": 2.9549498772092275,
-      "grad_norm": 0.0020690930541604757,
-      "learning_rate": 0.00019999569601150409,
-      "loss": 46.0,
-      "step": 18349
-    },
-    {
-      "epoch": 2.955110914288015,
-      "grad_norm": 0.0021843689028173685,
-      "learning_rate": 0.00019999569554211372,
-      "loss": 46.0,
-      "step": 18350
-    },
-    {
-      "epoch": 2.955271951366802,
-      "grad_norm": 0.004255227744579315,
-      "learning_rate": 0.0001999956950726978,
-      "loss": 46.0,
-      "step": 18351
-    },
-    {
-      "epoch": 2.9554329884455894,
-      "grad_norm": 0.002831714926287532,
-      "learning_rate": 0.00019999569460325622,
-      "loss": 46.0,
-      "step": 18352
-    },
-    {
-      "epoch": 2.955594025524377,
-      "grad_norm": 0.0016571101732552052,
-      "learning_rate": 0.00019999569413378903,
-      "loss": 46.0,
-      "step": 18353
-    },
-    {
-      "epoch": 2.9557550626031643,
-      "grad_norm": 0.004907633177936077,
-      "learning_rate": 0.00019999569366429632,
-      "loss": 46.0,
-      "step": 18354
-    },
-    {
-      "epoch": 2.955916099681952,
-      "grad_norm": 0.0016868987586349249,
-      "learning_rate": 0.000199995693194778,
-      "loss": 46.0,
-      "step": 18355
-    },
-    {
-      "epoch": 2.9560771367607392,
-      "grad_norm": 0.003417165484279394,
-      "learning_rate": 0.00019999569272523404,
-      "loss": 46.0,
-      "step": 18356
-    },
-    {
-      "epoch": 2.9562381738395267,
-      "grad_norm": 0.006100281607359648,
-      "learning_rate": 0.00019999569225566453,
-      "loss": 46.0,
-      "step": 18357
-    },
-    {
-      "epoch": 2.956399210918314,
-      "grad_norm": 0.0016772671369835734,
-      "learning_rate": 0.00019999569178606944,
-      "loss": 46.0,
-      "step": 18358
-    },
-    {
-      "epoch": 2.9565602479971016,
-      "grad_norm": 0.0018922797171398997,
-      "learning_rate": 0.0001999956913164487,
-      "loss": 46.0,
-      "step": 18359
-    },
-    {
-      "epoch": 2.9567212850758886,
-      "grad_norm": 0.015465022064745426,
-      "learning_rate": 0.0001999956908468024,
-      "loss": 46.0,
-      "step": 18360
-    },
-    {
-      "epoch": 2.956882322154676,
-      "grad_norm": 0.0033949699718505144,
-      "learning_rate": 0.0001999956903771305,
-      "loss": 46.0,
-      "step": 18361
-    },
-    {
-      "epoch": 2.9570433592334635,
-      "grad_norm": 0.008320840075612068,
-      "learning_rate": 0.00019999568990743303,
-      "loss": 46.0,
-      "step": 18362
-    },
-    {
-      "epoch": 2.957204396312251,
-      "grad_norm": 0.0036002597771584988,
-      "learning_rate": 0.00019999568943770992,
-      "loss": 46.0,
-      "step": 18363
-    },
-    {
-      "epoch": 2.9573654333910384,
-      "grad_norm": 0.0005141058354638517,
-      "learning_rate": 0.00019999568896796123,
-      "loss": 46.0,
-      "step": 18364
-    },
-    {
-      "epoch": 2.9575264704698254,
-      "grad_norm": 0.006049633491784334,
-      "learning_rate": 0.00019999568849818697,
-      "loss": 46.0,
-      "step": 18365
-    },
-    {
-      "epoch": 2.957687507548613,
-      "grad_norm": 0.0035453508608043194,
-      "learning_rate": 0.0001999956880283871,
-      "loss": 46.0,
-      "step": 18366
-    },
-    {
-      "epoch": 2.9578485446274003,
-      "grad_norm": 0.003871844382956624,
-      "learning_rate": 0.00019999568755856167,
-      "loss": 46.0,
-      "step": 18367
-    },
-    {
-      "epoch": 2.9580095817061878,
-      "grad_norm": 0.004047180525958538,
-      "learning_rate": 0.0001999956870887106,
-      "loss": 46.0,
-      "step": 18368
-    },
-    {
-      "epoch": 2.9581706187849752,
-      "grad_norm": 0.0045349509455263615,
-      "learning_rate": 0.00019999568661883393,
-      "loss": 46.0,
-      "step": 18369
-    },
-    {
-      "epoch": 2.9583316558637627,
-      "grad_norm": 0.0023616482503712177,
-      "learning_rate": 0.00019999568614893168,
-      "loss": 46.0,
-      "step": 18370
-    },
-    {
-      "epoch": 2.95849269294255,
-      "grad_norm": 0.00441835867241025,
-      "learning_rate": 0.00019999568567900388,
-      "loss": 46.0,
-      "step": 18371
-    },
-    {
-      "epoch": 2.9586537300213376,
-      "grad_norm": 0.007372252177447081,
-      "learning_rate": 0.00019999568520905045,
-      "loss": 46.0,
-      "step": 18372
-    },
-    {
-      "epoch": 2.958814767100125,
-      "grad_norm": 0.004499177914112806,
-      "learning_rate": 0.00019999568473907142,
-      "loss": 46.0,
-      "step": 18373
-    },
-    {
-      "epoch": 2.958975804178912,
-      "grad_norm": 0.0015675516333431005,
-      "learning_rate": 0.0001999956842690668,
-      "loss": 46.0,
-      "step": 18374
-    },
-    {
-      "epoch": 2.9591368412576995,
-      "grad_norm": 0.0010757470736280084,
-      "learning_rate": 0.00019999568379903658,
-      "loss": 46.0,
-      "step": 18375
-    },
-    {
-      "epoch": 2.959297878336487,
-      "grad_norm": 0.001580989919602871,
-      "learning_rate": 0.0001999956833289808,
-      "loss": 46.0,
-      "step": 18376
-    },
-    {
-      "epoch": 2.9594589154152744,
-      "grad_norm": 0.0026291089598089457,
-      "learning_rate": 0.0001999956828588994,
-      "loss": 46.0,
-      "step": 18377
-    },
-    {
-      "epoch": 2.959619952494062,
-      "grad_norm": 0.001443927874788642,
-      "learning_rate": 0.00019999568238879238,
-      "loss": 46.0,
-      "step": 18378
-    },
-    {
-      "epoch": 2.9597809895728493,
-      "grad_norm": 0.0013021137565374374,
-      "learning_rate": 0.00019999568191865982,
-      "loss": 46.0,
-      "step": 18379
-    },
-    {
-      "epoch": 2.9599420266516363,
-      "grad_norm": 0.001310904510319233,
-      "learning_rate": 0.00019999568144850164,
-      "loss": 46.0,
-      "step": 18380
-    },
-    {
-      "epoch": 2.9601030637304238,
-      "grad_norm": 0.0023698126897215843,
-      "learning_rate": 0.00019999568097831786,
-      "loss": 46.0,
-      "step": 18381
-    },
-    {
-      "epoch": 2.960264100809211,
-      "grad_norm": 0.0039572035893797874,
-      "learning_rate": 0.0001999956805081085,
-      "loss": 46.0,
-      "step": 18382
-    },
-    {
-      "epoch": 2.9604251378879987,
-      "grad_norm": 0.007814143784344196,
-      "learning_rate": 0.00019999568003787355,
-      "loss": 46.0,
-      "step": 18383
-    },
-    {
-      "epoch": 2.960586174966786,
-      "grad_norm": 0.002072358736768365,
-      "learning_rate": 0.00019999567956761297,
-      "loss": 46.0,
-      "step": 18384
-    },
-    {
-      "epoch": 2.9607472120455736,
-      "grad_norm": 0.0022338759154081345,
-      "learning_rate": 0.00019999567909732683,
-      "loss": 46.0,
-      "step": 18385
-    },
-    {
-      "epoch": 2.960908249124361,
-      "grad_norm": 0.0014239565934985876,
-      "learning_rate": 0.00019999567862701508,
-      "loss": 46.0,
-      "step": 18386
-    },
-    {
-      "epoch": 2.9610692862031485,
-      "grad_norm": 0.00490310974419117,
-      "learning_rate": 0.00019999567815667777,
-      "loss": 46.0,
-      "step": 18387
-    },
-    {
-      "epoch": 2.961230323281936,
-      "grad_norm": 0.0010968658607453108,
-      "learning_rate": 0.00019999567768631482,
-      "loss": 46.0,
-      "step": 18388
-    },
-    {
-      "epoch": 2.961391360360723,
-      "grad_norm": 0.005245309788733721,
-      "learning_rate": 0.00019999567721592628,
-      "loss": 46.0,
-      "step": 18389
-    },
-    {
-      "epoch": 2.9615523974395104,
-      "grad_norm": 0.0024718816857784986,
-      "learning_rate": 0.00019999567674551218,
-      "loss": 46.0,
-      "step": 18390
-    },
-    {
-      "epoch": 2.961713434518298,
-      "grad_norm": 0.00124309619423002,
-      "learning_rate": 0.00019999567627507247,
-      "loss": 46.0,
-      "step": 18391
-    },
-    {
-      "epoch": 2.9618744715970853,
-      "grad_norm": 0.009283908642828465,
-      "learning_rate": 0.0001999956758046072,
-      "loss": 46.0,
-      "step": 18392
-    },
-    {
-      "epoch": 2.9620355086758727,
-      "grad_norm": 0.0023621288128197193,
-      "learning_rate": 0.00019999567533411628,
-      "loss": 46.0,
-      "step": 18393
-    },
-    {
-      "epoch": 2.9621965457546597,
-      "grad_norm": 0.005116816144436598,
-      "learning_rate": 0.00019999567486359978,
-      "loss": 46.0,
-      "step": 18394
-    },
-    {
-      "epoch": 2.962357582833447,
-      "grad_norm": 0.001094233593903482,
-      "learning_rate": 0.00019999567439305769,
-      "loss": 46.0,
-      "step": 18395
-    },
-    {
-      "epoch": 2.9625186199122346,
-      "grad_norm": 0.0023934142664074898,
-      "learning_rate": 0.00019999567392249,
-      "loss": 46.0,
-      "step": 18396
-    },
-    {
-      "epoch": 2.962679656991022,
-      "grad_norm": 0.0014208616921678185,
-      "learning_rate": 0.00019999567345189675,
-      "loss": 46.0,
-      "step": 18397
-    },
-    {
-      "epoch": 2.9628406940698095,
-      "grad_norm": 0.002365308813750744,
-      "learning_rate": 0.0001999956729812779,
-      "loss": 46.0,
-      "step": 18398
-    },
-    {
-      "epoch": 2.963001731148597,
-      "grad_norm": 0.0033565261401236057,
-      "learning_rate": 0.00019999567251063343,
-      "loss": 46.0,
-      "step": 18399
-    },
-    {
-      "epoch": 2.9631627682273844,
-      "grad_norm": 0.00140491989441216,
-      "learning_rate": 0.00019999567203996335,
-      "loss": 46.0,
-      "step": 18400
-    },
-    {
-      "epoch": 2.963323805306172,
-      "grad_norm": 0.0060475068166852,
-      "learning_rate": 0.00019999567156926774,
-      "loss": 46.0,
-      "step": 18401
-    },
-    {
-      "epoch": 2.9634848423849594,
-      "grad_norm": 0.0017807814292609692,
-      "learning_rate": 0.00019999567109854646,
-      "loss": 46.0,
-      "step": 18402
-    },
-    {
-      "epoch": 2.963645879463747,
-      "grad_norm": 0.0020399540662765503,
-      "learning_rate": 0.00019999567062779962,
-      "loss": 46.0,
-      "step": 18403
-    },
-    {
-      "epoch": 2.963806916542534,
-      "grad_norm": 0.0048352801240980625,
-      "learning_rate": 0.00019999567015702721,
-      "loss": 46.0,
-      "step": 18404
-    },
-    {
-      "epoch": 2.9639679536213213,
-      "grad_norm": 0.0077485013753175735,
-      "learning_rate": 0.00019999566968622917,
-      "loss": 46.0,
-      "step": 18405
-    },
-    {
-      "epoch": 2.9641289907001087,
-      "grad_norm": 0.005939498543739319,
-      "learning_rate": 0.00019999566921540557,
-      "loss": 46.0,
-      "step": 18406
-    },
-    {
-      "epoch": 2.964290027778896,
-      "grad_norm": 0.0010195238282904029,
-      "learning_rate": 0.00019999566874455635,
-      "loss": 46.0,
-      "step": 18407
-    },
-    {
-      "epoch": 2.9644510648576836,
-      "grad_norm": 0.0034312172792851925,
-      "learning_rate": 0.00019999566827368155,
-      "loss": 46.0,
-      "step": 18408
-    },
-    {
-      "epoch": 2.9646121019364706,
-      "grad_norm": 0.0017127074534073472,
-      "learning_rate": 0.00019999566780278113,
-      "loss": 46.0,
-      "step": 18409
-    },
-    {
-      "epoch": 2.964773139015258,
-      "grad_norm": 0.004184579476714134,
-      "learning_rate": 0.00019999566733185518,
-      "loss": 46.0,
-      "step": 18410
-    },
-    {
-      "epoch": 2.9649341760940455,
-      "grad_norm": 0.00745483348146081,
-      "learning_rate": 0.0001999956668609036,
-      "loss": 46.0,
-      "step": 18411
-    },
-    {
-      "epoch": 2.965095213172833,
-      "grad_norm": 0.0038978401571512222,
-      "learning_rate": 0.0001999956663899264,
-      "loss": 46.0,
-      "step": 18412
-    },
-    {
-      "epoch": 2.9652562502516204,
-      "grad_norm": 0.0011870965827256441,
-      "learning_rate": 0.0001999956659189236,
-      "loss": 46.0,
-      "step": 18413
-    },
-    {
-      "epoch": 2.965417287330408,
-      "grad_norm": 0.0010393648408353329,
-      "learning_rate": 0.00019999566544789526,
-      "loss": 46.0,
-      "step": 18414
-    },
-    {
-      "epoch": 2.9655783244091953,
-      "grad_norm": 0.0019145701080560684,
-      "learning_rate": 0.0001999956649768413,
-      "loss": 46.0,
-      "step": 18415
-    },
-    {
-      "epoch": 2.965739361487983,
-      "grad_norm": 0.0017798240296542645,
-      "learning_rate": 0.00019999566450576173,
-      "loss": 46.0,
-      "step": 18416
-    },
-    {
-      "epoch": 2.9659003985667702,
-      "grad_norm": 0.0005918376846238971,
-      "learning_rate": 0.0001999956640346566,
-      "loss": 46.0,
-      "step": 18417
-    },
-    {
-      "epoch": 2.9660614356455572,
-      "grad_norm": 0.0017097140662372112,
-      "learning_rate": 0.00019999566356352583,
-      "loss": 46.0,
-      "step": 18418
-    },
-    {
-      "epoch": 2.9662224727243447,
-      "grad_norm": 0.0015214142622426152,
-      "learning_rate": 0.00019999566309236952,
-      "loss": 46.0,
-      "step": 18419
-    },
-    {
-      "epoch": 2.966383509803132,
-      "grad_norm": 0.004899428226053715,
-      "learning_rate": 0.00019999566262118759,
-      "loss": 46.0,
-      "step": 18420
-    },
-    {
-      "epoch": 2.9665445468819196,
-      "grad_norm": 0.005743523593991995,
-      "learning_rate": 0.00019999566214998007,
-      "loss": 46.0,
-      "step": 18421
-    },
-    {
-      "epoch": 2.966705583960707,
-      "grad_norm": 0.0011276621371507645,
-      "learning_rate": 0.00019999566167874694,
-      "loss": 46.0,
-      "step": 18422
-    },
-    {
-      "epoch": 2.9668666210394945,
-      "grad_norm": 0.0027939225547015667,
-      "learning_rate": 0.00019999566120748822,
-      "loss": 46.0,
-      "step": 18423
-    },
-    {
-      "epoch": 2.9670276581182815,
-      "grad_norm": 0.008693045936524868,
-      "learning_rate": 0.00019999566073620394,
-      "loss": 46.0,
-      "step": 18424
-    },
-    {
-      "epoch": 2.967188695197069,
-      "grad_norm": 0.001919653033837676,
-      "learning_rate": 0.00019999566026489401,
-      "loss": 46.0,
-      "step": 18425
-    },
-    {
-      "epoch": 2.9673497322758564,
-      "grad_norm": 0.00215740199200809,
-      "learning_rate": 0.00019999565979355853,
-      "loss": 46.0,
-      "step": 18426
-    },
-    {
-      "epoch": 2.967510769354644,
-      "grad_norm": 0.0043504973873496056,
-      "learning_rate": 0.00019999565932219744,
-      "loss": 46.0,
-      "step": 18427
-    },
-    {
-      "epoch": 2.9676718064334313,
-      "grad_norm": 0.002733775181695819,
-      "learning_rate": 0.00019999565885081078,
-      "loss": 46.0,
-      "step": 18428
-    },
-    {
-      "epoch": 2.9678328435122188,
-      "grad_norm": 0.0008774618036113679,
-      "learning_rate": 0.00019999565837939848,
-      "loss": 46.0,
-      "step": 18429
-    },
-    {
-      "epoch": 2.967993880591006,
-      "grad_norm": 0.003374746534973383,
-      "learning_rate": 0.00019999565790796063,
-      "loss": 46.0,
-      "step": 18430
-    },
-    {
-      "epoch": 2.9681549176697937,
-      "grad_norm": 0.001083322917111218,
-      "learning_rate": 0.00019999565743649718,
-      "loss": 46.0,
-      "step": 18431
-    },
-    {
-      "epoch": 2.968315954748581,
-      "grad_norm": 0.0014616484986618161,
-      "learning_rate": 0.0001999956569650081,
-      "loss": 46.0,
-      "step": 18432
-    },
-    {
-      "epoch": 2.968476991827368,
-      "grad_norm": 0.004131474066525698,
-      "learning_rate": 0.00019999565649349345,
-      "loss": 46.0,
-      "step": 18433
-    },
-    {
-      "epoch": 2.9686380289061556,
-      "grad_norm": 0.0019515649182721972,
-      "learning_rate": 0.00019999565602195322,
-      "loss": 46.0,
-      "step": 18434
-    },
-    {
-      "epoch": 2.968799065984943,
-      "grad_norm": 0.0015557692386209965,
-      "learning_rate": 0.00019999565555038737,
-      "loss": 46.0,
-      "step": 18435
-    },
-    {
-      "epoch": 2.9689601030637305,
-      "grad_norm": 0.002823358168825507,
-      "learning_rate": 0.00019999565507879596,
-      "loss": 46.0,
-      "step": 18436
-    },
-    {
-      "epoch": 2.969121140142518,
-      "grad_norm": 0.014139360748231411,
-      "learning_rate": 0.00019999565460717894,
-      "loss": 46.0,
-      "step": 18437
-    },
-    {
-      "epoch": 2.969282177221305,
-      "grad_norm": 0.0016486848471686244,
-      "learning_rate": 0.0001999956541355363,
-      "loss": 46.0,
-      "step": 18438
-    },
-    {
-      "epoch": 2.9694432143000924,
-      "grad_norm": 0.01684490032494068,
-      "learning_rate": 0.0001999956536638681,
-      "loss": 46.0,
-      "step": 18439
-    },
-    {
-      "epoch": 2.96960425137888,
-      "grad_norm": 0.00124447257257998,
-      "learning_rate": 0.0001999956531921743,
-      "loss": 46.0,
-      "step": 18440
-    },
-    {
-      "epoch": 2.9697652884576673,
-      "grad_norm": 0.004933250602334738,
-      "learning_rate": 0.0001999956527204549,
-      "loss": 46.0,
-      "step": 18441
-    },
-    {
-      "epoch": 2.9699263255364547,
-      "grad_norm": 0.002164689125493169,
-      "learning_rate": 0.00019999565224870992,
-      "loss": 46.0,
-      "step": 18442
-    },
-    {
-      "epoch": 2.970087362615242,
-      "grad_norm": 0.0036833875346928835,
-      "learning_rate": 0.00019999565177693932,
-      "loss": 46.0,
-      "step": 18443
-    },
-    {
-      "epoch": 2.9702483996940297,
-      "grad_norm": 0.0016168163856491446,
-      "learning_rate": 0.00019999565130514313,
-      "loss": 46.0,
-      "step": 18444
-    },
-    {
-      "epoch": 2.970409436772817,
-      "grad_norm": 0.00426136888563633,
-      "learning_rate": 0.00019999565083332136,
-      "loss": 46.0,
-      "step": 18445
-    },
-    {
-      "epoch": 2.9705704738516046,
-      "grad_norm": 0.0007692842627875507,
-      "learning_rate": 0.000199995650361474,
-      "loss": 46.0,
-      "step": 18446
-    },
-    {
-      "epoch": 2.9707315109303916,
-      "grad_norm": 0.004225170239806175,
-      "learning_rate": 0.00019999564988960105,
-      "loss": 46.0,
-      "step": 18447
-    },
-    {
-      "epoch": 2.970892548009179,
-      "grad_norm": 0.005309720989316702,
-      "learning_rate": 0.0001999956494177025,
-      "loss": 46.0,
-      "step": 18448
-    },
-    {
-      "epoch": 2.9710535850879665,
-      "grad_norm": 0.001555133960209787,
-      "learning_rate": 0.00019999564894577834,
-      "loss": 46.0,
-      "step": 18449
-    },
-    {
-      "epoch": 2.971214622166754,
-      "grad_norm": 0.004254933912307024,
-      "learning_rate": 0.0001999956484738286,
-      "loss": 46.0,
-      "step": 18450
-    },
-    {
-      "epoch": 2.9713756592455414,
-      "grad_norm": 0.0007019075565040112,
-      "learning_rate": 0.00019999564800185325,
-      "loss": 46.0,
-      "step": 18451
-    },
-    {
-      "epoch": 2.971536696324329,
-      "grad_norm": 0.0023639341816306114,
-      "learning_rate": 0.00019999564752985234,
-      "loss": 46.0,
-      "step": 18452
-    },
-    {
-      "epoch": 2.971697733403116,
-      "grad_norm": 0.00486363610252738,
-      "learning_rate": 0.00019999564705782582,
-      "loss": 46.0,
-      "step": 18453
-    },
-    {
-      "epoch": 2.9718587704819033,
-      "grad_norm": 0.004407297819852829,
-      "learning_rate": 0.0001999956465857737,
-      "loss": 46.0,
-      "step": 18454
-    },
-    {
-      "epoch": 2.9720198075606907,
-      "grad_norm": 0.0015118003357201815,
-      "learning_rate": 0.00019999564611369598,
-      "loss": 46.0,
-      "step": 18455
-    },
-    {
-      "epoch": 2.972180844639478,
-      "grad_norm": 0.006015649531036615,
-      "learning_rate": 0.00019999564564159266,
-      "loss": 46.0,
-      "step": 18456
-    },
-    {
-      "epoch": 2.9723418817182656,
-      "grad_norm": 0.002204635413363576,
-      "learning_rate": 0.0001999956451694638,
-      "loss": 46.0,
-      "step": 18457
-    },
-    {
-      "epoch": 2.972502918797053,
-      "grad_norm": 0.003703678958117962,
-      "learning_rate": 0.0001999956446973093,
-      "loss": 46.0,
-      "step": 18458
-    },
-    {
-      "epoch": 2.9726639558758405,
-      "grad_norm": 0.003325363388285041,
-      "learning_rate": 0.0001999956442251292,
-      "loss": 46.0,
-      "step": 18459
-    },
-    {
-      "epoch": 2.972824992954628,
-      "grad_norm": 0.004380201920866966,
-      "learning_rate": 0.0001999956437529235,
-      "loss": 46.0,
-      "step": 18460
-    },
-    {
-      "epoch": 2.9729860300334154,
-      "grad_norm": 0.002804793184623122,
-      "learning_rate": 0.00019999564328069224,
-      "loss": 46.0,
-      "step": 18461
-    },
-    {
-      "epoch": 2.9731470671122024,
-      "grad_norm": 0.0028695680666714907,
-      "learning_rate": 0.0001999956428084354,
-      "loss": 46.0,
-      "step": 18462
-    },
-    {
-      "epoch": 2.97330810419099,
-      "grad_norm": 0.00822277832776308,
-      "learning_rate": 0.00019999564233615292,
-      "loss": 46.0,
-      "step": 18463
-    },
-    {
-      "epoch": 2.9734691412697773,
-      "grad_norm": 0.002469172002747655,
-      "learning_rate": 0.00019999564186384486,
-      "loss": 46.0,
-      "step": 18464
-    },
-    {
-      "epoch": 2.973630178348565,
-      "grad_norm": 0.0005723289796151221,
-      "learning_rate": 0.0001999956413915112,
-      "loss": 46.0,
-      "step": 18465
-    },
-    {
-      "epoch": 2.9737912154273523,
-      "grad_norm": 0.0018965598428621888,
-      "learning_rate": 0.00019999564091915196,
-      "loss": 46.0,
-      "step": 18466
-    },
-    {
-      "epoch": 2.9739522525061397,
-      "grad_norm": 0.00133640191052109,
-      "learning_rate": 0.00019999564044676714,
-      "loss": 46.0,
-      "step": 18467
-    },
-    {
-      "epoch": 2.9741132895849267,
-      "grad_norm": 0.00894093606621027,
-      "learning_rate": 0.0001999956399743567,
-      "loss": 46.0,
-      "step": 18468
-    },
-    {
-      "epoch": 2.974274326663714,
-      "grad_norm": 0.002226758748292923,
-      "learning_rate": 0.0001999956395019207,
-      "loss": 46.0,
-      "step": 18469
-    },
-    {
-      "epoch": 2.9744353637425016,
-      "grad_norm": 0.004566085524857044,
-      "learning_rate": 0.00019999563902945905,
-      "loss": 46.0,
-      "step": 18470
-    },
-    {
-      "epoch": 2.974596400821289,
-      "grad_norm": 0.006014668848365545,
-      "learning_rate": 0.00019999563855697185,
-      "loss": 46.0,
-      "step": 18471
-    },
-    {
-      "epoch": 2.9747574379000765,
-      "grad_norm": 0.002742860233411193,
-      "learning_rate": 0.00019999563808445903,
-      "loss": 46.0,
-      "step": 18472
-    },
-    {
-      "epoch": 2.974918474978864,
-      "grad_norm": 0.0037918067537248135,
-      "learning_rate": 0.00019999563761192066,
-      "loss": 46.0,
-      "step": 18473
-    },
-    {
-      "epoch": 2.9750795120576514,
-      "grad_norm": 0.004271207842975855,
-      "learning_rate": 0.00019999563713935664,
-      "loss": 46.0,
-      "step": 18474
-    },
-    {
-      "epoch": 2.975240549136439,
-      "grad_norm": 0.0038061838131397963,
-      "learning_rate": 0.00019999563666676706,
-      "loss": 46.0,
-      "step": 18475
-    },
-    {
-      "epoch": 2.9754015862152263,
-      "grad_norm": 0.001159599982202053,
-      "learning_rate": 0.0001999956361941519,
-      "loss": 46.0,
-      "step": 18476
-    },
-    {
-      "epoch": 2.9755626232940133,
-      "grad_norm": 0.0021892578806728125,
-      "learning_rate": 0.0001999956357215111,
-      "loss": 46.0,
-      "step": 18477
-    },
-    {
-      "epoch": 2.975723660372801,
-      "grad_norm": 0.0023154402151703835,
-      "learning_rate": 0.00019999563524884475,
-      "loss": 46.0,
-      "step": 18478
-    },
-    {
-      "epoch": 2.9758846974515882,
-      "grad_norm": 0.009547240100800991,
-      "learning_rate": 0.00019999563477615277,
-      "loss": 46.0,
-      "step": 18479
-    },
-    {
-      "epoch": 2.9760457345303757,
-      "grad_norm": 0.000793353421613574,
-      "learning_rate": 0.00019999563430343523,
-      "loss": 46.0,
-      "step": 18480
-    },
-    {
-      "epoch": 2.976206771609163,
-      "grad_norm": 0.006900258362293243,
-      "learning_rate": 0.0001999956338306921,
-      "loss": 46.0,
-      "step": 18481
-    },
-    {
-      "epoch": 2.97636780868795,
-      "grad_norm": 0.0014947090530768037,
-      "learning_rate": 0.00019999563335792333,
-      "loss": 46.0,
-      "step": 18482
-    },
-    {
-      "epoch": 2.9765288457667376,
-      "grad_norm": 0.004521749448031187,
-      "learning_rate": 0.000199995632885129,
-      "loss": 46.0,
-      "step": 18483
-    },
-    {
-      "epoch": 2.976689882845525,
-      "grad_norm": 0.001985024195164442,
-      "learning_rate": 0.00019999563241230908,
-      "loss": 46.0,
-      "step": 18484
-    },
-    {
-      "epoch": 2.9768509199243125,
-      "grad_norm": 0.001415968406945467,
-      "learning_rate": 0.00019999563193946353,
-      "loss": 46.0,
-      "step": 18485
-    },
-    {
-      "epoch": 2.9770119570031,
-      "grad_norm": 0.0012037853011861444,
-      "learning_rate": 0.00019999563146659243,
-      "loss": 46.0,
-      "step": 18486
-    },
-    {
-      "epoch": 2.9771729940818874,
-      "grad_norm": 0.0011512440396472812,
-      "learning_rate": 0.00019999563099369573,
-      "loss": 46.0,
-      "step": 18487
-    },
-    {
-      "epoch": 2.977334031160675,
-      "grad_norm": 0.0021884555462747812,
-      "learning_rate": 0.0001999956305207734,
-      "loss": 46.0,
-      "step": 18488
-    },
-    {
-      "epoch": 2.9774950682394623,
-      "grad_norm": 0.005605579353868961,
-      "learning_rate": 0.00019999563004782553,
-      "loss": 46.0,
-      "step": 18489
-    },
-    {
-      "epoch": 2.9776561053182498,
-      "grad_norm": 0.003936262335628271,
-      "learning_rate": 0.00019999562957485203,
-      "loss": 46.0,
-      "step": 18490
-    },
-    {
-      "epoch": 2.9778171423970368,
-      "grad_norm": 0.0010099461069330573,
-      "learning_rate": 0.00019999562910185295,
-      "loss": 46.0,
-      "step": 18491
-    },
-    {
-      "epoch": 2.977978179475824,
-      "grad_norm": 0.0025921945925801992,
-      "learning_rate": 0.00019999562862882826,
-      "loss": 46.0,
-      "step": 18492
-    },
-    {
-      "epoch": 2.9781392165546117,
-      "grad_norm": 0.0014950225595384836,
-      "learning_rate": 0.000199995628155778,
-      "loss": 46.0,
-      "step": 18493
-    },
-    {
-      "epoch": 2.978300253633399,
-      "grad_norm": 0.0087662935256958,
-      "learning_rate": 0.0001999956276827021,
-      "loss": 46.0,
-      "step": 18494
-    },
-    {
-      "epoch": 2.9784612907121866,
-      "grad_norm": 0.0038464716635644436,
-      "learning_rate": 0.00019999562720960065,
-      "loss": 46.0,
-      "step": 18495
-    },
-    {
-      "epoch": 2.978622327790974,
-      "grad_norm": 0.0013468189863488078,
-      "learning_rate": 0.0001999956267364736,
-      "loss": 46.0,
-      "step": 18496
-    },
-    {
-      "epoch": 2.978783364869761,
-      "grad_norm": 0.014001449570059776,
-      "learning_rate": 0.00019999562626332095,
-      "loss": 46.0,
-      "step": 18497
-    },
-    {
-      "epoch": 2.9789444019485485,
-      "grad_norm": 0.004450570326298475,
-      "learning_rate": 0.0001999956257901427,
-      "loss": 46.0,
-      "step": 18498
-    },
-    {
-      "epoch": 2.979105439027336,
-      "grad_norm": 0.0011002302635461092,
-      "learning_rate": 0.00019999562531693887,
-      "loss": 46.0,
-      "step": 18499
-    },
-    {
-      "epoch": 2.9792664761061234,
-      "grad_norm": 0.0022788571659475565,
-      "learning_rate": 0.00019999562484370945,
-      "loss": 46.0,
-      "step": 18500
-    },
-    {
-      "epoch": 2.979427513184911,
-      "grad_norm": 0.003244733205065131,
-      "learning_rate": 0.00019999562437045444,
-      "loss": 46.0,
-      "step": 18501
-    },
-    {
-      "epoch": 2.9795885502636983,
-      "grad_norm": 0.004255375359207392,
-      "learning_rate": 0.00019999562389717382,
-      "loss": 46.0,
-      "step": 18502
-    },
-    {
-      "epoch": 2.9797495873424857,
-      "grad_norm": 0.006406597327440977,
-      "learning_rate": 0.00019999562342386758,
-      "loss": 46.0,
-      "step": 18503
-    },
-    {
-      "epoch": 2.979910624421273,
-      "grad_norm": 0.0012827736791223288,
-      "learning_rate": 0.00019999562295053579,
-      "loss": 46.0,
-      "step": 18504
-    },
-    {
-      "epoch": 2.9800716615000606,
-      "grad_norm": 0.00554022379219532,
-      "learning_rate": 0.0001999956224771784,
-      "loss": 46.0,
-      "step": 18505
-    },
-    {
-      "epoch": 2.9802326985788476,
-      "grad_norm": 0.0045968820340931416,
-      "learning_rate": 0.00019999562200379538,
-      "loss": 46.0,
-      "step": 18506
-    },
-    {
-      "epoch": 2.980393735657635,
-      "grad_norm": 0.0027091035153716803,
-      "learning_rate": 0.00019999562153038682,
-      "loss": 46.0,
-      "step": 18507
-    },
-    {
-      "epoch": 2.9805547727364226,
-      "grad_norm": 0.014366013929247856,
-      "learning_rate": 0.00019999562105695265,
-      "loss": 46.0,
-      "step": 18508
-    },
-    {
-      "epoch": 2.98071580981521,
-      "grad_norm": 0.004478765185922384,
-      "learning_rate": 0.00019999562058349286,
-      "loss": 46.0,
-      "step": 18509
-    },
-    {
-      "epoch": 2.9808768468939975,
-      "grad_norm": 0.0019709656480699778,
-      "learning_rate": 0.0001999956201100075,
-      "loss": 46.0,
-      "step": 18510
-    },
-    {
-      "epoch": 2.9810378839727845,
-      "grad_norm": 0.001509210211224854,
-      "learning_rate": 0.00019999561963649653,
-      "loss": 46.0,
-      "step": 18511
-    },
-    {
-      "epoch": 2.981198921051572,
-      "grad_norm": 0.003124014940112829,
-      "learning_rate": 0.00019999561916295996,
-      "loss": 46.0,
-      "step": 18512
-    },
-    {
-      "epoch": 2.9813599581303594,
-      "grad_norm": 0.012680255807936192,
-      "learning_rate": 0.00019999561868939782,
-      "loss": 46.0,
-      "step": 18513
-    },
-    {
-      "epoch": 2.981520995209147,
-      "grad_norm": 0.004033701494336128,
-      "learning_rate": 0.00019999561821581007,
-      "loss": 46.0,
-      "step": 18514
-    },
-    {
-      "epoch": 2.9816820322879343,
-      "grad_norm": 0.00432217214256525,
-      "learning_rate": 0.00019999561774219676,
-      "loss": 46.0,
-      "step": 18515
-    },
-    {
-      "epoch": 2.9818430693667217,
-      "grad_norm": 0.00091934745432809,
-      "learning_rate": 0.0001999956172685578,
-      "loss": 46.0,
-      "step": 18516
-    },
-    {
-      "epoch": 2.982004106445509,
-      "grad_norm": 0.0025655035860836506,
-      "learning_rate": 0.0001999956167948933,
-      "loss": 46.0,
-      "step": 18517
-    },
-    {
-      "epoch": 2.9821651435242966,
-      "grad_norm": 0.0009710592566989362,
-      "learning_rate": 0.00019999561632120315,
-      "loss": 46.0,
-      "step": 18518
-    },
-    {
-      "epoch": 2.982326180603084,
-      "grad_norm": 0.002257666317746043,
-      "learning_rate": 0.00019999561584748747,
-      "loss": 46.0,
-      "step": 18519
-    },
-    {
-      "epoch": 2.9824872176818715,
-      "grad_norm": 0.003338546259328723,
-      "learning_rate": 0.00019999561537374617,
-      "loss": 46.0,
-      "step": 18520
-    },
-    {
-      "epoch": 2.9826482547606585,
-      "grad_norm": 0.003992764744907618,
-      "learning_rate": 0.00019999561489997925,
-      "loss": 46.0,
-      "step": 18521
-    },
-    {
-      "epoch": 2.982809291839446,
-      "grad_norm": 0.002327491994947195,
-      "learning_rate": 0.00019999561442618678,
-      "loss": 46.0,
-      "step": 18522
-    },
-    {
-      "epoch": 2.9829703289182334,
-      "grad_norm": 0.0035936960484832525,
-      "learning_rate": 0.00019999561395236867,
-      "loss": 46.0,
-      "step": 18523
-    },
-    {
-      "epoch": 2.983131365997021,
-      "grad_norm": 0.0011990476632490754,
-      "learning_rate": 0.000199995613478525,
-      "loss": 46.0,
-      "step": 18524
-    },
-    {
-      "epoch": 2.9832924030758083,
-      "grad_norm": 0.008181135170161724,
-      "learning_rate": 0.00019999561300465573,
-      "loss": 46.0,
-      "step": 18525
-    },
-    {
-      "epoch": 2.9834534401545953,
-      "grad_norm": 0.005735564976930618,
-      "learning_rate": 0.00019999561253076085,
-      "loss": 46.0,
-      "step": 18526
-    },
-    {
-      "epoch": 2.983614477233383,
-      "grad_norm": 0.0009439793648198247,
-      "learning_rate": 0.0001999956120568404,
-      "loss": 46.0,
-      "step": 18527
-    },
-    {
-      "epoch": 2.9837755143121703,
-      "grad_norm": 0.0027761005330830812,
-      "learning_rate": 0.00019999561158289434,
-      "loss": 46.0,
-      "step": 18528
-    },
-    {
-      "epoch": 2.9839365513909577,
-      "grad_norm": 0.00563826272264123,
-      "learning_rate": 0.00019999561110892267,
-      "loss": 46.0,
-      "step": 18529
-    },
-    {
-      "epoch": 2.984097588469745,
-      "grad_norm": 0.0022981413640081882,
-      "learning_rate": 0.00019999561063492545,
-      "loss": 46.0,
-      "step": 18530
-    },
-    {
-      "epoch": 2.9842586255485326,
-      "grad_norm": 0.0004431944980751723,
-      "learning_rate": 0.00019999561016090264,
-      "loss": 46.0,
-      "step": 18531
-    },
-    {
-      "epoch": 2.98441966262732,
-      "grad_norm": 0.002274968195706606,
-      "learning_rate": 0.00019999560968685418,
-      "loss": 46.0,
-      "step": 18532
-    },
-    {
-      "epoch": 2.9845806997061075,
-      "grad_norm": 0.0027848570607602596,
-      "learning_rate": 0.00019999560921278014,
-      "loss": 46.0,
-      "step": 18533
-    },
-    {
-      "epoch": 2.984741736784895,
-      "grad_norm": 0.0008386837434954941,
-      "learning_rate": 0.00019999560873868054,
-      "loss": 46.0,
-      "step": 18534
-    },
-    {
-      "epoch": 2.984902773863682,
-      "grad_norm": 0.006353685632348061,
-      "learning_rate": 0.00019999560826455533,
-      "loss": 46.0,
-      "step": 18535
-    },
-    {
-      "epoch": 2.9850638109424694,
-      "grad_norm": 0.0015298710204660892,
-      "learning_rate": 0.00019999560779040452,
-      "loss": 46.0,
-      "step": 18536
-    },
-    {
-      "epoch": 2.985224848021257,
-      "grad_norm": 0.010095451027154922,
-      "learning_rate": 0.00019999560731622814,
-      "loss": 46.0,
-      "step": 18537
-    },
-    {
-      "epoch": 2.9853858851000443,
-      "grad_norm": 0.0013281877618283033,
-      "learning_rate": 0.00019999560684202613,
-      "loss": 46.0,
-      "step": 18538
-    },
-    {
-      "epoch": 2.9855469221788318,
-      "grad_norm": 0.001293028355576098,
-      "learning_rate": 0.00019999560636779854,
-      "loss": 46.0,
-      "step": 18539
-    },
-    {
-      "epoch": 2.9857079592576192,
-      "grad_norm": 0.005644380114972591,
-      "learning_rate": 0.00019999560589354536,
-      "loss": 46.0,
-      "step": 18540
-    },
-    {
-      "epoch": 2.9858689963364062,
-      "grad_norm": 0.002164635807275772,
-      "learning_rate": 0.0001999956054192666,
-      "loss": 46.0,
-      "step": 18541
-    },
-    {
-      "epoch": 2.9860300334151937,
-      "grad_norm": 0.0010207355953752995,
-      "learning_rate": 0.00019999560494496225,
-      "loss": 46.0,
-      "step": 18542
-    },
-    {
-      "epoch": 2.986191070493981,
-      "grad_norm": 0.0008625152986496687,
-      "learning_rate": 0.00019999560447063228,
-      "loss": 46.0,
-      "step": 18543
-    },
-    {
-      "epoch": 2.9863521075727686,
-      "grad_norm": 0.0008602470043115318,
-      "learning_rate": 0.0001999956039962767,
-      "loss": 46.0,
-      "step": 18544
-    },
-    {
-      "epoch": 2.986513144651556,
-      "grad_norm": 0.0026290961541235447,
-      "learning_rate": 0.00019999560352189556,
-      "loss": 46.0,
-      "step": 18545
-    },
-    {
-      "epoch": 2.9866741817303435,
-      "grad_norm": 0.003948190715163946,
-      "learning_rate": 0.0001999956030474888,
-      "loss": 46.0,
-      "step": 18546
-    },
-    {
-      "epoch": 2.986835218809131,
-      "grad_norm": 0.0006668227142654359,
-      "learning_rate": 0.0001999956025730565,
-      "loss": 46.0,
-      "step": 18547
-    },
-    {
-      "epoch": 2.9869962558879184,
-      "grad_norm": 0.00663685193285346,
-      "learning_rate": 0.00019999560209859856,
-      "loss": 46.0,
-      "step": 18548
-    },
-    {
-      "epoch": 2.987157292966706,
-      "grad_norm": 0.006620586384087801,
-      "learning_rate": 0.00019999560162411504,
-      "loss": 46.0,
-      "step": 18549
-    },
-    {
-      "epoch": 2.987318330045493,
-      "grad_norm": 0.0052564614452421665,
-      "learning_rate": 0.0001999956011496059,
-      "loss": 46.0,
-      "step": 18550
-    },
-    {
-      "epoch": 2.9874793671242803,
-      "grad_norm": 0.0028195157647132874,
-      "learning_rate": 0.00019999560067507122,
-      "loss": 46.0,
-      "step": 18551
-    },
-    {
-      "epoch": 2.9876404042030678,
-      "grad_norm": 0.004455295391380787,
-      "learning_rate": 0.0001999956002005109,
-      "loss": 46.0,
-      "step": 18552
-    },
-    {
-      "epoch": 2.987801441281855,
-      "grad_norm": 0.007713985163718462,
-      "learning_rate": 0.00019999559972592502,
-      "loss": 46.0,
-      "step": 18553
-    },
-    {
-      "epoch": 2.9879624783606427,
-      "grad_norm": 0.002368438523262739,
-      "learning_rate": 0.00019999559925131351,
-      "loss": 46.0,
-      "step": 18554
-    },
-    {
-      "epoch": 2.9881235154394297,
-      "grad_norm": 0.0034312766510993242,
-      "learning_rate": 0.00019999559877667645,
-      "loss": 46.0,
-      "step": 18555
-    },
-    {
-      "epoch": 2.988284552518217,
-      "grad_norm": 0.0013110170839354396,
-      "learning_rate": 0.00019999559830201374,
-      "loss": 46.0,
-      "step": 18556
-    },
-    {
-      "epoch": 2.9884455895970046,
-      "grad_norm": 0.00045992841478437185,
-      "learning_rate": 0.0001999955978273255,
-      "loss": 46.0,
-      "step": 18557
-    },
-    {
-      "epoch": 2.988606626675792,
-      "grad_norm": 0.002845642389729619,
-      "learning_rate": 0.00019999559735261162,
-      "loss": 46.0,
-      "step": 18558
-    },
-    {
-      "epoch": 2.9887676637545795,
-      "grad_norm": 0.008416905999183655,
-      "learning_rate": 0.00019999559687787215,
-      "loss": 46.0,
-      "step": 18559
-    },
-    {
-      "epoch": 2.988928700833367,
-      "grad_norm": 0.0017323879292234778,
-      "learning_rate": 0.0001999955964031071,
-      "loss": 46.0,
-      "step": 18560
-    },
-    {
-      "epoch": 2.9890897379121544,
-      "grad_norm": 0.00040783046279102564,
-      "learning_rate": 0.00019999559592831644,
-      "loss": 46.0,
-      "step": 18561
-    },
-    {
-      "epoch": 2.989250774990942,
-      "grad_norm": 0.005353429354727268,
-      "learning_rate": 0.00019999559545350019,
-      "loss": 46.0,
-      "step": 18562
-    },
-    {
-      "epoch": 2.9894118120697293,
-      "grad_norm": 0.00431845523416996,
-      "learning_rate": 0.00019999559497865837,
-      "loss": 46.0,
-      "step": 18563
-    },
-    {
-      "epoch": 2.9895728491485163,
-      "grad_norm": 0.0011112710926681757,
-      "learning_rate": 0.00019999559450379093,
-      "loss": 46.0,
-      "step": 18564
-    },
-    {
-      "epoch": 2.9897338862273037,
-      "grad_norm": 0.009182791225612164,
-      "learning_rate": 0.00019999559402889791,
-      "loss": 46.0,
-      "step": 18565
-    },
-    {
-      "epoch": 2.989894923306091,
-      "grad_norm": 0.0023359821643680334,
-      "learning_rate": 0.0001999955935539793,
-      "loss": 46.0,
-      "step": 18566
-    },
-    {
-      "epoch": 2.9900559603848786,
-      "grad_norm": 0.0009438145789317787,
-      "learning_rate": 0.0001999955930790351,
-      "loss": 46.0,
-      "step": 18567
-    },
-    {
-      "epoch": 2.990216997463666,
-      "grad_norm": 0.005053383763879538,
-      "learning_rate": 0.00019999559260406525,
-      "loss": 46.0,
-      "step": 18568
-    },
-    {
-      "epoch": 2.9903780345424535,
-      "grad_norm": 0.004160483833402395,
-      "learning_rate": 0.00019999559212906988,
-      "loss": 46.0,
-      "step": 18569
-    },
-    {
-      "epoch": 2.9905390716212406,
-      "grad_norm": 0.0017807877156883478,
-      "learning_rate": 0.0001999955916540489,
-      "loss": 46.0,
-      "step": 18570
-    },
-    {
-      "epoch": 2.990700108700028,
-      "grad_norm": 0.0013404268538579345,
-      "learning_rate": 0.0001999955911790023,
-      "loss": 46.0,
-      "step": 18571
-    },
-    {
-      "epoch": 2.9908611457788155,
-      "grad_norm": 0.0020173233933746815,
-      "learning_rate": 0.00019999559070393012,
-      "loss": 46.0,
-      "step": 18572
-    },
-    {
-      "epoch": 2.991022182857603,
-      "grad_norm": 0.006248829886317253,
-      "learning_rate": 0.00019999559022883237,
-      "loss": 46.0,
-      "step": 18573
-    },
-    {
-      "epoch": 2.9911832199363904,
-      "grad_norm": 0.008519444614648819,
-      "learning_rate": 0.000199995589753709,
-      "loss": 46.0,
-      "step": 18574
-    },
-    {
-      "epoch": 2.991344257015178,
-      "grad_norm": 0.001395870465785265,
-      "learning_rate": 0.00019999558927856002,
-      "loss": 46.0,
-      "step": 18575
-    },
-    {
-      "epoch": 2.9915052940939653,
-      "grad_norm": 0.0026189512573182583,
-      "learning_rate": 0.00019999558880338548,
-      "loss": 46.0,
-      "step": 18576
-    },
-    {
-      "epoch": 2.9916663311727527,
-      "grad_norm": 0.005446328781545162,
-      "learning_rate": 0.00019999558832818533,
-      "loss": 46.0,
-      "step": 18577
-    },
-    {
-      "epoch": 2.99182736825154,
-      "grad_norm": 0.0017989056650549173,
-      "learning_rate": 0.00019999558785295957,
-      "loss": 46.0,
-      "step": 18578
-    },
-    {
-      "epoch": 2.991988405330327,
-      "grad_norm": 0.003046183381229639,
-      "learning_rate": 0.00019999558737770822,
-      "loss": 46.0,
-      "step": 18579
-    },
-    {
-      "epoch": 2.9921494424091146,
-      "grad_norm": 0.0068048653192818165,
-      "learning_rate": 0.0001999955869024313,
-      "loss": 46.0,
-      "step": 18580
-    },
-    {
-      "epoch": 2.992310479487902,
-      "grad_norm": 0.0014451493043452501,
-      "learning_rate": 0.0001999955864271288,
-      "loss": 46.0,
-      "step": 18581
-    },
-    {
-      "epoch": 2.9924715165666895,
-      "grad_norm": 0.006651455070823431,
-      "learning_rate": 0.00019999558595180066,
-      "loss": 46.0,
-      "step": 18582
-    },
-    {
-      "epoch": 2.992632553645477,
-      "grad_norm": 0.005244353320449591,
-      "learning_rate": 0.00019999558547644696,
-      "loss": 46.0,
-      "step": 18583
-    },
-    {
-      "epoch": 2.992793590724264,
-      "grad_norm": 0.0014879959635436535,
-      "learning_rate": 0.00019999558500106765,
-      "loss": 46.0,
-      "step": 18584
-    },
-    {
-      "epoch": 2.9929546278030514,
-      "grad_norm": 0.0012022884329780936,
-      "learning_rate": 0.00019999558452566275,
-      "loss": 46.0,
-      "step": 18585
-    },
-    {
-      "epoch": 2.993115664881839,
-      "grad_norm": 0.0038021320942789316,
-      "learning_rate": 0.00019999558405023226,
-      "loss": 46.0,
-      "step": 18586
-    },
-    {
-      "epoch": 2.9932767019606263,
-      "grad_norm": 0.0038193119689822197,
-      "learning_rate": 0.00019999558357477616,
-      "loss": 46.0,
-      "step": 18587
-    },
-    {
-      "epoch": 2.993437739039414,
-      "grad_norm": 0.002903102897107601,
-      "learning_rate": 0.0001999955830992945,
-      "loss": 46.0,
-      "step": 18588
-    },
-    {
-      "epoch": 2.9935987761182012,
-      "grad_norm": 0.004516868386417627,
-      "learning_rate": 0.00019999558262378723,
-      "loss": 46.0,
-      "step": 18589
-    },
-    {
-      "epoch": 2.9937598131969887,
-      "grad_norm": 0.0005320445052348077,
-      "learning_rate": 0.00019999558214825436,
-      "loss": 46.0,
-      "step": 18590
-    },
-    {
-      "epoch": 2.993920850275776,
-      "grad_norm": 0.002455269917845726,
-      "learning_rate": 0.0001999955816726959,
-      "loss": 46.0,
-      "step": 18591
-    },
-    {
-      "epoch": 2.9940818873545636,
-      "grad_norm": 0.0030678401235491037,
-      "learning_rate": 0.00019999558119711183,
-      "loss": 46.0,
-      "step": 18592
-    },
-    {
-      "epoch": 2.994242924433351,
-      "grad_norm": 0.0007584202685393393,
-      "learning_rate": 0.0001999955807215022,
-      "loss": 46.0,
-      "step": 18593
-    },
-    {
-      "epoch": 2.994403961512138,
-      "grad_norm": 0.0019652366172522306,
-      "learning_rate": 0.00019999558024586696,
-      "loss": 46.0,
-      "step": 18594
-    },
-    {
-      "epoch": 2.9945649985909255,
-      "grad_norm": 0.002631678245961666,
-      "learning_rate": 0.00019999557977020614,
-      "loss": 46.0,
-      "step": 18595
-    },
-    {
-      "epoch": 2.994726035669713,
-      "grad_norm": 0.0021489951759576797,
-      "learning_rate": 0.00019999557929451967,
-      "loss": 46.0,
-      "step": 18596
-    },
-    {
-      "epoch": 2.9948870727485004,
-      "grad_norm": 0.0026354810688644648,
-      "learning_rate": 0.00019999557881880765,
-      "loss": 46.0,
-      "step": 18597
-    },
-    {
-      "epoch": 2.995048109827288,
-      "grad_norm": 0.0016480355989187956,
-      "learning_rate": 0.00019999557834307006,
-      "loss": 46.0,
-      "step": 18598
-    },
-    {
-      "epoch": 2.995209146906075,
-      "grad_norm": 0.0025088803377002478,
-      "learning_rate": 0.00019999557786730683,
-      "loss": 46.0,
-      "step": 18599
-    },
-    {
-      "epoch": 2.9953701839848623,
-      "grad_norm": 0.0015875728568062186,
-      "learning_rate": 0.00019999557739151805,
-      "loss": 46.0,
-      "step": 18600
-    },
-    {
-      "epoch": 2.9955312210636498,
-      "grad_norm": 0.002014116384088993,
-      "learning_rate": 0.00019999557691570364,
-      "loss": 46.0,
-      "step": 18601
-    },
-    {
-      "epoch": 2.9956922581424372,
-      "grad_norm": 0.002914705779403448,
-      "learning_rate": 0.00019999557643986363,
-      "loss": 46.0,
-      "step": 18602
-    },
-    {
-      "epoch": 2.9958532952212247,
-      "grad_norm": 0.005247557535767555,
-      "learning_rate": 0.00019999557596399808,
-      "loss": 46.0,
-      "step": 18603
-    },
-    {
-      "epoch": 2.996014332300012,
-      "grad_norm": 0.0063356030732393265,
-      "learning_rate": 0.0001999955754881069,
-      "loss": 46.0,
-      "step": 18604
-    },
-    {
-      "epoch": 2.9961753693787996,
-      "grad_norm": 0.0030614477582275867,
-      "learning_rate": 0.0001999955750121901,
-      "loss": 46.0,
-      "step": 18605
-    },
-    {
-      "epoch": 2.996336406457587,
-      "grad_norm": 0.015649404376745224,
-      "learning_rate": 0.00019999557453624775,
-      "loss": 46.0,
-      "step": 18606
-    },
-    {
-      "epoch": 2.9964974435363745,
-      "grad_norm": 0.00719917006790638,
-      "learning_rate": 0.00019999557406027977,
-      "loss": 46.0,
-      "step": 18607
-    },
-    {
-      "epoch": 2.9966584806151615,
-      "grad_norm": 0.012719051912426949,
-      "learning_rate": 0.00019999557358428623,
-      "loss": 46.0,
-      "step": 18608
-    },
-    {
-      "epoch": 2.996819517693949,
-      "grad_norm": 0.0011661925818771124,
-      "learning_rate": 0.0001999955731082671,
-      "loss": 46.0,
-      "step": 18609
-    },
-    {
-      "epoch": 2.9969805547727364,
-      "grad_norm": 0.001103631337173283,
-      "learning_rate": 0.00019999557263222233,
-      "loss": 46.0,
-      "step": 18610
-    },
-    {
-      "epoch": 2.997141591851524,
-      "grad_norm": 0.0011683221673592925,
-      "learning_rate": 0.00019999557215615198,
-      "loss": 46.0,
-      "step": 18611
-    },
-    {
-      "epoch": 2.9973026289303113,
-      "grad_norm": 0.0024710465222597122,
-      "learning_rate": 0.00019999557168005606,
-      "loss": 46.0,
-      "step": 18612
-    },
-    {
-      "epoch": 2.9974636660090987,
-      "grad_norm": 0.013376552611589432,
-      "learning_rate": 0.00019999557120393456,
-      "loss": 46.0,
-      "step": 18613
-    },
-    {
-      "epoch": 2.9976247030878858,
-      "grad_norm": 0.0036130868829786777,
-      "learning_rate": 0.00019999557072778742,
-      "loss": 46.0,
-      "step": 18614
-    },
-    {
-      "epoch": 2.997785740166673,
-      "grad_norm": 0.009003588929772377,
-      "learning_rate": 0.00019999557025161474,
-      "loss": 46.0,
-      "step": 18615
-    },
-    {
-      "epoch": 2.9979467772454607,
-      "grad_norm": 0.007207829505205154,
-      "learning_rate": 0.00019999556977541642,
-      "loss": 46.0,
-      "step": 18616
-    },
-    {
-      "epoch": 2.998107814324248,
-      "grad_norm": 0.008652038872241974,
-      "learning_rate": 0.00019999556929919252,
-      "loss": 46.0,
-      "step": 18617
-    },
-    {
-      "epoch": 2.9982688514030356,
-      "grad_norm": 0.004082380793988705,
-      "learning_rate": 0.00019999556882294302,
-      "loss": 46.0,
-      "step": 18618
-    },
-    {
-      "epoch": 2.998429888481823,
-      "grad_norm": 0.0042772721499204636,
-      "learning_rate": 0.00019999556834666794,
-      "loss": 46.0,
-      "step": 18619
-    },
-    {
-      "epoch": 2.9985909255606105,
-      "grad_norm": 0.006474708206951618,
-      "learning_rate": 0.00019999556787036725,
-      "loss": 46.0,
-      "step": 18620
-    },
-    {
-      "epoch": 2.998751962639398,
-      "grad_norm": 0.014925863593816757,
-      "learning_rate": 0.000199995567394041,
-      "loss": 46.0,
-      "step": 18621
-    },
-    {
-      "epoch": 2.9989129997181854,
-      "grad_norm": 0.011634370312094688,
-      "learning_rate": 0.00019999556691768913,
-      "loss": 46.0,
-      "step": 18622
-    },
-    {
-      "epoch": 2.9990740367969724,
-      "grad_norm": 0.0018507307395339012,
-      "learning_rate": 0.00019999556644131165,
-      "loss": 46.0,
-      "step": 18623
-    },
-    {
-      "epoch": 2.99923507387576,
-      "grad_norm": 0.0019304245943203568,
-      "learning_rate": 0.0001999955659649086,
-      "loss": 46.0,
-      "step": 18624
-    },
-    {
-      "epoch": 2.9993961109545473,
-      "grad_norm": 0.0014087522868067026,
-      "learning_rate": 0.00019999556548847997,
-      "loss": 46.0,
-      "step": 18625
-    },
-    {
-      "epoch": 2.9995571480333347,
-      "grad_norm": 0.01107286848127842,
-      "learning_rate": 0.00019999556501202573,
-      "loss": 46.0,
-      "step": 18626
-    },
-    {
-      "epoch": 2.999718185112122,
-      "grad_norm": 0.0028081524651497602,
-      "learning_rate": 0.0001999955645355459,
-      "loss": 46.0,
-      "step": 18627
-    },
-    {
-      "epoch": 2.999879222190909,
-      "grad_norm": 0.005611950531601906,
-      "learning_rate": 0.00019999556405904048,
-      "loss": 46.0,
-      "step": 18628
-    },
-    {
-      "epoch": 2.999879222190909,
-      "eval_loss": 11.5,
-      "eval_runtime": 14.5744,
-      "eval_samples_per_second": 179.424,
-      "eval_steps_per_second": 89.746,
-      "step": 18628
-    },
-    {
-      "epoch": 3.0000805185393937,
-      "grad_norm": 0.001445271191187203,
-      "learning_rate": 0.00019999556358250945,
-      "loss": 46.0,
-      "step": 18629
-    },
-    {
-      "epoch": 3.000241555618181,
-      "grad_norm": 0.015089699998497963,
-      "learning_rate": 0.00019999556310595283,
-      "loss": 46.0,
-      "step": 18630
-    },
-    {
-      "epoch": 3.0004025926969686,
-      "grad_norm": 0.008645368739962578,
-      "learning_rate": 0.00019999556262937062,
-      "loss": 46.0,
-      "step": 18631
-    },
-    {
-      "epoch": 3.000563629775756,
-      "grad_norm": 0.009409420192241669,
-      "learning_rate": 0.0001999955621527628,
-      "loss": 46.0,
-      "step": 18632
-    },
-    {
-      "epoch": 3.000724666854543,
-      "grad_norm": 0.0022755623795092106,
-      "learning_rate": 0.0001999955616761294,
-      "loss": 46.0,
-      "step": 18633
-    },
-    {
-      "epoch": 3.0008857039333305,
-      "grad_norm": 0.003756490768864751,
-      "learning_rate": 0.00019999556119947042,
-      "loss": 46.0,
-      "step": 18634
-    },
-    {
-      "epoch": 3.001046741012118,
-      "grad_norm": 0.00836783368140459,
-      "learning_rate": 0.00019999556072278581,
-      "loss": 46.0,
-      "step": 18635
-    },
-    {
-      "epoch": 3.0012077780909054,
-      "grad_norm": 0.008819560520350933,
-      "learning_rate": 0.00019999556024607565,
-      "loss": 46.0,
-      "step": 18636
-    },
-    {
-      "epoch": 3.001368815169693,
-      "grad_norm": 0.0006767732556909323,
-      "learning_rate": 0.0001999955597693399,
-      "loss": 46.0,
-      "step": 18637
-    },
-    {
-      "epoch": 3.0015298522484803,
-      "grad_norm": 0.008921225555241108,
-      "learning_rate": 0.00019999555929257852,
-      "loss": 46.0,
-      "step": 18638
-    },
-    {
-      "epoch": 3.001690889327268,
-      "grad_norm": 0.003402446396648884,
-      "learning_rate": 0.00019999555881579159,
-      "loss": 46.0,
-      "step": 18639
-    },
-    {
-      "epoch": 3.001851926406055,
-      "grad_norm": 0.0027959682047367096,
-      "learning_rate": 0.000199995558338979,
-      "loss": 46.0,
-      "step": 18640
-    },
-    {
-      "epoch": 3.0020129634848423,
-      "grad_norm": 0.0032038178760558367,
-      "learning_rate": 0.00019999555786214085,
-      "loss": 46.0,
-      "step": 18641
-    },
-    {
-      "epoch": 3.0021740005636297,
-      "grad_norm": 0.001509590889327228,
-      "learning_rate": 0.0001999955573852771,
-      "loss": 46.0,
-      "step": 18642
-    },
-    {
-      "epoch": 3.002335037642417,
-      "grad_norm": 0.0013796513667330146,
-      "learning_rate": 0.00019999555690838777,
-      "loss": 46.0,
-      "step": 18643
-    },
-    {
-      "epoch": 3.0024960747212046,
-      "grad_norm": 0.0012693741591647267,
-      "learning_rate": 0.00019999555643147285,
-      "loss": 46.0,
-      "step": 18644
-    },
-    {
-      "epoch": 3.002657111799992,
-      "grad_norm": 0.0009839271660894156,
-      "learning_rate": 0.00019999555595453234,
-      "loss": 46.0,
-      "step": 18645
-    },
-    {
-      "epoch": 3.0028181488787795,
-      "grad_norm": 0.00999511405825615,
-      "learning_rate": 0.00019999555547756622,
-      "loss": 46.0,
-      "step": 18646
-    },
-    {
-      "epoch": 3.0029791859575665,
-      "grad_norm": 0.00944233126938343,
-      "learning_rate": 0.0001999955550005745,
-      "loss": 46.0,
-      "step": 18647
-    },
-    {
-      "epoch": 3.003140223036354,
-      "grad_norm": 0.002734616631641984,
-      "learning_rate": 0.00019999555452355718,
-      "loss": 46.0,
-      "step": 18648
-    },
-    {
-      "epoch": 3.0033012601151414,
-      "grad_norm": 0.024181615561246872,
-      "learning_rate": 0.0001999955540465143,
-      "loss": 46.0,
-      "step": 18649
-    },
-    {
-      "epoch": 3.003462297193929,
-      "grad_norm": 0.0066743940114974976,
-      "learning_rate": 0.0001999955535694458,
-      "loss": 46.0,
-      "step": 18650
-    },
-    {
-      "epoch": 3.0036233342727163,
-      "grad_norm": 0.0005242424085736275,
-      "learning_rate": 0.00019999555309235172,
-      "loss": 46.0,
-      "step": 18651
-    },
-    {
-      "epoch": 3.003784371351504,
-      "grad_norm": 0.007995721884071827,
-      "learning_rate": 0.00019999555261523204,
-      "loss": 46.0,
-      "step": 18652
-    },
-    {
-      "epoch": 3.0039454084302912,
-      "grad_norm": 0.0018966497154906392,
-      "learning_rate": 0.00019999555213808678,
-      "loss": 46.0,
-      "step": 18653
-    },
-    {
-      "epoch": 3.0041064455090787,
-      "grad_norm": 0.0017775587039068341,
-      "learning_rate": 0.0001999955516609159,
-      "loss": 46.0,
-      "step": 18654
-    },
-    {
-      "epoch": 3.0042674825878657,
-      "grad_norm": 0.005550839938223362,
-      "learning_rate": 0.00019999555118371945,
-      "loss": 46.0,
-      "step": 18655
-    },
-    {
-      "epoch": 3.004428519666653,
-      "grad_norm": 0.0010232527274638414,
-      "learning_rate": 0.0001999955507064974,
-      "loss": 46.0,
-      "step": 18656
-    },
-    {
-      "epoch": 3.0045895567454406,
-      "grad_norm": 0.0020085815340280533,
-      "learning_rate": 0.00019999555022924973,
-      "loss": 46.0,
-      "step": 18657
-    },
-    {
-      "epoch": 3.004750593824228,
-      "grad_norm": 0.003025443758815527,
-      "learning_rate": 0.00019999554975197648,
-      "loss": 46.0,
-      "step": 18658
-    },
-    {
-      "epoch": 3.0049116309030155,
-      "grad_norm": 0.004842628259211779,
-      "learning_rate": 0.00019999554927467767,
-      "loss": 46.0,
-      "step": 18659
-    },
-    {
-      "epoch": 3.005072667981803,
-      "grad_norm": 0.005136552266776562,
-      "learning_rate": 0.00019999554879735325,
-      "loss": 46.0,
-      "step": 18660
-    },
-    {
-      "epoch": 3.0052337050605904,
-      "grad_norm": 0.002330041490495205,
-      "learning_rate": 0.0001999955483200032,
-      "loss": 46.0,
-      "step": 18661
-    },
-    {
-      "epoch": 3.0053947421393774,
-      "grad_norm": 0.0011494910577312112,
-      "learning_rate": 0.00019999554784262758,
-      "loss": 46.0,
-      "step": 18662
-    },
-    {
-      "epoch": 3.005555779218165,
-      "grad_norm": 0.0015137909213081002,
-      "learning_rate": 0.0001999955473652264,
-      "loss": 46.0,
-      "step": 18663
-    },
-    {
-      "epoch": 3.0057168162969523,
-      "grad_norm": 0.0007970886072143912,
-      "learning_rate": 0.0001999955468877996,
-      "loss": 46.0,
-      "step": 18664
-    },
-    {
-      "epoch": 3.0058778533757398,
-      "grad_norm": 0.006009059492498636,
-      "learning_rate": 0.0001999955464103472,
-      "loss": 46.0,
-      "step": 18665
-    },
-    {
-      "epoch": 3.006038890454527,
-      "grad_norm": 0.0021226785611361265,
-      "learning_rate": 0.0001999955459328692,
-      "loss": 46.0,
-      "step": 18666
-    },
-    {
-      "epoch": 3.0061999275333147,
-      "grad_norm": 0.0033282616641372442,
-      "learning_rate": 0.00019999554545536562,
-      "loss": 46.0,
-      "step": 18667
-    },
-    {
-      "epoch": 3.006360964612102,
-      "grad_norm": 0.00737522728741169,
-      "learning_rate": 0.00019999554497783644,
-      "loss": 46.0,
-      "step": 18668
-    },
-    {
-      "epoch": 3.006522001690889,
-      "grad_norm": 0.0076141394674777985,
-      "learning_rate": 0.00019999554450028168,
-      "loss": 46.0,
-      "step": 18669
-    },
-    {
-      "epoch": 3.0066830387696766,
-      "grad_norm": 0.003813562449067831,
-      "learning_rate": 0.0001999955440227013,
-      "loss": 46.0,
-      "step": 18670
-    },
-    {
-      "epoch": 3.006844075848464,
-      "grad_norm": 0.006963587366044521,
-      "learning_rate": 0.00019999554354509537,
-      "loss": 46.0,
-      "step": 18671
-    },
-    {
-      "epoch": 3.0070051129272515,
-      "grad_norm": 0.010456070303916931,
-      "learning_rate": 0.00019999554306746379,
-      "loss": 46.0,
-      "step": 18672
-    },
-    {
-      "epoch": 3.007166150006039,
-      "grad_norm": 0.0037042913027107716,
-      "learning_rate": 0.00019999554258980667,
-      "loss": 46.0,
-      "step": 18673
-    },
-    {
-      "epoch": 3.0073271870848264,
-      "grad_norm": 0.0019664722494781017,
-      "learning_rate": 0.00019999554211212392,
-      "loss": 46.0,
-      "step": 18674
-    },
-    {
-      "epoch": 3.007488224163614,
-      "grad_norm": 0.0022088796831667423,
-      "learning_rate": 0.00019999554163441555,
-      "loss": 46.0,
-      "step": 18675
-    },
-    {
-      "epoch": 3.0076492612424013,
-      "grad_norm": 0.00705556757748127,
-      "learning_rate": 0.00019999554115668165,
-      "loss": 46.0,
-      "step": 18676
-    },
-    {
-      "epoch": 3.0078102983211883,
-      "grad_norm": 0.001226063002832234,
-      "learning_rate": 0.0001999955406789221,
-      "loss": 46.0,
-      "step": 18677
-    },
-    {
-      "epoch": 3.0079713353999757,
-      "grad_norm": 0.0009407844627276063,
-      "learning_rate": 0.000199995540201137,
-      "loss": 46.0,
-      "step": 18678
-    },
-    {
-      "epoch": 3.008132372478763,
-      "grad_norm": 0.0007975497283041477,
-      "learning_rate": 0.0001999955397233263,
-      "loss": 46.0,
-      "step": 18679
-    },
-    {
-      "epoch": 3.0082934095575506,
-      "grad_norm": 0.004013837780803442,
-      "learning_rate": 0.00019999553924549,
-      "loss": 46.0,
-      "step": 18680
-    },
-    {
-      "epoch": 3.008454446636338,
-      "grad_norm": 0.0006178237963467836,
-      "learning_rate": 0.0001999955387676281,
-      "loss": 46.0,
-      "step": 18681
-    },
-    {
-      "epoch": 3.0086154837151255,
-      "grad_norm": 0.0010368465445935726,
-      "learning_rate": 0.00019999553828974062,
-      "loss": 46.0,
-      "step": 18682
-    },
-    {
-      "epoch": 3.008776520793913,
-      "grad_norm": 0.00988208968192339,
-      "learning_rate": 0.00019999553781182753,
-      "loss": 46.0,
-      "step": 18683
-    },
-    {
-      "epoch": 3.0089375578727,
-      "grad_norm": 0.005435602739453316,
-      "learning_rate": 0.00019999553733388885,
-      "loss": 46.0,
-      "step": 18684
-    },
-    {
-      "epoch": 3.0090985949514875,
-      "grad_norm": 0.005490770097821951,
-      "learning_rate": 0.00019999553685592458,
-      "loss": 46.0,
-      "step": 18685
-    },
-    {
-      "epoch": 3.009259632030275,
-      "grad_norm": 0.009492075070738792,
-      "learning_rate": 0.0001999955363779347,
-      "loss": 46.0,
-      "step": 18686
-    },
-    {
-      "epoch": 3.0094206691090624,
-      "grad_norm": 0.0025694286450743675,
-      "learning_rate": 0.00019999553589991926,
-      "loss": 46.0,
-      "step": 18687
-    },
-    {
-      "epoch": 3.00958170618785,
-      "grad_norm": 0.001257500029169023,
-      "learning_rate": 0.00019999553542187818,
-      "loss": 46.0,
-      "step": 18688
-    },
-    {
-      "epoch": 3.0097427432666373,
-      "grad_norm": 0.020698247477412224,
-      "learning_rate": 0.00019999553494381156,
-      "loss": 46.0,
-      "step": 18689
-    },
-    {
-      "epoch": 3.0099037803454247,
-      "grad_norm": 0.002571235643699765,
-      "learning_rate": 0.0001999955344657193,
-      "loss": 46.0,
-      "step": 18690
-    },
-    {
-      "epoch": 3.0100648174242117,
-      "grad_norm": 0.001914593973197043,
-      "learning_rate": 0.0001999955339876015,
-      "loss": 46.0,
-      "step": 18691
-    },
-    {
-      "epoch": 3.010225854502999,
-      "grad_norm": 0.007552071940153837,
-      "learning_rate": 0.00019999553350945806,
-      "loss": 46.0,
-      "step": 18692
-    },
-    {
-      "epoch": 3.0103868915817866,
-      "grad_norm": 0.002395815448835492,
-      "learning_rate": 0.00019999553303128904,
-      "loss": 46.0,
-      "step": 18693
-    },
-    {
-      "epoch": 3.010547928660574,
-      "grad_norm": 0.002316900063306093,
-      "learning_rate": 0.0001999955325530944,
-      "loss": 46.0,
-      "step": 18694
-    },
-    {
-      "epoch": 3.0107089657393615,
-      "grad_norm": 0.000937719305511564,
-      "learning_rate": 0.00019999553207487422,
-      "loss": 46.0,
-      "step": 18695
-    },
-    {
-      "epoch": 3.010870002818149,
-      "grad_norm": 0.003263594349846244,
-      "learning_rate": 0.0001999955315966284,
-      "loss": 46.0,
-      "step": 18696
-    },
-    {
-      "epoch": 3.0110310398969364,
-      "grad_norm": 0.0007541762897744775,
-      "learning_rate": 0.00019999553111835702,
-      "loss": 46.0,
-      "step": 18697
-    },
-    {
-      "epoch": 3.0111920769757234,
-      "grad_norm": 0.002288545947521925,
-      "learning_rate": 0.00019999553064006,
-      "loss": 46.0,
-      "step": 18698
-    },
-    {
-      "epoch": 3.011353114054511,
-      "grad_norm": 0.00357170682400465,
-      "learning_rate": 0.0001999955301617374,
-      "loss": 46.0,
-      "step": 18699
-    },
-    {
-      "epoch": 3.0115141511332983,
-      "grad_norm": 0.0012144313659518957,
-      "learning_rate": 0.00019999552968338926,
-      "loss": 46.0,
-      "step": 18700
-    },
-    {
-      "epoch": 3.011675188212086,
-      "grad_norm": 0.001376515137962997,
-      "learning_rate": 0.0001999955292050155,
-      "loss": 46.0,
-      "step": 18701
-    },
-    {
-      "epoch": 3.0118362252908732,
-      "grad_norm": 0.001987120136618614,
-      "learning_rate": 0.0001999955287266161,
-      "loss": 46.0,
-      "step": 18702
-    },
-    {
-      "epoch": 3.0119972623696607,
-      "grad_norm": 0.0013515574391931295,
-      "learning_rate": 0.00019999552824819116,
-      "loss": 46.0,
-      "step": 18703
-    },
-    {
-      "epoch": 3.012158299448448,
-      "grad_norm": 0.005974846426397562,
-      "learning_rate": 0.0001999955277697406,
-      "loss": 46.0,
-      "step": 18704
-    },
-    {
-      "epoch": 3.0123193365272356,
-      "grad_norm": 0.0016517681069672108,
-      "learning_rate": 0.00019999552729126446,
-      "loss": 46.0,
-      "step": 18705
-    },
-    {
-      "epoch": 3.0124803736060226,
-      "grad_norm": 0.0008199273725040257,
-      "learning_rate": 0.00019999552681276272,
-      "loss": 46.0,
-      "step": 18706
-    },
-    {
-      "epoch": 3.01264141068481,
-      "grad_norm": 0.0004188912862446159,
-      "learning_rate": 0.0001999955263342354,
-      "loss": 46.0,
-      "step": 18707
-    },
-    {
-      "epoch": 3.0128024477635975,
-      "grad_norm": 0.0023702443577349186,
-      "learning_rate": 0.00019999552585568247,
-      "loss": 46.0,
-      "step": 18708
-    },
-    {
-      "epoch": 3.012963484842385,
-      "grad_norm": 0.004834124818444252,
-      "learning_rate": 0.00019999552537710392,
-      "loss": 46.0,
-      "step": 18709
-    },
-    {
-      "epoch": 3.0131245219211724,
-      "grad_norm": 0.0008324910886585712,
-      "learning_rate": 0.00019999552489849984,
-      "loss": 46.0,
-      "step": 18710
-    },
-    {
-      "epoch": 3.01328555899996,
-      "grad_norm": 0.0036300348583608866,
-      "learning_rate": 0.00019999552441987012,
-      "loss": 46.0,
-      "step": 18711
-    },
-    {
-      "epoch": 3.0134465960787473,
-      "grad_norm": 0.0016632035840302706,
-      "learning_rate": 0.0001999955239412148,
-      "loss": 46.0,
-      "step": 18712
-    },
-    {
-      "epoch": 3.0136076331575343,
-      "grad_norm": 0.0021418659016489983,
-      "learning_rate": 0.0001999955234625339,
-      "loss": 46.0,
-      "step": 18713
-    },
-    {
-      "epoch": 3.013768670236322,
-      "grad_norm": 0.0009746663854457438,
-      "learning_rate": 0.00019999552298382743,
-      "loss": 46.0,
-      "step": 18714
-    },
-    {
-      "epoch": 3.0139297073151092,
-      "grad_norm": 0.001135788974352181,
-      "learning_rate": 0.00019999552250509536,
-      "loss": 46.0,
-      "step": 18715
-    },
-    {
-      "epoch": 3.0140907443938967,
-      "grad_norm": 0.0006664918037131429,
-      "learning_rate": 0.0001999955220263377,
-      "loss": 46.0,
-      "step": 18716
-    },
-    {
-      "epoch": 3.014251781472684,
-      "grad_norm": 0.00045535352546721697,
-      "learning_rate": 0.0001999955215475544,
-      "loss": 46.0,
-      "step": 18717
-    },
-    {
-      "epoch": 3.0144128185514716,
-      "grad_norm": 0.005772685632109642,
-      "learning_rate": 0.00019999552106874556,
-      "loss": 46.0,
-      "step": 18718
-    },
-    {
-      "epoch": 3.014573855630259,
-      "grad_norm": 0.0056907483376562595,
-      "learning_rate": 0.00019999552058991106,
-      "loss": 46.0,
-      "step": 18719
-    },
-    {
-      "epoch": 3.014734892709046,
-      "grad_norm": 0.0018697796622291207,
-      "learning_rate": 0.00019999552011105103,
-      "loss": 46.0,
-      "step": 18720
-    },
-    {
-      "epoch": 3.0148959297878335,
-      "grad_norm": 0.0012618767796084285,
-      "learning_rate": 0.00019999551963216538,
-      "loss": 46.0,
-      "step": 18721
-    },
-    {
-      "epoch": 3.015056966866621,
-      "grad_norm": 0.004601877648383379,
-      "learning_rate": 0.00019999551915325414,
-      "loss": 46.0,
-      "step": 18722
-    },
-    {
-      "epoch": 3.0152180039454084,
-      "grad_norm": 0.0019418438896536827,
-      "learning_rate": 0.0001999955186743173,
-      "loss": 46.0,
-      "step": 18723
-    },
-    {
-      "epoch": 3.015379041024196,
-      "grad_norm": 0.002044722205027938,
-      "learning_rate": 0.00019999551819535486,
-      "loss": 46.0,
-      "step": 18724
-    },
-    {
-      "epoch": 3.0155400781029833,
-      "grad_norm": 0.0032433385495096445,
-      "learning_rate": 0.00019999551771636686,
-      "loss": 46.0,
-      "step": 18725
-    },
-    {
-      "epoch": 3.0157011151817708,
-      "grad_norm": 0.004019223619252443,
-      "learning_rate": 0.00019999551723735325,
-      "loss": 46.0,
-      "step": 18726
-    },
-    {
-      "epoch": 3.015862152260558,
-      "grad_norm": 0.0028463841881603003,
-      "learning_rate": 0.00019999551675831402,
-      "loss": 46.0,
-      "step": 18727
-    },
-    {
-      "epoch": 3.016023189339345,
-      "grad_norm": 0.0025055555161088705,
-      "learning_rate": 0.00019999551627924924,
-      "loss": 46.0,
-      "step": 18728
-    },
-    {
-      "epoch": 3.0161842264181327,
-      "grad_norm": 0.0006456353585235775,
-      "learning_rate": 0.00019999551580015884,
-      "loss": 46.0,
-      "step": 18729
-    },
-    {
-      "epoch": 3.01634526349692,
-      "grad_norm": 0.0033867210149765015,
-      "learning_rate": 0.00019999551532104285,
-      "loss": 46.0,
-      "step": 18730
-    },
-    {
-      "epoch": 3.0165063005757076,
-      "grad_norm": 0.0036401154939085245,
-      "learning_rate": 0.00019999551484190127,
-      "loss": 46.0,
-      "step": 18731
-    },
-    {
-      "epoch": 3.016667337654495,
-      "grad_norm": 0.006417823024094105,
-      "learning_rate": 0.00019999551436273409,
-      "loss": 46.0,
-      "step": 18732
-    },
-    {
-      "epoch": 3.0168283747332825,
-      "grad_norm": 0.002617353107780218,
-      "learning_rate": 0.00019999551388354134,
-      "loss": 46.0,
-      "step": 18733
-    },
-    {
-      "epoch": 3.01698941181207,
-      "grad_norm": 0.002480825176462531,
-      "learning_rate": 0.00019999551340432297,
-      "loss": 46.0,
-      "step": 18734
-    },
-    {
-      "epoch": 3.017150448890857,
-      "grad_norm": 0.0038271714001893997,
-      "learning_rate": 0.000199995512925079,
-      "loss": 46.0,
-      "step": 18735
-    },
-    {
-      "epoch": 3.0173114859696444,
-      "grad_norm": 0.019938984885811806,
-      "learning_rate": 0.00019999551244580946,
-      "loss": 46.0,
-      "step": 18736
-    },
-    {
-      "epoch": 3.017472523048432,
-      "grad_norm": 0.00265054265037179,
-      "learning_rate": 0.0001999955119665143,
-      "loss": 46.0,
-      "step": 18737
-    },
-    {
-      "epoch": 3.0176335601272193,
-      "grad_norm": 0.007168011274188757,
-      "learning_rate": 0.00019999551148719357,
-      "loss": 46.0,
-      "step": 18738
-    },
-    {
-      "epoch": 3.0177945972060067,
-      "grad_norm": 0.0014649019576609135,
-      "learning_rate": 0.00019999551100784724,
-      "loss": 46.0,
-      "step": 18739
-    },
-    {
-      "epoch": 3.017955634284794,
-      "grad_norm": 0.003237906377762556,
-      "learning_rate": 0.0001999955105284753,
-      "loss": 46.0,
-      "step": 18740
-    },
-    {
-      "epoch": 3.0181166713635816,
-      "grad_norm": 0.0016598334768787026,
-      "learning_rate": 0.0001999955100490778,
-      "loss": 46.0,
-      "step": 18741
-    },
-    {
-      "epoch": 3.0182777084423686,
-      "grad_norm": 0.0014350726269185543,
-      "learning_rate": 0.00019999550956965466,
-      "loss": 46.0,
-      "step": 18742
-    },
-    {
-      "epoch": 3.018438745521156,
-      "grad_norm": 0.0027885762974619865,
-      "learning_rate": 0.00019999550909020596,
-      "loss": 46.0,
-      "step": 18743
-    },
-    {
-      "epoch": 3.0185997825999435,
-      "grad_norm": 0.005374143365770578,
-      "learning_rate": 0.00019999550861073167,
-      "loss": 46.0,
-      "step": 18744
-    },
-    {
-      "epoch": 3.018760819678731,
-      "grad_norm": 0.004725174512714148,
-      "learning_rate": 0.0001999955081312318,
-      "loss": 46.0,
-      "step": 18745
-    },
-    {
-      "epoch": 3.0189218567575185,
-      "grad_norm": 0.004258784465491772,
-      "learning_rate": 0.0001999955076517063,
-      "loss": 46.0,
-      "step": 18746
-    },
-    {
-      "epoch": 3.019082893836306,
-      "grad_norm": 0.003756534308195114,
-      "learning_rate": 0.0001999955071721552,
-      "loss": 46.0,
-      "step": 18747
-    },
-    {
-      "epoch": 3.0192439309150934,
-      "grad_norm": 0.0032782878261059523,
-      "learning_rate": 0.00019999550669257853,
-      "loss": 46.0,
-      "step": 18748
-    },
-    {
-      "epoch": 3.019404967993881,
-      "grad_norm": 0.000980779412202537,
-      "learning_rate": 0.00019999550621297628,
-      "loss": 46.0,
-      "step": 18749
-    },
-    {
-      "epoch": 3.019566005072668,
-      "grad_norm": 0.003476217156276107,
-      "learning_rate": 0.00019999550573334839,
-      "loss": 46.0,
-      "step": 18750
-    },
-    {
-      "epoch": 3.0197270421514553,
-      "grad_norm": 0.005282679107040167,
-      "learning_rate": 0.00019999550525369496,
-      "loss": 46.0,
-      "step": 18751
-    },
-    {
-      "epoch": 3.0198880792302427,
-      "grad_norm": 0.007214238867163658,
-      "learning_rate": 0.0001999955047740159,
-      "loss": 46.0,
-      "step": 18752
-    },
-    {
-      "epoch": 3.02004911630903,
-      "grad_norm": 0.0019638959784060717,
-      "learning_rate": 0.00019999550429431124,
-      "loss": 46.0,
-      "step": 18753
-    },
-    {
-      "epoch": 3.0202101533878176,
-      "grad_norm": 0.005872043315321207,
-      "learning_rate": 0.00019999550381458102,
-      "loss": 46.0,
-      "step": 18754
-    },
-    {
-      "epoch": 3.020371190466605,
-      "grad_norm": 0.007495333906263113,
-      "learning_rate": 0.00019999550333482517,
-      "loss": 46.0,
-      "step": 18755
-    },
-    {
-      "epoch": 3.0205322275453925,
-      "grad_norm": 0.006799828726798296,
-      "learning_rate": 0.00019999550285504375,
-      "loss": 46.0,
-      "step": 18756
-    },
-    {
-      "epoch": 3.0206932646241795,
-      "grad_norm": 0.00994163192808628,
-      "learning_rate": 0.00019999550237523672,
-      "loss": 46.0,
-      "step": 18757
-    },
-    {
-      "epoch": 3.020854301702967,
-      "grad_norm": 0.0038626380264759064,
-      "learning_rate": 0.0001999955018954041,
-      "loss": 46.0,
-      "step": 18758
-    },
-    {
-      "epoch": 3.0210153387817544,
-      "grad_norm": 0.004801772069185972,
-      "learning_rate": 0.0001999955014155459,
-      "loss": 46.0,
-      "step": 18759
-    },
-    {
-      "epoch": 3.021176375860542,
-      "grad_norm": 0.001676554442383349,
-      "learning_rate": 0.0001999955009356621,
-      "loss": 46.0,
-      "step": 18760
-    },
-    {
-      "epoch": 3.0213374129393293,
-      "grad_norm": 0.003552692010998726,
-      "learning_rate": 0.0001999955004557527,
-      "loss": 46.0,
-      "step": 18761
-    },
-    {
-      "epoch": 3.021498450018117,
-      "grad_norm": 0.0015632720896974206,
-      "learning_rate": 0.00019999549997581773,
-      "loss": 46.0,
-      "step": 18762
-    },
-    {
-      "epoch": 3.0216594870969042,
-      "grad_norm": 0.007069546729326248,
-      "learning_rate": 0.00019999549949585715,
-      "loss": 46.0,
-      "step": 18763
-    },
-    {
-      "epoch": 3.0218205241756912,
-      "grad_norm": 0.005410053767263889,
-      "learning_rate": 0.00019999549901587095,
-      "loss": 46.0,
-      "step": 18764
-    },
-    {
-      "epoch": 3.0219815612544787,
-      "grad_norm": 0.0020497876685112715,
-      "learning_rate": 0.0001999954985358592,
-      "loss": 46.0,
-      "step": 18765
-    },
-    {
-      "epoch": 3.022142598333266,
-      "grad_norm": 0.0029364845249801874,
-      "learning_rate": 0.00019999549805582183,
-      "loss": 46.0,
-      "step": 18766
-    },
-    {
-      "epoch": 3.0223036354120536,
-      "grad_norm": 0.006781401578336954,
-      "learning_rate": 0.00019999549757575887,
-      "loss": 46.0,
-      "step": 18767
-    },
-    {
-      "epoch": 3.022464672490841,
-      "grad_norm": 0.004405970685184002,
-      "learning_rate": 0.00019999549709567033,
-      "loss": 46.0,
-      "step": 18768
-    },
-    {
-      "epoch": 3.0226257095696285,
-      "grad_norm": 0.003249007510021329,
-      "learning_rate": 0.00019999549661555617,
-      "loss": 46.0,
-      "step": 18769
-    },
-    {
-      "epoch": 3.022786746648416,
-      "grad_norm": 0.0037679567467421293,
-      "learning_rate": 0.00019999549613541643,
-      "loss": 46.0,
-      "step": 18770
-    },
-    {
-      "epoch": 3.0229477837272034,
-      "grad_norm": 0.0030146290082484484,
-      "learning_rate": 0.00019999549565525112,
-      "loss": 46.0,
-      "step": 18771
-    },
-    {
-      "epoch": 3.0231088208059904,
-      "grad_norm": 0.003766910172998905,
-      "learning_rate": 0.00019999549517506017,
-      "loss": 46.0,
-      "step": 18772
-    },
-    {
-      "epoch": 3.023269857884778,
-      "grad_norm": 0.004299858585000038,
-      "learning_rate": 0.00019999549469484367,
-      "loss": 46.0,
-      "step": 18773
-    },
-    {
-      "epoch": 3.0234308949635653,
-      "grad_norm": 0.010553083382546902,
-      "learning_rate": 0.00019999549421460155,
-      "loss": 46.0,
-      "step": 18774
-    },
-    {
-      "epoch": 3.0235919320423528,
-      "grad_norm": 0.002172702457755804,
-      "learning_rate": 0.00019999549373433384,
-      "loss": 46.0,
-      "step": 18775
-    },
-    {
-      "epoch": 3.02375296912114,
-      "grad_norm": 0.01052631251513958,
-      "learning_rate": 0.00019999549325404054,
-      "loss": 46.0,
-      "step": 18776
-    },
-    {
-      "epoch": 3.0239140061999277,
-      "grad_norm": 0.0020577621180564165,
-      "learning_rate": 0.00019999549277372166,
-      "loss": 46.0,
-      "step": 18777
-    },
-    {
-      "epoch": 3.024075043278715,
-      "grad_norm": 0.009155402891337872,
-      "learning_rate": 0.00019999549229337716,
-      "loss": 46.0,
-      "step": 18778
-    },
-    {
-      "epoch": 3.024236080357502,
-      "grad_norm": 0.006087441463023424,
-      "learning_rate": 0.0001999954918130071,
-      "loss": 46.0,
-      "step": 18779
-    },
-    {
-      "epoch": 3.0243971174362896,
-      "grad_norm": 0.003831608220934868,
-      "learning_rate": 0.0001999954913326114,
-      "loss": 46.0,
-      "step": 18780
-    },
-    {
-      "epoch": 3.024558154515077,
-      "grad_norm": 0.0005081454291939735,
-      "learning_rate": 0.00019999549085219015,
-      "loss": 46.0,
-      "step": 18781
-    },
-    {
-      "epoch": 3.0247191915938645,
-      "grad_norm": 0.012818376533687115,
-      "learning_rate": 0.00019999549037174328,
-      "loss": 46.0,
-      "step": 18782
-    },
-    {
-      "epoch": 3.024880228672652,
-      "grad_norm": 0.007063471246510744,
-      "learning_rate": 0.00019999548989127082,
-      "loss": 46.0,
-      "step": 18783
-    },
-    {
-      "epoch": 3.0250412657514394,
-      "grad_norm": 0.0033594570122659206,
-      "learning_rate": 0.00019999548941077277,
-      "loss": 46.0,
-      "step": 18784
-    },
-    {
-      "epoch": 3.025202302830227,
-      "grad_norm": 0.007352369837462902,
-      "learning_rate": 0.00019999548893024914,
-      "loss": 46.0,
-      "step": 18785
-    },
-    {
-      "epoch": 3.025363339909014,
-      "grad_norm": 0.0030238190665841103,
-      "learning_rate": 0.00019999548844969992,
-      "loss": 46.0,
-      "step": 18786
-    },
-    {
-      "epoch": 3.0255243769878013,
-      "grad_norm": 0.002009775722399354,
-      "learning_rate": 0.00019999548796912505,
-      "loss": 46.0,
-      "step": 18787
-    },
-    {
-      "epoch": 3.0256854140665888,
-      "grad_norm": 0.002662340411916375,
-      "learning_rate": 0.00019999548748852463,
-      "loss": 46.0,
-      "step": 18788
-    },
-    {
-      "epoch": 3.025846451145376,
-      "grad_norm": 0.0009317048243246973,
-      "learning_rate": 0.00019999548700789862,
-      "loss": 46.0,
-      "step": 18789
-    },
-    {
-      "epoch": 3.0260074882241637,
-      "grad_norm": 0.0025243947748094797,
-      "learning_rate": 0.000199995486527247,
-      "loss": 46.0,
-      "step": 18790
-    },
-    {
-      "epoch": 3.026168525302951,
-      "grad_norm": 0.005571222864091396,
-      "learning_rate": 0.0001999954860465698,
-      "loss": 46.0,
-      "step": 18791
-    },
-    {
-      "epoch": 3.0263295623817386,
-      "grad_norm": 0.0019863948691636324,
-      "learning_rate": 0.000199995485565867,
-      "loss": 46.0,
-      "step": 18792
-    },
-    {
-      "epoch": 3.0264905994605256,
-      "grad_norm": 0.001520515768788755,
-      "learning_rate": 0.0001999954850851386,
-      "loss": 46.0,
-      "step": 18793
-    },
-    {
-      "epoch": 3.026651636539313,
-      "grad_norm": 0.0017588242189958692,
-      "learning_rate": 0.00019999548460438463,
-      "loss": 46.0,
-      "step": 18794
-    },
-    {
-      "epoch": 3.0268126736181005,
-      "grad_norm": 0.0005717701860703528,
-      "learning_rate": 0.00019999548412360504,
-      "loss": 46.0,
-      "step": 18795
-    },
-    {
-      "epoch": 3.026973710696888,
-      "grad_norm": 0.0031505892984569073,
-      "learning_rate": 0.00019999548364279987,
-      "loss": 46.0,
-      "step": 18796
-    },
-    {
-      "epoch": 3.0271347477756754,
-      "grad_norm": 0.002870397176593542,
-      "learning_rate": 0.0001999954831619691,
-      "loss": 46.0,
-      "step": 18797
-    },
-    {
-      "epoch": 3.027295784854463,
-      "grad_norm": 0.0012424769811332226,
-      "learning_rate": 0.00019999548268111276,
-      "loss": 46.0,
-      "step": 18798
-    },
-    {
-      "epoch": 3.0274568219332503,
-      "grad_norm": 0.0012431961949914694,
-      "learning_rate": 0.0001999954822002308,
-      "loss": 46.0,
-      "step": 18799
-    },
-    {
-      "epoch": 3.0276178590120377,
-      "grad_norm": 0.014016887173056602,
-      "learning_rate": 0.00019999548171932324,
-      "loss": 46.0,
-      "step": 18800
-    },
-    {
-      "epoch": 3.0277788960908247,
-      "grad_norm": 0.0014831143198534846,
-      "learning_rate": 0.0001999954812383901,
-      "loss": 46.0,
-      "step": 18801
-    },
-    {
-      "epoch": 3.027939933169612,
-      "grad_norm": 0.004771762993186712,
-      "learning_rate": 0.00019999548075743135,
-      "loss": 46.0,
-      "step": 18802
-    },
-    {
-      "epoch": 3.0281009702483996,
-      "grad_norm": 0.0055425758473575115,
-      "learning_rate": 0.00019999548027644704,
-      "loss": 46.0,
-      "step": 18803
-    },
-    {
-      "epoch": 3.028262007327187,
-      "grad_norm": 0.003356646979227662,
-      "learning_rate": 0.0001999954797954371,
-      "loss": 46.0,
-      "step": 18804
-    },
-    {
-      "epoch": 3.0284230444059745,
-      "grad_norm": 0.001957725500687957,
-      "learning_rate": 0.00019999547931440158,
-      "loss": 46.0,
-      "step": 18805
-    },
-    {
-      "epoch": 3.028584081484762,
-      "grad_norm": 0.0031272321939468384,
-      "learning_rate": 0.00019999547883334048,
-      "loss": 46.0,
-      "step": 18806
-    },
-    {
-      "epoch": 3.0287451185635494,
-      "grad_norm": 0.00486453901976347,
-      "learning_rate": 0.00019999547835225376,
-      "loss": 46.0,
-      "step": 18807
-    },
-    {
-      "epoch": 3.0289061556423365,
-      "grad_norm": 0.0048525077290833,
-      "learning_rate": 0.00019999547787114146,
-      "loss": 46.0,
-      "step": 18808
-    },
-    {
-      "epoch": 3.029067192721124,
-      "grad_norm": 0.01169600710272789,
-      "learning_rate": 0.00019999547739000357,
-      "loss": 46.0,
-      "step": 18809
-    },
-    {
-      "epoch": 3.0292282297999114,
-      "grad_norm": 0.002238035900518298,
-      "learning_rate": 0.0001999954769088401,
-      "loss": 46.0,
-      "step": 18810
-    },
-    {
-      "epoch": 3.029389266878699,
-      "grad_norm": 0.0010283234296366572,
-      "learning_rate": 0.000199995476427651,
-      "loss": 46.0,
-      "step": 18811
-    },
-    {
-      "epoch": 3.0295503039574863,
-      "grad_norm": 0.00341443857178092,
-      "learning_rate": 0.00019999547594643633,
-      "loss": 46.0,
-      "step": 18812
-    },
-    {
-      "epoch": 3.0297113410362737,
-      "grad_norm": 0.00789582822471857,
-      "learning_rate": 0.00019999547546519606,
-      "loss": 46.0,
-      "step": 18813
-    },
-    {
-      "epoch": 3.029872378115061,
-      "grad_norm": 0.004375720862299204,
-      "learning_rate": 0.0001999954749839302,
-      "loss": 46.0,
-      "step": 18814
-    },
-    {
-      "epoch": 3.030033415193848,
-      "grad_norm": 0.004484332632273436,
-      "learning_rate": 0.00019999547450263874,
-      "loss": 46.0,
-      "step": 18815
-    },
-    {
-      "epoch": 3.0301944522726356,
-      "grad_norm": 0.004468945786356926,
-      "learning_rate": 0.0001999954740213217,
-      "loss": 46.0,
-      "step": 18816
-    },
-    {
-      "epoch": 3.030355489351423,
-      "grad_norm": 0.0017636411357671022,
-      "learning_rate": 0.00019999547353997905,
-      "loss": 46.0,
-      "step": 18817
-    },
-    {
-      "epoch": 3.0305165264302105,
-      "grad_norm": 0.0004101128433831036,
-      "learning_rate": 0.00019999547305861082,
-      "loss": 46.0,
-      "step": 18818
-    },
-    {
-      "epoch": 3.030677563508998,
-      "grad_norm": 0.002128963591530919,
-      "learning_rate": 0.00019999547257721698,
-      "loss": 46.0,
-      "step": 18819
-    },
-    {
-      "epoch": 3.0308386005877854,
-      "grad_norm": 0.0025925440713763237,
-      "learning_rate": 0.00019999547209579755,
-      "loss": 46.0,
-      "step": 18820
-    },
-    {
-      "epoch": 3.030999637666573,
-      "grad_norm": 0.002610244555398822,
-      "learning_rate": 0.00019999547161435254,
-      "loss": 46.0,
-      "step": 18821
-    },
-    {
-      "epoch": 3.0311606747453603,
-      "grad_norm": 0.0006201292271725833,
-      "learning_rate": 0.00019999547113288193,
-      "loss": 46.0,
-      "step": 18822
-    },
-    {
-      "epoch": 3.0313217118241473,
-      "grad_norm": 0.00097772351000458,
-      "learning_rate": 0.00019999547065138574,
-      "loss": 46.0,
-      "step": 18823
-    },
-    {
-      "epoch": 3.031482748902935,
-      "grad_norm": 0.0012892846716567874,
-      "learning_rate": 0.00019999547016986394,
-      "loss": 46.0,
-      "step": 18824
-    },
-    {
-      "epoch": 3.0316437859817222,
-      "grad_norm": 0.003242732025682926,
-      "learning_rate": 0.00019999546968831652,
-      "loss": 46.0,
-      "step": 18825
-    },
-    {
-      "epoch": 3.0318048230605097,
-      "grad_norm": 0.0049265166744589806,
-      "learning_rate": 0.00019999546920674357,
-      "loss": 46.0,
-      "step": 18826
-    },
-    {
-      "epoch": 3.031965860139297,
-      "grad_norm": 0.005905521102249622,
-      "learning_rate": 0.00019999546872514497,
-      "loss": 46.0,
-      "step": 18827
-    },
-    {
-      "epoch": 3.0321268972180846,
-      "grad_norm": 0.003319772658869624,
-      "learning_rate": 0.0001999954682435208,
-      "loss": 46.0,
-      "step": 18828
-    },
-    {
-      "epoch": 3.032287934296872,
-      "grad_norm": 0.0021305489353835583,
-      "learning_rate": 0.00019999546776187102,
-      "loss": 46.0,
-      "step": 18829
-    },
-    {
-      "epoch": 3.032448971375659,
-      "grad_norm": 0.0016076359897851944,
-      "learning_rate": 0.00019999546728019567,
-      "loss": 46.0,
-      "step": 18830
-    },
-    {
-      "epoch": 3.0326100084544465,
-      "grad_norm": 0.004042502958327532,
-      "learning_rate": 0.00019999546679849473,
-      "loss": 46.0,
-      "step": 18831
-    },
-    {
-      "epoch": 3.032771045533234,
-      "grad_norm": 0.004754027351737022,
-      "learning_rate": 0.00019999546631676817,
-      "loss": 46.0,
-      "step": 18832
-    },
-    {
-      "epoch": 3.0329320826120214,
-      "grad_norm": 0.003226912347599864,
-      "learning_rate": 0.00019999546583501603,
-      "loss": 46.0,
-      "step": 18833
-    },
-    {
-      "epoch": 3.033093119690809,
-      "grad_norm": 0.027438033372163773,
-      "learning_rate": 0.0001999954653532383,
-      "loss": 46.0,
-      "step": 18834
-    },
-    {
-      "epoch": 3.0332541567695963,
-      "grad_norm": 0.0032605247106403112,
-      "learning_rate": 0.00019999546487143498,
-      "loss": 46.0,
-      "step": 18835
-    },
-    {
-      "epoch": 3.0334151938483838,
-      "grad_norm": 0.004450343083590269,
-      "learning_rate": 0.00019999546438960605,
-      "loss": 46.0,
-      "step": 18836
-    },
-    {
-      "epoch": 3.0335762309271708,
-      "grad_norm": 0.0030439465772360563,
-      "learning_rate": 0.00019999546390775153,
-      "loss": 46.0,
-      "step": 18837
-    },
-    {
-      "epoch": 3.033737268005958,
-      "grad_norm": 0.0020435668993741274,
-      "learning_rate": 0.00019999546342587142,
-      "loss": 46.0,
-      "step": 18838
-    },
-    {
-      "epoch": 3.0338983050847457,
-      "grad_norm": 0.0011062038829550147,
-      "learning_rate": 0.00019999546294396573,
-      "loss": 46.0,
-      "step": 18839
-    },
-    {
-      "epoch": 3.034059342163533,
-      "grad_norm": 0.006097679957747459,
-      "learning_rate": 0.00019999546246203442,
-      "loss": 46.0,
-      "step": 18840
-    },
-    {
-      "epoch": 3.0342203792423206,
-      "grad_norm": 0.005517052952200174,
-      "learning_rate": 0.00019999546198007752,
-      "loss": 46.0,
-      "step": 18841
-    },
-    {
-      "epoch": 3.034381416321108,
-      "grad_norm": 0.0010438545141369104,
-      "learning_rate": 0.00019999546149809504,
-      "loss": 46.0,
-      "step": 18842
-    },
-    {
-      "epoch": 3.0345424533998955,
-      "grad_norm": 0.0011138978879898787,
-      "learning_rate": 0.00019999546101608697,
-      "loss": 46.0,
-      "step": 18843
-    },
-    {
-      "epoch": 3.034703490478683,
-      "grad_norm": 0.007553721312433481,
-      "learning_rate": 0.0001999954605340533,
-      "loss": 46.0,
-      "step": 18844
-    },
-    {
-      "epoch": 3.03486452755747,
-      "grad_norm": 0.012318179942667484,
-      "learning_rate": 0.00019999546005199404,
-      "loss": 46.0,
-      "step": 18845
-    },
-    {
-      "epoch": 3.0350255646362574,
-      "grad_norm": 0.008240317925810814,
-      "learning_rate": 0.00019999545956990916,
-      "loss": 46.0,
-      "step": 18846
-    },
-    {
-      "epoch": 3.035186601715045,
-      "grad_norm": 0.0020806973334401846,
-      "learning_rate": 0.0001999954590877987,
-      "loss": 46.0,
-      "step": 18847
-    },
-    {
-      "epoch": 3.0353476387938323,
-      "grad_norm": 0.0019417419098317623,
-      "learning_rate": 0.00019999545860566265,
-      "loss": 46.0,
-      "step": 18848
-    },
-    {
-      "epoch": 3.0355086758726197,
-      "grad_norm": 0.00509099243208766,
-      "learning_rate": 0.00019999545812350103,
-      "loss": 46.0,
-      "step": 18849
-    },
-    {
-      "epoch": 3.035669712951407,
-      "grad_norm": 0.003336383495479822,
-      "learning_rate": 0.0001999954576413138,
-      "loss": 46.0,
-      "step": 18850
-    },
-    {
-      "epoch": 3.0358307500301946,
-      "grad_norm": 0.0034357572440057993,
-      "learning_rate": 0.00019999545715910095,
-      "loss": 46.0,
-      "step": 18851
-    },
-    {
-      "epoch": 3.0359917871089817,
-      "grad_norm": 0.007890604436397552,
-      "learning_rate": 0.00019999545667686252,
-      "loss": 46.0,
-      "step": 18852
-    },
-    {
-      "epoch": 3.036152824187769,
-      "grad_norm": 0.006422179751098156,
-      "learning_rate": 0.00019999545619459852,
-      "loss": 46.0,
-      "step": 18853
-    },
-    {
-      "epoch": 3.0363138612665566,
-      "grad_norm": 0.0009265092085115612,
-      "learning_rate": 0.0001999954557123089,
-      "loss": 46.0,
-      "step": 18854
-    },
-    {
-      "epoch": 3.036474898345344,
-      "grad_norm": 0.012349448166787624,
-      "learning_rate": 0.00019999545522999371,
-      "loss": 46.0,
-      "step": 18855
-    },
-    {
-      "epoch": 3.0366359354241315,
-      "grad_norm": 0.0021277754567563534,
-      "learning_rate": 0.0001999954547476529,
-      "loss": 46.0,
-      "step": 18856
-    },
-    {
-      "epoch": 3.036796972502919,
-      "grad_norm": 0.0139263104647398,
-      "learning_rate": 0.00019999545426528653,
-      "loss": 46.0,
-      "step": 18857
-    },
-    {
-      "epoch": 3.0369580095817064,
-      "grad_norm": 0.0029302441980689764,
-      "learning_rate": 0.00019999545378289452,
-      "loss": 46.0,
-      "step": 18858
-    },
-    {
-      "epoch": 3.0371190466604934,
-      "grad_norm": 0.0013795375125482678,
-      "learning_rate": 0.00019999545330047695,
-      "loss": 46.0,
-      "step": 18859
-    },
-    {
-      "epoch": 3.037280083739281,
-      "grad_norm": 0.007578321732580662,
-      "learning_rate": 0.00019999545281803381,
-      "loss": 46.0,
-      "step": 18860
-    },
-    {
-      "epoch": 3.0374411208180683,
-      "grad_norm": 0.0029816231690347195,
-      "learning_rate": 0.000199995452335565,
-      "loss": 46.0,
-      "step": 18861
-    },
-    {
-      "epoch": 3.0376021578968557,
-      "grad_norm": 0.0016293191583827138,
-      "learning_rate": 0.00019999545185307068,
-      "loss": 46.0,
-      "step": 18862
-    },
-    {
-      "epoch": 3.037763194975643,
-      "grad_norm": 0.0036395173519849777,
-      "learning_rate": 0.0001999954513705507,
-      "loss": 46.0,
-      "step": 18863
-    },
-    {
-      "epoch": 3.0379242320544306,
-      "grad_norm": 0.010569578036665916,
-      "learning_rate": 0.00019999545088800517,
-      "loss": 46.0,
-      "step": 18864
-    },
-    {
-      "epoch": 3.038085269133218,
-      "grad_norm": 0.001031906926073134,
-      "learning_rate": 0.00019999545040543402,
-      "loss": 46.0,
-      "step": 18865
-    },
-    {
-      "epoch": 3.0382463062120055,
-      "grad_norm": 0.008065774105489254,
-      "learning_rate": 0.00019999544992283728,
-      "loss": 46.0,
-      "step": 18866
-    },
-    {
-      "epoch": 3.0384073432907925,
-      "grad_norm": 0.0038250968791544437,
-      "learning_rate": 0.00019999544944021496,
-      "loss": 46.0,
-      "step": 18867
-    },
-    {
-      "epoch": 3.03856838036958,
-      "grad_norm": 0.007237450685352087,
-      "learning_rate": 0.00019999544895756704,
-      "loss": 46.0,
-      "step": 18868
-    },
-    {
-      "epoch": 3.0387294174483674,
-      "grad_norm": 0.0022652449551969767,
-      "learning_rate": 0.00019999544847489352,
-      "loss": 46.0,
-      "step": 18869
-    },
-    {
-      "epoch": 3.038890454527155,
-      "grad_norm": 0.005361523479223251,
-      "learning_rate": 0.00019999544799219443,
-      "loss": 46.0,
-      "step": 18870
-    },
-    {
-      "epoch": 3.0390514916059423,
-      "grad_norm": 0.0017716326983645558,
-      "learning_rate": 0.00019999544750946973,
-      "loss": 46.0,
-      "step": 18871
-    },
-    {
-      "epoch": 3.03921252868473,
-      "grad_norm": 0.0035376125015318394,
-      "learning_rate": 0.00019999544702671942,
-      "loss": 46.0,
-      "step": 18872
-    },
-    {
-      "epoch": 3.0393735657635172,
-      "grad_norm": 0.0018904684111475945,
-      "learning_rate": 0.00019999544654394354,
-      "loss": 46.0,
-      "step": 18873
-    },
-    {
-      "epoch": 3.0395346028423043,
-      "grad_norm": 0.013746327720582485,
-      "learning_rate": 0.00019999544606114206,
-      "loss": 46.0,
-      "step": 18874
-    },
-    {
-      "epoch": 3.0396956399210917,
-      "grad_norm": 0.0016141963424161077,
-      "learning_rate": 0.00019999544557831498,
-      "loss": 46.0,
-      "step": 18875
-    },
-    {
-      "epoch": 3.039856676999879,
-      "grad_norm": 0.007423622068017721,
-      "learning_rate": 0.0001999954450954623,
-      "loss": 46.0,
-      "step": 18876
-    },
-    {
-      "epoch": 3.0400177140786666,
-      "grad_norm": 0.0047676232643425465,
-      "learning_rate": 0.000199995444612584,
-      "loss": 46.0,
-      "step": 18877
-    },
-    {
-      "epoch": 3.040178751157454,
-      "grad_norm": 0.006295583210885525,
-      "learning_rate": 0.00019999544412968017,
-      "loss": 46.0,
-      "step": 18878
-    },
-    {
-      "epoch": 3.0403397882362415,
-      "grad_norm": 0.0018172429408878088,
-      "learning_rate": 0.00019999544364675072,
-      "loss": 46.0,
-      "step": 18879
-    },
-    {
-      "epoch": 3.040500825315029,
-      "grad_norm": 0.005919110961258411,
-      "learning_rate": 0.00019999544316379568,
-      "loss": 46.0,
-      "step": 18880
-    },
-    {
-      "epoch": 3.040661862393816,
-      "grad_norm": 0.0012296631466597319,
-      "learning_rate": 0.00019999544268081506,
-      "loss": 46.0,
-      "step": 18881
-    },
-    {
-      "epoch": 3.0408228994726034,
-      "grad_norm": 0.007597137708216906,
-      "learning_rate": 0.0001999954421978088,
-      "loss": 46.0,
-      "step": 18882
-    },
-    {
-      "epoch": 3.040983936551391,
-      "grad_norm": 0.0016721213469281793,
-      "learning_rate": 0.00019999544171477696,
-      "loss": 46.0,
-      "step": 18883
-    },
-    {
-      "epoch": 3.0411449736301783,
-      "grad_norm": 0.0032726868521422148,
-      "learning_rate": 0.00019999544123171955,
-      "loss": 46.0,
-      "step": 18884
-    },
-    {
-      "epoch": 3.041306010708966,
-      "grad_norm": 0.0037985057570040226,
-      "learning_rate": 0.00019999544074863655,
-      "loss": 46.0,
-      "step": 18885
-    },
-    {
-      "epoch": 3.0414670477877532,
-      "grad_norm": 0.005214225966483355,
-      "learning_rate": 0.00019999544026552793,
-      "loss": 46.0,
-      "step": 18886
-    },
-    {
-      "epoch": 3.0416280848665407,
-      "grad_norm": 0.0033551345113664865,
-      "learning_rate": 0.00019999543978239373,
-      "loss": 46.0,
-      "step": 18887
-    },
-    {
-      "epoch": 3.041789121945328,
-      "grad_norm": 0.0038539241068065166,
-      "learning_rate": 0.00019999543929923394,
-      "loss": 46.0,
-      "step": 18888
-    },
-    {
-      "epoch": 3.041950159024115,
-      "grad_norm": 0.004168844781816006,
-      "learning_rate": 0.00019999543881604853,
-      "loss": 46.0,
-      "step": 18889
-    },
-    {
-      "epoch": 3.0421111961029026,
-      "grad_norm": 0.003933880012482405,
-      "learning_rate": 0.00019999543833283757,
-      "loss": 46.0,
-      "step": 18890
-    },
-    {
-      "epoch": 3.04227223318169,
-      "grad_norm": 0.001300912699662149,
-      "learning_rate": 0.000199995437849601,
-      "loss": 46.0,
-      "step": 18891
-    },
-    {
-      "epoch": 3.0424332702604775,
-      "grad_norm": 0.0013871589908376336,
-      "learning_rate": 0.00019999543736633882,
-      "loss": 46.0,
-      "step": 18892
-    },
-    {
-      "epoch": 3.042594307339265,
-      "grad_norm": 0.00045556091936305165,
-      "learning_rate": 0.00019999543688305104,
-      "loss": 46.0,
-      "step": 18893
-    },
-    {
-      "epoch": 3.0427553444180524,
-      "grad_norm": 0.005769932176917791,
-      "learning_rate": 0.0001999954363997377,
-      "loss": 46.0,
-      "step": 18894
-    },
-    {
-      "epoch": 3.04291638149684,
-      "grad_norm": 0.003269238630309701,
-      "learning_rate": 0.00019999543591639872,
-      "loss": 46.0,
-      "step": 18895
-    },
-    {
-      "epoch": 3.043077418575627,
-      "grad_norm": 0.00249124295078218,
-      "learning_rate": 0.00019999543543303418,
-      "loss": 46.0,
-      "step": 18896
-    },
-    {
-      "epoch": 3.0432384556544143,
-      "grad_norm": 0.004293342120945454,
-      "learning_rate": 0.00019999543494964405,
-      "loss": 46.0,
-      "step": 18897
-    },
-    {
-      "epoch": 3.0433994927332018,
-      "grad_norm": 0.004066897556185722,
-      "learning_rate": 0.00019999543446622833,
-      "loss": 46.0,
-      "step": 18898
-    },
-    {
-      "epoch": 3.043560529811989,
-      "grad_norm": 0.002417211188003421,
-      "learning_rate": 0.000199995433982787,
-      "loss": 46.0,
-      "step": 18899
-    },
-    {
-      "epoch": 3.0437215668907767,
-      "grad_norm": 0.002351326635107398,
-      "learning_rate": 0.00019999543349932006,
-      "loss": 46.0,
-      "step": 18900
-    },
-    {
-      "epoch": 3.043882603969564,
-      "grad_norm": 0.005535287782549858,
-      "learning_rate": 0.00019999543301582758,
-      "loss": 46.0,
-      "step": 18901
-    },
-    {
-      "epoch": 3.0440436410483516,
-      "grad_norm": 0.0021434519439935684,
-      "learning_rate": 0.00019999543253230946,
-      "loss": 46.0,
-      "step": 18902
-    },
-    {
-      "epoch": 3.0442046781271386,
-      "grad_norm": 0.0018834639340639114,
-      "learning_rate": 0.00019999543204876575,
-      "loss": 46.0,
-      "step": 18903
-    },
-    {
-      "epoch": 3.044365715205926,
-      "grad_norm": 0.017911570146679878,
-      "learning_rate": 0.00019999543156519643,
-      "loss": 46.0,
-      "step": 18904
-    },
-    {
-      "epoch": 3.0445267522847135,
-      "grad_norm": 0.003974526654928923,
-      "learning_rate": 0.00019999543108160158,
-      "loss": 46.0,
-      "step": 18905
-    },
-    {
-      "epoch": 3.044687789363501,
-      "grad_norm": 0.0032931321766227484,
-      "learning_rate": 0.00019999543059798108,
-      "loss": 46.0,
-      "step": 18906
-    },
-    {
-      "epoch": 3.0448488264422884,
-      "grad_norm": 0.0022926926612854004,
-      "learning_rate": 0.000199995430114335,
-      "loss": 46.0,
-      "step": 18907
-    },
-    {
-      "epoch": 3.045009863521076,
-      "grad_norm": 0.0005494573269970715,
-      "learning_rate": 0.00019999542963066333,
-      "loss": 46.0,
-      "step": 18908
-    },
-    {
-      "epoch": 3.0451709005998633,
-      "grad_norm": 0.010758903808891773,
-      "learning_rate": 0.00019999542914696608,
-      "loss": 46.0,
-      "step": 18909
-    },
-    {
-      "epoch": 3.0453319376786503,
-      "grad_norm": 0.010110465809702873,
-      "learning_rate": 0.00019999542866324323,
-      "loss": 46.0,
-      "step": 18910
-    },
-    {
-      "epoch": 3.0454929747574377,
-      "grad_norm": 0.0005439491360448301,
-      "learning_rate": 0.00019999542817949477,
-      "loss": 46.0,
-      "step": 18911
-    },
-    {
-      "epoch": 3.045654011836225,
-      "grad_norm": 0.0037642328534275293,
-      "learning_rate": 0.00019999542769572073,
-      "loss": 46.0,
-      "step": 18912
-    },
-    {
-      "epoch": 3.0458150489150126,
-      "grad_norm": 0.0009757018415257335,
-      "learning_rate": 0.0001999954272119211,
-      "loss": 46.0,
-      "step": 18913
-    },
-    {
-      "epoch": 3.0459760859938,
-      "grad_norm": 0.010762848891317844,
-      "learning_rate": 0.00019999542672809585,
-      "loss": 46.0,
-      "step": 18914
-    },
-    {
-      "epoch": 3.0461371230725875,
-      "grad_norm": 0.003563212463632226,
-      "learning_rate": 0.00019999542624424504,
-      "loss": 46.0,
-      "step": 18915
-    },
-    {
-      "epoch": 3.046298160151375,
-      "grad_norm": 0.003709547221660614,
-      "learning_rate": 0.00019999542576036862,
-      "loss": 46.0,
-      "step": 18916
-    },
-    {
-      "epoch": 3.0464591972301625,
-      "grad_norm": 0.006384704262018204,
-      "learning_rate": 0.0001999954252764666,
-      "loss": 46.0,
-      "step": 18917
-    },
-    {
-      "epoch": 3.0466202343089495,
-      "grad_norm": 0.0021330039016902447,
-      "learning_rate": 0.000199995424792539,
-      "loss": 46.0,
-      "step": 18918
-    },
-    {
-      "epoch": 3.046781271387737,
-      "grad_norm": 0.0026270258240401745,
-      "learning_rate": 0.0001999954243085858,
-      "loss": 46.0,
-      "step": 18919
-    },
-    {
-      "epoch": 3.0469423084665244,
-      "grad_norm": 0.003202127991244197,
-      "learning_rate": 0.000199995423824607,
-      "loss": 46.0,
-      "step": 18920
-    },
-    {
-      "epoch": 3.047103345545312,
-      "grad_norm": 0.001913890358991921,
-      "learning_rate": 0.00019999542334060263,
-      "loss": 46.0,
-      "step": 18921
-    },
-    {
-      "epoch": 3.0472643826240993,
-      "grad_norm": 0.0006028263596817851,
-      "learning_rate": 0.00019999542285657263,
-      "loss": 46.0,
-      "step": 18922
-    },
-    {
-      "epoch": 3.0474254197028867,
-      "grad_norm": 0.003960226662456989,
-      "learning_rate": 0.00019999542237251707,
-      "loss": 46.0,
-      "step": 18923
-    },
-    {
-      "epoch": 3.047586456781674,
-      "grad_norm": 0.005489690694957972,
-      "learning_rate": 0.0001999954218884359,
-      "loss": 46.0,
-      "step": 18924
-    },
-    {
-      "epoch": 3.047747493860461,
-      "grad_norm": 0.0021123201586306095,
-      "learning_rate": 0.00019999542140432914,
-      "loss": 46.0,
-      "step": 18925
-    },
-    {
-      "epoch": 3.0479085309392486,
-      "grad_norm": 0.0012478644493967295,
-      "learning_rate": 0.00019999542092019677,
-      "loss": 46.0,
-      "step": 18926
-    },
-    {
-      "epoch": 3.048069568018036,
-      "grad_norm": 0.012074966914951801,
-      "learning_rate": 0.0001999954204360388,
-      "loss": 46.0,
-      "step": 18927
-    },
-    {
-      "epoch": 3.0482306050968235,
-      "grad_norm": 0.001110857934691012,
-      "learning_rate": 0.00019999541995185526,
-      "loss": 46.0,
-      "step": 18928
-    },
-    {
-      "epoch": 3.048391642175611,
-      "grad_norm": 0.011390984058380127,
-      "learning_rate": 0.00019999541946764615,
-      "loss": 46.0,
-      "step": 18929
-    },
-    {
-      "epoch": 3.0485526792543984,
-      "grad_norm": 0.0035159746184945107,
-      "learning_rate": 0.0001999954189834114,
-      "loss": 46.0,
-      "step": 18930
-    },
-    {
-      "epoch": 3.048713716333186,
-      "grad_norm": 0.00884826760739088,
-      "learning_rate": 0.0001999954184991511,
-      "loss": 46.0,
-      "step": 18931
-    },
-    {
-      "epoch": 3.048874753411973,
-      "grad_norm": 0.007816282100975513,
-      "learning_rate": 0.00019999541801486517,
-      "loss": 46.0,
-      "step": 18932
-    },
-    {
-      "epoch": 3.0490357904907603,
-      "grad_norm": 0.007879632525146008,
-      "learning_rate": 0.00019999541753055365,
-      "loss": 46.0,
-      "step": 18933
-    },
-    {
-      "epoch": 3.049196827569548,
-      "grad_norm": 0.002085912972688675,
-      "learning_rate": 0.00019999541704621656,
-      "loss": 46.0,
-      "step": 18934
-    },
-    {
-      "epoch": 3.0493578646483352,
-      "grad_norm": 0.0005128393531776965,
-      "learning_rate": 0.00019999541656185384,
-      "loss": 46.0,
-      "step": 18935
-    },
-    {
-      "epoch": 3.0495189017271227,
-      "grad_norm": 0.0019889206159859896,
-      "learning_rate": 0.00019999541607746557,
-      "loss": 46.0,
-      "step": 18936
-    },
-    {
-      "epoch": 3.04967993880591,
-      "grad_norm": 0.002107777167111635,
-      "learning_rate": 0.00019999541559305168,
-      "loss": 46.0,
-      "step": 18937
-    },
-    {
-      "epoch": 3.0498409758846976,
-      "grad_norm": 0.0053547583520412445,
-      "learning_rate": 0.00019999541510861218,
-      "loss": 46.0,
-      "step": 18938
-    },
-    {
-      "epoch": 3.050002012963485,
-      "grad_norm": 0.007552565541118383,
-      "learning_rate": 0.00019999541462414712,
-      "loss": 46.0,
-      "step": 18939
-    },
-    {
-      "epoch": 3.050163050042272,
-      "grad_norm": 0.005827803630381823,
-      "learning_rate": 0.00019999541413965645,
-      "loss": 46.0,
-      "step": 18940
-    },
-    {
-      "epoch": 3.0503240871210595,
-      "grad_norm": 0.003172443248331547,
-      "learning_rate": 0.00019999541365514018,
-      "loss": 46.0,
-      "step": 18941
-    },
-    {
-      "epoch": 3.050485124199847,
-      "grad_norm": 0.0031627239659428596,
-      "learning_rate": 0.00019999541317059833,
-      "loss": 46.0,
-      "step": 18942
-    },
-    {
-      "epoch": 3.0506461612786344,
-      "grad_norm": 0.007425693329423666,
-      "learning_rate": 0.00019999541268603087,
-      "loss": 46.0,
-      "step": 18943
-    },
-    {
-      "epoch": 3.050807198357422,
-      "grad_norm": 0.003946628887206316,
-      "learning_rate": 0.00019999541220143784,
-      "loss": 46.0,
-      "step": 18944
-    },
-    {
-      "epoch": 3.0509682354362093,
-      "grad_norm": 0.004186005797237158,
-      "learning_rate": 0.0001999954117168192,
-      "loss": 46.0,
-      "step": 18945
-    },
-    {
-      "epoch": 3.0511292725149968,
-      "grad_norm": 0.0015321825630962849,
-      "learning_rate": 0.00019999541123217498,
-      "loss": 46.0,
-      "step": 18946
-    },
-    {
-      "epoch": 3.051290309593784,
-      "grad_norm": 0.0016872929409146309,
-      "learning_rate": 0.00019999541074750514,
-      "loss": 46.0,
-      "step": 18947
-    },
-    {
-      "epoch": 3.0514513466725712,
-      "grad_norm": 0.0020634718239307404,
-      "learning_rate": 0.00019999541026280974,
-      "loss": 46.0,
-      "step": 18948
-    },
-    {
-      "epoch": 3.0516123837513587,
-      "grad_norm": 0.0044443742372095585,
-      "learning_rate": 0.00019999540977808873,
-      "loss": 46.0,
-      "step": 18949
-    },
-    {
-      "epoch": 3.051773420830146,
-      "grad_norm": 0.0053870584815740585,
-      "learning_rate": 0.00019999540929334212,
-      "loss": 46.0,
-      "step": 18950
-    },
-    {
-      "epoch": 3.0519344579089336,
-      "grad_norm": 0.0009179266635328531,
-      "learning_rate": 0.0001999954088085699,
-      "loss": 46.0,
-      "step": 18951
-    },
-    {
-      "epoch": 3.052095494987721,
-      "grad_norm": 0.007610139437019825,
-      "learning_rate": 0.00019999540832377213,
-      "loss": 46.0,
-      "step": 18952
-    },
-    {
-      "epoch": 3.0522565320665085,
-      "grad_norm": 0.009230876341462135,
-      "learning_rate": 0.00019999540783894874,
-      "loss": 46.0,
-      "step": 18953
-    },
-    {
-      "epoch": 3.0524175691452955,
-      "grad_norm": 0.0012778345262631774,
-      "learning_rate": 0.00019999540735409976,
-      "loss": 46.0,
-      "step": 18954
-    },
-    {
-      "epoch": 3.052578606224083,
-      "grad_norm": 0.006814607884734869,
-      "learning_rate": 0.00019999540686922517,
-      "loss": 46.0,
-      "step": 18955
-    },
-    {
-      "epoch": 3.0527396433028704,
-      "grad_norm": 0.004744658712297678,
-      "learning_rate": 0.00019999540638432502,
-      "loss": 46.0,
-      "step": 18956
-    },
-    {
-      "epoch": 3.052900680381658,
-      "grad_norm": 0.00651804031804204,
-      "learning_rate": 0.00019999540589939928,
-      "loss": 46.0,
-      "step": 18957
-    },
-    {
-      "epoch": 3.0530617174604453,
-      "grad_norm": 0.0016507715918123722,
-      "learning_rate": 0.0001999954054144479,
-      "loss": 46.0,
-      "step": 18958
-    },
-    {
-      "epoch": 3.0532227545392328,
-      "grad_norm": 0.0009587332606315613,
-      "learning_rate": 0.00019999540492947096,
-      "loss": 46.0,
-      "step": 18959
-    },
-    {
-      "epoch": 3.05338379161802,
-      "grad_norm": 0.005978851579129696,
-      "learning_rate": 0.00019999540444446843,
-      "loss": 46.0,
-      "step": 18960
-    },
-    {
-      "epoch": 3.0535448286968077,
-      "grad_norm": 0.005107056815177202,
-      "learning_rate": 0.00019999540395944026,
-      "loss": 46.0,
-      "step": 18961
-    },
-    {
-      "epoch": 3.0537058657755947,
-      "grad_norm": 0.0059722792357206345,
-      "learning_rate": 0.00019999540347438654,
-      "loss": 46.0,
-      "step": 18962
-    },
-    {
-      "epoch": 3.053866902854382,
-      "grad_norm": 0.007648978848010302,
-      "learning_rate": 0.00019999540298930725,
-      "loss": 46.0,
-      "step": 18963
-    },
-    {
-      "epoch": 3.0540279399331696,
-      "grad_norm": 0.00248450948856771,
-      "learning_rate": 0.0001999954025042023,
-      "loss": 46.0,
-      "step": 18964
-    },
-    {
-      "epoch": 3.054188977011957,
-      "grad_norm": 0.0018780005630105734,
-      "learning_rate": 0.0001999954020190718,
-      "loss": 46.0,
-      "step": 18965
-    },
-    {
-      "epoch": 3.0543500140907445,
-      "grad_norm": 0.0059234327636659145,
-      "learning_rate": 0.0001999954015339157,
-      "loss": 46.0,
-      "step": 18966
-    },
-    {
-      "epoch": 3.054511051169532,
-      "grad_norm": 0.0022601732052862644,
-      "learning_rate": 0.000199995401048734,
-      "loss": 46.0,
-      "step": 18967
-    },
-    {
-      "epoch": 3.0546720882483194,
-      "grad_norm": 0.001392781618051231,
-      "learning_rate": 0.00019999540056352672,
-      "loss": 46.0,
-      "step": 18968
-    },
-    {
-      "epoch": 3.0548331253271064,
-      "grad_norm": 0.0023862901143729687,
-      "learning_rate": 0.00019999540007829383,
-      "loss": 46.0,
-      "step": 18969
-    },
-    {
-      "epoch": 3.054994162405894,
-      "grad_norm": 0.005048906430602074,
-      "learning_rate": 0.00019999539959303535,
-      "loss": 46.0,
-      "step": 18970
-    },
-    {
-      "epoch": 3.0551551994846813,
-      "grad_norm": 0.003923005424439907,
-      "learning_rate": 0.00019999539910775128,
-      "loss": 46.0,
-      "step": 18971
-    },
-    {
-      "epoch": 3.0553162365634687,
-      "grad_norm": 0.0021009675692766905,
-      "learning_rate": 0.0001999953986224416,
-      "loss": 46.0,
-      "step": 18972
-    },
-    {
-      "epoch": 3.055477273642256,
-      "grad_norm": 0.0029160145204514265,
-      "learning_rate": 0.00019999539813710636,
-      "loss": 46.0,
-      "step": 18973
-    },
-    {
-      "epoch": 3.0556383107210436,
-      "grad_norm": 0.002000607782974839,
-      "learning_rate": 0.0001999953976517455,
-      "loss": 46.0,
-      "step": 18974
-    },
-    {
-      "epoch": 3.055799347799831,
-      "grad_norm": 0.003365874756127596,
-      "learning_rate": 0.00019999539716635906,
-      "loss": 46.0,
-      "step": 18975
-    },
-    {
-      "epoch": 3.055960384878618,
-      "grad_norm": 0.002717716619372368,
-      "learning_rate": 0.000199995396680947,
-      "loss": 46.0,
-      "step": 18976
-    },
-    {
-      "epoch": 3.0561214219574055,
-      "grad_norm": 0.001957734115421772,
-      "learning_rate": 0.00019999539619550938,
-      "loss": 46.0,
-      "step": 18977
-    },
-    {
-      "epoch": 3.056282459036193,
-      "grad_norm": 0.0007823628839105368,
-      "learning_rate": 0.00019999539571004612,
-      "loss": 46.0,
-      "step": 18978
-    },
-    {
-      "epoch": 3.0564434961149805,
-      "grad_norm": 0.007330520544201136,
-      "learning_rate": 0.00019999539522455733,
-      "loss": 46.0,
-      "step": 18979
-    },
-    {
-      "epoch": 3.056604533193768,
-      "grad_norm": 0.000815938925370574,
-      "learning_rate": 0.0001999953947390429,
-      "loss": 46.0,
-      "step": 18980
-    },
-    {
-      "epoch": 3.0567655702725554,
-      "grad_norm": 0.002668933942914009,
-      "learning_rate": 0.00019999539425350288,
-      "loss": 46.0,
-      "step": 18981
-    },
-    {
-      "epoch": 3.056926607351343,
-      "grad_norm": 0.0015341124963015318,
-      "learning_rate": 0.0001999953937679373,
-      "loss": 46.0,
-      "step": 18982
-    },
-    {
-      "epoch": 3.05708764443013,
-      "grad_norm": 0.0014781627105548978,
-      "learning_rate": 0.0001999953932823461,
-      "loss": 46.0,
-      "step": 18983
-    },
-    {
-      "epoch": 3.0572486815089173,
-      "grad_norm": 0.0025400114245712757,
-      "learning_rate": 0.0001999953927967293,
-      "loss": 46.0,
-      "step": 18984
-    },
-    {
-      "epoch": 3.0574097185877047,
-      "grad_norm": 0.01987171173095703,
-      "learning_rate": 0.00019999539231108692,
-      "loss": 46.0,
-      "step": 18985
-    },
-    {
-      "epoch": 3.057570755666492,
-      "grad_norm": 0.002904577413573861,
-      "learning_rate": 0.00019999539182541894,
-      "loss": 46.0,
-      "step": 18986
-    },
-    {
-      "epoch": 3.0577317927452796,
-      "grad_norm": 0.0034273648634552956,
-      "learning_rate": 0.00019999539133972537,
-      "loss": 46.0,
-      "step": 18987
-    },
-    {
-      "epoch": 3.057892829824067,
-      "grad_norm": 0.005315632093697786,
-      "learning_rate": 0.0001999953908540062,
-      "loss": 46.0,
-      "step": 18988
-    },
-    {
-      "epoch": 3.0580538669028545,
-      "grad_norm": 0.0017369046108797193,
-      "learning_rate": 0.00019999539036826144,
-      "loss": 46.0,
-      "step": 18989
-    },
-    {
-      "epoch": 3.058214903981642,
-      "grad_norm": 0.003790066344663501,
-      "learning_rate": 0.0001999953898824911,
-      "loss": 46.0,
-      "step": 18990
-    },
-    {
-      "epoch": 3.058375941060429,
-      "grad_norm": 0.004123042803257704,
-      "learning_rate": 0.00019999538939669516,
-      "loss": 46.0,
-      "step": 18991
-    },
-    {
-      "epoch": 3.0585369781392164,
-      "grad_norm": 0.0032799781765788794,
-      "learning_rate": 0.00019999538891087362,
-      "loss": 46.0,
-      "step": 18992
-    },
-    {
-      "epoch": 3.058698015218004,
-      "grad_norm": 0.000961768499109894,
-      "learning_rate": 0.00019999538842502648,
-      "loss": 46.0,
-      "step": 18993
-    },
-    {
-      "epoch": 3.0588590522967913,
-      "grad_norm": 0.0013210336910560727,
-      "learning_rate": 0.00019999538793915377,
-      "loss": 46.0,
-      "step": 18994
-    },
-    {
-      "epoch": 3.059020089375579,
-      "grad_norm": 0.015877913683652878,
-      "learning_rate": 0.00019999538745325545,
-      "loss": 46.0,
-      "step": 18995
-    },
-    {
-      "epoch": 3.0591811264543662,
-      "grad_norm": 0.002419903641566634,
-      "learning_rate": 0.0001999953869673315,
-      "loss": 46.0,
-      "step": 18996
-    },
-    {
-      "epoch": 3.0593421635331537,
-      "grad_norm": 0.002488799626007676,
-      "learning_rate": 0.000199995386481382,
-      "loss": 46.0,
-      "step": 18997
-    },
-    {
-      "epoch": 3.0595032006119407,
-      "grad_norm": 0.002177231013774872,
-      "learning_rate": 0.0001999953859954069,
-      "loss": 46.0,
-      "step": 18998
-    },
-    {
-      "epoch": 3.059664237690728,
-      "grad_norm": 0.008219020441174507,
-      "learning_rate": 0.0001999953855094062,
-      "loss": 46.0,
-      "step": 18999
-    },
-    {
-      "epoch": 3.0598252747695156,
-      "grad_norm": 0.00157107412815094,
-      "learning_rate": 0.00019999538502337992,
-      "loss": 46.0,
-      "step": 19000
-    },
-    {
-      "epoch": 3.059986311848303,
-      "grad_norm": 0.0013721369905397296,
-      "learning_rate": 0.00019999538453732804,
-      "loss": 46.0,
-      "step": 19001
-    },
-    {
-      "epoch": 3.0601473489270905,
-      "grad_norm": 0.003930696751922369,
-      "learning_rate": 0.00019999538405125056,
-      "loss": 46.0,
-      "step": 19002
-    },
-    {
-      "epoch": 3.060308386005878,
-      "grad_norm": 0.005053957924246788,
-      "learning_rate": 0.00019999538356514748,
-      "loss": 46.0,
-      "step": 19003
-    },
-    {
-      "epoch": 3.0604694230846654,
-      "grad_norm": 0.0066785081289708614,
-      "learning_rate": 0.00019999538307901885,
-      "loss": 46.0,
-      "step": 19004
-    },
-    {
-      "epoch": 3.060630460163453,
-      "grad_norm": 0.0007520662620663643,
-      "learning_rate": 0.00019999538259286457,
-      "loss": 46.0,
-      "step": 19005
-    },
-    {
-      "epoch": 3.06079149724224,
-      "grad_norm": 0.0018778488738462329,
-      "learning_rate": 0.0001999953821066847,
-      "loss": 46.0,
-      "step": 19006
-    },
-    {
-      "epoch": 3.0609525343210273,
-      "grad_norm": 0.002723777899518609,
-      "learning_rate": 0.0001999953816204793,
-      "loss": 46.0,
-      "step": 19007
-    },
-    {
-      "epoch": 3.0611135713998148,
-      "grad_norm": 0.0024173678830266,
-      "learning_rate": 0.00019999538113424822,
-      "loss": 46.0,
-      "step": 19008
-    },
-    {
-      "epoch": 3.061274608478602,
-      "grad_norm": 0.0023062312975525856,
-      "learning_rate": 0.0001999953806479916,
-      "loss": 46.0,
-      "step": 19009
-    },
-    {
-      "epoch": 3.0614356455573897,
-      "grad_norm": 0.005425425246357918,
-      "learning_rate": 0.0001999953801617094,
-      "loss": 46.0,
-      "step": 19010
-    },
-    {
-      "epoch": 3.061596682636177,
-      "grad_norm": 0.00368660781532526,
-      "learning_rate": 0.00019999537967540154,
-      "loss": 46.0,
-      "step": 19011
-    },
-    {
-      "epoch": 3.0617577197149646,
-      "grad_norm": 0.005451664328575134,
-      "learning_rate": 0.00019999537918906815,
-      "loss": 46.0,
-      "step": 19012
-    },
-    {
-      "epoch": 3.0619187567937516,
-      "grad_norm": 0.005232551135122776,
-      "learning_rate": 0.00019999537870270912,
-      "loss": 46.0,
-      "step": 19013
-    },
-    {
-      "epoch": 3.062079793872539,
-      "grad_norm": 0.002887663897126913,
-      "learning_rate": 0.00019999537821632454,
-      "loss": 46.0,
-      "step": 19014
-    },
-    {
-      "epoch": 3.0622408309513265,
-      "grad_norm": 0.0026603098958730698,
-      "learning_rate": 0.00019999537772991434,
-      "loss": 46.0,
-      "step": 19015
-    },
-    {
-      "epoch": 3.062401868030114,
-      "grad_norm": 0.0033915177918970585,
-      "learning_rate": 0.00019999537724347857,
-      "loss": 46.0,
-      "step": 19016
-    },
-    {
-      "epoch": 3.0625629051089014,
-      "grad_norm": 0.010073371231555939,
-      "learning_rate": 0.00019999537675701717,
-      "loss": 46.0,
-      "step": 19017
-    },
-    {
-      "epoch": 3.062723942187689,
-      "grad_norm": 0.011034855619072914,
-      "learning_rate": 0.00019999537627053018,
-      "loss": 46.0,
-      "step": 19018
-    },
-    {
-      "epoch": 3.0628849792664763,
-      "grad_norm": 0.002768860897049308,
-      "learning_rate": 0.0001999953757840176,
-      "loss": 46.0,
-      "step": 19019
-    },
-    {
-      "epoch": 3.0630460163452633,
-      "grad_norm": 0.0018350209575146437,
-      "learning_rate": 0.00019999537529747947,
-      "loss": 46.0,
-      "step": 19020
-    },
-    {
-      "epoch": 3.0632070534240508,
-      "grad_norm": 0.000755294575355947,
-      "learning_rate": 0.00019999537481091571,
-      "loss": 46.0,
-      "step": 19021
-    },
-    {
-      "epoch": 3.063368090502838,
-      "grad_norm": 0.001323820324614644,
-      "learning_rate": 0.00019999537432432635,
-      "loss": 46.0,
-      "step": 19022
-    },
-    {
-      "epoch": 3.0635291275816257,
-      "grad_norm": 0.002677367767319083,
-      "learning_rate": 0.0001999953738377114,
-      "loss": 46.0,
-      "step": 19023
-    },
-    {
-      "epoch": 3.063690164660413,
-      "grad_norm": 0.00567894522100687,
-      "learning_rate": 0.00019999537335107088,
-      "loss": 46.0,
-      "step": 19024
-    },
-    {
-      "epoch": 3.0638512017392006,
-      "grad_norm": 0.0025094083976000547,
-      "learning_rate": 0.00019999537286440476,
-      "loss": 46.0,
-      "step": 19025
-    },
-    {
-      "epoch": 3.064012238817988,
-      "grad_norm": 0.0041719041764736176,
-      "learning_rate": 0.00019999537237771301,
-      "loss": 46.0,
-      "step": 19026
-    },
-    {
-      "epoch": 3.064173275896775,
-      "grad_norm": 0.001999187283217907,
-      "learning_rate": 0.0001999953718909957,
-      "loss": 46.0,
-      "step": 19027
-    },
-    {
-      "epoch": 3.0643343129755625,
-      "grad_norm": 0.0008871268946677446,
-      "learning_rate": 0.0001999953714042528,
-      "loss": 46.0,
-      "step": 19028
-    },
-    {
-      "epoch": 3.06449535005435,
-      "grad_norm": 0.0022008302621543407,
-      "learning_rate": 0.0001999953709174843,
-      "loss": 46.0,
-      "step": 19029
-    },
-    {
-      "epoch": 3.0646563871331374,
-      "grad_norm": 0.0015959098236635327,
-      "learning_rate": 0.00019999537043069018,
-      "loss": 46.0,
-      "step": 19030
-    },
-    {
-      "epoch": 3.064817424211925,
-      "grad_norm": 0.005434796214103699,
-      "learning_rate": 0.0001999953699438705,
-      "loss": 46.0,
-      "step": 19031
-    },
-    {
-      "epoch": 3.0649784612907123,
-      "grad_norm": 0.0031477482989430428,
-      "learning_rate": 0.0001999953694570252,
-      "loss": 46.0,
-      "step": 19032
-    },
-    {
-      "epoch": 3.0651394983694997,
-      "grad_norm": 0.003985613118857145,
-      "learning_rate": 0.00019999536897015433,
-      "loss": 46.0,
-      "step": 19033
-    },
-    {
-      "epoch": 3.065300535448287,
-      "grad_norm": 0.0015886675100773573,
-      "learning_rate": 0.00019999536848325783,
-      "loss": 46.0,
-      "step": 19034
-    },
-    {
-      "epoch": 3.065461572527074,
-      "grad_norm": 0.003924034535884857,
-      "learning_rate": 0.00019999536799633578,
-      "loss": 46.0,
-      "step": 19035
-    },
-    {
-      "epoch": 3.0656226096058616,
-      "grad_norm": 0.005063912365585566,
-      "learning_rate": 0.0001999953675093881,
-      "loss": 46.0,
-      "step": 19036
-    },
-    {
-      "epoch": 3.065783646684649,
-      "grad_norm": 0.0011748622637242079,
-      "learning_rate": 0.00019999536702241486,
-      "loss": 46.0,
-      "step": 19037
-    },
-    {
-      "epoch": 3.0659446837634365,
-      "grad_norm": 0.003807569621130824,
-      "learning_rate": 0.000199995366535416,
-      "loss": 46.0,
-      "step": 19038
-    },
-    {
-      "epoch": 3.066105720842224,
-      "grad_norm": 0.002891937969252467,
-      "learning_rate": 0.00019999536604839156,
-      "loss": 46.0,
-      "step": 19039
-    },
-    {
-      "epoch": 3.0662667579210114,
-      "grad_norm": 0.004636032972484827,
-      "learning_rate": 0.00019999536556134152,
-      "loss": 46.0,
-      "step": 19040
-    },
-    {
-      "epoch": 3.066427794999799,
-      "grad_norm": 0.003022376913577318,
-      "learning_rate": 0.00019999536507426591,
-      "loss": 46.0,
-      "step": 19041
-    },
-    {
-      "epoch": 3.066588832078586,
-      "grad_norm": 0.0022735067177563906,
-      "learning_rate": 0.00019999536458716467,
-      "loss": 46.0,
-      "step": 19042
-    },
-    {
-      "epoch": 3.0667498691573734,
-      "grad_norm": 0.002569965785369277,
-      "learning_rate": 0.00019999536410003787,
-      "loss": 46.0,
-      "step": 19043
-    },
-    {
-      "epoch": 3.066910906236161,
-      "grad_norm": 0.004043196327984333,
-      "learning_rate": 0.00019999536361288545,
-      "loss": 46.0,
-      "step": 19044
-    },
-    {
-      "epoch": 3.0670719433149483,
-      "grad_norm": 0.0024684718810021877,
-      "learning_rate": 0.00019999536312570744,
-      "loss": 46.0,
-      "step": 19045
-    },
-    {
-      "epoch": 3.0672329803937357,
-      "grad_norm": 0.0024127268698066473,
-      "learning_rate": 0.00019999536263850382,
-      "loss": 46.0,
-      "step": 19046
-    },
-    {
-      "epoch": 3.067394017472523,
-      "grad_norm": 0.002998227719217539,
-      "learning_rate": 0.00019999536215127464,
-      "loss": 46.0,
-      "step": 19047
-    },
-    {
-      "epoch": 3.0675550545513106,
-      "grad_norm": 0.0041608987376093864,
-      "learning_rate": 0.00019999536166401987,
-      "loss": 46.0,
-      "step": 19048
-    },
-    {
-      "epoch": 3.0677160916300976,
-      "grad_norm": 0.0011057513765990734,
-      "learning_rate": 0.00019999536117673946,
-      "loss": 46.0,
-      "step": 19049
-    },
-    {
-      "epoch": 3.067877128708885,
-      "grad_norm": 0.007981859147548676,
-      "learning_rate": 0.0001999953606894335,
-      "loss": 46.0,
-      "step": 19050
-    },
-    {
-      "epoch": 3.0680381657876725,
-      "grad_norm": 0.004173414781689644,
-      "learning_rate": 0.00019999536020210194,
-      "loss": 46.0,
-      "step": 19051
-    },
-    {
-      "epoch": 3.06819920286646,
-      "grad_norm": 0.007655048277229071,
-      "learning_rate": 0.00019999535971474477,
-      "loss": 46.0,
-      "step": 19052
-    },
-    {
-      "epoch": 3.0683602399452474,
-      "grad_norm": 0.006797769572585821,
-      "learning_rate": 0.000199995359227362,
-      "loss": 46.0,
-      "step": 19053
-    },
-    {
-      "epoch": 3.068521277024035,
-      "grad_norm": 0.004133382346481085,
-      "learning_rate": 0.00019999535873995366,
-      "loss": 46.0,
-      "step": 19054
-    },
-    {
-      "epoch": 3.0686823141028223,
-      "grad_norm": 0.0009931714739650488,
-      "learning_rate": 0.00019999535825251973,
-      "loss": 46.0,
-      "step": 19055
-    },
-    {
-      "epoch": 3.06884335118161,
-      "grad_norm": 0.005739261396229267,
-      "learning_rate": 0.0001999953577650602,
-      "loss": 46.0,
-      "step": 19056
-    },
-    {
-      "epoch": 3.069004388260397,
-      "grad_norm": 0.004316538106650114,
-      "learning_rate": 0.00019999535727757505,
-      "loss": 46.0,
-      "step": 19057
-    },
-    {
-      "epoch": 3.0691654253391842,
-      "grad_norm": 0.0019707847386598587,
-      "learning_rate": 0.00019999535679006436,
-      "loss": 46.0,
-      "step": 19058
-    },
-    {
-      "epoch": 3.0693264624179717,
-      "grad_norm": 0.005369000602513552,
-      "learning_rate": 0.00019999535630252805,
-      "loss": 46.0,
-      "step": 19059
-    },
-    {
-      "epoch": 3.069487499496759,
-      "grad_norm": 0.00812921766191721,
-      "learning_rate": 0.0001999953558149661,
-      "loss": 46.0,
-      "step": 19060
-    },
-    {
-      "epoch": 3.0696485365755466,
-      "grad_norm": 0.010572869330644608,
-      "learning_rate": 0.00019999535532737862,
-      "loss": 46.0,
-      "step": 19061
-    },
-    {
-      "epoch": 3.069809573654334,
-      "grad_norm": 0.002061629667878151,
-      "learning_rate": 0.0001999953548397655,
-      "loss": 46.0,
-      "step": 19062
-    },
-    {
-      "epoch": 3.0699706107331215,
-      "grad_norm": 0.005293956026434898,
-      "learning_rate": 0.0001999953543521268,
-      "loss": 46.0,
-      "step": 19063
-    },
-    {
-      "epoch": 3.0701316478119085,
-      "grad_norm": 0.0014090079348534346,
-      "learning_rate": 0.00019999535386446257,
-      "loss": 46.0,
-      "step": 19064
-    },
-    {
-      "epoch": 3.070292684890696,
-      "grad_norm": 0.0025850203819572926,
-      "learning_rate": 0.00019999535337677265,
-      "loss": 46.0,
-      "step": 19065
-    },
-    {
-      "epoch": 3.0704537219694834,
-      "grad_norm": 0.0011749042896553874,
-      "learning_rate": 0.0001999953528890572,
-      "loss": 46.0,
-      "step": 19066
-    },
-    {
-      "epoch": 3.070614759048271,
-      "grad_norm": 0.014405600726604462,
-      "learning_rate": 0.00019999535240131612,
-      "loss": 46.0,
-      "step": 19067
-    },
-    {
-      "epoch": 3.0707757961270583,
-      "grad_norm": 0.0006996396114118397,
-      "learning_rate": 0.00019999535191354947,
-      "loss": 46.0,
-      "step": 19068
-    },
-    {
-      "epoch": 3.0709368332058458,
-      "grad_norm": 0.0014793199952691793,
-      "learning_rate": 0.0001999953514257572,
-      "loss": 46.0,
-      "step": 19069
-    },
-    {
-      "epoch": 3.071097870284633,
-      "grad_norm": 0.002936831908300519,
-      "learning_rate": 0.00019999535093793937,
-      "loss": 46.0,
-      "step": 19070
-    },
-    {
-      "epoch": 3.07125890736342,
-      "grad_norm": 0.0036574266850948334,
-      "learning_rate": 0.00019999535045009593,
-      "loss": 46.0,
-      "step": 19071
-    },
-    {
-      "epoch": 3.0714199444422077,
-      "grad_norm": 0.008239139802753925,
-      "learning_rate": 0.00019999534996222688,
-      "loss": 46.0,
-      "step": 19072
-    },
-    {
-      "epoch": 3.071580981520995,
-      "grad_norm": 0.00396543787792325,
-      "learning_rate": 0.00019999534947433224,
-      "loss": 46.0,
-      "step": 19073
-    },
-    {
-      "epoch": 3.0717420185997826,
-      "grad_norm": 0.0028600955847650766,
-      "learning_rate": 0.00019999534898641205,
-      "loss": 46.0,
-      "step": 19074
-    },
-    {
-      "epoch": 3.07190305567857,
-      "grad_norm": 0.0016765448963269591,
-      "learning_rate": 0.0001999953484984662,
-      "loss": 46.0,
-      "step": 19075
-    },
-    {
-      "epoch": 3.0720640927573575,
-      "grad_norm": 0.0019038845784962177,
-      "learning_rate": 0.0001999953480104948,
-      "loss": 46.0,
-      "step": 19076
-    },
-    {
-      "epoch": 3.072225129836145,
-      "grad_norm": 0.005017343908548355,
-      "learning_rate": 0.00019999534752249782,
-      "loss": 46.0,
-      "step": 19077
-    },
-    {
-      "epoch": 3.0723861669149324,
-      "grad_norm": 0.002789796097204089,
-      "learning_rate": 0.00019999534703447522,
-      "loss": 46.0,
-      "step": 19078
-    },
-    {
-      "epoch": 3.0725472039937194,
-      "grad_norm": 0.013695823960006237,
-      "learning_rate": 0.000199995346546427,
-      "loss": 46.0,
-      "step": 19079
-    },
-    {
-      "epoch": 3.072708241072507,
-      "grad_norm": 0.0012747837463393807,
-      "learning_rate": 0.00019999534605835324,
-      "loss": 46.0,
-      "step": 19080
-    },
-    {
-      "epoch": 3.0728692781512943,
-      "grad_norm": 0.0010314909741282463,
-      "learning_rate": 0.00019999534557025385,
-      "loss": 46.0,
-      "step": 19081
-    },
-    {
-      "epoch": 3.0730303152300817,
-      "grad_norm": 0.0006177301402203739,
-      "learning_rate": 0.0001999953450821289,
-      "loss": 46.0,
-      "step": 19082
-    },
-    {
-      "epoch": 3.073191352308869,
-      "grad_norm": 0.0034139796625822783,
-      "learning_rate": 0.00019999534459397834,
-      "loss": 46.0,
-      "step": 19083
-    },
-    {
-      "epoch": 3.0733523893876566,
-      "grad_norm": 0.003577446099370718,
-      "learning_rate": 0.00019999534410580216,
-      "loss": 46.0,
-      "step": 19084
-    },
-    {
-      "epoch": 3.073513426466444,
-      "grad_norm": 0.008248341269791126,
-      "learning_rate": 0.00019999534361760042,
-      "loss": 46.0,
-      "step": 19085
-    },
-    {
-      "epoch": 3.073674463545231,
-      "grad_norm": 0.00657343864440918,
-      "learning_rate": 0.00019999534312937307,
-      "loss": 46.0,
-      "step": 19086
-    },
-    {
-      "epoch": 3.0738355006240186,
-      "grad_norm": 0.00760771194472909,
-      "learning_rate": 0.00019999534264112013,
-      "loss": 46.0,
-      "step": 19087
-    },
-    {
-      "epoch": 3.073996537702806,
-      "grad_norm": 0.0028835826087743044,
-      "learning_rate": 0.00019999534215284158,
-      "loss": 46.0,
-      "step": 19088
-    },
-    {
-      "epoch": 3.0741575747815935,
-      "grad_norm": 0.001613772357814014,
-      "learning_rate": 0.00019999534166453747,
-      "loss": 46.0,
-      "step": 19089
-    },
-    {
-      "epoch": 3.074318611860381,
-      "grad_norm": 0.0063636754639446735,
-      "learning_rate": 0.00019999534117620774,
-      "loss": 46.0,
-      "step": 19090
-    },
-    {
-      "epoch": 3.0744796489391684,
-      "grad_norm": 0.0012422911822795868,
-      "learning_rate": 0.00019999534068785243,
-      "loss": 46.0,
-      "step": 19091
-    },
-    {
-      "epoch": 3.074640686017956,
-      "grad_norm": 0.017037486657500267,
-      "learning_rate": 0.00019999534019947152,
-      "loss": 46.0,
-      "step": 19092
-    },
-    {
-      "epoch": 3.074801723096743,
-      "grad_norm": 0.0009254264296032488,
-      "learning_rate": 0.00019999533971106504,
-      "loss": 46.0,
-      "step": 19093
-    },
-    {
-      "epoch": 3.0749627601755303,
-      "grad_norm": 0.0015830083284527063,
-      "learning_rate": 0.00019999533922263293,
-      "loss": 46.0,
-      "step": 19094
-    },
-    {
-      "epoch": 3.0751237972543177,
-      "grad_norm": 0.0027457610704004765,
-      "learning_rate": 0.00019999533873417524,
-      "loss": 46.0,
-      "step": 19095
-    },
-    {
-      "epoch": 3.075284834333105,
-      "grad_norm": 0.0009954030392691493,
-      "learning_rate": 0.00019999533824569197,
-      "loss": 46.0,
-      "step": 19096
-    },
-    {
-      "epoch": 3.0754458714118926,
-      "grad_norm": 0.0013651358895003796,
-      "learning_rate": 0.0001999953377571831,
-      "loss": 46.0,
-      "step": 19097
-    },
-    {
-      "epoch": 3.07560690849068,
-      "grad_norm": 0.005430291406810284,
-      "learning_rate": 0.0001999953372686486,
-      "loss": 46.0,
-      "step": 19098
-    },
-    {
-      "epoch": 3.0757679455694675,
-      "grad_norm": 0.006404591724276543,
-      "learning_rate": 0.00019999533678008856,
-      "loss": 46.0,
-      "step": 19099
-    },
-    {
-      "epoch": 3.0759289826482545,
-      "grad_norm": 0.0009197668405249715,
-      "learning_rate": 0.00019999533629150288,
-      "loss": 46.0,
-      "step": 19100
-    },
-    {
-      "epoch": 3.076090019727042,
-      "grad_norm": 0.002305516740307212,
-      "learning_rate": 0.00019999533580289164,
-      "loss": 46.0,
-      "step": 19101
-    },
-    {
-      "epoch": 3.0762510568058294,
-      "grad_norm": 0.0021574273705482483,
-      "learning_rate": 0.00019999533531425478,
-      "loss": 46.0,
-      "step": 19102
-    },
-    {
-      "epoch": 3.076412093884617,
-      "grad_norm": 0.0009340643882751465,
-      "learning_rate": 0.00019999533482559234,
-      "loss": 46.0,
-      "step": 19103
-    },
-    {
-      "epoch": 3.0765731309634043,
-      "grad_norm": 0.0005300783668644726,
-      "learning_rate": 0.0001999953343369043,
-      "loss": 46.0,
-      "step": 19104
-    },
-    {
-      "epoch": 3.076734168042192,
-      "grad_norm": 0.0010991146555170417,
-      "learning_rate": 0.0001999953338481907,
-      "loss": 46.0,
-      "step": 19105
-    },
-    {
-      "epoch": 3.0768952051209792,
-      "grad_norm": 0.002562117064371705,
-      "learning_rate": 0.0001999953333594515,
-      "loss": 46.0,
-      "step": 19106
-    },
-    {
-      "epoch": 3.0770562421997667,
-      "grad_norm": 0.003121523419395089,
-      "learning_rate": 0.00019999533287068665,
-      "loss": 46.0,
-      "step": 19107
-    },
-    {
-      "epoch": 3.0772172792785537,
-      "grad_norm": 0.0018334800843149424,
-      "learning_rate": 0.00019999533238189624,
-      "loss": 46.0,
-      "step": 19108
-    },
-    {
-      "epoch": 3.077378316357341,
-      "grad_norm": 0.002573322504758835,
-      "learning_rate": 0.00019999533189308025,
-      "loss": 46.0,
-      "step": 19109
-    },
-    {
-      "epoch": 3.0775393534361286,
-      "grad_norm": 0.00606690626591444,
-      "learning_rate": 0.00019999533140423864,
-      "loss": 46.0,
-      "step": 19110
-    },
-    {
-      "epoch": 3.077700390514916,
-      "grad_norm": 0.007160976063460112,
-      "learning_rate": 0.00019999533091537148,
-      "loss": 46.0,
-      "step": 19111
-    },
-    {
-      "epoch": 3.0778614275937035,
-      "grad_norm": 0.005840762052685022,
-      "learning_rate": 0.0001999953304264787,
-      "loss": 46.0,
-      "step": 19112
-    },
-    {
-      "epoch": 3.078022464672491,
-      "grad_norm": 0.0024405154399573803,
-      "learning_rate": 0.0001999953299375603,
-      "loss": 46.0,
-      "step": 19113
-    },
-    {
-      "epoch": 3.0781835017512784,
-      "grad_norm": 0.0037817552220076323,
-      "learning_rate": 0.00019999532944861635,
-      "loss": 46.0,
-      "step": 19114
-    },
-    {
-      "epoch": 3.0783445388300654,
-      "grad_norm": 0.0029663031455129385,
-      "learning_rate": 0.00019999532895964678,
-      "loss": 46.0,
-      "step": 19115
-    },
-    {
-      "epoch": 3.078505575908853,
-      "grad_norm": 0.0030656771268695593,
-      "learning_rate": 0.00019999532847065162,
-      "loss": 46.0,
-      "step": 19116
-    },
-    {
-      "epoch": 3.0786666129876403,
-      "grad_norm": 0.0016240831464529037,
-      "learning_rate": 0.00019999532798163088,
-      "loss": 46.0,
-      "step": 19117
-    },
-    {
-      "epoch": 3.078827650066428,
-      "grad_norm": 0.00464076129719615,
-      "learning_rate": 0.00019999532749258455,
-      "loss": 46.0,
-      "step": 19118
-    },
-    {
-      "epoch": 3.0789886871452152,
-      "grad_norm": 0.0023065260611474514,
-      "learning_rate": 0.0001999953270035126,
-      "loss": 46.0,
-      "step": 19119
-    },
-    {
-      "epoch": 3.0791497242240027,
-      "grad_norm": 0.005764540750533342,
-      "learning_rate": 0.00019999532651441507,
-      "loss": 46.0,
-      "step": 19120
-    },
-    {
-      "epoch": 3.07931076130279,
-      "grad_norm": 0.0040515111759305,
-      "learning_rate": 0.00019999532602529195,
-      "loss": 46.0,
-      "step": 19121
-    },
-    {
-      "epoch": 3.0794717983815776,
-      "grad_norm": 0.007397829554975033,
-      "learning_rate": 0.00019999532553614322,
-      "loss": 46.0,
-      "step": 19122
-    },
-    {
-      "epoch": 3.0796328354603646,
-      "grad_norm": 0.003961391746997833,
-      "learning_rate": 0.00019999532504696893,
-      "loss": 46.0,
-      "step": 19123
-    },
-    {
-      "epoch": 3.079793872539152,
-      "grad_norm": 0.0016536946641281247,
-      "learning_rate": 0.000199995324557769,
-      "loss": 46.0,
-      "step": 19124
-    },
-    {
-      "epoch": 3.0799549096179395,
-      "grad_norm": 0.0029958044178783894,
-      "learning_rate": 0.00019999532406854352,
-      "loss": 46.0,
-      "step": 19125
-    },
-    {
-      "epoch": 3.080115946696727,
-      "grad_norm": 0.0008020913228392601,
-      "learning_rate": 0.00019999532357929241,
-      "loss": 46.0,
-      "step": 19126
-    },
-    {
-      "epoch": 3.0802769837755144,
-      "grad_norm": 0.0029873119201511145,
-      "learning_rate": 0.00019999532309001575,
-      "loss": 46.0,
-      "step": 19127
-    },
-    {
-      "epoch": 3.080438020854302,
-      "grad_norm": 0.0005914492066949606,
-      "learning_rate": 0.00019999532260071346,
-      "loss": 46.0,
-      "step": 19128
-    },
-    {
-      "epoch": 3.0805990579330893,
-      "grad_norm": 0.010760613717138767,
-      "learning_rate": 0.0001999953221113856,
-      "loss": 46.0,
-      "step": 19129
-    },
-    {
-      "epoch": 3.0807600950118763,
-      "grad_norm": 0.003076407825574279,
-      "learning_rate": 0.0001999953216220321,
-      "loss": 46.0,
-      "step": 19130
-    },
-    {
-      "epoch": 3.0809211320906638,
-      "grad_norm": 0.008108134381473064,
-      "learning_rate": 0.00019999532113265303,
-      "loss": 46.0,
-      "step": 19131
-    },
-    {
-      "epoch": 3.081082169169451,
-      "grad_norm": 0.0010789695661514997,
-      "learning_rate": 0.00019999532064324837,
-      "loss": 46.0,
-      "step": 19132
-    },
-    {
-      "epoch": 3.0812432062482387,
-      "grad_norm": 0.0018693257588893175,
-      "learning_rate": 0.00019999532015381815,
-      "loss": 46.0,
-      "step": 19133
-    },
-    {
-      "epoch": 3.081404243327026,
-      "grad_norm": 0.006474814377725124,
-      "learning_rate": 0.0001999953196643623,
-      "loss": 46.0,
-      "step": 19134
-    },
-    {
-      "epoch": 3.0815652804058136,
-      "grad_norm": 0.0029409306589514017,
-      "learning_rate": 0.00019999531917488085,
-      "loss": 46.0,
-      "step": 19135
-    },
-    {
-      "epoch": 3.081726317484601,
-      "grad_norm": 0.010037630796432495,
-      "learning_rate": 0.00019999531868537384,
-      "loss": 46.0,
-      "step": 19136
-    },
-    {
-      "epoch": 3.081887354563388,
-      "grad_norm": 0.003775344928726554,
-      "learning_rate": 0.00019999531819584122,
-      "loss": 46.0,
-      "step": 19137
-    },
-    {
-      "epoch": 3.0820483916421755,
-      "grad_norm": 0.001168327871710062,
-      "learning_rate": 0.00019999531770628298,
-      "loss": 46.0,
-      "step": 19138
-    },
-    {
-      "epoch": 3.082209428720963,
-      "grad_norm": 0.005866238847374916,
-      "learning_rate": 0.00019999531721669918,
-      "loss": 46.0,
-      "step": 19139
-    },
-    {
-      "epoch": 3.0823704657997504,
-      "grad_norm": 0.0037726270966231823,
-      "learning_rate": 0.00019999531672708977,
-      "loss": 46.0,
-      "step": 19140
-    },
-    {
-      "epoch": 3.082531502878538,
-      "grad_norm": 0.0007196885999292135,
-      "learning_rate": 0.00019999531623745477,
-      "loss": 46.0,
-      "step": 19141
-    },
-    {
-      "epoch": 3.0826925399573253,
-      "grad_norm": 0.0031213753391057253,
-      "learning_rate": 0.0001999953157477942,
-      "loss": 46.0,
-      "step": 19142
-    },
-    {
-      "epoch": 3.0828535770361127,
-      "grad_norm": 0.0013955564936622977,
-      "learning_rate": 0.000199995315258108,
-      "loss": 46.0,
-      "step": 19143
-    },
-    {
-      "epoch": 3.0830146141148997,
-      "grad_norm": 0.012316612526774406,
-      "learning_rate": 0.0001999953147683962,
-      "loss": 46.0,
-      "step": 19144
-    },
-    {
-      "epoch": 3.083175651193687,
-      "grad_norm": 0.002138487994670868,
-      "learning_rate": 0.00019999531427865885,
-      "loss": 46.0,
-      "step": 19145
-    },
-    {
-      "epoch": 3.0833366882724746,
-      "grad_norm": 0.0034975174348801374,
-      "learning_rate": 0.00019999531378889587,
-      "loss": 46.0,
-      "step": 19146
-    },
-    {
-      "epoch": 3.083497725351262,
-      "grad_norm": 0.0024908240884542465,
-      "learning_rate": 0.00019999531329910732,
-      "loss": 46.0,
-      "step": 19147
-    },
-    {
-      "epoch": 3.0836587624300495,
-      "grad_norm": 0.004345782566815615,
-      "learning_rate": 0.00019999531280929316,
-      "loss": 46.0,
-      "step": 19148
-    },
-    {
-      "epoch": 3.083819799508837,
-      "grad_norm": 0.001221984624862671,
-      "learning_rate": 0.00019999531231945343,
-      "loss": 46.0,
-      "step": 19149
-    },
-    {
-      "epoch": 3.0839808365876245,
-      "grad_norm": 0.0036923657171428204,
-      "learning_rate": 0.0001999953118295881,
-      "loss": 46.0,
-      "step": 19150
-    },
-    {
-      "epoch": 3.084141873666412,
-      "grad_norm": 0.00238083116710186,
-      "learning_rate": 0.00019999531133969715,
-      "loss": 46.0,
-      "step": 19151
-    },
-    {
-      "epoch": 3.084302910745199,
-      "grad_norm": 0.0021271277219057083,
-      "learning_rate": 0.0001999953108497806,
-      "loss": 46.0,
-      "step": 19152
-    },
-    {
-      "epoch": 3.0844639478239864,
-      "grad_norm": 0.005991783458739519,
-      "learning_rate": 0.00019999531035983848,
-      "loss": 46.0,
-      "step": 19153
-    },
-    {
-      "epoch": 3.084624984902774,
-      "grad_norm": 0.002737136324867606,
-      "learning_rate": 0.00019999530986987077,
-      "loss": 46.0,
-      "step": 19154
-    },
-    {
-      "epoch": 3.0847860219815613,
-      "grad_norm": 0.005372087936848402,
-      "learning_rate": 0.00019999530937987747,
-      "loss": 46.0,
-      "step": 19155
-    },
-    {
-      "epoch": 3.0849470590603487,
-      "grad_norm": 0.003594535170122981,
-      "learning_rate": 0.00019999530888985855,
-      "loss": 46.0,
-      "step": 19156
-    },
-    {
-      "epoch": 3.085108096139136,
-      "grad_norm": 0.0014518267707899213,
-      "learning_rate": 0.00019999530839981405,
-      "loss": 46.0,
-      "step": 19157
-    },
-    {
-      "epoch": 3.0852691332179236,
-      "grad_norm": 0.0033178445883095264,
-      "learning_rate": 0.000199995307909744,
-      "loss": 46.0,
-      "step": 19158
-    },
-    {
-      "epoch": 3.0854301702967106,
-      "grad_norm": 0.005538743920624256,
-      "learning_rate": 0.0001999953074196483,
-      "loss": 46.0,
-      "step": 19159
-    },
-    {
-      "epoch": 3.085591207375498,
-      "grad_norm": 0.0009418170084245503,
-      "learning_rate": 0.00019999530692952703,
-      "loss": 46.0,
-      "step": 19160
-    },
-    {
-      "epoch": 3.0857522444542855,
-      "grad_norm": 0.005124872550368309,
-      "learning_rate": 0.00019999530643938015,
-      "loss": 46.0,
-      "step": 19161
-    },
-    {
-      "epoch": 3.085913281533073,
-      "grad_norm": 0.0015647117979824543,
-      "learning_rate": 0.00019999530594920769,
-      "loss": 46.0,
-      "step": 19162
-    },
-    {
-      "epoch": 3.0860743186118604,
-      "grad_norm": 0.0010755390394479036,
-      "learning_rate": 0.0001999953054590096,
-      "loss": 46.0,
-      "step": 19163
-    },
-    {
-      "epoch": 3.086235355690648,
-      "grad_norm": 0.003597024828195572,
-      "learning_rate": 0.00019999530496878597,
-      "loss": 46.0,
-      "step": 19164
-    },
-    {
-      "epoch": 3.0863963927694353,
-      "grad_norm": 0.005692549515515566,
-      "learning_rate": 0.00019999530447853672,
-      "loss": 46.0,
-      "step": 19165
-    },
-    {
-      "epoch": 3.0865574298482223,
-      "grad_norm": 0.00314165442250669,
-      "learning_rate": 0.00019999530398826188,
-      "loss": 46.0,
-      "step": 19166
-    },
-    {
-      "epoch": 3.08671846692701,
-      "grad_norm": 0.0061389184556901455,
-      "learning_rate": 0.00019999530349796145,
-      "loss": 46.0,
-      "step": 19167
-    },
-    {
-      "epoch": 3.0868795040057972,
-      "grad_norm": 0.005200805142521858,
-      "learning_rate": 0.00019999530300763543,
-      "loss": 46.0,
-      "step": 19168
-    },
-    {
-      "epoch": 3.0870405410845847,
-      "grad_norm": 0.011876535601913929,
-      "learning_rate": 0.0001999953025172838,
-      "loss": 46.0,
-      "step": 19169
-    },
-    {
-      "epoch": 3.087201578163372,
-      "grad_norm": 0.003240771358832717,
-      "learning_rate": 0.00019999530202690656,
-      "loss": 46.0,
-      "step": 19170
-    },
-    {
-      "epoch": 3.0873626152421596,
-      "grad_norm": 0.00891954731196165,
-      "learning_rate": 0.00019999530153650379,
-      "loss": 46.0,
-      "step": 19171
-    },
-    {
-      "epoch": 3.087523652320947,
-      "grad_norm": 0.002669870387762785,
-      "learning_rate": 0.00019999530104607537,
-      "loss": 46.0,
-      "step": 19172
-    },
-    {
-      "epoch": 3.087684689399734,
-      "grad_norm": 0.0055496483109891415,
-      "learning_rate": 0.0001999953005556214,
-      "loss": 46.0,
-      "step": 19173
-    },
-    {
-      "epoch": 3.0878457264785215,
-      "grad_norm": 0.004981720820069313,
-      "learning_rate": 0.00019999530006514177,
-      "loss": 46.0,
-      "step": 19174
-    },
-    {
-      "epoch": 3.088006763557309,
-      "grad_norm": 0.0006577274180017412,
-      "learning_rate": 0.00019999529957463662,
-      "loss": 46.0,
-      "step": 19175
-    },
-    {
-      "epoch": 3.0881678006360964,
-      "grad_norm": 0.0017516218358650804,
-      "learning_rate": 0.00019999529908410583,
-      "loss": 46.0,
-      "step": 19176
-    },
-    {
-      "epoch": 3.088328837714884,
-      "grad_norm": 0.0026046959683299065,
-      "learning_rate": 0.00019999529859354947,
-      "loss": 46.0,
-      "step": 19177
-    },
-    {
-      "epoch": 3.0884898747936713,
-      "grad_norm": 0.004757405258715153,
-      "learning_rate": 0.00019999529810296748,
-      "loss": 46.0,
-      "step": 19178
-    },
-    {
-      "epoch": 3.0886509118724588,
-      "grad_norm": 0.011958839371800423,
-      "learning_rate": 0.00019999529761235993,
-      "loss": 46.0,
-      "step": 19179
-    },
-    {
-      "epoch": 3.0888119489512462,
-      "grad_norm": 0.0006466091726906598,
-      "learning_rate": 0.0001999952971217268,
-      "loss": 46.0,
-      "step": 19180
-    },
-    {
-      "epoch": 3.0889729860300332,
-      "grad_norm": 0.0010535804321989417,
-      "learning_rate": 0.00019999529663106803,
-      "loss": 46.0,
-      "step": 19181
-    },
-    {
-      "epoch": 3.0891340231088207,
-      "grad_norm": 0.0012020752765238285,
-      "learning_rate": 0.0001999952961403837,
-      "loss": 46.0,
-      "step": 19182
-    },
-    {
-      "epoch": 3.089295060187608,
-      "grad_norm": 0.007841324433684349,
-      "learning_rate": 0.00019999529564967378,
-      "loss": 46.0,
-      "step": 19183
-    },
-    {
-      "epoch": 3.0894560972663956,
-      "grad_norm": 0.004968698602169752,
-      "learning_rate": 0.00019999529515893824,
-      "loss": 46.0,
-      "step": 19184
-    },
-    {
-      "epoch": 3.089617134345183,
-      "grad_norm": 0.002287528244778514,
-      "learning_rate": 0.00019999529466817713,
-      "loss": 46.0,
-      "step": 19185
-    },
-    {
-      "epoch": 3.0897781714239705,
-      "grad_norm": 0.001124705420807004,
-      "learning_rate": 0.0001999952941773904,
-      "loss": 46.0,
-      "step": 19186
-    },
-    {
-      "epoch": 3.089939208502758,
-      "grad_norm": 0.0076405955478549,
-      "learning_rate": 0.0001999952936865781,
-      "loss": 46.0,
-      "step": 19187
-    },
-    {
-      "epoch": 3.090100245581545,
-      "grad_norm": 0.004483805038034916,
-      "learning_rate": 0.0001999952931957402,
-      "loss": 46.0,
-      "step": 19188
-    },
-    {
-      "epoch": 3.0902612826603324,
-      "grad_norm": 0.002658623270690441,
-      "learning_rate": 0.0001999952927048767,
-      "loss": 46.0,
-      "step": 19189
-    },
-    {
-      "epoch": 3.09042231973912,
-      "grad_norm": 0.002346122870221734,
-      "learning_rate": 0.00019999529221398764,
-      "loss": 46.0,
-      "step": 19190
-    },
-    {
-      "epoch": 3.0905833568179073,
-      "grad_norm": 0.0032003053929656744,
-      "learning_rate": 0.00019999529172307293,
-      "loss": 46.0,
-      "step": 19191
-    },
-    {
-      "epoch": 3.0907443938966948,
-      "grad_norm": 0.007972792722284794,
-      "learning_rate": 0.0001999952912321327,
-      "loss": 46.0,
-      "step": 19192
-    },
-    {
-      "epoch": 3.090905430975482,
-      "grad_norm": 0.002378620905801654,
-      "learning_rate": 0.0001999952907411668,
-      "loss": 46.0,
-      "step": 19193
-    },
-    {
-      "epoch": 3.0910664680542697,
-      "grad_norm": 0.0011053212219849229,
-      "learning_rate": 0.00019999529025017533,
-      "loss": 46.0,
-      "step": 19194
-    },
-    {
-      "epoch": 3.091227505133057,
-      "grad_norm": 0.0009945402853190899,
-      "learning_rate": 0.00019999528975915828,
-      "loss": 46.0,
-      "step": 19195
-    },
-    {
-      "epoch": 3.091388542211844,
-      "grad_norm": 0.000790278660133481,
-      "learning_rate": 0.00019999528926811563,
-      "loss": 46.0,
-      "step": 19196
-    },
-    {
-      "epoch": 3.0915495792906316,
-      "grad_norm": 0.002588866511359811,
-      "learning_rate": 0.00019999528877704737,
-      "loss": 46.0,
-      "step": 19197
-    },
-    {
-      "epoch": 3.091710616369419,
-      "grad_norm": 0.0033637951128184795,
-      "learning_rate": 0.00019999528828595355,
-      "loss": 46.0,
-      "step": 19198
-    },
-    {
-      "epoch": 3.0918716534482065,
-      "grad_norm": 0.002044109860435128,
-      "learning_rate": 0.00019999528779483412,
-      "loss": 46.0,
-      "step": 19199
-    },
-    {
-      "epoch": 3.092032690526994,
-      "grad_norm": 0.0008687538793310523,
-      "learning_rate": 0.00019999528730368907,
-      "loss": 46.0,
-      "step": 19200
-    },
-    {
-      "epoch": 3.0921937276057814,
-      "grad_norm": 0.005857252515852451,
-      "learning_rate": 0.00019999528681251846,
-      "loss": 46.0,
-      "step": 19201
-    },
-    {
-      "epoch": 3.092354764684569,
-      "grad_norm": 0.001143820583820343,
-      "learning_rate": 0.00019999528632132227,
-      "loss": 46.0,
-      "step": 19202
-    },
-    {
-      "epoch": 3.092515801763356,
-      "grad_norm": 0.004507384262979031,
-      "learning_rate": 0.00019999528583010046,
-      "loss": 46.0,
-      "step": 19203
-    },
-    {
-      "epoch": 3.0926768388421433,
-      "grad_norm": 0.0022199093364179134,
-      "learning_rate": 0.00019999528533885306,
-      "loss": 46.0,
-      "step": 19204
-    },
-    {
-      "epoch": 3.0928378759209307,
-      "grad_norm": 0.0016401053871959448,
-      "learning_rate": 0.00019999528484758008,
-      "loss": 46.0,
-      "step": 19205
-    },
-    {
-      "epoch": 3.092998912999718,
-      "grad_norm": 0.0032426516991108656,
-      "learning_rate": 0.00019999528435628148,
-      "loss": 46.0,
-      "step": 19206
-    },
-    {
-      "epoch": 3.0931599500785056,
-      "grad_norm": 0.001714535872451961,
-      "learning_rate": 0.0001999952838649573,
-      "loss": 46.0,
-      "step": 19207
-    },
-    {
-      "epoch": 3.093320987157293,
-      "grad_norm": 0.0034224370028823614,
-      "learning_rate": 0.00019999528337360752,
-      "loss": 46.0,
-      "step": 19208
-    },
-    {
-      "epoch": 3.0934820242360805,
-      "grad_norm": 0.003591538406908512,
-      "learning_rate": 0.00019999528288223214,
-      "loss": 46.0,
-      "step": 19209
-    },
-    {
-      "epoch": 3.0936430613148675,
-      "grad_norm": 0.005739274434745312,
-      "learning_rate": 0.00019999528239083122,
-      "loss": 46.0,
-      "step": 19210
-    },
-    {
-      "epoch": 3.093804098393655,
-      "grad_norm": 0.001627480611205101,
-      "learning_rate": 0.00019999528189940463,
-      "loss": 46.0,
-      "step": 19211
-    },
-    {
-      "epoch": 3.0939651354724425,
-      "grad_norm": 0.0010919008636847138,
-      "learning_rate": 0.0001999952814079525,
-      "loss": 46.0,
-      "step": 19212
-    },
-    {
-      "epoch": 3.09412617255123,
-      "grad_norm": 0.008490863256156445,
-      "learning_rate": 0.00019999528091647477,
-      "loss": 46.0,
-      "step": 19213
-    },
-    {
-      "epoch": 3.0942872096300174,
-      "grad_norm": 0.005402341019362211,
-      "learning_rate": 0.00019999528042497142,
-      "loss": 46.0,
-      "step": 19214
-    },
-    {
-      "epoch": 3.094448246708805,
-      "grad_norm": 0.0018791462061926723,
-      "learning_rate": 0.0001999952799334425,
-      "loss": 46.0,
-      "step": 19215
-    },
-    {
-      "epoch": 3.0946092837875923,
-      "grad_norm": 0.0056406366638839245,
-      "learning_rate": 0.00019999527944188796,
-      "loss": 46.0,
-      "step": 19216
-    },
-    {
-      "epoch": 3.0947703208663793,
-      "grad_norm": 0.0020144961308687925,
-      "learning_rate": 0.00019999527895030785,
-      "loss": 46.0,
-      "step": 19217
-    },
-    {
-      "epoch": 3.0949313579451667,
-      "grad_norm": 0.003603936405852437,
-      "learning_rate": 0.00019999527845870213,
-      "loss": 46.0,
-      "step": 19218
-    },
-    {
-      "epoch": 3.095092395023954,
-      "grad_norm": 0.001043665106408298,
-      "learning_rate": 0.00019999527796707084,
-      "loss": 46.0,
-      "step": 19219
-    },
-    {
-      "epoch": 3.0952534321027416,
-      "grad_norm": 0.006417769473046064,
-      "learning_rate": 0.00019999527747541397,
-      "loss": 46.0,
-      "step": 19220
-    },
-    {
-      "epoch": 3.095414469181529,
-      "grad_norm": 0.007026469800621271,
-      "learning_rate": 0.00019999527698373143,
-      "loss": 46.0,
-      "step": 19221
-    },
-    {
-      "epoch": 3.0955755062603165,
-      "grad_norm": 0.005228106398135424,
-      "learning_rate": 0.00019999527649202335,
-      "loss": 46.0,
-      "step": 19222
-    },
-    {
-      "epoch": 3.095736543339104,
-      "grad_norm": 0.0035425720270723104,
-      "learning_rate": 0.0001999952760002897,
-      "loss": 46.0,
-      "step": 19223
-    },
-    {
-      "epoch": 3.0958975804178914,
-      "grad_norm": 0.003815435105934739,
-      "learning_rate": 0.00019999527550853042,
-      "loss": 46.0,
-      "step": 19224
-    },
-    {
-      "epoch": 3.0960586174966784,
-      "grad_norm": 0.0014954148791730404,
-      "learning_rate": 0.00019999527501674553,
-      "loss": 46.0,
-      "step": 19225
-    },
-    {
-      "epoch": 3.096219654575466,
-      "grad_norm": 0.011047880165278912,
-      "learning_rate": 0.00019999527452493508,
-      "loss": 46.0,
-      "step": 19226
-    },
-    {
-      "epoch": 3.0963806916542533,
-      "grad_norm": 0.001661021844483912,
-      "learning_rate": 0.00019999527403309904,
-      "loss": 46.0,
-      "step": 19227
-    },
-    {
-      "epoch": 3.096541728733041,
-      "grad_norm": 0.0013593442272394896,
-      "learning_rate": 0.0001999952735412374,
-      "loss": 46.0,
-      "step": 19228
-    },
-    {
-      "epoch": 3.0967027658118282,
-      "grad_norm": 0.009929660707712173,
-      "learning_rate": 0.00019999527304935015,
-      "loss": 46.0,
-      "step": 19229
-    },
-    {
-      "epoch": 3.0968638028906157,
-      "grad_norm": 0.0025440549943596125,
-      "learning_rate": 0.0001999952725574373,
-      "loss": 46.0,
-      "step": 19230
-    },
-    {
-      "epoch": 3.097024839969403,
-      "grad_norm": 0.0015816837549209595,
-      "learning_rate": 0.00019999527206549889,
-      "loss": 46.0,
-      "step": 19231
-    },
-    {
-      "epoch": 3.09718587704819,
-      "grad_norm": 0.003044366603717208,
-      "learning_rate": 0.00019999527157353486,
-      "loss": 46.0,
-      "step": 19232
-    },
-    {
-      "epoch": 3.0973469141269776,
-      "grad_norm": 0.003409534227102995,
-      "learning_rate": 0.00019999527108154524,
-      "loss": 46.0,
-      "step": 19233
-    },
-    {
-      "epoch": 3.097507951205765,
-      "grad_norm": 0.0026044296100735664,
-      "learning_rate": 0.00019999527058953004,
-      "loss": 46.0,
-      "step": 19234
-    },
-    {
-      "epoch": 3.0976689882845525,
-      "grad_norm": 0.0004715148243121803,
-      "learning_rate": 0.00019999527009748923,
-      "loss": 46.0,
-      "step": 19235
-    },
-    {
-      "epoch": 3.09783002536334,
-      "grad_norm": 0.0017794615123420954,
-      "learning_rate": 0.00019999526960542285,
-      "loss": 46.0,
-      "step": 19236
-    },
-    {
-      "epoch": 3.0979910624421274,
-      "grad_norm": 0.0054541523568332195,
-      "learning_rate": 0.00019999526911333083,
-      "loss": 46.0,
-      "step": 19237
-    },
-    {
-      "epoch": 3.098152099520915,
-      "grad_norm": 0.004682239145040512,
-      "learning_rate": 0.00019999526862121328,
-      "loss": 46.0,
-      "step": 19238
-    },
-    {
-      "epoch": 3.098313136599702,
-      "grad_norm": 0.006904846057295799,
-      "learning_rate": 0.00019999526812907007,
-      "loss": 46.0,
-      "step": 19239
-    },
-    {
-      "epoch": 3.0984741736784893,
-      "grad_norm": 0.004432819318026304,
-      "learning_rate": 0.00019999526763690134,
-      "loss": 46.0,
-      "step": 19240
-    },
-    {
-      "epoch": 3.0986352107572768,
-      "grad_norm": 0.0037251566536724567,
-      "learning_rate": 0.00019999526714470695,
-      "loss": 46.0,
-      "step": 19241
-    },
-    {
-      "epoch": 3.098796247836064,
-      "grad_norm": 0.004127888940274715,
-      "learning_rate": 0.000199995266652487,
-      "loss": 46.0,
-      "step": 19242
-    },
-    {
-      "epoch": 3.0989572849148517,
-      "grad_norm": 0.0057748123072087765,
-      "learning_rate": 0.00019999526616024146,
-      "loss": 46.0,
-      "step": 19243
-    },
-    {
-      "epoch": 3.099118321993639,
-      "grad_norm": 0.0029337021987885237,
-      "learning_rate": 0.0001999952656679703,
-      "loss": 46.0,
-      "step": 19244
-    },
-    {
-      "epoch": 3.0992793590724266,
-      "grad_norm": 0.002829350531101227,
-      "learning_rate": 0.0001999952651756736,
-      "loss": 46.0,
-      "step": 19245
-    },
-    {
-      "epoch": 3.099440396151214,
-      "grad_norm": 0.003779337042942643,
-      "learning_rate": 0.0001999952646833512,
-      "loss": 46.0,
-      "step": 19246
-    },
-    {
-      "epoch": 3.099601433230001,
-      "grad_norm": 0.009024163708090782,
-      "learning_rate": 0.00019999526419100331,
-      "loss": 46.0,
-      "step": 19247
-    },
-    {
-      "epoch": 3.0997624703087885,
-      "grad_norm": 0.004311509430408478,
-      "learning_rate": 0.00019999526369862978,
-      "loss": 46.0,
-      "step": 19248
-    },
-    {
-      "epoch": 3.099923507387576,
-      "grad_norm": 0.0021545386407524347,
-      "learning_rate": 0.00019999526320623067,
-      "loss": 46.0,
-      "step": 19249
-    },
-    {
-      "epoch": 3.1000845444663634,
-      "grad_norm": 0.001936333836056292,
-      "learning_rate": 0.00019999526271380596,
-      "loss": 46.0,
-      "step": 19250
-    },
-    {
-      "epoch": 3.100245581545151,
-      "grad_norm": 0.002758914837613702,
-      "learning_rate": 0.00019999526222135567,
-      "loss": 46.0,
-      "step": 19251
-    },
-    {
-      "epoch": 3.1004066186239383,
-      "grad_norm": 0.0029531673062592745,
-      "learning_rate": 0.00019999526172887977,
-      "loss": 46.0,
-      "step": 19252
-    },
-    {
-      "epoch": 3.1005676557027257,
-      "grad_norm": 0.005109885707497597,
-      "learning_rate": 0.00019999526123637827,
-      "loss": 46.0,
-      "step": 19253
-    },
-    {
-      "epoch": 3.1007286927815128,
-      "grad_norm": 0.004946262110024691,
-      "learning_rate": 0.0001999952607438512,
-      "loss": 46.0,
-      "step": 19254
-    },
-    {
-      "epoch": 3.1008897298603,
-      "grad_norm": 0.0007333557005040348,
-      "learning_rate": 0.00019999526025129852,
-      "loss": 46.0,
-      "step": 19255
-    },
-    {
-      "epoch": 3.1010507669390877,
-      "grad_norm": 0.0016733099473640323,
-      "learning_rate": 0.00019999525975872024,
-      "loss": 46.0,
-      "step": 19256
-    },
-    {
-      "epoch": 3.101211804017875,
-      "grad_norm": 0.0009643870871514082,
-      "learning_rate": 0.00019999525926611637,
-      "loss": 46.0,
-      "step": 19257
-    },
-    {
-      "epoch": 3.1013728410966626,
-      "grad_norm": 0.005745363887399435,
-      "learning_rate": 0.00019999525877348692,
-      "loss": 46.0,
-      "step": 19258
-    },
-    {
-      "epoch": 3.10153387817545,
-      "grad_norm": 0.010808480903506279,
-      "learning_rate": 0.00019999525828083187,
-      "loss": 46.0,
-      "step": 19259
-    },
-    {
-      "epoch": 3.1016949152542375,
-      "grad_norm": 0.0015271289739757776,
-      "learning_rate": 0.00019999525778815124,
-      "loss": 46.0,
-      "step": 19260
-    },
-    {
-      "epoch": 3.1018559523330245,
-      "grad_norm": 0.0021282541565597057,
-      "learning_rate": 0.000199995257295445,
-      "loss": 46.0,
-      "step": 19261
-    },
-    {
-      "epoch": 3.102016989411812,
-      "grad_norm": 0.0022353986278176308,
-      "learning_rate": 0.00019999525680271317,
-      "loss": 46.0,
-      "step": 19262
-    },
-    {
-      "epoch": 3.1021780264905994,
-      "grad_norm": 0.0010247172322124243,
-      "learning_rate": 0.00019999525630995572,
-      "loss": 46.0,
-      "step": 19263
-    },
-    {
-      "epoch": 3.102339063569387,
-      "grad_norm": 0.0007396000437438488,
-      "learning_rate": 0.00019999525581717271,
-      "loss": 46.0,
-      "step": 19264
-    },
-    {
-      "epoch": 3.1025001006481743,
-      "grad_norm": 0.002451027510687709,
-      "learning_rate": 0.00019999525532436412,
-      "loss": 46.0,
-      "step": 19265
-    },
-    {
-      "epoch": 3.1026611377269617,
-      "grad_norm": 0.008563502691686153,
-      "learning_rate": 0.00019999525483152989,
-      "loss": 46.0,
-      "step": 19266
-    },
-    {
-      "epoch": 3.102822174805749,
-      "grad_norm": 0.0020906063728034496,
-      "learning_rate": 0.0001999952543386701,
-      "loss": 46.0,
-      "step": 19267
-    },
-    {
-      "epoch": 3.1029832118845366,
-      "grad_norm": 0.005402501206845045,
-      "learning_rate": 0.00019999525384578474,
-      "loss": 46.0,
-      "step": 19268
-    },
-    {
-      "epoch": 3.1031442489633236,
-      "grad_norm": 0.0021859346888959408,
-      "learning_rate": 0.0001999952533528737,
-      "loss": 46.0,
-      "step": 19269
-    },
-    {
-      "epoch": 3.103305286042111,
-      "grad_norm": 0.011181621812283993,
-      "learning_rate": 0.00019999525285993716,
-      "loss": 46.0,
-      "step": 19270
-    },
-    {
-      "epoch": 3.1034663231208985,
-      "grad_norm": 0.0070154485292732716,
-      "learning_rate": 0.00019999525236697499,
-      "loss": 46.0,
-      "step": 19271
-    },
-    {
-      "epoch": 3.103627360199686,
-      "grad_norm": 0.0041419873014092445,
-      "learning_rate": 0.0001999952518739872,
-      "loss": 46.0,
-      "step": 19272
-    },
-    {
-      "epoch": 3.1037883972784734,
-      "grad_norm": 0.005718898959457874,
-      "learning_rate": 0.00019999525138097383,
-      "loss": 46.0,
-      "step": 19273
-    },
-    {
-      "epoch": 3.103949434357261,
-      "grad_norm": 0.003628492821007967,
-      "learning_rate": 0.0001999952508879349,
-      "loss": 46.0,
-      "step": 19274
-    },
-    {
-      "epoch": 3.1041104714360483,
-      "grad_norm": 0.003389259334653616,
-      "learning_rate": 0.00019999525039487032,
-      "loss": 46.0,
-      "step": 19275
-    },
-    {
-      "epoch": 3.1042715085148354,
-      "grad_norm": 0.0023660468868911266,
-      "learning_rate": 0.0001999952499017802,
-      "loss": 46.0,
-      "step": 19276
-    },
-    {
-      "epoch": 3.104432545593623,
-      "grad_norm": 0.00660520326346159,
-      "learning_rate": 0.00019999524940866444,
-      "loss": 46.0,
-      "step": 19277
-    },
-    {
-      "epoch": 3.1045935826724103,
-      "grad_norm": 0.0017505967989563942,
-      "learning_rate": 0.00019999524891552313,
-      "loss": 46.0,
-      "step": 19278
-    },
-    {
-      "epoch": 3.1047546197511977,
-      "grad_norm": 0.011438223533332348,
-      "learning_rate": 0.0001999952484223562,
-      "loss": 46.0,
-      "step": 19279
-    },
-    {
-      "epoch": 3.104915656829985,
-      "grad_norm": 0.0010729064233601093,
-      "learning_rate": 0.00019999524792916368,
-      "loss": 46.0,
-      "step": 19280
-    },
-    {
-      "epoch": 3.1050766939087726,
-      "grad_norm": 0.0010524013778194785,
-      "learning_rate": 0.00019999524743594558,
-      "loss": 46.0,
-      "step": 19281
-    },
-    {
-      "epoch": 3.10523773098756,
-      "grad_norm": 0.0019382539903745055,
-      "learning_rate": 0.00019999524694270187,
-      "loss": 46.0,
-      "step": 19282
-    },
-    {
-      "epoch": 3.105398768066347,
-      "grad_norm": 0.002451792825013399,
-      "learning_rate": 0.00019999524644943257,
-      "loss": 46.0,
-      "step": 19283
-    },
-    {
-      "epoch": 3.1055598051451345,
-      "grad_norm": 0.008010496385395527,
-      "learning_rate": 0.0001999952459561377,
-      "loss": 46.0,
-      "step": 19284
-    },
-    {
-      "epoch": 3.105720842223922,
-      "grad_norm": 0.0073457336984574795,
-      "learning_rate": 0.0001999952454628172,
-      "loss": 46.0,
-      "step": 19285
-    },
-    {
-      "epoch": 3.1058818793027094,
-      "grad_norm": 0.0011635952396318316,
-      "learning_rate": 0.00019999524496947113,
-      "loss": 46.0,
-      "step": 19286
-    },
-    {
-      "epoch": 3.106042916381497,
-      "grad_norm": 0.002992767607793212,
-      "learning_rate": 0.00019999524447609945,
-      "loss": 46.0,
-      "step": 19287
-    },
-    {
-      "epoch": 3.1062039534602843,
-      "grad_norm": 0.0036717213224619627,
-      "learning_rate": 0.0001999952439827022,
-      "loss": 46.0,
-      "step": 19288
-    },
-    {
-      "epoch": 3.106364990539072,
-      "grad_norm": 0.0024175650905817747,
-      "learning_rate": 0.00019999524348927934,
-      "loss": 46.0,
-      "step": 19289
-    },
-    {
-      "epoch": 3.106526027617859,
-      "grad_norm": 0.003215793753042817,
-      "learning_rate": 0.00019999524299583085,
-      "loss": 46.0,
-      "step": 19290
-    },
-    {
-      "epoch": 3.1066870646966462,
-      "grad_norm": 0.004355255514383316,
-      "learning_rate": 0.0001999952425023568,
-      "loss": 46.0,
-      "step": 19291
-    },
-    {
-      "epoch": 3.1068481017754337,
-      "grad_norm": 0.0015961091266945004,
-      "learning_rate": 0.00019999524200885717,
-      "loss": 46.0,
-      "step": 19292
-    },
-    {
-      "epoch": 3.107009138854221,
-      "grad_norm": 0.0033468592446297407,
-      "learning_rate": 0.00019999524151533194,
-      "loss": 46.0,
-      "step": 19293
-    },
-    {
-      "epoch": 3.1071701759330086,
-      "grad_norm": 0.0016286567552015185,
-      "learning_rate": 0.00019999524102178113,
-      "loss": 46.0,
-      "step": 19294
-    },
-    {
-      "epoch": 3.107331213011796,
-      "grad_norm": 0.0033984172623604536,
-      "learning_rate": 0.0001999952405282047,
-      "loss": 46.0,
-      "step": 19295
-    },
-    {
-      "epoch": 3.1074922500905835,
-      "grad_norm": 0.005804160609841347,
-      "learning_rate": 0.0001999952400346027,
-      "loss": 46.0,
-      "step": 19296
-    },
-    {
-      "epoch": 3.107653287169371,
-      "grad_norm": 0.005598474759608507,
-      "learning_rate": 0.00019999523954097507,
-      "loss": 46.0,
-      "step": 19297
-    },
-    {
-      "epoch": 3.107814324248158,
-      "grad_norm": 0.0022115791216492653,
-      "learning_rate": 0.00019999523904732188,
-      "loss": 46.0,
-      "step": 19298
-    },
-    {
-      "epoch": 3.1079753613269454,
-      "grad_norm": 0.008368846029043198,
-      "learning_rate": 0.00019999523855364308,
-      "loss": 46.0,
-      "step": 19299
-    },
-    {
-      "epoch": 3.108136398405733,
-      "grad_norm": 0.0006773097557015717,
-      "learning_rate": 0.00019999523805993866,
-      "loss": 46.0,
-      "step": 19300
-    },
-    {
-      "epoch": 3.1082974354845203,
-      "grad_norm": 0.0026804094668477774,
-      "learning_rate": 0.0001999952375662087,
-      "loss": 46.0,
-      "step": 19301
-    },
-    {
-      "epoch": 3.1084584725633078,
-      "grad_norm": 0.0038352832198143005,
-      "learning_rate": 0.00019999523707245313,
-      "loss": 46.0,
-      "step": 19302
-    },
-    {
-      "epoch": 3.108619509642095,
-      "grad_norm": 0.00030896722455509007,
-      "learning_rate": 0.00019999523657867198,
-      "loss": 46.0,
-      "step": 19303
-    },
-    {
-      "epoch": 3.1087805467208827,
-      "grad_norm": 0.004510022699832916,
-      "learning_rate": 0.00019999523608486518,
-      "loss": 46.0,
-      "step": 19304
-    },
-    {
-      "epoch": 3.1089415837996697,
-      "grad_norm": 0.004488205537199974,
-      "learning_rate": 0.00019999523559103283,
-      "loss": 46.0,
-      "step": 19305
-    },
-    {
-      "epoch": 3.109102620878457,
-      "grad_norm": 0.0026863000821322203,
-      "learning_rate": 0.00019999523509717487,
-      "loss": 46.0,
-      "step": 19306
-    },
-    {
-      "epoch": 3.1092636579572446,
-      "grad_norm": 0.0026817957405000925,
-      "learning_rate": 0.00019999523460329131,
-      "loss": 46.0,
-      "step": 19307
-    },
-    {
-      "epoch": 3.109424695036032,
-      "grad_norm": 0.0047181290574371815,
-      "learning_rate": 0.0001999952341093822,
-      "loss": 46.0,
-      "step": 19308
-    },
-    {
-      "epoch": 3.1095857321148195,
-      "grad_norm": 0.008078353479504585,
-      "learning_rate": 0.00019999523361544745,
-      "loss": 46.0,
-      "step": 19309
-    },
-    {
-      "epoch": 3.109746769193607,
-      "grad_norm": 0.0031496433075517416,
-      "learning_rate": 0.00019999523312148713,
-      "loss": 46.0,
-      "step": 19310
-    },
-    {
-      "epoch": 3.1099078062723944,
-      "grad_norm": 0.0007472268189303577,
-      "learning_rate": 0.0001999952326275012,
-      "loss": 46.0,
-      "step": 19311
-    },
-    {
-      "epoch": 3.110068843351182,
-      "grad_norm": 0.0006258013308979571,
-      "learning_rate": 0.0001999952321334897,
-      "loss": 46.0,
-      "step": 19312
-    },
-    {
-      "epoch": 3.110229880429969,
-      "grad_norm": 0.0041410732083022594,
-      "learning_rate": 0.00019999523163945259,
-      "loss": 46.0,
-      "step": 19313
-    },
-    {
-      "epoch": 3.1103909175087563,
-      "grad_norm": 0.001808946137316525,
-      "learning_rate": 0.00019999523114538987,
-      "loss": 46.0,
-      "step": 19314
-    },
-    {
-      "epoch": 3.1105519545875437,
-      "grad_norm": 0.0007960887160152197,
-      "learning_rate": 0.0001999952306513016,
-      "loss": 46.0,
-      "step": 19315
-    },
-    {
-      "epoch": 3.110712991666331,
-      "grad_norm": 0.0029116705991327763,
-      "learning_rate": 0.0001999952301571877,
-      "loss": 46.0,
-      "step": 19316
-    },
-    {
-      "epoch": 3.1108740287451186,
-      "grad_norm": 0.0008015538332983851,
-      "learning_rate": 0.00019999522966304822,
-      "loss": 46.0,
-      "step": 19317
-    },
-    {
-      "epoch": 3.111035065823906,
-      "grad_norm": 0.00940077193081379,
-      "learning_rate": 0.00019999522916888313,
-      "loss": 46.0,
-      "step": 19318
-    },
-    {
-      "epoch": 3.1111961029026935,
-      "grad_norm": 0.002370478818193078,
-      "learning_rate": 0.0001999952286746925,
-      "loss": 46.0,
-      "step": 19319
-    },
-    {
-      "epoch": 3.1113571399814806,
-      "grad_norm": 0.0028829616494476795,
-      "learning_rate": 0.00019999522818047624,
-      "loss": 46.0,
-      "step": 19320
-    },
-    {
-      "epoch": 3.111518177060268,
-      "grad_norm": 0.001068557146936655,
-      "learning_rate": 0.00019999522768623435,
-      "loss": 46.0,
-      "step": 19321
-    },
-    {
-      "epoch": 3.1116792141390555,
-      "grad_norm": 0.002573475707322359,
-      "learning_rate": 0.0001999952271919669,
-      "loss": 46.0,
-      "step": 19322
-    },
-    {
-      "epoch": 3.111840251217843,
-      "grad_norm": 0.0021205635275691748,
-      "learning_rate": 0.00019999522669767386,
-      "loss": 46.0,
-      "step": 19323
-    },
-    {
-      "epoch": 3.1120012882966304,
-      "grad_norm": 0.00813312828540802,
-      "learning_rate": 0.0001999952262033552,
-      "loss": 46.0,
-      "step": 19324
-    },
-    {
-      "epoch": 3.112162325375418,
-      "grad_norm": 0.004118777811527252,
-      "learning_rate": 0.00019999522570901098,
-      "loss": 46.0,
-      "step": 19325
-    },
-    {
-      "epoch": 3.1123233624542053,
-      "grad_norm": 0.0015542840119451284,
-      "learning_rate": 0.0001999952252146412,
-      "loss": 46.0,
-      "step": 19326
-    },
-    {
-      "epoch": 3.1124843995329923,
-      "grad_norm": 0.007820246741175652,
-      "learning_rate": 0.00019999522472024576,
-      "loss": 46.0,
-      "step": 19327
-    },
-    {
-      "epoch": 3.1126454366117797,
-      "grad_norm": 0.002289773663505912,
-      "learning_rate": 0.00019999522422582474,
-      "loss": 46.0,
-      "step": 19328
-    },
-    {
-      "epoch": 3.112806473690567,
-      "grad_norm": 0.009453109465539455,
-      "learning_rate": 0.00019999522373137814,
-      "loss": 46.0,
-      "step": 19329
-    },
-    {
-      "epoch": 3.1129675107693546,
-      "grad_norm": 0.004741839598864317,
-      "learning_rate": 0.00019999522323690594,
-      "loss": 46.0,
-      "step": 19330
-    },
-    {
-      "epoch": 3.113128547848142,
-      "grad_norm": 0.0024593514390289783,
-      "learning_rate": 0.00019999522274240813,
-      "loss": 46.0,
-      "step": 19331
-    },
-    {
-      "epoch": 3.1132895849269295,
-      "grad_norm": 0.0018467070767655969,
-      "learning_rate": 0.00019999522224788474,
-      "loss": 46.0,
-      "step": 19332
-    },
-    {
-      "epoch": 3.113450622005717,
-      "grad_norm": 0.003571482840925455,
-      "learning_rate": 0.00019999522175333576,
-      "loss": 46.0,
-      "step": 19333
-    },
-    {
-      "epoch": 3.113611659084504,
-      "grad_norm": 0.0040740324184298515,
-      "learning_rate": 0.0001999952212587612,
-      "loss": 46.0,
-      "step": 19334
-    },
-    {
-      "epoch": 3.1137726961632914,
-      "grad_norm": 0.0014620781876146793,
-      "learning_rate": 0.000199995220764161,
-      "loss": 46.0,
-      "step": 19335
-    },
-    {
-      "epoch": 3.113933733242079,
-      "grad_norm": 0.0028638429939746857,
-      "learning_rate": 0.00019999522026953526,
-      "loss": 46.0,
-      "step": 19336
-    },
-    {
-      "epoch": 3.1140947703208663,
-      "grad_norm": 0.001697445404715836,
-      "learning_rate": 0.0001999952197748839,
-      "loss": 46.0,
-      "step": 19337
-    },
-    {
-      "epoch": 3.114255807399654,
-      "grad_norm": 0.004957149736583233,
-      "learning_rate": 0.00019999521928020696,
-      "loss": 46.0,
-      "step": 19338
-    },
-    {
-      "epoch": 3.1144168444784412,
-      "grad_norm": 0.0012377359671518207,
-      "learning_rate": 0.00019999521878550443,
-      "loss": 46.0,
-      "step": 19339
-    },
-    {
-      "epoch": 3.1145778815572287,
-      "grad_norm": 0.005835159216076136,
-      "learning_rate": 0.00019999521829077628,
-      "loss": 46.0,
-      "step": 19340
-    },
-    {
-      "epoch": 3.114738918636016,
-      "grad_norm": 0.0023562470450997353,
-      "learning_rate": 0.00019999521779602258,
-      "loss": 46.0,
-      "step": 19341
-    },
-    {
-      "epoch": 3.114899955714803,
-      "grad_norm": 0.004596192389726639,
-      "learning_rate": 0.00019999521730124326,
-      "loss": 46.0,
-      "step": 19342
-    },
-    {
-      "epoch": 3.1150609927935906,
-      "grad_norm": 0.0037462601903826,
-      "learning_rate": 0.00019999521680643832,
-      "loss": 46.0,
-      "step": 19343
-    },
-    {
-      "epoch": 3.115222029872378,
-      "grad_norm": 0.0022621031384915113,
-      "learning_rate": 0.0001999952163116078,
-      "loss": 46.0,
-      "step": 19344
-    },
-    {
-      "epoch": 3.1153830669511655,
-      "grad_norm": 0.0028835791163146496,
-      "learning_rate": 0.0001999952158167517,
-      "loss": 46.0,
-      "step": 19345
-    },
-    {
-      "epoch": 3.115544104029953,
-      "grad_norm": 0.0014843278331682086,
-      "learning_rate": 0.00019999521532187,
-      "loss": 46.0,
-      "step": 19346
-    },
-    {
-      "epoch": 3.1157051411087404,
-      "grad_norm": 0.004026864189654589,
-      "learning_rate": 0.00019999521482696274,
-      "loss": 46.0,
-      "step": 19347
-    },
-    {
-      "epoch": 3.115866178187528,
-      "grad_norm": 0.0040955375880002975,
-      "learning_rate": 0.00019999521433202984,
-      "loss": 46.0,
-      "step": 19348
-    },
-    {
-      "epoch": 3.116027215266315,
-      "grad_norm": 0.0012072190875187516,
-      "learning_rate": 0.00019999521383707135,
-      "loss": 46.0,
-      "step": 19349
-    },
-    {
-      "epoch": 3.1161882523451023,
-      "grad_norm": 0.007035098038613796,
-      "learning_rate": 0.00019999521334208725,
-      "loss": 46.0,
-      "step": 19350
-    },
-    {
-      "epoch": 3.11634928942389,
-      "grad_norm": 0.0031054222490638494,
-      "learning_rate": 0.0001999952128470776,
-      "loss": 46.0,
-      "step": 19351
-    },
-    {
-      "epoch": 3.1165103265026772,
-      "grad_norm": 0.0011674832785502076,
-      "learning_rate": 0.00019999521235204235,
-      "loss": 46.0,
-      "step": 19352
-    },
-    {
-      "epoch": 3.1166713635814647,
-      "grad_norm": 0.0017470386810600758,
-      "learning_rate": 0.0001999952118569815,
-      "loss": 46.0,
-      "step": 19353
-    },
-    {
-      "epoch": 3.116832400660252,
-      "grad_norm": 0.009978740476071835,
-      "learning_rate": 0.00019999521136189504,
-      "loss": 46.0,
-      "step": 19354
-    },
-    {
-      "epoch": 3.1169934377390396,
-      "grad_norm": 0.0017703694757074118,
-      "learning_rate": 0.000199995210866783,
-      "loss": 46.0,
-      "step": 19355
-    },
-    {
-      "epoch": 3.1171544748178266,
-      "grad_norm": 0.0020083021372556686,
-      "learning_rate": 0.00019999521037164538,
-      "loss": 46.0,
-      "step": 19356
-    },
-    {
-      "epoch": 3.117315511896614,
-      "grad_norm": 0.0012780020479112864,
-      "learning_rate": 0.00019999520987648215,
-      "loss": 46.0,
-      "step": 19357
-    },
-    {
-      "epoch": 3.1174765489754015,
-      "grad_norm": 0.002130125882104039,
-      "learning_rate": 0.00019999520938129332,
-      "loss": 46.0,
-      "step": 19358
-    },
-    {
-      "epoch": 3.117637586054189,
-      "grad_norm": 0.001066154451109469,
-      "learning_rate": 0.0001999952088860789,
-      "loss": 46.0,
-      "step": 19359
-    },
-    {
-      "epoch": 3.1177986231329764,
-      "grad_norm": 0.0016177851939573884,
-      "learning_rate": 0.0001999952083908389,
-      "loss": 46.0,
-      "step": 19360
-    },
-    {
-      "epoch": 3.117959660211764,
-      "grad_norm": 0.0016625664429739118,
-      "learning_rate": 0.0001999952078955733,
-      "loss": 46.0,
-      "step": 19361
-    },
-    {
-      "epoch": 3.1181206972905513,
-      "grad_norm": 0.0017413728637620807,
-      "learning_rate": 0.0001999952074002821,
-      "loss": 46.0,
-      "step": 19362
-    },
-    {
-      "epoch": 3.1182817343693388,
-      "grad_norm": 0.013530614785850048,
-      "learning_rate": 0.00019999520690496531,
-      "loss": 46.0,
-      "step": 19363
-    },
-    {
-      "epoch": 3.1184427714481258,
-      "grad_norm": 0.003448938950896263,
-      "learning_rate": 0.0001999952064096229,
-      "loss": 46.0,
-      "step": 19364
-    },
-    {
-      "epoch": 3.118603808526913,
-      "grad_norm": 0.004797689151018858,
-      "learning_rate": 0.00019999520591425495,
-      "loss": 46.0,
-      "step": 19365
-    },
-    {
-      "epoch": 3.1187648456057007,
-      "grad_norm": 0.003533191978931427,
-      "learning_rate": 0.00019999520541886138,
-      "loss": 46.0,
-      "step": 19366
-    },
-    {
-      "epoch": 3.118925882684488,
-      "grad_norm": 0.0009692597086541355,
-      "learning_rate": 0.00019999520492344221,
-      "loss": 46.0,
-      "step": 19367
-    },
-    {
-      "epoch": 3.1190869197632756,
-      "grad_norm": 0.006410129833966494,
-      "learning_rate": 0.00019999520442799746,
-      "loss": 46.0,
-      "step": 19368
-    },
-    {
-      "epoch": 3.119247956842063,
-      "grad_norm": 0.001012086751870811,
-      "learning_rate": 0.0001999952039325271,
-      "loss": 46.0,
-      "step": 19369
-    },
-    {
-      "epoch": 3.1194089939208505,
-      "grad_norm": 0.0004513339954428375,
-      "learning_rate": 0.00019999520343703115,
-      "loss": 46.0,
-      "step": 19370
-    },
-    {
-      "epoch": 3.1195700309996375,
-      "grad_norm": 0.00431870250031352,
-      "learning_rate": 0.0001999952029415096,
-      "loss": 46.0,
-      "step": 19371
-    },
-    {
-      "epoch": 3.119731068078425,
-      "grad_norm": 0.0027193366549909115,
-      "learning_rate": 0.0001999952024459625,
-      "loss": 46.0,
-      "step": 19372
-    },
-    {
-      "epoch": 3.1198921051572124,
-      "grad_norm": 0.0023377921897917986,
-      "learning_rate": 0.00019999520195038977,
-      "loss": 46.0,
-      "step": 19373
-    },
-    {
-      "epoch": 3.120053142236,
-      "grad_norm": 0.001031764899380505,
-      "learning_rate": 0.00019999520145479145,
-      "loss": 46.0,
-      "step": 19374
-    },
-    {
-      "epoch": 3.1202141793147873,
-      "grad_norm": 0.0024424795992672443,
-      "learning_rate": 0.00019999520095916753,
-      "loss": 46.0,
-      "step": 19375
-    },
-    {
-      "epoch": 3.1203752163935747,
-      "grad_norm": 0.0005334223387762904,
-      "learning_rate": 0.00019999520046351803,
-      "loss": 46.0,
-      "step": 19376
-    },
-    {
-      "epoch": 3.120536253472362,
-      "grad_norm": 0.005903015844523907,
-      "learning_rate": 0.00019999519996784291,
-      "loss": 46.0,
-      "step": 19377
-    },
-    {
-      "epoch": 3.120697290551149,
-      "grad_norm": 0.0012081346940249205,
-      "learning_rate": 0.0001999951994721422,
-      "loss": 46.0,
-      "step": 19378
-    },
-    {
-      "epoch": 3.1208583276299366,
-      "grad_norm": 0.0010159856174141169,
-      "learning_rate": 0.00019999519897641595,
-      "loss": 46.0,
-      "step": 19379
-    },
-    {
-      "epoch": 3.121019364708724,
-      "grad_norm": 0.0038528433069586754,
-      "learning_rate": 0.00019999519848066405,
-      "loss": 46.0,
-      "step": 19380
-    },
-    {
-      "epoch": 3.1211804017875115,
-      "grad_norm": 0.0011607845081016421,
-      "learning_rate": 0.0001999951979848866,
-      "loss": 46.0,
-      "step": 19381
-    },
-    {
-      "epoch": 3.121341438866299,
-      "grad_norm": 0.002512323437258601,
-      "learning_rate": 0.0001999951974890835,
-      "loss": 46.0,
-      "step": 19382
-    },
-    {
-      "epoch": 3.1215024759450865,
-      "grad_norm": 0.0061342050321400166,
-      "learning_rate": 0.00019999519699325484,
-      "loss": 46.0,
-      "step": 19383
-    },
-    {
-      "epoch": 3.121663513023874,
-      "grad_norm": 0.004224579315632582,
-      "learning_rate": 0.00019999519649740058,
-      "loss": 46.0,
-      "step": 19384
-    },
-    {
-      "epoch": 3.1218245501026614,
-      "grad_norm": 0.010057457722723484,
-      "learning_rate": 0.00019999519600152074,
-      "loss": 46.0,
-      "step": 19385
-    },
-    {
-      "epoch": 3.1219855871814484,
-      "grad_norm": 0.0026779703330248594,
-      "learning_rate": 0.00019999519550561532,
-      "loss": 46.0,
-      "step": 19386
-    },
-    {
-      "epoch": 3.122146624260236,
-      "grad_norm": 0.006408275105059147,
-      "learning_rate": 0.00019999519500968428,
-      "loss": 46.0,
-      "step": 19387
-    },
-    {
-      "epoch": 3.1223076613390233,
-      "grad_norm": 0.0007667880854569376,
-      "learning_rate": 0.00019999519451372765,
-      "loss": 46.0,
-      "step": 19388
-    },
-    {
-      "epoch": 3.1224686984178107,
-      "grad_norm": 0.008423435501754284,
-      "learning_rate": 0.0001999951940177454,
-      "loss": 46.0,
-      "step": 19389
-    },
-    {
-      "epoch": 3.122629735496598,
-      "grad_norm": 0.001704515889286995,
-      "learning_rate": 0.0001999951935217376,
-      "loss": 46.0,
-      "step": 19390
-    },
-    {
-      "epoch": 3.1227907725753856,
-      "grad_norm": 0.0006694623734802008,
-      "learning_rate": 0.00019999519302570418,
-      "loss": 46.0,
-      "step": 19391
-    },
-    {
-      "epoch": 3.122951809654173,
-      "grad_norm": 0.001778366044163704,
-      "learning_rate": 0.00019999519252964518,
-      "loss": 46.0,
-      "step": 19392
-    },
-    {
-      "epoch": 3.12311284673296,
-      "grad_norm": 0.005164226517081261,
-      "learning_rate": 0.0001999951920335606,
-      "loss": 46.0,
-      "step": 19393
-    },
-    {
-      "epoch": 3.1232738838117475,
-      "grad_norm": 0.00195777858607471,
-      "learning_rate": 0.00019999519153745038,
-      "loss": 46.0,
-      "step": 19394
-    },
-    {
-      "epoch": 3.123434920890535,
-      "grad_norm": 0.004751776345074177,
-      "learning_rate": 0.00019999519104131462,
-      "loss": 46.0,
-      "step": 19395
-    },
-    {
-      "epoch": 3.1235959579693224,
-      "grad_norm": 0.007243297062814236,
-      "learning_rate": 0.00019999519054515324,
-      "loss": 46.0,
-      "step": 19396
-    },
-    {
-      "epoch": 3.12375699504811,
-      "grad_norm": 0.009361706674098969,
-      "learning_rate": 0.00019999519004896624,
-      "loss": 46.0,
-      "step": 19397
-    },
-    {
-      "epoch": 3.1239180321268973,
-      "grad_norm": 0.004897703416645527,
-      "learning_rate": 0.0001999951895527537,
-      "loss": 46.0,
-      "step": 19398
-    },
-    {
-      "epoch": 3.124079069205685,
-      "grad_norm": 0.007208153605461121,
-      "learning_rate": 0.00019999518905651552,
-      "loss": 46.0,
-      "step": 19399
-    },
-    {
-      "epoch": 3.124240106284472,
-      "grad_norm": 0.004397696815431118,
-      "learning_rate": 0.00019999518856025176,
-      "loss": 46.0,
-      "step": 19400
-    },
-    {
-      "epoch": 3.1244011433632592,
-      "grad_norm": 0.0012879247078672051,
-      "learning_rate": 0.00019999518806396242,
-      "loss": 46.0,
-      "step": 19401
-    },
-    {
-      "epoch": 3.1245621804420467,
-      "grad_norm": 0.002920987317338586,
-      "learning_rate": 0.0001999951875676475,
-      "loss": 46.0,
-      "step": 19402
-    },
-    {
-      "epoch": 3.124723217520834,
-      "grad_norm": 0.003686473472043872,
-      "learning_rate": 0.00019999518707130694,
-      "loss": 46.0,
-      "step": 19403
-    },
-    {
-      "epoch": 3.1248842545996216,
-      "grad_norm": 0.0036910930648446083,
-      "learning_rate": 0.0001999951865749408,
-      "loss": 46.0,
-      "step": 19404
-    },
-    {
-      "epoch": 3.125045291678409,
-      "grad_norm": 0.0038740267045795918,
-      "learning_rate": 0.0001999951860785491,
-      "loss": 46.0,
-      "step": 19405
-    },
-    {
-      "epoch": 3.1252063287571965,
-      "grad_norm": 0.009565374813973904,
-      "learning_rate": 0.00019999518558213176,
-      "loss": 46.0,
-      "step": 19406
-    },
-    {
-      "epoch": 3.1253673658359835,
-      "grad_norm": 0.002369903726503253,
-      "learning_rate": 0.00019999518508568887,
-      "loss": 46.0,
-      "step": 19407
-    },
-    {
-      "epoch": 3.125528402914771,
-      "grad_norm": 0.0029599997214972973,
-      "learning_rate": 0.00019999518458922036,
-      "loss": 46.0,
-      "step": 19408
-    },
-    {
-      "epoch": 3.1256894399935584,
-      "grad_norm": 0.0015792009653523564,
-      "learning_rate": 0.00019999518409272626,
-      "loss": 46.0,
-      "step": 19409
-    },
-    {
-      "epoch": 3.125850477072346,
-      "grad_norm": 0.0016246461309492588,
-      "learning_rate": 0.00019999518359620658,
-      "loss": 46.0,
-      "step": 19410
-    },
-    {
-      "epoch": 3.1260115141511333,
-      "grad_norm": 0.0009963264456018806,
-      "learning_rate": 0.00019999518309966129,
-      "loss": 46.0,
-      "step": 19411
-    },
-    {
-      "epoch": 3.1261725512299208,
-      "grad_norm": 0.0022691329941153526,
-      "learning_rate": 0.0001999951826030904,
-      "loss": 46.0,
-      "step": 19412
-    },
-    {
-      "epoch": 3.1263335883087082,
-      "grad_norm": 0.001222363207489252,
-      "learning_rate": 0.00019999518210649396,
-      "loss": 46.0,
-      "step": 19413
-    },
-    {
-      "epoch": 3.1264946253874957,
-      "grad_norm": 0.0015487669734284282,
-      "learning_rate": 0.00019999518160987187,
-      "loss": 46.0,
-      "step": 19414
-    },
-    {
-      "epoch": 3.1266556624662827,
-      "grad_norm": 0.005512857344001532,
-      "learning_rate": 0.00019999518111322423,
-      "loss": 46.0,
-      "step": 19415
-    },
-    {
-      "epoch": 3.12681669954507,
-      "grad_norm": 0.009297710843384266,
-      "learning_rate": 0.00019999518061655097,
-      "loss": 46.0,
-      "step": 19416
-    },
-    {
-      "epoch": 3.1269777366238576,
-      "grad_norm": 0.01024394016712904,
-      "learning_rate": 0.00019999518011985212,
-      "loss": 46.0,
-      "step": 19417
-    },
-    {
-      "epoch": 3.127138773702645,
-      "grad_norm": 0.006012026220560074,
-      "learning_rate": 0.0001999951796231277,
-      "loss": 46.0,
-      "step": 19418
-    },
-    {
-      "epoch": 3.1272998107814325,
-      "grad_norm": 0.001367154880426824,
-      "learning_rate": 0.00019999517912637764,
-      "loss": 46.0,
-      "step": 19419
-    },
-    {
-      "epoch": 3.12746084786022,
-      "grad_norm": 0.036792658269405365,
-      "learning_rate": 0.000199995178629602,
-      "loss": 46.0,
-      "step": 19420
-    },
-    {
-      "epoch": 3.1276218849390074,
-      "grad_norm": 0.007829583249986172,
-      "learning_rate": 0.00019999517813280079,
-      "loss": 46.0,
-      "step": 19421
-    },
-    {
-      "epoch": 3.1277829220177944,
-      "grad_norm": 0.004285490605980158,
-      "learning_rate": 0.00019999517763597398,
-      "loss": 46.0,
-      "step": 19422
-    },
-    {
-      "epoch": 3.127943959096582,
-      "grad_norm": 0.001631292630918324,
-      "learning_rate": 0.00019999517713912158,
-      "loss": 46.0,
-      "step": 19423
-    },
-    {
-      "epoch": 3.1281049961753693,
-      "grad_norm": 0.008447431959211826,
-      "learning_rate": 0.00019999517664224357,
-      "loss": 46.0,
-      "step": 19424
-    },
-    {
-      "epoch": 3.1282660332541568,
-      "grad_norm": 0.00588274234905839,
-      "learning_rate": 0.00019999517614533997,
-      "loss": 46.0,
-      "step": 19425
-    },
-    {
-      "epoch": 3.128427070332944,
-      "grad_norm": 0.002601267071440816,
-      "learning_rate": 0.00019999517564841076,
-      "loss": 46.0,
-      "step": 19426
-    },
-    {
-      "epoch": 3.1285881074117317,
-      "grad_norm": 0.007402703166007996,
-      "learning_rate": 0.000199995175151456,
-      "loss": 46.0,
-      "step": 19427
-    },
-    {
-      "epoch": 3.128749144490519,
-      "grad_norm": 0.004526689648628235,
-      "learning_rate": 0.00019999517465447563,
-      "loss": 46.0,
-      "step": 19428
-    },
-    {
-      "epoch": 3.1289101815693066,
-      "grad_norm": 0.0032449415884912014,
-      "learning_rate": 0.00019999517415746965,
-      "loss": 46.0,
-      "step": 19429
-    },
-    {
-      "epoch": 3.1290712186480936,
-      "grad_norm": 0.0029983920976519585,
-      "learning_rate": 0.0001999951736604381,
-      "loss": 46.0,
-      "step": 19430
-    },
-    {
-      "epoch": 3.129232255726881,
-      "grad_norm": 0.00463849538937211,
-      "learning_rate": 0.00019999517316338094,
-      "loss": 46.0,
-      "step": 19431
-    },
-    {
-      "epoch": 3.1293932928056685,
-      "grad_norm": 0.002553614554926753,
-      "learning_rate": 0.00019999517266629818,
-      "loss": 46.0,
-      "step": 19432
-    },
-    {
-      "epoch": 3.129554329884456,
-      "grad_norm": 0.0014503775164484978,
-      "learning_rate": 0.00019999517216918983,
-      "loss": 46.0,
-      "step": 19433
-    },
-    {
-      "epoch": 3.1297153669632434,
-      "grad_norm": 0.003753715194761753,
-      "learning_rate": 0.0001999951716720559,
-      "loss": 46.0,
-      "step": 19434
-    },
-    {
-      "epoch": 3.129876404042031,
-      "grad_norm": 0.0017359053017571568,
-      "learning_rate": 0.00019999517117489637,
-      "loss": 46.0,
-      "step": 19435
-    },
-    {
-      "epoch": 3.1300374411208183,
-      "grad_norm": 0.00038306316128000617,
-      "learning_rate": 0.00019999517067771123,
-      "loss": 46.0,
-      "step": 19436
-    },
-    {
-      "epoch": 3.1301984781996053,
-      "grad_norm": 0.0006059609586372972,
-      "learning_rate": 0.0001999951701805005,
-      "loss": 46.0,
-      "step": 19437
-    },
-    {
-      "epoch": 3.1303595152783927,
-      "grad_norm": 0.001608496531844139,
-      "learning_rate": 0.00019999516968326422,
-      "loss": 46.0,
-      "step": 19438
-    },
-    {
-      "epoch": 3.13052055235718,
-      "grad_norm": 0.001452327356673777,
-      "learning_rate": 0.0001999951691860023,
-      "loss": 46.0,
-      "step": 19439
-    },
-    {
-      "epoch": 3.1306815894359676,
-      "grad_norm": 0.0008440740057267249,
-      "learning_rate": 0.00019999516868871478,
-      "loss": 46.0,
-      "step": 19440
-    },
-    {
-      "epoch": 3.130842626514755,
-      "grad_norm": 0.000956208270508796,
-      "learning_rate": 0.0001999951681914017,
-      "loss": 46.0,
-      "step": 19441
-    },
-    {
-      "epoch": 3.1310036635935425,
-      "grad_norm": 0.004498219583183527,
-      "learning_rate": 0.00019999516769406302,
-      "loss": 46.0,
-      "step": 19442
-    },
-    {
-      "epoch": 3.13116470067233,
-      "grad_norm": 0.0007291391375474632,
-      "learning_rate": 0.00019999516719669872,
-      "loss": 46.0,
-      "step": 19443
-    },
-    {
-      "epoch": 3.131325737751117,
-      "grad_norm": 0.008510365150868893,
-      "learning_rate": 0.00019999516669930886,
-      "loss": 46.0,
-      "step": 19444
-    },
-    {
-      "epoch": 3.1314867748299045,
-      "grad_norm": 0.004410738591104746,
-      "learning_rate": 0.0001999951662018934,
-      "loss": 46.0,
-      "step": 19445
-    },
-    {
-      "epoch": 3.131647811908692,
-      "grad_norm": 0.0032752861734479666,
-      "learning_rate": 0.00019999516570445234,
-      "loss": 46.0,
-      "step": 19446
-    },
-    {
-      "epoch": 3.1318088489874794,
-      "grad_norm": 0.004121949430555105,
-      "learning_rate": 0.00019999516520698566,
-      "loss": 46.0,
-      "step": 19447
-    },
-    {
-      "epoch": 3.131969886066267,
-      "grad_norm": 0.0009498496074229479,
-      "learning_rate": 0.00019999516470949345,
-      "loss": 46.0,
-      "step": 19448
-    },
-    {
-      "epoch": 3.1321309231450543,
-      "grad_norm": 0.007713876198977232,
-      "learning_rate": 0.0001999951642119756,
-      "loss": 46.0,
-      "step": 19449
-    },
-    {
-      "epoch": 3.1322919602238417,
-      "grad_norm": 0.004907692316919565,
-      "learning_rate": 0.00019999516371443216,
-      "loss": 46.0,
-      "step": 19450
-    },
-    {
-      "epoch": 3.1324529973026287,
-      "grad_norm": 0.0027944075409322977,
-      "learning_rate": 0.0001999951632168631,
-      "loss": 46.0,
-      "step": 19451
-    },
-    {
-      "epoch": 3.132614034381416,
-      "grad_norm": 0.0088083790615201,
-      "learning_rate": 0.0001999951627192685,
-      "loss": 46.0,
-      "step": 19452
-    },
-    {
-      "epoch": 3.1327750714602036,
-      "grad_norm": 0.0007252052309922874,
-      "learning_rate": 0.00019999516222164826,
-      "loss": 46.0,
-      "step": 19453
-    },
-    {
-      "epoch": 3.132936108538991,
-      "grad_norm": 0.008067525923252106,
-      "learning_rate": 0.00019999516172400247,
-      "loss": 46.0,
-      "step": 19454
-    },
-    {
-      "epoch": 3.1330971456177785,
-      "grad_norm": 0.001005639904178679,
-      "learning_rate": 0.00019999516122633107,
-      "loss": 46.0,
-      "step": 19455
-    },
-    {
-      "epoch": 3.133258182696566,
-      "grad_norm": 0.006840166635811329,
-      "learning_rate": 0.00019999516072863405,
-      "loss": 46.0,
-      "step": 19456
-    },
-    {
-      "epoch": 3.1334192197753534,
-      "grad_norm": 0.013018138706684113,
-      "learning_rate": 0.0001999951602309115,
-      "loss": 46.0,
-      "step": 19457
-    },
-    {
-      "epoch": 3.133580256854141,
-      "grad_norm": 0.0010031365090981126,
-      "learning_rate": 0.00019999515973316329,
-      "loss": 46.0,
-      "step": 19458
-    },
-    {
-      "epoch": 3.133741293932928,
-      "grad_norm": 0.003623103955760598,
-      "learning_rate": 0.0001999951592353895,
-      "loss": 46.0,
-      "step": 19459
-    },
-    {
-      "epoch": 3.1339023310117153,
-      "grad_norm": 0.0018743551336228848,
-      "learning_rate": 0.00019999515873759014,
-      "loss": 46.0,
-      "step": 19460
-    },
-    {
-      "epoch": 3.134063368090503,
-      "grad_norm": 0.008622311055660248,
-      "learning_rate": 0.0001999951582397652,
-      "loss": 46.0,
-      "step": 19461
-    },
-    {
-      "epoch": 3.1342244051692902,
-      "grad_norm": 0.004155608359724283,
-      "learning_rate": 0.00019999515774191462,
-      "loss": 46.0,
-      "step": 19462
-    },
-    {
-      "epoch": 3.1343854422480777,
-      "grad_norm": 0.003546161809936166,
-      "learning_rate": 0.00019999515724403847,
-      "loss": 46.0,
-      "step": 19463
-    },
-    {
-      "epoch": 3.134546479326865,
-      "grad_norm": 0.009766785427927971,
-      "learning_rate": 0.00019999515674613673,
-      "loss": 46.0,
-      "step": 19464
-    },
-    {
-      "epoch": 3.1347075164056526,
-      "grad_norm": 0.0019418501760810614,
-      "learning_rate": 0.00019999515624820937,
-      "loss": 46.0,
-      "step": 19465
-    },
-    {
-      "epoch": 3.1348685534844396,
-      "grad_norm": 0.001314381486736238,
-      "learning_rate": 0.00019999515575025645,
-      "loss": 46.0,
-      "step": 19466
-    },
-    {
-      "epoch": 3.135029590563227,
-      "grad_norm": 0.004811578895896673,
-      "learning_rate": 0.0001999951552522779,
-      "loss": 46.0,
-      "step": 19467
-    },
-    {
-      "epoch": 3.1351906276420145,
-      "grad_norm": 0.009981920011341572,
-      "learning_rate": 0.0001999951547542738,
-      "loss": 46.0,
-      "step": 19468
-    },
-    {
-      "epoch": 3.135351664720802,
-      "grad_norm": 0.00500292656943202,
-      "learning_rate": 0.00019999515425624408,
-      "loss": 46.0,
-      "step": 19469
-    },
-    {
-      "epoch": 3.1355127017995894,
-      "grad_norm": 0.0019361722515895963,
-      "learning_rate": 0.00019999515375818878,
-      "loss": 46.0,
-      "step": 19470
-    },
-    {
-      "epoch": 3.135673738878377,
-      "grad_norm": 0.0033515747636556625,
-      "learning_rate": 0.00019999515326010788,
-      "loss": 46.0,
-      "step": 19471
-    },
-    {
-      "epoch": 3.1358347759571643,
-      "grad_norm": 0.002729524625465274,
-      "learning_rate": 0.00019999515276200138,
-      "loss": 46.0,
-      "step": 19472
-    },
-    {
-      "epoch": 3.1359958130359518,
-      "grad_norm": 0.0046296501532197,
-      "learning_rate": 0.00019999515226386928,
-      "loss": 46.0,
-      "step": 19473
-    },
-    {
-      "epoch": 3.1361568501147388,
-      "grad_norm": 0.0030332240276038647,
-      "learning_rate": 0.0001999951517657116,
-      "loss": 46.0,
-      "step": 19474
-    },
-    {
-      "epoch": 3.136317887193526,
-      "grad_norm": 0.00582721084356308,
-      "learning_rate": 0.00019999515126752833,
-      "loss": 46.0,
-      "step": 19475
-    },
-    {
-      "epoch": 3.1364789242723137,
-      "grad_norm": 0.004632348660379648,
-      "learning_rate": 0.00019999515076931946,
-      "loss": 46.0,
-      "step": 19476
-    },
-    {
-      "epoch": 3.136639961351101,
-      "grad_norm": 0.0017418634379282594,
-      "learning_rate": 0.00019999515027108498,
-      "loss": 46.0,
-      "step": 19477
-    },
-    {
-      "epoch": 3.1368009984298886,
-      "grad_norm": 0.0028542291838675737,
-      "learning_rate": 0.00019999514977282493,
-      "loss": 46.0,
-      "step": 19478
-    },
-    {
-      "epoch": 3.136962035508676,
-      "grad_norm": 0.006918790750205517,
-      "learning_rate": 0.00019999514927453927,
-      "loss": 46.0,
-      "step": 19479
-    },
-    {
-      "epoch": 3.137123072587463,
-      "grad_norm": 0.004386961460113525,
-      "learning_rate": 0.00019999514877622803,
-      "loss": 46.0,
-      "step": 19480
-    },
-    {
-      "epoch": 3.1372841096662505,
-      "grad_norm": 0.01744384691119194,
-      "learning_rate": 0.0001999951482778912,
-      "loss": 46.0,
-      "step": 19481
-    },
-    {
-      "epoch": 3.137445146745038,
-      "grad_norm": 0.007464381866157055,
-      "learning_rate": 0.00019999514777952878,
-      "loss": 46.0,
-      "step": 19482
-    },
-    {
-      "epoch": 3.1376061838238254,
-      "grad_norm": 0.005523985717445612,
-      "learning_rate": 0.00019999514728114072,
-      "loss": 46.0,
-      "step": 19483
-    },
-    {
-      "epoch": 3.137767220902613,
-      "grad_norm": 0.0058706956915557384,
-      "learning_rate": 0.0001999951467827271,
-      "loss": 46.0,
-      "step": 19484
-    },
-    {
-      "epoch": 3.1379282579814003,
-      "grad_norm": 0.0017817047191783786,
-      "learning_rate": 0.0001999951462842879,
-      "loss": 46.0,
-      "step": 19485
-    },
-    {
-      "epoch": 3.1380892950601877,
-      "grad_norm": 0.003794813994318247,
-      "learning_rate": 0.00019999514578582307,
-      "loss": 46.0,
-      "step": 19486
-    },
-    {
-      "epoch": 3.138250332138975,
-      "grad_norm": 0.00484858313575387,
-      "learning_rate": 0.0001999951452873327,
-      "loss": 46.0,
-      "step": 19487
-    },
-    {
-      "epoch": 3.138411369217762,
-      "grad_norm": 0.0012541485484689474,
-      "learning_rate": 0.0001999951447888167,
-      "loss": 46.0,
-      "step": 19488
-    },
-    {
-      "epoch": 3.1385724062965497,
-      "grad_norm": 0.003293959889560938,
-      "learning_rate": 0.00019999514429027513,
-      "loss": 46.0,
-      "step": 19489
-    },
-    {
-      "epoch": 3.138733443375337,
-      "grad_norm": 0.0022064840886741877,
-      "learning_rate": 0.00019999514379170794,
-      "loss": 46.0,
-      "step": 19490
-    },
-    {
-      "epoch": 3.1388944804541246,
-      "grad_norm": 0.0025604343973100185,
-      "learning_rate": 0.00019999514329311518,
-      "loss": 46.0,
-      "step": 19491
-    },
-    {
-      "epoch": 3.139055517532912,
-      "grad_norm": 0.0015439606504514813,
-      "learning_rate": 0.00019999514279449678,
-      "loss": 46.0,
-      "step": 19492
-    },
-    {
-      "epoch": 3.1392165546116995,
-      "grad_norm": 0.004343516193330288,
-      "learning_rate": 0.00019999514229585282,
-      "loss": 46.0,
-      "step": 19493
-    },
-    {
-      "epoch": 3.139377591690487,
-      "grad_norm": 0.0020656301639974117,
-      "learning_rate": 0.00019999514179718327,
-      "loss": 46.0,
-      "step": 19494
-    },
-    {
-      "epoch": 3.139538628769274,
-      "grad_norm": 0.005326667334884405,
-      "learning_rate": 0.00019999514129848814,
-      "loss": 46.0,
-      "step": 19495
-    },
-    {
-      "epoch": 3.1396996658480614,
-      "grad_norm": 0.003601297503337264,
-      "learning_rate": 0.00019999514079976737,
-      "loss": 46.0,
-      "step": 19496
-    },
-    {
-      "epoch": 3.139860702926849,
-      "grad_norm": 0.0009913978865370154,
-      "learning_rate": 0.00019999514030102106,
-      "loss": 46.0,
-      "step": 19497
-    },
-    {
-      "epoch": 3.1400217400056363,
-      "grad_norm": 0.0022651536855846643,
-      "learning_rate": 0.0001999951398022491,
-      "loss": 46.0,
-      "step": 19498
-    },
-    {
-      "epoch": 3.1401827770844237,
-      "grad_norm": 0.0034134145826101303,
-      "learning_rate": 0.0001999951393034516,
-      "loss": 46.0,
-      "step": 19499
-    },
-    {
-      "epoch": 3.140343814163211,
-      "grad_norm": 0.0014328915858641267,
-      "learning_rate": 0.00019999513880462847,
-      "loss": 46.0,
-      "step": 19500
-    },
-    {
-      "epoch": 3.1405048512419986,
-      "grad_norm": 0.008060412481427193,
-      "learning_rate": 0.00019999513830577976,
-      "loss": 46.0,
-      "step": 19501
-    },
-    {
-      "epoch": 3.140665888320786,
-      "grad_norm": 0.003635668894276023,
-      "learning_rate": 0.00019999513780690547,
-      "loss": 46.0,
-      "step": 19502
-    },
-    {
-      "epoch": 3.140826925399573,
-      "grad_norm": 0.0037198769859969616,
-      "learning_rate": 0.00019999513730800555,
-      "loss": 46.0,
-      "step": 19503
-    },
-    {
-      "epoch": 3.1409879624783605,
-      "grad_norm": 0.00278861285187304,
-      "learning_rate": 0.00019999513680908005,
-      "loss": 46.0,
-      "step": 19504
-    },
-    {
-      "epoch": 3.141148999557148,
-      "grad_norm": 0.004973365925252438,
-      "learning_rate": 0.000199995136310129,
-      "loss": 46.0,
-      "step": 19505
-    },
-    {
-      "epoch": 3.1413100366359354,
-      "grad_norm": 0.006277016829699278,
-      "learning_rate": 0.0001999951358111523,
-      "loss": 46.0,
-      "step": 19506
-    },
-    {
-      "epoch": 3.141471073714723,
-      "grad_norm": 0.004592635203152895,
-      "learning_rate": 0.00019999513531215003,
-      "loss": 46.0,
-      "step": 19507
-    },
-    {
-      "epoch": 3.1416321107935103,
-      "grad_norm": 0.0012343861162662506,
-      "learning_rate": 0.00019999513481312216,
-      "loss": 46.0,
-      "step": 19508
-    },
-    {
-      "epoch": 3.141793147872298,
-      "grad_norm": 0.0009086685022339225,
-      "learning_rate": 0.0001999951343140687,
-      "loss": 46.0,
-      "step": 19509
-    },
-    {
-      "epoch": 3.141954184951085,
-      "grad_norm": 0.0006483003380708396,
-      "learning_rate": 0.00019999513381498965,
-      "loss": 46.0,
-      "step": 19510
-    },
-    {
-      "epoch": 3.1421152220298723,
-      "grad_norm": 0.00268872850574553,
-      "learning_rate": 0.000199995133315885,
-      "loss": 46.0,
-      "step": 19511
-    },
-    {
-      "epoch": 3.1422762591086597,
-      "grad_norm": 0.002022774191573262,
-      "learning_rate": 0.00019999513281675476,
-      "loss": 46.0,
-      "step": 19512
-    },
-    {
-      "epoch": 3.142437296187447,
-      "grad_norm": 0.007730285171419382,
-      "learning_rate": 0.00019999513231759892,
-      "loss": 46.0,
-      "step": 19513
-    },
-    {
-      "epoch": 3.1425983332662346,
-      "grad_norm": 0.006866161711513996,
-      "learning_rate": 0.0001999951318184175,
-      "loss": 46.0,
-      "step": 19514
-    },
-    {
-      "epoch": 3.142759370345022,
-      "grad_norm": 0.003307870589196682,
-      "learning_rate": 0.00019999513131921045,
-      "loss": 46.0,
-      "step": 19515
-    },
-    {
-      "epoch": 3.1429204074238095,
-      "grad_norm": 0.003311126260086894,
-      "learning_rate": 0.00019999513081997786,
-      "loss": 46.0,
-      "step": 19516
-    },
-    {
-      "epoch": 3.1430814445025965,
-      "grad_norm": 0.000960042467340827,
-      "learning_rate": 0.00019999513032071967,
-      "loss": 46.0,
-      "step": 19517
-    },
-    {
-      "epoch": 3.143242481581384,
-      "grad_norm": 0.001321278978139162,
-      "learning_rate": 0.00019999512982143584,
-      "loss": 46.0,
-      "step": 19518
-    },
-    {
-      "epoch": 3.1434035186601714,
-      "grad_norm": 0.001282593933865428,
-      "learning_rate": 0.00019999512932212645,
-      "loss": 46.0,
-      "step": 19519
-    },
-    {
-      "epoch": 3.143564555738959,
-      "grad_norm": 0.0018429771298542619,
-      "learning_rate": 0.00019999512882279145,
-      "loss": 46.0,
-      "step": 19520
-    },
-    {
-      "epoch": 3.1437255928177463,
-      "grad_norm": 0.00237071979790926,
-      "learning_rate": 0.0001999951283234309,
-      "loss": 46.0,
-      "step": 19521
-    },
-    {
-      "epoch": 3.143886629896534,
-      "grad_norm": 0.0008535197703167796,
-      "learning_rate": 0.0001999951278240447,
-      "loss": 46.0,
-      "step": 19522
-    },
-    {
-      "epoch": 3.1440476669753212,
-      "grad_norm": 0.0074363392777740955,
-      "learning_rate": 0.00019999512732463292,
-      "loss": 46.0,
-      "step": 19523
-    },
-    {
-      "epoch": 3.1442087040541082,
-      "grad_norm": 0.00275426940061152,
-      "learning_rate": 0.00019999512682519557,
-      "loss": 46.0,
-      "step": 19524
-    },
-    {
-      "epoch": 3.1443697411328957,
-      "grad_norm": 0.0010301631409674883,
-      "learning_rate": 0.0001999951263257326,
-      "loss": 46.0,
-      "step": 19525
-    },
-    {
-      "epoch": 3.144530778211683,
-      "grad_norm": 0.008522147312760353,
-      "learning_rate": 0.00019999512582624405,
-      "loss": 46.0,
-      "step": 19526
-    },
-    {
-      "epoch": 3.1446918152904706,
-      "grad_norm": 0.0034903259947896004,
-      "learning_rate": 0.0001999951253267299,
-      "loss": 46.0,
-      "step": 19527
-    },
-    {
-      "epoch": 3.144852852369258,
-      "grad_norm": 0.007606164086610079,
-      "learning_rate": 0.00019999512482719015,
-      "loss": 46.0,
-      "step": 19528
-    },
-    {
-      "epoch": 3.1450138894480455,
-      "grad_norm": 0.0007449169643223286,
-      "learning_rate": 0.00019999512432762484,
-      "loss": 46.0,
-      "step": 19529
-    },
-    {
-      "epoch": 3.145174926526833,
-      "grad_norm": 0.003868551691994071,
-      "learning_rate": 0.00019999512382803388,
-      "loss": 46.0,
-      "step": 19530
-    },
-    {
-      "epoch": 3.1453359636056204,
-      "grad_norm": 0.0020239658188074827,
-      "learning_rate": 0.0001999951233284174,
-      "loss": 46.0,
-      "step": 19531
-    },
-    {
-      "epoch": 3.1454970006844074,
-      "grad_norm": 0.0023372266441583633,
-      "learning_rate": 0.00019999512282877527,
-      "loss": 46.0,
-      "step": 19532
-    },
-    {
-      "epoch": 3.145658037763195,
-      "grad_norm": 0.0050746239721775055,
-      "learning_rate": 0.00019999512232910758,
-      "loss": 46.0,
-      "step": 19533
-    },
-    {
-      "epoch": 3.1458190748419823,
-      "grad_norm": 0.013485323637723923,
-      "learning_rate": 0.00019999512182941427,
-      "loss": 46.0,
-      "step": 19534
-    },
-    {
-      "epoch": 3.1459801119207698,
-      "grad_norm": 0.004309992305934429,
-      "learning_rate": 0.00019999512132969538,
-      "loss": 46.0,
-      "step": 19535
-    },
-    {
-      "epoch": 3.146141148999557,
-      "grad_norm": 0.013468729332089424,
-      "learning_rate": 0.00019999512082995087,
-      "loss": 46.0,
-      "step": 19536
-    },
-    {
-      "epoch": 3.1463021860783447,
-      "grad_norm": 0.0070412238128483295,
-      "learning_rate": 0.0001999951203301808,
-      "loss": 46.0,
-      "step": 19537
-    },
-    {
-      "epoch": 3.146463223157132,
-      "grad_norm": 0.003107303287833929,
-      "learning_rate": 0.00019999511983038513,
-      "loss": 46.0,
-      "step": 19538
-    },
-    {
-      "epoch": 3.146624260235919,
-      "grad_norm": 0.0010528797283768654,
-      "learning_rate": 0.00019999511933056386,
-      "loss": 46.0,
-      "step": 19539
-    },
-    {
-      "epoch": 3.1467852973147066,
-      "grad_norm": 0.0016254577785730362,
-      "learning_rate": 0.000199995118830717,
-      "loss": 46.0,
-      "step": 19540
-    },
-    {
-      "epoch": 3.146946334393494,
-      "grad_norm": 0.008398851379752159,
-      "learning_rate": 0.0001999951183308445,
-      "loss": 46.0,
-      "step": 19541
-    },
-    {
-      "epoch": 3.1471073714722815,
-      "grad_norm": 0.0010431804694235325,
-      "learning_rate": 0.00019999511783094648,
-      "loss": 46.0,
-      "step": 19542
-    },
-    {
-      "epoch": 3.147268408551069,
-      "grad_norm": 0.002881931606680155,
-      "learning_rate": 0.00019999511733102284,
-      "loss": 46.0,
-      "step": 19543
-    },
-    {
-      "epoch": 3.1474294456298564,
-      "grad_norm": 0.001314807333983481,
-      "learning_rate": 0.00019999511683107358,
-      "loss": 46.0,
-      "step": 19544
-    },
-    {
-      "epoch": 3.147590482708644,
-      "grad_norm": 0.0066570742055773735,
-      "learning_rate": 0.00019999511633109876,
-      "loss": 46.0,
-      "step": 19545
-    },
-    {
-      "epoch": 3.1477515197874313,
-      "grad_norm": 0.001074617262929678,
-      "learning_rate": 0.00019999511583109836,
-      "loss": 46.0,
-      "step": 19546
-    },
-    {
-      "epoch": 3.1479125568662183,
-      "grad_norm": 0.0010014285799115896,
-      "learning_rate": 0.00019999511533107234,
-      "loss": 46.0,
-      "step": 19547
-    },
-    {
-      "epoch": 3.1480735939450057,
-      "grad_norm": 0.005359780043363571,
-      "learning_rate": 0.0001999951148310207,
-      "loss": 46.0,
-      "step": 19548
-    },
-    {
-      "epoch": 3.148234631023793,
-      "grad_norm": 0.004768101964145899,
-      "learning_rate": 0.00019999511433094354,
-      "loss": 46.0,
-      "step": 19549
-    },
-    {
-      "epoch": 3.1483956681025806,
-      "grad_norm": 0.0020312040578573942,
-      "learning_rate": 0.00019999511383084073,
-      "loss": 46.0,
-      "step": 19550
-    },
-    {
-      "epoch": 3.148556705181368,
-      "grad_norm": 0.001919933594763279,
-      "learning_rate": 0.0001999951133307123,
-      "loss": 46.0,
-      "step": 19551
-    },
-    {
-      "epoch": 3.1487177422601555,
-      "grad_norm": 0.004075990524142981,
-      "learning_rate": 0.00019999511283055833,
-      "loss": 46.0,
-      "step": 19552
-    },
-    {
-      "epoch": 3.1488787793389426,
-      "grad_norm": 0.007960813120007515,
-      "learning_rate": 0.00019999511233037876,
-      "loss": 46.0,
-      "step": 19553
-    },
-    {
-      "epoch": 3.14903981641773,
-      "grad_norm": 0.0069361524656414986,
-      "learning_rate": 0.00019999511183017358,
-      "loss": 46.0,
-      "step": 19554
-    },
-    {
-      "epoch": 3.1492008534965175,
-      "grad_norm": 0.003147019073367119,
-      "learning_rate": 0.0001999951113299428,
-      "loss": 46.0,
-      "step": 19555
-    },
-    {
-      "epoch": 3.149361890575305,
-      "grad_norm": 0.0016937522450461984,
-      "learning_rate": 0.00019999511082968648,
-      "loss": 46.0,
-      "step": 19556
-    },
-    {
-      "epoch": 3.1495229276540924,
-      "grad_norm": 0.0017041638493537903,
-      "learning_rate": 0.00019999511032940453,
-      "loss": 46.0,
-      "step": 19557
-    },
-    {
-      "epoch": 3.14968396473288,
-      "grad_norm": 0.007754228077828884,
-      "learning_rate": 0.00019999510982909695,
-      "loss": 46.0,
-      "step": 19558
-    },
-    {
-      "epoch": 3.1498450018116673,
-      "grad_norm": 0.005011291243135929,
-      "learning_rate": 0.00019999510932876383,
-      "loss": 46.0,
-      "step": 19559
-    },
-    {
-      "epoch": 3.1500060388904547,
-      "grad_norm": 0.0050039710476994514,
-      "learning_rate": 0.0001999951088284051,
-      "loss": 46.0,
-      "step": 19560
-    },
-    {
-      "epoch": 3.1501670759692417,
-      "grad_norm": 0.004077255725860596,
-      "learning_rate": 0.00019999510832802075,
-      "loss": 46.0,
-      "step": 19561
-    },
-    {
-      "epoch": 3.150328113048029,
-      "grad_norm": 0.0010054357117041945,
-      "learning_rate": 0.00019999510782761084,
-      "loss": 46.0,
-      "step": 19562
-    },
-    {
-      "epoch": 3.1504891501268166,
-      "grad_norm": 0.0013901363126933575,
-      "learning_rate": 0.00019999510732717532,
-      "loss": 46.0,
-      "step": 19563
-    },
-    {
-      "epoch": 3.150650187205604,
-      "grad_norm": 0.003362879855558276,
-      "learning_rate": 0.0001999951068267142,
-      "loss": 46.0,
-      "step": 19564
-    },
-    {
-      "epoch": 3.1508112242843915,
-      "grad_norm": 0.005998195614665747,
-      "learning_rate": 0.0001999951063262275,
-      "loss": 46.0,
-      "step": 19565
-    },
-    {
-      "epoch": 3.150972261363179,
-      "grad_norm": 0.0047590890899300575,
-      "learning_rate": 0.0001999951058257152,
-      "loss": 46.0,
-      "step": 19566
-    },
-    {
-      "epoch": 3.1511332984419664,
-      "grad_norm": 0.0019848798401653767,
-      "learning_rate": 0.0001999951053251773,
-      "loss": 46.0,
-      "step": 19567
-    },
-    {
-      "epoch": 3.1512943355207534,
-      "grad_norm": 0.0020959326066076756,
-      "learning_rate": 0.00019999510482461385,
-      "loss": 46.0,
-      "step": 19568
-    },
-    {
-      "epoch": 3.151455372599541,
-      "grad_norm": 0.010506482794880867,
-      "learning_rate": 0.00019999510432402477,
-      "loss": 46.0,
-      "step": 19569
-    },
-    {
-      "epoch": 3.1516164096783283,
-      "grad_norm": 0.007895024493336678,
-      "learning_rate": 0.0001999951038234101,
-      "loss": 46.0,
-      "step": 19570
-    },
-    {
-      "epoch": 3.151777446757116,
-      "grad_norm": 0.0019134761532768607,
-      "learning_rate": 0.00019999510332276984,
-      "loss": 46.0,
-      "step": 19571
-    },
-    {
-      "epoch": 3.1519384838359032,
-      "grad_norm": 0.020334867760539055,
-      "learning_rate": 0.00019999510282210395,
-      "loss": 46.0,
-      "step": 19572
-    },
-    {
-      "epoch": 3.1520995209146907,
-      "grad_norm": 0.018368765711784363,
-      "learning_rate": 0.00019999510232141253,
-      "loss": 46.0,
-      "step": 19573
-    },
-    {
-      "epoch": 3.152260557993478,
-      "grad_norm": 0.0007146418211050332,
-      "learning_rate": 0.00019999510182069544,
-      "loss": 46.0,
-      "step": 19574
-    },
-    {
-      "epoch": 3.1524215950722656,
-      "grad_norm": 0.0016286029713228345,
-      "learning_rate": 0.00019999510131995282,
-      "loss": 46.0,
-      "step": 19575
-    },
-    {
-      "epoch": 3.1525826321510526,
-      "grad_norm": 0.006618593819439411,
-      "learning_rate": 0.0001999951008191846,
-      "loss": 46.0,
-      "step": 19576
-    },
-    {
-      "epoch": 3.15274366922984,
-      "grad_norm": 0.0036968053318560123,
-      "learning_rate": 0.00019999510031839076,
-      "loss": 46.0,
-      "step": 19577
-    },
-    {
-      "epoch": 3.1529047063086275,
-      "grad_norm": 0.008197592571377754,
-      "learning_rate": 0.00019999509981757135,
-      "loss": 46.0,
-      "step": 19578
-    },
-    {
-      "epoch": 3.153065743387415,
-      "grad_norm": 0.005064803175628185,
-      "learning_rate": 0.00019999509931672632,
-      "loss": 46.0,
-      "step": 19579
-    },
-    {
-      "epoch": 3.1532267804662024,
-      "grad_norm": 0.0037036787252873182,
-      "learning_rate": 0.00019999509881585574,
-      "loss": 46.0,
-      "step": 19580
-    },
-    {
-      "epoch": 3.15338781754499,
-      "grad_norm": 0.001896088826470077,
-      "learning_rate": 0.00019999509831495954,
-      "loss": 46.0,
-      "step": 19581
-    },
-    {
-      "epoch": 3.1535488546237773,
-      "grad_norm": 0.0016632025362923741,
-      "learning_rate": 0.00019999509781403772,
-      "loss": 46.0,
-      "step": 19582
-    },
-    {
-      "epoch": 3.1537098917025643,
-      "grad_norm": 0.0014822020893916488,
-      "learning_rate": 0.00019999509731309035,
-      "loss": 46.0,
-      "step": 19583
-    },
-    {
-      "epoch": 3.153870928781352,
-      "grad_norm": 0.0025838366709649563,
-      "learning_rate": 0.00019999509681211736,
-      "loss": 46.0,
-      "step": 19584
-    },
-    {
-      "epoch": 3.1540319658601392,
-      "grad_norm": 0.00504272012040019,
-      "learning_rate": 0.0001999950963111188,
-      "loss": 46.0,
-      "step": 19585
-    },
-    {
-      "epoch": 3.1541930029389267,
-      "grad_norm": 0.0012272970052435994,
-      "learning_rate": 0.00019999509581009462,
-      "loss": 46.0,
-      "step": 19586
-    },
-    {
-      "epoch": 3.154354040017714,
-      "grad_norm": 0.0008530979393981397,
-      "learning_rate": 0.00019999509530904485,
-      "loss": 46.0,
-      "step": 19587
-    },
-    {
-      "epoch": 3.1545150770965016,
-      "grad_norm": 0.0011674538254737854,
-      "learning_rate": 0.00019999509480796948,
-      "loss": 46.0,
-      "step": 19588
-    },
-    {
-      "epoch": 3.154676114175289,
-      "grad_norm": 0.0010253567015752196,
-      "learning_rate": 0.00019999509430686856,
-      "loss": 46.0,
-      "step": 19589
-    },
-    {
-      "epoch": 3.154837151254076,
-      "grad_norm": 0.0034656180068850517,
-      "learning_rate": 0.000199995093805742,
-      "loss": 46.0,
-      "step": 19590
-    },
-    {
-      "epoch": 3.1549981883328635,
-      "grad_norm": 0.00491347024217248,
-      "learning_rate": 0.00019999509330458987,
-      "loss": 46.0,
-      "step": 19591
-    },
-    {
-      "epoch": 3.155159225411651,
-      "grad_norm": 0.006057028193026781,
-      "learning_rate": 0.00019999509280341213,
-      "loss": 46.0,
-      "step": 19592
-    },
-    {
-      "epoch": 3.1553202624904384,
-      "grad_norm": 0.002717996947467327,
-      "learning_rate": 0.00019999509230220883,
-      "loss": 46.0,
-      "step": 19593
-    },
-    {
-      "epoch": 3.155481299569226,
-      "grad_norm": 0.0031754192896187305,
-      "learning_rate": 0.00019999509180097992,
-      "loss": 46.0,
-      "step": 19594
-    },
-    {
-      "epoch": 3.1556423366480133,
-      "grad_norm": 0.005409469828009605,
-      "learning_rate": 0.00019999509129972541,
-      "loss": 46.0,
-      "step": 19595
-    },
-    {
-      "epoch": 3.1558033737268008,
-      "grad_norm": 0.002848827512934804,
-      "learning_rate": 0.00019999509079844527,
-      "loss": 46.0,
-      "step": 19596
-    },
-    {
-      "epoch": 3.1559644108055878,
-      "grad_norm": 0.00689384201541543,
-      "learning_rate": 0.0001999950902971396,
-      "loss": 46.0,
-      "step": 19597
-    },
-    {
-      "epoch": 3.156125447884375,
-      "grad_norm": 0.016865484416484833,
-      "learning_rate": 0.00019999508979580828,
-      "loss": 46.0,
-      "step": 19598
-    },
-    {
-      "epoch": 3.1562864849631627,
-      "grad_norm": 0.003080186899751425,
-      "learning_rate": 0.0001999950892944514,
-      "loss": 46.0,
-      "step": 19599
-    },
-    {
-      "epoch": 3.15644752204195,
-      "grad_norm": 0.0034886961802840233,
-      "learning_rate": 0.0001999950887930689,
-      "loss": 46.0,
-      "step": 19600
-    },
-    {
-      "epoch": 3.1566085591207376,
-      "grad_norm": 0.0019938715267926455,
-      "learning_rate": 0.00019999508829166086,
-      "loss": 46.0,
-      "step": 19601
-    },
-    {
-      "epoch": 3.156769596199525,
-      "grad_norm": 0.0033503801096230745,
-      "learning_rate": 0.0001999950877902272,
-      "loss": 46.0,
-      "step": 19602
-    },
-    {
-      "epoch": 3.1569306332783125,
-      "grad_norm": 0.0061528743244707584,
-      "learning_rate": 0.00019999508728876791,
-      "loss": 46.0,
-      "step": 19603
-    },
-    {
-      "epoch": 3.1570916703571,
-      "grad_norm": 0.0010879081673920155,
-      "learning_rate": 0.00019999508678728307,
-      "loss": 46.0,
-      "step": 19604
-    },
-    {
-      "epoch": 3.157252707435887,
-      "grad_norm": 0.007956474088132381,
-      "learning_rate": 0.00019999508628577265,
-      "loss": 46.0,
-      "step": 19605
-    },
-    {
-      "epoch": 3.1574137445146744,
-      "grad_norm": 0.007138898596167564,
-      "learning_rate": 0.00019999508578423658,
-      "loss": 46.0,
-      "step": 19606
-    },
-    {
-      "epoch": 3.157574781593462,
-      "grad_norm": 0.00251348246820271,
-      "learning_rate": 0.00019999508528267495,
-      "loss": 46.0,
-      "step": 19607
-    },
-    {
-      "epoch": 3.1577358186722493,
-      "grad_norm": 0.006476631388068199,
-      "learning_rate": 0.00019999508478108773,
-      "loss": 46.0,
-      "step": 19608
-    },
-    {
-      "epoch": 3.1578968557510367,
-      "grad_norm": 0.002998229581862688,
-      "learning_rate": 0.00019999508427947488,
-      "loss": 46.0,
-      "step": 19609
-    },
-    {
-      "epoch": 3.158057892829824,
-      "grad_norm": 0.004335950128734112,
-      "learning_rate": 0.00019999508377783646,
-      "loss": 46.0,
-      "step": 19610
-    },
-    {
-      "epoch": 3.1582189299086116,
-      "grad_norm": 0.0023647022899240255,
-      "learning_rate": 0.00019999508327617246,
-      "loss": 46.0,
-      "step": 19611
-    },
-    {
-      "epoch": 3.1583799669873986,
-      "grad_norm": 0.004215963184833527,
-      "learning_rate": 0.00019999508277448286,
-      "loss": 46.0,
-      "step": 19612
-    },
-    {
-      "epoch": 3.158541004066186,
-      "grad_norm": 0.00445952732115984,
-      "learning_rate": 0.00019999508227276769,
-      "loss": 46.0,
-      "step": 19613
-    },
-    {
-      "epoch": 3.1587020411449735,
-      "grad_norm": 0.0020399068016558886,
-      "learning_rate": 0.0001999950817710269,
-      "loss": 46.0,
-      "step": 19614
-    },
-    {
-      "epoch": 3.158863078223761,
-      "grad_norm": 0.0036151974927634,
-      "learning_rate": 0.00019999508126926048,
-      "loss": 46.0,
-      "step": 19615
-    },
-    {
-      "epoch": 3.1590241153025485,
-      "grad_norm": 0.004897549748420715,
-      "learning_rate": 0.00019999508076746852,
-      "loss": 46.0,
-      "step": 19616
-    },
-    {
-      "epoch": 3.159185152381336,
-      "grad_norm": 0.001708691823296249,
-      "learning_rate": 0.00019999508026565094,
-      "loss": 46.0,
-      "step": 19617
-    },
-    {
-      "epoch": 3.1593461894601234,
-      "grad_norm": 0.0015553038101643324,
-      "learning_rate": 0.0001999950797638078,
-      "loss": 46.0,
-      "step": 19618
-    },
-    {
-      "epoch": 3.159507226538911,
-      "grad_norm": 0.0017440100200474262,
-      "learning_rate": 0.00019999507926193904,
-      "loss": 46.0,
-      "step": 19619
-    },
-    {
-      "epoch": 3.159668263617698,
-      "grad_norm": 0.0018952075624838471,
-      "learning_rate": 0.0001999950787600447,
-      "loss": 46.0,
-      "step": 19620
-    },
-    {
-      "epoch": 3.1598293006964853,
-      "grad_norm": 0.011986377649009228,
-      "learning_rate": 0.00019999507825812474,
-      "loss": 46.0,
-      "step": 19621
-    },
-    {
-      "epoch": 3.1599903377752727,
-      "grad_norm": 0.0020185604225844145,
-      "learning_rate": 0.0001999950777561792,
-      "loss": 46.0,
-      "step": 19622
-    },
-    {
-      "epoch": 3.16015137485406,
-      "grad_norm": 0.00163203664124012,
-      "learning_rate": 0.00019999507725420806,
-      "loss": 46.0,
-      "step": 19623
-    },
-    {
-      "epoch": 3.1603124119328476,
-      "grad_norm": 0.006958762649446726,
-      "learning_rate": 0.00019999507675221134,
-      "loss": 46.0,
-      "step": 19624
-    },
-    {
-      "epoch": 3.160473449011635,
-      "grad_norm": 0.0051189446821808815,
-      "learning_rate": 0.00019999507625018904,
-      "loss": 46.0,
-      "step": 19625
-    },
-    {
-      "epoch": 3.1606344860904225,
-      "grad_norm": 0.003895426169037819,
-      "learning_rate": 0.00019999507574814112,
-      "loss": 46.0,
-      "step": 19626
-    },
-    {
-      "epoch": 3.1607955231692095,
-      "grad_norm": 0.01487557590007782,
-      "learning_rate": 0.0001999950752460676,
-      "loss": 46.0,
-      "step": 19627
-    },
-    {
-      "epoch": 3.160956560247997,
-      "grad_norm": 0.0015625845408067107,
-      "learning_rate": 0.00019999507474396851,
-      "loss": 46.0,
-      "step": 19628
-    },
-    {
-      "epoch": 3.1611175973267844,
-      "grad_norm": 0.0030629541724920273,
-      "learning_rate": 0.0001999950742418438,
-      "loss": 46.0,
-      "step": 19629
-    },
-    {
-      "epoch": 3.161278634405572,
-      "grad_norm": 0.002921532606706023,
-      "learning_rate": 0.00019999507373969354,
-      "loss": 46.0,
-      "step": 19630
-    },
-    {
-      "epoch": 3.1614396714843593,
-      "grad_norm": 0.0029720959719270468,
-      "learning_rate": 0.00019999507323751763,
-      "loss": 46.0,
-      "step": 19631
-    },
-    {
-      "epoch": 3.161600708563147,
-      "grad_norm": 0.0026841016951948404,
-      "learning_rate": 0.00019999507273531615,
-      "loss": 46.0,
-      "step": 19632
-    },
-    {
-      "epoch": 3.1617617456419342,
-      "grad_norm": 0.024503102526068687,
-      "learning_rate": 0.0001999950722330891,
-      "loss": 46.0,
-      "step": 19633
-    },
-    {
-      "epoch": 3.1619227827207212,
-      "grad_norm": 0.0014513133792206645,
-      "learning_rate": 0.00019999507173083642,
-      "loss": 46.0,
-      "step": 19634
-    },
-    {
-      "epoch": 3.1620838197995087,
-      "grad_norm": 0.0015343984123319387,
-      "learning_rate": 0.0001999950712285582,
-      "loss": 46.0,
-      "step": 19635
-    },
-    {
-      "epoch": 3.162244856878296,
-      "grad_norm": 0.0011266530491411686,
-      "learning_rate": 0.00019999507072625432,
-      "loss": 46.0,
-      "step": 19636
-    },
-    {
-      "epoch": 3.1624058939570836,
-      "grad_norm": 0.006266784854233265,
-      "learning_rate": 0.00019999507022392489,
-      "loss": 46.0,
-      "step": 19637
-    },
-    {
-      "epoch": 3.162566931035871,
-      "grad_norm": 0.0020720544271171093,
-      "learning_rate": 0.00019999506972156984,
-      "loss": 46.0,
-      "step": 19638
-    },
-    {
-      "epoch": 3.1627279681146585,
-      "grad_norm": 0.0019829608500003815,
-      "learning_rate": 0.0001999950692191892,
-      "loss": 46.0,
-      "step": 19639
-    },
-    {
-      "epoch": 3.162889005193446,
-      "grad_norm": 0.0038157084491103888,
-      "learning_rate": 0.000199995068716783,
-      "loss": 46.0,
-      "step": 19640
-    },
-    {
-      "epoch": 3.163050042272233,
-      "grad_norm": 0.0044404580257833,
-      "learning_rate": 0.00019999506821435117,
-      "loss": 46.0,
-      "step": 19641
-    },
-    {
-      "epoch": 3.1632110793510204,
-      "grad_norm": 0.002125629223883152,
-      "learning_rate": 0.00019999506771189375,
-      "loss": 46.0,
-      "step": 19642
-    },
-    {
-      "epoch": 3.163372116429808,
-      "grad_norm": 0.0030336007475852966,
-      "learning_rate": 0.00019999506720941076,
-      "loss": 46.0,
-      "step": 19643
-    },
-    {
-      "epoch": 3.1635331535085953,
-      "grad_norm": 0.003360901726409793,
-      "learning_rate": 0.00019999506670690217,
-      "loss": 46.0,
-      "step": 19644
-    },
-    {
-      "epoch": 3.1636941905873828,
-      "grad_norm": 0.0054570771753787994,
-      "learning_rate": 0.00019999506620436795,
-      "loss": 46.0,
-      "step": 19645
-    },
-    {
-      "epoch": 3.1638552276661702,
-      "grad_norm": 0.0010881581110879779,
-      "learning_rate": 0.00019999506570180818,
-      "loss": 46.0,
-      "step": 19646
-    },
-    {
-      "epoch": 3.1640162647449577,
-      "grad_norm": 0.002101463731378317,
-      "learning_rate": 0.0001999950651992228,
-      "loss": 46.0,
-      "step": 19647
-    },
-    {
-      "epoch": 3.164177301823745,
-      "grad_norm": 0.0011758340988308191,
-      "learning_rate": 0.00019999506469661182,
-      "loss": 46.0,
-      "step": 19648
-    },
-    {
-      "epoch": 3.164338338902532,
-      "grad_norm": 0.005233744625002146,
-      "learning_rate": 0.00019999506419397526,
-      "loss": 46.0,
-      "step": 19649
-    },
-    {
-      "epoch": 3.1644993759813196,
-      "grad_norm": 0.0012121361214667559,
-      "learning_rate": 0.0001999950636913131,
-      "loss": 46.0,
-      "step": 19650
-    },
-    {
-      "epoch": 3.164660413060107,
-      "grad_norm": 0.010114284232258797,
-      "learning_rate": 0.00019999506318862535,
-      "loss": 46.0,
-      "step": 19651
-    },
-    {
-      "epoch": 3.1648214501388945,
-      "grad_norm": 0.011123177595436573,
-      "learning_rate": 0.000199995062685912,
-      "loss": 46.0,
-      "step": 19652
-    },
-    {
-      "epoch": 3.164982487217682,
-      "grad_norm": 0.0038304938934743404,
-      "learning_rate": 0.00019999506218317307,
-      "loss": 46.0,
-      "step": 19653
-    },
-    {
-      "epoch": 3.1651435242964694,
-      "grad_norm": 0.0006281990790739655,
-      "learning_rate": 0.00019999506168040854,
-      "loss": 46.0,
-      "step": 19654
-    },
-    {
-      "epoch": 3.165304561375257,
-      "grad_norm": 0.005770376417785883,
-      "learning_rate": 0.0001999950611776184,
-      "loss": 46.0,
-      "step": 19655
-    },
-    {
-      "epoch": 3.165465598454044,
-      "grad_norm": 0.002181033371016383,
-      "learning_rate": 0.00019999506067480268,
-      "loss": 46.0,
-      "step": 19656
-    },
-    {
-      "epoch": 3.1656266355328313,
-      "grad_norm": 0.0061495923437178135,
-      "learning_rate": 0.00019999506017196134,
-      "loss": 46.0,
-      "step": 19657
-    },
-    {
-      "epoch": 3.1657876726116188,
-      "grad_norm": 0.012446665205061436,
-      "learning_rate": 0.00019999505966909444,
-      "loss": 46.0,
-      "step": 19658
-    },
-    {
-      "epoch": 3.165948709690406,
-      "grad_norm": 0.0030658040195703506,
-      "learning_rate": 0.00019999505916620195,
-      "loss": 46.0,
-      "step": 19659
-    },
-    {
-      "epoch": 3.1661097467691937,
-      "grad_norm": 0.0023766884114593267,
-      "learning_rate": 0.00019999505866328385,
-      "loss": 46.0,
-      "step": 19660
-    },
-    {
-      "epoch": 3.166270783847981,
-      "grad_norm": 0.007102517876774073,
-      "learning_rate": 0.00019999505816034016,
-      "loss": 46.0,
-      "step": 19661
-    },
-    {
-      "epoch": 3.1664318209267686,
-      "grad_norm": 0.005342569667845964,
-      "learning_rate": 0.0001999950576573709,
-      "loss": 46.0,
-      "step": 19662
-    },
-    {
-      "epoch": 3.166592858005556,
-      "grad_norm": 0.0024085398763418198,
-      "learning_rate": 0.000199995057154376,
-      "loss": 46.0,
-      "step": 19663
-    },
-    {
-      "epoch": 3.166753895084343,
-      "grad_norm": 0.006693227216601372,
-      "learning_rate": 0.00019999505665135555,
-      "loss": 46.0,
-      "step": 19664
-    },
-    {
-      "epoch": 3.1669149321631305,
-      "grad_norm": 0.0009110403480008245,
-      "learning_rate": 0.00019999505614830946,
-      "loss": 46.0,
-      "step": 19665
-    },
-    {
-      "epoch": 3.167075969241918,
-      "grad_norm": 0.0009074477129615843,
-      "learning_rate": 0.0001999950556452378,
-      "loss": 46.0,
-      "step": 19666
-    },
-    {
-      "epoch": 3.1672370063207054,
-      "grad_norm": 0.002921093488112092,
-      "learning_rate": 0.00019999505514214057,
-      "loss": 46.0,
-      "step": 19667
-    },
-    {
-      "epoch": 3.167398043399493,
-      "grad_norm": 0.006306065712124109,
-      "learning_rate": 0.00019999505463901774,
-      "loss": 46.0,
-      "step": 19668
-    },
-    {
-      "epoch": 3.1675590804782803,
-      "grad_norm": 0.002673779847100377,
-      "learning_rate": 0.00019999505413586928,
-      "loss": 46.0,
-      "step": 19669
-    },
-    {
-      "epoch": 3.1677201175570673,
-      "grad_norm": 0.002985946135595441,
-      "learning_rate": 0.00019999505363269525,
-      "loss": 46.0,
-      "step": 19670
-    },
-    {
-      "epoch": 3.1678811546358547,
-      "grad_norm": 0.0014855649787932634,
-      "learning_rate": 0.00019999505312949563,
-      "loss": 46.0,
-      "step": 19671
-    },
-    {
-      "epoch": 3.168042191714642,
-      "grad_norm": 0.0018030541250482202,
-      "learning_rate": 0.0001999950526262704,
-      "loss": 46.0,
-      "step": 19672
-    },
-    {
-      "epoch": 3.1682032287934296,
-      "grad_norm": 0.0012917385902255774,
-      "learning_rate": 0.0001999950521230196,
-      "loss": 46.0,
-      "step": 19673
-    },
-    {
-      "epoch": 3.168364265872217,
-      "grad_norm": 0.002281046472489834,
-      "learning_rate": 0.00019999505161974316,
-      "loss": 46.0,
-      "step": 19674
-    },
-    {
-      "epoch": 3.1685253029510045,
-      "grad_norm": 0.005363738164305687,
-      "learning_rate": 0.0001999950511164412,
-      "loss": 46.0,
-      "step": 19675
-    },
-    {
-      "epoch": 3.168686340029792,
-      "grad_norm": 0.010731049813330173,
-      "learning_rate": 0.0001999950506131136,
-      "loss": 46.0,
-      "step": 19676
-    },
-    {
-      "epoch": 3.1688473771085794,
-      "grad_norm": 0.008554967120289803,
-      "learning_rate": 0.0001999950501097604,
-      "loss": 46.0,
-      "step": 19677
-    },
-    {
-      "epoch": 3.1690084141873665,
-      "grad_norm": 0.01828489452600479,
-      "learning_rate": 0.00019999504960638162,
-      "loss": 46.0,
-      "step": 19678
-    },
-    {
-      "epoch": 3.169169451266154,
-      "grad_norm": 0.0017399616772308946,
-      "learning_rate": 0.00019999504910297726,
-      "loss": 46.0,
-      "step": 19679
-    },
-    {
-      "epoch": 3.1693304883449414,
-      "grad_norm": 0.004214193671941757,
-      "learning_rate": 0.00019999504859954728,
-      "loss": 46.0,
-      "step": 19680
-    },
-    {
-      "epoch": 3.169491525423729,
-      "grad_norm": 0.004244233947247267,
-      "learning_rate": 0.0001999950480960917,
-      "loss": 46.0,
-      "step": 19681
-    },
-    {
-      "epoch": 3.1696525625025163,
-      "grad_norm": 0.009747356176376343,
-      "learning_rate": 0.00019999504759261056,
-      "loss": 46.0,
-      "step": 19682
-    },
-    {
-      "epoch": 3.1698135995813037,
-      "grad_norm": 0.01339562889188528,
-      "learning_rate": 0.00019999504708910381,
-      "loss": 46.0,
-      "step": 19683
-    },
-    {
-      "epoch": 3.169974636660091,
-      "grad_norm": 0.0027845578733831644,
-      "learning_rate": 0.00019999504658557146,
-      "loss": 46.0,
-      "step": 19684
-    },
-    {
-      "epoch": 3.170135673738878,
-      "grad_norm": 0.0005256269359961152,
-      "learning_rate": 0.00019999504608201354,
-      "loss": 46.0,
-      "step": 19685
-    },
-    {
-      "epoch": 3.1702967108176656,
-      "grad_norm": 0.0026178061962127686,
-      "learning_rate": 0.00019999504557843,
-      "loss": 46.0,
-      "step": 19686
-    },
-    {
-      "epoch": 3.170457747896453,
-      "grad_norm": 0.016922473907470703,
-      "learning_rate": 0.00019999504507482087,
-      "loss": 46.0,
-      "step": 19687
-    },
-    {
-      "epoch": 3.1706187849752405,
-      "grad_norm": 0.001348848338238895,
-      "learning_rate": 0.00019999504457118616,
-      "loss": 46.0,
-      "step": 19688
-    },
-    {
-      "epoch": 3.170779822054028,
-      "grad_norm": 0.002857778687030077,
-      "learning_rate": 0.00019999504406752587,
-      "loss": 46.0,
-      "step": 19689
-    },
-    {
-      "epoch": 3.1709408591328154,
-      "grad_norm": 0.00183499651029706,
-      "learning_rate": 0.00019999504356383997,
-      "loss": 46.0,
-      "step": 19690
-    },
-    {
-      "epoch": 3.171101896211603,
-      "grad_norm": 0.0021241542417556047,
-      "learning_rate": 0.00019999504306012847,
-      "loss": 46.0,
-      "step": 19691
-    },
-    {
-      "epoch": 3.1712629332903903,
-      "grad_norm": 0.02402188442647457,
-      "learning_rate": 0.00019999504255639137,
-      "loss": 46.0,
-      "step": 19692
-    },
-    {
-      "epoch": 3.1714239703691773,
-      "grad_norm": 0.005322204902768135,
-      "learning_rate": 0.0001999950420526287,
-      "loss": 46.0,
-      "step": 19693
-    },
-    {
-      "epoch": 3.171585007447965,
-      "grad_norm": 0.0005124992458149791,
-      "learning_rate": 0.0001999950415488404,
-      "loss": 46.0,
-      "step": 19694
-    },
-    {
-      "epoch": 3.1717460445267522,
-      "grad_norm": 0.0014059654204174876,
-      "learning_rate": 0.00019999504104502652,
-      "loss": 46.0,
-      "step": 19695
-    },
-    {
-      "epoch": 3.1719070816055397,
-      "grad_norm": 0.0015810240292921662,
-      "learning_rate": 0.00019999504054118704,
-      "loss": 46.0,
-      "step": 19696
-    },
-    {
-      "epoch": 3.172068118684327,
-      "grad_norm": 0.002150014042854309,
-      "learning_rate": 0.00019999504003732202,
-      "loss": 46.0,
-      "step": 19697
-    },
-    {
-      "epoch": 3.1722291557631146,
-      "grad_norm": 0.005694233346730471,
-      "learning_rate": 0.00019999503953343134,
-      "loss": 46.0,
-      "step": 19698
-    },
-    {
-      "epoch": 3.172390192841902,
-      "grad_norm": 0.004789724014699459,
-      "learning_rate": 0.00019999503902951512,
-      "loss": 46.0,
-      "step": 19699
-    },
-    {
-      "epoch": 3.172551229920689,
-      "grad_norm": 0.0008061034604907036,
-      "learning_rate": 0.00019999503852557326,
-      "loss": 46.0,
-      "step": 19700
-    },
-    {
-      "epoch": 3.1727122669994765,
-      "grad_norm": 0.0015269630821421742,
-      "learning_rate": 0.00019999503802160584,
-      "loss": 46.0,
-      "step": 19701
-    },
-    {
-      "epoch": 3.172873304078264,
-      "grad_norm": 0.0024221769999712706,
-      "learning_rate": 0.0001999950375176128,
-      "loss": 46.0,
-      "step": 19702
-    },
-    {
-      "epoch": 3.1730343411570514,
-      "grad_norm": 0.0031108269467949867,
-      "learning_rate": 0.00019999503701359416,
-      "loss": 46.0,
-      "step": 19703
-    },
-    {
-      "epoch": 3.173195378235839,
-      "grad_norm": 0.0027087212074548006,
-      "learning_rate": 0.00019999503650954998,
-      "loss": 46.0,
-      "step": 19704
-    },
-    {
-      "epoch": 3.1733564153146263,
-      "grad_norm": 0.0023988927714526653,
-      "learning_rate": 0.00019999503600548019,
-      "loss": 46.0,
-      "step": 19705
-    },
-    {
-      "epoch": 3.1735174523934138,
-      "grad_norm": 0.0014148681657388806,
-      "learning_rate": 0.00019999503550138475,
-      "loss": 46.0,
-      "step": 19706
-    },
-    {
-      "epoch": 3.1736784894722008,
-      "grad_norm": 0.007136817555874586,
-      "learning_rate": 0.00019999503499726378,
-      "loss": 46.0,
-      "step": 19707
-    },
-    {
-      "epoch": 3.173839526550988,
-      "grad_norm": 0.002934227231889963,
-      "learning_rate": 0.0001999950344931172,
-      "loss": 46.0,
-      "step": 19708
-    },
-    {
-      "epoch": 3.1740005636297757,
-      "grad_norm": 0.008164871484041214,
-      "learning_rate": 0.000199995033988945,
-      "loss": 46.0,
-      "step": 19709
-    },
-    {
-      "epoch": 3.174161600708563,
-      "grad_norm": 0.011660341173410416,
-      "learning_rate": 0.00019999503348474721,
-      "loss": 46.0,
-      "step": 19710
-    },
-    {
-      "epoch": 3.1743226377873506,
-      "grad_norm": 0.0023767270613461733,
-      "learning_rate": 0.00019999503298052384,
-      "loss": 46.0,
-      "step": 19711
-    },
-    {
-      "epoch": 3.174483674866138,
-      "grad_norm": 0.0013685694430023432,
-      "learning_rate": 0.00019999503247627488,
-      "loss": 46.0,
-      "step": 19712
-    },
-    {
-      "epoch": 3.1746447119449255,
-      "grad_norm": 0.0022953858133405447,
-      "learning_rate": 0.0001999950319720003,
-      "loss": 46.0,
-      "step": 19713
-    },
-    {
-      "epoch": 3.1748057490237125,
-      "grad_norm": 0.0007009068503975868,
-      "learning_rate": 0.00019999503146770018,
-      "loss": 46.0,
-      "step": 19714
-    },
-    {
-      "epoch": 3.1749667861025,
-      "grad_norm": 0.0010385161731392145,
-      "learning_rate": 0.00019999503096337443,
-      "loss": 46.0,
-      "step": 19715
-    },
-    {
-      "epoch": 3.1751278231812874,
-      "grad_norm": 0.0037897727452218533,
-      "learning_rate": 0.0001999950304590231,
-      "loss": 46.0,
-      "step": 19716
-    },
-    {
-      "epoch": 3.175288860260075,
-      "grad_norm": 0.005629927385598421,
-      "learning_rate": 0.00019999502995464617,
-      "loss": 46.0,
-      "step": 19717
-    },
-    {
-      "epoch": 3.1754498973388623,
-      "grad_norm": 0.0012219124473631382,
-      "learning_rate": 0.00019999502945024363,
-      "loss": 46.0,
-      "step": 19718
-    },
-    {
-      "epoch": 3.1756109344176497,
-      "grad_norm": 0.002781735733151436,
-      "learning_rate": 0.0001999950289458155,
-      "loss": 46.0,
-      "step": 19719
-    },
-    {
-      "epoch": 3.175771971496437,
-      "grad_norm": 0.003969190176576376,
-      "learning_rate": 0.00019999502844136183,
-      "loss": 46.0,
-      "step": 19720
-    },
-    {
-      "epoch": 3.1759330085752246,
-      "grad_norm": 0.0014375975588336587,
-      "learning_rate": 0.00019999502793688253,
-      "loss": 46.0,
-      "step": 19721
-    },
-    {
-      "epoch": 3.1760940456540117,
-      "grad_norm": 0.0033689807169139385,
-      "learning_rate": 0.0001999950274323776,
-      "loss": 46.0,
-      "step": 19722
-    },
-    {
-      "epoch": 3.176255082732799,
-      "grad_norm": 0.0009516612044535577,
-      "learning_rate": 0.00019999502692784712,
-      "loss": 46.0,
-      "step": 19723
-    },
-    {
-      "epoch": 3.1764161198115866,
-      "grad_norm": 0.003003350691869855,
-      "learning_rate": 0.00019999502642329103,
-      "loss": 46.0,
-      "step": 19724
-    },
-    {
-      "epoch": 3.176577156890374,
-      "grad_norm": 0.0013025281950831413,
-      "learning_rate": 0.00019999502591870933,
-      "loss": 46.0,
-      "step": 19725
-    },
-    {
-      "epoch": 3.1767381939691615,
-      "grad_norm": 0.0008496972732245922,
-      "learning_rate": 0.00019999502541410207,
-      "loss": 46.0,
-      "step": 19726
-    },
-    {
-      "epoch": 3.176899231047949,
-      "grad_norm": 0.0034412203822284937,
-      "learning_rate": 0.00019999502490946922,
-      "loss": 46.0,
-      "step": 19727
-    },
-    {
-      "epoch": 3.1770602681267364,
-      "grad_norm": 0.003885657526552677,
-      "learning_rate": 0.00019999502440481073,
-      "loss": 46.0,
-      "step": 19728
-    },
-    {
-      "epoch": 3.1772213052055234,
-      "grad_norm": 0.0019904214423149824,
-      "learning_rate": 0.0001999950239001267,
-      "loss": 46.0,
-      "step": 19729
-    },
-    {
-      "epoch": 3.177382342284311,
-      "grad_norm": 0.007293752394616604,
-      "learning_rate": 0.00019999502339541704,
-      "loss": 46.0,
-      "step": 19730
-    },
-    {
-      "epoch": 3.1775433793630983,
-      "grad_norm": 0.0009874945972114801,
-      "learning_rate": 0.00019999502289068182,
-      "loss": 46.0,
-      "step": 19731
-    },
-    {
-      "epoch": 3.1777044164418857,
-      "grad_norm": 0.0016249839682132006,
-      "learning_rate": 0.00019999502238592096,
-      "loss": 46.0,
-      "step": 19732
-    },
-    {
-      "epoch": 3.177865453520673,
-      "grad_norm": 0.0013751467922702432,
-      "learning_rate": 0.00019999502188113453,
-      "loss": 46.0,
-      "step": 19733
-    },
-    {
-      "epoch": 3.1780264905994606,
-      "grad_norm": 0.007252424024045467,
-      "learning_rate": 0.00019999502137632252,
-      "loss": 46.0,
-      "step": 19734
-    },
-    {
-      "epoch": 3.178187527678248,
-      "grad_norm": 0.004219766706228256,
-      "learning_rate": 0.0001999950208714849,
-      "loss": 46.0,
-      "step": 19735
-    },
-    {
-      "epoch": 3.1783485647570355,
-      "grad_norm": 0.00420208228752017,
-      "learning_rate": 0.0001999950203666217,
-      "loss": 46.0,
-      "step": 19736
-    },
-    {
-      "epoch": 3.1785096018358225,
-      "grad_norm": 0.003922350239008665,
-      "learning_rate": 0.00019999501986173288,
-      "loss": 46.0,
-      "step": 19737
-    },
-    {
-      "epoch": 3.17867063891461,
-      "grad_norm": 0.006449609529227018,
-      "learning_rate": 0.0001999950193568185,
-      "loss": 46.0,
-      "step": 19738
-    },
-    {
-      "epoch": 3.1788316759933974,
-      "grad_norm": 0.011483429931104183,
-      "learning_rate": 0.0001999950188518785,
-      "loss": 46.0,
-      "step": 19739
-    },
-    {
-      "epoch": 3.178992713072185,
-      "grad_norm": 0.004094013012945652,
-      "learning_rate": 0.0001999950183469129,
-      "loss": 46.0,
-      "step": 19740
-    },
-    {
-      "epoch": 3.1791537501509723,
-      "grad_norm": 0.003709926037117839,
-      "learning_rate": 0.00019999501784192175,
-      "loss": 46.0,
-      "step": 19741
-    },
-    {
-      "epoch": 3.17931478722976,
-      "grad_norm": 0.0010222380515187979,
-      "learning_rate": 0.00019999501733690495,
-      "loss": 46.0,
-      "step": 19742
-    },
-    {
-      "epoch": 3.1794758243085472,
-      "grad_norm": 0.0028213472105562687,
-      "learning_rate": 0.0001999950168318626,
-      "loss": 46.0,
-      "step": 19743
-    },
-    {
-      "epoch": 3.1796368613873343,
-      "grad_norm": 0.008547100238502026,
-      "learning_rate": 0.00019999501632679464,
-      "loss": 46.0,
-      "step": 19744
-    },
-    {
-      "epoch": 3.1797978984661217,
-      "grad_norm": 0.003768955823034048,
-      "learning_rate": 0.00019999501582170108,
-      "loss": 46.0,
-      "step": 19745
-    },
-    {
-      "epoch": 3.179958935544909,
-      "grad_norm": 0.002996401395648718,
-      "learning_rate": 0.00019999501531658195,
-      "loss": 46.0,
-      "step": 19746
-    },
-    {
-      "epoch": 3.1801199726236966,
-      "grad_norm": 0.004124962259083986,
-      "learning_rate": 0.00019999501481143722,
-      "loss": 46.0,
-      "step": 19747
-    },
-    {
-      "epoch": 3.180281009702484,
-      "grad_norm": 0.0015998920425772667,
-      "learning_rate": 0.00019999501430626685,
-      "loss": 46.0,
-      "step": 19748
-    },
-    {
-      "epoch": 3.1804420467812715,
-      "grad_norm": 0.002692669630050659,
-      "learning_rate": 0.00019999501380107095,
-      "loss": 46.0,
-      "step": 19749
-    },
-    {
-      "epoch": 3.180603083860059,
-      "grad_norm": 0.005620558280497789,
-      "learning_rate": 0.00019999501329584943,
-      "loss": 46.0,
-      "step": 19750
-    },
-    {
-      "epoch": 3.180764120938846,
-      "grad_norm": 0.011771776713430882,
-      "learning_rate": 0.0001999950127906023,
-      "loss": 46.0,
-      "step": 19751
-    },
-    {
-      "epoch": 3.1809251580176334,
-      "grad_norm": 0.002610383555293083,
-      "learning_rate": 0.0001999950122853296,
-      "loss": 46.0,
-      "step": 19752
-    },
-    {
-      "epoch": 3.181086195096421,
-      "grad_norm": 0.0015096035785973072,
-      "learning_rate": 0.0001999950117800313,
-      "loss": 46.0,
-      "step": 19753
-    },
-    {
-      "epoch": 3.1812472321752083,
-      "grad_norm": 0.0026770683471113443,
-      "learning_rate": 0.00019999501127470742,
-      "loss": 46.0,
-      "step": 19754
-    },
-    {
-      "epoch": 3.181408269253996,
-      "grad_norm": 0.0055971224792301655,
-      "learning_rate": 0.0001999950107693579,
-      "loss": 46.0,
-      "step": 19755
-    },
-    {
-      "epoch": 3.1815693063327832,
-      "grad_norm": 0.004829136189073324,
-      "learning_rate": 0.00019999501026398285,
-      "loss": 46.0,
-      "step": 19756
-    },
-    {
-      "epoch": 3.1817303434115707,
-      "grad_norm": 0.001391859375871718,
-      "learning_rate": 0.00019999500975858217,
-      "loss": 46.0,
-      "step": 19757
-    },
-    {
-      "epoch": 3.1818913804903577,
-      "grad_norm": 0.0016644347924739122,
-      "learning_rate": 0.0001999950092531559,
-      "loss": 46.0,
-      "step": 19758
-    },
-    {
-      "epoch": 3.182052417569145,
-      "grad_norm": 0.002994673326611519,
-      "learning_rate": 0.00019999500874770404,
-      "loss": 46.0,
-      "step": 19759
-    },
-    {
-      "epoch": 3.1822134546479326,
-      "grad_norm": 0.003188950940966606,
-      "learning_rate": 0.00019999500824222657,
-      "loss": 46.0,
-      "step": 19760
-    },
-    {
-      "epoch": 3.18237449172672,
-      "grad_norm": 0.008136718533933163,
-      "learning_rate": 0.00019999500773672352,
-      "loss": 46.0,
-      "step": 19761
-    },
-    {
-      "epoch": 3.1825355288055075,
-      "grad_norm": 0.0007128171855583787,
-      "learning_rate": 0.00019999500723119488,
-      "loss": 46.0,
-      "step": 19762
-    },
-    {
-      "epoch": 3.182696565884295,
-      "grad_norm": 0.001606535050086677,
-      "learning_rate": 0.00019999500672564065,
-      "loss": 46.0,
-      "step": 19763
-    },
-    {
-      "epoch": 3.1828576029630824,
-      "grad_norm": 0.0011022391263395548,
-      "learning_rate": 0.00019999500622006083,
-      "loss": 46.0,
-      "step": 19764
-    },
-    {
-      "epoch": 3.18301864004187,
-      "grad_norm": 0.0026857571210712194,
-      "learning_rate": 0.0001999950057144554,
-      "loss": 46.0,
-      "step": 19765
-    },
-    {
-      "epoch": 3.183179677120657,
-      "grad_norm": 0.0010084382956847548,
-      "learning_rate": 0.00019999500520882438,
-      "loss": 46.0,
-      "step": 19766
-    },
-    {
-      "epoch": 3.1833407141994443,
-      "grad_norm": 0.005197951104491949,
-      "learning_rate": 0.0001999950047031678,
-      "loss": 46.0,
-      "step": 19767
-    },
-    {
-      "epoch": 3.1835017512782318,
-      "grad_norm": 0.004256872925907373,
-      "learning_rate": 0.00019999500419748558,
-      "loss": 46.0,
-      "step": 19768
-    },
-    {
-      "epoch": 3.183662788357019,
-      "grad_norm": 0.007161410059779882,
-      "learning_rate": 0.00019999500369177777,
-      "loss": 46.0,
-      "step": 19769
-    },
-    {
-      "epoch": 3.1838238254358067,
-      "grad_norm": 0.0011229043593630195,
-      "learning_rate": 0.00019999500318604438,
-      "loss": 46.0,
-      "step": 19770
-    },
-    {
-      "epoch": 3.183984862514594,
-      "grad_norm": 0.0037400221917778254,
-      "learning_rate": 0.0001999950026802854,
-      "loss": 46.0,
-      "step": 19771
-    },
-    {
-      "epoch": 3.1841458995933816,
-      "grad_norm": 0.00586453964933753,
-      "learning_rate": 0.00019999500217450083,
-      "loss": 46.0,
-      "step": 19772
-    },
-    {
-      "epoch": 3.1843069366721686,
-      "grad_norm": 0.0017640376463532448,
-      "learning_rate": 0.00019999500166869062,
-      "loss": 46.0,
-      "step": 19773
-    },
-    {
-      "epoch": 3.184467973750956,
-      "grad_norm": 0.005383512005209923,
-      "learning_rate": 0.00019999500116285487,
-      "loss": 46.0,
-      "step": 19774
-    },
-    {
-      "epoch": 3.1846290108297435,
-      "grad_norm": 0.0017542550340294838,
-      "learning_rate": 0.00019999500065699351,
-      "loss": 46.0,
-      "step": 19775
-    },
-    {
-      "epoch": 3.184790047908531,
-      "grad_norm": 0.005183705594390631,
-      "learning_rate": 0.00019999500015110657,
-      "loss": 46.0,
-      "step": 19776
-    },
-    {
-      "epoch": 3.1849510849873184,
-      "grad_norm": 0.0045154201798141,
-      "learning_rate": 0.000199994999645194,
-      "loss": 46.0,
-      "step": 19777
-    },
-    {
-      "epoch": 3.185112122066106,
-      "grad_norm": 0.0083194300532341,
-      "learning_rate": 0.0001999949991392559,
-      "loss": 46.0,
-      "step": 19778
-    },
-    {
-      "epoch": 3.1852731591448933,
-      "grad_norm": 0.009240538813173771,
-      "learning_rate": 0.00019999499863329213,
-      "loss": 46.0,
-      "step": 19779
-    },
-    {
-      "epoch": 3.1854341962236807,
-      "grad_norm": 0.001111158519051969,
-      "learning_rate": 0.0001999949981273028,
-      "loss": 46.0,
-      "step": 19780
-    },
-    {
-      "epoch": 3.1855952333024677,
-      "grad_norm": 0.006247317884117365,
-      "learning_rate": 0.0001999949976212879,
-      "loss": 46.0,
-      "step": 19781
-    },
-    {
-      "epoch": 3.185756270381255,
-      "grad_norm": 0.008027937263250351,
-      "learning_rate": 0.00019999499711524735,
-      "loss": 46.0,
-      "step": 19782
-    },
-    {
-      "epoch": 3.1859173074600426,
-      "grad_norm": 0.0036614250857383013,
-      "learning_rate": 0.00019999499660918127,
-      "loss": 46.0,
-      "step": 19783
-    },
-    {
-      "epoch": 3.18607834453883,
-      "grad_norm": 0.00796531606465578,
-      "learning_rate": 0.00019999499610308958,
-      "loss": 46.0,
-      "step": 19784
-    },
-    {
-      "epoch": 3.1862393816176175,
-      "grad_norm": 0.013205653987824917,
-      "learning_rate": 0.00019999499559697227,
-      "loss": 46.0,
-      "step": 19785
-    },
-    {
-      "epoch": 3.186400418696405,
-      "grad_norm": 0.0025140228681266308,
-      "learning_rate": 0.0001999949950908294,
-      "loss": 46.0,
-      "step": 19786
-    },
-    {
-      "epoch": 3.186561455775192,
-      "grad_norm": 0.011965310201048851,
-      "learning_rate": 0.0001999949945846609,
-      "loss": 46.0,
-      "step": 19787
-    },
-    {
-      "epoch": 3.1867224928539795,
-      "grad_norm": 0.004593444988131523,
-      "learning_rate": 0.00019999499407846684,
-      "loss": 46.0,
-      "step": 19788
-    },
-    {
-      "epoch": 3.186883529932767,
-      "grad_norm": 0.005895540118217468,
-      "learning_rate": 0.00019999499357224715,
-      "loss": 46.0,
-      "step": 19789
-    },
-    {
-      "epoch": 3.1870445670115544,
-      "grad_norm": 0.002427153056487441,
-      "learning_rate": 0.00019999499306600188,
-      "loss": 46.0,
-      "step": 19790
-    },
-    {
-      "epoch": 3.187205604090342,
-      "grad_norm": 0.009925429709255695,
-      "learning_rate": 0.00019999499255973104,
-      "loss": 46.0,
-      "step": 19791
-    },
-    {
-      "epoch": 3.1873666411691293,
-      "grad_norm": 0.002062993822619319,
-      "learning_rate": 0.00019999499205343457,
-      "loss": 46.0,
-      "step": 19792
-    },
-    {
-      "epoch": 3.1875276782479167,
-      "grad_norm": 0.002389253117144108,
-      "learning_rate": 0.00019999499154711253,
-      "loss": 46.0,
-      "step": 19793
-    },
-    {
-      "epoch": 3.187688715326704,
-      "grad_norm": 0.003960807342082262,
-      "learning_rate": 0.00019999499104076488,
-      "loss": 46.0,
-      "step": 19794
-    },
-    {
-      "epoch": 3.187849752405491,
-      "grad_norm": 0.0035039905924350023,
-      "learning_rate": 0.00019999499053439167,
-      "loss": 46.0,
-      "step": 19795
-    },
-    {
-      "epoch": 3.1880107894842786,
-      "grad_norm": 0.002037300728261471,
-      "learning_rate": 0.00019999499002799285,
-      "loss": 46.0,
-      "step": 19796
-    },
-    {
-      "epoch": 3.188171826563066,
-      "grad_norm": 0.0019222765695303679,
-      "learning_rate": 0.00019999498952156844,
-      "loss": 46.0,
-      "step": 19797
-    },
-    {
-      "epoch": 3.1883328636418535,
-      "grad_norm": 0.0019225234864279628,
-      "learning_rate": 0.00019999498901511842,
-      "loss": 46.0,
-      "step": 19798
-    },
-    {
-      "epoch": 3.188493900720641,
-      "grad_norm": 0.01437341421842575,
-      "learning_rate": 0.00019999498850864283,
-      "loss": 46.0,
-      "step": 19799
-    },
-    {
-      "epoch": 3.1886549377994284,
-      "grad_norm": 0.008268388919532299,
-      "learning_rate": 0.00019999498800214163,
-      "loss": 46.0,
-      "step": 19800
-    },
-    {
-      "epoch": 3.188815974878216,
-      "grad_norm": 0.00804196484386921,
-      "learning_rate": 0.00019999498749561482,
-      "loss": 46.0,
-      "step": 19801
-    },
-    {
-      "epoch": 3.188977011957003,
-      "grad_norm": 0.005860279779881239,
-      "learning_rate": 0.00019999498698906242,
-      "loss": 46.0,
-      "step": 19802
-    },
-    {
-      "epoch": 3.1891380490357903,
-      "grad_norm": 0.0023101232945919037,
-      "learning_rate": 0.00019999498648248445,
-      "loss": 46.0,
-      "step": 19803
-    },
-    {
-      "epoch": 3.189299086114578,
-      "grad_norm": 0.0016302675940096378,
-      "learning_rate": 0.00019999498597588088,
-      "loss": 46.0,
-      "step": 19804
-    },
-    {
-      "epoch": 3.1894601231933652,
-      "grad_norm": 0.00838552601635456,
-      "learning_rate": 0.00019999498546925172,
-      "loss": 46.0,
-      "step": 19805
-    },
-    {
-      "epoch": 3.1896211602721527,
-      "grad_norm": 0.015919389203190804,
-      "learning_rate": 0.00019999498496259697,
-      "loss": 46.0,
-      "step": 19806
-    },
-    {
-      "epoch": 3.18978219735094,
-      "grad_norm": 0.0009158725151792169,
-      "learning_rate": 0.00019999498445591658,
-      "loss": 46.0,
-      "step": 19807
-    },
-    {
-      "epoch": 3.1899432344297276,
-      "grad_norm": 0.012250055558979511,
-      "learning_rate": 0.00019999498394921065,
-      "loss": 46.0,
-      "step": 19808
-    },
-    {
-      "epoch": 3.190104271508515,
-      "grad_norm": 0.0009662067168392241,
-      "learning_rate": 0.00019999498344247911,
-      "loss": 46.0,
-      "step": 19809
-    },
-    {
-      "epoch": 3.190265308587302,
-      "grad_norm": 0.005286928731948137,
-      "learning_rate": 0.00019999498293572196,
-      "loss": 46.0,
-      "step": 19810
-    },
-    {
-      "epoch": 3.1904263456660895,
-      "grad_norm": 0.00653979042544961,
-      "learning_rate": 0.00019999498242893925,
-      "loss": 46.0,
-      "step": 19811
-    },
-    {
-      "epoch": 3.190587382744877,
-      "grad_norm": 0.0033032251521945,
-      "learning_rate": 0.0001999949819221309,
-      "loss": 46.0,
-      "step": 19812
-    },
-    {
-      "epoch": 3.1907484198236644,
-      "grad_norm": 0.0006170686101540923,
-      "learning_rate": 0.000199994981415297,
-      "loss": 46.0,
-      "step": 19813
-    },
-    {
-      "epoch": 3.190909456902452,
-      "grad_norm": 0.006778727751225233,
-      "learning_rate": 0.00019999498090843748,
-      "loss": 46.0,
-      "step": 19814
-    },
-    {
-      "epoch": 3.1910704939812393,
-      "grad_norm": 0.006944861263036728,
-      "learning_rate": 0.0001999949804015524,
-      "loss": 46.0,
-      "step": 19815
-    },
-    {
-      "epoch": 3.1912315310600268,
-      "grad_norm": 0.002142457291483879,
-      "learning_rate": 0.00019999497989464172,
-      "loss": 46.0,
-      "step": 19816
-    },
-    {
-      "epoch": 3.191392568138814,
-      "grad_norm": 0.0023954843636602163,
-      "learning_rate": 0.0001999949793877054,
-      "loss": 46.0,
-      "step": 19817
-    },
-    {
-      "epoch": 3.1915536052176012,
-      "grad_norm": 0.0009656427428126335,
-      "learning_rate": 0.00019999497888074352,
-      "loss": 46.0,
-      "step": 19818
-    },
-    {
-      "epoch": 3.1917146422963887,
-      "grad_norm": 0.008898722007870674,
-      "learning_rate": 0.00019999497837375603,
-      "loss": 46.0,
-      "step": 19819
-    },
-    {
-      "epoch": 3.191875679375176,
-      "grad_norm": 0.0036675261799246073,
-      "learning_rate": 0.00019999497786674298,
-      "loss": 46.0,
-      "step": 19820
-    },
-    {
-      "epoch": 3.1920367164539636,
-      "grad_norm": 0.007081733550876379,
-      "learning_rate": 0.0001999949773597043,
-      "loss": 46.0,
-      "step": 19821
-    },
-    {
-      "epoch": 3.192197753532751,
-      "grad_norm": 0.0029735665302723646,
-      "learning_rate": 0.00019999497685264003,
-      "loss": 46.0,
-      "step": 19822
-    },
-    {
-      "epoch": 3.1923587906115385,
-      "grad_norm": 0.003738351399078965,
-      "learning_rate": 0.0001999949763455502,
-      "loss": 46.0,
-      "step": 19823
-    },
-    {
-      "epoch": 3.1925198276903255,
-      "grad_norm": 0.0013856987934559584,
-      "learning_rate": 0.00019999497583843474,
-      "loss": 46.0,
-      "step": 19824
-    },
-    {
-      "epoch": 3.192680864769113,
-      "grad_norm": 0.0017351857386529446,
-      "learning_rate": 0.0001999949753312937,
-      "loss": 46.0,
-      "step": 19825
-    },
-    {
-      "epoch": 3.1928419018479004,
-      "grad_norm": 0.017825378105044365,
-      "learning_rate": 0.00019999497482412707,
-      "loss": 46.0,
-      "step": 19826
-    },
-    {
-      "epoch": 3.193002938926688,
-      "grad_norm": 0.0014785912353545427,
-      "learning_rate": 0.00019999497431693485,
-      "loss": 46.0,
-      "step": 19827
-    },
-    {
-      "epoch": 3.1931639760054753,
-      "grad_norm": 0.0010605257702991366,
-      "learning_rate": 0.00019999497380971702,
-      "loss": 46.0,
-      "step": 19828
-    },
-    {
-      "epoch": 3.1933250130842628,
-      "grad_norm": 0.004136411938816309,
-      "learning_rate": 0.0001999949733024736,
-      "loss": 46.0,
-      "step": 19829
-    },
-    {
-      "epoch": 3.19348605016305,
-      "grad_norm": 0.0009177191532216966,
-      "learning_rate": 0.00019999497279520463,
-      "loss": 46.0,
-      "step": 19830
-    },
-    {
-      "epoch": 3.193647087241837,
-      "grad_norm": 0.004069517366588116,
-      "learning_rate": 0.00019999497228791,
-      "loss": 46.0,
-      "step": 19831
-    },
-    {
-      "epoch": 3.1938081243206247,
-      "grad_norm": 0.0018785659922286868,
-      "learning_rate": 0.00019999497178058983,
-      "loss": 46.0,
-      "step": 19832
-    },
-    {
-      "epoch": 3.193969161399412,
-      "grad_norm": 0.000989588093943894,
-      "learning_rate": 0.000199994971273244,
-      "loss": 46.0,
-      "step": 19833
-    },
-    {
-      "epoch": 3.1941301984781996,
-      "grad_norm": 0.008440427482128143,
-      "learning_rate": 0.00019999497076587263,
-      "loss": 46.0,
-      "step": 19834
-    },
-    {
-      "epoch": 3.194291235556987,
-      "grad_norm": 0.0034890423994511366,
-      "learning_rate": 0.00019999497025847566,
-      "loss": 46.0,
-      "step": 19835
-    },
-    {
-      "epoch": 3.1944522726357745,
-      "grad_norm": 0.0015511981910094619,
-      "learning_rate": 0.0001999949697510531,
-      "loss": 46.0,
-      "step": 19836
-    },
-    {
-      "epoch": 3.194613309714562,
-      "grad_norm": 0.002736080205067992,
-      "learning_rate": 0.00019999496924360494,
-      "loss": 46.0,
-      "step": 19837
-    },
-    {
-      "epoch": 3.1947743467933494,
-      "grad_norm": 0.0018079903675243258,
-      "learning_rate": 0.00019999496873613119,
-      "loss": 46.0,
-      "step": 19838
-    },
-    {
-      "epoch": 3.1949353838721364,
-      "grad_norm": 0.0037863650359213352,
-      "learning_rate": 0.00019999496822863184,
-      "loss": 46.0,
-      "step": 19839
-    },
-    {
-      "epoch": 3.195096420950924,
-      "grad_norm": 0.0018491890514269471,
-      "learning_rate": 0.0001999949677211069,
-      "loss": 46.0,
-      "step": 19840
-    },
-    {
-      "epoch": 3.1952574580297113,
-      "grad_norm": 0.0014581484720110893,
-      "learning_rate": 0.00019999496721355637,
-      "loss": 46.0,
-      "step": 19841
-    },
-    {
-      "epoch": 3.1954184951084987,
-      "grad_norm": 0.0014562910655513406,
-      "learning_rate": 0.00019999496670598024,
-      "loss": 46.0,
-      "step": 19842
-    },
-    {
-      "epoch": 3.195579532187286,
-      "grad_norm": 0.003640740644186735,
-      "learning_rate": 0.00019999496619837852,
-      "loss": 46.0,
-      "step": 19843
-    },
-    {
-      "epoch": 3.1957405692660736,
-      "grad_norm": 0.0029499635566025972,
-      "learning_rate": 0.0001999949656907512,
-      "loss": 46.0,
-      "step": 19844
-    },
-    {
-      "epoch": 3.195901606344861,
-      "grad_norm": 0.0012260916410014033,
-      "learning_rate": 0.0001999949651830983,
-      "loss": 46.0,
-      "step": 19845
-    },
-    {
-      "epoch": 3.196062643423648,
-      "grad_norm": 0.0007619918324053288,
-      "learning_rate": 0.00019999496467541978,
-      "loss": 46.0,
-      "step": 19846
-    },
-    {
-      "epoch": 3.1962236805024355,
-      "grad_norm": 0.005921292584389448,
-      "learning_rate": 0.0001999949641677157,
-      "loss": 46.0,
-      "step": 19847
-    },
-    {
-      "epoch": 3.196384717581223,
-      "grad_norm": 0.0020439988002181053,
-      "learning_rate": 0.000199994963659986,
-      "loss": 46.0,
-      "step": 19848
-    },
-    {
-      "epoch": 3.1965457546600105,
-      "grad_norm": 0.008453522808849812,
-      "learning_rate": 0.00019999496315223068,
-      "loss": 46.0,
-      "step": 19849
-    },
-    {
-      "epoch": 3.196706791738798,
-      "grad_norm": 0.007206012029200792,
-      "learning_rate": 0.0001999949626444498,
-      "loss": 46.0,
-      "step": 19850
-    },
-    {
-      "epoch": 3.1968678288175854,
-      "grad_norm": 0.010151001624763012,
-      "learning_rate": 0.00019999496213664336,
-      "loss": 46.0,
-      "step": 19851
-    },
-    {
-      "epoch": 3.197028865896373,
-      "grad_norm": 0.0028867933433502913,
-      "learning_rate": 0.00019999496162881127,
-      "loss": 46.0,
-      "step": 19852
-    },
-    {
-      "epoch": 3.1971899029751603,
-      "grad_norm": 0.005977341905236244,
-      "learning_rate": 0.00019999496112095363,
-      "loss": 46.0,
-      "step": 19853
-    },
-    {
-      "epoch": 3.1973509400539473,
-      "grad_norm": 0.007132588420063257,
-      "learning_rate": 0.0001999949606130704,
-      "loss": 46.0,
-      "step": 19854
-    },
-    {
-      "epoch": 3.1975119771327347,
-      "grad_norm": 0.0011176193365827203,
-      "learning_rate": 0.00019999496010516152,
-      "loss": 46.0,
-      "step": 19855
-    },
-    {
-      "epoch": 3.197673014211522,
-      "grad_norm": 0.003553074086084962,
-      "learning_rate": 0.0001999949595972271,
-      "loss": 46.0,
-      "step": 19856
-    },
-    {
-      "epoch": 3.1978340512903096,
-      "grad_norm": 0.001030263607390225,
-      "learning_rate": 0.00019999495908926704,
-      "loss": 46.0,
-      "step": 19857
-    },
-    {
-      "epoch": 3.197995088369097,
-      "grad_norm": 0.0005515654920600355,
-      "learning_rate": 0.00019999495858128143,
-      "loss": 46.0,
-      "step": 19858
-    },
-    {
-      "epoch": 3.1981561254478845,
-      "grad_norm": 0.0038948762230575085,
-      "learning_rate": 0.0001999949580732702,
-      "loss": 46.0,
-      "step": 19859
-    },
-    {
-      "epoch": 3.1983171625266715,
-      "grad_norm": 0.012143001891672611,
-      "learning_rate": 0.00019999495756523337,
-      "loss": 46.0,
-      "step": 19860
-    },
-    {
-      "epoch": 3.198478199605459,
-      "grad_norm": 0.0031428919173777103,
-      "learning_rate": 0.000199994957057171,
-      "loss": 46.0,
-      "step": 19861
-    },
-    {
-      "epoch": 3.1986392366842464,
-      "grad_norm": 0.007391807623207569,
-      "learning_rate": 0.00019999495654908297,
-      "loss": 46.0,
-      "step": 19862
-    },
-    {
-      "epoch": 3.198800273763034,
-      "grad_norm": 0.0018393341451883316,
-      "learning_rate": 0.0001999949560409694,
-      "loss": 46.0,
-      "step": 19863
-    },
-    {
-      "epoch": 3.1989613108418213,
-      "grad_norm": 0.0036613158881664276,
-      "learning_rate": 0.0001999949555328302,
-      "loss": 46.0,
-      "step": 19864
-    },
-    {
-      "epoch": 3.199122347920609,
-      "grad_norm": 0.001508887391537428,
-      "learning_rate": 0.0001999949550246654,
-      "loss": 46.0,
-      "step": 19865
-    },
-    {
-      "epoch": 3.1992833849993962,
-      "grad_norm": 0.005387998651713133,
-      "learning_rate": 0.00019999495451647503,
-      "loss": 46.0,
-      "step": 19866
-    },
-    {
-      "epoch": 3.1994444220781837,
-      "grad_norm": 0.01063913106918335,
-      "learning_rate": 0.00019999495400825908,
-      "loss": 46.0,
-      "step": 19867
-    },
-    {
-      "epoch": 3.1996054591569707,
-      "grad_norm": 0.0027136628050357103,
-      "learning_rate": 0.00019999495350001752,
-      "loss": 46.0,
-      "step": 19868
-    },
-    {
-      "epoch": 3.199766496235758,
-      "grad_norm": 0.0026820707134902477,
-      "learning_rate": 0.00019999495299175034,
-      "loss": 46.0,
-      "step": 19869
-    },
-    {
-      "epoch": 3.1999275333145456,
-      "grad_norm": 0.0017326547531411052,
-      "learning_rate": 0.0001999949524834576,
-      "loss": 46.0,
-      "step": 19870
-    },
-    {
-      "epoch": 3.200088570393333,
-      "grad_norm": 0.004446511156857014,
-      "learning_rate": 0.00019999495197513926,
-      "loss": 46.0,
-      "step": 19871
-    },
-    {
-      "epoch": 3.2002496074721205,
-      "grad_norm": 0.0013527220580726862,
-      "learning_rate": 0.00019999495146679532,
-      "loss": 46.0,
-      "step": 19872
-    },
-    {
-      "epoch": 3.200410644550908,
-      "grad_norm": 0.008834746666252613,
-      "learning_rate": 0.0001999949509584258,
-      "loss": 46.0,
-      "step": 19873
-    },
-    {
-      "epoch": 3.2005716816296954,
-      "grad_norm": 0.010460826568305492,
-      "learning_rate": 0.00019999495045003066,
-      "loss": 46.0,
-      "step": 19874
-    },
-    {
-      "epoch": 3.2007327187084824,
-      "grad_norm": 0.0028877719305455685,
-      "learning_rate": 0.00019999494994160996,
-      "loss": 46.0,
-      "step": 19875
-    },
-    {
-      "epoch": 3.20089375578727,
-      "grad_norm": 0.0012199478223919868,
-      "learning_rate": 0.00019999494943316365,
-      "loss": 46.0,
-      "step": 19876
-    },
-    {
-      "epoch": 3.2010547928660573,
-      "grad_norm": 0.0007976951310411096,
-      "learning_rate": 0.00019999494892469172,
-      "loss": 46.0,
-      "step": 19877
-    },
-    {
-      "epoch": 3.2012158299448448,
-      "grad_norm": 0.0009848412591964006,
-      "learning_rate": 0.00019999494841619424,
-      "loss": 46.0,
-      "step": 19878
-    },
-    {
-      "epoch": 3.2013768670236322,
-      "grad_norm": 0.0006164132500998676,
-      "learning_rate": 0.00019999494790767114,
-      "loss": 46.0,
-      "step": 19879
-    },
-    {
-      "epoch": 3.2015379041024197,
-      "grad_norm": 0.006771137472242117,
-      "learning_rate": 0.00019999494739912245,
-      "loss": 46.0,
-      "step": 19880
-    },
-    {
-      "epoch": 3.201698941181207,
-      "grad_norm": 0.0028572995215654373,
-      "learning_rate": 0.00019999494689054817,
-      "loss": 46.0,
-      "step": 19881
-    },
-    {
-      "epoch": 3.2018599782599946,
-      "grad_norm": 0.0038422876968979836,
-      "learning_rate": 0.0001999949463819483,
-      "loss": 46.0,
-      "step": 19882
-    },
-    {
-      "epoch": 3.2020210153387816,
-      "grad_norm": 0.013455407693982124,
-      "learning_rate": 0.00019999494587332283,
-      "loss": 46.0,
-      "step": 19883
-    },
-    {
-      "epoch": 3.202182052417569,
-      "grad_norm": 0.01469449419528246,
-      "learning_rate": 0.0001999949453646718,
-      "loss": 46.0,
-      "step": 19884
-    },
-    {
-      "epoch": 3.2023430894963565,
-      "grad_norm": 0.0018090845551341772,
-      "learning_rate": 0.00019999494485599512,
-      "loss": 46.0,
-      "step": 19885
-    },
-    {
-      "epoch": 3.202504126575144,
-      "grad_norm": 0.000849371834192425,
-      "learning_rate": 0.00019999494434729288,
-      "loss": 46.0,
-      "step": 19886
-    },
-    {
-      "epoch": 3.2026651636539314,
-      "grad_norm": 0.0017237755237147212,
-      "learning_rate": 0.00019999494383856505,
-      "loss": 46.0,
-      "step": 19887
-    },
-    {
-      "epoch": 3.202826200732719,
-      "grad_norm": 0.0022191668394953012,
-      "learning_rate": 0.0001999949433298116,
-      "loss": 46.0,
-      "step": 19888
-    },
-    {
-      "epoch": 3.2029872378115063,
-      "grad_norm": 0.008016888052225113,
-      "learning_rate": 0.00019999494282103256,
-      "loss": 46.0,
-      "step": 19889
-    },
-    {
-      "epoch": 3.2031482748902933,
-      "grad_norm": 0.0024024019949138165,
-      "learning_rate": 0.00019999494231222795,
-      "loss": 46.0,
-      "step": 19890
-    },
-    {
-      "epoch": 3.2033093119690808,
-      "grad_norm": 0.001229557441547513,
-      "learning_rate": 0.00019999494180339772,
-      "loss": 46.0,
-      "step": 19891
-    },
-    {
-      "epoch": 3.203470349047868,
-      "grad_norm": 0.0033974347170442343,
-      "learning_rate": 0.00019999494129454193,
-      "loss": 46.0,
-      "step": 19892
-    },
-    {
-      "epoch": 3.2036313861266557,
-      "grad_norm": 0.0015074391849339008,
-      "learning_rate": 0.00019999494078566055,
-      "loss": 46.0,
-      "step": 19893
-    },
-    {
-      "epoch": 3.203792423205443,
-      "grad_norm": 0.008552758023142815,
-      "learning_rate": 0.00019999494027675354,
-      "loss": 46.0,
-      "step": 19894
-    },
-    {
-      "epoch": 3.2039534602842306,
-      "grad_norm": 0.002506379736587405,
-      "learning_rate": 0.00019999493976782096,
-      "loss": 46.0,
-      "step": 19895
-    },
-    {
-      "epoch": 3.204114497363018,
-      "grad_norm": 0.0010894174920395017,
-      "learning_rate": 0.00019999493925886274,
-      "loss": 46.0,
-      "step": 19896
-    },
-    {
-      "epoch": 3.204275534441805,
-      "grad_norm": 0.0034105246886610985,
-      "learning_rate": 0.000199994938749879,
-      "loss": 46.0,
-      "step": 19897
-    },
-    {
-      "epoch": 3.2044365715205925,
-      "grad_norm": 0.02548542059957981,
-      "learning_rate": 0.00019999493824086962,
-      "loss": 46.0,
-      "step": 19898
-    },
-    {
-      "epoch": 3.20459760859938,
-      "grad_norm": 0.012117553502321243,
-      "learning_rate": 0.00019999493773183464,
-      "loss": 46.0,
-      "step": 19899
-    },
-    {
-      "epoch": 3.2047586456781674,
-      "grad_norm": 0.002581066219136119,
-      "learning_rate": 0.0001999949372227741,
-      "loss": 46.0,
-      "step": 19900
-    },
-    {
-      "epoch": 3.204919682756955,
-      "grad_norm": 0.0008938151295296848,
-      "learning_rate": 0.00019999493671368795,
-      "loss": 46.0,
-      "step": 19901
-    },
-    {
-      "epoch": 3.2050807198357423,
-      "grad_norm": 0.0030661774799227715,
-      "learning_rate": 0.00019999493620457618,
-      "loss": 46.0,
-      "step": 19902
-    },
-    {
-      "epoch": 3.2052417569145297,
-      "grad_norm": 0.0034973423462361097,
-      "learning_rate": 0.00019999493569543885,
-      "loss": 46.0,
-      "step": 19903
-    },
-    {
-      "epoch": 3.2054027939933167,
-      "grad_norm": 0.001700281398370862,
-      "learning_rate": 0.00019999493518627594,
-      "loss": 46.0,
-      "step": 19904
-    },
-    {
-      "epoch": 3.205563831072104,
-      "grad_norm": 0.012507494539022446,
-      "learning_rate": 0.00019999493467708738,
-      "loss": 46.0,
-      "step": 19905
-    },
-    {
-      "epoch": 3.2057248681508916,
-      "grad_norm": 0.002298508770763874,
-      "learning_rate": 0.0001999949341678733,
-      "loss": 46.0,
-      "step": 19906
-    },
-    {
-      "epoch": 3.205885905229679,
-      "grad_norm": 0.0019406152423471212,
-      "learning_rate": 0.00019999493365863358,
-      "loss": 46.0,
-      "step": 19907
-    },
-    {
-      "epoch": 3.2060469423084665,
-      "grad_norm": 0.0018927935743704438,
-      "learning_rate": 0.0001999949331493683,
-      "loss": 46.0,
-      "step": 19908
-    },
-    {
-      "epoch": 3.206207979387254,
-      "grad_norm": 0.0038575297221541405,
-      "learning_rate": 0.00019999493264007736,
-      "loss": 46.0,
-      "step": 19909
-    },
-    {
-      "epoch": 3.2063690164660414,
-      "grad_norm": 0.003207950619980693,
-      "learning_rate": 0.00019999493213076087,
-      "loss": 46.0,
-      "step": 19910
-    },
-    {
-      "epoch": 3.206530053544829,
-      "grad_norm": 0.0022205302957445383,
-      "learning_rate": 0.00019999493162141879,
-      "loss": 46.0,
-      "step": 19911
-    },
-    {
-      "epoch": 3.206691090623616,
-      "grad_norm": 0.0017520474502816796,
-      "learning_rate": 0.00019999493111205112,
-      "loss": 46.0,
-      "step": 19912
-    },
-    {
-      "epoch": 3.2068521277024034,
-      "grad_norm": 0.0014618110144510865,
-      "learning_rate": 0.00019999493060265784,
-      "loss": 46.0,
-      "step": 19913
-    },
-    {
-      "epoch": 3.207013164781191,
-      "grad_norm": 0.00253496365621686,
-      "learning_rate": 0.00019999493009323897,
-      "loss": 46.0,
-      "step": 19914
-    },
-    {
-      "epoch": 3.2071742018599783,
-      "grad_norm": 0.0010251352796331048,
-      "learning_rate": 0.0001999949295837945,
-      "loss": 46.0,
-      "step": 19915
-    },
-    {
-      "epoch": 3.2073352389387657,
-      "grad_norm": 0.017363103106617928,
-      "learning_rate": 0.00019999492907432447,
-      "loss": 46.0,
-      "step": 19916
-    },
-    {
-      "epoch": 3.207496276017553,
-      "grad_norm": 0.006027654744684696,
-      "learning_rate": 0.0001999949285648288,
-      "loss": 46.0,
-      "step": 19917
-    },
-    {
-      "epoch": 3.2076573130963406,
-      "grad_norm": 0.005080610979348421,
-      "learning_rate": 0.00019999492805530754,
-      "loss": 46.0,
-      "step": 19918
-    },
-    {
-      "epoch": 3.2078183501751276,
-      "grad_norm": 0.008073265664279461,
-      "learning_rate": 0.00019999492754576074,
-      "loss": 46.0,
-      "step": 19919
-    },
-    {
-      "epoch": 3.207979387253915,
-      "grad_norm": 0.0013680735137313604,
-      "learning_rate": 0.0001999949270361883,
-      "loss": 46.0,
-      "step": 19920
-    },
-    {
-      "epoch": 3.2081404243327025,
-      "grad_norm": 0.00872149970382452,
-      "learning_rate": 0.00019999492652659026,
-      "loss": 46.0,
-      "step": 19921
-    },
-    {
-      "epoch": 3.20830146141149,
-      "grad_norm": 0.01572323590517044,
-      "learning_rate": 0.00019999492601696666,
-      "loss": 46.0,
-      "step": 19922
-    },
-    {
-      "epoch": 3.2084624984902774,
-      "grad_norm": 0.004353813827037811,
-      "learning_rate": 0.00019999492550731745,
-      "loss": 46.0,
-      "step": 19923
-    },
-    {
-      "epoch": 3.208623535569065,
-      "grad_norm": 0.0010871444828808308,
-      "learning_rate": 0.00019999492499764266,
-      "loss": 46.0,
-      "step": 19924
-    },
-    {
-      "epoch": 3.2087845726478523,
-      "grad_norm": 0.0056698499247431755,
-      "learning_rate": 0.00019999492448794225,
-      "loss": 46.0,
-      "step": 19925
-    },
-    {
-      "epoch": 3.20894560972664,
-      "grad_norm": 0.0003940163878723979,
-      "learning_rate": 0.00019999492397821625,
-      "loss": 46.0,
-      "step": 19926
-    },
-    {
-      "epoch": 3.209106646805427,
-      "grad_norm": 0.00767147121950984,
-      "learning_rate": 0.0001999949234684647,
-      "loss": 46.0,
-      "step": 19927
-    },
-    {
-      "epoch": 3.2092676838842142,
-      "grad_norm": 0.0029793891590088606,
-      "learning_rate": 0.0001999949229586875,
-      "loss": 46.0,
-      "step": 19928
-    },
-    {
-      "epoch": 3.2094287209630017,
-      "grad_norm": 0.002021720865741372,
-      "learning_rate": 0.00019999492244888472,
-      "loss": 46.0,
-      "step": 19929
-    },
-    {
-      "epoch": 3.209589758041789,
-      "grad_norm": 0.005881303455680609,
-      "learning_rate": 0.00019999492193905634,
-      "loss": 46.0,
-      "step": 19930
-    },
-    {
-      "epoch": 3.2097507951205766,
-      "grad_norm": 0.0014934591017663479,
-      "learning_rate": 0.0001999949214292024,
-      "loss": 46.0,
-      "step": 19931
-    },
-    {
-      "epoch": 3.209911832199364,
-      "grad_norm": 0.004109697882086039,
-      "learning_rate": 0.00019999492091932287,
-      "loss": 46.0,
-      "step": 19932
-    },
-    {
-      "epoch": 3.2100728692781515,
-      "grad_norm": 0.0014857043279334903,
-      "learning_rate": 0.00019999492040941773,
-      "loss": 46.0,
-      "step": 19933
-    },
-    {
-      "epoch": 3.2102339063569385,
-      "grad_norm": 0.0015882095322012901,
-      "learning_rate": 0.00019999491989948696,
-      "loss": 46.0,
-      "step": 19934
-    },
-    {
-      "epoch": 3.210394943435726,
-      "grad_norm": 0.01587427780032158,
-      "learning_rate": 0.00019999491938953065,
-      "loss": 46.0,
-      "step": 19935
-    },
-    {
-      "epoch": 3.2105559805145134,
-      "grad_norm": 0.0029281058814376593,
-      "learning_rate": 0.0001999949188795487,
-      "loss": 46.0,
-      "step": 19936
-    },
-    {
-      "epoch": 3.210717017593301,
-      "grad_norm": 0.001928345998749137,
-      "learning_rate": 0.0001999949183695412,
-      "loss": 46.0,
-      "step": 19937
-    },
-    {
-      "epoch": 3.2108780546720883,
-      "grad_norm": 0.012835285626351833,
-      "learning_rate": 0.00019999491785950807,
-      "loss": 46.0,
-      "step": 19938
-    },
-    {
-      "epoch": 3.2110390917508758,
-      "grad_norm": 0.0009133763378486037,
-      "learning_rate": 0.00019999491734944938,
-      "loss": 46.0,
-      "step": 19939
-    },
-    {
-      "epoch": 3.211200128829663,
-      "grad_norm": 0.0032668181229382753,
-      "learning_rate": 0.00019999491683936506,
-      "loss": 46.0,
-      "step": 19940
-    },
-    {
-      "epoch": 3.21136116590845,
-      "grad_norm": 0.003891577711328864,
-      "learning_rate": 0.00019999491632925517,
-      "loss": 46.0,
-      "step": 19941
-    },
-    {
-      "epoch": 3.2115222029872377,
-      "grad_norm": 0.0018773623742163181,
-      "learning_rate": 0.0001999949158191197,
-      "loss": 46.0,
-      "step": 19942
-    },
-    {
-      "epoch": 3.211683240066025,
-      "grad_norm": 0.0008994661038741469,
-      "learning_rate": 0.0001999949153089586,
-      "loss": 46.0,
-      "step": 19943
-    },
-    {
-      "epoch": 3.2118442771448126,
-      "grad_norm": 0.0037934351712465286,
-      "learning_rate": 0.00019999491479877194,
-      "loss": 46.0,
-      "step": 19944
-    },
-    {
-      "epoch": 3.2120053142236,
-      "grad_norm": 0.000621344312094152,
-      "learning_rate": 0.00019999491428855968,
-      "loss": 46.0,
-      "step": 19945
-    },
-    {
-      "epoch": 3.2121663513023875,
-      "grad_norm": 0.0006715527852065861,
-      "learning_rate": 0.00019999491377832183,
-      "loss": 46.0,
-      "step": 19946
-    },
-    {
-      "epoch": 3.212327388381175,
-      "grad_norm": 0.007685655262321234,
-      "learning_rate": 0.00019999491326805834,
-      "loss": 46.0,
-      "step": 19947
-    },
-    {
-      "epoch": 3.212488425459962,
-      "grad_norm": 0.003946251701563597,
-      "learning_rate": 0.0001999949127577693,
-      "loss": 46.0,
-      "step": 19948
-    },
-    {
-      "epoch": 3.2126494625387494,
-      "grad_norm": 0.004356798250228167,
-      "learning_rate": 0.00019999491224745466,
-      "loss": 46.0,
-      "step": 19949
-    },
-    {
-      "epoch": 3.212810499617537,
-      "grad_norm": 0.003964099567383528,
-      "learning_rate": 0.00019999491173711443,
-      "loss": 46.0,
-      "step": 19950
-    },
-    {
-      "epoch": 3.2129715366963243,
-      "grad_norm": 0.0004672715731430799,
-      "learning_rate": 0.0001999949112267486,
-      "loss": 46.0,
-      "step": 19951
-    },
-    {
-      "epoch": 3.2131325737751117,
-      "grad_norm": 0.0037271385081112385,
-      "learning_rate": 0.0001999949107163572,
-      "loss": 46.0,
-      "step": 19952
-    },
-    {
-      "epoch": 3.213293610853899,
-      "grad_norm": 0.001955863321200013,
-      "learning_rate": 0.00019999491020594016,
-      "loss": 46.0,
-      "step": 19953
-    },
-    {
-      "epoch": 3.2134546479326866,
-      "grad_norm": 0.004761870950460434,
-      "learning_rate": 0.00019999490969549756,
-      "loss": 46.0,
-      "step": 19954
-    },
-    {
-      "epoch": 3.213615685011474,
-      "grad_norm": 0.005952753592282534,
-      "learning_rate": 0.00019999490918502934,
-      "loss": 46.0,
-      "step": 19955
-    },
-    {
-      "epoch": 3.213776722090261,
-      "grad_norm": 0.00965246744453907,
-      "learning_rate": 0.00019999490867453554,
-      "loss": 46.0,
-      "step": 19956
-    },
-    {
-      "epoch": 3.2139377591690486,
-      "grad_norm": 0.007838096469640732,
-      "learning_rate": 0.00019999490816401615,
-      "loss": 46.0,
-      "step": 19957
-    },
-    {
-      "epoch": 3.214098796247836,
-      "grad_norm": 0.0017091804184019566,
-      "learning_rate": 0.00019999490765347118,
-      "loss": 46.0,
-      "step": 19958
-    },
-    {
-      "epoch": 3.2142598333266235,
-      "grad_norm": 0.0024706751573830843,
-      "learning_rate": 0.0001999949071429006,
-      "loss": 46.0,
-      "step": 19959
-    },
-    {
-      "epoch": 3.214420870405411,
-      "grad_norm": 0.003551687579602003,
-      "learning_rate": 0.0001999949066323044,
-      "loss": 46.0,
-      "step": 19960
-    },
-    {
-      "epoch": 3.2145819074841984,
-      "grad_norm": 0.00403203209862113,
-      "learning_rate": 0.00019999490612168265,
-      "loss": 46.0,
-      "step": 19961
-    },
-    {
-      "epoch": 3.214742944562986,
-      "grad_norm": 0.0007498914492316544,
-      "learning_rate": 0.0001999949056110353,
-      "loss": 46.0,
-      "step": 19962
-    },
-    {
-      "epoch": 3.214903981641773,
-      "grad_norm": 0.0055478839203715324,
-      "learning_rate": 0.00019999490510036236,
-      "loss": 46.0,
-      "step": 19963
-    },
-    {
-      "epoch": 3.2150650187205603,
-      "grad_norm": 0.0013714124215766788,
-      "learning_rate": 0.0001999949045896638,
-      "loss": 46.0,
-      "step": 19964
-    },
-    {
-      "epoch": 3.2152260557993477,
-      "grad_norm": 0.002369512338191271,
-      "learning_rate": 0.00019999490407893967,
-      "loss": 46.0,
-      "step": 19965
-    },
-    {
-      "epoch": 3.215387092878135,
-      "grad_norm": 0.005854782182723284,
-      "learning_rate": 0.00019999490356818994,
-      "loss": 46.0,
-      "step": 19966
-    },
-    {
-      "epoch": 3.2155481299569226,
-      "grad_norm": 0.002678430173546076,
-      "learning_rate": 0.0001999949030574146,
-      "loss": 46.0,
-      "step": 19967
-    },
-    {
-      "epoch": 3.21570916703571,
-      "grad_norm": 0.0008392783929593861,
-      "learning_rate": 0.0001999949025466137,
-      "loss": 46.0,
-      "step": 19968
-    },
-    {
-      "epoch": 3.2158702041144975,
-      "grad_norm": 0.015743868425488472,
-      "learning_rate": 0.00019999490203578718,
-      "loss": 46.0,
-      "step": 19969
-    },
-    {
-      "epoch": 3.216031241193285,
-      "grad_norm": 0.001104449387639761,
-      "learning_rate": 0.00019999490152493508,
-      "loss": 46.0,
-      "step": 19970
-    },
-    {
-      "epoch": 3.216192278272072,
-      "grad_norm": 0.0076009598560631275,
-      "learning_rate": 0.0001999949010140574,
-      "loss": 46.0,
-      "step": 19971
-    },
-    {
-      "epoch": 3.2163533153508594,
-      "grad_norm": 0.003827478038147092,
-      "learning_rate": 0.00019999490050315406,
-      "loss": 46.0,
-      "step": 19972
-    },
-    {
-      "epoch": 3.216514352429647,
-      "grad_norm": 0.004382527898997068,
-      "learning_rate": 0.0001999948999922252,
-      "loss": 46.0,
-      "step": 19973
-    },
-    {
-      "epoch": 3.2166753895084343,
-      "grad_norm": 0.004581006243824959,
-      "learning_rate": 0.0001999948994812707,
-      "loss": 46.0,
-      "step": 19974
-    },
-    {
-      "epoch": 3.216836426587222,
-      "grad_norm": 0.002011304022744298,
-      "learning_rate": 0.00019999489897029062,
-      "loss": 46.0,
-      "step": 19975
-    },
-    {
-      "epoch": 3.2169974636660092,
-      "grad_norm": 0.005309367086738348,
-      "learning_rate": 0.00019999489845928494,
-      "loss": 46.0,
-      "step": 19976
-    },
-    {
-      "epoch": 3.2171585007447963,
-      "grad_norm": 0.0010016907472163439,
-      "learning_rate": 0.00019999489794825367,
-      "loss": 46.0,
-      "step": 19977
-    },
-    {
-      "epoch": 3.2173195378235837,
-      "grad_norm": 0.0038207981269806623,
-      "learning_rate": 0.00019999489743719685,
-      "loss": 46.0,
-      "step": 19978
-    },
-    {
-      "epoch": 3.217480574902371,
-      "grad_norm": 0.0033337429631501436,
-      "learning_rate": 0.0001999948969261144,
-      "loss": 46.0,
-      "step": 19979
-    },
-    {
-      "epoch": 3.2176416119811586,
-      "grad_norm": 0.011218244209885597,
-      "learning_rate": 0.00019999489641500635,
-      "loss": 46.0,
-      "step": 19980
-    },
-    {
-      "epoch": 3.217802649059946,
-      "grad_norm": 0.009353700093925,
-      "learning_rate": 0.0001999948959038727,
-      "loss": 46.0,
-      "step": 19981
-    },
-    {
-      "epoch": 3.2179636861387335,
-      "grad_norm": 0.005936374422162771,
-      "learning_rate": 0.00019999489539271347,
-      "loss": 46.0,
-      "step": 19982
-    },
-    {
-      "epoch": 3.218124723217521,
-      "grad_norm": 0.008977899327874184,
-      "learning_rate": 0.00019999489488152868,
-      "loss": 46.0,
-      "step": 19983
-    },
-    {
-      "epoch": 3.2182857602963084,
-      "grad_norm": 0.007583903148770332,
-      "learning_rate": 0.00019999489437031825,
-      "loss": 46.0,
-      "step": 19984
-    },
-    {
-      "epoch": 3.2184467973750954,
-      "grad_norm": 0.006460958626121283,
-      "learning_rate": 0.00019999489385908223,
-      "loss": 46.0,
-      "step": 19985
-    },
-    {
-      "epoch": 3.218607834453883,
-      "grad_norm": 0.00982884131371975,
-      "learning_rate": 0.00019999489334782063,
-      "loss": 46.0,
-      "step": 19986
-    },
-    {
-      "epoch": 3.2187688715326703,
-      "grad_norm": 0.0028576948679983616,
-      "learning_rate": 0.00019999489283653344,
-      "loss": 46.0,
-      "step": 19987
-    },
-    {
-      "epoch": 3.218929908611458,
-      "grad_norm": 0.005682682618498802,
-      "learning_rate": 0.00019999489232522063,
-      "loss": 46.0,
-      "step": 19988
-    },
-    {
-      "epoch": 3.2190909456902452,
-      "grad_norm": 0.005526925902813673,
-      "learning_rate": 0.00019999489181388226,
-      "loss": 46.0,
-      "step": 19989
-    },
-    {
-      "epoch": 3.2192519827690327,
-      "grad_norm": 0.0011081219417974353,
-      "learning_rate": 0.00019999489130251828,
-      "loss": 46.0,
-      "step": 19990
-    },
-    {
-      "epoch": 3.21941301984782,
-      "grad_norm": 0.002907171845436096,
-      "learning_rate": 0.0001999948907911287,
-      "loss": 46.0,
-      "step": 19991
-    },
-    {
-      "epoch": 3.219574056926607,
-      "grad_norm": 0.0026894756592810154,
-      "learning_rate": 0.00019999489027971353,
-      "loss": 46.0,
-      "step": 19992
-    },
-    {
-      "epoch": 3.2197350940053946,
-      "grad_norm": 0.002433567075058818,
-      "learning_rate": 0.00019999488976827276,
-      "loss": 46.0,
-      "step": 19993
-    },
-    {
-      "epoch": 3.219896131084182,
-      "grad_norm": 0.0030539012514054775,
-      "learning_rate": 0.00019999488925680643,
-      "loss": 46.0,
-      "step": 19994
-    },
-    {
-      "epoch": 3.2200571681629695,
-      "grad_norm": 0.0007772623794153333,
-      "learning_rate": 0.00019999488874531448,
-      "loss": 46.0,
-      "step": 19995
-    },
-    {
-      "epoch": 3.220218205241757,
-      "grad_norm": 0.0018702943343669176,
-      "learning_rate": 0.00019999488823379692,
-      "loss": 46.0,
-      "step": 19996
-    },
-    {
-      "epoch": 3.2203792423205444,
-      "grad_norm": 0.001668772310949862,
-      "learning_rate": 0.0001999948877222538,
-      "loss": 46.0,
-      "step": 19997
-    },
-    {
-      "epoch": 3.220540279399332,
-      "grad_norm": 0.0018319544615224004,
-      "learning_rate": 0.00019999488721068507,
-      "loss": 46.0,
-      "step": 19998
-    },
-    {
-      "epoch": 3.2207013164781193,
-      "grad_norm": 0.002652161056175828,
-      "learning_rate": 0.00019999488669909075,
-      "loss": 46.0,
-      "step": 19999
-    },
-    {
-      "epoch": 3.2208623535569063,
-      "grad_norm": 0.0018790222238749266,
-      "learning_rate": 0.00019999488618747084,
-      "loss": 46.0,
-      "step": 20000
-    },
-    {
-      "epoch": 3.2210233906356938,
-      "grad_norm": 0.009816550649702549,
-      "learning_rate": 0.00019999488567582535,
-      "loss": 46.0,
-      "step": 20001
-    },
-    {
-      "epoch": 3.221184427714481,
-      "grad_norm": 0.0014191671507433057,
-      "learning_rate": 0.00019999488516415424,
-      "loss": 46.0,
-      "step": 20002
-    },
-    {
-      "epoch": 3.2213454647932687,
-      "grad_norm": 0.002027905313298106,
-      "learning_rate": 0.00019999488465245751,
-      "loss": 46.0,
-      "step": 20003
-    },
-    {
-      "epoch": 3.221506501872056,
-      "grad_norm": 0.006543018389493227,
-      "learning_rate": 0.00019999488414073526,
-      "loss": 46.0,
-      "step": 20004
-    },
-    {
-      "epoch": 3.2216675389508436,
-      "grad_norm": 0.0009746000287123024,
-      "learning_rate": 0.00019999488362898736,
-      "loss": 46.0,
-      "step": 20005
-    },
-    {
-      "epoch": 3.221828576029631,
-      "grad_norm": 0.0026498446241021156,
-      "learning_rate": 0.0001999948831172139,
-      "loss": 46.0,
-      "step": 20006
-    },
-    {
-      "epoch": 3.221989613108418,
-      "grad_norm": 0.0013738729758188128,
-      "learning_rate": 0.0001999948826054148,
-      "loss": 46.0,
-      "step": 20007
-    },
-    {
-      "epoch": 3.2221506501872055,
-      "grad_norm": 0.004459552001208067,
-      "learning_rate": 0.00019999488209359017,
-      "loss": 46.0,
-      "step": 20008
-    },
-    {
-      "epoch": 3.222311687265993,
-      "grad_norm": 0.0033183731138706207,
-      "learning_rate": 0.00019999488158173992,
-      "loss": 46.0,
-      "step": 20009
-    },
-    {
-      "epoch": 3.2224727243447804,
-      "grad_norm": 0.002151679014787078,
-      "learning_rate": 0.00019999488106986406,
-      "loss": 46.0,
-      "step": 20010
-    },
-    {
-      "epoch": 3.222633761423568,
-      "grad_norm": 0.0021975645795464516,
-      "learning_rate": 0.00019999488055796262,
-      "loss": 46.0,
-      "step": 20011
-    },
-    {
-      "epoch": 3.2227947985023553,
-      "grad_norm": 0.004420607816427946,
-      "learning_rate": 0.00019999488004603558,
-      "loss": 46.0,
-      "step": 20012
-    },
-    {
-      "epoch": 3.2229558355811427,
-      "grad_norm": 0.006460062228143215,
-      "learning_rate": 0.00019999487953408296,
-      "loss": 46.0,
-      "step": 20013
-    },
-    {
-      "epoch": 3.2231168726599297,
-      "grad_norm": 0.00763485673815012,
-      "learning_rate": 0.00019999487902210472,
-      "loss": 46.0,
-      "step": 20014
-    },
-    {
-      "epoch": 3.223277909738717,
-      "grad_norm": 0.003281361423432827,
-      "learning_rate": 0.0001999948785101009,
-      "loss": 46.0,
-      "step": 20015
-    },
-    {
-      "epoch": 3.2234389468175046,
-      "grad_norm": 0.0022874479182064533,
-      "learning_rate": 0.00019999487799807151,
-      "loss": 46.0,
-      "step": 20016
-    },
-    {
-      "epoch": 3.223599983896292,
-      "grad_norm": 0.0014630361692979932,
-      "learning_rate": 0.0001999948774860165,
-      "loss": 46.0,
-      "step": 20017
-    },
-    {
-      "epoch": 3.2237610209750795,
-      "grad_norm": 0.009154632687568665,
-      "learning_rate": 0.00019999487697393588,
-      "loss": 46.0,
-      "step": 20018
-    },
-    {
-      "epoch": 3.223922058053867,
-      "grad_norm": 0.0021597500890493393,
-      "learning_rate": 0.0001999948764618297,
-      "loss": 46.0,
-      "step": 20019
-    },
-    {
-      "epoch": 3.2240830951326545,
-      "grad_norm": 0.005203419830650091,
-      "learning_rate": 0.0001999948759496979,
-      "loss": 46.0,
-      "step": 20020
-    },
-    {
-      "epoch": 3.2242441322114415,
-      "grad_norm": 0.0008442361722700298,
-      "learning_rate": 0.00019999487543754054,
-      "loss": 46.0,
-      "step": 20021
-    },
-    {
-      "epoch": 3.224405169290229,
-      "grad_norm": 0.000546953349839896,
-      "learning_rate": 0.00019999487492535756,
-      "loss": 46.0,
-      "step": 20022
-    },
-    {
-      "epoch": 3.2245662063690164,
-      "grad_norm": 0.0030808565206825733,
-      "learning_rate": 0.000199994874413149,
-      "loss": 46.0,
-      "step": 20023
-    },
-    {
-      "epoch": 3.224727243447804,
-      "grad_norm": 0.00448093144223094,
-      "learning_rate": 0.00019999487390091485,
-      "loss": 46.0,
-      "step": 20024
-    },
-    {
-      "epoch": 3.2248882805265913,
-      "grad_norm": 0.002609221264719963,
-      "learning_rate": 0.0001999948733886551,
-      "loss": 46.0,
-      "step": 20025
-    },
-    {
-      "epoch": 3.2250493176053787,
-      "grad_norm": 0.0006908430950716138,
-      "learning_rate": 0.00019999487287636973,
-      "loss": 46.0,
-      "step": 20026
-    },
-    {
-      "epoch": 3.225210354684166,
-      "grad_norm": 0.0007829205715097487,
-      "learning_rate": 0.0001999948723640588,
-      "loss": 46.0,
-      "step": 20027
-    },
-    {
-      "epoch": 3.2253713917629536,
-      "grad_norm": 0.004746154882013798,
-      "learning_rate": 0.00019999487185172224,
-      "loss": 46.0,
-      "step": 20028
-    },
-    {
-      "epoch": 3.2255324288417406,
-      "grad_norm": 0.0032963696867227554,
-      "learning_rate": 0.00019999487133936014,
-      "loss": 46.0,
-      "step": 20029
-    },
-    {
-      "epoch": 3.225693465920528,
-      "grad_norm": 0.0027679477352648973,
-      "learning_rate": 0.0001999948708269724,
-      "loss": 46.0,
-      "step": 20030
-    },
-    {
-      "epoch": 3.2258545029993155,
-      "grad_norm": 0.003499365644529462,
-      "learning_rate": 0.0001999948703145591,
-      "loss": 46.0,
-      "step": 20031
-    },
-    {
-      "epoch": 3.226015540078103,
-      "grad_norm": 0.0031455690041184425,
-      "learning_rate": 0.00019999486980212022,
-      "loss": 46.0,
-      "step": 20032
-    },
-    {
-      "epoch": 3.2261765771568904,
-      "grad_norm": 0.006072757765650749,
-      "learning_rate": 0.0001999948692896557,
-      "loss": 46.0,
-      "step": 20033
-    },
-    {
-      "epoch": 3.226337614235678,
-      "grad_norm": 0.002154717454686761,
-      "learning_rate": 0.0001999948687771656,
-      "loss": 46.0,
-      "step": 20034
-    },
-    {
-      "epoch": 3.2264986513144653,
-      "grad_norm": 0.0023426564875990152,
-      "learning_rate": 0.00019999486826464993,
-      "loss": 46.0,
-      "step": 20035
-    },
-    {
-      "epoch": 3.2266596883932523,
-      "grad_norm": 0.003374747931957245,
-      "learning_rate": 0.00019999486775210864,
-      "loss": 46.0,
-      "step": 20036
-    },
-    {
-      "epoch": 3.22682072547204,
-      "grad_norm": 0.0027175594586879015,
-      "learning_rate": 0.00019999486723954176,
-      "loss": 46.0,
-      "step": 20037
-    },
-    {
-      "epoch": 3.2269817625508272,
-      "grad_norm": 0.001925921649672091,
-      "learning_rate": 0.0001999948667269493,
-      "loss": 46.0,
-      "step": 20038
-    },
-    {
-      "epoch": 3.2271427996296147,
-      "grad_norm": 0.004148443695157766,
-      "learning_rate": 0.00019999486621433122,
-      "loss": 46.0,
-      "step": 20039
-    },
-    {
-      "epoch": 3.227303836708402,
-      "grad_norm": 0.0011738608591258526,
-      "learning_rate": 0.00019999486570168758,
-      "loss": 46.0,
-      "step": 20040
-    },
-    {
-      "epoch": 3.2274648737871896,
-      "grad_norm": 0.0023432662710547447,
-      "learning_rate": 0.0001999948651890183,
-      "loss": 46.0,
-      "step": 20041
-    },
-    {
-      "epoch": 3.227625910865977,
-      "grad_norm": 0.0011990058701485395,
-      "learning_rate": 0.00019999486467632349,
-      "loss": 46.0,
-      "step": 20042
-    },
-    {
-      "epoch": 3.2277869479447645,
-      "grad_norm": 0.0035896487534046173,
-      "learning_rate": 0.00019999486416360306,
-      "loss": 46.0,
-      "step": 20043
-    },
-    {
-      "epoch": 3.2279479850235515,
-      "grad_norm": 0.0007968611898832023,
-      "learning_rate": 0.00019999486365085702,
-      "loss": 46.0,
-      "step": 20044
-    },
-    {
-      "epoch": 3.228109022102339,
-      "grad_norm": 0.006229350343346596,
-      "learning_rate": 0.0001999948631380854,
-      "loss": 46.0,
-      "step": 20045
-    },
-    {
-      "epoch": 3.2282700591811264,
-      "grad_norm": 0.0033001210540533066,
-      "learning_rate": 0.00019999486262528817,
-      "loss": 46.0,
-      "step": 20046
-    },
-    {
-      "epoch": 3.228431096259914,
-      "grad_norm": 0.002701087621971965,
-      "learning_rate": 0.00019999486211246534,
-      "loss": 46.0,
-      "step": 20047
-    },
-    {
-      "epoch": 3.2285921333387013,
-      "grad_norm": 0.0010901725618168712,
-      "learning_rate": 0.00019999486159961695,
-      "loss": 46.0,
-      "step": 20048
-    },
-    {
-      "epoch": 3.2287531704174888,
-      "grad_norm": 0.0037155996542423964,
-      "learning_rate": 0.00019999486108674295,
-      "loss": 46.0,
-      "step": 20049
-    },
-    {
-      "epoch": 3.2289142074962762,
-      "grad_norm": 0.0026516560465097427,
-      "learning_rate": 0.00019999486057384336,
-      "loss": 46.0,
-      "step": 20050
-    },
-    {
-      "epoch": 3.2290752445750632,
-      "grad_norm": 0.0006564582581631839,
-      "learning_rate": 0.00019999486006091818,
-      "loss": 46.0,
-      "step": 20051
-    },
-    {
-      "epoch": 3.2292362816538507,
-      "grad_norm": 0.0029622321017086506,
-      "learning_rate": 0.00019999485954796739,
-      "loss": 46.0,
-      "step": 20052
-    },
-    {
-      "epoch": 3.229397318732638,
-      "grad_norm": 0.005908967927098274,
-      "learning_rate": 0.00019999485903499103,
-      "loss": 46.0,
-      "step": 20053
-    },
-    {
-      "epoch": 3.2295583558114256,
-      "grad_norm": 0.00589406955987215,
-      "learning_rate": 0.00019999485852198906,
-      "loss": 46.0,
-      "step": 20054
-    },
-    {
-      "epoch": 3.229719392890213,
-      "grad_norm": 0.0016828267835080624,
-      "learning_rate": 0.00019999485800896148,
-      "loss": 46.0,
-      "step": 20055
-    },
-    {
-      "epoch": 3.2298804299690005,
-      "grad_norm": 0.0031218871008604765,
-      "learning_rate": 0.00019999485749590834,
-      "loss": 46.0,
-      "step": 20056
-    },
-    {
-      "epoch": 3.230041467047788,
-      "grad_norm": 0.0008189807995222509,
-      "learning_rate": 0.0001999948569828296,
-      "loss": 46.0,
-      "step": 20057
-    },
-    {
-      "epoch": 3.230202504126575,
-      "grad_norm": 0.0042694020085036755,
-      "learning_rate": 0.00019999485646972524,
-      "loss": 46.0,
-      "step": 20058
-    },
-    {
-      "epoch": 3.2303635412053624,
-      "grad_norm": 0.004502309486269951,
-      "learning_rate": 0.00019999485595659534,
-      "loss": 46.0,
-      "step": 20059
-    },
-    {
-      "epoch": 3.23052457828415,
-      "grad_norm": 0.0020823825616389513,
-      "learning_rate": 0.0001999948554434398,
-      "loss": 46.0,
-      "step": 20060
-    },
-    {
-      "epoch": 3.2306856153629373,
-      "grad_norm": 0.012499809265136719,
-      "learning_rate": 0.0001999948549302587,
-      "loss": 46.0,
-      "step": 20061
-    },
-    {
-      "epoch": 3.2308466524417248,
-      "grad_norm": 0.00125453423243016,
-      "learning_rate": 0.00019999485441705197,
-      "loss": 46.0,
-      "step": 20062
-    },
-    {
-      "epoch": 3.231007689520512,
-      "grad_norm": 0.0053497618064284325,
-      "learning_rate": 0.00019999485390381967,
-      "loss": 46.0,
-      "step": 20063
-    },
-    {
-      "epoch": 3.2311687265992997,
-      "grad_norm": 0.001731640542857349,
-      "learning_rate": 0.00019999485339056175,
-      "loss": 46.0,
-      "step": 20064
-    },
-    {
-      "epoch": 3.2313297636780867,
-      "grad_norm": 0.01032224204391241,
-      "learning_rate": 0.00019999485287727826,
-      "loss": 46.0,
-      "step": 20065
-    },
-    {
-      "epoch": 3.231490800756874,
-      "grad_norm": 0.0032352639827877283,
-      "learning_rate": 0.00019999485236396914,
-      "loss": 46.0,
-      "step": 20066
-    },
-    {
-      "epoch": 3.2316518378356616,
-      "grad_norm": 0.0018381490372121334,
-      "learning_rate": 0.0001999948518506345,
-      "loss": 46.0,
-      "step": 20067
-    },
-    {
-      "epoch": 3.231812874914449,
-      "grad_norm": 0.003487006528303027,
-      "learning_rate": 0.0001999948513372742,
-      "loss": 46.0,
-      "step": 20068
-    },
-    {
-      "epoch": 3.2319739119932365,
-      "grad_norm": 0.007057672366499901,
-      "learning_rate": 0.00019999485082388834,
-      "loss": 46.0,
-      "step": 20069
-    },
-    {
-      "epoch": 3.232134949072024,
-      "grad_norm": 0.007005220744758844,
-      "learning_rate": 0.0001999948503104769,
-      "loss": 46.0,
-      "step": 20070
-    },
-    {
-      "epoch": 3.2322959861508114,
-      "grad_norm": 0.0008962152060121298,
-      "learning_rate": 0.0001999948497970398,
-      "loss": 46.0,
-      "step": 20071
-    },
-    {
-      "epoch": 3.232457023229599,
-      "grad_norm": 0.007168622687458992,
-      "learning_rate": 0.00019999484928357716,
-      "loss": 46.0,
-      "step": 20072
-    },
-    {
-      "epoch": 3.232618060308386,
-      "grad_norm": 0.0006051333621144295,
-      "learning_rate": 0.00019999484877008893,
-      "loss": 46.0,
-      "step": 20073
-    },
-    {
-      "epoch": 3.2327790973871733,
-      "grad_norm": 0.010350801050662994,
-      "learning_rate": 0.00019999484825657508,
-      "loss": 46.0,
-      "step": 20074
-    },
-    {
-      "epoch": 3.2329401344659607,
-      "grad_norm": 0.0038665507454425097,
-      "learning_rate": 0.00019999484774303565,
-      "loss": 46.0,
-      "step": 20075
-    },
-    {
-      "epoch": 3.233101171544748,
-      "grad_norm": 0.0037180643994361162,
-      "learning_rate": 0.00019999484722947063,
-      "loss": 46.0,
-      "step": 20076
-    },
-    {
-      "epoch": 3.2332622086235356,
-      "grad_norm": 0.0025976565666496754,
-      "learning_rate": 0.00019999484671588002,
-      "loss": 46.0,
-      "step": 20077
-    },
-    {
-      "epoch": 3.233423245702323,
-      "grad_norm": 0.006441825069487095,
-      "learning_rate": 0.0001999948462022638,
-      "loss": 46.0,
-      "step": 20078
-    },
-    {
-      "epoch": 3.2335842827811105,
-      "grad_norm": 0.005928924772888422,
-      "learning_rate": 0.000199994845688622,
-      "loss": 46.0,
-      "step": 20079
-    },
-    {
-      "epoch": 3.2337453198598975,
-      "grad_norm": 0.012846838682889938,
-      "learning_rate": 0.00019999484517495462,
-      "loss": 46.0,
-      "step": 20080
-    },
-    {
-      "epoch": 3.233906356938685,
-      "grad_norm": 0.0015549047384411097,
-      "learning_rate": 0.0001999948446612616,
-      "loss": 46.0,
-      "step": 20081
-    },
-    {
-      "epoch": 3.2340673940174725,
-      "grad_norm": 0.008976079523563385,
-      "learning_rate": 0.00019999484414754304,
-      "loss": 46.0,
-      "step": 20082
-    },
-    {
-      "epoch": 3.23422843109626,
-      "grad_norm": 0.0017978926189243793,
-      "learning_rate": 0.00019999484363379883,
-      "loss": 46.0,
-      "step": 20083
-    },
-    {
-      "epoch": 3.2343894681750474,
-      "grad_norm": 0.007539292331784964,
-      "learning_rate": 0.00019999484312002908,
-      "loss": 46.0,
-      "step": 20084
-    },
-    {
-      "epoch": 3.234550505253835,
-      "grad_norm": 0.0007488857954740524,
-      "learning_rate": 0.00019999484260623372,
-      "loss": 46.0,
-      "step": 20085
-    },
-    {
-      "epoch": 3.2347115423326223,
-      "grad_norm": 0.00465942919254303,
-      "learning_rate": 0.00019999484209241275,
-      "loss": 46.0,
-      "step": 20086
-    },
-    {
-      "epoch": 3.2348725794114097,
-      "grad_norm": 0.005457698367536068,
-      "learning_rate": 0.0001999948415785662,
-      "loss": 46.0,
-      "step": 20087
-    },
-    {
-      "epoch": 3.2350336164901967,
-      "grad_norm": 0.004229591228067875,
-      "learning_rate": 0.00019999484106469404,
-      "loss": 46.0,
-      "step": 20088
-    },
-    {
-      "epoch": 3.235194653568984,
-      "grad_norm": 0.011333261616528034,
-      "learning_rate": 0.0001999948405507963,
-      "loss": 46.0,
-      "step": 20089
-    },
-    {
-      "epoch": 3.2353556906477716,
-      "grad_norm": 0.0010983553947880864,
-      "learning_rate": 0.00019999484003687296,
-      "loss": 46.0,
-      "step": 20090
-    },
-    {
-      "epoch": 3.235516727726559,
-      "grad_norm": 0.0011428326833993196,
-      "learning_rate": 0.00019999483952292405,
-      "loss": 46.0,
-      "step": 20091
-    },
-    {
-      "epoch": 3.2356777648053465,
-      "grad_norm": 0.0009357786038890481,
-      "learning_rate": 0.00019999483900894952,
-      "loss": 46.0,
-      "step": 20092
-    },
-    {
-      "epoch": 3.235838801884134,
-      "grad_norm": 0.007837435230612755,
-      "learning_rate": 0.0001999948384949494,
-      "loss": 46.0,
-      "step": 20093
-    },
-    {
-      "epoch": 3.235999838962921,
-      "grad_norm": 0.0028274888172745705,
-      "learning_rate": 0.00019999483798092372,
-      "loss": 46.0,
-      "step": 20094
-    },
-    {
-      "epoch": 3.2361608760417084,
-      "grad_norm": 0.004880876280367374,
-      "learning_rate": 0.0001999948374668724,
-      "loss": 46.0,
-      "step": 20095
-    },
-    {
-      "epoch": 3.236321913120496,
-      "grad_norm": 0.0009277403587475419,
-      "learning_rate": 0.0001999948369527955,
-      "loss": 46.0,
-      "step": 20096
-    },
-    {
-      "epoch": 3.2364829501992833,
-      "grad_norm": 0.0020233565010130405,
-      "learning_rate": 0.00019999483643869304,
-      "loss": 46.0,
-      "step": 20097
-    },
-    {
-      "epoch": 3.236643987278071,
-      "grad_norm": 0.003471504896879196,
-      "learning_rate": 0.00019999483592456494,
-      "loss": 46.0,
-      "step": 20098
-    },
-    {
-      "epoch": 3.2368050243568582,
-      "grad_norm": 0.000386683561373502,
-      "learning_rate": 0.00019999483541041126,
-      "loss": 46.0,
-      "step": 20099
-    },
-    {
-      "epoch": 3.2369660614356457,
-      "grad_norm": 0.004984088707715273,
-      "learning_rate": 0.00019999483489623198,
-      "loss": 46.0,
-      "step": 20100
-    },
-    {
-      "epoch": 3.237127098514433,
-      "grad_norm": 0.0024535865522921085,
-      "learning_rate": 0.00019999483438202712,
-      "loss": 46.0,
-      "step": 20101
-    },
-    {
-      "epoch": 3.23728813559322,
-      "grad_norm": 0.0008911170880310237,
-      "learning_rate": 0.00019999483386779667,
-      "loss": 46.0,
-      "step": 20102
-    },
-    {
-      "epoch": 3.2374491726720076,
-      "grad_norm": 0.0036983625032007694,
-      "learning_rate": 0.00019999483335354063,
-      "loss": 46.0,
-      "step": 20103
-    },
-    {
-      "epoch": 3.237610209750795,
-      "grad_norm": 0.0018291639862582088,
-      "learning_rate": 0.000199994832839259,
-      "loss": 46.0,
-      "step": 20104
-    },
-    {
-      "epoch": 3.2377712468295825,
-      "grad_norm": 0.003597784787416458,
-      "learning_rate": 0.00019999483232495174,
-      "loss": 46.0,
-      "step": 20105
-    },
-    {
-      "epoch": 3.23793228390837,
-      "grad_norm": 0.0036381843965500593,
-      "learning_rate": 0.00019999483181061892,
-      "loss": 46.0,
-      "step": 20106
-    },
-    {
-      "epoch": 3.2380933209871574,
-      "grad_norm": 0.0008736033923923969,
-      "learning_rate": 0.00019999483129626048,
-      "loss": 46.0,
-      "step": 20107
-    },
-    {
-      "epoch": 3.238254358065945,
-      "grad_norm": 0.0012813173234462738,
-      "learning_rate": 0.00019999483078187648,
-      "loss": 46.0,
-      "step": 20108
-    },
-    {
-      "epoch": 3.238415395144732,
-      "grad_norm": 0.004531661048531532,
-      "learning_rate": 0.00019999483026746686,
-      "loss": 46.0,
-      "step": 20109
-    },
-    {
-      "epoch": 3.2385764322235193,
-      "grad_norm": 0.0021447124890983105,
-      "learning_rate": 0.00019999482975303166,
-      "loss": 46.0,
-      "step": 20110
-    },
-    {
-      "epoch": 3.2387374693023068,
-      "grad_norm": 0.010657808743417263,
-      "learning_rate": 0.00019999482923857085,
-      "loss": 46.0,
-      "step": 20111
-    },
-    {
-      "epoch": 3.2388985063810942,
-      "grad_norm": 0.0017723551718518138,
-      "learning_rate": 0.00019999482872408447,
-      "loss": 46.0,
-      "step": 20112
-    },
-    {
-      "epoch": 3.2390595434598817,
-      "grad_norm": 0.002096558688208461,
-      "learning_rate": 0.00019999482820957248,
-      "loss": 46.0,
-      "step": 20113
-    },
-    {
-      "epoch": 3.239220580538669,
-      "grad_norm": 0.0033797612413764,
-      "learning_rate": 0.0001999948276950349,
-      "loss": 46.0,
-      "step": 20114
-    },
-    {
-      "epoch": 3.2393816176174566,
-      "grad_norm": 0.0021784815471619368,
-      "learning_rate": 0.00019999482718047174,
-      "loss": 46.0,
-      "step": 20115
-    },
-    {
-      "epoch": 3.239542654696244,
-      "grad_norm": 0.004739631898701191,
-      "learning_rate": 0.00019999482666588296,
-      "loss": 46.0,
-      "step": 20116
-    },
-    {
-      "epoch": 3.239703691775031,
-      "grad_norm": 0.005034257657825947,
-      "learning_rate": 0.00019999482615126862,
-      "loss": 46.0,
-      "step": 20117
-    },
-    {
-      "epoch": 3.2398647288538185,
-      "grad_norm": 0.005472995340824127,
-      "learning_rate": 0.00019999482563662864,
-      "loss": 46.0,
-      "step": 20118
-    },
-    {
-      "epoch": 3.240025765932606,
-      "grad_norm": 0.001128678792156279,
-      "learning_rate": 0.0001999948251219631,
-      "loss": 46.0,
-      "step": 20119
-    },
-    {
-      "epoch": 3.2401868030113934,
-      "grad_norm": 0.0017618119018152356,
-      "learning_rate": 0.00019999482460727195,
-      "loss": 46.0,
-      "step": 20120
-    },
-    {
-      "epoch": 3.240347840090181,
-      "grad_norm": 0.0010891658021137118,
-      "learning_rate": 0.00019999482409255523,
-      "loss": 46.0,
-      "step": 20121
-    },
-    {
-      "epoch": 3.2405088771689683,
-      "grad_norm": 0.0016678066458553076,
-      "learning_rate": 0.0001999948235778129,
-      "loss": 46.0,
-      "step": 20122
-    },
-    {
-      "epoch": 3.2406699142477557,
-      "grad_norm": 0.006061645690351725,
-      "learning_rate": 0.000199994823063045,
-      "loss": 46.0,
-      "step": 20123
-    },
-    {
-      "epoch": 3.2408309513265428,
-      "grad_norm": 0.010113246738910675,
-      "learning_rate": 0.00019999482254825149,
-      "loss": 46.0,
-      "step": 20124
-    },
-    {
-      "epoch": 3.24099198840533,
-      "grad_norm": 0.003650469006970525,
-      "learning_rate": 0.00019999482203343237,
-      "loss": 46.0,
-      "step": 20125
-    },
-    {
-      "epoch": 3.2411530254841177,
-      "grad_norm": 0.0164493415504694,
-      "learning_rate": 0.00019999482151858766,
-      "loss": 46.0,
-      "step": 20126
-    },
-    {
-      "epoch": 3.241314062562905,
-      "grad_norm": 0.0008639883599244058,
-      "learning_rate": 0.00019999482100371735,
-      "loss": 46.0,
-      "step": 20127
-    },
-    {
-      "epoch": 3.2414750996416926,
-      "grad_norm": 0.003550648922100663,
-      "learning_rate": 0.00019999482048882147,
-      "loss": 46.0,
-      "step": 20128
-    },
-    {
-      "epoch": 3.24163613672048,
-      "grad_norm": 0.003403973998501897,
-      "learning_rate": 0.0001999948199739,
-      "loss": 46.0,
-      "step": 20129
-    },
-    {
-      "epoch": 3.2417971737992675,
-      "grad_norm": 0.0042611416429281235,
-      "learning_rate": 0.00019999481945895292,
-      "loss": 46.0,
-      "step": 20130
-    },
-    {
-      "epoch": 3.2419582108780545,
-      "grad_norm": 0.0019452397245913744,
-      "learning_rate": 0.00019999481894398025,
-      "loss": 46.0,
-      "step": 20131
-    },
-    {
-      "epoch": 3.242119247956842,
-      "grad_norm": 0.003827241715043783,
-      "learning_rate": 0.00019999481842898197,
-      "loss": 46.0,
-      "step": 20132
-    },
-    {
-      "epoch": 3.2422802850356294,
-      "grad_norm": 0.0025218441151082516,
-      "learning_rate": 0.00019999481791395813,
-      "loss": 46.0,
-      "step": 20133
-    },
-    {
-      "epoch": 3.242441322114417,
-      "grad_norm": 0.006772161927074194,
-      "learning_rate": 0.00019999481739890867,
-      "loss": 46.0,
-      "step": 20134
-    },
-    {
-      "epoch": 3.2426023591932043,
-      "grad_norm": 0.007145663257688284,
-      "learning_rate": 0.00019999481688383363,
-      "loss": 46.0,
-      "step": 20135
-    },
-    {
-      "epoch": 3.2427633962719917,
-      "grad_norm": 0.001702795852907002,
-      "learning_rate": 0.00019999481636873297,
-      "loss": 46.0,
-      "step": 20136
-    },
-    {
-      "epoch": 3.242924433350779,
-      "grad_norm": 0.014489852823317051,
-      "learning_rate": 0.00019999481585360678,
-      "loss": 46.0,
-      "step": 20137
-    },
-    {
-      "epoch": 3.243085470429566,
-      "grad_norm": 0.0037290044128894806,
-      "learning_rate": 0.00019999481533845495,
-      "loss": 46.0,
-      "step": 20138
-    },
-    {
-      "epoch": 3.2432465075083536,
-      "grad_norm": 0.0073197754099965096,
-      "learning_rate": 0.00019999481482327753,
-      "loss": 46.0,
-      "step": 20139
-    },
-    {
-      "epoch": 3.243407544587141,
-      "grad_norm": 0.00508636562153697,
-      "learning_rate": 0.00019999481430807453,
-      "loss": 46.0,
-      "step": 20140
-    },
-    {
-      "epoch": 3.2435685816659285,
-      "grad_norm": 0.0035158093087375164,
-      "learning_rate": 0.00019999481379284593,
-      "loss": 46.0,
-      "step": 20141
-    },
-    {
-      "epoch": 3.243729618744716,
-      "grad_norm": 0.003287319326773286,
-      "learning_rate": 0.0001999948132775917,
-      "loss": 46.0,
-      "step": 20142
-    },
-    {
-      "epoch": 3.2438906558235034,
-      "grad_norm": 0.004204927012324333,
-      "learning_rate": 0.0001999948127623119,
-      "loss": 46.0,
-      "step": 20143
-    },
-    {
-      "epoch": 3.244051692902291,
-      "grad_norm": 0.0014636429259553552,
-      "learning_rate": 0.00019999481224700655,
-      "loss": 46.0,
-      "step": 20144
-    },
-    {
-      "epoch": 3.2442127299810783,
-      "grad_norm": 0.001967041287571192,
-      "learning_rate": 0.00019999481173167555,
-      "loss": 46.0,
-      "step": 20145
-    },
-    {
-      "epoch": 3.2443737670598654,
-      "grad_norm": 0.0039205183275043964,
-      "learning_rate": 0.00019999481121631897,
-      "loss": 46.0,
-      "step": 20146
-    },
-    {
-      "epoch": 3.244534804138653,
-      "grad_norm": 0.0016264102887362242,
-      "learning_rate": 0.00019999481070093683,
-      "loss": 46.0,
-      "step": 20147
-    },
-    {
-      "epoch": 3.2446958412174403,
-      "grad_norm": 0.0010595329804345965,
-      "learning_rate": 0.00019999481018552907,
-      "loss": 46.0,
-      "step": 20148
-    },
-    {
-      "epoch": 3.2448568782962277,
-      "grad_norm": 0.0006715874187648296,
-      "learning_rate": 0.0001999948096700957,
-      "loss": 46.0,
-      "step": 20149
-    },
-    {
-      "epoch": 3.245017915375015,
-      "grad_norm": 0.008115099743008614,
-      "learning_rate": 0.00019999480915463676,
-      "loss": 46.0,
-      "step": 20150
-    },
-    {
-      "epoch": 3.2451789524538026,
-      "grad_norm": 0.0025919293984770775,
-      "learning_rate": 0.00019999480863915224,
-      "loss": 46.0,
-      "step": 20151
-    },
-    {
-      "epoch": 3.24533998953259,
-      "grad_norm": 0.0009808718459680676,
-      "learning_rate": 0.00019999480812364208,
-      "loss": 46.0,
-      "step": 20152
-    },
-    {
-      "epoch": 3.245501026611377,
-      "grad_norm": 0.003987419884651899,
-      "learning_rate": 0.00019999480760810636,
-      "loss": 46.0,
-      "step": 20153
-    },
-    {
-      "epoch": 3.2456620636901645,
-      "grad_norm": 0.010316704399883747,
-      "learning_rate": 0.00019999480709254503,
-      "loss": 46.0,
-      "step": 20154
-    },
-    {
-      "epoch": 3.245823100768952,
-      "grad_norm": 0.005609240382909775,
-      "learning_rate": 0.00019999480657695813,
-      "loss": 46.0,
-      "step": 20155
-    },
-    {
-      "epoch": 3.2459841378477394,
-      "grad_norm": 0.008579924702644348,
-      "learning_rate": 0.00019999480606134562,
-      "loss": 46.0,
-      "step": 20156
-    },
-    {
-      "epoch": 3.246145174926527,
-      "grad_norm": 0.004717460833489895,
-      "learning_rate": 0.00019999480554570753,
-      "loss": 46.0,
-      "step": 20157
-    },
-    {
-      "epoch": 3.2463062120053143,
-      "grad_norm": 0.015387299470603466,
-      "learning_rate": 0.00019999480503004382,
-      "loss": 46.0,
-      "step": 20158
-    },
-    {
-      "epoch": 3.246467249084102,
-      "grad_norm": 0.0026779938489198685,
-      "learning_rate": 0.00019999480451435452,
-      "loss": 46.0,
-      "step": 20159
-    },
-    {
-      "epoch": 3.2466282861628892,
-      "grad_norm": 0.0012958986917510629,
-      "learning_rate": 0.00019999480399863963,
-      "loss": 46.0,
-      "step": 20160
-    },
-    {
-      "epoch": 3.2467893232416762,
-      "grad_norm": 0.010639621876180172,
-      "learning_rate": 0.00019999480348289916,
-      "loss": 46.0,
-      "step": 20161
-    },
-    {
-      "epoch": 3.2469503603204637,
-      "grad_norm": 0.001926374388858676,
-      "learning_rate": 0.0001999948029671331,
-      "loss": 46.0,
-      "step": 20162
-    },
-    {
-      "epoch": 3.247111397399251,
-      "grad_norm": 0.0015899967402219772,
-      "learning_rate": 0.00019999480245134143,
-      "loss": 46.0,
-      "step": 20163
-    },
-    {
-      "epoch": 3.2472724344780386,
-      "grad_norm": 0.001845594379119575,
-      "learning_rate": 0.00019999480193552417,
-      "loss": 46.0,
-      "step": 20164
-    },
-    {
-      "epoch": 3.247433471556826,
-      "grad_norm": 0.004314039368182421,
-      "learning_rate": 0.00019999480141968135,
-      "loss": 46.0,
-      "step": 20165
-    },
-    {
-      "epoch": 3.2475945086356135,
-      "grad_norm": 0.004243573639541864,
-      "learning_rate": 0.00019999480090381288,
-      "loss": 46.0,
-      "step": 20166
-    },
-    {
-      "epoch": 3.247755545714401,
-      "grad_norm": 0.0025196417700499296,
-      "learning_rate": 0.00019999480038791883,
-      "loss": 46.0,
-      "step": 20167
-    },
-    {
-      "epoch": 3.247916582793188,
-      "grad_norm": 0.005762726068496704,
-      "learning_rate": 0.00019999479987199923,
-      "loss": 46.0,
-      "step": 20168
-    },
-    {
-      "epoch": 3.2480776198719754,
-      "grad_norm": 0.0007746221963316202,
-      "learning_rate": 0.000199994799356054,
-      "loss": 46.0,
-      "step": 20169
-    },
-    {
-      "epoch": 3.248238656950763,
-      "grad_norm": 0.0020391386933624744,
-      "learning_rate": 0.0001999947988400832,
-      "loss": 46.0,
-      "step": 20170
-    },
-    {
-      "epoch": 3.2483996940295503,
-      "grad_norm": 0.004793431144207716,
-      "learning_rate": 0.00019999479832408677,
-      "loss": 46.0,
-      "step": 20171
-    },
-    {
-      "epoch": 3.2485607311083378,
-      "grad_norm": 0.0006186505197547376,
-      "learning_rate": 0.00019999479780806478,
-      "loss": 46.0,
-      "step": 20172
-    },
-    {
-      "epoch": 3.248721768187125,
-      "grad_norm": 0.004951554350554943,
-      "learning_rate": 0.00019999479729201718,
-      "loss": 46.0,
-      "step": 20173
-    },
-    {
-      "epoch": 3.2488828052659127,
-      "grad_norm": 0.004281815141439438,
-      "learning_rate": 0.000199994796775944,
-      "loss": 46.0,
-      "step": 20174
-    },
-    {
-      "epoch": 3.2490438423446997,
-      "grad_norm": 0.001294878195039928,
-      "learning_rate": 0.0001999947962598452,
-      "loss": 46.0,
-      "step": 20175
-    },
-    {
-      "epoch": 3.249204879423487,
-      "grad_norm": 0.0044972640462219715,
-      "learning_rate": 0.00019999479574372083,
-      "loss": 46.0,
-      "step": 20176
-    },
-    {
-      "epoch": 3.2493659165022746,
-      "grad_norm": 0.003758159698918462,
-      "learning_rate": 0.00019999479522757083,
-      "loss": 46.0,
-      "step": 20177
-    },
-    {
-      "epoch": 3.249526953581062,
-      "grad_norm": 0.002592820441350341,
-      "learning_rate": 0.0001999947947113953,
-      "loss": 46.0,
-      "step": 20178
-    },
-    {
-      "epoch": 3.2496879906598495,
-      "grad_norm": 0.006812127307057381,
-      "learning_rate": 0.00019999479419519412,
-      "loss": 46.0,
-      "step": 20179
-    },
-    {
-      "epoch": 3.249849027738637,
-      "grad_norm": 0.001565458718687296,
-      "learning_rate": 0.00019999479367896735,
-      "loss": 46.0,
-      "step": 20180
-    },
-    {
-      "epoch": 3.2500100648174244,
-      "grad_norm": 0.004508504178375006,
-      "learning_rate": 0.00019999479316271503,
-      "loss": 46.0,
-      "step": 20181
-    },
-    {
-      "epoch": 3.2501711018962114,
-      "grad_norm": 0.00215457403101027,
-      "learning_rate": 0.0001999947926464371,
-      "loss": 46.0,
-      "step": 20182
-    },
-    {
-      "epoch": 3.250332138974999,
-      "grad_norm": 0.0008396839839406312,
-      "learning_rate": 0.00019999479213013354,
-      "loss": 46.0,
-      "step": 20183
-    },
-    {
-      "epoch": 3.2504931760537863,
-      "grad_norm": 0.002159419935196638,
-      "learning_rate": 0.00019999479161380442,
-      "loss": 46.0,
-      "step": 20184
-    },
-    {
-      "epoch": 3.2506542131325737,
-      "grad_norm": 0.0007195051875896752,
-      "learning_rate": 0.0001999947910974497,
-      "loss": 46.0,
-      "step": 20185
-    },
-    {
-      "epoch": 3.250815250211361,
-      "grad_norm": 0.00041729031363502145,
-      "learning_rate": 0.00019999479058106938,
-      "loss": 46.0,
-      "step": 20186
-    },
-    {
-      "epoch": 3.2509762872901486,
-      "grad_norm": 0.0020426579285413027,
-      "learning_rate": 0.00019999479006466348,
-      "loss": 46.0,
-      "step": 20187
-    },
-    {
-      "epoch": 3.251137324368936,
-      "grad_norm": 0.0027418043464422226,
-      "learning_rate": 0.00019999478954823197,
-      "loss": 46.0,
-      "step": 20188
-    },
-    {
-      "epoch": 3.2512983614477236,
-      "grad_norm": 0.001304731471464038,
-      "learning_rate": 0.0001999947890317749,
-      "loss": 46.0,
-      "step": 20189
-    },
-    {
-      "epoch": 3.2514593985265106,
-      "grad_norm": 0.0023751843255013227,
-      "learning_rate": 0.00019999478851529217,
-      "loss": 46.0,
-      "step": 20190
-    },
-    {
-      "epoch": 3.251620435605298,
-      "grad_norm": 0.0017244471237063408,
-      "learning_rate": 0.00019999478799878392,
-      "loss": 46.0,
-      "step": 20191
-    },
-    {
-      "epoch": 3.2517814726840855,
-      "grad_norm": 0.004116279538720846,
-      "learning_rate": 0.00019999478748225003,
-      "loss": 46.0,
-      "step": 20192
-    },
-    {
-      "epoch": 3.251942509762873,
-      "grad_norm": 0.0018614638829603791,
-      "learning_rate": 0.00019999478696569058,
-      "loss": 46.0,
-      "step": 20193
-    },
-    {
-      "epoch": 3.2521035468416604,
-      "grad_norm": 0.001488811569288373,
-      "learning_rate": 0.0001999947864491055,
-      "loss": 46.0,
-      "step": 20194
-    },
-    {
-      "epoch": 3.252264583920448,
-      "grad_norm": 0.0035445280373096466,
-      "learning_rate": 0.00019999478593249486,
-      "loss": 46.0,
-      "step": 20195
-    },
-    {
-      "epoch": 3.2524256209992353,
-      "grad_norm": 0.005500501021742821,
-      "learning_rate": 0.00019999478541585862,
-      "loss": 46.0,
-      "step": 20196
-    },
-    {
-      "epoch": 3.2525866580780223,
-      "grad_norm": 0.0012397834798321128,
-      "learning_rate": 0.00019999478489919677,
-      "loss": 46.0,
-      "step": 20197
-    },
-    {
-      "epoch": 3.2527476951568097,
-      "grad_norm": 0.004841955844312906,
-      "learning_rate": 0.00019999478438250933,
-      "loss": 46.0,
-      "step": 20198
-    },
-    {
-      "epoch": 3.252908732235597,
-      "grad_norm": 0.0013978432398289442,
-      "learning_rate": 0.0001999947838657963,
-      "loss": 46.0,
-      "step": 20199
-    },
-    {
-      "epoch": 3.2530697693143846,
-      "grad_norm": 0.005983384791761637,
-      "learning_rate": 0.00019999478334905768,
-      "loss": 46.0,
-      "step": 20200
-    },
-    {
-      "epoch": 3.253230806393172,
-      "grad_norm": 0.0022298377007246017,
-      "learning_rate": 0.00019999478283229345,
-      "loss": 46.0,
-      "step": 20201
-    },
-    {
-      "epoch": 3.2533918434719595,
-      "grad_norm": 0.002500379690900445,
-      "learning_rate": 0.00019999478231550363,
-      "loss": 46.0,
-      "step": 20202
-    },
-    {
-      "epoch": 3.253552880550747,
-      "grad_norm": 0.00448454637080431,
-      "learning_rate": 0.00019999478179868826,
-      "loss": 46.0,
-      "step": 20203
-    },
-    {
-      "epoch": 3.2537139176295344,
-      "grad_norm": 0.0016677109524607658,
-      "learning_rate": 0.00019999478128184724,
-      "loss": 46.0,
-      "step": 20204
-    },
-    {
-      "epoch": 3.2538749547083214,
-      "grad_norm": 0.0017818021588027477,
-      "learning_rate": 0.00019999478076498063,
-      "loss": 46.0,
-      "step": 20205
-    },
-    {
-      "epoch": 3.254035991787109,
-      "grad_norm": 0.005662537645548582,
-      "learning_rate": 0.00019999478024808847,
-      "loss": 46.0,
-      "step": 20206
-    },
-    {
-      "epoch": 3.2541970288658963,
-      "grad_norm": 0.0025463267229497433,
-      "learning_rate": 0.00019999477973117069,
-      "loss": 46.0,
-      "step": 20207
-    },
-    {
-      "epoch": 3.254358065944684,
-      "grad_norm": 0.002630396280437708,
-      "learning_rate": 0.00019999477921422735,
-      "loss": 46.0,
-      "step": 20208
-    },
-    {
-      "epoch": 3.2545191030234712,
-      "grad_norm": 0.0034333046060055494,
-      "learning_rate": 0.00019999477869725836,
-      "loss": 46.0,
-      "step": 20209
-    },
-    {
-      "epoch": 3.2546801401022587,
-      "grad_norm": 0.0009719359804876149,
-      "learning_rate": 0.0001999947781802638,
-      "loss": 46.0,
-      "step": 20210
-    },
-    {
-      "epoch": 3.2548411771810457,
-      "grad_norm": 0.00383931677788496,
-      "learning_rate": 0.00019999477766324364,
-      "loss": 46.0,
-      "step": 20211
-    },
-    {
-      "epoch": 3.255002214259833,
-      "grad_norm": 0.0007152375183068216,
-      "learning_rate": 0.0001999947771461979,
-      "loss": 46.0,
-      "step": 20212
-    },
-    {
-      "epoch": 3.2551632513386206,
-      "grad_norm": 0.002020742278546095,
-      "learning_rate": 0.00019999477662912657,
-      "loss": 46.0,
-      "step": 20213
-    },
-    {
-      "epoch": 3.255324288417408,
-      "grad_norm": 0.002276879968121648,
-      "learning_rate": 0.00019999477611202965,
-      "loss": 46.0,
-      "step": 20214
-    },
-    {
-      "epoch": 3.2554853254961955,
-      "grad_norm": 0.002185808029025793,
-      "learning_rate": 0.00019999477559490712,
-      "loss": 46.0,
-      "step": 20215
-    },
-    {
-      "epoch": 3.255646362574983,
-      "grad_norm": 0.0019452732522040606,
-      "learning_rate": 0.00019999477507775897,
-      "loss": 46.0,
-      "step": 20216
-    },
-    {
-      "epoch": 3.2558073996537704,
-      "grad_norm": 0.0019966145046055317,
-      "learning_rate": 0.00019999477456058526,
-      "loss": 46.0,
-      "step": 20217
-    },
-    {
-      "epoch": 3.255968436732558,
-      "grad_norm": 0.003022721502929926,
-      "learning_rate": 0.00019999477404338597,
-      "loss": 46.0,
-      "step": 20218
-    },
-    {
-      "epoch": 3.256129473811345,
-      "grad_norm": 0.0034912684932351112,
-      "learning_rate": 0.00019999477352616106,
-      "loss": 46.0,
-      "step": 20219
-    },
-    {
-      "epoch": 3.2562905108901323,
-      "grad_norm": 0.003489930182695389,
-      "learning_rate": 0.00019999477300891057,
-      "loss": 46.0,
-      "step": 20220
-    },
-    {
-      "epoch": 3.25645154796892,
-      "grad_norm": 0.00244498229585588,
-      "learning_rate": 0.00019999477249163449,
-      "loss": 46.0,
-      "step": 20221
-    },
-    {
-      "epoch": 3.2566125850477072,
-      "grad_norm": 0.004198648501187563,
-      "learning_rate": 0.0001999947719743328,
-      "loss": 46.0,
-      "step": 20222
-    },
-    {
-      "epoch": 3.2567736221264947,
-      "grad_norm": 0.010327229276299477,
-      "learning_rate": 0.00019999477145700553,
-      "loss": 46.0,
-      "step": 20223
-    },
-    {
-      "epoch": 3.256934659205282,
-      "grad_norm": 0.003642395371571183,
-      "learning_rate": 0.00019999477093965266,
-      "loss": 46.0,
-      "step": 20224
-    },
-    {
-      "epoch": 3.2570956962840696,
-      "grad_norm": 0.002781534567475319,
-      "learning_rate": 0.0001999947704222742,
-      "loss": 46.0,
-      "step": 20225
-    },
-    {
-      "epoch": 3.2572567333628566,
-      "grad_norm": 0.0007274140953086317,
-      "learning_rate": 0.00019999476990487013,
-      "loss": 46.0,
-      "step": 20226
-    },
-    {
-      "epoch": 3.257417770441644,
-      "grad_norm": 0.002433524699881673,
-      "learning_rate": 0.00019999476938744047,
-      "loss": 46.0,
-      "step": 20227
-    },
-    {
-      "epoch": 3.2575788075204315,
-      "grad_norm": 0.0006209313869476318,
-      "learning_rate": 0.00019999476886998525,
-      "loss": 46.0,
-      "step": 20228
-    },
-    {
-      "epoch": 3.257739844599219,
-      "grad_norm": 0.0038785021752119064,
-      "learning_rate": 0.00019999476835250442,
-      "loss": 46.0,
-      "step": 20229
-    },
-    {
-      "epoch": 3.2579008816780064,
-      "grad_norm": 0.0014134763041511178,
-      "learning_rate": 0.00019999476783499797,
-      "loss": 46.0,
-      "step": 20230
-    },
-    {
-      "epoch": 3.258061918756794,
-      "grad_norm": 0.00252518686465919,
-      "learning_rate": 0.00019999476731746596,
-      "loss": 46.0,
-      "step": 20231
-    },
-    {
-      "epoch": 3.2582229558355813,
-      "grad_norm": 0.0059619261883199215,
-      "learning_rate": 0.00019999476679990834,
-      "loss": 46.0,
-      "step": 20232
-    },
-    {
-      "epoch": 3.2583839929143688,
-      "grad_norm": 0.0005177116836421192,
-      "learning_rate": 0.0001999947662823251,
-      "loss": 46.0,
-      "step": 20233
-    },
-    {
-      "epoch": 3.2585450299931558,
-      "grad_norm": 0.0035683391615748405,
-      "learning_rate": 0.0001999947657647163,
-      "loss": 46.0,
-      "step": 20234
-    },
-    {
-      "epoch": 3.258706067071943,
-      "grad_norm": 0.007995419204235077,
-      "learning_rate": 0.00019999476524708193,
-      "loss": 46.0,
-      "step": 20235
-    },
-    {
-      "epoch": 3.2588671041507307,
-      "grad_norm": 0.0016613744664937258,
-      "learning_rate": 0.00019999476472942193,
-      "loss": 46.0,
-      "step": 20236
-    },
-    {
-      "epoch": 3.259028141229518,
-      "grad_norm": 0.0028535693418234587,
-      "learning_rate": 0.00019999476421173634,
-      "loss": 46.0,
-      "step": 20237
-    },
-    {
-      "epoch": 3.2591891783083056,
-      "grad_norm": 0.0012323361588642001,
-      "learning_rate": 0.0001999947636940252,
-      "loss": 46.0,
-      "step": 20238
-    },
-    {
-      "epoch": 3.259350215387093,
-      "grad_norm": 0.005233832634985447,
-      "learning_rate": 0.0001999947631762884,
-      "loss": 46.0,
-      "step": 20239
-    },
-    {
-      "epoch": 3.25951125246588,
-      "grad_norm": 0.008632045239210129,
-      "learning_rate": 0.00019999476265852604,
-      "loss": 46.0,
-      "step": 20240
-    },
-    {
-      "epoch": 3.2596722895446675,
-      "grad_norm": 0.000558165367692709,
-      "learning_rate": 0.0001999947621407381,
-      "loss": 46.0,
-      "step": 20241
-    },
-    {
-      "epoch": 3.259833326623455,
-      "grad_norm": 0.007402414921671152,
-      "learning_rate": 0.00019999476162292453,
-      "loss": 46.0,
-      "step": 20242
-    },
-    {
-      "epoch": 3.2599943637022424,
-      "grad_norm": 0.0020012983586639166,
-      "learning_rate": 0.00019999476110508537,
-      "loss": 46.0,
-      "step": 20243
-    },
-    {
-      "epoch": 3.26015540078103,
-      "grad_norm": 0.008699571713805199,
-      "learning_rate": 0.00019999476058722064,
-      "loss": 46.0,
-      "step": 20244
-    },
-    {
-      "epoch": 3.2603164378598173,
-      "grad_norm": 0.0037860579323023558,
-      "learning_rate": 0.0001999947600693303,
-      "loss": 46.0,
-      "step": 20245
-    },
-    {
-      "epoch": 3.2604774749386047,
-      "grad_norm": 0.0028904483187943697,
-      "learning_rate": 0.0001999947595514144,
-      "loss": 46.0,
-      "step": 20246
-    },
-    {
-      "epoch": 3.260638512017392,
-      "grad_norm": 0.009074817411601543,
-      "learning_rate": 0.00019999475903347287,
-      "loss": 46.0,
-      "step": 20247
-    },
-    {
-      "epoch": 3.2607995490961796,
-      "grad_norm": 0.002737767994403839,
-      "learning_rate": 0.00019999475851550575,
-      "loss": 46.0,
-      "step": 20248
-    },
-    {
-      "epoch": 3.2609605861749666,
-      "grad_norm": 0.0028765045572072268,
-      "learning_rate": 0.00019999475799751303,
-      "loss": 46.0,
-      "step": 20249
-    },
-    {
-      "epoch": 3.261121623253754,
-      "grad_norm": 0.0038798353634774685,
-      "learning_rate": 0.00019999475747949473,
-      "loss": 46.0,
-      "step": 20250
-    },
-    {
-      "epoch": 3.2612826603325415,
-      "grad_norm": 0.001062415773048997,
-      "learning_rate": 0.00019999475696145087,
-      "loss": 46.0,
-      "step": 20251
-    },
-    {
-      "epoch": 3.261443697411329,
-      "grad_norm": 0.005231064278632402,
-      "learning_rate": 0.00019999475644338137,
-      "loss": 46.0,
-      "step": 20252
-    },
-    {
-      "epoch": 3.2616047344901165,
-      "grad_norm": 0.0025006080977618694,
-      "learning_rate": 0.00019999475592528628,
-      "loss": 46.0,
-      "step": 20253
-    },
-    {
-      "epoch": 3.261765771568904,
-      "grad_norm": 0.0017447317950427532,
-      "learning_rate": 0.0001999947554071656,
-      "loss": 46.0,
-      "step": 20254
-    },
-    {
-      "epoch": 3.261926808647691,
-      "grad_norm": 0.003976028878241777,
-      "learning_rate": 0.00019999475488901935,
-      "loss": 46.0,
-      "step": 20255
-    },
-    {
-      "epoch": 3.2620878457264784,
-      "grad_norm": 0.004502329509705305,
-      "learning_rate": 0.0001999947543708475,
-      "loss": 46.0,
-      "step": 20256
-    },
-    {
-      "epoch": 3.262248882805266,
-      "grad_norm": 0.0025826182682067156,
-      "learning_rate": 0.00019999475385265003,
-      "loss": 46.0,
-      "step": 20257
-    },
-    {
-      "epoch": 3.2624099198840533,
-      "grad_norm": 0.0040931422263383865,
-      "learning_rate": 0.00019999475333442698,
-      "loss": 46.0,
-      "step": 20258
-    },
-    {
-      "epoch": 3.2625709569628407,
-      "grad_norm": 0.005306491162627935,
-      "learning_rate": 0.00019999475281617834,
-      "loss": 46.0,
-      "step": 20259
-    },
-    {
-      "epoch": 3.262731994041628,
-      "grad_norm": 0.00695073651149869,
-      "learning_rate": 0.00019999475229790412,
-      "loss": 46.0,
-      "step": 20260
-    },
-    {
-      "epoch": 3.2628930311204156,
-      "grad_norm": 0.003346773562952876,
-      "learning_rate": 0.00019999475177960428,
-      "loss": 46.0,
-      "step": 20261
-    },
-    {
-      "epoch": 3.263054068199203,
-      "grad_norm": 0.008357105776667595,
-      "learning_rate": 0.00019999475126127885,
-      "loss": 46.0,
-      "step": 20262
-    },
-    {
-      "epoch": 3.26321510527799,
-      "grad_norm": 0.005580850411206484,
-      "learning_rate": 0.00019999475074292784,
-      "loss": 46.0,
-      "step": 20263
-    },
-    {
-      "epoch": 3.2633761423567775,
-      "grad_norm": 0.003865674138069153,
-      "learning_rate": 0.00019999475022455123,
-      "loss": 46.0,
-      "step": 20264
-    },
-    {
-      "epoch": 3.263537179435565,
-      "grad_norm": 0.007624385412782431,
-      "learning_rate": 0.00019999474970614905,
-      "loss": 46.0,
-      "step": 20265
-    },
-    {
-      "epoch": 3.2636982165143524,
-      "grad_norm": 0.0013032362330704927,
-      "learning_rate": 0.00019999474918772124,
-      "loss": 46.0,
-      "step": 20266
-    },
-    {
-      "epoch": 3.26385925359314,
-      "grad_norm": 0.005241406615823507,
-      "learning_rate": 0.00019999474866926785,
-      "loss": 46.0,
-      "step": 20267
-    },
-    {
-      "epoch": 3.2640202906719273,
-      "grad_norm": 0.00929312314838171,
-      "learning_rate": 0.00019999474815078885,
-      "loss": 46.0,
-      "step": 20268
-    },
-    {
-      "epoch": 3.264181327750715,
-      "grad_norm": 0.003043997334316373,
-      "learning_rate": 0.00019999474763228428,
-      "loss": 46.0,
-      "step": 20269
-    },
-    {
-      "epoch": 3.264342364829502,
-      "grad_norm": 0.002271262463182211,
-      "learning_rate": 0.00019999474711375413,
-      "loss": 46.0,
-      "step": 20270
-    },
-    {
-      "epoch": 3.2645034019082892,
-      "grad_norm": 0.00334842293523252,
-      "learning_rate": 0.00019999474659519834,
-      "loss": 46.0,
-      "step": 20271
-    },
-    {
-      "epoch": 3.2646644389870767,
-      "grad_norm": 0.0010278024710714817,
-      "learning_rate": 0.000199994746076617,
-      "loss": 46.0,
-      "step": 20272
-    },
-    {
-      "epoch": 3.264825476065864,
-      "grad_norm": 0.0020910913590341806,
-      "learning_rate": 0.00019999474555801005,
-      "loss": 46.0,
-      "step": 20273
-    },
-    {
-      "epoch": 3.2649865131446516,
-      "grad_norm": 0.003604491939768195,
-      "learning_rate": 0.0001999947450393775,
-      "loss": 46.0,
-      "step": 20274
-    },
-    {
-      "epoch": 3.265147550223439,
-      "grad_norm": 0.003708265256136656,
-      "learning_rate": 0.00019999474452071935,
-      "loss": 46.0,
-      "step": 20275
-    },
-    {
-      "epoch": 3.2653085873022265,
-      "grad_norm": 0.002162586199119687,
-      "learning_rate": 0.00019999474400203565,
-      "loss": 46.0,
-      "step": 20276
-    },
-    {
-      "epoch": 3.265469624381014,
-      "grad_norm": 0.005173577461391687,
-      "learning_rate": 0.0001999947434833263,
-      "loss": 46.0,
-      "step": 20277
-    },
-    {
-      "epoch": 3.265630661459801,
-      "grad_norm": 0.004262245260179043,
-      "learning_rate": 0.00019999474296459138,
-      "loss": 46.0,
-      "step": 20278
-    },
-    {
-      "epoch": 3.2657916985385884,
-      "grad_norm": 0.002357808407396078,
-      "learning_rate": 0.00019999474244583089,
-      "loss": 46.0,
-      "step": 20279
-    },
-    {
-      "epoch": 3.265952735617376,
-      "grad_norm": 0.0014336516615003347,
-      "learning_rate": 0.00019999474192704478,
-      "loss": 46.0,
-      "step": 20280
-    },
-    {
-      "epoch": 3.2661137726961633,
-      "grad_norm": 0.0024729929864406586,
-      "learning_rate": 0.00019999474140823306,
-      "loss": 46.0,
-      "step": 20281
-    },
-    {
-      "epoch": 3.2662748097749508,
-      "grad_norm": 0.0009040281875059009,
-      "learning_rate": 0.00019999474088939576,
-      "loss": 46.0,
-      "step": 20282
-    },
-    {
-      "epoch": 3.2664358468537382,
-      "grad_norm": 0.010032703168690205,
-      "learning_rate": 0.0001999947403705329,
-      "loss": 46.0,
-      "step": 20283
-    },
-    {
-      "epoch": 3.2665968839325252,
-      "grad_norm": 0.0009144143550656736,
-      "learning_rate": 0.0001999947398516444,
-      "loss": 46.0,
-      "step": 20284
-    },
-    {
-      "epoch": 3.2667579210113127,
-      "grad_norm": 0.0071673267520964146,
-      "learning_rate": 0.00019999473933273034,
-      "loss": 46.0,
-      "step": 20285
-    },
-    {
-      "epoch": 3.2669189580901,
-      "grad_norm": 0.006829691119492054,
-      "learning_rate": 0.00019999473881379066,
-      "loss": 46.0,
-      "step": 20286
-    },
-    {
-      "epoch": 3.2670799951688876,
-      "grad_norm": 0.0014621352311223745,
-      "learning_rate": 0.00019999473829482542,
-      "loss": 46.0,
-      "step": 20287
-    },
-    {
-      "epoch": 3.267241032247675,
-      "grad_norm": 0.0033235775772482157,
-      "learning_rate": 0.00019999473777583456,
-      "loss": 46.0,
-      "step": 20288
-    },
-    {
-      "epoch": 3.2674020693264625,
-      "grad_norm": 0.002992296824231744,
-      "learning_rate": 0.0001999947372568181,
-      "loss": 46.0,
-      "step": 20289
-    },
-    {
-      "epoch": 3.26756310640525,
-      "grad_norm": 0.005860642530024052,
-      "learning_rate": 0.00019999473673777606,
-      "loss": 46.0,
-      "step": 20290
-    },
-    {
-      "epoch": 3.2677241434840374,
-      "grad_norm": 0.001699344371445477,
-      "learning_rate": 0.00019999473621870844,
-      "loss": 46.0,
-      "step": 20291
-    },
-    {
-      "epoch": 3.2678851805628244,
-      "grad_norm": 0.0011733805295079947,
-      "learning_rate": 0.0001999947356996152,
-      "loss": 46.0,
-      "step": 20292
-    },
-    {
-      "epoch": 3.268046217641612,
-      "grad_norm": 0.0076037305407226086,
-      "learning_rate": 0.0001999947351804964,
-      "loss": 46.0,
-      "step": 20293
-    },
-    {
-      "epoch": 3.2682072547203993,
-      "grad_norm": 0.00441789161413908,
-      "learning_rate": 0.00019999473466135196,
-      "loss": 46.0,
-      "step": 20294
-    },
-    {
-      "epoch": 3.2683682917991868,
-      "grad_norm": 0.005370750557631254,
-      "learning_rate": 0.00019999473414218194,
-      "loss": 46.0,
-      "step": 20295
-    },
-    {
-      "epoch": 3.268529328877974,
-      "grad_norm": 0.011534000746905804,
-      "learning_rate": 0.00019999473362298636,
-      "loss": 46.0,
-      "step": 20296
-    },
-    {
-      "epoch": 3.2686903659567617,
-      "grad_norm": 0.0047613345086574554,
-      "learning_rate": 0.00019999473310376516,
-      "loss": 46.0,
-      "step": 20297
-    },
-    {
-      "epoch": 3.268851403035549,
-      "grad_norm": 0.0040600174106657505,
-      "learning_rate": 0.00019999473258451838,
-      "loss": 46.0,
-      "step": 20298
-    },
-    {
-      "epoch": 3.269012440114336,
-      "grad_norm": 0.005267912056297064,
-      "learning_rate": 0.00019999473206524598,
-      "loss": 46.0,
-      "step": 20299
-    },
-    {
-      "epoch": 3.2691734771931236,
-      "grad_norm": 0.0023280796594917774,
-      "learning_rate": 0.000199994731545948,
-      "loss": 46.0,
-      "step": 20300
-    },
-    {
-      "epoch": 3.269334514271911,
-      "grad_norm": 0.00952787883579731,
-      "learning_rate": 0.00019999473102662446,
-      "loss": 46.0,
-      "step": 20301
-    },
-    {
-      "epoch": 3.2694955513506985,
-      "grad_norm": 0.002269686432555318,
-      "learning_rate": 0.00019999473050727527,
-      "loss": 46.0,
-      "step": 20302
-    },
-    {
-      "epoch": 3.269656588429486,
-      "grad_norm": 0.00834378320723772,
-      "learning_rate": 0.00019999472998790053,
-      "loss": 46.0,
-      "step": 20303
-    },
-    {
-      "epoch": 3.2698176255082734,
-      "grad_norm": 0.0015597398160025477,
-      "learning_rate": 0.00019999472946850017,
-      "loss": 46.0,
-      "step": 20304
-    },
-    {
-      "epoch": 3.269978662587061,
-      "grad_norm": 0.0025558231864124537,
-      "learning_rate": 0.00019999472894907422,
-      "loss": 46.0,
-      "step": 20305
-    },
-    {
-      "epoch": 3.2701396996658483,
-      "grad_norm": 0.003061908995732665,
-      "learning_rate": 0.0001999947284296227,
-      "loss": 46.0,
-      "step": 20306
-    },
-    {
-      "epoch": 3.2703007367446353,
-      "grad_norm": 0.008666201494634151,
-      "learning_rate": 0.00019999472791014554,
-      "loss": 46.0,
-      "step": 20307
-    },
-    {
-      "epoch": 3.2704617738234227,
-      "grad_norm": 0.0032079971861094236,
-      "learning_rate": 0.00019999472739064283,
-      "loss": 46.0,
-      "step": 20308
-    },
-    {
-      "epoch": 3.27062281090221,
-      "grad_norm": 0.0030128522776067257,
-      "learning_rate": 0.0001999947268711145,
-      "loss": 46.0,
-      "step": 20309
-    },
-    {
-      "epoch": 3.2707838479809976,
-      "grad_norm": 0.005816400051116943,
-      "learning_rate": 0.0001999947263515606,
-      "loss": 46.0,
-      "step": 20310
-    },
-    {
-      "epoch": 3.270944885059785,
-      "grad_norm": 0.003315188456326723,
-      "learning_rate": 0.0001999947258319811,
-      "loss": 46.0,
-      "step": 20311
-    },
-    {
-      "epoch": 3.2711059221385725,
-      "grad_norm": 0.0037742629647254944,
-      "learning_rate": 0.000199994725312376,
-      "loss": 46.0,
-      "step": 20312
-    },
-    {
-      "epoch": 3.2712669592173595,
-      "grad_norm": 0.002245318843051791,
-      "learning_rate": 0.0001999947247927453,
-      "loss": 46.0,
-      "step": 20313
-    },
-    {
-      "epoch": 3.271427996296147,
-      "grad_norm": 0.0032121038530021906,
-      "learning_rate": 0.00019999472427308904,
-      "loss": 46.0,
-      "step": 20314
-    },
-    {
-      "epoch": 3.2715890333749345,
-      "grad_norm": 0.006445621140301228,
-      "learning_rate": 0.00019999472375340716,
-      "loss": 46.0,
-      "step": 20315
-    },
-    {
-      "epoch": 3.271750070453722,
-      "grad_norm": 0.004758812487125397,
-      "learning_rate": 0.00019999472323369968,
-      "loss": 46.0,
-      "step": 20316
-    },
-    {
-      "epoch": 3.2719111075325094,
-      "grad_norm": 0.011252738535404205,
-      "learning_rate": 0.0001999947227139666,
-      "loss": 46.0,
-      "step": 20317
-    },
-    {
-      "epoch": 3.272072144611297,
-      "grad_norm": 0.00478716753423214,
-      "learning_rate": 0.00019999472219420797,
-      "loss": 46.0,
-      "step": 20318
-    },
-    {
-      "epoch": 3.2722331816900843,
-      "grad_norm": 0.009921620599925518,
-      "learning_rate": 0.0001999947216744237,
-      "loss": 46.0,
-      "step": 20319
-    },
-    {
-      "epoch": 3.2723942187688717,
-      "grad_norm": 0.001872863038443029,
-      "learning_rate": 0.00019999472115461386,
-      "loss": 46.0,
-      "step": 20320
-    },
-    {
-      "epoch": 3.272555255847659,
-      "grad_norm": 0.0028653375338763,
-      "learning_rate": 0.0001999947206347784,
-      "loss": 46.0,
-      "step": 20321
-    },
-    {
-      "epoch": 3.272716292926446,
-      "grad_norm": 0.0021373596973717213,
-      "learning_rate": 0.00019999472011491737,
-      "loss": 46.0,
-      "step": 20322
-    },
-    {
-      "epoch": 3.2728773300052336,
-      "grad_norm": 0.00657370500266552,
-      "learning_rate": 0.00019999471959503075,
-      "loss": 46.0,
-      "step": 20323
-    },
-    {
-      "epoch": 3.273038367084021,
-      "grad_norm": 0.001477471087127924,
-      "learning_rate": 0.00019999471907511854,
-      "loss": 46.0,
-      "step": 20324
-    },
-    {
-      "epoch": 3.2731994041628085,
-      "grad_norm": 0.0005806459812447429,
-      "learning_rate": 0.0001999947185551807,
-      "loss": 46.0,
-      "step": 20325
-    },
-    {
-      "epoch": 3.273360441241596,
-      "grad_norm": 0.000988700077868998,
-      "learning_rate": 0.0001999947180352173,
-      "loss": 46.0,
-      "step": 20326
-    },
-    {
-      "epoch": 3.2735214783203834,
-      "grad_norm": 0.001616235706023872,
-      "learning_rate": 0.00019999471751522833,
-      "loss": 46.0,
-      "step": 20327
-    },
-    {
-      "epoch": 3.2736825153991704,
-      "grad_norm": 0.004434907343238592,
-      "learning_rate": 0.0001999947169952137,
-      "loss": 46.0,
-      "step": 20328
-    },
-    {
-      "epoch": 3.273843552477958,
-      "grad_norm": 0.011576223187148571,
-      "learning_rate": 0.0001999947164751735,
-      "loss": 46.0,
-      "step": 20329
-    },
-    {
-      "epoch": 3.2740045895567453,
-      "grad_norm": 0.0007288128254003823,
-      "learning_rate": 0.00019999471595510775,
-      "loss": 46.0,
-      "step": 20330
-    },
-    {
-      "epoch": 3.274165626635533,
-      "grad_norm": 0.00329859578050673,
-      "learning_rate": 0.00019999471543501635,
-      "loss": 46.0,
-      "step": 20331
-    },
-    {
-      "epoch": 3.2743266637143202,
-      "grad_norm": 0.008171114139258862,
-      "learning_rate": 0.0001999947149148994,
-      "loss": 46.0,
-      "step": 20332
-    },
-    {
-      "epoch": 3.2744877007931077,
-      "grad_norm": 0.0018043456366285682,
-      "learning_rate": 0.00019999471439475683,
-      "loss": 46.0,
-      "step": 20333
-    },
-    {
-      "epoch": 3.274648737871895,
-      "grad_norm": 0.006257224828004837,
-      "learning_rate": 0.00019999471387458867,
-      "loss": 46.0,
-      "step": 20334
-    },
-    {
-      "epoch": 3.2748097749506826,
-      "grad_norm": 0.004654455464333296,
-      "learning_rate": 0.00019999471335439495,
-      "loss": 46.0,
-      "step": 20335
-    },
-    {
-      "epoch": 3.2749708120294696,
-      "grad_norm": 0.002779880305752158,
-      "learning_rate": 0.00019999471283417558,
-      "loss": 46.0,
-      "step": 20336
-    },
-    {
-      "epoch": 3.275131849108257,
-      "grad_norm": 0.0016182982362806797,
-      "learning_rate": 0.00019999471231393066,
-      "loss": 46.0,
-      "step": 20337
-    },
-    {
-      "epoch": 3.2752928861870445,
-      "grad_norm": 0.008424478583037853,
-      "learning_rate": 0.00019999471179366012,
-      "loss": 46.0,
-      "step": 20338
-    },
-    {
-      "epoch": 3.275453923265832,
-      "grad_norm": 0.003382084658369422,
-      "learning_rate": 0.000199994711273364,
-      "loss": 46.0,
-      "step": 20339
-    },
-    {
-      "epoch": 3.2756149603446194,
-      "grad_norm": 0.0006228085840120912,
-      "learning_rate": 0.00019999471075304227,
-      "loss": 46.0,
-      "step": 20340
-    },
-    {
-      "epoch": 3.275775997423407,
-      "grad_norm": 0.0029146980959922075,
-      "learning_rate": 0.00019999471023269497,
-      "loss": 46.0,
-      "step": 20341
-    },
-    {
-      "epoch": 3.2759370345021943,
-      "grad_norm": 0.003615321358665824,
-      "learning_rate": 0.00019999470971232208,
-      "loss": 46.0,
-      "step": 20342
-    },
-    {
-      "epoch": 3.2760980715809813,
-      "grad_norm": 0.004053561948239803,
-      "learning_rate": 0.00019999470919192355,
-      "loss": 46.0,
-      "step": 20343
-    },
-    {
-      "epoch": 3.2762591086597688,
-      "grad_norm": 0.00173073576297611,
-      "learning_rate": 0.0001999947086714995,
-      "loss": 46.0,
-      "step": 20344
-    },
-    {
-      "epoch": 3.2764201457385562,
-      "grad_norm": 0.0049613104201853275,
-      "learning_rate": 0.0001999947081510498,
-      "loss": 46.0,
-      "step": 20345
-    },
-    {
-      "epoch": 3.2765811828173437,
-      "grad_norm": 0.0024762353859841824,
-      "learning_rate": 0.0001999947076305745,
-      "loss": 46.0,
-      "step": 20346
-    },
-    {
-      "epoch": 3.276742219896131,
-      "grad_norm": 0.015306377783417702,
-      "learning_rate": 0.00019999470711007367,
-      "loss": 46.0,
-      "step": 20347
-    },
-    {
-      "epoch": 3.2769032569749186,
-      "grad_norm": 0.0014777373289689422,
-      "learning_rate": 0.0001999947065895472,
-      "loss": 46.0,
-      "step": 20348
-    },
-    {
-      "epoch": 3.277064294053706,
-      "grad_norm": 0.0022636540234088898,
-      "learning_rate": 0.00019999470606899512,
-      "loss": 46.0,
-      "step": 20349
-    },
-    {
-      "epoch": 3.2772253311324935,
-      "grad_norm": 0.0004711646761279553,
-      "learning_rate": 0.00019999470554841748,
-      "loss": 46.0,
-      "step": 20350
-    },
-    {
-      "epoch": 3.2773863682112805,
-      "grad_norm": 0.0030130173545330763,
-      "learning_rate": 0.00019999470502781423,
-      "loss": 46.0,
-      "step": 20351
-    },
-    {
-      "epoch": 3.277547405290068,
-      "grad_norm": 0.0039952476508915424,
-      "learning_rate": 0.00019999470450718541,
-      "loss": 46.0,
-      "step": 20352
-    },
-    {
-      "epoch": 3.2777084423688554,
-      "grad_norm": 0.00043090531835332513,
-      "learning_rate": 0.00019999470398653096,
-      "loss": 46.0,
-      "step": 20353
-    },
-    {
-      "epoch": 3.277869479447643,
-      "grad_norm": 0.002169425366446376,
-      "learning_rate": 0.00019999470346585092,
-      "loss": 46.0,
-      "step": 20354
-    },
-    {
-      "epoch": 3.2780305165264303,
-      "grad_norm": 0.0063762301579117775,
-      "learning_rate": 0.0001999947029451453,
-      "loss": 46.0,
-      "step": 20355
-    },
-    {
-      "epoch": 3.2781915536052177,
-      "grad_norm": 0.0014614181127399206,
-      "learning_rate": 0.0001999947024244141,
-      "loss": 46.0,
-      "step": 20356
-    },
-    {
-      "epoch": 3.2783525906840048,
-      "grad_norm": 0.015582134015858173,
-      "learning_rate": 0.0001999947019036573,
-      "loss": 46.0,
-      "step": 20357
-    },
-    {
-      "epoch": 3.278513627762792,
-      "grad_norm": 0.001045424840413034,
-      "learning_rate": 0.0001999947013828749,
-      "loss": 46.0,
-      "step": 20358
-    },
-    {
-      "epoch": 3.2786746648415797,
-      "grad_norm": 0.006103664170950651,
-      "learning_rate": 0.0001999947008620669,
-      "loss": 46.0,
-      "step": 20359
-    },
-    {
-      "epoch": 3.278835701920367,
-      "grad_norm": 0.007698753383010626,
-      "learning_rate": 0.0001999947003412333,
-      "loss": 46.0,
-      "step": 20360
-    },
-    {
-      "epoch": 3.2789967389991546,
-      "grad_norm": 0.003799412166699767,
-      "learning_rate": 0.00019999469982037412,
-      "loss": 46.0,
-      "step": 20361
-    },
-    {
-      "epoch": 3.279157776077942,
-      "grad_norm": 0.0030289930291473866,
-      "learning_rate": 0.00019999469929948935,
-      "loss": 46.0,
-      "step": 20362
-    },
-    {
-      "epoch": 3.2793188131567295,
-      "grad_norm": 0.0024464610032737255,
-      "learning_rate": 0.000199994698778579,
-      "loss": 46.0,
-      "step": 20363
-    },
-    {
-      "epoch": 3.279479850235517,
-      "grad_norm": 0.0048452941700816154,
-      "learning_rate": 0.00019999469825764303,
-      "loss": 46.0,
-      "step": 20364
-    },
-    {
-      "epoch": 3.2796408873143044,
-      "grad_norm": 0.0014272892149165273,
-      "learning_rate": 0.00019999469773668147,
-      "loss": 46.0,
-      "step": 20365
-    },
-    {
-      "epoch": 3.2798019243930914,
-      "grad_norm": 0.004037264734506607,
-      "learning_rate": 0.00019999469721569433,
-      "loss": 46.0,
-      "step": 20366
-    },
-    {
-      "epoch": 3.279962961471879,
-      "grad_norm": 0.002270028693601489,
-      "learning_rate": 0.00019999469669468157,
-      "loss": 46.0,
-      "step": 20367
-    },
-    {
-      "epoch": 3.2801239985506663,
-      "grad_norm": 0.0011525989975780249,
-      "learning_rate": 0.00019999469617364323,
-      "loss": 46.0,
-      "step": 20368
-    },
-    {
-      "epoch": 3.2802850356294537,
-      "grad_norm": 0.015889935195446014,
-      "learning_rate": 0.00019999469565257932,
-      "loss": 46.0,
-      "step": 20369
-    },
-    {
-      "epoch": 3.280446072708241,
-      "grad_norm": 0.0024111676029860973,
-      "learning_rate": 0.0001999946951314898,
-      "loss": 46.0,
-      "step": 20370
-    },
-    {
-      "epoch": 3.2806071097870286,
-      "grad_norm": 0.005994080100208521,
-      "learning_rate": 0.00019999469461037467,
-      "loss": 46.0,
-      "step": 20371
-    },
-    {
-      "epoch": 3.2807681468658156,
-      "grad_norm": 0.002925029257312417,
-      "learning_rate": 0.00019999469408923397,
-      "loss": 46.0,
-      "step": 20372
-    },
-    {
-      "epoch": 3.280929183944603,
-      "grad_norm": 0.00670371251180768,
-      "learning_rate": 0.00019999469356806767,
-      "loss": 46.0,
-      "step": 20373
-    },
-    {
-      "epoch": 3.2810902210233905,
-      "grad_norm": 0.005363027565181255,
-      "learning_rate": 0.00019999469304687577,
-      "loss": 46.0,
-      "step": 20374
-    },
-    {
-      "epoch": 3.281251258102178,
-      "grad_norm": 0.005633344408124685,
-      "learning_rate": 0.0001999946925256583,
-      "loss": 46.0,
-      "step": 20375
-    },
-    {
-      "epoch": 3.2814122951809654,
-      "grad_norm": 0.00604011956602335,
-      "learning_rate": 0.00019999469200441522,
-      "loss": 46.0,
-      "step": 20376
-    },
-    {
-      "epoch": 3.281573332259753,
-      "grad_norm": 0.0011458328226581216,
-      "learning_rate": 0.00019999469148314654,
-      "loss": 46.0,
-      "step": 20377
-    },
-    {
-      "epoch": 3.2817343693385403,
-      "grad_norm": 0.007230387534946203,
-      "learning_rate": 0.00019999469096185224,
-      "loss": 46.0,
-      "step": 20378
-    },
-    {
-      "epoch": 3.281895406417328,
-      "grad_norm": 0.0023067896254360676,
-      "learning_rate": 0.00019999469044053238,
-      "loss": 46.0,
-      "step": 20379
-    },
-    {
-      "epoch": 3.282056443496115,
-      "grad_norm": 0.004255686420947313,
-      "learning_rate": 0.00019999468991918694,
-      "loss": 46.0,
-      "step": 20380
-    },
-    {
-      "epoch": 3.2822174805749023,
-      "grad_norm": 0.0031535234302282333,
-      "learning_rate": 0.00019999468939781588,
-      "loss": 46.0,
-      "step": 20381
-    },
-    {
-      "epoch": 3.2823785176536897,
-      "grad_norm": 0.00668123085051775,
-      "learning_rate": 0.00019999468887641923,
-      "loss": 46.0,
-      "step": 20382
-    },
-    {
-      "epoch": 3.282539554732477,
-      "grad_norm": 0.002975334180518985,
-      "learning_rate": 0.000199994688354997,
-      "loss": 46.0,
-      "step": 20383
-    },
-    {
-      "epoch": 3.2827005918112646,
-      "grad_norm": 0.00376451900228858,
-      "learning_rate": 0.00019999468783354915,
-      "loss": 46.0,
-      "step": 20384
-    },
-    {
-      "epoch": 3.282861628890052,
-      "grad_norm": 0.002300912281498313,
-      "learning_rate": 0.00019999468731207577,
-      "loss": 46.0,
-      "step": 20385
-    },
-    {
-      "epoch": 3.2830226659688395,
-      "grad_norm": 0.002907482208684087,
-      "learning_rate": 0.00019999468679057674,
-      "loss": 46.0,
-      "step": 20386
-    },
-    {
-      "epoch": 3.2831837030476265,
-      "grad_norm": 0.0014460888924077153,
-      "learning_rate": 0.0001999946862690521,
-      "loss": 46.0,
-      "step": 20387
-    },
-    {
-      "epoch": 3.283344740126414,
-      "grad_norm": 0.0010876660235226154,
-      "learning_rate": 0.0001999946857475019,
-      "loss": 46.0,
-      "step": 20388
-    },
-    {
-      "epoch": 3.2835057772052014,
-      "grad_norm": 0.0024834831710904837,
-      "learning_rate": 0.0001999946852259261,
-      "loss": 46.0,
-      "step": 20389
-    },
-    {
-      "epoch": 3.283666814283989,
-      "grad_norm": 0.0009719815570861101,
-      "learning_rate": 0.00019999468470432473,
-      "loss": 46.0,
-      "step": 20390
-    },
-    {
-      "epoch": 3.2838278513627763,
-      "grad_norm": 0.009112821891903877,
-      "learning_rate": 0.00019999468418269774,
-      "loss": 46.0,
-      "step": 20391
-    },
-    {
-      "epoch": 3.283988888441564,
-      "grad_norm": 0.0022545999381691217,
-      "learning_rate": 0.00019999468366104514,
-      "loss": 46.0,
-      "step": 20392
-    },
-    {
-      "epoch": 3.2841499255203512,
-      "grad_norm": 0.000750127190258354,
-      "learning_rate": 0.00019999468313936698,
-      "loss": 46.0,
-      "step": 20393
-    },
-    {
-      "epoch": 3.2843109625991387,
-      "grad_norm": 0.002832061145454645,
-      "learning_rate": 0.0001999946826176632,
-      "loss": 46.0,
-      "step": 20394
-    },
-    {
-      "epoch": 3.2844719996779257,
-      "grad_norm": 0.0018956970889121294,
-      "learning_rate": 0.00019999468209593385,
-      "loss": 46.0,
-      "step": 20395
-    },
-    {
-      "epoch": 3.284633036756713,
-      "grad_norm": 0.0022145244292914867,
-      "learning_rate": 0.0001999946815741789,
-      "loss": 46.0,
-      "step": 20396
-    },
-    {
-      "epoch": 3.2847940738355006,
-      "grad_norm": 0.011509504169225693,
-      "learning_rate": 0.00019999468105239837,
-      "loss": 46.0,
-      "step": 20397
-    },
-    {
-      "epoch": 3.284955110914288,
-      "grad_norm": 0.002674112329259515,
-      "learning_rate": 0.00019999468053059222,
-      "loss": 46.0,
-      "step": 20398
-    },
-    {
-      "epoch": 3.2851161479930755,
-      "grad_norm": 0.007826443761587143,
-      "learning_rate": 0.00019999468000876048,
-      "loss": 46.0,
-      "step": 20399
-    },
-    {
-      "epoch": 3.285277185071863,
-      "grad_norm": 0.0027850372716784477,
-      "learning_rate": 0.00019999467948690315,
-      "loss": 46.0,
-      "step": 20400
-    },
-    {
-      "epoch": 3.28543822215065,
-      "grad_norm": 0.003031131811439991,
-      "learning_rate": 0.00019999467896502022,
-      "loss": 46.0,
-      "step": 20401
-    },
-    {
-      "epoch": 3.2855992592294374,
-      "grad_norm": 0.0012760359095409513,
-      "learning_rate": 0.00019999467844311172,
-      "loss": 46.0,
-      "step": 20402
-    },
-    {
-      "epoch": 3.285760296308225,
-      "grad_norm": 0.007209960371255875,
-      "learning_rate": 0.0001999946779211776,
-      "loss": 46.0,
-      "step": 20403
-    },
-    {
-      "epoch": 3.2859213333870123,
-      "grad_norm": 0.0016973179299384356,
-      "learning_rate": 0.00019999467739921788,
-      "loss": 46.0,
-      "step": 20404
-    },
-    {
-      "epoch": 3.2860823704657998,
-      "grad_norm": 0.004956655669957399,
-      "learning_rate": 0.00019999467687723262,
-      "loss": 46.0,
-      "step": 20405
-    },
-    {
-      "epoch": 3.286243407544587,
-      "grad_norm": 0.0013060332275927067,
-      "learning_rate": 0.00019999467635522172,
-      "loss": 46.0,
-      "step": 20406
-    },
-    {
-      "epoch": 3.2864044446233747,
-      "grad_norm": 0.003149711061269045,
-      "learning_rate": 0.00019999467583318526,
-      "loss": 46.0,
-      "step": 20407
-    },
-    {
-      "epoch": 3.286565481702162,
-      "grad_norm": 0.000669082859531045,
-      "learning_rate": 0.00019999467531112318,
-      "loss": 46.0,
-      "step": 20408
-    },
-    {
-      "epoch": 3.286726518780949,
-      "grad_norm": 0.004308716394007206,
-      "learning_rate": 0.00019999467478903552,
-      "loss": 46.0,
-      "step": 20409
-    },
-    {
-      "epoch": 3.2868875558597366,
-      "grad_norm": 0.0035340306349098682,
-      "learning_rate": 0.0001999946742669222,
-      "loss": 46.0,
-      "step": 20410
-    },
-    {
-      "epoch": 3.287048592938524,
-      "grad_norm": 0.008656601421535015,
-      "learning_rate": 0.00019999467374478338,
-      "loss": 46.0,
-      "step": 20411
-    },
-    {
-      "epoch": 3.2872096300173115,
-      "grad_norm": 0.004896919708698988,
-      "learning_rate": 0.00019999467322261892,
-      "loss": 46.0,
-      "step": 20412
-    },
-    {
-      "epoch": 3.287370667096099,
-      "grad_norm": 0.001092921826057136,
-      "learning_rate": 0.00019999467270042889,
-      "loss": 46.0,
-      "step": 20413
-    },
-    {
-      "epoch": 3.2875317041748864,
-      "grad_norm": 0.0023733729030936956,
-      "learning_rate": 0.00019999467217821323,
-      "loss": 46.0,
-      "step": 20414
-    },
-    {
-      "epoch": 3.287692741253674,
-      "grad_norm": 0.012586906552314758,
-      "learning_rate": 0.00019999467165597202,
-      "loss": 46.0,
-      "step": 20415
-    },
-    {
-      "epoch": 3.287853778332461,
-      "grad_norm": 0.00931586418300867,
-      "learning_rate": 0.0001999946711337052,
-      "loss": 46.0,
-      "step": 20416
-    },
-    {
-      "epoch": 3.2880148154112483,
-      "grad_norm": 0.00513248797506094,
-      "learning_rate": 0.00019999467061141278,
-      "loss": 46.0,
-      "step": 20417
-    },
-    {
-      "epoch": 3.2881758524900357,
-      "grad_norm": 0.0017419853247702122,
-      "learning_rate": 0.00019999467008909475,
-      "loss": 46.0,
-      "step": 20418
-    },
-    {
-      "epoch": 3.288336889568823,
-      "grad_norm": 0.0012986648362129927,
-      "learning_rate": 0.00019999466956675116,
-      "loss": 46.0,
-      "step": 20419
-    },
-    {
-      "epoch": 3.2884979266476106,
-      "grad_norm": 0.005398717243224382,
-      "learning_rate": 0.00019999466904438196,
-      "loss": 46.0,
-      "step": 20420
-    },
-    {
-      "epoch": 3.288658963726398,
-      "grad_norm": 0.0048441351391375065,
-      "learning_rate": 0.00019999466852198717,
-      "loss": 46.0,
-      "step": 20421
-    },
-    {
-      "epoch": 3.2888200008051856,
-      "grad_norm": 0.0017884616972878575,
-      "learning_rate": 0.0001999946679995668,
-      "loss": 46.0,
-      "step": 20422
-    },
-    {
-      "epoch": 3.288981037883973,
-      "grad_norm": 0.003438309533521533,
-      "learning_rate": 0.0001999946674771208,
-      "loss": 46.0,
-      "step": 20423
-    },
-    {
-      "epoch": 3.28914207496276,
-      "grad_norm": 0.010533241555094719,
-      "learning_rate": 0.00019999466695464925,
-      "loss": 46.0,
-      "step": 20424
-    },
-    {
-      "epoch": 3.2893031120415475,
-      "grad_norm": 0.0010520125506445765,
-      "learning_rate": 0.00019999466643215205,
-      "loss": 46.0,
-      "step": 20425
-    },
-    {
-      "epoch": 3.289464149120335,
-      "grad_norm": 0.010319353081285954,
-      "learning_rate": 0.00019999466590962933,
-      "loss": 46.0,
-      "step": 20426
-    },
-    {
-      "epoch": 3.2896251861991224,
-      "grad_norm": 0.0007083070231601596,
-      "learning_rate": 0.00019999466538708096,
-      "loss": 46.0,
-      "step": 20427
-    },
-    {
-      "epoch": 3.28978622327791,
-      "grad_norm": 0.005335061810910702,
-      "learning_rate": 0.00019999466486450703,
-      "loss": 46.0,
-      "step": 20428
-    },
-    {
-      "epoch": 3.2899472603566973,
-      "grad_norm": 0.0022226402070373297,
-      "learning_rate": 0.0001999946643419075,
-      "loss": 46.0,
-      "step": 20429
-    },
-    {
-      "epoch": 3.2901082974354843,
-      "grad_norm": 0.0023903241381049156,
-      "learning_rate": 0.00019999466381928236,
-      "loss": 46.0,
-      "step": 20430
-    },
-    {
-      "epoch": 3.2902693345142717,
-      "grad_norm": 0.0005466773873195052,
-      "learning_rate": 0.00019999466329663162,
-      "loss": 46.0,
-      "step": 20431
-    },
-    {
-      "epoch": 3.290430371593059,
-      "grad_norm": 0.0009620157652534544,
-      "learning_rate": 0.0001999946627739553,
-      "loss": 46.0,
-      "step": 20432
-    },
-    {
-      "epoch": 3.2905914086718466,
-      "grad_norm": 0.004091755487024784,
-      "learning_rate": 0.00019999466225125337,
-      "loss": 46.0,
-      "step": 20433
-    },
-    {
-      "epoch": 3.290752445750634,
-      "grad_norm": 0.002535921987146139,
-      "learning_rate": 0.00019999466172852587,
-      "loss": 46.0,
-      "step": 20434
-    },
-    {
-      "epoch": 3.2909134828294215,
-      "grad_norm": 0.0007435278384946287,
-      "learning_rate": 0.00019999466120577277,
-      "loss": 46.0,
-      "step": 20435
-    },
-    {
-      "epoch": 3.291074519908209,
-      "grad_norm": 0.001149357995018363,
-      "learning_rate": 0.0001999946606829941,
-      "loss": 46.0,
-      "step": 20436
-    },
-    {
-      "epoch": 3.2912355569869964,
-      "grad_norm": 0.003767345566302538,
-      "learning_rate": 0.0001999946601601898,
-      "loss": 46.0,
-      "step": 20437
-    },
-    {
-      "epoch": 3.291396594065784,
-      "grad_norm": 0.0013400643365457654,
-      "learning_rate": 0.0001999946596373599,
-      "loss": 46.0,
-      "step": 20438
-    },
-    {
-      "epoch": 3.291557631144571,
-      "grad_norm": 0.0023799403570592403,
-      "learning_rate": 0.00019999465911450445,
-      "loss": 46.0,
-      "step": 20439
-    },
-    {
-      "epoch": 3.2917186682233583,
-      "grad_norm": 0.0017399901989847422,
-      "learning_rate": 0.00019999465859162337,
-      "loss": 46.0,
-      "step": 20440
-    },
-    {
-      "epoch": 3.291879705302146,
-      "grad_norm": 0.0016450092662125826,
-      "learning_rate": 0.0001999946580687167,
-      "loss": 46.0,
-      "step": 20441
-    },
-    {
-      "epoch": 3.2920407423809332,
-      "grad_norm": 0.0027623374480754137,
-      "learning_rate": 0.00019999465754578445,
-      "loss": 46.0,
-      "step": 20442
-    },
-    {
-      "epoch": 3.2922017794597207,
-      "grad_norm": 0.009613106027245522,
-      "learning_rate": 0.0001999946570228266,
-      "loss": 46.0,
-      "step": 20443
-    },
-    {
-      "epoch": 3.292362816538508,
-      "grad_norm": 0.008135140873491764,
-      "learning_rate": 0.00019999465649984317,
-      "loss": 46.0,
-      "step": 20444
-    },
-    {
-      "epoch": 3.292523853617295,
-      "grad_norm": 0.009226481430232525,
-      "learning_rate": 0.00019999465597683413,
-      "loss": 46.0,
-      "step": 20445
-    },
-    {
-      "epoch": 3.2926848906960826,
-      "grad_norm": 0.002246605232357979,
-      "learning_rate": 0.00019999465545379952,
-      "loss": 46.0,
-      "step": 20446
-    },
-    {
-      "epoch": 3.29284592777487,
-      "grad_norm": 0.002180477837100625,
-      "learning_rate": 0.00019999465493073928,
-      "loss": 46.0,
-      "step": 20447
-    },
-    {
-      "epoch": 3.2930069648536575,
-      "grad_norm": 0.012579609639942646,
-      "learning_rate": 0.00019999465440765347,
-      "loss": 46.0,
-      "step": 20448
-    },
-    {
-      "epoch": 3.293168001932445,
-      "grad_norm": 0.001363204326480627,
-      "learning_rate": 0.00019999465388454205,
-      "loss": 46.0,
-      "step": 20449
-    },
-    {
-      "epoch": 3.2933290390112324,
-      "grad_norm": 0.0010487326653674245,
-      "learning_rate": 0.00019999465336140507,
-      "loss": 46.0,
-      "step": 20450
-    },
-    {
-      "epoch": 3.29349007609002,
-      "grad_norm": 0.002048888709396124,
-      "learning_rate": 0.00019999465283824245,
-      "loss": 46.0,
-      "step": 20451
-    },
-    {
-      "epoch": 3.2936511131688073,
-      "grad_norm": 0.007130644749850035,
-      "learning_rate": 0.00019999465231505426,
-      "loss": 46.0,
-      "step": 20452
-    },
-    {
-      "epoch": 3.2938121502475943,
-      "grad_norm": 0.0039036625530570745,
-      "learning_rate": 0.00019999465179184047,
-      "loss": 46.0,
-      "step": 20453
-    },
-    {
-      "epoch": 3.293973187326382,
-      "grad_norm": 0.0017658513970673084,
-      "learning_rate": 0.0001999946512686011,
-      "loss": 46.0,
-      "step": 20454
-    },
-    {
-      "epoch": 3.2941342244051692,
-      "grad_norm": 0.001026177080348134,
-      "learning_rate": 0.00019999465074533614,
-      "loss": 46.0,
-      "step": 20455
-    },
-    {
-      "epoch": 3.2942952614839567,
-      "grad_norm": 0.0037942954804748297,
-      "learning_rate": 0.00019999465022204558,
-      "loss": 46.0,
-      "step": 20456
-    },
-    {
-      "epoch": 3.294456298562744,
-      "grad_norm": 0.009627923369407654,
-      "learning_rate": 0.0001999946496987294,
-      "loss": 46.0,
-      "step": 20457
-    },
-    {
-      "epoch": 3.2946173356415316,
-      "grad_norm": 0.004906713031232357,
-      "learning_rate": 0.00019999464917538768,
-      "loss": 46.0,
-      "step": 20458
-    },
-    {
-      "epoch": 3.294778372720319,
-      "grad_norm": 0.0030307569541037083,
-      "learning_rate": 0.00019999464865202033,
-      "loss": 46.0,
-      "step": 20459
-    },
-    {
-      "epoch": 3.294939409799106,
-      "grad_norm": 0.005650836508721113,
-      "learning_rate": 0.00019999464812862737,
-      "loss": 46.0,
-      "step": 20460
-    },
-    {
-      "epoch": 3.2951004468778935,
-      "grad_norm": 0.0017931369366124272,
-      "learning_rate": 0.00019999464760520885,
-      "loss": 46.0,
-      "step": 20461
-    },
-    {
-      "epoch": 3.295261483956681,
-      "grad_norm": 0.0019347493071109056,
-      "learning_rate": 0.0001999946470817647,
-      "loss": 46.0,
-      "step": 20462
-    },
-    {
-      "epoch": 3.2954225210354684,
-      "grad_norm": 0.003170775016769767,
-      "learning_rate": 0.00019999464655829502,
-      "loss": 46.0,
-      "step": 20463
-    },
-    {
-      "epoch": 3.295583558114256,
-      "grad_norm": 0.0015064164763316512,
-      "learning_rate": 0.0001999946460347997,
-      "loss": 46.0,
-      "step": 20464
-    },
-    {
-      "epoch": 3.2957445951930433,
-      "grad_norm": 0.009030257351696491,
-      "learning_rate": 0.0001999946455112788,
-      "loss": 46.0,
-      "step": 20465
-    },
-    {
-      "epoch": 3.2959056322718308,
-      "grad_norm": 0.006091805640608072,
-      "learning_rate": 0.0001999946449877323,
-      "loss": 46.0,
-      "step": 20466
-    },
-    {
-      "epoch": 3.296066669350618,
-      "grad_norm": 0.001362030627205968,
-      "learning_rate": 0.0001999946444641602,
-      "loss": 46.0,
-      "step": 20467
-    },
-    {
-      "epoch": 3.296227706429405,
-      "grad_norm": 0.0007975163171067834,
-      "learning_rate": 0.00019999464394056254,
-      "loss": 46.0,
-      "step": 20468
-    },
-    {
-      "epoch": 3.2963887435081927,
-      "grad_norm": 0.0028633701149374247,
-      "learning_rate": 0.00019999464341693924,
-      "loss": 46.0,
-      "step": 20469
-    },
-    {
-      "epoch": 3.29654978058698,
-      "grad_norm": 0.0022130045108497143,
-      "learning_rate": 0.00019999464289329036,
-      "loss": 46.0,
-      "step": 20470
-    },
-    {
-      "epoch": 3.2967108176657676,
-      "grad_norm": 0.002059884835034609,
-      "learning_rate": 0.0001999946423696159,
-      "loss": 46.0,
-      "step": 20471
-    },
-    {
-      "epoch": 3.296871854744555,
-      "grad_norm": 0.0026290572714060545,
-      "learning_rate": 0.00019999464184591585,
-      "loss": 46.0,
-      "step": 20472
-    },
-    {
-      "epoch": 3.2970328918233425,
-      "grad_norm": 0.0024627484381198883,
-      "learning_rate": 0.0001999946413221902,
-      "loss": 46.0,
-      "step": 20473
-    },
-    {
-      "epoch": 3.2971939289021295,
-      "grad_norm": 0.0047419723123312,
-      "learning_rate": 0.00019999464079843894,
-      "loss": 46.0,
-      "step": 20474
-    },
-    {
-      "epoch": 3.297354965980917,
-      "grad_norm": 0.0022318412084132433,
-      "learning_rate": 0.00019999464027466212,
-      "loss": 46.0,
-      "step": 20475
-    },
-    {
-      "epoch": 3.2975160030597044,
-      "grad_norm": 0.012493956834077835,
-      "learning_rate": 0.00019999463975085968,
-      "loss": 46.0,
-      "step": 20476
-    },
-    {
-      "epoch": 3.297677040138492,
-      "grad_norm": 0.001505105523392558,
-      "learning_rate": 0.00019999463922703163,
-      "loss": 46.0,
-      "step": 20477
-    },
-    {
-      "epoch": 3.2978380772172793,
-      "grad_norm": 0.0020163014996796846,
-      "learning_rate": 0.00019999463870317802,
-      "loss": 46.0,
-      "step": 20478
-    },
-    {
-      "epoch": 3.2979991142960667,
-      "grad_norm": 0.004755750764161348,
-      "learning_rate": 0.0001999946381792988,
-      "loss": 46.0,
-      "step": 20479
-    },
-    {
-      "epoch": 3.298160151374854,
-      "grad_norm": 0.0021764670964330435,
-      "learning_rate": 0.000199994637655394,
-      "loss": 46.0,
-      "step": 20480
-    },
-    {
-      "epoch": 3.2983211884536416,
-      "grad_norm": 0.002002222230657935,
-      "learning_rate": 0.0001999946371314636,
-      "loss": 46.0,
-      "step": 20481
-    },
-    {
-      "epoch": 3.298482225532429,
-      "grad_norm": 0.006661428604274988,
-      "learning_rate": 0.00019999463660750762,
-      "loss": 46.0,
-      "step": 20482
-    },
-    {
-      "epoch": 3.298643262611216,
-      "grad_norm": 0.007542894221842289,
-      "learning_rate": 0.000199994636083526,
-      "loss": 46.0,
-      "step": 20483
-    },
-    {
-      "epoch": 3.2988042996900035,
-      "grad_norm": 0.015227694995701313,
-      "learning_rate": 0.00019999463555951883,
-      "loss": 46.0,
-      "step": 20484
-    },
-    {
-      "epoch": 3.298965336768791,
-      "grad_norm": 0.012763544917106628,
-      "learning_rate": 0.00019999463503548605,
-      "loss": 46.0,
-      "step": 20485
-    },
-    {
-      "epoch": 3.2991263738475785,
-      "grad_norm": 0.0019030938856303692,
-      "learning_rate": 0.0001999946345114277,
-      "loss": 46.0,
-      "step": 20486
-    },
-    {
-      "epoch": 3.299287410926366,
-      "grad_norm": 0.003876202739775181,
-      "learning_rate": 0.00019999463398734374,
-      "loss": 46.0,
-      "step": 20487
-    },
-    {
-      "epoch": 3.2994484480051534,
-      "grad_norm": 0.002691423986107111,
-      "learning_rate": 0.00019999463346323418,
-      "loss": 46.0,
-      "step": 20488
-    },
-    {
-      "epoch": 3.2996094850839404,
-      "grad_norm": 0.0016912458231672645,
-      "learning_rate": 0.00019999463293909902,
-      "loss": 46.0,
-      "step": 20489
-    },
-    {
-      "epoch": 3.299770522162728,
-      "grad_norm": 0.0032249304931610823,
-      "learning_rate": 0.00019999463241493828,
-      "loss": 46.0,
-      "step": 20490
-    },
-    {
-      "epoch": 3.2999315592415153,
-      "grad_norm": 0.004676357842981815,
-      "learning_rate": 0.00019999463189075193,
-      "loss": 46.0,
-      "step": 20491
-    },
-    {
-      "epoch": 3.3000925963203027,
-      "grad_norm": 0.012644380331039429,
-      "learning_rate": 0.00019999463136654,
-      "loss": 46.0,
-      "step": 20492
-    },
-    {
-      "epoch": 3.30025363339909,
-      "grad_norm": 0.002780510112643242,
-      "learning_rate": 0.0001999946308423025,
-      "loss": 46.0,
-      "step": 20493
-    },
-    {
-      "epoch": 3.3004146704778776,
-      "grad_norm": 0.005221215542405844,
-      "learning_rate": 0.00019999463031803938,
-      "loss": 46.0,
-      "step": 20494
-    },
-    {
-      "epoch": 3.300575707556665,
-      "grad_norm": 0.0029816976748406887,
-      "learning_rate": 0.00019999462979375068,
-      "loss": 46.0,
-      "step": 20495
-    },
-    {
-      "epoch": 3.3007367446354525,
-      "grad_norm": 0.006407435983419418,
-      "learning_rate": 0.00019999462926943636,
-      "loss": 46.0,
-      "step": 20496
-    },
-    {
-      "epoch": 3.3008977817142395,
-      "grad_norm": 0.006033220328390598,
-      "learning_rate": 0.00019999462874509646,
-      "loss": 46.0,
-      "step": 20497
-    },
-    {
-      "epoch": 3.301058818793027,
-      "grad_norm": 0.008315898478031158,
-      "learning_rate": 0.00019999462822073097,
-      "loss": 46.0,
-      "step": 20498
-    },
-    {
-      "epoch": 3.3012198558718144,
-      "grad_norm": 0.004438674543052912,
-      "learning_rate": 0.0001999946276963399,
-      "loss": 46.0,
-      "step": 20499
-    },
-    {
-      "epoch": 3.301380892950602,
-      "grad_norm": 0.0006123550701886415,
-      "learning_rate": 0.0001999946271719232,
-      "loss": 46.0,
-      "step": 20500
-    },
-    {
-      "epoch": 3.3015419300293893,
-      "grad_norm": 0.0035484223626554012,
-      "learning_rate": 0.00019999462664748095,
-      "loss": 46.0,
-      "step": 20501
-    },
-    {
-      "epoch": 3.301702967108177,
-      "grad_norm": 0.0021711597219109535,
-      "learning_rate": 0.00019999462612301308,
-      "loss": 46.0,
-      "step": 20502
-    },
-    {
-      "epoch": 3.3018640041869642,
-      "grad_norm": 0.0013619862729683518,
-      "learning_rate": 0.00019999462559851963,
-      "loss": 46.0,
-      "step": 20503
-    },
-    {
-      "epoch": 3.3020250412657512,
-      "grad_norm": 0.004725265316665173,
-      "learning_rate": 0.00019999462507400059,
-      "loss": 46.0,
-      "step": 20504
-    },
-    {
-      "epoch": 3.3021860783445387,
-      "grad_norm": 0.011984646320343018,
-      "learning_rate": 0.00019999462454945593,
-      "loss": 46.0,
-      "step": 20505
-    },
-    {
-      "epoch": 3.302347115423326,
-      "grad_norm": 0.0004883562796749175,
-      "learning_rate": 0.0001999946240248857,
-      "loss": 46.0,
-      "step": 20506
-    },
-    {
-      "epoch": 3.3025081525021136,
-      "grad_norm": 0.002777355955913663,
-      "learning_rate": 0.00019999462350028986,
-      "loss": 46.0,
-      "step": 20507
-    },
-    {
-      "epoch": 3.302669189580901,
-      "grad_norm": 0.01274262834340334,
-      "learning_rate": 0.00019999462297566842,
-      "loss": 46.0,
-      "step": 20508
-    },
-    {
-      "epoch": 3.3028302266596885,
-      "grad_norm": 0.0017357562901452184,
-      "learning_rate": 0.0001999946224510214,
-      "loss": 46.0,
-      "step": 20509
-    },
-    {
-      "epoch": 3.302991263738476,
-      "grad_norm": 0.012472463771700859,
-      "learning_rate": 0.0001999946219263488,
-      "loss": 46.0,
-      "step": 20510
-    },
-    {
-      "epoch": 3.3031523008172634,
-      "grad_norm": 0.0036122689489275217,
-      "learning_rate": 0.0001999946214016506,
-      "loss": 46.0,
-      "step": 20511
-    },
-    {
-      "epoch": 3.3033133378960504,
-      "grad_norm": 0.0019842092879116535,
-      "learning_rate": 0.00019999462087692677,
-      "loss": 46.0,
-      "step": 20512
-    },
-    {
-      "epoch": 3.303474374974838,
-      "grad_norm": 0.0065961251966655254,
-      "learning_rate": 0.0001999946203521774,
-      "loss": 46.0,
-      "step": 20513
-    },
-    {
-      "epoch": 3.3036354120536253,
-      "grad_norm": 0.01669849455356598,
-      "learning_rate": 0.0001999946198274024,
-      "loss": 46.0,
-      "step": 20514
-    },
-    {
-      "epoch": 3.3037964491324128,
-      "grad_norm": 0.0029924034606665373,
-      "learning_rate": 0.00019999461930260185,
-      "loss": 46.0,
-      "step": 20515
-    },
-    {
-      "epoch": 3.3039574862112002,
-      "grad_norm": 0.0035890196450054646,
-      "learning_rate": 0.00019999461877777565,
-      "loss": 46.0,
-      "step": 20516
-    },
-    {
-      "epoch": 3.3041185232899877,
-      "grad_norm": 0.008518245071172714,
-      "learning_rate": 0.0001999946182529239,
-      "loss": 46.0,
-      "step": 20517
-    },
-    {
-      "epoch": 3.3042795603687747,
-      "grad_norm": 0.005013282876461744,
-      "learning_rate": 0.00019999461772804653,
-      "loss": 46.0,
-      "step": 20518
-    },
-    {
-      "epoch": 3.304440597447562,
-      "grad_norm": 0.005528761073946953,
-      "learning_rate": 0.0001999946172031436,
-      "loss": 46.0,
-      "step": 20519
-    },
-    {
-      "epoch": 3.3046016345263496,
-      "grad_norm": 0.0031828877981752157,
-      "learning_rate": 0.00019999461667821505,
-      "loss": 46.0,
-      "step": 20520
-    },
-    {
-      "epoch": 3.304762671605137,
-      "grad_norm": 0.00280867051333189,
-      "learning_rate": 0.00019999461615326087,
-      "loss": 46.0,
-      "step": 20521
-    },
-    {
-      "epoch": 3.3049237086839245,
-      "grad_norm": 0.007805526256561279,
-      "learning_rate": 0.00019999461562828118,
-      "loss": 46.0,
-      "step": 20522
-    },
-    {
-      "epoch": 3.305084745762712,
-      "grad_norm": 0.004497086629271507,
-      "learning_rate": 0.00019999461510327585,
-      "loss": 46.0,
-      "step": 20523
-    },
-    {
-      "epoch": 3.3052457828414994,
-      "grad_norm": 0.0031976636964827776,
-      "learning_rate": 0.0001999946145782449,
-      "loss": 46.0,
-      "step": 20524
-    },
-    {
-      "epoch": 3.305406819920287,
-      "grad_norm": 0.0019408968510106206,
-      "learning_rate": 0.0001999946140531884,
-      "loss": 46.0,
-      "step": 20525
-    },
-    {
-      "epoch": 3.305567856999074,
-      "grad_norm": 0.004233045503497124,
-      "learning_rate": 0.0001999946135281063,
-      "loss": 46.0,
-      "step": 20526
-    },
-    {
-      "epoch": 3.3057288940778613,
-      "grad_norm": 0.0015091991517692804,
-      "learning_rate": 0.00019999461300299856,
-      "loss": 46.0,
-      "step": 20527
-    },
-    {
-      "epoch": 3.3058899311566488,
-      "grad_norm": 0.008871247991919518,
-      "learning_rate": 0.0001999946124778653,
-      "loss": 46.0,
-      "step": 20528
-    },
-    {
-      "epoch": 3.306050968235436,
-      "grad_norm": 0.0023937553633004427,
-      "learning_rate": 0.0001999946119527064,
-      "loss": 46.0,
-      "step": 20529
-    },
-    {
-      "epoch": 3.3062120053142237,
-      "grad_norm": 0.0022780275903642178,
-      "learning_rate": 0.00019999461142752192,
-      "loss": 46.0,
-      "step": 20530
-    },
-    {
-      "epoch": 3.306373042393011,
-      "grad_norm": 0.00475579546764493,
-      "learning_rate": 0.0001999946109023118,
-      "loss": 46.0,
-      "step": 20531
-    },
-    {
-      "epoch": 3.3065340794717986,
-      "grad_norm": 0.0043720765970647335,
-      "learning_rate": 0.00019999461037707616,
-      "loss": 46.0,
-      "step": 20532
-    },
-    {
-      "epoch": 3.3066951165505856,
-      "grad_norm": 0.000885319896042347,
-      "learning_rate": 0.0001999946098518149,
-      "loss": 46.0,
-      "step": 20533
-    },
-    {
-      "epoch": 3.306856153629373,
-      "grad_norm": 0.0032566478475928307,
-      "learning_rate": 0.00019999460932652806,
-      "loss": 46.0,
-      "step": 20534
-    },
-    {
-      "epoch": 3.3070171907081605,
-      "grad_norm": 0.003298891708254814,
-      "learning_rate": 0.0001999946088012156,
-      "loss": 46.0,
-      "step": 20535
-    },
-    {
-      "epoch": 3.307178227786948,
-      "grad_norm": 0.0016318131238222122,
-      "learning_rate": 0.00019999460827587755,
-      "loss": 46.0,
-      "step": 20536
-    },
-    {
-      "epoch": 3.3073392648657354,
-      "grad_norm": 0.00962746050208807,
-      "learning_rate": 0.00019999460775051392,
-      "loss": 46.0,
-      "step": 20537
-    },
-    {
-      "epoch": 3.307500301944523,
-      "grad_norm": 0.003954700194299221,
-      "learning_rate": 0.0001999946072251247,
-      "loss": 46.0,
-      "step": 20538
-    },
-    {
-      "epoch": 3.3076613390233103,
-      "grad_norm": 0.00856354646384716,
-      "learning_rate": 0.00019999460669970986,
-      "loss": 46.0,
-      "step": 20539
-    },
-    {
-      "epoch": 3.3078223761020977,
-      "grad_norm": 0.0009719360386952758,
-      "learning_rate": 0.00019999460617426947,
-      "loss": 46.0,
-      "step": 20540
-    },
-    {
-      "epoch": 3.3079834131808847,
-      "grad_norm": 0.01439482718706131,
-      "learning_rate": 0.00019999460564880343,
-      "loss": 46.0,
-      "step": 20541
-    },
-    {
-      "epoch": 3.308144450259672,
-      "grad_norm": 0.004348213318735361,
-      "learning_rate": 0.00019999460512331183,
-      "loss": 46.0,
-      "step": 20542
-    },
-    {
-      "epoch": 3.3083054873384596,
-      "grad_norm": 0.002071383874863386,
-      "learning_rate": 0.00019999460459779465,
-      "loss": 46.0,
-      "step": 20543
-    },
-    {
-      "epoch": 3.308466524417247,
-      "grad_norm": 0.0018467125482857227,
-      "learning_rate": 0.00019999460407225182,
-      "loss": 46.0,
-      "step": 20544
-    },
-    {
-      "epoch": 3.3086275614960345,
-      "grad_norm": 0.002804292133077979,
-      "learning_rate": 0.00019999460354668344,
-      "loss": 46.0,
-      "step": 20545
-    },
-    {
-      "epoch": 3.308788598574822,
-      "grad_norm": 0.000851544551551342,
-      "learning_rate": 0.00019999460302108947,
-      "loss": 46.0,
-      "step": 20546
-    },
-    {
-      "epoch": 3.308949635653609,
-      "grad_norm": 0.007816920056939125,
-      "learning_rate": 0.0001999946024954699,
-      "loss": 46.0,
-      "step": 20547
-    },
-    {
-      "epoch": 3.3091106727323965,
-      "grad_norm": 0.001477054669521749,
-      "learning_rate": 0.00019999460196982476,
-      "loss": 46.0,
-      "step": 20548
-    },
-    {
-      "epoch": 3.309271709811184,
-      "grad_norm": 0.001662776106968522,
-      "learning_rate": 0.00019999460144415397,
-      "loss": 46.0,
-      "step": 20549
-    },
-    {
-      "epoch": 3.3094327468899714,
-      "grad_norm": 0.006226846016943455,
-      "learning_rate": 0.00019999460091845762,
-      "loss": 46.0,
-      "step": 20550
-    },
-    {
-      "epoch": 3.309593783968759,
-      "grad_norm": 0.001959709683433175,
-      "learning_rate": 0.0001999946003927357,
-      "loss": 46.0,
-      "step": 20551
-    },
-    {
-      "epoch": 3.3097548210475463,
-      "grad_norm": 0.0023772441782057285,
-      "learning_rate": 0.00019999459986698814,
-      "loss": 46.0,
-      "step": 20552
-    },
-    {
-      "epoch": 3.3099158581263337,
-      "grad_norm": 0.0011946274898946285,
-      "learning_rate": 0.00019999459934121503,
-      "loss": 46.0,
-      "step": 20553
-    },
-    {
-      "epoch": 3.310076895205121,
-      "grad_norm": 0.006707313004881144,
-      "learning_rate": 0.0001999945988154163,
-      "loss": 46.0,
-      "step": 20554
-    },
-    {
-      "epoch": 3.3102379322839086,
-      "grad_norm": 0.006173006258904934,
-      "learning_rate": 0.00019999459828959196,
-      "loss": 46.0,
-      "step": 20555
-    },
-    {
-      "epoch": 3.3103989693626956,
-      "grad_norm": 0.004015904385596514,
-      "learning_rate": 0.00019999459776374207,
-      "loss": 46.0,
-      "step": 20556
-    },
-    {
-      "epoch": 3.310560006441483,
-      "grad_norm": 0.00606947997584939,
-      "learning_rate": 0.00019999459723786655,
-      "loss": 46.0,
-      "step": 20557
-    },
-    {
-      "epoch": 3.3107210435202705,
-      "grad_norm": 0.009014831855893135,
-      "learning_rate": 0.00019999459671196545,
-      "loss": 46.0,
-      "step": 20558
-    },
-    {
-      "epoch": 3.310882080599058,
-      "grad_norm": 0.005148408003151417,
-      "learning_rate": 0.00019999459618603874,
-      "loss": 46.0,
-      "step": 20559
-    },
-    {
-      "epoch": 3.3110431176778454,
-      "grad_norm": 0.0055821118876338005,
-      "learning_rate": 0.00019999459566008646,
-      "loss": 46.0,
-      "step": 20560
-    },
-    {
-      "epoch": 3.311204154756633,
-      "grad_norm": 0.0021294988691806793,
-      "learning_rate": 0.00019999459513410858,
-      "loss": 46.0,
-      "step": 20561
-    },
-    {
-      "epoch": 3.31136519183542,
-      "grad_norm": 0.0019224905408918858,
-      "learning_rate": 0.00019999459460810513,
-      "loss": 46.0,
-      "step": 20562
-    },
-    {
-      "epoch": 3.3115262289142073,
-      "grad_norm": 0.005129345692694187,
-      "learning_rate": 0.00019999459408207604,
-      "loss": 46.0,
-      "step": 20563
-    },
-    {
-      "epoch": 3.311687265992995,
-      "grad_norm": 0.010406536050140858,
-      "learning_rate": 0.00019999459355602136,
-      "loss": 46.0,
-      "step": 20564
-    },
-    {
-      "epoch": 3.3118483030717822,
-      "grad_norm": 0.0063169002532958984,
-      "learning_rate": 0.00019999459302994112,
-      "loss": 46.0,
-      "step": 20565
-    },
-    {
-      "epoch": 3.3120093401505697,
-      "grad_norm": 0.004996041301637888,
-      "learning_rate": 0.0001999945925038353,
-      "loss": 46.0,
-      "step": 20566
-    },
-    {
-      "epoch": 3.312170377229357,
-      "grad_norm": 0.00403920654207468,
-      "learning_rate": 0.00019999459197770386,
-      "loss": 46.0,
-      "step": 20567
-    },
-    {
-      "epoch": 3.3123314143081446,
-      "grad_norm": 0.0018020266434177756,
-      "learning_rate": 0.00019999459145154683,
-      "loss": 46.0,
-      "step": 20568
-    },
-    {
-      "epoch": 3.312492451386932,
-      "grad_norm": 0.0040070051327347755,
-      "learning_rate": 0.00019999459092536417,
-      "loss": 46.0,
-      "step": 20569
-    },
-    {
-      "epoch": 3.312653488465719,
-      "grad_norm": 0.005236563738435507,
-      "learning_rate": 0.00019999459039915597,
-      "loss": 46.0,
-      "step": 20570
-    },
-    {
-      "epoch": 3.3128145255445065,
-      "grad_norm": 0.003927438519895077,
-      "learning_rate": 0.00019999458987292215,
-      "loss": 46.0,
-      "step": 20571
-    },
-    {
-      "epoch": 3.312975562623294,
-      "grad_norm": 0.014588958583772182,
-      "learning_rate": 0.00019999458934666272,
-      "loss": 46.0,
-      "step": 20572
-    },
-    {
-      "epoch": 3.3131365997020814,
-      "grad_norm": 0.002326664514839649,
-      "learning_rate": 0.0001999945888203777,
-      "loss": 46.0,
-      "step": 20573
-    },
-    {
-      "epoch": 3.313297636780869,
-      "grad_norm": 0.0018098813015967607,
-      "learning_rate": 0.00019999458829406713,
-      "loss": 46.0,
-      "step": 20574
-    },
-    {
-      "epoch": 3.3134586738596563,
-      "grad_norm": 0.003908189013600349,
-      "learning_rate": 0.00019999458776773094,
-      "loss": 46.0,
-      "step": 20575
-    },
-    {
-      "epoch": 3.3136197109384438,
-      "grad_norm": 0.015249301679432392,
-      "learning_rate": 0.00019999458724136916,
-      "loss": 46.0,
-      "step": 20576
-    },
-    {
-      "epoch": 3.3137807480172308,
-      "grad_norm": 0.0030181773472577333,
-      "learning_rate": 0.0001999945867149818,
-      "loss": 46.0,
-      "step": 20577
-    },
-    {
-      "epoch": 3.3139417850960182,
-      "grad_norm": 0.0014944358263164759,
-      "learning_rate": 0.00019999458618856882,
-      "loss": 46.0,
-      "step": 20578
-    },
-    {
-      "epoch": 3.3141028221748057,
-      "grad_norm": 0.00282674515619874,
-      "learning_rate": 0.00019999458566213025,
-      "loss": 46.0,
-      "step": 20579
-    },
-    {
-      "epoch": 3.314263859253593,
-      "grad_norm": 0.0022494960576295853,
-      "learning_rate": 0.0001999945851356661,
-      "loss": 46.0,
-      "step": 20580
-    },
-    {
-      "epoch": 3.3144248963323806,
-      "grad_norm": 0.0031505771912634373,
-      "learning_rate": 0.00019999458460917636,
-      "loss": 46.0,
-      "step": 20581
-    },
-    {
-      "epoch": 3.314585933411168,
-      "grad_norm": 0.002489285310730338,
-      "learning_rate": 0.000199994584082661,
-      "loss": 46.0,
-      "step": 20582
-    },
-    {
-      "epoch": 3.3147469704899555,
-      "grad_norm": 0.005234418902546167,
-      "learning_rate": 0.00019999458355612006,
-      "loss": 46.0,
-      "step": 20583
-    },
-    {
-      "epoch": 3.314908007568743,
-      "grad_norm": 0.0018298632930964231,
-      "learning_rate": 0.00019999458302955353,
-      "loss": 46.0,
-      "step": 20584
-    },
-    {
-      "epoch": 3.31506904464753,
-      "grad_norm": 0.004291005432605743,
-      "learning_rate": 0.00019999458250296142,
-      "loss": 46.0,
-      "step": 20585
-    },
-    {
-      "epoch": 3.3152300817263174,
-      "grad_norm": 0.002526351949200034,
-      "learning_rate": 0.0001999945819763437,
-      "loss": 46.0,
-      "step": 20586
-    },
-    {
-      "epoch": 3.315391118805105,
-      "grad_norm": 0.004323775414377451,
-      "learning_rate": 0.00019999458144970037,
-      "loss": 46.0,
-      "step": 20587
-    },
-    {
-      "epoch": 3.3155521558838923,
-      "grad_norm": 0.003098824294283986,
-      "learning_rate": 0.00019999458092303146,
-      "loss": 46.0,
-      "step": 20588
-    },
-    {
-      "epoch": 3.3157131929626797,
-      "grad_norm": 0.003980815876275301,
-      "learning_rate": 0.00019999458039633697,
-      "loss": 46.0,
-      "step": 20589
-    },
-    {
-      "epoch": 3.315874230041467,
-      "grad_norm": 0.0017970160115510225,
-      "learning_rate": 0.0001999945798696169,
-      "loss": 46.0,
-      "step": 20590
-    },
-    {
-      "epoch": 3.316035267120254,
-      "grad_norm": 0.0006784077850170434,
-      "learning_rate": 0.0001999945793428712,
-      "loss": 46.0,
-      "step": 20591
-    },
-    {
-      "epoch": 3.3161963041990417,
-      "grad_norm": 0.004842281807214022,
-      "learning_rate": 0.00019999457881609992,
-      "loss": 46.0,
-      "step": 20592
-    },
-    {
-      "epoch": 3.316357341277829,
-      "grad_norm": 0.0013882593484595418,
-      "learning_rate": 0.00019999457828930305,
-      "loss": 46.0,
-      "step": 20593
-    },
-    {
-      "epoch": 3.3165183783566166,
-      "grad_norm": 0.0033856176305562258,
-      "learning_rate": 0.0001999945777624806,
-      "loss": 46.0,
-      "step": 20594
-    },
-    {
-      "epoch": 3.316679415435404,
-      "grad_norm": 0.006645835470408201,
-      "learning_rate": 0.00019999457723563253,
-      "loss": 46.0,
-      "step": 20595
-    },
-    {
-      "epoch": 3.3168404525141915,
-      "grad_norm": 0.008744263090193272,
-      "learning_rate": 0.00019999457670875887,
-      "loss": 46.0,
-      "step": 20596
-    },
-    {
-      "epoch": 3.317001489592979,
-      "grad_norm": 0.004656627308577299,
-      "learning_rate": 0.00019999457618185963,
-      "loss": 46.0,
-      "step": 20597
-    },
-    {
-      "epoch": 3.3171625266717664,
-      "grad_norm": 0.0016099194763228297,
-      "learning_rate": 0.0001999945756549348,
-      "loss": 46.0,
-      "step": 20598
-    },
-    {
-      "epoch": 3.3173235637505534,
-      "grad_norm": 0.0004129723529331386,
-      "learning_rate": 0.00019999457512798435,
-      "loss": 46.0,
-      "step": 20599
-    },
-    {
-      "epoch": 3.317484600829341,
-      "grad_norm": 0.002357868477702141,
-      "learning_rate": 0.00019999457460100834,
-      "loss": 46.0,
-      "step": 20600
-    },
-    {
-      "epoch": 3.3176456379081283,
-      "grad_norm": 0.0013515668688341975,
-      "learning_rate": 0.0001999945740740067,
-      "loss": 46.0,
-      "step": 20601
-    },
-    {
-      "epoch": 3.3178066749869157,
-      "grad_norm": 0.0037109958939254284,
-      "learning_rate": 0.0001999945735469795,
-      "loss": 46.0,
-      "step": 20602
-    },
-    {
-      "epoch": 3.317967712065703,
-      "grad_norm": 0.010511173866689205,
-      "learning_rate": 0.00019999457301992667,
-      "loss": 46.0,
-      "step": 20603
-    },
-    {
-      "epoch": 3.3181287491444906,
-      "grad_norm": 0.002971078734844923,
-      "learning_rate": 0.0001999945724928483,
-      "loss": 46.0,
-      "step": 20604
-    },
-    {
-      "epoch": 3.318289786223278,
-      "grad_norm": 0.0021232753060758114,
-      "learning_rate": 0.0001999945719657443,
-      "loss": 46.0,
-      "step": 20605
-    },
-    {
-      "epoch": 3.318450823302065,
-      "grad_norm": 0.005472460761666298,
-      "learning_rate": 0.0001999945714386147,
-      "loss": 46.0,
-      "step": 20606
-    },
-    {
-      "epoch": 3.3186118603808525,
-      "grad_norm": 0.002818155800923705,
-      "learning_rate": 0.00019999457091145954,
-      "loss": 46.0,
-      "step": 20607
-    },
-    {
-      "epoch": 3.31877289745964,
-      "grad_norm": 0.002014744793996215,
-      "learning_rate": 0.00019999457038427876,
-      "loss": 46.0,
-      "step": 20608
-    },
-    {
-      "epoch": 3.3189339345384274,
-      "grad_norm": 0.003153668250888586,
-      "learning_rate": 0.0001999945698570724,
-      "loss": 46.0,
-      "step": 20609
-    },
-    {
-      "epoch": 3.319094971617215,
-      "grad_norm": 0.0007961684023030102,
-      "learning_rate": 0.00019999456932984043,
-      "loss": 46.0,
-      "step": 20610
-    },
-    {
-      "epoch": 3.3192560086960023,
-      "grad_norm": 0.0024413957726210356,
-      "learning_rate": 0.00019999456880258288,
-      "loss": 46.0,
-      "step": 20611
-    },
-    {
-      "epoch": 3.31941704577479,
-      "grad_norm": 0.004205880220979452,
-      "learning_rate": 0.00019999456827529972,
-      "loss": 46.0,
-      "step": 20612
-    },
-    {
-      "epoch": 3.3195780828535772,
-      "grad_norm": 0.0039676483720541,
-      "learning_rate": 0.00019999456774799098,
-      "loss": 46.0,
-      "step": 20613
-    },
-    {
-      "epoch": 3.3197391199323643,
-      "grad_norm": 0.0021052660886198282,
-      "learning_rate": 0.00019999456722065664,
-      "loss": 46.0,
-      "step": 20614
-    },
-    {
-      "epoch": 3.3199001570111517,
-      "grad_norm": 0.003295567352324724,
-      "learning_rate": 0.00019999456669329672,
-      "loss": 46.0,
-      "step": 20615
-    },
-    {
-      "epoch": 3.320061194089939,
-      "grad_norm": 0.0043800389394164085,
-      "learning_rate": 0.0001999945661659112,
-      "loss": 46.0,
-      "step": 20616
-    },
-    {
-      "epoch": 3.3202222311687266,
-      "grad_norm": 0.004360324703156948,
-      "learning_rate": 0.0001999945656385001,
-      "loss": 46.0,
-      "step": 20617
-    },
-    {
-      "epoch": 3.320383268247514,
-      "grad_norm": 0.002035733312368393,
-      "learning_rate": 0.00019999456511106338,
-      "loss": 46.0,
-      "step": 20618
-    },
-    {
-      "epoch": 3.3205443053263015,
-      "grad_norm": 0.0013910542475059628,
-      "learning_rate": 0.00019999456458360109,
-      "loss": 46.0,
-      "step": 20619
-    },
-    {
-      "epoch": 3.3207053424050885,
-      "grad_norm": 0.0024826517328619957,
-      "learning_rate": 0.00019999456405611317,
-      "loss": 46.0,
-      "step": 20620
-    },
-    {
-      "epoch": 3.320866379483876,
-      "grad_norm": 0.00580619927495718,
-      "learning_rate": 0.00019999456352859968,
-      "loss": 46.0,
-      "step": 20621
-    },
-    {
-      "epoch": 3.3210274165626634,
-      "grad_norm": 0.001343421288765967,
-      "learning_rate": 0.00019999456300106062,
-      "loss": 46.0,
-      "step": 20622
-    },
-    {
-      "epoch": 3.321188453641451,
-      "grad_norm": 0.0010835020802915096,
-      "learning_rate": 0.00019999456247349592,
-      "loss": 46.0,
-      "step": 20623
-    },
-    {
-      "epoch": 3.3213494907202383,
-      "grad_norm": 0.002239476889371872,
-      "learning_rate": 0.00019999456194590566,
-      "loss": 46.0,
-      "step": 20624
-    },
-    {
-      "epoch": 3.321510527799026,
-      "grad_norm": 0.0007758995052427053,
-      "learning_rate": 0.00019999456141828978,
-      "loss": 46.0,
-      "step": 20625
-    },
-    {
-      "epoch": 3.3216715648778132,
-      "grad_norm": 0.002318687504157424,
-      "learning_rate": 0.00019999456089064835,
-      "loss": 46.0,
-      "step": 20626
-    },
-    {
-      "epoch": 3.3218326019566007,
-      "grad_norm": 0.003191697411239147,
-      "learning_rate": 0.00019999456036298127,
-      "loss": 46.0,
-      "step": 20627
-    },
-    {
-      "epoch": 3.321993639035388,
-      "grad_norm": 0.0048633115366101265,
-      "learning_rate": 0.00019999455983528864,
-      "loss": 46.0,
-      "step": 20628
-    },
-    {
-      "epoch": 3.322154676114175,
-      "grad_norm": 0.004585013259202242,
-      "learning_rate": 0.00019999455930757042,
-      "loss": 46.0,
-      "step": 20629
-    },
-    {
-      "epoch": 3.3223157131929626,
-      "grad_norm": 0.002711718901991844,
-      "learning_rate": 0.00019999455877982658,
-      "loss": 46.0,
-      "step": 20630
-    },
-    {
-      "epoch": 3.32247675027175,
-      "grad_norm": 0.0012398186372593045,
-      "learning_rate": 0.00019999455825205715,
-      "loss": 46.0,
-      "step": 20631
-    },
-    {
-      "epoch": 3.3226377873505375,
-      "grad_norm": 0.0022244194988161325,
-      "learning_rate": 0.00019999455772426212,
-      "loss": 46.0,
-      "step": 20632
-    },
-    {
-      "epoch": 3.322798824429325,
-      "grad_norm": 0.000605227192863822,
-      "learning_rate": 0.00019999455719644154,
-      "loss": 46.0,
-      "step": 20633
-    },
-    {
-      "epoch": 3.3229598615081124,
-      "grad_norm": 0.013281482271850109,
-      "learning_rate": 0.00019999455666859533,
-      "loss": 46.0,
-      "step": 20634
-    },
-    {
-      "epoch": 3.3231208985868994,
-      "grad_norm": 0.003237626515328884,
-      "learning_rate": 0.00019999455614072353,
-      "loss": 46.0,
-      "step": 20635
-    },
-    {
-      "epoch": 3.323281935665687,
-      "grad_norm": 0.0018632387509569526,
-      "learning_rate": 0.00019999455561282612,
-      "loss": 46.0,
-      "step": 20636
-    },
-    {
-      "epoch": 3.3234429727444743,
-      "grad_norm": 0.0033254828304052353,
-      "learning_rate": 0.00019999455508490314,
-      "loss": 46.0,
-      "step": 20637
-    },
-    {
-      "epoch": 3.3236040098232618,
-      "grad_norm": 0.0015562031185254455,
-      "learning_rate": 0.00019999455455695458,
-      "loss": 46.0,
-      "step": 20638
-    },
-    {
-      "epoch": 3.323765046902049,
-      "grad_norm": 0.0019269655458629131,
-      "learning_rate": 0.0001999945540289804,
-      "loss": 46.0,
-      "step": 20639
-    },
-    {
-      "epoch": 3.3239260839808367,
-      "grad_norm": 0.004350072238594294,
-      "learning_rate": 0.00019999455350098061,
-      "loss": 46.0,
-      "step": 20640
-    },
-    {
-      "epoch": 3.324087121059624,
-      "grad_norm": 0.004888691008090973,
-      "learning_rate": 0.00019999455297295526,
-      "loss": 46.0,
-      "step": 20641
-    },
-    {
-      "epoch": 3.3242481581384116,
-      "grad_norm": 0.003112856298685074,
-      "learning_rate": 0.00019999455244490433,
-      "loss": 46.0,
-      "step": 20642
-    },
-    {
-      "epoch": 3.3244091952171986,
-      "grad_norm": 0.003854363923892379,
-      "learning_rate": 0.00019999455191682777,
-      "loss": 46.0,
-      "step": 20643
-    },
-    {
-      "epoch": 3.324570232295986,
-      "grad_norm": 0.010935456492006779,
-      "learning_rate": 0.00019999455138872564,
-      "loss": 46.0,
-      "step": 20644
-    },
-    {
-      "epoch": 3.3247312693747735,
-      "grad_norm": 0.004687285982072353,
-      "learning_rate": 0.0001999945508605979,
-      "loss": 46.0,
-      "step": 20645
-    },
-    {
-      "epoch": 3.324892306453561,
-      "grad_norm": 0.001487278612330556,
-      "learning_rate": 0.00019999455033244457,
-      "loss": 46.0,
-      "step": 20646
-    },
-    {
-      "epoch": 3.3250533435323484,
-      "grad_norm": 0.0009150960249826312,
-      "learning_rate": 0.00019999454980426564,
-      "loss": 46.0,
-      "step": 20647
-    },
-    {
-      "epoch": 3.325214380611136,
-      "grad_norm": 0.0037873967085033655,
-      "learning_rate": 0.00019999454927606113,
-      "loss": 46.0,
-      "step": 20648
-    },
-    {
-      "epoch": 3.3253754176899233,
-      "grad_norm": 0.0072565763257443905,
-      "learning_rate": 0.00019999454874783103,
-      "loss": 46.0,
-      "step": 20649
-    },
-    {
-      "epoch": 3.3255364547687103,
-      "grad_norm": 0.0011376062175258994,
-      "learning_rate": 0.00019999454821957536,
-      "loss": 46.0,
-      "step": 20650
-    },
-    {
-      "epoch": 3.3256974918474977,
-      "grad_norm": 0.0016701099229976535,
-      "learning_rate": 0.00019999454769129403,
-      "loss": 46.0,
-      "step": 20651
-    },
-    {
-      "epoch": 3.325858528926285,
-      "grad_norm": 0.000620406586676836,
-      "learning_rate": 0.00019999454716298714,
-      "loss": 46.0,
-      "step": 20652
-    },
-    {
-      "epoch": 3.3260195660050726,
-      "grad_norm": 0.006020907778292894,
-      "learning_rate": 0.00019999454663465467,
-      "loss": 46.0,
-      "step": 20653
-    },
-    {
-      "epoch": 3.32618060308386,
-      "grad_norm": 0.003065197728574276,
-      "learning_rate": 0.0001999945461062966,
-      "loss": 46.0,
-      "step": 20654
-    },
-    {
-      "epoch": 3.3263416401626476,
-      "grad_norm": 0.004784727469086647,
-      "learning_rate": 0.00019999454557791292,
-      "loss": 46.0,
-      "step": 20655
-    },
-    {
-      "epoch": 3.326502677241435,
-      "grad_norm": 0.0049361675046384335,
-      "learning_rate": 0.00019999454504950368,
-      "loss": 46.0,
-      "step": 20656
-    },
-    {
-      "epoch": 3.3266637143202225,
-      "grad_norm": 0.002360751386731863,
-      "learning_rate": 0.00019999454452106883,
-      "loss": 46.0,
-      "step": 20657
-    },
-    {
-      "epoch": 3.3268247513990095,
-      "grad_norm": 0.005002843681722879,
-      "learning_rate": 0.00019999454399260836,
-      "loss": 46.0,
-      "step": 20658
-    },
-    {
-      "epoch": 3.326985788477797,
-      "grad_norm": 0.005297200754284859,
-      "learning_rate": 0.00019999454346412233,
-      "loss": 46.0,
-      "step": 20659
-    },
-    {
-      "epoch": 3.3271468255565844,
-      "grad_norm": 0.004770212806761265,
-      "learning_rate": 0.00019999454293561072,
-      "loss": 46.0,
-      "step": 20660
-    },
-    {
-      "epoch": 3.327307862635372,
-      "grad_norm": 0.0026987900491803885,
-      "learning_rate": 0.00019999454240707346,
-      "loss": 46.0,
-      "step": 20661
-    },
-    {
-      "epoch": 3.3274688997141593,
-      "grad_norm": 0.0032691967207938433,
-      "learning_rate": 0.00019999454187851065,
-      "loss": 46.0,
-      "step": 20662
-    },
-    {
-      "epoch": 3.3276299367929467,
-      "grad_norm": 0.0004995659692212939,
-      "learning_rate": 0.00019999454134992224,
-      "loss": 46.0,
-      "step": 20663
-    },
-    {
-      "epoch": 3.3277909738717337,
-      "grad_norm": 0.005385675933212042,
-      "learning_rate": 0.00019999454082130822,
-      "loss": 46.0,
-      "step": 20664
-    },
-    {
-      "epoch": 3.327952010950521,
-      "grad_norm": 0.0010919248452410102,
-      "learning_rate": 0.00019999454029266862,
-      "loss": 46.0,
-      "step": 20665
-    },
-    {
-      "epoch": 3.3281130480293086,
-      "grad_norm": 0.0014182814629748464,
-      "learning_rate": 0.00019999453976400343,
-      "loss": 46.0,
-      "step": 20666
-    },
-    {
-      "epoch": 3.328274085108096,
-      "grad_norm": 0.0022046626545488834,
-      "learning_rate": 0.00019999453923531265,
-      "loss": 46.0,
-      "step": 20667
-    },
-    {
-      "epoch": 3.3284351221868835,
-      "grad_norm": 0.008756585419178009,
-      "learning_rate": 0.00019999453870659623,
-      "loss": 46.0,
-      "step": 20668
-    },
-    {
-      "epoch": 3.328596159265671,
-      "grad_norm": 0.0010858243331313133,
-      "learning_rate": 0.00019999453817785427,
-      "loss": 46.0,
-      "step": 20669
-    },
-    {
-      "epoch": 3.3287571963444584,
-      "grad_norm": 0.004240847192704678,
-      "learning_rate": 0.00019999453764908668,
-      "loss": 46.0,
-      "step": 20670
-    },
-    {
-      "epoch": 3.328918233423246,
-      "grad_norm": 0.004645610228180885,
-      "learning_rate": 0.00019999453712029355,
-      "loss": 46.0,
-      "step": 20671
-    },
-    {
-      "epoch": 3.3290792705020333,
-      "grad_norm": 0.00560349877923727,
-      "learning_rate": 0.00019999453659147478,
-      "loss": 46.0,
-      "step": 20672
-    },
-    {
-      "epoch": 3.3292403075808203,
-      "grad_norm": 0.0004197901289444417,
-      "learning_rate": 0.00019999453606263043,
-      "loss": 46.0,
-      "step": 20673
-    },
-    {
-      "epoch": 3.329401344659608,
-      "grad_norm": 0.0031594678293913603,
-      "learning_rate": 0.0001999945355337605,
-      "loss": 46.0,
-      "step": 20674
-    },
-    {
-      "epoch": 3.3295623817383952,
-      "grad_norm": 0.0054811895824968815,
-      "learning_rate": 0.00019999453500486495,
-      "loss": 46.0,
-      "step": 20675
-    },
-    {
-      "epoch": 3.3297234188171827,
-      "grad_norm": 0.006125556770712137,
-      "learning_rate": 0.00019999453447594383,
-      "loss": 46.0,
-      "step": 20676
-    },
-    {
-      "epoch": 3.32988445589597,
-      "grad_norm": 0.0014067491283640265,
-      "learning_rate": 0.0001999945339469971,
-      "loss": 46.0,
-      "step": 20677
-    },
-    {
-      "epoch": 3.3300454929747576,
-      "grad_norm": 0.004283860791474581,
-      "learning_rate": 0.00019999453341802478,
-      "loss": 46.0,
-      "step": 20678
-    },
-    {
-      "epoch": 3.3302065300535446,
-      "grad_norm": 0.006565620191395283,
-      "learning_rate": 0.00019999453288902687,
-      "loss": 46.0,
-      "step": 20679
-    },
-    {
-      "epoch": 3.330367567132332,
-      "grad_norm": 0.002439651405438781,
-      "learning_rate": 0.00019999453236000338,
-      "loss": 46.0,
-      "step": 20680
-    },
-    {
-      "epoch": 3.3305286042111195,
-      "grad_norm": 0.002741681644693017,
-      "learning_rate": 0.00019999453183095427,
-      "loss": 46.0,
-      "step": 20681
-    },
-    {
-      "epoch": 3.330689641289907,
-      "grad_norm": 0.0022727784235030413,
-      "learning_rate": 0.00019999453130187958,
-      "loss": 46.0,
-      "step": 20682
-    },
-    {
-      "epoch": 3.3308506783686944,
-      "grad_norm": 0.002211730694398284,
-      "learning_rate": 0.0001999945307727793,
-      "loss": 46.0,
-      "step": 20683
-    },
-    {
-      "epoch": 3.331011715447482,
-      "grad_norm": 0.0033187856897711754,
-      "learning_rate": 0.0001999945302436534,
-      "loss": 46.0,
-      "step": 20684
-    },
-    {
-      "epoch": 3.3311727525262693,
-      "grad_norm": 0.004431689623743296,
-      "learning_rate": 0.00019999452971450194,
-      "loss": 46.0,
-      "step": 20685
-    },
-    {
-      "epoch": 3.3313337896050568,
-      "grad_norm": 0.0006730607710778713,
-      "learning_rate": 0.00019999452918532487,
-      "loss": 46.0,
-      "step": 20686
-    },
-    {
-      "epoch": 3.331494826683844,
-      "grad_norm": 0.009347470477223396,
-      "learning_rate": 0.00019999452865612224,
-      "loss": 46.0,
-      "step": 20687
-    },
-    {
-      "epoch": 3.3316558637626312,
-      "grad_norm": 0.006930544041097164,
-      "learning_rate": 0.00019999452812689397,
-      "loss": 46.0,
-      "step": 20688
-    },
-    {
-      "epoch": 3.3318169008414187,
-      "grad_norm": 0.0034979088231921196,
-      "learning_rate": 0.00019999452759764013,
-      "loss": 46.0,
-      "step": 20689
-    },
-    {
-      "epoch": 3.331977937920206,
-      "grad_norm": 0.0031090572010725737,
-      "learning_rate": 0.0001999945270683607,
-      "loss": 46.0,
-      "step": 20690
-    },
-    {
-      "epoch": 3.3321389749989936,
-      "grad_norm": 0.0035414446610957384,
-      "learning_rate": 0.00019999452653905565,
-      "loss": 46.0,
-      "step": 20691
-    },
-    {
-      "epoch": 3.332300012077781,
-      "grad_norm": 0.0014024300035089254,
-      "learning_rate": 0.00019999452600972503,
-      "loss": 46.0,
-      "step": 20692
-    },
-    {
-      "epoch": 3.3324610491565685,
-      "grad_norm": 0.002385796280577779,
-      "learning_rate": 0.00019999452548036882,
-      "loss": 46.0,
-      "step": 20693
-    },
-    {
-      "epoch": 3.3326220862353555,
-      "grad_norm": 0.0024936727713793516,
-      "learning_rate": 0.000199994524950987,
-      "loss": 46.0,
-      "step": 20694
-    },
-    {
-      "epoch": 3.332783123314143,
-      "grad_norm": 0.0014235054841265082,
-      "learning_rate": 0.0001999945244215796,
-      "loss": 46.0,
-      "step": 20695
-    },
-    {
-      "epoch": 3.3329441603929304,
-      "grad_norm": 0.0037812914233654737,
-      "learning_rate": 0.0001999945238921466,
-      "loss": 46.0,
-      "step": 20696
-    },
-    {
-      "epoch": 3.333105197471718,
-      "grad_norm": 0.0016779991565272212,
-      "learning_rate": 0.000199994523362688,
-      "loss": 46.0,
-      "step": 20697
-    },
-    {
-      "epoch": 3.3332662345505053,
-      "grad_norm": 0.0014002828393131495,
-      "learning_rate": 0.0001999945228332038,
-      "loss": 46.0,
-      "step": 20698
-    },
-    {
-      "epoch": 3.3334272716292928,
-      "grad_norm": 0.00628651911392808,
-      "learning_rate": 0.00019999452230369405,
-      "loss": 46.0,
-      "step": 20699
-    },
-    {
-      "epoch": 3.33358830870808,
-      "grad_norm": 0.009666165336966515,
-      "learning_rate": 0.00019999452177415865,
-      "loss": 46.0,
-      "step": 20700
-    },
-    {
-      "epoch": 3.3337493457868677,
-      "grad_norm": 0.0023739002645015717,
-      "learning_rate": 0.00019999452124459767,
-      "loss": 46.0,
-      "step": 20701
-    },
-    {
-      "epoch": 3.3339103828656547,
-      "grad_norm": 0.0019127086270600557,
-      "learning_rate": 0.00019999452071501112,
-      "loss": 46.0,
-      "step": 20702
-    },
-    {
-      "epoch": 3.334071419944442,
-      "grad_norm": 0.002191151725128293,
-      "learning_rate": 0.00019999452018539898,
-      "loss": 46.0,
-      "step": 20703
-    },
-    {
-      "epoch": 3.3342324570232296,
-      "grad_norm": 0.004061200190335512,
-      "learning_rate": 0.00019999451965576124,
-      "loss": 46.0,
-      "step": 20704
-    },
-    {
-      "epoch": 3.334393494102017,
-      "grad_norm": 0.0008081065607257187,
-      "learning_rate": 0.00019999451912609787,
-      "loss": 46.0,
-      "step": 20705
-    },
-    {
-      "epoch": 3.3345545311808045,
-      "grad_norm": 0.001797973527573049,
-      "learning_rate": 0.00019999451859640895,
-      "loss": 46.0,
-      "step": 20706
-    },
-    {
-      "epoch": 3.334715568259592,
-      "grad_norm": 0.0018977973377332091,
-      "learning_rate": 0.0001999945180666944,
-      "loss": 46.0,
-      "step": 20707
-    },
-    {
-      "epoch": 3.334876605338379,
-      "grad_norm": 0.0018908806378021836,
-      "learning_rate": 0.0001999945175369543,
-      "loss": 46.0,
-      "step": 20708
-    },
-    {
-      "epoch": 3.3350376424171664,
-      "grad_norm": 0.001165832974947989,
-      "learning_rate": 0.00019999451700718858,
-      "loss": 46.0,
-      "step": 20709
-    },
-    {
-      "epoch": 3.335198679495954,
-      "grad_norm": 0.002274847123771906,
-      "learning_rate": 0.00019999451647739728,
-      "loss": 46.0,
-      "step": 20710
-    },
-    {
-      "epoch": 3.3353597165747413,
-      "grad_norm": 0.004373036324977875,
-      "learning_rate": 0.00019999451594758036,
-      "loss": 46.0,
-      "step": 20711
-    },
-    {
-      "epoch": 3.3355207536535287,
-      "grad_norm": 0.004236109554767609,
-      "learning_rate": 0.0001999945154177379,
-      "loss": 46.0,
-      "step": 20712
-    },
-    {
-      "epoch": 3.335681790732316,
-      "grad_norm": 0.005360749550163746,
-      "learning_rate": 0.00019999451488786978,
-      "loss": 46.0,
-      "step": 20713
-    },
-    {
-      "epoch": 3.3358428278111036,
-      "grad_norm": 0.0015582973137497902,
-      "learning_rate": 0.0001999945143579761,
-      "loss": 46.0,
-      "step": 20714
-    },
-    {
-      "epoch": 3.336003864889891,
-      "grad_norm": 0.007665487006306648,
-      "learning_rate": 0.00019999451382805684,
-      "loss": 46.0,
-      "step": 20715
-    },
-    {
-      "epoch": 3.336164901968678,
-      "grad_norm": 0.011600405909121037,
-      "learning_rate": 0.00019999451329811196,
-      "loss": 46.0,
-      "step": 20716
-    },
-    {
-      "epoch": 3.3363259390474655,
-      "grad_norm": 0.004321750719100237,
-      "learning_rate": 0.00019999451276814153,
-      "loss": 46.0,
-      "step": 20717
-    },
-    {
-      "epoch": 3.336486976126253,
-      "grad_norm": 0.008660581894218922,
-      "learning_rate": 0.00019999451223814545,
-      "loss": 46.0,
-      "step": 20718
-    },
-    {
-      "epoch": 3.3366480132050405,
-      "grad_norm": 0.003269096603617072,
-      "learning_rate": 0.0001999945117081238,
-      "loss": 46.0,
-      "step": 20719
-    },
-    {
-      "epoch": 3.336809050283828,
-      "grad_norm": 0.0034555543679744005,
-      "learning_rate": 0.00019999451117807656,
-      "loss": 46.0,
-      "step": 20720
-    },
-    {
-      "epoch": 3.3369700873626154,
-      "grad_norm": 0.001743566826917231,
-      "learning_rate": 0.00019999451064800372,
-      "loss": 46.0,
-      "step": 20721
-    },
-    {
-      "epoch": 3.337131124441403,
-      "grad_norm": 0.0014382983790710568,
-      "learning_rate": 0.0001999945101179053,
-      "loss": 46.0,
-      "step": 20722
-    },
-    {
-      "epoch": 3.33729216152019,
-      "grad_norm": 0.009087533690035343,
-      "learning_rate": 0.00019999450958778128,
-      "loss": 46.0,
-      "step": 20723
-    },
-    {
-      "epoch": 3.3374531985989773,
-      "grad_norm": 0.003041849471628666,
-      "learning_rate": 0.00019999450905763166,
-      "loss": 46.0,
-      "step": 20724
-    },
-    {
-      "epoch": 3.3376142356777647,
-      "grad_norm": 0.00196527736261487,
-      "learning_rate": 0.00019999450852745641,
-      "loss": 46.0,
-      "step": 20725
-    },
-    {
-      "epoch": 3.337775272756552,
-      "grad_norm": 0.003818437224254012,
-      "learning_rate": 0.00019999450799725564,
-      "loss": 46.0,
-      "step": 20726
-    },
-    {
-      "epoch": 3.3379363098353396,
-      "grad_norm": 0.004665777087211609,
-      "learning_rate": 0.00019999450746702922,
-      "loss": 46.0,
-      "step": 20727
-    },
-    {
-      "epoch": 3.338097346914127,
-      "grad_norm": 0.002855050377547741,
-      "learning_rate": 0.00019999450693677725,
-      "loss": 46.0,
-      "step": 20728
-    },
-    {
-      "epoch": 3.3382583839929145,
-      "grad_norm": 0.0007667517638765275,
-      "learning_rate": 0.00019999450640649966,
-      "loss": 46.0,
-      "step": 20729
-    },
-    {
-      "epoch": 3.338419421071702,
-      "grad_norm": 0.001363030169159174,
-      "learning_rate": 0.00019999450587619648,
-      "loss": 46.0,
-      "step": 20730
-    },
-    {
-      "epoch": 3.338580458150489,
-      "grad_norm": 0.0018201707862317562,
-      "learning_rate": 0.0001999945053458677,
-      "loss": 46.0,
-      "step": 20731
-    },
-    {
-      "epoch": 3.3387414952292764,
-      "grad_norm": 0.0019068780820816755,
-      "learning_rate": 0.00019999450481551336,
-      "loss": 46.0,
-      "step": 20732
-    },
-    {
-      "epoch": 3.338902532308064,
-      "grad_norm": 0.0014781117206439376,
-      "learning_rate": 0.0001999945042851334,
-      "loss": 46.0,
-      "step": 20733
-    },
-    {
-      "epoch": 3.3390635693868513,
-      "grad_norm": 0.017115868628025055,
-      "learning_rate": 0.00019999450375472784,
-      "loss": 46.0,
-      "step": 20734
-    },
-    {
-      "epoch": 3.339224606465639,
-      "grad_norm": 0.0011176539119333029,
-      "learning_rate": 0.0001999945032242967,
-      "loss": 46.0,
-      "step": 20735
-    },
-    {
-      "epoch": 3.3393856435444262,
-      "grad_norm": 0.002341602463275194,
-      "learning_rate": 0.00019999450269383995,
-      "loss": 46.0,
-      "step": 20736
-    },
-    {
-      "epoch": 3.3395466806232132,
-      "grad_norm": 0.0017520219553261995,
-      "learning_rate": 0.00019999450216335763,
-      "loss": 46.0,
-      "step": 20737
-    },
-    {
-      "epoch": 3.3397077177020007,
-      "grad_norm": 0.002885375637561083,
-      "learning_rate": 0.0001999945016328497,
-      "loss": 46.0,
-      "step": 20738
-    },
-    {
-      "epoch": 3.339868754780788,
-      "grad_norm": 0.005546938627958298,
-      "learning_rate": 0.00019999450110231619,
-      "loss": 46.0,
-      "step": 20739
-    },
-    {
-      "epoch": 3.3400297918595756,
-      "grad_norm": 0.002033561235293746,
-      "learning_rate": 0.00019999450057175708,
-      "loss": 46.0,
-      "step": 20740
-    },
-    {
-      "epoch": 3.340190828938363,
-      "grad_norm": 0.0027569993399083614,
-      "learning_rate": 0.00019999450004117234,
-      "loss": 46.0,
-      "step": 20741
-    },
-    {
-      "epoch": 3.3403518660171505,
-      "grad_norm": 0.00869830884039402,
-      "learning_rate": 0.00019999449951056206,
-      "loss": 46.0,
-      "step": 20742
-    },
-    {
-      "epoch": 3.340512903095938,
-      "grad_norm": 0.01141760777682066,
-      "learning_rate": 0.00019999449897992617,
-      "loss": 46.0,
-      "step": 20743
-    },
-    {
-      "epoch": 3.3406739401747254,
-      "grad_norm": 0.006766782607883215,
-      "learning_rate": 0.0001999944984492647,
-      "loss": 46.0,
-      "step": 20744
-    },
-    {
-      "epoch": 3.340834977253513,
-      "grad_norm": 0.007144279312342405,
-      "learning_rate": 0.0001999944979185776,
-      "loss": 46.0,
-      "step": 20745
-    },
-    {
-      "epoch": 3.3409960143323,
-      "grad_norm": 0.006176311988383532,
-      "learning_rate": 0.00019999449738786494,
-      "loss": 46.0,
-      "step": 20746
-    },
-    {
-      "epoch": 3.3411570514110873,
-      "grad_norm": 0.006303990725427866,
-      "learning_rate": 0.00019999449685712665,
-      "loss": 46.0,
-      "step": 20747
-    },
-    {
-      "epoch": 3.3413180884898748,
-      "grad_norm": 0.0009887849446386099,
-      "learning_rate": 0.0001999944963263628,
-      "loss": 46.0,
-      "step": 20748
-    },
-    {
-      "epoch": 3.3414791255686622,
-      "grad_norm": 0.004154487047344446,
-      "learning_rate": 0.00019999449579557338,
-      "loss": 46.0,
-      "step": 20749
-    },
-    {
-      "epoch": 3.3416401626474497,
-      "grad_norm": 0.002034200122579932,
-      "learning_rate": 0.00019999449526475832,
-      "loss": 46.0,
-      "step": 20750
-    },
-    {
-      "epoch": 3.341801199726237,
-      "grad_norm": 0.004265953321009874,
-      "learning_rate": 0.00019999449473391767,
-      "loss": 46.0,
-      "step": 20751
-    },
-    {
-      "epoch": 3.341962236805024,
-      "grad_norm": 0.0021867260802537203,
-      "learning_rate": 0.00019999449420305142,
-      "loss": 46.0,
-      "step": 20752
-    },
-    {
-      "epoch": 3.3421232738838116,
-      "grad_norm": 0.011533687822520733,
-      "learning_rate": 0.0001999944936721596,
-      "loss": 46.0,
-      "step": 20753
-    },
-    {
-      "epoch": 3.342284310962599,
-      "grad_norm": 0.0027390571776777506,
-      "learning_rate": 0.00019999449314124217,
-      "loss": 46.0,
-      "step": 20754
-    },
-    {
-      "epoch": 3.3424453480413865,
-      "grad_norm": 0.005125917959958315,
-      "learning_rate": 0.00019999449261029917,
-      "loss": 46.0,
-      "step": 20755
-    },
-    {
-      "epoch": 3.342606385120174,
-      "grad_norm": 0.0035561162512749434,
-      "learning_rate": 0.00019999449207933057,
-      "loss": 46.0,
-      "step": 20756
-    },
-    {
-      "epoch": 3.3427674221989614,
-      "grad_norm": 0.001143692061305046,
-      "learning_rate": 0.00019999449154833637,
-      "loss": 46.0,
-      "step": 20757
-    },
-    {
-      "epoch": 3.342928459277749,
-      "grad_norm": 0.0011206611525267363,
-      "learning_rate": 0.00019999449101731656,
-      "loss": 46.0,
-      "step": 20758
-    },
-    {
-      "epoch": 3.3430894963565363,
-      "grad_norm": 0.004940640646964312,
-      "learning_rate": 0.00019999449048627117,
-      "loss": 46.0,
-      "step": 20759
-    },
-    {
-      "epoch": 3.3432505334353233,
-      "grad_norm": 0.009246529079973698,
-      "learning_rate": 0.00019999448995520021,
-      "loss": 46.0,
-      "step": 20760
-    },
-    {
-      "epoch": 3.3434115705141108,
-      "grad_norm": 0.008048427291214466,
-      "learning_rate": 0.00019999448942410362,
-      "loss": 46.0,
-      "step": 20761
-    },
-    {
-      "epoch": 3.343572607592898,
-      "grad_norm": 0.004151604138314724,
-      "learning_rate": 0.00019999448889298143,
-      "loss": 46.0,
-      "step": 20762
-    },
-    {
-      "epoch": 3.3437336446716857,
-      "grad_norm": 0.002345653250813484,
-      "learning_rate": 0.0001999944883618337,
-      "loss": 46.0,
-      "step": 20763
-    },
-    {
-      "epoch": 3.343894681750473,
-      "grad_norm": 0.0022222164552658796,
-      "learning_rate": 0.00019999448783066033,
-      "loss": 46.0,
-      "step": 20764
-    },
-    {
-      "epoch": 3.3440557188292606,
-      "grad_norm": 0.0024431883357465267,
-      "learning_rate": 0.00019999448729946139,
-      "loss": 46.0,
-      "step": 20765
-    },
-    {
-      "epoch": 3.344216755908048,
-      "grad_norm": 0.0074889385141432285,
-      "learning_rate": 0.00019999448676823685,
-      "loss": 46.0,
-      "step": 20766
-    },
-    {
-      "epoch": 3.344377792986835,
-      "grad_norm": 0.0017557577230036259,
-      "learning_rate": 0.0001999944862369867,
-      "loss": 46.0,
-      "step": 20767
-    },
-    {
-      "epoch": 3.3445388300656225,
-      "grad_norm": 0.020771101117134094,
-      "learning_rate": 0.00019999448570571097,
-      "loss": 46.0,
-      "step": 20768
-    },
-    {
-      "epoch": 3.34469986714441,
-      "grad_norm": 0.008961359970271587,
-      "learning_rate": 0.00019999448517440965,
-      "loss": 46.0,
-      "step": 20769
-    },
-    {
-      "epoch": 3.3448609042231974,
-      "grad_norm": 0.002128560794517398,
-      "learning_rate": 0.00019999448464308274,
-      "loss": 46.0,
-      "step": 20770
-    },
-    {
-      "epoch": 3.345021941301985,
-      "grad_norm": 0.0015349421883001924,
-      "learning_rate": 0.00019999448411173022,
-      "loss": 46.0,
-      "step": 20771
-    },
-    {
-      "epoch": 3.3451829783807723,
-      "grad_norm": 0.0042289504781365395,
-      "learning_rate": 0.00019999448358035214,
-      "loss": 46.0,
-      "step": 20772
-    },
-    {
-      "epoch": 3.3453440154595597,
-      "grad_norm": 0.0038474167231470346,
-      "learning_rate": 0.00019999448304894844,
-      "loss": 46.0,
-      "step": 20773
-    },
-    {
-      "epoch": 3.345505052538347,
-      "grad_norm": 0.00923189613968134,
-      "learning_rate": 0.00019999448251751916,
-      "loss": 46.0,
-      "step": 20774
-    },
-    {
-      "epoch": 3.345666089617134,
-      "grad_norm": 0.0013017089804634452,
-      "learning_rate": 0.00019999448198606428,
-      "loss": 46.0,
-      "step": 20775
-    },
-    {
-      "epoch": 3.3458271266959216,
-      "grad_norm": 0.0007115930202417076,
-      "learning_rate": 0.00019999448145458377,
-      "loss": 46.0,
-      "step": 20776
-    },
-    {
-      "epoch": 3.345988163774709,
-      "grad_norm": 0.0033399956300854683,
-      "learning_rate": 0.0001999944809230777,
-      "loss": 46.0,
-      "step": 20777
-    },
-    {
-      "epoch": 3.3461492008534965,
-      "grad_norm": 0.001808327273465693,
-      "learning_rate": 0.00019999448039154604,
-      "loss": 46.0,
-      "step": 20778
-    },
-    {
-      "epoch": 3.346310237932284,
-      "grad_norm": 0.00222820439375937,
-      "learning_rate": 0.0001999944798599888,
-      "loss": 46.0,
-      "step": 20779
-    },
-    {
-      "epoch": 3.3464712750110714,
-      "grad_norm": 0.0009919166332110763,
-      "learning_rate": 0.00019999447932840593,
-      "loss": 46.0,
-      "step": 20780
-    },
-    {
-      "epoch": 3.3466323120898585,
-      "grad_norm": 0.0009279330843128264,
-      "learning_rate": 0.0001999944787967975,
-      "loss": 46.0,
-      "step": 20781
-    },
-    {
-      "epoch": 3.346793349168646,
-      "grad_norm": 0.0031397975981235504,
-      "learning_rate": 0.00019999447826516347,
-      "loss": 46.0,
-      "step": 20782
-    },
-    {
-      "epoch": 3.3469543862474334,
-      "grad_norm": 0.002643154002726078,
-      "learning_rate": 0.00019999447773350385,
-      "loss": 46.0,
-      "step": 20783
-    },
-    {
-      "epoch": 3.347115423326221,
-      "grad_norm": 0.0035772614646703005,
-      "learning_rate": 0.0001999944772018186,
-      "loss": 46.0,
-      "step": 20784
-    },
-    {
-      "epoch": 3.3472764604050083,
-      "grad_norm": 0.0010152192553505301,
-      "learning_rate": 0.0001999944766701078,
-      "loss": 46.0,
-      "step": 20785
-    },
-    {
-      "epoch": 3.3474374974837957,
-      "grad_norm": 0.0020548859611153603,
-      "learning_rate": 0.00019999447613837138,
-      "loss": 46.0,
-      "step": 20786
-    },
-    {
-      "epoch": 3.347598534562583,
-      "grad_norm": 0.0012068054638803005,
-      "learning_rate": 0.00019999447560660935,
-      "loss": 46.0,
-      "step": 20787
-    },
-    {
-      "epoch": 3.3477595716413706,
-      "grad_norm": 0.0028345868922770023,
-      "learning_rate": 0.00019999447507482176,
-      "loss": 46.0,
-      "step": 20788
-    },
-    {
-      "epoch": 3.347920608720158,
-      "grad_norm": 0.0026454576291143894,
-      "learning_rate": 0.0001999944745430086,
-      "loss": 46.0,
-      "step": 20789
-    },
-    {
-      "epoch": 3.348081645798945,
-      "grad_norm": 0.009147074073553085,
-      "learning_rate": 0.0001999944740111698,
-      "loss": 46.0,
-      "step": 20790
-    },
-    {
-      "epoch": 3.3482426828777325,
-      "grad_norm": 0.002347939182072878,
-      "learning_rate": 0.00019999447347930543,
-      "loss": 46.0,
-      "step": 20791
-    },
-    {
-      "epoch": 3.34840371995652,
-      "grad_norm": 0.004727926570922136,
-      "learning_rate": 0.00019999447294741544,
-      "loss": 46.0,
-      "step": 20792
-    },
-    {
-      "epoch": 3.3485647570353074,
-      "grad_norm": 0.002594444202259183,
-      "learning_rate": 0.0001999944724154999,
-      "loss": 46.0,
-      "step": 20793
-    },
-    {
-      "epoch": 3.348725794114095,
-      "grad_norm": 0.0014013260370120406,
-      "learning_rate": 0.00019999447188355873,
-      "loss": 46.0,
-      "step": 20794
-    },
-    {
-      "epoch": 3.3488868311928823,
-      "grad_norm": 0.005070163868367672,
-      "learning_rate": 0.00019999447135159198,
-      "loss": 46.0,
-      "step": 20795
-    },
-    {
-      "epoch": 3.3490478682716693,
-      "grad_norm": 0.0053039295598864555,
-      "learning_rate": 0.00019999447081959964,
-      "loss": 46.0,
-      "step": 20796
-    },
-    {
-      "epoch": 3.349208905350457,
-      "grad_norm": 0.005232988856732845,
-      "learning_rate": 0.0001999944702875817,
-      "loss": 46.0,
-      "step": 20797
-    },
-    {
-      "epoch": 3.3493699424292442,
-      "grad_norm": 0.0017715356079861522,
-      "learning_rate": 0.00019999446975553815,
-      "loss": 46.0,
-      "step": 20798
-    },
-    {
-      "epoch": 3.3495309795080317,
-      "grad_norm": 0.006321721710264683,
-      "learning_rate": 0.00019999446922346902,
-      "loss": 46.0,
-      "step": 20799
-    },
-    {
-      "epoch": 3.349692016586819,
-      "grad_norm": 0.00569496676325798,
-      "learning_rate": 0.0001999944686913743,
-      "loss": 46.0,
-      "step": 20800
-    },
-    {
-      "epoch": 3.3498530536656066,
-      "grad_norm": 0.005051599815487862,
-      "learning_rate": 0.000199994468159254,
-      "loss": 46.0,
-      "step": 20801
-    },
-    {
-      "epoch": 3.350014090744394,
-      "grad_norm": 0.003661056747660041,
-      "learning_rate": 0.00019999446762710807,
-      "loss": 46.0,
-      "step": 20802
-    },
-    {
-      "epoch": 3.3501751278231815,
-      "grad_norm": 0.0013093366287648678,
-      "learning_rate": 0.0001999944670949366,
-      "loss": 46.0,
-      "step": 20803
-    },
-    {
-      "epoch": 3.3503361649019685,
-      "grad_norm": 0.0018988519441336393,
-      "learning_rate": 0.00019999446656273948,
-      "loss": 46.0,
-      "step": 20804
-    },
-    {
-      "epoch": 3.350497201980756,
-      "grad_norm": 0.0051345848478376865,
-      "learning_rate": 0.00019999446603051683,
-      "loss": 46.0,
-      "step": 20805
-    },
-    {
-      "epoch": 3.3506582390595434,
-      "grad_norm": 0.0069570839405059814,
-      "learning_rate": 0.00019999446549826854,
-      "loss": 46.0,
-      "step": 20806
-    },
-    {
-      "epoch": 3.350819276138331,
-      "grad_norm": 0.008283072151243687,
-      "learning_rate": 0.00019999446496599466,
-      "loss": 46.0,
-      "step": 20807
-    },
-    {
-      "epoch": 3.3509803132171183,
-      "grad_norm": 0.0020841099321842194,
-      "learning_rate": 0.0001999944644336952,
-      "loss": 46.0,
-      "step": 20808
-    },
-    {
-      "epoch": 3.3511413502959058,
-      "grad_norm": 0.002222476527094841,
-      "learning_rate": 0.00019999446390137014,
-      "loss": 46.0,
-      "step": 20809
-    },
-    {
-      "epoch": 3.351302387374693,
-      "grad_norm": 0.005603512283414602,
-      "learning_rate": 0.00019999446336901947,
-      "loss": 46.0,
-      "step": 20810
-    },
-    {
-      "epoch": 3.3514634244534802,
-      "grad_norm": 0.008383144624531269,
-      "learning_rate": 0.0001999944628366432,
-      "loss": 46.0,
-      "step": 20811
-    },
-    {
-      "epoch": 3.3516244615322677,
-      "grad_norm": 0.0036360309459269047,
-      "learning_rate": 0.00019999446230424138,
-      "loss": 46.0,
-      "step": 20812
-    },
-    {
-      "epoch": 3.351785498611055,
-      "grad_norm": 0.005416340660303831,
-      "learning_rate": 0.00019999446177181395,
-      "loss": 46.0,
-      "step": 20813
-    },
-    {
-      "epoch": 3.3519465356898426,
-      "grad_norm": 0.002239226596429944,
-      "learning_rate": 0.0001999944612393609,
-      "loss": 46.0,
-      "step": 20814
-    },
-    {
-      "epoch": 3.35210757276863,
-      "grad_norm": 0.002920815022662282,
-      "learning_rate": 0.0001999944607068823,
-      "loss": 46.0,
-      "step": 20815
-    },
-    {
-      "epoch": 3.3522686098474175,
-      "grad_norm": 0.0020466644782572985,
-      "learning_rate": 0.0001999944601743781,
-      "loss": 46.0,
-      "step": 20816
-    },
-    {
-      "epoch": 3.352429646926205,
-      "grad_norm": 0.0018387290183454752,
-      "learning_rate": 0.0001999944596418483,
-      "loss": 46.0,
-      "step": 20817
-    },
-    {
-      "epoch": 3.3525906840049924,
-      "grad_norm": 0.003895149566233158,
-      "learning_rate": 0.00019999445910929287,
-      "loss": 46.0,
-      "step": 20818
-    },
-    {
-      "epoch": 3.3527517210837794,
-      "grad_norm": 0.002318927086889744,
-      "learning_rate": 0.00019999445857671186,
-      "loss": 46.0,
-      "step": 20819
-    },
-    {
-      "epoch": 3.352912758162567,
-      "grad_norm": 0.006851447746157646,
-      "learning_rate": 0.00019999445804410527,
-      "loss": 46.0,
-      "step": 20820
-    },
-    {
-      "epoch": 3.3530737952413543,
-      "grad_norm": 0.012200540862977505,
-      "learning_rate": 0.0001999944575114731,
-      "loss": 46.0,
-      "step": 20821
-    },
-    {
-      "epoch": 3.3532348323201417,
-      "grad_norm": 0.001614811597391963,
-      "learning_rate": 0.00019999445697881533,
-      "loss": 46.0,
-      "step": 20822
-    },
-    {
-      "epoch": 3.353395869398929,
-      "grad_norm": 0.00449097016826272,
-      "learning_rate": 0.00019999445644613197,
-      "loss": 46.0,
-      "step": 20823
-    },
-    {
-      "epoch": 3.3535569064777166,
-      "grad_norm": 0.005182918161153793,
-      "learning_rate": 0.00019999445591342298,
-      "loss": 46.0,
-      "step": 20824
-    },
-    {
-      "epoch": 3.3537179435565037,
-      "grad_norm": 0.001550475601106882,
-      "learning_rate": 0.00019999445538068845,
-      "loss": 46.0,
-      "step": 20825
-    },
-    {
-      "epoch": 3.353878980635291,
-      "grad_norm": 0.0004955654731020331,
-      "learning_rate": 0.00019999445484792828,
-      "loss": 46.0,
-      "step": 20826
-    },
-    {
-      "epoch": 3.3540400177140786,
-      "grad_norm": 0.017394132912158966,
-      "learning_rate": 0.00019999445431514255,
-      "loss": 46.0,
-      "step": 20827
-    },
-    {
-      "epoch": 3.354201054792866,
-      "grad_norm": 0.00375940278172493,
-      "learning_rate": 0.0001999944537823312,
-      "loss": 46.0,
-      "step": 20828
-    },
-    {
-      "epoch": 3.3543620918716535,
-      "grad_norm": 0.002217361005023122,
-      "learning_rate": 0.00019999445324949428,
-      "loss": 46.0,
-      "step": 20829
-    },
-    {
-      "epoch": 3.354523128950441,
-      "grad_norm": 0.007536798715591431,
-      "learning_rate": 0.00019999445271663176,
-      "loss": 46.0,
-      "step": 20830
-    },
-    {
-      "epoch": 3.3546841660292284,
-      "grad_norm": 0.0009419141570106149,
-      "learning_rate": 0.00019999445218374365,
-      "loss": 46.0,
-      "step": 20831
-    },
-    {
-      "epoch": 3.354845203108016,
-      "grad_norm": 0.0032349408138543367,
-      "learning_rate": 0.00019999445165082993,
-      "loss": 46.0,
-      "step": 20832
-    },
-    {
-      "epoch": 3.355006240186803,
-      "grad_norm": 0.007819443941116333,
-      "learning_rate": 0.00019999445111789063,
-      "loss": 46.0,
-      "step": 20833
-    },
-    {
-      "epoch": 3.3551672772655903,
-      "grad_norm": 0.014505263417959213,
-      "learning_rate": 0.00019999445058492573,
-      "loss": 46.0,
-      "step": 20834
-    },
-    {
-      "epoch": 3.3553283143443777,
-      "grad_norm": 0.001194495940580964,
-      "learning_rate": 0.00019999445005193525,
-      "loss": 46.0,
-      "step": 20835
-    },
-    {
-      "epoch": 3.355489351423165,
-      "grad_norm": 0.006805042736232281,
-      "learning_rate": 0.00019999444951891918,
-      "loss": 46.0,
-      "step": 20836
-    },
-    {
-      "epoch": 3.3556503885019526,
-      "grad_norm": 0.0037581122014671564,
-      "learning_rate": 0.00019999444898587747,
-      "loss": 46.0,
-      "step": 20837
-    },
-    {
-      "epoch": 3.35581142558074,
-      "grad_norm": 0.0020235648844391108,
-      "learning_rate": 0.0001999944484528102,
-      "loss": 46.0,
-      "step": 20838
-    },
-    {
-      "epoch": 3.3559724626595275,
-      "grad_norm": 0.00293528800830245,
-      "learning_rate": 0.00019999444791971732,
-      "loss": 46.0,
-      "step": 20839
-    },
-    {
-      "epoch": 3.3561334997383145,
-      "grad_norm": 0.008518904447555542,
-      "learning_rate": 0.0001999944473865989,
-      "loss": 46.0,
-      "step": 20840
-    },
-    {
-      "epoch": 3.356294536817102,
-      "grad_norm": 0.002349145244807005,
-      "learning_rate": 0.00019999444685345484,
-      "loss": 46.0,
-      "step": 20841
-    },
-    {
-      "epoch": 3.3564555738958894,
-      "grad_norm": 0.0014410704607143998,
-      "learning_rate": 0.0001999944463202852,
-      "loss": 46.0,
-      "step": 20842
-    },
-    {
-      "epoch": 3.356616610974677,
-      "grad_norm": 0.0021259256172925234,
-      "learning_rate": 0.00019999444578708996,
-      "loss": 46.0,
-      "step": 20843
-    },
-    {
-      "epoch": 3.3567776480534643,
-      "grad_norm": 0.007543244399130344,
-      "learning_rate": 0.00019999444525386912,
-      "loss": 46.0,
-      "step": 20844
-    },
-    {
-      "epoch": 3.356938685132252,
-      "grad_norm": 0.003808224806562066,
-      "learning_rate": 0.0001999944447206227,
-      "loss": 46.0,
-      "step": 20845
-    },
-    {
-      "epoch": 3.3570997222110392,
-      "grad_norm": 0.001223452971316874,
-      "learning_rate": 0.0001999944441873507,
-      "loss": 46.0,
-      "step": 20846
-    },
-    {
-      "epoch": 3.3572607592898267,
-      "grad_norm": 0.0026060177478939295,
-      "learning_rate": 0.00019999444365405305,
-      "loss": 46.0,
-      "step": 20847
-    },
-    {
-      "epoch": 3.3574217963686137,
-      "grad_norm": 0.0025034029968082905,
-      "learning_rate": 0.00019999444312072986,
-      "loss": 46.0,
-      "step": 20848
-    },
-    {
-      "epoch": 3.357582833447401,
-      "grad_norm": 0.004101970698684454,
-      "learning_rate": 0.00019999444258738107,
-      "loss": 46.0,
-      "step": 20849
-    },
-    {
-      "epoch": 3.3577438705261886,
-      "grad_norm": 0.006617759820073843,
-      "learning_rate": 0.00019999444205400665,
-      "loss": 46.0,
-      "step": 20850
-    },
-    {
-      "epoch": 3.357904907604976,
-      "grad_norm": 0.002592572011053562,
-      "learning_rate": 0.0001999944415206067,
-      "loss": 46.0,
-      "step": 20851
-    },
-    {
-      "epoch": 3.3580659446837635,
-      "grad_norm": 0.007662277668714523,
-      "learning_rate": 0.0001999944409871811,
-      "loss": 46.0,
-      "step": 20852
-    },
-    {
-      "epoch": 3.358226981762551,
-      "grad_norm": 0.0012603906216099858,
-      "learning_rate": 0.00019999444045372994,
-      "loss": 46.0,
-      "step": 20853
-    },
-    {
-      "epoch": 3.358388018841338,
-      "grad_norm": 0.0030396704096347094,
-      "learning_rate": 0.00019999443992025316,
-      "loss": 46.0,
-      "step": 20854
-    },
-    {
-      "epoch": 3.3585490559201254,
-      "grad_norm": 0.003122115507721901,
-      "learning_rate": 0.00019999443938675083,
-      "loss": 46.0,
-      "step": 20855
-    },
-    {
-      "epoch": 3.358710092998913,
-      "grad_norm": 0.0007894922746345401,
-      "learning_rate": 0.00019999443885322286,
-      "loss": 46.0,
-      "step": 20856
-    },
-    {
-      "epoch": 3.3588711300777003,
-      "grad_norm": 0.0028509756084531546,
-      "learning_rate": 0.00019999443831966932,
-      "loss": 46.0,
-      "step": 20857
-    },
-    {
-      "epoch": 3.359032167156488,
-      "grad_norm": 0.00408986397087574,
-      "learning_rate": 0.00019999443778609017,
-      "loss": 46.0,
-      "step": 20858
-    },
-    {
-      "epoch": 3.3591932042352752,
-      "grad_norm": 0.004244456999003887,
-      "learning_rate": 0.00019999443725248544,
-      "loss": 46.0,
-      "step": 20859
-    },
-    {
-      "epoch": 3.3593542413140627,
-      "grad_norm": 0.002518328605219722,
-      "learning_rate": 0.00019999443671885509,
-      "loss": 46.0,
-      "step": 20860
-    },
-    {
-      "epoch": 3.35951527839285,
-      "grad_norm": 0.0038628955371677876,
-      "learning_rate": 0.00019999443618519918,
-      "loss": 46.0,
-      "step": 20861
-    },
-    {
-      "epoch": 3.3596763154716376,
-      "grad_norm": 0.008417327888309956,
-      "learning_rate": 0.00019999443565151768,
-      "loss": 46.0,
-      "step": 20862
-    },
-    {
-      "epoch": 3.3598373525504246,
-      "grad_norm": 0.005671640392392874,
-      "learning_rate": 0.00019999443511781057,
-      "loss": 46.0,
-      "step": 20863
-    },
-    {
-      "epoch": 3.359998389629212,
-      "grad_norm": 0.0021164407953619957,
-      "learning_rate": 0.00019999443458407787,
-      "loss": 46.0,
-      "step": 20864
-    },
-    {
-      "epoch": 3.3601594267079995,
-      "grad_norm": 0.0012921376619488,
-      "learning_rate": 0.00019999443405031955,
-      "loss": 46.0,
-      "step": 20865
-    },
-    {
-      "epoch": 3.360320463786787,
-      "grad_norm": 0.005236590281128883,
-      "learning_rate": 0.00019999443351653568,
-      "loss": 46.0,
-      "step": 20866
-    },
-    {
-      "epoch": 3.3604815008655744,
-      "grad_norm": 0.0020093233324587345,
-      "learning_rate": 0.00019999443298272622,
-      "loss": 46.0,
-      "step": 20867
-    },
-    {
-      "epoch": 3.360642537944362,
-      "grad_norm": 0.0008796803303994238,
-      "learning_rate": 0.00019999443244889115,
-      "loss": 46.0,
-      "step": 20868
-    },
-    {
-      "epoch": 3.360803575023149,
-      "grad_norm": 0.0035525192506611347,
-      "learning_rate": 0.00019999443191503046,
-      "loss": 46.0,
-      "step": 20869
-    },
-    {
-      "epoch": 3.3609646121019363,
-      "grad_norm": 0.007032199762761593,
-      "learning_rate": 0.0001999944313811442,
-      "loss": 46.0,
-      "step": 20870
-    },
-    {
-      "epoch": 3.3611256491807238,
-      "grad_norm": 0.008084863424301147,
-      "learning_rate": 0.00019999443084723234,
-      "loss": 46.0,
-      "step": 20871
-    },
-    {
-      "epoch": 3.361286686259511,
-      "grad_norm": 0.0008347261464223266,
-      "learning_rate": 0.0001999944303132949,
-      "loss": 46.0,
-      "step": 20872
-    },
-    {
-      "epoch": 3.3614477233382987,
-      "grad_norm": 0.004060279577970505,
-      "learning_rate": 0.00019999442977933188,
-      "loss": 46.0,
-      "step": 20873
-    },
-    {
-      "epoch": 3.361608760417086,
-      "grad_norm": 0.0007594830240122974,
-      "learning_rate": 0.00019999442924534323,
-      "loss": 46.0,
-      "step": 20874
-    },
-    {
-      "epoch": 3.3617697974958736,
-      "grad_norm": 0.0027352285105735064,
-      "learning_rate": 0.000199994428711329,
-      "loss": 46.0,
-      "step": 20875
-    },
-    {
-      "epoch": 3.361930834574661,
-      "grad_norm": 0.0022483242210000753,
-      "learning_rate": 0.00019999442817728917,
-      "loss": 46.0,
-      "step": 20876
-    },
-    {
-      "epoch": 3.362091871653448,
-      "grad_norm": 0.001823390950448811,
-      "learning_rate": 0.00019999442764322375,
-      "loss": 46.0,
-      "step": 20877
-    },
-    {
-      "epoch": 3.3622529087322355,
-      "grad_norm": 0.003105304203927517,
-      "learning_rate": 0.00019999442710913275,
-      "loss": 46.0,
-      "step": 20878
-    },
-    {
-      "epoch": 3.362413945811023,
-      "grad_norm": 0.003525971435010433,
-      "learning_rate": 0.00019999442657501614,
-      "loss": 46.0,
-      "step": 20879
-    },
-    {
-      "epoch": 3.3625749828898104,
-      "grad_norm": 0.00048445616266690195,
-      "learning_rate": 0.00019999442604087396,
-      "loss": 46.0,
-      "step": 20880
-    },
-    {
-      "epoch": 3.362736019968598,
-      "grad_norm": 0.0010100880172103643,
-      "learning_rate": 0.00019999442550670615,
-      "loss": 46.0,
-      "step": 20881
-    },
-    {
-      "epoch": 3.3628970570473853,
-      "grad_norm": 0.003239578567445278,
-      "learning_rate": 0.0001999944249725128,
-      "loss": 46.0,
-      "step": 20882
-    },
-    {
-      "epoch": 3.3630580941261727,
-      "grad_norm": 0.0023300175089389086,
-      "learning_rate": 0.0001999944244382938,
-      "loss": 46.0,
-      "step": 20883
-    },
-    {
-      "epoch": 3.3632191312049597,
-      "grad_norm": 0.004318100865930319,
-      "learning_rate": 0.00019999442390404923,
-      "loss": 46.0,
-      "step": 20884
-    },
-    {
-      "epoch": 3.363380168283747,
-      "grad_norm": 0.0025229519233107567,
-      "learning_rate": 0.00019999442336977906,
-      "loss": 46.0,
-      "step": 20885
-    },
-    {
-      "epoch": 3.3635412053625346,
-      "grad_norm": 0.002387227723374963,
-      "learning_rate": 0.0001999944228354833,
-      "loss": 46.0,
-      "step": 20886
-    },
-    {
-      "epoch": 3.363702242441322,
-      "grad_norm": 0.00975829642266035,
-      "learning_rate": 0.00019999442230116197,
-      "loss": 46.0,
-      "step": 20887
-    },
-    {
-      "epoch": 3.3638632795201096,
-      "grad_norm": 0.0012357309460639954,
-      "learning_rate": 0.00019999442176681502,
-      "loss": 46.0,
-      "step": 20888
-    },
-    {
-      "epoch": 3.364024316598897,
-      "grad_norm": 0.002482104115188122,
-      "learning_rate": 0.0001999944212324425,
-      "loss": 46.0,
-      "step": 20889
-    },
-    {
-      "epoch": 3.3641853536776845,
-      "grad_norm": 0.0033045695163309574,
-      "learning_rate": 0.00019999442069804438,
-      "loss": 46.0,
-      "step": 20890
-    },
-    {
-      "epoch": 3.364346390756472,
-      "grad_norm": 0.01630895771086216,
-      "learning_rate": 0.00019999442016362063,
-      "loss": 46.0,
-      "step": 20891
-    },
-    {
-      "epoch": 3.364507427835259,
-      "grad_norm": 0.004521742928773165,
-      "learning_rate": 0.0001999944196291713,
-      "loss": 46.0,
-      "step": 20892
-    },
-    {
-      "epoch": 3.3646684649140464,
-      "grad_norm": 0.007557746022939682,
-      "learning_rate": 0.00019999441909469642,
-      "loss": 46.0,
-      "step": 20893
-    },
-    {
-      "epoch": 3.364829501992834,
-      "grad_norm": 0.002224199241027236,
-      "learning_rate": 0.00019999441856019589,
-      "loss": 46.0,
-      "step": 20894
-    },
-    {
-      "epoch": 3.3649905390716213,
-      "grad_norm": 0.001543963560834527,
-      "learning_rate": 0.0001999944180256698,
-      "loss": 46.0,
-      "step": 20895
-    },
-    {
-      "epoch": 3.3651515761504087,
-      "grad_norm": 0.001412329962477088,
-      "learning_rate": 0.00019999441749111812,
-      "loss": 46.0,
-      "step": 20896
-    },
-    {
-      "epoch": 3.365312613229196,
-      "grad_norm": 0.005306669976562262,
-      "learning_rate": 0.00019999441695654082,
-      "loss": 46.0,
-      "step": 20897
-    },
-    {
-      "epoch": 3.365473650307983,
-      "grad_norm": 0.007320174016058445,
-      "learning_rate": 0.00019999441642193797,
-      "loss": 46.0,
-      "step": 20898
-    },
-    {
-      "epoch": 3.3656346873867706,
-      "grad_norm": 0.0006209679995663464,
-      "learning_rate": 0.0001999944158873095,
-      "loss": 46.0,
-      "step": 20899
-    },
-    {
-      "epoch": 3.365795724465558,
-      "grad_norm": 0.0018192294519394636,
-      "learning_rate": 0.00019999441535265542,
-      "loss": 46.0,
-      "step": 20900
-    },
-    {
-      "epoch": 3.3659567615443455,
-      "grad_norm": 0.0006589615368284285,
-      "learning_rate": 0.00019999441481797576,
-      "loss": 46.0,
-      "step": 20901
-    },
-    {
-      "epoch": 3.366117798623133,
-      "grad_norm": 0.005652967374771833,
-      "learning_rate": 0.0001999944142832705,
-      "loss": 46.0,
-      "step": 20902
-    },
-    {
-      "epoch": 3.3662788357019204,
-      "grad_norm": 0.0008445366402156651,
-      "learning_rate": 0.00019999441374853966,
-      "loss": 46.0,
-      "step": 20903
-    },
-    {
-      "epoch": 3.366439872780708,
-      "grad_norm": 0.001166544039733708,
-      "learning_rate": 0.00019999441321378323,
-      "loss": 46.0,
-      "step": 20904
-    },
-    {
-      "epoch": 3.3666009098594953,
-      "grad_norm": 0.008133132942020893,
-      "learning_rate": 0.0001999944126790012,
-      "loss": 46.0,
-      "step": 20905
-    },
-    {
-      "epoch": 3.3667619469382823,
-      "grad_norm": 0.005981351714581251,
-      "learning_rate": 0.00019999441214419358,
-      "loss": 46.0,
-      "step": 20906
-    },
-    {
-      "epoch": 3.36692298401707,
-      "grad_norm": 0.0025186811108142138,
-      "learning_rate": 0.00019999441160936033,
-      "loss": 46.0,
-      "step": 20907
-    },
-    {
-      "epoch": 3.3670840210958572,
-      "grad_norm": 0.010102360509335995,
-      "learning_rate": 0.00019999441107450153,
-      "loss": 46.0,
-      "step": 20908
-    },
-    {
-      "epoch": 3.3672450581746447,
-      "grad_norm": 0.002637322759255767,
-      "learning_rate": 0.00019999441053961714,
-      "loss": 46.0,
-      "step": 20909
-    },
-    {
-      "epoch": 3.367406095253432,
-      "grad_norm": 0.010602711699903011,
-      "learning_rate": 0.00019999441000470713,
-      "loss": 46.0,
-      "step": 20910
-    },
-    {
-      "epoch": 3.3675671323322196,
-      "grad_norm": 0.006769978441298008,
-      "learning_rate": 0.00019999440946977154,
-      "loss": 46.0,
-      "step": 20911
-    },
-    {
-      "epoch": 3.367728169411007,
-      "grad_norm": 0.005505494307726622,
-      "learning_rate": 0.00019999440893481035,
-      "loss": 46.0,
-      "step": 20912
-    },
-    {
-      "epoch": 3.367889206489794,
-      "grad_norm": 0.0021205723751336336,
-      "learning_rate": 0.00019999440839982359,
-      "loss": 46.0,
-      "step": 20913
-    },
-    {
-      "epoch": 3.3680502435685815,
-      "grad_norm": 0.0019522298825904727,
-      "learning_rate": 0.0001999944078648112,
-      "loss": 46.0,
-      "step": 20914
-    },
-    {
-      "epoch": 3.368211280647369,
-      "grad_norm": 0.0017423428362235427,
-      "learning_rate": 0.00019999440732977326,
-      "loss": 46.0,
-      "step": 20915
-    },
-    {
-      "epoch": 3.3683723177261564,
-      "grad_norm": 0.002849302487447858,
-      "learning_rate": 0.00019999440679470968,
-      "loss": 46.0,
-      "step": 20916
-    },
-    {
-      "epoch": 3.368533354804944,
-      "grad_norm": 0.009484389796853065,
-      "learning_rate": 0.00019999440625962053,
-      "loss": 46.0,
-      "step": 20917
-    },
-    {
-      "epoch": 3.3686943918837313,
-      "grad_norm": 0.008852027356624603,
-      "learning_rate": 0.00019999440572450577,
-      "loss": 46.0,
-      "step": 20918
-    },
-    {
-      "epoch": 3.3688554289625188,
-      "grad_norm": 0.003986544441431761,
-      "learning_rate": 0.00019999440518936545,
-      "loss": 46.0,
-      "step": 20919
-    },
-    {
-      "epoch": 3.3690164660413062,
-      "grad_norm": 0.004097459837794304,
-      "learning_rate": 0.00019999440465419952,
-      "loss": 46.0,
-      "step": 20920
-    },
-    {
-      "epoch": 3.3691775031200932,
-      "grad_norm": 0.001590346684679389,
-      "learning_rate": 0.00019999440411900797,
-      "loss": 46.0,
-      "step": 20921
-    },
-    {
-      "epoch": 3.3693385401988807,
-      "grad_norm": 0.002898836974054575,
-      "learning_rate": 0.00019999440358379087,
-      "loss": 46.0,
-      "step": 20922
-    },
-    {
-      "epoch": 3.369499577277668,
-      "grad_norm": 0.0013032472925260663,
-      "learning_rate": 0.00019999440304854815,
-      "loss": 46.0,
-      "step": 20923
-    },
-    {
-      "epoch": 3.3696606143564556,
-      "grad_norm": 0.0042036776430904865,
-      "learning_rate": 0.00019999440251327984,
-      "loss": 46.0,
-      "step": 20924
-    },
-    {
-      "epoch": 3.369821651435243,
-      "grad_norm": 0.003879842348396778,
-      "learning_rate": 0.00019999440197798594,
-      "loss": 46.0,
-      "step": 20925
-    },
-    {
-      "epoch": 3.3699826885140305,
-      "grad_norm": 0.007399814669042826,
-      "learning_rate": 0.00019999440144266646,
-      "loss": 46.0,
-      "step": 20926
-    },
-    {
-      "epoch": 3.370143725592818,
-      "grad_norm": 0.00479771988466382,
-      "learning_rate": 0.00019999440090732136,
-      "loss": 46.0,
-      "step": 20927
-    },
-    {
-      "epoch": 3.370304762671605,
-      "grad_norm": 0.001091778976842761,
-      "learning_rate": 0.00019999440037195068,
-      "loss": 46.0,
-      "step": 20928
-    },
-    {
-      "epoch": 3.3704657997503924,
-      "grad_norm": 0.0009164019138552248,
-      "learning_rate": 0.0001999943998365544,
-      "loss": 46.0,
-      "step": 20929
-    },
-    {
-      "epoch": 3.37062683682918,
-      "grad_norm": 0.01292500365525484,
-      "learning_rate": 0.00019999439930113252,
-      "loss": 46.0,
-      "step": 20930
-    },
-    {
-      "epoch": 3.3707878739079673,
-      "grad_norm": 0.0014164549065753818,
-      "learning_rate": 0.00019999439876568507,
-      "loss": 46.0,
-      "step": 20931
-    },
-    {
-      "epoch": 3.3709489109867548,
-      "grad_norm": 0.0019916431047022343,
-      "learning_rate": 0.00019999439823021201,
-      "loss": 46.0,
-      "step": 20932
-    },
-    {
-      "epoch": 3.371109948065542,
-      "grad_norm": 0.002783977659419179,
-      "learning_rate": 0.00019999439769471337,
-      "loss": 46.0,
-      "step": 20933
-    },
-    {
-      "epoch": 3.3712709851443297,
-      "grad_norm": 0.012138224206864834,
-      "learning_rate": 0.00019999439715918916,
-      "loss": 46.0,
-      "step": 20934
-    },
-    {
-      "epoch": 3.371432022223117,
-      "grad_norm": 0.004607359878718853,
-      "learning_rate": 0.0001999943966236393,
-      "loss": 46.0,
-      "step": 20935
-    },
-    {
-      "epoch": 3.371593059301904,
-      "grad_norm": 0.0011318891774863005,
-      "learning_rate": 0.00019999439608806387,
-      "loss": 46.0,
-      "step": 20936
-    },
-    {
-      "epoch": 3.3717540963806916,
-      "grad_norm": 0.00155333848670125,
-      "learning_rate": 0.00019999439555246285,
-      "loss": 46.0,
-      "step": 20937
-    },
-    {
-      "epoch": 3.371915133459479,
-      "grad_norm": 0.011060883291065693,
-      "learning_rate": 0.00019999439501683624,
-      "loss": 46.0,
-      "step": 20938
-    },
-    {
-      "epoch": 3.3720761705382665,
-      "grad_norm": 0.003010694170370698,
-      "learning_rate": 0.00019999439448118404,
-      "loss": 46.0,
-      "step": 20939
-    },
-    {
-      "epoch": 3.372237207617054,
-      "grad_norm": 0.0024879195261746645,
-      "learning_rate": 0.00019999439394550623,
-      "loss": 46.0,
-      "step": 20940
-    },
-    {
-      "epoch": 3.3723982446958414,
-      "grad_norm": 0.0032771851401776075,
-      "learning_rate": 0.00019999439340980283,
-      "loss": 46.0,
-      "step": 20941
-    },
-    {
-      "epoch": 3.3725592817746284,
-      "grad_norm": 0.007326935417950153,
-      "learning_rate": 0.00019999439287407385,
-      "loss": 46.0,
-      "step": 20942
-    },
-    {
-      "epoch": 3.372720318853416,
-      "grad_norm": 0.0035892713349312544,
-      "learning_rate": 0.00019999439233831927,
-      "loss": 46.0,
-      "step": 20943
-    },
-    {
-      "epoch": 3.3728813559322033,
-      "grad_norm": 0.00965859368443489,
-      "learning_rate": 0.00019999439180253909,
-      "loss": 46.0,
-      "step": 20944
-    },
-    {
-      "epoch": 3.3730423930109907,
-      "grad_norm": 0.001588193466886878,
-      "learning_rate": 0.0001999943912667333,
-      "loss": 46.0,
-      "step": 20945
-    },
-    {
-      "epoch": 3.373203430089778,
-      "grad_norm": 0.012424630112946033,
-      "learning_rate": 0.00019999439073090195,
-      "loss": 46.0,
-      "step": 20946
-    },
-    {
-      "epoch": 3.3733644671685656,
-      "grad_norm": 0.0005508853355422616,
-      "learning_rate": 0.000199994390195045,
-      "loss": 46.0,
-      "step": 20947
-    },
-    {
-      "epoch": 3.373525504247353,
-      "grad_norm": 0.0025416258722543716,
-      "learning_rate": 0.00019999438965916246,
-      "loss": 46.0,
-      "step": 20948
-    },
-    {
-      "epoch": 3.3736865413261405,
-      "grad_norm": 0.0033665881492197514,
-      "learning_rate": 0.0001999943891232543,
-      "loss": 46.0,
-      "step": 20949
-    },
-    {
-      "epoch": 3.3738475784049275,
-      "grad_norm": 0.0008465584251098335,
-      "learning_rate": 0.00019999438858732055,
-      "loss": 46.0,
-      "step": 20950
-    },
-    {
-      "epoch": 3.374008615483715,
-      "grad_norm": 0.009099594317376614,
-      "learning_rate": 0.00019999438805136125,
-      "loss": 46.0,
-      "step": 20951
-    },
-    {
-      "epoch": 3.3741696525625025,
-      "grad_norm": 0.0026550244074314833,
-      "learning_rate": 0.00019999438751537634,
-      "loss": 46.0,
-      "step": 20952
-    },
-    {
-      "epoch": 3.37433068964129,
-      "grad_norm": 0.000848502735607326,
-      "learning_rate": 0.0001999943869793658,
-      "loss": 46.0,
-      "step": 20953
-    },
-    {
-      "epoch": 3.3744917267200774,
-      "grad_norm": 0.0026028442662209272,
-      "learning_rate": 0.0001999943864433297,
-      "loss": 46.0,
-      "step": 20954
-    },
-    {
-      "epoch": 3.374652763798865,
-      "grad_norm": 0.0013412300031632185,
-      "learning_rate": 0.00019999438590726797,
-      "loss": 46.0,
-      "step": 20955
-    },
-    {
-      "epoch": 3.3748138008776523,
-      "grad_norm": 0.005789418239146471,
-      "learning_rate": 0.00019999438537118068,
-      "loss": 46.0,
-      "step": 20956
-    },
-    {
-      "epoch": 3.3749748379564393,
-      "grad_norm": 0.0027159962337464094,
-      "learning_rate": 0.0001999943848350678,
-      "loss": 46.0,
-      "step": 20957
-    },
-    {
-      "epoch": 3.3751358750352267,
-      "grad_norm": 0.002325424226000905,
-      "learning_rate": 0.00019999438429892932,
-      "loss": 46.0,
-      "step": 20958
-    },
-    {
-      "epoch": 3.375296912114014,
-      "grad_norm": 0.0017380093922838569,
-      "learning_rate": 0.00019999438376276524,
-      "loss": 46.0,
-      "step": 20959
-    },
-    {
-      "epoch": 3.3754579491928016,
-      "grad_norm": 0.0010504063684493303,
-      "learning_rate": 0.00019999438322657558,
-      "loss": 46.0,
-      "step": 20960
-    },
-    {
-      "epoch": 3.375618986271589,
-      "grad_norm": 0.0008453310583718121,
-      "learning_rate": 0.0001999943826903603,
-      "loss": 46.0,
-      "step": 20961
-    },
-    {
-      "epoch": 3.3757800233503765,
-      "grad_norm": 0.009664828889071941,
-      "learning_rate": 0.00019999438215411946,
-      "loss": 46.0,
-      "step": 20962
-    },
-    {
-      "epoch": 3.375941060429164,
-      "grad_norm": 0.005448425188660622,
-      "learning_rate": 0.000199994381617853,
-      "loss": 46.0,
-      "step": 20963
-    },
-    {
-      "epoch": 3.3761020975079514,
-      "grad_norm": 0.0057162814773619175,
-      "learning_rate": 0.00019999438108156094,
-      "loss": 46.0,
-      "step": 20964
-    },
-    {
-      "epoch": 3.3762631345867384,
-      "grad_norm": 0.009896941483020782,
-      "learning_rate": 0.00019999438054524334,
-      "loss": 46.0,
-      "step": 20965
-    },
-    {
-      "epoch": 3.376424171665526,
-      "grad_norm": 0.0012061800807714462,
-      "learning_rate": 0.0001999943800089001,
-      "loss": 46.0,
-      "step": 20966
-    },
-    {
-      "epoch": 3.3765852087443133,
-      "grad_norm": 0.0017518760869279504,
-      "learning_rate": 0.00019999437947253128,
-      "loss": 46.0,
-      "step": 20967
-    },
-    {
-      "epoch": 3.376746245823101,
-      "grad_norm": 0.001236538402736187,
-      "learning_rate": 0.00019999437893613683,
-      "loss": 46.0,
-      "step": 20968
-    },
-    {
-      "epoch": 3.3769072829018882,
-      "grad_norm": 0.002920648083090782,
-      "learning_rate": 0.00019999437839971683,
-      "loss": 46.0,
-      "step": 20969
-    },
-    {
-      "epoch": 3.3770683199806757,
-      "grad_norm": 0.0034797561820596457,
-      "learning_rate": 0.00019999437786327121,
-      "loss": 46.0,
-      "step": 20970
-    },
-    {
-      "epoch": 3.3772293570594627,
-      "grad_norm": 0.004465105012059212,
-      "learning_rate": 0.00019999437732680004,
-      "loss": 46.0,
-      "step": 20971
-    },
-    {
-      "epoch": 3.37739039413825,
-      "grad_norm": 0.0009538336889818311,
-      "learning_rate": 0.00019999437679030325,
-      "loss": 46.0,
-      "step": 20972
-    },
-    {
-      "epoch": 3.3775514312170376,
-      "grad_norm": 0.001914582448080182,
-      "learning_rate": 0.00019999437625378087,
-      "loss": 46.0,
-      "step": 20973
-    },
-    {
-      "epoch": 3.377712468295825,
-      "grad_norm": 0.0038070983719080687,
-      "learning_rate": 0.00019999437571723288,
-      "loss": 46.0,
-      "step": 20974
-    },
-    {
-      "epoch": 3.3778735053746125,
-      "grad_norm": 0.0018001802964136004,
-      "learning_rate": 0.00019999437518065932,
-      "loss": 46.0,
-      "step": 20975
-    },
-    {
-      "epoch": 3.3780345424534,
-      "grad_norm": 0.001459547784179449,
-      "learning_rate": 0.00019999437464406016,
-      "loss": 46.0,
-      "step": 20976
-    },
-    {
-      "epoch": 3.3781955795321874,
-      "grad_norm": 0.0007161207031458616,
-      "learning_rate": 0.00019999437410743537,
-      "loss": 46.0,
-      "step": 20977
-    },
-    {
-      "epoch": 3.378356616610975,
-      "grad_norm": 0.002150939544662833,
-      "learning_rate": 0.00019999437357078503,
-      "loss": 46.0,
-      "step": 20978
-    },
-    {
-      "epoch": 3.3785176536897623,
-      "grad_norm": 0.002507770899683237,
-      "learning_rate": 0.00019999437303410908,
-      "loss": 46.0,
-      "step": 20979
-    },
-    {
-      "epoch": 3.3786786907685493,
-      "grad_norm": 0.0030368685256689787,
-      "learning_rate": 0.00019999437249740756,
-      "loss": 46.0,
-      "step": 20980
-    },
-    {
-      "epoch": 3.3788397278473368,
-      "grad_norm": 0.0024243092630058527,
-      "learning_rate": 0.0001999943719606804,
-      "loss": 46.0,
-      "step": 20981
-    },
-    {
-      "epoch": 3.3790007649261242,
-      "grad_norm": 0.013493322767317295,
-      "learning_rate": 0.00019999437142392766,
-      "loss": 46.0,
-      "step": 20982
-    },
-    {
-      "epoch": 3.3791618020049117,
-      "grad_norm": 0.001397455227561295,
-      "learning_rate": 0.00019999437088714936,
-      "loss": 46.0,
-      "step": 20983
-    },
-    {
-      "epoch": 3.379322839083699,
-      "grad_norm": 0.002155978698283434,
-      "learning_rate": 0.00019999437035034544,
-      "loss": 46.0,
-      "step": 20984
-    },
-    {
-      "epoch": 3.3794838761624866,
-      "grad_norm": 0.0027984476182609797,
-      "learning_rate": 0.00019999436981351596,
-      "loss": 46.0,
-      "step": 20985
-    },
-    {
-      "epoch": 3.3796449132412736,
-      "grad_norm": 0.012609382160007954,
-      "learning_rate": 0.00019999436927666084,
-      "loss": 46.0,
-      "step": 20986
-    },
-    {
-      "epoch": 3.379805950320061,
-      "grad_norm": 0.005818732548505068,
-      "learning_rate": 0.00019999436873978013,
-      "loss": 46.0,
-      "step": 20987
-    },
-    {
-      "epoch": 3.3799669873988485,
-      "grad_norm": 0.004531625658273697,
-      "learning_rate": 0.00019999436820287387,
-      "loss": 46.0,
-      "step": 20988
-    },
-    {
-      "epoch": 3.380128024477636,
-      "grad_norm": 0.007118010427802801,
-      "learning_rate": 0.00019999436766594199,
-      "loss": 46.0,
-      "step": 20989
-    },
-    {
-      "epoch": 3.3802890615564234,
-      "grad_norm": 0.0020114232320338488,
-      "learning_rate": 0.00019999436712898452,
-      "loss": 46.0,
-      "step": 20990
-    },
-    {
-      "epoch": 3.380450098635211,
-      "grad_norm": 0.00118837074842304,
-      "learning_rate": 0.00019999436659200143,
-      "loss": 46.0,
-      "step": 20991
-    },
-    {
-      "epoch": 3.3806111357139983,
-      "grad_norm": 0.006640010047703981,
-      "learning_rate": 0.00019999436605499276,
-      "loss": 46.0,
-      "step": 20992
-    },
-    {
-      "epoch": 3.3807721727927857,
-      "grad_norm": 0.006626845803111792,
-      "learning_rate": 0.00019999436551795853,
-      "loss": 46.0,
-      "step": 20993
-    },
-    {
-      "epoch": 3.3809332098715728,
-      "grad_norm": 0.0017630504444241524,
-      "learning_rate": 0.00019999436498089866,
-      "loss": 46.0,
-      "step": 20994
-    },
-    {
-      "epoch": 3.38109424695036,
-      "grad_norm": 0.008797770366072655,
-      "learning_rate": 0.00019999436444381323,
-      "loss": 46.0,
-      "step": 20995
-    },
-    {
-      "epoch": 3.3812552840291477,
-      "grad_norm": 0.0009718845249153674,
-      "learning_rate": 0.0001999943639067022,
-      "loss": 46.0,
-      "step": 20996
-    },
-    {
-      "epoch": 3.381416321107935,
-      "grad_norm": 0.008871011435985565,
-      "learning_rate": 0.00019999436336956558,
-      "loss": 46.0,
-      "step": 20997
-    },
-    {
-      "epoch": 3.3815773581867226,
-      "grad_norm": 0.0015421169809997082,
-      "learning_rate": 0.00019999436283240336,
-      "loss": 46.0,
-      "step": 20998
-    },
-    {
-      "epoch": 3.38173839526551,
-      "grad_norm": 0.0038646755274385214,
-      "learning_rate": 0.00019999436229521555,
-      "loss": 46.0,
-      "step": 20999
-    },
-    {
-      "epoch": 3.3818994323442975,
-      "grad_norm": 0.004479492548853159,
-      "learning_rate": 0.00019999436175800216,
-      "loss": 46.0,
-      "step": 21000
-    },
-    {
-      "epoch": 3.3820604694230845,
-      "grad_norm": 0.005288648419082165,
-      "learning_rate": 0.00019999436122076315,
-      "loss": 46.0,
-      "step": 21001
-    },
-    {
-      "epoch": 3.382221506501872,
-      "grad_norm": 0.000657967641018331,
-      "learning_rate": 0.00019999436068349855,
-      "loss": 46.0,
-      "step": 21002
-    },
-    {
-      "epoch": 3.3823825435806594,
-      "grad_norm": 0.0023342259228229523,
-      "learning_rate": 0.00019999436014620837,
-      "loss": 46.0,
-      "step": 21003
-    },
-    {
-      "epoch": 3.382543580659447,
-      "grad_norm": 0.0032856722828000784,
-      "learning_rate": 0.0001999943596088926,
-      "loss": 46.0,
-      "step": 21004
-    },
-    {
-      "epoch": 3.3827046177382343,
-      "grad_norm": 0.007843364961445332,
-      "learning_rate": 0.00019999435907155122,
-      "loss": 46.0,
-      "step": 21005
-    },
-    {
-      "epoch": 3.3828656548170217,
-      "grad_norm": 0.008105429820716381,
-      "learning_rate": 0.00019999435853418422,
-      "loss": 46.0,
-      "step": 21006
-    },
-    {
-      "epoch": 3.383026691895809,
-      "grad_norm": 0.007616278249770403,
-      "learning_rate": 0.00019999435799679166,
-      "loss": 46.0,
-      "step": 21007
-    },
-    {
-      "epoch": 3.3831877289745966,
-      "grad_norm": 0.0010869266698136926,
-      "learning_rate": 0.0001999943574593735,
-      "loss": 46.0,
-      "step": 21008
-    },
-    {
-      "epoch": 3.3833487660533836,
-      "grad_norm": 0.0077989003621041775,
-      "learning_rate": 0.00019999435692192975,
-      "loss": 46.0,
-      "step": 21009
-    },
-    {
-      "epoch": 3.383509803132171,
-      "grad_norm": 0.006215512752532959,
-      "learning_rate": 0.00019999435638446043,
-      "loss": 46.0,
-      "step": 21010
-    },
-    {
-      "epoch": 3.3836708402109585,
-      "grad_norm": 0.0038109011948108673,
-      "learning_rate": 0.0001999943558469655,
-      "loss": 46.0,
-      "step": 21011
-    },
-    {
-      "epoch": 3.383831877289746,
-      "grad_norm": 0.00961876567453146,
-      "learning_rate": 0.00019999435530944498,
-      "loss": 46.0,
-      "step": 21012
-    },
-    {
-      "epoch": 3.3839929143685334,
-      "grad_norm": 0.0007203230634331703,
-      "learning_rate": 0.00019999435477189884,
-      "loss": 46.0,
-      "step": 21013
-    },
-    {
-      "epoch": 3.384153951447321,
-      "grad_norm": 0.005657418165355921,
-      "learning_rate": 0.00019999435423432714,
-      "loss": 46.0,
-      "step": 21014
-    },
-    {
-      "epoch": 3.384314988526108,
-      "grad_norm": 0.004697679076343775,
-      "learning_rate": 0.00019999435369672983,
-      "loss": 46.0,
-      "step": 21015
-    },
-    {
-      "epoch": 3.3844760256048954,
-      "grad_norm": 0.005938044749200344,
-      "learning_rate": 0.0001999943531591069,
-      "loss": 46.0,
-      "step": 21016
-    },
-    {
-      "epoch": 3.384637062683683,
-      "grad_norm": 0.0028380025178194046,
-      "learning_rate": 0.0001999943526214584,
-      "loss": 46.0,
-      "step": 21017
-    },
-    {
-      "epoch": 3.3847980997624703,
-      "grad_norm": 0.003145261202007532,
-      "learning_rate": 0.00019999435208378432,
-      "loss": 46.0,
-      "step": 21018
-    },
-    {
-      "epoch": 3.3849591368412577,
-      "grad_norm": 0.006757540628314018,
-      "learning_rate": 0.00019999435154608467,
-      "loss": 46.0,
-      "step": 21019
-    },
-    {
-      "epoch": 3.385120173920045,
-      "grad_norm": 0.010384135879576206,
-      "learning_rate": 0.00019999435100835936,
-      "loss": 46.0,
-      "step": 21020
-    },
-    {
-      "epoch": 3.3852812109988326,
-      "grad_norm": 0.006183552090078592,
-      "learning_rate": 0.0001999943504706085,
-      "loss": 46.0,
-      "step": 21021
-    },
-    {
-      "epoch": 3.38544224807762,
-      "grad_norm": 0.0012846965109929442,
-      "learning_rate": 0.00019999434993283203,
-      "loss": 46.0,
-      "step": 21022
-    },
-    {
-      "epoch": 3.385603285156407,
-      "grad_norm": 0.00730533292517066,
-      "learning_rate": 0.00019999434939503,
-      "loss": 46.0,
-      "step": 21023
-    },
-    {
-      "epoch": 3.3857643222351945,
-      "grad_norm": 0.005410051438957453,
-      "learning_rate": 0.00019999434885720235,
-      "loss": 46.0,
-      "step": 21024
-    },
-    {
-      "epoch": 3.385925359313982,
-      "grad_norm": 0.0018060830188915133,
-      "learning_rate": 0.00019999434831934908,
-      "loss": 46.0,
-      "step": 21025
-    },
-    {
-      "epoch": 3.3860863963927694,
-      "grad_norm": 0.009844711050391197,
-      "learning_rate": 0.00019999434778147026,
-      "loss": 46.0,
-      "step": 21026
-    },
-    {
-      "epoch": 3.386247433471557,
-      "grad_norm": 0.0045937164686620235,
-      "learning_rate": 0.00019999434724356582,
-      "loss": 46.0,
-      "step": 21027
-    },
-    {
-      "epoch": 3.3864084705503443,
-      "grad_norm": 0.0038391142152249813,
-      "learning_rate": 0.00019999434670563582,
-      "loss": 46.0,
-      "step": 21028
-    },
-    {
-      "epoch": 3.386569507629132,
-      "grad_norm": 0.0021910294890403748,
-      "learning_rate": 0.00019999434616768018,
-      "loss": 46.0,
-      "step": 21029
-    },
-    {
-      "epoch": 3.386730544707919,
-      "grad_norm": 0.007108280900865793,
-      "learning_rate": 0.00019999434562969896,
-      "loss": 46.0,
-      "step": 21030
-    },
-    {
-      "epoch": 3.3868915817867062,
-      "grad_norm": 0.007247003726661205,
-      "learning_rate": 0.00019999434509169217,
-      "loss": 46.0,
-      "step": 21031
-    },
-    {
-      "epoch": 3.3870526188654937,
-      "grad_norm": 0.004079220816493034,
-      "learning_rate": 0.00019999434455365977,
-      "loss": 46.0,
-      "step": 21032
-    },
-    {
-      "epoch": 3.387213655944281,
-      "grad_norm": 0.0011850569862872362,
-      "learning_rate": 0.00019999434401560178,
-      "loss": 46.0,
-      "step": 21033
-    },
-    {
-      "epoch": 3.3873746930230686,
-      "grad_norm": 0.003189290640875697,
-      "learning_rate": 0.0001999943434775182,
-      "loss": 46.0,
-      "step": 21034
-    },
-    {
-      "epoch": 3.387535730101856,
-      "grad_norm": 0.0027853238862007856,
-      "learning_rate": 0.00019999434293940902,
-      "loss": 46.0,
-      "step": 21035
-    },
-    {
-      "epoch": 3.3876967671806435,
-      "grad_norm": 0.0017952574416995049,
-      "learning_rate": 0.00019999434240127424,
-      "loss": 46.0,
-      "step": 21036
-    },
-    {
-      "epoch": 3.387857804259431,
-      "grad_norm": 0.0017393914749845862,
-      "learning_rate": 0.00019999434186311388,
-      "loss": 46.0,
-      "step": 21037
-    },
-    {
-      "epoch": 3.388018841338218,
-      "grad_norm": 0.008094176650047302,
-      "learning_rate": 0.00019999434132492793,
-      "loss": 46.0,
-      "step": 21038
-    },
-    {
-      "epoch": 3.3881798784170054,
-      "grad_norm": 0.010397003963589668,
-      "learning_rate": 0.00019999434078671636,
-      "loss": 46.0,
-      "step": 21039
-    },
-    {
-      "epoch": 3.388340915495793,
-      "grad_norm": 0.001662988099269569,
-      "learning_rate": 0.00019999434024847924,
-      "loss": 46.0,
-      "step": 21040
-    },
-    {
-      "epoch": 3.3885019525745803,
-      "grad_norm": 0.0037286649458110332,
-      "learning_rate": 0.0001999943397102165,
-      "loss": 46.0,
-      "step": 21041
-    },
-    {
-      "epoch": 3.3886629896533678,
-      "grad_norm": 0.002169365994632244,
-      "learning_rate": 0.00019999433917192817,
-      "loss": 46.0,
-      "step": 21042
-    },
-    {
-      "epoch": 3.388824026732155,
-      "grad_norm": 0.010649395175278187,
-      "learning_rate": 0.00019999433863361423,
-      "loss": 46.0,
-      "step": 21043
-    },
-    {
-      "epoch": 3.3889850638109422,
-      "grad_norm": 0.004098736215382814,
-      "learning_rate": 0.00019999433809527473,
-      "loss": 46.0,
-      "step": 21044
-    },
-    {
-      "epoch": 3.3891461008897297,
-      "grad_norm": 0.002906162291765213,
-      "learning_rate": 0.00019999433755690959,
-      "loss": 46.0,
-      "step": 21045
-    },
-    {
-      "epoch": 3.389307137968517,
-      "grad_norm": 0.0025292462669312954,
-      "learning_rate": 0.0001999943370185189,
-      "loss": 46.0,
-      "step": 21046
-    },
-    {
-      "epoch": 3.3894681750473046,
-      "grad_norm": 0.0036772822495549917,
-      "learning_rate": 0.0001999943364801026,
-      "loss": 46.0,
-      "step": 21047
-    },
-    {
-      "epoch": 3.389629212126092,
-      "grad_norm": 0.0025302080903202295,
-      "learning_rate": 0.00019999433594166072,
-      "loss": 46.0,
-      "step": 21048
-    },
-    {
-      "epoch": 3.3897902492048795,
-      "grad_norm": 0.011087237857282162,
-      "learning_rate": 0.00019999433540319323,
-      "loss": 46.0,
-      "step": 21049
-    },
-    {
-      "epoch": 3.389951286283667,
-      "grad_norm": 0.003570837900042534,
-      "learning_rate": 0.00019999433486470017,
-      "loss": 46.0,
-      "step": 21050
-    },
-    {
-      "epoch": 3.3901123233624544,
-      "grad_norm": 0.005096086300909519,
-      "learning_rate": 0.00019999433432618145,
-      "loss": 46.0,
-      "step": 21051
-    },
-    {
-      "epoch": 3.390273360441242,
-      "grad_norm": 0.011205860413610935,
-      "learning_rate": 0.0001999943337876372,
-      "loss": 46.0,
-      "step": 21052
-    },
-    {
-      "epoch": 3.390434397520029,
-      "grad_norm": 0.0014744886429980397,
-      "learning_rate": 0.00019999433324906736,
-      "loss": 46.0,
-      "step": 21053
-    },
-    {
-      "epoch": 3.3905954345988163,
-      "grad_norm": 0.0060256351716816425,
-      "learning_rate": 0.0001999943327104719,
-      "loss": 46.0,
-      "step": 21054
-    },
-    {
-      "epoch": 3.3907564716776037,
-      "grad_norm": 0.0015035916585475206,
-      "learning_rate": 0.00019999433217185084,
-      "loss": 46.0,
-      "step": 21055
-    },
-    {
-      "epoch": 3.390917508756391,
-      "grad_norm": 0.0009834583615884185,
-      "learning_rate": 0.0001999943316332042,
-      "loss": 46.0,
-      "step": 21056
-    },
-    {
-      "epoch": 3.3910785458351786,
-      "grad_norm": 0.0048959203995764256,
-      "learning_rate": 0.000199994331094532,
-      "loss": 46.0,
-      "step": 21057
-    },
-    {
-      "epoch": 3.391239582913966,
-      "grad_norm": 0.002037177560850978,
-      "learning_rate": 0.00019999433055583414,
-      "loss": 46.0,
-      "step": 21058
-    },
-    {
-      "epoch": 3.391400619992753,
-      "grad_norm": 0.0025556283071637154,
-      "learning_rate": 0.00019999433001711072,
-      "loss": 46.0,
-      "step": 21059
-    },
-    {
-      "epoch": 3.3915616570715406,
-      "grad_norm": 0.012726373039186,
-      "learning_rate": 0.00019999432947836172,
-      "loss": 46.0,
-      "step": 21060
-    },
-    {
-      "epoch": 3.391722694150328,
-      "grad_norm": 0.0016494044102728367,
-      "learning_rate": 0.00019999432893958713,
-      "loss": 46.0,
-      "step": 21061
-    },
-    {
-      "epoch": 3.3918837312291155,
-      "grad_norm": 0.006000559311360121,
-      "learning_rate": 0.00019999432840078692,
-      "loss": 46.0,
-      "step": 21062
-    },
-    {
-      "epoch": 3.392044768307903,
-      "grad_norm": 0.003741178661584854,
-      "learning_rate": 0.0001999943278619611,
-      "loss": 46.0,
-      "step": 21063
-    },
-    {
-      "epoch": 3.3922058053866904,
-      "grad_norm": 0.002221198519691825,
-      "learning_rate": 0.00019999432732310975,
-      "loss": 46.0,
-      "step": 21064
-    },
-    {
-      "epoch": 3.392366842465478,
-      "grad_norm": 0.0015539692249149084,
-      "learning_rate": 0.00019999432678423275,
-      "loss": 46.0,
-      "step": 21065
-    },
-    {
-      "epoch": 3.3925278795442653,
-      "grad_norm": 0.0028189888689666986,
-      "learning_rate": 0.0001999943262453302,
-      "loss": 46.0,
-      "step": 21066
-    },
-    {
-      "epoch": 3.3926889166230523,
-      "grad_norm": 0.004286462441086769,
-      "learning_rate": 0.00019999432570640203,
-      "loss": 46.0,
-      "step": 21067
-    },
-    {
-      "epoch": 3.3928499537018397,
-      "grad_norm": 0.002745330799371004,
-      "learning_rate": 0.00019999432516744828,
-      "loss": 46.0,
-      "step": 21068
-    },
-    {
-      "epoch": 3.393010990780627,
-      "grad_norm": 0.009761913679540157,
-      "learning_rate": 0.0001999943246284689,
-      "loss": 46.0,
-      "step": 21069
-    },
-    {
-      "epoch": 3.3931720278594146,
-      "grad_norm": 0.009391088038682938,
-      "learning_rate": 0.00019999432408946395,
-      "loss": 46.0,
-      "step": 21070
-    },
-    {
-      "epoch": 3.393333064938202,
-      "grad_norm": 0.005231519229710102,
-      "learning_rate": 0.0001999943235504334,
-      "loss": 46.0,
-      "step": 21071
-    },
-    {
-      "epoch": 3.3934941020169895,
-      "grad_norm": 0.0017369450069963932,
-      "learning_rate": 0.00019999432301137727,
-      "loss": 46.0,
-      "step": 21072
-    },
-    {
-      "epoch": 3.393655139095777,
-      "grad_norm": 0.0015287279384210706,
-      "learning_rate": 0.00019999432247229555,
-      "loss": 46.0,
-      "step": 21073
-    },
-    {
-      "epoch": 3.393816176174564,
-      "grad_norm": 0.005664782132953405,
-      "learning_rate": 0.00019999432193318822,
-      "loss": 46.0,
-      "step": 21074
-    },
-    {
-      "epoch": 3.3939772132533514,
-      "grad_norm": 0.00287757464684546,
-      "learning_rate": 0.00019999432139405533,
-      "loss": 46.0,
-      "step": 21075
-    },
-    {
-      "epoch": 3.394138250332139,
-      "grad_norm": 0.00043558908510021865,
-      "learning_rate": 0.0001999943208548968,
-      "loss": 46.0,
-      "step": 21076
-    },
-    {
-      "epoch": 3.3942992874109263,
-      "grad_norm": 0.006284990813583136,
-      "learning_rate": 0.00019999432031571273,
-      "loss": 46.0,
-      "step": 21077
-    },
-    {
-      "epoch": 3.394460324489714,
-      "grad_norm": 0.0016005446668714285,
-      "learning_rate": 0.00019999431977650302,
-      "loss": 46.0,
-      "step": 21078
-    },
-    {
-      "epoch": 3.3946213615685012,
-      "grad_norm": 0.005038704257458448,
-      "learning_rate": 0.00019999431923726772,
-      "loss": 46.0,
-      "step": 21079
-    },
-    {
-      "epoch": 3.3947823986472887,
-      "grad_norm": 0.0013049077242612839,
-      "learning_rate": 0.00019999431869800684,
-      "loss": 46.0,
-      "step": 21080
-    },
-    {
-      "epoch": 3.394943435726076,
-      "grad_norm": 0.0030376403592526913,
-      "learning_rate": 0.00019999431815872037,
-      "loss": 46.0,
-      "step": 21081
-    },
-    {
-      "epoch": 3.395104472804863,
-      "grad_norm": 0.001290869782678783,
-      "learning_rate": 0.0001999943176194083,
-      "loss": 46.0,
-      "step": 21082
-    },
-    {
-      "epoch": 3.3952655098836506,
-      "grad_norm": 0.0034194362815469503,
-      "learning_rate": 0.00019999431708007064,
-      "loss": 46.0,
-      "step": 21083
-    },
-    {
-      "epoch": 3.395426546962438,
-      "grad_norm": 0.0015584477223455906,
-      "learning_rate": 0.0001999943165407074,
-      "loss": 46.0,
-      "step": 21084
-    },
-    {
-      "epoch": 3.3955875840412255,
-      "grad_norm": 0.008205071091651917,
-      "learning_rate": 0.00019999431600131856,
-      "loss": 46.0,
-      "step": 21085
-    },
-    {
-      "epoch": 3.395748621120013,
-      "grad_norm": 0.005005842074751854,
-      "learning_rate": 0.00019999431546190407,
-      "loss": 46.0,
-      "step": 21086
-    },
-    {
-      "epoch": 3.3959096581988004,
-      "grad_norm": 0.00037571846041828394,
-      "learning_rate": 0.00019999431492246408,
-      "loss": 46.0,
-      "step": 21087
-    },
-    {
-      "epoch": 3.3960706952775874,
-      "grad_norm": 0.0031257993541657925,
-      "learning_rate": 0.00019999431438299844,
-      "loss": 46.0,
-      "step": 21088
-    },
-    {
-      "epoch": 3.396231732356375,
-      "grad_norm": 0.002363610314205289,
-      "learning_rate": 0.0001999943138435072,
-      "loss": 46.0,
-      "step": 21089
-    },
-    {
-      "epoch": 3.3963927694351623,
-      "grad_norm": 0.004282078705728054,
-      "learning_rate": 0.00019999431330399041,
-      "loss": 46.0,
-      "step": 21090
-    },
-    {
-      "epoch": 3.39655380651395,
-      "grad_norm": 0.004170003812760115,
-      "learning_rate": 0.000199994312764448,
-      "loss": 46.0,
-      "step": 21091
-    },
-    {
-      "epoch": 3.3967148435927372,
-      "grad_norm": 0.004156829323619604,
-      "learning_rate": 0.00019999431222487998,
-      "loss": 46.0,
-      "step": 21092
-    },
-    {
-      "epoch": 3.3968758806715247,
-      "grad_norm": 0.0017569194315001369,
-      "learning_rate": 0.00019999431168528638,
-      "loss": 46.0,
-      "step": 21093
-    },
-    {
-      "epoch": 3.397036917750312,
-      "grad_norm": 0.009720665402710438,
-      "learning_rate": 0.0001999943111456672,
-      "loss": 46.0,
-      "step": 21094
-    },
-    {
-      "epoch": 3.3971979548290996,
-      "grad_norm": 0.0007093818858265877,
-      "learning_rate": 0.00019999431060602243,
-      "loss": 46.0,
-      "step": 21095
-    },
-    {
-      "epoch": 3.397358991907887,
-      "grad_norm": 0.002441916149109602,
-      "learning_rate": 0.00019999431006635204,
-      "loss": 46.0,
-      "step": 21096
-    },
-    {
-      "epoch": 3.397520028986674,
-      "grad_norm": 0.003565569408237934,
-      "learning_rate": 0.00019999430952665607,
-      "loss": 46.0,
-      "step": 21097
-    },
-    {
-      "epoch": 3.3976810660654615,
-      "grad_norm": 0.0024843630380928516,
-      "learning_rate": 0.0001999943089869345,
-      "loss": 46.0,
-      "step": 21098
-    },
-    {
-      "epoch": 3.397842103144249,
-      "grad_norm": 0.0024970900267362595,
-      "learning_rate": 0.00019999430844718736,
-      "loss": 46.0,
-      "step": 21099
-    },
-    {
-      "epoch": 3.3980031402230364,
-      "grad_norm": 0.00679752416908741,
-      "learning_rate": 0.0001999943079074146,
-      "loss": 46.0,
-      "step": 21100
-    },
-    {
-      "epoch": 3.398164177301824,
-      "grad_norm": 0.0013325853506103158,
-      "learning_rate": 0.00019999430736761627,
-      "loss": 46.0,
-      "step": 21101
-    },
-    {
-      "epoch": 3.3983252143806113,
-      "grad_norm": 0.0015867049805819988,
-      "learning_rate": 0.0001999943068277923,
-      "loss": 46.0,
-      "step": 21102
-    },
-    {
-      "epoch": 3.3984862514593983,
-      "grad_norm": 0.002462285803630948,
-      "learning_rate": 0.0001999943062879428,
-      "loss": 46.0,
-      "step": 21103
-    },
-    {
-      "epoch": 3.3986472885381858,
-      "grad_norm": 0.0008760459604673088,
-      "learning_rate": 0.00019999430574806768,
-      "loss": 46.0,
-      "step": 21104
-    },
-    {
-      "epoch": 3.398808325616973,
-      "grad_norm": 0.008111543022096157,
-      "learning_rate": 0.00019999430520816695,
-      "loss": 46.0,
-      "step": 21105
-    },
-    {
-      "epoch": 3.3989693626957607,
-      "grad_norm": 0.0011252565309405327,
-      "learning_rate": 0.00019999430466824064,
-      "loss": 46.0,
-      "step": 21106
-    },
-    {
-      "epoch": 3.399130399774548,
-      "grad_norm": 0.0054007177241146564,
-      "learning_rate": 0.00019999430412828874,
-      "loss": 46.0,
-      "step": 21107
-    },
-    {
-      "epoch": 3.3992914368533356,
-      "grad_norm": 0.0022789279464632273,
-      "learning_rate": 0.00019999430358831125,
-      "loss": 46.0,
-      "step": 21108
-    },
-    {
-      "epoch": 3.399452473932123,
-      "grad_norm": 0.0034397232811897993,
-      "learning_rate": 0.00019999430304830815,
-      "loss": 46.0,
-      "step": 21109
-    },
-    {
-      "epoch": 3.3996135110109105,
-      "grad_norm": 0.002536230254918337,
-      "learning_rate": 0.00019999430250827946,
-      "loss": 46.0,
-      "step": 21110
-    },
-    {
-      "epoch": 3.3997745480896975,
-      "grad_norm": 0.0027080022264271975,
-      "learning_rate": 0.0001999943019682252,
-      "loss": 46.0,
-      "step": 21111
-    },
-    {
-      "epoch": 3.399935585168485,
-      "grad_norm": 0.002194308442994952,
-      "learning_rate": 0.00019999430142814533,
-      "loss": 46.0,
-      "step": 21112
-    },
-    {
-      "epoch": 3.4000966222472724,
-      "grad_norm": 0.0021683378145098686,
-      "learning_rate": 0.00019999430088803988,
-      "loss": 46.0,
-      "step": 21113
-    },
-    {
-      "epoch": 3.40025765932606,
-      "grad_norm": 0.0029717704746872187,
-      "learning_rate": 0.0001999943003479088,
-      "loss": 46.0,
-      "step": 21114
-    },
-    {
-      "epoch": 3.4004186964048473,
-      "grad_norm": 0.001527405926026404,
-      "learning_rate": 0.00019999429980775216,
-      "loss": 46.0,
-      "step": 21115
-    },
-    {
-      "epoch": 3.4005797334836347,
-      "grad_norm": 0.002910199109464884,
-      "learning_rate": 0.0001999942992675699,
-      "loss": 46.0,
-      "step": 21116
-    },
-    {
-      "epoch": 3.400740770562422,
-      "grad_norm": 0.004136513452976942,
-      "learning_rate": 0.00019999429872736207,
-      "loss": 46.0,
-      "step": 21117
-    },
-    {
-      "epoch": 3.400901807641209,
-      "grad_norm": 0.007301343604922295,
-      "learning_rate": 0.00019999429818712866,
-      "loss": 46.0,
-      "step": 21118
-    },
-    {
-      "epoch": 3.4010628447199966,
-      "grad_norm": 0.008149239234626293,
-      "learning_rate": 0.0001999942976468696,
-      "loss": 46.0,
-      "step": 21119
-    },
-    {
-      "epoch": 3.401223881798784,
-      "grad_norm": 0.0015666043618693948,
-      "learning_rate": 0.000199994297106585,
-      "loss": 46.0,
-      "step": 21120
-    },
-    {
-      "epoch": 3.4013849188775716,
-      "grad_norm": 0.019910618662834167,
-      "learning_rate": 0.0001999942965662748,
-      "loss": 46.0,
-      "step": 21121
-    },
-    {
-      "epoch": 3.401545955956359,
-      "grad_norm": 0.0013587672729045153,
-      "learning_rate": 0.00019999429602593897,
-      "loss": 46.0,
-      "step": 21122
-    },
-    {
-      "epoch": 3.4017069930351465,
-      "grad_norm": 0.0022173745092004538,
-      "learning_rate": 0.00019999429548557757,
-      "loss": 46.0,
-      "step": 21123
-    },
-    {
-      "epoch": 3.401868030113934,
-      "grad_norm": 0.008212946355342865,
-      "learning_rate": 0.0001999942949451906,
-      "loss": 46.0,
-      "step": 21124
-    },
-    {
-      "epoch": 3.4020290671927214,
-      "grad_norm": 0.0037013136316090822,
-      "learning_rate": 0.000199994294404778,
-      "loss": 46.0,
-      "step": 21125
-    },
-    {
-      "epoch": 3.4021901042715084,
-      "grad_norm": 0.0040328605100512505,
-      "learning_rate": 0.00019999429386433984,
-      "loss": 46.0,
-      "step": 21126
-    },
-    {
-      "epoch": 3.402351141350296,
-      "grad_norm": 0.0009341384284198284,
-      "learning_rate": 0.00019999429332387604,
-      "loss": 46.0,
-      "step": 21127
-    },
-    {
-      "epoch": 3.4025121784290833,
-      "grad_norm": 0.006971676833927631,
-      "learning_rate": 0.0001999942927833867,
-      "loss": 46.0,
-      "step": 21128
-    },
-    {
-      "epoch": 3.4026732155078707,
-      "grad_norm": 0.007393034175038338,
-      "learning_rate": 0.00019999429224287174,
-      "loss": 46.0,
-      "step": 21129
-    },
-    {
-      "epoch": 3.402834252586658,
-      "grad_norm": 0.006465106271207333,
-      "learning_rate": 0.0001999942917023312,
-      "loss": 46.0,
-      "step": 21130
-    },
-    {
-      "epoch": 3.4029952896654456,
-      "grad_norm": 0.001826530322432518,
-      "learning_rate": 0.00019999429116176502,
-      "loss": 46.0,
-      "step": 21131
-    },
-    {
-      "epoch": 3.4031563267442326,
-      "grad_norm": 0.0026242146268486977,
-      "learning_rate": 0.0001999942906211733,
-      "loss": 46.0,
-      "step": 21132
-    },
-    {
-      "epoch": 3.40331736382302,
-      "grad_norm": 0.0050507173873484135,
-      "learning_rate": 0.00019999429008055598,
-      "loss": 46.0,
-      "step": 21133
-    },
-    {
-      "epoch": 3.4034784009018075,
-      "grad_norm": 0.00919990986585617,
-      "learning_rate": 0.00019999428953991304,
-      "loss": 46.0,
-      "step": 21134
-    },
-    {
-      "epoch": 3.403639437980595,
-      "grad_norm": 0.010222394950687885,
-      "learning_rate": 0.0001999942889992445,
-      "loss": 46.0,
-      "step": 21135
-    },
-    {
-      "epoch": 3.4038004750593824,
-      "grad_norm": 0.0015832679346203804,
-      "learning_rate": 0.0001999942884585504,
-      "loss": 46.0,
-      "step": 21136
-    },
-    {
-      "epoch": 3.40396151213817,
-      "grad_norm": 0.00965300016105175,
-      "learning_rate": 0.00019999428791783071,
-      "loss": 46.0,
-      "step": 21137
-    },
-    {
-      "epoch": 3.4041225492169573,
-      "grad_norm": 0.001704067224636674,
-      "learning_rate": 0.00019999428737708542,
-      "loss": 46.0,
-      "step": 21138
-    },
-    {
-      "epoch": 3.404283586295745,
-      "grad_norm": 0.0012747389264404774,
-      "learning_rate": 0.00019999428683631452,
-      "loss": 46.0,
-      "step": 21139
-    },
-    {
-      "epoch": 3.404444623374532,
-      "grad_norm": 0.0014449880691245198,
-      "learning_rate": 0.00019999428629551802,
-      "loss": 46.0,
-      "step": 21140
-    },
-    {
-      "epoch": 3.4046056604533192,
-      "grad_norm": 0.0006542811752296984,
-      "learning_rate": 0.00019999428575469594,
-      "loss": 46.0,
-      "step": 21141
-    },
-    {
-      "epoch": 3.4047666975321067,
-      "grad_norm": 0.0023396213073283434,
-      "learning_rate": 0.00019999428521384828,
-      "loss": 46.0,
-      "step": 21142
-    },
-    {
-      "epoch": 3.404927734610894,
-      "grad_norm": 0.001497275079600513,
-      "learning_rate": 0.000199994284672975,
-      "loss": 46.0,
-      "step": 21143
-    },
-    {
-      "epoch": 3.4050887716896816,
-      "grad_norm": 0.0011699807364493608,
-      "learning_rate": 0.00019999428413207615,
-      "loss": 46.0,
-      "step": 21144
-    },
-    {
-      "epoch": 3.405249808768469,
-      "grad_norm": 0.005118026863783598,
-      "learning_rate": 0.0001999942835911517,
-      "loss": 46.0,
-      "step": 21145
-    },
-    {
-      "epoch": 3.4054108458472565,
-      "grad_norm": 0.0187362190335989,
-      "learning_rate": 0.00019999428305020165,
-      "loss": 46.0,
-      "step": 21146
-    },
-    {
-      "epoch": 3.4055718829260435,
-      "grad_norm": 0.008424741216003895,
-      "learning_rate": 0.000199994282509226,
-      "loss": 46.0,
-      "step": 21147
-    },
-    {
-      "epoch": 3.405732920004831,
-      "grad_norm": 0.0019717966206371784,
-      "learning_rate": 0.00019999428196822478,
-      "loss": 46.0,
-      "step": 21148
-    },
-    {
-      "epoch": 3.4058939570836184,
-      "grad_norm": 0.010692927986383438,
-      "learning_rate": 0.00019999428142719795,
-      "loss": 46.0,
-      "step": 21149
-    },
-    {
-      "epoch": 3.406054994162406,
-      "grad_norm": 0.0025357073172926903,
-      "learning_rate": 0.00019999428088614553,
-      "loss": 46.0,
-      "step": 21150
-    },
-    {
-      "epoch": 3.4062160312411933,
-      "grad_norm": 0.0007546897977590561,
-      "learning_rate": 0.00019999428034506752,
-      "loss": 46.0,
-      "step": 21151
-    },
-    {
-      "epoch": 3.4063770683199808,
-      "grad_norm": 0.0032554753124713898,
-      "learning_rate": 0.00019999427980396393,
-      "loss": 46.0,
-      "step": 21152
-    },
-    {
-      "epoch": 3.4065381053987682,
-      "grad_norm": 0.002403033198788762,
-      "learning_rate": 0.0001999942792628347,
-      "loss": 46.0,
-      "step": 21153
-    },
-    {
-      "epoch": 3.4066991424775557,
-      "grad_norm": 0.007734985090792179,
-      "learning_rate": 0.00019999427872167993,
-      "loss": 46.0,
-      "step": 21154
-    },
-    {
-      "epoch": 3.4068601795563427,
-      "grad_norm": 0.006565115414559841,
-      "learning_rate": 0.00019999427818049955,
-      "loss": 46.0,
-      "step": 21155
-    },
-    {
-      "epoch": 3.40702121663513,
-      "grad_norm": 0.005572181195020676,
-      "learning_rate": 0.00019999427763929355,
-      "loss": 46.0,
-      "step": 21156
-    },
-    {
-      "epoch": 3.4071822537139176,
-      "grad_norm": 0.0022491810377687216,
-      "learning_rate": 0.00019999427709806197,
-      "loss": 46.0,
-      "step": 21157
-    },
-    {
-      "epoch": 3.407343290792705,
-      "grad_norm": 0.0040094442665576935,
-      "learning_rate": 0.0001999942765568048,
-      "loss": 46.0,
-      "step": 21158
-    },
-    {
-      "epoch": 3.4075043278714925,
-      "grad_norm": 0.012897714972496033,
-      "learning_rate": 0.00019999427601552204,
-      "loss": 46.0,
-      "step": 21159
-    },
-    {
-      "epoch": 3.40766536495028,
-      "grad_norm": 0.004199519287794828,
-      "learning_rate": 0.0001999942754742137,
-      "loss": 46.0,
-      "step": 21160
-    },
-    {
-      "epoch": 3.407826402029067,
-      "grad_norm": 0.0031029866077005863,
-      "learning_rate": 0.00019999427493287974,
-      "loss": 46.0,
-      "step": 21161
-    },
-    {
-      "epoch": 3.4079874391078544,
-      "grad_norm": 0.003604508936405182,
-      "learning_rate": 0.00019999427439152022,
-      "loss": 46.0,
-      "step": 21162
-    },
-    {
-      "epoch": 3.408148476186642,
-      "grad_norm": 0.0038238721899688244,
-      "learning_rate": 0.00019999427385013508,
-      "loss": 46.0,
-      "step": 21163
-    },
-    {
-      "epoch": 3.4083095132654293,
-      "grad_norm": 0.0029934379272162914,
-      "learning_rate": 0.00019999427330872434,
-      "loss": 46.0,
-      "step": 21164
-    },
-    {
-      "epoch": 3.4084705503442168,
-      "grad_norm": 0.004953837022185326,
-      "learning_rate": 0.00019999427276728803,
-      "loss": 46.0,
-      "step": 21165
-    },
-    {
-      "epoch": 3.408631587423004,
-      "grad_norm": 0.00626890454441309,
-      "learning_rate": 0.0001999942722258261,
-      "loss": 46.0,
-      "step": 21166
-    },
-    {
-      "epoch": 3.4087926245017917,
-      "grad_norm": 0.0042636338621377945,
-      "learning_rate": 0.0001999942716843386,
-      "loss": 46.0,
-      "step": 21167
-    },
-    {
-      "epoch": 3.408953661580579,
-      "grad_norm": 0.0049273790791630745,
-      "learning_rate": 0.0001999942711428255,
-      "loss": 46.0,
-      "step": 21168
-    },
-    {
-      "epoch": 3.4091146986593666,
-      "grad_norm": 0.007492925971746445,
-      "learning_rate": 0.00019999427060128682,
-      "loss": 46.0,
-      "step": 21169
-    },
-    {
-      "epoch": 3.4092757357381536,
-      "grad_norm": 0.005819031503051519,
-      "learning_rate": 0.00019999427005972252,
-      "loss": 46.0,
-      "step": 21170
-    },
-    {
-      "epoch": 3.409436772816941,
-      "grad_norm": 0.0044487239792943,
-      "learning_rate": 0.00019999426951813263,
-      "loss": 46.0,
-      "step": 21171
-    },
-    {
-      "epoch": 3.4095978098957285,
-      "grad_norm": 0.0006704427069053054,
-      "learning_rate": 0.00019999426897651716,
-      "loss": 46.0,
-      "step": 21172
-    },
-    {
-      "epoch": 3.409758846974516,
-      "grad_norm": 0.0037129633128643036,
-      "learning_rate": 0.0001999942684348761,
-      "loss": 46.0,
-      "step": 21173
-    },
-    {
-      "epoch": 3.4099198840533034,
-      "grad_norm": 0.004275023937225342,
-      "learning_rate": 0.00019999426789320943,
-      "loss": 46.0,
-      "step": 21174
-    },
-    {
-      "epoch": 3.410080921132091,
-      "grad_norm": 0.0008190679363906384,
-      "learning_rate": 0.00019999426735151717,
-      "loss": 46.0,
-      "step": 21175
-    },
-    {
-      "epoch": 3.410241958210878,
-      "grad_norm": 0.0014929737662896514,
-      "learning_rate": 0.00019999426680979935,
-      "loss": 46.0,
-      "step": 21176
-    },
-    {
-      "epoch": 3.4104029952896653,
-      "grad_norm": 0.008549555204808712,
-      "learning_rate": 0.00019999426626805589,
-      "loss": 46.0,
-      "step": 21177
-    },
-    {
-      "epoch": 3.4105640323684527,
-      "grad_norm": 0.0017408184939995408,
-      "learning_rate": 0.00019999426572628686,
-      "loss": 46.0,
-      "step": 21178
-    },
-    {
-      "epoch": 3.41072506944724,
-      "grad_norm": 0.0015750967431813478,
-      "learning_rate": 0.00019999426518449223,
-      "loss": 46.0,
-      "step": 21179
-    },
-    {
-      "epoch": 3.4108861065260276,
-      "grad_norm": 0.0037548651453107595,
-      "learning_rate": 0.000199994264642672,
-      "loss": 46.0,
-      "step": 21180
-    },
-    {
-      "epoch": 3.411047143604815,
-      "grad_norm": 0.0021783686242997646,
-      "learning_rate": 0.00019999426410082617,
-      "loss": 46.0,
-      "step": 21181
-    },
-    {
-      "epoch": 3.4112081806836025,
-      "grad_norm": 0.013517848215997219,
-      "learning_rate": 0.00019999426355895477,
-      "loss": 46.0,
-      "step": 21182
-    },
-    {
-      "epoch": 3.41136921776239,
-      "grad_norm": 0.008675516583025455,
-      "learning_rate": 0.00019999426301705778,
-      "loss": 46.0,
-      "step": 21183
-    },
-    {
-      "epoch": 3.411530254841177,
-      "grad_norm": 0.006291860248893499,
-      "learning_rate": 0.00019999426247513518,
-      "loss": 46.0,
-      "step": 21184
-    },
-    {
-      "epoch": 3.4116912919199645,
-      "grad_norm": 0.002247094875201583,
-      "learning_rate": 0.00019999426193318697,
-      "loss": 46.0,
-      "step": 21185
-    },
-    {
-      "epoch": 3.411852328998752,
-      "grad_norm": 0.00042392892646603286,
-      "learning_rate": 0.0001999942613912132,
-      "loss": 46.0,
-      "step": 21186
-    },
-    {
-      "epoch": 3.4120133660775394,
-      "grad_norm": 0.003107220632955432,
-      "learning_rate": 0.00019999426084921383,
-      "loss": 46.0,
-      "step": 21187
-    },
-    {
-      "epoch": 3.412174403156327,
-      "grad_norm": 0.008849323727190495,
-      "learning_rate": 0.00019999426030718886,
-      "loss": 46.0,
-      "step": 21188
-    },
-    {
-      "epoch": 3.4123354402351143,
-      "grad_norm": 0.006033473648130894,
-      "learning_rate": 0.0001999942597651383,
-      "loss": 46.0,
-      "step": 21189
-    },
-    {
-      "epoch": 3.4124964773139017,
-      "grad_norm": 0.0020630371291190386,
-      "learning_rate": 0.00019999425922306215,
-      "loss": 46.0,
-      "step": 21190
-    },
-    {
-      "epoch": 3.4126575143926887,
-      "grad_norm": 0.0033785656560212374,
-      "learning_rate": 0.0001999942586809604,
-      "loss": 46.0,
-      "step": 21191
-    },
-    {
-      "epoch": 3.412818551471476,
-      "grad_norm": 0.013755476102232933,
-      "learning_rate": 0.00019999425813883303,
-      "loss": 46.0,
-      "step": 21192
-    },
-    {
-      "epoch": 3.4129795885502636,
-      "grad_norm": 0.006385887041687965,
-      "learning_rate": 0.00019999425759668012,
-      "loss": 46.0,
-      "step": 21193
-    },
-    {
-      "epoch": 3.413140625629051,
-      "grad_norm": 0.002654741518199444,
-      "learning_rate": 0.0001999942570545016,
-      "loss": 46.0,
-      "step": 21194
-    },
-    {
-      "epoch": 3.4133016627078385,
-      "grad_norm": 0.003198145655915141,
-      "learning_rate": 0.00019999425651229746,
-      "loss": 46.0,
-      "step": 21195
-    },
-    {
-      "epoch": 3.413462699786626,
-      "grad_norm": 0.0029647655319422483,
-      "learning_rate": 0.00019999425597006773,
-      "loss": 46.0,
-      "step": 21196
-    },
-    {
-      "epoch": 3.4136237368654134,
-      "grad_norm": 0.013421309180557728,
-      "learning_rate": 0.00019999425542781242,
-      "loss": 46.0,
-      "step": 21197
-    },
-    {
-      "epoch": 3.413784773944201,
-      "grad_norm": 0.0011526071466505527,
-      "learning_rate": 0.00019999425488553154,
-      "loss": 46.0,
-      "step": 21198
-    },
-    {
-      "epoch": 3.413945811022988,
-      "grad_norm": 0.015555518679320812,
-      "learning_rate": 0.00019999425434322503,
-      "loss": 46.0,
-      "step": 21199
-    },
-    {
-      "epoch": 3.4141068481017753,
-      "grad_norm": 0.002551355864852667,
-      "learning_rate": 0.00019999425380089295,
-      "loss": 46.0,
-      "step": 21200
-    },
-    {
-      "epoch": 3.414267885180563,
-      "grad_norm": 0.0026627839542925358,
-      "learning_rate": 0.00019999425325853526,
-      "loss": 46.0,
-      "step": 21201
-    },
-    {
-      "epoch": 3.4144289222593502,
-      "grad_norm": 0.0009539287420921028,
-      "learning_rate": 0.00019999425271615198,
-      "loss": 46.0,
-      "step": 21202
-    },
-    {
-      "epoch": 3.4145899593381377,
-      "grad_norm": 0.0022598677314817905,
-      "learning_rate": 0.0001999942521737431,
-      "loss": 46.0,
-      "step": 21203
-    },
-    {
-      "epoch": 3.414750996416925,
-      "grad_norm": 0.0080317547544837,
-      "learning_rate": 0.00019999425163130864,
-      "loss": 46.0,
-      "step": 21204
-    },
-    {
-      "epoch": 3.414912033495712,
-      "grad_norm": 0.011915184557437897,
-      "learning_rate": 0.0001999942510888486,
-      "loss": 46.0,
-      "step": 21205
-    },
-    {
-      "epoch": 3.4150730705744996,
-      "grad_norm": 0.005996193736791611,
-      "learning_rate": 0.00019999425054636295,
-      "loss": 46.0,
-      "step": 21206
-    },
-    {
-      "epoch": 3.415234107653287,
-      "grad_norm": 0.004526139236986637,
-      "learning_rate": 0.0001999942500038517,
-      "loss": 46.0,
-      "step": 21207
-    },
-    {
-      "epoch": 3.4153951447320745,
-      "grad_norm": 0.004340293817222118,
-      "learning_rate": 0.00019999424946131486,
-      "loss": 46.0,
-      "step": 21208
-    },
-    {
-      "epoch": 3.415556181810862,
-      "grad_norm": 0.02035890333354473,
-      "learning_rate": 0.00019999424891875241,
-      "loss": 46.0,
-      "step": 21209
-    },
-    {
-      "epoch": 3.4157172188896494,
-      "grad_norm": 0.0010484060039743781,
-      "learning_rate": 0.0001999942483761644,
-      "loss": 46.0,
-      "step": 21210
-    },
-    {
-      "epoch": 3.415878255968437,
-      "grad_norm": 0.0016059390036389232,
-      "learning_rate": 0.0001999942478335508,
-      "loss": 46.0,
-      "step": 21211
-    },
-    {
-      "epoch": 3.4160392930472243,
-      "grad_norm": 0.009394737891852856,
-      "learning_rate": 0.0001999942472909116,
-      "loss": 46.0,
-      "step": 21212
-    },
-    {
-      "epoch": 3.4162003301260118,
-      "grad_norm": 0.0021858173422515392,
-      "learning_rate": 0.00019999424674824678,
-      "loss": 46.0,
-      "step": 21213
-    },
-    {
-      "epoch": 3.4163613672047988,
-      "grad_norm": 0.0031701961997896433,
-      "learning_rate": 0.00019999424620555637,
-      "loss": 46.0,
-      "step": 21214
-    },
-    {
-      "epoch": 3.4165224042835862,
-      "grad_norm": 0.00033398676896467805,
-      "learning_rate": 0.0001999942456628404,
-      "loss": 46.0,
-      "step": 21215
-    },
-    {
-      "epoch": 3.4166834413623737,
-      "grad_norm": 0.0027331404853612185,
-      "learning_rate": 0.0001999942451200988,
-      "loss": 46.0,
-      "step": 21216
-    },
-    {
-      "epoch": 3.416844478441161,
-      "grad_norm": 0.0019151372835040092,
-      "learning_rate": 0.00019999424457733164,
-      "loss": 46.0,
-      "step": 21217
-    },
-    {
-      "epoch": 3.4170055155199486,
-      "grad_norm": 0.006675058975815773,
-      "learning_rate": 0.00019999424403453886,
-      "loss": 46.0,
-      "step": 21218
-    },
-    {
-      "epoch": 3.417166552598736,
-      "grad_norm": 0.002675826195627451,
-      "learning_rate": 0.0001999942434917205,
-      "loss": 46.0,
-      "step": 21219
-    },
-    {
-      "epoch": 3.417327589677523,
-      "grad_norm": 0.005070195533335209,
-      "learning_rate": 0.00019999424294887653,
-      "loss": 46.0,
-      "step": 21220
-    },
-    {
-      "epoch": 3.4174886267563105,
-      "grad_norm": 0.0016517506446689367,
-      "learning_rate": 0.00019999424240600697,
-      "loss": 46.0,
-      "step": 21221
-    },
-    {
-      "epoch": 3.417649663835098,
-      "grad_norm": 0.0018827365711331367,
-      "learning_rate": 0.0001999942418631118,
-      "loss": 46.0,
-      "step": 21222
-    },
-    {
-      "epoch": 3.4178107009138854,
-      "grad_norm": 0.002622930333018303,
-      "learning_rate": 0.0001999942413201911,
-      "loss": 46.0,
-      "step": 21223
-    },
-    {
-      "epoch": 3.417971737992673,
-      "grad_norm": 0.006782460492104292,
-      "learning_rate": 0.0001999942407772448,
-      "loss": 46.0,
-      "step": 21224
-    },
-    {
-      "epoch": 3.4181327750714603,
-      "grad_norm": 0.0010855470318347216,
-      "learning_rate": 0.00019999424023427285,
-      "loss": 46.0,
-      "step": 21225
-    },
-    {
-      "epoch": 3.4182938121502477,
-      "grad_norm": 0.0012147071538493037,
-      "learning_rate": 0.00019999423969127534,
-      "loss": 46.0,
-      "step": 21226
-    },
-    {
-      "epoch": 3.418454849229035,
-      "grad_norm": 0.010797152295708656,
-      "learning_rate": 0.0001999942391482522,
-      "loss": 46.0,
-      "step": 21227
-    },
-    {
-      "epoch": 3.418615886307822,
-      "grad_norm": 0.002079980680719018,
-      "learning_rate": 0.0001999942386052035,
-      "loss": 46.0,
-      "step": 21228
-    },
-    {
-      "epoch": 3.4187769233866097,
-      "grad_norm": 0.0034605904947966337,
-      "learning_rate": 0.0001999942380621292,
-      "loss": 46.0,
-      "step": 21229
-    },
-    {
-      "epoch": 3.418937960465397,
-      "grad_norm": 0.008832719177007675,
-      "learning_rate": 0.00019999423751902932,
-      "loss": 46.0,
-      "step": 21230
-    },
-    {
-      "epoch": 3.4190989975441846,
-      "grad_norm": 0.0039038187824189663,
-      "learning_rate": 0.00019999423697590382,
-      "loss": 46.0,
-      "step": 21231
-    },
-    {
-      "epoch": 3.419260034622972,
-      "grad_norm": 0.0018081951420754194,
-      "learning_rate": 0.00019999423643275274,
-      "loss": 46.0,
-      "step": 21232
-    },
-    {
-      "epoch": 3.4194210717017595,
-      "grad_norm": 0.0006106903892941773,
-      "learning_rate": 0.00019999423588957607,
-      "loss": 46.0,
-      "step": 21233
-    },
-    {
-      "epoch": 3.419582108780547,
-      "grad_norm": 0.001498400466516614,
-      "learning_rate": 0.00019999423534637381,
-      "loss": 46.0,
-      "step": 21234
-    },
-    {
-      "epoch": 3.419743145859334,
-      "grad_norm": 0.0012723600957542658,
-      "learning_rate": 0.00019999423480314597,
-      "loss": 46.0,
-      "step": 21235
-    },
-    {
-      "epoch": 3.4199041829381214,
-      "grad_norm": 0.004716400522738695,
-      "learning_rate": 0.0001999942342598925,
-      "loss": 46.0,
-      "step": 21236
-    },
-    {
-      "epoch": 3.420065220016909,
-      "grad_norm": 0.014160579070448875,
-      "learning_rate": 0.00019999423371661344,
-      "loss": 46.0,
-      "step": 21237
-    },
-    {
-      "epoch": 3.4202262570956963,
-      "grad_norm": 0.014337912201881409,
-      "learning_rate": 0.0001999942331733088,
-      "loss": 46.0,
-      "step": 21238
-    },
-    {
-      "epoch": 3.4203872941744837,
-      "grad_norm": 0.0017312603304162621,
-      "learning_rate": 0.0001999942326299786,
-      "loss": 46.0,
-      "step": 21239
-    },
-    {
-      "epoch": 3.420548331253271,
-      "grad_norm": 0.0028110118582844734,
-      "learning_rate": 0.00019999423208662276,
-      "loss": 46.0,
-      "step": 21240
-    },
-    {
-      "epoch": 3.4207093683320586,
-      "grad_norm": 0.000721524003893137,
-      "learning_rate": 0.00019999423154324136,
-      "loss": 46.0,
-      "step": 21241
-    },
-    {
-      "epoch": 3.420870405410846,
-      "grad_norm": 0.007434132043272257,
-      "learning_rate": 0.00019999423099983433,
-      "loss": 46.0,
-      "step": 21242
-    },
-    {
-      "epoch": 3.421031442489633,
-      "grad_norm": 0.00933587271720171,
-      "learning_rate": 0.0001999942304564017,
-      "loss": 46.0,
-      "step": 21243
-    },
-    {
-      "epoch": 3.4211924795684205,
-      "grad_norm": 0.014096463099122047,
-      "learning_rate": 0.00019999422991294352,
-      "loss": 46.0,
-      "step": 21244
-    },
-    {
-      "epoch": 3.421353516647208,
-      "grad_norm": 0.003654714673757553,
-      "learning_rate": 0.00019999422936945973,
-      "loss": 46.0,
-      "step": 21245
-    },
-    {
-      "epoch": 3.4215145537259954,
-      "grad_norm": 0.004328264854848385,
-      "learning_rate": 0.00019999422882595034,
-      "loss": 46.0,
-      "step": 21246
-    },
-    {
-      "epoch": 3.421675590804783,
-      "grad_norm": 0.005616902373731136,
-      "learning_rate": 0.00019999422828241537,
-      "loss": 46.0,
-      "step": 21247
-    },
-    {
-      "epoch": 3.4218366278835703,
-      "grad_norm": 0.0061208028346300125,
-      "learning_rate": 0.0001999942277388548,
-      "loss": 46.0,
-      "step": 21248
-    },
-    {
-      "epoch": 3.4219976649623574,
-      "grad_norm": 0.008044317364692688,
-      "learning_rate": 0.00019999422719526862,
-      "loss": 46.0,
-      "step": 21249
-    },
-    {
-      "epoch": 3.422158702041145,
-      "grad_norm": 0.006132552865892649,
-      "learning_rate": 0.00019999422665165686,
-      "loss": 46.0,
-      "step": 21250
-    },
-    {
-      "epoch": 3.4223197391199323,
-      "grad_norm": 0.009378168731927872,
-      "learning_rate": 0.0001999942261080195,
-      "loss": 46.0,
-      "step": 21251
-    },
-    {
-      "epoch": 3.4224807761987197,
-      "grad_norm": 0.003203070955350995,
-      "learning_rate": 0.00019999422556435655,
-      "loss": 46.0,
-      "step": 21252
-    },
-    {
-      "epoch": 3.422641813277507,
-      "grad_norm": 0.0010591407772153616,
-      "learning_rate": 0.00019999422502066803,
-      "loss": 46.0,
-      "step": 21253
-    },
-    {
-      "epoch": 3.4228028503562946,
-      "grad_norm": 0.003611624240875244,
-      "learning_rate": 0.0001999942244769539,
-      "loss": 46.0,
-      "step": 21254
-    },
-    {
-      "epoch": 3.422963887435082,
-      "grad_norm": 0.0016722215805202723,
-      "learning_rate": 0.00019999422393321417,
-      "loss": 46.0,
-      "step": 21255
-    },
-    {
-      "epoch": 3.4231249245138695,
-      "grad_norm": 0.0010098578641191125,
-      "learning_rate": 0.00019999422338944884,
-      "loss": 46.0,
-      "step": 21256
-    },
-    {
-      "epoch": 3.4232859615926565,
-      "grad_norm": 0.003147292183712125,
-      "learning_rate": 0.00019999422284565794,
-      "loss": 46.0,
-      "step": 21257
-    },
-    {
-      "epoch": 3.423446998671444,
-      "grad_norm": 0.0037701742257922888,
-      "learning_rate": 0.00019999422230184143,
-      "loss": 46.0,
-      "step": 21258
-    },
-    {
-      "epoch": 3.4236080357502314,
-      "grad_norm": 0.0050891381688416,
-      "learning_rate": 0.0001999942217579993,
-      "loss": 46.0,
-      "step": 21259
-    },
-    {
-      "epoch": 3.423769072829019,
-      "grad_norm": 0.004334925673902035,
-      "learning_rate": 0.00019999422121413162,
-      "loss": 46.0,
-      "step": 21260
-    },
-    {
-      "epoch": 3.4239301099078063,
-      "grad_norm": 0.0013067360268905759,
-      "learning_rate": 0.00019999422067023834,
-      "loss": 46.0,
-      "step": 21261
-    },
-    {
-      "epoch": 3.424091146986594,
-      "grad_norm": 0.010193086229264736,
-      "learning_rate": 0.00019999422012631946,
-      "loss": 46.0,
-      "step": 21262
-    },
-    {
-      "epoch": 3.4242521840653812,
-      "grad_norm": 0.009990700520575047,
-      "learning_rate": 0.00019999421958237498,
-      "loss": 46.0,
-      "step": 21263
-    },
-    {
-      "epoch": 3.4244132211441682,
-      "grad_norm": 0.0032884918618947268,
-      "learning_rate": 0.00019999421903840492,
-      "loss": 46.0,
-      "step": 21264
-    },
-    {
-      "epoch": 3.4245742582229557,
-      "grad_norm": 0.0024800316896289587,
-      "learning_rate": 0.00019999421849440925,
-      "loss": 46.0,
-      "step": 21265
-    },
-    {
-      "epoch": 3.424735295301743,
-      "grad_norm": 0.0038425729144364595,
-      "learning_rate": 0.000199994217950388,
-      "loss": 46.0,
-      "step": 21266
-    },
-    {
-      "epoch": 3.4248963323805306,
-      "grad_norm": 0.000754939392209053,
-      "learning_rate": 0.00019999421740634116,
-      "loss": 46.0,
-      "step": 21267
-    },
-    {
-      "epoch": 3.425057369459318,
-      "grad_norm": 0.0014265953795984387,
-      "learning_rate": 0.00019999421686226873,
-      "loss": 46.0,
-      "step": 21268
-    },
-    {
-      "epoch": 3.4252184065381055,
-      "grad_norm": 0.0008738969336263835,
-      "learning_rate": 0.00019999421631817067,
-      "loss": 46.0,
-      "step": 21269
-    },
-    {
-      "epoch": 3.425379443616893,
-      "grad_norm": 0.0017767586978152394,
-      "learning_rate": 0.00019999421577404704,
-      "loss": 46.0,
-      "step": 21270
-    },
-    {
-      "epoch": 3.4255404806956804,
-      "grad_norm": 0.0005299150943756104,
-      "learning_rate": 0.00019999421522989784,
-      "loss": 46.0,
-      "step": 21271
-    },
-    {
-      "epoch": 3.4257015177744674,
-      "grad_norm": 0.005189400166273117,
-      "learning_rate": 0.00019999421468572303,
-      "loss": 46.0,
-      "step": 21272
-    },
-    {
-      "epoch": 3.425862554853255,
-      "grad_norm": 0.008974839001893997,
-      "learning_rate": 0.00019999421414152263,
-      "loss": 46.0,
-      "step": 21273
-    },
-    {
-      "epoch": 3.4260235919320423,
-      "grad_norm": 0.00281995115801692,
-      "learning_rate": 0.0001999942135972966,
-      "loss": 46.0,
-      "step": 21274
-    },
-    {
-      "epoch": 3.4261846290108298,
-      "grad_norm": 0.0035095063503831625,
-      "learning_rate": 0.000199994213053045,
-      "loss": 46.0,
-      "step": 21275
-    },
-    {
-      "epoch": 3.426345666089617,
-      "grad_norm": 0.002488999627530575,
-      "learning_rate": 0.00019999421250876785,
-      "loss": 46.0,
-      "step": 21276
-    },
-    {
-      "epoch": 3.4265067031684047,
-      "grad_norm": 0.00954401958733797,
-      "learning_rate": 0.00019999421196446507,
-      "loss": 46.0,
-      "step": 21277
-    },
-    {
-      "epoch": 3.4266677402471917,
-      "grad_norm": 0.006119216792285442,
-      "learning_rate": 0.00019999421142013666,
-      "loss": 46.0,
-      "step": 21278
-    },
-    {
-      "epoch": 3.426828777325979,
-      "grad_norm": 0.001271423534490168,
-      "learning_rate": 0.0001999942108757827,
-      "loss": 46.0,
-      "step": 21279
-    },
-    {
-      "epoch": 3.4269898144047666,
-      "grad_norm": 0.0037192278541624546,
-      "learning_rate": 0.00019999421033140314,
-      "loss": 46.0,
-      "step": 21280
-    },
-    {
-      "epoch": 3.427150851483554,
-      "grad_norm": 0.008798657916486263,
-      "learning_rate": 0.000199994209786998,
-      "loss": 46.0,
-      "step": 21281
-    },
-    {
-      "epoch": 3.4273118885623415,
-      "grad_norm": 0.0012121403124183416,
-      "learning_rate": 0.00019999420924256725,
-      "loss": 46.0,
-      "step": 21282
-    },
-    {
-      "epoch": 3.427472925641129,
-      "grad_norm": 0.002615212230011821,
-      "learning_rate": 0.0001999942086981109,
-      "loss": 46.0,
-      "step": 21283
-    },
-    {
-      "epoch": 3.4276339627199164,
-      "grad_norm": 0.005895392037928104,
-      "learning_rate": 0.00019999420815362896,
-      "loss": 46.0,
-      "step": 21284
-    },
-    {
-      "epoch": 3.427794999798704,
-      "grad_norm": 0.0015287752030417323,
-      "learning_rate": 0.00019999420760912143,
-      "loss": 46.0,
-      "step": 21285
-    },
-    {
-      "epoch": 3.4279560368774913,
-      "grad_norm": 0.0014868566067889333,
-      "learning_rate": 0.00019999420706458832,
-      "loss": 46.0,
-      "step": 21286
-    },
-    {
-      "epoch": 3.4281170739562783,
-      "grad_norm": 0.0016907043755054474,
-      "learning_rate": 0.0001999942065200296,
-      "loss": 46.0,
-      "step": 21287
-    },
-    {
-      "epoch": 3.4282781110350657,
-      "grad_norm": 0.007533952593803406,
-      "learning_rate": 0.0001999942059754453,
-      "loss": 46.0,
-      "step": 21288
-    },
-    {
-      "epoch": 3.428439148113853,
-      "grad_norm": 0.0011690924875438213,
-      "learning_rate": 0.00019999420543083537,
-      "loss": 46.0,
-      "step": 21289
-    },
-    {
-      "epoch": 3.4286001851926406,
-      "grad_norm": 0.007620951626449823,
-      "learning_rate": 0.00019999420488619988,
-      "loss": 46.0,
-      "step": 21290
-    },
-    {
-      "epoch": 3.428761222271428,
-      "grad_norm": 0.005403009708970785,
-      "learning_rate": 0.0001999942043415388,
-      "loss": 46.0,
-      "step": 21291
-    },
-    {
-      "epoch": 3.4289222593502156,
-      "grad_norm": 0.005006282590329647,
-      "learning_rate": 0.0001999942037968521,
-      "loss": 46.0,
-      "step": 21292
-    },
-    {
-      "epoch": 3.4290832964290026,
-      "grad_norm": 0.00840732455253601,
-      "learning_rate": 0.00019999420325213983,
-      "loss": 46.0,
-      "step": 21293
-    },
-    {
-      "epoch": 3.42924433350779,
-      "grad_norm": 0.004210444167256355,
-      "learning_rate": 0.000199994202707402,
-      "loss": 46.0,
-      "step": 21294
-    },
-    {
-      "epoch": 3.4294053705865775,
-      "grad_norm": 0.00911111943423748,
-      "learning_rate": 0.00019999420216263851,
-      "loss": 46.0,
-      "step": 21295
-    },
-    {
-      "epoch": 3.429566407665365,
-      "grad_norm": 0.0019327015615999699,
-      "learning_rate": 0.00019999420161784945,
-      "loss": 46.0,
-      "step": 21296
-    },
-    {
-      "epoch": 3.4297274447441524,
-      "grad_norm": 0.002551912795752287,
-      "learning_rate": 0.00019999420107303482,
-      "loss": 46.0,
-      "step": 21297
-    },
-    {
-      "epoch": 3.42988848182294,
-      "grad_norm": 0.0005681714392267168,
-      "learning_rate": 0.00019999420052819455,
-      "loss": 46.0,
-      "step": 21298
-    },
-    {
-      "epoch": 3.4300495189017273,
-      "grad_norm": 0.004439387936145067,
-      "learning_rate": 0.00019999419998332872,
-      "loss": 46.0,
-      "step": 21299
-    },
-    {
-      "epoch": 3.4302105559805147,
-      "grad_norm": 0.002708984771743417,
-      "learning_rate": 0.00019999419943843728,
-      "loss": 46.0,
-      "step": 21300
-    },
-    {
-      "epoch": 3.4303715930593017,
-      "grad_norm": 0.0018625356024131179,
-      "learning_rate": 0.0001999941988935203,
-      "loss": 46.0,
-      "step": 21301
-    },
-    {
-      "epoch": 3.430532630138089,
-      "grad_norm": 0.007309530396014452,
-      "learning_rate": 0.00019999419834857766,
-      "loss": 46.0,
-      "step": 21302
-    },
-    {
-      "epoch": 3.4306936672168766,
-      "grad_norm": 0.001809528679586947,
-      "learning_rate": 0.00019999419780360945,
-      "loss": 46.0,
-      "step": 21303
-    },
-    {
-      "epoch": 3.430854704295664,
-      "grad_norm": 0.00235823355615139,
-      "learning_rate": 0.00019999419725861563,
-      "loss": 46.0,
-      "step": 21304
-    },
-    {
-      "epoch": 3.4310157413744515,
-      "grad_norm": 0.003477443940937519,
-      "learning_rate": 0.00019999419671359625,
-      "loss": 46.0,
-      "step": 21305
-    },
-    {
-      "epoch": 3.431176778453239,
-      "grad_norm": 0.005481565371155739,
-      "learning_rate": 0.00019999419616855126,
-      "loss": 46.0,
-      "step": 21306
-    },
-    {
-      "epoch": 3.4313378155320264,
-      "grad_norm": 0.003387550823390484,
-      "learning_rate": 0.00019999419562348068,
-      "loss": 46.0,
-      "step": 21307
-    },
-    {
-      "epoch": 3.4314988526108134,
-      "grad_norm": 0.001911480096168816,
-      "learning_rate": 0.0001999941950783845,
-      "loss": 46.0,
-      "step": 21308
-    },
-    {
-      "epoch": 3.431659889689601,
-      "grad_norm": 0.0013350839726626873,
-      "learning_rate": 0.00019999419453326273,
-      "loss": 46.0,
-      "step": 21309
-    },
-    {
-      "epoch": 3.4318209267683883,
-      "grad_norm": 0.004663375671952963,
-      "learning_rate": 0.00019999419398811536,
-      "loss": 46.0,
-      "step": 21310
-    },
-    {
-      "epoch": 3.431981963847176,
-      "grad_norm": 0.008582983165979385,
-      "learning_rate": 0.0001999941934429424,
-      "loss": 46.0,
-      "step": 21311
-    },
-    {
-      "epoch": 3.4321430009259632,
-      "grad_norm": 0.00385684659704566,
-      "learning_rate": 0.0001999941928977439,
-      "loss": 46.0,
-      "step": 21312
-    },
-    {
-      "epoch": 3.4323040380047507,
-      "grad_norm": 0.0014047992881387472,
-      "learning_rate": 0.00019999419235251973,
-      "loss": 46.0,
-      "step": 21313
-    },
-    {
-      "epoch": 3.432465075083538,
-      "grad_norm": 0.005968023557215929,
-      "learning_rate": 0.00019999419180727,
-      "loss": 46.0,
-      "step": 21314
-    },
-    {
-      "epoch": 3.4326261121623256,
-      "grad_norm": 0.004191408399492502,
-      "learning_rate": 0.00019999419126199466,
-      "loss": 46.0,
-      "step": 21315
-    },
-    {
-      "epoch": 3.4327871492411126,
-      "grad_norm": 0.003994282800704241,
-      "learning_rate": 0.00019999419071669374,
-      "loss": 46.0,
-      "step": 21316
-    },
-    {
-      "epoch": 3.4329481863199,
-      "grad_norm": 0.011227629147469997,
-      "learning_rate": 0.0001999941901713672,
-      "loss": 46.0,
-      "step": 21317
-    },
-    {
-      "epoch": 3.4331092233986875,
-      "grad_norm": 0.005359116476029158,
-      "learning_rate": 0.0001999941896260151,
-      "loss": 46.0,
-      "step": 21318
-    },
-    {
-      "epoch": 3.433270260477475,
-      "grad_norm": 0.0013284857850521803,
-      "learning_rate": 0.0001999941890806374,
-      "loss": 46.0,
-      "step": 21319
-    },
-    {
-      "epoch": 3.4334312975562624,
-      "grad_norm": 0.00040370863280259073,
-      "learning_rate": 0.0001999941885352341,
-      "loss": 46.0,
-      "step": 21320
-    },
-    {
-      "epoch": 3.43359233463505,
-      "grad_norm": 0.009429258294403553,
-      "learning_rate": 0.0001999941879898052,
-      "loss": 46.0,
-      "step": 21321
-    },
-    {
-      "epoch": 3.433753371713837,
-      "grad_norm": 0.0114449979737401,
-      "learning_rate": 0.00019999418744435073,
-      "loss": 46.0,
-      "step": 21322
-    },
-    {
-      "epoch": 3.4339144087926243,
-      "grad_norm": 0.0015785489231348038,
-      "learning_rate": 0.00019999418689887067,
-      "loss": 46.0,
-      "step": 21323
-    },
-    {
-      "epoch": 3.434075445871412,
-      "grad_norm": 0.0022865927312523127,
-      "learning_rate": 0.00019999418635336497,
-      "loss": 46.0,
-      "step": 21324
-    },
-    {
-      "epoch": 3.4342364829501992,
-      "grad_norm": 0.004068242851644754,
-      "learning_rate": 0.00019999418580783372,
-      "loss": 46.0,
-      "step": 21325
-    },
-    {
-      "epoch": 3.4343975200289867,
-      "grad_norm": 0.006889275275170803,
-      "learning_rate": 0.00019999418526227687,
-      "loss": 46.0,
-      "step": 21326
-    },
-    {
-      "epoch": 3.434558557107774,
-      "grad_norm": 0.005282598081976175,
-      "learning_rate": 0.0001999941847166944,
-      "loss": 46.0,
-      "step": 21327
-    },
-    {
-      "epoch": 3.4347195941865616,
-      "grad_norm": 0.011786254122853279,
-      "learning_rate": 0.00019999418417108636,
-      "loss": 46.0,
-      "step": 21328
-    },
-    {
-      "epoch": 3.434880631265349,
-      "grad_norm": 0.0010525865945965052,
-      "learning_rate": 0.00019999418362545273,
-      "loss": 46.0,
-      "step": 21329
-    },
-    {
-      "epoch": 3.435041668344136,
-      "grad_norm": 0.01248264592140913,
-      "learning_rate": 0.0001999941830797935,
-      "loss": 46.0,
-      "step": 21330
-    },
-    {
-      "epoch": 3.4352027054229235,
-      "grad_norm": 0.0005735012236982584,
-      "learning_rate": 0.00019999418253410867,
-      "loss": 46.0,
-      "step": 21331
-    },
-    {
-      "epoch": 3.435363742501711,
-      "grad_norm": 0.00212243665009737,
-      "learning_rate": 0.00019999418198839828,
-      "loss": 46.0,
-      "step": 21332
-    },
-    {
-      "epoch": 3.4355247795804984,
-      "grad_norm": 0.0075389849953353405,
-      "learning_rate": 0.00019999418144266224,
-      "loss": 46.0,
-      "step": 21333
-    },
-    {
-      "epoch": 3.435685816659286,
-      "grad_norm": 0.002500149654224515,
-      "learning_rate": 0.00019999418089690062,
-      "loss": 46.0,
-      "step": 21334
-    },
-    {
-      "epoch": 3.4358468537380733,
-      "grad_norm": 0.008484416641294956,
-      "learning_rate": 0.00019999418035111343,
-      "loss": 46.0,
-      "step": 21335
-    },
-    {
-      "epoch": 3.4360078908168608,
-      "grad_norm": 0.0014663385227322578,
-      "learning_rate": 0.00019999417980530066,
-      "loss": 46.0,
-      "step": 21336
-    },
-    {
-      "epoch": 3.4361689278956478,
-      "grad_norm": 0.0013378283474594355,
-      "learning_rate": 0.00019999417925946228,
-      "loss": 46.0,
-      "step": 21337
-    },
-    {
-      "epoch": 3.436329964974435,
-      "grad_norm": 0.008930770680308342,
-      "learning_rate": 0.00019999417871359828,
-      "loss": 46.0,
-      "step": 21338
-    },
-    {
-      "epoch": 3.4364910020532227,
-      "grad_norm": 0.0026706643402576447,
-      "learning_rate": 0.00019999417816770875,
-      "loss": 46.0,
-      "step": 21339
-    },
-    {
-      "epoch": 3.43665203913201,
-      "grad_norm": 0.0024902811273932457,
-      "learning_rate": 0.00019999417762179354,
-      "loss": 46.0,
-      "step": 21340
-    },
-    {
-      "epoch": 3.4368130762107976,
-      "grad_norm": 0.005879633594304323,
-      "learning_rate": 0.00019999417707585278,
-      "loss": 46.0,
-      "step": 21341
-    },
-    {
-      "epoch": 3.436974113289585,
-      "grad_norm": 0.003570025088265538,
-      "learning_rate": 0.00019999417652988646,
-      "loss": 46.0,
-      "step": 21342
-    },
-    {
-      "epoch": 3.4371351503683725,
-      "grad_norm": 0.00975044071674347,
-      "learning_rate": 0.0001999941759838945,
-      "loss": 46.0,
-      "step": 21343
-    },
-    {
-      "epoch": 3.43729618744716,
-      "grad_norm": 0.0015033313538879156,
-      "learning_rate": 0.00019999417543787695,
-      "loss": 46.0,
-      "step": 21344
-    },
-    {
-      "epoch": 3.437457224525947,
-      "grad_norm": 0.0007017470779828727,
-      "learning_rate": 0.00019999417489183384,
-      "loss": 46.0,
-      "step": 21345
-    },
-    {
-      "epoch": 3.4376182616047344,
-      "grad_norm": 0.0010633855126798153,
-      "learning_rate": 0.00019999417434576512,
-      "loss": 46.0,
-      "step": 21346
-    },
-    {
-      "epoch": 3.437779298683522,
-      "grad_norm": 0.001470943447202444,
-      "learning_rate": 0.0001999941737996708,
-      "loss": 46.0,
-      "step": 21347
-    },
-    {
-      "epoch": 3.4379403357623093,
-      "grad_norm": 0.010627185925841331,
-      "learning_rate": 0.0001999941732535509,
-      "loss": 46.0,
-      "step": 21348
-    },
-    {
-      "epoch": 3.4381013728410967,
-      "grad_norm": 0.0013186817523092031,
-      "learning_rate": 0.00019999417270740537,
-      "loss": 46.0,
-      "step": 21349
-    },
-    {
-      "epoch": 3.438262409919884,
-      "grad_norm": 0.0008128781919367611,
-      "learning_rate": 0.0001999941721612343,
-      "loss": 46.0,
-      "step": 21350
-    },
-    {
-      "epoch": 3.438423446998671,
-      "grad_norm": 0.0015869452618062496,
-      "learning_rate": 0.00019999417161503758,
-      "loss": 46.0,
-      "step": 21351
-    },
-    {
-      "epoch": 3.4385844840774586,
-      "grad_norm": 0.006966514978557825,
-      "learning_rate": 0.0001999941710688153,
-      "loss": 46.0,
-      "step": 21352
-    },
-    {
-      "epoch": 3.438745521156246,
-      "grad_norm": 0.0005937858950346708,
-      "learning_rate": 0.00019999417052256742,
-      "loss": 46.0,
-      "step": 21353
-    },
-    {
-      "epoch": 3.4389065582350336,
-      "grad_norm": 0.006463267374783754,
-      "learning_rate": 0.00019999416997629394,
-      "loss": 46.0,
-      "step": 21354
-    },
-    {
-      "epoch": 3.439067595313821,
-      "grad_norm": 0.002489438047632575,
-      "learning_rate": 0.0001999941694299949,
-      "loss": 46.0,
-      "step": 21355
-    },
-    {
-      "epoch": 3.4392286323926085,
-      "grad_norm": 0.007998412474989891,
-      "learning_rate": 0.00019999416888367023,
-      "loss": 46.0,
-      "step": 21356
-    },
-    {
-      "epoch": 3.439389669471396,
-      "grad_norm": 0.006187098566442728,
-      "learning_rate": 0.00019999416833731996,
-      "loss": 46.0,
-      "step": 21357
-    },
-    {
-      "epoch": 3.4395507065501834,
-      "grad_norm": 0.012148981913924217,
-      "learning_rate": 0.00019999416779094414,
-      "loss": 46.0,
-      "step": 21358
-    },
-    {
-      "epoch": 3.439711743628971,
-      "grad_norm": 0.02225356362760067,
-      "learning_rate": 0.0001999941672445427,
-      "loss": 46.0,
-      "step": 21359
-    },
-    {
-      "epoch": 3.439872780707758,
-      "grad_norm": 0.0018617149908095598,
-      "learning_rate": 0.00019999416669811565,
-      "loss": 46.0,
-      "step": 21360
-    },
-    {
-      "epoch": 3.4400338177865453,
-      "grad_norm": 0.004765899386256933,
-      "learning_rate": 0.000199994166151663,
-      "loss": 46.0,
-      "step": 21361
-    },
-    {
-      "epoch": 3.4401948548653327,
-      "grad_norm": 0.0018667754484340549,
-      "learning_rate": 0.00019999416560518484,
-      "loss": 46.0,
-      "step": 21362
-    },
-    {
-      "epoch": 3.44035589194412,
-      "grad_norm": 0.004824182949960232,
-      "learning_rate": 0.000199994165058681,
-      "loss": 46.0,
-      "step": 21363
-    },
-    {
-      "epoch": 3.4405169290229076,
-      "grad_norm": 0.0029847051482647657,
-      "learning_rate": 0.0001999941645121516,
-      "loss": 46.0,
-      "step": 21364
-    },
-    {
-      "epoch": 3.440677966101695,
-      "grad_norm": 0.004837886895984411,
-      "learning_rate": 0.0001999941639655966,
-      "loss": 46.0,
-      "step": 21365
-    },
-    {
-      "epoch": 3.440839003180482,
-      "grad_norm": 0.005056528374552727,
-      "learning_rate": 0.00019999416341901603,
-      "loss": 46.0,
-      "step": 21366
-    },
-    {
-      "epoch": 3.4410000402592695,
-      "grad_norm": 0.007738937623798847,
-      "learning_rate": 0.0001999941628724098,
-      "loss": 46.0,
-      "step": 21367
-    },
-    {
-      "epoch": 3.441161077338057,
-      "grad_norm": 0.0034253697376698256,
-      "learning_rate": 0.00019999416232577803,
-      "loss": 46.0,
-      "step": 21368
-    },
-    {
-      "epoch": 3.4413221144168444,
-      "grad_norm": 0.0038828575052320957,
-      "learning_rate": 0.00019999416177912067,
-      "loss": 46.0,
-      "step": 21369
-    },
-    {
-      "epoch": 3.441483151495632,
-      "grad_norm": 0.003389406716451049,
-      "learning_rate": 0.00019999416123243772,
-      "loss": 46.0,
-      "step": 21370
-    },
-    {
-      "epoch": 3.4416441885744193,
-      "grad_norm": 0.0033943778835237026,
-      "learning_rate": 0.00019999416068572915,
-      "loss": 46.0,
-      "step": 21371
-    },
-    {
-      "epoch": 3.441805225653207,
-      "grad_norm": 0.002610705094411969,
-      "learning_rate": 0.000199994160138995,
-      "loss": 46.0,
-      "step": 21372
-    },
-    {
-      "epoch": 3.4419662627319942,
-      "grad_norm": 0.0014638545690104365,
-      "learning_rate": 0.00019999415959223526,
-      "loss": 46.0,
-      "step": 21373
-    },
-    {
-      "epoch": 3.4421272998107812,
-      "grad_norm": 0.0014145808527246118,
-      "learning_rate": 0.0001999941590454499,
-      "loss": 46.0,
-      "step": 21374
-    },
-    {
-      "epoch": 3.4422883368895687,
-      "grad_norm": 0.010198313742876053,
-      "learning_rate": 0.000199994158498639,
-      "loss": 46.0,
-      "step": 21375
-    },
-    {
-      "epoch": 3.442449373968356,
-      "grad_norm": 0.011615744791924953,
-      "learning_rate": 0.00019999415795180246,
-      "loss": 46.0,
-      "step": 21376
-    },
-    {
-      "epoch": 3.4426104110471436,
-      "grad_norm": 0.008249531500041485,
-      "learning_rate": 0.00019999415740494034,
-      "loss": 46.0,
-      "step": 21377
-    },
-    {
-      "epoch": 3.442771448125931,
-      "grad_norm": 0.0031641407404094934,
-      "learning_rate": 0.0001999941568580526,
-      "loss": 46.0,
-      "step": 21378
-    },
-    {
-      "epoch": 3.4429324852047185,
-      "grad_norm": 0.008500061929225922,
-      "learning_rate": 0.00019999415631113932,
-      "loss": 46.0,
-      "step": 21379
-    },
-    {
-      "epoch": 3.443093522283506,
-      "grad_norm": 0.006334215868264437,
-      "learning_rate": 0.0001999941557642004,
-      "loss": 46.0,
-      "step": 21380
-    },
-    {
-      "epoch": 3.443254559362293,
-      "grad_norm": 0.0016233554342761636,
-      "learning_rate": 0.00019999415521723592,
-      "loss": 46.0,
-      "step": 21381
-    },
-    {
-      "epoch": 3.4434155964410804,
-      "grad_norm": 0.002812237711623311,
-      "learning_rate": 0.00019999415467024582,
-      "loss": 46.0,
-      "step": 21382
-    },
-    {
-      "epoch": 3.443576633519868,
-      "grad_norm": 0.0022897920571267605,
-      "learning_rate": 0.00019999415412323015,
-      "loss": 46.0,
-      "step": 21383
-    },
-    {
-      "epoch": 3.4437376705986553,
-      "grad_norm": 0.003015248803421855,
-      "learning_rate": 0.0001999941535761889,
-      "loss": 46.0,
-      "step": 21384
-    },
-    {
-      "epoch": 3.4438987076774428,
-      "grad_norm": 0.0016156338388100266,
-      "learning_rate": 0.000199994153029122,
-      "loss": 46.0,
-      "step": 21385
-    },
-    {
-      "epoch": 3.4440597447562302,
-      "grad_norm": 0.003549477318301797,
-      "learning_rate": 0.00019999415248202955,
-      "loss": 46.0,
-      "step": 21386
-    },
-    {
-      "epoch": 3.4442207818350177,
-      "grad_norm": 0.001294828369282186,
-      "learning_rate": 0.0001999941519349115,
-      "loss": 46.0,
-      "step": 21387
-    },
-    {
-      "epoch": 3.444381818913805,
-      "grad_norm": 0.0013177813962101936,
-      "learning_rate": 0.00019999415138776785,
-      "loss": 46.0,
-      "step": 21388
-    },
-    {
-      "epoch": 3.444542855992592,
-      "grad_norm": 0.001766753033734858,
-      "learning_rate": 0.0001999941508405986,
-      "loss": 46.0,
-      "step": 21389
-    },
-    {
-      "epoch": 3.4447038930713796,
-      "grad_norm": 0.004019097425043583,
-      "learning_rate": 0.00019999415029340378,
-      "loss": 46.0,
-      "step": 21390
-    },
-    {
-      "epoch": 3.444864930150167,
-      "grad_norm": 0.004501206334680319,
-      "learning_rate": 0.00019999414974618336,
-      "loss": 46.0,
-      "step": 21391
-    },
-    {
-      "epoch": 3.4450259672289545,
-      "grad_norm": 0.0019878707826137543,
-      "learning_rate": 0.00019999414919893733,
-      "loss": 46.0,
-      "step": 21392
-    },
-    {
-      "epoch": 3.445187004307742,
-      "grad_norm": 0.011708315461874008,
-      "learning_rate": 0.00019999414865166573,
-      "loss": 46.0,
-      "step": 21393
-    },
-    {
-      "epoch": 3.4453480413865294,
-      "grad_norm": 0.0007050794083625078,
-      "learning_rate": 0.00019999414810436853,
-      "loss": 46.0,
-      "step": 21394
-    },
-    {
-      "epoch": 3.4455090784653164,
-      "grad_norm": 0.0035162714775651693,
-      "learning_rate": 0.0001999941475570457,
-      "loss": 46.0,
-      "step": 21395
-    },
-    {
-      "epoch": 3.445670115544104,
-      "grad_norm": 0.002092215232551098,
-      "learning_rate": 0.00019999414700969733,
-      "loss": 46.0,
-      "step": 21396
-    },
-    {
-      "epoch": 3.4458311526228913,
-      "grad_norm": 0.004627359099686146,
-      "learning_rate": 0.00019999414646232336,
-      "loss": 46.0,
-      "step": 21397
-    },
-    {
-      "epoch": 3.4459921897016788,
-      "grad_norm": 0.0012548464583232999,
-      "learning_rate": 0.00019999414591492375,
-      "loss": 46.0,
-      "step": 21398
-    },
-    {
-      "epoch": 3.446153226780466,
-      "grad_norm": 0.0010451386915519834,
-      "learning_rate": 0.00019999414536749858,
-      "loss": 46.0,
-      "step": 21399
-    },
-    {
-      "epoch": 3.4463142638592537,
-      "grad_norm": 0.001638429588638246,
-      "learning_rate": 0.00019999414482004782,
-      "loss": 46.0,
-      "step": 21400
-    },
-    {
-      "epoch": 3.446475300938041,
-      "grad_norm": 0.0039406828582286835,
-      "learning_rate": 0.00019999414427257145,
-      "loss": 46.0,
-      "step": 21401
-    },
-    {
-      "epoch": 3.4466363380168286,
-      "grad_norm": 0.0031661326065659523,
-      "learning_rate": 0.0001999941437250695,
-      "loss": 46.0,
-      "step": 21402
-    },
-    {
-      "epoch": 3.446797375095616,
-      "grad_norm": 0.008524918928742409,
-      "learning_rate": 0.00019999414317754195,
-      "loss": 46.0,
-      "step": 21403
-    },
-    {
-      "epoch": 3.446958412174403,
-      "grad_norm": 0.00435027526691556,
-      "learning_rate": 0.00019999414262998882,
-      "loss": 46.0,
-      "step": 21404
-    },
-    {
-      "epoch": 3.4471194492531905,
-      "grad_norm": 0.001814637565985322,
-      "learning_rate": 0.00019999414208241007,
-      "loss": 46.0,
-      "step": 21405
-    },
-    {
-      "epoch": 3.447280486331978,
-      "grad_norm": 0.0016347779892385006,
-      "learning_rate": 0.00019999414153480576,
-      "loss": 46.0,
-      "step": 21406
-    },
-    {
-      "epoch": 3.4474415234107654,
-      "grad_norm": 0.0037243920378386974,
-      "learning_rate": 0.00019999414098717581,
-      "loss": 46.0,
-      "step": 21407
-    },
-    {
-      "epoch": 3.447602560489553,
-      "grad_norm": 0.003347005695104599,
-      "learning_rate": 0.0001999941404395203,
-      "loss": 46.0,
-      "step": 21408
-    },
-    {
-      "epoch": 3.4477635975683403,
-      "grad_norm": 0.0016809668159112334,
-      "learning_rate": 0.00019999413989183918,
-      "loss": 46.0,
-      "step": 21409
-    },
-    {
-      "epoch": 3.4479246346471273,
-      "grad_norm": 0.006221352610737085,
-      "learning_rate": 0.0001999941393441325,
-      "loss": 46.0,
-      "step": 21410
-    },
-    {
-      "epoch": 3.4480856717259147,
-      "grad_norm": 0.001083912793546915,
-      "learning_rate": 0.0001999941387964002,
-      "loss": 46.0,
-      "step": 21411
-    },
-    {
-      "epoch": 3.448246708804702,
-      "grad_norm": 0.012469722889363766,
-      "learning_rate": 0.00019999413824864232,
-      "loss": 46.0,
-      "step": 21412
-    },
-    {
-      "epoch": 3.4484077458834896,
-      "grad_norm": 0.002171517815440893,
-      "learning_rate": 0.00019999413770085885,
-      "loss": 46.0,
-      "step": 21413
-    },
-    {
-      "epoch": 3.448568782962277,
-      "grad_norm": 0.005600192584097385,
-      "learning_rate": 0.00019999413715304976,
-      "loss": 46.0,
-      "step": 21414
-    },
-    {
-      "epoch": 3.4487298200410645,
-      "grad_norm": 0.0004889547126367688,
-      "learning_rate": 0.00019999413660521512,
-      "loss": 46.0,
-      "step": 21415
-    },
-    {
-      "epoch": 3.448890857119852,
-      "grad_norm": 0.0048583378084003925,
-      "learning_rate": 0.00019999413605735483,
-      "loss": 46.0,
-      "step": 21416
-    },
-    {
-      "epoch": 3.4490518941986394,
-      "grad_norm": 0.0009475324186496437,
-      "learning_rate": 0.00019999413550946899,
-      "loss": 46.0,
-      "step": 21417
-    },
-    {
-      "epoch": 3.4492129312774265,
-      "grad_norm": 0.0048226118087768555,
-      "learning_rate": 0.00019999413496155755,
-      "loss": 46.0,
-      "step": 21418
-    },
-    {
-      "epoch": 3.449373968356214,
-      "grad_norm": 0.0025488710962235928,
-      "learning_rate": 0.0001999941344136205,
-      "loss": 46.0,
-      "step": 21419
-    },
-    {
-      "epoch": 3.4495350054350014,
-      "grad_norm": 0.004803691990673542,
-      "learning_rate": 0.00019999413386565787,
-      "loss": 46.0,
-      "step": 21420
-    },
-    {
-      "epoch": 3.449696042513789,
-      "grad_norm": 0.000948400585912168,
-      "learning_rate": 0.00019999413331766962,
-      "loss": 46.0,
-      "step": 21421
-    },
-    {
-      "epoch": 3.4498570795925763,
-      "grad_norm": 0.0008942551212385297,
-      "learning_rate": 0.0001999941327696558,
-      "loss": 46.0,
-      "step": 21422
-    },
-    {
-      "epoch": 3.4500181166713637,
-      "grad_norm": 0.0011097368551418185,
-      "learning_rate": 0.0001999941322216164,
-      "loss": 46.0,
-      "step": 21423
-    },
-    {
-      "epoch": 3.450179153750151,
-      "grad_norm": 0.0043246932327747345,
-      "learning_rate": 0.0001999941316735514,
-      "loss": 46.0,
-      "step": 21424
-    },
-    {
-      "epoch": 3.450340190828938,
-      "grad_norm": 0.004683345090597868,
-      "learning_rate": 0.0001999941311254608,
-      "loss": 46.0,
-      "step": 21425
-    },
-    {
-      "epoch": 3.4505012279077256,
-      "grad_norm": 0.0017776319291442633,
-      "learning_rate": 0.0001999941305773446,
-      "loss": 46.0,
-      "step": 21426
-    },
-    {
-      "epoch": 3.450662264986513,
-      "grad_norm": 0.008514122106134892,
-      "learning_rate": 0.0001999941300292028,
-      "loss": 46.0,
-      "step": 21427
-    },
-    {
-      "epoch": 3.4508233020653005,
-      "grad_norm": 0.0015808396274223924,
-      "learning_rate": 0.00019999412948103543,
-      "loss": 46.0,
-      "step": 21428
-    },
-    {
-      "epoch": 3.450984339144088,
-      "grad_norm": 0.0011398581555113196,
-      "learning_rate": 0.00019999412893284245,
-      "loss": 46.0,
-      "step": 21429
-    },
-    {
-      "epoch": 3.4511453762228754,
-      "grad_norm": 0.013294342905282974,
-      "learning_rate": 0.0001999941283846239,
-      "loss": 46.0,
-      "step": 21430
-    },
-    {
-      "epoch": 3.451306413301663,
-      "grad_norm": 0.005192117765545845,
-      "learning_rate": 0.00019999412783637972,
-      "loss": 46.0,
-      "step": 21431
-    },
-    {
-      "epoch": 3.4514674503804503,
-      "grad_norm": 0.0011622088495641947,
-      "learning_rate": 0.00019999412728810998,
-      "loss": 46.0,
-      "step": 21432
-    },
-    {
-      "epoch": 3.4516284874592373,
-      "grad_norm": 0.0026824085507541895,
-      "learning_rate": 0.0001999941267398146,
-      "loss": 46.0,
-      "step": 21433
-    },
-    {
-      "epoch": 3.451789524538025,
-      "grad_norm": 0.004787949845194817,
-      "learning_rate": 0.00019999412619149367,
-      "loss": 46.0,
-      "step": 21434
-    },
-    {
-      "epoch": 3.4519505616168122,
-      "grad_norm": 0.0019320844439789653,
-      "learning_rate": 0.00019999412564314714,
-      "loss": 46.0,
-      "step": 21435
-    },
-    {
-      "epoch": 3.4521115986955997,
-      "grad_norm": 0.0012487316271290183,
-      "learning_rate": 0.00019999412509477503,
-      "loss": 46.0,
-      "step": 21436
-    },
-    {
-      "epoch": 3.452272635774387,
-      "grad_norm": 0.002958685392513871,
-      "learning_rate": 0.0001999941245463773,
-      "loss": 46.0,
-      "step": 21437
-    },
-    {
-      "epoch": 3.4524336728531746,
-      "grad_norm": 0.001750450930558145,
-      "learning_rate": 0.000199994123997954,
-      "loss": 46.0,
-      "step": 21438
-    },
-    {
-      "epoch": 3.4525947099319616,
-      "grad_norm": 0.0037361423019319773,
-      "learning_rate": 0.0001999941234495051,
-      "loss": 46.0,
-      "step": 21439
-    },
-    {
-      "epoch": 3.452755747010749,
-      "grad_norm": 0.0035370357800275087,
-      "learning_rate": 0.00019999412290103058,
-      "loss": 46.0,
-      "step": 21440
-    },
-    {
-      "epoch": 3.4529167840895365,
-      "grad_norm": 0.008352371864020824,
-      "learning_rate": 0.00019999412235253047,
-      "loss": 46.0,
-      "step": 21441
-    },
-    {
-      "epoch": 3.453077821168324,
-      "grad_norm": 0.005913614295423031,
-      "learning_rate": 0.00019999412180400479,
-      "loss": 46.0,
-      "step": 21442
-    },
-    {
-      "epoch": 3.4532388582471114,
-      "grad_norm": 0.01322778407484293,
-      "learning_rate": 0.0001999941212554535,
-      "loss": 46.0,
-      "step": 21443
-    },
-    {
-      "epoch": 3.453399895325899,
-      "grad_norm": 0.003756898920983076,
-      "learning_rate": 0.00019999412070687665,
-      "loss": 46.0,
-      "step": 21444
-    },
-    {
-      "epoch": 3.4535609324046863,
-      "grad_norm": 0.009156583808362484,
-      "learning_rate": 0.00019999412015827417,
-      "loss": 46.0,
-      "step": 21445
-    },
-    {
-      "epoch": 3.4537219694834738,
-      "grad_norm": 0.005296287126839161,
-      "learning_rate": 0.0001999941196096461,
-      "loss": 46.0,
-      "step": 21446
-    },
-    {
-      "epoch": 3.4538830065622608,
-      "grad_norm": 0.0010049953125417233,
-      "learning_rate": 0.00019999411906099245,
-      "loss": 46.0,
-      "step": 21447
-    },
-    {
-      "epoch": 3.4540440436410482,
-      "grad_norm": 0.001782875508069992,
-      "learning_rate": 0.0001999941185123132,
-      "loss": 46.0,
-      "step": 21448
-    },
-    {
-      "epoch": 3.4542050807198357,
-      "grad_norm": 0.005515060853213072,
-      "learning_rate": 0.00019999411796360836,
-      "loss": 46.0,
-      "step": 21449
-    },
-    {
-      "epoch": 3.454366117798623,
-      "grad_norm": 0.0038124844431877136,
-      "learning_rate": 0.00019999411741487792,
-      "loss": 46.0,
-      "step": 21450
-    },
-    {
-      "epoch": 3.4545271548774106,
-      "grad_norm": 0.011111240833997726,
-      "learning_rate": 0.0001999941168661219,
-      "loss": 46.0,
-      "step": 21451
-    },
-    {
-      "epoch": 3.454688191956198,
-      "grad_norm": 0.013927358202636242,
-      "learning_rate": 0.00019999411631734025,
-      "loss": 46.0,
-      "step": 21452
-    },
-    {
-      "epoch": 3.4548492290349855,
-      "grad_norm": 0.003935371525585651,
-      "learning_rate": 0.00019999411576853305,
-      "loss": 46.0,
-      "step": 21453
-    },
-    {
-      "epoch": 3.4550102661137725,
-      "grad_norm": 0.01726355217397213,
-      "learning_rate": 0.00019999411521970023,
-      "loss": 46.0,
-      "step": 21454
-    },
-    {
-      "epoch": 3.45517130319256,
-      "grad_norm": 0.005137636326253414,
-      "learning_rate": 0.00019999411467084186,
-      "loss": 46.0,
-      "step": 21455
-    },
-    {
-      "epoch": 3.4553323402713474,
-      "grad_norm": 0.0037466150242835283,
-      "learning_rate": 0.00019999411412195787,
-      "loss": 46.0,
-      "step": 21456
-    },
-    {
-      "epoch": 3.455493377350135,
-      "grad_norm": 0.0023211603984236717,
-      "learning_rate": 0.00019999411357304826,
-      "loss": 46.0,
-      "step": 21457
-    },
-    {
-      "epoch": 3.4556544144289223,
-      "grad_norm": 0.011249782517552376,
-      "learning_rate": 0.0001999941130241131,
-      "loss": 46.0,
-      "step": 21458
-    },
-    {
-      "epoch": 3.4558154515077097,
-      "grad_norm": 0.007194583769887686,
-      "learning_rate": 0.00019999411247515232,
-      "loss": 46.0,
-      "step": 21459
-    },
-    {
-      "epoch": 3.455976488586497,
-      "grad_norm": 0.0006801488343626261,
-      "learning_rate": 0.00019999411192616595,
-      "loss": 46.0,
-      "step": 21460
-    },
-    {
-      "epoch": 3.4561375256652846,
-      "grad_norm": 0.004215317778289318,
-      "learning_rate": 0.00019999411137715397,
-      "loss": 46.0,
-      "step": 21461
-    },
-    {
-      "epoch": 3.4562985627440717,
-      "grad_norm": 0.00459174532443285,
-      "learning_rate": 0.00019999411082811643,
-      "loss": 46.0,
-      "step": 21462
-    },
-    {
-      "epoch": 3.456459599822859,
-      "grad_norm": 0.0021795067004859447,
-      "learning_rate": 0.00019999411027905327,
-      "loss": 46.0,
-      "step": 21463
-    },
-    {
-      "epoch": 3.4566206369016466,
-      "grad_norm": 0.004822002723813057,
-      "learning_rate": 0.00019999410972996453,
-      "loss": 46.0,
-      "step": 21464
-    },
-    {
-      "epoch": 3.456781673980434,
-      "grad_norm": 0.006667776964604855,
-      "learning_rate": 0.0001999941091808502,
-      "loss": 46.0,
-      "step": 21465
-    },
-    {
-      "epoch": 3.4569427110592215,
-      "grad_norm": 0.012152226641774178,
-      "learning_rate": 0.00019999410863171026,
-      "loss": 46.0,
-      "step": 21466
-    },
-    {
-      "epoch": 3.457103748138009,
-      "grad_norm": 0.0074287299066782,
-      "learning_rate": 0.00019999410808254476,
-      "loss": 46.0,
-      "step": 21467
-    },
-    {
-      "epoch": 3.457264785216796,
-      "grad_norm": 0.0028173604514449835,
-      "learning_rate": 0.0001999941075333536,
-      "loss": 46.0,
-      "step": 21468
-    },
-    {
-      "epoch": 3.4574258222955834,
-      "grad_norm": 0.0025672900956124067,
-      "learning_rate": 0.0001999941069841369,
-      "loss": 46.0,
-      "step": 21469
-    },
-    {
-      "epoch": 3.457586859374371,
-      "grad_norm": 0.002528794575482607,
-      "learning_rate": 0.0001999941064348946,
-      "loss": 46.0,
-      "step": 21470
-    },
-    {
-      "epoch": 3.4577478964531583,
-      "grad_norm": 0.0007392846746370196,
-      "learning_rate": 0.00019999410588562673,
-      "loss": 46.0,
-      "step": 21471
-    },
-    {
-      "epoch": 3.4579089335319457,
-      "grad_norm": 0.0010529011487960815,
-      "learning_rate": 0.00019999410533633324,
-      "loss": 46.0,
-      "step": 21472
-    },
-    {
-      "epoch": 3.458069970610733,
-      "grad_norm": 0.00391079718247056,
-      "learning_rate": 0.00019999410478701416,
-      "loss": 46.0,
-      "step": 21473
-    },
-    {
-      "epoch": 3.4582310076895206,
-      "grad_norm": 0.001630091923289001,
-      "learning_rate": 0.00019999410423766946,
-      "loss": 46.0,
-      "step": 21474
-    },
-    {
-      "epoch": 3.458392044768308,
-      "grad_norm": 0.0015246031107380986,
-      "learning_rate": 0.0001999941036882992,
-      "loss": 46.0,
-      "step": 21475
-    },
-    {
-      "epoch": 3.4585530818470955,
-      "grad_norm": 0.0031663798727095127,
-      "learning_rate": 0.00019999410313890334,
-      "loss": 46.0,
-      "step": 21476
-    },
-    {
-      "epoch": 3.4587141189258825,
-      "grad_norm": 0.012410926632583141,
-      "learning_rate": 0.00019999410258948188,
-      "loss": 46.0,
-      "step": 21477
-    },
-    {
-      "epoch": 3.45887515600467,
-      "grad_norm": 0.007828181609511375,
-      "learning_rate": 0.00019999410204003484,
-      "loss": 46.0,
-      "step": 21478
-    },
-    {
-      "epoch": 3.4590361930834574,
-      "grad_norm": 0.0018423181027173996,
-      "learning_rate": 0.00019999410149056218,
-      "loss": 46.0,
-      "step": 21479
-    },
-    {
-      "epoch": 3.459197230162245,
-      "grad_norm": 0.0009254949982278049,
-      "learning_rate": 0.00019999410094106396,
-      "loss": 46.0,
-      "step": 21480
-    },
-    {
-      "epoch": 3.4593582672410323,
-      "grad_norm": 0.0020012175664305687,
-      "learning_rate": 0.00019999410039154013,
-      "loss": 46.0,
-      "step": 21481
-    },
-    {
-      "epoch": 3.45951930431982,
-      "grad_norm": 0.000831014767754823,
-      "learning_rate": 0.00019999409984199068,
-      "loss": 46.0,
-      "step": 21482
-    },
-    {
-      "epoch": 3.459680341398607,
-      "grad_norm": 0.004109567031264305,
-      "learning_rate": 0.00019999409929241568,
-      "loss": 46.0,
-      "step": 21483
-    },
-    {
-      "epoch": 3.4598413784773943,
-      "grad_norm": 0.0009089052909985185,
-      "learning_rate": 0.00019999409874281508,
-      "loss": 46.0,
-      "step": 21484
-    },
-    {
-      "epoch": 3.4600024155561817,
-      "grad_norm": 0.011840258724987507,
-      "learning_rate": 0.00019999409819318887,
-      "loss": 46.0,
-      "step": 21485
-    },
-    {
-      "epoch": 3.460163452634969,
-      "grad_norm": 0.0074869440868496895,
-      "learning_rate": 0.00019999409764353708,
-      "loss": 46.0,
-      "step": 21486
-    },
-    {
-      "epoch": 3.4603244897137566,
-      "grad_norm": 0.00601461436599493,
-      "learning_rate": 0.00019999409709385967,
-      "loss": 46.0,
-      "step": 21487
-    },
-    {
-      "epoch": 3.460485526792544,
-      "grad_norm": 0.001918905065394938,
-      "learning_rate": 0.00019999409654415667,
-      "loss": 46.0,
-      "step": 21488
-    },
-    {
-      "epoch": 3.4606465638713315,
-      "grad_norm": 0.010926822200417519,
-      "learning_rate": 0.00019999409599442811,
-      "loss": 46.0,
-      "step": 21489
-    },
-    {
-      "epoch": 3.460807600950119,
-      "grad_norm": 0.0026856965851038694,
-      "learning_rate": 0.00019999409544467394,
-      "loss": 46.0,
-      "step": 21490
-    },
-    {
-      "epoch": 3.460968638028906,
-      "grad_norm": 0.004998111166059971,
-      "learning_rate": 0.00019999409489489419,
-      "loss": 46.0,
-      "step": 21491
-    },
-    {
-      "epoch": 3.4611296751076934,
-      "grad_norm": 0.005897205322980881,
-      "learning_rate": 0.0001999940943450888,
-      "loss": 46.0,
-      "step": 21492
-    },
-    {
-      "epoch": 3.461290712186481,
-      "grad_norm": 0.008031473495066166,
-      "learning_rate": 0.00019999409379525788,
-      "loss": 46.0,
-      "step": 21493
-    },
-    {
-      "epoch": 3.4614517492652683,
-      "grad_norm": 0.0031186360865831375,
-      "learning_rate": 0.00019999409324540133,
-      "loss": 46.0,
-      "step": 21494
-    },
-    {
-      "epoch": 3.461612786344056,
-      "grad_norm": 0.004655584692955017,
-      "learning_rate": 0.0001999940926955192,
-      "loss": 46.0,
-      "step": 21495
-    },
-    {
-      "epoch": 3.4617738234228432,
-      "grad_norm": 0.0033942044246941805,
-      "learning_rate": 0.00019999409214561148,
-      "loss": 46.0,
-      "step": 21496
-    },
-    {
-      "epoch": 3.4619348605016307,
-      "grad_norm": 0.00680797453969717,
-      "learning_rate": 0.00019999409159567817,
-      "loss": 46.0,
-      "step": 21497
-    },
-    {
-      "epoch": 3.4620958975804177,
-      "grad_norm": 0.00296114943921566,
-      "learning_rate": 0.00019999409104571925,
-      "loss": 46.0,
-      "step": 21498
-    },
-    {
-      "epoch": 3.462256934659205,
-      "grad_norm": 0.0053552561439573765,
-      "learning_rate": 0.0001999940904957347,
-      "loss": 46.0,
-      "step": 21499
-    },
-    {
-      "epoch": 3.4624179717379926,
-      "grad_norm": 0.0022706228774040937,
-      "learning_rate": 0.0001999940899457246,
-      "loss": 46.0,
-      "step": 21500
-    },
-    {
-      "epoch": 3.46257900881678,
-      "grad_norm": 0.004283626563847065,
-      "learning_rate": 0.0001999940893956889,
-      "loss": 46.0,
-      "step": 21501
-    },
-    {
-      "epoch": 3.4627400458955675,
-      "grad_norm": 0.003928447607904673,
-      "learning_rate": 0.00019999408884562763,
-      "loss": 46.0,
-      "step": 21502
-    },
-    {
-      "epoch": 3.462901082974355,
-      "grad_norm": 0.002243251074105501,
-      "learning_rate": 0.00019999408829554077,
-      "loss": 46.0,
-      "step": 21503
-    },
-    {
-      "epoch": 3.4630621200531424,
-      "grad_norm": 0.0020860950462520123,
-      "learning_rate": 0.00019999408774542827,
-      "loss": 46.0,
-      "step": 21504
-    },
-    {
-      "epoch": 3.46322315713193,
-      "grad_norm": 0.0032327515073120594,
-      "learning_rate": 0.0001999940871952902,
-      "loss": 46.0,
-      "step": 21505
-    },
-    {
-      "epoch": 3.463384194210717,
-      "grad_norm": 0.0012006328906863928,
-      "learning_rate": 0.00019999408664512656,
-      "loss": 46.0,
-      "step": 21506
-    },
-    {
-      "epoch": 3.4635452312895043,
-      "grad_norm": 0.005046685691922903,
-      "learning_rate": 0.00019999408609493727,
-      "loss": 46.0,
-      "step": 21507
-    },
-    {
-      "epoch": 3.4637062683682918,
-      "grad_norm": 0.000724576530046761,
-      "learning_rate": 0.00019999408554472243,
-      "loss": 46.0,
-      "step": 21508
-    },
-    {
-      "epoch": 3.463867305447079,
-      "grad_norm": 0.0018612323328852654,
-      "learning_rate": 0.000199994084994482,
-      "loss": 46.0,
-      "step": 21509
-    },
-    {
-      "epoch": 3.4640283425258667,
-      "grad_norm": 0.0018015929963439703,
-      "learning_rate": 0.00019999408444421594,
-      "loss": 46.0,
-      "step": 21510
-    },
-    {
-      "epoch": 3.464189379604654,
-      "grad_norm": 0.003694532671943307,
-      "learning_rate": 0.0001999940838939243,
-      "loss": 46.0,
-      "step": 21511
-    },
-    {
-      "epoch": 3.464350416683441,
-      "grad_norm": 0.0059257992543280125,
-      "learning_rate": 0.0001999940833436071,
-      "loss": 46.0,
-      "step": 21512
-    },
-    {
-      "epoch": 3.4645114537622286,
-      "grad_norm": 0.006660667713731527,
-      "learning_rate": 0.0001999940827932643,
-      "loss": 46.0,
-      "step": 21513
-    },
-    {
-      "epoch": 3.464672490841016,
-      "grad_norm": 0.010414987802505493,
-      "learning_rate": 0.0001999940822428959,
-      "loss": 46.0,
-      "step": 21514
-    },
-    {
-      "epoch": 3.4648335279198035,
-      "grad_norm": 0.002524890238419175,
-      "learning_rate": 0.00019999408169250185,
-      "loss": 46.0,
-      "step": 21515
-    },
-    {
-      "epoch": 3.464994564998591,
-      "grad_norm": 0.0014486188301816583,
-      "learning_rate": 0.00019999408114208228,
-      "loss": 46.0,
-      "step": 21516
-    },
-    {
-      "epoch": 3.4651556020773784,
-      "grad_norm": 0.002709099790081382,
-      "learning_rate": 0.00019999408059163707,
-      "loss": 46.0,
-      "step": 21517
-    },
-    {
-      "epoch": 3.465316639156166,
-      "grad_norm": 0.0012349470052868128,
-      "learning_rate": 0.0001999940800411663,
-      "loss": 46.0,
-      "step": 21518
-    },
-    {
-      "epoch": 3.4654776762349533,
-      "grad_norm": 0.003846916602924466,
-      "learning_rate": 0.00019999407949066993,
-      "loss": 46.0,
-      "step": 21519
-    },
-    {
-      "epoch": 3.4656387133137407,
-      "grad_norm": 0.0047247339971363544,
-      "learning_rate": 0.00019999407894014795,
-      "loss": 46.0,
-      "step": 21520
-    },
-    {
-      "epoch": 3.4657997503925277,
-      "grad_norm": 0.00406165886670351,
-      "learning_rate": 0.0001999940783896004,
-      "loss": 46.0,
-      "step": 21521
-    },
-    {
-      "epoch": 3.465960787471315,
-      "grad_norm": 0.0019498143810778856,
-      "learning_rate": 0.00019999407783902724,
-      "loss": 46.0,
-      "step": 21522
-    },
-    {
-      "epoch": 3.4661218245501026,
-      "grad_norm": 0.0012497896095737815,
-      "learning_rate": 0.00019999407728842847,
-      "loss": 46.0,
-      "step": 21523
-    },
-    {
-      "epoch": 3.46628286162889,
-      "grad_norm": 0.003848744323477149,
-      "learning_rate": 0.00019999407673780412,
-      "loss": 46.0,
-      "step": 21524
-    },
-    {
-      "epoch": 3.4664438987076776,
-      "grad_norm": 0.0031660227105021477,
-      "learning_rate": 0.0001999940761871542,
-      "loss": 46.0,
-      "step": 21525
-    },
-    {
-      "epoch": 3.466604935786465,
-      "grad_norm": 0.001507553388364613,
-      "learning_rate": 0.00019999407563647868,
-      "loss": 46.0,
-      "step": 21526
-    },
-    {
-      "epoch": 3.466765972865252,
-      "grad_norm": 0.002985989674925804,
-      "learning_rate": 0.00019999407508577754,
-      "loss": 46.0,
-      "step": 21527
-    },
-    {
-      "epoch": 3.4669270099440395,
-      "grad_norm": 0.003641576273366809,
-      "learning_rate": 0.00019999407453505082,
-      "loss": 46.0,
-      "step": 21528
-    },
-    {
-      "epoch": 3.467088047022827,
-      "grad_norm": 0.00188413355499506,
-      "learning_rate": 0.00019999407398429853,
-      "loss": 46.0,
-      "step": 21529
-    },
-    {
-      "epoch": 3.4672490841016144,
-      "grad_norm": 0.0012499837903305888,
-      "learning_rate": 0.0001999940734335206,
-      "loss": 46.0,
-      "step": 21530
-    },
-    {
-      "epoch": 3.467410121180402,
-      "grad_norm": 0.00933252926915884,
-      "learning_rate": 0.0001999940728827171,
-      "loss": 46.0,
-      "step": 21531
-    },
-    {
-      "epoch": 3.4675711582591893,
-      "grad_norm": 0.006399719975888729,
-      "learning_rate": 0.00019999407233188803,
-      "loss": 46.0,
-      "step": 21532
-    },
-    {
-      "epoch": 3.4677321953379767,
-      "grad_norm": 0.0053024268709123135,
-      "learning_rate": 0.00019999407178103334,
-      "loss": 46.0,
-      "step": 21533
-    },
-    {
-      "epoch": 3.467893232416764,
-      "grad_norm": 0.007915375754237175,
-      "learning_rate": 0.0001999940712301531,
-      "loss": 46.0,
-      "step": 21534
-    },
-    {
-      "epoch": 3.468054269495551,
-      "grad_norm": 0.0030044482555240393,
-      "learning_rate": 0.00019999407067924717,
-      "loss": 46.0,
-      "step": 21535
-    },
-    {
-      "epoch": 3.4682153065743386,
-      "grad_norm": 0.01124865934252739,
-      "learning_rate": 0.00019999407012831572,
-      "loss": 46.0,
-      "step": 21536
-    },
-    {
-      "epoch": 3.468376343653126,
-      "grad_norm": 0.00439730379730463,
-      "learning_rate": 0.00019999406957735866,
-      "loss": 46.0,
-      "step": 21537
-    },
-    {
-      "epoch": 3.4685373807319135,
-      "grad_norm": 0.0021364956628531218,
-      "learning_rate": 0.000199994069026376,
-      "loss": 46.0,
-      "step": 21538
-    },
-    {
-      "epoch": 3.468698417810701,
-      "grad_norm": 0.009805376641452312,
-      "learning_rate": 0.0001999940684753678,
-      "loss": 46.0,
-      "step": 21539
-    },
-    {
-      "epoch": 3.4688594548894884,
-      "grad_norm": 0.002298088278621435,
-      "learning_rate": 0.00019999406792433396,
-      "loss": 46.0,
-      "step": 21540
-    },
-    {
-      "epoch": 3.469020491968276,
-      "grad_norm": 0.0025894942227751017,
-      "learning_rate": 0.00019999406737327452,
-      "loss": 46.0,
-      "step": 21541
-    },
-    {
-      "epoch": 3.469181529047063,
-      "grad_norm": 0.006048859562724829,
-      "learning_rate": 0.0001999940668221895,
-      "loss": 46.0,
-      "step": 21542
-    },
-    {
-      "epoch": 3.4693425661258503,
-      "grad_norm": 0.004307756666094065,
-      "learning_rate": 0.00019999406627107887,
-      "loss": 46.0,
-      "step": 21543
-    },
-    {
-      "epoch": 3.469503603204638,
-      "grad_norm": 0.001994300400838256,
-      "learning_rate": 0.00019999406571994267,
-      "loss": 46.0,
-      "step": 21544
-    },
-    {
-      "epoch": 3.4696646402834252,
-      "grad_norm": 0.003949034959077835,
-      "learning_rate": 0.00019999406516878085,
-      "loss": 46.0,
-      "step": 21545
-    },
-    {
-      "epoch": 3.4698256773622127,
-      "grad_norm": 0.0018495999975129962,
-      "learning_rate": 0.00019999406461759347,
-      "loss": 46.0,
-      "step": 21546
-    },
-    {
-      "epoch": 3.469986714441,
-      "grad_norm": 0.00537153659388423,
-      "learning_rate": 0.00019999406406638045,
-      "loss": 46.0,
-      "step": 21547
-    },
-    {
-      "epoch": 3.4701477515197876,
-      "grad_norm": 0.014439959079027176,
-      "learning_rate": 0.00019999406351514187,
-      "loss": 46.0,
-      "step": 21548
-    },
-    {
-      "epoch": 3.470308788598575,
-      "grad_norm": 0.0014400511281564832,
-      "learning_rate": 0.0001999940629638777,
-      "loss": 46.0,
-      "step": 21549
-    },
-    {
-      "epoch": 3.470469825677362,
-      "grad_norm": 0.003036829177290201,
-      "learning_rate": 0.00019999406241258793,
-      "loss": 46.0,
-      "step": 21550
-    },
-    {
-      "epoch": 3.4706308627561495,
-      "grad_norm": 0.01739700324833393,
-      "learning_rate": 0.00019999406186127256,
-      "loss": 46.0,
-      "step": 21551
-    },
-    {
-      "epoch": 3.470791899834937,
-      "grad_norm": 0.0022692920174449682,
-      "learning_rate": 0.0001999940613099316,
-      "loss": 46.0,
-      "step": 21552
-    },
-    {
-      "epoch": 3.4709529369137244,
-      "grad_norm": 0.007472389843314886,
-      "learning_rate": 0.00019999406075856506,
-      "loss": 46.0,
-      "step": 21553
-    },
-    {
-      "epoch": 3.471113973992512,
-      "grad_norm": 0.010819156654179096,
-      "learning_rate": 0.0001999940602071729,
-      "loss": 46.0,
-      "step": 21554
-    },
-    {
-      "epoch": 3.4712750110712993,
-      "grad_norm": 0.010931729339063168,
-      "learning_rate": 0.00019999405965575519,
-      "loss": 46.0,
-      "step": 21555
-    },
-    {
-      "epoch": 3.4714360481500863,
-      "grad_norm": 0.0035244664177298546,
-      "learning_rate": 0.00019999405910431185,
-      "loss": 46.0,
-      "step": 21556
-    },
-    {
-      "epoch": 3.471597085228874,
-      "grad_norm": 0.016931017860770226,
-      "learning_rate": 0.0001999940585528429,
-      "loss": 46.0,
-      "step": 21557
-    },
-    {
-      "epoch": 3.4717581223076612,
-      "grad_norm": 0.001266232691705227,
-      "learning_rate": 0.0001999940580013484,
-      "loss": 46.0,
-      "step": 21558
-    },
-    {
-      "epoch": 3.4719191593864487,
-      "grad_norm": 0.006256175227463245,
-      "learning_rate": 0.00019999405744982828,
-      "loss": 46.0,
-      "step": 21559
-    },
-    {
-      "epoch": 3.472080196465236,
-      "grad_norm": 0.0016495035961270332,
-      "learning_rate": 0.00019999405689828258,
-      "loss": 46.0,
-      "step": 21560
-    },
-    {
-      "epoch": 3.4722412335440236,
-      "grad_norm": 0.0034828991629183292,
-      "learning_rate": 0.00019999405634671128,
-      "loss": 46.0,
-      "step": 21561
-    },
-    {
-      "epoch": 3.472402270622811,
-      "grad_norm": 0.0018091370584443212,
-      "learning_rate": 0.0001999940557951144,
-      "loss": 46.0,
-      "step": 21562
-    },
-    {
-      "epoch": 3.4725633077015985,
-      "grad_norm": 0.0017083127750083804,
-      "learning_rate": 0.00019999405524349193,
-      "loss": 46.0,
-      "step": 21563
-    },
-    {
-      "epoch": 3.4727243447803855,
-      "grad_norm": 0.0019094264134764671,
-      "learning_rate": 0.00019999405469184382,
-      "loss": 46.0,
-      "step": 21564
-    },
-    {
-      "epoch": 3.472885381859173,
-      "grad_norm": 0.011336550116539001,
-      "learning_rate": 0.00019999405414017015,
-      "loss": 46.0,
-      "step": 21565
-    },
-    {
-      "epoch": 3.4730464189379604,
-      "grad_norm": 0.004751401487737894,
-      "learning_rate": 0.0001999940535884709,
-      "loss": 46.0,
-      "step": 21566
-    },
-    {
-      "epoch": 3.473207456016748,
-      "grad_norm": 0.001113905687816441,
-      "learning_rate": 0.00019999405303674603,
-      "loss": 46.0,
-      "step": 21567
-    },
-    {
-      "epoch": 3.4733684930955353,
-      "grad_norm": 0.0026605126913636923,
-      "learning_rate": 0.00019999405248499562,
-      "loss": 46.0,
-      "step": 21568
-    },
-    {
-      "epoch": 3.4735295301743228,
-      "grad_norm": 0.013115840964019299,
-      "learning_rate": 0.00019999405193321955,
-      "loss": 46.0,
-      "step": 21569
-    },
-    {
-      "epoch": 3.47369056725311,
-      "grad_norm": 0.0037995046004652977,
-      "learning_rate": 0.00019999405138141792,
-      "loss": 46.0,
-      "step": 21570
-    },
-    {
-      "epoch": 3.473851604331897,
-      "grad_norm": 0.004185269121080637,
-      "learning_rate": 0.0001999940508295907,
-      "loss": 46.0,
-      "step": 21571
-    },
-    {
-      "epoch": 3.4740126414106847,
-      "grad_norm": 0.00208605220541358,
-      "learning_rate": 0.00019999405027773786,
-      "loss": 46.0,
-      "step": 21572
-    },
-    {
-      "epoch": 3.474173678489472,
-      "grad_norm": 0.01016261987388134,
-      "learning_rate": 0.00019999404972585944,
-      "loss": 46.0,
-      "step": 21573
-    },
-    {
-      "epoch": 3.4743347155682596,
-      "grad_norm": 0.003020795062184334,
-      "learning_rate": 0.0001999940491739554,
-      "loss": 46.0,
-      "step": 21574
-    },
-    {
-      "epoch": 3.474495752647047,
-      "grad_norm": 0.012632275000214577,
-      "learning_rate": 0.0001999940486220258,
-      "loss": 46.0,
-      "step": 21575
-    },
-    {
-      "epoch": 3.4746567897258345,
-      "grad_norm": 0.007691246923059225,
-      "learning_rate": 0.00019999404807007063,
-      "loss": 46.0,
-      "step": 21576
-    },
-    {
-      "epoch": 3.474817826804622,
-      "grad_norm": 0.007989048957824707,
-      "learning_rate": 0.00019999404751808983,
-      "loss": 46.0,
-      "step": 21577
-    },
-    {
-      "epoch": 3.4749788638834094,
-      "grad_norm": 0.003225174266844988,
-      "learning_rate": 0.00019999404696608344,
-      "loss": 46.0,
-      "step": 21578
-    },
-    {
-      "epoch": 3.4751399009621964,
-      "grad_norm": 0.0073609910905361176,
-      "learning_rate": 0.00019999404641405147,
-      "loss": 46.0,
-      "step": 21579
-    },
-    {
-      "epoch": 3.475300938040984,
-      "grad_norm": 0.00303995655849576,
-      "learning_rate": 0.00019999404586199391,
-      "loss": 46.0,
-      "step": 21580
-    },
-    {
-      "epoch": 3.4754619751197713,
-      "grad_norm": 0.0037534211296588182,
-      "learning_rate": 0.0001999940453099107,
-      "loss": 46.0,
-      "step": 21581
-    },
-    {
-      "epoch": 3.4756230121985587,
-      "grad_norm": 0.004336799029260874,
-      "learning_rate": 0.00019999404475780198,
-      "loss": 46.0,
-      "step": 21582
-    },
-    {
-      "epoch": 3.475784049277346,
-      "grad_norm": 0.002320753177627921,
-      "learning_rate": 0.00019999404420566763,
-      "loss": 46.0,
-      "step": 21583
-    },
-    {
-      "epoch": 3.4759450863561336,
-      "grad_norm": 0.004559011198580265,
-      "learning_rate": 0.00019999404365350767,
-      "loss": 46.0,
-      "step": 21584
-    },
-    {
-      "epoch": 3.4761061234349206,
-      "grad_norm": 0.016857832670211792,
-      "learning_rate": 0.00019999404310132212,
-      "loss": 46.0,
-      "step": 21585
-    },
-    {
-      "epoch": 3.476267160513708,
-      "grad_norm": 0.001042317831888795,
-      "learning_rate": 0.00019999404254911101,
-      "loss": 46.0,
-      "step": 21586
-    },
-    {
-      "epoch": 3.4764281975924956,
-      "grad_norm": 0.015265410766005516,
-      "learning_rate": 0.0001999940419968743,
-      "loss": 46.0,
-      "step": 21587
-    },
-    {
-      "epoch": 3.476589234671283,
-      "grad_norm": 0.0011762077920138836,
-      "learning_rate": 0.00019999404144461198,
-      "loss": 46.0,
-      "step": 21588
-    },
-    {
-      "epoch": 3.4767502717500705,
-      "grad_norm": 0.0070272888988256454,
-      "learning_rate": 0.00019999404089232406,
-      "loss": 46.0,
-      "step": 21589
-    },
-    {
-      "epoch": 3.476911308828858,
-      "grad_norm": 0.0014839920913800597,
-      "learning_rate": 0.00019999404034001054,
-      "loss": 46.0,
-      "step": 21590
-    },
-    {
-      "epoch": 3.4770723459076454,
-      "grad_norm": 0.0043205576948821545,
-      "learning_rate": 0.00019999403978767144,
-      "loss": 46.0,
-      "step": 21591
-    },
-    {
-      "epoch": 3.477233382986433,
-      "grad_norm": 0.0053464495576918125,
-      "learning_rate": 0.00019999403923530676,
-      "loss": 46.0,
-      "step": 21592
-    },
-    {
-      "epoch": 3.4773944200652203,
-      "grad_norm": 0.00060122279683128,
-      "learning_rate": 0.00019999403868291649,
-      "loss": 46.0,
-      "step": 21593
-    },
-    {
-      "epoch": 3.4775554571440073,
-      "grad_norm": 0.007110511884093285,
-      "learning_rate": 0.0001999940381305006,
-      "loss": 46.0,
-      "step": 21594
-    },
-    {
-      "epoch": 3.4777164942227947,
-      "grad_norm": 0.002072003437206149,
-      "learning_rate": 0.00019999403757805912,
-      "loss": 46.0,
-      "step": 21595
-    },
-    {
-      "epoch": 3.477877531301582,
-      "grad_norm": 0.004132483154535294,
-      "learning_rate": 0.00019999403702559206,
-      "loss": 46.0,
-      "step": 21596
-    },
-    {
-      "epoch": 3.4780385683803696,
-      "grad_norm": 0.0025334444362670183,
-      "learning_rate": 0.0001999940364730994,
-      "loss": 46.0,
-      "step": 21597
-    },
-    {
-      "epoch": 3.478199605459157,
-      "grad_norm": 0.0011895459610968828,
-      "learning_rate": 0.00019999403592058115,
-      "loss": 46.0,
-      "step": 21598
-    },
-    {
-      "epoch": 3.4783606425379445,
-      "grad_norm": 0.005745948292315006,
-      "learning_rate": 0.00019999403536803733,
-      "loss": 46.0,
-      "step": 21599
-    },
-    {
-      "epoch": 3.4785216796167315,
-      "grad_norm": 0.0020802633371204138,
-      "learning_rate": 0.00019999403481546786,
-      "loss": 46.0,
-      "step": 21600
-    },
-    {
-      "epoch": 3.478682716695519,
-      "grad_norm": 0.003612247761338949,
-      "learning_rate": 0.0001999940342628728,
-      "loss": 46.0,
-      "step": 21601
-    },
-    {
-      "epoch": 3.4788437537743064,
-      "grad_norm": 0.0027700415812432766,
-      "learning_rate": 0.0001999940337102522,
-      "loss": 46.0,
-      "step": 21602
-    },
-    {
-      "epoch": 3.479004790853094,
-      "grad_norm": 0.0009137579472735524,
-      "learning_rate": 0.000199994033157606,
-      "loss": 46.0,
-      "step": 21603
-    },
-    {
-      "epoch": 3.4791658279318813,
-      "grad_norm": 0.002333424286916852,
-      "learning_rate": 0.00019999403260493418,
-      "loss": 46.0,
-      "step": 21604
-    },
-    {
-      "epoch": 3.479326865010669,
-      "grad_norm": 0.0031451343093067408,
-      "learning_rate": 0.00019999403205223678,
-      "loss": 46.0,
-      "step": 21605
-    },
-    {
-      "epoch": 3.4794879020894562,
-      "grad_norm": 0.0020351496059447527,
-      "learning_rate": 0.00019999403149951377,
-      "loss": 46.0,
-      "step": 21606
-    },
-    {
-      "epoch": 3.4796489391682437,
-      "grad_norm": 0.011380152776837349,
-      "learning_rate": 0.0001999940309467652,
-      "loss": 46.0,
-      "step": 21607
-    },
-    {
-      "epoch": 3.4798099762470307,
-      "grad_norm": 0.01850689947605133,
-      "learning_rate": 0.000199994030393991,
-      "loss": 46.0,
-      "step": 21608
-    },
-    {
-      "epoch": 3.479971013325818,
-      "grad_norm": 0.003453419543802738,
-      "learning_rate": 0.00019999402984119123,
-      "loss": 46.0,
-      "step": 21609
-    },
-    {
-      "epoch": 3.4801320504046056,
-      "grad_norm": 0.000979285454377532,
-      "learning_rate": 0.00019999402928836584,
-      "loss": 46.0,
-      "step": 21610
-    },
-    {
-      "epoch": 3.480293087483393,
-      "grad_norm": 0.0060765682719647884,
-      "learning_rate": 0.00019999402873551486,
-      "loss": 46.0,
-      "step": 21611
-    },
-    {
-      "epoch": 3.4804541245621805,
-      "grad_norm": 0.005047384649515152,
-      "learning_rate": 0.00019999402818263832,
-      "loss": 46.0,
-      "step": 21612
-    },
-    {
-      "epoch": 3.480615161640968,
-      "grad_norm": 0.012718395330011845,
-      "learning_rate": 0.00019999402762973617,
-      "loss": 46.0,
-      "step": 21613
-    },
-    {
-      "epoch": 3.4807761987197554,
-      "grad_norm": 0.006007462739944458,
-      "learning_rate": 0.00019999402707680843,
-      "loss": 46.0,
-      "step": 21614
-    },
-    {
-      "epoch": 3.4809372357985424,
-      "grad_norm": 0.004431604873389006,
-      "learning_rate": 0.00019999402652385508,
-      "loss": 46.0,
-      "step": 21615
-    },
-    {
-      "epoch": 3.48109827287733,
-      "grad_norm": 0.004615690093487501,
-      "learning_rate": 0.00019999402597087616,
-      "loss": 46.0,
-      "step": 21616
-    },
-    {
-      "epoch": 3.4812593099561173,
-      "grad_norm": 0.00464807590469718,
-      "learning_rate": 0.00019999402541787163,
-      "loss": 46.0,
-      "step": 21617
-    },
-    {
-      "epoch": 3.4814203470349048,
-      "grad_norm": 0.00957720447331667,
-      "learning_rate": 0.00019999402486484155,
-      "loss": 46.0,
-      "step": 21618
-    },
-    {
-      "epoch": 3.4815813841136922,
-      "grad_norm": 0.010536537505686283,
-      "learning_rate": 0.0001999940243117858,
-      "loss": 46.0,
-      "step": 21619
-    },
-    {
-      "epoch": 3.4817424211924797,
-      "grad_norm": 0.007567035965621471,
-      "learning_rate": 0.00019999402375870447,
-      "loss": 46.0,
-      "step": 21620
-    },
-    {
-      "epoch": 3.481903458271267,
-      "grad_norm": 0.003694914048537612,
-      "learning_rate": 0.0001999940232055976,
-      "loss": 46.0,
-      "step": 21621
-    },
-    {
-      "epoch": 3.4820644953500546,
-      "grad_norm": 0.013131159357726574,
-      "learning_rate": 0.0001999940226524651,
-      "loss": 46.0,
-      "step": 21622
-    },
-    {
-      "epoch": 3.4822255324288416,
-      "grad_norm": 0.0056958869099617004,
-      "learning_rate": 0.00019999402209930703,
-      "loss": 46.0,
-      "step": 21623
-    },
-    {
-      "epoch": 3.482386569507629,
-      "grad_norm": 0.0023376031313091516,
-      "learning_rate": 0.00019999402154612333,
-      "loss": 46.0,
-      "step": 21624
-    },
-    {
-      "epoch": 3.4825476065864165,
-      "grad_norm": 0.0026437039487063885,
-      "learning_rate": 0.00019999402099291408,
-      "loss": 46.0,
-      "step": 21625
-    },
-    {
-      "epoch": 3.482708643665204,
-      "grad_norm": 0.0037356445100158453,
-      "learning_rate": 0.00019999402043967922,
-      "loss": 46.0,
-      "step": 21626
-    },
-    {
-      "epoch": 3.4828696807439914,
-      "grad_norm": 0.0034827461931854486,
-      "learning_rate": 0.00019999401988641876,
-      "loss": 46.0,
-      "step": 21627
-    },
-    {
-      "epoch": 3.483030717822779,
-      "grad_norm": 0.0021162594202905893,
-      "learning_rate": 0.00019999401933313272,
-      "loss": 46.0,
-      "step": 21628
-    },
-    {
-      "epoch": 3.483191754901566,
-      "grad_norm": 0.0013706154422834516,
-      "learning_rate": 0.00019999401877982104,
-      "loss": 46.0,
-      "step": 21629
-    },
-    {
-      "epoch": 3.4833527919803533,
-      "grad_norm": 0.0022472376003861427,
-      "learning_rate": 0.0001999940182264838,
-      "loss": 46.0,
-      "step": 21630
-    },
-    {
-      "epoch": 3.4835138290591408,
-      "grad_norm": 0.005469535477459431,
-      "learning_rate": 0.000199994017673121,
-      "loss": 46.0,
-      "step": 21631
-    },
-    {
-      "epoch": 3.483674866137928,
-      "grad_norm": 0.00773253757506609,
-      "learning_rate": 0.00019999401711973255,
-      "loss": 46.0,
-      "step": 21632
-    },
-    {
-      "epoch": 3.4838359032167157,
-      "grad_norm": 0.003926425240933895,
-      "learning_rate": 0.00019999401656631854,
-      "loss": 46.0,
-      "step": 21633
-    },
-    {
-      "epoch": 3.483996940295503,
-      "grad_norm": 0.001304979552514851,
-      "learning_rate": 0.00019999401601287893,
-      "loss": 46.0,
-      "step": 21634
-    },
-    {
-      "epoch": 3.4841579773742906,
-      "grad_norm": 0.004565785638988018,
-      "learning_rate": 0.00019999401545941372,
-      "loss": 46.0,
-      "step": 21635
-    },
-    {
-      "epoch": 3.484319014453078,
-      "grad_norm": 0.003365050535649061,
-      "learning_rate": 0.0001999940149059229,
-      "loss": 46.0,
-      "step": 21636
-    },
-    {
-      "epoch": 3.484480051531865,
-      "grad_norm": 0.004340991843491793,
-      "learning_rate": 0.00019999401435240652,
-      "loss": 46.0,
-      "step": 21637
-    },
-    {
-      "epoch": 3.4846410886106525,
-      "grad_norm": 0.002922011073678732,
-      "learning_rate": 0.00019999401379886455,
-      "loss": 46.0,
-      "step": 21638
-    },
-    {
-      "epoch": 3.48480212568944,
-      "grad_norm": 0.002363980980589986,
-      "learning_rate": 0.00019999401324529697,
-      "loss": 46.0,
-      "step": 21639
-    },
-    {
-      "epoch": 3.4849631627682274,
-      "grad_norm": 0.0019061536295339465,
-      "learning_rate": 0.0001999940126917038,
-      "loss": 46.0,
-      "step": 21640
-    },
-    {
-      "epoch": 3.485124199847015,
-      "grad_norm": 0.007439350243657827,
-      "learning_rate": 0.00019999401213808505,
-      "loss": 46.0,
-      "step": 21641
-    },
-    {
-      "epoch": 3.4852852369258023,
-      "grad_norm": 0.0047789039090275764,
-      "learning_rate": 0.00019999401158444068,
-      "loss": 46.0,
-      "step": 21642
-    },
-    {
-      "epoch": 3.4854462740045897,
-      "grad_norm": 0.0035275130067020655,
-      "learning_rate": 0.00019999401103077072,
-      "loss": 46.0,
-      "step": 21643
-    },
-    {
-      "epoch": 3.4856073110833767,
-      "grad_norm": 0.0026339481119066477,
-      "learning_rate": 0.00019999401047707518,
-      "loss": 46.0,
-      "step": 21644
-    },
-    {
-      "epoch": 3.485768348162164,
-      "grad_norm": 0.00210524071007967,
-      "learning_rate": 0.00019999400992335405,
-      "loss": 46.0,
-      "step": 21645
-    },
-    {
-      "epoch": 3.4859293852409516,
-      "grad_norm": 0.015096982009708881,
-      "learning_rate": 0.0001999940093696073,
-      "loss": 46.0,
-      "step": 21646
-    },
-    {
-      "epoch": 3.486090422319739,
-      "grad_norm": 0.0012922576861456037,
-      "learning_rate": 0.000199994008815835,
-      "loss": 46.0,
-      "step": 21647
-    },
-    {
-      "epoch": 3.4862514593985265,
-      "grad_norm": 0.009506190195679665,
-      "learning_rate": 0.00019999400826203707,
-      "loss": 46.0,
-      "step": 21648
-    },
-    {
-      "epoch": 3.486412496477314,
-      "grad_norm": 0.004630511626601219,
-      "learning_rate": 0.00019999400770821354,
-      "loss": 46.0,
-      "step": 21649
-    },
-    {
-      "epoch": 3.4865735335561014,
-      "grad_norm": 0.007189747411757708,
-      "learning_rate": 0.00019999400715436445,
-      "loss": 46.0,
-      "step": 21650
-    },
-    {
-      "epoch": 3.486734570634889,
-      "grad_norm": 0.004328939598053694,
-      "learning_rate": 0.00019999400660048974,
-      "loss": 46.0,
-      "step": 21651
-    },
-    {
-      "epoch": 3.486895607713676,
-      "grad_norm": 0.0012114219134673476,
-      "learning_rate": 0.00019999400604658947,
-      "loss": 46.0,
-      "step": 21652
-    },
-    {
-      "epoch": 3.4870566447924634,
-      "grad_norm": 0.009022900834679604,
-      "learning_rate": 0.00019999400549266358,
-      "loss": 46.0,
-      "step": 21653
-    },
-    {
-      "epoch": 3.487217681871251,
-      "grad_norm": 0.002417529234662652,
-      "learning_rate": 0.0001999940049387121,
-      "loss": 46.0,
-      "step": 21654
-    },
-    {
-      "epoch": 3.4873787189500383,
-      "grad_norm": 0.0017306219087913632,
-      "learning_rate": 0.00019999400438473503,
-      "loss": 46.0,
-      "step": 21655
-    },
-    {
-      "epoch": 3.4875397560288257,
-      "grad_norm": 0.005544361192733049,
-      "learning_rate": 0.00019999400383073236,
-      "loss": 46.0,
-      "step": 21656
-    },
-    {
-      "epoch": 3.487700793107613,
-      "grad_norm": 0.005513602867722511,
-      "learning_rate": 0.0001999940032767041,
-      "loss": 46.0,
-      "step": 21657
-    },
-    {
-      "epoch": 3.4878618301864,
-      "grad_norm": 0.008863819763064384,
-      "learning_rate": 0.00019999400272265025,
-      "loss": 46.0,
-      "step": 21658
-    },
-    {
-      "epoch": 3.4880228672651876,
-      "grad_norm": 0.0014070312026888132,
-      "learning_rate": 0.0001999940021685708,
-      "loss": 46.0,
-      "step": 21659
-    },
-    {
-      "epoch": 3.488183904343975,
-      "grad_norm": 0.0028806105256080627,
-      "learning_rate": 0.00019999400161446577,
-      "loss": 46.0,
-      "step": 21660
-    },
-    {
-      "epoch": 3.4883449414227625,
-      "grad_norm": 0.0046287281438708305,
-      "learning_rate": 0.0001999940010603351,
-      "loss": 46.0,
-      "step": 21661
-    },
-    {
-      "epoch": 3.48850597850155,
-      "grad_norm": 0.0024114635307341814,
-      "learning_rate": 0.0001999940005061789,
-      "loss": 46.0,
-      "step": 21662
-    },
-    {
-      "epoch": 3.4886670155803374,
-      "grad_norm": 0.002221218543127179,
-      "learning_rate": 0.00019999399995199708,
-      "loss": 46.0,
-      "step": 21663
-    },
-    {
-      "epoch": 3.488828052659125,
-      "grad_norm": 0.004017350729554892,
-      "learning_rate": 0.00019999399939778968,
-      "loss": 46.0,
-      "step": 21664
-    },
-    {
-      "epoch": 3.4889890897379123,
-      "grad_norm": 0.0010599031811580062,
-      "learning_rate": 0.00019999399884355667,
-      "loss": 46.0,
-      "step": 21665
-    },
-    {
-      "epoch": 3.4891501268167,
-      "grad_norm": 0.005866139195859432,
-      "learning_rate": 0.00019999399828929808,
-      "loss": 46.0,
-      "step": 21666
-    },
-    {
-      "epoch": 3.489311163895487,
-      "grad_norm": 0.006685503292828798,
-      "learning_rate": 0.0001999939977350139,
-      "loss": 46.0,
-      "step": 21667
-    },
-    {
-      "epoch": 3.4894722009742742,
-      "grad_norm": 0.0009195986785925925,
-      "learning_rate": 0.00019999399718070412,
-      "loss": 46.0,
-      "step": 21668
-    },
-    {
-      "epoch": 3.4896332380530617,
-      "grad_norm": 0.002200279850512743,
-      "learning_rate": 0.00019999399662636873,
-      "loss": 46.0,
-      "step": 21669
-    },
-    {
-      "epoch": 3.489794275131849,
-      "grad_norm": 0.005131826736032963,
-      "learning_rate": 0.00019999399607200773,
-      "loss": 46.0,
-      "step": 21670
-    },
-    {
-      "epoch": 3.4899553122106366,
-      "grad_norm": 0.0012679877690970898,
-      "learning_rate": 0.0001999939955176212,
-      "loss": 46.0,
-      "step": 21671
-    },
-    {
-      "epoch": 3.490116349289424,
-      "grad_norm": 0.00447316188365221,
-      "learning_rate": 0.00019999399496320903,
-      "loss": 46.0,
-      "step": 21672
-    },
-    {
-      "epoch": 3.490277386368211,
-      "grad_norm": 0.0012581481132656336,
-      "learning_rate": 0.00019999399440877126,
-      "loss": 46.0,
-      "step": 21673
-    },
-    {
-      "epoch": 3.4904384234469985,
-      "grad_norm": 0.005315429996699095,
-      "learning_rate": 0.00019999399385430791,
-      "loss": 46.0,
-      "step": 21674
-    },
-    {
-      "epoch": 3.490599460525786,
-      "grad_norm": 0.007032371591776609,
-      "learning_rate": 0.00019999399329981898,
-      "loss": 46.0,
-      "step": 21675
-    },
-    {
-      "epoch": 3.4907604976045734,
-      "grad_norm": 0.0013241710839793086,
-      "learning_rate": 0.00019999399274530445,
-      "loss": 46.0,
-      "step": 21676
-    },
-    {
-      "epoch": 3.490921534683361,
-      "grad_norm": 0.01054837740957737,
-      "learning_rate": 0.00019999399219076432,
-      "loss": 46.0,
-      "step": 21677
-    },
-    {
-      "epoch": 3.4910825717621483,
-      "grad_norm": 0.012286514975130558,
-      "learning_rate": 0.00019999399163619862,
-      "loss": 46.0,
-      "step": 21678
-    },
-    {
-      "epoch": 3.4912436088409358,
-      "grad_norm": 0.001990196295082569,
-      "learning_rate": 0.0001999939910816073,
-      "loss": 46.0,
-      "step": 21679
-    },
-    {
-      "epoch": 3.491404645919723,
-      "grad_norm": 0.009741852059960365,
-      "learning_rate": 0.00019999399052699038,
-      "loss": 46.0,
-      "step": 21680
-    },
-    {
-      "epoch": 3.4915656829985102,
-      "grad_norm": 0.0013365488266572356,
-      "learning_rate": 0.0001999939899723479,
-      "loss": 46.0,
-      "step": 21681
-    },
-    {
-      "epoch": 3.4917267200772977,
-      "grad_norm": 0.0032115608919411898,
-      "learning_rate": 0.0001999939894176798,
-      "loss": 46.0,
-      "step": 21682
-    },
-    {
-      "epoch": 3.491887757156085,
-      "grad_norm": 0.006293166894465685,
-      "learning_rate": 0.00019999398886298613,
-      "loss": 46.0,
-      "step": 21683
-    },
-    {
-      "epoch": 3.4920487942348726,
-      "grad_norm": 0.00734447967261076,
-      "learning_rate": 0.00019999398830826683,
-      "loss": 46.0,
-      "step": 21684
-    },
-    {
-      "epoch": 3.49220983131366,
-      "grad_norm": 0.0027137629222124815,
-      "learning_rate": 0.00019999398775352197,
-      "loss": 46.0,
-      "step": 21685
-    },
-    {
-      "epoch": 3.4923708683924475,
-      "grad_norm": 0.006050765514373779,
-      "learning_rate": 0.00019999398719875152,
-      "loss": 46.0,
-      "step": 21686
-    },
-    {
-      "epoch": 3.492531905471235,
-      "grad_norm": 0.002486526034772396,
-      "learning_rate": 0.00019999398664395543,
-      "loss": 46.0,
-      "step": 21687
-    },
-    {
-      "epoch": 3.492692942550022,
-      "grad_norm": 0.005964949261397123,
-      "learning_rate": 0.00019999398608913378,
-      "loss": 46.0,
-      "step": 21688
-    },
-    {
-      "epoch": 3.4928539796288094,
-      "grad_norm": 0.003359105670824647,
-      "learning_rate": 0.00019999398553428654,
-      "loss": 46.0,
-      "step": 21689
-    },
-    {
-      "epoch": 3.493015016707597,
-      "grad_norm": 0.0008333980804309249,
-      "learning_rate": 0.0001999939849794137,
-      "loss": 46.0,
-      "step": 21690
-    },
-    {
-      "epoch": 3.4931760537863843,
-      "grad_norm": 0.00577760674059391,
-      "learning_rate": 0.0001999939844245153,
-      "loss": 46.0,
-      "step": 21691
-    },
-    {
-      "epoch": 3.4933370908651717,
-      "grad_norm": 0.008563734591007233,
-      "learning_rate": 0.00019999398386959125,
-      "loss": 46.0,
-      "step": 21692
-    },
-    {
-      "epoch": 3.493498127943959,
-      "grad_norm": 0.0029687120113521814,
-      "learning_rate": 0.00019999398331464163,
-      "loss": 46.0,
-      "step": 21693
-    },
-    {
-      "epoch": 3.4936591650227466,
-      "grad_norm": 0.020417695865035057,
-      "learning_rate": 0.0001999939827596664,
-      "loss": 46.0,
-      "step": 21694
-    },
-    {
-      "epoch": 3.493820202101534,
-      "grad_norm": 0.007189034018665552,
-      "learning_rate": 0.00019999398220466562,
-      "loss": 46.0,
-      "step": 21695
-    },
-    {
-      "epoch": 3.493981239180321,
-      "grad_norm": 0.0016737758414819837,
-      "learning_rate": 0.00019999398164963921,
-      "loss": 46.0,
-      "step": 21696
-    },
-    {
-      "epoch": 3.4941422762591086,
-      "grad_norm": 0.004586674738675356,
-      "learning_rate": 0.00019999398109458723,
-      "loss": 46.0,
-      "step": 21697
-    },
-    {
-      "epoch": 3.494303313337896,
-      "grad_norm": 0.0012549713719636202,
-      "learning_rate": 0.00019999398053950965,
-      "loss": 46.0,
-      "step": 21698
-    },
-    {
-      "epoch": 3.4944643504166835,
-      "grad_norm": 0.0020423668902367353,
-      "learning_rate": 0.00019999397998440648,
-      "loss": 46.0,
-      "step": 21699
-    },
-    {
-      "epoch": 3.494625387495471,
-      "grad_norm": 0.004017159342765808,
-      "learning_rate": 0.00019999397942927768,
-      "loss": 46.0,
-      "step": 21700
-    },
-    {
-      "epoch": 3.4947864245742584,
-      "grad_norm": 0.009589380584657192,
-      "learning_rate": 0.00019999397887412331,
-      "loss": 46.0,
-      "step": 21701
-    },
-    {
-      "epoch": 3.4949474616530454,
-      "grad_norm": 0.006555621977895498,
-      "learning_rate": 0.0001999939783189434,
-      "loss": 46.0,
-      "step": 21702
-    },
-    {
-      "epoch": 3.495108498731833,
-      "grad_norm": 0.0013683857396245003,
-      "learning_rate": 0.0001999939777637378,
-      "loss": 46.0,
-      "step": 21703
-    },
-    {
-      "epoch": 3.4952695358106203,
-      "grad_norm": 0.0011143877636641264,
-      "learning_rate": 0.00019999397720850667,
-      "loss": 46.0,
-      "step": 21704
-    },
-    {
-      "epoch": 3.4954305728894077,
-      "grad_norm": 0.017129722982645035,
-      "learning_rate": 0.00019999397665324993,
-      "loss": 46.0,
-      "step": 21705
-    },
-    {
-      "epoch": 3.495591609968195,
-      "grad_norm": 0.0059161619283258915,
-      "learning_rate": 0.0001999939760979676,
-      "loss": 46.0,
-      "step": 21706
-    },
-    {
-      "epoch": 3.4957526470469826,
-      "grad_norm": 0.004295977298170328,
-      "learning_rate": 0.00019999397554265968,
-      "loss": 46.0,
-      "step": 21707
-    },
-    {
-      "epoch": 3.49591368412577,
-      "grad_norm": 0.004680391866713762,
-      "learning_rate": 0.00019999397498732618,
-      "loss": 46.0,
-      "step": 21708
-    },
-    {
-      "epoch": 3.4960747212045575,
-      "grad_norm": 0.005192800424993038,
-      "learning_rate": 0.00019999397443196704,
-      "loss": 46.0,
-      "step": 21709
-    },
-    {
-      "epoch": 3.496235758283345,
-      "grad_norm": 0.0005860895616933703,
-      "learning_rate": 0.00019999397387658233,
-      "loss": 46.0,
-      "step": 21710
-    },
-    {
-      "epoch": 3.496396795362132,
-      "grad_norm": 0.007153054233640432,
-      "learning_rate": 0.00019999397332117204,
-      "loss": 46.0,
-      "step": 21711
-    },
-    {
-      "epoch": 3.4965578324409194,
-      "grad_norm": 0.005331411026418209,
-      "learning_rate": 0.00019999397276573616,
-      "loss": 46.0,
-      "step": 21712
-    },
-    {
-      "epoch": 3.496718869519707,
-      "grad_norm": 0.008422144688665867,
-      "learning_rate": 0.00019999397221027464,
-      "loss": 46.0,
-      "step": 21713
-    },
-    {
-      "epoch": 3.4968799065984943,
-      "grad_norm": 0.006241340655833483,
-      "learning_rate": 0.00019999397165478756,
-      "loss": 46.0,
-      "step": 21714
-    },
-    {
-      "epoch": 3.497040943677282,
-      "grad_norm": 0.014558670111000538,
-      "learning_rate": 0.00019999397109927492,
-      "loss": 46.0,
-      "step": 21715
-    },
-    {
-      "epoch": 3.4972019807560693,
-      "grad_norm": 0.0011415955377742648,
-      "learning_rate": 0.00019999397054373664,
-      "loss": 46.0,
-      "step": 21716
-    },
-    {
-      "epoch": 3.4973630178348563,
-      "grad_norm": 0.010599583387374878,
-      "learning_rate": 0.0001999939699881728,
-      "loss": 46.0,
-      "step": 21717
-    },
-    {
-      "epoch": 3.4975240549136437,
-      "grad_norm": 0.003147561103105545,
-      "learning_rate": 0.00019999396943258332,
-      "loss": 46.0,
-      "step": 21718
-    },
-    {
-      "epoch": 3.497685091992431,
-      "grad_norm": 0.007108535151928663,
-      "learning_rate": 0.0001999939688769683,
-      "loss": 46.0,
-      "step": 21719
-    },
-    {
-      "epoch": 3.4978461290712186,
-      "grad_norm": 0.0010863060597330332,
-      "learning_rate": 0.00019999396832132767,
-      "loss": 46.0,
-      "step": 21720
-    },
-    {
-      "epoch": 3.498007166150006,
-      "grad_norm": 0.002283665118739009,
-      "learning_rate": 0.00019999396776566143,
-      "loss": 46.0,
-      "step": 21721
-    },
-    {
-      "epoch": 3.4981682032287935,
-      "grad_norm": 0.004716191906481981,
-      "learning_rate": 0.0001999939672099696,
-      "loss": 46.0,
-      "step": 21722
-    },
-    {
-      "epoch": 3.498329240307581,
-      "grad_norm": 0.001842303667217493,
-      "learning_rate": 0.00019999396665425218,
-      "loss": 46.0,
-      "step": 21723
-    },
-    {
-      "epoch": 3.4984902773863684,
-      "grad_norm": 0.004425540566444397,
-      "learning_rate": 0.00019999396609850914,
-      "loss": 46.0,
-      "step": 21724
-    },
-    {
-      "epoch": 3.4986513144651554,
-      "grad_norm": 0.006759363692253828,
-      "learning_rate": 0.00019999396554274058,
-      "loss": 46.0,
-      "step": 21725
-    },
-    {
-      "epoch": 3.498812351543943,
-      "grad_norm": 0.00343524431809783,
-      "learning_rate": 0.00019999396498694637,
-      "loss": 46.0,
-      "step": 21726
-    },
-    {
-      "epoch": 3.4989733886227303,
-      "grad_norm": 0.010361671447753906,
-      "learning_rate": 0.00019999396443112657,
-      "loss": 46.0,
-      "step": 21727
-    },
-    {
-      "epoch": 3.499134425701518,
-      "grad_norm": 0.004125670995563269,
-      "learning_rate": 0.0001999939638752812,
-      "loss": 46.0,
-      "step": 21728
-    },
-    {
-      "epoch": 3.4992954627803052,
-      "grad_norm": 0.004922444466501474,
-      "learning_rate": 0.0001999939633194102,
-      "loss": 46.0,
-      "step": 21729
-    },
-    {
-      "epoch": 3.4994564998590927,
-      "grad_norm": 0.003544427454471588,
-      "learning_rate": 0.00019999396276351364,
-      "loss": 46.0,
-      "step": 21730
-    },
-    {
-      "epoch": 3.49961753693788,
-      "grad_norm": 0.001314751454629004,
-      "learning_rate": 0.00019999396220759147,
-      "loss": 46.0,
-      "step": 21731
-    },
-    {
-      "epoch": 3.499778574016667,
-      "grad_norm": 0.0017822623485699296,
-      "learning_rate": 0.0001999939616516437,
-      "loss": 46.0,
-      "step": 21732
-    },
-    {
-      "epoch": 3.4999396110954546,
-      "grad_norm": 0.009229912422597408,
-      "learning_rate": 0.00019999396109567037,
-      "loss": 46.0,
-      "step": 21733
-    },
-    {
-      "epoch": 3.500100648174242,
-      "grad_norm": 0.016182197257876396,
-      "learning_rate": 0.0001999939605396714,
-      "loss": 46.0,
-      "step": 21734
-    },
-    {
-      "epoch": 3.5002616852530295,
-      "grad_norm": 0.0021189632825553417,
-      "learning_rate": 0.0001999939599836469,
-      "loss": 46.0,
-      "step": 21735
-    },
-    {
-      "epoch": 3.500422722331817,
-      "grad_norm": 0.002374927746132016,
-      "learning_rate": 0.00019999395942759676,
-      "loss": 46.0,
-      "step": 21736
-    },
-    {
-      "epoch": 3.5005837594106044,
-      "grad_norm": 0.00620607566088438,
-      "learning_rate": 0.00019999395887152104,
-      "loss": 46.0,
-      "step": 21737
-    },
-    {
-      "epoch": 3.500744796489392,
-      "grad_norm": 0.006021073553711176,
-      "learning_rate": 0.0001999939583154197,
-      "loss": 46.0,
-      "step": 21738
-    },
-    {
-      "epoch": 3.5009058335681793,
-      "grad_norm": 0.007702481932938099,
-      "learning_rate": 0.00019999395775929278,
-      "loss": 46.0,
-      "step": 21739
-    },
-    {
-      "epoch": 3.5010668706469663,
-      "grad_norm": 0.0007809145608916879,
-      "learning_rate": 0.0001999939572031403,
-      "loss": 46.0,
-      "step": 21740
-    },
-    {
-      "epoch": 3.5012279077257538,
-      "grad_norm": 0.002246558666229248,
-      "learning_rate": 0.00019999395664696218,
-      "loss": 46.0,
-      "step": 21741
-    },
-    {
-      "epoch": 3.501388944804541,
-      "grad_norm": 0.0044090342707931995,
-      "learning_rate": 0.0001999939560907585,
-      "loss": 46.0,
-      "step": 21742
-    },
-    {
-      "epoch": 3.5015499818833287,
-      "grad_norm": 0.004719299264252186,
-      "learning_rate": 0.00019999395553452922,
-      "loss": 46.0,
-      "step": 21743
-    },
-    {
-      "epoch": 3.501711018962116,
-      "grad_norm": 0.0011664681369438767,
-      "learning_rate": 0.00019999395497827434,
-      "loss": 46.0,
-      "step": 21744
-    },
-    {
-      "epoch": 3.5018720560409036,
-      "grad_norm": 0.0008961429120972753,
-      "learning_rate": 0.00019999395442199386,
-      "loss": 46.0,
-      "step": 21745
-    },
-    {
-      "epoch": 3.5020330931196906,
-      "grad_norm": 0.004494569730013609,
-      "learning_rate": 0.0001999939538656878,
-      "loss": 46.0,
-      "step": 21746
-    },
-    {
-      "epoch": 3.502194130198478,
-      "grad_norm": 0.01446344144642353,
-      "learning_rate": 0.00019999395330935613,
-      "loss": 46.0,
-      "step": 21747
-    },
-    {
-      "epoch": 3.5023551672772655,
-      "grad_norm": 0.0036420905962586403,
-      "learning_rate": 0.0001999939527529989,
-      "loss": 46.0,
-      "step": 21748
-    },
-    {
-      "epoch": 3.502516204356053,
-      "grad_norm": 0.0023566405288875103,
-      "learning_rate": 0.00019999395219661605,
-      "loss": 46.0,
-      "step": 21749
-    },
-    {
-      "epoch": 3.5026772414348404,
-      "grad_norm": 0.001361611532047391,
-      "learning_rate": 0.00019999395164020762,
-      "loss": 46.0,
-      "step": 21750
-    },
-    {
-      "epoch": 3.502838278513628,
-      "grad_norm": 0.009291104972362518,
-      "learning_rate": 0.00019999395108377357,
-      "loss": 46.0,
-      "step": 21751
-    },
-    {
-      "epoch": 3.5029993155924153,
-      "grad_norm": 0.009813607670366764,
-      "learning_rate": 0.00019999395052731396,
-      "loss": 46.0,
-      "step": 21752
-    },
-    {
-      "epoch": 3.5031603526712027,
-      "grad_norm": 0.0020451825112104416,
-      "learning_rate": 0.00019999394997082873,
-      "loss": 46.0,
-      "step": 21753
-    },
-    {
-      "epoch": 3.50332138974999,
-      "grad_norm": 0.005616406444460154,
-      "learning_rate": 0.00019999394941431792,
-      "loss": 46.0,
-      "step": 21754
-    },
-    {
-      "epoch": 3.503482426828777,
-      "grad_norm": 0.002111807931214571,
-      "learning_rate": 0.0001999939488577815,
-      "loss": 46.0,
-      "step": 21755
-    },
-    {
-      "epoch": 3.5036434639075646,
-      "grad_norm": 0.001281952252611518,
-      "learning_rate": 0.0001999939483012195,
-      "loss": 46.0,
-      "step": 21756
-    },
-    {
-      "epoch": 3.503804500986352,
-      "grad_norm": 0.0019127260893583298,
-      "learning_rate": 0.00019999394774463194,
-      "loss": 46.0,
-      "step": 21757
-    },
-    {
-      "epoch": 3.5039655380651396,
-      "grad_norm": 0.0012628089170902967,
-      "learning_rate": 0.00019999394718801873,
-      "loss": 46.0,
-      "step": 21758
-    },
-    {
-      "epoch": 3.504126575143927,
-      "grad_norm": 0.001724792760796845,
-      "learning_rate": 0.00019999394663137995,
-      "loss": 46.0,
-      "step": 21759
-    },
-    {
-      "epoch": 3.504287612222714,
-      "grad_norm": 0.003039327682927251,
-      "learning_rate": 0.0001999939460747156,
-      "loss": 46.0,
-      "step": 21760
-    },
-    {
-      "epoch": 3.5044486493015015,
-      "grad_norm": 0.0009637277689762414,
-      "learning_rate": 0.00019999394551802564,
-      "loss": 46.0,
-      "step": 21761
-    },
-    {
-      "epoch": 3.504609686380289,
-      "grad_norm": 0.0008546972530893981,
-      "learning_rate": 0.00019999394496131008,
-      "loss": 46.0,
-      "step": 21762
-    },
-    {
-      "epoch": 3.5047707234590764,
-      "grad_norm": 0.008093750104308128,
-      "learning_rate": 0.0001999939444045689,
-      "loss": 46.0,
-      "step": 21763
-    },
-    {
-      "epoch": 3.504931760537864,
-      "grad_norm": 0.006316937506198883,
-      "learning_rate": 0.0001999939438478022,
-      "loss": 46.0,
-      "step": 21764
-    },
-    {
-      "epoch": 3.5050927976166513,
-      "grad_norm": 0.003119508270174265,
-      "learning_rate": 0.00019999394329100985,
-      "loss": 46.0,
-      "step": 21765
-    },
-    {
-      "epoch": 3.5052538346954387,
-      "grad_norm": 0.004355509765446186,
-      "learning_rate": 0.0001999939427341919,
-      "loss": 46.0,
-      "step": 21766
-    },
-    {
-      "epoch": 3.505414871774226,
-      "grad_norm": 0.002460635267198086,
-      "learning_rate": 0.00019999394217734838,
-      "loss": 46.0,
-      "step": 21767
-    },
-    {
-      "epoch": 3.5055759088530136,
-      "grad_norm": 0.003075494896620512,
-      "learning_rate": 0.0001999939416204793,
-      "loss": 46.0,
-      "step": 21768
-    },
-    {
-      "epoch": 3.505736945931801,
-      "grad_norm": 0.003784006228670478,
-      "learning_rate": 0.00019999394106358454,
-      "loss": 46.0,
-      "step": 21769
-    },
-    {
-      "epoch": 3.505897983010588,
-      "grad_norm": 0.005429904907941818,
-      "learning_rate": 0.00019999394050666426,
-      "loss": 46.0,
-      "step": 21770
-    },
-    {
-      "epoch": 3.5060590200893755,
-      "grad_norm": 0.006247547920793295,
-      "learning_rate": 0.00019999393994971835,
-      "loss": 46.0,
-      "step": 21771
-    },
-    {
-      "epoch": 3.506220057168163,
-      "grad_norm": 0.00142267148476094,
-      "learning_rate": 0.00019999393939274687,
-      "loss": 46.0,
-      "step": 21772
-    },
-    {
-      "epoch": 3.5063810942469504,
-      "grad_norm": 0.010881015099585056,
-      "learning_rate": 0.00019999393883574976,
-      "loss": 46.0,
-      "step": 21773
-    },
-    {
-      "epoch": 3.506542131325738,
-      "grad_norm": 0.006131685804575682,
-      "learning_rate": 0.0001999939382787271,
-      "loss": 46.0,
-      "step": 21774
-    },
-    {
-      "epoch": 3.506703168404525,
-      "grad_norm": 0.0019825256895273924,
-      "learning_rate": 0.00019999393772167885,
-      "loss": 46.0,
-      "step": 21775
-    },
-    {
-      "epoch": 3.5068642054833123,
-      "grad_norm": 0.002985789906233549,
-      "learning_rate": 0.00019999393716460496,
-      "loss": 46.0,
-      "step": 21776
-    },
-    {
-      "epoch": 3.5070252425621,
-      "grad_norm": 0.0010978684294968843,
-      "learning_rate": 0.0001999939366075055,
-      "loss": 46.0,
-      "step": 21777
-    },
-    {
-      "epoch": 3.5071862796408872,
-      "grad_norm": 0.0014760037884116173,
-      "learning_rate": 0.00019999393605038047,
-      "loss": 46.0,
-      "step": 21778
-    },
-    {
-      "epoch": 3.5073473167196747,
-      "grad_norm": 0.0018413078505545855,
-      "learning_rate": 0.00019999393549322982,
-      "loss": 46.0,
-      "step": 21779
-    },
-    {
-      "epoch": 3.507508353798462,
-      "grad_norm": 0.0013160889502614737,
-      "learning_rate": 0.00019999393493605358,
-      "loss": 46.0,
-      "step": 21780
-    },
-    {
-      "epoch": 3.5076693908772496,
-      "grad_norm": 0.004745855461806059,
-      "learning_rate": 0.00019999393437885175,
-      "loss": 46.0,
-      "step": 21781
-    },
-    {
-      "epoch": 3.507830427956037,
-      "grad_norm": 0.00353249697946012,
-      "learning_rate": 0.00019999393382162434,
-      "loss": 46.0,
-      "step": 21782
-    },
-    {
-      "epoch": 3.5079914650348245,
-      "grad_norm": 0.0032081259414553642,
-      "learning_rate": 0.0001999939332643713,
-      "loss": 46.0,
-      "step": 21783
-    },
-    {
-      "epoch": 3.5081525021136115,
-      "grad_norm": 0.008396645076572895,
-      "learning_rate": 0.00019999393270709267,
-      "loss": 46.0,
-      "step": 21784
-    },
-    {
-      "epoch": 3.508313539192399,
-      "grad_norm": 0.002217477886006236,
-      "learning_rate": 0.00019999393214978846,
-      "loss": 46.0,
-      "step": 21785
-    },
-    {
-      "epoch": 3.5084745762711864,
-      "grad_norm": 0.0020149927586317062,
-      "learning_rate": 0.0001999939315924587,
-      "loss": 46.0,
-      "step": 21786
-    },
-    {
-      "epoch": 3.508635613349974,
-      "grad_norm": 0.0017513299826532602,
-      "learning_rate": 0.0001999939310351033,
-      "loss": 46.0,
-      "step": 21787
-    },
-    {
-      "epoch": 3.5087966504287613,
-      "grad_norm": 0.0013605781132355332,
-      "learning_rate": 0.00019999393047772233,
-      "loss": 46.0,
-      "step": 21788
-    },
-    {
-      "epoch": 3.5089576875075488,
-      "grad_norm": 0.0031284557189792395,
-      "learning_rate": 0.00019999392992031573,
-      "loss": 46.0,
-      "step": 21789
-    },
-    {
-      "epoch": 3.509118724586336,
-      "grad_norm": 0.0033485747408121824,
-      "learning_rate": 0.00019999392936288356,
-      "loss": 46.0,
-      "step": 21790
-    },
-    {
-      "epoch": 3.5092797616651232,
-      "grad_norm": 0.0019145788392052054,
-      "learning_rate": 0.00019999392880542583,
-      "loss": 46.0,
-      "step": 21791
-    },
-    {
-      "epoch": 3.5094407987439107,
-      "grad_norm": 0.0011674072593450546,
-      "learning_rate": 0.00019999392824794244,
-      "loss": 46.0,
-      "step": 21792
-    },
-    {
-      "epoch": 3.509601835822698,
-      "grad_norm": 0.0018394498620182276,
-      "learning_rate": 0.0001999939276904335,
-      "loss": 46.0,
-      "step": 21793
-    },
-    {
-      "epoch": 3.5097628729014856,
-      "grad_norm": 0.002685382729396224,
-      "learning_rate": 0.00019999392713289894,
-      "loss": 46.0,
-      "step": 21794
-    },
-    {
-      "epoch": 3.509923909980273,
-      "grad_norm": 0.003416551975533366,
-      "learning_rate": 0.00019999392657533882,
-      "loss": 46.0,
-      "step": 21795
-    },
-    {
-      "epoch": 3.5100849470590605,
-      "grad_norm": 0.0033724710810929537,
-      "learning_rate": 0.00019999392601775307,
-      "loss": 46.0,
-      "step": 21796
-    },
-    {
-      "epoch": 3.510245984137848,
-      "grad_norm": 0.004365818575024605,
-      "learning_rate": 0.00019999392546014177,
-      "loss": 46.0,
-      "step": 21797
-    },
-    {
-      "epoch": 3.5104070212166354,
-      "grad_norm": 0.018542608246207237,
-      "learning_rate": 0.00019999392490250485,
-      "loss": 46.0,
-      "step": 21798
-    },
-    {
-      "epoch": 3.5105680582954224,
-      "grad_norm": 0.0024413387291133404,
-      "learning_rate": 0.00019999392434484232,
-      "loss": 46.0,
-      "step": 21799
-    },
-    {
-      "epoch": 3.51072909537421,
-      "grad_norm": 0.0019974494352936745,
-      "learning_rate": 0.00019999392378715426,
-      "loss": 46.0,
-      "step": 21800
-    },
-    {
-      "epoch": 3.5108901324529973,
-      "grad_norm": 0.0056679206900298595,
-      "learning_rate": 0.00019999392322944055,
-      "loss": 46.0,
-      "step": 21801
-    },
-    {
-      "epoch": 3.5110511695317848,
-      "grad_norm": 0.0060379174537956715,
-      "learning_rate": 0.00019999392267170123,
-      "loss": 46.0,
-      "step": 21802
-    },
-    {
-      "epoch": 3.511212206610572,
-      "grad_norm": 0.0011425487464293838,
-      "learning_rate": 0.00019999392211393638,
-      "loss": 46.0,
-      "step": 21803
-    },
-    {
-      "epoch": 3.511373243689359,
-      "grad_norm": 0.001393473008647561,
-      "learning_rate": 0.00019999392155614588,
-      "loss": 46.0,
-      "step": 21804
-    },
-    {
-      "epoch": 3.5115342807681467,
-      "grad_norm": 0.01002795435488224,
-      "learning_rate": 0.0001999939209983298,
-      "loss": 46.0,
-      "step": 21805
-    },
-    {
-      "epoch": 3.511695317846934,
-      "grad_norm": 0.006619532126933336,
-      "learning_rate": 0.00019999392044048816,
-      "loss": 46.0,
-      "step": 21806
-    },
-    {
-      "epoch": 3.5118563549257216,
-      "grad_norm": 0.0060730138793587685,
-      "learning_rate": 0.00019999391988262088,
-      "loss": 46.0,
-      "step": 21807
-    },
-    {
-      "epoch": 3.512017392004509,
-      "grad_norm": 0.006986928638070822,
-      "learning_rate": 0.00019999391932472803,
-      "loss": 46.0,
-      "step": 21808
-    },
-    {
-      "epoch": 3.5121784290832965,
-      "grad_norm": 0.0018528768559917808,
-      "learning_rate": 0.0001999939187668096,
-      "loss": 46.0,
-      "step": 21809
-    },
-    {
-      "epoch": 3.512339466162084,
-      "grad_norm": 0.001747790607623756,
-      "learning_rate": 0.00019999391820886558,
-      "loss": 46.0,
-      "step": 21810
-    },
-    {
-      "epoch": 3.5125005032408714,
-      "grad_norm": 0.00402861600741744,
-      "learning_rate": 0.00019999391765089595,
-      "loss": 46.0,
-      "step": 21811
-    },
-    {
-      "epoch": 3.512661540319659,
-      "grad_norm": 0.008654254488646984,
-      "learning_rate": 0.0001999939170929007,
-      "loss": 46.0,
-      "step": 21812
-    },
-    {
-      "epoch": 3.512822577398446,
-      "grad_norm": 0.004390066955238581,
-      "learning_rate": 0.0001999939165348799,
-      "loss": 46.0,
-      "step": 21813
-    },
-    {
-      "epoch": 3.5129836144772333,
-      "grad_norm": 0.005463108420372009,
-      "learning_rate": 0.00019999391597683348,
-      "loss": 46.0,
-      "step": 21814
-    },
-    {
-      "epoch": 3.5131446515560207,
-      "grad_norm": 0.0008358869818039238,
-      "learning_rate": 0.00019999391541876147,
-      "loss": 46.0,
-      "step": 21815
-    },
-    {
-      "epoch": 3.513305688634808,
-      "grad_norm": 0.002591717289760709,
-      "learning_rate": 0.00019999391486066388,
-      "loss": 46.0,
-      "step": 21816
-    },
-    {
-      "epoch": 3.5134667257135956,
-      "grad_norm": 0.009204046800732613,
-      "learning_rate": 0.0001999939143025407,
-      "loss": 46.0,
-      "step": 21817
-    },
-    {
-      "epoch": 3.513627762792383,
-      "grad_norm": 0.0013155278284102678,
-      "learning_rate": 0.0001999939137443919,
-      "loss": 46.0,
-      "step": 21818
-    },
-    {
-      "epoch": 3.51378879987117,
-      "grad_norm": 0.00045975917601026595,
-      "learning_rate": 0.00019999391318621754,
-      "loss": 46.0,
-      "step": 21819
-    },
-    {
-      "epoch": 3.5139498369499576,
-      "grad_norm": 0.0005814654286950827,
-      "learning_rate": 0.00019999391262801757,
-      "loss": 46.0,
-      "step": 21820
-    },
-    {
-      "epoch": 3.514110874028745,
-      "grad_norm": 0.0014132732758298516,
-      "learning_rate": 0.00019999391206979201,
-      "loss": 46.0,
-      "step": 21821
-    },
-    {
-      "epoch": 3.5142719111075325,
-      "grad_norm": 0.003257524687796831,
-      "learning_rate": 0.00019999391151154084,
-      "loss": 46.0,
-      "step": 21822
-    },
-    {
-      "epoch": 3.51443294818632,
-      "grad_norm": 0.001962295500561595,
-      "learning_rate": 0.00019999391095326408,
-      "loss": 46.0,
-      "step": 21823
-    },
-    {
-      "epoch": 3.5145939852651074,
-      "grad_norm": 0.008457900024950504,
-      "learning_rate": 0.00019999391039496174,
-      "loss": 46.0,
-      "step": 21824
-    },
-    {
-      "epoch": 3.514755022343895,
-      "grad_norm": 0.005166944116353989,
-      "learning_rate": 0.00019999390983663383,
-      "loss": 46.0,
-      "step": 21825
-    },
-    {
-      "epoch": 3.5149160594226823,
-      "grad_norm": 0.004429202061146498,
-      "learning_rate": 0.00019999390927828028,
-      "loss": 46.0,
-      "step": 21826
-    },
-    {
-      "epoch": 3.5150770965014697,
-      "grad_norm": 0.008082548156380653,
-      "learning_rate": 0.00019999390871990115,
-      "loss": 46.0,
-      "step": 21827
-    },
-    {
-      "epoch": 3.5152381335802567,
-      "grad_norm": 0.004783708602190018,
-      "learning_rate": 0.00019999390816149642,
-      "loss": 46.0,
-      "step": 21828
-    },
-    {
-      "epoch": 3.515399170659044,
-      "grad_norm": 0.007234132383018732,
-      "learning_rate": 0.00019999390760306614,
-      "loss": 46.0,
-      "step": 21829
-    },
-    {
-      "epoch": 3.5155602077378316,
-      "grad_norm": 0.006134225986897945,
-      "learning_rate": 0.00019999390704461022,
-      "loss": 46.0,
-      "step": 21830
-    },
-    {
-      "epoch": 3.515721244816619,
-      "grad_norm": 0.007768153678625822,
-      "learning_rate": 0.00019999390648612874,
-      "loss": 46.0,
-      "step": 21831
-    },
-    {
-      "epoch": 3.5158822818954065,
-      "grad_norm": 0.016721682623028755,
-      "learning_rate": 0.00019999390592762164,
-      "loss": 46.0,
-      "step": 21832
-    },
-    {
-      "epoch": 3.5160433189741935,
-      "grad_norm": 0.005124760791659355,
-      "learning_rate": 0.00019999390536908895,
-      "loss": 46.0,
-      "step": 21833
-    },
-    {
-      "epoch": 3.516204356052981,
-      "grad_norm": 0.0076878671534359455,
-      "learning_rate": 0.00019999390481053068,
-      "loss": 46.0,
-      "step": 21834
-    },
-    {
-      "epoch": 3.5163653931317684,
-      "grad_norm": 0.00998994242399931,
-      "learning_rate": 0.0001999939042519468,
-      "loss": 46.0,
-      "step": 21835
-    },
-    {
-      "epoch": 3.516526430210556,
-      "grad_norm": 0.0011731629492715001,
-      "learning_rate": 0.00019999390369333735,
-      "loss": 46.0,
-      "step": 21836
-    },
-    {
-      "epoch": 3.5166874672893433,
-      "grad_norm": 0.007419189903885126,
-      "learning_rate": 0.00019999390313470229,
-      "loss": 46.0,
-      "step": 21837
-    },
-    {
-      "epoch": 3.516848504368131,
-      "grad_norm": 0.0008965274901129305,
-      "learning_rate": 0.00019999390257604164,
-      "loss": 46.0,
-      "step": 21838
-    },
-    {
-      "epoch": 3.5170095414469182,
-      "grad_norm": 0.0019052295247092843,
-      "learning_rate": 0.0001999939020173554,
-      "loss": 46.0,
-      "step": 21839
-    },
-    {
-      "epoch": 3.5171705785257057,
-      "grad_norm": 0.003496067365631461,
-      "learning_rate": 0.00019999390145864355,
-      "loss": 46.0,
-      "step": 21840
-    },
-    {
-      "epoch": 3.517331615604493,
-      "grad_norm": 0.0008139025303535163,
-      "learning_rate": 0.00019999390089990612,
-      "loss": 46.0,
-      "step": 21841
-    },
-    {
-      "epoch": 3.5174926526832806,
-      "grad_norm": 0.0031868871301412582,
-      "learning_rate": 0.00019999390034114312,
-      "loss": 46.0,
-      "step": 21842
-    },
-    {
-      "epoch": 3.5176536897620676,
-      "grad_norm": 0.0012927409261465073,
-      "learning_rate": 0.0001999938997823545,
-      "loss": 46.0,
-      "step": 21843
-    },
-    {
-      "epoch": 3.517814726840855,
-      "grad_norm": 0.004887499380856752,
-      "learning_rate": 0.00019999389922354028,
-      "loss": 46.0,
-      "step": 21844
-    },
-    {
-      "epoch": 3.5179757639196425,
-      "grad_norm": 0.004289343021810055,
-      "learning_rate": 0.0001999938986647005,
-      "loss": 46.0,
-      "step": 21845
-    },
-    {
-      "epoch": 3.51813680099843,
-      "grad_norm": 0.007844308391213417,
-      "learning_rate": 0.0001999938981058351,
-      "loss": 46.0,
-      "step": 21846
-    },
-    {
-      "epoch": 3.5182978380772174,
-      "grad_norm": 0.0016463998472318053,
-      "learning_rate": 0.0001999938975469441,
-      "loss": 46.0,
-      "step": 21847
-    },
-    {
-      "epoch": 3.5184588751560044,
-      "grad_norm": 0.0053892736323177814,
-      "learning_rate": 0.0001999938969880275,
-      "loss": 46.0,
-      "step": 21848
-    },
-    {
-      "epoch": 3.518619912234792,
-      "grad_norm": 0.006787822116166353,
-      "learning_rate": 0.00019999389642908532,
-      "loss": 46.0,
-      "step": 21849
-    },
-    {
-      "epoch": 3.5187809493135793,
-      "grad_norm": 0.002457831520587206,
-      "learning_rate": 0.00019999389587011757,
-      "loss": 46.0,
-      "step": 21850
-    },
-    {
-      "epoch": 3.5189419863923668,
-      "grad_norm": 0.013491563498973846,
-      "learning_rate": 0.0001999938953111242,
-      "loss": 46.0,
-      "step": 21851
-    },
-    {
-      "epoch": 3.5191030234711542,
-      "grad_norm": 0.009038117714226246,
-      "learning_rate": 0.00019999389475210526,
-      "loss": 46.0,
-      "step": 21852
-    },
-    {
-      "epoch": 3.5192640605499417,
-      "grad_norm": 0.0032559256069362164,
-      "learning_rate": 0.0001999938941930607,
-      "loss": 46.0,
-      "step": 21853
-    },
-    {
-      "epoch": 3.519425097628729,
-      "grad_norm": 0.002037127036601305,
-      "learning_rate": 0.00019999389363399054,
-      "loss": 46.0,
-      "step": 21854
-    },
-    {
-      "epoch": 3.5195861347075166,
-      "grad_norm": 0.0012325349962338805,
-      "learning_rate": 0.00019999389307489483,
-      "loss": 46.0,
-      "step": 21855
-    },
-    {
-      "epoch": 3.519747171786304,
-      "grad_norm": 0.009791783057153225,
-      "learning_rate": 0.0001999938925157735,
-      "loss": 46.0,
-      "step": 21856
-    },
-    {
-      "epoch": 3.519908208865091,
-      "grad_norm": 0.0015503528993576765,
-      "learning_rate": 0.00019999389195662657,
-      "loss": 46.0,
-      "step": 21857
-    },
-    {
-      "epoch": 3.5200692459438785,
-      "grad_norm": 0.01240500621497631,
-      "learning_rate": 0.00019999389139745404,
-      "loss": 46.0,
-      "step": 21858
-    },
-    {
-      "epoch": 3.520230283022666,
-      "grad_norm": 0.005723054055124521,
-      "learning_rate": 0.00019999389083825595,
-      "loss": 46.0,
-      "step": 21859
-    },
-    {
-      "epoch": 3.5203913201014534,
-      "grad_norm": 0.002513917861506343,
-      "learning_rate": 0.00019999389027903222,
-      "loss": 46.0,
-      "step": 21860
-    },
-    {
-      "epoch": 3.520552357180241,
-      "grad_norm": 0.009991351515054703,
-      "learning_rate": 0.00019999388971978294,
-      "loss": 46.0,
-      "step": 21861
-    },
-    {
-      "epoch": 3.5207133942590283,
-      "grad_norm": 0.00209894054569304,
-      "learning_rate": 0.00019999388916050806,
-      "loss": 46.0,
-      "step": 21862
-    },
-    {
-      "epoch": 3.5208744313378153,
-      "grad_norm": 0.004655070602893829,
-      "learning_rate": 0.00019999388860120757,
-      "loss": 46.0,
-      "step": 21863
-    },
-    {
-      "epoch": 3.5210354684166028,
-      "grad_norm": 0.0017953141359612346,
-      "learning_rate": 0.0001999938880418815,
-      "loss": 46.0,
-      "step": 21864
-    },
-    {
-      "epoch": 3.52119650549539,
-      "grad_norm": 0.008750016801059246,
-      "learning_rate": 0.00019999388748252983,
-      "loss": 46.0,
-      "step": 21865
-    },
-    {
-      "epoch": 3.5213575425741777,
-      "grad_norm": 0.0009777933591976762,
-      "learning_rate": 0.00019999388692315258,
-      "loss": 46.0,
-      "step": 21866
-    },
-    {
-      "epoch": 3.521518579652965,
-      "grad_norm": 0.0009155089501291513,
-      "learning_rate": 0.0001999938863637497,
-      "loss": 46.0,
-      "step": 21867
-    },
-    {
-      "epoch": 3.5216796167317526,
-      "grad_norm": 0.0011881996178999543,
-      "learning_rate": 0.00019999388580432126,
-      "loss": 46.0,
-      "step": 21868
-    },
-    {
-      "epoch": 3.52184065381054,
-      "grad_norm": 0.007826287299394608,
-      "learning_rate": 0.00019999388524486722,
-      "loss": 46.0,
-      "step": 21869
-    },
-    {
-      "epoch": 3.5220016908893275,
-      "grad_norm": 0.0018998130690306425,
-      "learning_rate": 0.0001999938846853876,
-      "loss": 46.0,
-      "step": 21870
-    },
-    {
-      "epoch": 3.522162727968115,
-      "grad_norm": 0.0011056291405111551,
-      "learning_rate": 0.00019999388412588235,
-      "loss": 46.0,
-      "step": 21871
-    },
-    {
-      "epoch": 3.522323765046902,
-      "grad_norm": 0.004786670673638582,
-      "learning_rate": 0.00019999388356635155,
-      "loss": 46.0,
-      "step": 21872
-    },
-    {
-      "epoch": 3.5224848021256894,
-      "grad_norm": 0.0033293634187430143,
-      "learning_rate": 0.0001999938830067951,
-      "loss": 46.0,
-      "step": 21873
-    },
-    {
-      "epoch": 3.522645839204477,
-      "grad_norm": 0.0009380434639751911,
-      "learning_rate": 0.0001999938824472131,
-      "loss": 46.0,
-      "step": 21874
-    },
-    {
-      "epoch": 3.5228068762832643,
-      "grad_norm": 0.005221309140324593,
-      "learning_rate": 0.0001999938818876055,
-      "loss": 46.0,
-      "step": 21875
-    },
-    {
-      "epoch": 3.5229679133620517,
-      "grad_norm": 0.0027566649951040745,
-      "learning_rate": 0.0001999938813279723,
-      "loss": 46.0,
-      "step": 21876
-    },
-    {
-      "epoch": 3.5231289504408387,
-      "grad_norm": 0.0045267026871442795,
-      "learning_rate": 0.00019999388076831354,
-      "loss": 46.0,
-      "step": 21877
-    },
-    {
-      "epoch": 3.523289987519626,
-      "grad_norm": 0.002671432215720415,
-      "learning_rate": 0.00019999388020862914,
-      "loss": 46.0,
-      "step": 21878
-    },
-    {
-      "epoch": 3.5234510245984136,
-      "grad_norm": 0.0039221071638166904,
-      "learning_rate": 0.00019999387964891917,
-      "loss": 46.0,
-      "step": 21879
-    },
-    {
-      "epoch": 3.523612061677201,
-      "grad_norm": 0.002746603451669216,
-      "learning_rate": 0.0001999938790891836,
-      "loss": 46.0,
-      "step": 21880
-    },
-    {
-      "epoch": 3.5237730987559885,
-      "grad_norm": 0.0025265049189329147,
-      "learning_rate": 0.00019999387852942245,
-      "loss": 46.0,
-      "step": 21881
-    },
-    {
-      "epoch": 3.523934135834776,
-      "grad_norm": 0.0025974467862397432,
-      "learning_rate": 0.00019999387796963567,
-      "loss": 46.0,
-      "step": 21882
-    },
-    {
-      "epoch": 3.5240951729135634,
-      "grad_norm": 0.015995929017663002,
-      "learning_rate": 0.00019999387740982333,
-      "loss": 46.0,
-      "step": 21883
-    },
-    {
-      "epoch": 3.524256209992351,
-      "grad_norm": 0.0018486761255189776,
-      "learning_rate": 0.0001999938768499854,
-      "loss": 46.0,
-      "step": 21884
-    },
-    {
-      "epoch": 3.5244172470711383,
-      "grad_norm": 0.0051537067629396915,
-      "learning_rate": 0.00019999387629012185,
-      "loss": 46.0,
-      "step": 21885
-    },
-    {
-      "epoch": 3.5245782841499254,
-      "grad_norm": 0.0029616758693009615,
-      "learning_rate": 0.00019999387573023275,
-      "loss": 46.0,
-      "step": 21886
-    },
-    {
-      "epoch": 3.524739321228713,
-      "grad_norm": 0.0038254093378782272,
-      "learning_rate": 0.000199993875170318,
-      "loss": 46.0,
-      "step": 21887
-    },
-    {
-      "epoch": 3.5249003583075003,
-      "grad_norm": 0.007253831252455711,
-      "learning_rate": 0.00019999387461037767,
-      "loss": 46.0,
-      "step": 21888
-    },
-    {
-      "epoch": 3.5250613953862877,
-      "grad_norm": 0.007263258099555969,
-      "learning_rate": 0.00019999387405041178,
-      "loss": 46.0,
-      "step": 21889
-    },
-    {
-      "epoch": 3.525222432465075,
-      "grad_norm": 0.0022202448453754187,
-      "learning_rate": 0.00019999387349042027,
-      "loss": 46.0,
-      "step": 21890
-    },
-    {
-      "epoch": 3.5253834695438626,
-      "grad_norm": 0.006503438111394644,
-      "learning_rate": 0.00019999387293040318,
-      "loss": 46.0,
-      "step": 21891
-    },
-    {
-      "epoch": 3.5255445066226496,
-      "grad_norm": 0.004652680829167366,
-      "learning_rate": 0.0001999938723703605,
-      "loss": 46.0,
-      "step": 21892
-    },
-    {
-      "epoch": 3.525705543701437,
-      "grad_norm": 0.0049568782560527325,
-      "learning_rate": 0.00019999387181029223,
-      "loss": 46.0,
-      "step": 21893
-    },
-    {
-      "epoch": 3.5258665807802245,
-      "grad_norm": 0.0015088096261024475,
-      "learning_rate": 0.00019999387125019835,
-      "loss": 46.0,
-      "step": 21894
-    },
-    {
-      "epoch": 3.526027617859012,
-      "grad_norm": 0.017559386789798737,
-      "learning_rate": 0.00019999387069007888,
-      "loss": 46.0,
-      "step": 21895
-    },
-    {
-      "epoch": 3.5261886549377994,
-      "grad_norm": 0.02956242486834526,
-      "learning_rate": 0.0001999938701299338,
-      "loss": 46.0,
-      "step": 21896
-    },
-    {
-      "epoch": 3.526349692016587,
-      "grad_norm": 0.001476361881941557,
-      "learning_rate": 0.00019999386956976315,
-      "loss": 46.0,
-      "step": 21897
-    },
-    {
-      "epoch": 3.5265107290953743,
-      "grad_norm": 0.0020461431704461575,
-      "learning_rate": 0.00019999386900956692,
-      "loss": 46.0,
-      "step": 21898
-    },
-    {
-      "epoch": 3.526671766174162,
-      "grad_norm": 0.001690125442110002,
-      "learning_rate": 0.00019999386844934508,
-      "loss": 46.0,
-      "step": 21899
-    },
-    {
-      "epoch": 3.5268328032529492,
-      "grad_norm": 0.0011963190045207739,
-      "learning_rate": 0.00019999386788909762,
-      "loss": 46.0,
-      "step": 21900
-    },
-    {
-      "epoch": 3.5269938403317362,
-      "grad_norm": 0.0076559740118682384,
-      "learning_rate": 0.0001999938673288246,
-      "loss": 46.0,
-      "step": 21901
-    },
-    {
-      "epoch": 3.5271548774105237,
-      "grad_norm": 0.005284005310386419,
-      "learning_rate": 0.000199993866768526,
-      "loss": 46.0,
-      "step": 21902
-    },
-    {
-      "epoch": 3.527315914489311,
-      "grad_norm": 0.012967171147465706,
-      "learning_rate": 0.00019999386620820177,
-      "loss": 46.0,
-      "step": 21903
-    },
-    {
-      "epoch": 3.5274769515680986,
-      "grad_norm": 0.0018241682555526495,
-      "learning_rate": 0.00019999386564785196,
-      "loss": 46.0,
-      "step": 21904
-    },
-    {
-      "epoch": 3.527637988646886,
-      "grad_norm": 0.002812751103192568,
-      "learning_rate": 0.00019999386508747654,
-      "loss": 46.0,
-      "step": 21905
-    },
-    {
-      "epoch": 3.5277990257256735,
-      "grad_norm": 0.003000043798238039,
-      "learning_rate": 0.0001999938645270756,
-      "loss": 46.0,
-      "step": 21906
-    },
-    {
-      "epoch": 3.5279600628044605,
-      "grad_norm": 0.001961148576810956,
-      "learning_rate": 0.000199993863966649,
-      "loss": 46.0,
-      "step": 21907
-    },
-    {
-      "epoch": 3.528121099883248,
-      "grad_norm": 0.009626796469092369,
-      "learning_rate": 0.0001999938634061968,
-      "loss": 46.0,
-      "step": 21908
-    },
-    {
-      "epoch": 3.5282821369620354,
-      "grad_norm": 0.0048358505591750145,
-      "learning_rate": 0.000199993862845719,
-      "loss": 46.0,
-      "step": 21909
-    },
-    {
-      "epoch": 3.528443174040823,
-      "grad_norm": 0.0034459978342056274,
-      "learning_rate": 0.00019999386228521565,
-      "loss": 46.0,
-      "step": 21910
-    },
-    {
-      "epoch": 3.5286042111196103,
-      "grad_norm": 0.0021168130915611982,
-      "learning_rate": 0.00019999386172468668,
-      "loss": 46.0,
-      "step": 21911
-    },
-    {
-      "epoch": 3.5287652481983978,
-      "grad_norm": 0.006441343575716019,
-      "learning_rate": 0.00019999386116413215,
-      "loss": 46.0,
-      "step": 21912
-    },
-    {
-      "epoch": 3.528926285277185,
-      "grad_norm": 0.0010449185501784086,
-      "learning_rate": 0.000199993860603552,
-      "loss": 46.0,
-      "step": 21913
-    },
-    {
-      "epoch": 3.5290873223559727,
-      "grad_norm": 0.0031315439846366644,
-      "learning_rate": 0.00019999386004294624,
-      "loss": 46.0,
-      "step": 21914
-    },
-    {
-      "epoch": 3.52924835943476,
-      "grad_norm": 0.002528954530134797,
-      "learning_rate": 0.00019999385948231492,
-      "loss": 46.0,
-      "step": 21915
-    },
-    {
-      "epoch": 3.529409396513547,
-      "grad_norm": 0.008319096639752388,
-      "learning_rate": 0.00019999385892165798,
-      "loss": 46.0,
-      "step": 21916
-    },
-    {
-      "epoch": 3.5295704335923346,
-      "grad_norm": 0.011010879650712013,
-      "learning_rate": 0.00019999385836097546,
-      "loss": 46.0,
-      "step": 21917
-    },
-    {
-      "epoch": 3.529731470671122,
-      "grad_norm": 0.002860456705093384,
-      "learning_rate": 0.00019999385780026735,
-      "loss": 46.0,
-      "step": 21918
-    },
-    {
-      "epoch": 3.5298925077499095,
-      "grad_norm": 0.0017834343016147614,
-      "learning_rate": 0.00019999385723953365,
-      "loss": 46.0,
-      "step": 21919
-    },
-    {
-      "epoch": 3.530053544828697,
-      "grad_norm": 0.0068252370692789555,
-      "learning_rate": 0.00019999385667877434,
-      "loss": 46.0,
-      "step": 21920
-    },
-    {
-      "epoch": 3.530214581907484,
-      "grad_norm": 0.002756329718977213,
-      "learning_rate": 0.00019999385611798944,
-      "loss": 46.0,
-      "step": 21921
-    },
-    {
-      "epoch": 3.5303756189862714,
-      "grad_norm": 0.001273614470846951,
-      "learning_rate": 0.00019999385555717898,
-      "loss": 46.0,
-      "step": 21922
-    },
-    {
-      "epoch": 3.530536656065059,
-      "grad_norm": 0.008705164305865765,
-      "learning_rate": 0.00019999385499634288,
-      "loss": 46.0,
-      "step": 21923
-    },
-    {
-      "epoch": 3.5306976931438463,
-      "grad_norm": 0.005385234951972961,
-      "learning_rate": 0.00019999385443548122,
-      "loss": 46.0,
-      "step": 21924
-    },
-    {
-      "epoch": 3.5308587302226337,
-      "grad_norm": 0.014770200476050377,
-      "learning_rate": 0.00019999385387459395,
-      "loss": 46.0,
-      "step": 21925
-    },
-    {
-      "epoch": 3.531019767301421,
-      "grad_norm": 0.0028039859607815742,
-      "learning_rate": 0.0001999938533136811,
-      "loss": 46.0,
-      "step": 21926
-    },
-    {
-      "epoch": 3.5311808043802086,
-      "grad_norm": 0.003740460379049182,
-      "learning_rate": 0.00019999385275274264,
-      "loss": 46.0,
-      "step": 21927
-    },
-    {
-      "epoch": 3.531341841458996,
-      "grad_norm": 0.008975562639534473,
-      "learning_rate": 0.00019999385219177858,
-      "loss": 46.0,
-      "step": 21928
-    },
-    {
-      "epoch": 3.5315028785377836,
-      "grad_norm": 0.0015580275794491172,
-      "learning_rate": 0.00019999385163078895,
-      "loss": 46.0,
-      "step": 21929
-    },
-    {
-      "epoch": 3.5316639156165706,
-      "grad_norm": 0.0022880041506141424,
-      "learning_rate": 0.00019999385106977372,
-      "loss": 46.0,
-      "step": 21930
-    },
-    {
-      "epoch": 3.531824952695358,
-      "grad_norm": 0.0011812779121100903,
-      "learning_rate": 0.0001999938505087329,
-      "loss": 46.0,
-      "step": 21931
-    },
-    {
-      "epoch": 3.5319859897741455,
-      "grad_norm": 0.0015835604863241315,
-      "learning_rate": 0.00019999384994766648,
-      "loss": 46.0,
-      "step": 21932
-    },
-    {
-      "epoch": 3.532147026852933,
-      "grad_norm": 0.004070988390594721,
-      "learning_rate": 0.00019999384938657446,
-      "loss": 46.0,
-      "step": 21933
-    },
-    {
-      "epoch": 3.5323080639317204,
-      "grad_norm": 0.01473455224186182,
-      "learning_rate": 0.00019999384882545687,
-      "loss": 46.0,
-      "step": 21934
-    },
-    {
-      "epoch": 3.532469101010508,
-      "grad_norm": 0.002096122596412897,
-      "learning_rate": 0.00019999384826431364,
-      "loss": 46.0,
-      "step": 21935
-    },
-    {
-      "epoch": 3.532630138089295,
-      "grad_norm": 0.0006884551839902997,
-      "learning_rate": 0.00019999384770314486,
-      "loss": 46.0,
-      "step": 21936
-    },
-    {
-      "epoch": 3.5327911751680823,
-      "grad_norm": 0.0013674602378159761,
-      "learning_rate": 0.00019999384714195045,
-      "loss": 46.0,
-      "step": 21937
-    },
-    {
-      "epoch": 3.5329522122468697,
-      "grad_norm": 0.012345785275101662,
-      "learning_rate": 0.0001999938465807305,
-      "loss": 46.0,
-      "step": 21938
-    },
-    {
-      "epoch": 3.533113249325657,
-      "grad_norm": 0.005417299922555685,
-      "learning_rate": 0.00019999384601948492,
-      "loss": 46.0,
-      "step": 21939
-    },
-    {
-      "epoch": 3.5332742864044446,
-      "grad_norm": 0.002332830335944891,
-      "learning_rate": 0.00019999384545821375,
-      "loss": 46.0,
-      "step": 21940
-    },
-    {
-      "epoch": 3.533435323483232,
-      "grad_norm": 0.0035465157125145197,
-      "learning_rate": 0.000199993844896917,
-      "loss": 46.0,
-      "step": 21941
-    },
-    {
-      "epoch": 3.5335963605620195,
-      "grad_norm": 0.003067759098485112,
-      "learning_rate": 0.00019999384433559464,
-      "loss": 46.0,
-      "step": 21942
-    },
-    {
-      "epoch": 3.533757397640807,
-      "grad_norm": 0.0008980606799013913,
-      "learning_rate": 0.00019999384377424672,
-      "loss": 46.0,
-      "step": 21943
-    },
-    {
-      "epoch": 3.5339184347195944,
-      "grad_norm": 0.0015847431495785713,
-      "learning_rate": 0.00019999384321287315,
-      "loss": 46.0,
-      "step": 21944
-    },
-    {
-      "epoch": 3.5340794717983814,
-      "grad_norm": 0.006590429227799177,
-      "learning_rate": 0.00019999384265147402,
-      "loss": 46.0,
-      "step": 21945
-    },
-    {
-      "epoch": 3.534240508877169,
-      "grad_norm": 0.0019933029543608427,
-      "learning_rate": 0.0001999938420900493,
-      "loss": 46.0,
-      "step": 21946
-    },
-    {
-      "epoch": 3.5344015459559563,
-      "grad_norm": 0.0012358695967122912,
-      "learning_rate": 0.00019999384152859896,
-      "loss": 46.0,
-      "step": 21947
-    },
-    {
-      "epoch": 3.534562583034744,
-      "grad_norm": 0.006818469613790512,
-      "learning_rate": 0.00019999384096712307,
-      "loss": 46.0,
-      "step": 21948
-    },
-    {
-      "epoch": 3.5347236201135313,
-      "grad_norm": 0.005868944805115461,
-      "learning_rate": 0.00019999384040562154,
-      "loss": 46.0,
-      "step": 21949
-    },
-    {
-      "epoch": 3.5348846571923183,
-      "grad_norm": 0.003276033792644739,
-      "learning_rate": 0.00019999383984409445,
-      "loss": 46.0,
-      "step": 21950
-    },
-    {
-      "epoch": 3.5350456942711057,
-      "grad_norm": 0.003228364046663046,
-      "learning_rate": 0.00019999383928254177,
-      "loss": 46.0,
-      "step": 21951
-    },
-    {
-      "epoch": 3.535206731349893,
-      "grad_norm": 0.0011344053782522678,
-      "learning_rate": 0.00019999383872096348,
-      "loss": 46.0,
-      "step": 21952
-    },
-    {
-      "epoch": 3.5353677684286806,
-      "grad_norm": 0.004558082204312086,
-      "learning_rate": 0.00019999383815935958,
-      "loss": 46.0,
-      "step": 21953
-    },
-    {
-      "epoch": 3.535528805507468,
-      "grad_norm": 0.004243944771587849,
-      "learning_rate": 0.00019999383759773011,
-      "loss": 46.0,
-      "step": 21954
-    },
-    {
-      "epoch": 3.5356898425862555,
-      "grad_norm": 0.008999931626021862,
-      "learning_rate": 0.00019999383703607506,
-      "loss": 46.0,
-      "step": 21955
-    },
-    {
-      "epoch": 3.535850879665043,
-      "grad_norm": 0.002882792381569743,
-      "learning_rate": 0.0001999938364743944,
-      "loss": 46.0,
-      "step": 21956
-    },
-    {
-      "epoch": 3.5360119167438304,
-      "grad_norm": 0.0042806644923985004,
-      "learning_rate": 0.00019999383591268814,
-      "loss": 46.0,
-      "step": 21957
-    },
-    {
-      "epoch": 3.536172953822618,
-      "grad_norm": 0.0050038970075547695,
-      "learning_rate": 0.0001999938353509563,
-      "loss": 46.0,
-      "step": 21958
-    },
-    {
-      "epoch": 3.5363339909014053,
-      "grad_norm": 0.0014869271544739604,
-      "learning_rate": 0.00019999383478919887,
-      "loss": 46.0,
-      "step": 21959
-    },
-    {
-      "epoch": 3.5364950279801923,
-      "grad_norm": 0.006352775730192661,
-      "learning_rate": 0.00019999383422741583,
-      "loss": 46.0,
-      "step": 21960
-    },
-    {
-      "epoch": 3.53665606505898,
-      "grad_norm": 0.011851463466882706,
-      "learning_rate": 0.0001999938336656072,
-      "loss": 46.0,
-      "step": 21961
-    },
-    {
-      "epoch": 3.5368171021377672,
-      "grad_norm": 0.0010939991334453225,
-      "learning_rate": 0.000199993833103773,
-      "loss": 46.0,
-      "step": 21962
-    },
-    {
-      "epoch": 3.5369781392165547,
-      "grad_norm": 0.0022448664531111717,
-      "learning_rate": 0.00019999383254191318,
-      "loss": 46.0,
-      "step": 21963
-    },
-    {
-      "epoch": 3.537139176295342,
-      "grad_norm": 0.0050621237605810165,
-      "learning_rate": 0.00019999383198002777,
-      "loss": 46.0,
-      "step": 21964
-    },
-    {
-      "epoch": 3.537300213374129,
-      "grad_norm": 0.002287766896188259,
-      "learning_rate": 0.00019999383141811676,
-      "loss": 46.0,
-      "step": 21965
-    },
-    {
-      "epoch": 3.5374612504529166,
-      "grad_norm": 0.004195712506771088,
-      "learning_rate": 0.00019999383085618017,
-      "loss": 46.0,
-      "step": 21966
-    },
-    {
-      "epoch": 3.537622287531704,
-      "grad_norm": 0.00612446665763855,
-      "learning_rate": 0.000199993830294218,
-      "loss": 46.0,
-      "step": 21967
-    },
-    {
-      "epoch": 3.5377833246104915,
-      "grad_norm": 0.004373784642666578,
-      "learning_rate": 0.00019999382973223022,
-      "loss": 46.0,
-      "step": 21968
-    },
-    {
-      "epoch": 3.537944361689279,
-      "grad_norm": 0.0027409158647060394,
-      "learning_rate": 0.00019999382917021684,
-      "loss": 46.0,
-      "step": 21969
-    },
-    {
-      "epoch": 3.5381053987680664,
-      "grad_norm": 0.006163121201097965,
-      "learning_rate": 0.00019999382860817788,
-      "loss": 46.0,
-      "step": 21970
-    },
-    {
-      "epoch": 3.538266435846854,
-      "grad_norm": 0.0014037862420082092,
-      "learning_rate": 0.00019999382804611332,
-      "loss": 46.0,
-      "step": 21971
-    },
-    {
-      "epoch": 3.5384274729256413,
-      "grad_norm": 0.00676178652793169,
-      "learning_rate": 0.00019999382748402318,
-      "loss": 46.0,
-      "step": 21972
-    },
-    {
-      "epoch": 3.5385885100044288,
-      "grad_norm": 0.002491881838068366,
-      "learning_rate": 0.00019999382692190742,
-      "loss": 46.0,
-      "step": 21973
-    },
-    {
-      "epoch": 3.5387495470832158,
-      "grad_norm": 0.004652262199670076,
-      "learning_rate": 0.00019999382635976608,
-      "loss": 46.0,
-      "step": 21974
-    },
-    {
-      "epoch": 3.538910584162003,
-      "grad_norm": 0.0011893954360857606,
-      "learning_rate": 0.00019999382579759915,
-      "loss": 46.0,
-      "step": 21975
-    },
-    {
-      "epoch": 3.5390716212407907,
-      "grad_norm": 0.0014562904834747314,
-      "learning_rate": 0.00019999382523540663,
-      "loss": 46.0,
-      "step": 21976
-    },
-    {
-      "epoch": 3.539232658319578,
-      "grad_norm": 0.006258485373109579,
-      "learning_rate": 0.0001999938246731885,
-      "loss": 46.0,
-      "step": 21977
-    },
-    {
-      "epoch": 3.5393936953983656,
-      "grad_norm": 0.006208564154803753,
-      "learning_rate": 0.00019999382411094478,
-      "loss": 46.0,
-      "step": 21978
-    },
-    {
-      "epoch": 3.539554732477153,
-      "grad_norm": 0.017545703798532486,
-      "learning_rate": 0.0001999938235486755,
-      "loss": 46.0,
-      "step": 21979
-    },
-    {
-      "epoch": 3.53971576955594,
-      "grad_norm": 0.0008426980930380523,
-      "learning_rate": 0.0001999938229863806,
-      "loss": 46.0,
-      "step": 21980
-    },
-    {
-      "epoch": 3.5398768066347275,
-      "grad_norm": 0.00999778974801302,
-      "learning_rate": 0.0001999938224240601,
-      "loss": 46.0,
-      "step": 21981
-    },
-    {
-      "epoch": 3.540037843713515,
-      "grad_norm": 0.004886379465460777,
-      "learning_rate": 0.00019999382186171403,
-      "loss": 46.0,
-      "step": 21982
-    },
-    {
-      "epoch": 3.5401988807923024,
-      "grad_norm": 0.0026424226816743612,
-      "learning_rate": 0.00019999382129934234,
-      "loss": 46.0,
-      "step": 21983
-    },
-    {
-      "epoch": 3.54035991787109,
-      "grad_norm": 0.010750465095043182,
-      "learning_rate": 0.00019999382073694508,
-      "loss": 46.0,
-      "step": 21984
-    },
-    {
-      "epoch": 3.5405209549498773,
-      "grad_norm": 0.014166489243507385,
-      "learning_rate": 0.00019999382017452222,
-      "loss": 46.0,
-      "step": 21985
-    },
-    {
-      "epoch": 3.5406819920286647,
-      "grad_norm": 0.0014444702537730336,
-      "learning_rate": 0.00019999381961207375,
-      "loss": 46.0,
-      "step": 21986
-    },
-    {
-      "epoch": 3.540843029107452,
-      "grad_norm": 0.002986549399793148,
-      "learning_rate": 0.00019999381904959972,
-      "loss": 46.0,
-      "step": 21987
-    },
-    {
-      "epoch": 3.5410040661862396,
-      "grad_norm": 0.0007509914576075971,
-      "learning_rate": 0.00019999381848710007,
-      "loss": 46.0,
-      "step": 21988
-    },
-    {
-      "epoch": 3.5411651032650266,
-      "grad_norm": 0.00562213035300374,
-      "learning_rate": 0.0001999938179245748,
-      "loss": 46.0,
-      "step": 21989
-    },
-    {
-      "epoch": 3.541326140343814,
-      "grad_norm": 0.009790195152163506,
-      "learning_rate": 0.000199993817362024,
-      "loss": 46.0,
-      "step": 21990
-    },
-    {
-      "epoch": 3.5414871774226016,
-      "grad_norm": 0.006605955772101879,
-      "learning_rate": 0.00019999381679944756,
-      "loss": 46.0,
-      "step": 21991
-    },
-    {
-      "epoch": 3.541648214501389,
-      "grad_norm": 0.003194431308656931,
-      "learning_rate": 0.00019999381623684554,
-      "loss": 46.0,
-      "step": 21992
-    },
-    {
-      "epoch": 3.5418092515801765,
-      "grad_norm": 0.0016847802326083183,
-      "learning_rate": 0.00019999381567421793,
-      "loss": 46.0,
-      "step": 21993
-    },
-    {
-      "epoch": 3.5419702886589635,
-      "grad_norm": 0.001483723521232605,
-      "learning_rate": 0.00019999381511156473,
-      "loss": 46.0,
-      "step": 21994
-    },
-    {
-      "epoch": 3.542131325737751,
-      "grad_norm": 0.005787108093500137,
-      "learning_rate": 0.00019999381454888592,
-      "loss": 46.0,
-      "step": 21995
-    },
-    {
-      "epoch": 3.5422923628165384,
-      "grad_norm": 0.0010338110150769353,
-      "learning_rate": 0.00019999381398618153,
-      "loss": 46.0,
-      "step": 21996
-    },
-    {
-      "epoch": 3.542453399895326,
-      "grad_norm": 0.006773020140826702,
-      "learning_rate": 0.00019999381342345157,
-      "loss": 46.0,
-      "step": 21997
-    },
-    {
-      "epoch": 3.5426144369741133,
-      "grad_norm": 0.0016948096454143524,
-      "learning_rate": 0.00019999381286069597,
-      "loss": 46.0,
-      "step": 21998
-    },
-    {
-      "epoch": 3.5427754740529007,
-      "grad_norm": 0.00602155365049839,
-      "learning_rate": 0.0001999938122979148,
-      "loss": 46.0,
-      "step": 21999
-    },
-    {
-      "epoch": 3.542936511131688,
-      "grad_norm": 0.0042361048981547356,
-      "learning_rate": 0.00019999381173510804,
-      "loss": 46.0,
-      "step": 22000
-    },
-    {
-      "epoch": 3.5430975482104756,
-      "grad_norm": 0.0024669792037457228,
-      "learning_rate": 0.00019999381117227568,
-      "loss": 46.0,
-      "step": 22001
-    },
-    {
-      "epoch": 3.543258585289263,
-      "grad_norm": 0.005744407419115305,
-      "learning_rate": 0.00019999381060941773,
-      "loss": 46.0,
-      "step": 22002
-    },
-    {
-      "epoch": 3.54341962236805,
-      "grad_norm": 0.0015349757159128785,
-      "learning_rate": 0.0001999938100465342,
-      "loss": 46.0,
-      "step": 22003
-    },
-    {
-      "epoch": 3.5435806594468375,
-      "grad_norm": 0.008937159553170204,
-      "learning_rate": 0.00019999380948362505,
-      "loss": 46.0,
-      "step": 22004
-    },
-    {
-      "epoch": 3.543741696525625,
-      "grad_norm": 0.0010599809465929866,
-      "learning_rate": 0.00019999380892069034,
-      "loss": 46.0,
-      "step": 22005
-    },
-    {
-      "epoch": 3.5439027336044124,
-      "grad_norm": 0.0013040232006460428,
-      "learning_rate": 0.00019999380835773002,
-      "loss": 46.0,
-      "step": 22006
-    },
-    {
-      "epoch": 3.5440637706832,
-      "grad_norm": 0.001095239887945354,
-      "learning_rate": 0.00019999380779474408,
-      "loss": 46.0,
-      "step": 22007
-    },
-    {
-      "epoch": 3.5442248077619873,
-      "grad_norm": 0.0011410439619794488,
-      "learning_rate": 0.00019999380723173258,
-      "loss": 46.0,
-      "step": 22008
-    },
-    {
-      "epoch": 3.5443858448407743,
-      "grad_norm": 0.004196349997073412,
-      "learning_rate": 0.00019999380666869547,
-      "loss": 46.0,
-      "step": 22009
-    },
-    {
-      "epoch": 3.544546881919562,
-      "grad_norm": 0.0010497071780264378,
-      "learning_rate": 0.00019999380610563277,
-      "loss": 46.0,
-      "step": 22010
-    },
-    {
-      "epoch": 3.5447079189983492,
-      "grad_norm": 0.001308976672589779,
-      "learning_rate": 0.0001999938055425445,
-      "loss": 46.0,
-      "step": 22011
-    },
-    {
-      "epoch": 3.5448689560771367,
-      "grad_norm": 0.0029437458142638206,
-      "learning_rate": 0.0001999938049794306,
-      "loss": 46.0,
-      "step": 22012
-    },
-    {
-      "epoch": 3.545029993155924,
-      "grad_norm": 0.007164867129176855,
-      "learning_rate": 0.00019999380441629113,
-      "loss": 46.0,
-      "step": 22013
-    },
-    {
-      "epoch": 3.5451910302347116,
-      "grad_norm": 0.0010761006269603968,
-      "learning_rate": 0.00019999380385312608,
-      "loss": 46.0,
-      "step": 22014
-    },
-    {
-      "epoch": 3.545352067313499,
-      "grad_norm": 0.003038262017071247,
-      "learning_rate": 0.0001999938032899354,
-      "loss": 46.0,
-      "step": 22015
-    },
-    {
-      "epoch": 3.5455131043922865,
-      "grad_norm": 0.004080723039805889,
-      "learning_rate": 0.00019999380272671914,
-      "loss": 46.0,
-      "step": 22016
-    },
-    {
-      "epoch": 3.545674141471074,
-      "grad_norm": 0.007721003144979477,
-      "learning_rate": 0.0001999938021634773,
-      "loss": 46.0,
-      "step": 22017
-    },
-    {
-      "epoch": 3.545835178549861,
-      "grad_norm": 0.00481992494314909,
-      "learning_rate": 0.00019999380160020983,
-      "loss": 46.0,
-      "step": 22018
-    },
-    {
-      "epoch": 3.5459962156286484,
-      "grad_norm": 0.0033730817958712578,
-      "learning_rate": 0.00019999380103691682,
-      "loss": 46.0,
-      "step": 22019
-    },
-    {
-      "epoch": 3.546157252707436,
-      "grad_norm": 0.00402312446385622,
-      "learning_rate": 0.00019999380047359817,
-      "loss": 46.0,
-      "step": 22020
-    },
-    {
-      "epoch": 3.5463182897862233,
-      "grad_norm": 0.0019284755690023303,
-      "learning_rate": 0.00019999379991025395,
-      "loss": 46.0,
-      "step": 22021
-    },
-    {
-      "epoch": 3.5464793268650108,
-      "grad_norm": 0.004927211441099644,
-      "learning_rate": 0.00019999379934688413,
-      "loss": 46.0,
-      "step": 22022
-    },
-    {
-      "epoch": 3.5466403639437982,
-      "grad_norm": 0.005154408048838377,
-      "learning_rate": 0.00019999379878348874,
-      "loss": 46.0,
-      "step": 22023
-    },
-    {
-      "epoch": 3.5468014010225852,
-      "grad_norm": 0.001954978331923485,
-      "learning_rate": 0.00019999379822006774,
-      "loss": 46.0,
-      "step": 22024
-    },
-    {
-      "epoch": 3.5469624381013727,
-      "grad_norm": 0.0043811374343931675,
-      "learning_rate": 0.00019999379765662113,
-      "loss": 46.0,
-      "step": 22025
-    },
-    {
-      "epoch": 3.54712347518016,
-      "grad_norm": 0.0025637864600867033,
-      "learning_rate": 0.00019999379709314893,
-      "loss": 46.0,
-      "step": 22026
-    },
-    {
-      "epoch": 3.5472845122589476,
-      "grad_norm": 0.002644615015015006,
-      "learning_rate": 0.00019999379652965116,
-      "loss": 46.0,
-      "step": 22027
-    },
-    {
-      "epoch": 3.547445549337735,
-      "grad_norm": 0.022094856947660446,
-      "learning_rate": 0.00019999379596612776,
-      "loss": 46.0,
-      "step": 22028
-    },
-    {
-      "epoch": 3.5476065864165225,
-      "grad_norm": 0.0016870190156623721,
-      "learning_rate": 0.0001999937954025788,
-      "loss": 46.0,
-      "step": 22029
-    },
-    {
-      "epoch": 3.54776762349531,
-      "grad_norm": 0.0030062077566981316,
-      "learning_rate": 0.00019999379483900427,
-      "loss": 46.0,
-      "step": 22030
-    },
-    {
-      "epoch": 3.5479286605740974,
-      "grad_norm": 0.0015344704734161496,
-      "learning_rate": 0.0001999937942754041,
-      "loss": 46.0,
-      "step": 22031
-    },
-    {
-      "epoch": 3.548089697652885,
-      "grad_norm": 0.0033449300099164248,
-      "learning_rate": 0.00019999379371177836,
-      "loss": 46.0,
-      "step": 22032
-    },
-    {
-      "epoch": 3.548250734731672,
-      "grad_norm": 0.005942798685282469,
-      "learning_rate": 0.00019999379314812702,
-      "loss": 46.0,
-      "step": 22033
-    },
-    {
-      "epoch": 3.5484117718104593,
-      "grad_norm": 0.0012079703155905008,
-      "learning_rate": 0.0001999937925844501,
-      "loss": 46.0,
-      "step": 22034
-    },
-    {
-      "epoch": 3.5485728088892468,
-      "grad_norm": 0.007327585481107235,
-      "learning_rate": 0.00019999379202074752,
-      "loss": 46.0,
-      "step": 22035
-    },
-    {
-      "epoch": 3.548733845968034,
-      "grad_norm": 0.0015873860102146864,
-      "learning_rate": 0.00019999379145701942,
-      "loss": 46.0,
-      "step": 22036
-    },
-    {
-      "epoch": 3.5488948830468217,
-      "grad_norm": 0.00800500251352787,
-      "learning_rate": 0.0001999937908932657,
-      "loss": 46.0,
-      "step": 22037
-    },
-    {
-      "epoch": 3.5490559201256087,
-      "grad_norm": 0.010370759293437004,
-      "learning_rate": 0.0001999937903294864,
-      "loss": 46.0,
-      "step": 22038
-    },
-    {
-      "epoch": 3.549216957204396,
-      "grad_norm": 0.01294117234647274,
-      "learning_rate": 0.0001999937897656815,
-      "loss": 46.0,
-      "step": 22039
-    },
-    {
-      "epoch": 3.5493779942831836,
-      "grad_norm": 0.006117288023233414,
-      "learning_rate": 0.00019999378920185099,
-      "loss": 46.0,
-      "step": 22040
-    },
-    {
-      "epoch": 3.549539031361971,
-      "grad_norm": 0.006722374353557825,
-      "learning_rate": 0.00019999378863799492,
-      "loss": 46.0,
-      "step": 22041
-    },
-    {
-      "epoch": 3.5497000684407585,
-      "grad_norm": 0.003587601939216256,
-      "learning_rate": 0.00019999378807411322,
-      "loss": 46.0,
-      "step": 22042
-    },
-    {
-      "epoch": 3.549861105519546,
-      "grad_norm": 0.0036452380008995533,
-      "learning_rate": 0.00019999378751020595,
-      "loss": 46.0,
-      "step": 22043
-    },
-    {
-      "epoch": 3.5500221425983334,
-      "grad_norm": 0.006890326738357544,
-      "learning_rate": 0.0001999937869462731,
-      "loss": 46.0,
-      "step": 22044
-    },
-    {
-      "epoch": 3.550183179677121,
-      "grad_norm": 0.003870619460940361,
-      "learning_rate": 0.00019999378638231463,
-      "loss": 46.0,
-      "step": 22045
-    },
-    {
-      "epoch": 3.5503442167559083,
-      "grad_norm": 0.015682950615882874,
-      "learning_rate": 0.00019999378581833058,
-      "loss": 46.0,
-      "step": 22046
-    },
-    {
-      "epoch": 3.5505052538346953,
-      "grad_norm": 0.0009960850002244115,
-      "learning_rate": 0.0001999937852543209,
-      "loss": 46.0,
-      "step": 22047
-    },
-    {
-      "epoch": 3.5506662909134827,
-      "grad_norm": 0.002020682441070676,
-      "learning_rate": 0.00019999378469028568,
-      "loss": 46.0,
-      "step": 22048
-    },
-    {
-      "epoch": 3.55082732799227,
-      "grad_norm": 0.00999528355896473,
-      "learning_rate": 0.00019999378412622484,
-      "loss": 46.0,
-      "step": 22049
-    },
-    {
-      "epoch": 3.5509883650710576,
-      "grad_norm": 0.0014531823107972741,
-      "learning_rate": 0.0001999937835621384,
-      "loss": 46.0,
-      "step": 22050
-    },
-    {
-      "epoch": 3.551149402149845,
-      "grad_norm": 0.0036034679505974054,
-      "learning_rate": 0.0001999937829980264,
-      "loss": 46.0,
-      "step": 22051
-    },
-    {
-      "epoch": 3.5513104392286325,
-      "grad_norm": 0.002651785733178258,
-      "learning_rate": 0.0001999937824338888,
-      "loss": 46.0,
-      "step": 22052
-    },
-    {
-      "epoch": 3.5514714763074196,
-      "grad_norm": 0.0015067094936966896,
-      "learning_rate": 0.00019999378186972557,
-      "loss": 46.0,
-      "step": 22053
-    },
-    {
-      "epoch": 3.551632513386207,
-      "grad_norm": 0.0019392030080780387,
-      "learning_rate": 0.00019999378130553677,
-      "loss": 46.0,
-      "step": 22054
-    },
-    {
-      "epoch": 3.5517935504649945,
-      "grad_norm": 0.0034423451870679855,
-      "learning_rate": 0.00019999378074132238,
-      "loss": 46.0,
-      "step": 22055
-    },
-    {
-      "epoch": 3.551954587543782,
-      "grad_norm": 0.002157733077183366,
-      "learning_rate": 0.0001999937801770824,
-      "loss": 46.0,
-      "step": 22056
-    },
-    {
-      "epoch": 3.5521156246225694,
-      "grad_norm": 0.008837435394525528,
-      "learning_rate": 0.0001999937796128168,
-      "loss": 46.0,
-      "step": 22057
-    },
-    {
-      "epoch": 3.552276661701357,
-      "grad_norm": 0.003991037141531706,
-      "learning_rate": 0.00019999377904852565,
-      "loss": 46.0,
-      "step": 22058
-    },
-    {
-      "epoch": 3.5524376987801443,
-      "grad_norm": 0.0008896852377802134,
-      "learning_rate": 0.00019999377848420888,
-      "loss": 46.0,
-      "step": 22059
-    },
-    {
-      "epoch": 3.5525987358589317,
-      "grad_norm": 0.0017367544351145625,
-      "learning_rate": 0.0001999937779198665,
-      "loss": 46.0,
-      "step": 22060
-    },
-    {
-      "epoch": 3.552759772937719,
-      "grad_norm": 0.0024984185583889484,
-      "learning_rate": 0.00019999377735549856,
-      "loss": 46.0,
-      "step": 22061
-    },
-    {
-      "epoch": 3.552920810016506,
-      "grad_norm": 0.002191744511947036,
-      "learning_rate": 0.000199993776791105,
-      "loss": 46.0,
-      "step": 22062
-    },
-    {
-      "epoch": 3.5530818470952936,
-      "grad_norm": 0.004819722380489111,
-      "learning_rate": 0.00019999377622668588,
-      "loss": 46.0,
-      "step": 22063
-    },
-    {
-      "epoch": 3.553242884174081,
-      "grad_norm": 0.00223218509927392,
-      "learning_rate": 0.00019999377566224112,
-      "loss": 46.0,
-      "step": 22064
-    },
-    {
-      "epoch": 3.5534039212528685,
-      "grad_norm": 0.0016874123830348253,
-      "learning_rate": 0.00019999377509777083,
-      "loss": 46.0,
-      "step": 22065
-    },
-    {
-      "epoch": 3.553564958331656,
-      "grad_norm": 0.0043662418611347675,
-      "learning_rate": 0.0001999937745332749,
-      "loss": 46.0,
-      "step": 22066
-    },
-    {
-      "epoch": 3.553725995410443,
-      "grad_norm": 0.008035285398364067,
-      "learning_rate": 0.00019999377396875335,
-      "loss": 46.0,
-      "step": 22067
-    },
-    {
-      "epoch": 3.5538870324892304,
-      "grad_norm": 0.000836613355204463,
-      "learning_rate": 0.00019999377340420627,
-      "loss": 46.0,
-      "step": 22068
-    },
-    {
-      "epoch": 3.554048069568018,
-      "grad_norm": 0.004797850735485554,
-      "learning_rate": 0.00019999377283963355,
-      "loss": 46.0,
-      "step": 22069
-    },
-    {
-      "epoch": 3.5542091066468053,
-      "grad_norm": 0.002900391351431608,
-      "learning_rate": 0.00019999377227503527,
-      "loss": 46.0,
-      "step": 22070
-    },
-    {
-      "epoch": 3.554370143725593,
-      "grad_norm": 0.0030640731565654278,
-      "learning_rate": 0.00019999377171041137,
-      "loss": 46.0,
-      "step": 22071
-    },
-    {
-      "epoch": 3.5545311808043802,
-      "grad_norm": 0.012759199365973473,
-      "learning_rate": 0.00019999377114576192,
-      "loss": 46.0,
-      "step": 22072
-    },
-    {
-      "epoch": 3.5546922178831677,
-      "grad_norm": 0.00236497912555933,
-      "learning_rate": 0.00019999377058108685,
-      "loss": 46.0,
-      "step": 22073
-    },
-    {
-      "epoch": 3.554853254961955,
-      "grad_norm": 0.006937025114893913,
-      "learning_rate": 0.00019999377001638616,
-      "loss": 46.0,
-      "step": 22074
-    },
-    {
-      "epoch": 3.5550142920407426,
-      "grad_norm": 0.005155367311090231,
-      "learning_rate": 0.00019999376945165992,
-      "loss": 46.0,
-      "step": 22075
-    },
-    {
-      "epoch": 3.55517532911953,
-      "grad_norm": 0.004127301275730133,
-      "learning_rate": 0.00019999376888690806,
-      "loss": 46.0,
-      "step": 22076
-    },
-    {
-      "epoch": 3.555336366198317,
-      "grad_norm": 0.007588563486933708,
-      "learning_rate": 0.0001999937683221306,
-      "loss": 46.0,
-      "step": 22077
-    },
-    {
-      "epoch": 3.5554974032771045,
-      "grad_norm": 0.00534937996417284,
-      "learning_rate": 0.00019999376775732756,
-      "loss": 46.0,
-      "step": 22078
-    },
-    {
-      "epoch": 3.555658440355892,
-      "grad_norm": 0.006973283365368843,
-      "learning_rate": 0.00019999376719249894,
-      "loss": 46.0,
-      "step": 22079
-    },
-    {
-      "epoch": 3.5558194774346794,
-      "grad_norm": 0.0032623731531202793,
-      "learning_rate": 0.0001999937666276447,
-      "loss": 46.0,
-      "step": 22080
-    },
-    {
-      "epoch": 3.555980514513467,
-      "grad_norm": 0.006116578821092844,
-      "learning_rate": 0.00019999376606276488,
-      "loss": 46.0,
-      "step": 22081
-    },
-    {
-      "epoch": 3.556141551592254,
-      "grad_norm": 0.007953274063766003,
-      "learning_rate": 0.00019999376549785947,
-      "loss": 46.0,
-      "step": 22082
-    },
-    {
-      "epoch": 3.5563025886710413,
-      "grad_norm": 0.005206270609050989,
-      "learning_rate": 0.00019999376493292848,
-      "loss": 46.0,
-      "step": 22083
-    },
-    {
-      "epoch": 3.5564636257498288,
-      "grad_norm": 0.004771303851157427,
-      "learning_rate": 0.00019999376436797187,
-      "loss": 46.0,
-      "step": 22084
-    },
-    {
-      "epoch": 3.5566246628286162,
-      "grad_norm": 0.0026180166751146317,
-      "learning_rate": 0.00019999376380298967,
-      "loss": 46.0,
-      "step": 22085
-    },
-    {
-      "epoch": 3.5567856999074037,
-      "grad_norm": 0.0017999495612457395,
-      "learning_rate": 0.00019999376323798188,
-      "loss": 46.0,
-      "step": 22086
-    },
-    {
-      "epoch": 3.556946736986191,
-      "grad_norm": 0.0015788378659635782,
-      "learning_rate": 0.0001999937626729485,
-      "loss": 46.0,
-      "step": 22087
-    },
-    {
-      "epoch": 3.5571077740649786,
-      "grad_norm": 0.00141175277531147,
-      "learning_rate": 0.00019999376210788953,
-      "loss": 46.0,
-      "step": 22088
-    },
-    {
-      "epoch": 3.557268811143766,
-      "grad_norm": 0.00064327463041991,
-      "learning_rate": 0.00019999376154280498,
-      "loss": 46.0,
-      "step": 22089
-    },
-    {
-      "epoch": 3.5574298482225535,
-      "grad_norm": 0.0036487989127635956,
-      "learning_rate": 0.00019999376097769482,
-      "loss": 46.0,
-      "step": 22090
-    },
-    {
-      "epoch": 3.5575908853013405,
-      "grad_norm": 0.00048203894402831793,
-      "learning_rate": 0.00019999376041255907,
-      "loss": 46.0,
-      "step": 22091
-    },
-    {
-      "epoch": 3.557751922380128,
-      "grad_norm": 0.0007147903670556843,
-      "learning_rate": 0.0001999937598473977,
-      "loss": 46.0,
-      "step": 22092
-    },
-    {
-      "epoch": 3.5579129594589154,
-      "grad_norm": 0.003537654411047697,
-      "learning_rate": 0.0001999937592822108,
-      "loss": 46.0,
-      "step": 22093
-    },
-    {
-      "epoch": 3.558073996537703,
-      "grad_norm": 0.01155766285955906,
-      "learning_rate": 0.00019999375871699825,
-      "loss": 46.0,
-      "step": 22094
-    },
-    {
-      "epoch": 3.5582350336164903,
-      "grad_norm": 0.0012491022935137153,
-      "learning_rate": 0.00019999375815176013,
-      "loss": 46.0,
-      "step": 22095
-    },
-    {
-      "epoch": 3.5583960706952777,
-      "grad_norm": 0.0009142942726612091,
-      "learning_rate": 0.00019999375758649642,
-      "loss": 46.0,
-      "step": 22096
-    },
-    {
-      "epoch": 3.5585571077740648,
-      "grad_norm": 0.000657429569400847,
-      "learning_rate": 0.0001999937570212071,
-      "loss": 46.0,
-      "step": 22097
-    },
-    {
-      "epoch": 3.558718144852852,
-      "grad_norm": 0.0018619636539369822,
-      "learning_rate": 0.0001999937564558922,
-      "loss": 46.0,
-      "step": 22098
-    },
-    {
-      "epoch": 3.5588791819316397,
-      "grad_norm": 0.0061980425380170345,
-      "learning_rate": 0.00019999375589055168,
-      "loss": 46.0,
-      "step": 22099
-    },
-    {
-      "epoch": 3.559040219010427,
-      "grad_norm": 0.001235340372659266,
-      "learning_rate": 0.0001999937553251856,
-      "loss": 46.0,
-      "step": 22100
-    },
-    {
-      "epoch": 3.5592012560892146,
-      "grad_norm": 0.002880545100197196,
-      "learning_rate": 0.00019999375475979392,
-      "loss": 46.0,
-      "step": 22101
-    },
-    {
-      "epoch": 3.559362293168002,
-      "grad_norm": 0.0030691258143633604,
-      "learning_rate": 0.00019999375419437664,
-      "loss": 46.0,
-      "step": 22102
-    },
-    {
-      "epoch": 3.5595233302467895,
-      "grad_norm": 0.007847237400710583,
-      "learning_rate": 0.00019999375362893376,
-      "loss": 46.0,
-      "step": 22103
-    },
-    {
-      "epoch": 3.559684367325577,
-      "grad_norm": 0.0024644576478749514,
-      "learning_rate": 0.0001999937530634653,
-      "loss": 46.0,
-      "step": 22104
-    },
-    {
-      "epoch": 3.5598454044043644,
-      "grad_norm": 0.005099670495837927,
-      "learning_rate": 0.00019999375249797122,
-      "loss": 46.0,
-      "step": 22105
-    },
-    {
-      "epoch": 3.5600064414831514,
-      "grad_norm": 0.0015072562964633107,
-      "learning_rate": 0.00019999375193245161,
-      "loss": 46.0,
-      "step": 22106
-    },
-    {
-      "epoch": 3.560167478561939,
-      "grad_norm": 0.010963588021695614,
-      "learning_rate": 0.00019999375136690634,
-      "loss": 46.0,
-      "step": 22107
-    },
-    {
-      "epoch": 3.5603285156407263,
-      "grad_norm": 0.0012547129299491644,
-      "learning_rate": 0.0001999937508013355,
-      "loss": 46.0,
-      "step": 22108
-    },
-    {
-      "epoch": 3.5604895527195137,
-      "grad_norm": 0.006265297997742891,
-      "learning_rate": 0.00019999375023573907,
-      "loss": 46.0,
-      "step": 22109
-    },
-    {
-      "epoch": 3.560650589798301,
-      "grad_norm": 0.0017919526435434818,
-      "learning_rate": 0.00019999374967011706,
-      "loss": 46.0,
-      "step": 22110
-    },
-    {
-      "epoch": 3.560811626877088,
-      "grad_norm": 0.003235555486753583,
-      "learning_rate": 0.00019999374910446944,
-      "loss": 46.0,
-      "step": 22111
-    },
-    {
-      "epoch": 3.5609726639558756,
-      "grad_norm": 0.0021480501163750887,
-      "learning_rate": 0.00019999374853879622,
-      "loss": 46.0,
-      "step": 22112
-    },
-    {
-      "epoch": 3.561133701034663,
-      "grad_norm": 0.0016245265724137425,
-      "learning_rate": 0.00019999374797309742,
-      "loss": 46.0,
-      "step": 22113
-    },
-    {
-      "epoch": 3.5612947381134505,
-      "grad_norm": 0.0018488435307517648,
-      "learning_rate": 0.00019999374740737303,
-      "loss": 46.0,
-      "step": 22114
-    },
-    {
-      "epoch": 3.561455775192238,
-      "grad_norm": 0.00484824925661087,
-      "learning_rate": 0.00019999374684162303,
-      "loss": 46.0,
-      "step": 22115
-    },
-    {
-      "epoch": 3.5616168122710254,
-      "grad_norm": 0.0064271786250174046,
-      "learning_rate": 0.00019999374627584744,
-      "loss": 46.0,
-      "step": 22116
-    },
-    {
-      "epoch": 3.561777849349813,
-      "grad_norm": 0.0033978682477027178,
-      "learning_rate": 0.00019999374571004626,
-      "loss": 46.0,
-      "step": 22117
-    },
-    {
-      "epoch": 3.5619388864286003,
-      "grad_norm": 0.00105980911757797,
-      "learning_rate": 0.0001999937451442195,
-      "loss": 46.0,
-      "step": 22118
-    },
-    {
-      "epoch": 3.562099923507388,
-      "grad_norm": 0.0012255714973434806,
-      "learning_rate": 0.00019999374457836715,
-      "loss": 46.0,
-      "step": 22119
-    },
-    {
-      "epoch": 3.562260960586175,
-      "grad_norm": 0.0011637681163847446,
-      "learning_rate": 0.00019999374401248919,
-      "loss": 46.0,
-      "step": 22120
-    },
-    {
-      "epoch": 3.5624219976649623,
-      "grad_norm": 0.0009744488052092493,
-      "learning_rate": 0.00019999374344658563,
-      "loss": 46.0,
-      "step": 22121
-    },
-    {
-      "epoch": 3.5625830347437497,
-      "grad_norm": 0.012898662127554417,
-      "learning_rate": 0.0001999937428806565,
-      "loss": 46.0,
-      "step": 22122
-    },
-    {
-      "epoch": 3.562744071822537,
-      "grad_norm": 0.0022565291728824377,
-      "learning_rate": 0.00019999374231470177,
-      "loss": 46.0,
-      "step": 22123
-    },
-    {
-      "epoch": 3.5629051089013246,
-      "grad_norm": 0.002504949225112796,
-      "learning_rate": 0.00019999374174872143,
-      "loss": 46.0,
-      "step": 22124
-    },
-    {
-      "epoch": 3.563066145980112,
-      "grad_norm": 0.0019123753299936652,
-      "learning_rate": 0.0001999937411827155,
-      "loss": 46.0,
-      "step": 22125
-    },
-    {
-      "epoch": 3.563227183058899,
-      "grad_norm": 0.0017348452238366008,
-      "learning_rate": 0.00019999374061668398,
-      "loss": 46.0,
-      "step": 22126
-    },
-    {
-      "epoch": 3.5633882201376865,
-      "grad_norm": 0.0009706579148769379,
-      "learning_rate": 0.00019999374005062688,
-      "loss": 46.0,
-      "step": 22127
-    },
-    {
-      "epoch": 3.563549257216474,
-      "grad_norm": 0.010791081003844738,
-      "learning_rate": 0.00019999373948454416,
-      "loss": 46.0,
-      "step": 22128
-    },
-    {
-      "epoch": 3.5637102942952614,
-      "grad_norm": 0.0021594117861241102,
-      "learning_rate": 0.00019999373891843586,
-      "loss": 46.0,
-      "step": 22129
-    },
-    {
-      "epoch": 3.563871331374049,
-      "grad_norm": 0.007558625657111406,
-      "learning_rate": 0.00019999373835230197,
-      "loss": 46.0,
-      "step": 22130
-    },
-    {
-      "epoch": 3.5640323684528363,
-      "grad_norm": 0.008450819179415703,
-      "learning_rate": 0.0001999937377861425,
-      "loss": 46.0,
-      "step": 22131
-    },
-    {
-      "epoch": 3.564193405531624,
-      "grad_norm": 0.00552174961194396,
-      "learning_rate": 0.00019999373721995742,
-      "loss": 46.0,
-      "step": 22132
-    },
-    {
-      "epoch": 3.5643544426104112,
-      "grad_norm": 0.004941765684634447,
-      "learning_rate": 0.00019999373665374677,
-      "loss": 46.0,
-      "step": 22133
-    },
-    {
-      "epoch": 3.5645154796891987,
-      "grad_norm": 0.004483547993004322,
-      "learning_rate": 0.00019999373608751048,
-      "loss": 46.0,
-      "step": 22134
-    },
-    {
-      "epoch": 3.5646765167679857,
-      "grad_norm": 0.00358788575977087,
-      "learning_rate": 0.00019999373552124862,
-      "loss": 46.0,
-      "step": 22135
-    },
-    {
-      "epoch": 3.564837553846773,
-      "grad_norm": 0.0013157286448404193,
-      "learning_rate": 0.00019999373495496118,
-      "loss": 46.0,
-      "step": 22136
-    },
-    {
-      "epoch": 3.5649985909255606,
-      "grad_norm": 0.007234329357743263,
-      "learning_rate": 0.00019999373438864815,
-      "loss": 46.0,
-      "step": 22137
-    },
-    {
-      "epoch": 3.565159628004348,
-      "grad_norm": 0.0037574535235762596,
-      "learning_rate": 0.0001999937338223095,
-      "loss": 46.0,
-      "step": 22138
-    },
-    {
-      "epoch": 3.5653206650831355,
-      "grad_norm": 0.0013072319561615586,
-      "learning_rate": 0.00019999373325594528,
-      "loss": 46.0,
-      "step": 22139
-    },
-    {
-      "epoch": 3.5654817021619225,
-      "grad_norm": 0.005404674913734198,
-      "learning_rate": 0.00019999373268955547,
-      "loss": 46.0,
-      "step": 22140
-    },
-    {
-      "epoch": 3.56564273924071,
-      "grad_norm": 0.003842002945020795,
-      "learning_rate": 0.00019999373212314003,
-      "loss": 46.0,
-      "step": 22141
-    },
-    {
-      "epoch": 3.5658037763194974,
-      "grad_norm": 0.004123489372432232,
-      "learning_rate": 0.00019999373155669902,
-      "loss": 46.0,
-      "step": 22142
-    },
-    {
-      "epoch": 3.565964813398285,
-      "grad_norm": 0.004591065458953381,
-      "learning_rate": 0.00019999373099023244,
-      "loss": 46.0,
-      "step": 22143
-    },
-    {
-      "epoch": 3.5661258504770723,
-      "grad_norm": 0.01184425875544548,
-      "learning_rate": 0.00019999373042374024,
-      "loss": 46.0,
-      "step": 22144
-    },
-    {
-      "epoch": 3.5662868875558598,
-      "grad_norm": 0.005107114091515541,
-      "learning_rate": 0.00019999372985722244,
-      "loss": 46.0,
-      "step": 22145
-    },
-    {
-      "epoch": 3.566447924634647,
-      "grad_norm": 0.0010000468464568257,
-      "learning_rate": 0.00019999372929067904,
-      "loss": 46.0,
-      "step": 22146
-    },
-    {
-      "epoch": 3.5666089617134347,
-      "grad_norm": 0.003872596425935626,
-      "learning_rate": 0.0001999937287241101,
-      "loss": 46.0,
-      "step": 22147
-    },
-    {
-      "epoch": 3.566769998792222,
-      "grad_norm": 0.0057190777733922005,
-      "learning_rate": 0.00019999372815751552,
-      "loss": 46.0,
-      "step": 22148
-    },
-    {
-      "epoch": 3.5669310358710096,
-      "grad_norm": 0.007215212564915419,
-      "learning_rate": 0.00019999372759089536,
-      "loss": 46.0,
-      "step": 22149
-    },
-    {
-      "epoch": 3.5670920729497966,
-      "grad_norm": 0.0023381332866847515,
-      "learning_rate": 0.00019999372702424962,
-      "loss": 46.0,
-      "step": 22150
-    },
-    {
-      "epoch": 3.567253110028584,
-      "grad_norm": 0.006481852848082781,
-      "learning_rate": 0.00019999372645757826,
-      "loss": 46.0,
-      "step": 22151
-    },
-    {
-      "epoch": 3.5674141471073715,
-      "grad_norm": 0.0016577077331021428,
-      "learning_rate": 0.00019999372589088132,
-      "loss": 46.0,
-      "step": 22152
-    },
-    {
-      "epoch": 3.567575184186159,
-      "grad_norm": 0.004699667915701866,
-      "learning_rate": 0.0001999937253241588,
-      "loss": 46.0,
-      "step": 22153
-    },
-    {
-      "epoch": 3.5677362212649464,
-      "grad_norm": 0.0026436829939484596,
-      "learning_rate": 0.00019999372475741064,
-      "loss": 46.0,
-      "step": 22154
-    },
-    {
-      "epoch": 3.5678972583437334,
-      "grad_norm": 0.0045671770349144936,
-      "learning_rate": 0.00019999372419063694,
-      "loss": 46.0,
-      "step": 22155
-    },
-    {
-      "epoch": 3.568058295422521,
-      "grad_norm": 0.0021484759636223316,
-      "learning_rate": 0.00019999372362383762,
-      "loss": 46.0,
-      "step": 22156
-    },
-    {
-      "epoch": 3.5682193325013083,
-      "grad_norm": 0.0014306337106972933,
-      "learning_rate": 0.0001999937230570127,
-      "loss": 46.0,
-      "step": 22157
-    },
-    {
-      "epoch": 3.5683803695800957,
-      "grad_norm": 0.00875294953584671,
-      "learning_rate": 0.0001999937224901622,
-      "loss": 46.0,
-      "step": 22158
-    },
-    {
-      "epoch": 3.568541406658883,
-      "grad_norm": 0.0014109316980466247,
-      "learning_rate": 0.0001999937219232861,
-      "loss": 46.0,
-      "step": 22159
-    },
-    {
-      "epoch": 3.5687024437376706,
-      "grad_norm": 0.001319185714237392,
-      "learning_rate": 0.0001999937213563844,
-      "loss": 46.0,
-      "step": 22160
-    },
-    {
-      "epoch": 3.568863480816458,
-      "grad_norm": 0.005401642993092537,
-      "learning_rate": 0.00019999372078945715,
-      "loss": 46.0,
-      "step": 22161
-    },
-    {
-      "epoch": 3.5690245178952456,
-      "grad_norm": 0.0022573701571673155,
-      "learning_rate": 0.00019999372022250426,
-      "loss": 46.0,
-      "step": 22162
-    },
-    {
-      "epoch": 3.569185554974033,
-      "grad_norm": 0.00423166248947382,
-      "learning_rate": 0.0001999937196555258,
-      "loss": 46.0,
-      "step": 22163
-    },
-    {
-      "epoch": 3.56934659205282,
-      "grad_norm": 0.004280682187527418,
-      "learning_rate": 0.00019999371908852173,
-      "loss": 46.0,
-      "step": 22164
-    },
-    {
-      "epoch": 3.5695076291316075,
-      "grad_norm": 0.007625343278050423,
-      "learning_rate": 0.0001999937185214921,
-      "loss": 46.0,
-      "step": 22165
-    },
-    {
-      "epoch": 3.569668666210395,
-      "grad_norm": 0.002587782684713602,
-      "learning_rate": 0.00019999371795443685,
-      "loss": 46.0,
-      "step": 22166
-    },
-    {
-      "epoch": 3.5698297032891824,
-      "grad_norm": 0.0010397207224741578,
-      "learning_rate": 0.000199993717387356,
-      "loss": 46.0,
-      "step": 22167
-    },
-    {
-      "epoch": 3.56999074036797,
-      "grad_norm": 0.001793302595615387,
-      "learning_rate": 0.00019999371682024957,
-      "loss": 46.0,
-      "step": 22168
-    },
-    {
-      "epoch": 3.5701517774467573,
-      "grad_norm": 0.010664881207048893,
-      "learning_rate": 0.00019999371625311753,
-      "loss": 46.0,
-      "step": 22169
-    },
-    {
-      "epoch": 3.5703128145255443,
-      "grad_norm": 0.004823134280741215,
-      "learning_rate": 0.00019999371568595994,
-      "loss": 46.0,
-      "step": 22170
-    },
-    {
-      "epoch": 3.5704738516043317,
-      "grad_norm": 0.008988243527710438,
-      "learning_rate": 0.0001999937151187767,
-      "loss": 46.0,
-      "step": 22171
-    },
-    {
-      "epoch": 3.570634888683119,
-      "grad_norm": 0.004488545004278421,
-      "learning_rate": 0.0001999937145515679,
-      "loss": 46.0,
-      "step": 22172
-    },
-    {
-      "epoch": 3.5707959257619066,
-      "grad_norm": 0.00324258953332901,
-      "learning_rate": 0.0001999937139843335,
-      "loss": 46.0,
-      "step": 22173
-    },
-    {
-      "epoch": 3.570956962840694,
-      "grad_norm": 0.0019249595934525132,
-      "learning_rate": 0.00019999371341707353,
-      "loss": 46.0,
-      "step": 22174
-    },
-    {
-      "epoch": 3.5711179999194815,
-      "grad_norm": 0.0022335245739668608,
-      "learning_rate": 0.00019999371284978791,
-      "loss": 46.0,
-      "step": 22175
-    },
-    {
-      "epoch": 3.571279036998269,
-      "grad_norm": 0.0029795197769999504,
-      "learning_rate": 0.00019999371228247674,
-      "loss": 46.0,
-      "step": 22176
-    },
-    {
-      "epoch": 3.5714400740770564,
-      "grad_norm": 0.0024626704398542643,
-      "learning_rate": 0.00019999371171513996,
-      "loss": 46.0,
-      "step": 22177
-    },
-    {
-      "epoch": 3.571601111155844,
-      "grad_norm": 0.0039766328409314156,
-      "learning_rate": 0.0001999937111477776,
-      "loss": 46.0,
-      "step": 22178
-    },
-    {
-      "epoch": 3.571762148234631,
-      "grad_norm": 0.0012694451725110412,
-      "learning_rate": 0.00019999371058038965,
-      "loss": 46.0,
-      "step": 22179
-    },
-    {
-      "epoch": 3.5719231853134183,
-      "grad_norm": 0.0032486470881849527,
-      "learning_rate": 0.0001999937100129761,
-      "loss": 46.0,
-      "step": 22180
-    },
-    {
-      "epoch": 3.572084222392206,
-      "grad_norm": 0.0004951511509716511,
-      "learning_rate": 0.00019999370944553694,
-      "loss": 46.0,
-      "step": 22181
-    },
-    {
-      "epoch": 3.5722452594709933,
-      "grad_norm": 0.0026333224959671497,
-      "learning_rate": 0.00019999370887807222,
-      "loss": 46.0,
-      "step": 22182
-    },
-    {
-      "epoch": 3.5724062965497807,
-      "grad_norm": 0.006623640190809965,
-      "learning_rate": 0.00019999370831058188,
-      "loss": 46.0,
-      "step": 22183
-    },
-    {
-      "epoch": 3.5725673336285677,
-      "grad_norm": 0.008266623131930828,
-      "learning_rate": 0.00019999370774306593,
-      "loss": 46.0,
-      "step": 22184
-    },
-    {
-      "epoch": 3.572728370707355,
-      "grad_norm": 0.005043573211878538,
-      "learning_rate": 0.00019999370717552442,
-      "loss": 46.0,
-      "step": 22185
-    },
-    {
-      "epoch": 3.5728894077861426,
-      "grad_norm": 0.003764103865250945,
-      "learning_rate": 0.00019999370660795732,
-      "loss": 46.0,
-      "step": 22186
-    },
-    {
-      "epoch": 3.57305044486493,
-      "grad_norm": 0.006669378373771906,
-      "learning_rate": 0.0001999937060403646,
-      "loss": 46.0,
-      "step": 22187
-    },
-    {
-      "epoch": 3.5732114819437175,
-      "grad_norm": 0.0010816592257469893,
-      "learning_rate": 0.0001999937054727463,
-      "loss": 46.0,
-      "step": 22188
-    },
-    {
-      "epoch": 3.573372519022505,
-      "grad_norm": 0.0041970666497945786,
-      "learning_rate": 0.00019999370490510242,
-      "loss": 46.0,
-      "step": 22189
-    },
-    {
-      "epoch": 3.5735335561012924,
-      "grad_norm": 0.012897015549242496,
-      "learning_rate": 0.00019999370433743292,
-      "loss": 46.0,
-      "step": 22190
-    },
-    {
-      "epoch": 3.57369459318008,
-      "grad_norm": 0.004952274262905121,
-      "learning_rate": 0.00019999370376973786,
-      "loss": 46.0,
-      "step": 22191
-    },
-    {
-      "epoch": 3.5738556302588673,
-      "grad_norm": 0.003932930063456297,
-      "learning_rate": 0.00019999370320201719,
-      "loss": 46.0,
-      "step": 22192
-    },
-    {
-      "epoch": 3.5740166673376543,
-      "grad_norm": 0.0011036846553906798,
-      "learning_rate": 0.0001999937026342709,
-      "loss": 46.0,
-      "step": 22193
-    },
-    {
-      "epoch": 3.574177704416442,
-      "grad_norm": 0.0029764981009066105,
-      "learning_rate": 0.00019999370206649905,
-      "loss": 46.0,
-      "step": 22194
-    },
-    {
-      "epoch": 3.5743387414952292,
-      "grad_norm": 0.002830793149769306,
-      "learning_rate": 0.0001999937014987016,
-      "loss": 46.0,
-      "step": 22195
-    },
-    {
-      "epoch": 3.5744997785740167,
-      "grad_norm": 0.01494094729423523,
-      "learning_rate": 0.00019999370093087856,
-      "loss": 46.0,
-      "step": 22196
-    },
-    {
-      "epoch": 3.574660815652804,
-      "grad_norm": 0.0011509484611451626,
-      "learning_rate": 0.00019999370036302992,
-      "loss": 46.0,
-      "step": 22197
-    },
-    {
-      "epoch": 3.5748218527315916,
-      "grad_norm": 0.002567818621173501,
-      "learning_rate": 0.0001999936997951557,
-      "loss": 46.0,
-      "step": 22198
-    },
-    {
-      "epoch": 3.5749828898103786,
-      "grad_norm": 0.0015267434064298868,
-      "learning_rate": 0.00019999369922725588,
-      "loss": 46.0,
-      "step": 22199
-    },
-    {
-      "epoch": 3.575143926889166,
-      "grad_norm": 0.0023083004634827375,
-      "learning_rate": 0.00019999369865933043,
-      "loss": 46.0,
-      "step": 22200
-    },
-    {
-      "epoch": 3.5753049639679535,
-      "grad_norm": 0.008898667059838772,
-      "learning_rate": 0.00019999369809137942,
-      "loss": 46.0,
-      "step": 22201
-    },
-    {
-      "epoch": 3.575466001046741,
-      "grad_norm": 0.003886572550982237,
-      "learning_rate": 0.00019999369752340281,
-      "loss": 46.0,
-      "step": 22202
-    },
-    {
-      "epoch": 3.5756270381255284,
-      "grad_norm": 0.008097635582089424,
-      "learning_rate": 0.00019999369695540065,
-      "loss": 46.0,
-      "step": 22203
-    },
-    {
-      "epoch": 3.575788075204316,
-      "grad_norm": 0.0014989827759563923,
-      "learning_rate": 0.00019999369638737285,
-      "loss": 46.0,
-      "step": 22204
-    },
-    {
-      "epoch": 3.5759491122831033,
-      "grad_norm": 0.0005136894178576767,
-      "learning_rate": 0.00019999369581931946,
-      "loss": 46.0,
-      "step": 22205
-    },
-    {
-      "epoch": 3.5761101493618908,
-      "grad_norm": 0.0012887001503258944,
-      "learning_rate": 0.0001999936952512405,
-      "loss": 46.0,
-      "step": 22206
-    },
-    {
-      "epoch": 3.576271186440678,
-      "grad_norm": 0.008748008869588375,
-      "learning_rate": 0.00019999369468313592,
-      "loss": 46.0,
-      "step": 22207
-    },
-    {
-      "epoch": 3.576432223519465,
-      "grad_norm": 0.006734238937497139,
-      "learning_rate": 0.00019999369411500574,
-      "loss": 46.0,
-      "step": 22208
-    },
-    {
-      "epoch": 3.5765932605982527,
-      "grad_norm": 0.004319359548389912,
-      "learning_rate": 0.00019999369354685,
-      "loss": 46.0,
-      "step": 22209
-    },
-    {
-      "epoch": 3.57675429767704,
-      "grad_norm": 0.0028410619124770164,
-      "learning_rate": 0.00019999369297866862,
-      "loss": 46.0,
-      "step": 22210
-    },
-    {
-      "epoch": 3.5769153347558276,
-      "grad_norm": 0.0024277695920318365,
-      "learning_rate": 0.00019999369241046168,
-      "loss": 46.0,
-      "step": 22211
-    },
-    {
-      "epoch": 3.577076371834615,
-      "grad_norm": 0.004947687964886427,
-      "learning_rate": 0.00019999369184222916,
-      "loss": 46.0,
-      "step": 22212
-    },
-    {
-      "epoch": 3.5772374089134025,
-      "grad_norm": 0.010552571155130863,
-      "learning_rate": 0.00019999369127397102,
-      "loss": 46.0,
-      "step": 22213
-    },
-    {
-      "epoch": 3.5773984459921895,
-      "grad_norm": 0.007573977112770081,
-      "learning_rate": 0.00019999369070568731,
-      "loss": 46.0,
-      "step": 22214
-    },
-    {
-      "epoch": 3.577559483070977,
-      "grad_norm": 0.002108183689415455,
-      "learning_rate": 0.000199993690137378,
-      "loss": 46.0,
-      "step": 22215
-    },
-    {
-      "epoch": 3.5777205201497644,
-      "grad_norm": 0.0011062516132369637,
-      "learning_rate": 0.00019999368956904307,
-      "loss": 46.0,
-      "step": 22216
-    },
-    {
-      "epoch": 3.577881557228552,
-      "grad_norm": 0.0013869494432583451,
-      "learning_rate": 0.00019999368900068255,
-      "loss": 46.0,
-      "step": 22217
-    },
-    {
-      "epoch": 3.5780425943073393,
-      "grad_norm": 0.005112015642225742,
-      "learning_rate": 0.00019999368843229647,
-      "loss": 46.0,
-      "step": 22218
-    },
-    {
-      "epoch": 3.5782036313861267,
-      "grad_norm": 0.0071973856538534164,
-      "learning_rate": 0.00019999368786388478,
-      "loss": 46.0,
-      "step": 22219
-    },
-    {
-      "epoch": 3.578364668464914,
-      "grad_norm": 0.007174696307629347,
-      "learning_rate": 0.0001999936872954475,
-      "loss": 46.0,
-      "step": 22220
-    },
-    {
-      "epoch": 3.5785257055437016,
-      "grad_norm": 0.0036894932854920626,
-      "learning_rate": 0.0001999936867269846,
-      "loss": 46.0,
-      "step": 22221
-    },
-    {
-      "epoch": 3.578686742622489,
-      "grad_norm": 0.004031945485621691,
-      "learning_rate": 0.00019999368615849616,
-      "loss": 46.0,
-      "step": 22222
-    },
-    {
-      "epoch": 3.578847779701276,
-      "grad_norm": 0.00776843074709177,
-      "learning_rate": 0.0001999936855899821,
-      "loss": 46.0,
-      "step": 22223
-    },
-    {
-      "epoch": 3.5790088167800636,
-      "grad_norm": 0.01725834608078003,
-      "learning_rate": 0.00019999368502144244,
-      "loss": 46.0,
-      "step": 22224
-    },
-    {
-      "epoch": 3.579169853858851,
-      "grad_norm": 0.0049018352292478085,
-      "learning_rate": 0.0001999936844528772,
-      "loss": 46.0,
-      "step": 22225
-    },
-    {
-      "epoch": 3.5793308909376385,
-      "grad_norm": 0.0024316362105309963,
-      "learning_rate": 0.00019999368388428637,
-      "loss": 46.0,
-      "step": 22226
-    },
-    {
-      "epoch": 3.579491928016426,
-      "grad_norm": 0.0011868409346789122,
-      "learning_rate": 0.00019999368331566992,
-      "loss": 46.0,
-      "step": 22227
-    },
-    {
-      "epoch": 3.579652965095213,
-      "grad_norm": 0.004764891229569912,
-      "learning_rate": 0.0001999936827470279,
-      "loss": 46.0,
-      "step": 22228
-    },
-    {
-      "epoch": 3.5798140021740004,
-      "grad_norm": 0.005578783340752125,
-      "learning_rate": 0.00019999368217836025,
-      "loss": 46.0,
-      "step": 22229
-    },
-    {
-      "epoch": 3.579975039252788,
-      "grad_norm": 0.0029380801133811474,
-      "learning_rate": 0.00019999368160966704,
-      "loss": 46.0,
-      "step": 22230
-    },
-    {
-      "epoch": 3.5801360763315753,
-      "grad_norm": 0.005727612879127264,
-      "learning_rate": 0.00019999368104094823,
-      "loss": 46.0,
-      "step": 22231
-    },
-    {
-      "epoch": 3.5802971134103627,
-      "grad_norm": 0.01643526181578636,
-      "learning_rate": 0.00019999368047220385,
-      "loss": 46.0,
-      "step": 22232
-    },
-    {
-      "epoch": 3.58045815048915,
-      "grad_norm": 0.00436242762953043,
-      "learning_rate": 0.00019999367990343383,
-      "loss": 46.0,
-      "step": 22233
-    },
-    {
-      "epoch": 3.5806191875679376,
-      "grad_norm": 0.004011080134660006,
-      "learning_rate": 0.00019999367933463825,
-      "loss": 46.0,
-      "step": 22234
-    },
-    {
-      "epoch": 3.580780224646725,
-      "grad_norm": 0.0024378944654017687,
-      "learning_rate": 0.00019999367876581708,
-      "loss": 46.0,
-      "step": 22235
-    },
-    {
-      "epoch": 3.5809412617255125,
-      "grad_norm": 0.0018386357696726918,
-      "learning_rate": 0.00019999367819697027,
-      "loss": 46.0,
-      "step": 22236
-    },
-    {
-      "epoch": 3.5811022988042995,
-      "grad_norm": 0.003888638922944665,
-      "learning_rate": 0.00019999367762809793,
-      "loss": 46.0,
-      "step": 22237
-    },
-    {
-      "epoch": 3.581263335883087,
-      "grad_norm": 0.0023935684002935886,
-      "learning_rate": 0.00019999367705919995,
-      "loss": 46.0,
-      "step": 22238
-    },
-    {
-      "epoch": 3.5814243729618744,
-      "grad_norm": 0.005769809242337942,
-      "learning_rate": 0.0001999936764902764,
-      "loss": 46.0,
-      "step": 22239
-    },
-    {
-      "epoch": 3.581585410040662,
-      "grad_norm": 0.0028870890382677317,
-      "learning_rate": 0.00019999367592132727,
-      "loss": 46.0,
-      "step": 22240
-    },
-    {
-      "epoch": 3.5817464471194493,
-      "grad_norm": 0.003583308309316635,
-      "learning_rate": 0.00019999367535235253,
-      "loss": 46.0,
-      "step": 22241
-    },
-    {
-      "epoch": 3.581907484198237,
-      "grad_norm": 0.006188003346323967,
-      "learning_rate": 0.00019999367478335217,
-      "loss": 46.0,
-      "step": 22242
-    },
-    {
-      "epoch": 3.582068521277024,
-      "grad_norm": 0.004220211878418922,
-      "learning_rate": 0.00019999367421432625,
-      "loss": 46.0,
-      "step": 22243
-    },
-    {
-      "epoch": 3.5822295583558112,
-      "grad_norm": 0.0036809046287089586,
-      "learning_rate": 0.00019999367364527472,
-      "loss": 46.0,
-      "step": 22244
-    },
-    {
-      "epoch": 3.5823905954345987,
-      "grad_norm": 0.01166660524904728,
-      "learning_rate": 0.00019999367307619762,
-      "loss": 46.0,
-      "step": 22245
-    },
-    {
-      "epoch": 3.582551632513386,
-      "grad_norm": 0.008661210536956787,
-      "learning_rate": 0.00019999367250709491,
-      "loss": 46.0,
-      "step": 22246
-    },
-    {
-      "epoch": 3.5827126695921736,
-      "grad_norm": 0.00344940391369164,
-      "learning_rate": 0.00019999367193796662,
-      "loss": 46.0,
-      "step": 22247
-    },
-    {
-      "epoch": 3.582873706670961,
-      "grad_norm": 0.005918078124523163,
-      "learning_rate": 0.00019999367136881274,
-      "loss": 46.0,
-      "step": 22248
-    },
-    {
-      "epoch": 3.5830347437497485,
-      "grad_norm": 0.008551975712180138,
-      "learning_rate": 0.00019999367079963321,
-      "loss": 46.0,
-      "step": 22249
-    },
-    {
-      "epoch": 3.583195780828536,
-      "grad_norm": 0.0004640891856979579,
-      "learning_rate": 0.00019999367023042813,
-      "loss": 46.0,
-      "step": 22250
-    },
-    {
-      "epoch": 3.5833568179073234,
-      "grad_norm": 0.0029130319599062204,
-      "learning_rate": 0.0001999936696611975,
-      "loss": 46.0,
-      "step": 22251
-    },
-    {
-      "epoch": 3.5835178549861104,
-      "grad_norm": 0.003528533736243844,
-      "learning_rate": 0.0001999936690919412,
-      "loss": 46.0,
-      "step": 22252
-    },
-    {
-      "epoch": 3.583678892064898,
-      "grad_norm": 0.005134756676852703,
-      "learning_rate": 0.00019999366852265936,
-      "loss": 46.0,
-      "step": 22253
-    },
-    {
-      "epoch": 3.5838399291436853,
-      "grad_norm": 0.006688056513667107,
-      "learning_rate": 0.00019999366795335187,
-      "loss": 46.0,
-      "step": 22254
-    },
-    {
-      "epoch": 3.5840009662224728,
-      "grad_norm": 0.002679308643564582,
-      "learning_rate": 0.00019999366738401885,
-      "loss": 46.0,
-      "step": 22255
-    },
-    {
-      "epoch": 3.5841620033012602,
-      "grad_norm": 0.0009020857396535575,
-      "learning_rate": 0.00019999366681466022,
-      "loss": 46.0,
-      "step": 22256
-    },
-    {
-      "epoch": 3.5843230403800472,
-      "grad_norm": 0.003891663858667016,
-      "learning_rate": 0.00019999366624527597,
-      "loss": 46.0,
-      "step": 22257
-    },
-    {
-      "epoch": 3.5844840774588347,
-      "grad_norm": 0.006270218174904585,
-      "learning_rate": 0.00019999366567586614,
-      "loss": 46.0,
-      "step": 22258
-    },
-    {
-      "epoch": 3.584645114537622,
-      "grad_norm": 0.012125732377171516,
-      "learning_rate": 0.00019999366510643074,
-      "loss": 46.0,
-      "step": 22259
-    },
-    {
-      "epoch": 3.5848061516164096,
-      "grad_norm": 0.0018512671813368797,
-      "learning_rate": 0.00019999366453696973,
-      "loss": 46.0,
-      "step": 22260
-    },
-    {
-      "epoch": 3.584967188695197,
-      "grad_norm": 0.00933990627527237,
-      "learning_rate": 0.0001999936639674831,
-      "loss": 46.0,
-      "step": 22261
-    },
-    {
-      "epoch": 3.5851282257739845,
-      "grad_norm": 0.002727724378928542,
-      "learning_rate": 0.00019999366339797092,
-      "loss": 46.0,
-      "step": 22262
-    },
-    {
-      "epoch": 3.585289262852772,
-      "grad_norm": 0.001745673012919724,
-      "learning_rate": 0.00019999366282843313,
-      "loss": 46.0,
-      "step": 22263
-    },
-    {
-      "epoch": 3.5854502999315594,
-      "grad_norm": 0.0011512573109939694,
-      "learning_rate": 0.00019999366225886974,
-      "loss": 46.0,
-      "step": 22264
-    },
-    {
-      "epoch": 3.585611337010347,
-      "grad_norm": 0.0033175284042954445,
-      "learning_rate": 0.00019999366168928074,
-      "loss": 46.0,
-      "step": 22265
-    },
-    {
-      "epoch": 3.5857723740891343,
-      "grad_norm": 0.0012629320845007896,
-      "learning_rate": 0.00019999366111966618,
-      "loss": 46.0,
-      "step": 22266
-    },
-    {
-      "epoch": 3.5859334111679213,
-      "grad_norm": 0.007327020168304443,
-      "learning_rate": 0.000199993660550026,
-      "loss": 46.0,
-      "step": 22267
-    },
-    {
-      "epoch": 3.5860944482467088,
-      "grad_norm": 0.0014256129506975412,
-      "learning_rate": 0.00019999365998036025,
-      "loss": 46.0,
-      "step": 22268
-    },
-    {
-      "epoch": 3.586255485325496,
-      "grad_norm": 0.0027817138470709324,
-      "learning_rate": 0.00019999365941066893,
-      "loss": 46.0,
-      "step": 22269
-    },
-    {
-      "epoch": 3.5864165224042837,
-      "grad_norm": 0.001265592174604535,
-      "learning_rate": 0.00019999365884095196,
-      "loss": 46.0,
-      "step": 22270
-    },
-    {
-      "epoch": 3.586577559483071,
-      "grad_norm": 0.008917558006942272,
-      "learning_rate": 0.00019999365827120941,
-      "loss": 46.0,
-      "step": 22271
-    },
-    {
-      "epoch": 3.586738596561858,
-      "grad_norm": 0.0007122106617316604,
-      "learning_rate": 0.00019999365770144128,
-      "loss": 46.0,
-      "step": 22272
-    },
-    {
-      "epoch": 3.5868996336406456,
-      "grad_norm": 0.004836119245737791,
-      "learning_rate": 0.00019999365713164755,
-      "loss": 46.0,
-      "step": 22273
-    },
-    {
-      "epoch": 3.587060670719433,
-      "grad_norm": 0.0038063570391386747,
-      "learning_rate": 0.00019999365656182824,
-      "loss": 46.0,
-      "step": 22274
-    },
-    {
-      "epoch": 3.5872217077982205,
-      "grad_norm": 0.0018937435233965516,
-      "learning_rate": 0.00019999365599198332,
-      "loss": 46.0,
-      "step": 22275
-    },
-    {
-      "epoch": 3.587382744877008,
-      "grad_norm": 0.006482832133769989,
-      "learning_rate": 0.0001999936554221128,
-      "loss": 46.0,
-      "step": 22276
-    },
-    {
-      "epoch": 3.5875437819557954,
-      "grad_norm": 0.00454520620405674,
-      "learning_rate": 0.00019999365485221673,
-      "loss": 46.0,
-      "step": 22277
-    },
-    {
-      "epoch": 3.587704819034583,
-      "grad_norm": 0.0012868338963016868,
-      "learning_rate": 0.00019999365428229504,
-      "loss": 46.0,
-      "step": 22278
-    },
-    {
-      "epoch": 3.5878658561133703,
-      "grad_norm": 0.021897710859775543,
-      "learning_rate": 0.00019999365371234774,
-      "loss": 46.0,
-      "step": 22279
-    },
-    {
-      "epoch": 3.5880268931921577,
-      "grad_norm": 0.0013524509267881513,
-      "learning_rate": 0.00019999365314237485,
-      "loss": 46.0,
-      "step": 22280
-    },
-    {
-      "epoch": 3.5881879302709447,
-      "grad_norm": 0.0031641307286918163,
-      "learning_rate": 0.0001999936525723764,
-      "loss": 46.0,
-      "step": 22281
-    },
-    {
-      "epoch": 3.588348967349732,
-      "grad_norm": 0.001093899947591126,
-      "learning_rate": 0.00019999365200235234,
-      "loss": 46.0,
-      "step": 22282
-    },
-    {
-      "epoch": 3.5885100044285196,
-      "grad_norm": 0.0018966369098052382,
-      "learning_rate": 0.0001999936514323027,
-      "loss": 46.0,
-      "step": 22283
-    },
-    {
-      "epoch": 3.588671041507307,
-      "grad_norm": 0.0013478579930961132,
-      "learning_rate": 0.00019999365086222743,
-      "loss": 46.0,
-      "step": 22284
-    },
-    {
-      "epoch": 3.5888320785860945,
-      "grad_norm": 0.008609525859355927,
-      "learning_rate": 0.0001999936502921266,
-      "loss": 46.0,
-      "step": 22285
-    },
-    {
-      "epoch": 3.588993115664882,
-      "grad_norm": 0.006135663483291864,
-      "learning_rate": 0.00019999364972200017,
-      "loss": 46.0,
-      "step": 22286
-    },
-    {
-      "epoch": 3.589154152743669,
-      "grad_norm": 0.0014241492608562112,
-      "learning_rate": 0.00019999364915184811,
-      "loss": 46.0,
-      "step": 22287
-    },
-    {
-      "epoch": 3.5893151898224565,
-      "grad_norm": 0.006008440162986517,
-      "learning_rate": 0.0001999936485816705,
-      "loss": 46.0,
-      "step": 22288
-    },
-    {
-      "epoch": 3.589476226901244,
-      "grad_norm": 0.002906738081946969,
-      "learning_rate": 0.0001999936480114673,
-      "loss": 46.0,
-      "step": 22289
-    },
-    {
-      "epoch": 3.5896372639800314,
-      "grad_norm": 0.0026723395567387342,
-      "learning_rate": 0.00019999364744123846,
-      "loss": 46.0,
-      "step": 22290
-    },
-    {
-      "epoch": 3.589798301058819,
-      "grad_norm": 0.0034699132665991783,
-      "learning_rate": 0.00019999364687098409,
-      "loss": 46.0,
-      "step": 22291
-    },
-    {
-      "epoch": 3.5899593381376063,
-      "grad_norm": 0.001396158360876143,
-      "learning_rate": 0.00019999364630070407,
-      "loss": 46.0,
-      "step": 22292
-    },
-    {
-      "epoch": 3.5901203752163937,
-      "grad_norm": 0.0033217184245586395,
-      "learning_rate": 0.0001999936457303985,
-      "loss": 46.0,
-      "step": 22293
-    },
-    {
-      "epoch": 3.590281412295181,
-      "grad_norm": 0.014631196856498718,
-      "learning_rate": 0.00019999364516006728,
-      "loss": 46.0,
-      "step": 22294
-    },
-    {
-      "epoch": 3.5904424493739686,
-      "grad_norm": 0.001453693606890738,
-      "learning_rate": 0.0001999936445897105,
-      "loss": 46.0,
-      "step": 22295
-    },
-    {
-      "epoch": 3.5906034864527556,
-      "grad_norm": 0.0015400357078760862,
-      "learning_rate": 0.00019999364401932814,
-      "loss": 46.0,
-      "step": 22296
-    },
-    {
-      "epoch": 3.590764523531543,
-      "grad_norm": 0.005252873990684748,
-      "learning_rate": 0.0001999936434489202,
-      "loss": 46.0,
-      "step": 22297
-    },
-    {
-      "epoch": 3.5909255606103305,
-      "grad_norm": 0.0033839144743978977,
-      "learning_rate": 0.00019999364287848662,
-      "loss": 46.0,
-      "step": 22298
-    },
-    {
-      "epoch": 3.591086597689118,
-      "grad_norm": 0.005861698184162378,
-      "learning_rate": 0.0001999936423080275,
-      "loss": 46.0,
-      "step": 22299
-    },
-    {
-      "epoch": 3.5912476347679054,
-      "grad_norm": 0.001217004144564271,
-      "learning_rate": 0.00019999364173754273,
-      "loss": 46.0,
-      "step": 22300
-    },
-    {
-      "epoch": 3.5914086718466924,
-      "grad_norm": 0.0015326649881899357,
-      "learning_rate": 0.0001999936411670324,
-      "loss": 46.0,
-      "step": 22301
-    },
-    {
-      "epoch": 3.59156970892548,
-      "grad_norm": 0.0013569958973675966,
-      "learning_rate": 0.00019999364059649648,
-      "loss": 46.0,
-      "step": 22302
-    },
-    {
-      "epoch": 3.5917307460042673,
-      "grad_norm": 0.007940462790429592,
-      "learning_rate": 0.00019999364002593495,
-      "loss": 46.0,
-      "step": 22303
-    },
-    {
-      "epoch": 3.591891783083055,
-      "grad_norm": 0.0034608205314725637,
-      "learning_rate": 0.00019999363945534784,
-      "loss": 46.0,
-      "step": 22304
-    },
-    {
-      "epoch": 3.5920528201618422,
-      "grad_norm": 0.004165707156062126,
-      "learning_rate": 0.00019999363888473514,
-      "loss": 46.0,
-      "step": 22305
-    },
-    {
-      "epoch": 3.5922138572406297,
-      "grad_norm": 0.0016852917615324259,
-      "learning_rate": 0.00019999363831409685,
-      "loss": 46.0,
-      "step": 22306
-    },
-    {
-      "epoch": 3.592374894319417,
-      "grad_norm": 0.0013217369560152292,
-      "learning_rate": 0.0001999936377434329,
-      "loss": 46.0,
-      "step": 22307
-    },
-    {
-      "epoch": 3.5925359313982046,
-      "grad_norm": 0.001958205597475171,
-      "learning_rate": 0.00019999363717274342,
-      "loss": 46.0,
-      "step": 22308
-    },
-    {
-      "epoch": 3.592696968476992,
-      "grad_norm": 0.007180316373705864,
-      "learning_rate": 0.00019999363660202834,
-      "loss": 46.0,
-      "step": 22309
-    },
-    {
-      "epoch": 3.592858005555779,
-      "grad_norm": 0.0033955092076212168,
-      "learning_rate": 0.00019999363603128768,
-      "loss": 46.0,
-      "step": 22310
-    },
-    {
-      "epoch": 3.5930190426345665,
-      "grad_norm": 0.004092859569936991,
-      "learning_rate": 0.00019999363546052142,
-      "loss": 46.0,
-      "step": 22311
-    },
-    {
-      "epoch": 3.593180079713354,
-      "grad_norm": 0.0018009928753599524,
-      "learning_rate": 0.00019999363488972956,
-      "loss": 46.0,
-      "step": 22312
-    },
-    {
-      "epoch": 3.5933411167921414,
-      "grad_norm": 0.009082680568099022,
-      "learning_rate": 0.0001999936343189121,
-      "loss": 46.0,
-      "step": 22313
-    },
-    {
-      "epoch": 3.593502153870929,
-      "grad_norm": 0.001087721437215805,
-      "learning_rate": 0.00019999363374806906,
-      "loss": 46.0,
-      "step": 22314
-    },
-    {
-      "epoch": 3.5936631909497163,
-      "grad_norm": 0.003085762495175004,
-      "learning_rate": 0.0001999936331772004,
-      "loss": 46.0,
-      "step": 22315
-    },
-    {
-      "epoch": 3.5938242280285033,
-      "grad_norm": 0.007600707001984119,
-      "learning_rate": 0.00019999363260630616,
-      "loss": 46.0,
-      "step": 22316
-    },
-    {
-      "epoch": 3.5939852651072908,
-      "grad_norm": 0.0040770480409264565,
-      "learning_rate": 0.00019999363203538636,
-      "loss": 46.0,
-      "step": 22317
-    },
-    {
-      "epoch": 3.5941463021860782,
-      "grad_norm": 0.00740771135315299,
-      "learning_rate": 0.0001999936314644409,
-      "loss": 46.0,
-      "step": 22318
-    },
-    {
-      "epoch": 3.5943073392648657,
-      "grad_norm": 0.00436649052426219,
-      "learning_rate": 0.0001999936308934699,
-      "loss": 46.0,
-      "step": 22319
-    },
-    {
-      "epoch": 3.594468376343653,
-      "grad_norm": 0.004158633295446634,
-      "learning_rate": 0.0001999936303224733,
-      "loss": 46.0,
-      "step": 22320
-    },
-    {
-      "epoch": 3.5946294134224406,
-      "grad_norm": 0.005858105607330799,
-      "learning_rate": 0.00019999362975145108,
-      "loss": 46.0,
-      "step": 22321
-    },
-    {
-      "epoch": 3.594790450501228,
-      "grad_norm": 0.004049082286655903,
-      "learning_rate": 0.0001999936291804033,
-      "loss": 46.0,
-      "step": 22322
-    },
-    {
-      "epoch": 3.5949514875800155,
-      "grad_norm": 0.003086875192821026,
-      "learning_rate": 0.00019999362860932993,
-      "loss": 46.0,
-      "step": 22323
-    },
-    {
-      "epoch": 3.595112524658803,
-      "grad_norm": 0.007564725819975138,
-      "learning_rate": 0.00019999362803823094,
-      "loss": 46.0,
-      "step": 22324
-    },
-    {
-      "epoch": 3.59527356173759,
-      "grad_norm": 0.0008835851913318038,
-      "learning_rate": 0.00019999362746710636,
-      "loss": 46.0,
-      "step": 22325
-    },
-    {
-      "epoch": 3.5954345988163774,
-      "grad_norm": 0.0029259880539029837,
-      "learning_rate": 0.0001999936268959562,
-      "loss": 46.0,
-      "step": 22326
-    },
-    {
-      "epoch": 3.595595635895165,
-      "grad_norm": 0.002532773884013295,
-      "learning_rate": 0.00019999362632478046,
-      "loss": 46.0,
-      "step": 22327
-    },
-    {
-      "epoch": 3.5957566729739523,
-      "grad_norm": 0.006726366933435202,
-      "learning_rate": 0.0001999936257535791,
-      "loss": 46.0,
-      "step": 22328
-    },
-    {
-      "epoch": 3.5959177100527397,
-      "grad_norm": 0.002553165890276432,
-      "learning_rate": 0.00019999362518235213,
-      "loss": 46.0,
-      "step": 22329
-    },
-    {
-      "epoch": 3.596078747131527,
-      "grad_norm": 0.0013454457512125373,
-      "learning_rate": 0.0001999936246110996,
-      "loss": 46.0,
-      "step": 22330
-    },
-    {
-      "epoch": 3.596239784210314,
-      "grad_norm": 0.004314367659389973,
-      "learning_rate": 0.00019999362403982148,
-      "loss": 46.0,
-      "step": 22331
-    },
-    {
-      "epoch": 3.5964008212891017,
-      "grad_norm": 0.0022494480945169926,
-      "learning_rate": 0.00019999362346851773,
-      "loss": 46.0,
-      "step": 22332
-    },
-    {
-      "epoch": 3.596561858367889,
-      "grad_norm": 0.0018538335571065545,
-      "learning_rate": 0.00019999362289718843,
-      "loss": 46.0,
-      "step": 22333
-    },
-    {
-      "epoch": 3.5967228954466766,
-      "grad_norm": 0.011925830505788326,
-      "learning_rate": 0.0001999936223258335,
-      "loss": 46.0,
-      "step": 22334
-    },
-    {
-      "epoch": 3.596883932525464,
-      "grad_norm": 0.008757457137107849,
-      "learning_rate": 0.000199993621754453,
-      "loss": 46.0,
-      "step": 22335
-    },
-    {
-      "epoch": 3.5970449696042515,
-      "grad_norm": 0.015112852677702904,
-      "learning_rate": 0.0001999936211830469,
-      "loss": 46.0,
-      "step": 22336
-    },
-    {
-      "epoch": 3.597206006683039,
-      "grad_norm": 0.006661464925855398,
-      "learning_rate": 0.00019999362061161522,
-      "loss": 46.0,
-      "step": 22337
-    },
-    {
-      "epoch": 3.5973670437618264,
-      "grad_norm": 0.0035234594251960516,
-      "learning_rate": 0.00019999362004015792,
-      "loss": 46.0,
-      "step": 22338
-    },
-    {
-      "epoch": 3.597528080840614,
-      "grad_norm": 0.003458704799413681,
-      "learning_rate": 0.00019999361946867504,
-      "loss": 46.0,
-      "step": 22339
-    },
-    {
-      "epoch": 3.597689117919401,
-      "grad_norm": 0.0035416092723608017,
-      "learning_rate": 0.00019999361889716657,
-      "loss": 46.0,
-      "step": 22340
-    },
-    {
-      "epoch": 3.5978501549981883,
-      "grad_norm": 0.0010801405878737569,
-      "learning_rate": 0.0001999936183256325,
-      "loss": 46.0,
-      "step": 22341
-    },
-    {
-      "epoch": 3.5980111920769757,
-      "grad_norm": 0.0035934043116867542,
-      "learning_rate": 0.00019999361775407284,
-      "loss": 46.0,
-      "step": 22342
-    },
-    {
-      "epoch": 3.598172229155763,
-      "grad_norm": 0.0009040441946126521,
-      "learning_rate": 0.00019999361718248758,
-      "loss": 46.0,
-      "step": 22343
-    },
-    {
-      "epoch": 3.5983332662345506,
-      "grad_norm": 0.00912285316735506,
-      "learning_rate": 0.00019999361661087673,
-      "loss": 46.0,
-      "step": 22344
-    },
-    {
-      "epoch": 3.5984943033133376,
-      "grad_norm": 0.006449673790484667,
-      "learning_rate": 0.0001999936160392403,
-      "loss": 46.0,
-      "step": 22345
-    },
-    {
-      "epoch": 3.598655340392125,
-      "grad_norm": 0.009721963666379452,
-      "learning_rate": 0.00019999361546757828,
-      "loss": 46.0,
-      "step": 22346
-    },
-    {
-      "epoch": 3.5988163774709125,
-      "grad_norm": 0.0009141361224465072,
-      "learning_rate": 0.00019999361489589067,
-      "loss": 46.0,
-      "step": 22347
-    },
-    {
-      "epoch": 3.5989774145497,
-      "grad_norm": 0.0021284744143486023,
-      "learning_rate": 0.00019999361432417742,
-      "loss": 46.0,
-      "step": 22348
-    },
-    {
-      "epoch": 3.5991384516284874,
-      "grad_norm": 0.001282767509110272,
-      "learning_rate": 0.0001999936137524386,
-      "loss": 46.0,
-      "step": 22349
-    },
-    {
-      "epoch": 3.599299488707275,
-      "grad_norm": 0.002047148533165455,
-      "learning_rate": 0.0001999936131806742,
-      "loss": 46.0,
-      "step": 22350
-    },
-    {
-      "epoch": 3.5994605257860623,
-      "grad_norm": 0.001878916984423995,
-      "learning_rate": 0.0001999936126088842,
-      "loss": 46.0,
-      "step": 22351
-    },
-    {
-      "epoch": 3.59962156286485,
-      "grad_norm": 0.0034665060229599476,
-      "learning_rate": 0.00019999361203706863,
-      "loss": 46.0,
-      "step": 22352
-    },
-    {
-      "epoch": 3.5997825999436373,
-      "grad_norm": 0.0017352188006043434,
-      "learning_rate": 0.00019999361146522744,
-      "loss": 46.0,
-      "step": 22353
-    },
-    {
-      "epoch": 3.5999436370224243,
-      "grad_norm": 0.002765955636277795,
-      "learning_rate": 0.00019999361089336067,
-      "loss": 46.0,
-      "step": 22354
-    },
-    {
-      "epoch": 3.6001046741012117,
-      "grad_norm": 0.003913640510290861,
-      "learning_rate": 0.0001999936103214683,
-      "loss": 46.0,
-      "step": 22355
-    },
-    {
-      "epoch": 3.600265711179999,
-      "grad_norm": 0.00532541424036026,
-      "learning_rate": 0.00019999360974955034,
-      "loss": 46.0,
-      "step": 22356
-    },
-    {
-      "epoch": 3.6004267482587866,
-      "grad_norm": 0.00533573841676116,
-      "learning_rate": 0.00019999360917760678,
-      "loss": 46.0,
-      "step": 22357
-    },
-    {
-      "epoch": 3.600587785337574,
-      "grad_norm": 0.001956868451088667,
-      "learning_rate": 0.00019999360860563763,
-      "loss": 46.0,
-      "step": 22358
-    },
-    {
-      "epoch": 3.6007488224163615,
-      "grad_norm": 0.0009051914094015956,
-      "learning_rate": 0.00019999360803364287,
-      "loss": 46.0,
-      "step": 22359
-    },
-    {
-      "epoch": 3.6009098594951485,
-      "grad_norm": 0.003430198645219207,
-      "learning_rate": 0.00019999360746162252,
-      "loss": 46.0,
-      "step": 22360
-    },
-    {
-      "epoch": 3.601070896573936,
-      "grad_norm": 0.0006446716142818332,
-      "learning_rate": 0.0001999936068895766,
-      "loss": 46.0,
-      "step": 22361
-    },
-    {
-      "epoch": 3.6012319336527234,
-      "grad_norm": 0.006032352335751057,
-      "learning_rate": 0.0001999936063175051,
-      "loss": 46.0,
-      "step": 22362
-    },
-    {
-      "epoch": 3.601392970731511,
-      "grad_norm": 0.00455701956525445,
-      "learning_rate": 0.00019999360574540797,
-      "loss": 46.0,
-      "step": 22363
-    },
-    {
-      "epoch": 3.6015540078102983,
-      "grad_norm": 0.002215759363025427,
-      "learning_rate": 0.00019999360517328527,
-      "loss": 46.0,
-      "step": 22364
-    },
-    {
-      "epoch": 3.601715044889086,
-      "grad_norm": 0.002742103999480605,
-      "learning_rate": 0.00019999360460113693,
-      "loss": 46.0,
-      "step": 22365
-    },
-    {
-      "epoch": 3.6018760819678732,
-      "grad_norm": 0.002330132992938161,
-      "learning_rate": 0.00019999360402896306,
-      "loss": 46.0,
-      "step": 22366
-    },
-    {
-      "epoch": 3.6020371190466607,
-      "grad_norm": 0.011597679927945137,
-      "learning_rate": 0.00019999360345676357,
-      "loss": 46.0,
-      "step": 22367
-    },
-    {
-      "epoch": 3.602198156125448,
-      "grad_norm": 0.002671410096809268,
-      "learning_rate": 0.00019999360288453847,
-      "loss": 46.0,
-      "step": 22368
-    },
-    {
-      "epoch": 3.602359193204235,
-      "grad_norm": 0.0009399244445376098,
-      "learning_rate": 0.0001999936023122878,
-      "loss": 46.0,
-      "step": 22369
-    },
-    {
-      "epoch": 3.6025202302830226,
-      "grad_norm": 0.002231578342616558,
-      "learning_rate": 0.00019999360174001153,
-      "loss": 46.0,
-      "step": 22370
-    },
-    {
-      "epoch": 3.60268126736181,
-      "grad_norm": 0.00409920047968626,
-      "learning_rate": 0.00019999360116770967,
-      "loss": 46.0,
-      "step": 22371
-    },
-    {
-      "epoch": 3.6028423044405975,
-      "grad_norm": 0.001007885904982686,
-      "learning_rate": 0.00019999360059538225,
-      "loss": 46.0,
-      "step": 22372
-    },
-    {
-      "epoch": 3.603003341519385,
-      "grad_norm": 0.00783870555460453,
-      "learning_rate": 0.00019999360002302918,
-      "loss": 46.0,
-      "step": 22373
-    },
-    {
-      "epoch": 3.603164378598172,
-      "grad_norm": 0.005030629690736532,
-      "learning_rate": 0.00019999359945065053,
-      "loss": 46.0,
-      "step": 22374
-    },
-    {
-      "epoch": 3.6033254156769594,
-      "grad_norm": 0.005600613541901112,
-      "learning_rate": 0.0001999935988782463,
-      "loss": 46.0,
-      "step": 22375
-    },
-    {
-      "epoch": 3.603486452755747,
-      "grad_norm": 0.0009224683162756264,
-      "learning_rate": 0.00019999359830581647,
-      "loss": 46.0,
-      "step": 22376
-    },
-    {
-      "epoch": 3.6036474898345343,
-      "grad_norm": 0.002280866727232933,
-      "learning_rate": 0.00019999359773336102,
-      "loss": 46.0,
-      "step": 22377
-    },
-    {
-      "epoch": 3.6038085269133218,
-      "grad_norm": 0.0023356250021606684,
-      "learning_rate": 0.00019999359716088005,
-      "loss": 46.0,
-      "step": 22378
-    },
-    {
-      "epoch": 3.603969563992109,
-      "grad_norm": 0.001748585607856512,
-      "learning_rate": 0.0001999935965883734,
-      "loss": 46.0,
-      "step": 22379
-    },
-    {
-      "epoch": 3.6041306010708967,
-      "grad_norm": 0.003502840409055352,
-      "learning_rate": 0.00019999359601584124,
-      "loss": 46.0,
-      "step": 22380
-    },
-    {
-      "epoch": 3.604291638149684,
-      "grad_norm": 0.002135103801265359,
-      "learning_rate": 0.00019999359544328342,
-      "loss": 46.0,
-      "step": 22381
-    },
-    {
-      "epoch": 3.6044526752284716,
-      "grad_norm": 0.002440153155475855,
-      "learning_rate": 0.00019999359487070004,
-      "loss": 46.0,
-      "step": 22382
-    },
-    {
-      "epoch": 3.604613712307259,
-      "grad_norm": 0.0034764546435326338,
-      "learning_rate": 0.00019999359429809105,
-      "loss": 46.0,
-      "step": 22383
-    },
-    {
-      "epoch": 3.604774749386046,
-      "grad_norm": 0.0025229365564882755,
-      "learning_rate": 0.00019999359372545648,
-      "loss": 46.0,
-      "step": 22384
-    },
-    {
-      "epoch": 3.6049357864648335,
-      "grad_norm": 0.0026008551940321922,
-      "learning_rate": 0.0001999935931527963,
-      "loss": 46.0,
-      "step": 22385
-    },
-    {
-      "epoch": 3.605096823543621,
-      "grad_norm": 0.0062901414930820465,
-      "learning_rate": 0.00019999359258011053,
-      "loss": 46.0,
-      "step": 22386
-    },
-    {
-      "epoch": 3.6052578606224084,
-      "grad_norm": 0.004310450050979853,
-      "learning_rate": 0.0001999935920073992,
-      "loss": 46.0,
-      "step": 22387
-    },
-    {
-      "epoch": 3.605418897701196,
-      "grad_norm": 0.007083808537572622,
-      "learning_rate": 0.00019999359143466227,
-      "loss": 46.0,
-      "step": 22388
-    },
-    {
-      "epoch": 3.605579934779983,
-      "grad_norm": 0.0016448672395199537,
-      "learning_rate": 0.0001999935908618997,
-      "loss": 46.0,
-      "step": 22389
-    },
-    {
-      "epoch": 3.6057409718587703,
-      "grad_norm": 0.0026761542540043592,
-      "learning_rate": 0.00019999359028911157,
-      "loss": 46.0,
-      "step": 22390
-    },
-    {
-      "epoch": 3.6059020089375577,
-      "grad_norm": 0.0023413097951561213,
-      "learning_rate": 0.00019999358971629786,
-      "loss": 46.0,
-      "step": 22391
-    },
-    {
-      "epoch": 3.606063046016345,
-      "grad_norm": 0.00468074344098568,
-      "learning_rate": 0.00019999358914345853,
-      "loss": 46.0,
-      "step": 22392
-    },
-    {
-      "epoch": 3.6062240830951326,
-      "grad_norm": 0.004899135325103998,
-      "learning_rate": 0.0001999935885705936,
-      "loss": 46.0,
-      "step": 22393
-    },
-    {
-      "epoch": 3.60638512017392,
-      "grad_norm": 0.001748733571730554,
-      "learning_rate": 0.0001999935879977031,
-      "loss": 46.0,
-      "step": 22394
-    },
-    {
-      "epoch": 3.6065461572527076,
-      "grad_norm": 0.0028976218309253454,
-      "learning_rate": 0.000199993587424787,
-      "loss": 46.0,
-      "step": 22395
-    },
-    {
-      "epoch": 3.606707194331495,
-      "grad_norm": 0.001199536956846714,
-      "learning_rate": 0.0001999935868518453,
-      "loss": 46.0,
-      "step": 22396
-    },
-    {
-      "epoch": 3.6068682314102825,
-      "grad_norm": 0.003271740395575762,
-      "learning_rate": 0.00019999358627887802,
-      "loss": 46.0,
-      "step": 22397
-    },
-    {
-      "epoch": 3.6070292684890695,
-      "grad_norm": 0.002882123226299882,
-      "learning_rate": 0.00019999358570588514,
-      "loss": 46.0,
-      "step": 22398
-    },
-    {
-      "epoch": 3.607190305567857,
-      "grad_norm": 0.0012152903946116567,
-      "learning_rate": 0.00019999358513286665,
-      "loss": 46.0,
-      "step": 22399
-    },
-    {
-      "epoch": 3.6073513426466444,
-      "grad_norm": 0.0051603373140096664,
-      "learning_rate": 0.0001999935845598226,
-      "loss": 46.0,
-      "step": 22400
-    },
-    {
-      "epoch": 3.607512379725432,
-      "grad_norm": 0.0010274507803842425,
-      "learning_rate": 0.00019999358398675295,
-      "loss": 46.0,
-      "step": 22401
-    },
-    {
-      "epoch": 3.6076734168042193,
-      "grad_norm": 0.0016115158796310425,
-      "learning_rate": 0.0001999935834136577,
-      "loss": 46.0,
-      "step": 22402
-    },
-    {
-      "epoch": 3.6078344538830067,
-      "grad_norm": 0.0038267478812485933,
-      "learning_rate": 0.00019999358284053685,
-      "loss": 46.0,
-      "step": 22403
-    },
-    {
-      "epoch": 3.6079954909617937,
-      "grad_norm": 0.0030010014306753874,
-      "learning_rate": 0.0001999935822673904,
-      "loss": 46.0,
-      "step": 22404
-    },
-    {
-      "epoch": 3.608156528040581,
-      "grad_norm": 0.0015449613565579057,
-      "learning_rate": 0.00019999358169421838,
-      "loss": 46.0,
-      "step": 22405
-    },
-    {
-      "epoch": 3.6083175651193686,
-      "grad_norm": 0.0016683997819200158,
-      "learning_rate": 0.00019999358112102072,
-      "loss": 46.0,
-      "step": 22406
-    },
-    {
-      "epoch": 3.608478602198156,
-      "grad_norm": 0.0031114269513636827,
-      "learning_rate": 0.00019999358054779753,
-      "loss": 46.0,
-      "step": 22407
-    },
-    {
-      "epoch": 3.6086396392769435,
-      "grad_norm": 0.0033562202006578445,
-      "learning_rate": 0.0001999935799745487,
-      "loss": 46.0,
-      "step": 22408
-    },
-    {
-      "epoch": 3.608800676355731,
-      "grad_norm": 0.0034233522601425648,
-      "learning_rate": 0.0001999935794012743,
-      "loss": 46.0,
-      "step": 22409
-    },
-    {
-      "epoch": 3.6089617134345184,
-      "grad_norm": 0.007132889237254858,
-      "learning_rate": 0.0001999935788279743,
-      "loss": 46.0,
-      "step": 22410
-    },
-    {
-      "epoch": 3.609122750513306,
-      "grad_norm": 0.0042678555473685265,
-      "learning_rate": 0.00019999357825464873,
-      "loss": 46.0,
-      "step": 22411
-    },
-    {
-      "epoch": 3.6092837875920933,
-      "grad_norm": 0.00155303452629596,
-      "learning_rate": 0.00019999357768129753,
-      "loss": 46.0,
-      "step": 22412
-    },
-    {
-      "epoch": 3.6094448246708803,
-      "grad_norm": 0.0023707281798124313,
-      "learning_rate": 0.00019999357710792076,
-      "loss": 46.0,
-      "step": 22413
-    },
-    {
-      "epoch": 3.609605861749668,
-      "grad_norm": 0.0056295557878911495,
-      "learning_rate": 0.0001999935765345184,
-      "loss": 46.0,
-      "step": 22414
-    },
-    {
-      "epoch": 3.6097668988284553,
-      "grad_norm": 0.013550247065722942,
-      "learning_rate": 0.00019999357596109043,
-      "loss": 46.0,
-      "step": 22415
-    },
-    {
-      "epoch": 3.6099279359072427,
-      "grad_norm": 0.007372173015028238,
-      "learning_rate": 0.00019999357538763688,
-      "loss": 46.0,
-      "step": 22416
-    },
-    {
-      "epoch": 3.61008897298603,
-      "grad_norm": 0.001707296003587544,
-      "learning_rate": 0.0001999935748141577,
-      "loss": 46.0,
-      "step": 22417
-    },
-    {
-      "epoch": 3.610250010064817,
-      "grad_norm": 0.004992510657757521,
-      "learning_rate": 0.00019999357424065297,
-      "loss": 46.0,
-      "step": 22418
-    },
-    {
-      "epoch": 3.6104110471436046,
-      "grad_norm": 0.009980971924960613,
-      "learning_rate": 0.00019999357366712263,
-      "loss": 46.0,
-      "step": 22419
-    },
-    {
-      "epoch": 3.610572084222392,
-      "grad_norm": 0.00764093641191721,
-      "learning_rate": 0.0001999935730935667,
-      "loss": 46.0,
-      "step": 22420
-    },
-    {
-      "epoch": 3.6107331213011795,
-      "grad_norm": 0.002733397763222456,
-      "learning_rate": 0.00019999357251998518,
-      "loss": 46.0,
-      "step": 22421
-    },
-    {
-      "epoch": 3.610894158379967,
-      "grad_norm": 0.0007677925750613213,
-      "learning_rate": 0.00019999357194637807,
-      "loss": 46.0,
-      "step": 22422
-    },
-    {
-      "epoch": 3.6110551954587544,
-      "grad_norm": 0.005811657290905714,
-      "learning_rate": 0.00019999357137274535,
-      "loss": 46.0,
-      "step": 22423
-    },
-    {
-      "epoch": 3.611216232537542,
-      "grad_norm": 0.0019189569866284728,
-      "learning_rate": 0.00019999357079908704,
-      "loss": 46.0,
-      "step": 22424
-    },
-    {
-      "epoch": 3.6113772696163293,
-      "grad_norm": 0.007289526518434286,
-      "learning_rate": 0.00019999357022540314,
-      "loss": 46.0,
-      "step": 22425
-    },
-    {
-      "epoch": 3.6115383066951168,
-      "grad_norm": 0.0019161478849127889,
-      "learning_rate": 0.00019999356965169364,
-      "loss": 46.0,
-      "step": 22426
-    },
-    {
-      "epoch": 3.611699343773904,
-      "grad_norm": 0.0037309550680220127,
-      "learning_rate": 0.0001999935690779586,
-      "loss": 46.0,
-      "step": 22427
-    },
-    {
-      "epoch": 3.6118603808526912,
-      "grad_norm": 0.001906686113215983,
-      "learning_rate": 0.0001999935685041979,
-      "loss": 46.0,
-      "step": 22428
-    },
-    {
-      "epoch": 3.6120214179314787,
-      "grad_norm": 0.0019001959590241313,
-      "learning_rate": 0.00019999356793041164,
-      "loss": 46.0,
-      "step": 22429
-    },
-    {
-      "epoch": 3.612182455010266,
-      "grad_norm": 0.003228002693504095,
-      "learning_rate": 0.00019999356735659975,
-      "loss": 46.0,
-      "step": 22430
-    },
-    {
-      "epoch": 3.6123434920890536,
-      "grad_norm": 0.01450062170624733,
-      "learning_rate": 0.0001999935667827623,
-      "loss": 46.0,
-      "step": 22431
-    },
-    {
-      "epoch": 3.612504529167841,
-      "grad_norm": 0.005800783168524504,
-      "learning_rate": 0.00019999356620889927,
-      "loss": 46.0,
-      "step": 22432
-    },
-    {
-      "epoch": 3.612665566246628,
-      "grad_norm": 0.005044400691986084,
-      "learning_rate": 0.00019999356563501063,
-      "loss": 46.0,
-      "step": 22433
-    },
-    {
-      "epoch": 3.6128266033254155,
-      "grad_norm": 0.00861057173460722,
-      "learning_rate": 0.0001999935650610964,
-      "loss": 46.0,
-      "step": 22434
-    },
-    {
-      "epoch": 3.612987640404203,
-      "grad_norm": 0.008550865575671196,
-      "learning_rate": 0.00019999356448715654,
-      "loss": 46.0,
-      "step": 22435
-    },
-    {
-      "epoch": 3.6131486774829904,
-      "grad_norm": 0.010306457057595253,
-      "learning_rate": 0.00019999356391319113,
-      "loss": 46.0,
-      "step": 22436
-    },
-    {
-      "epoch": 3.613309714561778,
-      "grad_norm": 0.007211598102003336,
-      "learning_rate": 0.0001999935633392001,
-      "loss": 46.0,
-      "step": 22437
-    },
-    {
-      "epoch": 3.6134707516405653,
-      "grad_norm": 0.007781315129250288,
-      "learning_rate": 0.0001999935627651835,
-      "loss": 46.0,
-      "step": 22438
-    },
-    {
-      "epoch": 3.6136317887193528,
-      "grad_norm": 0.0017539658583700657,
-      "learning_rate": 0.0001999935621911413,
-      "loss": 46.0,
-      "step": 22439
-    },
-    {
-      "epoch": 3.61379282579814,
-      "grad_norm": 0.003925439901649952,
-      "learning_rate": 0.00019999356161707352,
-      "loss": 46.0,
-      "step": 22440
-    },
-    {
-      "epoch": 3.6139538628769277,
-      "grad_norm": 0.0048811486922204494,
-      "learning_rate": 0.00019999356104298012,
-      "loss": 46.0,
-      "step": 22441
-    },
-    {
-      "epoch": 3.6141148999557147,
-      "grad_norm": 0.0011200091103091836,
-      "learning_rate": 0.00019999356046886116,
-      "loss": 46.0,
-      "step": 22442
-    },
-    {
-      "epoch": 3.614275937034502,
-      "grad_norm": 0.0029285536147654057,
-      "learning_rate": 0.00019999355989471656,
-      "loss": 46.0,
-      "step": 22443
-    },
-    {
-      "epoch": 3.6144369741132896,
-      "grad_norm": 0.0019486559322103858,
-      "learning_rate": 0.0001999935593205464,
-      "loss": 46.0,
-      "step": 22444
-    },
-    {
-      "epoch": 3.614598011192077,
-      "grad_norm": 0.0021144987549632788,
-      "learning_rate": 0.00019999355874635063,
-      "loss": 46.0,
-      "step": 22445
-    },
-    {
-      "epoch": 3.6147590482708645,
-      "grad_norm": 0.0012204087106510997,
-      "learning_rate": 0.0001999935581721293,
-      "loss": 46.0,
-      "step": 22446
-    },
-    {
-      "epoch": 3.614920085349652,
-      "grad_norm": 0.008844881318509579,
-      "learning_rate": 0.00019999355759788232,
-      "loss": 46.0,
-      "step": 22447
-    },
-    {
-      "epoch": 3.615081122428439,
-      "grad_norm": 0.0020680285524576902,
-      "learning_rate": 0.00019999355702360978,
-      "loss": 46.0,
-      "step": 22448
-    },
-    {
-      "epoch": 3.6152421595072264,
-      "grad_norm": 0.004936013836413622,
-      "learning_rate": 0.00019999355644931166,
-      "loss": 46.0,
-      "step": 22449
-    },
-    {
-      "epoch": 3.615403196586014,
-      "grad_norm": 0.0047101108357310295,
-      "learning_rate": 0.00019999355587498792,
-      "loss": 46.0,
-      "step": 22450
-    },
-    {
-      "epoch": 3.6155642336648013,
-      "grad_norm": 0.005580500699579716,
-      "learning_rate": 0.00019999355530063862,
-      "loss": 46.0,
-      "step": 22451
-    },
-    {
-      "epoch": 3.6157252707435887,
-      "grad_norm": 0.007232708390802145,
-      "learning_rate": 0.00019999355472626368,
-      "loss": 46.0,
-      "step": 22452
-    },
-    {
-      "epoch": 3.615886307822376,
-      "grad_norm": 0.005742973182350397,
-      "learning_rate": 0.0001999935541518632,
-      "loss": 46.0,
-      "step": 22453
-    },
-    {
-      "epoch": 3.6160473449011636,
-      "grad_norm": 0.0022650575265288353,
-      "learning_rate": 0.0001999935535774371,
-      "loss": 46.0,
-      "step": 22454
-    },
-    {
-      "epoch": 3.616208381979951,
-      "grad_norm": 0.00304702902212739,
-      "learning_rate": 0.0001999935530029854,
-      "loss": 46.0,
-      "step": 22455
-    },
-    {
-      "epoch": 3.6163694190587385,
-      "grad_norm": 0.0013421442126855254,
-      "learning_rate": 0.0001999935524285081,
-      "loss": 46.0,
-      "step": 22456
-    },
-    {
-      "epoch": 3.6165304561375256,
-      "grad_norm": 0.003503709565848112,
-      "learning_rate": 0.0001999935518540052,
-      "loss": 46.0,
-      "step": 22457
-    },
-    {
-      "epoch": 3.616691493216313,
-      "grad_norm": 0.0025620809756219387,
-      "learning_rate": 0.00019999355127947675,
-      "loss": 46.0,
-      "step": 22458
-    },
-    {
-      "epoch": 3.6168525302951005,
-      "grad_norm": 0.0027202838100492954,
-      "learning_rate": 0.0001999935507049227,
-      "loss": 46.0,
-      "step": 22459
-    },
-    {
-      "epoch": 3.617013567373888,
-      "grad_norm": 0.008707068860530853,
-      "learning_rate": 0.00019999355013034303,
-      "loss": 46.0,
-      "step": 22460
-    },
-    {
-      "epoch": 3.6171746044526754,
-      "grad_norm": 0.003709145588800311,
-      "learning_rate": 0.00019999354955573776,
-      "loss": 46.0,
-      "step": 22461
-    },
-    {
-      "epoch": 3.6173356415314624,
-      "grad_norm": 0.0035873798187822104,
-      "learning_rate": 0.00019999354898110695,
-      "loss": 46.0,
-      "step": 22462
-    },
-    {
-      "epoch": 3.61749667861025,
-      "grad_norm": 0.0026390121784061193,
-      "learning_rate": 0.0001999935484064505,
-      "loss": 46.0,
-      "step": 22463
-    },
-    {
-      "epoch": 3.6176577156890373,
-      "grad_norm": 0.0010557264322414994,
-      "learning_rate": 0.00019999354783176846,
-      "loss": 46.0,
-      "step": 22464
-    },
-    {
-      "epoch": 3.6178187527678247,
-      "grad_norm": 0.0024030692875385284,
-      "learning_rate": 0.00019999354725706083,
-      "loss": 46.0,
-      "step": 22465
-    },
-    {
-      "epoch": 3.617979789846612,
-      "grad_norm": 0.002708001993596554,
-      "learning_rate": 0.00019999354668232762,
-      "loss": 46.0,
-      "step": 22466
-    },
-    {
-      "epoch": 3.6181408269253996,
-      "grad_norm": 0.007245389278978109,
-      "learning_rate": 0.0001999935461075688,
-      "loss": 46.0,
-      "step": 22467
-    },
-    {
-      "epoch": 3.618301864004187,
-      "grad_norm": 0.001983353868126869,
-      "learning_rate": 0.0001999935455327844,
-      "loss": 46.0,
-      "step": 22468
-    },
-    {
-      "epoch": 3.6184629010829745,
-      "grad_norm": 0.013206932693719864,
-      "learning_rate": 0.0001999935449579744,
-      "loss": 46.0,
-      "step": 22469
-    },
-    {
-      "epoch": 3.618623938161762,
-      "grad_norm": 0.004026846028864384,
-      "learning_rate": 0.0001999935443831388,
-      "loss": 46.0,
-      "step": 22470
-    },
-    {
-      "epoch": 3.618784975240549,
-      "grad_norm": 0.0019163753604516387,
-      "learning_rate": 0.0001999935438082776,
-      "loss": 46.0,
-      "step": 22471
-    },
-    {
-      "epoch": 3.6189460123193364,
-      "grad_norm": 0.0031291507184505463,
-      "learning_rate": 0.00019999354323339087,
-      "loss": 46.0,
-      "step": 22472
-    },
-    {
-      "epoch": 3.619107049398124,
-      "grad_norm": 0.0070044719614088535,
-      "learning_rate": 0.00019999354265847846,
-      "loss": 46.0,
-      "step": 22473
-    },
-    {
-      "epoch": 3.6192680864769113,
-      "grad_norm": 0.0052513424307107925,
-      "learning_rate": 0.0001999935420835405,
-      "loss": 46.0,
-      "step": 22474
-    },
-    {
-      "epoch": 3.619429123555699,
-      "grad_norm": 0.0043580965138971806,
-      "learning_rate": 0.00019999354150857695,
-      "loss": 46.0,
-      "step": 22475
-    },
-    {
-      "epoch": 3.6195901606344862,
-      "grad_norm": 0.006770983338356018,
-      "learning_rate": 0.0001999935409335878,
-      "loss": 46.0,
-      "step": 22476
-    },
-    {
-      "epoch": 3.6197511977132732,
-      "grad_norm": 0.005758809391409159,
-      "learning_rate": 0.00019999354035857305,
-      "loss": 46.0,
-      "step": 22477
-    },
-    {
-      "epoch": 3.6199122347920607,
-      "grad_norm": 0.013581417500972748,
-      "learning_rate": 0.0001999935397835327,
-      "loss": 46.0,
-      "step": 22478
-    },
-    {
-      "epoch": 3.620073271870848,
-      "grad_norm": 0.012798821553587914,
-      "learning_rate": 0.00019999353920846678,
-      "loss": 46.0,
-      "step": 22479
-    },
-    {
-      "epoch": 3.6202343089496356,
-      "grad_norm": 0.005479360464960337,
-      "learning_rate": 0.00019999353863337524,
-      "loss": 46.0,
-      "step": 22480
-    },
-    {
-      "epoch": 3.620395346028423,
-      "grad_norm": 0.01105831004679203,
-      "learning_rate": 0.00019999353805825814,
-      "loss": 46.0,
-      "step": 22481
-    },
-    {
-      "epoch": 3.6205563831072105,
-      "grad_norm": 0.004589867778122425,
-      "learning_rate": 0.00019999353748311545,
-      "loss": 46.0,
-      "step": 22482
-    },
-    {
-      "epoch": 3.620717420185998,
-      "grad_norm": 0.006644019857048988,
-      "learning_rate": 0.00019999353690794712,
-      "loss": 46.0,
-      "step": 22483
-    },
-    {
-      "epoch": 3.6208784572647854,
-      "grad_norm": 0.0012000190326943994,
-      "learning_rate": 0.00019999353633275323,
-      "loss": 46.0,
-      "step": 22484
-    },
-    {
-      "epoch": 3.621039494343573,
-      "grad_norm": 0.0040437402203679085,
-      "learning_rate": 0.00019999353575753375,
-      "loss": 46.0,
-      "step": 22485
-    },
-    {
-      "epoch": 3.62120053142236,
-      "grad_norm": 0.001889971666969359,
-      "learning_rate": 0.00019999353518228868,
-      "loss": 46.0,
-      "step": 22486
-    },
-    {
-      "epoch": 3.6213615685011473,
-      "grad_norm": 0.005469277035444975,
-      "learning_rate": 0.000199993534607018,
-      "loss": 46.0,
-      "step": 22487
-    },
-    {
-      "epoch": 3.6215226055799348,
-      "grad_norm": 0.003779173828661442,
-      "learning_rate": 0.00019999353403172168,
-      "loss": 46.0,
-      "step": 22488
-    },
-    {
-      "epoch": 3.6216836426587222,
-      "grad_norm": 0.004029697738587856,
-      "learning_rate": 0.00019999353345639986,
-      "loss": 46.0,
-      "step": 22489
-    },
-    {
-      "epoch": 3.6218446797375097,
-      "grad_norm": 0.009057889692485332,
-      "learning_rate": 0.00019999353288105242,
-      "loss": 46.0,
-      "step": 22490
-    },
-    {
-      "epoch": 3.6220057168162967,
-      "grad_norm": 0.0011810354189947248,
-      "learning_rate": 0.00019999353230567933,
-      "loss": 46.0,
-      "step": 22491
-    },
-    {
-      "epoch": 3.622166753895084,
-      "grad_norm": 0.010252796113491058,
-      "learning_rate": 0.0001999935317302807,
-      "loss": 46.0,
-      "step": 22492
-    },
-    {
-      "epoch": 3.6223277909738716,
-      "grad_norm": 0.002155910013243556,
-      "learning_rate": 0.00019999353115485646,
-      "loss": 46.0,
-      "step": 22493
-    },
-    {
-      "epoch": 3.622488828052659,
-      "grad_norm": 0.003056129440665245,
-      "learning_rate": 0.00019999353057940664,
-      "loss": 46.0,
-      "step": 22494
-    },
-    {
-      "epoch": 3.6226498651314465,
-      "grad_norm": 0.0030514122918248177,
-      "learning_rate": 0.0001999935300039312,
-      "loss": 46.0,
-      "step": 22495
-    },
-    {
-      "epoch": 3.622810902210234,
-      "grad_norm": 0.00620141951367259,
-      "learning_rate": 0.00019999352942843022,
-      "loss": 46.0,
-      "step": 22496
-    },
-    {
-      "epoch": 3.6229719392890214,
-      "grad_norm": 0.00171517429407686,
-      "learning_rate": 0.0001999935288529036,
-      "loss": 46.0,
-      "step": 22497
-    },
-    {
-      "epoch": 3.623132976367809,
-      "grad_norm": 0.004122231155633926,
-      "learning_rate": 0.0001999935282773514,
-      "loss": 46.0,
-      "step": 22498
-    },
-    {
-      "epoch": 3.6232940134465963,
-      "grad_norm": 0.005840396508574486,
-      "learning_rate": 0.0001999935277017736,
-      "loss": 46.0,
-      "step": 22499
-    },
-    {
-      "epoch": 3.6234550505253833,
-      "grad_norm": 0.01794731430709362,
-      "learning_rate": 0.00019999352712617022,
-      "loss": 46.0,
-      "step": 22500
-    },
-    {
-      "epoch": 3.6236160876041708,
-      "grad_norm": 0.007381750736385584,
-      "learning_rate": 0.00019999352655054121,
-      "loss": 46.0,
-      "step": 22501
-    },
-    {
-      "epoch": 3.623777124682958,
-      "grad_norm": 0.00266893464140594,
-      "learning_rate": 0.00019999352597488665,
-      "loss": 46.0,
-      "step": 22502
-    },
-    {
-      "epoch": 3.6239381617617457,
-      "grad_norm": 0.002578998450189829,
-      "learning_rate": 0.0001999935253992065,
-      "loss": 46.0,
-      "step": 22503
-    },
-    {
-      "epoch": 3.624099198840533,
-      "grad_norm": 0.004135339520871639,
-      "learning_rate": 0.00019999352482350072,
-      "loss": 46.0,
-      "step": 22504
-    },
-    {
-      "epoch": 3.6242602359193206,
-      "grad_norm": 0.005464186891913414,
-      "learning_rate": 0.0001999935242477694,
-      "loss": 46.0,
-      "step": 22505
-    },
-    {
-      "epoch": 3.6244212729981076,
-      "grad_norm": 0.0039016236551105976,
-      "learning_rate": 0.00019999352367201242,
-      "loss": 46.0,
-      "step": 22506
-    },
-    {
-      "epoch": 3.624582310076895,
-      "grad_norm": 0.0026380294002592564,
-      "learning_rate": 0.0001999935230962299,
-      "loss": 46.0,
-      "step": 22507
-    },
-    {
-      "epoch": 3.6247433471556825,
-      "grad_norm": 0.0013833181001245975,
-      "learning_rate": 0.00019999352252042174,
-      "loss": 46.0,
-      "step": 22508
-    },
-    {
-      "epoch": 3.62490438423447,
-      "grad_norm": 0.005701172631233931,
-      "learning_rate": 0.000199993521944588,
-      "loss": 46.0,
-      "step": 22509
-    },
-    {
-      "epoch": 3.6250654213132574,
-      "grad_norm": 0.0013801305321976542,
-      "learning_rate": 0.0001999935213687287,
-      "loss": 46.0,
-      "step": 22510
-    },
-    {
-      "epoch": 3.625226458392045,
-      "grad_norm": 0.0017217963468283415,
-      "learning_rate": 0.00019999352079284378,
-      "loss": 46.0,
-      "step": 22511
-    },
-    {
-      "epoch": 3.6253874954708323,
-      "grad_norm": 0.0034533457364887,
-      "learning_rate": 0.00019999352021693326,
-      "loss": 46.0,
-      "step": 22512
-    },
-    {
-      "epoch": 3.6255485325496197,
-      "grad_norm": 0.003828912042081356,
-      "learning_rate": 0.00019999351964099718,
-      "loss": 46.0,
-      "step": 22513
-    },
-    {
-      "epoch": 3.625709569628407,
-      "grad_norm": 0.0025549596175551414,
-      "learning_rate": 0.00019999351906503548,
-      "loss": 46.0,
-      "step": 22514
-    },
-    {
-      "epoch": 3.625870606707194,
-      "grad_norm": 0.013700297102332115,
-      "learning_rate": 0.0001999935184890482,
-      "loss": 46.0,
-      "step": 22515
-    },
-    {
-      "epoch": 3.6260316437859816,
-      "grad_norm": 0.002976646414026618,
-      "learning_rate": 0.0001999935179130353,
-      "loss": 46.0,
-      "step": 22516
-    },
-    {
-      "epoch": 3.626192680864769,
-      "grad_norm": 0.004166321363300085,
-      "learning_rate": 0.00019999351733699684,
-      "loss": 46.0,
-      "step": 22517
-    },
-    {
-      "epoch": 3.6263537179435565,
-      "grad_norm": 0.004388280212879181,
-      "learning_rate": 0.00019999351676093277,
-      "loss": 46.0,
-      "step": 22518
-    },
-    {
-      "epoch": 3.626514755022344,
-      "grad_norm": 0.01295491773635149,
-      "learning_rate": 0.00019999351618484311,
-      "loss": 46.0,
-      "step": 22519
-    },
-    {
-      "epoch": 3.6266757921011314,
-      "grad_norm": 0.0008753619622439146,
-      "learning_rate": 0.00019999351560872787,
-      "loss": 46.0,
-      "step": 22520
-    },
-    {
-      "epoch": 3.6268368291799185,
-      "grad_norm": 0.006063347682356834,
-      "learning_rate": 0.00019999351503258703,
-      "loss": 46.0,
-      "step": 22521
-    },
-    {
-      "epoch": 3.626997866258706,
-      "grad_norm": 0.002712126588448882,
-      "learning_rate": 0.0001999935144564206,
-      "loss": 46.0,
-      "step": 22522
-    },
-    {
-      "epoch": 3.6271589033374934,
-      "grad_norm": 0.0017489992314949632,
-      "learning_rate": 0.00019999351388022855,
-      "loss": 46.0,
-      "step": 22523
-    },
-    {
-      "epoch": 3.627319940416281,
-      "grad_norm": 0.020075609907507896,
-      "learning_rate": 0.00019999351330401093,
-      "loss": 46.0,
-      "step": 22524
-    },
-    {
-      "epoch": 3.6274809774950683,
-      "grad_norm": 0.008219759911298752,
-      "learning_rate": 0.00019999351272776772,
-      "loss": 46.0,
-      "step": 22525
-    },
-    {
-      "epoch": 3.6276420145738557,
-      "grad_norm": 0.0033160836901515722,
-      "learning_rate": 0.0001999935121514989,
-      "loss": 46.0,
-      "step": 22526
-    },
-    {
-      "epoch": 3.627803051652643,
-      "grad_norm": 0.002313353819772601,
-      "learning_rate": 0.00019999351157520446,
-      "loss": 46.0,
-      "step": 22527
-    },
-    {
-      "epoch": 3.6279640887314306,
-      "grad_norm": 0.0032709252554923296,
-      "learning_rate": 0.0001999935109988845,
-      "loss": 46.0,
-      "step": 22528
-    },
-    {
-      "epoch": 3.628125125810218,
-      "grad_norm": 0.0021192855201661587,
-      "learning_rate": 0.0001999935104225389,
-      "loss": 46.0,
-      "step": 22529
-    },
-    {
-      "epoch": 3.628286162889005,
-      "grad_norm": 0.0017459650989621878,
-      "learning_rate": 0.0001999935098461677,
-      "loss": 46.0,
-      "step": 22530
-    },
-    {
-      "epoch": 3.6284471999677925,
-      "grad_norm": 0.0024031877983361483,
-      "learning_rate": 0.00019999350926977095,
-      "loss": 46.0,
-      "step": 22531
-    },
-    {
-      "epoch": 3.62860823704658,
-      "grad_norm": 0.005624727811664343,
-      "learning_rate": 0.00019999350869334857,
-      "loss": 46.0,
-      "step": 22532
-    },
-    {
-      "epoch": 3.6287692741253674,
-      "grad_norm": 0.0025609242729842663,
-      "learning_rate": 0.0001999935081169006,
-      "loss": 46.0,
-      "step": 22533
-    },
-    {
-      "epoch": 3.628930311204155,
-      "grad_norm": 0.0017392337322235107,
-      "learning_rate": 0.00019999350754042704,
-      "loss": 46.0,
-      "step": 22534
-    },
-    {
-      "epoch": 3.629091348282942,
-      "grad_norm": 0.00645176088437438,
-      "learning_rate": 0.00019999350696392788,
-      "loss": 46.0,
-      "step": 22535
-    },
-    {
-      "epoch": 3.6292523853617293,
-      "grad_norm": 0.0018682080553844571,
-      "learning_rate": 0.00019999350638740315,
-      "loss": 46.0,
-      "step": 22536
-    },
-    {
-      "epoch": 3.629413422440517,
-      "grad_norm": 0.004116647411137819,
-      "learning_rate": 0.00019999350581085282,
-      "loss": 46.0,
-      "step": 22537
-    },
-    {
-      "epoch": 3.6295744595193042,
-      "grad_norm": 0.001600995077751577,
-      "learning_rate": 0.00019999350523427687,
-      "loss": 46.0,
-      "step": 22538
-    },
-    {
-      "epoch": 3.6297354965980917,
-      "grad_norm": 0.0016844253987073898,
-      "learning_rate": 0.00019999350465767536,
-      "loss": 46.0,
-      "step": 22539
-    },
-    {
-      "epoch": 3.629896533676879,
-      "grad_norm": 0.0014293370768427849,
-      "learning_rate": 0.00019999350408104823,
-      "loss": 46.0,
-      "step": 22540
-    },
-    {
-      "epoch": 3.6300575707556666,
-      "grad_norm": 0.00366139505058527,
-      "learning_rate": 0.00019999350350439552,
-      "loss": 46.0,
-      "step": 22541
-    },
-    {
-      "epoch": 3.630218607834454,
-      "grad_norm": 0.004743609577417374,
-      "learning_rate": 0.00019999350292771725,
-      "loss": 46.0,
-      "step": 22542
-    },
-    {
-      "epoch": 3.6303796449132415,
-      "grad_norm": 0.0012500014854595065,
-      "learning_rate": 0.00019999350235101333,
-      "loss": 46.0,
-      "step": 22543
-    },
-    {
-      "epoch": 3.6305406819920285,
-      "grad_norm": 0.002645831787958741,
-      "learning_rate": 0.00019999350177428383,
-      "loss": 46.0,
-      "step": 22544
-    },
-    {
-      "epoch": 3.630701719070816,
-      "grad_norm": 0.0033212376292794943,
-      "learning_rate": 0.00019999350119752877,
-      "loss": 46.0,
-      "step": 22545
-    },
-    {
-      "epoch": 3.6308627561496034,
-      "grad_norm": 0.0028429566882550716,
-      "learning_rate": 0.0001999935006207481,
-      "loss": 46.0,
-      "step": 22546
-    },
-    {
-      "epoch": 3.631023793228391,
-      "grad_norm": 0.0028267125599086285,
-      "learning_rate": 0.0001999935000439418,
-      "loss": 46.0,
-      "step": 22547
-    },
-    {
-      "epoch": 3.6311848303071783,
-      "grad_norm": 0.0005956628010608256,
-      "learning_rate": 0.00019999349946710996,
-      "loss": 46.0,
-      "step": 22548
-    },
-    {
-      "epoch": 3.6313458673859658,
-      "grad_norm": 0.005481266416609287,
-      "learning_rate": 0.0001999934988902525,
-      "loss": 46.0,
-      "step": 22549
-    },
-    {
-      "epoch": 3.6315069044647528,
-      "grad_norm": 0.000716834794729948,
-      "learning_rate": 0.00019999349831336947,
-      "loss": 46.0,
-      "step": 22550
-    },
-    {
-      "epoch": 3.6316679415435402,
-      "grad_norm": 0.0015818835236132145,
-      "learning_rate": 0.0001999934977364608,
-      "loss": 46.0,
-      "step": 22551
-    },
-    {
-      "epoch": 3.6318289786223277,
-      "grad_norm": 0.0025429250672459602,
-      "learning_rate": 0.00019999349715952658,
-      "loss": 46.0,
-      "step": 22552
-    },
-    {
-      "epoch": 3.631990015701115,
-      "grad_norm": 0.003357258625328541,
-      "learning_rate": 0.00019999349658256674,
-      "loss": 46.0,
-      "step": 22553
-    },
-    {
-      "epoch": 3.6321510527799026,
-      "grad_norm": 0.0018024045275524259,
-      "learning_rate": 0.00019999349600558131,
-      "loss": 46.0,
-      "step": 22554
-    },
-    {
-      "epoch": 3.63231208985869,
-      "grad_norm": 0.0032567300368100405,
-      "learning_rate": 0.0001999934954285703,
-      "loss": 46.0,
-      "step": 22555
-    },
-    {
-      "epoch": 3.6324731269374775,
-      "grad_norm": 0.005689000245183706,
-      "learning_rate": 0.00019999349485153373,
-      "loss": 46.0,
-      "step": 22556
-    },
-    {
-      "epoch": 3.632634164016265,
-      "grad_norm": 0.0036533710081130266,
-      "learning_rate": 0.00019999349427447148,
-      "loss": 46.0,
-      "step": 22557
-    },
-    {
-      "epoch": 3.6327952010950524,
-      "grad_norm": 0.000923704297747463,
-      "learning_rate": 0.00019999349369738368,
-      "loss": 46.0,
-      "step": 22558
-    },
-    {
-      "epoch": 3.6329562381738394,
-      "grad_norm": 0.0011604356113821268,
-      "learning_rate": 0.00019999349312027032,
-      "loss": 46.0,
-      "step": 22559
-    },
-    {
-      "epoch": 3.633117275252627,
-      "grad_norm": 0.008964471518993378,
-      "learning_rate": 0.00019999349254313134,
-      "loss": 46.0,
-      "step": 22560
-    },
-    {
-      "epoch": 3.6332783123314143,
-      "grad_norm": 0.000986400875262916,
-      "learning_rate": 0.00019999349196596678,
-      "loss": 46.0,
-      "step": 22561
-    },
-    {
-      "epoch": 3.6334393494102017,
-      "grad_norm": 0.0027297590859234333,
-      "learning_rate": 0.0001999934913887766,
-      "loss": 46.0,
-      "step": 22562
-    },
-    {
-      "epoch": 3.633600386488989,
-      "grad_norm": 0.001780897960998118,
-      "learning_rate": 0.00019999349081156084,
-      "loss": 46.0,
-      "step": 22563
-    },
-    {
-      "epoch": 3.633761423567776,
-      "grad_norm": 0.0015038541750982404,
-      "learning_rate": 0.00019999349023431948,
-      "loss": 46.0,
-      "step": 22564
-    },
-    {
-      "epoch": 3.6339224606465637,
-      "grad_norm": 0.00709498580545187,
-      "learning_rate": 0.00019999348965705254,
-      "loss": 46.0,
-      "step": 22565
-    },
-    {
-      "epoch": 3.634083497725351,
-      "grad_norm": 0.0028029889799654484,
-      "learning_rate": 0.00019999348907976,
-      "loss": 46.0,
-      "step": 22566
-    },
-    {
-      "epoch": 3.6342445348041386,
-      "grad_norm": 0.001673622871749103,
-      "learning_rate": 0.00019999348850244187,
-      "loss": 46.0,
-      "step": 22567
-    },
-    {
-      "epoch": 3.634405571882926,
-      "grad_norm": 0.0035601984709501266,
-      "learning_rate": 0.00019999348792509815,
-      "loss": 46.0,
-      "step": 22568
-    },
-    {
-      "epoch": 3.6345666089617135,
-      "grad_norm": 0.0008578160195611417,
-      "learning_rate": 0.0001999934873477288,
-      "loss": 46.0,
-      "step": 22569
-    },
-    {
-      "epoch": 3.634727646040501,
-      "grad_norm": 0.0053849415853619576,
-      "learning_rate": 0.00019999348677033393,
-      "loss": 46.0,
-      "step": 22570
-    },
-    {
-      "epoch": 3.6348886831192884,
-      "grad_norm": 0.003762367647141218,
-      "learning_rate": 0.0001999934861929134,
-      "loss": 46.0,
-      "step": 22571
-    },
-    {
-      "epoch": 3.635049720198076,
-      "grad_norm": 0.002221018308773637,
-      "learning_rate": 0.0001999934856154673,
-      "loss": 46.0,
-      "step": 22572
-    },
-    {
-      "epoch": 3.6352107572768633,
-      "grad_norm": 0.0036930839996784925,
-      "learning_rate": 0.0001999934850379956,
-      "loss": 46.0,
-      "step": 22573
-    },
-    {
-      "epoch": 3.6353717943556503,
-      "grad_norm": 0.003176774363964796,
-      "learning_rate": 0.00019999348446049834,
-      "loss": 46.0,
-      "step": 22574
-    },
-    {
-      "epoch": 3.6355328314344377,
-      "grad_norm": 0.004482978954911232,
-      "learning_rate": 0.00019999348388297542,
-      "loss": 46.0,
-      "step": 22575
-    },
-    {
-      "epoch": 3.635693868513225,
-      "grad_norm": 0.003068018238991499,
-      "learning_rate": 0.00019999348330542696,
-      "loss": 46.0,
-      "step": 22576
-    },
-    {
-      "epoch": 3.6358549055920126,
-      "grad_norm": 0.0014689676463603973,
-      "learning_rate": 0.0001999934827278529,
-      "loss": 46.0,
-      "step": 22577
-    },
-    {
-      "epoch": 3.6360159426708,
-      "grad_norm": 0.0028130931314080954,
-      "learning_rate": 0.00019999348215025324,
-      "loss": 46.0,
-      "step": 22578
-    },
-    {
-      "epoch": 3.636176979749587,
-      "grad_norm": 0.0009816274978220463,
-      "learning_rate": 0.000199993481572628,
-      "loss": 46.0,
-      "step": 22579
-    },
-    {
-      "epoch": 3.6363380168283745,
-      "grad_norm": 0.0030423817224800587,
-      "learning_rate": 0.00019999348099497715,
-      "loss": 46.0,
-      "step": 22580
-    },
-    {
-      "epoch": 3.636499053907162,
-      "grad_norm": 0.01929938606917858,
-      "learning_rate": 0.0001999934804173007,
-      "loss": 46.0,
-      "step": 22581
-    },
-    {
-      "epoch": 3.6366600909859494,
-      "grad_norm": 0.0015261085936799645,
-      "learning_rate": 0.00019999347983959867,
-      "loss": 46.0,
-      "step": 22582
-    },
-    {
-      "epoch": 3.636821128064737,
-      "grad_norm": 0.0012616056483238935,
-      "learning_rate": 0.00019999347926187106,
-      "loss": 46.0,
-      "step": 22583
-    },
-    {
-      "epoch": 3.6369821651435243,
-      "grad_norm": 0.006738286931067705,
-      "learning_rate": 0.00019999347868411785,
-      "loss": 46.0,
-      "step": 22584
-    },
-    {
-      "epoch": 3.637143202222312,
-      "grad_norm": 0.004843592178076506,
-      "learning_rate": 0.000199993478106339,
-      "loss": 46.0,
-      "step": 22585
-    },
-    {
-      "epoch": 3.6373042393010993,
-      "grad_norm": 0.005053396802395582,
-      "learning_rate": 0.00019999347752853463,
-      "loss": 46.0,
-      "step": 22586
-    },
-    {
-      "epoch": 3.6374652763798867,
-      "grad_norm": 0.0014434074983000755,
-      "learning_rate": 0.00019999347695070463,
-      "loss": 46.0,
-      "step": 22587
-    },
-    {
-      "epoch": 3.6376263134586737,
-      "grad_norm": 0.0031371042132377625,
-      "learning_rate": 0.00019999347637284905,
-      "loss": 46.0,
-      "step": 22588
-    },
-    {
-      "epoch": 3.637787350537461,
-      "grad_norm": 0.004133532755076885,
-      "learning_rate": 0.00019999347579496786,
-      "loss": 46.0,
-      "step": 22589
-    },
-    {
-      "epoch": 3.6379483876162486,
-      "grad_norm": 0.008824850432574749,
-      "learning_rate": 0.00019999347521706108,
-      "loss": 46.0,
-      "step": 22590
-    },
-    {
-      "epoch": 3.638109424695036,
-      "grad_norm": 0.00482999486848712,
-      "learning_rate": 0.0001999934746391287,
-      "loss": 46.0,
-      "step": 22591
-    },
-    {
-      "epoch": 3.6382704617738235,
-      "grad_norm": 0.004027730785310268,
-      "learning_rate": 0.00019999347406117072,
-      "loss": 46.0,
-      "step": 22592
-    },
-    {
-      "epoch": 3.638431498852611,
-      "grad_norm": 0.005034802481532097,
-      "learning_rate": 0.0001999934734831872,
-      "loss": 46.0,
-      "step": 22593
-    },
-    {
-      "epoch": 3.638592535931398,
-      "grad_norm": 0.0011818541679531336,
-      "learning_rate": 0.00019999347290517805,
-      "loss": 46.0,
-      "step": 22594
-    },
-    {
-      "epoch": 3.6387535730101854,
-      "grad_norm": 0.001119572902098298,
-      "learning_rate": 0.0001999934723271433,
-      "loss": 46.0,
-      "step": 22595
-    },
-    {
-      "epoch": 3.638914610088973,
-      "grad_norm": 0.0011435940396040678,
-      "learning_rate": 0.00019999347174908295,
-      "loss": 46.0,
-      "step": 22596
-    },
-    {
-      "epoch": 3.6390756471677603,
-      "grad_norm": 0.026443036273121834,
-      "learning_rate": 0.00019999347117099703,
-      "loss": 46.0,
-      "step": 22597
-    },
-    {
-      "epoch": 3.639236684246548,
-      "grad_norm": 0.007674022577702999,
-      "learning_rate": 0.00019999347059288552,
-      "loss": 46.0,
-      "step": 22598
-    },
-    {
-      "epoch": 3.6393977213253352,
-      "grad_norm": 0.0014888565056025982,
-      "learning_rate": 0.0001999934700147484,
-      "loss": 46.0,
-      "step": 22599
-    },
-    {
-      "epoch": 3.6395587584041227,
-      "grad_norm": 0.00165718758944422,
-      "learning_rate": 0.00019999346943658567,
-      "loss": 46.0,
-      "step": 22600
-    },
-    {
-      "epoch": 3.63971979548291,
-      "grad_norm": 0.003740053391084075,
-      "learning_rate": 0.00019999346885839737,
-      "loss": 46.0,
-      "step": 22601
-    },
-    {
-      "epoch": 3.6398808325616976,
-      "grad_norm": 0.01177155040204525,
-      "learning_rate": 0.0001999934682801835,
-      "loss": 46.0,
-      "step": 22602
-    },
-    {
-      "epoch": 3.6400418696404846,
-      "grad_norm": 0.0014776635216549039,
-      "learning_rate": 0.000199993467701944,
-      "loss": 46.0,
-      "step": 22603
-    },
-    {
-      "epoch": 3.640202906719272,
-      "grad_norm": 0.021222947165369987,
-      "learning_rate": 0.0001999934671236789,
-      "loss": 46.0,
-      "step": 22604
-    },
-    {
-      "epoch": 3.6403639437980595,
-      "grad_norm": 0.006313573103398085,
-      "learning_rate": 0.00019999346654538824,
-      "loss": 46.0,
-      "step": 22605
-    },
-    {
-      "epoch": 3.640524980876847,
-      "grad_norm": 0.014756487682461739,
-      "learning_rate": 0.00019999346596707196,
-      "loss": 46.0,
-      "step": 22606
-    },
-    {
-      "epoch": 3.6406860179556344,
-      "grad_norm": 0.0023969586472958326,
-      "learning_rate": 0.0001999934653887301,
-      "loss": 46.0,
-      "step": 22607
-    },
-    {
-      "epoch": 3.6408470550344214,
-      "grad_norm": 0.004246568772941828,
-      "learning_rate": 0.00019999346481036265,
-      "loss": 46.0,
-      "step": 22608
-    },
-    {
-      "epoch": 3.641008092113209,
-      "grad_norm": 0.002121901838108897,
-      "learning_rate": 0.0001999934642319696,
-      "loss": 46.0,
-      "step": 22609
-    },
-    {
-      "epoch": 3.6411691291919963,
-      "grad_norm": 0.0021776400972157717,
-      "learning_rate": 0.00019999346365355097,
-      "loss": 46.0,
-      "step": 22610
-    },
-    {
-      "epoch": 3.6413301662707838,
-      "grad_norm": 0.00777210108935833,
-      "learning_rate": 0.00019999346307510672,
-      "loss": 46.0,
-      "step": 22611
-    },
-    {
-      "epoch": 3.641491203349571,
-      "grad_norm": 0.0012758438242599368,
-      "learning_rate": 0.00019999346249663692,
-      "loss": 46.0,
-      "step": 22612
-    },
-    {
-      "epoch": 3.6416522404283587,
-      "grad_norm": 0.018482966348528862,
-      "learning_rate": 0.00019999346191814147,
-      "loss": 46.0,
-      "step": 22613
-    },
-    {
-      "epoch": 3.641813277507146,
-      "grad_norm": 0.0029913706239312887,
-      "learning_rate": 0.00019999346133962046,
-      "loss": 46.0,
-      "step": 22614
-    },
-    {
-      "epoch": 3.6419743145859336,
-      "grad_norm": 0.008060762658715248,
-      "learning_rate": 0.0001999934607610739,
-      "loss": 46.0,
-      "step": 22615
-    },
-    {
-      "epoch": 3.642135351664721,
-      "grad_norm": 0.003985735587775707,
-      "learning_rate": 0.00019999346018250168,
-      "loss": 46.0,
-      "step": 22616
-    },
-    {
-      "epoch": 3.642296388743508,
-      "grad_norm": 0.012810556218028069,
-      "learning_rate": 0.00019999345960390388,
-      "loss": 46.0,
-      "step": 22617
-    },
-    {
-      "epoch": 3.6424574258222955,
-      "grad_norm": 0.005633770488202572,
-      "learning_rate": 0.0001999934590252805,
-      "loss": 46.0,
-      "step": 22618
-    },
-    {
-      "epoch": 3.642618462901083,
-      "grad_norm": 0.0047984495759010315,
-      "learning_rate": 0.00019999345844663152,
-      "loss": 46.0,
-      "step": 22619
-    },
-    {
-      "epoch": 3.6427794999798704,
-      "grad_norm": 0.0024316892959177494,
-      "learning_rate": 0.00019999345786795694,
-      "loss": 46.0,
-      "step": 22620
-    },
-    {
-      "epoch": 3.642940537058658,
-      "grad_norm": 0.003199871862307191,
-      "learning_rate": 0.0001999934572892568,
-      "loss": 46.0,
-      "step": 22621
-    },
-    {
-      "epoch": 3.6431015741374453,
-      "grad_norm": 0.002251243218779564,
-      "learning_rate": 0.000199993456710531,
-      "loss": 46.0,
-      "step": 22622
-    },
-    {
-      "epoch": 3.6432626112162323,
-      "grad_norm": 0.005125878844410181,
-      "learning_rate": 0.00019999345613177965,
-      "loss": 46.0,
-      "step": 22623
-    },
-    {
-      "epoch": 3.6434236482950197,
-      "grad_norm": 0.003056485904380679,
-      "learning_rate": 0.00019999345555300272,
-      "loss": 46.0,
-      "step": 22624
-    },
-    {
-      "epoch": 3.643584685373807,
-      "grad_norm": 0.005509105511009693,
-      "learning_rate": 0.0001999934549742002,
-      "loss": 46.0,
-      "step": 22625
-    },
-    {
-      "epoch": 3.6437457224525946,
-      "grad_norm": 0.004110974259674549,
-      "learning_rate": 0.00019999345439537206,
-      "loss": 46.0,
-      "step": 22626
-    },
-    {
-      "epoch": 3.643906759531382,
-      "grad_norm": 0.0008788192644715309,
-      "learning_rate": 0.00019999345381651833,
-      "loss": 46.0,
-      "step": 22627
-    },
-    {
-      "epoch": 3.6440677966101696,
-      "grad_norm": 0.00850562285631895,
-      "learning_rate": 0.000199993453237639,
-      "loss": 46.0,
-      "step": 22628
-    },
-    {
-      "epoch": 3.644228833688957,
-      "grad_norm": 0.0026348510291427374,
-      "learning_rate": 0.0001999934526587341,
-      "loss": 46.0,
-      "step": 22629
-    },
-    {
-      "epoch": 3.6443898707677445,
-      "grad_norm": 0.008174367249011993,
-      "learning_rate": 0.0001999934520798036,
-      "loss": 46.0,
-      "step": 22630
-    },
-    {
-      "epoch": 3.644550907846532,
-      "grad_norm": 0.0071050627157092094,
-      "learning_rate": 0.00019999345150084749,
-      "loss": 46.0,
-      "step": 22631
-    },
-    {
-      "epoch": 3.644711944925319,
-      "grad_norm": 0.005144836381077766,
-      "learning_rate": 0.0001999934509218658,
-      "loss": 46.0,
-      "step": 22632
-    },
-    {
-      "epoch": 3.6448729820041064,
-      "grad_norm": 0.0004223309806548059,
-      "learning_rate": 0.00019999345034285853,
-      "loss": 46.0,
-      "step": 22633
-    },
-    {
-      "epoch": 3.645034019082894,
-      "grad_norm": 0.006718894932419062,
-      "learning_rate": 0.00019999344976382564,
-      "loss": 46.0,
-      "step": 22634
-    },
-    {
-      "epoch": 3.6451950561616813,
-      "grad_norm": 0.01032285112887621,
-      "learning_rate": 0.00019999344918476716,
-      "loss": 46.0,
-      "step": 22635
-    },
-    {
-      "epoch": 3.6453560932404687,
-      "grad_norm": 0.005285908468067646,
-      "learning_rate": 0.0001999934486056831,
-      "loss": 46.0,
-      "step": 22636
-    },
-    {
-      "epoch": 3.645517130319256,
-      "grad_norm": 0.0031795850954949856,
-      "learning_rate": 0.00019999344802657345,
-      "loss": 46.0,
-      "step": 22637
-    },
-    {
-      "epoch": 3.645678167398043,
-      "grad_norm": 0.0035892643500119448,
-      "learning_rate": 0.00019999344744743819,
-      "loss": 46.0,
-      "step": 22638
-    },
-    {
-      "epoch": 3.6458392044768306,
-      "grad_norm": 0.004392752889543772,
-      "learning_rate": 0.00019999344686827733,
-      "loss": 46.0,
-      "step": 22639
-    },
-    {
-      "epoch": 3.646000241555618,
-      "grad_norm": 0.005292119458317757,
-      "learning_rate": 0.00019999344628909092,
-      "loss": 46.0,
-      "step": 22640
-    },
-    {
-      "epoch": 3.6461612786344055,
-      "grad_norm": 0.0012819821713492274,
-      "learning_rate": 0.00019999344570987887,
-      "loss": 46.0,
-      "step": 22641
-    },
-    {
-      "epoch": 3.646322315713193,
-      "grad_norm": 0.0009550033137202263,
-      "learning_rate": 0.00019999344513064123,
-      "loss": 46.0,
-      "step": 22642
-    },
-    {
-      "epoch": 3.6464833527919804,
-      "grad_norm": 0.009785851463675499,
-      "learning_rate": 0.00019999344455137803,
-      "loss": 46.0,
-      "step": 22643
-    },
-    {
-      "epoch": 3.646644389870768,
-      "grad_norm": 0.0019457914168015122,
-      "learning_rate": 0.00019999344397208924,
-      "loss": 46.0,
-      "step": 22644
-    },
-    {
-      "epoch": 3.6468054269495553,
-      "grad_norm": 0.0035382984206080437,
-      "learning_rate": 0.00019999344339277481,
-      "loss": 46.0,
-      "step": 22645
-    },
-    {
-      "epoch": 3.646966464028343,
-      "grad_norm": 0.003357143606990576,
-      "learning_rate": 0.00019999344281343483,
-      "loss": 46.0,
-      "step": 22646
-    },
-    {
-      "epoch": 3.64712750110713,
-      "grad_norm": 0.002282208763062954,
-      "learning_rate": 0.00019999344223406925,
-      "loss": 46.0,
-      "step": 22647
-    },
-    {
-      "epoch": 3.6472885381859173,
-      "grad_norm": 0.003200829029083252,
-      "learning_rate": 0.00019999344165467806,
-      "loss": 46.0,
-      "step": 22648
-    },
-    {
-      "epoch": 3.6474495752647047,
-      "grad_norm": 0.00044119858648627996,
-      "learning_rate": 0.0001999934410752613,
-      "loss": 46.0,
-      "step": 22649
-    },
-    {
-      "epoch": 3.647610612343492,
-      "grad_norm": 0.0016092200530692935,
-      "learning_rate": 0.00019999344049581892,
-      "loss": 46.0,
-      "step": 22650
-    },
-    {
-      "epoch": 3.6477716494222796,
-      "grad_norm": 0.0016649457393214107,
-      "learning_rate": 0.00019999343991635097,
-      "loss": 46.0,
-      "step": 22651
-    },
-    {
-      "epoch": 3.6479326865010666,
-      "grad_norm": 0.008032422512769699,
-      "learning_rate": 0.0001999934393368574,
-      "loss": 46.0,
-      "step": 22652
-    },
-    {
-      "epoch": 3.648093723579854,
-      "grad_norm": 0.001791222020983696,
-      "learning_rate": 0.00019999343875733825,
-      "loss": 46.0,
-      "step": 22653
-    },
-    {
-      "epoch": 3.6482547606586415,
-      "grad_norm": 0.001990992808714509,
-      "learning_rate": 0.00019999343817779354,
-      "loss": 46.0,
-      "step": 22654
-    },
-    {
-      "epoch": 3.648415797737429,
-      "grad_norm": 0.0013464881340041757,
-      "learning_rate": 0.00019999343759822318,
-      "loss": 46.0,
-      "step": 22655
-    },
-    {
-      "epoch": 3.6485768348162164,
-      "grad_norm": 0.0010086926631629467,
-      "learning_rate": 0.00019999343701862724,
-      "loss": 46.0,
-      "step": 22656
-    },
-    {
-      "epoch": 3.648737871895004,
-      "grad_norm": 0.005623297765851021,
-      "learning_rate": 0.00019999343643900574,
-      "loss": 46.0,
-      "step": 22657
-    },
-    {
-      "epoch": 3.6488989089737913,
-      "grad_norm": 0.0007925299578346312,
-      "learning_rate": 0.00019999343585935862,
-      "loss": 46.0,
-      "step": 22658
-    },
-    {
-      "epoch": 3.6490599460525788,
-      "grad_norm": 0.0023344990331679583,
-      "learning_rate": 0.0001999934352796859,
-      "loss": 46.0,
-      "step": 22659
-    },
-    {
-      "epoch": 3.6492209831313662,
-      "grad_norm": 0.0036977152340114117,
-      "learning_rate": 0.0001999934346999876,
-      "loss": 46.0,
-      "step": 22660
-    },
-    {
-      "epoch": 3.6493820202101532,
-      "grad_norm": 0.004310498014092445,
-      "learning_rate": 0.00019999343412026372,
-      "loss": 46.0,
-      "step": 22661
-    },
-    {
-      "epoch": 3.6495430572889407,
-      "grad_norm": 0.0026043320540338755,
-      "learning_rate": 0.00019999343354051423,
-      "loss": 46.0,
-      "step": 22662
-    },
-    {
-      "epoch": 3.649704094367728,
-      "grad_norm": 0.0022007126826792955,
-      "learning_rate": 0.00019999343296073913,
-      "loss": 46.0,
-      "step": 22663
-    },
-    {
-      "epoch": 3.6498651314465156,
-      "grad_norm": 0.004146067891269922,
-      "learning_rate": 0.0001999934323809385,
-      "loss": 46.0,
-      "step": 22664
-    },
-    {
-      "epoch": 3.650026168525303,
-      "grad_norm": 0.0017406063852831721,
-      "learning_rate": 0.0001999934318011122,
-      "loss": 46.0,
-      "step": 22665
-    },
-    {
-      "epoch": 3.6501872056040905,
-      "grad_norm": 0.006764392368495464,
-      "learning_rate": 0.00019999343122126034,
-      "loss": 46.0,
-      "step": 22666
-    },
-    {
-      "epoch": 3.6503482426828775,
-      "grad_norm": 0.004552258178591728,
-      "learning_rate": 0.00019999343064138289,
-      "loss": 46.0,
-      "step": 22667
-    },
-    {
-      "epoch": 3.650509279761665,
-      "grad_norm": 0.004710499197244644,
-      "learning_rate": 0.00019999343006147987,
-      "loss": 46.0,
-      "step": 22668
-    },
-    {
-      "epoch": 3.6506703168404524,
-      "grad_norm": 0.0016715852543711662,
-      "learning_rate": 0.00019999342948155122,
-      "loss": 46.0,
-      "step": 22669
-    },
-    {
-      "epoch": 3.65083135391924,
-      "grad_norm": 0.0037318391259759665,
-      "learning_rate": 0.00019999342890159697,
-      "loss": 46.0,
-      "step": 22670
-    },
-    {
-      "epoch": 3.6509923909980273,
-      "grad_norm": 0.0016722220461815596,
-      "learning_rate": 0.00019999342832161717,
-      "loss": 46.0,
-      "step": 22671
-    },
-    {
-      "epoch": 3.6511534280768148,
-      "grad_norm": 0.0030264039523899555,
-      "learning_rate": 0.00019999342774161175,
-      "loss": 46.0,
-      "step": 22672
-    },
-    {
-      "epoch": 3.651314465155602,
-      "grad_norm": 0.007123306393623352,
-      "learning_rate": 0.00019999342716158075,
-      "loss": 46.0,
-      "step": 22673
-    },
-    {
-      "epoch": 3.6514755022343897,
-      "grad_norm": 0.006374321412295103,
-      "learning_rate": 0.0001999934265815241,
-      "loss": 46.0,
-      "step": 22674
-    },
-    {
-      "epoch": 3.651636539313177,
-      "grad_norm": 0.002023697830736637,
-      "learning_rate": 0.00019999342600144192,
-      "loss": 46.0,
-      "step": 22675
-    },
-    {
-      "epoch": 3.651797576391964,
-      "grad_norm": 0.003963215742260218,
-      "learning_rate": 0.00019999342542133413,
-      "loss": 46.0,
-      "step": 22676
-    },
-    {
-      "epoch": 3.6519586134707516,
-      "grad_norm": 0.001480724080465734,
-      "learning_rate": 0.00019999342484120077,
-      "loss": 46.0,
-      "step": 22677
-    },
-    {
-      "epoch": 3.652119650549539,
-      "grad_norm": 0.01023608073592186,
-      "learning_rate": 0.00019999342426104178,
-      "loss": 46.0,
-      "step": 22678
-    },
-    {
-      "epoch": 3.6522806876283265,
-      "grad_norm": 0.00320429471321404,
-      "learning_rate": 0.0001999934236808572,
-      "loss": 46.0,
-      "step": 22679
-    },
-    {
-      "epoch": 3.652441724707114,
-      "grad_norm": 0.004085510969161987,
-      "learning_rate": 0.00019999342310064703,
-      "loss": 46.0,
-      "step": 22680
-    },
-    {
-      "epoch": 3.652602761785901,
-      "grad_norm": 0.016598118469119072,
-      "learning_rate": 0.0001999934225204113,
-      "loss": 46.0,
-      "step": 22681
-    },
-    {
-      "epoch": 3.6527637988646884,
-      "grad_norm": 0.0022013175766915083,
-      "learning_rate": 0.00019999342194014993,
-      "loss": 46.0,
-      "step": 22682
-    },
-    {
-      "epoch": 3.652924835943476,
-      "grad_norm": 0.009513985365629196,
-      "learning_rate": 0.000199993421359863,
-      "loss": 46.0,
-      "step": 22683
-    },
-    {
-      "epoch": 3.6530858730222633,
-      "grad_norm": 0.015665218234062195,
-      "learning_rate": 0.00019999342077955045,
-      "loss": 46.0,
-      "step": 22684
-    },
-    {
-      "epoch": 3.6532469101010507,
-      "grad_norm": 0.00104470644146204,
-      "learning_rate": 0.00019999342019921232,
-      "loss": 46.0,
-      "step": 22685
-    },
-    {
-      "epoch": 3.653407947179838,
-      "grad_norm": 0.0038735801354050636,
-      "learning_rate": 0.0001999934196188486,
-      "loss": 46.0,
-      "step": 22686
-    },
-    {
-      "epoch": 3.6535689842586256,
-      "grad_norm": 0.001528553431853652,
-      "learning_rate": 0.00019999341903845927,
-      "loss": 46.0,
-      "step": 22687
-    },
-    {
-      "epoch": 3.653730021337413,
-      "grad_norm": 0.023378996178507805,
-      "learning_rate": 0.00019999341845804435,
-      "loss": 46.0,
-      "step": 22688
-    },
-    {
-      "epoch": 3.6538910584162005,
-      "grad_norm": 0.008008182048797607,
-      "learning_rate": 0.00019999341787760387,
-      "loss": 46.0,
-      "step": 22689
-    },
-    {
-      "epoch": 3.654052095494988,
-      "grad_norm": 0.007209602277725935,
-      "learning_rate": 0.00019999341729713777,
-      "loss": 46.0,
-      "step": 22690
-    },
-    {
-      "epoch": 3.654213132573775,
-      "grad_norm": 0.002953160787001252,
-      "learning_rate": 0.0001999934167166461,
-      "loss": 46.0,
-      "step": 22691
-    },
-    {
-      "epoch": 3.6543741696525625,
-      "grad_norm": 0.0018243159865960479,
-      "learning_rate": 0.00019999341613612882,
-      "loss": 46.0,
-      "step": 22692
-    },
-    {
-      "epoch": 3.65453520673135,
-      "grad_norm": 0.002337329089641571,
-      "learning_rate": 0.0001999934155555859,
-      "loss": 46.0,
-      "step": 22693
-    },
-    {
-      "epoch": 3.6546962438101374,
-      "grad_norm": 0.0028950306586921215,
-      "learning_rate": 0.00019999341497501746,
-      "loss": 46.0,
-      "step": 22694
-    },
-    {
-      "epoch": 3.654857280888925,
-      "grad_norm": 0.002868007170036435,
-      "learning_rate": 0.0001999934143944234,
-      "loss": 46.0,
-      "step": 22695
-    },
-    {
-      "epoch": 3.655018317967712,
-      "grad_norm": 0.0024472856894135475,
-      "learning_rate": 0.00019999341381380373,
-      "loss": 46.0,
-      "step": 22696
-    },
-    {
-      "epoch": 3.6551793550464993,
-      "grad_norm": 0.0033766641281545162,
-      "learning_rate": 0.0001999934132331585,
-      "loss": 46.0,
-      "step": 22697
-    },
-    {
-      "epoch": 3.6553403921252867,
-      "grad_norm": 0.00121620437130332,
-      "learning_rate": 0.00019999341265248765,
-      "loss": 46.0,
-      "step": 22698
-    },
-    {
-      "epoch": 3.655501429204074,
-      "grad_norm": 0.006206164136528969,
-      "learning_rate": 0.00019999341207179122,
-      "loss": 46.0,
-      "step": 22699
-    },
-    {
-      "epoch": 3.6556624662828616,
-      "grad_norm": 0.00413085101172328,
-      "learning_rate": 0.00019999341149106917,
-      "loss": 46.0,
-      "step": 22700
-    },
-    {
-      "epoch": 3.655823503361649,
-      "grad_norm": 0.004665317479521036,
-      "learning_rate": 0.00019999341091032156,
-      "loss": 46.0,
-      "step": 22701
-    },
-    {
-      "epoch": 3.6559845404404365,
-      "grad_norm": 0.002895243000239134,
-      "learning_rate": 0.00019999341032954834,
-      "loss": 46.0,
-      "step": 22702
-    },
-    {
-      "epoch": 3.656145577519224,
-      "grad_norm": 0.0028475916478782892,
-      "learning_rate": 0.00019999340974874953,
-      "loss": 46.0,
-      "step": 22703
-    },
-    {
-      "epoch": 3.6563066145980114,
-      "grad_norm": 0.00355276046320796,
-      "learning_rate": 0.00019999340916792516,
-      "loss": 46.0,
-      "step": 22704
-    },
-    {
-      "epoch": 3.6564676516767984,
-      "grad_norm": 0.0028274033684283495,
-      "learning_rate": 0.00019999340858707515,
-      "loss": 46.0,
-      "step": 22705
-    },
-    {
-      "epoch": 3.656628688755586,
-      "grad_norm": 0.0059944987297058105,
-      "learning_rate": 0.00019999340800619955,
-      "loss": 46.0,
-      "step": 22706
-    },
-    {
-      "epoch": 3.6567897258343733,
-      "grad_norm": 0.009118501096963882,
-      "learning_rate": 0.0001999934074252984,
-      "loss": 46.0,
-      "step": 22707
-    },
-    {
-      "epoch": 3.656950762913161,
-      "grad_norm": 0.0010084727546200156,
-      "learning_rate": 0.00019999340684437162,
-      "loss": 46.0,
-      "step": 22708
-    },
-    {
-      "epoch": 3.6571117999919482,
-      "grad_norm": 0.0014942653942853212,
-      "learning_rate": 0.00019999340626341924,
-      "loss": 46.0,
-      "step": 22709
-    },
-    {
-      "epoch": 3.6572728370707357,
-      "grad_norm": 0.001462449668906629,
-      "learning_rate": 0.0001999934056824413,
-      "loss": 46.0,
-      "step": 22710
-    },
-    {
-      "epoch": 3.6574338741495227,
-      "grad_norm": 0.004902323242276907,
-      "learning_rate": 0.00019999340510143773,
-      "loss": 46.0,
-      "step": 22711
-    },
-    {
-      "epoch": 3.65759491122831,
-      "grad_norm": 0.0022143602836877108,
-      "learning_rate": 0.00019999340452040858,
-      "loss": 46.0,
-      "step": 22712
-    },
-    {
-      "epoch": 3.6577559483070976,
-      "grad_norm": 0.0005657552974298596,
-      "learning_rate": 0.00019999340393935384,
-      "loss": 46.0,
-      "step": 22713
-    },
-    {
-      "epoch": 3.657916985385885,
-      "grad_norm": 0.004769987426698208,
-      "learning_rate": 0.0001999934033582735,
-      "loss": 46.0,
-      "step": 22714
-    },
-    {
-      "epoch": 3.6580780224646725,
-      "grad_norm": 0.002967857290059328,
-      "learning_rate": 0.00019999340277716758,
-      "loss": 46.0,
-      "step": 22715
-    },
-    {
-      "epoch": 3.65823905954346,
-      "grad_norm": 0.010575953871011734,
-      "learning_rate": 0.00019999340219603609,
-      "loss": 46.0,
-      "step": 22716
-    },
-    {
-      "epoch": 3.6584000966222474,
-      "grad_norm": 0.0020869679283350706,
-      "learning_rate": 0.00019999340161487895,
-      "loss": 46.0,
-      "step": 22717
-    },
-    {
-      "epoch": 3.658561133701035,
-      "grad_norm": 0.00603108387440443,
-      "learning_rate": 0.00019999340103369625,
-      "loss": 46.0,
-      "step": 22718
-    },
-    {
-      "epoch": 3.6587221707798223,
-      "grad_norm": 0.0023619262501597404,
-      "learning_rate": 0.00019999340045248796,
-      "loss": 46.0,
-      "step": 22719
-    },
-    {
-      "epoch": 3.6588832078586093,
-      "grad_norm": 0.0019366255728527904,
-      "learning_rate": 0.00019999339987125406,
-      "loss": 46.0,
-      "step": 22720
-    },
-    {
-      "epoch": 3.6590442449373968,
-      "grad_norm": 0.002384476363658905,
-      "learning_rate": 0.00019999339928999458,
-      "loss": 46.0,
-      "step": 22721
-    },
-    {
-      "epoch": 3.6592052820161842,
-      "grad_norm": 0.003032886190339923,
-      "learning_rate": 0.0001999933987087095,
-      "loss": 46.0,
-      "step": 22722
-    },
-    {
-      "epoch": 3.6593663190949717,
-      "grad_norm": 0.007723737508058548,
-      "learning_rate": 0.00019999339812739884,
-      "loss": 46.0,
-      "step": 22723
-    },
-    {
-      "epoch": 3.659527356173759,
-      "grad_norm": 0.0014552040956914425,
-      "learning_rate": 0.00019999339754606256,
-      "loss": 46.0,
-      "step": 22724
-    },
-    {
-      "epoch": 3.659688393252546,
-      "grad_norm": 0.0017237073043361306,
-      "learning_rate": 0.00019999339696470073,
-      "loss": 46.0,
-      "step": 22725
-    },
-    {
-      "epoch": 3.6598494303313336,
-      "grad_norm": 0.0014080456458032131,
-      "learning_rate": 0.00019999339638331325,
-      "loss": 46.0,
-      "step": 22726
-    },
-    {
-      "epoch": 3.660010467410121,
-      "grad_norm": 0.002336216624826193,
-      "learning_rate": 0.0001999933958019002,
-      "loss": 46.0,
-      "step": 22727
-    },
-    {
-      "epoch": 3.6601715044889085,
-      "grad_norm": 0.006157963536679745,
-      "learning_rate": 0.00019999339522046156,
-      "loss": 46.0,
-      "step": 22728
-    },
-    {
-      "epoch": 3.660332541567696,
-      "grad_norm": 0.0021693038288503885,
-      "learning_rate": 0.00019999339463899735,
-      "loss": 46.0,
-      "step": 22729
-    },
-    {
-      "epoch": 3.6604935786464834,
-      "grad_norm": 0.01331277471035719,
-      "learning_rate": 0.00019999339405750753,
-      "loss": 46.0,
-      "step": 22730
-    },
-    {
-      "epoch": 3.660654615725271,
-      "grad_norm": 0.007339002098888159,
-      "learning_rate": 0.0001999933934759921,
-      "loss": 46.0,
-      "step": 22731
-    },
-    {
-      "epoch": 3.6608156528040583,
-      "grad_norm": 0.0050214375369250774,
-      "learning_rate": 0.00019999339289445108,
-      "loss": 46.0,
-      "step": 22732
-    },
-    {
-      "epoch": 3.6609766898828457,
-      "grad_norm": 0.011197277344763279,
-      "learning_rate": 0.0001999933923128845,
-      "loss": 46.0,
-      "step": 22733
-    },
-    {
-      "epoch": 3.6611377269616328,
-      "grad_norm": 0.006236036773771048,
-      "learning_rate": 0.0001999933917312923,
-      "loss": 46.0,
-      "step": 22734
-    },
-    {
-      "epoch": 3.66129876404042,
-      "grad_norm": 0.0030410068575292826,
-      "learning_rate": 0.0001999933911496745,
-      "loss": 46.0,
-      "step": 22735
-    },
-    {
-      "epoch": 3.6614598011192077,
-      "grad_norm": 0.00892493687570095,
-      "learning_rate": 0.00019999339056803113,
-      "loss": 46.0,
-      "step": 22736
-    },
-    {
-      "epoch": 3.661620838197995,
-      "grad_norm": 0.002359558828175068,
-      "learning_rate": 0.00019999338998636214,
-      "loss": 46.0,
-      "step": 22737
-    },
-    {
-      "epoch": 3.6617818752767826,
-      "grad_norm": 0.0013390594394877553,
-      "learning_rate": 0.0001999933894046676,
-      "loss": 46.0,
-      "step": 22738
-    },
-    {
-      "epoch": 3.66194291235557,
-      "grad_norm": 0.0014586279867216945,
-      "learning_rate": 0.00019999338882294743,
-      "loss": 46.0,
-      "step": 22739
-    },
-    {
-      "epoch": 3.662103949434357,
-      "grad_norm": 0.003718315390869975,
-      "learning_rate": 0.00019999338824120167,
-      "loss": 46.0,
-      "step": 22740
-    },
-    {
-      "epoch": 3.6622649865131445,
-      "grad_norm": 0.0015525443013757467,
-      "learning_rate": 0.0001999933876594303,
-      "loss": 46.0,
-      "step": 22741
-    },
-    {
-      "epoch": 3.662426023591932,
-      "grad_norm": 0.0017321790801361203,
-      "learning_rate": 0.00019999338707763338,
-      "loss": 46.0,
-      "step": 22742
-    },
-    {
-      "epoch": 3.6625870606707194,
-      "grad_norm": 0.004759287461638451,
-      "learning_rate": 0.00019999338649581084,
-      "loss": 46.0,
-      "step": 22743
-    },
-    {
-      "epoch": 3.662748097749507,
-      "grad_norm": 0.0031508072279393673,
-      "learning_rate": 0.0001999933859139627,
-      "loss": 46.0,
-      "step": 22744
-    },
-    {
-      "epoch": 3.6629091348282943,
-      "grad_norm": 0.001341440831311047,
-      "learning_rate": 0.000199993385332089,
-      "loss": 46.0,
-      "step": 22745
-    },
-    {
-      "epoch": 3.6630701719070817,
-      "grad_norm": 0.011570530012249947,
-      "learning_rate": 0.00019999338475018967,
-      "loss": 46.0,
-      "step": 22746
-    },
-    {
-      "epoch": 3.663231208985869,
-      "grad_norm": 0.0018969964003190398,
-      "learning_rate": 0.00019999338416826475,
-      "loss": 46.0,
-      "step": 22747
-    },
-    {
-      "epoch": 3.6633922460646566,
-      "grad_norm": 0.0043100775219500065,
-      "learning_rate": 0.00019999338358631428,
-      "loss": 46.0,
-      "step": 22748
-    },
-    {
-      "epoch": 3.6635532831434436,
-      "grad_norm": 0.0054091825149953365,
-      "learning_rate": 0.00019999338300433816,
-      "loss": 46.0,
-      "step": 22749
-    },
-    {
-      "epoch": 3.663714320222231,
-      "grad_norm": 0.0025551102589815855,
-      "learning_rate": 0.00019999338242233648,
-      "loss": 46.0,
-      "step": 22750
-    },
-    {
-      "epoch": 3.6638753573010185,
-      "grad_norm": 0.009712565690279007,
-      "learning_rate": 0.0001999933818403092,
-      "loss": 46.0,
-      "step": 22751
-    },
-    {
-      "epoch": 3.664036394379806,
-      "grad_norm": 0.0016537596238777041,
-      "learning_rate": 0.00019999338125825636,
-      "loss": 46.0,
-      "step": 22752
-    },
-    {
-      "epoch": 3.6641974314585934,
-      "grad_norm": 0.003614256391301751,
-      "learning_rate": 0.00019999338067617787,
-      "loss": 46.0,
-      "step": 22753
-    },
-    {
-      "epoch": 3.664358468537381,
-      "grad_norm": 0.007853053510189056,
-      "learning_rate": 0.00019999338009407382,
-      "loss": 46.0,
-      "step": 22754
-    },
-    {
-      "epoch": 3.664519505616168,
-      "grad_norm": 0.002018106635659933,
-      "learning_rate": 0.00019999337951194415,
-      "loss": 46.0,
-      "step": 22755
-    },
-    {
-      "epoch": 3.6646805426949554,
-      "grad_norm": 0.00292217917740345,
-      "learning_rate": 0.00019999337892978892,
-      "loss": 46.0,
-      "step": 22756
-    },
-    {
-      "epoch": 3.664841579773743,
-      "grad_norm": 0.005501434672623873,
-      "learning_rate": 0.00019999337834760808,
-      "loss": 46.0,
-      "step": 22757
-    },
-    {
-      "epoch": 3.6650026168525303,
-      "grad_norm": 0.009175337851047516,
-      "learning_rate": 0.00019999337776540162,
-      "loss": 46.0,
-      "step": 22758
-    },
-    {
-      "epoch": 3.6651636539313177,
-      "grad_norm": 0.0013669782783836126,
-      "learning_rate": 0.0001999933771831696,
-      "loss": 46.0,
-      "step": 22759
-    },
-    {
-      "epoch": 3.665324691010105,
-      "grad_norm": 0.0018125309143215418,
-      "learning_rate": 0.000199993376600912,
-      "loss": 46.0,
-      "step": 22760
-    },
-    {
-      "epoch": 3.6654857280888926,
-      "grad_norm": 0.002656195079907775,
-      "learning_rate": 0.00019999337601862878,
-      "loss": 46.0,
-      "step": 22761
-    },
-    {
-      "epoch": 3.66564676516768,
-      "grad_norm": 0.004852490499615669,
-      "learning_rate": 0.00019999337543632,
-      "loss": 46.0,
-      "step": 22762
-    },
-    {
-      "epoch": 3.6658078022464675,
-      "grad_norm": 0.004935932345688343,
-      "learning_rate": 0.0001999933748539856,
-      "loss": 46.0,
-      "step": 22763
-    },
-    {
-      "epoch": 3.6659688393252545,
-      "grad_norm": 0.006723203230649233,
-      "learning_rate": 0.00019999337427162558,
-      "loss": 46.0,
-      "step": 22764
-    },
-    {
-      "epoch": 3.666129876404042,
-      "grad_norm": 0.00183191429823637,
-      "learning_rate": 0.00019999337368924001,
-      "loss": 46.0,
-      "step": 22765
-    },
-    {
-      "epoch": 3.6662909134828294,
-      "grad_norm": 0.0024352543987333775,
-      "learning_rate": 0.0001999933731068288,
-      "loss": 46.0,
-      "step": 22766
-    },
-    {
-      "epoch": 3.666451950561617,
-      "grad_norm": 0.010108031332492828,
-      "learning_rate": 0.00019999337252439207,
-      "loss": 46.0,
-      "step": 22767
-    },
-    {
-      "epoch": 3.6666129876404043,
-      "grad_norm": 0.015142385847866535,
-      "learning_rate": 0.0001999933719419297,
-      "loss": 46.0,
-      "step": 22768
-    },
-    {
-      "epoch": 3.6667740247191913,
-      "grad_norm": 0.0029771251138299704,
-      "learning_rate": 0.00019999337135944174,
-      "loss": 46.0,
-      "step": 22769
-    },
-    {
-      "epoch": 3.666935061797979,
-      "grad_norm": 0.00233553652651608,
-      "learning_rate": 0.00019999337077692818,
-      "loss": 46.0,
-      "step": 22770
-    },
-    {
-      "epoch": 3.6670960988767662,
-      "grad_norm": 0.0017812150763347745,
-      "learning_rate": 0.00019999337019438907,
-      "loss": 46.0,
-      "step": 22771
-    },
-    {
-      "epoch": 3.6672571359555537,
-      "grad_norm": 0.002252899343147874,
-      "learning_rate": 0.0001999933696118243,
-      "loss": 46.0,
-      "step": 22772
-    },
-    {
-      "epoch": 3.667418173034341,
-      "grad_norm": 0.0011877642245963216,
-      "learning_rate": 0.000199993369029234,
-      "loss": 46.0,
-      "step": 22773
-    },
-    {
-      "epoch": 3.6675792101131286,
-      "grad_norm": 0.0007877996540628374,
-      "learning_rate": 0.00019999336844661806,
-      "loss": 46.0,
-      "step": 22774
-    },
-    {
-      "epoch": 3.667740247191916,
-      "grad_norm": 0.001983505440875888,
-      "learning_rate": 0.00019999336786397654,
-      "loss": 46.0,
-      "step": 22775
-    },
-    {
-      "epoch": 3.6679012842707035,
-      "grad_norm": 0.005540449172258377,
-      "learning_rate": 0.00019999336728130943,
-      "loss": 46.0,
-      "step": 22776
-    },
-    {
-      "epoch": 3.668062321349491,
-      "grad_norm": 0.010100196115672588,
-      "learning_rate": 0.00019999336669861674,
-      "loss": 46.0,
-      "step": 22777
-    },
-    {
-      "epoch": 3.668223358428278,
-      "grad_norm": 0.005990343634039164,
-      "learning_rate": 0.00019999336611589843,
-      "loss": 46.0,
-      "step": 22778
-    },
-    {
-      "epoch": 3.6683843955070654,
-      "grad_norm": 0.0009399388218298554,
-      "learning_rate": 0.00019999336553315456,
-      "loss": 46.0,
-      "step": 22779
-    },
-    {
-      "epoch": 3.668545432585853,
-      "grad_norm": 0.0028247591108083725,
-      "learning_rate": 0.00019999336495038508,
-      "loss": 46.0,
-      "step": 22780
-    },
-    {
-      "epoch": 3.6687064696646403,
-      "grad_norm": 0.0014354345621541142,
-      "learning_rate": 0.00019999336436758998,
-      "loss": 46.0,
-      "step": 22781
-    },
-    {
-      "epoch": 3.6688675067434278,
-      "grad_norm": 0.002890727249905467,
-      "learning_rate": 0.00019999336378476932,
-      "loss": 46.0,
-      "step": 22782
-    },
-    {
-      "epoch": 3.669028543822215,
-      "grad_norm": 0.023550447076559067,
-      "learning_rate": 0.00019999336320192305,
-      "loss": 46.0,
-      "step": 22783
-    },
-    {
-      "epoch": 3.6691895809010022,
-      "grad_norm": 0.002335806144401431,
-      "learning_rate": 0.0001999933626190512,
-      "loss": 46.0,
-      "step": 22784
-    },
-    {
-      "epoch": 3.6693506179797897,
-      "grad_norm": 0.0026196411345154047,
-      "learning_rate": 0.00019999336203615375,
-      "loss": 46.0,
-      "step": 22785
-    },
-    {
-      "epoch": 3.669511655058577,
-      "grad_norm": 0.0011785373790189624,
-      "learning_rate": 0.00019999336145323072,
-      "loss": 46.0,
-      "step": 22786
-    },
-    {
-      "epoch": 3.6696726921373646,
-      "grad_norm": 0.00959487073123455,
-      "learning_rate": 0.00019999336087028207,
-      "loss": 46.0,
-      "step": 22787
-    },
-    {
-      "epoch": 3.669833729216152,
-      "grad_norm": 0.016064202412962914,
-      "learning_rate": 0.00019999336028730786,
-      "loss": 46.0,
-      "step": 22788
-    },
-    {
-      "epoch": 3.6699947662949395,
-      "grad_norm": 0.0059156883507966995,
-      "learning_rate": 0.00019999335970430804,
-      "loss": 46.0,
-      "step": 22789
-    },
-    {
-      "epoch": 3.670155803373727,
-      "grad_norm": 0.0074160280637443066,
-      "learning_rate": 0.00019999335912128263,
-      "loss": 46.0,
-      "step": 22790
-    },
-    {
-      "epoch": 3.6703168404525144,
-      "grad_norm": 0.0012285802513360977,
-      "learning_rate": 0.0001999933585382316,
-      "loss": 46.0,
-      "step": 22791
-    },
-    {
-      "epoch": 3.670477877531302,
-      "grad_norm": 0.0019702634308487177,
-      "learning_rate": 0.00019999335795515503,
-      "loss": 46.0,
-      "step": 22792
-    },
-    {
-      "epoch": 3.670638914610089,
-      "grad_norm": 0.002632237272337079,
-      "learning_rate": 0.0001999933573720528,
-      "loss": 46.0,
-      "step": 22793
-    },
-    {
-      "epoch": 3.6707999516888763,
-      "grad_norm": 0.01589401438832283,
-      "learning_rate": 0.00019999335678892505,
-      "loss": 46.0,
-      "step": 22794
-    },
-    {
-      "epoch": 3.6709609887676637,
-      "grad_norm": 0.0021344409324228764,
-      "learning_rate": 0.00019999335620577165,
-      "loss": 46.0,
-      "step": 22795
-    },
-    {
-      "epoch": 3.671122025846451,
-      "grad_norm": 0.004890965297818184,
-      "learning_rate": 0.00019999335562259266,
-      "loss": 46.0,
-      "step": 22796
-    },
-    {
-      "epoch": 3.6712830629252386,
-      "grad_norm": 0.015462430194020271,
-      "learning_rate": 0.00019999335503938811,
-      "loss": 46.0,
-      "step": 22797
-    },
-    {
-      "epoch": 3.6714441000040257,
-      "grad_norm": 0.00885778572410345,
-      "learning_rate": 0.00019999335445615795,
-      "loss": 46.0,
-      "step": 22798
-    },
-    {
-      "epoch": 3.671605137082813,
-      "grad_norm": 0.002308700466528535,
-      "learning_rate": 0.00019999335387290218,
-      "loss": 46.0,
-      "step": 22799
-    },
-    {
-      "epoch": 3.6717661741616006,
-      "grad_norm": 0.0008633616380393505,
-      "learning_rate": 0.00019999335328962085,
-      "loss": 46.0,
-      "step": 22800
-    },
-    {
-      "epoch": 3.671927211240388,
-      "grad_norm": 0.0028252352494746447,
-      "learning_rate": 0.00019999335270631392,
-      "loss": 46.0,
-      "step": 22801
-    },
-    {
-      "epoch": 3.6720882483191755,
-      "grad_norm": 0.0023980673868209124,
-      "learning_rate": 0.00019999335212298142,
-      "loss": 46.0,
-      "step": 22802
-    },
-    {
-      "epoch": 3.672249285397963,
-      "grad_norm": 0.008981111459434032,
-      "learning_rate": 0.00019999335153962327,
-      "loss": 46.0,
-      "step": 22803
-    },
-    {
-      "epoch": 3.6724103224767504,
-      "grad_norm": 0.016783330589532852,
-      "learning_rate": 0.00019999335095623956,
-      "loss": 46.0,
-      "step": 22804
-    },
-    {
-      "epoch": 3.672571359555538,
-      "grad_norm": 0.004916629754006863,
-      "learning_rate": 0.00019999335037283023,
-      "loss": 46.0,
-      "step": 22805
-    },
-    {
-      "epoch": 3.6727323966343253,
-      "grad_norm": 0.002610010327771306,
-      "learning_rate": 0.00019999334978939532,
-      "loss": 46.0,
-      "step": 22806
-    },
-    {
-      "epoch": 3.6728934337131127,
-      "grad_norm": 0.0028579039499163628,
-      "learning_rate": 0.00019999334920593485,
-      "loss": 46.0,
-      "step": 22807
-    },
-    {
-      "epoch": 3.6730544707918997,
-      "grad_norm": 0.0032562275882810354,
-      "learning_rate": 0.00019999334862244873,
-      "loss": 46.0,
-      "step": 22808
-    },
-    {
-      "epoch": 3.673215507870687,
-      "grad_norm": 0.003705337643623352,
-      "learning_rate": 0.00019999334803893706,
-      "loss": 46.0,
-      "step": 22809
-    },
-    {
-      "epoch": 3.6733765449494746,
-      "grad_norm": 0.012126673012971878,
-      "learning_rate": 0.0001999933474553998,
-      "loss": 46.0,
-      "step": 22810
-    },
-    {
-      "epoch": 3.673537582028262,
-      "grad_norm": 0.006201410666108131,
-      "learning_rate": 0.0001999933468718369,
-      "loss": 46.0,
-      "step": 22811
-    },
-    {
-      "epoch": 3.6736986191070495,
-      "grad_norm": 0.003494767239317298,
-      "learning_rate": 0.00019999334628824844,
-      "loss": 46.0,
-      "step": 22812
-    },
-    {
-      "epoch": 3.6738596561858365,
-      "grad_norm": 0.0015345205320045352,
-      "learning_rate": 0.0001999933457046344,
-      "loss": 46.0,
-      "step": 22813
-    },
-    {
-      "epoch": 3.674020693264624,
-      "grad_norm": 0.0024018334224820137,
-      "learning_rate": 0.00019999334512099475,
-      "loss": 46.0,
-      "step": 22814
-    },
-    {
-      "epoch": 3.6741817303434114,
-      "grad_norm": 0.004173180088400841,
-      "learning_rate": 0.0001999933445373295,
-      "loss": 46.0,
-      "step": 22815
-    },
-    {
-      "epoch": 3.674342767422199,
-      "grad_norm": 0.010577252134680748,
-      "learning_rate": 0.00019999334395363866,
-      "loss": 46.0,
-      "step": 22816
-    },
-    {
-      "epoch": 3.6745038045009863,
-      "grad_norm": 0.0017120972042903304,
-      "learning_rate": 0.00019999334336992224,
-      "loss": 46.0,
-      "step": 22817
-    },
-    {
-      "epoch": 3.674664841579774,
-      "grad_norm": 0.004271545447409153,
-      "learning_rate": 0.0001999933427861802,
-      "loss": 46.0,
-      "step": 22818
-    },
-    {
-      "epoch": 3.6748258786585613,
-      "grad_norm": 0.001638551359064877,
-      "learning_rate": 0.0001999933422024126,
-      "loss": 46.0,
-      "step": 22819
-    },
-    {
-      "epoch": 3.6749869157373487,
-      "grad_norm": 0.0024707496631890535,
-      "learning_rate": 0.00019999334161861941,
-      "loss": 46.0,
-      "step": 22820
-    },
-    {
-      "epoch": 3.675147952816136,
-      "grad_norm": 0.003369895974174142,
-      "learning_rate": 0.00019999334103480059,
-      "loss": 46.0,
-      "step": 22821
-    },
-    {
-      "epoch": 3.675308989894923,
-      "grad_norm": 0.002848181175068021,
-      "learning_rate": 0.0001999933404509562,
-      "loss": 46.0,
-      "step": 22822
-    },
-    {
-      "epoch": 3.6754700269737106,
-      "grad_norm": 0.0007846674998290837,
-      "learning_rate": 0.00019999333986708622,
-      "loss": 46.0,
-      "step": 22823
-    },
-    {
-      "epoch": 3.675631064052498,
-      "grad_norm": 0.009992103092372417,
-      "learning_rate": 0.0001999933392831906,
-      "loss": 46.0,
-      "step": 22824
-    },
-    {
-      "epoch": 3.6757921011312855,
-      "grad_norm": 0.004940797109156847,
-      "learning_rate": 0.00019999333869926943,
-      "loss": 46.0,
-      "step": 22825
-    },
-    {
-      "epoch": 3.675953138210073,
-      "grad_norm": 0.0035269337240606546,
-      "learning_rate": 0.0001999933381153227,
-      "loss": 46.0,
-      "step": 22826
-    },
-    {
-      "epoch": 3.6761141752888604,
-      "grad_norm": 0.0008664656197652221,
-      "learning_rate": 0.00019999333753135032,
-      "loss": 46.0,
-      "step": 22827
-    },
-    {
-      "epoch": 3.6762752123676474,
-      "grad_norm": 0.0016531017608940601,
-      "learning_rate": 0.00019999333694735235,
-      "loss": 46.0,
-      "step": 22828
-    },
-    {
-      "epoch": 3.676436249446435,
-      "grad_norm": 0.002864454174414277,
-      "learning_rate": 0.00019999333636332883,
-      "loss": 46.0,
-      "step": 22829
-    },
-    {
-      "epoch": 3.6765972865252223,
-      "grad_norm": 0.004423864651471376,
-      "learning_rate": 0.0001999933357792797,
-      "loss": 46.0,
-      "step": 22830
-    },
-    {
-      "epoch": 3.67675832360401,
-      "grad_norm": 0.006899977568536997,
-      "learning_rate": 0.00019999333519520493,
-      "loss": 46.0,
-      "step": 22831
-    },
-    {
-      "epoch": 3.6769193606827972,
-      "grad_norm": 0.00943421944975853,
-      "learning_rate": 0.00019999333461110462,
-      "loss": 46.0,
-      "step": 22832
-    },
-    {
-      "epoch": 3.6770803977615847,
-      "grad_norm": 0.0018641840433701873,
-      "learning_rate": 0.00019999333402697872,
-      "loss": 46.0,
-      "step": 22833
-    },
-    {
-      "epoch": 3.677241434840372,
-      "grad_norm": 0.0019748711492866278,
-      "learning_rate": 0.00019999333344282718,
-      "loss": 46.0,
-      "step": 22834
-    },
-    {
-      "epoch": 3.6774024719191596,
-      "grad_norm": 0.002223348943516612,
-      "learning_rate": 0.00019999333285865007,
-      "loss": 46.0,
-      "step": 22835
-    },
-    {
-      "epoch": 3.677563508997947,
-      "grad_norm": 0.0017048557056114078,
-      "learning_rate": 0.00019999333227444738,
-      "loss": 46.0,
-      "step": 22836
-    },
-    {
-      "epoch": 3.677724546076734,
-      "grad_norm": 0.001784069580025971,
-      "learning_rate": 0.00019999333169021908,
-      "loss": 46.0,
-      "step": 22837
-    },
-    {
-      "epoch": 3.6778855831555215,
-      "grad_norm": 0.00346533115953207,
-      "learning_rate": 0.0001999933311059652,
-      "loss": 46.0,
-      "step": 22838
-    },
-    {
-      "epoch": 3.678046620234309,
-      "grad_norm": 0.001329404185526073,
-      "learning_rate": 0.0001999933305216857,
-      "loss": 46.0,
-      "step": 22839
-    },
-    {
-      "epoch": 3.6782076573130964,
-      "grad_norm": 0.0019629541784524918,
-      "learning_rate": 0.00019999332993738065,
-      "loss": 46.0,
-      "step": 22840
-    },
-    {
-      "epoch": 3.678368694391884,
-      "grad_norm": 0.00393206812441349,
-      "learning_rate": 0.00019999332935305,
-      "loss": 46.0,
-      "step": 22841
-    },
-    {
-      "epoch": 3.678529731470671,
-      "grad_norm": 0.0031270713079720736,
-      "learning_rate": 0.00019999332876869373,
-      "loss": 46.0,
-      "step": 22842
-    },
-    {
-      "epoch": 3.6786907685494583,
-      "grad_norm": 0.002972578862681985,
-      "learning_rate": 0.00019999332818431187,
-      "loss": 46.0,
-      "step": 22843
-    },
-    {
-      "epoch": 3.6788518056282458,
-      "grad_norm": 0.0027528656646609306,
-      "learning_rate": 0.00019999332759990443,
-      "loss": 46.0,
-      "step": 22844
-    },
-    {
-      "epoch": 3.679012842707033,
-      "grad_norm": 0.003687758930027485,
-      "learning_rate": 0.00019999332701547138,
-      "loss": 46.0,
-      "step": 22845
-    },
-    {
-      "epoch": 3.6791738797858207,
-      "grad_norm": 0.0015499844448640943,
-      "learning_rate": 0.00019999332643101276,
-      "loss": 46.0,
-      "step": 22846
-    },
-    {
-      "epoch": 3.679334916864608,
-      "grad_norm": 0.006773034110665321,
-      "learning_rate": 0.00019999332584652853,
-      "loss": 46.0,
-      "step": 22847
-    },
-    {
-      "epoch": 3.6794959539433956,
-      "grad_norm": 0.00490977056324482,
-      "learning_rate": 0.00019999332526201872,
-      "loss": 46.0,
-      "step": 22848
-    },
-    {
-      "epoch": 3.679656991022183,
-      "grad_norm": 0.002157560782507062,
-      "learning_rate": 0.0001999933246774833,
-      "loss": 46.0,
-      "step": 22849
-    },
-    {
-      "epoch": 3.6798180281009705,
-      "grad_norm": 0.0025769590865820646,
-      "learning_rate": 0.00019999332409292232,
-      "loss": 46.0,
-      "step": 22850
-    },
-    {
-      "epoch": 3.6799790651797575,
-      "grad_norm": 0.0049954731948673725,
-      "learning_rate": 0.00019999332350833571,
-      "loss": 46.0,
-      "step": 22851
-    },
-    {
-      "epoch": 3.680140102258545,
-      "grad_norm": 0.00099918432533741,
-      "learning_rate": 0.00019999332292372352,
-      "loss": 46.0,
-      "step": 22852
-    },
-    {
-      "epoch": 3.6803011393373324,
-      "grad_norm": 0.002734280424192548,
-      "learning_rate": 0.00019999332233908574,
-      "loss": 46.0,
-      "step": 22853
-    },
-    {
-      "epoch": 3.68046217641612,
-      "grad_norm": 0.01219154428690672,
-      "learning_rate": 0.00019999332175442235,
-      "loss": 46.0,
-      "step": 22854
-    },
-    {
-      "epoch": 3.6806232134949073,
-      "grad_norm": 0.0005331409047357738,
-      "learning_rate": 0.0001999933211697334,
-      "loss": 46.0,
-      "step": 22855
-    },
-    {
-      "epoch": 3.6807842505736947,
-      "grad_norm": 0.006523890886455774,
-      "learning_rate": 0.00019999332058501883,
-      "loss": 46.0,
-      "step": 22856
-    },
-    {
-      "epoch": 3.6809452876524817,
-      "grad_norm": 0.0036100205034017563,
-      "learning_rate": 0.00019999332000027867,
-      "loss": 46.0,
-      "step": 22857
-    },
-    {
-      "epoch": 3.681106324731269,
-      "grad_norm": 0.0022948416881263256,
-      "learning_rate": 0.00019999331941551293,
-      "loss": 46.0,
-      "step": 22858
-    },
-    {
-      "epoch": 3.6812673618100566,
-      "grad_norm": 0.01894768513739109,
-      "learning_rate": 0.0001999933188307216,
-      "loss": 46.0,
-      "step": 22859
-    },
-    {
-      "epoch": 3.681428398888844,
-      "grad_norm": 0.00970908347517252,
-      "learning_rate": 0.00019999331824590465,
-      "loss": 46.0,
-      "step": 22860
-    },
-    {
-      "epoch": 3.6815894359676316,
-      "grad_norm": 0.0023871108423918486,
-      "learning_rate": 0.00019999331766106212,
-      "loss": 46.0,
-      "step": 22861
-    },
-    {
-      "epoch": 3.681750473046419,
-      "grad_norm": 0.0014528199099004269,
-      "learning_rate": 0.000199993317076194,
-      "loss": 46.0,
-      "step": 22862
-    },
-    {
-      "epoch": 3.6819115101252065,
-      "grad_norm": 0.005543417762964964,
-      "learning_rate": 0.0001999933164913003,
-      "loss": 46.0,
-      "step": 22863
-    },
-    {
-      "epoch": 3.682072547203994,
-      "grad_norm": 0.0044127884320914745,
-      "learning_rate": 0.00019999331590638098,
-      "loss": 46.0,
-      "step": 22864
-    },
-    {
-      "epoch": 3.6822335842827814,
-      "grad_norm": 0.0009569072281010449,
-      "learning_rate": 0.00019999331532143607,
-      "loss": 46.0,
-      "step": 22865
-    },
-    {
-      "epoch": 3.6823946213615684,
-      "grad_norm": 0.004044860135763884,
-      "learning_rate": 0.00019999331473646558,
-      "loss": 46.0,
-      "step": 22866
-    },
-    {
-      "epoch": 3.682555658440356,
-      "grad_norm": 0.006370786111801863,
-      "learning_rate": 0.0001999933141514695,
-      "loss": 46.0,
-      "step": 22867
-    },
-    {
-      "epoch": 3.6827166955191433,
-      "grad_norm": 0.0022870569955557585,
-      "learning_rate": 0.00019999331356644783,
-      "loss": 46.0,
-      "step": 22868
-    },
-    {
-      "epoch": 3.6828777325979307,
-      "grad_norm": 0.003003267338499427,
-      "learning_rate": 0.00019999331298140054,
-      "loss": 46.0,
-      "step": 22869
-    },
-    {
-      "epoch": 3.683038769676718,
-      "grad_norm": 0.0029237500857561827,
-      "learning_rate": 0.00019999331239632765,
-      "loss": 46.0,
-      "step": 22870
-    },
-    {
-      "epoch": 3.683199806755505,
-      "grad_norm": 0.0016908635152503848,
-      "learning_rate": 0.0001999933118112292,
-      "loss": 46.0,
-      "step": 22871
-    },
-    {
-      "epoch": 3.6833608438342926,
-      "grad_norm": 0.0012524742633104324,
-      "learning_rate": 0.00019999331122610517,
-      "loss": 46.0,
-      "step": 22872
-    },
-    {
-      "epoch": 3.68352188091308,
-      "grad_norm": 0.0029560548719018698,
-      "learning_rate": 0.0001999933106409555,
-      "loss": 46.0,
-      "step": 22873
-    },
-    {
-      "epoch": 3.6836829179918675,
-      "grad_norm": 0.00247136689722538,
-      "learning_rate": 0.00019999331005578027,
-      "loss": 46.0,
-      "step": 22874
-    },
-    {
-      "epoch": 3.683843955070655,
-      "grad_norm": 0.018603768199682236,
-      "learning_rate": 0.00019999330947057943,
-      "loss": 46.0,
-      "step": 22875
-    },
-    {
-      "epoch": 3.6840049921494424,
-      "grad_norm": 0.002960095414891839,
-      "learning_rate": 0.000199993308885353,
-      "loss": 46.0,
-      "step": 22876
-    },
-    {
-      "epoch": 3.68416602922823,
-      "grad_norm": 0.005040630232542753,
-      "learning_rate": 0.000199993308300101,
-      "loss": 46.0,
-      "step": 22877
-    },
-    {
-      "epoch": 3.6843270663070173,
-      "grad_norm": 0.0059850625693798065,
-      "learning_rate": 0.00019999330771482338,
-      "loss": 46.0,
-      "step": 22878
-    },
-    {
-      "epoch": 3.684488103385805,
-      "grad_norm": 0.0017186513869091868,
-      "learning_rate": 0.00019999330712952017,
-      "loss": 46.0,
-      "step": 22879
-    },
-    {
-      "epoch": 3.6846491404645922,
-      "grad_norm": 0.013015073724091053,
-      "learning_rate": 0.00019999330654419138,
-      "loss": 46.0,
-      "step": 22880
-    },
-    {
-      "epoch": 3.6848101775433793,
-      "grad_norm": 0.004506245721131563,
-      "learning_rate": 0.000199993305958837,
-      "loss": 46.0,
-      "step": 22881
-    },
-    {
-      "epoch": 3.6849712146221667,
-      "grad_norm": 0.0019291311036795378,
-      "learning_rate": 0.00019999330537345702,
-      "loss": 46.0,
-      "step": 22882
-    },
-    {
-      "epoch": 3.685132251700954,
-      "grad_norm": 0.007094101049005985,
-      "learning_rate": 0.00019999330478805144,
-      "loss": 46.0,
-      "step": 22883
-    },
-    {
-      "epoch": 3.6852932887797416,
-      "grad_norm": 0.0028363880701363087,
-      "learning_rate": 0.00019999330420262027,
-      "loss": 46.0,
-      "step": 22884
-    },
-    {
-      "epoch": 3.685454325858529,
-      "grad_norm": 0.005502133630216122,
-      "learning_rate": 0.0001999933036171635,
-      "loss": 46.0,
-      "step": 22885
-    },
-    {
-      "epoch": 3.685615362937316,
-      "grad_norm": 0.005104415584355593,
-      "learning_rate": 0.00019999330303168114,
-      "loss": 46.0,
-      "step": 22886
-    },
-    {
-      "epoch": 3.6857764000161035,
-      "grad_norm": 0.005357289686799049,
-      "learning_rate": 0.00019999330244617318,
-      "loss": 46.0,
-      "step": 22887
-    },
-    {
-      "epoch": 3.685937437094891,
-      "grad_norm": 0.0006300599779933691,
-      "learning_rate": 0.00019999330186063966,
-      "loss": 46.0,
-      "step": 22888
-    },
-    {
-      "epoch": 3.6860984741736784,
-      "grad_norm": 0.013662843964993954,
-      "learning_rate": 0.0001999933012750805,
-      "loss": 46.0,
-      "step": 22889
-    },
-    {
-      "epoch": 3.686259511252466,
-      "grad_norm": 0.0019073965959250927,
-      "learning_rate": 0.00019999330068949577,
-      "loss": 46.0,
-      "step": 22890
-    },
-    {
-      "epoch": 3.6864205483312533,
-      "grad_norm": 0.0030715062748640776,
-      "learning_rate": 0.00019999330010388543,
-      "loss": 46.0,
-      "step": 22891
-    },
-    {
-      "epoch": 3.6865815854100408,
-      "grad_norm": 0.0018682855879887938,
-      "learning_rate": 0.00019999329951824954,
-      "loss": 46.0,
-      "step": 22892
-    },
-    {
-      "epoch": 3.6867426224888282,
-      "grad_norm": 0.0003367076569702476,
-      "learning_rate": 0.00019999329893258803,
-      "loss": 46.0,
-      "step": 22893
-    },
-    {
-      "epoch": 3.6869036595676157,
-      "grad_norm": 0.005779862403869629,
-      "learning_rate": 0.00019999329834690093,
-      "loss": 46.0,
-      "step": 22894
-    },
-    {
-      "epoch": 3.6870646966464027,
-      "grad_norm": 0.0035860328935086727,
-      "learning_rate": 0.00019999329776118822,
-      "loss": 46.0,
-      "step": 22895
-    },
-    {
-      "epoch": 3.68722573372519,
-      "grad_norm": 0.0022047758102416992,
-      "learning_rate": 0.00019999329717544995,
-      "loss": 46.0,
-      "step": 22896
-    },
-    {
-      "epoch": 3.6873867708039776,
-      "grad_norm": 0.0029248481150716543,
-      "learning_rate": 0.00019999329658968606,
-      "loss": 46.0,
-      "step": 22897
-    },
-    {
-      "epoch": 3.687547807882765,
-      "grad_norm": 0.015097597613930702,
-      "learning_rate": 0.00019999329600389659,
-      "loss": 46.0,
-      "step": 22898
-    },
-    {
-      "epoch": 3.6877088449615525,
-      "grad_norm": 0.0012811818160116673,
-      "learning_rate": 0.00019999329541808153,
-      "loss": 46.0,
-      "step": 22899
-    },
-    {
-      "epoch": 3.68786988204034,
-      "grad_norm": 0.00263396929949522,
-      "learning_rate": 0.00019999329483224085,
-      "loss": 46.0,
-      "step": 22900
-    },
-    {
-      "epoch": 3.688030919119127,
-      "grad_norm": 0.004833109211176634,
-      "learning_rate": 0.00019999329424637462,
-      "loss": 46.0,
-      "step": 22901
-    },
-    {
-      "epoch": 3.6881919561979144,
-      "grad_norm": 0.0014522494748234749,
-      "learning_rate": 0.00019999329366048274,
-      "loss": 46.0,
-      "step": 22902
-    },
-    {
-      "epoch": 3.688352993276702,
-      "grad_norm": 0.0032425944227725267,
-      "learning_rate": 0.0001999932930745653,
-      "loss": 46.0,
-      "step": 22903
-    },
-    {
-      "epoch": 3.6885140303554893,
-      "grad_norm": 0.0021811225451529026,
-      "learning_rate": 0.00019999329248862225,
-      "loss": 46.0,
-      "step": 22904
-    },
-    {
-      "epoch": 3.6886750674342768,
-      "grad_norm": 0.0019636524375528097,
-      "learning_rate": 0.00019999329190265367,
-      "loss": 46.0,
-      "step": 22905
-    },
-    {
-      "epoch": 3.688836104513064,
-      "grad_norm": 0.003221196820959449,
-      "learning_rate": 0.00019999329131665942,
-      "loss": 46.0,
-      "step": 22906
-    },
-    {
-      "epoch": 3.6889971415918517,
-      "grad_norm": 0.004156569018959999,
-      "learning_rate": 0.0001999932907306396,
-      "loss": 46.0,
-      "step": 22907
-    },
-    {
-      "epoch": 3.689158178670639,
-      "grad_norm": 0.0012089312076568604,
-      "learning_rate": 0.00019999329014459423,
-      "loss": 46.0,
-      "step": 22908
-    },
-    {
-      "epoch": 3.6893192157494266,
-      "grad_norm": 0.0009855962125584483,
-      "learning_rate": 0.00019999328955852322,
-      "loss": 46.0,
-      "step": 22909
-    },
-    {
-      "epoch": 3.6894802528282136,
-      "grad_norm": 0.004043465480208397,
-      "learning_rate": 0.00019999328897242662,
-      "loss": 46.0,
-      "step": 22910
-    },
-    {
-      "epoch": 3.689641289907001,
-      "grad_norm": 0.002063443884253502,
-      "learning_rate": 0.00019999328838630445,
-      "loss": 46.0,
-      "step": 22911
-    },
-    {
-      "epoch": 3.6898023269857885,
-      "grad_norm": 0.0078023886308074,
-      "learning_rate": 0.00019999328780015665,
-      "loss": 46.0,
-      "step": 22912
-    },
-    {
-      "epoch": 3.689963364064576,
-      "grad_norm": 0.003180890344083309,
-      "learning_rate": 0.00019999328721398326,
-      "loss": 46.0,
-      "step": 22913
-    },
-    {
-      "epoch": 3.6901244011433634,
-      "grad_norm": 0.0017783313523977995,
-      "learning_rate": 0.0001999932866277843,
-      "loss": 46.0,
-      "step": 22914
-    },
-    {
-      "epoch": 3.6902854382221504,
-      "grad_norm": 0.005659760441631079,
-      "learning_rate": 0.00019999328604155975,
-      "loss": 46.0,
-      "step": 22915
-    },
-    {
-      "epoch": 3.690446475300938,
-      "grad_norm": 0.0030311914160847664,
-      "learning_rate": 0.00019999328545530963,
-      "loss": 46.0,
-      "step": 22916
-    },
-    {
-      "epoch": 3.6906075123797253,
-      "grad_norm": 0.002965526655316353,
-      "learning_rate": 0.00019999328486903386,
-      "loss": 46.0,
-      "step": 22917
-    },
-    {
-      "epoch": 3.6907685494585127,
-      "grad_norm": 0.002880080370232463,
-      "learning_rate": 0.0001999932842827325,
-      "loss": 46.0,
-      "step": 22918
-    },
-    {
-      "epoch": 3.6909295865373,
-      "grad_norm": 0.005357214715331793,
-      "learning_rate": 0.00019999328369640557,
-      "loss": 46.0,
-      "step": 22919
-    },
-    {
-      "epoch": 3.6910906236160876,
-      "grad_norm": 0.0023911993484944105,
-      "learning_rate": 0.00019999328311005304,
-      "loss": 46.0,
-      "step": 22920
-    },
-    {
-      "epoch": 3.691251660694875,
-      "grad_norm": 0.01646076887845993,
-      "learning_rate": 0.00019999328252367495,
-      "loss": 46.0,
-      "step": 22921
-    },
-    {
-      "epoch": 3.6914126977736625,
-      "grad_norm": 0.02226894348859787,
-      "learning_rate": 0.00019999328193727123,
-      "loss": 46.0,
-      "step": 22922
-    },
-    {
-      "epoch": 3.69157373485245,
-      "grad_norm": 0.0019311883952468634,
-      "learning_rate": 0.0001999932813508419,
-      "loss": 46.0,
-      "step": 22923
-    },
-    {
-      "epoch": 3.691734771931237,
-      "grad_norm": 0.005404660943895578,
-      "learning_rate": 0.00019999328076438703,
-      "loss": 46.0,
-      "step": 22924
-    },
-    {
-      "epoch": 3.6918958090100245,
-      "grad_norm": 0.0089335385710001,
-      "learning_rate": 0.00019999328017790655,
-      "loss": 46.0,
-      "step": 22925
-    },
-    {
-      "epoch": 3.692056846088812,
-      "grad_norm": 0.0027518875431269407,
-      "learning_rate": 0.00019999327959140047,
-      "loss": 46.0,
-      "step": 22926
-    },
-    {
-      "epoch": 3.6922178831675994,
-      "grad_norm": 0.008946986868977547,
-      "learning_rate": 0.00019999327900486878,
-      "loss": 46.0,
-      "step": 22927
-    },
-    {
-      "epoch": 3.692378920246387,
-      "grad_norm": 0.003374751191586256,
-      "learning_rate": 0.00019999327841831153,
-      "loss": 46.0,
-      "step": 22928
-    },
-    {
-      "epoch": 3.6925399573251743,
-      "grad_norm": 0.0034910866525024176,
-      "learning_rate": 0.00019999327783172866,
-      "loss": 46.0,
-      "step": 22929
-    },
-    {
-      "epoch": 3.6927009944039613,
-      "grad_norm": 0.005131212528795004,
-      "learning_rate": 0.0001999932772451202,
-      "loss": 46.0,
-      "step": 22930
-    },
-    {
-      "epoch": 3.6928620314827487,
-      "grad_norm": 0.0021010444033890963,
-      "learning_rate": 0.00019999327665848614,
-      "loss": 46.0,
-      "step": 22931
-    },
-    {
-      "epoch": 3.693023068561536,
-      "grad_norm": 0.0015383217250928283,
-      "learning_rate": 0.0001999932760718265,
-      "loss": 46.0,
-      "step": 22932
-    },
-    {
-      "epoch": 3.6931841056403236,
-      "grad_norm": 0.006102294661104679,
-      "learning_rate": 0.00019999327548514127,
-      "loss": 46.0,
-      "step": 22933
-    },
-    {
-      "epoch": 3.693345142719111,
-      "grad_norm": 0.0040518539026379585,
-      "learning_rate": 0.00019999327489843042,
-      "loss": 46.0,
-      "step": 22934
-    },
-    {
-      "epoch": 3.6935061797978985,
-      "grad_norm": 0.008036128245294094,
-      "learning_rate": 0.00019999327431169403,
-      "loss": 46.0,
-      "step": 22935
-    },
-    {
-      "epoch": 3.693667216876686,
-      "grad_norm": 0.003131825476884842,
-      "learning_rate": 0.00019999327372493202,
-      "loss": 46.0,
-      "step": 22936
-    },
-    {
-      "epoch": 3.6938282539554734,
-      "grad_norm": 0.0014921118272468448,
-      "learning_rate": 0.0001999932731381444,
-      "loss": 46.0,
-      "step": 22937
-    },
-    {
-      "epoch": 3.693989291034261,
-      "grad_norm": 0.001481468672864139,
-      "learning_rate": 0.0001999932725513312,
-      "loss": 46.0,
-      "step": 22938
-    },
-    {
-      "epoch": 3.694150328113048,
-      "grad_norm": 0.005379470065236092,
-      "learning_rate": 0.0001999932719644924,
-      "loss": 46.0,
-      "step": 22939
-    },
-    {
-      "epoch": 3.6943113651918353,
-      "grad_norm": 0.00589486351236701,
-      "learning_rate": 0.00019999327137762803,
-      "loss": 46.0,
-      "step": 22940
-    },
-    {
-      "epoch": 3.694472402270623,
-      "grad_norm": 0.005143916234374046,
-      "learning_rate": 0.00019999327079073804,
-      "loss": 46.0,
-      "step": 22941
-    },
-    {
-      "epoch": 3.6946334393494102,
-      "grad_norm": 0.00259154150262475,
-      "learning_rate": 0.00019999327020382246,
-      "loss": 46.0,
-      "step": 22942
-    },
-    {
-      "epoch": 3.6947944764281977,
-      "grad_norm": 0.003596675116568804,
-      "learning_rate": 0.0001999932696168813,
-      "loss": 46.0,
-      "step": 22943
-    },
-    {
-      "epoch": 3.694955513506985,
-      "grad_norm": 0.0015946300700306892,
-      "learning_rate": 0.00019999326902991454,
-      "loss": 46.0,
-      "step": 22944
-    },
-    {
-      "epoch": 3.695116550585772,
-      "grad_norm": 0.0016247582389041781,
-      "learning_rate": 0.0001999932684429222,
-      "loss": 46.0,
-      "step": 22945
-    },
-    {
-      "epoch": 3.6952775876645596,
-      "grad_norm": 0.0006699913647025824,
-      "learning_rate": 0.0001999932678559042,
-      "loss": 46.0,
-      "step": 22946
-    },
-    {
-      "epoch": 3.695438624743347,
-      "grad_norm": 0.001453114440664649,
-      "learning_rate": 0.0001999932672688607,
-      "loss": 46.0,
-      "step": 22947
-    },
-    {
-      "epoch": 3.6955996618221345,
-      "grad_norm": 0.005377592984586954,
-      "learning_rate": 0.00019999326668179157,
-      "loss": 46.0,
-      "step": 22948
-    },
-    {
-      "epoch": 3.695760698900922,
-      "grad_norm": 0.004424140322953463,
-      "learning_rate": 0.00019999326609469685,
-      "loss": 46.0,
-      "step": 22949
-    },
-    {
-      "epoch": 3.6959217359797094,
-      "grad_norm": 0.010377160273492336,
-      "learning_rate": 0.00019999326550757652,
-      "loss": 46.0,
-      "step": 22950
-    },
-    {
-      "epoch": 3.696082773058497,
-      "grad_norm": 0.0022835538256913424,
-      "learning_rate": 0.00019999326492043063,
-      "loss": 46.0,
-      "step": 22951
-    },
-    {
-      "epoch": 3.6962438101372843,
-      "grad_norm": 0.003912990912795067,
-      "learning_rate": 0.00019999326433325912,
-      "loss": 46.0,
-      "step": 22952
-    },
-    {
-      "epoch": 3.6964048472160718,
-      "grad_norm": 0.011481009423732758,
-      "learning_rate": 0.00019999326374606203,
-      "loss": 46.0,
-      "step": 22953
-    },
-    {
-      "epoch": 3.6965658842948588,
-      "grad_norm": 0.007846255786716938,
-      "learning_rate": 0.00019999326315883932,
-      "loss": 46.0,
-      "step": 22954
-    },
-    {
-      "epoch": 3.6967269213736462,
-      "grad_norm": 0.0004045862006023526,
-      "learning_rate": 0.00019999326257159102,
-      "loss": 46.0,
-      "step": 22955
-    },
-    {
-      "epoch": 3.6968879584524337,
-      "grad_norm": 0.002795801032334566,
-      "learning_rate": 0.00019999326198431714,
-      "loss": 46.0,
-      "step": 22956
-    },
-    {
-      "epoch": 3.697048995531221,
-      "grad_norm": 0.004743854980915785,
-      "learning_rate": 0.00019999326139701767,
-      "loss": 46.0,
-      "step": 22957
-    },
-    {
-      "epoch": 3.6972100326100086,
-      "grad_norm": 0.004762520547956228,
-      "learning_rate": 0.00019999326080969262,
-      "loss": 46.0,
-      "step": 22958
-    },
-    {
-      "epoch": 3.6973710696887956,
-      "grad_norm": 0.013277491554617882,
-      "learning_rate": 0.00019999326022234195,
-      "loss": 46.0,
-      "step": 22959
-    },
-    {
-      "epoch": 3.697532106767583,
-      "grad_norm": 0.004574695136398077,
-      "learning_rate": 0.00019999325963496572,
-      "loss": 46.0,
-      "step": 22960
-    },
-    {
-      "epoch": 3.6976931438463705,
-      "grad_norm": 0.008152835071086884,
-      "learning_rate": 0.00019999325904756384,
-      "loss": 46.0,
-      "step": 22961
-    },
-    {
-      "epoch": 3.697854180925158,
-      "grad_norm": 0.0036802445538342,
-      "learning_rate": 0.00019999325846013644,
-      "loss": 46.0,
-      "step": 22962
-    },
-    {
-      "epoch": 3.6980152180039454,
-      "grad_norm": 0.003269928740337491,
-      "learning_rate": 0.0001999932578726834,
-      "loss": 46.0,
-      "step": 22963
-    },
-    {
-      "epoch": 3.698176255082733,
-      "grad_norm": 0.0023690026719123125,
-      "learning_rate": 0.00019999325728520476,
-      "loss": 46.0,
-      "step": 22964
-    },
-    {
-      "epoch": 3.6983372921615203,
-      "grad_norm": 0.002575025660917163,
-      "learning_rate": 0.00019999325669770057,
-      "loss": 46.0,
-      "step": 22965
-    },
-    {
-      "epoch": 3.6984983292403077,
-      "grad_norm": 0.0053553506731987,
-      "learning_rate": 0.00019999325611017076,
-      "loss": 46.0,
-      "step": 22966
-    },
-    {
-      "epoch": 3.698659366319095,
-      "grad_norm": 0.0030848453752696514,
-      "learning_rate": 0.00019999325552261534,
-      "loss": 46.0,
-      "step": 22967
-    },
-    {
-      "epoch": 3.698820403397882,
-      "grad_norm": 0.004077146295458078,
-      "learning_rate": 0.00019999325493503435,
-      "loss": 46.0,
-      "step": 22968
-    },
-    {
-      "epoch": 3.6989814404766697,
-      "grad_norm": 0.011258543469011784,
-      "learning_rate": 0.00019999325434742776,
-      "loss": 46.0,
-      "step": 22969
-    },
-    {
-      "epoch": 3.699142477555457,
-      "grad_norm": 0.001314330380409956,
-      "learning_rate": 0.00019999325375979557,
-      "loss": 46.0,
-      "step": 22970
-    },
-    {
-      "epoch": 3.6993035146342446,
-      "grad_norm": 0.0031963076908141375,
-      "learning_rate": 0.0001999932531721378,
-      "loss": 46.0,
-      "step": 22971
-    },
-    {
-      "epoch": 3.699464551713032,
-      "grad_norm": 0.0034081123303622007,
-      "learning_rate": 0.00019999325258445442,
-      "loss": 46.0,
-      "step": 22972
-    },
-    {
-      "epoch": 3.6996255887918195,
-      "grad_norm": 0.005806222092360258,
-      "learning_rate": 0.00019999325199674547,
-      "loss": 46.0,
-      "step": 22973
-    },
-    {
-      "epoch": 3.6997866258706065,
-      "grad_norm": 0.0014548038598150015,
-      "learning_rate": 0.00019999325140901094,
-      "loss": 46.0,
-      "step": 22974
-    },
-    {
-      "epoch": 3.699947662949394,
-      "grad_norm": 0.003776298835873604,
-      "learning_rate": 0.00019999325082125077,
-      "loss": 46.0,
-      "step": 22975
-    },
-    {
-      "epoch": 3.7001087000281814,
-      "grad_norm": 0.0030455931555479765,
-      "learning_rate": 0.00019999325023346503,
-      "loss": 46.0,
-      "step": 22976
-    },
-    {
-      "epoch": 3.700269737106969,
-      "grad_norm": 0.008189480751752853,
-      "learning_rate": 0.0001999932496456537,
-      "loss": 46.0,
-      "step": 22977
-    },
-    {
-      "epoch": 3.7004307741857563,
-      "grad_norm": 0.004415380302816629,
-      "learning_rate": 0.00019999324905781678,
-      "loss": 46.0,
-      "step": 22978
-    },
-    {
-      "epoch": 3.7005918112645437,
-      "grad_norm": 0.0030416182707995176,
-      "learning_rate": 0.00019999324846995425,
-      "loss": 46.0,
-      "step": 22979
-    },
-    {
-      "epoch": 3.700752848343331,
-      "grad_norm": 0.015108397230505943,
-      "learning_rate": 0.00019999324788206615,
-      "loss": 46.0,
-      "step": 22980
-    },
-    {
-      "epoch": 3.7009138854221186,
-      "grad_norm": 0.011023464612662792,
-      "learning_rate": 0.00019999324729415245,
-      "loss": 46.0,
-      "step": 22981
-    },
-    {
-      "epoch": 3.701074922500906,
-      "grad_norm": 0.005132417194545269,
-      "learning_rate": 0.00019999324670621314,
-      "loss": 46.0,
-      "step": 22982
-    },
-    {
-      "epoch": 3.701235959579693,
-      "grad_norm": 0.0005180487642064691,
-      "learning_rate": 0.00019999324611824824,
-      "loss": 46.0,
-      "step": 22983
-    },
-    {
-      "epoch": 3.7013969966584805,
-      "grad_norm": 0.002225810894742608,
-      "learning_rate": 0.00019999324553025775,
-      "loss": 46.0,
-      "step": 22984
-    },
-    {
-      "epoch": 3.701558033737268,
-      "grad_norm": 0.005416846368461847,
-      "learning_rate": 0.0001999932449422417,
-      "loss": 46.0,
-      "step": 22985
-    },
-    {
-      "epoch": 3.7017190708160554,
-      "grad_norm": 0.00812110211700201,
-      "learning_rate": 0.0001999932443542,
-      "loss": 46.0,
-      "step": 22986
-    },
-    {
-      "epoch": 3.701880107894843,
-      "grad_norm": 0.0024413838982582092,
-      "learning_rate": 0.00019999324376613275,
-      "loss": 46.0,
-      "step": 22987
-    },
-    {
-      "epoch": 3.70204114497363,
-      "grad_norm": 0.0050889188423752785,
-      "learning_rate": 0.0001999932431780399,
-      "loss": 46.0,
-      "step": 22988
-    },
-    {
-      "epoch": 3.7022021820524174,
-      "grad_norm": 0.002696953946724534,
-      "learning_rate": 0.00019999324258992144,
-      "loss": 46.0,
-      "step": 22989
-    },
-    {
-      "epoch": 3.702363219131205,
-      "grad_norm": 0.002093171002343297,
-      "learning_rate": 0.00019999324200177738,
-      "loss": 46.0,
-      "step": 22990
-    },
-    {
-      "epoch": 3.7025242562099923,
-      "grad_norm": 0.0025020111352205276,
-      "learning_rate": 0.00019999324141360775,
-      "loss": 46.0,
-      "step": 22991
-    },
-    {
-      "epoch": 3.7026852932887797,
-      "grad_norm": 0.0035636061802506447,
-      "learning_rate": 0.00019999324082541252,
-      "loss": 46.0,
-      "step": 22992
-    },
-    {
-      "epoch": 3.702846330367567,
-      "grad_norm": 0.003597965929657221,
-      "learning_rate": 0.0001999932402371917,
-      "loss": 46.0,
-      "step": 22993
-    },
-    {
-      "epoch": 3.7030073674463546,
-      "grad_norm": 0.002469766652211547,
-      "learning_rate": 0.00019999323964894528,
-      "loss": 46.0,
-      "step": 22994
-    },
-    {
-      "epoch": 3.703168404525142,
-      "grad_norm": 0.0026130781043320894,
-      "learning_rate": 0.00019999323906067326,
-      "loss": 46.0,
-      "step": 22995
-    },
-    {
-      "epoch": 3.7033294416039295,
-      "grad_norm": 0.008627819828689098,
-      "learning_rate": 0.00019999323847237567,
-      "loss": 46.0,
-      "step": 22996
-    },
-    {
-      "epoch": 3.703490478682717,
-      "grad_norm": 0.006470984313637018,
-      "learning_rate": 0.00019999323788405247,
-      "loss": 46.0,
-      "step": 22997
-    },
-    {
-      "epoch": 3.703651515761504,
-      "grad_norm": 0.013201608322560787,
-      "learning_rate": 0.00019999323729570368,
-      "loss": 46.0,
-      "step": 22998
-    },
-    {
-      "epoch": 3.7038125528402914,
-      "grad_norm": 0.0015339754754677415,
-      "learning_rate": 0.0001999932367073293,
-      "loss": 46.0,
-      "step": 22999
-    },
-    {
-      "epoch": 3.703973589919079,
-      "grad_norm": 0.008167789317667484,
-      "learning_rate": 0.0001999932361189293,
-      "loss": 46.0,
-      "step": 23000
-    },
-    {
-      "epoch": 3.7041346269978663,
-      "grad_norm": 0.001999037340283394,
-      "learning_rate": 0.00019999323553050375,
-      "loss": 46.0,
-      "step": 23001
-    },
-    {
-      "epoch": 3.704295664076654,
-      "grad_norm": 0.0035351796541363,
-      "learning_rate": 0.00019999323494205258,
-      "loss": 46.0,
-      "step": 23002
-    },
-    {
-      "epoch": 3.704456701155441,
-      "grad_norm": 0.008765707723796368,
-      "learning_rate": 0.00019999323435357583,
-      "loss": 46.0,
-      "step": 23003
-    },
-    {
-      "epoch": 3.7046177382342282,
-      "grad_norm": 0.005180766806006432,
-      "learning_rate": 0.0001999932337650735,
-      "loss": 46.0,
-      "step": 23004
-    },
-    {
-      "epoch": 3.7047787753130157,
-      "grad_norm": 0.0035774833522737026,
-      "learning_rate": 0.00019999323317654554,
-      "loss": 46.0,
-      "step": 23005
-    },
-    {
-      "epoch": 3.704939812391803,
-      "grad_norm": 0.0008684717467986047,
-      "learning_rate": 0.000199993232587992,
-      "loss": 46.0,
-      "step": 23006
-    },
-    {
-      "epoch": 3.7051008494705906,
-      "grad_norm": 0.0015333372866734862,
-      "learning_rate": 0.00019999323199941288,
-      "loss": 46.0,
-      "step": 23007
-    },
-    {
-      "epoch": 3.705261886549378,
-      "grad_norm": 0.0018458024132996798,
-      "learning_rate": 0.00019999323141080817,
-      "loss": 46.0,
-      "step": 23008
-    },
-    {
-      "epoch": 3.7054229236281655,
-      "grad_norm": 0.0025055750738829374,
-      "learning_rate": 0.00019999323082217784,
-      "loss": 46.0,
-      "step": 23009
-    },
-    {
-      "epoch": 3.705583960706953,
-      "grad_norm": 0.010124053806066513,
-      "learning_rate": 0.00019999323023352195,
-      "loss": 46.0,
-      "step": 23010
-    },
-    {
-      "epoch": 3.7057449977857404,
-      "grad_norm": 0.0055197617039084435,
-      "learning_rate": 0.00019999322964484045,
-      "loss": 46.0,
-      "step": 23011
-    },
-    {
-      "epoch": 3.7059060348645274,
-      "grad_norm": 0.006251504644751549,
-      "learning_rate": 0.00019999322905613336,
-      "loss": 46.0,
-      "step": 23012
-    },
-    {
-      "epoch": 3.706067071943315,
-      "grad_norm": 0.0018859145930036902,
-      "learning_rate": 0.00019999322846740066,
-      "loss": 46.0,
-      "step": 23013
-    },
-    {
-      "epoch": 3.7062281090221023,
-      "grad_norm": 0.005661732517182827,
-      "learning_rate": 0.00019999322787864237,
-      "loss": 46.0,
-      "step": 23014
-    },
-    {
-      "epoch": 3.7063891461008898,
-      "grad_norm": 0.005678261164575815,
-      "learning_rate": 0.0001999932272898585,
-      "loss": 46.0,
-      "step": 23015
-    },
-    {
-      "epoch": 3.706550183179677,
-      "grad_norm": 0.004850441589951515,
-      "learning_rate": 0.00019999322670104905,
-      "loss": 46.0,
-      "step": 23016
-    },
-    {
-      "epoch": 3.7067112202584647,
-      "grad_norm": 0.0026961183175444603,
-      "learning_rate": 0.00019999322611221398,
-      "loss": 46.0,
-      "step": 23017
-    },
-    {
-      "epoch": 3.7068722573372517,
-      "grad_norm": 0.000925944943446666,
-      "learning_rate": 0.00019999322552335334,
-      "loss": 46.0,
-      "step": 23018
-    },
-    {
-      "epoch": 3.707033294416039,
-      "grad_norm": 0.0021295666228979826,
-      "learning_rate": 0.00019999322493446708,
-      "loss": 46.0,
-      "step": 23019
-    },
-    {
-      "epoch": 3.7071943314948266,
-      "grad_norm": 0.0023116902448236942,
-      "learning_rate": 0.00019999322434555524,
-      "loss": 46.0,
-      "step": 23020
-    },
-    {
-      "epoch": 3.707355368573614,
-      "grad_norm": 0.0020218666177242994,
-      "learning_rate": 0.00019999322375661782,
-      "loss": 46.0,
-      "step": 23021
-    },
-    {
-      "epoch": 3.7075164056524015,
-      "grad_norm": 0.0015538507141172886,
-      "learning_rate": 0.00019999322316765477,
-      "loss": 46.0,
-      "step": 23022
-    },
-    {
-      "epoch": 3.707677442731189,
-      "grad_norm": 0.0073905824683606625,
-      "learning_rate": 0.00019999322257866617,
-      "loss": 46.0,
-      "step": 23023
-    },
-    {
-      "epoch": 3.7078384798099764,
-      "grad_norm": 0.0058021689765155315,
-      "learning_rate": 0.00019999322198965196,
-      "loss": 46.0,
-      "step": 23024
-    },
-    {
-      "epoch": 3.707999516888764,
-      "grad_norm": 0.0016744486056268215,
-      "learning_rate": 0.00019999322140061215,
-      "loss": 46.0,
-      "step": 23025
-    },
-    {
-      "epoch": 3.7081605539675513,
-      "grad_norm": 0.003271284978836775,
-      "learning_rate": 0.00019999322081154674,
-      "loss": 46.0,
-      "step": 23026
-    },
-    {
-      "epoch": 3.7083215910463383,
-      "grad_norm": 0.002253343816846609,
-      "learning_rate": 0.00019999322022245576,
-      "loss": 46.0,
-      "step": 23027
-    },
-    {
-      "epoch": 3.7084826281251257,
-      "grad_norm": 0.0013409563107416034,
-      "learning_rate": 0.00019999321963333917,
-      "loss": 46.0,
-      "step": 23028
-    },
-    {
-      "epoch": 3.708643665203913,
-      "grad_norm": 0.0016886061057448387,
-      "learning_rate": 0.00019999321904419701,
-      "loss": 46.0,
-      "step": 23029
-    },
-    {
-      "epoch": 3.7088047022827006,
-      "grad_norm": 0.006010679993778467,
-      "learning_rate": 0.00019999321845502922,
-      "loss": 46.0,
-      "step": 23030
-    },
-    {
-      "epoch": 3.708965739361488,
-      "grad_norm": 0.005514186341315508,
-      "learning_rate": 0.00019999321786583587,
-      "loss": 46.0,
-      "step": 23031
-    },
-    {
-      "epoch": 3.709126776440275,
-      "grad_norm": 0.004206299316138029,
-      "learning_rate": 0.00019999321727661693,
-      "loss": 46.0,
-      "step": 23032
-    },
-    {
-      "epoch": 3.7092878135190626,
-      "grad_norm": 0.0025531467981636524,
-      "learning_rate": 0.00019999321668737237,
-      "loss": 46.0,
-      "step": 23033
-    },
-    {
-      "epoch": 3.70944885059785,
-      "grad_norm": 0.0012883504386991262,
-      "learning_rate": 0.00019999321609810223,
-      "loss": 46.0,
-      "step": 23034
-    },
-    {
-      "epoch": 3.7096098876766375,
-      "grad_norm": 0.002383655169978738,
-      "learning_rate": 0.0001999932155088065,
-      "loss": 46.0,
-      "step": 23035
-    },
-    {
-      "epoch": 3.709770924755425,
-      "grad_norm": 0.0033795132767409086,
-      "learning_rate": 0.00019999321491948516,
-      "loss": 46.0,
-      "step": 23036
-    },
-    {
-      "epoch": 3.7099319618342124,
-      "grad_norm": 0.001663564471527934,
-      "learning_rate": 0.00019999321433013823,
-      "loss": 46.0,
-      "step": 23037
-    },
-    {
-      "epoch": 3.710092998913,
-      "grad_norm": 0.008418679237365723,
-      "learning_rate": 0.0001999932137407657,
-      "loss": 46.0,
-      "step": 23038
-    },
-    {
-      "epoch": 3.7102540359917873,
-      "grad_norm": 0.0033010654151439667,
-      "learning_rate": 0.00019999321315136763,
-      "loss": 46.0,
-      "step": 23039
-    },
-    {
-      "epoch": 3.7104150730705747,
-      "grad_norm": 0.005354561842978001,
-      "learning_rate": 0.0001999932125619439,
-      "loss": 46.0,
-      "step": 23040
-    },
-    {
-      "epoch": 3.7105761101493617,
-      "grad_norm": 0.008223689161241055,
-      "learning_rate": 0.00019999321197249463,
-      "loss": 46.0,
-      "step": 23041
-    },
-    {
-      "epoch": 3.710737147228149,
-      "grad_norm": 0.004386438522487879,
-      "learning_rate": 0.0001999932113830197,
-      "loss": 46.0,
-      "step": 23042
-    },
-    {
-      "epoch": 3.7108981843069366,
-      "grad_norm": 0.0010133887408301234,
-      "learning_rate": 0.00019999321079351926,
-      "loss": 46.0,
-      "step": 23043
-    },
-    {
-      "epoch": 3.711059221385724,
-      "grad_norm": 0.006740883458405733,
-      "learning_rate": 0.00019999321020399316,
-      "loss": 46.0,
-      "step": 23044
-    },
-    {
-      "epoch": 3.7112202584645115,
-      "grad_norm": 0.0010446736123412848,
-      "learning_rate": 0.0001999932096144415,
-      "loss": 46.0,
-      "step": 23045
-    },
-    {
-      "epoch": 3.711381295543299,
-      "grad_norm": 0.00254206545650959,
-      "learning_rate": 0.00019999320902486424,
-      "loss": 46.0,
-      "step": 23046
-    },
-    {
-      "epoch": 3.711542332622086,
-      "grad_norm": 0.009789474308490753,
-      "learning_rate": 0.0001999932084352614,
-      "loss": 46.0,
-      "step": 23047
-    },
-    {
-      "epoch": 3.7117033697008734,
-      "grad_norm": 0.001101772882975638,
-      "learning_rate": 0.00019999320784563297,
-      "loss": 46.0,
-      "step": 23048
-    },
-    {
-      "epoch": 3.711864406779661,
-      "grad_norm": 0.0047613405622541904,
-      "learning_rate": 0.0001999932072559789,
-      "loss": 46.0,
-      "step": 23049
-    },
-    {
-      "epoch": 3.7120254438584483,
-      "grad_norm": 0.001397734391503036,
-      "learning_rate": 0.0001999932066662993,
-      "loss": 46.0,
-      "step": 23050
-    },
-    {
-      "epoch": 3.712186480937236,
-      "grad_norm": 0.00730739114806056,
-      "learning_rate": 0.00019999320607659406,
-      "loss": 46.0,
-      "step": 23051
-    },
-    {
-      "epoch": 3.7123475180160233,
-      "grad_norm": 0.009815474972128868,
-      "learning_rate": 0.0001999932054868632,
-      "loss": 46.0,
-      "step": 23052
-    },
-    {
-      "epoch": 3.7125085550948107,
-      "grad_norm": 0.0046633221209049225,
-      "learning_rate": 0.00019999320489710683,
-      "loss": 46.0,
-      "step": 23053
-    },
-    {
-      "epoch": 3.712669592173598,
-      "grad_norm": 0.004195145331323147,
-      "learning_rate": 0.0001999932043073248,
-      "loss": 46.0,
-      "step": 23054
-    },
-    {
-      "epoch": 3.7128306292523856,
-      "grad_norm": 0.015399031341075897,
-      "learning_rate": 0.00019999320371751723,
-      "loss": 46.0,
-      "step": 23055
-    },
-    {
-      "epoch": 3.7129916663311726,
-      "grad_norm": 0.0033258432522416115,
-      "learning_rate": 0.000199993203127684,
-      "loss": 46.0,
-      "step": 23056
-    },
-    {
-      "epoch": 3.71315270340996,
-      "grad_norm": 0.0014301176415756345,
-      "learning_rate": 0.00019999320253782522,
-      "loss": 46.0,
-      "step": 23057
-    },
-    {
-      "epoch": 3.7133137404887475,
-      "grad_norm": 0.0015852913493290544,
-      "learning_rate": 0.00019999320194794085,
-      "loss": 46.0,
-      "step": 23058
-    },
-    {
-      "epoch": 3.713474777567535,
-      "grad_norm": 0.009821073152124882,
-      "learning_rate": 0.00019999320135803087,
-      "loss": 46.0,
-      "step": 23059
-    },
-    {
-      "epoch": 3.7136358146463224,
-      "grad_norm": 0.008045095019042492,
-      "learning_rate": 0.00019999320076809533,
-      "loss": 46.0,
-      "step": 23060
-    },
-    {
-      "epoch": 3.71379685172511,
-      "grad_norm": 0.002824934897944331,
-      "learning_rate": 0.00019999320017813417,
-      "loss": 46.0,
-      "step": 23061
-    },
-    {
-      "epoch": 3.713957888803897,
-      "grad_norm": 0.002330234507098794,
-      "learning_rate": 0.00019999319958814745,
-      "loss": 46.0,
-      "step": 23062
-    },
-    {
-      "epoch": 3.7141189258826843,
-      "grad_norm": 0.0033003368880599737,
-      "learning_rate": 0.0001999931989981351,
-      "loss": 46.0,
-      "step": 23063
-    },
-    {
-      "epoch": 3.714279962961472,
-      "grad_norm": 0.0009408897021785378,
-      "learning_rate": 0.00019999319840809717,
-      "loss": 46.0,
-      "step": 23064
-    },
-    {
-      "epoch": 3.7144410000402592,
-      "grad_norm": 0.002168254228308797,
-      "learning_rate": 0.0001999931978180336,
-      "loss": 46.0,
-      "step": 23065
-    },
-    {
-      "epoch": 3.7146020371190467,
-      "grad_norm": 0.004309351556003094,
-      "learning_rate": 0.0001999931972279445,
-      "loss": 46.0,
-      "step": 23066
-    },
-    {
-      "epoch": 3.714763074197834,
-      "grad_norm": 0.003641866846010089,
-      "learning_rate": 0.00019999319663782975,
-      "loss": 46.0,
-      "step": 23067
-    },
-    {
-      "epoch": 3.7149241112766216,
-      "grad_norm": 0.015055439434945583,
-      "learning_rate": 0.00019999319604768946,
-      "loss": 46.0,
-      "step": 23068
-    },
-    {
-      "epoch": 3.715085148355409,
-      "grad_norm": 0.0008330951677635312,
-      "learning_rate": 0.00019999319545752355,
-      "loss": 46.0,
-      "step": 23069
-    },
-    {
-      "epoch": 3.7152461854341965,
-      "grad_norm": 0.0017588816117495298,
-      "learning_rate": 0.00019999319486733205,
-      "loss": 46.0,
-      "step": 23070
-    },
-    {
-      "epoch": 3.7154072225129835,
-      "grad_norm": 0.0015629057306796312,
-      "learning_rate": 0.000199993194277115,
-      "loss": 46.0,
-      "step": 23071
-    },
-    {
-      "epoch": 3.715568259591771,
-      "grad_norm": 0.003930437844246626,
-      "learning_rate": 0.0001999931936868723,
-      "loss": 46.0,
-      "step": 23072
-    },
-    {
-      "epoch": 3.7157292966705584,
-      "grad_norm": 0.016792690381407738,
-      "learning_rate": 0.00019999319309660404,
-      "loss": 46.0,
-      "step": 23073
-    },
-    {
-      "epoch": 3.715890333749346,
-      "grad_norm": 0.004370777402073145,
-      "learning_rate": 0.00019999319250631014,
-      "loss": 46.0,
-      "step": 23074
-    },
-    {
-      "epoch": 3.7160513708281333,
-      "grad_norm": 0.0033224904909729958,
-      "learning_rate": 0.0001999931919159907,
-      "loss": 46.0,
-      "step": 23075
-    },
-    {
-      "epoch": 3.7162124079069203,
-      "grad_norm": 0.0037025297060608864,
-      "learning_rate": 0.00019999319132564566,
-      "loss": 46.0,
-      "step": 23076
-    },
-    {
-      "epoch": 3.7163734449857078,
-      "grad_norm": 0.0005642303149215877,
-      "learning_rate": 0.000199993190735275,
-      "loss": 46.0,
-      "step": 23077
-    },
-    {
-      "epoch": 3.716534482064495,
-      "grad_norm": 0.004197478760033846,
-      "learning_rate": 0.00019999319014487877,
-      "loss": 46.0,
-      "step": 23078
-    },
-    {
-      "epoch": 3.7166955191432827,
-      "grad_norm": 0.0027420350816100836,
-      "learning_rate": 0.00019999318955445694,
-      "loss": 46.0,
-      "step": 23079
-    },
-    {
-      "epoch": 3.71685655622207,
-      "grad_norm": 0.0030675502493977547,
-      "learning_rate": 0.0001999931889640095,
-      "loss": 46.0,
-      "step": 23080
-    },
-    {
-      "epoch": 3.7170175933008576,
-      "grad_norm": 0.013590652495622635,
-      "learning_rate": 0.00019999318837353648,
-      "loss": 46.0,
-      "step": 23081
-    },
-    {
-      "epoch": 3.717178630379645,
-      "grad_norm": 0.003240320598706603,
-      "learning_rate": 0.00019999318778303785,
-      "loss": 46.0,
-      "step": 23082
-    },
-    {
-      "epoch": 3.7173396674584325,
-      "grad_norm": 0.002534376922994852,
-      "learning_rate": 0.00019999318719251364,
-      "loss": 46.0,
-      "step": 23083
-    },
-    {
-      "epoch": 3.71750070453722,
-      "grad_norm": 0.003964030183851719,
-      "learning_rate": 0.00019999318660196384,
-      "loss": 46.0,
-      "step": 23084
-    },
-    {
-      "epoch": 3.717661741616007,
-      "grad_norm": 0.00806252472102642,
-      "learning_rate": 0.00019999318601138848,
-      "loss": 46.0,
-      "step": 23085
-    },
-    {
-      "epoch": 3.7178227786947944,
-      "grad_norm": 0.004998079035431147,
-      "learning_rate": 0.00019999318542078748,
-      "loss": 46.0,
-      "step": 23086
-    },
-    {
-      "epoch": 3.717983815773582,
-      "grad_norm": 0.0014475666685029864,
-      "learning_rate": 0.00019999318483016092,
-      "loss": 46.0,
-      "step": 23087
-    },
-    {
-      "epoch": 3.7181448528523693,
-      "grad_norm": 0.006805706303566694,
-      "learning_rate": 0.00019999318423950872,
-      "loss": 46.0,
-      "step": 23088
-    },
-    {
-      "epoch": 3.7183058899311567,
-      "grad_norm": 0.0072632646188139915,
-      "learning_rate": 0.00019999318364883096,
-      "loss": 46.0,
-      "step": 23089
-    },
-    {
-      "epoch": 3.718466927009944,
-      "grad_norm": 0.009562850929796696,
-      "learning_rate": 0.0001999931830581276,
-      "loss": 46.0,
-      "step": 23090
-    },
-    {
-      "epoch": 3.718627964088731,
-      "grad_norm": 0.0031430628150701523,
-      "learning_rate": 0.00019999318246739867,
-      "loss": 46.0,
-      "step": 23091
-    },
-    {
-      "epoch": 3.7187890011675186,
-      "grad_norm": 0.005490915384143591,
-      "learning_rate": 0.0001999931818766441,
-      "loss": 46.0,
-      "step": 23092
-    },
-    {
-      "epoch": 3.718950038246306,
-      "grad_norm": 0.0036144601181149483,
-      "learning_rate": 0.00019999318128586396,
-      "loss": 46.0,
-      "step": 23093
-    },
-    {
-      "epoch": 3.7191110753250936,
-      "grad_norm": 0.002688084961846471,
-      "learning_rate": 0.00019999318069505826,
-      "loss": 46.0,
-      "step": 23094
-    },
-    {
-      "epoch": 3.719272112403881,
-      "grad_norm": 0.007703827228397131,
-      "learning_rate": 0.00019999318010422692,
-      "loss": 46.0,
-      "step": 23095
-    },
-    {
-      "epoch": 3.7194331494826685,
-      "grad_norm": 0.0013276677345857024,
-      "learning_rate": 0.00019999317951337,
-      "loss": 46.0,
-      "step": 23096
-    },
-    {
-      "epoch": 3.719594186561456,
-      "grad_norm": 0.0012462981976568699,
-      "learning_rate": 0.0001999931789224875,
-      "loss": 46.0,
-      "step": 23097
-    },
-    {
-      "epoch": 3.7197552236402434,
-      "grad_norm": 0.005808768328279257,
-      "learning_rate": 0.0001999931783315794,
-      "loss": 46.0,
-      "step": 23098
-    },
-    {
-      "epoch": 3.719916260719031,
-      "grad_norm": 0.005260785110294819,
-      "learning_rate": 0.0001999931777406457,
-      "loss": 46.0,
-      "step": 23099
-    },
-    {
-      "epoch": 3.720077297797818,
-      "grad_norm": 0.0008696753066033125,
-      "learning_rate": 0.00019999317714968642,
-      "loss": 46.0,
-      "step": 23100
-    },
-    {
-      "epoch": 3.7202383348766053,
-      "grad_norm": 0.002557438099756837,
-      "learning_rate": 0.00019999317655870153,
-      "loss": 46.0,
-      "step": 23101
-    },
-    {
-      "epoch": 3.7203993719553927,
-      "grad_norm": 0.005627449601888657,
-      "learning_rate": 0.00019999317596769106,
-      "loss": 46.0,
-      "step": 23102
-    },
-    {
-      "epoch": 3.72056040903418,
-      "grad_norm": 0.004101383034139872,
-      "learning_rate": 0.000199993175376655,
-      "loss": 46.0,
-      "step": 23103
-    },
-    {
-      "epoch": 3.7207214461129676,
-      "grad_norm": 0.0019744932651519775,
-      "learning_rate": 0.00019999317478559334,
-      "loss": 46.0,
-      "step": 23104
-    },
-    {
-      "epoch": 3.7208824831917546,
-      "grad_norm": 0.002369930734857917,
-      "learning_rate": 0.00019999317419450608,
-      "loss": 46.0,
-      "step": 23105
-    },
-    {
-      "epoch": 3.721043520270542,
-      "grad_norm": 0.004191146697849035,
-      "learning_rate": 0.00019999317360339322,
-      "loss": 46.0,
-      "step": 23106
-    },
-    {
-      "epoch": 3.7212045573493295,
-      "grad_norm": 0.0037796811666339636,
-      "learning_rate": 0.00019999317301225479,
-      "loss": 46.0,
-      "step": 23107
-    },
-    {
-      "epoch": 3.721365594428117,
-      "grad_norm": 0.0052701495587825775,
-      "learning_rate": 0.00019999317242109076,
-      "loss": 46.0,
-      "step": 23108
-    },
-    {
-      "epoch": 3.7215266315069044,
-      "grad_norm": 0.0016494878800585866,
-      "learning_rate": 0.00019999317182990112,
-      "loss": 46.0,
-      "step": 23109
-    },
-    {
-      "epoch": 3.721687668585692,
-      "grad_norm": 0.0018284859834238887,
-      "learning_rate": 0.00019999317123868592,
-      "loss": 46.0,
-      "step": 23110
-    },
-    {
-      "epoch": 3.7218487056644793,
-      "grad_norm": 0.0016238036332651973,
-      "learning_rate": 0.00019999317064744508,
-      "loss": 46.0,
-      "step": 23111
-    },
-    {
-      "epoch": 3.722009742743267,
-      "grad_norm": 0.00222844653762877,
-      "learning_rate": 0.00019999317005617867,
-      "loss": 46.0,
-      "step": 23112
-    },
-    {
-      "epoch": 3.7221707798220542,
-      "grad_norm": 0.017566602677106857,
-      "learning_rate": 0.00019999316946488666,
-      "loss": 46.0,
-      "step": 23113
-    },
-    {
-      "epoch": 3.7223318169008417,
-      "grad_norm": 0.0033925585448741913,
-      "learning_rate": 0.00019999316887356908,
-      "loss": 46.0,
-      "step": 23114
-    },
-    {
-      "epoch": 3.7224928539796287,
-      "grad_norm": 0.0023799650371074677,
-      "learning_rate": 0.0001999931682822259,
-      "loss": 46.0,
-      "step": 23115
-    },
-    {
-      "epoch": 3.722653891058416,
-      "grad_norm": 0.0009959266753867269,
-      "learning_rate": 0.0001999931676908571,
-      "loss": 46.0,
-      "step": 23116
-    },
-    {
-      "epoch": 3.7228149281372036,
-      "grad_norm": 0.004000527318567038,
-      "learning_rate": 0.00019999316709946277,
-      "loss": 46.0,
-      "step": 23117
-    },
-    {
-      "epoch": 3.722975965215991,
-      "grad_norm": 0.0026175882667303085,
-      "learning_rate": 0.0001999931665080428,
-      "loss": 46.0,
-      "step": 23118
-    },
-    {
-      "epoch": 3.7231370022947785,
-      "grad_norm": 0.002381636295467615,
-      "learning_rate": 0.00019999316591659723,
-      "loss": 46.0,
-      "step": 23119
-    },
-    {
-      "epoch": 3.7232980393735655,
-      "grad_norm": 0.011801318265497684,
-      "learning_rate": 0.00019999316532512607,
-      "loss": 46.0,
-      "step": 23120
-    },
-    {
-      "epoch": 3.723459076452353,
-      "grad_norm": 0.003851519664749503,
-      "learning_rate": 0.00019999316473362933,
-      "loss": 46.0,
-      "step": 23121
-    },
-    {
-      "epoch": 3.7236201135311404,
-      "grad_norm": 0.0029754312708973885,
-      "learning_rate": 0.00019999316414210698,
-      "loss": 46.0,
-      "step": 23122
-    },
-    {
-      "epoch": 3.723781150609928,
-      "grad_norm": 0.005113998427987099,
-      "learning_rate": 0.00019999316355055906,
-      "loss": 46.0,
-      "step": 23123
-    },
-    {
-      "epoch": 3.7239421876887153,
-      "grad_norm": 0.0030976508278399706,
-      "learning_rate": 0.00019999316295898553,
-      "loss": 46.0,
-      "step": 23124
-    },
-    {
-      "epoch": 3.7241032247675028,
-      "grad_norm": 0.0025091266725212336,
-      "learning_rate": 0.00019999316236738644,
-      "loss": 46.0,
-      "step": 23125
-    },
-    {
-      "epoch": 3.7242642618462902,
-      "grad_norm": 0.011517970822751522,
-      "learning_rate": 0.00019999316177576174,
-      "loss": 46.0,
-      "step": 23126
-    },
-    {
-      "epoch": 3.7244252989250777,
-      "grad_norm": 0.0062842643819749355,
-      "learning_rate": 0.0001999931611841114,
-      "loss": 46.0,
-      "step": 23127
-    },
-    {
-      "epoch": 3.724586336003865,
-      "grad_norm": 0.00368786184117198,
-      "learning_rate": 0.00019999316059243552,
-      "loss": 46.0,
-      "step": 23128
-    },
-    {
-      "epoch": 3.724747373082652,
-      "grad_norm": 0.008995292708277702,
-      "learning_rate": 0.00019999316000073402,
-      "loss": 46.0,
-      "step": 23129
-    },
-    {
-      "epoch": 3.7249084101614396,
-      "grad_norm": 0.0016693223733454943,
-      "learning_rate": 0.00019999315940900694,
-      "loss": 46.0,
-      "step": 23130
-    },
-    {
-      "epoch": 3.725069447240227,
-      "grad_norm": 0.0010223229182884097,
-      "learning_rate": 0.00019999315881725428,
-      "loss": 46.0,
-      "step": 23131
-    },
-    {
-      "epoch": 3.7252304843190145,
-      "grad_norm": 0.00498659722507,
-      "learning_rate": 0.000199993158225476,
-      "loss": 46.0,
-      "step": 23132
-    },
-    {
-      "epoch": 3.725391521397802,
-      "grad_norm": 0.005208555608987808,
-      "learning_rate": 0.00019999315763367213,
-      "loss": 46.0,
-      "step": 23133
-    },
-    {
-      "epoch": 3.7255525584765894,
-      "grad_norm": 0.0015590887051075697,
-      "learning_rate": 0.0001999931570418427,
-      "loss": 46.0,
-      "step": 23134
-    },
-    {
-      "epoch": 3.7257135955553764,
-      "grad_norm": 0.008314471691846848,
-      "learning_rate": 0.00019999315644998763,
-      "loss": 46.0,
-      "step": 23135
-    },
-    {
-      "epoch": 3.725874632634164,
-      "grad_norm": 0.015822215005755424,
-      "learning_rate": 0.000199993155858107,
-      "loss": 46.0,
-      "step": 23136
-    },
-    {
-      "epoch": 3.7260356697129513,
-      "grad_norm": 0.022752514109015465,
-      "learning_rate": 0.00019999315526620076,
-      "loss": 46.0,
-      "step": 23137
-    },
-    {
-      "epoch": 3.7261967067917388,
-      "grad_norm": 0.0013210473116487265,
-      "learning_rate": 0.00019999315467426892,
-      "loss": 46.0,
-      "step": 23138
-    },
-    {
-      "epoch": 3.726357743870526,
-      "grad_norm": 0.008984082378447056,
-      "learning_rate": 0.0001999931540823115,
-      "loss": 46.0,
-      "step": 23139
-    },
-    {
-      "epoch": 3.7265187809493137,
-      "grad_norm": 0.001040121540427208,
-      "learning_rate": 0.0001999931534903285,
-      "loss": 46.0,
-      "step": 23140
-    },
-    {
-      "epoch": 3.726679818028101,
-      "grad_norm": 0.0020976795349270105,
-      "learning_rate": 0.00019999315289831988,
-      "loss": 46.0,
-      "step": 23141
-    },
-    {
-      "epoch": 3.7268408551068886,
-      "grad_norm": 0.009401296265423298,
-      "learning_rate": 0.00019999315230628567,
-      "loss": 46.0,
-      "step": 23142
-    },
-    {
-      "epoch": 3.727001892185676,
-      "grad_norm": 0.00599231431260705,
-      "learning_rate": 0.0001999931517142259,
-      "loss": 46.0,
-      "step": 23143
-    },
-    {
-      "epoch": 3.727162929264463,
-      "grad_norm": 0.003892629174515605,
-      "learning_rate": 0.0001999931511221405,
-      "loss": 46.0,
-      "step": 23144
-    },
-    {
-      "epoch": 3.7273239663432505,
-      "grad_norm": 0.0030356482602655888,
-      "learning_rate": 0.0001999931505300295,
-      "loss": 46.0,
-      "step": 23145
-    },
-    {
-      "epoch": 3.727485003422038,
-      "grad_norm": 0.0031859027221798897,
-      "learning_rate": 0.00019999314993789295,
-      "loss": 46.0,
-      "step": 23146
-    },
-    {
-      "epoch": 3.7276460405008254,
-      "grad_norm": 0.015063718892633915,
-      "learning_rate": 0.00019999314934573078,
-      "loss": 46.0,
-      "step": 23147
-    },
-    {
-      "epoch": 3.727807077579613,
-      "grad_norm": 0.002772016217932105,
-      "learning_rate": 0.00019999314875354305,
-      "loss": 46.0,
-      "step": 23148
-    },
-    {
-      "epoch": 3.7279681146584,
-      "grad_norm": 0.0008904255810193717,
-      "learning_rate": 0.00019999314816132968,
-      "loss": 46.0,
-      "step": 23149
-    },
-    {
-      "epoch": 3.7281291517371873,
-      "grad_norm": 0.0014147789916023612,
-      "learning_rate": 0.00019999314756909074,
-      "loss": 46.0,
-      "step": 23150
-    },
-    {
-      "epoch": 3.7282901888159747,
-      "grad_norm": 0.0038090685848146677,
-      "learning_rate": 0.00019999314697682617,
-      "loss": 46.0,
-      "step": 23151
-    },
-    {
-      "epoch": 3.728451225894762,
-      "grad_norm": 0.0018200924387201667,
-      "learning_rate": 0.00019999314638453607,
-      "loss": 46.0,
-      "step": 23152
-    },
-    {
-      "epoch": 3.7286122629735496,
-      "grad_norm": 0.003739455947652459,
-      "learning_rate": 0.00019999314579222035,
-      "loss": 46.0,
-      "step": 23153
-    },
-    {
-      "epoch": 3.728773300052337,
-      "grad_norm": 0.013962958008050919,
-      "learning_rate": 0.000199993145199879,
-      "loss": 46.0,
-      "step": 23154
-    },
-    {
-      "epoch": 3.7289343371311245,
-      "grad_norm": 0.0044592684134840965,
-      "learning_rate": 0.00019999314460751212,
-      "loss": 46.0,
-      "step": 23155
-    },
-    {
-      "epoch": 3.729095374209912,
-      "grad_norm": 0.00839564111083746,
-      "learning_rate": 0.0001999931440151196,
-      "loss": 46.0,
-      "step": 23156
-    },
-    {
-      "epoch": 3.7292564112886994,
-      "grad_norm": 0.011357411742210388,
-      "learning_rate": 0.0001999931434227015,
-      "loss": 46.0,
-      "step": 23157
-    },
-    {
-      "epoch": 3.7294174483674865,
-      "grad_norm": 0.0008929366013035178,
-      "learning_rate": 0.00019999314283025783,
-      "loss": 46.0,
-      "step": 23158
-    },
-    {
-      "epoch": 3.729578485446274,
-      "grad_norm": 0.0022368384525179863,
-      "learning_rate": 0.00019999314223778853,
-      "loss": 46.0,
-      "step": 23159
-    },
-    {
-      "epoch": 3.7297395225250614,
-      "grad_norm": 0.00394734600558877,
-      "learning_rate": 0.00019999314164529367,
-      "loss": 46.0,
-      "step": 23160
-    },
-    {
-      "epoch": 3.729900559603849,
-      "grad_norm": 0.002173580462113023,
-      "learning_rate": 0.0001999931410527732,
-      "loss": 46.0,
-      "step": 23161
-    },
-    {
-      "epoch": 3.7300615966826363,
-      "grad_norm": 0.0026979295071214437,
-      "learning_rate": 0.00019999314046022714,
-      "loss": 46.0,
-      "step": 23162
-    },
-    {
-      "epoch": 3.7302226337614237,
-      "grad_norm": 0.0017331313574686646,
-      "learning_rate": 0.00019999313986765547,
-      "loss": 46.0,
-      "step": 23163
-    },
-    {
-      "epoch": 3.7303836708402107,
-      "grad_norm": 0.0057273744605481625,
-      "learning_rate": 0.00019999313927505824,
-      "loss": 46.0,
-      "step": 23164
-    },
-    {
-      "epoch": 3.730544707918998,
-      "grad_norm": 0.012292359955608845,
-      "learning_rate": 0.0001999931386824354,
-      "loss": 46.0,
-      "step": 23165
-    },
-    {
-      "epoch": 3.7307057449977856,
-      "grad_norm": 0.006643029861152172,
-      "learning_rate": 0.00019999313808978695,
-      "loss": 46.0,
-      "step": 23166
-    },
-    {
-      "epoch": 3.730866782076573,
-      "grad_norm": 0.0033334516920149326,
-      "learning_rate": 0.00019999313749711293,
-      "loss": 46.0,
-      "step": 23167
-    },
-    {
-      "epoch": 3.7310278191553605,
-      "grad_norm": 0.0030398694798350334,
-      "learning_rate": 0.0001999931369044133,
-      "loss": 46.0,
-      "step": 23168
-    },
-    {
-      "epoch": 3.731188856234148,
-      "grad_norm": 0.006750424858182669,
-      "learning_rate": 0.00019999313631168807,
-      "loss": 46.0,
-      "step": 23169
-    },
-    {
-      "epoch": 3.7313498933129354,
-      "grad_norm": 0.012335010804235935,
-      "learning_rate": 0.0001999931357189373,
-      "loss": 46.0,
-      "step": 23170
-    },
-    {
-      "epoch": 3.731510930391723,
-      "grad_norm": 0.010827120393514633,
-      "learning_rate": 0.0001999931351261609,
-      "loss": 46.0,
-      "step": 23171
-    },
-    {
-      "epoch": 3.7316719674705103,
-      "grad_norm": 0.003413174068555236,
-      "learning_rate": 0.0001999931345333589,
-      "loss": 46.0,
-      "step": 23172
-    },
-    {
-      "epoch": 3.7318330045492973,
-      "grad_norm": 0.005273792427033186,
-      "learning_rate": 0.0001999931339405313,
-      "loss": 46.0,
-      "step": 23173
-    },
-    {
-      "epoch": 3.731994041628085,
-      "grad_norm": 0.0027752777095884085,
-      "learning_rate": 0.00019999313334767812,
-      "loss": 46.0,
-      "step": 23174
-    },
-    {
-      "epoch": 3.7321550787068722,
-      "grad_norm": 0.00590770086273551,
-      "learning_rate": 0.00019999313275479935,
-      "loss": 46.0,
-      "step": 23175
-    },
-    {
-      "epoch": 3.7323161157856597,
-      "grad_norm": 0.007024338003247976,
-      "learning_rate": 0.000199993132161895,
-      "loss": 46.0,
-      "step": 23176
-    },
-    {
-      "epoch": 3.732477152864447,
-      "grad_norm": 0.0032840659841895103,
-      "learning_rate": 0.000199993131568965,
-      "loss": 46.0,
-      "step": 23177
-    },
-    {
-      "epoch": 3.732638189943234,
-      "grad_norm": 0.004606082569807768,
-      "learning_rate": 0.00019999313097600948,
-      "loss": 46.0,
-      "step": 23178
-    },
-    {
-      "epoch": 3.7327992270220216,
-      "grad_norm": 0.0018258269410580397,
-      "learning_rate": 0.00019999313038302833,
-      "loss": 46.0,
-      "step": 23179
-    },
-    {
-      "epoch": 3.732960264100809,
-      "grad_norm": 0.001961832633242011,
-      "learning_rate": 0.0001999931297900216,
-      "loss": 46.0,
-      "step": 23180
-    },
-    {
-      "epoch": 3.7331213011795965,
-      "grad_norm": 0.003131021047011018,
-      "learning_rate": 0.00019999312919698924,
-      "loss": 46.0,
-      "step": 23181
-    },
-    {
-      "epoch": 3.733282338258384,
-      "grad_norm": 0.003571550128981471,
-      "learning_rate": 0.00019999312860393133,
-      "loss": 46.0,
-      "step": 23182
-    },
-    {
-      "epoch": 3.7334433753371714,
-      "grad_norm": 0.002121776808053255,
-      "learning_rate": 0.0001999931280108478,
-      "loss": 46.0,
-      "step": 23183
-    },
-    {
-      "epoch": 3.733604412415959,
-      "grad_norm": 0.0038131901528686285,
-      "learning_rate": 0.00019999312741773872,
-      "loss": 46.0,
-      "step": 23184
-    },
-    {
-      "epoch": 3.7337654494947463,
-      "grad_norm": 0.00986417755484581,
-      "learning_rate": 0.000199993126824604,
-      "loss": 46.0,
-      "step": 23185
-    },
-    {
-      "epoch": 3.7339264865735338,
-      "grad_norm": 0.002036524470895529,
-      "learning_rate": 0.0001999931262314437,
-      "loss": 46.0,
-      "step": 23186
-    },
-    {
-      "epoch": 3.734087523652321,
-      "grad_norm": 0.010031732730567455,
-      "learning_rate": 0.0001999931256382578,
-      "loss": 46.0,
-      "step": 23187
-    },
-    {
-      "epoch": 3.7342485607311082,
-      "grad_norm": 0.003502281615510583,
-      "learning_rate": 0.00019999312504504632,
-      "loss": 46.0,
-      "step": 23188
-    },
-    {
-      "epoch": 3.7344095978098957,
-      "grad_norm": 0.006705968640744686,
-      "learning_rate": 0.00019999312445180927,
-      "loss": 46.0,
-      "step": 23189
-    },
-    {
-      "epoch": 3.734570634888683,
-      "grad_norm": 0.0026294023264199495,
-      "learning_rate": 0.00019999312385854658,
-      "loss": 46.0,
-      "step": 23190
-    },
-    {
-      "epoch": 3.7347316719674706,
-      "grad_norm": 0.004401372279971838,
-      "learning_rate": 0.00019999312326525833,
-      "loss": 46.0,
-      "step": 23191
-    },
-    {
-      "epoch": 3.734892709046258,
-      "grad_norm": 0.007621718104928732,
-      "learning_rate": 0.00019999312267194447,
-      "loss": 46.0,
-      "step": 23192
-    },
-    {
-      "epoch": 3.735053746125045,
-      "grad_norm": 0.0008155046962201595,
-      "learning_rate": 0.00019999312207860502,
-      "loss": 46.0,
-      "step": 23193
-    },
-    {
-      "epoch": 3.7352147832038325,
-      "grad_norm": 0.001990322256460786,
-      "learning_rate": 0.00019999312148523995,
-      "loss": 46.0,
-      "step": 23194
-    },
-    {
-      "epoch": 3.73537582028262,
-      "grad_norm": 0.0029427832923829556,
-      "learning_rate": 0.00019999312089184933,
-      "loss": 46.0,
-      "step": 23195
-    },
-    {
-      "epoch": 3.7355368573614074,
-      "grad_norm": 0.01669461838901043,
-      "learning_rate": 0.00019999312029843312,
-      "loss": 46.0,
-      "step": 23196
-    },
-    {
-      "epoch": 3.735697894440195,
-      "grad_norm": 0.0012514758855104446,
-      "learning_rate": 0.00019999311970499132,
-      "loss": 46.0,
-      "step": 23197
-    },
-    {
-      "epoch": 3.7358589315189823,
-      "grad_norm": 0.0012641752837225795,
-      "learning_rate": 0.0001999931191115239,
-      "loss": 46.0,
-      "step": 23198
-    },
-    {
-      "epoch": 3.7360199685977697,
-      "grad_norm": 0.0016014878638088703,
-      "learning_rate": 0.00019999311851803087,
-      "loss": 46.0,
-      "step": 23199
-    },
-    {
-      "epoch": 3.736181005676557,
-      "grad_norm": 0.005642854608595371,
-      "learning_rate": 0.00019999311792451229,
-      "loss": 46.0,
-      "step": 23200
-    },
-    {
-      "epoch": 3.7363420427553447,
-      "grad_norm": 0.0017803406808525324,
-      "learning_rate": 0.00019999311733096808,
-      "loss": 46.0,
-      "step": 23201
-    },
-    {
-      "epoch": 3.7365030798341317,
-      "grad_norm": 0.003669108496978879,
-      "learning_rate": 0.0001999931167373983,
-      "loss": 46.0,
-      "step": 23202
-    },
-    {
-      "epoch": 3.736664116912919,
-      "grad_norm": 0.004593828693032265,
-      "learning_rate": 0.00019999311614380292,
-      "loss": 46.0,
-      "step": 23203
-    },
-    {
-      "epoch": 3.7368251539917066,
-      "grad_norm": 0.0029747160151600838,
-      "learning_rate": 0.00019999311555018195,
-      "loss": 46.0,
-      "step": 23204
-    },
-    {
-      "epoch": 3.736986191070494,
-      "grad_norm": 0.003492704126983881,
-      "learning_rate": 0.00019999311495653537,
-      "loss": 46.0,
-      "step": 23205
-    },
-    {
-      "epoch": 3.7371472281492815,
-      "grad_norm": 0.002444078680127859,
-      "learning_rate": 0.00019999311436286324,
-      "loss": 46.0,
-      "step": 23206
-    },
-    {
-      "epoch": 3.737308265228069,
-      "grad_norm": 0.0004748831270262599,
-      "learning_rate": 0.00019999311376916548,
-      "loss": 46.0,
-      "step": 23207
-    },
-    {
-      "epoch": 3.737469302306856,
-      "grad_norm": 0.0024681733921170235,
-      "learning_rate": 0.00019999311317544212,
-      "loss": 46.0,
-      "step": 23208
-    },
-    {
-      "epoch": 3.7376303393856434,
-      "grad_norm": 0.007616622839123011,
-      "learning_rate": 0.0001999931125816932,
-      "loss": 46.0,
-      "step": 23209
-    },
-    {
-      "epoch": 3.737791376464431,
-      "grad_norm": 0.006673749536275864,
-      "learning_rate": 0.00019999311198791865,
-      "loss": 46.0,
-      "step": 23210
-    },
-    {
-      "epoch": 3.7379524135432183,
-      "grad_norm": 0.0015464113093912601,
-      "learning_rate": 0.00019999311139411852,
-      "loss": 46.0,
-      "step": 23211
-    },
-    {
-      "epoch": 3.7381134506220057,
-      "grad_norm": 0.003553914837539196,
-      "learning_rate": 0.0001999931108002928,
-      "loss": 46.0,
-      "step": 23212
-    },
-    {
-      "epoch": 3.738274487700793,
-      "grad_norm": 0.004218549467623234,
-      "learning_rate": 0.0001999931102064415,
-      "loss": 46.0,
-      "step": 23213
-    },
-    {
-      "epoch": 3.7384355247795806,
-      "grad_norm": 0.000705431797541678,
-      "learning_rate": 0.00019999310961256458,
-      "loss": 46.0,
-      "step": 23214
-    },
-    {
-      "epoch": 3.738596561858368,
-      "grad_norm": 0.004536414984613657,
-      "learning_rate": 0.0001999931090186621,
-      "loss": 46.0,
-      "step": 23215
-    },
-    {
-      "epoch": 3.7387575989371555,
-      "grad_norm": 0.0033399872481822968,
-      "learning_rate": 0.000199993108424734,
-      "loss": 46.0,
-      "step": 23216
-    },
-    {
-      "epoch": 3.7389186360159425,
-      "grad_norm": 0.002238695975393057,
-      "learning_rate": 0.0001999931078307803,
-      "loss": 46.0,
-      "step": 23217
-    },
-    {
-      "epoch": 3.73907967309473,
-      "grad_norm": 0.008804631419479847,
-      "learning_rate": 0.00019999310723680104,
-      "loss": 46.0,
-      "step": 23218
-    },
-    {
-      "epoch": 3.7392407101735174,
-      "grad_norm": 0.0005120678688399494,
-      "learning_rate": 0.0001999931066427962,
-      "loss": 46.0,
-      "step": 23219
-    },
-    {
-      "epoch": 3.739401747252305,
-      "grad_norm": 0.006160099990665913,
-      "learning_rate": 0.0001999931060487657,
-      "loss": 46.0,
-      "step": 23220
-    },
-    {
-      "epoch": 3.7395627843310923,
-      "grad_norm": 0.0013807618524879217,
-      "learning_rate": 0.00019999310545470967,
-      "loss": 46.0,
-      "step": 23221
-    },
-    {
-      "epoch": 3.7397238214098794,
-      "grad_norm": 0.0012846424942836165,
-      "learning_rate": 0.000199993104860628,
-      "loss": 46.0,
-      "step": 23222
-    },
-    {
-      "epoch": 3.739884858488667,
-      "grad_norm": 0.0010511906584724784,
-      "learning_rate": 0.00019999310426652077,
-      "loss": 46.0,
-      "step": 23223
-    },
-    {
-      "epoch": 3.7400458955674543,
-      "grad_norm": 0.0033722578082233667,
-      "learning_rate": 0.00019999310367238793,
-      "loss": 46.0,
-      "step": 23224
-    },
-    {
-      "epoch": 3.7402069326462417,
-      "grad_norm": 0.0027915111277252436,
-      "learning_rate": 0.00019999310307822953,
-      "loss": 46.0,
-      "step": 23225
-    },
-    {
-      "epoch": 3.740367969725029,
-      "grad_norm": 0.0012366075534373522,
-      "learning_rate": 0.00019999310248404548,
-      "loss": 46.0,
-      "step": 23226
-    },
-    {
-      "epoch": 3.7405290068038166,
-      "grad_norm": 0.008099728263914585,
-      "learning_rate": 0.00019999310188983588,
-      "loss": 46.0,
-      "step": 23227
-    },
-    {
-      "epoch": 3.740690043882604,
-      "grad_norm": 0.013237777166068554,
-      "learning_rate": 0.00019999310129560066,
-      "loss": 46.0,
-      "step": 23228
-    },
-    {
-      "epoch": 3.7408510809613915,
-      "grad_norm": 0.0014385567046701908,
-      "learning_rate": 0.00019999310070133985,
-      "loss": 46.0,
-      "step": 23229
-    },
-    {
-      "epoch": 3.741012118040179,
-      "grad_norm": 0.005096824374049902,
-      "learning_rate": 0.00019999310010705346,
-      "loss": 46.0,
-      "step": 23230
-    },
-    {
-      "epoch": 3.741173155118966,
-      "grad_norm": 0.014582911506295204,
-      "learning_rate": 0.00019999309951274148,
-      "loss": 46.0,
-      "step": 23231
-    },
-    {
-      "epoch": 3.7413341921977534,
-      "grad_norm": 0.007373742293566465,
-      "learning_rate": 0.00019999309891840388,
-      "loss": 46.0,
-      "step": 23232
-    },
-    {
-      "epoch": 3.741495229276541,
-      "grad_norm": 0.0041610137559473515,
-      "learning_rate": 0.00019999309832404073,
-      "loss": 46.0,
-      "step": 23233
-    },
-    {
-      "epoch": 3.7416562663553283,
-      "grad_norm": 0.0027232591528445482,
-      "learning_rate": 0.00019999309772965196,
-      "loss": 46.0,
-      "step": 23234
-    },
-    {
-      "epoch": 3.741817303434116,
-      "grad_norm": 0.010877788066864014,
-      "learning_rate": 0.0001999930971352376,
-      "loss": 46.0,
-      "step": 23235
-    },
-    {
-      "epoch": 3.7419783405129032,
-      "grad_norm": 0.005337840877473354,
-      "learning_rate": 0.00019999309654079766,
-      "loss": 46.0,
-      "step": 23236
-    },
-    {
-      "epoch": 3.7421393775916902,
-      "grad_norm": 0.005943040829151869,
-      "learning_rate": 0.00019999309594633208,
-      "loss": 46.0,
-      "step": 23237
-    },
-    {
-      "epoch": 3.7423004146704777,
-      "grad_norm": 0.0037458178121596575,
-      "learning_rate": 0.00019999309535184096,
-      "loss": 46.0,
-      "step": 23238
-    },
-    {
-      "epoch": 3.742461451749265,
-      "grad_norm": 0.0009623331134207547,
-      "learning_rate": 0.00019999309475732423,
-      "loss": 46.0,
-      "step": 23239
-    },
-    {
-      "epoch": 3.7426224888280526,
-      "grad_norm": 0.0021316311322152615,
-      "learning_rate": 0.0001999930941627819,
-      "loss": 46.0,
-      "step": 23240
-    },
-    {
-      "epoch": 3.74278352590684,
-      "grad_norm": 0.01692894659936428,
-      "learning_rate": 0.00019999309356821398,
-      "loss": 46.0,
-      "step": 23241
-    },
-    {
-      "epoch": 3.7429445629856275,
-      "grad_norm": 0.0046026455238461494,
-      "learning_rate": 0.00019999309297362048,
-      "loss": 46.0,
-      "step": 23242
-    },
-    {
-      "epoch": 3.743105600064415,
-      "grad_norm": 0.000602768617682159,
-      "learning_rate": 0.0001999930923790014,
-      "loss": 46.0,
-      "step": 23243
-    },
-    {
-      "epoch": 3.7432666371432024,
-      "grad_norm": 0.012639382854104042,
-      "learning_rate": 0.00019999309178435668,
-      "loss": 46.0,
-      "step": 23244
-    },
-    {
-      "epoch": 3.74342767422199,
-      "grad_norm": 0.0024291942827403545,
-      "learning_rate": 0.0001999930911896864,
-      "loss": 46.0,
-      "step": 23245
-    },
-    {
-      "epoch": 3.743588711300777,
-      "grad_norm": 0.0007512205629609525,
-      "learning_rate": 0.0001999930905949905,
-      "loss": 46.0,
-      "step": 23246
-    },
-    {
-      "epoch": 3.7437497483795643,
-      "grad_norm": 0.0024907661136239767,
-      "learning_rate": 0.00019999309000026902,
-      "loss": 46.0,
-      "step": 23247
-    },
-    {
-      "epoch": 3.7439107854583518,
-      "grad_norm": 0.0008014555787667632,
-      "learning_rate": 0.00019999308940552197,
-      "loss": 46.0,
-      "step": 23248
-    },
-    {
-      "epoch": 3.744071822537139,
-      "grad_norm": 0.0010756963165476918,
-      "learning_rate": 0.0001999930888107493,
-      "loss": 46.0,
-      "step": 23249
-    },
-    {
-      "epoch": 3.7442328596159267,
-      "grad_norm": 0.005392158403992653,
-      "learning_rate": 0.00019999308821595104,
-      "loss": 46.0,
-      "step": 23250
-    },
-    {
-      "epoch": 3.744393896694714,
-      "grad_norm": 0.004219592548906803,
-      "learning_rate": 0.0001999930876211272,
-      "loss": 46.0,
-      "step": 23251
-    },
-    {
-      "epoch": 3.744554933773501,
-      "grad_norm": 0.0041700867004692554,
-      "learning_rate": 0.00019999308702627774,
-      "loss": 46.0,
-      "step": 23252
-    },
-    {
-      "epoch": 3.7447159708522886,
-      "grad_norm": 0.0034497117158025503,
-      "learning_rate": 0.0001999930864314027,
-      "loss": 46.0,
-      "step": 23253
-    },
-    {
-      "epoch": 3.744877007931076,
-      "grad_norm": 0.0019165397388860583,
-      "learning_rate": 0.00019999308583650208,
-      "loss": 46.0,
-      "step": 23254
-    },
-    {
-      "epoch": 3.7450380450098635,
-      "grad_norm": 0.0017872173339128494,
-      "learning_rate": 0.00019999308524157585,
-      "loss": 46.0,
-      "step": 23255
-    },
-    {
-      "epoch": 3.745199082088651,
-      "grad_norm": 0.0020129105541855097,
-      "learning_rate": 0.00019999308464662405,
-      "loss": 46.0,
-      "step": 23256
-    },
-    {
-      "epoch": 3.7453601191674384,
-      "grad_norm": 0.0017803329974412918,
-      "learning_rate": 0.00019999308405164662,
-      "loss": 46.0,
-      "step": 23257
-    },
-    {
-      "epoch": 3.745521156246226,
-      "grad_norm": 0.0025569559074938297,
-      "learning_rate": 0.00019999308345664365,
-      "loss": 46.0,
-      "step": 23258
-    },
-    {
-      "epoch": 3.7456821933250133,
-      "grad_norm": 0.002819600747898221,
-      "learning_rate": 0.00019999308286161504,
-      "loss": 46.0,
-      "step": 23259
-    },
-    {
-      "epoch": 3.7458432304038007,
-      "grad_norm": 0.0029949620366096497,
-      "learning_rate": 0.00019999308226656087,
-      "loss": 46.0,
-      "step": 23260
-    },
-    {
-      "epoch": 3.7460042674825877,
-      "grad_norm": 0.0017388558480888605,
-      "learning_rate": 0.0001999930816714811,
-      "loss": 46.0,
-      "step": 23261
-    },
-    {
-      "epoch": 3.746165304561375,
-      "grad_norm": 0.0017265565693378448,
-      "learning_rate": 0.0001999930810763757,
-      "loss": 46.0,
-      "step": 23262
-    },
-    {
-      "epoch": 3.7463263416401626,
-      "grad_norm": 0.005802724976092577,
-      "learning_rate": 0.00019999308048124475,
-      "loss": 46.0,
-      "step": 23263
-    },
-    {
-      "epoch": 3.74648737871895,
-      "grad_norm": 0.0030983497854322195,
-      "learning_rate": 0.0001999930798860882,
-      "loss": 46.0,
-      "step": 23264
-    },
-    {
-      "epoch": 3.7466484157977376,
-      "grad_norm": 0.006051764823496342,
-      "learning_rate": 0.00019999307929090602,
-      "loss": 46.0,
-      "step": 23265
-    },
-    {
-      "epoch": 3.7468094528765246,
-      "grad_norm": 0.008463297039270401,
-      "learning_rate": 0.00019999307869569827,
-      "loss": 46.0,
-      "step": 23266
-    },
-    {
-      "epoch": 3.746970489955312,
-      "grad_norm": 0.0011268469970673323,
-      "learning_rate": 0.00019999307810046496,
-      "loss": 46.0,
-      "step": 23267
-    },
-    {
-      "epoch": 3.7471315270340995,
-      "grad_norm": 0.004161206539720297,
-      "learning_rate": 0.000199993077505206,
-      "loss": 46.0,
-      "step": 23268
-    },
-    {
-      "epoch": 3.747292564112887,
-      "grad_norm": 0.0012801686534658074,
-      "learning_rate": 0.0001999930769099215,
-      "loss": 46.0,
-      "step": 23269
-    },
-    {
-      "epoch": 3.7474536011916744,
-      "grad_norm": 0.009900976903736591,
-      "learning_rate": 0.0001999930763146114,
-      "loss": 46.0,
-      "step": 23270
-    },
-    {
-      "epoch": 3.747614638270462,
-      "grad_norm": 0.0022036691661924124,
-      "learning_rate": 0.00019999307571927567,
-      "loss": 46.0,
-      "step": 23271
-    },
-    {
-      "epoch": 3.7477756753492493,
-      "grad_norm": 0.0027693714946508408,
-      "learning_rate": 0.00019999307512391437,
-      "loss": 46.0,
-      "step": 23272
-    },
-    {
-      "epoch": 3.7479367124280367,
-      "grad_norm": 0.0030390287283807993,
-      "learning_rate": 0.00019999307452852748,
-      "loss": 46.0,
-      "step": 23273
-    },
-    {
-      "epoch": 3.748097749506824,
-      "grad_norm": 0.0028306397143751383,
-      "learning_rate": 0.00019999307393311498,
-      "loss": 46.0,
-      "step": 23274
-    },
-    {
-      "epoch": 3.748258786585611,
-      "grad_norm": 0.010642561130225658,
-      "learning_rate": 0.0001999930733376769,
-      "loss": 46.0,
-      "step": 23275
-    },
-    {
-      "epoch": 3.7484198236643986,
-      "grad_norm": 0.0030794998165220022,
-      "learning_rate": 0.00019999307274221322,
-      "loss": 46.0,
-      "step": 23276
-    },
-    {
-      "epoch": 3.748580860743186,
-      "grad_norm": 0.002850885270163417,
-      "learning_rate": 0.00019999307214672393,
-      "loss": 46.0,
-      "step": 23277
-    },
-    {
-      "epoch": 3.7487418978219735,
-      "grad_norm": 0.0037070740945637226,
-      "learning_rate": 0.0001999930715512091,
-      "loss": 46.0,
-      "step": 23278
-    },
-    {
-      "epoch": 3.748902934900761,
-      "grad_norm": 0.006732048466801643,
-      "learning_rate": 0.00019999307095566865,
-      "loss": 46.0,
-      "step": 23279
-    },
-    {
-      "epoch": 3.7490639719795484,
-      "grad_norm": 0.002443492878228426,
-      "learning_rate": 0.0001999930703601026,
-      "loss": 46.0,
-      "step": 23280
-    },
-    {
-      "epoch": 3.7492250090583354,
-      "grad_norm": 0.003433868521824479,
-      "learning_rate": 0.00019999306976451093,
-      "loss": 46.0,
-      "step": 23281
-    },
-    {
-      "epoch": 3.749386046137123,
-      "grad_norm": 0.0033519493881613016,
-      "learning_rate": 0.0001999930691688937,
-      "loss": 46.0,
-      "step": 23282
-    },
-    {
-      "epoch": 3.7495470832159103,
-      "grad_norm": 0.005566973704844713,
-      "learning_rate": 0.00019999306857325087,
-      "loss": 46.0,
-      "step": 23283
-    },
-    {
-      "epoch": 3.749708120294698,
-      "grad_norm": 0.002177475718781352,
-      "learning_rate": 0.00019999306797758245,
-      "loss": 46.0,
-      "step": 23284
-    },
-    {
-      "epoch": 3.7498691573734853,
-      "grad_norm": 0.00495817419141531,
-      "learning_rate": 0.00019999306738188843,
-      "loss": 46.0,
-      "step": 23285
-    },
-    {
-      "epoch": 3.7500301944522727,
-      "grad_norm": 0.003249040339142084,
-      "learning_rate": 0.00019999306678616883,
-      "loss": 46.0,
-      "step": 23286
-    },
-    {
-      "epoch": 3.75019123153106,
-      "grad_norm": 0.002706686733290553,
-      "learning_rate": 0.00019999306619042365,
-      "loss": 46.0,
-      "step": 23287
-    },
-    {
-      "epoch": 3.7503522686098476,
-      "grad_norm": 0.003501946572214365,
-      "learning_rate": 0.00019999306559465282,
-      "loss": 46.0,
-      "step": 23288
-    },
-    {
-      "epoch": 3.750513305688635,
-      "grad_norm": 0.001713958103209734,
-      "learning_rate": 0.00019999306499885643,
-      "loss": 46.0,
-      "step": 23289
-    },
-    {
-      "epoch": 3.750674342767422,
-      "grad_norm": 0.0043103257194161415,
-      "learning_rate": 0.00019999306440303448,
-      "loss": 46.0,
-      "step": 23290
-    },
-    {
-      "epoch": 3.7508353798462095,
-      "grad_norm": 0.0015746662393212318,
-      "learning_rate": 0.0001999930638071869,
-      "loss": 46.0,
-      "step": 23291
-    },
-    {
-      "epoch": 3.750996416924997,
-      "grad_norm": 0.006887136492878199,
-      "learning_rate": 0.00019999306321131374,
-      "loss": 46.0,
-      "step": 23292
-    },
-    {
-      "epoch": 3.7511574540037844,
-      "grad_norm": 0.010822443291544914,
-      "learning_rate": 0.00019999306261541498,
-      "loss": 46.0,
-      "step": 23293
-    },
-    {
-      "epoch": 3.751318491082572,
-      "grad_norm": 0.0028583186212927103,
-      "learning_rate": 0.00019999306201949063,
-      "loss": 46.0,
-      "step": 23294
-    },
-    {
-      "epoch": 3.751479528161359,
-      "grad_norm": 0.009037405252456665,
-      "learning_rate": 0.00019999306142354069,
-      "loss": 46.0,
-      "step": 23295
-    },
-    {
-      "epoch": 3.7516405652401463,
-      "grad_norm": 0.006317655090242624,
-      "learning_rate": 0.00019999306082756513,
-      "loss": 46.0,
-      "step": 23296
-    },
-    {
-      "epoch": 3.751801602318934,
-      "grad_norm": 0.0026851806323975325,
-      "learning_rate": 0.00019999306023156402,
-      "loss": 46.0,
-      "step": 23297
-    },
-    {
-      "epoch": 3.7519626393977212,
-      "grad_norm": 0.0034984173253178596,
-      "learning_rate": 0.00019999305963553727,
-      "loss": 46.0,
-      "step": 23298
-    },
-    {
-      "epoch": 3.7521236764765087,
-      "grad_norm": 0.003804260166361928,
-      "learning_rate": 0.00019999305903948495,
-      "loss": 46.0,
-      "step": 23299
-    },
-    {
-      "epoch": 3.752284713555296,
-      "grad_norm": 0.003835939336568117,
-      "learning_rate": 0.00019999305844340705,
-      "loss": 46.0,
-      "step": 23300
-    },
-    {
-      "epoch": 3.7524457506340836,
-      "grad_norm": 0.007036760449409485,
-      "learning_rate": 0.00019999305784730356,
-      "loss": 46.0,
-      "step": 23301
-    },
-    {
-      "epoch": 3.752606787712871,
-      "grad_norm": 0.011795183643698692,
-      "learning_rate": 0.00019999305725117446,
-      "loss": 46.0,
-      "step": 23302
-    },
-    {
-      "epoch": 3.7527678247916585,
-      "grad_norm": 0.031054265797138214,
-      "learning_rate": 0.00019999305665501977,
-      "loss": 46.0,
-      "step": 23303
-    },
-    {
-      "epoch": 3.752928861870446,
-      "grad_norm": 0.0031320760026574135,
-      "learning_rate": 0.0001999930560588395,
-      "loss": 46.0,
-      "step": 23304
-    },
-    {
-      "epoch": 3.753089898949233,
-      "grad_norm": 0.004960140213370323,
-      "learning_rate": 0.0001999930554626336,
-      "loss": 46.0,
-      "step": 23305
-    },
-    {
-      "epoch": 3.7532509360280204,
-      "grad_norm": 0.012674725614488125,
-      "learning_rate": 0.00019999305486640215,
-      "loss": 46.0,
-      "step": 23306
-    },
-    {
-      "epoch": 3.753411973106808,
-      "grad_norm": 0.003215216798707843,
-      "learning_rate": 0.00019999305427014508,
-      "loss": 46.0,
-      "step": 23307
-    },
-    {
-      "epoch": 3.7535730101855953,
-      "grad_norm": 0.004510294180363417,
-      "learning_rate": 0.00019999305367386243,
-      "loss": 46.0,
-      "step": 23308
-    },
-    {
-      "epoch": 3.7537340472643828,
-      "grad_norm": 0.000803169037681073,
-      "learning_rate": 0.0001999930530775542,
-      "loss": 46.0,
-      "step": 23309
-    },
-    {
-      "epoch": 3.7538950843431698,
-      "grad_norm": 0.0030744210816919804,
-      "learning_rate": 0.00019999305248122033,
-      "loss": 46.0,
-      "step": 23310
-    },
-    {
-      "epoch": 3.754056121421957,
-      "grad_norm": 0.001205191481858492,
-      "learning_rate": 0.00019999305188486092,
-      "loss": 46.0,
-      "step": 23311
-    },
-    {
-      "epoch": 3.7542171585007447,
-      "grad_norm": 0.00221211276948452,
-      "learning_rate": 0.00019999305128847586,
-      "loss": 46.0,
-      "step": 23312
-    },
-    {
-      "epoch": 3.754378195579532,
-      "grad_norm": 0.00478509534150362,
-      "learning_rate": 0.00019999305069206527,
-      "loss": 46.0,
-      "step": 23313
-    },
-    {
-      "epoch": 3.7545392326583196,
-      "grad_norm": 0.001582092372700572,
-      "learning_rate": 0.00019999305009562904,
-      "loss": 46.0,
-      "step": 23314
-    },
-    {
-      "epoch": 3.754700269737107,
-      "grad_norm": 0.003600649069994688,
-      "learning_rate": 0.00019999304949916722,
-      "loss": 46.0,
-      "step": 23315
-    },
-    {
-      "epoch": 3.7548613068158945,
-      "grad_norm": 0.0018423496512696147,
-      "learning_rate": 0.00019999304890267982,
-      "loss": 46.0,
-      "step": 23316
-    },
-    {
-      "epoch": 3.755022343894682,
-      "grad_norm": 0.0012430831557139754,
-      "learning_rate": 0.00019999304830616685,
-      "loss": 46.0,
-      "step": 23317
-    },
-    {
-      "epoch": 3.7551833809734694,
-      "grad_norm": 0.0021214019507169724,
-      "learning_rate": 0.00019999304770962825,
-      "loss": 46.0,
-      "step": 23318
-    },
-    {
-      "epoch": 3.7553444180522564,
-      "grad_norm": 0.0031343905720859766,
-      "learning_rate": 0.00019999304711306408,
-      "loss": 46.0,
-      "step": 23319
-    },
-    {
-      "epoch": 3.755505455131044,
-      "grad_norm": 0.006793077103793621,
-      "learning_rate": 0.0001999930465164743,
-      "loss": 46.0,
-      "step": 23320
-    },
-    {
-      "epoch": 3.7556664922098313,
-      "grad_norm": 0.008207363076508045,
-      "learning_rate": 0.00019999304591985893,
-      "loss": 46.0,
-      "step": 23321
-    },
-    {
-      "epoch": 3.7558275292886187,
-      "grad_norm": 0.006455515045672655,
-      "learning_rate": 0.00019999304532321798,
-      "loss": 46.0,
-      "step": 23322
-    },
-    {
-      "epoch": 3.755988566367406,
-      "grad_norm": 0.005089758895337582,
-      "learning_rate": 0.00019999304472655144,
-      "loss": 46.0,
-      "step": 23323
-    },
-    {
-      "epoch": 3.7561496034461936,
-      "grad_norm": 0.01651889644563198,
-      "learning_rate": 0.00019999304412985928,
-      "loss": 46.0,
-      "step": 23324
-    },
-    {
-      "epoch": 3.7563106405249806,
-      "grad_norm": 0.014437532983720303,
-      "learning_rate": 0.00019999304353314154,
-      "loss": 46.0,
-      "step": 23325
-    },
-    {
-      "epoch": 3.756471677603768,
-      "grad_norm": 0.0029440789949148893,
-      "learning_rate": 0.00019999304293639823,
-      "loss": 46.0,
-      "step": 23326
-    },
-    {
-      "epoch": 3.7566327146825556,
-      "grad_norm": 0.0014560960698872805,
-      "learning_rate": 0.0001999930423396293,
-      "loss": 46.0,
-      "step": 23327
-    },
-    {
-      "epoch": 3.756793751761343,
-      "grad_norm": 0.006653280928730965,
-      "learning_rate": 0.00019999304174283478,
-      "loss": 46.0,
-      "step": 23328
-    },
-    {
-      "epoch": 3.7569547888401305,
-      "grad_norm": 0.01361078955233097,
-      "learning_rate": 0.00019999304114601464,
-      "loss": 46.0,
-      "step": 23329
-    },
-    {
-      "epoch": 3.757115825918918,
-      "grad_norm": 0.004327659495174885,
-      "learning_rate": 0.00019999304054916896,
-      "loss": 46.0,
-      "step": 23330
-    },
-    {
-      "epoch": 3.7572768629977054,
-      "grad_norm": 0.0010913473088294268,
-      "learning_rate": 0.00019999303995229766,
-      "loss": 46.0,
-      "step": 23331
-    },
-    {
-      "epoch": 3.757437900076493,
-      "grad_norm": 0.003111462341621518,
-      "learning_rate": 0.00019999303935540078,
-      "loss": 46.0,
-      "step": 23332
-    },
-    {
-      "epoch": 3.7575989371552803,
-      "grad_norm": 0.005549356807023287,
-      "learning_rate": 0.00019999303875847831,
-      "loss": 46.0,
-      "step": 23333
-    },
-    {
-      "epoch": 3.7577599742340673,
-      "grad_norm": 0.0056270030327141285,
-      "learning_rate": 0.0001999930381615302,
-      "loss": 46.0,
-      "step": 23334
-    },
-    {
-      "epoch": 3.7579210113128547,
-      "grad_norm": 0.021341169252991676,
-      "learning_rate": 0.00019999303756455654,
-      "loss": 46.0,
-      "step": 23335
-    },
-    {
-      "epoch": 3.758082048391642,
-      "grad_norm": 0.005347100086510181,
-      "learning_rate": 0.00019999303696755728,
-      "loss": 46.0,
-      "step": 23336
-    },
-    {
-      "epoch": 3.7582430854704296,
-      "grad_norm": 0.0026312253903597593,
-      "learning_rate": 0.0001999930363705324,
-      "loss": 46.0,
-      "step": 23337
-    },
-    {
-      "epoch": 3.758404122549217,
-      "grad_norm": 0.012708326801657677,
-      "learning_rate": 0.00019999303577348198,
-      "loss": 46.0,
-      "step": 23338
-    },
-    {
-      "epoch": 3.758565159628004,
-      "grad_norm": 0.0008886596187949181,
-      "learning_rate": 0.00019999303517640593,
-      "loss": 46.0,
-      "step": 23339
-    },
-    {
-      "epoch": 3.7587261967067915,
-      "grad_norm": 0.013958428986370564,
-      "learning_rate": 0.00019999303457930433,
-      "loss": 46.0,
-      "step": 23340
-    },
-    {
-      "epoch": 3.758887233785579,
-      "grad_norm": 0.0020769955590367317,
-      "learning_rate": 0.00019999303398217708,
-      "loss": 46.0,
-      "step": 23341
-    },
-    {
-      "epoch": 3.7590482708643664,
-      "grad_norm": 0.012553153559565544,
-      "learning_rate": 0.00019999303338502427,
-      "loss": 46.0,
-      "step": 23342
-    },
-    {
-      "epoch": 3.759209307943154,
-      "grad_norm": 0.003156843362376094,
-      "learning_rate": 0.00019999303278784585,
-      "loss": 46.0,
-      "step": 23343
-    },
-    {
-      "epoch": 3.7593703450219413,
-      "grad_norm": 0.0006373069481924176,
-      "learning_rate": 0.00019999303219064184,
-      "loss": 46.0,
-      "step": 23344
-    },
-    {
-      "epoch": 3.759531382100729,
-      "grad_norm": 0.002016419544816017,
-      "learning_rate": 0.00019999303159341225,
-      "loss": 46.0,
-      "step": 23345
-    },
-    {
-      "epoch": 3.7596924191795162,
-      "grad_norm": 0.001978748245164752,
-      "learning_rate": 0.00019999303099615704,
-      "loss": 46.0,
-      "step": 23346
-    },
-    {
-      "epoch": 3.7598534562583037,
-      "grad_norm": 0.0043363929726183414,
-      "learning_rate": 0.00019999303039887627,
-      "loss": 46.0,
-      "step": 23347
-    },
-    {
-      "epoch": 3.7600144933370907,
-      "grad_norm": 0.0014998463448137045,
-      "learning_rate": 0.00019999302980156988,
-      "loss": 46.0,
-      "step": 23348
-    },
-    {
-      "epoch": 3.760175530415878,
-      "grad_norm": 0.004350574221462011,
-      "learning_rate": 0.0001999930292042379,
-      "loss": 46.0,
-      "step": 23349
-    },
-    {
-      "epoch": 3.7603365674946656,
-      "grad_norm": 0.0032098707742989063,
-      "learning_rate": 0.00019999302860688033,
-      "loss": 46.0,
-      "step": 23350
-    },
-    {
-      "epoch": 3.760497604573453,
-      "grad_norm": 0.005226675420999527,
-      "learning_rate": 0.00019999302800949718,
-      "loss": 46.0,
-      "step": 23351
-    },
-    {
-      "epoch": 3.7606586416522405,
-      "grad_norm": 0.0010043485090136528,
-      "learning_rate": 0.00019999302741208842,
-      "loss": 46.0,
-      "step": 23352
-    },
-    {
-      "epoch": 3.760819678731028,
-      "grad_norm": 0.006235708016902208,
-      "learning_rate": 0.0001999930268146541,
-      "loss": 46.0,
-      "step": 23353
-    },
-    {
-      "epoch": 3.760980715809815,
-      "grad_norm": 0.005920794326812029,
-      "learning_rate": 0.00019999302621719417,
-      "loss": 46.0,
-      "step": 23354
-    },
-    {
-      "epoch": 3.7611417528886024,
-      "grad_norm": 0.0019254216458648443,
-      "learning_rate": 0.00019999302561970862,
-      "loss": 46.0,
-      "step": 23355
-    },
-    {
-      "epoch": 3.76130278996739,
-      "grad_norm": 0.0037464576307684183,
-      "learning_rate": 0.0001999930250221975,
-      "loss": 46.0,
-      "step": 23356
-    },
-    {
-      "epoch": 3.7614638270461773,
-      "grad_norm": 0.004395162221044302,
-      "learning_rate": 0.0001999930244246608,
-      "loss": 46.0,
-      "step": 23357
-    },
-    {
-      "epoch": 3.7616248641249648,
-      "grad_norm": 0.00424527982249856,
-      "learning_rate": 0.00019999302382709848,
-      "loss": 46.0,
-      "step": 23358
-    },
-    {
-      "epoch": 3.7617859012037522,
-      "grad_norm": 0.0008832887979224324,
-      "learning_rate": 0.00019999302322951058,
-      "loss": 46.0,
-      "step": 23359
-    },
-    {
-      "epoch": 3.7619469382825397,
-      "grad_norm": 0.0057721748016774654,
-      "learning_rate": 0.00019999302263189707,
-      "loss": 46.0,
-      "step": 23360
-    },
-    {
-      "epoch": 3.762107975361327,
-      "grad_norm": 0.004645981825888157,
-      "learning_rate": 0.00019999302203425797,
-      "loss": 46.0,
-      "step": 23361
-    },
-    {
-      "epoch": 3.7622690124401146,
-      "grad_norm": 0.007081453688442707,
-      "learning_rate": 0.0001999930214365933,
-      "loss": 46.0,
-      "step": 23362
-    },
-    {
-      "epoch": 3.7624300495189016,
-      "grad_norm": 0.0038249497301876545,
-      "learning_rate": 0.00019999302083890304,
-      "loss": 46.0,
-      "step": 23363
-    },
-    {
-      "epoch": 3.762591086597689,
-      "grad_norm": 0.0052764806896448135,
-      "learning_rate": 0.00019999302024118718,
-      "loss": 46.0,
-      "step": 23364
-    },
-    {
-      "epoch": 3.7627521236764765,
-      "grad_norm": 0.0020675412379205227,
-      "learning_rate": 0.0001999930196434457,
-      "loss": 46.0,
-      "step": 23365
-    },
-    {
-      "epoch": 3.762913160755264,
-      "grad_norm": 0.0055201854556798935,
-      "learning_rate": 0.00019999301904567864,
-      "loss": 46.0,
-      "step": 23366
-    },
-    {
-      "epoch": 3.7630741978340514,
-      "grad_norm": 0.0030638568568974733,
-      "learning_rate": 0.000199993018447886,
-      "loss": 46.0,
-      "step": 23367
-    },
-    {
-      "epoch": 3.763235234912839,
-      "grad_norm": 0.003037769813090563,
-      "learning_rate": 0.00019999301785006776,
-      "loss": 46.0,
-      "step": 23368
-    },
-    {
-      "epoch": 3.763396271991626,
-      "grad_norm": 0.0026919208467006683,
-      "learning_rate": 0.00019999301725222394,
-      "loss": 46.0,
-      "step": 23369
-    },
-    {
-      "epoch": 3.7635573090704133,
-      "grad_norm": 0.0021743851248174906,
-      "learning_rate": 0.0001999930166543545,
-      "loss": 46.0,
-      "step": 23370
-    },
-    {
-      "epoch": 3.7637183461492008,
-      "grad_norm": 0.005291048903018236,
-      "learning_rate": 0.00019999301605645947,
-      "loss": 46.0,
-      "step": 23371
-    },
-    {
-      "epoch": 3.763879383227988,
-      "grad_norm": 0.003063508542254567,
-      "learning_rate": 0.00019999301545853886,
-      "loss": 46.0,
-      "step": 23372
-    },
-    {
-      "epoch": 3.7640404203067757,
-      "grad_norm": 0.004322974011301994,
-      "learning_rate": 0.00019999301486059266,
-      "loss": 46.0,
-      "step": 23373
-    },
-    {
-      "epoch": 3.764201457385563,
-      "grad_norm": 0.004668680485337973,
-      "learning_rate": 0.00019999301426262088,
-      "loss": 46.0,
-      "step": 23374
-    },
-    {
-      "epoch": 3.7643624944643506,
-      "grad_norm": 0.002153209876269102,
-      "learning_rate": 0.00019999301366462348,
-      "loss": 46.0,
-      "step": 23375
-    },
-    {
-      "epoch": 3.764523531543138,
-      "grad_norm": 0.0044435737654566765,
-      "learning_rate": 0.0001999930130666005,
-      "loss": 46.0,
-      "step": 23376
-    },
-    {
-      "epoch": 3.7646845686219255,
-      "grad_norm": 0.004240675829350948,
-      "learning_rate": 0.00019999301246855191,
-      "loss": 46.0,
-      "step": 23377
-    },
-    {
-      "epoch": 3.7648456057007125,
-      "grad_norm": 0.010793029330670834,
-      "learning_rate": 0.00019999301187047775,
-      "loss": 46.0,
-      "step": 23378
-    },
-    {
-      "epoch": 3.7650066427795,
-      "grad_norm": 0.001335603534244001,
-      "learning_rate": 0.00019999301127237798,
-      "loss": 46.0,
-      "step": 23379
-    },
-    {
-      "epoch": 3.7651676798582874,
-      "grad_norm": 0.0051144990138709545,
-      "learning_rate": 0.00019999301067425261,
-      "loss": 46.0,
-      "step": 23380
-    },
-    {
-      "epoch": 3.765328716937075,
-      "grad_norm": 0.011687594465911388,
-      "learning_rate": 0.00019999301007610166,
-      "loss": 46.0,
-      "step": 23381
-    },
-    {
-      "epoch": 3.7654897540158623,
-      "grad_norm": 0.014922913163900375,
-      "learning_rate": 0.00019999300947792513,
-      "loss": 46.0,
-      "step": 23382
-    },
-    {
-      "epoch": 3.7656507910946493,
-      "grad_norm": 0.003803300904110074,
-      "learning_rate": 0.00019999300887972297,
-      "loss": 46.0,
-      "step": 23383
-    },
-    {
-      "epoch": 3.7658118281734367,
-      "grad_norm": 0.001957479165866971,
-      "learning_rate": 0.00019999300828149526,
-      "loss": 46.0,
-      "step": 23384
-    },
-    {
-      "epoch": 3.765972865252224,
-      "grad_norm": 0.00957445427775383,
-      "learning_rate": 0.0001999930076832419,
-      "loss": 46.0,
-      "step": 23385
-    },
-    {
-      "epoch": 3.7661339023310116,
-      "grad_norm": 0.0011465654242783785,
-      "learning_rate": 0.00019999300708496302,
-      "loss": 46.0,
-      "step": 23386
-    },
-    {
-      "epoch": 3.766294939409799,
-      "grad_norm": 0.007577102165669203,
-      "learning_rate": 0.0001999930064866585,
-      "loss": 46.0,
-      "step": 23387
-    },
-    {
-      "epoch": 3.7664559764885865,
-      "grad_norm": 0.0029909831937402487,
-      "learning_rate": 0.0001999930058883284,
-      "loss": 46.0,
-      "step": 23388
-    },
-    {
-      "epoch": 3.766617013567374,
-      "grad_norm": 0.005610207095742226,
-      "learning_rate": 0.0001999930052899727,
-      "loss": 46.0,
-      "step": 23389
-    },
-    {
-      "epoch": 3.7667780506461614,
-      "grad_norm": 0.010227243416011333,
-      "learning_rate": 0.00019999300469159144,
-      "loss": 46.0,
-      "step": 23390
-    },
-    {
-      "epoch": 3.766939087724949,
-      "grad_norm": 0.006348138675093651,
-      "learning_rate": 0.00019999300409318457,
-      "loss": 46.0,
-      "step": 23391
-    },
-    {
-      "epoch": 3.767100124803736,
-      "grad_norm": 0.002757695270702243,
-      "learning_rate": 0.00019999300349475205,
-      "loss": 46.0,
-      "step": 23392
-    },
-    {
-      "epoch": 3.7672611618825234,
-      "grad_norm": 0.0056867399252951145,
-      "learning_rate": 0.00019999300289629403,
-      "loss": 46.0,
-      "step": 23393
-    },
-    {
-      "epoch": 3.767422198961311,
-      "grad_norm": 0.0012817836832255125,
-      "learning_rate": 0.00019999300229781036,
-      "loss": 46.0,
-      "step": 23394
-    },
-    {
-      "epoch": 3.7675832360400983,
-      "grad_norm": 0.01238818746060133,
-      "learning_rate": 0.00019999300169930108,
-      "loss": 46.0,
-      "step": 23395
-    },
-    {
-      "epoch": 3.7677442731188857,
-      "grad_norm": 0.003599542658776045,
-      "learning_rate": 0.00019999300110076624,
-      "loss": 46.0,
-      "step": 23396
-    },
-    {
-      "epoch": 3.767905310197673,
-      "grad_norm": 0.0031063018832355738,
-      "learning_rate": 0.0001999930005022058,
-      "loss": 46.0,
-      "step": 23397
-    },
-    {
-      "epoch": 3.76806634727646,
-      "grad_norm": 0.0021299864165484905,
-      "learning_rate": 0.00019999299990361977,
-      "loss": 46.0,
-      "step": 23398
-    },
-    {
-      "epoch": 3.7682273843552476,
-      "grad_norm": 0.025733893737196922,
-      "learning_rate": 0.00019999299930500815,
-      "loss": 46.0,
-      "step": 23399
-    },
-    {
-      "epoch": 3.768388421434035,
-      "grad_norm": 0.005361327435821295,
-      "learning_rate": 0.00019999299870637093,
-      "loss": 46.0,
-      "step": 23400
-    },
-    {
-      "epoch": 3.7685494585128225,
-      "grad_norm": 0.01232514251023531,
-      "learning_rate": 0.0001999929981077081,
-      "loss": 46.0,
-      "step": 23401
-    },
-    {
-      "epoch": 3.76871049559161,
-      "grad_norm": 0.003512471215799451,
-      "learning_rate": 0.00019999299750901969,
-      "loss": 46.0,
-      "step": 23402
-    },
-    {
-      "epoch": 3.7688715326703974,
-      "grad_norm": 0.0026258009020239115,
-      "learning_rate": 0.0001999929969103057,
-      "loss": 46.0,
-      "step": 23403
-    },
-    {
-      "epoch": 3.769032569749185,
-      "grad_norm": 0.00991733092814684,
-      "learning_rate": 0.00019999299631156612,
-      "loss": 46.0,
-      "step": 23404
-    },
-    {
-      "epoch": 3.7691936068279723,
-      "grad_norm": 0.0013436574954539537,
-      "learning_rate": 0.0001999929957128009,
-      "loss": 46.0,
-      "step": 23405
-    },
-    {
-      "epoch": 3.76935464390676,
-      "grad_norm": 0.004663341678678989,
-      "learning_rate": 0.00019999299511401015,
-      "loss": 46.0,
-      "step": 23406
-    },
-    {
-      "epoch": 3.769515680985547,
-      "grad_norm": 0.004056538455188274,
-      "learning_rate": 0.0001999929945151938,
-      "loss": 46.0,
-      "step": 23407
-    },
-    {
-      "epoch": 3.7696767180643342,
-      "grad_norm": 0.0011676936410367489,
-      "learning_rate": 0.00019999299391635183,
-      "loss": 46.0,
-      "step": 23408
-    },
-    {
-      "epoch": 3.7698377551431217,
-      "grad_norm": 0.0011463186237961054,
-      "learning_rate": 0.00019999299331748427,
-      "loss": 46.0,
-      "step": 23409
-    },
-    {
-      "epoch": 3.769998792221909,
-      "grad_norm": 0.005641614552587271,
-      "learning_rate": 0.0001999929927185911,
-      "loss": 46.0,
-      "step": 23410
-    },
-    {
-      "epoch": 3.7701598293006966,
-      "grad_norm": 0.005202275235205889,
-      "learning_rate": 0.00019999299211967238,
-      "loss": 46.0,
-      "step": 23411
-    },
-    {
-      "epoch": 3.7703208663794836,
-      "grad_norm": 0.0037886020727455616,
-      "learning_rate": 0.00019999299152072804,
-      "loss": 46.0,
-      "step": 23412
-    },
-    {
-      "epoch": 3.770481903458271,
-      "grad_norm": 0.0007066461257636547,
-      "learning_rate": 0.0001999929909217581,
-      "loss": 46.0,
-      "step": 23413
-    },
-    {
-      "epoch": 3.7706429405370585,
-      "grad_norm": 0.005618244409561157,
-      "learning_rate": 0.0001999929903227626,
-      "loss": 46.0,
-      "step": 23414
-    },
-    {
-      "epoch": 3.770803977615846,
-      "grad_norm": 0.00250281416811049,
-      "learning_rate": 0.00019999298972374146,
-      "loss": 46.0,
-      "step": 23415
-    },
-    {
-      "epoch": 3.7709650146946334,
-      "grad_norm": 0.0019470519619062543,
-      "learning_rate": 0.00019999298912469477,
-      "loss": 46.0,
-      "step": 23416
-    },
-    {
-      "epoch": 3.771126051773421,
-      "grad_norm": 0.0028976253233850002,
-      "learning_rate": 0.00019999298852562246,
-      "loss": 46.0,
-      "step": 23417
-    },
-    {
-      "epoch": 3.7712870888522083,
-      "grad_norm": 0.007943235337734222,
-      "learning_rate": 0.00019999298792652457,
-      "loss": 46.0,
-      "step": 23418
-    },
-    {
-      "epoch": 3.7714481259309958,
-      "grad_norm": 0.0029335098806768656,
-      "learning_rate": 0.0001999929873274011,
-      "loss": 46.0,
-      "step": 23419
-    },
-    {
-      "epoch": 3.771609163009783,
-      "grad_norm": 0.009886838495731354,
-      "learning_rate": 0.000199992986728252,
-      "loss": 46.0,
-      "step": 23420
-    },
-    {
-      "epoch": 3.7717702000885707,
-      "grad_norm": 0.0015666900435462594,
-      "learning_rate": 0.00019999298612907734,
-      "loss": 46.0,
-      "step": 23421
-    },
-    {
-      "epoch": 3.7719312371673577,
-      "grad_norm": 0.007241830229759216,
-      "learning_rate": 0.00019999298552987705,
-      "loss": 46.0,
-      "step": 23422
-    },
-    {
-      "epoch": 3.772092274246145,
-      "grad_norm": 0.007948859594762325,
-      "learning_rate": 0.0001999929849306512,
-      "loss": 46.0,
-      "step": 23423
-    },
-    {
-      "epoch": 3.7722533113249326,
-      "grad_norm": 0.0016738002886995673,
-      "learning_rate": 0.00019999298433139975,
-      "loss": 46.0,
-      "step": 23424
-    },
-    {
-      "epoch": 3.77241434840372,
-      "grad_norm": 0.001979612046852708,
-      "learning_rate": 0.0001999929837321227,
-      "loss": 46.0,
-      "step": 23425
-    },
-    {
-      "epoch": 3.7725753854825075,
-      "grad_norm": 0.0015118021983653307,
-      "learning_rate": 0.00019999298313282005,
-      "loss": 46.0,
-      "step": 23426
-    },
-    {
-      "epoch": 3.7727364225612945,
-      "grad_norm": 0.0010859890608116984,
-      "learning_rate": 0.00019999298253349181,
-      "loss": 46.0,
-      "step": 23427
-    },
-    {
-      "epoch": 3.772897459640082,
-      "grad_norm": 0.002263714326545596,
-      "learning_rate": 0.000199992981934138,
-      "loss": 46.0,
-      "step": 23428
-    },
-    {
-      "epoch": 3.7730584967188694,
-      "grad_norm": 0.0005644666380248964,
-      "learning_rate": 0.0001999929813347586,
-      "loss": 46.0,
-      "step": 23429
-    },
-    {
-      "epoch": 3.773219533797657,
-      "grad_norm": 0.0010720707941800356,
-      "learning_rate": 0.00019999298073535357,
-      "loss": 46.0,
-      "step": 23430
-    },
-    {
-      "epoch": 3.7733805708764443,
-      "grad_norm": 0.002276270417496562,
-      "learning_rate": 0.00019999298013592296,
-      "loss": 46.0,
-      "step": 23431
-    },
-    {
-      "epoch": 3.7735416079552317,
-      "grad_norm": 0.0007864724029786885,
-      "learning_rate": 0.0001999929795364668,
-      "loss": 46.0,
-      "step": 23432
-    },
-    {
-      "epoch": 3.773702645034019,
-      "grad_norm": 0.0009934685658663511,
-      "learning_rate": 0.000199992978936985,
-      "loss": 46.0,
-      "step": 23433
-    },
-    {
-      "epoch": 3.7738636821128067,
-      "grad_norm": 0.005271119065582752,
-      "learning_rate": 0.00019999297833747762,
-      "loss": 46.0,
-      "step": 23434
-    },
-    {
-      "epoch": 3.774024719191594,
-      "grad_norm": 0.01132645457983017,
-      "learning_rate": 0.0001999929777379446,
-      "loss": 46.0,
-      "step": 23435
-    },
-    {
-      "epoch": 3.774185756270381,
-      "grad_norm": 0.0023303860798478127,
-      "learning_rate": 0.00019999297713838604,
-      "loss": 46.0,
-      "step": 23436
-    },
-    {
-      "epoch": 3.7743467933491686,
-      "grad_norm": 0.00591937405988574,
-      "learning_rate": 0.00019999297653880188,
-      "loss": 46.0,
-      "step": 23437
-    },
-    {
-      "epoch": 3.774507830427956,
-      "grad_norm": 0.005823095329105854,
-      "learning_rate": 0.00019999297593919214,
-      "loss": 46.0,
-      "step": 23438
-    },
-    {
-      "epoch": 3.7746688675067435,
-      "grad_norm": 0.0056540220975875854,
-      "learning_rate": 0.00019999297533955678,
-      "loss": 46.0,
-      "step": 23439
-    },
-    {
-      "epoch": 3.774829904585531,
-      "grad_norm": 0.00987009983509779,
-      "learning_rate": 0.00019999297473989583,
-      "loss": 46.0,
-      "step": 23440
-    },
-    {
-      "epoch": 3.7749909416643184,
-      "grad_norm": 0.0023566654417663813,
-      "learning_rate": 0.0001999929741402093,
-      "loss": 46.0,
-      "step": 23441
-    },
-    {
-      "epoch": 3.7751519787431054,
-      "grad_norm": 0.0013051513815298676,
-      "learning_rate": 0.00019999297354049718,
-      "loss": 46.0,
-      "step": 23442
-    },
-    {
-      "epoch": 3.775313015821893,
-      "grad_norm": 0.0034219883382320404,
-      "learning_rate": 0.00019999297294075947,
-      "loss": 46.0,
-      "step": 23443
-    },
-    {
-      "epoch": 3.7754740529006803,
-      "grad_norm": 0.0007844906067475677,
-      "learning_rate": 0.00019999297234099615,
-      "loss": 46.0,
-      "step": 23444
-    },
-    {
-      "epoch": 3.7756350899794677,
-      "grad_norm": 0.004213909152895212,
-      "learning_rate": 0.00019999297174120724,
-      "loss": 46.0,
-      "step": 23445
-    },
-    {
-      "epoch": 3.775796127058255,
-      "grad_norm": 0.002701682737097144,
-      "learning_rate": 0.00019999297114139272,
-      "loss": 46.0,
-      "step": 23446
-    },
-    {
-      "epoch": 3.7759571641370426,
-      "grad_norm": 0.002492605010047555,
-      "learning_rate": 0.00019999297054155263,
-      "loss": 46.0,
-      "step": 23447
-    },
-    {
-      "epoch": 3.77611820121583,
-      "grad_norm": 0.0019003769848495722,
-      "learning_rate": 0.00019999296994168696,
-      "loss": 46.0,
-      "step": 23448
-    },
-    {
-      "epoch": 3.7762792382946175,
-      "grad_norm": 0.003208177164196968,
-      "learning_rate": 0.00019999296934179568,
-      "loss": 46.0,
-      "step": 23449
-    },
-    {
-      "epoch": 3.776440275373405,
-      "grad_norm": 0.0016427161172032356,
-      "learning_rate": 0.0001999929687418788,
-      "loss": 46.0,
-      "step": 23450
-    },
-    {
-      "epoch": 3.776601312452192,
-      "grad_norm": 0.005170580465346575,
-      "learning_rate": 0.00019999296814193635,
-      "loss": 46.0,
-      "step": 23451
-    },
-    {
-      "epoch": 3.7767623495309794,
-      "grad_norm": 0.004875520244240761,
-      "learning_rate": 0.00019999296754196828,
-      "loss": 46.0,
-      "step": 23452
-    },
-    {
-      "epoch": 3.776923386609767,
-      "grad_norm": 0.0016431326512247324,
-      "learning_rate": 0.00019999296694197462,
-      "loss": 46.0,
-      "step": 23453
-    },
-    {
-      "epoch": 3.7770844236885543,
-      "grad_norm": 0.007538162637501955,
-      "learning_rate": 0.0001999929663419554,
-      "loss": 46.0,
-      "step": 23454
-    },
-    {
-      "epoch": 3.777245460767342,
-      "grad_norm": 0.0017951886402443051,
-      "learning_rate": 0.00019999296574191053,
-      "loss": 46.0,
-      "step": 23455
-    },
-    {
-      "epoch": 3.777406497846129,
-      "grad_norm": 0.0034587865229696035,
-      "learning_rate": 0.0001999929651418401,
-      "loss": 46.0,
-      "step": 23456
-    },
-    {
-      "epoch": 3.7775675349249163,
-      "grad_norm": 0.0014397423947229981,
-      "learning_rate": 0.00019999296454174407,
-      "loss": 46.0,
-      "step": 23457
-    },
-    {
-      "epoch": 3.7777285720037037,
-      "grad_norm": 0.007553314324468374,
-      "learning_rate": 0.00019999296394162248,
-      "loss": 46.0,
-      "step": 23458
-    },
-    {
-      "epoch": 3.777889609082491,
-      "grad_norm": 0.0010361107997596264,
-      "learning_rate": 0.00019999296334147527,
-      "loss": 46.0,
-      "step": 23459
-    },
-    {
-      "epoch": 3.7780506461612786,
-      "grad_norm": 0.008487565442919731,
-      "learning_rate": 0.00019999296274130247,
-      "loss": 46.0,
-      "step": 23460
-    },
-    {
-      "epoch": 3.778211683240066,
-      "grad_norm": 0.0008957805694080889,
-      "learning_rate": 0.00019999296214110409,
-      "loss": 46.0,
-      "step": 23461
-    },
-    {
-      "epoch": 3.7783727203188535,
-      "grad_norm": 0.002249722369015217,
-      "learning_rate": 0.00019999296154088009,
-      "loss": 46.0,
-      "step": 23462
-    },
-    {
-      "epoch": 3.778533757397641,
-      "grad_norm": 0.0034861310850828886,
-      "learning_rate": 0.0001999929609406305,
-      "loss": 46.0,
-      "step": 23463
-    },
-    {
-      "epoch": 3.7786947944764284,
-      "grad_norm": 0.008774700574576855,
-      "learning_rate": 0.00019999296034035533,
-      "loss": 46.0,
-      "step": 23464
-    },
-    {
-      "epoch": 3.7788558315552154,
-      "grad_norm": 0.001866961014457047,
-      "learning_rate": 0.00019999295974005454,
-      "loss": 46.0,
-      "step": 23465
-    },
-    {
-      "epoch": 3.779016868634003,
-      "grad_norm": 0.004746279213577509,
-      "learning_rate": 0.00019999295913972816,
-      "loss": 46.0,
-      "step": 23466
-    },
-    {
-      "epoch": 3.7791779057127903,
-      "grad_norm": 0.0006533092819154263,
-      "learning_rate": 0.00019999295853937623,
-      "loss": 46.0,
-      "step": 23467
-    },
-    {
-      "epoch": 3.779338942791578,
-      "grad_norm": 0.001827763393521309,
-      "learning_rate": 0.00019999295793899868,
-      "loss": 46.0,
-      "step": 23468
-    },
-    {
-      "epoch": 3.7794999798703652,
-      "grad_norm": 0.0011072917841374874,
-      "learning_rate": 0.00019999295733859554,
-      "loss": 46.0,
-      "step": 23469
-    },
-    {
-      "epoch": 3.7796610169491527,
-      "grad_norm": 0.020383261144161224,
-      "learning_rate": 0.0001999929567381668,
-      "loss": 46.0,
-      "step": 23470
-    },
-    {
-      "epoch": 3.7798220540279397,
-      "grad_norm": 0.005471076816320419,
-      "learning_rate": 0.00019999295613771248,
-      "loss": 46.0,
-      "step": 23471
-    },
-    {
-      "epoch": 3.779983091106727,
-      "grad_norm": 0.005412416532635689,
-      "learning_rate": 0.00019999295553723256,
-      "loss": 46.0,
-      "step": 23472
-    },
-    {
-      "epoch": 3.7801441281855146,
-      "grad_norm": 0.009214709512889385,
-      "learning_rate": 0.00019999295493672704,
-      "loss": 46.0,
-      "step": 23473
-    },
-    {
-      "epoch": 3.780305165264302,
-      "grad_norm": 0.009223414584994316,
-      "learning_rate": 0.00019999295433619594,
-      "loss": 46.0,
-      "step": 23474
-    },
-    {
-      "epoch": 3.7804662023430895,
-      "grad_norm": 0.00212464714422822,
-      "learning_rate": 0.00019999295373563923,
-      "loss": 46.0,
-      "step": 23475
-    },
-    {
-      "epoch": 3.780627239421877,
-      "grad_norm": 0.00892906729131937,
-      "learning_rate": 0.00019999295313505696,
-      "loss": 46.0,
-      "step": 23476
-    },
-    {
-      "epoch": 3.7807882765006644,
-      "grad_norm": 0.008025085553526878,
-      "learning_rate": 0.00019999295253444904,
-      "loss": 46.0,
-      "step": 23477
-    },
-    {
-      "epoch": 3.780949313579452,
-      "grad_norm": 0.0014605334727093577,
-      "learning_rate": 0.0001999929519338156,
-      "loss": 46.0,
-      "step": 23478
-    },
-    {
-      "epoch": 3.7811103506582393,
-      "grad_norm": 0.003567602252587676,
-      "learning_rate": 0.00019999295133315653,
-      "loss": 46.0,
-      "step": 23479
-    },
-    {
-      "epoch": 3.7812713877370263,
-      "grad_norm": 0.0009228388662450016,
-      "learning_rate": 0.00019999295073247185,
-      "loss": 46.0,
-      "step": 23480
-    },
-    {
-      "epoch": 3.7814324248158138,
-      "grad_norm": 0.009420409798622131,
-      "learning_rate": 0.0001999929501317616,
-      "loss": 46.0,
-      "step": 23481
-    },
-    {
-      "epoch": 3.781593461894601,
-      "grad_norm": 0.0041953325271606445,
-      "learning_rate": 0.00019999294953102574,
-      "loss": 46.0,
-      "step": 23482
-    },
-    {
-      "epoch": 3.7817544989733887,
-      "grad_norm": 0.0046717398799955845,
-      "learning_rate": 0.0001999929489302643,
-      "loss": 46.0,
-      "step": 23483
-    },
-    {
-      "epoch": 3.781915536052176,
-      "grad_norm": 0.002981435274705291,
-      "learning_rate": 0.00019999294832947725,
-      "loss": 46.0,
-      "step": 23484
-    },
-    {
-      "epoch": 3.7820765731309636,
-      "grad_norm": 0.008540144190192223,
-      "learning_rate": 0.0001999929477286646,
-      "loss": 46.0,
-      "step": 23485
-    },
-    {
-      "epoch": 3.7822376102097506,
-      "grad_norm": 0.0069182561710476875,
-      "learning_rate": 0.00019999294712782638,
-      "loss": 46.0,
-      "step": 23486
-    },
-    {
-      "epoch": 3.782398647288538,
-      "grad_norm": 0.001253429683856666,
-      "learning_rate": 0.0001999929465269626,
-      "loss": 46.0,
-      "step": 23487
-    },
-    {
-      "epoch": 3.7825596843673255,
-      "grad_norm": 0.005894755478948355,
-      "learning_rate": 0.0001999929459260732,
-      "loss": 46.0,
-      "step": 23488
-    },
-    {
-      "epoch": 3.782720721446113,
-      "grad_norm": 0.0023197641130536795,
-      "learning_rate": 0.00019999294532515818,
-      "loss": 46.0,
-      "step": 23489
-    },
-    {
-      "epoch": 3.7828817585249004,
-      "grad_norm": 0.01157285738736391,
-      "learning_rate": 0.0001999929447242176,
-      "loss": 46.0,
-      "step": 23490
-    },
-    {
-      "epoch": 3.783042795603688,
-      "grad_norm": 0.0069605023600161076,
-      "learning_rate": 0.0001999929441232514,
-      "loss": 46.0,
-      "step": 23491
-    },
-    {
-      "epoch": 3.7832038326824753,
-      "grad_norm": 0.011911396868526936,
-      "learning_rate": 0.0001999929435222596,
-      "loss": 46.0,
-      "step": 23492
-    },
-    {
-      "epoch": 3.7833648697612627,
-      "grad_norm": 0.0012840928975492716,
-      "learning_rate": 0.00019999294292124224,
-      "loss": 46.0,
-      "step": 23493
-    },
-    {
-      "epoch": 3.78352590684005,
-      "grad_norm": 0.0012501165037974715,
-      "learning_rate": 0.00019999294232019927,
-      "loss": 46.0,
-      "step": 23494
-    },
-    {
-      "epoch": 3.783686943918837,
-      "grad_norm": 0.0013155360938981175,
-      "learning_rate": 0.00019999294171913073,
-      "loss": 46.0,
-      "step": 23495
-    },
-    {
-      "epoch": 3.7838479809976246,
-      "grad_norm": 0.0072088055312633514,
-      "learning_rate": 0.00019999294111803658,
-      "loss": 46.0,
-      "step": 23496
-    },
-    {
-      "epoch": 3.784009018076412,
-      "grad_norm": 0.007935171946883202,
-      "learning_rate": 0.0001999929405169168,
-      "loss": 46.0,
-      "step": 23497
-    },
-    {
-      "epoch": 3.7841700551551996,
-      "grad_norm": 0.0071401833556592464,
-      "learning_rate": 0.00019999293991577148,
-      "loss": 46.0,
-      "step": 23498
-    },
-    {
-      "epoch": 3.784331092233987,
-      "grad_norm": 0.0016121533699333668,
-      "learning_rate": 0.00019999293931460054,
-      "loss": 46.0,
-      "step": 23499
-    },
-    {
-      "epoch": 3.784492129312774,
-      "grad_norm": 0.01656302809715271,
-      "learning_rate": 0.000199992938713404,
-      "loss": 46.0,
-      "step": 23500
-    },
-    {
-      "epoch": 3.7846531663915615,
-      "grad_norm": 0.0021986172068864107,
-      "learning_rate": 0.0001999929381121819,
-      "loss": 46.0,
-      "step": 23501
-    },
-    {
-      "epoch": 3.784814203470349,
-      "grad_norm": 0.001959993503987789,
-      "learning_rate": 0.00019999293751093417,
-      "loss": 46.0,
-      "step": 23502
-    },
-    {
-      "epoch": 3.7849752405491364,
-      "grad_norm": 0.0034746001474559307,
-      "learning_rate": 0.00019999293690966088,
-      "loss": 46.0,
-      "step": 23503
-    },
-    {
-      "epoch": 3.785136277627924,
-      "grad_norm": 0.001077381893992424,
-      "learning_rate": 0.00019999293630836197,
-      "loss": 46.0,
-      "step": 23504
-    },
-    {
-      "epoch": 3.7852973147067113,
-      "grad_norm": 0.010049964301288128,
-      "learning_rate": 0.0001999929357070375,
-      "loss": 46.0,
-      "step": 23505
-    },
-    {
-      "epoch": 3.7854583517854987,
-      "grad_norm": 0.004013793542981148,
-      "learning_rate": 0.00019999293510568743,
-      "loss": 46.0,
-      "step": 23506
-    },
-    {
-      "epoch": 3.785619388864286,
-      "grad_norm": 0.004395932424813509,
-      "learning_rate": 0.0001999929345043117,
-      "loss": 46.0,
-      "step": 23507
-    },
-    {
-      "epoch": 3.7857804259430736,
-      "grad_norm": 0.0012676736805588007,
-      "learning_rate": 0.00019999293390291045,
-      "loss": 46.0,
-      "step": 23508
-    },
-    {
-      "epoch": 3.7859414630218606,
-      "grad_norm": 0.0021775392815470695,
-      "learning_rate": 0.00019999293330148361,
-      "loss": 46.0,
-      "step": 23509
-    },
-    {
-      "epoch": 3.786102500100648,
-      "grad_norm": 0.0024661454372107983,
-      "learning_rate": 0.00019999293270003113,
-      "loss": 46.0,
-      "step": 23510
-    },
-    {
-      "epoch": 3.7862635371794355,
-      "grad_norm": 0.009182114154100418,
-      "learning_rate": 0.0001999929320985531,
-      "loss": 46.0,
-      "step": 23511
-    },
-    {
-      "epoch": 3.786424574258223,
-      "grad_norm": 0.0041510877199471,
-      "learning_rate": 0.00019999293149704946,
-      "loss": 46.0,
-      "step": 23512
-    },
-    {
-      "epoch": 3.7865856113370104,
-      "grad_norm": 0.010162005200982094,
-      "learning_rate": 0.0001999929308955202,
-      "loss": 46.0,
-      "step": 23513
-    },
-    {
-      "epoch": 3.786746648415798,
-      "grad_norm": 0.0013867107918486,
-      "learning_rate": 0.00019999293029396536,
-      "loss": 46.0,
-      "step": 23514
-    },
-    {
-      "epoch": 3.786907685494585,
-      "grad_norm": 0.001820576493628323,
-      "learning_rate": 0.00019999292969238494,
-      "loss": 46.0,
-      "step": 23515
-    },
-    {
-      "epoch": 3.7870687225733723,
-      "grad_norm": 0.01022383477538824,
-      "learning_rate": 0.00019999292909077894,
-      "loss": 46.0,
-      "step": 23516
-    },
-    {
-      "epoch": 3.78722975965216,
-      "grad_norm": 0.008014820516109467,
-      "learning_rate": 0.00019999292848914735,
-      "loss": 46.0,
-      "step": 23517
-    },
-    {
-      "epoch": 3.7873907967309473,
-      "grad_norm": 0.001742486609145999,
-      "learning_rate": 0.00019999292788749014,
-      "loss": 46.0,
-      "step": 23518
-    },
-    {
-      "epoch": 3.7875518338097347,
-      "grad_norm": 0.0037293757777661085,
-      "learning_rate": 0.00019999292728580735,
-      "loss": 46.0,
-      "step": 23519
-    },
-    {
-      "epoch": 3.787712870888522,
-      "grad_norm": 0.005462685599923134,
-      "learning_rate": 0.00019999292668409897,
-      "loss": 46.0,
-      "step": 23520
-    },
-    {
-      "epoch": 3.7878739079673096,
-      "grad_norm": 0.019822850823402405,
-      "learning_rate": 0.00019999292608236497,
-      "loss": 46.0,
-      "step": 23521
-    },
-    {
-      "epoch": 3.788034945046097,
-      "grad_norm": 0.0037660361267626286,
-      "learning_rate": 0.0001999929254806054,
-      "loss": 46.0,
-      "step": 23522
-    },
-    {
-      "epoch": 3.7881959821248845,
-      "grad_norm": 0.008371307514607906,
-      "learning_rate": 0.00019999292487882025,
-      "loss": 46.0,
-      "step": 23523
-    },
-    {
-      "epoch": 3.7883570192036715,
-      "grad_norm": 0.0005686399526894093,
-      "learning_rate": 0.00019999292427700947,
-      "loss": 46.0,
-      "step": 23524
-    },
-    {
-      "epoch": 3.788518056282459,
-      "grad_norm": 0.005702432710677385,
-      "learning_rate": 0.00019999292367517312,
-      "loss": 46.0,
-      "step": 23525
-    },
-    {
-      "epoch": 3.7886790933612464,
-      "grad_norm": 0.004375653341412544,
-      "learning_rate": 0.00019999292307331117,
-      "loss": 46.0,
-      "step": 23526
-    },
-    {
-      "epoch": 3.788840130440034,
-      "grad_norm": 0.0016482442151755095,
-      "learning_rate": 0.00019999292247142365,
-      "loss": 46.0,
-      "step": 23527
-    },
-    {
-      "epoch": 3.7890011675188213,
-      "grad_norm": 0.007545525208115578,
-      "learning_rate": 0.00019999292186951052,
-      "loss": 46.0,
-      "step": 23528
-    },
-    {
-      "epoch": 3.7891622045976083,
-      "grad_norm": 0.00484321266412735,
-      "learning_rate": 0.0001999929212675718,
-      "loss": 46.0,
-      "step": 23529
-    },
-    {
-      "epoch": 3.789323241676396,
-      "grad_norm": 0.006865546572953463,
-      "learning_rate": 0.00019999292066560746,
-      "loss": 46.0,
-      "step": 23530
-    },
-    {
-      "epoch": 3.7894842787551832,
-      "grad_norm": 0.0033078778069466352,
-      "learning_rate": 0.00019999292006361757,
-      "loss": 46.0,
-      "step": 23531
-    },
-    {
-      "epoch": 3.7896453158339707,
-      "grad_norm": 0.0017028943402692676,
-      "learning_rate": 0.00019999291946160206,
-      "loss": 46.0,
-      "step": 23532
-    },
-    {
-      "epoch": 3.789806352912758,
-      "grad_norm": 0.0033133940305560827,
-      "learning_rate": 0.00019999291885956094,
-      "loss": 46.0,
-      "step": 23533
-    },
-    {
-      "epoch": 3.7899673899915456,
-      "grad_norm": 0.009623431600630283,
-      "learning_rate": 0.00019999291825749426,
-      "loss": 46.0,
-      "step": 23534
-    },
-    {
-      "epoch": 3.790128427070333,
-      "grad_norm": 0.0015019259881228209,
-      "learning_rate": 0.000199992917655402,
-      "loss": 46.0,
-      "step": 23535
-    },
-    {
-      "epoch": 3.7902894641491205,
-      "grad_norm": 0.0028213916812092066,
-      "learning_rate": 0.0001999929170532841,
-      "loss": 46.0,
-      "step": 23536
-    },
-    {
-      "epoch": 3.790450501227908,
-      "grad_norm": 0.013229393400251865,
-      "learning_rate": 0.00019999291645114064,
-      "loss": 46.0,
-      "step": 23537
-    },
-    {
-      "epoch": 3.7906115383066954,
-      "grad_norm": 0.005452876444905996,
-      "learning_rate": 0.00019999291584897158,
-      "loss": 46.0,
-      "step": 23538
-    },
-    {
-      "epoch": 3.7907725753854824,
-      "grad_norm": 0.006903988774865866,
-      "learning_rate": 0.0001999929152467769,
-      "loss": 46.0,
-      "step": 23539
-    },
-    {
-      "epoch": 3.79093361246427,
-      "grad_norm": 0.0015960269374772906,
-      "learning_rate": 0.00019999291464455667,
-      "loss": 46.0,
-      "step": 23540
-    },
-    {
-      "epoch": 3.7910946495430573,
-      "grad_norm": 0.0014241775497794151,
-      "learning_rate": 0.00019999291404231083,
-      "loss": 46.0,
-      "step": 23541
-    },
-    {
-      "epoch": 3.7912556866218448,
-      "grad_norm": 0.0013796341372653842,
-      "learning_rate": 0.0001999929134400394,
-      "loss": 46.0,
-      "step": 23542
-    },
-    {
-      "epoch": 3.791416723700632,
-      "grad_norm": 0.006222702097147703,
-      "learning_rate": 0.00019999291283774237,
-      "loss": 46.0,
-      "step": 23543
-    },
-    {
-      "epoch": 3.791577760779419,
-      "grad_norm": 0.0026718589942902327,
-      "learning_rate": 0.00019999291223541974,
-      "loss": 46.0,
-      "step": 23544
-    },
-    {
-      "epoch": 3.7917387978582067,
-      "grad_norm": 0.005570186302065849,
-      "learning_rate": 0.00019999291163307151,
-      "loss": 46.0,
-      "step": 23545
-    },
-    {
-      "epoch": 3.791899834936994,
-      "grad_norm": 0.0015155143337324262,
-      "learning_rate": 0.00019999291103069773,
-      "loss": 46.0,
-      "step": 23546
-    },
-    {
-      "epoch": 3.7920608720157816,
-      "grad_norm": 0.004810272250324488,
-      "learning_rate": 0.00019999291042829834,
-      "loss": 46.0,
-      "step": 23547
-    },
-    {
-      "epoch": 3.792221909094569,
-      "grad_norm": 0.0016728319460526109,
-      "learning_rate": 0.00019999290982587332,
-      "loss": 46.0,
-      "step": 23548
-    },
-    {
-      "epoch": 3.7923829461733565,
-      "grad_norm": 0.002594210673123598,
-      "learning_rate": 0.00019999290922342273,
-      "loss": 46.0,
-      "step": 23549
-    },
-    {
-      "epoch": 3.792543983252144,
-      "grad_norm": 0.00584800262004137,
-      "learning_rate": 0.00019999290862094657,
-      "loss": 46.0,
-      "step": 23550
-    },
-    {
-      "epoch": 3.7927050203309314,
-      "grad_norm": 0.0037820106372237206,
-      "learning_rate": 0.0001999929080184448,
-      "loss": 46.0,
-      "step": 23551
-    },
-    {
-      "epoch": 3.792866057409719,
-      "grad_norm": 0.0065300436690449715,
-      "learning_rate": 0.00019999290741591744,
-      "loss": 46.0,
-      "step": 23552
-    },
-    {
-      "epoch": 3.793027094488506,
-      "grad_norm": 0.002413803245872259,
-      "learning_rate": 0.0001999929068133645,
-      "loss": 46.0,
-      "step": 23553
-    },
-    {
-      "epoch": 3.7931881315672933,
-      "grad_norm": 0.0037758895196020603,
-      "learning_rate": 0.00019999290621078593,
-      "loss": 46.0,
-      "step": 23554
-    },
-    {
-      "epoch": 3.7933491686460807,
-      "grad_norm": 0.0014778823824599385,
-      "learning_rate": 0.00019999290560818178,
-      "loss": 46.0,
-      "step": 23555
-    },
-    {
-      "epoch": 3.793510205724868,
-      "grad_norm": 0.0028374451212584972,
-      "learning_rate": 0.00019999290500555204,
-      "loss": 46.0,
-      "step": 23556
-    },
-    {
-      "epoch": 3.7936712428036556,
-      "grad_norm": 0.003306510392576456,
-      "learning_rate": 0.0001999929044028967,
-      "loss": 46.0,
-      "step": 23557
-    },
-    {
-      "epoch": 3.793832279882443,
-      "grad_norm": 0.0024255476891994476,
-      "learning_rate": 0.00019999290380021579,
-      "loss": 46.0,
-      "step": 23558
-    },
-    {
-      "epoch": 3.79399331696123,
-      "grad_norm": 0.005692268721759319,
-      "learning_rate": 0.00019999290319750926,
-      "loss": 46.0,
-      "step": 23559
-    },
-    {
-      "epoch": 3.7941543540400176,
-      "grad_norm": 0.003182213054969907,
-      "learning_rate": 0.00019999290259477718,
-      "loss": 46.0,
-      "step": 23560
-    },
-    {
-      "epoch": 3.794315391118805,
-      "grad_norm": 0.004343269858509302,
-      "learning_rate": 0.00019999290199201948,
-      "loss": 46.0,
-      "step": 23561
-    },
-    {
-      "epoch": 3.7944764281975925,
-      "grad_norm": 0.001727725611999631,
-      "learning_rate": 0.00019999290138923617,
-      "loss": 46.0,
-      "step": 23562
-    },
-    {
-      "epoch": 3.79463746527638,
-      "grad_norm": 0.0019260860281065106,
-      "learning_rate": 0.00019999290078642727,
-      "loss": 46.0,
-      "step": 23563
-    },
-    {
-      "epoch": 3.7947985023551674,
-      "grad_norm": 0.008498795330524445,
-      "learning_rate": 0.00019999290018359278,
-      "loss": 46.0,
-      "step": 23564
-    },
-    {
-      "epoch": 3.794959539433955,
-      "grad_norm": 0.004467620514333248,
-      "learning_rate": 0.0001999928995807327,
-      "loss": 46.0,
-      "step": 23565
-    },
-    {
-      "epoch": 3.7951205765127423,
-      "grad_norm": 0.008226759731769562,
-      "learning_rate": 0.00019999289897784704,
-      "loss": 46.0,
-      "step": 23566
-    },
-    {
-      "epoch": 3.7952816135915297,
-      "grad_norm": 0.004658955615013838,
-      "learning_rate": 0.00019999289837493577,
-      "loss": 46.0,
-      "step": 23567
-    },
-    {
-      "epoch": 3.7954426506703167,
-      "grad_norm": 0.0036381580866873264,
-      "learning_rate": 0.00019999289777199893,
-      "loss": 46.0,
-      "step": 23568
-    },
-    {
-      "epoch": 3.795603687749104,
-      "grad_norm": 0.001321410178206861,
-      "learning_rate": 0.00019999289716903648,
-      "loss": 46.0,
-      "step": 23569
-    },
-    {
-      "epoch": 3.7957647248278916,
-      "grad_norm": 0.0018899224232882261,
-      "learning_rate": 0.00019999289656604845,
-      "loss": 46.0,
-      "step": 23570
-    },
-    {
-      "epoch": 3.795925761906679,
-      "grad_norm": 0.008362204767763615,
-      "learning_rate": 0.00019999289596303482,
-      "loss": 46.0,
-      "step": 23571
-    },
-    {
-      "epoch": 3.7960867989854665,
-      "grad_norm": 0.001967393094673753,
-      "learning_rate": 0.00019999289535999558,
-      "loss": 46.0,
-      "step": 23572
-    },
-    {
-      "epoch": 3.7962478360642535,
-      "grad_norm": 0.020419172942638397,
-      "learning_rate": 0.00019999289475693078,
-      "loss": 46.0,
-      "step": 23573
-    },
-    {
-      "epoch": 3.796408873143041,
-      "grad_norm": 0.007248731795698404,
-      "learning_rate": 0.00019999289415384034,
-      "loss": 46.0,
-      "step": 23574
-    },
-    {
-      "epoch": 3.7965699102218284,
-      "grad_norm": 0.0038336936850100756,
-      "learning_rate": 0.00019999289355072434,
-      "loss": 46.0,
-      "step": 23575
-    },
-    {
-      "epoch": 3.796730947300616,
-      "grad_norm": 0.0012371959164738655,
-      "learning_rate": 0.00019999289294758276,
-      "loss": 46.0,
-      "step": 23576
-    },
-    {
-      "epoch": 3.7968919843794033,
-      "grad_norm": 0.005969126243144274,
-      "learning_rate": 0.00019999289234441556,
-      "loss": 46.0,
-      "step": 23577
-    },
-    {
-      "epoch": 3.797053021458191,
-      "grad_norm": 0.001394742401316762,
-      "learning_rate": 0.00019999289174122277,
-      "loss": 46.0,
-      "step": 23578
-    },
-    {
-      "epoch": 3.7972140585369782,
-      "grad_norm": 0.006398888770490885,
-      "learning_rate": 0.00019999289113800442,
-      "loss": 46.0,
-      "step": 23579
-    },
-    {
-      "epoch": 3.7973750956157657,
-      "grad_norm": 0.002249662997201085,
-      "learning_rate": 0.00019999289053476043,
-      "loss": 46.0,
-      "step": 23580
-    },
-    {
-      "epoch": 3.797536132694553,
-      "grad_norm": 0.011105634272098541,
-      "learning_rate": 0.00019999288993149085,
-      "loss": 46.0,
-      "step": 23581
-    },
-    {
-      "epoch": 3.79769716977334,
-      "grad_norm": 0.004338633269071579,
-      "learning_rate": 0.0001999928893281957,
-      "loss": 46.0,
-      "step": 23582
-    },
-    {
-      "epoch": 3.7978582068521276,
-      "grad_norm": 0.002980279503390193,
-      "learning_rate": 0.00019999288872487496,
-      "loss": 46.0,
-      "step": 23583
-    },
-    {
-      "epoch": 3.798019243930915,
-      "grad_norm": 0.002213576342910528,
-      "learning_rate": 0.00019999288812152862,
-      "loss": 46.0,
-      "step": 23584
-    },
-    {
-      "epoch": 3.7981802810097025,
-      "grad_norm": 0.0026220022700726986,
-      "learning_rate": 0.0001999928875181567,
-      "loss": 46.0,
-      "step": 23585
-    },
-    {
-      "epoch": 3.79834131808849,
-      "grad_norm": 0.003433849662542343,
-      "learning_rate": 0.00019999288691475916,
-      "loss": 46.0,
-      "step": 23586
-    },
-    {
-      "epoch": 3.7985023551672774,
-      "grad_norm": 0.006517481058835983,
-      "learning_rate": 0.00019999288631133603,
-      "loss": 46.0,
-      "step": 23587
-    },
-    {
-      "epoch": 3.7986633922460644,
-      "grad_norm": 0.0023642985615879297,
-      "learning_rate": 0.00019999288570788731,
-      "loss": 46.0,
-      "step": 23588
-    },
-    {
-      "epoch": 3.798824429324852,
-      "grad_norm": 0.003381990361958742,
-      "learning_rate": 0.000199992885104413,
-      "loss": 46.0,
-      "step": 23589
-    },
-    {
-      "epoch": 3.7989854664036393,
-      "grad_norm": 0.0054267458617687225,
-      "learning_rate": 0.00019999288450091312,
-      "loss": 46.0,
-      "step": 23590
-    },
-    {
-      "epoch": 3.7991465034824268,
-      "grad_norm": 0.0018450272036716342,
-      "learning_rate": 0.00019999288389738762,
-      "loss": 46.0,
-      "step": 23591
-    },
-    {
-      "epoch": 3.7993075405612142,
-      "grad_norm": 0.0011486992007121444,
-      "learning_rate": 0.00019999288329383653,
-      "loss": 46.0,
-      "step": 23592
-    },
-    {
-      "epoch": 3.7994685776400017,
-      "grad_norm": 0.0042379009537398815,
-      "learning_rate": 0.00019999288269025985,
-      "loss": 46.0,
-      "step": 23593
-    },
-    {
-      "epoch": 3.799629614718789,
-      "grad_norm": 0.004558446817100048,
-      "learning_rate": 0.0001999928820866576,
-      "loss": 46.0,
-      "step": 23594
-    },
-    {
-      "epoch": 3.7997906517975766,
-      "grad_norm": 0.0015566786751151085,
-      "learning_rate": 0.00019999288148302974,
-      "loss": 46.0,
-      "step": 23595
-    },
-    {
-      "epoch": 3.799951688876364,
-      "grad_norm": 0.0025924034416675568,
-      "learning_rate": 0.00019999288087937624,
-      "loss": 46.0,
-      "step": 23596
-    },
-    {
-      "epoch": 3.800112725955151,
-      "grad_norm": 0.008190043270587921,
-      "learning_rate": 0.00019999288027569722,
-      "loss": 46.0,
-      "step": 23597
-    },
-    {
-      "epoch": 3.8002737630339385,
-      "grad_norm": 0.00351817742921412,
-      "learning_rate": 0.00019999287967199255,
-      "loss": 46.0,
-      "step": 23598
-    },
-    {
-      "epoch": 3.800434800112726,
-      "grad_norm": 0.001109899254515767,
-      "learning_rate": 0.00019999287906826232,
-      "loss": 46.0,
-      "step": 23599
-    },
-    {
-      "epoch": 3.8005958371915134,
-      "grad_norm": 0.0076414719223976135,
-      "learning_rate": 0.00019999287846450648,
-      "loss": 46.0,
-      "step": 23600
-    },
-    {
-      "epoch": 3.800756874270301,
-      "grad_norm": 0.0014270510291680694,
-      "learning_rate": 0.00019999287786072505,
-      "loss": 46.0,
-      "step": 23601
-    },
-    {
-      "epoch": 3.800917911349088,
-      "grad_norm": 0.003094103652983904,
-      "learning_rate": 0.00019999287725691803,
-      "loss": 46.0,
-      "step": 23602
-    },
-    {
-      "epoch": 3.8010789484278753,
-      "grad_norm": 0.008699733763933182,
-      "learning_rate": 0.00019999287665308543,
-      "loss": 46.0,
-      "step": 23603
-    },
-    {
-      "epoch": 3.8012399855066628,
-      "grad_norm": 0.004201273433864117,
-      "learning_rate": 0.00019999287604922724,
-      "loss": 46.0,
-      "step": 23604
-    },
-    {
-      "epoch": 3.80140102258545,
-      "grad_norm": 0.00544446986168623,
-      "learning_rate": 0.0001999928754453434,
-      "loss": 46.0,
-      "step": 23605
-    },
-    {
-      "epoch": 3.8015620596642377,
-      "grad_norm": 0.0029939047526568174,
-      "learning_rate": 0.00019999287484143404,
-      "loss": 46.0,
-      "step": 23606
-    },
-    {
-      "epoch": 3.801723096743025,
-      "grad_norm": 0.004269227851182222,
-      "learning_rate": 0.00019999287423749906,
-      "loss": 46.0,
-      "step": 23607
-    },
-    {
-      "epoch": 3.8018841338218126,
-      "grad_norm": 0.008536243811249733,
-      "learning_rate": 0.0001999928736335385,
-      "loss": 46.0,
-      "step": 23608
-    },
-    {
-      "epoch": 3.8020451709006,
-      "grad_norm": 0.0010491583961993456,
-      "learning_rate": 0.00019999287302955232,
-      "loss": 46.0,
-      "step": 23609
-    },
-    {
-      "epoch": 3.8022062079793875,
-      "grad_norm": 0.0008426276035606861,
-      "learning_rate": 0.00019999287242554057,
-      "loss": 46.0,
-      "step": 23610
-    },
-    {
-      "epoch": 3.802367245058175,
-      "grad_norm": 0.0027947211638092995,
-      "learning_rate": 0.00019999287182150322,
-      "loss": 46.0,
-      "step": 23611
-    },
-    {
-      "epoch": 3.802528282136962,
-      "grad_norm": 0.0009417349356226623,
-      "learning_rate": 0.00019999287121744025,
-      "loss": 46.0,
-      "step": 23612
-    },
-    {
-      "epoch": 3.8026893192157494,
-      "grad_norm": 0.003538094460964203,
-      "learning_rate": 0.00019999287061335172,
-      "loss": 46.0,
-      "step": 23613
-    },
-    {
-      "epoch": 3.802850356294537,
-      "grad_norm": 0.0024768486618995667,
-      "learning_rate": 0.00019999287000923758,
-      "loss": 46.0,
-      "step": 23614
-    },
-    {
-      "epoch": 3.8030113933733243,
-      "grad_norm": 0.004929191432893276,
-      "learning_rate": 0.00019999286940509785,
-      "loss": 46.0,
-      "step": 23615
-    },
-    {
-      "epoch": 3.8031724304521117,
-      "grad_norm": 0.0013620893005281687,
-      "learning_rate": 0.00019999286880093253,
-      "loss": 46.0,
-      "step": 23616
-    },
-    {
-      "epoch": 3.8033334675308987,
-      "grad_norm": 0.0025519547052681446,
-      "learning_rate": 0.00019999286819674162,
-      "loss": 46.0,
-      "step": 23617
-    },
-    {
-      "epoch": 3.803494504609686,
-      "grad_norm": 0.002054918324574828,
-      "learning_rate": 0.0001999928675925251,
-      "loss": 46.0,
-      "step": 23618
-    },
-    {
-      "epoch": 3.8036555416884736,
-      "grad_norm": 0.007278750650584698,
-      "learning_rate": 0.00019999286698828302,
-      "loss": 46.0,
-      "step": 23619
-    },
-    {
-      "epoch": 3.803816578767261,
-      "grad_norm": 0.01253379974514246,
-      "learning_rate": 0.00019999286638401533,
-      "loss": 46.0,
-      "step": 23620
-    },
-    {
-      "epoch": 3.8039776158460485,
-      "grad_norm": 0.0026233820244669914,
-      "learning_rate": 0.00019999286577972205,
-      "loss": 46.0,
-      "step": 23621
-    },
-    {
-      "epoch": 3.804138652924836,
-      "grad_norm": 0.00379826407879591,
-      "learning_rate": 0.00019999286517540315,
-      "loss": 46.0,
-      "step": 23622
-    },
-    {
-      "epoch": 3.8042996900036234,
-      "grad_norm": 0.0019124912796542048,
-      "learning_rate": 0.0001999928645710587,
-      "loss": 46.0,
-      "step": 23623
-    },
-    {
-      "epoch": 3.804460727082411,
-      "grad_norm": 0.009216277860105038,
-      "learning_rate": 0.00019999286396668863,
-      "loss": 46.0,
-      "step": 23624
-    },
-    {
-      "epoch": 3.8046217641611983,
-      "grad_norm": 0.003284048056229949,
-      "learning_rate": 0.00019999286336229297,
-      "loss": 46.0,
-      "step": 23625
-    },
-    {
-      "epoch": 3.8047828012399854,
-      "grad_norm": 0.004107843153178692,
-      "learning_rate": 0.00019999286275787173,
-      "loss": 46.0,
-      "step": 23626
-    },
-    {
-      "epoch": 3.804943838318773,
-      "grad_norm": 0.005023350939154625,
-      "learning_rate": 0.00019999286215342487,
-      "loss": 46.0,
-      "step": 23627
-    },
-    {
-      "epoch": 3.8051048753975603,
-      "grad_norm": 0.020202776417136192,
-      "learning_rate": 0.00019999286154895245,
-      "loss": 46.0,
-      "step": 23628
-    },
-    {
-      "epoch": 3.8052659124763477,
-      "grad_norm": 0.004103577695786953,
-      "learning_rate": 0.0001999928609444544,
-      "loss": 46.0,
-      "step": 23629
-    },
-    {
-      "epoch": 3.805426949555135,
-      "grad_norm": 0.0029470762237906456,
-      "learning_rate": 0.0001999928603399308,
-      "loss": 46.0,
-      "step": 23630
-    },
-    {
-      "epoch": 3.8055879866339226,
-      "grad_norm": 0.0005229293019510806,
-      "learning_rate": 0.00019999285973538157,
-      "loss": 46.0,
-      "step": 23631
-    },
-    {
-      "epoch": 3.8057490237127096,
-      "grad_norm": 0.002074221847578883,
-      "learning_rate": 0.00019999285913080677,
-      "loss": 46.0,
-      "step": 23632
-    },
-    {
-      "epoch": 3.805910060791497,
-      "grad_norm": 0.0034410192165523767,
-      "learning_rate": 0.00019999285852620636,
-      "loss": 46.0,
-      "step": 23633
-    },
-    {
-      "epoch": 3.8060710978702845,
-      "grad_norm": 0.007800699677318335,
-      "learning_rate": 0.00019999285792158037,
-      "loss": 46.0,
-      "step": 23634
-    },
-    {
-      "epoch": 3.806232134949072,
-      "grad_norm": 0.0005563221056945622,
-      "learning_rate": 0.00019999285731692878,
-      "loss": 46.0,
-      "step": 23635
-    },
-    {
-      "epoch": 3.8063931720278594,
-      "grad_norm": 0.008285616524517536,
-      "learning_rate": 0.0001999928567122516,
-      "loss": 46.0,
-      "step": 23636
-    },
-    {
-      "epoch": 3.806554209106647,
-      "grad_norm": 0.0011474619386717677,
-      "learning_rate": 0.0001999928561075488,
-      "loss": 46.0,
-      "step": 23637
-    },
-    {
-      "epoch": 3.8067152461854343,
-      "grad_norm": 0.008062407374382019,
-      "learning_rate": 0.00019999285550282046,
-      "loss": 46.0,
-      "step": 23638
-    },
-    {
-      "epoch": 3.806876283264222,
-      "grad_norm": 0.002201765775680542,
-      "learning_rate": 0.0001999928548980665,
-      "loss": 46.0,
-      "step": 23639
-    },
-    {
-      "epoch": 3.8070373203430092,
-      "grad_norm": 0.001507507637143135,
-      "learning_rate": 0.00019999285429328693,
-      "loss": 46.0,
-      "step": 23640
-    },
-    {
-      "epoch": 3.8071983574217962,
-      "grad_norm": 0.0014076469233259559,
-      "learning_rate": 0.0001999928536884818,
-      "loss": 46.0,
-      "step": 23641
-    },
-    {
-      "epoch": 3.8073593945005837,
-      "grad_norm": 0.0054513937793672085,
-      "learning_rate": 0.00019999285308365105,
-      "loss": 46.0,
-      "step": 23642
-    },
-    {
-      "epoch": 3.807520431579371,
-      "grad_norm": 0.005643813870847225,
-      "learning_rate": 0.00019999285247879471,
-      "loss": 46.0,
-      "step": 23643
-    },
-    {
-      "epoch": 3.8076814686581586,
-      "grad_norm": 0.007829139940440655,
-      "learning_rate": 0.0001999928518739128,
-      "loss": 46.0,
-      "step": 23644
-    },
-    {
-      "epoch": 3.807842505736946,
-      "grad_norm": 0.0011677708243951201,
-      "learning_rate": 0.00019999285126900528,
-      "loss": 46.0,
-      "step": 23645
-    },
-    {
-      "epoch": 3.808003542815733,
-      "grad_norm": 0.0022964433301240206,
-      "learning_rate": 0.00019999285066407216,
-      "loss": 46.0,
-      "step": 23646
-    },
-    {
-      "epoch": 3.8081645798945205,
-      "grad_norm": 0.001790171256288886,
-      "learning_rate": 0.00019999285005911345,
-      "loss": 46.0,
-      "step": 23647
-    },
-    {
-      "epoch": 3.808325616973308,
-      "grad_norm": 0.0019000850152224302,
-      "learning_rate": 0.00019999284945412918,
-      "loss": 46.0,
-      "step": 23648
-    },
-    {
-      "epoch": 3.8084866540520954,
-      "grad_norm": 0.0055053322575986385,
-      "learning_rate": 0.00019999284884911927,
-      "loss": 46.0,
-      "step": 23649
-    },
-    {
-      "epoch": 3.808647691130883,
-      "grad_norm": 0.002368929563090205,
-      "learning_rate": 0.0001999928482440838,
-      "loss": 46.0,
-      "step": 23650
-    },
-    {
-      "epoch": 3.8088087282096703,
-      "grad_norm": 0.004623609595000744,
-      "learning_rate": 0.0001999928476390227,
-      "loss": 46.0,
-      "step": 23651
-    },
-    {
-      "epoch": 3.8089697652884578,
-      "grad_norm": 0.006825034040957689,
-      "learning_rate": 0.00019999284703393607,
-      "loss": 46.0,
-      "step": 23652
-    },
-    {
-      "epoch": 3.809130802367245,
-      "grad_norm": 0.0015381784178316593,
-      "learning_rate": 0.00019999284642882378,
-      "loss": 46.0,
-      "step": 23653
-    },
-    {
-      "epoch": 3.8092918394460327,
-      "grad_norm": 0.007325910031795502,
-      "learning_rate": 0.00019999284582368593,
-      "loss": 46.0,
-      "step": 23654
-    },
-    {
-      "epoch": 3.8094528765248197,
-      "grad_norm": 0.014854847453534603,
-      "learning_rate": 0.0001999928452185225,
-      "loss": 46.0,
-      "step": 23655
-    },
-    {
-      "epoch": 3.809613913603607,
-      "grad_norm": 0.0021445502061396837,
-      "learning_rate": 0.00019999284461333345,
-      "loss": 46.0,
-      "step": 23656
-    },
-    {
-      "epoch": 3.8097749506823946,
-      "grad_norm": 0.005499368533492088,
-      "learning_rate": 0.0001999928440081188,
-      "loss": 46.0,
-      "step": 23657
-    },
-    {
-      "epoch": 3.809935987761182,
-      "grad_norm": 0.0010063834488391876,
-      "learning_rate": 0.00019999284340287856,
-      "loss": 46.0,
-      "step": 23658
-    },
-    {
-      "epoch": 3.8100970248399695,
-      "grad_norm": 0.0063238199800252914,
-      "learning_rate": 0.00019999284279761278,
-      "loss": 46.0,
-      "step": 23659
-    },
-    {
-      "epoch": 3.810258061918757,
-      "grad_norm": 0.0030097351409494877,
-      "learning_rate": 0.00019999284219232135,
-      "loss": 46.0,
-      "step": 23660
-    },
-    {
-      "epoch": 3.810419098997544,
-      "grad_norm": 0.003349976846948266,
-      "learning_rate": 0.00019999284158700434,
-      "loss": 46.0,
-      "step": 23661
-    },
-    {
-      "epoch": 3.8105801360763314,
-      "grad_norm": 0.0038509576115757227,
-      "learning_rate": 0.00019999284098166174,
-      "loss": 46.0,
-      "step": 23662
-    },
-    {
-      "epoch": 3.810741173155119,
-      "grad_norm": 0.006076053716242313,
-      "learning_rate": 0.00019999284037629353,
-      "loss": 46.0,
-      "step": 23663
-    },
-    {
-      "epoch": 3.8109022102339063,
-      "grad_norm": 0.00255609885789454,
-      "learning_rate": 0.00019999283977089976,
-      "loss": 46.0,
-      "step": 23664
-    },
-    {
-      "epoch": 3.8110632473126937,
-      "grad_norm": 0.005714502651244402,
-      "learning_rate": 0.00019999283916548037,
-      "loss": 46.0,
-      "step": 23665
-    },
-    {
-      "epoch": 3.811224284391481,
-      "grad_norm": 0.0011660543968901038,
-      "learning_rate": 0.00019999283856003542,
-      "loss": 46.0,
-      "step": 23666
-    },
-    {
-      "epoch": 3.8113853214702687,
-      "grad_norm": 0.005012121517211199,
-      "learning_rate": 0.00019999283795456486,
-      "loss": 46.0,
-      "step": 23667
-    },
-    {
-      "epoch": 3.811546358549056,
-      "grad_norm": 0.0010740929283201694,
-      "learning_rate": 0.00019999283734906868,
-      "loss": 46.0,
-      "step": 23668
-    },
-    {
-      "epoch": 3.8117073956278436,
-      "grad_norm": 0.0013253248762339354,
-      "learning_rate": 0.00019999283674354695,
-      "loss": 46.0,
-      "step": 23669
-    },
-    {
-      "epoch": 3.8118684327066306,
-      "grad_norm": 0.0023376839235424995,
-      "learning_rate": 0.0001999928361379996,
-      "loss": 46.0,
-      "step": 23670
-    },
-    {
-      "epoch": 3.812029469785418,
-      "grad_norm": 0.007657058071345091,
-      "learning_rate": 0.00019999283553242666,
-      "loss": 46.0,
-      "step": 23671
-    },
-    {
-      "epoch": 3.8121905068642055,
-      "grad_norm": 0.004085401073098183,
-      "learning_rate": 0.00019999283492682813,
-      "loss": 46.0,
-      "step": 23672
-    },
-    {
-      "epoch": 3.812351543942993,
-      "grad_norm": 0.0011746814707294106,
-      "learning_rate": 0.000199992834321204,
-      "loss": 46.0,
-      "step": 23673
-    },
-    {
-      "epoch": 3.8125125810217804,
-      "grad_norm": 0.004453752189874649,
-      "learning_rate": 0.0001999928337155543,
-      "loss": 46.0,
-      "step": 23674
-    },
-    {
-      "epoch": 3.812673618100568,
-      "grad_norm": 0.01846867799758911,
-      "learning_rate": 0.00019999283310987898,
-      "loss": 46.0,
-      "step": 23675
-    },
-    {
-      "epoch": 3.812834655179355,
-      "grad_norm": 0.0049817850813269615,
-      "learning_rate": 0.00019999283250417805,
-      "loss": 46.0,
-      "step": 23676
-    },
-    {
-      "epoch": 3.8129956922581423,
-      "grad_norm": 0.002020810265094042,
-      "learning_rate": 0.00019999283189845157,
-      "loss": 46.0,
-      "step": 23677
-    },
-    {
-      "epoch": 3.8131567293369297,
-      "grad_norm": 0.0018876423127949238,
-      "learning_rate": 0.0001999928312926995,
-      "loss": 46.0,
-      "step": 23678
-    },
-    {
-      "epoch": 3.813317766415717,
-      "grad_norm": 0.005340597592294216,
-      "learning_rate": 0.0001999928306869218,
-      "loss": 46.0,
-      "step": 23679
-    },
-    {
-      "epoch": 3.8134788034945046,
-      "grad_norm": 0.0054100798442959785,
-      "learning_rate": 0.00019999283008111853,
-      "loss": 46.0,
-      "step": 23680
-    },
-    {
-      "epoch": 3.813639840573292,
-      "grad_norm": 0.0010715537937358022,
-      "learning_rate": 0.00019999282947528966,
-      "loss": 46.0,
-      "step": 23681
-    },
-    {
-      "epoch": 3.8138008776520795,
-      "grad_norm": 0.005323282442986965,
-      "learning_rate": 0.0001999928288694352,
-      "loss": 46.0,
-      "step": 23682
-    },
-    {
-      "epoch": 3.813961914730867,
-      "grad_norm": 0.016577554866671562,
-      "learning_rate": 0.00019999282826355515,
-      "loss": 46.0,
-      "step": 23683
-    },
-    {
-      "epoch": 3.8141229518096544,
-      "grad_norm": 0.005348172504454851,
-      "learning_rate": 0.0001999928276576495,
-      "loss": 46.0,
-      "step": 23684
-    },
-    {
-      "epoch": 3.8142839888884414,
-      "grad_norm": 0.004378584213554859,
-      "learning_rate": 0.00019999282705171828,
-      "loss": 46.0,
-      "step": 23685
-    },
-    {
-      "epoch": 3.814445025967229,
-      "grad_norm": 0.002704095793887973,
-      "learning_rate": 0.00019999282644576143,
-      "loss": 46.0,
-      "step": 23686
-    },
-    {
-      "epoch": 3.8146060630460163,
-      "grad_norm": 0.003326993901282549,
-      "learning_rate": 0.00019999282583977901,
-      "loss": 46.0,
-      "step": 23687
-    },
-    {
-      "epoch": 3.814767100124804,
-      "grad_norm": 0.006455174647271633,
-      "learning_rate": 0.00019999282523377099,
-      "loss": 46.0,
-      "step": 23688
-    },
-    {
-      "epoch": 3.8149281372035913,
-      "grad_norm": 0.010551992803812027,
-      "learning_rate": 0.0001999928246277374,
-      "loss": 46.0,
-      "step": 23689
-    },
-    {
-      "epoch": 3.8150891742823783,
-      "grad_norm": 0.0036160375457257032,
-      "learning_rate": 0.00019999282402167817,
-      "loss": 46.0,
-      "step": 23690
-    },
-    {
-      "epoch": 3.8152502113611657,
-      "grad_norm": 0.0015594776486977935,
-      "learning_rate": 0.00019999282341559335,
-      "loss": 46.0,
-      "step": 23691
-    },
-    {
-      "epoch": 3.815411248439953,
-      "grad_norm": 0.0038541185203939676,
-      "learning_rate": 0.00019999282280948297,
-      "loss": 46.0,
-      "step": 23692
-    },
-    {
-      "epoch": 3.8155722855187406,
-      "grad_norm": 0.0035496940836310387,
-      "learning_rate": 0.000199992822203347,
-      "loss": 46.0,
-      "step": 23693
-    },
-    {
-      "epoch": 3.815733322597528,
-      "grad_norm": 0.0014210037188604474,
-      "learning_rate": 0.00019999282159718543,
-      "loss": 46.0,
-      "step": 23694
-    },
-    {
-      "epoch": 3.8158943596763155,
-      "grad_norm": 0.00115022377576679,
-      "learning_rate": 0.00019999282099099824,
-      "loss": 46.0,
-      "step": 23695
-    },
-    {
-      "epoch": 3.816055396755103,
-      "grad_norm": 0.0019606153946369886,
-      "learning_rate": 0.0001999928203847855,
-      "loss": 46.0,
-      "step": 23696
-    },
-    {
-      "epoch": 3.8162164338338904,
-      "grad_norm": 0.001526747248135507,
-      "learning_rate": 0.00019999281977854715,
-      "loss": 46.0,
-      "step": 23697
-    },
-    {
-      "epoch": 3.816377470912678,
-      "grad_norm": 0.003063860582187772,
-      "learning_rate": 0.00019999281917228317,
-      "loss": 46.0,
-      "step": 23698
-    },
-    {
-      "epoch": 3.816538507991465,
-      "grad_norm": 0.001384721021167934,
-      "learning_rate": 0.00019999281856599363,
-      "loss": 46.0,
-      "step": 23699
-    },
-    {
-      "epoch": 3.8166995450702523,
-      "grad_norm": 0.008011037483811378,
-      "learning_rate": 0.0001999928179596785,
-      "loss": 46.0,
-      "step": 23700
-    },
-    {
-      "epoch": 3.81686058214904,
-      "grad_norm": 0.0017826680559664965,
-      "learning_rate": 0.00019999281735333778,
-      "loss": 46.0,
-      "step": 23701
-    },
-    {
-      "epoch": 3.8170216192278272,
-      "grad_norm": 0.0010142020182684064,
-      "learning_rate": 0.00019999281674697142,
-      "loss": 46.0,
-      "step": 23702
-    },
-    {
-      "epoch": 3.8171826563066147,
-      "grad_norm": 0.0013554822653532028,
-      "learning_rate": 0.0001999928161405795,
-      "loss": 46.0,
-      "step": 23703
-    },
-    {
-      "epoch": 3.817343693385402,
-      "grad_norm": 0.001434633624739945,
-      "learning_rate": 0.00019999281553416203,
-      "loss": 46.0,
-      "step": 23704
-    },
-    {
-      "epoch": 3.817504730464189,
-      "grad_norm": 0.005213681608438492,
-      "learning_rate": 0.00019999281492771894,
-      "loss": 46.0,
-      "step": 23705
-    },
-    {
-      "epoch": 3.8176657675429766,
-      "grad_norm": 0.0018418292747810483,
-      "learning_rate": 0.00019999281432125023,
-      "loss": 46.0,
-      "step": 23706
-    },
-    {
-      "epoch": 3.817826804621764,
-      "grad_norm": 0.006393212825059891,
-      "learning_rate": 0.00019999281371475594,
-      "loss": 46.0,
-      "step": 23707
-    },
-    {
-      "epoch": 3.8179878417005515,
-      "grad_norm": 0.0016864193603396416,
-      "learning_rate": 0.0001999928131082361,
-      "loss": 46.0,
-      "step": 23708
-    },
-    {
-      "epoch": 3.818148878779339,
-      "grad_norm": 0.005730402190238237,
-      "learning_rate": 0.0001999928125016906,
-      "loss": 46.0,
-      "step": 23709
-    },
-    {
-      "epoch": 3.8183099158581264,
-      "grad_norm": 0.0013925766106694937,
-      "learning_rate": 0.00019999281189511954,
-      "loss": 46.0,
-      "step": 23710
-    },
-    {
-      "epoch": 3.818470952936914,
-      "grad_norm": 0.005298233591020107,
-      "learning_rate": 0.00019999281128852287,
-      "loss": 46.0,
-      "step": 23711
-    },
-    {
-      "epoch": 3.8186319900157013,
-      "grad_norm": 0.01740388572216034,
-      "learning_rate": 0.00019999281068190064,
-      "loss": 46.0,
-      "step": 23712
-    },
-    {
-      "epoch": 3.8187930270944888,
-      "grad_norm": 0.014286418445408344,
-      "learning_rate": 0.0001999928100752528,
-      "loss": 46.0,
-      "step": 23713
-    },
-    {
-      "epoch": 3.8189540641732758,
-      "grad_norm": 0.002689335960894823,
-      "learning_rate": 0.00019999280946857932,
-      "loss": 46.0,
-      "step": 23714
-    },
-    {
-      "epoch": 3.819115101252063,
-      "grad_norm": 0.0017703790217638016,
-      "learning_rate": 0.00019999280886188033,
-      "loss": 46.0,
-      "step": 23715
-    },
-    {
-      "epoch": 3.8192761383308507,
-      "grad_norm": 0.0038326107896864414,
-      "learning_rate": 0.0001999928082551557,
-      "loss": 46.0,
-      "step": 23716
-    },
-    {
-      "epoch": 3.819437175409638,
-      "grad_norm": 0.00924014300107956,
-      "learning_rate": 0.00019999280764840548,
-      "loss": 46.0,
-      "step": 23717
-    },
-    {
-      "epoch": 3.8195982124884256,
-      "grad_norm": 0.0068095168098807335,
-      "learning_rate": 0.00019999280704162967,
-      "loss": 46.0,
-      "step": 23718
-    },
-    {
-      "epoch": 3.8197592495672126,
-      "grad_norm": 0.004377112258225679,
-      "learning_rate": 0.00019999280643482825,
-      "loss": 46.0,
-      "step": 23719
-    },
-    {
-      "epoch": 3.819920286646,
-      "grad_norm": 0.004944744054228067,
-      "learning_rate": 0.00019999280582800127,
-      "loss": 46.0,
-      "step": 23720
-    },
-    {
-      "epoch": 3.8200813237247875,
-      "grad_norm": 0.0025594928301870823,
-      "learning_rate": 0.00019999280522114868,
-      "loss": 46.0,
-      "step": 23721
-    },
-    {
-      "epoch": 3.820242360803575,
-      "grad_norm": 0.0010990941664204001,
-      "learning_rate": 0.0001999928046142705,
-      "loss": 46.0,
-      "step": 23722
-    },
-    {
-      "epoch": 3.8204033978823624,
-      "grad_norm": 0.0021668653935194016,
-      "learning_rate": 0.00019999280400736672,
-      "loss": 46.0,
-      "step": 23723
-    },
-    {
-      "epoch": 3.82056443496115,
-      "grad_norm": 0.005260012578219175,
-      "learning_rate": 0.00019999280340043734,
-      "loss": 46.0,
-      "step": 23724
-    },
-    {
-      "epoch": 3.8207254720399373,
-      "grad_norm": 0.004265293013304472,
-      "learning_rate": 0.00019999280279348237,
-      "loss": 46.0,
-      "step": 23725
-    },
-    {
-      "epoch": 3.8208865091187247,
-      "grad_norm": 0.0022187649738043547,
-      "learning_rate": 0.00019999280218650184,
-      "loss": 46.0,
-      "step": 23726
-    },
-    {
-      "epoch": 3.821047546197512,
-      "grad_norm": 0.006004838272929192,
-      "learning_rate": 0.00019999280157949567,
-      "loss": 46.0,
-      "step": 23727
-    },
-    {
-      "epoch": 3.8212085832762996,
-      "grad_norm": 0.004536593332886696,
-      "learning_rate": 0.00019999280097246397,
-      "loss": 46.0,
-      "step": 23728
-    },
-    {
-      "epoch": 3.8213696203550866,
-      "grad_norm": 0.0023089649621397257,
-      "learning_rate": 0.00019999280036540662,
-      "loss": 46.0,
-      "step": 23729
-    },
-    {
-      "epoch": 3.821530657433874,
-      "grad_norm": 0.001987896393984556,
-      "learning_rate": 0.0001999927997583237,
-      "loss": 46.0,
-      "step": 23730
-    },
-    {
-      "epoch": 3.8216916945126616,
-      "grad_norm": 0.006554876919835806,
-      "learning_rate": 0.00019999279915121517,
-      "loss": 46.0,
-      "step": 23731
-    },
-    {
-      "epoch": 3.821852731591449,
-      "grad_norm": 0.009937213733792305,
-      "learning_rate": 0.00019999279854408103,
-      "loss": 46.0,
-      "step": 23732
-    },
-    {
-      "epoch": 3.8220137686702365,
-      "grad_norm": 0.0013865112559869885,
-      "learning_rate": 0.00019999279793692134,
-      "loss": 46.0,
-      "step": 23733
-    },
-    {
-      "epoch": 3.8221748057490235,
-      "grad_norm": 0.0072349668480455875,
-      "learning_rate": 0.00019999279732973605,
-      "loss": 46.0,
-      "step": 23734
-    },
-    {
-      "epoch": 3.822335842827811,
-      "grad_norm": 0.006639469880610704,
-      "learning_rate": 0.00019999279672252516,
-      "loss": 46.0,
-      "step": 23735
-    },
-    {
-      "epoch": 3.8224968799065984,
-      "grad_norm": 0.005938095971941948,
-      "learning_rate": 0.00019999279611528867,
-      "loss": 46.0,
-      "step": 23736
-    },
-    {
-      "epoch": 3.822657916985386,
-      "grad_norm": 0.002632892457768321,
-      "learning_rate": 0.0001999927955080266,
-      "loss": 46.0,
-      "step": 23737
-    },
-    {
-      "epoch": 3.8228189540641733,
-      "grad_norm": 0.004561766516417265,
-      "learning_rate": 0.00019999279490073892,
-      "loss": 46.0,
-      "step": 23738
-    },
-    {
-      "epoch": 3.8229799911429607,
-      "grad_norm": 0.0014203216414898634,
-      "learning_rate": 0.00019999279429342565,
-      "loss": 46.0,
-      "step": 23739
-    },
-    {
-      "epoch": 3.823141028221748,
-      "grad_norm": 0.003651438746601343,
-      "learning_rate": 0.00019999279368608682,
-      "loss": 46.0,
-      "step": 23740
-    },
-    {
-      "epoch": 3.8233020653005356,
-      "grad_norm": 0.003425365313887596,
-      "learning_rate": 0.00019999279307872237,
-      "loss": 46.0,
-      "step": 23741
-    },
-    {
-      "epoch": 3.823463102379323,
-      "grad_norm": 0.011294897645711899,
-      "learning_rate": 0.0001999927924713323,
-      "loss": 46.0,
-      "step": 23742
-    },
-    {
-      "epoch": 3.82362413945811,
-      "grad_norm": 0.013482105918228626,
-      "learning_rate": 0.0001999927918639167,
-      "loss": 46.0,
-      "step": 23743
-    },
-    {
-      "epoch": 3.8237851765368975,
-      "grad_norm": 0.0008809698047116399,
-      "learning_rate": 0.00019999279125647545,
-      "loss": 46.0,
-      "step": 23744
-    },
-    {
-      "epoch": 3.823946213615685,
-      "grad_norm": 0.004353642929345369,
-      "learning_rate": 0.00019999279064900863,
-      "loss": 46.0,
-      "step": 23745
-    },
-    {
-      "epoch": 3.8241072506944724,
-      "grad_norm": 0.00181178271304816,
-      "learning_rate": 0.00019999279004151622,
-      "loss": 46.0,
-      "step": 23746
-    },
-    {
-      "epoch": 3.82426828777326,
-      "grad_norm": 0.0008638782892376184,
-      "learning_rate": 0.00019999278943399823,
-      "loss": 46.0,
-      "step": 23747
-    },
-    {
-      "epoch": 3.8244293248520473,
-      "grad_norm": 0.0035788519307971,
-      "learning_rate": 0.00019999278882645461,
-      "loss": 46.0,
-      "step": 23748
-    },
-    {
-      "epoch": 3.8245903619308343,
-      "grad_norm": 0.0019932491704821587,
-      "learning_rate": 0.00019999278821888542,
-      "loss": 46.0,
-      "step": 23749
-    },
-    {
-      "epoch": 3.824751399009622,
-      "grad_norm": 0.005432513076812029,
-      "learning_rate": 0.00019999278761129063,
-      "loss": 46.0,
-      "step": 23750
-    },
-    {
-      "epoch": 3.8249124360884093,
-      "grad_norm": 0.00308479112572968,
-      "learning_rate": 0.00019999278700367026,
-      "loss": 46.0,
-      "step": 23751
-    },
-    {
-      "epoch": 3.8250734731671967,
-      "grad_norm": 0.0024276836775243282,
-      "learning_rate": 0.00019999278639602427,
-      "loss": 46.0,
-      "step": 23752
-    },
-    {
-      "epoch": 3.825234510245984,
-      "grad_norm": 0.0011070839827880263,
-      "learning_rate": 0.0001999927857883527,
-      "loss": 46.0,
-      "step": 23753
-    },
-    {
-      "epoch": 3.8253955473247716,
-      "grad_norm": 0.004445291124284267,
-      "learning_rate": 0.00019999278518065557,
-      "loss": 46.0,
-      "step": 23754
-    },
-    {
-      "epoch": 3.825556584403559,
-      "grad_norm": 0.001863056910224259,
-      "learning_rate": 0.0001999927845729328,
-      "loss": 46.0,
-      "step": 23755
-    },
-    {
-      "epoch": 3.8257176214823465,
-      "grad_norm": 0.001070081372745335,
-      "learning_rate": 0.00019999278396518446,
-      "loss": 46.0,
-      "step": 23756
-    },
-    {
-      "epoch": 3.825878658561134,
-      "grad_norm": 0.002958189230412245,
-      "learning_rate": 0.0001999927833574105,
-      "loss": 46.0,
-      "step": 23757
-    },
-    {
-      "epoch": 3.826039695639921,
-      "grad_norm": 0.0024542456958442926,
-      "learning_rate": 0.000199992782749611,
-      "loss": 46.0,
-      "step": 23758
-    },
-    {
-      "epoch": 3.8262007327187084,
-      "grad_norm": 0.0032551358453929424,
-      "learning_rate": 0.00019999278214178587,
-      "loss": 46.0,
-      "step": 23759
-    },
-    {
-      "epoch": 3.826361769797496,
-      "grad_norm": 0.008647998794913292,
-      "learning_rate": 0.00019999278153393513,
-      "loss": 46.0,
-      "step": 23760
-    },
-    {
-      "epoch": 3.8265228068762833,
-      "grad_norm": 0.00384552706964314,
-      "learning_rate": 0.00019999278092605884,
-      "loss": 46.0,
-      "step": 23761
-    },
-    {
-      "epoch": 3.8266838439550708,
-      "grad_norm": 0.0018242730293422937,
-      "learning_rate": 0.00019999278031815692,
-      "loss": 46.0,
-      "step": 23762
-    },
-    {
-      "epoch": 3.826844881033858,
-      "grad_norm": 0.0024351656902581453,
-      "learning_rate": 0.00019999277971022945,
-      "loss": 46.0,
-      "step": 23763
-    },
-    {
-      "epoch": 3.8270059181126452,
-      "grad_norm": 0.00894754845649004,
-      "learning_rate": 0.00019999277910227634,
-      "loss": 46.0,
-      "step": 23764
-    },
-    {
-      "epoch": 3.8271669551914327,
-      "grad_norm": 0.019980011507868767,
-      "learning_rate": 0.00019999277849429766,
-      "loss": 46.0,
-      "step": 23765
-    },
-    {
-      "epoch": 3.82732799227022,
-      "grad_norm": 0.0037471551913768053,
-      "learning_rate": 0.00019999277788629338,
-      "loss": 46.0,
-      "step": 23766
-    },
-    {
-      "epoch": 3.8274890293490076,
-      "grad_norm": 0.013219069689512253,
-      "learning_rate": 0.00019999277727826353,
-      "loss": 46.0,
-      "step": 23767
-    },
-    {
-      "epoch": 3.827650066427795,
-      "grad_norm": 0.0070031085051596165,
-      "learning_rate": 0.00019999277667020807,
-      "loss": 46.0,
-      "step": 23768
-    },
-    {
-      "epoch": 3.8278111035065825,
-      "grad_norm": 0.005571064539253712,
-      "learning_rate": 0.00019999277606212702,
-      "loss": 46.0,
-      "step": 23769
-    },
-    {
-      "epoch": 3.82797214058537,
-      "grad_norm": 0.0038876717444509268,
-      "learning_rate": 0.00019999277545402038,
-      "loss": 46.0,
-      "step": 23770
-    },
-    {
-      "epoch": 3.8281331776641574,
-      "grad_norm": 0.004201279021799564,
-      "learning_rate": 0.00019999277484588813,
-      "loss": 46.0,
-      "step": 23771
-    },
-    {
-      "epoch": 3.8282942147429444,
-      "grad_norm": 0.0024333479814231396,
-      "learning_rate": 0.0001999927742377303,
-      "loss": 46.0,
-      "step": 23772
-    },
-    {
-      "epoch": 3.828455251821732,
-      "grad_norm": 0.004609811119735241,
-      "learning_rate": 0.00019999277362954686,
-      "loss": 46.0,
-      "step": 23773
-    },
-    {
-      "epoch": 3.8286162889005193,
-      "grad_norm": 0.005001090466976166,
-      "learning_rate": 0.00019999277302133782,
-      "loss": 46.0,
-      "step": 23774
-    },
-    {
-      "epoch": 3.8287773259793068,
-      "grad_norm": 0.002505801385268569,
-      "learning_rate": 0.00019999277241310322,
-      "loss": 46.0,
-      "step": 23775
-    },
-    {
-      "epoch": 3.828938363058094,
-      "grad_norm": 0.014222148805856705,
-      "learning_rate": 0.000199992771804843,
-      "loss": 46.0,
-      "step": 23776
-    },
-    {
-      "epoch": 3.8290994001368817,
-      "grad_norm": 0.0076241218484938145,
-      "learning_rate": 0.00019999277119655724,
-      "loss": 46.0,
-      "step": 23777
-    },
-    {
-      "epoch": 3.8292604372156687,
-      "grad_norm": 0.003420913591980934,
-      "learning_rate": 0.00019999277058824585,
-      "loss": 46.0,
-      "step": 23778
-    },
-    {
-      "epoch": 3.829421474294456,
-      "grad_norm": 0.005837075877934694,
-      "learning_rate": 0.00019999276997990884,
-      "loss": 46.0,
-      "step": 23779
-    },
-    {
-      "epoch": 3.8295825113732436,
-      "grad_norm": 0.005029190797358751,
-      "learning_rate": 0.0001999927693715463,
-      "loss": 46.0,
-      "step": 23780
-    },
-    {
-      "epoch": 3.829743548452031,
-      "grad_norm": 0.0019204431446269155,
-      "learning_rate": 0.0001999927687631581,
-      "loss": 46.0,
-      "step": 23781
-    },
-    {
-      "epoch": 3.8299045855308185,
-      "grad_norm": 0.013321777805685997,
-      "learning_rate": 0.00019999276815474434,
-      "loss": 46.0,
-      "step": 23782
-    },
-    {
-      "epoch": 3.830065622609606,
-      "grad_norm": 0.0017787304241210222,
-      "learning_rate": 0.00019999276754630502,
-      "loss": 46.0,
-      "step": 23783
-    },
-    {
-      "epoch": 3.8302266596883934,
-      "grad_norm": 0.00221823388710618,
-      "learning_rate": 0.00019999276693784005,
-      "loss": 46.0,
-      "step": 23784
-    },
-    {
-      "epoch": 3.830387696767181,
-      "grad_norm": 0.010900537483394146,
-      "learning_rate": 0.0001999927663293495,
-      "loss": 46.0,
-      "step": 23785
-    },
-    {
-      "epoch": 3.8305487338459683,
-      "grad_norm": 0.004711878951638937,
-      "learning_rate": 0.00019999276572083336,
-      "loss": 46.0,
-      "step": 23786
-    },
-    {
-      "epoch": 3.8307097709247553,
-      "grad_norm": 0.010935375466942787,
-      "learning_rate": 0.00019999276511229163,
-      "loss": 46.0,
-      "step": 23787
-    },
-    {
-      "epoch": 3.8308708080035427,
-      "grad_norm": 0.001460511703044176,
-      "learning_rate": 0.00019999276450372432,
-      "loss": 46.0,
-      "step": 23788
-    },
-    {
-      "epoch": 3.83103184508233,
-      "grad_norm": 0.008396551944315434,
-      "learning_rate": 0.0001999927638951314,
-      "loss": 46.0,
-      "step": 23789
-    },
-    {
-      "epoch": 3.8311928821611176,
-      "grad_norm": 0.0016302214935421944,
-      "learning_rate": 0.0001999927632865129,
-      "loss": 46.0,
-      "step": 23790
-    },
-    {
-      "epoch": 3.831353919239905,
-      "grad_norm": 0.002367028733715415,
-      "learning_rate": 0.0001999927626778688,
-      "loss": 46.0,
-      "step": 23791
-    },
-    {
-      "epoch": 3.8315149563186925,
-      "grad_norm": 0.0015716521302238107,
-      "learning_rate": 0.0001999927620691991,
-      "loss": 46.0,
-      "step": 23792
-    },
-    {
-      "epoch": 3.8316759933974796,
-      "grad_norm": 0.005456264596432447,
-      "learning_rate": 0.0001999927614605038,
-      "loss": 46.0,
-      "step": 23793
-    },
-    {
-      "epoch": 3.831837030476267,
-      "grad_norm": 0.0031718614045530558,
-      "learning_rate": 0.00019999276085178294,
-      "loss": 46.0,
-      "step": 23794
-    },
-    {
-      "epoch": 3.8319980675550545,
-      "grad_norm": 0.002720343181863427,
-      "learning_rate": 0.00019999276024303646,
-      "loss": 46.0,
-      "step": 23795
-    },
-    {
-      "epoch": 3.832159104633842,
-      "grad_norm": 0.002677729818969965,
-      "learning_rate": 0.0001999927596342644,
-      "loss": 46.0,
-      "step": 23796
-    },
-    {
-      "epoch": 3.8323201417126294,
-      "grad_norm": 0.00726128788664937,
-      "learning_rate": 0.00019999275902546674,
-      "loss": 46.0,
-      "step": 23797
-    },
-    {
-      "epoch": 3.832481178791417,
-      "grad_norm": 0.008322179317474365,
-      "learning_rate": 0.0001999927584166435,
-      "loss": 46.0,
-      "step": 23798
-    },
-    {
-      "epoch": 3.8326422158702043,
-      "grad_norm": 0.002563060726970434,
-      "learning_rate": 0.00019999275780779464,
-      "loss": 46.0,
-      "step": 23799
-    },
-    {
-      "epoch": 3.8328032529489917,
-      "grad_norm": 0.0039304327219724655,
-      "learning_rate": 0.0001999927571989202,
-      "loss": 46.0,
-      "step": 23800
-    },
-    {
-      "epoch": 3.832964290027779,
-      "grad_norm": 0.007026908919215202,
-      "learning_rate": 0.00019999275659002017,
-      "loss": 46.0,
-      "step": 23801
-    },
-    {
-      "epoch": 3.833125327106566,
-      "grad_norm": 0.000881332263816148,
-      "learning_rate": 0.00019999275598109456,
-      "loss": 46.0,
-      "step": 23802
-    },
-    {
-      "epoch": 3.8332863641853536,
-      "grad_norm": 0.008376401849091053,
-      "learning_rate": 0.00019999275537214333,
-      "loss": 46.0,
-      "step": 23803
-    },
-    {
-      "epoch": 3.833447401264141,
-      "grad_norm": 0.0030209124088287354,
-      "learning_rate": 0.00019999275476316654,
-      "loss": 46.0,
-      "step": 23804
-    },
-    {
-      "epoch": 3.8336084383429285,
-      "grad_norm": 0.0019772592931985855,
-      "learning_rate": 0.00019999275415416413,
-      "loss": 46.0,
-      "step": 23805
-    },
-    {
-      "epoch": 3.833769475421716,
-      "grad_norm": 0.006213182583451271,
-      "learning_rate": 0.0001999927535451361,
-      "loss": 46.0,
-      "step": 23806
-    },
-    {
-      "epoch": 3.833930512500503,
-      "grad_norm": 0.005444422364234924,
-      "learning_rate": 0.00019999275293608256,
-      "loss": 46.0,
-      "step": 23807
-    },
-    {
-      "epoch": 3.8340915495792904,
-      "grad_norm": 0.0014355507446452975,
-      "learning_rate": 0.00019999275232700337,
-      "loss": 46.0,
-      "step": 23808
-    },
-    {
-      "epoch": 3.834252586658078,
-      "grad_norm": 0.0015205810777842999,
-      "learning_rate": 0.0001999927517178986,
-      "loss": 46.0,
-      "step": 23809
-    },
-    {
-      "epoch": 3.8344136237368653,
-      "grad_norm": 0.003284480655565858,
-      "learning_rate": 0.00019999275110876822,
-      "loss": 46.0,
-      "step": 23810
-    },
-    {
-      "epoch": 3.834574660815653,
-      "grad_norm": 0.0027947314083576202,
-      "learning_rate": 0.00019999275049961224,
-      "loss": 46.0,
-      "step": 23811
-    },
-    {
-      "epoch": 3.8347356978944402,
-      "grad_norm": 0.008528749458491802,
-      "learning_rate": 0.0001999927498904307,
-      "loss": 46.0,
-      "step": 23812
-    },
-    {
-      "epoch": 3.8348967349732277,
-      "grad_norm": 0.0013023784849792719,
-      "learning_rate": 0.00019999274928122357,
-      "loss": 46.0,
-      "step": 23813
-    },
-    {
-      "epoch": 3.835057772052015,
-      "grad_norm": 0.005034647881984711,
-      "learning_rate": 0.0001999927486719908,
-      "loss": 46.0,
-      "step": 23814
-    },
-    {
-      "epoch": 3.8352188091308026,
-      "grad_norm": 0.008998457342386246,
-      "learning_rate": 0.00019999274806273246,
-      "loss": 46.0,
-      "step": 23815
-    },
-    {
-      "epoch": 3.8353798462095896,
-      "grad_norm": 0.005973034538328648,
-      "learning_rate": 0.00019999274745344855,
-      "loss": 46.0,
-      "step": 23816
-    },
-    {
-      "epoch": 3.835540883288377,
-      "grad_norm": 0.00256621022708714,
-      "learning_rate": 0.00019999274684413901,
-      "loss": 46.0,
-      "step": 23817
-    },
-    {
-      "epoch": 3.8357019203671645,
-      "grad_norm": 0.007949058897793293,
-      "learning_rate": 0.0001999927462348039,
-      "loss": 46.0,
-      "step": 23818
-    },
-    {
-      "epoch": 3.835862957445952,
-      "grad_norm": 0.0015344663988798857,
-      "learning_rate": 0.00019999274562544322,
-      "loss": 46.0,
-      "step": 23819
-    },
-    {
-      "epoch": 3.8360239945247394,
-      "grad_norm": 0.0013722279109060764,
-      "learning_rate": 0.00019999274501605695,
-      "loss": 46.0,
-      "step": 23820
-    },
-    {
-      "epoch": 3.836185031603527,
-      "grad_norm": 0.0012689040740951896,
-      "learning_rate": 0.00019999274440664504,
-      "loss": 46.0,
-      "step": 23821
-    },
-    {
-      "epoch": 3.836346068682314,
-      "grad_norm": 0.015141929499804974,
-      "learning_rate": 0.00019999274379720755,
-      "loss": 46.0,
-      "step": 23822
-    },
-    {
-      "epoch": 3.8365071057611013,
-      "grad_norm": 0.011476409621536732,
-      "learning_rate": 0.00019999274318774446,
-      "loss": 46.0,
-      "step": 23823
-    },
-    {
-      "epoch": 3.8366681428398888,
-      "grad_norm": 0.0041524856351315975,
-      "learning_rate": 0.0001999927425782558,
-      "loss": 46.0,
-      "step": 23824
-    },
-    {
-      "epoch": 3.8368291799186762,
-      "grad_norm": 0.004125271923840046,
-      "learning_rate": 0.00019999274196874154,
-      "loss": 46.0,
-      "step": 23825
-    },
-    {
-      "epoch": 3.8369902169974637,
-      "grad_norm": 0.01737317629158497,
-      "learning_rate": 0.00019999274135920167,
-      "loss": 46.0,
-      "step": 23826
-    },
-    {
-      "epoch": 3.837151254076251,
-      "grad_norm": 0.006977675016969442,
-      "learning_rate": 0.00019999274074963624,
-      "loss": 46.0,
-      "step": 23827
-    },
-    {
-      "epoch": 3.8373122911550386,
-      "grad_norm": 0.007337931077927351,
-      "learning_rate": 0.0001999927401400452,
-      "loss": 46.0,
-      "step": 23828
-    },
-    {
-      "epoch": 3.837473328233826,
-      "grad_norm": 0.007270507980138063,
-      "learning_rate": 0.00019999273953042853,
-      "loss": 46.0,
-      "step": 23829
-    },
-    {
-      "epoch": 3.8376343653126135,
-      "grad_norm": 0.005673321429640055,
-      "learning_rate": 0.00019999273892078634,
-      "loss": 46.0,
-      "step": 23830
-    },
-    {
-      "epoch": 3.8377954023914005,
-      "grad_norm": 0.0019528295379132032,
-      "learning_rate": 0.0001999927383111185,
-      "loss": 46.0,
-      "step": 23831
-    },
-    {
-      "epoch": 3.837956439470188,
-      "grad_norm": 0.002949316054582596,
-      "learning_rate": 0.0001999927377014251,
-      "loss": 46.0,
-      "step": 23832
-    },
-    {
-      "epoch": 3.8381174765489754,
-      "grad_norm": 0.010876092128455639,
-      "learning_rate": 0.00019999273709170608,
-      "loss": 46.0,
-      "step": 23833
-    },
-    {
-      "epoch": 3.838278513627763,
-      "grad_norm": 0.001497737132012844,
-      "learning_rate": 0.00019999273648196148,
-      "loss": 46.0,
-      "step": 23834
-    },
-    {
-      "epoch": 3.8384395507065503,
-      "grad_norm": 0.003304415615275502,
-      "learning_rate": 0.0001999927358721913,
-      "loss": 46.0,
-      "step": 23835
-    },
-    {
-      "epoch": 3.8386005877853373,
-      "grad_norm": 0.0009059087606146932,
-      "learning_rate": 0.0001999927352623955,
-      "loss": 46.0,
-      "step": 23836
-    },
-    {
-      "epoch": 3.8387616248641248,
-      "grad_norm": 0.0013369123917073011,
-      "learning_rate": 0.00019999273465257412,
-      "loss": 46.0,
-      "step": 23837
-    },
-    {
-      "epoch": 3.838922661942912,
-      "grad_norm": 0.0016327179037034512,
-      "learning_rate": 0.00019999273404272715,
-      "loss": 46.0,
-      "step": 23838
-    },
-    {
-      "epoch": 3.8390836990216997,
-      "grad_norm": 0.004308358766138554,
-      "learning_rate": 0.0001999927334328546,
-      "loss": 46.0,
-      "step": 23839
-    },
-    {
-      "epoch": 3.839244736100487,
-      "grad_norm": 0.0034348871558904648,
-      "learning_rate": 0.00019999273282295644,
-      "loss": 46.0,
-      "step": 23840
-    },
-    {
-      "epoch": 3.8394057731792746,
-      "grad_norm": 0.0012503493344411254,
-      "learning_rate": 0.00019999273221303268,
-      "loss": 46.0,
-      "step": 23841
-    },
-    {
-      "epoch": 3.839566810258062,
-      "grad_norm": 0.002518642693758011,
-      "learning_rate": 0.00019999273160308334,
-      "loss": 46.0,
-      "step": 23842
-    },
-    {
-      "epoch": 3.8397278473368495,
-      "grad_norm": 0.0028336290270090103,
-      "learning_rate": 0.0001999927309931084,
-      "loss": 46.0,
-      "step": 23843
-    },
-    {
-      "epoch": 3.839888884415637,
-      "grad_norm": 0.0015037036500871181,
-      "learning_rate": 0.00019999273038310785,
-      "loss": 46.0,
-      "step": 23844
-    },
-    {
-      "epoch": 3.8400499214944244,
-      "grad_norm": 0.001164411660283804,
-      "learning_rate": 0.00019999272977308175,
-      "loss": 46.0,
-      "step": 23845
-    },
-    {
-      "epoch": 3.8402109585732114,
-      "grad_norm": 0.006247709970921278,
-      "learning_rate": 0.00019999272916303005,
-      "loss": 46.0,
-      "step": 23846
-    },
-    {
-      "epoch": 3.840371995651999,
-      "grad_norm": 0.0024582005571573973,
-      "learning_rate": 0.00019999272855295271,
-      "loss": 46.0,
-      "step": 23847
-    },
-    {
-      "epoch": 3.8405330327307863,
-      "grad_norm": 0.009390462189912796,
-      "learning_rate": 0.00019999272794284982,
-      "loss": 46.0,
-      "step": 23848
-    },
-    {
-      "epoch": 3.8406940698095737,
-      "grad_norm": 0.001749495742842555,
-      "learning_rate": 0.00019999272733272133,
-      "loss": 46.0,
-      "step": 23849
-    },
-    {
-      "epoch": 3.840855106888361,
-      "grad_norm": 0.0022811098024249077,
-      "learning_rate": 0.00019999272672256723,
-      "loss": 46.0,
-      "step": 23850
-    },
-    {
-      "epoch": 3.841016143967148,
-      "grad_norm": 0.001285584643483162,
-      "learning_rate": 0.00019999272611238755,
-      "loss": 46.0,
-      "step": 23851
-    },
-    {
-      "epoch": 3.8411771810459356,
-      "grad_norm": 0.0026719607412815094,
-      "learning_rate": 0.00019999272550218228,
-      "loss": 46.0,
-      "step": 23852
-    },
-    {
-      "epoch": 3.841338218124723,
-      "grad_norm": 0.004525855649262667,
-      "learning_rate": 0.00019999272489195142,
-      "loss": 46.0,
-      "step": 23853
-    },
-    {
-      "epoch": 3.8414992552035105,
-      "grad_norm": 0.0026460273656994104,
-      "learning_rate": 0.00019999272428169494,
-      "loss": 46.0,
-      "step": 23854
-    },
-    {
-      "epoch": 3.841660292282298,
-      "grad_norm": 0.007868615910410881,
-      "learning_rate": 0.0001999927236714129,
-      "loss": 46.0,
-      "step": 23855
-    },
-    {
-      "epoch": 3.8418213293610854,
-      "grad_norm": 0.002740424359217286,
-      "learning_rate": 0.00019999272306110526,
-      "loss": 46.0,
-      "step": 23856
-    },
-    {
-      "epoch": 3.841982366439873,
-      "grad_norm": 0.005905696656554937,
-      "learning_rate": 0.000199992722450772,
-      "loss": 46.0,
-      "step": 23857
-    },
-    {
-      "epoch": 3.8421434035186603,
-      "grad_norm": 0.008847422897815704,
-      "learning_rate": 0.00019999272184041317,
-      "loss": 46.0,
-      "step": 23858
-    },
-    {
-      "epoch": 3.842304440597448,
-      "grad_norm": 0.007343965116888285,
-      "learning_rate": 0.00019999272123002876,
-      "loss": 46.0,
-      "step": 23859
-    },
-    {
-      "epoch": 3.842465477676235,
-      "grad_norm": 0.006909176241606474,
-      "learning_rate": 0.00019999272061961874,
-      "loss": 46.0,
-      "step": 23860
-    },
-    {
-      "epoch": 3.8426265147550223,
-      "grad_norm": 0.003095058025792241,
-      "learning_rate": 0.00019999272000918313,
-      "loss": 46.0,
-      "step": 23861
-    },
-    {
-      "epoch": 3.8427875518338097,
-      "grad_norm": 0.0021113466937094927,
-      "learning_rate": 0.00019999271939872193,
-      "loss": 46.0,
-      "step": 23862
-    },
-    {
-      "epoch": 3.842948588912597,
-      "grad_norm": 0.0008507101447321475,
-      "learning_rate": 0.00019999271878823514,
-      "loss": 46.0,
-      "step": 23863
-    },
-    {
-      "epoch": 3.8431096259913846,
-      "grad_norm": 0.0067399670369923115,
-      "learning_rate": 0.00019999271817772274,
-      "loss": 46.0,
-      "step": 23864
-    },
-    {
-      "epoch": 3.843270663070172,
-      "grad_norm": 0.002260900568217039,
-      "learning_rate": 0.00019999271756718475,
-      "loss": 46.0,
-      "step": 23865
-    },
-    {
-      "epoch": 3.843431700148959,
-      "grad_norm": 0.008987069129943848,
-      "learning_rate": 0.00019999271695662118,
-      "loss": 46.0,
-      "step": 23866
-    },
-    {
-      "epoch": 3.8435927372277465,
-      "grad_norm": 0.010576714761555195,
-      "learning_rate": 0.00019999271634603202,
-      "loss": 46.0,
-      "step": 23867
-    },
-    {
-      "epoch": 3.843753774306534,
-      "grad_norm": 0.00532457185909152,
-      "learning_rate": 0.00019999271573541724,
-      "loss": 46.0,
-      "step": 23868
-    },
-    {
-      "epoch": 3.8439148113853214,
-      "grad_norm": 0.003059093374758959,
-      "learning_rate": 0.0001999927151247769,
-      "loss": 46.0,
-      "step": 23869
-    },
-    {
-      "epoch": 3.844075848464109,
-      "grad_norm": 0.021839655935764313,
-      "learning_rate": 0.00019999271451411096,
-      "loss": 46.0,
-      "step": 23870
-    },
-    {
-      "epoch": 3.8442368855428963,
-      "grad_norm": 0.01567326858639717,
-      "learning_rate": 0.0001999927139034194,
-      "loss": 46.0,
-      "step": 23871
-    },
-    {
-      "epoch": 3.844397922621684,
-      "grad_norm": 0.000972215726505965,
-      "learning_rate": 0.0001999927132927023,
-      "loss": 46.0,
-      "step": 23872
-    },
-    {
-      "epoch": 3.8445589597004712,
-      "grad_norm": 0.008044194430112839,
-      "learning_rate": 0.00019999271268195953,
-      "loss": 46.0,
-      "step": 23873
-    },
-    {
-      "epoch": 3.8447199967792587,
-      "grad_norm": 0.0038501210510730743,
-      "learning_rate": 0.0001999927120711912,
-      "loss": 46.0,
-      "step": 23874
-    },
-    {
-      "epoch": 3.8448810338580457,
-      "grad_norm": 0.014655952341854572,
-      "learning_rate": 0.00019999271146039732,
-      "loss": 46.0,
-      "step": 23875
-    },
-    {
-      "epoch": 3.845042070936833,
-      "grad_norm": 0.009296517819166183,
-      "learning_rate": 0.00019999271084957781,
-      "loss": 46.0,
-      "step": 23876
-    },
-    {
-      "epoch": 3.8452031080156206,
-      "grad_norm": 0.006652393378317356,
-      "learning_rate": 0.00019999271023873273,
-      "loss": 46.0,
-      "step": 23877
-    },
-    {
-      "epoch": 3.845364145094408,
-      "grad_norm": 0.005066597834229469,
-      "learning_rate": 0.00019999270962786203,
-      "loss": 46.0,
-      "step": 23878
-    },
-    {
-      "epoch": 3.8455251821731955,
-      "grad_norm": 0.00882706232368946,
-      "learning_rate": 0.0001999927090169657,
-      "loss": 46.0,
-      "step": 23879
-    },
-    {
-      "epoch": 3.8456862192519825,
-      "grad_norm": 0.012673420831561089,
-      "learning_rate": 0.00019999270840604386,
-      "loss": 46.0,
-      "step": 23880
-    },
-    {
-      "epoch": 3.84584725633077,
-      "grad_norm": 0.0029137085657566786,
-      "learning_rate": 0.00019999270779509637,
-      "loss": 46.0,
-      "step": 23881
-    },
-    {
-      "epoch": 3.8460082934095574,
-      "grad_norm": 0.010180748999118805,
-      "learning_rate": 0.00019999270718412332,
-      "loss": 46.0,
-      "step": 23882
-    },
-    {
-      "epoch": 3.846169330488345,
-      "grad_norm": 0.002074816729873419,
-      "learning_rate": 0.00019999270657312465,
-      "loss": 46.0,
-      "step": 23883
-    },
-    {
-      "epoch": 3.8463303675671323,
-      "grad_norm": 0.008947465568780899,
-      "learning_rate": 0.0001999927059621004,
-      "loss": 46.0,
-      "step": 23884
-    },
-    {
-      "epoch": 3.8464914046459198,
-      "grad_norm": 0.0037822150625288486,
-      "learning_rate": 0.00019999270535105054,
-      "loss": 46.0,
-      "step": 23885
-    },
-    {
-      "epoch": 3.846652441724707,
-      "grad_norm": 0.0004745628102682531,
-      "learning_rate": 0.00019999270473997514,
-      "loss": 46.0,
-      "step": 23886
-    },
-    {
-      "epoch": 3.8468134788034947,
-      "grad_norm": 0.002333694836124778,
-      "learning_rate": 0.0001999927041288741,
-      "loss": 46.0,
-      "step": 23887
-    },
-    {
-      "epoch": 3.846974515882282,
-      "grad_norm": 0.008906783536076546,
-      "learning_rate": 0.0001999927035177475,
-      "loss": 46.0,
-      "step": 23888
-    },
-    {
-      "epoch": 3.847135552961069,
-      "grad_norm": 0.0016794158145785332,
-      "learning_rate": 0.00019999270290659525,
-      "loss": 46.0,
-      "step": 23889
-    },
-    {
-      "epoch": 3.8472965900398566,
-      "grad_norm": 0.00228791288100183,
-      "learning_rate": 0.00019999270229541748,
-      "loss": 46.0,
-      "step": 23890
-    },
-    {
-      "epoch": 3.847457627118644,
-      "grad_norm": 0.005565068684518337,
-      "learning_rate": 0.00019999270168421403,
-      "loss": 46.0,
-      "step": 23891
-    },
-    {
-      "epoch": 3.8476186641974315,
-      "grad_norm": 0.008348161354660988,
-      "learning_rate": 0.00019999270107298506,
-      "loss": 46.0,
-      "step": 23892
-    },
-    {
-      "epoch": 3.847779701276219,
-      "grad_norm": 0.006248812191188335,
-      "learning_rate": 0.00019999270046173047,
-      "loss": 46.0,
-      "step": 23893
-    },
-    {
-      "epoch": 3.8479407383550064,
-      "grad_norm": 0.004908010829240084,
-      "learning_rate": 0.00019999269985045032,
-      "loss": 46.0,
-      "step": 23894
-    },
-    {
-      "epoch": 3.8481017754337934,
-      "grad_norm": 0.004707091022282839,
-      "learning_rate": 0.00019999269923914452,
-      "loss": 46.0,
-      "step": 23895
-    },
-    {
-      "epoch": 3.848262812512581,
-      "grad_norm": 0.015538440085947514,
-      "learning_rate": 0.00019999269862781314,
-      "loss": 46.0,
-      "step": 23896
-    },
-    {
-      "epoch": 3.8484238495913683,
-      "grad_norm": 0.005997444968670607,
-      "learning_rate": 0.0001999926980164562,
-      "loss": 46.0,
-      "step": 23897
-    },
-    {
-      "epoch": 3.8485848866701557,
-      "grad_norm": 0.004419758915901184,
-      "learning_rate": 0.00019999269740507365,
-      "loss": 46.0,
-      "step": 23898
-    },
-    {
-      "epoch": 3.848745923748943,
-      "grad_norm": 0.009846421889960766,
-      "learning_rate": 0.00019999269679366548,
-      "loss": 46.0,
-      "step": 23899
-    },
-    {
-      "epoch": 3.8489069608277307,
-      "grad_norm": 0.003134183818474412,
-      "learning_rate": 0.00019999269618223176,
-      "loss": 46.0,
-      "step": 23900
-    },
-    {
-      "epoch": 3.849067997906518,
-      "grad_norm": 0.0008235535933636129,
-      "learning_rate": 0.0001999926955707724,
-      "loss": 46.0,
-      "step": 23901
-    },
-    {
-      "epoch": 3.8492290349853056,
-      "grad_norm": 0.009524377062916756,
-      "learning_rate": 0.00019999269495928748,
-      "loss": 46.0,
-      "step": 23902
-    },
-    {
-      "epoch": 3.849390072064093,
-      "grad_norm": 0.005838321056216955,
-      "learning_rate": 0.000199992694347777,
-      "loss": 46.0,
-      "step": 23903
-    },
-    {
-      "epoch": 3.84955110914288,
-      "grad_norm": 0.002070268616080284,
-      "learning_rate": 0.0001999926937362409,
-      "loss": 46.0,
-      "step": 23904
-    },
-    {
-      "epoch": 3.8497121462216675,
-      "grad_norm": 0.007831402122974396,
-      "learning_rate": 0.00019999269312467917,
-      "loss": 46.0,
-      "step": 23905
-    },
-    {
-      "epoch": 3.849873183300455,
-      "grad_norm": 0.015754563733935356,
-      "learning_rate": 0.00019999269251309187,
-      "loss": 46.0,
-      "step": 23906
-    },
-    {
-      "epoch": 3.8500342203792424,
-      "grad_norm": 0.010111791081726551,
-      "learning_rate": 0.00019999269190147897,
-      "loss": 46.0,
-      "step": 23907
-    },
-    {
-      "epoch": 3.85019525745803,
-      "grad_norm": 0.001547613414004445,
-      "learning_rate": 0.0001999926912898405,
-      "loss": 46.0,
-      "step": 23908
-    },
-    {
-      "epoch": 3.850356294536817,
-      "grad_norm": 0.0014617778360843658,
-      "learning_rate": 0.00019999269067817643,
-      "loss": 46.0,
-      "step": 23909
-    },
-    {
-      "epoch": 3.8505173316156043,
-      "grad_norm": 0.0030381157994270325,
-      "learning_rate": 0.00019999269006648675,
-      "loss": 46.0,
-      "step": 23910
-    },
-    {
-      "epoch": 3.8506783686943917,
-      "grad_norm": 0.0017007588176056743,
-      "learning_rate": 0.0001999926894547715,
-      "loss": 46.0,
-      "step": 23911
-    },
-    {
-      "epoch": 3.850839405773179,
-      "grad_norm": 0.009313547052443027,
-      "learning_rate": 0.00019999268884303065,
-      "loss": 46.0,
-      "step": 23912
-    },
-    {
-      "epoch": 3.8510004428519666,
-      "grad_norm": 0.0019377215066924691,
-      "learning_rate": 0.0001999926882312642,
-      "loss": 46.0,
-      "step": 23913
-    },
-    {
-      "epoch": 3.851161479930754,
-      "grad_norm": 0.0011417100904509425,
-      "learning_rate": 0.00019999268761947215,
-      "loss": 46.0,
-      "step": 23914
-    },
-    {
-      "epoch": 3.8513225170095415,
-      "grad_norm": 0.0040351650677621365,
-      "learning_rate": 0.00019999268700765453,
-      "loss": 46.0,
-      "step": 23915
-    },
-    {
-      "epoch": 3.851483554088329,
-      "grad_norm": 0.00386438868008554,
-      "learning_rate": 0.0001999926863958113,
-      "loss": 46.0,
-      "step": 23916
-    },
-    {
-      "epoch": 3.8516445911671164,
-      "grad_norm": 0.002421856392174959,
-      "learning_rate": 0.00019999268578394248,
-      "loss": 46.0,
-      "step": 23917
-    },
-    {
-      "epoch": 3.851805628245904,
-      "grad_norm": 0.005682044662535191,
-      "learning_rate": 0.00019999268517204808,
-      "loss": 46.0,
-      "step": 23918
-    },
-    {
-      "epoch": 3.851966665324691,
-      "grad_norm": 0.0033794965129345655,
-      "learning_rate": 0.00019999268456012806,
-      "loss": 46.0,
-      "step": 23919
-    },
-    {
-      "epoch": 3.8521277024034783,
-      "grad_norm": 0.002941686660051346,
-      "learning_rate": 0.00019999268394818245,
-      "loss": 46.0,
-      "step": 23920
-    },
-    {
-      "epoch": 3.852288739482266,
-      "grad_norm": 0.0006447505438700318,
-      "learning_rate": 0.00019999268333621126,
-      "loss": 46.0,
-      "step": 23921
-    },
-    {
-      "epoch": 3.8524497765610533,
-      "grad_norm": 0.009037910960614681,
-      "learning_rate": 0.0001999926827242145,
-      "loss": 46.0,
-      "step": 23922
-    },
-    {
-      "epoch": 3.8526108136398407,
-      "grad_norm": 0.0015992186963558197,
-      "learning_rate": 0.0001999926821121921,
-      "loss": 46.0,
-      "step": 23923
-    },
-    {
-      "epoch": 3.8527718507186277,
-      "grad_norm": 0.002157653449103236,
-      "learning_rate": 0.00019999268150014415,
-      "loss": 46.0,
-      "step": 23924
-    },
-    {
-      "epoch": 3.852932887797415,
-      "grad_norm": 0.009380200877785683,
-      "learning_rate": 0.00019999268088807058,
-      "loss": 46.0,
-      "step": 23925
-    },
-    {
-      "epoch": 3.8530939248762026,
-      "grad_norm": 0.005483757704496384,
-      "learning_rate": 0.00019999268027597142,
-      "loss": 46.0,
-      "step": 23926
-    },
-    {
-      "epoch": 3.85325496195499,
-      "grad_norm": 0.0015798325184732676,
-      "learning_rate": 0.00019999267966384668,
-      "loss": 46.0,
-      "step": 23927
-    },
-    {
-      "epoch": 3.8534159990337775,
-      "grad_norm": 0.0031880135647952557,
-      "learning_rate": 0.00019999267905169635,
-      "loss": 46.0,
-      "step": 23928
-    },
-    {
-      "epoch": 3.853577036112565,
-      "grad_norm": 0.006490484811365604,
-      "learning_rate": 0.00019999267843952038,
-      "loss": 46.0,
-      "step": 23929
-    },
-    {
-      "epoch": 3.8537380731913524,
-      "grad_norm": 0.0019183688564226031,
-      "learning_rate": 0.00019999267782731884,
-      "loss": 46.0,
-      "step": 23930
-    },
-    {
-      "epoch": 3.85389911027014,
-      "grad_norm": 0.0011442864779382944,
-      "learning_rate": 0.00019999267721509175,
-      "loss": 46.0,
-      "step": 23931
-    },
-    {
-      "epoch": 3.8540601473489273,
-      "grad_norm": 0.0018853372894227505,
-      "learning_rate": 0.00019999267660283901,
-      "loss": 46.0,
-      "step": 23932
-    },
-    {
-      "epoch": 3.8542211844277143,
-      "grad_norm": 0.0011674242559820414,
-      "learning_rate": 0.00019999267599056072,
-      "loss": 46.0,
-      "step": 23933
-    },
-    {
-      "epoch": 3.854382221506502,
-      "grad_norm": 0.0023582978174090385,
-      "learning_rate": 0.00019999267537825684,
-      "loss": 46.0,
-      "step": 23934
-    },
-    {
-      "epoch": 3.8545432585852892,
-      "grad_norm": 0.0017070864560082555,
-      "learning_rate": 0.00019999267476592732,
-      "loss": 46.0,
-      "step": 23935
-    },
-    {
-      "epoch": 3.8547042956640767,
-      "grad_norm": 0.0015629647532477975,
-      "learning_rate": 0.00019999267415357226,
-      "loss": 46.0,
-      "step": 23936
-    },
-    {
-      "epoch": 3.854865332742864,
-      "grad_norm": 0.004717658273875713,
-      "learning_rate": 0.0001999926735411916,
-      "loss": 46.0,
-      "step": 23937
-    },
-    {
-      "epoch": 3.8550263698216516,
-      "grad_norm": 0.0037513435818254948,
-      "learning_rate": 0.0001999926729287853,
-      "loss": 46.0,
-      "step": 23938
-    },
-    {
-      "epoch": 3.8551874069004386,
-      "grad_norm": 0.004000397399067879,
-      "learning_rate": 0.00019999267231635343,
-      "loss": 46.0,
-      "step": 23939
-    },
-    {
-      "epoch": 3.855348443979226,
-      "grad_norm": 0.002302933717146516,
-      "learning_rate": 0.00019999267170389597,
-      "loss": 46.0,
-      "step": 23940
-    },
-    {
-      "epoch": 3.8555094810580135,
-      "grad_norm": 0.007076386362314224,
-      "learning_rate": 0.00019999267109141293,
-      "loss": 46.0,
-      "step": 23941
-    },
-    {
-      "epoch": 3.855670518136801,
-      "grad_norm": 0.0018582346383482218,
-      "learning_rate": 0.00019999267047890427,
-      "loss": 46.0,
-      "step": 23942
-    },
-    {
-      "epoch": 3.8558315552155884,
-      "grad_norm": 0.0020525604486465454,
-      "learning_rate": 0.00019999266986637002,
-      "loss": 46.0,
-      "step": 23943
-    },
-    {
-      "epoch": 3.855992592294376,
-      "grad_norm": 0.0007765043410472572,
-      "learning_rate": 0.00019999266925381019,
-      "loss": 46.0,
-      "step": 23944
-    },
-    {
-      "epoch": 3.8561536293731633,
-      "grad_norm": 0.010290330275893211,
-      "learning_rate": 0.0001999926686412248,
-      "loss": 46.0,
-      "step": 23945
-    },
-    {
-      "epoch": 3.8563146664519508,
-      "grad_norm": 0.0020926615688949823,
-      "learning_rate": 0.00019999266802861375,
-      "loss": 46.0,
-      "step": 23946
-    },
-    {
-      "epoch": 3.856475703530738,
-      "grad_norm": 0.0040846941992640495,
-      "learning_rate": 0.00019999266741597716,
-      "loss": 46.0,
-      "step": 23947
-    },
-    {
-      "epoch": 3.856636740609525,
-      "grad_norm": 0.002246662974357605,
-      "learning_rate": 0.00019999266680331498,
-      "loss": 46.0,
-      "step": 23948
-    },
-    {
-      "epoch": 3.8567977776883127,
-      "grad_norm": 0.014789251610636711,
-      "learning_rate": 0.00019999266619062715,
-      "loss": 46.0,
-      "step": 23949
-    },
-    {
-      "epoch": 3.8569588147671,
-      "grad_norm": 0.0029551174957305193,
-      "learning_rate": 0.00019999266557791377,
-      "loss": 46.0,
-      "step": 23950
-    },
-    {
-      "epoch": 3.8571198518458876,
-      "grad_norm": 0.0022438555024564266,
-      "learning_rate": 0.0001999926649651748,
-      "loss": 46.0,
-      "step": 23951
-    },
-    {
-      "epoch": 3.857280888924675,
-      "grad_norm": 0.0008192301611416042,
-      "learning_rate": 0.0001999926643524102,
-      "loss": 46.0,
-      "step": 23952
-    },
-    {
-      "epoch": 3.857441926003462,
-      "grad_norm": 0.0035310029052197933,
-      "learning_rate": 0.00019999266373962003,
-      "loss": 46.0,
-      "step": 23953
-    },
-    {
-      "epoch": 3.8576029630822495,
-      "grad_norm": 0.012899009510874748,
-      "learning_rate": 0.0001999926631268043,
-      "loss": 46.0,
-      "step": 23954
-    },
-    {
-      "epoch": 3.857764000161037,
-      "grad_norm": 0.0018951051170006394,
-      "learning_rate": 0.00019999266251396293,
-      "loss": 46.0,
-      "step": 23955
-    },
-    {
-      "epoch": 3.8579250372398244,
-      "grad_norm": 0.001851825974881649,
-      "learning_rate": 0.000199992661901096,
-      "loss": 46.0,
-      "step": 23956
-    },
-    {
-      "epoch": 3.858086074318612,
-      "grad_norm": 0.002390582114458084,
-      "learning_rate": 0.00019999266128820344,
-      "loss": 46.0,
-      "step": 23957
-    },
-    {
-      "epoch": 3.8582471113973993,
-      "grad_norm": 0.0011908072046935558,
-      "learning_rate": 0.00019999266067528533,
-      "loss": 46.0,
-      "step": 23958
-    },
-    {
-      "epoch": 3.8584081484761867,
-      "grad_norm": 0.0025828955695033073,
-      "learning_rate": 0.00019999266006234158,
-      "loss": 46.0,
-      "step": 23959
-    },
-    {
-      "epoch": 3.858569185554974,
-      "grad_norm": 0.0038658420089632273,
-      "learning_rate": 0.00019999265944937227,
-      "loss": 46.0,
-      "step": 23960
-    },
-    {
-      "epoch": 3.8587302226337616,
-      "grad_norm": 0.003765959059819579,
-      "learning_rate": 0.00019999265883637737,
-      "loss": 46.0,
-      "step": 23961
-    },
-    {
-      "epoch": 3.8588912597125486,
-      "grad_norm": 0.0012922107707709074,
-      "learning_rate": 0.00019999265822335686,
-      "loss": 46.0,
-      "step": 23962
-    },
-    {
-      "epoch": 3.859052296791336,
-      "grad_norm": 0.0026274758856743574,
-      "learning_rate": 0.00019999265761031076,
-      "loss": 46.0,
-      "step": 23963
-    },
-    {
-      "epoch": 3.8592133338701236,
-      "grad_norm": 0.00278154737316072,
-      "learning_rate": 0.00019999265699723908,
-      "loss": 46.0,
-      "step": 23964
-    },
-    {
-      "epoch": 3.859374370948911,
-      "grad_norm": 0.0021455432288348675,
-      "learning_rate": 0.0001999926563841418,
-      "loss": 46.0,
-      "step": 23965
-    },
-    {
-      "epoch": 3.8595354080276985,
-      "grad_norm": 0.0019393404945731163,
-      "learning_rate": 0.0001999926557710189,
-      "loss": 46.0,
-      "step": 23966
-    },
-    {
-      "epoch": 3.859696445106486,
-      "grad_norm": 0.002570522017776966,
-      "learning_rate": 0.00019999265515787044,
-      "loss": 46.0,
-      "step": 23967
-    },
-    {
-      "epoch": 3.859857482185273,
-      "grad_norm": 0.021145427599549294,
-      "learning_rate": 0.0001999926545446964,
-      "loss": 46.0,
-      "step": 23968
-    },
-    {
-      "epoch": 3.8600185192640604,
-      "grad_norm": 0.0017528659664094448,
-      "learning_rate": 0.00019999265393149673,
-      "loss": 46.0,
-      "step": 23969
-    },
-    {
-      "epoch": 3.860179556342848,
-      "grad_norm": 0.003110961988568306,
-      "learning_rate": 0.0001999926533182715,
-      "loss": 46.0,
-      "step": 23970
-    },
-    {
-      "epoch": 3.8603405934216353,
-      "grad_norm": 0.0026708340737968683,
-      "learning_rate": 0.00019999265270502067,
-      "loss": 46.0,
-      "step": 23971
-    },
-    {
-      "epoch": 3.8605016305004227,
-      "grad_norm": 0.015935273841023445,
-      "learning_rate": 0.0001999926520917442,
-      "loss": 46.0,
-      "step": 23972
-    },
-    {
-      "epoch": 3.86066266757921,
-      "grad_norm": 0.0038627285975962877,
-      "learning_rate": 0.00019999265147844218,
-      "loss": 46.0,
-      "step": 23973
-    },
-    {
-      "epoch": 3.8608237046579976,
-      "grad_norm": 0.004917750600725412,
-      "learning_rate": 0.00019999265086511456,
-      "loss": 46.0,
-      "step": 23974
-    },
-    {
-      "epoch": 3.860984741736785,
-      "grad_norm": 0.001065968768671155,
-      "learning_rate": 0.00019999265025176134,
-      "loss": 46.0,
-      "step": 23975
-    },
-    {
-      "epoch": 3.8611457788155725,
-      "grad_norm": 0.0022499931510537863,
-      "learning_rate": 0.00019999264963838255,
-      "loss": 46.0,
-      "step": 23976
-    },
-    {
-      "epoch": 3.8613068158943595,
-      "grad_norm": 0.0013169817393645644,
-      "learning_rate": 0.00019999264902497815,
-      "loss": 46.0,
-      "step": 23977
-    },
-    {
-      "epoch": 3.861467852973147,
-      "grad_norm": 0.007300612982362509,
-      "learning_rate": 0.00019999264841154813,
-      "loss": 46.0,
-      "step": 23978
-    },
-    {
-      "epoch": 3.8616288900519344,
-      "grad_norm": 0.007505542133003473,
-      "learning_rate": 0.00019999264779809256,
-      "loss": 46.0,
-      "step": 23979
-    },
-    {
-      "epoch": 3.861789927130722,
-      "grad_norm": 0.0012584192445501685,
-      "learning_rate": 0.0001999926471846114,
-      "loss": 46.0,
-      "step": 23980
-    },
-    {
-      "epoch": 3.8619509642095093,
-      "grad_norm": 0.008367362432181835,
-      "learning_rate": 0.00019999264657110462,
-      "loss": 46.0,
-      "step": 23981
-    },
-    {
-      "epoch": 3.862112001288297,
-      "grad_norm": 0.008066481910645962,
-      "learning_rate": 0.00019999264595757225,
-      "loss": 46.0,
-      "step": 23982
-    },
-    {
-      "epoch": 3.862273038367084,
-      "grad_norm": 0.006368101108819246,
-      "learning_rate": 0.0001999926453440143,
-      "loss": 46.0,
-      "step": 23983
-    },
-    {
-      "epoch": 3.8624340754458713,
-      "grad_norm": 0.015335018746554852,
-      "learning_rate": 0.00019999264473043074,
-      "loss": 46.0,
-      "step": 23984
-    },
-    {
-      "epoch": 3.8625951125246587,
-      "grad_norm": 0.007345108315348625,
-      "learning_rate": 0.0001999926441168216,
-      "loss": 46.0,
-      "step": 23985
-    },
-    {
-      "epoch": 3.862756149603446,
-      "grad_norm": 0.002686293562874198,
-      "learning_rate": 0.00019999264350318687,
-      "loss": 46.0,
-      "step": 23986
-    },
-    {
-      "epoch": 3.8629171866822336,
-      "grad_norm": 0.002146095037460327,
-      "learning_rate": 0.00019999264288952655,
-      "loss": 46.0,
-      "step": 23987
-    },
-    {
-      "epoch": 3.863078223761021,
-      "grad_norm": 0.004183642100542784,
-      "learning_rate": 0.00019999264227584063,
-      "loss": 46.0,
-      "step": 23988
-    },
-    {
-      "epoch": 3.8632392608398085,
-      "grad_norm": 0.0020193124655634165,
-      "learning_rate": 0.00019999264166212913,
-      "loss": 46.0,
-      "step": 23989
-    },
-    {
-      "epoch": 3.863400297918596,
-      "grad_norm": 0.00040674267802387476,
-      "learning_rate": 0.00019999264104839201,
-      "loss": 46.0,
-      "step": 23990
-    },
-    {
-      "epoch": 3.8635613349973834,
-      "grad_norm": 0.0014019823865965009,
-      "learning_rate": 0.00019999264043462928,
-      "loss": 46.0,
-      "step": 23991
-    },
-    {
-      "epoch": 3.8637223720761704,
-      "grad_norm": 0.014791633933782578,
-      "learning_rate": 0.00019999263982084102,
-      "loss": 46.0,
-      "step": 23992
-    },
-    {
-      "epoch": 3.863883409154958,
-      "grad_norm": 0.0013207363663241267,
-      "learning_rate": 0.00019999263920702714,
-      "loss": 46.0,
-      "step": 23993
-    },
-    {
-      "epoch": 3.8640444462337453,
-      "grad_norm": 0.004629611968994141,
-      "learning_rate": 0.00019999263859318765,
-      "loss": 46.0,
-      "step": 23994
-    },
-    {
-      "epoch": 3.8642054833125328,
-      "grad_norm": 0.0037192164454609156,
-      "learning_rate": 0.00019999263797932257,
-      "loss": 46.0,
-      "step": 23995
-    },
-    {
-      "epoch": 3.8643665203913202,
-      "grad_norm": 0.005566635634750128,
-      "learning_rate": 0.0001999926373654319,
-      "loss": 46.0,
-      "step": 23996
-    },
-    {
-      "epoch": 3.8645275574701072,
-      "grad_norm": 0.003593568690121174,
-      "learning_rate": 0.00019999263675151566,
-      "loss": 46.0,
-      "step": 23997
-    },
-    {
-      "epoch": 3.8646885945488947,
-      "grad_norm": 0.003664148971438408,
-      "learning_rate": 0.0001999926361375738,
-      "loss": 46.0,
-      "step": 23998
-    },
-    {
-      "epoch": 3.864849631627682,
-      "grad_norm": 0.0015974607085809112,
-      "learning_rate": 0.00019999263552360633,
-      "loss": 46.0,
-      "step": 23999
-    },
-    {
-      "epoch": 3.8650106687064696,
-      "grad_norm": 0.0035389717668294907,
-      "learning_rate": 0.00019999263490961332,
-      "loss": 46.0,
-      "step": 24000
-    },
-    {
-      "epoch": 3.865171705785257,
-      "grad_norm": 0.006094140466302633,
-      "learning_rate": 0.00019999263429559466,
-      "loss": 46.0,
-      "step": 24001
-    },
-    {
-      "epoch": 3.8653327428640445,
-      "grad_norm": 0.009808441624045372,
-      "learning_rate": 0.00019999263368155045,
-      "loss": 46.0,
-      "step": 24002
-    },
-    {
-      "epoch": 3.865493779942832,
-      "grad_norm": 0.0015333130722865462,
-      "learning_rate": 0.00019999263306748065,
-      "loss": 46.0,
-      "step": 24003
-    },
-    {
-      "epoch": 3.8656548170216194,
-      "grad_norm": 0.004158667754381895,
-      "learning_rate": 0.00019999263245338523,
-      "loss": 46.0,
-      "step": 24004
-    },
-    {
-      "epoch": 3.865815854100407,
-      "grad_norm": 0.0019012149423360825,
-      "learning_rate": 0.00019999263183926425,
-      "loss": 46.0,
-      "step": 24005
-    },
-    {
-      "epoch": 3.865976891179194,
-      "grad_norm": 0.004446346778422594,
-      "learning_rate": 0.00019999263122511763,
-      "loss": 46.0,
-      "step": 24006
-    },
-    {
-      "epoch": 3.8661379282579813,
-      "grad_norm": 0.006250792648643255,
-      "learning_rate": 0.00019999263061094545,
-      "loss": 46.0,
-      "step": 24007
-    },
-    {
-      "epoch": 3.8662989653367688,
-      "grad_norm": 0.012037362903356552,
-      "learning_rate": 0.00019999262999674766,
-      "loss": 46.0,
-      "step": 24008
-    },
-    {
-      "epoch": 3.866460002415556,
-      "grad_norm": 0.0010690775234252214,
-      "learning_rate": 0.00019999262938252428,
-      "loss": 46.0,
-      "step": 24009
-    },
-    {
-      "epoch": 3.8666210394943437,
-      "grad_norm": 0.01186061929911375,
-      "learning_rate": 0.0001999926287682753,
-      "loss": 46.0,
-      "step": 24010
-    },
-    {
-      "epoch": 3.866782076573131,
-      "grad_norm": 0.011861158534884453,
-      "learning_rate": 0.00019999262815400078,
-      "loss": 46.0,
-      "step": 24011
-    },
-    {
-      "epoch": 3.866943113651918,
-      "grad_norm": 0.0011435867054387927,
-      "learning_rate": 0.00019999262753970062,
-      "loss": 46.0,
-      "step": 24012
-    },
-    {
-      "epoch": 3.8671041507307056,
-      "grad_norm": 0.00245189992710948,
-      "learning_rate": 0.00019999262692537486,
-      "loss": 46.0,
-      "step": 24013
-    },
-    {
-      "epoch": 3.867265187809493,
-      "grad_norm": 0.001089563244022429,
-      "learning_rate": 0.00019999262631102354,
-      "loss": 46.0,
-      "step": 24014
-    },
-    {
-      "epoch": 3.8674262248882805,
-      "grad_norm": 0.0014150417409837246,
-      "learning_rate": 0.00019999262569664659,
-      "loss": 46.0,
-      "step": 24015
-    },
-    {
-      "epoch": 3.867587261967068,
-      "grad_norm": 0.008513418957591057,
-      "learning_rate": 0.00019999262508224407,
-      "loss": 46.0,
-      "step": 24016
-    },
-    {
-      "epoch": 3.8677482990458554,
-      "grad_norm": 0.0006640064530074596,
-      "learning_rate": 0.00019999262446781594,
-      "loss": 46.0,
-      "step": 24017
-    },
-    {
-      "epoch": 3.867909336124643,
-      "grad_norm": 0.0021558396983891726,
-      "learning_rate": 0.00019999262385336224,
-      "loss": 46.0,
-      "step": 24018
-    },
-    {
-      "epoch": 3.8680703732034303,
-      "grad_norm": 0.006785494741052389,
-      "learning_rate": 0.00019999262323888294,
-      "loss": 46.0,
-      "step": 24019
-    },
-    {
-      "epoch": 3.8682314102822177,
-      "grad_norm": 0.001933910883963108,
-      "learning_rate": 0.00019999262262437805,
-      "loss": 46.0,
-      "step": 24020
-    },
-    {
-      "epoch": 3.8683924473610047,
-      "grad_norm": 0.005459047853946686,
-      "learning_rate": 0.00019999262200984754,
-      "loss": 46.0,
-      "step": 24021
-    },
-    {
-      "epoch": 3.868553484439792,
-      "grad_norm": 0.0012136929435655475,
-      "learning_rate": 0.00019999262139529144,
-      "loss": 46.0,
-      "step": 24022
-    },
-    {
-      "epoch": 3.8687145215185796,
-      "grad_norm": 0.002100676530972123,
-      "learning_rate": 0.00019999262078070976,
-      "loss": 46.0,
-      "step": 24023
-    },
-    {
-      "epoch": 3.868875558597367,
-      "grad_norm": 0.001325149554759264,
-      "learning_rate": 0.00019999262016610252,
-      "loss": 46.0,
-      "step": 24024
-    },
-    {
-      "epoch": 3.8690365956761545,
-      "grad_norm": 0.001594682689756155,
-      "learning_rate": 0.00019999261955146964,
-      "loss": 46.0,
-      "step": 24025
-    },
-    {
-      "epoch": 3.8691976327549416,
-      "grad_norm": 0.0010829733218997717,
-      "learning_rate": 0.0001999926189368112,
-      "loss": 46.0,
-      "step": 24026
-    },
-    {
-      "epoch": 3.869358669833729,
-      "grad_norm": 0.010376001708209515,
-      "learning_rate": 0.00019999261832212713,
-      "loss": 46.0,
-      "step": 24027
-    },
-    {
-      "epoch": 3.8695197069125165,
-      "grad_norm": 0.0011982464930042624,
-      "learning_rate": 0.00019999261770741752,
-      "loss": 46.0,
-      "step": 24028
-    },
-    {
-      "epoch": 3.869680743991304,
-      "grad_norm": 0.005371060688048601,
-      "learning_rate": 0.00019999261709268228,
-      "loss": 46.0,
-      "step": 24029
-    },
-    {
-      "epoch": 3.8698417810700914,
-      "grad_norm": 0.006046678870916367,
-      "learning_rate": 0.00019999261647792144,
-      "loss": 46.0,
-      "step": 24030
-    },
-    {
-      "epoch": 3.870002818148879,
-      "grad_norm": 0.0029259948059916496,
-      "learning_rate": 0.00019999261586313503,
-      "loss": 46.0,
-      "step": 24031
-    },
-    {
-      "epoch": 3.8701638552276663,
-      "grad_norm": 0.001116109313443303,
-      "learning_rate": 0.000199992615248323,
-      "loss": 46.0,
-      "step": 24032
-    },
-    {
-      "epoch": 3.8703248923064537,
-      "grad_norm": 0.0017653099494054914,
-      "learning_rate": 0.0001999926146334854,
-      "loss": 46.0,
-      "step": 24033
-    },
-    {
-      "epoch": 3.870485929385241,
-      "grad_norm": 0.004673819988965988,
-      "learning_rate": 0.0001999926140186222,
-      "loss": 46.0,
-      "step": 24034
-    },
-    {
-      "epoch": 3.8706469664640286,
-      "grad_norm": 0.003567418549209833,
-      "learning_rate": 0.0001999926134037334,
-      "loss": 46.0,
-      "step": 24035
-    },
-    {
-      "epoch": 3.8708080035428156,
-      "grad_norm": 0.0019304970046505332,
-      "learning_rate": 0.00019999261278881903,
-      "loss": 46.0,
-      "step": 24036
-    },
-    {
-      "epoch": 3.870969040621603,
-      "grad_norm": 0.0025090614799410105,
-      "learning_rate": 0.00019999261217387905,
-      "loss": 46.0,
-      "step": 24037
-    },
-    {
-      "epoch": 3.8711300777003905,
-      "grad_norm": 0.002176633570343256,
-      "learning_rate": 0.0001999926115589135,
-      "loss": 46.0,
-      "step": 24038
-    },
-    {
-      "epoch": 3.871291114779178,
-      "grad_norm": 0.002901740139350295,
-      "learning_rate": 0.00019999261094392232,
-      "loss": 46.0,
-      "step": 24039
-    },
-    {
-      "epoch": 3.8714521518579654,
-      "grad_norm": 0.009482808411121368,
-      "learning_rate": 0.00019999261032890557,
-      "loss": 46.0,
-      "step": 24040
-    },
-    {
-      "epoch": 3.8716131889367524,
-      "grad_norm": 0.0035449410788714886,
-      "learning_rate": 0.0001999926097138632,
-      "loss": 46.0,
-      "step": 24041
-    },
-    {
-      "epoch": 3.87177422601554,
-      "grad_norm": 0.004730087239295244,
-      "learning_rate": 0.00019999260909879526,
-      "loss": 46.0,
-      "step": 24042
-    },
-    {
-      "epoch": 3.8719352630943273,
-      "grad_norm": 0.0021472552325576544,
-      "learning_rate": 0.00019999260848370173,
-      "loss": 46.0,
-      "step": 24043
-    },
-    {
-      "epoch": 3.872096300173115,
-      "grad_norm": 0.010897508822381496,
-      "learning_rate": 0.00019999260786858264,
-      "loss": 46.0,
-      "step": 24044
-    },
-    {
-      "epoch": 3.8722573372519022,
-      "grad_norm": 0.01087038591504097,
-      "learning_rate": 0.0001999926072534379,
-      "loss": 46.0,
-      "step": 24045
-    },
-    {
-      "epoch": 3.8724183743306897,
-      "grad_norm": 0.00884197372943163,
-      "learning_rate": 0.00019999260663826758,
-      "loss": 46.0,
-      "step": 24046
-    },
-    {
-      "epoch": 3.872579411409477,
-      "grad_norm": 0.009836284443736076,
-      "learning_rate": 0.00019999260602307167,
-      "loss": 46.0,
-      "step": 24047
-    },
-    {
-      "epoch": 3.8727404484882646,
-      "grad_norm": 0.002207430312409997,
-      "learning_rate": 0.00019999260540785017,
-      "loss": 46.0,
-      "step": 24048
-    },
-    {
-      "epoch": 3.872901485567052,
-      "grad_norm": 0.004285207949578762,
-      "learning_rate": 0.00019999260479260306,
-      "loss": 46.0,
-      "step": 24049
-    },
-    {
-      "epoch": 3.873062522645839,
-      "grad_norm": 0.0009452085359953344,
-      "learning_rate": 0.0001999926041773304,
-      "loss": 46.0,
-      "step": 24050
-    },
-    {
-      "epoch": 3.8732235597246265,
-      "grad_norm": 0.003883578348904848,
-      "learning_rate": 0.0001999926035620321,
-      "loss": 46.0,
-      "step": 24051
-    },
-    {
-      "epoch": 3.873384596803414,
-      "grad_norm": 0.004387306049466133,
-      "learning_rate": 0.00019999260294670826,
-      "loss": 46.0,
-      "step": 24052
-    },
-    {
-      "epoch": 3.8735456338822014,
-      "grad_norm": 0.004749350715428591,
-      "learning_rate": 0.00019999260233135877,
-      "loss": 46.0,
-      "step": 24053
-    },
-    {
-      "epoch": 3.873706670960989,
-      "grad_norm": 0.017119865864515305,
-      "learning_rate": 0.0001999926017159837,
-      "loss": 46.0,
-      "step": 24054
-    },
-    {
-      "epoch": 3.8738677080397763,
-      "grad_norm": 0.01100144349038601,
-      "learning_rate": 0.00019999260110058306,
-      "loss": 46.0,
-      "step": 24055
-    },
-    {
-      "epoch": 3.8740287451185633,
-      "grad_norm": 0.007789523806422949,
-      "learning_rate": 0.0001999926004851568,
-      "loss": 46.0,
-      "step": 24056
-    },
-    {
-      "epoch": 3.8741897821973508,
-      "grad_norm": 0.0015500987647101283,
-      "learning_rate": 0.00019999259986970498,
-      "loss": 46.0,
-      "step": 24057
-    },
-    {
-      "epoch": 3.8743508192761382,
-      "grad_norm": 0.0019097868353128433,
-      "learning_rate": 0.00019999259925422756,
-      "loss": 46.0,
-      "step": 24058
-    },
-    {
-      "epoch": 3.8745118563549257,
-      "grad_norm": 0.004824074916541576,
-      "learning_rate": 0.00019999259863872452,
-      "loss": 46.0,
-      "step": 24059
-    },
-    {
-      "epoch": 3.874672893433713,
-      "grad_norm": 0.0035186184104532003,
-      "learning_rate": 0.0001999925980231959,
-      "loss": 46.0,
-      "step": 24060
-    },
-    {
-      "epoch": 3.8748339305125006,
-      "grad_norm": 0.005075378343462944,
-      "learning_rate": 0.0001999925974076417,
-      "loss": 46.0,
-      "step": 24061
-    },
-    {
-      "epoch": 3.874994967591288,
-      "grad_norm": 0.0016036900924518704,
-      "learning_rate": 0.00019999259679206191,
-      "loss": 46.0,
-      "step": 24062
-    },
-    {
-      "epoch": 3.8751560046700755,
-      "grad_norm": 0.009347415529191494,
-      "learning_rate": 0.0001999925961764565,
-      "loss": 46.0,
-      "step": 24063
-    },
-    {
-      "epoch": 3.875317041748863,
-      "grad_norm": 0.0012929226504638791,
-      "learning_rate": 0.0001999925955608255,
-      "loss": 46.0,
-      "step": 24064
-    },
-    {
-      "epoch": 3.87547807882765,
-      "grad_norm": 0.002047156449407339,
-      "learning_rate": 0.00019999259494516894,
-      "loss": 46.0,
-      "step": 24065
-    },
-    {
-      "epoch": 3.8756391159064374,
-      "grad_norm": 0.012128115631639957,
-      "learning_rate": 0.00019999259432948677,
-      "loss": 46.0,
-      "step": 24066
-    },
-    {
-      "epoch": 3.875800152985225,
-      "grad_norm": 0.0029642682056874037,
-      "learning_rate": 0.00019999259371377898,
-      "loss": 46.0,
-      "step": 24067
-    },
-    {
-      "epoch": 3.8759611900640123,
-      "grad_norm": 0.0023666848428547382,
-      "learning_rate": 0.00019999259309804563,
-      "loss": 46.0,
-      "step": 24068
-    },
-    {
-      "epoch": 3.8761222271427997,
-      "grad_norm": 0.008460167795419693,
-      "learning_rate": 0.00019999259248228667,
-      "loss": 46.0,
-      "step": 24069
-    },
-    {
-      "epoch": 3.8762832642215868,
-      "grad_norm": 0.004398632328957319,
-      "learning_rate": 0.00019999259186650212,
-      "loss": 46.0,
-      "step": 24070
-    },
-    {
-      "epoch": 3.876444301300374,
-      "grad_norm": 0.0014474081108346581,
-      "learning_rate": 0.000199992591250692,
-      "loss": 46.0,
-      "step": 24071
-    },
-    {
-      "epoch": 3.8766053383791617,
-      "grad_norm": 0.0017836528131738305,
-      "learning_rate": 0.00019999259063485628,
-      "loss": 46.0,
-      "step": 24072
-    },
-    {
-      "epoch": 3.876766375457949,
-      "grad_norm": 0.008612348698079586,
-      "learning_rate": 0.00019999259001899494,
-      "loss": 46.0,
-      "step": 24073
-    },
-    {
-      "epoch": 3.8769274125367366,
-      "grad_norm": 0.00445523951202631,
-      "learning_rate": 0.00019999258940310802,
-      "loss": 46.0,
-      "step": 24074
-    },
-    {
-      "epoch": 3.877088449615524,
-      "grad_norm": 0.0028079834301024675,
-      "learning_rate": 0.0001999925887871955,
-      "loss": 46.0,
-      "step": 24075
-    },
-    {
-      "epoch": 3.8772494866943115,
-      "grad_norm": 0.001897752983495593,
-      "learning_rate": 0.0001999925881712574,
-      "loss": 46.0,
-      "step": 24076
-    },
-    {
-      "epoch": 3.877410523773099,
-      "grad_norm": 0.0069461390376091,
-      "learning_rate": 0.00019999258755529372,
-      "loss": 46.0,
-      "step": 24077
-    },
-    {
-      "epoch": 3.8775715608518864,
-      "grad_norm": 0.0010076443431898952,
-      "learning_rate": 0.00019999258693930442,
-      "loss": 46.0,
-      "step": 24078
-    },
-    {
-      "epoch": 3.8777325979306734,
-      "grad_norm": 0.0057729510590434074,
-      "learning_rate": 0.00019999258632328953,
-      "loss": 46.0,
-      "step": 24079
-    },
-    {
-      "epoch": 3.877893635009461,
-      "grad_norm": 0.00535427313297987,
-      "learning_rate": 0.00019999258570724905,
-      "loss": 46.0,
-      "step": 24080
-    },
-    {
-      "epoch": 3.8780546720882483,
-      "grad_norm": 0.010499185882508755,
-      "learning_rate": 0.000199992585091183,
-      "loss": 46.0,
-      "step": 24081
-    },
-    {
-      "epoch": 3.8782157091670357,
-      "grad_norm": 0.012383539229631424,
-      "learning_rate": 0.00019999258447509134,
-      "loss": 46.0,
-      "step": 24082
-    },
-    {
-      "epoch": 3.878376746245823,
-      "grad_norm": 0.0036151634994894266,
-      "learning_rate": 0.00019999258385897407,
-      "loss": 46.0,
-      "step": 24083
-    },
-    {
-      "epoch": 3.8785377833246106,
-      "grad_norm": 0.0011932895286008716,
-      "learning_rate": 0.00019999258324283122,
-      "loss": 46.0,
-      "step": 24084
-    },
-    {
-      "epoch": 3.8786988204033976,
-      "grad_norm": 0.0017412648303434253,
-      "learning_rate": 0.00019999258262666278,
-      "loss": 46.0,
-      "step": 24085
-    },
-    {
-      "epoch": 3.878859857482185,
-      "grad_norm": 0.0036719886120408773,
-      "learning_rate": 0.00019999258201046876,
-      "loss": 46.0,
-      "step": 24086
-    },
-    {
-      "epoch": 3.8790208945609725,
-      "grad_norm": 0.004792661406099796,
-      "learning_rate": 0.00019999258139424912,
-      "loss": 46.0,
-      "step": 24087
-    },
-    {
-      "epoch": 3.87918193163976,
-      "grad_norm": 0.003962106071412563,
-      "learning_rate": 0.00019999258077800391,
-      "loss": 46.0,
-      "step": 24088
-    },
-    {
-      "epoch": 3.8793429687185474,
-      "grad_norm": 0.004261672031134367,
-      "learning_rate": 0.0001999925801617331,
-      "loss": 46.0,
-      "step": 24089
-    },
-    {
-      "epoch": 3.879504005797335,
-      "grad_norm": 0.01727946475148201,
-      "learning_rate": 0.0001999925795454367,
-      "loss": 46.0,
-      "step": 24090
-    },
-    {
-      "epoch": 3.8796650428761223,
-      "grad_norm": 0.0010890386765822768,
-      "learning_rate": 0.00019999257892911468,
-      "loss": 46.0,
-      "step": 24091
-    },
-    {
-      "epoch": 3.87982607995491,
-      "grad_norm": 0.00723354471847415,
-      "learning_rate": 0.0001999925783127671,
-      "loss": 46.0,
-      "step": 24092
-    },
-    {
-      "epoch": 3.8799871170336973,
-      "grad_norm": 0.004591763485223055,
-      "learning_rate": 0.00019999257769639391,
-      "loss": 46.0,
-      "step": 24093
-    },
-    {
-      "epoch": 3.8801481541124843,
-      "grad_norm": 0.0029744494240731,
-      "learning_rate": 0.00019999257707999514,
-      "loss": 46.0,
-      "step": 24094
-    },
-    {
-      "epoch": 3.8803091911912717,
-      "grad_norm": 0.008001652546226978,
-      "learning_rate": 0.00019999257646357077,
-      "loss": 46.0,
-      "step": 24095
-    },
-    {
-      "epoch": 3.880470228270059,
-      "grad_norm": 0.005045961122959852,
-      "learning_rate": 0.00019999257584712082,
-      "loss": 46.0,
-      "step": 24096
-    },
-    {
-      "epoch": 3.8806312653488466,
-      "grad_norm": 0.0023090653121471405,
-      "learning_rate": 0.00019999257523064525,
-      "loss": 46.0,
-      "step": 24097
-    },
-    {
-      "epoch": 3.880792302427634,
-      "grad_norm": 0.002473338507115841,
-      "learning_rate": 0.0001999925746141441,
-      "loss": 46.0,
-      "step": 24098
-    },
-    {
-      "epoch": 3.8809533395064215,
-      "grad_norm": 0.002843037713319063,
-      "learning_rate": 0.00019999257399761736,
-      "loss": 46.0,
-      "step": 24099
-    },
-    {
-      "epoch": 3.8811143765852085,
-      "grad_norm": 0.003936846740543842,
-      "learning_rate": 0.00019999257338106503,
-      "loss": 46.0,
-      "step": 24100
-    },
-    {
-      "epoch": 3.881275413663996,
-      "grad_norm": 0.0013656708179041743,
-      "learning_rate": 0.0001999925727644871,
-      "loss": 46.0,
-      "step": 24101
-    },
-    {
-      "epoch": 3.8814364507427834,
-      "grad_norm": 0.0062868306413292885,
-      "learning_rate": 0.00019999257214788356,
-      "loss": 46.0,
-      "step": 24102
-    },
-    {
-      "epoch": 3.881597487821571,
-      "grad_norm": 0.0016342218732461333,
-      "learning_rate": 0.00019999257153125447,
-      "loss": 46.0,
-      "step": 24103
-    },
-    {
-      "epoch": 3.8817585249003583,
-      "grad_norm": 0.004117118194699287,
-      "learning_rate": 0.00019999257091459974,
-      "loss": 46.0,
-      "step": 24104
-    },
-    {
-      "epoch": 3.881919561979146,
-      "grad_norm": 0.002345234854146838,
-      "learning_rate": 0.00019999257029791948,
-      "loss": 46.0,
-      "step": 24105
-    },
-    {
-      "epoch": 3.8820805990579332,
-      "grad_norm": 0.008430286310613155,
-      "learning_rate": 0.00019999256968121357,
-      "loss": 46.0,
-      "step": 24106
-    },
-    {
-      "epoch": 3.8822416361367207,
-      "grad_norm": 0.005393289960920811,
-      "learning_rate": 0.00019999256906448208,
-      "loss": 46.0,
-      "step": 24107
-    },
-    {
-      "epoch": 3.882402673215508,
-      "grad_norm": 0.0008756909519433975,
-      "learning_rate": 0.00019999256844772497,
-      "loss": 46.0,
-      "step": 24108
-    },
-    {
-      "epoch": 3.882563710294295,
-      "grad_norm": 0.010313312523066998,
-      "learning_rate": 0.0001999925678309423,
-      "loss": 46.0,
-      "step": 24109
-    },
-    {
-      "epoch": 3.8827247473730826,
-      "grad_norm": 0.0035547094885259867,
-      "learning_rate": 0.00019999256721413405,
-      "loss": 46.0,
-      "step": 24110
-    },
-    {
-      "epoch": 3.88288578445187,
-      "grad_norm": 0.004055333323776722,
-      "learning_rate": 0.00019999256659730018,
-      "loss": 46.0,
-      "step": 24111
-    },
-    {
-      "epoch": 3.8830468215306575,
-      "grad_norm": 0.0021817300003021955,
-      "learning_rate": 0.00019999256598044073,
-      "loss": 46.0,
-      "step": 24112
-    },
-    {
-      "epoch": 3.883207858609445,
-      "grad_norm": 0.006538146175444126,
-      "learning_rate": 0.00019999256536355571,
-      "loss": 46.0,
-      "step": 24113
-    },
-    {
-      "epoch": 3.883368895688232,
-      "grad_norm": 0.0006481777527369559,
-      "learning_rate": 0.00019999256474664508,
-      "loss": 46.0,
-      "step": 24114
-    },
-    {
-      "epoch": 3.8835299327670194,
-      "grad_norm": 0.0010808173101395369,
-      "learning_rate": 0.00019999256412970887,
-      "loss": 46.0,
-      "step": 24115
-    },
-    {
-      "epoch": 3.883690969845807,
-      "grad_norm": 0.007094038650393486,
-      "learning_rate": 0.000199992563512747,
-      "loss": 46.0,
-      "step": 24116
-    },
-    {
-      "epoch": 3.8838520069245943,
-      "grad_norm": 0.007232999429106712,
-      "learning_rate": 0.00019999256289575962,
-      "loss": 46.0,
-      "step": 24117
-    },
-    {
-      "epoch": 3.8840130440033818,
-      "grad_norm": 0.0025392400566488504,
-      "learning_rate": 0.00019999256227874659,
-      "loss": 46.0,
-      "step": 24118
-    },
-    {
-      "epoch": 3.884174081082169,
-      "grad_norm": 0.01313191931694746,
-      "learning_rate": 0.000199992561661708,
-      "loss": 46.0,
-      "step": 24119
-    },
-    {
-      "epoch": 3.8843351181609567,
-      "grad_norm": 0.002989828120917082,
-      "learning_rate": 0.0001999925610446438,
-      "loss": 46.0,
-      "step": 24120
-    },
-    {
-      "epoch": 3.884496155239744,
-      "grad_norm": 0.0030406485311686993,
-      "learning_rate": 0.00019999256042755402,
-      "loss": 46.0,
-      "step": 24121
-    },
-    {
-      "epoch": 3.8846571923185316,
-      "grad_norm": 0.001560604083351791,
-      "learning_rate": 0.00019999255981043864,
-      "loss": 46.0,
-      "step": 24122
-    },
-    {
-      "epoch": 3.8848182293973186,
-      "grad_norm": 0.003048016456887126,
-      "learning_rate": 0.00019999255919329767,
-      "loss": 46.0,
-      "step": 24123
-    },
-    {
-      "epoch": 3.884979266476106,
-      "grad_norm": 0.005632190499454737,
-      "learning_rate": 0.0001999925585761311,
-      "loss": 46.0,
-      "step": 24124
-    },
-    {
-      "epoch": 3.8851403035548935,
-      "grad_norm": 0.008868822827935219,
-      "learning_rate": 0.00019999255795893892,
-      "loss": 46.0,
-      "step": 24125
-    },
-    {
-      "epoch": 3.885301340633681,
-      "grad_norm": 0.0019880905747413635,
-      "learning_rate": 0.0001999925573417212,
-      "loss": 46.0,
-      "step": 24126
-    },
-    {
-      "epoch": 3.8854623777124684,
-      "grad_norm": 0.012027781456708908,
-      "learning_rate": 0.00019999255672447784,
-      "loss": 46.0,
-      "step": 24127
-    },
-    {
-      "epoch": 3.885623414791256,
-      "grad_norm": 0.0021413180511444807,
-      "learning_rate": 0.0001999925561072089,
-      "loss": 46.0,
-      "step": 24128
-    },
-    {
-      "epoch": 3.885784451870043,
-      "grad_norm": 0.0014022750547155738,
-      "learning_rate": 0.00019999255548991437,
-      "loss": 46.0,
-      "step": 24129
-    },
-    {
-      "epoch": 3.8859454889488303,
-      "grad_norm": 0.0031765063758939505,
-      "learning_rate": 0.00019999255487259423,
-      "loss": 46.0,
-      "step": 24130
-    },
-    {
-      "epoch": 3.8861065260276177,
-      "grad_norm": 0.009840128943324089,
-      "learning_rate": 0.00019999255425524854,
-      "loss": 46.0,
-      "step": 24131
-    },
-    {
-      "epoch": 3.886267563106405,
-      "grad_norm": 0.0036061331629753113,
-      "learning_rate": 0.00019999255363787723,
-      "loss": 46.0,
-      "step": 24132
-    },
-    {
-      "epoch": 3.8864286001851926,
-      "grad_norm": 0.00471268268302083,
-      "learning_rate": 0.0001999925530204803,
-      "loss": 46.0,
-      "step": 24133
-    },
-    {
-      "epoch": 3.88658963726398,
-      "grad_norm": 0.0011690089013427496,
-      "learning_rate": 0.00019999255240305783,
-      "loss": 46.0,
-      "step": 24134
-    },
-    {
-      "epoch": 3.8867506743427676,
-      "grad_norm": 0.004206649027764797,
-      "learning_rate": 0.00019999255178560974,
-      "loss": 46.0,
-      "step": 24135
-    },
-    {
-      "epoch": 3.886911711421555,
-      "grad_norm": 0.010350743308663368,
-      "learning_rate": 0.00019999255116813605,
-      "loss": 46.0,
-      "step": 24136
-    },
-    {
-      "epoch": 3.8870727485003425,
-      "grad_norm": 0.008397089317440987,
-      "learning_rate": 0.00019999255055063678,
-      "loss": 46.0,
-      "step": 24137
-    },
-    {
-      "epoch": 3.8872337855791295,
-      "grad_norm": 0.004742264281958342,
-      "learning_rate": 0.0001999925499331119,
-      "loss": 46.0,
-      "step": 24138
-    },
-    {
-      "epoch": 3.887394822657917,
-      "grad_norm": 0.007493521552532911,
-      "learning_rate": 0.00019999254931556143,
-      "loss": 46.0,
-      "step": 24139
-    },
-    {
-      "epoch": 3.8875558597367044,
-      "grad_norm": 0.002483305288478732,
-      "learning_rate": 0.0001999925486979854,
-      "loss": 46.0,
-      "step": 24140
-    },
-    {
-      "epoch": 3.887716896815492,
-      "grad_norm": 0.011176660656929016,
-      "learning_rate": 0.00019999254808038375,
-      "loss": 46.0,
-      "step": 24141
-    },
-    {
-      "epoch": 3.8878779338942793,
-      "grad_norm": 0.00583136361092329,
-      "learning_rate": 0.00019999254746275652,
-      "loss": 46.0,
-      "step": 24142
-    },
-    {
-      "epoch": 3.8880389709730663,
-      "grad_norm": 0.0037059015594422817,
-      "learning_rate": 0.00019999254684510367,
-      "loss": 46.0,
-      "step": 24143
-    },
-    {
-      "epoch": 3.8882000080518537,
-      "grad_norm": 0.013053884729743004,
-      "learning_rate": 0.00019999254622742524,
-      "loss": 46.0,
-      "step": 24144
-    },
-    {
-      "epoch": 3.888361045130641,
-      "grad_norm": 0.008035755716264248,
-      "learning_rate": 0.0001999925456097212,
-      "loss": 46.0,
-      "step": 24145
-    },
-    {
-      "epoch": 3.8885220822094286,
-      "grad_norm": 0.0010997587814927101,
-      "learning_rate": 0.00019999254499199163,
-      "loss": 46.0,
-      "step": 24146
-    },
-    {
-      "epoch": 3.888683119288216,
-      "grad_norm": 0.001482553081586957,
-      "learning_rate": 0.0001999925443742364,
-      "loss": 46.0,
-      "step": 24147
-    },
-    {
-      "epoch": 3.8888441563670035,
-      "grad_norm": 0.0019616500940173864,
-      "learning_rate": 0.00019999254375645563,
-      "loss": 46.0,
-      "step": 24148
-    },
-    {
-      "epoch": 3.889005193445791,
-      "grad_norm": 0.003192444331943989,
-      "learning_rate": 0.0001999925431386492,
-      "loss": 46.0,
-      "step": 24149
-    },
-    {
-      "epoch": 3.8891662305245784,
-      "grad_norm": 0.0011868590954691172,
-      "learning_rate": 0.00019999254252081724,
-      "loss": 46.0,
-      "step": 24150
-    },
-    {
-      "epoch": 3.889327267603366,
-      "grad_norm": 0.007910474203526974,
-      "learning_rate": 0.00019999254190295967,
-      "loss": 46.0,
-      "step": 24151
-    },
-    {
-      "epoch": 3.8894883046821533,
-      "grad_norm": 0.015386927872896194,
-      "learning_rate": 0.00019999254128507649,
-      "loss": 46.0,
-      "step": 24152
-    },
-    {
-      "epoch": 3.8896493417609403,
-      "grad_norm": 0.010213050059974194,
-      "learning_rate": 0.00019999254066716774,
-      "loss": 46.0,
-      "step": 24153
-    },
-    {
-      "epoch": 3.889810378839728,
-      "grad_norm": 0.003139836946502328,
-      "learning_rate": 0.00019999254004923338,
-      "loss": 46.0,
-      "step": 24154
-    },
-    {
-      "epoch": 3.8899714159185153,
-      "grad_norm": 0.0035195823293179274,
-      "learning_rate": 0.00019999253943127343,
-      "loss": 46.0,
-      "step": 24155
-    },
-    {
-      "epoch": 3.8901324529973027,
-      "grad_norm": 0.0009600928169675171,
-      "learning_rate": 0.0001999925388132879,
-      "loss": 46.0,
-      "step": 24156
-    },
-    {
-      "epoch": 3.89029349007609,
-      "grad_norm": 0.002001772169023752,
-      "learning_rate": 0.00019999253819527675,
-      "loss": 46.0,
-      "step": 24157
-    },
-    {
-      "epoch": 3.890454527154877,
-      "grad_norm": 0.0035237520933151245,
-      "learning_rate": 0.00019999253757724,
-      "loss": 46.0,
-      "step": 24158
-    },
-    {
-      "epoch": 3.8906155642336646,
-      "grad_norm": 0.0014567676698789,
-      "learning_rate": 0.0001999925369591777,
-      "loss": 46.0,
-      "step": 24159
-    },
-    {
-      "epoch": 3.890776601312452,
-      "grad_norm": 0.0017159712733700871,
-      "learning_rate": 0.0001999925363410898,
-      "loss": 46.0,
-      "step": 24160
-    },
-    {
-      "epoch": 3.8909376383912395,
-      "grad_norm": 0.005806478671729565,
-      "learning_rate": 0.00019999253572297628,
-      "loss": 46.0,
-      "step": 24161
-    },
-    {
-      "epoch": 3.891098675470027,
-      "grad_norm": 0.007622143253684044,
-      "learning_rate": 0.0001999925351048372,
-      "loss": 46.0,
-      "step": 24162
-    },
-    {
-      "epoch": 3.8912597125488144,
-      "grad_norm": 0.0026005215477198362,
-      "learning_rate": 0.0001999925344866725,
-      "loss": 46.0,
-      "step": 24163
-    },
-    {
-      "epoch": 3.891420749627602,
-      "grad_norm": 0.0015932057285681367,
-      "learning_rate": 0.0001999925338684822,
-      "loss": 46.0,
-      "step": 24164
-    },
-    {
-      "epoch": 3.8915817867063893,
-      "grad_norm": 0.007183215580880642,
-      "learning_rate": 0.00019999253325026634,
-      "loss": 46.0,
-      "step": 24165
-    },
-    {
-      "epoch": 3.8917428237851768,
-      "grad_norm": 0.0019393969560042024,
-      "learning_rate": 0.00019999253263202488,
-      "loss": 46.0,
-      "step": 24166
-    },
-    {
-      "epoch": 3.891903860863964,
-      "grad_norm": 0.004852348007261753,
-      "learning_rate": 0.0001999925320137578,
-      "loss": 46.0,
-      "step": 24167
-    },
-    {
-      "epoch": 3.8920648979427512,
-      "grad_norm": 0.007236093748360872,
-      "learning_rate": 0.00019999253139546514,
-      "loss": 46.0,
-      "step": 24168
-    },
-    {
-      "epoch": 3.8922259350215387,
-      "grad_norm": 0.007589172106236219,
-      "learning_rate": 0.0001999925307771469,
-      "loss": 46.0,
-      "step": 24169
-    },
-    {
-      "epoch": 3.892386972100326,
-      "grad_norm": 0.006314924452453852,
-      "learning_rate": 0.00019999253015880305,
-      "loss": 46.0,
-      "step": 24170
-    },
-    {
-      "epoch": 3.8925480091791136,
-      "grad_norm": 0.0028942995704710484,
-      "learning_rate": 0.0001999925295404336,
-      "loss": 46.0,
-      "step": 24171
-    },
-    {
-      "epoch": 3.892709046257901,
-      "grad_norm": 0.0020921321120113134,
-      "learning_rate": 0.0001999925289220386,
-      "loss": 46.0,
-      "step": 24172
-    },
-    {
-      "epoch": 3.892870083336688,
-      "grad_norm": 0.0008537573739886284,
-      "learning_rate": 0.00019999252830361797,
-      "loss": 46.0,
-      "step": 24173
-    },
-    {
-      "epoch": 3.8930311204154755,
-      "grad_norm": 0.002026214264333248,
-      "learning_rate": 0.00019999252768517175,
-      "loss": 46.0,
-      "step": 24174
-    },
-    {
-      "epoch": 3.893192157494263,
-      "grad_norm": 0.0033443926367908716,
-      "learning_rate": 0.00019999252706669996,
-      "loss": 46.0,
-      "step": 24175
-    },
-    {
-      "epoch": 3.8933531945730504,
-      "grad_norm": 0.0008840114460326731,
-      "learning_rate": 0.00019999252644820254,
-      "loss": 46.0,
-      "step": 24176
-    },
-    {
-      "epoch": 3.893514231651838,
-      "grad_norm": 0.0081992968916893,
-      "learning_rate": 0.00019999252582967957,
-      "loss": 46.0,
-      "step": 24177
-    },
-    {
-      "epoch": 3.8936752687306253,
-      "grad_norm": 0.0019973102025687695,
-      "learning_rate": 0.00019999252521113095,
-      "loss": 46.0,
-      "step": 24178
-    },
-    {
-      "epoch": 3.8938363058094128,
-      "grad_norm": 0.000560144369956106,
-      "learning_rate": 0.0001999925245925568,
-      "loss": 46.0,
-      "step": 24179
-    },
-    {
-      "epoch": 3.8939973428882,
-      "grad_norm": 0.006176150869578123,
-      "learning_rate": 0.00019999252397395702,
-      "loss": 46.0,
-      "step": 24180
-    },
-    {
-      "epoch": 3.8941583799669877,
-      "grad_norm": 0.00864804070442915,
-      "learning_rate": 0.00019999252335533164,
-      "loss": 46.0,
-      "step": 24181
-    },
-    {
-      "epoch": 3.8943194170457747,
-      "grad_norm": 0.009942607954144478,
-      "learning_rate": 0.0001999925227366807,
-      "loss": 46.0,
-      "step": 24182
-    },
-    {
-      "epoch": 3.894480454124562,
-      "grad_norm": 0.004692689515650272,
-      "learning_rate": 0.00019999252211800415,
-      "loss": 46.0,
-      "step": 24183
-    },
-    {
-      "epoch": 3.8946414912033496,
-      "grad_norm": 0.002124243415892124,
-      "learning_rate": 0.000199992521499302,
-      "loss": 46.0,
-      "step": 24184
-    },
-    {
-      "epoch": 3.894802528282137,
-      "grad_norm": 0.00414430582895875,
-      "learning_rate": 0.0001999925208805743,
-      "loss": 46.0,
-      "step": 24185
-    },
-    {
-      "epoch": 3.8949635653609245,
-      "grad_norm": 0.004519076552242041,
-      "learning_rate": 0.00019999252026182095,
-      "loss": 46.0,
-      "step": 24186
-    },
-    {
-      "epoch": 3.8951246024397115,
-      "grad_norm": 0.002755421446636319,
-      "learning_rate": 0.00019999251964304202,
-      "loss": 46.0,
-      "step": 24187
-    },
-    {
-      "epoch": 3.895285639518499,
-      "grad_norm": 0.00133071793243289,
-      "learning_rate": 0.0001999925190242375,
-      "loss": 46.0,
-      "step": 24188
-    },
-    {
-      "epoch": 3.8954466765972864,
-      "grad_norm": 0.003784241620451212,
-      "learning_rate": 0.0001999925184054074,
-      "loss": 46.0,
-      "step": 24189
-    },
-    {
-      "epoch": 3.895607713676074,
-      "grad_norm": 0.012115403078496456,
-      "learning_rate": 0.0001999925177865517,
-      "loss": 46.0,
-      "step": 24190
-    },
-    {
-      "epoch": 3.8957687507548613,
-      "grad_norm": 0.0018362364498898387,
-      "learning_rate": 0.0001999925171676704,
-      "loss": 46.0,
-      "step": 24191
-    },
-    {
-      "epoch": 3.8959297878336487,
-      "grad_norm": 0.005691426806151867,
-      "learning_rate": 0.0001999925165487635,
-      "loss": 46.0,
-      "step": 24192
-    },
-    {
-      "epoch": 3.896090824912436,
-      "grad_norm": 0.007418483030050993,
-      "learning_rate": 0.00019999251592983105,
-      "loss": 46.0,
-      "step": 24193
-    },
-    {
-      "epoch": 3.8962518619912236,
-      "grad_norm": 0.003390252124518156,
-      "learning_rate": 0.00019999251531087296,
-      "loss": 46.0,
-      "step": 24194
-    },
-    {
-      "epoch": 3.896412899070011,
-      "grad_norm": 0.0013939245836809278,
-      "learning_rate": 0.00019999251469188928,
-      "loss": 46.0,
-      "step": 24195
-    },
-    {
-      "epoch": 3.896573936148798,
-      "grad_norm": 0.0035515360068529844,
-      "learning_rate": 0.00019999251407288001,
-      "loss": 46.0,
-      "step": 24196
-    },
-    {
-      "epoch": 3.8967349732275856,
-      "grad_norm": 0.0044871107675135136,
-      "learning_rate": 0.0001999925134538452,
-      "loss": 46.0,
-      "step": 24197
-    },
-    {
-      "epoch": 3.896896010306373,
-      "grad_norm": 0.0023952689953148365,
-      "learning_rate": 0.00019999251283478472,
-      "loss": 46.0,
-      "step": 24198
-    },
-    {
-      "epoch": 3.8970570473851605,
-      "grad_norm": 0.002432761713862419,
-      "learning_rate": 0.0001999925122156987,
-      "loss": 46.0,
-      "step": 24199
-    },
-    {
-      "epoch": 3.897218084463948,
-      "grad_norm": 0.004769071005284786,
-      "learning_rate": 0.00019999251159658705,
-      "loss": 46.0,
-      "step": 24200
-    },
-    {
-      "epoch": 3.8973791215427354,
-      "grad_norm": 0.013915703631937504,
-      "learning_rate": 0.00019999251097744982,
-      "loss": 46.0,
-      "step": 24201
-    },
-    {
-      "epoch": 3.8975401586215224,
-      "grad_norm": 0.005264056380838156,
-      "learning_rate": 0.000199992510358287,
-      "loss": 46.0,
-      "step": 24202
-    },
-    {
-      "epoch": 3.89770119570031,
-      "grad_norm": 0.007863735780119896,
-      "learning_rate": 0.00019999250973909863,
-      "loss": 46.0,
-      "step": 24203
-    },
-    {
-      "epoch": 3.8978622327790973,
-      "grad_norm": 0.001511303591541946,
-      "learning_rate": 0.00019999250911988462,
-      "loss": 46.0,
-      "step": 24204
-    },
-    {
-      "epoch": 3.8980232698578847,
-      "grad_norm": 0.004624978173524141,
-      "learning_rate": 0.000199992508500645,
-      "loss": 46.0,
-      "step": 24205
-    },
-    {
-      "epoch": 3.898184306936672,
-      "grad_norm": 0.01257304847240448,
-      "learning_rate": 0.00019999250788137982,
-      "loss": 46.0,
-      "step": 24206
-    },
-    {
-      "epoch": 3.8983453440154596,
-      "grad_norm": 0.012954912148416042,
-      "learning_rate": 0.00019999250726208904,
-      "loss": 46.0,
-      "step": 24207
-    },
-    {
-      "epoch": 3.898506381094247,
-      "grad_norm": 0.0005362546071410179,
-      "learning_rate": 0.00019999250664277268,
-      "loss": 46.0,
-      "step": 24208
-    },
-    {
-      "epoch": 3.8986674181730345,
-      "grad_norm": 0.0005457444349303842,
-      "learning_rate": 0.0001999925060234307,
-      "loss": 46.0,
-      "step": 24209
-    },
-    {
-      "epoch": 3.898828455251822,
-      "grad_norm": 0.004380771424621344,
-      "learning_rate": 0.00019999250540406313,
-      "loss": 46.0,
-      "step": 24210
-    },
-    {
-      "epoch": 3.898989492330609,
-      "grad_norm": 0.0032124503049999475,
-      "learning_rate": 0.00019999250478466997,
-      "loss": 46.0,
-      "step": 24211
-    },
-    {
-      "epoch": 3.8991505294093964,
-      "grad_norm": 0.004542514681816101,
-      "learning_rate": 0.00019999250416525123,
-      "loss": 46.0,
-      "step": 24212
-    },
-    {
-      "epoch": 3.899311566488184,
-      "grad_norm": 0.0030122194439172745,
-      "learning_rate": 0.0001999925035458069,
-      "loss": 46.0,
-      "step": 24213
-    },
-    {
-      "epoch": 3.8994726035669713,
-      "grad_norm": 0.015683606266975403,
-      "learning_rate": 0.000199992502926337,
-      "loss": 46.0,
-      "step": 24214
-    },
-    {
-      "epoch": 3.899633640645759,
-      "grad_norm": 0.018528258427977562,
-      "learning_rate": 0.00019999250230684143,
-      "loss": 46.0,
-      "step": 24215
-    },
-    {
-      "epoch": 3.899794677724546,
-      "grad_norm": 0.005893013905733824,
-      "learning_rate": 0.00019999250168732031,
-      "loss": 46.0,
-      "step": 24216
-    },
-    {
-      "epoch": 3.8999557148033333,
-      "grad_norm": 0.008164685219526291,
-      "learning_rate": 0.0001999925010677736,
-      "loss": 46.0,
-      "step": 24217
-    },
-    {
-      "epoch": 3.9001167518821207,
-      "grad_norm": 0.005075582303106785,
-      "learning_rate": 0.0001999925004482013,
-      "loss": 46.0,
-      "step": 24218
-    },
-    {
-      "epoch": 3.900277788960908,
-      "grad_norm": 0.004707506392151117,
-      "learning_rate": 0.00019999249982860338,
-      "loss": 46.0,
-      "step": 24219
-    },
-    {
-      "epoch": 3.9004388260396956,
-      "grad_norm": 0.003324424382299185,
-      "learning_rate": 0.00019999249920897995,
-      "loss": 46.0,
-      "step": 24220
-    },
-    {
-      "epoch": 3.900599863118483,
-      "grad_norm": 0.00880496110767126,
-      "learning_rate": 0.00019999249858933084,
-      "loss": 46.0,
-      "step": 24221
-    },
-    {
-      "epoch": 3.9007609001972705,
-      "grad_norm": 0.0033119728323072195,
-      "learning_rate": 0.00019999249796965614,
-      "loss": 46.0,
-      "step": 24222
-    },
-    {
-      "epoch": 3.900921937276058,
-      "grad_norm": 0.0024106313940137625,
-      "learning_rate": 0.0001999924973499559,
-      "loss": 46.0,
-      "step": 24223
-    },
-    {
-      "epoch": 3.9010829743548454,
-      "grad_norm": 0.005441464018076658,
-      "learning_rate": 0.00019999249673023002,
-      "loss": 46.0,
-      "step": 24224
-    },
-    {
-      "epoch": 3.901244011433633,
-      "grad_norm": 0.002073982497677207,
-      "learning_rate": 0.00019999249611047856,
-      "loss": 46.0,
-      "step": 24225
-    },
-    {
-      "epoch": 3.90140504851242,
-      "grad_norm": 0.009327485226094723,
-      "learning_rate": 0.00019999249549070155,
-      "loss": 46.0,
-      "step": 24226
-    },
-    {
-      "epoch": 3.9015660855912073,
-      "grad_norm": 0.001985186478123069,
-      "learning_rate": 0.0001999924948708989,
-      "loss": 46.0,
-      "step": 24227
-    },
-    {
-      "epoch": 3.9017271226699948,
-      "grad_norm": 0.0021331205498427153,
-      "learning_rate": 0.00019999249425107065,
-      "loss": 46.0,
-      "step": 24228
-    },
-    {
-      "epoch": 3.9018881597487822,
-      "grad_norm": 0.0018735101912170649,
-      "learning_rate": 0.00019999249363121684,
-      "loss": 46.0,
-      "step": 24229
-    },
-    {
-      "epoch": 3.9020491968275697,
-      "grad_norm": 0.0019107733387500048,
-      "learning_rate": 0.0001999924930113374,
-      "loss": 46.0,
-      "step": 24230
-    },
-    {
-      "epoch": 3.9022102339063567,
-      "grad_norm": 0.012786966748535633,
-      "learning_rate": 0.0001999924923914324,
-      "loss": 46.0,
-      "step": 24231
-    },
-    {
-      "epoch": 3.902371270985144,
-      "grad_norm": 0.0006641882355324924,
-      "learning_rate": 0.0001999924917715018,
-      "loss": 46.0,
-      "step": 24232
-    },
-    {
-      "epoch": 3.9025323080639316,
-      "grad_norm": 0.0053881434723734856,
-      "learning_rate": 0.0001999924911515456,
-      "loss": 46.0,
-      "step": 24233
-    },
-    {
-      "epoch": 3.902693345142719,
-      "grad_norm": 0.0014949383912608027,
-      "learning_rate": 0.0001999924905315638,
-      "loss": 46.0,
-      "step": 24234
-    },
-    {
-      "epoch": 3.9028543822215065,
-      "grad_norm": 0.008667722344398499,
-      "learning_rate": 0.0001999924899115564,
-      "loss": 46.0,
-      "step": 24235
-    },
-    {
-      "epoch": 3.903015419300294,
-      "grad_norm": 0.003046006429940462,
-      "learning_rate": 0.00019999248929152344,
-      "loss": 46.0,
-      "step": 24236
-    },
-    {
-      "epoch": 3.9031764563790814,
-      "grad_norm": 0.0012053512036800385,
-      "learning_rate": 0.00019999248867146486,
-      "loss": 46.0,
-      "step": 24237
-    },
-    {
-      "epoch": 3.903337493457869,
-      "grad_norm": 0.010604547336697578,
-      "learning_rate": 0.0001999924880513807,
-      "loss": 46.0,
-      "step": 24238
-    },
-    {
-      "epoch": 3.9034985305366563,
-      "grad_norm": 0.00573544017970562,
-      "learning_rate": 0.00019999248743127096,
-      "loss": 46.0,
-      "step": 24239
-    },
-    {
-      "epoch": 3.9036595676154433,
-      "grad_norm": 0.002507448196411133,
-      "learning_rate": 0.00019999248681113558,
-      "loss": 46.0,
-      "step": 24240
-    },
-    {
-      "epoch": 3.9038206046942308,
-      "grad_norm": 0.0019390815868973732,
-      "learning_rate": 0.00019999248619097462,
-      "loss": 46.0,
-      "step": 24241
-    },
-    {
-      "epoch": 3.903981641773018,
-      "grad_norm": 0.007382815703749657,
-      "learning_rate": 0.0001999924855707881,
-      "loss": 46.0,
-      "step": 24242
-    },
-    {
-      "epoch": 3.9041426788518057,
-      "grad_norm": 0.001997357001528144,
-      "learning_rate": 0.000199992484950576,
-      "loss": 46.0,
-      "step": 24243
-    },
-    {
-      "epoch": 3.904303715930593,
-      "grad_norm": 0.0026674431283026934,
-      "learning_rate": 0.00019999248433033828,
-      "loss": 46.0,
-      "step": 24244
-    },
-    {
-      "epoch": 3.9044647530093806,
-      "grad_norm": 0.0023212411906570196,
-      "learning_rate": 0.00019999248371007494,
-      "loss": 46.0,
-      "step": 24245
-    },
-    {
-      "epoch": 3.9046257900881676,
-      "grad_norm": 0.0011796843027696013,
-      "learning_rate": 0.00019999248308978605,
-      "loss": 46.0,
-      "step": 24246
-    },
-    {
-      "epoch": 3.904786827166955,
-      "grad_norm": 0.0010502564255148172,
-      "learning_rate": 0.00019999248246947157,
-      "loss": 46.0,
-      "step": 24247
-    },
-    {
-      "epoch": 3.9049478642457425,
-      "grad_norm": 0.00227341684512794,
-      "learning_rate": 0.00019999248184913144,
-      "loss": 46.0,
-      "step": 24248
-    },
-    {
-      "epoch": 3.90510890132453,
-      "grad_norm": 0.0022540211211889982,
-      "learning_rate": 0.00019999248122876576,
-      "loss": 46.0,
-      "step": 24249
-    },
-    {
-      "epoch": 3.9052699384033174,
-      "grad_norm": 0.0015600745100528002,
-      "learning_rate": 0.0001999924806083745,
-      "loss": 46.0,
-      "step": 24250
-    },
-    {
-      "epoch": 3.905430975482105,
-      "grad_norm": 0.007016841322183609,
-      "learning_rate": 0.00019999247998795763,
-      "loss": 46.0,
-      "step": 24251
-    },
-    {
-      "epoch": 3.9055920125608923,
-      "grad_norm": 0.007967165671288967,
-      "learning_rate": 0.00019999247936751516,
-      "loss": 46.0,
-      "step": 24252
-    },
-    {
-      "epoch": 3.9057530496396797,
-      "grad_norm": 0.003511112881824374,
-      "learning_rate": 0.0001999924787470471,
-      "loss": 46.0,
-      "step": 24253
-    },
-    {
-      "epoch": 3.905914086718467,
-      "grad_norm": 0.0021445807069540024,
-      "learning_rate": 0.00019999247812655345,
-      "loss": 46.0,
-      "step": 24254
-    },
-    {
-      "epoch": 3.906075123797254,
-      "grad_norm": 0.0020800428465008736,
-      "learning_rate": 0.00019999247750603422,
-      "loss": 46.0,
-      "step": 24255
-    },
-    {
-      "epoch": 3.9062361608760416,
-      "grad_norm": 0.0035484626423567533,
-      "learning_rate": 0.00019999247688548937,
-      "loss": 46.0,
-      "step": 24256
-    },
-    {
-      "epoch": 3.906397197954829,
-      "grad_norm": 0.010253670625388622,
-      "learning_rate": 0.00019999247626491893,
-      "loss": 46.0,
-      "step": 24257
-    },
-    {
-      "epoch": 3.9065582350336165,
-      "grad_norm": 0.0008009191369637847,
-      "learning_rate": 0.0001999924756443229,
-      "loss": 46.0,
-      "step": 24258
-    },
-    {
-      "epoch": 3.906719272112404,
-      "grad_norm": 0.004874709993600845,
-      "learning_rate": 0.0001999924750237013,
-      "loss": 46.0,
-      "step": 24259
-    },
-    {
-      "epoch": 3.906880309191191,
-      "grad_norm": 0.006979524157941341,
-      "learning_rate": 0.00019999247440305408,
-      "loss": 46.0,
-      "step": 24260
-    },
-    {
-      "epoch": 3.9070413462699785,
-      "grad_norm": 0.008102000690996647,
-      "learning_rate": 0.0001999924737823813,
-      "loss": 46.0,
-      "step": 24261
-    },
-    {
-      "epoch": 3.907202383348766,
-      "grad_norm": 0.005288542248308659,
-      "learning_rate": 0.0001999924731616829,
-      "loss": 46.0,
-      "step": 24262
-    },
-    {
-      "epoch": 3.9073634204275534,
-      "grad_norm": 0.0025655077770352364,
-      "learning_rate": 0.0001999924725409589,
-      "loss": 46.0,
-      "step": 24263
-    },
-    {
-      "epoch": 3.907524457506341,
-      "grad_norm": 0.0030293199233710766,
-      "learning_rate": 0.00019999247192020934,
-      "loss": 46.0,
-      "step": 24264
-    },
-    {
-      "epoch": 3.9076854945851283,
-      "grad_norm": 0.005259111989289522,
-      "learning_rate": 0.00019999247129943415,
-      "loss": 46.0,
-      "step": 24265
-    },
-    {
-      "epoch": 3.9078465316639157,
-      "grad_norm": 0.006712483707815409,
-      "learning_rate": 0.0001999924706786334,
-      "loss": 46.0,
-      "step": 24266
-    },
-    {
-      "epoch": 3.908007568742703,
-      "grad_norm": 0.002392790513113141,
-      "learning_rate": 0.00019999247005780704,
-      "loss": 46.0,
-      "step": 24267
-    },
-    {
-      "epoch": 3.9081686058214906,
-      "grad_norm": 0.005893141031265259,
-      "learning_rate": 0.0001999924694369551,
-      "loss": 46.0,
-      "step": 24268
-    },
-    {
-      "epoch": 3.9083296429002776,
-      "grad_norm": 0.002989998087286949,
-      "learning_rate": 0.00019999246881607753,
-      "loss": 46.0,
-      "step": 24269
-    },
-    {
-      "epoch": 3.908490679979065,
-      "grad_norm": 0.0043844375759363174,
-      "learning_rate": 0.0001999924681951744,
-      "loss": 46.0,
-      "step": 24270
-    },
-    {
-      "epoch": 3.9086517170578525,
-      "grad_norm": 0.0005724730435758829,
-      "learning_rate": 0.00019999246757424567,
-      "loss": 46.0,
-      "step": 24271
-    },
-    {
-      "epoch": 3.90881275413664,
-      "grad_norm": 0.0026805636007338762,
-      "learning_rate": 0.00019999246695329135,
-      "loss": 46.0,
-      "step": 24272
-    },
-    {
-      "epoch": 3.9089737912154274,
-      "grad_norm": 0.006760813295841217,
-      "learning_rate": 0.00019999246633231144,
-      "loss": 46.0,
-      "step": 24273
-    },
-    {
-      "epoch": 3.909134828294215,
-      "grad_norm": 0.0062536257319152355,
-      "learning_rate": 0.0001999924657113059,
-      "loss": 46.0,
-      "step": 24274
-    },
-    {
-      "epoch": 3.909295865373002,
-      "grad_norm": 0.006175358314067125,
-      "learning_rate": 0.00019999246509027482,
-      "loss": 46.0,
-      "step": 24275
-    },
-    {
-      "epoch": 3.9094569024517893,
-      "grad_norm": 0.0011312743881717324,
-      "learning_rate": 0.00019999246446921812,
-      "loss": 46.0,
-      "step": 24276
-    },
-    {
-      "epoch": 3.909617939530577,
-      "grad_norm": 0.005588274449110031,
-      "learning_rate": 0.00019999246384813584,
-      "loss": 46.0,
-      "step": 24277
-    },
-    {
-      "epoch": 3.9097789766093642,
-      "grad_norm": 0.004909217823296785,
-      "learning_rate": 0.00019999246322702796,
-      "loss": 46.0,
-      "step": 24278
-    },
-    {
-      "epoch": 3.9099400136881517,
-      "grad_norm": 0.006424983497709036,
-      "learning_rate": 0.00019999246260589447,
-      "loss": 46.0,
-      "step": 24279
-    },
-    {
-      "epoch": 3.910101050766939,
-      "grad_norm": 0.0027578778099268675,
-      "learning_rate": 0.00019999246198473543,
-      "loss": 46.0,
-      "step": 24280
-    },
-    {
-      "epoch": 3.9102620878457266,
-      "grad_norm": 0.0024216067977249622,
-      "learning_rate": 0.00019999246136355076,
-      "loss": 46.0,
-      "step": 24281
-    },
-    {
-      "epoch": 3.910423124924514,
-      "grad_norm": 0.0019010824616998434,
-      "learning_rate": 0.0001999924607423405,
-      "loss": 46.0,
-      "step": 24282
-    },
-    {
-      "epoch": 3.9105841620033015,
-      "grad_norm": 0.008727259933948517,
-      "learning_rate": 0.00019999246012110468,
-      "loss": 46.0,
-      "step": 24283
-    },
-    {
-      "epoch": 3.9107451990820885,
-      "grad_norm": 0.003990017343312502,
-      "learning_rate": 0.00019999245949984322,
-      "loss": 46.0,
-      "step": 24284
-    },
-    {
-      "epoch": 3.910906236160876,
-      "grad_norm": 0.0037385544274002314,
-      "learning_rate": 0.00019999245887855618,
-      "loss": 46.0,
-      "step": 24285
-    },
-    {
-      "epoch": 3.9110672732396634,
-      "grad_norm": 0.0029327156953513622,
-      "learning_rate": 0.00019999245825724359,
-      "loss": 46.0,
-      "step": 24286
-    },
-    {
-      "epoch": 3.911228310318451,
-      "grad_norm": 0.001799881225451827,
-      "learning_rate": 0.00019999245763590535,
-      "loss": 46.0,
-      "step": 24287
-    },
-    {
-      "epoch": 3.9113893473972383,
-      "grad_norm": 0.009433192200958729,
-      "learning_rate": 0.00019999245701454155,
-      "loss": 46.0,
-      "step": 24288
-    },
-    {
-      "epoch": 3.9115503844760258,
-      "grad_norm": 0.0045669106766581535,
-      "learning_rate": 0.00019999245639315213,
-      "loss": 46.0,
-      "step": 24289
-    },
-    {
-      "epoch": 3.9117114215548128,
-      "grad_norm": 0.001397787593305111,
-      "learning_rate": 0.00019999245577173716,
-      "loss": 46.0,
-      "step": 24290
-    },
-    {
-      "epoch": 3.9118724586336002,
-      "grad_norm": 0.0020416441839188337,
-      "learning_rate": 0.00019999245515029657,
-      "loss": 46.0,
-      "step": 24291
-    },
-    {
-      "epoch": 3.9120334957123877,
-      "grad_norm": 0.002587964292615652,
-      "learning_rate": 0.00019999245452883036,
-      "loss": 46.0,
-      "step": 24292
-    },
-    {
-      "epoch": 3.912194532791175,
-      "grad_norm": 0.005450607743114233,
-      "learning_rate": 0.0001999924539073386,
-      "loss": 46.0,
-      "step": 24293
-    },
-    {
-      "epoch": 3.9123555698699626,
-      "grad_norm": 0.005157710053026676,
-      "learning_rate": 0.00019999245328582122,
-      "loss": 46.0,
-      "step": 24294
-    },
-    {
-      "epoch": 3.91251660694875,
-      "grad_norm": 0.004069255664944649,
-      "learning_rate": 0.00019999245266427826,
-      "loss": 46.0,
-      "step": 24295
-    },
-    {
-      "epoch": 3.9126776440275375,
-      "grad_norm": 0.010862830094993114,
-      "learning_rate": 0.0001999924520427097,
-      "loss": 46.0,
-      "step": 24296
-    },
-    {
-      "epoch": 3.912838681106325,
-      "grad_norm": 0.003951295278966427,
-      "learning_rate": 0.00019999245142111557,
-      "loss": 46.0,
-      "step": 24297
-    },
-    {
-      "epoch": 3.9129997181851124,
-      "grad_norm": 0.01030823215842247,
-      "learning_rate": 0.00019999245079949581,
-      "loss": 46.0,
-      "step": 24298
-    },
-    {
-      "epoch": 3.9131607552638994,
-      "grad_norm": 0.006517235189676285,
-      "learning_rate": 0.00019999245017785047,
-      "loss": 46.0,
-      "step": 24299
-    },
-    {
-      "epoch": 3.913321792342687,
-      "grad_norm": 0.008500062860548496,
-      "learning_rate": 0.00019999244955617957,
-      "loss": 46.0,
-      "step": 24300
-    },
-    {
-      "epoch": 3.9134828294214743,
-      "grad_norm": 0.0017892306204885244,
-      "learning_rate": 0.00019999244893448306,
-      "loss": 46.0,
-      "step": 24301
-    },
-    {
-      "epoch": 3.9136438665002617,
-      "grad_norm": 0.002678181976079941,
-      "learning_rate": 0.00019999244831276093,
-      "loss": 46.0,
-      "step": 24302
-    },
-    {
-      "epoch": 3.913804903579049,
-      "grad_norm": 0.01573646441102028,
-      "learning_rate": 0.00019999244769101324,
-      "loss": 46.0,
-      "step": 24303
-    },
-    {
-      "epoch": 3.913965940657836,
-      "grad_norm": 0.000764210766647011,
-      "learning_rate": 0.00019999244706923994,
-      "loss": 46.0,
-      "step": 24304
-    },
-    {
-      "epoch": 3.9141269777366237,
-      "grad_norm": 0.003422482404857874,
-      "learning_rate": 0.00019999244644744107,
-      "loss": 46.0,
-      "step": 24305
-    },
-    {
-      "epoch": 3.914288014815411,
-      "grad_norm": 0.008215613663196564,
-      "learning_rate": 0.00019999244582561657,
-      "loss": 46.0,
-      "step": 24306
-    },
-    {
-      "epoch": 3.9144490518941986,
-      "grad_norm": 0.0015661359066143632,
-      "learning_rate": 0.00019999244520376648,
-      "loss": 46.0,
-      "step": 24307
-    },
-    {
-      "epoch": 3.914610088972986,
-      "grad_norm": 0.007579847238957882,
-      "learning_rate": 0.00019999244458189082,
-      "loss": 46.0,
-      "step": 24308
-    },
-    {
-      "epoch": 3.9147711260517735,
-      "grad_norm": 0.0019457826856523752,
-      "learning_rate": 0.00019999244395998956,
-      "loss": 46.0,
-      "step": 24309
-    },
-    {
-      "epoch": 3.914932163130561,
-      "grad_norm": 0.0009566232329234481,
-      "learning_rate": 0.0001999924433380627,
-      "loss": 46.0,
-      "step": 24310
-    },
-    {
-      "epoch": 3.9150932002093484,
-      "grad_norm": 0.007234182208776474,
-      "learning_rate": 0.00019999244271611026,
-      "loss": 46.0,
-      "step": 24311
-    },
-    {
-      "epoch": 3.915254237288136,
-      "grad_norm": 0.004438658244907856,
-      "learning_rate": 0.0001999924420941322,
-      "loss": 46.0,
-      "step": 24312
-    },
-    {
-      "epoch": 3.915415274366923,
-      "grad_norm": 0.002196710556745529,
-      "learning_rate": 0.0001999924414721286,
-      "loss": 46.0,
-      "step": 24313
-    },
-    {
-      "epoch": 3.9155763114457103,
-      "grad_norm": 0.0016576610505580902,
-      "learning_rate": 0.00019999244085009936,
-      "loss": 46.0,
-      "step": 24314
-    },
-    {
-      "epoch": 3.9157373485244977,
-      "grad_norm": 0.013751697726547718,
-      "learning_rate": 0.00019999244022804452,
-      "loss": 46.0,
-      "step": 24315
-    },
-    {
-      "epoch": 3.915898385603285,
-      "grad_norm": 0.00655261380597949,
-      "learning_rate": 0.00019999243960596412,
-      "loss": 46.0,
-      "step": 24316
-    },
-    {
-      "epoch": 3.9160594226820726,
-      "grad_norm": 0.0025038616731762886,
-      "learning_rate": 0.0001999924389838581,
-      "loss": 46.0,
-      "step": 24317
-    },
-    {
-      "epoch": 3.91622045976086,
-      "grad_norm": 0.008263389579951763,
-      "learning_rate": 0.00019999243836172652,
-      "loss": 46.0,
-      "step": 24318
-    },
-    {
-      "epoch": 3.916381496839647,
-      "grad_norm": 0.00229754694737494,
-      "learning_rate": 0.00019999243773956933,
-      "loss": 46.0,
-      "step": 24319
-    },
-    {
-      "epoch": 3.9165425339184345,
-      "grad_norm": 0.004807076882570982,
-      "learning_rate": 0.00019999243711738655,
-      "loss": 46.0,
-      "step": 24320
-    },
-    {
-      "epoch": 3.916703570997222,
-      "grad_norm": 0.0017659561708569527,
-      "learning_rate": 0.00019999243649517816,
-      "loss": 46.0,
-      "step": 24321
-    },
-    {
-      "epoch": 3.9168646080760094,
-      "grad_norm": 0.0021025885362178087,
-      "learning_rate": 0.00019999243587294418,
-      "loss": 46.0,
-      "step": 24322
-    },
-    {
-      "epoch": 3.917025645154797,
-      "grad_norm": 0.013332934118807316,
-      "learning_rate": 0.00019999243525068463,
-      "loss": 46.0,
-      "step": 24323
-    },
-    {
-      "epoch": 3.9171866822335843,
-      "grad_norm": 0.0029037711210548878,
-      "learning_rate": 0.00019999243462839948,
-      "loss": 46.0,
-      "step": 24324
-    },
-    {
-      "epoch": 3.917347719312372,
-      "grad_norm": 0.01141924038529396,
-      "learning_rate": 0.0001999924340060887,
-      "loss": 46.0,
-      "step": 24325
-    },
-    {
-      "epoch": 3.9175087563911593,
-      "grad_norm": 0.004519221372902393,
-      "learning_rate": 0.0001999924333837524,
-      "loss": 46.0,
-      "step": 24326
-    },
-    {
-      "epoch": 3.9176697934699467,
-      "grad_norm": 0.0029633843805640936,
-      "learning_rate": 0.00019999243276139046,
-      "loss": 46.0,
-      "step": 24327
-    },
-    {
-      "epoch": 3.9178308305487337,
-      "grad_norm": 0.002741101896390319,
-      "learning_rate": 0.0001999924321390029,
-      "loss": 46.0,
-      "step": 24328
-    },
-    {
-      "epoch": 3.917991867627521,
-      "grad_norm": 0.003616885282099247,
-      "learning_rate": 0.0001999924315165898,
-      "loss": 46.0,
-      "step": 24329
-    },
-    {
-      "epoch": 3.9181529047063086,
-      "grad_norm": 0.0010131315793842077,
-      "learning_rate": 0.00019999243089415108,
-      "loss": 46.0,
-      "step": 24330
-    },
-    {
-      "epoch": 3.918313941785096,
-      "grad_norm": 0.0015564961358904839,
-      "learning_rate": 0.00019999243027168676,
-      "loss": 46.0,
-      "step": 24331
-    },
-    {
-      "epoch": 3.9184749788638835,
-      "grad_norm": 0.016632836312055588,
-      "learning_rate": 0.00019999242964919688,
-      "loss": 46.0,
-      "step": 24332
-    },
-    {
-      "epoch": 3.9186360159426705,
-      "grad_norm": 0.010603503324091434,
-      "learning_rate": 0.0001999924290266814,
-      "loss": 46.0,
-      "step": 24333
-    },
-    {
-      "epoch": 3.918797053021458,
-      "grad_norm": 0.0012143533676862717,
-      "learning_rate": 0.0001999924284041403,
-      "loss": 46.0,
-      "step": 24334
-    },
-    {
-      "epoch": 3.9189580901002454,
-      "grad_norm": 0.004277637228369713,
-      "learning_rate": 0.00019999242778157364,
-      "loss": 46.0,
-      "step": 24335
-    },
-    {
-      "epoch": 3.919119127179033,
-      "grad_norm": 0.0029652700759470463,
-      "learning_rate": 0.00019999242715898136,
-      "loss": 46.0,
-      "step": 24336
-    },
-    {
-      "epoch": 3.9192801642578203,
-      "grad_norm": 0.011836004443466663,
-      "learning_rate": 0.00019999242653636349,
-      "loss": 46.0,
-      "step": 24337
-    },
-    {
-      "epoch": 3.919441201336608,
-      "grad_norm": 0.0017603442538529634,
-      "learning_rate": 0.00019999242591372003,
-      "loss": 46.0,
-      "step": 24338
-    },
-    {
-      "epoch": 3.9196022384153952,
-      "grad_norm": 0.005851692520081997,
-      "learning_rate": 0.00019999242529105098,
-      "loss": 46.0,
-      "step": 24339
-    },
-    {
-      "epoch": 3.9197632754941827,
-      "grad_norm": 0.0037130527198314667,
-      "learning_rate": 0.00019999242466835635,
-      "loss": 46.0,
-      "step": 24340
-    },
-    {
-      "epoch": 3.91992431257297,
-      "grad_norm": 0.0024980197194963694,
-      "learning_rate": 0.0001999924240456361,
-      "loss": 46.0,
-      "step": 24341
-    },
-    {
-      "epoch": 3.9200853496517576,
-      "grad_norm": 0.0038946177810430527,
-      "learning_rate": 0.00019999242342289027,
-      "loss": 46.0,
-      "step": 24342
-    },
-    {
-      "epoch": 3.9202463867305446,
-      "grad_norm": 0.0033849331084638834,
-      "learning_rate": 0.00019999242280011886,
-      "loss": 46.0,
-      "step": 24343
-    },
-    {
-      "epoch": 3.920407423809332,
-      "grad_norm": 0.01179580483585596,
-      "learning_rate": 0.00019999242217732185,
-      "loss": 46.0,
-      "step": 24344
-    },
-    {
-      "epoch": 3.9205684608881195,
-      "grad_norm": 0.001581982127390802,
-      "learning_rate": 0.00019999242155449923,
-      "loss": 46.0,
-      "step": 24345
-    },
-    {
-      "epoch": 3.920729497966907,
-      "grad_norm": 0.00549320550635457,
-      "learning_rate": 0.00019999242093165102,
-      "loss": 46.0,
-      "step": 24346
-    },
-    {
-      "epoch": 3.9208905350456944,
-      "grad_norm": 0.0030619860626757145,
-      "learning_rate": 0.00019999242030877725,
-      "loss": 46.0,
-      "step": 24347
-    },
-    {
-      "epoch": 3.9210515721244814,
-      "grad_norm": 0.005441068671643734,
-      "learning_rate": 0.00019999241968587784,
-      "loss": 46.0,
-      "step": 24348
-    },
-    {
-      "epoch": 3.921212609203269,
-      "grad_norm": 0.0016338042914867401,
-      "learning_rate": 0.00019999241906295287,
-      "loss": 46.0,
-      "step": 24349
-    },
-    {
-      "epoch": 3.9213736462820563,
-      "grad_norm": 0.0029275708366185427,
-      "learning_rate": 0.00019999241844000228,
-      "loss": 46.0,
-      "step": 24350
-    },
-    {
-      "epoch": 3.9215346833608438,
-      "grad_norm": 0.0021279742941260338,
-      "learning_rate": 0.00019999241781702614,
-      "loss": 46.0,
-      "step": 24351
-    },
-    {
-      "epoch": 3.921695720439631,
-      "grad_norm": 0.002017433987930417,
-      "learning_rate": 0.00019999241719402438,
-      "loss": 46.0,
-      "step": 24352
-    },
-    {
-      "epoch": 3.9218567575184187,
-      "grad_norm": 0.0036244397051632404,
-      "learning_rate": 0.00019999241657099704,
-      "loss": 46.0,
-      "step": 24353
-    },
-    {
-      "epoch": 3.922017794597206,
-      "grad_norm": 0.0064416504465043545,
-      "learning_rate": 0.0001999924159479441,
-      "loss": 46.0,
-      "step": 24354
-    },
-    {
-      "epoch": 3.9221788316759936,
-      "grad_norm": 0.006330688018351793,
-      "learning_rate": 0.00019999241532486556,
-      "loss": 46.0,
-      "step": 24355
-    },
-    {
-      "epoch": 3.922339868754781,
-      "grad_norm": 0.003200170351192355,
-      "learning_rate": 0.00019999241470176142,
-      "loss": 46.0,
-      "step": 24356
-    },
-    {
-      "epoch": 3.922500905833568,
-      "grad_norm": 0.006174854934215546,
-      "learning_rate": 0.0001999924140786317,
-      "loss": 46.0,
-      "step": 24357
-    },
-    {
-      "epoch": 3.9226619429123555,
-      "grad_norm": 0.007807197980582714,
-      "learning_rate": 0.0001999924134554764,
-      "loss": 46.0,
-      "step": 24358
-    },
-    {
-      "epoch": 3.922822979991143,
-      "grad_norm": 0.005494216922670603,
-      "learning_rate": 0.00019999241283229547,
-      "loss": 46.0,
-      "step": 24359
-    },
-    {
-      "epoch": 3.9229840170699304,
-      "grad_norm": 0.009967494755983353,
-      "learning_rate": 0.00019999241220908896,
-      "loss": 46.0,
-      "step": 24360
-    },
-    {
-      "epoch": 3.923145054148718,
-      "grad_norm": 0.004544772207736969,
-      "learning_rate": 0.00019999241158585686,
-      "loss": 46.0,
-      "step": 24361
-    },
-    {
-      "epoch": 3.9233060912275053,
-      "grad_norm": 0.005755240563303232,
-      "learning_rate": 0.0001999924109625992,
-      "loss": 46.0,
-      "step": 24362
-    },
-    {
-      "epoch": 3.9234671283062923,
-      "grad_norm": 0.002001445507630706,
-      "learning_rate": 0.0001999924103393159,
-      "loss": 46.0,
-      "step": 24363
-    },
-    {
-      "epoch": 3.9236281653850797,
-      "grad_norm": 0.003334293607622385,
-      "learning_rate": 0.00019999240971600704,
-      "loss": 46.0,
-      "step": 24364
-    },
-    {
-      "epoch": 3.923789202463867,
-      "grad_norm": 0.0029219326097518206,
-      "learning_rate": 0.00019999240909267257,
-      "loss": 46.0,
-      "step": 24365
-    },
-    {
-      "epoch": 3.9239502395426546,
-      "grad_norm": 0.0026663679163903,
-      "learning_rate": 0.0001999924084693125,
-      "loss": 46.0,
-      "step": 24366
-    },
-    {
-      "epoch": 3.924111276621442,
-      "grad_norm": 0.002864918438717723,
-      "learning_rate": 0.00019999240784592686,
-      "loss": 46.0,
-      "step": 24367
-    },
-    {
-      "epoch": 3.9242723137002296,
-      "grad_norm": 0.01613437570631504,
-      "learning_rate": 0.00019999240722251563,
-      "loss": 46.0,
-      "step": 24368
-    },
-    {
-      "epoch": 3.924433350779017,
-      "grad_norm": 0.023917198181152344,
-      "learning_rate": 0.00019999240659907875,
-      "loss": 46.0,
-      "step": 24369
-    },
-    {
-      "epoch": 3.9245943878578045,
-      "grad_norm": 0.017110902816057205,
-      "learning_rate": 0.00019999240597561632,
-      "loss": 46.0,
-      "step": 24370
-    },
-    {
-      "epoch": 3.924755424936592,
-      "grad_norm": 0.001088926917873323,
-      "learning_rate": 0.00019999240535212832,
-      "loss": 46.0,
-      "step": 24371
-    },
-    {
-      "epoch": 3.924916462015379,
-      "grad_norm": 0.001707250252366066,
-      "learning_rate": 0.0001999924047286147,
-      "loss": 46.0,
-      "step": 24372
-    },
-    {
-      "epoch": 3.9250774990941664,
-      "grad_norm": 0.007229944691061974,
-      "learning_rate": 0.00019999240410507548,
-      "loss": 46.0,
-      "step": 24373
-    },
-    {
-      "epoch": 3.925238536172954,
-      "grad_norm": 0.0019223009003326297,
-      "learning_rate": 0.0001999924034815107,
-      "loss": 46.0,
-      "step": 24374
-    },
-    {
-      "epoch": 3.9253995732517413,
-      "grad_norm": 0.014989964663982391,
-      "learning_rate": 0.0001999924028579203,
-      "loss": 46.0,
-      "step": 24375
-    },
-    {
-      "epoch": 3.9255606103305287,
-      "grad_norm": 0.005130224861204624,
-      "learning_rate": 0.00019999240223430431,
-      "loss": 46.0,
-      "step": 24376
-    },
-    {
-      "epoch": 3.9257216474093157,
-      "grad_norm": 0.009179562330245972,
-      "learning_rate": 0.00019999240161066274,
-      "loss": 46.0,
-      "step": 24377
-    },
-    {
-      "epoch": 3.925882684488103,
-      "grad_norm": 0.003586171194911003,
-      "learning_rate": 0.00019999240098699555,
-      "loss": 46.0,
-      "step": 24378
-    },
-    {
-      "epoch": 3.9260437215668906,
-      "grad_norm": 0.001972099067643285,
-      "learning_rate": 0.00019999240036330278,
-      "loss": 46.0,
-      "step": 24379
-    },
-    {
-      "epoch": 3.926204758645678,
-      "grad_norm": 0.0045670573599636555,
-      "learning_rate": 0.00019999239973958442,
-      "loss": 46.0,
-      "step": 24380
-    },
-    {
-      "epoch": 3.9263657957244655,
-      "grad_norm": 0.003906788770109415,
-      "learning_rate": 0.00019999239911584047,
-      "loss": 46.0,
-      "step": 24381
-    },
-    {
-      "epoch": 3.926526832803253,
-      "grad_norm": 0.001660844311118126,
-      "learning_rate": 0.00019999239849207093,
-      "loss": 46.0,
-      "step": 24382
-    },
-    {
-      "epoch": 3.9266878698820404,
-      "grad_norm": 0.0007775754784233868,
-      "learning_rate": 0.0001999923978682758,
-      "loss": 46.0,
-      "step": 24383
-    },
-    {
-      "epoch": 3.926848906960828,
-      "grad_norm": 0.007136431522667408,
-      "learning_rate": 0.00019999239724445507,
-      "loss": 46.0,
-      "step": 24384
-    },
-    {
-      "epoch": 3.9270099440396153,
-      "grad_norm": 0.003681618720293045,
-      "learning_rate": 0.00019999239662060874,
-      "loss": 46.0,
-      "step": 24385
-    },
-    {
-      "epoch": 3.9271709811184023,
-      "grad_norm": 0.0019532719161361456,
-      "learning_rate": 0.00019999239599673683,
-      "loss": 46.0,
-      "step": 24386
-    },
-    {
-      "epoch": 3.92733201819719,
-      "grad_norm": 0.006563209462910891,
-      "learning_rate": 0.0001999923953728393,
-      "loss": 46.0,
-      "step": 24387
-    },
-    {
-      "epoch": 3.9274930552759773,
-      "grad_norm": 0.008773471228778362,
-      "learning_rate": 0.0001999923947489162,
-      "loss": 46.0,
-      "step": 24388
-    },
-    {
-      "epoch": 3.9276540923547647,
-      "grad_norm": 0.001396200037561357,
-      "learning_rate": 0.00019999239412496752,
-      "loss": 46.0,
-      "step": 24389
-    },
-    {
-      "epoch": 3.927815129433552,
-      "grad_norm": 0.007127637974917889,
-      "learning_rate": 0.0001999923935009932,
-      "loss": 46.0,
-      "step": 24390
-    },
-    {
-      "epoch": 3.9279761665123396,
-      "grad_norm": 0.017086312174797058,
-      "learning_rate": 0.00019999239287699333,
-      "loss": 46.0,
-      "step": 24391
-    },
-    {
-      "epoch": 3.9281372035911266,
-      "grad_norm": 0.005270370282232761,
-      "learning_rate": 0.00019999239225296784,
-      "loss": 46.0,
-      "step": 24392
-    },
-    {
-      "epoch": 3.928298240669914,
-      "grad_norm": 0.0016827555373311043,
-      "learning_rate": 0.0001999923916289168,
-      "loss": 46.0,
-      "step": 24393
-    },
-    {
-      "epoch": 3.9284592777487015,
-      "grad_norm": 0.003553206566721201,
-      "learning_rate": 0.00019999239100484012,
-      "loss": 46.0,
-      "step": 24394
-    },
-    {
-      "epoch": 3.928620314827489,
-      "grad_norm": 0.0048052165657281876,
-      "learning_rate": 0.00019999239038073787,
-      "loss": 46.0,
-      "step": 24395
-    },
-    {
-      "epoch": 3.9287813519062764,
-      "grad_norm": 0.004999968688935041,
-      "learning_rate": 0.00019999238975661004,
-      "loss": 46.0,
-      "step": 24396
-    },
-    {
-      "epoch": 3.928942388985064,
-      "grad_norm": 0.005583801306784153,
-      "learning_rate": 0.00019999238913245658,
-      "loss": 46.0,
-      "step": 24397
-    },
-    {
-      "epoch": 3.9291034260638513,
-      "grad_norm": 0.0017018058570101857,
-      "learning_rate": 0.00019999238850827754,
-      "loss": 46.0,
-      "step": 24398
-    },
-    {
-      "epoch": 3.9292644631426388,
-      "grad_norm": 0.00243575731292367,
-      "learning_rate": 0.00019999238788407292,
-      "loss": 46.0,
-      "step": 24399
-    },
-    {
-      "epoch": 3.9294255002214262,
-      "grad_norm": 0.003056165063753724,
-      "learning_rate": 0.0001999923872598427,
-      "loss": 46.0,
-      "step": 24400
-    },
-    {
-      "epoch": 3.9295865373002132,
-      "grad_norm": 0.005382818635553122,
-      "learning_rate": 0.0001999923866355869,
-      "loss": 46.0,
-      "step": 24401
-    },
-    {
-      "epoch": 3.9297475743790007,
-      "grad_norm": 0.0014281562762334943,
-      "learning_rate": 0.00019999238601130549,
-      "loss": 46.0,
-      "step": 24402
-    },
-    {
-      "epoch": 3.929908611457788,
-      "grad_norm": 0.0013499307679012418,
-      "learning_rate": 0.00019999238538699848,
-      "loss": 46.0,
-      "step": 24403
-    },
-    {
-      "epoch": 3.9300696485365756,
-      "grad_norm": 0.0035716991405934095,
-      "learning_rate": 0.00019999238476266592,
-      "loss": 46.0,
-      "step": 24404
-    },
-    {
-      "epoch": 3.930230685615363,
-      "grad_norm": 0.009319575503468513,
-      "learning_rate": 0.00019999238413830772,
-      "loss": 46.0,
-      "step": 24405
-    },
-    {
-      "epoch": 3.9303917226941505,
-      "grad_norm": 0.00225442205555737,
-      "learning_rate": 0.00019999238351392395,
-      "loss": 46.0,
-      "step": 24406
-    },
-    {
-      "epoch": 3.9305527597729375,
-      "grad_norm": 0.0031867618672549725,
-      "learning_rate": 0.00019999238288951457,
-      "loss": 46.0,
-      "step": 24407
-    },
-    {
-      "epoch": 3.930713796851725,
-      "grad_norm": 0.00477884104475379,
-      "learning_rate": 0.0001999923822650796,
-      "loss": 46.0,
-      "step": 24408
-    },
-    {
-      "epoch": 3.9308748339305124,
-      "grad_norm": 0.001682958216406405,
-      "learning_rate": 0.00019999238164061906,
-      "loss": 46.0,
-      "step": 24409
-    },
-    {
-      "epoch": 3.9310358710093,
-      "grad_norm": 0.002937573241069913,
-      "learning_rate": 0.00019999238101613292,
-      "loss": 46.0,
-      "step": 24410
-    },
-    {
-      "epoch": 3.9311969080880873,
-      "grad_norm": 0.004121054895222187,
-      "learning_rate": 0.0001999923803916212,
-      "loss": 46.0,
-      "step": 24411
-    },
-    {
-      "epoch": 3.9313579451668748,
-      "grad_norm": 0.006640677340328693,
-      "learning_rate": 0.00019999237976708385,
-      "loss": 46.0,
-      "step": 24412
-    },
-    {
-      "epoch": 3.931518982245662,
-      "grad_norm": 0.0036491509526968002,
-      "learning_rate": 0.00019999237914252092,
-      "loss": 46.0,
-      "step": 24413
-    },
-    {
-      "epoch": 3.9316800193244497,
-      "grad_norm": 0.0027274528983980417,
-      "learning_rate": 0.0001999923785179324,
-      "loss": 46.0,
-      "step": 24414
-    },
-    {
-      "epoch": 3.931841056403237,
-      "grad_norm": 0.004347385838627815,
-      "learning_rate": 0.00019999237789331827,
-      "loss": 46.0,
-      "step": 24415
-    },
-    {
-      "epoch": 3.932002093482024,
-      "grad_norm": 0.0015313193434849381,
-      "learning_rate": 0.00019999237726867856,
-      "loss": 46.0,
-      "step": 24416
-    },
-    {
-      "epoch": 3.9321631305608116,
-      "grad_norm": 0.027371564880013466,
-      "learning_rate": 0.00019999237664401328,
-      "loss": 46.0,
-      "step": 24417
-    },
-    {
-      "epoch": 3.932324167639599,
-      "grad_norm": 0.0009167072712443769,
-      "learning_rate": 0.0001999923760193224,
-      "loss": 46.0,
-      "step": 24418
-    },
-    {
-      "epoch": 3.9324852047183865,
-      "grad_norm": 0.003430148120969534,
-      "learning_rate": 0.00019999237539460588,
-      "loss": 46.0,
-      "step": 24419
-    },
-    {
-      "epoch": 3.932646241797174,
-      "grad_norm": 0.00242228084243834,
-      "learning_rate": 0.00019999237476986384,
-      "loss": 46.0,
-      "step": 24420
-    },
-    {
-      "epoch": 3.932807278875961,
-      "grad_norm": 0.010304367169737816,
-      "learning_rate": 0.00019999237414509616,
-      "loss": 46.0,
-      "step": 24421
-    },
-    {
-      "epoch": 3.9329683159547484,
-      "grad_norm": 0.006370606832206249,
-      "learning_rate": 0.00019999237352030292,
-      "loss": 46.0,
-      "step": 24422
-    },
-    {
-      "epoch": 3.933129353033536,
-      "grad_norm": 0.0054320283234119415,
-      "learning_rate": 0.00019999237289548404,
-      "loss": 46.0,
-      "step": 24423
-    },
-    {
-      "epoch": 3.9332903901123233,
-      "grad_norm": 0.0010298618581146002,
-      "learning_rate": 0.0001999923722706396,
-      "loss": 46.0,
-      "step": 24424
-    },
-    {
-      "epoch": 3.9334514271911107,
-      "grad_norm": 0.016750281676650047,
-      "learning_rate": 0.00019999237164576954,
-      "loss": 46.0,
-      "step": 24425
-    },
-    {
-      "epoch": 3.933612464269898,
-      "grad_norm": 0.0008130322676151991,
-      "learning_rate": 0.0001999923710208739,
-      "loss": 46.0,
-      "step": 24426
-    },
-    {
-      "epoch": 3.9337735013486856,
-      "grad_norm": 0.001588718849234283,
-      "learning_rate": 0.0001999923703959527,
-      "loss": 46.0,
-      "step": 24427
-    },
-    {
-      "epoch": 3.933934538427473,
-      "grad_norm": 0.006425799336284399,
-      "learning_rate": 0.00019999236977100585,
-      "loss": 46.0,
-      "step": 24428
-    },
-    {
-      "epoch": 3.9340955755062605,
-      "grad_norm": 0.005509865004569292,
-      "learning_rate": 0.00019999236914603345,
-      "loss": 46.0,
-      "step": 24429
-    },
-    {
-      "epoch": 3.9342566125850476,
-      "grad_norm": 0.0029557920061051846,
-      "learning_rate": 0.00019999236852103546,
-      "loss": 46.0,
-      "step": 24430
-    },
-    {
-      "epoch": 3.934417649663835,
-      "grad_norm": 0.0070237042382359505,
-      "learning_rate": 0.00019999236789601185,
-      "loss": 46.0,
-      "step": 24431
-    },
-    {
-      "epoch": 3.9345786867426225,
-      "grad_norm": 0.0018422477878630161,
-      "learning_rate": 0.00019999236727096266,
-      "loss": 46.0,
-      "step": 24432
-    },
-    {
-      "epoch": 3.93473972382141,
-      "grad_norm": 0.008246928453445435,
-      "learning_rate": 0.00019999236664588788,
-      "loss": 46.0,
-      "step": 24433
-    },
-    {
-      "epoch": 3.9349007609001974,
-      "grad_norm": 0.005493798293173313,
-      "learning_rate": 0.0001999923660207875,
-      "loss": 46.0,
-      "step": 24434
-    },
-    {
-      "epoch": 3.935061797978985,
-      "grad_norm": 0.0015698785427957773,
-      "learning_rate": 0.00019999236539566153,
-      "loss": 46.0,
-      "step": 24435
-    },
-    {
-      "epoch": 3.935222835057772,
-      "grad_norm": 0.005580568220466375,
-      "learning_rate": 0.00019999236477050999,
-      "loss": 46.0,
-      "step": 24436
-    },
-    {
-      "epoch": 3.9353838721365593,
-      "grad_norm": 0.006116837728768587,
-      "learning_rate": 0.00019999236414533283,
-      "loss": 46.0,
-      "step": 24437
-    },
-    {
-      "epoch": 3.9355449092153467,
-      "grad_norm": 0.0014182840241119266,
-      "learning_rate": 0.00019999236352013003,
-      "loss": 46.0,
-      "step": 24438
-    },
-    {
-      "epoch": 3.935705946294134,
-      "grad_norm": 0.007212547119706869,
-      "learning_rate": 0.00019999236289490173,
-      "loss": 46.0,
-      "step": 24439
-    },
-    {
-      "epoch": 3.9358669833729216,
-      "grad_norm": 0.002889544004574418,
-      "learning_rate": 0.00019999236226964778,
-      "loss": 46.0,
-      "step": 24440
-    },
-    {
-      "epoch": 3.936028020451709,
-      "grad_norm": 0.0021446554455906153,
-      "learning_rate": 0.00019999236164436825,
-      "loss": 46.0,
-      "step": 24441
-    },
-    {
-      "epoch": 3.9361890575304965,
-      "grad_norm": 0.009180703200399876,
-      "learning_rate": 0.00019999236101906313,
-      "loss": 46.0,
-      "step": 24442
-    },
-    {
-      "epoch": 3.936350094609284,
-      "grad_norm": 0.0017527807503938675,
-      "learning_rate": 0.00019999236039373243,
-      "loss": 46.0,
-      "step": 24443
-    },
-    {
-      "epoch": 3.9365111316880714,
-      "grad_norm": 0.00906195305287838,
-      "learning_rate": 0.0001999923597683761,
-      "loss": 46.0,
-      "step": 24444
-    },
-    {
-      "epoch": 3.9366721687668584,
-      "grad_norm": 0.01186458207666874,
-      "learning_rate": 0.0001999923591429942,
-      "loss": 46.0,
-      "step": 24445
-    },
-    {
-      "epoch": 3.936833205845646,
-      "grad_norm": 0.011776086874306202,
-      "learning_rate": 0.00019999235851758673,
-      "loss": 46.0,
-      "step": 24446
-    },
-    {
-      "epoch": 3.9369942429244333,
-      "grad_norm": 0.0017151961801573634,
-      "learning_rate": 0.00019999235789215365,
-      "loss": 46.0,
-      "step": 24447
-    },
-    {
-      "epoch": 3.937155280003221,
-      "grad_norm": 0.0012179887853562832,
-      "learning_rate": 0.00019999235726669495,
-      "loss": 46.0,
-      "step": 24448
-    },
-    {
-      "epoch": 3.9373163170820082,
-      "grad_norm": 0.006051843985915184,
-      "learning_rate": 0.0001999923566412107,
-      "loss": 46.0,
-      "step": 24449
-    },
-    {
-      "epoch": 3.9374773541607953,
-      "grad_norm": 0.00758137134835124,
-      "learning_rate": 0.00019999235601570083,
-      "loss": 46.0,
-      "step": 24450
-    },
-    {
-      "epoch": 3.9376383912395827,
-      "grad_norm": 0.006233640015125275,
-      "learning_rate": 0.00019999235539016537,
-      "loss": 46.0,
-      "step": 24451
-    },
-    {
-      "epoch": 3.93779942831837,
-      "grad_norm": 0.008672326803207397,
-      "learning_rate": 0.00019999235476460432,
-      "loss": 46.0,
-      "step": 24452
-    },
-    {
-      "epoch": 3.9379604653971576,
-      "grad_norm": 0.005385734140872955,
-      "learning_rate": 0.00019999235413901766,
-      "loss": 46.0,
-      "step": 24453
-    },
-    {
-      "epoch": 3.938121502475945,
-      "grad_norm": 0.0013677530223503709,
-      "learning_rate": 0.00019999235351340544,
-      "loss": 46.0,
-      "step": 24454
-    },
-    {
-      "epoch": 3.9382825395547325,
-      "grad_norm": 0.00402272492647171,
-      "learning_rate": 0.0001999923528877676,
-      "loss": 46.0,
-      "step": 24455
-    },
-    {
-      "epoch": 3.93844357663352,
-      "grad_norm": 0.004740627016872168,
-      "learning_rate": 0.0001999923522621042,
-      "loss": 46.0,
-      "step": 24456
-    },
-    {
-      "epoch": 3.9386046137123074,
-      "grad_norm": 0.009174207225441933,
-      "learning_rate": 0.00019999235163641515,
-      "loss": 46.0,
-      "step": 24457
-    },
-    {
-      "epoch": 3.938765650791095,
-      "grad_norm": 0.014905828982591629,
-      "learning_rate": 0.00019999235101070056,
-      "loss": 46.0,
-      "step": 24458
-    },
-    {
-      "epoch": 3.9389266878698823,
-      "grad_norm": 0.0029241740703582764,
-      "learning_rate": 0.00019999235038496035,
-      "loss": 46.0,
-      "step": 24459
-    },
-    {
-      "epoch": 3.9390877249486693,
-      "grad_norm": 0.00474179582670331,
-      "learning_rate": 0.00019999234975919455,
-      "loss": 46.0,
-      "step": 24460
-    },
-    {
-      "epoch": 3.9392487620274568,
-      "grad_norm": 0.004056048113852739,
-      "learning_rate": 0.00019999234913340317,
-      "loss": 46.0,
-      "step": 24461
-    },
-    {
-      "epoch": 3.9394097991062442,
-      "grad_norm": 0.005026824306696653,
-      "learning_rate": 0.00019999234850758623,
-      "loss": 46.0,
-      "step": 24462
-    },
-    {
-      "epoch": 3.9395708361850317,
-      "grad_norm": 0.0023352550342679024,
-      "learning_rate": 0.00019999234788174364,
-      "loss": 46.0,
-      "step": 24463
-    },
-    {
-      "epoch": 3.939731873263819,
-      "grad_norm": 0.0009777604136615992,
-      "learning_rate": 0.00019999234725587547,
-      "loss": 46.0,
-      "step": 24464
-    },
-    {
-      "epoch": 3.939892910342606,
-      "grad_norm": 0.006058571394532919,
-      "learning_rate": 0.0001999923466299817,
-      "loss": 46.0,
-      "step": 24465
-    },
-    {
-      "epoch": 3.9400539474213936,
-      "grad_norm": 0.004813102073967457,
-      "learning_rate": 0.00019999234600406236,
-      "loss": 46.0,
-      "step": 24466
-    },
-    {
-      "epoch": 3.940214984500181,
-      "grad_norm": 0.0020553087815642357,
-      "learning_rate": 0.00019999234537811743,
-      "loss": 46.0,
-      "step": 24467
-    },
-    {
-      "epoch": 3.9403760215789685,
-      "grad_norm": 0.0022767093032598495,
-      "learning_rate": 0.00019999234475214688,
-      "loss": 46.0,
-      "step": 24468
-    },
-    {
-      "epoch": 3.940537058657756,
-      "grad_norm": 0.012391808442771435,
-      "learning_rate": 0.00019999234412615074,
-      "loss": 46.0,
-      "step": 24469
-    },
-    {
-      "epoch": 3.9406980957365434,
-      "grad_norm": 0.005296351388096809,
-      "learning_rate": 0.00019999234350012902,
-      "loss": 46.0,
-      "step": 24470
-    },
-    {
-      "epoch": 3.940859132815331,
-      "grad_norm": 0.003683160524815321,
-      "learning_rate": 0.0001999923428740817,
-      "loss": 46.0,
-      "step": 24471
-    },
-    {
-      "epoch": 3.9410201698941183,
-      "grad_norm": 0.0030060908757150173,
-      "learning_rate": 0.0001999923422480088,
-      "loss": 46.0,
-      "step": 24472
-    },
-    {
-      "epoch": 3.9411812069729057,
-      "grad_norm": 0.003344297409057617,
-      "learning_rate": 0.0001999923416219103,
-      "loss": 46.0,
-      "step": 24473
-    },
-    {
-      "epoch": 3.9413422440516928,
-      "grad_norm": 0.005463795270770788,
-      "learning_rate": 0.0001999923409957862,
-      "loss": 46.0,
-      "step": 24474
-    },
-    {
-      "epoch": 3.94150328113048,
-      "grad_norm": 0.006249964702874422,
-      "learning_rate": 0.00019999234036963652,
-      "loss": 46.0,
-      "step": 24475
-    },
-    {
-      "epoch": 3.9416643182092677,
-      "grad_norm": 0.002506648190319538,
-      "learning_rate": 0.00019999233974346122,
-      "loss": 46.0,
-      "step": 24476
-    },
-    {
-      "epoch": 3.941825355288055,
-      "grad_norm": 0.0032582751009613276,
-      "learning_rate": 0.00019999233911726036,
-      "loss": 46.0,
-      "step": 24477
-    },
-    {
-      "epoch": 3.9419863923668426,
-      "grad_norm": 0.004647792316973209,
-      "learning_rate": 0.00019999233849103391,
-      "loss": 46.0,
-      "step": 24478
-    },
-    {
-      "epoch": 3.94214742944563,
-      "grad_norm": 0.009381710551679134,
-      "learning_rate": 0.00019999233786478185,
-      "loss": 46.0,
-      "step": 24479
-    },
-    {
-      "epoch": 3.942308466524417,
-      "grad_norm": 0.006575780920684338,
-      "learning_rate": 0.00019999233723850418,
-      "loss": 46.0,
-      "step": 24480
-    },
-    {
-      "epoch": 3.9424695036032045,
-      "grad_norm": 0.0029790003318339586,
-      "learning_rate": 0.00019999233661220097,
-      "loss": 46.0,
-      "step": 24481
-    },
-    {
-      "epoch": 3.942630540681992,
-      "grad_norm": 0.013919372111558914,
-      "learning_rate": 0.0001999923359858721,
-      "loss": 46.0,
-      "step": 24482
-    },
-    {
-      "epoch": 3.9427915777607794,
-      "grad_norm": 0.011686764657497406,
-      "learning_rate": 0.00019999233535951768,
-      "loss": 46.0,
-      "step": 24483
-    },
-    {
-      "epoch": 3.942952614839567,
-      "grad_norm": 0.008118031546473503,
-      "learning_rate": 0.00019999233473313766,
-      "loss": 46.0,
-      "step": 24484
-    },
-    {
-      "epoch": 3.9431136519183543,
-      "grad_norm": 0.00322963809594512,
-      "learning_rate": 0.00019999233410673205,
-      "loss": 46.0,
-      "step": 24485
-    },
-    {
-      "epoch": 3.9432746889971417,
-      "grad_norm": 0.003362247720360756,
-      "learning_rate": 0.00019999233348030085,
-      "loss": 46.0,
-      "step": 24486
-    },
-    {
-      "epoch": 3.943435726075929,
-      "grad_norm": 0.006708750035613775,
-      "learning_rate": 0.00019999233285384403,
-      "loss": 46.0,
-      "step": 24487
-    },
-    {
-      "epoch": 3.9435967631547166,
-      "grad_norm": 0.005984371528029442,
-      "learning_rate": 0.00019999233222736163,
-      "loss": 46.0,
-      "step": 24488
-    },
-    {
-      "epoch": 3.9437578002335036,
-      "grad_norm": 0.003902870463207364,
-      "learning_rate": 0.00019999233160085365,
-      "loss": 46.0,
-      "step": 24489
-    },
-    {
-      "epoch": 3.943918837312291,
-      "grad_norm": 0.005402877926826477,
-      "learning_rate": 0.00019999233097432007,
-      "loss": 46.0,
-      "step": 24490
-    },
-    {
-      "epoch": 3.9440798743910785,
-      "grad_norm": 0.004179893992841244,
-      "learning_rate": 0.00019999233034776088,
-      "loss": 46.0,
-      "step": 24491
-    },
-    {
-      "epoch": 3.944240911469866,
-      "grad_norm": 0.0032317708246409893,
-      "learning_rate": 0.00019999232972117614,
-      "loss": 46.0,
-      "step": 24492
-    },
-    {
-      "epoch": 3.9444019485486534,
-      "grad_norm": 0.003148757154121995,
-      "learning_rate": 0.00019999232909456577,
-      "loss": 46.0,
-      "step": 24493
-    },
-    {
-      "epoch": 3.9445629856274405,
-      "grad_norm": 0.001753403339534998,
-      "learning_rate": 0.00019999232846792982,
-      "loss": 46.0,
-      "step": 24494
-    },
-    {
-      "epoch": 3.944724022706228,
-      "grad_norm": 0.004594752565026283,
-      "learning_rate": 0.00019999232784126826,
-      "loss": 46.0,
-      "step": 24495
-    },
-    {
-      "epoch": 3.9448850597850154,
-      "grad_norm": 0.003652523271739483,
-      "learning_rate": 0.00019999232721458113,
-      "loss": 46.0,
-      "step": 24496
-    },
-    {
-      "epoch": 3.945046096863803,
-      "grad_norm": 0.0020665600895881653,
-      "learning_rate": 0.0001999923265878684,
-      "loss": 46.0,
-      "step": 24497
-    },
-    {
-      "epoch": 3.9452071339425903,
-      "grad_norm": 0.00592459924519062,
-      "learning_rate": 0.0001999923259611301,
-      "loss": 46.0,
-      "step": 24498
-    },
-    {
-      "epoch": 3.9453681710213777,
-      "grad_norm": 0.001933505991473794,
-      "learning_rate": 0.00019999232533436616,
-      "loss": 46.0,
-      "step": 24499
-    },
-    {
-      "epoch": 3.945529208100165,
-      "grad_norm": 0.0006409965571947396,
-      "learning_rate": 0.00019999232470757666,
-      "loss": 46.0,
-      "step": 24500
-    },
-    {
-      "epoch": 3.9456902451789526,
-      "grad_norm": 0.001832569600082934,
-      "learning_rate": 0.00019999232408076157,
-      "loss": 46.0,
-      "step": 24501
-    },
-    {
-      "epoch": 3.94585128225774,
-      "grad_norm": 0.004537877626717091,
-      "learning_rate": 0.00019999232345392084,
-      "loss": 46.0,
-      "step": 24502
-    },
-    {
-      "epoch": 3.946012319336527,
-      "grad_norm": 0.0005695093423128128,
-      "learning_rate": 0.00019999232282705455,
-      "loss": 46.0,
-      "step": 24503
-    },
-    {
-      "epoch": 3.9461733564153145,
-      "grad_norm": 0.0008433074108324945,
-      "learning_rate": 0.00019999232220016267,
-      "loss": 46.0,
-      "step": 24504
-    },
-    {
-      "epoch": 3.946334393494102,
-      "grad_norm": 0.0032459928188472986,
-      "learning_rate": 0.0001999923215732452,
-      "loss": 46.0,
-      "step": 24505
-    },
-    {
-      "epoch": 3.9464954305728894,
-      "grad_norm": 0.0019439419265836477,
-      "learning_rate": 0.00019999232094630213,
-      "loss": 46.0,
-      "step": 24506
-    },
-    {
-      "epoch": 3.946656467651677,
-      "grad_norm": 0.0017491135513409972,
-      "learning_rate": 0.00019999232031933347,
-      "loss": 46.0,
-      "step": 24507
-    },
-    {
-      "epoch": 3.9468175047304643,
-      "grad_norm": 0.005312608554959297,
-      "learning_rate": 0.00019999231969233922,
-      "loss": 46.0,
-      "step": 24508
-    },
-    {
-      "epoch": 3.9469785418092513,
-      "grad_norm": 0.0019482218194752932,
-      "learning_rate": 0.00019999231906531938,
-      "loss": 46.0,
-      "step": 24509
-    },
-    {
-      "epoch": 3.947139578888039,
-      "grad_norm": 0.0024019565898925066,
-      "learning_rate": 0.00019999231843827392,
-      "loss": 46.0,
-      "step": 24510
-    },
-    {
-      "epoch": 3.9473006159668262,
-      "grad_norm": 0.003191335592418909,
-      "learning_rate": 0.0001999923178112029,
-      "loss": 46.0,
-      "step": 24511
-    },
-    {
-      "epoch": 3.9474616530456137,
-      "grad_norm": 0.010148586705327034,
-      "learning_rate": 0.00019999231718410628,
-      "loss": 46.0,
-      "step": 24512
-    },
-    {
-      "epoch": 3.947622690124401,
-      "grad_norm": 0.005791592411696911,
-      "learning_rate": 0.00019999231655698407,
-      "loss": 46.0,
-      "step": 24513
-    },
-    {
-      "epoch": 3.9477837272031886,
-      "grad_norm": 0.02290402166545391,
-      "learning_rate": 0.00019999231592983624,
-      "loss": 46.0,
-      "step": 24514
-    },
-    {
-      "epoch": 3.947944764281976,
-      "grad_norm": 0.007506233640015125,
-      "learning_rate": 0.00019999231530266288,
-      "loss": 46.0,
-      "step": 24515
-    },
-    {
-      "epoch": 3.9481058013607635,
-      "grad_norm": 0.00459651043638587,
-      "learning_rate": 0.00019999231467546387,
-      "loss": 46.0,
-      "step": 24516
-    },
-    {
-      "epoch": 3.948266838439551,
-      "grad_norm": 0.013507161289453506,
-      "learning_rate": 0.00019999231404823928,
-      "loss": 46.0,
-      "step": 24517
-    },
-    {
-      "epoch": 3.948427875518338,
-      "grad_norm": 0.01366760116070509,
-      "learning_rate": 0.00019999231342098908,
-      "loss": 46.0,
-      "step": 24518
-    },
-    {
-      "epoch": 3.9485889125971254,
-      "grad_norm": 0.0063840365037322044,
-      "learning_rate": 0.0001999923127937133,
-      "loss": 46.0,
-      "step": 24519
-    },
-    {
-      "epoch": 3.948749949675913,
-      "grad_norm": 0.001921529765240848,
-      "learning_rate": 0.00019999231216641193,
-      "loss": 46.0,
-      "step": 24520
-    },
-    {
-      "epoch": 3.9489109867547003,
-      "grad_norm": 0.0056505585089325905,
-      "learning_rate": 0.00019999231153908497,
-      "loss": 46.0,
-      "step": 24521
-    },
-    {
-      "epoch": 3.9490720238334878,
-      "grad_norm": 0.005920773837715387,
-      "learning_rate": 0.00019999231091173244,
-      "loss": 46.0,
-      "step": 24522
-    },
-    {
-      "epoch": 3.949233060912275,
-      "grad_norm": 0.0006380933336913586,
-      "learning_rate": 0.0001999923102843543,
-      "loss": 46.0,
-      "step": 24523
-    },
-    {
-      "epoch": 3.9493940979910622,
-      "grad_norm": 0.005779617466032505,
-      "learning_rate": 0.00019999230965695055,
-      "loss": 46.0,
-      "step": 24524
-    },
-    {
-      "epoch": 3.9495551350698497,
-      "grad_norm": 0.0075210267677903175,
-      "learning_rate": 0.00019999230902952123,
-      "loss": 46.0,
-      "step": 24525
-    },
-    {
-      "epoch": 3.949716172148637,
-      "grad_norm": 0.0014395912876352668,
-      "learning_rate": 0.00019999230840206627,
-      "loss": 46.0,
-      "step": 24526
-    },
-    {
-      "epoch": 3.9498772092274246,
-      "grad_norm": 0.002967552747577429,
-      "learning_rate": 0.00019999230777458578,
-      "loss": 46.0,
-      "step": 24527
-    },
-    {
-      "epoch": 3.950038246306212,
-      "grad_norm": 0.0018580231117084622,
-      "learning_rate": 0.00019999230714707965,
-      "loss": 46.0,
-      "step": 24528
-    },
-    {
-      "epoch": 3.9501992833849995,
-      "grad_norm": 0.003680624533444643,
-      "learning_rate": 0.00019999230651954796,
-      "loss": 46.0,
-      "step": 24529
-    },
-    {
-      "epoch": 3.950360320463787,
-      "grad_norm": 0.004533518571406603,
-      "learning_rate": 0.00019999230589199068,
-      "loss": 46.0,
-      "step": 24530
-    },
-    {
-      "epoch": 3.9505213575425744,
-      "grad_norm": 0.0060940757393836975,
-      "learning_rate": 0.00019999230526440776,
-      "loss": 46.0,
-      "step": 24531
-    },
-    {
-      "epoch": 3.950682394621362,
-      "grad_norm": 0.0031006347853690386,
-      "learning_rate": 0.0001999923046367993,
-      "loss": 46.0,
-      "step": 24532
-    },
-    {
-      "epoch": 3.950843431700149,
-      "grad_norm": 0.0012043003225699067,
-      "learning_rate": 0.00019999230400916522,
-      "loss": 46.0,
-      "step": 24533
-    },
-    {
-      "epoch": 3.9510044687789363,
-      "grad_norm": 0.003294929163530469,
-      "learning_rate": 0.00019999230338150554,
-      "loss": 46.0,
-      "step": 24534
-    },
-    {
-      "epoch": 3.9511655058577237,
-      "grad_norm": 0.006164142396301031,
-      "learning_rate": 0.00019999230275382027,
-      "loss": 46.0,
-      "step": 24535
-    },
-    {
-      "epoch": 3.951326542936511,
-      "grad_norm": 0.008812782354652882,
-      "learning_rate": 0.00019999230212610941,
-      "loss": 46.0,
-      "step": 24536
-    },
-    {
-      "epoch": 3.9514875800152987,
-      "grad_norm": 0.007229702081531286,
-      "learning_rate": 0.00019999230149837297,
-      "loss": 46.0,
-      "step": 24537
-    },
-    {
-      "epoch": 3.9516486170940857,
-      "grad_norm": 0.014134583063423634,
-      "learning_rate": 0.00019999230087061094,
-      "loss": 46.0,
-      "step": 24538
-    },
-    {
-      "epoch": 3.951809654172873,
-      "grad_norm": 0.0012406106106936932,
-      "learning_rate": 0.00019999230024282333,
-      "loss": 46.0,
-      "step": 24539
-    },
-    {
-      "epoch": 3.9519706912516606,
-      "grad_norm": 0.000965129176620394,
-      "learning_rate": 0.0001999922996150101,
-      "loss": 46.0,
-      "step": 24540
-    },
-    {
-      "epoch": 3.952131728330448,
-      "grad_norm": 0.0019423657795414329,
-      "learning_rate": 0.00019999229898717128,
-      "loss": 46.0,
-      "step": 24541
-    },
-    {
-      "epoch": 3.9522927654092355,
-      "grad_norm": 0.0010240275878459215,
-      "learning_rate": 0.00019999229835930687,
-      "loss": 46.0,
-      "step": 24542
-    },
-    {
-      "epoch": 3.952453802488023,
-      "grad_norm": 0.006287777796387672,
-      "learning_rate": 0.00019999229773141685,
-      "loss": 46.0,
-      "step": 24543
-    },
-    {
-      "epoch": 3.9526148395668104,
-      "grad_norm": 0.004004526883363724,
-      "learning_rate": 0.00019999229710350124,
-      "loss": 46.0,
-      "step": 24544
-    },
-    {
-      "epoch": 3.952775876645598,
-      "grad_norm": 0.004683574661612511,
-      "learning_rate": 0.00019999229647556005,
-      "loss": 46.0,
-      "step": 24545
-    },
-    {
-      "epoch": 3.9529369137243853,
-      "grad_norm": 0.005028977990150452,
-      "learning_rate": 0.00019999229584759327,
-      "loss": 46.0,
-      "step": 24546
-    },
-    {
-      "epoch": 3.9530979508031723,
-      "grad_norm": 0.0026760261971503496,
-      "learning_rate": 0.0001999922952196009,
-      "loss": 46.0,
-      "step": 24547
-    },
-    {
-      "epoch": 3.9532589878819597,
-      "grad_norm": 0.005508644971996546,
-      "learning_rate": 0.00019999229459158292,
-      "loss": 46.0,
-      "step": 24548
-    },
-    {
-      "epoch": 3.953420024960747,
-      "grad_norm": 0.0029298197478055954,
-      "learning_rate": 0.00019999229396353938,
-      "loss": 46.0,
-      "step": 24549
-    },
-    {
-      "epoch": 3.9535810620395346,
-      "grad_norm": 0.0018909123027697206,
-      "learning_rate": 0.00019999229333547022,
-      "loss": 46.0,
-      "step": 24550
-    },
-    {
-      "epoch": 3.953742099118322,
-      "grad_norm": 0.009362361393868923,
-      "learning_rate": 0.00019999229270737547,
-      "loss": 46.0,
-      "step": 24551
-    },
-    {
-      "epoch": 3.9539031361971095,
-      "grad_norm": 0.003478787373751402,
-      "learning_rate": 0.00019999229207925514,
-      "loss": 46.0,
-      "step": 24552
-    },
-    {
-      "epoch": 3.9540641732758965,
-      "grad_norm": 0.004581904038786888,
-      "learning_rate": 0.0001999922914511092,
-      "loss": 46.0,
-      "step": 24553
-    },
-    {
-      "epoch": 3.954225210354684,
-      "grad_norm": 0.00929262489080429,
-      "learning_rate": 0.00019999229082293766,
-      "loss": 46.0,
-      "step": 24554
-    },
-    {
-      "epoch": 3.9543862474334714,
-      "grad_norm": 0.004574401304125786,
-      "learning_rate": 0.00019999229019474054,
-      "loss": 46.0,
-      "step": 24555
-    },
-    {
-      "epoch": 3.954547284512259,
-      "grad_norm": 0.0038825373630970716,
-      "learning_rate": 0.00019999228956651784,
-      "loss": 46.0,
-      "step": 24556
-    },
-    {
-      "epoch": 3.9547083215910463,
-      "grad_norm": 0.002296898514032364,
-      "learning_rate": 0.00019999228893826954,
-      "loss": 46.0,
-      "step": 24557
-    },
-    {
-      "epoch": 3.954869358669834,
-      "grad_norm": 0.0063306717202067375,
-      "learning_rate": 0.00019999228830999563,
-      "loss": 46.0,
-      "step": 24558
-    },
-    {
-      "epoch": 3.9550303957486213,
-      "grad_norm": 0.004092125222086906,
-      "learning_rate": 0.00019999228768169617,
-      "loss": 46.0,
-      "step": 24559
-    },
-    {
-      "epoch": 3.9551914328274087,
-      "grad_norm": 0.000549173797480762,
-      "learning_rate": 0.00019999228705337108,
-      "loss": 46.0,
-      "step": 24560
-    },
-    {
-      "epoch": 3.955352469906196,
-      "grad_norm": 0.00241836067289114,
-      "learning_rate": 0.00019999228642502036,
-      "loss": 46.0,
-      "step": 24561
-    },
-    {
-      "epoch": 3.955513506984983,
-      "grad_norm": 0.002404788276180625,
-      "learning_rate": 0.00019999228579664413,
-      "loss": 46.0,
-      "step": 24562
-    },
-    {
-      "epoch": 3.9556745440637706,
-      "grad_norm": 0.001841920311562717,
-      "learning_rate": 0.00019999228516824226,
-      "loss": 46.0,
-      "step": 24563
-    },
-    {
-      "epoch": 3.955835581142558,
-      "grad_norm": 0.0011428429279476404,
-      "learning_rate": 0.0001999922845398148,
-      "loss": 46.0,
-      "step": 24564
-    },
-    {
-      "epoch": 3.9559966182213455,
-      "grad_norm": 0.001579323667101562,
-      "learning_rate": 0.00019999228391136175,
-      "loss": 46.0,
-      "step": 24565
-    },
-    {
-      "epoch": 3.956157655300133,
-      "grad_norm": 0.004542314447462559,
-      "learning_rate": 0.00019999228328288312,
-      "loss": 46.0,
-      "step": 24566
-    },
-    {
-      "epoch": 3.95631869237892,
-      "grad_norm": 0.00798245519399643,
-      "learning_rate": 0.00019999228265437887,
-      "loss": 46.0,
-      "step": 24567
-    },
-    {
-      "epoch": 3.9564797294577074,
-      "grad_norm": 0.003039659233763814,
-      "learning_rate": 0.00019999228202584904,
-      "loss": 46.0,
-      "step": 24568
-    },
-    {
-      "epoch": 3.956640766536495,
-      "grad_norm": 0.0036243193317204714,
-      "learning_rate": 0.00019999228139729362,
-      "loss": 46.0,
-      "step": 24569
-    },
-    {
-      "epoch": 3.9568018036152823,
-      "grad_norm": 0.003200292820110917,
-      "learning_rate": 0.0001999922807687126,
-      "loss": 46.0,
-      "step": 24570
-    },
-    {
-      "epoch": 3.95696284069407,
-      "grad_norm": 0.003493676660582423,
-      "learning_rate": 0.000199992280140106,
-      "loss": 46.0,
-      "step": 24571
-    },
-    {
-      "epoch": 3.9571238777728572,
-      "grad_norm": 0.006319560110569,
-      "learning_rate": 0.0001999922795114738,
-      "loss": 46.0,
-      "step": 24572
-    },
-    {
-      "epoch": 3.9572849148516447,
-      "grad_norm": 0.0011464391136541963,
-      "learning_rate": 0.00019999227888281603,
-      "loss": 46.0,
-      "step": 24573
-    },
-    {
-      "epoch": 3.957445951930432,
-      "grad_norm": 0.0019397110445424914,
-      "learning_rate": 0.00019999227825413262,
-      "loss": 46.0,
-      "step": 24574
-    },
-    {
-      "epoch": 3.9576069890092196,
-      "grad_norm": 0.0020150644704699516,
-      "learning_rate": 0.00019999227762542367,
-      "loss": 46.0,
-      "step": 24575
-    },
-    {
-      "epoch": 3.957768026088007,
-      "grad_norm": 0.006844239309430122,
-      "learning_rate": 0.0001999922769966891,
-      "loss": 46.0,
-      "step": 24576
-    },
-    {
-      "epoch": 3.957929063166794,
-      "grad_norm": 0.0008079410181380808,
-      "learning_rate": 0.00019999227636792891,
-      "loss": 46.0,
-      "step": 24577
-    },
-    {
-      "epoch": 3.9580901002455815,
-      "grad_norm": 0.005341343581676483,
-      "learning_rate": 0.00019999227573914315,
-      "loss": 46.0,
-      "step": 24578
-    },
-    {
-      "epoch": 3.958251137324369,
-      "grad_norm": 0.007375328801572323,
-      "learning_rate": 0.0001999922751103318,
-      "loss": 46.0,
-      "step": 24579
-    },
-    {
-      "epoch": 3.9584121744031564,
-      "grad_norm": 0.0037683420814573765,
-      "learning_rate": 0.00019999227448149487,
-      "loss": 46.0,
-      "step": 24580
-    },
-    {
-      "epoch": 3.958573211481944,
-      "grad_norm": 0.0030840453691780567,
-      "learning_rate": 0.00019999227385263235,
-      "loss": 46.0,
-      "step": 24581
-    },
-    {
-      "epoch": 3.958734248560731,
-      "grad_norm": 0.0016249476466327906,
-      "learning_rate": 0.0001999922732237442,
-      "loss": 46.0,
-      "step": 24582
-    },
-    {
-      "epoch": 3.9588952856395183,
-      "grad_norm": 0.0052049485966563225,
-      "learning_rate": 0.0001999922725948305,
-      "loss": 46.0,
-      "step": 24583
-    },
-    {
-      "epoch": 3.9590563227183058,
-      "grad_norm": 0.003728102194145322,
-      "learning_rate": 0.00019999227196589118,
-      "loss": 46.0,
-      "step": 24584
-    },
-    {
-      "epoch": 3.959217359797093,
-      "grad_norm": 0.008609190583229065,
-      "learning_rate": 0.00019999227133692625,
-      "loss": 46.0,
-      "step": 24585
-    },
-    {
-      "epoch": 3.9593783968758807,
-      "grad_norm": 0.0010269450722262263,
-      "learning_rate": 0.00019999227070793577,
-      "loss": 46.0,
-      "step": 24586
-    },
-    {
-      "epoch": 3.959539433954668,
-      "grad_norm": 0.017943529412150383,
-      "learning_rate": 0.00019999227007891967,
-      "loss": 46.0,
-      "step": 24587
-    },
-    {
-      "epoch": 3.9597004710334556,
-      "grad_norm": 0.003252143505960703,
-      "learning_rate": 0.000199992269449878,
-      "loss": 46.0,
-      "step": 24588
-    },
-    {
-      "epoch": 3.959861508112243,
-      "grad_norm": 0.01167239248752594,
-      "learning_rate": 0.00019999226882081074,
-      "loss": 46.0,
-      "step": 24589
-    },
-    {
-      "epoch": 3.9600225451910305,
-      "grad_norm": 0.009695515036582947,
-      "learning_rate": 0.00019999226819171785,
-      "loss": 46.0,
-      "step": 24590
-    },
-    {
-      "epoch": 3.9601835822698175,
-      "grad_norm": 0.0008362962398678064,
-      "learning_rate": 0.0001999922675625994,
-      "loss": 46.0,
-      "step": 24591
-    },
-    {
-      "epoch": 3.960344619348605,
-      "grad_norm": 0.0036508759949356318,
-      "learning_rate": 0.00019999226693345534,
-      "loss": 46.0,
-      "step": 24592
-    },
-    {
-      "epoch": 3.9605056564273924,
-      "grad_norm": 0.003990464378148317,
-      "learning_rate": 0.0001999922663042857,
-      "loss": 46.0,
-      "step": 24593
-    },
-    {
-      "epoch": 3.96066669350618,
-      "grad_norm": 0.003227963810786605,
-      "learning_rate": 0.00019999226567509045,
-      "loss": 46.0,
-      "step": 24594
-    },
-    {
-      "epoch": 3.9608277305849673,
-      "grad_norm": 0.00426216097548604,
-      "learning_rate": 0.0001999922650458696,
-      "loss": 46.0,
-      "step": 24595
-    },
-    {
-      "epoch": 3.9609887676637547,
-      "grad_norm": 0.0027961486484855413,
-      "learning_rate": 0.00019999226441662316,
-      "loss": 46.0,
-      "step": 24596
-    },
-    {
-      "epoch": 3.9611498047425417,
-      "grad_norm": 0.000826875155325979,
-      "learning_rate": 0.00019999226378735117,
-      "loss": 46.0,
-      "step": 24597
-    },
-    {
-      "epoch": 3.961310841821329,
-      "grad_norm": 0.01634635031223297,
-      "learning_rate": 0.00019999226315805355,
-      "loss": 46.0,
-      "step": 24598
-    },
-    {
-      "epoch": 3.9614718789001166,
-      "grad_norm": 0.003508588531985879,
-      "learning_rate": 0.00019999226252873035,
-      "loss": 46.0,
-      "step": 24599
-    },
-    {
-      "epoch": 3.961632915978904,
-      "grad_norm": 0.0032635731622576714,
-      "learning_rate": 0.00019999226189938154,
-      "loss": 46.0,
-      "step": 24600
-    },
-    {
-      "epoch": 3.9617939530576916,
-      "grad_norm": 0.004523720126599073,
-      "learning_rate": 0.00019999226127000714,
-      "loss": 46.0,
-      "step": 24601
-    },
-    {
-      "epoch": 3.961954990136479,
-      "grad_norm": 0.0037904521450400352,
-      "learning_rate": 0.00019999226064060715,
-      "loss": 46.0,
-      "step": 24602
-    },
-    {
-      "epoch": 3.9621160272152665,
-      "grad_norm": 0.003829203313216567,
-      "learning_rate": 0.00019999226001118158,
-      "loss": 46.0,
-      "step": 24603
-    },
-    {
-      "epoch": 3.962277064294054,
-      "grad_norm": 0.015329036861658096,
-      "learning_rate": 0.00019999225938173041,
-      "loss": 46.0,
-      "step": 24604
-    },
-    {
-      "epoch": 3.9624381013728414,
-      "grad_norm": 0.001176963560283184,
-      "learning_rate": 0.00019999225875225364,
-      "loss": 46.0,
-      "step": 24605
-    },
-    {
-      "epoch": 3.9625991384516284,
-      "grad_norm": 0.008821663446724415,
-      "learning_rate": 0.0001999922581227513,
-      "loss": 46.0,
-      "step": 24606
-    },
-    {
-      "epoch": 3.962760175530416,
-      "grad_norm": 0.006941392086446285,
-      "learning_rate": 0.00019999225749322335,
-      "loss": 46.0,
-      "step": 24607
-    },
-    {
-      "epoch": 3.9629212126092033,
-      "grad_norm": 0.005642466712743044,
-      "learning_rate": 0.0001999922568636698,
-      "loss": 46.0,
-      "step": 24608
-    },
-    {
-      "epoch": 3.9630822496879907,
-      "grad_norm": 0.0016563436947762966,
-      "learning_rate": 0.00019999225623409066,
-      "loss": 46.0,
-      "step": 24609
-    },
-    {
-      "epoch": 3.963243286766778,
-      "grad_norm": 0.004143841564655304,
-      "learning_rate": 0.00019999225560448595,
-      "loss": 46.0,
-      "step": 24610
-    },
-    {
-      "epoch": 3.963404323845565,
-      "grad_norm": 0.008736458607017994,
-      "learning_rate": 0.00019999225497485562,
-      "loss": 46.0,
-      "step": 24611
-    },
-    {
-      "epoch": 3.9635653609243526,
-      "grad_norm": 0.005676708649843931,
-      "learning_rate": 0.0001999922543451997,
-      "loss": 46.0,
-      "step": 24612
-    },
-    {
-      "epoch": 3.96372639800314,
-      "grad_norm": 0.003004975849762559,
-      "learning_rate": 0.0001999922537155182,
-      "loss": 46.0,
-      "step": 24613
-    },
-    {
-      "epoch": 3.9638874350819275,
-      "grad_norm": 0.005604591220617294,
-      "learning_rate": 0.00019999225308581112,
-      "loss": 46.0,
-      "step": 24614
-    },
-    {
-      "epoch": 3.964048472160715,
-      "grad_norm": 0.011630786582827568,
-      "learning_rate": 0.0001999922524560784,
-      "loss": 46.0,
-      "step": 24615
-    },
-    {
-      "epoch": 3.9642095092395024,
-      "grad_norm": 0.00235688965767622,
-      "learning_rate": 0.0001999922518263201,
-      "loss": 46.0,
-      "step": 24616
-    },
-    {
-      "epoch": 3.96437054631829,
-      "grad_norm": 0.0023954608477652073,
-      "learning_rate": 0.00019999225119653625,
-      "loss": 46.0,
-      "step": 24617
-    },
-    {
-      "epoch": 3.9645315833970773,
-      "grad_norm": 0.002856738632544875,
-      "learning_rate": 0.00019999225056672676,
-      "loss": 46.0,
-      "step": 24618
-    },
-    {
-      "epoch": 3.964692620475865,
-      "grad_norm": 0.0027815313078463078,
-      "learning_rate": 0.0001999922499368917,
-      "loss": 46.0,
-      "step": 24619
-    },
-    {
-      "epoch": 3.964853657554652,
-      "grad_norm": 0.008760727010667324,
-      "learning_rate": 0.00019999224930703107,
-      "loss": 46.0,
-      "step": 24620
-    },
-    {
-      "epoch": 3.9650146946334393,
-      "grad_norm": 0.002955361269414425,
-      "learning_rate": 0.00019999224867714479,
-      "loss": 46.0,
-      "step": 24621
-    },
-    {
-      "epoch": 3.9651757317122267,
-      "grad_norm": 0.0180122759193182,
-      "learning_rate": 0.00019999224804723295,
-      "loss": 46.0,
-      "step": 24622
-    },
-    {
-      "epoch": 3.965336768791014,
-      "grad_norm": 0.002538575790822506,
-      "learning_rate": 0.00019999224741729552,
-      "loss": 46.0,
-      "step": 24623
-    },
-    {
-      "epoch": 3.9654978058698016,
-      "grad_norm": 0.0071937888860702515,
-      "learning_rate": 0.00019999224678733248,
-      "loss": 46.0,
-      "step": 24624
-    },
-    {
-      "epoch": 3.965658842948589,
-      "grad_norm": 0.009937465190887451,
-      "learning_rate": 0.00019999224615734388,
-      "loss": 46.0,
-      "step": 24625
-    },
-    {
-      "epoch": 3.965819880027376,
-      "grad_norm": 0.01640373282134533,
-      "learning_rate": 0.00019999224552732966,
-      "loss": 46.0,
-      "step": 24626
-    },
-    {
-      "epoch": 3.9659809171061635,
-      "grad_norm": 0.005190275143831968,
-      "learning_rate": 0.00019999224489728986,
-      "loss": 46.0,
-      "step": 24627
-    },
-    {
-      "epoch": 3.966141954184951,
-      "grad_norm": 0.0014798413030803204,
-      "learning_rate": 0.00019999224426722444,
-      "loss": 46.0,
-      "step": 24628
-    },
-    {
-      "epoch": 3.9663029912637384,
-      "grad_norm": 0.010416446253657341,
-      "learning_rate": 0.00019999224363713344,
-      "loss": 46.0,
-      "step": 24629
-    },
-    {
-      "epoch": 3.966464028342526,
-      "grad_norm": 0.00212200079113245,
-      "learning_rate": 0.00019999224300701687,
-      "loss": 46.0,
-      "step": 24630
-    },
-    {
-      "epoch": 3.9666250654213133,
-      "grad_norm": 0.005232592113316059,
-      "learning_rate": 0.0001999922423768747,
-      "loss": 46.0,
-      "step": 24631
-    },
-    {
-      "epoch": 3.9667861025001008,
-      "grad_norm": 0.002631609793752432,
-      "learning_rate": 0.00019999224174670693,
-      "loss": 46.0,
-      "step": 24632
-    },
-    {
-      "epoch": 3.9669471395788882,
-      "grad_norm": 0.0026403837837278843,
-      "learning_rate": 0.00019999224111651355,
-      "loss": 46.0,
-      "step": 24633
-    },
-    {
-      "epoch": 3.9671081766576757,
-      "grad_norm": 0.0011286140652373433,
-      "learning_rate": 0.0001999922404862946,
-      "loss": 46.0,
-      "step": 24634
-    },
-    {
-      "epoch": 3.9672692137364627,
-      "grad_norm": 0.0037339525297284126,
-      "learning_rate": 0.00019999223985605005,
-      "loss": 46.0,
-      "step": 24635
-    },
-    {
-      "epoch": 3.96743025081525,
-      "grad_norm": 0.0016117809573188424,
-      "learning_rate": 0.0001999922392257799,
-      "loss": 46.0,
-      "step": 24636
-    },
-    {
-      "epoch": 3.9675912878940376,
-      "grad_norm": 0.003521806327626109,
-      "learning_rate": 0.00019999223859548418,
-      "loss": 46.0,
-      "step": 24637
-    },
-    {
-      "epoch": 3.967752324972825,
-      "grad_norm": 0.002333017997443676,
-      "learning_rate": 0.00019999223796516284,
-      "loss": 46.0,
-      "step": 24638
-    },
-    {
-      "epoch": 3.9679133620516125,
-      "grad_norm": 0.0025561561342328787,
-      "learning_rate": 0.0001999922373348159,
-      "loss": 46.0,
-      "step": 24639
-    },
-    {
-      "epoch": 3.9680743991303995,
-      "grad_norm": 0.0010061677312478423,
-      "learning_rate": 0.0001999922367044434,
-      "loss": 46.0,
-      "step": 24640
-    },
-    {
-      "epoch": 3.968235436209187,
-      "grad_norm": 0.011254752986133099,
-      "learning_rate": 0.0001999922360740453,
-      "loss": 46.0,
-      "step": 24641
-    },
-    {
-      "epoch": 3.9683964732879744,
-      "grad_norm": 0.008519008755683899,
-      "learning_rate": 0.00019999223544362157,
-      "loss": 46.0,
-      "step": 24642
-    },
-    {
-      "epoch": 3.968557510366762,
-      "grad_norm": 0.00339347287081182,
-      "learning_rate": 0.0001999922348131723,
-      "loss": 46.0,
-      "step": 24643
-    },
-    {
-      "epoch": 3.9687185474455493,
-      "grad_norm": 0.0018427064642310143,
-      "learning_rate": 0.00019999223418269742,
-      "loss": 46.0,
-      "step": 24644
-    },
-    {
-      "epoch": 3.9688795845243368,
-      "grad_norm": 0.008539896458387375,
-      "learning_rate": 0.00019999223355219694,
-      "loss": 46.0,
-      "step": 24645
-    },
-    {
-      "epoch": 3.969040621603124,
-      "grad_norm": 0.0031247620936483145,
-      "learning_rate": 0.00019999223292167088,
-      "loss": 46.0,
-      "step": 24646
-    },
-    {
-      "epoch": 3.9692016586819117,
-      "grad_norm": 0.002595973666757345,
-      "learning_rate": 0.0001999922322911192,
-      "loss": 46.0,
-      "step": 24647
-    },
-    {
-      "epoch": 3.969362695760699,
-      "grad_norm": 0.003589972387999296,
-      "learning_rate": 0.00019999223166054195,
-      "loss": 46.0,
-      "step": 24648
-    },
-    {
-      "epoch": 3.9695237328394866,
-      "grad_norm": 0.008597678504884243,
-      "learning_rate": 0.0001999922310299391,
-      "loss": 46.0,
-      "step": 24649
-    },
-    {
-      "epoch": 3.9696847699182736,
-      "grad_norm": 0.0037852467503398657,
-      "learning_rate": 0.00019999223039931065,
-      "loss": 46.0,
-      "step": 24650
-    },
-    {
-      "epoch": 3.969845806997061,
-      "grad_norm": 0.016174133867025375,
-      "learning_rate": 0.00019999222976865662,
-      "loss": 46.0,
-      "step": 24651
-    },
-    {
-      "epoch": 3.9700068440758485,
-      "grad_norm": 0.0014885629061609507,
-      "learning_rate": 0.000199992229137977,
-      "loss": 46.0,
-      "step": 24652
-    },
-    {
-      "epoch": 3.970167881154636,
-      "grad_norm": 0.0023910885211080313,
-      "learning_rate": 0.00019999222850727177,
-      "loss": 46.0,
-      "step": 24653
-    },
-    {
-      "epoch": 3.9703289182334234,
-      "grad_norm": 0.0009760893881320953,
-      "learning_rate": 0.00019999222787654095,
-      "loss": 46.0,
-      "step": 24654
-    },
-    {
-      "epoch": 3.9704899553122104,
-      "grad_norm": 0.005014242138713598,
-      "learning_rate": 0.00019999222724578455,
-      "loss": 46.0,
-      "step": 24655
-    },
-    {
-      "epoch": 3.970650992390998,
-      "grad_norm": 0.004406902473419905,
-      "learning_rate": 0.00019999222661500255,
-      "loss": 46.0,
-      "step": 24656
-    },
-    {
-      "epoch": 3.9708120294697853,
-      "grad_norm": 0.010985944420099258,
-      "learning_rate": 0.00019999222598419494,
-      "loss": 46.0,
-      "step": 24657
-    },
-    {
-      "epoch": 3.9709730665485727,
-      "grad_norm": 0.000462591735413298,
-      "learning_rate": 0.00019999222535336175,
-      "loss": 46.0,
-      "step": 24658
-    },
-    {
-      "epoch": 3.97113410362736,
-      "grad_norm": 0.01707618683576584,
-      "learning_rate": 0.00019999222472250297,
-      "loss": 46.0,
-      "step": 24659
-    },
-    {
-      "epoch": 3.9712951407061476,
-      "grad_norm": 0.002642994513735175,
-      "learning_rate": 0.0001999922240916186,
-      "loss": 46.0,
-      "step": 24660
-    },
-    {
-      "epoch": 3.971456177784935,
-      "grad_norm": 0.004335297271609306,
-      "learning_rate": 0.00019999222346070864,
-      "loss": 46.0,
-      "step": 24661
-    },
-    {
-      "epoch": 3.9716172148637225,
-      "grad_norm": 0.020558614283800125,
-      "learning_rate": 0.00019999222282977307,
-      "loss": 46.0,
-      "step": 24662
-    },
-    {
-      "epoch": 3.97177825194251,
-      "grad_norm": 0.008380230516195297,
-      "learning_rate": 0.00019999222219881194,
-      "loss": 46.0,
-      "step": 24663
-    },
-    {
-      "epoch": 3.971939289021297,
-      "grad_norm": 0.004801325965672731,
-      "learning_rate": 0.0001999922215678252,
-      "loss": 46.0,
-      "step": 24664
-    },
-    {
-      "epoch": 3.9721003261000845,
-      "grad_norm": 0.0026345206424593925,
-      "learning_rate": 0.00019999222093681284,
-      "loss": 46.0,
-      "step": 24665
-    },
-    {
-      "epoch": 3.972261363178872,
-      "grad_norm": 0.0064878519624471664,
-      "learning_rate": 0.00019999222030577492,
-      "loss": 46.0,
-      "step": 24666
-    },
-    {
-      "epoch": 3.9724224002576594,
-      "grad_norm": 0.001823552418500185,
-      "learning_rate": 0.0001999922196747114,
-      "loss": 46.0,
-      "step": 24667
-    },
-    {
-      "epoch": 3.972583437336447,
-      "grad_norm": 0.004214639309793711,
-      "learning_rate": 0.0001999922190436223,
-      "loss": 46.0,
-      "step": 24668
-    },
-    {
-      "epoch": 3.9727444744152343,
-      "grad_norm": 0.01371967326849699,
-      "learning_rate": 0.00019999221841250758,
-      "loss": 46.0,
-      "step": 24669
-    },
-    {
-      "epoch": 3.9729055114940213,
-      "grad_norm": 0.007138080894947052,
-      "learning_rate": 0.00019999221778136729,
-      "loss": 46.0,
-      "step": 24670
-    },
-    {
-      "epoch": 3.9730665485728087,
-      "grad_norm": 0.002246492076665163,
-      "learning_rate": 0.00019999221715020138,
-      "loss": 46.0,
-      "step": 24671
-    },
-    {
-      "epoch": 3.973227585651596,
-      "grad_norm": 0.0008088405011221766,
-      "learning_rate": 0.00019999221651900988,
-      "loss": 46.0,
-      "step": 24672
-    },
-    {
-      "epoch": 3.9733886227303836,
-      "grad_norm": 0.0047655063681304455,
-      "learning_rate": 0.0001999922158877928,
-      "loss": 46.0,
-      "step": 24673
-    },
-    {
-      "epoch": 3.973549659809171,
-      "grad_norm": 0.0012428220361471176,
-      "learning_rate": 0.00019999221525655015,
-      "loss": 46.0,
-      "step": 24674
-    },
-    {
-      "epoch": 3.9737106968879585,
-      "grad_norm": 0.0027536784764379263,
-      "learning_rate": 0.00019999221462528187,
-      "loss": 46.0,
-      "step": 24675
-    },
-    {
-      "epoch": 3.973871733966746,
-      "grad_norm": 0.0021960961166769266,
-      "learning_rate": 0.000199992213993988,
-      "loss": 46.0,
-      "step": 24676
-    },
-    {
-      "epoch": 3.9740327710455334,
-      "grad_norm": 0.0020430022850632668,
-      "learning_rate": 0.00019999221336266856,
-      "loss": 46.0,
-      "step": 24677
-    },
-    {
-      "epoch": 3.974193808124321,
-      "grad_norm": 0.010779774747788906,
-      "learning_rate": 0.00019999221273132352,
-      "loss": 46.0,
-      "step": 24678
-    },
-    {
-      "epoch": 3.974354845203108,
-      "grad_norm": 0.0004967165878042579,
-      "learning_rate": 0.00019999221209995288,
-      "loss": 46.0,
-      "step": 24679
-    },
-    {
-      "epoch": 3.9745158822818953,
-      "grad_norm": 0.004149053245782852,
-      "learning_rate": 0.00019999221146855666,
-      "loss": 46.0,
-      "step": 24680
-    },
-    {
-      "epoch": 3.974676919360683,
-      "grad_norm": 0.01024789921939373,
-      "learning_rate": 0.00019999221083713483,
-      "loss": 46.0,
-      "step": 24681
-    },
-    {
-      "epoch": 3.9748379564394702,
-      "grad_norm": 0.004827845375984907,
-      "learning_rate": 0.0001999922102056874,
-      "loss": 46.0,
-      "step": 24682
-    },
-    {
-      "epoch": 3.9749989935182577,
-      "grad_norm": 0.0033513468224555254,
-      "learning_rate": 0.00019999220957421442,
-      "loss": 46.0,
-      "step": 24683
-    },
-    {
-      "epoch": 3.9751600305970447,
-      "grad_norm": 0.006375516764819622,
-      "learning_rate": 0.0001999922089427158,
-      "loss": 46.0,
-      "step": 24684
-    },
-    {
-      "epoch": 3.975321067675832,
-      "grad_norm": 0.0020544766448438168,
-      "learning_rate": 0.0001999922083111916,
-      "loss": 46.0,
-      "step": 24685
-    },
-    {
-      "epoch": 3.9754821047546196,
-      "grad_norm": 0.0003264143888372928,
-      "learning_rate": 0.0001999922076796418,
-      "loss": 46.0,
-      "step": 24686
-    },
-    {
-      "epoch": 3.975643141833407,
-      "grad_norm": 0.010692698881030083,
-      "learning_rate": 0.00019999220704806643,
-      "loss": 46.0,
-      "step": 24687
-    },
-    {
-      "epoch": 3.9758041789121945,
-      "grad_norm": 0.0098728621378541,
-      "learning_rate": 0.00019999220641646548,
-      "loss": 46.0,
-      "step": 24688
-    },
-    {
-      "epoch": 3.975965215990982,
-      "grad_norm": 0.007909636944532394,
-      "learning_rate": 0.0001999922057848389,
-      "loss": 46.0,
-      "step": 24689
-    },
-    {
-      "epoch": 3.9761262530697694,
-      "grad_norm": 0.0012305461568757892,
-      "learning_rate": 0.00019999220515318675,
-      "loss": 46.0,
-      "step": 24690
-    },
-    {
-      "epoch": 3.976287290148557,
-      "grad_norm": 0.010655613616108894,
-      "learning_rate": 0.000199992204521509,
-      "loss": 46.0,
-      "step": 24691
-    },
-    {
-      "epoch": 3.9764483272273443,
-      "grad_norm": 0.011089557781815529,
-      "learning_rate": 0.00019999220388980567,
-      "loss": 46.0,
-      "step": 24692
-    },
-    {
-      "epoch": 3.9766093643061313,
-      "grad_norm": 0.0222268495708704,
-      "learning_rate": 0.00019999220325807673,
-      "loss": 46.0,
-      "step": 24693
-    },
-    {
-      "epoch": 3.9767704013849188,
-      "grad_norm": 0.0025264096911996603,
-      "learning_rate": 0.00019999220262632218,
-      "loss": 46.0,
-      "step": 24694
-    },
-    {
-      "epoch": 3.9769314384637062,
-      "grad_norm": 0.004113536793738604,
-      "learning_rate": 0.00019999220199454207,
-      "loss": 46.0,
-      "step": 24695
-    },
-    {
-      "epoch": 3.9770924755424937,
-      "grad_norm": 0.012270519509911537,
-      "learning_rate": 0.00019999220136273635,
-      "loss": 46.0,
-      "step": 24696
-    },
-    {
-      "epoch": 3.977253512621281,
-      "grad_norm": 0.0021190920379012823,
-      "learning_rate": 0.00019999220073090503,
-      "loss": 46.0,
-      "step": 24697
-    },
-    {
-      "epoch": 3.9774145497000686,
-      "grad_norm": 0.0017293510027229786,
-      "learning_rate": 0.00019999220009904814,
-      "loss": 46.0,
-      "step": 24698
-    },
-    {
-      "epoch": 3.9775755867788556,
-      "grad_norm": 0.0018818129319697618,
-      "learning_rate": 0.00019999219946716562,
-      "loss": 46.0,
-      "step": 24699
-    },
-    {
-      "epoch": 3.977736623857643,
-      "grad_norm": 0.0004118960350751877,
-      "learning_rate": 0.00019999219883525755,
-      "loss": 46.0,
-      "step": 24700
-    },
-    {
-      "epoch": 3.9778976609364305,
-      "grad_norm": 0.0017546769231557846,
-      "learning_rate": 0.00019999219820332386,
-      "loss": 46.0,
-      "step": 24701
-    },
-    {
-      "epoch": 3.978058698015218,
-      "grad_norm": 0.0035819578915834427,
-      "learning_rate": 0.00019999219757136462,
-      "loss": 46.0,
-      "step": 24702
-    },
-    {
-      "epoch": 3.9782197350940054,
-      "grad_norm": 0.003807461354881525,
-      "learning_rate": 0.00019999219693937973,
-      "loss": 46.0,
-      "step": 24703
-    },
-    {
-      "epoch": 3.978380772172793,
-      "grad_norm": 0.006211885716766119,
-      "learning_rate": 0.00019999219630736928,
-      "loss": 46.0,
-      "step": 24704
-    },
-    {
-      "epoch": 3.9785418092515803,
-      "grad_norm": 0.0023693828843533993,
-      "learning_rate": 0.00019999219567533321,
-      "loss": 46.0,
-      "step": 24705
-    },
-    {
-      "epoch": 3.9787028463303677,
-      "grad_norm": 0.0019062282517552376,
-      "learning_rate": 0.0001999921950432716,
-      "loss": 46.0,
-      "step": 24706
-    },
-    {
-      "epoch": 3.978863883409155,
-      "grad_norm": 0.008016892708837986,
-      "learning_rate": 0.00019999219441118435,
-      "loss": 46.0,
-      "step": 24707
-    },
-    {
-      "epoch": 3.979024920487942,
-      "grad_norm": 0.005275571718811989,
-      "learning_rate": 0.0001999921937790715,
-      "loss": 46.0,
-      "step": 24708
-    },
-    {
-      "epoch": 3.9791859575667297,
-      "grad_norm": 0.002864799229428172,
-      "learning_rate": 0.00019999219314693306,
-      "loss": 46.0,
-      "step": 24709
-    },
-    {
-      "epoch": 3.979346994645517,
-      "grad_norm": 0.0022006540093570948,
-      "learning_rate": 0.00019999219251476907,
-      "loss": 46.0,
-      "step": 24710
-    },
-    {
-      "epoch": 3.9795080317243046,
-      "grad_norm": 0.0026308512315154076,
-      "learning_rate": 0.00019999219188257948,
-      "loss": 46.0,
-      "step": 24711
-    },
-    {
-      "epoch": 3.979669068803092,
-      "grad_norm": 0.0034603977110236883,
-      "learning_rate": 0.00019999219125036425,
-      "loss": 46.0,
-      "step": 24712
-    },
-    {
-      "epoch": 3.9798301058818795,
-      "grad_norm": 0.00779740558937192,
-      "learning_rate": 0.00019999219061812346,
-      "loss": 46.0,
-      "step": 24713
-    },
-    {
-      "epoch": 3.9799911429606665,
-      "grad_norm": 0.013969328254461288,
-      "learning_rate": 0.0001999921899858571,
-      "loss": 46.0,
-      "step": 24714
-    },
-    {
-      "epoch": 3.980152180039454,
-      "grad_norm": 0.0053104558028280735,
-      "learning_rate": 0.0001999921893535651,
-      "loss": 46.0,
-      "step": 24715
-    },
-    {
-      "epoch": 3.9803132171182414,
-      "grad_norm": 0.0016471610870212317,
-      "learning_rate": 0.00019999218872124752,
-      "loss": 46.0,
-      "step": 24716
-    },
-    {
-      "epoch": 3.980474254197029,
-      "grad_norm": 0.000960015517193824,
-      "learning_rate": 0.00019999218808890436,
-      "loss": 46.0,
-      "step": 24717
-    },
-    {
-      "epoch": 3.9806352912758163,
-      "grad_norm": 0.003994822036474943,
-      "learning_rate": 0.00019999218745653558,
-      "loss": 46.0,
-      "step": 24718
-    },
-    {
-      "epoch": 3.9807963283546037,
-      "grad_norm": 0.001035549445077777,
-      "learning_rate": 0.00019999218682414125,
-      "loss": 46.0,
-      "step": 24719
-    },
-    {
-      "epoch": 3.980957365433391,
-      "grad_norm": 0.009329813532531261,
-      "learning_rate": 0.0001999921861917213,
-      "loss": 46.0,
-      "step": 24720
-    },
-    {
-      "epoch": 3.9811184025121786,
-      "grad_norm": 0.0053340899758040905,
-      "learning_rate": 0.00019999218555927578,
-      "loss": 46.0,
-      "step": 24721
-    },
-    {
-      "epoch": 3.981279439590966,
-      "grad_norm": 0.011015500873327255,
-      "learning_rate": 0.00019999218492680463,
-      "loss": 46.0,
-      "step": 24722
-    },
-    {
-      "epoch": 3.981440476669753,
-      "grad_norm": 0.012094903737306595,
-      "learning_rate": 0.00019999218429430792,
-      "loss": 46.0,
-      "step": 24723
-    },
-    {
-      "epoch": 3.9816015137485405,
-      "grad_norm": 0.005665578413754702,
-      "learning_rate": 0.00019999218366178561,
-      "loss": 46.0,
-      "step": 24724
-    },
-    {
-      "epoch": 3.981762550827328,
-      "grad_norm": 0.0025657762307673693,
-      "learning_rate": 0.00019999218302923767,
-      "loss": 46.0,
-      "step": 24725
-    },
-    {
-      "epoch": 3.9819235879061154,
-      "grad_norm": 0.005956851411610842,
-      "learning_rate": 0.00019999218239666417,
-      "loss": 46.0,
-      "step": 24726
-    },
-    {
-      "epoch": 3.982084624984903,
-      "grad_norm": 0.0009998215828090906,
-      "learning_rate": 0.00019999218176406508,
-      "loss": 46.0,
-      "step": 24727
-    },
-    {
-      "epoch": 3.98224566206369,
-      "grad_norm": 0.015045387670397758,
-      "learning_rate": 0.0001999921811314404,
-      "loss": 46.0,
-      "step": 24728
-    },
-    {
-      "epoch": 3.9824066991424774,
-      "grad_norm": 0.0013141731033101678,
-      "learning_rate": 0.00019999218049879011,
-      "loss": 46.0,
-      "step": 24729
-    },
-    {
-      "epoch": 3.982567736221265,
-      "grad_norm": 0.002103073289617896,
-      "learning_rate": 0.00019999217986611424,
-      "loss": 46.0,
-      "step": 24730
-    },
-    {
-      "epoch": 3.9827287733000523,
-      "grad_norm": 0.0019091975409537554,
-      "learning_rate": 0.00019999217923341277,
-      "loss": 46.0,
-      "step": 24731
-    },
-    {
-      "epoch": 3.9828898103788397,
-      "grad_norm": 0.006263358052819967,
-      "learning_rate": 0.00019999217860068572,
-      "loss": 46.0,
-      "step": 24732
-    },
-    {
-      "epoch": 3.983050847457627,
-      "grad_norm": 0.0015347658190876245,
-      "learning_rate": 0.00019999217796793305,
-      "loss": 46.0,
-      "step": 24733
-    },
-    {
-      "epoch": 3.9832118845364146,
-      "grad_norm": 0.0006784264696761966,
-      "learning_rate": 0.00019999217733515483,
-      "loss": 46.0,
-      "step": 24734
-    },
-    {
-      "epoch": 3.983372921615202,
-      "grad_norm": 0.004705656319856644,
-      "learning_rate": 0.00019999217670235099,
-      "loss": 46.0,
-      "step": 24735
-    },
-    {
-      "epoch": 3.9835339586939895,
-      "grad_norm": 0.002923637628555298,
-      "learning_rate": 0.00019999217606952153,
-      "loss": 46.0,
-      "step": 24736
-    },
-    {
-      "epoch": 3.9836949957727765,
-      "grad_norm": 0.0025600737426429987,
-      "learning_rate": 0.00019999217543666654,
-      "loss": 46.0,
-      "step": 24737
-    },
-    {
-      "epoch": 3.983856032851564,
-      "grad_norm": 0.0014375726459547877,
-      "learning_rate": 0.0001999921748037859,
-      "loss": 46.0,
-      "step": 24738
-    },
-    {
-      "epoch": 3.9840170699303514,
-      "grad_norm": 0.0026932931505143642,
-      "learning_rate": 0.00019999217417087972,
-      "loss": 46.0,
-      "step": 24739
-    },
-    {
-      "epoch": 3.984178107009139,
-      "grad_norm": 0.0025541021022945642,
-      "learning_rate": 0.0001999921735379479,
-      "loss": 46.0,
-      "step": 24740
-    },
-    {
-      "epoch": 3.9843391440879263,
-      "grad_norm": 0.0012892563827335835,
-      "learning_rate": 0.0001999921729049905,
-      "loss": 46.0,
-      "step": 24741
-    },
-    {
-      "epoch": 3.984500181166714,
-      "grad_norm": 0.002662871964275837,
-      "learning_rate": 0.0001999921722720075,
-      "loss": 46.0,
-      "step": 24742
-    },
-    {
-      "epoch": 3.984661218245501,
-      "grad_norm": 0.010451670736074448,
-      "learning_rate": 0.00019999217163899893,
-      "loss": 46.0,
-      "step": 24743
-    },
-    {
-      "epoch": 3.9848222553242882,
-      "grad_norm": 0.016197005286812782,
-      "learning_rate": 0.00019999217100596478,
-      "loss": 46.0,
-      "step": 24744
-    },
-    {
-      "epoch": 3.9849832924030757,
-      "grad_norm": 0.001326465979218483,
-      "learning_rate": 0.00019999217037290499,
-      "loss": 46.0,
-      "step": 24745
-    },
-    {
-      "epoch": 3.985144329481863,
-      "grad_norm": 0.0023445934057235718,
-      "learning_rate": 0.00019999216973981963,
-      "loss": 46.0,
-      "step": 24746
-    },
-    {
-      "epoch": 3.9853053665606506,
-      "grad_norm": 0.00758154783397913,
-      "learning_rate": 0.0001999921691067087,
-      "loss": 46.0,
-      "step": 24747
-    },
-    {
-      "epoch": 3.985466403639438,
-      "grad_norm": 0.007461497560143471,
-      "learning_rate": 0.00019999216847357213,
-      "loss": 46.0,
-      "step": 24748
-    },
-    {
-      "epoch": 3.9856274407182255,
-      "grad_norm": 0.012365833856165409,
-      "learning_rate": 0.00019999216784041002,
-      "loss": 46.0,
-      "step": 24749
-    },
-    {
-      "epoch": 3.985788477797013,
-      "grad_norm": 0.0039055971428751945,
-      "learning_rate": 0.00019999216720722226,
-      "loss": 46.0,
-      "step": 24750
-    },
-    {
-      "epoch": 3.9859495148758004,
-      "grad_norm": 0.003098601708188653,
-      "learning_rate": 0.00019999216657400895,
-      "loss": 46.0,
-      "step": 24751
-    },
-    {
-      "epoch": 3.9861105519545874,
-      "grad_norm": 0.0011022333055734634,
-      "learning_rate": 0.00019999216594077004,
-      "loss": 46.0,
-      "step": 24752
-    },
-    {
-      "epoch": 3.986271589033375,
-      "grad_norm": 0.01449637021869421,
-      "learning_rate": 0.0001999921653075055,
-      "loss": 46.0,
-      "step": 24753
-    },
-    {
-      "epoch": 3.9864326261121623,
-      "grad_norm": 0.0024452845100313425,
-      "learning_rate": 0.00019999216467421542,
-      "loss": 46.0,
-      "step": 24754
-    },
-    {
-      "epoch": 3.9865936631909498,
-      "grad_norm": 0.012003865092992783,
-      "learning_rate": 0.0001999921640408997,
-      "loss": 46.0,
-      "step": 24755
-    },
-    {
-      "epoch": 3.986754700269737,
-      "grad_norm": 0.0013558976352214813,
-      "learning_rate": 0.00019999216340755845,
-      "loss": 46.0,
-      "step": 24756
-    },
-    {
-      "epoch": 3.9869157373485242,
-      "grad_norm": 0.004575369879603386,
-      "learning_rate": 0.00019999216277419155,
-      "loss": 46.0,
-      "step": 24757
-    },
-    {
-      "epoch": 3.9870767744273117,
-      "grad_norm": 0.006203664466738701,
-      "learning_rate": 0.0001999921621407991,
-      "loss": 46.0,
-      "step": 24758
-    },
-    {
-      "epoch": 3.987237811506099,
-      "grad_norm": 0.0019251321209594607,
-      "learning_rate": 0.00019999216150738103,
-      "loss": 46.0,
-      "step": 24759
-    },
-    {
-      "epoch": 3.9873988485848866,
-      "grad_norm": 0.009185073897242546,
-      "learning_rate": 0.00019999216087393737,
-      "loss": 46.0,
-      "step": 24760
-    },
-    {
-      "epoch": 3.987559885663674,
-      "grad_norm": 0.0031341221183538437,
-      "learning_rate": 0.0001999921602404681,
-      "loss": 46.0,
-      "step": 24761
-    },
-    {
-      "epoch": 3.9877209227424615,
-      "grad_norm": 0.004984932020306587,
-      "learning_rate": 0.00019999215960697327,
-      "loss": 46.0,
-      "step": 24762
-    },
-    {
-      "epoch": 3.987881959821249,
-      "grad_norm": 0.0034871690440922976,
-      "learning_rate": 0.00019999215897345283,
-      "loss": 46.0,
-      "step": 24763
-    },
-    {
-      "epoch": 3.9880429969000364,
-      "grad_norm": 0.0013955391477793455,
-      "learning_rate": 0.0001999921583399068,
-      "loss": 46.0,
-      "step": 24764
-    },
-    {
-      "epoch": 3.988204033978824,
-      "grad_norm": 0.009611187502741814,
-      "learning_rate": 0.00019999215770633518,
-      "loss": 46.0,
-      "step": 24765
-    },
-    {
-      "epoch": 3.9883650710576113,
-      "grad_norm": 0.001299932599067688,
-      "learning_rate": 0.00019999215707273795,
-      "loss": 46.0,
-      "step": 24766
-    },
-    {
-      "epoch": 3.9885261081363983,
-      "grad_norm": 0.0028664679266512394,
-      "learning_rate": 0.00019999215643911513,
-      "loss": 46.0,
-      "step": 24767
-    },
-    {
-      "epoch": 3.9886871452151857,
-      "grad_norm": 0.007663626689463854,
-      "learning_rate": 0.00019999215580546675,
-      "loss": 46.0,
-      "step": 24768
-    },
-    {
-      "epoch": 3.988848182293973,
-      "grad_norm": 0.0063698128797113895,
-      "learning_rate": 0.00019999215517179275,
-      "loss": 46.0,
-      "step": 24769
-    },
-    {
-      "epoch": 3.9890092193727607,
-      "grad_norm": 0.0028165553230792284,
-      "learning_rate": 0.00019999215453809317,
-      "loss": 46.0,
-      "step": 24770
-    },
-    {
-      "epoch": 3.989170256451548,
-      "grad_norm": 0.0068365428596735,
-      "learning_rate": 0.00019999215390436797,
-      "loss": 46.0,
-      "step": 24771
-    },
-    {
-      "epoch": 3.989331293530335,
-      "grad_norm": 0.0009981007315218449,
-      "learning_rate": 0.0001999921532706172,
-      "loss": 46.0,
-      "step": 24772
-    },
-    {
-      "epoch": 3.9894923306091226,
-      "grad_norm": 0.002645881613716483,
-      "learning_rate": 0.00019999215263684082,
-      "loss": 46.0,
-      "step": 24773
-    },
-    {
-      "epoch": 3.98965336768791,
-      "grad_norm": 0.001725579029880464,
-      "learning_rate": 0.00019999215200303886,
-      "loss": 46.0,
-      "step": 24774
-    },
-    {
-      "epoch": 3.9898144047666975,
-      "grad_norm": 0.00524672819301486,
-      "learning_rate": 0.00019999215136921132,
-      "loss": 46.0,
-      "step": 24775
-    },
-    {
-      "epoch": 3.989975441845485,
-      "grad_norm": 0.0017213233513757586,
-      "learning_rate": 0.0001999921507353582,
-      "loss": 46.0,
-      "step": 24776
-    },
-    {
-      "epoch": 3.9901364789242724,
-      "grad_norm": 0.003459772327914834,
-      "learning_rate": 0.00019999215010147941,
-      "loss": 46.0,
-      "step": 24777
-    },
-    {
-      "epoch": 3.99029751600306,
-      "grad_norm": 0.00529498141258955,
-      "learning_rate": 0.0001999921494675751,
-      "loss": 46.0,
-      "step": 24778
-    },
-    {
-      "epoch": 3.9904585530818473,
-      "grad_norm": 0.01653357408940792,
-      "learning_rate": 0.0001999921488336452,
-      "loss": 46.0,
-      "step": 24779
-    },
-    {
-      "epoch": 3.9906195901606347,
-      "grad_norm": 0.0013534490717574954,
-      "learning_rate": 0.00019999214819968968,
-      "loss": 46.0,
-      "step": 24780
-    },
-    {
-      "epoch": 3.9907806272394217,
-      "grad_norm": 0.004747649189084768,
-      "learning_rate": 0.00019999214756570853,
-      "loss": 46.0,
-      "step": 24781
-    },
-    {
-      "epoch": 3.990941664318209,
-      "grad_norm": 0.004367772489786148,
-      "learning_rate": 0.00019999214693170185,
-      "loss": 46.0,
-      "step": 24782
-    },
-    {
-      "epoch": 3.9911027013969966,
-      "grad_norm": 0.008277861401438713,
-      "learning_rate": 0.00019999214629766955,
-      "loss": 46.0,
-      "step": 24783
-    },
-    {
-      "epoch": 3.991263738475784,
-      "grad_norm": 0.006896821781992912,
-      "learning_rate": 0.00019999214566361167,
-      "loss": 46.0,
-      "step": 24784
-    },
-    {
-      "epoch": 3.9914247755545715,
-      "grad_norm": 0.0029220758005976677,
-      "learning_rate": 0.0001999921450295282,
-      "loss": 46.0,
-      "step": 24785
-    },
-    {
-      "epoch": 3.991585812633359,
-      "grad_norm": 0.003588699968531728,
-      "learning_rate": 0.00019999214439541914,
-      "loss": 46.0,
-      "step": 24786
-    },
-    {
-      "epoch": 3.991746849712146,
-      "grad_norm": 0.004981989040970802,
-      "learning_rate": 0.00019999214376128444,
-      "loss": 46.0,
-      "step": 24787
-    },
-    {
-      "epoch": 3.9919078867909334,
-      "grad_norm": 0.005065221339464188,
-      "learning_rate": 0.00019999214312712419,
-      "loss": 46.0,
-      "step": 24788
-    },
-    {
-      "epoch": 3.992068923869721,
-      "grad_norm": 0.0030243347864598036,
-      "learning_rate": 0.00019999214249293834,
-      "loss": 46.0,
-      "step": 24789
-    },
-    {
-      "epoch": 3.9922299609485083,
-      "grad_norm": 0.003149948548525572,
-      "learning_rate": 0.0001999921418587269,
-      "loss": 46.0,
-      "step": 24790
-    },
-    {
-      "epoch": 3.992390998027296,
-      "grad_norm": 0.0061769732274115086,
-      "learning_rate": 0.00019999214122448986,
-      "loss": 46.0,
-      "step": 24791
-    },
-    {
-      "epoch": 3.9925520351060833,
-      "grad_norm": 0.004211899358779192,
-      "learning_rate": 0.00019999214059022725,
-      "loss": 46.0,
-      "step": 24792
-    },
-    {
-      "epoch": 3.9927130721848707,
-      "grad_norm": 0.004260799381881952,
-      "learning_rate": 0.00019999213995593903,
-      "loss": 46.0,
-      "step": 24793
-    },
-    {
-      "epoch": 3.992874109263658,
-      "grad_norm": 0.0047255465760827065,
-      "learning_rate": 0.00019999213932162517,
-      "loss": 46.0,
-      "step": 24794
-    },
-    {
-      "epoch": 3.9930351463424456,
-      "grad_norm": 0.004821198061108589,
-      "learning_rate": 0.0001999921386872858,
-      "loss": 46.0,
-      "step": 24795
-    },
-    {
-      "epoch": 3.9931961834212326,
-      "grad_norm": 0.002056732075288892,
-      "learning_rate": 0.00019999213805292079,
-      "loss": 46.0,
-      "step": 24796
-    },
-    {
-      "epoch": 3.99335722050002,
-      "grad_norm": 0.001037733512930572,
-      "learning_rate": 0.0001999921374185302,
-      "loss": 46.0,
-      "step": 24797
-    },
-    {
-      "epoch": 3.9935182575788075,
-      "grad_norm": 0.006771755404770374,
-      "learning_rate": 0.000199992136784114,
-      "loss": 46.0,
-      "step": 24798
-    },
-    {
-      "epoch": 3.993679294657595,
-      "grad_norm": 0.006890963297337294,
-      "learning_rate": 0.0001999921361496722,
-      "loss": 46.0,
-      "step": 24799
-    },
-    {
-      "epoch": 3.9938403317363824,
-      "grad_norm": 0.004899971187114716,
-      "learning_rate": 0.00019999213551520484,
-      "loss": 46.0,
-      "step": 24800
-    },
-    {
-      "epoch": 3.9940013688151694,
-      "grad_norm": 0.0019697954412549734,
-      "learning_rate": 0.00019999213488071187,
-      "loss": 46.0,
-      "step": 24801
-    },
-    {
-      "epoch": 3.994162405893957,
-      "grad_norm": 0.0011517596431076527,
-      "learning_rate": 0.0001999921342461933,
-      "loss": 46.0,
-      "step": 24802
-    },
-    {
-      "epoch": 3.9943234429727443,
-      "grad_norm": 0.0033353553153574467,
-      "learning_rate": 0.00019999213361164916,
-      "loss": 46.0,
-      "step": 24803
-    },
-    {
-      "epoch": 3.994484480051532,
-      "grad_norm": 0.009231653064489365,
-      "learning_rate": 0.00019999213297707943,
-      "loss": 46.0,
-      "step": 24804
-    },
-    {
-      "epoch": 3.9946455171303192,
-      "grad_norm": 0.005704787094146013,
-      "learning_rate": 0.00019999213234248408,
-      "loss": 46.0,
-      "step": 24805
-    },
-    {
-      "epoch": 3.9948065542091067,
-      "grad_norm": 0.0023751973640173674,
-      "learning_rate": 0.00019999213170786314,
-      "loss": 46.0,
-      "step": 24806
-    },
-    {
-      "epoch": 3.994967591287894,
-      "grad_norm": 0.0035583116114139557,
-      "learning_rate": 0.00019999213107321662,
-      "loss": 46.0,
-      "step": 24807
-    },
-    {
-      "epoch": 3.9951286283666816,
-      "grad_norm": 0.004200700670480728,
-      "learning_rate": 0.0001999921304385445,
-      "loss": 46.0,
-      "step": 24808
-    },
-    {
-      "epoch": 3.995289665445469,
-      "grad_norm": 0.0019821873866021633,
-      "learning_rate": 0.00019999212980384678,
-      "loss": 46.0,
-      "step": 24809
-    },
-    {
-      "epoch": 3.995450702524256,
-      "grad_norm": 0.01348206214606762,
-      "learning_rate": 0.0001999921291691235,
-      "loss": 46.0,
-      "step": 24810
-    },
-    {
-      "epoch": 3.9956117396030435,
-      "grad_norm": 0.004082653671503067,
-      "learning_rate": 0.0001999921285343746,
-      "loss": 46.0,
-      "step": 24811
-    },
-    {
-      "epoch": 3.995772776681831,
-      "grad_norm": 0.008464104495942593,
-      "learning_rate": 0.0001999921278996001,
-      "loss": 46.0,
-      "step": 24812
-    },
-    {
-      "epoch": 3.9959338137606184,
-      "grad_norm": 0.0035004690289497375,
-      "learning_rate": 0.00019999212726480004,
-      "loss": 46.0,
-      "step": 24813
-    },
-    {
-      "epoch": 3.996094850839406,
-      "grad_norm": 0.002037183614447713,
-      "learning_rate": 0.00019999212662997435,
-      "loss": 46.0,
-      "step": 24814
-    },
-    {
-      "epoch": 3.9962558879181933,
-      "grad_norm": 0.001355841406621039,
-      "learning_rate": 0.0001999921259951231,
-      "loss": 46.0,
-      "step": 24815
-    },
-    {
-      "epoch": 3.9964169249969803,
-      "grad_norm": 0.0014639337314292789,
-      "learning_rate": 0.00019999212536024624,
-      "loss": 46.0,
-      "step": 24816
-    },
-    {
-      "epoch": 3.9965779620757678,
-      "grad_norm": 0.020203057676553726,
-      "learning_rate": 0.0001999921247253438,
-      "loss": 46.0,
-      "step": 24817
-    },
-    {
-      "epoch": 3.996738999154555,
-      "grad_norm": 0.002857648767530918,
-      "learning_rate": 0.00019999212409041572,
-      "loss": 46.0,
-      "step": 24818
-    },
-    {
-      "epoch": 3.9969000362333427,
-      "grad_norm": 0.0017090535257011652,
-      "learning_rate": 0.00019999212345546207,
-      "loss": 46.0,
-      "step": 24819
-    },
-    {
-      "epoch": 3.99706107331213,
-      "grad_norm": 0.0034088161773979664,
-      "learning_rate": 0.00019999212282048286,
-      "loss": 46.0,
-      "step": 24820
-    },
-    {
-      "epoch": 3.9972221103909176,
-      "grad_norm": 0.00397028774023056,
-      "learning_rate": 0.00019999212218547804,
-      "loss": 46.0,
-      "step": 24821
-    },
-    {
-      "epoch": 3.997383147469705,
-      "grad_norm": 0.002675285330042243,
-      "learning_rate": 0.00019999212155044762,
-      "loss": 46.0,
-      "step": 24822
-    },
-    {
-      "epoch": 3.9975441845484925,
-      "grad_norm": 0.0015281352680176497,
-      "learning_rate": 0.00019999212091539162,
-      "loss": 46.0,
-      "step": 24823
-    },
-    {
-      "epoch": 3.99770522162728,
-      "grad_norm": 0.001216026023030281,
-      "learning_rate": 0.00019999212028031,
-      "loss": 46.0,
-      "step": 24824
-    },
-    {
-      "epoch": 3.997866258706067,
-      "grad_norm": 0.0036450561601668596,
-      "learning_rate": 0.0001999921196452028,
-      "loss": 46.0,
-      "step": 24825
-    },
-    {
-      "epoch": 3.9980272957848544,
-      "grad_norm": 0.01008449587970972,
-      "learning_rate": 0.00019999211901007002,
-      "loss": 46.0,
-      "step": 24826
-    },
-    {
-      "epoch": 3.998188332863642,
-      "grad_norm": 0.005471523851156235,
-      "learning_rate": 0.00019999211837491164,
-      "loss": 46.0,
-      "step": 24827
-    },
-    {
-      "epoch": 3.9983493699424293,
-      "grad_norm": 0.002226640470325947,
-      "learning_rate": 0.00019999211773972765,
-      "loss": 46.0,
-      "step": 24828
-    },
-    {
-      "epoch": 3.9985104070212167,
-      "grad_norm": 0.012375475838780403,
-      "learning_rate": 0.0001999921171045181,
-      "loss": 46.0,
-      "step": 24829
-    },
-    {
-      "epoch": 3.998671444100004,
-      "grad_norm": 0.003869743784889579,
-      "learning_rate": 0.00019999211646928294,
-      "loss": 46.0,
-      "step": 24830
-    },
-    {
-      "epoch": 3.998832481178791,
-      "grad_norm": 0.0021538231521844864,
-      "learning_rate": 0.0001999921158340222,
-      "loss": 46.0,
-      "step": 24831
-    },
-    {
-      "epoch": 3.9989935182575786,
-      "grad_norm": 0.004209042061120272,
-      "learning_rate": 0.00019999211519873585,
-      "loss": 46.0,
-      "step": 24832
-    },
-    {
-      "epoch": 3.999154555336366,
-      "grad_norm": 0.004311444237828255,
-      "learning_rate": 0.0001999921145634239,
-      "loss": 46.0,
-      "step": 24833
-    },
-    {
-      "epoch": 3.9993155924151536,
-      "grad_norm": 0.004337720572948456,
-      "learning_rate": 0.00019999211392808639,
-      "loss": 46.0,
-      "step": 24834
-    },
-    {
-      "epoch": 3.999476629493941,
-      "grad_norm": 0.0011222240282222629,
-      "learning_rate": 0.00019999211329272326,
-      "loss": 46.0,
-      "step": 24835
-    },
-    {
-      "epoch": 3.9996376665727285,
-      "grad_norm": 0.012477749958634377,
-      "learning_rate": 0.00019999211265733452,
-      "loss": 46.0,
-      "step": 24836
-    },
-    {
-      "epoch": 3.999798703651516,
-      "grad_norm": 0.003405028721317649,
-      "learning_rate": 0.00019999211202192024,
-      "loss": 46.0,
-      "step": 24837
-    },
-    {
-      "epoch": 3.9999597407303034,
-      "grad_norm": 0.0043719857931137085,
-      "learning_rate": 0.00019999211138648033,
-      "loss": 46.0,
-      "step": 24838
-    },
-    {
-      "epoch": 3.9999597407303034,
-      "eval_loss": 11.5,
-      "eval_runtime": 14.8473,
-      "eval_samples_per_second": 176.126,
-      "eval_steps_per_second": 88.097,
-      "step": 24838
-    },
-    {
-      "epoch": 4.0001610370787875,
-      "grad_norm": 0.013840921223163605,
-      "learning_rate": 0.00019999211075101482,
-      "loss": 46.0,
-      "step": 24839
-    },
-    {
-      "epoch": 4.000322074157575,
-      "grad_norm": 0.0011507347226142883,
-      "learning_rate": 0.00019999211011552373,
-      "loss": 46.0,
-      "step": 24840
-    },
-    {
-      "epoch": 4.000483111236362,
-      "grad_norm": 0.01128339022397995,
-      "learning_rate": 0.00019999210948000706,
-      "loss": 46.0,
-      "step": 24841
-    },
-    {
-      "epoch": 4.00064414831515,
-      "grad_norm": 0.006644542794674635,
-      "learning_rate": 0.00019999210884446476,
-      "loss": 46.0,
-      "step": 24842
-    },
-    {
-      "epoch": 4.000805185393937,
-      "grad_norm": 0.005154333543032408,
-      "learning_rate": 0.0001999921082088969,
-      "loss": 46.0,
-      "step": 24843
-    },
-    {
-      "epoch": 4.000966222472725,
-      "grad_norm": 0.004086259752511978,
-      "learning_rate": 0.00019999210757330347,
-      "loss": 46.0,
-      "step": 24844
-    },
-    {
-      "epoch": 4.001127259551512,
-      "grad_norm": 0.0042493208311498165,
-      "learning_rate": 0.0001999921069376844,
-      "loss": 46.0,
-      "step": 24845
-    },
-    {
-      "epoch": 4.001288296630299,
-      "grad_norm": 0.004113773349672556,
-      "learning_rate": 0.00019999210630203975,
-      "loss": 46.0,
-      "step": 24846
-    },
-    {
-      "epoch": 4.001449333709086,
-      "grad_norm": 0.004083591513335705,
-      "learning_rate": 0.00019999210566636953,
-      "loss": 46.0,
-      "step": 24847
-    },
-    {
-      "epoch": 4.001610370787874,
-      "grad_norm": 0.010515168309211731,
-      "learning_rate": 0.00019999210503067368,
-      "loss": 46.0,
-      "step": 24848
-    },
-    {
-      "epoch": 4.001771407866661,
-      "grad_norm": 0.0006976528675295413,
-      "learning_rate": 0.00019999210439495228,
-      "loss": 46.0,
-      "step": 24849
-    },
-    {
-      "epoch": 4.0019324449454485,
-      "grad_norm": 0.00829688087105751,
-      "learning_rate": 0.00019999210375920524,
-      "loss": 46.0,
-      "step": 24850
-    },
-    {
-      "epoch": 4.002093482024236,
-      "grad_norm": 0.011194777674973011,
-      "learning_rate": 0.00019999210312343264,
-      "loss": 46.0,
-      "step": 24851
-    },
-    {
-      "epoch": 4.002254519103023,
-      "grad_norm": 0.0022454794961959124,
-      "learning_rate": 0.00019999210248763445,
-      "loss": 46.0,
-      "step": 24852
-    },
-    {
-      "epoch": 4.002415556181811,
-      "grad_norm": 0.0011609733337536454,
-      "learning_rate": 0.00019999210185181064,
-      "loss": 46.0,
-      "step": 24853
-    },
-    {
-      "epoch": 4.002576593260598,
-      "grad_norm": 0.0038015421014279127,
-      "learning_rate": 0.00019999210121596125,
-      "loss": 46.0,
-      "step": 24854
-    },
-    {
-      "epoch": 4.002737630339386,
-      "grad_norm": 0.001069945632480085,
-      "learning_rate": 0.00019999210058008627,
-      "loss": 46.0,
-      "step": 24855
-    },
-    {
-      "epoch": 4.002898667418173,
-      "grad_norm": 0.005756659433245659,
-      "learning_rate": 0.0001999920999441857,
-      "loss": 46.0,
-      "step": 24856
-    },
-    {
-      "epoch": 4.003059704496961,
-      "grad_norm": 0.012916953302919865,
-      "learning_rate": 0.00019999209930825952,
-      "loss": 46.0,
-      "step": 24857
-    },
-    {
-      "epoch": 4.003220741575748,
-      "grad_norm": 0.0026917133945971727,
-      "learning_rate": 0.00019999209867230778,
-      "loss": 46.0,
-      "step": 24858
-    },
-    {
-      "epoch": 4.003381778654536,
-      "grad_norm": 0.017080552875995636,
-      "learning_rate": 0.00019999209803633043,
-      "loss": 46.0,
-      "step": 24859
-    },
-    {
-      "epoch": 4.003542815733322,
-      "grad_norm": 0.010402707383036613,
-      "learning_rate": 0.0001999920974003275,
-      "loss": 46.0,
-      "step": 24860
-    },
-    {
-      "epoch": 4.00370385281211,
-      "grad_norm": 0.007182632572948933,
-      "learning_rate": 0.00019999209676429896,
-      "loss": 46.0,
-      "step": 24861
-    },
-    {
-      "epoch": 4.003864889890897,
-      "grad_norm": 0.005855188705027103,
-      "learning_rate": 0.00019999209612824484,
-      "loss": 46.0,
-      "step": 24862
-    },
-    {
-      "epoch": 4.0040259269696845,
-      "grad_norm": 0.0009144256473518908,
-      "learning_rate": 0.0001999920954921651,
-      "loss": 46.0,
-      "step": 24863
-    },
-    {
-      "epoch": 4.004186964048472,
-      "grad_norm": 0.0022613827604800463,
-      "learning_rate": 0.00019999209485605977,
-      "loss": 46.0,
-      "step": 24864
-    },
-    {
-      "epoch": 4.004348001127259,
-      "grad_norm": 0.002389657311141491,
-      "learning_rate": 0.0001999920942199289,
-      "loss": 46.0,
-      "step": 24865
-    },
-    {
-      "epoch": 4.004509038206047,
-      "grad_norm": 0.006755763199180365,
-      "learning_rate": 0.00019999209358377237,
-      "loss": 46.0,
-      "step": 24866
-    },
-    {
-      "epoch": 4.004670075284834,
-      "grad_norm": 0.0031629065051674843,
-      "learning_rate": 0.00019999209294759026,
-      "loss": 46.0,
-      "step": 24867
-    },
-    {
-      "epoch": 4.004831112363622,
-      "grad_norm": 0.0027789652813225985,
-      "learning_rate": 0.0001999920923113826,
-      "loss": 46.0,
-      "step": 24868
-    },
-    {
-      "epoch": 4.004992149442409,
-      "grad_norm": 0.006508123595267534,
-      "learning_rate": 0.0001999920916751493,
-      "loss": 46.0,
-      "step": 24869
-    },
-    {
-      "epoch": 4.005153186521197,
-      "grad_norm": 0.003923442680388689,
-      "learning_rate": 0.00019999209103889042,
-      "loss": 46.0,
-      "step": 24870
-    },
-    {
-      "epoch": 4.005314223599984,
-      "grad_norm": 0.004182691685855389,
-      "learning_rate": 0.000199992090402606,
-      "loss": 46.0,
-      "step": 24871
-    },
-    {
-      "epoch": 4.005475260678772,
-      "grad_norm": 0.01167905144393444,
-      "learning_rate": 0.0001999920897662959,
-      "loss": 46.0,
-      "step": 24872
-    },
-    {
-      "epoch": 4.005636297757559,
-      "grad_norm": 0.009988836012780666,
-      "learning_rate": 0.00019999208912996027,
-      "loss": 46.0,
-      "step": 24873
-    },
-    {
-      "epoch": 4.0057973348363465,
-      "grad_norm": 0.009333707392215729,
-      "learning_rate": 0.000199992088493599,
-      "loss": 46.0,
-      "step": 24874
-    },
-    {
-      "epoch": 4.005958371915133,
-      "grad_norm": 0.008291992358863354,
-      "learning_rate": 0.00019999208785721217,
-      "loss": 46.0,
-      "step": 24875
-    },
-    {
-      "epoch": 4.0061194089939205,
-      "grad_norm": 0.0009046962368302047,
-      "learning_rate": 0.00019999208722079975,
-      "loss": 46.0,
-      "step": 24876
-    },
-    {
-      "epoch": 4.006280446072708,
-      "grad_norm": 0.008579776622354984,
-      "learning_rate": 0.00019999208658436172,
-      "loss": 46.0,
-      "step": 24877
-    },
-    {
-      "epoch": 4.006441483151495,
-      "grad_norm": 0.0016432763077318668,
-      "learning_rate": 0.0001999920859478981,
-      "loss": 46.0,
-      "step": 24878
-    },
-    {
-      "epoch": 4.006602520230283,
-      "grad_norm": 0.008889056742191315,
-      "learning_rate": 0.0001999920853114089,
-      "loss": 46.0,
-      "step": 24879
-    },
-    {
-      "epoch": 4.00676355730907,
-      "grad_norm": 0.0012846224708482623,
-      "learning_rate": 0.00019999208467489407,
-      "loss": 46.0,
-      "step": 24880
-    },
-    {
-      "epoch": 4.006924594387858,
-      "grad_norm": 0.0062545230612158775,
-      "learning_rate": 0.0001999920840383537,
-      "loss": 46.0,
-      "step": 24881
-    },
-    {
-      "epoch": 4.007085631466645,
-      "grad_norm": 0.005658826325088739,
-      "learning_rate": 0.0001999920834017877,
-      "loss": 46.0,
-      "step": 24882
-    },
-    {
-      "epoch": 4.007246668545433,
-      "grad_norm": 0.0049173543229699135,
-      "learning_rate": 0.0001999920827651961,
-      "loss": 46.0,
-      "step": 24883
-    },
-    {
-      "epoch": 4.00740770562422,
-      "grad_norm": 0.006444412283599377,
-      "learning_rate": 0.00019999208212857897,
-      "loss": 46.0,
-      "step": 24884
-    },
-    {
-      "epoch": 4.007568742703008,
-      "grad_norm": 0.0014707319205626845,
-      "learning_rate": 0.00019999208149193618,
-      "loss": 46.0,
-      "step": 24885
-    },
-    {
-      "epoch": 4.007729779781795,
-      "grad_norm": 0.001380774425342679,
-      "learning_rate": 0.0001999920808552678,
-      "loss": 46.0,
-      "step": 24886
-    },
-    {
-      "epoch": 4.0078908168605825,
-      "grad_norm": 0.0029165069572627544,
-      "learning_rate": 0.00019999208021857388,
-      "loss": 46.0,
-      "step": 24887
-    },
-    {
-      "epoch": 4.00805185393937,
-      "grad_norm": 0.001213430194184184,
-      "learning_rate": 0.00019999207958185434,
-      "loss": 46.0,
-      "step": 24888
-    },
-    {
-      "epoch": 4.008212891018157,
-      "grad_norm": 0.0045651583932340145,
-      "learning_rate": 0.0001999920789451092,
-      "loss": 46.0,
-      "step": 24889
-    },
-    {
-      "epoch": 4.008373928096944,
-      "grad_norm": 0.005638986360281706,
-      "learning_rate": 0.00019999207830833848,
-      "loss": 46.0,
-      "step": 24890
-    },
-    {
-      "epoch": 4.008534965175731,
-      "grad_norm": 0.005020777229219675,
-      "learning_rate": 0.00019999207767154212,
-      "loss": 46.0,
-      "step": 24891
-    },
-    {
-      "epoch": 4.008696002254519,
-      "grad_norm": 0.0014963076682761312,
-      "learning_rate": 0.00019999207703472023,
-      "loss": 46.0,
-      "step": 24892
-    },
-    {
-      "epoch": 4.008857039333306,
-      "grad_norm": 0.006173077039420605,
-      "learning_rate": 0.00019999207639787272,
-      "loss": 46.0,
-      "step": 24893
-    },
-    {
-      "epoch": 4.009018076412094,
-      "grad_norm": 0.004271052777767181,
-      "learning_rate": 0.0001999920757609996,
-      "loss": 46.0,
-      "step": 24894
-    },
-    {
-      "epoch": 4.009179113490881,
-      "grad_norm": 0.00589071586728096,
-      "learning_rate": 0.00019999207512410092,
-      "loss": 46.0,
-      "step": 24895
-    },
-    {
-      "epoch": 4.009340150569669,
-      "grad_norm": 0.0028419415466487408,
-      "learning_rate": 0.00019999207448717665,
-      "loss": 46.0,
-      "step": 24896
-    },
-    {
-      "epoch": 4.009501187648456,
-      "grad_norm": 0.004104808904230595,
-      "learning_rate": 0.00019999207385022676,
-      "loss": 46.0,
-      "step": 24897
-    },
-    {
-      "epoch": 4.0096622247272435,
-      "grad_norm": 0.0016246193554252386,
-      "learning_rate": 0.0001999920732132513,
-      "loss": 46.0,
-      "step": 24898
-    },
-    {
-      "epoch": 4.009823261806031,
-      "grad_norm": 0.005655268207192421,
-      "learning_rate": 0.0001999920725762502,
-      "loss": 46.0,
-      "step": 24899
-    },
-    {
-      "epoch": 4.009984298884818,
-      "grad_norm": 0.007813077419996262,
-      "learning_rate": 0.00019999207193922356,
-      "loss": 46.0,
-      "step": 24900
-    },
-    {
-      "epoch": 4.010145335963606,
-      "grad_norm": 0.006974251940846443,
-      "learning_rate": 0.0001999920713021713,
-      "loss": 46.0,
-      "step": 24901
-    },
-    {
-      "epoch": 4.010306373042393,
-      "grad_norm": 0.002973036142066121,
-      "learning_rate": 0.00019999207066509348,
-      "loss": 46.0,
-      "step": 24902
-    },
-    {
-      "epoch": 4.010467410121181,
-      "grad_norm": 0.0018903139280155301,
-      "learning_rate": 0.00019999207002799002,
-      "loss": 46.0,
-      "step": 24903
-    },
-    {
-      "epoch": 4.010628447199967,
-      "grad_norm": 0.0013387489598244429,
-      "learning_rate": 0.000199992069390861,
-      "loss": 46.0,
-      "step": 24904
-    },
-    {
-      "epoch": 4.010789484278755,
-      "grad_norm": 0.00622066855430603,
-      "learning_rate": 0.0001999920687537064,
-      "loss": 46.0,
-      "step": 24905
-    },
-    {
-      "epoch": 4.010950521357542,
-      "grad_norm": 0.01253416296094656,
-      "learning_rate": 0.00019999206811652616,
-      "loss": 46.0,
-      "step": 24906
-    },
-    {
-      "epoch": 4.01111155843633,
-      "grad_norm": 0.0027325714472681284,
-      "learning_rate": 0.00019999206747932035,
-      "loss": 46.0,
-      "step": 24907
-    },
-    {
-      "epoch": 4.011272595515117,
-      "grad_norm": 0.006488606799393892,
-      "learning_rate": 0.00019999206684208893,
-      "loss": 46.0,
-      "step": 24908
-    },
-    {
-      "epoch": 4.011433632593905,
-      "grad_norm": 0.002637316007167101,
-      "learning_rate": 0.00019999206620483195,
-      "loss": 46.0,
-      "step": 24909
-    },
-    {
-      "epoch": 4.011594669672692,
-      "grad_norm": 0.004580279812216759,
-      "learning_rate": 0.00019999206556754937,
-      "loss": 46.0,
-      "step": 24910
-    },
-    {
-      "epoch": 4.0117557067514795,
-      "grad_norm": 0.0018053918611258268,
-      "learning_rate": 0.0001999920649302412,
-      "loss": 46.0,
-      "step": 24911
-    },
-    {
-      "epoch": 4.011916743830267,
-      "grad_norm": 0.004670732654631138,
-      "learning_rate": 0.0001999920642929074,
-      "loss": 46.0,
-      "step": 24912
-    },
-    {
-      "epoch": 4.012077780909054,
-      "grad_norm": 0.004771932028234005,
-      "learning_rate": 0.00019999206365554803,
-      "loss": 46.0,
-      "step": 24913
-    },
-    {
-      "epoch": 4.012238817987842,
-      "grad_norm": 0.0013259975239634514,
-      "learning_rate": 0.00019999206301816308,
-      "loss": 46.0,
-      "step": 24914
-    },
-    {
-      "epoch": 4.012399855066629,
-      "grad_norm": 0.004627016838639975,
-      "learning_rate": 0.00019999206238075255,
-      "loss": 46.0,
-      "step": 24915
-    },
-    {
-      "epoch": 4.012560892145417,
-      "grad_norm": 0.006728042382746935,
-      "learning_rate": 0.0001999920617433164,
-      "loss": 46.0,
-      "step": 24916
-    },
-    {
-      "epoch": 4.012721929224204,
-      "grad_norm": 0.0013815994607284665,
-      "learning_rate": 0.00019999206110585463,
-      "loss": 46.0,
-      "step": 24917
-    },
-    {
-      "epoch": 4.012882966302992,
-      "grad_norm": 0.008304712362587452,
-      "learning_rate": 0.00019999206046836734,
-      "loss": 46.0,
-      "step": 24918
-    },
-    {
-      "epoch": 4.013044003381778,
-      "grad_norm": 0.0037253228947520256,
-      "learning_rate": 0.0001999920598308544,
-      "loss": 46.0,
-      "step": 24919
-    },
-    {
-      "epoch": 4.013205040460566,
-      "grad_norm": 0.003998242784291506,
-      "learning_rate": 0.0001999920591933159,
-      "loss": 46.0,
-      "step": 24920
-    },
-    {
-      "epoch": 4.013366077539353,
-      "grad_norm": 0.0059352112002670765,
-      "learning_rate": 0.0001999920585557518,
-      "loss": 46.0,
-      "step": 24921
-    },
-    {
-      "epoch": 4.013527114618141,
-      "grad_norm": 0.004196269437670708,
-      "learning_rate": 0.0001999920579181621,
-      "loss": 46.0,
-      "step": 24922
-    },
-    {
-      "epoch": 4.013688151696928,
-      "grad_norm": 0.0044249314814805984,
-      "learning_rate": 0.0001999920572805468,
-      "loss": 46.0,
-      "step": 24923
-    },
-    {
-      "epoch": 4.0138491887757155,
-      "grad_norm": 0.001224488951265812,
-      "learning_rate": 0.0001999920566429059,
-      "loss": 46.0,
-      "step": 24924
-    },
-    {
-      "epoch": 4.014010225854503,
-      "grad_norm": 0.01328927744179964,
-      "learning_rate": 0.00019999205600523942,
-      "loss": 46.0,
-      "step": 24925
-    },
-    {
-      "epoch": 4.01417126293329,
-      "grad_norm": 0.0035148095339536667,
-      "learning_rate": 0.00019999205536754734,
-      "loss": 46.0,
-      "step": 24926
-    },
-    {
-      "epoch": 4.014332300012078,
-      "grad_norm": 0.0024550568778067827,
-      "learning_rate": 0.00019999205472982968,
-      "loss": 46.0,
-      "step": 24927
-    },
-    {
-      "epoch": 4.014493337090865,
-      "grad_norm": 0.002423433819785714,
-      "learning_rate": 0.0001999920540920864,
-      "loss": 46.0,
-      "step": 24928
-    },
-    {
-      "epoch": 4.014654374169653,
-      "grad_norm": 0.001050095190294087,
-      "learning_rate": 0.00019999205345431757,
-      "loss": 46.0,
-      "step": 24929
-    },
-    {
-      "epoch": 4.01481541124844,
-      "grad_norm": 0.005810847505927086,
-      "learning_rate": 0.00019999205281652315,
-      "loss": 46.0,
-      "step": 24930
-    },
-    {
-      "epoch": 4.014976448327228,
-      "grad_norm": 0.002049764385446906,
-      "learning_rate": 0.00019999205217870308,
-      "loss": 46.0,
-      "step": 24931
-    },
-    {
-      "epoch": 4.015137485406015,
-      "grad_norm": 0.0035755804274231195,
-      "learning_rate": 0.00019999205154085746,
-      "loss": 46.0,
-      "step": 24932
-    },
-    {
-      "epoch": 4.015298522484803,
-      "grad_norm": 0.0034430858213454485,
-      "learning_rate": 0.00019999205090298622,
-      "loss": 46.0,
-      "step": 24933
-    },
-    {
-      "epoch": 4.015459559563589,
-      "grad_norm": 0.021088844165205956,
-      "learning_rate": 0.00019999205026508942,
-      "loss": 46.0,
-      "step": 24934
-    },
-    {
-      "epoch": 4.015620596642377,
-      "grad_norm": 0.0028874108102172613,
-      "learning_rate": 0.00019999204962716703,
-      "loss": 46.0,
-      "step": 24935
-    },
-    {
-      "epoch": 4.015781633721164,
-      "grad_norm": 0.011025993153452873,
-      "learning_rate": 0.000199992048989219,
-      "loss": 46.0,
-      "step": 24936
-    },
-    {
-      "epoch": 4.0159426707999515,
-      "grad_norm": 0.02030302956700325,
-      "learning_rate": 0.00019999204835124542,
-      "loss": 46.0,
-      "step": 24937
-    },
-    {
-      "epoch": 4.016103707878739,
-      "grad_norm": 0.0024919710122048855,
-      "learning_rate": 0.00019999204771324624,
-      "loss": 46.0,
-      "step": 24938
-    },
-    {
-      "epoch": 4.016264744957526,
-      "grad_norm": 0.005026031751185656,
-      "learning_rate": 0.00019999204707522145,
-      "loss": 46.0,
-      "step": 24939
-    },
-    {
-      "epoch": 4.016425782036314,
-      "grad_norm": 0.005144732538610697,
-      "learning_rate": 0.00019999204643717108,
-      "loss": 46.0,
-      "step": 24940
-    },
-    {
-      "epoch": 4.016586819115101,
-      "grad_norm": 0.0009662075317464769,
-      "learning_rate": 0.0001999920457990951,
-      "loss": 46.0,
-      "step": 24941
-    },
-    {
-      "epoch": 4.016747856193889,
-      "grad_norm": 0.010334135964512825,
-      "learning_rate": 0.00019999204516099356,
-      "loss": 46.0,
-      "step": 24942
-    },
-    {
-      "epoch": 4.016908893272676,
-      "grad_norm": 0.0017033818876370788,
-      "learning_rate": 0.00019999204452286642,
-      "loss": 46.0,
-      "step": 24943
-    },
-    {
-      "epoch": 4.017069930351464,
-      "grad_norm": 0.007641301490366459,
-      "learning_rate": 0.0001999920438847137,
-      "loss": 46.0,
-      "step": 24944
-    },
-    {
-      "epoch": 4.017230967430251,
-      "grad_norm": 0.0037254299968481064,
-      "learning_rate": 0.00019999204324653533,
-      "loss": 46.0,
-      "step": 24945
-    },
-    {
-      "epoch": 4.0173920045090385,
-      "grad_norm": 0.002317138249054551,
-      "learning_rate": 0.0001999920426083314,
-      "loss": 46.0,
-      "step": 24946
-    },
-    {
-      "epoch": 4.017553041587826,
-      "grad_norm": 0.0012140922481194139,
-      "learning_rate": 0.0001999920419701019,
-      "loss": 46.0,
-      "step": 24947
-    },
-    {
-      "epoch": 4.017714078666613,
-      "grad_norm": 0.003595631802454591,
-      "learning_rate": 0.00019999204133184676,
-      "loss": 46.0,
-      "step": 24948
-    },
-    {
-      "epoch": 4.0178751157454,
-      "grad_norm": 0.015302340500056744,
-      "learning_rate": 0.00019999204069356608,
-      "loss": 46.0,
-      "step": 24949
-    },
-    {
-      "epoch": 4.0180361528241875,
-      "grad_norm": 0.003522982355207205,
-      "learning_rate": 0.00019999204005525975,
-      "loss": 46.0,
-      "step": 24950
-    },
-    {
-      "epoch": 4.018197189902975,
-      "grad_norm": 0.008157297968864441,
-      "learning_rate": 0.00019999203941692786,
-      "loss": 46.0,
-      "step": 24951
-    },
-    {
-      "epoch": 4.018358226981762,
-      "grad_norm": 0.004223973024636507,
-      "learning_rate": 0.00019999203877857038,
-      "loss": 46.0,
-      "step": 24952
-    },
-    {
-      "epoch": 4.01851926406055,
-      "grad_norm": 0.011615936644375324,
-      "learning_rate": 0.0001999920381401873,
-      "loss": 46.0,
-      "step": 24953
-    },
-    {
-      "epoch": 4.018680301139337,
-      "grad_norm": 0.0013157908106222749,
-      "learning_rate": 0.0001999920375017786,
-      "loss": 46.0,
-      "step": 24954
-    },
-    {
-      "epoch": 4.018841338218125,
-      "grad_norm": 0.0023194567766040564,
-      "learning_rate": 0.00019999203686334437,
-      "loss": 46.0,
-      "step": 24955
-    },
-    {
-      "epoch": 4.019002375296912,
-      "grad_norm": 0.0052051842212677,
-      "learning_rate": 0.00019999203622488452,
-      "loss": 46.0,
-      "step": 24956
-    },
-    {
-      "epoch": 4.0191634123757,
-      "grad_norm": 0.0013000270118936896,
-      "learning_rate": 0.00019999203558639908,
-      "loss": 46.0,
-      "step": 24957
-    },
-    {
-      "epoch": 4.019324449454487,
-      "grad_norm": 0.00223491876386106,
-      "learning_rate": 0.000199992034947888,
-      "loss": 46.0,
-      "step": 24958
-    },
-    {
-      "epoch": 4.0194854865332745,
-      "grad_norm": 0.0036144836340099573,
-      "learning_rate": 0.00019999203430935139,
-      "loss": 46.0,
-      "step": 24959
-    },
-    {
-      "epoch": 4.019646523612062,
-      "grad_norm": 0.0016257986426353455,
-      "learning_rate": 0.00019999203367078913,
-      "loss": 46.0,
-      "step": 24960
-    },
-    {
-      "epoch": 4.019807560690849,
-      "grad_norm": 0.0010201609693467617,
-      "learning_rate": 0.00019999203303220134,
-      "loss": 46.0,
-      "step": 24961
-    },
-    {
-      "epoch": 4.019968597769637,
-      "grad_norm": 0.001547725056298077,
-      "learning_rate": 0.0001999920323935879,
-      "loss": 46.0,
-      "step": 24962
-    },
-    {
-      "epoch": 4.0201296348484234,
-      "grad_norm": 0.006314681842923164,
-      "learning_rate": 0.0001999920317549489,
-      "loss": 46.0,
-      "step": 24963
-    },
-    {
-      "epoch": 4.020290671927211,
-      "grad_norm": 0.0016107544070109725,
-      "learning_rate": 0.00019999203111628432,
-      "loss": 46.0,
-      "step": 24964
-    },
-    {
-      "epoch": 4.020451709005998,
-      "grad_norm": 0.0024504787288606167,
-      "learning_rate": 0.00019999203047759413,
-      "loss": 46.0,
-      "step": 24965
-    },
-    {
-      "epoch": 4.020612746084786,
-      "grad_norm": 0.002583805937319994,
-      "learning_rate": 0.00019999202983887832,
-      "loss": 46.0,
-      "step": 24966
-    },
-    {
-      "epoch": 4.020773783163573,
-      "grad_norm": 0.004775282461196184,
-      "learning_rate": 0.00019999202920013696,
-      "loss": 46.0,
-      "step": 24967
-    },
-    {
-      "epoch": 4.020934820242361,
-      "grad_norm": 0.015760183334350586,
-      "learning_rate": 0.00019999202856136998,
-      "loss": 46.0,
-      "step": 24968
-    },
-    {
-      "epoch": 4.021095857321148,
-      "grad_norm": 0.0076279910281300545,
-      "learning_rate": 0.0001999920279225774,
-      "loss": 46.0,
-      "step": 24969
-    },
-    {
-      "epoch": 4.021256894399936,
-      "grad_norm": 0.002553044818341732,
-      "learning_rate": 0.00019999202728375926,
-      "loss": 46.0,
-      "step": 24970
-    },
-    {
-      "epoch": 4.021417931478723,
-      "grad_norm": 0.002563176676630974,
-      "learning_rate": 0.00019999202664491551,
-      "loss": 46.0,
-      "step": 24971
-    },
-    {
-      "epoch": 4.0215789685575105,
-      "grad_norm": 0.010500002652406693,
-      "learning_rate": 0.00019999202600604616,
-      "loss": 46.0,
-      "step": 24972
-    },
-    {
-      "epoch": 4.021740005636298,
-      "grad_norm": 0.0006657430785708129,
-      "learning_rate": 0.00019999202536715122,
-      "loss": 46.0,
-      "step": 24973
-    },
-    {
-      "epoch": 4.021901042715085,
-      "grad_norm": 0.0030671637505292892,
-      "learning_rate": 0.0001999920247282307,
-      "loss": 46.0,
-      "step": 24974
-    },
-    {
-      "epoch": 4.022062079793873,
-      "grad_norm": 0.00227194931358099,
-      "learning_rate": 0.00019999202408928457,
-      "loss": 46.0,
-      "step": 24975
-    },
-    {
-      "epoch": 4.02222311687266,
-      "grad_norm": 0.01887577958405018,
-      "learning_rate": 0.00019999202345031286,
-      "loss": 46.0,
-      "step": 24976
-    },
-    {
-      "epoch": 4.022384153951447,
-      "grad_norm": 0.002493244130164385,
-      "learning_rate": 0.00019999202281131555,
-      "loss": 46.0,
-      "step": 24977
-    },
-    {
-      "epoch": 4.022545191030234,
-      "grad_norm": 0.0019332539523020387,
-      "learning_rate": 0.00019999202217229267,
-      "loss": 46.0,
-      "step": 24978
-    },
-    {
-      "epoch": 4.022706228109022,
-      "grad_norm": 0.01769905537366867,
-      "learning_rate": 0.00019999202153324417,
-      "loss": 46.0,
-      "step": 24979
-    },
-    {
-      "epoch": 4.022867265187809,
-      "grad_norm": 0.001634084852412343,
-      "learning_rate": 0.0001999920208941701,
-      "loss": 46.0,
-      "step": 24980
-    },
-    {
-      "epoch": 4.023028302266597,
-      "grad_norm": 0.004199749790132046,
-      "learning_rate": 0.0001999920202550704,
-      "loss": 46.0,
-      "step": 24981
-    },
-    {
-      "epoch": 4.023189339345384,
-      "grad_norm": 0.0019532626029103994,
-      "learning_rate": 0.00019999201961594514,
-      "loss": 46.0,
-      "step": 24982
-    },
-    {
-      "epoch": 4.023350376424172,
-      "grad_norm": 0.002360276412218809,
-      "learning_rate": 0.00019999201897679428,
-      "loss": 46.0,
-      "step": 24983
-    },
-    {
-      "epoch": 4.023511413502959,
-      "grad_norm": 0.010797782801091671,
-      "learning_rate": 0.00019999201833761782,
-      "loss": 46.0,
-      "step": 24984
-    },
-    {
-      "epoch": 4.0236724505817465,
-      "grad_norm": 0.0034726497251540422,
-      "learning_rate": 0.00019999201769841575,
-      "loss": 46.0,
-      "step": 24985
-    },
-    {
-      "epoch": 4.023833487660534,
-      "grad_norm": 0.008773855865001678,
-      "learning_rate": 0.00019999201705918812,
-      "loss": 46.0,
-      "step": 24986
-    },
-    {
-      "epoch": 4.023994524739321,
-      "grad_norm": 0.0036492610815912485,
-      "learning_rate": 0.00019999201641993488,
-      "loss": 46.0,
-      "step": 24987
-    },
-    {
-      "epoch": 4.024155561818109,
-      "grad_norm": 0.005425607319921255,
-      "learning_rate": 0.00019999201578065607,
-      "loss": 46.0,
-      "step": 24988
-    },
-    {
-      "epoch": 4.024316598896896,
-      "grad_norm": 0.0035622103605419397,
-      "learning_rate": 0.00019999201514135165,
-      "loss": 46.0,
-      "step": 24989
-    },
-    {
-      "epoch": 4.024477635975684,
-      "grad_norm": 0.004134961403906345,
-      "learning_rate": 0.00019999201450202162,
-      "loss": 46.0,
-      "step": 24990
-    },
-    {
-      "epoch": 4.024638673054471,
-      "grad_norm": 0.009632492437958717,
-      "learning_rate": 0.00019999201386266603,
-      "loss": 46.0,
-      "step": 24991
-    },
-    {
-      "epoch": 4.024799710133258,
-      "grad_norm": 0.0043112351559102535,
-      "learning_rate": 0.00019999201322328482,
-      "loss": 46.0,
-      "step": 24992
-    },
-    {
-      "epoch": 4.024960747212045,
-      "grad_norm": 0.001084771123714745,
-      "learning_rate": 0.00019999201258387802,
-      "loss": 46.0,
-      "step": 24993
-    },
-    {
-      "epoch": 4.025121784290833,
-      "grad_norm": 0.011641322635114193,
-      "learning_rate": 0.00019999201194444567,
-      "loss": 46.0,
-      "step": 24994
-    },
-    {
-      "epoch": 4.02528282136962,
-      "grad_norm": 0.006057314109057188,
-      "learning_rate": 0.00019999201130498765,
-      "loss": 46.0,
-      "step": 24995
-    },
-    {
-      "epoch": 4.025443858448408,
-      "grad_norm": 0.0020053863991051912,
-      "learning_rate": 0.0001999920106655041,
-      "loss": 46.0,
-      "step": 24996
-    },
-    {
-      "epoch": 4.025604895527195,
-      "grad_norm": 0.005097714252769947,
-      "learning_rate": 0.00019999201002599492,
-      "loss": 46.0,
-      "step": 24997
-    },
-    {
-      "epoch": 4.0257659326059825,
-      "grad_norm": 0.00591740757226944,
-      "learning_rate": 0.0001999920093864602,
-      "loss": 46.0,
-      "step": 24998
-    },
-    {
-      "epoch": 4.02592696968477,
-      "grad_norm": 0.00429281173273921,
-      "learning_rate": 0.00019999200874689982,
-      "loss": 46.0,
-      "step": 24999
-    },
-    {
-      "epoch": 4.026088006763557,
-      "grad_norm": 0.009144529700279236,
-      "learning_rate": 0.00019999200810731388,
-      "loss": 46.0,
-      "step": 25000
-    },
-    {
-      "epoch": 4.026249043842345,
-      "grad_norm": 0.002104668179526925,
-      "learning_rate": 0.00019999200746770234,
-      "loss": 46.0,
-      "step": 25001
-    },
-    {
-      "epoch": 4.026410080921132,
-      "grad_norm": 0.0022079164627939463,
-      "learning_rate": 0.00019999200682806523,
-      "loss": 46.0,
-      "step": 25002
-    },
-    {
-      "epoch": 4.02657111799992,
-      "grad_norm": 0.0010310064535588026,
-      "learning_rate": 0.0001999920061884025,
-      "loss": 46.0,
-      "step": 25003
-    },
-    {
-      "epoch": 4.026732155078707,
-      "grad_norm": 0.0016190280439332128,
-      "learning_rate": 0.0001999920055487142,
-      "loss": 46.0,
-      "step": 25004
-    },
-    {
-      "epoch": 4.026893192157495,
-      "grad_norm": 0.0067587136290967464,
-      "learning_rate": 0.00019999200490900028,
-      "loss": 46.0,
-      "step": 25005
-    },
-    {
-      "epoch": 4.027054229236282,
-      "grad_norm": 0.005166187416762114,
-      "learning_rate": 0.00019999200426926077,
-      "loss": 46.0,
-      "step": 25006
-    },
-    {
-      "epoch": 4.027215266315069,
-      "grad_norm": 0.012126859277486801,
-      "learning_rate": 0.00019999200362949568,
-      "loss": 46.0,
-      "step": 25007
-    },
-    {
-      "epoch": 4.027376303393856,
-      "grad_norm": 0.0038047428242862225,
-      "learning_rate": 0.00019999200298970502,
-      "loss": 46.0,
-      "step": 25008
-    },
-    {
-      "epoch": 4.027537340472644,
-      "grad_norm": 0.008614459075033665,
-      "learning_rate": 0.00019999200234988872,
-      "loss": 46.0,
-      "step": 25009
-    },
-    {
-      "epoch": 4.027698377551431,
-      "grad_norm": 0.0038431184366345406,
-      "learning_rate": 0.00019999200171004686,
-      "loss": 46.0,
-      "step": 25010
-    },
-    {
-      "epoch": 4.0278594146302185,
-      "grad_norm": 0.004231929313391447,
-      "learning_rate": 0.00019999200107017942,
-      "loss": 46.0,
-      "step": 25011
-    },
-    {
-      "epoch": 4.028020451709006,
-      "grad_norm": 0.0010132197057828307,
-      "learning_rate": 0.00019999200043028633,
-      "loss": 46.0,
-      "step": 25012
-    },
-    {
-      "epoch": 4.028181488787793,
-      "grad_norm": 0.001272825407795608,
-      "learning_rate": 0.00019999199979036769,
-      "loss": 46.0,
-      "step": 25013
-    },
-    {
-      "epoch": 4.028342525866581,
-      "grad_norm": 0.0025782862212508917,
-      "learning_rate": 0.00019999199915042345,
-      "loss": 46.0,
-      "step": 25014
-    },
-    {
-      "epoch": 4.028503562945368,
-      "grad_norm": 0.0019017639569938183,
-      "learning_rate": 0.0001999919985104536,
-      "loss": 46.0,
-      "step": 25015
-    },
-    {
-      "epoch": 4.028664600024156,
-      "grad_norm": 0.0020026471465826035,
-      "learning_rate": 0.0001999919978704582,
-      "loss": 46.0,
-      "step": 25016
-    },
-    {
-      "epoch": 4.028825637102943,
-      "grad_norm": 0.0023075321223586798,
-      "learning_rate": 0.00019999199723043715,
-      "loss": 46.0,
-      "step": 25017
-    },
-    {
-      "epoch": 4.028986674181731,
-      "grad_norm": 0.002112357411533594,
-      "learning_rate": 0.00019999199659039056,
-      "loss": 46.0,
-      "step": 25018
-    },
-    {
-      "epoch": 4.029147711260518,
-      "grad_norm": 0.002465772209689021,
-      "learning_rate": 0.00019999199595031834,
-      "loss": 46.0,
-      "step": 25019
-    },
-    {
-      "epoch": 4.0293087483393055,
-      "grad_norm": 0.010843264870345592,
-      "learning_rate": 0.00019999199531022056,
-      "loss": 46.0,
-      "step": 25020
-    },
-    {
-      "epoch": 4.029469785418092,
-      "grad_norm": 0.0015449357451871037,
-      "learning_rate": 0.00019999199467009716,
-      "loss": 46.0,
-      "step": 25021
-    },
-    {
-      "epoch": 4.0296308224968795,
-      "grad_norm": 0.006690714042633772,
-      "learning_rate": 0.00019999199402994817,
-      "loss": 46.0,
-      "step": 25022
-    },
-    {
-      "epoch": 4.029791859575667,
-      "grad_norm": 0.006099675316363573,
-      "learning_rate": 0.0001999919933897736,
-      "loss": 46.0,
-      "step": 25023
-    },
-    {
-      "epoch": 4.029952896654454,
-      "grad_norm": 0.015103408135473728,
-      "learning_rate": 0.00019999199274957344,
-      "loss": 46.0,
-      "step": 25024
-    },
-    {
-      "epoch": 4.030113933733242,
-      "grad_norm": 0.004041378851979971,
-      "learning_rate": 0.00019999199210934767,
-      "loss": 46.0,
-      "step": 25025
-    },
-    {
-      "epoch": 4.030274970812029,
-      "grad_norm": 0.0014146839966997504,
-      "learning_rate": 0.00019999199146909633,
-      "loss": 46.0,
-      "step": 25026
-    },
-    {
-      "epoch": 4.030436007890817,
-      "grad_norm": 0.010885761119425297,
-      "learning_rate": 0.00019999199082881936,
-      "loss": 46.0,
-      "step": 25027
-    },
-    {
-      "epoch": 4.030597044969604,
-      "grad_norm": 0.002577926032245159,
-      "learning_rate": 0.00019999199018851682,
-      "loss": 46.0,
-      "step": 25028
-    },
-    {
-      "epoch": 4.030758082048392,
-      "grad_norm": 0.002156562637537718,
-      "learning_rate": 0.0001999919895481887,
-      "loss": 46.0,
-      "step": 25029
-    },
-    {
-      "epoch": 4.030919119127179,
-      "grad_norm": 0.006228687707334757,
-      "learning_rate": 0.00019999198890783496,
-      "loss": 46.0,
-      "step": 25030
-    },
-    {
-      "epoch": 4.031080156205967,
-      "grad_norm": 0.002927055349573493,
-      "learning_rate": 0.00019999198826745564,
-      "loss": 46.0,
-      "step": 25031
-    },
-    {
-      "epoch": 4.031241193284754,
-      "grad_norm": 0.015131741762161255,
-      "learning_rate": 0.00019999198762705073,
-      "loss": 46.0,
-      "step": 25032
-    },
-    {
-      "epoch": 4.0314022303635415,
-      "grad_norm": 0.008905814960598946,
-      "learning_rate": 0.00019999198698662023,
-      "loss": 46.0,
-      "step": 25033
-    },
-    {
-      "epoch": 4.031563267442329,
-      "grad_norm": 0.0007261548307724297,
-      "learning_rate": 0.00019999198634616412,
-      "loss": 46.0,
-      "step": 25034
-    },
-    {
-      "epoch": 4.031724304521116,
-      "grad_norm": 0.009940962307155132,
-      "learning_rate": 0.00019999198570568244,
-      "loss": 46.0,
-      "step": 25035
-    },
-    {
-      "epoch": 4.031885341599903,
-      "grad_norm": 0.0009617822943255305,
-      "learning_rate": 0.00019999198506517516,
-      "loss": 46.0,
-      "step": 25036
-    },
-    {
-      "epoch": 4.03204637867869,
-      "grad_norm": 0.002603802364319563,
-      "learning_rate": 0.00019999198442464228,
-      "loss": 46.0,
-      "step": 25037
-    },
-    {
-      "epoch": 4.032207415757478,
-      "grad_norm": 0.000773383944761008,
-      "learning_rate": 0.00019999198378408382,
-      "loss": 46.0,
-      "step": 25038
-    },
-    {
-      "epoch": 4.032368452836265,
-      "grad_norm": 0.0027535571716725826,
-      "learning_rate": 0.00019999198314349977,
-      "loss": 46.0,
-      "step": 25039
-    },
-    {
-      "epoch": 4.032529489915053,
-      "grad_norm": 0.0035872994922101498,
-      "learning_rate": 0.0001999919825028901,
-      "loss": 46.0,
-      "step": 25040
-    },
-    {
-      "epoch": 4.03269052699384,
-      "grad_norm": 0.008839412592351437,
-      "learning_rate": 0.00019999198186225486,
-      "loss": 46.0,
-      "step": 25041
-    },
-    {
-      "epoch": 4.032851564072628,
-      "grad_norm": 0.00111858278978616,
-      "learning_rate": 0.00019999198122159402,
-      "loss": 46.0,
-      "step": 25042
-    },
-    {
-      "epoch": 4.033012601151415,
-      "grad_norm": 0.0018319826340302825,
-      "learning_rate": 0.00019999198058090757,
-      "loss": 46.0,
-      "step": 25043
-    },
-    {
-      "epoch": 4.033173638230203,
-      "grad_norm": 0.003526741173118353,
-      "learning_rate": 0.00019999197994019556,
-      "loss": 46.0,
-      "step": 25044
-    },
-    {
-      "epoch": 4.03333467530899,
-      "grad_norm": 0.0012674665777012706,
-      "learning_rate": 0.0001999919792994579,
-      "loss": 46.0,
-      "step": 25045
-    },
-    {
-      "epoch": 4.0334957123877775,
-      "grad_norm": 0.0018450462957844138,
-      "learning_rate": 0.00019999197865869472,
-      "loss": 46.0,
-      "step": 25046
-    },
-    {
-      "epoch": 4.033656749466565,
-      "grad_norm": 0.004494448192417622,
-      "learning_rate": 0.0001999919780179059,
-      "loss": 46.0,
-      "step": 25047
-    },
-    {
-      "epoch": 4.033817786545352,
-      "grad_norm": 0.0012645210372284055,
-      "learning_rate": 0.0001999919773770915,
-      "loss": 46.0,
-      "step": 25048
-    },
-    {
-      "epoch": 4.03397882362414,
-      "grad_norm": 0.00114197819493711,
-      "learning_rate": 0.0001999919767362515,
-      "loss": 46.0,
-      "step": 25049
-    },
-    {
-      "epoch": 4.034139860702927,
-      "grad_norm": 0.0013513603480532765,
-      "learning_rate": 0.00019999197609538595,
-      "loss": 46.0,
-      "step": 25050
-    },
-    {
-      "epoch": 4.034300897781714,
-      "grad_norm": 0.006062310189008713,
-      "learning_rate": 0.00019999197545449477,
-      "loss": 46.0,
-      "step": 25051
-    },
-    {
-      "epoch": 4.034461934860501,
-      "grad_norm": 0.002342484425753355,
-      "learning_rate": 0.00019999197481357798,
-      "loss": 46.0,
-      "step": 25052
-    },
-    {
-      "epoch": 4.034622971939289,
-      "grad_norm": 0.00553166214376688,
-      "learning_rate": 0.00019999197417263566,
-      "loss": 46.0,
-      "step": 25053
-    },
-    {
-      "epoch": 4.034784009018076,
-      "grad_norm": 0.016203725710511208,
-      "learning_rate": 0.00019999197353166766,
-      "loss": 46.0,
-      "step": 25054
-    },
-    {
-      "epoch": 4.034945046096864,
-      "grad_norm": 0.008070418611168861,
-      "learning_rate": 0.00019999197289067414,
-      "loss": 46.0,
-      "step": 25055
-    },
-    {
-      "epoch": 4.035106083175651,
-      "grad_norm": 0.01761159673333168,
-      "learning_rate": 0.000199991972249655,
-      "loss": 46.0,
-      "step": 25056
-    },
-    {
-      "epoch": 4.035267120254439,
-      "grad_norm": 0.0008269277750514448,
-      "learning_rate": 0.00019999197160861028,
-      "loss": 46.0,
-      "step": 25057
-    },
-    {
-      "epoch": 4.035428157333226,
-      "grad_norm": 0.010538355447351933,
-      "learning_rate": 0.00019999197096753994,
-      "loss": 46.0,
-      "step": 25058
-    },
-    {
-      "epoch": 4.0355891944120135,
-      "grad_norm": 0.0006135565345175564,
-      "learning_rate": 0.00019999197032644403,
-      "loss": 46.0,
-      "step": 25059
-    },
-    {
-      "epoch": 4.035750231490801,
-      "grad_norm": 0.002369600348174572,
-      "learning_rate": 0.00019999196968532252,
-      "loss": 46.0,
-      "step": 25060
-    },
-    {
-      "epoch": 4.035911268569588,
-      "grad_norm": 0.0025276204105466604,
-      "learning_rate": 0.00019999196904417542,
-      "loss": 46.0,
-      "step": 25061
-    },
-    {
-      "epoch": 4.036072305648376,
-      "grad_norm": 0.007783102337270975,
-      "learning_rate": 0.00019999196840300273,
-      "loss": 46.0,
-      "step": 25062
-    },
-    {
-      "epoch": 4.036233342727163,
-      "grad_norm": 0.002237935084849596,
-      "learning_rate": 0.00019999196776180443,
-      "loss": 46.0,
-      "step": 25063
-    },
-    {
-      "epoch": 4.036394379805951,
-      "grad_norm": 0.0032588159665465355,
-      "learning_rate": 0.00019999196712058054,
-      "loss": 46.0,
-      "step": 25064
-    },
-    {
-      "epoch": 4.036555416884737,
-      "grad_norm": 0.0023829364217817783,
-      "learning_rate": 0.00019999196647933109,
-      "loss": 46.0,
-      "step": 25065
-    },
-    {
-      "epoch": 4.036716453963525,
-      "grad_norm": 0.020837897434830666,
-      "learning_rate": 0.000199991965838056,
-      "loss": 46.0,
-      "step": 25066
-    },
-    {
-      "epoch": 4.036877491042312,
-      "grad_norm": 0.004435418639332056,
-      "learning_rate": 0.00019999196519675534,
-      "loss": 46.0,
-      "step": 25067
-    },
-    {
-      "epoch": 4.0370385281211,
-      "grad_norm": 0.0006732947658747435,
-      "learning_rate": 0.00019999196455542908,
-      "loss": 46.0,
-      "step": 25068
-    },
-    {
-      "epoch": 4.037199565199887,
-      "grad_norm": 0.004389289300888777,
-      "learning_rate": 0.00019999196391407725,
-      "loss": 46.0,
-      "step": 25069
-    },
-    {
-      "epoch": 4.0373606022786745,
-      "grad_norm": 0.004719639662653208,
-      "learning_rate": 0.0001999919632726998,
-      "loss": 46.0,
-      "step": 25070
-    },
-    {
-      "epoch": 4.037521639357462,
-      "grad_norm": 0.006348883267492056,
-      "learning_rate": 0.00019999196263129675,
-      "loss": 46.0,
-      "step": 25071
-    },
-    {
-      "epoch": 4.0376826764362495,
-      "grad_norm": 0.0005346002290025353,
-      "learning_rate": 0.00019999196198986817,
-      "loss": 46.0,
-      "step": 25072
-    },
-    {
-      "epoch": 4.037843713515037,
-      "grad_norm": 0.01372213289141655,
-      "learning_rate": 0.00019999196134841394,
-      "loss": 46.0,
-      "step": 25073
-    },
-    {
-      "epoch": 4.038004750593824,
-      "grad_norm": 0.0025717909447848797,
-      "learning_rate": 0.00019999196070693412,
-      "loss": 46.0,
-      "step": 25074
-    },
-    {
-      "epoch": 4.038165787672612,
-      "grad_norm": 0.0019021675689145923,
-      "learning_rate": 0.00019999196006542872,
-      "loss": 46.0,
-      "step": 25075
-    },
-    {
-      "epoch": 4.038326824751399,
-      "grad_norm": 0.0017507259035483003,
-      "learning_rate": 0.0001999919594238977,
-      "loss": 46.0,
-      "step": 25076
-    },
-    {
-      "epoch": 4.038487861830187,
-      "grad_norm": 0.0019939723424613476,
-      "learning_rate": 0.00019999195878234112,
-      "loss": 46.0,
-      "step": 25077
-    },
-    {
-      "epoch": 4.038648898908974,
-      "grad_norm": 0.0016106757102534175,
-      "learning_rate": 0.00019999195814075896,
-      "loss": 46.0,
-      "step": 25078
-    },
-    {
-      "epoch": 4.038809935987762,
-      "grad_norm": 0.0068449340760707855,
-      "learning_rate": 0.00019999195749915115,
-      "loss": 46.0,
-      "step": 25079
-    },
-    {
-      "epoch": 4.038970973066548,
-      "grad_norm": 0.007759213447570801,
-      "learning_rate": 0.00019999195685751778,
-      "loss": 46.0,
-      "step": 25080
-    },
-    {
-      "epoch": 4.039132010145336,
-      "grad_norm": 0.0021500838920474052,
-      "learning_rate": 0.00019999195621585886,
-      "loss": 46.0,
-      "step": 25081
-    },
-    {
-      "epoch": 4.039293047224123,
-      "grad_norm": 0.0017113228095695376,
-      "learning_rate": 0.0001999919555741743,
-      "loss": 46.0,
-      "step": 25082
-    },
-    {
-      "epoch": 4.0394540843029105,
-      "grad_norm": 0.0013238689862191677,
-      "learning_rate": 0.00019999195493246413,
-      "loss": 46.0,
-      "step": 25083
-    },
-    {
-      "epoch": 4.039615121381698,
-      "grad_norm": 0.0016637734370306134,
-      "learning_rate": 0.00019999195429072842,
-      "loss": 46.0,
-      "step": 25084
-    },
-    {
-      "epoch": 4.039776158460485,
-      "grad_norm": 0.00834680162370205,
-      "learning_rate": 0.0001999919536489671,
-      "loss": 46.0,
-      "step": 25085
-    },
-    {
-      "epoch": 4.039937195539273,
-      "grad_norm": 0.0008648328948765993,
-      "learning_rate": 0.00019999195300718015,
-      "loss": 46.0,
-      "step": 25086
-    },
-    {
-      "epoch": 4.04009823261806,
-      "grad_norm": 0.0017787946853786707,
-      "learning_rate": 0.00019999195236536764,
-      "loss": 46.0,
-      "step": 25087
-    },
-    {
-      "epoch": 4.040259269696848,
-      "grad_norm": 0.003584889229387045,
-      "learning_rate": 0.00019999195172352952,
-      "loss": 46.0,
-      "step": 25088
-    },
-    {
-      "epoch": 4.040420306775635,
-      "grad_norm": 0.007776655722409487,
-      "learning_rate": 0.00019999195108166582,
-      "loss": 46.0,
-      "step": 25089
-    },
-    {
-      "epoch": 4.040581343854423,
-      "grad_norm": 0.0030605753418058157,
-      "learning_rate": 0.00019999195043977653,
-      "loss": 46.0,
-      "step": 25090
-    },
-    {
-      "epoch": 4.04074238093321,
-      "grad_norm": 0.0029180629644542933,
-      "learning_rate": 0.00019999194979786165,
-      "loss": 46.0,
-      "step": 25091
-    },
-    {
-      "epoch": 4.040903418011998,
-      "grad_norm": 0.001516191172413528,
-      "learning_rate": 0.00019999194915592115,
-      "loss": 46.0,
-      "step": 25092
-    },
-    {
-      "epoch": 4.041064455090785,
-      "grad_norm": 0.0032532494515180588,
-      "learning_rate": 0.00019999194851395507,
-      "loss": 46.0,
-      "step": 25093
-    },
-    {
-      "epoch": 4.041225492169572,
-      "grad_norm": 0.007593729067593813,
-      "learning_rate": 0.00019999194787196343,
-      "loss": 46.0,
-      "step": 25094
-    },
-    {
-      "epoch": 4.041386529248359,
-      "grad_norm": 0.003355221124365926,
-      "learning_rate": 0.00019999194722994618,
-      "loss": 46.0,
-      "step": 25095
-    },
-    {
-      "epoch": 4.0415475663271465,
-      "grad_norm": 0.008278166875243187,
-      "learning_rate": 0.0001999919465879033,
-      "loss": 46.0,
-      "step": 25096
-    },
-    {
-      "epoch": 4.041708603405934,
-      "grad_norm": 0.004897734150290489,
-      "learning_rate": 0.00019999194594583485,
-      "loss": 46.0,
-      "step": 25097
-    },
-    {
-      "epoch": 4.041869640484721,
-      "grad_norm": 0.001760690938681364,
-      "learning_rate": 0.00019999194530374083,
-      "loss": 46.0,
-      "step": 25098
-    },
-    {
-      "epoch": 4.042030677563509,
-      "grad_norm": 0.0021013529039919376,
-      "learning_rate": 0.0001999919446616212,
-      "loss": 46.0,
-      "step": 25099
-    },
-    {
-      "epoch": 4.042191714642296,
-      "grad_norm": 0.006535171065479517,
-      "learning_rate": 0.00019999194401947596,
-      "loss": 46.0,
-      "step": 25100
-    },
-    {
-      "epoch": 4.042352751721084,
-      "grad_norm": 0.0026875522453337908,
-      "learning_rate": 0.00019999194337730515,
-      "loss": 46.0,
-      "step": 25101
-    },
-    {
-      "epoch": 4.042513788799871,
-      "grad_norm": 0.002320497529581189,
-      "learning_rate": 0.00019999194273510876,
-      "loss": 46.0,
-      "step": 25102
-    },
-    {
-      "epoch": 4.042674825878659,
-      "grad_norm": 0.003637721762061119,
-      "learning_rate": 0.00019999194209288675,
-      "loss": 46.0,
-      "step": 25103
-    },
-    {
-      "epoch": 4.042835862957446,
-      "grad_norm": 0.0026988149620592594,
-      "learning_rate": 0.00019999194145063916,
-      "loss": 46.0,
-      "step": 25104
-    },
-    {
-      "epoch": 4.042996900036234,
-      "grad_norm": 0.006968206260353327,
-      "learning_rate": 0.00019999194080836598,
-      "loss": 46.0,
-      "step": 25105
-    },
-    {
-      "epoch": 4.043157937115021,
-      "grad_norm": 0.0019634298514574766,
-      "learning_rate": 0.00019999194016606718,
-      "loss": 46.0,
-      "step": 25106
-    },
-    {
-      "epoch": 4.0433189741938085,
-      "grad_norm": 0.004155104514211416,
-      "learning_rate": 0.0001999919395237428,
-      "loss": 46.0,
-      "step": 25107
-    },
-    {
-      "epoch": 4.043480011272596,
-      "grad_norm": 0.010871573351323605,
-      "learning_rate": 0.00019999193888139285,
-      "loss": 46.0,
-      "step": 25108
-    },
-    {
-      "epoch": 4.0436410483513825,
-      "grad_norm": 0.013126935809850693,
-      "learning_rate": 0.00019999193823901727,
-      "loss": 46.0,
-      "step": 25109
-    },
-    {
-      "epoch": 4.04380208543017,
-      "grad_norm": 0.0015862892614677548,
-      "learning_rate": 0.00019999193759661613,
-      "loss": 46.0,
-      "step": 25110
-    },
-    {
-      "epoch": 4.043963122508957,
-      "grad_norm": 0.03452477976679802,
-      "learning_rate": 0.0001999919369541894,
-      "loss": 46.0,
-      "step": 25111
-    },
-    {
-      "epoch": 4.044124159587745,
-      "grad_norm": 0.013962416909635067,
-      "learning_rate": 0.00019999193631173705,
-      "loss": 46.0,
-      "step": 25112
-    },
-    {
-      "epoch": 4.044285196666532,
-      "grad_norm": 0.0023345299996435642,
-      "learning_rate": 0.00019999193566925914,
-      "loss": 46.0,
-      "step": 25113
-    },
-    {
-      "epoch": 4.04444623374532,
-      "grad_norm": 0.005403893534094095,
-      "learning_rate": 0.00019999193502675562,
-      "loss": 46.0,
-      "step": 25114
-    },
-    {
-      "epoch": 4.044607270824107,
-      "grad_norm": 0.009967776946723461,
-      "learning_rate": 0.00019999193438422646,
-      "loss": 46.0,
-      "step": 25115
-    },
-    {
-      "epoch": 4.044768307902895,
-      "grad_norm": 0.005063614808022976,
-      "learning_rate": 0.00019999193374167177,
-      "loss": 46.0,
-      "step": 25116
-    },
-    {
-      "epoch": 4.044929344981682,
-      "grad_norm": 0.008629816584289074,
-      "learning_rate": 0.0001999919330990915,
-      "loss": 46.0,
-      "step": 25117
-    },
-    {
-      "epoch": 4.04509038206047,
-      "grad_norm": 0.0034128918778151274,
-      "learning_rate": 0.00019999193245648556,
-      "loss": 46.0,
-      "step": 25118
-    },
-    {
-      "epoch": 4.045251419139257,
-      "grad_norm": 0.0012624525697901845,
-      "learning_rate": 0.0001999919318138541,
-      "loss": 46.0,
-      "step": 25119
-    },
-    {
-      "epoch": 4.0454124562180445,
-      "grad_norm": 0.015965472906827927,
-      "learning_rate": 0.000199991931171197,
-      "loss": 46.0,
-      "step": 25120
-    },
-    {
-      "epoch": 4.045573493296832,
-      "grad_norm": 0.006691340357065201,
-      "learning_rate": 0.00019999193052851435,
-      "loss": 46.0,
-      "step": 25121
-    },
-    {
-      "epoch": 4.045734530375619,
-      "grad_norm": 0.0012196802999824286,
-      "learning_rate": 0.00019999192988580608,
-      "loss": 46.0,
-      "step": 25122
-    },
-    {
-      "epoch": 4.045895567454407,
-      "grad_norm": 0.0007714468520134687,
-      "learning_rate": 0.00019999192924307222,
-      "loss": 46.0,
-      "step": 25123
-    },
-    {
-      "epoch": 4.046056604533193,
-      "grad_norm": 0.008993716910481453,
-      "learning_rate": 0.00019999192860031278,
-      "loss": 46.0,
-      "step": 25124
-    },
-    {
-      "epoch": 4.046217641611981,
-      "grad_norm": 0.004828222095966339,
-      "learning_rate": 0.00019999192795752772,
-      "loss": 46.0,
-      "step": 25125
-    },
-    {
-      "epoch": 4.046378678690768,
-      "grad_norm": 0.0037623513489961624,
-      "learning_rate": 0.0001999919273147171,
-      "loss": 46.0,
-      "step": 25126
-    },
-    {
-      "epoch": 4.046539715769556,
-      "grad_norm": 0.004046284593641758,
-      "learning_rate": 0.00019999192667188086,
-      "loss": 46.0,
-      "step": 25127
-    },
-    {
-      "epoch": 4.046700752848343,
-      "grad_norm": 0.005544481333345175,
-      "learning_rate": 0.00019999192602901904,
-      "loss": 46.0,
-      "step": 25128
-    },
-    {
-      "epoch": 4.046861789927131,
-      "grad_norm": 0.008518224582076073,
-      "learning_rate": 0.00019999192538613163,
-      "loss": 46.0,
-      "step": 25129
-    },
-    {
-      "epoch": 4.047022827005918,
-      "grad_norm": 0.009400081820786,
-      "learning_rate": 0.0001999919247432186,
-      "loss": 46.0,
-      "step": 25130
-    },
-    {
-      "epoch": 4.0471838640847055,
-      "grad_norm": 0.00755870109423995,
-      "learning_rate": 0.00019999192410028002,
-      "loss": 46.0,
-      "step": 25131
-    },
-    {
-      "epoch": 4.047344901163493,
-      "grad_norm": 0.004427696578204632,
-      "learning_rate": 0.0001999919234573158,
-      "loss": 46.0,
-      "step": 25132
-    },
-    {
-      "epoch": 4.04750593824228,
-      "grad_norm": 0.003742228029295802,
-      "learning_rate": 0.00019999192281432604,
-      "loss": 46.0,
-      "step": 25133
-    },
-    {
-      "epoch": 4.047666975321068,
-      "grad_norm": 0.005632624961435795,
-      "learning_rate": 0.00019999192217131067,
-      "loss": 46.0,
-      "step": 25134
-    },
-    {
-      "epoch": 4.047828012399855,
-      "grad_norm": 0.0042465548031032085,
-      "learning_rate": 0.00019999192152826968,
-      "loss": 46.0,
-      "step": 25135
-    },
-    {
-      "epoch": 4.047989049478643,
-      "grad_norm": 0.0019142965320497751,
-      "learning_rate": 0.0001999919208852031,
-      "loss": 46.0,
-      "step": 25136
-    },
-    {
-      "epoch": 4.04815008655743,
-      "grad_norm": 0.006928254850208759,
-      "learning_rate": 0.00019999192024211095,
-      "loss": 46.0,
-      "step": 25137
-    },
-    {
-      "epoch": 4.048311123636217,
-      "grad_norm": 0.001466908841393888,
-      "learning_rate": 0.0001999919195989932,
-      "loss": 46.0,
-      "step": 25138
-    },
-    {
-      "epoch": 4.048472160715004,
-      "grad_norm": 0.001917113084346056,
-      "learning_rate": 0.00019999191895584987,
-      "loss": 46.0,
-      "step": 25139
-    },
-    {
-      "epoch": 4.048633197793792,
-      "grad_norm": 0.0011072380002588034,
-      "learning_rate": 0.00019999191831268095,
-      "loss": 46.0,
-      "step": 25140
-    },
-    {
-      "epoch": 4.048794234872579,
-      "grad_norm": 0.0024496724363416433,
-      "learning_rate": 0.00019999191766948638,
-      "loss": 46.0,
-      "step": 25141
-    },
-    {
-      "epoch": 4.048955271951367,
-      "grad_norm": 0.0028447695076465607,
-      "learning_rate": 0.0001999919170262663,
-      "loss": 46.0,
-      "step": 25142
-    },
-    {
-      "epoch": 4.049116309030154,
-      "grad_norm": 0.002107927342876792,
-      "learning_rate": 0.00019999191638302058,
-      "loss": 46.0,
-      "step": 25143
-    },
-    {
-      "epoch": 4.0492773461089415,
-      "grad_norm": 0.006156007759273052,
-      "learning_rate": 0.00019999191573974928,
-      "loss": 46.0,
-      "step": 25144
-    },
-    {
-      "epoch": 4.049438383187729,
-      "grad_norm": 0.0030636582523584366,
-      "learning_rate": 0.00019999191509645237,
-      "loss": 46.0,
-      "step": 25145
-    },
-    {
-      "epoch": 4.049599420266516,
-      "grad_norm": 0.0011971063213422894,
-      "learning_rate": 0.00019999191445312987,
-      "loss": 46.0,
-      "step": 25146
-    },
-    {
-      "epoch": 4.049760457345304,
-      "grad_norm": 0.0026841172948479652,
-      "learning_rate": 0.0001999919138097818,
-      "loss": 46.0,
-      "step": 25147
-    },
-    {
-      "epoch": 4.049921494424091,
-      "grad_norm": 0.00591725530102849,
-      "learning_rate": 0.00019999191316640813,
-      "loss": 46.0,
-      "step": 25148
-    },
-    {
-      "epoch": 4.050082531502879,
-      "grad_norm": 0.0026979632675647736,
-      "learning_rate": 0.00019999191252300885,
-      "loss": 46.0,
-      "step": 25149
-    },
-    {
-      "epoch": 4.050243568581666,
-      "grad_norm": 0.001508760149590671,
-      "learning_rate": 0.00019999191187958397,
-      "loss": 46.0,
-      "step": 25150
-    },
-    {
-      "epoch": 4.050404605660454,
-      "grad_norm": 0.0017535273218527436,
-      "learning_rate": 0.0001999919112361335,
-      "loss": 46.0,
-      "step": 25151
-    },
-    {
-      "epoch": 4.050565642739241,
-      "grad_norm": 0.009976713918149471,
-      "learning_rate": 0.00019999191059265746,
-      "loss": 46.0,
-      "step": 25152
-    },
-    {
-      "epoch": 4.050726679818028,
-      "grad_norm": 0.007307478692382574,
-      "learning_rate": 0.00019999190994915585,
-      "loss": 46.0,
-      "step": 25153
-    },
-    {
-      "epoch": 4.050887716896815,
-      "grad_norm": 0.002895388752222061,
-      "learning_rate": 0.0001999919093056286,
-      "loss": 46.0,
-      "step": 25154
-    },
-    {
-      "epoch": 4.051048753975603,
-      "grad_norm": 0.003612358821555972,
-      "learning_rate": 0.00019999190866207576,
-      "loss": 46.0,
-      "step": 25155
-    },
-    {
-      "epoch": 4.05120979105439,
-      "grad_norm": 0.006851687096059322,
-      "learning_rate": 0.00019999190801849736,
-      "loss": 46.0,
-      "step": 25156
-    },
-    {
-      "epoch": 4.0513708281331775,
-      "grad_norm": 0.0065048676915466785,
-      "learning_rate": 0.00019999190737489335,
-      "loss": 46.0,
-      "step": 25157
-    },
-    {
-      "epoch": 4.051531865211965,
-      "grad_norm": 0.002576471772044897,
-      "learning_rate": 0.00019999190673126372,
-      "loss": 46.0,
-      "step": 25158
-    },
-    {
-      "epoch": 4.051692902290752,
-      "grad_norm": 0.0021585344802588224,
-      "learning_rate": 0.00019999190608760854,
-      "loss": 46.0,
-      "step": 25159
-    },
-    {
-      "epoch": 4.05185393936954,
-      "grad_norm": 0.0053537990897893906,
-      "learning_rate": 0.00019999190544392774,
-      "loss": 46.0,
-      "step": 25160
-    },
-    {
-      "epoch": 4.052014976448327,
-      "grad_norm": 0.008653052151203156,
-      "learning_rate": 0.00019999190480022138,
-      "loss": 46.0,
-      "step": 25161
-    },
-    {
-      "epoch": 4.052176013527115,
-      "grad_norm": 0.0014121340354904532,
-      "learning_rate": 0.00019999190415648937,
-      "loss": 46.0,
-      "step": 25162
-    },
-    {
-      "epoch": 4.052337050605902,
-      "grad_norm": 0.0024111121892929077,
-      "learning_rate": 0.00019999190351273184,
-      "loss": 46.0,
-      "step": 25163
-    },
-    {
-      "epoch": 4.05249808768469,
-      "grad_norm": 0.013518192805349827,
-      "learning_rate": 0.00019999190286894863,
-      "loss": 46.0,
-      "step": 25164
-    },
-    {
-      "epoch": 4.052659124763477,
-      "grad_norm": 0.004137784242630005,
-      "learning_rate": 0.00019999190222513987,
-      "loss": 46.0,
-      "step": 25165
-    },
-    {
-      "epoch": 4.052820161842265,
-      "grad_norm": 0.0022434331476688385,
-      "learning_rate": 0.00019999190158130555,
-      "loss": 46.0,
-      "step": 25166
-    },
-    {
-      "epoch": 4.052981198921051,
-      "grad_norm": 0.007980064488947392,
-      "learning_rate": 0.0001999919009374456,
-      "loss": 46.0,
-      "step": 25167
-    },
-    {
-      "epoch": 4.053142235999839,
-      "grad_norm": 0.002234174171462655,
-      "learning_rate": 0.00019999190029356006,
-      "loss": 46.0,
-      "step": 25168
-    },
-    {
-      "epoch": 4.053303273078626,
-      "grad_norm": 0.00829023215919733,
-      "learning_rate": 0.00019999189964964894,
-      "loss": 46.0,
-      "step": 25169
-    },
-    {
-      "epoch": 4.0534643101574135,
-      "grad_norm": 0.002040259772911668,
-      "learning_rate": 0.00019999189900571222,
-      "loss": 46.0,
-      "step": 25170
-    },
-    {
-      "epoch": 4.053625347236201,
-      "grad_norm": 0.0046273223124444485,
-      "learning_rate": 0.00019999189836174993,
-      "loss": 46.0,
-      "step": 25171
-    },
-    {
-      "epoch": 4.053786384314988,
-      "grad_norm": 0.006136543117463589,
-      "learning_rate": 0.000199991897717762,
-      "loss": 46.0,
-      "step": 25172
-    },
-    {
-      "epoch": 4.053947421393776,
-      "grad_norm": 0.0066335368901491165,
-      "learning_rate": 0.0001999918970737485,
-      "loss": 46.0,
-      "step": 25173
-    },
-    {
-      "epoch": 4.054108458472563,
-      "grad_norm": 0.0004941087099723518,
-      "learning_rate": 0.0001999918964297094,
-      "loss": 46.0,
-      "step": 25174
-    },
-    {
-      "epoch": 4.054269495551351,
-      "grad_norm": 0.002243608236312866,
-      "learning_rate": 0.00019999189578564472,
-      "loss": 46.0,
-      "step": 25175
-    },
-    {
-      "epoch": 4.054430532630138,
-      "grad_norm": 0.007276506163179874,
-      "learning_rate": 0.00019999189514155444,
-      "loss": 46.0,
-      "step": 25176
-    },
-    {
-      "epoch": 4.054591569708926,
-      "grad_norm": 0.00997980684041977,
-      "learning_rate": 0.0001999918944974386,
-      "loss": 46.0,
-      "step": 25177
-    },
-    {
-      "epoch": 4.054752606787713,
-      "grad_norm": 0.0020957200322300196,
-      "learning_rate": 0.00019999189385329713,
-      "loss": 46.0,
-      "step": 25178
-    },
-    {
-      "epoch": 4.0549136438665005,
-      "grad_norm": 0.003410806180909276,
-      "learning_rate": 0.0001999918932091301,
-      "loss": 46.0,
-      "step": 25179
-    },
-    {
-      "epoch": 4.055074680945288,
-      "grad_norm": 0.004063980188220739,
-      "learning_rate": 0.0001999918925649374,
-      "loss": 46.0,
-      "step": 25180
-    },
-    {
-      "epoch": 4.0552357180240755,
-      "grad_norm": 0.0035752933472394943,
-      "learning_rate": 0.0001999918919207192,
-      "loss": 46.0,
-      "step": 25181
-    },
-    {
-      "epoch": 4.055396755102862,
-      "grad_norm": 0.006180457770824432,
-      "learning_rate": 0.00019999189127647534,
-      "loss": 46.0,
-      "step": 25182
-    },
-    {
-      "epoch": 4.0555577921816495,
-      "grad_norm": 0.002256970154121518,
-      "learning_rate": 0.00019999189063220593,
-      "loss": 46.0,
-      "step": 25183
-    },
-    {
-      "epoch": 4.055718829260437,
-      "grad_norm": 0.0028089622501283884,
-      "learning_rate": 0.00019999188998791093,
-      "loss": 46.0,
-      "step": 25184
-    },
-    {
-      "epoch": 4.055879866339224,
-      "grad_norm": 0.004492702428251505,
-      "learning_rate": 0.0001999918893435903,
-      "loss": 46.0,
-      "step": 25185
-    },
-    {
-      "epoch": 4.056040903418012,
-      "grad_norm": 0.009128935635089874,
-      "learning_rate": 0.0001999918886992441,
-      "loss": 46.0,
-      "step": 25186
-    },
-    {
-      "epoch": 4.056201940496799,
-      "grad_norm": 0.006036663427948952,
-      "learning_rate": 0.0001999918880548723,
-      "loss": 46.0,
-      "step": 25187
-    },
-    {
-      "epoch": 4.056362977575587,
-      "grad_norm": 0.0018259697826579213,
-      "learning_rate": 0.00019999188741047491,
-      "loss": 46.0,
-      "step": 25188
-    },
-    {
-      "epoch": 4.056524014654374,
-      "grad_norm": 0.0036418833769857883,
-      "learning_rate": 0.00019999188676605192,
-      "loss": 46.0,
-      "step": 25189
-    },
-    {
-      "epoch": 4.056685051733162,
-      "grad_norm": 0.004148347303271294,
-      "learning_rate": 0.00019999188612160334,
-      "loss": 46.0,
-      "step": 25190
-    },
-    {
-      "epoch": 4.056846088811949,
-      "grad_norm": 0.00329596851952374,
-      "learning_rate": 0.00019999188547712918,
-      "loss": 46.0,
-      "step": 25191
-    },
-    {
-      "epoch": 4.0570071258907365,
-      "grad_norm": 0.0062295617535710335,
-      "learning_rate": 0.00019999188483262942,
-      "loss": 46.0,
-      "step": 25192
-    },
-    {
-      "epoch": 4.057168162969524,
-      "grad_norm": 0.007119002752006054,
-      "learning_rate": 0.00019999188418810406,
-      "loss": 46.0,
-      "step": 25193
-    },
-    {
-      "epoch": 4.057329200048311,
-      "grad_norm": 0.001718393643386662,
-      "learning_rate": 0.0001999918835435531,
-      "loss": 46.0,
-      "step": 25194
-    },
-    {
-      "epoch": 4.057490237127099,
-      "grad_norm": 0.02157345972955227,
-      "learning_rate": 0.0001999918828989766,
-      "loss": 46.0,
-      "step": 25195
-    },
-    {
-      "epoch": 4.057651274205886,
-      "grad_norm": 0.0031764970626682043,
-      "learning_rate": 0.00019999188225437443,
-      "loss": 46.0,
-      "step": 25196
-    },
-    {
-      "epoch": 4.057812311284673,
-      "grad_norm": 0.0010836933506652713,
-      "learning_rate": 0.00019999188160974672,
-      "loss": 46.0,
-      "step": 25197
-    },
-    {
-      "epoch": 4.05797334836346,
-      "grad_norm": 0.009517136961221695,
-      "learning_rate": 0.0001999918809650934,
-      "loss": 46.0,
-      "step": 25198
-    },
-    {
-      "epoch": 4.058134385442248,
-      "grad_norm": 0.0042744833044707775,
-      "learning_rate": 0.0001999918803204145,
-      "loss": 46.0,
-      "step": 25199
-    },
-    {
-      "epoch": 4.058295422521035,
-      "grad_norm": 0.0010750008514150977,
-      "learning_rate": 0.00019999187967571,
-      "loss": 46.0,
-      "step": 25200
-    },
-    {
-      "epoch": 4.058456459599823,
-      "grad_norm": 0.0023556307423859835,
-      "learning_rate": 0.0001999918790309799,
-      "loss": 46.0,
-      "step": 25201
-    },
-    {
-      "epoch": 4.05861749667861,
-      "grad_norm": 0.0028152463492006063,
-      "learning_rate": 0.0001999918783862242,
-      "loss": 46.0,
-      "step": 25202
-    },
-    {
-      "epoch": 4.058778533757398,
-      "grad_norm": 0.001785579021088779,
-      "learning_rate": 0.00019999187774144293,
-      "loss": 46.0,
-      "step": 25203
-    },
-    {
-      "epoch": 4.058939570836185,
-      "grad_norm": 0.006174648180603981,
-      "learning_rate": 0.00019999187709663605,
-      "loss": 46.0,
-      "step": 25204
-    },
-    {
-      "epoch": 4.0591006079149725,
-      "grad_norm": 0.008724072016775608,
-      "learning_rate": 0.00019999187645180358,
-      "loss": 46.0,
-      "step": 25205
-    },
-    {
-      "epoch": 4.05926164499376,
-      "grad_norm": 0.00929219275712967,
-      "learning_rate": 0.00019999187580694555,
-      "loss": 46.0,
-      "step": 25206
-    },
-    {
-      "epoch": 4.059422682072547,
-      "grad_norm": 0.0054887025617063046,
-      "learning_rate": 0.00019999187516206188,
-      "loss": 46.0,
-      "step": 25207
-    },
-    {
-      "epoch": 4.059583719151335,
-      "grad_norm": 0.003904205746948719,
-      "learning_rate": 0.00019999187451715263,
-      "loss": 46.0,
-      "step": 25208
-    },
-    {
-      "epoch": 4.059744756230122,
-      "grad_norm": 0.0022380314767360687,
-      "learning_rate": 0.00019999187387221778,
-      "loss": 46.0,
-      "step": 25209
-    },
-    {
-      "epoch": 4.05990579330891,
-      "grad_norm": 0.0006175815942697227,
-      "learning_rate": 0.00019999187322725735,
-      "loss": 46.0,
-      "step": 25210
-    },
-    {
-      "epoch": 4.060066830387696,
-      "grad_norm": 0.005174033343791962,
-      "learning_rate": 0.00019999187258227133,
-      "loss": 46.0,
-      "step": 25211
-    },
-    {
-      "epoch": 4.060227867466484,
-      "grad_norm": 0.002573292935267091,
-      "learning_rate": 0.00019999187193725973,
-      "loss": 46.0,
-      "step": 25212
-    },
-    {
-      "epoch": 4.060388904545271,
-      "grad_norm": 0.003381407354027033,
-      "learning_rate": 0.0001999918712922225,
-      "loss": 46.0,
-      "step": 25213
-    },
-    {
-      "epoch": 4.060549941624059,
-      "grad_norm": 0.0033199312165379524,
-      "learning_rate": 0.0001999918706471597,
-      "loss": 46.0,
-      "step": 25214
-    },
-    {
-      "epoch": 4.060710978702846,
-      "grad_norm": 0.005421431735157967,
-      "learning_rate": 0.0001999918700020713,
-      "loss": 46.0,
-      "step": 25215
-    },
-    {
-      "epoch": 4.060872015781634,
-      "grad_norm": 0.005486281588673592,
-      "learning_rate": 0.0001999918693569573,
-      "loss": 46.0,
-      "step": 25216
-    },
-    {
-      "epoch": 4.061033052860421,
-      "grad_norm": 0.007144741248339415,
-      "learning_rate": 0.00019999186871181773,
-      "loss": 46.0,
-      "step": 25217
-    },
-    {
-      "epoch": 4.0611940899392085,
-      "grad_norm": 0.018770670518279076,
-      "learning_rate": 0.00019999186806665255,
-      "loss": 46.0,
-      "step": 25218
-    },
-    {
-      "epoch": 4.061355127017996,
-      "grad_norm": 0.0036431788466870785,
-      "learning_rate": 0.0001999918674214618,
-      "loss": 46.0,
-      "step": 25219
-    },
-    {
-      "epoch": 4.061516164096783,
-      "grad_norm": 0.0012442056322470307,
-      "learning_rate": 0.00019999186677624543,
-      "loss": 46.0,
-      "step": 25220
-    },
-    {
-      "epoch": 4.061677201175571,
-      "grad_norm": 0.0025082076899707317,
-      "learning_rate": 0.00019999186613100348,
-      "loss": 46.0,
-      "step": 25221
-    },
-    {
-      "epoch": 4.061838238254358,
-      "grad_norm": 0.006033538840711117,
-      "learning_rate": 0.00019999186548573595,
-      "loss": 46.0,
-      "step": 25222
-    },
-    {
-      "epoch": 4.061999275333146,
-      "grad_norm": 0.007237591315060854,
-      "learning_rate": 0.0001999918648404428,
-      "loss": 46.0,
-      "step": 25223
-    },
-    {
-      "epoch": 4.062160312411933,
-      "grad_norm": 0.011093477718532085,
-      "learning_rate": 0.00019999186419512408,
-      "loss": 46.0,
-      "step": 25224
-    },
-    {
-      "epoch": 4.062321349490721,
-      "grad_norm": 0.001368444412946701,
-      "learning_rate": 0.00019999186354977976,
-      "loss": 46.0,
-      "step": 25225
-    },
-    {
-      "epoch": 4.062482386569507,
-      "grad_norm": 0.00213375105522573,
-      "learning_rate": 0.00019999186290440982,
-      "loss": 46.0,
-      "step": 25226
-    },
-    {
-      "epoch": 4.062643423648295,
-      "grad_norm": 0.0029317731969058514,
-      "learning_rate": 0.00019999186225901433,
-      "loss": 46.0,
-      "step": 25227
-    },
-    {
-      "epoch": 4.062804460727082,
-      "grad_norm": 0.0009794340003281832,
-      "learning_rate": 0.00019999186161359322,
-      "loss": 46.0,
-      "step": 25228
-    },
-    {
-      "epoch": 4.06296549780587,
-      "grad_norm": 0.00515911728143692,
-      "learning_rate": 0.00019999186096814655,
-      "loss": 46.0,
-      "step": 25229
-    },
-    {
-      "epoch": 4.063126534884657,
-      "grad_norm": 0.010796592570841312,
-      "learning_rate": 0.00019999186032267424,
-      "loss": 46.0,
-      "step": 25230
-    },
-    {
-      "epoch": 4.0632875719634445,
-      "grad_norm": 0.0009916980052366853,
-      "learning_rate": 0.00019999185967717637,
-      "loss": 46.0,
-      "step": 25231
-    },
-    {
-      "epoch": 4.063448609042232,
-      "grad_norm": 0.002051148796454072,
-      "learning_rate": 0.0001999918590316529,
-      "loss": 46.0,
-      "step": 25232
-    },
-    {
-      "epoch": 4.063609646121019,
-      "grad_norm": 0.0016804755432531238,
-      "learning_rate": 0.00019999185838610385,
-      "loss": 46.0,
-      "step": 25233
-    },
-    {
-      "epoch": 4.063770683199807,
-      "grad_norm": 0.000529679236933589,
-      "learning_rate": 0.0001999918577405292,
-      "loss": 46.0,
-      "step": 25234
-    },
-    {
-      "epoch": 4.063931720278594,
-      "grad_norm": 0.0062715704552829266,
-      "learning_rate": 0.00019999185709492894,
-      "loss": 46.0,
-      "step": 25235
-    },
-    {
-      "epoch": 4.064092757357382,
-      "grad_norm": 0.0011910318862646818,
-      "learning_rate": 0.00019999185644930308,
-      "loss": 46.0,
-      "step": 25236
-    },
-    {
-      "epoch": 4.064253794436169,
-      "grad_norm": 0.009606339037418365,
-      "learning_rate": 0.00019999185580365164,
-      "loss": 46.0,
-      "step": 25237
-    },
-    {
-      "epoch": 4.064414831514957,
-      "grad_norm": 0.004720029886811972,
-      "learning_rate": 0.0001999918551579746,
-      "loss": 46.0,
-      "step": 25238
-    },
-    {
-      "epoch": 4.064575868593744,
-      "grad_norm": 0.008531820960342884,
-      "learning_rate": 0.000199991854512272,
-      "loss": 46.0,
-      "step": 25239
-    },
-    {
-      "epoch": 4.064736905672531,
-      "grad_norm": 0.001511663431301713,
-      "learning_rate": 0.00019999185386654377,
-      "loss": 46.0,
-      "step": 25240
-    },
-    {
-      "epoch": 4.064897942751318,
-      "grad_norm": 0.014855904504656792,
-      "learning_rate": 0.00019999185322078998,
-      "loss": 46.0,
-      "step": 25241
-    },
-    {
-      "epoch": 4.065058979830106,
-      "grad_norm": 0.0005880754324607551,
-      "learning_rate": 0.00019999185257501057,
-      "loss": 46.0,
-      "step": 25242
-    },
-    {
-      "epoch": 4.065220016908893,
-      "grad_norm": 0.0004652971983887255,
-      "learning_rate": 0.00019999185192920557,
-      "loss": 46.0,
-      "step": 25243
-    },
-    {
-      "epoch": 4.0653810539876805,
-      "grad_norm": 0.0011934635695070028,
-      "learning_rate": 0.00019999185128337499,
-      "loss": 46.0,
-      "step": 25244
-    },
-    {
-      "epoch": 4.065542091066468,
-      "grad_norm": 0.0019407463259994984,
-      "learning_rate": 0.00019999185063751881,
-      "loss": 46.0,
-      "step": 25245
-    },
-    {
-      "epoch": 4.065703128145255,
-      "grad_norm": 0.0012099232990294695,
-      "learning_rate": 0.00019999184999163703,
-      "loss": 46.0,
-      "step": 25246
-    },
-    {
-      "epoch": 4.065864165224043,
-      "grad_norm": 0.023715445771813393,
-      "learning_rate": 0.00019999184934572968,
-      "loss": 46.0,
-      "step": 25247
-    },
-    {
-      "epoch": 4.06602520230283,
-      "grad_norm": 0.014227522537112236,
-      "learning_rate": 0.0001999918486997967,
-      "loss": 46.0,
-      "step": 25248
-    },
-    {
-      "epoch": 4.066186239381618,
-      "grad_norm": 0.0035511336755007505,
-      "learning_rate": 0.00019999184805383818,
-      "loss": 46.0,
-      "step": 25249
-    },
-    {
-      "epoch": 4.066347276460405,
-      "grad_norm": 0.005493625998497009,
-      "learning_rate": 0.00019999184740785401,
-      "loss": 46.0,
-      "step": 25250
-    },
-    {
-      "epoch": 4.066508313539193,
-      "grad_norm": 0.005144867580384016,
-      "learning_rate": 0.0001999918467618443,
-      "loss": 46.0,
-      "step": 25251
-    },
-    {
-      "epoch": 4.06666935061798,
-      "grad_norm": 0.0035444507375359535,
-      "learning_rate": 0.00019999184611580896,
-      "loss": 46.0,
-      "step": 25252
-    },
-    {
-      "epoch": 4.0668303876967675,
-      "grad_norm": 0.0022409844677895308,
-      "learning_rate": 0.00019999184546974803,
-      "loss": 46.0,
-      "step": 25253
-    },
-    {
-      "epoch": 4.066991424775555,
-      "grad_norm": 0.0017127110622823238,
-      "learning_rate": 0.00019999184482366152,
-      "loss": 46.0,
-      "step": 25254
-    },
-    {
-      "epoch": 4.0671524618543415,
-      "grad_norm": 0.004753213841468096,
-      "learning_rate": 0.00019999184417754943,
-      "loss": 46.0,
-      "step": 25255
-    },
-    {
-      "epoch": 4.067313498933129,
-      "grad_norm": 0.0015547281363978982,
-      "learning_rate": 0.00019999184353141172,
-      "loss": 46.0,
-      "step": 25256
-    },
-    {
-      "epoch": 4.067474536011916,
-      "grad_norm": 0.0014243695186451077,
-      "learning_rate": 0.00019999184288524844,
-      "loss": 46.0,
-      "step": 25257
-    },
-    {
-      "epoch": 4.067635573090704,
-      "grad_norm": 0.0010112738236784935,
-      "learning_rate": 0.00019999184223905956,
-      "loss": 46.0,
-      "step": 25258
-    },
-    {
-      "epoch": 4.067796610169491,
-      "grad_norm": 0.013485829345881939,
-      "learning_rate": 0.00019999184159284506,
-      "loss": 46.0,
-      "step": 25259
-    },
-    {
-      "epoch": 4.067957647248279,
-      "grad_norm": 0.0068550691939890385,
-      "learning_rate": 0.000199991840946605,
-      "loss": 46.0,
-      "step": 25260
-    },
-    {
-      "epoch": 4.068118684327066,
-      "grad_norm": 0.000990360975265503,
-      "learning_rate": 0.00019999184030033932,
-      "loss": 46.0,
-      "step": 25261
-    },
-    {
-      "epoch": 4.068279721405854,
-      "grad_norm": 0.002246738411486149,
-      "learning_rate": 0.00019999183965404806,
-      "loss": 46.0,
-      "step": 25262
-    },
-    {
-      "epoch": 4.068440758484641,
-      "grad_norm": 0.005762267857789993,
-      "learning_rate": 0.00019999183900773124,
-      "loss": 46.0,
-      "step": 25263
-    },
-    {
-      "epoch": 4.068601795563429,
-      "grad_norm": 0.0022816245909780264,
-      "learning_rate": 0.0001999918383613888,
-      "loss": 46.0,
-      "step": 25264
-    },
-    {
-      "epoch": 4.068762832642216,
-      "grad_norm": 0.0005689717945642769,
-      "learning_rate": 0.00019999183771502075,
-      "loss": 46.0,
-      "step": 25265
-    },
-    {
-      "epoch": 4.0689238697210035,
-      "grad_norm": 0.0046620857901871204,
-      "learning_rate": 0.00019999183706862712,
-      "loss": 46.0,
-      "step": 25266
-    },
-    {
-      "epoch": 4.069084906799791,
-      "grad_norm": 0.002422595862299204,
-      "learning_rate": 0.0001999918364222079,
-      "loss": 46.0,
-      "step": 25267
-    },
-    {
-      "epoch": 4.069245943878578,
-      "grad_norm": 0.007057536393404007,
-      "learning_rate": 0.00019999183577576308,
-      "loss": 46.0,
-      "step": 25268
-    },
-    {
-      "epoch": 4.069406980957366,
-      "grad_norm": 0.003205882851034403,
-      "learning_rate": 0.00019999183512929268,
-      "loss": 46.0,
-      "step": 25269
-    },
-    {
-      "epoch": 4.069568018036152,
-      "grad_norm": 0.008235041052103043,
-      "learning_rate": 0.00019999183448279667,
-      "loss": 46.0,
-      "step": 25270
-    },
-    {
-      "epoch": 4.06972905511494,
-      "grad_norm": 0.0087369279935956,
-      "learning_rate": 0.0001999918338362751,
-      "loss": 46.0,
-      "step": 25271
-    },
-    {
-      "epoch": 4.069890092193727,
-      "grad_norm": 0.003213796764612198,
-      "learning_rate": 0.0001999918331897279,
-      "loss": 46.0,
-      "step": 25272
-    },
-    {
-      "epoch": 4.070051129272515,
-      "grad_norm": 0.0017283365596085787,
-      "learning_rate": 0.00019999183254315513,
-      "loss": 46.0,
-      "step": 25273
-    },
-    {
-      "epoch": 4.070212166351302,
-      "grad_norm": 0.004078635945916176,
-      "learning_rate": 0.00019999183189655677,
-      "loss": 46.0,
-      "step": 25274
-    },
-    {
-      "epoch": 4.07037320343009,
-      "grad_norm": 0.0010716464603319764,
-      "learning_rate": 0.0001999918312499328,
-      "loss": 46.0,
-      "step": 25275
-    },
-    {
-      "epoch": 4.070534240508877,
-      "grad_norm": 0.0020703410264104605,
-      "learning_rate": 0.00019999183060328326,
-      "loss": 46.0,
-      "step": 25276
-    },
-    {
-      "epoch": 4.070695277587665,
-      "grad_norm": 0.003401249647140503,
-      "learning_rate": 0.0001999918299566081,
-      "loss": 46.0,
-      "step": 25277
-    },
-    {
-      "epoch": 4.070856314666452,
-      "grad_norm": 0.002651548944413662,
-      "learning_rate": 0.00019999182930990737,
-      "loss": 46.0,
-      "step": 25278
-    },
-    {
-      "epoch": 4.0710173517452395,
-      "grad_norm": 0.0022555135656148195,
-      "learning_rate": 0.00019999182866318105,
-      "loss": 46.0,
-      "step": 25279
-    },
-    {
-      "epoch": 4.071178388824027,
-      "grad_norm": 0.0022942586801946163,
-      "learning_rate": 0.0001999918280164291,
-      "loss": 46.0,
-      "step": 25280
-    },
-    {
-      "epoch": 4.071339425902814,
-      "grad_norm": 0.006510575767606497,
-      "learning_rate": 0.00019999182736965158,
-      "loss": 46.0,
-      "step": 25281
-    },
-    {
-      "epoch": 4.071500462981602,
-      "grad_norm": 0.00912565365433693,
-      "learning_rate": 0.0001999918267228485,
-      "loss": 46.0,
-      "step": 25282
-    },
-    {
-      "epoch": 4.071661500060389,
-      "grad_norm": 0.009387057274580002,
-      "learning_rate": 0.00019999182607601977,
-      "loss": 46.0,
-      "step": 25283
-    },
-    {
-      "epoch": 4.071822537139177,
-      "grad_norm": 0.008505571633577347,
-      "learning_rate": 0.00019999182542916545,
-      "loss": 46.0,
-      "step": 25284
-    },
-    {
-      "epoch": 4.071983574217963,
-      "grad_norm": 0.011253810487687588,
-      "learning_rate": 0.00019999182478228558,
-      "loss": 46.0,
-      "step": 25285
-    },
-    {
-      "epoch": 4.072144611296751,
-      "grad_norm": 0.006155780982226133,
-      "learning_rate": 0.00019999182413538011,
-      "loss": 46.0,
-      "step": 25286
-    },
-    {
-      "epoch": 4.072305648375538,
-      "grad_norm": 0.003153882920742035,
-      "learning_rate": 0.00019999182348844904,
-      "loss": 46.0,
-      "step": 25287
-    },
-    {
-      "epoch": 4.072466685454326,
-      "grad_norm": 0.0036026728339493275,
-      "learning_rate": 0.00019999182284149235,
-      "loss": 46.0,
-      "step": 25288
-    },
-    {
-      "epoch": 4.072627722533113,
-      "grad_norm": 0.0011451862519606948,
-      "learning_rate": 0.0001999918221945101,
-      "loss": 46.0,
-      "step": 25289
-    },
-    {
-      "epoch": 4.072788759611901,
-      "grad_norm": 0.010938707739114761,
-      "learning_rate": 0.00019999182154750223,
-      "loss": 46.0,
-      "step": 25290
-    },
-    {
-      "epoch": 4.072949796690688,
-      "grad_norm": 0.004895662888884544,
-      "learning_rate": 0.0001999918209004688,
-      "loss": 46.0,
-      "step": 25291
-    },
-    {
-      "epoch": 4.0731108337694755,
-      "grad_norm": 0.01065182127058506,
-      "learning_rate": 0.00019999182025340977,
-      "loss": 46.0,
-      "step": 25292
-    },
-    {
-      "epoch": 4.073271870848263,
-      "grad_norm": 0.0026786313392221928,
-      "learning_rate": 0.0001999918196063251,
-      "loss": 46.0,
-      "step": 25293
-    },
-    {
-      "epoch": 4.07343290792705,
-      "grad_norm": 0.00778460456058383,
-      "learning_rate": 0.0001999918189592149,
-      "loss": 46.0,
-      "step": 25294
-    },
-    {
-      "epoch": 4.073593945005838,
-      "grad_norm": 0.003203046740964055,
-      "learning_rate": 0.00019999181831207907,
-      "loss": 46.0,
-      "step": 25295
-    },
-    {
-      "epoch": 4.073754982084625,
-      "grad_norm": 0.001484760083258152,
-      "learning_rate": 0.00019999181766491768,
-      "loss": 46.0,
-      "step": 25296
-    },
-    {
-      "epoch": 4.073916019163413,
-      "grad_norm": 0.006206273566931486,
-      "learning_rate": 0.00019999181701773065,
-      "loss": 46.0,
-      "step": 25297
-    },
-    {
-      "epoch": 4.0740770562422,
-      "grad_norm": 0.0030963770113885403,
-      "learning_rate": 0.00019999181637051806,
-      "loss": 46.0,
-      "step": 25298
-    },
-    {
-      "epoch": 4.074238093320987,
-      "grad_norm": 0.0029490015003830194,
-      "learning_rate": 0.0001999918157232799,
-      "loss": 46.0,
-      "step": 25299
-    },
-    {
-      "epoch": 4.074399130399774,
-      "grad_norm": 0.005219497252255678,
-      "learning_rate": 0.0001999918150760161,
-      "loss": 46.0,
-      "step": 25300
-    },
-    {
-      "epoch": 4.074560167478562,
-      "grad_norm": 0.0019559382926672697,
-      "learning_rate": 0.00019999181442872672,
-      "loss": 46.0,
-      "step": 25301
-    },
-    {
-      "epoch": 4.074721204557349,
-      "grad_norm": 0.0020601944997906685,
-      "learning_rate": 0.00019999181378141175,
-      "loss": 46.0,
-      "step": 25302
-    },
-    {
-      "epoch": 4.0748822416361365,
-      "grad_norm": 0.005165477283298969,
-      "learning_rate": 0.0001999918131340712,
-      "loss": 46.0,
-      "step": 25303
-    },
-    {
-      "epoch": 4.075043278714924,
-      "grad_norm": 0.002091160509735346,
-      "learning_rate": 0.00019999181248670503,
-      "loss": 46.0,
-      "step": 25304
-    },
-    {
-      "epoch": 4.0752043157937115,
-      "grad_norm": 0.012431016191840172,
-      "learning_rate": 0.0001999918118393133,
-      "loss": 46.0,
-      "step": 25305
-    },
-    {
-      "epoch": 4.075365352872499,
-      "grad_norm": 0.0012369583128020167,
-      "learning_rate": 0.00019999181119189597,
-      "loss": 46.0,
-      "step": 25306
-    },
-    {
-      "epoch": 4.075526389951286,
-      "grad_norm": 0.011434996500611305,
-      "learning_rate": 0.000199991810544453,
-      "loss": 46.0,
-      "step": 25307
-    },
-    {
-      "epoch": 4.075687427030074,
-      "grad_norm": 0.005263522733002901,
-      "learning_rate": 0.0001999918098969845,
-      "loss": 46.0,
-      "step": 25308
-    },
-    {
-      "epoch": 4.075848464108861,
-      "grad_norm": 0.0013070355635136366,
-      "learning_rate": 0.0001999918092494904,
-      "loss": 46.0,
-      "step": 25309
-    },
-    {
-      "epoch": 4.076009501187649,
-      "grad_norm": 0.001119088614359498,
-      "learning_rate": 0.00019999180860197068,
-      "loss": 46.0,
-      "step": 25310
-    },
-    {
-      "epoch": 4.076170538266436,
-      "grad_norm": 0.00222882186062634,
-      "learning_rate": 0.00019999180795442537,
-      "loss": 46.0,
-      "step": 25311
-    },
-    {
-      "epoch": 4.076331575345224,
-      "grad_norm": 0.0012359506217762828,
-      "learning_rate": 0.00019999180730685448,
-      "loss": 46.0,
-      "step": 25312
-    },
-    {
-      "epoch": 4.076492612424011,
-      "grad_norm": 0.012910140678286552,
-      "learning_rate": 0.000199991806659258,
-      "loss": 46.0,
-      "step": 25313
-    },
-    {
-      "epoch": 4.076653649502798,
-      "grad_norm": 0.003943998832255602,
-      "learning_rate": 0.0001999918060116359,
-      "loss": 46.0,
-      "step": 25314
-    },
-    {
-      "epoch": 4.076814686581585,
-      "grad_norm": 0.002708324696868658,
-      "learning_rate": 0.00019999180536398826,
-      "loss": 46.0,
-      "step": 25315
-    },
-    {
-      "epoch": 4.0769757236603725,
-      "grad_norm": 0.0022963974624872208,
-      "learning_rate": 0.000199991804716315,
-      "loss": 46.0,
-      "step": 25316
-    },
-    {
-      "epoch": 4.07713676073916,
-      "grad_norm": 0.003000213298946619,
-      "learning_rate": 0.00019999180406861614,
-      "loss": 46.0,
-      "step": 25317
-    },
-    {
-      "epoch": 4.077297797817947,
-      "grad_norm": 0.0036189432721585035,
-      "learning_rate": 0.00019999180342089167,
-      "loss": 46.0,
-      "step": 25318
-    },
-    {
-      "epoch": 4.077458834896735,
-      "grad_norm": 0.0041221617721021175,
-      "learning_rate": 0.0001999918027731416,
-      "loss": 46.0,
-      "step": 25319
-    },
-    {
-      "epoch": 4.077619871975522,
-      "grad_norm": 0.006861657369881868,
-      "learning_rate": 0.000199991802125366,
-      "loss": 46.0,
-      "step": 25320
-    },
-    {
-      "epoch": 4.07778090905431,
-      "grad_norm": 0.0033206609077751637,
-      "learning_rate": 0.00019999180147756477,
-      "loss": 46.0,
-      "step": 25321
-    },
-    {
-      "epoch": 4.077941946133097,
-      "grad_norm": 0.0022297140676528215,
-      "learning_rate": 0.00019999180082973795,
-      "loss": 46.0,
-      "step": 25322
-    },
-    {
-      "epoch": 4.078102983211885,
-      "grad_norm": 0.0053413985297083855,
-      "learning_rate": 0.00019999180018188552,
-      "loss": 46.0,
-      "step": 25323
-    },
-    {
-      "epoch": 4.078264020290672,
-      "grad_norm": 0.016457371413707733,
-      "learning_rate": 0.00019999179953400753,
-      "loss": 46.0,
-      "step": 25324
-    },
-    {
-      "epoch": 4.07842505736946,
-      "grad_norm": 0.01563914865255356,
-      "learning_rate": 0.00019999179888610392,
-      "loss": 46.0,
-      "step": 25325
-    },
-    {
-      "epoch": 4.078586094448247,
-      "grad_norm": 0.003230681410059333,
-      "learning_rate": 0.0001999917982381747,
-      "loss": 46.0,
-      "step": 25326
-    },
-    {
-      "epoch": 4.0787471315270345,
-      "grad_norm": 0.0024274399038404226,
-      "learning_rate": 0.00019999179759021992,
-      "loss": 46.0,
-      "step": 25327
-    },
-    {
-      "epoch": 4.078908168605821,
-      "grad_norm": 0.015444042161107063,
-      "learning_rate": 0.00019999179694223955,
-      "loss": 46.0,
-      "step": 25328
-    },
-    {
-      "epoch": 4.0790692056846085,
-      "grad_norm": 0.007372250314801931,
-      "learning_rate": 0.00019999179629423357,
-      "loss": 46.0,
-      "step": 25329
-    },
-    {
-      "epoch": 4.079230242763396,
-      "grad_norm": 0.002808860968798399,
-      "learning_rate": 0.00019999179564620203,
-      "loss": 46.0,
-      "step": 25330
-    },
-    {
-      "epoch": 4.079391279842183,
-      "grad_norm": 0.011987869627773762,
-      "learning_rate": 0.00019999179499814487,
-      "loss": 46.0,
-      "step": 25331
-    },
-    {
-      "epoch": 4.079552316920971,
-      "grad_norm": 0.0008494609501212835,
-      "learning_rate": 0.00019999179435006213,
-      "loss": 46.0,
-      "step": 25332
-    },
-    {
-      "epoch": 4.079713353999758,
-      "grad_norm": 0.002371512819081545,
-      "learning_rate": 0.00019999179370195375,
-      "loss": 46.0,
-      "step": 25333
-    },
-    {
-      "epoch": 4.079874391078546,
-      "grad_norm": 0.004722840152680874,
-      "learning_rate": 0.00019999179305381983,
-      "loss": 46.0,
-      "step": 25334
-    },
-    {
-      "epoch": 4.080035428157333,
-      "grad_norm": 0.0013415272114798427,
-      "learning_rate": 0.0001999917924056603,
-      "loss": 46.0,
-      "step": 25335
-    },
-    {
-      "epoch": 4.080196465236121,
-      "grad_norm": 0.0025180336087942123,
-      "learning_rate": 0.00019999179175747518,
-      "loss": 46.0,
-      "step": 25336
-    },
-    {
-      "epoch": 4.080357502314908,
-      "grad_norm": 0.005454804748296738,
-      "learning_rate": 0.00019999179110926447,
-      "loss": 46.0,
-      "step": 25337
-    },
-    {
-      "epoch": 4.080518539393696,
-      "grad_norm": 0.0020784582011401653,
-      "learning_rate": 0.00019999179046102815,
-      "loss": 46.0,
-      "step": 25338
-    },
-    {
-      "epoch": 4.080679576472483,
-      "grad_norm": 0.004487820900976658,
-      "learning_rate": 0.00019999178981276624,
-      "loss": 46.0,
-      "step": 25339
-    },
-    {
-      "epoch": 4.0808406135512705,
-      "grad_norm": 0.004285737406462431,
-      "learning_rate": 0.00019999178916447878,
-      "loss": 46.0,
-      "step": 25340
-    },
-    {
-      "epoch": 4.081001650630058,
-      "grad_norm": 0.002617442747578025,
-      "learning_rate": 0.0001999917885161657,
-      "loss": 46.0,
-      "step": 25341
-    },
-    {
-      "epoch": 4.081162687708845,
-      "grad_norm": 0.0018709114519879222,
-      "learning_rate": 0.00019999178786782703,
-      "loss": 46.0,
-      "step": 25342
-    },
-    {
-      "epoch": 4.081323724787632,
-      "grad_norm": 0.0020592729561030865,
-      "learning_rate": 0.00019999178721946272,
-      "loss": 46.0,
-      "step": 25343
-    },
-    {
-      "epoch": 4.081484761866419,
-      "grad_norm": 0.007648243103176355,
-      "learning_rate": 0.00019999178657107287,
-      "loss": 46.0,
-      "step": 25344
-    },
-    {
-      "epoch": 4.081645798945207,
-      "grad_norm": 0.006734162103384733,
-      "learning_rate": 0.00019999178592265741,
-      "loss": 46.0,
-      "step": 25345
-    },
-    {
-      "epoch": 4.081806836023994,
-      "grad_norm": 0.0062074181623756886,
-      "learning_rate": 0.00019999178527421637,
-      "loss": 46.0,
-      "step": 25346
-    },
-    {
-      "epoch": 4.081967873102782,
-      "grad_norm": 0.004024601075798273,
-      "learning_rate": 0.0001999917846257497,
-      "loss": 46.0,
-      "step": 25347
-    },
-    {
-      "epoch": 4.082128910181569,
-      "grad_norm": 0.011393807828426361,
-      "learning_rate": 0.00019999178397725752,
-      "loss": 46.0,
-      "step": 25348
-    },
-    {
-      "epoch": 4.082289947260357,
-      "grad_norm": 0.009684430435299873,
-      "learning_rate": 0.00019999178332873966,
-      "loss": 46.0,
-      "step": 25349
-    },
-    {
-      "epoch": 4.082450984339144,
-      "grad_norm": 0.005655083805322647,
-      "learning_rate": 0.00019999178268019624,
-      "loss": 46.0,
-      "step": 25350
-    },
-    {
-      "epoch": 4.082612021417932,
-      "grad_norm": 0.023631690070033073,
-      "learning_rate": 0.00019999178203162723,
-      "loss": 46.0,
-      "step": 25351
-    },
-    {
-      "epoch": 4.082773058496719,
-      "grad_norm": 0.005316827446222305,
-      "learning_rate": 0.00019999178138303264,
-      "loss": 46.0,
-      "step": 25352
-    },
-    {
-      "epoch": 4.0829340955755065,
-      "grad_norm": 0.008111649192869663,
-      "learning_rate": 0.00019999178073441243,
-      "loss": 46.0,
-      "step": 25353
-    },
-    {
-      "epoch": 4.083095132654294,
-      "grad_norm": 0.007422955706715584,
-      "learning_rate": 0.00019999178008576666,
-      "loss": 46.0,
-      "step": 25354
-    },
-    {
-      "epoch": 4.083256169733081,
-      "grad_norm": 0.005682788789272308,
-      "learning_rate": 0.00019999177943709527,
-      "loss": 46.0,
-      "step": 25355
-    },
-    {
-      "epoch": 4.083417206811869,
-      "grad_norm": 0.006704689469188452,
-      "learning_rate": 0.00019999177878839828,
-      "loss": 46.0,
-      "step": 25356
-    },
-    {
-      "epoch": 4.083578243890656,
-      "grad_norm": 0.006221194285899401,
-      "learning_rate": 0.00019999177813967572,
-      "loss": 46.0,
-      "step": 25357
-    },
-    {
-      "epoch": 4.083739280969443,
-      "grad_norm": 0.006843688897788525,
-      "learning_rate": 0.00019999177749092755,
-      "loss": 46.0,
-      "step": 25358
-    },
-    {
-      "epoch": 4.08390031804823,
-      "grad_norm": 0.00809916015714407,
-      "learning_rate": 0.00019999177684215379,
-      "loss": 46.0,
-      "step": 25359
-    },
-    {
-      "epoch": 4.084061355127018,
-      "grad_norm": 0.0032513048499822617,
-      "learning_rate": 0.00019999177619335444,
-      "loss": 46.0,
-      "step": 25360
-    },
-    {
-      "epoch": 4.084222392205805,
-      "grad_norm": 0.007452653255313635,
-      "learning_rate": 0.0001999917755445295,
-      "loss": 46.0,
-      "step": 25361
-    },
-    {
-      "epoch": 4.084383429284593,
-      "grad_norm": 0.004726361948996782,
-      "learning_rate": 0.00019999177489567899,
-      "loss": 46.0,
-      "step": 25362
-    },
-    {
-      "epoch": 4.08454446636338,
-      "grad_norm": 0.0025916057638823986,
-      "learning_rate": 0.00019999177424680285,
-      "loss": 46.0,
-      "step": 25363
-    },
-    {
-      "epoch": 4.0847055034421675,
-      "grad_norm": 0.0011292851995676756,
-      "learning_rate": 0.0001999917735979011,
-      "loss": 46.0,
-      "step": 25364
-    },
-    {
-      "epoch": 4.084866540520955,
-      "grad_norm": 0.0009537922451272607,
-      "learning_rate": 0.0001999917729489738,
-      "loss": 46.0,
-      "step": 25365
-    },
-    {
-      "epoch": 4.085027577599742,
-      "grad_norm": 0.002446504542604089,
-      "learning_rate": 0.0001999917723000209,
-      "loss": 46.0,
-      "step": 25366
-    },
-    {
-      "epoch": 4.08518861467853,
-      "grad_norm": 0.017476635053753853,
-      "learning_rate": 0.0001999917716510424,
-      "loss": 46.0,
-      "step": 25367
-    },
-    {
-      "epoch": 4.085349651757317,
-      "grad_norm": 0.0022909617982804775,
-      "learning_rate": 0.0001999917710020383,
-      "loss": 46.0,
-      "step": 25368
-    },
-    {
-      "epoch": 4.085510688836105,
-      "grad_norm": 0.0018288498977199197,
-      "learning_rate": 0.00019999177035300865,
-      "loss": 46.0,
-      "step": 25369
-    },
-    {
-      "epoch": 4.085671725914892,
-      "grad_norm": 0.0016076794127002358,
-      "learning_rate": 0.00019999176970395338,
-      "loss": 46.0,
-      "step": 25370
-    },
-    {
-      "epoch": 4.08583276299368,
-      "grad_norm": 0.000692850211635232,
-      "learning_rate": 0.0001999917690548725,
-      "loss": 46.0,
-      "step": 25371
-    },
-    {
-      "epoch": 4.085993800072466,
-      "grad_norm": 0.0034705435391515493,
-      "learning_rate": 0.00019999176840576602,
-      "loss": 46.0,
-      "step": 25372
-    },
-    {
-      "epoch": 4.086154837151254,
-      "grad_norm": 0.0033773507457226515,
-      "learning_rate": 0.000199991767756634,
-      "loss": 46.0,
-      "step": 25373
-    },
-    {
-      "epoch": 4.086315874230041,
-      "grad_norm": 0.00610406044870615,
-      "learning_rate": 0.00019999176710747634,
-      "loss": 46.0,
-      "step": 25374
-    },
-    {
-      "epoch": 4.086476911308829,
-      "grad_norm": 0.009026920422911644,
-      "learning_rate": 0.0001999917664582931,
-      "loss": 46.0,
-      "step": 25375
-    },
-    {
-      "epoch": 4.086637948387616,
-      "grad_norm": 0.003868024330586195,
-      "learning_rate": 0.00019999176580908428,
-      "loss": 46.0,
-      "step": 25376
-    },
-    {
-      "epoch": 4.0867989854664035,
-      "grad_norm": 0.003424036782234907,
-      "learning_rate": 0.00019999176515984982,
-      "loss": 46.0,
-      "step": 25377
-    },
-    {
-      "epoch": 4.086960022545191,
-      "grad_norm": 0.008421915583312511,
-      "learning_rate": 0.00019999176451058982,
-      "loss": 46.0,
-      "step": 25378
-    },
-    {
-      "epoch": 4.087121059623978,
-      "grad_norm": 0.005649816710501909,
-      "learning_rate": 0.0001999917638613042,
-      "loss": 46.0,
-      "step": 25379
-    },
-    {
-      "epoch": 4.087282096702766,
-      "grad_norm": 0.003014152869582176,
-      "learning_rate": 0.000199991763211993,
-      "loss": 46.0,
-      "step": 25380
-    },
-    {
-      "epoch": 4.087443133781553,
-      "grad_norm": 0.0066256653517484665,
-      "learning_rate": 0.00019999176256265623,
-      "loss": 46.0,
-      "step": 25381
-    },
-    {
-      "epoch": 4.087604170860341,
-      "grad_norm": 0.005814736243337393,
-      "learning_rate": 0.00019999176191329383,
-      "loss": 46.0,
-      "step": 25382
-    },
-    {
-      "epoch": 4.087765207939128,
-      "grad_norm": 0.0035878093913197517,
-      "learning_rate": 0.00019999176126390584,
-      "loss": 46.0,
-      "step": 25383
-    },
-    {
-      "epoch": 4.087926245017916,
-      "grad_norm": 0.00922172237187624,
-      "learning_rate": 0.00019999176061449226,
-      "loss": 46.0,
-      "step": 25384
-    },
-    {
-      "epoch": 4.088087282096703,
-      "grad_norm": 0.004588711075484753,
-      "learning_rate": 0.0001999917599650531,
-      "loss": 46.0,
-      "step": 25385
-    },
-    {
-      "epoch": 4.088248319175491,
-      "grad_norm": 0.00894121639430523,
-      "learning_rate": 0.00019999175931558835,
-      "loss": 46.0,
-      "step": 25386
-    },
-    {
-      "epoch": 4.088409356254277,
-      "grad_norm": 0.0024799557868391275,
-      "learning_rate": 0.000199991758666098,
-      "loss": 46.0,
-      "step": 25387
-    },
-    {
-      "epoch": 4.088570393333065,
-      "grad_norm": 0.007205348927527666,
-      "learning_rate": 0.00019999175801658204,
-      "loss": 46.0,
-      "step": 25388
-    },
-    {
-      "epoch": 4.088731430411852,
-      "grad_norm": 0.00122648268006742,
-      "learning_rate": 0.0001999917573670405,
-      "loss": 46.0,
-      "step": 25389
-    },
-    {
-      "epoch": 4.0888924674906395,
-      "grad_norm": 0.0010580611415207386,
-      "learning_rate": 0.00019999175671747338,
-      "loss": 46.0,
-      "step": 25390
-    },
-    {
-      "epoch": 4.089053504569427,
-      "grad_norm": 0.0029637455008924007,
-      "learning_rate": 0.00019999175606788064,
-      "loss": 46.0,
-      "step": 25391
-    },
-    {
-      "epoch": 4.089214541648214,
-      "grad_norm": 0.0019093838054686785,
-      "learning_rate": 0.00019999175541826234,
-      "loss": 46.0,
-      "step": 25392
-    },
-    {
-      "epoch": 4.089375578727002,
-      "grad_norm": 0.0026252414099872112,
-      "learning_rate": 0.00019999175476861846,
-      "loss": 46.0,
-      "step": 25393
-    },
-    {
-      "epoch": 4.089536615805789,
-      "grad_norm": 0.003417298197746277,
-      "learning_rate": 0.00019999175411894893,
-      "loss": 46.0,
-      "step": 25394
-    },
-    {
-      "epoch": 4.089697652884577,
-      "grad_norm": 0.004173026420176029,
-      "learning_rate": 0.00019999175346925382,
-      "loss": 46.0,
-      "step": 25395
-    },
-    {
-      "epoch": 4.089858689963364,
-      "grad_norm": 0.0015862321015447378,
-      "learning_rate": 0.00019999175281953314,
-      "loss": 46.0,
-      "step": 25396
-    },
-    {
-      "epoch": 4.090019727042152,
-      "grad_norm": 0.004064126871526241,
-      "learning_rate": 0.00019999175216978688,
-      "loss": 46.0,
-      "step": 25397
-    },
-    {
-      "epoch": 4.090180764120939,
-      "grad_norm": 0.0025165406987071037,
-      "learning_rate": 0.000199991751520015,
-      "loss": 46.0,
-      "step": 25398
-    },
-    {
-      "epoch": 4.090341801199727,
-      "grad_norm": 0.014682458713650703,
-      "learning_rate": 0.00019999175087021754,
-      "loss": 46.0,
-      "step": 25399
-    },
-    {
-      "epoch": 4.090502838278514,
-      "grad_norm": 0.0016353977844119072,
-      "learning_rate": 0.00019999175022039446,
-      "loss": 46.0,
-      "step": 25400
-    },
-    {
-      "epoch": 4.090663875357301,
-      "grad_norm": 0.002113655675202608,
-      "learning_rate": 0.0001999917495705458,
-      "loss": 46.0,
-      "step": 25401
-    },
-    {
-      "epoch": 4.090824912436088,
-      "grad_norm": 0.00485873082652688,
-      "learning_rate": 0.00019999174892067158,
-      "loss": 46.0,
-      "step": 25402
-    },
-    {
-      "epoch": 4.0909859495148755,
-      "grad_norm": 0.007050623185932636,
-      "learning_rate": 0.00019999174827077174,
-      "loss": 46.0,
-      "step": 25403
-    },
-    {
-      "epoch": 4.091146986593663,
-      "grad_norm": 0.013660944066941738,
-      "learning_rate": 0.0001999917476208463,
-      "loss": 46.0,
-      "step": 25404
-    },
-    {
-      "epoch": 4.09130802367245,
-      "grad_norm": 0.004856301471590996,
-      "learning_rate": 0.0001999917469708953,
-      "loss": 46.0,
-      "step": 25405
-    },
-    {
-      "epoch": 4.091469060751238,
-      "grad_norm": 0.002449101535603404,
-      "learning_rate": 0.00019999174632091867,
-      "loss": 46.0,
-      "step": 25406
-    },
-    {
-      "epoch": 4.091630097830025,
-      "grad_norm": 0.0032331738620996475,
-      "learning_rate": 0.00019999174567091648,
-      "loss": 46.0,
-      "step": 25407
-    },
-    {
-      "epoch": 4.091791134908813,
-      "grad_norm": 0.004462626297026873,
-      "learning_rate": 0.00019999174502088865,
-      "loss": 46.0,
-      "step": 25408
-    },
-    {
-      "epoch": 4.0919521719876,
-      "grad_norm": 0.0015092194080352783,
-      "learning_rate": 0.00019999174437083526,
-      "loss": 46.0,
-      "step": 25409
-    },
-    {
-      "epoch": 4.092113209066388,
-      "grad_norm": 0.012744955718517303,
-      "learning_rate": 0.0001999917437207563,
-      "loss": 46.0,
-      "step": 25410
-    },
-    {
-      "epoch": 4.092274246145175,
-      "grad_norm": 0.0034201527014374733,
-      "learning_rate": 0.00019999174307065173,
-      "loss": 46.0,
-      "step": 25411
-    },
-    {
-      "epoch": 4.0924352832239625,
-      "grad_norm": 0.0012307639699429274,
-      "learning_rate": 0.00019999174242052155,
-      "loss": 46.0,
-      "step": 25412
-    },
-    {
-      "epoch": 4.09259632030275,
-      "grad_norm": 0.0033559866715222597,
-      "learning_rate": 0.00019999174177036578,
-      "loss": 46.0,
-      "step": 25413
-    },
-    {
-      "epoch": 4.0927573573815375,
-      "grad_norm": 0.0020630250219255686,
-      "learning_rate": 0.0001999917411201844,
-      "loss": 46.0,
-      "step": 25414
-    },
-    {
-      "epoch": 4.092918394460325,
-      "grad_norm": 0.0012368483003228903,
-      "learning_rate": 0.00019999174046997747,
-      "loss": 46.0,
-      "step": 25415
-    },
-    {
-      "epoch": 4.0930794315391115,
-      "grad_norm": 0.0018089090008288622,
-      "learning_rate": 0.00019999173981974494,
-      "loss": 46.0,
-      "step": 25416
-    },
-    {
-      "epoch": 4.093240468617899,
-      "grad_norm": 0.013004492968320847,
-      "learning_rate": 0.0001999917391694868,
-      "loss": 46.0,
-      "step": 25417
-    },
-    {
-      "epoch": 4.093401505696686,
-      "grad_norm": 0.008340667001903057,
-      "learning_rate": 0.00019999173851920307,
-      "loss": 46.0,
-      "step": 25418
-    },
-    {
-      "epoch": 4.093562542775474,
-      "grad_norm": 0.002470900071784854,
-      "learning_rate": 0.00019999173786889376,
-      "loss": 46.0,
-      "step": 25419
-    },
-    {
-      "epoch": 4.093723579854261,
-      "grad_norm": 0.00087618303950876,
-      "learning_rate": 0.00019999173721855883,
-      "loss": 46.0,
-      "step": 25420
-    },
-    {
-      "epoch": 4.093884616933049,
-      "grad_norm": 0.004502859897911549,
-      "learning_rate": 0.00019999173656819834,
-      "loss": 46.0,
-      "step": 25421
-    },
-    {
-      "epoch": 4.094045654011836,
-      "grad_norm": 0.0018137659644708037,
-      "learning_rate": 0.00019999173591781223,
-      "loss": 46.0,
-      "step": 25422
-    },
-    {
-      "epoch": 4.094206691090624,
-      "grad_norm": 0.001247807638719678,
-      "learning_rate": 0.00019999173526740054,
-      "loss": 46.0,
-      "step": 25423
-    },
-    {
-      "epoch": 4.094367728169411,
-      "grad_norm": 0.003618153277784586,
-      "learning_rate": 0.00019999173461696327,
-      "loss": 46.0,
-      "step": 25424
-    },
-    {
-      "epoch": 4.0945287652481985,
-      "grad_norm": 0.0016069714911282063,
-      "learning_rate": 0.00019999173396650037,
-      "loss": 46.0,
-      "step": 25425
-    },
-    {
-      "epoch": 4.094689802326986,
-      "grad_norm": 0.0011741031194105744,
-      "learning_rate": 0.00019999173331601192,
-      "loss": 46.0,
-      "step": 25426
-    },
-    {
-      "epoch": 4.094850839405773,
-      "grad_norm": 0.001205502892844379,
-      "learning_rate": 0.00019999173266549785,
-      "loss": 46.0,
-      "step": 25427
-    },
-    {
-      "epoch": 4.095011876484561,
-      "grad_norm": 0.0017412963788956404,
-      "learning_rate": 0.0001999917320149582,
-      "loss": 46.0,
-      "step": 25428
-    },
-    {
-      "epoch": 4.095172913563348,
-      "grad_norm": 0.0013506341492757201,
-      "learning_rate": 0.00019999173136439296,
-      "loss": 46.0,
-      "step": 25429
-    },
-    {
-      "epoch": 4.095333950642136,
-      "grad_norm": 0.0027436986565589905,
-      "learning_rate": 0.00019999173071380208,
-      "loss": 46.0,
-      "step": 25430
-    },
-    {
-      "epoch": 4.095494987720922,
-      "grad_norm": 0.002620898885652423,
-      "learning_rate": 0.0001999917300631857,
-      "loss": 46.0,
-      "step": 25431
-    },
-    {
-      "epoch": 4.09565602479971,
-      "grad_norm": 0.01690833829343319,
-      "learning_rate": 0.00019999172941254366,
-      "loss": 46.0,
-      "step": 25432
-    },
-    {
-      "epoch": 4.095817061878497,
-      "grad_norm": 0.0021657529287040234,
-      "learning_rate": 0.00019999172876187602,
-      "loss": 46.0,
-      "step": 25433
-    },
-    {
-      "epoch": 4.095978098957285,
-      "grad_norm": 0.013728772290050983,
-      "learning_rate": 0.0001999917281111828,
-      "loss": 46.0,
-      "step": 25434
-    },
-    {
-      "epoch": 4.096139136036072,
-      "grad_norm": 0.008101335726678371,
-      "learning_rate": 0.000199991727460464,
-      "loss": 46.0,
-      "step": 25435
-    },
-    {
-      "epoch": 4.09630017311486,
-      "grad_norm": 0.0013107523554936051,
-      "learning_rate": 0.00019999172680971962,
-      "loss": 46.0,
-      "step": 25436
-    },
-    {
-      "epoch": 4.096461210193647,
-      "grad_norm": 0.003535193158313632,
-      "learning_rate": 0.00019999172615894962,
-      "loss": 46.0,
-      "step": 25437
-    },
-    {
-      "epoch": 4.0966222472724345,
-      "grad_norm": 0.009353358298540115,
-      "learning_rate": 0.00019999172550815405,
-      "loss": 46.0,
-      "step": 25438
-    },
-    {
-      "epoch": 4.096783284351222,
-      "grad_norm": 0.0039663491770625114,
-      "learning_rate": 0.00019999172485733285,
-      "loss": 46.0,
-      "step": 25439
-    },
-    {
-      "epoch": 4.096944321430009,
-      "grad_norm": 0.0015022153966128826,
-      "learning_rate": 0.0001999917242064861,
-      "loss": 46.0,
-      "step": 25440
-    },
-    {
-      "epoch": 4.097105358508797,
-      "grad_norm": 0.0021316735073924065,
-      "learning_rate": 0.00019999172355561373,
-      "loss": 46.0,
-      "step": 25441
-    },
-    {
-      "epoch": 4.097266395587584,
-      "grad_norm": 0.005879572592675686,
-      "learning_rate": 0.00019999172290471577,
-      "loss": 46.0,
-      "step": 25442
-    },
-    {
-      "epoch": 4.097427432666372,
-      "grad_norm": 0.005232840310782194,
-      "learning_rate": 0.00019999172225379223,
-      "loss": 46.0,
-      "step": 25443
-    },
-    {
-      "epoch": 4.097588469745159,
-      "grad_norm": 0.002854888793081045,
-      "learning_rate": 0.00019999172160284307,
-      "loss": 46.0,
-      "step": 25444
-    },
-    {
-      "epoch": 4.097749506823946,
-      "grad_norm": 0.010229691863059998,
-      "learning_rate": 0.00019999172095186836,
-      "loss": 46.0,
-      "step": 25445
-    },
-    {
-      "epoch": 4.097910543902733,
-      "grad_norm": 0.0035191874485462904,
-      "learning_rate": 0.00019999172030086803,
-      "loss": 46.0,
-      "step": 25446
-    },
-    {
-      "epoch": 4.098071580981521,
-      "grad_norm": 0.017856374382972717,
-      "learning_rate": 0.0001999917196498421,
-      "loss": 46.0,
-      "step": 25447
-    },
-    {
-      "epoch": 4.098232618060308,
-      "grad_norm": 0.0037848991341888905,
-      "learning_rate": 0.00019999171899879058,
-      "loss": 46.0,
-      "step": 25448
-    },
-    {
-      "epoch": 4.098393655139096,
-      "grad_norm": 0.01623384840786457,
-      "learning_rate": 0.0001999917183477135,
-      "loss": 46.0,
-      "step": 25449
-    },
-    {
-      "epoch": 4.098554692217883,
-      "grad_norm": 0.007264601066708565,
-      "learning_rate": 0.0001999917176966108,
-      "loss": 46.0,
-      "step": 25450
-    },
-    {
-      "epoch": 4.0987157292966705,
-      "grad_norm": 0.0008852232131175697,
-      "learning_rate": 0.0001999917170454825,
-      "loss": 46.0,
-      "step": 25451
-    },
-    {
-      "epoch": 4.098876766375458,
-      "grad_norm": 0.0022032661363482475,
-      "learning_rate": 0.00019999171639432863,
-      "loss": 46.0,
-      "step": 25452
-    },
-    {
-      "epoch": 4.099037803454245,
-      "grad_norm": 0.008992953225970268,
-      "learning_rate": 0.00019999171574314916,
-      "loss": 46.0,
-      "step": 25453
-    },
-    {
-      "epoch": 4.099198840533033,
-      "grad_norm": 0.00500809820368886,
-      "learning_rate": 0.0001999917150919441,
-      "loss": 46.0,
-      "step": 25454
-    },
-    {
-      "epoch": 4.09935987761182,
-      "grad_norm": 0.007882903330028057,
-      "learning_rate": 0.00019999171444071343,
-      "loss": 46.0,
-      "step": 25455
-    },
-    {
-      "epoch": 4.099520914690608,
-      "grad_norm": 0.011945423670113087,
-      "learning_rate": 0.00019999171378945718,
-      "loss": 46.0,
-      "step": 25456
-    },
-    {
-      "epoch": 4.099681951769395,
-      "grad_norm": 0.0037769211921840906,
-      "learning_rate": 0.00019999171313817536,
-      "loss": 46.0,
-      "step": 25457
-    },
-    {
-      "epoch": 4.099842988848183,
-      "grad_norm": 0.003988330252468586,
-      "learning_rate": 0.0001999917124868679,
-      "loss": 46.0,
-      "step": 25458
-    },
-    {
-      "epoch": 4.10000402592697,
-      "grad_norm": 0.004875501152127981,
-      "learning_rate": 0.00019999171183553486,
-      "loss": 46.0,
-      "step": 25459
-    },
-    {
-      "epoch": 4.100165063005757,
-      "grad_norm": 0.0023024508263915777,
-      "learning_rate": 0.00019999171118417625,
-      "loss": 46.0,
-      "step": 25460
-    },
-    {
-      "epoch": 4.100326100084544,
-      "grad_norm": 0.00654234504327178,
-      "learning_rate": 0.00019999171053279203,
-      "loss": 46.0,
-      "step": 25461
-    },
-    {
-      "epoch": 4.100487137163332,
-      "grad_norm": 0.002930715214461088,
-      "learning_rate": 0.00019999170988138222,
-      "loss": 46.0,
-      "step": 25462
-    },
-    {
-      "epoch": 4.100648174242119,
-      "grad_norm": 0.009244444780051708,
-      "learning_rate": 0.0001999917092299468,
-      "loss": 46.0,
-      "step": 25463
-    },
-    {
-      "epoch": 4.1008092113209065,
-      "grad_norm": 0.0007282950100488961,
-      "learning_rate": 0.00019999170857848585,
-      "loss": 46.0,
-      "step": 25464
-    },
-    {
-      "epoch": 4.100970248399694,
-      "grad_norm": 0.005571243353188038,
-      "learning_rate": 0.00019999170792699925,
-      "loss": 46.0,
-      "step": 25465
-    },
-    {
-      "epoch": 4.101131285478481,
-      "grad_norm": 0.00936428364366293,
-      "learning_rate": 0.00019999170727548707,
-      "loss": 46.0,
-      "step": 25466
-    },
-    {
-      "epoch": 4.101292322557269,
-      "grad_norm": 0.0033771838061511517,
-      "learning_rate": 0.0001999917066239493,
-      "loss": 46.0,
-      "step": 25467
-    },
-    {
-      "epoch": 4.101453359636056,
-      "grad_norm": 0.006844170391559601,
-      "learning_rate": 0.00019999170597238591,
-      "loss": 46.0,
-      "step": 25468
-    },
-    {
-      "epoch": 4.101614396714844,
-      "grad_norm": 0.0017905480926856399,
-      "learning_rate": 0.00019999170532079697,
-      "loss": 46.0,
-      "step": 25469
-    },
-    {
-      "epoch": 4.101775433793631,
-      "grad_norm": 0.0075187901966273785,
-      "learning_rate": 0.0001999917046691824,
-      "loss": 46.0,
-      "step": 25470
-    },
-    {
-      "epoch": 4.101936470872419,
-      "grad_norm": 0.005218985956162214,
-      "learning_rate": 0.00019999170401754226,
-      "loss": 46.0,
-      "step": 25471
-    },
-    {
-      "epoch": 4.102097507951206,
-      "grad_norm": 0.005367629695683718,
-      "learning_rate": 0.00019999170336587653,
-      "loss": 46.0,
-      "step": 25472
-    },
-    {
-      "epoch": 4.1022585450299935,
-      "grad_norm": 0.010221910662949085,
-      "learning_rate": 0.00019999170271418518,
-      "loss": 46.0,
-      "step": 25473
-    },
-    {
-      "epoch": 4.10241958210878,
-      "grad_norm": 0.004604964051395655,
-      "learning_rate": 0.00019999170206246828,
-      "loss": 46.0,
-      "step": 25474
-    },
-    {
-      "epoch": 4.102580619187568,
-      "grad_norm": 0.003356988076120615,
-      "learning_rate": 0.00019999170141072576,
-      "loss": 46.0,
-      "step": 25475
-    },
-    {
-      "epoch": 4.102741656266355,
-      "grad_norm": 0.0037764401640743017,
-      "learning_rate": 0.00019999170075895765,
-      "loss": 46.0,
-      "step": 25476
-    },
-    {
-      "epoch": 4.1029026933451425,
-      "grad_norm": 0.0042011430487036705,
-      "learning_rate": 0.00019999170010716398,
-      "loss": 46.0,
-      "step": 25477
-    },
-    {
-      "epoch": 4.10306373042393,
-      "grad_norm": 0.0023326058872044086,
-      "learning_rate": 0.00019999169945534467,
-      "loss": 46.0,
-      "step": 25478
-    },
-    {
-      "epoch": 4.103224767502717,
-      "grad_norm": 0.004028123803436756,
-      "learning_rate": 0.00019999169880349977,
-      "loss": 46.0,
-      "step": 25479
-    },
-    {
-      "epoch": 4.103385804581505,
-      "grad_norm": 0.004875359125435352,
-      "learning_rate": 0.0001999916981516293,
-      "loss": 46.0,
-      "step": 25480
-    },
-    {
-      "epoch": 4.103546841660292,
-      "grad_norm": 0.007139843422919512,
-      "learning_rate": 0.00019999169749973324,
-      "loss": 46.0,
-      "step": 25481
-    },
-    {
-      "epoch": 4.10370787873908,
-      "grad_norm": 0.008145545609295368,
-      "learning_rate": 0.00019999169684781158,
-      "loss": 46.0,
-      "step": 25482
-    },
-    {
-      "epoch": 4.103868915817867,
-      "grad_norm": 0.003824364859610796,
-      "learning_rate": 0.00019999169619586434,
-      "loss": 46.0,
-      "step": 25483
-    },
-    {
-      "epoch": 4.104029952896655,
-      "grad_norm": 0.0010947324335575104,
-      "learning_rate": 0.00019999169554389145,
-      "loss": 46.0,
-      "step": 25484
-    },
-    {
-      "epoch": 4.104190989975442,
-      "grad_norm": 0.001225115149281919,
-      "learning_rate": 0.00019999169489189303,
-      "loss": 46.0,
-      "step": 25485
-    },
-    {
-      "epoch": 4.1043520270542295,
-      "grad_norm": 0.004524054005742073,
-      "learning_rate": 0.000199991694239869,
-      "loss": 46.0,
-      "step": 25486
-    },
-    {
-      "epoch": 4.104513064133017,
-      "grad_norm": 0.004667694214731455,
-      "learning_rate": 0.00019999169358781934,
-      "loss": 46.0,
-      "step": 25487
-    },
-    {
-      "epoch": 4.104674101211804,
-      "grad_norm": 0.00201613362878561,
-      "learning_rate": 0.00019999169293574416,
-      "loss": 46.0,
-      "step": 25488
-    },
-    {
-      "epoch": 4.104835138290591,
-      "grad_norm": 0.007933225482702255,
-      "learning_rate": 0.00019999169228364334,
-      "loss": 46.0,
-      "step": 25489
-    },
-    {
-      "epoch": 4.104996175369378,
-      "grad_norm": 0.005335974041372538,
-      "learning_rate": 0.00019999169163151693,
-      "loss": 46.0,
-      "step": 25490
-    },
-    {
-      "epoch": 4.105157212448166,
-      "grad_norm": 0.001674870727583766,
-      "learning_rate": 0.00019999169097936493,
-      "loss": 46.0,
-      "step": 25491
-    },
-    {
-      "epoch": 4.105318249526953,
-      "grad_norm": 0.0024037370458245277,
-      "learning_rate": 0.00019999169032718732,
-      "loss": 46.0,
-      "step": 25492
-    },
-    {
-      "epoch": 4.105479286605741,
-      "grad_norm": 0.006767706014215946,
-      "learning_rate": 0.00019999168967498415,
-      "loss": 46.0,
-      "step": 25493
-    },
-    {
-      "epoch": 4.105640323684528,
-      "grad_norm": 0.002236138330772519,
-      "learning_rate": 0.00019999168902275539,
-      "loss": 46.0,
-      "step": 25494
-    },
-    {
-      "epoch": 4.105801360763316,
-      "grad_norm": 0.001654462656006217,
-      "learning_rate": 0.00019999168837050099,
-      "loss": 46.0,
-      "step": 25495
-    },
-    {
-      "epoch": 4.105962397842103,
-      "grad_norm": 0.014386680908501148,
-      "learning_rate": 0.00019999168771822102,
-      "loss": 46.0,
-      "step": 25496
-    },
-    {
-      "epoch": 4.106123434920891,
-      "grad_norm": 0.003681241301819682,
-      "learning_rate": 0.00019999168706591548,
-      "loss": 46.0,
-      "step": 25497
-    },
-    {
-      "epoch": 4.106284471999678,
-      "grad_norm": 0.0005037786322645843,
-      "learning_rate": 0.00019999168641358434,
-      "loss": 46.0,
-      "step": 25498
-    },
-    {
-      "epoch": 4.1064455090784655,
-      "grad_norm": 0.007998153567314148,
-      "learning_rate": 0.0001999916857612276,
-      "loss": 46.0,
-      "step": 25499
-    },
-    {
-      "epoch": 4.106606546157253,
-      "grad_norm": 0.006519208196550608,
-      "learning_rate": 0.00019999168510884528,
-      "loss": 46.0,
-      "step": 25500
-    },
-    {
-      "epoch": 4.10676758323604,
-      "grad_norm": 0.012437457218766212,
-      "learning_rate": 0.00019999168445643733,
-      "loss": 46.0,
-      "step": 25501
-    },
-    {
-      "epoch": 4.106928620314828,
-      "grad_norm": 0.0024969023652374744,
-      "learning_rate": 0.00019999168380400382,
-      "loss": 46.0,
-      "step": 25502
-    },
-    {
-      "epoch": 4.107089657393615,
-      "grad_norm": 0.007852721959352493,
-      "learning_rate": 0.0001999916831515447,
-      "loss": 46.0,
-      "step": 25503
-    },
-    {
-      "epoch": 4.107250694472402,
-      "grad_norm": 0.0032842422369867563,
-      "learning_rate": 0.00019999168249906,
-      "loss": 46.0,
-      "step": 25504
-    },
-    {
-      "epoch": 4.107411731551189,
-      "grad_norm": 0.001273187343031168,
-      "learning_rate": 0.0001999916818465497,
-      "loss": 46.0,
-      "step": 25505
-    },
-    {
-      "epoch": 4.107572768629977,
-      "grad_norm": 0.01275511085987091,
-      "learning_rate": 0.00019999168119401382,
-      "loss": 46.0,
-      "step": 25506
-    },
-    {
-      "epoch": 4.107733805708764,
-      "grad_norm": 0.004364375956356525,
-      "learning_rate": 0.00019999168054145235,
-      "loss": 46.0,
-      "step": 25507
-    },
-    {
-      "epoch": 4.107894842787552,
-      "grad_norm": 0.0128211984410882,
-      "learning_rate": 0.00019999167988886526,
-      "loss": 46.0,
-      "step": 25508
-    },
-    {
-      "epoch": 4.108055879866339,
-      "grad_norm": 0.014299626462161541,
-      "learning_rate": 0.0001999916792362526,
-      "loss": 46.0,
-      "step": 25509
-    },
-    {
-      "epoch": 4.108216916945127,
-      "grad_norm": 0.0030823044944554567,
-      "learning_rate": 0.00019999167858361435,
-      "loss": 46.0,
-      "step": 25510
-    },
-    {
-      "epoch": 4.108377954023914,
-      "grad_norm": 0.0015012375079095364,
-      "learning_rate": 0.00019999167793095048,
-      "loss": 46.0,
-      "step": 25511
-    },
-    {
-      "epoch": 4.1085389911027015,
-      "grad_norm": 0.0028856382705271244,
-      "learning_rate": 0.000199991677278261,
-      "loss": 46.0,
-      "step": 25512
-    },
-    {
-      "epoch": 4.108700028181489,
-      "grad_norm": 0.001829297747462988,
-      "learning_rate": 0.000199991676625546,
-      "loss": 46.0,
-      "step": 25513
-    },
-    {
-      "epoch": 4.108861065260276,
-      "grad_norm": 0.001995287137106061,
-      "learning_rate": 0.00019999167597280538,
-      "loss": 46.0,
-      "step": 25514
-    },
-    {
-      "epoch": 4.109022102339064,
-      "grad_norm": 0.004471911583095789,
-      "learning_rate": 0.00019999167532003912,
-      "loss": 46.0,
-      "step": 25515
-    },
-    {
-      "epoch": 4.109183139417851,
-      "grad_norm": 0.0019047876121476293,
-      "learning_rate": 0.00019999167466724734,
-      "loss": 46.0,
-      "step": 25516
-    },
-    {
-      "epoch": 4.109344176496639,
-      "grad_norm": 0.0047268345952034,
-      "learning_rate": 0.0001999916740144299,
-      "loss": 46.0,
-      "step": 25517
-    },
-    {
-      "epoch": 4.109505213575425,
-      "grad_norm": 0.0009277617791667581,
-      "learning_rate": 0.00019999167336158692,
-      "loss": 46.0,
-      "step": 25518
-    },
-    {
-      "epoch": 4.109666250654213,
-      "grad_norm": 0.006467899307608604,
-      "learning_rate": 0.00019999167270871832,
-      "loss": 46.0,
-      "step": 25519
-    },
-    {
-      "epoch": 4.109827287733,
-      "grad_norm": 0.00186252873390913,
-      "learning_rate": 0.00019999167205582413,
-      "loss": 46.0,
-      "step": 25520
-    },
-    {
-      "epoch": 4.109988324811788,
-      "grad_norm": 0.0011660184245556593,
-      "learning_rate": 0.00019999167140290436,
-      "loss": 46.0,
-      "step": 25521
-    },
-    {
-      "epoch": 4.110149361890575,
-      "grad_norm": 0.004667657893151045,
-      "learning_rate": 0.000199991670749959,
-      "loss": 46.0,
-      "step": 25522
-    },
-    {
-      "epoch": 4.110310398969363,
-      "grad_norm": 0.002636682940647006,
-      "learning_rate": 0.00019999167009698802,
-      "loss": 46.0,
-      "step": 25523
-    },
-    {
-      "epoch": 4.11047143604815,
-      "grad_norm": 0.002740677446126938,
-      "learning_rate": 0.00019999166944399146,
-      "loss": 46.0,
-      "step": 25524
-    },
-    {
-      "epoch": 4.1106324731269375,
-      "grad_norm": 0.001717817853204906,
-      "learning_rate": 0.0001999916687909693,
-      "loss": 46.0,
-      "step": 25525
-    },
-    {
-      "epoch": 4.110793510205725,
-      "grad_norm": 0.006547986064106226,
-      "learning_rate": 0.00019999166813792157,
-      "loss": 46.0,
-      "step": 25526
-    },
-    {
-      "epoch": 4.110954547284512,
-      "grad_norm": 0.0006603376241400838,
-      "learning_rate": 0.00019999166748484824,
-      "loss": 46.0,
-      "step": 25527
-    },
-    {
-      "epoch": 4.1111155843633,
-      "grad_norm": 0.013379744254052639,
-      "learning_rate": 0.00019999166683174927,
-      "loss": 46.0,
-      "step": 25528
-    },
-    {
-      "epoch": 4.111276621442087,
-      "grad_norm": 0.0016734004020690918,
-      "learning_rate": 0.00019999166617862477,
-      "loss": 46.0,
-      "step": 25529
-    },
-    {
-      "epoch": 4.111437658520875,
-      "grad_norm": 0.008375810459256172,
-      "learning_rate": 0.00019999166552547466,
-      "loss": 46.0,
-      "step": 25530
-    },
-    {
-      "epoch": 4.111598695599662,
-      "grad_norm": 0.0014235987327992916,
-      "learning_rate": 0.00019999166487229896,
-      "loss": 46.0,
-      "step": 25531
-    },
-    {
-      "epoch": 4.11175973267845,
-      "grad_norm": 0.005986319854855537,
-      "learning_rate": 0.00019999166421909764,
-      "loss": 46.0,
-      "step": 25532
-    },
-    {
-      "epoch": 4.111920769757236,
-      "grad_norm": 0.006131486501544714,
-      "learning_rate": 0.00019999166356587077,
-      "loss": 46.0,
-      "step": 25533
-    },
-    {
-      "epoch": 4.112081806836024,
-      "grad_norm": 0.008920717053115368,
-      "learning_rate": 0.00019999166291261825,
-      "loss": 46.0,
-      "step": 25534
-    },
-    {
-      "epoch": 4.112242843914811,
-      "grad_norm": 0.0019028628012165427,
-      "learning_rate": 0.00019999166225934017,
-      "loss": 46.0,
-      "step": 25535
-    },
-    {
-      "epoch": 4.1124038809935985,
-      "grad_norm": 0.002718705916777253,
-      "learning_rate": 0.00019999166160603653,
-      "loss": 46.0,
-      "step": 25536
-    },
-    {
-      "epoch": 4.112564918072386,
-      "grad_norm": 0.0020678136497735977,
-      "learning_rate": 0.00019999166095270726,
-      "loss": 46.0,
-      "step": 25537
-    },
-    {
-      "epoch": 4.1127259551511735,
-      "grad_norm": 0.0060973805375397205,
-      "learning_rate": 0.0001999916602993524,
-      "loss": 46.0,
-      "step": 25538
-    },
-    {
-      "epoch": 4.112886992229961,
-      "grad_norm": 0.00440619932487607,
-      "learning_rate": 0.00019999165964597194,
-      "loss": 46.0,
-      "step": 25539
-    },
-    {
-      "epoch": 4.113048029308748,
-      "grad_norm": 0.0037918188609182835,
-      "learning_rate": 0.0001999916589925659,
-      "loss": 46.0,
-      "step": 25540
-    },
-    {
-      "epoch": 4.113209066387536,
-      "grad_norm": 0.001046994817443192,
-      "learning_rate": 0.00019999165833913427,
-      "loss": 46.0,
-      "step": 25541
-    },
-    {
-      "epoch": 4.113370103466323,
-      "grad_norm": 0.006555836647748947,
-      "learning_rate": 0.00019999165768567703,
-      "loss": 46.0,
-      "step": 25542
-    },
-    {
-      "epoch": 4.113531140545111,
-      "grad_norm": 0.0010875486768782139,
-      "learning_rate": 0.00019999165703219423,
-      "loss": 46.0,
-      "step": 25543
-    },
-    {
-      "epoch": 4.113692177623898,
-      "grad_norm": 0.0016576419584453106,
-      "learning_rate": 0.0001999916563786858,
-      "loss": 46.0,
-      "step": 25544
-    },
-    {
-      "epoch": 4.113853214702686,
-      "grad_norm": 0.004397212993353605,
-      "learning_rate": 0.0001999916557251518,
-      "loss": 46.0,
-      "step": 25545
-    },
-    {
-      "epoch": 4.114014251781473,
-      "grad_norm": 0.003139763604849577,
-      "learning_rate": 0.00019999165507159222,
-      "loss": 46.0,
-      "step": 25546
-    },
-    {
-      "epoch": 4.11417528886026,
-      "grad_norm": 0.007025458384305239,
-      "learning_rate": 0.000199991654418007,
-      "loss": 46.0,
-      "step": 25547
-    },
-    {
-      "epoch": 4.114336325939047,
-      "grad_norm": 0.001014004461467266,
-      "learning_rate": 0.00019999165376439622,
-      "loss": 46.0,
-      "step": 25548
-    },
-    {
-      "epoch": 4.1144973630178345,
-      "grad_norm": 0.0013552741147577763,
-      "learning_rate": 0.00019999165311075984,
-      "loss": 46.0,
-      "step": 25549
-    },
-    {
-      "epoch": 4.114658400096622,
-      "grad_norm": 0.004622077569365501,
-      "learning_rate": 0.00019999165245709787,
-      "loss": 46.0,
-      "step": 25550
-    },
-    {
-      "epoch": 4.114819437175409,
-      "grad_norm": 0.004090077243745327,
-      "learning_rate": 0.0001999916518034103,
-      "loss": 46.0,
-      "step": 25551
-    },
-    {
-      "epoch": 4.114980474254197,
-      "grad_norm": 0.02914608083665371,
-      "learning_rate": 0.00019999165114969715,
-      "loss": 46.0,
-      "step": 25552
-    },
-    {
-      "epoch": 4.115141511332984,
-      "grad_norm": 0.000994830159470439,
-      "learning_rate": 0.00019999165049595843,
-      "loss": 46.0,
-      "step": 25553
-    },
-    {
-      "epoch": 4.115302548411772,
-      "grad_norm": 0.0025630900636315346,
-      "learning_rate": 0.00019999164984219408,
-      "loss": 46.0,
-      "step": 25554
-    },
-    {
-      "epoch": 4.115463585490559,
-      "grad_norm": 0.001853022607974708,
-      "learning_rate": 0.00019999164918840413,
-      "loss": 46.0,
-      "step": 25555
-    },
-    {
-      "epoch": 4.115624622569347,
-      "grad_norm": 0.00142704788595438,
-      "learning_rate": 0.0001999916485345886,
-      "loss": 46.0,
-      "step": 25556
-    },
-    {
-      "epoch": 4.115785659648134,
-      "grad_norm": 0.009458962827920914,
-      "learning_rate": 0.00019999164788074748,
-      "loss": 46.0,
-      "step": 25557
-    },
-    {
-      "epoch": 4.115946696726922,
-      "grad_norm": 0.0011380433570593596,
-      "learning_rate": 0.0001999916472268808,
-      "loss": 46.0,
-      "step": 25558
-    },
-    {
-      "epoch": 4.116107733805709,
-      "grad_norm": 0.0017686458304524422,
-      "learning_rate": 0.00019999164657298849,
-      "loss": 46.0,
-      "step": 25559
-    },
-    {
-      "epoch": 4.1162687708844965,
-      "grad_norm": 0.005422645248472691,
-      "learning_rate": 0.0001999916459190706,
-      "loss": 46.0,
-      "step": 25560
-    },
-    {
-      "epoch": 4.116429807963284,
-      "grad_norm": 0.0033317890483886003,
-      "learning_rate": 0.00019999164526512711,
-      "loss": 46.0,
-      "step": 25561
-    },
-    {
-      "epoch": 4.1165908450420705,
-      "grad_norm": 0.0007542499806731939,
-      "learning_rate": 0.000199991644611158,
-      "loss": 46.0,
-      "step": 25562
-    },
-    {
-      "epoch": 4.116751882120858,
-      "grad_norm": 0.015065771527588367,
-      "learning_rate": 0.0001999916439571633,
-      "loss": 46.0,
-      "step": 25563
-    },
-    {
-      "epoch": 4.116912919199645,
-      "grad_norm": 0.003225766122341156,
-      "learning_rate": 0.00019999164330314307,
-      "loss": 46.0,
-      "step": 25564
-    },
-    {
-      "epoch": 4.117073956278433,
-      "grad_norm": 0.0027500493451952934,
-      "learning_rate": 0.0001999916426490972,
-      "loss": 46.0,
-      "step": 25565
-    },
-    {
-      "epoch": 4.11723499335722,
-      "grad_norm": 0.0054383049719035625,
-      "learning_rate": 0.00019999164199502575,
-      "loss": 46.0,
-      "step": 25566
-    },
-    {
-      "epoch": 4.117396030436008,
-      "grad_norm": 0.00944253895431757,
-      "learning_rate": 0.00019999164134092872,
-      "loss": 46.0,
-      "step": 25567
-    },
-    {
-      "epoch": 4.117557067514795,
-      "grad_norm": 0.004202804993838072,
-      "learning_rate": 0.00019999164068680605,
-      "loss": 46.0,
-      "step": 25568
-    },
-    {
-      "epoch": 4.117718104593583,
-      "grad_norm": 0.0015304501866921782,
-      "learning_rate": 0.00019999164003265784,
-      "loss": 46.0,
-      "step": 25569
-    },
-    {
-      "epoch": 4.11787914167237,
-      "grad_norm": 0.011983905918896198,
-      "learning_rate": 0.00019999163937848403,
-      "loss": 46.0,
-      "step": 25570
-    },
-    {
-      "epoch": 4.118040178751158,
-      "grad_norm": 0.002110941568389535,
-      "learning_rate": 0.0001999916387242846,
-      "loss": 46.0,
-      "step": 25571
-    },
-    {
-      "epoch": 4.118201215829945,
-      "grad_norm": 0.006360162515193224,
-      "learning_rate": 0.00019999163807005957,
-      "loss": 46.0,
-      "step": 25572
-    },
-    {
-      "epoch": 4.1183622529087325,
-      "grad_norm": 0.0006603975198231637,
-      "learning_rate": 0.00019999163741580897,
-      "loss": 46.0,
-      "step": 25573
-    },
-    {
-      "epoch": 4.11852328998752,
-      "grad_norm": 0.007174971979111433,
-      "learning_rate": 0.00019999163676153277,
-      "loss": 46.0,
-      "step": 25574
-    },
-    {
-      "epoch": 4.118684327066307,
-      "grad_norm": 0.002847258932888508,
-      "learning_rate": 0.000199991636107231,
-      "loss": 46.0,
-      "step": 25575
-    },
-    {
-      "epoch": 4.118845364145095,
-      "grad_norm": 0.0034133384469896555,
-      "learning_rate": 0.00019999163545290362,
-      "loss": 46.0,
-      "step": 25576
-    },
-    {
-      "epoch": 4.119006401223881,
-      "grad_norm": 0.0021964781917631626,
-      "learning_rate": 0.00019999163479855064,
-      "loss": 46.0,
-      "step": 25577
-    },
-    {
-      "epoch": 4.119167438302669,
-      "grad_norm": 0.002158648334443569,
-      "learning_rate": 0.00019999163414417205,
-      "loss": 46.0,
-      "step": 25578
-    },
-    {
-      "epoch": 4.119328475381456,
-      "grad_norm": 0.002800311893224716,
-      "learning_rate": 0.00019999163348976792,
-      "loss": 46.0,
-      "step": 25579
-    },
-    {
-      "epoch": 4.119489512460244,
-      "grad_norm": 0.012722013518214226,
-      "learning_rate": 0.00019999163283533817,
-      "loss": 46.0,
-      "step": 25580
-    },
-    {
-      "epoch": 4.119650549539031,
-      "grad_norm": 0.013124719262123108,
-      "learning_rate": 0.00019999163218088281,
-      "loss": 46.0,
-      "step": 25581
-    },
-    {
-      "epoch": 4.119811586617819,
-      "grad_norm": 0.0015910544898360968,
-      "learning_rate": 0.00019999163152640187,
-      "loss": 46.0,
-      "step": 25582
-    },
-    {
-      "epoch": 4.119972623696606,
-      "grad_norm": 0.003991268575191498,
-      "learning_rate": 0.00019999163087189534,
-      "loss": 46.0,
-      "step": 25583
-    },
-    {
-      "epoch": 4.120133660775394,
-      "grad_norm": 0.005277678370475769,
-      "learning_rate": 0.00019999163021736322,
-      "loss": 46.0,
-      "step": 25584
-    },
-    {
-      "epoch": 4.120294697854181,
-      "grad_norm": 0.0029431104194372892,
-      "learning_rate": 0.00019999162956280554,
-      "loss": 46.0,
-      "step": 25585
-    },
-    {
-      "epoch": 4.1204557349329685,
-      "grad_norm": 0.006882273126393557,
-      "learning_rate": 0.00019999162890822222,
-      "loss": 46.0,
-      "step": 25586
-    },
-    {
-      "epoch": 4.120616772011756,
-      "grad_norm": 0.011436086148023605,
-      "learning_rate": 0.0001999916282536133,
-      "loss": 46.0,
-      "step": 25587
-    },
-    {
-      "epoch": 4.120777809090543,
-      "grad_norm": 0.005408705677837133,
-      "learning_rate": 0.00019999162759897884,
-      "loss": 46.0,
-      "step": 25588
-    },
-    {
-      "epoch": 4.120938846169331,
-      "grad_norm": 0.0021844564471393824,
-      "learning_rate": 0.00019999162694431873,
-      "loss": 46.0,
-      "step": 25589
-    },
-    {
-      "epoch": 4.121099883248118,
-      "grad_norm": 0.0036464945878833532,
-      "learning_rate": 0.00019999162628963303,
-      "loss": 46.0,
-      "step": 25590
-    },
-    {
-      "epoch": 4.121260920326906,
-      "grad_norm": 0.0017811524448916316,
-      "learning_rate": 0.00019999162563492178,
-      "loss": 46.0,
-      "step": 25591
-    },
-    {
-      "epoch": 4.121421957405692,
-      "grad_norm": 0.0027819517999887466,
-      "learning_rate": 0.0001999916249801849,
-      "loss": 46.0,
-      "step": 25592
-    },
-    {
-      "epoch": 4.12158299448448,
-      "grad_norm": 0.0021310080774128437,
-      "learning_rate": 0.00019999162432542247,
-      "loss": 46.0,
-      "step": 25593
-    },
-    {
-      "epoch": 4.121744031563267,
-      "grad_norm": 0.00754657294601202,
-      "learning_rate": 0.0001999916236706344,
-      "loss": 46.0,
-      "step": 25594
-    },
-    {
-      "epoch": 4.121905068642055,
-      "grad_norm": 0.002623029751703143,
-      "learning_rate": 0.00019999162301582074,
-      "loss": 46.0,
-      "step": 25595
-    },
-    {
-      "epoch": 4.122066105720842,
-      "grad_norm": 0.002840273082256317,
-      "learning_rate": 0.00019999162236098152,
-      "loss": 46.0,
-      "step": 25596
-    },
-    {
-      "epoch": 4.1222271427996295,
-      "grad_norm": 0.005038201343268156,
-      "learning_rate": 0.00019999162170611669,
-      "loss": 46.0,
-      "step": 25597
-    },
-    {
-      "epoch": 4.122388179878417,
-      "grad_norm": 0.016149956732988358,
-      "learning_rate": 0.00019999162105122626,
-      "loss": 46.0,
-      "step": 25598
-    },
-    {
-      "epoch": 4.122549216957204,
-      "grad_norm": 0.002412924077361822,
-      "learning_rate": 0.00019999162039631028,
-      "loss": 46.0,
-      "step": 25599
-    },
-    {
-      "epoch": 4.122710254035992,
-      "grad_norm": 0.0009178838809020817,
-      "learning_rate": 0.00019999161974136866,
-      "loss": 46.0,
-      "step": 25600
-    },
-    {
-      "epoch": 4.122871291114779,
-      "grad_norm": 0.0012937730643898249,
-      "learning_rate": 0.00019999161908640148,
-      "loss": 46.0,
-      "step": 25601
-    },
-    {
-      "epoch": 4.123032328193567,
-      "grad_norm": 0.0023727118968963623,
-      "learning_rate": 0.00019999161843140865,
-      "loss": 46.0,
-      "step": 25602
-    },
-    {
-      "epoch": 4.123193365272354,
-      "grad_norm": 0.0032714291010051966,
-      "learning_rate": 0.0001999916177763903,
-      "loss": 46.0,
-      "step": 25603
-    },
-    {
-      "epoch": 4.123354402351142,
-      "grad_norm": 0.00729818781837821,
-      "learning_rate": 0.0001999916171213463,
-      "loss": 46.0,
-      "step": 25604
-    },
-    {
-      "epoch": 4.123515439429929,
-      "grad_norm": 0.003525134874507785,
-      "learning_rate": 0.00019999161646627674,
-      "loss": 46.0,
-      "step": 25605
-    },
-    {
-      "epoch": 4.123676476508716,
-      "grad_norm": 0.005336896050721407,
-      "learning_rate": 0.00019999161581118157,
-      "loss": 46.0,
-      "step": 25606
-    },
-    {
-      "epoch": 4.123837513587503,
-      "grad_norm": 0.005608264822512865,
-      "learning_rate": 0.00019999161515606083,
-      "loss": 46.0,
-      "step": 25607
-    },
-    {
-      "epoch": 4.123998550666291,
-      "grad_norm": 0.001506253145635128,
-      "learning_rate": 0.00019999161450091446,
-      "loss": 46.0,
-      "step": 25608
-    },
-    {
-      "epoch": 4.124159587745078,
-      "grad_norm": 0.00699825631454587,
-      "learning_rate": 0.00019999161384574252,
-      "loss": 46.0,
-      "step": 25609
-    },
-    {
-      "epoch": 4.1243206248238655,
-      "grad_norm": 0.018045881763100624,
-      "learning_rate": 0.000199991613190545,
-      "loss": 46.0,
-      "step": 25610
-    },
-    {
-      "epoch": 4.124481661902653,
-      "grad_norm": 0.003668231889605522,
-      "learning_rate": 0.0001999916125353219,
-      "loss": 46.0,
-      "step": 25611
-    },
-    {
-      "epoch": 4.12464269898144,
-      "grad_norm": 0.00043198675848543644,
-      "learning_rate": 0.00019999161188007314,
-      "loss": 46.0,
-      "step": 25612
-    },
-    {
-      "epoch": 4.124803736060228,
-      "grad_norm": 0.0017844585236161947,
-      "learning_rate": 0.00019999161122479883,
-      "loss": 46.0,
-      "step": 25613
-    },
-    {
-      "epoch": 4.124964773139015,
-      "grad_norm": 0.007753214333206415,
-      "learning_rate": 0.00019999161056949893,
-      "loss": 46.0,
-      "step": 25614
-    },
-    {
-      "epoch": 4.125125810217803,
-      "grad_norm": 0.0009150505647994578,
-      "learning_rate": 0.00019999160991417342,
-      "loss": 46.0,
-      "step": 25615
-    },
-    {
-      "epoch": 4.12528684729659,
-      "grad_norm": 0.00484734121710062,
-      "learning_rate": 0.00019999160925882235,
-      "loss": 46.0,
-      "step": 25616
-    },
-    {
-      "epoch": 4.125447884375378,
-      "grad_norm": 0.0038678511045873165,
-      "learning_rate": 0.00019999160860344567,
-      "loss": 46.0,
-      "step": 25617
-    },
-    {
-      "epoch": 4.125608921454165,
-      "grad_norm": 0.003355953376740217,
-      "learning_rate": 0.0001999916079480434,
-      "loss": 46.0,
-      "step": 25618
-    },
-    {
-      "epoch": 4.125769958532953,
-      "grad_norm": 0.014977559447288513,
-      "learning_rate": 0.00019999160729261553,
-      "loss": 46.0,
-      "step": 25619
-    },
-    {
-      "epoch": 4.125930995611739,
-      "grad_norm": 0.0042245956137776375,
-      "learning_rate": 0.00019999160663716206,
-      "loss": 46.0,
-      "step": 25620
-    },
-    {
-      "epoch": 4.126092032690527,
-      "grad_norm": 0.002997392090037465,
-      "learning_rate": 0.000199991605981683,
-      "loss": 46.0,
-      "step": 25621
-    },
-    {
-      "epoch": 4.126253069769314,
-      "grad_norm": 0.009510383009910583,
-      "learning_rate": 0.00019999160532617835,
-      "loss": 46.0,
-      "step": 25622
-    },
-    {
-      "epoch": 4.1264141068481015,
-      "grad_norm": 0.0033083290327340364,
-      "learning_rate": 0.00019999160467064814,
-      "loss": 46.0,
-      "step": 25623
-    },
-    {
-      "epoch": 4.126575143926889,
-      "grad_norm": 0.006896724924445152,
-      "learning_rate": 0.00019999160401509229,
-      "loss": 46.0,
-      "step": 25624
-    },
-    {
-      "epoch": 4.126736181005676,
-      "grad_norm": 0.0031561029609292746,
-      "learning_rate": 0.00019999160335951085,
-      "loss": 46.0,
-      "step": 25625
-    },
-    {
-      "epoch": 4.126897218084464,
-      "grad_norm": 0.001359561923891306,
-      "learning_rate": 0.00019999160270390385,
-      "loss": 46.0,
-      "step": 25626
-    },
-    {
-      "epoch": 4.127058255163251,
-      "grad_norm": 0.008148791268467903,
-      "learning_rate": 0.00019999160204827124,
-      "loss": 46.0,
-      "step": 25627
-    },
-    {
-      "epoch": 4.127219292242039,
-      "grad_norm": 0.002030620351433754,
-      "learning_rate": 0.00019999160139261304,
-      "loss": 46.0,
-      "step": 25628
-    },
-    {
-      "epoch": 4.127380329320826,
-      "grad_norm": 0.009539092890918255,
-      "learning_rate": 0.00019999160073692925,
-      "loss": 46.0,
-      "step": 25629
-    },
-    {
-      "epoch": 4.127541366399614,
-      "grad_norm": 0.002409338718280196,
-      "learning_rate": 0.00019999160008121985,
-      "loss": 46.0,
-      "step": 25630
-    },
-    {
-      "epoch": 4.127702403478401,
-      "grad_norm": 0.0011922902194783092,
-      "learning_rate": 0.00019999159942548486,
-      "loss": 46.0,
-      "step": 25631
-    },
-    {
-      "epoch": 4.127863440557189,
-      "grad_norm": 0.003443419933319092,
-      "learning_rate": 0.0001999915987697243,
-      "loss": 46.0,
-      "step": 25632
-    },
-    {
-      "epoch": 4.128024477635976,
-      "grad_norm": 0.004027842544019222,
-      "learning_rate": 0.00019999159811393813,
-      "loss": 46.0,
-      "step": 25633
-    },
-    {
-      "epoch": 4.1281855147147635,
-      "grad_norm": 0.002964272629469633,
-      "learning_rate": 0.00019999159745812638,
-      "loss": 46.0,
-      "step": 25634
-    },
-    {
-      "epoch": 4.12834655179355,
-      "grad_norm": 0.00287226471118629,
-      "learning_rate": 0.00019999159680228901,
-      "loss": 46.0,
-      "step": 25635
-    },
-    {
-      "epoch": 4.1285075888723375,
-      "grad_norm": 0.0044133830815553665,
-      "learning_rate": 0.0001999915961464261,
-      "loss": 46.0,
-      "step": 25636
-    },
-    {
-      "epoch": 4.128668625951125,
-      "grad_norm": 0.0032337193842977285,
-      "learning_rate": 0.00019999159549053755,
-      "loss": 46.0,
-      "step": 25637
-    },
-    {
-      "epoch": 4.128829663029912,
-      "grad_norm": 0.007156878709793091,
-      "learning_rate": 0.00019999159483462343,
-      "loss": 46.0,
-      "step": 25638
-    },
-    {
-      "epoch": 4.1289907001087,
-      "grad_norm": 0.0004423867794685066,
-      "learning_rate": 0.00019999159417868366,
-      "loss": 46.0,
-      "step": 25639
-    },
-    {
-      "epoch": 4.129151737187487,
-      "grad_norm": 0.004037897102534771,
-      "learning_rate": 0.00019999159352271836,
-      "loss": 46.0,
-      "step": 25640
-    },
-    {
-      "epoch": 4.129312774266275,
-      "grad_norm": 0.0017991041531786323,
-      "learning_rate": 0.00019999159286672745,
-      "loss": 46.0,
-      "step": 25641
-    },
-    {
-      "epoch": 4.129473811345062,
-      "grad_norm": 0.007354862987995148,
-      "learning_rate": 0.00019999159221071095,
-      "loss": 46.0,
-      "step": 25642
-    },
-    {
-      "epoch": 4.12963484842385,
-      "grad_norm": 0.008874555118381977,
-      "learning_rate": 0.00019999159155466886,
-      "loss": 46.0,
-      "step": 25643
-    },
-    {
-      "epoch": 4.129795885502637,
-      "grad_norm": 0.001482005463913083,
-      "learning_rate": 0.0001999915908986012,
-      "loss": 46.0,
-      "step": 25644
-    },
-    {
-      "epoch": 4.1299569225814245,
-      "grad_norm": 0.005707482807338238,
-      "learning_rate": 0.00019999159024250792,
-      "loss": 46.0,
-      "step": 25645
-    },
-    {
-      "epoch": 4.130117959660212,
-      "grad_norm": 0.004640547558665276,
-      "learning_rate": 0.00019999158958638905,
-      "loss": 46.0,
-      "step": 25646
-    },
-    {
-      "epoch": 4.1302789967389995,
-      "grad_norm": 0.002107742242515087,
-      "learning_rate": 0.00019999158893024455,
-      "loss": 46.0,
-      "step": 25647
-    },
-    {
-      "epoch": 4.130440033817787,
-      "grad_norm": 0.006413899827748537,
-      "learning_rate": 0.0001999915882740745,
-      "loss": 46.0,
-      "step": 25648
-    },
-    {
-      "epoch": 4.130601070896574,
-      "grad_norm": 0.00154272117651999,
-      "learning_rate": 0.00019999158761787887,
-      "loss": 46.0,
-      "step": 25649
-    },
-    {
-      "epoch": 4.130762107975361,
-      "grad_norm": 0.002160134259611368,
-      "learning_rate": 0.00019999158696165764,
-      "loss": 46.0,
-      "step": 25650
-    },
-    {
-      "epoch": 4.130923145054148,
-      "grad_norm": 0.0016933969454839826,
-      "learning_rate": 0.0001999915863054108,
-      "loss": 46.0,
-      "step": 25651
-    },
-    {
-      "epoch": 4.131084182132936,
-      "grad_norm": 0.002816577907651663,
-      "learning_rate": 0.00019999158564913835,
-      "loss": 46.0,
-      "step": 25652
-    },
-    {
-      "epoch": 4.131245219211723,
-      "grad_norm": 0.0024627463426440954,
-      "learning_rate": 0.00019999158499284033,
-      "loss": 46.0,
-      "step": 25653
-    },
-    {
-      "epoch": 4.131406256290511,
-      "grad_norm": 0.0020824482198804617,
-      "learning_rate": 0.00019999158433651673,
-      "loss": 46.0,
-      "step": 25654
-    },
-    {
-      "epoch": 4.131567293369298,
-      "grad_norm": 0.00415445351973176,
-      "learning_rate": 0.00019999158368016752,
-      "loss": 46.0,
-      "step": 25655
-    },
-    {
-      "epoch": 4.131728330448086,
-      "grad_norm": 0.0021817239467054605,
-      "learning_rate": 0.0001999915830237927,
-      "loss": 46.0,
-      "step": 25656
-    },
-    {
-      "epoch": 4.131889367526873,
-      "grad_norm": 0.002095539588481188,
-      "learning_rate": 0.00019999158236739232,
-      "loss": 46.0,
-      "step": 25657
-    },
-    {
-      "epoch": 4.1320504046056605,
-      "grad_norm": 0.006299429573118687,
-      "learning_rate": 0.00019999158171096635,
-      "loss": 46.0,
-      "step": 25658
-    },
-    {
-      "epoch": 4.132211441684448,
-      "grad_norm": 0.0012190460693091154,
-      "learning_rate": 0.00019999158105451476,
-      "loss": 46.0,
-      "step": 25659
-    },
-    {
-      "epoch": 4.132372478763235,
-      "grad_norm": 0.0038476467598229647,
-      "learning_rate": 0.0001999915803980376,
-      "loss": 46.0,
-      "step": 25660
-    },
-    {
-      "epoch": 4.132533515842023,
-      "grad_norm": 0.004012128803879023,
-      "learning_rate": 0.00019999157974153484,
-      "loss": 46.0,
-      "step": 25661
-    },
-    {
-      "epoch": 4.13269455292081,
-      "grad_norm": 0.001898324117064476,
-      "learning_rate": 0.00019999157908500649,
-      "loss": 46.0,
-      "step": 25662
-    },
-    {
-      "epoch": 4.132855589999598,
-      "grad_norm": 0.002790663857012987,
-      "learning_rate": 0.00019999157842845252,
-      "loss": 46.0,
-      "step": 25663
-    },
-    {
-      "epoch": 4.133016627078385,
-      "grad_norm": 0.007841236889362335,
-      "learning_rate": 0.00019999157777187302,
-      "loss": 46.0,
-      "step": 25664
-    },
-    {
-      "epoch": 4.133177664157172,
-      "grad_norm": 0.010113999247550964,
-      "learning_rate": 0.00019999157711526785,
-      "loss": 46.0,
-      "step": 25665
-    },
-    {
-      "epoch": 4.133338701235959,
-      "grad_norm": 0.010194149799644947,
-      "learning_rate": 0.00019999157645863712,
-      "loss": 46.0,
-      "step": 25666
-    },
-    {
-      "epoch": 4.133499738314747,
-      "grad_norm": 0.0036781923845410347,
-      "learning_rate": 0.00019999157580198083,
-      "loss": 46.0,
-      "step": 25667
-    },
-    {
-      "epoch": 4.133660775393534,
-      "grad_norm": 0.005262341815978289,
-      "learning_rate": 0.0001999915751452989,
-      "loss": 46.0,
-      "step": 25668
-    },
-    {
-      "epoch": 4.133821812472322,
-      "grad_norm": 0.00517252879217267,
-      "learning_rate": 0.0001999915744885914,
-      "loss": 46.0,
-      "step": 25669
-    },
-    {
-      "epoch": 4.133982849551109,
-      "grad_norm": 0.00330323725938797,
-      "learning_rate": 0.00019999157383185828,
-      "loss": 46.0,
-      "step": 25670
-    },
-    {
-      "epoch": 4.1341438866298965,
-      "grad_norm": 0.0017552450299263,
-      "learning_rate": 0.00019999157317509962,
-      "loss": 46.0,
-      "step": 25671
-    },
-    {
-      "epoch": 4.134304923708684,
-      "grad_norm": 0.005632500629872084,
-      "learning_rate": 0.00019999157251831534,
-      "loss": 46.0,
-      "step": 25672
-    },
-    {
-      "epoch": 4.134465960787471,
-      "grad_norm": 0.004503052681684494,
-      "learning_rate": 0.00019999157186150545,
-      "loss": 46.0,
-      "step": 25673
-    },
-    {
-      "epoch": 4.134626997866259,
-      "grad_norm": 0.01812112331390381,
-      "learning_rate": 0.00019999157120467,
-      "loss": 46.0,
-      "step": 25674
-    },
-    {
-      "epoch": 4.134788034945046,
-      "grad_norm": 0.0025516224559396505,
-      "learning_rate": 0.00019999157054780895,
-      "loss": 46.0,
-      "step": 25675
-    },
-    {
-      "epoch": 4.134949072023834,
-      "grad_norm": 0.004877248778939247,
-      "learning_rate": 0.00019999156989092227,
-      "loss": 46.0,
-      "step": 25676
-    },
-    {
-      "epoch": 4.135110109102621,
-      "grad_norm": 0.0030104005709290504,
-      "learning_rate": 0.00019999156923401003,
-      "loss": 46.0,
-      "step": 25677
-    },
-    {
-      "epoch": 4.135271146181409,
-      "grad_norm": 0.0052670701406896114,
-      "learning_rate": 0.0001999915685770722,
-      "loss": 46.0,
-      "step": 25678
-    },
-    {
-      "epoch": 4.135432183260195,
-      "grad_norm": 0.004815518390387297,
-      "learning_rate": 0.00019999156792010876,
-      "loss": 46.0,
-      "step": 25679
-    },
-    {
-      "epoch": 4.135593220338983,
-      "grad_norm": 0.011307308450341225,
-      "learning_rate": 0.00019999156726311973,
-      "loss": 46.0,
-      "step": 25680
-    },
-    {
-      "epoch": 4.13575425741777,
-      "grad_norm": 0.0015974113484844565,
-      "learning_rate": 0.00019999156660610514,
-      "loss": 46.0,
-      "step": 25681
-    },
-    {
-      "epoch": 4.135915294496558,
-      "grad_norm": 0.0063629476353526115,
-      "learning_rate": 0.0001999915659490649,
-      "loss": 46.0,
-      "step": 25682
-    },
-    {
-      "epoch": 4.136076331575345,
-      "grad_norm": 0.006022559013217688,
-      "learning_rate": 0.0001999915652919991,
-      "loss": 46.0,
-      "step": 25683
-    },
-    {
-      "epoch": 4.1362373686541325,
-      "grad_norm": 0.0039491672068834305,
-      "learning_rate": 0.0001999915646349077,
-      "loss": 46.0,
-      "step": 25684
-    },
-    {
-      "epoch": 4.13639840573292,
-      "grad_norm": 0.004771003033965826,
-      "learning_rate": 0.00019999156397779074,
-      "loss": 46.0,
-      "step": 25685
-    },
-    {
-      "epoch": 4.136559442811707,
-      "grad_norm": 0.0039065321907401085,
-      "learning_rate": 0.00019999156332064816,
-      "loss": 46.0,
-      "step": 25686
-    },
-    {
-      "epoch": 4.136720479890495,
-      "grad_norm": 0.0047430070117115974,
-      "learning_rate": 0.00019999156266348,
-      "loss": 46.0,
-      "step": 25687
-    },
-    {
-      "epoch": 4.136881516969282,
-      "grad_norm": 0.00212700292468071,
-      "learning_rate": 0.0001999915620062862,
-      "loss": 46.0,
-      "step": 25688
-    },
-    {
-      "epoch": 4.13704255404807,
-      "grad_norm": 0.001397873042151332,
-      "learning_rate": 0.00019999156134906687,
-      "loss": 46.0,
-      "step": 25689
-    },
-    {
-      "epoch": 4.137203591126857,
-      "grad_norm": 0.000960984849371016,
-      "learning_rate": 0.0001999915606918219,
-      "loss": 46.0,
-      "step": 25690
-    },
-    {
-      "epoch": 4.137364628205645,
-      "grad_norm": 0.0022925168741494417,
-      "learning_rate": 0.00019999156003455134,
-      "loss": 46.0,
-      "step": 25691
-    },
-    {
-      "epoch": 4.137525665284432,
-      "grad_norm": 0.010183616541326046,
-      "learning_rate": 0.0001999915593772552,
-      "loss": 46.0,
-      "step": 25692
-    },
-    {
-      "epoch": 4.13768670236322,
-      "grad_norm": 0.009615905582904816,
-      "learning_rate": 0.00019999155871993352,
-      "loss": 46.0,
-      "step": 25693
-    },
-    {
-      "epoch": 4.137847739442006,
-      "grad_norm": 0.0008418922079727054,
-      "learning_rate": 0.0001999915580625862,
-      "loss": 46.0,
-      "step": 25694
-    },
-    {
-      "epoch": 4.138008776520794,
-      "grad_norm": 0.003879308234900236,
-      "learning_rate": 0.00019999155740521327,
-      "loss": 46.0,
-      "step": 25695
-    },
-    {
-      "epoch": 4.138169813599581,
-      "grad_norm": 0.003745253663510084,
-      "learning_rate": 0.00019999155674781476,
-      "loss": 46.0,
-      "step": 25696
-    },
-    {
-      "epoch": 4.1383308506783685,
-      "grad_norm": 0.002811028389260173,
-      "learning_rate": 0.00019999155609039067,
-      "loss": 46.0,
-      "step": 25697
-    },
-    {
-      "epoch": 4.138491887757156,
-      "grad_norm": 0.014671149663627148,
-      "learning_rate": 0.00019999155543294096,
-      "loss": 46.0,
-      "step": 25698
-    },
-    {
-      "epoch": 4.138652924835943,
-      "grad_norm": 0.003004190046340227,
-      "learning_rate": 0.00019999155477546567,
-      "loss": 46.0,
-      "step": 25699
-    },
-    {
-      "epoch": 4.138813961914731,
-      "grad_norm": 0.009994974359869957,
-      "learning_rate": 0.0001999915541179648,
-      "loss": 46.0,
-      "step": 25700
-    },
-    {
-      "epoch": 4.138974998993518,
-      "grad_norm": 0.003434967016801238,
-      "learning_rate": 0.00019999155346043832,
-      "loss": 46.0,
-      "step": 25701
-    },
-    {
-      "epoch": 4.139136036072306,
-      "grad_norm": 0.004942676518112421,
-      "learning_rate": 0.00019999155280288626,
-      "loss": 46.0,
-      "step": 25702
-    },
-    {
-      "epoch": 4.139297073151093,
-      "grad_norm": 0.0008012540056370199,
-      "learning_rate": 0.00019999155214530862,
-      "loss": 46.0,
-      "step": 25703
-    },
-    {
-      "epoch": 4.139458110229881,
-      "grad_norm": 0.007842950522899628,
-      "learning_rate": 0.00019999155148770536,
-      "loss": 46.0,
-      "step": 25704
-    },
-    {
-      "epoch": 4.139619147308668,
-      "grad_norm": 0.0017749844118952751,
-      "learning_rate": 0.00019999155083007651,
-      "loss": 46.0,
-      "step": 25705
-    },
-    {
-      "epoch": 4.1397801843874555,
-      "grad_norm": 0.010829790495336056,
-      "learning_rate": 0.00019999155017242208,
-      "loss": 46.0,
-      "step": 25706
-    },
-    {
-      "epoch": 4.139941221466243,
-      "grad_norm": 0.001926140277646482,
-      "learning_rate": 0.00019999154951474206,
-      "loss": 46.0,
-      "step": 25707
-    },
-    {
-      "epoch": 4.14010225854503,
-      "grad_norm": 0.001501739607192576,
-      "learning_rate": 0.00019999154885703643,
-      "loss": 46.0,
-      "step": 25708
-    },
-    {
-      "epoch": 4.140263295623817,
-      "grad_norm": 0.0071657029911875725,
-      "learning_rate": 0.00019999154819930524,
-      "loss": 46.0,
-      "step": 25709
-    },
-    {
-      "epoch": 4.1404243327026045,
-      "grad_norm": 0.0005854924093000591,
-      "learning_rate": 0.0001999915475415484,
-      "loss": 46.0,
-      "step": 25710
-    },
-    {
-      "epoch": 4.140585369781392,
-      "grad_norm": 0.008494430221617222,
-      "learning_rate": 0.00019999154688376603,
-      "loss": 46.0,
-      "step": 25711
-    },
-    {
-      "epoch": 4.140746406860179,
-      "grad_norm": 0.005708741489797831,
-      "learning_rate": 0.00019999154622595805,
-      "loss": 46.0,
-      "step": 25712
-    },
-    {
-      "epoch": 4.140907443938967,
-      "grad_norm": 0.001477561891078949,
-      "learning_rate": 0.00019999154556812445,
-      "loss": 46.0,
-      "step": 25713
-    },
-    {
-      "epoch": 4.141068481017754,
-      "grad_norm": 0.003391193225979805,
-      "learning_rate": 0.00019999154491026527,
-      "loss": 46.0,
-      "step": 25714
-    },
-    {
-      "epoch": 4.141229518096542,
-      "grad_norm": 0.010698160156607628,
-      "learning_rate": 0.0001999915442523805,
-      "loss": 46.0,
-      "step": 25715
-    },
-    {
-      "epoch": 4.141390555175329,
-      "grad_norm": 0.007413472048938274,
-      "learning_rate": 0.00019999154359447014,
-      "loss": 46.0,
-      "step": 25716
-    },
-    {
-      "epoch": 4.141551592254117,
-      "grad_norm": 0.005606756079941988,
-      "learning_rate": 0.0001999915429365342,
-      "loss": 46.0,
-      "step": 25717
-    },
-    {
-      "epoch": 4.141712629332904,
-      "grad_norm": 0.0029826743993908167,
-      "learning_rate": 0.00019999154227857264,
-      "loss": 46.0,
-      "step": 25718
-    },
-    {
-      "epoch": 4.1418736664116915,
-      "grad_norm": 0.0037592726293951273,
-      "learning_rate": 0.00019999154162058552,
-      "loss": 46.0,
-      "step": 25719
-    },
-    {
-      "epoch": 4.142034703490479,
-      "grad_norm": 0.001650240388698876,
-      "learning_rate": 0.00019999154096257278,
-      "loss": 46.0,
-      "step": 25720
-    },
-    {
-      "epoch": 4.142195740569266,
-      "grad_norm": 0.0015117531875148416,
-      "learning_rate": 0.00019999154030453443,
-      "loss": 46.0,
-      "step": 25721
-    },
-    {
-      "epoch": 4.142356777648054,
-      "grad_norm": 0.007537306286394596,
-      "learning_rate": 0.00019999153964647053,
-      "loss": 46.0,
-      "step": 25722
-    },
-    {
-      "epoch": 4.14251781472684,
-      "grad_norm": 0.007101949770003557,
-      "learning_rate": 0.00019999153898838103,
-      "loss": 46.0,
-      "step": 25723
-    },
-    {
-      "epoch": 4.142678851805628,
-      "grad_norm": 0.0075798360630869865,
-      "learning_rate": 0.00019999153833026592,
-      "loss": 46.0,
-      "step": 25724
-    },
-    {
-      "epoch": 4.142839888884415,
-      "grad_norm": 0.00363737135194242,
-      "learning_rate": 0.00019999153767212522,
-      "loss": 46.0,
-      "step": 25725
-    },
-    {
-      "epoch": 4.143000925963203,
-      "grad_norm": 0.0017418122151866555,
-      "learning_rate": 0.00019999153701395894,
-      "loss": 46.0,
-      "step": 25726
-    },
-    {
-      "epoch": 4.14316196304199,
-      "grad_norm": 0.004063683561980724,
-      "learning_rate": 0.00019999153635576704,
-      "loss": 46.0,
-      "step": 25727
-    },
-    {
-      "epoch": 4.143323000120778,
-      "grad_norm": 0.0010686945170164108,
-      "learning_rate": 0.00019999153569754958,
-      "loss": 46.0,
-      "step": 25728
-    },
-    {
-      "epoch": 4.143484037199565,
-      "grad_norm": 0.0055069816298782825,
-      "learning_rate": 0.0001999915350393065,
-      "loss": 46.0,
-      "step": 25729
-    },
-    {
-      "epoch": 4.143645074278353,
-      "grad_norm": 0.005126043688505888,
-      "learning_rate": 0.00019999153438103785,
-      "loss": 46.0,
-      "step": 25730
-    },
-    {
-      "epoch": 4.14380611135714,
-      "grad_norm": 0.004279494751244783,
-      "learning_rate": 0.0001999915337227436,
-      "loss": 46.0,
-      "step": 25731
-    },
-    {
-      "epoch": 4.1439671484359275,
-      "grad_norm": 0.00860437098890543,
-      "learning_rate": 0.00019999153306442374,
-      "loss": 46.0,
-      "step": 25732
-    },
-    {
-      "epoch": 4.144128185514715,
-      "grad_norm": 0.0013098431518301368,
-      "learning_rate": 0.00019999153240607832,
-      "loss": 46.0,
-      "step": 25733
-    },
-    {
-      "epoch": 4.144289222593502,
-      "grad_norm": 0.014905581250786781,
-      "learning_rate": 0.00019999153174770728,
-      "loss": 46.0,
-      "step": 25734
-    },
-    {
-      "epoch": 4.14445025967229,
-      "grad_norm": 0.004219368100166321,
-      "learning_rate": 0.00019999153108931066,
-      "loss": 46.0,
-      "step": 25735
-    },
-    {
-      "epoch": 4.144611296751077,
-      "grad_norm": 0.0006310707540251315,
-      "learning_rate": 0.00019999153043088845,
-      "loss": 46.0,
-      "step": 25736
-    },
-    {
-      "epoch": 4.144772333829865,
-      "grad_norm": 0.0008926604641601443,
-      "learning_rate": 0.00019999152977244065,
-      "loss": 46.0,
-      "step": 25737
-    },
-    {
-      "epoch": 4.144933370908651,
-      "grad_norm": 0.005972984712570906,
-      "learning_rate": 0.00019999152911396727,
-      "loss": 46.0,
-      "step": 25738
-    },
-    {
-      "epoch": 4.145094407987439,
-      "grad_norm": 0.0018255076138302684,
-      "learning_rate": 0.00019999152845546827,
-      "loss": 46.0,
-      "step": 25739
-    },
-    {
-      "epoch": 4.145255445066226,
-      "grad_norm": 0.001713650650344789,
-      "learning_rate": 0.00019999152779694366,
-      "loss": 46.0,
-      "step": 25740
-    },
-    {
-      "epoch": 4.145416482145014,
-      "grad_norm": 0.0027448616456240416,
-      "learning_rate": 0.00019999152713839349,
-      "loss": 46.0,
-      "step": 25741
-    },
-    {
-      "epoch": 4.145577519223801,
-      "grad_norm": 0.0012780750403180718,
-      "learning_rate": 0.00019999152647981773,
-      "loss": 46.0,
-      "step": 25742
-    },
-    {
-      "epoch": 4.145738556302589,
-      "grad_norm": 0.0037260556127876043,
-      "learning_rate": 0.00019999152582121635,
-      "loss": 46.0,
-      "step": 25743
-    },
-    {
-      "epoch": 4.145899593381376,
-      "grad_norm": 0.0010752436937764287,
-      "learning_rate": 0.00019999152516258942,
-      "loss": 46.0,
-      "step": 25744
-    },
-    {
-      "epoch": 4.1460606304601635,
-      "grad_norm": 0.003949676640331745,
-      "learning_rate": 0.00019999152450393687,
-      "loss": 46.0,
-      "step": 25745
-    },
-    {
-      "epoch": 4.146221667538951,
-      "grad_norm": 0.004104393068701029,
-      "learning_rate": 0.00019999152384525873,
-      "loss": 46.0,
-      "step": 25746
-    },
-    {
-      "epoch": 4.146382704617738,
-      "grad_norm": 0.006399809382855892,
-      "learning_rate": 0.00019999152318655496,
-      "loss": 46.0,
-      "step": 25747
-    },
-    {
-      "epoch": 4.146543741696526,
-      "grad_norm": 0.0018477010307833552,
-      "learning_rate": 0.00019999152252782564,
-      "loss": 46.0,
-      "step": 25748
-    },
-    {
-      "epoch": 4.146704778775313,
-      "grad_norm": 0.010605844669044018,
-      "learning_rate": 0.00019999152186907072,
-      "loss": 46.0,
-      "step": 25749
-    },
-    {
-      "epoch": 4.146865815854101,
-      "grad_norm": 0.005574748385697603,
-      "learning_rate": 0.0001999915212102902,
-      "loss": 46.0,
-      "step": 25750
-    },
-    {
-      "epoch": 4.147026852932888,
-      "grad_norm": 0.0009573210845701396,
-      "learning_rate": 0.00019999152055148408,
-      "loss": 46.0,
-      "step": 25751
-    },
-    {
-      "epoch": 4.147187890011675,
-      "grad_norm": 0.005036060698330402,
-      "learning_rate": 0.0001999915198926524,
-      "loss": 46.0,
-      "step": 25752
-    },
-    {
-      "epoch": 4.147348927090462,
-      "grad_norm": 0.005134321749210358,
-      "learning_rate": 0.00019999151923379512,
-      "loss": 46.0,
-      "step": 25753
-    },
-    {
-      "epoch": 4.14750996416925,
-      "grad_norm": 0.0035497955977916718,
-      "learning_rate": 0.00019999151857491224,
-      "loss": 46.0,
-      "step": 25754
-    },
-    {
-      "epoch": 4.147671001248037,
-      "grad_norm": 0.0021118242293596268,
-      "learning_rate": 0.00019999151791600376,
-      "loss": 46.0,
-      "step": 25755
-    },
-    {
-      "epoch": 4.147832038326825,
-      "grad_norm": 0.003920841496437788,
-      "learning_rate": 0.00019999151725706967,
-      "loss": 46.0,
-      "step": 25756
-    },
-    {
-      "epoch": 4.147993075405612,
-      "grad_norm": 0.003977254964411259,
-      "learning_rate": 0.00019999151659811002,
-      "loss": 46.0,
-      "step": 25757
-    },
-    {
-      "epoch": 4.1481541124843995,
-      "grad_norm": 0.0011857778299599886,
-      "learning_rate": 0.0001999915159391248,
-      "loss": 46.0,
-      "step": 25758
-    },
-    {
-      "epoch": 4.148315149563187,
-      "grad_norm": 0.004219596274197102,
-      "learning_rate": 0.0001999915152801139,
-      "loss": 46.0,
-      "step": 25759
-    },
-    {
-      "epoch": 4.148476186641974,
-      "grad_norm": 0.003712916513904929,
-      "learning_rate": 0.00019999151462107747,
-      "loss": 46.0,
-      "step": 25760
-    },
-    {
-      "epoch": 4.148637223720762,
-      "grad_norm": 0.01058608666062355,
-      "learning_rate": 0.00019999151396201545,
-      "loss": 46.0,
-      "step": 25761
-    },
-    {
-      "epoch": 4.148798260799549,
-      "grad_norm": 0.0012363122077658772,
-      "learning_rate": 0.00019999151330292784,
-      "loss": 46.0,
-      "step": 25762
-    },
-    {
-      "epoch": 4.148959297878337,
-      "grad_norm": 0.0029525563586503267,
-      "learning_rate": 0.00019999151264381458,
-      "loss": 46.0,
-      "step": 25763
-    },
-    {
-      "epoch": 4.149120334957124,
-      "grad_norm": 0.007817268371582031,
-      "learning_rate": 0.0001999915119846758,
-      "loss": 46.0,
-      "step": 25764
-    },
-    {
-      "epoch": 4.149281372035912,
-      "grad_norm": 0.00783733930438757,
-      "learning_rate": 0.0001999915113255114,
-      "loss": 46.0,
-      "step": 25765
-    },
-    {
-      "epoch": 4.149442409114699,
-      "grad_norm": 0.004292396828532219,
-      "learning_rate": 0.0001999915106663214,
-      "loss": 46.0,
-      "step": 25766
-    },
-    {
-      "epoch": 4.149603446193486,
-      "grad_norm": 0.005185423418879509,
-      "learning_rate": 0.0001999915100071058,
-      "loss": 46.0,
-      "step": 25767
-    },
-    {
-      "epoch": 4.149764483272273,
-      "grad_norm": 0.0025544508825987577,
-      "learning_rate": 0.00019999150934786464,
-      "loss": 46.0,
-      "step": 25768
-    },
-    {
-      "epoch": 4.1499255203510605,
-      "grad_norm": 0.019969942048192024,
-      "learning_rate": 0.00019999150868859787,
-      "loss": 46.0,
-      "step": 25769
-    },
-    {
-      "epoch": 4.150086557429848,
-      "grad_norm": 0.004226475954055786,
-      "learning_rate": 0.00019999150802930548,
-      "loss": 46.0,
-      "step": 25770
-    },
-    {
-      "epoch": 4.1502475945086355,
-      "grad_norm": 0.003954090178012848,
-      "learning_rate": 0.00019999150736998753,
-      "loss": 46.0,
-      "step": 25771
-    },
-    {
-      "epoch": 4.150408631587423,
-      "grad_norm": 0.011470174416899681,
-      "learning_rate": 0.00019999150671064396,
-      "loss": 46.0,
-      "step": 25772
-    },
-    {
-      "epoch": 4.15056966866621,
-      "grad_norm": 0.0014393340097740293,
-      "learning_rate": 0.00019999150605127484,
-      "loss": 46.0,
-      "step": 25773
-    },
-    {
-      "epoch": 4.150730705744998,
-      "grad_norm": 0.021243076771497726,
-      "learning_rate": 0.00019999150539188007,
-      "loss": 46.0,
-      "step": 25774
-    },
-    {
-      "epoch": 4.150891742823785,
-      "grad_norm": 0.011256651021540165,
-      "learning_rate": 0.00019999150473245974,
-      "loss": 46.0,
-      "step": 25775
-    },
-    {
-      "epoch": 4.151052779902573,
-      "grad_norm": 0.011080256663262844,
-      "learning_rate": 0.00019999150407301383,
-      "loss": 46.0,
-      "step": 25776
-    },
-    {
-      "epoch": 4.15121381698136,
-      "grad_norm": 0.004300955682992935,
-      "learning_rate": 0.0001999915034135423,
-      "loss": 46.0,
-      "step": 25777
-    },
-    {
-      "epoch": 4.151374854060148,
-      "grad_norm": 0.002081675687804818,
-      "learning_rate": 0.0001999915027540452,
-      "loss": 46.0,
-      "step": 25778
-    },
-    {
-      "epoch": 4.151535891138935,
-      "grad_norm": 0.002691699657589197,
-      "learning_rate": 0.0001999915020945225,
-      "loss": 46.0,
-      "step": 25779
-    },
-    {
-      "epoch": 4.1516969282177225,
-      "grad_norm": 0.0034728620667010546,
-      "learning_rate": 0.0001999915014349742,
-      "loss": 46.0,
-      "step": 25780
-    },
-    {
-      "epoch": 4.151857965296509,
-      "grad_norm": 0.004659638274461031,
-      "learning_rate": 0.00019999150077540034,
-      "loss": 46.0,
-      "step": 25781
-    },
-    {
-      "epoch": 4.1520190023752965,
-      "grad_norm": 0.0035389536060392857,
-      "learning_rate": 0.00019999150011580085,
-      "loss": 46.0,
-      "step": 25782
-    },
-    {
-      "epoch": 4.152180039454084,
-      "grad_norm": 0.003957580309361219,
-      "learning_rate": 0.0001999914994561758,
-      "loss": 46.0,
-      "step": 25783
-    },
-    {
-      "epoch": 4.152341076532871,
-      "grad_norm": 0.00241960771381855,
-      "learning_rate": 0.0001999914987965251,
-      "loss": 46.0,
-      "step": 25784
-    },
-    {
-      "epoch": 4.152502113611659,
-      "grad_norm": 0.008114400319755077,
-      "learning_rate": 0.00019999149813684886,
-      "loss": 46.0,
-      "step": 25785
-    },
-    {
-      "epoch": 4.152663150690446,
-      "grad_norm": 0.0025531696155667305,
-      "learning_rate": 0.00019999149747714702,
-      "loss": 46.0,
-      "step": 25786
-    },
-    {
-      "epoch": 4.152824187769234,
-      "grad_norm": 0.003837055992335081,
-      "learning_rate": 0.00019999149681741956,
-      "loss": 46.0,
-      "step": 25787
-    },
-    {
-      "epoch": 4.152985224848021,
-      "grad_norm": 0.00905930157750845,
-      "learning_rate": 0.00019999149615766652,
-      "loss": 46.0,
-      "step": 25788
-    },
-    {
-      "epoch": 4.153146261926809,
-      "grad_norm": 0.00726577453315258,
-      "learning_rate": 0.00019999149549788792,
-      "loss": 46.0,
-      "step": 25789
-    },
-    {
-      "epoch": 4.153307299005596,
-      "grad_norm": 0.002783138770610094,
-      "learning_rate": 0.00019999149483808368,
-      "loss": 46.0,
-      "step": 25790
-    },
-    {
-      "epoch": 4.153468336084384,
-      "grad_norm": 0.001132084522396326,
-      "learning_rate": 0.00019999149417825388,
-      "loss": 46.0,
-      "step": 25791
-    },
-    {
-      "epoch": 4.153629373163171,
-      "grad_norm": 0.0028803006280213594,
-      "learning_rate": 0.00019999149351839846,
-      "loss": 46.0,
-      "step": 25792
-    },
-    {
-      "epoch": 4.1537904102419585,
-      "grad_norm": 0.003320316318422556,
-      "learning_rate": 0.00019999149285851746,
-      "loss": 46.0,
-      "step": 25793
-    },
-    {
-      "epoch": 4.153951447320746,
-      "grad_norm": 0.011789404787123203,
-      "learning_rate": 0.00019999149219861087,
-      "loss": 46.0,
-      "step": 25794
-    },
-    {
-      "epoch": 4.154112484399533,
-      "grad_norm": 0.0020085216965526342,
-      "learning_rate": 0.0001999914915386787,
-      "loss": 46.0,
-      "step": 25795
-    },
-    {
-      "epoch": 4.15427352147832,
-      "grad_norm": 0.02010505087673664,
-      "learning_rate": 0.00019999149087872092,
-      "loss": 46.0,
-      "step": 25796
-    },
-    {
-      "epoch": 4.154434558557107,
-      "grad_norm": 0.0045603010803461075,
-      "learning_rate": 0.00019999149021873754,
-      "loss": 46.0,
-      "step": 25797
-    },
-    {
-      "epoch": 4.154595595635895,
-      "grad_norm": 0.0024973773397505283,
-      "learning_rate": 0.00019999148955872857,
-      "loss": 46.0,
-      "step": 25798
-    },
-    {
-      "epoch": 4.154756632714682,
-      "grad_norm": 0.0042458027601242065,
-      "learning_rate": 0.00019999148889869402,
-      "loss": 46.0,
-      "step": 25799
-    },
-    {
-      "epoch": 4.15491766979347,
-      "grad_norm": 0.0034933092538267374,
-      "learning_rate": 0.00019999148823863388,
-      "loss": 46.0,
-      "step": 25800
-    },
-    {
-      "epoch": 4.155078706872257,
-      "grad_norm": 0.005349744576960802,
-      "learning_rate": 0.00019999148757854812,
-      "loss": 46.0,
-      "step": 25801
-    },
-    {
-      "epoch": 4.155239743951045,
-      "grad_norm": 0.006178024224936962,
-      "learning_rate": 0.0001999914869184368,
-      "loss": 46.0,
-      "step": 25802
-    },
-    {
-      "epoch": 4.155400781029832,
-      "grad_norm": 0.00198973435908556,
-      "learning_rate": 0.00019999148625829988,
-      "loss": 46.0,
-      "step": 25803
-    },
-    {
-      "epoch": 4.15556181810862,
-      "grad_norm": 0.0013105213874951005,
-      "learning_rate": 0.00019999148559813733,
-      "loss": 46.0,
-      "step": 25804
-    },
-    {
-      "epoch": 4.155722855187407,
-      "grad_norm": 0.0008887567091733217,
-      "learning_rate": 0.00019999148493794926,
-      "loss": 46.0,
-      "step": 25805
-    },
-    {
-      "epoch": 4.1558838922661945,
-      "grad_norm": 0.00624180119484663,
-      "learning_rate": 0.00019999148427773556,
-      "loss": 46.0,
-      "step": 25806
-    },
-    {
-      "epoch": 4.156044929344982,
-      "grad_norm": 0.004844765178859234,
-      "learning_rate": 0.00019999148361749623,
-      "loss": 46.0,
-      "step": 25807
-    },
-    {
-      "epoch": 4.156205966423769,
-      "grad_norm": 0.006591136567294598,
-      "learning_rate": 0.00019999148295723137,
-      "loss": 46.0,
-      "step": 25808
-    },
-    {
-      "epoch": 4.156367003502557,
-      "grad_norm": 0.004222767893224955,
-      "learning_rate": 0.00019999148229694086,
-      "loss": 46.0,
-      "step": 25809
-    },
-    {
-      "epoch": 4.156528040581344,
-      "grad_norm": 0.0019136391347274184,
-      "learning_rate": 0.0001999914816366248,
-      "loss": 46.0,
-      "step": 25810
-    },
-    {
-      "epoch": 4.156689077660131,
-      "grad_norm": 0.001856248127296567,
-      "learning_rate": 0.00019999148097628314,
-      "loss": 46.0,
-      "step": 25811
-    },
-    {
-      "epoch": 4.156850114738918,
-      "grad_norm": 0.008216975256800652,
-      "learning_rate": 0.00019999148031591587,
-      "loss": 46.0,
-      "step": 25812
-    },
-    {
-      "epoch": 4.157011151817706,
-      "grad_norm": 0.006173674017190933,
-      "learning_rate": 0.00019999147965552304,
-      "loss": 46.0,
-      "step": 25813
-    },
-    {
-      "epoch": 4.157172188896493,
-      "grad_norm": 0.003741530468687415,
-      "learning_rate": 0.00019999147899510457,
-      "loss": 46.0,
-      "step": 25814
-    },
-    {
-      "epoch": 4.157333225975281,
-      "grad_norm": 0.005394606851041317,
-      "learning_rate": 0.00019999147833466054,
-      "loss": 46.0,
-      "step": 25815
-    },
-    {
-      "epoch": 4.157494263054068,
-      "grad_norm": 0.0011143069714307785,
-      "learning_rate": 0.0001999914776741909,
-      "loss": 46.0,
-      "step": 25816
-    },
-    {
-      "epoch": 4.157655300132856,
-      "grad_norm": 0.001310808351263404,
-      "learning_rate": 0.0001999914770136957,
-      "loss": 46.0,
-      "step": 25817
-    },
-    {
-      "epoch": 4.157816337211643,
-      "grad_norm": 0.002002993132919073,
-      "learning_rate": 0.00019999147635317484,
-      "loss": 46.0,
-      "step": 25818
-    },
-    {
-      "epoch": 4.1579773742904305,
-      "grad_norm": 0.003437912557274103,
-      "learning_rate": 0.00019999147569262844,
-      "loss": 46.0,
-      "step": 25819
-    },
-    {
-      "epoch": 4.158138411369218,
-      "grad_norm": 0.007866820320487022,
-      "learning_rate": 0.00019999147503205645,
-      "loss": 46.0,
-      "step": 25820
-    },
-    {
-      "epoch": 4.158299448448005,
-      "grad_norm": 0.00494722742587328,
-      "learning_rate": 0.00019999147437145887,
-      "loss": 46.0,
-      "step": 25821
-    },
-    {
-      "epoch": 4.158460485526793,
-      "grad_norm": 0.0016632643528282642,
-      "learning_rate": 0.00019999147371083567,
-      "loss": 46.0,
-      "step": 25822
-    },
-    {
-      "epoch": 4.15862152260558,
-      "grad_norm": 0.0009520756429992616,
-      "learning_rate": 0.0001999914730501869,
-      "loss": 46.0,
-      "step": 25823
-    },
-    {
-      "epoch": 4.158782559684368,
-      "grad_norm": 0.014313461259007454,
-      "learning_rate": 0.00019999147238951252,
-      "loss": 46.0,
-      "step": 25824
-    },
-    {
-      "epoch": 4.158943596763155,
-      "grad_norm": 0.0028570096474140882,
-      "learning_rate": 0.00019999147172881254,
-      "loss": 46.0,
-      "step": 25825
-    },
-    {
-      "epoch": 4.159104633841942,
-      "grad_norm": 0.002582564949989319,
-      "learning_rate": 0.000199991471068087,
-      "loss": 46.0,
-      "step": 25826
-    },
-    {
-      "epoch": 4.159265670920729,
-      "grad_norm": 0.003907115198671818,
-      "learning_rate": 0.00019999147040733584,
-      "loss": 46.0,
-      "step": 25827
-    },
-    {
-      "epoch": 4.159426707999517,
-      "grad_norm": 0.0027338529471307993,
-      "learning_rate": 0.00019999146974655912,
-      "loss": 46.0,
-      "step": 25828
-    },
-    {
-      "epoch": 4.159587745078304,
-      "grad_norm": 0.007554525509476662,
-      "learning_rate": 0.00019999146908575676,
-      "loss": 46.0,
-      "step": 25829
-    },
-    {
-      "epoch": 4.1597487821570915,
-      "grad_norm": 0.0034019004087895155,
-      "learning_rate": 0.00019999146842492884,
-      "loss": 46.0,
-      "step": 25830
-    },
-    {
-      "epoch": 4.159909819235879,
-      "grad_norm": 0.006143419072031975,
-      "learning_rate": 0.00019999146776407533,
-      "loss": 46.0,
-      "step": 25831
-    },
-    {
-      "epoch": 4.160070856314666,
-      "grad_norm": 0.0030050959903746843,
-      "learning_rate": 0.0001999914671031962,
-      "loss": 46.0,
-      "step": 25832
-    },
-    {
-      "epoch": 4.160231893393454,
-      "grad_norm": 0.004732530098408461,
-      "learning_rate": 0.0001999914664422915,
-      "loss": 46.0,
-      "step": 25833
-    },
-    {
-      "epoch": 4.160392930472241,
-      "grad_norm": 0.00803361739963293,
-      "learning_rate": 0.0001999914657813612,
-      "loss": 46.0,
-      "step": 25834
-    },
-    {
-      "epoch": 4.160553967551029,
-      "grad_norm": 0.0039372858591377735,
-      "learning_rate": 0.0001999914651204053,
-      "loss": 46.0,
-      "step": 25835
-    },
-    {
-      "epoch": 4.160715004629816,
-      "grad_norm": 0.006818860769271851,
-      "learning_rate": 0.00019999146445942383,
-      "loss": 46.0,
-      "step": 25836
-    },
-    {
-      "epoch": 4.160876041708604,
-      "grad_norm": 0.0009600259363651276,
-      "learning_rate": 0.00019999146379841672,
-      "loss": 46.0,
-      "step": 25837
-    },
-    {
-      "epoch": 4.161037078787391,
-      "grad_norm": 0.0019614805933088064,
-      "learning_rate": 0.00019999146313738408,
-      "loss": 46.0,
-      "step": 25838
-    },
-    {
-      "epoch": 4.161198115866179,
-      "grad_norm": 0.003422726644203067,
-      "learning_rate": 0.00019999146247632582,
-      "loss": 46.0,
-      "step": 25839
-    },
-    {
-      "epoch": 4.161359152944965,
-      "grad_norm": 0.01131085678935051,
-      "learning_rate": 0.00019999146181524195,
-      "loss": 46.0,
-      "step": 25840
-    },
-    {
-      "epoch": 4.161520190023753,
-      "grad_norm": 0.0016847657971084118,
-      "learning_rate": 0.00019999146115413252,
-      "loss": 46.0,
-      "step": 25841
-    },
-    {
-      "epoch": 4.16168122710254,
-      "grad_norm": 0.0014023496769368649,
-      "learning_rate": 0.00019999146049299747,
-      "loss": 46.0,
-      "step": 25842
-    },
-    {
-      "epoch": 4.1618422641813275,
-      "grad_norm": 0.0047528911381959915,
-      "learning_rate": 0.00019999145983183684,
-      "loss": 46.0,
-      "step": 25843
-    },
-    {
-      "epoch": 4.162003301260115,
-      "grad_norm": 0.002841240493580699,
-      "learning_rate": 0.00019999145917065062,
-      "loss": 46.0,
-      "step": 25844
-    },
-    {
-      "epoch": 4.162164338338902,
-      "grad_norm": 0.005046960897743702,
-      "learning_rate": 0.00019999145850943878,
-      "loss": 46.0,
-      "step": 25845
-    },
-    {
-      "epoch": 4.16232537541769,
-      "grad_norm": 0.001826892956160009,
-      "learning_rate": 0.00019999145784820139,
-      "loss": 46.0,
-      "step": 25846
-    },
-    {
-      "epoch": 4.162486412496477,
-      "grad_norm": 0.0008144687162712216,
-      "learning_rate": 0.0001999914571869384,
-      "loss": 46.0,
-      "step": 25847
-    },
-    {
-      "epoch": 4.162647449575265,
-      "grad_norm": 0.002259097760543227,
-      "learning_rate": 0.00019999145652564978,
-      "loss": 46.0,
-      "step": 25848
-    },
-    {
-      "epoch": 4.162808486654052,
-      "grad_norm": 0.008259464055299759,
-      "learning_rate": 0.0001999914558643356,
-      "loss": 46.0,
-      "step": 25849
-    },
-    {
-      "epoch": 4.16296952373284,
-      "grad_norm": 0.009847515262663364,
-      "learning_rate": 0.0001999914552029958,
-      "loss": 46.0,
-      "step": 25850
-    },
-    {
-      "epoch": 4.163130560811627,
-      "grad_norm": 0.0024707375559955835,
-      "learning_rate": 0.0001999914545416304,
-      "loss": 46.0,
-      "step": 25851
-    },
-    {
-      "epoch": 4.163291597890415,
-      "grad_norm": 0.001153877703472972,
-      "learning_rate": 0.00019999145388023947,
-      "loss": 46.0,
-      "step": 25852
-    },
-    {
-      "epoch": 4.163452634969202,
-      "grad_norm": 0.0018085596384480596,
-      "learning_rate": 0.0001999914532188229,
-      "loss": 46.0,
-      "step": 25853
-    },
-    {
-      "epoch": 4.163613672047989,
-      "grad_norm": 0.004764970857650042,
-      "learning_rate": 0.00019999145255738076,
-      "loss": 46.0,
-      "step": 25854
-    },
-    {
-      "epoch": 4.163774709126776,
-      "grad_norm": 0.0024147185031324625,
-      "learning_rate": 0.000199991451895913,
-      "loss": 46.0,
-      "step": 25855
-    },
-    {
-      "epoch": 4.1639357462055635,
-      "grad_norm": 0.005599981173872948,
-      "learning_rate": 0.00019999145123441968,
-      "loss": 46.0,
-      "step": 25856
-    },
-    {
-      "epoch": 4.164096783284351,
-      "grad_norm": 0.0013498428743332624,
-      "learning_rate": 0.00019999145057290074,
-      "loss": 46.0,
-      "step": 25857
-    },
-    {
-      "epoch": 4.164257820363138,
-      "grad_norm": 0.006182997953146696,
-      "learning_rate": 0.00019999144991135622,
-      "loss": 46.0,
-      "step": 25858
-    },
-    {
-      "epoch": 4.164418857441926,
-      "grad_norm": 0.0023755005095154047,
-      "learning_rate": 0.0001999914492497861,
-      "loss": 46.0,
-      "step": 25859
-    },
-    {
-      "epoch": 4.164579894520713,
-      "grad_norm": 0.008315110579133034,
-      "learning_rate": 0.0001999914485881904,
-      "loss": 46.0,
-      "step": 25860
-    },
-    {
-      "epoch": 4.164740931599501,
-      "grad_norm": 0.009556795470416546,
-      "learning_rate": 0.00019999144792656908,
-      "loss": 46.0,
-      "step": 25861
-    },
-    {
-      "epoch": 4.164901968678288,
-      "grad_norm": 0.002784101292490959,
-      "learning_rate": 0.0001999914472649222,
-      "loss": 46.0,
-      "step": 25862
-    },
-    {
-      "epoch": 4.165063005757076,
-      "grad_norm": 0.00457377266138792,
-      "learning_rate": 0.00019999144660324972,
-      "loss": 46.0,
-      "step": 25863
-    },
-    {
-      "epoch": 4.165224042835863,
-      "grad_norm": 0.002098704455420375,
-      "learning_rate": 0.00019999144594155165,
-      "loss": 46.0,
-      "step": 25864
-    },
-    {
-      "epoch": 4.165385079914651,
-      "grad_norm": 0.002842725021764636,
-      "learning_rate": 0.00019999144527982796,
-      "loss": 46.0,
-      "step": 25865
-    },
-    {
-      "epoch": 4.165546116993438,
-      "grad_norm": 0.0026035672053694725,
-      "learning_rate": 0.0001999914446180787,
-      "loss": 46.0,
-      "step": 25866
-    },
-    {
-      "epoch": 4.1657071540722255,
-      "grad_norm": 0.008512424305081367,
-      "learning_rate": 0.00019999144395630385,
-      "loss": 46.0,
-      "step": 25867
-    },
-    {
-      "epoch": 4.165868191151013,
-      "grad_norm": 0.0016673884820193052,
-      "learning_rate": 0.00019999144329450338,
-      "loss": 46.0,
-      "step": 25868
-    },
-    {
-      "epoch": 4.1660292282297995,
-      "grad_norm": 0.00853794626891613,
-      "learning_rate": 0.00019999144263267734,
-      "loss": 46.0,
-      "step": 25869
-    },
-    {
-      "epoch": 4.166190265308587,
-      "grad_norm": 0.005087920464575291,
-      "learning_rate": 0.00019999144197082572,
-      "loss": 46.0,
-      "step": 25870
-    },
-    {
-      "epoch": 4.166351302387374,
-      "grad_norm": 0.004212118219584227,
-      "learning_rate": 0.00019999144130894848,
-      "loss": 46.0,
-      "step": 25871
-    },
-    {
-      "epoch": 4.166512339466162,
-      "grad_norm": 0.005717687774449587,
-      "learning_rate": 0.00019999144064704566,
-      "loss": 46.0,
-      "step": 25872
-    },
-    {
-      "epoch": 4.166673376544949,
-      "grad_norm": 0.004749008920043707,
-      "learning_rate": 0.00019999143998511725,
-      "loss": 46.0,
-      "step": 25873
-    },
-    {
-      "epoch": 4.166834413623737,
-      "grad_norm": 0.003818746190518141,
-      "learning_rate": 0.00019999143932316322,
-      "loss": 46.0,
-      "step": 25874
-    },
-    {
-      "epoch": 4.166995450702524,
-      "grad_norm": 0.011723394505679607,
-      "learning_rate": 0.00019999143866118363,
-      "loss": 46.0,
-      "step": 25875
-    },
-    {
-      "epoch": 4.167156487781312,
-      "grad_norm": 0.003460391191765666,
-      "learning_rate": 0.00019999143799917843,
-      "loss": 46.0,
-      "step": 25876
-    },
-    {
-      "epoch": 4.167317524860099,
-      "grad_norm": 0.0020264191552996635,
-      "learning_rate": 0.00019999143733714765,
-      "loss": 46.0,
-      "step": 25877
-    },
-    {
-      "epoch": 4.1674785619388865,
-      "grad_norm": 0.0030912510119378567,
-      "learning_rate": 0.00019999143667509127,
-      "loss": 46.0,
-      "step": 25878
-    },
-    {
-      "epoch": 4.167639599017674,
-      "grad_norm": 0.016076941043138504,
-      "learning_rate": 0.0001999914360130093,
-      "loss": 46.0,
-      "step": 25879
-    },
-    {
-      "epoch": 4.1678006360964615,
-      "grad_norm": 0.002340406645089388,
-      "learning_rate": 0.00019999143535090173,
-      "loss": 46.0,
-      "step": 25880
-    },
-    {
-      "epoch": 4.167961673175249,
-      "grad_norm": 0.002248638542369008,
-      "learning_rate": 0.0001999914346887686,
-      "loss": 46.0,
-      "step": 25881
-    },
-    {
-      "epoch": 4.168122710254036,
-      "grad_norm": 0.004312028642743826,
-      "learning_rate": 0.00019999143402660985,
-      "loss": 46.0,
-      "step": 25882
-    },
-    {
-      "epoch": 4.168283747332824,
-      "grad_norm": 0.007768033538013697,
-      "learning_rate": 0.00019999143336442548,
-      "loss": 46.0,
-      "step": 25883
-    },
-    {
-      "epoch": 4.16844478441161,
-      "grad_norm": 0.008499770425260067,
-      "learning_rate": 0.00019999143270221556,
-      "loss": 46.0,
-      "step": 25884
-    },
-    {
-      "epoch": 4.168605821490398,
-      "grad_norm": 0.0028355568647384644,
-      "learning_rate": 0.00019999143203998005,
-      "loss": 46.0,
-      "step": 25885
-    },
-    {
-      "epoch": 4.168766858569185,
-      "grad_norm": 0.004336782265454531,
-      "learning_rate": 0.00019999143137771892,
-      "loss": 46.0,
-      "step": 25886
-    },
-    {
-      "epoch": 4.168927895647973,
-      "grad_norm": 0.0016734452219679952,
-      "learning_rate": 0.0001999914307154322,
-      "loss": 46.0,
-      "step": 25887
-    },
-    {
-      "epoch": 4.16908893272676,
-      "grad_norm": 0.005384510848671198,
-      "learning_rate": 0.0001999914300531199,
-      "loss": 46.0,
-      "step": 25888
-    },
-    {
-      "epoch": 4.169249969805548,
-      "grad_norm": 0.0032869975548237562,
-      "learning_rate": 0.00019999142939078202,
-      "loss": 46.0,
-      "step": 25889
-    },
-    {
-      "epoch": 4.169411006884335,
-      "grad_norm": 0.002316078171133995,
-      "learning_rate": 0.00019999142872841855,
-      "loss": 46.0,
-      "step": 25890
-    },
-    {
-      "epoch": 4.1695720439631225,
-      "grad_norm": 0.011938927695155144,
-      "learning_rate": 0.00019999142806602946,
-      "loss": 46.0,
-      "step": 25891
-    },
-    {
-      "epoch": 4.16973308104191,
-      "grad_norm": 0.002336437115445733,
-      "learning_rate": 0.00019999142740361475,
-      "loss": 46.0,
-      "step": 25892
-    },
-    {
-      "epoch": 4.169894118120697,
-      "grad_norm": 0.010293380357325077,
-      "learning_rate": 0.0001999914267411745,
-      "loss": 46.0,
-      "step": 25893
-    },
-    {
-      "epoch": 4.170055155199485,
-      "grad_norm": 0.006510206498205662,
-      "learning_rate": 0.00019999142607870864,
-      "loss": 46.0,
-      "step": 25894
-    },
-    {
-      "epoch": 4.170216192278272,
-      "grad_norm": 0.00653754873201251,
-      "learning_rate": 0.00019999142541621718,
-      "loss": 46.0,
-      "step": 25895
-    },
-    {
-      "epoch": 4.17037722935706,
-      "grad_norm": 0.0033379278611391783,
-      "learning_rate": 0.00019999142475370015,
-      "loss": 46.0,
-      "step": 25896
-    },
-    {
-      "epoch": 4.170538266435847,
-      "grad_norm": 0.00745510496199131,
-      "learning_rate": 0.0001999914240911575,
-      "loss": 46.0,
-      "step": 25897
-    },
-    {
-      "epoch": 4.170699303514635,
-      "grad_norm": 0.0014718774473294616,
-      "learning_rate": 0.0001999914234285893,
-      "loss": 46.0,
-      "step": 25898
-    },
-    {
-      "epoch": 4.170860340593421,
-      "grad_norm": 0.0015339652309194207,
-      "learning_rate": 0.00019999142276599545,
-      "loss": 46.0,
-      "step": 25899
-    },
-    {
-      "epoch": 4.171021377672209,
-      "grad_norm": 0.0006713454495184124,
-      "learning_rate": 0.00019999142210337602,
-      "loss": 46.0,
-      "step": 25900
-    },
-    {
-      "epoch": 4.171182414750996,
-      "grad_norm": 0.0007346144411712885,
-      "learning_rate": 0.00019999142144073103,
-      "loss": 46.0,
-      "step": 25901
-    },
-    {
-      "epoch": 4.171343451829784,
-      "grad_norm": 0.004603332839906216,
-      "learning_rate": 0.00019999142077806043,
-      "loss": 46.0,
-      "step": 25902
-    },
-    {
-      "epoch": 4.171504488908571,
-      "grad_norm": 0.023984026163816452,
-      "learning_rate": 0.0001999914201153642,
-      "loss": 46.0,
-      "step": 25903
-    },
-    {
-      "epoch": 4.1716655259873585,
-      "grad_norm": 0.002608900424093008,
-      "learning_rate": 0.00019999141945264244,
-      "loss": 46.0,
-      "step": 25904
-    },
-    {
-      "epoch": 4.171826563066146,
-      "grad_norm": 0.004634843673557043,
-      "learning_rate": 0.00019999141878989507,
-      "loss": 46.0,
-      "step": 25905
-    },
-    {
-      "epoch": 4.171987600144933,
-      "grad_norm": 0.00967329554259777,
-      "learning_rate": 0.0001999914181271221,
-      "loss": 46.0,
-      "step": 25906
-    },
-    {
-      "epoch": 4.172148637223721,
-      "grad_norm": 0.0016136339399963617,
-      "learning_rate": 0.00019999141746432356,
-      "loss": 46.0,
-      "step": 25907
-    },
-    {
-      "epoch": 4.172309674302508,
-      "grad_norm": 0.0045011527836322784,
-      "learning_rate": 0.00019999141680149938,
-      "loss": 46.0,
-      "step": 25908
-    },
-    {
-      "epoch": 4.172470711381296,
-      "grad_norm": 0.0010537662310525775,
-      "learning_rate": 0.0001999914161386496,
-      "loss": 46.0,
-      "step": 25909
-    },
-    {
-      "epoch": 4.172631748460083,
-      "grad_norm": 0.0017520738765597343,
-      "learning_rate": 0.00019999141547577426,
-      "loss": 46.0,
-      "step": 25910
-    },
-    {
-      "epoch": 4.172792785538871,
-      "grad_norm": 0.006552866194397211,
-      "learning_rate": 0.00019999141481287334,
-      "loss": 46.0,
-      "step": 25911
-    },
-    {
-      "epoch": 4.172953822617658,
-      "grad_norm": 0.004470761399716139,
-      "learning_rate": 0.00019999141414994682,
-      "loss": 46.0,
-      "step": 25912
-    },
-    {
-      "epoch": 4.173114859696445,
-      "grad_norm": 0.0013951323926448822,
-      "learning_rate": 0.00019999141348699467,
-      "loss": 46.0,
-      "step": 25913
-    },
-    {
-      "epoch": 4.173275896775232,
-      "grad_norm": 0.004247300326824188,
-      "learning_rate": 0.000199991412824017,
-      "loss": 46.0,
-      "step": 25914
-    },
-    {
-      "epoch": 4.17343693385402,
-      "grad_norm": 0.0028988532721996307,
-      "learning_rate": 0.00019999141216101368,
-      "loss": 46.0,
-      "step": 25915
-    },
-    {
-      "epoch": 4.173597970932807,
-      "grad_norm": 0.0037056265864521265,
-      "learning_rate": 0.0001999914114979848,
-      "loss": 46.0,
-      "step": 25916
-    },
-    {
-      "epoch": 4.1737590080115945,
-      "grad_norm": 0.0030603790655732155,
-      "learning_rate": 0.0001999914108349303,
-      "loss": 46.0,
-      "step": 25917
-    },
-    {
-      "epoch": 4.173920045090382,
-      "grad_norm": 0.004141510464251041,
-      "learning_rate": 0.0001999914101718502,
-      "loss": 46.0,
-      "step": 25918
-    },
-    {
-      "epoch": 4.174081082169169,
-      "grad_norm": 0.004250417463481426,
-      "learning_rate": 0.00019999140950874454,
-      "loss": 46.0,
-      "step": 25919
-    },
-    {
-      "epoch": 4.174242119247957,
-      "grad_norm": 0.012560942210257053,
-      "learning_rate": 0.00019999140884561327,
-      "loss": 46.0,
-      "step": 25920
-    },
-    {
-      "epoch": 4.174403156326744,
-      "grad_norm": 0.0010623267153277993,
-      "learning_rate": 0.0001999914081824564,
-      "loss": 46.0,
-      "step": 25921
-    },
-    {
-      "epoch": 4.174564193405532,
-      "grad_norm": 0.0036591319367289543,
-      "learning_rate": 0.00019999140751927394,
-      "loss": 46.0,
-      "step": 25922
-    },
-    {
-      "epoch": 4.174725230484319,
-      "grad_norm": 0.0076692369766533375,
-      "learning_rate": 0.0001999914068560659,
-      "loss": 46.0,
-      "step": 25923
-    },
-    {
-      "epoch": 4.174886267563107,
-      "grad_norm": 0.0024209062103182077,
-      "learning_rate": 0.00019999140619283225,
-      "loss": 46.0,
-      "step": 25924
-    },
-    {
-      "epoch": 4.175047304641894,
-      "grad_norm": 0.004220735281705856,
-      "learning_rate": 0.00019999140552957303,
-      "loss": 46.0,
-      "step": 25925
-    },
-    {
-      "epoch": 4.175208341720682,
-      "grad_norm": 0.010969248600304127,
-      "learning_rate": 0.0001999914048662882,
-      "loss": 46.0,
-      "step": 25926
-    },
-    {
-      "epoch": 4.175369378799468,
-      "grad_norm": 0.0036004185676574707,
-      "learning_rate": 0.0001999914042029778,
-      "loss": 46.0,
-      "step": 25927
-    },
-    {
-      "epoch": 4.175530415878256,
-      "grad_norm": 0.003311747219413519,
-      "learning_rate": 0.00019999140353964176,
-      "loss": 46.0,
-      "step": 25928
-    },
-    {
-      "epoch": 4.175691452957043,
-      "grad_norm": 0.0037359322886914015,
-      "learning_rate": 0.00019999140287628016,
-      "loss": 46.0,
-      "step": 25929
-    },
-    {
-      "epoch": 4.1758524900358305,
-      "grad_norm": 0.0011967142345383763,
-      "learning_rate": 0.00019999140221289296,
-      "loss": 46.0,
-      "step": 25930
-    },
-    {
-      "epoch": 4.176013527114618,
-      "grad_norm": 0.002566169947385788,
-      "learning_rate": 0.0001999914015494802,
-      "loss": 46.0,
-      "step": 25931
-    },
-    {
-      "epoch": 4.176174564193405,
-      "grad_norm": 0.0010997374774888158,
-      "learning_rate": 0.00019999140088604181,
-      "loss": 46.0,
-      "step": 25932
-    },
-    {
-      "epoch": 4.176335601272193,
-      "grad_norm": 0.004429088439792395,
-      "learning_rate": 0.00019999140022257782,
-      "loss": 46.0,
-      "step": 25933
-    },
-    {
-      "epoch": 4.17649663835098,
-      "grad_norm": 0.004189136903733015,
-      "learning_rate": 0.00019999139955908827,
-      "loss": 46.0,
-      "step": 25934
-    },
-    {
-      "epoch": 4.176657675429768,
-      "grad_norm": 0.0036843728739768267,
-      "learning_rate": 0.0001999913988955731,
-      "loss": 46.0,
-      "step": 25935
-    },
-    {
-      "epoch": 4.176818712508555,
-      "grad_norm": 0.002175193279981613,
-      "learning_rate": 0.00019999139823203234,
-      "loss": 46.0,
-      "step": 25936
-    },
-    {
-      "epoch": 4.176979749587343,
-      "grad_norm": 0.0028209525626152754,
-      "learning_rate": 0.000199991397568466,
-      "loss": 46.0,
-      "step": 25937
-    },
-    {
-      "epoch": 4.17714078666613,
-      "grad_norm": 0.0025836420245468616,
-      "learning_rate": 0.00019999139690487407,
-      "loss": 46.0,
-      "step": 25938
-    },
-    {
-      "epoch": 4.1773018237449175,
-      "grad_norm": 0.0052740322425961494,
-      "learning_rate": 0.00019999139624125653,
-      "loss": 46.0,
-      "step": 25939
-    },
-    {
-      "epoch": 4.177462860823705,
-      "grad_norm": 0.006183016579598188,
-      "learning_rate": 0.00019999139557761342,
-      "loss": 46.0,
-      "step": 25940
-    },
-    {
-      "epoch": 4.1776238979024924,
-      "grad_norm": 0.0026666871272027493,
-      "learning_rate": 0.0001999913949139447,
-      "loss": 46.0,
-      "step": 25941
-    },
-    {
-      "epoch": 4.177784934981279,
-      "grad_norm": 0.0018965437775477767,
-      "learning_rate": 0.0001999913942502504,
-      "loss": 46.0,
-      "step": 25942
-    },
-    {
-      "epoch": 4.1779459720600665,
-      "grad_norm": 0.0069977957755327225,
-      "learning_rate": 0.00019999139358653048,
-      "loss": 46.0,
-      "step": 25943
-    },
-    {
-      "epoch": 4.178107009138854,
-      "grad_norm": 0.0058150095865130424,
-      "learning_rate": 0.000199991392922785,
-      "loss": 46.0,
-      "step": 25944
-    },
-    {
-      "epoch": 4.178268046217641,
-      "grad_norm": 0.0025091199204325676,
-      "learning_rate": 0.0001999913922590139,
-      "loss": 46.0,
-      "step": 25945
-    },
-    {
-      "epoch": 4.178429083296429,
-      "grad_norm": 0.004277421161532402,
-      "learning_rate": 0.00019999139159521722,
-      "loss": 46.0,
-      "step": 25946
-    },
-    {
-      "epoch": 4.178590120375216,
-      "grad_norm": 0.002977653406560421,
-      "learning_rate": 0.00019999139093139495,
-      "loss": 46.0,
-      "step": 25947
-    },
-    {
-      "epoch": 4.178751157454004,
-      "grad_norm": 0.0017982794670388103,
-      "learning_rate": 0.0001999913902675471,
-      "loss": 46.0,
-      "step": 25948
-    },
-    {
-      "epoch": 4.178912194532791,
-      "grad_norm": 0.005572856403887272,
-      "learning_rate": 0.00019999138960367363,
-      "loss": 46.0,
-      "step": 25949
-    },
-    {
-      "epoch": 4.179073231611579,
-      "grad_norm": 0.001722762011922896,
-      "learning_rate": 0.00019999138893977457,
-      "loss": 46.0,
-      "step": 25950
-    },
-    {
-      "epoch": 4.179234268690366,
-      "grad_norm": 0.015083261765539646,
-      "learning_rate": 0.00019999138827584995,
-      "loss": 46.0,
-      "step": 25951
-    },
-    {
-      "epoch": 4.1793953057691535,
-      "grad_norm": 0.0018014807719737291,
-      "learning_rate": 0.0001999913876118997,
-      "loss": 46.0,
-      "step": 25952
-    },
-    {
-      "epoch": 4.179556342847941,
-      "grad_norm": 0.015364490449428558,
-      "learning_rate": 0.00019999138694792387,
-      "loss": 46.0,
-      "step": 25953
-    },
-    {
-      "epoch": 4.179717379926728,
-      "grad_norm": 0.002354520605877042,
-      "learning_rate": 0.00019999138628392247,
-      "loss": 46.0,
-      "step": 25954
-    },
-    {
-      "epoch": 4.179878417005516,
-      "grad_norm": 0.00257012783549726,
-      "learning_rate": 0.00019999138561989545,
-      "loss": 46.0,
-      "step": 25955
-    },
-    {
-      "epoch": 4.180039454084303,
-      "grad_norm": 0.006188462022691965,
-      "learning_rate": 0.00019999138495584284,
-      "loss": 46.0,
-      "step": 25956
-    },
-    {
-      "epoch": 4.18020049116309,
-      "grad_norm": 0.012042365036904812,
-      "learning_rate": 0.00019999138429176465,
-      "loss": 46.0,
-      "step": 25957
-    },
-    {
-      "epoch": 4.180361528241877,
-      "grad_norm": 0.004100468475371599,
-      "learning_rate": 0.00019999138362766086,
-      "loss": 46.0,
-      "step": 25958
-    },
-    {
-      "epoch": 4.180522565320665,
-      "grad_norm": 0.0027238535694777966,
-      "learning_rate": 0.00019999138296353147,
-      "loss": 46.0,
-      "step": 25959
-    },
-    {
-      "epoch": 4.180683602399452,
-      "grad_norm": 0.0014333584113046527,
-      "learning_rate": 0.0001999913822993765,
-      "loss": 46.0,
-      "step": 25960
-    },
-    {
-      "epoch": 4.18084463947824,
-      "grad_norm": 0.006552796810865402,
-      "learning_rate": 0.00019999138163519594,
-      "loss": 46.0,
-      "step": 25961
-    },
-    {
-      "epoch": 4.181005676557027,
-      "grad_norm": 0.001441031345166266,
-      "learning_rate": 0.00019999138097098978,
-      "loss": 46.0,
-      "step": 25962
-    },
-    {
-      "epoch": 4.181166713635815,
-      "grad_norm": 0.003725338028743863,
-      "learning_rate": 0.000199991380306758,
-      "loss": 46.0,
-      "step": 25963
-    },
-    {
-      "epoch": 4.181327750714602,
-      "grad_norm": 0.007430018857121468,
-      "learning_rate": 0.00019999137964250068,
-      "loss": 46.0,
-      "step": 25964
-    },
-    {
-      "epoch": 4.1814887877933895,
-      "grad_norm": 0.003349131438881159,
-      "learning_rate": 0.00019999137897821773,
-      "loss": 46.0,
-      "step": 25965
-    },
-    {
-      "epoch": 4.181649824872177,
-      "grad_norm": 0.004528583958745003,
-      "learning_rate": 0.0001999913783139092,
-      "loss": 46.0,
-      "step": 25966
-    },
-    {
-      "epoch": 4.181810861950964,
-      "grad_norm": 0.007112594787031412,
-      "learning_rate": 0.00019999137764957508,
-      "loss": 46.0,
-      "step": 25967
-    },
-    {
-      "epoch": 4.181971899029752,
-      "grad_norm": 0.011922754347324371,
-      "learning_rate": 0.00019999137698521535,
-      "loss": 46.0,
-      "step": 25968
-    },
-    {
-      "epoch": 4.182132936108539,
-      "grad_norm": 0.0030051604844629765,
-      "learning_rate": 0.00019999137632083005,
-      "loss": 46.0,
-      "step": 25969
-    },
-    {
-      "epoch": 4.182293973187327,
-      "grad_norm": 0.0033459635451436043,
-      "learning_rate": 0.00019999137565641914,
-      "loss": 46.0,
-      "step": 25970
-    },
-    {
-      "epoch": 4.182455010266114,
-      "grad_norm": 0.02054242603480816,
-      "learning_rate": 0.00019999137499198267,
-      "loss": 46.0,
-      "step": 25971
-    },
-    {
-      "epoch": 4.182616047344901,
-      "grad_norm": 0.004144029226154089,
-      "learning_rate": 0.00019999137432752056,
-      "loss": 46.0,
-      "step": 25972
-    },
-    {
-      "epoch": 4.182777084423688,
-      "grad_norm": 0.0021680439822375774,
-      "learning_rate": 0.00019999137366303286,
-      "loss": 46.0,
-      "step": 25973
-    },
-    {
-      "epoch": 4.182938121502476,
-      "grad_norm": 0.007987151853740215,
-      "learning_rate": 0.0001999913729985196,
-      "loss": 46.0,
-      "step": 25974
-    },
-    {
-      "epoch": 4.183099158581263,
-      "grad_norm": 0.007736022584140301,
-      "learning_rate": 0.00019999137233398073,
-      "loss": 46.0,
-      "step": 25975
-    },
-    {
-      "epoch": 4.183260195660051,
-      "grad_norm": 0.0018217575270682573,
-      "learning_rate": 0.00019999137166941628,
-      "loss": 46.0,
-      "step": 25976
-    },
-    {
-      "epoch": 4.183421232738838,
-      "grad_norm": 0.00365527905523777,
-      "learning_rate": 0.00019999137100482623,
-      "loss": 46.0,
-      "step": 25977
-    },
-    {
-      "epoch": 4.1835822698176255,
-      "grad_norm": 0.006252591032534838,
-      "learning_rate": 0.00019999137034021057,
-      "loss": 46.0,
-      "step": 25978
-    },
-    {
-      "epoch": 4.183743306896413,
-      "grad_norm": 0.0016039125621318817,
-      "learning_rate": 0.00019999136967556935,
-      "loss": 46.0,
-      "step": 25979
-    },
-    {
-      "epoch": 4.1839043439752,
-      "grad_norm": 0.002131569664925337,
-      "learning_rate": 0.0001999913690109025,
-      "loss": 46.0,
-      "step": 25980
-    },
-    {
-      "epoch": 4.184065381053988,
-      "grad_norm": 0.0071007730439305305,
-      "learning_rate": 0.00019999136834621012,
-      "loss": 46.0,
-      "step": 25981
-    },
-    {
-      "epoch": 4.184226418132775,
-      "grad_norm": 0.005144949536770582,
-      "learning_rate": 0.00019999136768149208,
-      "loss": 46.0,
-      "step": 25982
-    },
-    {
-      "epoch": 4.184387455211563,
-      "grad_norm": 0.0024578217417001724,
-      "learning_rate": 0.00019999136701674848,
-      "loss": 46.0,
-      "step": 25983
-    },
-    {
-      "epoch": 4.18454849229035,
-      "grad_norm": 0.0034769868943840265,
-      "learning_rate": 0.00019999136635197927,
-      "loss": 46.0,
-      "step": 25984
-    },
-    {
-      "epoch": 4.184709529369138,
-      "grad_norm": 0.0010151652386412024,
-      "learning_rate": 0.00019999136568718448,
-      "loss": 46.0,
-      "step": 25985
-    },
-    {
-      "epoch": 4.184870566447924,
-      "grad_norm": 0.0035235495306551456,
-      "learning_rate": 0.00019999136502236412,
-      "loss": 46.0,
-      "step": 25986
-    },
-    {
-      "epoch": 4.185031603526712,
-      "grad_norm": 0.0016252518398687243,
-      "learning_rate": 0.00019999136435751812,
-      "loss": 46.0,
-      "step": 25987
-    },
-    {
-      "epoch": 4.185192640605499,
-      "grad_norm": 0.007759832311421633,
-      "learning_rate": 0.00019999136369264656,
-      "loss": 46.0,
-      "step": 25988
-    },
-    {
-      "epoch": 4.185353677684287,
-      "grad_norm": 0.005069665610790253,
-      "learning_rate": 0.00019999136302774936,
-      "loss": 46.0,
-      "step": 25989
-    },
-    {
-      "epoch": 4.185514714763074,
-      "grad_norm": 0.0035398025065660477,
-      "learning_rate": 0.00019999136236282662,
-      "loss": 46.0,
-      "step": 25990
-    },
-    {
-      "epoch": 4.1856757518418615,
-      "grad_norm": 0.009738585911691189,
-      "learning_rate": 0.00019999136169787828,
-      "loss": 46.0,
-      "step": 25991
-    },
-    {
-      "epoch": 4.185836788920649,
-      "grad_norm": 0.0013918853364884853,
-      "learning_rate": 0.00019999136103290431,
-      "loss": 46.0,
-      "step": 25992
-    },
-    {
-      "epoch": 4.185997825999436,
-      "grad_norm": 0.0032958476804196835,
-      "learning_rate": 0.0001999913603679048,
-      "loss": 46.0,
-      "step": 25993
-    },
-    {
-      "epoch": 4.186158863078224,
-      "grad_norm": 0.0035923910327255726,
-      "learning_rate": 0.00019999135970287968,
-      "loss": 46.0,
-      "step": 25994
-    },
-    {
-      "epoch": 4.186319900157011,
-      "grad_norm": 0.00758514367043972,
-      "learning_rate": 0.00019999135903782898,
-      "loss": 46.0,
-      "step": 25995
-    },
-    {
-      "epoch": 4.186480937235799,
-      "grad_norm": 0.0036441378761082888,
-      "learning_rate": 0.00019999135837275265,
-      "loss": 46.0,
-      "step": 25996
-    },
-    {
-      "epoch": 4.186641974314586,
-      "grad_norm": 0.002245335839688778,
-      "learning_rate": 0.00019999135770765075,
-      "loss": 46.0,
-      "step": 25997
-    },
-    {
-      "epoch": 4.186803011393374,
-      "grad_norm": 0.0045206849463284016,
-      "learning_rate": 0.00019999135704252324,
-      "loss": 46.0,
-      "step": 25998
-    },
-    {
-      "epoch": 4.186964048472161,
-      "grad_norm": 0.0017208128701895475,
-      "learning_rate": 0.00019999135637737014,
-      "loss": 46.0,
-      "step": 25999
-    },
-    {
-      "epoch": 4.1871250855509485,
-      "grad_norm": 0.002556613879278302,
-      "learning_rate": 0.00019999135571219148,
-      "loss": 46.0,
-      "step": 26000
-    },
-    {
-      "epoch": 4.187286122629735,
-      "grad_norm": 0.0019640920218080282,
-      "learning_rate": 0.00019999135504698718,
-      "loss": 46.0,
-      "step": 26001
-    },
-    {
-      "epoch": 4.1874471597085225,
-      "grad_norm": 0.004104936961084604,
-      "learning_rate": 0.00019999135438175734,
-      "loss": 46.0,
-      "step": 26002
-    },
-    {
-      "epoch": 4.18760819678731,
-      "grad_norm": 0.005912536755204201,
-      "learning_rate": 0.00019999135371650187,
-      "loss": 46.0,
-      "step": 26003
-    },
-    {
-      "epoch": 4.1877692338660975,
-      "grad_norm": 0.005457097664475441,
-      "learning_rate": 0.00019999135305122083,
-      "loss": 46.0,
-      "step": 26004
-    },
-    {
-      "epoch": 4.187930270944885,
-      "grad_norm": 0.013931632973253727,
-      "learning_rate": 0.00019999135238591418,
-      "loss": 46.0,
-      "step": 26005
-    },
-    {
-      "epoch": 4.188091308023672,
-      "grad_norm": 0.007091052830219269,
-      "learning_rate": 0.00019999135172058194,
-      "loss": 46.0,
-      "step": 26006
-    },
-    {
-      "epoch": 4.18825234510246,
-      "grad_norm": 0.0008303801296278834,
-      "learning_rate": 0.0001999913510552241,
-      "loss": 46.0,
-      "step": 26007
-    },
-    {
-      "epoch": 4.188413382181247,
-      "grad_norm": 0.015890713781118393,
-      "learning_rate": 0.00019999135038984068,
-      "loss": 46.0,
-      "step": 26008
-    },
-    {
-      "epoch": 4.188574419260035,
-      "grad_norm": 0.0024414905346930027,
-      "learning_rate": 0.00019999134972443168,
-      "loss": 46.0,
-      "step": 26009
-    },
-    {
-      "epoch": 4.188735456338822,
-      "grad_norm": 0.0013518971391022205,
-      "learning_rate": 0.00019999134905899704,
-      "loss": 46.0,
-      "step": 26010
-    },
-    {
-      "epoch": 4.18889649341761,
-      "grad_norm": 0.006290055345743895,
-      "learning_rate": 0.00019999134839353687,
-      "loss": 46.0,
-      "step": 26011
-    },
-    {
-      "epoch": 4.189057530496397,
-      "grad_norm": 0.0018630078993737698,
-      "learning_rate": 0.00019999134772805106,
-      "loss": 46.0,
-      "step": 26012
-    },
-    {
-      "epoch": 4.1892185675751845,
-      "grad_norm": 0.008745655417442322,
-      "learning_rate": 0.00019999134706253968,
-      "loss": 46.0,
-      "step": 26013
-    },
-    {
-      "epoch": 4.189379604653972,
-      "grad_norm": 0.0027428437024354935,
-      "learning_rate": 0.00019999134639700267,
-      "loss": 46.0,
-      "step": 26014
-    },
-    {
-      "epoch": 4.1895406417327585,
-      "grad_norm": 0.0038365055806934834,
-      "learning_rate": 0.00019999134573144012,
-      "loss": 46.0,
-      "step": 26015
-    },
-    {
-      "epoch": 4.189701678811546,
-      "grad_norm": 0.0038948110304772854,
-      "learning_rate": 0.00019999134506585195,
-      "loss": 46.0,
-      "step": 26016
-    },
-    {
-      "epoch": 4.189862715890333,
-      "grad_norm": 0.0009821417042985559,
-      "learning_rate": 0.00019999134440023818,
-      "loss": 46.0,
-      "step": 26017
-    },
-    {
-      "epoch": 4.190023752969121,
-      "grad_norm": 0.001074650906957686,
-      "learning_rate": 0.00019999134373459884,
-      "loss": 46.0,
-      "step": 26018
-    },
-    {
-      "epoch": 4.190184790047908,
-      "grad_norm": 0.0015313818585127592,
-      "learning_rate": 0.0001999913430689339,
-      "loss": 46.0,
-      "step": 26019
-    },
-    {
-      "epoch": 4.190345827126696,
-      "grad_norm": 0.0018712281016632915,
-      "learning_rate": 0.00019999134240324335,
-      "loss": 46.0,
-      "step": 26020
-    },
-    {
-      "epoch": 4.190506864205483,
-      "grad_norm": 0.0031497799791395664,
-      "learning_rate": 0.00019999134173752725,
-      "loss": 46.0,
-      "step": 26021
-    },
-    {
-      "epoch": 4.190667901284271,
-      "grad_norm": 0.006303925067186356,
-      "learning_rate": 0.0001999913410717855,
-      "loss": 46.0,
-      "step": 26022
-    },
-    {
-      "epoch": 4.190828938363058,
-      "grad_norm": 0.0014620608417317271,
-      "learning_rate": 0.0001999913404060182,
-      "loss": 46.0,
-      "step": 26023
-    },
-    {
-      "epoch": 4.190989975441846,
-      "grad_norm": 0.004717608913779259,
-      "learning_rate": 0.00019999133974022527,
-      "loss": 46.0,
-      "step": 26024
-    },
-    {
-      "epoch": 4.191151012520633,
-      "grad_norm": 0.0011663367040455341,
-      "learning_rate": 0.00019999133907440682,
-      "loss": 46.0,
-      "step": 26025
-    },
-    {
-      "epoch": 4.1913120495994205,
-      "grad_norm": 0.0015338532393798232,
-      "learning_rate": 0.00019999133840856273,
-      "loss": 46.0,
-      "step": 26026
-    },
-    {
-      "epoch": 4.191473086678208,
-      "grad_norm": 0.0032446887344121933,
-      "learning_rate": 0.000199991337742693,
-      "loss": 46.0,
-      "step": 26027
-    },
-    {
-      "epoch": 4.191634123756995,
-      "grad_norm": 0.007691057864576578,
-      "learning_rate": 0.00019999133707679777,
-      "loss": 46.0,
-      "step": 26028
-    },
-    {
-      "epoch": 4.191795160835783,
-      "grad_norm": 0.0011235671117901802,
-      "learning_rate": 0.0001999913364108769,
-      "loss": 46.0,
-      "step": 26029
-    },
-    {
-      "epoch": 4.191956197914569,
-      "grad_norm": 0.011805804446339607,
-      "learning_rate": 0.00019999133574493042,
-      "loss": 46.0,
-      "step": 26030
-    },
-    {
-      "epoch": 4.192117234993357,
-      "grad_norm": 0.009845497086644173,
-      "learning_rate": 0.00019999133507895834,
-      "loss": 46.0,
-      "step": 26031
-    },
-    {
-      "epoch": 4.192278272072144,
-      "grad_norm": 0.005638534668833017,
-      "learning_rate": 0.0001999913344129607,
-      "loss": 46.0,
-      "step": 26032
-    },
-    {
-      "epoch": 4.192439309150932,
-      "grad_norm": 0.002307936316356063,
-      "learning_rate": 0.00019999133374693745,
-      "loss": 46.0,
-      "step": 26033
-    },
-    {
-      "epoch": 4.192600346229719,
-      "grad_norm": 0.0071575697511434555,
-      "learning_rate": 0.00019999133308088864,
-      "loss": 46.0,
-      "step": 26034
-    },
-    {
-      "epoch": 4.192761383308507,
-      "grad_norm": 0.000621053040958941,
-      "learning_rate": 0.0001999913324148142,
-      "loss": 46.0,
-      "step": 26035
-    },
-    {
-      "epoch": 4.192922420387294,
-      "grad_norm": 0.004900424275547266,
-      "learning_rate": 0.0001999913317487142,
-      "loss": 46.0,
-      "step": 26036
-    },
-    {
-      "epoch": 4.193083457466082,
-      "grad_norm": 0.0015247509581968188,
-      "learning_rate": 0.0001999913310825886,
-      "loss": 46.0,
-      "step": 26037
-    },
-    {
-      "epoch": 4.193244494544869,
-      "grad_norm": 0.005635363049805164,
-      "learning_rate": 0.00019999133041643738,
-      "loss": 46.0,
-      "step": 26038
-    },
-    {
-      "epoch": 4.1934055316236565,
-      "grad_norm": 0.006605396047234535,
-      "learning_rate": 0.00019999132975026057,
-      "loss": 46.0,
-      "step": 26039
-    },
-    {
-      "epoch": 4.193566568702444,
-      "grad_norm": 0.005611578933894634,
-      "learning_rate": 0.0001999913290840582,
-      "loss": 46.0,
-      "step": 26040
-    },
-    {
-      "epoch": 4.193727605781231,
-      "grad_norm": 0.004594260361045599,
-      "learning_rate": 0.0001999913284178302,
-      "loss": 46.0,
-      "step": 26041
-    },
-    {
-      "epoch": 4.193888642860019,
-      "grad_norm": 0.003198167309165001,
-      "learning_rate": 0.00019999132775157664,
-      "loss": 46.0,
-      "step": 26042
-    },
-    {
-      "epoch": 4.194049679938806,
-      "grad_norm": 0.00684497831389308,
-      "learning_rate": 0.00019999132708529746,
-      "loss": 46.0,
-      "step": 26043
-    },
-    {
-      "epoch": 4.194210717017594,
-      "grad_norm": 0.003949678037315607,
-      "learning_rate": 0.00019999132641899272,
-      "loss": 46.0,
-      "step": 26044
-    },
-    {
-      "epoch": 4.19437175409638,
-      "grad_norm": 0.0029609936755150557,
-      "learning_rate": 0.00019999132575266234,
-      "loss": 46.0,
-      "step": 26045
-    },
-    {
-      "epoch": 4.194532791175168,
-      "grad_norm": 0.013808994553983212,
-      "learning_rate": 0.00019999132508630643,
-      "loss": 46.0,
-      "step": 26046
-    },
-    {
-      "epoch": 4.194693828253955,
-      "grad_norm": 0.0022515649907290936,
-      "learning_rate": 0.00019999132441992487,
-      "loss": 46.0,
-      "step": 26047
-    },
-    {
-      "epoch": 4.194854865332743,
-      "grad_norm": 0.0062814923003315926,
-      "learning_rate": 0.00019999132375351776,
-      "loss": 46.0,
-      "step": 26048
-    },
-    {
-      "epoch": 4.19501590241153,
-      "grad_norm": 0.005903781391680241,
-      "learning_rate": 0.00019999132308708503,
-      "loss": 46.0,
-      "step": 26049
-    },
-    {
-      "epoch": 4.195176939490318,
-      "grad_norm": 0.014126066118478775,
-      "learning_rate": 0.0001999913224206267,
-      "loss": 46.0,
-      "step": 26050
-    },
-    {
-      "epoch": 4.195337976569105,
-      "grad_norm": 0.01271476037800312,
-      "learning_rate": 0.0001999913217541428,
-      "loss": 46.0,
-      "step": 26051
-    },
-    {
-      "epoch": 4.1954990136478925,
-      "grad_norm": 0.004971700720489025,
-      "learning_rate": 0.00019999132108763332,
-      "loss": 46.0,
-      "step": 26052
-    },
-    {
-      "epoch": 4.19566005072668,
-      "grad_norm": 0.00850827619433403,
-      "learning_rate": 0.00019999132042109824,
-      "loss": 46.0,
-      "step": 26053
-    },
-    {
-      "epoch": 4.195821087805467,
-      "grad_norm": 0.005059359595179558,
-      "learning_rate": 0.00019999131975453752,
-      "loss": 46.0,
-      "step": 26054
-    },
-    {
-      "epoch": 4.195982124884255,
-      "grad_norm": 0.0032999475952237844,
-      "learning_rate": 0.00019999131908795127,
-      "loss": 46.0,
-      "step": 26055
-    },
-    {
-      "epoch": 4.196143161963042,
-      "grad_norm": 0.002233864739537239,
-      "learning_rate": 0.00019999131842133938,
-      "loss": 46.0,
-      "step": 26056
-    },
-    {
-      "epoch": 4.19630419904183,
-      "grad_norm": 0.007805574685335159,
-      "learning_rate": 0.00019999131775470192,
-      "loss": 46.0,
-      "step": 26057
-    },
-    {
-      "epoch": 4.196465236120617,
-      "grad_norm": 0.00768780754879117,
-      "learning_rate": 0.00019999131708803885,
-      "loss": 46.0,
-      "step": 26058
-    },
-    {
-      "epoch": 4.196626273199404,
-      "grad_norm": 0.007372531108558178,
-      "learning_rate": 0.0001999913164213502,
-      "loss": 46.0,
-      "step": 26059
-    },
-    {
-      "epoch": 4.196787310278191,
-      "grad_norm": 0.0017058696830645204,
-      "learning_rate": 0.00019999131575463596,
-      "loss": 46.0,
-      "step": 26060
-    },
-    {
-      "epoch": 4.196948347356979,
-      "grad_norm": 0.0016624911222606897,
-      "learning_rate": 0.00019999131508789613,
-      "loss": 46.0,
-      "step": 26061
-    },
-    {
-      "epoch": 4.197109384435766,
-      "grad_norm": 0.0008786537800915539,
-      "learning_rate": 0.0001999913144211307,
-      "loss": 46.0,
-      "step": 26062
-    },
-    {
-      "epoch": 4.1972704215145535,
-      "grad_norm": 0.0029030819423496723,
-      "learning_rate": 0.0001999913137543397,
-      "loss": 46.0,
-      "step": 26063
-    },
-    {
-      "epoch": 4.197431458593341,
-      "grad_norm": 0.003672652179375291,
-      "learning_rate": 0.0001999913130875231,
-      "loss": 46.0,
-      "step": 26064
-    },
-    {
-      "epoch": 4.197592495672128,
-      "grad_norm": 0.009355917572975159,
-      "learning_rate": 0.00019999131242068088,
-      "loss": 46.0,
-      "step": 26065
-    },
-    {
-      "epoch": 4.197753532750916,
-      "grad_norm": 0.01713617891073227,
-      "learning_rate": 0.00019999131175381306,
-      "loss": 46.0,
-      "step": 26066
-    },
-    {
-      "epoch": 4.197914569829703,
-      "grad_norm": 0.0029010928701609373,
-      "learning_rate": 0.00019999131108691968,
-      "loss": 46.0,
-      "step": 26067
-    },
-    {
-      "epoch": 4.198075606908491,
-      "grad_norm": 0.0013409628299996257,
-      "learning_rate": 0.0001999913104200007,
-      "loss": 46.0,
-      "step": 26068
-    },
-    {
-      "epoch": 4.198236643987278,
-      "grad_norm": 0.004620407707989216,
-      "learning_rate": 0.00019999130975305613,
-      "loss": 46.0,
-      "step": 26069
-    },
-    {
-      "epoch": 4.198397681066066,
-      "grad_norm": 0.015187554992735386,
-      "learning_rate": 0.00019999130908608597,
-      "loss": 46.0,
-      "step": 26070
-    },
-    {
-      "epoch": 4.198558718144853,
-      "grad_norm": 0.002239885739982128,
-      "learning_rate": 0.00019999130841909018,
-      "loss": 46.0,
-      "step": 26071
-    },
-    {
-      "epoch": 4.198719755223641,
-      "grad_norm": 0.004163686186075211,
-      "learning_rate": 0.00019999130775206884,
-      "loss": 46.0,
-      "step": 26072
-    },
-    {
-      "epoch": 4.198880792302428,
-      "grad_norm": 0.0018952051177620888,
-      "learning_rate": 0.00019999130708502188,
-      "loss": 46.0,
-      "step": 26073
-    },
-    {
-      "epoch": 4.199041829381215,
-      "grad_norm": 0.0025391767267137766,
-      "learning_rate": 0.00019999130641794937,
-      "loss": 46.0,
-      "step": 26074
-    },
-    {
-      "epoch": 4.199202866460002,
-      "grad_norm": 0.0011277308221906424,
-      "learning_rate": 0.00019999130575085124,
-      "loss": 46.0,
-      "step": 26075
-    },
-    {
-      "epoch": 4.1993639035387895,
-      "grad_norm": 0.0027454053051769733,
-      "learning_rate": 0.00019999130508372752,
-      "loss": 46.0,
-      "step": 26076
-    },
-    {
-      "epoch": 4.199524940617577,
-      "grad_norm": 0.002370435744524002,
-      "learning_rate": 0.00019999130441657818,
-      "loss": 46.0,
-      "step": 26077
-    },
-    {
-      "epoch": 4.199685977696364,
-      "grad_norm": 0.0012260667281225324,
-      "learning_rate": 0.00019999130374940326,
-      "loss": 46.0,
-      "step": 26078
-    },
-    {
-      "epoch": 4.199847014775152,
-      "grad_norm": 0.005955003201961517,
-      "learning_rate": 0.00019999130308220278,
-      "loss": 46.0,
-      "step": 26079
-    },
-    {
-      "epoch": 4.200008051853939,
-      "grad_norm": 0.0014732145937159657,
-      "learning_rate": 0.0001999913024149767,
-      "loss": 46.0,
-      "step": 26080
-    },
-    {
-      "epoch": 4.200169088932727,
-      "grad_norm": 0.003810550319030881,
-      "learning_rate": 0.00019999130174772498,
-      "loss": 46.0,
-      "step": 26081
-    },
-    {
-      "epoch": 4.200330126011514,
-      "grad_norm": 0.01232217438519001,
-      "learning_rate": 0.0001999913010804477,
-      "loss": 46.0,
-      "step": 26082
-    },
-    {
-      "epoch": 4.200491163090302,
-      "grad_norm": 0.0014904760755598545,
-      "learning_rate": 0.00019999130041314486,
-      "loss": 46.0,
-      "step": 26083
-    },
-    {
-      "epoch": 4.200652200169089,
-      "grad_norm": 0.002004674170166254,
-      "learning_rate": 0.0001999912997458164,
-      "loss": 46.0,
-      "step": 26084
-    },
-    {
-      "epoch": 4.200813237247877,
-      "grad_norm": 0.00470479391515255,
-      "learning_rate": 0.00019999129907846233,
-      "loss": 46.0,
-      "step": 26085
-    },
-    {
-      "epoch": 4.200974274326664,
-      "grad_norm": 0.007371739484369755,
-      "learning_rate": 0.00019999129841108269,
-      "loss": 46.0,
-      "step": 26086
-    },
-    {
-      "epoch": 4.2011353114054515,
-      "grad_norm": 0.01456478238105774,
-      "learning_rate": 0.00019999129774367745,
-      "loss": 46.0,
-      "step": 26087
-    },
-    {
-      "epoch": 4.201296348484238,
-      "grad_norm": 0.0057002645917236805,
-      "learning_rate": 0.00019999129707624658,
-      "loss": 46.0,
-      "step": 26088
-    },
-    {
-      "epoch": 4.2014573855630255,
-      "grad_norm": 0.002759326249361038,
-      "learning_rate": 0.00019999129640879018,
-      "loss": 46.0,
-      "step": 26089
-    },
-    {
-      "epoch": 4.201618422641813,
-      "grad_norm": 0.0014946566661819816,
-      "learning_rate": 0.00019999129574130816,
-      "loss": 46.0,
-      "step": 26090
-    },
-    {
-      "epoch": 4.2017794597206,
-      "grad_norm": 0.0024217774625867605,
-      "learning_rate": 0.00019999129507380055,
-      "loss": 46.0,
-      "step": 26091
-    },
-    {
-      "epoch": 4.201940496799388,
-      "grad_norm": 0.0016628142911940813,
-      "learning_rate": 0.00019999129440626733,
-      "loss": 46.0,
-      "step": 26092
-    },
-    {
-      "epoch": 4.202101533878175,
-      "grad_norm": 0.006629998330026865,
-      "learning_rate": 0.00019999129373870852,
-      "loss": 46.0,
-      "step": 26093
-    },
-    {
-      "epoch": 4.202262570956963,
-      "grad_norm": 0.0035316506400704384,
-      "learning_rate": 0.00019999129307112412,
-      "loss": 46.0,
-      "step": 26094
-    },
-    {
-      "epoch": 4.20242360803575,
-      "grad_norm": 0.0024842265993356705,
-      "learning_rate": 0.00019999129240351417,
-      "loss": 46.0,
-      "step": 26095
-    },
-    {
-      "epoch": 4.202584645114538,
-      "grad_norm": 0.005518514197319746,
-      "learning_rate": 0.00019999129173587857,
-      "loss": 46.0,
-      "step": 26096
-    },
-    {
-      "epoch": 4.202745682193325,
-      "grad_norm": 0.0038072566967457533,
-      "learning_rate": 0.00019999129106821738,
-      "loss": 46.0,
-      "step": 26097
-    },
-    {
-      "epoch": 4.202906719272113,
-      "grad_norm": 0.007764378096908331,
-      "learning_rate": 0.00019999129040053064,
-      "loss": 46.0,
-      "step": 26098
-    },
-    {
-      "epoch": 4.2030677563509,
-      "grad_norm": 0.005074677523225546,
-      "learning_rate": 0.00019999128973281828,
-      "loss": 46.0,
-      "step": 26099
-    },
-    {
-      "epoch": 4.2032287934296875,
-      "grad_norm": 0.0011158455163240433,
-      "learning_rate": 0.00019999128906508033,
-      "loss": 46.0,
-      "step": 26100
-    },
-    {
-      "epoch": 4.203389830508475,
-      "grad_norm": 0.0019502454670146108,
-      "learning_rate": 0.0001999912883973168,
-      "loss": 46.0,
-      "step": 26101
-    },
-    {
-      "epoch": 4.203550867587262,
-      "grad_norm": 0.0078010293655097485,
-      "learning_rate": 0.00019999128772952765,
-      "loss": 46.0,
-      "step": 26102
-    },
-    {
-      "epoch": 4.203711904666049,
-      "grad_norm": 0.010709693655371666,
-      "learning_rate": 0.00019999128706171292,
-      "loss": 46.0,
-      "step": 26103
-    },
-    {
-      "epoch": 4.203872941744836,
-      "grad_norm": 0.005534814670681953,
-      "learning_rate": 0.00019999128639387262,
-      "loss": 46.0,
-      "step": 26104
-    },
-    {
-      "epoch": 4.204033978823624,
-      "grad_norm": 0.0022047837264835835,
-      "learning_rate": 0.00019999128572600671,
-      "loss": 46.0,
-      "step": 26105
-    },
-    {
-      "epoch": 4.204195015902411,
-      "grad_norm": 0.0052042207680642605,
-      "learning_rate": 0.0001999912850581152,
-      "loss": 46.0,
-      "step": 26106
-    },
-    {
-      "epoch": 4.204356052981199,
-      "grad_norm": 0.011884715408086777,
-      "learning_rate": 0.0001999912843901981,
-      "loss": 46.0,
-      "step": 26107
-    },
-    {
-      "epoch": 4.204517090059986,
-      "grad_norm": 0.0017527422169223428,
-      "learning_rate": 0.0001999912837222554,
-      "loss": 46.0,
-      "step": 26108
-    },
-    {
-      "epoch": 4.204678127138774,
-      "grad_norm": 0.005948703270405531,
-      "learning_rate": 0.00019999128305428715,
-      "loss": 46.0,
-      "step": 26109
-    },
-    {
-      "epoch": 4.204839164217561,
-      "grad_norm": 0.0041191610507667065,
-      "learning_rate": 0.00019999128238629328,
-      "loss": 46.0,
-      "step": 26110
-    },
-    {
-      "epoch": 4.2050002012963485,
-      "grad_norm": 0.007457006257027388,
-      "learning_rate": 0.00019999128171827382,
-      "loss": 46.0,
-      "step": 26111
-    },
-    {
-      "epoch": 4.205161238375136,
-      "grad_norm": 0.006035392638295889,
-      "learning_rate": 0.00019999128105022872,
-      "loss": 46.0,
-      "step": 26112
-    },
-    {
-      "epoch": 4.2053222754539235,
-      "grad_norm": 0.013747980818152428,
-      "learning_rate": 0.00019999128038215811,
-      "loss": 46.0,
-      "step": 26113
-    },
-    {
-      "epoch": 4.205483312532711,
-      "grad_norm": 0.0013991923769935966,
-      "learning_rate": 0.00019999127971406187,
-      "loss": 46.0,
-      "step": 26114
-    },
-    {
-      "epoch": 4.205644349611498,
-      "grad_norm": 0.0037634915206581354,
-      "learning_rate": 0.00019999127904594,
-      "loss": 46.0,
-      "step": 26115
-    },
-    {
-      "epoch": 4.205805386690286,
-      "grad_norm": 0.005118495784699917,
-      "learning_rate": 0.00019999127837779258,
-      "loss": 46.0,
-      "step": 26116
-    },
-    {
-      "epoch": 4.205966423769073,
-      "grad_norm": 0.00338145368732512,
-      "learning_rate": 0.00019999127770961957,
-      "loss": 46.0,
-      "step": 26117
-    },
-    {
-      "epoch": 4.20612746084786,
-      "grad_norm": 0.005637587048113346,
-      "learning_rate": 0.00019999127704142095,
-      "loss": 46.0,
-      "step": 26118
-    },
-    {
-      "epoch": 4.206288497926647,
-      "grad_norm": 0.007844589650630951,
-      "learning_rate": 0.0001999912763731967,
-      "loss": 46.0,
-      "step": 26119
-    },
-    {
-      "epoch": 4.206449535005435,
-      "grad_norm": 0.00894088577479124,
-      "learning_rate": 0.00019999127570494694,
-      "loss": 46.0,
-      "step": 26120
-    },
-    {
-      "epoch": 4.206610572084222,
-      "grad_norm": 0.0014346856623888016,
-      "learning_rate": 0.00019999127503667156,
-      "loss": 46.0,
-      "step": 26121
-    },
-    {
-      "epoch": 4.20677160916301,
-      "grad_norm": 0.004349898546934128,
-      "learning_rate": 0.00019999127436837056,
-      "loss": 46.0,
-      "step": 26122
-    },
-    {
-      "epoch": 4.206932646241797,
-      "grad_norm": 0.0029623203445225954,
-      "learning_rate": 0.00019999127370004397,
-      "loss": 46.0,
-      "step": 26123
-    },
-    {
-      "epoch": 4.2070936833205845,
-      "grad_norm": 0.0021699510980397463,
-      "learning_rate": 0.00019999127303169182,
-      "loss": 46.0,
-      "step": 26124
-    },
-    {
-      "epoch": 4.207254720399372,
-      "grad_norm": 0.004261610563844442,
-      "learning_rate": 0.00019999127236331406,
-      "loss": 46.0,
-      "step": 26125
-    },
-    {
-      "epoch": 4.207415757478159,
-      "grad_norm": 0.0021020250860601664,
-      "learning_rate": 0.0001999912716949107,
-      "loss": 46.0,
-      "step": 26126
-    },
-    {
-      "epoch": 4.207576794556947,
-      "grad_norm": 0.004452476277947426,
-      "learning_rate": 0.00019999127102648175,
-      "loss": 46.0,
-      "step": 26127
-    },
-    {
-      "epoch": 4.207737831635734,
-      "grad_norm": 0.002101828809827566,
-      "learning_rate": 0.0001999912703580272,
-      "loss": 46.0,
-      "step": 26128
-    },
-    {
-      "epoch": 4.207898868714522,
-      "grad_norm": 0.006327989976853132,
-      "learning_rate": 0.00019999126968954707,
-      "loss": 46.0,
-      "step": 26129
-    },
-    {
-      "epoch": 4.208059905793309,
-      "grad_norm": 0.010032350197434425,
-      "learning_rate": 0.00019999126902104134,
-      "loss": 46.0,
-      "step": 26130
-    },
-    {
-      "epoch": 4.208220942872097,
-      "grad_norm": 0.00930061750113964,
-      "learning_rate": 0.00019999126835251003,
-      "loss": 46.0,
-      "step": 26131
-    },
-    {
-      "epoch": 4.208381979950884,
-      "grad_norm": 0.005770361516624689,
-      "learning_rate": 0.0001999912676839531,
-      "loss": 46.0,
-      "step": 26132
-    },
-    {
-      "epoch": 4.208543017029671,
-      "grad_norm": 0.0010209655156359076,
-      "learning_rate": 0.00019999126701537062,
-      "loss": 46.0,
-      "step": 26133
-    },
-    {
-      "epoch": 4.208704054108458,
-      "grad_norm": 0.006891271099448204,
-      "learning_rate": 0.00019999126634676252,
-      "loss": 46.0,
-      "step": 26134
-    },
-    {
-      "epoch": 4.208865091187246,
-      "grad_norm": 0.002525690710172057,
-      "learning_rate": 0.00019999126567812883,
-      "loss": 46.0,
-      "step": 26135
-    },
-    {
-      "epoch": 4.209026128266033,
-      "grad_norm": 0.0023005539551377296,
-      "learning_rate": 0.00019999126500946953,
-      "loss": 46.0,
-      "step": 26136
-    },
-    {
-      "epoch": 4.2091871653448205,
-      "grad_norm": 0.004584599751979113,
-      "learning_rate": 0.00019999126434078467,
-      "loss": 46.0,
-      "step": 26137
-    },
-    {
-      "epoch": 4.209348202423608,
-      "grad_norm": 0.0007366443751379848,
-      "learning_rate": 0.00019999126367207422,
-      "loss": 46.0,
-      "step": 26138
-    },
-    {
-      "epoch": 4.209509239502395,
-      "grad_norm": 0.0011814075987786055,
-      "learning_rate": 0.00019999126300333813,
-      "loss": 46.0,
-      "step": 26139
-    },
-    {
-      "epoch": 4.209670276581183,
-      "grad_norm": 0.004154510796070099,
-      "learning_rate": 0.0001999912623345765,
-      "loss": 46.0,
-      "step": 26140
-    },
-    {
-      "epoch": 4.20983131365997,
-      "grad_norm": 0.00594858406111598,
-      "learning_rate": 0.00019999126166578925,
-      "loss": 46.0,
-      "step": 26141
-    },
-    {
-      "epoch": 4.209992350738758,
-      "grad_norm": 0.004756337963044643,
-      "learning_rate": 0.0001999912609969764,
-      "loss": 46.0,
-      "step": 26142
-    },
-    {
-      "epoch": 4.210153387817545,
-      "grad_norm": 0.015073360875248909,
-      "learning_rate": 0.00019999126032813798,
-      "loss": 46.0,
-      "step": 26143
-    },
-    {
-      "epoch": 4.210314424896333,
-      "grad_norm": 0.007323187310248613,
-      "learning_rate": 0.00019999125965927393,
-      "loss": 46.0,
-      "step": 26144
-    },
-    {
-      "epoch": 4.21047546197512,
-      "grad_norm": 0.003056993242353201,
-      "learning_rate": 0.00019999125899038434,
-      "loss": 46.0,
-      "step": 26145
-    },
-    {
-      "epoch": 4.210636499053908,
-      "grad_norm": 0.00145050382707268,
-      "learning_rate": 0.00019999125832146912,
-      "loss": 46.0,
-      "step": 26146
-    },
-    {
-      "epoch": 4.210797536132694,
-      "grad_norm": 0.004086241591721773,
-      "learning_rate": 0.00019999125765252833,
-      "loss": 46.0,
-      "step": 26147
-    },
-    {
-      "epoch": 4.210958573211482,
-      "grad_norm": 0.00959494523704052,
-      "learning_rate": 0.00019999125698356193,
-      "loss": 46.0,
-      "step": 26148
-    },
-    {
-      "epoch": 4.211119610290269,
-      "grad_norm": 0.0040459055453538895,
-      "learning_rate": 0.00019999125631456994,
-      "loss": 46.0,
-      "step": 26149
-    },
-    {
-      "epoch": 4.2112806473690565,
-      "grad_norm": 0.0015438953414559364,
-      "learning_rate": 0.00019999125564555237,
-      "loss": 46.0,
-      "step": 26150
-    },
-    {
-      "epoch": 4.211441684447844,
-      "grad_norm": 0.002931318012997508,
-      "learning_rate": 0.0001999912549765092,
-      "loss": 46.0,
-      "step": 26151
-    },
-    {
-      "epoch": 4.211602721526631,
-      "grad_norm": 0.0008940070983953774,
-      "learning_rate": 0.00019999125430744043,
-      "loss": 46.0,
-      "step": 26152
-    },
-    {
-      "epoch": 4.211763758605419,
-      "grad_norm": 0.0026036081835627556,
-      "learning_rate": 0.00019999125363834606,
-      "loss": 46.0,
-      "step": 26153
-    },
-    {
-      "epoch": 4.211924795684206,
-      "grad_norm": 0.008871166966855526,
-      "learning_rate": 0.0001999912529692261,
-      "loss": 46.0,
-      "step": 26154
-    },
-    {
-      "epoch": 4.212085832762994,
-      "grad_norm": 0.004649289418011904,
-      "learning_rate": 0.00019999125230008057,
-      "loss": 46.0,
-      "step": 26155
-    },
-    {
-      "epoch": 4.212246869841781,
-      "grad_norm": 0.0029622167348861694,
-      "learning_rate": 0.00019999125163090944,
-      "loss": 46.0,
-      "step": 26156
-    },
-    {
-      "epoch": 4.212407906920569,
-      "grad_norm": 0.009121979586780071,
-      "learning_rate": 0.0001999912509617127,
-      "loss": 46.0,
-      "step": 26157
-    },
-    {
-      "epoch": 4.212568943999356,
-      "grad_norm": 0.0007816595025360584,
-      "learning_rate": 0.0001999912502924904,
-      "loss": 46.0,
-      "step": 26158
-    },
-    {
-      "epoch": 4.212729981078144,
-      "grad_norm": 0.006691058166325092,
-      "learning_rate": 0.00019999124962324246,
-      "loss": 46.0,
-      "step": 26159
-    },
-    {
-      "epoch": 4.212891018156931,
-      "grad_norm": 0.008115086704492569,
-      "learning_rate": 0.000199991248953969,
-      "loss": 46.0,
-      "step": 26160
-    },
-    {
-      "epoch": 4.213052055235718,
-      "grad_norm": 0.005389650817960501,
-      "learning_rate": 0.00019999124828466987,
-      "loss": 46.0,
-      "step": 26161
-    },
-    {
-      "epoch": 4.213213092314505,
-      "grad_norm": 0.008649742230772972,
-      "learning_rate": 0.0001999912476153452,
-      "loss": 46.0,
-      "step": 26162
-    },
-    {
-      "epoch": 4.2133741293932925,
-      "grad_norm": 0.002736145630478859,
-      "learning_rate": 0.00019999124694599488,
-      "loss": 46.0,
-      "step": 26163
-    },
-    {
-      "epoch": 4.21353516647208,
-      "grad_norm": 0.0196493249386549,
-      "learning_rate": 0.00019999124627661903,
-      "loss": 46.0,
-      "step": 26164
-    },
-    {
-      "epoch": 4.213696203550867,
-      "grad_norm": 0.005836357828229666,
-      "learning_rate": 0.00019999124560721756,
-      "loss": 46.0,
-      "step": 26165
-    },
-    {
-      "epoch": 4.213857240629655,
-      "grad_norm": 0.008302832953631878,
-      "learning_rate": 0.00019999124493779049,
-      "loss": 46.0,
-      "step": 26166
-    },
-    {
-      "epoch": 4.214018277708442,
-      "grad_norm": 0.0025642337277531624,
-      "learning_rate": 0.00019999124426833787,
-      "loss": 46.0,
-      "step": 26167
-    },
-    {
-      "epoch": 4.21417931478723,
-      "grad_norm": 0.0027002268470823765,
-      "learning_rate": 0.00019999124359885962,
-      "loss": 46.0,
-      "step": 26168
-    },
-    {
-      "epoch": 4.214340351866017,
-      "grad_norm": 0.002399461343884468,
-      "learning_rate": 0.00019999124292935575,
-      "loss": 46.0,
-      "step": 26169
-    },
-    {
-      "epoch": 4.214501388944805,
-      "grad_norm": 0.0019487167010083795,
-      "learning_rate": 0.00019999124225982633,
-      "loss": 46.0,
-      "step": 26170
-    },
-    {
-      "epoch": 4.214662426023592,
-      "grad_norm": 0.0008998162229545414,
-      "learning_rate": 0.00019999124159027128,
-      "loss": 46.0,
-      "step": 26171
-    },
-    {
-      "epoch": 4.2148234631023795,
-      "grad_norm": 0.005393853411078453,
-      "learning_rate": 0.00019999124092069068,
-      "loss": 46.0,
-      "step": 26172
-    },
-    {
-      "epoch": 4.214984500181167,
-      "grad_norm": 0.0061567011289298534,
-      "learning_rate": 0.0001999912402510845,
-      "loss": 46.0,
-      "step": 26173
-    },
-    {
-      "epoch": 4.2151455372599544,
-      "grad_norm": 0.006997460033744574,
-      "learning_rate": 0.00019999123958145266,
-      "loss": 46.0,
-      "step": 26174
-    },
-    {
-      "epoch": 4.215306574338742,
-      "grad_norm": 0.004063728731125593,
-      "learning_rate": 0.00019999123891179527,
-      "loss": 46.0,
-      "step": 26175
-    },
-    {
-      "epoch": 4.2154676114175285,
-      "grad_norm": 0.001859088079072535,
-      "learning_rate": 0.0001999912382421123,
-      "loss": 46.0,
-      "step": 26176
-    },
-    {
-      "epoch": 4.215628648496316,
-      "grad_norm": 0.0017855068435892463,
-      "learning_rate": 0.0001999912375724037,
-      "loss": 46.0,
-      "step": 26177
-    },
-    {
-      "epoch": 4.215789685575103,
-      "grad_norm": 0.0018491818336769938,
-      "learning_rate": 0.00019999123690266955,
-      "loss": 46.0,
-      "step": 26178
-    },
-    {
-      "epoch": 4.215950722653891,
-      "grad_norm": 0.004011351149529219,
-      "learning_rate": 0.00019999123623290976,
-      "loss": 46.0,
-      "step": 26179
-    },
-    {
-      "epoch": 4.216111759732678,
-      "grad_norm": 0.0011792851146310568,
-      "learning_rate": 0.00019999123556312443,
-      "loss": 46.0,
-      "step": 26180
-    },
-    {
-      "epoch": 4.216272796811466,
-      "grad_norm": 0.01074582152068615,
-      "learning_rate": 0.00019999123489331346,
-      "loss": 46.0,
-      "step": 26181
-    },
-    {
-      "epoch": 4.216433833890253,
-      "grad_norm": 0.013922140933573246,
-      "learning_rate": 0.0001999912342234769,
-      "loss": 46.0,
-      "step": 26182
-    },
-    {
-      "epoch": 4.216594870969041,
-      "grad_norm": 0.002625813940539956,
-      "learning_rate": 0.0001999912335536148,
-      "loss": 46.0,
-      "step": 26183
-    },
-    {
-      "epoch": 4.216755908047828,
-      "grad_norm": 0.01154703926295042,
-      "learning_rate": 0.00019999123288372706,
-      "loss": 46.0,
-      "step": 26184
-    },
-    {
-      "epoch": 4.2169169451266155,
-      "grad_norm": 0.0023028203286230564,
-      "learning_rate": 0.00019999123221381372,
-      "loss": 46.0,
-      "step": 26185
-    },
-    {
-      "epoch": 4.217077982205403,
-      "grad_norm": 0.003878034418448806,
-      "learning_rate": 0.00019999123154387482,
-      "loss": 46.0,
-      "step": 26186
-    },
-    {
-      "epoch": 4.21723901928419,
-      "grad_norm": 0.0031337235122919083,
-      "learning_rate": 0.00019999123087391033,
-      "loss": 46.0,
-      "step": 26187
-    },
-    {
-      "epoch": 4.217400056362978,
-      "grad_norm": 0.003292187349870801,
-      "learning_rate": 0.00019999123020392022,
-      "loss": 46.0,
-      "step": 26188
-    },
-    {
-      "epoch": 4.217561093441765,
-      "grad_norm": 0.003771548392251134,
-      "learning_rate": 0.00019999122953390455,
-      "loss": 46.0,
-      "step": 26189
-    },
-    {
-      "epoch": 4.217722130520553,
-      "grad_norm": 0.005378633737564087,
-      "learning_rate": 0.00019999122886386322,
-      "loss": 46.0,
-      "step": 26190
-    },
-    {
-      "epoch": 4.217883167599339,
-      "grad_norm": 0.004960936959832907,
-      "learning_rate": 0.00019999122819379635,
-      "loss": 46.0,
-      "step": 26191
-    },
-    {
-      "epoch": 4.218044204678127,
-      "grad_norm": 0.004270308651030064,
-      "learning_rate": 0.00019999122752370393,
-      "loss": 46.0,
-      "step": 26192
-    },
-    {
-      "epoch": 4.218205241756914,
-      "grad_norm": 0.0016056271269917488,
-      "learning_rate": 0.00019999122685358583,
-      "loss": 46.0,
-      "step": 26193
-    },
-    {
-      "epoch": 4.218366278835702,
-      "grad_norm": 0.01219590101391077,
-      "learning_rate": 0.0001999912261834422,
-      "loss": 46.0,
-      "step": 26194
-    },
-    {
-      "epoch": 4.218527315914489,
-      "grad_norm": 0.002686714520677924,
-      "learning_rate": 0.00019999122551327294,
-      "loss": 46.0,
-      "step": 26195
-    },
-    {
-      "epoch": 4.218688352993277,
-      "grad_norm": 0.0009120117174461484,
-      "learning_rate": 0.0001999912248430781,
-      "loss": 46.0,
-      "step": 26196
-    },
-    {
-      "epoch": 4.218849390072064,
-      "grad_norm": 0.002591249532997608,
-      "learning_rate": 0.00019999122417285766,
-      "loss": 46.0,
-      "step": 26197
-    },
-    {
-      "epoch": 4.2190104271508515,
-      "grad_norm": 0.003160852240398526,
-      "learning_rate": 0.00019999122350261163,
-      "loss": 46.0,
-      "step": 26198
-    },
-    {
-      "epoch": 4.219171464229639,
-      "grad_norm": 0.0018156197620555758,
-      "learning_rate": 0.00019999122283234,
-      "loss": 46.0,
-      "step": 26199
-    },
-    {
-      "epoch": 4.219332501308426,
-      "grad_norm": 0.0076890247873961926,
-      "learning_rate": 0.0001999912221620428,
-      "loss": 46.0,
-      "step": 26200
-    },
-    {
-      "epoch": 4.219493538387214,
-      "grad_norm": 0.002638919046148658,
-      "learning_rate": 0.00019999122149172002,
-      "loss": 46.0,
-      "step": 26201
-    },
-    {
-      "epoch": 4.219654575466001,
-      "grad_norm": 0.008569651283323765,
-      "learning_rate": 0.00019999122082137164,
-      "loss": 46.0,
-      "step": 26202
-    },
-    {
-      "epoch": 4.219815612544789,
-      "grad_norm": 0.008760536089539528,
-      "learning_rate": 0.00019999122015099761,
-      "loss": 46.0,
-      "step": 26203
-    },
-    {
-      "epoch": 4.219976649623576,
-      "grad_norm": 0.00193566654343158,
-      "learning_rate": 0.00019999121948059803,
-      "loss": 46.0,
-      "step": 26204
-    },
-    {
-      "epoch": 4.220137686702364,
-      "grad_norm": 0.006570394616574049,
-      "learning_rate": 0.00019999121881017286,
-      "loss": 46.0,
-      "step": 26205
-    },
-    {
-      "epoch": 4.22029872378115,
-      "grad_norm": 0.0019867916125804186,
-      "learning_rate": 0.0001999912181397221,
-      "loss": 46.0,
-      "step": 26206
-    },
-    {
-      "epoch": 4.220459760859938,
-      "grad_norm": 0.0013249969342723489,
-      "learning_rate": 0.00019999121746924574,
-      "loss": 46.0,
-      "step": 26207
-    },
-    {
-      "epoch": 4.220620797938725,
-      "grad_norm": 0.002329988870769739,
-      "learning_rate": 0.0001999912167987438,
-      "loss": 46.0,
-      "step": 26208
-    },
-    {
-      "epoch": 4.220781835017513,
-      "grad_norm": 0.0028690302278846502,
-      "learning_rate": 0.00019999121612821624,
-      "loss": 46.0,
-      "step": 26209
-    },
-    {
-      "epoch": 4.2209428720963,
-      "grad_norm": 0.00662878667935729,
-      "learning_rate": 0.00019999121545766313,
-      "loss": 46.0,
-      "step": 26210
-    },
-    {
-      "epoch": 4.2211039091750875,
-      "grad_norm": 0.004774064291268587,
-      "learning_rate": 0.0001999912147870844,
-      "loss": 46.0,
-      "step": 26211
-    },
-    {
-      "epoch": 4.221264946253875,
-      "grad_norm": 0.004695198033004999,
-      "learning_rate": 0.00019999121411648006,
-      "loss": 46.0,
-      "step": 26212
-    },
-    {
-      "epoch": 4.221425983332662,
-      "grad_norm": 0.0052057174034416676,
-      "learning_rate": 0.00019999121344585014,
-      "loss": 46.0,
-      "step": 26213
-    },
-    {
-      "epoch": 4.22158702041145,
-      "grad_norm": 0.006722983904182911,
-      "learning_rate": 0.00019999121277519463,
-      "loss": 46.0,
-      "step": 26214
-    },
-    {
-      "epoch": 4.221748057490237,
-      "grad_norm": 0.002662076847627759,
-      "learning_rate": 0.00019999121210451354,
-      "loss": 46.0,
-      "step": 26215
-    },
-    {
-      "epoch": 4.221909094569025,
-      "grad_norm": 0.005133450962603092,
-      "learning_rate": 0.00019999121143380685,
-      "loss": 46.0,
-      "step": 26216
-    },
-    {
-      "epoch": 4.222070131647812,
-      "grad_norm": 0.010997754521667957,
-      "learning_rate": 0.00019999121076307456,
-      "loss": 46.0,
-      "step": 26217
-    },
-    {
-      "epoch": 4.2222311687266,
-      "grad_norm": 0.001192996627651155,
-      "learning_rate": 0.0001999912100923167,
-      "loss": 46.0,
-      "step": 26218
-    },
-    {
-      "epoch": 4.222392205805387,
-      "grad_norm": 0.0022864823695272207,
-      "learning_rate": 0.00019999120942153323,
-      "loss": 46.0,
-      "step": 26219
-    },
-    {
-      "epoch": 4.222553242884174,
-      "grad_norm": 0.0033758634235709906,
-      "learning_rate": 0.00019999120875072418,
-      "loss": 46.0,
-      "step": 26220
-    },
-    {
-      "epoch": 4.222714279962961,
-      "grad_norm": 0.005932088941335678,
-      "learning_rate": 0.0001999912080798895,
-      "loss": 46.0,
-      "step": 26221
-    },
-    {
-      "epoch": 4.222875317041749,
-      "grad_norm": 0.0035091752652078867,
-      "learning_rate": 0.00019999120740902925,
-      "loss": 46.0,
-      "step": 26222
-    },
-    {
-      "epoch": 4.223036354120536,
-      "grad_norm": 0.0006309850141406059,
-      "learning_rate": 0.0001999912067381434,
-      "loss": 46.0,
-      "step": 26223
-    },
-    {
-      "epoch": 4.2231973911993235,
-      "grad_norm": 0.008238116279244423,
-      "learning_rate": 0.000199991206067232,
-      "loss": 46.0,
-      "step": 26224
-    },
-    {
-      "epoch": 4.223358428278111,
-      "grad_norm": 0.0032019237987697124,
-      "learning_rate": 0.00019999120539629495,
-      "loss": 46.0,
-      "step": 26225
-    },
-    {
-      "epoch": 4.223519465356898,
-      "grad_norm": 0.005181713495403528,
-      "learning_rate": 0.00019999120472533234,
-      "loss": 46.0,
-      "step": 26226
-    },
-    {
-      "epoch": 4.223680502435686,
-      "grad_norm": 0.0021451578941196203,
-      "learning_rate": 0.00019999120405434412,
-      "loss": 46.0,
-      "step": 26227
-    },
-    {
-      "epoch": 4.223841539514473,
-      "grad_norm": 0.007014904171228409,
-      "learning_rate": 0.0001999912033833303,
-      "loss": 46.0,
-      "step": 26228
-    },
-    {
-      "epoch": 4.224002576593261,
-      "grad_norm": 0.002018124796450138,
-      "learning_rate": 0.0001999912027122909,
-      "loss": 46.0,
-      "step": 26229
-    },
-    {
-      "epoch": 4.224163613672048,
-      "grad_norm": 0.003496370278298855,
-      "learning_rate": 0.00019999120204122593,
-      "loss": 46.0,
-      "step": 26230
-    },
-    {
-      "epoch": 4.224324650750836,
-      "grad_norm": 0.0058142924681305885,
-      "learning_rate": 0.00019999120137013536,
-      "loss": 46.0,
-      "step": 26231
-    },
-    {
-      "epoch": 4.224485687829623,
-      "grad_norm": 0.006493426393717527,
-      "learning_rate": 0.00019999120069901917,
-      "loss": 46.0,
-      "step": 26232
-    },
-    {
-      "epoch": 4.2246467249084105,
-      "grad_norm": 0.002277251798659563,
-      "learning_rate": 0.0001999912000278774,
-      "loss": 46.0,
-      "step": 26233
-    },
-    {
-      "epoch": 4.224807761987197,
-      "grad_norm": 0.008645284920930862,
-      "learning_rate": 0.00019999119935671004,
-      "loss": 46.0,
-      "step": 26234
-    },
-    {
-      "epoch": 4.2249687990659845,
-      "grad_norm": 0.0012503694742918015,
-      "learning_rate": 0.0001999911986855171,
-      "loss": 46.0,
-      "step": 26235
-    },
-    {
-      "epoch": 4.225129836144772,
-      "grad_norm": 0.0045026084408164024,
-      "learning_rate": 0.00019999119801429856,
-      "loss": 46.0,
-      "step": 26236
-    },
-    {
-      "epoch": 4.2252908732235595,
-      "grad_norm": 0.0014557895483449101,
-      "learning_rate": 0.00019999119734305442,
-      "loss": 46.0,
-      "step": 26237
-    },
-    {
-      "epoch": 4.225451910302347,
-      "grad_norm": 0.008168893866240978,
-      "learning_rate": 0.00019999119667178468,
-      "loss": 46.0,
-      "step": 26238
-    },
-    {
-      "epoch": 4.225612947381134,
-      "grad_norm": 0.0018320472445338964,
-      "learning_rate": 0.00019999119600048936,
-      "loss": 46.0,
-      "step": 26239
-    },
-    {
-      "epoch": 4.225773984459922,
-      "grad_norm": 0.006885674316436052,
-      "learning_rate": 0.00019999119532916842,
-      "loss": 46.0,
-      "step": 26240
-    },
-    {
-      "epoch": 4.225935021538709,
-      "grad_norm": 0.00178134860470891,
-      "learning_rate": 0.00019999119465782196,
-      "loss": 46.0,
-      "step": 26241
-    },
-    {
-      "epoch": 4.226096058617497,
-      "grad_norm": 0.0034408732317388058,
-      "learning_rate": 0.00019999119398644985,
-      "loss": 46.0,
-      "step": 26242
-    },
-    {
-      "epoch": 4.226257095696284,
-      "grad_norm": 0.009279229678213596,
-      "learning_rate": 0.00019999119331505215,
-      "loss": 46.0,
-      "step": 26243
-    },
-    {
-      "epoch": 4.226418132775072,
-      "grad_norm": 0.002921992912888527,
-      "learning_rate": 0.00019999119264362886,
-      "loss": 46.0,
-      "step": 26244
-    },
-    {
-      "epoch": 4.226579169853859,
-      "grad_norm": 0.00111789267975837,
-      "learning_rate": 0.00019999119197217996,
-      "loss": 46.0,
-      "step": 26245
-    },
-    {
-      "epoch": 4.2267402069326465,
-      "grad_norm": 0.00583791546523571,
-      "learning_rate": 0.0001999911913007055,
-      "loss": 46.0,
-      "step": 26246
-    },
-    {
-      "epoch": 4.226901244011434,
-      "grad_norm": 0.0016119221691042185,
-      "learning_rate": 0.00019999119062920543,
-      "loss": 46.0,
-      "step": 26247
-    },
-    {
-      "epoch": 4.227062281090221,
-      "grad_norm": 0.004340555984526873,
-      "learning_rate": 0.0001999911899576798,
-      "loss": 46.0,
-      "step": 26248
-    },
-    {
-      "epoch": 4.227223318169008,
-      "grad_norm": 0.012931088916957378,
-      "learning_rate": 0.00019999118928612855,
-      "loss": 46.0,
-      "step": 26249
-    },
-    {
-      "epoch": 4.227384355247795,
-      "grad_norm": 0.0026871436275541782,
-      "learning_rate": 0.00019999118861455171,
-      "loss": 46.0,
-      "step": 26250
-    },
-    {
-      "epoch": 4.227545392326583,
-      "grad_norm": 0.0041232905350625515,
-      "learning_rate": 0.00019999118794294927,
-      "loss": 46.0,
-      "step": 26251
-    },
-    {
-      "epoch": 4.22770642940537,
-      "grad_norm": 0.0015007088659331203,
-      "learning_rate": 0.00019999118727132126,
-      "loss": 46.0,
-      "step": 26252
-    },
-    {
-      "epoch": 4.227867466484158,
-      "grad_norm": 0.004195880610495806,
-      "learning_rate": 0.00019999118659966763,
-      "loss": 46.0,
-      "step": 26253
-    },
-    {
-      "epoch": 4.228028503562945,
-      "grad_norm": 0.008949180133640766,
-      "learning_rate": 0.00019999118592798842,
-      "loss": 46.0,
-      "step": 26254
-    },
-    {
-      "epoch": 4.228189540641733,
-      "grad_norm": 0.002593415090814233,
-      "learning_rate": 0.0001999911852562836,
-      "loss": 46.0,
-      "step": 26255
-    },
-    {
-      "epoch": 4.22835057772052,
-      "grad_norm": 0.0029092226177453995,
-      "learning_rate": 0.0001999911845845532,
-      "loss": 46.0,
-      "step": 26256
-    },
-    {
-      "epoch": 4.228511614799308,
-      "grad_norm": 0.0051793926395475864,
-      "learning_rate": 0.0001999911839127972,
-      "loss": 46.0,
-      "step": 26257
-    },
-    {
-      "epoch": 4.228672651878095,
-      "grad_norm": 0.0020929318852722645,
-      "learning_rate": 0.00019999118324101565,
-      "loss": 46.0,
-      "step": 26258
-    },
-    {
-      "epoch": 4.2288336889568825,
-      "grad_norm": 0.01367963757365942,
-      "learning_rate": 0.00019999118256920845,
-      "loss": 46.0,
-      "step": 26259
-    },
-    {
-      "epoch": 4.22899472603567,
-      "grad_norm": 0.005756871309131384,
-      "learning_rate": 0.0001999911818973757,
-      "loss": 46.0,
-      "step": 26260
-    },
-    {
-      "epoch": 4.229155763114457,
-      "grad_norm": 0.010185593739151955,
-      "learning_rate": 0.00019999118122551734,
-      "loss": 46.0,
-      "step": 26261
-    },
-    {
-      "epoch": 4.229316800193245,
-      "grad_norm": 0.0005250861286185682,
-      "learning_rate": 0.00019999118055363338,
-      "loss": 46.0,
-      "step": 26262
-    },
-    {
-      "epoch": 4.229477837272032,
-      "grad_norm": 0.005665406119078398,
-      "learning_rate": 0.00019999117988172383,
-      "loss": 46.0,
-      "step": 26263
-    },
-    {
-      "epoch": 4.229638874350819,
-      "grad_norm": 0.00416199816390872,
-      "learning_rate": 0.0001999911792097887,
-      "loss": 46.0,
-      "step": 26264
-    },
-    {
-      "epoch": 4.229799911429606,
-      "grad_norm": 0.004688936751335859,
-      "learning_rate": 0.00019999117853782797,
-      "loss": 46.0,
-      "step": 26265
-    },
-    {
-      "epoch": 4.229960948508394,
-      "grad_norm": 0.003609878709539771,
-      "learning_rate": 0.00019999117786584166,
-      "loss": 46.0,
-      "step": 26266
-    },
-    {
-      "epoch": 4.230121985587181,
-      "grad_norm": 0.001283125951886177,
-      "learning_rate": 0.00019999117719382974,
-      "loss": 46.0,
-      "step": 26267
-    },
-    {
-      "epoch": 4.230283022665969,
-      "grad_norm": 0.00844255369156599,
-      "learning_rate": 0.00019999117652179222,
-      "loss": 46.0,
-      "step": 26268
-    },
-    {
-      "epoch": 4.230444059744756,
-      "grad_norm": 0.0014399484498426318,
-      "learning_rate": 0.00019999117584972912,
-      "loss": 46.0,
-      "step": 26269
-    },
-    {
-      "epoch": 4.230605096823544,
-      "grad_norm": 0.0013790338998660445,
-      "learning_rate": 0.00019999117517764044,
-      "loss": 46.0,
-      "step": 26270
-    },
-    {
-      "epoch": 4.230766133902331,
-      "grad_norm": 0.0026121370028704405,
-      "learning_rate": 0.00019999117450552616,
-      "loss": 46.0,
-      "step": 26271
-    },
-    {
-      "epoch": 4.2309271709811185,
-      "grad_norm": 0.008186489343643188,
-      "learning_rate": 0.00019999117383338625,
-      "loss": 46.0,
-      "step": 26272
-    },
-    {
-      "epoch": 4.231088208059906,
-      "grad_norm": 0.007265111431479454,
-      "learning_rate": 0.00019999117316122077,
-      "loss": 46.0,
-      "step": 26273
-    },
-    {
-      "epoch": 4.231249245138693,
-      "grad_norm": 0.0024882478173822165,
-      "learning_rate": 0.0001999911724890297,
-      "loss": 46.0,
-      "step": 26274
-    },
-    {
-      "epoch": 4.231410282217481,
-      "grad_norm": 0.005082158371806145,
-      "learning_rate": 0.00019999117181681309,
-      "loss": 46.0,
-      "step": 26275
-    },
-    {
-      "epoch": 4.231571319296268,
-      "grad_norm": 0.00590904988348484,
-      "learning_rate": 0.00019999117114457082,
-      "loss": 46.0,
-      "step": 26276
-    },
-    {
-      "epoch": 4.231732356375056,
-      "grad_norm": 0.00529545359313488,
-      "learning_rate": 0.00019999117047230297,
-      "loss": 46.0,
-      "step": 26277
-    },
-    {
-      "epoch": 4.231893393453843,
-      "grad_norm": 0.0015442437725141644,
-      "learning_rate": 0.00019999116980000956,
-      "loss": 46.0,
-      "step": 26278
-    },
-    {
-      "epoch": 4.23205443053263,
-      "grad_norm": 0.0037009220104664564,
-      "learning_rate": 0.0001999911691276905,
-      "loss": 46.0,
-      "step": 26279
-    },
-    {
-      "epoch": 4.232215467611417,
-      "grad_norm": 0.005697970744222403,
-      "learning_rate": 0.0001999911684553459,
-      "loss": 46.0,
-      "step": 26280
-    },
-    {
-      "epoch": 4.232376504690205,
-      "grad_norm": 0.0007133521139621735,
-      "learning_rate": 0.0001999911677829757,
-      "loss": 46.0,
-      "step": 26281
-    },
-    {
-      "epoch": 4.232537541768992,
-      "grad_norm": 0.005868204869329929,
-      "learning_rate": 0.00019999116711057988,
-      "loss": 46.0,
-      "step": 26282
-    },
-    {
-      "epoch": 4.23269857884778,
-      "grad_norm": 0.0053249988704919815,
-      "learning_rate": 0.0001999911664381585,
-      "loss": 46.0,
-      "step": 26283
-    },
-    {
-      "epoch": 4.232859615926567,
-      "grad_norm": 0.0036060435231775045,
-      "learning_rate": 0.00019999116576571152,
-      "loss": 46.0,
-      "step": 26284
-    },
-    {
-      "epoch": 4.2330206530053545,
-      "grad_norm": 0.005285861436277628,
-      "learning_rate": 0.0001999911650932389,
-      "loss": 46.0,
-      "step": 26285
-    },
-    {
-      "epoch": 4.233181690084142,
-      "grad_norm": 0.0017002781387418509,
-      "learning_rate": 0.00019999116442074075,
-      "loss": 46.0,
-      "step": 26286
-    },
-    {
-      "epoch": 4.233342727162929,
-      "grad_norm": 0.0013269941555336118,
-      "learning_rate": 0.00019999116374821697,
-      "loss": 46.0,
-      "step": 26287
-    },
-    {
-      "epoch": 4.233503764241717,
-      "grad_norm": 0.001127561554312706,
-      "learning_rate": 0.0001999911630756676,
-      "loss": 46.0,
-      "step": 26288
-    },
-    {
-      "epoch": 4.233664801320504,
-      "grad_norm": 0.003587308106943965,
-      "learning_rate": 0.00019999116240309266,
-      "loss": 46.0,
-      "step": 26289
-    },
-    {
-      "epoch": 4.233825838399292,
-      "grad_norm": 0.0031553911976516247,
-      "learning_rate": 0.0001999911617304921,
-      "loss": 46.0,
-      "step": 26290
-    },
-    {
-      "epoch": 4.233986875478079,
-      "grad_norm": 0.0026676699053496122,
-      "learning_rate": 0.00019999116105786597,
-      "loss": 46.0,
-      "step": 26291
-    },
-    {
-      "epoch": 4.234147912556867,
-      "grad_norm": 0.012290775775909424,
-      "learning_rate": 0.00019999116038521423,
-      "loss": 46.0,
-      "step": 26292
-    },
-    {
-      "epoch": 4.234308949635653,
-      "grad_norm": 0.013185273855924606,
-      "learning_rate": 0.00019999115971253693,
-      "loss": 46.0,
-      "step": 26293
-    },
-    {
-      "epoch": 4.234469986714441,
-      "grad_norm": 0.0036340963561087847,
-      "learning_rate": 0.000199991159039834,
-      "loss": 46.0,
-      "step": 26294
-    },
-    {
-      "epoch": 4.234631023793228,
-      "grad_norm": 0.002658189507201314,
-      "learning_rate": 0.00019999115836710548,
-      "loss": 46.0,
-      "step": 26295
-    },
-    {
-      "epoch": 4.2347920608720155,
-      "grad_norm": 0.004227503202855587,
-      "learning_rate": 0.0001999911576943514,
-      "loss": 46.0,
-      "step": 26296
-    },
-    {
-      "epoch": 4.234953097950803,
-      "grad_norm": 0.013710539788007736,
-      "learning_rate": 0.0001999911570215717,
-      "loss": 46.0,
-      "step": 26297
-    },
-    {
-      "epoch": 4.23511413502959,
-      "grad_norm": 0.008512726984918118,
-      "learning_rate": 0.00019999115634876643,
-      "loss": 46.0,
-      "step": 26298
-    },
-    {
-      "epoch": 4.235275172108378,
-      "grad_norm": 0.005582069046795368,
-      "learning_rate": 0.00019999115567593552,
-      "loss": 46.0,
-      "step": 26299
-    },
-    {
-      "epoch": 4.235436209187165,
-      "grad_norm": 0.007138532120734453,
-      "learning_rate": 0.00019999115500307906,
-      "loss": 46.0,
-      "step": 26300
-    },
-    {
-      "epoch": 4.235597246265953,
-      "grad_norm": 0.003104856237769127,
-      "learning_rate": 0.00019999115433019698,
-      "loss": 46.0,
-      "step": 26301
-    },
-    {
-      "epoch": 4.23575828334474,
-      "grad_norm": 0.0017345809610560536,
-      "learning_rate": 0.00019999115365728934,
-      "loss": 46.0,
-      "step": 26302
-    },
-    {
-      "epoch": 4.235919320423528,
-      "grad_norm": 0.005846432875841856,
-      "learning_rate": 0.0001999911529843561,
-      "loss": 46.0,
-      "step": 26303
-    },
-    {
-      "epoch": 4.236080357502315,
-      "grad_norm": 0.009179968386888504,
-      "learning_rate": 0.00019999115231139725,
-      "loss": 46.0,
-      "step": 26304
-    },
-    {
-      "epoch": 4.236241394581103,
-      "grad_norm": 0.004852697718888521,
-      "learning_rate": 0.0001999911516384128,
-      "loss": 46.0,
-      "step": 26305
-    },
-    {
-      "epoch": 4.23640243165989,
-      "grad_norm": 0.0015597400488331914,
-      "learning_rate": 0.0001999911509654028,
-      "loss": 46.0,
-      "step": 26306
-    },
-    {
-      "epoch": 4.2365634687386775,
-      "grad_norm": 0.004277390893548727,
-      "learning_rate": 0.00019999115029236718,
-      "loss": 46.0,
-      "step": 26307
-    },
-    {
-      "epoch": 4.236724505817464,
-      "grad_norm": 0.002068681875243783,
-      "learning_rate": 0.00019999114961930596,
-      "loss": 46.0,
-      "step": 26308
-    },
-    {
-      "epoch": 4.2368855428962515,
-      "grad_norm": 0.008431944064795971,
-      "learning_rate": 0.00019999114894621913,
-      "loss": 46.0,
-      "step": 26309
-    },
-    {
-      "epoch": 4.237046579975039,
-      "grad_norm": 0.0029291450046002865,
-      "learning_rate": 0.00019999114827310674,
-      "loss": 46.0,
-      "step": 26310
-    },
-    {
-      "epoch": 4.237207617053826,
-      "grad_norm": 0.0039008574094623327,
-      "learning_rate": 0.00019999114759996876,
-      "loss": 46.0,
-      "step": 26311
-    },
-    {
-      "epoch": 4.237368654132614,
-      "grad_norm": 0.0014049317687749863,
-      "learning_rate": 0.00019999114692680517,
-      "loss": 46.0,
-      "step": 26312
-    },
-    {
-      "epoch": 4.237529691211401,
-      "grad_norm": 0.0016699208645150065,
-      "learning_rate": 0.00019999114625361602,
-      "loss": 46.0,
-      "step": 26313
-    },
-    {
-      "epoch": 4.237690728290189,
-      "grad_norm": 0.00632148003205657,
-      "learning_rate": 0.00019999114558040125,
-      "loss": 46.0,
-      "step": 26314
-    },
-    {
-      "epoch": 4.237851765368976,
-      "grad_norm": 0.0028135075699537992,
-      "learning_rate": 0.00019999114490716087,
-      "loss": 46.0,
-      "step": 26315
-    },
-    {
-      "epoch": 4.238012802447764,
-      "grad_norm": 0.012456709519028664,
-      "learning_rate": 0.0001999911442338949,
-      "loss": 46.0,
-      "step": 26316
-    },
-    {
-      "epoch": 4.238173839526551,
-      "grad_norm": 0.0014022429240867496,
-      "learning_rate": 0.00019999114356060335,
-      "loss": 46.0,
-      "step": 26317
-    },
-    {
-      "epoch": 4.238334876605339,
-      "grad_norm": 0.0030872586648911238,
-      "learning_rate": 0.0001999911428872862,
-      "loss": 46.0,
-      "step": 26318
-    },
-    {
-      "epoch": 4.238495913684126,
-      "grad_norm": 0.002695930888876319,
-      "learning_rate": 0.0001999911422139435,
-      "loss": 46.0,
-      "step": 26319
-    },
-    {
-      "epoch": 4.2386569507629135,
-      "grad_norm": 0.005902216769754887,
-      "learning_rate": 0.00019999114154057516,
-      "loss": 46.0,
-      "step": 26320
-    },
-    {
-      "epoch": 4.238817987841701,
-      "grad_norm": 0.002503668423742056,
-      "learning_rate": 0.00019999114086718126,
-      "loss": 46.0,
-      "step": 26321
-    },
-    {
-      "epoch": 4.2389790249204875,
-      "grad_norm": 0.006871354300528765,
-      "learning_rate": 0.00019999114019376174,
-      "loss": 46.0,
-      "step": 26322
-    },
-    {
-      "epoch": 4.239140061999275,
-      "grad_norm": 0.001992875011637807,
-      "learning_rate": 0.00019999113952031666,
-      "loss": 46.0,
-      "step": 26323
-    },
-    {
-      "epoch": 4.239301099078062,
-      "grad_norm": 0.008072340860962868,
-      "learning_rate": 0.00019999113884684597,
-      "loss": 46.0,
-      "step": 26324
-    },
-    {
-      "epoch": 4.23946213615685,
-      "grad_norm": 0.002895774319767952,
-      "learning_rate": 0.00019999113817334966,
-      "loss": 46.0,
-      "step": 26325
-    },
-    {
-      "epoch": 4.239623173235637,
-      "grad_norm": 0.008577320724725723,
-      "learning_rate": 0.0001999911374998278,
-      "loss": 46.0,
-      "step": 26326
-    },
-    {
-      "epoch": 4.239784210314425,
-      "grad_norm": 0.00791993085294962,
-      "learning_rate": 0.0001999911368262803,
-      "loss": 46.0,
-      "step": 26327
-    },
-    {
-      "epoch": 4.239945247393212,
-      "grad_norm": 0.0024015309754759073,
-      "learning_rate": 0.00019999113615270725,
-      "loss": 46.0,
-      "step": 26328
-    },
-    {
-      "epoch": 4.240106284472,
-      "grad_norm": 0.003368641948327422,
-      "learning_rate": 0.0001999911354791086,
-      "loss": 46.0,
-      "step": 26329
-    },
-    {
-      "epoch": 4.240267321550787,
-      "grad_norm": 0.005208450369536877,
-      "learning_rate": 0.00019999113480548435,
-      "loss": 46.0,
-      "step": 26330
-    },
-    {
-      "epoch": 4.240428358629575,
-      "grad_norm": 0.004403963685035706,
-      "learning_rate": 0.0001999911341318345,
-      "loss": 46.0,
-      "step": 26331
-    },
-    {
-      "epoch": 4.240589395708362,
-      "grad_norm": 0.01280742883682251,
-      "learning_rate": 0.00019999113345815905,
-      "loss": 46.0,
-      "step": 26332
-    },
-    {
-      "epoch": 4.2407504327871495,
-      "grad_norm": 0.014430827461183071,
-      "learning_rate": 0.00019999113278445805,
-      "loss": 46.0,
-      "step": 26333
-    },
-    {
-      "epoch": 4.240911469865937,
-      "grad_norm": 0.004967774264514446,
-      "learning_rate": 0.00019999113211073143,
-      "loss": 46.0,
-      "step": 26334
-    },
-    {
-      "epoch": 4.241072506944724,
-      "grad_norm": 0.0009713007602840662,
-      "learning_rate": 0.0001999911314369792,
-      "loss": 46.0,
-      "step": 26335
-    },
-    {
-      "epoch": 4.241233544023512,
-      "grad_norm": 0.004461093805730343,
-      "learning_rate": 0.0001999911307632014,
-      "loss": 46.0,
-      "step": 26336
-    },
-    {
-      "epoch": 4.241394581102298,
-      "grad_norm": 0.006361424457281828,
-      "learning_rate": 0.000199991130089398,
-      "loss": 46.0,
-      "step": 26337
-    },
-    {
-      "epoch": 4.241555618181086,
-      "grad_norm": 0.005571140442043543,
-      "learning_rate": 0.000199991129415569,
-      "loss": 46.0,
-      "step": 26338
-    },
-    {
-      "epoch": 4.241716655259873,
-      "grad_norm": 0.005587654188275337,
-      "learning_rate": 0.00019999112874171445,
-      "loss": 46.0,
-      "step": 26339
-    },
-    {
-      "epoch": 4.241877692338661,
-      "grad_norm": 0.007114875130355358,
-      "learning_rate": 0.00019999112806783425,
-      "loss": 46.0,
-      "step": 26340
-    },
-    {
-      "epoch": 4.242038729417448,
-      "grad_norm": 0.002650391310453415,
-      "learning_rate": 0.0001999911273939285,
-      "loss": 46.0,
-      "step": 26341
-    },
-    {
-      "epoch": 4.242199766496236,
-      "grad_norm": 0.0023981230333447456,
-      "learning_rate": 0.0001999911267199971,
-      "loss": 46.0,
-      "step": 26342
-    },
-    {
-      "epoch": 4.242360803575023,
-      "grad_norm": 0.003958090674132109,
-      "learning_rate": 0.00019999112604604017,
-      "loss": 46.0,
-      "step": 26343
-    },
-    {
-      "epoch": 4.2425218406538105,
-      "grad_norm": 0.0031815425027161837,
-      "learning_rate": 0.00019999112537205763,
-      "loss": 46.0,
-      "step": 26344
-    },
-    {
-      "epoch": 4.242682877732598,
-      "grad_norm": 0.003761272644624114,
-      "learning_rate": 0.00019999112469804947,
-      "loss": 46.0,
-      "step": 26345
-    },
-    {
-      "epoch": 4.2428439148113855,
-      "grad_norm": 0.003021727781742811,
-      "learning_rate": 0.00019999112402401575,
-      "loss": 46.0,
-      "step": 26346
-    },
-    {
-      "epoch": 4.243004951890173,
-      "grad_norm": 0.0035204461310058832,
-      "learning_rate": 0.00019999112334995642,
-      "loss": 46.0,
-      "step": 26347
-    },
-    {
-      "epoch": 4.24316598896896,
-      "grad_norm": 0.003963842056691647,
-      "learning_rate": 0.0001999911226758715,
-      "loss": 46.0,
-      "step": 26348
-    },
-    {
-      "epoch": 4.243327026047748,
-      "grad_norm": 0.006535484455525875,
-      "learning_rate": 0.000199991122001761,
-      "loss": 46.0,
-      "step": 26349
-    },
-    {
-      "epoch": 4.243488063126535,
-      "grad_norm": 0.0019848307128995657,
-      "learning_rate": 0.00019999112132762493,
-      "loss": 46.0,
-      "step": 26350
-    },
-    {
-      "epoch": 4.243649100205323,
-      "grad_norm": 0.014494430273771286,
-      "learning_rate": 0.00019999112065346322,
-      "loss": 46.0,
-      "step": 26351
-    },
-    {
-      "epoch": 4.243810137284109,
-      "grad_norm": 0.0017849645810201764,
-      "learning_rate": 0.00019999111997927592,
-      "loss": 46.0,
-      "step": 26352
-    },
-    {
-      "epoch": 4.243971174362897,
-      "grad_norm": 0.004987181629985571,
-      "learning_rate": 0.00019999111930506304,
-      "loss": 46.0,
-      "step": 26353
-    },
-    {
-      "epoch": 4.244132211441684,
-      "grad_norm": 0.0034145419485867023,
-      "learning_rate": 0.00019999111863082457,
-      "loss": 46.0,
-      "step": 26354
-    },
-    {
-      "epoch": 4.244293248520472,
-      "grad_norm": 0.002710689790546894,
-      "learning_rate": 0.0001999911179565605,
-      "loss": 46.0,
-      "step": 26355
-    },
-    {
-      "epoch": 4.244454285599259,
-      "grad_norm": 0.004202722106128931,
-      "learning_rate": 0.00019999111728227084,
-      "loss": 46.0,
-      "step": 26356
-    },
-    {
-      "epoch": 4.2446153226780465,
-      "grad_norm": 0.006209149956703186,
-      "learning_rate": 0.0001999911166079556,
-      "loss": 46.0,
-      "step": 26357
-    },
-    {
-      "epoch": 4.244776359756834,
-      "grad_norm": 0.0010890814010053873,
-      "learning_rate": 0.00019999111593361474,
-      "loss": 46.0,
-      "step": 26358
-    },
-    {
-      "epoch": 4.244937396835621,
-      "grad_norm": 0.00703685637563467,
-      "learning_rate": 0.00019999111525924833,
-      "loss": 46.0,
-      "step": 26359
-    },
-    {
-      "epoch": 4.245098433914409,
-      "grad_norm": 0.00873405672609806,
-      "learning_rate": 0.00019999111458485628,
-      "loss": 46.0,
-      "step": 26360
-    },
-    {
-      "epoch": 4.245259470993196,
-      "grad_norm": 0.0009693785104900599,
-      "learning_rate": 0.00019999111391043868,
-      "loss": 46.0,
-      "step": 26361
-    },
-    {
-      "epoch": 4.245420508071984,
-      "grad_norm": 0.008365047164261341,
-      "learning_rate": 0.00019999111323599545,
-      "loss": 46.0,
-      "step": 26362
-    },
-    {
-      "epoch": 4.245581545150771,
-      "grad_norm": 0.004657451994717121,
-      "learning_rate": 0.00019999111256152664,
-      "loss": 46.0,
-      "step": 26363
-    },
-    {
-      "epoch": 4.245742582229559,
-      "grad_norm": 0.007005263119935989,
-      "learning_rate": 0.00019999111188703225,
-      "loss": 46.0,
-      "step": 26364
-    },
-    {
-      "epoch": 4.245903619308346,
-      "grad_norm": 0.008059971034526825,
-      "learning_rate": 0.00019999111121251227,
-      "loss": 46.0,
-      "step": 26365
-    },
-    {
-      "epoch": 4.246064656387133,
-      "grad_norm": 0.009876109659671783,
-      "learning_rate": 0.0001999911105379667,
-      "loss": 46.0,
-      "step": 26366
-    },
-    {
-      "epoch": 4.24622569346592,
-      "grad_norm": 0.001157525461167097,
-      "learning_rate": 0.0001999911098633955,
-      "loss": 46.0,
-      "step": 26367
-    },
-    {
-      "epoch": 4.246386730544708,
-      "grad_norm": 0.004886473994702101,
-      "learning_rate": 0.00019999110918879877,
-      "loss": 46.0,
-      "step": 26368
-    },
-    {
-      "epoch": 4.246547767623495,
-      "grad_norm": 0.0037689292803406715,
-      "learning_rate": 0.00019999110851417638,
-      "loss": 46.0,
-      "step": 26369
-    },
-    {
-      "epoch": 4.2467088047022825,
-      "grad_norm": 0.006638484075665474,
-      "learning_rate": 0.0001999911078395284,
-      "loss": 46.0,
-      "step": 26370
-    },
-    {
-      "epoch": 4.24686984178107,
-      "grad_norm": 0.0046528177335858345,
-      "learning_rate": 0.00019999110716485487,
-      "loss": 46.0,
-      "step": 26371
-    },
-    {
-      "epoch": 4.247030878859857,
-      "grad_norm": 0.010144357569515705,
-      "learning_rate": 0.00019999110649015573,
-      "loss": 46.0,
-      "step": 26372
-    },
-    {
-      "epoch": 4.247191915938645,
-      "grad_norm": 0.0026838178746402264,
-      "learning_rate": 0.000199991105815431,
-      "loss": 46.0,
-      "step": 26373
-    },
-    {
-      "epoch": 4.247352953017432,
-      "grad_norm": 0.003250395180657506,
-      "learning_rate": 0.0001999911051406807,
-      "loss": 46.0,
-      "step": 26374
-    },
-    {
-      "epoch": 4.24751399009622,
-      "grad_norm": 0.010543163865804672,
-      "learning_rate": 0.00019999110446590476,
-      "loss": 46.0,
-      "step": 26375
-    },
-    {
-      "epoch": 4.247675027175007,
-      "grad_norm": 0.002511736936867237,
-      "learning_rate": 0.00019999110379110327,
-      "loss": 46.0,
-      "step": 26376
-    },
-    {
-      "epoch": 4.247836064253795,
-      "grad_norm": 0.00756703270599246,
-      "learning_rate": 0.00019999110311627615,
-      "loss": 46.0,
-      "step": 26377
-    },
-    {
-      "epoch": 4.247997101332582,
-      "grad_norm": 0.015099585056304932,
-      "learning_rate": 0.00019999110244142348,
-      "loss": 46.0,
-      "step": 26378
-    },
-    {
-      "epoch": 4.24815813841137,
-      "grad_norm": 0.008893853984773159,
-      "learning_rate": 0.00019999110176654517,
-      "loss": 46.0,
-      "step": 26379
-    },
-    {
-      "epoch": 4.248319175490157,
-      "grad_norm": 0.008545872755348682,
-      "learning_rate": 0.0001999911010916413,
-      "loss": 46.0,
-      "step": 26380
-    },
-    {
-      "epoch": 4.248480212568944,
-      "grad_norm": 0.0007415362633764744,
-      "learning_rate": 0.00019999110041671184,
-      "loss": 46.0,
-      "step": 26381
-    },
-    {
-      "epoch": 4.248641249647731,
-      "grad_norm": 0.004860542248934507,
-      "learning_rate": 0.00019999109974175677,
-      "loss": 46.0,
-      "step": 26382
-    },
-    {
-      "epoch": 4.2488022867265185,
-      "grad_norm": 0.007643924560397863,
-      "learning_rate": 0.0001999910990667761,
-      "loss": 46.0,
-      "step": 26383
-    },
-    {
-      "epoch": 4.248963323805306,
-      "grad_norm": 0.0016655635554343462,
-      "learning_rate": 0.00019999109839176983,
-      "loss": 46.0,
-      "step": 26384
-    },
-    {
-      "epoch": 4.249124360884093,
-      "grad_norm": 0.0036033783107995987,
-      "learning_rate": 0.000199991097716738,
-      "loss": 46.0,
-      "step": 26385
-    },
-    {
-      "epoch": 4.249285397962881,
-      "grad_norm": 0.005626278929412365,
-      "learning_rate": 0.00019999109704168058,
-      "loss": 46.0,
-      "step": 26386
-    },
-    {
-      "epoch": 4.249446435041668,
-      "grad_norm": 0.004101132042706013,
-      "learning_rate": 0.00019999109636659754,
-      "loss": 46.0,
-      "step": 26387
-    },
-    {
-      "epoch": 4.249607472120456,
-      "grad_norm": 0.003996509127318859,
-      "learning_rate": 0.00019999109569148892,
-      "loss": 46.0,
-      "step": 26388
-    },
-    {
-      "epoch": 4.249768509199243,
-      "grad_norm": 0.0023707302752882242,
-      "learning_rate": 0.0001999910950163547,
-      "loss": 46.0,
-      "step": 26389
-    },
-    {
-      "epoch": 4.249929546278031,
-      "grad_norm": 0.00349828670732677,
-      "learning_rate": 0.0001999910943411949,
-      "loss": 46.0,
-      "step": 26390
-    },
-    {
-      "epoch": 4.250090583356818,
-      "grad_norm": 0.00982285849750042,
-      "learning_rate": 0.00019999109366600952,
-      "loss": 46.0,
-      "step": 26391
-    },
-    {
-      "epoch": 4.250251620435606,
-      "grad_norm": 0.008573968894779682,
-      "learning_rate": 0.00019999109299079852,
-      "loss": 46.0,
-      "step": 26392
-    },
-    {
-      "epoch": 4.250412657514393,
-      "grad_norm": 0.013444820418953896,
-      "learning_rate": 0.00019999109231556194,
-      "loss": 46.0,
-      "step": 26393
-    },
-    {
-      "epoch": 4.2505736945931805,
-      "grad_norm": 0.011791357770562172,
-      "learning_rate": 0.00019999109164029974,
-      "loss": 46.0,
-      "step": 26394
-    },
-    {
-      "epoch": 4.250734731671967,
-      "grad_norm": 0.0014449275331571698,
-      "learning_rate": 0.00019999109096501197,
-      "loss": 46.0,
-      "step": 26395
-    },
-    {
-      "epoch": 4.2508957687507545,
-      "grad_norm": 0.0008376318146474659,
-      "learning_rate": 0.00019999109028969863,
-      "loss": 46.0,
-      "step": 26396
-    },
-    {
-      "epoch": 4.251056805829542,
-      "grad_norm": 0.0035114444326609373,
-      "learning_rate": 0.00019999108961435966,
-      "loss": 46.0,
-      "step": 26397
-    },
-    {
-      "epoch": 4.251217842908329,
-      "grad_norm": 0.003312203101813793,
-      "learning_rate": 0.00019999108893899514,
-      "loss": 46.0,
-      "step": 26398
-    },
-    {
-      "epoch": 4.251378879987117,
-      "grad_norm": 0.015752699226140976,
-      "learning_rate": 0.000199991088263605,
-      "loss": 46.0,
-      "step": 26399
-    },
-    {
-      "epoch": 4.251539917065904,
-      "grad_norm": 0.007732367608696222,
-      "learning_rate": 0.00019999108758818925,
-      "loss": 46.0,
-      "step": 26400
-    },
-    {
-      "epoch": 4.251700954144692,
-      "grad_norm": 0.007054062094539404,
-      "learning_rate": 0.00019999108691274792,
-      "loss": 46.0,
-      "step": 26401
-    },
-    {
-      "epoch": 4.251861991223479,
-      "grad_norm": 0.0006912920507602394,
-      "learning_rate": 0.000199991086237281,
-      "loss": 46.0,
-      "step": 26402
-    },
-    {
-      "epoch": 4.252023028302267,
-      "grad_norm": 0.007961279712617397,
-      "learning_rate": 0.0001999910855617885,
-      "loss": 46.0,
-      "step": 26403
-    },
-    {
-      "epoch": 4.252184065381054,
-      "grad_norm": 0.004229597747325897,
-      "learning_rate": 0.0001999910848862704,
-      "loss": 46.0,
-      "step": 26404
-    },
-    {
-      "epoch": 4.2523451024598415,
-      "grad_norm": 0.017234614118933678,
-      "learning_rate": 0.0001999910842107267,
-      "loss": 46.0,
-      "step": 26405
-    },
-    {
-      "epoch": 4.252506139538629,
-      "grad_norm": 0.014713821932673454,
-      "learning_rate": 0.00019999108353515742,
-      "loss": 46.0,
-      "step": 26406
-    },
-    {
-      "epoch": 4.2526671766174164,
-      "grad_norm": 0.008469310589134693,
-      "learning_rate": 0.00019999108285956253,
-      "loss": 46.0,
-      "step": 26407
-    },
-    {
-      "epoch": 4.252828213696204,
-      "grad_norm": 0.005367763806134462,
-      "learning_rate": 0.00019999108218394208,
-      "loss": 46.0,
-      "step": 26408
-    },
-    {
-      "epoch": 4.252989250774991,
-      "grad_norm": 0.0026243410538882017,
-      "learning_rate": 0.00019999108150829602,
-      "loss": 46.0,
-      "step": 26409
-    },
-    {
-      "epoch": 4.253150287853778,
-      "grad_norm": 0.0024996225256472826,
-      "learning_rate": 0.00019999108083262434,
-      "loss": 46.0,
-      "step": 26410
-    },
-    {
-      "epoch": 4.253311324932565,
-      "grad_norm": 0.0036348735447973013,
-      "learning_rate": 0.00019999108015692708,
-      "loss": 46.0,
-      "step": 26411
-    },
-    {
-      "epoch": 4.253472362011353,
-      "grad_norm": 0.016380146145820618,
-      "learning_rate": 0.00019999107948120426,
-      "loss": 46.0,
-      "step": 26412
-    },
-    {
-      "epoch": 4.25363339909014,
-      "grad_norm": 0.0023258281871676445,
-      "learning_rate": 0.00019999107880545582,
-      "loss": 46.0,
-      "step": 26413
-    },
-    {
-      "epoch": 4.253794436168928,
-      "grad_norm": 0.0019040658371523023,
-      "learning_rate": 0.0001999910781296818,
-      "loss": 46.0,
-      "step": 26414
-    },
-    {
-      "epoch": 4.253955473247715,
-      "grad_norm": 0.012818894349038601,
-      "learning_rate": 0.00019999107745388218,
-      "loss": 46.0,
-      "step": 26415
-    },
-    {
-      "epoch": 4.254116510326503,
-      "grad_norm": 0.006248683203011751,
-      "learning_rate": 0.00019999107677805698,
-      "loss": 46.0,
-      "step": 26416
-    },
-    {
-      "epoch": 4.25427754740529,
-      "grad_norm": 0.004687626846134663,
-      "learning_rate": 0.00019999107610220616,
-      "loss": 46.0,
-      "step": 26417
-    },
-    {
-      "epoch": 4.2544385844840775,
-      "grad_norm": 0.006617632694542408,
-      "learning_rate": 0.00019999107542632976,
-      "loss": 46.0,
-      "step": 26418
-    },
-    {
-      "epoch": 4.254599621562865,
-      "grad_norm": 0.001413870370015502,
-      "learning_rate": 0.00019999107475042775,
-      "loss": 46.0,
-      "step": 26419
-    },
-    {
-      "epoch": 4.254760658641652,
-      "grad_norm": 0.013979134149849415,
-      "learning_rate": 0.00019999107407450017,
-      "loss": 46.0,
-      "step": 26420
-    },
-    {
-      "epoch": 4.25492169572044,
-      "grad_norm": 0.010164640843868256,
-      "learning_rate": 0.000199991073398547,
-      "loss": 46.0,
-      "step": 26421
-    },
-    {
-      "epoch": 4.255082732799227,
-      "grad_norm": 0.003486220259219408,
-      "learning_rate": 0.00019999107272256823,
-      "loss": 46.0,
-      "step": 26422
-    },
-    {
-      "epoch": 4.255243769878015,
-      "grad_norm": 0.018868697807192802,
-      "learning_rate": 0.00019999107204656387,
-      "loss": 46.0,
-      "step": 26423
-    },
-    {
-      "epoch": 4.255404806956802,
-      "grad_norm": 0.010871385224163532,
-      "learning_rate": 0.00019999107137053392,
-      "loss": 46.0,
-      "step": 26424
-    },
-    {
-      "epoch": 4.255565844035589,
-      "grad_norm": 0.007027751766145229,
-      "learning_rate": 0.00019999107069447838,
-      "loss": 46.0,
-      "step": 26425
-    },
-    {
-      "epoch": 4.255726881114376,
-      "grad_norm": 0.0068663256242871284,
-      "learning_rate": 0.00019999107001839725,
-      "loss": 46.0,
-      "step": 26426
-    },
-    {
-      "epoch": 4.255887918193164,
-      "grad_norm": 0.0010301386937499046,
-      "learning_rate": 0.0001999910693422905,
-      "loss": 46.0,
-      "step": 26427
-    },
-    {
-      "epoch": 4.256048955271951,
-      "grad_norm": 0.005545147228986025,
-      "learning_rate": 0.00019999106866615818,
-      "loss": 46.0,
-      "step": 26428
-    },
-    {
-      "epoch": 4.256209992350739,
-      "grad_norm": 0.00392505619674921,
-      "learning_rate": 0.00019999106799000027,
-      "loss": 46.0,
-      "step": 26429
-    },
-    {
-      "epoch": 4.256371029429526,
-      "grad_norm": 0.0008495452348142862,
-      "learning_rate": 0.00019999106731381677,
-      "loss": 46.0,
-      "step": 26430
-    },
-    {
-      "epoch": 4.2565320665083135,
-      "grad_norm": 0.005274081602692604,
-      "learning_rate": 0.00019999106663760765,
-      "loss": 46.0,
-      "step": 26431
-    },
-    {
-      "epoch": 4.256693103587101,
-      "grad_norm": 0.003307618200778961,
-      "learning_rate": 0.00019999106596137298,
-      "loss": 46.0,
-      "step": 26432
-    },
-    {
-      "epoch": 4.256854140665888,
-      "grad_norm": 0.004555467516183853,
-      "learning_rate": 0.0001999910652851127,
-      "loss": 46.0,
-      "step": 26433
-    },
-    {
-      "epoch": 4.257015177744676,
-      "grad_norm": 0.004250813741236925,
-      "learning_rate": 0.00019999106460882678,
-      "loss": 46.0,
-      "step": 26434
-    },
-    {
-      "epoch": 4.257176214823463,
-      "grad_norm": 0.0038257124833762646,
-      "learning_rate": 0.00019999106393251532,
-      "loss": 46.0,
-      "step": 26435
-    },
-    {
-      "epoch": 4.257337251902251,
-      "grad_norm": 0.0037351020146161318,
-      "learning_rate": 0.00019999106325617824,
-      "loss": 46.0,
-      "step": 26436
-    },
-    {
-      "epoch": 4.257498288981038,
-      "grad_norm": 0.004851972684264183,
-      "learning_rate": 0.0001999910625798156,
-      "loss": 46.0,
-      "step": 26437
-    },
-    {
-      "epoch": 4.257659326059826,
-      "grad_norm": 0.004382154904305935,
-      "learning_rate": 0.00019999106190342735,
-      "loss": 46.0,
-      "step": 26438
-    },
-    {
-      "epoch": 4.257820363138613,
-      "grad_norm": 0.0042005497962236404,
-      "learning_rate": 0.0001999910612270135,
-      "loss": 46.0,
-      "step": 26439
-    },
-    {
-      "epoch": 4.2579814002174,
-      "grad_norm": 0.007045915815979242,
-      "learning_rate": 0.00019999106055057408,
-      "loss": 46.0,
-      "step": 26440
-    },
-    {
-      "epoch": 4.258142437296187,
-      "grad_norm": 0.005091259256005287,
-      "learning_rate": 0.00019999105987410904,
-      "loss": 46.0,
-      "step": 26441
-    },
-    {
-      "epoch": 4.258303474374975,
-      "grad_norm": 0.0036288921255618334,
-      "learning_rate": 0.00019999105919761844,
-      "loss": 46.0,
-      "step": 26442
-    },
-    {
-      "epoch": 4.258464511453762,
-      "grad_norm": 0.004293323494493961,
-      "learning_rate": 0.00019999105852110222,
-      "loss": 46.0,
-      "step": 26443
-    },
-    {
-      "epoch": 4.2586255485325495,
-      "grad_norm": 0.0037275587674230337,
-      "learning_rate": 0.00019999105784456042,
-      "loss": 46.0,
-      "step": 26444
-    },
-    {
-      "epoch": 4.258786585611337,
-      "grad_norm": 0.002928992733359337,
-      "learning_rate": 0.000199991057167993,
-      "loss": 46.0,
-      "step": 26445
-    },
-    {
-      "epoch": 4.258947622690124,
-      "grad_norm": 0.007753681857138872,
-      "learning_rate": 0.00019999105649140005,
-      "loss": 46.0,
-      "step": 26446
-    },
-    {
-      "epoch": 4.259108659768912,
-      "grad_norm": 0.005893183872103691,
-      "learning_rate": 0.00019999105581478146,
-      "loss": 46.0,
-      "step": 26447
-    },
-    {
-      "epoch": 4.259269696847699,
-      "grad_norm": 0.0024961617309600115,
-      "learning_rate": 0.00019999105513813725,
-      "loss": 46.0,
-      "step": 26448
-    },
-    {
-      "epoch": 4.259430733926487,
-      "grad_norm": 0.0051794699393212795,
-      "learning_rate": 0.0001999910544614675,
-      "loss": 46.0,
-      "step": 26449
-    },
-    {
-      "epoch": 4.259591771005274,
-      "grad_norm": 0.004589755553752184,
-      "learning_rate": 0.00019999105378477213,
-      "loss": 46.0,
-      "step": 26450
-    },
-    {
-      "epoch": 4.259752808084062,
-      "grad_norm": 0.0026035637129098177,
-      "learning_rate": 0.0001999910531080512,
-      "loss": 46.0,
-      "step": 26451
-    },
-    {
-      "epoch": 4.259913845162849,
-      "grad_norm": 0.0014852768508717418,
-      "learning_rate": 0.00019999105243130464,
-      "loss": 46.0,
-      "step": 26452
-    },
-    {
-      "epoch": 4.2600748822416366,
-      "grad_norm": 0.007767904549837112,
-      "learning_rate": 0.0001999910517545325,
-      "loss": 46.0,
-      "step": 26453
-    },
-    {
-      "epoch": 4.260235919320423,
-      "grad_norm": 0.0043710567988455296,
-      "learning_rate": 0.00019999105107773476,
-      "loss": 46.0,
-      "step": 26454
-    },
-    {
-      "epoch": 4.260396956399211,
-      "grad_norm": 0.0039058702532202005,
-      "learning_rate": 0.00019999105040091145,
-      "loss": 46.0,
-      "step": 26455
-    },
-    {
-      "epoch": 4.260557993477998,
-      "grad_norm": 0.003990390338003635,
-      "learning_rate": 0.00019999104972406252,
-      "loss": 46.0,
-      "step": 26456
-    },
-    {
-      "epoch": 4.2607190305567855,
-      "grad_norm": 0.0023479436058551073,
-      "learning_rate": 0.00019999104904718803,
-      "loss": 46.0,
-      "step": 26457
-    },
-    {
-      "epoch": 4.260880067635573,
-      "grad_norm": 0.005221907049417496,
-      "learning_rate": 0.00019999104837028792,
-      "loss": 46.0,
-      "step": 26458
-    },
-    {
-      "epoch": 4.26104110471436,
-      "grad_norm": 0.007145897950977087,
-      "learning_rate": 0.00019999104769336223,
-      "loss": 46.0,
-      "step": 26459
-    },
-    {
-      "epoch": 4.261202141793148,
-      "grad_norm": 0.015062019228935242,
-      "learning_rate": 0.00019999104701641095,
-      "loss": 46.0,
-      "step": 26460
-    },
-    {
-      "epoch": 4.261363178871935,
-      "grad_norm": 0.004843233153223991,
-      "learning_rate": 0.00019999104633943406,
-      "loss": 46.0,
-      "step": 26461
-    },
-    {
-      "epoch": 4.261524215950723,
-      "grad_norm": 0.0019011307740584016,
-      "learning_rate": 0.00019999104566243158,
-      "loss": 46.0,
-      "step": 26462
-    },
-    {
-      "epoch": 4.26168525302951,
-      "grad_norm": 0.002416796749457717,
-      "learning_rate": 0.00019999104498540354,
-      "loss": 46.0,
-      "step": 26463
-    },
-    {
-      "epoch": 4.261846290108298,
-      "grad_norm": 0.004971110727638006,
-      "learning_rate": 0.00019999104430834988,
-      "loss": 46.0,
-      "step": 26464
-    },
-    {
-      "epoch": 4.262007327187085,
-      "grad_norm": 0.008334905840456486,
-      "learning_rate": 0.00019999104363127064,
-      "loss": 46.0,
-      "step": 26465
-    },
-    {
-      "epoch": 4.2621683642658725,
-      "grad_norm": 0.0036073343362659216,
-      "learning_rate": 0.00019999104295416578,
-      "loss": 46.0,
-      "step": 26466
-    },
-    {
-      "epoch": 4.26232940134466,
-      "grad_norm": 0.002912555355578661,
-      "learning_rate": 0.00019999104227703534,
-      "loss": 46.0,
-      "step": 26467
-    },
-    {
-      "epoch": 4.2624904384234465,
-      "grad_norm": 0.003374774008989334,
-      "learning_rate": 0.0001999910415998793,
-      "loss": 46.0,
-      "step": 26468
-    },
-    {
-      "epoch": 4.262651475502234,
-      "grad_norm": 0.0019413164118304849,
-      "learning_rate": 0.0001999910409226977,
-      "loss": 46.0,
-      "step": 26469
-    },
-    {
-      "epoch": 4.2628125125810215,
-      "grad_norm": 0.00401551416143775,
-      "learning_rate": 0.00019999104024549048,
-      "loss": 46.0,
-      "step": 26470
-    },
-    {
-      "epoch": 4.262973549659809,
-      "grad_norm": 0.004251905716955662,
-      "learning_rate": 0.0001999910395682577,
-      "loss": 46.0,
-      "step": 26471
-    },
-    {
-      "epoch": 4.263134586738596,
-      "grad_norm": 0.0020622832234948874,
-      "learning_rate": 0.00019999103889099928,
-      "loss": 46.0,
-      "step": 26472
-    },
-    {
-      "epoch": 4.263295623817384,
-      "grad_norm": 0.0013198304222896695,
-      "learning_rate": 0.0001999910382137153,
-      "loss": 46.0,
-      "step": 26473
-    },
-    {
-      "epoch": 4.263456660896171,
-      "grad_norm": 0.005101666320115328,
-      "learning_rate": 0.0001999910375364057,
-      "loss": 46.0,
-      "step": 26474
-    },
-    {
-      "epoch": 4.263617697974959,
-      "grad_norm": 0.013275950215756893,
-      "learning_rate": 0.00019999103685907054,
-      "loss": 46.0,
-      "step": 26475
-    },
-    {
-      "epoch": 4.263778735053746,
-      "grad_norm": 0.010259171016514301,
-      "learning_rate": 0.00019999103618170976,
-      "loss": 46.0,
-      "step": 26476
-    },
-    {
-      "epoch": 4.263939772132534,
-      "grad_norm": 0.001979172695428133,
-      "learning_rate": 0.00019999103550432342,
-      "loss": 46.0,
-      "step": 26477
-    },
-    {
-      "epoch": 4.264100809211321,
-      "grad_norm": 0.009595053270459175,
-      "learning_rate": 0.00019999103482691143,
-      "loss": 46.0,
-      "step": 26478
-    },
-    {
-      "epoch": 4.2642618462901085,
-      "grad_norm": 0.002920767292380333,
-      "learning_rate": 0.00019999103414947392,
-      "loss": 46.0,
-      "step": 26479
-    },
-    {
-      "epoch": 4.264422883368896,
-      "grad_norm": 0.000645426451228559,
-      "learning_rate": 0.00019999103347201076,
-      "loss": 46.0,
-      "step": 26480
-    },
-    {
-      "epoch": 4.264583920447683,
-      "grad_norm": 0.002456387970596552,
-      "learning_rate": 0.00019999103279452204,
-      "loss": 46.0,
-      "step": 26481
-    },
-    {
-      "epoch": 4.264744957526471,
-      "grad_norm": 0.006661789491772652,
-      "learning_rate": 0.0001999910321170077,
-      "loss": 46.0,
-      "step": 26482
-    },
-    {
-      "epoch": 4.264905994605257,
-      "grad_norm": 0.02166179195046425,
-      "learning_rate": 0.0001999910314394678,
-      "loss": 46.0,
-      "step": 26483
-    },
-    {
-      "epoch": 4.265067031684045,
-      "grad_norm": 0.004659612663090229,
-      "learning_rate": 0.0001999910307619023,
-      "loss": 46.0,
-      "step": 26484
-    },
-    {
-      "epoch": 4.265228068762832,
-      "grad_norm": 0.004583099391311407,
-      "learning_rate": 0.00019999103008431118,
-      "loss": 46.0,
-      "step": 26485
-    },
-    {
-      "epoch": 4.26538910584162,
-      "grad_norm": 0.007043486926704645,
-      "learning_rate": 0.0001999910294066945,
-      "loss": 46.0,
-      "step": 26486
-    },
-    {
-      "epoch": 4.265550142920407,
-      "grad_norm": 0.006036832928657532,
-      "learning_rate": 0.0001999910287290522,
-      "loss": 46.0,
-      "step": 26487
-    },
-    {
-      "epoch": 4.265711179999195,
-      "grad_norm": 0.010575943626463413,
-      "learning_rate": 0.00019999102805138433,
-      "loss": 46.0,
-      "step": 26488
-    },
-    {
-      "epoch": 4.265872217077982,
-      "grad_norm": 0.011251966468989849,
-      "learning_rate": 0.00019999102737369086,
-      "loss": 46.0,
-      "step": 26489
-    },
-    {
-      "epoch": 4.26603325415677,
-      "grad_norm": 0.002662515500560403,
-      "learning_rate": 0.0001999910266959718,
-      "loss": 46.0,
-      "step": 26490
-    },
-    {
-      "epoch": 4.266194291235557,
-      "grad_norm": 0.0024956930428743362,
-      "learning_rate": 0.00019999102601822715,
-      "loss": 46.0,
-      "step": 26491
-    },
-    {
-      "epoch": 4.2663553283143445,
-      "grad_norm": 0.001442663837224245,
-      "learning_rate": 0.0001999910253404569,
-      "loss": 46.0,
-      "step": 26492
-    },
-    {
-      "epoch": 4.266516365393132,
-      "grad_norm": 0.0015323837287724018,
-      "learning_rate": 0.00019999102466266108,
-      "loss": 46.0,
-      "step": 26493
-    },
-    {
-      "epoch": 4.266677402471919,
-      "grad_norm": 0.003055414417758584,
-      "learning_rate": 0.00019999102398483962,
-      "loss": 46.0,
-      "step": 26494
-    },
-    {
-      "epoch": 4.266838439550707,
-      "grad_norm": 0.0038292952813208103,
-      "learning_rate": 0.0001999910233069926,
-      "loss": 46.0,
-      "step": 26495
-    },
-    {
-      "epoch": 4.266999476629494,
-      "grad_norm": 0.009217841550707817,
-      "learning_rate": 0.00019999102262911996,
-      "loss": 46.0,
-      "step": 26496
-    },
-    {
-      "epoch": 4.267160513708282,
-      "grad_norm": 0.0015858605038374662,
-      "learning_rate": 0.00019999102195122177,
-      "loss": 46.0,
-      "step": 26497
-    },
-    {
-      "epoch": 4.267321550787068,
-      "grad_norm": 0.00275441468693316,
-      "learning_rate": 0.00019999102127329796,
-      "loss": 46.0,
-      "step": 26498
-    },
-    {
-      "epoch": 4.267482587865856,
-      "grad_norm": 0.0003706343413796276,
-      "learning_rate": 0.00019999102059534856,
-      "loss": 46.0,
-      "step": 26499
-    },
-    {
-      "epoch": 4.267643624944643,
-      "grad_norm": 0.002610041992738843,
-      "learning_rate": 0.00019999101991737358,
-      "loss": 46.0,
-      "step": 26500
-    },
-    {
-      "epoch": 4.267804662023431,
-      "grad_norm": 0.0008353313314728439,
-      "learning_rate": 0.00019999101923937298,
-      "loss": 46.0,
-      "step": 26501
-    },
-    {
-      "epoch": 4.267965699102218,
-      "grad_norm": 0.002252095378935337,
-      "learning_rate": 0.0001999910185613468,
-      "loss": 46.0,
-      "step": 26502
-    },
-    {
-      "epoch": 4.268126736181006,
-      "grad_norm": 0.0016725927125662565,
-      "learning_rate": 0.00019999101788329505,
-      "loss": 46.0,
-      "step": 26503
-    },
-    {
-      "epoch": 4.268287773259793,
-      "grad_norm": 0.0031853329855948687,
-      "learning_rate": 0.0001999910172052177,
-      "loss": 46.0,
-      "step": 26504
-    },
-    {
-      "epoch": 4.2684488103385805,
-      "grad_norm": 0.0007579732919111848,
-      "learning_rate": 0.00019999101652711475,
-      "loss": 46.0,
-      "step": 26505
-    },
-    {
-      "epoch": 4.268609847417368,
-      "grad_norm": 0.002940551843494177,
-      "learning_rate": 0.00019999101584898622,
-      "loss": 46.0,
-      "step": 26506
-    },
-    {
-      "epoch": 4.268770884496155,
-      "grad_norm": 0.001293775043450296,
-      "learning_rate": 0.00019999101517083204,
-      "loss": 46.0,
-      "step": 26507
-    },
-    {
-      "epoch": 4.268931921574943,
-      "grad_norm": 0.00363354803994298,
-      "learning_rate": 0.0001999910144926523,
-      "loss": 46.0,
-      "step": 26508
-    },
-    {
-      "epoch": 4.26909295865373,
-      "grad_norm": 0.0032102023251354694,
-      "learning_rate": 0.000199991013814447,
-      "loss": 46.0,
-      "step": 26509
-    },
-    {
-      "epoch": 4.269253995732518,
-      "grad_norm": 0.005243254359811544,
-      "learning_rate": 0.00019999101313621608,
-      "loss": 46.0,
-      "step": 26510
-    },
-    {
-      "epoch": 4.269415032811305,
-      "grad_norm": 0.005274489521980286,
-      "learning_rate": 0.00019999101245795958,
-      "loss": 46.0,
-      "step": 26511
-    },
-    {
-      "epoch": 4.269576069890093,
-      "grad_norm": 0.008864698931574821,
-      "learning_rate": 0.0001999910117796775,
-      "loss": 46.0,
-      "step": 26512
-    },
-    {
-      "epoch": 4.269737106968879,
-      "grad_norm": 0.0021092614624649286,
-      "learning_rate": 0.00019999101110136977,
-      "loss": 46.0,
-      "step": 26513
-    },
-    {
-      "epoch": 4.269898144047667,
-      "grad_norm": 0.010687408968806267,
-      "learning_rate": 0.00019999101042303651,
-      "loss": 46.0,
-      "step": 26514
-    },
-    {
-      "epoch": 4.270059181126454,
-      "grad_norm": 0.008976457640528679,
-      "learning_rate": 0.00019999100974467764,
-      "loss": 46.0,
-      "step": 26515
-    },
-    {
-      "epoch": 4.270220218205242,
-      "grad_norm": 0.006882063113152981,
-      "learning_rate": 0.00019999100906629316,
-      "loss": 46.0,
-      "step": 26516
-    },
-    {
-      "epoch": 4.270381255284029,
-      "grad_norm": 0.002696617506444454,
-      "learning_rate": 0.00019999100838788308,
-      "loss": 46.0,
-      "step": 26517
-    },
-    {
-      "epoch": 4.2705422923628165,
-      "grad_norm": 0.0024096437264233828,
-      "learning_rate": 0.00019999100770944742,
-      "loss": 46.0,
-      "step": 26518
-    },
-    {
-      "epoch": 4.270703329441604,
-      "grad_norm": 0.0037499365862458944,
-      "learning_rate": 0.00019999100703098618,
-      "loss": 46.0,
-      "step": 26519
-    },
-    {
-      "epoch": 4.270864366520391,
-      "grad_norm": 0.0049476902931928635,
-      "learning_rate": 0.00019999100635249934,
-      "loss": 46.0,
-      "step": 26520
-    },
-    {
-      "epoch": 4.271025403599179,
-      "grad_norm": 0.003276398405432701,
-      "learning_rate": 0.00019999100567398692,
-      "loss": 46.0,
-      "step": 26521
-    },
-    {
-      "epoch": 4.271186440677966,
-      "grad_norm": 0.009021335281431675,
-      "learning_rate": 0.00019999100499544888,
-      "loss": 46.0,
-      "step": 26522
-    },
-    {
-      "epoch": 4.271347477756754,
-      "grad_norm": 0.0009816967649385333,
-      "learning_rate": 0.00019999100431688526,
-      "loss": 46.0,
-      "step": 26523
-    },
-    {
-      "epoch": 4.271508514835541,
-      "grad_norm": 0.0010538423666730523,
-      "learning_rate": 0.00019999100363829605,
-      "loss": 46.0,
-      "step": 26524
-    },
-    {
-      "epoch": 4.271669551914329,
-      "grad_norm": 0.005478078033775091,
-      "learning_rate": 0.00019999100295968125,
-      "loss": 46.0,
-      "step": 26525
-    },
-    {
-      "epoch": 4.271830588993116,
-      "grad_norm": 0.0025303044822067022,
-      "learning_rate": 0.00019999100228104087,
-      "loss": 46.0,
-      "step": 26526
-    },
-    {
-      "epoch": 4.2719916260719035,
-      "grad_norm": 0.0023073945194482803,
-      "learning_rate": 0.00019999100160237487,
-      "loss": 46.0,
-      "step": 26527
-    },
-    {
-      "epoch": 4.27215266315069,
-      "grad_norm": 0.007825607433915138,
-      "learning_rate": 0.0001999910009236833,
-      "loss": 46.0,
-      "step": 26528
-    },
-    {
-      "epoch": 4.2723137002294775,
-      "grad_norm": 0.003671275684610009,
-      "learning_rate": 0.0001999910002449661,
-      "loss": 46.0,
-      "step": 26529
-    },
-    {
-      "epoch": 4.272474737308265,
-      "grad_norm": 0.0028167180716991425,
-      "learning_rate": 0.00019999099956622335,
-      "loss": 46.0,
-      "step": 26530
-    },
-    {
-      "epoch": 4.272635774387052,
-      "grad_norm": 0.0026043830439448357,
-      "learning_rate": 0.00019999099888745497,
-      "loss": 46.0,
-      "step": 26531
-    },
-    {
-      "epoch": 4.27279681146584,
-      "grad_norm": 0.006476242560893297,
-      "learning_rate": 0.00019999099820866106,
-      "loss": 46.0,
-      "step": 26532
-    },
-    {
-      "epoch": 4.272957848544627,
-      "grad_norm": 0.011442218907177448,
-      "learning_rate": 0.00019999099752984149,
-      "loss": 46.0,
-      "step": 26533
-    },
-    {
-      "epoch": 4.273118885623415,
-      "grad_norm": 0.004513797350227833,
-      "learning_rate": 0.00019999099685099635,
-      "loss": 46.0,
-      "step": 26534
-    },
-    {
-      "epoch": 4.273279922702202,
-      "grad_norm": 0.007934493012726307,
-      "learning_rate": 0.00019999099617212563,
-      "loss": 46.0,
-      "step": 26535
-    },
-    {
-      "epoch": 4.27344095978099,
-      "grad_norm": 0.0025239503011107445,
-      "learning_rate": 0.00019999099549322931,
-      "loss": 46.0,
-      "step": 26536
-    },
-    {
-      "epoch": 4.273601996859777,
-      "grad_norm": 0.002945407759398222,
-      "learning_rate": 0.0001999909948143074,
-      "loss": 46.0,
-      "step": 26537
-    },
-    {
-      "epoch": 4.273763033938565,
-      "grad_norm": 0.010019524022936821,
-      "learning_rate": 0.00019999099413535988,
-      "loss": 46.0,
-      "step": 26538
-    },
-    {
-      "epoch": 4.273924071017352,
-      "grad_norm": 0.005350045394152403,
-      "learning_rate": 0.00019999099345638678,
-      "loss": 46.0,
-      "step": 26539
-    },
-    {
-      "epoch": 4.2740851080961395,
-      "grad_norm": 0.002620598068460822,
-      "learning_rate": 0.00019999099277738812,
-      "loss": 46.0,
-      "step": 26540
-    },
-    {
-      "epoch": 4.274246145174926,
-      "grad_norm": 0.011673769913613796,
-      "learning_rate": 0.00019999099209836382,
-      "loss": 46.0,
-      "step": 26541
-    },
-    {
-      "epoch": 4.2744071822537135,
-      "grad_norm": 0.004122912418097258,
-      "learning_rate": 0.00019999099141931393,
-      "loss": 46.0,
-      "step": 26542
-    },
-    {
-      "epoch": 4.274568219332501,
-      "grad_norm": 0.0028292795177549124,
-      "learning_rate": 0.00019999099074023845,
-      "loss": 46.0,
-      "step": 26543
-    },
-    {
-      "epoch": 4.274729256411288,
-      "grad_norm": 0.0029606770258396864,
-      "learning_rate": 0.0001999909900611374,
-      "loss": 46.0,
-      "step": 26544
-    },
-    {
-      "epoch": 4.274890293490076,
-      "grad_norm": 0.004221840295940638,
-      "learning_rate": 0.00019999098938201074,
-      "loss": 46.0,
-      "step": 26545
-    },
-    {
-      "epoch": 4.275051330568863,
-      "grad_norm": 0.0016699066618457437,
-      "learning_rate": 0.0001999909887028585,
-      "loss": 46.0,
-      "step": 26546
-    },
-    {
-      "epoch": 4.275212367647651,
-      "grad_norm": 0.005708546377718449,
-      "learning_rate": 0.00019999098802368065,
-      "loss": 46.0,
-      "step": 26547
-    },
-    {
-      "epoch": 4.275373404726438,
-      "grad_norm": 0.003101825015619397,
-      "learning_rate": 0.00019999098734447724,
-      "loss": 46.0,
-      "step": 26548
-    },
-    {
-      "epoch": 4.275534441805226,
-      "grad_norm": 0.006787819787859917,
-      "learning_rate": 0.00019999098666524822,
-      "loss": 46.0,
-      "step": 26549
-    },
-    {
-      "epoch": 4.275695478884013,
-      "grad_norm": 0.003859987249597907,
-      "learning_rate": 0.0001999909859859936,
-      "loss": 46.0,
-      "step": 26550
-    },
-    {
-      "epoch": 4.275856515962801,
-      "grad_norm": 0.004384852945804596,
-      "learning_rate": 0.00019999098530671335,
-      "loss": 46.0,
-      "step": 26551
-    },
-    {
-      "epoch": 4.276017553041588,
-      "grad_norm": 0.014640344306826591,
-      "learning_rate": 0.00019999098462740756,
-      "loss": 46.0,
-      "step": 26552
-    },
-    {
-      "epoch": 4.2761785901203755,
-      "grad_norm": 0.0023615416139364243,
-      "learning_rate": 0.00019999098394807616,
-      "loss": 46.0,
-      "step": 26553
-    },
-    {
-      "epoch": 4.276339627199163,
-      "grad_norm": 0.00537398224696517,
-      "learning_rate": 0.0001999909832687192,
-      "loss": 46.0,
-      "step": 26554
-    },
-    {
-      "epoch": 4.27650066427795,
-      "grad_norm": 0.007670649327337742,
-      "learning_rate": 0.0001999909825893366,
-      "loss": 46.0,
-      "step": 26555
-    },
-    {
-      "epoch": 4.276661701356737,
-      "grad_norm": 0.004995721857994795,
-      "learning_rate": 0.0001999909819099284,
-      "loss": 46.0,
-      "step": 26556
-    },
-    {
-      "epoch": 4.276822738435524,
-      "grad_norm": 0.004026542883366346,
-      "learning_rate": 0.00019999098123049466,
-      "loss": 46.0,
-      "step": 26557
-    },
-    {
-      "epoch": 4.276983775514312,
-      "grad_norm": 0.002549421042203903,
-      "learning_rate": 0.00019999098055103532,
-      "loss": 46.0,
-      "step": 26558
-    },
-    {
-      "epoch": 4.277144812593099,
-      "grad_norm": 0.0008053979836404324,
-      "learning_rate": 0.00019999097987155034,
-      "loss": 46.0,
-      "step": 26559
-    },
-    {
-      "epoch": 4.277305849671887,
-      "grad_norm": 0.015385954640805721,
-      "learning_rate": 0.00019999097919203983,
-      "loss": 46.0,
-      "step": 26560
-    },
-    {
-      "epoch": 4.277466886750674,
-      "grad_norm": 0.0027865078300237656,
-      "learning_rate": 0.00019999097851250368,
-      "loss": 46.0,
-      "step": 26561
-    },
-    {
-      "epoch": 4.277627923829462,
-      "grad_norm": 0.0029884926043450832,
-      "learning_rate": 0.00019999097783294194,
-      "loss": 46.0,
-      "step": 26562
-    },
-    {
-      "epoch": 4.277788960908249,
-      "grad_norm": 0.002812744118273258,
-      "learning_rate": 0.0001999909771533546,
-      "loss": 46.0,
-      "step": 26563
-    },
-    {
-      "epoch": 4.277949997987037,
-      "grad_norm": 0.0017508711898699403,
-      "learning_rate": 0.00019999097647374172,
-      "loss": 46.0,
-      "step": 26564
-    },
-    {
-      "epoch": 4.278111035065824,
-      "grad_norm": 0.008622458204627037,
-      "learning_rate": 0.0001999909757941032,
-      "loss": 46.0,
-      "step": 26565
-    },
-    {
-      "epoch": 4.2782720721446115,
-      "grad_norm": 0.0008098440011963248,
-      "learning_rate": 0.0001999909751144391,
-      "loss": 46.0,
-      "step": 26566
-    },
-    {
-      "epoch": 4.278433109223399,
-      "grad_norm": 0.005315616726875305,
-      "learning_rate": 0.0001999909744347494,
-      "loss": 46.0,
-      "step": 26567
-    },
-    {
-      "epoch": 4.278594146302186,
-      "grad_norm": 0.0026556290686130524,
-      "learning_rate": 0.00019999097375503414,
-      "loss": 46.0,
-      "step": 26568
-    },
-    {
-      "epoch": 4.278755183380974,
-      "grad_norm": 0.004052278585731983,
-      "learning_rate": 0.00019999097307529326,
-      "loss": 46.0,
-      "step": 26569
-    },
-    {
-      "epoch": 4.278916220459761,
-      "grad_norm": 0.017848588526248932,
-      "learning_rate": 0.0001999909723955268,
-      "loss": 46.0,
-      "step": 26570
-    },
-    {
-      "epoch": 4.279077257538548,
-      "grad_norm": 0.0021919936407357454,
-      "learning_rate": 0.00019999097171573472,
-      "loss": 46.0,
-      "step": 26571
-    },
-    {
-      "epoch": 4.279238294617335,
-      "grad_norm": 0.004026074893772602,
-      "learning_rate": 0.00019999097103591708,
-      "loss": 46.0,
-      "step": 26572
-    },
-    {
-      "epoch": 4.279399331696123,
-      "grad_norm": 0.0068419137969613075,
-      "learning_rate": 0.00019999097035607383,
-      "loss": 46.0,
-      "step": 26573
-    },
-    {
-      "epoch": 4.27956036877491,
-      "grad_norm": 0.007841920480132103,
-      "learning_rate": 0.000199990969676205,
-      "loss": 46.0,
-      "step": 26574
-    },
-    {
-      "epoch": 4.279721405853698,
-      "grad_norm": 0.0017415997572243214,
-      "learning_rate": 0.00019999096899631056,
-      "loss": 46.0,
-      "step": 26575
-    },
-    {
-      "epoch": 4.279882442932485,
-      "grad_norm": 0.0034500970505177975,
-      "learning_rate": 0.00019999096831639052,
-      "loss": 46.0,
-      "step": 26576
-    },
-    {
-      "epoch": 4.2800434800112725,
-      "grad_norm": 0.004743452183902264,
-      "learning_rate": 0.00019999096763644492,
-      "loss": 46.0,
-      "step": 26577
-    },
-    {
-      "epoch": 4.28020451709006,
-      "grad_norm": 0.001275358721613884,
-      "learning_rate": 0.0001999909669564737,
-      "loss": 46.0,
-      "step": 26578
-    },
-    {
-      "epoch": 4.2803655541688475,
-      "grad_norm": 0.003845681669190526,
-      "learning_rate": 0.0001999909662764769,
-      "loss": 46.0,
-      "step": 26579
-    },
-    {
-      "epoch": 4.280526591247635,
-      "grad_norm": 0.0025372994132339954,
-      "learning_rate": 0.0001999909655964545,
-      "loss": 46.0,
-      "step": 26580
-    },
-    {
-      "epoch": 4.280687628326422,
-      "grad_norm": 0.014806030318140984,
-      "learning_rate": 0.00019999096491640653,
-      "loss": 46.0,
-      "step": 26581
-    },
-    {
-      "epoch": 4.28084866540521,
-      "grad_norm": 0.006652695592492819,
-      "learning_rate": 0.00019999096423633294,
-      "loss": 46.0,
-      "step": 26582
-    },
-    {
-      "epoch": 4.281009702483997,
-      "grad_norm": 0.0010407832451164722,
-      "learning_rate": 0.00019999096355623376,
-      "loss": 46.0,
-      "step": 26583
-    },
-    {
-      "epoch": 4.281170739562785,
-      "grad_norm": 0.011185847222805023,
-      "learning_rate": 0.00019999096287610902,
-      "loss": 46.0,
-      "step": 26584
-    },
-    {
-      "epoch": 4.281331776641572,
-      "grad_norm": 0.004261564929038286,
-      "learning_rate": 0.00019999096219595867,
-      "loss": 46.0,
-      "step": 26585
-    },
-    {
-      "epoch": 4.281492813720359,
-      "grad_norm": 0.002568823052570224,
-      "learning_rate": 0.00019999096151578273,
-      "loss": 46.0,
-      "step": 26586
-    },
-    {
-      "epoch": 4.281653850799146,
-      "grad_norm": 0.0023710024543106556,
-      "learning_rate": 0.00019999096083558114,
-      "loss": 46.0,
-      "step": 26587
-    },
-    {
-      "epoch": 4.281814887877934,
-      "grad_norm": 0.0030266176909208298,
-      "learning_rate": 0.000199990960155354,
-      "loss": 46.0,
-      "step": 26588
-    },
-    {
-      "epoch": 4.281975924956721,
-      "grad_norm": 0.007652450352907181,
-      "learning_rate": 0.0001999909594751013,
-      "loss": 46.0,
-      "step": 26589
-    },
-    {
-      "epoch": 4.2821369620355085,
-      "grad_norm": 0.005180449690669775,
-      "learning_rate": 0.00019999095879482298,
-      "loss": 46.0,
-      "step": 26590
-    },
-    {
-      "epoch": 4.282297999114296,
-      "grad_norm": 0.0025742785073816776,
-      "learning_rate": 0.00019999095811451908,
-      "loss": 46.0,
-      "step": 26591
-    },
-    {
-      "epoch": 4.282459036193083,
-      "grad_norm": 0.008738684467971325,
-      "learning_rate": 0.00019999095743418956,
-      "loss": 46.0,
-      "step": 26592
-    },
-    {
-      "epoch": 4.282620073271871,
-      "grad_norm": 0.001745415385812521,
-      "learning_rate": 0.00019999095675383446,
-      "loss": 46.0,
-      "step": 26593
-    },
-    {
-      "epoch": 4.282781110350658,
-      "grad_norm": 0.004507759120315313,
-      "learning_rate": 0.00019999095607345377,
-      "loss": 46.0,
-      "step": 26594
-    },
-    {
-      "epoch": 4.282942147429446,
-      "grad_norm": 0.002628570655360818,
-      "learning_rate": 0.00019999095539304749,
-      "loss": 46.0,
-      "step": 26595
-    },
-    {
-      "epoch": 4.283103184508233,
-      "grad_norm": 0.002590820426121354,
-      "learning_rate": 0.00019999095471261562,
-      "loss": 46.0,
-      "step": 26596
-    },
-    {
-      "epoch": 4.283264221587021,
-      "grad_norm": 0.00393479922786355,
-      "learning_rate": 0.00019999095403215814,
-      "loss": 46.0,
-      "step": 26597
-    },
-    {
-      "epoch": 4.283425258665808,
-      "grad_norm": 0.005723224487155676,
-      "learning_rate": 0.00019999095335167507,
-      "loss": 46.0,
-      "step": 26598
-    },
-    {
-      "epoch": 4.283586295744596,
-      "grad_norm": 0.010294840671122074,
-      "learning_rate": 0.00019999095267116644,
-      "loss": 46.0,
-      "step": 26599
-    },
-    {
-      "epoch": 4.283747332823383,
-      "grad_norm": 0.0033415881916880608,
-      "learning_rate": 0.00019999095199063217,
-      "loss": 46.0,
-      "step": 26600
-    },
-    {
-      "epoch": 4.28390836990217,
-      "grad_norm": 0.0013351988745853305,
-      "learning_rate": 0.00019999095131007234,
-      "loss": 46.0,
-      "step": 26601
-    },
-    {
-      "epoch": 4.284069406980957,
-      "grad_norm": 0.003475687000900507,
-      "learning_rate": 0.00019999095062948693,
-      "loss": 46.0,
-      "step": 26602
-    },
-    {
-      "epoch": 4.2842304440597445,
-      "grad_norm": 0.017013397067785263,
-      "learning_rate": 0.0001999909499488759,
-      "loss": 46.0,
-      "step": 26603
-    },
-    {
-      "epoch": 4.284391481138532,
-      "grad_norm": 0.012152256444096565,
-      "learning_rate": 0.00019999094926823928,
-      "loss": 46.0,
-      "step": 26604
-    },
-    {
-      "epoch": 4.284552518217319,
-      "grad_norm": 0.003864055033773184,
-      "learning_rate": 0.00019999094858757705,
-      "loss": 46.0,
-      "step": 26605
-    },
-    {
-      "epoch": 4.284713555296107,
-      "grad_norm": 0.002758234040811658,
-      "learning_rate": 0.00019999094790688928,
-      "loss": 46.0,
-      "step": 26606
-    },
-    {
-      "epoch": 4.284874592374894,
-      "grad_norm": 0.00917613785713911,
-      "learning_rate": 0.00019999094722617587,
-      "loss": 46.0,
-      "step": 26607
-    },
-    {
-      "epoch": 4.285035629453682,
-      "grad_norm": 0.0014187961351126432,
-      "learning_rate": 0.00019999094654543688,
-      "loss": 46.0,
-      "step": 26608
-    },
-    {
-      "epoch": 4.285196666532469,
-      "grad_norm": 0.006978214252740145,
-      "learning_rate": 0.0001999909458646723,
-      "loss": 46.0,
-      "step": 26609
-    },
-    {
-      "epoch": 4.285357703611257,
-      "grad_norm": 0.006249192636460066,
-      "learning_rate": 0.00019999094518388213,
-      "loss": 46.0,
-      "step": 26610
-    },
-    {
-      "epoch": 4.285518740690044,
-      "grad_norm": 0.004516422748565674,
-      "learning_rate": 0.00019999094450306638,
-      "loss": 46.0,
-      "step": 26611
-    },
-    {
-      "epoch": 4.285679777768832,
-      "grad_norm": 0.00408960273489356,
-      "learning_rate": 0.00019999094382222503,
-      "loss": 46.0,
-      "step": 26612
-    },
-    {
-      "epoch": 4.285840814847619,
-      "grad_norm": 0.008581865578889847,
-      "learning_rate": 0.00019999094314135808,
-      "loss": 46.0,
-      "step": 26613
-    },
-    {
-      "epoch": 4.286001851926406,
-      "grad_norm": 0.0033353972248733044,
-      "learning_rate": 0.00019999094246046553,
-      "loss": 46.0,
-      "step": 26614
-    },
-    {
-      "epoch": 4.286162889005193,
-      "grad_norm": 0.005265991669148207,
-      "learning_rate": 0.0001999909417795474,
-      "loss": 46.0,
-      "step": 26615
-    },
-    {
-      "epoch": 4.2863239260839805,
-      "grad_norm": 0.0015589615795761347,
-      "learning_rate": 0.00019999094109860366,
-      "loss": 46.0,
-      "step": 26616
-    },
-    {
-      "epoch": 4.286484963162768,
-      "grad_norm": 0.006723297759890556,
-      "learning_rate": 0.00019999094041763435,
-      "loss": 46.0,
-      "step": 26617
-    },
-    {
-      "epoch": 4.286646000241555,
-      "grad_norm": 0.005083768162876368,
-      "learning_rate": 0.00019999093973663943,
-      "loss": 46.0,
-      "step": 26618
-    },
-    {
-      "epoch": 4.286807037320343,
-      "grad_norm": 0.005982173141092062,
-      "learning_rate": 0.00019999093905561895,
-      "loss": 46.0,
-      "step": 26619
-    },
-    {
-      "epoch": 4.28696807439913,
-      "grad_norm": 0.0018009596969932318,
-      "learning_rate": 0.00019999093837457283,
-      "loss": 46.0,
-      "step": 26620
-    },
-    {
-      "epoch": 4.287129111477918,
-      "grad_norm": 0.0015570831019431353,
-      "learning_rate": 0.00019999093769350115,
-      "loss": 46.0,
-      "step": 26621
-    },
-    {
-      "epoch": 4.287290148556705,
-      "grad_norm": 0.0029178354889154434,
-      "learning_rate": 0.00019999093701240388,
-      "loss": 46.0,
-      "step": 26622
-    },
-    {
-      "epoch": 4.287451185635493,
-      "grad_norm": 0.0030241538770496845,
-      "learning_rate": 0.000199990936331281,
-      "loss": 46.0,
-      "step": 26623
-    },
-    {
-      "epoch": 4.28761222271428,
-      "grad_norm": 0.02319498546421528,
-      "learning_rate": 0.00019999093565013253,
-      "loss": 46.0,
-      "step": 26624
-    },
-    {
-      "epoch": 4.287773259793068,
-      "grad_norm": 0.0027693198062479496,
-      "learning_rate": 0.00019999093496895847,
-      "loss": 46.0,
-      "step": 26625
-    },
-    {
-      "epoch": 4.287934296871855,
-      "grad_norm": 0.0031212191097438335,
-      "learning_rate": 0.00019999093428775883,
-      "loss": 46.0,
-      "step": 26626
-    },
-    {
-      "epoch": 4.2880953339506425,
-      "grad_norm": 0.0008244689088314772,
-      "learning_rate": 0.00019999093360653357,
-      "loss": 46.0,
-      "step": 26627
-    },
-    {
-      "epoch": 4.28825637102943,
-      "grad_norm": 0.004984673112630844,
-      "learning_rate": 0.00019999093292528275,
-      "loss": 46.0,
-      "step": 26628
-    },
-    {
-      "epoch": 4.2884174081082165,
-      "grad_norm": 0.0034086136147379875,
-      "learning_rate": 0.0001999909322440063,
-      "loss": 46.0,
-      "step": 26629
-    },
-    {
-      "epoch": 4.288578445187004,
-      "grad_norm": 0.01127166859805584,
-      "learning_rate": 0.00019999093156270427,
-      "loss": 46.0,
-      "step": 26630
-    },
-    {
-      "epoch": 4.288739482265791,
-      "grad_norm": 0.011049147695302963,
-      "learning_rate": 0.0001999909308813767,
-      "loss": 46.0,
-      "step": 26631
-    },
-    {
-      "epoch": 4.288900519344579,
-      "grad_norm": 0.0017384352395310998,
-      "learning_rate": 0.00019999093020002347,
-      "loss": 46.0,
-      "step": 26632
-    },
-    {
-      "epoch": 4.289061556423366,
-      "grad_norm": 0.002624780172482133,
-      "learning_rate": 0.00019999092951864466,
-      "loss": 46.0,
-      "step": 26633
-    },
-    {
-      "epoch": 4.289222593502154,
-      "grad_norm": 0.006158050149679184,
-      "learning_rate": 0.00019999092883724026,
-      "loss": 46.0,
-      "step": 26634
-    },
-    {
-      "epoch": 4.289383630580941,
-      "grad_norm": 0.007796927355229855,
-      "learning_rate": 0.00019999092815581028,
-      "loss": 46.0,
-      "step": 26635
-    },
-    {
-      "epoch": 4.289544667659729,
-      "grad_norm": 0.0029653809033334255,
-      "learning_rate": 0.0001999909274743547,
-      "loss": 46.0,
-      "step": 26636
-    },
-    {
-      "epoch": 4.289705704738516,
-      "grad_norm": 0.0018333265325054526,
-      "learning_rate": 0.00019999092679287355,
-      "loss": 46.0,
-      "step": 26637
-    },
-    {
-      "epoch": 4.2898667418173035,
-      "grad_norm": 0.00667930580675602,
-      "learning_rate": 0.00019999092611136678,
-      "loss": 46.0,
-      "step": 26638
-    },
-    {
-      "epoch": 4.290027778896091,
-      "grad_norm": 0.0036499658599495888,
-      "learning_rate": 0.00019999092542983442,
-      "loss": 46.0,
-      "step": 26639
-    },
-    {
-      "epoch": 4.2901888159748784,
-      "grad_norm": 0.004853421822190285,
-      "learning_rate": 0.00019999092474827648,
-      "loss": 46.0,
-      "step": 26640
-    },
-    {
-      "epoch": 4.290349853053666,
-      "grad_norm": 0.003092606784775853,
-      "learning_rate": 0.00019999092406669294,
-      "loss": 46.0,
-      "step": 26641
-    },
-    {
-      "epoch": 4.290510890132453,
-      "grad_norm": 0.01001638825982809,
-      "learning_rate": 0.0001999909233850838,
-      "loss": 46.0,
-      "step": 26642
-    },
-    {
-      "epoch": 4.290671927211241,
-      "grad_norm": 0.00549060944467783,
-      "learning_rate": 0.00019999092270344906,
-      "loss": 46.0,
-      "step": 26643
-    },
-    {
-      "epoch": 4.290832964290027,
-      "grad_norm": 0.00893006008118391,
-      "learning_rate": 0.00019999092202178874,
-      "loss": 46.0,
-      "step": 26644
-    },
-    {
-      "epoch": 4.290994001368815,
-      "grad_norm": 0.0012227289844304323,
-      "learning_rate": 0.00019999092134010283,
-      "loss": 46.0,
-      "step": 26645
-    },
-    {
-      "epoch": 4.291155038447602,
-      "grad_norm": 0.00520157627761364,
-      "learning_rate": 0.00019999092065839134,
-      "loss": 46.0,
-      "step": 26646
-    },
-    {
-      "epoch": 4.29131607552639,
-      "grad_norm": 0.0033616607543081045,
-      "learning_rate": 0.00019999091997665425,
-      "loss": 46.0,
-      "step": 26647
-    },
-    {
-      "epoch": 4.291477112605177,
-      "grad_norm": 0.0013342745369300246,
-      "learning_rate": 0.00019999091929489155,
-      "loss": 46.0,
-      "step": 26648
-    },
-    {
-      "epoch": 4.291638149683965,
-      "grad_norm": 0.00819773506373167,
-      "learning_rate": 0.00019999091861310327,
-      "loss": 46.0,
-      "step": 26649
-    },
-    {
-      "epoch": 4.291799186762752,
-      "grad_norm": 0.008123735897243023,
-      "learning_rate": 0.00019999091793128942,
-      "loss": 46.0,
-      "step": 26650
-    },
-    {
-      "epoch": 4.2919602238415395,
-      "grad_norm": 0.005695602390915155,
-      "learning_rate": 0.00019999091724944994,
-      "loss": 46.0,
-      "step": 26651
-    },
-    {
-      "epoch": 4.292121260920327,
-      "grad_norm": 0.002651038346812129,
-      "learning_rate": 0.0001999909165675849,
-      "loss": 46.0,
-      "step": 26652
-    },
-    {
-      "epoch": 4.292282297999114,
-      "grad_norm": 0.003819429548457265,
-      "learning_rate": 0.00019999091588569423,
-      "loss": 46.0,
-      "step": 26653
-    },
-    {
-      "epoch": 4.292443335077902,
-      "grad_norm": 0.006745167542248964,
-      "learning_rate": 0.00019999091520377798,
-      "loss": 46.0,
-      "step": 26654
-    },
-    {
-      "epoch": 4.292604372156689,
-      "grad_norm": 0.0028554874006658792,
-      "learning_rate": 0.00019999091452183615,
-      "loss": 46.0,
-      "step": 26655
-    },
-    {
-      "epoch": 4.292765409235477,
-      "grad_norm": 0.005565503612160683,
-      "learning_rate": 0.00019999091383986873,
-      "loss": 46.0,
-      "step": 26656
-    },
-    {
-      "epoch": 4.292926446314264,
-      "grad_norm": 0.003558178199455142,
-      "learning_rate": 0.0001999909131578757,
-      "loss": 46.0,
-      "step": 26657
-    },
-    {
-      "epoch": 4.293087483393052,
-      "grad_norm": 0.006929673254489899,
-      "learning_rate": 0.0001999909124758571,
-      "loss": 46.0,
-      "step": 26658
-    },
-    {
-      "epoch": 4.293248520471838,
-      "grad_norm": 0.012406129390001297,
-      "learning_rate": 0.00019999091179381288,
-      "loss": 46.0,
-      "step": 26659
-    },
-    {
-      "epoch": 4.293409557550626,
-      "grad_norm": 0.0014742364874109626,
-      "learning_rate": 0.0001999909111117431,
-      "loss": 46.0,
-      "step": 26660
-    },
-    {
-      "epoch": 4.293570594629413,
-      "grad_norm": 0.0027932864613831043,
-      "learning_rate": 0.0001999909104296477,
-      "loss": 46.0,
-      "step": 26661
-    },
-    {
-      "epoch": 4.293731631708201,
-      "grad_norm": 0.006052783690392971,
-      "learning_rate": 0.00019999090974752673,
-      "loss": 46.0,
-      "step": 26662
-    },
-    {
-      "epoch": 4.293892668786988,
-      "grad_norm": 0.002156896749511361,
-      "learning_rate": 0.00019999090906538014,
-      "loss": 46.0,
-      "step": 26663
-    },
-    {
-      "epoch": 4.2940537058657755,
-      "grad_norm": 0.004768858663737774,
-      "learning_rate": 0.00019999090838320797,
-      "loss": 46.0,
-      "step": 26664
-    },
-    {
-      "epoch": 4.294214742944563,
-      "grad_norm": 0.0023126413580030203,
-      "learning_rate": 0.0001999909077010102,
-      "loss": 46.0,
-      "step": 26665
-    },
-    {
-      "epoch": 4.29437578002335,
-      "grad_norm": 0.005086436867713928,
-      "learning_rate": 0.00019999090701878686,
-      "loss": 46.0,
-      "step": 26666
-    },
-    {
-      "epoch": 4.294536817102138,
-      "grad_norm": 0.003239473793655634,
-      "learning_rate": 0.0001999909063365379,
-      "loss": 46.0,
-      "step": 26667
-    },
-    {
-      "epoch": 4.294697854180925,
-      "grad_norm": 0.009535335004329681,
-      "learning_rate": 0.00019999090565426335,
-      "loss": 46.0,
-      "step": 26668
-    },
-    {
-      "epoch": 4.294858891259713,
-      "grad_norm": 0.0024792891927063465,
-      "learning_rate": 0.0001999909049719632,
-      "loss": 46.0,
-      "step": 26669
-    },
-    {
-      "epoch": 4.2950199283385,
-      "grad_norm": 0.0019711507484316826,
-      "learning_rate": 0.00019999090428963752,
-      "loss": 46.0,
-      "step": 26670
-    },
-    {
-      "epoch": 4.295180965417288,
-      "grad_norm": 0.0029612237121909857,
-      "learning_rate": 0.00019999090360728618,
-      "loss": 46.0,
-      "step": 26671
-    },
-    {
-      "epoch": 4.295342002496075,
-      "grad_norm": 0.013222194276750088,
-      "learning_rate": 0.00019999090292490928,
-      "loss": 46.0,
-      "step": 26672
-    },
-    {
-      "epoch": 4.295503039574863,
-      "grad_norm": 0.00529817771166563,
-      "learning_rate": 0.0001999909022425068,
-      "loss": 46.0,
-      "step": 26673
-    },
-    {
-      "epoch": 4.295664076653649,
-      "grad_norm": 0.017362484708428383,
-      "learning_rate": 0.0001999909015600787,
-      "loss": 46.0,
-      "step": 26674
-    },
-    {
-      "epoch": 4.295825113732437,
-      "grad_norm": 0.005261871498078108,
-      "learning_rate": 0.000199990900877625,
-      "loss": 46.0,
-      "step": 26675
-    },
-    {
-      "epoch": 4.295986150811224,
-      "grad_norm": 0.005306694190949202,
-      "learning_rate": 0.00019999090019514574,
-      "loss": 46.0,
-      "step": 26676
-    },
-    {
-      "epoch": 4.2961471878900115,
-      "grad_norm": 0.007739758584648371,
-      "learning_rate": 0.00019999089951264085,
-      "loss": 46.0,
-      "step": 26677
-    },
-    {
-      "epoch": 4.296308224968799,
-      "grad_norm": 0.0015890548238530755,
-      "learning_rate": 0.00019999089883011037,
-      "loss": 46.0,
-      "step": 26678
-    },
-    {
-      "epoch": 4.296469262047586,
-      "grad_norm": 0.017876725643873215,
-      "learning_rate": 0.0001999908981475543,
-      "loss": 46.0,
-      "step": 26679
-    },
-    {
-      "epoch": 4.296630299126374,
-      "grad_norm": 0.0012373109348118305,
-      "learning_rate": 0.0001999908974649727,
-      "loss": 46.0,
-      "step": 26680
-    },
-    {
-      "epoch": 4.296791336205161,
-      "grad_norm": 0.0043030050583183765,
-      "learning_rate": 0.00019999089678236545,
-      "loss": 46.0,
-      "step": 26681
-    },
-    {
-      "epoch": 4.296952373283949,
-      "grad_norm": 0.008796032518148422,
-      "learning_rate": 0.0001999908960997326,
-      "loss": 46.0,
-      "step": 26682
-    },
-    {
-      "epoch": 4.297113410362736,
-      "grad_norm": 0.016864174976944923,
-      "learning_rate": 0.00019999089541707416,
-      "loss": 46.0,
-      "step": 26683
-    },
-    {
-      "epoch": 4.297274447441524,
-      "grad_norm": 0.001528097433038056,
-      "learning_rate": 0.00019999089473439016,
-      "loss": 46.0,
-      "step": 26684
-    },
-    {
-      "epoch": 4.297435484520311,
-      "grad_norm": 0.0013680357951670885,
-      "learning_rate": 0.00019999089405168055,
-      "loss": 46.0,
-      "step": 26685
-    },
-    {
-      "epoch": 4.2975965215990986,
-      "grad_norm": 0.00332814222201705,
-      "learning_rate": 0.00019999089336894535,
-      "loss": 46.0,
-      "step": 26686
-    },
-    {
-      "epoch": 4.297757558677885,
-      "grad_norm": 0.0034406413324177265,
-      "learning_rate": 0.00019999089268618456,
-      "loss": 46.0,
-      "step": 26687
-    },
-    {
-      "epoch": 4.297918595756673,
-      "grad_norm": 0.0029496520292013884,
-      "learning_rate": 0.00019999089200339816,
-      "loss": 46.0,
-      "step": 26688
-    },
-    {
-      "epoch": 4.29807963283546,
-      "grad_norm": 0.006994324736297131,
-      "learning_rate": 0.00019999089132058617,
-      "loss": 46.0,
-      "step": 26689
-    },
-    {
-      "epoch": 4.2982406699142475,
-      "grad_norm": 0.008817637339234352,
-      "learning_rate": 0.0001999908906377486,
-      "loss": 46.0,
-      "step": 26690
-    },
-    {
-      "epoch": 4.298401706993035,
-      "grad_norm": 0.005084660369902849,
-      "learning_rate": 0.0001999908899548854,
-      "loss": 46.0,
-      "step": 26691
-    },
-    {
-      "epoch": 4.298562744071822,
-      "grad_norm": 0.0020231306552886963,
-      "learning_rate": 0.00019999088927199669,
-      "loss": 46.0,
-      "step": 26692
-    },
-    {
-      "epoch": 4.29872378115061,
-      "grad_norm": 0.013756733387708664,
-      "learning_rate": 0.0001999908885890823,
-      "loss": 46.0,
-      "step": 26693
-    },
-    {
-      "epoch": 4.298884818229397,
-      "grad_norm": 0.005016148090362549,
-      "learning_rate": 0.00019999088790614237,
-      "loss": 46.0,
-      "step": 26694
-    },
-    {
-      "epoch": 4.299045855308185,
-      "grad_norm": 0.002878783503547311,
-      "learning_rate": 0.00019999088722317683,
-      "loss": 46.0,
-      "step": 26695
-    },
-    {
-      "epoch": 4.299206892386972,
-      "grad_norm": 0.016500653699040413,
-      "learning_rate": 0.0001999908865401857,
-      "loss": 46.0,
-      "step": 26696
-    },
-    {
-      "epoch": 4.29936792946576,
-      "grad_norm": 0.003728938288986683,
-      "learning_rate": 0.000199990885857169,
-      "loss": 46.0,
-      "step": 26697
-    },
-    {
-      "epoch": 4.299528966544547,
-      "grad_norm": 0.003530869958922267,
-      "learning_rate": 0.00019999088517412667,
-      "loss": 46.0,
-      "step": 26698
-    },
-    {
-      "epoch": 4.2996900036233345,
-      "grad_norm": 0.0028254776261746883,
-      "learning_rate": 0.00019999088449105875,
-      "loss": 46.0,
-      "step": 26699
-    },
-    {
-      "epoch": 4.299851040702122,
-      "grad_norm": 0.0009359674877487123,
-      "learning_rate": 0.00019999088380796528,
-      "loss": 46.0,
-      "step": 26700
-    },
-    {
-      "epoch": 4.300012077780909,
-      "grad_norm": 0.0008102217689156532,
-      "learning_rate": 0.00019999088312484616,
-      "loss": 46.0,
-      "step": 26701
-    },
-    {
-      "epoch": 4.300173114859696,
-      "grad_norm": 0.0013406844809651375,
-      "learning_rate": 0.00019999088244170149,
-      "loss": 46.0,
-      "step": 26702
-    },
-    {
-      "epoch": 4.3003341519384835,
-      "grad_norm": 0.003481874242424965,
-      "learning_rate": 0.0001999908817585312,
-      "loss": 46.0,
-      "step": 26703
-    },
-    {
-      "epoch": 4.300495189017271,
-      "grad_norm": 0.0012658028863370419,
-      "learning_rate": 0.00019999088107533535,
-      "loss": 46.0,
-      "step": 26704
-    },
-    {
-      "epoch": 4.300656226096058,
-      "grad_norm": 0.0047179111279547215,
-      "learning_rate": 0.00019999088039211385,
-      "loss": 46.0,
-      "step": 26705
-    },
-    {
-      "epoch": 4.300817263174846,
-      "grad_norm": 0.004426430445164442,
-      "learning_rate": 0.0001999908797088668,
-      "loss": 46.0,
-      "step": 26706
-    },
-    {
-      "epoch": 4.300978300253633,
-      "grad_norm": 0.0029422191437333822,
-      "learning_rate": 0.00019999087902559416,
-      "loss": 46.0,
-      "step": 26707
-    },
-    {
-      "epoch": 4.301139337332421,
-      "grad_norm": 0.016356289386749268,
-      "learning_rate": 0.00019999087834229594,
-      "loss": 46.0,
-      "step": 26708
-    },
-    {
-      "epoch": 4.301300374411208,
-      "grad_norm": 0.0074377222917973995,
-      "learning_rate": 0.0001999908776589721,
-      "loss": 46.0,
-      "step": 26709
-    },
-    {
-      "epoch": 4.301461411489996,
-      "grad_norm": 0.0068101114593446255,
-      "learning_rate": 0.0001999908769756227,
-      "loss": 46.0,
-      "step": 26710
-    },
-    {
-      "epoch": 4.301622448568783,
-      "grad_norm": 0.002824036404490471,
-      "learning_rate": 0.00019999087629224766,
-      "loss": 46.0,
-      "step": 26711
-    },
-    {
-      "epoch": 4.3017834856475705,
-      "grad_norm": 0.005177484359592199,
-      "learning_rate": 0.00019999087560884705,
-      "loss": 46.0,
-      "step": 26712
-    },
-    {
-      "epoch": 4.301944522726358,
-      "grad_norm": 0.0039574452675879,
-      "learning_rate": 0.00019999087492542086,
-      "loss": 46.0,
-      "step": 26713
-    },
-    {
-      "epoch": 4.302105559805145,
-      "grad_norm": 0.003296489827334881,
-      "learning_rate": 0.00019999087424196906,
-      "loss": 46.0,
-      "step": 26714
-    },
-    {
-      "epoch": 4.302266596883933,
-      "grad_norm": 0.001527283457107842,
-      "learning_rate": 0.00019999087355849167,
-      "loss": 46.0,
-      "step": 26715
-    },
-    {
-      "epoch": 4.30242763396272,
-      "grad_norm": 0.0034816754050552845,
-      "learning_rate": 0.00019999087287498867,
-      "loss": 46.0,
-      "step": 26716
-    },
-    {
-      "epoch": 4.302588671041507,
-      "grad_norm": 0.006492525804787874,
-      "learning_rate": 0.0001999908721914601,
-      "loss": 46.0,
-      "step": 26717
-    },
-    {
-      "epoch": 4.302749708120294,
-      "grad_norm": 0.005838372744619846,
-      "learning_rate": 0.00019999087150790595,
-      "loss": 46.0,
-      "step": 26718
-    },
-    {
-      "epoch": 4.302910745199082,
-      "grad_norm": 0.013641547411680222,
-      "learning_rate": 0.00019999087082432618,
-      "loss": 46.0,
-      "step": 26719
-    },
-    {
-      "epoch": 4.303071782277869,
-      "grad_norm": 0.0020314499270170927,
-      "learning_rate": 0.00019999087014072083,
-      "loss": 46.0,
-      "step": 26720
-    },
-    {
-      "epoch": 4.303232819356657,
-      "grad_norm": 0.0036813009064644575,
-      "learning_rate": 0.0001999908694570899,
-      "loss": 46.0,
-      "step": 26721
-    },
-    {
-      "epoch": 4.303393856435444,
-      "grad_norm": 0.0015707028796896338,
-      "learning_rate": 0.00019999086877343336,
-      "loss": 46.0,
-      "step": 26722
-    },
-    {
-      "epoch": 4.303554893514232,
-      "grad_norm": 0.0005612073582597077,
-      "learning_rate": 0.00019999086808975122,
-      "loss": 46.0,
-      "step": 26723
-    },
-    {
-      "epoch": 4.303715930593019,
-      "grad_norm": 0.0027560086455196142,
-      "learning_rate": 0.00019999086740604352,
-      "loss": 46.0,
-      "step": 26724
-    },
-    {
-      "epoch": 4.3038769676718065,
-      "grad_norm": 0.0025809984654188156,
-      "learning_rate": 0.00019999086672231017,
-      "loss": 46.0,
-      "step": 26725
-    },
-    {
-      "epoch": 4.304038004750594,
-      "grad_norm": 0.0029472627211362123,
-      "learning_rate": 0.00019999086603855127,
-      "loss": 46.0,
-      "step": 26726
-    },
-    {
-      "epoch": 4.304199041829381,
-      "grad_norm": 0.007135242223739624,
-      "learning_rate": 0.00019999086535476678,
-      "loss": 46.0,
-      "step": 26727
-    },
-    {
-      "epoch": 4.304360078908169,
-      "grad_norm": 0.000985802966170013,
-      "learning_rate": 0.0001999908646709567,
-      "loss": 46.0,
-      "step": 26728
-    },
-    {
-      "epoch": 4.304521115986956,
-      "grad_norm": 0.0017725750803947449,
-      "learning_rate": 0.000199990863987121,
-      "loss": 46.0,
-      "step": 26729
-    },
-    {
-      "epoch": 4.304682153065744,
-      "grad_norm": 0.002109637251123786,
-      "learning_rate": 0.00019999086330325973,
-      "loss": 46.0,
-      "step": 26730
-    },
-    {
-      "epoch": 4.304843190144531,
-      "grad_norm": 0.0025911603588610888,
-      "learning_rate": 0.00019999086261937287,
-      "loss": 46.0,
-      "step": 26731
-    },
-    {
-      "epoch": 4.305004227223318,
-      "grad_norm": 0.0013003478525206447,
-      "learning_rate": 0.0001999908619354604,
-      "loss": 46.0,
-      "step": 26732
-    },
-    {
-      "epoch": 4.305165264302105,
-      "grad_norm": 0.005003707483410835,
-      "learning_rate": 0.00019999086125152234,
-      "loss": 46.0,
-      "step": 26733
-    },
-    {
-      "epoch": 4.305326301380893,
-      "grad_norm": 0.0056510204449296,
-      "learning_rate": 0.00019999086056755872,
-      "loss": 46.0,
-      "step": 26734
-    },
-    {
-      "epoch": 4.30548733845968,
-      "grad_norm": 0.0020021649543195963,
-      "learning_rate": 0.00019999085988356947,
-      "loss": 46.0,
-      "step": 26735
-    },
-    {
-      "epoch": 4.305648375538468,
-      "grad_norm": 0.001760657294653356,
-      "learning_rate": 0.00019999085919955464,
-      "loss": 46.0,
-      "step": 26736
-    },
-    {
-      "epoch": 4.305809412617255,
-      "grad_norm": 0.01574433594942093,
-      "learning_rate": 0.00019999085851551423,
-      "loss": 46.0,
-      "step": 26737
-    },
-    {
-      "epoch": 4.3059704496960425,
-      "grad_norm": 0.004249979741871357,
-      "learning_rate": 0.0001999908578314482,
-      "loss": 46.0,
-      "step": 26738
-    },
-    {
-      "epoch": 4.30613148677483,
-      "grad_norm": 0.02154911682009697,
-      "learning_rate": 0.00019999085714735658,
-      "loss": 46.0,
-      "step": 26739
-    },
-    {
-      "epoch": 4.306292523853617,
-      "grad_norm": 0.0044340468011796474,
-      "learning_rate": 0.00019999085646323937,
-      "loss": 46.0,
-      "step": 26740
-    },
-    {
-      "epoch": 4.306453560932405,
-      "grad_norm": 0.009184077382087708,
-      "learning_rate": 0.0001999908557790966,
-      "loss": 46.0,
-      "step": 26741
-    },
-    {
-      "epoch": 4.306614598011192,
-      "grad_norm": 0.0012312076287344098,
-      "learning_rate": 0.0001999908550949282,
-      "loss": 46.0,
-      "step": 26742
-    },
-    {
-      "epoch": 4.30677563508998,
-      "grad_norm": 0.0018530371598899364,
-      "learning_rate": 0.00019999085441073424,
-      "loss": 46.0,
-      "step": 26743
-    },
-    {
-      "epoch": 4.306936672168767,
-      "grad_norm": 0.002606758149340749,
-      "learning_rate": 0.00019999085372651468,
-      "loss": 46.0,
-      "step": 26744
-    },
-    {
-      "epoch": 4.307097709247555,
-      "grad_norm": 0.0008032356854528189,
-      "learning_rate": 0.00019999085304226949,
-      "loss": 46.0,
-      "step": 26745
-    },
-    {
-      "epoch": 4.307258746326342,
-      "grad_norm": 0.003036564914509654,
-      "learning_rate": 0.00019999085235799876,
-      "loss": 46.0,
-      "step": 26746
-    },
-    {
-      "epoch": 4.307419783405129,
-      "grad_norm": 0.00476501788944006,
-      "learning_rate": 0.0001999908516737024,
-      "loss": 46.0,
-      "step": 26747
-    },
-    {
-      "epoch": 4.307580820483916,
-      "grad_norm": 0.006617344915866852,
-      "learning_rate": 0.00019999085098938043,
-      "loss": 46.0,
-      "step": 26748
-    },
-    {
-      "epoch": 4.307741857562704,
-      "grad_norm": 0.007015538401901722,
-      "learning_rate": 0.00019999085030503292,
-      "loss": 46.0,
-      "step": 26749
-    },
-    {
-      "epoch": 4.307902894641491,
-      "grad_norm": 0.004610313102602959,
-      "learning_rate": 0.00019999084962065976,
-      "loss": 46.0,
-      "step": 26750
-    },
-    {
-      "epoch": 4.3080639317202785,
-      "grad_norm": 0.0031711666379123926,
-      "learning_rate": 0.00019999084893626107,
-      "loss": 46.0,
-      "step": 26751
-    },
-    {
-      "epoch": 4.308224968799066,
-      "grad_norm": 0.006278675980865955,
-      "learning_rate": 0.00019999084825183676,
-      "loss": 46.0,
-      "step": 26752
-    },
-    {
-      "epoch": 4.308386005877853,
-      "grad_norm": 0.0010566464625298977,
-      "learning_rate": 0.00019999084756738684,
-      "loss": 46.0,
-      "step": 26753
-    },
-    {
-      "epoch": 4.308547042956641,
-      "grad_norm": 0.006965600419789553,
-      "learning_rate": 0.00019999084688291133,
-      "loss": 46.0,
-      "step": 26754
-    },
-    {
-      "epoch": 4.308708080035428,
-      "grad_norm": 0.005000283941626549,
-      "learning_rate": 0.00019999084619841027,
-      "loss": 46.0,
-      "step": 26755
-    },
-    {
-      "epoch": 4.308869117114216,
-      "grad_norm": 0.005617290735244751,
-      "learning_rate": 0.00019999084551388359,
-      "loss": 46.0,
-      "step": 26756
-    },
-    {
-      "epoch": 4.309030154193003,
-      "grad_norm": 0.008802504278719425,
-      "learning_rate": 0.0001999908448293313,
-      "loss": 46.0,
-      "step": 26757
-    },
-    {
-      "epoch": 4.309191191271791,
-      "grad_norm": 0.0014073667116463184,
-      "learning_rate": 0.00019999084414475343,
-      "loss": 46.0,
-      "step": 26758
-    },
-    {
-      "epoch": 4.309352228350578,
-      "grad_norm": 0.003529355162754655,
-      "learning_rate": 0.00019999084346015,
-      "loss": 46.0,
-      "step": 26759
-    },
-    {
-      "epoch": 4.3095132654293655,
-      "grad_norm": 0.0074369837529957294,
-      "learning_rate": 0.0001999908427755209,
-      "loss": 46.0,
-      "step": 26760
-    },
-    {
-      "epoch": 4.309674302508152,
-      "grad_norm": 0.011568040587008,
-      "learning_rate": 0.00019999084209086626,
-      "loss": 46.0,
-      "step": 26761
-    },
-    {
-      "epoch": 4.3098353395869395,
-      "grad_norm": 0.0017909719608724117,
-      "learning_rate": 0.00019999084140618606,
-      "loss": 46.0,
-      "step": 26762
-    },
-    {
-      "epoch": 4.309996376665727,
-      "grad_norm": 0.001900325994938612,
-      "learning_rate": 0.00019999084072148021,
-      "loss": 46.0,
-      "step": 26763
-    },
-    {
-      "epoch": 4.310157413744514,
-      "grad_norm": 0.005974930711090565,
-      "learning_rate": 0.00019999084003674878,
-      "loss": 46.0,
-      "step": 26764
-    },
-    {
-      "epoch": 4.310318450823302,
-      "grad_norm": 0.00584437744691968,
-      "learning_rate": 0.00019999083935199176,
-      "loss": 46.0,
-      "step": 26765
-    },
-    {
-      "epoch": 4.310479487902089,
-      "grad_norm": 0.003897067392244935,
-      "learning_rate": 0.00019999083866720918,
-      "loss": 46.0,
-      "step": 26766
-    },
-    {
-      "epoch": 4.310640524980877,
-      "grad_norm": 0.0007997590582817793,
-      "learning_rate": 0.00019999083798240096,
-      "loss": 46.0,
-      "step": 26767
-    },
-    {
-      "epoch": 4.310801562059664,
-      "grad_norm": 0.0026893229223787785,
-      "learning_rate": 0.00019999083729756718,
-      "loss": 46.0,
-      "step": 26768
-    },
-    {
-      "epoch": 4.310962599138452,
-      "grad_norm": 0.008987111039459705,
-      "learning_rate": 0.0001999908366127078,
-      "loss": 46.0,
-      "step": 26769
-    },
-    {
-      "epoch": 4.311123636217239,
-      "grad_norm": 0.0015317542711272836,
-      "learning_rate": 0.00019999083592782283,
-      "loss": 46.0,
-      "step": 26770
-    },
-    {
-      "epoch": 4.311284673296027,
-      "grad_norm": 0.003954547457396984,
-      "learning_rate": 0.00019999083524291226,
-      "loss": 46.0,
-      "step": 26771
-    },
-    {
-      "epoch": 4.311445710374814,
-      "grad_norm": 0.012025341391563416,
-      "learning_rate": 0.0001999908345579761,
-      "loss": 46.0,
-      "step": 26772
-    },
-    {
-      "epoch": 4.3116067474536015,
-      "grad_norm": 0.004102442879229784,
-      "learning_rate": 0.00019999083387301433,
-      "loss": 46.0,
-      "step": 26773
-    },
-    {
-      "epoch": 4.311767784532389,
-      "grad_norm": 0.0034901387989521027,
-      "learning_rate": 0.000199990833188027,
-      "loss": 46.0,
-      "step": 26774
-    },
-    {
-      "epoch": 4.3119288216111755,
-      "grad_norm": 0.011448844335973263,
-      "learning_rate": 0.00019999083250301405,
-      "loss": 46.0,
-      "step": 26775
-    },
-    {
-      "epoch": 4.312089858689963,
-      "grad_norm": 0.007840665057301521,
-      "learning_rate": 0.00019999083181797552,
-      "loss": 46.0,
-      "step": 26776
-    },
-    {
-      "epoch": 4.31225089576875,
-      "grad_norm": 0.005458701401948929,
-      "learning_rate": 0.0001999908311329114,
-      "loss": 46.0,
-      "step": 26777
-    },
-    {
-      "epoch": 4.312411932847538,
-      "grad_norm": 0.0008306221570819616,
-      "learning_rate": 0.00019999083044782166,
-      "loss": 46.0,
-      "step": 26778
-    },
-    {
-      "epoch": 4.312572969926325,
-      "grad_norm": 0.011773806065320969,
-      "learning_rate": 0.00019999082976270637,
-      "loss": 46.0,
-      "step": 26779
-    },
-    {
-      "epoch": 4.312734007005113,
-      "grad_norm": 0.0029615748208016157,
-      "learning_rate": 0.00019999082907756546,
-      "loss": 46.0,
-      "step": 26780
-    },
-    {
-      "epoch": 4.3128950440839,
-      "grad_norm": 0.0031708742026239634,
-      "learning_rate": 0.000199990828392399,
-      "loss": 46.0,
-      "step": 26781
-    },
-    {
-      "epoch": 4.313056081162688,
-      "grad_norm": 0.0038983039557933807,
-      "learning_rate": 0.00019999082770720688,
-      "loss": 46.0,
-      "step": 26782
-    },
-    {
-      "epoch": 4.313217118241475,
-      "grad_norm": 0.0013287054607644677,
-      "learning_rate": 0.00019999082702198918,
-      "loss": 46.0,
-      "step": 26783
-    },
-    {
-      "epoch": 4.313378155320263,
-      "grad_norm": 0.002075494732707739,
-      "learning_rate": 0.00019999082633674592,
-      "loss": 46.0,
-      "step": 26784
-    },
-    {
-      "epoch": 4.31353919239905,
-      "grad_norm": 0.007053275592625141,
-      "learning_rate": 0.00019999082565147705,
-      "loss": 46.0,
-      "step": 26785
-    },
-    {
-      "epoch": 4.3137002294778375,
-      "grad_norm": 0.004405648913234472,
-      "learning_rate": 0.0001999908249661826,
-      "loss": 46.0,
-      "step": 26786
-    },
-    {
-      "epoch": 4.313861266556625,
-      "grad_norm": 0.002393609844148159,
-      "learning_rate": 0.00019999082428086255,
-      "loss": 46.0,
-      "step": 26787
-    },
-    {
-      "epoch": 4.314022303635412,
-      "grad_norm": 0.0010131018934771419,
-      "learning_rate": 0.00019999082359551691,
-      "loss": 46.0,
-      "step": 26788
-    },
-    {
-      "epoch": 4.3141833407142,
-      "grad_norm": 0.005945413839071989,
-      "learning_rate": 0.0001999908229101457,
-      "loss": 46.0,
-      "step": 26789
-    },
-    {
-      "epoch": 4.314344377792986,
-      "grad_norm": 0.002798818750306964,
-      "learning_rate": 0.00019999082222474886,
-      "loss": 46.0,
-      "step": 26790
-    },
-    {
-      "epoch": 4.314505414871774,
-      "grad_norm": 0.002516764448955655,
-      "learning_rate": 0.0001999908215393264,
-      "loss": 46.0,
-      "step": 26791
-    },
-    {
-      "epoch": 4.314666451950561,
-      "grad_norm": 0.002391841495409608,
-      "learning_rate": 0.0001999908208538784,
-      "loss": 46.0,
-      "step": 26792
-    },
-    {
-      "epoch": 4.314827489029349,
-      "grad_norm": 0.010734972544014454,
-      "learning_rate": 0.00019999082016840483,
-      "loss": 46.0,
-      "step": 26793
-    },
-    {
-      "epoch": 4.314988526108136,
-      "grad_norm": 0.002970040775835514,
-      "learning_rate": 0.00019999081948290562,
-      "loss": 46.0,
-      "step": 26794
-    },
-    {
-      "epoch": 4.315149563186924,
-      "grad_norm": 0.0029096766375005245,
-      "learning_rate": 0.00019999081879738082,
-      "loss": 46.0,
-      "step": 26795
-    },
-    {
-      "epoch": 4.315310600265711,
-      "grad_norm": 0.003329734317958355,
-      "learning_rate": 0.00019999081811183044,
-      "loss": 46.0,
-      "step": 26796
-    },
-    {
-      "epoch": 4.315471637344499,
-      "grad_norm": 0.0018089211080223322,
-      "learning_rate": 0.00019999081742625444,
-      "loss": 46.0,
-      "step": 26797
-    },
-    {
-      "epoch": 4.315632674423286,
-      "grad_norm": 0.0014738165773451328,
-      "learning_rate": 0.0001999908167406529,
-      "loss": 46.0,
-      "step": 26798
-    },
-    {
-      "epoch": 4.3157937115020735,
-      "grad_norm": 0.002472248859703541,
-      "learning_rate": 0.00019999081605502573,
-      "loss": 46.0,
-      "step": 26799
-    },
-    {
-      "epoch": 4.315954748580861,
-      "grad_norm": 0.0014111031778156757,
-      "learning_rate": 0.00019999081536937297,
-      "loss": 46.0,
-      "step": 26800
-    },
-    {
-      "epoch": 4.316115785659648,
-      "grad_norm": 0.007189365569502115,
-      "learning_rate": 0.00019999081468369463,
-      "loss": 46.0,
-      "step": 26801
-    },
-    {
-      "epoch": 4.316276822738436,
-      "grad_norm": 0.002926510525867343,
-      "learning_rate": 0.0001999908139979907,
-      "loss": 46.0,
-      "step": 26802
-    },
-    {
-      "epoch": 4.316437859817223,
-      "grad_norm": 0.009579833596944809,
-      "learning_rate": 0.00019999081331226117,
-      "loss": 46.0,
-      "step": 26803
-    },
-    {
-      "epoch": 4.316598896896011,
-      "grad_norm": 0.00287226727232337,
-      "learning_rate": 0.00019999081262650606,
-      "loss": 46.0,
-      "step": 26804
-    },
-    {
-      "epoch": 4.316759933974797,
-      "grad_norm": 0.011395324021577835,
-      "learning_rate": 0.00019999081194072533,
-      "loss": 46.0,
-      "step": 26805
-    },
-    {
-      "epoch": 4.316920971053585,
-      "grad_norm": 0.0065136500634253025,
-      "learning_rate": 0.00019999081125491902,
-      "loss": 46.0,
-      "step": 26806
-    },
-    {
-      "epoch": 4.317082008132372,
-      "grad_norm": 0.0027885367162525654,
-      "learning_rate": 0.00019999081056908713,
-      "loss": 46.0,
-      "step": 26807
-    },
-    {
-      "epoch": 4.31724304521116,
-      "grad_norm": 0.003528163768351078,
-      "learning_rate": 0.00019999080988322961,
-      "loss": 46.0,
-      "step": 26808
-    },
-    {
-      "epoch": 4.317404082289947,
-      "grad_norm": 0.0013375038979575038,
-      "learning_rate": 0.00019999080919734654,
-      "loss": 46.0,
-      "step": 26809
-    },
-    {
-      "epoch": 4.3175651193687345,
-      "grad_norm": 0.004187578801065683,
-      "learning_rate": 0.00019999080851143788,
-      "loss": 46.0,
-      "step": 26810
-    },
-    {
-      "epoch": 4.317726156447522,
-      "grad_norm": 0.004302739631384611,
-      "learning_rate": 0.00019999080782550358,
-      "loss": 46.0,
-      "step": 26811
-    },
-    {
-      "epoch": 4.3178871935263095,
-      "grad_norm": 0.0018789521418511868,
-      "learning_rate": 0.00019999080713954372,
-      "loss": 46.0,
-      "step": 26812
-    },
-    {
-      "epoch": 4.318048230605097,
-      "grad_norm": 0.00590970553457737,
-      "learning_rate": 0.00019999080645355827,
-      "loss": 46.0,
-      "step": 26813
-    },
-    {
-      "epoch": 4.318209267683884,
-      "grad_norm": 0.011279669590294361,
-      "learning_rate": 0.00019999080576754724,
-      "loss": 46.0,
-      "step": 26814
-    },
-    {
-      "epoch": 4.318370304762672,
-      "grad_norm": 0.020957602187991142,
-      "learning_rate": 0.00019999080508151061,
-      "loss": 46.0,
-      "step": 26815
-    },
-    {
-      "epoch": 4.318531341841459,
-      "grad_norm": 0.004916503094136715,
-      "learning_rate": 0.00019999080439544838,
-      "loss": 46.0,
-      "step": 26816
-    },
-    {
-      "epoch": 4.318692378920247,
-      "grad_norm": 0.0006075624260120094,
-      "learning_rate": 0.00019999080370936053,
-      "loss": 46.0,
-      "step": 26817
-    },
-    {
-      "epoch": 4.318853415999034,
-      "grad_norm": 0.0012110215611755848,
-      "learning_rate": 0.00019999080302324714,
-      "loss": 46.0,
-      "step": 26818
-    },
-    {
-      "epoch": 4.319014453077822,
-      "grad_norm": 0.005387540906667709,
-      "learning_rate": 0.00019999080233710812,
-      "loss": 46.0,
-      "step": 26819
-    },
-    {
-      "epoch": 4.319175490156608,
-      "grad_norm": 0.0017627552151679993,
-      "learning_rate": 0.0001999908016509435,
-      "loss": 46.0,
-      "step": 26820
-    },
-    {
-      "epoch": 4.319336527235396,
-      "grad_norm": 0.0028475935105234385,
-      "learning_rate": 0.0001999908009647533,
-      "loss": 46.0,
-      "step": 26821
-    },
-    {
-      "epoch": 4.319497564314183,
-      "grad_norm": 0.0012488632928580046,
-      "learning_rate": 0.00019999080027853752,
-      "loss": 46.0,
-      "step": 26822
-    },
-    {
-      "epoch": 4.3196586013929705,
-      "grad_norm": 0.0023156683892011642,
-      "learning_rate": 0.00019999079959229614,
-      "loss": 46.0,
-      "step": 26823
-    },
-    {
-      "epoch": 4.319819638471758,
-      "grad_norm": 0.001889853272587061,
-      "learning_rate": 0.00019999079890602916,
-      "loss": 46.0,
-      "step": 26824
-    },
-    {
-      "epoch": 4.319980675550545,
-      "grad_norm": 0.002063898602500558,
-      "learning_rate": 0.0001999907982197366,
-      "loss": 46.0,
-      "step": 26825
-    },
-    {
-      "epoch": 4.320141712629333,
-      "grad_norm": 0.0012143782805651426,
-      "learning_rate": 0.00019999079753341844,
-      "loss": 46.0,
-      "step": 26826
-    },
-    {
-      "epoch": 4.32030274970812,
-      "grad_norm": 0.0013486681273207068,
-      "learning_rate": 0.0001999907968470747,
-      "loss": 46.0,
-      "step": 26827
-    },
-    {
-      "epoch": 4.320463786786908,
-      "grad_norm": 0.008054655976593494,
-      "learning_rate": 0.00019999079616070536,
-      "loss": 46.0,
-      "step": 26828
-    },
-    {
-      "epoch": 4.320624823865695,
-      "grad_norm": 0.0014415632467716932,
-      "learning_rate": 0.0001999907954743104,
-      "loss": 46.0,
-      "step": 26829
-    },
-    {
-      "epoch": 4.320785860944483,
-      "grad_norm": 0.0015651858411729336,
-      "learning_rate": 0.0001999907947878899,
-      "loss": 46.0,
-      "step": 26830
-    },
-    {
-      "epoch": 4.32094689802327,
-      "grad_norm": 0.003934207838028669,
-      "learning_rate": 0.00019999079410144374,
-      "loss": 46.0,
-      "step": 26831
-    },
-    {
-      "epoch": 4.321107935102058,
-      "grad_norm": 0.0015107926446944475,
-      "learning_rate": 0.00019999079341497206,
-      "loss": 46.0,
-      "step": 26832
-    },
-    {
-      "epoch": 4.321268972180845,
-      "grad_norm": 0.002687514293938875,
-      "learning_rate": 0.00019999079272847473,
-      "loss": 46.0,
-      "step": 26833
-    },
-    {
-      "epoch": 4.3214300092596325,
-      "grad_norm": 0.0033835256472229958,
-      "learning_rate": 0.00019999079204195184,
-      "loss": 46.0,
-      "step": 26834
-    },
-    {
-      "epoch": 4.321591046338419,
-      "grad_norm": 0.005868611391633749,
-      "learning_rate": 0.00019999079135540334,
-      "loss": 46.0,
-      "step": 26835
-    },
-    {
-      "epoch": 4.3217520834172065,
-      "grad_norm": 0.006264748051762581,
-      "learning_rate": 0.00019999079066882928,
-      "loss": 46.0,
-      "step": 26836
-    },
-    {
-      "epoch": 4.321913120495994,
-      "grad_norm": 0.002757747657597065,
-      "learning_rate": 0.0001999907899822296,
-      "loss": 46.0,
-      "step": 26837
-    },
-    {
-      "epoch": 4.322074157574781,
-      "grad_norm": 0.005958610214293003,
-      "learning_rate": 0.00019999078929560434,
-      "loss": 46.0,
-      "step": 26838
-    },
-    {
-      "epoch": 4.322235194653569,
-      "grad_norm": 0.004291116259992123,
-      "learning_rate": 0.00019999078860895346,
-      "loss": 46.0,
-      "step": 26839
-    },
-    {
-      "epoch": 4.322396231732356,
-      "grad_norm": 0.0012658450286835432,
-      "learning_rate": 0.000199990787922277,
-      "loss": 46.0,
-      "step": 26840
-    },
-    {
-      "epoch": 4.322557268811144,
-      "grad_norm": 0.0006600506021641195,
-      "learning_rate": 0.00019999078723557497,
-      "loss": 46.0,
-      "step": 26841
-    },
-    {
-      "epoch": 4.322718305889931,
-      "grad_norm": 0.006315393839031458,
-      "learning_rate": 0.00019999078654884734,
-      "loss": 46.0,
-      "step": 26842
-    },
-    {
-      "epoch": 4.322879342968719,
-      "grad_norm": 0.012058478780090809,
-      "learning_rate": 0.0001999907858620941,
-      "loss": 46.0,
-      "step": 26843
-    },
-    {
-      "epoch": 4.323040380047506,
-      "grad_norm": 0.0019039458129554987,
-      "learning_rate": 0.0001999907851753153,
-      "loss": 46.0,
-      "step": 26844
-    },
-    {
-      "epoch": 4.323201417126294,
-      "grad_norm": 0.0019165630219504237,
-      "learning_rate": 0.00019999078448851084,
-      "loss": 46.0,
-      "step": 26845
-    },
-    {
-      "epoch": 4.323362454205081,
-      "grad_norm": 0.0020978557877242565,
-      "learning_rate": 0.00019999078380168083,
-      "loss": 46.0,
-      "step": 26846
-    },
-    {
-      "epoch": 4.3235234912838685,
-      "grad_norm": 0.006783506833016872,
-      "learning_rate": 0.00019999078311482525,
-      "loss": 46.0,
-      "step": 26847
-    },
-    {
-      "epoch": 4.323684528362655,
-      "grad_norm": 0.009774120524525642,
-      "learning_rate": 0.00019999078242794406,
-      "loss": 46.0,
-      "step": 26848
-    },
-    {
-      "epoch": 4.3238455654414425,
-      "grad_norm": 0.0018163928762078285,
-      "learning_rate": 0.00019999078174103726,
-      "loss": 46.0,
-      "step": 26849
-    },
-    {
-      "epoch": 4.32400660252023,
-      "grad_norm": 0.0021257009357213974,
-      "learning_rate": 0.0001999907810541049,
-      "loss": 46.0,
-      "step": 26850
-    },
-    {
-      "epoch": 4.324167639599017,
-      "grad_norm": 0.004128101281821728,
-      "learning_rate": 0.00019999078036714695,
-      "loss": 46.0,
-      "step": 26851
-    },
-    {
-      "epoch": 4.324328676677805,
-      "grad_norm": 0.004893376957625151,
-      "learning_rate": 0.00019999077968016336,
-      "loss": 46.0,
-      "step": 26852
-    },
-    {
-      "epoch": 4.324489713756592,
-      "grad_norm": 0.0020878843497484922,
-      "learning_rate": 0.0001999907789931542,
-      "loss": 46.0,
-      "step": 26853
-    },
-    {
-      "epoch": 4.32465075083538,
-      "grad_norm": 0.004224977921694517,
-      "learning_rate": 0.00019999077830611946,
-      "loss": 46.0,
-      "step": 26854
-    },
-    {
-      "epoch": 4.324811787914167,
-      "grad_norm": 0.0030052324291318655,
-      "learning_rate": 0.0001999907776190591,
-      "loss": 46.0,
-      "step": 26855
-    },
-    {
-      "epoch": 4.324972824992955,
-      "grad_norm": 0.015664605423808098,
-      "learning_rate": 0.0001999907769319732,
-      "loss": 46.0,
-      "step": 26856
-    },
-    {
-      "epoch": 4.325133862071742,
-      "grad_norm": 0.020176611840724945,
-      "learning_rate": 0.00019999077624486167,
-      "loss": 46.0,
-      "step": 26857
-    },
-    {
-      "epoch": 4.32529489915053,
-      "grad_norm": 0.009067884646356106,
-      "learning_rate": 0.00019999077555772453,
-      "loss": 46.0,
-      "step": 26858
-    },
-    {
-      "epoch": 4.325455936229317,
-      "grad_norm": 0.011108560487627983,
-      "learning_rate": 0.00019999077487056183,
-      "loss": 46.0,
-      "step": 26859
-    },
-    {
-      "epoch": 4.3256169733081045,
-      "grad_norm": 0.002188031328842044,
-      "learning_rate": 0.00019999077418337354,
-      "loss": 46.0,
-      "step": 26860
-    },
-    {
-      "epoch": 4.325778010386892,
-      "grad_norm": 0.007436261046677828,
-      "learning_rate": 0.00019999077349615963,
-      "loss": 46.0,
-      "step": 26861
-    },
-    {
-      "epoch": 4.325939047465679,
-      "grad_norm": 0.0025805607438087463,
-      "learning_rate": 0.00019999077280892014,
-      "loss": 46.0,
-      "step": 26862
-    },
-    {
-      "epoch": 4.326100084544466,
-      "grad_norm": 0.003826677333563566,
-      "learning_rate": 0.00019999077212165504,
-      "loss": 46.0,
-      "step": 26863
-    },
-    {
-      "epoch": 4.326261121623253,
-      "grad_norm": 0.004517124034464359,
-      "learning_rate": 0.00019999077143436437,
-      "loss": 46.0,
-      "step": 26864
-    },
-    {
-      "epoch": 4.326422158702041,
-      "grad_norm": 0.002376899356022477,
-      "learning_rate": 0.00019999077074704812,
-      "loss": 46.0,
-      "step": 26865
-    },
-    {
-      "epoch": 4.326583195780828,
-      "grad_norm": 0.004488472826778889,
-      "learning_rate": 0.00019999077005970626,
-      "loss": 46.0,
-      "step": 26866
-    },
-    {
-      "epoch": 4.326744232859616,
-      "grad_norm": 0.003915915731340647,
-      "learning_rate": 0.0001999907693723388,
-      "loss": 46.0,
-      "step": 26867
-    },
-    {
-      "epoch": 4.326905269938403,
-      "grad_norm": 0.0010685756569728255,
-      "learning_rate": 0.00019999076868494576,
-      "loss": 46.0,
-      "step": 26868
-    },
-    {
-      "epoch": 4.327066307017191,
-      "grad_norm": 0.0020829103887081146,
-      "learning_rate": 0.0001999907679975271,
-      "loss": 46.0,
-      "step": 26869
-    },
-    {
-      "epoch": 4.327227344095978,
-      "grad_norm": 0.0040535940788686275,
-      "learning_rate": 0.0001999907673100829,
-      "loss": 46.0,
-      "step": 26870
-    },
-    {
-      "epoch": 4.3273883811747655,
-      "grad_norm": 0.010978455655276775,
-      "learning_rate": 0.00019999076662261306,
-      "loss": 46.0,
-      "step": 26871
-    },
-    {
-      "epoch": 4.327549418253553,
-      "grad_norm": 0.0008613588288426399,
-      "learning_rate": 0.00019999076593511767,
-      "loss": 46.0,
-      "step": 26872
-    },
-    {
-      "epoch": 4.3277104553323404,
-      "grad_norm": 0.003091295948252082,
-      "learning_rate": 0.00019999076524759667,
-      "loss": 46.0,
-      "step": 26873
-    },
-    {
-      "epoch": 4.327871492411128,
-      "grad_norm": 0.004010102245956659,
-      "learning_rate": 0.00019999076456005005,
-      "loss": 46.0,
-      "step": 26874
-    },
-    {
-      "epoch": 4.328032529489915,
-      "grad_norm": 0.0014021600363776088,
-      "learning_rate": 0.00019999076387247787,
-      "loss": 46.0,
-      "step": 26875
-    },
-    {
-      "epoch": 4.328193566568703,
-      "grad_norm": 0.0030932254157960415,
-      "learning_rate": 0.00019999076318488008,
-      "loss": 46.0,
-      "step": 26876
-    },
-    {
-      "epoch": 4.32835460364749,
-      "grad_norm": 0.0017301643965765834,
-      "learning_rate": 0.0001999907624972567,
-      "loss": 46.0,
-      "step": 26877
-    },
-    {
-      "epoch": 4.328515640726277,
-      "grad_norm": 0.006087945774197578,
-      "learning_rate": 0.00019999076180960773,
-      "loss": 46.0,
-      "step": 26878
-    },
-    {
-      "epoch": 4.328676677805064,
-      "grad_norm": 0.003597947070375085,
-      "learning_rate": 0.00019999076112193318,
-      "loss": 46.0,
-      "step": 26879
-    },
-    {
-      "epoch": 4.328837714883852,
-      "grad_norm": 0.03220827132463455,
-      "learning_rate": 0.00019999076043423304,
-      "loss": 46.0,
-      "step": 26880
-    },
-    {
-      "epoch": 4.328998751962639,
-      "grad_norm": 0.005902218632400036,
-      "learning_rate": 0.00019999075974650728,
-      "loss": 46.0,
-      "step": 26881
-    },
-    {
-      "epoch": 4.329159789041427,
-      "grad_norm": 0.008851482532918453,
-      "learning_rate": 0.00019999075905875591,
-      "loss": 46.0,
-      "step": 26882
-    },
-    {
-      "epoch": 4.329320826120214,
-      "grad_norm": 0.006018993444740772,
-      "learning_rate": 0.000199990758370979,
-      "loss": 46.0,
-      "step": 26883
-    },
-    {
-      "epoch": 4.3294818631990015,
-      "grad_norm": 0.0024594860151410103,
-      "learning_rate": 0.00019999075768317647,
-      "loss": 46.0,
-      "step": 26884
-    },
-    {
-      "epoch": 4.329642900277789,
-      "grad_norm": 0.003404195187613368,
-      "learning_rate": 0.00019999075699534836,
-      "loss": 46.0,
-      "step": 26885
-    },
-    {
-      "epoch": 4.329803937356576,
-      "grad_norm": 0.003302328521385789,
-      "learning_rate": 0.00019999075630749467,
-      "loss": 46.0,
-      "step": 26886
-    },
-    {
-      "epoch": 4.329964974435364,
-      "grad_norm": 0.0013579169753938913,
-      "learning_rate": 0.00019999075561961537,
-      "loss": 46.0,
-      "step": 26887
-    },
-    {
-      "epoch": 4.330126011514151,
-      "grad_norm": 0.0020914459601044655,
-      "learning_rate": 0.00019999075493171048,
-      "loss": 46.0,
-      "step": 26888
-    },
-    {
-      "epoch": 4.330287048592939,
-      "grad_norm": 0.005633731372654438,
-      "learning_rate": 0.00019999075424377997,
-      "loss": 46.0,
-      "step": 26889
-    },
-    {
-      "epoch": 4.330448085671726,
-      "grad_norm": 0.00464225048199296,
-      "learning_rate": 0.0001999907535558239,
-      "loss": 46.0,
-      "step": 26890
-    },
-    {
-      "epoch": 4.330609122750514,
-      "grad_norm": 0.0026460890658199787,
-      "learning_rate": 0.00019999075286784225,
-      "loss": 46.0,
-      "step": 26891
-    },
-    {
-      "epoch": 4.330770159829301,
-      "grad_norm": 0.004468322731554508,
-      "learning_rate": 0.00019999075217983495,
-      "loss": 46.0,
-      "step": 26892
-    },
-    {
-      "epoch": 4.330931196908088,
-      "grad_norm": 0.002827275777235627,
-      "learning_rate": 0.0001999907514918021,
-      "loss": 46.0,
-      "step": 26893
-    },
-    {
-      "epoch": 4.331092233986875,
-      "grad_norm": 0.007451838813722134,
-      "learning_rate": 0.00019999075080374365,
-      "loss": 46.0,
-      "step": 26894
-    },
-    {
-      "epoch": 4.331253271065663,
-      "grad_norm": 0.012124565429985523,
-      "learning_rate": 0.00019999075011565962,
-      "loss": 46.0,
-      "step": 26895
-    },
-    {
-      "epoch": 4.33141430814445,
-      "grad_norm": 0.002243731403723359,
-      "learning_rate": 0.00019999074942754998,
-      "loss": 46.0,
-      "step": 26896
-    },
-    {
-      "epoch": 4.3315753452232375,
-      "grad_norm": 0.0029568923637270927,
-      "learning_rate": 0.00019999074873941478,
-      "loss": 46.0,
-      "step": 26897
-    },
-    {
-      "epoch": 4.331736382302025,
-      "grad_norm": 0.008408868685364723,
-      "learning_rate": 0.00019999074805125396,
-      "loss": 46.0,
-      "step": 26898
-    },
-    {
-      "epoch": 4.331897419380812,
-      "grad_norm": 0.021713390946388245,
-      "learning_rate": 0.00019999074736306752,
-      "loss": 46.0,
-      "step": 26899
-    },
-    {
-      "epoch": 4.3320584564596,
-      "grad_norm": 0.0030565140768885612,
-      "learning_rate": 0.00019999074667485553,
-      "loss": 46.0,
-      "step": 26900
-    },
-    {
-      "epoch": 4.332219493538387,
-      "grad_norm": 0.010622644796967506,
-      "learning_rate": 0.00019999074598661792,
-      "loss": 46.0,
-      "step": 26901
-    },
-    {
-      "epoch": 4.332380530617175,
-      "grad_norm": 0.0027988175861537457,
-      "learning_rate": 0.00019999074529835476,
-      "loss": 46.0,
-      "step": 26902
-    },
-    {
-      "epoch": 4.332541567695962,
-      "grad_norm": 0.0014003480318933725,
-      "learning_rate": 0.00019999074461006595,
-      "loss": 46.0,
-      "step": 26903
-    },
-    {
-      "epoch": 4.33270260477475,
-      "grad_norm": 0.002058149315416813,
-      "learning_rate": 0.00019999074392175158,
-      "loss": 46.0,
-      "step": 26904
-    },
-    {
-      "epoch": 4.332863641853537,
-      "grad_norm": 0.002338825026527047,
-      "learning_rate": 0.0001999907432334116,
-      "loss": 46.0,
-      "step": 26905
-    },
-    {
-      "epoch": 4.333024678932325,
-      "grad_norm": 0.0016602956457063556,
-      "learning_rate": 0.00019999074254504605,
-      "loss": 46.0,
-      "step": 26906
-    },
-    {
-      "epoch": 4.333185716011112,
-      "grad_norm": 0.011720615439116955,
-      "learning_rate": 0.0001999907418566549,
-      "loss": 46.0,
-      "step": 26907
-    },
-    {
-      "epoch": 4.333346753089899,
-      "grad_norm": 0.0005688276141881943,
-      "learning_rate": 0.00019999074116823815,
-      "loss": 46.0,
-      "step": 26908
-    },
-    {
-      "epoch": 4.333507790168686,
-      "grad_norm": 0.002005399437621236,
-      "learning_rate": 0.00019999074047979582,
-      "loss": 46.0,
-      "step": 26909
-    },
-    {
-      "epoch": 4.3336688272474735,
-      "grad_norm": 0.0006501065799966455,
-      "learning_rate": 0.0001999907397913279,
-      "loss": 46.0,
-      "step": 26910
-    },
-    {
-      "epoch": 4.333829864326261,
-      "grad_norm": 0.006243292707949877,
-      "learning_rate": 0.0001999907391028344,
-      "loss": 46.0,
-      "step": 26911
-    },
-    {
-      "epoch": 4.333990901405048,
-      "grad_norm": 0.0007953814929351211,
-      "learning_rate": 0.00019999073841431525,
-      "loss": 46.0,
-      "step": 26912
-    },
-    {
-      "epoch": 4.334151938483836,
-      "grad_norm": 0.0014161694562062621,
-      "learning_rate": 0.00019999073772577054,
-      "loss": 46.0,
-      "step": 26913
-    },
-    {
-      "epoch": 4.334312975562623,
-      "grad_norm": 0.0015716011403128505,
-      "learning_rate": 0.00019999073703720025,
-      "loss": 46.0,
-      "step": 26914
-    },
-    {
-      "epoch": 4.334474012641411,
-      "grad_norm": 0.007059035357087851,
-      "learning_rate": 0.00019999073634860437,
-      "loss": 46.0,
-      "step": 26915
-    },
-    {
-      "epoch": 4.334635049720198,
-      "grad_norm": 0.00320522696711123,
-      "learning_rate": 0.00019999073565998287,
-      "loss": 46.0,
-      "step": 26916
-    },
-    {
-      "epoch": 4.334796086798986,
-      "grad_norm": 0.0024273826275020838,
-      "learning_rate": 0.00019999073497133581,
-      "loss": 46.0,
-      "step": 26917
-    },
-    {
-      "epoch": 4.334957123877773,
-      "grad_norm": 0.006734349764883518,
-      "learning_rate": 0.00019999073428266314,
-      "loss": 46.0,
-      "step": 26918
-    },
-    {
-      "epoch": 4.3351181609565606,
-      "grad_norm": 0.004628733266144991,
-      "learning_rate": 0.00019999073359396489,
-      "loss": 46.0,
-      "step": 26919
-    },
-    {
-      "epoch": 4.335279198035348,
-      "grad_norm": 0.00761287659406662,
-      "learning_rate": 0.00019999073290524101,
-      "loss": 46.0,
-      "step": 26920
-    },
-    {
-      "epoch": 4.335440235114135,
-      "grad_norm": 0.011748096905648708,
-      "learning_rate": 0.00019999073221649158,
-      "loss": 46.0,
-      "step": 26921
-    },
-    {
-      "epoch": 4.335601272192922,
-      "grad_norm": 0.011996623128652573,
-      "learning_rate": 0.00019999073152771654,
-      "loss": 46.0,
-      "step": 26922
-    },
-    {
-      "epoch": 4.3357623092717095,
-      "grad_norm": 0.005394014064222574,
-      "learning_rate": 0.0001999907308389159,
-      "loss": 46.0,
-      "step": 26923
-    },
-    {
-      "epoch": 4.335923346350497,
-      "grad_norm": 0.003779176389798522,
-      "learning_rate": 0.00019999073015008968,
-      "loss": 46.0,
-      "step": 26924
-    },
-    {
-      "epoch": 4.336084383429284,
-      "grad_norm": 0.0032469138968735933,
-      "learning_rate": 0.00019999072946123788,
-      "loss": 46.0,
-      "step": 26925
-    },
-    {
-      "epoch": 4.336245420508072,
-      "grad_norm": 0.002185650635510683,
-      "learning_rate": 0.00019999072877236045,
-      "loss": 46.0,
-      "step": 26926
-    },
-    {
-      "epoch": 4.336406457586859,
-      "grad_norm": 0.001808663597330451,
-      "learning_rate": 0.00019999072808345747,
-      "loss": 46.0,
-      "step": 26927
-    },
-    {
-      "epoch": 4.336567494665647,
-      "grad_norm": 0.0010025043739005923,
-      "learning_rate": 0.00019999072739452887,
-      "loss": 46.0,
-      "step": 26928
-    },
-    {
-      "epoch": 4.336728531744434,
-      "grad_norm": 0.0033047625329345465,
-      "learning_rate": 0.0001999907267055747,
-      "loss": 46.0,
-      "step": 26929
-    },
-    {
-      "epoch": 4.336889568823222,
-      "grad_norm": 0.013785278424620628,
-      "learning_rate": 0.0001999907260165949,
-      "loss": 46.0,
-      "step": 26930
-    },
-    {
-      "epoch": 4.337050605902009,
-      "grad_norm": 0.005489861126989126,
-      "learning_rate": 0.00019999072532758954,
-      "loss": 46.0,
-      "step": 26931
-    },
-    {
-      "epoch": 4.3372116429807965,
-      "grad_norm": 0.004444201942533255,
-      "learning_rate": 0.00019999072463855856,
-      "loss": 46.0,
-      "step": 26932
-    },
-    {
-      "epoch": 4.337372680059584,
-      "grad_norm": 0.0026683283504098654,
-      "learning_rate": 0.00019999072394950203,
-      "loss": 46.0,
-      "step": 26933
-    },
-    {
-      "epoch": 4.337533717138371,
-      "grad_norm": 0.0037998268380761147,
-      "learning_rate": 0.00019999072326041986,
-      "loss": 46.0,
-      "step": 26934
-    },
-    {
-      "epoch": 4.337694754217159,
-      "grad_norm": 0.009480730630457401,
-      "learning_rate": 0.0001999907225713121,
-      "loss": 46.0,
-      "step": 26935
-    },
-    {
-      "epoch": 4.3378557912959455,
-      "grad_norm": 0.015174102038145065,
-      "learning_rate": 0.00019999072188217878,
-      "loss": 46.0,
-      "step": 26936
-    },
-    {
-      "epoch": 4.338016828374733,
-      "grad_norm": 0.0023719877935945988,
-      "learning_rate": 0.00019999072119301987,
-      "loss": 46.0,
-      "step": 26937
-    },
-    {
-      "epoch": 4.33817786545352,
-      "grad_norm": 0.008152016438543797,
-      "learning_rate": 0.00019999072050383535,
-      "loss": 46.0,
-      "step": 26938
-    },
-    {
-      "epoch": 4.338338902532308,
-      "grad_norm": 0.0032736333087086678,
-      "learning_rate": 0.00019999071981462524,
-      "loss": 46.0,
-      "step": 26939
-    },
-    {
-      "epoch": 4.338499939611095,
-      "grad_norm": 0.008397176861763,
-      "learning_rate": 0.00019999071912538954,
-      "loss": 46.0,
-      "step": 26940
-    },
-    {
-      "epoch": 4.338660976689883,
-      "grad_norm": 0.000996448565274477,
-      "learning_rate": 0.00019999071843612823,
-      "loss": 46.0,
-      "step": 26941
-    },
-    {
-      "epoch": 4.33882201376867,
-      "grad_norm": 0.007727055344730616,
-      "learning_rate": 0.00019999071774684133,
-      "loss": 46.0,
-      "step": 26942
-    },
-    {
-      "epoch": 4.338983050847458,
-      "grad_norm": 0.004088437184691429,
-      "learning_rate": 0.00019999071705752885,
-      "loss": 46.0,
-      "step": 26943
-    },
-    {
-      "epoch": 4.339144087926245,
-      "grad_norm": 0.0048957462422549725,
-      "learning_rate": 0.0001999907163681908,
-      "loss": 46.0,
-      "step": 26944
-    },
-    {
-      "epoch": 4.3393051250050325,
-      "grad_norm": 0.00665948074311018,
-      "learning_rate": 0.00019999071567882712,
-      "loss": 46.0,
-      "step": 26945
-    },
-    {
-      "epoch": 4.33946616208382,
-      "grad_norm": 0.010340073145925999,
-      "learning_rate": 0.00019999071498943787,
-      "loss": 46.0,
-      "step": 26946
-    },
-    {
-      "epoch": 4.339627199162607,
-      "grad_norm": 0.00218792911618948,
-      "learning_rate": 0.000199990714300023,
-      "loss": 46.0,
-      "step": 26947
-    },
-    {
-      "epoch": 4.339788236241395,
-      "grad_norm": 0.009167652577161789,
-      "learning_rate": 0.00019999071361058256,
-      "loss": 46.0,
-      "step": 26948
-    },
-    {
-      "epoch": 4.339949273320182,
-      "grad_norm": 0.003372750710695982,
-      "learning_rate": 0.00019999071292111652,
-      "loss": 46.0,
-      "step": 26949
-    },
-    {
-      "epoch": 4.34011031039897,
-      "grad_norm": 0.005344822537153959,
-      "learning_rate": 0.0001999907122316249,
-      "loss": 46.0,
-      "step": 26950
-    },
-    {
-      "epoch": 4.340271347477756,
-      "grad_norm": 0.00749730970710516,
-      "learning_rate": 0.0001999907115421077,
-      "loss": 46.0,
-      "step": 26951
-    },
-    {
-      "epoch": 4.340432384556544,
-      "grad_norm": 0.0029990843031555414,
-      "learning_rate": 0.00019999071085256487,
-      "loss": 46.0,
-      "step": 26952
-    },
-    {
-      "epoch": 4.340593421635331,
-      "grad_norm": 0.0047389790415763855,
-      "learning_rate": 0.00019999071016299646,
-      "loss": 46.0,
-      "step": 26953
-    },
-    {
-      "epoch": 4.340754458714119,
-      "grad_norm": 0.005430468823760748,
-      "learning_rate": 0.00019999070947340246,
-      "loss": 46.0,
-      "step": 26954
-    },
-    {
-      "epoch": 4.340915495792906,
-      "grad_norm": 0.0030369332525879145,
-      "learning_rate": 0.00019999070878378287,
-      "loss": 46.0,
-      "step": 26955
-    },
-    {
-      "epoch": 4.341076532871694,
-      "grad_norm": 0.002362895989790559,
-      "learning_rate": 0.0001999907080941377,
-      "loss": 46.0,
-      "step": 26956
-    },
-    {
-      "epoch": 4.341237569950481,
-      "grad_norm": 0.0035293116234242916,
-      "learning_rate": 0.00019999070740446691,
-      "loss": 46.0,
-      "step": 26957
-    },
-    {
-      "epoch": 4.3413986070292685,
-      "grad_norm": 0.0018974484410136938,
-      "learning_rate": 0.00019999070671477057,
-      "loss": 46.0,
-      "step": 26958
-    },
-    {
-      "epoch": 4.341559644108056,
-      "grad_norm": 0.005015982314944267,
-      "learning_rate": 0.00019999070602504858,
-      "loss": 46.0,
-      "step": 26959
-    },
-    {
-      "epoch": 4.341720681186843,
-      "grad_norm": 0.001991160213947296,
-      "learning_rate": 0.00019999070533530103,
-      "loss": 46.0,
-      "step": 26960
-    },
-    {
-      "epoch": 4.341881718265631,
-      "grad_norm": 0.004959786776453257,
-      "learning_rate": 0.00019999070464552787,
-      "loss": 46.0,
-      "step": 26961
-    },
-    {
-      "epoch": 4.342042755344418,
-      "grad_norm": 0.002750984625890851,
-      "learning_rate": 0.00019999070395572917,
-      "loss": 46.0,
-      "step": 26962
-    },
-    {
-      "epoch": 4.342203792423206,
-      "grad_norm": 0.010296100750565529,
-      "learning_rate": 0.00019999070326590484,
-      "loss": 46.0,
-      "step": 26963
-    },
-    {
-      "epoch": 4.342364829501993,
-      "grad_norm": 0.00532003166154027,
-      "learning_rate": 0.00019999070257605488,
-      "loss": 46.0,
-      "step": 26964
-    },
-    {
-      "epoch": 4.342525866580781,
-      "grad_norm": 0.009289945475757122,
-      "learning_rate": 0.0001999907018861794,
-      "loss": 46.0,
-      "step": 26965
-    },
-    {
-      "epoch": 4.342686903659567,
-      "grad_norm": 0.010900177992880344,
-      "learning_rate": 0.00019999070119627827,
-      "loss": 46.0,
-      "step": 26966
-    },
-    {
-      "epoch": 4.342847940738355,
-      "grad_norm": 0.0026615154929459095,
-      "learning_rate": 0.0001999907005063516,
-      "loss": 46.0,
-      "step": 26967
-    },
-    {
-      "epoch": 4.343008977817142,
-      "grad_norm": 0.006060689687728882,
-      "learning_rate": 0.00019999069981639926,
-      "loss": 46.0,
-      "step": 26968
-    },
-    {
-      "epoch": 4.34317001489593,
-      "grad_norm": 0.004727237857878208,
-      "learning_rate": 0.00019999069912642137,
-      "loss": 46.0,
-      "step": 26969
-    },
-    {
-      "epoch": 4.343331051974717,
-      "grad_norm": 0.0015523300971835852,
-      "learning_rate": 0.0001999906984364179,
-      "loss": 46.0,
-      "step": 26970
-    },
-    {
-      "epoch": 4.3434920890535045,
-      "grad_norm": 0.010069606825709343,
-      "learning_rate": 0.00019999069774638884,
-      "loss": 46.0,
-      "step": 26971
-    },
-    {
-      "epoch": 4.343653126132292,
-      "grad_norm": 0.0022206956055015326,
-      "learning_rate": 0.00019999069705633416,
-      "loss": 46.0,
-      "step": 26972
-    },
-    {
-      "epoch": 4.343814163211079,
-      "grad_norm": 0.004480462521314621,
-      "learning_rate": 0.00019999069636625393,
-      "loss": 46.0,
-      "step": 26973
-    },
-    {
-      "epoch": 4.343975200289867,
-      "grad_norm": 0.005475439131259918,
-      "learning_rate": 0.00019999069567614808,
-      "loss": 46.0,
-      "step": 26974
-    },
-    {
-      "epoch": 4.344136237368654,
-      "grad_norm": 0.0070614987052977085,
-      "learning_rate": 0.00019999069498601664,
-      "loss": 46.0,
-      "step": 26975
-    },
-    {
-      "epoch": 4.344297274447442,
-      "grad_norm": 0.002243722788989544,
-      "learning_rate": 0.0001999906942958596,
-      "loss": 46.0,
-      "step": 26976
-    },
-    {
-      "epoch": 4.344458311526229,
-      "grad_norm": 0.006641100160777569,
-      "learning_rate": 0.00019999069360567698,
-      "loss": 46.0,
-      "step": 26977
-    },
-    {
-      "epoch": 4.344619348605017,
-      "grad_norm": 0.0038843858055770397,
-      "learning_rate": 0.00019999069291546878,
-      "loss": 46.0,
-      "step": 26978
-    },
-    {
-      "epoch": 4.344780385683804,
-      "grad_norm": 0.003903406672179699,
-      "learning_rate": 0.00019999069222523494,
-      "loss": 46.0,
-      "step": 26979
-    },
-    {
-      "epoch": 4.3449414227625915,
-      "grad_norm": 0.006455189548432827,
-      "learning_rate": 0.00019999069153497554,
-      "loss": 46.0,
-      "step": 26980
-    },
-    {
-      "epoch": 4.345102459841378,
-      "grad_norm": 0.006002070847898722,
-      "learning_rate": 0.00019999069084469055,
-      "loss": 46.0,
-      "step": 26981
-    },
-    {
-      "epoch": 4.345263496920166,
-      "grad_norm": 0.0072707985527813435,
-      "learning_rate": 0.00019999069015437995,
-      "loss": 46.0,
-      "step": 26982
-    },
-    {
-      "epoch": 4.345424533998953,
-      "grad_norm": 0.004200181923806667,
-      "learning_rate": 0.0001999906894640438,
-      "loss": 46.0,
-      "step": 26983
-    },
-    {
-      "epoch": 4.3455855710777405,
-      "grad_norm": 0.015590599738061428,
-      "learning_rate": 0.00019999068877368198,
-      "loss": 46.0,
-      "step": 26984
-    },
-    {
-      "epoch": 4.345746608156528,
-      "grad_norm": 0.007098034955561161,
-      "learning_rate": 0.00019999068808329465,
-      "loss": 46.0,
-      "step": 26985
-    },
-    {
-      "epoch": 4.345907645235315,
-      "grad_norm": 0.014213014394044876,
-      "learning_rate": 0.00019999068739288167,
-      "loss": 46.0,
-      "step": 26986
-    },
-    {
-      "epoch": 4.346068682314103,
-      "grad_norm": 0.007169815246015787,
-      "learning_rate": 0.0001999906867024431,
-      "loss": 46.0,
-      "step": 26987
-    },
-    {
-      "epoch": 4.34622971939289,
-      "grad_norm": 0.0040504042990505695,
-      "learning_rate": 0.00019999068601197898,
-      "loss": 46.0,
-      "step": 26988
-    },
-    {
-      "epoch": 4.346390756471678,
-      "grad_norm": 0.004480308853089809,
-      "learning_rate": 0.00019999068532148924,
-      "loss": 46.0,
-      "step": 26989
-    },
-    {
-      "epoch": 4.346551793550465,
-      "grad_norm": 0.009324566461145878,
-      "learning_rate": 0.00019999068463097392,
-      "loss": 46.0,
-      "step": 26990
-    },
-    {
-      "epoch": 4.346712830629253,
-      "grad_norm": 0.000650244124699384,
-      "learning_rate": 0.00019999068394043298,
-      "loss": 46.0,
-      "step": 26991
-    },
-    {
-      "epoch": 4.34687386770804,
-      "grad_norm": 0.0036535440012812614,
-      "learning_rate": 0.0001999906832498665,
-      "loss": 46.0,
-      "step": 26992
-    },
-    {
-      "epoch": 4.3470349047868275,
-      "grad_norm": 0.007548793684691191,
-      "learning_rate": 0.0001999906825592744,
-      "loss": 46.0,
-      "step": 26993
-    },
-    {
-      "epoch": 4.347195941865614,
-      "grad_norm": 0.0009921301389113069,
-      "learning_rate": 0.00019999068186865669,
-      "loss": 46.0,
-      "step": 26994
-    },
-    {
-      "epoch": 4.3473569789444015,
-      "grad_norm": 0.007178795523941517,
-      "learning_rate": 0.0001999906811780134,
-      "loss": 46.0,
-      "step": 26995
-    },
-    {
-      "epoch": 4.347518016023189,
-      "grad_norm": 0.030864425003528595,
-      "learning_rate": 0.0001999906804873445,
-      "loss": 46.0,
-      "step": 26996
-    },
-    {
-      "epoch": 4.347679053101976,
-      "grad_norm": 0.00576575193554163,
-      "learning_rate": 0.00019999067979665003,
-      "loss": 46.0,
-      "step": 26997
-    },
-    {
-      "epoch": 4.347840090180764,
-      "grad_norm": 0.0037891629617661238,
-      "learning_rate": 0.00019999067910592998,
-      "loss": 46.0,
-      "step": 26998
-    },
-    {
-      "epoch": 4.348001127259551,
-      "grad_norm": 0.0030787428840994835,
-      "learning_rate": 0.00019999067841518431,
-      "loss": 46.0,
-      "step": 26999
-    },
-    {
-      "epoch": 4.348162164338339,
-      "grad_norm": 0.002335413359105587,
-      "learning_rate": 0.00019999067772441306,
-      "loss": 46.0,
-      "step": 27000
-    },
-    {
-      "epoch": 4.348323201417126,
-      "grad_norm": 0.0076245530508458614,
-      "learning_rate": 0.00019999067703361622,
-      "loss": 46.0,
-      "step": 27001
-    },
-    {
-      "epoch": 4.348484238495914,
-      "grad_norm": 0.007962411269545555,
-      "learning_rate": 0.00019999067634279377,
-      "loss": 46.0,
-      "step": 27002
-    },
-    {
-      "epoch": 4.348645275574701,
-      "grad_norm": 0.02506794035434723,
-      "learning_rate": 0.00019999067565194576,
-      "loss": 46.0,
-      "step": 27003
-    },
-    {
-      "epoch": 4.348806312653489,
-      "grad_norm": 0.0025293612852692604,
-      "learning_rate": 0.00019999067496107213,
-      "loss": 46.0,
-      "step": 27004
-    },
-    {
-      "epoch": 4.348967349732276,
-      "grad_norm": 0.0010694734519347548,
-      "learning_rate": 0.00019999067427017291,
-      "loss": 46.0,
-      "step": 27005
-    },
-    {
-      "epoch": 4.3491283868110635,
-      "grad_norm": 0.002649099100381136,
-      "learning_rate": 0.0001999906735792481,
-      "loss": 46.0,
-      "step": 27006
-    },
-    {
-      "epoch": 4.349289423889851,
-      "grad_norm": 0.010377351194620132,
-      "learning_rate": 0.0001999906728882977,
-      "loss": 46.0,
-      "step": 27007
-    },
-    {
-      "epoch": 4.349450460968638,
-      "grad_norm": 0.0047653838992118835,
-      "learning_rate": 0.0001999906721973217,
-      "loss": 46.0,
-      "step": 27008
-    },
-    {
-      "epoch": 4.349611498047425,
-      "grad_norm": 0.009352420456707478,
-      "learning_rate": 0.00019999067150632013,
-      "loss": 46.0,
-      "step": 27009
-    },
-    {
-      "epoch": 4.349772535126212,
-      "grad_norm": 0.006724266801029444,
-      "learning_rate": 0.00019999067081529295,
-      "loss": 46.0,
-      "step": 27010
-    },
-    {
-      "epoch": 4.349933572205,
-      "grad_norm": 0.00318281352519989,
-      "learning_rate": 0.00019999067012424018,
-      "loss": 46.0,
-      "step": 27011
-    },
-    {
-      "epoch": 4.350094609283787,
-      "grad_norm": 0.002402494428679347,
-      "learning_rate": 0.0001999906694331618,
-      "loss": 46.0,
-      "step": 27012
-    },
-    {
-      "epoch": 4.350255646362575,
-      "grad_norm": 0.0007097293855622411,
-      "learning_rate": 0.00019999066874205784,
-      "loss": 46.0,
-      "step": 27013
-    },
-    {
-      "epoch": 4.350416683441362,
-      "grad_norm": 0.008624077774584293,
-      "learning_rate": 0.00019999066805092828,
-      "loss": 46.0,
-      "step": 27014
-    },
-    {
-      "epoch": 4.35057772052015,
-      "grad_norm": 0.005293683148920536,
-      "learning_rate": 0.00019999066735977317,
-      "loss": 46.0,
-      "step": 27015
-    },
-    {
-      "epoch": 4.350738757598937,
-      "grad_norm": 0.0036781809758394957,
-      "learning_rate": 0.00019999066666859244,
-      "loss": 46.0,
-      "step": 27016
-    },
-    {
-      "epoch": 4.350899794677725,
-      "grad_norm": 0.0006104833446443081,
-      "learning_rate": 0.00019999066597738613,
-      "loss": 46.0,
-      "step": 27017
-    },
-    {
-      "epoch": 4.351060831756512,
-      "grad_norm": 0.008044463582336903,
-      "learning_rate": 0.0001999906652861542,
-      "loss": 46.0,
-      "step": 27018
-    },
-    {
-      "epoch": 4.3512218688352995,
-      "grad_norm": 0.0030127991922199726,
-      "learning_rate": 0.00019999066459489668,
-      "loss": 46.0,
-      "step": 27019
-    },
-    {
-      "epoch": 4.351382905914087,
-      "grad_norm": 0.002214134205132723,
-      "learning_rate": 0.00019999066390361358,
-      "loss": 46.0,
-      "step": 27020
-    },
-    {
-      "epoch": 4.351543942992874,
-      "grad_norm": 0.0007719038985669613,
-      "learning_rate": 0.00019999066321230489,
-      "loss": 46.0,
-      "step": 27021
-    },
-    {
-      "epoch": 4.351704980071662,
-      "grad_norm": 0.004441957920789719,
-      "learning_rate": 0.0001999906625209706,
-      "loss": 46.0,
-      "step": 27022
-    },
-    {
-      "epoch": 4.351866017150449,
-      "grad_norm": 0.00208304007537663,
-      "learning_rate": 0.00019999066182961072,
-      "loss": 46.0,
-      "step": 27023
-    },
-    {
-      "epoch": 4.352027054229236,
-      "grad_norm": 0.0017706903163343668,
-      "learning_rate": 0.00019999066113822524,
-      "loss": 46.0,
-      "step": 27024
-    },
-    {
-      "epoch": 4.352188091308023,
-      "grad_norm": 0.005134752951562405,
-      "learning_rate": 0.00019999066044681417,
-      "loss": 46.0,
-      "step": 27025
-    },
-    {
-      "epoch": 4.352349128386811,
-      "grad_norm": 0.0017050745664164424,
-      "learning_rate": 0.00019999065975537751,
-      "loss": 46.0,
-      "step": 27026
-    },
-    {
-      "epoch": 4.352510165465598,
-      "grad_norm": 0.0014627579366788268,
-      "learning_rate": 0.00019999065906391527,
-      "loss": 46.0,
-      "step": 27027
-    },
-    {
-      "epoch": 4.352671202544386,
-      "grad_norm": 0.0031340501736849546,
-      "learning_rate": 0.00019999065837242742,
-      "loss": 46.0,
-      "step": 27028
-    },
-    {
-      "epoch": 4.352832239623173,
-      "grad_norm": 0.01404839288443327,
-      "learning_rate": 0.00019999065768091398,
-      "loss": 46.0,
-      "step": 27029
-    },
-    {
-      "epoch": 4.352993276701961,
-      "grad_norm": 0.008472476154565811,
-      "learning_rate": 0.00019999065698937497,
-      "loss": 46.0,
-      "step": 27030
-    },
-    {
-      "epoch": 4.353154313780748,
-      "grad_norm": 0.002201506169512868,
-      "learning_rate": 0.00019999065629781036,
-      "loss": 46.0,
-      "step": 27031
-    },
-    {
-      "epoch": 4.3533153508595355,
-      "grad_norm": 0.003494028467684984,
-      "learning_rate": 0.00019999065560622015,
-      "loss": 46.0,
-      "step": 27032
-    },
-    {
-      "epoch": 4.353476387938323,
-      "grad_norm": 0.003972860053181648,
-      "learning_rate": 0.0001999906549146043,
-      "loss": 46.0,
-      "step": 27033
-    },
-    {
-      "epoch": 4.35363742501711,
-      "grad_norm": 0.0031147426925599575,
-      "learning_rate": 0.00019999065422296295,
-      "loss": 46.0,
-      "step": 27034
-    },
-    {
-      "epoch": 4.353798462095898,
-      "grad_norm": 0.0014362650690600276,
-      "learning_rate": 0.00019999065353129596,
-      "loss": 46.0,
-      "step": 27035
-    },
-    {
-      "epoch": 4.353959499174685,
-      "grad_norm": 0.00355161027982831,
-      "learning_rate": 0.00019999065283960336,
-      "loss": 46.0,
-      "step": 27036
-    },
-    {
-      "epoch": 4.354120536253473,
-      "grad_norm": 0.006990562193095684,
-      "learning_rate": 0.0001999906521478852,
-      "loss": 46.0,
-      "step": 27037
-    },
-    {
-      "epoch": 4.35428157333226,
-      "grad_norm": 0.006435507442802191,
-      "learning_rate": 0.00019999065145614143,
-      "loss": 46.0,
-      "step": 27038
-    },
-    {
-      "epoch": 4.354442610411047,
-      "grad_norm": 0.002615630626678467,
-      "learning_rate": 0.00019999065076437207,
-      "loss": 46.0,
-      "step": 27039
-    },
-    {
-      "epoch": 4.354603647489834,
-      "grad_norm": 0.0010898062027990818,
-      "learning_rate": 0.0001999906500725771,
-      "loss": 46.0,
-      "step": 27040
-    },
-    {
-      "epoch": 4.354764684568622,
-      "grad_norm": 0.0026471030432730913,
-      "learning_rate": 0.00019999064938075657,
-      "loss": 46.0,
-      "step": 27041
-    },
-    {
-      "epoch": 4.354925721647409,
-      "grad_norm": 0.006431697867810726,
-      "learning_rate": 0.0001999906486889104,
-      "loss": 46.0,
-      "step": 27042
-    },
-    {
-      "epoch": 4.3550867587261965,
-      "grad_norm": 0.0005420433008112013,
-      "learning_rate": 0.0001999906479970387,
-      "loss": 46.0,
-      "step": 27043
-    },
-    {
-      "epoch": 4.355247795804984,
-      "grad_norm": 0.013869138434529305,
-      "learning_rate": 0.0001999906473051414,
-      "loss": 46.0,
-      "step": 27044
-    },
-    {
-      "epoch": 4.3554088328837715,
-      "grad_norm": 0.014153920114040375,
-      "learning_rate": 0.00019999064661321847,
-      "loss": 46.0,
-      "step": 27045
-    },
-    {
-      "epoch": 4.355569869962559,
-      "grad_norm": 0.0006360609550029039,
-      "learning_rate": 0.00019999064592126994,
-      "loss": 46.0,
-      "step": 27046
-    },
-    {
-      "epoch": 4.355730907041346,
-      "grad_norm": 0.0047181532718241215,
-      "learning_rate": 0.00019999064522929584,
-      "loss": 46.0,
-      "step": 27047
-    },
-    {
-      "epoch": 4.355891944120134,
-      "grad_norm": 0.00938387680798769,
-      "learning_rate": 0.00019999064453729614,
-      "loss": 46.0,
-      "step": 27048
-    },
-    {
-      "epoch": 4.356052981198921,
-      "grad_norm": 0.0032553032506257296,
-      "learning_rate": 0.00019999064384527087,
-      "loss": 46.0,
-      "step": 27049
-    },
-    {
-      "epoch": 4.356214018277709,
-      "grad_norm": 0.003998341970145702,
-      "learning_rate": 0.00019999064315322001,
-      "loss": 46.0,
-      "step": 27050
-    },
-    {
-      "epoch": 4.356375055356496,
-      "grad_norm": 0.0007415534928441048,
-      "learning_rate": 0.00019999064246114352,
-      "loss": 46.0,
-      "step": 27051
-    },
-    {
-      "epoch": 4.356536092435284,
-      "grad_norm": 0.009251092560589314,
-      "learning_rate": 0.00019999064176904146,
-      "loss": 46.0,
-      "step": 27052
-    },
-    {
-      "epoch": 4.356697129514071,
-      "grad_norm": 0.0022210204042494297,
-      "learning_rate": 0.00019999064107691382,
-      "loss": 46.0,
-      "step": 27053
-    },
-    {
-      "epoch": 4.356858166592858,
-      "grad_norm": 0.0012636297615244985,
-      "learning_rate": 0.00019999064038476056,
-      "loss": 46.0,
-      "step": 27054
-    },
-    {
-      "epoch": 4.357019203671645,
-      "grad_norm": 0.0026149656623601913,
-      "learning_rate": 0.00019999063969258174,
-      "loss": 46.0,
-      "step": 27055
-    },
-    {
-      "epoch": 4.3571802407504325,
-      "grad_norm": 0.002562863053753972,
-      "learning_rate": 0.0001999906390003773,
-      "loss": 46.0,
-      "step": 27056
-    },
-    {
-      "epoch": 4.35734127782922,
-      "grad_norm": 0.005267334636300802,
-      "learning_rate": 0.00019999063830814727,
-      "loss": 46.0,
-      "step": 27057
-    },
-    {
-      "epoch": 4.357502314908007,
-      "grad_norm": 0.0026908330619335175,
-      "learning_rate": 0.00019999063761589166,
-      "loss": 46.0,
-      "step": 27058
-    },
-    {
-      "epoch": 4.357663351986795,
-      "grad_norm": 0.002155652502551675,
-      "learning_rate": 0.00019999063692361047,
-      "loss": 46.0,
-      "step": 27059
-    },
-    {
-      "epoch": 4.357824389065582,
-      "grad_norm": 0.003818254452198744,
-      "learning_rate": 0.00019999063623130363,
-      "loss": 46.0,
-      "step": 27060
-    },
-    {
-      "epoch": 4.35798542614437,
-      "grad_norm": 0.0036588225048035383,
-      "learning_rate": 0.00019999063553897126,
-      "loss": 46.0,
-      "step": 27061
-    },
-    {
-      "epoch": 4.358146463223157,
-      "grad_norm": 0.0072400267235934734,
-      "learning_rate": 0.00019999063484661325,
-      "loss": 46.0,
-      "step": 27062
-    },
-    {
-      "epoch": 4.358307500301945,
-      "grad_norm": 0.0027110599912703037,
-      "learning_rate": 0.00019999063415422969,
-      "loss": 46.0,
-      "step": 27063
-    },
-    {
-      "epoch": 4.358468537380732,
-      "grad_norm": 0.014049648307263851,
-      "learning_rate": 0.00019999063346182053,
-      "loss": 46.0,
-      "step": 27064
-    },
-    {
-      "epoch": 4.35862957445952,
-      "grad_norm": 0.0014710122486576438,
-      "learning_rate": 0.00019999063276938576,
-      "loss": 46.0,
-      "step": 27065
-    },
-    {
-      "epoch": 4.358790611538307,
-      "grad_norm": 0.001651704660616815,
-      "learning_rate": 0.0001999906320769254,
-      "loss": 46.0,
-      "step": 27066
-    },
-    {
-      "epoch": 4.3589516486170945,
-      "grad_norm": 0.008998613804578781,
-      "learning_rate": 0.00019999063138443943,
-      "loss": 46.0,
-      "step": 27067
-    },
-    {
-      "epoch": 4.359112685695881,
-      "grad_norm": 0.003404980758205056,
-      "learning_rate": 0.00019999063069192792,
-      "loss": 46.0,
-      "step": 27068
-    },
-    {
-      "epoch": 4.3592737227746685,
-      "grad_norm": 0.009843757376074791,
-      "learning_rate": 0.00019999062999939078,
-      "loss": 46.0,
-      "step": 27069
-    },
-    {
-      "epoch": 4.359434759853456,
-      "grad_norm": 0.00931534357368946,
-      "learning_rate": 0.00019999062930682807,
-      "loss": 46.0,
-      "step": 27070
-    },
-    {
-      "epoch": 4.359595796932243,
-      "grad_norm": 0.0031456982251256704,
-      "learning_rate": 0.00019999062861423972,
-      "loss": 46.0,
-      "step": 27071
-    },
-    {
-      "epoch": 4.359756834011031,
-      "grad_norm": 0.008781283162534237,
-      "learning_rate": 0.00019999062792162582,
-      "loss": 46.0,
-      "step": 27072
-    },
-    {
-      "epoch": 4.359917871089818,
-      "grad_norm": 0.0014256745344027877,
-      "learning_rate": 0.00019999062722898632,
-      "loss": 46.0,
-      "step": 27073
-    },
-    {
-      "epoch": 4.360078908168606,
-      "grad_norm": 0.002624319400638342,
-      "learning_rate": 0.0001999906265363212,
-      "loss": 46.0,
-      "step": 27074
-    },
-    {
-      "epoch": 4.360239945247393,
-      "grad_norm": 0.0021495830733329058,
-      "learning_rate": 0.00019999062584363054,
-      "loss": 46.0,
-      "step": 27075
-    },
-    {
-      "epoch": 4.360400982326181,
-      "grad_norm": 0.002483322052285075,
-      "learning_rate": 0.00019999062515091426,
-      "loss": 46.0,
-      "step": 27076
-    },
-    {
-      "epoch": 4.360562019404968,
-      "grad_norm": 0.00913215521723032,
-      "learning_rate": 0.00019999062445817236,
-      "loss": 46.0,
-      "step": 27077
-    },
-    {
-      "epoch": 4.360723056483756,
-      "grad_norm": 0.00351537112146616,
-      "learning_rate": 0.00019999062376540493,
-      "loss": 46.0,
-      "step": 27078
-    },
-    {
-      "epoch": 4.360884093562543,
-      "grad_norm": 0.0006506695644930005,
-      "learning_rate": 0.00019999062307261186,
-      "loss": 46.0,
-      "step": 27079
-    },
-    {
-      "epoch": 4.3610451306413305,
-      "grad_norm": 0.009654135443270206,
-      "learning_rate": 0.0001999906223797932,
-      "loss": 46.0,
-      "step": 27080
-    },
-    {
-      "epoch": 4.361206167720118,
-      "grad_norm": 0.0036119548603892326,
-      "learning_rate": 0.00019999062168694895,
-      "loss": 46.0,
-      "step": 27081
-    },
-    {
-      "epoch": 4.3613672047989045,
-      "grad_norm": 0.010684236884117126,
-      "learning_rate": 0.00019999062099407912,
-      "loss": 46.0,
-      "step": 27082
-    },
-    {
-      "epoch": 4.361528241877692,
-      "grad_norm": 0.0027164663188159466,
-      "learning_rate": 0.0001999906203011837,
-      "loss": 46.0,
-      "step": 27083
-    },
-    {
-      "epoch": 4.361689278956479,
-      "grad_norm": 0.0009681042865850031,
-      "learning_rate": 0.00019999061960826266,
-      "loss": 46.0,
-      "step": 27084
-    },
-    {
-      "epoch": 4.361850316035267,
-      "grad_norm": 0.004821220878511667,
-      "learning_rate": 0.00019999061891531604,
-      "loss": 46.0,
-      "step": 27085
-    },
-    {
-      "epoch": 4.362011353114054,
-      "grad_norm": 0.002124957274645567,
-      "learning_rate": 0.00019999061822234383,
-      "loss": 46.0,
-      "step": 27086
-    },
-    {
-      "epoch": 4.362172390192842,
-      "grad_norm": 0.005472272168844938,
-      "learning_rate": 0.00019999061752934606,
-      "loss": 46.0,
-      "step": 27087
-    },
-    {
-      "epoch": 4.362333427271629,
-      "grad_norm": 0.0031505899969488382,
-      "learning_rate": 0.00019999061683632268,
-      "loss": 46.0,
-      "step": 27088
-    },
-    {
-      "epoch": 4.362494464350417,
-      "grad_norm": 0.0014355042949318886,
-      "learning_rate": 0.00019999061614327368,
-      "loss": 46.0,
-      "step": 27089
-    },
-    {
-      "epoch": 4.362655501429204,
-      "grad_norm": 0.005559311248362064,
-      "learning_rate": 0.0001999906154501991,
-      "loss": 46.0,
-      "step": 27090
-    },
-    {
-      "epoch": 4.362816538507992,
-      "grad_norm": 0.004858406260609627,
-      "learning_rate": 0.00019999061475709895,
-      "loss": 46.0,
-      "step": 27091
-    },
-    {
-      "epoch": 4.362977575586779,
-      "grad_norm": 0.002242660615593195,
-      "learning_rate": 0.00019999061406397316,
-      "loss": 46.0,
-      "step": 27092
-    },
-    {
-      "epoch": 4.3631386126655665,
-      "grad_norm": 0.00249241734854877,
-      "learning_rate": 0.0001999906133708218,
-      "loss": 46.0,
-      "step": 27093
-    },
-    {
-      "epoch": 4.363299649744354,
-      "grad_norm": 0.008938068524003029,
-      "learning_rate": 0.00019999061267764488,
-      "loss": 46.0,
-      "step": 27094
-    },
-    {
-      "epoch": 4.363460686823141,
-      "grad_norm": 0.0026307005900889635,
-      "learning_rate": 0.00019999061198444233,
-      "loss": 46.0,
-      "step": 27095
-    },
-    {
-      "epoch": 4.363621723901929,
-      "grad_norm": 0.004502871073782444,
-      "learning_rate": 0.00019999061129121422,
-      "loss": 46.0,
-      "step": 27096
-    },
-    {
-      "epoch": 4.363782760980715,
-      "grad_norm": 0.013487651012837887,
-      "learning_rate": 0.0001999906105979605,
-      "loss": 46.0,
-      "step": 27097
-    },
-    {
-      "epoch": 4.363943798059503,
-      "grad_norm": 0.006255973130464554,
-      "learning_rate": 0.00019999060990468116,
-      "loss": 46.0,
-      "step": 27098
-    },
-    {
-      "epoch": 4.36410483513829,
-      "grad_norm": 0.004796335007995367,
-      "learning_rate": 0.00019999060921137627,
-      "loss": 46.0,
-      "step": 27099
-    },
-    {
-      "epoch": 4.364265872217078,
-      "grad_norm": 0.004446865059435368,
-      "learning_rate": 0.00019999060851804575,
-      "loss": 46.0,
-      "step": 27100
-    },
-    {
-      "epoch": 4.364426909295865,
-      "grad_norm": 0.007209441624581814,
-      "learning_rate": 0.00019999060782468968,
-      "loss": 46.0,
-      "step": 27101
-    },
-    {
-      "epoch": 4.364587946374653,
-      "grad_norm": 0.000757641508243978,
-      "learning_rate": 0.00019999060713130797,
-      "loss": 46.0,
-      "step": 27102
-    },
-    {
-      "epoch": 4.36474898345344,
-      "grad_norm": 0.0013446417870000005,
-      "learning_rate": 0.0001999906064379007,
-      "loss": 46.0,
-      "step": 27103
-    },
-    {
-      "epoch": 4.3649100205322275,
-      "grad_norm": 0.019902419298887253,
-      "learning_rate": 0.0001999906057444678,
-      "loss": 46.0,
-      "step": 27104
-    },
-    {
-      "epoch": 4.365071057611015,
-      "grad_norm": 0.0032546038273721933,
-      "learning_rate": 0.00019999060505100936,
-      "loss": 46.0,
-      "step": 27105
-    },
-    {
-      "epoch": 4.3652320946898024,
-      "grad_norm": 0.0037054584827274084,
-      "learning_rate": 0.0001999906043575253,
-      "loss": 46.0,
-      "step": 27106
-    },
-    {
-      "epoch": 4.36539313176859,
-      "grad_norm": 0.0005276603042148054,
-      "learning_rate": 0.00019999060366401565,
-      "loss": 46.0,
-      "step": 27107
-    },
-    {
-      "epoch": 4.365554168847377,
-      "grad_norm": 0.0007191240438260138,
-      "learning_rate": 0.00019999060297048042,
-      "loss": 46.0,
-      "step": 27108
-    },
-    {
-      "epoch": 4.365715205926165,
-      "grad_norm": 0.005270296707749367,
-      "learning_rate": 0.0001999906022769196,
-      "loss": 46.0,
-      "step": 27109
-    },
-    {
-      "epoch": 4.365876243004952,
-      "grad_norm": 0.008191713131964207,
-      "learning_rate": 0.00019999060158333318,
-      "loss": 46.0,
-      "step": 27110
-    },
-    {
-      "epoch": 4.36603728008374,
-      "grad_norm": 0.0016720653511583805,
-      "learning_rate": 0.00019999060088972116,
-      "loss": 46.0,
-      "step": 27111
-    },
-    {
-      "epoch": 4.366198317162526,
-      "grad_norm": 0.0019134688191115856,
-      "learning_rate": 0.00019999060019608352,
-      "loss": 46.0,
-      "step": 27112
-    },
-    {
-      "epoch": 4.366359354241314,
-      "grad_norm": 0.007254447788000107,
-      "learning_rate": 0.00019999059950242035,
-      "loss": 46.0,
-      "step": 27113
-    },
-    {
-      "epoch": 4.366520391320101,
-      "grad_norm": 0.0008195106638595462,
-      "learning_rate": 0.00019999059880873154,
-      "loss": 46.0,
-      "step": 27114
-    },
-    {
-      "epoch": 4.366681428398889,
-      "grad_norm": 0.006284576375037432,
-      "learning_rate": 0.00019999059811501714,
-      "loss": 46.0,
-      "step": 27115
-    },
-    {
-      "epoch": 4.366842465477676,
-      "grad_norm": 0.008425384759902954,
-      "learning_rate": 0.0001999905974212772,
-      "loss": 46.0,
-      "step": 27116
-    },
-    {
-      "epoch": 4.3670035025564635,
-      "grad_norm": 0.0022102666553109884,
-      "learning_rate": 0.0001999905967275116,
-      "loss": 46.0,
-      "step": 27117
-    },
-    {
-      "epoch": 4.367164539635251,
-      "grad_norm": 0.004138268996030092,
-      "learning_rate": 0.00019999059603372047,
-      "loss": 46.0,
-      "step": 27118
-    },
-    {
-      "epoch": 4.367325576714038,
-      "grad_norm": 0.00965751800686121,
-      "learning_rate": 0.00019999059533990366,
-      "loss": 46.0,
-      "step": 27119
-    },
-    {
-      "epoch": 4.367486613792826,
-      "grad_norm": 0.004430947359651327,
-      "learning_rate": 0.00019999059464606133,
-      "loss": 46.0,
-      "step": 27120
-    },
-    {
-      "epoch": 4.367647650871613,
-      "grad_norm": 0.0010899221524596214,
-      "learning_rate": 0.0001999905939521934,
-      "loss": 46.0,
-      "step": 27121
-    },
-    {
-      "epoch": 4.367808687950401,
-      "grad_norm": 0.006405069027096033,
-      "learning_rate": 0.00019999059325829984,
-      "loss": 46.0,
-      "step": 27122
-    },
-    {
-      "epoch": 4.367969725029188,
-      "grad_norm": 0.013673920184373856,
-      "learning_rate": 0.00019999059256438072,
-      "loss": 46.0,
-      "step": 27123
-    },
-    {
-      "epoch": 4.368130762107976,
-      "grad_norm": 0.006496686954051256,
-      "learning_rate": 0.00019999059187043603,
-      "loss": 46.0,
-      "step": 27124
-    },
-    {
-      "epoch": 4.368291799186763,
-      "grad_norm": 0.007036920636892319,
-      "learning_rate": 0.0001999905911764657,
-      "loss": 46.0,
-      "step": 27125
-    },
-    {
-      "epoch": 4.368452836265551,
-      "grad_norm": 0.0066603501327335835,
-      "learning_rate": 0.0001999905904824698,
-      "loss": 46.0,
-      "step": 27126
-    },
-    {
-      "epoch": 4.368613873344337,
-      "grad_norm": 0.0025863044429570436,
-      "learning_rate": 0.0001999905897884483,
-      "loss": 46.0,
-      "step": 27127
-    },
-    {
-      "epoch": 4.368774910423125,
-      "grad_norm": 0.0008355535101145506,
-      "learning_rate": 0.0001999905890944012,
-      "loss": 46.0,
-      "step": 27128
-    },
-    {
-      "epoch": 4.368935947501912,
-      "grad_norm": 0.0017207521013915539,
-      "learning_rate": 0.00019999058840032853,
-      "loss": 46.0,
-      "step": 27129
-    },
-    {
-      "epoch": 4.3690969845806995,
-      "grad_norm": 0.0031994956079870462,
-      "learning_rate": 0.00019999058770623024,
-      "loss": 46.0,
-      "step": 27130
-    },
-    {
-      "epoch": 4.369258021659487,
-      "grad_norm": 0.004539301618933678,
-      "learning_rate": 0.0001999905870121064,
-      "loss": 46.0,
-      "step": 27131
-    },
-    {
-      "epoch": 4.369419058738274,
-      "grad_norm": 0.018510285764932632,
-      "learning_rate": 0.00019999058631795693,
-      "loss": 46.0,
-      "step": 27132
-    },
-    {
-      "epoch": 4.369580095817062,
-      "grad_norm": 0.0016801398014649749,
-      "learning_rate": 0.00019999058562378188,
-      "loss": 46.0,
-      "step": 27133
-    },
-    {
-      "epoch": 4.369741132895849,
-      "grad_norm": 0.0031483913771808147,
-      "learning_rate": 0.00019999058492958124,
-      "loss": 46.0,
-      "step": 27134
-    },
-    {
-      "epoch": 4.369902169974637,
-      "grad_norm": 0.01651756837964058,
-      "learning_rate": 0.000199990584235355,
-      "loss": 46.0,
-      "step": 27135
-    },
-    {
-      "epoch": 4.370063207053424,
-      "grad_norm": 0.0009231962030753493,
-      "learning_rate": 0.00019999058354110315,
-      "loss": 46.0,
-      "step": 27136
-    },
-    {
-      "epoch": 4.370224244132212,
-      "grad_norm": 0.0010743794264271855,
-      "learning_rate": 0.00019999058284682575,
-      "loss": 46.0,
-      "step": 27137
-    },
-    {
-      "epoch": 4.370385281210999,
-      "grad_norm": 0.0020308492239564657,
-      "learning_rate": 0.00019999058215252274,
-      "loss": 46.0,
-      "step": 27138
-    },
-    {
-      "epoch": 4.370546318289787,
-      "grad_norm": 0.0009223786764778197,
-      "learning_rate": 0.0001999905814581941,
-      "loss": 46.0,
-      "step": 27139
-    },
-    {
-      "epoch": 4.370707355368574,
-      "grad_norm": 0.003088445169851184,
-      "learning_rate": 0.00019999058076383995,
-      "loss": 46.0,
-      "step": 27140
-    },
-    {
-      "epoch": 4.3708683924473615,
-      "grad_norm": 0.0049479054287076,
-      "learning_rate": 0.00019999058006946015,
-      "loss": 46.0,
-      "step": 27141
-    },
-    {
-      "epoch": 4.371029429526148,
-      "grad_norm": 0.017845382913947105,
-      "learning_rate": 0.00019999057937505476,
-      "loss": 46.0,
-      "step": 27142
-    },
-    {
-      "epoch": 4.3711904666049355,
-      "grad_norm": 0.009426828473806381,
-      "learning_rate": 0.00019999057868062375,
-      "loss": 46.0,
-      "step": 27143
-    },
-    {
-      "epoch": 4.371351503683723,
-      "grad_norm": 0.004196139983832836,
-      "learning_rate": 0.00019999057798616722,
-      "loss": 46.0,
-      "step": 27144
-    },
-    {
-      "epoch": 4.37151254076251,
-      "grad_norm": 0.0027773615438491106,
-      "learning_rate": 0.00019999057729168504,
-      "loss": 46.0,
-      "step": 27145
-    },
-    {
-      "epoch": 4.371673577841298,
-      "grad_norm": 0.005441885907202959,
-      "learning_rate": 0.00019999057659717727,
-      "loss": 46.0,
-      "step": 27146
-    },
-    {
-      "epoch": 4.371834614920085,
-      "grad_norm": 0.0028727292083203793,
-      "learning_rate": 0.00019999057590264397,
-      "loss": 46.0,
-      "step": 27147
-    },
-    {
-      "epoch": 4.371995651998873,
-      "grad_norm": 0.001167637063190341,
-      "learning_rate": 0.000199990575208085,
-      "loss": 46.0,
-      "step": 27148
-    },
-    {
-      "epoch": 4.37215668907766,
-      "grad_norm": 0.005727018229663372,
-      "learning_rate": 0.00019999057451350048,
-      "loss": 46.0,
-      "step": 27149
-    },
-    {
-      "epoch": 4.372317726156448,
-      "grad_norm": 0.006780306342989206,
-      "learning_rate": 0.00019999057381889034,
-      "loss": 46.0,
-      "step": 27150
-    },
-    {
-      "epoch": 4.372478763235235,
-      "grad_norm": 0.0014827157137915492,
-      "learning_rate": 0.00019999057312425464,
-      "loss": 46.0,
-      "step": 27151
-    },
-    {
-      "epoch": 4.3726398003140226,
-      "grad_norm": 0.004794629756361246,
-      "learning_rate": 0.00019999057242959332,
-      "loss": 46.0,
-      "step": 27152
-    },
-    {
-      "epoch": 4.37280083739281,
-      "grad_norm": 0.003373529762029648,
-      "learning_rate": 0.00019999057173490642,
-      "loss": 46.0,
-      "step": 27153
-    },
-    {
-      "epoch": 4.3729618744715975,
-      "grad_norm": 0.0029637019615620375,
-      "learning_rate": 0.00019999057104019393,
-      "loss": 46.0,
-      "step": 27154
-    },
-    {
-      "epoch": 4.373122911550384,
-      "grad_norm": 0.0037493992131203413,
-      "learning_rate": 0.00019999057034545586,
-      "loss": 46.0,
-      "step": 27155
-    },
-    {
-      "epoch": 4.3732839486291715,
-      "grad_norm": 0.006962976418435574,
-      "learning_rate": 0.00019999056965069216,
-      "loss": 46.0,
-      "step": 27156
-    },
-    {
-      "epoch": 4.373444985707959,
-      "grad_norm": 0.0015373227652162313,
-      "learning_rate": 0.0001999905689559029,
-      "loss": 46.0,
-      "step": 27157
-    },
-    {
-      "epoch": 4.373606022786746,
-      "grad_norm": 0.0017439405201002955,
-      "learning_rate": 0.00019999056826108805,
-      "loss": 46.0,
-      "step": 27158
-    },
-    {
-      "epoch": 4.373767059865534,
-      "grad_norm": 0.005726260133087635,
-      "learning_rate": 0.0001999905675662476,
-      "loss": 46.0,
-      "step": 27159
-    },
-    {
-      "epoch": 4.373928096944321,
-      "grad_norm": 0.0012014490785077214,
-      "learning_rate": 0.00019999056687138153,
-      "loss": 46.0,
-      "step": 27160
-    },
-    {
-      "epoch": 4.374089134023109,
-      "grad_norm": 0.005440492648631334,
-      "learning_rate": 0.0001999905661764899,
-      "loss": 46.0,
-      "step": 27161
-    },
-    {
-      "epoch": 4.374250171101896,
-      "grad_norm": 0.01880985125899315,
-      "learning_rate": 0.00019999056548157266,
-      "loss": 46.0,
-      "step": 27162
-    },
-    {
-      "epoch": 4.374411208180684,
-      "grad_norm": 0.010553323663771152,
-      "learning_rate": 0.00019999056478662986,
-      "loss": 46.0,
-      "step": 27163
-    },
-    {
-      "epoch": 4.374572245259471,
-      "grad_norm": 0.0013744422467425466,
-      "learning_rate": 0.00019999056409166142,
-      "loss": 46.0,
-      "step": 27164
-    },
-    {
-      "epoch": 4.3747332823382585,
-      "grad_norm": 0.01420077495276928,
-      "learning_rate": 0.0001999905633966674,
-      "loss": 46.0,
-      "step": 27165
-    },
-    {
-      "epoch": 4.374894319417046,
-      "grad_norm": 0.002734562149271369,
-      "learning_rate": 0.0001999905627016478,
-      "loss": 46.0,
-      "step": 27166
-    },
-    {
-      "epoch": 4.375055356495833,
-      "grad_norm": 0.013370763510465622,
-      "learning_rate": 0.00019999056200660262,
-      "loss": 46.0,
-      "step": 27167
-    },
-    {
-      "epoch": 4.375216393574621,
-      "grad_norm": 0.0037429702933877707,
-      "learning_rate": 0.00019999056131153183,
-      "loss": 46.0,
-      "step": 27168
-    },
-    {
-      "epoch": 4.375377430653408,
-      "grad_norm": 0.017090575769543648,
-      "learning_rate": 0.00019999056061643545,
-      "loss": 46.0,
-      "step": 27169
-    },
-    {
-      "epoch": 4.375538467732195,
-      "grad_norm": 0.007569979876279831,
-      "learning_rate": 0.00019999055992131348,
-      "loss": 46.0,
-      "step": 27170
-    },
-    {
-      "epoch": 4.375699504810982,
-      "grad_norm": 0.02025243639945984,
-      "learning_rate": 0.0001999905592261659,
-      "loss": 46.0,
-      "step": 27171
-    },
-    {
-      "epoch": 4.37586054188977,
-      "grad_norm": 0.0019129186403006315,
-      "learning_rate": 0.00019999055853099274,
-      "loss": 46.0,
-      "step": 27172
-    },
-    {
-      "epoch": 4.376021578968557,
-      "grad_norm": 0.011236516758799553,
-      "learning_rate": 0.00019999055783579398,
-      "loss": 46.0,
-      "step": 27173
-    },
-    {
-      "epoch": 4.376182616047345,
-      "grad_norm": 0.004096119664609432,
-      "learning_rate": 0.00019999055714056967,
-      "loss": 46.0,
-      "step": 27174
-    },
-    {
-      "epoch": 4.376343653126132,
-      "grad_norm": 0.011911129578948021,
-      "learning_rate": 0.00019999055644531971,
-      "loss": 46.0,
-      "step": 27175
-    },
-    {
-      "epoch": 4.37650469020492,
-      "grad_norm": 0.0014494559727609158,
-      "learning_rate": 0.0001999905557500442,
-      "loss": 46.0,
-      "step": 27176
-    },
-    {
-      "epoch": 4.376665727283707,
-      "grad_norm": 0.01516756508499384,
-      "learning_rate": 0.0001999905550547431,
-      "loss": 46.0,
-      "step": 27177
-    },
-    {
-      "epoch": 4.3768267643624945,
-      "grad_norm": 0.007629004307091236,
-      "learning_rate": 0.00019999055435941635,
-      "loss": 46.0,
-      "step": 27178
-    },
-    {
-      "epoch": 4.376987801441282,
-      "grad_norm": 0.004359136335551739,
-      "learning_rate": 0.00019999055366406405,
-      "loss": 46.0,
-      "step": 27179
-    },
-    {
-      "epoch": 4.377148838520069,
-      "grad_norm": 0.0033742839004844427,
-      "learning_rate": 0.00019999055296868615,
-      "loss": 46.0,
-      "step": 27180
-    },
-    {
-      "epoch": 4.377309875598857,
-      "grad_norm": 0.007984524592757225,
-      "learning_rate": 0.00019999055227328268,
-      "loss": 46.0,
-      "step": 27181
-    },
-    {
-      "epoch": 4.377470912677644,
-      "grad_norm": 0.0045737153850495815,
-      "learning_rate": 0.0001999905515778536,
-      "loss": 46.0,
-      "step": 27182
-    },
-    {
-      "epoch": 4.377631949756432,
-      "grad_norm": 0.015872295945882797,
-      "learning_rate": 0.00019999055088239893,
-      "loss": 46.0,
-      "step": 27183
-    },
-    {
-      "epoch": 4.377792986835219,
-      "grad_norm": 0.0030408913735300303,
-      "learning_rate": 0.00019999055018691864,
-      "loss": 46.0,
-      "step": 27184
-    },
-    {
-      "epoch": 4.377954023914006,
-      "grad_norm": 0.005618977826088667,
-      "learning_rate": 0.0001999905494914128,
-      "loss": 46.0,
-      "step": 27185
-    },
-    {
-      "epoch": 4.378115060992793,
-      "grad_norm": 0.0023326631635427475,
-      "learning_rate": 0.00019999054879588134,
-      "loss": 46.0,
-      "step": 27186
-    },
-    {
-      "epoch": 4.378276098071581,
-      "grad_norm": 0.001446245238184929,
-      "learning_rate": 0.00019999054810032428,
-      "loss": 46.0,
-      "step": 27187
-    },
-    {
-      "epoch": 4.378437135150368,
-      "grad_norm": 0.023635298013687134,
-      "learning_rate": 0.00019999054740474167,
-      "loss": 46.0,
-      "step": 27188
-    },
-    {
-      "epoch": 4.378598172229156,
-      "grad_norm": 0.001972927013412118,
-      "learning_rate": 0.0001999905467091334,
-      "loss": 46.0,
-      "step": 27189
-    },
-    {
-      "epoch": 4.378759209307943,
-      "grad_norm": 0.0035482419189065695,
-      "learning_rate": 0.0001999905460134996,
-      "loss": 46.0,
-      "step": 27190
-    },
-    {
-      "epoch": 4.3789202463867305,
-      "grad_norm": 0.003834902308881283,
-      "learning_rate": 0.00019999054531784021,
-      "loss": 46.0,
-      "step": 27191
-    },
-    {
-      "epoch": 4.379081283465518,
-      "grad_norm": 0.003968066535890102,
-      "learning_rate": 0.0001999905446221552,
-      "loss": 46.0,
-      "step": 27192
-    },
-    {
-      "epoch": 4.379242320544305,
-      "grad_norm": 0.0021446545142680407,
-      "learning_rate": 0.0001999905439264446,
-      "loss": 46.0,
-      "step": 27193
-    },
-    {
-      "epoch": 4.379403357623093,
-      "grad_norm": 0.002743483753874898,
-      "learning_rate": 0.0001999905432307084,
-      "loss": 46.0,
-      "step": 27194
-    },
-    {
-      "epoch": 4.37956439470188,
-      "grad_norm": 0.0024175182916224003,
-      "learning_rate": 0.0001999905425349466,
-      "loss": 46.0,
-      "step": 27195
-    },
-    {
-      "epoch": 4.379725431780668,
-      "grad_norm": 0.003960182424634695,
-      "learning_rate": 0.00019999054183915925,
-      "loss": 46.0,
-      "step": 27196
-    },
-    {
-      "epoch": 4.379886468859455,
-      "grad_norm": 0.005331596825271845,
-      "learning_rate": 0.00019999054114334627,
-      "loss": 46.0,
-      "step": 27197
-    },
-    {
-      "epoch": 4.380047505938243,
-      "grad_norm": 0.004465213045477867,
-      "learning_rate": 0.00019999054044750772,
-      "loss": 46.0,
-      "step": 27198
-    },
-    {
-      "epoch": 4.38020854301703,
-      "grad_norm": 0.005457528866827488,
-      "learning_rate": 0.00019999053975164354,
-      "loss": 46.0,
-      "step": 27199
-    },
-    {
-      "epoch": 4.380369580095817,
-      "grad_norm": 0.0036655315198004246,
-      "learning_rate": 0.0001999905390557538,
-      "loss": 46.0,
-      "step": 27200
-    },
-    {
-      "epoch": 4.380530617174604,
-      "grad_norm": 0.0014014887856319547,
-      "learning_rate": 0.00019999053835983847,
-      "loss": 46.0,
-      "step": 27201
-    },
-    {
-      "epoch": 4.380691654253392,
-      "grad_norm": 0.007068261504173279,
-      "learning_rate": 0.00019999053766389752,
-      "loss": 46.0,
-      "step": 27202
-    },
-    {
-      "epoch": 4.380852691332179,
-      "grad_norm": 0.001446820329874754,
-      "learning_rate": 0.00019999053696793102,
-      "loss": 46.0,
-      "step": 27203
-    },
-    {
-      "epoch": 4.3810137284109665,
-      "grad_norm": 0.005662822164595127,
-      "learning_rate": 0.0001999905362719389,
-      "loss": 46.0,
-      "step": 27204
-    },
-    {
-      "epoch": 4.381174765489754,
-      "grad_norm": 0.008554995059967041,
-      "learning_rate": 0.0001999905355759212,
-      "loss": 46.0,
-      "step": 27205
-    },
-    {
-      "epoch": 4.381335802568541,
-      "grad_norm": 0.008206910453736782,
-      "learning_rate": 0.0001999905348798779,
-      "loss": 46.0,
-      "step": 27206
-    },
-    {
-      "epoch": 4.381496839647329,
-      "grad_norm": 0.001227443921379745,
-      "learning_rate": 0.000199990534183809,
-      "loss": 46.0,
-      "step": 27207
-    },
-    {
-      "epoch": 4.381657876726116,
-      "grad_norm": 0.007651353720575571,
-      "learning_rate": 0.00019999053348771452,
-      "loss": 46.0,
-      "step": 27208
-    },
-    {
-      "epoch": 4.381818913804904,
-      "grad_norm": 0.0031927514355629683,
-      "learning_rate": 0.00019999053279159444,
-      "loss": 46.0,
-      "step": 27209
-    },
-    {
-      "epoch": 4.381979950883691,
-      "grad_norm": 0.005342401098459959,
-      "learning_rate": 0.00019999053209544874,
-      "loss": 46.0,
-      "step": 27210
-    },
-    {
-      "epoch": 4.382140987962479,
-      "grad_norm": 0.0032986472360789776,
-      "learning_rate": 0.00019999053139927751,
-      "loss": 46.0,
-      "step": 27211
-    },
-    {
-      "epoch": 4.382302025041266,
-      "grad_norm": 0.004678192548453808,
-      "learning_rate": 0.00019999053070308064,
-      "loss": 46.0,
-      "step": 27212
-    },
-    {
-      "epoch": 4.3824630621200535,
-      "grad_norm": 0.0019402520265430212,
-      "learning_rate": 0.0001999905300068582,
-      "loss": 46.0,
-      "step": 27213
-    },
-    {
-      "epoch": 4.382624099198841,
-      "grad_norm": 0.0028347631450742483,
-      "learning_rate": 0.00019999052931061014,
-      "loss": 46.0,
-      "step": 27214
-    },
-    {
-      "epoch": 4.382785136277628,
-      "grad_norm": 0.0008329771808348596,
-      "learning_rate": 0.0001999905286143365,
-      "loss": 46.0,
-      "step": 27215
-    },
-    {
-      "epoch": 4.382946173356415,
-      "grad_norm": 0.0035379512701183558,
-      "learning_rate": 0.0001999905279180373,
-      "loss": 46.0,
-      "step": 27216
-    },
-    {
-      "epoch": 4.3831072104352025,
-      "grad_norm": 0.0015824425499886274,
-      "learning_rate": 0.0001999905272217125,
-      "loss": 46.0,
-      "step": 27217
-    },
-    {
-      "epoch": 4.38326824751399,
-      "grad_norm": 0.00535028288140893,
-      "learning_rate": 0.0001999905265253621,
-      "loss": 46.0,
-      "step": 27218
-    },
-    {
-      "epoch": 4.383429284592777,
-      "grad_norm": 0.007845137268304825,
-      "learning_rate": 0.00019999052582898608,
-      "loss": 46.0,
-      "step": 27219
-    },
-    {
-      "epoch": 4.383590321671565,
-      "grad_norm": 0.006032073870301247,
-      "learning_rate": 0.00019999052513258446,
-      "loss": 46.0,
-      "step": 27220
-    },
-    {
-      "epoch": 4.383751358750352,
-      "grad_norm": 0.002569495001807809,
-      "learning_rate": 0.00019999052443615728,
-      "loss": 46.0,
-      "step": 27221
-    },
-    {
-      "epoch": 4.38391239582914,
-      "grad_norm": 0.0013232954079285264,
-      "learning_rate": 0.0001999905237397045,
-      "loss": 46.0,
-      "step": 27222
-    },
-    {
-      "epoch": 4.384073432907927,
-      "grad_norm": 0.001953263534232974,
-      "learning_rate": 0.00019999052304322615,
-      "loss": 46.0,
-      "step": 27223
-    },
-    {
-      "epoch": 4.384234469986715,
-      "grad_norm": 0.009116681292653084,
-      "learning_rate": 0.00019999052234672218,
-      "loss": 46.0,
-      "step": 27224
-    },
-    {
-      "epoch": 4.384395507065502,
-      "grad_norm": 0.00343763898126781,
-      "learning_rate": 0.00019999052165019262,
-      "loss": 46.0,
-      "step": 27225
-    },
-    {
-      "epoch": 4.3845565441442895,
-      "grad_norm": 0.016197996214032173,
-      "learning_rate": 0.00019999052095363745,
-      "loss": 46.0,
-      "step": 27226
-    },
-    {
-      "epoch": 4.384717581223077,
-      "grad_norm": 0.0060784099623560905,
-      "learning_rate": 0.00019999052025705672,
-      "loss": 46.0,
-      "step": 27227
-    },
-    {
-      "epoch": 4.3848786183018635,
-      "grad_norm": 0.011557098478078842,
-      "learning_rate": 0.0001999905195604504,
-      "loss": 46.0,
-      "step": 27228
-    },
-    {
-      "epoch": 4.385039655380651,
-      "grad_norm": 0.009646390564739704,
-      "learning_rate": 0.0001999905188638185,
-      "loss": 46.0,
-      "step": 27229
-    },
-    {
-      "epoch": 4.385200692459438,
-      "grad_norm": 0.012289199978113174,
-      "learning_rate": 0.00019999051816716094,
-      "loss": 46.0,
-      "step": 27230
-    },
-    {
-      "epoch": 4.385361729538226,
-      "grad_norm": 0.002778079593554139,
-      "learning_rate": 0.00019999051747047783,
-      "loss": 46.0,
-      "step": 27231
-    },
-    {
-      "epoch": 4.385522766617013,
-      "grad_norm": 0.004068195354193449,
-      "learning_rate": 0.0001999905167737691,
-      "loss": 46.0,
-      "step": 27232
-    },
-    {
-      "epoch": 4.385683803695801,
-      "grad_norm": 0.004521645605564117,
-      "learning_rate": 0.00019999051607703483,
-      "loss": 46.0,
-      "step": 27233
-    },
-    {
-      "epoch": 4.385844840774588,
-      "grad_norm": 0.0013227129820734262,
-      "learning_rate": 0.00019999051538027493,
-      "loss": 46.0,
-      "step": 27234
-    },
-    {
-      "epoch": 4.386005877853376,
-      "grad_norm": 0.011316705495119095,
-      "learning_rate": 0.00019999051468348947,
-      "loss": 46.0,
-      "step": 27235
-    },
-    {
-      "epoch": 4.386166914932163,
-      "grad_norm": 0.0023729936219751835,
-      "learning_rate": 0.00019999051398667837,
-      "loss": 46.0,
-      "step": 27236
-    },
-    {
-      "epoch": 4.386327952010951,
-      "grad_norm": 0.0026012288872152567,
-      "learning_rate": 0.00019999051328984174,
-      "loss": 46.0,
-      "step": 27237
-    },
-    {
-      "epoch": 4.386488989089738,
-      "grad_norm": 0.0019430381944403052,
-      "learning_rate": 0.00019999051259297947,
-      "loss": 46.0,
-      "step": 27238
-    },
-    {
-      "epoch": 4.3866500261685255,
-      "grad_norm": 0.003454585326835513,
-      "learning_rate": 0.00019999051189609158,
-      "loss": 46.0,
-      "step": 27239
-    },
-    {
-      "epoch": 4.386811063247313,
-      "grad_norm": 0.00687748659402132,
-      "learning_rate": 0.00019999051119917816,
-      "loss": 46.0,
-      "step": 27240
-    },
-    {
-      "epoch": 4.3869721003261,
-      "grad_norm": 0.0034899902530014515,
-      "learning_rate": 0.00019999051050223913,
-      "loss": 46.0,
-      "step": 27241
-    },
-    {
-      "epoch": 4.387133137404888,
-      "grad_norm": 0.00203665136359632,
-      "learning_rate": 0.0001999905098052745,
-      "loss": 46.0,
-      "step": 27242
-    },
-    {
-      "epoch": 4.387294174483674,
-      "grad_norm": 0.002735869726166129,
-      "learning_rate": 0.00019999050910828427,
-      "loss": 46.0,
-      "step": 27243
-    },
-    {
-      "epoch": 4.387455211562462,
-      "grad_norm": 0.002177219605073333,
-      "learning_rate": 0.00019999050841126845,
-      "loss": 46.0,
-      "step": 27244
-    },
-    {
-      "epoch": 4.387616248641249,
-      "grad_norm": 0.0011689657112583518,
-      "learning_rate": 0.00019999050771422704,
-      "loss": 46.0,
-      "step": 27245
-    },
-    {
-      "epoch": 4.387777285720037,
-      "grad_norm": 0.006479209288954735,
-      "learning_rate": 0.00019999050701716007,
-      "loss": 46.0,
-      "step": 27246
-    },
-    {
-      "epoch": 4.387938322798824,
-      "grad_norm": 0.008870921097695827,
-      "learning_rate": 0.00019999050632006746,
-      "loss": 46.0,
-      "step": 27247
-    },
-    {
-      "epoch": 4.388099359877612,
-      "grad_norm": 0.007648714818060398,
-      "learning_rate": 0.00019999050562294929,
-      "loss": 46.0,
-      "step": 27248
-    },
-    {
-      "epoch": 4.388260396956399,
-      "grad_norm": 0.0023880142252892256,
-      "learning_rate": 0.0001999905049258055,
-      "loss": 46.0,
-      "step": 27249
-    },
-    {
-      "epoch": 4.388421434035187,
-      "grad_norm": 0.008994984440505505,
-      "learning_rate": 0.00019999050422863615,
-      "loss": 46.0,
-      "step": 27250
-    },
-    {
-      "epoch": 4.388582471113974,
-      "grad_norm": 0.003872970351949334,
-      "learning_rate": 0.0001999905035314412,
-      "loss": 46.0,
-      "step": 27251
-    },
-    {
-      "epoch": 4.3887435081927615,
-      "grad_norm": 0.00975795928388834,
-      "learning_rate": 0.00019999050283422062,
-      "loss": 46.0,
-      "step": 27252
-    },
-    {
-      "epoch": 4.388904545271549,
-      "grad_norm": 0.00978830549865961,
-      "learning_rate": 0.00019999050213697449,
-      "loss": 46.0,
-      "step": 27253
-    },
-    {
-      "epoch": 4.389065582350336,
-      "grad_norm": 0.011856836266815662,
-      "learning_rate": 0.00019999050143970274,
-      "loss": 46.0,
-      "step": 27254
-    },
-    {
-      "epoch": 4.389226619429124,
-      "grad_norm": 0.0014544740552082658,
-      "learning_rate": 0.00019999050074240543,
-      "loss": 46.0,
-      "step": 27255
-    },
-    {
-      "epoch": 4.389387656507911,
-      "grad_norm": 0.00330316717736423,
-      "learning_rate": 0.00019999050004508248,
-      "loss": 46.0,
-      "step": 27256
-    },
-    {
-      "epoch": 4.389548693586699,
-      "grad_norm": 0.0014017503708600998,
-      "learning_rate": 0.000199990499347734,
-      "loss": 46.0,
-      "step": 27257
-    },
-    {
-      "epoch": 4.389709730665485,
-      "grad_norm": 0.0007029243861325085,
-      "learning_rate": 0.00019999049865035987,
-      "loss": 46.0,
-      "step": 27258
-    },
-    {
-      "epoch": 4.389870767744273,
-      "grad_norm": 0.004955565556883812,
-      "learning_rate": 0.00019999049795296016,
-      "loss": 46.0,
-      "step": 27259
-    },
-    {
-      "epoch": 4.39003180482306,
-      "grad_norm": 0.0024757403880357742,
-      "learning_rate": 0.00019999049725553486,
-      "loss": 46.0,
-      "step": 27260
-    },
-    {
-      "epoch": 4.390192841901848,
-      "grad_norm": 0.0020914538763463497,
-      "learning_rate": 0.000199990496558084,
-      "loss": 46.0,
-      "step": 27261
-    },
-    {
-      "epoch": 4.390353878980635,
-      "grad_norm": 0.002094556577503681,
-      "learning_rate": 0.00019999049586060753,
-      "loss": 46.0,
-      "step": 27262
-    },
-    {
-      "epoch": 4.390514916059423,
-      "grad_norm": 0.00696517201140523,
-      "learning_rate": 0.00019999049516310544,
-      "loss": 46.0,
-      "step": 27263
-    },
-    {
-      "epoch": 4.39067595313821,
-      "grad_norm": 0.0029635962564498186,
-      "learning_rate": 0.0001999904944655778,
-      "loss": 46.0,
-      "step": 27264
-    },
-    {
-      "epoch": 4.3908369902169975,
-      "grad_norm": 0.0019418272422626615,
-      "learning_rate": 0.00019999049376802456,
-      "loss": 46.0,
-      "step": 27265
-    },
-    {
-      "epoch": 4.390998027295785,
-      "grad_norm": 0.0016454599099233747,
-      "learning_rate": 0.00019999049307044568,
-      "loss": 46.0,
-      "step": 27266
-    },
-    {
-      "epoch": 4.391159064374572,
-      "grad_norm": 0.00554841011762619,
-      "learning_rate": 0.00019999049237284124,
-      "loss": 46.0,
-      "step": 27267
-    },
-    {
-      "epoch": 4.39132010145336,
-      "grad_norm": 0.003708989592269063,
-      "learning_rate": 0.0001999904916752112,
-      "loss": 46.0,
-      "step": 27268
-    },
-    {
-      "epoch": 4.391481138532147,
-      "grad_norm": 0.003114469349384308,
-      "learning_rate": 0.00019999049097755558,
-      "loss": 46.0,
-      "step": 27269
-    },
-    {
-      "epoch": 4.391642175610935,
-      "grad_norm": 0.008489621803164482,
-      "learning_rate": 0.00019999049027987439,
-      "loss": 46.0,
-      "step": 27270
-    },
-    {
-      "epoch": 4.391803212689722,
-      "grad_norm": 0.017003241926431656,
-      "learning_rate": 0.00019999048958216755,
-      "loss": 46.0,
-      "step": 27271
-    },
-    {
-      "epoch": 4.39196424976851,
-      "grad_norm": 0.006560994312167168,
-      "learning_rate": 0.00019999048888443515,
-      "loss": 46.0,
-      "step": 27272
-    },
-    {
-      "epoch": 4.392125286847296,
-      "grad_norm": 0.007608387619256973,
-      "learning_rate": 0.00019999048818667716,
-      "loss": 46.0,
-      "step": 27273
-    },
-    {
-      "epoch": 4.392286323926084,
-      "grad_norm": 0.0036803975235670805,
-      "learning_rate": 0.0001999904874888936,
-      "loss": 46.0,
-      "step": 27274
-    },
-    {
-      "epoch": 4.392447361004871,
-      "grad_norm": 0.0007890161359682679,
-      "learning_rate": 0.0001999904867910844,
-      "loss": 46.0,
-      "step": 27275
-    },
-    {
-      "epoch": 4.3926083980836585,
-      "grad_norm": 0.009090281091630459,
-      "learning_rate": 0.00019999048609324965,
-      "loss": 46.0,
-      "step": 27276
-    },
-    {
-      "epoch": 4.392769435162446,
-      "grad_norm": 0.0024434817023575306,
-      "learning_rate": 0.00019999048539538926,
-      "loss": 46.0,
-      "step": 27277
-    },
-    {
-      "epoch": 4.3929304722412335,
-      "grad_norm": 0.005545934662222862,
-      "learning_rate": 0.0001999904846975033,
-      "loss": 46.0,
-      "step": 27278
-    },
-    {
-      "epoch": 4.393091509320021,
-      "grad_norm": 0.01016025710850954,
-      "learning_rate": 0.00019999048399959178,
-      "loss": 46.0,
-      "step": 27279
-    },
-    {
-      "epoch": 4.393252546398808,
-      "grad_norm": 0.01716170646250248,
-      "learning_rate": 0.00019999048330165463,
-      "loss": 46.0,
-      "step": 27280
-    },
-    {
-      "epoch": 4.393413583477596,
-      "grad_norm": 0.0040521337650716305,
-      "learning_rate": 0.0001999904826036919,
-      "loss": 46.0,
-      "step": 27281
-    },
-    {
-      "epoch": 4.393574620556383,
-      "grad_norm": 0.0064969779923558235,
-      "learning_rate": 0.00019999048190570356,
-      "loss": 46.0,
-      "step": 27282
-    },
-    {
-      "epoch": 4.393735657635171,
-      "grad_norm": 0.0027617421001195908,
-      "learning_rate": 0.00019999048120768965,
-      "loss": 46.0,
-      "step": 27283
-    },
-    {
-      "epoch": 4.393896694713958,
-      "grad_norm": 0.0012256621848791838,
-      "learning_rate": 0.00019999048050965012,
-      "loss": 46.0,
-      "step": 27284
-    },
-    {
-      "epoch": 4.394057731792746,
-      "grad_norm": 0.003906638827174902,
-      "learning_rate": 0.00019999047981158504,
-      "loss": 46.0,
-      "step": 27285
-    },
-    {
-      "epoch": 4.394218768871533,
-      "grad_norm": 0.004821930546313524,
-      "learning_rate": 0.00019999047911349434,
-      "loss": 46.0,
-      "step": 27286
-    },
-    {
-      "epoch": 4.3943798059503205,
-      "grad_norm": 0.00246910797432065,
-      "learning_rate": 0.00019999047841537805,
-      "loss": 46.0,
-      "step": 27287
-    },
-    {
-      "epoch": 4.394540843029107,
-      "grad_norm": 0.002774229273200035,
-      "learning_rate": 0.00019999047771723615,
-      "loss": 46.0,
-      "step": 27288
-    },
-    {
-      "epoch": 4.3947018801078945,
-      "grad_norm": 0.0012742452090606093,
-      "learning_rate": 0.00019999047701906868,
-      "loss": 46.0,
-      "step": 27289
-    },
-    {
-      "epoch": 4.394862917186682,
-      "grad_norm": 0.010323584079742432,
-      "learning_rate": 0.0001999904763208756,
-      "loss": 46.0,
-      "step": 27290
-    },
-    {
-      "epoch": 4.395023954265469,
-      "grad_norm": 0.0025238331872969866,
-      "learning_rate": 0.00019999047562265697,
-      "loss": 46.0,
-      "step": 27291
-    },
-    {
-      "epoch": 4.395184991344257,
-      "grad_norm": 0.005917206406593323,
-      "learning_rate": 0.00019999047492441272,
-      "loss": 46.0,
-      "step": 27292
-    },
-    {
-      "epoch": 4.395346028423044,
-      "grad_norm": 0.0033823102712631226,
-      "learning_rate": 0.00019999047422614288,
-      "loss": 46.0,
-      "step": 27293
-    },
-    {
-      "epoch": 4.395507065501832,
-      "grad_norm": 0.01712118834257126,
-      "learning_rate": 0.00019999047352784743,
-      "loss": 46.0,
-      "step": 27294
-    },
-    {
-      "epoch": 4.395668102580619,
-      "grad_norm": 0.006925694178789854,
-      "learning_rate": 0.00019999047282952642,
-      "loss": 46.0,
-      "step": 27295
-    },
-    {
-      "epoch": 4.395829139659407,
-      "grad_norm": 0.003581579774618149,
-      "learning_rate": 0.0001999904721311798,
-      "loss": 46.0,
-      "step": 27296
-    },
-    {
-      "epoch": 4.395990176738194,
-      "grad_norm": 0.015719318762421608,
-      "learning_rate": 0.00019999047143280757,
-      "loss": 46.0,
-      "step": 27297
-    },
-    {
-      "epoch": 4.396151213816982,
-      "grad_norm": 0.002585549606010318,
-      "learning_rate": 0.00019999047073440977,
-      "loss": 46.0,
-      "step": 27298
-    },
-    {
-      "epoch": 4.396312250895769,
-      "grad_norm": 0.004612908698618412,
-      "learning_rate": 0.00019999047003598638,
-      "loss": 46.0,
-      "step": 27299
-    },
-    {
-      "epoch": 4.3964732879745565,
-      "grad_norm": 0.004402322694659233,
-      "learning_rate": 0.00019999046933753738,
-      "loss": 46.0,
-      "step": 27300
-    },
-    {
-      "epoch": 4.396634325053343,
-      "grad_norm": 0.012610073201358318,
-      "learning_rate": 0.0001999904686390628,
-      "loss": 46.0,
-      "step": 27301
-    },
-    {
-      "epoch": 4.3967953621321305,
-      "grad_norm": 0.0031418197322636843,
-      "learning_rate": 0.00019999046794056262,
-      "loss": 46.0,
-      "step": 27302
-    },
-    {
-      "epoch": 4.396956399210918,
-      "grad_norm": 0.006623541936278343,
-      "learning_rate": 0.00019999046724203688,
-      "loss": 46.0,
-      "step": 27303
-    },
-    {
-      "epoch": 4.397117436289705,
-      "grad_norm": 0.005509393289685249,
-      "learning_rate": 0.0001999904665434855,
-      "loss": 46.0,
-      "step": 27304
-    },
-    {
-      "epoch": 4.397278473368493,
-      "grad_norm": 0.0046736011281609535,
-      "learning_rate": 0.00019999046584490853,
-      "loss": 46.0,
-      "step": 27305
-    },
-    {
-      "epoch": 4.39743951044728,
-      "grad_norm": 0.01256459392607212,
-      "learning_rate": 0.00019999046514630598,
-      "loss": 46.0,
-      "step": 27306
-    },
-    {
-      "epoch": 4.397600547526068,
-      "grad_norm": 0.0038980746176093817,
-      "learning_rate": 0.00019999046444767787,
-      "loss": 46.0,
-      "step": 27307
-    },
-    {
-      "epoch": 4.397761584604855,
-      "grad_norm": 0.0073218620382249355,
-      "learning_rate": 0.00019999046374902414,
-      "loss": 46.0,
-      "step": 27308
-    },
-    {
-      "epoch": 4.397922621683643,
-      "grad_norm": 0.002032952383160591,
-      "learning_rate": 0.0001999904630503448,
-      "loss": 46.0,
-      "step": 27309
-    },
-    {
-      "epoch": 4.39808365876243,
-      "grad_norm": 0.012131624855101109,
-      "learning_rate": 0.0001999904623516399,
-      "loss": 46.0,
-      "step": 27310
-    },
-    {
-      "epoch": 4.398244695841218,
-      "grad_norm": 0.008933956734836102,
-      "learning_rate": 0.0001999904616529094,
-      "loss": 46.0,
-      "step": 27311
-    },
-    {
-      "epoch": 4.398405732920005,
-      "grad_norm": 0.0033500180579721928,
-      "learning_rate": 0.00019999046095415328,
-      "loss": 46.0,
-      "step": 27312
-    },
-    {
-      "epoch": 4.3985667699987925,
-      "grad_norm": 0.0023320051841437817,
-      "learning_rate": 0.0001999904602553716,
-      "loss": 46.0,
-      "step": 27313
-    },
-    {
-      "epoch": 4.39872780707758,
-      "grad_norm": 0.002560779917985201,
-      "learning_rate": 0.0001999904595565643,
-      "loss": 46.0,
-      "step": 27314
-    },
-    {
-      "epoch": 4.398888844156367,
-      "grad_norm": 0.012891615740954876,
-      "learning_rate": 0.00019999045885773145,
-      "loss": 46.0,
-      "step": 27315
-    },
-    {
-      "epoch": 4.399049881235154,
-      "grad_norm": 0.008207086473703384,
-      "learning_rate": 0.00019999045815887297,
-      "loss": 46.0,
-      "step": 27316
-    },
-    {
-      "epoch": 4.399210918313941,
-      "grad_norm": 0.0012559652095660567,
-      "learning_rate": 0.0001999904574599889,
-      "loss": 46.0,
-      "step": 27317
-    },
-    {
-      "epoch": 4.399371955392729,
-      "grad_norm": 0.001945055671967566,
-      "learning_rate": 0.00019999045676107925,
-      "loss": 46.0,
-      "step": 27318
-    },
-    {
-      "epoch": 4.399532992471516,
-      "grad_norm": 0.0026621054857969284,
-      "learning_rate": 0.000199990456062144,
-      "loss": 46.0,
-      "step": 27319
-    },
-    {
-      "epoch": 4.399694029550304,
-      "grad_norm": 0.002348612993955612,
-      "learning_rate": 0.00019999045536318318,
-      "loss": 46.0,
-      "step": 27320
-    },
-    {
-      "epoch": 4.399855066629091,
-      "grad_norm": 0.01567247323691845,
-      "learning_rate": 0.00019999045466419674,
-      "loss": 46.0,
-      "step": 27321
-    },
-    {
-      "epoch": 4.400016103707879,
-      "grad_norm": 0.004743401426821947,
-      "learning_rate": 0.0001999904539651847,
-      "loss": 46.0,
-      "step": 27322
-    },
-    {
-      "epoch": 4.400177140786666,
-      "grad_norm": 0.011594566516578197,
-      "learning_rate": 0.0001999904532661471,
-      "loss": 46.0,
-      "step": 27323
-    },
-    {
-      "epoch": 4.400338177865454,
-      "grad_norm": 0.004417700227349997,
-      "learning_rate": 0.0001999904525670839,
-      "loss": 46.0,
-      "step": 27324
-    },
-    {
-      "epoch": 4.400499214944241,
-      "grad_norm": 0.0013746990589424968,
-      "learning_rate": 0.0001999904518679951,
-      "loss": 46.0,
-      "step": 27325
-    },
-    {
-      "epoch": 4.4006602520230285,
-      "grad_norm": 0.0018068519420921803,
-      "learning_rate": 0.0001999904511688807,
-      "loss": 46.0,
-      "step": 27326
-    },
-    {
-      "epoch": 4.400821289101816,
-      "grad_norm": 0.0023612671066075563,
-      "learning_rate": 0.0001999904504697407,
-      "loss": 46.0,
-      "step": 27327
-    },
-    {
-      "epoch": 4.400982326180603,
-      "grad_norm": 0.0010737653356045485,
-      "learning_rate": 0.00019999044977057513,
-      "loss": 46.0,
-      "step": 27328
-    },
-    {
-      "epoch": 4.401143363259391,
-      "grad_norm": 0.007300083991140127,
-      "learning_rate": 0.000199990449071384,
-      "loss": 46.0,
-      "step": 27329
-    },
-    {
-      "epoch": 4.401304400338178,
-      "grad_norm": 0.004863124340772629,
-      "learning_rate": 0.0001999904483721672,
-      "loss": 46.0,
-      "step": 27330
-    },
-    {
-      "epoch": 4.401465437416965,
-      "grad_norm": 0.0029079876840114594,
-      "learning_rate": 0.00019999044767292484,
-      "loss": 46.0,
-      "step": 27331
-    },
-    {
-      "epoch": 4.401626474495752,
-      "grad_norm": 0.002203600015491247,
-      "learning_rate": 0.0001999904469736569,
-      "loss": 46.0,
-      "step": 27332
-    },
-    {
-      "epoch": 4.40178751157454,
-      "grad_norm": 0.0017512129852548242,
-      "learning_rate": 0.00019999044627436337,
-      "loss": 46.0,
-      "step": 27333
-    },
-    {
-      "epoch": 4.401948548653327,
-      "grad_norm": 0.004532170481979847,
-      "learning_rate": 0.00019999044557504422,
-      "loss": 46.0,
-      "step": 27334
-    },
-    {
-      "epoch": 4.402109585732115,
-      "grad_norm": 0.012387637980282307,
-      "learning_rate": 0.0001999904448756995,
-      "loss": 46.0,
-      "step": 27335
-    },
-    {
-      "epoch": 4.402270622810902,
-      "grad_norm": 0.0029165635351091623,
-      "learning_rate": 0.0001999904441763292,
-      "loss": 46.0,
-      "step": 27336
-    },
-    {
-      "epoch": 4.4024316598896895,
-      "grad_norm": 0.003268257947638631,
-      "learning_rate": 0.0001999904434769333,
-      "loss": 46.0,
-      "step": 27337
-    },
-    {
-      "epoch": 4.402592696968477,
-      "grad_norm": 0.028318816795945168,
-      "learning_rate": 0.00019999044277751177,
-      "loss": 46.0,
-      "step": 27338
-    },
-    {
-      "epoch": 4.4027537340472644,
-      "grad_norm": 0.0035808617249131203,
-      "learning_rate": 0.00019999044207806468,
-      "loss": 46.0,
-      "step": 27339
-    },
-    {
-      "epoch": 4.402914771126052,
-      "grad_norm": 0.00875371228903532,
-      "learning_rate": 0.000199990441378592,
-      "loss": 46.0,
-      "step": 27340
-    },
-    {
-      "epoch": 4.403075808204839,
-      "grad_norm": 0.0037388845812529325,
-      "learning_rate": 0.0001999904406790937,
-      "loss": 46.0,
-      "step": 27341
-    },
-    {
-      "epoch": 4.403236845283627,
-      "grad_norm": 0.00576428696513176,
-      "learning_rate": 0.00019999043997956983,
-      "loss": 46.0,
-      "step": 27342
-    },
-    {
-      "epoch": 4.403397882362414,
-      "grad_norm": 0.0019498420879244804,
-      "learning_rate": 0.0001999904392800204,
-      "loss": 46.0,
-      "step": 27343
-    },
-    {
-      "epoch": 4.403558919441202,
-      "grad_norm": 0.0016144951805472374,
-      "learning_rate": 0.00019999043858044533,
-      "loss": 46.0,
-      "step": 27344
-    },
-    {
-      "epoch": 4.403719956519989,
-      "grad_norm": 0.005327403545379639,
-      "learning_rate": 0.0001999904378808447,
-      "loss": 46.0,
-      "step": 27345
-    },
-    {
-      "epoch": 4.403880993598776,
-      "grad_norm": 0.0029097970109432936,
-      "learning_rate": 0.00019999043718121843,
-      "loss": 46.0,
-      "step": 27346
-    },
-    {
-      "epoch": 4.404042030677563,
-      "grad_norm": 0.005063321907073259,
-      "learning_rate": 0.00019999043648156662,
-      "loss": 46.0,
-      "step": 27347
-    },
-    {
-      "epoch": 4.404203067756351,
-      "grad_norm": 0.0027337747160345316,
-      "learning_rate": 0.00019999043578188919,
-      "loss": 46.0,
-      "step": 27348
-    },
-    {
-      "epoch": 4.404364104835138,
-      "grad_norm": 0.00688413018360734,
-      "learning_rate": 0.00019999043508218617,
-      "loss": 46.0,
-      "step": 27349
-    },
-    {
-      "epoch": 4.4045251419139255,
-      "grad_norm": 0.0009647997212596238,
-      "learning_rate": 0.00019999043438245754,
-      "loss": 46.0,
-      "step": 27350
-    },
-    {
-      "epoch": 4.404686178992713,
-      "grad_norm": 0.0017396489856764674,
-      "learning_rate": 0.00019999043368270334,
-      "loss": 46.0,
-      "step": 27351
-    },
-    {
-      "epoch": 4.4048472160715,
-      "grad_norm": 0.0025180771481245756,
-      "learning_rate": 0.00019999043298292354,
-      "loss": 46.0,
-      "step": 27352
-    },
-    {
-      "epoch": 4.405008253150288,
-      "grad_norm": 0.0048584109172225,
-      "learning_rate": 0.00019999043228311814,
-      "loss": 46.0,
-      "step": 27353
-    },
-    {
-      "epoch": 4.405169290229075,
-      "grad_norm": 0.001719545922242105,
-      "learning_rate": 0.0001999904315832872,
-      "loss": 46.0,
-      "step": 27354
-    },
-    {
-      "epoch": 4.405330327307863,
-      "grad_norm": 0.0025414868723601103,
-      "learning_rate": 0.00019999043088343062,
-      "loss": 46.0,
-      "step": 27355
-    },
-    {
-      "epoch": 4.40549136438665,
-      "grad_norm": 0.011448708362877369,
-      "learning_rate": 0.00019999043018354844,
-      "loss": 46.0,
-      "step": 27356
-    },
-    {
-      "epoch": 4.405652401465438,
-      "grad_norm": 0.005573517642915249,
-      "learning_rate": 0.0001999904294836407,
-      "loss": 46.0,
-      "step": 27357
-    },
-    {
-      "epoch": 4.405813438544225,
-      "grad_norm": 0.0020006783306598663,
-      "learning_rate": 0.00019999042878370734,
-      "loss": 46.0,
-      "step": 27358
-    },
-    {
-      "epoch": 4.405974475623013,
-      "grad_norm": 0.01607229746878147,
-      "learning_rate": 0.0001999904280837484,
-      "loss": 46.0,
-      "step": 27359
-    },
-    {
-      "epoch": 4.4061355127018,
-      "grad_norm": 0.0024870475754141808,
-      "learning_rate": 0.00019999042738376386,
-      "loss": 46.0,
-      "step": 27360
-    },
-    {
-      "epoch": 4.406296549780587,
-      "grad_norm": 0.011495199985802174,
-      "learning_rate": 0.00019999042668375375,
-      "loss": 46.0,
-      "step": 27361
-    },
-    {
-      "epoch": 4.406457586859374,
-      "grad_norm": 0.0029110966715961695,
-      "learning_rate": 0.000199990425983718,
-      "loss": 46.0,
-      "step": 27362
-    },
-    {
-      "epoch": 4.4066186239381615,
-      "grad_norm": 0.006751770153641701,
-      "learning_rate": 0.00019999042528365667,
-      "loss": 46.0,
-      "step": 27363
-    },
-    {
-      "epoch": 4.406779661016949,
-      "grad_norm": 0.010595368221402168,
-      "learning_rate": 0.00019999042458356981,
-      "loss": 46.0,
-      "step": 27364
-    },
-    {
-      "epoch": 4.406940698095736,
-      "grad_norm": 0.0059717013500630856,
-      "learning_rate": 0.0001999904238834573,
-      "loss": 46.0,
-      "step": 27365
-    },
-    {
-      "epoch": 4.407101735174524,
-      "grad_norm": 0.002517386106774211,
-      "learning_rate": 0.00019999042318331919,
-      "loss": 46.0,
-      "step": 27366
-    },
-    {
-      "epoch": 4.407262772253311,
-      "grad_norm": 0.0035003568045794964,
-      "learning_rate": 0.00019999042248315552,
-      "loss": 46.0,
-      "step": 27367
-    },
-    {
-      "epoch": 4.407423809332099,
-      "grad_norm": 0.006473209243267775,
-      "learning_rate": 0.00019999042178296623,
-      "loss": 46.0,
-      "step": 27368
-    },
-    {
-      "epoch": 4.407584846410886,
-      "grad_norm": 0.011653761379420757,
-      "learning_rate": 0.00019999042108275136,
-      "loss": 46.0,
-      "step": 27369
-    },
-    {
-      "epoch": 4.407745883489674,
-      "grad_norm": 0.007314703427255154,
-      "learning_rate": 0.00019999042038251093,
-      "loss": 46.0,
-      "step": 27370
-    },
-    {
-      "epoch": 4.407906920568461,
-      "grad_norm": 0.002933237934485078,
-      "learning_rate": 0.00019999041968224486,
-      "loss": 46.0,
-      "step": 27371
-    },
-    {
-      "epoch": 4.408067957647249,
-      "grad_norm": 0.003286631777882576,
-      "learning_rate": 0.0001999904189819532,
-      "loss": 46.0,
-      "step": 27372
-    },
-    {
-      "epoch": 4.408228994726036,
-      "grad_norm": 0.0019488940015435219,
-      "learning_rate": 0.00019999041828163599,
-      "loss": 46.0,
-      "step": 27373
-    },
-    {
-      "epoch": 4.4083900318048235,
-      "grad_norm": 0.002086965600028634,
-      "learning_rate": 0.00019999041758129315,
-      "loss": 46.0,
-      "step": 27374
-    },
-    {
-      "epoch": 4.40855106888361,
-      "grad_norm": 0.0017261735629290342,
-      "learning_rate": 0.00019999041688092473,
-      "loss": 46.0,
-      "step": 27375
-    },
-    {
-      "epoch": 4.4087121059623975,
-      "grad_norm": 0.006090558134019375,
-      "learning_rate": 0.00019999041618053073,
-      "loss": 46.0,
-      "step": 27376
-    },
-    {
-      "epoch": 4.408873143041185,
-      "grad_norm": 0.0021092109382152557,
-      "learning_rate": 0.0001999904154801111,
-      "loss": 46.0,
-      "step": 27377
-    },
-    {
-      "epoch": 4.409034180119972,
-      "grad_norm": 0.006767811253666878,
-      "learning_rate": 0.00019999041477966592,
-      "loss": 46.0,
-      "step": 27378
-    },
-    {
-      "epoch": 4.40919521719876,
-      "grad_norm": 0.002752867992967367,
-      "learning_rate": 0.00019999041407919513,
-      "loss": 46.0,
-      "step": 27379
-    },
-    {
-      "epoch": 4.409356254277547,
-      "grad_norm": 0.004678305238485336,
-      "learning_rate": 0.00019999041337869872,
-      "loss": 46.0,
-      "step": 27380
-    },
-    {
-      "epoch": 4.409517291356335,
-      "grad_norm": 0.0012461270671337843,
-      "learning_rate": 0.00019999041267817677,
-      "loss": 46.0,
-      "step": 27381
-    },
-    {
-      "epoch": 4.409678328435122,
-      "grad_norm": 0.0024450754281133413,
-      "learning_rate": 0.0001999904119776292,
-      "loss": 46.0,
-      "step": 27382
-    },
-    {
-      "epoch": 4.40983936551391,
-      "grad_norm": 0.006418305449187756,
-      "learning_rate": 0.00019999041127705605,
-      "loss": 46.0,
-      "step": 27383
-    },
-    {
-      "epoch": 4.410000402592697,
-      "grad_norm": 0.012624701485037804,
-      "learning_rate": 0.00019999041057645726,
-      "loss": 46.0,
-      "step": 27384
-    },
-    {
-      "epoch": 4.4101614396714846,
-      "grad_norm": 0.0018664766103029251,
-      "learning_rate": 0.00019999040987583294,
-      "loss": 46.0,
-      "step": 27385
-    },
-    {
-      "epoch": 4.410322476750272,
-      "grad_norm": 0.003344783326610923,
-      "learning_rate": 0.000199990409175183,
-      "loss": 46.0,
-      "step": 27386
-    },
-    {
-      "epoch": 4.4104835138290595,
-      "grad_norm": 0.003047558479011059,
-      "learning_rate": 0.00019999040847450746,
-      "loss": 46.0,
-      "step": 27387
-    },
-    {
-      "epoch": 4.410644550907847,
-      "grad_norm": 0.006820410490036011,
-      "learning_rate": 0.00019999040777380635,
-      "loss": 46.0,
-      "step": 27388
-    },
-    {
-      "epoch": 4.4108055879866335,
-      "grad_norm": 0.008419794030487537,
-      "learning_rate": 0.00019999040707307963,
-      "loss": 46.0,
-      "step": 27389
-    },
-    {
-      "epoch": 4.410966625065421,
-      "grad_norm": 0.006269561592489481,
-      "learning_rate": 0.00019999040637232732,
-      "loss": 46.0,
-      "step": 27390
-    },
-    {
-      "epoch": 4.411127662144208,
-      "grad_norm": 0.0024414879735559225,
-      "learning_rate": 0.0001999904056715494,
-      "loss": 46.0,
-      "step": 27391
-    },
-    {
-      "epoch": 4.411288699222996,
-      "grad_norm": 0.0022647599689662457,
-      "learning_rate": 0.00019999040497074592,
-      "loss": 46.0,
-      "step": 27392
-    },
-    {
-      "epoch": 4.411449736301783,
-      "grad_norm": 0.0038857299368828535,
-      "learning_rate": 0.00019999040426991685,
-      "loss": 46.0,
-      "step": 27393
-    },
-    {
-      "epoch": 4.411610773380571,
-      "grad_norm": 0.0017655508127063513,
-      "learning_rate": 0.00019999040356906216,
-      "loss": 46.0,
-      "step": 27394
-    },
-    {
-      "epoch": 4.411771810459358,
-      "grad_norm": 0.0027361814863979816,
-      "learning_rate": 0.0001999904028681819,
-      "loss": 46.0,
-      "step": 27395
-    },
-    {
-      "epoch": 4.411932847538146,
-      "grad_norm": 0.001051982631906867,
-      "learning_rate": 0.00019999040216727606,
-      "loss": 46.0,
-      "step": 27396
-    },
-    {
-      "epoch": 4.412093884616933,
-      "grad_norm": 0.0018975939601659775,
-      "learning_rate": 0.00019999040146634459,
-      "loss": 46.0,
-      "step": 27397
-    },
-    {
-      "epoch": 4.4122549216957205,
-      "grad_norm": 0.0038895714096724987,
-      "learning_rate": 0.00019999040076538752,
-      "loss": 46.0,
-      "step": 27398
-    },
-    {
-      "epoch": 4.412415958774508,
-      "grad_norm": 0.0020249593071639538,
-      "learning_rate": 0.00019999040006440488,
-      "loss": 46.0,
-      "step": 27399
-    },
-    {
-      "epoch": 4.412576995853295,
-      "grad_norm": 0.0033499368000775576,
-      "learning_rate": 0.00019999039936339664,
-      "loss": 46.0,
-      "step": 27400
-    },
-    {
-      "epoch": 4.412738032932083,
-      "grad_norm": 0.0009048962383531034,
-      "learning_rate": 0.00019999039866236285,
-      "loss": 46.0,
-      "step": 27401
-    },
-    {
-      "epoch": 4.41289907001087,
-      "grad_norm": 0.002805883763357997,
-      "learning_rate": 0.0001999903979613034,
-      "loss": 46.0,
-      "step": 27402
-    },
-    {
-      "epoch": 4.413060107089658,
-      "grad_norm": 0.004113157745450735,
-      "learning_rate": 0.0001999903972602184,
-      "loss": 46.0,
-      "step": 27403
-    },
-    {
-      "epoch": 4.413221144168444,
-      "grad_norm": 0.005982518196105957,
-      "learning_rate": 0.0001999903965591078,
-      "loss": 46.0,
-      "step": 27404
-    },
-    {
-      "epoch": 4.413382181247232,
-      "grad_norm": 0.001797487959265709,
-      "learning_rate": 0.0001999903958579716,
-      "loss": 46.0,
-      "step": 27405
-    },
-    {
-      "epoch": 4.413543218326019,
-      "grad_norm": 0.003818618832156062,
-      "learning_rate": 0.00019999039515680985,
-      "loss": 46.0,
-      "step": 27406
-    },
-    {
-      "epoch": 4.413704255404807,
-      "grad_norm": 0.004484761506319046,
-      "learning_rate": 0.00019999039445562248,
-      "loss": 46.0,
-      "step": 27407
-    },
-    {
-      "epoch": 4.413865292483594,
-      "grad_norm": 0.027530241757631302,
-      "learning_rate": 0.00019999039375440946,
-      "loss": 46.0,
-      "step": 27408
-    },
-    {
-      "epoch": 4.414026329562382,
-      "grad_norm": 0.010995124466717243,
-      "learning_rate": 0.00019999039305317092,
-      "loss": 46.0,
-      "step": 27409
-    },
-    {
-      "epoch": 4.414187366641169,
-      "grad_norm": 0.004521520342677832,
-      "learning_rate": 0.00019999039235190676,
-      "loss": 46.0,
-      "step": 27410
-    },
-    {
-      "epoch": 4.4143484037199565,
-      "grad_norm": 0.0023264845367521048,
-      "learning_rate": 0.00019999039165061698,
-      "loss": 46.0,
-      "step": 27411
-    },
-    {
-      "epoch": 4.414509440798744,
-      "grad_norm": 0.002906639128923416,
-      "learning_rate": 0.00019999039094930167,
-      "loss": 46.0,
-      "step": 27412
-    },
-    {
-      "epoch": 4.414670477877531,
-      "grad_norm": 0.002049673581495881,
-      "learning_rate": 0.00019999039024796072,
-      "loss": 46.0,
-      "step": 27413
-    },
-    {
-      "epoch": 4.414831514956319,
-      "grad_norm": 0.007087324280291796,
-      "learning_rate": 0.00019999038954659422,
-      "loss": 46.0,
-      "step": 27414
-    },
-    {
-      "epoch": 4.414992552035106,
-      "grad_norm": 0.0018439792329445481,
-      "learning_rate": 0.0001999903888452021,
-      "loss": 46.0,
-      "step": 27415
-    },
-    {
-      "epoch": 4.415153589113894,
-      "grad_norm": 0.0035840305499732494,
-      "learning_rate": 0.00019999038814378438,
-      "loss": 46.0,
-      "step": 27416
-    },
-    {
-      "epoch": 4.415314626192681,
-      "grad_norm": 0.012445604428648949,
-      "learning_rate": 0.00019999038744234108,
-      "loss": 46.0,
-      "step": 27417
-    },
-    {
-      "epoch": 4.415475663271469,
-      "grad_norm": 0.013269487768411636,
-      "learning_rate": 0.00019999038674087217,
-      "loss": 46.0,
-      "step": 27418
-    },
-    {
-      "epoch": 4.415636700350255,
-      "grad_norm": 0.00528492359444499,
-      "learning_rate": 0.0001999903860393777,
-      "loss": 46.0,
-      "step": 27419
-    },
-    {
-      "epoch": 4.415797737429043,
-      "grad_norm": 0.0030354666523635387,
-      "learning_rate": 0.0001999903853378576,
-      "loss": 46.0,
-      "step": 27420
-    },
-    {
-      "epoch": 4.41595877450783,
-      "grad_norm": 0.00844779796898365,
-      "learning_rate": 0.00019999038463631194,
-      "loss": 46.0,
-      "step": 27421
-    },
-    {
-      "epoch": 4.416119811586618,
-      "grad_norm": 0.0012517308350652456,
-      "learning_rate": 0.00019999038393474068,
-      "loss": 46.0,
-      "step": 27422
-    },
-    {
-      "epoch": 4.416280848665405,
-      "grad_norm": 0.007626294158399105,
-      "learning_rate": 0.00019999038323314383,
-      "loss": 46.0,
-      "step": 27423
-    },
-    {
-      "epoch": 4.4164418857441925,
-      "grad_norm": 0.005166039802134037,
-      "learning_rate": 0.0001999903825315214,
-      "loss": 46.0,
-      "step": 27424
-    },
-    {
-      "epoch": 4.41660292282298,
-      "grad_norm": 0.003854632843285799,
-      "learning_rate": 0.00019999038182987332,
-      "loss": 46.0,
-      "step": 27425
-    },
-    {
-      "epoch": 4.416763959901767,
-      "grad_norm": 0.00282242801040411,
-      "learning_rate": 0.0001999903811281997,
-      "loss": 46.0,
-      "step": 27426
-    },
-    {
-      "epoch": 4.416924996980555,
-      "grad_norm": 0.009118017740547657,
-      "learning_rate": 0.00019999038042650046,
-      "loss": 46.0,
-      "step": 27427
-    },
-    {
-      "epoch": 4.417086034059342,
-      "grad_norm": 0.005772007163614035,
-      "learning_rate": 0.00019999037972477565,
-      "loss": 46.0,
-      "step": 27428
-    },
-    {
-      "epoch": 4.41724707113813,
-      "grad_norm": 0.0037051483523100615,
-      "learning_rate": 0.00019999037902302522,
-      "loss": 46.0,
-      "step": 27429
-    },
-    {
-      "epoch": 4.417408108216917,
-      "grad_norm": 0.0013796199345961213,
-      "learning_rate": 0.00019999037832124923,
-      "loss": 46.0,
-      "step": 27430
-    },
-    {
-      "epoch": 4.417569145295705,
-      "grad_norm": 0.008439082652330399,
-      "learning_rate": 0.00019999037761944763,
-      "loss": 46.0,
-      "step": 27431
-    },
-    {
-      "epoch": 4.417730182374492,
-      "grad_norm": 0.009900303557515144,
-      "learning_rate": 0.00019999037691762045,
-      "loss": 46.0,
-      "step": 27432
-    },
-    {
-      "epoch": 4.41789121945328,
-      "grad_norm": 0.001965191913768649,
-      "learning_rate": 0.00019999037621576767,
-      "loss": 46.0,
-      "step": 27433
-    },
-    {
-      "epoch": 4.418052256532066,
-      "grad_norm": 0.00998140498995781,
-      "learning_rate": 0.0001999903755138893,
-      "loss": 46.0,
-      "step": 27434
-    },
-    {
-      "epoch": 4.418213293610854,
-      "grad_norm": 0.0008698838646523654,
-      "learning_rate": 0.00019999037481198534,
-      "loss": 46.0,
-      "step": 27435
-    },
-    {
-      "epoch": 4.418374330689641,
-      "grad_norm": 0.007133046630769968,
-      "learning_rate": 0.00019999037411005577,
-      "loss": 46.0,
-      "step": 27436
-    },
-    {
-      "epoch": 4.4185353677684285,
-      "grad_norm": 0.0007299024146050215,
-      "learning_rate": 0.0001999903734081006,
-      "loss": 46.0,
-      "step": 27437
-    },
-    {
-      "epoch": 4.418696404847216,
-      "grad_norm": 0.003916271962225437,
-      "learning_rate": 0.00019999037270611986,
-      "loss": 46.0,
-      "step": 27438
-    },
-    {
-      "epoch": 4.418857441926003,
-      "grad_norm": 0.001871993183158338,
-      "learning_rate": 0.00019999037200411354,
-      "loss": 46.0,
-      "step": 27439
-    },
-    {
-      "epoch": 4.419018479004791,
-      "grad_norm": 0.004256522282958031,
-      "learning_rate": 0.0001999903713020816,
-      "loss": 46.0,
-      "step": 27440
-    },
-    {
-      "epoch": 4.419179516083578,
-      "grad_norm": 0.0019722296856343746,
-      "learning_rate": 0.00019999037060002407,
-      "loss": 46.0,
-      "step": 27441
-    },
-    {
-      "epoch": 4.419340553162366,
-      "grad_norm": 0.004146620165556669,
-      "learning_rate": 0.000199990369897941,
-      "loss": 46.0,
-      "step": 27442
-    },
-    {
-      "epoch": 4.419501590241153,
-      "grad_norm": 0.010430843569338322,
-      "learning_rate": 0.00019999036919583226,
-      "loss": 46.0,
-      "step": 27443
-    },
-    {
-      "epoch": 4.419662627319941,
-      "grad_norm": 0.001501597696915269,
-      "learning_rate": 0.00019999036849369797,
-      "loss": 46.0,
-      "step": 27444
-    },
-    {
-      "epoch": 4.419823664398728,
-      "grad_norm": 0.013357128947973251,
-      "learning_rate": 0.00019999036779153807,
-      "loss": 46.0,
-      "step": 27445
-    },
-    {
-      "epoch": 4.4199847014775155,
-      "grad_norm": 0.005096310283988714,
-      "learning_rate": 0.00019999036708935258,
-      "loss": 46.0,
-      "step": 27446
-    },
-    {
-      "epoch": 4.420145738556303,
-      "grad_norm": 0.008011817000806332,
-      "learning_rate": 0.00019999036638714153,
-      "loss": 46.0,
-      "step": 27447
-    },
-    {
-      "epoch": 4.4203067756350904,
-      "grad_norm": 0.004170471336692572,
-      "learning_rate": 0.00019999036568490484,
-      "loss": 46.0,
-      "step": 27448
-    },
-    {
-      "epoch": 4.420467812713877,
-      "grad_norm": 0.0007771399104967713,
-      "learning_rate": 0.0001999903649826426,
-      "loss": 46.0,
-      "step": 27449
-    },
-    {
-      "epoch": 4.4206288497926645,
-      "grad_norm": 0.005197890568524599,
-      "learning_rate": 0.00019999036428035476,
-      "loss": 46.0,
-      "step": 27450
-    },
-    {
-      "epoch": 4.420789886871452,
-      "grad_norm": 0.0008401761297136545,
-      "learning_rate": 0.0001999903635780413,
-      "loss": 46.0,
-      "step": 27451
-    },
-    {
-      "epoch": 4.420950923950239,
-      "grad_norm": 0.003124451031908393,
-      "learning_rate": 0.00019999036287570227,
-      "loss": 46.0,
-      "step": 27452
-    },
-    {
-      "epoch": 4.421111961029027,
-      "grad_norm": 0.006259698886424303,
-      "learning_rate": 0.0001999903621733376,
-      "loss": 46.0,
-      "step": 27453
-    },
-    {
-      "epoch": 4.421272998107814,
-      "grad_norm": 0.004135970026254654,
-      "learning_rate": 0.0001999903614709474,
-      "loss": 46.0,
-      "step": 27454
-    },
-    {
-      "epoch": 4.421434035186602,
-      "grad_norm": 0.008526977151632309,
-      "learning_rate": 0.0001999903607685316,
-      "loss": 46.0,
-      "step": 27455
-    },
-    {
-      "epoch": 4.421595072265389,
-      "grad_norm": 0.003540364094078541,
-      "learning_rate": 0.00019999036006609019,
-      "loss": 46.0,
-      "step": 27456
-    },
-    {
-      "epoch": 4.421756109344177,
-      "grad_norm": 0.00501538161188364,
-      "learning_rate": 0.0001999903593636232,
-      "loss": 46.0,
-      "step": 27457
-    },
-    {
-      "epoch": 4.421917146422964,
-      "grad_norm": 0.004424090497195721,
-      "learning_rate": 0.0001999903586611306,
-      "loss": 46.0,
-      "step": 27458
-    },
-    {
-      "epoch": 4.4220781835017515,
-      "grad_norm": 0.005353490822017193,
-      "learning_rate": 0.00019999035795861242,
-      "loss": 46.0,
-      "step": 27459
-    },
-    {
-      "epoch": 4.422239220580539,
-      "grad_norm": 0.007835939526557922,
-      "learning_rate": 0.00019999035725606863,
-      "loss": 46.0,
-      "step": 27460
-    },
-    {
-      "epoch": 4.422400257659326,
-      "grad_norm": 0.004030084237456322,
-      "learning_rate": 0.00019999035655349928,
-      "loss": 46.0,
-      "step": 27461
-    },
-    {
-      "epoch": 4.422561294738113,
-      "grad_norm": 0.012688993476331234,
-      "learning_rate": 0.00019999035585090431,
-      "loss": 46.0,
-      "step": 27462
-    },
-    {
-      "epoch": 4.4227223318169,
-      "grad_norm": 0.0035829246044158936,
-      "learning_rate": 0.00019999035514828373,
-      "loss": 46.0,
-      "step": 27463
-    },
-    {
-      "epoch": 4.422883368895688,
-      "grad_norm": 0.0062058595940470695,
-      "learning_rate": 0.0001999903544456376,
-      "loss": 46.0,
-      "step": 27464
-    },
-    {
-      "epoch": 4.423044405974475,
-      "grad_norm": 0.011897902935743332,
-      "learning_rate": 0.0001999903537429659,
-      "loss": 46.0,
-      "step": 27465
-    },
-    {
-      "epoch": 4.423205443053263,
-      "grad_norm": 0.0048384033143520355,
-      "learning_rate": 0.00019999035304026853,
-      "loss": 46.0,
-      "step": 27466
-    },
-    {
-      "epoch": 4.42336648013205,
-      "grad_norm": 0.0071312733925879,
-      "learning_rate": 0.00019999035233754563,
-      "loss": 46.0,
-      "step": 27467
-    },
-    {
-      "epoch": 4.423527517210838,
-      "grad_norm": 0.005477431230247021,
-      "learning_rate": 0.0001999903516347971,
-      "loss": 46.0,
-      "step": 27468
-    },
-    {
-      "epoch": 4.423688554289625,
-      "grad_norm": 0.013101095333695412,
-      "learning_rate": 0.00019999035093202298,
-      "loss": 46.0,
-      "step": 27469
-    },
-    {
-      "epoch": 4.423849591368413,
-      "grad_norm": 0.006468312349170446,
-      "learning_rate": 0.0001999903502292233,
-      "loss": 46.0,
-      "step": 27470
-    },
-    {
-      "epoch": 4.4240106284472,
-      "grad_norm": 0.010020007379353046,
-      "learning_rate": 0.000199990349526398,
-      "loss": 46.0,
-      "step": 27471
-    },
-    {
-      "epoch": 4.4241716655259875,
-      "grad_norm": 0.005001259967684746,
-      "learning_rate": 0.0001999903488235471,
-      "loss": 46.0,
-      "step": 27472
-    },
-    {
-      "epoch": 4.424332702604775,
-      "grad_norm": 0.0012007616460323334,
-      "learning_rate": 0.00019999034812067062,
-      "loss": 46.0,
-      "step": 27473
-    },
-    {
-      "epoch": 4.424493739683562,
-      "grad_norm": 0.00495382584631443,
-      "learning_rate": 0.00019999034741776856,
-      "loss": 46.0,
-      "step": 27474
-    },
-    {
-      "epoch": 4.42465477676235,
-      "grad_norm": 0.011137055233120918,
-      "learning_rate": 0.0001999903467148409,
-      "loss": 46.0,
-      "step": 27475
-    },
-    {
-      "epoch": 4.424815813841137,
-      "grad_norm": 0.0017682421021163464,
-      "learning_rate": 0.00019999034601188767,
-      "loss": 46.0,
-      "step": 27476
-    },
-    {
-      "epoch": 4.424976850919924,
-      "grad_norm": 0.005838059354573488,
-      "learning_rate": 0.0001999903453089088,
-      "loss": 46.0,
-      "step": 27477
-    },
-    {
-      "epoch": 4.425137887998711,
-      "grad_norm": 0.0027915502432733774,
-      "learning_rate": 0.00019999034460590437,
-      "loss": 46.0,
-      "step": 27478
-    },
-    {
-      "epoch": 4.425298925077499,
-      "grad_norm": 0.003008077386766672,
-      "learning_rate": 0.00019999034390287435,
-      "loss": 46.0,
-      "step": 27479
-    },
-    {
-      "epoch": 4.425459962156286,
-      "grad_norm": 0.00393042154610157,
-      "learning_rate": 0.0001999903431998187,
-      "loss": 46.0,
-      "step": 27480
-    },
-    {
-      "epoch": 4.425620999235074,
-      "grad_norm": 0.004699901212006807,
-      "learning_rate": 0.0001999903424967375,
-      "loss": 46.0,
-      "step": 27481
-    },
-    {
-      "epoch": 4.425782036313861,
-      "grad_norm": 0.004551334772258997,
-      "learning_rate": 0.00019999034179363069,
-      "loss": 46.0,
-      "step": 27482
-    },
-    {
-      "epoch": 4.425943073392649,
-      "grad_norm": 0.008647710084915161,
-      "learning_rate": 0.00019999034109049828,
-      "loss": 46.0,
-      "step": 27483
-    },
-    {
-      "epoch": 4.426104110471436,
-      "grad_norm": 0.00466693053022027,
-      "learning_rate": 0.0001999903403873403,
-      "loss": 46.0,
-      "step": 27484
-    },
-    {
-      "epoch": 4.4262651475502235,
-      "grad_norm": 0.009514503180980682,
-      "learning_rate": 0.0001999903396841567,
-      "loss": 46.0,
-      "step": 27485
-    },
-    {
-      "epoch": 4.426426184629011,
-      "grad_norm": 0.010345939546823502,
-      "learning_rate": 0.00019999033898094752,
-      "loss": 46.0,
-      "step": 27486
-    },
-    {
-      "epoch": 4.426587221707798,
-      "grad_norm": 0.0046316394582390785,
-      "learning_rate": 0.00019999033827771277,
-      "loss": 46.0,
-      "step": 27487
-    },
-    {
-      "epoch": 4.426748258786586,
-      "grad_norm": 0.015254509635269642,
-      "learning_rate": 0.00019999033757445237,
-      "loss": 46.0,
-      "step": 27488
-    },
-    {
-      "epoch": 4.426909295865373,
-      "grad_norm": 0.0034906507935374975,
-      "learning_rate": 0.00019999033687116642,
-      "loss": 46.0,
-      "step": 27489
-    },
-    {
-      "epoch": 4.427070332944161,
-      "grad_norm": 0.0015474860556423664,
-      "learning_rate": 0.00019999033616785488,
-      "loss": 46.0,
-      "step": 27490
-    },
-    {
-      "epoch": 4.427231370022948,
-      "grad_norm": 0.0026442024391144514,
-      "learning_rate": 0.00019999033546451775,
-      "loss": 46.0,
-      "step": 27491
-    },
-    {
-      "epoch": 4.427392407101735,
-      "grad_norm": 0.013181975111365318,
-      "learning_rate": 0.000199990334761155,
-      "loss": 46.0,
-      "step": 27492
-    },
-    {
-      "epoch": 4.427553444180522,
-      "grad_norm": 0.003948105964809656,
-      "learning_rate": 0.00019999033405776668,
-      "loss": 46.0,
-      "step": 27493
-    },
-    {
-      "epoch": 4.42771448125931,
-      "grad_norm": 0.013080221600830555,
-      "learning_rate": 0.0001999903333543528,
-      "loss": 46.0,
-      "step": 27494
-    },
-    {
-      "epoch": 4.427875518338097,
-      "grad_norm": 0.002306859940290451,
-      "learning_rate": 0.00019999033265091326,
-      "loss": 46.0,
-      "step": 27495
-    },
-    {
-      "epoch": 4.428036555416885,
-      "grad_norm": 0.004247264936566353,
-      "learning_rate": 0.00019999033194744814,
-      "loss": 46.0,
-      "step": 27496
-    },
-    {
-      "epoch": 4.428197592495672,
-      "grad_norm": 0.014462347142398357,
-      "learning_rate": 0.00019999033124395746,
-      "loss": 46.0,
-      "step": 27497
-    },
-    {
-      "epoch": 4.4283586295744595,
-      "grad_norm": 0.0006957217119634151,
-      "learning_rate": 0.0001999903305404412,
-      "loss": 46.0,
-      "step": 27498
-    },
-    {
-      "epoch": 4.428519666653247,
-      "grad_norm": 0.0009768378222361207,
-      "learning_rate": 0.0001999903298368993,
-      "loss": 46.0,
-      "step": 27499
-    },
-    {
-      "epoch": 4.428680703732034,
-      "grad_norm": 0.00218068715184927,
-      "learning_rate": 0.00019999032913333182,
-      "loss": 46.0,
-      "step": 27500
-    },
-    {
-      "epoch": 4.428841740810822,
-      "grad_norm": 0.005291748326271772,
-      "learning_rate": 0.00019999032842973874,
-      "loss": 46.0,
-      "step": 27501
-    },
-    {
-      "epoch": 4.429002777889609,
-      "grad_norm": 0.0007746998453512788,
-      "learning_rate": 0.0001999903277261201,
-      "loss": 46.0,
-      "step": 27502
-    },
-    {
-      "epoch": 4.429163814968397,
-      "grad_norm": 0.006851615384221077,
-      "learning_rate": 0.00019999032702247585,
-      "loss": 46.0,
-      "step": 27503
-    },
-    {
-      "epoch": 4.429324852047184,
-      "grad_norm": 0.0024214282166212797,
-      "learning_rate": 0.000199990326318806,
-      "loss": 46.0,
-      "step": 27504
-    },
-    {
-      "epoch": 4.429485889125972,
-      "grad_norm": 0.0007645718287676573,
-      "learning_rate": 0.00019999032561511057,
-      "loss": 46.0,
-      "step": 27505
-    },
-    {
-      "epoch": 4.429646926204759,
-      "grad_norm": 0.009464041329920292,
-      "learning_rate": 0.00019999032491138956,
-      "loss": 46.0,
-      "step": 27506
-    },
-    {
-      "epoch": 4.429807963283546,
-      "grad_norm": 0.008248715661466122,
-      "learning_rate": 0.00019999032420764295,
-      "loss": 46.0,
-      "step": 27507
-    },
-    {
-      "epoch": 4.429969000362333,
-      "grad_norm": 0.0009513927507214248,
-      "learning_rate": 0.0001999903235038707,
-      "loss": 46.0,
-      "step": 27508
-    },
-    {
-      "epoch": 4.4301300374411205,
-      "grad_norm": 0.0029127642046660185,
-      "learning_rate": 0.00019999032280007293,
-      "loss": 46.0,
-      "step": 27509
-    },
-    {
-      "epoch": 4.430291074519908,
-      "grad_norm": 0.015009592287242413,
-      "learning_rate": 0.00019999032209624953,
-      "loss": 46.0,
-      "step": 27510
-    },
-    {
-      "epoch": 4.4304521115986955,
-      "grad_norm": 0.003507836489006877,
-      "learning_rate": 0.00019999032139240053,
-      "loss": 46.0,
-      "step": 27511
-    },
-    {
-      "epoch": 4.430613148677483,
-      "grad_norm": 0.009676394052803516,
-      "learning_rate": 0.00019999032068852593,
-      "loss": 46.0,
-      "step": 27512
-    },
-    {
-      "epoch": 4.43077418575627,
-      "grad_norm": 0.0035322660114616156,
-      "learning_rate": 0.00019999031998462575,
-      "loss": 46.0,
-      "step": 27513
-    },
-    {
-      "epoch": 4.430935222835058,
-      "grad_norm": 0.005746645852923393,
-      "learning_rate": 0.0001999903192807,
-      "loss": 46.0,
-      "step": 27514
-    },
-    {
-      "epoch": 4.431096259913845,
-      "grad_norm": 0.0014641284942626953,
-      "learning_rate": 0.00019999031857674865,
-      "loss": 46.0,
-      "step": 27515
-    },
-    {
-      "epoch": 4.431257296992633,
-      "grad_norm": 0.003112539416179061,
-      "learning_rate": 0.00019999031787277168,
-      "loss": 46.0,
-      "step": 27516
-    },
-    {
-      "epoch": 4.43141833407142,
-      "grad_norm": 0.005320766009390354,
-      "learning_rate": 0.00019999031716876915,
-      "loss": 46.0,
-      "step": 27517
-    },
-    {
-      "epoch": 4.431579371150208,
-      "grad_norm": 0.0018871987704187632,
-      "learning_rate": 0.000199990316464741,
-      "loss": 46.0,
-      "step": 27518
-    },
-    {
-      "epoch": 4.431740408228995,
-      "grad_norm": 0.013681559823453426,
-      "learning_rate": 0.0001999903157606873,
-      "loss": 46.0,
-      "step": 27519
-    },
-    {
-      "epoch": 4.4319014453077825,
-      "grad_norm": 0.0063905734568834305,
-      "learning_rate": 0.00019999031505660796,
-      "loss": 46.0,
-      "step": 27520
-    },
-    {
-      "epoch": 4.43206248238657,
-      "grad_norm": 0.003211074275895953,
-      "learning_rate": 0.00019999031435250305,
-      "loss": 46.0,
-      "step": 27521
-    },
-    {
-      "epoch": 4.4322235194653565,
-      "grad_norm": 0.0175580233335495,
-      "learning_rate": 0.00019999031364837256,
-      "loss": 46.0,
-      "step": 27522
-    },
-    {
-      "epoch": 4.432384556544144,
-      "grad_norm": 0.001359413145110011,
-      "learning_rate": 0.00019999031294421645,
-      "loss": 46.0,
-      "step": 27523
-    },
-    {
-      "epoch": 4.432545593622931,
-      "grad_norm": 0.0013794455444440246,
-      "learning_rate": 0.00019999031224003476,
-      "loss": 46.0,
-      "step": 27524
-    },
-    {
-      "epoch": 4.432706630701719,
-      "grad_norm": 0.00516211474314332,
-      "learning_rate": 0.00019999031153582748,
-      "loss": 46.0,
-      "step": 27525
-    },
-    {
-      "epoch": 4.432867667780506,
-      "grad_norm": 0.004926779307425022,
-      "learning_rate": 0.0001999903108315946,
-      "loss": 46.0,
-      "step": 27526
-    },
-    {
-      "epoch": 4.433028704859294,
-      "grad_norm": 0.003878003219142556,
-      "learning_rate": 0.00019999031012733615,
-      "loss": 46.0,
-      "step": 27527
-    },
-    {
-      "epoch": 4.433189741938081,
-      "grad_norm": 0.004260852932929993,
-      "learning_rate": 0.0001999903094230521,
-      "loss": 46.0,
-      "step": 27528
-    },
-    {
-      "epoch": 4.433350779016869,
-      "grad_norm": 0.001436437712982297,
-      "learning_rate": 0.00019999030871874243,
-      "loss": 46.0,
-      "step": 27529
-    },
-    {
-      "epoch": 4.433511816095656,
-      "grad_norm": 0.005477797240018845,
-      "learning_rate": 0.00019999030801440718,
-      "loss": 46.0,
-      "step": 27530
-    },
-    {
-      "epoch": 4.433672853174444,
-      "grad_norm": 0.006780264433473349,
-      "learning_rate": 0.00019999030731004635,
-      "loss": 46.0,
-      "step": 27531
-    },
-    {
-      "epoch": 4.433833890253231,
-      "grad_norm": 0.0016470797127112746,
-      "learning_rate": 0.00019999030660565993,
-      "loss": 46.0,
-      "step": 27532
-    },
-    {
-      "epoch": 4.4339949273320185,
-      "grad_norm": 0.0032933943439275026,
-      "learning_rate": 0.00019999030590124792,
-      "loss": 46.0,
-      "step": 27533
-    },
-    {
-      "epoch": 4.434155964410806,
-      "grad_norm": 0.008356256410479546,
-      "learning_rate": 0.0001999903051968103,
-      "loss": 46.0,
-      "step": 27534
-    },
-    {
-      "epoch": 4.4343170014895925,
-      "grad_norm": 0.0032204226590692997,
-      "learning_rate": 0.00019999030449234707,
-      "loss": 46.0,
-      "step": 27535
-    },
-    {
-      "epoch": 4.43447803856838,
-      "grad_norm": 0.007282535079866648,
-      "learning_rate": 0.00019999030378785827,
-      "loss": 46.0,
-      "step": 27536
-    },
-    {
-      "epoch": 4.434639075647167,
-      "grad_norm": 0.002672413596883416,
-      "learning_rate": 0.0001999903030833439,
-      "loss": 46.0,
-      "step": 27537
-    },
-    {
-      "epoch": 4.434800112725955,
-      "grad_norm": 0.008341788314282894,
-      "learning_rate": 0.00019999030237880392,
-      "loss": 46.0,
-      "step": 27538
-    },
-    {
-      "epoch": 4.434961149804742,
-      "grad_norm": 0.005341968033462763,
-      "learning_rate": 0.00019999030167423834,
-      "loss": 46.0,
-      "step": 27539
-    },
-    {
-      "epoch": 4.43512218688353,
-      "grad_norm": 0.002406828571110964,
-      "learning_rate": 0.00019999030096964717,
-      "loss": 46.0,
-      "step": 27540
-    },
-    {
-      "epoch": 4.435283223962317,
-      "grad_norm": 0.004608490038663149,
-      "learning_rate": 0.0001999903002650304,
-      "loss": 46.0,
-      "step": 27541
-    },
-    {
-      "epoch": 4.435444261041105,
-      "grad_norm": 0.003845945931971073,
-      "learning_rate": 0.00019999029956038807,
-      "loss": 46.0,
-      "step": 27542
-    },
-    {
-      "epoch": 4.435605298119892,
-      "grad_norm": 0.011631255969405174,
-      "learning_rate": 0.0001999902988557201,
-      "loss": 46.0,
-      "step": 27543
-    },
-    {
-      "epoch": 4.43576633519868,
-      "grad_norm": 0.018674982711672783,
-      "learning_rate": 0.00019999029815102656,
-      "loss": 46.0,
-      "step": 27544
-    },
-    {
-      "epoch": 4.435927372277467,
-      "grad_norm": 0.0025312798097729683,
-      "learning_rate": 0.00019999029744630743,
-      "loss": 46.0,
-      "step": 27545
-    },
-    {
-      "epoch": 4.4360884093562545,
-      "grad_norm": 0.003297422779724002,
-      "learning_rate": 0.00019999029674156273,
-      "loss": 46.0,
-      "step": 27546
-    },
-    {
-      "epoch": 4.436249446435042,
-      "grad_norm": 0.015062783844769001,
-      "learning_rate": 0.0001999902960367924,
-      "loss": 46.0,
-      "step": 27547
-    },
-    {
-      "epoch": 4.436410483513829,
-      "grad_norm": 0.004186402540653944,
-      "learning_rate": 0.0001999902953319965,
-      "loss": 46.0,
-      "step": 27548
-    },
-    {
-      "epoch": 4.436571520592617,
-      "grad_norm": 0.005380000453442335,
-      "learning_rate": 0.000199990294627175,
-      "loss": 46.0,
-      "step": 27549
-    },
-    {
-      "epoch": 4.436732557671403,
-      "grad_norm": 0.0017875123303383589,
-      "learning_rate": 0.0001999902939223279,
-      "loss": 46.0,
-      "step": 27550
-    },
-    {
-      "epoch": 4.436893594750191,
-      "grad_norm": 0.002819065237417817,
-      "learning_rate": 0.00019999029321745524,
-      "loss": 46.0,
-      "step": 27551
-    },
-    {
-      "epoch": 4.437054631828978,
-      "grad_norm": 0.004958902485668659,
-      "learning_rate": 0.00019999029251255697,
-      "loss": 46.0,
-      "step": 27552
-    },
-    {
-      "epoch": 4.437215668907766,
-      "grad_norm": 0.008189218118786812,
-      "learning_rate": 0.00019999029180763306,
-      "loss": 46.0,
-      "step": 27553
-    },
-    {
-      "epoch": 4.437376705986553,
-      "grad_norm": 0.003609184641391039,
-      "learning_rate": 0.00019999029110268364,
-      "loss": 46.0,
-      "step": 27554
-    },
-    {
-      "epoch": 4.437537743065341,
-      "grad_norm": 0.0032892327290028334,
-      "learning_rate": 0.00019999029039770858,
-      "loss": 46.0,
-      "step": 27555
-    },
-    {
-      "epoch": 4.437698780144128,
-      "grad_norm": 0.0021846629679203033,
-      "learning_rate": 0.00019999028969270793,
-      "loss": 46.0,
-      "step": 27556
-    },
-    {
-      "epoch": 4.437859817222916,
-      "grad_norm": 0.0025701646227389574,
-      "learning_rate": 0.00019999028898768167,
-      "loss": 46.0,
-      "step": 27557
-    },
-    {
-      "epoch": 4.438020854301703,
-      "grad_norm": 0.0029579834081232548,
-      "learning_rate": 0.00019999028828262985,
-      "loss": 46.0,
-      "step": 27558
-    },
-    {
-      "epoch": 4.4381818913804905,
-      "grad_norm": 0.0034569271374493837,
-      "learning_rate": 0.00019999028757755242,
-      "loss": 46.0,
-      "step": 27559
-    },
-    {
-      "epoch": 4.438342928459278,
-      "grad_norm": 0.00441031064838171,
-      "learning_rate": 0.00019999028687244942,
-      "loss": 46.0,
-      "step": 27560
-    },
-    {
-      "epoch": 4.438503965538065,
-      "grad_norm": 0.003183339722454548,
-      "learning_rate": 0.0001999902861673208,
-      "loss": 46.0,
-      "step": 27561
-    },
-    {
-      "epoch": 4.438665002616853,
-      "grad_norm": 0.015945790335536003,
-      "learning_rate": 0.00019999028546216661,
-      "loss": 46.0,
-      "step": 27562
-    },
-    {
-      "epoch": 4.43882603969564,
-      "grad_norm": 0.0013115203473716974,
-      "learning_rate": 0.00019999028475698683,
-      "loss": 46.0,
-      "step": 27563
-    },
-    {
-      "epoch": 4.438987076774428,
-      "grad_norm": 0.0027518514543771744,
-      "learning_rate": 0.0001999902840517814,
-      "loss": 46.0,
-      "step": 27564
-    },
-    {
-      "epoch": 4.439148113853214,
-      "grad_norm": 0.010220200754702091,
-      "learning_rate": 0.00019999028334655042,
-      "loss": 46.0,
-      "step": 27565
-    },
-    {
-      "epoch": 4.439309150932002,
-      "grad_norm": 0.0070998347364366055,
-      "learning_rate": 0.00019999028264129387,
-      "loss": 46.0,
-      "step": 27566
-    },
-    {
-      "epoch": 4.439470188010789,
-      "grad_norm": 0.0011235951678827405,
-      "learning_rate": 0.00019999028193601172,
-      "loss": 46.0,
-      "step": 27567
-    },
-    {
-      "epoch": 4.439631225089577,
-      "grad_norm": 0.009345768950879574,
-      "learning_rate": 0.00019999028123070397,
-      "loss": 46.0,
-      "step": 27568
-    },
-    {
-      "epoch": 4.439792262168364,
-      "grad_norm": 0.004411500412970781,
-      "learning_rate": 0.0001999902805253706,
-      "loss": 46.0,
-      "step": 27569
-    },
-    {
-      "epoch": 4.4399532992471515,
-      "grad_norm": 0.009427009150385857,
-      "learning_rate": 0.00019999027982001169,
-      "loss": 46.0,
-      "step": 27570
-    },
-    {
-      "epoch": 4.440114336325939,
-      "grad_norm": 0.006120585836470127,
-      "learning_rate": 0.00019999027911462712,
-      "loss": 46.0,
-      "step": 27571
-    },
-    {
-      "epoch": 4.4402753734047264,
-      "grad_norm": 0.002826493699103594,
-      "learning_rate": 0.00019999027840921703,
-      "loss": 46.0,
-      "step": 27572
-    },
-    {
-      "epoch": 4.440436410483514,
-      "grad_norm": 0.005567345768213272,
-      "learning_rate": 0.0001999902777037813,
-      "loss": 46.0,
-      "step": 27573
-    },
-    {
-      "epoch": 4.440597447562301,
-      "grad_norm": 0.004630262032151222,
-      "learning_rate": 0.00019999027699832,
-      "loss": 46.0,
-      "step": 27574
-    },
-    {
-      "epoch": 4.440758484641089,
-      "grad_norm": 0.00944850966334343,
-      "learning_rate": 0.00019999027629283308,
-      "loss": 46.0,
-      "step": 27575
-    },
-    {
-      "epoch": 4.440919521719876,
-      "grad_norm": 0.0021597445011138916,
-      "learning_rate": 0.00019999027558732058,
-      "loss": 46.0,
-      "step": 27576
-    },
-    {
-      "epoch": 4.441080558798664,
-      "grad_norm": 0.0020268631633371115,
-      "learning_rate": 0.00019999027488178253,
-      "loss": 46.0,
-      "step": 27577
-    },
-    {
-      "epoch": 4.441241595877451,
-      "grad_norm": 0.0027508866041898727,
-      "learning_rate": 0.00019999027417621883,
-      "loss": 46.0,
-      "step": 27578
-    },
-    {
-      "epoch": 4.441402632956239,
-      "grad_norm": 0.0073135183192789555,
-      "learning_rate": 0.00019999027347062957,
-      "loss": 46.0,
-      "step": 27579
-    },
-    {
-      "epoch": 4.441563670035025,
-      "grad_norm": 0.008463002741336823,
-      "learning_rate": 0.0001999902727650147,
-      "loss": 46.0,
-      "step": 27580
-    },
-    {
-      "epoch": 4.441724707113813,
-      "grad_norm": 0.004070807248353958,
-      "learning_rate": 0.00019999027205937426,
-      "loss": 46.0,
-      "step": 27581
-    },
-    {
-      "epoch": 4.4418857441926,
-      "grad_norm": 0.0016556144692003727,
-      "learning_rate": 0.0001999902713537082,
-      "loss": 46.0,
-      "step": 27582
-    },
-    {
-      "epoch": 4.4420467812713875,
-      "grad_norm": 0.003913614433258772,
-      "learning_rate": 0.00019999027064801657,
-      "loss": 46.0,
-      "step": 27583
-    },
-    {
-      "epoch": 4.442207818350175,
-      "grad_norm": 0.0021010919008404016,
-      "learning_rate": 0.00019999026994229932,
-      "loss": 46.0,
-      "step": 27584
-    },
-    {
-      "epoch": 4.442368855428962,
-      "grad_norm": 0.0022471530828624964,
-      "learning_rate": 0.0001999902692365565,
-      "loss": 46.0,
-      "step": 27585
-    },
-    {
-      "epoch": 4.44252989250775,
-      "grad_norm": 0.0034856132697314024,
-      "learning_rate": 0.0001999902685307881,
-      "loss": 46.0,
-      "step": 27586
-    },
-    {
-      "epoch": 4.442690929586537,
-      "grad_norm": 0.015602142550051212,
-      "learning_rate": 0.00019999026782499408,
-      "loss": 46.0,
-      "step": 27587
-    },
-    {
-      "epoch": 4.442851966665325,
-      "grad_norm": 0.006289406679570675,
-      "learning_rate": 0.00019999026711917445,
-      "loss": 46.0,
-      "step": 27588
-    },
-    {
-      "epoch": 4.443013003744112,
-      "grad_norm": 0.013078049756586552,
-      "learning_rate": 0.00019999026641332926,
-      "loss": 46.0,
-      "step": 27589
-    },
-    {
-      "epoch": 4.4431740408229,
-      "grad_norm": 0.0013088486157357693,
-      "learning_rate": 0.0001999902657074585,
-      "loss": 46.0,
-      "step": 27590
-    },
-    {
-      "epoch": 4.443335077901687,
-      "grad_norm": 0.003772629890590906,
-      "learning_rate": 0.0001999902650015621,
-      "loss": 46.0,
-      "step": 27591
-    },
-    {
-      "epoch": 4.443496114980475,
-      "grad_norm": 0.02117474004626274,
-      "learning_rate": 0.00019999026429564013,
-      "loss": 46.0,
-      "step": 27592
-    },
-    {
-      "epoch": 4.443657152059262,
-      "grad_norm": 0.004630803130567074,
-      "learning_rate": 0.00019999026358969257,
-      "loss": 46.0,
-      "step": 27593
-    },
-    {
-      "epoch": 4.4438181891380495,
-      "grad_norm": 0.006949576549232006,
-      "learning_rate": 0.0001999902628837194,
-      "loss": 46.0,
-      "step": 27594
-    },
-    {
-      "epoch": 4.443979226216836,
-      "grad_norm": 0.008330770768225193,
-      "learning_rate": 0.00019999026217772068,
-      "loss": 46.0,
-      "step": 27595
-    },
-    {
-      "epoch": 4.4441402632956235,
-      "grad_norm": 0.015360256657004356,
-      "learning_rate": 0.00019999026147169633,
-      "loss": 46.0,
-      "step": 27596
-    },
-    {
-      "epoch": 4.444301300374411,
-      "grad_norm": 0.00623265840113163,
-      "learning_rate": 0.0001999902607656464,
-      "loss": 46.0,
-      "step": 27597
-    },
-    {
-      "epoch": 4.444462337453198,
-      "grad_norm": 0.005557015538215637,
-      "learning_rate": 0.00019999026005957087,
-      "loss": 46.0,
-      "step": 27598
-    },
-    {
-      "epoch": 4.444623374531986,
-      "grad_norm": 0.0055144005455076694,
-      "learning_rate": 0.00019999025935346976,
-      "loss": 46.0,
-      "step": 27599
-    },
-    {
-      "epoch": 4.444784411610773,
-      "grad_norm": 0.00569901941344142,
-      "learning_rate": 0.00019999025864734303,
-      "loss": 46.0,
-      "step": 27600
-    },
-    {
-      "epoch": 4.444945448689561,
-      "grad_norm": 0.00342980376444757,
-      "learning_rate": 0.00019999025794119075,
-      "loss": 46.0,
-      "step": 27601
-    },
-    {
-      "epoch": 4.445106485768348,
-      "grad_norm": 0.000592138385400176,
-      "learning_rate": 0.00019999025723501285,
-      "loss": 46.0,
-      "step": 27602
-    },
-    {
-      "epoch": 4.445267522847136,
-      "grad_norm": 0.004448787309229374,
-      "learning_rate": 0.00019999025652880933,
-      "loss": 46.0,
-      "step": 27603
-    },
-    {
-      "epoch": 4.445428559925923,
-      "grad_norm": 0.004147749859839678,
-      "learning_rate": 0.00019999025582258026,
-      "loss": 46.0,
-      "step": 27604
-    },
-    {
-      "epoch": 4.445589597004711,
-      "grad_norm": 0.004227131139487028,
-      "learning_rate": 0.00019999025511632562,
-      "loss": 46.0,
-      "step": 27605
-    },
-    {
-      "epoch": 4.445750634083498,
-      "grad_norm": 0.006812307517975569,
-      "learning_rate": 0.00019999025441004532,
-      "loss": 46.0,
-      "step": 27606
-    },
-    {
-      "epoch": 4.4459116711622855,
-      "grad_norm": 0.005776179488748312,
-      "learning_rate": 0.00019999025370373948,
-      "loss": 46.0,
-      "step": 27607
-    },
-    {
-      "epoch": 4.446072708241072,
-      "grad_norm": 0.012053025886416435,
-      "learning_rate": 0.00019999025299740803,
-      "loss": 46.0,
-      "step": 27608
-    },
-    {
-      "epoch": 4.4462337453198595,
-      "grad_norm": 0.0008147385669872165,
-      "learning_rate": 0.00019999025229105097,
-      "loss": 46.0,
-      "step": 27609
-    },
-    {
-      "epoch": 4.446394782398647,
-      "grad_norm": 0.008392365649342537,
-      "learning_rate": 0.00019999025158466834,
-      "loss": 46.0,
-      "step": 27610
-    },
-    {
-      "epoch": 4.446555819477434,
-      "grad_norm": 0.001987253315746784,
-      "learning_rate": 0.0001999902508782601,
-      "loss": 46.0,
-      "step": 27611
-    },
-    {
-      "epoch": 4.446716856556222,
-      "grad_norm": 0.006051927804946899,
-      "learning_rate": 0.0001999902501718263,
-      "loss": 46.0,
-      "step": 27612
-    },
-    {
-      "epoch": 4.446877893635009,
-      "grad_norm": 0.004000207409262657,
-      "learning_rate": 0.0001999902494653669,
-      "loss": 46.0,
-      "step": 27613
-    },
-    {
-      "epoch": 4.447038930713797,
-      "grad_norm": 0.008097521029412746,
-      "learning_rate": 0.0001999902487588819,
-      "loss": 46.0,
-      "step": 27614
-    },
-    {
-      "epoch": 4.447199967792584,
-      "grad_norm": 0.00424214918166399,
-      "learning_rate": 0.0001999902480523713,
-      "loss": 46.0,
-      "step": 27615
-    },
-    {
-      "epoch": 4.447361004871372,
-      "grad_norm": 0.006864896509796381,
-      "learning_rate": 0.0001999902473458351,
-      "loss": 46.0,
-      "step": 27616
-    },
-    {
-      "epoch": 4.447522041950159,
-      "grad_norm": 0.0033597571309655905,
-      "learning_rate": 0.00019999024663927334,
-      "loss": 46.0,
-      "step": 27617
-    },
-    {
-      "epoch": 4.4476830790289466,
-      "grad_norm": 0.0025569158606231213,
-      "learning_rate": 0.00019999024593268593,
-      "loss": 46.0,
-      "step": 27618
-    },
-    {
-      "epoch": 4.447844116107734,
-      "grad_norm": 0.002775526838377118,
-      "learning_rate": 0.000199990245226073,
-      "loss": 46.0,
-      "step": 27619
-    },
-    {
-      "epoch": 4.4480051531865215,
-      "grad_norm": 0.0027869076002389193,
-      "learning_rate": 0.00019999024451943442,
-      "loss": 46.0,
-      "step": 27620
-    },
-    {
-      "epoch": 4.448166190265309,
-      "grad_norm": 0.0014000433729961514,
-      "learning_rate": 0.00019999024381277028,
-      "loss": 46.0,
-      "step": 27621
-    },
-    {
-      "epoch": 4.448327227344096,
-      "grad_norm": 0.004964140709489584,
-      "learning_rate": 0.00019999024310608056,
-      "loss": 46.0,
-      "step": 27622
-    },
-    {
-      "epoch": 4.448488264422883,
-      "grad_norm": 0.001712570432573557,
-      "learning_rate": 0.0001999902423993652,
-      "loss": 46.0,
-      "step": 27623
-    },
-    {
-      "epoch": 4.44864930150167,
-      "grad_norm": 0.0012439304264262319,
-      "learning_rate": 0.00019999024169262426,
-      "loss": 46.0,
-      "step": 27624
-    },
-    {
-      "epoch": 4.448810338580458,
-      "grad_norm": 0.002051472896710038,
-      "learning_rate": 0.00019999024098585775,
-      "loss": 46.0,
-      "step": 27625
-    },
-    {
-      "epoch": 4.448971375659245,
-      "grad_norm": 0.004451302345842123,
-      "learning_rate": 0.00019999024027906565,
-      "loss": 46.0,
-      "step": 27626
-    },
-    {
-      "epoch": 4.449132412738033,
-      "grad_norm": 0.005217827390879393,
-      "learning_rate": 0.00019999023957224793,
-      "loss": 46.0,
-      "step": 27627
-    },
-    {
-      "epoch": 4.44929344981682,
-      "grad_norm": 0.002307931426912546,
-      "learning_rate": 0.00019999023886540466,
-      "loss": 46.0,
-      "step": 27628
-    },
-    {
-      "epoch": 4.449454486895608,
-      "grad_norm": 0.008777051232755184,
-      "learning_rate": 0.00019999023815853574,
-      "loss": 46.0,
-      "step": 27629
-    },
-    {
-      "epoch": 4.449615523974395,
-      "grad_norm": 0.009207247756421566,
-      "learning_rate": 0.00019999023745164126,
-      "loss": 46.0,
-      "step": 27630
-    },
-    {
-      "epoch": 4.4497765610531825,
-      "grad_norm": 0.01339022722095251,
-      "learning_rate": 0.0001999902367447212,
-      "loss": 46.0,
-      "step": 27631
-    },
-    {
-      "epoch": 4.44993759813197,
-      "grad_norm": 0.007821612060070038,
-      "learning_rate": 0.00019999023603777555,
-      "loss": 46.0,
-      "step": 27632
-    },
-    {
-      "epoch": 4.450098635210757,
-      "grad_norm": 0.005127355922013521,
-      "learning_rate": 0.00019999023533080426,
-      "loss": 46.0,
-      "step": 27633
-    },
-    {
-      "epoch": 4.450259672289545,
-      "grad_norm": 0.003538827644661069,
-      "learning_rate": 0.0001999902346238074,
-      "loss": 46.0,
-      "step": 27634
-    },
-    {
-      "epoch": 4.450420709368332,
-      "grad_norm": 0.0018289581639692187,
-      "learning_rate": 0.00019999023391678496,
-      "loss": 46.0,
-      "step": 27635
-    },
-    {
-      "epoch": 4.45058174644712,
-      "grad_norm": 0.0012734842021018267,
-      "learning_rate": 0.00019999023320973694,
-      "loss": 46.0,
-      "step": 27636
-    },
-    {
-      "epoch": 4.450742783525907,
-      "grad_norm": 0.004097527824342251,
-      "learning_rate": 0.00019999023250266332,
-      "loss": 46.0,
-      "step": 27637
-    },
-    {
-      "epoch": 4.450903820604694,
-      "grad_norm": 0.0016025390941649675,
-      "learning_rate": 0.0001999902317955641,
-      "loss": 46.0,
-      "step": 27638
-    },
-    {
-      "epoch": 4.451064857683481,
-      "grad_norm": 0.00741965277120471,
-      "learning_rate": 0.00019999023108843925,
-      "loss": 46.0,
-      "step": 27639
-    },
-    {
-      "epoch": 4.451225894762269,
-      "grad_norm": 0.0016680487897247076,
-      "learning_rate": 0.00019999023038128888,
-      "loss": 46.0,
-      "step": 27640
-    },
-    {
-      "epoch": 4.451386931841056,
-      "grad_norm": 0.0036669380497187376,
-      "learning_rate": 0.00019999022967411286,
-      "loss": 46.0,
-      "step": 27641
-    },
-    {
-      "epoch": 4.451547968919844,
-      "grad_norm": 0.0036178594455122948,
-      "learning_rate": 0.00019999022896691128,
-      "loss": 46.0,
-      "step": 27642
-    },
-    {
-      "epoch": 4.451709005998631,
-      "grad_norm": 0.0038552344776690006,
-      "learning_rate": 0.0001999902282596841,
-      "loss": 46.0,
-      "step": 27643
-    },
-    {
-      "epoch": 4.4518700430774185,
-      "grad_norm": 0.004150839056819677,
-      "learning_rate": 0.00019999022755243134,
-      "loss": 46.0,
-      "step": 27644
-    },
-    {
-      "epoch": 4.452031080156206,
-      "grad_norm": 0.009294205345213413,
-      "learning_rate": 0.00019999022684515297,
-      "loss": 46.0,
-      "step": 27645
-    },
-    {
-      "epoch": 4.452192117234993,
-      "grad_norm": 0.010999877005815506,
-      "learning_rate": 0.000199990226137849,
-      "loss": 46.0,
-      "step": 27646
-    },
-    {
-      "epoch": 4.452353154313781,
-      "grad_norm": 0.0011699118185788393,
-      "learning_rate": 0.00019999022543051943,
-      "loss": 46.0,
-      "step": 27647
-    },
-    {
-      "epoch": 4.452514191392568,
-      "grad_norm": 0.004467427730560303,
-      "learning_rate": 0.0001999902247231643,
-      "loss": 46.0,
-      "step": 27648
-    },
-    {
-      "epoch": 4.452675228471356,
-      "grad_norm": 0.004868346732109785,
-      "learning_rate": 0.00019999022401578356,
-      "loss": 46.0,
-      "step": 27649
-    },
-    {
-      "epoch": 4.452836265550143,
-      "grad_norm": 0.0024809425231069326,
-      "learning_rate": 0.00019999022330837726,
-      "loss": 46.0,
-      "step": 27650
-    },
-    {
-      "epoch": 4.452997302628931,
-      "grad_norm": 0.005295577459037304,
-      "learning_rate": 0.00019999022260094531,
-      "loss": 46.0,
-      "step": 27651
-    },
-    {
-      "epoch": 4.453158339707718,
-      "grad_norm": 0.004362618084996939,
-      "learning_rate": 0.00019999022189348778,
-      "loss": 46.0,
-      "step": 27652
-    },
-    {
-      "epoch": 4.453319376786505,
-      "grad_norm": 0.0030936384573578835,
-      "learning_rate": 0.0001999902211860047,
-      "loss": 46.0,
-      "step": 27653
-    },
-    {
-      "epoch": 4.453480413865292,
-      "grad_norm": 0.0027114993426948786,
-      "learning_rate": 0.000199990220478496,
-      "loss": 46.0,
-      "step": 27654
-    },
-    {
-      "epoch": 4.45364145094408,
-      "grad_norm": 0.020321976393461227,
-      "learning_rate": 0.0001999902197709617,
-      "loss": 46.0,
-      "step": 27655
-    },
-    {
-      "epoch": 4.453802488022867,
-      "grad_norm": 0.004300059285014868,
-      "learning_rate": 0.00019999021906340182,
-      "loss": 46.0,
-      "step": 27656
-    },
-    {
-      "epoch": 4.4539635251016545,
-      "grad_norm": 0.007416091859340668,
-      "learning_rate": 0.00019999021835581635,
-      "loss": 46.0,
-      "step": 27657
-    },
-    {
-      "epoch": 4.454124562180442,
-      "grad_norm": 0.012945530004799366,
-      "learning_rate": 0.00019999021764820527,
-      "loss": 46.0,
-      "step": 27658
-    },
-    {
-      "epoch": 4.454285599259229,
-      "grad_norm": 0.004318093881011009,
-      "learning_rate": 0.0001999902169405686,
-      "loss": 46.0,
-      "step": 27659
-    },
-    {
-      "epoch": 4.454446636338017,
-      "grad_norm": 0.003392336890101433,
-      "learning_rate": 0.00019999021623290638,
-      "loss": 46.0,
-      "step": 27660
-    },
-    {
-      "epoch": 4.454607673416804,
-      "grad_norm": 0.0029355748556554317,
-      "learning_rate": 0.0001999902155252185,
-      "loss": 46.0,
-      "step": 27661
-    },
-    {
-      "epoch": 4.454768710495592,
-      "grad_norm": 0.004893424920737743,
-      "learning_rate": 0.00019999021481750505,
-      "loss": 46.0,
-      "step": 27662
-    },
-    {
-      "epoch": 4.454929747574379,
-      "grad_norm": 0.014779835939407349,
-      "learning_rate": 0.00019999021410976604,
-      "loss": 46.0,
-      "step": 27663
-    },
-    {
-      "epoch": 4.455090784653167,
-      "grad_norm": 0.004990631714463234,
-      "learning_rate": 0.00019999021340200143,
-      "loss": 46.0,
-      "step": 27664
-    },
-    {
-      "epoch": 4.455251821731954,
-      "grad_norm": 0.0011120865819975734,
-      "learning_rate": 0.00019999021269421122,
-      "loss": 46.0,
-      "step": 27665
-    },
-    {
-      "epoch": 4.455412858810742,
-      "grad_norm": 0.0024424034636467695,
-      "learning_rate": 0.0001999902119863954,
-      "loss": 46.0,
-      "step": 27666
-    },
-    {
-      "epoch": 4.455573895889529,
-      "grad_norm": 0.005659935064613819,
-      "learning_rate": 0.00019999021127855402,
-      "loss": 46.0,
-      "step": 27667
-    },
-    {
-      "epoch": 4.455734932968316,
-      "grad_norm": 0.004742661956697702,
-      "learning_rate": 0.00019999021057068701,
-      "loss": 46.0,
-      "step": 27668
-    },
-    {
-      "epoch": 4.455895970047103,
-      "grad_norm": 0.001482888008467853,
-      "learning_rate": 0.00019999020986279442,
-      "loss": 46.0,
-      "step": 27669
-    },
-    {
-      "epoch": 4.4560570071258905,
-      "grad_norm": 0.0017236744752153754,
-      "learning_rate": 0.00019999020915487624,
-      "loss": 46.0,
-      "step": 27670
-    },
-    {
-      "epoch": 4.456218044204678,
-      "grad_norm": 0.004353113938122988,
-      "learning_rate": 0.00019999020844693247,
-      "loss": 46.0,
-      "step": 27671
-    },
-    {
-      "epoch": 4.456379081283465,
-      "grad_norm": 0.005677455570548773,
-      "learning_rate": 0.00019999020773896312,
-      "loss": 46.0,
-      "step": 27672
-    },
-    {
-      "epoch": 4.456540118362253,
-      "grad_norm": 0.0056228190660476685,
-      "learning_rate": 0.00019999020703096818,
-      "loss": 46.0,
-      "step": 27673
-    },
-    {
-      "epoch": 4.45670115544104,
-      "grad_norm": 0.00876994151622057,
-      "learning_rate": 0.0001999902063229476,
-      "loss": 46.0,
-      "step": 27674
-    },
-    {
-      "epoch": 4.456862192519828,
-      "grad_norm": 0.0034961539786309004,
-      "learning_rate": 0.0001999902056149015,
-      "loss": 46.0,
-      "step": 27675
-    },
-    {
-      "epoch": 4.457023229598615,
-      "grad_norm": 0.005490646697580814,
-      "learning_rate": 0.00019999020490682975,
-      "loss": 46.0,
-      "step": 27676
-    },
-    {
-      "epoch": 4.457184266677403,
-      "grad_norm": 0.015552288852632046,
-      "learning_rate": 0.00019999020419873243,
-      "loss": 46.0,
-      "step": 27677
-    },
-    {
-      "epoch": 4.45734530375619,
-      "grad_norm": 0.004894928075373173,
-      "learning_rate": 0.0001999902034906095,
-      "loss": 46.0,
-      "step": 27678
-    },
-    {
-      "epoch": 4.4575063408349775,
-      "grad_norm": 0.002772238804027438,
-      "learning_rate": 0.00019999020278246098,
-      "loss": 46.0,
-      "step": 27679
-    },
-    {
-      "epoch": 4.457667377913765,
-      "grad_norm": 0.003138558939099312,
-      "learning_rate": 0.0001999902020742869,
-      "loss": 46.0,
-      "step": 27680
-    },
-    {
-      "epoch": 4.4578284149925524,
-      "grad_norm": 0.008120818994939327,
-      "learning_rate": 0.0001999902013660872,
-      "loss": 46.0,
-      "step": 27681
-    },
-    {
-      "epoch": 4.45798945207134,
-      "grad_norm": 0.002648937748745084,
-      "learning_rate": 0.00019999020065786192,
-      "loss": 46.0,
-      "step": 27682
-    },
-    {
-      "epoch": 4.4581504891501265,
-      "grad_norm": 0.004655967932194471,
-      "learning_rate": 0.00019999019994961103,
-      "loss": 46.0,
-      "step": 27683
-    },
-    {
-      "epoch": 4.458311526228914,
-      "grad_norm": 0.008802182972431183,
-      "learning_rate": 0.00019999019924133457,
-      "loss": 46.0,
-      "step": 27684
-    },
-    {
-      "epoch": 4.458472563307701,
-      "grad_norm": 0.009249554947018623,
-      "learning_rate": 0.0001999901985330325,
-      "loss": 46.0,
-      "step": 27685
-    },
-    {
-      "epoch": 4.458633600386489,
-      "grad_norm": 0.003689124481752515,
-      "learning_rate": 0.00019999019782470487,
-      "loss": 46.0,
-      "step": 27686
-    },
-    {
-      "epoch": 4.458794637465276,
-      "grad_norm": 0.002156637143343687,
-      "learning_rate": 0.00019999019711635163,
-      "loss": 46.0,
-      "step": 27687
-    },
-    {
-      "epoch": 4.458955674544064,
-      "grad_norm": 0.0007555747870355844,
-      "learning_rate": 0.00019999019640797277,
-      "loss": 46.0,
-      "step": 27688
-    },
-    {
-      "epoch": 4.459116711622851,
-      "grad_norm": 0.0037337748799473047,
-      "learning_rate": 0.00019999019569956835,
-      "loss": 46.0,
-      "step": 27689
-    },
-    {
-      "epoch": 4.459277748701639,
-      "grad_norm": 0.008690381422638893,
-      "learning_rate": 0.00019999019499113832,
-      "loss": 46.0,
-      "step": 27690
-    },
-    {
-      "epoch": 4.459438785780426,
-      "grad_norm": 0.002384126652032137,
-      "learning_rate": 0.0001999901942826827,
-      "loss": 46.0,
-      "step": 27691
-    },
-    {
-      "epoch": 4.4595998228592135,
-      "grad_norm": 0.0035090213641524315,
-      "learning_rate": 0.0001999901935742015,
-      "loss": 46.0,
-      "step": 27692
-    },
-    {
-      "epoch": 4.459760859938001,
-      "grad_norm": 0.011557115241885185,
-      "learning_rate": 0.0001999901928656947,
-      "loss": 46.0,
-      "step": 27693
-    },
-    {
-      "epoch": 4.459921897016788,
-      "grad_norm": 0.0022580409422516823,
-      "learning_rate": 0.0001999901921571623,
-      "loss": 46.0,
-      "step": 27694
-    },
-    {
-      "epoch": 4.460082934095576,
-      "grad_norm": 0.003985777031630278,
-      "learning_rate": 0.00019999019144860433,
-      "loss": 46.0,
-      "step": 27695
-    },
-    {
-      "epoch": 4.460243971174362,
-      "grad_norm": 0.01426723413169384,
-      "learning_rate": 0.00019999019074002074,
-      "loss": 46.0,
-      "step": 27696
-    },
-    {
-      "epoch": 4.46040500825315,
-      "grad_norm": 0.010837400332093239,
-      "learning_rate": 0.00019999019003141157,
-      "loss": 46.0,
-      "step": 27697
-    },
-    {
-      "epoch": 4.460566045331937,
-      "grad_norm": 0.025637198239564896,
-      "learning_rate": 0.00019999018932277682,
-      "loss": 46.0,
-      "step": 27698
-    },
-    {
-      "epoch": 4.460727082410725,
-      "grad_norm": 0.0017911669565364718,
-      "learning_rate": 0.00019999018861411645,
-      "loss": 46.0,
-      "step": 27699
-    },
-    {
-      "epoch": 4.460888119489512,
-      "grad_norm": 0.003228267887607217,
-      "learning_rate": 0.0001999901879054305,
-      "loss": 46.0,
-      "step": 27700
-    },
-    {
-      "epoch": 4.4610491565683,
-      "grad_norm": 0.008008738048374653,
-      "learning_rate": 0.00019999018719671897,
-      "loss": 46.0,
-      "step": 27701
-    },
-    {
-      "epoch": 4.461210193647087,
-      "grad_norm": 0.006623091176152229,
-      "learning_rate": 0.00019999018648798184,
-      "loss": 46.0,
-      "step": 27702
-    },
-    {
-      "epoch": 4.461371230725875,
-      "grad_norm": 0.00327286752872169,
-      "learning_rate": 0.00019999018577921912,
-      "loss": 46.0,
-      "step": 27703
-    },
-    {
-      "epoch": 4.461532267804662,
-      "grad_norm": 0.010388505645096302,
-      "learning_rate": 0.00019999018507043078,
-      "loss": 46.0,
-      "step": 27704
-    },
-    {
-      "epoch": 4.4616933048834495,
-      "grad_norm": 0.006412656512111425,
-      "learning_rate": 0.0001999901843616169,
-      "loss": 46.0,
-      "step": 27705
-    },
-    {
-      "epoch": 4.461854341962237,
-      "grad_norm": 0.004264703020453453,
-      "learning_rate": 0.00019999018365277738,
-      "loss": 46.0,
-      "step": 27706
-    },
-    {
-      "epoch": 4.462015379041024,
-      "grad_norm": 0.005853310227394104,
-      "learning_rate": 0.00019999018294391229,
-      "loss": 46.0,
-      "step": 27707
-    },
-    {
-      "epoch": 4.462176416119812,
-      "grad_norm": 0.0006503654294647276,
-      "learning_rate": 0.0001999901822350216,
-      "loss": 46.0,
-      "step": 27708
-    },
-    {
-      "epoch": 4.462337453198599,
-      "grad_norm": 0.0021483106538653374,
-      "learning_rate": 0.00019999018152610533,
-      "loss": 46.0,
-      "step": 27709
-    },
-    {
-      "epoch": 4.462498490277387,
-      "grad_norm": 0.008021563291549683,
-      "learning_rate": 0.00019999018081716345,
-      "loss": 46.0,
-      "step": 27710
-    },
-    {
-      "epoch": 4.462659527356173,
-      "grad_norm": 0.013105835765600204,
-      "learning_rate": 0.00019999018010819598,
-      "loss": 46.0,
-      "step": 27711
-    },
-    {
-      "epoch": 4.462820564434961,
-      "grad_norm": 0.0016825452912598848,
-      "learning_rate": 0.00019999017939920295,
-      "loss": 46.0,
-      "step": 27712
-    },
-    {
-      "epoch": 4.462981601513748,
-      "grad_norm": 0.004741394426673651,
-      "learning_rate": 0.00019999017869018427,
-      "loss": 46.0,
-      "step": 27713
-    },
-    {
-      "epoch": 4.463142638592536,
-      "grad_norm": 0.001727042836137116,
-      "learning_rate": 0.00019999017798114004,
-      "loss": 46.0,
-      "step": 27714
-    },
-    {
-      "epoch": 4.463303675671323,
-      "grad_norm": 0.0026482846587896347,
-      "learning_rate": 0.00019999017727207022,
-      "loss": 46.0,
-      "step": 27715
-    },
-    {
-      "epoch": 4.463464712750111,
-      "grad_norm": 0.010358884930610657,
-      "learning_rate": 0.00019999017656297479,
-      "loss": 46.0,
-      "step": 27716
-    },
-    {
-      "epoch": 4.463625749828898,
-      "grad_norm": 0.002391768153756857,
-      "learning_rate": 0.0001999901758538538,
-      "loss": 46.0,
-      "step": 27717
-    },
-    {
-      "epoch": 4.4637867869076855,
-      "grad_norm": 0.0013181371614336967,
-      "learning_rate": 0.00019999017514470718,
-      "loss": 46.0,
-      "step": 27718
-    },
-    {
-      "epoch": 4.463947823986473,
-      "grad_norm": 0.0069139860570430756,
-      "learning_rate": 0.00019999017443553496,
-      "loss": 46.0,
-      "step": 27719
-    },
-    {
-      "epoch": 4.46410886106526,
-      "grad_norm": 0.004040156025439501,
-      "learning_rate": 0.00019999017372633718,
-      "loss": 46.0,
-      "step": 27720
-    },
-    {
-      "epoch": 4.464269898144048,
-      "grad_norm": 0.005994816310703754,
-      "learning_rate": 0.00019999017301711378,
-      "loss": 46.0,
-      "step": 27721
-    },
-    {
-      "epoch": 4.464430935222835,
-      "grad_norm": 0.010004755109548569,
-      "learning_rate": 0.00019999017230786482,
-      "loss": 46.0,
-      "step": 27722
-    },
-    {
-      "epoch": 4.464591972301623,
-      "grad_norm": 0.002008899115025997,
-      "learning_rate": 0.00019999017159859022,
-      "loss": 46.0,
-      "step": 27723
-    },
-    {
-      "epoch": 4.46475300938041,
-      "grad_norm": 0.0023983328137546778,
-      "learning_rate": 0.00019999017088929007,
-      "loss": 46.0,
-      "step": 27724
-    },
-    {
-      "epoch": 4.464914046459198,
-      "grad_norm": 0.0013118039350956678,
-      "learning_rate": 0.00019999017017996432,
-      "loss": 46.0,
-      "step": 27725
-    },
-    {
-      "epoch": 4.465075083537984,
-      "grad_norm": 0.0072244820185005665,
-      "learning_rate": 0.00019999016947061296,
-      "loss": 46.0,
-      "step": 27726
-    },
-    {
-      "epoch": 4.465236120616772,
-      "grad_norm": 0.001990034943446517,
-      "learning_rate": 0.000199990168761236,
-      "loss": 46.0,
-      "step": 27727
-    },
-    {
-      "epoch": 4.465397157695559,
-      "grad_norm": 0.0032743357587605715,
-      "learning_rate": 0.00019999016805183348,
-      "loss": 46.0,
-      "step": 27728
-    },
-    {
-      "epoch": 4.465558194774347,
-      "grad_norm": 0.0027179557364434004,
-      "learning_rate": 0.00019999016734240536,
-      "loss": 46.0,
-      "step": 27729
-    },
-    {
-      "epoch": 4.465719231853134,
-      "grad_norm": 0.0036493882071226835,
-      "learning_rate": 0.00019999016663295165,
-      "loss": 46.0,
-      "step": 27730
-    },
-    {
-      "epoch": 4.4658802689319215,
-      "grad_norm": 0.021345365792512894,
-      "learning_rate": 0.00019999016592347232,
-      "loss": 46.0,
-      "step": 27731
-    },
-    {
-      "epoch": 4.466041306010709,
-      "grad_norm": 0.0017449143342673779,
-      "learning_rate": 0.00019999016521396744,
-      "loss": 46.0,
-      "step": 27732
-    },
-    {
-      "epoch": 4.466202343089496,
-      "grad_norm": 0.0019290444906800985,
-      "learning_rate": 0.00019999016450443692,
-      "loss": 46.0,
-      "step": 27733
-    },
-    {
-      "epoch": 4.466363380168284,
-      "grad_norm": 0.008033310994505882,
-      "learning_rate": 0.00019999016379488086,
-      "loss": 46.0,
-      "step": 27734
-    },
-    {
-      "epoch": 4.466524417247071,
-      "grad_norm": 0.008996175602078438,
-      "learning_rate": 0.00019999016308529916,
-      "loss": 46.0,
-      "step": 27735
-    },
-    {
-      "epoch": 4.466685454325859,
-      "grad_norm": 0.0116839949041605,
-      "learning_rate": 0.0001999901623756919,
-      "loss": 46.0,
-      "step": 27736
-    },
-    {
-      "epoch": 4.466846491404646,
-      "grad_norm": 0.0013488969998434186,
-      "learning_rate": 0.00019999016166605903,
-      "loss": 46.0,
-      "step": 27737
-    },
-    {
-      "epoch": 4.467007528483434,
-      "grad_norm": 0.005387578159570694,
-      "learning_rate": 0.00019999016095640057,
-      "loss": 46.0,
-      "step": 27738
-    },
-    {
-      "epoch": 4.467168565562221,
-      "grad_norm": 0.01853710040450096,
-      "learning_rate": 0.00019999016024671652,
-      "loss": 46.0,
-      "step": 27739
-    },
-    {
-      "epoch": 4.4673296026410085,
-      "grad_norm": 0.0019915702287107706,
-      "learning_rate": 0.00019999015953700688,
-      "loss": 46.0,
-      "step": 27740
-    },
-    {
-      "epoch": 4.467490639719795,
-      "grad_norm": 0.006848345045000315,
-      "learning_rate": 0.00019999015882727163,
-      "loss": 46.0,
-      "step": 27741
-    },
-    {
-      "epoch": 4.4676516767985825,
-      "grad_norm": 0.005931831430643797,
-      "learning_rate": 0.0001999901581175108,
-      "loss": 46.0,
-      "step": 27742
-    },
-    {
-      "epoch": 4.46781271387737,
-      "grad_norm": 0.013120879419147968,
-      "learning_rate": 0.0001999901574077244,
-      "loss": 46.0,
-      "step": 27743
-    },
-    {
-      "epoch": 4.4679737509561575,
-      "grad_norm": 0.011298086494207382,
-      "learning_rate": 0.0001999901566979124,
-      "loss": 46.0,
-      "step": 27744
-    },
-    {
-      "epoch": 4.468134788034945,
-      "grad_norm": 0.004889341536909342,
-      "learning_rate": 0.0001999901559880748,
-      "loss": 46.0,
-      "step": 27745
-    },
-    {
-      "epoch": 4.468295825113732,
-      "grad_norm": 0.0023594731464982033,
-      "learning_rate": 0.00019999015527821158,
-      "loss": 46.0,
-      "step": 27746
-    },
-    {
-      "epoch": 4.46845686219252,
-      "grad_norm": 0.0008554062806069851,
-      "learning_rate": 0.0001999901545683228,
-      "loss": 46.0,
-      "step": 27747
-    },
-    {
-      "epoch": 4.468617899271307,
-      "grad_norm": 0.009103850461542606,
-      "learning_rate": 0.00019999015385840842,
-      "loss": 46.0,
-      "step": 27748
-    },
-    {
-      "epoch": 4.468778936350095,
-      "grad_norm": 0.010141601786017418,
-      "learning_rate": 0.00019999015314846845,
-      "loss": 46.0,
-      "step": 27749
-    },
-    {
-      "epoch": 4.468939973428882,
-      "grad_norm": 0.009308332577347755,
-      "learning_rate": 0.0001999901524385029,
-      "loss": 46.0,
-      "step": 27750
-    },
-    {
-      "epoch": 4.46910101050767,
-      "grad_norm": 0.0033651439007371664,
-      "learning_rate": 0.00019999015172851174,
-      "loss": 46.0,
-      "step": 27751
-    },
-    {
-      "epoch": 4.469262047586457,
-      "grad_norm": 0.002752259373664856,
-      "learning_rate": 0.00019999015101849498,
-      "loss": 46.0,
-      "step": 27752
-    },
-    {
-      "epoch": 4.4694230846652445,
-      "grad_norm": 0.0032868387643247843,
-      "learning_rate": 0.00019999015030845266,
-      "loss": 46.0,
-      "step": 27753
-    },
-    {
-      "epoch": 4.469584121744032,
-      "grad_norm": 0.0008634725236333907,
-      "learning_rate": 0.0001999901495983847,
-      "loss": 46.0,
-      "step": 27754
-    },
-    {
-      "epoch": 4.469745158822819,
-      "grad_norm": 0.00149381288792938,
-      "learning_rate": 0.00019999014888829117,
-      "loss": 46.0,
-      "step": 27755
-    },
-    {
-      "epoch": 4.469906195901606,
-      "grad_norm": 0.00942729227244854,
-      "learning_rate": 0.00019999014817817206,
-      "loss": 46.0,
-      "step": 27756
-    },
-    {
-      "epoch": 4.470067232980393,
-      "grad_norm": 0.003926991950720549,
-      "learning_rate": 0.00019999014746802733,
-      "loss": 46.0,
-      "step": 27757
-    },
-    {
-      "epoch": 4.470228270059181,
-      "grad_norm": 0.0026579948607832193,
-      "learning_rate": 0.00019999014675785705,
-      "loss": 46.0,
-      "step": 27758
-    },
-    {
-      "epoch": 4.470389307137968,
-      "grad_norm": 0.010880478657782078,
-      "learning_rate": 0.00019999014604766115,
-      "loss": 46.0,
-      "step": 27759
-    },
-    {
-      "epoch": 4.470550344216756,
-      "grad_norm": 0.010115723125636578,
-      "learning_rate": 0.00019999014533743966,
-      "loss": 46.0,
-      "step": 27760
-    },
-    {
-      "epoch": 4.470711381295543,
-      "grad_norm": 0.006723347585648298,
-      "learning_rate": 0.00019999014462719256,
-      "loss": 46.0,
-      "step": 27761
-    },
-    {
-      "epoch": 4.470872418374331,
-      "grad_norm": 0.0032588550820946693,
-      "learning_rate": 0.0001999901439169199,
-      "loss": 46.0,
-      "step": 27762
-    },
-    {
-      "epoch": 4.471033455453118,
-      "grad_norm": 0.006792832165956497,
-      "learning_rate": 0.00019999014320662165,
-      "loss": 46.0,
-      "step": 27763
-    },
-    {
-      "epoch": 4.471194492531906,
-      "grad_norm": 0.005623354576528072,
-      "learning_rate": 0.00019999014249629776,
-      "loss": 46.0,
-      "step": 27764
-    },
-    {
-      "epoch": 4.471355529610693,
-      "grad_norm": 0.0070923203602433205,
-      "learning_rate": 0.0001999901417859483,
-      "loss": 46.0,
-      "step": 27765
-    },
-    {
-      "epoch": 4.4715165666894805,
-      "grad_norm": 0.00419378699734807,
-      "learning_rate": 0.0001999901410755733,
-      "loss": 46.0,
-      "step": 27766
-    },
-    {
-      "epoch": 4.471677603768268,
-      "grad_norm": 0.0019027515081688762,
-      "learning_rate": 0.00019999014036517265,
-      "loss": 46.0,
-      "step": 27767
-    },
-    {
-      "epoch": 4.471838640847055,
-      "grad_norm": 0.0009966065408661962,
-      "learning_rate": 0.00019999013965474641,
-      "loss": 46.0,
-      "step": 27768
-    },
-    {
-      "epoch": 4.471999677925842,
-      "grad_norm": 0.006412910763174295,
-      "learning_rate": 0.0001999901389442946,
-      "loss": 46.0,
-      "step": 27769
-    },
-    {
-      "epoch": 4.472160715004629,
-      "grad_norm": 0.013252461329102516,
-      "learning_rate": 0.00019999013823381718,
-      "loss": 46.0,
-      "step": 27770
-    },
-    {
-      "epoch": 4.472321752083417,
-      "grad_norm": 0.002783434931188822,
-      "learning_rate": 0.0001999901375233142,
-      "loss": 46.0,
-      "step": 27771
-    },
-    {
-      "epoch": 4.472482789162204,
-      "grad_norm": 0.001420151675119996,
-      "learning_rate": 0.0001999901368127856,
-      "loss": 46.0,
-      "step": 27772
-    },
-    {
-      "epoch": 4.472643826240992,
-      "grad_norm": 0.002665820997208357,
-      "learning_rate": 0.0001999901361022314,
-      "loss": 46.0,
-      "step": 27773
-    },
-    {
-      "epoch": 4.472804863319779,
-      "grad_norm": 0.008886474184691906,
-      "learning_rate": 0.0001999901353916516,
-      "loss": 46.0,
-      "step": 27774
-    },
-    {
-      "epoch": 4.472965900398567,
-      "grad_norm": 0.00343678193166852,
-      "learning_rate": 0.00019999013468104625,
-      "loss": 46.0,
-      "step": 27775
-    },
-    {
-      "epoch": 4.473126937477354,
-      "grad_norm": 0.004481021780520678,
-      "learning_rate": 0.0001999901339704153,
-      "loss": 46.0,
-      "step": 27776
-    },
-    {
-      "epoch": 4.473287974556142,
-      "grad_norm": 0.009889495559036732,
-      "learning_rate": 0.00019999013325975874,
-      "loss": 46.0,
-      "step": 27777
-    },
-    {
-      "epoch": 4.473449011634929,
-      "grad_norm": 0.007639762479811907,
-      "learning_rate": 0.00019999013254907657,
-      "loss": 46.0,
-      "step": 27778
-    },
-    {
-      "epoch": 4.4736100487137165,
-      "grad_norm": 0.0052644130773842335,
-      "learning_rate": 0.00019999013183836882,
-      "loss": 46.0,
-      "step": 27779
-    },
-    {
-      "epoch": 4.473771085792504,
-      "grad_norm": 0.015168738551437855,
-      "learning_rate": 0.00019999013112763548,
-      "loss": 46.0,
-      "step": 27780
-    },
-    {
-      "epoch": 4.473932122871291,
-      "grad_norm": 0.001831917092204094,
-      "learning_rate": 0.00019999013041687659,
-      "loss": 46.0,
-      "step": 27781
-    },
-    {
-      "epoch": 4.474093159950079,
-      "grad_norm": 0.0044548953883349895,
-      "learning_rate": 0.00019999012970609205,
-      "loss": 46.0,
-      "step": 27782
-    },
-    {
-      "epoch": 4.474254197028866,
-      "grad_norm": 0.0151712941005826,
-      "learning_rate": 0.00019999012899528195,
-      "loss": 46.0,
-      "step": 27783
-    },
-    {
-      "epoch": 4.474415234107653,
-      "grad_norm": 0.005104460753500462,
-      "learning_rate": 0.00019999012828444623,
-      "loss": 46.0,
-      "step": 27784
-    },
-    {
-      "epoch": 4.47457627118644,
-      "grad_norm": 0.0052878339774906635,
-      "learning_rate": 0.00019999012757358493,
-      "loss": 46.0,
-      "step": 27785
-    },
-    {
-      "epoch": 4.474737308265228,
-      "grad_norm": 0.004600665532052517,
-      "learning_rate": 0.00019999012686269804,
-      "loss": 46.0,
-      "step": 27786
-    },
-    {
-      "epoch": 4.474898345344015,
-      "grad_norm": 0.0005238695885054767,
-      "learning_rate": 0.00019999012615178557,
-      "loss": 46.0,
-      "step": 27787
-    },
-    {
-      "epoch": 4.475059382422803,
-      "grad_norm": 0.0009664374520070851,
-      "learning_rate": 0.00019999012544084748,
-      "loss": 46.0,
-      "step": 27788
-    },
-    {
-      "epoch": 4.47522041950159,
-      "grad_norm": 0.012209751643240452,
-      "learning_rate": 0.00019999012472988383,
-      "loss": 46.0,
-      "step": 27789
-    },
-    {
-      "epoch": 4.475381456580378,
-      "grad_norm": 0.015519886277616024,
-      "learning_rate": 0.00019999012401889456,
-      "loss": 46.0,
-      "step": 27790
-    },
-    {
-      "epoch": 4.475542493659165,
-      "grad_norm": 0.008853236213326454,
-      "learning_rate": 0.0001999901233078797,
-      "loss": 46.0,
-      "step": 27791
-    },
-    {
-      "epoch": 4.4757035307379525,
-      "grad_norm": 0.00215332699008286,
-      "learning_rate": 0.00019999012259683927,
-      "loss": 46.0,
-      "step": 27792
-    },
-    {
-      "epoch": 4.47586456781674,
-      "grad_norm": 0.0018610305851325393,
-      "learning_rate": 0.00019999012188577324,
-      "loss": 46.0,
-      "step": 27793
-    },
-    {
-      "epoch": 4.476025604895527,
-      "grad_norm": 0.006760372314602137,
-      "learning_rate": 0.0001999901211746816,
-      "loss": 46.0,
-      "step": 27794
-    },
-    {
-      "epoch": 4.476186641974315,
-      "grad_norm": 0.011449955403804779,
-      "learning_rate": 0.00019999012046356438,
-      "loss": 46.0,
-      "step": 27795
-    },
-    {
-      "epoch": 4.476347679053102,
-      "grad_norm": 0.0010572896571829915,
-      "learning_rate": 0.00019999011975242156,
-      "loss": 46.0,
-      "step": 27796
-    },
-    {
-      "epoch": 4.47650871613189,
-      "grad_norm": 0.0009545032517053187,
-      "learning_rate": 0.00019999011904125316,
-      "loss": 46.0,
-      "step": 27797
-    },
-    {
-      "epoch": 4.476669753210677,
-      "grad_norm": 0.005871484987437725,
-      "learning_rate": 0.00019999011833005917,
-      "loss": 46.0,
-      "step": 27798
-    },
-    {
-      "epoch": 4.476830790289464,
-      "grad_norm": 0.0022825472988188267,
-      "learning_rate": 0.0001999901176188396,
-      "loss": 46.0,
-      "step": 27799
-    },
-    {
-      "epoch": 4.476991827368251,
-      "grad_norm": 0.002660314552485943,
-      "learning_rate": 0.0001999901169075944,
-      "loss": 46.0,
-      "step": 27800
-    },
-    {
-      "epoch": 4.477152864447039,
-      "grad_norm": 0.0111022237688303,
-      "learning_rate": 0.00019999011619632363,
-      "loss": 46.0,
-      "step": 27801
-    },
-    {
-      "epoch": 4.477313901525826,
-      "grad_norm": 0.0045442297123372555,
-      "learning_rate": 0.00019999011548502726,
-      "loss": 46.0,
-      "step": 27802
-    },
-    {
-      "epoch": 4.4774749386046135,
-      "grad_norm": 0.001113284844905138,
-      "learning_rate": 0.00019999011477370528,
-      "loss": 46.0,
-      "step": 27803
-    },
-    {
-      "epoch": 4.477635975683401,
-      "grad_norm": 0.005483392626047134,
-      "learning_rate": 0.00019999011406235772,
-      "loss": 46.0,
-      "step": 27804
-    },
-    {
-      "epoch": 4.4777970127621884,
-      "grad_norm": 0.0027010401245206594,
-      "learning_rate": 0.0001999901133509846,
-      "loss": 46.0,
-      "step": 27805
-    },
-    {
-      "epoch": 4.477958049840976,
-      "grad_norm": 0.0022945499513298273,
-      "learning_rate": 0.00019999011263958588,
-      "loss": 46.0,
-      "step": 27806
-    },
-    {
-      "epoch": 4.478119086919763,
-      "grad_norm": 0.0059724366292357445,
-      "learning_rate": 0.00019999011192816152,
-      "loss": 46.0,
-      "step": 27807
-    },
-    {
-      "epoch": 4.478280123998551,
-      "grad_norm": 0.0036939126439392567,
-      "learning_rate": 0.0001999901112167116,
-      "loss": 46.0,
-      "step": 27808
-    },
-    {
-      "epoch": 4.478441161077338,
-      "grad_norm": 0.002406270941719413,
-      "learning_rate": 0.0001999901105052361,
-      "loss": 46.0,
-      "step": 27809
-    },
-    {
-      "epoch": 4.478602198156126,
-      "grad_norm": 0.017762577161192894,
-      "learning_rate": 0.000199990109793735,
-      "loss": 46.0,
-      "step": 27810
-    },
-    {
-      "epoch": 4.478763235234913,
-      "grad_norm": 0.0014806096442043781,
-      "learning_rate": 0.0001999901090822083,
-      "loss": 46.0,
-      "step": 27811
-    },
-    {
-      "epoch": 4.478924272313701,
-      "grad_norm": 0.003524838015437126,
-      "learning_rate": 0.000199990108370656,
-      "loss": 46.0,
-      "step": 27812
-    },
-    {
-      "epoch": 4.479085309392488,
-      "grad_norm": 0.0012625409290194511,
-      "learning_rate": 0.00019999010765907812,
-      "loss": 46.0,
-      "step": 27813
-    },
-    {
-      "epoch": 4.479246346471275,
-      "grad_norm": 0.003366118762642145,
-      "learning_rate": 0.00019999010694747465,
-      "loss": 46.0,
-      "step": 27814
-    },
-    {
-      "epoch": 4.479407383550062,
-      "grad_norm": 0.005326335318386555,
-      "learning_rate": 0.00019999010623584557,
-      "loss": 46.0,
-      "step": 27815
-    },
-    {
-      "epoch": 4.4795684206288495,
-      "grad_norm": 0.017563287168741226,
-      "learning_rate": 0.00019999010552419094,
-      "loss": 46.0,
-      "step": 27816
-    },
-    {
-      "epoch": 4.479729457707637,
-      "grad_norm": 0.004145849496126175,
-      "learning_rate": 0.00019999010481251065,
-      "loss": 46.0,
-      "step": 27817
-    },
-    {
-      "epoch": 4.479890494786424,
-      "grad_norm": 0.003242343896999955,
-      "learning_rate": 0.00019999010410080481,
-      "loss": 46.0,
-      "step": 27818
-    },
-    {
-      "epoch": 4.480051531865212,
-      "grad_norm": 0.008767087012529373,
-      "learning_rate": 0.00019999010338907339,
-      "loss": 46.0,
-      "step": 27819
-    },
-    {
-      "epoch": 4.480212568943999,
-      "grad_norm": 0.0042501818388700485,
-      "learning_rate": 0.00019999010267731637,
-      "loss": 46.0,
-      "step": 27820
-    },
-    {
-      "epoch": 4.480373606022787,
-      "grad_norm": 0.0023322668857872486,
-      "learning_rate": 0.00019999010196553374,
-      "loss": 46.0,
-      "step": 27821
-    },
-    {
-      "epoch": 4.480534643101574,
-      "grad_norm": 0.0037910579703748226,
-      "learning_rate": 0.00019999010125372553,
-      "loss": 46.0,
-      "step": 27822
-    },
-    {
-      "epoch": 4.480695680180362,
-      "grad_norm": 0.003508639056235552,
-      "learning_rate": 0.00019999010054189172,
-      "loss": 46.0,
-      "step": 27823
-    },
-    {
-      "epoch": 4.480856717259149,
-      "grad_norm": 0.0035073955077677965,
-      "learning_rate": 0.00019999009983003233,
-      "loss": 46.0,
-      "step": 27824
-    },
-    {
-      "epoch": 4.481017754337937,
-      "grad_norm": 0.0024367840960621834,
-      "learning_rate": 0.00019999009911814733,
-      "loss": 46.0,
-      "step": 27825
-    },
-    {
-      "epoch": 4.481178791416724,
-      "grad_norm": 0.003021844429895282,
-      "learning_rate": 0.00019999009840623676,
-      "loss": 46.0,
-      "step": 27826
-    },
-    {
-      "epoch": 4.4813398284955115,
-      "grad_norm": 0.0015001606661826372,
-      "learning_rate": 0.00019999009769430058,
-      "loss": 46.0,
-      "step": 27827
-    },
-    {
-      "epoch": 4.481500865574299,
-      "grad_norm": 0.0023999549448490143,
-      "learning_rate": 0.00019999009698233881,
-      "loss": 46.0,
-      "step": 27828
-    },
-    {
-      "epoch": 4.4816619026530855,
-      "grad_norm": 0.0007765580667182803,
-      "learning_rate": 0.00019999009627035146,
-      "loss": 46.0,
-      "step": 27829
-    },
-    {
-      "epoch": 4.481822939731873,
-      "grad_norm": 0.005309755448251963,
-      "learning_rate": 0.00019999009555833846,
-      "loss": 46.0,
-      "step": 27830
-    },
-    {
-      "epoch": 4.48198397681066,
-      "grad_norm": 0.0027437207754701376,
-      "learning_rate": 0.00019999009484629996,
-      "loss": 46.0,
-      "step": 27831
-    },
-    {
-      "epoch": 4.482145013889448,
-      "grad_norm": 0.002044303109869361,
-      "learning_rate": 0.00019999009413423582,
-      "loss": 46.0,
-      "step": 27832
-    },
-    {
-      "epoch": 4.482306050968235,
-      "grad_norm": 0.001303886529058218,
-      "learning_rate": 0.00019999009342214606,
-      "loss": 46.0,
-      "step": 27833
-    },
-    {
-      "epoch": 4.482467088047023,
-      "grad_norm": 0.002399574965238571,
-      "learning_rate": 0.00019999009271003075,
-      "loss": 46.0,
-      "step": 27834
-    },
-    {
-      "epoch": 4.48262812512581,
-      "grad_norm": 0.0034279238898307085,
-      "learning_rate": 0.00019999009199788982,
-      "loss": 46.0,
-      "step": 27835
-    },
-    {
-      "epoch": 4.482789162204598,
-      "grad_norm": 0.005639585666358471,
-      "learning_rate": 0.00019999009128572333,
-      "loss": 46.0,
-      "step": 27836
-    },
-    {
-      "epoch": 4.482950199283385,
-      "grad_norm": 0.018510637804865837,
-      "learning_rate": 0.00019999009057353125,
-      "loss": 46.0,
-      "step": 27837
-    },
-    {
-      "epoch": 4.483111236362173,
-      "grad_norm": 0.004043771885335445,
-      "learning_rate": 0.00019999008986131353,
-      "loss": 46.0,
-      "step": 27838
-    },
-    {
-      "epoch": 4.48327227344096,
-      "grad_norm": 0.0028538291808217764,
-      "learning_rate": 0.00019999008914907025,
-      "loss": 46.0,
-      "step": 27839
-    },
-    {
-      "epoch": 4.4834333105197475,
-      "grad_norm": 0.0019156353082507849,
-      "learning_rate": 0.00019999008843680138,
-      "loss": 46.0,
-      "step": 27840
-    },
-    {
-      "epoch": 4.483594347598535,
-      "grad_norm": 0.0016860762843862176,
-      "learning_rate": 0.00019999008772450693,
-      "loss": 46.0,
-      "step": 27841
-    },
-    {
-      "epoch": 4.4837553846773215,
-      "grad_norm": 0.0018055159598588943,
-      "learning_rate": 0.00019999008701218686,
-      "loss": 46.0,
-      "step": 27842
-    },
-    {
-      "epoch": 4.483916421756109,
-      "grad_norm": 0.004454894922673702,
-      "learning_rate": 0.00019999008629984123,
-      "loss": 46.0,
-      "step": 27843
-    },
-    {
-      "epoch": 4.484077458834896,
-      "grad_norm": 0.001974615501239896,
-      "learning_rate": 0.00019999008558746996,
-      "loss": 46.0,
-      "step": 27844
-    },
-    {
-      "epoch": 4.484238495913684,
-      "grad_norm": 0.0023097912780940533,
-      "learning_rate": 0.00019999008487507313,
-      "loss": 46.0,
-      "step": 27845
-    },
-    {
-      "epoch": 4.484399532992471,
-      "grad_norm": 0.015262618660926819,
-      "learning_rate": 0.0001999900841626507,
-      "loss": 46.0,
-      "step": 27846
-    },
-    {
-      "epoch": 4.484560570071259,
-      "grad_norm": 0.012720214203000069,
-      "learning_rate": 0.00019999008345020268,
-      "loss": 46.0,
-      "step": 27847
-    },
-    {
-      "epoch": 4.484721607150046,
-      "grad_norm": 0.004694274626672268,
-      "learning_rate": 0.00019999008273772906,
-      "loss": 46.0,
-      "step": 27848
-    },
-    {
-      "epoch": 4.484882644228834,
-      "grad_norm": 0.009466389194130898,
-      "learning_rate": 0.00019999008202522983,
-      "loss": 46.0,
-      "step": 27849
-    },
-    {
-      "epoch": 4.485043681307621,
-      "grad_norm": 0.0029717348515987396,
-      "learning_rate": 0.00019999008131270504,
-      "loss": 46.0,
-      "step": 27850
-    },
-    {
-      "epoch": 4.4852047183864086,
-      "grad_norm": 0.0018321602838113904,
-      "learning_rate": 0.00019999008060015468,
-      "loss": 46.0,
-      "step": 27851
-    },
-    {
-      "epoch": 4.485365755465196,
-      "grad_norm": 0.0048034158535301685,
-      "learning_rate": 0.00019999007988757866,
-      "loss": 46.0,
-      "step": 27852
-    },
-    {
-      "epoch": 4.4855267925439835,
-      "grad_norm": 0.003441430628299713,
-      "learning_rate": 0.0001999900791749771,
-      "loss": 46.0,
-      "step": 27853
-    },
-    {
-      "epoch": 4.485687829622771,
-      "grad_norm": 0.0032759748864918947,
-      "learning_rate": 0.00019999007846234994,
-      "loss": 46.0,
-      "step": 27854
-    },
-    {
-      "epoch": 4.485848866701558,
-      "grad_norm": 0.002269813558086753,
-      "learning_rate": 0.00019999007774969716,
-      "loss": 46.0,
-      "step": 27855
-    },
-    {
-      "epoch": 4.486009903780346,
-      "grad_norm": 0.0019601918756961823,
-      "learning_rate": 0.00019999007703701881,
-      "loss": 46.0,
-      "step": 27856
-    },
-    {
-      "epoch": 4.486170940859132,
-      "grad_norm": 0.0030982508324086666,
-      "learning_rate": 0.00019999007632431488,
-      "loss": 46.0,
-      "step": 27857
-    },
-    {
-      "epoch": 4.48633197793792,
-      "grad_norm": 0.010010729543864727,
-      "learning_rate": 0.00019999007561158534,
-      "loss": 46.0,
-      "step": 27858
-    },
-    {
-      "epoch": 4.486493015016707,
-      "grad_norm": 0.007289108820259571,
-      "learning_rate": 0.00019999007489883018,
-      "loss": 46.0,
-      "step": 27859
-    },
-    {
-      "epoch": 4.486654052095495,
-      "grad_norm": 0.016130438074469566,
-      "learning_rate": 0.00019999007418604946,
-      "loss": 46.0,
-      "step": 27860
-    },
-    {
-      "epoch": 4.486815089174282,
-      "grad_norm": 0.002619587816298008,
-      "learning_rate": 0.00019999007347324316,
-      "loss": 46.0,
-      "step": 27861
-    },
-    {
-      "epoch": 4.48697612625307,
-      "grad_norm": 0.010935296304523945,
-      "learning_rate": 0.00019999007276041126,
-      "loss": 46.0,
-      "step": 27862
-    },
-    {
-      "epoch": 4.487137163331857,
-      "grad_norm": 0.002434429246932268,
-      "learning_rate": 0.00019999007204755378,
-      "loss": 46.0,
-      "step": 27863
-    },
-    {
-      "epoch": 4.4872982004106445,
-      "grad_norm": 0.005705788731575012,
-      "learning_rate": 0.00019999007133467066,
-      "loss": 46.0,
-      "step": 27864
-    },
-    {
-      "epoch": 4.487459237489432,
-      "grad_norm": 0.007801066618412733,
-      "learning_rate": 0.00019999007062176198,
-      "loss": 46.0,
-      "step": 27865
-    },
-    {
-      "epoch": 4.487620274568219,
-      "grad_norm": 0.012204373255372047,
-      "learning_rate": 0.0001999900699088277,
-      "loss": 46.0,
-      "step": 27866
-    },
-    {
-      "epoch": 4.487781311647007,
-      "grad_norm": 0.0011020316742360592,
-      "learning_rate": 0.00019999006919586785,
-      "loss": 46.0,
-      "step": 27867
-    },
-    {
-      "epoch": 4.487942348725794,
-      "grad_norm": 0.005237429868429899,
-      "learning_rate": 0.00019999006848288238,
-      "loss": 46.0,
-      "step": 27868
-    },
-    {
-      "epoch": 4.488103385804582,
-      "grad_norm": 0.002027320908382535,
-      "learning_rate": 0.00019999006776987133,
-      "loss": 46.0,
-      "step": 27869
-    },
-    {
-      "epoch": 4.488264422883369,
-      "grad_norm": 0.01505147386342287,
-      "learning_rate": 0.0001999900670568347,
-      "loss": 46.0,
-      "step": 27870
-    },
-    {
-      "epoch": 4.488425459962157,
-      "grad_norm": 0.0028392779640853405,
-      "learning_rate": 0.00019999006634377245,
-      "loss": 46.0,
-      "step": 27871
-    },
-    {
-      "epoch": 4.488586497040943,
-      "grad_norm": 0.0022702605929225683,
-      "learning_rate": 0.0001999900656306846,
-      "loss": 46.0,
-      "step": 27872
-    },
-    {
-      "epoch": 4.488747534119731,
-      "grad_norm": 0.0028342497535049915,
-      "learning_rate": 0.0001999900649175712,
-      "loss": 46.0,
-      "step": 27873
-    },
-    {
-      "epoch": 4.488908571198518,
-      "grad_norm": 0.0015059520956128836,
-      "learning_rate": 0.00019999006420443218,
-      "loss": 46.0,
-      "step": 27874
-    },
-    {
-      "epoch": 4.489069608277306,
-      "grad_norm": 0.0023169571068137884,
-      "learning_rate": 0.00019999006349126757,
-      "loss": 46.0,
-      "step": 27875
-    },
-    {
-      "epoch": 4.489230645356093,
-      "grad_norm": 0.0031651421450078487,
-      "learning_rate": 0.00019999006277807737,
-      "loss": 46.0,
-      "step": 27876
-    },
-    {
-      "epoch": 4.4893916824348805,
-      "grad_norm": 0.0031909889075905085,
-      "learning_rate": 0.0001999900620648616,
-      "loss": 46.0,
-      "step": 27877
-    },
-    {
-      "epoch": 4.489552719513668,
-      "grad_norm": 0.007121788803488016,
-      "learning_rate": 0.0001999900613516202,
-      "loss": 46.0,
-      "step": 27878
-    },
-    {
-      "epoch": 4.489713756592455,
-      "grad_norm": 0.0050599356181919575,
-      "learning_rate": 0.0001999900606383532,
-      "loss": 46.0,
-      "step": 27879
-    },
-    {
-      "epoch": 4.489874793671243,
-      "grad_norm": 0.011040827259421349,
-      "learning_rate": 0.00019999005992506064,
-      "loss": 46.0,
-      "step": 27880
-    },
-    {
-      "epoch": 4.49003583075003,
-      "grad_norm": 0.0010725490283221006,
-      "learning_rate": 0.00019999005921174248,
-      "loss": 46.0,
-      "step": 27881
-    },
-    {
-      "epoch": 4.490196867828818,
-      "grad_norm": 0.0033081036526709795,
-      "learning_rate": 0.00019999005849839874,
-      "loss": 46.0,
-      "step": 27882
-    },
-    {
-      "epoch": 4.490357904907605,
-      "grad_norm": 0.006857835687696934,
-      "learning_rate": 0.0001999900577850294,
-      "loss": 46.0,
-      "step": 27883
-    },
-    {
-      "epoch": 4.490518941986393,
-      "grad_norm": 0.0021876965183764696,
-      "learning_rate": 0.00019999005707163446,
-      "loss": 46.0,
-      "step": 27884
-    },
-    {
-      "epoch": 4.49067997906518,
-      "grad_norm": 0.007280195597559214,
-      "learning_rate": 0.0001999900563582139,
-      "loss": 46.0,
-      "step": 27885
-    },
-    {
-      "epoch": 4.490841016143968,
-      "grad_norm": 0.002429440850391984,
-      "learning_rate": 0.0001999900556447678,
-      "loss": 46.0,
-      "step": 27886
-    },
-    {
-      "epoch": 4.491002053222754,
-      "grad_norm": 0.0017278622835874557,
-      "learning_rate": 0.00019999005493129607,
-      "loss": 46.0,
-      "step": 27887
-    },
-    {
-      "epoch": 4.491163090301542,
-      "grad_norm": 0.005135042127221823,
-      "learning_rate": 0.00019999005421779878,
-      "loss": 46.0,
-      "step": 27888
-    },
-    {
-      "epoch": 4.491324127380329,
-      "grad_norm": 0.02064802125096321,
-      "learning_rate": 0.0001999900535042759,
-      "loss": 46.0,
-      "step": 27889
-    },
-    {
-      "epoch": 4.4914851644591165,
-      "grad_norm": 0.002363914856687188,
-      "learning_rate": 0.00019999005279072737,
-      "loss": 46.0,
-      "step": 27890
-    },
-    {
-      "epoch": 4.491646201537904,
-      "grad_norm": 0.0016027031233534217,
-      "learning_rate": 0.0001999900520771533,
-      "loss": 46.0,
-      "step": 27891
-    },
-    {
-      "epoch": 4.491807238616691,
-      "grad_norm": 0.002568686380982399,
-      "learning_rate": 0.00019999005136355362,
-      "loss": 46.0,
-      "step": 27892
-    },
-    {
-      "epoch": 4.491968275695479,
-      "grad_norm": 0.00444758078083396,
-      "learning_rate": 0.00019999005064992836,
-      "loss": 46.0,
-      "step": 27893
-    },
-    {
-      "epoch": 4.492129312774266,
-      "grad_norm": 0.009934254921972752,
-      "learning_rate": 0.0001999900499362775,
-      "loss": 46.0,
-      "step": 27894
-    },
-    {
-      "epoch": 4.492290349853054,
-      "grad_norm": 0.004778178408741951,
-      "learning_rate": 0.00019999004922260105,
-      "loss": 46.0,
-      "step": 27895
-    },
-    {
-      "epoch": 4.492451386931841,
-      "grad_norm": 0.0018919466529041529,
-      "learning_rate": 0.000199990048508899,
-      "loss": 46.0,
-      "step": 27896
-    },
-    {
-      "epoch": 4.492612424010629,
-      "grad_norm": 0.010147398337721825,
-      "learning_rate": 0.00019999004779517137,
-      "loss": 46.0,
-      "step": 27897
-    },
-    {
-      "epoch": 4.492773461089416,
-      "grad_norm": 0.002310806652531028,
-      "learning_rate": 0.00019999004708141812,
-      "loss": 46.0,
-      "step": 27898
-    },
-    {
-      "epoch": 4.492934498168204,
-      "grad_norm": 0.011375637724995613,
-      "learning_rate": 0.00019999004636763932,
-      "loss": 46.0,
-      "step": 27899
-    },
-    {
-      "epoch": 4.493095535246991,
-      "grad_norm": 0.0022303175646811724,
-      "learning_rate": 0.00019999004565383487,
-      "loss": 46.0,
-      "step": 27900
-    },
-    {
-      "epoch": 4.4932565723257785,
-      "grad_norm": 0.005507065914571285,
-      "learning_rate": 0.00019999004494000486,
-      "loss": 46.0,
-      "step": 27901
-    },
-    {
-      "epoch": 4.493417609404565,
-      "grad_norm": 0.0024187129456549883,
-      "learning_rate": 0.00019999004422614926,
-      "loss": 46.0,
-      "step": 27902
-    },
-    {
-      "epoch": 4.4935786464833525,
-      "grad_norm": 0.00918637402355671,
-      "learning_rate": 0.00019999004351226808,
-      "loss": 46.0,
-      "step": 27903
-    },
-    {
-      "epoch": 4.49373968356214,
-      "grad_norm": 0.0023364771623164415,
-      "learning_rate": 0.00019999004279836128,
-      "loss": 46.0,
-      "step": 27904
-    },
-    {
-      "epoch": 4.493900720640927,
-      "grad_norm": 0.0009917255956679583,
-      "learning_rate": 0.00019999004208442892,
-      "loss": 46.0,
-      "step": 27905
-    },
-    {
-      "epoch": 4.494061757719715,
-      "grad_norm": 0.0013017618330195546,
-      "learning_rate": 0.00019999004137047095,
-      "loss": 46.0,
-      "step": 27906
-    },
-    {
-      "epoch": 4.494222794798502,
-      "grad_norm": 0.0037942298222333193,
-      "learning_rate": 0.00019999004065648741,
-      "loss": 46.0,
-      "step": 27907
-    },
-    {
-      "epoch": 4.49438383187729,
-      "grad_norm": 0.00743611017242074,
-      "learning_rate": 0.00019999003994247824,
-      "loss": 46.0,
-      "step": 27908
-    },
-    {
-      "epoch": 4.494544868956077,
-      "grad_norm": 0.004279667977243662,
-      "learning_rate": 0.0001999900392284435,
-      "loss": 46.0,
-      "step": 27909
-    },
-    {
-      "epoch": 4.494705906034865,
-      "grad_norm": 0.0021074344404041767,
-      "learning_rate": 0.00019999003851438316,
-      "loss": 46.0,
-      "step": 27910
-    },
-    {
-      "epoch": 4.494866943113652,
-      "grad_norm": 0.0032866366673260927,
-      "learning_rate": 0.00019999003780029725,
-      "loss": 46.0,
-      "step": 27911
-    },
-    {
-      "epoch": 4.4950279801924395,
-      "grad_norm": 0.001152546377852559,
-      "learning_rate": 0.0001999900370861857,
-      "loss": 46.0,
-      "step": 27912
-    },
-    {
-      "epoch": 4.495189017271227,
-      "grad_norm": 0.002169529441744089,
-      "learning_rate": 0.0001999900363720486,
-      "loss": 46.0,
-      "step": 27913
-    },
-    {
-      "epoch": 4.4953500543500144,
-      "grad_norm": 0.004141353536397219,
-      "learning_rate": 0.00019999003565788587,
-      "loss": 46.0,
-      "step": 27914
-    },
-    {
-      "epoch": 4.495511091428802,
-      "grad_norm": 0.008625070564448833,
-      "learning_rate": 0.00019999003494369758,
-      "loss": 46.0,
-      "step": 27915
-    },
-    {
-      "epoch": 4.4956721285075885,
-      "grad_norm": 0.024159444496035576,
-      "learning_rate": 0.00019999003422948368,
-      "loss": 46.0,
-      "step": 27916
-    },
-    {
-      "epoch": 4.495833165586376,
-      "grad_norm": 0.0017476622015237808,
-      "learning_rate": 0.00019999003351524422,
-      "loss": 46.0,
-      "step": 27917
-    },
-    {
-      "epoch": 4.495994202665163,
-      "grad_norm": 0.0041218833066523075,
-      "learning_rate": 0.00019999003280097912,
-      "loss": 46.0,
-      "step": 27918
-    },
-    {
-      "epoch": 4.496155239743951,
-      "grad_norm": 0.030301189050078392,
-      "learning_rate": 0.00019999003208668844,
-      "loss": 46.0,
-      "step": 27919
-    },
-    {
-      "epoch": 4.496316276822738,
-      "grad_norm": 0.003254748648032546,
-      "learning_rate": 0.0001999900313723722,
-      "loss": 46.0,
-      "step": 27920
-    },
-    {
-      "epoch": 4.496477313901526,
-      "grad_norm": 0.0028789632488042116,
-      "learning_rate": 0.00019999003065803033,
-      "loss": 46.0,
-      "step": 27921
-    },
-    {
-      "epoch": 4.496638350980313,
-      "grad_norm": 0.0015815565129742026,
-      "learning_rate": 0.00019999002994366288,
-      "loss": 46.0,
-      "step": 27922
-    },
-    {
-      "epoch": 4.496799388059101,
-      "grad_norm": 0.00667013693600893,
-      "learning_rate": 0.00019999002922926984,
-      "loss": 46.0,
-      "step": 27923
-    },
-    {
-      "epoch": 4.496960425137888,
-      "grad_norm": 0.00903643574565649,
-      "learning_rate": 0.00019999002851485122,
-      "loss": 46.0,
-      "step": 27924
-    },
-    {
-      "epoch": 4.4971214622166755,
-      "grad_norm": 0.011258893646299839,
-      "learning_rate": 0.00019999002780040698,
-      "loss": 46.0,
-      "step": 27925
-    },
-    {
-      "epoch": 4.497282499295463,
-      "grad_norm": 0.007935581728816032,
-      "learning_rate": 0.00019999002708593716,
-      "loss": 46.0,
-      "step": 27926
-    },
-    {
-      "epoch": 4.49744353637425,
-      "grad_norm": 0.0037196953780949116,
-      "learning_rate": 0.00019999002637144177,
-      "loss": 46.0,
-      "step": 27927
-    },
-    {
-      "epoch": 4.497604573453038,
-      "grad_norm": 0.0012101439060643315,
-      "learning_rate": 0.00019999002565692078,
-      "loss": 46.0,
-      "step": 27928
-    },
-    {
-      "epoch": 4.497765610531825,
-      "grad_norm": 0.018383676186203957,
-      "learning_rate": 0.0001999900249423742,
-      "loss": 46.0,
-      "step": 27929
-    },
-    {
-      "epoch": 4.497926647610612,
-      "grad_norm": 0.0027241890784353018,
-      "learning_rate": 0.000199990024227802,
-      "loss": 46.0,
-      "step": 27930
-    },
-    {
-      "epoch": 4.498087684689399,
-      "grad_norm": 0.0075929309241473675,
-      "learning_rate": 0.00019999002351320423,
-      "loss": 46.0,
-      "step": 27931
-    },
-    {
-      "epoch": 4.498248721768187,
-      "grad_norm": 0.003081547562032938,
-      "learning_rate": 0.00019999002279858085,
-      "loss": 46.0,
-      "step": 27932
-    },
-    {
-      "epoch": 4.498409758846974,
-      "grad_norm": 0.007031906396150589,
-      "learning_rate": 0.0001999900220839319,
-      "loss": 46.0,
-      "step": 27933
-    },
-    {
-      "epoch": 4.498570795925762,
-      "grad_norm": 0.0015045617474243045,
-      "learning_rate": 0.00019999002136925732,
-      "loss": 46.0,
-      "step": 27934
-    },
-    {
-      "epoch": 4.498731833004549,
-      "grad_norm": 0.002424549777060747,
-      "learning_rate": 0.00019999002065455718,
-      "loss": 46.0,
-      "step": 27935
-    },
-    {
-      "epoch": 4.498892870083337,
-      "grad_norm": 0.00467715784907341,
-      "learning_rate": 0.00019999001993983146,
-      "loss": 46.0,
-      "step": 27936
-    },
-    {
-      "epoch": 4.499053907162124,
-      "grad_norm": 0.002440407406538725,
-      "learning_rate": 0.00019999001922508012,
-      "loss": 46.0,
-      "step": 27937
-    },
-    {
-      "epoch": 4.4992149442409115,
-      "grad_norm": 0.00235581211745739,
-      "learning_rate": 0.0001999900185103032,
-      "loss": 46.0,
-      "step": 27938
-    },
-    {
-      "epoch": 4.499375981319699,
-      "grad_norm": 0.004303944297134876,
-      "learning_rate": 0.00019999001779550068,
-      "loss": 46.0,
-      "step": 27939
-    },
-    {
-      "epoch": 4.499537018398486,
-      "grad_norm": 0.002360617509111762,
-      "learning_rate": 0.00019999001708067258,
-      "loss": 46.0,
-      "step": 27940
-    },
-    {
-      "epoch": 4.499698055477274,
-      "grad_norm": 0.0028185902629047632,
-      "learning_rate": 0.00019999001636581887,
-      "loss": 46.0,
-      "step": 27941
-    },
-    {
-      "epoch": 4.499859092556061,
-      "grad_norm": 0.002025514841079712,
-      "learning_rate": 0.00019999001565093957,
-      "loss": 46.0,
-      "step": 27942
-    },
-    {
-      "epoch": 4.500020129634849,
-      "grad_norm": 0.004514962900429964,
-      "learning_rate": 0.0001999900149360347,
-      "loss": 46.0,
-      "step": 27943
-    },
-    {
-      "epoch": 4.500181166713636,
-      "grad_norm": 0.0014514346839860082,
-      "learning_rate": 0.0001999900142211042,
-      "loss": 46.0,
-      "step": 27944
-    },
-    {
-      "epoch": 4.500342203792423,
-      "grad_norm": 0.015978841111063957,
-      "learning_rate": 0.00019999001350614815,
-      "loss": 46.0,
-      "step": 27945
-    },
-    {
-      "epoch": 4.50050324087121,
-      "grad_norm": 0.016491329297423363,
-      "learning_rate": 0.00019999001279116647,
-      "loss": 46.0,
-      "step": 27946
-    },
-    {
-      "epoch": 4.500664277949998,
-      "grad_norm": 0.002122324425727129,
-      "learning_rate": 0.0001999900120761592,
-      "loss": 46.0,
-      "step": 27947
-    },
-    {
-      "epoch": 4.500825315028785,
-      "grad_norm": 0.0022590658627450466,
-      "learning_rate": 0.00019999001136112638,
-      "loss": 46.0,
-      "step": 27948
-    },
-    {
-      "epoch": 4.500986352107573,
-      "grad_norm": 0.002661405596882105,
-      "learning_rate": 0.00019999001064606794,
-      "loss": 46.0,
-      "step": 27949
-    },
-    {
-      "epoch": 4.50114738918636,
-      "grad_norm": 0.0037559238262474537,
-      "learning_rate": 0.00019999000993098392,
-      "loss": 46.0,
-      "step": 27950
-    },
-    {
-      "epoch": 4.5013084262651475,
-      "grad_norm": 0.0017206493066623807,
-      "learning_rate": 0.0001999900092158743,
-      "loss": 46.0,
-      "step": 27951
-    },
-    {
-      "epoch": 4.501469463343935,
-      "grad_norm": 0.006683338899165392,
-      "learning_rate": 0.00019999000850073908,
-      "loss": 46.0,
-      "step": 27952
-    },
-    {
-      "epoch": 4.501630500422722,
-      "grad_norm": 0.00048455732758156955,
-      "learning_rate": 0.00019999000778557827,
-      "loss": 46.0,
-      "step": 27953
-    },
-    {
-      "epoch": 4.50179153750151,
-      "grad_norm": 0.006423015613108873,
-      "learning_rate": 0.00019999000707039187,
-      "loss": 46.0,
-      "step": 27954
-    },
-    {
-      "epoch": 4.501952574580297,
-      "grad_norm": 0.002031286945566535,
-      "learning_rate": 0.00019999000635517988,
-      "loss": 46.0,
-      "step": 27955
-    },
-    {
-      "epoch": 4.502113611659085,
-      "grad_norm": 0.01293118018656969,
-      "learning_rate": 0.00019999000563994228,
-      "loss": 46.0,
-      "step": 27956
-    },
-    {
-      "epoch": 4.502274648737872,
-      "grad_norm": 0.0038702180609107018,
-      "learning_rate": 0.0001999900049246791,
-      "loss": 46.0,
-      "step": 27957
-    },
-    {
-      "epoch": 4.50243568581666,
-      "grad_norm": 0.00194447988178581,
-      "learning_rate": 0.00019999000420939034,
-      "loss": 46.0,
-      "step": 27958
-    },
-    {
-      "epoch": 4.502596722895447,
-      "grad_norm": 0.002438580384477973,
-      "learning_rate": 0.000199990003494076,
-      "loss": 46.0,
-      "step": 27959
-    },
-    {
-      "epoch": 4.502757759974234,
-      "grad_norm": 0.005687889643013477,
-      "learning_rate": 0.00019999000277873602,
-      "loss": 46.0,
-      "step": 27960
-    },
-    {
-      "epoch": 4.502918797053021,
-      "grad_norm": 0.006321834400296211,
-      "learning_rate": 0.00019999000206337049,
-      "loss": 46.0,
-      "step": 27961
-    },
-    {
-      "epoch": 4.503079834131809,
-      "grad_norm": 0.005697234533727169,
-      "learning_rate": 0.00019999000134797933,
-      "loss": 46.0,
-      "step": 27962
-    },
-    {
-      "epoch": 4.503240871210596,
-      "grad_norm": 0.011142939329147339,
-      "learning_rate": 0.00019999000063256262,
-      "loss": 46.0,
-      "step": 27963
-    },
-    {
-      "epoch": 4.5034019082893835,
-      "grad_norm": 0.01310784462839365,
-      "learning_rate": 0.0001999899999171203,
-      "loss": 46.0,
-      "step": 27964
-    },
-    {
-      "epoch": 4.503562945368171,
-      "grad_norm": 0.011725926771759987,
-      "learning_rate": 0.00019998999920165238,
-      "loss": 46.0,
-      "step": 27965
-    },
-    {
-      "epoch": 4.503723982446958,
-      "grad_norm": 0.00788667518645525,
-      "learning_rate": 0.00019998999848615885,
-      "loss": 46.0,
-      "step": 27966
-    },
-    {
-      "epoch": 4.503885019525746,
-      "grad_norm": 0.0017495183274149895,
-      "learning_rate": 0.00019998999777063976,
-      "loss": 46.0,
-      "step": 27967
-    },
-    {
-      "epoch": 4.504046056604533,
-      "grad_norm": 0.0035278829745948315,
-      "learning_rate": 0.00019998999705509506,
-      "loss": 46.0,
-      "step": 27968
-    },
-    {
-      "epoch": 4.504207093683321,
-      "grad_norm": 0.009284217841923237,
-      "learning_rate": 0.00019998999633952477,
-      "loss": 46.0,
-      "step": 27969
-    },
-    {
-      "epoch": 4.504368130762108,
-      "grad_norm": 0.0018602310447022319,
-      "learning_rate": 0.0001999899956239289,
-      "loss": 46.0,
-      "step": 27970
-    },
-    {
-      "epoch": 4.504529167840896,
-      "grad_norm": 0.0038876133039593697,
-      "learning_rate": 0.00019998999490830746,
-      "loss": 46.0,
-      "step": 27971
-    },
-    {
-      "epoch": 4.504690204919683,
-      "grad_norm": 0.00523751974105835,
-      "learning_rate": 0.00019998999419266038,
-      "loss": 46.0,
-      "step": 27972
-    },
-    {
-      "epoch": 4.5048512419984705,
-      "grad_norm": 0.015857869759202003,
-      "learning_rate": 0.00019998999347698771,
-      "loss": 46.0,
-      "step": 27973
-    },
-    {
-      "epoch": 4.505012279077258,
-      "grad_norm": 0.0032813113648444414,
-      "learning_rate": 0.00019998999276128946,
-      "loss": 46.0,
-      "step": 27974
-    },
-    {
-      "epoch": 4.5051733161560445,
-      "grad_norm": 0.0012366772862151265,
-      "learning_rate": 0.00019998999204556562,
-      "loss": 46.0,
-      "step": 27975
-    },
-    {
-      "epoch": 4.505334353234832,
-      "grad_norm": 0.0009771501645445824,
-      "learning_rate": 0.00019998999132981622,
-      "loss": 46.0,
-      "step": 27976
-    },
-    {
-      "epoch": 4.5054953903136195,
-      "grad_norm": 0.001074718777090311,
-      "learning_rate": 0.00019998999061404118,
-      "loss": 46.0,
-      "step": 27977
-    },
-    {
-      "epoch": 4.505656427392407,
-      "grad_norm": 0.014997282065451145,
-      "learning_rate": 0.00019998998989824055,
-      "loss": 46.0,
-      "step": 27978
-    },
-    {
-      "epoch": 4.505817464471194,
-      "grad_norm": 0.0018334087217226624,
-      "learning_rate": 0.00019998998918241436,
-      "loss": 46.0,
-      "step": 27979
-    },
-    {
-      "epoch": 4.505978501549982,
-      "grad_norm": 0.003739659208804369,
-      "learning_rate": 0.00019998998846656256,
-      "loss": 46.0,
-      "step": 27980
-    },
-    {
-      "epoch": 4.506139538628769,
-      "grad_norm": 0.0026020260993391275,
-      "learning_rate": 0.00019998998775068517,
-      "loss": 46.0,
-      "step": 27981
-    },
-    {
-      "epoch": 4.506300575707557,
-      "grad_norm": 0.003148492192849517,
-      "learning_rate": 0.00019998998703478217,
-      "loss": 46.0,
-      "step": 27982
-    },
-    {
-      "epoch": 4.506461612786344,
-      "grad_norm": 0.003607948310673237,
-      "learning_rate": 0.0001999899863188536,
-      "loss": 46.0,
-      "step": 27983
-    },
-    {
-      "epoch": 4.506622649865132,
-      "grad_norm": 0.004364179912954569,
-      "learning_rate": 0.00019998998560289942,
-      "loss": 46.0,
-      "step": 27984
-    },
-    {
-      "epoch": 4.506783686943919,
-      "grad_norm": 0.0030395081266760826,
-      "learning_rate": 0.00019998998488691966,
-      "loss": 46.0,
-      "step": 27985
-    },
-    {
-      "epoch": 4.5069447240227065,
-      "grad_norm": 0.00841621682047844,
-      "learning_rate": 0.0001999899841709143,
-      "loss": 46.0,
-      "step": 27986
-    },
-    {
-      "epoch": 4.507105761101494,
-      "grad_norm": 0.00864341575652361,
-      "learning_rate": 0.00019998998345488337,
-      "loss": 46.0,
-      "step": 27987
-    },
-    {
-      "epoch": 4.5072667981802805,
-      "grad_norm": 0.0041961087845265865,
-      "learning_rate": 0.00019998998273882684,
-      "loss": 46.0,
-      "step": 27988
-    },
-    {
-      "epoch": 4.507427835259069,
-      "grad_norm": 0.0029167085886001587,
-      "learning_rate": 0.0001999899820227447,
-      "loss": 46.0,
-      "step": 27989
-    },
-    {
-      "epoch": 4.507588872337855,
-      "grad_norm": 0.002331846859306097,
-      "learning_rate": 0.00019998998130663697,
-      "loss": 46.0,
-      "step": 27990
-    },
-    {
-      "epoch": 4.507749909416643,
-      "grad_norm": 0.0008793384185992181,
-      "learning_rate": 0.00019998998059050368,
-      "loss": 46.0,
-      "step": 27991
-    },
-    {
-      "epoch": 4.50791094649543,
-      "grad_norm": 0.004005909897387028,
-      "learning_rate": 0.00019998997987434475,
-      "loss": 46.0,
-      "step": 27992
-    },
-    {
-      "epoch": 4.508071983574218,
-      "grad_norm": 0.0031180153600871563,
-      "learning_rate": 0.00019998997915816026,
-      "loss": 46.0,
-      "step": 27993
-    },
-    {
-      "epoch": 4.508233020653005,
-      "grad_norm": 0.004932167008519173,
-      "learning_rate": 0.00019998997844195015,
-      "loss": 46.0,
-      "step": 27994
-    },
-    {
-      "epoch": 4.508394057731793,
-      "grad_norm": 0.0017057282384485006,
-      "learning_rate": 0.00019998997772571446,
-      "loss": 46.0,
-      "step": 27995
-    },
-    {
-      "epoch": 4.50855509481058,
-      "grad_norm": 0.007572742644697428,
-      "learning_rate": 0.00019998997700945318,
-      "loss": 46.0,
-      "step": 27996
-    },
-    {
-      "epoch": 4.508716131889368,
-      "grad_norm": 0.004785297904163599,
-      "learning_rate": 0.00019998997629316635,
-      "loss": 46.0,
-      "step": 27997
-    },
-    {
-      "epoch": 4.508877168968155,
-      "grad_norm": 0.0014636771520599723,
-      "learning_rate": 0.00019998997557685386,
-      "loss": 46.0,
-      "step": 27998
-    },
-    {
-      "epoch": 4.5090382060469425,
-      "grad_norm": 0.0012036453699693084,
-      "learning_rate": 0.0001999899748605158,
-      "loss": 46.0,
-      "step": 27999
-    },
-    {
-      "epoch": 4.50919924312573,
-      "grad_norm": 0.001644710311666131,
-      "learning_rate": 0.00019998997414415217,
-      "loss": 46.0,
-      "step": 28000
-    },
-    {
-      "epoch": 4.509360280204517,
-      "grad_norm": 0.003134592669084668,
-      "learning_rate": 0.0001999899734277629,
-      "loss": 46.0,
-      "step": 28001
-    },
-    {
-      "epoch": 4.509521317283305,
-      "grad_norm": 0.0018005961319431663,
-      "learning_rate": 0.0001999899727113481,
-      "loss": 46.0,
-      "step": 28002
-    },
-    {
-      "epoch": 4.509682354362091,
-      "grad_norm": 0.008437184616923332,
-      "learning_rate": 0.00019998997199490768,
-      "loss": 46.0,
-      "step": 28003
-    },
-    {
-      "epoch": 4.50984339144088,
-      "grad_norm": 0.003125728340819478,
-      "learning_rate": 0.00019998997127844165,
-      "loss": 46.0,
-      "step": 28004
-    },
-    {
-      "epoch": 4.510004428519666,
-      "grad_norm": 0.0017466152785345912,
-      "learning_rate": 0.00019998997056195004,
-      "loss": 46.0,
-      "step": 28005
-    },
-    {
-      "epoch": 4.510165465598454,
-      "grad_norm": 0.0020630578510463238,
-      "learning_rate": 0.00019998996984543283,
-      "loss": 46.0,
-      "step": 28006
-    },
-    {
-      "epoch": 4.510326502677241,
-      "grad_norm": 0.003983749076724052,
-      "learning_rate": 0.00019998996912889007,
-      "loss": 46.0,
-      "step": 28007
-    },
-    {
-      "epoch": 4.510487539756029,
-      "grad_norm": 0.009786240756511688,
-      "learning_rate": 0.00019998996841232166,
-      "loss": 46.0,
-      "step": 28008
-    },
-    {
-      "epoch": 4.510648576834816,
-      "grad_norm": 0.00785856880247593,
-      "learning_rate": 0.00019998996769572772,
-      "loss": 46.0,
-      "step": 28009
-    },
-    {
-      "epoch": 4.510809613913604,
-      "grad_norm": 0.004561574198305607,
-      "learning_rate": 0.0001999899669791081,
-      "loss": 46.0,
-      "step": 28010
-    },
-    {
-      "epoch": 4.510970650992391,
-      "grad_norm": 0.00855952873826027,
-      "learning_rate": 0.00019998996626246295,
-      "loss": 46.0,
-      "step": 28011
-    },
-    {
-      "epoch": 4.5111316880711785,
-      "grad_norm": 0.002436211798340082,
-      "learning_rate": 0.00019998996554579222,
-      "loss": 46.0,
-      "step": 28012
-    },
-    {
-      "epoch": 4.511292725149966,
-      "grad_norm": 0.0026596460957080126,
-      "learning_rate": 0.00019998996482909588,
-      "loss": 46.0,
-      "step": 28013
-    },
-    {
-      "epoch": 4.511453762228753,
-      "grad_norm": 0.001462169224396348,
-      "learning_rate": 0.00019998996411237392,
-      "loss": 46.0,
-      "step": 28014
-    },
-    {
-      "epoch": 4.511614799307541,
-      "grad_norm": 0.0017911311006173491,
-      "learning_rate": 0.0001999899633956264,
-      "loss": 46.0,
-      "step": 28015
-    },
-    {
-      "epoch": 4.511775836386328,
-      "grad_norm": 0.026311425492167473,
-      "learning_rate": 0.00019998996267885327,
-      "loss": 46.0,
-      "step": 28016
-    },
-    {
-      "epoch": 4.511936873465116,
-      "grad_norm": 0.0023388275876641273,
-      "learning_rate": 0.00019998996196205455,
-      "loss": 46.0,
-      "step": 28017
-    },
-    {
-      "epoch": 4.512097910543902,
-      "grad_norm": 0.0042266868986189365,
-      "learning_rate": 0.00019998996124523025,
-      "loss": 46.0,
-      "step": 28018
-    },
-    {
-      "epoch": 4.51225894762269,
-      "grad_norm": 0.005381749011576176,
-      "learning_rate": 0.00019998996052838036,
-      "loss": 46.0,
-      "step": 28019
-    },
-    {
-      "epoch": 4.512419984701477,
-      "grad_norm": 0.006414171773940325,
-      "learning_rate": 0.00019998995981150485,
-      "loss": 46.0,
-      "step": 28020
-    },
-    {
-      "epoch": 4.512581021780265,
-      "grad_norm": 0.00540187768638134,
-      "learning_rate": 0.00019998995909460376,
-      "loss": 46.0,
-      "step": 28021
-    },
-    {
-      "epoch": 4.512742058859052,
-      "grad_norm": 0.004719736985862255,
-      "learning_rate": 0.00019998995837767708,
-      "loss": 46.0,
-      "step": 28022
-    },
-    {
-      "epoch": 4.51290309593784,
-      "grad_norm": 0.0017974013462662697,
-      "learning_rate": 0.00019998995766072484,
-      "loss": 46.0,
-      "step": 28023
-    },
-    {
-      "epoch": 4.513064133016627,
-      "grad_norm": 0.002993357367813587,
-      "learning_rate": 0.00019998995694374695,
-      "loss": 46.0,
-      "step": 28024
-    },
-    {
-      "epoch": 4.5132251700954145,
-      "grad_norm": 0.005367578938603401,
-      "learning_rate": 0.0001999899562267435,
-      "loss": 46.0,
-      "step": 28025
-    },
-    {
-      "epoch": 4.513386207174202,
-      "grad_norm": 0.007388588506728411,
-      "learning_rate": 0.00019998995550971445,
-      "loss": 46.0,
-      "step": 28026
-    },
-    {
-      "epoch": 4.513547244252989,
-      "grad_norm": 0.011032975278794765,
-      "learning_rate": 0.0001999899547926598,
-      "loss": 46.0,
-      "step": 28027
-    },
-    {
-      "epoch": 4.513708281331777,
-      "grad_norm": 0.00642789714038372,
-      "learning_rate": 0.00019998995407557958,
-      "loss": 46.0,
-      "step": 28028
-    },
-    {
-      "epoch": 4.513869318410564,
-      "grad_norm": 0.007025600876659155,
-      "learning_rate": 0.00019998995335847376,
-      "loss": 46.0,
-      "step": 28029
-    },
-    {
-      "epoch": 4.514030355489352,
-      "grad_norm": 0.0006311052129603922,
-      "learning_rate": 0.00019998995264134236,
-      "loss": 46.0,
-      "step": 28030
-    },
-    {
-      "epoch": 4.514191392568139,
-      "grad_norm": 0.009003935381770134,
-      "learning_rate": 0.00019998995192418534,
-      "loss": 46.0,
-      "step": 28031
-    },
-    {
-      "epoch": 4.514352429646927,
-      "grad_norm": 0.01081156637519598,
-      "learning_rate": 0.00019998995120700276,
-      "loss": 46.0,
-      "step": 28032
-    },
-    {
-      "epoch": 4.514513466725713,
-      "grad_norm": 0.005827121902257204,
-      "learning_rate": 0.00019998995048979454,
-      "loss": 46.0,
-      "step": 28033
-    },
-    {
-      "epoch": 4.514674503804501,
-      "grad_norm": 0.0017113096546381712,
-      "learning_rate": 0.00019998994977256076,
-      "loss": 46.0,
-      "step": 28034
-    },
-    {
-      "epoch": 4.514835540883288,
-      "grad_norm": 0.003421470522880554,
-      "learning_rate": 0.0001999899490553014,
-      "loss": 46.0,
-      "step": 28035
-    },
-    {
-      "epoch": 4.5149965779620755,
-      "grad_norm": 0.0013876743614673615,
-      "learning_rate": 0.0001999899483380164,
-      "loss": 46.0,
-      "step": 28036
-    },
-    {
-      "epoch": 4.515157615040863,
-      "grad_norm": 0.0020678043365478516,
-      "learning_rate": 0.00019998994762070583,
-      "loss": 46.0,
-      "step": 28037
-    },
-    {
-      "epoch": 4.5153186521196504,
-      "grad_norm": 0.0021930222865194082,
-      "learning_rate": 0.0001999899469033697,
-      "loss": 46.0,
-      "step": 28038
-    },
-    {
-      "epoch": 4.515479689198438,
-      "grad_norm": 0.002599215367808938,
-      "learning_rate": 0.00019998994618600796,
-      "loss": 46.0,
-      "step": 28039
-    },
-    {
-      "epoch": 4.515640726277225,
-      "grad_norm": 0.01092553324997425,
-      "learning_rate": 0.00019998994546862063,
-      "loss": 46.0,
-      "step": 28040
-    },
-    {
-      "epoch": 4.515801763356013,
-      "grad_norm": 0.005444251466542482,
-      "learning_rate": 0.00019998994475120768,
-      "loss": 46.0,
-      "step": 28041
-    },
-    {
-      "epoch": 4.5159628004348,
-      "grad_norm": 0.007895600982010365,
-      "learning_rate": 0.00019998994403376915,
-      "loss": 46.0,
-      "step": 28042
-    },
-    {
-      "epoch": 4.516123837513588,
-      "grad_norm": 0.0010517495684325695,
-      "learning_rate": 0.00019998994331630506,
-      "loss": 46.0,
-      "step": 28043
-    },
-    {
-      "epoch": 4.516284874592375,
-      "grad_norm": 0.011224798858165741,
-      "learning_rate": 0.00019998994259881535,
-      "loss": 46.0,
-      "step": 28044
-    },
-    {
-      "epoch": 4.516445911671163,
-      "grad_norm": 0.007543615065515041,
-      "learning_rate": 0.00019998994188130006,
-      "loss": 46.0,
-      "step": 28045
-    },
-    {
-      "epoch": 4.51660694874995,
-      "grad_norm": 0.003957655280828476,
-      "learning_rate": 0.00019998994116375915,
-      "loss": 46.0,
-      "step": 28046
-    },
-    {
-      "epoch": 4.5167679858287375,
-      "grad_norm": 0.0025699941907078028,
-      "learning_rate": 0.00019998994044619268,
-      "loss": 46.0,
-      "step": 28047
-    },
-    {
-      "epoch": 4.516929022907524,
-      "grad_norm": 0.009581594727933407,
-      "learning_rate": 0.0001999899397286006,
-      "loss": 46.0,
-      "step": 28048
-    },
-    {
-      "epoch": 4.5170900599863115,
-      "grad_norm": 0.0027826293371617794,
-      "learning_rate": 0.0001999899390109829,
-      "loss": 46.0,
-      "step": 28049
-    },
-    {
-      "epoch": 4.517251097065099,
-      "grad_norm": 0.0009113684063777328,
-      "learning_rate": 0.00019998993829333967,
-      "loss": 46.0,
-      "step": 28050
-    },
-    {
-      "epoch": 4.517412134143886,
-      "grad_norm": 0.0033036170061677694,
-      "learning_rate": 0.00019998993757567083,
-      "loss": 46.0,
-      "step": 28051
-    },
-    {
-      "epoch": 4.517573171222674,
-      "grad_norm": 0.002945908345282078,
-      "learning_rate": 0.00019998993685797637,
-      "loss": 46.0,
-      "step": 28052
-    },
-    {
-      "epoch": 4.517734208301461,
-      "grad_norm": 0.006020206492394209,
-      "learning_rate": 0.00019998993614025632,
-      "loss": 46.0,
-      "step": 28053
-    },
-    {
-      "epoch": 4.517895245380249,
-      "grad_norm": 0.0065395161509513855,
-      "learning_rate": 0.0001999899354225107,
-      "loss": 46.0,
-      "step": 28054
-    },
-    {
-      "epoch": 4.518056282459036,
-      "grad_norm": 0.0010534306056797504,
-      "learning_rate": 0.00019998993470473947,
-      "loss": 46.0,
-      "step": 28055
-    },
-    {
-      "epoch": 4.518217319537824,
-      "grad_norm": 0.003093145089223981,
-      "learning_rate": 0.00019998993398694266,
-      "loss": 46.0,
-      "step": 28056
-    },
-    {
-      "epoch": 4.518378356616611,
-      "grad_norm": 0.0070256819017231464,
-      "learning_rate": 0.00019998993326912024,
-      "loss": 46.0,
-      "step": 28057
-    },
-    {
-      "epoch": 4.518539393695399,
-      "grad_norm": 0.0031234880443662405,
-      "learning_rate": 0.00019998993255127223,
-      "loss": 46.0,
-      "step": 28058
-    },
-    {
-      "epoch": 4.518700430774186,
-      "grad_norm": 0.004970559850335121,
-      "learning_rate": 0.00019998993183339864,
-      "loss": 46.0,
-      "step": 28059
-    },
-    {
-      "epoch": 4.5188614678529735,
-      "grad_norm": 0.0023421715013682842,
-      "learning_rate": 0.00019998993111549948,
-      "loss": 46.0,
-      "step": 28060
-    },
-    {
-      "epoch": 4.51902250493176,
-      "grad_norm": 0.003251435933634639,
-      "learning_rate": 0.0001999899303975747,
-      "loss": 46.0,
-      "step": 28061
-    },
-    {
-      "epoch": 4.519183542010548,
-      "grad_norm": 0.005180378910154104,
-      "learning_rate": 0.00019998992967962433,
-      "loss": 46.0,
-      "step": 28062
-    },
-    {
-      "epoch": 4.519344579089335,
-      "grad_norm": 0.002574390731751919,
-      "learning_rate": 0.00019998992896164835,
-      "loss": 46.0,
-      "step": 28063
-    },
-    {
-      "epoch": 4.519505616168122,
-      "grad_norm": 0.008656046353280544,
-      "learning_rate": 0.00019998992824364682,
-      "loss": 46.0,
-      "step": 28064
-    },
-    {
-      "epoch": 4.51966665324691,
-      "grad_norm": 0.00193415442481637,
-      "learning_rate": 0.00019998992752561968,
-      "loss": 46.0,
-      "step": 28065
-    },
-    {
-      "epoch": 4.519827690325697,
-      "grad_norm": 0.014596354216337204,
-      "learning_rate": 0.00019998992680756694,
-      "loss": 46.0,
-      "step": 28066
-    },
-    {
-      "epoch": 4.519988727404485,
-      "grad_norm": 0.005065843928605318,
-      "learning_rate": 0.0001999899260894886,
-      "loss": 46.0,
-      "step": 28067
-    },
-    {
-      "epoch": 4.520149764483272,
-      "grad_norm": 0.0014835710171610117,
-      "learning_rate": 0.00019998992537138466,
-      "loss": 46.0,
-      "step": 28068
-    },
-    {
-      "epoch": 4.52031080156206,
-      "grad_norm": 0.0024256010074168444,
-      "learning_rate": 0.00019998992465325514,
-      "loss": 46.0,
-      "step": 28069
-    },
-    {
-      "epoch": 4.520471838640847,
-      "grad_norm": 0.003363888943567872,
-      "learning_rate": 0.00019998992393510006,
-      "loss": 46.0,
-      "step": 28070
-    },
-    {
-      "epoch": 4.520632875719635,
-      "grad_norm": 0.01180997584015131,
-      "learning_rate": 0.00019998992321691933,
-      "loss": 46.0,
-      "step": 28071
-    },
-    {
-      "epoch": 4.520793912798422,
-      "grad_norm": 0.004813733045011759,
-      "learning_rate": 0.00019998992249871305,
-      "loss": 46.0,
-      "step": 28072
-    },
-    {
-      "epoch": 4.5209549498772095,
-      "grad_norm": 0.01116902381181717,
-      "learning_rate": 0.00019998992178048118,
-      "loss": 46.0,
-      "step": 28073
-    },
-    {
-      "epoch": 4.521115986955997,
-      "grad_norm": 0.0022108505945652723,
-      "learning_rate": 0.0001999899210622237,
-      "loss": 46.0,
-      "step": 28074
-    },
-    {
-      "epoch": 4.521277024034784,
-      "grad_norm": 0.0007912889122962952,
-      "learning_rate": 0.00019998992034394065,
-      "loss": 46.0,
-      "step": 28075
-    },
-    {
-      "epoch": 4.521438061113571,
-      "grad_norm": 0.006706543732434511,
-      "learning_rate": 0.000199989919625632,
-      "loss": 46.0,
-      "step": 28076
-    },
-    {
-      "epoch": 4.521599098192359,
-      "grad_norm": 0.003491014242172241,
-      "learning_rate": 0.0001999899189072977,
-      "loss": 46.0,
-      "step": 28077
-    },
-    {
-      "epoch": 4.521760135271146,
-      "grad_norm": 0.0022573666647076607,
-      "learning_rate": 0.00019998991818893786,
-      "loss": 46.0,
-      "step": 28078
-    },
-    {
-      "epoch": 4.521921172349933,
-      "grad_norm": 0.004572810139507055,
-      "learning_rate": 0.00019998991747055244,
-      "loss": 46.0,
-      "step": 28079
-    },
-    {
-      "epoch": 4.522082209428721,
-      "grad_norm": 0.0068563357926905155,
-      "learning_rate": 0.0001999899167521414,
-      "loss": 46.0,
-      "step": 28080
-    },
-    {
-      "epoch": 4.522243246507508,
-      "grad_norm": 0.0007289955392479897,
-      "learning_rate": 0.00019998991603370478,
-      "loss": 46.0,
-      "step": 28081
-    },
-    {
-      "epoch": 4.522404283586296,
-      "grad_norm": 0.0036731078289449215,
-      "learning_rate": 0.00019998991531524257,
-      "loss": 46.0,
-      "step": 28082
-    },
-    {
-      "epoch": 4.522565320665083,
-      "grad_norm": 0.006915264762938023,
-      "learning_rate": 0.00019998991459675475,
-      "loss": 46.0,
-      "step": 28083
-    },
-    {
-      "epoch": 4.5227263577438706,
-      "grad_norm": 0.01765129156410694,
-      "learning_rate": 0.00019998991387824136,
-      "loss": 46.0,
-      "step": 28084
-    },
-    {
-      "epoch": 4.522887394822658,
-      "grad_norm": 0.004520501010119915,
-      "learning_rate": 0.00019998991315970237,
-      "loss": 46.0,
-      "step": 28085
-    },
-    {
-      "epoch": 4.5230484319014455,
-      "grad_norm": 0.002556120976805687,
-      "learning_rate": 0.00019998991244113778,
-      "loss": 46.0,
-      "step": 28086
-    },
-    {
-      "epoch": 4.523209468980233,
-      "grad_norm": 0.0046248831786215305,
-      "learning_rate": 0.0001999899117225476,
-      "loss": 46.0,
-      "step": 28087
-    },
-    {
-      "epoch": 4.52337050605902,
-      "grad_norm": 0.005944658536463976,
-      "learning_rate": 0.00019998991100393182,
-      "loss": 46.0,
-      "step": 28088
-    },
-    {
-      "epoch": 4.523531543137808,
-      "grad_norm": 0.0027721503283828497,
-      "learning_rate": 0.00019998991028529045,
-      "loss": 46.0,
-      "step": 28089
-    },
-    {
-      "epoch": 4.523692580216595,
-      "grad_norm": 0.0020460374653339386,
-      "learning_rate": 0.00019998990956662352,
-      "loss": 46.0,
-      "step": 28090
-    },
-    {
-      "epoch": 4.523853617295382,
-      "grad_norm": 0.0018385767471045256,
-      "learning_rate": 0.000199989908847931,
-      "loss": 46.0,
-      "step": 28091
-    },
-    {
-      "epoch": 4.524014654374169,
-      "grad_norm": 0.0014394954778254032,
-      "learning_rate": 0.00019998990812921283,
-      "loss": 46.0,
-      "step": 28092
-    },
-    {
-      "epoch": 4.524175691452957,
-      "grad_norm": 0.007078861817717552,
-      "learning_rate": 0.0001999899074104691,
-      "loss": 46.0,
-      "step": 28093
-    },
-    {
-      "epoch": 4.524336728531744,
-      "grad_norm": 0.005528643261641264,
-      "learning_rate": 0.0001999899066916998,
-      "loss": 46.0,
-      "step": 28094
-    },
-    {
-      "epoch": 4.524497765610532,
-      "grad_norm": 0.0033406035508960485,
-      "learning_rate": 0.00019998990597290488,
-      "loss": 46.0,
-      "step": 28095
-    },
-    {
-      "epoch": 4.524658802689319,
-      "grad_norm": 0.0012600012123584747,
-      "learning_rate": 0.00019998990525408437,
-      "loss": 46.0,
-      "step": 28096
-    },
-    {
-      "epoch": 4.5248198397681065,
-      "grad_norm": 0.0047652460634708405,
-      "learning_rate": 0.00019998990453523827,
-      "loss": 46.0,
-      "step": 28097
-    },
-    {
-      "epoch": 4.524980876846894,
-      "grad_norm": 0.0027372546028345823,
-      "learning_rate": 0.00019998990381636658,
-      "loss": 46.0,
-      "step": 28098
-    },
-    {
-      "epoch": 4.525141913925681,
-      "grad_norm": 0.001851135166361928,
-      "learning_rate": 0.00019998990309746928,
-      "loss": 46.0,
-      "step": 28099
-    },
-    {
-      "epoch": 4.525302951004469,
-      "grad_norm": 0.0025847184006124735,
-      "learning_rate": 0.0001999899023785464,
-      "loss": 46.0,
-      "step": 28100
-    },
-    {
-      "epoch": 4.525463988083256,
-      "grad_norm": 0.01216368842869997,
-      "learning_rate": 0.00019998990165959795,
-      "loss": 46.0,
-      "step": 28101
-    },
-    {
-      "epoch": 4.525625025162044,
-      "grad_norm": 0.006062064785510302,
-      "learning_rate": 0.0001999899009406239,
-      "loss": 46.0,
-      "step": 28102
-    },
-    {
-      "epoch": 4.525786062240831,
-      "grad_norm": 0.011608520522713661,
-      "learning_rate": 0.00019998990022162424,
-      "loss": 46.0,
-      "step": 28103
-    },
-    {
-      "epoch": 4.525947099319619,
-      "grad_norm": 0.0017695939168334007,
-      "learning_rate": 0.00019998989950259898,
-      "loss": 46.0,
-      "step": 28104
-    },
-    {
-      "epoch": 4.526108136398406,
-      "grad_norm": 0.003787803230807185,
-      "learning_rate": 0.00019998989878354816,
-      "loss": 46.0,
-      "step": 28105
-    },
-    {
-      "epoch": 4.526269173477193,
-      "grad_norm": 0.003946825861930847,
-      "learning_rate": 0.00019998989806447172,
-      "loss": 46.0,
-      "step": 28106
-    },
-    {
-      "epoch": 4.52643021055598,
-      "grad_norm": 0.0008840218069963157,
-      "learning_rate": 0.0001999898973453697,
-      "loss": 46.0,
-      "step": 28107
-    },
-    {
-      "epoch": 4.526591247634768,
-      "grad_norm": 0.004806368611752987,
-      "learning_rate": 0.00019998989662624208,
-      "loss": 46.0,
-      "step": 28108
-    },
-    {
-      "epoch": 4.526752284713555,
-      "grad_norm": 0.010537315160036087,
-      "learning_rate": 0.00019998989590708888,
-      "loss": 46.0,
-      "step": 28109
-    },
-    {
-      "epoch": 4.5269133217923425,
-      "grad_norm": 0.005134051665663719,
-      "learning_rate": 0.0001999898951879101,
-      "loss": 46.0,
-      "step": 28110
-    },
-    {
-      "epoch": 4.52707435887113,
-      "grad_norm": 0.018456293269991875,
-      "learning_rate": 0.0001999898944687057,
-      "loss": 46.0,
-      "step": 28111
-    },
-    {
-      "epoch": 4.527235395949917,
-      "grad_norm": 0.007777049671858549,
-      "learning_rate": 0.00019998989374947569,
-      "loss": 46.0,
-      "step": 28112
-    },
-    {
-      "epoch": 4.527396433028705,
-      "grad_norm": 0.002604363253340125,
-      "learning_rate": 0.00019998989303022014,
-      "loss": 46.0,
-      "step": 28113
-    },
-    {
-      "epoch": 4.527557470107492,
-      "grad_norm": 0.001909418380819261,
-      "learning_rate": 0.00019998989231093898,
-      "loss": 46.0,
-      "step": 28114
-    },
-    {
-      "epoch": 4.52771850718628,
-      "grad_norm": 0.011187337338924408,
-      "learning_rate": 0.0001999898915916322,
-      "loss": 46.0,
-      "step": 28115
-    },
-    {
-      "epoch": 4.527879544265067,
-      "grad_norm": 0.0013596827629953623,
-      "learning_rate": 0.00019998989087229987,
-      "loss": 46.0,
-      "step": 28116
-    },
-    {
-      "epoch": 4.528040581343855,
-      "grad_norm": 0.012069943360984325,
-      "learning_rate": 0.00019998989015294192,
-      "loss": 46.0,
-      "step": 28117
-    },
-    {
-      "epoch": 4.528201618422642,
-      "grad_norm": 0.02140405774116516,
-      "learning_rate": 0.00019998988943355838,
-      "loss": 46.0,
-      "step": 28118
-    },
-    {
-      "epoch": 4.52836265550143,
-      "grad_norm": 0.0013159465743228793,
-      "learning_rate": 0.00019998988871414923,
-      "loss": 46.0,
-      "step": 28119
-    },
-    {
-      "epoch": 4.528523692580217,
-      "grad_norm": 0.002588863717392087,
-      "learning_rate": 0.00019998988799471452,
-      "loss": 46.0,
-      "step": 28120
-    },
-    {
-      "epoch": 4.528684729659004,
-      "grad_norm": 0.005136772524565458,
-      "learning_rate": 0.00019998988727525422,
-      "loss": 46.0,
-      "step": 28121
-    },
-    {
-      "epoch": 4.528845766737791,
-      "grad_norm": 0.017035972326993942,
-      "learning_rate": 0.0001999898865557683,
-      "loss": 46.0,
-      "step": 28122
-    },
-    {
-      "epoch": 4.5290068038165785,
-      "grad_norm": 0.0044188774190843105,
-      "learning_rate": 0.00019998988583625683,
-      "loss": 46.0,
-      "step": 28123
-    },
-    {
-      "epoch": 4.529167840895366,
-      "grad_norm": 0.002522155875340104,
-      "learning_rate": 0.00019998988511671972,
-      "loss": 46.0,
-      "step": 28124
-    },
-    {
-      "epoch": 4.529328877974153,
-      "grad_norm": 0.006839301902800798,
-      "learning_rate": 0.00019998988439715702,
-      "loss": 46.0,
-      "step": 28125
-    },
-    {
-      "epoch": 4.529489915052941,
-      "grad_norm": 0.004980405792593956,
-      "learning_rate": 0.00019998988367756875,
-      "loss": 46.0,
-      "step": 28126
-    },
-    {
-      "epoch": 4.529650952131728,
-      "grad_norm": 0.004197351634502411,
-      "learning_rate": 0.00019998988295795488,
-      "loss": 46.0,
-      "step": 28127
-    },
-    {
-      "epoch": 4.529811989210516,
-      "grad_norm": 0.017260542139410973,
-      "learning_rate": 0.00019998988223831544,
-      "loss": 46.0,
-      "step": 28128
-    },
-    {
-      "epoch": 4.529973026289303,
-      "grad_norm": 0.0015277242055162787,
-      "learning_rate": 0.00019998988151865036,
-      "loss": 46.0,
-      "step": 28129
-    },
-    {
-      "epoch": 4.530134063368091,
-      "grad_norm": 0.00429127924144268,
-      "learning_rate": 0.00019998988079895973,
-      "loss": 46.0,
-      "step": 28130
-    },
-    {
-      "epoch": 4.530295100446878,
-      "grad_norm": 0.005581240635365248,
-      "learning_rate": 0.0001999898800792435,
-      "loss": 46.0,
-      "step": 28131
-    },
-    {
-      "epoch": 4.530456137525666,
-      "grad_norm": 0.0016650476027280092,
-      "learning_rate": 0.00019998987935950166,
-      "loss": 46.0,
-      "step": 28132
-    },
-    {
-      "epoch": 4.530617174604453,
-      "grad_norm": 0.011678727343678474,
-      "learning_rate": 0.00019998987863973424,
-      "loss": 46.0,
-      "step": 28133
-    },
-    {
-      "epoch": 4.53077821168324,
-      "grad_norm": 0.001121566048823297,
-      "learning_rate": 0.00019998987791994122,
-      "loss": 46.0,
-      "step": 28134
-    },
-    {
-      "epoch": 4.530939248762028,
-      "grad_norm": 0.0030708073172718287,
-      "learning_rate": 0.0001999898772001226,
-      "loss": 46.0,
-      "step": 28135
-    },
-    {
-      "epoch": 4.5311002858408145,
-      "grad_norm": 0.0036146959755569696,
-      "learning_rate": 0.00019998987648027844,
-      "loss": 46.0,
-      "step": 28136
-    },
-    {
-      "epoch": 4.531261322919602,
-      "grad_norm": 0.01375151239335537,
-      "learning_rate": 0.00019998987576040864,
-      "loss": 46.0,
-      "step": 28137
-    },
-    {
-      "epoch": 4.531422359998389,
-      "grad_norm": 0.008395235054194927,
-      "learning_rate": 0.00019998987504051325,
-      "loss": 46.0,
-      "step": 28138
-    },
-    {
-      "epoch": 4.531583397077177,
-      "grad_norm": 0.0024923905730247498,
-      "learning_rate": 0.00019998987432059227,
-      "loss": 46.0,
-      "step": 28139
-    },
-    {
-      "epoch": 4.531744434155964,
-      "grad_norm": 0.008777826093137264,
-      "learning_rate": 0.0001999898736006457,
-      "loss": 46.0,
-      "step": 28140
-    },
-    {
-      "epoch": 4.531905471234752,
-      "grad_norm": 0.0028207488358020782,
-      "learning_rate": 0.00019998987288067356,
-      "loss": 46.0,
-      "step": 28141
-    },
-    {
-      "epoch": 4.532066508313539,
-      "grad_norm": 0.005800921469926834,
-      "learning_rate": 0.0001999898721606758,
-      "loss": 46.0,
-      "step": 28142
-    },
-    {
-      "epoch": 4.532227545392327,
-      "grad_norm": 0.0027123247273266315,
-      "learning_rate": 0.00019998987144065247,
-      "loss": 46.0,
-      "step": 28143
-    },
-    {
-      "epoch": 4.532388582471114,
-      "grad_norm": 0.0012545640347525477,
-      "learning_rate": 0.0001999898707206035,
-      "loss": 46.0,
-      "step": 28144
-    },
-    {
-      "epoch": 4.5325496195499015,
-      "grad_norm": 0.002751567168161273,
-      "learning_rate": 0.00019998987000052897,
-      "loss": 46.0,
-      "step": 28145
-    },
-    {
-      "epoch": 4.532710656628689,
-      "grad_norm": 0.00078404980013147,
-      "learning_rate": 0.00019998986928042886,
-      "loss": 46.0,
-      "step": 28146
-    },
-    {
-      "epoch": 4.5328716937074764,
-      "grad_norm": 0.005834897048771381,
-      "learning_rate": 0.00019998986856030316,
-      "loss": 46.0,
-      "step": 28147
-    },
-    {
-      "epoch": 4.533032730786264,
-      "grad_norm": 0.004490547347813845,
-      "learning_rate": 0.00019998986784015185,
-      "loss": 46.0,
-      "step": 28148
-    },
-    {
-      "epoch": 4.5331937678650505,
-      "grad_norm": 0.004180487710982561,
-      "learning_rate": 0.00019998986711997497,
-      "loss": 46.0,
-      "step": 28149
-    },
-    {
-      "epoch": 4.533354804943839,
-      "grad_norm": 0.008809946477413177,
-      "learning_rate": 0.00019998986639977245,
-      "loss": 46.0,
-      "step": 28150
-    },
-    {
-      "epoch": 4.533515842022625,
-      "grad_norm": 0.012196926400065422,
-      "learning_rate": 0.00019998986567954438,
-      "loss": 46.0,
-      "step": 28151
-    },
-    {
-      "epoch": 4.533676879101413,
-      "grad_norm": 0.0025184727273881435,
-      "learning_rate": 0.00019998986495929071,
-      "loss": 46.0,
-      "step": 28152
-    },
-    {
-      "epoch": 4.5338379161802,
-      "grad_norm": 0.0071161356754601,
-      "learning_rate": 0.00019998986423901144,
-      "loss": 46.0,
-      "step": 28153
-    },
-    {
-      "epoch": 4.533998953258988,
-      "grad_norm": 0.0047623696736991405,
-      "learning_rate": 0.00019998986351870657,
-      "loss": 46.0,
-      "step": 28154
-    },
-    {
-      "epoch": 4.534159990337775,
-      "grad_norm": 0.007362185511738062,
-      "learning_rate": 0.00019998986279837612,
-      "loss": 46.0,
-      "step": 28155
-    },
-    {
-      "epoch": 4.534321027416563,
-      "grad_norm": 0.004340854473412037,
-      "learning_rate": 0.00019998986207802008,
-      "loss": 46.0,
-      "step": 28156
-    },
-    {
-      "epoch": 4.53448206449535,
-      "grad_norm": 0.0020779420156031847,
-      "learning_rate": 0.00019998986135763845,
-      "loss": 46.0,
-      "step": 28157
-    },
-    {
-      "epoch": 4.5346431015741375,
-      "grad_norm": 0.00365415564738214,
-      "learning_rate": 0.0001999898606372312,
-      "loss": 46.0,
-      "step": 28158
-    },
-    {
-      "epoch": 4.534804138652925,
-      "grad_norm": 0.004089410416781902,
-      "learning_rate": 0.0001999898599167984,
-      "loss": 46.0,
-      "step": 28159
-    },
-    {
-      "epoch": 4.534965175731712,
-      "grad_norm": 0.006561561953276396,
-      "learning_rate": 0.00019998985919634,
-      "loss": 46.0,
-      "step": 28160
-    },
-    {
-      "epoch": 4.5351262128105,
-      "grad_norm": 0.0030093041714280844,
-      "learning_rate": 0.000199989858475856,
-      "loss": 46.0,
-      "step": 28161
-    },
-    {
-      "epoch": 4.535287249889287,
-      "grad_norm": 0.007026029285043478,
-      "learning_rate": 0.00019998985775534638,
-      "loss": 46.0,
-      "step": 28162
-    },
-    {
-      "epoch": 4.535448286968075,
-      "grad_norm": 0.0024453147780150175,
-      "learning_rate": 0.0001999898570348112,
-      "loss": 46.0,
-      "step": 28163
-    },
-    {
-      "epoch": 4.535609324046861,
-      "grad_norm": 0.0025690484326332808,
-      "learning_rate": 0.0001999898563142504,
-      "loss": 46.0,
-      "step": 28164
-    },
-    {
-      "epoch": 4.535770361125649,
-      "grad_norm": 0.002705293009057641,
-      "learning_rate": 0.00019998985559366403,
-      "loss": 46.0,
-      "step": 28165
-    },
-    {
-      "epoch": 4.535931398204436,
-      "grad_norm": 0.0009811506606638432,
-      "learning_rate": 0.00019998985487305207,
-      "loss": 46.0,
-      "step": 28166
-    },
-    {
-      "epoch": 4.536092435283224,
-      "grad_norm": 0.0018293160246685147,
-      "learning_rate": 0.00019998985415241451,
-      "loss": 46.0,
-      "step": 28167
-    },
-    {
-      "epoch": 4.536253472362011,
-      "grad_norm": 0.003863105084747076,
-      "learning_rate": 0.00019998985343175137,
-      "loss": 46.0,
-      "step": 28168
-    },
-    {
-      "epoch": 4.536414509440799,
-      "grad_norm": 0.0053612226620316505,
-      "learning_rate": 0.00019998985271106262,
-      "loss": 46.0,
-      "step": 28169
-    },
-    {
-      "epoch": 4.536575546519586,
-      "grad_norm": 0.01269327849149704,
-      "learning_rate": 0.0001999898519903483,
-      "loss": 46.0,
-      "step": 28170
-    },
-    {
-      "epoch": 4.5367365835983735,
-      "grad_norm": 0.002893842291086912,
-      "learning_rate": 0.00019998985126960835,
-      "loss": 46.0,
-      "step": 28171
-    },
-    {
-      "epoch": 4.536897620677161,
-      "grad_norm": 0.0032377797178924084,
-      "learning_rate": 0.00019998985054884286,
-      "loss": 46.0,
-      "step": 28172
-    },
-    {
-      "epoch": 4.537058657755948,
-      "grad_norm": 0.0026010910514742136,
-      "learning_rate": 0.00019998984982805173,
-      "loss": 46.0,
-      "step": 28173
-    },
-    {
-      "epoch": 4.537219694834736,
-      "grad_norm": 0.005723660346120596,
-      "learning_rate": 0.00019998984910723504,
-      "loss": 46.0,
-      "step": 28174
-    },
-    {
-      "epoch": 4.537380731913523,
-      "grad_norm": 0.0017634622054174542,
-      "learning_rate": 0.0001999898483863927,
-      "loss": 46.0,
-      "step": 28175
-    },
-    {
-      "epoch": 4.537541768992311,
-      "grad_norm": 0.002203563926741481,
-      "learning_rate": 0.00019998984766552485,
-      "loss": 46.0,
-      "step": 28176
-    },
-    {
-      "epoch": 4.537702806071098,
-      "grad_norm": 0.005214695353060961,
-      "learning_rate": 0.00019998984694463134,
-      "loss": 46.0,
-      "step": 28177
-    },
-    {
-      "epoch": 4.537863843149886,
-      "grad_norm": 0.0027587739750742912,
-      "learning_rate": 0.00019998984622371228,
-      "loss": 46.0,
-      "step": 28178
-    },
-    {
-      "epoch": 4.538024880228672,
-      "grad_norm": 0.006451367400586605,
-      "learning_rate": 0.0001999898455027676,
-      "loss": 46.0,
-      "step": 28179
-    },
-    {
-      "epoch": 4.53818591730746,
-      "grad_norm": 0.00657782144844532,
-      "learning_rate": 0.00019998984478179736,
-      "loss": 46.0,
-      "step": 28180
-    },
-    {
-      "epoch": 4.538346954386247,
-      "grad_norm": 0.023852072656154633,
-      "learning_rate": 0.0001999898440608015,
-      "loss": 46.0,
-      "step": 28181
-    },
-    {
-      "epoch": 4.538507991465035,
-      "grad_norm": 0.002822832204401493,
-      "learning_rate": 0.0001999898433397801,
-      "loss": 46.0,
-      "step": 28182
-    },
-    {
-      "epoch": 4.538669028543822,
-      "grad_norm": 0.007189903873950243,
-      "learning_rate": 0.00019998984261873303,
-      "loss": 46.0,
-      "step": 28183
-    },
-    {
-      "epoch": 4.5388300656226095,
-      "grad_norm": 0.0042359549552202225,
-      "learning_rate": 0.00019998984189766042,
-      "loss": 46.0,
-      "step": 28184
-    },
-    {
-      "epoch": 4.538991102701397,
-      "grad_norm": 0.0004599812673404813,
-      "learning_rate": 0.0001999898411765622,
-      "loss": 46.0,
-      "step": 28185
-    },
-    {
-      "epoch": 4.539152139780184,
-      "grad_norm": 0.003450341522693634,
-      "learning_rate": 0.00019998984045543837,
-      "loss": 46.0,
-      "step": 28186
-    },
-    {
-      "epoch": 4.539313176858972,
-      "grad_norm": 0.0021386267617344856,
-      "learning_rate": 0.00019998983973428897,
-      "loss": 46.0,
-      "step": 28187
-    },
-    {
-      "epoch": 4.539474213937759,
-      "grad_norm": 0.007107846904546022,
-      "learning_rate": 0.000199989839013114,
-      "loss": 46.0,
-      "step": 28188
-    },
-    {
-      "epoch": 4.539635251016547,
-      "grad_norm": 0.0049768113531172276,
-      "learning_rate": 0.00019998983829191337,
-      "loss": 46.0,
-      "step": 28189
-    },
-    {
-      "epoch": 4.539796288095334,
-      "grad_norm": 0.004822993651032448,
-      "learning_rate": 0.00019998983757068718,
-      "loss": 46.0,
-      "step": 28190
-    },
-    {
-      "epoch": 4.539957325174122,
-      "grad_norm": 0.00548260472714901,
-      "learning_rate": 0.00019998983684943542,
-      "loss": 46.0,
-      "step": 28191
-    },
-    {
-      "epoch": 4.540118362252909,
-      "grad_norm": 0.008749093860387802,
-      "learning_rate": 0.00019998983612815808,
-      "loss": 46.0,
-      "step": 28192
-    },
-    {
-      "epoch": 4.5402793993316966,
-      "grad_norm": 0.003197180340066552,
-      "learning_rate": 0.0001999898354068551,
-      "loss": 46.0,
-      "step": 28193
-    },
-    {
-      "epoch": 4.540440436410483,
-      "grad_norm": 0.001760354032739997,
-      "learning_rate": 0.00019998983468552653,
-      "loss": 46.0,
-      "step": 28194
-    },
-    {
-      "epoch": 4.540601473489271,
-      "grad_norm": 0.002885441994294524,
-      "learning_rate": 0.0001999898339641724,
-      "loss": 46.0,
-      "step": 28195
-    },
-    {
-      "epoch": 4.540762510568058,
-      "grad_norm": 0.0035731117241084576,
-      "learning_rate": 0.0001999898332427927,
-      "loss": 46.0,
-      "step": 28196
-    },
-    {
-      "epoch": 4.5409235476468455,
-      "grad_norm": 0.0033794636838138103,
-      "learning_rate": 0.00019998983252138736,
-      "loss": 46.0,
-      "step": 28197
-    },
-    {
-      "epoch": 4.541084584725633,
-      "grad_norm": 0.0027073032688349485,
-      "learning_rate": 0.00019998983179995644,
-      "loss": 46.0,
-      "step": 28198
-    },
-    {
-      "epoch": 4.54124562180442,
-      "grad_norm": 0.002348700538277626,
-      "learning_rate": 0.0001999898310784999,
-      "loss": 46.0,
-      "step": 28199
-    },
-    {
-      "epoch": 4.541406658883208,
-      "grad_norm": 0.004205639939755201,
-      "learning_rate": 0.00019998983035701784,
-      "loss": 46.0,
-      "step": 28200
-    },
-    {
-      "epoch": 4.541567695961995,
-      "grad_norm": 0.004746836610138416,
-      "learning_rate": 0.00019998982963551014,
-      "loss": 46.0,
-      "step": 28201
-    },
-    {
-      "epoch": 4.541728733040783,
-      "grad_norm": 0.002171910833567381,
-      "learning_rate": 0.00019998982891397685,
-      "loss": 46.0,
-      "step": 28202
-    },
-    {
-      "epoch": 4.54188977011957,
-      "grad_norm": 0.005923191085457802,
-      "learning_rate": 0.00019998982819241796,
-      "loss": 46.0,
-      "step": 28203
-    },
-    {
-      "epoch": 4.542050807198358,
-      "grad_norm": 0.0036464245058596134,
-      "learning_rate": 0.00019998982747083347,
-      "loss": 46.0,
-      "step": 28204
-    },
-    {
-      "epoch": 4.542211844277145,
-      "grad_norm": 0.0044078282080590725,
-      "learning_rate": 0.00019998982674922342,
-      "loss": 46.0,
-      "step": 28205
-    },
-    {
-      "epoch": 4.5423728813559325,
-      "grad_norm": 0.003571434412151575,
-      "learning_rate": 0.00019998982602758777,
-      "loss": 46.0,
-      "step": 28206
-    },
-    {
-      "epoch": 4.542533918434719,
-      "grad_norm": 0.005233215633779764,
-      "learning_rate": 0.00019998982530592652,
-      "loss": 46.0,
-      "step": 28207
-    },
-    {
-      "epoch": 4.542694955513507,
-      "grad_norm": 0.003917828667908907,
-      "learning_rate": 0.00019998982458423967,
-      "loss": 46.0,
-      "step": 28208
-    },
-    {
-      "epoch": 4.542855992592294,
-      "grad_norm": 0.013388173654675484,
-      "learning_rate": 0.00019998982386252724,
-      "loss": 46.0,
-      "step": 28209
-    },
-    {
-      "epoch": 4.5430170296710815,
-      "grad_norm": 0.0016912242863327265,
-      "learning_rate": 0.0001999898231407892,
-      "loss": 46.0,
-      "step": 28210
-    },
-    {
-      "epoch": 4.543178066749869,
-      "grad_norm": 0.007283136248588562,
-      "learning_rate": 0.0001999898224190256,
-      "loss": 46.0,
-      "step": 28211
-    },
-    {
-      "epoch": 4.543339103828656,
-      "grad_norm": 0.0037675367202609777,
-      "learning_rate": 0.0001999898216972364,
-      "loss": 46.0,
-      "step": 28212
-    },
-    {
-      "epoch": 4.543500140907444,
-      "grad_norm": 0.005025290884077549,
-      "learning_rate": 0.0001999898209754216,
-      "loss": 46.0,
-      "step": 28213
-    },
-    {
-      "epoch": 4.543661177986231,
-      "grad_norm": 0.001676195184700191,
-      "learning_rate": 0.00019998982025358118,
-      "loss": 46.0,
-      "step": 28214
-    },
-    {
-      "epoch": 4.543822215065019,
-      "grad_norm": 0.005545006599277258,
-      "learning_rate": 0.0001999898195317152,
-      "loss": 46.0,
-      "step": 28215
-    },
-    {
-      "epoch": 4.543983252143806,
-      "grad_norm": 0.004245630465447903,
-      "learning_rate": 0.00019998981880982363,
-      "loss": 46.0,
-      "step": 28216
-    },
-    {
-      "epoch": 4.544144289222594,
-      "grad_norm": 0.004485202021896839,
-      "learning_rate": 0.00019998981808790644,
-      "loss": 46.0,
-      "step": 28217
-    },
-    {
-      "epoch": 4.544305326301381,
-      "grad_norm": 0.0124098751693964,
-      "learning_rate": 0.0001999898173659637,
-      "loss": 46.0,
-      "step": 28218
-    },
-    {
-      "epoch": 4.5444663633801685,
-      "grad_norm": 0.0062004439532756805,
-      "learning_rate": 0.00019998981664399535,
-      "loss": 46.0,
-      "step": 28219
-    },
-    {
-      "epoch": 4.544627400458956,
-      "grad_norm": 0.0012701530940830708,
-      "learning_rate": 0.00019998981592200138,
-      "loss": 46.0,
-      "step": 28220
-    },
-    {
-      "epoch": 4.544788437537743,
-      "grad_norm": 0.0059409174136817455,
-      "learning_rate": 0.00019998981519998184,
-      "loss": 46.0,
-      "step": 28221
-    },
-    {
-      "epoch": 4.54494947461653,
-      "grad_norm": 0.001416877843439579,
-      "learning_rate": 0.00019998981447793673,
-      "loss": 46.0,
-      "step": 28222
-    },
-    {
-      "epoch": 4.545110511695318,
-      "grad_norm": 0.007426689378917217,
-      "learning_rate": 0.00019998981375586597,
-      "loss": 46.0,
-      "step": 28223
-    },
-    {
-      "epoch": 4.545271548774105,
-      "grad_norm": 0.003515574848279357,
-      "learning_rate": 0.00019998981303376968,
-      "loss": 46.0,
-      "step": 28224
-    },
-    {
-      "epoch": 4.545432585852892,
-      "grad_norm": 0.0019585704430937767,
-      "learning_rate": 0.00019998981231164774,
-      "loss": 46.0,
-      "step": 28225
-    },
-    {
-      "epoch": 4.54559362293168,
-      "grad_norm": 0.006768498569726944,
-      "learning_rate": 0.00019998981158950025,
-      "loss": 46.0,
-      "step": 28226
-    },
-    {
-      "epoch": 4.545754660010467,
-      "grad_norm": 0.0034896666184067726,
-      "learning_rate": 0.00019998981086732717,
-      "loss": 46.0,
-      "step": 28227
-    },
-    {
-      "epoch": 4.545915697089255,
-      "grad_norm": 0.006893771700561047,
-      "learning_rate": 0.00019998981014512847,
-      "loss": 46.0,
-      "step": 28228
-    },
-    {
-      "epoch": 4.546076734168042,
-      "grad_norm": 0.014092490077018738,
-      "learning_rate": 0.0001999898094229042,
-      "loss": 46.0,
-      "step": 28229
-    },
-    {
-      "epoch": 4.54623777124683,
-      "grad_norm": 0.0021846434101462364,
-      "learning_rate": 0.00019998980870065432,
-      "loss": 46.0,
-      "step": 28230
-    },
-    {
-      "epoch": 4.546398808325617,
-      "grad_norm": 0.002295448910444975,
-      "learning_rate": 0.00019998980797837886,
-      "loss": 46.0,
-      "step": 28231
-    },
-    {
-      "epoch": 4.5465598454044045,
-      "grad_norm": 0.012366644106805325,
-      "learning_rate": 0.0001999898072560778,
-      "loss": 46.0,
-      "step": 28232
-    },
-    {
-      "epoch": 4.546720882483192,
-      "grad_norm": 0.0015726208221167326,
-      "learning_rate": 0.00019998980653375113,
-      "loss": 46.0,
-      "step": 28233
-    },
-    {
-      "epoch": 4.546881919561979,
-      "grad_norm": 0.009916739538311958,
-      "learning_rate": 0.00019998980581139892,
-      "loss": 46.0,
-      "step": 28234
-    },
-    {
-      "epoch": 4.547042956640767,
-      "grad_norm": 0.0029857633635401726,
-      "learning_rate": 0.00019998980508902108,
-      "loss": 46.0,
-      "step": 28235
-    },
-    {
-      "epoch": 4.547203993719554,
-      "grad_norm": 0.004519087262451649,
-      "learning_rate": 0.00019998980436661764,
-      "loss": 46.0,
-      "step": 28236
-    },
-    {
-      "epoch": 4.547365030798341,
-      "grad_norm": 0.002939286408945918,
-      "learning_rate": 0.00019998980364418866,
-      "loss": 46.0,
-      "step": 28237
-    },
-    {
-      "epoch": 4.547526067877129,
-      "grad_norm": 0.0009906847262755036,
-      "learning_rate": 0.000199989802921734,
-      "loss": 46.0,
-      "step": 28238
-    },
-    {
-      "epoch": 4.547687104955916,
-      "grad_norm": 0.004644661210477352,
-      "learning_rate": 0.00019998980219925383,
-      "loss": 46.0,
-      "step": 28239
-    },
-    {
-      "epoch": 4.547848142034703,
-      "grad_norm": 0.006151839159429073,
-      "learning_rate": 0.000199989801476748,
-      "loss": 46.0,
-      "step": 28240
-    },
-    {
-      "epoch": 4.548009179113491,
-      "grad_norm": 0.010135608725249767,
-      "learning_rate": 0.00019998980075421665,
-      "loss": 46.0,
-      "step": 28241
-    },
-    {
-      "epoch": 4.548170216192278,
-      "grad_norm": 0.002157302340492606,
-      "learning_rate": 0.00019998980003165965,
-      "loss": 46.0,
-      "step": 28242
-    },
-    {
-      "epoch": 4.548331253271066,
-      "grad_norm": 0.004398688208311796,
-      "learning_rate": 0.0001999897993090771,
-      "loss": 46.0,
-      "step": 28243
-    },
-    {
-      "epoch": 4.548492290349853,
-      "grad_norm": 0.006782538257539272,
-      "learning_rate": 0.00019998979858646892,
-      "loss": 46.0,
-      "step": 28244
-    },
-    {
-      "epoch": 4.5486533274286405,
-      "grad_norm": 0.006548644043505192,
-      "learning_rate": 0.00019998979786383517,
-      "loss": 46.0,
-      "step": 28245
-    },
-    {
-      "epoch": 4.548814364507428,
-      "grad_norm": 0.0014312970452010632,
-      "learning_rate": 0.00019998979714117582,
-      "loss": 46.0,
-      "step": 28246
-    },
-    {
-      "epoch": 4.548975401586215,
-      "grad_norm": 0.001444373861886561,
-      "learning_rate": 0.0001999897964184909,
-      "loss": 46.0,
-      "step": 28247
-    },
-    {
-      "epoch": 4.549136438665003,
-      "grad_norm": 0.002922563813626766,
-      "learning_rate": 0.00019998979569578037,
-      "loss": 46.0,
-      "step": 28248
-    },
-    {
-      "epoch": 4.54929747574379,
-      "grad_norm": 0.0067072720266878605,
-      "learning_rate": 0.0001999897949730442,
-      "loss": 46.0,
-      "step": 28249
-    },
-    {
-      "epoch": 4.549458512822578,
-      "grad_norm": 0.02529129944741726,
-      "learning_rate": 0.0001999897942502825,
-      "loss": 46.0,
-      "step": 28250
-    },
-    {
-      "epoch": 4.549619549901365,
-      "grad_norm": 0.0025627086870372295,
-      "learning_rate": 0.0001999897935274952,
-      "loss": 46.0,
-      "step": 28251
-    },
-    {
-      "epoch": 4.549780586980152,
-      "grad_norm": 0.00597706763073802,
-      "learning_rate": 0.0001999897928046823,
-      "loss": 46.0,
-      "step": 28252
-    },
-    {
-      "epoch": 4.549941624058939,
-      "grad_norm": 0.0023587120231240988,
-      "learning_rate": 0.0001999897920818438,
-      "loss": 46.0,
-      "step": 28253
-    },
-    {
-      "epoch": 4.550102661137727,
-      "grad_norm": 0.0034493152052164078,
-      "learning_rate": 0.00019998979135897973,
-      "loss": 46.0,
-      "step": 28254
-    },
-    {
-      "epoch": 4.550263698216514,
-      "grad_norm": 0.006048236973583698,
-      "learning_rate": 0.00019998979063609004,
-      "loss": 46.0,
-      "step": 28255
-    },
-    {
-      "epoch": 4.550424735295302,
-      "grad_norm": 0.005193533841520548,
-      "learning_rate": 0.00019998978991317477,
-      "loss": 46.0,
-      "step": 28256
-    },
-    {
-      "epoch": 4.550585772374089,
-      "grad_norm": 0.011304478161036968,
-      "learning_rate": 0.00019998978919023391,
-      "loss": 46.0,
-      "step": 28257
-    },
-    {
-      "epoch": 4.5507468094528765,
-      "grad_norm": 0.009198413230478764,
-      "learning_rate": 0.00019998978846726747,
-      "loss": 46.0,
-      "step": 28258
-    },
-    {
-      "epoch": 4.550907846531664,
-      "grad_norm": 0.005907850340008736,
-      "learning_rate": 0.0001999897877442754,
-      "loss": 46.0,
-      "step": 28259
-    },
-    {
-      "epoch": 4.551068883610451,
-      "grad_norm": 0.005806928034871817,
-      "learning_rate": 0.0001999897870212578,
-      "loss": 46.0,
-      "step": 28260
-    },
-    {
-      "epoch": 4.551229920689239,
-      "grad_norm": 0.007388319820165634,
-      "learning_rate": 0.00019998978629821455,
-      "loss": 46.0,
-      "step": 28261
-    },
-    {
-      "epoch": 4.551390957768026,
-      "grad_norm": 0.002225305885076523,
-      "learning_rate": 0.0001999897855751457,
-      "loss": 46.0,
-      "step": 28262
-    },
-    {
-      "epoch": 4.551551994846814,
-      "grad_norm": 0.0031608748249709606,
-      "learning_rate": 0.00019998978485205132,
-      "loss": 46.0,
-      "step": 28263
-    },
-    {
-      "epoch": 4.551713031925601,
-      "grad_norm": 0.006002143025398254,
-      "learning_rate": 0.0001999897841289313,
-      "loss": 46.0,
-      "step": 28264
-    },
-    {
-      "epoch": 4.551874069004389,
-      "grad_norm": 0.0024387154262512922,
-      "learning_rate": 0.0001999897834057857,
-      "loss": 46.0,
-      "step": 28265
-    },
-    {
-      "epoch": 4.552035106083176,
-      "grad_norm": 0.008900858461856842,
-      "learning_rate": 0.0001999897826826145,
-      "loss": 46.0,
-      "step": 28266
-    },
-    {
-      "epoch": 4.552196143161963,
-      "grad_norm": 0.007654973771423101,
-      "learning_rate": 0.00019998978195941773,
-      "loss": 46.0,
-      "step": 28267
-    },
-    {
-      "epoch": 4.55235718024075,
-      "grad_norm": 0.0014580423012375832,
-      "learning_rate": 0.00019998978123619534,
-      "loss": 46.0,
-      "step": 28268
-    },
-    {
-      "epoch": 4.5525182173195375,
-      "grad_norm": 0.001803286257199943,
-      "learning_rate": 0.00019998978051294738,
-      "loss": 46.0,
-      "step": 28269
-    },
-    {
-      "epoch": 4.552679254398325,
-      "grad_norm": 0.0034424234181642532,
-      "learning_rate": 0.0001999897797896738,
-      "loss": 46.0,
-      "step": 28270
-    },
-    {
-      "epoch": 4.5528402914771124,
-      "grad_norm": 0.0038513776380568743,
-      "learning_rate": 0.00019998977906637467,
-      "loss": 46.0,
-      "step": 28271
-    },
-    {
-      "epoch": 4.5530013285559,
-      "grad_norm": 0.005881298333406448,
-      "learning_rate": 0.0001999897783430499,
-      "loss": 46.0,
-      "step": 28272
-    },
-    {
-      "epoch": 4.553162365634687,
-      "grad_norm": 0.002593496348708868,
-      "learning_rate": 0.00019998977761969956,
-      "loss": 46.0,
-      "step": 28273
-    },
-    {
-      "epoch": 4.553323402713475,
-      "grad_norm": 0.01781601831316948,
-      "learning_rate": 0.00019998977689632367,
-      "loss": 46.0,
-      "step": 28274
-    },
-    {
-      "epoch": 4.553484439792262,
-      "grad_norm": 0.0012850770726799965,
-      "learning_rate": 0.0001999897761729221,
-      "loss": 46.0,
-      "step": 28275
-    },
-    {
-      "epoch": 4.55364547687105,
-      "grad_norm": 0.005866596475243568,
-      "learning_rate": 0.000199989775449495,
-      "loss": 46.0,
-      "step": 28276
-    },
-    {
-      "epoch": 4.553806513949837,
-      "grad_norm": 0.006958988960832357,
-      "learning_rate": 0.0001999897747260423,
-      "loss": 46.0,
-      "step": 28277
-    },
-    {
-      "epoch": 4.553967551028625,
-      "grad_norm": 0.00456677470356226,
-      "learning_rate": 0.000199989774002564,
-      "loss": 46.0,
-      "step": 28278
-    },
-    {
-      "epoch": 4.554128588107412,
-      "grad_norm": 0.014959155581891537,
-      "learning_rate": 0.0001999897732790601,
-      "loss": 46.0,
-      "step": 28279
-    },
-    {
-      "epoch": 4.5542896251861995,
-      "grad_norm": 0.002239905996248126,
-      "learning_rate": 0.00019998977255553061,
-      "loss": 46.0,
-      "step": 28280
-    },
-    {
-      "epoch": 4.554450662264987,
-      "grad_norm": 0.011093229055404663,
-      "learning_rate": 0.00019998977183197556,
-      "loss": 46.0,
-      "step": 28281
-    },
-    {
-      "epoch": 4.5546116993437735,
-      "grad_norm": 0.006913878954946995,
-      "learning_rate": 0.00019998977110839488,
-      "loss": 46.0,
-      "step": 28282
-    },
-    {
-      "epoch": 4.554772736422561,
-      "grad_norm": 0.004666912369430065,
-      "learning_rate": 0.00019998977038478862,
-      "loss": 46.0,
-      "step": 28283
-    },
-    {
-      "epoch": 4.554933773501348,
-      "grad_norm": 0.0013718479312956333,
-      "learning_rate": 0.00019998976966115677,
-      "loss": 46.0,
-      "step": 28284
-    },
-    {
-      "epoch": 4.555094810580136,
-      "grad_norm": 0.0023330957628786564,
-      "learning_rate": 0.0001999897689374993,
-      "loss": 46.0,
-      "step": 28285
-    },
-    {
-      "epoch": 4.555255847658923,
-      "grad_norm": 0.017625343054533005,
-      "learning_rate": 0.00019998976821381626,
-      "loss": 46.0,
-      "step": 28286
-    },
-    {
-      "epoch": 4.555416884737711,
-      "grad_norm": 0.009019560180604458,
-      "learning_rate": 0.00019998976749010765,
-      "loss": 46.0,
-      "step": 28287
-    },
-    {
-      "epoch": 4.555577921816498,
-      "grad_norm": 0.007749797776341438,
-      "learning_rate": 0.00019998976676637343,
-      "loss": 46.0,
-      "step": 28288
-    },
-    {
-      "epoch": 4.555738958895286,
-      "grad_norm": 0.0019633814226835966,
-      "learning_rate": 0.00019998976604261362,
-      "loss": 46.0,
-      "step": 28289
-    },
-    {
-      "epoch": 4.555899995974073,
-      "grad_norm": 0.017374303191900253,
-      "learning_rate": 0.0001999897653188282,
-      "loss": 46.0,
-      "step": 28290
-    },
-    {
-      "epoch": 4.556061033052861,
-      "grad_norm": 0.002849637996405363,
-      "learning_rate": 0.00019998976459501718,
-      "loss": 46.0,
-      "step": 28291
-    },
-    {
-      "epoch": 4.556222070131648,
-      "grad_norm": 0.000549272692296654,
-      "learning_rate": 0.0001999897638711806,
-      "loss": 46.0,
-      "step": 28292
-    },
-    {
-      "epoch": 4.5563831072104355,
-      "grad_norm": 0.00421734107658267,
-      "learning_rate": 0.00019998976314731842,
-      "loss": 46.0,
-      "step": 28293
-    },
-    {
-      "epoch": 4.556544144289223,
-      "grad_norm": 0.003836626885458827,
-      "learning_rate": 0.00019998976242343065,
-      "loss": 46.0,
-      "step": 28294
-    },
-    {
-      "epoch": 4.5567051813680095,
-      "grad_norm": 0.002068571513518691,
-      "learning_rate": 0.0001999897616995173,
-      "loss": 46.0,
-      "step": 28295
-    },
-    {
-      "epoch": 4.556866218446798,
-      "grad_norm": 0.00625478383153677,
-      "learning_rate": 0.00019998976097557834,
-      "loss": 46.0,
-      "step": 28296
-    },
-    {
-      "epoch": 4.557027255525584,
-      "grad_norm": 0.006283685099333525,
-      "learning_rate": 0.00019998976025161378,
-      "loss": 46.0,
-      "step": 28297
-    },
-    {
-      "epoch": 4.557188292604372,
-      "grad_norm": 0.009408128447830677,
-      "learning_rate": 0.00019998975952762363,
-      "loss": 46.0,
-      "step": 28298
-    },
-    {
-      "epoch": 4.557349329683159,
-      "grad_norm": 0.0022420056629925966,
-      "learning_rate": 0.0001999897588036079,
-      "loss": 46.0,
-      "step": 28299
-    },
-    {
-      "epoch": 4.557510366761947,
-      "grad_norm": 0.0017376190517097712,
-      "learning_rate": 0.00019998975807956657,
-      "loss": 46.0,
-      "step": 28300
-    },
-    {
-      "epoch": 4.557671403840734,
-      "grad_norm": 0.006508963648229837,
-      "learning_rate": 0.00019998975735549963,
-      "loss": 46.0,
-      "step": 28301
-    },
-    {
-      "epoch": 4.557832440919522,
-      "grad_norm": 0.002009720541536808,
-      "learning_rate": 0.00019998975663140713,
-      "loss": 46.0,
-      "step": 28302
-    },
-    {
-      "epoch": 4.557993477998309,
-      "grad_norm": 0.008927403017878532,
-      "learning_rate": 0.00019998975590728902,
-      "loss": 46.0,
-      "step": 28303
-    },
-    {
-      "epoch": 4.558154515077097,
-      "grad_norm": 0.009754546917974949,
-      "learning_rate": 0.00019998975518314535,
-      "loss": 46.0,
-      "step": 28304
-    },
-    {
-      "epoch": 4.558315552155884,
-      "grad_norm": 0.0027473876252770424,
-      "learning_rate": 0.00019998975445897603,
-      "loss": 46.0,
-      "step": 28305
-    },
-    {
-      "epoch": 4.5584765892346715,
-      "grad_norm": 0.0027844624128192663,
-      "learning_rate": 0.00019998975373478116,
-      "loss": 46.0,
-      "step": 28306
-    },
-    {
-      "epoch": 4.558637626313459,
-      "grad_norm": 0.009226133115589619,
-      "learning_rate": 0.0001999897530105607,
-      "loss": 46.0,
-      "step": 28307
-    },
-    {
-      "epoch": 4.558798663392246,
-      "grad_norm": 0.0023362061474472284,
-      "learning_rate": 0.0001999897522863146,
-      "loss": 46.0,
-      "step": 28308
-    },
-    {
-      "epoch": 4.558959700471034,
-      "grad_norm": 0.01723659038543701,
-      "learning_rate": 0.00019998975156204293,
-      "loss": 46.0,
-      "step": 28309
-    },
-    {
-      "epoch": 4.55912073754982,
-      "grad_norm": 0.00814008992165327,
-      "learning_rate": 0.00019998975083774568,
-      "loss": 46.0,
-      "step": 28310
-    },
-    {
-      "epoch": 4.559281774628609,
-      "grad_norm": 0.0026714010164141655,
-      "learning_rate": 0.00019998975011342287,
-      "loss": 46.0,
-      "step": 28311
-    },
-    {
-      "epoch": 4.559442811707395,
-      "grad_norm": 0.003975267522037029,
-      "learning_rate": 0.00019998974938907442,
-      "loss": 46.0,
-      "step": 28312
-    },
-    {
-      "epoch": 4.559603848786183,
-      "grad_norm": 0.013119656592607498,
-      "learning_rate": 0.00019998974866470038,
-      "loss": 46.0,
-      "step": 28313
-    },
-    {
-      "epoch": 4.55976488586497,
-      "grad_norm": 0.0019531873986124992,
-      "learning_rate": 0.00019998974794030078,
-      "loss": 46.0,
-      "step": 28314
-    },
-    {
-      "epoch": 4.559925922943758,
-      "grad_norm": 0.0026776501908898354,
-      "learning_rate": 0.00019998974721587554,
-      "loss": 46.0,
-      "step": 28315
-    },
-    {
-      "epoch": 4.560086960022545,
-      "grad_norm": 0.0036813144106417894,
-      "learning_rate": 0.00019998974649142474,
-      "loss": 46.0,
-      "step": 28316
-    },
-    {
-      "epoch": 4.5602479971013326,
-      "grad_norm": 0.0014915541978552938,
-      "learning_rate": 0.00019998974576694836,
-      "loss": 46.0,
-      "step": 28317
-    },
-    {
-      "epoch": 4.56040903418012,
-      "grad_norm": 0.006268836557865143,
-      "learning_rate": 0.00019998974504244636,
-      "loss": 46.0,
-      "step": 28318
-    },
-    {
-      "epoch": 4.5605700712589075,
-      "grad_norm": 0.007194567006081343,
-      "learning_rate": 0.0001999897443179188,
-      "loss": 46.0,
-      "step": 28319
-    },
-    {
-      "epoch": 4.560731108337695,
-      "grad_norm": 0.0015810198383405805,
-      "learning_rate": 0.00019998974359336562,
-      "loss": 46.0,
-      "step": 28320
-    },
-    {
-      "epoch": 4.560892145416482,
-      "grad_norm": 0.006441446952521801,
-      "learning_rate": 0.00019998974286878686,
-      "loss": 46.0,
-      "step": 28321
-    },
-    {
-      "epoch": 4.56105318249527,
-      "grad_norm": 0.00707559147849679,
-      "learning_rate": 0.00019998974214418248,
-      "loss": 46.0,
-      "step": 28322
-    },
-    {
-      "epoch": 4.561214219574057,
-      "grad_norm": 0.006631402298808098,
-      "learning_rate": 0.00019998974141955254,
-      "loss": 46.0,
-      "step": 28323
-    },
-    {
-      "epoch": 4.561375256652845,
-      "grad_norm": 0.0061634317971765995,
-      "learning_rate": 0.000199989740694897,
-      "loss": 46.0,
-      "step": 28324
-    },
-    {
-      "epoch": 4.561536293731631,
-      "grad_norm": 0.008553068153560162,
-      "learning_rate": 0.00019998973997021585,
-      "loss": 46.0,
-      "step": 28325
-    },
-    {
-      "epoch": 4.561697330810419,
-      "grad_norm": 0.005489504896104336,
-      "learning_rate": 0.00019998973924550915,
-      "loss": 46.0,
-      "step": 28326
-    },
-    {
-      "epoch": 4.561858367889206,
-      "grad_norm": 0.005943399388343096,
-      "learning_rate": 0.0001999897385207768,
-      "loss": 46.0,
-      "step": 28327
-    },
-    {
-      "epoch": 4.562019404967994,
-      "grad_norm": 0.005850377958267927,
-      "learning_rate": 0.00019998973779601889,
-      "loss": 46.0,
-      "step": 28328
-    },
-    {
-      "epoch": 4.562180442046781,
-      "grad_norm": 0.0027680303901433945,
-      "learning_rate": 0.0001999897370712354,
-      "loss": 46.0,
-      "step": 28329
-    },
-    {
-      "epoch": 4.5623414791255685,
-      "grad_norm": 0.0008222645265050232,
-      "learning_rate": 0.0001999897363464263,
-      "loss": 46.0,
-      "step": 28330
-    },
-    {
-      "epoch": 4.562502516204356,
-      "grad_norm": 0.0037766112945973873,
-      "learning_rate": 0.00019998973562159163,
-      "loss": 46.0,
-      "step": 28331
-    },
-    {
-      "epoch": 4.562663553283143,
-      "grad_norm": 0.004440687131136656,
-      "learning_rate": 0.0001999897348967313,
-      "loss": 46.0,
-      "step": 28332
-    },
-    {
-      "epoch": 4.562824590361931,
-      "grad_norm": 0.009641860611736774,
-      "learning_rate": 0.00019998973417184547,
-      "loss": 46.0,
-      "step": 28333
-    },
-    {
-      "epoch": 4.562985627440718,
-      "grad_norm": 0.005281203892081976,
-      "learning_rate": 0.000199989733446934,
-      "loss": 46.0,
-      "step": 28334
-    },
-    {
-      "epoch": 4.563146664519506,
-      "grad_norm": 0.0026960819959640503,
-      "learning_rate": 0.00019998973272199693,
-      "loss": 46.0,
-      "step": 28335
-    },
-    {
-      "epoch": 4.563307701598293,
-      "grad_norm": 0.0054990132339298725,
-      "learning_rate": 0.00019998973199703427,
-      "loss": 46.0,
-      "step": 28336
-    },
-    {
-      "epoch": 4.563468738677081,
-      "grad_norm": 0.0021799386013299227,
-      "learning_rate": 0.00019998973127204606,
-      "loss": 46.0,
-      "step": 28337
-    },
-    {
-      "epoch": 4.563629775755868,
-      "grad_norm": 0.004018097184598446,
-      "learning_rate": 0.0001999897305470322,
-      "loss": 46.0,
-      "step": 28338
-    },
-    {
-      "epoch": 4.563790812834656,
-      "grad_norm": 0.0018440329004079103,
-      "learning_rate": 0.00019998972982199277,
-      "loss": 46.0,
-      "step": 28339
-    },
-    {
-      "epoch": 4.563951849913442,
-      "grad_norm": 0.0031687328591942787,
-      "learning_rate": 0.00019998972909692774,
-      "loss": 46.0,
-      "step": 28340
-    },
-    {
-      "epoch": 4.56411288699223,
-      "grad_norm": 0.0034760283306241035,
-      "learning_rate": 0.00019998972837183713,
-      "loss": 46.0,
-      "step": 28341
-    },
-    {
-      "epoch": 4.564273924071017,
-      "grad_norm": 0.009181971661746502,
-      "learning_rate": 0.00019998972764672095,
-      "loss": 46.0,
-      "step": 28342
-    },
-    {
-      "epoch": 4.5644349611498045,
-      "grad_norm": 0.006531584542244673,
-      "learning_rate": 0.00019998972692157914,
-      "loss": 46.0,
-      "step": 28343
-    },
-    {
-      "epoch": 4.564595998228592,
-      "grad_norm": 0.005438131280243397,
-      "learning_rate": 0.00019998972619641176,
-      "loss": 46.0,
-      "step": 28344
-    },
-    {
-      "epoch": 4.564757035307379,
-      "grad_norm": 0.013236350379884243,
-      "learning_rate": 0.00019998972547121877,
-      "loss": 46.0,
-      "step": 28345
-    },
-    {
-      "epoch": 4.564918072386167,
-      "grad_norm": 0.0014279453316703439,
-      "learning_rate": 0.00019998972474600022,
-      "loss": 46.0,
-      "step": 28346
-    },
-    {
-      "epoch": 4.565079109464954,
-      "grad_norm": 0.003309838240966201,
-      "learning_rate": 0.00019998972402075605,
-      "loss": 46.0,
-      "step": 28347
-    },
-    {
-      "epoch": 4.565240146543742,
-      "grad_norm": 0.0037735984660685062,
-      "learning_rate": 0.00019998972329548627,
-      "loss": 46.0,
-      "step": 28348
-    },
-    {
-      "epoch": 4.565401183622529,
-      "grad_norm": 0.005396547727286816,
-      "learning_rate": 0.00019998972257019096,
-      "loss": 46.0,
-      "step": 28349
-    },
-    {
-      "epoch": 4.565562220701317,
-      "grad_norm": 0.012511735782027245,
-      "learning_rate": 0.00019998972184486998,
-      "loss": 46.0,
-      "step": 28350
-    },
-    {
-      "epoch": 4.565723257780104,
-      "grad_norm": 0.0031030962709337473,
-      "learning_rate": 0.00019998972111952344,
-      "loss": 46.0,
-      "step": 28351
-    },
-    {
-      "epoch": 4.565884294858892,
-      "grad_norm": 0.005439350847154856,
-      "learning_rate": 0.00019998972039415134,
-      "loss": 46.0,
-      "step": 28352
-    },
-    {
-      "epoch": 4.566045331937679,
-      "grad_norm": 0.0023892424069344997,
-      "learning_rate": 0.00019998971966875363,
-      "loss": 46.0,
-      "step": 28353
-    },
-    {
-      "epoch": 4.5662063690164665,
-      "grad_norm": 0.002230866579338908,
-      "learning_rate": 0.00019998971894333032,
-      "loss": 46.0,
-      "step": 28354
-    },
-    {
-      "epoch": 4.566367406095253,
-      "grad_norm": 0.00886800978332758,
-      "learning_rate": 0.0001999897182178814,
-      "loss": 46.0,
-      "step": 28355
-    },
-    {
-      "epoch": 4.5665284431740405,
-      "grad_norm": 0.010969109833240509,
-      "learning_rate": 0.0001999897174924069,
-      "loss": 46.0,
-      "step": 28356
-    },
-    {
-      "epoch": 4.566689480252828,
-      "grad_norm": 0.0030753680039197206,
-      "learning_rate": 0.00019998971676690684,
-      "loss": 46.0,
-      "step": 28357
-    },
-    {
-      "epoch": 4.566850517331615,
-      "grad_norm": 0.005177609156817198,
-      "learning_rate": 0.00019998971604138113,
-      "loss": 46.0,
-      "step": 28358
-    },
-    {
-      "epoch": 4.567011554410403,
-      "grad_norm": 0.003990458324551582,
-      "learning_rate": 0.0001999897153158299,
-      "loss": 46.0,
-      "step": 28359
-    },
-    {
-      "epoch": 4.56717259148919,
-      "grad_norm": 0.005395010579377413,
-      "learning_rate": 0.000199989714590253,
-      "loss": 46.0,
-      "step": 28360
-    },
-    {
-      "epoch": 4.567333628567978,
-      "grad_norm": 0.006897305138409138,
-      "learning_rate": 0.00019998971386465055,
-      "loss": 46.0,
-      "step": 28361
-    },
-    {
-      "epoch": 4.567494665646765,
-      "grad_norm": 0.004040502477437258,
-      "learning_rate": 0.0001999897131390225,
-      "loss": 46.0,
-      "step": 28362
-    },
-    {
-      "epoch": 4.567655702725553,
-      "grad_norm": 0.003765776753425598,
-      "learning_rate": 0.00019998971241336886,
-      "loss": 46.0,
-      "step": 28363
-    },
-    {
-      "epoch": 4.56781673980434,
-      "grad_norm": 0.0033694072626531124,
-      "learning_rate": 0.00019998971168768966,
-      "loss": 46.0,
-      "step": 28364
-    },
-    {
-      "epoch": 4.567977776883128,
-      "grad_norm": 0.005281837657094002,
-      "learning_rate": 0.0001999897109619848,
-      "loss": 46.0,
-      "step": 28365
-    },
-    {
-      "epoch": 4.568138813961915,
-      "grad_norm": 0.003135291626676917,
-      "learning_rate": 0.00019998971023625438,
-      "loss": 46.0,
-      "step": 28366
-    },
-    {
-      "epoch": 4.5682998510407025,
-      "grad_norm": 0.001260228455066681,
-      "learning_rate": 0.00019998970951049837,
-      "loss": 46.0,
-      "step": 28367
-    },
-    {
-      "epoch": 4.568460888119489,
-      "grad_norm": 0.004873974714428186,
-      "learning_rate": 0.00019998970878471676,
-      "loss": 46.0,
-      "step": 28368
-    },
-    {
-      "epoch": 4.568621925198277,
-      "grad_norm": 0.0039630369283258915,
-      "learning_rate": 0.00019998970805890957,
-      "loss": 46.0,
-      "step": 28369
-    },
-    {
-      "epoch": 4.568782962277064,
-      "grad_norm": 0.005855859722942114,
-      "learning_rate": 0.0001999897073330768,
-      "loss": 46.0,
-      "step": 28370
-    },
-    {
-      "epoch": 4.568943999355851,
-      "grad_norm": 0.0012839700793847442,
-      "learning_rate": 0.00019998970660721843,
-      "loss": 46.0,
-      "step": 28371
-    },
-    {
-      "epoch": 4.569105036434639,
-      "grad_norm": 0.0016850670799613,
-      "learning_rate": 0.00019998970588133445,
-      "loss": 46.0,
-      "step": 28372
-    },
-    {
-      "epoch": 4.569266073513426,
-      "grad_norm": 0.0019511821446940303,
-      "learning_rate": 0.00019998970515542488,
-      "loss": 46.0,
-      "step": 28373
-    },
-    {
-      "epoch": 4.569427110592214,
-      "grad_norm": 0.003716961480677128,
-      "learning_rate": 0.00019998970442948975,
-      "loss": 46.0,
-      "step": 28374
-    },
-    {
-      "epoch": 4.569588147671001,
-      "grad_norm": 0.005065771751105785,
-      "learning_rate": 0.000199989703703529,
-      "loss": 46.0,
-      "step": 28375
-    },
-    {
-      "epoch": 4.569749184749789,
-      "grad_norm": 0.002809886122122407,
-      "learning_rate": 0.00019998970297754266,
-      "loss": 46.0,
-      "step": 28376
-    },
-    {
-      "epoch": 4.569910221828576,
-      "grad_norm": 0.011783783324062824,
-      "learning_rate": 0.00019998970225153074,
-      "loss": 46.0,
-      "step": 28377
-    },
-    {
-      "epoch": 4.5700712589073635,
-      "grad_norm": 0.006862163078039885,
-      "learning_rate": 0.00019998970152549319,
-      "loss": 46.0,
-      "step": 28378
-    },
-    {
-      "epoch": 4.570232295986151,
-      "grad_norm": 0.0027296182233840227,
-      "learning_rate": 0.00019998970079943007,
-      "loss": 46.0,
-      "step": 28379
-    },
-    {
-      "epoch": 4.5703933330649384,
-      "grad_norm": 0.004065383691340685,
-      "learning_rate": 0.00019998970007334136,
-      "loss": 46.0,
-      "step": 28380
-    },
-    {
-      "epoch": 4.570554370143726,
-      "grad_norm": 0.008041276596486568,
-      "learning_rate": 0.00019998969934722707,
-      "loss": 46.0,
-      "step": 28381
-    },
-    {
-      "epoch": 4.570715407222513,
-      "grad_norm": 0.005012942012399435,
-      "learning_rate": 0.0001999896986210872,
-      "loss": 46.0,
-      "step": 28382
-    },
-    {
-      "epoch": 4.5708764443013,
-      "grad_norm": 0.004440854303538799,
-      "learning_rate": 0.0001999896978949217,
-      "loss": 46.0,
-      "step": 28383
-    },
-    {
-      "epoch": 4.571037481380088,
-      "grad_norm": 0.0037269804161041975,
-      "learning_rate": 0.00019998969716873062,
-      "loss": 46.0,
-      "step": 28384
-    },
-    {
-      "epoch": 4.571198518458875,
-      "grad_norm": 0.0008085716981440783,
-      "learning_rate": 0.00019998969644251396,
-      "loss": 46.0,
-      "step": 28385
-    },
-    {
-      "epoch": 4.571359555537662,
-      "grad_norm": 0.004750309977680445,
-      "learning_rate": 0.00019998969571627167,
-      "loss": 46.0,
-      "step": 28386
-    },
-    {
-      "epoch": 4.57152059261645,
-      "grad_norm": 0.0015329812886193395,
-      "learning_rate": 0.00019998969499000383,
-      "loss": 46.0,
-      "step": 28387
-    },
-    {
-      "epoch": 4.571681629695237,
-      "grad_norm": 0.003719376865774393,
-      "learning_rate": 0.00019998969426371038,
-      "loss": 46.0,
-      "step": 28388
-    },
-    {
-      "epoch": 4.571842666774025,
-      "grad_norm": 0.009897838346660137,
-      "learning_rate": 0.00019998969353739133,
-      "loss": 46.0,
-      "step": 28389
-    },
-    {
-      "epoch": 4.572003703852812,
-      "grad_norm": 0.0012749492889270186,
-      "learning_rate": 0.0001999896928110467,
-      "loss": 46.0,
-      "step": 28390
-    },
-    {
-      "epoch": 4.5721647409315995,
-      "grad_norm": 0.005704416893422604,
-      "learning_rate": 0.0001999896920846765,
-      "loss": 46.0,
-      "step": 28391
-    },
-    {
-      "epoch": 4.572325778010387,
-      "grad_norm": 0.0023975849617272615,
-      "learning_rate": 0.00019998969135828068,
-      "loss": 46.0,
-      "step": 28392
-    },
-    {
-      "epoch": 4.572486815089174,
-      "grad_norm": 0.006974838208407164,
-      "learning_rate": 0.0001999896906318593,
-      "loss": 46.0,
-      "step": 28393
-    },
-    {
-      "epoch": 4.572647852167962,
-      "grad_norm": 0.004614113364368677,
-      "learning_rate": 0.00019998968990541226,
-      "loss": 46.0,
-      "step": 28394
-    },
-    {
-      "epoch": 4.572808889246749,
-      "grad_norm": 0.00568041205406189,
-      "learning_rate": 0.00019998968917893966,
-      "loss": 46.0,
-      "step": 28395
-    },
-    {
-      "epoch": 4.572969926325537,
-      "grad_norm": 0.008333171717822552,
-      "learning_rate": 0.00019998968845244148,
-      "loss": 46.0,
-      "step": 28396
-    },
-    {
-      "epoch": 4.573130963404324,
-      "grad_norm": 0.0035618366673588753,
-      "learning_rate": 0.00019998968772591772,
-      "loss": 46.0,
-      "step": 28397
-    },
-    {
-      "epoch": 4.573292000483111,
-      "grad_norm": 0.004362665116786957,
-      "learning_rate": 0.00019998968699936836,
-      "loss": 46.0,
-      "step": 28398
-    },
-    {
-      "epoch": 4.573453037561898,
-      "grad_norm": 0.00687659066170454,
-      "learning_rate": 0.00019998968627279337,
-      "loss": 46.0,
-      "step": 28399
-    },
-    {
-      "epoch": 4.573614074640686,
-      "grad_norm": 0.0024620285257697105,
-      "learning_rate": 0.00019998968554619284,
-      "loss": 46.0,
-      "step": 28400
-    },
-    {
-      "epoch": 4.573775111719473,
-      "grad_norm": 0.0033602954354137182,
-      "learning_rate": 0.00019998968481956667,
-      "loss": 46.0,
-      "step": 28401
-    },
-    {
-      "epoch": 4.573936148798261,
-      "grad_norm": 0.020862076431512833,
-      "learning_rate": 0.00019998968409291496,
-      "loss": 46.0,
-      "step": 28402
-    },
-    {
-      "epoch": 4.574097185877048,
-      "grad_norm": 0.005975116044282913,
-      "learning_rate": 0.00019998968336623762,
-      "loss": 46.0,
-      "step": 28403
-    },
-    {
-      "epoch": 4.5742582229558355,
-      "grad_norm": 0.0035918280482292175,
-      "learning_rate": 0.00019998968263953471,
-      "loss": 46.0,
-      "step": 28404
-    },
-    {
-      "epoch": 4.574419260034623,
-      "grad_norm": 0.011295353062450886,
-      "learning_rate": 0.00019998968191280617,
-      "loss": 46.0,
-      "step": 28405
-    },
-    {
-      "epoch": 4.57458029711341,
-      "grad_norm": 0.00641406886279583,
-      "learning_rate": 0.0001999896811860521,
-      "loss": 46.0,
-      "step": 28406
-    },
-    {
-      "epoch": 4.574741334192198,
-      "grad_norm": 0.003867147024720907,
-      "learning_rate": 0.00019998968045927237,
-      "loss": 46.0,
-      "step": 28407
-    },
-    {
-      "epoch": 4.574902371270985,
-      "grad_norm": 0.006757321301847696,
-      "learning_rate": 0.0001999896797324671,
-      "loss": 46.0,
-      "step": 28408
-    },
-    {
-      "epoch": 4.575063408349773,
-      "grad_norm": 0.004591462668031454,
-      "learning_rate": 0.00019998967900563622,
-      "loss": 46.0,
-      "step": 28409
-    },
-    {
-      "epoch": 4.57522444542856,
-      "grad_norm": 0.0016657090745866299,
-      "learning_rate": 0.00019998967827877974,
-      "loss": 46.0,
-      "step": 28410
-    },
-    {
-      "epoch": 4.575385482507348,
-      "grad_norm": 0.004417312331497669,
-      "learning_rate": 0.00019998967755189767,
-      "loss": 46.0,
-      "step": 28411
-    },
-    {
-      "epoch": 4.575546519586135,
-      "grad_norm": 0.01496258843690157,
-      "learning_rate": 0.00019998967682499,
-      "loss": 46.0,
-      "step": 28412
-    },
-    {
-      "epoch": 4.575707556664922,
-      "grad_norm": 0.004842329770326614,
-      "learning_rate": 0.00019998967609805677,
-      "loss": 46.0,
-      "step": 28413
-    },
-    {
-      "epoch": 4.575868593743709,
-      "grad_norm": 0.01410147175192833,
-      "learning_rate": 0.0001999896753710979,
-      "loss": 46.0,
-      "step": 28414
-    },
-    {
-      "epoch": 4.576029630822497,
-      "grad_norm": 0.001803920604288578,
-      "learning_rate": 0.0001999896746441135,
-      "loss": 46.0,
-      "step": 28415
-    },
-    {
-      "epoch": 4.576190667901284,
-      "grad_norm": 0.002926790388301015,
-      "learning_rate": 0.00019998967391710344,
-      "loss": 46.0,
-      "step": 28416
-    },
-    {
-      "epoch": 4.5763517049800715,
-      "grad_norm": 0.002284947084262967,
-      "learning_rate": 0.00019998967319006782,
-      "loss": 46.0,
-      "step": 28417
-    },
-    {
-      "epoch": 4.576512742058859,
-      "grad_norm": 0.002812547143548727,
-      "learning_rate": 0.00019998967246300658,
-      "loss": 46.0,
-      "step": 28418
-    },
-    {
-      "epoch": 4.576673779137646,
-      "grad_norm": 0.002657968085259199,
-      "learning_rate": 0.00019998967173591982,
-      "loss": 46.0,
-      "step": 28419
-    },
-    {
-      "epoch": 4.576834816216434,
-      "grad_norm": 0.0068285283632576466,
-      "learning_rate": 0.0001999896710088074,
-      "loss": 46.0,
-      "step": 28420
-    },
-    {
-      "epoch": 4.576995853295221,
-      "grad_norm": 0.007613273337483406,
-      "learning_rate": 0.00019998967028166941,
-      "loss": 46.0,
-      "step": 28421
-    },
-    {
-      "epoch": 4.577156890374009,
-      "grad_norm": 0.005652642343193293,
-      "learning_rate": 0.00019998966955450583,
-      "loss": 46.0,
-      "step": 28422
-    },
-    {
-      "epoch": 4.577317927452796,
-      "grad_norm": 0.0014848349383100867,
-      "learning_rate": 0.00019998966882731666,
-      "loss": 46.0,
-      "step": 28423
-    },
-    {
-      "epoch": 4.577478964531584,
-      "grad_norm": 0.0011207249481230974,
-      "learning_rate": 0.00019998966810010188,
-      "loss": 46.0,
-      "step": 28424
-    },
-    {
-      "epoch": 4.577640001610371,
-      "grad_norm": 0.019937967881560326,
-      "learning_rate": 0.00019998966737286154,
-      "loss": 46.0,
-      "step": 28425
-    },
-    {
-      "epoch": 4.5778010386891586,
-      "grad_norm": 0.0026541464030742645,
-      "learning_rate": 0.0001999896666455956,
-      "loss": 46.0,
-      "step": 28426
-    },
-    {
-      "epoch": 4.577962075767946,
-      "grad_norm": 0.002017723862081766,
-      "learning_rate": 0.00019998966591830403,
-      "loss": 46.0,
-      "step": 28427
-    },
-    {
-      "epoch": 4.578123112846733,
-      "grad_norm": 0.002365991473197937,
-      "learning_rate": 0.0001999896651909869,
-      "loss": 46.0,
-      "step": 28428
-    },
-    {
-      "epoch": 4.57828414992552,
-      "grad_norm": 0.0027964806649833918,
-      "learning_rate": 0.00019998966446364418,
-      "loss": 46.0,
-      "step": 28429
-    },
-    {
-      "epoch": 4.5784451870043075,
-      "grad_norm": 0.0025261621922254562,
-      "learning_rate": 0.00019998966373627585,
-      "loss": 46.0,
-      "step": 28430
-    },
-    {
-      "epoch": 4.578606224083095,
-      "grad_norm": 0.00337060634046793,
-      "learning_rate": 0.00019998966300888195,
-      "loss": 46.0,
-      "step": 28431
-    },
-    {
-      "epoch": 4.578767261161882,
-      "grad_norm": 0.003920095041394234,
-      "learning_rate": 0.00019998966228146244,
-      "loss": 46.0,
-      "step": 28432
-    },
-    {
-      "epoch": 4.57892829824067,
-      "grad_norm": 0.0017860674997791648,
-      "learning_rate": 0.00019998966155401735,
-      "loss": 46.0,
-      "step": 28433
-    },
-    {
-      "epoch": 4.579089335319457,
-      "grad_norm": 0.002171809785068035,
-      "learning_rate": 0.00019998966082654667,
-      "loss": 46.0,
-      "step": 28434
-    },
-    {
-      "epoch": 4.579250372398245,
-      "grad_norm": 0.009731261059641838,
-      "learning_rate": 0.00019998966009905037,
-      "loss": 46.0,
-      "step": 28435
-    },
-    {
-      "epoch": 4.579411409477032,
-      "grad_norm": 0.0015148160746321082,
-      "learning_rate": 0.00019998965937152851,
-      "loss": 46.0,
-      "step": 28436
-    },
-    {
-      "epoch": 4.57957244655582,
-      "grad_norm": 0.0018464995082467794,
-      "learning_rate": 0.00019998965864398102,
-      "loss": 46.0,
-      "step": 28437
-    },
-    {
-      "epoch": 4.579733483634607,
-      "grad_norm": 0.0063488297164440155,
-      "learning_rate": 0.00019998965791640798,
-      "loss": 46.0,
-      "step": 28438
-    },
-    {
-      "epoch": 4.5798945207133945,
-      "grad_norm": 0.01029153075069189,
-      "learning_rate": 0.00019998965718880934,
-      "loss": 46.0,
-      "step": 28439
-    },
-    {
-      "epoch": 4.580055557792182,
-      "grad_norm": 0.0009967988589778543,
-      "learning_rate": 0.00019998965646118508,
-      "loss": 46.0,
-      "step": 28440
-    },
-    {
-      "epoch": 4.5802165948709685,
-      "grad_norm": 0.003307321574538946,
-      "learning_rate": 0.00019998965573353526,
-      "loss": 46.0,
-      "step": 28441
-    },
-    {
-      "epoch": 4.580377631949757,
-      "grad_norm": 0.002107944805175066,
-      "learning_rate": 0.00019998965500585985,
-      "loss": 46.0,
-      "step": 28442
-    },
-    {
-      "epoch": 4.5805386690285435,
-      "grad_norm": 0.014695961959660053,
-      "learning_rate": 0.0001999896542781588,
-      "loss": 46.0,
-      "step": 28443
-    },
-    {
-      "epoch": 4.580699706107331,
-      "grad_norm": 0.002039693295955658,
-      "learning_rate": 0.0001999896535504322,
-      "loss": 46.0,
-      "step": 28444
-    },
-    {
-      "epoch": 4.580860743186118,
-      "grad_norm": 0.0021300979424268007,
-      "learning_rate": 0.00019998965282267997,
-      "loss": 46.0,
-      "step": 28445
-    },
-    {
-      "epoch": 4.581021780264906,
-      "grad_norm": 0.009845821186900139,
-      "learning_rate": 0.0001999896520949022,
-      "loss": 46.0,
-      "step": 28446
-    },
-    {
-      "epoch": 4.581182817343693,
-      "grad_norm": 0.006082086358219385,
-      "learning_rate": 0.00019998965136709882,
-      "loss": 46.0,
-      "step": 28447
-    },
-    {
-      "epoch": 4.581343854422481,
-      "grad_norm": 0.010245672427117825,
-      "learning_rate": 0.0001999896506392698,
-      "loss": 46.0,
-      "step": 28448
-    },
-    {
-      "epoch": 4.581504891501268,
-      "grad_norm": 0.005928464233875275,
-      "learning_rate": 0.00019998964991141527,
-      "loss": 46.0,
-      "step": 28449
-    },
-    {
-      "epoch": 4.581665928580056,
-      "grad_norm": 0.006840656977146864,
-      "learning_rate": 0.00019998964918353508,
-      "loss": 46.0,
-      "step": 28450
-    },
-    {
-      "epoch": 4.581826965658843,
-      "grad_norm": 0.0006742523983120918,
-      "learning_rate": 0.00019998964845562936,
-      "loss": 46.0,
-      "step": 28451
-    },
-    {
-      "epoch": 4.5819880027376305,
-      "grad_norm": 0.008705566637217999,
-      "learning_rate": 0.000199989647727698,
-      "loss": 46.0,
-      "step": 28452
-    },
-    {
-      "epoch": 4.582149039816418,
-      "grad_norm": 0.0007399450987577438,
-      "learning_rate": 0.00019998964699974105,
-      "loss": 46.0,
-      "step": 28453
-    },
-    {
-      "epoch": 4.582310076895205,
-      "grad_norm": 0.0040413024835288525,
-      "learning_rate": 0.00019998964627175855,
-      "loss": 46.0,
-      "step": 28454
-    },
-    {
-      "epoch": 4.582471113973993,
-      "grad_norm": 0.0055252546444535255,
-      "learning_rate": 0.0001999896455437504,
-      "loss": 46.0,
-      "step": 28455
-    },
-    {
-      "epoch": 4.582632151052779,
-      "grad_norm": 0.0025090461131185293,
-      "learning_rate": 0.00019998964481571666,
-      "loss": 46.0,
-      "step": 28456
-    },
-    {
-      "epoch": 4.582793188131568,
-      "grad_norm": 0.00297033553943038,
-      "learning_rate": 0.00019998964408765737,
-      "loss": 46.0,
-      "step": 28457
-    },
-    {
-      "epoch": 4.582954225210354,
-      "grad_norm": 0.004799159709364176,
-      "learning_rate": 0.00019998964335957246,
-      "loss": 46.0,
-      "step": 28458
-    },
-    {
-      "epoch": 4.583115262289142,
-      "grad_norm": 0.0039450605399906635,
-      "learning_rate": 0.000199989642631462,
-      "loss": 46.0,
-      "step": 28459
-    },
-    {
-      "epoch": 4.583276299367929,
-      "grad_norm": 0.006026839837431908,
-      "learning_rate": 0.0001999896419033259,
-      "loss": 46.0,
-      "step": 28460
-    },
-    {
-      "epoch": 4.583437336446717,
-      "grad_norm": 0.0046498412266373634,
-      "learning_rate": 0.0001999896411751642,
-      "loss": 46.0,
-      "step": 28461
-    },
-    {
-      "epoch": 4.583598373525504,
-      "grad_norm": 0.007260121870785952,
-      "learning_rate": 0.00019998964044697694,
-      "loss": 46.0,
-      "step": 28462
-    },
-    {
-      "epoch": 4.583759410604292,
-      "grad_norm": 0.0072744861245155334,
-      "learning_rate": 0.00019998963971876407,
-      "loss": 46.0,
-      "step": 28463
-    },
-    {
-      "epoch": 4.583920447683079,
-      "grad_norm": 0.004115426447242498,
-      "learning_rate": 0.00019998963899052564,
-      "loss": 46.0,
-      "step": 28464
-    },
-    {
-      "epoch": 4.5840814847618665,
-      "grad_norm": 0.006077341735363007,
-      "learning_rate": 0.00019998963826226156,
-      "loss": 46.0,
-      "step": 28465
-    },
-    {
-      "epoch": 4.584242521840654,
-      "grad_norm": 0.0024771972093731165,
-      "learning_rate": 0.00019998963753397193,
-      "loss": 46.0,
-      "step": 28466
-    },
-    {
-      "epoch": 4.584403558919441,
-      "grad_norm": 0.00158779660705477,
-      "learning_rate": 0.0001999896368056567,
-      "loss": 46.0,
-      "step": 28467
-    },
-    {
-      "epoch": 4.584564595998229,
-      "grad_norm": 0.0031605777330696583,
-      "learning_rate": 0.00019998963607731587,
-      "loss": 46.0,
-      "step": 28468
-    },
-    {
-      "epoch": 4.584725633077016,
-      "grad_norm": 0.005443513859063387,
-      "learning_rate": 0.00019998963534894945,
-      "loss": 46.0,
-      "step": 28469
-    },
-    {
-      "epoch": 4.584886670155804,
-      "grad_norm": 0.0021112828981131315,
-      "learning_rate": 0.00019998963462055747,
-      "loss": 46.0,
-      "step": 28470
-    },
-    {
-      "epoch": 4.58504770723459,
-      "grad_norm": 0.0065785483457148075,
-      "learning_rate": 0.00019998963389213984,
-      "loss": 46.0,
-      "step": 28471
-    },
-    {
-      "epoch": 4.585208744313378,
-      "grad_norm": 0.004789309576153755,
-      "learning_rate": 0.00019998963316369666,
-      "loss": 46.0,
-      "step": 28472
-    },
-    {
-      "epoch": 4.585369781392165,
-      "grad_norm": 0.0035931633319705725,
-      "learning_rate": 0.00019998963243522786,
-      "loss": 46.0,
-      "step": 28473
-    },
-    {
-      "epoch": 4.585530818470953,
-      "grad_norm": 0.005356601905077696,
-      "learning_rate": 0.00019998963170673347,
-      "loss": 46.0,
-      "step": 28474
-    },
-    {
-      "epoch": 4.58569185554974,
-      "grad_norm": 0.014345882460474968,
-      "learning_rate": 0.00019998963097821352,
-      "loss": 46.0,
-      "step": 28475
-    },
-    {
-      "epoch": 4.585852892628528,
-      "grad_norm": 0.003207128494977951,
-      "learning_rate": 0.00019998963024966794,
-      "loss": 46.0,
-      "step": 28476
-    },
-    {
-      "epoch": 4.586013929707315,
-      "grad_norm": 0.00799788348376751,
-      "learning_rate": 0.00019998962952109682,
-      "loss": 46.0,
-      "step": 28477
-    },
-    {
-      "epoch": 4.5861749667861025,
-      "grad_norm": 0.005665863864123821,
-      "learning_rate": 0.00019998962879250005,
-      "loss": 46.0,
-      "step": 28478
-    },
-    {
-      "epoch": 4.58633600386489,
-      "grad_norm": 0.0017914465861395001,
-      "learning_rate": 0.00019998962806387773,
-      "loss": 46.0,
-      "step": 28479
-    },
-    {
-      "epoch": 4.586497040943677,
-      "grad_norm": 0.000754940789192915,
-      "learning_rate": 0.0001999896273352298,
-      "loss": 46.0,
-      "step": 28480
-    },
-    {
-      "epoch": 4.586658078022465,
-      "grad_norm": 0.002521541900932789,
-      "learning_rate": 0.00019998962660655627,
-      "loss": 46.0,
-      "step": 28481
-    },
-    {
-      "epoch": 4.586819115101252,
-      "grad_norm": 0.008795598521828651,
-      "learning_rate": 0.00019998962587785716,
-      "loss": 46.0,
-      "step": 28482
-    },
-    {
-      "epoch": 4.58698015218004,
-      "grad_norm": 0.0020332983694970608,
-      "learning_rate": 0.00019998962514913244,
-      "loss": 46.0,
-      "step": 28483
-    },
-    {
-      "epoch": 4.587141189258827,
-      "grad_norm": 0.006329297088086605,
-      "learning_rate": 0.00019998962442038213,
-      "loss": 46.0,
-      "step": 28484
-    },
-    {
-      "epoch": 4.587302226337615,
-      "grad_norm": 0.0040099602192640305,
-      "learning_rate": 0.00019998962369160625,
-      "loss": 46.0,
-      "step": 28485
-    },
-    {
-      "epoch": 4.587463263416401,
-      "grad_norm": 0.003025706624612212,
-      "learning_rate": 0.00019998962296280477,
-      "loss": 46.0,
-      "step": 28486
-    },
-    {
-      "epoch": 4.587624300495189,
-      "grad_norm": 0.004924978595227003,
-      "learning_rate": 0.0001999896222339777,
-      "loss": 46.0,
-      "step": 28487
-    },
-    {
-      "epoch": 4.587785337573976,
-      "grad_norm": 0.0020735817961394787,
-      "learning_rate": 0.00019998962150512503,
-      "loss": 46.0,
-      "step": 28488
-    },
-    {
-      "epoch": 4.587946374652764,
-      "grad_norm": 0.0021971550304442644,
-      "learning_rate": 0.00019998962077624676,
-      "loss": 46.0,
-      "step": 28489
-    },
-    {
-      "epoch": 4.588107411731551,
-      "grad_norm": 0.0023650459479540586,
-      "learning_rate": 0.00019998962004734292,
-      "loss": 46.0,
-      "step": 28490
-    },
-    {
-      "epoch": 4.5882684488103385,
-      "grad_norm": 0.005954652093350887,
-      "learning_rate": 0.00019998961931841347,
-      "loss": 46.0,
-      "step": 28491
-    },
-    {
-      "epoch": 4.588429485889126,
-      "grad_norm": 0.002070229034870863,
-      "learning_rate": 0.00019998961858945844,
-      "loss": 46.0,
-      "step": 28492
-    },
-    {
-      "epoch": 4.588590522967913,
-      "grad_norm": 0.0014339193003252149,
-      "learning_rate": 0.00019998961786047779,
-      "loss": 46.0,
-      "step": 28493
-    },
-    {
-      "epoch": 4.588751560046701,
-      "grad_norm": 0.005268601235002279,
-      "learning_rate": 0.00019998961713147158,
-      "loss": 46.0,
-      "step": 28494
-    },
-    {
-      "epoch": 4.588912597125488,
-      "grad_norm": 0.006824306212365627,
-      "learning_rate": 0.00019998961640243975,
-      "loss": 46.0,
-      "step": 28495
-    },
-    {
-      "epoch": 4.589073634204276,
-      "grad_norm": 0.0018382209818810225,
-      "learning_rate": 0.00019998961567338234,
-      "loss": 46.0,
-      "step": 28496
-    },
-    {
-      "epoch": 4.589234671283063,
-      "grad_norm": 0.015351309441030025,
-      "learning_rate": 0.00019998961494429937,
-      "loss": 46.0,
-      "step": 28497
-    },
-    {
-      "epoch": 4.589395708361851,
-      "grad_norm": 0.0014979561092332006,
-      "learning_rate": 0.00019998961421519078,
-      "loss": 46.0,
-      "step": 28498
-    },
-    {
-      "epoch": 4.589556745440638,
-      "grad_norm": 0.0005707125528715551,
-      "learning_rate": 0.00019998961348605658,
-      "loss": 46.0,
-      "step": 28499
-    },
-    {
-      "epoch": 4.5897177825194255,
-      "grad_norm": 0.007236639969050884,
-      "learning_rate": 0.00019998961275689682,
-      "loss": 46.0,
-      "step": 28500
-    },
-    {
-      "epoch": 4.589878819598212,
-      "grad_norm": 0.0015959461452439427,
-      "learning_rate": 0.00019998961202771144,
-      "loss": 46.0,
-      "step": 28501
-    },
-    {
-      "epoch": 4.5900398566769995,
-      "grad_norm": 0.003295185277238488,
-      "learning_rate": 0.00019998961129850048,
-      "loss": 46.0,
-      "step": 28502
-    },
-    {
-      "epoch": 4.590200893755787,
-      "grad_norm": 0.004769337363541126,
-      "learning_rate": 0.00019998961056926393,
-      "loss": 46.0,
-      "step": 28503
-    },
-    {
-      "epoch": 4.5903619308345744,
-      "grad_norm": 0.0012110670795664191,
-      "learning_rate": 0.0001999896098400018,
-      "loss": 46.0,
-      "step": 28504
-    },
-    {
-      "epoch": 4.590522967913362,
-      "grad_norm": 0.014359165914356709,
-      "learning_rate": 0.00019998960911071404,
-      "loss": 46.0,
-      "step": 28505
-    },
-    {
-      "epoch": 4.590684004992149,
-      "grad_norm": 0.001631799153983593,
-      "learning_rate": 0.00019998960838140073,
-      "loss": 46.0,
-      "step": 28506
-    },
-    {
-      "epoch": 4.590845042070937,
-      "grad_norm": 0.005717022810131311,
-      "learning_rate": 0.0001999896076520618,
-      "loss": 46.0,
-      "step": 28507
-    },
-    {
-      "epoch": 4.591006079149724,
-      "grad_norm": 0.007634556386619806,
-      "learning_rate": 0.0001999896069226973,
-      "loss": 46.0,
-      "step": 28508
-    },
-    {
-      "epoch": 4.591167116228512,
-      "grad_norm": 0.004141835495829582,
-      "learning_rate": 0.00019998960619330717,
-      "loss": 46.0,
-      "step": 28509
-    },
-    {
-      "epoch": 4.591328153307299,
-      "grad_norm": 0.005560701712965965,
-      "learning_rate": 0.0001999896054638915,
-      "loss": 46.0,
-      "step": 28510
-    },
-    {
-      "epoch": 4.591489190386087,
-      "grad_norm": 0.002460506744682789,
-      "learning_rate": 0.00019998960473445018,
-      "loss": 46.0,
-      "step": 28511
-    },
-    {
-      "epoch": 4.591650227464874,
-      "grad_norm": 0.003538668854162097,
-      "learning_rate": 0.00019998960400498332,
-      "loss": 46.0,
-      "step": 28512
-    },
-    {
-      "epoch": 4.5918112645436615,
-      "grad_norm": 0.0013440271141007543,
-      "learning_rate": 0.00019998960327549084,
-      "loss": 46.0,
-      "step": 28513
-    },
-    {
-      "epoch": 4.591972301622448,
-      "grad_norm": 0.0025261377450078726,
-      "learning_rate": 0.00019998960254597278,
-      "loss": 46.0,
-      "step": 28514
-    },
-    {
-      "epoch": 4.592133338701236,
-      "grad_norm": 0.013304091058671474,
-      "learning_rate": 0.00019998960181642913,
-      "loss": 46.0,
-      "step": 28515
-    },
-    {
-      "epoch": 4.592294375780023,
-      "grad_norm": 0.003078119596466422,
-      "learning_rate": 0.00019998960108685986,
-      "loss": 46.0,
-      "step": 28516
-    },
-    {
-      "epoch": 4.59245541285881,
-      "grad_norm": 0.0027525704354047775,
-      "learning_rate": 0.00019998960035726504,
-      "loss": 46.0,
-      "step": 28517
-    },
-    {
-      "epoch": 4.592616449937598,
-      "grad_norm": 0.002791887614876032,
-      "learning_rate": 0.00019998959962764457,
-      "loss": 46.0,
-      "step": 28518
-    },
-    {
-      "epoch": 4.592777487016385,
-      "grad_norm": 0.0026647874619811773,
-      "learning_rate": 0.00019998959889799855,
-      "loss": 46.0,
-      "step": 28519
-    },
-    {
-      "epoch": 4.592938524095173,
-      "grad_norm": 0.001245353021658957,
-      "learning_rate": 0.00019998959816832694,
-      "loss": 46.0,
-      "step": 28520
-    },
-    {
-      "epoch": 4.59309956117396,
-      "grad_norm": 0.0008130603237077594,
-      "learning_rate": 0.00019998959743862976,
-      "loss": 46.0,
-      "step": 28521
-    },
-    {
-      "epoch": 4.593260598252748,
-      "grad_norm": 0.0005793844466097653,
-      "learning_rate": 0.00019998959670890692,
-      "loss": 46.0,
-      "step": 28522
-    },
-    {
-      "epoch": 4.593421635331535,
-      "grad_norm": 0.0012297973735257983,
-      "learning_rate": 0.00019998959597915855,
-      "loss": 46.0,
-      "step": 28523
-    },
-    {
-      "epoch": 4.593582672410323,
-      "grad_norm": 0.006347378715872765,
-      "learning_rate": 0.00019998959524938456,
-      "loss": 46.0,
-      "step": 28524
-    },
-    {
-      "epoch": 4.59374370948911,
-      "grad_norm": 0.003759901039302349,
-      "learning_rate": 0.00019998959451958496,
-      "loss": 46.0,
-      "step": 28525
-    },
-    {
-      "epoch": 4.5939047465678975,
-      "grad_norm": 0.006052514538168907,
-      "learning_rate": 0.0001999895937897598,
-      "loss": 46.0,
-      "step": 28526
-    },
-    {
-      "epoch": 4.594065783646685,
-      "grad_norm": 0.0020625251345336437,
-      "learning_rate": 0.00019998959305990904,
-      "loss": 46.0,
-      "step": 28527
-    },
-    {
-      "epoch": 4.594226820725472,
-      "grad_norm": 0.0017713834531605244,
-      "learning_rate": 0.00019998959233003268,
-      "loss": 46.0,
-      "step": 28528
-    },
-    {
-      "epoch": 4.594387857804259,
-      "grad_norm": 0.0014661067398265004,
-      "learning_rate": 0.00019998959160013073,
-      "loss": 46.0,
-      "step": 28529
-    },
-    {
-      "epoch": 4.594548894883047,
-      "grad_norm": 0.0017022364772856236,
-      "learning_rate": 0.0001999895908702032,
-      "loss": 46.0,
-      "step": 28530
-    },
-    {
-      "epoch": 4.594709931961834,
-      "grad_norm": 0.0057439361698925495,
-      "learning_rate": 0.00019998959014025006,
-      "loss": 46.0,
-      "step": 28531
-    },
-    {
-      "epoch": 4.594870969040621,
-      "grad_norm": 0.005177950952202082,
-      "learning_rate": 0.00019998958941027132,
-      "loss": 46.0,
-      "step": 28532
-    },
-    {
-      "epoch": 4.595032006119409,
-      "grad_norm": 0.006061226595193148,
-      "learning_rate": 0.000199989588680267,
-      "loss": 46.0,
-      "step": 28533
-    },
-    {
-      "epoch": 4.595193043198196,
-      "grad_norm": 0.006619125138968229,
-      "learning_rate": 0.0001999895879502371,
-      "loss": 46.0,
-      "step": 28534
-    },
-    {
-      "epoch": 4.595354080276984,
-      "grad_norm": 0.0022555554751306772,
-      "learning_rate": 0.0001999895872201816,
-      "loss": 46.0,
-      "step": 28535
-    },
-    {
-      "epoch": 4.595515117355771,
-      "grad_norm": 0.006830146536231041,
-      "learning_rate": 0.00019998958649010052,
-      "loss": 46.0,
-      "step": 28536
-    },
-    {
-      "epoch": 4.595676154434559,
-      "grad_norm": 0.007183676119893789,
-      "learning_rate": 0.0001999895857599938,
-      "loss": 46.0,
-      "step": 28537
-    },
-    {
-      "epoch": 4.595837191513346,
-      "grad_norm": 0.020978108048439026,
-      "learning_rate": 0.00019998958502986155,
-      "loss": 46.0,
-      "step": 28538
-    },
-    {
-      "epoch": 4.5959982285921335,
-      "grad_norm": 0.0036064700689166784,
-      "learning_rate": 0.00019998958429970367,
-      "loss": 46.0,
-      "step": 28539
-    },
-    {
-      "epoch": 4.596159265670921,
-      "grad_norm": 0.0031953167635947466,
-      "learning_rate": 0.0001999895835695202,
-      "loss": 46.0,
-      "step": 28540
-    },
-    {
-      "epoch": 4.596320302749708,
-      "grad_norm": 0.009189536795020103,
-      "learning_rate": 0.00019998958283931116,
-      "loss": 46.0,
-      "step": 28541
-    },
-    {
-      "epoch": 4.596481339828496,
-      "grad_norm": 0.0021295237820595503,
-      "learning_rate": 0.0001999895821090765,
-      "loss": 46.0,
-      "step": 28542
-    },
-    {
-      "epoch": 4.596642376907283,
-      "grad_norm": 0.004190010018646717,
-      "learning_rate": 0.00019998958137881626,
-      "loss": 46.0,
-      "step": 28543
-    },
-    {
-      "epoch": 4.59680341398607,
-      "grad_norm": 0.005775100085884333,
-      "learning_rate": 0.00019998958064853045,
-      "loss": 46.0,
-      "step": 28544
-    },
-    {
-      "epoch": 4.596964451064858,
-      "grad_norm": 0.004560623783618212,
-      "learning_rate": 0.000199989579918219,
-      "loss": 46.0,
-      "step": 28545
-    },
-    {
-      "epoch": 4.597125488143645,
-      "grad_norm": 0.010999364778399467,
-      "learning_rate": 0.00019998957918788198,
-      "loss": 46.0,
-      "step": 28546
-    },
-    {
-      "epoch": 4.597286525222432,
-      "grad_norm": 0.003572491928935051,
-      "learning_rate": 0.00019998957845751938,
-      "loss": 46.0,
-      "step": 28547
-    },
-    {
-      "epoch": 4.59744756230122,
-      "grad_norm": 0.0017069445457309484,
-      "learning_rate": 0.0001999895777271312,
-      "loss": 46.0,
-      "step": 28548
-    },
-    {
-      "epoch": 4.597608599380007,
-      "grad_norm": 0.004072308074682951,
-      "learning_rate": 0.00019998957699671739,
-      "loss": 46.0,
-      "step": 28549
-    },
-    {
-      "epoch": 4.5977696364587946,
-      "grad_norm": 0.005793790332973003,
-      "learning_rate": 0.00019998957626627802,
-      "loss": 46.0,
-      "step": 28550
-    },
-    {
-      "epoch": 4.597930673537582,
-      "grad_norm": 0.010596828535199165,
-      "learning_rate": 0.00019998957553581304,
-      "loss": 46.0,
-      "step": 28551
-    },
-    {
-      "epoch": 4.5980917106163695,
-      "grad_norm": 0.00041604763828217983,
-      "learning_rate": 0.00019998957480532245,
-      "loss": 46.0,
-      "step": 28552
-    },
-    {
-      "epoch": 4.598252747695157,
-      "grad_norm": 0.0032967927400022745,
-      "learning_rate": 0.0001999895740748063,
-      "loss": 46.0,
-      "step": 28553
-    },
-    {
-      "epoch": 4.598413784773944,
-      "grad_norm": 0.0014154138043522835,
-      "learning_rate": 0.00019998957334426453,
-      "loss": 46.0,
-      "step": 28554
-    },
-    {
-      "epoch": 4.598574821852732,
-      "grad_norm": 0.005878071766346693,
-      "learning_rate": 0.0001999895726136972,
-      "loss": 46.0,
-      "step": 28555
-    },
-    {
-      "epoch": 4.598735858931519,
-      "grad_norm": 0.0025634162593632936,
-      "learning_rate": 0.00019998957188310427,
-      "loss": 46.0,
-      "step": 28556
-    },
-    {
-      "epoch": 4.598896896010307,
-      "grad_norm": 0.004028363153338432,
-      "learning_rate": 0.00019998957115248574,
-      "loss": 46.0,
-      "step": 28557
-    },
-    {
-      "epoch": 4.599057933089094,
-      "grad_norm": 0.0024117566645145416,
-      "learning_rate": 0.00019998957042184162,
-      "loss": 46.0,
-      "step": 28558
-    },
-    {
-      "epoch": 4.599218970167881,
-      "grad_norm": 0.0055236779153347015,
-      "learning_rate": 0.0001999895696911719,
-      "loss": 46.0,
-      "step": 28559
-    },
-    {
-      "epoch": 4.599380007246668,
-      "grad_norm": 0.002532748971134424,
-      "learning_rate": 0.00019998956896047658,
-      "loss": 46.0,
-      "step": 28560
-    },
-    {
-      "epoch": 4.599541044325456,
-      "grad_norm": 0.001568284467794001,
-      "learning_rate": 0.00019998956822975567,
-      "loss": 46.0,
-      "step": 28561
-    },
-    {
-      "epoch": 4.599702081404243,
-      "grad_norm": 0.006054223515093327,
-      "learning_rate": 0.00019998956749900918,
-      "loss": 46.0,
-      "step": 28562
-    },
-    {
-      "epoch": 4.5998631184830305,
-      "grad_norm": 0.006243539042770863,
-      "learning_rate": 0.0001999895667682371,
-      "loss": 46.0,
-      "step": 28563
-    },
-    {
-      "epoch": 4.600024155561818,
-      "grad_norm": 0.015282531268894672,
-      "learning_rate": 0.00019998956603743944,
-      "loss": 46.0,
-      "step": 28564
-    },
-    {
-      "epoch": 4.600185192640605,
-      "grad_norm": 0.006074786651879549,
-      "learning_rate": 0.00019998956530661616,
-      "loss": 46.0,
-      "step": 28565
-    },
-    {
-      "epoch": 4.600346229719393,
-      "grad_norm": 0.00400999141857028,
-      "learning_rate": 0.0001999895645757673,
-      "loss": 46.0,
-      "step": 28566
-    },
-    {
-      "epoch": 4.60050726679818,
-      "grad_norm": 0.00989500805735588,
-      "learning_rate": 0.00019998956384489284,
-      "loss": 46.0,
-      "step": 28567
-    },
-    {
-      "epoch": 4.600668303876968,
-      "grad_norm": 0.004770105239003897,
-      "learning_rate": 0.0001999895631139928,
-      "loss": 46.0,
-      "step": 28568
-    },
-    {
-      "epoch": 4.600829340955755,
-      "grad_norm": 0.007653380744159222,
-      "learning_rate": 0.00019998956238306717,
-      "loss": 46.0,
-      "step": 28569
-    },
-    {
-      "epoch": 4.600990378034543,
-      "grad_norm": 0.005711417179554701,
-      "learning_rate": 0.00019998956165211593,
-      "loss": 46.0,
-      "step": 28570
-    },
-    {
-      "epoch": 4.60115141511333,
-      "grad_norm": 0.0066521973349153996,
-      "learning_rate": 0.00019998956092113913,
-      "loss": 46.0,
-      "step": 28571
-    },
-    {
-      "epoch": 4.601312452192118,
-      "grad_norm": 0.0035176360979676247,
-      "learning_rate": 0.0001999895601901367,
-      "loss": 46.0,
-      "step": 28572
-    },
-    {
-      "epoch": 4.601473489270905,
-      "grad_norm": 0.02367776818573475,
-      "learning_rate": 0.00019998955945910868,
-      "loss": 46.0,
-      "step": 28573
-    },
-    {
-      "epoch": 4.601634526349692,
-      "grad_norm": 0.0005685294163413346,
-      "learning_rate": 0.0001999895587280551,
-      "loss": 46.0,
-      "step": 28574
-    },
-    {
-      "epoch": 4.601795563428479,
-      "grad_norm": 0.0103748245164752,
-      "learning_rate": 0.0001999895579969759,
-      "loss": 46.0,
-      "step": 28575
-    },
-    {
-      "epoch": 4.6019566005072665,
-      "grad_norm": 0.006886561866849661,
-      "learning_rate": 0.00019998955726587112,
-      "loss": 46.0,
-      "step": 28576
-    },
-    {
-      "epoch": 4.602117637586054,
-      "grad_norm": 0.00479584326967597,
-      "learning_rate": 0.00019998955653474076,
-      "loss": 46.0,
-      "step": 28577
-    },
-    {
-      "epoch": 4.602278674664841,
-      "grad_norm": 0.006385931745171547,
-      "learning_rate": 0.0001999895558035848,
-      "loss": 46.0,
-      "step": 28578
-    },
-    {
-      "epoch": 4.602439711743629,
-      "grad_norm": 0.005352462641894817,
-      "learning_rate": 0.00019998955507240321,
-      "loss": 46.0,
-      "step": 28579
-    },
-    {
-      "epoch": 4.602600748822416,
-      "grad_norm": 0.0076520005241036415,
-      "learning_rate": 0.00019998955434119607,
-      "loss": 46.0,
-      "step": 28580
-    },
-    {
-      "epoch": 4.602761785901204,
-      "grad_norm": 0.0023984024301171303,
-      "learning_rate": 0.00019998955360996332,
-      "loss": 46.0,
-      "step": 28581
-    },
-    {
-      "epoch": 4.602922822979991,
-      "grad_norm": 0.007030025590211153,
-      "learning_rate": 0.00019998955287870498,
-      "loss": 46.0,
-      "step": 28582
-    },
-    {
-      "epoch": 4.603083860058779,
-      "grad_norm": 0.001687400508671999,
-      "learning_rate": 0.00019998955214742105,
-      "loss": 46.0,
-      "step": 28583
-    },
-    {
-      "epoch": 4.603244897137566,
-      "grad_norm": 0.008841497823596,
-      "learning_rate": 0.00019998955141611156,
-      "loss": 46.0,
-      "step": 28584
-    },
-    {
-      "epoch": 4.603405934216354,
-      "grad_norm": 0.001367873279377818,
-      "learning_rate": 0.00019998955068477642,
-      "loss": 46.0,
-      "step": 28585
-    },
-    {
-      "epoch": 4.603566971295141,
-      "grad_norm": 0.0015143114142119884,
-      "learning_rate": 0.00019998954995341573,
-      "loss": 46.0,
-      "step": 28586
-    },
-    {
-      "epoch": 4.6037280083739285,
-      "grad_norm": 0.0026926349382847548,
-      "learning_rate": 0.00019998954922202943,
-      "loss": 46.0,
-      "step": 28587
-    },
-    {
-      "epoch": 4.603889045452716,
-      "grad_norm": 0.00903333630412817,
-      "learning_rate": 0.00019998954849061753,
-      "loss": 46.0,
-      "step": 28588
-    },
-    {
-      "epoch": 4.6040500825315025,
-      "grad_norm": 0.0016941786743700504,
-      "learning_rate": 0.00019998954775918005,
-      "loss": 46.0,
-      "step": 28589
-    },
-    {
-      "epoch": 4.60421111961029,
-      "grad_norm": 0.01835503615438938,
-      "learning_rate": 0.00019998954702771696,
-      "loss": 46.0,
-      "step": 28590
-    },
-    {
-      "epoch": 4.604372156689077,
-      "grad_norm": 0.0007851842674426734,
-      "learning_rate": 0.00019998954629622833,
-      "loss": 46.0,
-      "step": 28591
-    },
-    {
-      "epoch": 4.604533193767865,
-      "grad_norm": 0.004995634313672781,
-      "learning_rate": 0.00019998954556471406,
-      "loss": 46.0,
-      "step": 28592
-    },
-    {
-      "epoch": 4.604694230846652,
-      "grad_norm": 0.007921691983938217,
-      "learning_rate": 0.0001999895448331742,
-      "loss": 46.0,
-      "step": 28593
-    },
-    {
-      "epoch": 4.60485526792544,
-      "grad_norm": 0.009077431634068489,
-      "learning_rate": 0.00019998954410160877,
-      "loss": 46.0,
-      "step": 28594
-    },
-    {
-      "epoch": 4.605016305004227,
-      "grad_norm": 0.009035086259245872,
-      "learning_rate": 0.00019998954337001773,
-      "loss": 46.0,
-      "step": 28595
-    },
-    {
-      "epoch": 4.605177342083015,
-      "grad_norm": 0.004378116223961115,
-      "learning_rate": 0.00019998954263840112,
-      "loss": 46.0,
-      "step": 28596
-    },
-    {
-      "epoch": 4.605338379161802,
-      "grad_norm": 0.006223103031516075,
-      "learning_rate": 0.00019998954190675889,
-      "loss": 46.0,
-      "step": 28597
-    },
-    {
-      "epoch": 4.60549941624059,
-      "grad_norm": 0.002344702137634158,
-      "learning_rate": 0.00019998954117509107,
-      "loss": 46.0,
-      "step": 28598
-    },
-    {
-      "epoch": 4.605660453319377,
-      "grad_norm": 0.005434425547719002,
-      "learning_rate": 0.0001999895404433977,
-      "loss": 46.0,
-      "step": 28599
-    },
-    {
-      "epoch": 4.6058214903981645,
-      "grad_norm": 0.005976187530905008,
-      "learning_rate": 0.0001999895397116787,
-      "loss": 46.0,
-      "step": 28600
-    },
-    {
-      "epoch": 4.605982527476952,
-      "grad_norm": 0.023467248305678368,
-      "learning_rate": 0.0001999895389799341,
-      "loss": 46.0,
-      "step": 28601
-    },
-    {
-      "epoch": 4.6061435645557385,
-      "grad_norm": 0.0029122892301529646,
-      "learning_rate": 0.00019998953824816392,
-      "loss": 46.0,
-      "step": 28602
-    },
-    {
-      "epoch": 4.606304601634527,
-      "grad_norm": 0.0015702040400356054,
-      "learning_rate": 0.00019998953751636817,
-      "loss": 46.0,
-      "step": 28603
-    },
-    {
-      "epoch": 4.606465638713313,
-      "grad_norm": 0.0009783548302948475,
-      "learning_rate": 0.0001999895367845468,
-      "loss": 46.0,
-      "step": 28604
-    },
-    {
-      "epoch": 4.606626675792101,
-      "grad_norm": 0.0019342172890901566,
-      "learning_rate": 0.00019998953605269981,
-      "loss": 46.0,
-      "step": 28605
-    },
-    {
-      "epoch": 4.606787712870888,
-      "grad_norm": 0.00224115327000618,
-      "learning_rate": 0.0001999895353208273,
-      "loss": 46.0,
-      "step": 28606
-    },
-    {
-      "epoch": 4.606948749949676,
-      "grad_norm": 0.005332854110747576,
-      "learning_rate": 0.00019998953458892914,
-      "loss": 46.0,
-      "step": 28607
-    },
-    {
-      "epoch": 4.607109787028463,
-      "grad_norm": 0.0061399005353450775,
-      "learning_rate": 0.0001999895338570054,
-      "loss": 46.0,
-      "step": 28608
-    },
-    {
-      "epoch": 4.607270824107251,
-      "grad_norm": 0.006406235974282026,
-      "learning_rate": 0.0001999895331250561,
-      "loss": 46.0,
-      "step": 28609
-    },
-    {
-      "epoch": 4.607431861186038,
-      "grad_norm": 0.012562812305986881,
-      "learning_rate": 0.00019998953239308117,
-      "loss": 46.0,
-      "step": 28610
-    },
-    {
-      "epoch": 4.6075928982648255,
-      "grad_norm": 0.004897661041468382,
-      "learning_rate": 0.00019998953166108067,
-      "loss": 46.0,
-      "step": 28611
-    },
-    {
-      "epoch": 4.607753935343613,
-      "grad_norm": 0.0029372184071689844,
-      "learning_rate": 0.00019998953092905457,
-      "loss": 46.0,
-      "step": 28612
-    },
-    {
-      "epoch": 4.6079149724224004,
-      "grad_norm": 0.0026479947846382856,
-      "learning_rate": 0.00019998953019700286,
-      "loss": 46.0,
-      "step": 28613
-    },
-    {
-      "epoch": 4.608076009501188,
-      "grad_norm": 0.005616776645183563,
-      "learning_rate": 0.00019998952946492557,
-      "loss": 46.0,
-      "step": 28614
-    },
-    {
-      "epoch": 4.608237046579975,
-      "grad_norm": 0.005181202199310064,
-      "learning_rate": 0.00019998952873282271,
-      "loss": 46.0,
-      "step": 28615
-    },
-    {
-      "epoch": 4.608398083658763,
-      "grad_norm": 0.0068702129647135735,
-      "learning_rate": 0.00019998952800069422,
-      "loss": 46.0,
-      "step": 28616
-    },
-    {
-      "epoch": 4.608559120737549,
-      "grad_norm": 0.001958029344677925,
-      "learning_rate": 0.00019998952726854016,
-      "loss": 46.0,
-      "step": 28617
-    },
-    {
-      "epoch": 4.608720157816338,
-      "grad_norm": 0.0015985690988600254,
-      "learning_rate": 0.0001999895265363605,
-      "loss": 46.0,
-      "step": 28618
-    },
-    {
-      "epoch": 4.608881194895124,
-      "grad_norm": 0.0019705630838871002,
-      "learning_rate": 0.00019998952580415526,
-      "loss": 46.0,
-      "step": 28619
-    },
-    {
-      "epoch": 4.609042231973912,
-      "grad_norm": 0.0012102543842047453,
-      "learning_rate": 0.00019998952507192442,
-      "loss": 46.0,
-      "step": 28620
-    },
-    {
-      "epoch": 4.609203269052699,
-      "grad_norm": 0.0038378890603780746,
-      "learning_rate": 0.00019998952433966798,
-      "loss": 46.0,
-      "step": 28621
-    },
-    {
-      "epoch": 4.609364306131487,
-      "grad_norm": 0.0026133048813790083,
-      "learning_rate": 0.00019998952360738596,
-      "loss": 46.0,
-      "step": 28622
-    },
-    {
-      "epoch": 4.609525343210274,
-      "grad_norm": 0.010411676950752735,
-      "learning_rate": 0.00019998952287507836,
-      "loss": 46.0,
-      "step": 28623
-    },
-    {
-      "epoch": 4.6096863802890615,
-      "grad_norm": 0.006977350451052189,
-      "learning_rate": 0.00019998952214274516,
-      "loss": 46.0,
-      "step": 28624
-    },
-    {
-      "epoch": 4.609847417367849,
-      "grad_norm": 0.0025255039799958467,
-      "learning_rate": 0.00019998952141038636,
-      "loss": 46.0,
-      "step": 28625
-    },
-    {
-      "epoch": 4.610008454446636,
-      "grad_norm": 0.006918781902641058,
-      "learning_rate": 0.00019998952067800193,
-      "loss": 46.0,
-      "step": 28626
-    },
-    {
-      "epoch": 4.610169491525424,
-      "grad_norm": 0.006830655969679356,
-      "learning_rate": 0.00019998951994559198,
-      "loss": 46.0,
-      "step": 28627
-    },
-    {
-      "epoch": 4.610330528604211,
-      "grad_norm": 0.00397122697904706,
-      "learning_rate": 0.0001999895192131564,
-      "loss": 46.0,
-      "step": 28628
-    },
-    {
-      "epoch": 4.610491565682999,
-      "grad_norm": 0.015576930716633797,
-      "learning_rate": 0.00019998951848069523,
-      "loss": 46.0,
-      "step": 28629
-    },
-    {
-      "epoch": 4.610652602761786,
-      "grad_norm": 0.002232639119029045,
-      "learning_rate": 0.00019998951774820845,
-      "loss": 46.0,
-      "step": 28630
-    },
-    {
-      "epoch": 4.610813639840574,
-      "grad_norm": 0.01757732778787613,
-      "learning_rate": 0.0001999895170156961,
-      "loss": 46.0,
-      "step": 28631
-    },
-    {
-      "epoch": 4.61097467691936,
-      "grad_norm": 0.008033502846956253,
-      "learning_rate": 0.00019998951628315815,
-      "loss": 46.0,
-      "step": 28632
-    },
-    {
-      "epoch": 4.611135713998148,
-      "grad_norm": 0.0028197630308568478,
-      "learning_rate": 0.00019998951555059465,
-      "loss": 46.0,
-      "step": 28633
-    },
-    {
-      "epoch": 4.611296751076935,
-      "grad_norm": 0.0018976012943312526,
-      "learning_rate": 0.0001999895148180055,
-      "loss": 46.0,
-      "step": 28634
-    },
-    {
-      "epoch": 4.611457788155723,
-      "grad_norm": 0.003416329389438033,
-      "learning_rate": 0.00019998951408539077,
-      "loss": 46.0,
-      "step": 28635
-    },
-    {
-      "epoch": 4.61161882523451,
-      "grad_norm": 0.008057817816734314,
-      "learning_rate": 0.00019998951335275047,
-      "loss": 46.0,
-      "step": 28636
-    },
-    {
-      "epoch": 4.6117798623132975,
-      "grad_norm": 0.009286249056458473,
-      "learning_rate": 0.00019998951262008456,
-      "loss": 46.0,
-      "step": 28637
-    },
-    {
-      "epoch": 4.611940899392085,
-      "grad_norm": 0.0068008312955498695,
-      "learning_rate": 0.00019998951188739307,
-      "loss": 46.0,
-      "step": 28638
-    },
-    {
-      "epoch": 4.612101936470872,
-      "grad_norm": 0.005206909030675888,
-      "learning_rate": 0.00019998951115467596,
-      "loss": 46.0,
-      "step": 28639
-    },
-    {
-      "epoch": 4.61226297354966,
-      "grad_norm": 0.005285155493766069,
-      "learning_rate": 0.0001999895104219333,
-      "loss": 46.0,
-      "step": 28640
-    },
-    {
-      "epoch": 4.612424010628447,
-      "grad_norm": 0.0038314827252179384,
-      "learning_rate": 0.000199989509689165,
-      "loss": 46.0,
-      "step": 28641
-    },
-    {
-      "epoch": 4.612585047707235,
-      "grad_norm": 0.0022522611543536186,
-      "learning_rate": 0.00019998950895637113,
-      "loss": 46.0,
-      "step": 28642
-    },
-    {
-      "epoch": 4.612746084786022,
-      "grad_norm": 0.002440350130200386,
-      "learning_rate": 0.00019998950822355168,
-      "loss": 46.0,
-      "step": 28643
-    },
-    {
-      "epoch": 4.61290712186481,
-      "grad_norm": 0.0048972065560519695,
-      "learning_rate": 0.00019998950749070663,
-      "loss": 46.0,
-      "step": 28644
-    },
-    {
-      "epoch": 4.613068158943597,
-      "grad_norm": 0.0019255060469731688,
-      "learning_rate": 0.000199989506757836,
-      "loss": 46.0,
-      "step": 28645
-    },
-    {
-      "epoch": 4.613229196022385,
-      "grad_norm": 0.003462818218395114,
-      "learning_rate": 0.00019998950602493975,
-      "loss": 46.0,
-      "step": 28646
-    },
-    {
-      "epoch": 4.613390233101171,
-      "grad_norm": 0.0044579338282346725,
-      "learning_rate": 0.00019998950529201794,
-      "loss": 46.0,
-      "step": 28647
-    },
-    {
-      "epoch": 4.613551270179959,
-      "grad_norm": 0.027491984888911247,
-      "learning_rate": 0.0001999895045590705,
-      "loss": 46.0,
-      "step": 28648
-    },
-    {
-      "epoch": 4.613712307258746,
-      "grad_norm": 0.017997391521930695,
-      "learning_rate": 0.00019998950382609749,
-      "loss": 46.0,
-      "step": 28649
-    },
-    {
-      "epoch": 4.6138733443375335,
-      "grad_norm": 0.005901267286390066,
-      "learning_rate": 0.0001999895030930989,
-      "loss": 46.0,
-      "step": 28650
-    },
-    {
-      "epoch": 4.614034381416321,
-      "grad_norm": 0.0022986619733273983,
-      "learning_rate": 0.0001999895023600747,
-      "loss": 46.0,
-      "step": 28651
-    },
-    {
-      "epoch": 4.614195418495108,
-      "grad_norm": 0.0024556294083595276,
-      "learning_rate": 0.0001999895016270249,
-      "loss": 46.0,
-      "step": 28652
-    },
-    {
-      "epoch": 4.614356455573896,
-      "grad_norm": 0.009372428059577942,
-      "learning_rate": 0.00019998950089394953,
-      "loss": 46.0,
-      "step": 28653
-    },
-    {
-      "epoch": 4.614517492652683,
-      "grad_norm": 0.004906050395220518,
-      "learning_rate": 0.00019998950016084855,
-      "loss": 46.0,
-      "step": 28654
-    },
-    {
-      "epoch": 4.614678529731471,
-      "grad_norm": 0.006026510149240494,
-      "learning_rate": 0.00019998949942772197,
-      "loss": 46.0,
-      "step": 28655
-    },
-    {
-      "epoch": 4.614839566810258,
-      "grad_norm": 0.00670495955273509,
-      "learning_rate": 0.00019998949869456985,
-      "loss": 46.0,
-      "step": 28656
-    },
-    {
-      "epoch": 4.615000603889046,
-      "grad_norm": 0.001298116403631866,
-      "learning_rate": 0.0001999894979613921,
-      "loss": 46.0,
-      "step": 28657
-    },
-    {
-      "epoch": 4.615161640967833,
-      "grad_norm": 0.01055904570966959,
-      "learning_rate": 0.00019998949722818874,
-      "loss": 46.0,
-      "step": 28658
-    },
-    {
-      "epoch": 4.6153226780466206,
-      "grad_norm": 0.0016208699671551585,
-      "learning_rate": 0.00019998949649495978,
-      "loss": 46.0,
-      "step": 28659
-    },
-    {
-      "epoch": 4.615483715125408,
-      "grad_norm": 0.008294446393847466,
-      "learning_rate": 0.00019998949576170526,
-      "loss": 46.0,
-      "step": 28660
-    },
-    {
-      "epoch": 4.6156447522041955,
-      "grad_norm": 0.004737715236842632,
-      "learning_rate": 0.00019998949502842518,
-      "loss": 46.0,
-      "step": 28661
-    },
-    {
-      "epoch": 4.615805789282982,
-      "grad_norm": 0.004632387775927782,
-      "learning_rate": 0.00019998949429511943,
-      "loss": 46.0,
-      "step": 28662
-    },
-    {
-      "epoch": 4.6159668263617695,
-      "grad_norm": 0.0025180247612297535,
-      "learning_rate": 0.00019998949356178815,
-      "loss": 46.0,
-      "step": 28663
-    },
-    {
-      "epoch": 4.616127863440557,
-      "grad_norm": 0.002578171668574214,
-      "learning_rate": 0.00019998949282843125,
-      "loss": 46.0,
-      "step": 28664
-    },
-    {
-      "epoch": 4.616288900519344,
-      "grad_norm": 0.011505522765219212,
-      "learning_rate": 0.00019998949209504876,
-      "loss": 46.0,
-      "step": 28665
-    },
-    {
-      "epoch": 4.616449937598132,
-      "grad_norm": 0.0042459312826395035,
-      "learning_rate": 0.0001999894913616407,
-      "loss": 46.0,
-      "step": 28666
-    },
-    {
-      "epoch": 4.616610974676919,
-      "grad_norm": 0.004854676313698292,
-      "learning_rate": 0.00019998949062820703,
-      "loss": 46.0,
-      "step": 28667
-    },
-    {
-      "epoch": 4.616772011755707,
-      "grad_norm": 0.009851894341409206,
-      "learning_rate": 0.00019998948989474776,
-      "loss": 46.0,
-      "step": 28668
-    },
-    {
-      "epoch": 4.616933048834494,
-      "grad_norm": 0.002170231193304062,
-      "learning_rate": 0.0001999894891612629,
-      "loss": 46.0,
-      "step": 28669
-    },
-    {
-      "epoch": 4.617094085913282,
-      "grad_norm": 0.0005476687219925225,
-      "learning_rate": 0.00019998948842775245,
-      "loss": 46.0,
-      "step": 28670
-    },
-    {
-      "epoch": 4.617255122992069,
-      "grad_norm": 0.004725928883999586,
-      "learning_rate": 0.00019998948769421642,
-      "loss": 46.0,
-      "step": 28671
-    },
-    {
-      "epoch": 4.6174161600708565,
-      "grad_norm": 0.0034042028710246086,
-      "learning_rate": 0.00019998948696065477,
-      "loss": 46.0,
-      "step": 28672
-    },
-    {
-      "epoch": 4.617577197149644,
-      "grad_norm": 0.004970040637999773,
-      "learning_rate": 0.00019998948622706756,
-      "loss": 46.0,
-      "step": 28673
-    },
-    {
-      "epoch": 4.617738234228431,
-      "grad_norm": 0.0006485159392468631,
-      "learning_rate": 0.00019998948549345474,
-      "loss": 46.0,
-      "step": 28674
-    },
-    {
-      "epoch": 4.617899271307218,
-      "grad_norm": 0.0031922643538564444,
-      "learning_rate": 0.00019998948475981633,
-      "loss": 46.0,
-      "step": 28675
-    },
-    {
-      "epoch": 4.618060308386006,
-      "grad_norm": 0.000634676543995738,
-      "learning_rate": 0.00019998948402615233,
-      "loss": 46.0,
-      "step": 28676
-    },
-    {
-      "epoch": 4.618221345464793,
-      "grad_norm": 0.004520370624959469,
-      "learning_rate": 0.00019998948329246274,
-      "loss": 46.0,
-      "step": 28677
-    },
-    {
-      "epoch": 4.61838238254358,
-      "grad_norm": 0.0052521973848342896,
-      "learning_rate": 0.00019998948255874754,
-      "loss": 46.0,
-      "step": 28678
-    },
-    {
-      "epoch": 4.618543419622368,
-      "grad_norm": 0.0035603551659733057,
-      "learning_rate": 0.00019998948182500676,
-      "loss": 46.0,
-      "step": 28679
-    },
-    {
-      "epoch": 4.618704456701155,
-      "grad_norm": 0.0029775267466902733,
-      "learning_rate": 0.00019998948109124038,
-      "loss": 46.0,
-      "step": 28680
-    },
-    {
-      "epoch": 4.618865493779943,
-      "grad_norm": 0.0051308441907167435,
-      "learning_rate": 0.00019998948035744842,
-      "loss": 46.0,
-      "step": 28681
-    },
-    {
-      "epoch": 4.61902653085873,
-      "grad_norm": 0.01388498768210411,
-      "learning_rate": 0.00019998947962363088,
-      "loss": 46.0,
-      "step": 28682
-    },
-    {
-      "epoch": 4.619187567937518,
-      "grad_norm": 0.0019024768844246864,
-      "learning_rate": 0.00019998947888978771,
-      "loss": 46.0,
-      "step": 28683
-    },
-    {
-      "epoch": 4.619348605016305,
-      "grad_norm": 0.00683908024802804,
-      "learning_rate": 0.000199989478155919,
-      "loss": 46.0,
-      "step": 28684
-    },
-    {
-      "epoch": 4.6195096420950925,
-      "grad_norm": 0.0029897105414420366,
-      "learning_rate": 0.00019998947742202466,
-      "loss": 46.0,
-      "step": 28685
-    },
-    {
-      "epoch": 4.61967067917388,
-      "grad_norm": 0.00455856928601861,
-      "learning_rate": 0.00019998947668810476,
-      "loss": 46.0,
-      "step": 28686
-    },
-    {
-      "epoch": 4.619831716252667,
-      "grad_norm": 0.0018517576390877366,
-      "learning_rate": 0.00019998947595415922,
-      "loss": 46.0,
-      "step": 28687
-    },
-    {
-      "epoch": 4.619992753331455,
-      "grad_norm": 0.004377380479127169,
-      "learning_rate": 0.00019998947522018812,
-      "loss": 46.0,
-      "step": 28688
-    },
-    {
-      "epoch": 4.620153790410242,
-      "grad_norm": 0.0011183862807229161,
-      "learning_rate": 0.00019998947448619144,
-      "loss": 46.0,
-      "step": 28689
-    },
-    {
-      "epoch": 4.620314827489029,
-      "grad_norm": 0.0066835167817771435,
-      "learning_rate": 0.0001999894737521691,
-      "loss": 46.0,
-      "step": 28690
-    },
-    {
-      "epoch": 4.620475864567817,
-      "grad_norm": 0.00665680505335331,
-      "learning_rate": 0.00019998947301812123,
-      "loss": 46.0,
-      "step": 28691
-    },
-    {
-      "epoch": 4.620636901646604,
-      "grad_norm": 0.0053182621486485004,
-      "learning_rate": 0.00019998947228404775,
-      "loss": 46.0,
-      "step": 28692
-    },
-    {
-      "epoch": 4.620797938725391,
-      "grad_norm": 0.0014270198298618197,
-      "learning_rate": 0.0001999894715499487,
-      "loss": 46.0,
-      "step": 28693
-    },
-    {
-      "epoch": 4.620958975804179,
-      "grad_norm": 0.001682116067968309,
-      "learning_rate": 0.00019998947081582404,
-      "loss": 46.0,
-      "step": 28694
-    },
-    {
-      "epoch": 4.621120012882966,
-      "grad_norm": 0.010620569810271263,
-      "learning_rate": 0.00019998947008167378,
-      "loss": 46.0,
-      "step": 28695
-    },
-    {
-      "epoch": 4.621281049961754,
-      "grad_norm": 0.002223008079454303,
-      "learning_rate": 0.00019998946934749793,
-      "loss": 46.0,
-      "step": 28696
-    },
-    {
-      "epoch": 4.621442087040541,
-      "grad_norm": 0.0020947102457284927,
-      "learning_rate": 0.0001999894686132965,
-      "loss": 46.0,
-      "step": 28697
-    },
-    {
-      "epoch": 4.6216031241193285,
-      "grad_norm": 0.008420648984611034,
-      "learning_rate": 0.00019998946787906947,
-      "loss": 46.0,
-      "step": 28698
-    },
-    {
-      "epoch": 4.621764161198116,
-      "grad_norm": 0.001620523165911436,
-      "learning_rate": 0.00019998946714481686,
-      "loss": 46.0,
-      "step": 28699
-    },
-    {
-      "epoch": 4.621925198276903,
-      "grad_norm": 0.004672982729971409,
-      "learning_rate": 0.00019998946641053864,
-      "loss": 46.0,
-      "step": 28700
-    },
-    {
-      "epoch": 4.622086235355691,
-      "grad_norm": 0.0009783884743228555,
-      "learning_rate": 0.00019998946567623485,
-      "loss": 46.0,
-      "step": 28701
-    },
-    {
-      "epoch": 4.622247272434478,
-      "grad_norm": 0.008539550937712193,
-      "learning_rate": 0.00019998946494190545,
-      "loss": 46.0,
-      "step": 28702
-    },
-    {
-      "epoch": 4.622408309513266,
-      "grad_norm": 0.004753672983497381,
-      "learning_rate": 0.00019998946420755044,
-      "loss": 46.0,
-      "step": 28703
-    },
-    {
-      "epoch": 4.622569346592053,
-      "grad_norm": 0.00302482140250504,
-      "learning_rate": 0.00019998946347316986,
-      "loss": 46.0,
-      "step": 28704
-    },
-    {
-      "epoch": 4.62273038367084,
-      "grad_norm": 0.0042535122483968735,
-      "learning_rate": 0.0001999894627387637,
-      "loss": 46.0,
-      "step": 28705
-    },
-    {
-      "epoch": 4.622891420749627,
-      "grad_norm": 0.019024042412638664,
-      "learning_rate": 0.00019998946200433193,
-      "loss": 46.0,
-      "step": 28706
-    },
-    {
-      "epoch": 4.623052457828415,
-      "grad_norm": 0.008529804646968842,
-      "learning_rate": 0.00019998946126987456,
-      "loss": 46.0,
-      "step": 28707
-    },
-    {
-      "epoch": 4.623213494907202,
-      "grad_norm": 0.007570905145257711,
-      "learning_rate": 0.00019998946053539161,
-      "loss": 46.0,
-      "step": 28708
-    },
-    {
-      "epoch": 4.62337453198599,
-      "grad_norm": 0.00918601918965578,
-      "learning_rate": 0.00019998945980088308,
-      "loss": 46.0,
-      "step": 28709
-    },
-    {
-      "epoch": 4.623535569064777,
-      "grad_norm": 0.0027174788992851973,
-      "learning_rate": 0.00019998945906634893,
-      "loss": 46.0,
-      "step": 28710
-    },
-    {
-      "epoch": 4.6236966061435645,
-      "grad_norm": 0.005719912238419056,
-      "learning_rate": 0.00019998945833178921,
-      "loss": 46.0,
-      "step": 28711
-    },
-    {
-      "epoch": 4.623857643222352,
-      "grad_norm": 0.005411564372479916,
-      "learning_rate": 0.0001999894575972039,
-      "loss": 46.0,
-      "step": 28712
-    },
-    {
-      "epoch": 4.624018680301139,
-      "grad_norm": 0.001239596982486546,
-      "learning_rate": 0.00019998945686259298,
-      "loss": 46.0,
-      "step": 28713
-    },
-    {
-      "epoch": 4.624179717379927,
-      "grad_norm": 0.0036991252563893795,
-      "learning_rate": 0.00019998945612795648,
-      "loss": 46.0,
-      "step": 28714
-    },
-    {
-      "epoch": 4.624340754458714,
-      "grad_norm": 0.003780174534767866,
-      "learning_rate": 0.0001999894553932944,
-      "loss": 46.0,
-      "step": 28715
-    },
-    {
-      "epoch": 4.624501791537502,
-      "grad_norm": 0.0045287394896149635,
-      "learning_rate": 0.0001999894546586067,
-      "loss": 46.0,
-      "step": 28716
-    },
-    {
-      "epoch": 4.624662828616289,
-      "grad_norm": 0.007606519851833582,
-      "learning_rate": 0.0001999894539238934,
-      "loss": 46.0,
-      "step": 28717
-    },
-    {
-      "epoch": 4.624823865695077,
-      "grad_norm": 0.002691676141694188,
-      "learning_rate": 0.00019998945318915455,
-      "loss": 46.0,
-      "step": 28718
-    },
-    {
-      "epoch": 4.624984902773864,
-      "grad_norm": 0.00417829304933548,
-      "learning_rate": 0.00019998945245439006,
-      "loss": 46.0,
-      "step": 28719
-    },
-    {
-      "epoch": 4.625145939852651,
-      "grad_norm": 0.005441527813673019,
-      "learning_rate": 0.0001999894517196,
-      "loss": 46.0,
-      "step": 28720
-    },
-    {
-      "epoch": 4.625306976931438,
-      "grad_norm": 0.0035826906096190214,
-      "learning_rate": 0.00019998945098478437,
-      "loss": 46.0,
-      "step": 28721
-    },
-    {
-      "epoch": 4.625468014010226,
-      "grad_norm": 0.00313868117518723,
-      "learning_rate": 0.00019998945024994312,
-      "loss": 46.0,
-      "step": 28722
-    },
-    {
-      "epoch": 4.625629051089013,
-      "grad_norm": 0.0020399545319378376,
-      "learning_rate": 0.0001999894495150763,
-      "loss": 46.0,
-      "step": 28723
-    },
-    {
-      "epoch": 4.6257900881678005,
-      "grad_norm": 0.002654819516465068,
-      "learning_rate": 0.00019998944878018389,
-      "loss": 46.0,
-      "step": 28724
-    },
-    {
-      "epoch": 4.625951125246588,
-      "grad_norm": 0.011123983189463615,
-      "learning_rate": 0.00019998944804526585,
-      "loss": 46.0,
-      "step": 28725
-    },
-    {
-      "epoch": 4.626112162325375,
-      "grad_norm": 0.0010415163123980165,
-      "learning_rate": 0.00019998944731032222,
-      "loss": 46.0,
-      "step": 28726
-    },
-    {
-      "epoch": 4.626273199404163,
-      "grad_norm": 0.0016799357254058123,
-      "learning_rate": 0.00019998944657535303,
-      "loss": 46.0,
-      "step": 28727
-    },
-    {
-      "epoch": 4.62643423648295,
-      "grad_norm": 0.001290600630454719,
-      "learning_rate": 0.00019998944584035826,
-      "loss": 46.0,
-      "step": 28728
-    },
-    {
-      "epoch": 4.626595273561738,
-      "grad_norm": 0.0064903805032372475,
-      "learning_rate": 0.00019998944510533787,
-      "loss": 46.0,
-      "step": 28729
-    },
-    {
-      "epoch": 4.626756310640525,
-      "grad_norm": 0.001428467221558094,
-      "learning_rate": 0.0001999894443702919,
-      "loss": 46.0,
-      "step": 28730
-    },
-    {
-      "epoch": 4.626917347719313,
-      "grad_norm": 0.0028896804433315992,
-      "learning_rate": 0.00019998944363522033,
-      "loss": 46.0,
-      "step": 28731
-    },
-    {
-      "epoch": 4.6270783847981,
-      "grad_norm": 0.0032285735942423344,
-      "learning_rate": 0.00019998944290012312,
-      "loss": 46.0,
-      "step": 28732
-    },
-    {
-      "epoch": 4.6272394218768875,
-      "grad_norm": 0.006240965332835913,
-      "learning_rate": 0.0001999894421650004,
-      "loss": 46.0,
-      "step": 28733
-    },
-    {
-      "epoch": 4.627400458955675,
-      "grad_norm": 0.003867104183882475,
-      "learning_rate": 0.00019998944142985203,
-      "loss": 46.0,
-      "step": 28734
-    },
-    {
-      "epoch": 4.6275614960344615,
-      "grad_norm": 0.0027933414094150066,
-      "learning_rate": 0.00019998944069467812,
-      "loss": 46.0,
-      "step": 28735
-    },
-    {
-      "epoch": 4.627722533113249,
-      "grad_norm": 0.004787487909197807,
-      "learning_rate": 0.00019998943995947857,
-      "loss": 46.0,
-      "step": 28736
-    },
-    {
-      "epoch": 4.6278835701920364,
-      "grad_norm": 0.004292582161724567,
-      "learning_rate": 0.00019998943922425343,
-      "loss": 46.0,
-      "step": 28737
-    },
-    {
-      "epoch": 4.628044607270824,
-      "grad_norm": 0.004687467589974403,
-      "learning_rate": 0.00019998943848900273,
-      "loss": 46.0,
-      "step": 28738
-    },
-    {
-      "epoch": 4.628205644349611,
-      "grad_norm": 0.005203989800065756,
-      "learning_rate": 0.00019998943775372641,
-      "loss": 46.0,
-      "step": 28739
-    },
-    {
-      "epoch": 4.628366681428399,
-      "grad_norm": 0.0018016740214079618,
-      "learning_rate": 0.00019998943701842454,
-      "loss": 46.0,
-      "step": 28740
-    },
-    {
-      "epoch": 4.628527718507186,
-      "grad_norm": 0.005282426718622446,
-      "learning_rate": 0.00019998943628309702,
-      "loss": 46.0,
-      "step": 28741
-    },
-    {
-      "epoch": 4.628688755585974,
-      "grad_norm": 0.001087265438400209,
-      "learning_rate": 0.00019998943554774395,
-      "loss": 46.0,
-      "step": 28742
-    },
-    {
-      "epoch": 4.628849792664761,
-      "grad_norm": 0.0044693793170154095,
-      "learning_rate": 0.00019998943481236528,
-      "loss": 46.0,
-      "step": 28743
-    },
-    {
-      "epoch": 4.629010829743549,
-      "grad_norm": 0.008755608461797237,
-      "learning_rate": 0.000199989434076961,
-      "loss": 46.0,
-      "step": 28744
-    },
-    {
-      "epoch": 4.629171866822336,
-      "grad_norm": 0.0053488025441765785,
-      "learning_rate": 0.00019998943334153114,
-      "loss": 46.0,
-      "step": 28745
-    },
-    {
-      "epoch": 4.6293329039011235,
-      "grad_norm": 0.006271216087043285,
-      "learning_rate": 0.0001999894326060757,
-      "loss": 46.0,
-      "step": 28746
-    },
-    {
-      "epoch": 4.629493940979911,
-      "grad_norm": 0.0045645213685929775,
-      "learning_rate": 0.00019998943187059465,
-      "loss": 46.0,
-      "step": 28747
-    },
-    {
-      "epoch": 4.6296549780586975,
-      "grad_norm": 0.0031551923602819443,
-      "learning_rate": 0.00019998943113508802,
-      "loss": 46.0,
-      "step": 28748
-    },
-    {
-      "epoch": 4.629816015137486,
-      "grad_norm": 0.0012918355641886592,
-      "learning_rate": 0.00019998943039955578,
-      "loss": 46.0,
-      "step": 28749
-    },
-    {
-      "epoch": 4.629977052216272,
-      "grad_norm": 0.0048337276093661785,
-      "learning_rate": 0.00019998942966399796,
-      "loss": 46.0,
-      "step": 28750
-    },
-    {
-      "epoch": 4.63013808929506,
-      "grad_norm": 0.004521054681390524,
-      "learning_rate": 0.00019998942892841454,
-      "loss": 46.0,
-      "step": 28751
-    },
-    {
-      "epoch": 4.630299126373847,
-      "grad_norm": 0.003795759519562125,
-      "learning_rate": 0.00019998942819280554,
-      "loss": 46.0,
-      "step": 28752
-    },
-    {
-      "epoch": 4.630460163452635,
-      "grad_norm": 0.004174003843218088,
-      "learning_rate": 0.00019998942745717092,
-      "loss": 46.0,
-      "step": 28753
-    },
-    {
-      "epoch": 4.630621200531422,
-      "grad_norm": 0.0024887763429433107,
-      "learning_rate": 0.00019998942672151075,
-      "loss": 46.0,
-      "step": 28754
-    },
-    {
-      "epoch": 4.63078223761021,
-      "grad_norm": 0.00228418642655015,
-      "learning_rate": 0.00019998942598582496,
-      "loss": 46.0,
-      "step": 28755
-    },
-    {
-      "epoch": 4.630943274688997,
-      "grad_norm": 0.0032761809416115284,
-      "learning_rate": 0.0001999894252501136,
-      "loss": 46.0,
-      "step": 28756
-    },
-    {
-      "epoch": 4.631104311767785,
-      "grad_norm": 0.00321602844633162,
-      "learning_rate": 0.0001999894245143766,
-      "loss": 46.0,
-      "step": 28757
-    },
-    {
-      "epoch": 4.631265348846572,
-      "grad_norm": 0.004598714876919985,
-      "learning_rate": 0.00019998942377861406,
-      "loss": 46.0,
-      "step": 28758
-    },
-    {
-      "epoch": 4.6314263859253595,
-      "grad_norm": 0.005074535496532917,
-      "learning_rate": 0.00019998942304282592,
-      "loss": 46.0,
-      "step": 28759
-    },
-    {
-      "epoch": 4.631587423004147,
-      "grad_norm": 0.024239646270871162,
-      "learning_rate": 0.0001999894223070122,
-      "loss": 46.0,
-      "step": 28760
-    },
-    {
-      "epoch": 4.631748460082934,
-      "grad_norm": 0.00471185939386487,
-      "learning_rate": 0.00019998942157117285,
-      "loss": 46.0,
-      "step": 28761
-    },
-    {
-      "epoch": 4.631909497161722,
-      "grad_norm": 0.011587814427912235,
-      "learning_rate": 0.00019998942083530793,
-      "loss": 46.0,
-      "step": 28762
-    },
-    {
-      "epoch": 4.632070534240508,
-      "grad_norm": 0.008011476136744022,
-      "learning_rate": 0.00019998942009941738,
-      "loss": 46.0,
-      "step": 28763
-    },
-    {
-      "epoch": 4.632231571319297,
-      "grad_norm": 0.0014787832042202353,
-      "learning_rate": 0.00019998941936350128,
-      "loss": 46.0,
-      "step": 28764
-    },
-    {
-      "epoch": 4.632392608398083,
-      "grad_norm": 0.011929119005799294,
-      "learning_rate": 0.00019998941862755956,
-      "loss": 46.0,
-      "step": 28765
-    },
-    {
-      "epoch": 4.632553645476871,
-      "grad_norm": 0.007643118966370821,
-      "learning_rate": 0.0001999894178915923,
-      "loss": 46.0,
-      "step": 28766
-    },
-    {
-      "epoch": 4.632714682555658,
-      "grad_norm": 0.002223399467766285,
-      "learning_rate": 0.00019998941715559937,
-      "loss": 46.0,
-      "step": 28767
-    },
-    {
-      "epoch": 4.632875719634446,
-      "grad_norm": 0.0020988464821130037,
-      "learning_rate": 0.00019998941641958092,
-      "loss": 46.0,
-      "step": 28768
-    },
-    {
-      "epoch": 4.633036756713233,
-      "grad_norm": 0.004369881935417652,
-      "learning_rate": 0.00019998941568353683,
-      "loss": 46.0,
-      "step": 28769
-    },
-    {
-      "epoch": 4.633197793792021,
-      "grad_norm": 0.0049811918288469315,
-      "learning_rate": 0.00019998941494746717,
-      "loss": 46.0,
-      "step": 28770
-    },
-    {
-      "epoch": 4.633358830870808,
-      "grad_norm": 0.004455159418284893,
-      "learning_rate": 0.0001999894142113719,
-      "loss": 46.0,
-      "step": 28771
-    },
-    {
-      "epoch": 4.6335198679495955,
-      "grad_norm": 0.002118463162332773,
-      "learning_rate": 0.00019998941347525105,
-      "loss": 46.0,
-      "step": 28772
-    },
-    {
-      "epoch": 4.633680905028383,
-      "grad_norm": 0.003406836884096265,
-      "learning_rate": 0.0001999894127391046,
-      "loss": 46.0,
-      "step": 28773
-    },
-    {
-      "epoch": 4.63384194210717,
-      "grad_norm": 0.006176305469125509,
-      "learning_rate": 0.00019998941200293258,
-      "loss": 46.0,
-      "step": 28774
-    },
-    {
-      "epoch": 4.634002979185958,
-      "grad_norm": 0.010172223672270775,
-      "learning_rate": 0.00019998941126673497,
-      "loss": 46.0,
-      "step": 28775
-    },
-    {
-      "epoch": 4.634164016264745,
-      "grad_norm": 0.00523246917873621,
-      "learning_rate": 0.00019998941053051174,
-      "loss": 46.0,
-      "step": 28776
-    },
-    {
-      "epoch": 4.634325053343533,
-      "grad_norm": 0.0025002325419336557,
-      "learning_rate": 0.00019998940979426295,
-      "loss": 46.0,
-      "step": 28777
-    },
-    {
-      "epoch": 4.634486090422319,
-      "grad_norm": 0.0034382734447717667,
-      "learning_rate": 0.00019998940905798852,
-      "loss": 46.0,
-      "step": 28778
-    },
-    {
-      "epoch": 4.634647127501107,
-      "grad_norm": 0.010849333368241787,
-      "learning_rate": 0.00019998940832168853,
-      "loss": 46.0,
-      "step": 28779
-    },
-    {
-      "epoch": 4.634808164579894,
-      "grad_norm": 0.001068843062967062,
-      "learning_rate": 0.00019998940758536295,
-      "loss": 46.0,
-      "step": 28780
-    },
-    {
-      "epoch": 4.634969201658682,
-      "grad_norm": 0.015925215557217598,
-      "learning_rate": 0.00019998940684901178,
-      "loss": 46.0,
-      "step": 28781
-    },
-    {
-      "epoch": 4.635130238737469,
-      "grad_norm": 0.011597731150686741,
-      "learning_rate": 0.000199989406112635,
-      "loss": 46.0,
-      "step": 28782
-    },
-    {
-      "epoch": 4.6352912758162566,
-      "grad_norm": 0.0019245428266003728,
-      "learning_rate": 0.00019998940537623266,
-      "loss": 46.0,
-      "step": 28783
-    },
-    {
-      "epoch": 4.635452312895044,
-      "grad_norm": 0.003491608891636133,
-      "learning_rate": 0.00019998940463980468,
-      "loss": 46.0,
-      "step": 28784
-    },
-    {
-      "epoch": 4.6356133499738315,
-      "grad_norm": 0.005695475731045008,
-      "learning_rate": 0.00019998940390335114,
-      "loss": 46.0,
-      "step": 28785
-    },
-    {
-      "epoch": 4.635774387052619,
-      "grad_norm": 0.003567585488781333,
-      "learning_rate": 0.000199989403166872,
-      "loss": 46.0,
-      "step": 28786
-    },
-    {
-      "epoch": 4.635935424131406,
-      "grad_norm": 0.0032246678601950407,
-      "learning_rate": 0.00019998940243036727,
-      "loss": 46.0,
-      "step": 28787
-    },
-    {
-      "epoch": 4.636096461210194,
-      "grad_norm": 0.0022226048167794943,
-      "learning_rate": 0.00019998940169383694,
-      "loss": 46.0,
-      "step": 28788
-    },
-    {
-      "epoch": 4.636257498288981,
-      "grad_norm": 0.002625696826726198,
-      "learning_rate": 0.00019998940095728102,
-      "loss": 46.0,
-      "step": 28789
-    },
-    {
-      "epoch": 4.636418535367769,
-      "grad_norm": 0.0031848677899688482,
-      "learning_rate": 0.00019998940022069952,
-      "loss": 46.0,
-      "step": 28790
-    },
-    {
-      "epoch": 4.636579572446556,
-      "grad_norm": 0.0010988399153575301,
-      "learning_rate": 0.0001999893994840924,
-      "loss": 46.0,
-      "step": 28791
-    },
-    {
-      "epoch": 4.636740609525344,
-      "grad_norm": 0.0026120019610971212,
-      "learning_rate": 0.00019998939874745972,
-      "loss": 46.0,
-      "step": 28792
-    },
-    {
-      "epoch": 4.63690164660413,
-      "grad_norm": 0.0035063589457422495,
-      "learning_rate": 0.00019998939801080145,
-      "loss": 46.0,
-      "step": 28793
-    },
-    {
-      "epoch": 4.637062683682918,
-      "grad_norm": 0.016091682016849518,
-      "learning_rate": 0.00019998939727411757,
-      "loss": 46.0,
-      "step": 28794
-    },
-    {
-      "epoch": 4.637223720761705,
-      "grad_norm": 0.010831695981323719,
-      "learning_rate": 0.00019998939653740808,
-      "loss": 46.0,
-      "step": 28795
-    },
-    {
-      "epoch": 4.6373847578404925,
-      "grad_norm": 0.002514464780688286,
-      "learning_rate": 0.00019998939580067302,
-      "loss": 46.0,
-      "step": 28796
-    },
-    {
-      "epoch": 4.63754579491928,
-      "grad_norm": 0.0066358535550534725,
-      "learning_rate": 0.00019998939506391236,
-      "loss": 46.0,
-      "step": 28797
-    },
-    {
-      "epoch": 4.637706831998067,
-      "grad_norm": 0.006156806834042072,
-      "learning_rate": 0.00019998939432712613,
-      "loss": 46.0,
-      "step": 28798
-    },
-    {
-      "epoch": 4.637867869076855,
-      "grad_norm": 0.002510531572625041,
-      "learning_rate": 0.0001999893935903143,
-      "loss": 46.0,
-      "step": 28799
-    },
-    {
-      "epoch": 4.638028906155642,
-      "grad_norm": 0.011003522202372551,
-      "learning_rate": 0.00019998939285347685,
-      "loss": 46.0,
-      "step": 28800
-    },
-    {
-      "epoch": 4.63818994323443,
-      "grad_norm": 0.021284379065036774,
-      "learning_rate": 0.00019998939211661386,
-      "loss": 46.0,
-      "step": 28801
-    },
-    {
-      "epoch": 4.638350980313217,
-      "grad_norm": 0.002376176882535219,
-      "learning_rate": 0.00019998939137972523,
-      "loss": 46.0,
-      "step": 28802
-    },
-    {
-      "epoch": 4.638512017392005,
-      "grad_norm": 0.006309289485216141,
-      "learning_rate": 0.000199989390642811,
-      "loss": 46.0,
-      "step": 28803
-    },
-    {
-      "epoch": 4.638673054470792,
-      "grad_norm": 0.011936895549297333,
-      "learning_rate": 0.0001999893899058712,
-      "loss": 46.0,
-      "step": 28804
-    },
-    {
-      "epoch": 4.63883409154958,
-      "grad_norm": 0.0042016212828457355,
-      "learning_rate": 0.00019998938916890584,
-      "loss": 46.0,
-      "step": 28805
-    },
-    {
-      "epoch": 4.638995128628367,
-      "grad_norm": 0.0020441431552171707,
-      "learning_rate": 0.00019998938843191483,
-      "loss": 46.0,
-      "step": 28806
-    },
-    {
-      "epoch": 4.6391561657071545,
-      "grad_norm": 0.002865507733076811,
-      "learning_rate": 0.00019998938769489827,
-      "loss": 46.0,
-      "step": 28807
-    },
-    {
-      "epoch": 4.639317202785941,
-      "grad_norm": 0.0020579732954502106,
-      "learning_rate": 0.00019998938695785608,
-      "loss": 46.0,
-      "step": 28808
-    },
-    {
-      "epoch": 4.6394782398647285,
-      "grad_norm": 0.0038868901319801807,
-      "learning_rate": 0.00019998938622078837,
-      "loss": 46.0,
-      "step": 28809
-    },
-    {
-      "epoch": 4.639639276943516,
-      "grad_norm": 0.004660432692617178,
-      "learning_rate": 0.00019998938548369499,
-      "loss": 46.0,
-      "step": 28810
-    },
-    {
-      "epoch": 4.639800314022303,
-      "grad_norm": 0.004483303986489773,
-      "learning_rate": 0.00019998938474657604,
-      "loss": 46.0,
-      "step": 28811
-    },
-    {
-      "epoch": 4.639961351101091,
-      "grad_norm": 0.007797194644808769,
-      "learning_rate": 0.0001999893840094315,
-      "loss": 46.0,
-      "step": 28812
-    },
-    {
-      "epoch": 4.640122388179878,
-      "grad_norm": 0.0037905557546764612,
-      "learning_rate": 0.0001999893832722614,
-      "loss": 46.0,
-      "step": 28813
-    },
-    {
-      "epoch": 4.640283425258666,
-      "grad_norm": 0.01063101552426815,
-      "learning_rate": 0.00019998938253506566,
-      "loss": 46.0,
-      "step": 28814
-    },
-    {
-      "epoch": 4.640444462337453,
-      "grad_norm": 0.000624926236923784,
-      "learning_rate": 0.00019998938179784437,
-      "loss": 46.0,
-      "step": 28815
-    },
-    {
-      "epoch": 4.640605499416241,
-      "grad_norm": 0.003992448095232248,
-      "learning_rate": 0.00019998938106059746,
-      "loss": 46.0,
-      "step": 28816
-    },
-    {
-      "epoch": 4.640766536495028,
-      "grad_norm": 0.004300958476960659,
-      "learning_rate": 0.00019998938032332494,
-      "loss": 46.0,
-      "step": 28817
-    },
-    {
-      "epoch": 4.640927573573816,
-      "grad_norm": 0.0032595705706626177,
-      "learning_rate": 0.00019998937958602686,
-      "loss": 46.0,
-      "step": 28818
-    },
-    {
-      "epoch": 4.641088610652603,
-      "grad_norm": 0.007325825281441212,
-      "learning_rate": 0.0001999893788487032,
-      "loss": 46.0,
-      "step": 28819
-    },
-    {
-      "epoch": 4.6412496477313905,
-      "grad_norm": 0.010010342113673687,
-      "learning_rate": 0.0001999893781113539,
-      "loss": 46.0,
-      "step": 28820
-    },
-    {
-      "epoch": 4.641410684810177,
-      "grad_norm": 0.0026441614609211683,
-      "learning_rate": 0.00019998937737397907,
-      "loss": 46.0,
-      "step": 28821
-    },
-    {
-      "epoch": 4.641571721888965,
-      "grad_norm": 0.0011363831581547856,
-      "learning_rate": 0.0001999893766365786,
-      "loss": 46.0,
-      "step": 28822
-    },
-    {
-      "epoch": 4.641732758967752,
-      "grad_norm": 0.002139223041012883,
-      "learning_rate": 0.00019998937589915254,
-      "loss": 46.0,
-      "step": 28823
-    },
-    {
-      "epoch": 4.641893796046539,
-      "grad_norm": 0.0007391894469037652,
-      "learning_rate": 0.0001999893751617009,
-      "loss": 46.0,
-      "step": 28824
-    },
-    {
-      "epoch": 4.642054833125327,
-      "grad_norm": 0.01802361011505127,
-      "learning_rate": 0.00019998937442422367,
-      "loss": 46.0,
-      "step": 28825
-    },
-    {
-      "epoch": 4.642215870204114,
-      "grad_norm": 0.001768721267580986,
-      "learning_rate": 0.00019998937368672083,
-      "loss": 46.0,
-      "step": 28826
-    },
-    {
-      "epoch": 4.642376907282902,
-      "grad_norm": 0.0010522359516471624,
-      "learning_rate": 0.00019998937294919244,
-      "loss": 46.0,
-      "step": 28827
-    },
-    {
-      "epoch": 4.642537944361689,
-      "grad_norm": 0.010329823940992355,
-      "learning_rate": 0.00019998937221163844,
-      "loss": 46.0,
-      "step": 28828
-    },
-    {
-      "epoch": 4.642698981440477,
-      "grad_norm": 0.010859774425625801,
-      "learning_rate": 0.00019998937147405884,
-      "loss": 46.0,
-      "step": 28829
-    },
-    {
-      "epoch": 4.642860018519264,
-      "grad_norm": 0.0082384143024683,
-      "learning_rate": 0.00019998937073645363,
-      "loss": 46.0,
-      "step": 28830
-    },
-    {
-      "epoch": 4.643021055598052,
-      "grad_norm": 0.017249522730708122,
-      "learning_rate": 0.00019998936999882284,
-      "loss": 46.0,
-      "step": 28831
-    },
-    {
-      "epoch": 4.643182092676839,
-      "grad_norm": 0.0026608631014823914,
-      "learning_rate": 0.00019998936926116648,
-      "loss": 46.0,
-      "step": 28832
-    },
-    {
-      "epoch": 4.6433431297556265,
-      "grad_norm": 0.009312928654253483,
-      "learning_rate": 0.0001999893685234845,
-      "loss": 46.0,
-      "step": 28833
-    },
-    {
-      "epoch": 4.643504166834414,
-      "grad_norm": 0.008810551837086678,
-      "learning_rate": 0.00019998936778577693,
-      "loss": 46.0,
-      "step": 28834
-    },
-    {
-      "epoch": 4.643665203913201,
-      "grad_norm": 0.002851797267794609,
-      "learning_rate": 0.00019998936704804376,
-      "loss": 46.0,
-      "step": 28835
-    },
-    {
-      "epoch": 4.643826240991988,
-      "grad_norm": 0.002766981953755021,
-      "learning_rate": 0.00019998936631028503,
-      "loss": 46.0,
-      "step": 28836
-    },
-    {
-      "epoch": 4.643987278070776,
-      "grad_norm": 0.00512302853167057,
-      "learning_rate": 0.0001999893655725007,
-      "loss": 46.0,
-      "step": 28837
-    },
-    {
-      "epoch": 4.644148315149563,
-      "grad_norm": 0.01607791893184185,
-      "learning_rate": 0.00019998936483469075,
-      "loss": 46.0,
-      "step": 28838
-    },
-    {
-      "epoch": 4.64430935222835,
-      "grad_norm": 0.005732540041208267,
-      "learning_rate": 0.00019998936409685523,
-      "loss": 46.0,
-      "step": 28839
-    },
-    {
-      "epoch": 4.644470389307138,
-      "grad_norm": 0.008913103491067886,
-      "learning_rate": 0.00019998936335899412,
-      "loss": 46.0,
-      "step": 28840
-    },
-    {
-      "epoch": 4.644631426385925,
-      "grad_norm": 0.01068598311394453,
-      "learning_rate": 0.0001999893626211074,
-      "loss": 46.0,
-      "step": 28841
-    },
-    {
-      "epoch": 4.644792463464713,
-      "grad_norm": 0.005982093047350645,
-      "learning_rate": 0.0001999893618831951,
-      "loss": 46.0,
-      "step": 28842
-    },
-    {
-      "epoch": 4.6449535005435,
-      "grad_norm": 0.0027696872130036354,
-      "learning_rate": 0.00019998936114525723,
-      "loss": 46.0,
-      "step": 28843
-    },
-    {
-      "epoch": 4.6451145376222875,
-      "grad_norm": 0.002058908576145768,
-      "learning_rate": 0.00019998936040729372,
-      "loss": 46.0,
-      "step": 28844
-    },
-    {
-      "epoch": 4.645275574701075,
-      "grad_norm": 0.0007687556208111346,
-      "learning_rate": 0.00019998935966930465,
-      "loss": 46.0,
-      "step": 28845
-    },
-    {
-      "epoch": 4.6454366117798624,
-      "grad_norm": 0.010753043927252293,
-      "learning_rate": 0.00019998935893129,
-      "loss": 46.0,
-      "step": 28846
-    },
-    {
-      "epoch": 4.64559764885865,
-      "grad_norm": 0.02487793378531933,
-      "learning_rate": 0.00019998935819324972,
-      "loss": 46.0,
-      "step": 28847
-    },
-    {
-      "epoch": 4.645758685937437,
-      "grad_norm": 0.004360293038189411,
-      "learning_rate": 0.0001999893574551839,
-      "loss": 46.0,
-      "step": 28848
-    },
-    {
-      "epoch": 4.645919723016225,
-      "grad_norm": 0.003180798841640353,
-      "learning_rate": 0.00019998935671709241,
-      "loss": 46.0,
-      "step": 28849
-    },
-    {
-      "epoch": 4.646080760095012,
-      "grad_norm": 0.006085132248699665,
-      "learning_rate": 0.00019998935597897538,
-      "loss": 46.0,
-      "step": 28850
-    },
-    {
-      "epoch": 4.646241797173799,
-      "grad_norm": 0.001889756415039301,
-      "learning_rate": 0.00019998935524083276,
-      "loss": 46.0,
-      "step": 28851
-    },
-    {
-      "epoch": 4.646402834252587,
-      "grad_norm": 0.009555756114423275,
-      "learning_rate": 0.00019998935450266453,
-      "loss": 46.0,
-      "step": 28852
-    },
-    {
-      "epoch": 4.646563871331374,
-      "grad_norm": 0.009193788282573223,
-      "learning_rate": 0.00019998935376447073,
-      "loss": 46.0,
-      "step": 28853
-    },
-    {
-      "epoch": 4.646724908410161,
-      "grad_norm": 0.0019975604955106974,
-      "learning_rate": 0.00019998935302625132,
-      "loss": 46.0,
-      "step": 28854
-    },
-    {
-      "epoch": 4.646885945488949,
-      "grad_norm": 0.01905670575797558,
-      "learning_rate": 0.00019998935228800633,
-      "loss": 46.0,
-      "step": 28855
-    },
-    {
-      "epoch": 4.647046982567736,
-      "grad_norm": 0.001570022781379521,
-      "learning_rate": 0.00019998935154973572,
-      "loss": 46.0,
-      "step": 28856
-    },
-    {
-      "epoch": 4.6472080196465235,
-      "grad_norm": 0.0019738497212529182,
-      "learning_rate": 0.00019998935081143955,
-      "loss": 46.0,
-      "step": 28857
-    },
-    {
-      "epoch": 4.647369056725311,
-      "grad_norm": 0.0029243184253573418,
-      "learning_rate": 0.00019998935007311776,
-      "loss": 46.0,
-      "step": 28858
-    },
-    {
-      "epoch": 4.647530093804098,
-      "grad_norm": 0.005150943994522095,
-      "learning_rate": 0.0001999893493347704,
-      "loss": 46.0,
-      "step": 28859
-    },
-    {
-      "epoch": 4.647691130882886,
-      "grad_norm": 0.012920072302222252,
-      "learning_rate": 0.00019998934859639746,
-      "loss": 46.0,
-      "step": 28860
-    },
-    {
-      "epoch": 4.647852167961673,
-      "grad_norm": 0.0070352330803871155,
-      "learning_rate": 0.0001999893478579989,
-      "loss": 46.0,
-      "step": 28861
-    },
-    {
-      "epoch": 4.648013205040461,
-      "grad_norm": 0.0014051893958821893,
-      "learning_rate": 0.00019998934711957476,
-      "loss": 46.0,
-      "step": 28862
-    },
-    {
-      "epoch": 4.648174242119248,
-      "grad_norm": 0.0026403588708490133,
-      "learning_rate": 0.000199989346381125,
-      "loss": 46.0,
-      "step": 28863
-    },
-    {
-      "epoch": 4.648335279198036,
-      "grad_norm": 0.001131717232055962,
-      "learning_rate": 0.00019998934564264967,
-      "loss": 46.0,
-      "step": 28864
-    },
-    {
-      "epoch": 4.648496316276823,
-      "grad_norm": 0.0032847800757735968,
-      "learning_rate": 0.00019998934490414878,
-      "loss": 46.0,
-      "step": 28865
-    },
-    {
-      "epoch": 4.64865735335561,
-      "grad_norm": 0.0187359731644392,
-      "learning_rate": 0.00019998934416562224,
-      "loss": 46.0,
-      "step": 28866
-    },
-    {
-      "epoch": 4.648818390434397,
-      "grad_norm": 0.0023506670258939266,
-      "learning_rate": 0.00019998934342707017,
-      "loss": 46.0,
-      "step": 28867
-    },
-    {
-      "epoch": 4.648979427513185,
-      "grad_norm": 0.0016623161500319839,
-      "learning_rate": 0.00019998934268849246,
-      "loss": 46.0,
-      "step": 28868
-    },
-    {
-      "epoch": 4.649140464591972,
-      "grad_norm": 0.01885918341577053,
-      "learning_rate": 0.0001999893419498892,
-      "loss": 46.0,
-      "step": 28869
-    },
-    {
-      "epoch": 4.6493015016707595,
-      "grad_norm": 0.0025795241817831993,
-      "learning_rate": 0.00019998934121126028,
-      "loss": 46.0,
-      "step": 28870
-    },
-    {
-      "epoch": 4.649462538749547,
-      "grad_norm": 0.005783949047327042,
-      "learning_rate": 0.00019998934047260584,
-      "loss": 46.0,
-      "step": 28871
-    },
-    {
-      "epoch": 4.649623575828334,
-      "grad_norm": 0.004076711367815733,
-      "learning_rate": 0.00019998933973392578,
-      "loss": 46.0,
-      "step": 28872
-    },
-    {
-      "epoch": 4.649784612907122,
-      "grad_norm": 0.002280758460983634,
-      "learning_rate": 0.0001999893389952201,
-      "loss": 46.0,
-      "step": 28873
-    },
-    {
-      "epoch": 4.649945649985909,
-      "grad_norm": 0.0033324796240776777,
-      "learning_rate": 0.00019998933825648884,
-      "loss": 46.0,
-      "step": 28874
-    },
-    {
-      "epoch": 4.650106687064697,
-      "grad_norm": 0.005482825450599194,
-      "learning_rate": 0.00019998933751773202,
-      "loss": 46.0,
-      "step": 28875
-    },
-    {
-      "epoch": 4.650267724143484,
-      "grad_norm": 0.0033118233550339937,
-      "learning_rate": 0.0001999893367789496,
-      "loss": 46.0,
-      "step": 28876
-    },
-    {
-      "epoch": 4.650428761222272,
-      "grad_norm": 0.0021592797711491585,
-      "learning_rate": 0.00019998933604014157,
-      "loss": 46.0,
-      "step": 28877
-    },
-    {
-      "epoch": 4.650589798301059,
-      "grad_norm": 0.001610046485438943,
-      "learning_rate": 0.00019998933530130796,
-      "loss": 46.0,
-      "step": 28878
-    },
-    {
-      "epoch": 4.650750835379847,
-      "grad_norm": 0.0034365851897746325,
-      "learning_rate": 0.00019998933456244873,
-      "loss": 46.0,
-      "step": 28879
-    },
-    {
-      "epoch": 4.650911872458634,
-      "grad_norm": 0.0019274357473477721,
-      "learning_rate": 0.00019998933382356392,
-      "loss": 46.0,
-      "step": 28880
-    },
-    {
-      "epoch": 4.651072909537421,
-      "grad_norm": 0.004763009026646614,
-      "learning_rate": 0.00019998933308465353,
-      "loss": 46.0,
-      "step": 28881
-    },
-    {
-      "epoch": 4.651233946616208,
-      "grad_norm": 0.006467279512435198,
-      "learning_rate": 0.00019998933234571754,
-      "loss": 46.0,
-      "step": 28882
-    },
-    {
-      "epoch": 4.6513949836949955,
-      "grad_norm": 0.011365518905222416,
-      "learning_rate": 0.00019998933160675597,
-      "loss": 46.0,
-      "step": 28883
-    },
-    {
-      "epoch": 4.651556020773783,
-      "grad_norm": 0.0031606238335371017,
-      "learning_rate": 0.0001999893308677688,
-      "loss": 46.0,
-      "step": 28884
-    },
-    {
-      "epoch": 4.65171705785257,
-      "grad_norm": 0.0038487704005092382,
-      "learning_rate": 0.00019998933012875604,
-      "loss": 46.0,
-      "step": 28885
-    },
-    {
-      "epoch": 4.651878094931358,
-      "grad_norm": 0.0037622677627950907,
-      "learning_rate": 0.00019998932938971768,
-      "loss": 46.0,
-      "step": 28886
-    },
-    {
-      "epoch": 4.652039132010145,
-      "grad_norm": 0.00617364514619112,
-      "learning_rate": 0.00019998932865065373,
-      "loss": 46.0,
-      "step": 28887
-    },
-    {
-      "epoch": 4.652200169088933,
-      "grad_norm": 0.0050621856935322285,
-      "learning_rate": 0.0001999893279115642,
-      "loss": 46.0,
-      "step": 28888
-    },
-    {
-      "epoch": 4.65236120616772,
-      "grad_norm": 0.0010590513702481985,
-      "learning_rate": 0.00019998932717244907,
-      "loss": 46.0,
-      "step": 28889
-    },
-    {
-      "epoch": 4.652522243246508,
-      "grad_norm": 0.00235868152230978,
-      "learning_rate": 0.00019998932643330833,
-      "loss": 46.0,
-      "step": 28890
-    },
-    {
-      "epoch": 4.652683280325295,
-      "grad_norm": 0.0073089152574539185,
-      "learning_rate": 0.00019998932569414204,
-      "loss": 46.0,
-      "step": 28891
-    },
-    {
-      "epoch": 4.6528443174040826,
-      "grad_norm": 0.006188882980495691,
-      "learning_rate": 0.00019998932495495013,
-      "loss": 46.0,
-      "step": 28892
-    },
-    {
-      "epoch": 4.65300535448287,
-      "grad_norm": 0.013392229564487934,
-      "learning_rate": 0.00019998932421573263,
-      "loss": 46.0,
-      "step": 28893
-    },
-    {
-      "epoch": 4.6531663915616575,
-      "grad_norm": 0.0023209438659250736,
-      "learning_rate": 0.00019998932347648954,
-      "loss": 46.0,
-      "step": 28894
-    },
-    {
-      "epoch": 4.653327428640445,
-      "grad_norm": 0.007212834898382425,
-      "learning_rate": 0.00019998932273722084,
-      "loss": 46.0,
-      "step": 28895
-    },
-    {
-      "epoch": 4.6534884657192315,
-      "grad_norm": 0.007052355445921421,
-      "learning_rate": 0.00019998932199792656,
-      "loss": 46.0,
-      "step": 28896
-    },
-    {
-      "epoch": 4.653649502798019,
-      "grad_norm": 0.0017822900554165244,
-      "learning_rate": 0.0001999893212586067,
-      "loss": 46.0,
-      "step": 28897
-    },
-    {
-      "epoch": 4.653810539876806,
-      "grad_norm": 0.0026739598251879215,
-      "learning_rate": 0.00019998932051926125,
-      "loss": 46.0,
-      "step": 28898
-    },
-    {
-      "epoch": 4.653971576955594,
-      "grad_norm": 0.001390158082358539,
-      "learning_rate": 0.0001999893197798902,
-      "loss": 46.0,
-      "step": 28899
-    },
-    {
-      "epoch": 4.654132614034381,
-      "grad_norm": 0.0026745223440229893,
-      "learning_rate": 0.00019998931904049357,
-      "loss": 46.0,
-      "step": 28900
-    },
-    {
-      "epoch": 4.654293651113169,
-      "grad_norm": 0.0022330102510750294,
-      "learning_rate": 0.00019998931830107132,
-      "loss": 46.0,
-      "step": 28901
-    },
-    {
-      "epoch": 4.654454688191956,
-      "grad_norm": 0.0026370638515800238,
-      "learning_rate": 0.00019998931756162348,
-      "loss": 46.0,
-      "step": 28902
-    },
-    {
-      "epoch": 4.654615725270744,
-      "grad_norm": 0.008907534182071686,
-      "learning_rate": 0.00019998931682215006,
-      "loss": 46.0,
-      "step": 28903
-    },
-    {
-      "epoch": 4.654776762349531,
-      "grad_norm": 0.0017651786329224706,
-      "learning_rate": 0.00019998931608265105,
-      "loss": 46.0,
-      "step": 28904
-    },
-    {
-      "epoch": 4.6549377994283185,
-      "grad_norm": 0.0031292897183448076,
-      "learning_rate": 0.00019998931534312645,
-      "loss": 46.0,
-      "step": 28905
-    },
-    {
-      "epoch": 4.655098836507106,
-      "grad_norm": 0.0037577941548079252,
-      "learning_rate": 0.00019998931460357624,
-      "loss": 46.0,
-      "step": 28906
-    },
-    {
-      "epoch": 4.655259873585893,
-      "grad_norm": 0.005130811594426632,
-      "learning_rate": 0.00019998931386400046,
-      "loss": 46.0,
-      "step": 28907
-    },
-    {
-      "epoch": 4.655420910664681,
-      "grad_norm": 0.007506816182285547,
-      "learning_rate": 0.00019998931312439908,
-      "loss": 46.0,
-      "step": 28908
-    },
-    {
-      "epoch": 4.6555819477434675,
-      "grad_norm": 0.0010403116466477513,
-      "learning_rate": 0.00019998931238477213,
-      "loss": 46.0,
-      "step": 28909
-    },
-    {
-      "epoch": 4.655742984822256,
-      "grad_norm": 0.003379139583557844,
-      "learning_rate": 0.00019998931164511954,
-      "loss": 46.0,
-      "step": 28910
-    },
-    {
-      "epoch": 4.655904021901042,
-      "grad_norm": 0.0038106029387563467,
-      "learning_rate": 0.00019998931090544136,
-      "loss": 46.0,
-      "step": 28911
-    },
-    {
-      "epoch": 4.65606505897983,
-      "grad_norm": 0.0013994182227179408,
-      "learning_rate": 0.00019998931016573763,
-      "loss": 46.0,
-      "step": 28912
-    },
-    {
-      "epoch": 4.656226096058617,
-      "grad_norm": 0.004259624518454075,
-      "learning_rate": 0.00019998930942600828,
-      "loss": 46.0,
-      "step": 28913
-    },
-    {
-      "epoch": 4.656387133137405,
-      "grad_norm": 0.003210079623386264,
-      "learning_rate": 0.00019998930868625337,
-      "loss": 46.0,
-      "step": 28914
-    },
-    {
-      "epoch": 4.656548170216192,
-      "grad_norm": 0.006245154421776533,
-      "learning_rate": 0.00019998930794647285,
-      "loss": 46.0,
-      "step": 28915
-    },
-    {
-      "epoch": 4.65670920729498,
-      "grad_norm": 0.01517784409224987,
-      "learning_rate": 0.0001999893072066667,
-      "loss": 46.0,
-      "step": 28916
-    },
-    {
-      "epoch": 4.656870244373767,
-      "grad_norm": 0.0025047510862350464,
-      "learning_rate": 0.000199989306466835,
-      "loss": 46.0,
-      "step": 28917
-    },
-    {
-      "epoch": 4.6570312814525545,
-      "grad_norm": 0.0025115013122558594,
-      "learning_rate": 0.00019998930572697772,
-      "loss": 46.0,
-      "step": 28918
-    },
-    {
-      "epoch": 4.657192318531342,
-      "grad_norm": 0.011892990209162235,
-      "learning_rate": 0.0001999893049870948,
-      "loss": 46.0,
-      "step": 28919
-    },
-    {
-      "epoch": 4.657353355610129,
-      "grad_norm": 0.0019797985441982746,
-      "learning_rate": 0.0001999893042471863,
-      "loss": 46.0,
-      "step": 28920
-    },
-    {
-      "epoch": 4.657514392688917,
-      "grad_norm": 0.005202241241931915,
-      "learning_rate": 0.00019998930350725223,
-      "loss": 46.0,
-      "step": 28921
-    },
-    {
-      "epoch": 4.657675429767704,
-      "grad_norm": 0.007219049613922834,
-      "learning_rate": 0.00019998930276729255,
-      "loss": 46.0,
-      "step": 28922
-    },
-    {
-      "epoch": 4.657836466846492,
-      "grad_norm": 0.005986146628856659,
-      "learning_rate": 0.0001999893020273073,
-      "loss": 46.0,
-      "step": 28923
-    },
-    {
-      "epoch": 4.657997503925278,
-      "grad_norm": 0.0044911266304552555,
-      "learning_rate": 0.00019998930128729646,
-      "loss": 46.0,
-      "step": 28924
-    },
-    {
-      "epoch": 4.658158541004067,
-      "grad_norm": 0.0028890916146337986,
-      "learning_rate": 0.00019998930054725998,
-      "loss": 46.0,
-      "step": 28925
-    },
-    {
-      "epoch": 4.658319578082853,
-      "grad_norm": 0.0012727015418931842,
-      "learning_rate": 0.00019998929980719797,
-      "loss": 46.0,
-      "step": 28926
-    },
-    {
-      "epoch": 4.658480615161641,
-      "grad_norm": 0.0037314246874302626,
-      "learning_rate": 0.00019998929906711035,
-      "loss": 46.0,
-      "step": 28927
-    },
-    {
-      "epoch": 4.658641652240428,
-      "grad_norm": 0.0020747019443660975,
-      "learning_rate": 0.0001999892983269971,
-      "loss": 46.0,
-      "step": 28928
-    },
-    {
-      "epoch": 4.658802689319216,
-      "grad_norm": 0.002910978626459837,
-      "learning_rate": 0.00019998929758685828,
-      "loss": 46.0,
-      "step": 28929
-    },
-    {
-      "epoch": 4.658963726398003,
-      "grad_norm": 0.007213153410702944,
-      "learning_rate": 0.00019998929684669387,
-      "loss": 46.0,
-      "step": 28930
-    },
-    {
-      "epoch": 4.6591247634767905,
-      "grad_norm": 0.0007391978288069367,
-      "learning_rate": 0.00019998929610650387,
-      "loss": 46.0,
-      "step": 28931
-    },
-    {
-      "epoch": 4.659285800555578,
-      "grad_norm": 0.00462718028575182,
-      "learning_rate": 0.0001999892953662883,
-      "loss": 46.0,
-      "step": 28932
-    },
-    {
-      "epoch": 4.659446837634365,
-      "grad_norm": 0.0071923150680959225,
-      "learning_rate": 0.0001999892946260471,
-      "loss": 46.0,
-      "step": 28933
-    },
-    {
-      "epoch": 4.659607874713153,
-      "grad_norm": 0.009800664149224758,
-      "learning_rate": 0.00019998929388578032,
-      "loss": 46.0,
-      "step": 28934
-    },
-    {
-      "epoch": 4.65976891179194,
-      "grad_norm": 0.0016293359221890569,
-      "learning_rate": 0.00019998929314548794,
-      "loss": 46.0,
-      "step": 28935
-    },
-    {
-      "epoch": 4.659929948870728,
-      "grad_norm": 0.013044043444097042,
-      "learning_rate": 0.00019998929240517,
-      "loss": 46.0,
-      "step": 28936
-    },
-    {
-      "epoch": 4.660090985949515,
-      "grad_norm": 0.0018472142983227968,
-      "learning_rate": 0.00019998929166482646,
-      "loss": 46.0,
-      "step": 28937
-    },
-    {
-      "epoch": 4.660252023028303,
-      "grad_norm": 0.0036964186001569033,
-      "learning_rate": 0.0001999892909244573,
-      "loss": 46.0,
-      "step": 28938
-    },
-    {
-      "epoch": 4.660413060107089,
-      "grad_norm": 0.007954008877277374,
-      "learning_rate": 0.00019998929018406257,
-      "loss": 46.0,
-      "step": 28939
-    },
-    {
-      "epoch": 4.660574097185877,
-      "grad_norm": 0.004352484364062548,
-      "learning_rate": 0.00019998928944364223,
-      "loss": 46.0,
-      "step": 28940
-    },
-    {
-      "epoch": 4.660735134264664,
-      "grad_norm": 0.004022127948701382,
-      "learning_rate": 0.0001999892887031963,
-      "loss": 46.0,
-      "step": 28941
-    },
-    {
-      "epoch": 4.660896171343452,
-      "grad_norm": 0.013982059434056282,
-      "learning_rate": 0.0001999892879627248,
-      "loss": 46.0,
-      "step": 28942
-    },
-    {
-      "epoch": 4.661057208422239,
-      "grad_norm": 0.006098475307226181,
-      "learning_rate": 0.0001999892872222277,
-      "loss": 46.0,
-      "step": 28943
-    },
-    {
-      "epoch": 4.6612182455010265,
-      "grad_norm": 0.0031336217653006315,
-      "learning_rate": 0.000199989286481705,
-      "loss": 46.0,
-      "step": 28944
-    },
-    {
-      "epoch": 4.661379282579814,
-      "grad_norm": 0.005195137578994036,
-      "learning_rate": 0.00019998928574115673,
-      "loss": 46.0,
-      "step": 28945
-    },
-    {
-      "epoch": 4.661540319658601,
-      "grad_norm": 0.003978671506047249,
-      "learning_rate": 0.00019998928500058284,
-      "loss": 46.0,
-      "step": 28946
-    },
-    {
-      "epoch": 4.661701356737389,
-      "grad_norm": 0.0018693781457841396,
-      "learning_rate": 0.00019998928425998336,
-      "loss": 46.0,
-      "step": 28947
-    },
-    {
-      "epoch": 4.661862393816176,
-      "grad_norm": 0.006021400447934866,
-      "learning_rate": 0.00019998928351935827,
-      "loss": 46.0,
-      "step": 28948
-    },
-    {
-      "epoch": 4.662023430894964,
-      "grad_norm": 0.00260544684715569,
-      "learning_rate": 0.00019998928277870762,
-      "loss": 46.0,
-      "step": 28949
-    },
-    {
-      "epoch": 4.662184467973751,
-      "grad_norm": 0.003724590176716447,
-      "learning_rate": 0.00019998928203803138,
-      "loss": 46.0,
-      "step": 28950
-    },
-    {
-      "epoch": 4.662345505052539,
-      "grad_norm": 0.0018055383116006851,
-      "learning_rate": 0.00019998928129732956,
-      "loss": 46.0,
-      "step": 28951
-    },
-    {
-      "epoch": 4.662506542131326,
-      "grad_norm": 0.002299807732924819,
-      "learning_rate": 0.00019998928055660212,
-      "loss": 46.0,
-      "step": 28952
-    },
-    {
-      "epoch": 4.6626675792101135,
-      "grad_norm": 0.0006217075861059129,
-      "learning_rate": 0.00019998927981584907,
-      "loss": 46.0,
-      "step": 28953
-    },
-    {
-      "epoch": 4.6628286162889,
-      "grad_norm": 0.006435351446270943,
-      "learning_rate": 0.00019998927907507045,
-      "loss": 46.0,
-      "step": 28954
-    },
-    {
-      "epoch": 4.662989653367688,
-      "grad_norm": 0.008741314522922039,
-      "learning_rate": 0.00019998927833426625,
-      "loss": 46.0,
-      "step": 28955
-    },
-    {
-      "epoch": 4.663150690446475,
-      "grad_norm": 0.003162316745147109,
-      "learning_rate": 0.00019998927759343644,
-      "loss": 46.0,
-      "step": 28956
-    },
-    {
-      "epoch": 4.6633117275252625,
-      "grad_norm": 0.0022366230841726065,
-      "learning_rate": 0.00019998927685258103,
-      "loss": 46.0,
-      "step": 28957
-    },
-    {
-      "epoch": 4.66347276460405,
-      "grad_norm": 0.0020759697072207928,
-      "learning_rate": 0.00019998927611170007,
-      "loss": 46.0,
-      "step": 28958
-    },
-    {
-      "epoch": 4.663633801682837,
-      "grad_norm": 0.012824960052967072,
-      "learning_rate": 0.0001999892753707935,
-      "loss": 46.0,
-      "step": 28959
-    },
-    {
-      "epoch": 4.663794838761625,
-      "grad_norm": 0.003920450806617737,
-      "learning_rate": 0.0001999892746298613,
-      "loss": 46.0,
-      "step": 28960
-    },
-    {
-      "epoch": 4.663955875840412,
-      "grad_norm": 0.002151580061763525,
-      "learning_rate": 0.00019998927388890353,
-      "loss": 46.0,
-      "step": 28961
-    },
-    {
-      "epoch": 4.6641169129192,
-      "grad_norm": 0.0014872851315885782,
-      "learning_rate": 0.0001999892731479202,
-      "loss": 46.0,
-      "step": 28962
-    },
-    {
-      "epoch": 4.664277949997987,
-      "grad_norm": 0.009645538404583931,
-      "learning_rate": 0.00019998927240691124,
-      "loss": 46.0,
-      "step": 28963
-    },
-    {
-      "epoch": 4.664438987076775,
-      "grad_norm": 0.002582182874903083,
-      "learning_rate": 0.0001999892716658767,
-      "loss": 46.0,
-      "step": 28964
-    },
-    {
-      "epoch": 4.664600024155562,
-      "grad_norm": 0.001219027559272945,
-      "learning_rate": 0.00019998927092481657,
-      "loss": 46.0,
-      "step": 28965
-    },
-    {
-      "epoch": 4.6647610612343495,
-      "grad_norm": 0.0031504600774496794,
-      "learning_rate": 0.00019998927018373083,
-      "loss": 46.0,
-      "step": 28966
-    },
-    {
-      "epoch": 4.664922098313137,
-      "grad_norm": 0.0031122055370360613,
-      "learning_rate": 0.00019998926944261953,
-      "loss": 46.0,
-      "step": 28967
-    },
-    {
-      "epoch": 4.665083135391924,
-      "grad_norm": 0.0015825171722099185,
-      "learning_rate": 0.0001999892687014826,
-      "loss": 46.0,
-      "step": 28968
-    },
-    {
-      "epoch": 4.665244172470711,
-      "grad_norm": 0.0006179422489367425,
-      "learning_rate": 0.00019998926796032008,
-      "loss": 46.0,
-      "step": 28969
-    },
-    {
-      "epoch": 4.6654052095494984,
-      "grad_norm": 0.002954722149297595,
-      "learning_rate": 0.000199989267219132,
-      "loss": 46.0,
-      "step": 28970
-    },
-    {
-      "epoch": 4.665566246628286,
-      "grad_norm": 0.0018114696722477674,
-      "learning_rate": 0.00019998926647791834,
-      "loss": 46.0,
-      "step": 28971
-    },
-    {
-      "epoch": 4.665727283707073,
-      "grad_norm": 0.002110788132995367,
-      "learning_rate": 0.00019998926573667905,
-      "loss": 46.0,
-      "step": 28972
-    },
-    {
-      "epoch": 4.665888320785861,
-      "grad_norm": 0.009134180843830109,
-      "learning_rate": 0.00019998926499541417,
-      "loss": 46.0,
-      "step": 28973
-    },
-    {
-      "epoch": 4.666049357864648,
-      "grad_norm": 0.01235402375459671,
-      "learning_rate": 0.0001999892642541237,
-      "loss": 46.0,
-      "step": 28974
-    },
-    {
-      "epoch": 4.666210394943436,
-      "grad_norm": 0.005200848914682865,
-      "learning_rate": 0.00019998926351280766,
-      "loss": 46.0,
-      "step": 28975
-    },
-    {
-      "epoch": 4.666371432022223,
-      "grad_norm": 0.0018218058394268155,
-      "learning_rate": 0.00019998926277146604,
-      "loss": 46.0,
-      "step": 28976
-    },
-    {
-      "epoch": 4.666532469101011,
-      "grad_norm": 0.009023908525705338,
-      "learning_rate": 0.0001999892620300988,
-      "loss": 46.0,
-      "step": 28977
-    },
-    {
-      "epoch": 4.666693506179798,
-      "grad_norm": 0.0024534212425351143,
-      "learning_rate": 0.00019998926128870595,
-      "loss": 46.0,
-      "step": 28978
-    },
-    {
-      "epoch": 4.6668545432585855,
-      "grad_norm": 0.013477045111358166,
-      "learning_rate": 0.00019998926054728752,
-      "loss": 46.0,
-      "step": 28979
-    },
-    {
-      "epoch": 4.667015580337373,
-      "grad_norm": 0.004790612496435642,
-      "learning_rate": 0.0001999892598058435,
-      "loss": 46.0,
-      "step": 28980
-    },
-    {
-      "epoch": 4.66717661741616,
-      "grad_norm": 0.004339388106018305,
-      "learning_rate": 0.0001999892590643739,
-      "loss": 46.0,
-      "step": 28981
-    },
-    {
-      "epoch": 4.667337654494947,
-      "grad_norm": 0.00179983489215374,
-      "learning_rate": 0.0001999892583228787,
-      "loss": 46.0,
-      "step": 28982
-    },
-    {
-      "epoch": 4.667498691573735,
-      "grad_norm": 0.0025336977560073137,
-      "learning_rate": 0.0001999892575813579,
-      "loss": 46.0,
-      "step": 28983
-    },
-    {
-      "epoch": 4.667659728652522,
-      "grad_norm": 0.009031030349433422,
-      "learning_rate": 0.00019998925683981154,
-      "loss": 46.0,
-      "step": 28984
-    },
-    {
-      "epoch": 4.667820765731309,
-      "grad_norm": 0.002823849208652973,
-      "learning_rate": 0.00019998925609823956,
-      "loss": 46.0,
-      "step": 28985
-    },
-    {
-      "epoch": 4.667981802810097,
-      "grad_norm": 0.0065071796998381615,
-      "learning_rate": 0.000199989255356642,
-      "loss": 46.0,
-      "step": 28986
-    },
-    {
-      "epoch": 4.668142839888884,
-      "grad_norm": 0.0005395651096478105,
-      "learning_rate": 0.00019998925461501882,
-      "loss": 46.0,
-      "step": 28987
-    },
-    {
-      "epoch": 4.668303876967672,
-      "grad_norm": 0.005989130586385727,
-      "learning_rate": 0.00019998925387337008,
-      "loss": 46.0,
-      "step": 28988
-    },
-    {
-      "epoch": 4.668464914046459,
-      "grad_norm": 0.002158117713406682,
-      "learning_rate": 0.00019998925313169575,
-      "loss": 46.0,
-      "step": 28989
-    },
-    {
-      "epoch": 4.668625951125247,
-      "grad_norm": 0.004594181198626757,
-      "learning_rate": 0.0001999892523899958,
-      "loss": 46.0,
-      "step": 28990
-    },
-    {
-      "epoch": 4.668786988204034,
-      "grad_norm": 0.0063536809757351875,
-      "learning_rate": 0.00019998925164827028,
-      "loss": 46.0,
-      "step": 28991
-    },
-    {
-      "epoch": 4.6689480252828215,
-      "grad_norm": 0.002324148081243038,
-      "learning_rate": 0.00019998925090651914,
-      "loss": 46.0,
-      "step": 28992
-    },
-    {
-      "epoch": 4.669109062361609,
-      "grad_norm": 0.0036857454106211662,
-      "learning_rate": 0.00019998925016474246,
-      "loss": 46.0,
-      "step": 28993
-    },
-    {
-      "epoch": 4.669270099440396,
-      "grad_norm": 0.0008407070417888463,
-      "learning_rate": 0.00019998924942294015,
-      "loss": 46.0,
-      "step": 28994
-    },
-    {
-      "epoch": 4.669431136519184,
-      "grad_norm": 0.009376151487231255,
-      "learning_rate": 0.00019998924868111224,
-      "loss": 46.0,
-      "step": 28995
-    },
-    {
-      "epoch": 4.669592173597971,
-      "grad_norm": 0.003775755176320672,
-      "learning_rate": 0.00019998924793925875,
-      "loss": 46.0,
-      "step": 28996
-    },
-    {
-      "epoch": 4.669753210676758,
-      "grad_norm": 0.004688691347837448,
-      "learning_rate": 0.00019998924719737967,
-      "loss": 46.0,
-      "step": 28997
-    },
-    {
-      "epoch": 4.669914247755546,
-      "grad_norm": 0.02092566527426243,
-      "learning_rate": 0.00019998924645547498,
-      "loss": 46.0,
-      "step": 28998
-    },
-    {
-      "epoch": 4.670075284834333,
-      "grad_norm": 0.012101748958230019,
-      "learning_rate": 0.00019998924571354472,
-      "loss": 46.0,
-      "step": 28999
-    },
-    {
-      "epoch": 4.67023632191312,
-      "grad_norm": 0.014231319539248943,
-      "learning_rate": 0.00019998924497158888,
-      "loss": 46.0,
-      "step": 29000
-    },
-    {
-      "epoch": 4.670397358991908,
-      "grad_norm": 0.004863038659095764,
-      "learning_rate": 0.00019998924422960743,
-      "loss": 46.0,
-      "step": 29001
-    },
-    {
-      "epoch": 4.670558396070695,
-      "grad_norm": 0.0008185673505067825,
-      "learning_rate": 0.0001999892434876004,
-      "loss": 46.0,
-      "step": 29002
-    },
-    {
-      "epoch": 4.670719433149483,
-      "grad_norm": 0.009100433439016342,
-      "learning_rate": 0.00019998924274556776,
-      "loss": 46.0,
-      "step": 29003
-    },
-    {
-      "epoch": 4.67088047022827,
-      "grad_norm": 0.004728530067950487,
-      "learning_rate": 0.00019998924200350952,
-      "loss": 46.0,
-      "step": 29004
-    },
-    {
-      "epoch": 4.6710415073070575,
-      "grad_norm": 0.0031026548240333796,
-      "learning_rate": 0.0001999892412614257,
-      "loss": 46.0,
-      "step": 29005
-    },
-    {
-      "epoch": 4.671202544385845,
-      "grad_norm": 0.0032068907748907804,
-      "learning_rate": 0.00019998924051931632,
-      "loss": 46.0,
-      "step": 29006
-    },
-    {
-      "epoch": 4.671363581464632,
-      "grad_norm": 0.005066788289695978,
-      "learning_rate": 0.0001999892397771813,
-      "loss": 46.0,
-      "step": 29007
-    },
-    {
-      "epoch": 4.67152461854342,
-      "grad_norm": 0.004904876463115215,
-      "learning_rate": 0.00019998923903502073,
-      "loss": 46.0,
-      "step": 29008
-    },
-    {
-      "epoch": 4.671685655622207,
-      "grad_norm": 0.0179563257843256,
-      "learning_rate": 0.00019998923829283452,
-      "loss": 46.0,
-      "step": 29009
-    },
-    {
-      "epoch": 4.671846692700995,
-      "grad_norm": 0.004103793762624264,
-      "learning_rate": 0.00019998923755062275,
-      "loss": 46.0,
-      "step": 29010
-    },
-    {
-      "epoch": 4.672007729779782,
-      "grad_norm": 0.0026706280186772346,
-      "learning_rate": 0.00019998923680838537,
-      "loss": 46.0,
-      "step": 29011
-    },
-    {
-      "epoch": 4.672168766858569,
-      "grad_norm": 0.005859110038727522,
-      "learning_rate": 0.0001999892360661224,
-      "loss": 46.0,
-      "step": 29012
-    },
-    {
-      "epoch": 4.672329803937356,
-      "grad_norm": 0.0018263488309457898,
-      "learning_rate": 0.00019998923532383388,
-      "loss": 46.0,
-      "step": 29013
-    },
-    {
-      "epoch": 4.672490841016144,
-      "grad_norm": 0.011946558952331543,
-      "learning_rate": 0.0001999892345815197,
-      "loss": 46.0,
-      "step": 29014
-    },
-    {
-      "epoch": 4.672651878094931,
-      "grad_norm": 0.0055854138918221,
-      "learning_rate": 0.00019998923383917998,
-      "loss": 46.0,
-      "step": 29015
-    },
-    {
-      "epoch": 4.6728129151737186,
-      "grad_norm": 0.0018073191167786717,
-      "learning_rate": 0.00019998923309681464,
-      "loss": 46.0,
-      "step": 29016
-    },
-    {
-      "epoch": 4.672973952252506,
-      "grad_norm": 0.006920395884662867,
-      "learning_rate": 0.00019998923235442373,
-      "loss": 46.0,
-      "step": 29017
-    },
-    {
-      "epoch": 4.6731349893312935,
-      "grad_norm": 0.002207123441621661,
-      "learning_rate": 0.0001999892316120072,
-      "loss": 46.0,
-      "step": 29018
-    },
-    {
-      "epoch": 4.673296026410081,
-      "grad_norm": 0.0018793592462316155,
-      "learning_rate": 0.00019998923086956508,
-      "loss": 46.0,
-      "step": 29019
-    },
-    {
-      "epoch": 4.673457063488868,
-      "grad_norm": 0.0027318818029016256,
-      "learning_rate": 0.00019998923012709742,
-      "loss": 46.0,
-      "step": 29020
-    },
-    {
-      "epoch": 4.673618100567656,
-      "grad_norm": 0.0016887737438082695,
-      "learning_rate": 0.0001999892293846041,
-      "loss": 46.0,
-      "step": 29021
-    },
-    {
-      "epoch": 4.673779137646443,
-      "grad_norm": 0.0023246132768690586,
-      "learning_rate": 0.00019998922864208522,
-      "loss": 46.0,
-      "step": 29022
-    },
-    {
-      "epoch": 4.673940174725231,
-      "grad_norm": 0.004542097914963961,
-      "learning_rate": 0.00019998922789954076,
-      "loss": 46.0,
-      "step": 29023
-    },
-    {
-      "epoch": 4.674101211804018,
-      "grad_norm": 0.02634265273809433,
-      "learning_rate": 0.0001999892271569707,
-      "loss": 46.0,
-      "step": 29024
-    },
-    {
-      "epoch": 4.674262248882806,
-      "grad_norm": 0.003157178871333599,
-      "learning_rate": 0.00019998922641437504,
-      "loss": 46.0,
-      "step": 29025
-    },
-    {
-      "epoch": 4.674423285961593,
-      "grad_norm": 0.004033325240015984,
-      "learning_rate": 0.00019998922567175377,
-      "loss": 46.0,
-      "step": 29026
-    },
-    {
-      "epoch": 4.67458432304038,
-      "grad_norm": 0.01731760799884796,
-      "learning_rate": 0.00019998922492910691,
-      "loss": 46.0,
-      "step": 29027
-    },
-    {
-      "epoch": 4.674745360119167,
-      "grad_norm": 0.002872991608455777,
-      "learning_rate": 0.0001999892241864345,
-      "loss": 46.0,
-      "step": 29028
-    },
-    {
-      "epoch": 4.6749063971979545,
-      "grad_norm": 0.01205460261553526,
-      "learning_rate": 0.00019998922344373647,
-      "loss": 46.0,
-      "step": 29029
-    },
-    {
-      "epoch": 4.675067434276742,
-      "grad_norm": 0.004222155548632145,
-      "learning_rate": 0.00019998922270101285,
-      "loss": 46.0,
-      "step": 29030
-    },
-    {
-      "epoch": 4.675228471355529,
-      "grad_norm": 0.00934133306145668,
-      "learning_rate": 0.00019998922195826362,
-      "loss": 46.0,
-      "step": 29031
-    },
-    {
-      "epoch": 4.675389508434317,
-      "grad_norm": 0.011708592996001244,
-      "learning_rate": 0.00019998922121548885,
-      "loss": 46.0,
-      "step": 29032
-    },
-    {
-      "epoch": 4.675550545513104,
-      "grad_norm": 0.0011638246942311525,
-      "learning_rate": 0.00019998922047268844,
-      "loss": 46.0,
-      "step": 29033
-    },
-    {
-      "epoch": 4.675711582591892,
-      "grad_norm": 0.0007008335669524968,
-      "learning_rate": 0.00019998921972986245,
-      "loss": 46.0,
-      "step": 29034
-    },
-    {
-      "epoch": 4.675872619670679,
-      "grad_norm": 0.006141862832009792,
-      "learning_rate": 0.00019998921898701087,
-      "loss": 46.0,
-      "step": 29035
-    },
-    {
-      "epoch": 4.676033656749467,
-      "grad_norm": 0.004861264023929834,
-      "learning_rate": 0.00019998921824413367,
-      "loss": 46.0,
-      "step": 29036
-    },
-    {
-      "epoch": 4.676194693828254,
-      "grad_norm": 0.009285642765462399,
-      "learning_rate": 0.00019998921750123092,
-      "loss": 46.0,
-      "step": 29037
-    },
-    {
-      "epoch": 4.676355730907042,
-      "grad_norm": 0.0068975286558270454,
-      "learning_rate": 0.00019998921675830258,
-      "loss": 46.0,
-      "step": 29038
-    },
-    {
-      "epoch": 4.676516767985829,
-      "grad_norm": 0.005794209893792868,
-      "learning_rate": 0.00019998921601534862,
-      "loss": 46.0,
-      "step": 29039
-    },
-    {
-      "epoch": 4.6766778050646165,
-      "grad_norm": 0.009501967579126358,
-      "learning_rate": 0.00019998921527236908,
-      "loss": 46.0,
-      "step": 29040
-    },
-    {
-      "epoch": 4.676838842143404,
-      "grad_norm": 0.010775511153042316,
-      "learning_rate": 0.00019998921452936394,
-      "loss": 46.0,
-      "step": 29041
-    },
-    {
-      "epoch": 4.6769998792221905,
-      "grad_norm": 0.0021371403709053993,
-      "learning_rate": 0.00019998921378633323,
-      "loss": 46.0,
-      "step": 29042
-    },
-    {
-      "epoch": 4.677160916300978,
-      "grad_norm": 0.0052638365887105465,
-      "learning_rate": 0.00019998921304327692,
-      "loss": 46.0,
-      "step": 29043
-    },
-    {
-      "epoch": 4.677321953379765,
-      "grad_norm": 0.011134116910398006,
-      "learning_rate": 0.000199989212300195,
-      "loss": 46.0,
-      "step": 29044
-    },
-    {
-      "epoch": 4.677482990458553,
-      "grad_norm": 0.002875329228118062,
-      "learning_rate": 0.00019998921155708747,
-      "loss": 46.0,
-      "step": 29045
-    },
-    {
-      "epoch": 4.67764402753734,
-      "grad_norm": 0.0009744235430844128,
-      "learning_rate": 0.0001999892108139544,
-      "loss": 46.0,
-      "step": 29046
-    },
-    {
-      "epoch": 4.677805064616128,
-      "grad_norm": 0.005998586770147085,
-      "learning_rate": 0.00019998921007079572,
-      "loss": 46.0,
-      "step": 29047
-    },
-    {
-      "epoch": 4.677966101694915,
-      "grad_norm": 0.0025747613981366158,
-      "learning_rate": 0.00019998920932761142,
-      "loss": 46.0,
-      "step": 29048
-    },
-    {
-      "epoch": 4.678127138773703,
-      "grad_norm": 0.005435957573354244,
-      "learning_rate": 0.00019998920858440157,
-      "loss": 46.0,
-      "step": 29049
-    },
-    {
-      "epoch": 4.67828817585249,
-      "grad_norm": 0.002497670240700245,
-      "learning_rate": 0.0001999892078411661,
-      "loss": 46.0,
-      "step": 29050
-    },
-    {
-      "epoch": 4.678449212931278,
-      "grad_norm": 0.004089960362762213,
-      "learning_rate": 0.00019998920709790504,
-      "loss": 46.0,
-      "step": 29051
-    },
-    {
-      "epoch": 4.678610250010065,
-      "grad_norm": 0.0021418847609311342,
-      "learning_rate": 0.0001999892063546184,
-      "loss": 46.0,
-      "step": 29052
-    },
-    {
-      "epoch": 4.6787712870888525,
-      "grad_norm": 0.004259703680872917,
-      "learning_rate": 0.00019998920561130616,
-      "loss": 46.0,
-      "step": 29053
-    },
-    {
-      "epoch": 4.67893232416764,
-      "grad_norm": 0.0005282812635414302,
-      "learning_rate": 0.00019998920486796832,
-      "loss": 46.0,
-      "step": 29054
-    },
-    {
-      "epoch": 4.6790933612464265,
-      "grad_norm": 0.010999402031302452,
-      "learning_rate": 0.00019998920412460489,
-      "loss": 46.0,
-      "step": 29055
-    },
-    {
-      "epoch": 4.679254398325215,
-      "grad_norm": 0.00427471986040473,
-      "learning_rate": 0.0001999892033812159,
-      "loss": 46.0,
-      "step": 29056
-    },
-    {
-      "epoch": 4.679415435404001,
-      "grad_norm": 0.001892294967547059,
-      "learning_rate": 0.00019998920263780129,
-      "loss": 46.0,
-      "step": 29057
-    },
-    {
-      "epoch": 4.679576472482789,
-      "grad_norm": 0.009502123109996319,
-      "learning_rate": 0.00019998920189436106,
-      "loss": 46.0,
-      "step": 29058
-    },
-    {
-      "epoch": 4.679737509561576,
-      "grad_norm": 0.001962902257218957,
-      "learning_rate": 0.0001999892011508953,
-      "loss": 46.0,
-      "step": 29059
-    },
-    {
-      "epoch": 4.679898546640364,
-      "grad_norm": 0.004759308882057667,
-      "learning_rate": 0.0001999892004074039,
-      "loss": 46.0,
-      "step": 29060
-    },
-    {
-      "epoch": 4.680059583719151,
-      "grad_norm": 0.001699597924016416,
-      "learning_rate": 0.00019998919966388693,
-      "loss": 46.0,
-      "step": 29061
-    },
-    {
-      "epoch": 4.680220620797939,
-      "grad_norm": 0.0034312093630433083,
-      "learning_rate": 0.00019998919892034436,
-      "loss": 46.0,
-      "step": 29062
-    },
-    {
-      "epoch": 4.680381657876726,
-      "grad_norm": 0.00467008538544178,
-      "learning_rate": 0.0001999891981767762,
-      "loss": 46.0,
-      "step": 29063
-    },
-    {
-      "epoch": 4.680542694955514,
-      "grad_norm": 0.002616120269522071,
-      "learning_rate": 0.00019998919743318246,
-      "loss": 46.0,
-      "step": 29064
-    },
-    {
-      "epoch": 4.680703732034301,
-      "grad_norm": 0.0087283905595541,
-      "learning_rate": 0.0001999891966895631,
-      "loss": 46.0,
-      "step": 29065
-    },
-    {
-      "epoch": 4.6808647691130885,
-      "grad_norm": 0.006890458986163139,
-      "learning_rate": 0.00019998919594591818,
-      "loss": 46.0,
-      "step": 29066
-    },
-    {
-      "epoch": 4.681025806191876,
-      "grad_norm": 0.0010276816319674253,
-      "learning_rate": 0.00019998919520224765,
-      "loss": 46.0,
-      "step": 29067
-    },
-    {
-      "epoch": 4.681186843270663,
-      "grad_norm": 0.0011364122619852424,
-      "learning_rate": 0.00019998919445855153,
-      "loss": 46.0,
-      "step": 29068
-    },
-    {
-      "epoch": 4.681347880349451,
-      "grad_norm": 0.005245715845376253,
-      "learning_rate": 0.0001999891937148298,
-      "loss": 46.0,
-      "step": 29069
-    },
-    {
-      "epoch": 4.681508917428237,
-      "grad_norm": 0.0015774654457345605,
-      "learning_rate": 0.00019998919297108253,
-      "loss": 46.0,
-      "step": 29070
-    },
-    {
-      "epoch": 4.681669954507026,
-      "grad_norm": 0.004939381033182144,
-      "learning_rate": 0.00019998919222730962,
-      "loss": 46.0,
-      "step": 29071
-    },
-    {
-      "epoch": 4.681830991585812,
-      "grad_norm": 0.010158554650843143,
-      "learning_rate": 0.00019998919148351112,
-      "loss": 46.0,
-      "step": 29072
-    },
-    {
-      "epoch": 4.6819920286646,
-      "grad_norm": 0.013227131217718124,
-      "learning_rate": 0.00019998919073968704,
-      "loss": 46.0,
-      "step": 29073
-    },
-    {
-      "epoch": 4.682153065743387,
-      "grad_norm": 0.010011455975472927,
-      "learning_rate": 0.00019998918999583737,
-      "loss": 46.0,
-      "step": 29074
-    },
-    {
-      "epoch": 4.682314102822175,
-      "grad_norm": 0.002737136324867606,
-      "learning_rate": 0.0001999891892519621,
-      "loss": 46.0,
-      "step": 29075
-    },
-    {
-      "epoch": 4.682475139900962,
-      "grad_norm": 0.0052657099440693855,
-      "learning_rate": 0.00019998918850806126,
-      "loss": 46.0,
-      "step": 29076
-    },
-    {
-      "epoch": 4.6826361769797495,
-      "grad_norm": 0.0023131216876208782,
-      "learning_rate": 0.0001999891877641348,
-      "loss": 46.0,
-      "step": 29077
-    },
-    {
-      "epoch": 4.682797214058537,
-      "grad_norm": 0.0012860159622505307,
-      "learning_rate": 0.00019998918702018273,
-      "loss": 46.0,
-      "step": 29078
-    },
-    {
-      "epoch": 4.6829582511373244,
-      "grad_norm": 0.004784207325428724,
-      "learning_rate": 0.0001999891862762051,
-      "loss": 46.0,
-      "step": 29079
-    },
-    {
-      "epoch": 4.683119288216112,
-      "grad_norm": 0.0028047605883330107,
-      "learning_rate": 0.0001999891855322019,
-      "loss": 46.0,
-      "step": 29080
-    },
-    {
-      "epoch": 4.683280325294899,
-      "grad_norm": 0.0030267657712101936,
-      "learning_rate": 0.00019998918478817307,
-      "loss": 46.0,
-      "step": 29081
-    },
-    {
-      "epoch": 4.683441362373687,
-      "grad_norm": 0.004298881161957979,
-      "learning_rate": 0.00019998918404411868,
-      "loss": 46.0,
-      "step": 29082
-    },
-    {
-      "epoch": 4.683602399452474,
-      "grad_norm": 0.0023336512967944145,
-      "learning_rate": 0.0001999891833000387,
-      "loss": 46.0,
-      "step": 29083
-    },
-    {
-      "epoch": 4.683763436531262,
-      "grad_norm": 0.004733358975499868,
-      "learning_rate": 0.00019998918255593307,
-      "loss": 46.0,
-      "step": 29084
-    },
-    {
-      "epoch": 4.683924473610048,
-      "grad_norm": 0.01219208724796772,
-      "learning_rate": 0.00019998918181180191,
-      "loss": 46.0,
-      "step": 29085
-    },
-    {
-      "epoch": 4.684085510688836,
-      "grad_norm": 0.002086380496621132,
-      "learning_rate": 0.00019998918106764512,
-      "loss": 46.0,
-      "step": 29086
-    },
-    {
-      "epoch": 4.684246547767623,
-      "grad_norm": 0.004757540766149759,
-      "learning_rate": 0.00019998918032346273,
-      "loss": 46.0,
-      "step": 29087
-    },
-    {
-      "epoch": 4.684407584846411,
-      "grad_norm": 0.0022966107353568077,
-      "learning_rate": 0.00019998917957925479,
-      "loss": 46.0,
-      "step": 29088
-    },
-    {
-      "epoch": 4.684568621925198,
-      "grad_norm": 0.006842173635959625,
-      "learning_rate": 0.00019998917883502123,
-      "loss": 46.0,
-      "step": 29089
-    },
-    {
-      "epoch": 4.6847296590039855,
-      "grad_norm": 0.012283913791179657,
-      "learning_rate": 0.00019998917809076208,
-      "loss": 46.0,
-      "step": 29090
-    },
-    {
-      "epoch": 4.684890696082773,
-      "grad_norm": 0.000713318819180131,
-      "learning_rate": 0.00019998917734647735,
-      "loss": 46.0,
-      "step": 29091
-    },
-    {
-      "epoch": 4.68505173316156,
-      "grad_norm": 0.004016088787466288,
-      "learning_rate": 0.000199989176602167,
-      "loss": 46.0,
-      "step": 29092
-    },
-    {
-      "epoch": 4.685212770240348,
-      "grad_norm": 0.006681870669126511,
-      "learning_rate": 0.0001999891758578311,
-      "loss": 46.0,
-      "step": 29093
-    },
-    {
-      "epoch": 4.685373807319135,
-      "grad_norm": 0.002273829421028495,
-      "learning_rate": 0.00019998917511346957,
-      "loss": 46.0,
-      "step": 29094
-    },
-    {
-      "epoch": 4.685534844397923,
-      "grad_norm": 0.0017024537082761526,
-      "learning_rate": 0.00019998917436908246,
-      "loss": 46.0,
-      "step": 29095
-    },
-    {
-      "epoch": 4.68569588147671,
-      "grad_norm": 0.0027831485494971275,
-      "learning_rate": 0.0001999891736246698,
-      "loss": 46.0,
-      "step": 29096
-    },
-    {
-      "epoch": 4.685856918555498,
-      "grad_norm": 0.0107642887160182,
-      "learning_rate": 0.00019998917288023148,
-      "loss": 46.0,
-      "step": 29097
-    },
-    {
-      "epoch": 4.686017955634285,
-      "grad_norm": 0.003889176994562149,
-      "learning_rate": 0.00019998917213576758,
-      "loss": 46.0,
-      "step": 29098
-    },
-    {
-      "epoch": 4.686178992713073,
-      "grad_norm": 0.0033913159277290106,
-      "learning_rate": 0.00019998917139127812,
-      "loss": 46.0,
-      "step": 29099
-    },
-    {
-      "epoch": 4.686340029791859,
-      "grad_norm": 0.0023318175226449966,
-      "learning_rate": 0.00019998917064676305,
-      "loss": 46.0,
-      "step": 29100
-    },
-    {
-      "epoch": 4.686501066870647,
-      "grad_norm": 0.001813381677493453,
-      "learning_rate": 0.00019998916990222242,
-      "loss": 46.0,
-      "step": 29101
-    },
-    {
-      "epoch": 4.686662103949434,
-      "grad_norm": 0.001382240210659802,
-      "learning_rate": 0.00019998916915765614,
-      "loss": 46.0,
-      "step": 29102
-    },
-    {
-      "epoch": 4.6868231410282215,
-      "grad_norm": 0.0021454552188515663,
-      "learning_rate": 0.0001999891684130643,
-      "loss": 46.0,
-      "step": 29103
-    },
-    {
-      "epoch": 4.686984178107009,
-      "grad_norm": 0.006069576367735863,
-      "learning_rate": 0.00019998916766844686,
-      "loss": 46.0,
-      "step": 29104
-    },
-    {
-      "epoch": 4.687145215185796,
-      "grad_norm": 0.01616569608449936,
-      "learning_rate": 0.00019998916692380383,
-      "loss": 46.0,
-      "step": 29105
-    },
-    {
-      "epoch": 4.687306252264584,
-      "grad_norm": 0.006197620648890734,
-      "learning_rate": 0.0001999891661791352,
-      "loss": 46.0,
-      "step": 29106
-    },
-    {
-      "epoch": 4.687467289343371,
-      "grad_norm": 0.00939467828720808,
-      "learning_rate": 0.000199989165434441,
-      "loss": 46.0,
-      "step": 29107
-    },
-    {
-      "epoch": 4.687628326422159,
-      "grad_norm": 0.011693950742483139,
-      "learning_rate": 0.0001999891646897212,
-      "loss": 46.0,
-      "step": 29108
-    },
-    {
-      "epoch": 4.687789363500946,
-      "grad_norm": 0.00068492820719257,
-      "learning_rate": 0.0001999891639449758,
-      "loss": 46.0,
-      "step": 29109
-    },
-    {
-      "epoch": 4.687950400579734,
-      "grad_norm": 0.007851121947169304,
-      "learning_rate": 0.0001999891632002048,
-      "loss": 46.0,
-      "step": 29110
-    },
-    {
-      "epoch": 4.688111437658521,
-      "grad_norm": 0.0035151569172739983,
-      "learning_rate": 0.00019998916245540823,
-      "loss": 46.0,
-      "step": 29111
-    },
-    {
-      "epoch": 4.688272474737309,
-      "grad_norm": 0.0029837789479643106,
-      "learning_rate": 0.00019998916171058606,
-      "loss": 46.0,
-      "step": 29112
-    },
-    {
-      "epoch": 4.688433511816096,
-      "grad_norm": 0.003076000139117241,
-      "learning_rate": 0.00019998916096573827,
-      "loss": 46.0,
-      "step": 29113
-    },
-    {
-      "epoch": 4.6885945488948835,
-      "grad_norm": 0.0014842329546809196,
-      "learning_rate": 0.00019998916022086492,
-      "loss": 46.0,
-      "step": 29114
-    },
-    {
-      "epoch": 4.68875558597367,
-      "grad_norm": 0.0019867236260324717,
-      "learning_rate": 0.00019998915947596596,
-      "loss": 46.0,
-      "step": 29115
-    },
-    {
-      "epoch": 4.6889166230524575,
-      "grad_norm": 0.0016855103895068169,
-      "learning_rate": 0.00019998915873104144,
-      "loss": 46.0,
-      "step": 29116
-    },
-    {
-      "epoch": 4.689077660131245,
-      "grad_norm": 0.009796492755413055,
-      "learning_rate": 0.0001999891579860913,
-      "loss": 46.0,
-      "step": 29117
-    },
-    {
-      "epoch": 4.689238697210032,
-      "grad_norm": 0.0004139193333685398,
-      "learning_rate": 0.00019998915724111558,
-      "loss": 46.0,
-      "step": 29118
-    },
-    {
-      "epoch": 4.68939973428882,
-      "grad_norm": 0.0029851384460926056,
-      "learning_rate": 0.00019998915649611427,
-      "loss": 46.0,
-      "step": 29119
-    },
-    {
-      "epoch": 4.689560771367607,
-      "grad_norm": 0.008399692364037037,
-      "learning_rate": 0.00019998915575108737,
-      "loss": 46.0,
-      "step": 29120
-    },
-    {
-      "epoch": 4.689721808446395,
-      "grad_norm": 0.005843984428793192,
-      "learning_rate": 0.00019998915500603486,
-      "loss": 46.0,
-      "step": 29121
-    },
-    {
-      "epoch": 4.689882845525182,
-      "grad_norm": 0.0018409896874800324,
-      "learning_rate": 0.00019998915426095676,
-      "loss": 46.0,
-      "step": 29122
-    },
-    {
-      "epoch": 4.69004388260397,
-      "grad_norm": 0.0005754969897679985,
-      "learning_rate": 0.00019998915351585308,
-      "loss": 46.0,
-      "step": 29123
-    },
-    {
-      "epoch": 4.690204919682757,
-      "grad_norm": 0.0020168640185147524,
-      "learning_rate": 0.00019998915277072378,
-      "loss": 46.0,
-      "step": 29124
-    },
-    {
-      "epoch": 4.6903659567615446,
-      "grad_norm": 0.0038000138010829687,
-      "learning_rate": 0.00019998915202556894,
-      "loss": 46.0,
-      "step": 29125
-    },
-    {
-      "epoch": 4.690526993840332,
-      "grad_norm": 0.0032192859798669815,
-      "learning_rate": 0.00019998915128038847,
-      "loss": 46.0,
-      "step": 29126
-    },
-    {
-      "epoch": 4.6906880309191195,
-      "grad_norm": 0.0025296914391219616,
-      "learning_rate": 0.0001999891505351824,
-      "loss": 46.0,
-      "step": 29127
-    },
-    {
-      "epoch": 4.690849067997906,
-      "grad_norm": 0.0039049312472343445,
-      "learning_rate": 0.0001999891497899508,
-      "loss": 46.0,
-      "step": 29128
-    },
-    {
-      "epoch": 4.691010105076694,
-      "grad_norm": 0.001298865769058466,
-      "learning_rate": 0.00019998914904469355,
-      "loss": 46.0,
-      "step": 29129
-    },
-    {
-      "epoch": 4.691171142155481,
-      "grad_norm": 0.0031968820840120316,
-      "learning_rate": 0.0001999891482994107,
-      "loss": 46.0,
-      "step": 29130
-    },
-    {
-      "epoch": 4.691332179234268,
-      "grad_norm": 0.0017479286761954427,
-      "learning_rate": 0.00019998914755410226,
-      "loss": 46.0,
-      "step": 29131
-    },
-    {
-      "epoch": 4.691493216313056,
-      "grad_norm": 0.006804362870752811,
-      "learning_rate": 0.00019998914680876827,
-      "loss": 46.0,
-      "step": 29132
-    },
-    {
-      "epoch": 4.691654253391843,
-      "grad_norm": 0.0038377020973712206,
-      "learning_rate": 0.00019998914606340865,
-      "loss": 46.0,
-      "step": 29133
-    },
-    {
-      "epoch": 4.691815290470631,
-      "grad_norm": 0.006302055437117815,
-      "learning_rate": 0.00019998914531802346,
-      "loss": 46.0,
-      "step": 29134
-    },
-    {
-      "epoch": 4.691976327549418,
-      "grad_norm": 0.007912084460258484,
-      "learning_rate": 0.00019998914457261267,
-      "loss": 46.0,
-      "step": 29135
-    },
-    {
-      "epoch": 4.692137364628206,
-      "grad_norm": 0.0017726949881762266,
-      "learning_rate": 0.0001999891438271763,
-      "loss": 46.0,
-      "step": 29136
-    },
-    {
-      "epoch": 4.692298401706993,
-      "grad_norm": 0.0031725927256047726,
-      "learning_rate": 0.00019998914308171434,
-      "loss": 46.0,
-      "step": 29137
-    },
-    {
-      "epoch": 4.6924594387857805,
-      "grad_norm": 0.0046320389956235886,
-      "learning_rate": 0.00019998914233622673,
-      "loss": 46.0,
-      "step": 29138
-    },
-    {
-      "epoch": 4.692620475864568,
-      "grad_norm": 0.0028918073512613773,
-      "learning_rate": 0.00019998914159071357,
-      "loss": 46.0,
-      "step": 29139
-    },
-    {
-      "epoch": 4.692781512943355,
-      "grad_norm": 0.002753194887191057,
-      "learning_rate": 0.00019998914084517485,
-      "loss": 46.0,
-      "step": 29140
-    },
-    {
-      "epoch": 4.692942550022143,
-      "grad_norm": 0.006393604911863804,
-      "learning_rate": 0.0001999891400996105,
-      "loss": 46.0,
-      "step": 29141
-    },
-    {
-      "epoch": 4.69310358710093,
-      "grad_norm": 0.001441923901438713,
-      "learning_rate": 0.00019998913935402056,
-      "loss": 46.0,
-      "step": 29142
-    },
-    {
-      "epoch": 4.693264624179717,
-      "grad_norm": 0.008365838788449764,
-      "learning_rate": 0.00019998913860840503,
-      "loss": 46.0,
-      "step": 29143
-    },
-    {
-      "epoch": 4.693425661258505,
-      "grad_norm": 0.008358891122043133,
-      "learning_rate": 0.0001999891378627639,
-      "loss": 46.0,
-      "step": 29144
-    },
-    {
-      "epoch": 4.693586698337292,
-      "grad_norm": 0.0032135392539203167,
-      "learning_rate": 0.0001999891371170972,
-      "loss": 46.0,
-      "step": 29145
-    },
-    {
-      "epoch": 4.693747735416079,
-      "grad_norm": 0.01464059203863144,
-      "learning_rate": 0.00019998913637140492,
-      "loss": 46.0,
-      "step": 29146
-    },
-    {
-      "epoch": 4.693908772494867,
-      "grad_norm": 0.003912672866135836,
-      "learning_rate": 0.000199989135625687,
-      "loss": 46.0,
-      "step": 29147
-    },
-    {
-      "epoch": 4.694069809573654,
-      "grad_norm": 0.004959849175065756,
-      "learning_rate": 0.0001999891348799435,
-      "loss": 46.0,
-      "step": 29148
-    },
-    {
-      "epoch": 4.694230846652442,
-      "grad_norm": 0.0054310900159180164,
-      "learning_rate": 0.00019998913413417442,
-      "loss": 46.0,
-      "step": 29149
-    },
-    {
-      "epoch": 4.694391883731229,
-      "grad_norm": 0.005952246021479368,
-      "learning_rate": 0.00019998913338837974,
-      "loss": 46.0,
-      "step": 29150
-    },
-    {
-      "epoch": 4.6945529208100165,
-      "grad_norm": 0.006007140036672354,
-      "learning_rate": 0.0001999891326425595,
-      "loss": 46.0,
-      "step": 29151
-    },
-    {
-      "epoch": 4.694713957888804,
-      "grad_norm": 0.0020258279982954264,
-      "learning_rate": 0.00019998913189671363,
-      "loss": 46.0,
-      "step": 29152
-    },
-    {
-      "epoch": 4.694874994967591,
-      "grad_norm": 0.005397609435021877,
-      "learning_rate": 0.0001999891311508422,
-      "loss": 46.0,
-      "step": 29153
-    },
-    {
-      "epoch": 4.695036032046379,
-      "grad_norm": 0.011329378932714462,
-      "learning_rate": 0.00019998913040494514,
-      "loss": 46.0,
-      "step": 29154
-    },
-    {
-      "epoch": 4.695197069125166,
-      "grad_norm": 0.0007836871664039791,
-      "learning_rate": 0.00019998912965902253,
-      "loss": 46.0,
-      "step": 29155
-    },
-    {
-      "epoch": 4.695358106203954,
-      "grad_norm": 0.01549210213124752,
-      "learning_rate": 0.00019998912891307428,
-      "loss": 46.0,
-      "step": 29156
-    },
-    {
-      "epoch": 4.695519143282741,
-      "grad_norm": 0.0009757225052453578,
-      "learning_rate": 0.00019998912816710047,
-      "loss": 46.0,
-      "step": 29157
-    },
-    {
-      "epoch": 4.695680180361528,
-      "grad_norm": 0.007227322552353144,
-      "learning_rate": 0.00019998912742110107,
-      "loss": 46.0,
-      "step": 29158
-    },
-    {
-      "epoch": 4.695841217440316,
-      "grad_norm": 0.005283804144710302,
-      "learning_rate": 0.00019998912667507605,
-      "loss": 46.0,
-      "step": 29159
-    },
-    {
-      "epoch": 4.696002254519103,
-      "grad_norm": 0.002289191819727421,
-      "learning_rate": 0.00019998912592902545,
-      "loss": 46.0,
-      "step": 29160
-    },
-    {
-      "epoch": 4.69616329159789,
-      "grad_norm": 0.004192507825791836,
-      "learning_rate": 0.00019998912518294926,
-      "loss": 46.0,
-      "step": 29161
-    },
-    {
-      "epoch": 4.696324328676678,
-      "grad_norm": 0.00043169312993995845,
-      "learning_rate": 0.00019998912443684752,
-      "loss": 46.0,
-      "step": 29162
-    },
-    {
-      "epoch": 4.696485365755465,
-      "grad_norm": 0.006443547084927559,
-      "learning_rate": 0.00019998912369072013,
-      "loss": 46.0,
-      "step": 29163
-    },
-    {
-      "epoch": 4.6966464028342525,
-      "grad_norm": 0.00981169193983078,
-      "learning_rate": 0.00019998912294456718,
-      "loss": 46.0,
-      "step": 29164
-    },
-    {
-      "epoch": 4.69680743991304,
-      "grad_norm": 0.008081848733127117,
-      "learning_rate": 0.00019998912219838864,
-      "loss": 46.0,
-      "step": 29165
-    },
-    {
-      "epoch": 4.696968476991827,
-      "grad_norm": 0.0027035672683268785,
-      "learning_rate": 0.00019998912145218446,
-      "loss": 46.0,
-      "step": 29166
-    },
-    {
-      "epoch": 4.697129514070615,
-      "grad_norm": 0.014703537337481976,
-      "learning_rate": 0.00019998912070595475,
-      "loss": 46.0,
-      "step": 29167
-    },
-    {
-      "epoch": 4.697290551149402,
-      "grad_norm": 0.004574276506900787,
-      "learning_rate": 0.0001999891199596994,
-      "loss": 46.0,
-      "step": 29168
-    },
-    {
-      "epoch": 4.69745158822819,
-      "grad_norm": 0.0051938071846961975,
-      "learning_rate": 0.0001999891192134185,
-      "loss": 46.0,
-      "step": 29169
-    },
-    {
-      "epoch": 4.697612625306977,
-      "grad_norm": 0.0034741077106446028,
-      "learning_rate": 0.00019998911846711196,
-      "loss": 46.0,
-      "step": 29170
-    },
-    {
-      "epoch": 4.697773662385765,
-      "grad_norm": 0.0009269822621718049,
-      "learning_rate": 0.00019998911772077987,
-      "loss": 46.0,
-      "step": 29171
-    },
-    {
-      "epoch": 4.697934699464552,
-      "grad_norm": 0.00685663940384984,
-      "learning_rate": 0.00019998911697442217,
-      "loss": 46.0,
-      "step": 29172
-    },
-    {
-      "epoch": 4.698095736543339,
-      "grad_norm": 0.0053397961892187595,
-      "learning_rate": 0.00019998911622803888,
-      "loss": 46.0,
-      "step": 29173
-    },
-    {
-      "epoch": 4.698256773622126,
-      "grad_norm": 0.02413760870695114,
-      "learning_rate": 0.00019998911548163,
-      "loss": 46.0,
-      "step": 29174
-    },
-    {
-      "epoch": 4.698417810700914,
-      "grad_norm": 0.003894147463142872,
-      "learning_rate": 0.00019998911473519552,
-      "loss": 46.0,
-      "step": 29175
-    },
-    {
-      "epoch": 4.698578847779701,
-      "grad_norm": 0.000604710599873215,
-      "learning_rate": 0.00019998911398873544,
-      "loss": 46.0,
-      "step": 29176
-    },
-    {
-      "epoch": 4.6987398848584885,
-      "grad_norm": 0.0025112025905400515,
-      "learning_rate": 0.0001999891132422498,
-      "loss": 46.0,
-      "step": 29177
-    },
-    {
-      "epoch": 4.698900921937276,
-      "grad_norm": 0.010967286303639412,
-      "learning_rate": 0.00019998911249573855,
-      "loss": 46.0,
-      "step": 29178
-    },
-    {
-      "epoch": 4.699061959016063,
-      "grad_norm": 0.0010603993432596326,
-      "learning_rate": 0.0001999891117492017,
-      "loss": 46.0,
-      "step": 29179
-    },
-    {
-      "epoch": 4.699222996094851,
-      "grad_norm": 0.001997011713683605,
-      "learning_rate": 0.0001999891110026393,
-      "loss": 46.0,
-      "step": 29180
-    },
-    {
-      "epoch": 4.699384033173638,
-      "grad_norm": 0.007600634824484587,
-      "learning_rate": 0.00019998911025605125,
-      "loss": 46.0,
-      "step": 29181
-    },
-    {
-      "epoch": 4.699545070252426,
-      "grad_norm": 0.008752559311687946,
-      "learning_rate": 0.00019998910950943762,
-      "loss": 46.0,
-      "step": 29182
-    },
-    {
-      "epoch": 4.699706107331213,
-      "grad_norm": 0.006707174703478813,
-      "learning_rate": 0.0001999891087627984,
-      "loss": 46.0,
-      "step": 29183
-    },
-    {
-      "epoch": 4.699867144410001,
-      "grad_norm": 0.003188934875652194,
-      "learning_rate": 0.0001999891080161336,
-      "loss": 46.0,
-      "step": 29184
-    },
-    {
-      "epoch": 4.700028181488788,
-      "grad_norm": 0.0010247660102322698,
-      "learning_rate": 0.00019998910726944322,
-      "loss": 46.0,
-      "step": 29185
-    },
-    {
-      "epoch": 4.7001892185675755,
-      "grad_norm": 0.003599941497668624,
-      "learning_rate": 0.00019998910652272724,
-      "loss": 46.0,
-      "step": 29186
-    },
-    {
-      "epoch": 4.700350255646363,
-      "grad_norm": 0.005215603392571211,
-      "learning_rate": 0.00019998910577598565,
-      "loss": 46.0,
-      "step": 29187
-    },
-    {
-      "epoch": 4.70051129272515,
-      "grad_norm": 0.0033436438534408808,
-      "learning_rate": 0.00019998910502921848,
-      "loss": 46.0,
-      "step": 29188
-    },
-    {
-      "epoch": 4.700672329803937,
-      "grad_norm": 0.00279989093542099,
-      "learning_rate": 0.0001999891042824257,
-      "loss": 46.0,
-      "step": 29189
-    },
-    {
-      "epoch": 4.7008333668827245,
-      "grad_norm": 0.005491578485816717,
-      "learning_rate": 0.00019998910353560736,
-      "loss": 46.0,
-      "step": 29190
-    },
-    {
-      "epoch": 4.700994403961512,
-      "grad_norm": 0.004783495794981718,
-      "learning_rate": 0.00019998910278876342,
-      "loss": 46.0,
-      "step": 29191
-    },
-    {
-      "epoch": 4.701155441040299,
-      "grad_norm": 0.0027589083183556795,
-      "learning_rate": 0.00019998910204189387,
-      "loss": 46.0,
-      "step": 29192
-    },
-    {
-      "epoch": 4.701316478119087,
-      "grad_norm": 0.007185005582869053,
-      "learning_rate": 0.00019998910129499876,
-      "loss": 46.0,
-      "step": 29193
-    },
-    {
-      "epoch": 4.701477515197874,
-      "grad_norm": 0.0050863465294241905,
-      "learning_rate": 0.00019998910054807803,
-      "loss": 46.0,
-      "step": 29194
-    },
-    {
-      "epoch": 4.701638552276662,
-      "grad_norm": 0.0014377750921994448,
-      "learning_rate": 0.00019998909980113169,
-      "loss": 46.0,
-      "step": 29195
-    },
-    {
-      "epoch": 4.701799589355449,
-      "grad_norm": 0.0034104082733392715,
-      "learning_rate": 0.00019998909905415978,
-      "loss": 46.0,
-      "step": 29196
-    },
-    {
-      "epoch": 4.701960626434237,
-      "grad_norm": 0.008074586279690266,
-      "learning_rate": 0.0001999890983071623,
-      "loss": 46.0,
-      "step": 29197
-    },
-    {
-      "epoch": 4.702121663513024,
-      "grad_norm": 0.0018546972423791885,
-      "learning_rate": 0.0001999890975601392,
-      "loss": 46.0,
-      "step": 29198
-    },
-    {
-      "epoch": 4.7022827005918115,
-      "grad_norm": 0.005170183256268501,
-      "learning_rate": 0.00019998909681309053,
-      "loss": 46.0,
-      "step": 29199
-    },
-    {
-      "epoch": 4.702443737670599,
-      "grad_norm": 0.0027615155559033155,
-      "learning_rate": 0.00019998909606601622,
-      "loss": 46.0,
-      "step": 29200
-    },
-    {
-      "epoch": 4.702604774749386,
-      "grad_norm": 0.002034719567745924,
-      "learning_rate": 0.0001999890953189164,
-      "loss": 46.0,
-      "step": 29201
-    },
-    {
-      "epoch": 4.702765811828174,
-      "grad_norm": 0.010149680078029633,
-      "learning_rate": 0.0001999890945717909,
-      "loss": 46.0,
-      "step": 29202
-    },
-    {
-      "epoch": 4.7029268489069604,
-      "grad_norm": 0.003622225718572736,
-      "learning_rate": 0.00019998909382463987,
-      "loss": 46.0,
-      "step": 29203
-    },
-    {
-      "epoch": 4.703087885985748,
-      "grad_norm": 0.0034116932656615973,
-      "learning_rate": 0.00019998909307746322,
-      "loss": 46.0,
-      "step": 29204
-    },
-    {
-      "epoch": 4.703248923064535,
-      "grad_norm": 0.006020493805408478,
-      "learning_rate": 0.00019998909233026098,
-      "loss": 46.0,
-      "step": 29205
-    },
-    {
-      "epoch": 4.703409960143323,
-      "grad_norm": 0.002413938520476222,
-      "learning_rate": 0.00019998909158303315,
-      "loss": 46.0,
-      "step": 29206
-    },
-    {
-      "epoch": 4.70357099722211,
-      "grad_norm": 0.006398188881576061,
-      "learning_rate": 0.00019998909083577973,
-      "loss": 46.0,
-      "step": 29207
-    },
-    {
-      "epoch": 4.703732034300898,
-      "grad_norm": 0.00417255237698555,
-      "learning_rate": 0.00019998909008850073,
-      "loss": 46.0,
-      "step": 29208
-    },
-    {
-      "epoch": 4.703893071379685,
-      "grad_norm": 0.002781786723062396,
-      "learning_rate": 0.00019998908934119612,
-      "loss": 46.0,
-      "step": 29209
-    },
-    {
-      "epoch": 4.704054108458473,
-      "grad_norm": 0.003387131029739976,
-      "learning_rate": 0.0001999890885938659,
-      "loss": 46.0,
-      "step": 29210
-    },
-    {
-      "epoch": 4.70421514553726,
-      "grad_norm": 0.003240004414692521,
-      "learning_rate": 0.00019998908784651015,
-      "loss": 46.0,
-      "step": 29211
-    },
-    {
-      "epoch": 4.7043761826160475,
-      "grad_norm": 0.0025013363920152187,
-      "learning_rate": 0.00019998908709912874,
-      "loss": 46.0,
-      "step": 29212
-    },
-    {
-      "epoch": 4.704537219694835,
-      "grad_norm": 0.002901485189795494,
-      "learning_rate": 0.00019998908635172178,
-      "loss": 46.0,
-      "step": 29213
-    },
-    {
-      "epoch": 4.704698256773622,
-      "grad_norm": 0.00529779028147459,
-      "learning_rate": 0.0001999890856042892,
-      "loss": 46.0,
-      "step": 29214
-    },
-    {
-      "epoch": 4.70485929385241,
-      "grad_norm": 0.012361081317067146,
-      "learning_rate": 0.00019998908485683106,
-      "loss": 46.0,
-      "step": 29215
-    },
-    {
-      "epoch": 4.705020330931196,
-      "grad_norm": 0.01101479772478342,
-      "learning_rate": 0.00019998908410934728,
-      "loss": 46.0,
-      "step": 29216
-    },
-    {
-      "epoch": 4.705181368009985,
-      "grad_norm": 0.017921939492225647,
-      "learning_rate": 0.00019998908336183794,
-      "loss": 46.0,
-      "step": 29217
-    },
-    {
-      "epoch": 4.705342405088771,
-      "grad_norm": 0.0027513159438967705,
-      "learning_rate": 0.000199989082614303,
-      "loss": 46.0,
-      "step": 29218
-    },
-    {
-      "epoch": 4.705503442167559,
-      "grad_norm": 0.0020752442069351673,
-      "learning_rate": 0.0001999890818667425,
-      "loss": 46.0,
-      "step": 29219
-    },
-    {
-      "epoch": 4.705664479246346,
-      "grad_norm": 0.009334743954241276,
-      "learning_rate": 0.0001999890811191564,
-      "loss": 46.0,
-      "step": 29220
-    },
-    {
-      "epoch": 4.705825516325134,
-      "grad_norm": 0.006994061172008514,
-      "learning_rate": 0.00019998908037154468,
-      "loss": 46.0,
-      "step": 29221
-    },
-    {
-      "epoch": 4.705986553403921,
-      "grad_norm": 0.007131069432944059,
-      "learning_rate": 0.00019998907962390735,
-      "loss": 46.0,
-      "step": 29222
-    },
-    {
-      "epoch": 4.706147590482709,
-      "grad_norm": 0.002933935495093465,
-      "learning_rate": 0.00019998907887624448,
-      "loss": 46.0,
-      "step": 29223
-    },
-    {
-      "epoch": 4.706308627561496,
-      "grad_norm": 0.006688784807920456,
-      "learning_rate": 0.000199989078128556,
-      "loss": 46.0,
-      "step": 29224
-    },
-    {
-      "epoch": 4.7064696646402835,
-      "grad_norm": 0.0032244024332612753,
-      "learning_rate": 0.0001999890773808419,
-      "loss": 46.0,
-      "step": 29225
-    },
-    {
-      "epoch": 4.706630701719071,
-      "grad_norm": 0.009899741038680077,
-      "learning_rate": 0.00019998907663310223,
-      "loss": 46.0,
-      "step": 29226
-    },
-    {
-      "epoch": 4.706791738797858,
-      "grad_norm": 0.004525924101471901,
-      "learning_rate": 0.00019998907588533697,
-      "loss": 46.0,
-      "step": 29227
-    },
-    {
-      "epoch": 4.706952775876646,
-      "grad_norm": 0.0018723371904343367,
-      "learning_rate": 0.0001999890751375461,
-      "loss": 46.0,
-      "step": 29228
-    },
-    {
-      "epoch": 4.707113812955433,
-      "grad_norm": 0.0017093382775783539,
-      "learning_rate": 0.0001999890743897297,
-      "loss": 46.0,
-      "step": 29229
-    },
-    {
-      "epoch": 4.707274850034221,
-      "grad_norm": 0.003120125737041235,
-      "learning_rate": 0.00019998907364188764,
-      "loss": 46.0,
-      "step": 29230
-    },
-    {
-      "epoch": 4.707435887113007,
-      "grad_norm": 0.001238172990269959,
-      "learning_rate": 0.00019998907289402,
-      "loss": 46.0,
-      "step": 29231
-    },
-    {
-      "epoch": 4.707596924191796,
-      "grad_norm": 0.0034787170588970184,
-      "learning_rate": 0.0001999890721461268,
-      "loss": 46.0,
-      "step": 29232
-    },
-    {
-      "epoch": 4.707757961270582,
-      "grad_norm": 0.004233032464981079,
-      "learning_rate": 0.00019998907139820798,
-      "loss": 46.0,
-      "step": 29233
-    },
-    {
-      "epoch": 4.70791899834937,
-      "grad_norm": 0.002697734860703349,
-      "learning_rate": 0.00019998907065026358,
-      "loss": 46.0,
-      "step": 29234
-    },
-    {
-      "epoch": 4.708080035428157,
-      "grad_norm": 0.005337487440556288,
-      "learning_rate": 0.0001999890699022936,
-      "loss": 46.0,
-      "step": 29235
-    },
-    {
-      "epoch": 4.708241072506945,
-      "grad_norm": 0.0030276046600192785,
-      "learning_rate": 0.00019998906915429798,
-      "loss": 46.0,
-      "step": 29236
-    },
-    {
-      "epoch": 4.708402109585732,
-      "grad_norm": 0.0014151401119306684,
-      "learning_rate": 0.0001999890684062768,
-      "loss": 46.0,
-      "step": 29237
-    },
-    {
-      "epoch": 4.7085631466645195,
-      "grad_norm": 0.002331359777599573,
-      "learning_rate": 0.00019998906765823,
-      "loss": 46.0,
-      "step": 29238
-    },
-    {
-      "epoch": 4.708724183743307,
-      "grad_norm": 0.007948044687509537,
-      "learning_rate": 0.00019998906691015767,
-      "loss": 46.0,
-      "step": 29239
-    },
-    {
-      "epoch": 4.708885220822094,
-      "grad_norm": 0.003867212450131774,
-      "learning_rate": 0.0001999890661620597,
-      "loss": 46.0,
-      "step": 29240
-    },
-    {
-      "epoch": 4.709046257900882,
-      "grad_norm": 0.007837440818548203,
-      "learning_rate": 0.00019998906541393615,
-      "loss": 46.0,
-      "step": 29241
-    },
-    {
-      "epoch": 4.709207294979669,
-      "grad_norm": 0.003048261860385537,
-      "learning_rate": 0.000199989064665787,
-      "loss": 46.0,
-      "step": 29242
-    },
-    {
-      "epoch": 4.709368332058457,
-      "grad_norm": 0.0028265388682484627,
-      "learning_rate": 0.00019998906391761225,
-      "loss": 46.0,
-      "step": 29243
-    },
-    {
-      "epoch": 4.709529369137244,
-      "grad_norm": 0.0034669069573283195,
-      "learning_rate": 0.00019998906316941195,
-      "loss": 46.0,
-      "step": 29244
-    },
-    {
-      "epoch": 4.709690406216032,
-      "grad_norm": 0.0012782543199136853,
-      "learning_rate": 0.00019998906242118603,
-      "loss": 46.0,
-      "step": 29245
-    },
-    {
-      "epoch": 4.709851443294818,
-      "grad_norm": 0.0034581669606268406,
-      "learning_rate": 0.00019998906167293453,
-      "loss": 46.0,
-      "step": 29246
-    },
-    {
-      "epoch": 4.710012480373606,
-      "grad_norm": 0.0022560099605470896,
-      "learning_rate": 0.00019998906092465738,
-      "loss": 46.0,
-      "step": 29247
-    },
-    {
-      "epoch": 4.710173517452393,
-      "grad_norm": 0.00527240289375186,
-      "learning_rate": 0.0001999890601763547,
-      "loss": 46.0,
-      "step": 29248
-    },
-    {
-      "epoch": 4.7103345545311806,
-      "grad_norm": 0.004832895006984472,
-      "learning_rate": 0.0001999890594280264,
-      "loss": 46.0,
-      "step": 29249
-    },
-    {
-      "epoch": 4.710495591609968,
-      "grad_norm": 0.0005298354080878198,
-      "learning_rate": 0.00019998905867967253,
-      "loss": 46.0,
-      "step": 29250
-    },
-    {
-      "epoch": 4.7106566286887555,
-      "grad_norm": 0.0026842583902180195,
-      "learning_rate": 0.0001999890579312931,
-      "loss": 46.0,
-      "step": 29251
-    },
-    {
-      "epoch": 4.710817665767543,
-      "grad_norm": 0.0022797740530222654,
-      "learning_rate": 0.000199989057182888,
-      "loss": 46.0,
-      "step": 29252
-    },
-    {
-      "epoch": 4.71097870284633,
-      "grad_norm": 0.0024207711685448885,
-      "learning_rate": 0.00019998905643445737,
-      "loss": 46.0,
-      "step": 29253
-    },
-    {
-      "epoch": 4.711139739925118,
-      "grad_norm": 0.0023152839858084917,
-      "learning_rate": 0.0001999890556860011,
-      "loss": 46.0,
-      "step": 29254
-    },
-    {
-      "epoch": 4.711300777003905,
-      "grad_norm": 0.007540998049080372,
-      "learning_rate": 0.00019998905493751927,
-      "loss": 46.0,
-      "step": 29255
-    },
-    {
-      "epoch": 4.711461814082693,
-      "grad_norm": 0.007167678326368332,
-      "learning_rate": 0.00019998905418901184,
-      "loss": 46.0,
-      "step": 29256
-    },
-    {
-      "epoch": 4.71162285116148,
-      "grad_norm": 0.002133867470547557,
-      "learning_rate": 0.0001999890534404788,
-      "loss": 46.0,
-      "step": 29257
-    },
-    {
-      "epoch": 4.711783888240268,
-      "grad_norm": 0.0020729033276438713,
-      "learning_rate": 0.0001999890526919202,
-      "loss": 46.0,
-      "step": 29258
-    },
-    {
-      "epoch": 4.711944925319055,
-      "grad_norm": 0.0018183771753683686,
-      "learning_rate": 0.00019998905194333597,
-      "loss": 46.0,
-      "step": 29259
-    },
-    {
-      "epoch": 4.7121059623978425,
-      "grad_norm": 0.003653302090242505,
-      "learning_rate": 0.00019998905119472616,
-      "loss": 46.0,
-      "step": 29260
-    },
-    {
-      "epoch": 4.712266999476629,
-      "grad_norm": 0.0081028426066041,
-      "learning_rate": 0.0001999890504460908,
-      "loss": 46.0,
-      "step": 29261
-    },
-    {
-      "epoch": 4.7124280365554165,
-      "grad_norm": 0.004289961885660887,
-      "learning_rate": 0.0001999890496974298,
-      "loss": 46.0,
-      "step": 29262
-    },
-    {
-      "epoch": 4.712589073634204,
-      "grad_norm": 0.00216194661334157,
-      "learning_rate": 0.00019998904894874322,
-      "loss": 46.0,
-      "step": 29263
-    },
-    {
-      "epoch": 4.712750110712991,
-      "grad_norm": 0.004811721853911877,
-      "learning_rate": 0.00019998904820003107,
-      "loss": 46.0,
-      "step": 29264
-    },
-    {
-      "epoch": 4.712911147791779,
-      "grad_norm": 0.0019144603284075856,
-      "learning_rate": 0.00019998904745129327,
-      "loss": 46.0,
-      "step": 29265
-    },
-    {
-      "epoch": 4.713072184870566,
-      "grad_norm": 0.01709693856537342,
-      "learning_rate": 0.00019998904670252994,
-      "loss": 46.0,
-      "step": 29266
-    },
-    {
-      "epoch": 4.713233221949354,
-      "grad_norm": 0.0048108892515301704,
-      "learning_rate": 0.000199989045953741,
-      "loss": 46.0,
-      "step": 29267
-    },
-    {
-      "epoch": 4.713394259028141,
-      "grad_norm": 0.0031093235593289137,
-      "learning_rate": 0.00019998904520492644,
-      "loss": 46.0,
-      "step": 29268
-    },
-    {
-      "epoch": 4.713555296106929,
-      "grad_norm": 0.007624708581715822,
-      "learning_rate": 0.00019998904445608633,
-      "loss": 46.0,
-      "step": 29269
-    },
-    {
-      "epoch": 4.713716333185716,
-      "grad_norm": 0.0036514580715447664,
-      "learning_rate": 0.0001999890437072206,
-      "loss": 46.0,
-      "step": 29270
-    },
-    {
-      "epoch": 4.713877370264504,
-      "grad_norm": 0.0028236950747668743,
-      "learning_rate": 0.0001999890429583293,
-      "loss": 46.0,
-      "step": 29271
-    },
-    {
-      "epoch": 4.714038407343291,
-      "grad_norm": 0.000693981593940407,
-      "learning_rate": 0.0001999890422094124,
-      "loss": 46.0,
-      "step": 29272
-    },
-    {
-      "epoch": 4.7141994444220785,
-      "grad_norm": 0.004591328091919422,
-      "learning_rate": 0.00019998904146046988,
-      "loss": 46.0,
-      "step": 29273
-    },
-    {
-      "epoch": 4.714360481500866,
-      "grad_norm": 0.003871242282912135,
-      "learning_rate": 0.0001999890407115018,
-      "loss": 46.0,
-      "step": 29274
-    },
-    {
-      "epoch": 4.714521518579653,
-      "grad_norm": 0.0035130351316183805,
-      "learning_rate": 0.00019998903996250813,
-      "loss": 46.0,
-      "step": 29275
-    },
-    {
-      "epoch": 4.71468255565844,
-      "grad_norm": 0.00098819297272712,
-      "learning_rate": 0.00019998903921348885,
-      "loss": 46.0,
-      "step": 29276
-    },
-    {
-      "epoch": 4.714843592737227,
-      "grad_norm": 0.0017879941733554006,
-      "learning_rate": 0.00019998903846444398,
-      "loss": 46.0,
-      "step": 29277
-    },
-    {
-      "epoch": 4.715004629816015,
-      "grad_norm": 0.005383829586207867,
-      "learning_rate": 0.00019998903771537352,
-      "loss": 46.0,
-      "step": 29278
-    },
-    {
-      "epoch": 4.715165666894802,
-      "grad_norm": 0.002313616219907999,
-      "learning_rate": 0.00019998903696627748,
-      "loss": 46.0,
-      "step": 29279
-    },
-    {
-      "epoch": 4.71532670397359,
-      "grad_norm": 0.002853547688573599,
-      "learning_rate": 0.00019998903621715585,
-      "loss": 46.0,
-      "step": 29280
-    },
-    {
-      "epoch": 4.715487741052377,
-      "grad_norm": 0.004320763982832432,
-      "learning_rate": 0.0001999890354680086,
-      "loss": 46.0,
-      "step": 29281
-    },
-    {
-      "epoch": 4.715648778131165,
-      "grad_norm": 0.0020063386764377356,
-      "learning_rate": 0.00019998903471883577,
-      "loss": 46.0,
-      "step": 29282
-    },
-    {
-      "epoch": 4.715809815209952,
-      "grad_norm": 0.0028386530466377735,
-      "learning_rate": 0.00019998903396963735,
-      "loss": 46.0,
-      "step": 29283
-    },
-    {
-      "epoch": 4.71597085228874,
-      "grad_norm": 0.020207369700074196,
-      "learning_rate": 0.00019998903322041334,
-      "loss": 46.0,
-      "step": 29284
-    },
-    {
-      "epoch": 4.716131889367527,
-      "grad_norm": 0.0034559525083750486,
-      "learning_rate": 0.00019998903247116375,
-      "loss": 46.0,
-      "step": 29285
-    },
-    {
-      "epoch": 4.7162929264463145,
-      "grad_norm": 0.00772892776876688,
-      "learning_rate": 0.00019998903172188854,
-      "loss": 46.0,
-      "step": 29286
-    },
-    {
-      "epoch": 4.716453963525102,
-      "grad_norm": 0.012257229536771774,
-      "learning_rate": 0.00019998903097258777,
-      "loss": 46.0,
-      "step": 29287
-    },
-    {
-      "epoch": 4.716615000603889,
-      "grad_norm": 0.01703144982457161,
-      "learning_rate": 0.0001999890302232614,
-      "loss": 46.0,
-      "step": 29288
-    },
-    {
-      "epoch": 4.716776037682676,
-      "grad_norm": 0.0026686512865126133,
-      "learning_rate": 0.0001999890294739094,
-      "loss": 46.0,
-      "step": 29289
-    },
-    {
-      "epoch": 4.716937074761464,
-      "grad_norm": 0.0013272403739392757,
-      "learning_rate": 0.00019998902872453184,
-      "loss": 46.0,
-      "step": 29290
-    },
-    {
-      "epoch": 4.717098111840251,
-      "grad_norm": 0.0008416443015448749,
-      "learning_rate": 0.0001999890279751287,
-      "loss": 46.0,
-      "step": 29291
-    },
-    {
-      "epoch": 4.717259148919038,
-      "grad_norm": 0.011600710451602936,
-      "learning_rate": 0.00019998902722569994,
-      "loss": 46.0,
-      "step": 29292
-    },
-    {
-      "epoch": 4.717420185997826,
-      "grad_norm": 0.003983397968113422,
-      "learning_rate": 0.00019998902647624562,
-      "loss": 46.0,
-      "step": 29293
-    },
-    {
-      "epoch": 4.717581223076613,
-      "grad_norm": 0.00986267626285553,
-      "learning_rate": 0.0001999890257267657,
-      "loss": 46.0,
-      "step": 29294
-    },
-    {
-      "epoch": 4.717742260155401,
-      "grad_norm": 0.0030516362749040127,
-      "learning_rate": 0.00019998902497726014,
-      "loss": 46.0,
-      "step": 29295
-    },
-    {
-      "epoch": 4.717903297234188,
-      "grad_norm": 0.0014918227680027485,
-      "learning_rate": 0.00019998902422772906,
-      "loss": 46.0,
-      "step": 29296
-    },
-    {
-      "epoch": 4.718064334312976,
-      "grad_norm": 0.003218598896637559,
-      "learning_rate": 0.0001999890234781723,
-      "loss": 46.0,
-      "step": 29297
-    },
-    {
-      "epoch": 4.718225371391763,
-      "grad_norm": 0.005359306000173092,
-      "learning_rate": 0.00019998902272859003,
-      "loss": 46.0,
-      "step": 29298
-    },
-    {
-      "epoch": 4.7183864084705505,
-      "grad_norm": 0.0026564777363091707,
-      "learning_rate": 0.00019998902197898214,
-      "loss": 46.0,
-      "step": 29299
-    },
-    {
-      "epoch": 4.718547445549338,
-      "grad_norm": 0.0033268441911786795,
-      "learning_rate": 0.00019998902122934865,
-      "loss": 46.0,
-      "step": 29300
-    },
-    {
-      "epoch": 4.718708482628125,
-      "grad_norm": 0.003321029245853424,
-      "learning_rate": 0.00019998902047968956,
-      "loss": 46.0,
-      "step": 29301
-    },
-    {
-      "epoch": 4.718869519706913,
-      "grad_norm": 0.004557216074317694,
-      "learning_rate": 0.0001999890197300049,
-      "loss": 46.0,
-      "step": 29302
-    },
-    {
-      "epoch": 4.7190305567857,
-      "grad_norm": 0.005502277985215187,
-      "learning_rate": 0.00019998901898029466,
-      "loss": 46.0,
-      "step": 29303
-    },
-    {
-      "epoch": 4.719191593864487,
-      "grad_norm": 0.0010434914147481322,
-      "learning_rate": 0.00019998901823055883,
-      "loss": 46.0,
-      "step": 29304
-    },
-    {
-      "epoch": 4.719352630943275,
-      "grad_norm": 0.01131720095872879,
-      "learning_rate": 0.00019998901748079738,
-      "loss": 46.0,
-      "step": 29305
-    },
-    {
-      "epoch": 4.719513668022062,
-      "grad_norm": 0.003906423691660166,
-      "learning_rate": 0.00019998901673101035,
-      "loss": 46.0,
-      "step": 29306
-    },
-    {
-      "epoch": 4.719674705100849,
-      "grad_norm": 0.005399438086897135,
-      "learning_rate": 0.0001999890159811977,
-      "loss": 46.0,
-      "step": 29307
-    },
-    {
-      "epoch": 4.719835742179637,
-      "grad_norm": 0.003501476952806115,
-      "learning_rate": 0.0001999890152313595,
-      "loss": 46.0,
-      "step": 29308
-    },
-    {
-      "epoch": 4.719996779258424,
-      "grad_norm": 0.002397339791059494,
-      "learning_rate": 0.00019998901448149567,
-      "loss": 46.0,
-      "step": 29309
-    },
-    {
-      "epoch": 4.7201578163372115,
-      "grad_norm": 0.011559433303773403,
-      "learning_rate": 0.0001999890137316063,
-      "loss": 46.0,
-      "step": 29310
-    },
-    {
-      "epoch": 4.720318853415999,
-      "grad_norm": 0.004457653500139713,
-      "learning_rate": 0.00019998901298169127,
-      "loss": 46.0,
-      "step": 29311
-    },
-    {
-      "epoch": 4.7204798904947864,
-      "grad_norm": 0.002904606517404318,
-      "learning_rate": 0.00019998901223175069,
-      "loss": 46.0,
-      "step": 29312
-    },
-    {
-      "epoch": 4.720640927573574,
-      "grad_norm": 0.003907191101461649,
-      "learning_rate": 0.00019998901148178452,
-      "loss": 46.0,
-      "step": 29313
-    },
-    {
-      "epoch": 4.720801964652361,
-      "grad_norm": 0.0123291639611125,
-      "learning_rate": 0.00019998901073179273,
-      "loss": 46.0,
-      "step": 29314
-    },
-    {
-      "epoch": 4.720963001731149,
-      "grad_norm": 0.0017887142021209002,
-      "learning_rate": 0.00019998900998177536,
-      "loss": 46.0,
-      "step": 29315
-    },
-    {
-      "epoch": 4.721124038809936,
-      "grad_norm": 0.00621689623221755,
-      "learning_rate": 0.0001999890092317324,
-      "loss": 46.0,
-      "step": 29316
-    },
-    {
-      "epoch": 4.721285075888724,
-      "grad_norm": 0.0029398349579423666,
-      "learning_rate": 0.00019998900848166386,
-      "loss": 46.0,
-      "step": 29317
-    },
-    {
-      "epoch": 4.721446112967511,
-      "grad_norm": 0.0005799714126624167,
-      "learning_rate": 0.00019998900773156972,
-      "loss": 46.0,
-      "step": 29318
-    },
-    {
-      "epoch": 4.721607150046298,
-      "grad_norm": 0.0037565052043646574,
-      "learning_rate": 0.00019998900698144998,
-      "loss": 46.0,
-      "step": 29319
-    },
-    {
-      "epoch": 4.721768187125085,
-      "grad_norm": 0.004496282432228327,
-      "learning_rate": 0.00019998900623130464,
-      "loss": 46.0,
-      "step": 29320
-    },
-    {
-      "epoch": 4.721929224203873,
-      "grad_norm": 0.003233895171433687,
-      "learning_rate": 0.00019998900548113375,
-      "loss": 46.0,
-      "step": 29321
-    },
-    {
-      "epoch": 4.72209026128266,
-      "grad_norm": 0.003190218936651945,
-      "learning_rate": 0.0001999890047309372,
-      "loss": 46.0,
-      "step": 29322
-    },
-    {
-      "epoch": 4.7222512983614475,
-      "grad_norm": 0.005417706910520792,
-      "learning_rate": 0.00019998900398071514,
-      "loss": 46.0,
-      "step": 29323
-    },
-    {
-      "epoch": 4.722412335440235,
-      "grad_norm": 0.0026388815604150295,
-      "learning_rate": 0.00019998900323046743,
-      "loss": 46.0,
-      "step": 29324
-    },
-    {
-      "epoch": 4.722573372519022,
-      "grad_norm": 0.002630612114444375,
-      "learning_rate": 0.00019998900248019416,
-      "loss": 46.0,
-      "step": 29325
-    },
-    {
-      "epoch": 4.72273440959781,
-      "grad_norm": 0.0053198328241705894,
-      "learning_rate": 0.00019998900172989528,
-      "loss": 46.0,
-      "step": 29326
-    },
-    {
-      "epoch": 4.722895446676597,
-      "grad_norm": 0.002506533870473504,
-      "learning_rate": 0.0001999890009795708,
-      "loss": 46.0,
-      "step": 29327
-    },
-    {
-      "epoch": 4.723056483755385,
-      "grad_norm": 0.003889096202328801,
-      "learning_rate": 0.00019998900022922075,
-      "loss": 46.0,
-      "step": 29328
-    },
-    {
-      "epoch": 4.723217520834172,
-      "grad_norm": 0.0018801233964040875,
-      "learning_rate": 0.0001999889994788451,
-      "loss": 46.0,
-      "step": 29329
-    },
-    {
-      "epoch": 4.72337855791296,
-      "grad_norm": 0.010475135408341885,
-      "learning_rate": 0.00019998899872844384,
-      "loss": 46.0,
-      "step": 29330
-    },
-    {
-      "epoch": 4.723539594991747,
-      "grad_norm": 0.0016895242733880877,
-      "learning_rate": 0.00019998899797801702,
-      "loss": 46.0,
-      "step": 29331
-    },
-    {
-      "epoch": 4.723700632070535,
-      "grad_norm": 0.0015805610455572605,
-      "learning_rate": 0.0001999889972275646,
-      "loss": 46.0,
-      "step": 29332
-    },
-    {
-      "epoch": 4.723861669149322,
-      "grad_norm": 0.001714044134132564,
-      "learning_rate": 0.00019998899647708657,
-      "loss": 46.0,
-      "step": 29333
-    },
-    {
-      "epoch": 4.724022706228109,
-      "grad_norm": 0.00274398154579103,
-      "learning_rate": 0.00019998899572658296,
-      "loss": 46.0,
-      "step": 29334
-    },
-    {
-      "epoch": 4.724183743306896,
-      "grad_norm": 0.004670158494263887,
-      "learning_rate": 0.00019998899497605373,
-      "loss": 46.0,
-      "step": 29335
-    },
-    {
-      "epoch": 4.7243447803856835,
-      "grad_norm": 0.00502600846812129,
-      "learning_rate": 0.00019998899422549898,
-      "loss": 46.0,
-      "step": 29336
-    },
-    {
-      "epoch": 4.724505817464471,
-      "grad_norm": 0.004184090998023748,
-      "learning_rate": 0.00019998899347491858,
-      "loss": 46.0,
-      "step": 29337
-    },
-    {
-      "epoch": 4.724666854543258,
-      "grad_norm": 0.0057343426160514355,
-      "learning_rate": 0.0001999889927243126,
-      "loss": 46.0,
-      "step": 29338
-    },
-    {
-      "epoch": 4.724827891622046,
-      "grad_norm": 0.014873027801513672,
-      "learning_rate": 0.00019998899197368102,
-      "loss": 46.0,
-      "step": 29339
-    },
-    {
-      "epoch": 4.724988928700833,
-      "grad_norm": 0.0017248686635866761,
-      "learning_rate": 0.00019998899122302386,
-      "loss": 46.0,
-      "step": 29340
-    },
-    {
-      "epoch": 4.725149965779621,
-      "grad_norm": 0.008649676106870174,
-      "learning_rate": 0.0001999889904723411,
-      "loss": 46.0,
-      "step": 29341
-    },
-    {
-      "epoch": 4.725311002858408,
-      "grad_norm": 0.008922945708036423,
-      "learning_rate": 0.00019998898972163276,
-      "loss": 46.0,
-      "step": 29342
-    },
-    {
-      "epoch": 4.725472039937196,
-      "grad_norm": 0.004946904256939888,
-      "learning_rate": 0.00019998898897089878,
-      "loss": 46.0,
-      "step": 29343
-    },
-    {
-      "epoch": 4.725633077015983,
-      "grad_norm": 0.0019709193147718906,
-      "learning_rate": 0.00019998898822013925,
-      "loss": 46.0,
-      "step": 29344
-    },
-    {
-      "epoch": 4.725794114094771,
-      "grad_norm": 0.005183913744986057,
-      "learning_rate": 0.00019998898746935413,
-      "loss": 46.0,
-      "step": 29345
-    },
-    {
-      "epoch": 4.725955151173558,
-      "grad_norm": 0.01176372542977333,
-      "learning_rate": 0.00019998898671854342,
-      "loss": 46.0,
-      "step": 29346
-    },
-    {
-      "epoch": 4.7261161882523455,
-      "grad_norm": 0.0020438064821064472,
-      "learning_rate": 0.0001999889859677071,
-      "loss": 46.0,
-      "step": 29347
-    },
-    {
-      "epoch": 4.726277225331133,
-      "grad_norm": 0.0004885127418674529,
-      "learning_rate": 0.0001999889852168452,
-      "loss": 46.0,
-      "step": 29348
-    },
-    {
-      "epoch": 4.7264382624099195,
-      "grad_norm": 0.010794468224048615,
-      "learning_rate": 0.0001999889844659577,
-      "loss": 46.0,
-      "step": 29349
-    },
-    {
-      "epoch": 4.726599299488707,
-      "grad_norm": 0.01155570987612009,
-      "learning_rate": 0.00019998898371504465,
-      "loss": 46.0,
-      "step": 29350
-    },
-    {
-      "epoch": 4.726760336567494,
-      "grad_norm": 0.003462859196588397,
-      "learning_rate": 0.00019998898296410595,
-      "loss": 46.0,
-      "step": 29351
-    },
-    {
-      "epoch": 4.726921373646282,
-      "grad_norm": 0.0027673481963574886,
-      "learning_rate": 0.0001999889822131417,
-      "loss": 46.0,
-      "step": 29352
-    },
-    {
-      "epoch": 4.727082410725069,
-      "grad_norm": 0.002472318010404706,
-      "learning_rate": 0.00019998898146215185,
-      "loss": 46.0,
-      "step": 29353
-    },
-    {
-      "epoch": 4.727243447803857,
-      "grad_norm": 0.0022224565036594868,
-      "learning_rate": 0.00019998898071113636,
-      "loss": 46.0,
-      "step": 29354
-    },
-    {
-      "epoch": 4.727404484882644,
-      "grad_norm": 0.0022227480076253414,
-      "learning_rate": 0.0001999889799600953,
-      "loss": 46.0,
-      "step": 29355
-    },
-    {
-      "epoch": 4.727565521961432,
-      "grad_norm": 0.0015774141065776348,
-      "learning_rate": 0.00019998897920902868,
-      "loss": 46.0,
-      "step": 29356
-    },
-    {
-      "epoch": 4.727726559040219,
-      "grad_norm": 0.005216037388890982,
-      "learning_rate": 0.00019998897845793646,
-      "loss": 46.0,
-      "step": 29357
-    },
-    {
-      "epoch": 4.7278875961190066,
-      "grad_norm": 0.0028596031479537487,
-      "learning_rate": 0.00019998897770681865,
-      "loss": 46.0,
-      "step": 29358
-    },
-    {
-      "epoch": 4.728048633197794,
-      "grad_norm": 0.0021019771229475737,
-      "learning_rate": 0.00019998897695567522,
-      "loss": 46.0,
-      "step": 29359
-    },
-    {
-      "epoch": 4.7282096702765815,
-      "grad_norm": 0.00817108154296875,
-      "learning_rate": 0.0001999889762045062,
-      "loss": 46.0,
-      "step": 29360
-    },
-    {
-      "epoch": 4.728370707355369,
-      "grad_norm": 0.004792229272425175,
-      "learning_rate": 0.00019998897545331161,
-      "loss": 46.0,
-      "step": 29361
-    },
-    {
-      "epoch": 4.7285317444341555,
-      "grad_norm": 0.0008420338854193687,
-      "learning_rate": 0.0001999889747020914,
-      "loss": 46.0,
-      "step": 29362
-    },
-    {
-      "epoch": 4.728692781512944,
-      "grad_norm": 0.015027117915451527,
-      "learning_rate": 0.00019998897395084563,
-      "loss": 46.0,
-      "step": 29363
-    },
-    {
-      "epoch": 4.72885381859173,
-      "grad_norm": 0.0037622731178998947,
-      "learning_rate": 0.00019998897319957427,
-      "loss": 46.0,
-      "step": 29364
-    },
-    {
-      "epoch": 4.729014855670518,
-      "grad_norm": 0.0054428973235189915,
-      "learning_rate": 0.0001999889724482773,
-      "loss": 46.0,
-      "step": 29365
-    },
-    {
-      "epoch": 4.729175892749305,
-      "grad_norm": 0.0036208373494446278,
-      "learning_rate": 0.00019998897169695473,
-      "loss": 46.0,
-      "step": 29366
-    },
-    {
-      "epoch": 4.729336929828093,
-      "grad_norm": 0.0008238909067586064,
-      "learning_rate": 0.0001999889709456066,
-      "loss": 46.0,
-      "step": 29367
-    },
-    {
-      "epoch": 4.72949796690688,
-      "grad_norm": 0.0014410164440050721,
-      "learning_rate": 0.00019998897019423285,
-      "loss": 46.0,
-      "step": 29368
-    },
-    {
-      "epoch": 4.729659003985668,
-      "grad_norm": 0.0032045100815594196,
-      "learning_rate": 0.00019998896944283353,
-      "loss": 46.0,
-      "step": 29369
-    },
-    {
-      "epoch": 4.729820041064455,
-      "grad_norm": 0.0012403222499415278,
-      "learning_rate": 0.00019998896869140856,
-      "loss": 46.0,
-      "step": 29370
-    },
-    {
-      "epoch": 4.7299810781432425,
-      "grad_norm": 0.011964606121182442,
-      "learning_rate": 0.00019998896793995807,
-      "loss": 46.0,
-      "step": 29371
-    },
-    {
-      "epoch": 4.73014211522203,
-      "grad_norm": 0.0021035645622760057,
-      "learning_rate": 0.00019998896718848196,
-      "loss": 46.0,
-      "step": 29372
-    },
-    {
-      "epoch": 4.730303152300817,
-      "grad_norm": 0.0036605617497116327,
-      "learning_rate": 0.00019998896643698023,
-      "loss": 46.0,
-      "step": 29373
-    },
-    {
-      "epoch": 4.730464189379605,
-      "grad_norm": 0.020130915567278862,
-      "learning_rate": 0.00019998896568545295,
-      "loss": 46.0,
-      "step": 29374
-    },
-    {
-      "epoch": 4.730625226458392,
-      "grad_norm": 0.001779328566044569,
-      "learning_rate": 0.00019998896493390007,
-      "loss": 46.0,
-      "step": 29375
-    },
-    {
-      "epoch": 4.73078626353718,
-      "grad_norm": 0.002841960871592164,
-      "learning_rate": 0.0001999889641823216,
-      "loss": 46.0,
-      "step": 29376
-    },
-    {
-      "epoch": 4.730947300615966,
-      "grad_norm": 0.003420154331251979,
-      "learning_rate": 0.00019998896343071754,
-      "loss": 46.0,
-      "step": 29377
-    },
-    {
-      "epoch": 4.731108337694755,
-      "grad_norm": 0.0027154451236128807,
-      "learning_rate": 0.00019998896267908785,
-      "loss": 46.0,
-      "step": 29378
-    },
-    {
-      "epoch": 4.731269374773541,
-      "grad_norm": 0.0006392297800630331,
-      "learning_rate": 0.00019998896192743263,
-      "loss": 46.0,
-      "step": 29379
-    },
-    {
-      "epoch": 4.731430411852329,
-      "grad_norm": 0.003128868294879794,
-      "learning_rate": 0.00019998896117575177,
-      "loss": 46.0,
-      "step": 29380
-    },
-    {
-      "epoch": 4.731591448931116,
-      "grad_norm": 0.002107715466991067,
-      "learning_rate": 0.00019998896042404532,
-      "loss": 46.0,
-      "step": 29381
-    },
-    {
-      "epoch": 4.731752486009904,
-      "grad_norm": 0.002084210980683565,
-      "learning_rate": 0.00019998895967231328,
-      "loss": 46.0,
-      "step": 29382
-    },
-    {
-      "epoch": 4.731913523088691,
-      "grad_norm": 0.005268099717795849,
-      "learning_rate": 0.00019998895892055566,
-      "loss": 46.0,
-      "step": 29383
-    },
-    {
-      "epoch": 4.7320745601674785,
-      "grad_norm": 0.011741421185433865,
-      "learning_rate": 0.00019998895816877244,
-      "loss": 46.0,
-      "step": 29384
-    },
-    {
-      "epoch": 4.732235597246266,
-      "grad_norm": 0.007093553431332111,
-      "learning_rate": 0.00019998895741696365,
-      "loss": 46.0,
-      "step": 29385
-    },
-    {
-      "epoch": 4.732396634325053,
-      "grad_norm": 0.002017336431890726,
-      "learning_rate": 0.00019998895666512923,
-      "loss": 46.0,
-      "step": 29386
-    },
-    {
-      "epoch": 4.732557671403841,
-      "grad_norm": 0.009207571856677532,
-      "learning_rate": 0.00019998895591326926,
-      "loss": 46.0,
-      "step": 29387
-    },
-    {
-      "epoch": 4.732718708482628,
-      "grad_norm": 0.0025740114506334066,
-      "learning_rate": 0.00019998895516138365,
-      "loss": 46.0,
-      "step": 29388
-    },
-    {
-      "epoch": 4.732879745561416,
-      "grad_norm": 0.0072769238613545895,
-      "learning_rate": 0.0001999889544094725,
-      "loss": 46.0,
-      "step": 29389
-    },
-    {
-      "epoch": 4.733040782640203,
-      "grad_norm": 0.004357141442596912,
-      "learning_rate": 0.0001999889536575357,
-      "loss": 46.0,
-      "step": 29390
-    },
-    {
-      "epoch": 4.733201819718991,
-      "grad_norm": 0.005643466021865606,
-      "learning_rate": 0.00019998895290557336,
-      "loss": 46.0,
-      "step": 29391
-    },
-    {
-      "epoch": 4.733362856797777,
-      "grad_norm": 0.003422433976083994,
-      "learning_rate": 0.00019998895215358542,
-      "loss": 46.0,
-      "step": 29392
-    },
-    {
-      "epoch": 4.733523893876565,
-      "grad_norm": 0.008912176825106144,
-      "learning_rate": 0.00019998895140157185,
-      "loss": 46.0,
-      "step": 29393
-    },
-    {
-      "epoch": 4.733684930955352,
-      "grad_norm": 0.002388990018516779,
-      "learning_rate": 0.0001999889506495327,
-      "loss": 46.0,
-      "step": 29394
-    },
-    {
-      "epoch": 4.73384596803414,
-      "grad_norm": 0.0008180595468729734,
-      "learning_rate": 0.00019998894989746799,
-      "loss": 46.0,
-      "step": 29395
-    },
-    {
-      "epoch": 4.734007005112927,
-      "grad_norm": 0.005835089832544327,
-      "learning_rate": 0.0001999889491453777,
-      "loss": 46.0,
-      "step": 29396
-    },
-    {
-      "epoch": 4.7341680421917145,
-      "grad_norm": 0.003755220677703619,
-      "learning_rate": 0.00019998894839326177,
-      "loss": 46.0,
-      "step": 29397
-    },
-    {
-      "epoch": 4.734329079270502,
-      "grad_norm": 0.003420136636123061,
-      "learning_rate": 0.00019998894764112026,
-      "loss": 46.0,
-      "step": 29398
-    },
-    {
-      "epoch": 4.734490116349289,
-      "grad_norm": 0.016514291986823082,
-      "learning_rate": 0.0001999889468889532,
-      "loss": 46.0,
-      "step": 29399
-    },
-    {
-      "epoch": 4.734651153428077,
-      "grad_norm": 0.0060422117821872234,
-      "learning_rate": 0.00019998894613676047,
-      "loss": 46.0,
-      "step": 29400
-    },
-    {
-      "epoch": 4.734812190506864,
-      "grad_norm": 0.009305658750236034,
-      "learning_rate": 0.0001999889453845422,
-      "loss": 46.0,
-      "step": 29401
-    },
-    {
-      "epoch": 4.734973227585652,
-      "grad_norm": 0.004312990698963404,
-      "learning_rate": 0.0001999889446322983,
-      "loss": 46.0,
-      "step": 29402
-    },
-    {
-      "epoch": 4.735134264664439,
-      "grad_norm": 0.0009957398287951946,
-      "learning_rate": 0.00019998894388002886,
-      "loss": 46.0,
-      "step": 29403
-    },
-    {
-      "epoch": 4.735295301743227,
-      "grad_norm": 0.009375032968819141,
-      "learning_rate": 0.0001999889431277338,
-      "loss": 46.0,
-      "step": 29404
-    },
-    {
-      "epoch": 4.735456338822014,
-      "grad_norm": 0.003856866853311658,
-      "learning_rate": 0.00019998894237541315,
-      "loss": 46.0,
-      "step": 29405
-    },
-    {
-      "epoch": 4.735617375900802,
-      "grad_norm": 0.001658300287090242,
-      "learning_rate": 0.0001999889416230669,
-      "loss": 46.0,
-      "step": 29406
-    },
-    {
-      "epoch": 4.735778412979588,
-      "grad_norm": 0.0038576764054596424,
-      "learning_rate": 0.00019998894087069509,
-      "loss": 46.0,
-      "step": 29407
-    },
-    {
-      "epoch": 4.735939450058376,
-      "grad_norm": 0.004734646063297987,
-      "learning_rate": 0.00019998894011829767,
-      "loss": 46.0,
-      "step": 29408
-    },
-    {
-      "epoch": 4.736100487137163,
-      "grad_norm": 0.002241219161078334,
-      "learning_rate": 0.00019998893936587467,
-      "loss": 46.0,
-      "step": 29409
-    },
-    {
-      "epoch": 4.7362615242159505,
-      "grad_norm": 0.0013894643634557724,
-      "learning_rate": 0.00019998893861342603,
-      "loss": 46.0,
-      "step": 29410
-    },
-    {
-      "epoch": 4.736422561294738,
-      "grad_norm": 0.010313589125871658,
-      "learning_rate": 0.00019998893786095186,
-      "loss": 46.0,
-      "step": 29411
-    },
-    {
-      "epoch": 4.736583598373525,
-      "grad_norm": 0.002238770015537739,
-      "learning_rate": 0.00019998893710845205,
-      "loss": 46.0,
-      "step": 29412
-    },
-    {
-      "epoch": 4.736744635452313,
-      "grad_norm": 0.004633741453289986,
-      "learning_rate": 0.00019998893635592667,
-      "loss": 46.0,
-      "step": 29413
-    },
-    {
-      "epoch": 4.7369056725311,
-      "grad_norm": 0.01556902565062046,
-      "learning_rate": 0.00019998893560337568,
-      "loss": 46.0,
-      "step": 29414
-    },
-    {
-      "epoch": 4.737066709609888,
-      "grad_norm": 0.028797954320907593,
-      "learning_rate": 0.0001999889348507991,
-      "loss": 46.0,
-      "step": 29415
-    },
-    {
-      "epoch": 4.737227746688675,
-      "grad_norm": 0.004877353087067604,
-      "learning_rate": 0.00019998893409819694,
-      "loss": 46.0,
-      "step": 29416
-    },
-    {
-      "epoch": 4.737388783767463,
-      "grad_norm": 0.0019912421703338623,
-      "learning_rate": 0.00019998893334556922,
-      "loss": 46.0,
-      "step": 29417
-    },
-    {
-      "epoch": 4.73754982084625,
-      "grad_norm": 0.007213990204036236,
-      "learning_rate": 0.00019998893259291585,
-      "loss": 46.0,
-      "step": 29418
-    },
-    {
-      "epoch": 4.7377108579250375,
-      "grad_norm": 0.0038081426173448563,
-      "learning_rate": 0.00019998893184023693,
-      "loss": 46.0,
-      "step": 29419
-    },
-    {
-      "epoch": 4.737871895003825,
-      "grad_norm": 0.0012381550623103976,
-      "learning_rate": 0.0001999889310875324,
-      "loss": 46.0,
-      "step": 29420
-    },
-    {
-      "epoch": 4.7380329320826124,
-      "grad_norm": 0.005331635940819979,
-      "learning_rate": 0.0001999889303348023,
-      "loss": 46.0,
-      "step": 29421
-    },
-    {
-      "epoch": 4.738193969161399,
-      "grad_norm": 0.0043566869571805,
-      "learning_rate": 0.00019998892958204655,
-      "loss": 46.0,
-      "step": 29422
-    },
-    {
-      "epoch": 4.7383550062401865,
-      "grad_norm": 0.002686574589461088,
-      "learning_rate": 0.00019998892882926525,
-      "loss": 46.0,
-      "step": 29423
-    },
-    {
-      "epoch": 4.738516043318974,
-      "grad_norm": 0.011207504197955132,
-      "learning_rate": 0.00019998892807645836,
-      "loss": 46.0,
-      "step": 29424
-    },
-    {
-      "epoch": 4.738677080397761,
-      "grad_norm": 0.003423671703785658,
-      "learning_rate": 0.00019998892732362586,
-      "loss": 46.0,
-      "step": 29425
-    },
-    {
-      "epoch": 4.738838117476549,
-      "grad_norm": 0.0017669046064838767,
-      "learning_rate": 0.00019998892657076777,
-      "loss": 46.0,
-      "step": 29426
-    },
-    {
-      "epoch": 4.738999154555336,
-      "grad_norm": 0.004224743228405714,
-      "learning_rate": 0.00019998892581788412,
-      "loss": 46.0,
-      "step": 29427
-    },
-    {
-      "epoch": 4.739160191634124,
-      "grad_norm": 0.0030419223476201296,
-      "learning_rate": 0.00019998892506497485,
-      "loss": 46.0,
-      "step": 29428
-    },
-    {
-      "epoch": 4.739321228712911,
-      "grad_norm": 0.0019750739447772503,
-      "learning_rate": 0.00019998892431204,
-      "loss": 46.0,
-      "step": 29429
-    },
-    {
-      "epoch": 4.739482265791699,
-      "grad_norm": 0.0037552407011389732,
-      "learning_rate": 0.00019998892355907954,
-      "loss": 46.0,
-      "step": 29430
-    },
-    {
-      "epoch": 4.739643302870486,
-      "grad_norm": 0.01344646792858839,
-      "learning_rate": 0.0001999889228060935,
-      "loss": 46.0,
-      "step": 29431
-    },
-    {
-      "epoch": 4.7398043399492735,
-      "grad_norm": 0.010561018250882626,
-      "learning_rate": 0.00019998892205308187,
-      "loss": 46.0,
-      "step": 29432
-    },
-    {
-      "epoch": 4.739965377028061,
-      "grad_norm": 0.001239353441633284,
-      "learning_rate": 0.00019998892130004462,
-      "loss": 46.0,
-      "step": 29433
-    },
-    {
-      "epoch": 4.740126414106848,
-      "grad_norm": 0.004819825757294893,
-      "learning_rate": 0.0001999889205469818,
-      "loss": 46.0,
-      "step": 29434
-    },
-    {
-      "epoch": 4.740287451185636,
-      "grad_norm": 0.008641612716019154,
-      "learning_rate": 0.00019998891979389343,
-      "loss": 46.0,
-      "step": 29435
-    },
-    {
-      "epoch": 4.740448488264423,
-      "grad_norm": 0.002989055821672082,
-      "learning_rate": 0.00019998891904077942,
-      "loss": 46.0,
-      "step": 29436
-    },
-    {
-      "epoch": 4.74060952534321,
-      "grad_norm": 0.0026798087637871504,
-      "learning_rate": 0.00019998891828763981,
-      "loss": 46.0,
-      "step": 29437
-    },
-    {
-      "epoch": 4.740770562421997,
-      "grad_norm": 0.0035659109707921743,
-      "learning_rate": 0.00019998891753447465,
-      "loss": 46.0,
-      "step": 29438
-    },
-    {
-      "epoch": 4.740931599500785,
-      "grad_norm": 0.0047483728267252445,
-      "learning_rate": 0.00019998891678128387,
-      "loss": 46.0,
-      "step": 29439
-    },
-    {
-      "epoch": 4.741092636579572,
-      "grad_norm": 0.002634359523653984,
-      "learning_rate": 0.00019998891602806748,
-      "loss": 46.0,
-      "step": 29440
-    },
-    {
-      "epoch": 4.74125367365836,
-      "grad_norm": 0.005374438129365444,
-      "learning_rate": 0.00019998891527482553,
-      "loss": 46.0,
-      "step": 29441
-    },
-    {
-      "epoch": 4.741414710737147,
-      "grad_norm": 0.00490146316587925,
-      "learning_rate": 0.00019998891452155797,
-      "loss": 46.0,
-      "step": 29442
-    },
-    {
-      "epoch": 4.741575747815935,
-      "grad_norm": 0.001773351919837296,
-      "learning_rate": 0.00019998891376826484,
-      "loss": 46.0,
-      "step": 29443
-    },
-    {
-      "epoch": 4.741736784894722,
-      "grad_norm": 0.009983288124203682,
-      "learning_rate": 0.0001999889130149461,
-      "loss": 46.0,
-      "step": 29444
-    },
-    {
-      "epoch": 4.7418978219735095,
-      "grad_norm": 0.005572212394326925,
-      "learning_rate": 0.00019998891226160175,
-      "loss": 46.0,
-      "step": 29445
-    },
-    {
-      "epoch": 4.742058859052297,
-      "grad_norm": 0.007713311351835728,
-      "learning_rate": 0.00019998891150823186,
-      "loss": 46.0,
-      "step": 29446
-    },
-    {
-      "epoch": 4.742219896131084,
-      "grad_norm": 0.002674366580322385,
-      "learning_rate": 0.00019998891075483633,
-      "loss": 46.0,
-      "step": 29447
-    },
-    {
-      "epoch": 4.742380933209872,
-      "grad_norm": 0.008505702018737793,
-      "learning_rate": 0.00019998891000141522,
-      "loss": 46.0,
-      "step": 29448
-    },
-    {
-      "epoch": 4.742541970288659,
-      "grad_norm": 0.004731995053589344,
-      "learning_rate": 0.00019998890924796854,
-      "loss": 46.0,
-      "step": 29449
-    },
-    {
-      "epoch": 4.742703007367446,
-      "grad_norm": 0.007525719236582518,
-      "learning_rate": 0.00019998890849449622,
-      "loss": 46.0,
-      "step": 29450
-    },
-    {
-      "epoch": 4.742864044446234,
-      "grad_norm": 0.006710035726428032,
-      "learning_rate": 0.00019998890774099835,
-      "loss": 46.0,
-      "step": 29451
-    },
-    {
-      "epoch": 4.743025081525021,
-      "grad_norm": 0.0023106341250240803,
-      "learning_rate": 0.00019998890698747488,
-      "loss": 46.0,
-      "step": 29452
-    },
-    {
-      "epoch": 4.743186118603808,
-      "grad_norm": 0.010832172818481922,
-      "learning_rate": 0.0001999889062339258,
-      "loss": 46.0,
-      "step": 29453
-    },
-    {
-      "epoch": 4.743347155682596,
-      "grad_norm": 0.001380281406454742,
-      "learning_rate": 0.00019998890548035116,
-      "loss": 46.0,
-      "step": 29454
-    },
-    {
-      "epoch": 4.743508192761383,
-      "grad_norm": 0.007656737230718136,
-      "learning_rate": 0.0001999889047267509,
-      "loss": 46.0,
-      "step": 29455
-    },
-    {
-      "epoch": 4.743669229840171,
-      "grad_norm": 0.0022427341900765896,
-      "learning_rate": 0.00019998890397312507,
-      "loss": 46.0,
-      "step": 29456
-    },
-    {
-      "epoch": 4.743830266918958,
-      "grad_norm": 0.005935526918619871,
-      "learning_rate": 0.00019998890321947364,
-      "loss": 46.0,
-      "step": 29457
-    },
-    {
-      "epoch": 4.7439913039977455,
-      "grad_norm": 0.007354166824370623,
-      "learning_rate": 0.0001999889024657966,
-      "loss": 46.0,
-      "step": 29458
-    },
-    {
-      "epoch": 4.744152341076533,
-      "grad_norm": 0.0018673599697649479,
-      "learning_rate": 0.000199988901712094,
-      "loss": 46.0,
-      "step": 29459
-    },
-    {
-      "epoch": 4.74431337815532,
-      "grad_norm": 0.003711445489898324,
-      "learning_rate": 0.00019998890095836578,
-      "loss": 46.0,
-      "step": 29460
-    },
-    {
-      "epoch": 4.744474415234108,
-      "grad_norm": 0.00214279699139297,
-      "learning_rate": 0.00019998890020461198,
-      "loss": 46.0,
-      "step": 29461
-    },
-    {
-      "epoch": 4.744635452312895,
-      "grad_norm": 0.006456930655986071,
-      "learning_rate": 0.0001999888994508326,
-      "loss": 46.0,
-      "step": 29462
-    },
-    {
-      "epoch": 4.744796489391683,
-      "grad_norm": 0.0016828099032863975,
-      "learning_rate": 0.0001999888986970276,
-      "loss": 46.0,
-      "step": 29463
-    },
-    {
-      "epoch": 4.74495752647047,
-      "grad_norm": 0.006650710012763739,
-      "learning_rate": 0.00019998889794319702,
-      "loss": 46.0,
-      "step": 29464
-    },
-    {
-      "epoch": 4.745118563549257,
-      "grad_norm": 0.0035515003837645054,
-      "learning_rate": 0.00019998889718934084,
-      "loss": 46.0,
-      "step": 29465
-    },
-    {
-      "epoch": 4.745279600628045,
-      "grad_norm": 0.0035023754462599754,
-      "learning_rate": 0.00019998889643545908,
-      "loss": 46.0,
-      "step": 29466
-    },
-    {
-      "epoch": 4.745440637706832,
-      "grad_norm": 0.0028557120822370052,
-      "learning_rate": 0.00019998889568155172,
-      "loss": 46.0,
-      "step": 29467
-    },
-    {
-      "epoch": 4.745601674785619,
-      "grad_norm": 0.01697532832622528,
-      "learning_rate": 0.00019998889492761878,
-      "loss": 46.0,
-      "step": 29468
-    },
-    {
-      "epoch": 4.745762711864407,
-      "grad_norm": 0.01211035717278719,
-      "learning_rate": 0.00019998889417366026,
-      "loss": 46.0,
-      "step": 29469
-    },
-    {
-      "epoch": 4.745923748943194,
-      "grad_norm": 0.005862020421773195,
-      "learning_rate": 0.00019998889341967611,
-      "loss": 46.0,
-      "step": 29470
-    },
-    {
-      "epoch": 4.7460847860219815,
-      "grad_norm": 0.007999500259757042,
-      "learning_rate": 0.00019998889266566638,
-      "loss": 46.0,
-      "step": 29471
-    },
-    {
-      "epoch": 4.746245823100769,
-      "grad_norm": 0.0044660987332463264,
-      "learning_rate": 0.00019998889191163107,
-      "loss": 46.0,
-      "step": 29472
-    },
-    {
-      "epoch": 4.746406860179556,
-      "grad_norm": 0.010455731302499771,
-      "learning_rate": 0.00019998889115757017,
-      "loss": 46.0,
-      "step": 29473
-    },
-    {
-      "epoch": 4.746567897258344,
-      "grad_norm": 0.004530957899987698,
-      "learning_rate": 0.00019998889040348367,
-      "loss": 46.0,
-      "step": 29474
-    },
-    {
-      "epoch": 4.746728934337131,
-      "grad_norm": 0.006481084506958723,
-      "learning_rate": 0.00019998888964937157,
-      "loss": 46.0,
-      "step": 29475
-    },
-    {
-      "epoch": 4.746889971415919,
-      "grad_norm": 0.0032839993946254253,
-      "learning_rate": 0.00019998888889523388,
-      "loss": 46.0,
-      "step": 29476
-    },
-    {
-      "epoch": 4.747051008494706,
-      "grad_norm": 0.002434573834761977,
-      "learning_rate": 0.00019998888814107063,
-      "loss": 46.0,
-      "step": 29477
-    },
-    {
-      "epoch": 4.747212045573494,
-      "grad_norm": 0.0016336912522092462,
-      "learning_rate": 0.00019998888738688176,
-      "loss": 46.0,
-      "step": 29478
-    },
-    {
-      "epoch": 4.747373082652281,
-      "grad_norm": 0.005440335255116224,
-      "learning_rate": 0.0001999888866326673,
-      "loss": 46.0,
-      "step": 29479
-    },
-    {
-      "epoch": 4.747534119731068,
-      "grad_norm": 0.010634477250277996,
-      "learning_rate": 0.00019998888587842727,
-      "loss": 46.0,
-      "step": 29480
-    },
-    {
-      "epoch": 4.747695156809855,
-      "grad_norm": 0.002356134122237563,
-      "learning_rate": 0.0001999888851241616,
-      "loss": 46.0,
-      "step": 29481
-    },
-    {
-      "epoch": 4.7478561938886426,
-      "grad_norm": 0.004118389915674925,
-      "learning_rate": 0.00019998888436987037,
-      "loss": 46.0,
-      "step": 29482
-    },
-    {
-      "epoch": 4.74801723096743,
-      "grad_norm": 0.009017317555844784,
-      "learning_rate": 0.00019998888361555354,
-      "loss": 46.0,
-      "step": 29483
-    },
-    {
-      "epoch": 4.7481782680462175,
-      "grad_norm": 0.001375826308503747,
-      "learning_rate": 0.00019998888286121112,
-      "loss": 46.0,
-      "step": 29484
-    },
-    {
-      "epoch": 4.748339305125005,
-      "grad_norm": 0.0009405128075741231,
-      "learning_rate": 0.0001999888821068431,
-      "loss": 46.0,
-      "step": 29485
-    },
-    {
-      "epoch": 4.748500342203792,
-      "grad_norm": 0.0032468524295836687,
-      "learning_rate": 0.0001999888813524495,
-      "loss": 46.0,
-      "step": 29486
-    },
-    {
-      "epoch": 4.74866137928258,
-      "grad_norm": 0.01063726469874382,
-      "learning_rate": 0.00019998888059803032,
-      "loss": 46.0,
-      "step": 29487
-    },
-    {
-      "epoch": 4.748822416361367,
-      "grad_norm": 0.0031899053137749434,
-      "learning_rate": 0.00019998887984358553,
-      "loss": 46.0,
-      "step": 29488
-    },
-    {
-      "epoch": 4.748983453440155,
-      "grad_norm": 0.02001010812819004,
-      "learning_rate": 0.00019998887908911515,
-      "loss": 46.0,
-      "step": 29489
-    },
-    {
-      "epoch": 4.749144490518942,
-      "grad_norm": 0.0026898193173110485,
-      "learning_rate": 0.00019998887833461919,
-      "loss": 46.0,
-      "step": 29490
-    },
-    {
-      "epoch": 4.74930552759773,
-      "grad_norm": 0.012236927635967731,
-      "learning_rate": 0.0001999888775800976,
-      "loss": 46.0,
-      "step": 29491
-    },
-    {
-      "epoch": 4.749466564676517,
-      "grad_norm": 0.005238463170826435,
-      "learning_rate": 0.00019998887682555046,
-      "loss": 46.0,
-      "step": 29492
-    },
-    {
-      "epoch": 4.7496276017553045,
-      "grad_norm": 0.0017704227939248085,
-      "learning_rate": 0.0001999888760709777,
-      "loss": 46.0,
-      "step": 29493
-    },
-    {
-      "epoch": 4.749788638834092,
-      "grad_norm": 0.0015627461252734065,
-      "learning_rate": 0.00019998887531637937,
-      "loss": 46.0,
-      "step": 29494
-    },
-    {
-      "epoch": 4.7499496759128785,
-      "grad_norm": 0.0007592069450765848,
-      "learning_rate": 0.00019998887456175544,
-      "loss": 46.0,
-      "step": 29495
-    },
-    {
-      "epoch": 4.750110712991666,
-      "grad_norm": 0.003729273797944188,
-      "learning_rate": 0.00019998887380710592,
-      "loss": 46.0,
-      "step": 29496
-    },
-    {
-      "epoch": 4.750271750070453,
-      "grad_norm": 0.00501279579475522,
-      "learning_rate": 0.00019998887305243082,
-      "loss": 46.0,
-      "step": 29497
-    },
-    {
-      "epoch": 4.750432787149241,
-      "grad_norm": 0.0038480714429169893,
-      "learning_rate": 0.00019998887229773012,
-      "loss": 46.0,
-      "step": 29498
-    },
-    {
-      "epoch": 4.750593824228028,
-      "grad_norm": 0.0017705995123833418,
-      "learning_rate": 0.0001999888715430038,
-      "loss": 46.0,
-      "step": 29499
-    },
-    {
-      "epoch": 4.750754861306816,
-      "grad_norm": 0.0019863443449139595,
-      "learning_rate": 0.0001999888707882519,
-      "loss": 46.0,
-      "step": 29500
-    },
-    {
-      "epoch": 4.750915898385603,
-      "grad_norm": 0.008791308850049973,
-      "learning_rate": 0.00019998887003347445,
-      "loss": 46.0,
-      "step": 29501
-    },
-    {
-      "epoch": 4.751076935464391,
-      "grad_norm": 0.0016202365513890982,
-      "learning_rate": 0.00019998886927867135,
-      "loss": 46.0,
-      "step": 29502
-    },
-    {
-      "epoch": 4.751237972543178,
-      "grad_norm": 0.004478315357118845,
-      "learning_rate": 0.0001999888685238427,
-      "loss": 46.0,
-      "step": 29503
-    },
-    {
-      "epoch": 4.751399009621966,
-      "grad_norm": 0.0031289015896618366,
-      "learning_rate": 0.00019998886776898843,
-      "loss": 46.0,
-      "step": 29504
-    },
-    {
-      "epoch": 4.751560046700753,
-      "grad_norm": 0.0047536808997392654,
-      "learning_rate": 0.00019998886701410858,
-      "loss": 46.0,
-      "step": 29505
-    },
-    {
-      "epoch": 4.7517210837795405,
-      "grad_norm": 0.0050652530044317245,
-      "learning_rate": 0.00019998886625920313,
-      "loss": 46.0,
-      "step": 29506
-    },
-    {
-      "epoch": 4.751882120858328,
-      "grad_norm": 0.004600538872182369,
-      "learning_rate": 0.0001999888655042721,
-      "loss": 46.0,
-      "step": 29507
-    },
-    {
-      "epoch": 4.752043157937115,
-      "grad_norm": 0.005968911107629538,
-      "learning_rate": 0.00019998886474931546,
-      "loss": 46.0,
-      "step": 29508
-    },
-    {
-      "epoch": 4.752204195015903,
-      "grad_norm": 0.002788045909255743,
-      "learning_rate": 0.00019998886399433326,
-      "loss": 46.0,
-      "step": 29509
-    },
-    {
-      "epoch": 4.752365232094689,
-      "grad_norm": 0.0011554601369425654,
-      "learning_rate": 0.0001999888632393254,
-      "loss": 46.0,
-      "step": 29510
-    },
-    {
-      "epoch": 4.752526269173477,
-      "grad_norm": 0.003151525044813752,
-      "learning_rate": 0.00019998886248429203,
-      "loss": 46.0,
-      "step": 29511
-    },
-    {
-      "epoch": 4.752687306252264,
-      "grad_norm": 0.0034211769234389067,
-      "learning_rate": 0.00019998886172923304,
-      "loss": 46.0,
-      "step": 29512
-    },
-    {
-      "epoch": 4.752848343331052,
-      "grad_norm": 0.013129165396094322,
-      "learning_rate": 0.00019998886097414843,
-      "loss": 46.0,
-      "step": 29513
-    },
-    {
-      "epoch": 4.753009380409839,
-      "grad_norm": 0.007085001096129417,
-      "learning_rate": 0.00019998886021903826,
-      "loss": 46.0,
-      "step": 29514
-    },
-    {
-      "epoch": 4.753170417488627,
-      "grad_norm": 0.0033717367332428694,
-      "learning_rate": 0.0001999888594639025,
-      "loss": 46.0,
-      "step": 29515
-    },
-    {
-      "epoch": 4.753331454567414,
-      "grad_norm": 0.007327381055802107,
-      "learning_rate": 0.00019998885870874114,
-      "loss": 46.0,
-      "step": 29516
-    },
-    {
-      "epoch": 4.753492491646202,
-      "grad_norm": 0.0016730257775634527,
-      "learning_rate": 0.00019998885795355419,
-      "loss": 46.0,
-      "step": 29517
-    },
-    {
-      "epoch": 4.753653528724989,
-      "grad_norm": 0.0070864129811525345,
-      "learning_rate": 0.00019998885719834162,
-      "loss": 46.0,
-      "step": 29518
-    },
-    {
-      "epoch": 4.7538145658037765,
-      "grad_norm": 0.002798378234729171,
-      "learning_rate": 0.00019998885644310349,
-      "loss": 46.0,
-      "step": 29519
-    },
-    {
-      "epoch": 4.753975602882564,
-      "grad_norm": 0.002048699650913477,
-      "learning_rate": 0.00019998885568783977,
-      "loss": 46.0,
-      "step": 29520
-    },
-    {
-      "epoch": 4.754136639961351,
-      "grad_norm": 0.00618006894364953,
-      "learning_rate": 0.00019998885493255044,
-      "loss": 46.0,
-      "step": 29521
-    },
-    {
-      "epoch": 4.754297677040139,
-      "grad_norm": 0.007030103355646133,
-      "learning_rate": 0.00019998885417723552,
-      "loss": 46.0,
-      "step": 29522
-    },
-    {
-      "epoch": 4.754458714118925,
-      "grad_norm": 0.004415110219269991,
-      "learning_rate": 0.000199988853421895,
-      "loss": 46.0,
-      "step": 29523
-    },
-    {
-      "epoch": 4.754619751197714,
-      "grad_norm": 0.006786801386624575,
-      "learning_rate": 0.00019998885266652892,
-      "loss": 46.0,
-      "step": 29524
-    },
-    {
-      "epoch": 4.7547807882765,
-      "grad_norm": 0.012729967013001442,
-      "learning_rate": 0.0001999888519111372,
-      "loss": 46.0,
-      "step": 29525
-    },
-    {
-      "epoch": 4.754941825355288,
-      "grad_norm": 0.0026426208205521107,
-      "learning_rate": 0.00019998885115571992,
-      "loss": 46.0,
-      "step": 29526
-    },
-    {
-      "epoch": 4.755102862434075,
-      "grad_norm": 0.00648786686360836,
-      "learning_rate": 0.00019998885040027706,
-      "loss": 46.0,
-      "step": 29527
-    },
-    {
-      "epoch": 4.755263899512863,
-      "grad_norm": 0.005379085429012775,
-      "learning_rate": 0.0001999888496448086,
-      "loss": 46.0,
-      "step": 29528
-    },
-    {
-      "epoch": 4.75542493659165,
-      "grad_norm": 0.010481463745236397,
-      "learning_rate": 0.00019998884888931454,
-      "loss": 46.0,
-      "step": 29529
-    },
-    {
-      "epoch": 4.755585973670438,
-      "grad_norm": 0.01833781599998474,
-      "learning_rate": 0.0001999888481337949,
-      "loss": 46.0,
-      "step": 29530
-    },
-    {
-      "epoch": 4.755747010749225,
-      "grad_norm": 0.002483892487362027,
-      "learning_rate": 0.00019998884737824964,
-      "loss": 46.0,
-      "step": 29531
-    },
-    {
-      "epoch": 4.7559080478280125,
-      "grad_norm": 0.0020950946491211653,
-      "learning_rate": 0.0001999888466226788,
-      "loss": 46.0,
-      "step": 29532
-    },
-    {
-      "epoch": 4.7560690849068,
-      "grad_norm": 0.004345500376075506,
-      "learning_rate": 0.0001999888458670824,
-      "loss": 46.0,
-      "step": 29533
-    },
-    {
-      "epoch": 4.756230121985587,
-      "grad_norm": 0.0018428814364597201,
-      "learning_rate": 0.00019998884511146034,
-      "loss": 46.0,
-      "step": 29534
-    },
-    {
-      "epoch": 4.756391159064375,
-      "grad_norm": 0.015519008040428162,
-      "learning_rate": 0.00019998884435581276,
-      "loss": 46.0,
-      "step": 29535
-    },
-    {
-      "epoch": 4.756552196143162,
-      "grad_norm": 0.006286387797445059,
-      "learning_rate": 0.00019998884360013954,
-      "loss": 46.0,
-      "step": 29536
-    },
-    {
-      "epoch": 4.75671323322195,
-      "grad_norm": 0.0033569256775081158,
-      "learning_rate": 0.00019998884284444076,
-      "loss": 46.0,
-      "step": 29537
-    },
-    {
-      "epoch": 4.756874270300736,
-      "grad_norm": 0.00317478203214705,
-      "learning_rate": 0.00019998884208871637,
-      "loss": 46.0,
-      "step": 29538
-    },
-    {
-      "epoch": 4.757035307379525,
-      "grad_norm": 0.004013438709080219,
-      "learning_rate": 0.00019998884133296639,
-      "loss": 46.0,
-      "step": 29539
-    },
-    {
-      "epoch": 4.757196344458311,
-      "grad_norm": 0.001528636901639402,
-      "learning_rate": 0.00019998884057719082,
-      "loss": 46.0,
-      "step": 29540
-    },
-    {
-      "epoch": 4.757357381537099,
-      "grad_norm": 0.005556702613830566,
-      "learning_rate": 0.00019998883982138966,
-      "loss": 46.0,
-      "step": 29541
-    },
-    {
-      "epoch": 4.757518418615886,
-      "grad_norm": 0.004374017007648945,
-      "learning_rate": 0.0001999888390655629,
-      "loss": 46.0,
-      "step": 29542
-    },
-    {
-      "epoch": 4.7576794556946735,
-      "grad_norm": 0.003192405914887786,
-      "learning_rate": 0.00019998883830971056,
-      "loss": 46.0,
-      "step": 29543
-    },
-    {
-      "epoch": 4.757840492773461,
-      "grad_norm": 0.0022569317370653152,
-      "learning_rate": 0.0001999888375538326,
-      "loss": 46.0,
-      "step": 29544
-    },
-    {
-      "epoch": 4.7580015298522484,
-      "grad_norm": 0.0034018929582089186,
-      "learning_rate": 0.00019998883679792906,
-      "loss": 46.0,
-      "step": 29545
-    },
-    {
-      "epoch": 4.758162566931036,
-      "grad_norm": 0.002782811876386404,
-      "learning_rate": 0.00019998883604199996,
-      "loss": 46.0,
-      "step": 29546
-    },
-    {
-      "epoch": 4.758323604009823,
-      "grad_norm": 0.0008092928910627961,
-      "learning_rate": 0.00019998883528604523,
-      "loss": 46.0,
-      "step": 29547
-    },
-    {
-      "epoch": 4.758484641088611,
-      "grad_norm": 0.0011253795819357038,
-      "learning_rate": 0.00019998883453006494,
-      "loss": 46.0,
-      "step": 29548
-    },
-    {
-      "epoch": 4.758645678167398,
-      "grad_norm": 0.0011445381678640842,
-      "learning_rate": 0.000199988833774059,
-      "loss": 46.0,
-      "step": 29549
-    },
-    {
-      "epoch": 4.758806715246186,
-      "grad_norm": 0.0024955698754638433,
-      "learning_rate": 0.00019998883301802753,
-      "loss": 46.0,
-      "step": 29550
-    },
-    {
-      "epoch": 4.758967752324973,
-      "grad_norm": 0.012237120419740677,
-      "learning_rate": 0.00019998883226197045,
-      "loss": 46.0,
-      "step": 29551
-    },
-    {
-      "epoch": 4.759128789403761,
-      "grad_norm": 0.02256646752357483,
-      "learning_rate": 0.00019998883150588775,
-      "loss": 46.0,
-      "step": 29552
-    },
-    {
-      "epoch": 4.759289826482547,
-      "grad_norm": 0.013496628031134605,
-      "learning_rate": 0.0001999888307497795,
-      "loss": 46.0,
-      "step": 29553
-    },
-    {
-      "epoch": 4.759450863561335,
-      "grad_norm": 0.012984919361770153,
-      "learning_rate": 0.00019998882999364565,
-      "loss": 46.0,
-      "step": 29554
-    },
-    {
-      "epoch": 4.759611900640122,
-      "grad_norm": 0.0016839482123032212,
-      "learning_rate": 0.0001999888292374862,
-      "loss": 46.0,
-      "step": 29555
-    },
-    {
-      "epoch": 4.7597729377189095,
-      "grad_norm": 0.0036112088710069656,
-      "learning_rate": 0.00019998882848130115,
-      "loss": 46.0,
-      "step": 29556
-    },
-    {
-      "epoch": 4.759933974797697,
-      "grad_norm": 0.0018007522448897362,
-      "learning_rate": 0.00019998882772509052,
-      "loss": 46.0,
-      "step": 29557
-    },
-    {
-      "epoch": 4.760095011876484,
-      "grad_norm": 0.00039856089279055595,
-      "learning_rate": 0.00019998882696885427,
-      "loss": 46.0,
-      "step": 29558
-    },
-    {
-      "epoch": 4.760256048955272,
-      "grad_norm": 0.0016185863642022014,
-      "learning_rate": 0.00019998882621259244,
-      "loss": 46.0,
-      "step": 29559
-    },
-    {
-      "epoch": 4.760417086034059,
-      "grad_norm": 0.0032558897510170937,
-      "learning_rate": 0.00019998882545630507,
-      "loss": 46.0,
-      "step": 29560
-    },
-    {
-      "epoch": 4.760578123112847,
-      "grad_norm": 0.0011197688290849328,
-      "learning_rate": 0.00019998882469999204,
-      "loss": 46.0,
-      "step": 29561
-    },
-    {
-      "epoch": 4.760739160191634,
-      "grad_norm": 0.00581378536298871,
-      "learning_rate": 0.00019998882394365347,
-      "loss": 46.0,
-      "step": 29562
-    },
-    {
-      "epoch": 4.760900197270422,
-      "grad_norm": 0.0063275909051299095,
-      "learning_rate": 0.00019998882318728926,
-      "loss": 46.0,
-      "step": 29563
-    },
-    {
-      "epoch": 4.761061234349209,
-      "grad_norm": 0.0014789545675739646,
-      "learning_rate": 0.0001999888224308995,
-      "loss": 46.0,
-      "step": 29564
-    },
-    {
-      "epoch": 4.761222271427997,
-      "grad_norm": 0.003132209647446871,
-      "learning_rate": 0.0001999888216744841,
-      "loss": 46.0,
-      "step": 29565
-    },
-    {
-      "epoch": 4.761383308506784,
-      "grad_norm": 0.002272936748340726,
-      "learning_rate": 0.00019998882091804316,
-      "loss": 46.0,
-      "step": 29566
-    },
-    {
-      "epoch": 4.7615443455855715,
-      "grad_norm": 0.004940418992191553,
-      "learning_rate": 0.00019998882016157658,
-      "loss": 46.0,
-      "step": 29567
-    },
-    {
-      "epoch": 4.761705382664358,
-      "grad_norm": 0.003442492103204131,
-      "learning_rate": 0.00019998881940508446,
-      "loss": 46.0,
-      "step": 29568
-    },
-    {
-      "epoch": 4.7618664197431455,
-      "grad_norm": 0.013806091621518135,
-      "learning_rate": 0.0001999888186485667,
-      "loss": 46.0,
-      "step": 29569
-    },
-    {
-      "epoch": 4.762027456821933,
-      "grad_norm": 0.0017356426687911153,
-      "learning_rate": 0.00019998881789202338,
-      "loss": 46.0,
-      "step": 29570
-    },
-    {
-      "epoch": 4.76218849390072,
-      "grad_norm": 0.001779056154191494,
-      "learning_rate": 0.00019998881713545444,
-      "loss": 46.0,
-      "step": 29571
-    },
-    {
-      "epoch": 4.762349530979508,
-      "grad_norm": 0.0057486738078296185,
-      "learning_rate": 0.00019998881637885992,
-      "loss": 46.0,
-      "step": 29572
-    },
-    {
-      "epoch": 4.762510568058295,
-      "grad_norm": 0.0010489927371963859,
-      "learning_rate": 0.00019998881562223981,
-      "loss": 46.0,
-      "step": 29573
-    },
-    {
-      "epoch": 4.762671605137083,
-      "grad_norm": 0.0031836184207350016,
-      "learning_rate": 0.00019998881486559412,
-      "loss": 46.0,
-      "step": 29574
-    },
-    {
-      "epoch": 4.76283264221587,
-      "grad_norm": 0.006618183571845293,
-      "learning_rate": 0.00019998881410892284,
-      "loss": 46.0,
-      "step": 29575
-    },
-    {
-      "epoch": 4.762993679294658,
-      "grad_norm": 0.0011528455652296543,
-      "learning_rate": 0.00019998881335222594,
-      "loss": 46.0,
-      "step": 29576
-    },
-    {
-      "epoch": 4.763154716373445,
-      "grad_norm": 0.02076326310634613,
-      "learning_rate": 0.00019998881259550348,
-      "loss": 46.0,
-      "step": 29577
-    },
-    {
-      "epoch": 4.763315753452233,
-      "grad_norm": 0.011185937561094761,
-      "learning_rate": 0.00019998881183875538,
-      "loss": 46.0,
-      "step": 29578
-    },
-    {
-      "epoch": 4.76347679053102,
-      "grad_norm": 0.006465322803705931,
-      "learning_rate": 0.00019998881108198173,
-      "loss": 46.0,
-      "step": 29579
-    },
-    {
-      "epoch": 4.7636378276098075,
-      "grad_norm": 0.00529022142291069,
-      "learning_rate": 0.00019998881032518248,
-      "loss": 46.0,
-      "step": 29580
-    },
-    {
-      "epoch": 4.763798864688595,
-      "grad_norm": 0.0014770481502637267,
-      "learning_rate": 0.00019998880956835765,
-      "loss": 46.0,
-      "step": 29581
-    },
-    {
-      "epoch": 4.763959901767382,
-      "grad_norm": 0.008048774674534798,
-      "learning_rate": 0.0001999888088115072,
-      "loss": 46.0,
-      "step": 29582
-    },
-    {
-      "epoch": 4.764120938846169,
-      "grad_norm": 0.0027655549347400665,
-      "learning_rate": 0.0001999888080546312,
-      "loss": 46.0,
-      "step": 29583
-    },
-    {
-      "epoch": 4.764281975924956,
-      "grad_norm": 0.0013003175845369697,
-      "learning_rate": 0.00019998880729772957,
-      "loss": 46.0,
-      "step": 29584
-    },
-    {
-      "epoch": 4.764443013003744,
-      "grad_norm": 0.002102261409163475,
-      "learning_rate": 0.00019998880654080234,
-      "loss": 46.0,
-      "step": 29585
-    },
-    {
-      "epoch": 4.764604050082531,
-      "grad_norm": 0.0018356706714257598,
-      "learning_rate": 0.00019998880578384954,
-      "loss": 46.0,
-      "step": 29586
-    },
-    {
-      "epoch": 4.764765087161319,
-      "grad_norm": 0.001584004727192223,
-      "learning_rate": 0.00019998880502687113,
-      "loss": 46.0,
-      "step": 29587
-    },
-    {
-      "epoch": 4.764926124240106,
-      "grad_norm": 0.005989392288029194,
-      "learning_rate": 0.00019998880426986716,
-      "loss": 46.0,
-      "step": 29588
-    },
-    {
-      "epoch": 4.765087161318894,
-      "grad_norm": 0.0026728089433163404,
-      "learning_rate": 0.00019998880351283758,
-      "loss": 46.0,
-      "step": 29589
-    },
-    {
-      "epoch": 4.765248198397681,
-      "grad_norm": 0.0032795623410493135,
-      "learning_rate": 0.0001999888027557824,
-      "loss": 46.0,
-      "step": 29590
-    },
-    {
-      "epoch": 4.7654092354764686,
-      "grad_norm": 0.007996369153261185,
-      "learning_rate": 0.00019998880199870164,
-      "loss": 46.0,
-      "step": 29591
-    },
-    {
-      "epoch": 4.765570272555256,
-      "grad_norm": 0.011026730760931969,
-      "learning_rate": 0.0001999888012415953,
-      "loss": 46.0,
-      "step": 29592
-    },
-    {
-      "epoch": 4.7657313096340435,
-      "grad_norm": 0.003427068470045924,
-      "learning_rate": 0.00019998880048446337,
-      "loss": 46.0,
-      "step": 29593
-    },
-    {
-      "epoch": 4.765892346712831,
-      "grad_norm": 0.009349680505692959,
-      "learning_rate": 0.0001999887997273058,
-      "loss": 46.0,
-      "step": 29594
-    },
-    {
-      "epoch": 4.766053383791618,
-      "grad_norm": 0.0038277243729680777,
-      "learning_rate": 0.00019998879897012266,
-      "loss": 46.0,
-      "step": 29595
-    },
-    {
-      "epoch": 4.766214420870405,
-      "grad_norm": 0.005733050871640444,
-      "learning_rate": 0.00019998879821291394,
-      "loss": 46.0,
-      "step": 29596
-    },
-    {
-      "epoch": 4.766375457949193,
-      "grad_norm": 0.0020615593530237675,
-      "learning_rate": 0.00019998879745567965,
-      "loss": 46.0,
-      "step": 29597
-    },
-    {
-      "epoch": 4.76653649502798,
-      "grad_norm": 0.00975791085511446,
-      "learning_rate": 0.00019998879669841973,
-      "loss": 46.0,
-      "step": 29598
-    },
-    {
-      "epoch": 4.766697532106767,
-      "grad_norm": 0.0031067838426679373,
-      "learning_rate": 0.00019998879594113422,
-      "loss": 46.0,
-      "step": 29599
-    },
-    {
-      "epoch": 4.766858569185555,
-      "grad_norm": 0.005844696890562773,
-      "learning_rate": 0.00019998879518382312,
-      "loss": 46.0,
-      "step": 29600
-    },
-    {
-      "epoch": 4.767019606264342,
-      "grad_norm": 0.005044762510806322,
-      "learning_rate": 0.00019998879442648644,
-      "loss": 46.0,
-      "step": 29601
-    },
-    {
-      "epoch": 4.76718064334313,
-      "grad_norm": 0.0058305230922997,
-      "learning_rate": 0.00019998879366912417,
-      "loss": 46.0,
-      "step": 29602
-    },
-    {
-      "epoch": 4.767341680421917,
-      "grad_norm": 0.01473147887736559,
-      "learning_rate": 0.00019998879291173628,
-      "loss": 46.0,
-      "step": 29603
-    },
-    {
-      "epoch": 4.7675027175007045,
-      "grad_norm": 0.0036365208216011524,
-      "learning_rate": 0.00019998879215432283,
-      "loss": 46.0,
-      "step": 29604
-    },
-    {
-      "epoch": 4.767663754579492,
-      "grad_norm": 0.004284884314984083,
-      "learning_rate": 0.00019998879139688377,
-      "loss": 46.0,
-      "step": 29605
-    },
-    {
-      "epoch": 4.767824791658279,
-      "grad_norm": 0.0049233026802539825,
-      "learning_rate": 0.00019998879063941915,
-      "loss": 46.0,
-      "step": 29606
-    },
-    {
-      "epoch": 4.767985828737067,
-      "grad_norm": 0.007909219712018967,
-      "learning_rate": 0.00019998878988192892,
-      "loss": 46.0,
-      "step": 29607
-    },
-    {
-      "epoch": 4.768146865815854,
-      "grad_norm": 0.010383238084614277,
-      "learning_rate": 0.00019998878912441307,
-      "loss": 46.0,
-      "step": 29608
-    },
-    {
-      "epoch": 4.768307902894642,
-      "grad_norm": 0.0022008235100656748,
-      "learning_rate": 0.00019998878836687166,
-      "loss": 46.0,
-      "step": 29609
-    },
-    {
-      "epoch": 4.768468939973429,
-      "grad_norm": 0.005689368117600679,
-      "learning_rate": 0.00019998878760930464,
-      "loss": 46.0,
-      "step": 29610
-    },
-    {
-      "epoch": 4.768629977052216,
-      "grad_norm": 0.0069381617940962315,
-      "learning_rate": 0.00019998878685171203,
-      "loss": 46.0,
-      "step": 29611
-    },
-    {
-      "epoch": 4.768791014131004,
-      "grad_norm": 0.0028861570172011852,
-      "learning_rate": 0.00019998878609409383,
-      "loss": 46.0,
-      "step": 29612
-    },
-    {
-      "epoch": 4.768952051209791,
-      "grad_norm": 0.011469135992228985,
-      "learning_rate": 0.00019998878533645007,
-      "loss": 46.0,
-      "step": 29613
-    },
-    {
-      "epoch": 4.769113088288578,
-      "grad_norm": 0.0030142394825816154,
-      "learning_rate": 0.00019998878457878067,
-      "loss": 46.0,
-      "step": 29614
-    },
-    {
-      "epoch": 4.769274125367366,
-      "grad_norm": 0.01402412448078394,
-      "learning_rate": 0.0001999887838210857,
-      "loss": 46.0,
-      "step": 29615
-    },
-    {
-      "epoch": 4.769435162446153,
-      "grad_norm": 0.004262520000338554,
-      "learning_rate": 0.00019998878306336514,
-      "loss": 46.0,
-      "step": 29616
-    },
-    {
-      "epoch": 4.7695961995249405,
-      "grad_norm": 0.002674600575119257,
-      "learning_rate": 0.00019998878230561898,
-      "loss": 46.0,
-      "step": 29617
-    },
-    {
-      "epoch": 4.769757236603728,
-      "grad_norm": 0.015288958325982094,
-      "learning_rate": 0.00019998878154784723,
-      "loss": 46.0,
-      "step": 29618
-    },
-    {
-      "epoch": 4.769918273682515,
-      "grad_norm": 0.0068214405328035355,
-      "learning_rate": 0.0001999887807900499,
-      "loss": 46.0,
-      "step": 29619
-    },
-    {
-      "epoch": 4.770079310761303,
-      "grad_norm": 0.006429290398955345,
-      "learning_rate": 0.00019998878003222697,
-      "loss": 46.0,
-      "step": 29620
-    },
-    {
-      "epoch": 4.77024034784009,
-      "grad_norm": 0.0014157138066366315,
-      "learning_rate": 0.00019998877927437844,
-      "loss": 46.0,
-      "step": 29621
-    },
-    {
-      "epoch": 4.770401384918878,
-      "grad_norm": 0.008382156491279602,
-      "learning_rate": 0.00019998877851650434,
-      "loss": 46.0,
-      "step": 29622
-    },
-    {
-      "epoch": 4.770562421997665,
-      "grad_norm": 0.0020573085639625788,
-      "learning_rate": 0.0001999887777586046,
-      "loss": 46.0,
-      "step": 29623
-    },
-    {
-      "epoch": 4.770723459076453,
-      "grad_norm": 0.0018315042834728956,
-      "learning_rate": 0.00019998877700067933,
-      "loss": 46.0,
-      "step": 29624
-    },
-    {
-      "epoch": 4.77088449615524,
-      "grad_norm": 0.0009660259820520878,
-      "learning_rate": 0.00019998877624272842,
-      "loss": 46.0,
-      "step": 29625
-    },
-    {
-      "epoch": 4.771045533234027,
-      "grad_norm": 0.0021420144475996494,
-      "learning_rate": 0.00019998877548475194,
-      "loss": 46.0,
-      "step": 29626
-    },
-    {
-      "epoch": 4.771206570312814,
-      "grad_norm": 0.0023009763099253178,
-      "learning_rate": 0.00019998877472674986,
-      "loss": 46.0,
-      "step": 29627
-    },
-    {
-      "epoch": 4.771367607391602,
-      "grad_norm": 0.0018324163975194097,
-      "learning_rate": 0.0001999887739687222,
-      "loss": 46.0,
-      "step": 29628
-    },
-    {
-      "epoch": 4.771528644470389,
-      "grad_norm": 0.020726684480905533,
-      "learning_rate": 0.00019998877321066892,
-      "loss": 46.0,
-      "step": 29629
-    },
-    {
-      "epoch": 4.7716896815491765,
-      "grad_norm": 0.005250627640634775,
-      "learning_rate": 0.0001999887724525901,
-      "loss": 46.0,
-      "step": 29630
-    },
-    {
-      "epoch": 4.771850718627964,
-      "grad_norm": 0.009093799628317356,
-      "learning_rate": 0.0001999887716944856,
-      "loss": 46.0,
-      "step": 29631
-    },
-    {
-      "epoch": 4.772011755706751,
-      "grad_norm": 0.0037383155431598425,
-      "learning_rate": 0.00019998877093635562,
-      "loss": 46.0,
-      "step": 29632
-    },
-    {
-      "epoch": 4.772172792785539,
-      "grad_norm": 0.008891158737242222,
-      "learning_rate": 0.00019998877017819998,
-      "loss": 46.0,
-      "step": 29633
-    },
-    {
-      "epoch": 4.772333829864326,
-      "grad_norm": 0.0023776928428560495,
-      "learning_rate": 0.00019998876942001873,
-      "loss": 46.0,
-      "step": 29634
-    },
-    {
-      "epoch": 4.772494866943114,
-      "grad_norm": 0.0023169522173702717,
-      "learning_rate": 0.00019998876866181195,
-      "loss": 46.0,
-      "step": 29635
-    },
-    {
-      "epoch": 4.772655904021901,
-      "grad_norm": 0.009000053629279137,
-      "learning_rate": 0.00019998876790357955,
-      "loss": 46.0,
-      "step": 29636
-    },
-    {
-      "epoch": 4.772816941100689,
-      "grad_norm": 0.0033633296843618155,
-      "learning_rate": 0.00019998876714532156,
-      "loss": 46.0,
-      "step": 29637
-    },
-    {
-      "epoch": 4.772977978179476,
-      "grad_norm": 0.014860347844660282,
-      "learning_rate": 0.00019998876638703793,
-      "loss": 46.0,
-      "step": 29638
-    },
-    {
-      "epoch": 4.773139015258264,
-      "grad_norm": 0.0018532221438363194,
-      "learning_rate": 0.00019998876562872877,
-      "loss": 46.0,
-      "step": 29639
-    },
-    {
-      "epoch": 4.773300052337051,
-      "grad_norm": 0.01112237386405468,
-      "learning_rate": 0.000199988764870394,
-      "loss": 46.0,
-      "step": 29640
-    },
-    {
-      "epoch": 4.773461089415838,
-      "grad_norm": 0.007151414640247822,
-      "learning_rate": 0.0001999887641120336,
-      "loss": 46.0,
-      "step": 29641
-    },
-    {
-      "epoch": 4.773622126494625,
-      "grad_norm": 0.0030269986018538475,
-      "learning_rate": 0.0001999887633536477,
-      "loss": 46.0,
-      "step": 29642
-    },
-    {
-      "epoch": 4.7737831635734125,
-      "grad_norm": 0.010572108440101147,
-      "learning_rate": 0.00019998876259523613,
-      "loss": 46.0,
-      "step": 29643
-    },
-    {
-      "epoch": 4.7739442006522,
-      "grad_norm": 0.015476390719413757,
-      "learning_rate": 0.000199988761836799,
-      "loss": 46.0,
-      "step": 29644
-    },
-    {
-      "epoch": 4.774105237730987,
-      "grad_norm": 0.008253595791757107,
-      "learning_rate": 0.00019998876107833627,
-      "loss": 46.0,
-      "step": 29645
-    },
-    {
-      "epoch": 4.774266274809775,
-      "grad_norm": 0.0030193838756531477,
-      "learning_rate": 0.00019998876031984794,
-      "loss": 46.0,
-      "step": 29646
-    },
-    {
-      "epoch": 4.774427311888562,
-      "grad_norm": 0.0036472859792411327,
-      "learning_rate": 0.00019998875956133403,
-      "loss": 46.0,
-      "step": 29647
-    },
-    {
-      "epoch": 4.77458834896735,
-      "grad_norm": 0.00540729146450758,
-      "learning_rate": 0.00019998875880279453,
-      "loss": 46.0,
-      "step": 29648
-    },
-    {
-      "epoch": 4.774749386046137,
-      "grad_norm": 0.001582067459821701,
-      "learning_rate": 0.0001999887580442294,
-      "loss": 46.0,
-      "step": 29649
-    },
-    {
-      "epoch": 4.774910423124925,
-      "grad_norm": 0.005634675268083811,
-      "learning_rate": 0.0001999887572856387,
-      "loss": 46.0,
-      "step": 29650
-    },
-    {
-      "epoch": 4.775071460203712,
-      "grad_norm": 0.019446276128292084,
-      "learning_rate": 0.00019998875652702243,
-      "loss": 46.0,
-      "step": 29651
-    },
-    {
-      "epoch": 4.7752324972824995,
-      "grad_norm": 0.0036128598731011152,
-      "learning_rate": 0.00019998875576838056,
-      "loss": 46.0,
-      "step": 29652
-    },
-    {
-      "epoch": 4.775393534361287,
-      "grad_norm": 0.004637226928025484,
-      "learning_rate": 0.00019998875500971307,
-      "loss": 46.0,
-      "step": 29653
-    },
-    {
-      "epoch": 4.7755545714400744,
-      "grad_norm": 0.005163829308003187,
-      "learning_rate": 0.00019998875425102002,
-      "loss": 46.0,
-      "step": 29654
-    },
-    {
-      "epoch": 4.775715608518862,
-      "grad_norm": 0.0034166902769356966,
-      "learning_rate": 0.00019998875349230136,
-      "loss": 46.0,
-      "step": 29655
-    },
-    {
-      "epoch": 4.7758766455976485,
-      "grad_norm": 0.0013359319418668747,
-      "learning_rate": 0.00019998875273355714,
-      "loss": 46.0,
-      "step": 29656
-    },
-    {
-      "epoch": 4.776037682676436,
-      "grad_norm": 0.010175767354667187,
-      "learning_rate": 0.0001999887519747873,
-      "loss": 46.0,
-      "step": 29657
-    },
-    {
-      "epoch": 4.776198719755223,
-      "grad_norm": 0.004424629267305136,
-      "learning_rate": 0.00019998875121599187,
-      "loss": 46.0,
-      "step": 29658
-    },
-    {
-      "epoch": 4.776359756834011,
-      "grad_norm": 0.0019444164354354143,
-      "learning_rate": 0.00019998875045717084,
-      "loss": 46.0,
-      "step": 29659
-    },
-    {
-      "epoch": 4.776520793912798,
-      "grad_norm": 0.003968354314565659,
-      "learning_rate": 0.00019998874969832424,
-      "loss": 46.0,
-      "step": 29660
-    },
-    {
-      "epoch": 4.776681830991586,
-      "grad_norm": 0.0024845667649060488,
-      "learning_rate": 0.00019998874893945202,
-      "loss": 46.0,
-      "step": 29661
-    },
-    {
-      "epoch": 4.776842868070373,
-      "grad_norm": 0.0068671079352498055,
-      "learning_rate": 0.00019998874818055425,
-      "loss": 46.0,
-      "step": 29662
-    },
-    {
-      "epoch": 4.777003905149161,
-      "grad_norm": 0.002554807346314192,
-      "learning_rate": 0.00019998874742163086,
-      "loss": 46.0,
-      "step": 29663
-    },
-    {
-      "epoch": 4.777164942227948,
-      "grad_norm": 0.0033804159611463547,
-      "learning_rate": 0.0001999887466626819,
-      "loss": 46.0,
-      "step": 29664
-    },
-    {
-      "epoch": 4.7773259793067355,
-      "grad_norm": 0.006721459329128265,
-      "learning_rate": 0.0001999887459037073,
-      "loss": 46.0,
-      "step": 29665
-    },
-    {
-      "epoch": 4.777487016385523,
-      "grad_norm": 0.0022280775010585785,
-      "learning_rate": 0.00019998874514470712,
-      "loss": 46.0,
-      "step": 29666
-    },
-    {
-      "epoch": 4.77764805346431,
-      "grad_norm": 0.0026382971554994583,
-      "learning_rate": 0.0001999887443856814,
-      "loss": 46.0,
-      "step": 29667
-    },
-    {
-      "epoch": 4.777809090543098,
-      "grad_norm": 0.0062921964563429356,
-      "learning_rate": 0.00019998874362663004,
-      "loss": 46.0,
-      "step": 29668
-    },
-    {
-      "epoch": 4.7779701276218844,
-      "grad_norm": 0.005004643928259611,
-      "learning_rate": 0.0001999887428675531,
-      "loss": 46.0,
-      "step": 29669
-    },
-    {
-      "epoch": 4.778131164700673,
-      "grad_norm": 0.0064356098882853985,
-      "learning_rate": 0.00019998874210845057,
-      "loss": 46.0,
-      "step": 29670
-    },
-    {
-      "epoch": 4.778292201779459,
-      "grad_norm": 0.007435160223394632,
-      "learning_rate": 0.00019998874134932244,
-      "loss": 46.0,
-      "step": 29671
-    },
-    {
-      "epoch": 4.778453238858247,
-      "grad_norm": 0.004613068420439959,
-      "learning_rate": 0.0001999887405901687,
-      "loss": 46.0,
-      "step": 29672
-    },
-    {
-      "epoch": 4.778614275937034,
-      "grad_norm": 0.010184877552092075,
-      "learning_rate": 0.00019998873983098942,
-      "loss": 46.0,
-      "step": 29673
-    },
-    {
-      "epoch": 4.778775313015822,
-      "grad_norm": 0.0011658653384074569,
-      "learning_rate": 0.0001999887390717845,
-      "loss": 46.0,
-      "step": 29674
-    },
-    {
-      "epoch": 4.778936350094609,
-      "grad_norm": 0.006491041276603937,
-      "learning_rate": 0.00019998873831255404,
-      "loss": 46.0,
-      "step": 29675
-    },
-    {
-      "epoch": 4.779097387173397,
-      "grad_norm": 0.008794019930064678,
-      "learning_rate": 0.00019998873755329793,
-      "loss": 46.0,
-      "step": 29676
-    },
-    {
-      "epoch": 4.779258424252184,
-      "grad_norm": 0.004227577243000269,
-      "learning_rate": 0.00019998873679401627,
-      "loss": 46.0,
-      "step": 29677
-    },
-    {
-      "epoch": 4.7794194613309715,
-      "grad_norm": 0.005861061159521341,
-      "learning_rate": 0.000199988736034709,
-      "loss": 46.0,
-      "step": 29678
-    },
-    {
-      "epoch": 4.779580498409759,
-      "grad_norm": 0.01167288888245821,
-      "learning_rate": 0.00019998873527537613,
-      "loss": 46.0,
-      "step": 29679
-    },
-    {
-      "epoch": 4.779741535488546,
-      "grad_norm": 0.0011174395913258195,
-      "learning_rate": 0.00019998873451601768,
-      "loss": 46.0,
-      "step": 29680
-    },
-    {
-      "epoch": 4.779902572567334,
-      "grad_norm": 0.0061180549673736095,
-      "learning_rate": 0.00019998873375663364,
-      "loss": 46.0,
-      "step": 29681
-    },
-    {
-      "epoch": 4.780063609646121,
-      "grad_norm": 0.023945102468132973,
-      "learning_rate": 0.00019998873299722402,
-      "loss": 46.0,
-      "step": 29682
-    },
-    {
-      "epoch": 4.780224646724909,
-      "grad_norm": 0.01428929902613163,
-      "learning_rate": 0.00019998873223778878,
-      "loss": 46.0,
-      "step": 29683
-    },
-    {
-      "epoch": 4.780385683803695,
-      "grad_norm": 0.00374038890004158,
-      "learning_rate": 0.00019998873147832798,
-      "loss": 46.0,
-      "step": 29684
-    },
-    {
-      "epoch": 4.780546720882484,
-      "grad_norm": 0.005625316873192787,
-      "learning_rate": 0.00019998873071884154,
-      "loss": 46.0,
-      "step": 29685
-    },
-    {
-      "epoch": 4.78070775796127,
-      "grad_norm": 0.010602965019643307,
-      "learning_rate": 0.00019998872995932954,
-      "loss": 46.0,
-      "step": 29686
-    },
-    {
-      "epoch": 4.780868795040058,
-      "grad_norm": 0.004356399178504944,
-      "learning_rate": 0.00019998872919979195,
-      "loss": 46.0,
-      "step": 29687
-    },
-    {
-      "epoch": 4.781029832118845,
-      "grad_norm": 0.005907748360186815,
-      "learning_rate": 0.00019998872844022875,
-      "loss": 46.0,
-      "step": 29688
-    },
-    {
-      "epoch": 4.781190869197633,
-      "grad_norm": 0.0007467250688932836,
-      "learning_rate": 0.00019998872768063998,
-      "loss": 46.0,
-      "step": 29689
-    },
-    {
-      "epoch": 4.78135190627642,
-      "grad_norm": 0.000649077002890408,
-      "learning_rate": 0.0001999887269210256,
-      "loss": 46.0,
-      "step": 29690
-    },
-    {
-      "epoch": 4.7815129433552075,
-      "grad_norm": 0.008154094219207764,
-      "learning_rate": 0.00019998872616138564,
-      "loss": 46.0,
-      "step": 29691
-    },
-    {
-      "epoch": 4.781673980433995,
-      "grad_norm": 0.011132962070405483,
-      "learning_rate": 0.0001999887254017201,
-      "loss": 46.0,
-      "step": 29692
-    },
-    {
-      "epoch": 4.781835017512782,
-      "grad_norm": 0.00263105146586895,
-      "learning_rate": 0.00019998872464202893,
-      "loss": 46.0,
-      "step": 29693
-    },
-    {
-      "epoch": 4.78199605459157,
-      "grad_norm": 0.01146366260945797,
-      "learning_rate": 0.0001999887238823122,
-      "loss": 46.0,
-      "step": 29694
-    },
-    {
-      "epoch": 4.782157091670357,
-      "grad_norm": 0.003947129473090172,
-      "learning_rate": 0.0001999887231225699,
-      "loss": 46.0,
-      "step": 29695
-    },
-    {
-      "epoch": 4.782318128749145,
-      "grad_norm": 0.0026669823564589024,
-      "learning_rate": 0.00019998872236280196,
-      "loss": 46.0,
-      "step": 29696
-    },
-    {
-      "epoch": 4.782479165827932,
-      "grad_norm": 0.0018738452345132828,
-      "learning_rate": 0.00019998872160300842,
-      "loss": 46.0,
-      "step": 29697
-    },
-    {
-      "epoch": 4.78264020290672,
-      "grad_norm": 0.004479184281080961,
-      "learning_rate": 0.00019998872084318935,
-      "loss": 46.0,
-      "step": 29698
-    },
-    {
-      "epoch": 4.782801239985506,
-      "grad_norm": 0.0032427734695374966,
-      "learning_rate": 0.0001999887200833446,
-      "loss": 46.0,
-      "step": 29699
-    },
-    {
-      "epoch": 4.782962277064294,
-      "grad_norm": 0.001625141710974276,
-      "learning_rate": 0.00019998871932347433,
-      "loss": 46.0,
-      "step": 29700
-    },
-    {
-      "epoch": 4.783123314143081,
-      "grad_norm": 0.006443955469876528,
-      "learning_rate": 0.00019998871856357844,
-      "loss": 46.0,
-      "step": 29701
-    },
-    {
-      "epoch": 4.783284351221869,
-      "grad_norm": 0.002301170490682125,
-      "learning_rate": 0.00019998871780365696,
-      "loss": 46.0,
-      "step": 29702
-    },
-    {
-      "epoch": 4.783445388300656,
-      "grad_norm": 0.004346245899796486,
-      "learning_rate": 0.0001999887170437099,
-      "loss": 46.0,
-      "step": 29703
-    },
-    {
-      "epoch": 4.7836064253794435,
-      "grad_norm": 0.0034879206214100122,
-      "learning_rate": 0.00019998871628373725,
-      "loss": 46.0,
-      "step": 29704
-    },
-    {
-      "epoch": 4.783767462458231,
-      "grad_norm": 0.005502117332071066,
-      "learning_rate": 0.00019998871552373898,
-      "loss": 46.0,
-      "step": 29705
-    },
-    {
-      "epoch": 4.783928499537018,
-      "grad_norm": 0.0031630699522793293,
-      "learning_rate": 0.00019998871476371516,
-      "loss": 46.0,
-      "step": 29706
-    },
-    {
-      "epoch": 4.784089536615806,
-      "grad_norm": 0.003849920816719532,
-      "learning_rate": 0.00019998871400366572,
-      "loss": 46.0,
-      "step": 29707
-    },
-    {
-      "epoch": 4.784250573694593,
-      "grad_norm": 0.01764640025794506,
-      "learning_rate": 0.0001999887132435907,
-      "loss": 46.0,
-      "step": 29708
-    },
-    {
-      "epoch": 4.784411610773381,
-      "grad_norm": 0.005238171201199293,
-      "learning_rate": 0.00019998871248349008,
-      "loss": 46.0,
-      "step": 29709
-    },
-    {
-      "epoch": 4.784572647852168,
-      "grad_norm": 0.004160532262176275,
-      "learning_rate": 0.00019998871172336388,
-      "loss": 46.0,
-      "step": 29710
-    },
-    {
-      "epoch": 4.784733684930956,
-      "grad_norm": 0.002927604829892516,
-      "learning_rate": 0.00019998871096321206,
-      "loss": 46.0,
-      "step": 29711
-    },
-    {
-      "epoch": 4.784894722009743,
-      "grad_norm": 0.005752460565418005,
-      "learning_rate": 0.00019998871020303466,
-      "loss": 46.0,
-      "step": 29712
-    },
-    {
-      "epoch": 4.7850557590885305,
-      "grad_norm": 0.0006801657727919519,
-      "learning_rate": 0.00019998870944283167,
-      "loss": 46.0,
-      "step": 29713
-    },
-    {
-      "epoch": 4.785216796167317,
-      "grad_norm": 0.0007681837887503207,
-      "learning_rate": 0.00019998870868260312,
-      "loss": 46.0,
-      "step": 29714
-    },
-    {
-      "epoch": 4.7853778332461046,
-      "grad_norm": 0.0038328443188220263,
-      "learning_rate": 0.00019998870792234893,
-      "loss": 46.0,
-      "step": 29715
-    },
-    {
-      "epoch": 4.785538870324892,
-      "grad_norm": 0.004479486029595137,
-      "learning_rate": 0.00019998870716206917,
-      "loss": 46.0,
-      "step": 29716
-    },
-    {
-      "epoch": 4.7856999074036795,
-      "grad_norm": 0.003548318985849619,
-      "learning_rate": 0.0001999887064017638,
-      "loss": 46.0,
-      "step": 29717
-    },
-    {
-      "epoch": 4.785860944482467,
-      "grad_norm": 0.0008701508049853146,
-      "learning_rate": 0.00019998870564143286,
-      "loss": 46.0,
-      "step": 29718
-    },
-    {
-      "epoch": 4.786021981561254,
-      "grad_norm": 0.0025080766063183546,
-      "learning_rate": 0.00019998870488107632,
-      "loss": 46.0,
-      "step": 29719
-    },
-    {
-      "epoch": 4.786183018640042,
-      "grad_norm": 0.008069365285336971,
-      "learning_rate": 0.0001999887041206942,
-      "loss": 46.0,
-      "step": 29720
-    },
-    {
-      "epoch": 4.786344055718829,
-      "grad_norm": 0.0009555853321217,
-      "learning_rate": 0.00019998870336028647,
-      "loss": 46.0,
-      "step": 29721
-    },
-    {
-      "epoch": 4.786505092797617,
-      "grad_norm": 0.0037297927774488926,
-      "learning_rate": 0.00019998870259985317,
-      "loss": 46.0,
-      "step": 29722
-    },
-    {
-      "epoch": 4.786666129876404,
-      "grad_norm": 0.007247936446219683,
-      "learning_rate": 0.00019998870183939426,
-      "loss": 46.0,
-      "step": 29723
-    },
-    {
-      "epoch": 4.786827166955192,
-      "grad_norm": 0.008992508985102177,
-      "learning_rate": 0.00019998870107890975,
-      "loss": 46.0,
-      "step": 29724
-    },
-    {
-      "epoch": 4.786988204033979,
-      "grad_norm": 0.001904545584693551,
-      "learning_rate": 0.00019998870031839964,
-      "loss": 46.0,
-      "step": 29725
-    },
-    {
-      "epoch": 4.7871492411127665,
-      "grad_norm": 0.022919410839676857,
-      "learning_rate": 0.00019998869955786396,
-      "loss": 46.0,
-      "step": 29726
-    },
-    {
-      "epoch": 4.787310278191554,
-      "grad_norm": 0.0012247557751834393,
-      "learning_rate": 0.00019998869879730272,
-      "loss": 46.0,
-      "step": 29727
-    },
-    {
-      "epoch": 4.787471315270341,
-      "grad_norm": 0.0011363471858203411,
-      "learning_rate": 0.00019998869803671581,
-      "loss": 46.0,
-      "step": 29728
-    },
-    {
-      "epoch": 4.787632352349128,
-      "grad_norm": 0.00351770780980587,
-      "learning_rate": 0.00019998869727610338,
-      "loss": 46.0,
-      "step": 29729
-    },
-    {
-      "epoch": 4.787793389427915,
-      "grad_norm": 0.02019556611776352,
-      "learning_rate": 0.00019998869651546532,
-      "loss": 46.0,
-      "step": 29730
-    },
-    {
-      "epoch": 4.787954426506703,
-      "grad_norm": 0.0014853174798190594,
-      "learning_rate": 0.00019998869575480168,
-      "loss": 46.0,
-      "step": 29731
-    },
-    {
-      "epoch": 4.78811546358549,
-      "grad_norm": 0.0020423715468496084,
-      "learning_rate": 0.00019998869499411245,
-      "loss": 46.0,
-      "step": 29732
-    },
-    {
-      "epoch": 4.788276500664278,
-      "grad_norm": 0.011394595727324486,
-      "learning_rate": 0.00019998869423339764,
-      "loss": 46.0,
-      "step": 29733
-    },
-    {
-      "epoch": 4.788437537743065,
-      "grad_norm": 0.004124745726585388,
-      "learning_rate": 0.0001999886934726572,
-      "loss": 46.0,
-      "step": 29734
-    },
-    {
-      "epoch": 4.788598574821853,
-      "grad_norm": 0.0058473628014326096,
-      "learning_rate": 0.0001999886927118912,
-      "loss": 46.0,
-      "step": 29735
-    },
-    {
-      "epoch": 4.78875961190064,
-      "grad_norm": 0.005639522336423397,
-      "learning_rate": 0.0001999886919510996,
-      "loss": 46.0,
-      "step": 29736
-    },
-    {
-      "epoch": 4.788920648979428,
-      "grad_norm": 0.00618215836584568,
-      "learning_rate": 0.0001999886911902824,
-      "loss": 46.0,
-      "step": 29737
-    },
-    {
-      "epoch": 4.789081686058215,
-      "grad_norm": 0.0023963251151144505,
-      "learning_rate": 0.0001999886904294396,
-      "loss": 46.0,
-      "step": 29738
-    },
-    {
-      "epoch": 4.7892427231370025,
-      "grad_norm": 0.008571038953959942,
-      "learning_rate": 0.00019998868966857123,
-      "loss": 46.0,
-      "step": 29739
-    },
-    {
-      "epoch": 4.78940376021579,
-      "grad_norm": 0.004283445421606302,
-      "learning_rate": 0.00019998868890767725,
-      "loss": 46.0,
-      "step": 29740
-    },
-    {
-      "epoch": 4.789564797294577,
-      "grad_norm": 0.003792413044720888,
-      "learning_rate": 0.00019998868814675769,
-      "loss": 46.0,
-      "step": 29741
-    },
-    {
-      "epoch": 4.789725834373365,
-      "grad_norm": 0.001151659176684916,
-      "learning_rate": 0.00019998868738581256,
-      "loss": 46.0,
-      "step": 29742
-    },
-    {
-      "epoch": 4.789886871452152,
-      "grad_norm": 0.006588222924619913,
-      "learning_rate": 0.0001999886866248418,
-      "loss": 46.0,
-      "step": 29743
-    },
-    {
-      "epoch": 4.790047908530939,
-      "grad_norm": 0.00819449219852686,
-      "learning_rate": 0.00019998868586384546,
-      "loss": 46.0,
-      "step": 29744
-    },
-    {
-      "epoch": 4.790208945609726,
-      "grad_norm": 0.00936033483594656,
-      "learning_rate": 0.00019998868510282352,
-      "loss": 46.0,
-      "step": 29745
-    },
-    {
-      "epoch": 4.790369982688514,
-      "grad_norm": 0.006183766294270754,
-      "learning_rate": 0.000199988684341776,
-      "loss": 46.0,
-      "step": 29746
-    },
-    {
-      "epoch": 4.790531019767301,
-      "grad_norm": 0.009324228391051292,
-      "learning_rate": 0.00019998868358070288,
-      "loss": 46.0,
-      "step": 29747
-    },
-    {
-      "epoch": 4.790692056846089,
-      "grad_norm": 0.008201288990676403,
-      "learning_rate": 0.0001999886828196042,
-      "loss": 46.0,
-      "step": 29748
-    },
-    {
-      "epoch": 4.790853093924876,
-      "grad_norm": 0.0046630133874714375,
-      "learning_rate": 0.00019998868205847988,
-      "loss": 46.0,
-      "step": 29749
-    },
-    {
-      "epoch": 4.791014131003664,
-      "grad_norm": 0.007685188669711351,
-      "learning_rate": 0.00019998868129733,
-      "loss": 46.0,
-      "step": 29750
-    },
-    {
-      "epoch": 4.791175168082451,
-      "grad_norm": 0.013726707547903061,
-      "learning_rate": 0.0001999886805361545,
-      "loss": 46.0,
-      "step": 29751
-    },
-    {
-      "epoch": 4.7913362051612385,
-      "grad_norm": 0.0086591225117445,
-      "learning_rate": 0.00019998867977495343,
-      "loss": 46.0,
-      "step": 29752
-    },
-    {
-      "epoch": 4.791497242240026,
-      "grad_norm": 0.01896575838327408,
-      "learning_rate": 0.00019998867901372677,
-      "loss": 46.0,
-      "step": 29753
-    },
-    {
-      "epoch": 4.791658279318813,
-      "grad_norm": 0.005633487366139889,
-      "learning_rate": 0.0001999886782524745,
-      "loss": 46.0,
-      "step": 29754
-    },
-    {
-      "epoch": 4.791819316397601,
-      "grad_norm": 0.005812176037579775,
-      "learning_rate": 0.00019998867749119667,
-      "loss": 46.0,
-      "step": 29755
-    },
-    {
-      "epoch": 4.791980353476388,
-      "grad_norm": 0.00278187426738441,
-      "learning_rate": 0.00019998867672989322,
-      "loss": 46.0,
-      "step": 29756
-    },
-    {
-      "epoch": 4.792141390555175,
-      "grad_norm": 0.008168050087988377,
-      "learning_rate": 0.00019998867596856417,
-      "loss": 46.0,
-      "step": 29757
-    },
-    {
-      "epoch": 4.792302427633963,
-      "grad_norm": 0.007299512624740601,
-      "learning_rate": 0.00019998867520720957,
-      "loss": 46.0,
-      "step": 29758
-    },
-    {
-      "epoch": 4.79246346471275,
-      "grad_norm": 0.011872775852680206,
-      "learning_rate": 0.00019998867444582933,
-      "loss": 46.0,
-      "step": 29759
-    },
-    {
-      "epoch": 4.792624501791537,
-      "grad_norm": 0.0030426299199461937,
-      "learning_rate": 0.00019998867368442352,
-      "loss": 46.0,
-      "step": 29760
-    },
-    {
-      "epoch": 4.792785538870325,
-      "grad_norm": 0.0028889866080135107,
-      "learning_rate": 0.00019998867292299216,
-      "loss": 46.0,
-      "step": 29761
-    },
-    {
-      "epoch": 4.792946575949112,
-      "grad_norm": 0.0021509702783077955,
-      "learning_rate": 0.00019998867216153513,
-      "loss": 46.0,
-      "step": 29762
-    },
-    {
-      "epoch": 4.7931076130279,
-      "grad_norm": 0.0053243329748511314,
-      "learning_rate": 0.00019998867140005256,
-      "loss": 46.0,
-      "step": 29763
-    },
-    {
-      "epoch": 4.793268650106687,
-      "grad_norm": 0.0022179947700351477,
-      "learning_rate": 0.00019998867063854436,
-      "loss": 46.0,
-      "step": 29764
-    },
-    {
-      "epoch": 4.7934296871854745,
-      "grad_norm": 0.0023544016294181347,
-      "learning_rate": 0.0001999886698770106,
-      "loss": 46.0,
-      "step": 29765
-    },
-    {
-      "epoch": 4.793590724264262,
-      "grad_norm": 0.003395197680220008,
-      "learning_rate": 0.00019998866911545123,
-      "loss": 46.0,
-      "step": 29766
-    },
-    {
-      "epoch": 4.793751761343049,
-      "grad_norm": 0.0029666482005268335,
-      "learning_rate": 0.0001999886683538663,
-      "loss": 46.0,
-      "step": 29767
-    },
-    {
-      "epoch": 4.793912798421837,
-      "grad_norm": 0.004383180756121874,
-      "learning_rate": 0.00019998866759225574,
-      "loss": 46.0,
-      "step": 29768
-    },
-    {
-      "epoch": 4.794073835500624,
-      "grad_norm": 0.016333680599927902,
-      "learning_rate": 0.0001999886668306196,
-      "loss": 46.0,
-      "step": 29769
-    },
-    {
-      "epoch": 4.794234872579412,
-      "grad_norm": 0.009403948672115803,
-      "learning_rate": 0.00019998866606895787,
-      "loss": 46.0,
-      "step": 29770
-    },
-    {
-      "epoch": 4.794395909658199,
-      "grad_norm": 0.004911371041089296,
-      "learning_rate": 0.00019998866530727055,
-      "loss": 46.0,
-      "step": 29771
-    },
-    {
-      "epoch": 4.794556946736986,
-      "grad_norm": 0.004054257180541754,
-      "learning_rate": 0.00019998866454555764,
-      "loss": 46.0,
-      "step": 29772
-    },
-    {
-      "epoch": 4.794717983815774,
-      "grad_norm": 0.0054565006867051125,
-      "learning_rate": 0.00019998866378381913,
-      "loss": 46.0,
-      "step": 29773
-    },
-    {
-      "epoch": 4.794879020894561,
-      "grad_norm": 0.012302064336836338,
-      "learning_rate": 0.00019998866302205505,
-      "loss": 46.0,
-      "step": 29774
-    },
-    {
-      "epoch": 4.795040057973348,
-      "grad_norm": 0.0037337648682296276,
-      "learning_rate": 0.00019998866226026536,
-      "loss": 46.0,
-      "step": 29775
-    },
-    {
-      "epoch": 4.7952010950521355,
-      "grad_norm": 0.005812222603708506,
-      "learning_rate": 0.00019998866149845008,
-      "loss": 46.0,
-      "step": 29776
-    },
-    {
-      "epoch": 4.795362132130923,
-      "grad_norm": 0.002526002237573266,
-      "learning_rate": 0.00019998866073660918,
-      "loss": 46.0,
-      "step": 29777
-    },
-    {
-      "epoch": 4.7955231692097104,
-      "grad_norm": 0.002947915345430374,
-      "learning_rate": 0.00019998865997474275,
-      "loss": 46.0,
-      "step": 29778
-    },
-    {
-      "epoch": 4.795684206288498,
-      "grad_norm": 0.010503734461963177,
-      "learning_rate": 0.00019998865921285069,
-      "loss": 46.0,
-      "step": 29779
-    },
-    {
-      "epoch": 4.795845243367285,
-      "grad_norm": 0.014623298309743404,
-      "learning_rate": 0.00019998865845093303,
-      "loss": 46.0,
-      "step": 29780
-    },
-    {
-      "epoch": 4.796006280446073,
-      "grad_norm": 0.004749680869281292,
-      "learning_rate": 0.0001999886576889898,
-      "loss": 46.0,
-      "step": 29781
-    },
-    {
-      "epoch": 4.79616731752486,
-      "grad_norm": 0.02176261506974697,
-      "learning_rate": 0.00019998865692702093,
-      "loss": 46.0,
-      "step": 29782
-    },
-    {
-      "epoch": 4.796328354603648,
-      "grad_norm": 0.0008449893211945891,
-      "learning_rate": 0.0001999886561650265,
-      "loss": 46.0,
-      "step": 29783
-    },
-    {
-      "epoch": 4.796489391682435,
-      "grad_norm": 0.010199923068284988,
-      "learning_rate": 0.0001999886554030065,
-      "loss": 46.0,
-      "step": 29784
-    },
-    {
-      "epoch": 4.796650428761223,
-      "grad_norm": 0.0008555551758036017,
-      "learning_rate": 0.0001999886546409609,
-      "loss": 46.0,
-      "step": 29785
-    },
-    {
-      "epoch": 4.79681146584001,
-      "grad_norm": 0.02349877543747425,
-      "learning_rate": 0.0001999886538788897,
-      "loss": 46.0,
-      "step": 29786
-    },
-    {
-      "epoch": 4.796972502918797,
-      "grad_norm": 0.011679049581289291,
-      "learning_rate": 0.0001999886531167929,
-      "loss": 46.0,
-      "step": 29787
-    },
-    {
-      "epoch": 4.797133539997584,
-      "grad_norm": 0.0020645330660045147,
-      "learning_rate": 0.00019998865235467054,
-      "loss": 46.0,
-      "step": 29788
-    },
-    {
-      "epoch": 4.7972945770763715,
-      "grad_norm": 0.004109121859073639,
-      "learning_rate": 0.00019998865159252254,
-      "loss": 46.0,
-      "step": 29789
-    },
-    {
-      "epoch": 4.797455614155159,
-      "grad_norm": 0.0012318691005930305,
-      "learning_rate": 0.00019998865083034896,
-      "loss": 46.0,
-      "step": 29790
-    },
-    {
-      "epoch": 4.797616651233946,
-      "grad_norm": 0.01024674903601408,
-      "learning_rate": 0.00019998865006814982,
-      "loss": 46.0,
-      "step": 29791
-    },
-    {
-      "epoch": 4.797777688312734,
-      "grad_norm": 0.0024883365258574486,
-      "learning_rate": 0.00019998864930592504,
-      "loss": 46.0,
-      "step": 29792
-    },
-    {
-      "epoch": 4.797938725391521,
-      "grad_norm": 0.00452685821801424,
-      "learning_rate": 0.00019998864854367472,
-      "loss": 46.0,
-      "step": 29793
-    },
-    {
-      "epoch": 4.798099762470309,
-      "grad_norm": 0.004877728410065174,
-      "learning_rate": 0.00019998864778139876,
-      "loss": 46.0,
-      "step": 29794
-    },
-    {
-      "epoch": 4.798260799549096,
-      "grad_norm": 0.0031970189884305,
-      "learning_rate": 0.00019998864701909725,
-      "loss": 46.0,
-      "step": 29795
-    },
-    {
-      "epoch": 4.798421836627884,
-      "grad_norm": 0.002675888128578663,
-      "learning_rate": 0.00019998864625677011,
-      "loss": 46.0,
-      "step": 29796
-    },
-    {
-      "epoch": 4.798582873706671,
-      "grad_norm": 0.0044942288659513,
-      "learning_rate": 0.0001999886454944174,
-      "loss": 46.0,
-      "step": 29797
-    },
-    {
-      "epoch": 4.798743910785459,
-      "grad_norm": 0.018630048260092735,
-      "learning_rate": 0.00019998864473203912,
-      "loss": 46.0,
-      "step": 29798
-    },
-    {
-      "epoch": 4.798904947864246,
-      "grad_norm": 0.0017461758106946945,
-      "learning_rate": 0.00019998864396963522,
-      "loss": 46.0,
-      "step": 29799
-    },
-    {
-      "epoch": 4.7990659849430335,
-      "grad_norm": 0.0017961953999474645,
-      "learning_rate": 0.00019998864320720574,
-      "loss": 46.0,
-      "step": 29800
-    },
-    {
-      "epoch": 4.799227022021821,
-      "grad_norm": 0.003009121399372816,
-      "learning_rate": 0.00019998864244475065,
-      "loss": 46.0,
-      "step": 29801
-    },
-    {
-      "epoch": 4.7993880591006075,
-      "grad_norm": 0.00737546943128109,
-      "learning_rate": 0.00019998864168226997,
-      "loss": 46.0,
-      "step": 29802
-    },
-    {
-      "epoch": 4.799549096179395,
-      "grad_norm": 0.0007850078400224447,
-      "learning_rate": 0.0001999886409197637,
-      "loss": 46.0,
-      "step": 29803
-    },
-    {
-      "epoch": 4.799710133258182,
-      "grad_norm": 0.0029610597994178534,
-      "learning_rate": 0.00019998864015723187,
-      "loss": 46.0,
-      "step": 29804
-    },
-    {
-      "epoch": 4.79987117033697,
-      "grad_norm": 0.0028766151517629623,
-      "learning_rate": 0.0001999886393946744,
-      "loss": 46.0,
-      "step": 29805
-    },
-    {
-      "epoch": 4.800032207415757,
-      "grad_norm": 0.003073472296819091,
-      "learning_rate": 0.00019998863863209137,
-      "loss": 46.0,
-      "step": 29806
-    },
-    {
-      "epoch": 4.800193244494545,
-      "grad_norm": 0.014405116438865662,
-      "learning_rate": 0.00019998863786948272,
-      "loss": 46.0,
-      "step": 29807
-    },
-    {
-      "epoch": 4.800354281573332,
-      "grad_norm": 0.005622957367449999,
-      "learning_rate": 0.0001999886371068485,
-      "loss": 46.0,
-      "step": 29808
-    },
-    {
-      "epoch": 4.80051531865212,
-      "grad_norm": 0.0038253404200077057,
-      "learning_rate": 0.0001999886363441887,
-      "loss": 46.0,
-      "step": 29809
-    },
-    {
-      "epoch": 4.800676355730907,
-      "grad_norm": 0.002177529502660036,
-      "learning_rate": 0.00019998863558150326,
-      "loss": 46.0,
-      "step": 29810
-    },
-    {
-      "epoch": 4.800837392809695,
-      "grad_norm": 0.0028617591597139835,
-      "learning_rate": 0.0001999886348187923,
-      "loss": 46.0,
-      "step": 29811
-    },
-    {
-      "epoch": 4.800998429888482,
-      "grad_norm": 0.003431614488363266,
-      "learning_rate": 0.0001999886340560557,
-      "loss": 46.0,
-      "step": 29812
-    },
-    {
-      "epoch": 4.8011594669672695,
-      "grad_norm": 0.001781382830813527,
-      "learning_rate": 0.00019998863329329352,
-      "loss": 46.0,
-      "step": 29813
-    },
-    {
-      "epoch": 4.801320504046057,
-      "grad_norm": 0.004645872861146927,
-      "learning_rate": 0.00019998863253050574,
-      "loss": 46.0,
-      "step": 29814
-    },
-    {
-      "epoch": 4.801481541124844,
-      "grad_norm": 0.0049195787869393826,
-      "learning_rate": 0.00019998863176769237,
-      "loss": 46.0,
-      "step": 29815
-    },
-    {
-      "epoch": 4.801642578203632,
-      "grad_norm": 0.004815096501260996,
-      "learning_rate": 0.00019998863100485341,
-      "loss": 46.0,
-      "step": 29816
-    },
-    {
-      "epoch": 4.801803615282418,
-      "grad_norm": 0.003890766529366374,
-      "learning_rate": 0.00019998863024198884,
-      "loss": 46.0,
-      "step": 29817
-    },
-    {
-      "epoch": 4.801964652361206,
-      "grad_norm": 0.008310000412166119,
-      "learning_rate": 0.0001999886294790987,
-      "loss": 46.0,
-      "step": 29818
-    },
-    {
-      "epoch": 4.802125689439993,
-      "grad_norm": 0.008424951694905758,
-      "learning_rate": 0.00019998862871618297,
-      "loss": 46.0,
-      "step": 29819
-    },
-    {
-      "epoch": 4.802286726518781,
-      "grad_norm": 0.006671548821032047,
-      "learning_rate": 0.00019998862795324163,
-      "loss": 46.0,
-      "step": 29820
-    },
-    {
-      "epoch": 4.802447763597568,
-      "grad_norm": 0.0012449411442503333,
-      "learning_rate": 0.00019998862719027472,
-      "loss": 46.0,
-      "step": 29821
-    },
-    {
-      "epoch": 4.802608800676356,
-      "grad_norm": 0.004106082487851381,
-      "learning_rate": 0.0001999886264272822,
-      "loss": 46.0,
-      "step": 29822
-    },
-    {
-      "epoch": 4.802769837755143,
-      "grad_norm": 0.004259491804987192,
-      "learning_rate": 0.00019998862566426411,
-      "loss": 46.0,
-      "step": 29823
-    },
-    {
-      "epoch": 4.8029308748339306,
-      "grad_norm": 0.004347400274127722,
-      "learning_rate": 0.0001999886249012204,
-      "loss": 46.0,
-      "step": 29824
-    },
-    {
-      "epoch": 4.803091911912718,
-      "grad_norm": 0.008030601777136326,
-      "learning_rate": 0.00019998862413815114,
-      "loss": 46.0,
-      "step": 29825
-    },
-    {
-      "epoch": 4.8032529489915055,
-      "grad_norm": 0.005460289306938648,
-      "learning_rate": 0.00019998862337505626,
-      "loss": 46.0,
-      "step": 29826
-    },
-    {
-      "epoch": 4.803413986070293,
-      "grad_norm": 0.0029346952214837074,
-      "learning_rate": 0.00019998862261193576,
-      "loss": 46.0,
-      "step": 29827
-    },
-    {
-      "epoch": 4.80357502314908,
-      "grad_norm": 0.006770648993551731,
-      "learning_rate": 0.00019998862184878973,
-      "loss": 46.0,
-      "step": 29828
-    },
-    {
-      "epoch": 4.803736060227868,
-      "grad_norm": 0.003570184577256441,
-      "learning_rate": 0.00019998862108561806,
-      "loss": 46.0,
-      "step": 29829
-    },
-    {
-      "epoch": 4.803897097306654,
-      "grad_norm": 0.001085524563677609,
-      "learning_rate": 0.0001999886203224208,
-      "loss": 46.0,
-      "step": 29830
-    },
-    {
-      "epoch": 4.804058134385443,
-      "grad_norm": 0.004232665058225393,
-      "learning_rate": 0.00019998861955919796,
-      "loss": 46.0,
-      "step": 29831
-    },
-    {
-      "epoch": 4.804219171464229,
-      "grad_norm": 0.001190761337056756,
-      "learning_rate": 0.00019998861879594952,
-      "loss": 46.0,
-      "step": 29832
-    },
-    {
-      "epoch": 4.804380208543017,
-      "grad_norm": 0.0029191027861088514,
-      "learning_rate": 0.0001999886180326755,
-      "loss": 46.0,
-      "step": 29833
-    },
-    {
-      "epoch": 4.804541245621804,
-      "grad_norm": 0.004711335524916649,
-      "learning_rate": 0.0001999886172693759,
-      "loss": 46.0,
-      "step": 29834
-    },
-    {
-      "epoch": 4.804702282700592,
-      "grad_norm": 0.002555497922003269,
-      "learning_rate": 0.0001999886165060507,
-      "loss": 46.0,
-      "step": 29835
-    },
-    {
-      "epoch": 4.804863319779379,
-      "grad_norm": 0.0087977834045887,
-      "learning_rate": 0.0001999886157426999,
-      "loss": 46.0,
-      "step": 29836
-    },
-    {
-      "epoch": 4.8050243568581665,
-      "grad_norm": 0.004582275170832872,
-      "learning_rate": 0.0001999886149793235,
-      "loss": 46.0,
-      "step": 29837
-    },
-    {
-      "epoch": 4.805185393936954,
-      "grad_norm": 0.01322112139314413,
-      "learning_rate": 0.00019998861421592152,
-      "loss": 46.0,
-      "step": 29838
-    },
-    {
-      "epoch": 4.805346431015741,
-      "grad_norm": 0.0019608517177402973,
-      "learning_rate": 0.00019998861345249395,
-      "loss": 46.0,
-      "step": 29839
-    },
-    {
-      "epoch": 4.805507468094529,
-      "grad_norm": 0.002760443137958646,
-      "learning_rate": 0.00019998861268904076,
-      "loss": 46.0,
-      "step": 29840
-    },
-    {
-      "epoch": 4.805668505173316,
-      "grad_norm": 0.001646880991756916,
-      "learning_rate": 0.00019998861192556202,
-      "loss": 46.0,
-      "step": 29841
-    },
-    {
-      "epoch": 4.805829542252104,
-      "grad_norm": 0.007906120270490646,
-      "learning_rate": 0.00019998861116205766,
-      "loss": 46.0,
-      "step": 29842
-    },
-    {
-      "epoch": 4.805990579330891,
-      "grad_norm": 0.02050287462770939,
-      "learning_rate": 0.0001999886103985277,
-      "loss": 46.0,
-      "step": 29843
-    },
-    {
-      "epoch": 4.806151616409679,
-      "grad_norm": 0.01248864084482193,
-      "learning_rate": 0.00019998860963497218,
-      "loss": 46.0,
-      "step": 29844
-    },
-    {
-      "epoch": 4.806312653488465,
-      "grad_norm": 0.004966906271874905,
-      "learning_rate": 0.00019998860887139106,
-      "loss": 46.0,
-      "step": 29845
-    },
-    {
-      "epoch": 4.806473690567254,
-      "grad_norm": 0.0009638596093282104,
-      "learning_rate": 0.00019998860810778435,
-      "loss": 46.0,
-      "step": 29846
-    },
-    {
-      "epoch": 4.80663472764604,
-      "grad_norm": 0.0024966069031506777,
-      "learning_rate": 0.00019998860734415203,
-      "loss": 46.0,
-      "step": 29847
-    },
-    {
-      "epoch": 4.806795764724828,
-      "grad_norm": 0.018226562067866325,
-      "learning_rate": 0.00019998860658049412,
-      "loss": 46.0,
-      "step": 29848
-    },
-    {
-      "epoch": 4.806956801803615,
-      "grad_norm": 0.007932223379611969,
-      "learning_rate": 0.00019998860581681065,
-      "loss": 46.0,
-      "step": 29849
-    },
-    {
-      "epoch": 4.8071178388824025,
-      "grad_norm": 0.007319622207432985,
-      "learning_rate": 0.00019998860505310154,
-      "loss": 46.0,
-      "step": 29850
-    },
-    {
-      "epoch": 4.80727887596119,
-      "grad_norm": 0.0018009509658440948,
-      "learning_rate": 0.00019998860428936687,
-      "loss": 46.0,
-      "step": 29851
-    },
-    {
-      "epoch": 4.807439913039977,
-      "grad_norm": 0.01569865643978119,
-      "learning_rate": 0.0001999886035256066,
-      "loss": 46.0,
-      "step": 29852
-    },
-    {
-      "epoch": 4.807600950118765,
-      "grad_norm": 0.0026742275804281235,
-      "learning_rate": 0.00019998860276182075,
-      "loss": 46.0,
-      "step": 29853
-    },
-    {
-      "epoch": 4.807761987197552,
-      "grad_norm": 0.004862932953983545,
-      "learning_rate": 0.0001999886019980093,
-      "loss": 46.0,
-      "step": 29854
-    },
-    {
-      "epoch": 4.80792302427634,
-      "grad_norm": 0.0064040180295705795,
-      "learning_rate": 0.00019998860123417224,
-      "loss": 46.0,
-      "step": 29855
-    },
-    {
-      "epoch": 4.808084061355127,
-      "grad_norm": 0.008049490861594677,
-      "learning_rate": 0.00019998860047030958,
-      "loss": 46.0,
-      "step": 29856
-    },
-    {
-      "epoch": 4.808245098433915,
-      "grad_norm": 0.005625725258141756,
-      "learning_rate": 0.0001999885997064214,
-      "loss": 46.0,
-      "step": 29857
-    },
-    {
-      "epoch": 4.808406135512702,
-      "grad_norm": 0.001973808743059635,
-      "learning_rate": 0.00019998859894250755,
-      "loss": 46.0,
-      "step": 29858
-    },
-    {
-      "epoch": 4.80856717259149,
-      "grad_norm": 0.01069825142621994,
-      "learning_rate": 0.00019998859817856813,
-      "loss": 46.0,
-      "step": 29859
-    },
-    {
-      "epoch": 4.808728209670276,
-      "grad_norm": 0.0028028555680066347,
-      "learning_rate": 0.00019998859741460315,
-      "loss": 46.0,
-      "step": 29860
-    },
-    {
-      "epoch": 4.808889246749064,
-      "grad_norm": 0.0026830544229596853,
-      "learning_rate": 0.00019998859665061253,
-      "loss": 46.0,
-      "step": 29861
-    },
-    {
-      "epoch": 4.809050283827851,
-      "grad_norm": 0.00322624109685421,
-      "learning_rate": 0.00019998859588659634,
-      "loss": 46.0,
-      "step": 29862
-    },
-    {
-      "epoch": 4.8092113209066385,
-      "grad_norm": 0.0058891805820167065,
-      "learning_rate": 0.00019998859512255455,
-      "loss": 46.0,
-      "step": 29863
-    },
-    {
-      "epoch": 4.809372357985426,
-      "grad_norm": 0.010692720301449299,
-      "learning_rate": 0.0001999885943584872,
-      "loss": 46.0,
-      "step": 29864
-    },
-    {
-      "epoch": 4.809533395064213,
-      "grad_norm": 0.0008916780934669077,
-      "learning_rate": 0.00019998859359439424,
-      "loss": 46.0,
-      "step": 29865
-    },
-    {
-      "epoch": 4.809694432143001,
-      "grad_norm": 0.0034505638759583235,
-      "learning_rate": 0.00019998859283027566,
-      "loss": 46.0,
-      "step": 29866
-    },
-    {
-      "epoch": 4.809855469221788,
-      "grad_norm": 0.00544386263936758,
-      "learning_rate": 0.0001999885920661315,
-      "loss": 46.0,
-      "step": 29867
-    },
-    {
-      "epoch": 4.810016506300576,
-      "grad_norm": 0.011903595179319382,
-      "learning_rate": 0.00019998859130196178,
-      "loss": 46.0,
-      "step": 29868
-    },
-    {
-      "epoch": 4.810177543379363,
-      "grad_norm": 0.008608249947428703,
-      "learning_rate": 0.00019998859053776643,
-      "loss": 46.0,
-      "step": 29869
-    },
-    {
-      "epoch": 4.810338580458151,
-      "grad_norm": 0.005572275724261999,
-      "learning_rate": 0.00019998858977354552,
-      "loss": 46.0,
-      "step": 29870
-    },
-    {
-      "epoch": 4.810499617536938,
-      "grad_norm": 0.002035081386566162,
-      "learning_rate": 0.000199988589009299,
-      "loss": 46.0,
-      "step": 29871
-    },
-    {
-      "epoch": 4.810660654615726,
-      "grad_norm": 0.012300747446715832,
-      "learning_rate": 0.0001999885882450269,
-      "loss": 46.0,
-      "step": 29872
-    },
-    {
-      "epoch": 4.810821691694513,
-      "grad_norm": 0.005571066867560148,
-      "learning_rate": 0.0001999885874807292,
-      "loss": 46.0,
-      "step": 29873
-    },
-    {
-      "epoch": 4.8109827287733005,
-      "grad_norm": 0.0026481610257178545,
-      "learning_rate": 0.00019998858671640588,
-      "loss": 46.0,
-      "step": 29874
-    },
-    {
-      "epoch": 4.811143765852087,
-      "grad_norm": 0.01004584040492773,
-      "learning_rate": 0.000199988585952057,
-      "loss": 46.0,
-      "step": 29875
-    },
-    {
-      "epoch": 4.8113048029308745,
-      "grad_norm": 0.005056702997535467,
-      "learning_rate": 0.00019998858518768253,
-      "loss": 46.0,
-      "step": 29876
-    },
-    {
-      "epoch": 4.811465840009662,
-      "grad_norm": 0.007630080915987492,
-      "learning_rate": 0.00019998858442328246,
-      "loss": 46.0,
-      "step": 29877
-    },
-    {
-      "epoch": 4.811626877088449,
-      "grad_norm": 0.01838238537311554,
-      "learning_rate": 0.0001999885836588568,
-      "loss": 46.0,
-      "step": 29878
-    },
-    {
-      "epoch": 4.811787914167237,
-      "grad_norm": 0.01748710870742798,
-      "learning_rate": 0.00019998858289440555,
-      "loss": 46.0,
-      "step": 29879
-    },
-    {
-      "epoch": 4.811948951246024,
-      "grad_norm": 0.002137694275006652,
-      "learning_rate": 0.0001999885821299287,
-      "loss": 46.0,
-      "step": 29880
-    },
-    {
-      "epoch": 4.812109988324812,
-      "grad_norm": 0.007220414467155933,
-      "learning_rate": 0.00019998858136542627,
-      "loss": 46.0,
-      "step": 29881
-    },
-    {
-      "epoch": 4.812271025403599,
-      "grad_norm": 0.0009207912953570485,
-      "learning_rate": 0.00019998858060089823,
-      "loss": 46.0,
-      "step": 29882
-    },
-    {
-      "epoch": 4.812432062482387,
-      "grad_norm": 0.0033582462929189205,
-      "learning_rate": 0.0001999885798363446,
-      "loss": 46.0,
-      "step": 29883
-    },
-    {
-      "epoch": 4.812593099561174,
-      "grad_norm": 0.0026050391606986523,
-      "learning_rate": 0.0001999885790717654,
-      "loss": 46.0,
-      "step": 29884
-    },
-    {
-      "epoch": 4.8127541366399615,
-      "grad_norm": 0.009508085437119007,
-      "learning_rate": 0.0001999885783071606,
-      "loss": 46.0,
-      "step": 29885
-    },
-    {
-      "epoch": 4.812915173718749,
-      "grad_norm": 0.003897358663380146,
-      "learning_rate": 0.0001999885775425302,
-      "loss": 46.0,
-      "step": 29886
-    },
-    {
-      "epoch": 4.8130762107975364,
-      "grad_norm": 0.008441883139312267,
-      "learning_rate": 0.00019998857677787422,
-      "loss": 46.0,
-      "step": 29887
-    },
-    {
-      "epoch": 4.813237247876324,
-      "grad_norm": 0.0009966617217287421,
-      "learning_rate": 0.00019998857601319264,
-      "loss": 46.0,
-      "step": 29888
-    },
-    {
-      "epoch": 4.813398284955111,
-      "grad_norm": 0.006653158459812403,
-      "learning_rate": 0.00019998857524848544,
-      "loss": 46.0,
-      "step": 29889
-    },
-    {
-      "epoch": 4.813559322033898,
-      "grad_norm": 0.003060739254578948,
-      "learning_rate": 0.00019998857448375268,
-      "loss": 46.0,
-      "step": 29890
-    },
-    {
-      "epoch": 4.813720359112685,
-      "grad_norm": 0.0009679802460595965,
-      "learning_rate": 0.00019998857371899433,
-      "loss": 46.0,
-      "step": 29891
-    },
-    {
-      "epoch": 4.813881396191473,
-      "grad_norm": 0.0020888003055006266,
-      "learning_rate": 0.00019998857295421037,
-      "loss": 46.0,
-      "step": 29892
-    },
-    {
-      "epoch": 4.81404243327026,
-      "grad_norm": 0.008264970034360886,
-      "learning_rate": 0.00019998857218940085,
-      "loss": 46.0,
-      "step": 29893
-    },
-    {
-      "epoch": 4.814203470349048,
-      "grad_norm": 0.005687571130692959,
-      "learning_rate": 0.00019998857142456572,
-      "loss": 46.0,
-      "step": 29894
-    },
-    {
-      "epoch": 4.814364507427835,
-      "grad_norm": 0.005523642525076866,
-      "learning_rate": 0.00019998857065970497,
-      "loss": 46.0,
-      "step": 29895
-    },
-    {
-      "epoch": 4.814525544506623,
-      "grad_norm": 0.004472050350159407,
-      "learning_rate": 0.00019998856989481868,
-      "loss": 46.0,
-      "step": 29896
-    },
-    {
-      "epoch": 4.81468658158541,
-      "grad_norm": 0.0013236034428700805,
-      "learning_rate": 0.0001999885691299068,
-      "loss": 46.0,
-      "step": 29897
-    },
-    {
-      "epoch": 4.8148476186641975,
-      "grad_norm": 0.005061403848230839,
-      "learning_rate": 0.00019998856836496928,
-      "loss": 46.0,
-      "step": 29898
-    },
-    {
-      "epoch": 4.815008655742985,
-      "grad_norm": 0.0015141351614147425,
-      "learning_rate": 0.00019998856760000618,
-      "loss": 46.0,
-      "step": 29899
-    },
-    {
-      "epoch": 4.815169692821772,
-      "grad_norm": 0.0074175018817186356,
-      "learning_rate": 0.00019998856683501746,
-      "loss": 46.0,
-      "step": 29900
-    },
-    {
-      "epoch": 4.81533072990056,
-      "grad_norm": 0.0039913528598845005,
-      "learning_rate": 0.0001999885660700032,
-      "loss": 46.0,
-      "step": 29901
-    },
-    {
-      "epoch": 4.815491766979347,
-      "grad_norm": 0.0038621709682047367,
-      "learning_rate": 0.00019998856530496336,
-      "loss": 46.0,
-      "step": 29902
-    },
-    {
-      "epoch": 4.815652804058134,
-      "grad_norm": 0.004586962517350912,
-      "learning_rate": 0.00019998856453989789,
-      "loss": 46.0,
-      "step": 29903
-    },
-    {
-      "epoch": 4.815813841136922,
-      "grad_norm": 0.0020976581145077944,
-      "learning_rate": 0.00019998856377480682,
-      "loss": 46.0,
-      "step": 29904
-    },
-    {
-      "epoch": 4.815974878215709,
-      "grad_norm": 0.017379876226186752,
-      "learning_rate": 0.0001999885630096902,
-      "loss": 46.0,
-      "step": 29905
-    },
-    {
-      "epoch": 4.816135915294496,
-      "grad_norm": 0.004537864588201046,
-      "learning_rate": 0.00019998856224454797,
-      "loss": 46.0,
-      "step": 29906
-    },
-    {
-      "epoch": 4.816296952373284,
-      "grad_norm": 0.001742177177220583,
-      "learning_rate": 0.00019998856147938012,
-      "loss": 46.0,
-      "step": 29907
-    },
-    {
-      "epoch": 4.816457989452071,
-      "grad_norm": 0.01234701182693243,
-      "learning_rate": 0.0001999885607141867,
-      "loss": 46.0,
-      "step": 29908
-    },
-    {
-      "epoch": 4.816619026530859,
-      "grad_norm": 0.0011545680463314056,
-      "learning_rate": 0.0001999885599489677,
-      "loss": 46.0,
-      "step": 29909
-    },
-    {
-      "epoch": 4.816780063609646,
-      "grad_norm": 0.011560438200831413,
-      "learning_rate": 0.0001999885591837231,
-      "loss": 46.0,
-      "step": 29910
-    },
-    {
-      "epoch": 4.8169411006884335,
-      "grad_norm": 0.00048677678569220006,
-      "learning_rate": 0.0001999885584184529,
-      "loss": 46.0,
-      "step": 29911
-    },
-    {
-      "epoch": 4.817102137767221,
-      "grad_norm": 0.0067277164198458195,
-      "learning_rate": 0.00019998855765315711,
-      "loss": 46.0,
-      "step": 29912
-    },
-    {
-      "epoch": 4.817263174846008,
-      "grad_norm": 0.0041071283631026745,
-      "learning_rate": 0.00019998855688783574,
-      "loss": 46.0,
-      "step": 29913
-    },
-    {
-      "epoch": 4.817424211924796,
-      "grad_norm": 0.0024829935282468796,
-      "learning_rate": 0.00019998855612248878,
-      "loss": 46.0,
-      "step": 29914
-    },
-    {
-      "epoch": 4.817585249003583,
-      "grad_norm": 0.00551704503595829,
-      "learning_rate": 0.0001999885553571162,
-      "loss": 46.0,
-      "step": 29915
-    },
-    {
-      "epoch": 4.817746286082371,
-      "grad_norm": 0.017143117263913155,
-      "learning_rate": 0.00019998855459171805,
-      "loss": 46.0,
-      "step": 29916
-    },
-    {
-      "epoch": 4.817907323161158,
-      "grad_norm": 0.010960116051137447,
-      "learning_rate": 0.00019998855382629432,
-      "loss": 46.0,
-      "step": 29917
-    },
-    {
-      "epoch": 4.818068360239945,
-      "grad_norm": 0.0022764671593904495,
-      "learning_rate": 0.00019998855306084496,
-      "loss": 46.0,
-      "step": 29918
-    },
-    {
-      "epoch": 4.818229397318733,
-      "grad_norm": 0.008843803778290749,
-      "learning_rate": 0.00019998855229537004,
-      "loss": 46.0,
-      "step": 29919
-    },
-    {
-      "epoch": 4.81839043439752,
-      "grad_norm": 0.005262303631752729,
-      "learning_rate": 0.0001999885515298695,
-      "loss": 46.0,
-      "step": 29920
-    },
-    {
-      "epoch": 4.818551471476307,
-      "grad_norm": 0.0019062585197389126,
-      "learning_rate": 0.0001999885507643434,
-      "loss": 46.0,
-      "step": 29921
-    },
-    {
-      "epoch": 4.818712508555095,
-      "grad_norm": 0.01564912311732769,
-      "learning_rate": 0.0001999885499987917,
-      "loss": 46.0,
-      "step": 29922
-    },
-    {
-      "epoch": 4.818873545633882,
-      "grad_norm": 0.001540483208373189,
-      "learning_rate": 0.0001999885492332144,
-      "loss": 46.0,
-      "step": 29923
-    },
-    {
-      "epoch": 4.8190345827126695,
-      "grad_norm": 0.0037821594160050154,
-      "learning_rate": 0.0001999885484676115,
-      "loss": 46.0,
-      "step": 29924
-    },
-    {
-      "epoch": 4.819195619791457,
-      "grad_norm": 0.0028067342936992645,
-      "learning_rate": 0.00019998854770198303,
-      "loss": 46.0,
-      "step": 29925
-    },
-    {
-      "epoch": 4.819356656870244,
-      "grad_norm": 0.021875912323594093,
-      "learning_rate": 0.00019998854693632895,
-      "loss": 46.0,
-      "step": 29926
-    },
-    {
-      "epoch": 4.819517693949032,
-      "grad_norm": 0.0023984243161976337,
-      "learning_rate": 0.00019998854617064927,
-      "loss": 46.0,
-      "step": 29927
-    },
-    {
-      "epoch": 4.819678731027819,
-      "grad_norm": 0.0029302360489964485,
-      "learning_rate": 0.00019998854540494404,
-      "loss": 46.0,
-      "step": 29928
-    },
-    {
-      "epoch": 4.819839768106607,
-      "grad_norm": 0.0022596800699830055,
-      "learning_rate": 0.00019998854463921316,
-      "loss": 46.0,
-      "step": 29929
-    },
-    {
-      "epoch": 4.820000805185394,
-      "grad_norm": 0.0018841740675270557,
-      "learning_rate": 0.00019998854387345673,
-      "loss": 46.0,
-      "step": 29930
-    },
-    {
-      "epoch": 4.820161842264182,
-      "grad_norm": 0.002793577965348959,
-      "learning_rate": 0.0001999885431076747,
-      "loss": 46.0,
-      "step": 29931
-    },
-    {
-      "epoch": 4.820322879342969,
-      "grad_norm": 0.004283386282622814,
-      "learning_rate": 0.0001999885423418671,
-      "loss": 46.0,
-      "step": 29932
-    },
-    {
-      "epoch": 4.820483916421756,
-      "grad_norm": 0.005727067124098539,
-      "learning_rate": 0.00019998854157603387,
-      "loss": 46.0,
-      "step": 29933
-    },
-    {
-      "epoch": 4.820644953500543,
-      "grad_norm": 0.007798322010785341,
-      "learning_rate": 0.00019998854081017503,
-      "loss": 46.0,
-      "step": 29934
-    },
-    {
-      "epoch": 4.820805990579331,
-      "grad_norm": 0.002008601790294051,
-      "learning_rate": 0.00019998854004429063,
-      "loss": 46.0,
-      "step": 29935
-    },
-    {
-      "epoch": 4.820967027658118,
-      "grad_norm": 0.0019837799482047558,
-      "learning_rate": 0.00019998853927838064,
-      "loss": 46.0,
-      "step": 29936
-    },
-    {
-      "epoch": 4.8211280647369055,
-      "grad_norm": 0.00276742922142148,
-      "learning_rate": 0.00019998853851244507,
-      "loss": 46.0,
-      "step": 29937
-    },
-    {
-      "epoch": 4.821289101815693,
-      "grad_norm": 0.007150628138333559,
-      "learning_rate": 0.0001999885377464839,
-      "loss": 46.0,
-      "step": 29938
-    },
-    {
-      "epoch": 4.82145013889448,
-      "grad_norm": 0.006005027797073126,
-      "learning_rate": 0.0001999885369804971,
-      "loss": 46.0,
-      "step": 29939
-    },
-    {
-      "epoch": 4.821611175973268,
-      "grad_norm": 0.0085380170494318,
-      "learning_rate": 0.00019998853621448474,
-      "loss": 46.0,
-      "step": 29940
-    },
-    {
-      "epoch": 4.821772213052055,
-      "grad_norm": 0.004049663431942463,
-      "learning_rate": 0.0001999885354484468,
-      "loss": 46.0,
-      "step": 29941
-    },
-    {
-      "epoch": 4.821933250130843,
-      "grad_norm": 0.0049303132109344006,
-      "learning_rate": 0.00019998853468238326,
-      "loss": 46.0,
-      "step": 29942
-    },
-    {
-      "epoch": 4.82209428720963,
-      "grad_norm": 0.009177030995488167,
-      "learning_rate": 0.00019998853391629413,
-      "loss": 46.0,
-      "step": 29943
-    },
-    {
-      "epoch": 4.822255324288418,
-      "grad_norm": 0.0006188903353177011,
-      "learning_rate": 0.00019998853315017937,
-      "loss": 46.0,
-      "step": 29944
-    },
-    {
-      "epoch": 4.822416361367205,
-      "grad_norm": 0.006716664414852858,
-      "learning_rate": 0.00019998853238403905,
-      "loss": 46.0,
-      "step": 29945
-    },
-    {
-      "epoch": 4.8225773984459925,
-      "grad_norm": 0.00544294249266386,
-      "learning_rate": 0.00019998853161787316,
-      "loss": 46.0,
-      "step": 29946
-    },
-    {
-      "epoch": 4.82273843552478,
-      "grad_norm": 0.0007383980555459857,
-      "learning_rate": 0.00019998853085168166,
-      "loss": 46.0,
-      "step": 29947
-    },
-    {
-      "epoch": 4.8228994726035666,
-      "grad_norm": 0.0011091517517343163,
-      "learning_rate": 0.00019998853008546457,
-      "loss": 46.0,
-      "step": 29948
-    },
-    {
-      "epoch": 4.823060509682354,
-      "grad_norm": 0.007518439553678036,
-      "learning_rate": 0.00019998852931922185,
-      "loss": 46.0,
-      "step": 29949
-    },
-    {
-      "epoch": 4.8232215467611415,
-      "grad_norm": 0.003972018137574196,
-      "learning_rate": 0.00019998852855295359,
-      "loss": 46.0,
-      "step": 29950
-    },
-    {
-      "epoch": 4.823382583839929,
-      "grad_norm": 0.009267158806324005,
-      "learning_rate": 0.0001999885277866597,
-      "loss": 46.0,
-      "step": 29951
-    },
-    {
-      "epoch": 4.823543620918716,
-      "grad_norm": 0.0057826293632388115,
-      "learning_rate": 0.00019998852702034022,
-      "loss": 46.0,
-      "step": 29952
-    },
-    {
-      "epoch": 4.823704657997504,
-      "grad_norm": 0.008165759034454823,
-      "learning_rate": 0.0001999885262539952,
-      "loss": 46.0,
-      "step": 29953
-    },
-    {
-      "epoch": 4.823865695076291,
-      "grad_norm": 0.006560510955750942,
-      "learning_rate": 0.00019998852548762453,
-      "loss": 46.0,
-      "step": 29954
-    },
-    {
-      "epoch": 4.824026732155079,
-      "grad_norm": 0.006294541526585817,
-      "learning_rate": 0.0001999885247212283,
-      "loss": 46.0,
-      "step": 29955
-    },
-    {
-      "epoch": 4.824187769233866,
-      "grad_norm": 0.015574057586491108,
-      "learning_rate": 0.00019998852395480645,
-      "loss": 46.0,
-      "step": 29956
-    },
-    {
-      "epoch": 4.824348806312654,
-      "grad_norm": 0.0024759573861956596,
-      "learning_rate": 0.00019998852318835905,
-      "loss": 46.0,
-      "step": 29957
-    },
-    {
-      "epoch": 4.824509843391441,
-      "grad_norm": 0.015002661384642124,
-      "learning_rate": 0.00019998852242188603,
-      "loss": 46.0,
-      "step": 29958
-    },
-    {
-      "epoch": 4.8246708804702285,
-      "grad_norm": 0.002515319036319852,
-      "learning_rate": 0.0001999885216553874,
-      "loss": 46.0,
-      "step": 29959
-    },
-    {
-      "epoch": 4.824831917549016,
-      "grad_norm": 0.0063401684165000916,
-      "learning_rate": 0.00019998852088886322,
-      "loss": 46.0,
-      "step": 29960
-    },
-    {
-      "epoch": 4.824992954627803,
-      "grad_norm": 0.01542993076145649,
-      "learning_rate": 0.00019998852012231342,
-      "loss": 46.0,
-      "step": 29961
-    },
-    {
-      "epoch": 4.825153991706591,
-      "grad_norm": 0.002219581510871649,
-      "learning_rate": 0.00019998851935573803,
-      "loss": 46.0,
-      "step": 29962
-    },
-    {
-      "epoch": 4.825315028785377,
-      "grad_norm": 0.004082961939275265,
-      "learning_rate": 0.00019998851858913708,
-      "loss": 46.0,
-      "step": 29963
-    },
-    {
-      "epoch": 4.825476065864165,
-      "grad_norm": 0.011154497042298317,
-      "learning_rate": 0.0001999885178225105,
-      "loss": 46.0,
-      "step": 29964
-    },
-    {
-      "epoch": 4.825637102942952,
-      "grad_norm": 0.004594013094902039,
-      "learning_rate": 0.00019998851705585832,
-      "loss": 46.0,
-      "step": 29965
-    },
-    {
-      "epoch": 4.82579814002174,
-      "grad_norm": 0.0013535520993173122,
-      "learning_rate": 0.00019998851628918058,
-      "loss": 46.0,
-      "step": 29966
-    },
-    {
-      "epoch": 4.825959177100527,
-      "grad_norm": 0.0030616759322583675,
-      "learning_rate": 0.00019998851552247723,
-      "loss": 46.0,
-      "step": 29967
-    },
-    {
-      "epoch": 4.826120214179315,
-      "grad_norm": 0.00381903862580657,
-      "learning_rate": 0.00019998851475574832,
-      "loss": 46.0,
-      "step": 29968
-    },
-    {
-      "epoch": 4.826281251258102,
-      "grad_norm": 0.010005421936511993,
-      "learning_rate": 0.0001999885139889938,
-      "loss": 46.0,
-      "step": 29969
-    },
-    {
-      "epoch": 4.82644228833689,
-      "grad_norm": 0.006665172521024942,
-      "learning_rate": 0.00019998851322221365,
-      "loss": 46.0,
-      "step": 29970
-    },
-    {
-      "epoch": 4.826603325415677,
-      "grad_norm": 0.003194404998794198,
-      "learning_rate": 0.00019998851245540795,
-      "loss": 46.0,
-      "step": 29971
-    },
-    {
-      "epoch": 4.8267643624944645,
-      "grad_norm": 0.018488630652427673,
-      "learning_rate": 0.00019998851168857664,
-      "loss": 46.0,
-      "step": 29972
-    },
-    {
-      "epoch": 4.826925399573252,
-      "grad_norm": 0.003305562539026141,
-      "learning_rate": 0.00019998851092171976,
-      "loss": 46.0,
-      "step": 29973
-    },
-    {
-      "epoch": 4.827086436652039,
-      "grad_norm": 0.00537894805893302,
-      "learning_rate": 0.00019998851015483725,
-      "loss": 46.0,
-      "step": 29974
-    },
-    {
-      "epoch": 4.827247473730827,
-      "grad_norm": 0.0015342716360464692,
-      "learning_rate": 0.0001999885093879292,
-      "loss": 46.0,
-      "step": 29975
-    },
-    {
-      "epoch": 4.827408510809613,
-      "grad_norm": 0.0036027287133038044,
-      "learning_rate": 0.0001999885086209955,
-      "loss": 46.0,
-      "step": 29976
-    },
-    {
-      "epoch": 4.827569547888402,
-      "grad_norm": 0.001188427908346057,
-      "learning_rate": 0.00019998850785403626,
-      "loss": 46.0,
-      "step": 29977
-    },
-    {
-      "epoch": 4.827730584967188,
-      "grad_norm": 0.0009356028749607503,
-      "learning_rate": 0.0001999885070870514,
-      "loss": 46.0,
-      "step": 29978
-    },
-    {
-      "epoch": 4.827891622045976,
-      "grad_norm": 0.015714846551418304,
-      "learning_rate": 0.00019998850632004094,
-      "loss": 46.0,
-      "step": 29979
-    },
-    {
-      "epoch": 4.828052659124763,
-      "grad_norm": 0.008404687978327274,
-      "learning_rate": 0.0001999885055530049,
-      "loss": 46.0,
-      "step": 29980
-    },
-    {
-      "epoch": 4.828213696203551,
-      "grad_norm": 0.004202132113277912,
-      "learning_rate": 0.00019998850478594328,
-      "loss": 46.0,
-      "step": 29981
-    },
-    {
-      "epoch": 4.828374733282338,
-      "grad_norm": 0.0030121146701276302,
-      "learning_rate": 0.00019998850401885604,
-      "loss": 46.0,
-      "step": 29982
-    },
-    {
-      "epoch": 4.828535770361126,
-      "grad_norm": 0.004591251723468304,
-      "learning_rate": 0.00019998850325174324,
-      "loss": 46.0,
-      "step": 29983
-    },
-    {
-      "epoch": 4.828696807439913,
-      "grad_norm": 0.006869460456073284,
-      "learning_rate": 0.00019998850248460483,
-      "loss": 46.0,
-      "step": 29984
-    },
-    {
-      "epoch": 4.8288578445187005,
-      "grad_norm": 0.001769445720128715,
-      "learning_rate": 0.00019998850171744082,
-      "loss": 46.0,
-      "step": 29985
-    },
-    {
-      "epoch": 4.829018881597488,
-      "grad_norm": 0.0017876693746075034,
-      "learning_rate": 0.00019998850095025126,
-      "loss": 46.0,
-      "step": 29986
-    },
-    {
-      "epoch": 4.829179918676275,
-      "grad_norm": 0.0012478568824008107,
-      "learning_rate": 0.00019998850018303609,
-      "loss": 46.0,
-      "step": 29987
-    },
-    {
-      "epoch": 4.829340955755063,
-      "grad_norm": 0.0021852231584489346,
-      "learning_rate": 0.0001999884994157953,
-      "loss": 46.0,
-      "step": 29988
-    },
-    {
-      "epoch": 4.82950199283385,
-      "grad_norm": 0.0010365054477006197,
-      "learning_rate": 0.00019998849864852895,
-      "loss": 46.0,
-      "step": 29989
-    },
-    {
-      "epoch": 4.829663029912638,
-      "grad_norm": 0.0026975921355187893,
-      "learning_rate": 0.00019998849788123698,
-      "loss": 46.0,
-      "step": 29990
-    },
-    {
-      "epoch": 4.829824066991424,
-      "grad_norm": 0.00727562140673399,
-      "learning_rate": 0.00019998849711391943,
-      "loss": 46.0,
-      "step": 29991
-    },
-    {
-      "epoch": 4.829985104070213,
-      "grad_norm": 0.0065017263405025005,
-      "learning_rate": 0.0001999884963465763,
-      "loss": 46.0,
-      "step": 29992
-    },
-    {
-      "epoch": 4.830146141148999,
-      "grad_norm": 0.008813540451228619,
-      "learning_rate": 0.00019998849557920756,
-      "loss": 46.0,
-      "step": 29993
-    },
-    {
-      "epoch": 4.830307178227787,
-      "grad_norm": 0.002243209630250931,
-      "learning_rate": 0.00019998849481181322,
-      "loss": 46.0,
-      "step": 29994
-    },
-    {
-      "epoch": 4.830468215306574,
-      "grad_norm": 0.002013383898884058,
-      "learning_rate": 0.0001999884940443933,
-      "loss": 46.0,
-      "step": 29995
-    },
-    {
-      "epoch": 4.830629252385362,
-      "grad_norm": 0.005970244761556387,
-      "learning_rate": 0.0001999884932769478,
-      "loss": 46.0,
-      "step": 29996
-    },
-    {
-      "epoch": 4.830790289464149,
-      "grad_norm": 0.007251771632581949,
-      "learning_rate": 0.00019998849250947668,
-      "loss": 46.0,
-      "step": 29997
-    },
-    {
-      "epoch": 4.8309513265429365,
-      "grad_norm": 0.00426494050770998,
-      "learning_rate": 0.00019998849174198,
-      "loss": 46.0,
-      "step": 29998
-    },
-    {
-      "epoch": 4.831112363621724,
-      "grad_norm": 0.0014846059493720531,
-      "learning_rate": 0.00019998849097445771,
-      "loss": 46.0,
-      "step": 29999
-    },
-    {
-      "epoch": 4.831273400700511,
-      "grad_norm": 0.006002186331897974,
-      "learning_rate": 0.00019998849020690985,
-      "loss": 46.0,
-      "step": 30000
-    },
-    {
-      "epoch": 4.831434437779299,
-      "grad_norm": 0.0027920566499233246,
-      "learning_rate": 0.00019998848943933637,
-      "loss": 46.0,
-      "step": 30001
-    },
-    {
-      "epoch": 4.831595474858086,
-      "grad_norm": 0.004230475518852472,
-      "learning_rate": 0.0001999884886717373,
-      "loss": 46.0,
-      "step": 30002
-    },
-    {
-      "epoch": 4.831756511936874,
-      "grad_norm": 0.008627585135400295,
-      "learning_rate": 0.00019998848790411263,
-      "loss": 46.0,
-      "step": 30003
-    },
-    {
-      "epoch": 4.831917549015661,
-      "grad_norm": 0.003110413206741214,
-      "learning_rate": 0.00019998848713646242,
-      "loss": 46.0,
-      "step": 30004
-    },
-    {
-      "epoch": 4.832078586094449,
-      "grad_norm": 0.0030101905576884747,
-      "learning_rate": 0.00019998848636878656,
-      "loss": 46.0,
-      "step": 30005
-    },
-    {
-      "epoch": 4.832239623173235,
-      "grad_norm": 0.003456271719187498,
-      "learning_rate": 0.00019998848560108515,
-      "loss": 46.0,
-      "step": 30006
-    },
-    {
-      "epoch": 4.8324006602520235,
-      "grad_norm": 0.0014123270520940423,
-      "learning_rate": 0.00019998848483335812,
-      "loss": 46.0,
-      "step": 30007
-    },
-    {
-      "epoch": 4.83256169733081,
-      "grad_norm": 0.012516341172158718,
-      "learning_rate": 0.0001999884840656055,
-      "loss": 46.0,
-      "step": 30008
-    },
-    {
-      "epoch": 4.8327227344095975,
-      "grad_norm": 0.006561226677149534,
-      "learning_rate": 0.00019998848329782728,
-      "loss": 46.0,
-      "step": 30009
-    },
-    {
-      "epoch": 4.832883771488385,
-      "grad_norm": 0.019363123923540115,
-      "learning_rate": 0.00019998848253002352,
-      "loss": 46.0,
-      "step": 30010
-    },
-    {
-      "epoch": 4.8330448085671724,
-      "grad_norm": 0.020547227934002876,
-      "learning_rate": 0.0001999884817621941,
-      "loss": 46.0,
-      "step": 30011
-    },
-    {
-      "epoch": 4.83320584564596,
-      "grad_norm": 0.0028513846918940544,
-      "learning_rate": 0.00019998848099433912,
-      "loss": 46.0,
-      "step": 30012
-    },
-    {
-      "epoch": 4.833366882724747,
-      "grad_norm": 0.006960470695048571,
-      "learning_rate": 0.00019998848022645857,
-      "loss": 46.0,
-      "step": 30013
-    },
-    {
-      "epoch": 4.833527919803535,
-      "grad_norm": 0.004035444930195808,
-      "learning_rate": 0.00019998847945855238,
-      "loss": 46.0,
-      "step": 30014
-    },
-    {
-      "epoch": 4.833688956882322,
-      "grad_norm": 0.010005823336541653,
-      "learning_rate": 0.00019998847869062065,
-      "loss": 46.0,
-      "step": 30015
-    },
-    {
-      "epoch": 4.83384999396111,
-      "grad_norm": 0.010618772357702255,
-      "learning_rate": 0.0001999884779226633,
-      "loss": 46.0,
-      "step": 30016
-    },
-    {
-      "epoch": 4.834011031039897,
-      "grad_norm": 0.000913581345230341,
-      "learning_rate": 0.00019998847715468036,
-      "loss": 46.0,
-      "step": 30017
-    },
-    {
-      "epoch": 4.834172068118685,
-      "grad_norm": 0.001419433276169002,
-      "learning_rate": 0.00019998847638667182,
-      "loss": 46.0,
-      "step": 30018
-    },
-    {
-      "epoch": 4.834333105197472,
-      "grad_norm": 0.002580949105322361,
-      "learning_rate": 0.0001999884756186377,
-      "loss": 46.0,
-      "step": 30019
-    },
-    {
-      "epoch": 4.8344941422762595,
-      "grad_norm": 0.010713114403188229,
-      "learning_rate": 0.00019998847485057798,
-      "loss": 46.0,
-      "step": 30020
-    },
-    {
-      "epoch": 4.834655179355046,
-      "grad_norm": 0.0009170554694719613,
-      "learning_rate": 0.00019998847408249267,
-      "loss": 46.0,
-      "step": 30021
-    },
-    {
-      "epoch": 4.8348162164338335,
-      "grad_norm": 0.0032446573022753,
-      "learning_rate": 0.00019998847331438178,
-      "loss": 46.0,
-      "step": 30022
-    },
-    {
-      "epoch": 4.834977253512621,
-      "grad_norm": 0.002422585152089596,
-      "learning_rate": 0.00019998847254624528,
-      "loss": 46.0,
-      "step": 30023
-    },
-    {
-      "epoch": 4.835138290591408,
-      "grad_norm": 0.01699347421526909,
-      "learning_rate": 0.0001999884717780832,
-      "loss": 46.0,
-      "step": 30024
-    },
-    {
-      "epoch": 4.835299327670196,
-      "grad_norm": 0.013538056053221226,
-      "learning_rate": 0.00019998847100989554,
-      "loss": 46.0,
-      "step": 30025
-    },
-    {
-      "epoch": 4.835460364748983,
-      "grad_norm": 0.002329036360606551,
-      "learning_rate": 0.00019998847024168227,
-      "loss": 46.0,
-      "step": 30026
-    },
-    {
-      "epoch": 4.835621401827771,
-      "grad_norm": 0.0018605185905471444,
-      "learning_rate": 0.0001999884694734434,
-      "loss": 46.0,
-      "step": 30027
-    },
-    {
-      "epoch": 4.835782438906558,
-      "grad_norm": 0.002526079071685672,
-      "learning_rate": 0.00019998846870517895,
-      "loss": 46.0,
-      "step": 30028
-    },
-    {
-      "epoch": 4.835943475985346,
-      "grad_norm": 0.0045343260280787945,
-      "learning_rate": 0.0001999884679368889,
-      "loss": 46.0,
-      "step": 30029
-    },
-    {
-      "epoch": 4.836104513064133,
-      "grad_norm": 0.004781306255608797,
-      "learning_rate": 0.00019998846716857326,
-      "loss": 46.0,
-      "step": 30030
-    },
-    {
-      "epoch": 4.836265550142921,
-      "grad_norm": 0.007476066239178181,
-      "learning_rate": 0.00019998846640023206,
-      "loss": 46.0,
-      "step": 30031
-    },
-    {
-      "epoch": 4.836426587221708,
-      "grad_norm": 0.019446317106485367,
-      "learning_rate": 0.00019998846563186522,
-      "loss": 46.0,
-      "step": 30032
-    },
-    {
-      "epoch": 4.8365876243004955,
-      "grad_norm": 0.008398410864174366,
-      "learning_rate": 0.0001999884648634728,
-      "loss": 46.0,
-      "step": 30033
-    },
-    {
-      "epoch": 4.836748661379283,
-      "grad_norm": 0.00261465716175735,
-      "learning_rate": 0.0001999884640950548,
-      "loss": 46.0,
-      "step": 30034
-    },
-    {
-      "epoch": 4.83690969845807,
-      "grad_norm": 0.009719041176140308,
-      "learning_rate": 0.00019998846332661122,
-      "loss": 46.0,
-      "step": 30035
-    },
-    {
-      "epoch": 4.837070735536857,
-      "grad_norm": 0.015909885987639427,
-      "learning_rate": 0.00019998846255814203,
-      "loss": 46.0,
-      "step": 30036
-    },
-    {
-      "epoch": 4.837231772615644,
-      "grad_norm": 0.0069299242459237576,
-      "learning_rate": 0.00019998846178964725,
-      "loss": 46.0,
-      "step": 30037
-    },
-    {
-      "epoch": 4.837392809694432,
-      "grad_norm": 0.002662136685103178,
-      "learning_rate": 0.00019998846102112688,
-      "loss": 46.0,
-      "step": 30038
-    },
-    {
-      "epoch": 4.837553846773219,
-      "grad_norm": 0.028122657909989357,
-      "learning_rate": 0.0001999884602525809,
-      "loss": 46.0,
-      "step": 30039
-    },
-    {
-      "epoch": 4.837714883852007,
-      "grad_norm": 0.0028424824122339487,
-      "learning_rate": 0.00019998845948400936,
-      "loss": 46.0,
-      "step": 30040
-    },
-    {
-      "epoch": 4.837875920930794,
-      "grad_norm": 0.006363618187606335,
-      "learning_rate": 0.00019998845871541224,
-      "loss": 46.0,
-      "step": 30041
-    },
-    {
-      "epoch": 4.838036958009582,
-      "grad_norm": 0.004169164225459099,
-      "learning_rate": 0.0001999884579467895,
-      "loss": 46.0,
-      "step": 30042
-    },
-    {
-      "epoch": 4.838197995088369,
-      "grad_norm": 0.002681769896298647,
-      "learning_rate": 0.00019998845717814117,
-      "loss": 46.0,
-      "step": 30043
-    },
-    {
-      "epoch": 4.838359032167157,
-      "grad_norm": 0.0017171020153909922,
-      "learning_rate": 0.00019998845640946725,
-      "loss": 46.0,
-      "step": 30044
-    },
-    {
-      "epoch": 4.838520069245944,
-      "grad_norm": 0.0024586499202996492,
-      "learning_rate": 0.00019998845564076772,
-      "loss": 46.0,
-      "step": 30045
-    },
-    {
-      "epoch": 4.8386811063247315,
-      "grad_norm": 0.0016594165936112404,
-      "learning_rate": 0.00019998845487204263,
-      "loss": 46.0,
-      "step": 30046
-    },
-    {
-      "epoch": 4.838842143403519,
-      "grad_norm": 0.004528514109551907,
-      "learning_rate": 0.00019998845410329193,
-      "loss": 46.0,
-      "step": 30047
-    },
-    {
-      "epoch": 4.839003180482306,
-      "grad_norm": 0.0020554414950311184,
-      "learning_rate": 0.00019998845333451563,
-      "loss": 46.0,
-      "step": 30048
-    },
-    {
-      "epoch": 4.839164217561094,
-      "grad_norm": 0.0022366635967046022,
-      "learning_rate": 0.00019998845256571376,
-      "loss": 46.0,
-      "step": 30049
-    },
-    {
-      "epoch": 4.839325254639881,
-      "grad_norm": 0.004545716568827629,
-      "learning_rate": 0.0001999884517968863,
-      "loss": 46.0,
-      "step": 30050
-    },
-    {
-      "epoch": 4.839486291718668,
-      "grad_norm": 0.0018451712094247341,
-      "learning_rate": 0.0001999884510280332,
-      "loss": 46.0,
-      "step": 30051
-    },
-    {
-      "epoch": 4.839647328797455,
-      "grad_norm": 0.004489523824304342,
-      "learning_rate": 0.00019998845025915457,
-      "loss": 46.0,
-      "step": 30052
-    },
-    {
-      "epoch": 4.839808365876243,
-      "grad_norm": 0.0024036963004618883,
-      "learning_rate": 0.00019998844949025032,
-      "loss": 46.0,
-      "step": 30053
-    },
-    {
-      "epoch": 4.83996940295503,
-      "grad_norm": 0.003908371087163687,
-      "learning_rate": 0.00019998844872132047,
-      "loss": 46.0,
-      "step": 30054
-    },
-    {
-      "epoch": 4.840130440033818,
-      "grad_norm": 0.006355820223689079,
-      "learning_rate": 0.00019998844795236504,
-      "loss": 46.0,
-      "step": 30055
-    },
-    {
-      "epoch": 4.840291477112605,
-      "grad_norm": 0.003235979937016964,
-      "learning_rate": 0.000199988447183384,
-      "loss": 46.0,
-      "step": 30056
-    },
-    {
-      "epoch": 4.8404525141913926,
-      "grad_norm": 0.0024941901210695505,
-      "learning_rate": 0.0001999884464143774,
-      "loss": 46.0,
-      "step": 30057
-    },
-    {
-      "epoch": 4.84061355127018,
-      "grad_norm": 0.01161390170454979,
-      "learning_rate": 0.00019998844564534518,
-      "loss": 46.0,
-      "step": 30058
-    },
-    {
-      "epoch": 4.8407745883489675,
-      "grad_norm": 0.003102522809058428,
-      "learning_rate": 0.00019998844487628738,
-      "loss": 46.0,
-      "step": 30059
-    },
-    {
-      "epoch": 4.840935625427755,
-      "grad_norm": 0.001814471441321075,
-      "learning_rate": 0.000199988444107204,
-      "loss": 46.0,
-      "step": 30060
-    },
-    {
-      "epoch": 4.841096662506542,
-      "grad_norm": 0.006588350981473923,
-      "learning_rate": 0.000199988443338095,
-      "loss": 46.0,
-      "step": 30061
-    },
-    {
-      "epoch": 4.84125769958533,
-      "grad_norm": 0.0020115876104682684,
-      "learning_rate": 0.00019998844256896044,
-      "loss": 46.0,
-      "step": 30062
-    },
-    {
-      "epoch": 4.841418736664117,
-      "grad_norm": 0.005327795632183552,
-      "learning_rate": 0.00019998844179980029,
-      "loss": 46.0,
-      "step": 30063
-    },
-    {
-      "epoch": 4.841579773742904,
-      "grad_norm": 0.008031301200389862,
-      "learning_rate": 0.0001999884410306145,
-      "loss": 46.0,
-      "step": 30064
-    },
-    {
-      "epoch": 4.841740810821692,
-      "grad_norm": 0.0035582315176725388,
-      "learning_rate": 0.00019998844026140316,
-      "loss": 46.0,
-      "step": 30065
-    },
-    {
-      "epoch": 4.841901847900479,
-      "grad_norm": 0.019608670845627785,
-      "learning_rate": 0.00019998843949216622,
-      "loss": 46.0,
-      "step": 30066
-    },
-    {
-      "epoch": 4.842062884979266,
-      "grad_norm": 0.0021964400075376034,
-      "learning_rate": 0.00019998843872290367,
-      "loss": 46.0,
-      "step": 30067
-    },
-    {
-      "epoch": 4.842223922058054,
-      "grad_norm": 0.014563900418579578,
-      "learning_rate": 0.00019998843795361555,
-      "loss": 46.0,
-      "step": 30068
-    },
-    {
-      "epoch": 4.842384959136841,
-      "grad_norm": 0.0036409483291208744,
-      "learning_rate": 0.00019998843718430182,
-      "loss": 46.0,
-      "step": 30069
-    },
-    {
-      "epoch": 4.8425459962156285,
-      "grad_norm": 0.010826571844518185,
-      "learning_rate": 0.00019998843641496253,
-      "loss": 46.0,
-      "step": 30070
-    },
-    {
-      "epoch": 4.842707033294416,
-      "grad_norm": 0.009756935760378838,
-      "learning_rate": 0.00019998843564559762,
-      "loss": 46.0,
-      "step": 30071
-    },
-    {
-      "epoch": 4.842868070373203,
-      "grad_norm": 0.006842344533652067,
-      "learning_rate": 0.0001999884348762071,
-      "loss": 46.0,
-      "step": 30072
-    },
-    {
-      "epoch": 4.843029107451991,
-      "grad_norm": 0.0023168267216533422,
-      "learning_rate": 0.00019998843410679102,
-      "loss": 46.0,
-      "step": 30073
-    },
-    {
-      "epoch": 4.843190144530778,
-      "grad_norm": 0.0026655595283955336,
-      "learning_rate": 0.00019998843333734936,
-      "loss": 46.0,
-      "step": 30074
-    },
-    {
-      "epoch": 4.843351181609566,
-      "grad_norm": 0.002843728754669428,
-      "learning_rate": 0.00019998843256788208,
-      "loss": 46.0,
-      "step": 30075
-    },
-    {
-      "epoch": 4.843512218688353,
-      "grad_norm": 0.021774034947156906,
-      "learning_rate": 0.0001999884317983892,
-      "loss": 46.0,
-      "step": 30076
-    },
-    {
-      "epoch": 4.843673255767141,
-      "grad_norm": 0.00770561071112752,
-      "learning_rate": 0.00019998843102887075,
-      "loss": 46.0,
-      "step": 30077
-    },
-    {
-      "epoch": 4.843834292845928,
-      "grad_norm": 0.0008823887910693884,
-      "learning_rate": 0.0001999884302593267,
-      "loss": 46.0,
-      "step": 30078
-    },
-    {
-      "epoch": 4.843995329924715,
-      "grad_norm": 0.0016032927669584751,
-      "learning_rate": 0.00019998842948975706,
-      "loss": 46.0,
-      "step": 30079
-    },
-    {
-      "epoch": 4.844156367003503,
-      "grad_norm": 0.004396903328597546,
-      "learning_rate": 0.00019998842872016184,
-      "loss": 46.0,
-      "step": 30080
-    },
-    {
-      "epoch": 4.84431740408229,
-      "grad_norm": 0.016406282782554626,
-      "learning_rate": 0.00019998842795054104,
-      "loss": 46.0,
-      "step": 30081
-    },
-    {
-      "epoch": 4.844478441161077,
-      "grad_norm": 0.0028925766237080097,
-      "learning_rate": 0.00019998842718089462,
-      "loss": 46.0,
-      "step": 30082
-    },
-    {
-      "epoch": 4.8446394782398645,
-      "grad_norm": 0.006768523715436459,
-      "learning_rate": 0.0001999884264112226,
-      "loss": 46.0,
-      "step": 30083
-    },
-    {
-      "epoch": 4.844800515318652,
-      "grad_norm": 0.0020876924972981215,
-      "learning_rate": 0.00019998842564152502,
-      "loss": 46.0,
-      "step": 30084
-    },
-    {
-      "epoch": 4.844961552397439,
-      "grad_norm": 0.0031605486292392015,
-      "learning_rate": 0.00019998842487180181,
-      "loss": 46.0,
-      "step": 30085
-    },
-    {
-      "epoch": 4.845122589476227,
-      "grad_norm": 0.0017508193850517273,
-      "learning_rate": 0.00019998842410205305,
-      "loss": 46.0,
-      "step": 30086
-    },
-    {
-      "epoch": 4.845283626555014,
-      "grad_norm": 0.0005341361975297332,
-      "learning_rate": 0.00019998842333227867,
-      "loss": 46.0,
-      "step": 30087
-    },
-    {
-      "epoch": 4.845444663633802,
-      "grad_norm": 0.004532895050942898,
-      "learning_rate": 0.0001999884225624787,
-      "loss": 46.0,
-      "step": 30088
-    },
-    {
-      "epoch": 4.845605700712589,
-      "grad_norm": 0.0018501088488847017,
-      "learning_rate": 0.00019998842179265314,
-      "loss": 46.0,
-      "step": 30089
-    },
-    {
-      "epoch": 4.845766737791377,
-      "grad_norm": 0.005854758433997631,
-      "learning_rate": 0.00019998842102280197,
-      "loss": 46.0,
-      "step": 30090
-    },
-    {
-      "epoch": 4.845927774870164,
-      "grad_norm": 0.004057453945279121,
-      "learning_rate": 0.00019998842025292524,
-      "loss": 46.0,
-      "step": 30091
-    },
-    {
-      "epoch": 4.846088811948952,
-      "grad_norm": 0.008809411898255348,
-      "learning_rate": 0.00019998841948302293,
-      "loss": 46.0,
-      "step": 30092
-    },
-    {
-      "epoch": 4.846249849027739,
-      "grad_norm": 0.004371872637420893,
-      "learning_rate": 0.000199988418713095,
-      "loss": 46.0,
-      "step": 30093
-    },
-    {
-      "epoch": 4.846410886106526,
-      "grad_norm": 0.005941818468272686,
-      "learning_rate": 0.00019998841794314148,
-      "loss": 46.0,
-      "step": 30094
-    },
-    {
-      "epoch": 4.846571923185313,
-      "grad_norm": 0.0021971401292830706,
-      "learning_rate": 0.00019998841717316237,
-      "loss": 46.0,
-      "step": 30095
-    },
-    {
-      "epoch": 4.8467329602641005,
-      "grad_norm": 0.0011607641354203224,
-      "learning_rate": 0.00019998841640315765,
-      "loss": 46.0,
-      "step": 30096
-    },
-    {
-      "epoch": 4.846893997342888,
-      "grad_norm": 0.0034573585726320744,
-      "learning_rate": 0.00019998841563312735,
-      "loss": 46.0,
-      "step": 30097
-    },
-    {
-      "epoch": 4.847055034421675,
-      "grad_norm": 0.0053519150242209435,
-      "learning_rate": 0.00019998841486307148,
-      "loss": 46.0,
-      "step": 30098
-    },
-    {
-      "epoch": 4.847216071500463,
-      "grad_norm": 0.007368908263742924,
-      "learning_rate": 0.00019998841409299,
-      "loss": 46.0,
-      "step": 30099
-    },
-    {
-      "epoch": 4.84737710857925,
-      "grad_norm": 0.008754079230129719,
-      "learning_rate": 0.00019998841332288293,
-      "loss": 46.0,
-      "step": 30100
-    },
-    {
-      "epoch": 4.847538145658038,
-      "grad_norm": 0.0011550977360457182,
-      "learning_rate": 0.0001999884125527503,
-      "loss": 46.0,
-      "step": 30101
-    },
-    {
-      "epoch": 4.847699182736825,
-      "grad_norm": 0.0049674613401293755,
-      "learning_rate": 0.00019998841178259206,
-      "loss": 46.0,
-      "step": 30102
-    },
-    {
-      "epoch": 4.847860219815613,
-      "grad_norm": 0.001759710954502225,
-      "learning_rate": 0.00019998841101240817,
-      "loss": 46.0,
-      "step": 30103
-    },
-    {
-      "epoch": 4.8480212568944,
-      "grad_norm": 0.0023341637570410967,
-      "learning_rate": 0.00019998841024219876,
-      "loss": 46.0,
-      "step": 30104
-    },
-    {
-      "epoch": 4.848182293973188,
-      "grad_norm": 0.0013407682999968529,
-      "learning_rate": 0.00019998840947196372,
-      "loss": 46.0,
-      "step": 30105
-    },
-    {
-      "epoch": 4.848343331051975,
-      "grad_norm": 0.0016165488632395864,
-      "learning_rate": 0.0001999884087017031,
-      "loss": 46.0,
-      "step": 30106
-    },
-    {
-      "epoch": 4.8485043681307625,
-      "grad_norm": 0.010297886095941067,
-      "learning_rate": 0.0001999884079314169,
-      "loss": 46.0,
-      "step": 30107
-    },
-    {
-      "epoch": 4.84866540520955,
-      "grad_norm": 0.004696574993431568,
-      "learning_rate": 0.00019998840716110508,
-      "loss": 46.0,
-      "step": 30108
-    },
-    {
-      "epoch": 4.8488264422883365,
-      "grad_norm": 0.013613591901957989,
-      "learning_rate": 0.00019998840639076767,
-      "loss": 46.0,
-      "step": 30109
-    },
-    {
-      "epoch": 4.848987479367124,
-      "grad_norm": 0.0024449399206787348,
-      "learning_rate": 0.0001999884056204047,
-      "loss": 46.0,
-      "step": 30110
-    },
-    {
-      "epoch": 4.849148516445911,
-      "grad_norm": 0.005865130107849836,
-      "learning_rate": 0.0001999884048500161,
-      "loss": 46.0,
-      "step": 30111
-    },
-    {
-      "epoch": 4.849309553524699,
-      "grad_norm": 0.004899154417216778,
-      "learning_rate": 0.00019998840407960196,
-      "loss": 46.0,
-      "step": 30112
-    },
-    {
-      "epoch": 4.849470590603486,
-      "grad_norm": 0.0012283639516681433,
-      "learning_rate": 0.00019998840330916217,
-      "loss": 46.0,
-      "step": 30113
-    },
-    {
-      "epoch": 4.849631627682274,
-      "grad_norm": 0.0009270452428609133,
-      "learning_rate": 0.00019998840253869683,
-      "loss": 46.0,
-      "step": 30114
-    },
-    {
-      "epoch": 4.849792664761061,
-      "grad_norm": 0.005930482409894466,
-      "learning_rate": 0.00019998840176820587,
-      "loss": 46.0,
-      "step": 30115
-    },
-    {
-      "epoch": 4.849953701839849,
-      "grad_norm": 0.005603109486401081,
-      "learning_rate": 0.00019998840099768933,
-      "loss": 46.0,
-      "step": 30116
-    },
-    {
-      "epoch": 4.850114738918636,
-      "grad_norm": 0.004156806971877813,
-      "learning_rate": 0.00019998840022714722,
-      "loss": 46.0,
-      "step": 30117
-    },
-    {
-      "epoch": 4.8502757759974235,
-      "grad_norm": 0.003789585782214999,
-      "learning_rate": 0.00019998839945657948,
-      "loss": 46.0,
-      "step": 30118
-    },
-    {
-      "epoch": 4.850436813076211,
-      "grad_norm": 0.005212790798395872,
-      "learning_rate": 0.00019998839868598617,
-      "loss": 46.0,
-      "step": 30119
-    },
-    {
-      "epoch": 4.8505978501549984,
-      "grad_norm": 0.007081334479153156,
-      "learning_rate": 0.00019998839791536728,
-      "loss": 46.0,
-      "step": 30120
-    },
-    {
-      "epoch": 4.850758887233786,
-      "grad_norm": 0.007732132915407419,
-      "learning_rate": 0.00019998839714472277,
-      "loss": 46.0,
-      "step": 30121
-    },
-    {
-      "epoch": 4.850919924312573,
-      "grad_norm": 0.005993806757032871,
-      "learning_rate": 0.00019998839637405268,
-      "loss": 46.0,
-      "step": 30122
-    },
-    {
-      "epoch": 4.851080961391361,
-      "grad_norm": 0.008535230532288551,
-      "learning_rate": 0.000199988395603357,
-      "loss": 46.0,
-      "step": 30123
-    },
-    {
-      "epoch": 4.851241998470147,
-      "grad_norm": 0.008338553830981255,
-      "learning_rate": 0.00019998839483263573,
-      "loss": 46.0,
-      "step": 30124
-    },
-    {
-      "epoch": 4.851403035548935,
-      "grad_norm": 0.004016068298369646,
-      "learning_rate": 0.00019998839406188887,
-      "loss": 46.0,
-      "step": 30125
-    },
-    {
-      "epoch": 4.851564072627722,
-      "grad_norm": 0.0038904640823602676,
-      "learning_rate": 0.0001999883932911164,
-      "loss": 46.0,
-      "step": 30126
-    },
-    {
-      "epoch": 4.85172510970651,
-      "grad_norm": 0.00815887562930584,
-      "learning_rate": 0.00019998839252031837,
-      "loss": 46.0,
-      "step": 30127
-    },
-    {
-      "epoch": 4.851886146785297,
-      "grad_norm": 0.008666375651955605,
-      "learning_rate": 0.0001999883917494947,
-      "loss": 46.0,
-      "step": 30128
-    },
-    {
-      "epoch": 4.852047183864085,
-      "grad_norm": 0.008966566994786263,
-      "learning_rate": 0.00019998839097864547,
-      "loss": 46.0,
-      "step": 30129
-    },
-    {
-      "epoch": 4.852208220942872,
-      "grad_norm": 0.001107836258597672,
-      "learning_rate": 0.00019998839020777065,
-      "loss": 46.0,
-      "step": 30130
-    },
-    {
-      "epoch": 4.8523692580216595,
-      "grad_norm": 0.012590380385518074,
-      "learning_rate": 0.00019998838943687024,
-      "loss": 46.0,
-      "step": 30131
-    },
-    {
-      "epoch": 4.852530295100447,
-      "grad_norm": 0.003554648719727993,
-      "learning_rate": 0.00019998838866594422,
-      "loss": 46.0,
-      "step": 30132
-    },
-    {
-      "epoch": 4.852691332179234,
-      "grad_norm": 0.007092298939824104,
-      "learning_rate": 0.00019998838789499264,
-      "loss": 46.0,
-      "step": 30133
-    },
-    {
-      "epoch": 4.852852369258022,
-      "grad_norm": 0.0020292927511036396,
-      "learning_rate": 0.00019998838712401544,
-      "loss": 46.0,
-      "step": 30134
-    },
-    {
-      "epoch": 4.853013406336809,
-      "grad_norm": 0.009691545739769936,
-      "learning_rate": 0.00019998838635301264,
-      "loss": 46.0,
-      "step": 30135
-    },
-    {
-      "epoch": 4.853174443415597,
-      "grad_norm": 0.011362004093825817,
-      "learning_rate": 0.00019998838558198427,
-      "loss": 46.0,
-      "step": 30136
-    },
-    {
-      "epoch": 4.853335480494383,
-      "grad_norm": 0.004084752406924963,
-      "learning_rate": 0.00019998838481093028,
-      "loss": 46.0,
-      "step": 30137
-    },
-    {
-      "epoch": 4.853496517573172,
-      "grad_norm": 0.013240302912890911,
-      "learning_rate": 0.00019998838403985074,
-      "loss": 46.0,
-      "step": 30138
-    },
-    {
-      "epoch": 4.853657554651958,
-      "grad_norm": 0.00785786472260952,
-      "learning_rate": 0.0001999883832687456,
-      "loss": 46.0,
-      "step": 30139
-    },
-    {
-      "epoch": 4.853818591730746,
-      "grad_norm": 0.0011068106396123767,
-      "learning_rate": 0.00019998838249761484,
-      "loss": 46.0,
-      "step": 30140
-    },
-    {
-      "epoch": 4.853979628809533,
-      "grad_norm": 0.001177026773802936,
-      "learning_rate": 0.0001999883817264585,
-      "loss": 46.0,
-      "step": 30141
-    },
-    {
-      "epoch": 4.854140665888321,
-      "grad_norm": 0.0051734596490859985,
-      "learning_rate": 0.00019998838095527658,
-      "loss": 46.0,
-      "step": 30142
-    },
-    {
-      "epoch": 4.854301702967108,
-      "grad_norm": 0.0038889681454747915,
-      "learning_rate": 0.00019998838018406905,
-      "loss": 46.0,
-      "step": 30143
-    },
-    {
-      "epoch": 4.8544627400458955,
-      "grad_norm": 0.0060275462456047535,
-      "learning_rate": 0.00019998837941283596,
-      "loss": 46.0,
-      "step": 30144
-    },
-    {
-      "epoch": 4.854623777124683,
-      "grad_norm": 0.011145524680614471,
-      "learning_rate": 0.00019998837864157725,
-      "loss": 46.0,
-      "step": 30145
-    },
-    {
-      "epoch": 4.85478481420347,
-      "grad_norm": 0.0065670073963701725,
-      "learning_rate": 0.00019998837787029295,
-      "loss": 46.0,
-      "step": 30146
-    },
-    {
-      "epoch": 4.854945851282258,
-      "grad_norm": 0.0009768144227564335,
-      "learning_rate": 0.00019998837709898307,
-      "loss": 46.0,
-      "step": 30147
-    },
-    {
-      "epoch": 4.855106888361045,
-      "grad_norm": 0.003639065660536289,
-      "learning_rate": 0.0001999883763276476,
-      "loss": 46.0,
-      "step": 30148
-    },
-    {
-      "epoch": 4.855267925439833,
-      "grad_norm": 0.015276482328772545,
-      "learning_rate": 0.00019998837555628652,
-      "loss": 46.0,
-      "step": 30149
-    },
-    {
-      "epoch": 4.85542896251862,
-      "grad_norm": 0.013655618764460087,
-      "learning_rate": 0.00019998837478489985,
-      "loss": 46.0,
-      "step": 30150
-    },
-    {
-      "epoch": 4.855589999597408,
-      "grad_norm": 0.00477282889187336,
-      "learning_rate": 0.0001999883740134876,
-      "loss": 46.0,
-      "step": 30151
-    },
-    {
-      "epoch": 4.855751036676194,
-      "grad_norm": 0.001977842766791582,
-      "learning_rate": 0.00019998837324204972,
-      "loss": 46.0,
-      "step": 30152
-    },
-    {
-      "epoch": 4.855912073754983,
-      "grad_norm": 0.0033407658338546753,
-      "learning_rate": 0.0001999883724705863,
-      "loss": 46.0,
-      "step": 30153
-    },
-    {
-      "epoch": 4.856073110833769,
-      "grad_norm": 0.0027965158224105835,
-      "learning_rate": 0.00019998837169909727,
-      "loss": 46.0,
-      "step": 30154
-    },
-    {
-      "epoch": 4.856234147912557,
-      "grad_norm": 0.0026018614880740643,
-      "learning_rate": 0.00019998837092758266,
-      "loss": 46.0,
-      "step": 30155
-    },
-    {
-      "epoch": 4.856395184991344,
-      "grad_norm": 0.001808958244509995,
-      "learning_rate": 0.00019998837015604244,
-      "loss": 46.0,
-      "step": 30156
-    },
-    {
-      "epoch": 4.8565562220701315,
-      "grad_norm": 0.023240918293595314,
-      "learning_rate": 0.00019998836938447663,
-      "loss": 46.0,
-      "step": 30157
-    },
-    {
-      "epoch": 4.856717259148919,
-      "grad_norm": 0.0030549464281648397,
-      "learning_rate": 0.00019998836861288523,
-      "loss": 46.0,
-      "step": 30158
-    },
-    {
-      "epoch": 4.856878296227706,
-      "grad_norm": 0.00876065343618393,
-      "learning_rate": 0.00019998836784126825,
-      "loss": 46.0,
-      "step": 30159
-    },
-    {
-      "epoch": 4.857039333306494,
-      "grad_norm": 0.0046285721473395824,
-      "learning_rate": 0.00019998836706962566,
-      "loss": 46.0,
-      "step": 30160
-    },
-    {
-      "epoch": 4.857200370385281,
-      "grad_norm": 0.005354483611881733,
-      "learning_rate": 0.0001999883662979575,
-      "loss": 46.0,
-      "step": 30161
-    },
-    {
-      "epoch": 4.857361407464069,
-      "grad_norm": 0.002823406597599387,
-      "learning_rate": 0.00019998836552626373,
-      "loss": 46.0,
-      "step": 30162
-    },
-    {
-      "epoch": 4.857522444542856,
-      "grad_norm": 0.006698140408843756,
-      "learning_rate": 0.00019998836475454437,
-      "loss": 46.0,
-      "step": 30163
-    },
-    {
-      "epoch": 4.857683481621644,
-      "grad_norm": 0.0060594757087528706,
-      "learning_rate": 0.0001999883639827994,
-      "loss": 46.0,
-      "step": 30164
-    },
-    {
-      "epoch": 4.857844518700431,
-      "grad_norm": 0.0035164153669029474,
-      "learning_rate": 0.00019998836321102886,
-      "loss": 46.0,
-      "step": 30165
-    },
-    {
-      "epoch": 4.8580055557792186,
-      "grad_norm": 0.0018185873050242662,
-      "learning_rate": 0.00019998836243923274,
-      "loss": 46.0,
-      "step": 30166
-    },
-    {
-      "epoch": 4.858166592858005,
-      "grad_norm": 0.0027507031336426735,
-      "learning_rate": 0.000199988361667411,
-      "loss": 46.0,
-      "step": 30167
-    },
-    {
-      "epoch": 4.858327629936793,
-      "grad_norm": 0.0015631286660209298,
-      "learning_rate": 0.0001999883608955637,
-      "loss": 46.0,
-      "step": 30168
-    },
-    {
-      "epoch": 4.85848866701558,
-      "grad_norm": 0.0015366551233455539,
-      "learning_rate": 0.0001999883601236908,
-      "loss": 46.0,
-      "step": 30169
-    },
-    {
-      "epoch": 4.8586497040943675,
-      "grad_norm": 0.00434156646952033,
-      "learning_rate": 0.00019998835935179228,
-      "loss": 46.0,
-      "step": 30170
-    },
-    {
-      "epoch": 4.858810741173155,
-      "grad_norm": 0.02606615424156189,
-      "learning_rate": 0.00019998835857986817,
-      "loss": 46.0,
-      "step": 30171
-    },
-    {
-      "epoch": 4.858971778251942,
-      "grad_norm": 0.0031105487141758204,
-      "learning_rate": 0.0001999883578079185,
-      "loss": 46.0,
-      "step": 30172
-    },
-    {
-      "epoch": 4.85913281533073,
-      "grad_norm": 0.004212117288261652,
-      "learning_rate": 0.00019998835703594322,
-      "loss": 46.0,
-      "step": 30173
-    },
-    {
-      "epoch": 4.859293852409517,
-      "grad_norm": 0.004099614452570677,
-      "learning_rate": 0.00019998835626394238,
-      "loss": 46.0,
-      "step": 30174
-    },
-    {
-      "epoch": 4.859454889488305,
-      "grad_norm": 0.011553723365068436,
-      "learning_rate": 0.0001999883554919159,
-      "loss": 46.0,
-      "step": 30175
-    },
-    {
-      "epoch": 4.859615926567092,
-      "grad_norm": 0.00280005089007318,
-      "learning_rate": 0.00019998835471986384,
-      "loss": 46.0,
-      "step": 30176
-    },
-    {
-      "epoch": 4.85977696364588,
-      "grad_norm": 0.003370868042111397,
-      "learning_rate": 0.0001999883539477862,
-      "loss": 46.0,
-      "step": 30177
-    },
-    {
-      "epoch": 4.859938000724667,
-      "grad_norm": 0.006544739007949829,
-      "learning_rate": 0.00019998835317568296,
-      "loss": 46.0,
-      "step": 30178
-    },
-    {
-      "epoch": 4.8600990378034545,
-      "grad_norm": 0.0028320089913904667,
-      "learning_rate": 0.00019998835240355413,
-      "loss": 46.0,
-      "step": 30179
-    },
-    {
-      "epoch": 4.860260074882242,
-      "grad_norm": 0.006709593813866377,
-      "learning_rate": 0.0001999883516313997,
-      "loss": 46.0,
-      "step": 30180
-    },
-    {
-      "epoch": 4.860421111961029,
-      "grad_norm": 0.009623527526855469,
-      "learning_rate": 0.0001999883508592197,
-      "loss": 46.0,
-      "step": 30181
-    },
-    {
-      "epoch": 4.860582149039816,
-      "grad_norm": 0.011304141022264957,
-      "learning_rate": 0.0001999883500870141,
-      "loss": 46.0,
-      "step": 30182
-    },
-    {
-      "epoch": 4.8607431861186035,
-      "grad_norm": 0.0010971177835017443,
-      "learning_rate": 0.0001999883493147829,
-      "loss": 46.0,
-      "step": 30183
-    },
-    {
-      "epoch": 4.860904223197391,
-      "grad_norm": 0.0016037834575399756,
-      "learning_rate": 0.0001999883485425261,
-      "loss": 46.0,
-      "step": 30184
-    },
-    {
-      "epoch": 4.861065260276178,
-      "grad_norm": 0.004147219937294722,
-      "learning_rate": 0.00019998834777024374,
-      "loss": 46.0,
-      "step": 30185
-    },
-    {
-      "epoch": 4.861226297354966,
-      "grad_norm": 0.00477145379409194,
-      "learning_rate": 0.00019998834699793577,
-      "loss": 46.0,
-      "step": 30186
-    },
-    {
-      "epoch": 4.861387334433753,
-      "grad_norm": 0.005681208334863186,
-      "learning_rate": 0.0001999883462256022,
-      "loss": 46.0,
-      "step": 30187
-    },
-    {
-      "epoch": 4.861548371512541,
-      "grad_norm": 0.011490699835121632,
-      "learning_rate": 0.00019998834545324306,
-      "loss": 46.0,
-      "step": 30188
-    },
-    {
-      "epoch": 4.861709408591328,
-      "grad_norm": 0.00845568347722292,
-      "learning_rate": 0.0001999883446808583,
-      "loss": 46.0,
-      "step": 30189
-    },
-    {
-      "epoch": 4.861870445670116,
-      "grad_norm": 0.005368630401790142,
-      "learning_rate": 0.00019998834390844795,
-      "loss": 46.0,
-      "step": 30190
-    },
-    {
-      "epoch": 4.862031482748903,
-      "grad_norm": 0.004007462877780199,
-      "learning_rate": 0.00019998834313601202,
-      "loss": 46.0,
-      "step": 30191
-    },
-    {
-      "epoch": 4.8621925198276905,
-      "grad_norm": 0.004882803652435541,
-      "learning_rate": 0.0001999883423635505,
-      "loss": 46.0,
-      "step": 30192
-    },
-    {
-      "epoch": 4.862353556906478,
-      "grad_norm": 0.004058310762047768,
-      "learning_rate": 0.0001999883415910634,
-      "loss": 46.0,
-      "step": 30193
-    },
-    {
-      "epoch": 4.862514593985265,
-      "grad_norm": 0.01806306280195713,
-      "learning_rate": 0.0001999883408185507,
-      "loss": 46.0,
-      "step": 30194
-    },
-    {
-      "epoch": 4.862675631064053,
-      "grad_norm": 0.004027595277875662,
-      "learning_rate": 0.00019998834004601238,
-      "loss": 46.0,
-      "step": 30195
-    },
-    {
-      "epoch": 4.86283666814284,
-      "grad_norm": 0.012487724423408508,
-      "learning_rate": 0.0001999883392734485,
-      "loss": 46.0,
-      "step": 30196
-    },
-    {
-      "epoch": 4.862997705221627,
-      "grad_norm": 0.01029012817889452,
-      "learning_rate": 0.000199988338500859,
-      "loss": 46.0,
-      "step": 30197
-    },
-    {
-      "epoch": 4.863158742300414,
-      "grad_norm": 0.002836928702890873,
-      "learning_rate": 0.00019998833772824393,
-      "loss": 46.0,
-      "step": 30198
-    },
-    {
-      "epoch": 4.863319779379202,
-      "grad_norm": 0.0032990477047860622,
-      "learning_rate": 0.00019998833695560327,
-      "loss": 46.0,
-      "step": 30199
-    },
-    {
-      "epoch": 4.863480816457989,
-      "grad_norm": 0.007273085415363312,
-      "learning_rate": 0.00019998833618293702,
-      "loss": 46.0,
-      "step": 30200
-    },
-    {
-      "epoch": 4.863641853536777,
-      "grad_norm": 0.0014787502586841583,
-      "learning_rate": 0.00019998833541024516,
-      "loss": 46.0,
-      "step": 30201
-    },
-    {
-      "epoch": 4.863802890615564,
-      "grad_norm": 0.00284698442555964,
-      "learning_rate": 0.00019998833463752774,
-      "loss": 46.0,
-      "step": 30202
-    },
-    {
-      "epoch": 4.863963927694352,
-      "grad_norm": 0.01878676563501358,
-      "learning_rate": 0.00019998833386478468,
-      "loss": 46.0,
-      "step": 30203
-    },
-    {
-      "epoch": 4.864124964773139,
-      "grad_norm": 0.0051757642067968845,
-      "learning_rate": 0.00019998833309201608,
-      "loss": 46.0,
-      "step": 30204
-    },
-    {
-      "epoch": 4.8642860018519265,
-      "grad_norm": 0.0012672676239162683,
-      "learning_rate": 0.00019998833231922185,
-      "loss": 46.0,
-      "step": 30205
-    },
-    {
-      "epoch": 4.864447038930714,
-      "grad_norm": 0.007116463501006365,
-      "learning_rate": 0.00019998833154640205,
-      "loss": 46.0,
-      "step": 30206
-    },
-    {
-      "epoch": 4.864608076009501,
-      "grad_norm": 0.000850141397677362,
-      "learning_rate": 0.00019998833077355664,
-      "loss": 46.0,
-      "step": 30207
-    },
-    {
-      "epoch": 4.864769113088289,
-      "grad_norm": 0.001165292109362781,
-      "learning_rate": 0.00019998833000068564,
-      "loss": 46.0,
-      "step": 30208
-    },
-    {
-      "epoch": 4.864930150167076,
-      "grad_norm": 0.001940658432431519,
-      "learning_rate": 0.00019998832922778906,
-      "loss": 46.0,
-      "step": 30209
-    },
-    {
-      "epoch": 4.865091187245863,
-      "grad_norm": 0.0036168291699141264,
-      "learning_rate": 0.00019998832845486688,
-      "loss": 46.0,
-      "step": 30210
-    },
-    {
-      "epoch": 4.865252224324651,
-      "grad_norm": 0.0011742841452360153,
-      "learning_rate": 0.00019998832768191913,
-      "loss": 46.0,
-      "step": 30211
-    },
-    {
-      "epoch": 4.865413261403438,
-      "grad_norm": 0.001702725188806653,
-      "learning_rate": 0.00019998832690894575,
-      "loss": 46.0,
-      "step": 30212
-    },
-    {
-      "epoch": 4.865574298482225,
-      "grad_norm": 0.0012292437022551894,
-      "learning_rate": 0.0001999883261359468,
-      "loss": 46.0,
-      "step": 30213
-    },
-    {
-      "epoch": 4.865735335561013,
-      "grad_norm": 0.0007949745049700141,
-      "learning_rate": 0.00019998832536292227,
-      "loss": 46.0,
-      "step": 30214
-    },
-    {
-      "epoch": 4.8658963726398,
-      "grad_norm": 0.007670976221561432,
-      "learning_rate": 0.0001999883245898721,
-      "loss": 46.0,
-      "step": 30215
-    },
-    {
-      "epoch": 4.866057409718588,
-      "grad_norm": 0.004126108251512051,
-      "learning_rate": 0.00019998832381679639,
-      "loss": 46.0,
-      "step": 30216
-    },
-    {
-      "epoch": 4.866218446797375,
-      "grad_norm": 0.0031773000955581665,
-      "learning_rate": 0.00019998832304369508,
-      "loss": 46.0,
-      "step": 30217
-    },
-    {
-      "epoch": 4.8663794838761625,
-      "grad_norm": 0.003949275240302086,
-      "learning_rate": 0.00019998832227056815,
-      "loss": 46.0,
-      "step": 30218
-    },
-    {
-      "epoch": 4.86654052095495,
-      "grad_norm": 0.004697037860751152,
-      "learning_rate": 0.00019998832149741567,
-      "loss": 46.0,
-      "step": 30219
-    },
-    {
-      "epoch": 4.866701558033737,
-      "grad_norm": 0.0032105708960443735,
-      "learning_rate": 0.00019998832072423757,
-      "loss": 46.0,
-      "step": 30220
-    },
-    {
-      "epoch": 4.866862595112525,
-      "grad_norm": 0.0019673542119562626,
-      "learning_rate": 0.00019998831995103386,
-      "loss": 46.0,
-      "step": 30221
-    },
-    {
-      "epoch": 4.867023632191312,
-      "grad_norm": 0.003639364615082741,
-      "learning_rate": 0.0001999883191778046,
-      "loss": 46.0,
-      "step": 30222
-    },
-    {
-      "epoch": 4.8671846692701,
-      "grad_norm": 0.0025361175648868084,
-      "learning_rate": 0.00019998831840454973,
-      "loss": 46.0,
-      "step": 30223
-    },
-    {
-      "epoch": 4.867345706348887,
-      "grad_norm": 0.0024909640196710825,
-      "learning_rate": 0.00019998831763126925,
-      "loss": 46.0,
-      "step": 30224
-    },
-    {
-      "epoch": 4.867506743427674,
-      "grad_norm": 0.0006870125071145594,
-      "learning_rate": 0.00019998831685796322,
-      "loss": 46.0,
-      "step": 30225
-    },
-    {
-      "epoch": 4.867667780506462,
-      "grad_norm": 0.002211096463724971,
-      "learning_rate": 0.00019998831608463154,
-      "loss": 46.0,
-      "step": 30226
-    },
-    {
-      "epoch": 4.867828817585249,
-      "grad_norm": 0.016503984108567238,
-      "learning_rate": 0.0001999883153112743,
-      "loss": 46.0,
-      "step": 30227
-    },
-    {
-      "epoch": 4.867989854664036,
-      "grad_norm": 0.0012430896749719977,
-      "learning_rate": 0.00019998831453789148,
-      "loss": 46.0,
-      "step": 30228
-    },
-    {
-      "epoch": 4.868150891742824,
-      "grad_norm": 0.007462533190846443,
-      "learning_rate": 0.00019998831376448307,
-      "loss": 46.0,
-      "step": 30229
-    },
-    {
-      "epoch": 4.868311928821611,
-      "grad_norm": 0.01666617952287197,
-      "learning_rate": 0.00019998831299104905,
-      "loss": 46.0,
-      "step": 30230
-    },
-    {
-      "epoch": 4.8684729659003985,
-      "grad_norm": 0.002064245520159602,
-      "learning_rate": 0.00019998831221758944,
-      "loss": 46.0,
-      "step": 30231
-    },
-    {
-      "epoch": 4.868634002979186,
-      "grad_norm": 0.005984019488096237,
-      "learning_rate": 0.00019998831144410427,
-      "loss": 46.0,
-      "step": 30232
-    },
-    {
-      "epoch": 4.868795040057973,
-      "grad_norm": 0.006874515675008297,
-      "learning_rate": 0.00019998831067059345,
-      "loss": 46.0,
-      "step": 30233
-    },
-    {
-      "epoch": 4.868956077136761,
-      "grad_norm": 0.0031324748415499926,
-      "learning_rate": 0.00019998830989705708,
-      "loss": 46.0,
-      "step": 30234
-    },
-    {
-      "epoch": 4.869117114215548,
-      "grad_norm": 0.009701925329864025,
-      "learning_rate": 0.00019998830912349512,
-      "loss": 46.0,
-      "step": 30235
-    },
-    {
-      "epoch": 4.869278151294336,
-      "grad_norm": 0.0017760891932994127,
-      "learning_rate": 0.00019998830834990755,
-      "loss": 46.0,
-      "step": 30236
-    },
-    {
-      "epoch": 4.869439188373123,
-      "grad_norm": 0.004007274284958839,
-      "learning_rate": 0.0001999883075762944,
-      "loss": 46.0,
-      "step": 30237
-    },
-    {
-      "epoch": 4.869600225451911,
-      "grad_norm": 0.008442658931016922,
-      "learning_rate": 0.00019998830680265564,
-      "loss": 46.0,
-      "step": 30238
-    },
-    {
-      "epoch": 4.869761262530698,
-      "grad_norm": 0.005054876673966646,
-      "learning_rate": 0.00019998830602899128,
-      "loss": 46.0,
-      "step": 30239
-    },
-    {
-      "epoch": 4.869922299609485,
-      "grad_norm": 0.004223821219056845,
-      "learning_rate": 0.00019998830525530138,
-      "loss": 46.0,
-      "step": 30240
-    },
-    {
-      "epoch": 4.870083336688272,
-      "grad_norm": 0.017017167061567307,
-      "learning_rate": 0.00019998830448158584,
-      "loss": 46.0,
-      "step": 30241
-    },
-    {
-      "epoch": 4.8702443737670595,
-      "grad_norm": 0.0032326721120625734,
-      "learning_rate": 0.00019998830370784472,
-      "loss": 46.0,
-      "step": 30242
-    },
-    {
-      "epoch": 4.870405410845847,
-      "grad_norm": 0.005026225931942463,
-      "learning_rate": 0.00019998830293407803,
-      "loss": 46.0,
-      "step": 30243
-    },
-    {
-      "epoch": 4.8705664479246344,
-      "grad_norm": 0.0020192961674183607,
-      "learning_rate": 0.0001999883021602857,
-      "loss": 46.0,
-      "step": 30244
-    },
-    {
-      "epoch": 4.870727485003422,
-      "grad_norm": 0.0018568228697404265,
-      "learning_rate": 0.0001999883013864678,
-      "loss": 46.0,
-      "step": 30245
-    },
-    {
-      "epoch": 4.870888522082209,
-      "grad_norm": 0.0020218132995069027,
-      "learning_rate": 0.00019998830061262432,
-      "loss": 46.0,
-      "step": 30246
-    },
-    {
-      "epoch": 4.871049559160997,
-      "grad_norm": 0.0018797758966684341,
-      "learning_rate": 0.00019998829983875526,
-      "loss": 46.0,
-      "step": 30247
-    },
-    {
-      "epoch": 4.871210596239784,
-      "grad_norm": 0.008374815806746483,
-      "learning_rate": 0.0001999882990648606,
-      "loss": 46.0,
-      "step": 30248
-    },
-    {
-      "epoch": 4.871371633318572,
-      "grad_norm": 0.0010333530372008681,
-      "learning_rate": 0.00019998829829094033,
-      "loss": 46.0,
-      "step": 30249
-    },
-    {
-      "epoch": 4.871532670397359,
-      "grad_norm": 0.0016382524045184255,
-      "learning_rate": 0.00019998829751699448,
-      "loss": 46.0,
-      "step": 30250
-    },
-    {
-      "epoch": 4.871693707476147,
-      "grad_norm": 0.006623008754104376,
-      "learning_rate": 0.00019998829674302301,
-      "loss": 46.0,
-      "step": 30251
-    },
-    {
-      "epoch": 4.871854744554934,
-      "grad_norm": 0.004867917858064175,
-      "learning_rate": 0.000199988295969026,
-      "loss": 46.0,
-      "step": 30252
-    },
-    {
-      "epoch": 4.8720157816337215,
-      "grad_norm": 0.010020680725574493,
-      "learning_rate": 0.00019998829519500338,
-      "loss": 46.0,
-      "step": 30253
-    },
-    {
-      "epoch": 4.872176818712509,
-      "grad_norm": 0.0018205250380560756,
-      "learning_rate": 0.00019998829442095516,
-      "loss": 46.0,
-      "step": 30254
-    },
-    {
-      "epoch": 4.8723378557912955,
-      "grad_norm": 0.0014964076690375805,
-      "learning_rate": 0.00019998829364688134,
-      "loss": 46.0,
-      "step": 30255
-    },
-    {
-      "epoch": 4.872498892870083,
-      "grad_norm": 0.006149259861558676,
-      "learning_rate": 0.00019998829287278194,
-      "loss": 46.0,
-      "step": 30256
-    },
-    {
-      "epoch": 4.87265992994887,
-      "grad_norm": 0.002652009017765522,
-      "learning_rate": 0.00019998829209865693,
-      "loss": 46.0,
-      "step": 30257
-    },
-    {
-      "epoch": 4.872820967027658,
-      "grad_norm": 0.007546828594058752,
-      "learning_rate": 0.00019998829132450636,
-      "loss": 46.0,
-      "step": 30258
-    },
-    {
-      "epoch": 4.872982004106445,
-      "grad_norm": 0.004417920019477606,
-      "learning_rate": 0.00019998829055033017,
-      "loss": 46.0,
-      "step": 30259
-    },
-    {
-      "epoch": 4.873143041185233,
-      "grad_norm": 0.0015945524210110307,
-      "learning_rate": 0.0001999882897761284,
-      "loss": 46.0,
-      "step": 30260
-    },
-    {
-      "epoch": 4.87330407826402,
-      "grad_norm": 0.0023466814309358597,
-      "learning_rate": 0.00019998828900190106,
-      "loss": 46.0,
-      "step": 30261
-    },
-    {
-      "epoch": 4.873465115342808,
-      "grad_norm": 0.012706050649285316,
-      "learning_rate": 0.00019998828822764809,
-      "loss": 46.0,
-      "step": 30262
-    },
-    {
-      "epoch": 4.873626152421595,
-      "grad_norm": 0.0036380223464220762,
-      "learning_rate": 0.00019998828745336955,
-      "loss": 46.0,
-      "step": 30263
-    },
-    {
-      "epoch": 4.873787189500383,
-      "grad_norm": 0.002249692566692829,
-      "learning_rate": 0.0001999882866790654,
-      "loss": 46.0,
-      "step": 30264
-    },
-    {
-      "epoch": 4.87394822657917,
-      "grad_norm": 0.005754155106842518,
-      "learning_rate": 0.0001999882859047357,
-      "loss": 46.0,
-      "step": 30265
-    },
-    {
-      "epoch": 4.8741092636579575,
-      "grad_norm": 0.0035027097910642624,
-      "learning_rate": 0.00019998828513038034,
-      "loss": 46.0,
-      "step": 30266
-    },
-    {
-      "epoch": 4.874270300736745,
-      "grad_norm": 0.00200637592934072,
-      "learning_rate": 0.00019998828435599942,
-      "loss": 46.0,
-      "step": 30267
-    },
-    {
-      "epoch": 4.874431337815532,
-      "grad_norm": 0.0037039672024548054,
-      "learning_rate": 0.00019998828358159295,
-      "loss": 46.0,
-      "step": 30268
-    },
-    {
-      "epoch": 4.87459237489432,
-      "grad_norm": 0.018483847379684448,
-      "learning_rate": 0.00019998828280716084,
-      "loss": 46.0,
-      "step": 30269
-    },
-    {
-      "epoch": 4.874753411973106,
-      "grad_norm": 0.009647499769926071,
-      "learning_rate": 0.00019998828203270314,
-      "loss": 46.0,
-      "step": 30270
-    },
-    {
-      "epoch": 4.874914449051894,
-      "grad_norm": 0.00447214487940073,
-      "learning_rate": 0.00019998828125821988,
-      "loss": 46.0,
-      "step": 30271
-    },
-    {
-      "epoch": 4.875075486130681,
-      "grad_norm": 0.0038623660802841187,
-      "learning_rate": 0.00019998828048371097,
-      "loss": 46.0,
-      "step": 30272
-    },
-    {
-      "epoch": 4.875236523209469,
-      "grad_norm": 0.0018283635145053267,
-      "learning_rate": 0.00019998827970917654,
-      "loss": 46.0,
-      "step": 30273
-    },
-    {
-      "epoch": 4.875397560288256,
-      "grad_norm": 0.003693078877404332,
-      "learning_rate": 0.00019998827893461646,
-      "loss": 46.0,
-      "step": 30274
-    },
-    {
-      "epoch": 4.875558597367044,
-      "grad_norm": 0.014787347987294197,
-      "learning_rate": 0.00019998827816003083,
-      "loss": 46.0,
-      "step": 30275
-    },
-    {
-      "epoch": 4.875719634445831,
-      "grad_norm": 0.012749990448355675,
-      "learning_rate": 0.00019998827738541958,
-      "loss": 46.0,
-      "step": 30276
-    },
-    {
-      "epoch": 4.875880671524619,
-      "grad_norm": 0.0014343708753585815,
-      "learning_rate": 0.00019998827661078274,
-      "loss": 46.0,
-      "step": 30277
-    },
-    {
-      "epoch": 4.876041708603406,
-      "grad_norm": 0.0012202727375552058,
-      "learning_rate": 0.0001999882758361203,
-      "loss": 46.0,
-      "step": 30278
-    },
-    {
-      "epoch": 4.8762027456821935,
-      "grad_norm": 0.0057300073094666,
-      "learning_rate": 0.00019998827506143233,
-      "loss": 46.0,
-      "step": 30279
-    },
-    {
-      "epoch": 4.876363782760981,
-      "grad_norm": 0.0021433464717119932,
-      "learning_rate": 0.0001999882742867187,
-      "loss": 46.0,
-      "step": 30280
-    },
-    {
-      "epoch": 4.876524819839768,
-      "grad_norm": 0.0040053422562778,
-      "learning_rate": 0.00019998827351197951,
-      "loss": 46.0,
-      "step": 30281
-    },
-    {
-      "epoch": 4.876685856918556,
-      "grad_norm": 0.006644417066127062,
-      "learning_rate": 0.00019998827273721469,
-      "loss": 46.0,
-      "step": 30282
-    },
-    {
-      "epoch": 4.876846893997342,
-      "grad_norm": 0.007489868439733982,
-      "learning_rate": 0.00019998827196242433,
-      "loss": 46.0,
-      "step": 30283
-    },
-    {
-      "epoch": 4.877007931076131,
-      "grad_norm": 0.0031959705520421267,
-      "learning_rate": 0.00019998827118760835,
-      "loss": 46.0,
-      "step": 30284
-    },
-    {
-      "epoch": 4.877168968154917,
-      "grad_norm": 0.009030255489051342,
-      "learning_rate": 0.0001999882704127668,
-      "loss": 46.0,
-      "step": 30285
-    },
-    {
-      "epoch": 4.877330005233705,
-      "grad_norm": 0.006461299955844879,
-      "learning_rate": 0.00019998826963789964,
-      "loss": 46.0,
-      "step": 30286
-    },
-    {
-      "epoch": 4.877491042312492,
-      "grad_norm": 0.00523214740678668,
-      "learning_rate": 0.00019998826886300687,
-      "loss": 46.0,
-      "step": 30287
-    },
-    {
-      "epoch": 4.87765207939128,
-      "grad_norm": 0.008426429703831673,
-      "learning_rate": 0.00019998826808808855,
-      "loss": 46.0,
-      "step": 30288
-    },
-    {
-      "epoch": 4.877813116470067,
-      "grad_norm": 0.016680296510457993,
-      "learning_rate": 0.00019998826731314458,
-      "loss": 46.0,
-      "step": 30289
-    },
-    {
-      "epoch": 4.8779741535488546,
-      "grad_norm": 0.008195668458938599,
-      "learning_rate": 0.00019998826653817506,
-      "loss": 46.0,
-      "step": 30290
-    },
-    {
-      "epoch": 4.878135190627642,
-      "grad_norm": 0.005405410658568144,
-      "learning_rate": 0.00019998826576317995,
-      "loss": 46.0,
-      "step": 30291
-    },
-    {
-      "epoch": 4.8782962277064295,
-      "grad_norm": 0.006686722859740257,
-      "learning_rate": 0.00019998826498815925,
-      "loss": 46.0,
-      "step": 30292
-    },
-    {
-      "epoch": 4.878457264785217,
-      "grad_norm": 0.0007601392571814358,
-      "learning_rate": 0.00019998826421311293,
-      "loss": 46.0,
-      "step": 30293
-    },
-    {
-      "epoch": 4.878618301864004,
-      "grad_norm": 0.017041867598891258,
-      "learning_rate": 0.00019998826343804106,
-      "loss": 46.0,
-      "step": 30294
-    },
-    {
-      "epoch": 4.878779338942792,
-      "grad_norm": 0.005406068172305822,
-      "learning_rate": 0.00019998826266294354,
-      "loss": 46.0,
-      "step": 30295
-    },
-    {
-      "epoch": 4.878940376021579,
-      "grad_norm": 0.0026037776842713356,
-      "learning_rate": 0.00019998826188782047,
-      "loss": 46.0,
-      "step": 30296
-    },
-    {
-      "epoch": 4.879101413100367,
-      "grad_norm": 0.005247111432254314,
-      "learning_rate": 0.0001999882611126718,
-      "loss": 46.0,
-      "step": 30297
-    },
-    {
-      "epoch": 4.879262450179153,
-      "grad_norm": 0.01520408596843481,
-      "learning_rate": 0.00019998826033749756,
-      "loss": 46.0,
-      "step": 30298
-    },
-    {
-      "epoch": 4.879423487257942,
-      "grad_norm": 0.003051883541047573,
-      "learning_rate": 0.0001999882595622977,
-      "loss": 46.0,
-      "step": 30299
-    },
-    {
-      "epoch": 4.879584524336728,
-      "grad_norm": 0.008032537996768951,
-      "learning_rate": 0.00019998825878707227,
-      "loss": 46.0,
-      "step": 30300
-    },
-    {
-      "epoch": 4.879745561415516,
-      "grad_norm": 0.010447860695421696,
-      "learning_rate": 0.00019998825801182123,
-      "loss": 46.0,
-      "step": 30301
-    },
-    {
-      "epoch": 4.879906598494303,
-      "grad_norm": 0.006183580029755831,
-      "learning_rate": 0.00019998825723654458,
-      "loss": 46.0,
-      "step": 30302
-    },
-    {
-      "epoch": 4.8800676355730905,
-      "grad_norm": 0.0260601919144392,
-      "learning_rate": 0.00019998825646124236,
-      "loss": 46.0,
-      "step": 30303
-    },
-    {
-      "epoch": 4.880228672651878,
-      "grad_norm": 0.004913515876978636,
-      "learning_rate": 0.00019998825568591456,
-      "loss": 46.0,
-      "step": 30304
-    },
-    {
-      "epoch": 4.880389709730665,
-      "grad_norm": 0.01729273796081543,
-      "learning_rate": 0.00019998825491056115,
-      "loss": 46.0,
-      "step": 30305
-    },
-    {
-      "epoch": 4.880550746809453,
-      "grad_norm": 0.0011975648812949657,
-      "learning_rate": 0.00019998825413518217,
-      "loss": 46.0,
-      "step": 30306
-    },
-    {
-      "epoch": 4.88071178388824,
-      "grad_norm": 0.0027460732962936163,
-      "learning_rate": 0.00019998825335977759,
-      "loss": 46.0,
-      "step": 30307
-    },
-    {
-      "epoch": 4.880872820967028,
-      "grad_norm": 0.004315624944865704,
-      "learning_rate": 0.00019998825258434738,
-      "loss": 46.0,
-      "step": 30308
-    },
-    {
-      "epoch": 4.881033858045815,
-      "grad_norm": 0.007370648439973593,
-      "learning_rate": 0.00019998825180889162,
-      "loss": 46.0,
-      "step": 30309
-    },
-    {
-      "epoch": 4.881194895124603,
-      "grad_norm": 0.003338668728247285,
-      "learning_rate": 0.00019998825103341027,
-      "loss": 46.0,
-      "step": 30310
-    },
-    {
-      "epoch": 4.88135593220339,
-      "grad_norm": 0.0010761988814920187,
-      "learning_rate": 0.0001999882502579033,
-      "loss": 46.0,
-      "step": 30311
-    },
-    {
-      "epoch": 4.881516969282178,
-      "grad_norm": 0.0032709992956370115,
-      "learning_rate": 0.00019998824948237073,
-      "loss": 46.0,
-      "step": 30312
-    },
-    {
-      "epoch": 4.881678006360964,
-      "grad_norm": 0.002058805199339986,
-      "learning_rate": 0.00019998824870681261,
-      "loss": 46.0,
-      "step": 30313
-    },
-    {
-      "epoch": 4.8818390434397525,
-      "grad_norm": 0.0021465301979333162,
-      "learning_rate": 0.0001999882479312289,
-      "loss": 46.0,
-      "step": 30314
-    },
-    {
-      "epoch": 4.882000080518539,
-      "grad_norm": 0.006435915362089872,
-      "learning_rate": 0.00019998824715561955,
-      "loss": 46.0,
-      "step": 30315
-    },
-    {
-      "epoch": 4.8821611175973265,
-      "grad_norm": 0.0026787524111568928,
-      "learning_rate": 0.00019998824637998462,
-      "loss": 46.0,
-      "step": 30316
-    },
-    {
-      "epoch": 4.882322154676114,
-      "grad_norm": 0.02055901288986206,
-      "learning_rate": 0.00019998824560432413,
-      "loss": 46.0,
-      "step": 30317
-    },
-    {
-      "epoch": 4.882483191754901,
-      "grad_norm": 0.004382592625916004,
-      "learning_rate": 0.00019998824482863803,
-      "loss": 46.0,
-      "step": 30318
-    },
-    {
-      "epoch": 4.882644228833689,
-      "grad_norm": 0.0028473767451941967,
-      "learning_rate": 0.00019998824405292634,
-      "loss": 46.0,
-      "step": 30319
-    },
-    {
-      "epoch": 4.882805265912476,
-      "grad_norm": 0.0022538998164236546,
-      "learning_rate": 0.00019998824327718907,
-      "loss": 46.0,
-      "step": 30320
-    },
-    {
-      "epoch": 4.882966302991264,
-      "grad_norm": 0.006421265192329884,
-      "learning_rate": 0.00019998824250142618,
-      "loss": 46.0,
-      "step": 30321
-    },
-    {
-      "epoch": 4.883127340070051,
-      "grad_norm": 0.02118576690554619,
-      "learning_rate": 0.0001999882417256377,
-      "loss": 46.0,
-      "step": 30322
-    },
-    {
-      "epoch": 4.883288377148839,
-      "grad_norm": 0.009731358848512173,
-      "learning_rate": 0.00019998824094982366,
-      "loss": 46.0,
-      "step": 30323
-    },
-    {
-      "epoch": 4.883449414227626,
-      "grad_norm": 0.0035421494394540787,
-      "learning_rate": 0.000199988240173984,
-      "loss": 46.0,
-      "step": 30324
-    },
-    {
-      "epoch": 4.883610451306414,
-      "grad_norm": 0.0027094639372080564,
-      "learning_rate": 0.00019998823939811877,
-      "loss": 46.0,
-      "step": 30325
-    },
-    {
-      "epoch": 4.883771488385201,
-      "grad_norm": 0.009699389338493347,
-      "learning_rate": 0.00019998823862222792,
-      "loss": 46.0,
-      "step": 30326
-    },
-    {
-      "epoch": 4.8839325254639885,
-      "grad_norm": 0.007638414856046438,
-      "learning_rate": 0.0001999882378463115,
-      "loss": 46.0,
-      "step": 30327
-    },
-    {
-      "epoch": 4.884093562542775,
-      "grad_norm": 0.012364639900624752,
-      "learning_rate": 0.00019998823707036948,
-      "loss": 46.0,
-      "step": 30328
-    },
-    {
-      "epoch": 4.8842545996215625,
-      "grad_norm": 0.003991544712334871,
-      "learning_rate": 0.0001999882362944019,
-      "loss": 46.0,
-      "step": 30329
-    },
-    {
-      "epoch": 4.88441563670035,
-      "grad_norm": 0.0011267001973465085,
-      "learning_rate": 0.0001999882355184087,
-      "loss": 46.0,
-      "step": 30330
-    },
-    {
-      "epoch": 4.884576673779137,
-      "grad_norm": 0.010877440683543682,
-      "learning_rate": 0.0001999882347423899,
-      "loss": 46.0,
-      "step": 30331
-    },
-    {
-      "epoch": 4.884737710857925,
-      "grad_norm": 0.0014555666130036116,
-      "learning_rate": 0.00019998823396634552,
-      "loss": 46.0,
-      "step": 30332
-    },
-    {
-      "epoch": 4.884898747936712,
-      "grad_norm": 0.008220436982810497,
-      "learning_rate": 0.00019998823319027553,
-      "loss": 46.0,
-      "step": 30333
-    },
-    {
-      "epoch": 4.8850597850155,
-      "grad_norm": 0.0010317128617316484,
-      "learning_rate": 0.00019998823241417995,
-      "loss": 46.0,
-      "step": 30334
-    },
-    {
-      "epoch": 4.885220822094287,
-      "grad_norm": 0.007349408231675625,
-      "learning_rate": 0.0001999882316380588,
-      "loss": 46.0,
-      "step": 30335
-    },
-    {
-      "epoch": 4.885381859173075,
-      "grad_norm": 0.004547438118606806,
-      "learning_rate": 0.00019998823086191206,
-      "loss": 46.0,
-      "step": 30336
-    },
-    {
-      "epoch": 4.885542896251862,
-      "grad_norm": 0.007054098881781101,
-      "learning_rate": 0.0001999882300857397,
-      "loss": 46.0,
-      "step": 30337
-    },
-    {
-      "epoch": 4.88570393333065,
-      "grad_norm": 0.005873312707990408,
-      "learning_rate": 0.00019998822930954177,
-      "loss": 46.0,
-      "step": 30338
-    },
-    {
-      "epoch": 4.885864970409437,
-      "grad_norm": 0.01052178256213665,
-      "learning_rate": 0.00019998822853331826,
-      "loss": 46.0,
-      "step": 30339
-    },
-    {
-      "epoch": 4.8860260074882245,
-      "grad_norm": 0.003539759898558259,
-      "learning_rate": 0.00019998822775706913,
-      "loss": 46.0,
-      "step": 30340
-    },
-    {
-      "epoch": 4.886187044567012,
-      "grad_norm": 0.005448522511869669,
-      "learning_rate": 0.0001999882269807944,
-      "loss": 46.0,
-      "step": 30341
-    },
-    {
-      "epoch": 4.886348081645799,
-      "grad_norm": 0.012556903064250946,
-      "learning_rate": 0.00019998822620449408,
-      "loss": 46.0,
-      "step": 30342
-    },
-    {
-      "epoch": 4.886509118724586,
-      "grad_norm": 0.00702200410887599,
-      "learning_rate": 0.0001999882254281682,
-      "loss": 46.0,
-      "step": 30343
-    },
-    {
-      "epoch": 4.886670155803373,
-      "grad_norm": 0.001821403275243938,
-      "learning_rate": 0.00019998822465181672,
-      "loss": 46.0,
-      "step": 30344
-    },
-    {
-      "epoch": 4.886831192882161,
-      "grad_norm": 0.001047898200340569,
-      "learning_rate": 0.00019998822387543963,
-      "loss": 46.0,
-      "step": 30345
-    },
-    {
-      "epoch": 4.886992229960948,
-      "grad_norm": 0.003435407066717744,
-      "learning_rate": 0.00019998822309903698,
-      "loss": 46.0,
-      "step": 30346
-    },
-    {
-      "epoch": 4.887153267039736,
-      "grad_norm": 0.0014386746333912015,
-      "learning_rate": 0.0001999882223226087,
-      "loss": 46.0,
-      "step": 30347
-    },
-    {
-      "epoch": 4.887314304118523,
-      "grad_norm": 0.006209913641214371,
-      "learning_rate": 0.00019998822154615486,
-      "loss": 46.0,
-      "step": 30348
-    },
-    {
-      "epoch": 4.887475341197311,
-      "grad_norm": 0.0028700323309749365,
-      "learning_rate": 0.0001999882207696754,
-      "loss": 46.0,
-      "step": 30349
-    },
-    {
-      "epoch": 4.887636378276098,
-      "grad_norm": 0.005138221196830273,
-      "learning_rate": 0.00019998821999317037,
-      "loss": 46.0,
-      "step": 30350
-    },
-    {
-      "epoch": 4.8877974153548855,
-      "grad_norm": 0.00448189489543438,
-      "learning_rate": 0.00019998821921663975,
-      "loss": 46.0,
-      "step": 30351
-    },
-    {
-      "epoch": 4.887958452433673,
-      "grad_norm": 0.004331518430262804,
-      "learning_rate": 0.0001999882184400835,
-      "loss": 46.0,
-      "step": 30352
-    },
-    {
-      "epoch": 4.8881194895124604,
-      "grad_norm": 0.0023716462310403585,
-      "learning_rate": 0.0001999882176635017,
-      "loss": 46.0,
-      "step": 30353
-    },
-    {
-      "epoch": 4.888280526591248,
-      "grad_norm": 0.002219144022092223,
-      "learning_rate": 0.00019998821688689428,
-      "loss": 46.0,
-      "step": 30354
-    },
-    {
-      "epoch": 4.888441563670035,
-      "grad_norm": 0.00531558133661747,
-      "learning_rate": 0.00019998821611026132,
-      "loss": 46.0,
-      "step": 30355
-    },
-    {
-      "epoch": 4.888602600748823,
-      "grad_norm": 0.0030300694052129984,
-      "learning_rate": 0.0001999882153336027,
-      "loss": 46.0,
-      "step": 30356
-    },
-    {
-      "epoch": 4.88876363782761,
-      "grad_norm": 0.0028026322834193707,
-      "learning_rate": 0.00019998821455691852,
-      "loss": 46.0,
-      "step": 30357
-    },
-    {
-      "epoch": 4.888924674906397,
-      "grad_norm": 0.011993777006864548,
-      "learning_rate": 0.00019998821378020874,
-      "loss": 46.0,
-      "step": 30358
-    },
-    {
-      "epoch": 4.889085711985184,
-      "grad_norm": 0.0022229549940675497,
-      "learning_rate": 0.0001999882130034734,
-      "loss": 46.0,
-      "step": 30359
-    },
-    {
-      "epoch": 4.889246749063972,
-      "grad_norm": 0.007727064657956362,
-      "learning_rate": 0.00019998821222671245,
-      "loss": 46.0,
-      "step": 30360
-    },
-    {
-      "epoch": 4.889407786142759,
-      "grad_norm": 0.006744609214365482,
-      "learning_rate": 0.0001999882114499259,
-      "loss": 46.0,
-      "step": 30361
-    },
-    {
-      "epoch": 4.889568823221547,
-      "grad_norm": 0.004535200539976358,
-      "learning_rate": 0.00019998821067311376,
-      "loss": 46.0,
-      "step": 30362
-    },
-    {
-      "epoch": 4.889729860300334,
-      "grad_norm": 0.008297115564346313,
-      "learning_rate": 0.00019998820989627602,
-      "loss": 46.0,
-      "step": 30363
-    },
-    {
-      "epoch": 4.8898908973791215,
-      "grad_norm": 0.0032407615799456835,
-      "learning_rate": 0.0001999882091194127,
-      "loss": 46.0,
-      "step": 30364
-    },
-    {
-      "epoch": 4.890051934457909,
-      "grad_norm": 0.002680008765310049,
-      "learning_rate": 0.0001999882083425238,
-      "loss": 46.0,
-      "step": 30365
-    },
-    {
-      "epoch": 4.890212971536696,
-      "grad_norm": 0.0033730214927345514,
-      "learning_rate": 0.00019998820756560924,
-      "loss": 46.0,
-      "step": 30366
-    },
-    {
-      "epoch": 4.890374008615484,
-      "grad_norm": 0.004781623836606741,
-      "learning_rate": 0.00019998820678866915,
-      "loss": 46.0,
-      "step": 30367
-    },
-    {
-      "epoch": 4.890535045694271,
-      "grad_norm": 0.005390890873968601,
-      "learning_rate": 0.00019998820601170347,
-      "loss": 46.0,
-      "step": 30368
-    },
-    {
-      "epoch": 4.890696082773059,
-      "grad_norm": 0.002060280181467533,
-      "learning_rate": 0.00019998820523471218,
-      "loss": 46.0,
-      "step": 30369
-    },
-    {
-      "epoch": 4.890857119851846,
-      "grad_norm": 0.006463841535151005,
-      "learning_rate": 0.00019998820445769533,
-      "loss": 46.0,
-      "step": 30370
-    },
-    {
-      "epoch": 4.891018156930633,
-      "grad_norm": 0.005537079647183418,
-      "learning_rate": 0.00019998820368065284,
-      "loss": 46.0,
-      "step": 30371
-    },
-    {
-      "epoch": 4.891179194009421,
-      "grad_norm": 0.006676413584500551,
-      "learning_rate": 0.00019998820290358479,
-      "loss": 46.0,
-      "step": 30372
-    },
-    {
-      "epoch": 4.891340231088208,
-      "grad_norm": 0.004183501936495304,
-      "learning_rate": 0.00019998820212649115,
-      "loss": 46.0,
-      "step": 30373
-    },
-    {
-      "epoch": 4.891501268166995,
-      "grad_norm": 0.002491398248821497,
-      "learning_rate": 0.00019998820134937192,
-      "loss": 46.0,
-      "step": 30374
-    },
-    {
-      "epoch": 4.891662305245783,
-      "grad_norm": 0.014169501140713692,
-      "learning_rate": 0.00019998820057222708,
-      "loss": 46.0,
-      "step": 30375
-    },
-    {
-      "epoch": 4.89182334232457,
-      "grad_norm": 0.0033617732115089893,
-      "learning_rate": 0.00019998819979505665,
-      "loss": 46.0,
-      "step": 30376
-    },
-    {
-      "epoch": 4.8919843794033575,
-      "grad_norm": 0.003594490699470043,
-      "learning_rate": 0.0001999881990178606,
-      "loss": 46.0,
-      "step": 30377
-    },
-    {
-      "epoch": 4.892145416482145,
-      "grad_norm": 0.003582483623176813,
-      "learning_rate": 0.000199988198240639,
-      "loss": 46.0,
-      "step": 30378
-    },
-    {
-      "epoch": 4.892306453560932,
-      "grad_norm": 0.003846733132377267,
-      "learning_rate": 0.0001999881974633918,
-      "loss": 46.0,
-      "step": 30379
-    },
-    {
-      "epoch": 4.89246749063972,
-      "grad_norm": 0.0031894822604954243,
-      "learning_rate": 0.000199988196686119,
-      "loss": 46.0,
-      "step": 30380
-    },
-    {
-      "epoch": 4.892628527718507,
-      "grad_norm": 0.021642964333295822,
-      "learning_rate": 0.00019998819590882064,
-      "loss": 46.0,
-      "step": 30381
-    },
-    {
-      "epoch": 4.892789564797295,
-      "grad_norm": 0.0023222407326102257,
-      "learning_rate": 0.00019998819513149664,
-      "loss": 46.0,
-      "step": 30382
-    },
-    {
-      "epoch": 4.892950601876082,
-      "grad_norm": 0.018712298944592476,
-      "learning_rate": 0.00019998819435414707,
-      "loss": 46.0,
-      "step": 30383
-    },
-    {
-      "epoch": 4.89311163895487,
-      "grad_norm": 0.009970182552933693,
-      "learning_rate": 0.00019998819357677192,
-      "loss": 46.0,
-      "step": 30384
-    },
-    {
-      "epoch": 4.893272676033657,
-      "grad_norm": 0.0011596559779718518,
-      "learning_rate": 0.00019998819279937115,
-      "loss": 46.0,
-      "step": 30385
-    },
-    {
-      "epoch": 4.893433713112444,
-      "grad_norm": 0.005341564305126667,
-      "learning_rate": 0.0001999881920219448,
-      "loss": 46.0,
-      "step": 30386
-    },
-    {
-      "epoch": 4.893594750191232,
-      "grad_norm": 0.0016422390472143888,
-      "learning_rate": 0.00019998819124449288,
-      "loss": 46.0,
-      "step": 30387
-    },
-    {
-      "epoch": 4.893755787270019,
-      "grad_norm": 0.002725841011852026,
-      "learning_rate": 0.00019998819046701535,
-      "loss": 46.0,
-      "step": 30388
-    },
-    {
-      "epoch": 4.893916824348806,
-      "grad_norm": 0.004834026098251343,
-      "learning_rate": 0.00019998818968951226,
-      "loss": 46.0,
-      "step": 30389
-    },
-    {
-      "epoch": 4.8940778614275935,
-      "grad_norm": 0.013006445951759815,
-      "learning_rate": 0.00019998818891198353,
-      "loss": 46.0,
-      "step": 30390
-    },
-    {
-      "epoch": 4.894238898506381,
-      "grad_norm": 0.005409400910139084,
-      "learning_rate": 0.00019998818813442922,
-      "loss": 46.0,
-      "step": 30391
-    },
-    {
-      "epoch": 4.894399935585168,
-      "grad_norm": 0.004750378895550966,
-      "learning_rate": 0.0001999881873568493,
-      "loss": 46.0,
-      "step": 30392
-    },
-    {
-      "epoch": 4.894560972663956,
-      "grad_norm": 0.001828087493777275,
-      "learning_rate": 0.00019998818657924385,
-      "loss": 46.0,
-      "step": 30393
-    },
-    {
-      "epoch": 4.894722009742743,
-      "grad_norm": 0.003092053811997175,
-      "learning_rate": 0.00019998818580161277,
-      "loss": 46.0,
-      "step": 30394
-    },
-    {
-      "epoch": 4.894883046821531,
-      "grad_norm": 0.0020606161560863256,
-      "learning_rate": 0.0001999881850239561,
-      "loss": 46.0,
-      "step": 30395
-    },
-    {
-      "epoch": 4.895044083900318,
-      "grad_norm": 0.003499452257528901,
-      "learning_rate": 0.00019998818424627382,
-      "loss": 46.0,
-      "step": 30396
-    },
-    {
-      "epoch": 4.895205120979106,
-      "grad_norm": 0.004869183525443077,
-      "learning_rate": 0.00019998818346856595,
-      "loss": 46.0,
-      "step": 30397
-    },
-    {
-      "epoch": 4.895366158057893,
-      "grad_norm": 0.011048783548176289,
-      "learning_rate": 0.00019998818269083252,
-      "loss": 46.0,
-      "step": 30398
-    },
-    {
-      "epoch": 4.8955271951366806,
-      "grad_norm": 0.005548267159610987,
-      "learning_rate": 0.00019998818191307348,
-      "loss": 46.0,
-      "step": 30399
-    },
-    {
-      "epoch": 4.895688232215468,
-      "grad_norm": 0.005866540130227804,
-      "learning_rate": 0.00019998818113528885,
-      "loss": 46.0,
-      "step": 30400
-    },
-    {
-      "epoch": 4.895849269294255,
-      "grad_norm": 0.0019323013257235289,
-      "learning_rate": 0.0001999881803574786,
-      "loss": 46.0,
-      "step": 30401
-    },
-    {
-      "epoch": 4.896010306373042,
-      "grad_norm": 0.005561450030654669,
-      "learning_rate": 0.00019998817957964283,
-      "loss": 46.0,
-      "step": 30402
-    },
-    {
-      "epoch": 4.8961713434518295,
-      "grad_norm": 0.013374553062021732,
-      "learning_rate": 0.0001999881788017814,
-      "loss": 46.0,
-      "step": 30403
-    },
-    {
-      "epoch": 4.896332380530617,
-      "grad_norm": 0.0043674553744494915,
-      "learning_rate": 0.0001999881780238944,
-      "loss": 46.0,
-      "step": 30404
-    },
-    {
-      "epoch": 4.896493417609404,
-      "grad_norm": 0.004610069561749697,
-      "learning_rate": 0.00019998817724598182,
-      "loss": 46.0,
-      "step": 30405
-    },
-    {
-      "epoch": 4.896654454688192,
-      "grad_norm": 0.0042014289647340775,
-      "learning_rate": 0.00019998817646804364,
-      "loss": 46.0,
-      "step": 30406
-    },
-    {
-      "epoch": 4.896815491766979,
-      "grad_norm": 0.005477037280797958,
-      "learning_rate": 0.00019998817569007987,
-      "loss": 46.0,
-      "step": 30407
-    },
-    {
-      "epoch": 4.896976528845767,
-      "grad_norm": 0.00274669355712831,
-      "learning_rate": 0.0001999881749120905,
-      "loss": 46.0,
-      "step": 30408
-    },
-    {
-      "epoch": 4.897137565924554,
-      "grad_norm": 0.0032716146670281887,
-      "learning_rate": 0.00019998817413407552,
-      "loss": 46.0,
-      "step": 30409
-    },
-    {
-      "epoch": 4.897298603003342,
-      "grad_norm": 0.003663887968286872,
-      "learning_rate": 0.000199988173356035,
-      "loss": 46.0,
-      "step": 30410
-    },
-    {
-      "epoch": 4.897459640082129,
-      "grad_norm": 0.0022849636152386665,
-      "learning_rate": 0.00019998817257796888,
-      "loss": 46.0,
-      "step": 30411
-    },
-    {
-      "epoch": 4.8976206771609165,
-      "grad_norm": 0.0036398752126842737,
-      "learning_rate": 0.00019998817179987712,
-      "loss": 46.0,
-      "step": 30412
-    },
-    {
-      "epoch": 4.897781714239704,
-      "grad_norm": 0.006911598611623049,
-      "learning_rate": 0.0001999881710217598,
-      "loss": 46.0,
-      "step": 30413
-    },
-    {
-      "epoch": 4.897942751318491,
-      "grad_norm": 0.007757857907563448,
-      "learning_rate": 0.00019998817024361688,
-      "loss": 46.0,
-      "step": 30414
-    },
-    {
-      "epoch": 4.898103788397279,
-      "grad_norm": 0.0018110171658918262,
-      "learning_rate": 0.00019998816946544836,
-      "loss": 46.0,
-      "step": 30415
-    },
-    {
-      "epoch": 4.8982648254760655,
-      "grad_norm": 0.008003453724086285,
-      "learning_rate": 0.00019998816868725428,
-      "loss": 46.0,
-      "step": 30416
-    },
-    {
-      "epoch": 4.898425862554853,
-      "grad_norm": 0.003984391689300537,
-      "learning_rate": 0.0001999881679090346,
-      "loss": 46.0,
-      "step": 30417
-    },
-    {
-      "epoch": 4.89858689963364,
-      "grad_norm": 0.0027316221967339516,
-      "learning_rate": 0.0001999881671307893,
-      "loss": 46.0,
-      "step": 30418
-    },
-    {
-      "epoch": 4.898747936712428,
-      "grad_norm": 0.006737871095538139,
-      "learning_rate": 0.0001999881663525184,
-      "loss": 46.0,
-      "step": 30419
-    },
-    {
-      "epoch": 4.898908973791215,
-      "grad_norm": 0.002492668339982629,
-      "learning_rate": 0.00019998816557422196,
-      "loss": 46.0,
-      "step": 30420
-    },
-    {
-      "epoch": 4.899070010870003,
-      "grad_norm": 0.002593198325484991,
-      "learning_rate": 0.0001999881647958999,
-      "loss": 46.0,
-      "step": 30421
-    },
-    {
-      "epoch": 4.89923104794879,
-      "grad_norm": 0.004081309773027897,
-      "learning_rate": 0.00019998816401755224,
-      "loss": 46.0,
-      "step": 30422
-    },
-    {
-      "epoch": 4.899392085027578,
-      "grad_norm": 0.0021048986818641424,
-      "learning_rate": 0.000199988163239179,
-      "loss": 46.0,
-      "step": 30423
-    },
-    {
-      "epoch": 4.899553122106365,
-      "grad_norm": 0.007963730953633785,
-      "learning_rate": 0.00019998816246078016,
-      "loss": 46.0,
-      "step": 30424
-    },
-    {
-      "epoch": 4.8997141591851525,
-      "grad_norm": 0.0086543383076787,
-      "learning_rate": 0.00019998816168235575,
-      "loss": 46.0,
-      "step": 30425
-    },
-    {
-      "epoch": 4.89987519626394,
-      "grad_norm": 0.0004533409664873034,
-      "learning_rate": 0.00019998816090390574,
-      "loss": 46.0,
-      "step": 30426
-    },
-    {
-      "epoch": 4.900036233342727,
-      "grad_norm": 0.0030244532972574234,
-      "learning_rate": 0.00019998816012543012,
-      "loss": 46.0,
-      "step": 30427
-    },
-    {
-      "epoch": 4.900197270421515,
-      "grad_norm": 0.0015222234651446342,
-      "learning_rate": 0.00019998815934692892,
-      "loss": 46.0,
-      "step": 30428
-    },
-    {
-      "epoch": 4.900358307500302,
-      "grad_norm": 0.0079489229246974,
-      "learning_rate": 0.00019998815856840213,
-      "loss": 46.0,
-      "step": 30429
-    },
-    {
-      "epoch": 4.90051934457909,
-      "grad_norm": 0.010724318213760853,
-      "learning_rate": 0.00019998815778984975,
-      "loss": 46.0,
-      "step": 30430
-    },
-    {
-      "epoch": 4.900680381657876,
-      "grad_norm": 0.015424591489136219,
-      "learning_rate": 0.00019998815701127172,
-      "loss": 46.0,
-      "step": 30431
-    },
-    {
-      "epoch": 4.900841418736664,
-      "grad_norm": 0.0312027707695961,
-      "learning_rate": 0.00019998815623266817,
-      "loss": 46.0,
-      "step": 30432
-    },
-    {
-      "epoch": 4.901002455815451,
-      "grad_norm": 0.0021878459956496954,
-      "learning_rate": 0.00019998815545403903,
-      "loss": 46.0,
-      "step": 30433
-    },
-    {
-      "epoch": 4.901163492894239,
-      "grad_norm": 0.0016812459798529744,
-      "learning_rate": 0.00019998815467538425,
-      "loss": 46.0,
-      "step": 30434
-    },
-    {
-      "epoch": 4.901324529973026,
-      "grad_norm": 0.00404763501137495,
-      "learning_rate": 0.00019998815389670393,
-      "loss": 46.0,
-      "step": 30435
-    },
-    {
-      "epoch": 4.901485567051814,
-      "grad_norm": 0.006528899073600769,
-      "learning_rate": 0.000199988153117998,
-      "loss": 46.0,
-      "step": 30436
-    },
-    {
-      "epoch": 4.901646604130601,
-      "grad_norm": 0.0030419824179261923,
-      "learning_rate": 0.00019998815233926646,
-      "loss": 46.0,
-      "step": 30437
-    },
-    {
-      "epoch": 4.9018076412093885,
-      "grad_norm": 0.00468186242505908,
-      "learning_rate": 0.00019998815156050932,
-      "loss": 46.0,
-      "step": 30438
-    },
-    {
-      "epoch": 4.901968678288176,
-      "grad_norm": 0.01168001163750887,
-      "learning_rate": 0.00019998815078172663,
-      "loss": 46.0,
-      "step": 30439
-    },
-    {
-      "epoch": 4.902129715366963,
-      "grad_norm": 0.004269199445843697,
-      "learning_rate": 0.00019998815000291833,
-      "loss": 46.0,
-      "step": 30440
-    },
-    {
-      "epoch": 4.902290752445751,
-      "grad_norm": 0.0019266594899818301,
-      "learning_rate": 0.0001999881492240844,
-      "loss": 46.0,
-      "step": 30441
-    },
-    {
-      "epoch": 4.902451789524538,
-      "grad_norm": 0.017323721200227737,
-      "learning_rate": 0.00019998814844522493,
-      "loss": 46.0,
-      "step": 30442
-    },
-    {
-      "epoch": 4.902612826603326,
-      "grad_norm": 0.0006327939918264747,
-      "learning_rate": 0.00019998814766633983,
-      "loss": 46.0,
-      "step": 30443
-    },
-    {
-      "epoch": 4.902773863682112,
-      "grad_norm": 0.0005349363200366497,
-      "learning_rate": 0.00019998814688742918,
-      "loss": 46.0,
-      "step": 30444
-    },
-    {
-      "epoch": 4.902934900760901,
-      "grad_norm": 0.0034096436575055122,
-      "learning_rate": 0.00019998814610849294,
-      "loss": 46.0,
-      "step": 30445
-    },
-    {
-      "epoch": 4.903095937839687,
-      "grad_norm": 0.01690039224922657,
-      "learning_rate": 0.00019998814532953105,
-      "loss": 46.0,
-      "step": 30446
-    },
-    {
-      "epoch": 4.903256974918475,
-      "grad_norm": 0.0039896718226373196,
-      "learning_rate": 0.0001999881445505436,
-      "loss": 46.0,
-      "step": 30447
-    },
-    {
-      "epoch": 4.903418011997262,
-      "grad_norm": 0.01275465078651905,
-      "learning_rate": 0.00019998814377153058,
-      "loss": 46.0,
-      "step": 30448
-    },
-    {
-      "epoch": 4.90357904907605,
-      "grad_norm": 0.009657173417508602,
-      "learning_rate": 0.00019998814299249196,
-      "loss": 46.0,
-      "step": 30449
-    },
-    {
-      "epoch": 4.903740086154837,
-      "grad_norm": 0.0028049226384609938,
-      "learning_rate": 0.00019998814221342773,
-      "loss": 46.0,
-      "step": 30450
-    },
-    {
-      "epoch": 4.9039011232336245,
-      "grad_norm": 0.004891689866781235,
-      "learning_rate": 0.0001999881414343379,
-      "loss": 46.0,
-      "step": 30451
-    },
-    {
-      "epoch": 4.904062160312412,
-      "grad_norm": 0.008742043748497963,
-      "learning_rate": 0.0001999881406552225,
-      "loss": 46.0,
-      "step": 30452
-    },
-    {
-      "epoch": 4.904223197391199,
-      "grad_norm": 0.005531701724976301,
-      "learning_rate": 0.00019998813987608149,
-      "loss": 46.0,
-      "step": 30453
-    },
-    {
-      "epoch": 4.904384234469987,
-      "grad_norm": 0.008188373409211636,
-      "learning_rate": 0.0001999881390969149,
-      "loss": 46.0,
-      "step": 30454
-    },
-    {
-      "epoch": 4.904545271548774,
-      "grad_norm": 0.013696274720132351,
-      "learning_rate": 0.00019998813831772274,
-      "loss": 46.0,
-      "step": 30455
-    },
-    {
-      "epoch": 4.904706308627562,
-      "grad_norm": 0.011236791498959064,
-      "learning_rate": 0.00019998813753850496,
-      "loss": 46.0,
-      "step": 30456
-    },
-    {
-      "epoch": 4.904867345706349,
-      "grad_norm": 0.007230163086205721,
-      "learning_rate": 0.00019998813675926159,
-      "loss": 46.0,
-      "step": 30457
-    },
-    {
-      "epoch": 4.905028382785137,
-      "grad_norm": 0.0028765499591827393,
-      "learning_rate": 0.00019998813597999263,
-      "loss": 46.0,
-      "step": 30458
-    },
-    {
-      "epoch": 4.905189419863923,
-      "grad_norm": 0.005269724875688553,
-      "learning_rate": 0.0001999881352006981,
-      "loss": 46.0,
-      "step": 30459
-    },
-    {
-      "epoch": 4.9053504569427115,
-      "grad_norm": 0.006079904735088348,
-      "learning_rate": 0.00019998813442137796,
-      "loss": 46.0,
-      "step": 30460
-    },
-    {
-      "epoch": 4.905511494021498,
-      "grad_norm": 0.0034910067915916443,
-      "learning_rate": 0.0001999881336420322,
-      "loss": 46.0,
-      "step": 30461
-    },
-    {
-      "epoch": 4.905672531100286,
-      "grad_norm": 0.0014764934312552214,
-      "learning_rate": 0.0001999881328626609,
-      "loss": 46.0,
-      "step": 30462
-    },
-    {
-      "epoch": 4.905833568179073,
-      "grad_norm": 0.025733614340424538,
-      "learning_rate": 0.000199988132083264,
-      "loss": 46.0,
-      "step": 30463
-    },
-    {
-      "epoch": 4.9059946052578605,
-      "grad_norm": 0.006812684237957001,
-      "learning_rate": 0.00019998813130384145,
-      "loss": 46.0,
-      "step": 30464
-    },
-    {
-      "epoch": 4.906155642336648,
-      "grad_norm": 0.00285906414501369,
-      "learning_rate": 0.00019998813052439336,
-      "loss": 46.0,
-      "step": 30465
-    },
-    {
-      "epoch": 4.906316679415435,
-      "grad_norm": 0.0029761367477476597,
-      "learning_rate": 0.00019998812974491968,
-      "loss": 46.0,
-      "step": 30466
-    },
-    {
-      "epoch": 4.906477716494223,
-      "grad_norm": 0.008120449259877205,
-      "learning_rate": 0.00019998812896542038,
-      "loss": 46.0,
-      "step": 30467
-    },
-    {
-      "epoch": 4.90663875357301,
-      "grad_norm": 0.00585969490930438,
-      "learning_rate": 0.00019998812818589553,
-      "loss": 46.0,
-      "step": 30468
-    },
-    {
-      "epoch": 4.906799790651798,
-      "grad_norm": 0.006378841586410999,
-      "learning_rate": 0.00019998812740634506,
-      "loss": 46.0,
-      "step": 30469
-    },
-    {
-      "epoch": 4.906960827730585,
-      "grad_norm": 0.0033306467812508345,
-      "learning_rate": 0.000199988126626769,
-      "loss": 46.0,
-      "step": 30470
-    },
-    {
-      "epoch": 4.907121864809373,
-      "grad_norm": 0.0010022582719102502,
-      "learning_rate": 0.00019998812584716733,
-      "loss": 46.0,
-      "step": 30471
-    },
-    {
-      "epoch": 4.90728290188816,
-      "grad_norm": 0.020576857030391693,
-      "learning_rate": 0.00019998812506754007,
-      "loss": 46.0,
-      "step": 30472
-    },
-    {
-      "epoch": 4.9074439389669475,
-      "grad_norm": 0.009301193989813328,
-      "learning_rate": 0.00019998812428788726,
-      "loss": 46.0,
-      "step": 30473
-    },
-    {
-      "epoch": 4.907604976045734,
-      "grad_norm": 0.01914830692112446,
-      "learning_rate": 0.00019998812350820882,
-      "loss": 46.0,
-      "step": 30474
-    },
-    {
-      "epoch": 4.9077660131245215,
-      "grad_norm": 0.0014181635342538357,
-      "learning_rate": 0.0001999881227285048,
-      "loss": 46.0,
-      "step": 30475
-    },
-    {
-      "epoch": 4.907927050203309,
-      "grad_norm": 0.005296367220580578,
-      "learning_rate": 0.0001999881219487752,
-      "loss": 46.0,
-      "step": 30476
-    },
-    {
-      "epoch": 4.9080880872820964,
-      "grad_norm": 0.005020085256546736,
-      "learning_rate": 0.00019998812116902,
-      "loss": 46.0,
-      "step": 30477
-    },
-    {
-      "epoch": 4.908249124360884,
-      "grad_norm": 0.0006146127707324922,
-      "learning_rate": 0.00019998812038923922,
-      "loss": 46.0,
-      "step": 30478
-    },
-    {
-      "epoch": 4.908410161439671,
-      "grad_norm": 0.0041870009154081345,
-      "learning_rate": 0.00019998811960943283,
-      "loss": 46.0,
-      "step": 30479
-    },
-    {
-      "epoch": 4.908571198518459,
-      "grad_norm": 0.001721163745969534,
-      "learning_rate": 0.00019998811882960085,
-      "loss": 46.0,
-      "step": 30480
-    },
-    {
-      "epoch": 4.908732235597246,
-      "grad_norm": 0.017062436789274216,
-      "learning_rate": 0.00019998811804974328,
-      "loss": 46.0,
-      "step": 30481
-    },
-    {
-      "epoch": 4.908893272676034,
-      "grad_norm": 0.0018941735615953803,
-      "learning_rate": 0.00019998811726986012,
-      "loss": 46.0,
-      "step": 30482
-    },
-    {
-      "epoch": 4.909054309754821,
-      "grad_norm": 0.0016774415271356702,
-      "learning_rate": 0.00019998811648995135,
-      "loss": 46.0,
-      "step": 30483
-    },
-    {
-      "epoch": 4.909215346833609,
-      "grad_norm": 0.012793129310011864,
-      "learning_rate": 0.00019998811571001702,
-      "loss": 46.0,
-      "step": 30484
-    },
-    {
-      "epoch": 4.909376383912396,
-      "grad_norm": 0.023086661472916603,
-      "learning_rate": 0.0001999881149300571,
-      "loss": 46.0,
-      "step": 30485
-    },
-    {
-      "epoch": 4.9095374209911835,
-      "grad_norm": 0.0045737880282104015,
-      "learning_rate": 0.00019998811415007154,
-      "loss": 46.0,
-      "step": 30486
-    },
-    {
-      "epoch": 4.909698458069971,
-      "grad_norm": 0.003975625149905682,
-      "learning_rate": 0.00019998811337006042,
-      "loss": 46.0,
-      "step": 30487
-    },
-    {
-      "epoch": 4.909859495148758,
-      "grad_norm": 0.005778086371719837,
-      "learning_rate": 0.0001999881125900237,
-      "loss": 46.0,
-      "step": 30488
-    },
-    {
-      "epoch": 4.910020532227545,
-      "grad_norm": 0.009252060204744339,
-      "learning_rate": 0.00019998811180996142,
-      "loss": 46.0,
-      "step": 30489
-    },
-    {
-      "epoch": 4.910181569306332,
-      "grad_norm": 0.00075905816629529,
-      "learning_rate": 0.0001999881110298735,
-      "loss": 46.0,
-      "step": 30490
-    },
-    {
-      "epoch": 4.91034260638512,
-      "grad_norm": 0.011408671736717224,
-      "learning_rate": 0.00019998811024976004,
-      "loss": 46.0,
-      "step": 30491
-    },
-    {
-      "epoch": 4.910503643463907,
-      "grad_norm": 0.006041468121111393,
-      "learning_rate": 0.00019998810946962093,
-      "loss": 46.0,
-      "step": 30492
-    },
-    {
-      "epoch": 4.910664680542695,
-      "grad_norm": 0.005167999770492315,
-      "learning_rate": 0.0001999881086894563,
-      "loss": 46.0,
-      "step": 30493
-    },
-    {
-      "epoch": 4.910825717621482,
-      "grad_norm": 0.011471273377537727,
-      "learning_rate": 0.00019998810790926603,
-      "loss": 46.0,
-      "step": 30494
-    },
-    {
-      "epoch": 4.91098675470027,
-      "grad_norm": 0.0029366835951805115,
-      "learning_rate": 0.00019998810712905016,
-      "loss": 46.0,
-      "step": 30495
-    },
-    {
-      "epoch": 4.911147791779057,
-      "grad_norm": 0.0074851494282484055,
-      "learning_rate": 0.0001999881063488087,
-      "loss": 46.0,
-      "step": 30496
-    },
-    {
-      "epoch": 4.911308828857845,
-      "grad_norm": 0.011722452938556671,
-      "learning_rate": 0.00019998810556854166,
-      "loss": 46.0,
-      "step": 30497
-    },
-    {
-      "epoch": 4.911469865936632,
-      "grad_norm": 0.0035150835756212473,
-      "learning_rate": 0.00019998810478824902,
-      "loss": 46.0,
-      "step": 30498
-    },
-    {
-      "epoch": 4.9116309030154195,
-      "grad_norm": 0.015447013080120087,
-      "learning_rate": 0.00019998810400793083,
-      "loss": 46.0,
-      "step": 30499
-    },
-    {
-      "epoch": 4.911791940094207,
-      "grad_norm": 0.002246592426672578,
-      "learning_rate": 0.000199988103227587,
-      "loss": 46.0,
-      "step": 30500
-    },
-    {
-      "epoch": 4.911952977172994,
-      "grad_norm": 0.001899111084640026,
-      "learning_rate": 0.00019998810244721763,
-      "loss": 46.0,
-      "step": 30501
-    },
-    {
-      "epoch": 4.912114014251782,
-      "grad_norm": 0.0031420388258993626,
-      "learning_rate": 0.0001999881016668226,
-      "loss": 46.0,
-      "step": 30502
-    },
-    {
-      "epoch": 4.912275051330569,
-      "grad_norm": 0.0027979863807559013,
-      "learning_rate": 0.000199988100886402,
-      "loss": 46.0,
-      "step": 30503
-    },
-    {
-      "epoch": 4.912436088409356,
-      "grad_norm": 0.003839815268293023,
-      "learning_rate": 0.00019998810010595584,
-      "loss": 46.0,
-      "step": 30504
-    },
-    {
-      "epoch": 4.912597125488143,
-      "grad_norm": 0.010382944718003273,
-      "learning_rate": 0.00019998809932548407,
-      "loss": 46.0,
-      "step": 30505
-    },
-    {
-      "epoch": 4.912758162566931,
-      "grad_norm": 0.0017271682154387236,
-      "learning_rate": 0.0001999880985449867,
-      "loss": 46.0,
-      "step": 30506
-    },
-    {
-      "epoch": 4.912919199645718,
-      "grad_norm": 0.006422657053917646,
-      "learning_rate": 0.00019998809776446374,
-      "loss": 46.0,
-      "step": 30507
-    },
-    {
-      "epoch": 4.913080236724506,
-      "grad_norm": 0.0037081504706293344,
-      "learning_rate": 0.00019998809698391519,
-      "loss": 46.0,
-      "step": 30508
-    },
-    {
-      "epoch": 4.913241273803293,
-      "grad_norm": 0.0043160878121852875,
-      "learning_rate": 0.00019998809620334104,
-      "loss": 46.0,
-      "step": 30509
-    },
-    {
-      "epoch": 4.913402310882081,
-      "grad_norm": 0.0044052740558981895,
-      "learning_rate": 0.00019998809542274133,
-      "loss": 46.0,
-      "step": 30510
-    },
-    {
-      "epoch": 4.913563347960868,
-      "grad_norm": 0.0050047775730490685,
-      "learning_rate": 0.00019998809464211599,
-      "loss": 46.0,
-      "step": 30511
-    },
-    {
-      "epoch": 4.9137243850396555,
-      "grad_norm": 0.009869280271232128,
-      "learning_rate": 0.00019998809386146505,
-      "loss": 46.0,
-      "step": 30512
-    },
-    {
-      "epoch": 4.913885422118443,
-      "grad_norm": 0.006626026704907417,
-      "learning_rate": 0.00019998809308078856,
-      "loss": 46.0,
-      "step": 30513
-    },
-    {
-      "epoch": 4.91404645919723,
-      "grad_norm": 0.004311206750571728,
-      "learning_rate": 0.00019998809230008648,
-      "loss": 46.0,
-      "step": 30514
-    },
-    {
-      "epoch": 4.914207496276018,
-      "grad_norm": 0.0018047093180939555,
-      "learning_rate": 0.00019998809151935878,
-      "loss": 46.0,
-      "step": 30515
-    },
-    {
-      "epoch": 4.914368533354805,
-      "grad_norm": 0.0034651169553399086,
-      "learning_rate": 0.0001999880907386055,
-      "loss": 46.0,
-      "step": 30516
-    },
-    {
-      "epoch": 4.914529570433592,
-      "grad_norm": 0.0027213257271796465,
-      "learning_rate": 0.00019998808995782663,
-      "loss": 46.0,
-      "step": 30517
-    },
-    {
-      "epoch": 4.91469060751238,
-      "grad_norm": 0.015327741391956806,
-      "learning_rate": 0.00019998808917702214,
-      "loss": 46.0,
-      "step": 30518
-    },
-    {
-      "epoch": 4.914851644591167,
-      "grad_norm": 0.003087273333221674,
-      "learning_rate": 0.0001999880883961921,
-      "loss": 46.0,
-      "step": 30519
-    },
-    {
-      "epoch": 4.915012681669954,
-      "grad_norm": 0.004218158312141895,
-      "learning_rate": 0.00019998808761533644,
-      "loss": 46.0,
-      "step": 30520
-    },
-    {
-      "epoch": 4.915173718748742,
-      "grad_norm": 0.008186622522771358,
-      "learning_rate": 0.0001999880868344552,
-      "loss": 46.0,
-      "step": 30521
-    },
-    {
-      "epoch": 4.915334755827529,
-      "grad_norm": 0.010806851089000702,
-      "learning_rate": 0.00019998808605354838,
-      "loss": 46.0,
-      "step": 30522
-    },
-    {
-      "epoch": 4.9154957929063166,
-      "grad_norm": 0.005453029647469521,
-      "learning_rate": 0.00019998808527261594,
-      "loss": 46.0,
-      "step": 30523
-    },
-    {
-      "epoch": 4.915656829985104,
-      "grad_norm": 0.001271772664040327,
-      "learning_rate": 0.0001999880844916579,
-      "loss": 46.0,
-      "step": 30524
-    },
-    {
-      "epoch": 4.9158178670638915,
-      "grad_norm": 0.0022089264821261168,
-      "learning_rate": 0.00019998808371067433,
-      "loss": 46.0,
-      "step": 30525
-    },
-    {
-      "epoch": 4.915978904142679,
-      "grad_norm": 0.003307298058643937,
-      "learning_rate": 0.0001999880829296651,
-      "loss": 46.0,
-      "step": 30526
-    },
-    {
-      "epoch": 4.916139941221466,
-      "grad_norm": 0.0011044098064303398,
-      "learning_rate": 0.00019998808214863033,
-      "loss": 46.0,
-      "step": 30527
-    },
-    {
-      "epoch": 4.916300978300254,
-      "grad_norm": 0.0006588261458091438,
-      "learning_rate": 0.00019998808136756995,
-      "loss": 46.0,
-      "step": 30528
-    },
-    {
-      "epoch": 4.916462015379041,
-      "grad_norm": 0.005067430902272463,
-      "learning_rate": 0.00019998808058648395,
-      "loss": 46.0,
-      "step": 30529
-    },
-    {
-      "epoch": 4.916623052457829,
-      "grad_norm": 0.016253668814897537,
-      "learning_rate": 0.0001999880798053724,
-      "loss": 46.0,
-      "step": 30530
-    },
-    {
-      "epoch": 4.916784089536616,
-      "grad_norm": 0.016832856461405754,
-      "learning_rate": 0.00019998807902423522,
-      "loss": 46.0,
-      "step": 30531
-    },
-    {
-      "epoch": 4.916945126615403,
-      "grad_norm": 0.01826532371342182,
-      "learning_rate": 0.00019998807824307249,
-      "loss": 46.0,
-      "step": 30532
-    },
-    {
-      "epoch": 4.917106163694191,
-      "grad_norm": 0.004080309998244047,
-      "learning_rate": 0.00019998807746188414,
-      "loss": 46.0,
-      "step": 30533
-    },
-    {
-      "epoch": 4.917267200772978,
-      "grad_norm": 0.010835053399205208,
-      "learning_rate": 0.00019998807668067018,
-      "loss": 46.0,
-      "step": 30534
-    },
-    {
-      "epoch": 4.917428237851765,
-      "grad_norm": 0.012455137446522713,
-      "learning_rate": 0.00019998807589943066,
-      "loss": 46.0,
-      "step": 30535
-    },
-    {
-      "epoch": 4.9175892749305525,
-      "grad_norm": 0.004329121671617031,
-      "learning_rate": 0.00019998807511816555,
-      "loss": 46.0,
-      "step": 30536
-    },
-    {
-      "epoch": 4.91775031200934,
-      "grad_norm": 0.004221884533762932,
-      "learning_rate": 0.00019998807433687486,
-      "loss": 46.0,
-      "step": 30537
-    },
-    {
-      "epoch": 4.917911349088127,
-      "grad_norm": 0.01379698421806097,
-      "learning_rate": 0.00019998807355555855,
-      "loss": 46.0,
-      "step": 30538
-    },
-    {
-      "epoch": 4.918072386166915,
-      "grad_norm": 0.008747509680688381,
-      "learning_rate": 0.00019998807277421662,
-      "loss": 46.0,
-      "step": 30539
-    },
-    {
-      "epoch": 4.918233423245702,
-      "grad_norm": 0.014827131293714046,
-      "learning_rate": 0.00019998807199284917,
-      "loss": 46.0,
-      "step": 30540
-    },
-    {
-      "epoch": 4.91839446032449,
-      "grad_norm": 0.006563459523022175,
-      "learning_rate": 0.0001999880712114561,
-      "loss": 46.0,
-      "step": 30541
-    },
-    {
-      "epoch": 4.918555497403277,
-      "grad_norm": 0.008277750574052334,
-      "learning_rate": 0.0001999880704300374,
-      "loss": 46.0,
-      "step": 30542
-    },
-    {
-      "epoch": 4.918716534482065,
-      "grad_norm": 0.0018741608364507556,
-      "learning_rate": 0.00019998806964859314,
-      "loss": 46.0,
-      "step": 30543
-    },
-    {
-      "epoch": 4.918877571560852,
-      "grad_norm": 0.0015922526363283396,
-      "learning_rate": 0.00019998806886712328,
-      "loss": 46.0,
-      "step": 30544
-    },
-    {
-      "epoch": 4.91903860863964,
-      "grad_norm": 0.004034386482089758,
-      "learning_rate": 0.00019998806808562786,
-      "loss": 46.0,
-      "step": 30545
-    },
-    {
-      "epoch": 4.919199645718427,
-      "grad_norm": 0.0011017619399353862,
-      "learning_rate": 0.0001999880673041068,
-      "loss": 46.0,
-      "step": 30546
-    },
-    {
-      "epoch": 4.919360682797214,
-      "grad_norm": 0.003767644288018346,
-      "learning_rate": 0.0001999880665225602,
-      "loss": 46.0,
-      "step": 30547
-    },
-    {
-      "epoch": 4.919521719876001,
-      "grad_norm": 0.0010831747204065323,
-      "learning_rate": 0.00019998806574098797,
-      "loss": 46.0,
-      "step": 30548
-    },
-    {
-      "epoch": 4.9196827569547885,
-      "grad_norm": 0.00435589998960495,
-      "learning_rate": 0.00019998806495939015,
-      "loss": 46.0,
-      "step": 30549
-    },
-    {
-      "epoch": 4.919843794033576,
-      "grad_norm": 0.002801221562549472,
-      "learning_rate": 0.00019998806417776674,
-      "loss": 46.0,
-      "step": 30550
-    },
-    {
-      "epoch": 4.920004831112363,
-      "grad_norm": 0.0012493854155763984,
-      "learning_rate": 0.00019998806339611777,
-      "loss": 46.0,
-      "step": 30551
-    },
-    {
-      "epoch": 4.920165868191151,
-      "grad_norm": 0.0074576702900230885,
-      "learning_rate": 0.00019998806261444315,
-      "loss": 46.0,
-      "step": 30552
-    },
-    {
-      "epoch": 4.920326905269938,
-      "grad_norm": 0.0026332444977015257,
-      "learning_rate": 0.00019998806183274296,
-      "loss": 46.0,
-      "step": 30553
-    },
-    {
-      "epoch": 4.920487942348726,
-      "grad_norm": 0.001999326515942812,
-      "learning_rate": 0.0001999880610510172,
-      "loss": 46.0,
-      "step": 30554
-    },
-    {
-      "epoch": 4.920648979427513,
-      "grad_norm": 0.0025490603875368834,
-      "learning_rate": 0.00019998806026926583,
-      "loss": 46.0,
-      "step": 30555
-    },
-    {
-      "epoch": 4.920810016506301,
-      "grad_norm": 0.008558065630495548,
-      "learning_rate": 0.00019998805948748887,
-      "loss": 46.0,
-      "step": 30556
-    },
-    {
-      "epoch": 4.920971053585088,
-      "grad_norm": 0.013050605542957783,
-      "learning_rate": 0.00019998805870568632,
-      "loss": 46.0,
-      "step": 30557
-    },
-    {
-      "epoch": 4.921132090663876,
-      "grad_norm": 0.0031150965951383114,
-      "learning_rate": 0.00019998805792385818,
-      "loss": 46.0,
-      "step": 30558
-    },
-    {
-      "epoch": 4.921293127742663,
-      "grad_norm": 0.0055672600865364075,
-      "learning_rate": 0.00019998805714200446,
-      "loss": 46.0,
-      "step": 30559
-    },
-    {
-      "epoch": 4.9214541648214505,
-      "grad_norm": 0.006387733854353428,
-      "learning_rate": 0.00019998805636012515,
-      "loss": 46.0,
-      "step": 30560
-    },
-    {
-      "epoch": 4.921615201900238,
-      "grad_norm": 0.01056070625782013,
-      "learning_rate": 0.00019998805557822023,
-      "loss": 46.0,
-      "step": 30561
-    },
-    {
-      "epoch": 4.9217762389790245,
-      "grad_norm": 0.00413871556520462,
-      "learning_rate": 0.0001999880547962897,
-      "loss": 46.0,
-      "step": 30562
-    },
-    {
-      "epoch": 4.921937276057812,
-      "grad_norm": 0.002815619111061096,
-      "learning_rate": 0.0001999880540143336,
-      "loss": 46.0,
-      "step": 30563
-    },
-    {
-      "epoch": 4.922098313136599,
-      "grad_norm": 0.011996772140264511,
-      "learning_rate": 0.00019998805323235191,
-      "loss": 46.0,
-      "step": 30564
-    },
-    {
-      "epoch": 4.922259350215387,
-      "grad_norm": 0.0037979588378220797,
-      "learning_rate": 0.00019998805245034464,
-      "loss": 46.0,
-      "step": 30565
-    },
-    {
-      "epoch": 4.922420387294174,
-      "grad_norm": 0.0024226531386375427,
-      "learning_rate": 0.00019998805166831178,
-      "loss": 46.0,
-      "step": 30566
-    },
-    {
-      "epoch": 4.922581424372962,
-      "grad_norm": 0.0066368901170790195,
-      "learning_rate": 0.0001999880508862533,
-      "loss": 46.0,
-      "step": 30567
-    },
-    {
-      "epoch": 4.922742461451749,
-      "grad_norm": 0.0017153420485556126,
-      "learning_rate": 0.00019998805010416922,
-      "loss": 46.0,
-      "step": 30568
-    },
-    {
-      "epoch": 4.922903498530537,
-      "grad_norm": 0.008662872947752476,
-      "learning_rate": 0.0001999880493220596,
-      "loss": 46.0,
-      "step": 30569
-    },
-    {
-      "epoch": 4.923064535609324,
-      "grad_norm": 0.005743962712585926,
-      "learning_rate": 0.00019998804853992437,
-      "loss": 46.0,
-      "step": 30570
-    },
-    {
-      "epoch": 4.923225572688112,
-      "grad_norm": 0.004111960995942354,
-      "learning_rate": 0.00019998804775776352,
-      "loss": 46.0,
-      "step": 30571
-    },
-    {
-      "epoch": 4.923386609766899,
-      "grad_norm": 0.015256236307322979,
-      "learning_rate": 0.00019998804697557708,
-      "loss": 46.0,
-      "step": 30572
-    },
-    {
-      "epoch": 4.9235476468456865,
-      "grad_norm": 0.005067076068371534,
-      "learning_rate": 0.0001999880461933651,
-      "loss": 46.0,
-      "step": 30573
-    },
-    {
-      "epoch": 4.923708683924474,
-      "grad_norm": 0.0018835312221199274,
-      "learning_rate": 0.00019998804541112748,
-      "loss": 46.0,
-      "step": 30574
-    },
-    {
-      "epoch": 4.923869721003261,
-      "grad_norm": 0.009835692122578621,
-      "learning_rate": 0.00019998804462886428,
-      "loss": 46.0,
-      "step": 30575
-    },
-    {
-      "epoch": 4.924030758082049,
-      "grad_norm": 0.01062731072306633,
-      "learning_rate": 0.00019998804384657547,
-      "loss": 46.0,
-      "step": 30576
-    },
-    {
-      "epoch": 4.924191795160835,
-      "grad_norm": 0.0022026204969733953,
-      "learning_rate": 0.0001999880430642611,
-      "loss": 46.0,
-      "step": 30577
-    },
-    {
-      "epoch": 4.924352832239623,
-      "grad_norm": 0.009752246551215649,
-      "learning_rate": 0.0001999880422819211,
-      "loss": 46.0,
-      "step": 30578
-    },
-    {
-      "epoch": 4.92451386931841,
-      "grad_norm": 0.007753752637654543,
-      "learning_rate": 0.00019998804149955554,
-      "loss": 46.0,
-      "step": 30579
-    },
-    {
-      "epoch": 4.924674906397198,
-      "grad_norm": 0.0047276997938752174,
-      "learning_rate": 0.0001999880407171644,
-      "loss": 46.0,
-      "step": 30580
-    },
-    {
-      "epoch": 4.924835943475985,
-      "grad_norm": 0.00304826139472425,
-      "learning_rate": 0.00019998803993474763,
-      "loss": 46.0,
-      "step": 30581
-    },
-    {
-      "epoch": 4.924996980554773,
-      "grad_norm": 0.007551336195319891,
-      "learning_rate": 0.0001999880391523053,
-      "loss": 46.0,
-      "step": 30582
-    },
-    {
-      "epoch": 4.92515801763356,
-      "grad_norm": 0.012198468670248985,
-      "learning_rate": 0.00019998803836983737,
-      "loss": 46.0,
-      "step": 30583
-    },
-    {
-      "epoch": 4.9253190547123475,
-      "grad_norm": 0.006854066625237465,
-      "learning_rate": 0.00019998803758734384,
-      "loss": 46.0,
-      "step": 30584
-    },
-    {
-      "epoch": 4.925480091791135,
-      "grad_norm": 0.002870082389563322,
-      "learning_rate": 0.00019998803680482471,
-      "loss": 46.0,
-      "step": 30585
-    },
-    {
-      "epoch": 4.9256411288699224,
-      "grad_norm": 0.010755311697721481,
-      "learning_rate": 0.00019998803602228003,
-      "loss": 46.0,
-      "step": 30586
-    },
-    {
-      "epoch": 4.92580216594871,
-      "grad_norm": 0.0014694230630993843,
-      "learning_rate": 0.0001999880352397097,
-      "loss": 46.0,
-      "step": 30587
-    },
-    {
-      "epoch": 4.925963203027497,
-      "grad_norm": 0.01283201202750206,
-      "learning_rate": 0.0001999880344571138,
-      "loss": 46.0,
-      "step": 30588
-    },
-    {
-      "epoch": 4.926124240106285,
-      "grad_norm": 0.002975200302898884,
-      "learning_rate": 0.00019998803367449232,
-      "loss": 46.0,
-      "step": 30589
-    },
-    {
-      "epoch": 4.926285277185071,
-      "grad_norm": 0.015369229018688202,
-      "learning_rate": 0.00019998803289184526,
-      "loss": 46.0,
-      "step": 30590
-    },
-    {
-      "epoch": 4.92644631426386,
-      "grad_norm": 0.002160497009754181,
-      "learning_rate": 0.0001999880321091726,
-      "loss": 46.0,
-      "step": 30591
-    },
-    {
-      "epoch": 4.926607351342646,
-      "grad_norm": 0.0012843343429267406,
-      "learning_rate": 0.00019998803132647433,
-      "loss": 46.0,
-      "step": 30592
-    },
-    {
-      "epoch": 4.926768388421434,
-      "grad_norm": 0.007385102566331625,
-      "learning_rate": 0.00019998803054375048,
-      "loss": 46.0,
-      "step": 30593
-    },
-    {
-      "epoch": 4.926929425500221,
-      "grad_norm": 0.0009941896423697472,
-      "learning_rate": 0.00019998802976100102,
-      "loss": 46.0,
-      "step": 30594
-    },
-    {
-      "epoch": 4.927090462579009,
-      "grad_norm": 0.0013169351732358336,
-      "learning_rate": 0.00019998802897822597,
-      "loss": 46.0,
-      "step": 30595
-    },
-    {
-      "epoch": 4.927251499657796,
-      "grad_norm": 0.010834407061338425,
-      "learning_rate": 0.00019998802819542533,
-      "loss": 46.0,
-      "step": 30596
-    },
-    {
-      "epoch": 4.9274125367365835,
-      "grad_norm": 0.002388420980423689,
-      "learning_rate": 0.00019998802741259914,
-      "loss": 46.0,
-      "step": 30597
-    },
-    {
-      "epoch": 4.927573573815371,
-      "grad_norm": 0.004795580171048641,
-      "learning_rate": 0.00019998802662974733,
-      "loss": 46.0,
-      "step": 30598
-    },
-    {
-      "epoch": 4.927734610894158,
-      "grad_norm": 0.002790645696222782,
-      "learning_rate": 0.0001999880258468699,
-      "loss": 46.0,
-      "step": 30599
-    },
-    {
-      "epoch": 4.927895647972946,
-      "grad_norm": 0.0010824664495885372,
-      "learning_rate": 0.00019998802506396692,
-      "loss": 46.0,
-      "step": 30600
-    },
-    {
-      "epoch": 4.928056685051733,
-      "grad_norm": 0.004720304161310196,
-      "learning_rate": 0.00019998802428103834,
-      "loss": 46.0,
-      "step": 30601
-    },
-    {
-      "epoch": 4.928217722130521,
-      "grad_norm": 0.005735001061111689,
-      "learning_rate": 0.00019998802349808416,
-      "loss": 46.0,
-      "step": 30602
-    },
-    {
-      "epoch": 4.928378759209308,
-      "grad_norm": 0.00919886864721775,
-      "learning_rate": 0.00019998802271510438,
-      "loss": 46.0,
-      "step": 30603
-    },
-    {
-      "epoch": 4.928539796288096,
-      "grad_norm": 0.0024084672331809998,
-      "learning_rate": 0.00019998802193209905,
-      "loss": 46.0,
-      "step": 30604
-    },
-    {
-      "epoch": 4.928700833366882,
-      "grad_norm": 0.0016177950892597437,
-      "learning_rate": 0.00019998802114906808,
-      "loss": 46.0,
-      "step": 30605
-    },
-    {
-      "epoch": 4.928861870445671,
-      "grad_norm": 0.011064939200878143,
-      "learning_rate": 0.0001999880203660115,
-      "loss": 46.0,
-      "step": 30606
-    },
-    {
-      "epoch": 4.929022907524457,
-      "grad_norm": 0.030213357880711555,
-      "learning_rate": 0.0001999880195829294,
-      "loss": 46.0,
-      "step": 30607
-    },
-    {
-      "epoch": 4.929183944603245,
-      "grad_norm": 0.004055551718920469,
-      "learning_rate": 0.00019998801879982168,
-      "loss": 46.0,
-      "step": 30608
-    },
-    {
-      "epoch": 4.929344981682032,
-      "grad_norm": 0.01130355428904295,
-      "learning_rate": 0.00019998801801668833,
-      "loss": 46.0,
-      "step": 30609
-    },
-    {
-      "epoch": 4.9295060187608195,
-      "grad_norm": 0.003402838483452797,
-      "learning_rate": 0.00019998801723352942,
-      "loss": 46.0,
-      "step": 30610
-    },
-    {
-      "epoch": 4.929667055839607,
-      "grad_norm": 0.006314252037554979,
-      "learning_rate": 0.00019998801645034492,
-      "loss": 46.0,
-      "step": 30611
-    },
-    {
-      "epoch": 4.929828092918394,
-      "grad_norm": 0.010180431418120861,
-      "learning_rate": 0.0001999880156671348,
-      "loss": 46.0,
-      "step": 30612
-    },
-    {
-      "epoch": 4.929989129997182,
-      "grad_norm": 0.0019255392253398895,
-      "learning_rate": 0.00019998801488389914,
-      "loss": 46.0,
-      "step": 30613
-    },
-    {
-      "epoch": 4.930150167075969,
-      "grad_norm": 0.008755774237215519,
-      "learning_rate": 0.00019998801410063785,
-      "loss": 46.0,
-      "step": 30614
-    },
-    {
-      "epoch": 4.930311204154757,
-      "grad_norm": 0.011707756668329239,
-      "learning_rate": 0.00019998801331735098,
-      "loss": 46.0,
-      "step": 30615
-    },
-    {
-      "epoch": 4.930472241233544,
-      "grad_norm": 0.004657205194234848,
-      "learning_rate": 0.00019998801253403851,
-      "loss": 46.0,
-      "step": 30616
-    },
-    {
-      "epoch": 4.930633278312332,
-      "grad_norm": 0.0028296730015426874,
-      "learning_rate": 0.00019998801175070047,
-      "loss": 46.0,
-      "step": 30617
-    },
-    {
-      "epoch": 4.930794315391119,
-      "grad_norm": 0.0026236700359731913,
-      "learning_rate": 0.0001999880109673368,
-      "loss": 46.0,
-      "step": 30618
-    },
-    {
-      "epoch": 4.930955352469907,
-      "grad_norm": 0.0018111473182216287,
-      "learning_rate": 0.00019998801018394758,
-      "loss": 46.0,
-      "step": 30619
-    },
-    {
-      "epoch": 4.931116389548693,
-      "grad_norm": 0.003679076209664345,
-      "learning_rate": 0.00019998800940053272,
-      "loss": 46.0,
-      "step": 30620
-    },
-    {
-      "epoch": 4.9312774266274815,
-      "grad_norm": 0.012358995154500008,
-      "learning_rate": 0.0001999880086170923,
-      "loss": 46.0,
-      "step": 30621
-    },
-    {
-      "epoch": 4.931438463706268,
-      "grad_norm": 0.0016086430987343192,
-      "learning_rate": 0.0001999880078336263,
-      "loss": 46.0,
-      "step": 30622
-    },
-    {
-      "epoch": 4.9315995007850555,
-      "grad_norm": 0.0025007708463817835,
-      "learning_rate": 0.00019998800705013468,
-      "loss": 46.0,
-      "step": 30623
-    },
-    {
-      "epoch": 4.931760537863843,
-      "grad_norm": 0.004497797694057226,
-      "learning_rate": 0.0001999880062666175,
-      "loss": 46.0,
-      "step": 30624
-    },
-    {
-      "epoch": 4.93192157494263,
-      "grad_norm": 0.005750203039497137,
-      "learning_rate": 0.0001999880054830747,
-      "loss": 46.0,
-      "step": 30625
-    },
-    {
-      "epoch": 4.932082612021418,
-      "grad_norm": 0.0030960417352616787,
-      "learning_rate": 0.0001999880046995063,
-      "loss": 46.0,
-      "step": 30626
-    },
-    {
-      "epoch": 4.932243649100205,
-      "grad_norm": 0.0053208451718091965,
-      "learning_rate": 0.00019998800391591234,
-      "loss": 46.0,
-      "step": 30627
-    },
-    {
-      "epoch": 4.932404686178993,
-      "grad_norm": 0.013970322906970978,
-      "learning_rate": 0.0001999880031322928,
-      "loss": 46.0,
-      "step": 30628
-    },
-    {
-      "epoch": 4.93256572325778,
-      "grad_norm": 0.003336176509037614,
-      "learning_rate": 0.00019998800234864763,
-      "loss": 46.0,
-      "step": 30629
-    },
-    {
-      "epoch": 4.932726760336568,
-      "grad_norm": 0.004502678755670786,
-      "learning_rate": 0.0001999880015649769,
-      "loss": 46.0,
-      "step": 30630
-    },
-    {
-      "epoch": 4.932887797415355,
-      "grad_norm": 0.003126248251646757,
-      "learning_rate": 0.00019998800078128051,
-      "loss": 46.0,
-      "step": 30631
-    },
-    {
-      "epoch": 4.9330488344941426,
-      "grad_norm": 0.004403343889862299,
-      "learning_rate": 0.0001999879999975586,
-      "loss": 46.0,
-      "step": 30632
-    },
-    {
-      "epoch": 4.93320987157293,
-      "grad_norm": 0.004321679472923279,
-      "learning_rate": 0.00019998799921381108,
-      "loss": 46.0,
-      "step": 30633
-    },
-    {
-      "epoch": 4.9333709086517175,
-      "grad_norm": 0.007235975936055183,
-      "learning_rate": 0.00019998799843003794,
-      "loss": 46.0,
-      "step": 30634
-    },
-    {
-      "epoch": 4.933531945730504,
-      "grad_norm": 0.0024102397728711367,
-      "learning_rate": 0.00019998799764623924,
-      "loss": 46.0,
-      "step": 30635
-    },
-    {
-      "epoch": 4.9336929828092915,
-      "grad_norm": 0.0024136831052601337,
-      "learning_rate": 0.00019998799686241493,
-      "loss": 46.0,
-      "step": 30636
-    },
-    {
-      "epoch": 4.933854019888079,
-      "grad_norm": 0.003616488305851817,
-      "learning_rate": 0.00019998799607856506,
-      "loss": 46.0,
-      "step": 30637
-    },
-    {
-      "epoch": 4.934015056966866,
-      "grad_norm": 0.0038302731700241566,
-      "learning_rate": 0.00019998799529468957,
-      "loss": 46.0,
-      "step": 30638
-    },
-    {
-      "epoch": 4.934176094045654,
-      "grad_norm": 0.0060085938312113285,
-      "learning_rate": 0.0001999879945107885,
-      "loss": 46.0,
-      "step": 30639
-    },
-    {
-      "epoch": 4.934337131124441,
-      "grad_norm": 0.008247928693890572,
-      "learning_rate": 0.00019998799372686183,
-      "loss": 46.0,
-      "step": 30640
-    },
-    {
-      "epoch": 4.934498168203229,
-      "grad_norm": 0.0027552242390811443,
-      "learning_rate": 0.00019998799294290958,
-      "loss": 46.0,
-      "step": 30641
-    },
-    {
-      "epoch": 4.934659205282016,
-      "grad_norm": 0.013369555585086346,
-      "learning_rate": 0.00019998799215893175,
-      "loss": 46.0,
-      "step": 30642
-    },
-    {
-      "epoch": 4.934820242360804,
-      "grad_norm": 0.020754095166921616,
-      "learning_rate": 0.0001999879913749283,
-      "loss": 46.0,
-      "step": 30643
-    },
-    {
-      "epoch": 4.934981279439591,
-      "grad_norm": 0.006118000019341707,
-      "learning_rate": 0.00019998799059089923,
-      "loss": 46.0,
-      "step": 30644
-    },
-    {
-      "epoch": 4.9351423165183785,
-      "grad_norm": 0.003953027073293924,
-      "learning_rate": 0.00019998798980684464,
-      "loss": 46.0,
-      "step": 30645
-    },
-    {
-      "epoch": 4.935303353597166,
-      "grad_norm": 0.006661976221948862,
-      "learning_rate": 0.00019998798902276442,
-      "loss": 46.0,
-      "step": 30646
-    },
-    {
-      "epoch": 4.935464390675953,
-      "grad_norm": 0.0008577602566219866,
-      "learning_rate": 0.0001999879882386586,
-      "loss": 46.0,
-      "step": 30647
-    },
-    {
-      "epoch": 4.935625427754741,
-      "grad_norm": 0.0018127582734450698,
-      "learning_rate": 0.00019998798745452718,
-      "loss": 46.0,
-      "step": 30648
-    },
-    {
-      "epoch": 4.935786464833528,
-      "grad_norm": 0.0017990750493481755,
-      "learning_rate": 0.00019998798667037018,
-      "loss": 46.0,
-      "step": 30649
-    },
-    {
-      "epoch": 4.935947501912315,
-      "grad_norm": 0.002674547955393791,
-      "learning_rate": 0.00019998798588618762,
-      "loss": 46.0,
-      "step": 30650
-    },
-    {
-      "epoch": 4.936108538991102,
-      "grad_norm": 0.0059504699893295765,
-      "learning_rate": 0.00019998798510197942,
-      "loss": 46.0,
-      "step": 30651
-    },
-    {
-      "epoch": 4.93626957606989,
-      "grad_norm": 0.005228621419519186,
-      "learning_rate": 0.0001999879843177457,
-      "loss": 46.0,
-      "step": 30652
-    },
-    {
-      "epoch": 4.936430613148677,
-      "grad_norm": 0.008515691384673119,
-      "learning_rate": 0.0001999879835334863,
-      "loss": 46.0,
-      "step": 30653
-    },
-    {
-      "epoch": 4.936591650227465,
-      "grad_norm": 0.001437501166947186,
-      "learning_rate": 0.00019998798274920137,
-      "loss": 46.0,
-      "step": 30654
-    },
-    {
-      "epoch": 4.936752687306252,
-      "grad_norm": 0.0013436836889013648,
-      "learning_rate": 0.0001999879819648908,
-      "loss": 46.0,
-      "step": 30655
-    },
-    {
-      "epoch": 4.93691372438504,
-      "grad_norm": 0.033359453082084656,
-      "learning_rate": 0.00019998798118055469,
-      "loss": 46.0,
-      "step": 30656
-    },
-    {
-      "epoch": 4.937074761463827,
-      "grad_norm": 0.008371630683541298,
-      "learning_rate": 0.00019998798039619296,
-      "loss": 46.0,
-      "step": 30657
-    },
-    {
-      "epoch": 4.9372357985426145,
-      "grad_norm": 0.004646202549338341,
-      "learning_rate": 0.00019998797961180562,
-      "loss": 46.0,
-      "step": 30658
-    },
-    {
-      "epoch": 4.937396835621402,
-      "grad_norm": 0.002892756136134267,
-      "learning_rate": 0.0001999879788273927,
-      "loss": 46.0,
-      "step": 30659
-    },
-    {
-      "epoch": 4.937557872700189,
-      "grad_norm": 0.0036805877462029457,
-      "learning_rate": 0.00019998797804295418,
-      "loss": 46.0,
-      "step": 30660
-    },
-    {
-      "epoch": 4.937718909778977,
-      "grad_norm": 0.009498110972344875,
-      "learning_rate": 0.0001999879772584901,
-      "loss": 46.0,
-      "step": 30661
-    },
-    {
-      "epoch": 4.937879946857764,
-      "grad_norm": 0.010308600030839443,
-      "learning_rate": 0.00019998797647400042,
-      "loss": 46.0,
-      "step": 30662
-    },
-    {
-      "epoch": 4.938040983936552,
-      "grad_norm": 0.0038641702849417925,
-      "learning_rate": 0.00019998797568948512,
-      "loss": 46.0,
-      "step": 30663
-    },
-    {
-      "epoch": 4.938202021015339,
-      "grad_norm": 0.0046220156364142895,
-      "learning_rate": 0.00019998797490494428,
-      "loss": 46.0,
-      "step": 30664
-    },
-    {
-      "epoch": 4.938363058094126,
-      "grad_norm": 0.000665553321596235,
-      "learning_rate": 0.0001999879741203778,
-      "loss": 46.0,
-      "step": 30665
-    },
-    {
-      "epoch": 4.938524095172913,
-      "grad_norm": 0.0027863301802426577,
-      "learning_rate": 0.00019998797333578577,
-      "loss": 46.0,
-      "step": 30666
-    },
-    {
-      "epoch": 4.938685132251701,
-      "grad_norm": 0.00626972783356905,
-      "learning_rate": 0.0001999879725511681,
-      "loss": 46.0,
-      "step": 30667
-    },
-    {
-      "epoch": 4.938846169330488,
-      "grad_norm": 0.001908987876959145,
-      "learning_rate": 0.00019998797176652486,
-      "loss": 46.0,
-      "step": 30668
-    },
-    {
-      "epoch": 4.939007206409276,
-      "grad_norm": 0.012596477754414082,
-      "learning_rate": 0.00019998797098185603,
-      "loss": 46.0,
-      "step": 30669
-    },
-    {
-      "epoch": 4.939168243488063,
-      "grad_norm": 0.00546701205894351,
-      "learning_rate": 0.00019998797019716162,
-      "loss": 46.0,
-      "step": 30670
-    },
-    {
-      "epoch": 4.9393292805668505,
-      "grad_norm": 0.010783938691020012,
-      "learning_rate": 0.00019998796941244162,
-      "loss": 46.0,
-      "step": 30671
-    },
-    {
-      "epoch": 4.939490317645638,
-      "grad_norm": 0.0013946013059467077,
-      "learning_rate": 0.00019998796862769603,
-      "loss": 46.0,
-      "step": 30672
-    },
-    {
-      "epoch": 4.939651354724425,
-      "grad_norm": 0.0008063954301178455,
-      "learning_rate": 0.0001999879678429248,
-      "loss": 46.0,
-      "step": 30673
-    },
-    {
-      "epoch": 4.939812391803213,
-      "grad_norm": 0.001811600406654179,
-      "learning_rate": 0.00019998796705812804,
-      "loss": 46.0,
-      "step": 30674
-    },
-    {
-      "epoch": 4.939973428882,
-      "grad_norm": 0.007117320317775011,
-      "learning_rate": 0.00019998796627330564,
-      "loss": 46.0,
-      "step": 30675
-    },
-    {
-      "epoch": 4.940134465960788,
-      "grad_norm": 0.002142674755305052,
-      "learning_rate": 0.00019998796548845768,
-      "loss": 46.0,
-      "step": 30676
-    },
-    {
-      "epoch": 4.940295503039575,
-      "grad_norm": 0.00502261146903038,
-      "learning_rate": 0.0001999879647035841,
-      "loss": 46.0,
-      "step": 30677
-    },
-    {
-      "epoch": 4.940456540118362,
-      "grad_norm": 0.0016893288120627403,
-      "learning_rate": 0.00019998796391868496,
-      "loss": 46.0,
-      "step": 30678
-    },
-    {
-      "epoch": 4.94061757719715,
-      "grad_norm": 0.012884542346000671,
-      "learning_rate": 0.0001999879631337602,
-      "loss": 46.0,
-      "step": 30679
-    },
-    {
-      "epoch": 4.940778614275937,
-      "grad_norm": 0.0037730729673057795,
-      "learning_rate": 0.00019998796234880985,
-      "loss": 46.0,
-      "step": 30680
-    },
-    {
-      "epoch": 4.940939651354724,
-      "grad_norm": 0.009059794247150421,
-      "learning_rate": 0.00019998796156383395,
-      "loss": 46.0,
-      "step": 30681
-    },
-    {
-      "epoch": 4.941100688433512,
-      "grad_norm": 0.005041288211941719,
-      "learning_rate": 0.0001999879607788324,
-      "loss": 46.0,
-      "step": 30682
-    },
-    {
-      "epoch": 4.941261725512299,
-      "grad_norm": 0.0025501365307718515,
-      "learning_rate": 0.0001999879599938053,
-      "loss": 46.0,
-      "step": 30683
-    },
-    {
-      "epoch": 4.9414227625910865,
-      "grad_norm": 0.003085311036556959,
-      "learning_rate": 0.00019998795920875257,
-      "loss": 46.0,
-      "step": 30684
-    },
-    {
-      "epoch": 4.941583799669874,
-      "grad_norm": 0.001115269260481,
-      "learning_rate": 0.0001999879584236743,
-      "loss": 46.0,
-      "step": 30685
-    },
-    {
-      "epoch": 4.941744836748661,
-      "grad_norm": 0.006796635687351227,
-      "learning_rate": 0.0001999879576385704,
-      "loss": 46.0,
-      "step": 30686
-    },
-    {
-      "epoch": 4.941905873827449,
-      "grad_norm": 0.011606014333665371,
-      "learning_rate": 0.0001999879568534409,
-      "loss": 46.0,
-      "step": 30687
-    },
-    {
-      "epoch": 4.942066910906236,
-      "grad_norm": 0.019449952989816666,
-      "learning_rate": 0.00019998795606828584,
-      "loss": 46.0,
-      "step": 30688
-    },
-    {
-      "epoch": 4.942227947985024,
-      "grad_norm": 0.006146786734461784,
-      "learning_rate": 0.00019998795528310516,
-      "loss": 46.0,
-      "step": 30689
-    },
-    {
-      "epoch": 4.942388985063811,
-      "grad_norm": 0.00306873582303524,
-      "learning_rate": 0.0001999879544978989,
-      "loss": 46.0,
-      "step": 30690
-    },
-    {
-      "epoch": 4.942550022142599,
-      "grad_norm": 0.002787427045404911,
-      "learning_rate": 0.00019998795371266708,
-      "loss": 46.0,
-      "step": 30691
-    },
-    {
-      "epoch": 4.942711059221386,
-      "grad_norm": 0.0028707613237202168,
-      "learning_rate": 0.00019998795292740964,
-      "loss": 46.0,
-      "step": 30692
-    },
-    {
-      "epoch": 4.942872096300173,
-      "grad_norm": 0.008118171244859695,
-      "learning_rate": 0.0001999879521421266,
-      "loss": 46.0,
-      "step": 30693
-    },
-    {
-      "epoch": 4.943033133378961,
-      "grad_norm": 0.0020772370044142008,
-      "learning_rate": 0.00019998795135681797,
-      "loss": 46.0,
-      "step": 30694
-    },
-    {
-      "epoch": 4.943194170457748,
-      "grad_norm": 0.006738371681421995,
-      "learning_rate": 0.00019998795057148377,
-      "loss": 46.0,
-      "step": 30695
-    },
-    {
-      "epoch": 4.943355207536535,
-      "grad_norm": 0.003445185488089919,
-      "learning_rate": 0.00019998794978612393,
-      "loss": 46.0,
-      "step": 30696
-    },
-    {
-      "epoch": 4.9435162446153225,
-      "grad_norm": 0.005194577854126692,
-      "learning_rate": 0.00019998794900073853,
-      "loss": 46.0,
-      "step": 30697
-    },
-    {
-      "epoch": 4.94367728169411,
-      "grad_norm": 0.0015515319537371397,
-      "learning_rate": 0.00019998794821532757,
-      "loss": 46.0,
-      "step": 30698
-    },
-    {
-      "epoch": 4.943838318772897,
-      "grad_norm": 0.012259879149496555,
-      "learning_rate": 0.00019998794742989096,
-      "loss": 46.0,
-      "step": 30699
-    },
-    {
-      "epoch": 4.943999355851685,
-      "grad_norm": 0.0016848270315676928,
-      "learning_rate": 0.0001999879466444288,
-      "loss": 46.0,
-      "step": 30700
-    },
-    {
-      "epoch": 4.944160392930472,
-      "grad_norm": 0.0017575763631612062,
-      "learning_rate": 0.00019998794585894105,
-      "loss": 46.0,
-      "step": 30701
-    },
-    {
-      "epoch": 4.94432143000926,
-      "grad_norm": 0.006004973314702511,
-      "learning_rate": 0.0001999879450734277,
-      "loss": 46.0,
-      "step": 30702
-    },
-    {
-      "epoch": 4.944482467088047,
-      "grad_norm": 0.0045472909696400166,
-      "learning_rate": 0.00019998794428788874,
-      "loss": 46.0,
-      "step": 30703
-    },
-    {
-      "epoch": 4.944643504166835,
-      "grad_norm": 0.003451672615483403,
-      "learning_rate": 0.0001999879435023242,
-      "loss": 46.0,
-      "step": 30704
-    },
-    {
-      "epoch": 4.944804541245622,
-      "grad_norm": 0.0039057403337210417,
-      "learning_rate": 0.00019998794271673407,
-      "loss": 46.0,
-      "step": 30705
-    },
-    {
-      "epoch": 4.9449655783244095,
-      "grad_norm": 0.013285440392792225,
-      "learning_rate": 0.0001999879419311183,
-      "loss": 46.0,
-      "step": 30706
-    },
-    {
-      "epoch": 4.945126615403197,
-      "grad_norm": 0.0026773347053676844,
-      "learning_rate": 0.000199987941145477,
-      "loss": 46.0,
-      "step": 30707
-    },
-    {
-      "epoch": 4.9452876524819835,
-      "grad_norm": 0.004525552503764629,
-      "learning_rate": 0.0001999879403598101,
-      "loss": 46.0,
-      "step": 30708
-    },
-    {
-      "epoch": 4.945448689560771,
-      "grad_norm": 0.004753525368869305,
-      "learning_rate": 0.0001999879395741176,
-      "loss": 46.0,
-      "step": 30709
-    },
-    {
-      "epoch": 4.9456097266395584,
-      "grad_norm": 0.0020778074394911528,
-      "learning_rate": 0.00019998793878839952,
-      "loss": 46.0,
-      "step": 30710
-    },
-    {
-      "epoch": 4.945770763718346,
-      "grad_norm": 0.005345955025404692,
-      "learning_rate": 0.00019998793800265582,
-      "loss": 46.0,
-      "step": 30711
-    },
-    {
-      "epoch": 4.945931800797133,
-      "grad_norm": 0.007204418536275625,
-      "learning_rate": 0.00019998793721688656,
-      "loss": 46.0,
-      "step": 30712
-    },
-    {
-      "epoch": 4.946092837875921,
-      "grad_norm": 0.01259730663150549,
-      "learning_rate": 0.00019998793643109168,
-      "loss": 46.0,
-      "step": 30713
-    },
-    {
-      "epoch": 4.946253874954708,
-      "grad_norm": 0.003062712959945202,
-      "learning_rate": 0.00019998793564527122,
-      "loss": 46.0,
-      "step": 30714
-    },
-    {
-      "epoch": 4.946414912033496,
-      "grad_norm": 0.006511330604553223,
-      "learning_rate": 0.00019998793485942516,
-      "loss": 46.0,
-      "step": 30715
-    },
-    {
-      "epoch": 4.946575949112283,
-      "grad_norm": 0.002218864159658551,
-      "learning_rate": 0.00019998793407355355,
-      "loss": 46.0,
-      "step": 30716
-    },
-    {
-      "epoch": 4.946736986191071,
-      "grad_norm": 0.0012778856325894594,
-      "learning_rate": 0.00019998793328765627,
-      "loss": 46.0,
-      "step": 30717
-    },
-    {
-      "epoch": 4.946898023269858,
-      "grad_norm": 0.0029207635670900345,
-      "learning_rate": 0.00019998793250173346,
-      "loss": 46.0,
-      "step": 30718
-    },
-    {
-      "epoch": 4.9470590603486455,
-      "grad_norm": 0.006804576143622398,
-      "learning_rate": 0.00019998793171578506,
-      "loss": 46.0,
-      "step": 30719
-    },
-    {
-      "epoch": 4.947220097427433,
-      "grad_norm": 0.004211138002574444,
-      "learning_rate": 0.00019998793092981102,
-      "loss": 46.0,
-      "step": 30720
-    },
-    {
-      "epoch": 4.94738113450622,
-      "grad_norm": 0.007896900177001953,
-      "learning_rate": 0.00019998793014381144,
-      "loss": 46.0,
-      "step": 30721
-    },
-    {
-      "epoch": 4.947542171585008,
-      "grad_norm": 0.004960453137755394,
-      "learning_rate": 0.00019998792935778625,
-      "loss": 46.0,
-      "step": 30722
-    },
-    {
-      "epoch": 4.947703208663794,
-      "grad_norm": 0.012006685137748718,
-      "learning_rate": 0.00019998792857173545,
-      "loss": 46.0,
-      "step": 30723
-    },
-    {
-      "epoch": 4.947864245742582,
-      "grad_norm": 0.005531883332878351,
-      "learning_rate": 0.00019998792778565906,
-      "loss": 46.0,
-      "step": 30724
-    },
-    {
-      "epoch": 4.948025282821369,
-      "grad_norm": 0.007913201116025448,
-      "learning_rate": 0.0001999879269995571,
-      "loss": 46.0,
-      "step": 30725
-    },
-    {
-      "epoch": 4.948186319900157,
-      "grad_norm": 0.0027838731184601784,
-      "learning_rate": 0.00019998792621342954,
-      "loss": 46.0,
-      "step": 30726
-    },
-    {
-      "epoch": 4.948347356978944,
-      "grad_norm": 0.00880567915737629,
-      "learning_rate": 0.0001999879254272764,
-      "loss": 46.0,
-      "step": 30727
-    },
-    {
-      "epoch": 4.948508394057732,
-      "grad_norm": 0.010923749767243862,
-      "learning_rate": 0.00019998792464109765,
-      "loss": 46.0,
-      "step": 30728
-    },
-    {
-      "epoch": 4.948669431136519,
-      "grad_norm": 0.00965795386582613,
-      "learning_rate": 0.00019998792385489332,
-      "loss": 46.0,
-      "step": 30729
-    },
-    {
-      "epoch": 4.948830468215307,
-      "grad_norm": 0.005377555266022682,
-      "learning_rate": 0.00019998792306866338,
-      "loss": 46.0,
-      "step": 30730
-    },
-    {
-      "epoch": 4.948991505294094,
-      "grad_norm": 0.002625696361064911,
-      "learning_rate": 0.00019998792228240786,
-      "loss": 46.0,
-      "step": 30731
-    },
-    {
-      "epoch": 4.9491525423728815,
-      "grad_norm": 0.0037672684993594885,
-      "learning_rate": 0.00019998792149612674,
-      "loss": 46.0,
-      "step": 30732
-    },
-    {
-      "epoch": 4.949313579451669,
-      "grad_norm": 0.01357579231262207,
-      "learning_rate": 0.00019998792070982004,
-      "loss": 46.0,
-      "step": 30733
-    },
-    {
-      "epoch": 4.949474616530456,
-      "grad_norm": 0.0016936531756073236,
-      "learning_rate": 0.00019998791992348775,
-      "loss": 46.0,
-      "step": 30734
-    },
-    {
-      "epoch": 4.949635653609244,
-      "grad_norm": 0.00652466993778944,
-      "learning_rate": 0.00019998791913712987,
-      "loss": 46.0,
-      "step": 30735
-    },
-    {
-      "epoch": 4.949796690688031,
-      "grad_norm": 0.0016508083790540695,
-      "learning_rate": 0.00019998791835074638,
-      "loss": 46.0,
-      "step": 30736
-    },
-    {
-      "epoch": 4.949957727766819,
-      "grad_norm": 0.014912517741322517,
-      "learning_rate": 0.00019998791756433733,
-      "loss": 46.0,
-      "step": 30737
-    },
-    {
-      "epoch": 4.950118764845605,
-      "grad_norm": 0.0018059872090816498,
-      "learning_rate": 0.00019998791677790264,
-      "loss": 46.0,
-      "step": 30738
-    },
-    {
-      "epoch": 4.950279801924393,
-      "grad_norm": 0.0017447017598897219,
-      "learning_rate": 0.00019998791599144239,
-      "loss": 46.0,
-      "step": 30739
-    },
-    {
-      "epoch": 4.95044083900318,
-      "grad_norm": 0.0019586447160691023,
-      "learning_rate": 0.00019998791520495655,
-      "loss": 46.0,
-      "step": 30740
-    },
-    {
-      "epoch": 4.950601876081968,
-      "grad_norm": 0.002781867515295744,
-      "learning_rate": 0.00019998791441844512,
-      "loss": 46.0,
-      "step": 30741
-    },
-    {
-      "epoch": 4.950762913160755,
-      "grad_norm": 0.004331487230956554,
-      "learning_rate": 0.0001999879136319081,
-      "loss": 46.0,
-      "step": 30742
-    },
-    {
-      "epoch": 4.950923950239543,
-      "grad_norm": 0.006971669849008322,
-      "learning_rate": 0.00019998791284534545,
-      "loss": 46.0,
-      "step": 30743
-    },
-    {
-      "epoch": 4.95108498731833,
-      "grad_norm": 0.007546829991042614,
-      "learning_rate": 0.00019998791205875724,
-      "loss": 46.0,
-      "step": 30744
-    },
-    {
-      "epoch": 4.9512460243971175,
-      "grad_norm": 0.00620879465714097,
-      "learning_rate": 0.00019998791127214346,
-      "loss": 46.0,
-      "step": 30745
-    },
-    {
-      "epoch": 4.951407061475905,
-      "grad_norm": 0.0024855416268110275,
-      "learning_rate": 0.00019998791048550404,
-      "loss": 46.0,
-      "step": 30746
-    },
-    {
-      "epoch": 4.951568098554692,
-      "grad_norm": 0.0022948093246668577,
-      "learning_rate": 0.00019998790969883907,
-      "loss": 46.0,
-      "step": 30747
-    },
-    {
-      "epoch": 4.95172913563348,
-      "grad_norm": 0.02409781888127327,
-      "learning_rate": 0.0001999879089121485,
-      "loss": 46.0,
-      "step": 30748
-    },
-    {
-      "epoch": 4.951890172712267,
-      "grad_norm": 0.0031440521124750376,
-      "learning_rate": 0.00019998790812543232,
-      "loss": 46.0,
-      "step": 30749
-    },
-    {
-      "epoch": 4.952051209791055,
-      "grad_norm": 0.004547193646430969,
-      "learning_rate": 0.00019998790733869056,
-      "loss": 46.0,
-      "step": 30750
-    },
-    {
-      "epoch": 4.952212246869841,
-      "grad_norm": 0.011125778779387474,
-      "learning_rate": 0.00019998790655192318,
-      "loss": 46.0,
-      "step": 30751
-    },
-    {
-      "epoch": 4.95237328394863,
-      "grad_norm": 0.002236815867945552,
-      "learning_rate": 0.00019998790576513024,
-      "loss": 46.0,
-      "step": 30752
-    },
-    {
-      "epoch": 4.952534321027416,
-      "grad_norm": 0.0036717483308166265,
-      "learning_rate": 0.0001999879049783117,
-      "loss": 46.0,
-      "step": 30753
-    },
-    {
-      "epoch": 4.952695358106204,
-      "grad_norm": 0.00991331972181797,
-      "learning_rate": 0.00019998790419146757,
-      "loss": 46.0,
-      "step": 30754
-    },
-    {
-      "epoch": 4.952856395184991,
-      "grad_norm": 0.0018810078036040068,
-      "learning_rate": 0.00019998790340459784,
-      "loss": 46.0,
-      "step": 30755
-    },
-    {
-      "epoch": 4.9530174322637786,
-      "grad_norm": 0.002276583109050989,
-      "learning_rate": 0.00019998790261770253,
-      "loss": 46.0,
-      "step": 30756
-    },
-    {
-      "epoch": 4.953178469342566,
-      "grad_norm": 0.005891216918826103,
-      "learning_rate": 0.00019998790183078162,
-      "loss": 46.0,
-      "step": 30757
-    },
-    {
-      "epoch": 4.9533395064213535,
-      "grad_norm": 0.001798367011360824,
-      "learning_rate": 0.0001999879010438351,
-      "loss": 46.0,
-      "step": 30758
-    },
-    {
-      "epoch": 4.953500543500141,
-      "grad_norm": 0.0049967155791819096,
-      "learning_rate": 0.00019998790025686303,
-      "loss": 46.0,
-      "step": 30759
-    },
-    {
-      "epoch": 4.953661580578928,
-      "grad_norm": 0.00212867627851665,
-      "learning_rate": 0.00019998789946986536,
-      "loss": 46.0,
-      "step": 30760
-    },
-    {
-      "epoch": 4.953822617657716,
-      "grad_norm": 0.005703443195670843,
-      "learning_rate": 0.00019998789868284206,
-      "loss": 46.0,
-      "step": 30761
-    },
-    {
-      "epoch": 4.953983654736503,
-      "grad_norm": 0.0011965708108618855,
-      "learning_rate": 0.00019998789789579322,
-      "loss": 46.0,
-      "step": 30762
-    },
-    {
-      "epoch": 4.954144691815291,
-      "grad_norm": 0.0067934091202914715,
-      "learning_rate": 0.00019998789710871874,
-      "loss": 46.0,
-      "step": 30763
-    },
-    {
-      "epoch": 4.954305728894078,
-      "grad_norm": 0.015066059306263924,
-      "learning_rate": 0.00019998789632161873,
-      "loss": 46.0,
-      "step": 30764
-    },
-    {
-      "epoch": 4.954466765972866,
-      "grad_norm": 0.005104829557240009,
-      "learning_rate": 0.00019998789553449307,
-      "loss": 46.0,
-      "step": 30765
-    },
-    {
-      "epoch": 4.954627803051652,
-      "grad_norm": 0.004770467057824135,
-      "learning_rate": 0.00019998789474734183,
-      "loss": 46.0,
-      "step": 30766
-    },
-    {
-      "epoch": 4.9547888401304405,
-      "grad_norm": 0.0018297730712220073,
-      "learning_rate": 0.00019998789396016503,
-      "loss": 46.0,
-      "step": 30767
-    },
-    {
-      "epoch": 4.954949877209227,
-      "grad_norm": 0.0027819073293358088,
-      "learning_rate": 0.00019998789317296261,
-      "loss": 46.0,
-      "step": 30768
-    },
-    {
-      "epoch": 4.9551109142880145,
-      "grad_norm": 0.013442805968225002,
-      "learning_rate": 0.00019998789238573458,
-      "loss": 46.0,
-      "step": 30769
-    },
-    {
-      "epoch": 4.955271951366802,
-      "grad_norm": 0.018889401108026505,
-      "learning_rate": 0.000199987891598481,
-      "loss": 46.0,
-      "step": 30770
-    },
-    {
-      "epoch": 4.955432988445589,
-      "grad_norm": 0.010513770394027233,
-      "learning_rate": 0.0001999878908112018,
-      "loss": 46.0,
-      "step": 30771
-    },
-    {
-      "epoch": 4.955594025524377,
-      "grad_norm": 0.0029280823655426502,
-      "learning_rate": 0.00019998789002389702,
-      "loss": 46.0,
-      "step": 30772
-    },
-    {
-      "epoch": 4.955755062603164,
-      "grad_norm": 0.005389626137912273,
-      "learning_rate": 0.00019998788923656665,
-      "loss": 46.0,
-      "step": 30773
-    },
-    {
-      "epoch": 4.955916099681952,
-      "grad_norm": 0.00129792932420969,
-      "learning_rate": 0.00019998788844921068,
-      "loss": 46.0,
-      "step": 30774
-    },
-    {
-      "epoch": 4.956077136760739,
-      "grad_norm": 0.00236967159435153,
-      "learning_rate": 0.00019998788766182913,
-      "loss": 46.0,
-      "step": 30775
-    },
-    {
-      "epoch": 4.956238173839527,
-      "grad_norm": 0.0024803578853607178,
-      "learning_rate": 0.00019998788687442196,
-      "loss": 46.0,
-      "step": 30776
-    },
-    {
-      "epoch": 4.956399210918314,
-      "grad_norm": 0.002935096388682723,
-      "learning_rate": 0.0001999878860869892,
-      "loss": 46.0,
-      "step": 30777
-    },
-    {
-      "epoch": 4.956560247997102,
-      "grad_norm": 0.005948344245553017,
-      "learning_rate": 0.0001999878852995309,
-      "loss": 46.0,
-      "step": 30778
-    },
-    {
-      "epoch": 4.956721285075889,
-      "grad_norm": 0.012712572701275349,
-      "learning_rate": 0.000199987884512047,
-      "loss": 46.0,
-      "step": 30779
-    },
-    {
-      "epoch": 4.9568823221546765,
-      "grad_norm": 0.00227820104919374,
-      "learning_rate": 0.00019998788372453744,
-      "loss": 46.0,
-      "step": 30780
-    },
-    {
-      "epoch": 4.957043359233463,
-      "grad_norm": 0.004744593519717455,
-      "learning_rate": 0.00019998788293700234,
-      "loss": 46.0,
-      "step": 30781
-    },
-    {
-      "epoch": 4.9572043963122505,
-      "grad_norm": 0.003750715870410204,
-      "learning_rate": 0.00019998788214944165,
-      "loss": 46.0,
-      "step": 30782
-    },
-    {
-      "epoch": 4.957365433391038,
-      "grad_norm": 0.0069166370667517185,
-      "learning_rate": 0.00019998788136185535,
-      "loss": 46.0,
-      "step": 30783
-    },
-    {
-      "epoch": 4.957526470469825,
-      "grad_norm": 0.002376657212153077,
-      "learning_rate": 0.00019998788057424348,
-      "loss": 46.0,
-      "step": 30784
-    },
-    {
-      "epoch": 4.957687507548613,
-      "grad_norm": 0.006184377241879702,
-      "learning_rate": 0.00019998787978660598,
-      "loss": 46.0,
-      "step": 30785
-    },
-    {
-      "epoch": 4.9578485446274,
-      "grad_norm": 0.0018443232402205467,
-      "learning_rate": 0.00019998787899894294,
-      "loss": 46.0,
-      "step": 30786
-    },
-    {
-      "epoch": 4.958009581706188,
-      "grad_norm": 0.02072012796998024,
-      "learning_rate": 0.00019998787821125426,
-      "loss": 46.0,
-      "step": 30787
-    },
-    {
-      "epoch": 4.958170618784975,
-      "grad_norm": 0.002755719004198909,
-      "learning_rate": 0.00019998787742354001,
-      "loss": 46.0,
-      "step": 30788
-    },
-    {
-      "epoch": 4.958331655863763,
-      "grad_norm": 0.004709905479103327,
-      "learning_rate": 0.0001999878766358002,
-      "loss": 46.0,
-      "step": 30789
-    },
-    {
-      "epoch": 4.95849269294255,
-      "grad_norm": 0.004154428839683533,
-      "learning_rate": 0.00019998787584803474,
-      "loss": 46.0,
-      "step": 30790
-    },
-    {
-      "epoch": 4.958653730021338,
-      "grad_norm": 0.001521439990028739,
-      "learning_rate": 0.0001999878750602437,
-      "loss": 46.0,
-      "step": 30791
-    },
-    {
-      "epoch": 4.958814767100125,
-      "grad_norm": 0.005281943362206221,
-      "learning_rate": 0.0001999878742724271,
-      "loss": 46.0,
-      "step": 30792
-    },
-    {
-      "epoch": 4.9589758041789125,
-      "grad_norm": 0.00515834242105484,
-      "learning_rate": 0.0001999878734845849,
-      "loss": 46.0,
-      "step": 30793
-    },
-    {
-      "epoch": 4.9591368412577,
-      "grad_norm": 0.0026498320512473583,
-      "learning_rate": 0.00019998787269671708,
-      "loss": 46.0,
-      "step": 30794
-    },
-    {
-      "epoch": 4.959297878336487,
-      "grad_norm": 0.0018939678557217121,
-      "learning_rate": 0.0001999878719088237,
-      "loss": 46.0,
-      "step": 30795
-    },
-    {
-      "epoch": 4.959458915415274,
-      "grad_norm": 0.006592052523046732,
-      "learning_rate": 0.0001999878711209047,
-      "loss": 46.0,
-      "step": 30796
-    },
-    {
-      "epoch": 4.959619952494061,
-      "grad_norm": 0.0020348874386399984,
-      "learning_rate": 0.0001999878703329601,
-      "loss": 46.0,
-      "step": 30797
-    },
-    {
-      "epoch": 4.959780989572849,
-      "grad_norm": 0.008914710022509098,
-      "learning_rate": 0.00019998786954498996,
-      "loss": 46.0,
-      "step": 30798
-    },
-    {
-      "epoch": 4.959942026651636,
-      "grad_norm": 0.0071667516604065895,
-      "learning_rate": 0.00019998786875699418,
-      "loss": 46.0,
-      "step": 30799
-    },
-    {
-      "epoch": 4.960103063730424,
-      "grad_norm": 0.006221315823495388,
-      "learning_rate": 0.00019998786796897284,
-      "loss": 46.0,
-      "step": 30800
-    },
-    {
-      "epoch": 4.960264100809211,
-      "grad_norm": 0.004883614834398031,
-      "learning_rate": 0.0001999878671809259,
-      "loss": 46.0,
-      "step": 30801
-    },
-    {
-      "epoch": 4.960425137887999,
-      "grad_norm": 0.008304298855364323,
-      "learning_rate": 0.00019998786639285337,
-      "loss": 46.0,
-      "step": 30802
-    },
-    {
-      "epoch": 4.960586174966786,
-      "grad_norm": 0.009872698225080967,
-      "learning_rate": 0.0001999878656047552,
-      "loss": 46.0,
-      "step": 30803
-    },
-    {
-      "epoch": 4.960747212045574,
-      "grad_norm": 0.0013597247889265418,
-      "learning_rate": 0.00019998786481663152,
-      "loss": 46.0,
-      "step": 30804
-    },
-    {
-      "epoch": 4.960908249124361,
-      "grad_norm": 0.0041055562905967236,
-      "learning_rate": 0.0001999878640284822,
-      "loss": 46.0,
-      "step": 30805
-    },
-    {
-      "epoch": 4.9610692862031485,
-      "grad_norm": 0.000781733135227114,
-      "learning_rate": 0.0001999878632403073,
-      "loss": 46.0,
-      "step": 30806
-    },
-    {
-      "epoch": 4.961230323281936,
-      "grad_norm": 0.010199559852480888,
-      "learning_rate": 0.00019998786245210682,
-      "loss": 46.0,
-      "step": 30807
-    },
-    {
-      "epoch": 4.961391360360723,
-      "grad_norm": 0.013601471669971943,
-      "learning_rate": 0.0001999878616638807,
-      "loss": 46.0,
-      "step": 30808
-    },
-    {
-      "epoch": 4.961552397439511,
-      "grad_norm": 0.005652326624840498,
-      "learning_rate": 0.00019998786087562905,
-      "loss": 46.0,
-      "step": 30809
-    },
-    {
-      "epoch": 4.961713434518298,
-      "grad_norm": 0.0021816270891577005,
-      "learning_rate": 0.00019998786008735178,
-      "loss": 46.0,
-      "step": 30810
-    },
-    {
-      "epoch": 4.961874471597085,
-      "grad_norm": 0.0051984842866659164,
-      "learning_rate": 0.0001999878592990489,
-      "loss": 46.0,
-      "step": 30811
-    },
-    {
-      "epoch": 4.962035508675872,
-      "grad_norm": 0.0021691059228032827,
-      "learning_rate": 0.00019998785851072046,
-      "loss": 46.0,
-      "step": 30812
-    },
-    {
-      "epoch": 4.96219654575466,
-      "grad_norm": 0.0024269712157547474,
-      "learning_rate": 0.0001999878577223664,
-      "loss": 46.0,
-      "step": 30813
-    },
-    {
-      "epoch": 4.962357582833447,
-      "grad_norm": 0.006437079980969429,
-      "learning_rate": 0.00019998785693398676,
-      "loss": 46.0,
-      "step": 30814
-    },
-    {
-      "epoch": 4.962518619912235,
-      "grad_norm": 0.0037849147338420153,
-      "learning_rate": 0.00019998785614558156,
-      "loss": 46.0,
-      "step": 30815
-    },
-    {
-      "epoch": 4.962679656991022,
-      "grad_norm": 0.006798509042710066,
-      "learning_rate": 0.00019998785535715072,
-      "loss": 46.0,
-      "step": 30816
-    },
-    {
-      "epoch": 4.9628406940698095,
-      "grad_norm": 0.004250046797096729,
-      "learning_rate": 0.0001999878545686943,
-      "loss": 46.0,
-      "step": 30817
-    },
-    {
-      "epoch": 4.963001731148597,
-      "grad_norm": 0.0032048027496784925,
-      "learning_rate": 0.00019998785378021232,
-      "loss": 46.0,
-      "step": 30818
-    },
-    {
-      "epoch": 4.9631627682273844,
-      "grad_norm": 0.008781819604337215,
-      "learning_rate": 0.00019998785299170472,
-      "loss": 46.0,
-      "step": 30819
-    },
-    {
-      "epoch": 4.963323805306172,
-      "grad_norm": 0.004050773102790117,
-      "learning_rate": 0.00019998785220317152,
-      "loss": 46.0,
-      "step": 30820
-    },
-    {
-      "epoch": 4.963484842384959,
-      "grad_norm": 0.00239588413387537,
-      "learning_rate": 0.00019998785141461274,
-      "loss": 46.0,
-      "step": 30821
-    },
-    {
-      "epoch": 4.963645879463747,
-      "grad_norm": 0.008822372183203697,
-      "learning_rate": 0.00019998785062602838,
-      "loss": 46.0,
-      "step": 30822
-    },
-    {
-      "epoch": 4.963806916542534,
-      "grad_norm": 0.003714524209499359,
-      "learning_rate": 0.0001999878498374184,
-      "loss": 46.0,
-      "step": 30823
-    },
-    {
-      "epoch": 4.963967953621321,
-      "grad_norm": 0.02254457026720047,
-      "learning_rate": 0.00019998784904878285,
-      "loss": 46.0,
-      "step": 30824
-    },
-    {
-      "epoch": 4.964128990700109,
-      "grad_norm": 0.001230532769113779,
-      "learning_rate": 0.00019998784826012173,
-      "loss": 46.0,
-      "step": 30825
-    },
-    {
-      "epoch": 4.964290027778896,
-      "grad_norm": 0.006136293988674879,
-      "learning_rate": 0.00019998784747143498,
-      "loss": 46.0,
-      "step": 30826
-    },
-    {
-      "epoch": 4.964451064857683,
-      "grad_norm": 0.005339915864169598,
-      "learning_rate": 0.00019998784668272265,
-      "loss": 46.0,
-      "step": 30827
-    },
-    {
-      "epoch": 4.964612101936471,
-      "grad_norm": 0.013644712045788765,
-      "learning_rate": 0.00019998784589398474,
-      "loss": 46.0,
-      "step": 30828
-    },
-    {
-      "epoch": 4.964773139015258,
-      "grad_norm": 0.008860011585056782,
-      "learning_rate": 0.0001999878451052212,
-      "loss": 46.0,
-      "step": 30829
-    },
-    {
-      "epoch": 4.9649341760940455,
-      "grad_norm": 0.002051092917099595,
-      "learning_rate": 0.0001999878443164321,
-      "loss": 46.0,
-      "step": 30830
-    },
-    {
-      "epoch": 4.965095213172833,
-      "grad_norm": 0.013038805685937405,
-      "learning_rate": 0.0001999878435276174,
-      "loss": 46.0,
-      "step": 30831
-    },
-    {
-      "epoch": 4.96525625025162,
-      "grad_norm": 0.004594277124851942,
-      "learning_rate": 0.00019998784273877714,
-      "loss": 46.0,
-      "step": 30832
-    },
-    {
-      "epoch": 4.965417287330408,
-      "grad_norm": 0.00135852734092623,
-      "learning_rate": 0.00019998784194991123,
-      "loss": 46.0,
-      "step": 30833
-    },
-    {
-      "epoch": 4.965578324409195,
-      "grad_norm": 0.003066158387809992,
-      "learning_rate": 0.00019998784116101977,
-      "loss": 46.0,
-      "step": 30834
-    },
-    {
-      "epoch": 4.965739361487983,
-      "grad_norm": 0.010096262209117413,
-      "learning_rate": 0.0001999878403721027,
-      "loss": 46.0,
-      "step": 30835
-    },
-    {
-      "epoch": 4.96590039856677,
-      "grad_norm": 0.0010591770987957716,
-      "learning_rate": 0.00019998783958316004,
-      "loss": 46.0,
-      "step": 30836
-    },
-    {
-      "epoch": 4.966061435645558,
-      "grad_norm": 0.008141896687448025,
-      "learning_rate": 0.00019998783879419181,
-      "loss": 46.0,
-      "step": 30837
-    },
-    {
-      "epoch": 4.966222472724345,
-      "grad_norm": 0.002683746861293912,
-      "learning_rate": 0.00019998783800519797,
-      "loss": 46.0,
-      "step": 30838
-    },
-    {
-      "epoch": 4.966383509803132,
-      "grad_norm": 0.005121730268001556,
-      "learning_rate": 0.00019998783721617854,
-      "loss": 46.0,
-      "step": 30839
-    },
-    {
-      "epoch": 4.96654454688192,
-      "grad_norm": 0.0031825529877096415,
-      "learning_rate": 0.00019998783642713352,
-      "loss": 46.0,
-      "step": 30840
-    },
-    {
-      "epoch": 4.966705583960707,
-      "grad_norm": 0.002786989789456129,
-      "learning_rate": 0.0001999878356380629,
-      "loss": 46.0,
-      "step": 30841
-    },
-    {
-      "epoch": 4.966866621039494,
-      "grad_norm": 0.010649020783603191,
-      "learning_rate": 0.00019998783484896667,
-      "loss": 46.0,
-      "step": 30842
-    },
-    {
-      "epoch": 4.9670276581182815,
-      "grad_norm": 0.004177742172032595,
-      "learning_rate": 0.0001999878340598449,
-      "loss": 46.0,
-      "step": 30843
-    },
-    {
-      "epoch": 4.967188695197069,
-      "grad_norm": 0.007972175255417824,
-      "learning_rate": 0.0001999878332706975,
-      "loss": 46.0,
-      "step": 30844
-    },
-    {
-      "epoch": 4.967349732275856,
-      "grad_norm": 0.002459653653204441,
-      "learning_rate": 0.00019998783248152452,
-      "loss": 46.0,
-      "step": 30845
-    },
-    {
-      "epoch": 4.967510769354644,
-      "grad_norm": 0.0047114272601902485,
-      "learning_rate": 0.00019998783169232595,
-      "loss": 46.0,
-      "step": 30846
-    },
-    {
-      "epoch": 4.967671806433431,
-      "grad_norm": 0.002555410610511899,
-      "learning_rate": 0.00019998783090310177,
-      "loss": 46.0,
-      "step": 30847
-    },
-    {
-      "epoch": 4.967832843512219,
-      "grad_norm": 0.0028972234576940536,
-      "learning_rate": 0.00019998783011385203,
-      "loss": 46.0,
-      "step": 30848
-    },
-    {
-      "epoch": 4.967993880591006,
-      "grad_norm": 0.01440118532627821,
-      "learning_rate": 0.00019998782932457667,
-      "loss": 46.0,
-      "step": 30849
-    },
-    {
-      "epoch": 4.968154917669794,
-      "grad_norm": 0.0037989113479852676,
-      "learning_rate": 0.00019998782853527573,
-      "loss": 46.0,
-      "step": 30850
-    },
-    {
-      "epoch": 4.968315954748581,
-      "grad_norm": 0.0014283166965469718,
-      "learning_rate": 0.0001999878277459492,
-      "loss": 46.0,
-      "step": 30851
-    },
-    {
-      "epoch": 4.968476991827369,
-      "grad_norm": 0.011698959395289421,
-      "learning_rate": 0.00019998782695659708,
-      "loss": 46.0,
-      "step": 30852
-    },
-    {
-      "epoch": 4.968638028906156,
-      "grad_norm": 0.0020634287502616644,
-      "learning_rate": 0.00019998782616721935,
-      "loss": 46.0,
-      "step": 30853
-    },
-    {
-      "epoch": 4.968799065984943,
-      "grad_norm": 0.015871427953243256,
-      "learning_rate": 0.00019998782537781606,
-      "loss": 46.0,
-      "step": 30854
-    },
-    {
-      "epoch": 4.96896010306373,
-      "grad_norm": 0.008170481771230698,
-      "learning_rate": 0.00019998782458838715,
-      "loss": 46.0,
-      "step": 30855
-    },
-    {
-      "epoch": 4.9691211401425175,
-      "grad_norm": 0.005524741485714912,
-      "learning_rate": 0.00019998782379893266,
-      "loss": 46.0,
-      "step": 30856
-    },
-    {
-      "epoch": 4.969282177221305,
-      "grad_norm": 0.001704831374809146,
-      "learning_rate": 0.00019998782300945258,
-      "loss": 46.0,
-      "step": 30857
-    },
-    {
-      "epoch": 4.969443214300092,
-      "grad_norm": 0.0036982260644435883,
-      "learning_rate": 0.0001999878222199469,
-      "loss": 46.0,
-      "step": 30858
-    },
-    {
-      "epoch": 4.96960425137888,
-      "grad_norm": 0.0019745659083127975,
-      "learning_rate": 0.00019998782143041565,
-      "loss": 46.0,
-      "step": 30859
-    },
-    {
-      "epoch": 4.969765288457667,
-      "grad_norm": 0.002938745077699423,
-      "learning_rate": 0.00019998782064085878,
-      "loss": 46.0,
-      "step": 30860
-    },
-    {
-      "epoch": 4.969926325536455,
-      "grad_norm": 0.008062013424932957,
-      "learning_rate": 0.00019998781985127635,
-      "loss": 46.0,
-      "step": 30861
-    },
-    {
-      "epoch": 4.970087362615242,
-      "grad_norm": 0.006624516565352678,
-      "learning_rate": 0.00019998781906166828,
-      "loss": 46.0,
-      "step": 30862
-    },
-    {
-      "epoch": 4.97024839969403,
-      "grad_norm": 0.0031087002716958523,
-      "learning_rate": 0.00019998781827203468,
-      "loss": 46.0,
-      "step": 30863
-    },
-    {
-      "epoch": 4.970409436772817,
-      "grad_norm": 0.005210024770349264,
-      "learning_rate": 0.00019998781748237546,
-      "loss": 46.0,
-      "step": 30864
-    },
-    {
-      "epoch": 4.9705704738516046,
-      "grad_norm": 0.0054557411931455135,
-      "learning_rate": 0.00019998781669269063,
-      "loss": 46.0,
-      "step": 30865
-    },
-    {
-      "epoch": 4.970731510930392,
-      "grad_norm": 0.005232595838606358,
-      "learning_rate": 0.0001999878159029802,
-      "loss": 46.0,
-      "step": 30866
-    },
-    {
-      "epoch": 4.9708925480091795,
-      "grad_norm": 0.00858256034553051,
-      "learning_rate": 0.0001999878151132442,
-      "loss": 46.0,
-      "step": 30867
-    },
-    {
-      "epoch": 4.971053585087967,
-      "grad_norm": 0.01136784814298153,
-      "learning_rate": 0.0001999878143234826,
-      "loss": 46.0,
-      "step": 30868
-    },
-    {
-      "epoch": 4.9712146221667535,
-      "grad_norm": 0.009713488630950451,
-      "learning_rate": 0.00019998781353369546,
-      "loss": 46.0,
-      "step": 30869
-    },
-    {
-      "epoch": 4.971375659245541,
-      "grad_norm": 0.0037244162522256374,
-      "learning_rate": 0.00019998781274388266,
-      "loss": 46.0,
-      "step": 30870
-    },
-    {
-      "epoch": 4.971536696324328,
-      "grad_norm": 0.010117919184267521,
-      "learning_rate": 0.00019998781195404428,
-      "loss": 46.0,
-      "step": 30871
-    },
-    {
-      "epoch": 4.971697733403116,
-      "grad_norm": 0.0011379426578059793,
-      "learning_rate": 0.00019998781116418034,
-      "loss": 46.0,
-      "step": 30872
-    },
-    {
-      "epoch": 4.971858770481903,
-      "grad_norm": 0.007301435340195894,
-      "learning_rate": 0.0001999878103742908,
-      "loss": 46.0,
-      "step": 30873
-    },
-    {
-      "epoch": 4.972019807560691,
-      "grad_norm": 0.026161599904298782,
-      "learning_rate": 0.00019998780958437563,
-      "loss": 46.0,
-      "step": 30874
-    },
-    {
-      "epoch": 4.972180844639478,
-      "grad_norm": 0.006837194785475731,
-      "learning_rate": 0.0001999878087944349,
-      "loss": 46.0,
-      "step": 30875
-    },
-    {
-      "epoch": 4.972341881718266,
-      "grad_norm": 0.002399842720478773,
-      "learning_rate": 0.00019998780800446858,
-      "loss": 46.0,
-      "step": 30876
-    },
-    {
-      "epoch": 4.972502918797053,
-      "grad_norm": 0.00424237409606576,
-      "learning_rate": 0.00019998780721447665,
-      "loss": 46.0,
-      "step": 30877
-    },
-    {
-      "epoch": 4.9726639558758405,
-      "grad_norm": 0.004987417720258236,
-      "learning_rate": 0.00019998780642445916,
-      "loss": 46.0,
-      "step": 30878
-    },
-    {
-      "epoch": 4.972824992954628,
-      "grad_norm": 0.021086450666189194,
-      "learning_rate": 0.00019998780563441605,
-      "loss": 46.0,
-      "step": 30879
-    },
-    {
-      "epoch": 4.972986030033415,
-      "grad_norm": 0.0005767957773059607,
-      "learning_rate": 0.00019998780484434736,
-      "loss": 46.0,
-      "step": 30880
-    },
-    {
-      "epoch": 4.973147067112203,
-      "grad_norm": 0.005905651021748781,
-      "learning_rate": 0.00019998780405425308,
-      "loss": 46.0,
-      "step": 30881
-    },
-    {
-      "epoch": 4.97330810419099,
-      "grad_norm": 0.006714705843478441,
-      "learning_rate": 0.0001999878032641332,
-      "loss": 46.0,
-      "step": 30882
-    },
-    {
-      "epoch": 4.973469141269778,
-      "grad_norm": 0.006292471196502447,
-      "learning_rate": 0.00019998780247398773,
-      "loss": 46.0,
-      "step": 30883
-    },
-    {
-      "epoch": 4.973630178348564,
-      "grad_norm": 0.005078427493572235,
-      "learning_rate": 0.00019998780168381668,
-      "loss": 46.0,
-      "step": 30884
-    },
-    {
-      "epoch": 4.973791215427352,
-      "grad_norm": 0.0019055759767070413,
-      "learning_rate": 0.00019998780089362003,
-      "loss": 46.0,
-      "step": 30885
-    },
-    {
-      "epoch": 4.973952252506139,
-      "grad_norm": 0.006134196650236845,
-      "learning_rate": 0.00019998780010339778,
-      "loss": 46.0,
-      "step": 30886
-    },
-    {
-      "epoch": 4.974113289584927,
-      "grad_norm": 0.0013564007822424173,
-      "learning_rate": 0.00019998779931314995,
-      "loss": 46.0,
-      "step": 30887
-    },
-    {
-      "epoch": 4.974274326663714,
-      "grad_norm": 0.007288997061550617,
-      "learning_rate": 0.0001999877985228765,
-      "loss": 46.0,
-      "step": 30888
-    },
-    {
-      "epoch": 4.974435363742502,
-      "grad_norm": 0.016209525987505913,
-      "learning_rate": 0.0001999877977325775,
-      "loss": 46.0,
-      "step": 30889
-    },
-    {
-      "epoch": 4.974596400821289,
-      "grad_norm": 0.003899284638464451,
-      "learning_rate": 0.00019998779694225288,
-      "loss": 46.0,
-      "step": 30890
-    },
-    {
-      "epoch": 4.9747574379000765,
-      "grad_norm": 0.010110035538673401,
-      "learning_rate": 0.00019998779615190267,
-      "loss": 46.0,
-      "step": 30891
-    },
-    {
-      "epoch": 4.974918474978864,
-      "grad_norm": 0.009592756628990173,
-      "learning_rate": 0.0001999877953615269,
-      "loss": 46.0,
-      "step": 30892
-    },
-    {
-      "epoch": 4.975079512057651,
-      "grad_norm": 0.00339070288464427,
-      "learning_rate": 0.0001999877945711255,
-      "loss": 46.0,
-      "step": 30893
-    },
-    {
-      "epoch": 4.975240549136439,
-      "grad_norm": 0.005003774538636208,
-      "learning_rate": 0.00019998779378069853,
-      "loss": 46.0,
-      "step": 30894
-    },
-    {
-      "epoch": 4.975401586215226,
-      "grad_norm": 0.005525850225239992,
-      "learning_rate": 0.00019998779299024597,
-      "loss": 46.0,
-      "step": 30895
-    },
-    {
-      "epoch": 4.975562623294014,
-      "grad_norm": 0.00617035198956728,
-      "learning_rate": 0.0001999877921997678,
-      "loss": 46.0,
-      "step": 30896
-    },
-    {
-      "epoch": 4.9757236603728,
-      "grad_norm": 0.00976397842168808,
-      "learning_rate": 0.00019998779140926405,
-      "loss": 46.0,
-      "step": 30897
-    },
-    {
-      "epoch": 4.975884697451589,
-      "grad_norm": 0.0016574239125475287,
-      "learning_rate": 0.00019998779061873473,
-      "loss": 46.0,
-      "step": 30898
-    },
-    {
-      "epoch": 4.976045734530375,
-      "grad_norm": 0.008125197142362595,
-      "learning_rate": 0.00019998778982817977,
-      "loss": 46.0,
-      "step": 30899
-    },
-    {
-      "epoch": 4.976206771609163,
-      "grad_norm": 0.005675680469721556,
-      "learning_rate": 0.00019998778903759922,
-      "loss": 46.0,
-      "step": 30900
-    },
-    {
-      "epoch": 4.97636780868795,
-      "grad_norm": 0.005658462177962065,
-      "learning_rate": 0.00019998778824699312,
-      "loss": 46.0,
-      "step": 30901
-    },
-    {
-      "epoch": 4.976528845766738,
-      "grad_norm": 0.001684830873273313,
-      "learning_rate": 0.00019998778745636142,
-      "loss": 46.0,
-      "step": 30902
-    },
-    {
-      "epoch": 4.976689882845525,
-      "grad_norm": 0.003426732262596488,
-      "learning_rate": 0.0001999877866657041,
-      "loss": 46.0,
-      "step": 30903
-    },
-    {
-      "epoch": 4.9768509199243125,
-      "grad_norm": 0.007448168471455574,
-      "learning_rate": 0.0001999877858750212,
-      "loss": 46.0,
-      "step": 30904
-    },
-    {
-      "epoch": 4.9770119570031,
-      "grad_norm": 0.003538034623488784,
-      "learning_rate": 0.00019998778508431274,
-      "loss": 46.0,
-      "step": 30905
-    },
-    {
-      "epoch": 4.977172994081887,
-      "grad_norm": 0.03049345500767231,
-      "learning_rate": 0.00019998778429357864,
-      "loss": 46.0,
-      "step": 30906
-    },
-    {
-      "epoch": 4.977334031160675,
-      "grad_norm": 0.004986871499568224,
-      "learning_rate": 0.000199987783502819,
-      "loss": 46.0,
-      "step": 30907
-    },
-    {
-      "epoch": 4.977495068239462,
-      "grad_norm": 0.0028877519071102142,
-      "learning_rate": 0.00019998778271203374,
-      "loss": 46.0,
-      "step": 30908
-    },
-    {
-      "epoch": 4.97765610531825,
-      "grad_norm": 0.01014393474906683,
-      "learning_rate": 0.00019998778192122286,
-      "loss": 46.0,
-      "step": 30909
-    },
-    {
-      "epoch": 4.977817142397037,
-      "grad_norm": 0.00386176654137671,
-      "learning_rate": 0.00019998778113038644,
-      "loss": 46.0,
-      "step": 30910
-    },
-    {
-      "epoch": 4.977978179475825,
-      "grad_norm": 0.0066736014559865,
-      "learning_rate": 0.00019998778033952438,
-      "loss": 46.0,
-      "step": 30911
-    },
-    {
-      "epoch": 4.978139216554611,
-      "grad_norm": 0.016126932576298714,
-      "learning_rate": 0.00019998777954863676,
-      "loss": 46.0,
-      "step": 30912
-    },
-    {
-      "epoch": 4.9783002536334,
-      "grad_norm": 0.004830478224903345,
-      "learning_rate": 0.00019998777875772353,
-      "loss": 46.0,
-      "step": 30913
-    },
-    {
-      "epoch": 4.978461290712186,
-      "grad_norm": 0.0030132890678942204,
-      "learning_rate": 0.0001999877779667847,
-      "loss": 46.0,
-      "step": 30914
-    },
-    {
-      "epoch": 4.978622327790974,
-      "grad_norm": 0.017756516113877296,
-      "learning_rate": 0.0001999877771758203,
-      "loss": 46.0,
-      "step": 30915
-    },
-    {
-      "epoch": 4.978783364869761,
-      "grad_norm": 0.012786378152668476,
-      "learning_rate": 0.0001999877763848303,
-      "loss": 46.0,
-      "step": 30916
-    },
-    {
-      "epoch": 4.9789444019485485,
-      "grad_norm": 0.005736165679991245,
-      "learning_rate": 0.00019998777559381475,
-      "loss": 46.0,
-      "step": 30917
-    },
-    {
-      "epoch": 4.979105439027336,
-      "grad_norm": 0.005925099831074476,
-      "learning_rate": 0.00019998777480277356,
-      "loss": 46.0,
-      "step": 30918
-    },
-    {
-      "epoch": 4.979266476106123,
-      "grad_norm": 0.004123123828321695,
-      "learning_rate": 0.00019998777401170677,
-      "loss": 46.0,
-      "step": 30919
-    },
-    {
-      "epoch": 4.979427513184911,
-      "grad_norm": 0.007774327881634235,
-      "learning_rate": 0.00019998777322061443,
-      "loss": 46.0,
-      "step": 30920
-    },
-    {
-      "epoch": 4.979588550263698,
-      "grad_norm": 0.00395645946264267,
-      "learning_rate": 0.00019998777242949647,
-      "loss": 46.0,
-      "step": 30921
-    },
-    {
-      "epoch": 4.979749587342486,
-      "grad_norm": 0.002001544926315546,
-      "learning_rate": 0.00019998777163835293,
-      "loss": 46.0,
-      "step": 30922
-    },
-    {
-      "epoch": 4.979910624421273,
-      "grad_norm": 0.009097103029489517,
-      "learning_rate": 0.0001999877708471838,
-      "loss": 46.0,
-      "step": 30923
-    },
-    {
-      "epoch": 4.980071661500061,
-      "grad_norm": 0.007415729574859142,
-      "learning_rate": 0.00019998777005598905,
-      "loss": 46.0,
-      "step": 30924
-    },
-    {
-      "epoch": 4.980232698578848,
-      "grad_norm": 0.0012105636997148395,
-      "learning_rate": 0.00019998776926476874,
-      "loss": 46.0,
-      "step": 30925
-    },
-    {
-      "epoch": 4.9803937356576355,
-      "grad_norm": 0.0026218751445412636,
-      "learning_rate": 0.00019998776847352282,
-      "loss": 46.0,
-      "step": 30926
-    },
-    {
-      "epoch": 4.980554772736422,
-      "grad_norm": 0.008122207596898079,
-      "learning_rate": 0.0001999877676822513,
-      "loss": 46.0,
-      "step": 30927
-    },
-    {
-      "epoch": 4.9807158098152104,
-      "grad_norm": 0.003486272180452943,
-      "learning_rate": 0.00019998776689095422,
-      "loss": 46.0,
-      "step": 30928
-    },
-    {
-      "epoch": 4.980876846893997,
-      "grad_norm": 0.0024311935994774103,
-      "learning_rate": 0.00019998776609963154,
-      "loss": 46.0,
-      "step": 30929
-    },
-    {
-      "epoch": 4.9810378839727845,
-      "grad_norm": 0.002731007756665349,
-      "learning_rate": 0.00019998776530828324,
-      "loss": 46.0,
-      "step": 30930
-    },
-    {
-      "epoch": 4.981198921051572,
-      "grad_norm": 0.0016660717083141208,
-      "learning_rate": 0.00019998776451690938,
-      "loss": 46.0,
-      "step": 30931
-    },
-    {
-      "epoch": 4.981359958130359,
-      "grad_norm": 0.011875721625983715,
-      "learning_rate": 0.0001999877637255099,
-      "loss": 46.0,
-      "step": 30932
-    },
-    {
-      "epoch": 4.981520995209147,
-      "grad_norm": 0.010018021799623966,
-      "learning_rate": 0.00019998776293408485,
-      "loss": 46.0,
-      "step": 30933
-    },
-    {
-      "epoch": 4.981682032287934,
-      "grad_norm": 0.004446617793291807,
-      "learning_rate": 0.0001999877621426342,
-      "loss": 46.0,
-      "step": 30934
-    },
-    {
-      "epoch": 4.981843069366722,
-      "grad_norm": 0.003570565953850746,
-      "learning_rate": 0.00019998776135115798,
-      "loss": 46.0,
-      "step": 30935
-    },
-    {
-      "epoch": 4.982004106445509,
-      "grad_norm": 0.001409512828104198,
-      "learning_rate": 0.00019998776055965616,
-      "loss": 46.0,
-      "step": 30936
-    },
-    {
-      "epoch": 4.982165143524297,
-      "grad_norm": 0.009436271153390408,
-      "learning_rate": 0.00019998775976812872,
-      "loss": 46.0,
-      "step": 30937
-    },
-    {
-      "epoch": 4.982326180603084,
-      "grad_norm": 0.0013256825041025877,
-      "learning_rate": 0.0001999877589765757,
-      "loss": 46.0,
-      "step": 30938
-    },
-    {
-      "epoch": 4.9824872176818715,
-      "grad_norm": 0.01066936831921339,
-      "learning_rate": 0.00019998775818499712,
-      "loss": 46.0,
-      "step": 30939
-    },
-    {
-      "epoch": 4.982648254760659,
-      "grad_norm": 0.0018738553626462817,
-      "learning_rate": 0.0001999877573933929,
-      "loss": 46.0,
-      "step": 30940
-    },
-    {
-      "epoch": 4.982809291839446,
-      "grad_norm": 0.003846400883048773,
-      "learning_rate": 0.00019998775660176312,
-      "loss": 46.0,
-      "step": 30941
-    },
-    {
-      "epoch": 4.982970328918233,
-      "grad_norm": 0.005156259518116713,
-      "learning_rate": 0.00019998775581010775,
-      "loss": 46.0,
-      "step": 30942
-    },
-    {
-      "epoch": 4.9831313659970204,
-      "grad_norm": 0.0021843810100108385,
-      "learning_rate": 0.00019998775501842676,
-      "loss": 46.0,
-      "step": 30943
-    },
-    {
-      "epoch": 4.983292403075808,
-      "grad_norm": 0.008953774347901344,
-      "learning_rate": 0.0001999877542267202,
-      "loss": 46.0,
-      "step": 30944
-    },
-    {
-      "epoch": 4.983453440154595,
-      "grad_norm": 0.013401445001363754,
-      "learning_rate": 0.00019998775343498806,
-      "loss": 46.0,
-      "step": 30945
-    },
-    {
-      "epoch": 4.983614477233383,
-      "grad_norm": 0.00585845485329628,
-      "learning_rate": 0.00019998775264323029,
-      "loss": 46.0,
-      "step": 30946
-    },
-    {
-      "epoch": 4.98377551431217,
-      "grad_norm": 0.0016057100147008896,
-      "learning_rate": 0.00019998775185144695,
-      "loss": 46.0,
-      "step": 30947
-    },
-    {
-      "epoch": 4.983936551390958,
-      "grad_norm": 0.0034353923983871937,
-      "learning_rate": 0.00019998775105963803,
-      "loss": 46.0,
-      "step": 30948
-    },
-    {
-      "epoch": 4.984097588469745,
-      "grad_norm": 0.0035877306945621967,
-      "learning_rate": 0.0001999877502678035,
-      "loss": 46.0,
-      "step": 30949
-    },
-    {
-      "epoch": 4.984258625548533,
-      "grad_norm": 0.002961766207590699,
-      "learning_rate": 0.00019998774947594338,
-      "loss": 46.0,
-      "step": 30950
-    },
-    {
-      "epoch": 4.98441966262732,
-      "grad_norm": 0.005100170150399208,
-      "learning_rate": 0.0001999877486840577,
-      "loss": 46.0,
-      "step": 30951
-    },
-    {
-      "epoch": 4.9845806997061075,
-      "grad_norm": 0.003901463933289051,
-      "learning_rate": 0.00019998774789214637,
-      "loss": 46.0,
-      "step": 30952
-    },
-    {
-      "epoch": 4.984741736784895,
-      "grad_norm": 0.003433484584093094,
-      "learning_rate": 0.0001999877471002095,
-      "loss": 46.0,
-      "step": 30953
-    },
-    {
-      "epoch": 4.984902773863682,
-      "grad_norm": 0.005060386378318071,
-      "learning_rate": 0.00019998774630824702,
-      "loss": 46.0,
-      "step": 30954
-    },
-    {
-      "epoch": 4.98506381094247,
-      "grad_norm": 0.004352663177996874,
-      "learning_rate": 0.00019998774551625894,
-      "loss": 46.0,
-      "step": 30955
-    },
-    {
-      "epoch": 4.985224848021257,
-      "grad_norm": 0.006874861195683479,
-      "learning_rate": 0.00019998774472424527,
-      "loss": 46.0,
-      "step": 30956
-    },
-    {
-      "epoch": 4.985385885100044,
-      "grad_norm": 0.006450616288930178,
-      "learning_rate": 0.00019998774393220603,
-      "loss": 46.0,
-      "step": 30957
-    },
-    {
-      "epoch": 4.985546922178831,
-      "grad_norm": 0.0057152630761265755,
-      "learning_rate": 0.00019998774314014116,
-      "loss": 46.0,
-      "step": 30958
-    },
-    {
-      "epoch": 4.985707959257619,
-      "grad_norm": 0.012109542265534401,
-      "learning_rate": 0.00019998774234805073,
-      "loss": 46.0,
-      "step": 30959
-    },
-    {
-      "epoch": 4.985868996336406,
-      "grad_norm": 0.0049367258325219154,
-      "learning_rate": 0.00019998774155593468,
-      "loss": 46.0,
-      "step": 30960
-    },
-    {
-      "epoch": 4.986030033415194,
-      "grad_norm": 0.0017551257042214274,
-      "learning_rate": 0.00019998774076379307,
-      "loss": 46.0,
-      "step": 30961
-    },
-    {
-      "epoch": 4.986191070493981,
-      "grad_norm": 0.006023516412824392,
-      "learning_rate": 0.00019998773997162585,
-      "loss": 46.0,
-      "step": 30962
-    },
-    {
-      "epoch": 4.986352107572769,
-      "grad_norm": 0.0025072949938476086,
-      "learning_rate": 0.00019998773917943304,
-      "loss": 46.0,
-      "step": 30963
-    },
-    {
-      "epoch": 4.986513144651556,
-      "grad_norm": 0.0038455044850707054,
-      "learning_rate": 0.00019998773838721465,
-      "loss": 46.0,
-      "step": 30964
-    },
-    {
-      "epoch": 4.9866741817303435,
-      "grad_norm": 0.004335750825703144,
-      "learning_rate": 0.00019998773759497066,
-      "loss": 46.0,
-      "step": 30965
-    },
-    {
-      "epoch": 4.986835218809131,
-      "grad_norm": 0.005389214958995581,
-      "learning_rate": 0.0001999877368027011,
-      "loss": 46.0,
-      "step": 30966
-    },
-    {
-      "epoch": 4.986996255887918,
-      "grad_norm": 0.010922829620540142,
-      "learning_rate": 0.0001999877360104059,
-      "loss": 46.0,
-      "step": 30967
-    },
-    {
-      "epoch": 4.987157292966706,
-      "grad_norm": 0.004410484340041876,
-      "learning_rate": 0.00019998773521808514,
-      "loss": 46.0,
-      "step": 30968
-    },
-    {
-      "epoch": 4.987318330045493,
-      "grad_norm": 0.0010219208197668195,
-      "learning_rate": 0.00019998773442573878,
-      "loss": 46.0,
-      "step": 30969
-    },
-    {
-      "epoch": 4.987479367124281,
-      "grad_norm": 0.004234891384840012,
-      "learning_rate": 0.00019998773363336683,
-      "loss": 46.0,
-      "step": 30970
-    },
-    {
-      "epoch": 4.987640404203068,
-      "grad_norm": 0.004864760208874941,
-      "learning_rate": 0.0001999877328409693,
-      "loss": 46.0,
-      "step": 30971
-    },
-    {
-      "epoch": 4.987801441281855,
-      "grad_norm": 0.007149561308324337,
-      "learning_rate": 0.00019998773204854615,
-      "loss": 46.0,
-      "step": 30972
-    },
-    {
-      "epoch": 4.987962478360642,
-      "grad_norm": 0.013999372720718384,
-      "learning_rate": 0.00019998773125609742,
-      "loss": 46.0,
-      "step": 30973
-    },
-    {
-      "epoch": 4.98812351543943,
-      "grad_norm": 0.0031189939472824335,
-      "learning_rate": 0.00019998773046362312,
-      "loss": 46.0,
-      "step": 30974
-    },
-    {
-      "epoch": 4.988284552518217,
-      "grad_norm": 0.007929399609565735,
-      "learning_rate": 0.00019998772967112321,
-      "loss": 46.0,
-      "step": 30975
-    },
-    {
-      "epoch": 4.988445589597005,
-      "grad_norm": 0.003621111623942852,
-      "learning_rate": 0.00019998772887859772,
-      "loss": 46.0,
-      "step": 30976
-    },
-    {
-      "epoch": 4.988606626675792,
-      "grad_norm": 0.0033273661974817514,
-      "learning_rate": 0.00019998772808604663,
-      "loss": 46.0,
-      "step": 30977
-    },
-    {
-      "epoch": 4.9887676637545795,
-      "grad_norm": 0.003927677869796753,
-      "learning_rate": 0.00019998772729346996,
-      "loss": 46.0,
-      "step": 30978
-    },
-    {
-      "epoch": 4.988928700833367,
-      "grad_norm": 0.001456919126212597,
-      "learning_rate": 0.00019998772650086765,
-      "loss": 46.0,
-      "step": 30979
-    },
-    {
-      "epoch": 4.989089737912154,
-      "grad_norm": 0.00237159407697618,
-      "learning_rate": 0.00019998772570823978,
-      "loss": 46.0,
-      "step": 30980
-    },
-    {
-      "epoch": 4.989250774990942,
-      "grad_norm": 0.01632472313940525,
-      "learning_rate": 0.00019998772491558632,
-      "loss": 46.0,
-      "step": 30981
-    },
-    {
-      "epoch": 4.989411812069729,
-      "grad_norm": 0.004613370168954134,
-      "learning_rate": 0.0001999877241229073,
-      "loss": 46.0,
-      "step": 30982
-    },
-    {
-      "epoch": 4.989572849148517,
-      "grad_norm": 0.006446159444749355,
-      "learning_rate": 0.00019998772333020264,
-      "loss": 46.0,
-      "step": 30983
-    },
-    {
-      "epoch": 4.989733886227304,
-      "grad_norm": 0.0079695088788867,
-      "learning_rate": 0.0001999877225374724,
-      "loss": 46.0,
-      "step": 30984
-    },
-    {
-      "epoch": 4.989894923306091,
-      "grad_norm": 0.0008088378235697746,
-      "learning_rate": 0.00019998772174471659,
-      "loss": 46.0,
-      "step": 30985
-    },
-    {
-      "epoch": 4.990055960384879,
-      "grad_norm": 0.003061580238863826,
-      "learning_rate": 0.0001999877209519352,
-      "loss": 46.0,
-      "step": 30986
-    },
-    {
-      "epoch": 4.990216997463666,
-      "grad_norm": 0.0015336141223087907,
-      "learning_rate": 0.00019998772015912815,
-      "loss": 46.0,
-      "step": 30987
-    },
-    {
-      "epoch": 4.990378034542453,
-      "grad_norm": 0.016620026901364326,
-      "learning_rate": 0.00019998771936629556,
-      "loss": 46.0,
-      "step": 30988
-    },
-    {
-      "epoch": 4.9905390716212406,
-      "grad_norm": 0.004312083590775728,
-      "learning_rate": 0.00019998771857343737,
-      "loss": 46.0,
-      "step": 30989
-    },
-    {
-      "epoch": 4.990700108700028,
-      "grad_norm": 0.012261553667485714,
-      "learning_rate": 0.00019998771778055358,
-      "loss": 46.0,
-      "step": 30990
-    },
-    {
-      "epoch": 4.9908611457788155,
-      "grad_norm": 0.0023435086477547884,
-      "learning_rate": 0.0001999877169876442,
-      "loss": 46.0,
-      "step": 30991
-    },
-    {
-      "epoch": 4.991022182857603,
-      "grad_norm": 0.0034246023278683424,
-      "learning_rate": 0.00019998771619470925,
-      "loss": 46.0,
-      "step": 30992
-    },
-    {
-      "epoch": 4.99118321993639,
-      "grad_norm": 0.0020012909080833197,
-      "learning_rate": 0.00019998771540174869,
-      "loss": 46.0,
-      "step": 30993
-    },
-    {
-      "epoch": 4.991344257015178,
-      "grad_norm": 0.001958669163286686,
-      "learning_rate": 0.00019998771460876254,
-      "loss": 46.0,
-      "step": 30994
-    },
-    {
-      "epoch": 4.991505294093965,
-      "grad_norm": 0.006679236888885498,
-      "learning_rate": 0.00019998771381575078,
-      "loss": 46.0,
-      "step": 30995
-    },
-    {
-      "epoch": 4.991666331172753,
-      "grad_norm": 0.004908496048301458,
-      "learning_rate": 0.00019998771302271346,
-      "loss": 46.0,
-      "step": 30996
-    },
-    {
-      "epoch": 4.99182736825154,
-      "grad_norm": 0.009609609842300415,
-      "learning_rate": 0.00019998771222965052,
-      "loss": 46.0,
-      "step": 30997
-    },
-    {
-      "epoch": 4.991988405330328,
-      "grad_norm": 0.008011335507035255,
-      "learning_rate": 0.000199987711436562,
-      "loss": 46.0,
-      "step": 30998
-    },
-    {
-      "epoch": 4.992149442409115,
-      "grad_norm": 0.007766086608171463,
-      "learning_rate": 0.0001999877106434479,
-      "loss": 46.0,
-      "step": 30999
-    },
-    {
-      "epoch": 4.992310479487902,
-      "grad_norm": 0.0015139388851821423,
-      "learning_rate": 0.0001999877098503082,
-      "loss": 46.0,
-      "step": 31000
-    },
-    {
-      "epoch": 4.99247151656669,
-      "grad_norm": 0.00339633715339005,
-      "learning_rate": 0.00019998770905714294,
-      "loss": 46.0,
-      "step": 31001
-    },
-    {
-      "epoch": 4.9926325536454765,
-      "grad_norm": 0.003292802022770047,
-      "learning_rate": 0.00019998770826395204,
-      "loss": 46.0,
-      "step": 31002
-    },
-    {
-      "epoch": 4.992793590724264,
-      "grad_norm": 0.008728443644940853,
-      "learning_rate": 0.00019998770747073555,
-      "loss": 46.0,
-      "step": 31003
-    },
-    {
-      "epoch": 4.992954627803051,
-      "grad_norm": 0.0023864663671702147,
-      "learning_rate": 0.00019998770667749348,
-      "loss": 46.0,
-      "step": 31004
-    },
-    {
-      "epoch": 4.993115664881839,
-      "grad_norm": 0.008173982612788677,
-      "learning_rate": 0.00019998770588422582,
-      "loss": 46.0,
-      "step": 31005
-    },
-    {
-      "epoch": 4.993276701960626,
-      "grad_norm": 0.0024232121650129557,
-      "learning_rate": 0.00019998770509093257,
-      "loss": 46.0,
-      "step": 31006
-    },
-    {
-      "epoch": 4.993437739039414,
-      "grad_norm": 0.008254874497652054,
-      "learning_rate": 0.00019998770429761374,
-      "loss": 46.0,
-      "step": 31007
-    },
-    {
-      "epoch": 4.993598776118201,
-      "grad_norm": 0.003584672464057803,
-      "learning_rate": 0.0001999877035042693,
-      "loss": 46.0,
-      "step": 31008
-    },
-    {
-      "epoch": 4.993759813196989,
-      "grad_norm": 0.04514848440885544,
-      "learning_rate": 0.00019998770271089926,
-      "loss": 46.0,
-      "step": 31009
-    },
-    {
-      "epoch": 4.993920850275776,
-      "grad_norm": 0.0050653438083827496,
-      "learning_rate": 0.00019998770191750363,
-      "loss": 46.0,
-      "step": 31010
-    },
-    {
-      "epoch": 4.994081887354564,
-      "grad_norm": 0.0041606230661273,
-      "learning_rate": 0.00019998770112408245,
-      "loss": 46.0,
-      "step": 31011
-    },
-    {
-      "epoch": 4.994242924433351,
-      "grad_norm": 0.008867884054780006,
-      "learning_rate": 0.00019998770033063563,
-      "loss": 46.0,
-      "step": 31012
-    },
-    {
-      "epoch": 4.9944039615121385,
-      "grad_norm": 0.0016287073958665133,
-      "learning_rate": 0.00019998769953716324,
-      "loss": 46.0,
-      "step": 31013
-    },
-    {
-      "epoch": 4.994564998590926,
-      "grad_norm": 0.0017289791721850634,
-      "learning_rate": 0.00019998769874366525,
-      "loss": 46.0,
-      "step": 31014
-    },
-    {
-      "epoch": 4.9947260356697125,
-      "grad_norm": 0.010679534636437893,
-      "learning_rate": 0.0001999876979501417,
-      "loss": 46.0,
-      "step": 31015
-    },
-    {
-      "epoch": 4.9948870727485,
-      "grad_norm": 0.006222427356988192,
-      "learning_rate": 0.00019998769715659251,
-      "loss": 46.0,
-      "step": 31016
-    },
-    {
-      "epoch": 4.995048109827287,
-      "grad_norm": 0.010056688450276852,
-      "learning_rate": 0.00019998769636301775,
-      "loss": 46.0,
-      "step": 31017
-    },
-    {
-      "epoch": 4.995209146906075,
-      "grad_norm": 0.004496424458920956,
-      "learning_rate": 0.0001999876955694174,
-      "loss": 46.0,
-      "step": 31018
-    },
-    {
-      "epoch": 4.995370183984862,
-      "grad_norm": 0.009411857463419437,
-      "learning_rate": 0.00019998769477579145,
-      "loss": 46.0,
-      "step": 31019
-    },
-    {
-      "epoch": 4.99553122106365,
-      "grad_norm": 0.0017789442790672183,
-      "learning_rate": 0.00019998769398213992,
-      "loss": 46.0,
-      "step": 31020
-    },
-    {
-      "epoch": 4.995692258142437,
-      "grad_norm": 0.005408520344644785,
-      "learning_rate": 0.0001999876931884628,
-      "loss": 46.0,
-      "step": 31021
-    },
-    {
-      "epoch": 4.995853295221225,
-      "grad_norm": 0.001626838929951191,
-      "learning_rate": 0.00019998769239476007,
-      "loss": 46.0,
-      "step": 31022
-    },
-    {
-      "epoch": 4.996014332300012,
-      "grad_norm": 0.017273297533392906,
-      "learning_rate": 0.00019998769160103176,
-      "loss": 46.0,
-      "step": 31023
-    },
-    {
-      "epoch": 4.9961753693788,
-      "grad_norm": 0.008846103213727474,
-      "learning_rate": 0.00019998769080727786,
-      "loss": 46.0,
-      "step": 31024
-    },
-    {
-      "epoch": 4.996336406457587,
-      "grad_norm": 0.0014638167340308428,
-      "learning_rate": 0.00019998769001349835,
-      "loss": 46.0,
-      "step": 31025
-    },
-    {
-      "epoch": 4.9964974435363745,
-      "grad_norm": 0.001910537132062018,
-      "learning_rate": 0.00019998768921969327,
-      "loss": 46.0,
-      "step": 31026
-    },
-    {
-      "epoch": 4.996658480615162,
-      "grad_norm": 0.009062094613909721,
-      "learning_rate": 0.0001999876884258626,
-      "loss": 46.0,
-      "step": 31027
-    },
-    {
-      "epoch": 4.996819517693949,
-      "grad_norm": 0.011714388616383076,
-      "learning_rate": 0.00019998768763200632,
-      "loss": 46.0,
-      "step": 31028
-    },
-    {
-      "epoch": 4.996980554772737,
-      "grad_norm": 0.0028681657277047634,
-      "learning_rate": 0.00019998768683812448,
-      "loss": 46.0,
-      "step": 31029
-    },
-    {
-      "epoch": 4.997141591851523,
-      "grad_norm": 0.0024218233302235603,
-      "learning_rate": 0.00019998768604421704,
-      "loss": 46.0,
-      "step": 31030
-    },
-    {
-      "epoch": 4.997302628930311,
-      "grad_norm": 0.012475757859647274,
-      "learning_rate": 0.00019998768525028397,
-      "loss": 46.0,
-      "step": 31031
-    },
-    {
-      "epoch": 4.997463666009098,
-      "grad_norm": 0.002611500909551978,
-      "learning_rate": 0.00019998768445632535,
-      "loss": 46.0,
-      "step": 31032
-    },
-    {
-      "epoch": 4.997624703087886,
-      "grad_norm": 0.0012553916312754154,
-      "learning_rate": 0.00019998768366234112,
-      "loss": 46.0,
-      "step": 31033
-    },
-    {
-      "epoch": 4.997785740166673,
-      "grad_norm": 0.0017350241541862488,
-      "learning_rate": 0.0001999876828683313,
-      "loss": 46.0,
-      "step": 31034
-    },
-    {
-      "epoch": 4.997946777245461,
-      "grad_norm": 0.005065203178673983,
-      "learning_rate": 0.00019998768207429588,
-      "loss": 46.0,
-      "step": 31035
-    },
-    {
-      "epoch": 4.998107814324248,
-      "grad_norm": 0.0024507571943104267,
-      "learning_rate": 0.00019998768128023489,
-      "loss": 46.0,
-      "step": 31036
-    },
-    {
-      "epoch": 4.998268851403036,
-      "grad_norm": 0.00270538660697639,
-      "learning_rate": 0.0001999876804861483,
-      "loss": 46.0,
-      "step": 31037
-    },
-    {
-      "epoch": 4.998429888481823,
-      "grad_norm": 0.006515856832265854,
-      "learning_rate": 0.0001999876796920361,
-      "loss": 46.0,
-      "step": 31038
-    },
-    {
-      "epoch": 4.9985909255606105,
-      "grad_norm": 0.007227238267660141,
-      "learning_rate": 0.00019998767889789832,
-      "loss": 46.0,
-      "step": 31039
-    },
-    {
-      "epoch": 4.998751962639398,
-      "grad_norm": 0.00120845518540591,
-      "learning_rate": 0.00019998767810373494,
-      "loss": 46.0,
-      "step": 31040
-    },
-    {
-      "epoch": 4.998912999718185,
-      "grad_norm": 0.0017949381144717336,
-      "learning_rate": 0.00019998767730954598,
-      "loss": 46.0,
-      "step": 31041
-    },
-    {
-      "epoch": 4.999074036796973,
-      "grad_norm": 0.001391031313687563,
-      "learning_rate": 0.00019998767651533143,
-      "loss": 46.0,
-      "step": 31042
-    },
-    {
-      "epoch": 4.99923507387576,
-      "grad_norm": 0.005238562356680632,
-      "learning_rate": 0.0001999876757210913,
-      "loss": 46.0,
-      "step": 31043
-    },
-    {
-      "epoch": 4.999396110954548,
-      "grad_norm": 0.001440562424249947,
-      "learning_rate": 0.00019998767492682555,
-      "loss": 46.0,
-      "step": 31044
-    },
-    {
-      "epoch": 4.999557148033334,
-      "grad_norm": 0.004359934478998184,
-      "learning_rate": 0.00019998767413253424,
-      "loss": 46.0,
-      "step": 31045
-    },
-    {
-      "epoch": 4.999718185112122,
-      "grad_norm": 0.017080603167414665,
-      "learning_rate": 0.00019998767333821732,
-      "loss": 46.0,
-      "step": 31046
-    },
-    {
-      "epoch": 4.999879222190909,
-      "grad_norm": 0.001398540218360722,
-      "learning_rate": 0.00019998767254387478,
-      "loss": 46.0,
-      "step": 31047
-    },
-    {
-      "epoch": 4.999879222190909,
-      "eval_loss": 11.5,
-      "eval_runtime": 15.2627,
-      "eval_samples_per_second": 171.333,
-      "eval_steps_per_second": 85.699,
-      "step": 31047
-    },
-    {
-      "epoch": 5.000080518539393,
-      "grad_norm": 0.0047907172702252865,
-      "learning_rate": 0.0001999876717495067,
-      "loss": 46.0,
-      "step": 31048
-    },
-    {
-      "epoch": 5.000241555618181,
-      "grad_norm": 0.003105146111920476,
-      "learning_rate": 0.000199987670955113,
-      "loss": 46.0,
-      "step": 31049
-    },
-    {
-      "epoch": 5.000402592696968,
-      "grad_norm": 0.0012852853396907449,
-      "learning_rate": 0.0001999876701606937,
-      "loss": 46.0,
-      "step": 31050
-    },
-    {
-      "epoch": 5.000563629775756,
-      "grad_norm": 0.005212837364524603,
-      "learning_rate": 0.00019998766936624884,
-      "loss": 46.0,
-      "step": 31051
-    },
-    {
-      "epoch": 5.000724666854543,
-      "grad_norm": 0.005906665697693825,
-      "learning_rate": 0.00019998766857177836,
-      "loss": 46.0,
-      "step": 31052
-    },
-    {
-      "epoch": 5.0008857039333305,
-      "grad_norm": 0.004046909511089325,
-      "learning_rate": 0.00019998766777728228,
-      "loss": 46.0,
-      "step": 31053
-    },
-    {
-      "epoch": 5.001046741012118,
-      "grad_norm": 0.006136893294751644,
-      "learning_rate": 0.00019998766698276063,
-      "loss": 46.0,
-      "step": 31054
-    },
-    {
-      "epoch": 5.001207778090905,
-      "grad_norm": 0.007463934365659952,
-      "learning_rate": 0.0001999876661882134,
-      "loss": 46.0,
-      "step": 31055
-    },
-    {
-      "epoch": 5.001368815169693,
-      "grad_norm": 0.006553539074957371,
-      "learning_rate": 0.00019998766539364054,
-      "loss": 46.0,
-      "step": 31056
-    },
-    {
-      "epoch": 5.00152985224848,
-      "grad_norm": 0.008825916796922684,
-      "learning_rate": 0.00019998766459904214,
-      "loss": 46.0,
-      "step": 31057
-    },
-    {
-      "epoch": 5.001690889327268,
-      "grad_norm": 0.006321832537651062,
-      "learning_rate": 0.0001999876638044181,
-      "loss": 46.0,
-      "step": 31058
-    },
-    {
-      "epoch": 5.001851926406055,
-      "grad_norm": 0.0011174511164426804,
-      "learning_rate": 0.0001999876630097685,
-      "loss": 46.0,
-      "step": 31059
-    },
-    {
-      "epoch": 5.002012963484843,
-      "grad_norm": 0.0032858788035809994,
-      "learning_rate": 0.0001999876622150933,
-      "loss": 46.0,
-      "step": 31060
-    },
-    {
-      "epoch": 5.00217400056363,
-      "grad_norm": 0.0013211389305070043,
-      "learning_rate": 0.00019998766142039252,
-      "loss": 46.0,
-      "step": 31061
-    },
-    {
-      "epoch": 5.002335037642418,
-      "grad_norm": 0.0038677456323057413,
-      "learning_rate": 0.0001999876606256661,
-      "loss": 46.0,
-      "step": 31062
-    },
-    {
-      "epoch": 5.002496074721204,
-      "grad_norm": 0.0029299515299499035,
-      "learning_rate": 0.00019998765983091413,
-      "loss": 46.0,
-      "step": 31063
-    },
-    {
-      "epoch": 5.002657111799992,
-      "grad_norm": 0.0014939854154363275,
-      "learning_rate": 0.00019998765903613656,
-      "loss": 46.0,
-      "step": 31064
-    },
-    {
-      "epoch": 5.002818148878779,
-      "grad_norm": 0.005311173386871815,
-      "learning_rate": 0.0001999876582413334,
-      "loss": 46.0,
-      "step": 31065
-    },
-    {
-      "epoch": 5.0029791859575665,
-      "grad_norm": 0.0051200916059315205,
-      "learning_rate": 0.00019998765744650462,
-      "loss": 46.0,
-      "step": 31066
-    },
-    {
-      "epoch": 5.003140223036354,
-      "grad_norm": 0.009907403960824013,
-      "learning_rate": 0.0001999876566516503,
-      "loss": 46.0,
-      "step": 31067
-    },
-    {
-      "epoch": 5.003301260115141,
-      "grad_norm": 0.003349072765558958,
-      "learning_rate": 0.00019998765585677036,
-      "loss": 46.0,
-      "step": 31068
-    },
-    {
-      "epoch": 5.003462297193929,
-      "grad_norm": 0.0053487312979996204,
-      "learning_rate": 0.00019998765506186483,
-      "loss": 46.0,
-      "step": 31069
-    },
-    {
-      "epoch": 5.003623334272716,
-      "grad_norm": 0.0007635327056050301,
-      "learning_rate": 0.0001999876542669337,
-      "loss": 46.0,
-      "step": 31070
-    },
-    {
-      "epoch": 5.003784371351504,
-      "grad_norm": 0.004554382991045713,
-      "learning_rate": 0.000199987653471977,
-      "loss": 46.0,
-      "step": 31071
-    },
-    {
-      "epoch": 5.003945408430291,
-      "grad_norm": 0.00705056544393301,
-      "learning_rate": 0.0001999876526769947,
-      "loss": 46.0,
-      "step": 31072
-    },
-    {
-      "epoch": 5.004106445509079,
-      "grad_norm": 0.0008869162993505597,
-      "learning_rate": 0.0001999876518819868,
-      "loss": 46.0,
-      "step": 31073
-    },
-    {
-      "epoch": 5.004267482587866,
-      "grad_norm": 0.002091227099299431,
-      "learning_rate": 0.0001999876510869533,
-      "loss": 46.0,
-      "step": 31074
-    },
-    {
-      "epoch": 5.004428519666654,
-      "grad_norm": 0.004489665850996971,
-      "learning_rate": 0.00019998765029189422,
-      "loss": 46.0,
-      "step": 31075
-    },
-    {
-      "epoch": 5.004589556745441,
-      "grad_norm": 0.00648216949775815,
-      "learning_rate": 0.00019998764949680955,
-      "loss": 46.0,
-      "step": 31076
-    },
-    {
-      "epoch": 5.004750593824228,
-      "grad_norm": 0.0076082851737737656,
-      "learning_rate": 0.0001999876487016993,
-      "loss": 46.0,
-      "step": 31077
-    },
-    {
-      "epoch": 5.004911630903015,
-      "grad_norm": 0.0014451676979660988,
-      "learning_rate": 0.00019998764790656344,
-      "loss": 46.0,
-      "step": 31078
-    },
-    {
-      "epoch": 5.0050726679818025,
-      "grad_norm": 0.0011559949489310384,
-      "learning_rate": 0.00019998764711140197,
-      "loss": 46.0,
-      "step": 31079
-    },
-    {
-      "epoch": 5.00523370506059,
-      "grad_norm": 0.008217500522732735,
-      "learning_rate": 0.00019998764631621495,
-      "loss": 46.0,
-      "step": 31080
-    },
-    {
-      "epoch": 5.005394742139377,
-      "grad_norm": 0.002597081009298563,
-      "learning_rate": 0.00019998764552100231,
-      "loss": 46.0,
-      "step": 31081
-    },
-    {
-      "epoch": 5.005555779218165,
-      "grad_norm": 0.002886929316446185,
-      "learning_rate": 0.0001999876447257641,
-      "loss": 46.0,
-      "step": 31082
-    },
-    {
-      "epoch": 5.005716816296952,
-      "grad_norm": 0.011141336522996426,
-      "learning_rate": 0.00019998764393050028,
-      "loss": 46.0,
-      "step": 31083
-    },
-    {
-      "epoch": 5.00587785337574,
-      "grad_norm": 0.008937749080359936,
-      "learning_rate": 0.00019998764313521088,
-      "loss": 46.0,
-      "step": 31084
-    },
-    {
-      "epoch": 5.006038890454527,
-      "grad_norm": 0.011391947977244854,
-      "learning_rate": 0.0001999876423398959,
-      "loss": 46.0,
-      "step": 31085
-    },
-    {
-      "epoch": 5.006199927533315,
-      "grad_norm": 0.0061226957477629185,
-      "learning_rate": 0.0001999876415445553,
-      "loss": 46.0,
-      "step": 31086
-    },
-    {
-      "epoch": 5.006360964612102,
-      "grad_norm": 0.0018338634399697185,
-      "learning_rate": 0.0001999876407491891,
-      "loss": 46.0,
-      "step": 31087
-    },
-    {
-      "epoch": 5.00652200169089,
-      "grad_norm": 0.006106365472078323,
-      "learning_rate": 0.00019998763995379733,
-      "loss": 46.0,
-      "step": 31088
-    },
-    {
-      "epoch": 5.006683038769677,
-      "grad_norm": 0.007486812770366669,
-      "learning_rate": 0.00019998763915837997,
-      "loss": 46.0,
-      "step": 31089
-    },
-    {
-      "epoch": 5.0068440758484645,
-      "grad_norm": 0.0017316878074780107,
-      "learning_rate": 0.00019998763836293702,
-      "loss": 46.0,
-      "step": 31090
-    },
-    {
-      "epoch": 5.007005112927252,
-      "grad_norm": 0.0035680984146893024,
-      "learning_rate": 0.0001999876375674685,
-      "loss": 46.0,
-      "step": 31091
-    },
-    {
-      "epoch": 5.0071661500060385,
-      "grad_norm": 0.00277877738699317,
-      "learning_rate": 0.00019998763677197434,
-      "loss": 46.0,
-      "step": 31092
-    },
-    {
-      "epoch": 5.007327187084826,
-      "grad_norm": 0.004459621850401163,
-      "learning_rate": 0.00019998763597645463,
-      "loss": 46.0,
-      "step": 31093
-    },
-    {
-      "epoch": 5.007488224163613,
-      "grad_norm": 0.006595935672521591,
-      "learning_rate": 0.0001999876351809093,
-      "loss": 46.0,
-      "step": 31094
-    },
-    {
-      "epoch": 5.007649261242401,
-      "grad_norm": 0.0010635115904733539,
-      "learning_rate": 0.00019998763438533836,
-      "loss": 46.0,
-      "step": 31095
-    },
-    {
-      "epoch": 5.007810298321188,
-      "grad_norm": 0.003824312472715974,
-      "learning_rate": 0.0001999876335897419,
-      "loss": 46.0,
-      "step": 31096
-    },
-    {
-      "epoch": 5.007971335399976,
-      "grad_norm": 0.00623067282140255,
-      "learning_rate": 0.00019998763279411978,
-      "loss": 46.0,
-      "step": 31097
-    },
-    {
-      "epoch": 5.008132372478763,
-      "grad_norm": 0.0035313498228788376,
-      "learning_rate": 0.00019998763199847208,
-      "loss": 46.0,
-      "step": 31098
-    },
-    {
-      "epoch": 5.008293409557551,
-      "grad_norm": 0.005680904723703861,
-      "learning_rate": 0.0001999876312027988,
-      "loss": 46.0,
-      "step": 31099
-    },
-    {
-      "epoch": 5.008454446636338,
-      "grad_norm": 0.006670357659459114,
-      "learning_rate": 0.00019998763040709994,
-      "loss": 46.0,
-      "step": 31100
-    },
-    {
-      "epoch": 5.0086154837151255,
-      "grad_norm": 0.004993006121367216,
-      "learning_rate": 0.00019998762961137548,
-      "loss": 46.0,
-      "step": 31101
-    },
-    {
-      "epoch": 5.008776520793913,
-      "grad_norm": 0.007767189759761095,
-      "learning_rate": 0.0001999876288156254,
-      "loss": 46.0,
-      "step": 31102
-    },
-    {
-      "epoch": 5.0089375578727005,
-      "grad_norm": 0.007792117074131966,
-      "learning_rate": 0.00019998762801984977,
-      "loss": 46.0,
-      "step": 31103
-    },
-    {
-      "epoch": 5.009098594951488,
-      "grad_norm": 0.003257140750065446,
-      "learning_rate": 0.00019998762722404855,
-      "loss": 46.0,
-      "step": 31104
-    },
-    {
-      "epoch": 5.009259632030275,
-      "grad_norm": 0.014960839413106441,
-      "learning_rate": 0.0001999876264282217,
-      "loss": 46.0,
-      "step": 31105
-    },
-    {
-      "epoch": 5.009420669109062,
-      "grad_norm": 0.008353774435818195,
-      "learning_rate": 0.00019998762563236926,
-      "loss": 46.0,
-      "step": 31106
-    },
-    {
-      "epoch": 5.009581706187849,
-      "grad_norm": 0.0054909707978367805,
-      "learning_rate": 0.00019998762483649127,
-      "loss": 46.0,
-      "step": 31107
-    },
-    {
-      "epoch": 5.009742743266637,
-      "grad_norm": 0.003979809116572142,
-      "learning_rate": 0.00019998762404058767,
-      "loss": 46.0,
-      "step": 31108
-    },
-    {
-      "epoch": 5.009903780345424,
-      "grad_norm": 0.005869802553206682,
-      "learning_rate": 0.00019998762324465846,
-      "loss": 46.0,
-      "step": 31109
-    },
-    {
-      "epoch": 5.010064817424212,
-      "grad_norm": 0.011984726414084435,
-      "learning_rate": 0.00019998762244870366,
-      "loss": 46.0,
-      "step": 31110
-    },
-    {
-      "epoch": 5.010225854502999,
-      "grad_norm": 0.001436824444681406,
-      "learning_rate": 0.00019998762165272327,
-      "loss": 46.0,
-      "step": 31111
-    },
-    {
-      "epoch": 5.010386891581787,
-      "grad_norm": 0.003839449491351843,
-      "learning_rate": 0.0001999876208567173,
-      "loss": 46.0,
-      "step": 31112
-    },
-    {
-      "epoch": 5.010547928660574,
-      "grad_norm": 0.017565922811627388,
-      "learning_rate": 0.00019998762006068573,
-      "loss": 46.0,
-      "step": 31113
-    },
-    {
-      "epoch": 5.0107089657393615,
-      "grad_norm": 0.0032677168492227793,
-      "learning_rate": 0.00019998761926462858,
-      "loss": 46.0,
-      "step": 31114
-    },
-    {
-      "epoch": 5.010870002818149,
-      "grad_norm": 0.0022214825730770826,
-      "learning_rate": 0.00019998761846854585,
-      "loss": 46.0,
-      "step": 31115
-    },
-    {
-      "epoch": 5.011031039896936,
-      "grad_norm": 0.006211854051798582,
-      "learning_rate": 0.0001999876176724375,
-      "loss": 46.0,
-      "step": 31116
-    },
-    {
-      "epoch": 5.011192076975724,
-      "grad_norm": 0.018349047750234604,
-      "learning_rate": 0.00019998761687630356,
-      "loss": 46.0,
-      "step": 31117
-    },
-    {
-      "epoch": 5.011353114054511,
-      "grad_norm": 0.00869948323816061,
-      "learning_rate": 0.00019998761608014403,
-      "loss": 46.0,
-      "step": 31118
-    },
-    {
-      "epoch": 5.011514151133299,
-      "grad_norm": 0.005436311475932598,
-      "learning_rate": 0.00019998761528395892,
-      "loss": 46.0,
-      "step": 31119
-    },
-    {
-      "epoch": 5.011675188212086,
-      "grad_norm": 0.0035337633453309536,
-      "learning_rate": 0.0001999876144877482,
-      "loss": 46.0,
-      "step": 31120
-    },
-    {
-      "epoch": 5.011836225290873,
-      "grad_norm": 0.003181623062118888,
-      "learning_rate": 0.00019998761369151194,
-      "loss": 46.0,
-      "step": 31121
-    },
-    {
-      "epoch": 5.01199726236966,
-      "grad_norm": 0.0017693424597382545,
-      "learning_rate": 0.00019998761289525004,
-      "loss": 46.0,
-      "step": 31122
-    },
-    {
-      "epoch": 5.012158299448448,
-      "grad_norm": 0.011082902550697327,
-      "learning_rate": 0.00019998761209896255,
-      "loss": 46.0,
-      "step": 31123
-    },
-    {
-      "epoch": 5.012319336527235,
-      "grad_norm": 0.001154599478468299,
-      "learning_rate": 0.0001999876113026495,
-      "loss": 46.0,
-      "step": 31124
-    },
-    {
-      "epoch": 5.012480373606023,
-      "grad_norm": 0.00754148792475462,
-      "learning_rate": 0.0001999876105063108,
-      "loss": 46.0,
-      "step": 31125
-    },
-    {
-      "epoch": 5.01264141068481,
-      "grad_norm": 0.0011923379497602582,
-      "learning_rate": 0.00019998760970994656,
-      "loss": 46.0,
-      "step": 31126
-    },
-    {
-      "epoch": 5.0128024477635975,
-      "grad_norm": 0.00216309423558414,
-      "learning_rate": 0.0001999876089135567,
-      "loss": 46.0,
-      "step": 31127
-    },
-    {
-      "epoch": 5.012963484842385,
-      "grad_norm": 0.014163664542138577,
-      "learning_rate": 0.00019998760811714127,
-      "loss": 46.0,
-      "step": 31128
-    },
-    {
-      "epoch": 5.013124521921172,
-      "grad_norm": 0.002531968289986253,
-      "learning_rate": 0.00019998760732070023,
-      "loss": 46.0,
-      "step": 31129
-    },
-    {
-      "epoch": 5.01328555899996,
-      "grad_norm": 0.0022920039482414722,
-      "learning_rate": 0.0001999876065242336,
-      "loss": 46.0,
-      "step": 31130
-    },
-    {
-      "epoch": 5.013446596078747,
-      "grad_norm": 0.004964747000485659,
-      "learning_rate": 0.0001999876057277414,
-      "loss": 46.0,
-      "step": 31131
-    },
-    {
-      "epoch": 5.013607633157535,
-      "grad_norm": 0.002050135051831603,
-      "learning_rate": 0.0001999876049312236,
-      "loss": 46.0,
-      "step": 31132
-    },
-    {
-      "epoch": 5.013768670236322,
-      "grad_norm": 0.014891921542584896,
-      "learning_rate": 0.0001999876041346802,
-      "loss": 46.0,
-      "step": 31133
-    },
-    {
-      "epoch": 5.01392970731511,
-      "grad_norm": 0.0050053102895617485,
-      "learning_rate": 0.0001999876033381112,
-      "loss": 46.0,
-      "step": 31134
-    },
-    {
-      "epoch": 5.014090744393897,
-      "grad_norm": 0.002103540115058422,
-      "learning_rate": 0.00019998760254151662,
-      "loss": 46.0,
-      "step": 31135
-    },
-    {
-      "epoch": 5.014251781472684,
-      "grad_norm": 0.013208774849772453,
-      "learning_rate": 0.00019998760174489645,
-      "loss": 46.0,
-      "step": 31136
-    },
-    {
-      "epoch": 5.014412818551471,
-      "grad_norm": 0.0017268150113523006,
-      "learning_rate": 0.00019998760094825068,
-      "loss": 46.0,
-      "step": 31137
-    },
-    {
-      "epoch": 5.014573855630259,
-      "grad_norm": 0.0022069367114454508,
-      "learning_rate": 0.00019998760015157933,
-      "loss": 46.0,
-      "step": 31138
-    },
-    {
-      "epoch": 5.014734892709046,
-      "grad_norm": 0.01026224810630083,
-      "learning_rate": 0.00019998759935488237,
-      "loss": 46.0,
-      "step": 31139
-    },
-    {
-      "epoch": 5.0148959297878335,
-      "grad_norm": 0.003364915493875742,
-      "learning_rate": 0.00019998759855815985,
-      "loss": 46.0,
-      "step": 31140
-    },
-    {
-      "epoch": 5.015056966866621,
-      "grad_norm": 0.0015189736150205135,
-      "learning_rate": 0.0001999875977614117,
-      "loss": 46.0,
-      "step": 31141
-    },
-    {
-      "epoch": 5.015218003945408,
-      "grad_norm": 0.001997397281229496,
-      "learning_rate": 0.000199987596964638,
-      "loss": 46.0,
-      "step": 31142
-    },
-    {
-      "epoch": 5.015379041024196,
-      "grad_norm": 0.003704332746565342,
-      "learning_rate": 0.00019998759616783867,
-      "loss": 46.0,
-      "step": 31143
-    },
-    {
-      "epoch": 5.015540078102983,
-      "grad_norm": 0.002264477778226137,
-      "learning_rate": 0.00019998759537101374,
-      "loss": 46.0,
-      "step": 31144
-    },
-    {
-      "epoch": 5.015701115181771,
-      "grad_norm": 0.0014337929897010326,
-      "learning_rate": 0.00019998759457416325,
-      "loss": 46.0,
-      "step": 31145
-    },
-    {
-      "epoch": 5.015862152260558,
-      "grad_norm": 0.0015657455660402775,
-      "learning_rate": 0.00019998759377728718,
-      "loss": 46.0,
-      "step": 31146
-    },
-    {
-      "epoch": 5.016023189339346,
-      "grad_norm": 0.01921684853732586,
-      "learning_rate": 0.0001999875929803855,
-      "loss": 46.0,
-      "step": 31147
-    },
-    {
-      "epoch": 5.016184226418133,
-      "grad_norm": 0.0017174036474898458,
-      "learning_rate": 0.00019998759218345822,
-      "loss": 46.0,
-      "step": 31148
-    },
-    {
-      "epoch": 5.016345263496921,
-      "grad_norm": 0.0022288498003035784,
-      "learning_rate": 0.00019998759138650535,
-      "loss": 46.0,
-      "step": 31149
-    },
-    {
-      "epoch": 5.016506300575707,
-      "grad_norm": 0.002555317245423794,
-      "learning_rate": 0.0001999875905895269,
-      "loss": 46.0,
-      "step": 31150
-    },
-    {
-      "epoch": 5.016667337654495,
-      "grad_norm": 0.00671594450250268,
-      "learning_rate": 0.00019998758979252284,
-      "loss": 46.0,
-      "step": 31151
-    },
-    {
-      "epoch": 5.016828374733282,
-      "grad_norm": 0.001396189909428358,
-      "learning_rate": 0.0001999875889954932,
-      "loss": 46.0,
-      "step": 31152
-    },
-    {
-      "epoch": 5.0169894118120695,
-      "grad_norm": 0.006512317340821028,
-      "learning_rate": 0.00019998758819843798,
-      "loss": 46.0,
-      "step": 31153
-    },
-    {
-      "epoch": 5.017150448890857,
-      "grad_norm": 0.007165046408772469,
-      "learning_rate": 0.00019998758740135715,
-      "loss": 46.0,
-      "step": 31154
-    },
-    {
-      "epoch": 5.017311485969644,
-      "grad_norm": 0.0015913145616650581,
-      "learning_rate": 0.00019998758660425074,
-      "loss": 46.0,
-      "step": 31155
-    },
-    {
-      "epoch": 5.017472523048432,
-      "grad_norm": 0.008072750642895699,
-      "learning_rate": 0.00019998758580711874,
-      "loss": 46.0,
-      "step": 31156
-    },
-    {
-      "epoch": 5.017633560127219,
-      "grad_norm": 0.01012209989130497,
-      "learning_rate": 0.00019998758500996112,
-      "loss": 46.0,
-      "step": 31157
-    },
-    {
-      "epoch": 5.017794597206007,
-      "grad_norm": 0.0016868539387360215,
-      "learning_rate": 0.00019998758421277795,
-      "loss": 46.0,
-      "step": 31158
-    },
-    {
-      "epoch": 5.017955634284794,
-      "grad_norm": 0.0039093708619475365,
-      "learning_rate": 0.00019998758341556916,
-      "loss": 46.0,
-      "step": 31159
-    },
-    {
-      "epoch": 5.018116671363582,
-      "grad_norm": 0.005279083736240864,
-      "learning_rate": 0.00019998758261833478,
-      "loss": 46.0,
-      "step": 31160
-    },
-    {
-      "epoch": 5.018277708442369,
-      "grad_norm": 0.01379369292408228,
-      "learning_rate": 0.0001999875818210748,
-      "loss": 46.0,
-      "step": 31161
-    },
-    {
-      "epoch": 5.0184387455211565,
-      "grad_norm": 0.004628817550837994,
-      "learning_rate": 0.00019998758102378927,
-      "loss": 46.0,
-      "step": 31162
-    },
-    {
-      "epoch": 5.018599782599944,
-      "grad_norm": 0.004593974910676479,
-      "learning_rate": 0.00019998758022647813,
-      "loss": 46.0,
-      "step": 31163
-    },
-    {
-      "epoch": 5.018760819678731,
-      "grad_norm": 0.007389673963189125,
-      "learning_rate": 0.00019998757942914138,
-      "loss": 46.0,
-      "step": 31164
-    },
-    {
-      "epoch": 5.018921856757518,
-      "grad_norm": 0.009068983606994152,
-      "learning_rate": 0.00019998757863177904,
-      "loss": 46.0,
-      "step": 31165
-    },
-    {
-      "epoch": 5.0190828938363055,
-      "grad_norm": 0.005277920980006456,
-      "learning_rate": 0.00019998757783439112,
-      "loss": 46.0,
-      "step": 31166
-    },
-    {
-      "epoch": 5.019243930915093,
-      "grad_norm": 0.0037230681627988815,
-      "learning_rate": 0.0001999875770369776,
-      "loss": 46.0,
-      "step": 31167
-    },
-    {
-      "epoch": 5.01940496799388,
-      "grad_norm": 0.0019074546871706843,
-      "learning_rate": 0.0001999875762395385,
-      "loss": 46.0,
-      "step": 31168
-    },
-    {
-      "epoch": 5.019566005072668,
-      "grad_norm": 0.008009098470211029,
-      "learning_rate": 0.0001999875754420738,
-      "loss": 46.0,
-      "step": 31169
-    },
-    {
-      "epoch": 5.019727042151455,
-      "grad_norm": 0.0022549203131347895,
-      "learning_rate": 0.00019998757464458349,
-      "loss": 46.0,
-      "step": 31170
-    },
-    {
-      "epoch": 5.019888079230243,
-      "grad_norm": 0.001208345522172749,
-      "learning_rate": 0.00019998757384706762,
-      "loss": 46.0,
-      "step": 31171
-    },
-    {
-      "epoch": 5.02004911630903,
-      "grad_norm": 0.00873639713972807,
-      "learning_rate": 0.00019998757304952612,
-      "loss": 46.0,
-      "step": 31172
-    },
-    {
-      "epoch": 5.020210153387818,
-      "grad_norm": 0.012708732858300209,
-      "learning_rate": 0.00019998757225195908,
-      "loss": 46.0,
-      "step": 31173
-    },
-    {
-      "epoch": 5.020371190466605,
-      "grad_norm": 0.008436281234025955,
-      "learning_rate": 0.0001999875714543664,
-      "loss": 46.0,
-      "step": 31174
-    },
-    {
-      "epoch": 5.0205322275453925,
-      "grad_norm": 0.013966712169349194,
-      "learning_rate": 0.00019998757065674817,
-      "loss": 46.0,
-      "step": 31175
-    },
-    {
-      "epoch": 5.02069326462418,
-      "grad_norm": 0.002289323601871729,
-      "learning_rate": 0.00019998756985910432,
-      "loss": 46.0,
-      "step": 31176
-    },
-    {
-      "epoch": 5.020854301702967,
-      "grad_norm": 0.008387654088437557,
-      "learning_rate": 0.00019998756906143488,
-      "loss": 46.0,
-      "step": 31177
-    },
-    {
-      "epoch": 5.021015338781755,
-      "grad_norm": 0.000529229873791337,
-      "learning_rate": 0.00019998756826373985,
-      "loss": 46.0,
-      "step": 31178
-    },
-    {
-      "epoch": 5.021176375860542,
-      "grad_norm": 0.005441870074719191,
-      "learning_rate": 0.00019998756746601926,
-      "loss": 46.0,
-      "step": 31179
-    },
-    {
-      "epoch": 5.021337412939329,
-      "grad_norm": 0.0012339279055595398,
-      "learning_rate": 0.00019998756666827304,
-      "loss": 46.0,
-      "step": 31180
-    },
-    {
-      "epoch": 5.021498450018116,
-      "grad_norm": 0.01439022645354271,
-      "learning_rate": 0.00019998756587050122,
-      "loss": 46.0,
-      "step": 31181
-    },
-    {
-      "epoch": 5.021659487096904,
-      "grad_norm": 0.002546433825045824,
-      "learning_rate": 0.00019998756507270385,
-      "loss": 46.0,
-      "step": 31182
-    },
-    {
-      "epoch": 5.021820524175691,
-      "grad_norm": 0.0054767681285738945,
-      "learning_rate": 0.00019998756427488088,
-      "loss": 46.0,
-      "step": 31183
-    },
-    {
-      "epoch": 5.021981561254479,
-      "grad_norm": 0.012280695140361786,
-      "learning_rate": 0.00019998756347703228,
-      "loss": 46.0,
-      "step": 31184
-    },
-    {
-      "epoch": 5.022142598333266,
-      "grad_norm": 0.013114583678543568,
-      "learning_rate": 0.00019998756267915812,
-      "loss": 46.0,
-      "step": 31185
-    },
-    {
-      "epoch": 5.022303635412054,
-      "grad_norm": 0.00488185603171587,
-      "learning_rate": 0.00019998756188125837,
-      "loss": 46.0,
-      "step": 31186
-    },
-    {
-      "epoch": 5.022464672490841,
-      "grad_norm": 0.0020085610449314117,
-      "learning_rate": 0.000199987561083333,
-      "loss": 46.0,
-      "step": 31187
-    },
-    {
-      "epoch": 5.0226257095696285,
-      "grad_norm": 0.011085419915616512,
-      "learning_rate": 0.00019998756028538208,
-      "loss": 46.0,
-      "step": 31188
-    },
-    {
-      "epoch": 5.022786746648416,
-      "grad_norm": 0.0025856555439531803,
-      "learning_rate": 0.00019998755948740554,
-      "loss": 46.0,
-      "step": 31189
-    },
-    {
-      "epoch": 5.022947783727203,
-      "grad_norm": 0.005422505084425211,
-      "learning_rate": 0.0001999875586894034,
-      "loss": 46.0,
-      "step": 31190
-    },
-    {
-      "epoch": 5.023108820805991,
-      "grad_norm": 0.0019389294320717454,
-      "learning_rate": 0.0001999875578913757,
-      "loss": 46.0,
-      "step": 31191
-    },
-    {
-      "epoch": 5.023269857884778,
-      "grad_norm": 0.005671205930411816,
-      "learning_rate": 0.0001999875570933224,
-      "loss": 46.0,
-      "step": 31192
-    },
-    {
-      "epoch": 5.023430894963566,
-      "grad_norm": 0.013513694517314434,
-      "learning_rate": 0.00019998755629524348,
-      "loss": 46.0,
-      "step": 31193
-    },
-    {
-      "epoch": 5.023591932042352,
-      "grad_norm": 0.005603449884802103,
-      "learning_rate": 0.000199987555497139,
-      "loss": 46.0,
-      "step": 31194
-    },
-    {
-      "epoch": 5.02375296912114,
-      "grad_norm": 0.003135878825560212,
-      "learning_rate": 0.00019998755469900892,
-      "loss": 46.0,
-      "step": 31195
-    },
-    {
-      "epoch": 5.023914006199927,
-      "grad_norm": 0.003170355688780546,
-      "learning_rate": 0.00019998755390085324,
-      "loss": 46.0,
-      "step": 31196
-    },
-    {
-      "epoch": 5.024075043278715,
-      "grad_norm": 0.009705073200166225,
-      "learning_rate": 0.00019998755310267197,
-      "loss": 46.0,
-      "step": 31197
-    },
-    {
-      "epoch": 5.024236080357502,
-      "grad_norm": 0.0016705053858458996,
-      "learning_rate": 0.0001999875523044651,
-      "loss": 46.0,
-      "step": 31198
-    },
-    {
-      "epoch": 5.02439711743629,
-      "grad_norm": 0.0015336641808971763,
-      "learning_rate": 0.00019998755150623269,
-      "loss": 46.0,
-      "step": 31199
-    },
-    {
-      "epoch": 5.024558154515077,
-      "grad_norm": 0.005949017126113176,
-      "learning_rate": 0.0001999875507079746,
-      "loss": 46.0,
-      "step": 31200
-    },
-    {
-      "epoch": 5.0247191915938645,
-      "grad_norm": 0.013331376947462559,
-      "learning_rate": 0.000199987549909691,
-      "loss": 46.0,
-      "step": 31201
-    },
-    {
-      "epoch": 5.024880228672652,
-      "grad_norm": 0.013253557495772839,
-      "learning_rate": 0.00019998754911138177,
-      "loss": 46.0,
-      "step": 31202
-    },
-    {
-      "epoch": 5.025041265751439,
-      "grad_norm": 0.0014974800869822502,
-      "learning_rate": 0.00019998754831304695,
-      "loss": 46.0,
-      "step": 31203
-    },
-    {
-      "epoch": 5.025202302830227,
-      "grad_norm": 0.004102327395230532,
-      "learning_rate": 0.00019998754751468652,
-      "loss": 46.0,
-      "step": 31204
-    },
-    {
-      "epoch": 5.025363339909014,
-      "grad_norm": 0.021181462332606316,
-      "learning_rate": 0.00019998754671630056,
-      "loss": 46.0,
-      "step": 31205
-    },
-    {
-      "epoch": 5.025524376987802,
-      "grad_norm": 0.01931837573647499,
-      "learning_rate": 0.00019998754591788896,
-      "loss": 46.0,
-      "step": 31206
-    },
-    {
-      "epoch": 5.025685414066589,
-      "grad_norm": 0.006445386912673712,
-      "learning_rate": 0.00019998754511945177,
-      "loss": 46.0,
-      "step": 31207
-    },
-    {
-      "epoch": 5.025846451145377,
-      "grad_norm": 0.004176761489361525,
-      "learning_rate": 0.00019998754432098902,
-      "loss": 46.0,
-      "step": 31208
-    },
-    {
-      "epoch": 5.026007488224163,
-      "grad_norm": 0.009595552459359169,
-      "learning_rate": 0.00019998754352250063,
-      "loss": 46.0,
-      "step": 31209
-    },
-    {
-      "epoch": 5.026168525302951,
-      "grad_norm": 0.001910917810164392,
-      "learning_rate": 0.00019998754272398665,
-      "loss": 46.0,
-      "step": 31210
-    },
-    {
-      "epoch": 5.026329562381738,
-      "grad_norm": 0.014299686998128891,
-      "learning_rate": 0.0001999875419254471,
-      "loss": 46.0,
-      "step": 31211
-    },
-    {
-      "epoch": 5.026490599460526,
-      "grad_norm": 0.002501022769138217,
-      "learning_rate": 0.00019998754112688199,
-      "loss": 46.0,
-      "step": 31212
-    },
-    {
-      "epoch": 5.026651636539313,
-      "grad_norm": 0.001806715619750321,
-      "learning_rate": 0.00019998754032829125,
-      "loss": 46.0,
-      "step": 31213
-    },
-    {
-      "epoch": 5.0268126736181005,
-      "grad_norm": 0.006467574741691351,
-      "learning_rate": 0.00019998753952967492,
-      "loss": 46.0,
-      "step": 31214
-    },
-    {
-      "epoch": 5.026973710696888,
-      "grad_norm": 0.005253058858215809,
-      "learning_rate": 0.000199987538731033,
-      "loss": 46.0,
-      "step": 31215
-    },
-    {
-      "epoch": 5.027134747775675,
-      "grad_norm": 0.009899591095745564,
-      "learning_rate": 0.00019998753793236548,
-      "loss": 46.0,
-      "step": 31216
-    },
-    {
-      "epoch": 5.027295784854463,
-      "grad_norm": 0.012147229164838791,
-      "learning_rate": 0.0001999875371336724,
-      "loss": 46.0,
-      "step": 31217
-    },
-    {
-      "epoch": 5.02745682193325,
-      "grad_norm": 0.0035418584011495113,
-      "learning_rate": 0.0001999875363349537,
-      "loss": 46.0,
-      "step": 31218
-    },
-    {
-      "epoch": 5.027617859012038,
-      "grad_norm": 0.007967657409608364,
-      "learning_rate": 0.00019998753553620942,
-      "loss": 46.0,
-      "step": 31219
-    },
-    {
-      "epoch": 5.027778896090825,
-      "grad_norm": 0.004422798287123442,
-      "learning_rate": 0.00019998753473743954,
-      "loss": 46.0,
-      "step": 31220
-    },
-    {
-      "epoch": 5.027939933169613,
-      "grad_norm": 0.0024494112003594637,
-      "learning_rate": 0.00019998753393864408,
-      "loss": 46.0,
-      "step": 31221
-    },
-    {
-      "epoch": 5.0281009702484,
-      "grad_norm": 0.002499592024832964,
-      "learning_rate": 0.000199987533139823,
-      "loss": 46.0,
-      "step": 31222
-    },
-    {
-      "epoch": 5.028262007327187,
-      "grad_norm": 0.017103072255849838,
-      "learning_rate": 0.0001999875323409764,
-      "loss": 46.0,
-      "step": 31223
-    },
-    {
-      "epoch": 5.028423044405974,
-      "grad_norm": 0.0033480413258075714,
-      "learning_rate": 0.00019998753154210414,
-      "loss": 46.0,
-      "step": 31224
-    },
-    {
-      "epoch": 5.0285840814847615,
-      "grad_norm": 0.012788372114300728,
-      "learning_rate": 0.0001999875307432063,
-      "loss": 46.0,
-      "step": 31225
-    },
-    {
-      "epoch": 5.028745118563549,
-      "grad_norm": 0.001563073368743062,
-      "learning_rate": 0.00019998752994428287,
-      "loss": 46.0,
-      "step": 31226
-    },
-    {
-      "epoch": 5.0289061556423365,
-      "grad_norm": 0.003520380472764373,
-      "learning_rate": 0.00019998752914533385,
-      "loss": 46.0,
-      "step": 31227
-    },
-    {
-      "epoch": 5.029067192721124,
-      "grad_norm": 0.0013068087864667177,
-      "learning_rate": 0.00019998752834635925,
-      "loss": 46.0,
-      "step": 31228
-    },
-    {
-      "epoch": 5.029228229799911,
-      "grad_norm": 0.0033790695015341043,
-      "learning_rate": 0.00019998752754735906,
-      "loss": 46.0,
-      "step": 31229
-    },
-    {
-      "epoch": 5.029389266878699,
-      "grad_norm": 0.008708048611879349,
-      "learning_rate": 0.00019998752674833323,
-      "loss": 46.0,
-      "step": 31230
-    },
-    {
-      "epoch": 5.029550303957486,
-      "grad_norm": 0.006931872572749853,
-      "learning_rate": 0.00019998752594928187,
-      "loss": 46.0,
-      "step": 31231
-    },
-    {
-      "epoch": 5.029711341036274,
-      "grad_norm": 0.0027630438562482595,
-      "learning_rate": 0.0001999875251502049,
-      "loss": 46.0,
-      "step": 31232
-    },
-    {
-      "epoch": 5.029872378115061,
-      "grad_norm": 0.009353955276310444,
-      "learning_rate": 0.00019998752435110233,
-      "loss": 46.0,
-      "step": 31233
-    },
-    {
-      "epoch": 5.030033415193849,
-      "grad_norm": 0.0018961328314617276,
-      "learning_rate": 0.00019998752355197418,
-      "loss": 46.0,
-      "step": 31234
-    },
-    {
-      "epoch": 5.030194452272636,
-      "grad_norm": 0.005243843421339989,
-      "learning_rate": 0.0001999875227528204,
-      "loss": 46.0,
-      "step": 31235
-    },
-    {
-      "epoch": 5.0303554893514235,
-      "grad_norm": 0.005383746698498726,
-      "learning_rate": 0.00019998752195364108,
-      "loss": 46.0,
-      "step": 31236
-    },
-    {
-      "epoch": 5.030516526430211,
-      "grad_norm": 0.0013867057859897614,
-      "learning_rate": 0.00019998752115443614,
-      "loss": 46.0,
-      "step": 31237
-    },
-    {
-      "epoch": 5.0306775635089975,
-      "grad_norm": 0.0018150400137528777,
-      "learning_rate": 0.00019998752035520561,
-      "loss": 46.0,
-      "step": 31238
-    },
-    {
-      "epoch": 5.030838600587785,
-      "grad_norm": 0.0033124228939414024,
-      "learning_rate": 0.00019998751955594947,
-      "loss": 46.0,
-      "step": 31239
-    },
-    {
-      "epoch": 5.030999637666572,
-      "grad_norm": 0.00721521582454443,
-      "learning_rate": 0.00019998751875666777,
-      "loss": 46.0,
-      "step": 31240
-    },
-    {
-      "epoch": 5.03116067474536,
-      "grad_norm": 0.007003186736255884,
-      "learning_rate": 0.00019998751795736048,
-      "loss": 46.0,
-      "step": 31241
-    },
-    {
-      "epoch": 5.031321711824147,
-      "grad_norm": 0.0025241137482225895,
-      "learning_rate": 0.0001999875171580276,
-      "loss": 46.0,
-      "step": 31242
-    },
-    {
-      "epoch": 5.031482748902935,
-      "grad_norm": 0.001630175393074751,
-      "learning_rate": 0.00019998751635866911,
-      "loss": 46.0,
-      "step": 31243
-    },
-    {
-      "epoch": 5.031643785981722,
-      "grad_norm": 0.0049359663389623165,
-      "learning_rate": 0.00019998751555928504,
-      "loss": 46.0,
-      "step": 31244
-    },
-    {
-      "epoch": 5.03180482306051,
-      "grad_norm": 0.009411569684743881,
-      "learning_rate": 0.00019998751475987537,
-      "loss": 46.0,
-      "step": 31245
-    },
-    {
-      "epoch": 5.031965860139297,
-      "grad_norm": 0.001628089346922934,
-      "learning_rate": 0.00019998751396044012,
-      "loss": 46.0,
-      "step": 31246
-    },
-    {
-      "epoch": 5.032126897218085,
-      "grad_norm": 0.002978523727506399,
-      "learning_rate": 0.00019998751316097925,
-      "loss": 46.0,
-      "step": 31247
-    },
-    {
-      "epoch": 5.032287934296872,
-      "grad_norm": 0.002056524157524109,
-      "learning_rate": 0.0001999875123614928,
-      "loss": 46.0,
-      "step": 31248
-    },
-    {
-      "epoch": 5.0324489713756595,
-      "grad_norm": 0.0015227656112983823,
-      "learning_rate": 0.00019998751156198079,
-      "loss": 46.0,
-      "step": 31249
-    },
-    {
-      "epoch": 5.032610008454447,
-      "grad_norm": 0.02733725495636463,
-      "learning_rate": 0.00019998751076244316,
-      "loss": 46.0,
-      "step": 31250
-    },
-    {
-      "epoch": 5.032771045533234,
-      "grad_norm": 0.003301442600786686,
-      "learning_rate": 0.00019998750996287994,
-      "loss": 46.0,
-      "step": 31251
-    },
-    {
-      "epoch": 5.032932082612022,
-      "grad_norm": 0.009448815137147903,
-      "learning_rate": 0.00019998750916329114,
-      "loss": 46.0,
-      "step": 31252
-    },
-    {
-      "epoch": 5.033093119690808,
-      "grad_norm": 0.006169270724058151,
-      "learning_rate": 0.00019998750836367672,
-      "loss": 46.0,
-      "step": 31253
-    },
-    {
-      "epoch": 5.033254156769596,
-      "grad_norm": 0.003445628099143505,
-      "learning_rate": 0.00019998750756403672,
-      "loss": 46.0,
-      "step": 31254
-    },
-    {
-      "epoch": 5.033415193848383,
-      "grad_norm": 0.003946996293962002,
-      "learning_rate": 0.00019998750676437116,
-      "loss": 46.0,
-      "step": 31255
-    },
-    {
-      "epoch": 5.033576230927171,
-      "grad_norm": 0.0019527708645910025,
-      "learning_rate": 0.00019998750596467998,
-      "loss": 46.0,
-      "step": 31256
-    },
-    {
-      "epoch": 5.033737268005958,
-      "grad_norm": 0.0045500583946704865,
-      "learning_rate": 0.0001999875051649632,
-      "loss": 46.0,
-      "step": 31257
-    },
-    {
-      "epoch": 5.033898305084746,
-      "grad_norm": 0.0023381556384265423,
-      "learning_rate": 0.00019998750436522083,
-      "loss": 46.0,
-      "step": 31258
-    },
-    {
-      "epoch": 5.034059342163533,
-      "grad_norm": 0.00200643646530807,
-      "learning_rate": 0.0001999875035654529,
-      "loss": 46.0,
-      "step": 31259
-    },
-    {
-      "epoch": 5.034220379242321,
-      "grad_norm": 0.011275083757936954,
-      "learning_rate": 0.00019998750276565934,
-      "loss": 46.0,
-      "step": 31260
-    },
-    {
-      "epoch": 5.034381416321108,
-      "grad_norm": 0.01734207756817341,
-      "learning_rate": 0.00019998750196584023,
-      "loss": 46.0,
-      "step": 31261
-    },
-    {
-      "epoch": 5.0345424533998955,
-      "grad_norm": 0.013983099721372128,
-      "learning_rate": 0.0001999875011659955,
-      "loss": 46.0,
-      "step": 31262
-    },
-    {
-      "epoch": 5.034703490478683,
-      "grad_norm": 0.01058198045939207,
-      "learning_rate": 0.00019998750036612516,
-      "loss": 46.0,
-      "step": 31263
-    },
-    {
-      "epoch": 5.03486452755747,
-      "grad_norm": 0.0030954291578382254,
-      "learning_rate": 0.00019998749956622928,
-      "loss": 46.0,
-      "step": 31264
-    },
-    {
-      "epoch": 5.035025564636258,
-      "grad_norm": 0.0050180512480437756,
-      "learning_rate": 0.0001999874987663078,
-      "loss": 46.0,
-      "step": 31265
-    },
-    {
-      "epoch": 5.035186601715045,
-      "grad_norm": 0.00824425183236599,
-      "learning_rate": 0.0001999874979663607,
-      "loss": 46.0,
-      "step": 31266
-    },
-    {
-      "epoch": 5.035347638793832,
-      "grad_norm": 0.009842642582952976,
-      "learning_rate": 0.000199987497166388,
-      "loss": 46.0,
-      "step": 31267
-    },
-    {
-      "epoch": 5.035508675872619,
-      "grad_norm": 0.0010818196460604668,
-      "learning_rate": 0.00019998749636638972,
-      "loss": 46.0,
-      "step": 31268
-    },
-    {
-      "epoch": 5.035669712951407,
-      "grad_norm": 0.0069105722941458225,
-      "learning_rate": 0.00019998749556636585,
-      "loss": 46.0,
-      "step": 31269
-    },
-    {
-      "epoch": 5.035830750030194,
-      "grad_norm": 0.003831706941127777,
-      "learning_rate": 0.0001999874947663164,
-      "loss": 46.0,
-      "step": 31270
-    },
-    {
-      "epoch": 5.035991787108982,
-      "grad_norm": 0.009741471149027348,
-      "learning_rate": 0.00019998749396624136,
-      "loss": 46.0,
-      "step": 31271
-    },
-    {
-      "epoch": 5.036152824187769,
-      "grad_norm": 0.01142609678208828,
-      "learning_rate": 0.0001999874931661407,
-      "loss": 46.0,
-      "step": 31272
-    },
-    {
-      "epoch": 5.036313861266557,
-      "grad_norm": 0.0040459115989506245,
-      "learning_rate": 0.00019998749236601446,
-      "loss": 46.0,
-      "step": 31273
-    },
-    {
-      "epoch": 5.036474898345344,
-      "grad_norm": 0.004744853358715773,
-      "learning_rate": 0.00019998749156586266,
-      "loss": 46.0,
-      "step": 31274
-    },
-    {
-      "epoch": 5.0366359354241315,
-      "grad_norm": 0.0057984162122011185,
-      "learning_rate": 0.00019998749076568522,
-      "loss": 46.0,
-      "step": 31275
-    },
-    {
-      "epoch": 5.036796972502919,
-      "grad_norm": 0.006191633641719818,
-      "learning_rate": 0.0001999874899654822,
-      "loss": 46.0,
-      "step": 31276
-    },
-    {
-      "epoch": 5.036958009581706,
-      "grad_norm": 0.003779152175411582,
-      "learning_rate": 0.0001999874891652536,
-      "loss": 46.0,
-      "step": 31277
-    },
-    {
-      "epoch": 5.037119046660494,
-      "grad_norm": 0.0015368192689493299,
-      "learning_rate": 0.00019998748836499942,
-      "loss": 46.0,
-      "step": 31278
-    },
-    {
-      "epoch": 5.037280083739281,
-      "grad_norm": 0.007674081716686487,
-      "learning_rate": 0.00019998748756471963,
-      "loss": 46.0,
-      "step": 31279
-    },
-    {
-      "epoch": 5.037441120818069,
-      "grad_norm": 0.0035461331717669964,
-      "learning_rate": 0.00019998748676441425,
-      "loss": 46.0,
-      "step": 31280
-    },
-    {
-      "epoch": 5.037602157896856,
-      "grad_norm": 0.0025449793320149183,
-      "learning_rate": 0.0001999874859640833,
-      "loss": 46.0,
-      "step": 31281
-    },
-    {
-      "epoch": 5.037763194975643,
-      "grad_norm": 0.006803569383919239,
-      "learning_rate": 0.00019998748516372673,
-      "loss": 46.0,
-      "step": 31282
-    },
-    {
-      "epoch": 5.03792423205443,
-      "grad_norm": 0.0013075171737000346,
-      "learning_rate": 0.00019998748436334457,
-      "loss": 46.0,
-      "step": 31283
-    },
-    {
-      "epoch": 5.038085269133218,
-      "grad_norm": 0.0029097353108227253,
-      "learning_rate": 0.00019998748356293684,
-      "loss": 46.0,
-      "step": 31284
-    },
-    {
-      "epoch": 5.038246306212005,
-      "grad_norm": 0.007517886348068714,
-      "learning_rate": 0.0001999874827625035,
-      "loss": 46.0,
-      "step": 31285
-    },
-    {
-      "epoch": 5.0384073432907925,
-      "grad_norm": 0.002215017331764102,
-      "learning_rate": 0.0001999874819620446,
-      "loss": 46.0,
-      "step": 31286
-    },
-    {
-      "epoch": 5.03856838036958,
-      "grad_norm": 0.004603625275194645,
-      "learning_rate": 0.00019998748116156008,
-      "loss": 46.0,
-      "step": 31287
-    },
-    {
-      "epoch": 5.038729417448367,
-      "grad_norm": 0.007704117335379124,
-      "learning_rate": 0.00019998748036104995,
-      "loss": 46.0,
-      "step": 31288
-    },
-    {
-      "epoch": 5.038890454527155,
-      "grad_norm": 0.009899219498038292,
-      "learning_rate": 0.00019998747956051426,
-      "loss": 46.0,
-      "step": 31289
-    },
-    {
-      "epoch": 5.039051491605942,
-      "grad_norm": 0.0007270171772688627,
-      "learning_rate": 0.00019998747875995295,
-      "loss": 46.0,
-      "step": 31290
-    },
-    {
-      "epoch": 5.03921252868473,
-      "grad_norm": 0.0006603830261155963,
-      "learning_rate": 0.00019998747795936606,
-      "loss": 46.0,
-      "step": 31291
-    },
-    {
-      "epoch": 5.039373565763517,
-      "grad_norm": 0.003224874846637249,
-      "learning_rate": 0.00019998747715875358,
-      "loss": 46.0,
-      "step": 31292
-    },
-    {
-      "epoch": 5.039534602842305,
-      "grad_norm": 0.02393312193453312,
-      "learning_rate": 0.00019998747635811555,
-      "loss": 46.0,
-      "step": 31293
-    },
-    {
-      "epoch": 5.039695639921092,
-      "grad_norm": 0.0041695875115692616,
-      "learning_rate": 0.00019998747555745187,
-      "loss": 46.0,
-      "step": 31294
-    },
-    {
-      "epoch": 5.03985667699988,
-      "grad_norm": 0.008520207367837429,
-      "learning_rate": 0.00019998747475676262,
-      "loss": 46.0,
-      "step": 31295
-    },
-    {
-      "epoch": 5.040017714078667,
-      "grad_norm": 0.015758665278553963,
-      "learning_rate": 0.0001999874739560478,
-      "loss": 46.0,
-      "step": 31296
-    },
-    {
-      "epoch": 5.040178751157454,
-      "grad_norm": 0.004376154858618975,
-      "learning_rate": 0.00019998747315530736,
-      "loss": 46.0,
-      "step": 31297
-    },
-    {
-      "epoch": 5.040339788236241,
-      "grad_norm": 0.007451072335243225,
-      "learning_rate": 0.00019998747235454133,
-      "loss": 46.0,
-      "step": 31298
-    },
-    {
-      "epoch": 5.0405008253150285,
-      "grad_norm": 0.005293107591569424,
-      "learning_rate": 0.0001999874715537497,
-      "loss": 46.0,
-      "step": 31299
-    },
-    {
-      "epoch": 5.040661862393816,
-      "grad_norm": 0.00759174395352602,
-      "learning_rate": 0.00019998747075293248,
-      "loss": 46.0,
-      "step": 31300
-    },
-    {
-      "epoch": 5.040822899472603,
-      "grad_norm": 0.0014590921346098185,
-      "learning_rate": 0.0001999874699520897,
-      "loss": 46.0,
-      "step": 31301
-    },
-    {
-      "epoch": 5.040983936551391,
-      "grad_norm": 0.003641568124294281,
-      "learning_rate": 0.0001999874691512213,
-      "loss": 46.0,
-      "step": 31302
-    },
-    {
-      "epoch": 5.041144973630178,
-      "grad_norm": 0.015508061274886131,
-      "learning_rate": 0.00019998746835032735,
-      "loss": 46.0,
-      "step": 31303
-    },
-    {
-      "epoch": 5.041306010708966,
-      "grad_norm": 0.0020176840480417013,
-      "learning_rate": 0.00019998746754940774,
-      "loss": 46.0,
-      "step": 31304
-    },
-    {
-      "epoch": 5.041467047787753,
-      "grad_norm": 0.003791556926444173,
-      "learning_rate": 0.00019998746674846257,
-      "loss": 46.0,
-      "step": 31305
-    },
-    {
-      "epoch": 5.041628084866541,
-      "grad_norm": 0.007079689763486385,
-      "learning_rate": 0.00019998746594749182,
-      "loss": 46.0,
-      "step": 31306
-    },
-    {
-      "epoch": 5.041789121945328,
-      "grad_norm": 0.003946996293962002,
-      "learning_rate": 0.00019998746514649548,
-      "loss": 46.0,
-      "step": 31307
-    },
-    {
-      "epoch": 5.041950159024116,
-      "grad_norm": 0.005046943202614784,
-      "learning_rate": 0.00019998746434547352,
-      "loss": 46.0,
-      "step": 31308
-    },
-    {
-      "epoch": 5.042111196102903,
-      "grad_norm": 0.0015233828453347087,
-      "learning_rate": 0.00019998746354442598,
-      "loss": 46.0,
-      "step": 31309
-    },
-    {
-      "epoch": 5.0422722331816905,
-      "grad_norm": 0.00573785649612546,
-      "learning_rate": 0.00019998746274335288,
-      "loss": 46.0,
-      "step": 31310
-    },
-    {
-      "epoch": 5.042433270260477,
-      "grad_norm": 0.0013896606396883726,
-      "learning_rate": 0.00019998746194225414,
-      "loss": 46.0,
-      "step": 31311
-    },
-    {
-      "epoch": 5.0425943073392645,
-      "grad_norm": 0.003102433867752552,
-      "learning_rate": 0.00019998746114112983,
-      "loss": 46.0,
-      "step": 31312
-    },
-    {
-      "epoch": 5.042755344418052,
-      "grad_norm": 0.0025348858907818794,
-      "learning_rate": 0.00019998746033997992,
-      "loss": 46.0,
-      "step": 31313
-    },
-    {
-      "epoch": 5.042916381496839,
-      "grad_norm": 0.0022277783136814833,
-      "learning_rate": 0.00019998745953880444,
-      "loss": 46.0,
-      "step": 31314
-    },
-    {
-      "epoch": 5.043077418575627,
-      "grad_norm": 0.008986001834273338,
-      "learning_rate": 0.00019998745873760337,
-      "loss": 46.0,
-      "step": 31315
-    },
-    {
-      "epoch": 5.043238455654414,
-      "grad_norm": 0.008521556854248047,
-      "learning_rate": 0.0001999874579363767,
-      "loss": 46.0,
-      "step": 31316
-    },
-    {
-      "epoch": 5.043399492733202,
-      "grad_norm": 0.002612863900139928,
-      "learning_rate": 0.00019998745713512443,
-      "loss": 46.0,
-      "step": 31317
-    },
-    {
-      "epoch": 5.043560529811989,
-      "grad_norm": 0.008150513283908367,
-      "learning_rate": 0.00019998745633384657,
-      "loss": 46.0,
-      "step": 31318
-    },
-    {
-      "epoch": 5.043721566890777,
-      "grad_norm": 0.004577559418976307,
-      "learning_rate": 0.00019998745553254313,
-      "loss": 46.0,
-      "step": 31319
-    },
-    {
-      "epoch": 5.043882603969564,
-      "grad_norm": 0.0018382498528808355,
-      "learning_rate": 0.00019998745473121408,
-      "loss": 46.0,
-      "step": 31320
-    },
-    {
-      "epoch": 5.044043641048352,
-      "grad_norm": 0.0040992083959281445,
-      "learning_rate": 0.00019998745392985943,
-      "loss": 46.0,
-      "step": 31321
-    },
-    {
-      "epoch": 5.044204678127139,
-      "grad_norm": 0.004502686206251383,
-      "learning_rate": 0.0001999874531284792,
-      "loss": 46.0,
-      "step": 31322
-    },
-    {
-      "epoch": 5.0443657152059265,
-      "grad_norm": 0.015286893583834171,
-      "learning_rate": 0.00019998745232707342,
-      "loss": 46.0,
-      "step": 31323
-    },
-    {
-      "epoch": 5.044526752284714,
-      "grad_norm": 0.0019343370804563165,
-      "learning_rate": 0.000199987451525642,
-      "loss": 46.0,
-      "step": 31324
-    },
-    {
-      "epoch": 5.044687789363501,
-      "grad_norm": 0.004592434968799353,
-      "learning_rate": 0.000199987450724185,
-      "loss": 46.0,
-      "step": 31325
-    },
-    {
-      "epoch": 5.044848826442288,
-      "grad_norm": 0.007957251742482185,
-      "learning_rate": 0.0001999874499227024,
-      "loss": 46.0,
-      "step": 31326
-    },
-    {
-      "epoch": 5.045009863521075,
-      "grad_norm": 0.0012921326560899615,
-      "learning_rate": 0.0001999874491211942,
-      "loss": 46.0,
-      "step": 31327
-    },
-    {
-      "epoch": 5.045170900599863,
-      "grad_norm": 0.007931535132229328,
-      "learning_rate": 0.00019998744831966044,
-      "loss": 46.0,
-      "step": 31328
-    },
-    {
-      "epoch": 5.04533193767865,
-      "grad_norm": 0.002153608947992325,
-      "learning_rate": 0.00019998744751810108,
-      "loss": 46.0,
-      "step": 31329
-    },
-    {
-      "epoch": 5.045492974757438,
-      "grad_norm": 0.00686339195817709,
-      "learning_rate": 0.0001999874467165161,
-      "loss": 46.0,
-      "step": 31330
-    },
-    {
-      "epoch": 5.045654011836225,
-      "grad_norm": 0.001382661983370781,
-      "learning_rate": 0.00019998744591490555,
-      "loss": 46.0,
-      "step": 31331
-    },
-    {
-      "epoch": 5.045815048915013,
-      "grad_norm": 0.005815770942717791,
-      "learning_rate": 0.0001999874451132694,
-      "loss": 46.0,
-      "step": 31332
-    },
-    {
-      "epoch": 5.0459760859938,
-      "grad_norm": 0.006721396930515766,
-      "learning_rate": 0.0001999874443116077,
-      "loss": 46.0,
-      "step": 31333
-    },
-    {
-      "epoch": 5.0461371230725875,
-      "grad_norm": 0.004813921172171831,
-      "learning_rate": 0.00019998744350992035,
-      "loss": 46.0,
-      "step": 31334
-    },
-    {
-      "epoch": 5.046298160151375,
-      "grad_norm": 0.003513301257044077,
-      "learning_rate": 0.00019998744270820744,
-      "loss": 46.0,
-      "step": 31335
-    },
-    {
-      "epoch": 5.0464591972301625,
-      "grad_norm": 0.017438331618905067,
-      "learning_rate": 0.00019998744190646893,
-      "loss": 46.0,
-      "step": 31336
-    },
-    {
-      "epoch": 5.04662023430895,
-      "grad_norm": 0.00219210353679955,
-      "learning_rate": 0.0001999874411047048,
-      "loss": 46.0,
-      "step": 31337
-    },
-    {
-      "epoch": 5.046781271387737,
-      "grad_norm": 0.0018554676789790392,
-      "learning_rate": 0.00019998744030291513,
-      "loss": 46.0,
-      "step": 31338
-    },
-    {
-      "epoch": 5.046942308466525,
-      "grad_norm": 0.0037925350479781628,
-      "learning_rate": 0.00019998743950109987,
-      "loss": 46.0,
-      "step": 31339
-    },
-    {
-      "epoch": 5.047103345545311,
-      "grad_norm": 0.002175368135794997,
-      "learning_rate": 0.000199987438699259,
-      "loss": 46.0,
-      "step": 31340
-    },
-    {
-      "epoch": 5.047264382624099,
-      "grad_norm": 0.007592346053570509,
-      "learning_rate": 0.00019998743789739252,
-      "loss": 46.0,
-      "step": 31341
-    },
-    {
-      "epoch": 5.047425419702886,
-      "grad_norm": 0.0024825402069836855,
-      "learning_rate": 0.00019998743709550044,
-      "loss": 46.0,
-      "step": 31342
-    },
-    {
-      "epoch": 5.047586456781674,
-      "grad_norm": 0.0038324189372360706,
-      "learning_rate": 0.0001999874362935828,
-      "loss": 46.0,
-      "step": 31343
-    },
-    {
-      "epoch": 5.047747493860461,
-      "grad_norm": 0.00716969883069396,
-      "learning_rate": 0.00019998743549163957,
-      "loss": 46.0,
-      "step": 31344
-    },
-    {
-      "epoch": 5.047908530939249,
-      "grad_norm": 0.011669233441352844,
-      "learning_rate": 0.00019998743468967073,
-      "loss": 46.0,
-      "step": 31345
-    },
-    {
-      "epoch": 5.048069568018036,
-      "grad_norm": 0.001614892389625311,
-      "learning_rate": 0.0001999874338876763,
-      "loss": 46.0,
-      "step": 31346
-    },
-    {
-      "epoch": 5.0482306050968235,
-      "grad_norm": 0.005077993031591177,
-      "learning_rate": 0.00019998743308565628,
-      "loss": 46.0,
-      "step": 31347
-    },
-    {
-      "epoch": 5.048391642175611,
-      "grad_norm": 0.0030572321265935898,
-      "learning_rate": 0.00019998743228361068,
-      "loss": 46.0,
-      "step": 31348
-    },
-    {
-      "epoch": 5.048552679254398,
-      "grad_norm": 0.015330816619098186,
-      "learning_rate": 0.00019998743148153948,
-      "loss": 46.0,
-      "step": 31349
-    },
-    {
-      "epoch": 5.048713716333186,
-      "grad_norm": 0.0018201855709776282,
-      "learning_rate": 0.0001999874306794427,
-      "loss": 46.0,
-      "step": 31350
-    },
-    {
-      "epoch": 5.048874753411973,
-      "grad_norm": 0.006428643129765987,
-      "learning_rate": 0.0001999874298773203,
-      "loss": 46.0,
-      "step": 31351
-    },
-    {
-      "epoch": 5.049035790490761,
-      "grad_norm": 0.0064857881516218185,
-      "learning_rate": 0.00019998742907517233,
-      "loss": 46.0,
-      "step": 31352
-    },
-    {
-      "epoch": 5.049196827569548,
-      "grad_norm": 0.006932168733328581,
-      "learning_rate": 0.00019998742827299877,
-      "loss": 46.0,
-      "step": 31353
-    },
-    {
-      "epoch": 5.049357864648336,
-      "grad_norm": 0.007753266952931881,
-      "learning_rate": 0.0001999874274707996,
-      "loss": 46.0,
-      "step": 31354
-    },
-    {
-      "epoch": 5.049518901727122,
-      "grad_norm": 0.011997136287391186,
-      "learning_rate": 0.00019998742666857484,
-      "loss": 46.0,
-      "step": 31355
-    },
-    {
-      "epoch": 5.04967993880591,
-      "grad_norm": 0.012536806985735893,
-      "learning_rate": 0.0001999874258663245,
-      "loss": 46.0,
-      "step": 31356
-    },
-    {
-      "epoch": 5.049840975884697,
-      "grad_norm": 0.004379378166049719,
-      "learning_rate": 0.0001999874250640486,
-      "loss": 46.0,
-      "step": 31357
-    },
-    {
-      "epoch": 5.050002012963485,
-      "grad_norm": 0.0015932349488139153,
-      "learning_rate": 0.00019998742426174707,
-      "loss": 46.0,
-      "step": 31358
-    },
-    {
-      "epoch": 5.050163050042272,
-      "grad_norm": 0.011047838255763054,
-      "learning_rate": 0.00019998742345941995,
-      "loss": 46.0,
-      "step": 31359
-    },
-    {
-      "epoch": 5.0503240871210595,
-      "grad_norm": 0.0020418083295226097,
-      "learning_rate": 0.00019998742265706722,
-      "loss": 46.0,
-      "step": 31360
-    },
-    {
-      "epoch": 5.050485124199847,
-      "grad_norm": 0.0030328778084367514,
-      "learning_rate": 0.00019998742185468893,
-      "loss": 46.0,
-      "step": 31361
-    },
-    {
-      "epoch": 5.050646161278634,
-      "grad_norm": 0.003227280220016837,
-      "learning_rate": 0.00019998742105228505,
-      "loss": 46.0,
-      "step": 31362
-    },
-    {
-      "epoch": 5.050807198357422,
-      "grad_norm": 0.004678688477724791,
-      "learning_rate": 0.00019998742024985555,
-      "loss": 46.0,
-      "step": 31363
-    },
-    {
-      "epoch": 5.050968235436209,
-      "grad_norm": 0.00585603155195713,
-      "learning_rate": 0.00019998741944740047,
-      "loss": 46.0,
-      "step": 31364
-    },
-    {
-      "epoch": 5.051129272514997,
-      "grad_norm": 0.004272001795470715,
-      "learning_rate": 0.0001999874186449198,
-      "loss": 46.0,
-      "step": 31365
-    },
-    {
-      "epoch": 5.051290309593784,
-      "grad_norm": 0.005335711408406496,
-      "learning_rate": 0.00019998741784241358,
-      "loss": 46.0,
-      "step": 31366
-    },
-    {
-      "epoch": 5.051451346672572,
-      "grad_norm": 0.005429176148027182,
-      "learning_rate": 0.0001999874170398817,
-      "loss": 46.0,
-      "step": 31367
-    },
-    {
-      "epoch": 5.051612383751359,
-      "grad_norm": 0.0013133641332387924,
-      "learning_rate": 0.00019998741623732425,
-      "loss": 46.0,
-      "step": 31368
-    },
-    {
-      "epoch": 5.051773420830147,
-      "grad_norm": 0.005599755793809891,
-      "learning_rate": 0.00019998741543474124,
-      "loss": 46.0,
-      "step": 31369
-    },
-    {
-      "epoch": 5.051934457908933,
-      "grad_norm": 0.004211513325572014,
-      "learning_rate": 0.0001999874146321326,
-      "loss": 46.0,
-      "step": 31370
-    },
-    {
-      "epoch": 5.052095494987721,
-      "grad_norm": 0.009187194518744946,
-      "learning_rate": 0.0001999874138294984,
-      "loss": 46.0,
-      "step": 31371
-    },
-    {
-      "epoch": 5.052256532066508,
-      "grad_norm": 0.004940877668559551,
-      "learning_rate": 0.00019998741302683856,
-      "loss": 46.0,
-      "step": 31372
-    },
-    {
-      "epoch": 5.0524175691452955,
-      "grad_norm": 0.012068133801221848,
-      "learning_rate": 0.0001999874122241532,
-      "loss": 46.0,
-      "step": 31373
-    },
-    {
-      "epoch": 5.052578606224083,
-      "grad_norm": 0.006456427741795778,
-      "learning_rate": 0.00019998741142144218,
-      "loss": 46.0,
-      "step": 31374
-    },
-    {
-      "epoch": 5.05273964330287,
-      "grad_norm": 0.004046276211738586,
-      "learning_rate": 0.00019998741061870562,
-      "loss": 46.0,
-      "step": 31375
-    },
-    {
-      "epoch": 5.052900680381658,
-      "grad_norm": 0.0027716588228940964,
-      "learning_rate": 0.0001999874098159434,
-      "loss": 46.0,
-      "step": 31376
-    },
-    {
-      "epoch": 5.053061717460445,
-      "grad_norm": 0.0017947101732715964,
-      "learning_rate": 0.00019998740901315567,
-      "loss": 46.0,
-      "step": 31377
-    },
-    {
-      "epoch": 5.053222754539233,
-      "grad_norm": 0.016897421330213547,
-      "learning_rate": 0.0001999874082103423,
-      "loss": 46.0,
-      "step": 31378
-    },
-    {
-      "epoch": 5.05338379161802,
-      "grad_norm": 0.024554461240768433,
-      "learning_rate": 0.00019998740740750334,
-      "loss": 46.0,
-      "step": 31379
-    },
-    {
-      "epoch": 5.053544828696808,
-      "grad_norm": 0.00404990091919899,
-      "learning_rate": 0.00019998740660463881,
-      "loss": 46.0,
-      "step": 31380
-    },
-    {
-      "epoch": 5.053705865775595,
-      "grad_norm": 0.005295802839100361,
-      "learning_rate": 0.00019998740580174867,
-      "loss": 46.0,
-      "step": 31381
-    },
-    {
-      "epoch": 5.053866902854383,
-      "grad_norm": 0.005360559560358524,
-      "learning_rate": 0.00019998740499883297,
-      "loss": 46.0,
-      "step": 31382
-    },
-    {
-      "epoch": 5.05402793993317,
-      "grad_norm": 0.0006004504975862801,
-      "learning_rate": 0.00019998740419589165,
-      "loss": 46.0,
-      "step": 31383
-    },
-    {
-      "epoch": 5.054188977011957,
-      "grad_norm": 0.015216195955872536,
-      "learning_rate": 0.00019998740339292472,
-      "loss": 46.0,
-      "step": 31384
-    },
-    {
-      "epoch": 5.054350014090744,
-      "grad_norm": 0.00952720083296299,
-      "learning_rate": 0.00019998740258993225,
-      "loss": 46.0,
-      "step": 31385
-    },
-    {
-      "epoch": 5.0545110511695315,
-      "grad_norm": 0.014766951091587543,
-      "learning_rate": 0.00019998740178691415,
-      "loss": 46.0,
-      "step": 31386
-    },
-    {
-      "epoch": 5.054672088248319,
-      "grad_norm": 0.0011706594377756119,
-      "learning_rate": 0.00019998740098387048,
-      "loss": 46.0,
-      "step": 31387
-    },
-    {
-      "epoch": 5.054833125327106,
-      "grad_norm": 0.00246440339833498,
-      "learning_rate": 0.00019998740018080117,
-      "loss": 46.0,
-      "step": 31388
-    },
-    {
-      "epoch": 5.054994162405894,
-      "grad_norm": 0.0016380480956286192,
-      "learning_rate": 0.00019998739937770633,
-      "loss": 46.0,
-      "step": 31389
-    },
-    {
-      "epoch": 5.055155199484681,
-      "grad_norm": 0.014208029955625534,
-      "learning_rate": 0.00019998739857458585,
-      "loss": 46.0,
-      "step": 31390
-    },
-    {
-      "epoch": 5.055316236563469,
-      "grad_norm": 0.004792451858520508,
-      "learning_rate": 0.0001999873977714398,
-      "loss": 46.0,
-      "step": 31391
-    },
-    {
-      "epoch": 5.055477273642256,
-      "grad_norm": 0.002148693660274148,
-      "learning_rate": 0.00019998739696826817,
-      "loss": 46.0,
-      "step": 31392
-    },
-    {
-      "epoch": 5.055638310721044,
-      "grad_norm": 0.0014832004671916366,
-      "learning_rate": 0.00019998739616507093,
-      "loss": 46.0,
-      "step": 31393
-    },
-    {
-      "epoch": 5.055799347799831,
-      "grad_norm": 0.004854957107454538,
-      "learning_rate": 0.00019998739536184813,
-      "loss": 46.0,
-      "step": 31394
-    },
-    {
-      "epoch": 5.0559603848786185,
-      "grad_norm": 0.0009185775998048484,
-      "learning_rate": 0.0001999873945585997,
-      "loss": 46.0,
-      "step": 31395
-    },
-    {
-      "epoch": 5.056121421957406,
-      "grad_norm": 0.0025046789087355137,
-      "learning_rate": 0.0001999873937553257,
-      "loss": 46.0,
-      "step": 31396
-    },
-    {
-      "epoch": 5.056282459036193,
-      "grad_norm": 0.008300804533064365,
-      "learning_rate": 0.00019998739295202608,
-      "loss": 46.0,
-      "step": 31397
-    },
-    {
-      "epoch": 5.056443496114981,
-      "grad_norm": 0.001007849583402276,
-      "learning_rate": 0.0001999873921487009,
-      "loss": 46.0,
-      "step": 31398
-    },
-    {
-      "epoch": 5.0566045331937675,
-      "grad_norm": 0.013126351870596409,
-      "learning_rate": 0.0001999873913453501,
-      "loss": 46.0,
-      "step": 31399
-    },
-    {
-      "epoch": 5.056765570272555,
-      "grad_norm": 0.002243891591206193,
-      "learning_rate": 0.00019998739054197376,
-      "loss": 46.0,
-      "step": 31400
-    },
-    {
-      "epoch": 5.056926607351342,
-      "grad_norm": 0.004190828185528517,
-      "learning_rate": 0.0001999873897385718,
-      "loss": 46.0,
-      "step": 31401
-    },
-    {
-      "epoch": 5.05708764443013,
-      "grad_norm": 0.007364446762949228,
-      "learning_rate": 0.00019998738893514423,
-      "loss": 46.0,
-      "step": 31402
-    },
-    {
-      "epoch": 5.057248681508917,
-      "grad_norm": 0.005762541200965643,
-      "learning_rate": 0.00019998738813169106,
-      "loss": 46.0,
-      "step": 31403
-    },
-    {
-      "epoch": 5.057409718587705,
-      "grad_norm": 0.0010841024341061711,
-      "learning_rate": 0.00019998738732821233,
-      "loss": 46.0,
-      "step": 31404
-    },
-    {
-      "epoch": 5.057570755666492,
-      "grad_norm": 0.008369272574782372,
-      "learning_rate": 0.000199987386524708,
-      "loss": 46.0,
-      "step": 31405
-    },
-    {
-      "epoch": 5.05773179274528,
-      "grad_norm": 0.010041013360023499,
-      "learning_rate": 0.00019998738572117808,
-      "loss": 46.0,
-      "step": 31406
-    },
-    {
-      "epoch": 5.057892829824067,
-      "grad_norm": 0.002257033484056592,
-      "learning_rate": 0.00019998738491762256,
-      "loss": 46.0,
-      "step": 31407
-    },
-    {
-      "epoch": 5.0580538669028545,
-      "grad_norm": 0.001278592273592949,
-      "learning_rate": 0.00019998738411404146,
-      "loss": 46.0,
-      "step": 31408
-    },
-    {
-      "epoch": 5.058214903981642,
-      "grad_norm": 0.002209677826613188,
-      "learning_rate": 0.00019998738331043477,
-      "loss": 46.0,
-      "step": 31409
-    },
-    {
-      "epoch": 5.058375941060429,
-      "grad_norm": 0.005247199907898903,
-      "learning_rate": 0.00019998738250680246,
-      "loss": 46.0,
-      "step": 31410
-    },
-    {
-      "epoch": 5.058536978139217,
-      "grad_norm": 0.0017442961689084768,
-      "learning_rate": 0.00019998738170314456,
-      "loss": 46.0,
-      "step": 31411
-    },
-    {
-      "epoch": 5.058698015218004,
-      "grad_norm": 0.0007388941594399512,
-      "learning_rate": 0.0001999873808994611,
-      "loss": 46.0,
-      "step": 31412
-    },
-    {
-      "epoch": 5.058859052296791,
-      "grad_norm": 0.0036482533905655146,
-      "learning_rate": 0.00019998738009575204,
-      "loss": 46.0,
-      "step": 31413
-    },
-    {
-      "epoch": 5.059020089375578,
-      "grad_norm": 0.00955873727798462,
-      "learning_rate": 0.00019998737929201738,
-      "loss": 46.0,
-      "step": 31414
-    },
-    {
-      "epoch": 5.059181126454366,
-      "grad_norm": 0.0017869167495518923,
-      "learning_rate": 0.00019998737848825714,
-      "loss": 46.0,
-      "step": 31415
-    },
-    {
-      "epoch": 5.059342163533153,
-      "grad_norm": 0.0059241121634840965,
-      "learning_rate": 0.0001999873776844713,
-      "loss": 46.0,
-      "step": 31416
-    },
-    {
-      "epoch": 5.059503200611941,
-      "grad_norm": 0.00184874318074435,
-      "learning_rate": 0.00019998737688065987,
-      "loss": 46.0,
-      "step": 31417
-    },
-    {
-      "epoch": 5.059664237690728,
-      "grad_norm": 0.003958504181355238,
-      "learning_rate": 0.00019998737607682286,
-      "loss": 46.0,
-      "step": 31418
-    },
-    {
-      "epoch": 5.059825274769516,
-      "grad_norm": 0.004700548946857452,
-      "learning_rate": 0.00019998737527296024,
-      "loss": 46.0,
-      "step": 31419
-    },
-    {
-      "epoch": 5.059986311848303,
-      "grad_norm": 0.0021934835240244865,
-      "learning_rate": 0.00019998737446907204,
-      "loss": 46.0,
-      "step": 31420
-    },
-    {
-      "epoch": 5.0601473489270905,
-      "grad_norm": 0.0026741705369204283,
-      "learning_rate": 0.00019998737366515822,
-      "loss": 46.0,
-      "step": 31421
-    },
-    {
-      "epoch": 5.060308386005878,
-      "grad_norm": 0.0029577219393104315,
-      "learning_rate": 0.00019998737286121884,
-      "loss": 46.0,
-      "step": 31422
-    },
-    {
-      "epoch": 5.060469423084665,
-      "grad_norm": 0.004920677747577429,
-      "learning_rate": 0.00019998737205725387,
-      "loss": 46.0,
-      "step": 31423
-    },
-    {
-      "epoch": 5.060630460163453,
-      "grad_norm": 0.010909433476626873,
-      "learning_rate": 0.0001999873712532633,
-      "loss": 46.0,
-      "step": 31424
-    },
-    {
-      "epoch": 5.06079149724224,
-      "grad_norm": 0.018570439890027046,
-      "learning_rate": 0.00019998737044924712,
-      "loss": 46.0,
-      "step": 31425
-    },
-    {
-      "epoch": 5.060952534321028,
-      "grad_norm": 0.014057573862373829,
-      "learning_rate": 0.00019998736964520534,
-      "loss": 46.0,
-      "step": 31426
-    },
-    {
-      "epoch": 5.061113571399815,
-      "grad_norm": 0.003790802089497447,
-      "learning_rate": 0.000199987368841138,
-      "loss": 46.0,
-      "step": 31427
-    },
-    {
-      "epoch": 5.061274608478602,
-      "grad_norm": 0.005988498218357563,
-      "learning_rate": 0.00019998736803704506,
-      "loss": 46.0,
-      "step": 31428
-    },
-    {
-      "epoch": 5.061435645557389,
-      "grad_norm": 0.004277796018868685,
-      "learning_rate": 0.00019998736723292652,
-      "loss": 46.0,
-      "step": 31429
-    },
-    {
-      "epoch": 5.061596682636177,
-      "grad_norm": 0.003905894234776497,
-      "learning_rate": 0.00019998736642878239,
-      "loss": 46.0,
-      "step": 31430
-    },
-    {
-      "epoch": 5.061757719714964,
-      "grad_norm": 0.0018447316251695156,
-      "learning_rate": 0.0001999873656246127,
-      "loss": 46.0,
-      "step": 31431
-    },
-    {
-      "epoch": 5.061918756793752,
-      "grad_norm": 0.002647134242579341,
-      "learning_rate": 0.0001999873648204174,
-      "loss": 46.0,
-      "step": 31432
-    },
-    {
-      "epoch": 5.062079793872539,
-      "grad_norm": 0.004798299167305231,
-      "learning_rate": 0.0001999873640161965,
-      "loss": 46.0,
-      "step": 31433
-    },
-    {
-      "epoch": 5.0622408309513265,
-      "grad_norm": 0.0034769484773278236,
-      "learning_rate": 0.00019998736321194999,
-      "loss": 46.0,
-      "step": 31434
-    },
-    {
-      "epoch": 5.062401868030114,
-      "grad_norm": 0.0018220728961750865,
-      "learning_rate": 0.00019998736240767792,
-      "loss": 46.0,
-      "step": 31435
-    },
-    {
-      "epoch": 5.062562905108901,
-      "grad_norm": 0.006823256146162748,
-      "learning_rate": 0.00019998736160338026,
-      "loss": 46.0,
-      "step": 31436
-    },
-    {
-      "epoch": 5.062723942187689,
-      "grad_norm": 0.0007807258516550064,
-      "learning_rate": 0.00019998736079905697,
-      "loss": 46.0,
-      "step": 31437
-    },
-    {
-      "epoch": 5.062884979266476,
-      "grad_norm": 0.0043825022876262665,
-      "learning_rate": 0.0001999873599947081,
-      "loss": 46.0,
-      "step": 31438
-    },
-    {
-      "epoch": 5.063046016345264,
-      "grad_norm": 0.0011262602638453245,
-      "learning_rate": 0.00019998735919033367,
-      "loss": 46.0,
-      "step": 31439
-    },
-    {
-      "epoch": 5.063207053424051,
-      "grad_norm": 0.006850048899650574,
-      "learning_rate": 0.00019998735838593366,
-      "loss": 46.0,
-      "step": 31440
-    },
-    {
-      "epoch": 5.063368090502839,
-      "grad_norm": 0.004780042450875044,
-      "learning_rate": 0.00019998735758150802,
-      "loss": 46.0,
-      "step": 31441
-    },
-    {
-      "epoch": 5.063529127581626,
-      "grad_norm": 0.002735063899308443,
-      "learning_rate": 0.0001999873567770568,
-      "loss": 46.0,
-      "step": 31442
-    },
-    {
-      "epoch": 5.063690164660413,
-      "grad_norm": 0.0023452742025256157,
-      "learning_rate": 0.00019998735597257997,
-      "loss": 46.0,
-      "step": 31443
-    },
-    {
-      "epoch": 5.0638512017392,
-      "grad_norm": 0.002004046458750963,
-      "learning_rate": 0.00019998735516807756,
-      "loss": 46.0,
-      "step": 31444
-    },
-    {
-      "epoch": 5.064012238817988,
-      "grad_norm": 0.010709072463214397,
-      "learning_rate": 0.00019998735436354957,
-      "loss": 46.0,
-      "step": 31445
-    },
-    {
-      "epoch": 5.064173275896775,
-      "grad_norm": 0.002725010272115469,
-      "learning_rate": 0.00019998735355899596,
-      "loss": 46.0,
-      "step": 31446
-    },
-    {
-      "epoch": 5.0643343129755625,
-      "grad_norm": 0.007613108493387699,
-      "learning_rate": 0.0001999873527544168,
-      "loss": 46.0,
-      "step": 31447
-    },
-    {
-      "epoch": 5.06449535005435,
-      "grad_norm": 0.004275319166481495,
-      "learning_rate": 0.000199987351949812,
-      "loss": 46.0,
-      "step": 31448
-    },
-    {
-      "epoch": 5.064656387133137,
-      "grad_norm": 0.005990230478346348,
-      "learning_rate": 0.00019998735114518167,
-      "loss": 46.0,
-      "step": 31449
-    },
-    {
-      "epoch": 5.064817424211925,
-      "grad_norm": 0.0027404134161770344,
-      "learning_rate": 0.0001999873503405257,
-      "loss": 46.0,
-      "step": 31450
-    },
-    {
-      "epoch": 5.064978461290712,
-      "grad_norm": 0.0028265758883208036,
-      "learning_rate": 0.00019998734953584414,
-      "loss": 46.0,
-      "step": 31451
-    },
-    {
-      "epoch": 5.0651394983695,
-      "grad_norm": 0.012204973958432674,
-      "learning_rate": 0.000199987348731137,
-      "loss": 46.0,
-      "step": 31452
-    },
-    {
-      "epoch": 5.065300535448287,
-      "grad_norm": 0.009151040576398373,
-      "learning_rate": 0.0001999873479264043,
-      "loss": 46.0,
-      "step": 31453
-    },
-    {
-      "epoch": 5.065461572527075,
-      "grad_norm": 0.0008236308349296451,
-      "learning_rate": 0.00019998734712164598,
-      "loss": 46.0,
-      "step": 31454
-    },
-    {
-      "epoch": 5.065622609605862,
-      "grad_norm": 0.016653602942824364,
-      "learning_rate": 0.00019998734631686204,
-      "loss": 46.0,
-      "step": 31455
-    },
-    {
-      "epoch": 5.0657836466846495,
-      "grad_norm": 0.0010466242674738169,
-      "learning_rate": 0.00019998734551205253,
-      "loss": 46.0,
-      "step": 31456
-    },
-    {
-      "epoch": 5.065944683763436,
-      "grad_norm": 0.0054195416159927845,
-      "learning_rate": 0.00019998734470721746,
-      "loss": 46.0,
-      "step": 31457
-    },
-    {
-      "epoch": 5.0661057208422235,
-      "grad_norm": 0.0033081627916544676,
-      "learning_rate": 0.00019998734390235675,
-      "loss": 46.0,
-      "step": 31458
-    },
-    {
-      "epoch": 5.066266757921011,
-      "grad_norm": 0.003239066805690527,
-      "learning_rate": 0.0001999873430974705,
-      "loss": 46.0,
-      "step": 31459
-    },
-    {
-      "epoch": 5.0664277949997985,
-      "grad_norm": 0.0018130347598344088,
-      "learning_rate": 0.0001999873422925586,
-      "loss": 46.0,
-      "step": 31460
-    },
-    {
-      "epoch": 5.066588832078586,
-      "grad_norm": 0.016576655209064484,
-      "learning_rate": 0.00019998734148762113,
-      "loss": 46.0,
-      "step": 31461
-    },
-    {
-      "epoch": 5.066749869157373,
-      "grad_norm": 0.008636985905468464,
-      "learning_rate": 0.0001999873406826581,
-      "loss": 46.0,
-      "step": 31462
-    },
-    {
-      "epoch": 5.066910906236161,
-      "grad_norm": 0.01158988382667303,
-      "learning_rate": 0.00019998733987766943,
-      "loss": 46.0,
-      "step": 31463
-    },
-    {
-      "epoch": 5.067071943314948,
-      "grad_norm": 0.0015201332280412316,
-      "learning_rate": 0.0001999873390726552,
-      "loss": 46.0,
-      "step": 31464
-    },
-    {
-      "epoch": 5.067232980393736,
-      "grad_norm": 0.0038251907099038363,
-      "learning_rate": 0.00019998733826761536,
-      "loss": 46.0,
-      "step": 31465
-    },
-    {
-      "epoch": 5.067394017472523,
-      "grad_norm": 0.0038515462074428797,
-      "learning_rate": 0.00019998733746254995,
-      "loss": 46.0,
-      "step": 31466
-    },
-    {
-      "epoch": 5.067555054551311,
-      "grad_norm": 0.011617777869105339,
-      "learning_rate": 0.00019998733665745896,
-      "loss": 46.0,
-      "step": 31467
-    },
-    {
-      "epoch": 5.067716091630098,
-      "grad_norm": 0.002525769639760256,
-      "learning_rate": 0.00019998733585234232,
-      "loss": 46.0,
-      "step": 31468
-    },
-    {
-      "epoch": 5.0678771287088855,
-      "grad_norm": 0.0031561728101223707,
-      "learning_rate": 0.00019998733504720013,
-      "loss": 46.0,
-      "step": 31469
-    },
-    {
-      "epoch": 5.068038165787673,
-      "grad_norm": 0.0010259057162329555,
-      "learning_rate": 0.00019998733424203232,
-      "loss": 46.0,
-      "step": 31470
-    },
-    {
-      "epoch": 5.06819920286646,
-      "grad_norm": 0.005360717419534922,
-      "learning_rate": 0.00019998733343683895,
-      "loss": 46.0,
-      "step": 31471
-    },
-    {
-      "epoch": 5.068360239945247,
-      "grad_norm": 0.001513156108558178,
-      "learning_rate": 0.00019998733263162,
-      "loss": 46.0,
-      "step": 31472
-    },
-    {
-      "epoch": 5.068521277024034,
-      "grad_norm": 0.014202897436916828,
-      "learning_rate": 0.0001999873318263754,
-      "loss": 46.0,
-      "step": 31473
-    },
-    {
-      "epoch": 5.068682314102822,
-      "grad_norm": 0.0009605882223695517,
-      "learning_rate": 0.00019998733102110527,
-      "loss": 46.0,
-      "step": 31474
-    },
-    {
-      "epoch": 5.068843351181609,
-      "grad_norm": 0.0026066009886562824,
-      "learning_rate": 0.0001999873302158095,
-      "loss": 46.0,
-      "step": 31475
-    },
-    {
-      "epoch": 5.069004388260397,
-      "grad_norm": 0.02905389666557312,
-      "learning_rate": 0.0001999873294104882,
-      "loss": 46.0,
-      "step": 31476
-    },
-    {
-      "epoch": 5.069165425339184,
-      "grad_norm": 0.005645869765430689,
-      "learning_rate": 0.00019998732860514125,
-      "loss": 46.0,
-      "step": 31477
-    },
-    {
-      "epoch": 5.069326462417972,
-      "grad_norm": 0.001611900283023715,
-      "learning_rate": 0.00019998732779976871,
-      "loss": 46.0,
-      "step": 31478
-    },
-    {
-      "epoch": 5.069487499496759,
-      "grad_norm": 0.009771431796252728,
-      "learning_rate": 0.00019998732699437062,
-      "loss": 46.0,
-      "step": 31479
-    },
-    {
-      "epoch": 5.069648536575547,
-      "grad_norm": 0.0006704932893626392,
-      "learning_rate": 0.00019998732618894691,
-      "loss": 46.0,
-      "step": 31480
-    },
-    {
-      "epoch": 5.069809573654334,
-      "grad_norm": 0.004354712553322315,
-      "learning_rate": 0.00019998732538349762,
-      "loss": 46.0,
-      "step": 31481
-    },
-    {
-      "epoch": 5.0699706107331215,
-      "grad_norm": 0.009478788822889328,
-      "learning_rate": 0.00019998732457802274,
-      "loss": 46.0,
-      "step": 31482
-    },
-    {
-      "epoch": 5.070131647811909,
-      "grad_norm": 0.002908652415499091,
-      "learning_rate": 0.00019998732377252224,
-      "loss": 46.0,
-      "step": 31483
-    },
-    {
-      "epoch": 5.070292684890696,
-      "grad_norm": 0.012400351464748383,
-      "learning_rate": 0.00019998732296699619,
-      "loss": 46.0,
-      "step": 31484
-    },
-    {
-      "epoch": 5.070453721969484,
-      "grad_norm": 0.00456674350425601,
-      "learning_rate": 0.00019998732216144452,
-      "loss": 46.0,
-      "step": 31485
-    },
-    {
-      "epoch": 5.07061475904827,
-      "grad_norm": 0.015737170353531837,
-      "learning_rate": 0.00019998732135586726,
-      "loss": 46.0,
-      "step": 31486
-    },
-    {
-      "epoch": 5.070775796127058,
-      "grad_norm": 0.007681137416511774,
-      "learning_rate": 0.00019998732055026441,
-      "loss": 46.0,
-      "step": 31487
-    },
-    {
-      "epoch": 5.070936833205845,
-      "grad_norm": 0.010516487993299961,
-      "learning_rate": 0.00019998731974463598,
-      "loss": 46.0,
-      "step": 31488
-    },
-    {
-      "epoch": 5.071097870284633,
-      "grad_norm": 0.012859782204031944,
-      "learning_rate": 0.00019998731893898194,
-      "loss": 46.0,
-      "step": 31489
-    },
-    {
-      "epoch": 5.07125890736342,
-      "grad_norm": 0.0018989465897902846,
-      "learning_rate": 0.00019998731813330233,
-      "loss": 46.0,
-      "step": 31490
-    },
-    {
-      "epoch": 5.071419944442208,
-      "grad_norm": 0.0024115622509270906,
-      "learning_rate": 0.0001999873173275971,
-      "loss": 46.0,
-      "step": 31491
-    },
-    {
-      "epoch": 5.071580981520995,
-      "grad_norm": 0.008649192750453949,
-      "learning_rate": 0.0001999873165218663,
-      "loss": 46.0,
-      "step": 31492
-    },
-    {
-      "epoch": 5.071742018599783,
-      "grad_norm": 0.0024184526409953833,
-      "learning_rate": 0.0001999873157161099,
-      "loss": 46.0,
-      "step": 31493
-    },
-    {
-      "epoch": 5.07190305567857,
-      "grad_norm": 0.009620298631489277,
-      "learning_rate": 0.00019998731491032793,
-      "loss": 46.0,
-      "step": 31494
-    },
-    {
-      "epoch": 5.0720640927573575,
-      "grad_norm": 0.0030542039312422276,
-      "learning_rate": 0.00019998731410452036,
-      "loss": 46.0,
-      "step": 31495
-    },
-    {
-      "epoch": 5.072225129836145,
-      "grad_norm": 0.001520965714007616,
-      "learning_rate": 0.00019998731329868717,
-      "loss": 46.0,
-      "step": 31496
-    },
-    {
-      "epoch": 5.072386166914932,
-      "grad_norm": 0.0036817335058003664,
-      "learning_rate": 0.00019998731249282843,
-      "loss": 46.0,
-      "step": 31497
-    },
-    {
-      "epoch": 5.07254720399372,
-      "grad_norm": 0.003716722596436739,
-      "learning_rate": 0.00019998731168694404,
-      "loss": 46.0,
-      "step": 31498
-    },
-    {
-      "epoch": 5.072708241072507,
-      "grad_norm": 0.0005912923370487988,
-      "learning_rate": 0.00019998731088103413,
-      "loss": 46.0,
-      "step": 31499
-    },
-    {
-      "epoch": 5.072869278151295,
-      "grad_norm": 0.0007993608014658093,
-      "learning_rate": 0.0001999873100750986,
-      "loss": 46.0,
-      "step": 31500
-    },
-    {
-      "epoch": 5.073030315230081,
-      "grad_norm": 0.011456293985247612,
-      "learning_rate": 0.00019998730926913745,
-      "loss": 46.0,
-      "step": 31501
-    },
-    {
-      "epoch": 5.073191352308869,
-      "grad_norm": 0.0019239726243540645,
-      "learning_rate": 0.00019998730846315071,
-      "loss": 46.0,
-      "step": 31502
-    },
-    {
-      "epoch": 5.073352389387656,
-      "grad_norm": 0.0032419771887362003,
-      "learning_rate": 0.00019998730765713842,
-      "loss": 46.0,
-      "step": 31503
-    },
-    {
-      "epoch": 5.073513426466444,
-      "grad_norm": 0.010542859323322773,
-      "learning_rate": 0.00019998730685110049,
-      "loss": 46.0,
-      "step": 31504
-    },
-    {
-      "epoch": 5.073674463545231,
-      "grad_norm": 0.0032794310245662928,
-      "learning_rate": 0.00019998730604503702,
-      "loss": 46.0,
-      "step": 31505
-    },
-    {
-      "epoch": 5.073835500624019,
-      "grad_norm": 0.006397836841642857,
-      "learning_rate": 0.00019998730523894794,
-      "loss": 46.0,
-      "step": 31506
-    },
-    {
-      "epoch": 5.073996537702806,
-      "grad_norm": 0.01285083219408989,
-      "learning_rate": 0.00019998730443283327,
-      "loss": 46.0,
-      "step": 31507
-    },
-    {
-      "epoch": 5.0741575747815935,
-      "grad_norm": 0.0035771497059613466,
-      "learning_rate": 0.00019998730362669298,
-      "loss": 46.0,
-      "step": 31508
-    },
-    {
-      "epoch": 5.074318611860381,
-      "grad_norm": 0.0007602878613397479,
-      "learning_rate": 0.0001999873028205271,
-      "loss": 46.0,
-      "step": 31509
-    },
-    {
-      "epoch": 5.074479648939168,
-      "grad_norm": 0.010783039033412933,
-      "learning_rate": 0.00019998730201433565,
-      "loss": 46.0,
-      "step": 31510
-    },
-    {
-      "epoch": 5.074640686017956,
-      "grad_norm": 0.005767618305981159,
-      "learning_rate": 0.00019998730120811864,
-      "loss": 46.0,
-      "step": 31511
-    },
-    {
-      "epoch": 5.074801723096743,
-      "grad_norm": 0.006642652675509453,
-      "learning_rate": 0.00019998730040187598,
-      "loss": 46.0,
-      "step": 31512
-    },
-    {
-      "epoch": 5.074962760175531,
-      "grad_norm": 0.005331065505743027,
-      "learning_rate": 0.00019998729959560776,
-      "loss": 46.0,
-      "step": 31513
-    },
-    {
-      "epoch": 5.075123797254318,
-      "grad_norm": 0.005481342785060406,
-      "learning_rate": 0.00019998729878931392,
-      "loss": 46.0,
-      "step": 31514
-    },
-    {
-      "epoch": 5.075284834333106,
-      "grad_norm": 0.007799872197210789,
-      "learning_rate": 0.0001999872979829945,
-      "loss": 46.0,
-      "step": 31515
-    },
-    {
-      "epoch": 5.075445871411892,
-      "grad_norm": 0.014986932277679443,
-      "learning_rate": 0.00019998729717664952,
-      "loss": 46.0,
-      "step": 31516
-    },
-    {
-      "epoch": 5.07560690849068,
-      "grad_norm": 0.0039672283455729485,
-      "learning_rate": 0.00019998729637027893,
-      "loss": 46.0,
-      "step": 31517
-    },
-    {
-      "epoch": 5.075767945569467,
-      "grad_norm": 0.009957185946404934,
-      "learning_rate": 0.00019998729556388274,
-      "loss": 46.0,
-      "step": 31518
-    },
-    {
-      "epoch": 5.0759289826482545,
-      "grad_norm": 0.009040186181664467,
-      "learning_rate": 0.00019998729475746097,
-      "loss": 46.0,
-      "step": 31519
-    },
-    {
-      "epoch": 5.076090019727042,
-      "grad_norm": 0.0041587925516068935,
-      "learning_rate": 0.0001999872939510136,
-      "loss": 46.0,
-      "step": 31520
-    },
-    {
-      "epoch": 5.076251056805829,
-      "grad_norm": 0.003826018190011382,
-      "learning_rate": 0.00019998729314454065,
-      "loss": 46.0,
-      "step": 31521
-    },
-    {
-      "epoch": 5.076412093884617,
-      "grad_norm": 0.0040887692011892796,
-      "learning_rate": 0.0001999872923380421,
-      "loss": 46.0,
-      "step": 31522
-    },
-    {
-      "epoch": 5.076573130963404,
-      "grad_norm": 0.009654448367655277,
-      "learning_rate": 0.00019998729153151794,
-      "loss": 46.0,
-      "step": 31523
-    },
-    {
-      "epoch": 5.076734168042192,
-      "grad_norm": 0.0019868065137416124,
-      "learning_rate": 0.0001999872907249682,
-      "loss": 46.0,
-      "step": 31524
-    },
-    {
-      "epoch": 5.076895205120979,
-      "grad_norm": 0.0022918428294360638,
-      "learning_rate": 0.00019998728991839286,
-      "loss": 46.0,
-      "step": 31525
-    },
-    {
-      "epoch": 5.077056242199767,
-      "grad_norm": 0.014718353748321533,
-      "learning_rate": 0.00019998728911179193,
-      "loss": 46.0,
-      "step": 31526
-    },
-    {
-      "epoch": 5.077217279278554,
-      "grad_norm": 0.004916117526590824,
-      "learning_rate": 0.00019998728830516543,
-      "loss": 46.0,
-      "step": 31527
-    },
-    {
-      "epoch": 5.077378316357342,
-      "grad_norm": 0.0077346679754555225,
-      "learning_rate": 0.00019998728749851335,
-      "loss": 46.0,
-      "step": 31528
-    },
-    {
-      "epoch": 5.077539353436129,
-      "grad_norm": 0.0023475319612771273,
-      "learning_rate": 0.00019998728669183563,
-      "loss": 46.0,
-      "step": 31529
-    },
-    {
-      "epoch": 5.0777003905149165,
-      "grad_norm": 0.017003672197461128,
-      "learning_rate": 0.00019998728588513235,
-      "loss": 46.0,
-      "step": 31530
-    },
-    {
-      "epoch": 5.077861427593703,
-      "grad_norm": 0.0027289206627756357,
-      "learning_rate": 0.00019998728507840348,
-      "loss": 46.0,
-      "step": 31531
-    },
-    {
-      "epoch": 5.0780224646724905,
-      "grad_norm": 0.006898499093949795,
-      "learning_rate": 0.00019998728427164902,
-      "loss": 46.0,
-      "step": 31532
-    },
-    {
-      "epoch": 5.078183501751278,
-      "grad_norm": 0.0038688050117343664,
-      "learning_rate": 0.00019998728346486895,
-      "loss": 46.0,
-      "step": 31533
-    },
-    {
-      "epoch": 5.078344538830065,
-      "grad_norm": 0.0016531539149582386,
-      "learning_rate": 0.0001999872826580633,
-      "loss": 46.0,
-      "step": 31534
-    },
-    {
-      "epoch": 5.078505575908853,
-      "grad_norm": 0.0011903924169018865,
-      "learning_rate": 0.00019998728185123205,
-      "loss": 46.0,
-      "step": 31535
-    },
-    {
-      "epoch": 5.07866661298764,
-      "grad_norm": 0.019421108067035675,
-      "learning_rate": 0.0001999872810443752,
-      "loss": 46.0,
-      "step": 31536
-    },
-    {
-      "epoch": 5.078827650066428,
-      "grad_norm": 0.0008002580143511295,
-      "learning_rate": 0.0001999872802374928,
-      "loss": 46.0,
-      "step": 31537
-    },
-    {
-      "epoch": 5.078988687145215,
-      "grad_norm": 0.01161094382405281,
-      "learning_rate": 0.00019998727943058479,
-      "loss": 46.0,
-      "step": 31538
-    },
-    {
-      "epoch": 5.079149724224003,
-      "grad_norm": 0.0034828488714993,
-      "learning_rate": 0.00019998727862365116,
-      "loss": 46.0,
-      "step": 31539
-    },
-    {
-      "epoch": 5.07931076130279,
-      "grad_norm": 0.002845194423571229,
-      "learning_rate": 0.00019998727781669198,
-      "loss": 46.0,
-      "step": 31540
-    },
-    {
-      "epoch": 5.079471798381578,
-      "grad_norm": 0.004946758504956961,
-      "learning_rate": 0.00019998727700970716,
-      "loss": 46.0,
-      "step": 31541
-    },
-    {
-      "epoch": 5.079632835460365,
-      "grad_norm": 0.0028560252394527197,
-      "learning_rate": 0.00019998727620269678,
-      "loss": 46.0,
-      "step": 31542
-    },
-    {
-      "epoch": 5.0797938725391525,
-      "grad_norm": 0.0023897234350442886,
-      "learning_rate": 0.00019998727539566083,
-      "loss": 46.0,
-      "step": 31543
-    },
-    {
-      "epoch": 5.07995490961794,
-      "grad_norm": 0.0018610517727211118,
-      "learning_rate": 0.00019998727458859925,
-      "loss": 46.0,
-      "step": 31544
-    },
-    {
-      "epoch": 5.0801159466967265,
-      "grad_norm": 0.01645711250603199,
-      "learning_rate": 0.00019998727378151208,
-      "loss": 46.0,
-      "step": 31545
-    },
-    {
-      "epoch": 5.080276983775514,
-      "grad_norm": 0.0027641672641038895,
-      "learning_rate": 0.00019998727297439932,
-      "loss": 46.0,
-      "step": 31546
-    },
-    {
-      "epoch": 5.080438020854301,
-      "grad_norm": 0.002907135523855686,
-      "learning_rate": 0.000199987272167261,
-      "loss": 46.0,
-      "step": 31547
-    },
-    {
-      "epoch": 5.080599057933089,
-      "grad_norm": 0.0017631857190281153,
-      "learning_rate": 0.00019998727136009704,
-      "loss": 46.0,
-      "step": 31548
-    },
-    {
-      "epoch": 5.080760095011876,
-      "grad_norm": 0.007124314084649086,
-      "learning_rate": 0.00019998727055290752,
-      "loss": 46.0,
-      "step": 31549
-    },
-    {
-      "epoch": 5.080921132090664,
-      "grad_norm": 0.002711135894060135,
-      "learning_rate": 0.0001999872697456924,
-      "loss": 46.0,
-      "step": 31550
-    },
-    {
-      "epoch": 5.081082169169451,
-      "grad_norm": 0.009224534034729004,
-      "learning_rate": 0.0001999872689384517,
-      "loss": 46.0,
-      "step": 31551
-    },
-    {
-      "epoch": 5.081243206248239,
-      "grad_norm": 0.004644982051104307,
-      "learning_rate": 0.0001999872681311854,
-      "loss": 46.0,
-      "step": 31552
-    },
-    {
-      "epoch": 5.081404243327026,
-      "grad_norm": 0.007895423099398613,
-      "learning_rate": 0.0001999872673238935,
-      "loss": 46.0,
-      "step": 31553
-    },
-    {
-      "epoch": 5.081565280405814,
-      "grad_norm": 0.004799280781298876,
-      "learning_rate": 0.00019998726651657603,
-      "loss": 46.0,
-      "step": 31554
-    },
-    {
-      "epoch": 5.081726317484601,
-      "grad_norm": 0.002282541710883379,
-      "learning_rate": 0.00019998726570923293,
-      "loss": 46.0,
-      "step": 31555
-    },
-    {
-      "epoch": 5.0818873545633885,
-      "grad_norm": 0.0023004882968962193,
-      "learning_rate": 0.00019998726490186427,
-      "loss": 46.0,
-      "step": 31556
-    },
-    {
-      "epoch": 5.082048391642176,
-      "grad_norm": 0.004254021681845188,
-      "learning_rate": 0.00019998726409447003,
-      "loss": 46.0,
-      "step": 31557
-    },
-    {
-      "epoch": 5.082209428720963,
-      "grad_norm": 0.008733365684747696,
-      "learning_rate": 0.00019998726328705017,
-      "loss": 46.0,
-      "step": 31558
-    },
-    {
-      "epoch": 5.082370465799751,
-      "grad_norm": 0.012953191995620728,
-      "learning_rate": 0.00019998726247960472,
-      "loss": 46.0,
-      "step": 31559
-    },
-    {
-      "epoch": 5.082531502878537,
-      "grad_norm": 0.003261252772063017,
-      "learning_rate": 0.0001999872616721337,
-      "loss": 46.0,
-      "step": 31560
-    },
-    {
-      "epoch": 5.082692539957325,
-      "grad_norm": 0.006477107293903828,
-      "learning_rate": 0.00019998726086463707,
-      "loss": 46.0,
-      "step": 31561
-    },
-    {
-      "epoch": 5.082853577036112,
-      "grad_norm": 0.007697482593357563,
-      "learning_rate": 0.00019998726005711483,
-      "loss": 46.0,
-      "step": 31562
-    },
-    {
-      "epoch": 5.0830146141149,
-      "grad_norm": 0.0009260574588552117,
-      "learning_rate": 0.00019998725924956704,
-      "loss": 46.0,
-      "step": 31563
-    },
-    {
-      "epoch": 5.083175651193687,
-      "grad_norm": 0.010477319359779358,
-      "learning_rate": 0.00019998725844199363,
-      "loss": 46.0,
-      "step": 31564
-    },
-    {
-      "epoch": 5.083336688272475,
-      "grad_norm": 0.011995733715593815,
-      "learning_rate": 0.00019998725763439463,
-      "loss": 46.0,
-      "step": 31565
-    },
-    {
-      "epoch": 5.083497725351262,
-      "grad_norm": 0.002470846753567457,
-      "learning_rate": 0.00019998725682677005,
-      "loss": 46.0,
-      "step": 31566
-    },
-    {
-      "epoch": 5.0836587624300495,
-      "grad_norm": 0.009876557625830173,
-      "learning_rate": 0.00019998725601911988,
-      "loss": 46.0,
-      "step": 31567
-    },
-    {
-      "epoch": 5.083819799508837,
-      "grad_norm": 0.006988706067204475,
-      "learning_rate": 0.00019998725521144412,
-      "loss": 46.0,
-      "step": 31568
-    },
-    {
-      "epoch": 5.0839808365876245,
-      "grad_norm": 0.0031932799611240625,
-      "learning_rate": 0.00019998725440374275,
-      "loss": 46.0,
-      "step": 31569
-    },
-    {
-      "epoch": 5.084141873666412,
-      "grad_norm": 0.002237739274278283,
-      "learning_rate": 0.0001999872535960158,
-      "loss": 46.0,
-      "step": 31570
-    },
-    {
-      "epoch": 5.084302910745199,
-      "grad_norm": 0.0008495231741108,
-      "learning_rate": 0.00019998725278826327,
-      "loss": 46.0,
-      "step": 31571
-    },
-    {
-      "epoch": 5.084463947823987,
-      "grad_norm": 0.007848595269024372,
-      "learning_rate": 0.0001999872519804851,
-      "loss": 46.0,
-      "step": 31572
-    },
-    {
-      "epoch": 5.084624984902774,
-      "grad_norm": 0.009854995645582676,
-      "learning_rate": 0.0001999872511726814,
-      "loss": 46.0,
-      "step": 31573
-    },
-    {
-      "epoch": 5.084786021981561,
-      "grad_norm": 0.011838048696517944,
-      "learning_rate": 0.00019998725036485208,
-      "loss": 46.0,
-      "step": 31574
-    },
-    {
-      "epoch": 5.084947059060348,
-      "grad_norm": 0.0029190005734562874,
-      "learning_rate": 0.00019998724955699716,
-      "loss": 46.0,
-      "step": 31575
-    },
-    {
-      "epoch": 5.085108096139136,
-      "grad_norm": 0.001822870341129601,
-      "learning_rate": 0.00019998724874911668,
-      "loss": 46.0,
-      "step": 31576
-    },
-    {
-      "epoch": 5.085269133217923,
-      "grad_norm": 0.003114506136626005,
-      "learning_rate": 0.0001999872479412106,
-      "loss": 46.0,
-      "step": 31577
-    },
-    {
-      "epoch": 5.085430170296711,
-      "grad_norm": 0.0015350612811744213,
-      "learning_rate": 0.0001999872471332789,
-      "loss": 46.0,
-      "step": 31578
-    },
-    {
-      "epoch": 5.085591207375498,
-      "grad_norm": 0.0020557488314807415,
-      "learning_rate": 0.00019998724632532163,
-      "loss": 46.0,
-      "step": 31579
-    },
-    {
-      "epoch": 5.0857522444542855,
-      "grad_norm": 0.004932016599923372,
-      "learning_rate": 0.00019998724551733877,
-      "loss": 46.0,
-      "step": 31580
-    },
-    {
-      "epoch": 5.085913281533073,
-      "grad_norm": 0.019641203805804253,
-      "learning_rate": 0.0001999872447093303,
-      "loss": 46.0,
-      "step": 31581
-    },
-    {
-      "epoch": 5.08607431861186,
-      "grad_norm": 0.004743436351418495,
-      "learning_rate": 0.00019998724390129624,
-      "loss": 46.0,
-      "step": 31582
-    },
-    {
-      "epoch": 5.086235355690648,
-      "grad_norm": 0.009794383309781551,
-      "learning_rate": 0.00019998724309323662,
-      "loss": 46.0,
-      "step": 31583
-    },
-    {
-      "epoch": 5.086396392769435,
-      "grad_norm": 0.0032204575836658478,
-      "learning_rate": 0.00019998724228515138,
-      "loss": 46.0,
-      "step": 31584
-    },
-    {
-      "epoch": 5.086557429848223,
-      "grad_norm": 0.005569989327341318,
-      "learning_rate": 0.00019998724147704056,
-      "loss": 46.0,
-      "step": 31585
-    },
-    {
-      "epoch": 5.08671846692701,
-      "grad_norm": 0.004238417837768793,
-      "learning_rate": 0.00019998724066890415,
-      "loss": 46.0,
-      "step": 31586
-    },
-    {
-      "epoch": 5.086879504005798,
-      "grad_norm": 0.002847732277587056,
-      "learning_rate": 0.00019998723986074213,
-      "loss": 46.0,
-      "step": 31587
-    },
-    {
-      "epoch": 5.087040541084585,
-      "grad_norm": 0.0033774760086089373,
-      "learning_rate": 0.00019998723905255452,
-      "loss": 46.0,
-      "step": 31588
-    },
-    {
-      "epoch": 5.087201578163372,
-      "grad_norm": 0.010860154405236244,
-      "learning_rate": 0.00019998723824434132,
-      "loss": 46.0,
-      "step": 31589
-    },
-    {
-      "epoch": 5.087362615242159,
-      "grad_norm": 0.002917844569310546,
-      "learning_rate": 0.00019998723743610257,
-      "loss": 46.0,
-      "step": 31590
-    },
-    {
-      "epoch": 5.087523652320947,
-      "grad_norm": 0.0014842735836282372,
-      "learning_rate": 0.00019998723662783817,
-      "loss": 46.0,
-      "step": 31591
-    },
-    {
-      "epoch": 5.087684689399734,
-      "grad_norm": 0.018754929304122925,
-      "learning_rate": 0.0001999872358195482,
-      "loss": 46.0,
-      "step": 31592
-    },
-    {
-      "epoch": 5.0878457264785215,
-      "grad_norm": 0.01046660728752613,
-      "learning_rate": 0.00019998723501123266,
-      "loss": 46.0,
-      "step": 31593
-    },
-    {
-      "epoch": 5.088006763557309,
-      "grad_norm": 0.00049332354683429,
-      "learning_rate": 0.0001999872342028915,
-      "loss": 46.0,
-      "step": 31594
-    },
-    {
-      "epoch": 5.088167800636096,
-      "grad_norm": 0.002214283449575305,
-      "learning_rate": 0.00019998723339452478,
-      "loss": 46.0,
-      "step": 31595
-    },
-    {
-      "epoch": 5.088328837714884,
-      "grad_norm": 0.0033477996475994587,
-      "learning_rate": 0.00019998723258613245,
-      "loss": 46.0,
-      "step": 31596
-    },
-    {
-      "epoch": 5.088489874793671,
-      "grad_norm": 0.013349231332540512,
-      "learning_rate": 0.00019998723177771453,
-      "loss": 46.0,
-      "step": 31597
-    },
-    {
-      "epoch": 5.088650911872459,
-      "grad_norm": 0.0033220131881535053,
-      "learning_rate": 0.00019998723096927102,
-      "loss": 46.0,
-      "step": 31598
-    },
-    {
-      "epoch": 5.088811948951246,
-      "grad_norm": 0.0012966557405889034,
-      "learning_rate": 0.0001999872301608019,
-      "loss": 46.0,
-      "step": 31599
-    },
-    {
-      "epoch": 5.088972986030034,
-      "grad_norm": 0.0010098421480506659,
-      "learning_rate": 0.00019998722935230721,
-      "loss": 46.0,
-      "step": 31600
-    },
-    {
-      "epoch": 5.089134023108821,
-      "grad_norm": 0.003335656365379691,
-      "learning_rate": 0.00019998722854378692,
-      "loss": 46.0,
-      "step": 31601
-    },
-    {
-      "epoch": 5.089295060187609,
-      "grad_norm": 0.006485785357654095,
-      "learning_rate": 0.00019998722773524103,
-      "loss": 46.0,
-      "step": 31602
-    },
-    {
-      "epoch": 5.089456097266396,
-      "grad_norm": 0.006169487722218037,
-      "learning_rate": 0.00019998722692666956,
-      "loss": 46.0,
-      "step": 31603
-    },
-    {
-      "epoch": 5.089617134345183,
-      "grad_norm": 0.004332922399044037,
-      "learning_rate": 0.0001999872261180725,
-      "loss": 46.0,
-      "step": 31604
-    },
-    {
-      "epoch": 5.08977817142397,
-      "grad_norm": 0.008071127347648144,
-      "learning_rate": 0.00019998722530944983,
-      "loss": 46.0,
-      "step": 31605
-    },
-    {
-      "epoch": 5.0899392085027575,
-      "grad_norm": 0.008919227868318558,
-      "learning_rate": 0.0001999872245008016,
-      "loss": 46.0,
-      "step": 31606
-    },
-    {
-      "epoch": 5.090100245581545,
-      "grad_norm": 0.002631570678204298,
-      "learning_rate": 0.00019998722369212775,
-      "loss": 46.0,
-      "step": 31607
-    },
-    {
-      "epoch": 5.090261282660332,
-      "grad_norm": 0.004740362521260977,
-      "learning_rate": 0.00019998722288342832,
-      "loss": 46.0,
-      "step": 31608
-    },
-    {
-      "epoch": 5.09042231973912,
-      "grad_norm": 0.001849306165240705,
-      "learning_rate": 0.0001999872220747033,
-      "loss": 46.0,
-      "step": 31609
-    },
-    {
-      "epoch": 5.090583356817907,
-      "grad_norm": 0.0038777433801442385,
-      "learning_rate": 0.00019998722126595268,
-      "loss": 46.0,
-      "step": 31610
-    },
-    {
-      "epoch": 5.090744393896695,
-      "grad_norm": 0.002898280043154955,
-      "learning_rate": 0.0001999872204571765,
-      "loss": 46.0,
-      "step": 31611
-    },
-    {
-      "epoch": 5.090905430975482,
-      "grad_norm": 0.010929821990430355,
-      "learning_rate": 0.00019998721964837468,
-      "loss": 46.0,
-      "step": 31612
-    },
-    {
-      "epoch": 5.09106646805427,
-      "grad_norm": 0.001335539622232318,
-      "learning_rate": 0.0001999872188395473,
-      "loss": 46.0,
-      "step": 31613
-    },
-    {
-      "epoch": 5.091227505133057,
-      "grad_norm": 0.008599543944001198,
-      "learning_rate": 0.00019998721803069432,
-      "loss": 46.0,
-      "step": 31614
-    },
-    {
-      "epoch": 5.091388542211845,
-      "grad_norm": 0.0010156950447708368,
-      "learning_rate": 0.00019998721722181575,
-      "loss": 46.0,
-      "step": 31615
-    },
-    {
-      "epoch": 5.091549579290632,
-      "grad_norm": 0.0017077663214877248,
-      "learning_rate": 0.00019998721641291157,
-      "loss": 46.0,
-      "step": 31616
-    },
-    {
-      "epoch": 5.0917106163694195,
-      "grad_norm": 0.003509596921503544,
-      "learning_rate": 0.00019998721560398182,
-      "loss": 46.0,
-      "step": 31617
-    },
-    {
-      "epoch": 5.091871653448206,
-      "grad_norm": 0.01000924501568079,
-      "learning_rate": 0.00019998721479502646,
-      "loss": 46.0,
-      "step": 31618
-    },
-    {
-      "epoch": 5.0920326905269935,
-      "grad_norm": 0.0036562189925462008,
-      "learning_rate": 0.0001999872139860455,
-      "loss": 46.0,
-      "step": 31619
-    },
-    {
-      "epoch": 5.092193727605781,
-      "grad_norm": 0.0006999990437179804,
-      "learning_rate": 0.000199987213177039,
-      "loss": 46.0,
-      "step": 31620
-    },
-    {
-      "epoch": 5.092354764684568,
-      "grad_norm": 0.007833797484636307,
-      "learning_rate": 0.00019998721236800688,
-      "loss": 46.0,
-      "step": 31621
-    },
-    {
-      "epoch": 5.092515801763356,
-      "grad_norm": 0.0033367800060659647,
-      "learning_rate": 0.00019998721155894917,
-      "loss": 46.0,
-      "step": 31622
-    },
-    {
-      "epoch": 5.092676838842143,
-      "grad_norm": 0.002594214165583253,
-      "learning_rate": 0.00019998721074986585,
-      "loss": 46.0,
-      "step": 31623
-    },
-    {
-      "epoch": 5.092837875920931,
-      "grad_norm": 0.004807803314179182,
-      "learning_rate": 0.00019998720994075697,
-      "loss": 46.0,
-      "step": 31624
-    },
-    {
-      "epoch": 5.092998912999718,
-      "grad_norm": 0.0019444884965196252,
-      "learning_rate": 0.00019998720913162247,
-      "loss": 46.0,
-      "step": 31625
-    },
-    {
-      "epoch": 5.093159950078506,
-      "grad_norm": 0.014486836269497871,
-      "learning_rate": 0.0001999872083224624,
-      "loss": 46.0,
-      "step": 31626
-    },
-    {
-      "epoch": 5.093320987157293,
-      "grad_norm": 0.001850353553891182,
-      "learning_rate": 0.0001999872075132767,
-      "loss": 46.0,
-      "step": 31627
-    },
-    {
-      "epoch": 5.0934820242360805,
-      "grad_norm": 0.003495913464576006,
-      "learning_rate": 0.00019998720670406545,
-      "loss": 46.0,
-      "step": 31628
-    },
-    {
-      "epoch": 5.093643061314868,
-      "grad_norm": 0.008592427708208561,
-      "learning_rate": 0.0001999872058948286,
-      "loss": 46.0,
-      "step": 31629
-    },
-    {
-      "epoch": 5.093804098393655,
-      "grad_norm": 0.0013628288870677352,
-      "learning_rate": 0.00019998720508556615,
-      "loss": 46.0,
-      "step": 31630
-    },
-    {
-      "epoch": 5.093965135472443,
-      "grad_norm": 0.0036802305839955807,
-      "learning_rate": 0.00019998720427627812,
-      "loss": 46.0,
-      "step": 31631
-    },
-    {
-      "epoch": 5.09412617255123,
-      "grad_norm": 0.0013586009154096246,
-      "learning_rate": 0.0001999872034669645,
-      "loss": 46.0,
-      "step": 31632
-    },
-    {
-      "epoch": 5.094287209630017,
-      "grad_norm": 0.0055445656180381775,
-      "learning_rate": 0.00019998720265762527,
-      "loss": 46.0,
-      "step": 31633
-    },
-    {
-      "epoch": 5.094448246708804,
-      "grad_norm": 0.003254644339904189,
-      "learning_rate": 0.00019998720184826046,
-      "loss": 46.0,
-      "step": 31634
-    },
-    {
-      "epoch": 5.094609283787592,
-      "grad_norm": 0.0018056377302855253,
-      "learning_rate": 0.00019998720103887006,
-      "loss": 46.0,
-      "step": 31635
-    },
-    {
-      "epoch": 5.094770320866379,
-      "grad_norm": 0.005883445963263512,
-      "learning_rate": 0.00019998720022945405,
-      "loss": 46.0,
-      "step": 31636
-    },
-    {
-      "epoch": 5.094931357945167,
-      "grad_norm": 0.01950792782008648,
-      "learning_rate": 0.00019998719942001248,
-      "loss": 46.0,
-      "step": 31637
-    },
-    {
-      "epoch": 5.095092395023954,
-      "grad_norm": 0.008549262769520283,
-      "learning_rate": 0.0001999871986105453,
-      "loss": 46.0,
-      "step": 31638
-    },
-    {
-      "epoch": 5.095253432102742,
-      "grad_norm": 0.00631296169012785,
-      "learning_rate": 0.00019998719780105252,
-      "loss": 46.0,
-      "step": 31639
-    },
-    {
-      "epoch": 5.095414469181529,
-      "grad_norm": 0.0015487271593883634,
-      "learning_rate": 0.00019998719699153416,
-      "loss": 46.0,
-      "step": 31640
-    },
-    {
-      "epoch": 5.0955755062603165,
-      "grad_norm": 0.011292981915175915,
-      "learning_rate": 0.0001999871961819902,
-      "loss": 46.0,
-      "step": 31641
-    },
-    {
-      "epoch": 5.095736543339104,
-      "grad_norm": 0.0037516835145652294,
-      "learning_rate": 0.00019998719537242066,
-      "loss": 46.0,
-      "step": 31642
-    },
-    {
-      "epoch": 5.095897580417891,
-      "grad_norm": 0.007287455257028341,
-      "learning_rate": 0.0001999871945628255,
-      "loss": 46.0,
-      "step": 31643
-    },
-    {
-      "epoch": 5.096058617496679,
-      "grad_norm": 0.0011183179449290037,
-      "learning_rate": 0.00019998719375320477,
-      "loss": 46.0,
-      "step": 31644
-    },
-    {
-      "epoch": 5.096219654575466,
-      "grad_norm": 0.0020118425600230694,
-      "learning_rate": 0.00019998719294355848,
-      "loss": 46.0,
-      "step": 31645
-    },
-    {
-      "epoch": 5.096380691654254,
-      "grad_norm": 0.007203317247331142,
-      "learning_rate": 0.00019998719213388654,
-      "loss": 46.0,
-      "step": 31646
-    },
-    {
-      "epoch": 5.09654172873304,
-      "grad_norm": 0.014000486582517624,
-      "learning_rate": 0.00019998719132418907,
-      "loss": 46.0,
-      "step": 31647
-    },
-    {
-      "epoch": 5.096702765811828,
-      "grad_norm": 0.02207367494702339,
-      "learning_rate": 0.00019998719051446596,
-      "loss": 46.0,
-      "step": 31648
-    },
-    {
-      "epoch": 5.096863802890615,
-      "grad_norm": 0.009998398832976818,
-      "learning_rate": 0.0001999871897047173,
-      "loss": 46.0,
-      "step": 31649
-    },
-    {
-      "epoch": 5.097024839969403,
-      "grad_norm": 0.00410973746329546,
-      "learning_rate": 0.00019998718889494298,
-      "loss": 46.0,
-      "step": 31650
-    },
-    {
-      "epoch": 5.09718587704819,
-      "grad_norm": 0.0008798366761766374,
-      "learning_rate": 0.00019998718808514313,
-      "loss": 46.0,
-      "step": 31651
-    },
-    {
-      "epoch": 5.097346914126978,
-      "grad_norm": 0.002491818042472005,
-      "learning_rate": 0.00019998718727531767,
-      "loss": 46.0,
-      "step": 31652
-    },
-    {
-      "epoch": 5.097507951205765,
-      "grad_norm": 0.003479825798422098,
-      "learning_rate": 0.00019998718646546662,
-      "loss": 46.0,
-      "step": 31653
-    },
-    {
-      "epoch": 5.0976689882845525,
-      "grad_norm": 0.0031058855820447206,
-      "learning_rate": 0.00019998718565559,
-      "loss": 46.0,
-      "step": 31654
-    },
-    {
-      "epoch": 5.09783002536334,
-      "grad_norm": 0.015748916193842888,
-      "learning_rate": 0.00019998718484568774,
-      "loss": 46.0,
-      "step": 31655
-    },
-    {
-      "epoch": 5.097991062442127,
-      "grad_norm": 0.006286006420850754,
-      "learning_rate": 0.00019998718403575988,
-      "loss": 46.0,
-      "step": 31656
-    },
-    {
-      "epoch": 5.098152099520915,
-      "grad_norm": 0.003368560690432787,
-      "learning_rate": 0.00019998718322580648,
-      "loss": 46.0,
-      "step": 31657
-    },
-    {
-      "epoch": 5.098313136599702,
-      "grad_norm": 0.001771627925336361,
-      "learning_rate": 0.00019998718241582747,
-      "loss": 46.0,
-      "step": 31658
-    },
-    {
-      "epoch": 5.09847417367849,
-      "grad_norm": 0.0017098980024456978,
-      "learning_rate": 0.00019998718160582288,
-      "loss": 46.0,
-      "step": 31659
-    },
-    {
-      "epoch": 5.098635210757277,
-      "grad_norm": 0.005052240565419197,
-      "learning_rate": 0.0001999871807957927,
-      "loss": 46.0,
-      "step": 31660
-    },
-    {
-      "epoch": 5.098796247836065,
-      "grad_norm": 0.002359638223424554,
-      "learning_rate": 0.0001999871799857369,
-      "loss": 46.0,
-      "step": 31661
-    },
-    {
-      "epoch": 5.098957284914851,
-      "grad_norm": 0.012051573023200035,
-      "learning_rate": 0.00019998717917565554,
-      "loss": 46.0,
-      "step": 31662
-    },
-    {
-      "epoch": 5.099118321993639,
-      "grad_norm": 0.00994238443672657,
-      "learning_rate": 0.00019998717836554854,
-      "loss": 46.0,
-      "step": 31663
-    },
-    {
-      "epoch": 5.099279359072426,
-      "grad_norm": 0.021187372505664825,
-      "learning_rate": 0.00019998717755541598,
-      "loss": 46.0,
-      "step": 31664
-    },
-    {
-      "epoch": 5.099440396151214,
-      "grad_norm": 0.003129806602373719,
-      "learning_rate": 0.00019998717674525783,
-      "loss": 46.0,
-      "step": 31665
-    },
-    {
-      "epoch": 5.099601433230001,
-      "grad_norm": 0.0065110777504742146,
-      "learning_rate": 0.00019998717593507407,
-      "loss": 46.0,
-      "step": 31666
-    },
-    {
-      "epoch": 5.0997624703087885,
-      "grad_norm": 0.006100281607359648,
-      "learning_rate": 0.00019998717512486475,
-      "loss": 46.0,
-      "step": 31667
-    },
-    {
-      "epoch": 5.099923507387576,
-      "grad_norm": 0.001959226094186306,
-      "learning_rate": 0.00019998717431462984,
-      "loss": 46.0,
-      "step": 31668
-    },
-    {
-      "epoch": 5.100084544466363,
-      "grad_norm": 0.0024511960800737143,
-      "learning_rate": 0.00019998717350436931,
-      "loss": 46.0,
-      "step": 31669
-    },
-    {
-      "epoch": 5.100245581545151,
-      "grad_norm": 0.0038314347621053457,
-      "learning_rate": 0.0001999871726940832,
-      "loss": 46.0,
-      "step": 31670
-    },
-    {
-      "epoch": 5.100406618623938,
-      "grad_norm": 0.006789130158722401,
-      "learning_rate": 0.0001999871718837715,
-      "loss": 46.0,
-      "step": 31671
-    },
-    {
-      "epoch": 5.100567655702726,
-      "grad_norm": 0.004750135820358992,
-      "learning_rate": 0.00019998717107343422,
-      "loss": 46.0,
-      "step": 31672
-    },
-    {
-      "epoch": 5.100728692781513,
-      "grad_norm": 0.003990810830146074,
-      "learning_rate": 0.00019998717026307132,
-      "loss": 46.0,
-      "step": 31673
-    },
-    {
-      "epoch": 5.100889729860301,
-      "grad_norm": 0.011149369180202484,
-      "learning_rate": 0.00019998716945268284,
-      "loss": 46.0,
-      "step": 31674
-    },
-    {
-      "epoch": 5.101050766939088,
-      "grad_norm": 0.004834349267184734,
-      "learning_rate": 0.0001999871686422688,
-      "loss": 46.0,
-      "step": 31675
-    },
-    {
-      "epoch": 5.1012118040178756,
-      "grad_norm": 0.0037851883098483086,
-      "learning_rate": 0.0001999871678318291,
-      "loss": 46.0,
-      "step": 31676
-    },
-    {
-      "epoch": 5.101372841096662,
-      "grad_norm": 0.0028146200347691774,
-      "learning_rate": 0.00019998716702136386,
-      "loss": 46.0,
-      "step": 31677
-    },
-    {
-      "epoch": 5.10153387817545,
-      "grad_norm": 0.006069815717637539,
-      "learning_rate": 0.00019998716621087302,
-      "loss": 46.0,
-      "step": 31678
-    },
-    {
-      "epoch": 5.101694915254237,
-      "grad_norm": 0.002303576096892357,
-      "learning_rate": 0.00019998716540035657,
-      "loss": 46.0,
-      "step": 31679
-    },
-    {
-      "epoch": 5.1018559523330245,
-      "grad_norm": 0.0019306502072140574,
-      "learning_rate": 0.00019998716458981456,
-      "loss": 46.0,
-      "step": 31680
-    },
-    {
-      "epoch": 5.102016989411812,
-      "grad_norm": 0.0035780046600848436,
-      "learning_rate": 0.00019998716377924694,
-      "loss": 46.0,
-      "step": 31681
-    },
-    {
-      "epoch": 5.102178026490599,
-      "grad_norm": 0.015230325981974602,
-      "learning_rate": 0.00019998716296865373,
-      "loss": 46.0,
-      "step": 31682
-    },
-    {
-      "epoch": 5.102339063569387,
-      "grad_norm": 0.003726790426298976,
-      "learning_rate": 0.00019998716215803493,
-      "loss": 46.0,
-      "step": 31683
-    },
-    {
-      "epoch": 5.102500100648174,
-      "grad_norm": 0.006447991821914911,
-      "learning_rate": 0.00019998716134739055,
-      "loss": 46.0,
-      "step": 31684
-    },
-    {
-      "epoch": 5.102661137726962,
-      "grad_norm": 0.007664111442863941,
-      "learning_rate": 0.00019998716053672055,
-      "loss": 46.0,
-      "step": 31685
-    },
-    {
-      "epoch": 5.102822174805749,
-      "grad_norm": 0.011768290773034096,
-      "learning_rate": 0.000199987159726025,
-      "loss": 46.0,
-      "step": 31686
-    },
-    {
-      "epoch": 5.102983211884537,
-      "grad_norm": 0.00295541831292212,
-      "learning_rate": 0.00019998715891530382,
-      "loss": 46.0,
-      "step": 31687
-    },
-    {
-      "epoch": 5.103144248963324,
-      "grad_norm": 0.004615542478859425,
-      "learning_rate": 0.00019998715810455706,
-      "loss": 46.0,
-      "step": 31688
-    },
-    {
-      "epoch": 5.1033052860421115,
-      "grad_norm": 0.0015148420352488756,
-      "learning_rate": 0.0001999871572937847,
-      "loss": 46.0,
-      "step": 31689
-    },
-    {
-      "epoch": 5.103466323120899,
-      "grad_norm": 0.0018063432071357965,
-      "learning_rate": 0.00019998715648298675,
-      "loss": 46.0,
-      "step": 31690
-    },
-    {
-      "epoch": 5.1036273601996855,
-      "grad_norm": 0.003022967604920268,
-      "learning_rate": 0.00019998715567216325,
-      "loss": 46.0,
-      "step": 31691
-    },
-    {
-      "epoch": 5.103788397278473,
-      "grad_norm": 0.00608081417158246,
-      "learning_rate": 0.0001999871548613141,
-      "loss": 46.0,
-      "step": 31692
-    },
-    {
-      "epoch": 5.1039494343572605,
-      "grad_norm": 0.01729155145585537,
-      "learning_rate": 0.0001999871540504394,
-      "loss": 46.0,
-      "step": 31693
-    },
-    {
-      "epoch": 5.104110471436048,
-      "grad_norm": 0.0017848373390734196,
-      "learning_rate": 0.00019998715323953908,
-      "loss": 46.0,
-      "step": 31694
-    },
-    {
-      "epoch": 5.104271508514835,
-      "grad_norm": 0.018515024334192276,
-      "learning_rate": 0.00019998715242861318,
-      "loss": 46.0,
-      "step": 31695
-    },
-    {
-      "epoch": 5.104432545593623,
-      "grad_norm": 0.0021290862932801247,
-      "learning_rate": 0.00019998715161766167,
-      "loss": 46.0,
-      "step": 31696
-    },
-    {
-      "epoch": 5.10459358267241,
-      "grad_norm": 0.0026584663428366184,
-      "learning_rate": 0.00019998715080668463,
-      "loss": 46.0,
-      "step": 31697
-    },
-    {
-      "epoch": 5.104754619751198,
-      "grad_norm": 0.0021825479343533516,
-      "learning_rate": 0.00019998714999568194,
-      "loss": 46.0,
-      "step": 31698
-    },
-    {
-      "epoch": 5.104915656829985,
-      "grad_norm": 0.005995746236294508,
-      "learning_rate": 0.00019998714918465367,
-      "loss": 46.0,
-      "step": 31699
-    },
-    {
-      "epoch": 5.105076693908773,
-      "grad_norm": 0.01741694286465645,
-      "learning_rate": 0.00019998714837359983,
-      "loss": 46.0,
-      "step": 31700
-    },
-    {
-      "epoch": 5.10523773098756,
-      "grad_norm": 0.009037435054779053,
-      "learning_rate": 0.00019998714756252038,
-      "loss": 46.0,
-      "step": 31701
-    },
-    {
-      "epoch": 5.1053987680663475,
-      "grad_norm": 0.0029692198149859905,
-      "learning_rate": 0.00019998714675141535,
-      "loss": 46.0,
-      "step": 31702
-    },
-    {
-      "epoch": 5.105559805145135,
-      "grad_norm": 0.004011984448879957,
-      "learning_rate": 0.0001999871459402847,
-      "loss": 46.0,
-      "step": 31703
-    },
-    {
-      "epoch": 5.105720842223922,
-      "grad_norm": 0.011755079962313175,
-      "learning_rate": 0.00019998714512912846,
-      "loss": 46.0,
-      "step": 31704
-    },
-    {
-      "epoch": 5.10588187930271,
-      "grad_norm": 0.0044697984121739864,
-      "learning_rate": 0.00019998714431794667,
-      "loss": 46.0,
-      "step": 31705
-    },
-    {
-      "epoch": 5.106042916381496,
-      "grad_norm": 0.004943686071783304,
-      "learning_rate": 0.00019998714350673926,
-      "loss": 46.0,
-      "step": 31706
-    },
-    {
-      "epoch": 5.106203953460284,
-      "grad_norm": 0.008940238505601883,
-      "learning_rate": 0.00019998714269550626,
-      "loss": 46.0,
-      "step": 31707
-    },
-    {
-      "epoch": 5.106364990539071,
-      "grad_norm": 0.0014674649573862553,
-      "learning_rate": 0.00019998714188424765,
-      "loss": 46.0,
-      "step": 31708
-    },
-    {
-      "epoch": 5.106526027617859,
-      "grad_norm": 0.003657751251012087,
-      "learning_rate": 0.00019998714107296347,
-      "loss": 46.0,
-      "step": 31709
-    },
-    {
-      "epoch": 5.106687064696646,
-      "grad_norm": 0.0029363837093114853,
-      "learning_rate": 0.0001999871402616537,
-      "loss": 46.0,
-      "step": 31710
-    },
-    {
-      "epoch": 5.106848101775434,
-      "grad_norm": 0.010432448238134384,
-      "learning_rate": 0.00019998713945031834,
-      "loss": 46.0,
-      "step": 31711
-    },
-    {
-      "epoch": 5.107009138854221,
-      "grad_norm": 0.006936706602573395,
-      "learning_rate": 0.0001999871386389574,
-      "loss": 46.0,
-      "step": 31712
-    },
-    {
-      "epoch": 5.107170175933009,
-      "grad_norm": 0.013828225433826447,
-      "learning_rate": 0.00019998713782757083,
-      "loss": 46.0,
-      "step": 31713
-    },
-    {
-      "epoch": 5.107331213011796,
-      "grad_norm": 0.013111437670886517,
-      "learning_rate": 0.0001999871370161587,
-      "loss": 46.0,
-      "step": 31714
-    },
-    {
-      "epoch": 5.1074922500905835,
-      "grad_norm": 0.010958925820887089,
-      "learning_rate": 0.00019998713620472097,
-      "loss": 46.0,
-      "step": 31715
-    },
-    {
-      "epoch": 5.107653287169371,
-      "grad_norm": 0.004703730810433626,
-      "learning_rate": 0.00019998713539325764,
-      "loss": 46.0,
-      "step": 31716
-    },
-    {
-      "epoch": 5.107814324248158,
-      "grad_norm": 0.0008039295789785683,
-      "learning_rate": 0.0001999871345817687,
-      "loss": 46.0,
-      "step": 31717
-    },
-    {
-      "epoch": 5.107975361326946,
-      "grad_norm": 0.01591426320374012,
-      "learning_rate": 0.00019998713377025423,
-      "loss": 46.0,
-      "step": 31718
-    },
-    {
-      "epoch": 5.108136398405733,
-      "grad_norm": 0.002451897133141756,
-      "learning_rate": 0.00019998713295871413,
-      "loss": 46.0,
-      "step": 31719
-    },
-    {
-      "epoch": 5.10829743548452,
-      "grad_norm": 0.0022533764131367207,
-      "learning_rate": 0.00019998713214714844,
-      "loss": 46.0,
-      "step": 31720
-    },
-    {
-      "epoch": 5.108458472563307,
-      "grad_norm": 0.017491359263658524,
-      "learning_rate": 0.00019998713133555714,
-      "loss": 46.0,
-      "step": 31721
-    },
-    {
-      "epoch": 5.108619509642095,
-      "grad_norm": 0.0038294882979243994,
-      "learning_rate": 0.00019998713052394028,
-      "loss": 46.0,
-      "step": 31722
-    },
-    {
-      "epoch": 5.108780546720882,
-      "grad_norm": 0.005774224177002907,
-      "learning_rate": 0.00019998712971229784,
-      "loss": 46.0,
-      "step": 31723
-    },
-    {
-      "epoch": 5.10894158379967,
-      "grad_norm": 0.0029772575944662094,
-      "learning_rate": 0.00019998712890062978,
-      "loss": 46.0,
-      "step": 31724
-    },
-    {
-      "epoch": 5.109102620878457,
-      "grad_norm": 0.004000029526650906,
-      "learning_rate": 0.00019998712808893613,
-      "loss": 46.0,
-      "step": 31725
-    },
-    {
-      "epoch": 5.109263657957245,
-      "grad_norm": 0.0035110164899379015,
-      "learning_rate": 0.00019998712727721686,
-      "loss": 46.0,
-      "step": 31726
-    },
-    {
-      "epoch": 5.109424695036032,
-      "grad_norm": 0.0008503383141942322,
-      "learning_rate": 0.00019998712646547207,
-      "loss": 46.0,
-      "step": 31727
-    },
-    {
-      "epoch": 5.1095857321148195,
-      "grad_norm": 0.004292924888432026,
-      "learning_rate": 0.00019998712565370163,
-      "loss": 46.0,
-      "step": 31728
-    },
-    {
-      "epoch": 5.109746769193607,
-      "grad_norm": 0.0012082025641575456,
-      "learning_rate": 0.00019998712484190563,
-      "loss": 46.0,
-      "step": 31729
-    },
-    {
-      "epoch": 5.109907806272394,
-      "grad_norm": 0.0033361774403601885,
-      "learning_rate": 0.00019998712403008402,
-      "loss": 46.0,
-      "step": 31730
-    },
-    {
-      "epoch": 5.110068843351182,
-      "grad_norm": 0.006076530087739229,
-      "learning_rate": 0.00019998712321823682,
-      "loss": 46.0,
-      "step": 31731
-    },
-    {
-      "epoch": 5.110229880429969,
-      "grad_norm": 0.002917871344834566,
-      "learning_rate": 0.00019998712240636404,
-      "loss": 46.0,
-      "step": 31732
-    },
-    {
-      "epoch": 5.110390917508757,
-      "grad_norm": 0.002042217645794153,
-      "learning_rate": 0.00019998712159446566,
-      "loss": 46.0,
-      "step": 31733
-    },
-    {
-      "epoch": 5.110551954587544,
-      "grad_norm": 0.004983257502317429,
-      "learning_rate": 0.00019998712078254168,
-      "loss": 46.0,
-      "step": 31734
-    },
-    {
-      "epoch": 5.110712991666331,
-      "grad_norm": 0.003702867776155472,
-      "learning_rate": 0.00019998711997059213,
-      "loss": 46.0,
-      "step": 31735
-    },
-    {
-      "epoch": 5.110874028745118,
-      "grad_norm": 0.006975711323320866,
-      "learning_rate": 0.00019998711915861697,
-      "loss": 46.0,
-      "step": 31736
-    },
-    {
-      "epoch": 5.111035065823906,
-      "grad_norm": 0.015306190587580204,
-      "learning_rate": 0.00019998711834661622,
-      "loss": 46.0,
-      "step": 31737
-    },
-    {
-      "epoch": 5.111196102902693,
-      "grad_norm": 0.0022727414034307003,
-      "learning_rate": 0.00019998711753458988,
-      "loss": 46.0,
-      "step": 31738
-    },
-    {
-      "epoch": 5.111357139981481,
-      "grad_norm": 0.01086303312331438,
-      "learning_rate": 0.00019998711672253796,
-      "loss": 46.0,
-      "step": 31739
-    },
-    {
-      "epoch": 5.111518177060268,
-      "grad_norm": 0.0005476285587064922,
-      "learning_rate": 0.00019998711591046045,
-      "loss": 46.0,
-      "step": 31740
-    },
-    {
-      "epoch": 5.1116792141390555,
-      "grad_norm": 0.019337452948093414,
-      "learning_rate": 0.0001999871150983573,
-      "loss": 46.0,
-      "step": 31741
-    },
-    {
-      "epoch": 5.111840251217843,
-      "grad_norm": 0.003269969252869487,
-      "learning_rate": 0.00019998711428622861,
-      "loss": 46.0,
-      "step": 31742
-    },
-    {
-      "epoch": 5.11200128829663,
-      "grad_norm": 0.0025042598135769367,
-      "learning_rate": 0.00019998711347407432,
-      "loss": 46.0,
-      "step": 31743
-    },
-    {
-      "epoch": 5.112162325375418,
-      "grad_norm": 0.00527442991733551,
-      "learning_rate": 0.00019998711266189446,
-      "loss": 46.0,
-      "step": 31744
-    },
-    {
-      "epoch": 5.112323362454205,
-      "grad_norm": 0.008749818429350853,
-      "learning_rate": 0.00019998711184968896,
-      "loss": 46.0,
-      "step": 31745
-    },
-    {
-      "epoch": 5.112484399532993,
-      "grad_norm": 0.006924590095877647,
-      "learning_rate": 0.0001999871110374579,
-      "loss": 46.0,
-      "step": 31746
-    },
-    {
-      "epoch": 5.11264543661178,
-      "grad_norm": 0.004711885005235672,
-      "learning_rate": 0.00019998711022520122,
-      "loss": 46.0,
-      "step": 31747
-    },
-    {
-      "epoch": 5.112806473690568,
-      "grad_norm": 0.0006080442690290511,
-      "learning_rate": 0.00019998710941291896,
-      "loss": 46.0,
-      "step": 31748
-    },
-    {
-      "epoch": 5.112967510769355,
-      "grad_norm": 0.0035481364466249943,
-      "learning_rate": 0.00019998710860061114,
-      "loss": 46.0,
-      "step": 31749
-    },
-    {
-      "epoch": 5.113128547848142,
-      "grad_norm": 0.02730635367333889,
-      "learning_rate": 0.0001999871077882777,
-      "loss": 46.0,
-      "step": 31750
-    },
-    {
-      "epoch": 5.113289584926929,
-      "grad_norm": 0.0027243653312325478,
-      "learning_rate": 0.00019998710697591868,
-      "loss": 46.0,
-      "step": 31751
-    },
-    {
-      "epoch": 5.1134506220057165,
-      "grad_norm": 0.01114586740732193,
-      "learning_rate": 0.00019998710616353404,
-      "loss": 46.0,
-      "step": 31752
-    },
-    {
-      "epoch": 5.113611659084504,
-      "grad_norm": 0.010896636173129082,
-      "learning_rate": 0.00019998710535112384,
-      "loss": 46.0,
-      "step": 31753
-    },
-    {
-      "epoch": 5.113772696163291,
-      "grad_norm": 0.007446507923305035,
-      "learning_rate": 0.00019998710453868803,
-      "loss": 46.0,
-      "step": 31754
-    },
-    {
-      "epoch": 5.113933733242079,
-      "grad_norm": 0.0055212113074958324,
-      "learning_rate": 0.00019998710372622663,
-      "loss": 46.0,
-      "step": 31755
-    },
-    {
-      "epoch": 5.114094770320866,
-      "grad_norm": 0.01968042179942131,
-      "learning_rate": 0.00019998710291373965,
-      "loss": 46.0,
-      "step": 31756
-    },
-    {
-      "epoch": 5.114255807399654,
-      "grad_norm": 0.004497861955314875,
-      "learning_rate": 0.0001999871021012271,
-      "loss": 46.0,
-      "step": 31757
-    },
-    {
-      "epoch": 5.114416844478441,
-      "grad_norm": 0.002537531778216362,
-      "learning_rate": 0.0001999871012886889,
-      "loss": 46.0,
-      "step": 31758
-    },
-    {
-      "epoch": 5.114577881557229,
-      "grad_norm": 0.0012752487091347575,
-      "learning_rate": 0.00019998710047612514,
-      "loss": 46.0,
-      "step": 31759
-    },
-    {
-      "epoch": 5.114738918636016,
-      "grad_norm": 0.0038238794077187777,
-      "learning_rate": 0.0001999870996635358,
-      "loss": 46.0,
-      "step": 31760
-    },
-    {
-      "epoch": 5.114899955714804,
-      "grad_norm": 0.002374745672568679,
-      "learning_rate": 0.00019998709885092085,
-      "loss": 46.0,
-      "step": 31761
-    },
-    {
-      "epoch": 5.115060992793591,
-      "grad_norm": 0.0030338556971400976,
-      "learning_rate": 0.00019998709803828032,
-      "loss": 46.0,
-      "step": 31762
-    },
-    {
-      "epoch": 5.1152220298723785,
-      "grad_norm": 0.004380014259368181,
-      "learning_rate": 0.0001999870972256142,
-      "loss": 46.0,
-      "step": 31763
-    },
-    {
-      "epoch": 5.115383066951165,
-      "grad_norm": 0.0029799898620694876,
-      "learning_rate": 0.00019998709641292245,
-      "loss": 46.0,
-      "step": 31764
-    },
-    {
-      "epoch": 5.1155441040299525,
-      "grad_norm": 0.010815626010298729,
-      "learning_rate": 0.00019998709560020516,
-      "loss": 46.0,
-      "step": 31765
-    },
-    {
-      "epoch": 5.11570514110874,
-      "grad_norm": 0.01009557954967022,
-      "learning_rate": 0.00019998709478746224,
-      "loss": 46.0,
-      "step": 31766
-    },
-    {
-      "epoch": 5.115866178187527,
-      "grad_norm": 0.006285614334046841,
-      "learning_rate": 0.00019998709397469377,
-      "loss": 46.0,
-      "step": 31767
-    },
-    {
-      "epoch": 5.116027215266315,
-      "grad_norm": 0.011339014396071434,
-      "learning_rate": 0.00019998709316189968,
-      "loss": 46.0,
-      "step": 31768
-    },
-    {
-      "epoch": 5.116188252345102,
-      "grad_norm": 0.0037707164883613586,
-      "learning_rate": 0.00019998709234907998,
-      "loss": 46.0,
-      "step": 31769
-    },
-    {
-      "epoch": 5.11634928942389,
-      "grad_norm": 0.005328834988176823,
-      "learning_rate": 0.00019998709153623475,
-      "loss": 46.0,
-      "step": 31770
-    },
-    {
-      "epoch": 5.116510326502677,
-      "grad_norm": 0.0035266312770545483,
-      "learning_rate": 0.00019998709072336387,
-      "loss": 46.0,
-      "step": 31771
-    },
-    {
-      "epoch": 5.116671363581465,
-      "grad_norm": 0.0072785671800374985,
-      "learning_rate": 0.00019998708991046744,
-      "loss": 46.0,
-      "step": 31772
-    },
-    {
-      "epoch": 5.116832400660252,
-      "grad_norm": 0.0034306710585951805,
-      "learning_rate": 0.0001999870890975454,
-      "loss": 46.0,
-      "step": 31773
-    },
-    {
-      "epoch": 5.11699343773904,
-      "grad_norm": 0.00595342181622982,
-      "learning_rate": 0.00019998708828459775,
-      "loss": 46.0,
-      "step": 31774
-    },
-    {
-      "epoch": 5.117154474817827,
-      "grad_norm": 0.003118218155577779,
-      "learning_rate": 0.00019998708747162453,
-      "loss": 46.0,
-      "step": 31775
-    },
-    {
-      "epoch": 5.1173155118966145,
-      "grad_norm": 0.010127179324626923,
-      "learning_rate": 0.00019998708665862572,
-      "loss": 46.0,
-      "step": 31776
-    },
-    {
-      "epoch": 5.117476548975402,
-      "grad_norm": 0.01014582347124815,
-      "learning_rate": 0.00019998708584560132,
-      "loss": 46.0,
-      "step": 31777
-    },
-    {
-      "epoch": 5.117637586054189,
-      "grad_norm": 0.00323995528742671,
-      "learning_rate": 0.0001999870850325513,
-      "loss": 46.0,
-      "step": 31778
-    },
-    {
-      "epoch": 5.117798623132976,
-      "grad_norm": 0.004588102921843529,
-      "learning_rate": 0.0001999870842194757,
-      "loss": 46.0,
-      "step": 31779
-    },
-    {
-      "epoch": 5.117959660211763,
-      "grad_norm": 0.006466016639024019,
-      "learning_rate": 0.00019998708340637452,
-      "loss": 46.0,
-      "step": 31780
-    },
-    {
-      "epoch": 5.118120697290551,
-      "grad_norm": 0.0020204796455800533,
-      "learning_rate": 0.00019998708259324777,
-      "loss": 46.0,
-      "step": 31781
-    },
-    {
-      "epoch": 5.118281734369338,
-      "grad_norm": 0.015112536028027534,
-      "learning_rate": 0.00019998708178009539,
-      "loss": 46.0,
-      "step": 31782
-    },
-    {
-      "epoch": 5.118442771448126,
-      "grad_norm": 0.0012880859430879354,
-      "learning_rate": 0.00019998708096691744,
-      "loss": 46.0,
-      "step": 31783
-    },
-    {
-      "epoch": 5.118603808526913,
-      "grad_norm": 0.0015607925597578287,
-      "learning_rate": 0.00019998708015371387,
-      "loss": 46.0,
-      "step": 31784
-    },
-    {
-      "epoch": 5.118764845605701,
-      "grad_norm": 0.004017500206828117,
-      "learning_rate": 0.00019998707934048472,
-      "loss": 46.0,
-      "step": 31785
-    },
-    {
-      "epoch": 5.118925882684488,
-      "grad_norm": 0.004509245976805687,
-      "learning_rate": 0.00019998707852723001,
-      "loss": 46.0,
-      "step": 31786
-    },
-    {
-      "epoch": 5.119086919763276,
-      "grad_norm": 0.005218183156102896,
-      "learning_rate": 0.0001999870777139497,
-      "loss": 46.0,
-      "step": 31787
-    },
-    {
-      "epoch": 5.119247956842063,
-      "grad_norm": 0.006368751637637615,
-      "learning_rate": 0.00019998707690064378,
-      "loss": 46.0,
-      "step": 31788
-    },
-    {
-      "epoch": 5.1194089939208505,
-      "grad_norm": 0.008672593161463737,
-      "learning_rate": 0.00019998707608731225,
-      "loss": 46.0,
-      "step": 31789
-    },
-    {
-      "epoch": 5.119570030999638,
-      "grad_norm": 0.010050750337541103,
-      "learning_rate": 0.00019998707527395517,
-      "loss": 46.0,
-      "step": 31790
-    },
-    {
-      "epoch": 5.119731068078425,
-      "grad_norm": 0.018477262929081917,
-      "learning_rate": 0.00019998707446057247,
-      "loss": 46.0,
-      "step": 31791
-    },
-    {
-      "epoch": 5.119892105157213,
-      "grad_norm": 0.0165387112647295,
-      "learning_rate": 0.0001999870736471642,
-      "loss": 46.0,
-      "step": 31792
-    },
-    {
-      "epoch": 5.120053142235999,
-      "grad_norm": 0.0012520333984866738,
-      "learning_rate": 0.0001999870728337303,
-      "loss": 46.0,
-      "step": 31793
-    },
-    {
-      "epoch": 5.120214179314787,
-      "grad_norm": 0.0046949079260230064,
-      "learning_rate": 0.00019998707202027085,
-      "loss": 46.0,
-      "step": 31794
-    },
-    {
-      "epoch": 5.120375216393574,
-      "grad_norm": 0.010558567009866238,
-      "learning_rate": 0.0001999870712067858,
-      "loss": 46.0,
-      "step": 31795
-    },
-    {
-      "epoch": 5.120536253472362,
-      "grad_norm": 0.024613451212644577,
-      "learning_rate": 0.00019998707039327513,
-      "loss": 46.0,
-      "step": 31796
-    },
-    {
-      "epoch": 5.120697290551149,
-      "grad_norm": 0.0015605506487190723,
-      "learning_rate": 0.0001999870695797389,
-      "loss": 46.0,
-      "step": 31797
-    },
-    {
-      "epoch": 5.120858327629937,
-      "grad_norm": 0.0032269759103655815,
-      "learning_rate": 0.00019998706876617707,
-      "loss": 46.0,
-      "step": 31798
-    },
-    {
-      "epoch": 5.121019364708724,
-      "grad_norm": 0.00508446479216218,
-      "learning_rate": 0.00019998706795258965,
-      "loss": 46.0,
-      "step": 31799
-    },
-    {
-      "epoch": 5.1211804017875115,
-      "grad_norm": 0.0070706079714000225,
-      "learning_rate": 0.00019998706713897664,
-      "loss": 46.0,
-      "step": 31800
-    },
-    {
-      "epoch": 5.121341438866299,
-      "grad_norm": 0.002075758296996355,
-      "learning_rate": 0.00019998706632533804,
-      "loss": 46.0,
-      "step": 31801
-    },
-    {
-      "epoch": 5.1215024759450865,
-      "grad_norm": 0.003729782300069928,
-      "learning_rate": 0.00019998706551167383,
-      "loss": 46.0,
-      "step": 31802
-    },
-    {
-      "epoch": 5.121663513023874,
-      "grad_norm": 0.0024954641703516245,
-      "learning_rate": 0.00019998706469798405,
-      "loss": 46.0,
-      "step": 31803
-    },
-    {
-      "epoch": 5.121824550102661,
-      "grad_norm": 0.0027922680601477623,
-      "learning_rate": 0.00019998706388426867,
-      "loss": 46.0,
-      "step": 31804
-    },
-    {
-      "epoch": 5.121985587181449,
-      "grad_norm": 0.0034518782049417496,
-      "learning_rate": 0.0001999870630705277,
-      "loss": 46.0,
-      "step": 31805
-    },
-    {
-      "epoch": 5.122146624260236,
-      "grad_norm": 0.008538051508367062,
-      "learning_rate": 0.00019998706225676113,
-      "loss": 46.0,
-      "step": 31806
-    },
-    {
-      "epoch": 5.122307661339024,
-      "grad_norm": 0.010760035365819931,
-      "learning_rate": 0.00019998706144296898,
-      "loss": 46.0,
-      "step": 31807
-    },
-    {
-      "epoch": 5.12246869841781,
-      "grad_norm": 0.0022751842625439167,
-      "learning_rate": 0.00019998706062915122,
-      "loss": 46.0,
-      "step": 31808
-    },
-    {
-      "epoch": 5.122629735496598,
-      "grad_norm": 0.0067843664437532425,
-      "learning_rate": 0.00019998705981530787,
-      "loss": 46.0,
-      "step": 31809
-    },
-    {
-      "epoch": 5.122790772575385,
-      "grad_norm": 0.017063351348042488,
-      "learning_rate": 0.00019998705900143896,
-      "loss": 46.0,
-      "step": 31810
-    },
-    {
-      "epoch": 5.122951809654173,
-      "grad_norm": 0.012322777882218361,
-      "learning_rate": 0.00019998705818754443,
-      "loss": 46.0,
-      "step": 31811
-    },
-    {
-      "epoch": 5.12311284673296,
-      "grad_norm": 0.014878259971737862,
-      "learning_rate": 0.00019998705737362432,
-      "loss": 46.0,
-      "step": 31812
-    },
-    {
-      "epoch": 5.1232738838117475,
-      "grad_norm": 0.0020468709990382195,
-      "learning_rate": 0.00019998705655967862,
-      "loss": 46.0,
-      "step": 31813
-    },
-    {
-      "epoch": 5.123434920890535,
-      "grad_norm": 0.0025103818625211716,
-      "learning_rate": 0.0001999870557457073,
-      "loss": 46.0,
-      "step": 31814
-    },
-    {
-      "epoch": 5.123595957969322,
-      "grad_norm": 0.0058688693679869175,
-      "learning_rate": 0.00019998705493171043,
-      "loss": 46.0,
-      "step": 31815
-    },
-    {
-      "epoch": 5.12375699504811,
-      "grad_norm": 0.004198704846203327,
-      "learning_rate": 0.00019998705411768795,
-      "loss": 46.0,
-      "step": 31816
-    },
-    {
-      "epoch": 5.123918032126897,
-      "grad_norm": 0.011426171287894249,
-      "learning_rate": 0.00019998705330363987,
-      "loss": 46.0,
-      "step": 31817
-    },
-    {
-      "epoch": 5.124079069205685,
-      "grad_norm": 0.006741536781191826,
-      "learning_rate": 0.0001999870524895662,
-      "loss": 46.0,
-      "step": 31818
-    },
-    {
-      "epoch": 5.124240106284472,
-      "grad_norm": 0.004225070588290691,
-      "learning_rate": 0.000199987051675467,
-      "loss": 46.0,
-      "step": 31819
-    },
-    {
-      "epoch": 5.12440114336326,
-      "grad_norm": 0.004235315602272749,
-      "learning_rate": 0.00019998705086134212,
-      "loss": 46.0,
-      "step": 31820
-    },
-    {
-      "epoch": 5.124562180442047,
-      "grad_norm": 0.002773455111309886,
-      "learning_rate": 0.0001999870500471917,
-      "loss": 46.0,
-      "step": 31821
-    },
-    {
-      "epoch": 5.124723217520835,
-      "grad_norm": 0.004156128969043493,
-      "learning_rate": 0.00019998704923301566,
-      "loss": 46.0,
-      "step": 31822
-    },
-    {
-      "epoch": 5.124884254599621,
-      "grad_norm": 0.004585111979395151,
-      "learning_rate": 0.00019998704841881404,
-      "loss": 46.0,
-      "step": 31823
-    },
-    {
-      "epoch": 5.125045291678409,
-      "grad_norm": 0.009455375373363495,
-      "learning_rate": 0.00019998704760458683,
-      "loss": 46.0,
-      "step": 31824
-    },
-    {
-      "epoch": 5.125206328757196,
-      "grad_norm": 0.004446776583790779,
-      "learning_rate": 0.00019998704679033403,
-      "loss": 46.0,
-      "step": 31825
-    },
-    {
-      "epoch": 5.1253673658359835,
-      "grad_norm": 0.005561869125813246,
-      "learning_rate": 0.00019998704597605564,
-      "loss": 46.0,
-      "step": 31826
-    },
-    {
-      "epoch": 5.125528402914771,
-      "grad_norm": 0.005660254042595625,
-      "learning_rate": 0.00019998704516175164,
-      "loss": 46.0,
-      "step": 31827
-    },
-    {
-      "epoch": 5.125689439993558,
-      "grad_norm": 0.003981557209044695,
-      "learning_rate": 0.00019998704434742205,
-      "loss": 46.0,
-      "step": 31828
-    },
-    {
-      "epoch": 5.125850477072346,
-      "grad_norm": 0.014550895430147648,
-      "learning_rate": 0.0001999870435330669,
-      "loss": 46.0,
-      "step": 31829
-    },
-    {
-      "epoch": 5.126011514151133,
-      "grad_norm": 0.0021189465187489986,
-      "learning_rate": 0.00019998704271868614,
-      "loss": 46.0,
-      "step": 31830
-    },
-    {
-      "epoch": 5.126172551229921,
-      "grad_norm": 0.01488464791327715,
-      "learning_rate": 0.0001999870419042798,
-      "loss": 46.0,
-      "step": 31831
-    },
-    {
-      "epoch": 5.126333588308708,
-      "grad_norm": 0.018151327967643738,
-      "learning_rate": 0.00019998704108984785,
-      "loss": 46.0,
-      "step": 31832
-    },
-    {
-      "epoch": 5.126494625387496,
-      "grad_norm": 0.011158701963722706,
-      "learning_rate": 0.0001999870402753903,
-      "loss": 46.0,
-      "step": 31833
-    },
-    {
-      "epoch": 5.126655662466283,
-      "grad_norm": 0.006380947772413492,
-      "learning_rate": 0.00019998703946090717,
-      "loss": 46.0,
-      "step": 31834
-    },
-    {
-      "epoch": 5.126816699545071,
-      "grad_norm": 0.0044798413291573524,
-      "learning_rate": 0.00019998703864639844,
-      "loss": 46.0,
-      "step": 31835
-    },
-    {
-      "epoch": 5.126977736623858,
-      "grad_norm": 0.005650964565575123,
-      "learning_rate": 0.00019998703783186416,
-      "loss": 46.0,
-      "step": 31836
-    },
-    {
-      "epoch": 5.1271387737026455,
-      "grad_norm": 0.0026274228002876043,
-      "learning_rate": 0.00019998703701730426,
-      "loss": 46.0,
-      "step": 31837
-    },
-    {
-      "epoch": 5.127299810781432,
-      "grad_norm": 0.0026627432089298964,
-      "learning_rate": 0.00019998703620271874,
-      "loss": 46.0,
-      "step": 31838
-    },
-    {
-      "epoch": 5.1274608478602195,
-      "grad_norm": 0.0013048682594671845,
-      "learning_rate": 0.00019998703538810767,
-      "loss": 46.0,
-      "step": 31839
-    },
-    {
-      "epoch": 5.127621884939007,
-      "grad_norm": 0.006113915238529444,
-      "learning_rate": 0.00019998703457347098,
-      "loss": 46.0,
-      "step": 31840
-    },
-    {
-      "epoch": 5.127782922017794,
-      "grad_norm": 0.0057538640685379505,
-      "learning_rate": 0.0001999870337588087,
-      "loss": 46.0,
-      "step": 31841
-    },
-    {
-      "epoch": 5.127943959096582,
-      "grad_norm": 0.0028739715926349163,
-      "learning_rate": 0.00019998703294412087,
-      "loss": 46.0,
-      "step": 31842
-    },
-    {
-      "epoch": 5.128104996175369,
-      "grad_norm": 0.003068583784624934,
-      "learning_rate": 0.00019998703212940742,
-      "loss": 46.0,
-      "step": 31843
-    },
-    {
-      "epoch": 5.128266033254157,
-      "grad_norm": 0.0021642097271978855,
-      "learning_rate": 0.00019998703131466838,
-      "loss": 46.0,
-      "step": 31844
-    },
-    {
-      "epoch": 5.128427070332944,
-      "grad_norm": 0.0014632302336394787,
-      "learning_rate": 0.00019998703049990373,
-      "loss": 46.0,
-      "step": 31845
-    },
-    {
-      "epoch": 5.128588107411732,
-      "grad_norm": 0.012482680380344391,
-      "learning_rate": 0.0001999870296851135,
-      "loss": 46.0,
-      "step": 31846
-    },
-    {
-      "epoch": 5.128749144490519,
-      "grad_norm": 0.02205357700586319,
-      "learning_rate": 0.00019998702887029767,
-      "loss": 46.0,
-      "step": 31847
-    },
-    {
-      "epoch": 5.128910181569307,
-      "grad_norm": 0.00596151826903224,
-      "learning_rate": 0.00019998702805545628,
-      "loss": 46.0,
-      "step": 31848
-    },
-    {
-      "epoch": 5.129071218648094,
-      "grad_norm": 0.017880428582429886,
-      "learning_rate": 0.00019998702724058928,
-      "loss": 46.0,
-      "step": 31849
-    },
-    {
-      "epoch": 5.1292322557268815,
-      "grad_norm": 0.004431016743183136,
-      "learning_rate": 0.0001999870264256967,
-      "loss": 46.0,
-      "step": 31850
-    },
-    {
-      "epoch": 5.129393292805669,
-      "grad_norm": 0.003324661171063781,
-      "learning_rate": 0.00019998702561077852,
-      "loss": 46.0,
-      "step": 31851
-    },
-    {
-      "epoch": 5.1295543298844555,
-      "grad_norm": 0.014620086178183556,
-      "learning_rate": 0.00019998702479583473,
-      "loss": 46.0,
-      "step": 31852
-    },
-    {
-      "epoch": 5.129715366963243,
-      "grad_norm": 0.0032435583416372538,
-      "learning_rate": 0.00019998702398086536,
-      "loss": 46.0,
-      "step": 31853
-    },
-    {
-      "epoch": 5.12987640404203,
-      "grad_norm": 0.005491221789270639,
-      "learning_rate": 0.0001999870231658704,
-      "loss": 46.0,
-      "step": 31854
-    },
-    {
-      "epoch": 5.130037441120818,
-      "grad_norm": 0.005614907015115023,
-      "learning_rate": 0.00019998702235084984,
-      "loss": 46.0,
-      "step": 31855
-    },
-    {
-      "epoch": 5.130198478199605,
-      "grad_norm": 0.0020053989719599485,
-      "learning_rate": 0.00019998702153580373,
-      "loss": 46.0,
-      "step": 31856
-    },
-    {
-      "epoch": 5.130359515278393,
-      "grad_norm": 0.0018744183471426368,
-      "learning_rate": 0.00019998702072073198,
-      "loss": 46.0,
-      "step": 31857
-    },
-    {
-      "epoch": 5.13052055235718,
-      "grad_norm": 0.015590890310704708,
-      "learning_rate": 0.00019998701990563467,
-      "loss": 46.0,
-      "step": 31858
-    },
-    {
-      "epoch": 5.130681589435968,
-      "grad_norm": 0.0010458175092935562,
-      "learning_rate": 0.00019998701909051175,
-      "loss": 46.0,
-      "step": 31859
-    },
-    {
-      "epoch": 5.130842626514755,
-      "grad_norm": 0.001125580514781177,
-      "learning_rate": 0.00019998701827536323,
-      "loss": 46.0,
-      "step": 31860
-    },
-    {
-      "epoch": 5.1310036635935425,
-      "grad_norm": 0.0012638133484870195,
-      "learning_rate": 0.00019998701746018913,
-      "loss": 46.0,
-      "step": 31861
-    },
-    {
-      "epoch": 5.13116470067233,
-      "grad_norm": 0.0016861631302163005,
-      "learning_rate": 0.00019998701664498944,
-      "loss": 46.0,
-      "step": 31862
-    },
-    {
-      "epoch": 5.131325737751117,
-      "grad_norm": 0.01926904357969761,
-      "learning_rate": 0.00019998701582976417,
-      "loss": 46.0,
-      "step": 31863
-    },
-    {
-      "epoch": 5.131486774829905,
-      "grad_norm": 0.005981662310659885,
-      "learning_rate": 0.00019998701501451328,
-      "loss": 46.0,
-      "step": 31864
-    },
-    {
-      "epoch": 5.131647811908692,
-      "grad_norm": 0.0062086321413517,
-      "learning_rate": 0.00019998701419923683,
-      "loss": 46.0,
-      "step": 31865
-    },
-    {
-      "epoch": 5.131808848987479,
-      "grad_norm": 0.006888432428240776,
-      "learning_rate": 0.00019998701338393477,
-      "loss": 46.0,
-      "step": 31866
-    },
-    {
-      "epoch": 5.131969886066266,
-      "grad_norm": 0.014934870414435863,
-      "learning_rate": 0.00019998701256860712,
-      "loss": 46.0,
-      "step": 31867
-    },
-    {
-      "epoch": 5.132130923145054,
-      "grad_norm": 0.007694324012845755,
-      "learning_rate": 0.00019998701175325388,
-      "loss": 46.0,
-      "step": 31868
-    },
-    {
-      "epoch": 5.132291960223841,
-      "grad_norm": 0.008090621791779995,
-      "learning_rate": 0.00019998701093787506,
-      "loss": 46.0,
-      "step": 31869
-    },
-    {
-      "epoch": 5.132452997302629,
-      "grad_norm": 0.0028628341387957335,
-      "learning_rate": 0.00019998701012247062,
-      "loss": 46.0,
-      "step": 31870
-    },
-    {
-      "epoch": 5.132614034381416,
-      "grad_norm": 0.005956295877695084,
-      "learning_rate": 0.00019998700930704062,
-      "loss": 46.0,
-      "step": 31871
-    },
-    {
-      "epoch": 5.132775071460204,
-      "grad_norm": 0.003277809824794531,
-      "learning_rate": 0.000199987008491585,
-      "loss": 46.0,
-      "step": 31872
-    },
-    {
-      "epoch": 5.132936108538991,
-      "grad_norm": 0.01019643247127533,
-      "learning_rate": 0.0001999870076761038,
-      "loss": 46.0,
-      "step": 31873
-    },
-    {
-      "epoch": 5.1330971456177785,
-      "grad_norm": 0.002147262915968895,
-      "learning_rate": 0.00019998700686059702,
-      "loss": 46.0,
-      "step": 31874
-    },
-    {
-      "epoch": 5.133258182696566,
-      "grad_norm": 0.0026059038937091827,
-      "learning_rate": 0.00019998700604506464,
-      "loss": 46.0,
-      "step": 31875
-    },
-    {
-      "epoch": 5.133419219775353,
-      "grad_norm": 0.003703092923387885,
-      "learning_rate": 0.00019998700522950666,
-      "loss": 46.0,
-      "step": 31876
-    },
-    {
-      "epoch": 5.133580256854141,
-      "grad_norm": 0.0031463506165891886,
-      "learning_rate": 0.0001999870044139231,
-      "loss": 46.0,
-      "step": 31877
-    },
-    {
-      "epoch": 5.133741293932928,
-      "grad_norm": 0.0009380585397593677,
-      "learning_rate": 0.00019998700359831394,
-      "loss": 46.0,
-      "step": 31878
-    },
-    {
-      "epoch": 5.133902331011716,
-      "grad_norm": 0.003604380413889885,
-      "learning_rate": 0.00019998700278267922,
-      "loss": 46.0,
-      "step": 31879
-    },
-    {
-      "epoch": 5.134063368090503,
-      "grad_norm": 0.022177448496222496,
-      "learning_rate": 0.00019998700196701888,
-      "loss": 46.0,
-      "step": 31880
-    },
-    {
-      "epoch": 5.13422440516929,
-      "grad_norm": 0.003096317872405052,
-      "learning_rate": 0.00019998700115133293,
-      "loss": 46.0,
-      "step": 31881
-    },
-    {
-      "epoch": 5.134385442248077,
-      "grad_norm": 0.007766911294311285,
-      "learning_rate": 0.00019998700033562142,
-      "loss": 46.0,
-      "step": 31882
-    },
-    {
-      "epoch": 5.134546479326865,
-      "grad_norm": 0.0021253577433526516,
-      "learning_rate": 0.0001999869995198843,
-      "loss": 46.0,
-      "step": 31883
-    },
-    {
-      "epoch": 5.134707516405652,
-      "grad_norm": 0.010086525231599808,
-      "learning_rate": 0.00019998699870412158,
-      "loss": 46.0,
-      "step": 31884
-    },
-    {
-      "epoch": 5.13486855348444,
-      "grad_norm": 0.0037321089766919613,
-      "learning_rate": 0.0001999869978883333,
-      "loss": 46.0,
-      "step": 31885
-    },
-    {
-      "epoch": 5.135029590563227,
-      "grad_norm": 0.0051113017834723,
-      "learning_rate": 0.0001999869970725194,
-      "loss": 46.0,
-      "step": 31886
-    },
-    {
-      "epoch": 5.1351906276420145,
-      "grad_norm": 0.004103382583707571,
-      "learning_rate": 0.00019998699625667991,
-      "loss": 46.0,
-      "step": 31887
-    },
-    {
-      "epoch": 5.135351664720802,
-      "grad_norm": 0.003515543881803751,
-      "learning_rate": 0.00019998699544081485,
-      "loss": 46.0,
-      "step": 31888
-    },
-    {
-      "epoch": 5.135512701799589,
-      "grad_norm": 0.002703124424442649,
-      "learning_rate": 0.0001999869946249242,
-      "loss": 46.0,
-      "step": 31889
-    },
-    {
-      "epoch": 5.135673738878377,
-      "grad_norm": 0.0009707590797916055,
-      "learning_rate": 0.00019998699380900794,
-      "loss": 46.0,
-      "step": 31890
-    },
-    {
-      "epoch": 5.135834775957164,
-      "grad_norm": 0.0026337311137467623,
-      "learning_rate": 0.0001999869929930661,
-      "loss": 46.0,
-      "step": 31891
-    },
-    {
-      "epoch": 5.135995813035952,
-      "grad_norm": 0.007333251181989908,
-      "learning_rate": 0.00019998699217709865,
-      "loss": 46.0,
-      "step": 31892
-    },
-    {
-      "epoch": 5.136156850114739,
-      "grad_norm": 0.010815009474754333,
-      "learning_rate": 0.00019998699136110563,
-      "loss": 46.0,
-      "step": 31893
-    },
-    {
-      "epoch": 5.136317887193527,
-      "grad_norm": 0.0009113884880207479,
-      "learning_rate": 0.00019998699054508704,
-      "loss": 46.0,
-      "step": 31894
-    },
-    {
-      "epoch": 5.136478924272314,
-      "grad_norm": 0.020153556019067764,
-      "learning_rate": 0.0001999869897290428,
-      "loss": 46.0,
-      "step": 31895
-    },
-    {
-      "epoch": 5.136639961351101,
-      "grad_norm": 0.003549706656485796,
-      "learning_rate": 0.00019998698891297297,
-      "loss": 46.0,
-      "step": 31896
-    },
-    {
-      "epoch": 5.136800998429888,
-      "grad_norm": 0.016075970605015755,
-      "learning_rate": 0.00019998698809687763,
-      "loss": 46.0,
-      "step": 31897
-    },
-    {
-      "epoch": 5.136962035508676,
-      "grad_norm": 0.003938252106308937,
-      "learning_rate": 0.0001999869872807566,
-      "loss": 46.0,
-      "step": 31898
-    },
-    {
-      "epoch": 5.137123072587463,
-      "grad_norm": 0.0031970986165106297,
-      "learning_rate": 0.00019998698646461004,
-      "loss": 46.0,
-      "step": 31899
-    },
-    {
-      "epoch": 5.1372841096662505,
-      "grad_norm": 0.005274391267448664,
-      "learning_rate": 0.00019998698564843787,
-      "loss": 46.0,
-      "step": 31900
-    },
-    {
-      "epoch": 5.137445146745038,
-      "grad_norm": 0.02253166027367115,
-      "learning_rate": 0.00019998698483224012,
-      "loss": 46.0,
-      "step": 31901
-    },
-    {
-      "epoch": 5.137606183823825,
-      "grad_norm": 0.004127433989197016,
-      "learning_rate": 0.00019998698401601676,
-      "loss": 46.0,
-      "step": 31902
-    },
-    {
-      "epoch": 5.137767220902613,
-      "grad_norm": 0.002368879271671176,
-      "learning_rate": 0.0001999869831997678,
-      "loss": 46.0,
-      "step": 31903
-    },
-    {
-      "epoch": 5.1379282579814,
-      "grad_norm": 0.0035266403574496508,
-      "learning_rate": 0.00019998698238349327,
-      "loss": 46.0,
-      "step": 31904
-    },
-    {
-      "epoch": 5.138089295060188,
-      "grad_norm": 0.009408049285411835,
-      "learning_rate": 0.00019998698156719315,
-      "loss": 46.0,
-      "step": 31905
-    },
-    {
-      "epoch": 5.138250332138975,
-      "grad_norm": 0.0008480396354570985,
-      "learning_rate": 0.00019998698075086743,
-      "loss": 46.0,
-      "step": 31906
-    },
-    {
-      "epoch": 5.138411369217763,
-      "grad_norm": 0.00807144958525896,
-      "learning_rate": 0.0001999869799345161,
-      "loss": 46.0,
-      "step": 31907
-    },
-    {
-      "epoch": 5.13857240629655,
-      "grad_norm": 0.009584181942045689,
-      "learning_rate": 0.00019998697911813922,
-      "loss": 46.0,
-      "step": 31908
-    },
-    {
-      "epoch": 5.1387334433753376,
-      "grad_norm": 0.015621276572346687,
-      "learning_rate": 0.00019998697830173675,
-      "loss": 46.0,
-      "step": 31909
-    },
-    {
-      "epoch": 5.138894480454125,
-      "grad_norm": 0.008341421373188496,
-      "learning_rate": 0.00019998697748530866,
-      "loss": 46.0,
-      "step": 31910
-    },
-    {
-      "epoch": 5.139055517532912,
-      "grad_norm": 0.0017939795507118106,
-      "learning_rate": 0.00019998697666885495,
-      "loss": 46.0,
-      "step": 31911
-    },
-    {
-      "epoch": 5.139216554611699,
-      "grad_norm": 0.003648452227935195,
-      "learning_rate": 0.00019998697585237572,
-      "loss": 46.0,
-      "step": 31912
-    },
-    {
-      "epoch": 5.1393775916904865,
-      "grad_norm": 0.002116371877491474,
-      "learning_rate": 0.00019998697503587084,
-      "loss": 46.0,
-      "step": 31913
-    },
-    {
-      "epoch": 5.139538628769274,
-      "grad_norm": 0.006568969693034887,
-      "learning_rate": 0.0001999869742193404,
-      "loss": 46.0,
-      "step": 31914
-    },
-    {
-      "epoch": 5.139699665848061,
-      "grad_norm": 0.007540897000581026,
-      "learning_rate": 0.00019998697340278433,
-      "loss": 46.0,
-      "step": 31915
-    },
-    {
-      "epoch": 5.139860702926849,
-      "grad_norm": 0.013246042653918266,
-      "learning_rate": 0.00019998697258620272,
-      "loss": 46.0,
-      "step": 31916
-    },
-    {
-      "epoch": 5.140021740005636,
-      "grad_norm": 0.0015266399132087827,
-      "learning_rate": 0.0001999869717695955,
-      "loss": 46.0,
-      "step": 31917
-    },
-    {
-      "epoch": 5.140182777084424,
-      "grad_norm": 0.012009619735181332,
-      "learning_rate": 0.00019998697095296265,
-      "loss": 46.0,
-      "step": 31918
-    },
-    {
-      "epoch": 5.140343814163211,
-      "grad_norm": 0.004252308048307896,
-      "learning_rate": 0.00019998697013630428,
-      "loss": 46.0,
-      "step": 31919
-    },
-    {
-      "epoch": 5.140504851241999,
-      "grad_norm": 0.01128622516989708,
-      "learning_rate": 0.00019998696931962026,
-      "loss": 46.0,
-      "step": 31920
-    },
-    {
-      "epoch": 5.140665888320786,
-      "grad_norm": 0.002854184480383992,
-      "learning_rate": 0.0001999869685029107,
-      "loss": 46.0,
-      "step": 31921
-    },
-    {
-      "epoch": 5.1408269253995735,
-      "grad_norm": 0.0050996169447898865,
-      "learning_rate": 0.00019998696768617547,
-      "loss": 46.0,
-      "step": 31922
-    },
-    {
-      "epoch": 5.140987962478361,
-      "grad_norm": 0.0017365822568535805,
-      "learning_rate": 0.00019998696686941472,
-      "loss": 46.0,
-      "step": 31923
-    },
-    {
-      "epoch": 5.141148999557148,
-      "grad_norm": 0.0038486134726554155,
-      "learning_rate": 0.00019998696605262833,
-      "loss": 46.0,
-      "step": 31924
-    },
-    {
-      "epoch": 5.141310036635935,
-      "grad_norm": 0.0014324215007945895,
-      "learning_rate": 0.00019998696523581638,
-      "loss": 46.0,
-      "step": 31925
-    },
-    {
-      "epoch": 5.1414710737147225,
-      "grad_norm": 0.003406471572816372,
-      "learning_rate": 0.00019998696441897885,
-      "loss": 46.0,
-      "step": 31926
-    },
-    {
-      "epoch": 5.14163211079351,
-      "grad_norm": 0.010486254468560219,
-      "learning_rate": 0.0001999869636021157,
-      "loss": 46.0,
-      "step": 31927
-    },
-    {
-      "epoch": 5.141793147872297,
-      "grad_norm": 0.0025951347779482603,
-      "learning_rate": 0.00019998696278522698,
-      "loss": 46.0,
-      "step": 31928
-    },
-    {
-      "epoch": 5.141954184951085,
-      "grad_norm": 0.005481914151459932,
-      "learning_rate": 0.00019998696196831266,
-      "loss": 46.0,
-      "step": 31929
-    },
-    {
-      "epoch": 5.142115222029872,
-      "grad_norm": 0.0021850187331438065,
-      "learning_rate": 0.00019998696115137272,
-      "loss": 46.0,
-      "step": 31930
-    },
-    {
-      "epoch": 5.14227625910866,
-      "grad_norm": 0.0020159154664725065,
-      "learning_rate": 0.00019998696033440722,
-      "loss": 46.0,
-      "step": 31931
-    },
-    {
-      "epoch": 5.142437296187447,
-      "grad_norm": 0.008351726457476616,
-      "learning_rate": 0.00019998695951741613,
-      "loss": 46.0,
-      "step": 31932
-    },
-    {
-      "epoch": 5.142598333266235,
-      "grad_norm": 0.0037066920194774866,
-      "learning_rate": 0.00019998695870039943,
-      "loss": 46.0,
-      "step": 31933
-    },
-    {
-      "epoch": 5.142759370345022,
-      "grad_norm": 0.004341783933341503,
-      "learning_rate": 0.00019998695788335714,
-      "loss": 46.0,
-      "step": 31934
-    },
-    {
-      "epoch": 5.1429204074238095,
-      "grad_norm": 0.015800854191184044,
-      "learning_rate": 0.0001999869570662893,
-      "loss": 46.0,
-      "step": 31935
-    },
-    {
-      "epoch": 5.143081444502597,
-      "grad_norm": 0.003397870808839798,
-      "learning_rate": 0.00019998695624919583,
-      "loss": 46.0,
-      "step": 31936
-    },
-    {
-      "epoch": 5.143242481581384,
-      "grad_norm": 0.0027612776029855013,
-      "learning_rate": 0.00019998695543207675,
-      "loss": 46.0,
-      "step": 31937
-    },
-    {
-      "epoch": 5.143403518660172,
-      "grad_norm": 0.008857367560267448,
-      "learning_rate": 0.0001999869546149321,
-      "loss": 46.0,
-      "step": 31938
-    },
-    {
-      "epoch": 5.143564555738959,
-      "grad_norm": 0.003905584802851081,
-      "learning_rate": 0.0001999869537977619,
-      "loss": 46.0,
-      "step": 31939
-    },
-    {
-      "epoch": 5.143725592817746,
-      "grad_norm": 0.008554146625101566,
-      "learning_rate": 0.00019998695298056605,
-      "loss": 46.0,
-      "step": 31940
-    },
-    {
-      "epoch": 5.143886629896533,
-      "grad_norm": 0.003602387849241495,
-      "learning_rate": 0.00019998695216334462,
-      "loss": 46.0,
-      "step": 31941
-    },
-    {
-      "epoch": 5.144047666975321,
-      "grad_norm": 0.00845616776496172,
-      "learning_rate": 0.0001999869513460976,
-      "loss": 46.0,
-      "step": 31942
-    },
-    {
-      "epoch": 5.144208704054108,
-      "grad_norm": 0.007769668009132147,
-      "learning_rate": 0.00019998695052882498,
-      "loss": 46.0,
-      "step": 31943
-    },
-    {
-      "epoch": 5.144369741132896,
-      "grad_norm": 0.013131647370755672,
-      "learning_rate": 0.0001999869497115268,
-      "loss": 46.0,
-      "step": 31944
-    },
-    {
-      "epoch": 5.144530778211683,
-      "grad_norm": 0.014707699418067932,
-      "learning_rate": 0.00019998694889420302,
-      "loss": 46.0,
-      "step": 31945
-    },
-    {
-      "epoch": 5.144691815290471,
-      "grad_norm": 0.002883515553548932,
-      "learning_rate": 0.00019998694807685366,
-      "loss": 46.0,
-      "step": 31946
-    },
-    {
-      "epoch": 5.144852852369258,
-      "grad_norm": 0.006109686568379402,
-      "learning_rate": 0.00019998694725947865,
-      "loss": 46.0,
-      "step": 31947
-    },
-    {
-      "epoch": 5.1450138894480455,
-      "grad_norm": 0.0022044205106794834,
-      "learning_rate": 0.0001999869464420781,
-      "loss": 46.0,
-      "step": 31948
-    },
-    {
-      "epoch": 5.145174926526833,
-      "grad_norm": 0.0054174126125872135,
-      "learning_rate": 0.00019998694562465197,
-      "loss": 46.0,
-      "step": 31949
-    },
-    {
-      "epoch": 5.14533596360562,
-      "grad_norm": 0.002054027281701565,
-      "learning_rate": 0.00019998694480720023,
-      "loss": 46.0,
-      "step": 31950
-    },
-    {
-      "epoch": 5.145497000684408,
-      "grad_norm": 0.0015779273817315698,
-      "learning_rate": 0.00019998694398972288,
-      "loss": 46.0,
-      "step": 31951
-    },
-    {
-      "epoch": 5.145658037763195,
-      "grad_norm": 0.001287986640818417,
-      "learning_rate": 0.00019998694317221994,
-      "loss": 46.0,
-      "step": 31952
-    },
-    {
-      "epoch": 5.145819074841983,
-      "grad_norm": 0.003587508574128151,
-      "learning_rate": 0.0001999869423546914,
-      "loss": 46.0,
-      "step": 31953
-    },
-    {
-      "epoch": 5.145980111920769,
-      "grad_norm": 0.0014130700146779418,
-      "learning_rate": 0.0001999869415371373,
-      "loss": 46.0,
-      "step": 31954
-    },
-    {
-      "epoch": 5.146141148999557,
-      "grad_norm": 0.010049944743514061,
-      "learning_rate": 0.0001999869407195576,
-      "loss": 46.0,
-      "step": 31955
-    },
-    {
-      "epoch": 5.146302186078344,
-      "grad_norm": 0.0018803478451445699,
-      "learning_rate": 0.0001999869399019523,
-      "loss": 46.0,
-      "step": 31956
-    },
-    {
-      "epoch": 5.146463223157132,
-      "grad_norm": 0.010292558930814266,
-      "learning_rate": 0.00019998693908432143,
-      "loss": 46.0,
-      "step": 31957
-    },
-    {
-      "epoch": 5.146624260235919,
-      "grad_norm": 0.004415625240653753,
-      "learning_rate": 0.00019998693826666494,
-      "loss": 46.0,
-      "step": 31958
-    },
-    {
-      "epoch": 5.146785297314707,
-      "grad_norm": 0.0029766466468572617,
-      "learning_rate": 0.00019998693744898287,
-      "loss": 46.0,
-      "step": 31959
-    },
-    {
-      "epoch": 5.146946334393494,
-      "grad_norm": 0.009527206420898438,
-      "learning_rate": 0.0001999869366312752,
-      "loss": 46.0,
-      "step": 31960
-    },
-    {
-      "epoch": 5.1471073714722815,
-      "grad_norm": 0.0058088358491659164,
-      "learning_rate": 0.00019998693581354195,
-      "loss": 46.0,
-      "step": 31961
-    },
-    {
-      "epoch": 5.147268408551069,
-      "grad_norm": 0.003082311013713479,
-      "learning_rate": 0.00019998693499578311,
-      "loss": 46.0,
-      "step": 31962
-    },
-    {
-      "epoch": 5.147429445629856,
-      "grad_norm": 0.007802627515047789,
-      "learning_rate": 0.00019998693417799866,
-      "loss": 46.0,
-      "step": 31963
-    },
-    {
-      "epoch": 5.147590482708644,
-      "grad_norm": 0.007802153471857309,
-      "learning_rate": 0.00019998693336018865,
-      "loss": 46.0,
-      "step": 31964
-    },
-    {
-      "epoch": 5.147751519787431,
-      "grad_norm": 0.0027400145772844553,
-      "learning_rate": 0.00019998693254235302,
-      "loss": 46.0,
-      "step": 31965
-    },
-    {
-      "epoch": 5.147912556866219,
-      "grad_norm": 0.0021923664025962353,
-      "learning_rate": 0.00019998693172449184,
-      "loss": 46.0,
-      "step": 31966
-    },
-    {
-      "epoch": 5.148073593945006,
-      "grad_norm": 0.0036887081805616617,
-      "learning_rate": 0.000199986930906605,
-      "loss": 46.0,
-      "step": 31967
-    },
-    {
-      "epoch": 5.148234631023794,
-      "grad_norm": 0.001792989089153707,
-      "learning_rate": 0.00019998693008869262,
-      "loss": 46.0,
-      "step": 31968
-    },
-    {
-      "epoch": 5.14839566810258,
-      "grad_norm": 0.010252018459141254,
-      "learning_rate": 0.00019998692927075464,
-      "loss": 46.0,
-      "step": 31969
-    },
-    {
-      "epoch": 5.148556705181368,
-      "grad_norm": 0.002269984222948551,
-      "learning_rate": 0.00019998692845279105,
-      "loss": 46.0,
-      "step": 31970
-    },
-    {
-      "epoch": 5.148717742260155,
-      "grad_norm": 0.002541786292567849,
-      "learning_rate": 0.0001999869276348019,
-      "loss": 46.0,
-      "step": 31971
-    },
-    {
-      "epoch": 5.148878779338943,
-      "grad_norm": 0.0016420090105384588,
-      "learning_rate": 0.00019998692681678714,
-      "loss": 46.0,
-      "step": 31972
-    },
-    {
-      "epoch": 5.14903981641773,
-      "grad_norm": 0.004215634427964687,
-      "learning_rate": 0.0001999869259987468,
-      "loss": 46.0,
-      "step": 31973
-    },
-    {
-      "epoch": 5.1492008534965175,
-      "grad_norm": 0.013805377297103405,
-      "learning_rate": 0.00019998692518068085,
-      "loss": 46.0,
-      "step": 31974
-    },
-    {
-      "epoch": 5.149361890575305,
-      "grad_norm": 0.006287030875682831,
-      "learning_rate": 0.00019998692436258933,
-      "loss": 46.0,
-      "step": 31975
-    },
-    {
-      "epoch": 5.149522927654092,
-      "grad_norm": 0.006337171886116266,
-      "learning_rate": 0.00019998692354447219,
-      "loss": 46.0,
-      "step": 31976
-    },
-    {
-      "epoch": 5.14968396473288,
-      "grad_norm": 0.0035442980006337166,
-      "learning_rate": 0.0001999869227263295,
-      "loss": 46.0,
-      "step": 31977
-    },
-    {
-      "epoch": 5.149845001811667,
-      "grad_norm": 0.004764437209814787,
-      "learning_rate": 0.00019998692190816115,
-      "loss": 46.0,
-      "step": 31978
-    },
-    {
-      "epoch": 5.150006038890455,
-      "grad_norm": 0.011036619544029236,
-      "learning_rate": 0.00019998692108996727,
-      "loss": 46.0,
-      "step": 31979
-    },
-    {
-      "epoch": 5.150167075969242,
-      "grad_norm": 0.0048725781962275505,
-      "learning_rate": 0.00019998692027174776,
-      "loss": 46.0,
-      "step": 31980
-    },
-    {
-      "epoch": 5.15032811304803,
-      "grad_norm": 0.002315012738108635,
-      "learning_rate": 0.00019998691945350268,
-      "loss": 46.0,
-      "step": 31981
-    },
-    {
-      "epoch": 5.150489150126817,
-      "grad_norm": 0.009306208230555058,
-      "learning_rate": 0.00019998691863523202,
-      "loss": 46.0,
-      "step": 31982
-    },
-    {
-      "epoch": 5.1506501872056045,
-      "grad_norm": 0.0018951231613755226,
-      "learning_rate": 0.00019998691781693574,
-      "loss": 46.0,
-      "step": 31983
-    },
-    {
-      "epoch": 5.150811224284391,
-      "grad_norm": 0.02272212691605091,
-      "learning_rate": 0.0001999869169986139,
-      "loss": 46.0,
-      "step": 31984
-    },
-    {
-      "epoch": 5.1509722613631785,
-      "grad_norm": 0.0015805884031578898,
-      "learning_rate": 0.00019998691618026642,
-      "loss": 46.0,
-      "step": 31985
-    },
-    {
-      "epoch": 5.151133298441966,
-      "grad_norm": 0.009673546068370342,
-      "learning_rate": 0.0001999869153618934,
-      "loss": 46.0,
-      "step": 31986
-    },
-    {
-      "epoch": 5.151294335520753,
-      "grad_norm": 0.017125146463513374,
-      "learning_rate": 0.00019998691454349476,
-      "loss": 46.0,
-      "step": 31987
-    },
-    {
-      "epoch": 5.151455372599541,
-      "grad_norm": 0.007747529074549675,
-      "learning_rate": 0.00019998691372507055,
-      "loss": 46.0,
-      "step": 31988
-    },
-    {
-      "epoch": 5.151616409678328,
-      "grad_norm": 0.0019121356308460236,
-      "learning_rate": 0.00019998691290662072,
-      "loss": 46.0,
-      "step": 31989
-    },
-    {
-      "epoch": 5.151777446757116,
-      "grad_norm": 0.020110275596380234,
-      "learning_rate": 0.00019998691208814533,
-      "loss": 46.0,
-      "step": 31990
-    },
-    {
-      "epoch": 5.151938483835903,
-      "grad_norm": 0.03902021422982216,
-      "learning_rate": 0.00019998691126964433,
-      "loss": 46.0,
-      "step": 31991
-    },
-    {
-      "epoch": 5.152099520914691,
-      "grad_norm": 0.003365870099514723,
-      "learning_rate": 0.00019998691045111774,
-      "loss": 46.0,
-      "step": 31992
-    },
-    {
-      "epoch": 5.152260557993478,
-      "grad_norm": 0.0011135706445202231,
-      "learning_rate": 0.00019998690963256554,
-      "loss": 46.0,
-      "step": 31993
-    },
-    {
-      "epoch": 5.152421595072266,
-      "grad_norm": 0.003256438300013542,
-      "learning_rate": 0.00019998690881398778,
-      "loss": 46.0,
-      "step": 31994
-    },
-    {
-      "epoch": 5.152582632151053,
-      "grad_norm": 0.007314857095479965,
-      "learning_rate": 0.00019998690799538443,
-      "loss": 46.0,
-      "step": 31995
-    },
-    {
-      "epoch": 5.1527436692298405,
-      "grad_norm": 0.0039177159778773785,
-      "learning_rate": 0.00019998690717675546,
-      "loss": 46.0,
-      "step": 31996
-    },
-    {
-      "epoch": 5.152904706308628,
-      "grad_norm": 0.0039351144805550575,
-      "learning_rate": 0.0001999869063581009,
-      "loss": 46.0,
-      "step": 31997
-    },
-    {
-      "epoch": 5.153065743387415,
-      "grad_norm": 0.005802639760077,
-      "learning_rate": 0.0001999869055394208,
-      "loss": 46.0,
-      "step": 31998
-    },
-    {
-      "epoch": 5.153226780466202,
-      "grad_norm": 0.0011089701438322663,
-      "learning_rate": 0.00019998690472071504,
-      "loss": 46.0,
-      "step": 31999
-    },
-    {
-      "epoch": 5.153387817544989,
-      "grad_norm": 0.007191171403974295,
-      "learning_rate": 0.0001999869039019837,
-      "loss": 46.0,
-      "step": 32000
-    },
-    {
-      "epoch": 5.153548854623777,
-      "grad_norm": 0.0034574205055832863,
-      "learning_rate": 0.0001999869030832268,
-      "loss": 46.0,
-      "step": 32001
-    },
-    {
-      "epoch": 5.153709891702564,
-      "grad_norm": 0.0017519757384434342,
-      "learning_rate": 0.0001999869022644443,
-      "loss": 46.0,
-      "step": 32002
-    },
-    {
-      "epoch": 5.153870928781352,
-      "grad_norm": 0.002987498650327325,
-      "learning_rate": 0.00019998690144563621,
-      "loss": 46.0,
-      "step": 32003
-    },
-    {
-      "epoch": 5.154031965860139,
-      "grad_norm": 0.007787894457578659,
-      "learning_rate": 0.00019998690062680253,
-      "loss": 46.0,
-      "step": 32004
-    },
-    {
-      "epoch": 5.154193002938927,
-      "grad_norm": 0.0026969073805958033,
-      "learning_rate": 0.00019998689980794322,
-      "loss": 46.0,
-      "step": 32005
-    },
-    {
-      "epoch": 5.154354040017714,
-      "grad_norm": 0.003432060359045863,
-      "learning_rate": 0.00019998689898905836,
-      "loss": 46.0,
-      "step": 32006
-    },
-    {
-      "epoch": 5.154515077096502,
-      "grad_norm": 0.006968228612095118,
-      "learning_rate": 0.0001999868981701479,
-      "loss": 46.0,
-      "step": 32007
-    },
-    {
-      "epoch": 5.154676114175289,
-      "grad_norm": 0.002099679782986641,
-      "learning_rate": 0.00019998689735121184,
-      "loss": 46.0,
-      "step": 32008
-    },
-    {
-      "epoch": 5.1548371512540765,
-      "grad_norm": 0.011155657470226288,
-      "learning_rate": 0.0001999868965322502,
-      "loss": 46.0,
-      "step": 32009
-    },
-    {
-      "epoch": 5.154998188332864,
-      "grad_norm": 0.002716090762987733,
-      "learning_rate": 0.00019998689571326298,
-      "loss": 46.0,
-      "step": 32010
-    },
-    {
-      "epoch": 5.155159225411651,
-      "grad_norm": 0.006175829563289881,
-      "learning_rate": 0.00019998689489425013,
-      "loss": 46.0,
-      "step": 32011
-    },
-    {
-      "epoch": 5.155320262490439,
-      "grad_norm": 0.019180694594979286,
-      "learning_rate": 0.0001999868940752117,
-      "loss": 46.0,
-      "step": 32012
-    },
-    {
-      "epoch": 5.155481299569225,
-      "grad_norm": 0.0019575918558984995,
-      "learning_rate": 0.0001999868932561477,
-      "loss": 46.0,
-      "step": 32013
-    },
-    {
-      "epoch": 5.155642336648013,
-      "grad_norm": 0.0014998972183093429,
-      "learning_rate": 0.0001999868924370581,
-      "loss": 46.0,
-      "step": 32014
-    },
-    {
-      "epoch": 5.1558033737268,
-      "grad_norm": 0.0029361594934016466,
-      "learning_rate": 0.0001999868916179429,
-      "loss": 46.0,
-      "step": 32015
-    },
-    {
-      "epoch": 5.155964410805588,
-      "grad_norm": 0.001516103744506836,
-      "learning_rate": 0.00019998689079880213,
-      "loss": 46.0,
-      "step": 32016
-    },
-    {
-      "epoch": 5.156125447884375,
-      "grad_norm": 0.002657112665474415,
-      "learning_rate": 0.00019998688997963576,
-      "loss": 46.0,
-      "step": 32017
-    },
-    {
-      "epoch": 5.156286484963163,
-      "grad_norm": 0.0008394195465371013,
-      "learning_rate": 0.00019998688916044377,
-      "loss": 46.0,
-      "step": 32018
-    },
-    {
-      "epoch": 5.15644752204195,
-      "grad_norm": 0.008360626175999641,
-      "learning_rate": 0.00019998688834122624,
-      "loss": 46.0,
-      "step": 32019
-    },
-    {
-      "epoch": 5.156608559120738,
-      "grad_norm": 0.0073290071450173855,
-      "learning_rate": 0.00019998688752198308,
-      "loss": 46.0,
-      "step": 32020
-    },
-    {
-      "epoch": 5.156769596199525,
-      "grad_norm": 0.0031881213653832674,
-      "learning_rate": 0.00019998688670271432,
-      "loss": 46.0,
-      "step": 32021
-    },
-    {
-      "epoch": 5.1569306332783125,
-      "grad_norm": 0.006068927701562643,
-      "learning_rate": 0.00019998688588341999,
-      "loss": 46.0,
-      "step": 32022
-    },
-    {
-      "epoch": 5.1570916703571,
-      "grad_norm": 0.012644579634070396,
-      "learning_rate": 0.00019998688506410006,
-      "loss": 46.0,
-      "step": 32023
-    },
-    {
-      "epoch": 5.157252707435887,
-      "grad_norm": 0.0051238052546978,
-      "learning_rate": 0.00019998688424475454,
-      "loss": 46.0,
-      "step": 32024
-    },
-    {
-      "epoch": 5.157413744514675,
-      "grad_norm": 0.0034228686708956957,
-      "learning_rate": 0.00019998688342538344,
-      "loss": 46.0,
-      "step": 32025
-    },
-    {
-      "epoch": 5.157574781593462,
-      "grad_norm": 0.0034452243708074093,
-      "learning_rate": 0.00019998688260598673,
-      "loss": 46.0,
-      "step": 32026
-    },
-    {
-      "epoch": 5.157735818672249,
-      "grad_norm": 0.0025913238059729338,
-      "learning_rate": 0.00019998688178656443,
-      "loss": 46.0,
-      "step": 32027
-    },
-    {
-      "epoch": 5.157896855751036,
-      "grad_norm": 0.006521561648696661,
-      "learning_rate": 0.00019998688096711656,
-      "loss": 46.0,
-      "step": 32028
-    },
-    {
-      "epoch": 5.158057892829824,
-      "grad_norm": 0.0017765399534255266,
-      "learning_rate": 0.0001999868801476431,
-      "loss": 46.0,
-      "step": 32029
-    },
-    {
-      "epoch": 5.158218929908611,
-      "grad_norm": 0.005090164951980114,
-      "learning_rate": 0.000199986879328144,
-      "loss": 46.0,
-      "step": 32030
-    },
-    {
-      "epoch": 5.158379966987399,
-      "grad_norm": 0.0036904604639858007,
-      "learning_rate": 0.00019998687850861935,
-      "loss": 46.0,
-      "step": 32031
-    },
-    {
-      "epoch": 5.158541004066186,
-      "grad_norm": 0.0014547455357387662,
-      "learning_rate": 0.0001999868776890691,
-      "loss": 46.0,
-      "step": 32032
-    },
-    {
-      "epoch": 5.1587020411449735,
-      "grad_norm": 0.0015070633962750435,
-      "learning_rate": 0.00019998687686949325,
-      "loss": 46.0,
-      "step": 32033
-    },
-    {
-      "epoch": 5.158863078223761,
-      "grad_norm": 0.0021063510794192553,
-      "learning_rate": 0.00019998687604989184,
-      "loss": 46.0,
-      "step": 32034
-    },
-    {
-      "epoch": 5.1590241153025485,
-      "grad_norm": 0.001814411603845656,
-      "learning_rate": 0.00019998687523026482,
-      "loss": 46.0,
-      "step": 32035
-    },
-    {
-      "epoch": 5.159185152381336,
-      "grad_norm": 0.003035438247025013,
-      "learning_rate": 0.0001999868744106122,
-      "loss": 46.0,
-      "step": 32036
-    },
-    {
-      "epoch": 5.159346189460123,
-      "grad_norm": 0.007734688930213451,
-      "learning_rate": 0.00019998687359093397,
-      "loss": 46.0,
-      "step": 32037
-    },
-    {
-      "epoch": 5.159507226538911,
-      "grad_norm": 0.008473081514239311,
-      "learning_rate": 0.00019998687277123018,
-      "loss": 46.0,
-      "step": 32038
-    },
-    {
-      "epoch": 5.159668263617698,
-      "grad_norm": 0.0015391348861157894,
-      "learning_rate": 0.00019998687195150078,
-      "loss": 46.0,
-      "step": 32039
-    },
-    {
-      "epoch": 5.159829300696486,
-      "grad_norm": 0.004602321423590183,
-      "learning_rate": 0.00019998687113174582,
-      "loss": 46.0,
-      "step": 32040
-    },
-    {
-      "epoch": 5.159990337775273,
-      "grad_norm": 0.0052553098648786545,
-      "learning_rate": 0.00019998687031196522,
-      "loss": 46.0,
-      "step": 32041
-    },
-    {
-      "epoch": 5.16015137485406,
-      "grad_norm": 0.0014426345005631447,
-      "learning_rate": 0.00019998686949215905,
-      "loss": 46.0,
-      "step": 32042
-    },
-    {
-      "epoch": 5.160312411932847,
-      "grad_norm": 0.0019206246361136436,
-      "learning_rate": 0.00019998686867232733,
-      "loss": 46.0,
-      "step": 32043
-    },
-    {
-      "epoch": 5.160473449011635,
-      "grad_norm": 0.0025787819176912308,
-      "learning_rate": 0.00019998686785246996,
-      "loss": 46.0,
-      "step": 32044
-    },
-    {
-      "epoch": 5.160634486090422,
-      "grad_norm": 0.01073532085865736,
-      "learning_rate": 0.000199986867032587,
-      "loss": 46.0,
-      "step": 32045
-    },
-    {
-      "epoch": 5.1607955231692095,
-      "grad_norm": 0.0022140785586088896,
-      "learning_rate": 0.00019998686621267847,
-      "loss": 46.0,
-      "step": 32046
-    },
-    {
-      "epoch": 5.160956560247997,
-      "grad_norm": 0.003601910313591361,
-      "learning_rate": 0.00019998686539274437,
-      "loss": 46.0,
-      "step": 32047
-    },
-    {
-      "epoch": 5.161117597326784,
-      "grad_norm": 0.007175914477556944,
-      "learning_rate": 0.00019998686457278465,
-      "loss": 46.0,
-      "step": 32048
-    },
-    {
-      "epoch": 5.161278634405572,
-      "grad_norm": 0.004147344268858433,
-      "learning_rate": 0.00019998686375279933,
-      "loss": 46.0,
-      "step": 32049
-    },
-    {
-      "epoch": 5.161439671484359,
-      "grad_norm": 0.004679791163653135,
-      "learning_rate": 0.00019998686293278844,
-      "loss": 46.0,
-      "step": 32050
-    },
-    {
-      "epoch": 5.161600708563147,
-      "grad_norm": 0.001714741694740951,
-      "learning_rate": 0.00019998686211275196,
-      "loss": 46.0,
-      "step": 32051
-    },
-    {
-      "epoch": 5.161761745641934,
-      "grad_norm": 0.0049721552059054375,
-      "learning_rate": 0.0001999868612926899,
-      "loss": 46.0,
-      "step": 32052
-    },
-    {
-      "epoch": 5.161922782720722,
-      "grad_norm": 0.033338431268930435,
-      "learning_rate": 0.0001999868604726022,
-      "loss": 46.0,
-      "step": 32053
-    },
-    {
-      "epoch": 5.162083819799509,
-      "grad_norm": 0.005200261250138283,
-      "learning_rate": 0.00019998685965248893,
-      "loss": 46.0,
-      "step": 32054
-    },
-    {
-      "epoch": 5.162244856878297,
-      "grad_norm": 0.019763480871915817,
-      "learning_rate": 0.00019998685883235008,
-      "loss": 46.0,
-      "step": 32055
-    },
-    {
-      "epoch": 5.162405893957084,
-      "grad_norm": 0.008143062703311443,
-      "learning_rate": 0.00019998685801218564,
-      "loss": 46.0,
-      "step": 32056
-    },
-    {
-      "epoch": 5.162566931035871,
-      "grad_norm": 0.007820661179721355,
-      "learning_rate": 0.0001999868571919956,
-      "loss": 46.0,
-      "step": 32057
-    },
-    {
-      "epoch": 5.162727968114658,
-      "grad_norm": 0.007466123439371586,
-      "learning_rate": 0.00019998685637177997,
-      "loss": 46.0,
-      "step": 32058
-    },
-    {
-      "epoch": 5.1628890051934455,
-      "grad_norm": 0.00197614636272192,
-      "learning_rate": 0.00019998685555153872,
-      "loss": 46.0,
-      "step": 32059
-    },
-    {
-      "epoch": 5.163050042272233,
-      "grad_norm": 0.0017582852160558105,
-      "learning_rate": 0.00019998685473127193,
-      "loss": 46.0,
-      "step": 32060
-    },
-    {
-      "epoch": 5.16321107935102,
-      "grad_norm": 0.002587753813713789,
-      "learning_rate": 0.00019998685391097953,
-      "loss": 46.0,
-      "step": 32061
-    },
-    {
-      "epoch": 5.163372116429808,
-      "grad_norm": 0.0031756702810525894,
-      "learning_rate": 0.00019998685309066154,
-      "loss": 46.0,
-      "step": 32062
-    },
-    {
-      "epoch": 5.163533153508595,
-      "grad_norm": 0.0023196216206997633,
-      "learning_rate": 0.00019998685227031794,
-      "loss": 46.0,
-      "step": 32063
-    },
-    {
-      "epoch": 5.163694190587383,
-      "grad_norm": 0.004183600191026926,
-      "learning_rate": 0.00019998685144994877,
-      "loss": 46.0,
-      "step": 32064
-    },
-    {
-      "epoch": 5.16385522766617,
-      "grad_norm": 0.004171449225395918,
-      "learning_rate": 0.000199986850629554,
-      "loss": 46.0,
-      "step": 32065
-    },
-    {
-      "epoch": 5.164016264744958,
-      "grad_norm": 0.003592939116060734,
-      "learning_rate": 0.00019998684980913366,
-      "loss": 46.0,
-      "step": 32066
-    },
-    {
-      "epoch": 5.164177301823745,
-      "grad_norm": 0.006181646138429642,
-      "learning_rate": 0.00019998684898868768,
-      "loss": 46.0,
-      "step": 32067
-    },
-    {
-      "epoch": 5.164338338902533,
-      "grad_norm": 0.0017797845648601651,
-      "learning_rate": 0.00019998684816821617,
-      "loss": 46.0,
-      "step": 32068
-    },
-    {
-      "epoch": 5.16449937598132,
-      "grad_norm": 0.008590904995799065,
-      "learning_rate": 0.00019998684734771901,
-      "loss": 46.0,
-      "step": 32069
-    },
-    {
-      "epoch": 5.1646604130601075,
-      "grad_norm": 0.006853925529867411,
-      "learning_rate": 0.00019998684652719627,
-      "loss": 46.0,
-      "step": 32070
-    },
-    {
-      "epoch": 5.164821450138895,
-      "grad_norm": 0.006140245124697685,
-      "learning_rate": 0.00019998684570664797,
-      "loss": 46.0,
-      "step": 32071
-    },
-    {
-      "epoch": 5.1649824872176815,
-      "grad_norm": 0.004573285114020109,
-      "learning_rate": 0.00019998684488607403,
-      "loss": 46.0,
-      "step": 32072
-    },
-    {
-      "epoch": 5.165143524296469,
-      "grad_norm": 0.010456668213009834,
-      "learning_rate": 0.00019998684406547453,
-      "loss": 46.0,
-      "step": 32073
-    },
-    {
-      "epoch": 5.165304561375256,
-      "grad_norm": 0.004685725551098585,
-      "learning_rate": 0.00019998684324484947,
-      "loss": 46.0,
-      "step": 32074
-    },
-    {
-      "epoch": 5.165465598454044,
-      "grad_norm": 0.009032877162098885,
-      "learning_rate": 0.00019998684242419876,
-      "loss": 46.0,
-      "step": 32075
-    },
-    {
-      "epoch": 5.165626635532831,
-      "grad_norm": 0.010874249041080475,
-      "learning_rate": 0.00019998684160352247,
-      "loss": 46.0,
-      "step": 32076
-    },
-    {
-      "epoch": 5.165787672611619,
-      "grad_norm": 0.0025444519706070423,
-      "learning_rate": 0.00019998684078282062,
-      "loss": 46.0,
-      "step": 32077
-    },
-    {
-      "epoch": 5.165948709690406,
-      "grad_norm": 0.0036635638680309057,
-      "learning_rate": 0.00019998683996209319,
-      "loss": 46.0,
-      "step": 32078
-    },
-    {
-      "epoch": 5.166109746769194,
-      "grad_norm": 0.012037809938192368,
-      "learning_rate": 0.0001999868391413401,
-      "loss": 46.0,
-      "step": 32079
-    },
-    {
-      "epoch": 5.166270783847981,
-      "grad_norm": 0.002927191089838743,
-      "learning_rate": 0.00019998683832056147,
-      "loss": 46.0,
-      "step": 32080
-    },
-    {
-      "epoch": 5.166431820926769,
-      "grad_norm": 0.01672675646841526,
-      "learning_rate": 0.00019998683749975724,
-      "loss": 46.0,
-      "step": 32081
-    },
-    {
-      "epoch": 5.166592858005556,
-      "grad_norm": 0.002120276214554906,
-      "learning_rate": 0.00019998683667892743,
-      "loss": 46.0,
-      "step": 32082
-    },
-    {
-      "epoch": 5.1667538950843435,
-      "grad_norm": 0.0024933675304055214,
-      "learning_rate": 0.000199986835858072,
-      "loss": 46.0,
-      "step": 32083
-    },
-    {
-      "epoch": 5.166914932163131,
-      "grad_norm": 0.0028369955252856016,
-      "learning_rate": 0.00019998683503719098,
-      "loss": 46.0,
-      "step": 32084
-    },
-    {
-      "epoch": 5.167075969241918,
-      "grad_norm": 0.024100981652736664,
-      "learning_rate": 0.00019998683421628438,
-      "loss": 46.0,
-      "step": 32085
-    },
-    {
-      "epoch": 5.167237006320705,
-      "grad_norm": 0.0043613119050860405,
-      "learning_rate": 0.0001999868333953522,
-      "loss": 46.0,
-      "step": 32086
-    },
-    {
-      "epoch": 5.167398043399492,
-      "grad_norm": 0.007613196037709713,
-      "learning_rate": 0.00019998683257439442,
-      "loss": 46.0,
-      "step": 32087
-    },
-    {
-      "epoch": 5.16755908047828,
-      "grad_norm": 0.005537358112633228,
-      "learning_rate": 0.00019998683175341105,
-      "loss": 46.0,
-      "step": 32088
-    },
-    {
-      "epoch": 5.167720117557067,
-      "grad_norm": 0.00807643961161375,
-      "learning_rate": 0.00019998683093240207,
-      "loss": 46.0,
-      "step": 32089
-    },
-    {
-      "epoch": 5.167881154635855,
-      "grad_norm": 0.0023925926070660353,
-      "learning_rate": 0.0001999868301113675,
-      "loss": 46.0,
-      "step": 32090
-    },
-    {
-      "epoch": 5.168042191714642,
-      "grad_norm": 0.0020383901428431273,
-      "learning_rate": 0.00019998682929030736,
-      "loss": 46.0,
-      "step": 32091
-    },
-    {
-      "epoch": 5.16820322879343,
-      "grad_norm": 0.0017731408588588238,
-      "learning_rate": 0.00019998682846922164,
-      "loss": 46.0,
-      "step": 32092
-    },
-    {
-      "epoch": 5.168364265872217,
-      "grad_norm": 0.006467212922871113,
-      "learning_rate": 0.00019998682764811026,
-      "loss": 46.0,
-      "step": 32093
-    },
-    {
-      "epoch": 5.1685253029510045,
-      "grad_norm": 0.0034874887205660343,
-      "learning_rate": 0.00019998682682697335,
-      "loss": 46.0,
-      "step": 32094
-    },
-    {
-      "epoch": 5.168686340029792,
-      "grad_norm": 0.0034083575010299683,
-      "learning_rate": 0.00019998682600581085,
-      "loss": 46.0,
-      "step": 32095
-    },
-    {
-      "epoch": 5.168847377108579,
-      "grad_norm": 0.003910582046955824,
-      "learning_rate": 0.00019998682518462273,
-      "loss": 46.0,
-      "step": 32096
-    },
-    {
-      "epoch": 5.169008414187367,
-      "grad_norm": 0.01141782570630312,
-      "learning_rate": 0.00019998682436340906,
-      "loss": 46.0,
-      "step": 32097
-    },
-    {
-      "epoch": 5.169169451266154,
-      "grad_norm": 0.003728589043021202,
-      "learning_rate": 0.00019998682354216977,
-      "loss": 46.0,
-      "step": 32098
-    },
-    {
-      "epoch": 5.169330488344942,
-      "grad_norm": 0.0011391510488465428,
-      "learning_rate": 0.00019998682272090486,
-      "loss": 46.0,
-      "step": 32099
-    },
-    {
-      "epoch": 5.169491525423728,
-      "grad_norm": 0.0020659365691244602,
-      "learning_rate": 0.0001999868218996144,
-      "loss": 46.0,
-      "step": 32100
-    },
-    {
-      "epoch": 5.169652562502516,
-      "grad_norm": 0.015897784382104874,
-      "learning_rate": 0.00019998682107829835,
-      "loss": 46.0,
-      "step": 32101
-    },
-    {
-      "epoch": 5.169813599581303,
-      "grad_norm": 0.00939464196562767,
-      "learning_rate": 0.00019998682025695668,
-      "loss": 46.0,
-      "step": 32102
-    },
-    {
-      "epoch": 5.169974636660091,
-      "grad_norm": 0.0017044766573235393,
-      "learning_rate": 0.00019998681943558943,
-      "loss": 46.0,
-      "step": 32103
-    },
-    {
-      "epoch": 5.170135673738878,
-      "grad_norm": 0.007020737510174513,
-      "learning_rate": 0.0001999868186141966,
-      "loss": 46.0,
-      "step": 32104
-    },
-    {
-      "epoch": 5.170296710817666,
-      "grad_norm": 0.001351483864709735,
-      "learning_rate": 0.00019998681779277816,
-      "loss": 46.0,
-      "step": 32105
-    },
-    {
-      "epoch": 5.170457747896453,
-      "grad_norm": 0.009765173308551311,
-      "learning_rate": 0.00019998681697133415,
-      "loss": 46.0,
-      "step": 32106
-    },
-    {
-      "epoch": 5.1706187849752405,
-      "grad_norm": 0.0012136913137510419,
-      "learning_rate": 0.00019998681614986455,
-      "loss": 46.0,
-      "step": 32107
-    },
-    {
-      "epoch": 5.170779822054028,
-      "grad_norm": 0.0009749795426614583,
-      "learning_rate": 0.0001999868153283693,
-      "loss": 46.0,
-      "step": 32108
-    },
-    {
-      "epoch": 5.170940859132815,
-      "grad_norm": 0.003672999795526266,
-      "learning_rate": 0.00019998681450684853,
-      "loss": 46.0,
-      "step": 32109
-    },
-    {
-      "epoch": 5.171101896211603,
-      "grad_norm": 0.010947924107313156,
-      "learning_rate": 0.00019998681368530214,
-      "loss": 46.0,
-      "step": 32110
-    },
-    {
-      "epoch": 5.17126293329039,
-      "grad_norm": 0.006620231084525585,
-      "learning_rate": 0.00019998681286373016,
-      "loss": 46.0,
-      "step": 32111
-    },
-    {
-      "epoch": 5.171423970369178,
-      "grad_norm": 0.015757473185658455,
-      "learning_rate": 0.0001999868120421326,
-      "loss": 46.0,
-      "step": 32112
-    },
-    {
-      "epoch": 5.171585007447965,
-      "grad_norm": 0.0023470637388527393,
-      "learning_rate": 0.00019998681122050945,
-      "loss": 46.0,
-      "step": 32113
-    },
-    {
-      "epoch": 5.171746044526753,
-      "grad_norm": 0.0022487491369247437,
-      "learning_rate": 0.00019998681039886068,
-      "loss": 46.0,
-      "step": 32114
-    },
-    {
-      "epoch": 5.171907081605539,
-      "grad_norm": 0.002365415683016181,
-      "learning_rate": 0.00019998680957718633,
-      "loss": 46.0,
-      "step": 32115
-    },
-    {
-      "epoch": 5.172068118684327,
-      "grad_norm": 0.0034747770987451077,
-      "learning_rate": 0.00019998680875548642,
-      "loss": 46.0,
-      "step": 32116
-    },
-    {
-      "epoch": 5.172229155763114,
-      "grad_norm": 0.0028694013599306345,
-      "learning_rate": 0.0001999868079337609,
-      "loss": 46.0,
-      "step": 32117
-    },
-    {
-      "epoch": 5.172390192841902,
-      "grad_norm": 0.0029435050673782825,
-      "learning_rate": 0.00019998680711200975,
-      "loss": 46.0,
-      "step": 32118
-    },
-    {
-      "epoch": 5.172551229920689,
-      "grad_norm": 0.013902498409152031,
-      "learning_rate": 0.00019998680629023302,
-      "loss": 46.0,
-      "step": 32119
-    },
-    {
-      "epoch": 5.1727122669994765,
-      "grad_norm": 0.0063034528866410255,
-      "learning_rate": 0.00019998680546843073,
-      "loss": 46.0,
-      "step": 32120
-    },
-    {
-      "epoch": 5.172873304078264,
-      "grad_norm": 0.005939032882452011,
-      "learning_rate": 0.00019998680464660285,
-      "loss": 46.0,
-      "step": 32121
-    },
-    {
-      "epoch": 5.173034341157051,
-      "grad_norm": 0.0005871170433238149,
-      "learning_rate": 0.00019998680382474934,
-      "loss": 46.0,
-      "step": 32122
-    },
-    {
-      "epoch": 5.173195378235839,
-      "grad_norm": 0.0053421612828969955,
-      "learning_rate": 0.00019998680300287029,
-      "loss": 46.0,
-      "step": 32123
-    },
-    {
-      "epoch": 5.173356415314626,
-      "grad_norm": 0.003946044947952032,
-      "learning_rate": 0.00019998680218096562,
-      "loss": 46.0,
-      "step": 32124
-    },
-    {
-      "epoch": 5.173517452393414,
-      "grad_norm": 0.012237252667546272,
-      "learning_rate": 0.00019998680135903537,
-      "loss": 46.0,
-      "step": 32125
-    },
-    {
-      "epoch": 5.173678489472201,
-      "grad_norm": 0.008623216301202774,
-      "learning_rate": 0.0001999868005370795,
-      "loss": 46.0,
-      "step": 32126
-    },
-    {
-      "epoch": 5.173839526550989,
-      "grad_norm": 0.006859529297798872,
-      "learning_rate": 0.00019998679971509805,
-      "loss": 46.0,
-      "step": 32127
-    },
-    {
-      "epoch": 5.174000563629776,
-      "grad_norm": 0.004463626071810722,
-      "learning_rate": 0.000199986798893091,
-      "loss": 46.0,
-      "step": 32128
-    },
-    {
-      "epoch": 5.174161600708564,
-      "grad_norm": 0.0021174491848796606,
-      "learning_rate": 0.0001999867980710584,
-      "loss": 46.0,
-      "step": 32129
-    },
-    {
-      "epoch": 5.17432263778735,
-      "grad_norm": 0.0017284222412854433,
-      "learning_rate": 0.00019998679724900017,
-      "loss": 46.0,
-      "step": 32130
-    },
-    {
-      "epoch": 5.174483674866138,
-      "grad_norm": 0.006913302466273308,
-      "learning_rate": 0.00019998679642691637,
-      "loss": 46.0,
-      "step": 32131
-    },
-    {
-      "epoch": 5.174644711944925,
-      "grad_norm": 0.003466990776360035,
-      "learning_rate": 0.00019998679560480698,
-      "loss": 46.0,
-      "step": 32132
-    },
-    {
-      "epoch": 5.1748057490237125,
-      "grad_norm": 0.0019665840081870556,
-      "learning_rate": 0.000199986794782672,
-      "loss": 46.0,
-      "step": 32133
-    },
-    {
-      "epoch": 5.1749667861025,
-      "grad_norm": 0.0018506355118006468,
-      "learning_rate": 0.00019998679396051138,
-      "loss": 46.0,
-      "step": 32134
-    },
-    {
-      "epoch": 5.175127823181287,
-      "grad_norm": 0.014390092343091965,
-      "learning_rate": 0.00019998679313832523,
-      "loss": 46.0,
-      "step": 32135
-    },
-    {
-      "epoch": 5.175288860260075,
-      "grad_norm": 0.003375882050022483,
-      "learning_rate": 0.00019998679231611344,
-      "loss": 46.0,
-      "step": 32136
-    },
-    {
-      "epoch": 5.175449897338862,
-      "grad_norm": 0.002697971649467945,
-      "learning_rate": 0.0001999867914938761,
-      "loss": 46.0,
-      "step": 32137
-    },
-    {
-      "epoch": 5.17561093441765,
-      "grad_norm": 0.00269668223336339,
-      "learning_rate": 0.00019998679067161315,
-      "loss": 46.0,
-      "step": 32138
-    },
-    {
-      "epoch": 5.175771971496437,
-      "grad_norm": 0.00459744269028306,
-      "learning_rate": 0.00019998678984932462,
-      "loss": 46.0,
-      "step": 32139
-    },
-    {
-      "epoch": 5.175933008575225,
-      "grad_norm": 0.013508311472833157,
-      "learning_rate": 0.00019998678902701046,
-      "loss": 46.0,
-      "step": 32140
-    },
-    {
-      "epoch": 5.176094045654012,
-      "grad_norm": 0.010211831890046597,
-      "learning_rate": 0.00019998678820467073,
-      "loss": 46.0,
-      "step": 32141
-    },
-    {
-      "epoch": 5.1762550827327996,
-      "grad_norm": 0.00506754033267498,
-      "learning_rate": 0.00019998678738230544,
-      "loss": 46.0,
-      "step": 32142
-    },
-    {
-      "epoch": 5.176416119811587,
-      "grad_norm": 0.0034224314149469137,
-      "learning_rate": 0.00019998678655991454,
-      "loss": 46.0,
-      "step": 32143
-    },
-    {
-      "epoch": 5.1765771568903745,
-      "grad_norm": 0.0038564172573387623,
-      "learning_rate": 0.00019998678573749802,
-      "loss": 46.0,
-      "step": 32144
-    },
-    {
-      "epoch": 5.176738193969161,
-      "grad_norm": 0.0031078457832336426,
-      "learning_rate": 0.00019998678491505594,
-      "loss": 46.0,
-      "step": 32145
-    },
-    {
-      "epoch": 5.1768992310479485,
-      "grad_norm": 0.006362107582390308,
-      "learning_rate": 0.00019998678409258825,
-      "loss": 46.0,
-      "step": 32146
-    },
-    {
-      "epoch": 5.177060268126736,
-      "grad_norm": 0.008865317329764366,
-      "learning_rate": 0.00019998678327009498,
-      "loss": 46.0,
-      "step": 32147
-    },
-    {
-      "epoch": 5.177221305205523,
-      "grad_norm": 0.0026274453848600388,
-      "learning_rate": 0.0001999867824475761,
-      "loss": 46.0,
-      "step": 32148
-    },
-    {
-      "epoch": 5.177382342284311,
-      "grad_norm": 0.013871009461581707,
-      "learning_rate": 0.00019998678162503166,
-      "loss": 46.0,
-      "step": 32149
-    },
-    {
-      "epoch": 5.177543379363098,
-      "grad_norm": 0.003805729327723384,
-      "learning_rate": 0.00019998678080246162,
-      "loss": 46.0,
-      "step": 32150
-    },
-    {
-      "epoch": 5.177704416441886,
-      "grad_norm": 0.0035232841037213802,
-      "learning_rate": 0.000199986779979866,
-      "loss": 46.0,
-      "step": 32151
-    },
-    {
-      "epoch": 5.177865453520673,
-      "grad_norm": 0.008128591813147068,
-      "learning_rate": 0.00019998677915724475,
-      "loss": 46.0,
-      "step": 32152
-    },
-    {
-      "epoch": 5.178026490599461,
-      "grad_norm": 0.0013356945710256696,
-      "learning_rate": 0.00019998677833459792,
-      "loss": 46.0,
-      "step": 32153
-    },
-    {
-      "epoch": 5.178187527678248,
-      "grad_norm": 0.005026698112487793,
-      "learning_rate": 0.0001999867775119255,
-      "loss": 46.0,
-      "step": 32154
-    },
-    {
-      "epoch": 5.1783485647570355,
-      "grad_norm": 0.00934301782399416,
-      "learning_rate": 0.0001999867766892275,
-      "loss": 46.0,
-      "step": 32155
-    },
-    {
-      "epoch": 5.178509601835823,
-      "grad_norm": 0.014060525223612785,
-      "learning_rate": 0.00019998677586650392,
-      "loss": 46.0,
-      "step": 32156
-    },
-    {
-      "epoch": 5.17867063891461,
-      "grad_norm": 0.005326290149241686,
-      "learning_rate": 0.00019998677504375474,
-      "loss": 46.0,
-      "step": 32157
-    },
-    {
-      "epoch": 5.178831675993398,
-      "grad_norm": 0.00762485945597291,
-      "learning_rate": 0.00019998677422097995,
-      "loss": 46.0,
-      "step": 32158
-    },
-    {
-      "epoch": 5.1789927130721845,
-      "grad_norm": 0.003433982143178582,
-      "learning_rate": 0.00019998677339817957,
-      "loss": 46.0,
-      "step": 32159
-    },
-    {
-      "epoch": 5.179153750150972,
-      "grad_norm": 0.003874999238178134,
-      "learning_rate": 0.0001999867725753536,
-      "loss": 46.0,
-      "step": 32160
-    },
-    {
-      "epoch": 5.179314787229759,
-      "grad_norm": 0.0026599785778671503,
-      "learning_rate": 0.00019998677175250205,
-      "loss": 46.0,
-      "step": 32161
-    },
-    {
-      "epoch": 5.179475824308547,
-      "grad_norm": 0.0011649243533611298,
-      "learning_rate": 0.0001999867709296249,
-      "loss": 46.0,
-      "step": 32162
-    },
-    {
-      "epoch": 5.179636861387334,
-      "grad_norm": 0.005228824447840452,
-      "learning_rate": 0.00019998677010672219,
-      "loss": 46.0,
-      "step": 32163
-    },
-    {
-      "epoch": 5.179797898466122,
-      "grad_norm": 0.009424913674592972,
-      "learning_rate": 0.00019998676928379384,
-      "loss": 46.0,
-      "step": 32164
-    },
-    {
-      "epoch": 5.179958935544909,
-      "grad_norm": 0.0023327148519456387,
-      "learning_rate": 0.00019998676846083994,
-      "loss": 46.0,
-      "step": 32165
-    },
-    {
-      "epoch": 5.180119972623697,
-      "grad_norm": 0.008002068847417831,
-      "learning_rate": 0.00019998676763786043,
-      "loss": 46.0,
-      "step": 32166
-    },
-    {
-      "epoch": 5.180281009702484,
-      "grad_norm": 0.021423954516649246,
-      "learning_rate": 0.0001999867668148553,
-      "loss": 46.0,
-      "step": 32167
-    },
-    {
-      "epoch": 5.1804420467812715,
-      "grad_norm": 0.0049739438109099865,
-      "learning_rate": 0.0001999867659918246,
-      "loss": 46.0,
-      "step": 32168
-    },
-    {
-      "epoch": 5.180603083860059,
-      "grad_norm": 0.0014739732723683119,
-      "learning_rate": 0.00019998676516876836,
-      "loss": 46.0,
-      "step": 32169
-    },
-    {
-      "epoch": 5.180764120938846,
-      "grad_norm": 0.003991612233221531,
-      "learning_rate": 0.00019998676434568647,
-      "loss": 46.0,
-      "step": 32170
-    },
-    {
-      "epoch": 5.180925158017634,
-      "grad_norm": 0.002354171359911561,
-      "learning_rate": 0.000199986763522579,
-      "loss": 46.0,
-      "step": 32171
-    },
-    {
-      "epoch": 5.181086195096421,
-      "grad_norm": 0.012767891399562359,
-      "learning_rate": 0.00019998676269944595,
-      "loss": 46.0,
-      "step": 32172
-    },
-    {
-      "epoch": 5.181247232175208,
-      "grad_norm": 0.0124012790620327,
-      "learning_rate": 0.0001999867618762873,
-      "loss": 46.0,
-      "step": 32173
-    },
-    {
-      "epoch": 5.181408269253995,
-      "grad_norm": 0.012858306989073753,
-      "learning_rate": 0.00019998676105310303,
-      "loss": 46.0,
-      "step": 32174
-    },
-    {
-      "epoch": 5.181569306332783,
-      "grad_norm": 0.003873368725180626,
-      "learning_rate": 0.00019998676022989323,
-      "loss": 46.0,
-      "step": 32175
-    },
-    {
-      "epoch": 5.18173034341157,
-      "grad_norm": 0.005934448912739754,
-      "learning_rate": 0.0001999867594066578,
-      "loss": 46.0,
-      "step": 32176
-    },
-    {
-      "epoch": 5.181891380490358,
-      "grad_norm": 0.014673309400677681,
-      "learning_rate": 0.00019998675858339676,
-      "loss": 46.0,
-      "step": 32177
-    },
-    {
-      "epoch": 5.182052417569145,
-      "grad_norm": 0.003604224883019924,
-      "learning_rate": 0.00019998675776011014,
-      "loss": 46.0,
-      "step": 32178
-    },
-    {
-      "epoch": 5.182213454647933,
-      "grad_norm": 0.012089215219020844,
-      "learning_rate": 0.00019998675693679797,
-      "loss": 46.0,
-      "step": 32179
-    },
-    {
-      "epoch": 5.18237449172672,
-      "grad_norm": 0.0029040127992630005,
-      "learning_rate": 0.00019998675611346015,
-      "loss": 46.0,
-      "step": 32180
-    },
-    {
-      "epoch": 5.1825355288055075,
-      "grad_norm": 0.006171551998704672,
-      "learning_rate": 0.00019998675529009677,
-      "loss": 46.0,
-      "step": 32181
-    },
-    {
-      "epoch": 5.182696565884295,
-      "grad_norm": 0.0026237000711262226,
-      "learning_rate": 0.0001999867544667078,
-      "loss": 46.0,
-      "step": 32182
-    },
-    {
-      "epoch": 5.182857602963082,
-      "grad_norm": 0.0030658748000860214,
-      "learning_rate": 0.00019998675364329326,
-      "loss": 46.0,
-      "step": 32183
-    },
-    {
-      "epoch": 5.18301864004187,
-      "grad_norm": 0.010112238116562366,
-      "learning_rate": 0.0001999867528198531,
-      "loss": 46.0,
-      "step": 32184
-    },
-    {
-      "epoch": 5.183179677120657,
-      "grad_norm": 0.012909832410514355,
-      "learning_rate": 0.00019998675199638734,
-      "loss": 46.0,
-      "step": 32185
-    },
-    {
-      "epoch": 5.183340714199445,
-      "grad_norm": 0.003441684879362583,
-      "learning_rate": 0.00019998675117289597,
-      "loss": 46.0,
-      "step": 32186
-    },
-    {
-      "epoch": 5.183501751278232,
-      "grad_norm": 0.005345614161342382,
-      "learning_rate": 0.00019998675034937907,
-      "loss": 46.0,
-      "step": 32187
-    },
-    {
-      "epoch": 5.183662788357019,
-      "grad_norm": 0.002593819983303547,
-      "learning_rate": 0.00019998674952583653,
-      "loss": 46.0,
-      "step": 32188
-    },
-    {
-      "epoch": 5.183823825435806,
-      "grad_norm": 0.0019140162039548159,
-      "learning_rate": 0.0001999867487022684,
-      "loss": 46.0,
-      "step": 32189
-    },
-    {
-      "epoch": 5.183984862514594,
-      "grad_norm": 0.0022274444345384836,
-      "learning_rate": 0.00019998674787867474,
-      "loss": 46.0,
-      "step": 32190
-    },
-    {
-      "epoch": 5.184145899593381,
-      "grad_norm": 0.006879804190248251,
-      "learning_rate": 0.00019998674705505544,
-      "loss": 46.0,
-      "step": 32191
-    },
-    {
-      "epoch": 5.184306936672169,
-      "grad_norm": 0.0018583649070933461,
-      "learning_rate": 0.00019998674623141055,
-      "loss": 46.0,
-      "step": 32192
-    },
-    {
-      "epoch": 5.184467973750956,
-      "grad_norm": 0.0010316588450223207,
-      "learning_rate": 0.00019998674540774004,
-      "loss": 46.0,
-      "step": 32193
-    },
-    {
-      "epoch": 5.1846290108297435,
-      "grad_norm": 0.015452869236469269,
-      "learning_rate": 0.00019998674458404398,
-      "loss": 46.0,
-      "step": 32194
-    },
-    {
-      "epoch": 5.184790047908531,
-      "grad_norm": 0.008038332685828209,
-      "learning_rate": 0.00019998674376032233,
-      "loss": 46.0,
-      "step": 32195
-    },
-    {
-      "epoch": 5.184951084987318,
-      "grad_norm": 0.0045012845657765865,
-      "learning_rate": 0.00019998674293657506,
-      "loss": 46.0,
-      "step": 32196
-    },
-    {
-      "epoch": 5.185112122066106,
-      "grad_norm": 0.0007633883506059647,
-      "learning_rate": 0.0001999867421128022,
-      "loss": 46.0,
-      "step": 32197
-    },
-    {
-      "epoch": 5.185273159144893,
-      "grad_norm": 0.004212799482047558,
-      "learning_rate": 0.0001999867412890038,
-      "loss": 46.0,
-      "step": 32198
-    },
-    {
-      "epoch": 5.185434196223681,
-      "grad_norm": 0.018131045624613762,
-      "learning_rate": 0.00019998674046517976,
-      "loss": 46.0,
-      "step": 32199
-    },
-    {
-      "epoch": 5.185595233302468,
-      "grad_norm": 0.0029433213640004396,
-      "learning_rate": 0.00019998673964133012,
-      "loss": 46.0,
-      "step": 32200
-    },
-    {
-      "epoch": 5.185756270381256,
-      "grad_norm": 0.005548818968236446,
-      "learning_rate": 0.00019998673881745492,
-      "loss": 46.0,
-      "step": 32201
-    },
-    {
-      "epoch": 5.185917307460043,
-      "grad_norm": 0.03727267310023308,
-      "learning_rate": 0.0001999867379935541,
-      "loss": 46.0,
-      "step": 32202
-    },
-    {
-      "epoch": 5.18607834453883,
-      "grad_norm": 0.0027350359596312046,
-      "learning_rate": 0.0001999867371696277,
-      "loss": 46.0,
-      "step": 32203
-    },
-    {
-      "epoch": 5.186239381617617,
-      "grad_norm": 0.01887637749314308,
-      "learning_rate": 0.00019998673634567574,
-      "loss": 46.0,
-      "step": 32204
-    },
-    {
-      "epoch": 5.186400418696405,
-      "grad_norm": 0.008631446398794651,
-      "learning_rate": 0.00019998673552169813,
-      "loss": 46.0,
-      "step": 32205
-    },
-    {
-      "epoch": 5.186561455775192,
-      "grad_norm": 0.002271883189678192,
-      "learning_rate": 0.00019998673469769497,
-      "loss": 46.0,
-      "step": 32206
-    },
-    {
-      "epoch": 5.1867224928539795,
-      "grad_norm": 0.0039206258952617645,
-      "learning_rate": 0.0001999867338736662,
-      "loss": 46.0,
-      "step": 32207
-    },
-    {
-      "epoch": 5.186883529932767,
-      "grad_norm": 0.0022147258277982473,
-      "learning_rate": 0.00019998673304961188,
-      "loss": 46.0,
-      "step": 32208
-    },
-    {
-      "epoch": 5.187044567011554,
-      "grad_norm": 0.006100933067500591,
-      "learning_rate": 0.0001999867322255319,
-      "loss": 46.0,
-      "step": 32209
-    },
-    {
-      "epoch": 5.187205604090342,
-      "grad_norm": 0.0021571496035903692,
-      "learning_rate": 0.00019998673140142635,
-      "loss": 46.0,
-      "step": 32210
-    },
-    {
-      "epoch": 5.187366641169129,
-      "grad_norm": 0.009839133359491825,
-      "learning_rate": 0.00019998673057729525,
-      "loss": 46.0,
-      "step": 32211
-    },
-    {
-      "epoch": 5.187527678247917,
-      "grad_norm": 0.004839913919568062,
-      "learning_rate": 0.00019998672975313854,
-      "loss": 46.0,
-      "step": 32212
-    },
-    {
-      "epoch": 5.187688715326704,
-      "grad_norm": 0.0029112743213772774,
-      "learning_rate": 0.00019998672892895624,
-      "loss": 46.0,
-      "step": 32213
-    },
-    {
-      "epoch": 5.187849752405492,
-      "grad_norm": 0.004325307905673981,
-      "learning_rate": 0.00019998672810474832,
-      "loss": 46.0,
-      "step": 32214
-    },
-    {
-      "epoch": 5.188010789484279,
-      "grad_norm": 0.001963454531505704,
-      "learning_rate": 0.00019998672728051484,
-      "loss": 46.0,
-      "step": 32215
-    },
-    {
-      "epoch": 5.1881718265630665,
-      "grad_norm": 0.013674069195985794,
-      "learning_rate": 0.00019998672645625573,
-      "loss": 46.0,
-      "step": 32216
-    },
-    {
-      "epoch": 5.188332863641854,
-      "grad_norm": 0.013424782082438469,
-      "learning_rate": 0.00019998672563197105,
-      "loss": 46.0,
-      "step": 32217
-    },
-    {
-      "epoch": 5.1884939007206405,
-      "grad_norm": 0.002722163451835513,
-      "learning_rate": 0.00019998672480766078,
-      "loss": 46.0,
-      "step": 32218
-    },
-    {
-      "epoch": 5.188654937799428,
-      "grad_norm": 0.0076047698967158794,
-      "learning_rate": 0.00019998672398332493,
-      "loss": 46.0,
-      "step": 32219
-    },
-    {
-      "epoch": 5.188815974878215,
-      "grad_norm": 0.004611724056303501,
-      "learning_rate": 0.00019998672315896346,
-      "loss": 46.0,
-      "step": 32220
-    },
-    {
-      "epoch": 5.188977011957003,
-      "grad_norm": 0.006655667908489704,
-      "learning_rate": 0.00019998672233457644,
-      "loss": 46.0,
-      "step": 32221
-    },
-    {
-      "epoch": 5.18913804903579,
-      "grad_norm": 0.0017513770144432783,
-      "learning_rate": 0.0001999867215101638,
-      "loss": 46.0,
-      "step": 32222
-    },
-    {
-      "epoch": 5.189299086114578,
-      "grad_norm": 0.007567969616502523,
-      "learning_rate": 0.0001999867206857256,
-      "loss": 46.0,
-      "step": 32223
-    },
-    {
-      "epoch": 5.189460123193365,
-      "grad_norm": 0.002437217626720667,
-      "learning_rate": 0.00019998671986126175,
-      "loss": 46.0,
-      "step": 32224
-    },
-    {
-      "epoch": 5.189621160272153,
-      "grad_norm": 0.007849694229662418,
-      "learning_rate": 0.00019998671903677235,
-      "loss": 46.0,
-      "step": 32225
-    },
-    {
-      "epoch": 5.18978219735094,
-      "grad_norm": 0.007074286229908466,
-      "learning_rate": 0.00019998671821225736,
-      "loss": 46.0,
-      "step": 32226
-    },
-    {
-      "epoch": 5.189943234429728,
-      "grad_norm": 0.004294555634260178,
-      "learning_rate": 0.00019998671738771678,
-      "loss": 46.0,
-      "step": 32227
-    },
-    {
-      "epoch": 5.190104271508515,
-      "grad_norm": 0.003116513602435589,
-      "learning_rate": 0.0001999867165631506,
-      "loss": 46.0,
-      "step": 32228
-    },
-    {
-      "epoch": 5.1902653085873025,
-      "grad_norm": 0.018665479496121407,
-      "learning_rate": 0.00019998671573855884,
-      "loss": 46.0,
-      "step": 32229
-    },
-    {
-      "epoch": 5.19042634566609,
-      "grad_norm": 0.005529359448701143,
-      "learning_rate": 0.00019998671491394145,
-      "loss": 46.0,
-      "step": 32230
-    },
-    {
-      "epoch": 5.190587382744877,
-      "grad_norm": 0.003929000813513994,
-      "learning_rate": 0.0001999867140892985,
-      "loss": 46.0,
-      "step": 32231
-    },
-    {
-      "epoch": 5.190748419823664,
-      "grad_norm": 0.001755247823894024,
-      "learning_rate": 0.00019998671326462996,
-      "loss": 46.0,
-      "step": 32232
-    },
-    {
-      "epoch": 5.190909456902451,
-      "grad_norm": 0.006418773904442787,
-      "learning_rate": 0.0001999867124399358,
-      "loss": 46.0,
-      "step": 32233
-    },
-    {
-      "epoch": 5.191070493981239,
-      "grad_norm": 0.002701960038393736,
-      "learning_rate": 0.00019998671161521606,
-      "loss": 46.0,
-      "step": 32234
-    },
-    {
-      "epoch": 5.191231531060026,
-      "grad_norm": 0.007042792625725269,
-      "learning_rate": 0.00019998671079047076,
-      "loss": 46.0,
-      "step": 32235
-    },
-    {
-      "epoch": 5.191392568138814,
-      "grad_norm": 0.004141178447753191,
-      "learning_rate": 0.00019998670996569984,
-      "loss": 46.0,
-      "step": 32236
-    },
-    {
-      "epoch": 5.191553605217601,
-      "grad_norm": 0.01399219036102295,
-      "learning_rate": 0.00019998670914090334,
-      "loss": 46.0,
-      "step": 32237
-    },
-    {
-      "epoch": 5.191714642296389,
-      "grad_norm": 0.001908156438730657,
-      "learning_rate": 0.00019998670831608125,
-      "loss": 46.0,
-      "step": 32238
-    },
-    {
-      "epoch": 5.191875679375176,
-      "grad_norm": 0.005616384092718363,
-      "learning_rate": 0.00019998670749123357,
-      "loss": 46.0,
-      "step": 32239
-    },
-    {
-      "epoch": 5.192036716453964,
-      "grad_norm": 0.006891006603837013,
-      "learning_rate": 0.00019998670666636028,
-      "loss": 46.0,
-      "step": 32240
-    },
-    {
-      "epoch": 5.192197753532751,
-      "grad_norm": 0.01339486800134182,
-      "learning_rate": 0.0001999867058414614,
-      "loss": 46.0,
-      "step": 32241
-    },
-    {
-      "epoch": 5.1923587906115385,
-      "grad_norm": 0.0019606389105319977,
-      "learning_rate": 0.00019998670501653694,
-      "loss": 46.0,
-      "step": 32242
-    },
-    {
-      "epoch": 5.192519827690326,
-      "grad_norm": 0.011777014471590519,
-      "learning_rate": 0.00019998670419158688,
-      "loss": 46.0,
-      "step": 32243
-    },
-    {
-      "epoch": 5.192680864769113,
-      "grad_norm": 0.0020849357824772596,
-      "learning_rate": 0.00019998670336661122,
-      "loss": 46.0,
-      "step": 32244
-    },
-    {
-      "epoch": 5.192841901847901,
-      "grad_norm": 0.012270960956811905,
-      "learning_rate": 0.00019998670254161002,
-      "loss": 46.0,
-      "step": 32245
-    },
-    {
-      "epoch": 5.193002938926688,
-      "grad_norm": 0.012650938704609871,
-      "learning_rate": 0.00019998670171658315,
-      "loss": 46.0,
-      "step": 32246
-    },
-    {
-      "epoch": 5.193163976005475,
-      "grad_norm": 0.009674279019236565,
-      "learning_rate": 0.00019998670089153072,
-      "loss": 46.0,
-      "step": 32247
-    },
-    {
-      "epoch": 5.193325013084262,
-      "grad_norm": 0.014590908773243427,
-      "learning_rate": 0.00019998670006645273,
-      "loss": 46.0,
-      "step": 32248
-    },
-    {
-      "epoch": 5.19348605016305,
-      "grad_norm": 0.0007428644457831979,
-      "learning_rate": 0.00019998669924134913,
-      "loss": 46.0,
-      "step": 32249
-    },
-    {
-      "epoch": 5.193647087241837,
-      "grad_norm": 0.014840604737401009,
-      "learning_rate": 0.00019998669841621994,
-      "loss": 46.0,
-      "step": 32250
-    },
-    {
-      "epoch": 5.193808124320625,
-      "grad_norm": 0.0019372176611796021,
-      "learning_rate": 0.00019998669759106513,
-      "loss": 46.0,
-      "step": 32251
-    },
-    {
-      "epoch": 5.193969161399412,
-      "grad_norm": 0.003242619801312685,
-      "learning_rate": 0.00019998669676588477,
-      "loss": 46.0,
-      "step": 32252
-    },
-    {
-      "epoch": 5.1941301984782,
-      "grad_norm": 0.006310049444437027,
-      "learning_rate": 0.00019998669594067882,
-      "loss": 46.0,
-      "step": 32253
-    },
-    {
-      "epoch": 5.194291235556987,
-      "grad_norm": 0.0031712206546217203,
-      "learning_rate": 0.00019998669511544725,
-      "loss": 46.0,
-      "step": 32254
-    },
-    {
-      "epoch": 5.1944522726357745,
-      "grad_norm": 0.026464257389307022,
-      "learning_rate": 0.0001999866942901901,
-      "loss": 46.0,
-      "step": 32255
-    },
-    {
-      "epoch": 5.194613309714562,
-      "grad_norm": 0.011263553984463215,
-      "learning_rate": 0.00019998669346490736,
-      "loss": 46.0,
-      "step": 32256
-    },
-    {
-      "epoch": 5.194774346793349,
-      "grad_norm": 0.0027723631355911493,
-      "learning_rate": 0.00019998669263959903,
-      "loss": 46.0,
-      "step": 32257
-    },
-    {
-      "epoch": 5.194935383872137,
-      "grad_norm": 0.0026018302887678146,
-      "learning_rate": 0.0001999866918142651,
-      "loss": 46.0,
-      "step": 32258
-    },
-    {
-      "epoch": 5.195096420950924,
-      "grad_norm": 0.006990714464336634,
-      "learning_rate": 0.00019998669098890556,
-      "loss": 46.0,
-      "step": 32259
-    },
-    {
-      "epoch": 5.195257458029712,
-      "grad_norm": 0.021532442420721054,
-      "learning_rate": 0.00019998669016352047,
-      "loss": 46.0,
-      "step": 32260
-    },
-    {
-      "epoch": 5.195418495108498,
-      "grad_norm": 0.0025244515854865313,
-      "learning_rate": 0.00019998668933810977,
-      "loss": 46.0,
-      "step": 32261
-    },
-    {
-      "epoch": 5.195579532187286,
-      "grad_norm": 0.0014430736191570759,
-      "learning_rate": 0.00019998668851267345,
-      "loss": 46.0,
-      "step": 32262
-    },
-    {
-      "epoch": 5.195740569266073,
-      "grad_norm": 0.0033355862833559513,
-      "learning_rate": 0.00019998668768721157,
-      "loss": 46.0,
-      "step": 32263
-    },
-    {
-      "epoch": 5.195901606344861,
-      "grad_norm": 0.0025123562663793564,
-      "learning_rate": 0.0001999866868617241,
-      "loss": 46.0,
-      "step": 32264
-    },
-    {
-      "epoch": 5.196062643423648,
-      "grad_norm": 0.003933960106223822,
-      "learning_rate": 0.00019998668603621105,
-      "loss": 46.0,
-      "step": 32265
-    },
-    {
-      "epoch": 5.1962236805024355,
-      "grad_norm": 0.005743119399994612,
-      "learning_rate": 0.0001999866852106724,
-      "loss": 46.0,
-      "step": 32266
-    },
-    {
-      "epoch": 5.196384717581223,
-      "grad_norm": 0.003340854775160551,
-      "learning_rate": 0.00019998668438510813,
-      "loss": 46.0,
-      "step": 32267
-    },
-    {
-      "epoch": 5.1965457546600105,
-      "grad_norm": 0.010091559961438179,
-      "learning_rate": 0.0001999866835595183,
-      "loss": 46.0,
-      "step": 32268
-    },
-    {
-      "epoch": 5.196706791738798,
-      "grad_norm": 0.001893533393740654,
-      "learning_rate": 0.00019998668273390284,
-      "loss": 46.0,
-      "step": 32269
-    },
-    {
-      "epoch": 5.196867828817585,
-      "grad_norm": 0.014940696768462658,
-      "learning_rate": 0.00019998668190826182,
-      "loss": 46.0,
-      "step": 32270
-    },
-    {
-      "epoch": 5.197028865896373,
-      "grad_norm": 0.003325200639665127,
-      "learning_rate": 0.00019998668108259522,
-      "loss": 46.0,
-      "step": 32271
-    },
-    {
-      "epoch": 5.19718990297516,
-      "grad_norm": 0.008352204225957394,
-      "learning_rate": 0.000199986680256903,
-      "loss": 46.0,
-      "step": 32272
-    },
-    {
-      "epoch": 5.197350940053948,
-      "grad_norm": 0.006152221467345953,
-      "learning_rate": 0.0001999866794311852,
-      "loss": 46.0,
-      "step": 32273
-    },
-    {
-      "epoch": 5.197511977132735,
-      "grad_norm": 0.0026792038697749376,
-      "learning_rate": 0.0001999866786054418,
-      "loss": 46.0,
-      "step": 32274
-    },
-    {
-      "epoch": 5.197673014211523,
-      "grad_norm": 0.0010549160651862621,
-      "learning_rate": 0.00019998667777967283,
-      "loss": 46.0,
-      "step": 32275
-    },
-    {
-      "epoch": 5.197834051290309,
-      "grad_norm": 0.03546413406729698,
-      "learning_rate": 0.00019998667695387827,
-      "loss": 46.0,
-      "step": 32276
-    },
-    {
-      "epoch": 5.197995088369097,
-      "grad_norm": 0.006419955287128687,
-      "learning_rate": 0.00019998667612805812,
-      "loss": 46.0,
-      "step": 32277
-    },
-    {
-      "epoch": 5.198156125447884,
-      "grad_norm": 0.00393195915967226,
-      "learning_rate": 0.00019998667530221232,
-      "loss": 46.0,
-      "step": 32278
-    },
-    {
-      "epoch": 5.1983171625266715,
-      "grad_norm": 0.00228667794726789,
-      "learning_rate": 0.000199986674476341,
-      "loss": 46.0,
-      "step": 32279
-    },
-    {
-      "epoch": 5.198478199605459,
-      "grad_norm": 0.013646171428263187,
-      "learning_rate": 0.00019998667365044405,
-      "loss": 46.0,
-      "step": 32280
-    },
-    {
-      "epoch": 5.198639236684246,
-      "grad_norm": 0.007861576974391937,
-      "learning_rate": 0.00019998667282452153,
-      "loss": 46.0,
-      "step": 32281
-    },
-    {
-      "epoch": 5.198800273763034,
-      "grad_norm": 0.021162524819374084,
-      "learning_rate": 0.0001999866719985734,
-      "loss": 46.0,
-      "step": 32282
-    },
-    {
-      "epoch": 5.198961310841821,
-      "grad_norm": 0.0029266648925840855,
-      "learning_rate": 0.0001999866711725997,
-      "loss": 46.0,
-      "step": 32283
-    },
-    {
-      "epoch": 5.199122347920609,
-      "grad_norm": 0.013833215460181236,
-      "learning_rate": 0.0001999866703466004,
-      "loss": 46.0,
-      "step": 32284
-    },
-    {
-      "epoch": 5.199283384999396,
-      "grad_norm": 0.009972468949854374,
-      "learning_rate": 0.0001999866695205755,
-      "loss": 46.0,
-      "step": 32285
-    },
-    {
-      "epoch": 5.199444422078184,
-      "grad_norm": 0.01192058902233839,
-      "learning_rate": 0.000199986668694525,
-      "loss": 46.0,
-      "step": 32286
-    },
-    {
-      "epoch": 5.199605459156971,
-      "grad_norm": 0.0017158373957499862,
-      "learning_rate": 0.00019998666786844892,
-      "loss": 46.0,
-      "step": 32287
-    },
-    {
-      "epoch": 5.199766496235759,
-      "grad_norm": 0.0027622284833341837,
-      "learning_rate": 0.00019998666704234728,
-      "loss": 46.0,
-      "step": 32288
-    },
-    {
-      "epoch": 5.199927533314546,
-      "grad_norm": 0.001802717219106853,
-      "learning_rate": 0.00019998666621622,
-      "loss": 46.0,
-      "step": 32289
-    },
-    {
-      "epoch": 5.2000885703933335,
-      "grad_norm": 0.0036672374699264765,
-      "learning_rate": 0.00019998666539006713,
-      "loss": 46.0,
-      "step": 32290
-    },
-    {
-      "epoch": 5.20024960747212,
-      "grad_norm": 0.012927980162203312,
-      "learning_rate": 0.0001999866645638887,
-      "loss": 46.0,
-      "step": 32291
-    },
-    {
-      "epoch": 5.2004106445509075,
-      "grad_norm": 0.0068999649956822395,
-      "learning_rate": 0.00019998666373768467,
-      "loss": 46.0,
-      "step": 32292
-    },
-    {
-      "epoch": 5.200571681629695,
-      "grad_norm": 0.002394864335656166,
-      "learning_rate": 0.00019998666291145504,
-      "loss": 46.0,
-      "step": 32293
-    },
-    {
-      "epoch": 5.200732718708482,
-      "grad_norm": 0.006971741560846567,
-      "learning_rate": 0.00019998666208519982,
-      "loss": 46.0,
-      "step": 32294
-    },
-    {
-      "epoch": 5.20089375578727,
-      "grad_norm": 0.00999845378100872,
-      "learning_rate": 0.000199986661258919,
-      "loss": 46.0,
-      "step": 32295
-    },
-    {
-      "epoch": 5.201054792866057,
-      "grad_norm": 0.014367658644914627,
-      "learning_rate": 0.0001999866604326126,
-      "loss": 46.0,
-      "step": 32296
-    },
-    {
-      "epoch": 5.201215829944845,
-      "grad_norm": 0.0010793162509799004,
-      "learning_rate": 0.00019998665960628063,
-      "loss": 46.0,
-      "step": 32297
-    },
-    {
-      "epoch": 5.201376867023632,
-      "grad_norm": 0.005863463971763849,
-      "learning_rate": 0.000199986658779923,
-      "loss": 46.0,
-      "step": 32298
-    },
-    {
-      "epoch": 5.20153790410242,
-      "grad_norm": 0.004061252810060978,
-      "learning_rate": 0.00019998665795353986,
-      "loss": 46.0,
-      "step": 32299
-    },
-    {
-      "epoch": 5.201698941181207,
-      "grad_norm": 0.012200465425848961,
-      "learning_rate": 0.00019998665712713106,
-      "loss": 46.0,
-      "step": 32300
-    },
-    {
-      "epoch": 5.201859978259995,
-      "grad_norm": 0.00080094113945961,
-      "learning_rate": 0.00019998665630069674,
-      "loss": 46.0,
-      "step": 32301
-    },
-    {
-      "epoch": 5.202021015338782,
-      "grad_norm": 0.0070303515531122684,
-      "learning_rate": 0.00019998665547423677,
-      "loss": 46.0,
-      "step": 32302
-    },
-    {
-      "epoch": 5.2021820524175695,
-      "grad_norm": 0.0014221030287444592,
-      "learning_rate": 0.00019998665464775124,
-      "loss": 46.0,
-      "step": 32303
-    },
-    {
-      "epoch": 5.202343089496357,
-      "grad_norm": 0.006524026393890381,
-      "learning_rate": 0.0001999866538212401,
-      "loss": 46.0,
-      "step": 32304
-    },
-    {
-      "epoch": 5.202504126575144,
-      "grad_norm": 0.003949557896703482,
-      "learning_rate": 0.00019998665299470337,
-      "loss": 46.0,
-      "step": 32305
-    },
-    {
-      "epoch": 5.202665163653931,
-      "grad_norm": 0.003168280702084303,
-      "learning_rate": 0.00019998665216814106,
-      "loss": 46.0,
-      "step": 32306
-    },
-    {
-      "epoch": 5.202826200732718,
-      "grad_norm": 0.001541856792755425,
-      "learning_rate": 0.00019998665134155315,
-      "loss": 46.0,
-      "step": 32307
-    },
-    {
-      "epoch": 5.202987237811506,
-      "grad_norm": 0.009249972179532051,
-      "learning_rate": 0.00019998665051493966,
-      "loss": 46.0,
-      "step": 32308
-    },
-    {
-      "epoch": 5.203148274890293,
-      "grad_norm": 0.0028835709672421217,
-      "learning_rate": 0.00019998664968830056,
-      "loss": 46.0,
-      "step": 32309
-    },
-    {
-      "epoch": 5.203309311969081,
-      "grad_norm": 0.003160462249070406,
-      "learning_rate": 0.0001999866488616359,
-      "loss": 46.0,
-      "step": 32310
-    },
-    {
-      "epoch": 5.203470349047868,
-      "grad_norm": 0.012947812676429749,
-      "learning_rate": 0.00019998664803494562,
-      "loss": 46.0,
-      "step": 32311
-    },
-    {
-      "epoch": 5.203631386126656,
-      "grad_norm": 0.005402395036071539,
-      "learning_rate": 0.00019998664720822972,
-      "loss": 46.0,
-      "step": 32312
-    },
-    {
-      "epoch": 5.203792423205443,
-      "grad_norm": 0.014578593894839287,
-      "learning_rate": 0.0001999866463814883,
-      "loss": 46.0,
-      "step": 32313
-    },
-    {
-      "epoch": 5.203953460284231,
-      "grad_norm": 0.0032998265232890844,
-      "learning_rate": 0.00019998664555472123,
-      "loss": 46.0,
-      "step": 32314
-    },
-    {
-      "epoch": 5.204114497363018,
-      "grad_norm": 0.00633576326072216,
-      "learning_rate": 0.0001999866447279286,
-      "loss": 46.0,
-      "step": 32315
-    },
-    {
-      "epoch": 5.2042755344418055,
-      "grad_norm": 0.0073896548710763454,
-      "learning_rate": 0.00019998664390111036,
-      "loss": 46.0,
-      "step": 32316
-    },
-    {
-      "epoch": 5.204436571520593,
-      "grad_norm": 0.006716161035001278,
-      "learning_rate": 0.00019998664307426656,
-      "loss": 46.0,
-      "step": 32317
-    },
-    {
-      "epoch": 5.20459760859938,
-      "grad_norm": 0.007239084225147963,
-      "learning_rate": 0.00019998664224739714,
-      "loss": 46.0,
-      "step": 32318
-    },
-    {
-      "epoch": 5.204758645678168,
-      "grad_norm": 0.003839393611997366,
-      "learning_rate": 0.00019998664142050214,
-      "loss": 46.0,
-      "step": 32319
-    },
-    {
-      "epoch": 5.204919682756954,
-      "grad_norm": 0.0026037844363600016,
-      "learning_rate": 0.00019998664059358152,
-      "loss": 46.0,
-      "step": 32320
-    },
-    {
-      "epoch": 5.205080719835742,
-      "grad_norm": 0.013562245294451714,
-      "learning_rate": 0.00019998663976663535,
-      "loss": 46.0,
-      "step": 32321
-    },
-    {
-      "epoch": 5.205241756914529,
-      "grad_norm": 0.0033752210438251495,
-      "learning_rate": 0.00019998663893966355,
-      "loss": 46.0,
-      "step": 32322
-    },
-    {
-      "epoch": 5.205402793993317,
-      "grad_norm": 0.008500763215124607,
-      "learning_rate": 0.00019998663811266617,
-      "loss": 46.0,
-      "step": 32323
-    },
-    {
-      "epoch": 5.205563831072104,
-      "grad_norm": 0.003265803214162588,
-      "learning_rate": 0.00019998663728564324,
-      "loss": 46.0,
-      "step": 32324
-    },
-    {
-      "epoch": 5.205724868150892,
-      "grad_norm": 0.002039253478869796,
-      "learning_rate": 0.00019998663645859468,
-      "loss": 46.0,
-      "step": 32325
-    },
-    {
-      "epoch": 5.205885905229679,
-      "grad_norm": 0.004213891923427582,
-      "learning_rate": 0.00019998663563152054,
-      "loss": 46.0,
-      "step": 32326
-    },
-    {
-      "epoch": 5.2060469423084665,
-      "grad_norm": 0.0023148925974965096,
-      "learning_rate": 0.0001999866348044208,
-      "loss": 46.0,
-      "step": 32327
-    },
-    {
-      "epoch": 5.206207979387254,
-      "grad_norm": 0.010200661607086658,
-      "learning_rate": 0.00019998663397729547,
-      "loss": 46.0,
-      "step": 32328
-    },
-    {
-      "epoch": 5.206369016466041,
-      "grad_norm": 0.0037055055145174265,
-      "learning_rate": 0.00019998663315014454,
-      "loss": 46.0,
-      "step": 32329
-    },
-    {
-      "epoch": 5.206530053544829,
-      "grad_norm": 0.0033718347549438477,
-      "learning_rate": 0.00019998663232296803,
-      "loss": 46.0,
-      "step": 32330
-    },
-    {
-      "epoch": 5.206691090623616,
-      "grad_norm": 0.00187536736484617,
-      "learning_rate": 0.00019998663149576592,
-      "loss": 46.0,
-      "step": 32331
-    },
-    {
-      "epoch": 5.206852127702404,
-      "grad_norm": 0.02099069394171238,
-      "learning_rate": 0.00019998663066853823,
-      "loss": 46.0,
-      "step": 32332
-    },
-    {
-      "epoch": 5.207013164781191,
-      "grad_norm": 0.007152533624321222,
-      "learning_rate": 0.00019998662984128495,
-      "loss": 46.0,
-      "step": 32333
-    },
-    {
-      "epoch": 5.207174201859978,
-      "grad_norm": 0.0036567014176398516,
-      "learning_rate": 0.0001999866290140061,
-      "loss": 46.0,
-      "step": 32334
-    },
-    {
-      "epoch": 5.207335238938765,
-      "grad_norm": 0.003594242734834552,
-      "learning_rate": 0.0001999866281867016,
-      "loss": 46.0,
-      "step": 32335
-    },
-    {
-      "epoch": 5.207496276017553,
-      "grad_norm": 0.0075663370080292225,
-      "learning_rate": 0.00019998662735937154,
-      "loss": 46.0,
-      "step": 32336
-    },
-    {
-      "epoch": 5.20765731309634,
-      "grad_norm": 0.005721709690988064,
-      "learning_rate": 0.0001999866265320159,
-      "loss": 46.0,
-      "step": 32337
-    },
-    {
-      "epoch": 5.207818350175128,
-      "grad_norm": 0.004641632083803415,
-      "learning_rate": 0.00019998662570463465,
-      "loss": 46.0,
-      "step": 32338
-    },
-    {
-      "epoch": 5.207979387253915,
-      "grad_norm": 0.0018439932027831674,
-      "learning_rate": 0.00019998662487722782,
-      "loss": 46.0,
-      "step": 32339
-    },
-    {
-      "epoch": 5.2081404243327025,
-      "grad_norm": 0.00573316216468811,
-      "learning_rate": 0.0001999866240497954,
-      "loss": 46.0,
-      "step": 32340
-    },
-    {
-      "epoch": 5.20830146141149,
-      "grad_norm": 0.007510887458920479,
-      "learning_rate": 0.00019998662322233738,
-      "loss": 46.0,
-      "step": 32341
-    },
-    {
-      "epoch": 5.208462498490277,
-      "grad_norm": 0.00951984990388155,
-      "learning_rate": 0.00019998662239485376,
-      "loss": 46.0,
-      "step": 32342
-    },
-    {
-      "epoch": 5.208623535569065,
-      "grad_norm": 0.0018545144703239202,
-      "learning_rate": 0.00019998662156734458,
-      "loss": 46.0,
-      "step": 32343
-    },
-    {
-      "epoch": 5.208784572647852,
-      "grad_norm": 0.00960698164999485,
-      "learning_rate": 0.0001999866207398098,
-      "loss": 46.0,
-      "step": 32344
-    },
-    {
-      "epoch": 5.20894560972664,
-      "grad_norm": 0.008608940988779068,
-      "learning_rate": 0.00019998661991224941,
-      "loss": 46.0,
-      "step": 32345
-    },
-    {
-      "epoch": 5.209106646805427,
-      "grad_norm": 0.005764259956777096,
-      "learning_rate": 0.0001999866190846634,
-      "loss": 46.0,
-      "step": 32346
-    },
-    {
-      "epoch": 5.209267683884215,
-      "grad_norm": 0.0076390160247683525,
-      "learning_rate": 0.00019998661825705184,
-      "loss": 46.0,
-      "step": 32347
-    },
-    {
-      "epoch": 5.209428720963002,
-      "grad_norm": 0.006310062017291784,
-      "learning_rate": 0.0001999866174294147,
-      "loss": 46.0,
-      "step": 32348
-    },
-    {
-      "epoch": 5.209589758041789,
-      "grad_norm": 0.0040053133852779865,
-      "learning_rate": 0.00019998661660175192,
-      "loss": 46.0,
-      "step": 32349
-    },
-    {
-      "epoch": 5.209750795120576,
-      "grad_norm": 0.002568659605458379,
-      "learning_rate": 0.0001999866157740636,
-      "loss": 46.0,
-      "step": 32350
-    },
-    {
-      "epoch": 5.209911832199364,
-      "grad_norm": 0.01154275145381689,
-      "learning_rate": 0.00019998661494634968,
-      "loss": 46.0,
-      "step": 32351
-    },
-    {
-      "epoch": 5.210072869278151,
-      "grad_norm": 0.012137651443481445,
-      "learning_rate": 0.00019998661411861014,
-      "loss": 46.0,
-      "step": 32352
-    },
-    {
-      "epoch": 5.2102339063569385,
-      "grad_norm": 0.003844765480607748,
-      "learning_rate": 0.00019998661329084504,
-      "loss": 46.0,
-      "step": 32353
-    },
-    {
-      "epoch": 5.210394943435726,
-      "grad_norm": 0.006151413079351187,
-      "learning_rate": 0.00019998661246305432,
-      "loss": 46.0,
-      "step": 32354
-    },
-    {
-      "epoch": 5.210555980514513,
-      "grad_norm": 0.0009811024647206068,
-      "learning_rate": 0.00019998661163523801,
-      "loss": 46.0,
-      "step": 32355
-    },
-    {
-      "epoch": 5.210717017593301,
-      "grad_norm": 0.02592357061803341,
-      "learning_rate": 0.00019998661080739612,
-      "loss": 46.0,
-      "step": 32356
-    },
-    {
-      "epoch": 5.210878054672088,
-      "grad_norm": 0.023151839151978493,
-      "learning_rate": 0.00019998660997952867,
-      "loss": 46.0,
-      "step": 32357
-    },
-    {
-      "epoch": 5.211039091750876,
-      "grad_norm": 0.003335629589855671,
-      "learning_rate": 0.00019998660915163558,
-      "loss": 46.0,
-      "step": 32358
-    },
-    {
-      "epoch": 5.211200128829663,
-      "grad_norm": 0.007951409555971622,
-      "learning_rate": 0.00019998660832371693,
-      "loss": 46.0,
-      "step": 32359
-    },
-    {
-      "epoch": 5.211361165908451,
-      "grad_norm": 0.008815682493150234,
-      "learning_rate": 0.00019998660749577266,
-      "loss": 46.0,
-      "step": 32360
-    },
-    {
-      "epoch": 5.211522202987238,
-      "grad_norm": 0.0033207014203071594,
-      "learning_rate": 0.00019998660666780283,
-      "loss": 46.0,
-      "step": 32361
-    },
-    {
-      "epoch": 5.211683240066026,
-      "grad_norm": 0.004477128852158785,
-      "learning_rate": 0.0001999866058398074,
-      "loss": 46.0,
-      "step": 32362
-    },
-    {
-      "epoch": 5.211844277144813,
-      "grad_norm": 0.0022156641352921724,
-      "learning_rate": 0.00019998660501178636,
-      "loss": 46.0,
-      "step": 32363
-    },
-    {
-      "epoch": 5.2120053142236,
-      "grad_norm": 0.00433557340875268,
-      "learning_rate": 0.00019998660418373975,
-      "loss": 46.0,
-      "step": 32364
-    },
-    {
-      "epoch": 5.212166351302387,
-      "grad_norm": 0.0011965023586526513,
-      "learning_rate": 0.00019998660335566752,
-      "loss": 46.0,
-      "step": 32365
-    },
-    {
-      "epoch": 5.2123273883811745,
-      "grad_norm": 0.011845373548567295,
-      "learning_rate": 0.00019998660252756973,
-      "loss": 46.0,
-      "step": 32366
-    },
-    {
-      "epoch": 5.212488425459962,
-      "grad_norm": 0.00622065644711256,
-      "learning_rate": 0.00019998660169944632,
-      "loss": 46.0,
-      "step": 32367
-    },
-    {
-      "epoch": 5.212649462538749,
-      "grad_norm": 0.004135714378207922,
-      "learning_rate": 0.00019998660087129736,
-      "loss": 46.0,
-      "step": 32368
-    },
-    {
-      "epoch": 5.212810499617537,
-      "grad_norm": 0.00586656853556633,
-      "learning_rate": 0.00019998660004312278,
-      "loss": 46.0,
-      "step": 32369
-    },
-    {
-      "epoch": 5.212971536696324,
-      "grad_norm": 0.0019765174947679043,
-      "learning_rate": 0.0001999865992149226,
-      "loss": 46.0,
-      "step": 32370
-    },
-    {
-      "epoch": 5.213132573775112,
-      "grad_norm": 0.005308425985276699,
-      "learning_rate": 0.00019998659838669683,
-      "loss": 46.0,
-      "step": 32371
-    },
-    {
-      "epoch": 5.213293610853899,
-      "grad_norm": 0.0031161182560026646,
-      "learning_rate": 0.0001999865975584455,
-      "loss": 46.0,
-      "step": 32372
-    },
-    {
-      "epoch": 5.213454647932687,
-      "grad_norm": 0.0043140072375535965,
-      "learning_rate": 0.00019998659673016854,
-      "loss": 46.0,
-      "step": 32373
-    },
-    {
-      "epoch": 5.213615685011474,
-      "grad_norm": 0.0022940579801797867,
-      "learning_rate": 0.00019998659590186603,
-      "loss": 46.0,
-      "step": 32374
-    },
-    {
-      "epoch": 5.2137767220902616,
-      "grad_norm": 0.0057122004218399525,
-      "learning_rate": 0.00019998659507353787,
-      "loss": 46.0,
-      "step": 32375
-    },
-    {
-      "epoch": 5.213937759169049,
-      "grad_norm": 0.010407740250229836,
-      "learning_rate": 0.00019998659424518418,
-      "loss": 46.0,
-      "step": 32376
-    },
-    {
-      "epoch": 5.2140987962478365,
-      "grad_norm": 0.0037147626280784607,
-      "learning_rate": 0.00019998659341680485,
-      "loss": 46.0,
-      "step": 32377
-    },
-    {
-      "epoch": 5.214259833326624,
-      "grad_norm": 0.0026124482974410057,
-      "learning_rate": 0.00019998659258839996,
-      "loss": 46.0,
-      "step": 32378
-    },
-    {
-      "epoch": 5.2144208704054105,
-      "grad_norm": 0.0011078146053478122,
-      "learning_rate": 0.00019998659175996948,
-      "loss": 46.0,
-      "step": 32379
-    },
-    {
-      "epoch": 5.214581907484198,
-      "grad_norm": 0.0126641308888793,
-      "learning_rate": 0.00019998659093151336,
-      "loss": 46.0,
-      "step": 32380
-    },
-    {
-      "epoch": 5.214742944562985,
-      "grad_norm": 0.014025966636836529,
-      "learning_rate": 0.0001999865901030317,
-      "loss": 46.0,
-      "step": 32381
-    },
-    {
-      "epoch": 5.214903981641773,
-      "grad_norm": 0.005641443654894829,
-      "learning_rate": 0.00019998658927452442,
-      "loss": 46.0,
-      "step": 32382
-    },
-    {
-      "epoch": 5.21506501872056,
-      "grad_norm": 0.0034278188832104206,
-      "learning_rate": 0.0001999865884459916,
-      "loss": 46.0,
-      "step": 32383
-    },
-    {
-      "epoch": 5.215226055799348,
-      "grad_norm": 0.00983754824846983,
-      "learning_rate": 0.00019998658761743315,
-      "loss": 46.0,
-      "step": 32384
-    },
-    {
-      "epoch": 5.215387092878135,
-      "grad_norm": 0.010542117990553379,
-      "learning_rate": 0.0001999865867888491,
-      "loss": 46.0,
-      "step": 32385
-    },
-    {
-      "epoch": 5.215548129956923,
-      "grad_norm": 0.0038853143341839314,
-      "learning_rate": 0.00019998658596023946,
-      "loss": 46.0,
-      "step": 32386
-    },
-    {
-      "epoch": 5.21570916703571,
-      "grad_norm": 0.0027490260545164347,
-      "learning_rate": 0.00019998658513160425,
-      "loss": 46.0,
-      "step": 32387
-    },
-    {
-      "epoch": 5.2158702041144975,
-      "grad_norm": 0.006599657703191042,
-      "learning_rate": 0.0001999865843029434,
-      "loss": 46.0,
-      "step": 32388
-    },
-    {
-      "epoch": 5.216031241193285,
-      "grad_norm": 0.010191729292273521,
-      "learning_rate": 0.000199986583474257,
-      "loss": 46.0,
-      "step": 32389
-    },
-    {
-      "epoch": 5.216192278272072,
-      "grad_norm": 0.0063103921711444855,
-      "learning_rate": 0.00019998658264554502,
-      "loss": 46.0,
-      "step": 32390
-    },
-    {
-      "epoch": 5.21635331535086,
-      "grad_norm": 0.0011862819083034992,
-      "learning_rate": 0.00019998658181680744,
-      "loss": 46.0,
-      "step": 32391
-    },
-    {
-      "epoch": 5.216514352429647,
-      "grad_norm": 0.005997427739202976,
-      "learning_rate": 0.00019998658098804425,
-      "loss": 46.0,
-      "step": 32392
-    },
-    {
-      "epoch": 5.216675389508434,
-      "grad_norm": 0.004797309171408415,
-      "learning_rate": 0.0001999865801592555,
-      "loss": 46.0,
-      "step": 32393
-    },
-    {
-      "epoch": 5.216836426587221,
-      "grad_norm": 0.01907212659716606,
-      "learning_rate": 0.0001999865793304411,
-      "loss": 46.0,
-      "step": 32394
-    },
-    {
-      "epoch": 5.216997463666009,
-      "grad_norm": 0.031080951914191246,
-      "learning_rate": 0.00019998657850160118,
-      "loss": 46.0,
-      "step": 32395
-    },
-    {
-      "epoch": 5.217158500744796,
-      "grad_norm": 0.0020203404128551483,
-      "learning_rate": 0.0001999865776727356,
-      "loss": 46.0,
-      "step": 32396
-    },
-    {
-      "epoch": 5.217319537823584,
-      "grad_norm": 0.013640389777719975,
-      "learning_rate": 0.00019998657684384448,
-      "loss": 46.0,
-      "step": 32397
-    },
-    {
-      "epoch": 5.217480574902371,
-      "grad_norm": 0.0028299863915890455,
-      "learning_rate": 0.00019998657601492776,
-      "loss": 46.0,
-      "step": 32398
-    },
-    {
-      "epoch": 5.217641611981159,
-      "grad_norm": 0.007689146790653467,
-      "learning_rate": 0.00019998657518598544,
-      "loss": 46.0,
-      "step": 32399
-    },
-    {
-      "epoch": 5.217802649059946,
-      "grad_norm": 0.003342656884342432,
-      "learning_rate": 0.00019998657435701752,
-      "loss": 46.0,
-      "step": 32400
-    },
-    {
-      "epoch": 5.2179636861387335,
-      "grad_norm": 0.0022177011705935,
-      "learning_rate": 0.00019998657352802402,
-      "loss": 46.0,
-      "step": 32401
-    },
-    {
-      "epoch": 5.218124723217521,
-      "grad_norm": 0.00589415431022644,
-      "learning_rate": 0.00019998657269900492,
-      "loss": 46.0,
-      "step": 32402
-    },
-    {
-      "epoch": 5.218285760296308,
-      "grad_norm": 0.016138875856995583,
-      "learning_rate": 0.00019998657186996025,
-      "loss": 46.0,
-      "step": 32403
-    },
-    {
-      "epoch": 5.218446797375096,
-      "grad_norm": 0.005968274548649788,
-      "learning_rate": 0.00019998657104088995,
-      "loss": 46.0,
-      "step": 32404
-    },
-    {
-      "epoch": 5.218607834453883,
-      "grad_norm": 0.006998554803431034,
-      "learning_rate": 0.0001999865702117941,
-      "loss": 46.0,
-      "step": 32405
-    },
-    {
-      "epoch": 5.218768871532671,
-      "grad_norm": 0.002928377129137516,
-      "learning_rate": 0.0001999865693826726,
-      "loss": 46.0,
-      "step": 32406
-    },
-    {
-      "epoch": 5.218929908611457,
-      "grad_norm": 0.0024839399848133326,
-      "learning_rate": 0.00019998656855352558,
-      "loss": 46.0,
-      "step": 32407
-    },
-    {
-      "epoch": 5.219090945690245,
-      "grad_norm": 0.0053073749877512455,
-      "learning_rate": 0.00019998656772435291,
-      "loss": 46.0,
-      "step": 32408
-    },
-    {
-      "epoch": 5.219251982769032,
-      "grad_norm": 0.004488674458116293,
-      "learning_rate": 0.00019998656689515469,
-      "loss": 46.0,
-      "step": 32409
-    },
-    {
-      "epoch": 5.21941301984782,
-      "grad_norm": 0.00266914046369493,
-      "learning_rate": 0.00019998656606593087,
-      "loss": 46.0,
-      "step": 32410
-    },
-    {
-      "epoch": 5.219574056926607,
-      "grad_norm": 0.00374097959138453,
-      "learning_rate": 0.00019998656523668147,
-      "loss": 46.0,
-      "step": 32411
-    },
-    {
-      "epoch": 5.219735094005395,
-      "grad_norm": 0.0014494635397568345,
-      "learning_rate": 0.00019998656440740645,
-      "loss": 46.0,
-      "step": 32412
-    },
-    {
-      "epoch": 5.219896131084182,
-      "grad_norm": 0.0036898301914334297,
-      "learning_rate": 0.00019998656357810585,
-      "loss": 46.0,
-      "step": 32413
-    },
-    {
-      "epoch": 5.2200571681629695,
-      "grad_norm": 0.002010650932788849,
-      "learning_rate": 0.00019998656274877966,
-      "loss": 46.0,
-      "step": 32414
-    },
-    {
-      "epoch": 5.220218205241757,
-      "grad_norm": 0.006113533396273851,
-      "learning_rate": 0.00019998656191942785,
-      "loss": 46.0,
-      "step": 32415
-    },
-    {
-      "epoch": 5.220379242320544,
-      "grad_norm": 0.004528458695858717,
-      "learning_rate": 0.00019998656109005049,
-      "loss": 46.0,
-      "step": 32416
-    },
-    {
-      "epoch": 5.220540279399332,
-      "grad_norm": 0.0020464423578232527,
-      "learning_rate": 0.00019998656026064753,
-      "loss": 46.0,
-      "step": 32417
-    },
-    {
-      "epoch": 5.220701316478119,
-      "grad_norm": 0.003225229447707534,
-      "learning_rate": 0.00019998655943121897,
-      "loss": 46.0,
-      "step": 32418
-    },
-    {
-      "epoch": 5.220862353556907,
-      "grad_norm": 0.0021926143672317266,
-      "learning_rate": 0.0001999865586017648,
-      "loss": 46.0,
-      "step": 32419
-    },
-    {
-      "epoch": 5.221023390635694,
-      "grad_norm": 0.006882375106215477,
-      "learning_rate": 0.00019998655777228507,
-      "loss": 46.0,
-      "step": 32420
-    },
-    {
-      "epoch": 5.221184427714482,
-      "grad_norm": 0.0009305743151344359,
-      "learning_rate": 0.00019998655694277974,
-      "loss": 46.0,
-      "step": 32421
-    },
-    {
-      "epoch": 5.221345464793268,
-      "grad_norm": 0.002799306297674775,
-      "learning_rate": 0.00019998655611324883,
-      "loss": 46.0,
-      "step": 32422
-    },
-    {
-      "epoch": 5.221506501872056,
-      "grad_norm": 0.0060835410840809345,
-      "learning_rate": 0.0001999865552836923,
-      "loss": 46.0,
-      "step": 32423
-    },
-    {
-      "epoch": 5.221667538950843,
-      "grad_norm": 0.0019642033148556948,
-      "learning_rate": 0.00019998655445411018,
-      "loss": 46.0,
-      "step": 32424
-    },
-    {
-      "epoch": 5.221828576029631,
-      "grad_norm": 0.00123660359531641,
-      "learning_rate": 0.0001999865536245025,
-      "loss": 46.0,
-      "step": 32425
-    },
-    {
-      "epoch": 5.221989613108418,
-      "grad_norm": 0.015515579842031002,
-      "learning_rate": 0.0001999865527948692,
-      "loss": 46.0,
-      "step": 32426
-    },
-    {
-      "epoch": 5.2221506501872055,
-      "grad_norm": 0.006554416846483946,
-      "learning_rate": 0.00019998655196521033,
-      "loss": 46.0,
-      "step": 32427
-    },
-    {
-      "epoch": 5.222311687265993,
-      "grad_norm": 0.0029202012810856104,
-      "learning_rate": 0.00019998655113552584,
-      "loss": 46.0,
-      "step": 32428
-    },
-    {
-      "epoch": 5.22247272434478,
-      "grad_norm": 0.0010103315580636263,
-      "learning_rate": 0.00019998655030581581,
-      "loss": 46.0,
-      "step": 32429
-    },
-    {
-      "epoch": 5.222633761423568,
-      "grad_norm": 0.024140065535902977,
-      "learning_rate": 0.00019998654947608015,
-      "loss": 46.0,
-      "step": 32430
-    },
-    {
-      "epoch": 5.222794798502355,
-      "grad_norm": 0.00398417329415679,
-      "learning_rate": 0.0001999865486463189,
-      "loss": 46.0,
-      "step": 32431
-    },
-    {
-      "epoch": 5.222955835581143,
-      "grad_norm": 0.006070380099117756,
-      "learning_rate": 0.00019998654781653205,
-      "loss": 46.0,
-      "step": 32432
-    },
-    {
-      "epoch": 5.22311687265993,
-      "grad_norm": 0.0076614501886069775,
-      "learning_rate": 0.00019998654698671962,
-      "loss": 46.0,
-      "step": 32433
-    },
-    {
-      "epoch": 5.223277909738718,
-      "grad_norm": 0.004210289567708969,
-      "learning_rate": 0.00019998654615688158,
-      "loss": 46.0,
-      "step": 32434
-    },
-    {
-      "epoch": 5.223438946817505,
-      "grad_norm": 0.0008750613196752965,
-      "learning_rate": 0.000199986545327018,
-      "loss": 46.0,
-      "step": 32435
-    },
-    {
-      "epoch": 5.2235999838962925,
-      "grad_norm": 0.0032051270827651024,
-      "learning_rate": 0.0001999865444971288,
-      "loss": 46.0,
-      "step": 32436
-    },
-    {
-      "epoch": 5.223761020975079,
-      "grad_norm": 0.004105807282030582,
-      "learning_rate": 0.000199986543667214,
-      "loss": 46.0,
-      "step": 32437
-    },
-    {
-      "epoch": 5.223922058053867,
-      "grad_norm": 0.02621007151901722,
-      "learning_rate": 0.00019998654283727362,
-      "loss": 46.0,
-      "step": 32438
-    },
-    {
-      "epoch": 5.224083095132654,
-      "grad_norm": 0.0033084868919104338,
-      "learning_rate": 0.0001999865420073076,
-      "loss": 46.0,
-      "step": 32439
-    },
-    {
-      "epoch": 5.2242441322114415,
-      "grad_norm": 0.007322421297430992,
-      "learning_rate": 0.00019998654117731607,
-      "loss": 46.0,
-      "step": 32440
-    },
-    {
-      "epoch": 5.224405169290229,
-      "grad_norm": 0.006035187281668186,
-      "learning_rate": 0.00019998654034729892,
-      "loss": 46.0,
-      "step": 32441
-    },
-    {
-      "epoch": 5.224566206369016,
-      "grad_norm": 0.0019522428046911955,
-      "learning_rate": 0.00019998653951725615,
-      "loss": 46.0,
-      "step": 32442
-    },
-    {
-      "epoch": 5.224727243447804,
-      "grad_norm": 0.004291558638215065,
-      "learning_rate": 0.0001999865386871878,
-      "loss": 46.0,
-      "step": 32443
-    },
-    {
-      "epoch": 5.224888280526591,
-      "grad_norm": 0.007751860190182924,
-      "learning_rate": 0.00019998653785709386,
-      "loss": 46.0,
-      "step": 32444
-    },
-    {
-      "epoch": 5.225049317605379,
-      "grad_norm": 0.015379723161458969,
-      "learning_rate": 0.00019998653702697436,
-      "loss": 46.0,
-      "step": 32445
-    },
-    {
-      "epoch": 5.225210354684166,
-      "grad_norm": 0.003933741711080074,
-      "learning_rate": 0.00019998653619682924,
-      "loss": 46.0,
-      "step": 32446
-    },
-    {
-      "epoch": 5.225371391762954,
-      "grad_norm": 0.007469188887625933,
-      "learning_rate": 0.00019998653536665854,
-      "loss": 46.0,
-      "step": 32447
-    },
-    {
-      "epoch": 5.225532428841741,
-      "grad_norm": 0.005155201535671949,
-      "learning_rate": 0.00019998653453646222,
-      "loss": 46.0,
-      "step": 32448
-    },
-    {
-      "epoch": 5.2256934659205285,
-      "grad_norm": 0.0011988560436293483,
-      "learning_rate": 0.00019998653370624031,
-      "loss": 46.0,
-      "step": 32449
-    },
-    {
-      "epoch": 5.225854502999316,
-      "grad_norm": 0.0026310391258448362,
-      "learning_rate": 0.00019998653287599285,
-      "loss": 46.0,
-      "step": 32450
-    },
-    {
-      "epoch": 5.226015540078103,
-      "grad_norm": 0.008084907196462154,
-      "learning_rate": 0.00019998653204571974,
-      "loss": 46.0,
-      "step": 32451
-    },
-    {
-      "epoch": 5.22617657715689,
-      "grad_norm": 0.004009178373962641,
-      "learning_rate": 0.0001999865312154211,
-      "loss": 46.0,
-      "step": 32452
-    },
-    {
-      "epoch": 5.226337614235677,
-      "grad_norm": 0.013323145918548107,
-      "learning_rate": 0.00019998653038509685,
-      "loss": 46.0,
-      "step": 32453
-    },
-    {
-      "epoch": 5.226498651314465,
-      "grad_norm": 0.002848051954060793,
-      "learning_rate": 0.00019998652955474698,
-      "loss": 46.0,
-      "step": 32454
-    },
-    {
-      "epoch": 5.226659688393252,
-      "grad_norm": 0.002332792617380619,
-      "learning_rate": 0.00019998652872437156,
-      "loss": 46.0,
-      "step": 32455
-    },
-    {
-      "epoch": 5.22682072547204,
-      "grad_norm": 0.021011171862483025,
-      "learning_rate": 0.0001999865278939705,
-      "loss": 46.0,
-      "step": 32456
-    },
-    {
-      "epoch": 5.226981762550827,
-      "grad_norm": 0.0024412842467427254,
-      "learning_rate": 0.0001999865270635439,
-      "loss": 46.0,
-      "step": 32457
-    },
-    {
-      "epoch": 5.227142799629615,
-      "grad_norm": 0.011419717222452164,
-      "learning_rate": 0.0001999865262330917,
-      "loss": 46.0,
-      "step": 32458
-    },
-    {
-      "epoch": 5.227303836708402,
-      "grad_norm": 0.011326770298182964,
-      "learning_rate": 0.00019998652540261386,
-      "loss": 46.0,
-      "step": 32459
-    },
-    {
-      "epoch": 5.22746487378719,
-      "grad_norm": 0.0008496668888255954,
-      "learning_rate": 0.00019998652457211047,
-      "loss": 46.0,
-      "step": 32460
-    },
-    {
-      "epoch": 5.227625910865977,
-      "grad_norm": 0.002399976132437587,
-      "learning_rate": 0.0001999865237415815,
-      "loss": 46.0,
-      "step": 32461
-    },
-    {
-      "epoch": 5.2277869479447645,
-      "grad_norm": 0.0010485263774171472,
-      "learning_rate": 0.00019998652291102693,
-      "loss": 46.0,
-      "step": 32462
-    },
-    {
-      "epoch": 5.227947985023552,
-      "grad_norm": 0.006310475990176201,
-      "learning_rate": 0.00019998652208044675,
-      "loss": 46.0,
-      "step": 32463
-    },
-    {
-      "epoch": 5.228109022102339,
-      "grad_norm": 0.005123792216181755,
-      "learning_rate": 0.00019998652124984098,
-      "loss": 46.0,
-      "step": 32464
-    },
-    {
-      "epoch": 5.228270059181127,
-      "grad_norm": 0.02080768719315529,
-      "learning_rate": 0.00019998652041920963,
-      "loss": 46.0,
-      "step": 32465
-    },
-    {
-      "epoch": 5.228431096259913,
-      "grad_norm": 0.0018601218471303582,
-      "learning_rate": 0.0001999865195885527,
-      "loss": 46.0,
-      "step": 32466
-    },
-    {
-      "epoch": 5.228592133338701,
-      "grad_norm": 0.0029565643053501844,
-      "learning_rate": 0.00019998651875787014,
-      "loss": 46.0,
-      "step": 32467
-    },
-    {
-      "epoch": 5.228753170417488,
-      "grad_norm": 0.0011048543965443969,
-      "learning_rate": 0.00019998651792716202,
-      "loss": 46.0,
-      "step": 32468
-    },
-    {
-      "epoch": 5.228914207496276,
-      "grad_norm": 0.004862792789936066,
-      "learning_rate": 0.00019998651709642832,
-      "loss": 46.0,
-      "step": 32469
-    },
-    {
-      "epoch": 5.229075244575063,
-      "grad_norm": 0.007608181331306696,
-      "learning_rate": 0.00019998651626566897,
-      "loss": 46.0,
-      "step": 32470
-    },
-    {
-      "epoch": 5.229236281653851,
-      "grad_norm": 0.0020849204156547785,
-      "learning_rate": 0.00019998651543488407,
-      "loss": 46.0,
-      "step": 32471
-    },
-    {
-      "epoch": 5.229397318732638,
-      "grad_norm": 0.009699077345430851,
-      "learning_rate": 0.00019998651460407358,
-      "loss": 46.0,
-      "step": 32472
-    },
-    {
-      "epoch": 5.229558355811426,
-      "grad_norm": 0.0067525506019592285,
-      "learning_rate": 0.0001999865137732375,
-      "loss": 46.0,
-      "step": 32473
-    },
-    {
-      "epoch": 5.229719392890213,
-      "grad_norm": 0.002224602969363332,
-      "learning_rate": 0.00019998651294237584,
-      "loss": 46.0,
-      "step": 32474
-    },
-    {
-      "epoch": 5.2298804299690005,
-      "grad_norm": 0.009725421667098999,
-      "learning_rate": 0.00019998651211148856,
-      "loss": 46.0,
-      "step": 32475
-    },
-    {
-      "epoch": 5.230041467047788,
-      "grad_norm": 0.004606904461979866,
-      "learning_rate": 0.0001999865112805757,
-      "loss": 46.0,
-      "step": 32476
-    },
-    {
-      "epoch": 5.230202504126575,
-      "grad_norm": 0.004499922040849924,
-      "learning_rate": 0.00019998651044963724,
-      "loss": 46.0,
-      "step": 32477
-    },
-    {
-      "epoch": 5.230363541205363,
-      "grad_norm": 0.0031274249777197838,
-      "learning_rate": 0.00019998650961867317,
-      "loss": 46.0,
-      "step": 32478
-    },
-    {
-      "epoch": 5.23052457828415,
-      "grad_norm": 0.0032461306545883417,
-      "learning_rate": 0.00019998650878768354,
-      "loss": 46.0,
-      "step": 32479
-    },
-    {
-      "epoch": 5.230685615362938,
-      "grad_norm": 0.0018505065236240625,
-      "learning_rate": 0.00019998650795666832,
-      "loss": 46.0,
-      "step": 32480
-    },
-    {
-      "epoch": 5.230846652441724,
-      "grad_norm": 0.0035431047435849905,
-      "learning_rate": 0.00019998650712562752,
-      "loss": 46.0,
-      "step": 32481
-    },
-    {
-      "epoch": 5.231007689520512,
-      "grad_norm": 0.0033981550950556993,
-      "learning_rate": 0.0001999865062945611,
-      "loss": 46.0,
-      "step": 32482
-    },
-    {
-      "epoch": 5.231168726599299,
-      "grad_norm": 0.003470989875495434,
-      "learning_rate": 0.0001999865054634691,
-      "loss": 46.0,
-      "step": 32483
-    },
-    {
-      "epoch": 5.231329763678087,
-      "grad_norm": 0.008644778281450272,
-      "learning_rate": 0.0001999865046323515,
-      "loss": 46.0,
-      "step": 32484
-    },
-    {
-      "epoch": 5.231490800756874,
-      "grad_norm": 0.0020351686980575323,
-      "learning_rate": 0.00019998650380120833,
-      "loss": 46.0,
-      "step": 32485
-    },
-    {
-      "epoch": 5.231651837835662,
-      "grad_norm": 0.002281755208969116,
-      "learning_rate": 0.00019998650297003957,
-      "loss": 46.0,
-      "step": 32486
-    },
-    {
-      "epoch": 5.231812874914449,
-      "grad_norm": 0.0020727887749671936,
-      "learning_rate": 0.00019998650213884519,
-      "loss": 46.0,
-      "step": 32487
-    },
-    {
-      "epoch": 5.2319739119932365,
-      "grad_norm": 0.0014243468176573515,
-      "learning_rate": 0.00019998650130762525,
-      "loss": 46.0,
-      "step": 32488
-    },
-    {
-      "epoch": 5.232134949072024,
-      "grad_norm": 0.0031394653487950563,
-      "learning_rate": 0.0001999865004763797,
-      "loss": 46.0,
-      "step": 32489
-    },
-    {
-      "epoch": 5.232295986150811,
-      "grad_norm": 0.0024528398644179106,
-      "learning_rate": 0.00019998649964510855,
-      "loss": 46.0,
-      "step": 32490
-    },
-    {
-      "epoch": 5.232457023229599,
-      "grad_norm": 0.006793811451643705,
-      "learning_rate": 0.00019998649881381182,
-      "loss": 46.0,
-      "step": 32491
-    },
-    {
-      "epoch": 5.232618060308386,
-      "grad_norm": 0.0043118358589708805,
-      "learning_rate": 0.0001999864979824895,
-      "loss": 46.0,
-      "step": 32492
-    },
-    {
-      "epoch": 5.232779097387174,
-      "grad_norm": 0.007221119478344917,
-      "learning_rate": 0.0001999864971511416,
-      "loss": 46.0,
-      "step": 32493
-    },
-    {
-      "epoch": 5.232940134465961,
-      "grad_norm": 0.012481052428483963,
-      "learning_rate": 0.00019998649631976806,
-      "loss": 46.0,
-      "step": 32494
-    },
-    {
-      "epoch": 5.233101171544748,
-      "grad_norm": 0.0038691589143127203,
-      "learning_rate": 0.00019998649548836895,
-      "loss": 46.0,
-      "step": 32495
-    },
-    {
-      "epoch": 5.233262208623535,
-      "grad_norm": 0.002644366817548871,
-      "learning_rate": 0.0001999864946569443,
-      "loss": 46.0,
-      "step": 32496
-    },
-    {
-      "epoch": 5.233423245702323,
-      "grad_norm": 0.005674370098859072,
-      "learning_rate": 0.000199986493825494,
-      "loss": 46.0,
-      "step": 32497
-    },
-    {
-      "epoch": 5.23358428278111,
-      "grad_norm": 0.008699550293385983,
-      "learning_rate": 0.00019998649299401815,
-      "loss": 46.0,
-      "step": 32498
-    },
-    {
-      "epoch": 5.2337453198598975,
-      "grad_norm": 0.007826943881809711,
-      "learning_rate": 0.00019998649216251667,
-      "loss": 46.0,
-      "step": 32499
-    },
-    {
-      "epoch": 5.233906356938685,
-      "grad_norm": 0.007207693997770548,
-      "learning_rate": 0.00019998649133098963,
-      "loss": 46.0,
-      "step": 32500
-    },
-    {
-      "epoch": 5.2340673940174725,
-      "grad_norm": 0.0035860768985003233,
-      "learning_rate": 0.00019998649049943697,
-      "loss": 46.0,
-      "step": 32501
-    },
-    {
-      "epoch": 5.23422843109626,
-      "grad_norm": 0.0014584943419322371,
-      "learning_rate": 0.00019998648966785873,
-      "loss": 46.0,
-      "step": 32502
-    },
-    {
-      "epoch": 5.234389468175047,
-      "grad_norm": 0.0023908785078674555,
-      "learning_rate": 0.0001999864888362549,
-      "loss": 46.0,
-      "step": 32503
-    },
-    {
-      "epoch": 5.234550505253835,
-      "grad_norm": 0.013223525136709213,
-      "learning_rate": 0.00019998648800462549,
-      "loss": 46.0,
-      "step": 32504
-    },
-    {
-      "epoch": 5.234711542332622,
-      "grad_norm": 0.002594653284177184,
-      "learning_rate": 0.00019998648717297045,
-      "loss": 46.0,
-      "step": 32505
-    },
-    {
-      "epoch": 5.23487257941141,
-      "grad_norm": 0.0025989434216171503,
-      "learning_rate": 0.00019998648634128986,
-      "loss": 46.0,
-      "step": 32506
-    },
-    {
-      "epoch": 5.235033616490197,
-      "grad_norm": 0.004163894336670637,
-      "learning_rate": 0.0001999864855095837,
-      "loss": 46.0,
-      "step": 32507
-    },
-    {
-      "epoch": 5.235194653568985,
-      "grad_norm": 0.006227575708180666,
-      "learning_rate": 0.0001999864846778519,
-      "loss": 46.0,
-      "step": 32508
-    },
-    {
-      "epoch": 5.235355690647772,
-      "grad_norm": 0.0022246032021939754,
-      "learning_rate": 0.00019998648384609452,
-      "loss": 46.0,
-      "step": 32509
-    },
-    {
-      "epoch": 5.235516727726559,
-      "grad_norm": 0.0065980893559753895,
-      "learning_rate": 0.00019998648301431152,
-      "loss": 46.0,
-      "step": 32510
-    },
-    {
-      "epoch": 5.235677764805346,
-      "grad_norm": 0.0042998818680644035,
-      "learning_rate": 0.000199986482182503,
-      "loss": 46.0,
-      "step": 32511
-    },
-    {
-      "epoch": 5.2358388018841335,
-      "grad_norm": 0.009482955560088158,
-      "learning_rate": 0.00019998648135066883,
-      "loss": 46.0,
-      "step": 32512
-    },
-    {
-      "epoch": 5.235999838962921,
-      "grad_norm": 0.006263263523578644,
-      "learning_rate": 0.0001999864805188091,
-      "loss": 46.0,
-      "step": 32513
-    },
-    {
-      "epoch": 5.236160876041708,
-      "grad_norm": 0.009389905259013176,
-      "learning_rate": 0.00019998647968692373,
-      "loss": 46.0,
-      "step": 32514
-    },
-    {
-      "epoch": 5.236321913120496,
-      "grad_norm": 0.0022588225547224283,
-      "learning_rate": 0.00019998647885501286,
-      "loss": 46.0,
-      "step": 32515
-    },
-    {
-      "epoch": 5.236482950199283,
-      "grad_norm": 0.001993166981264949,
-      "learning_rate": 0.00019998647802307634,
-      "loss": 46.0,
-      "step": 32516
-    },
-    {
-      "epoch": 5.236643987278071,
-      "grad_norm": 0.010135490447282791,
-      "learning_rate": 0.0001999864771911142,
-      "loss": 46.0,
-      "step": 32517
-    },
-    {
-      "epoch": 5.236805024356858,
-      "grad_norm": 0.001166346250101924,
-      "learning_rate": 0.0001999864763591265,
-      "loss": 46.0,
-      "step": 32518
-    },
-    {
-      "epoch": 5.236966061435646,
-      "grad_norm": 0.005321672651916742,
-      "learning_rate": 0.00019998647552711322,
-      "loss": 46.0,
-      "step": 32519
-    },
-    {
-      "epoch": 5.237127098514433,
-      "grad_norm": 0.006660795770585537,
-      "learning_rate": 0.00019998647469507432,
-      "loss": 46.0,
-      "step": 32520
-    },
-    {
-      "epoch": 5.237288135593221,
-      "grad_norm": 0.004549749661237001,
-      "learning_rate": 0.00019998647386300987,
-      "loss": 46.0,
-      "step": 32521
-    },
-    {
-      "epoch": 5.237449172672008,
-      "grad_norm": 0.020205644890666008,
-      "learning_rate": 0.0001999864730309198,
-      "loss": 46.0,
-      "step": 32522
-    },
-    {
-      "epoch": 5.2376102097507955,
-      "grad_norm": 0.018811682239174843,
-      "learning_rate": 0.00019998647219880415,
-      "loss": 46.0,
-      "step": 32523
-    },
-    {
-      "epoch": 5.237771246829583,
-      "grad_norm": 0.005193459801375866,
-      "learning_rate": 0.0001999864713666629,
-      "loss": 46.0,
-      "step": 32524
-    },
-    {
-      "epoch": 5.2379322839083695,
-      "grad_norm": 0.011831403709948063,
-      "learning_rate": 0.00019998647053449603,
-      "loss": 46.0,
-      "step": 32525
-    },
-    {
-      "epoch": 5.238093320987157,
-      "grad_norm": 0.002070151036605239,
-      "learning_rate": 0.00019998646970230364,
-      "loss": 46.0,
-      "step": 32526
-    },
-    {
-      "epoch": 5.238254358065944,
-      "grad_norm": 0.0035476593766361475,
-      "learning_rate": 0.0001999864688700856,
-      "loss": 46.0,
-      "step": 32527
-    },
-    {
-      "epoch": 5.238415395144732,
-      "grad_norm": 0.0023394327145069838,
-      "learning_rate": 0.000199986468037842,
-      "loss": 46.0,
-      "step": 32528
-    },
-    {
-      "epoch": 5.238576432223519,
-      "grad_norm": 0.0028536287136375904,
-      "learning_rate": 0.00019998646720557277,
-      "loss": 46.0,
-      "step": 32529
-    },
-    {
-      "epoch": 5.238737469302307,
-      "grad_norm": 0.004566666204482317,
-      "learning_rate": 0.00019998646637327797,
-      "loss": 46.0,
-      "step": 32530
-    },
-    {
-      "epoch": 5.238898506381094,
-      "grad_norm": 0.008364316076040268,
-      "learning_rate": 0.00019998646554095762,
-      "loss": 46.0,
-      "step": 32531
-    },
-    {
-      "epoch": 5.239059543459882,
-      "grad_norm": 0.008646366186439991,
-      "learning_rate": 0.00019998646470861163,
-      "loss": 46.0,
-      "step": 32532
-    },
-    {
-      "epoch": 5.239220580538669,
-      "grad_norm": 0.0030348661821335554,
-      "learning_rate": 0.00019998646387624005,
-      "loss": 46.0,
-      "step": 32533
-    },
-    {
-      "epoch": 5.239381617617457,
-      "grad_norm": 0.0020744376815855503,
-      "learning_rate": 0.00019998646304384289,
-      "loss": 46.0,
-      "step": 32534
-    },
-    {
-      "epoch": 5.239542654696244,
-      "grad_norm": 0.012931537814438343,
-      "learning_rate": 0.00019998646221142013,
-      "loss": 46.0,
-      "step": 32535
-    },
-    {
-      "epoch": 5.2397036917750315,
-      "grad_norm": 0.0035900769289582968,
-      "learning_rate": 0.0001999864613789718,
-      "loss": 46.0,
-      "step": 32536
-    },
-    {
-      "epoch": 5.239864728853819,
-      "grad_norm": 0.003176985075697303,
-      "learning_rate": 0.00019998646054649786,
-      "loss": 46.0,
-      "step": 32537
-    },
-    {
-      "epoch": 5.240025765932606,
-      "grad_norm": 0.003611267078667879,
-      "learning_rate": 0.00019998645971399832,
-      "loss": 46.0,
-      "step": 32538
-    },
-    {
-      "epoch": 5.240186803011393,
-      "grad_norm": 0.008464580401778221,
-      "learning_rate": 0.00019998645888147322,
-      "loss": 46.0,
-      "step": 32539
-    },
-    {
-      "epoch": 5.24034784009018,
-      "grad_norm": 0.009467690251767635,
-      "learning_rate": 0.00019998645804892247,
-      "loss": 46.0,
-      "step": 32540
-    },
-    {
-      "epoch": 5.240508877168968,
-      "grad_norm": 0.008480178192257881,
-      "learning_rate": 0.0001999864572163462,
-      "loss": 46.0,
-      "step": 32541
-    },
-    {
-      "epoch": 5.240669914247755,
-      "grad_norm": 0.0017373107839375734,
-      "learning_rate": 0.0001999864563837443,
-      "loss": 46.0,
-      "step": 32542
-    },
-    {
-      "epoch": 5.240830951326543,
-      "grad_norm": 0.0016843137564137578,
-      "learning_rate": 0.0001999864555511168,
-      "loss": 46.0,
-      "step": 32543
-    },
-    {
-      "epoch": 5.24099198840533,
-      "grad_norm": 0.02414952963590622,
-      "learning_rate": 0.00019998645471846374,
-      "loss": 46.0,
-      "step": 32544
-    },
-    {
-      "epoch": 5.241153025484118,
-      "grad_norm": 0.0015188314719125628,
-      "learning_rate": 0.00019998645388578508,
-      "loss": 46.0,
-      "step": 32545
-    },
-    {
-      "epoch": 5.241314062562905,
-      "grad_norm": 0.0023643230088055134,
-      "learning_rate": 0.00019998645305308082,
-      "loss": 46.0,
-      "step": 32546
-    },
-    {
-      "epoch": 5.241475099641693,
-      "grad_norm": 0.004094649571925402,
-      "learning_rate": 0.00019998645222035096,
-      "loss": 46.0,
-      "step": 32547
-    },
-    {
-      "epoch": 5.24163613672048,
-      "grad_norm": 0.018110547214746475,
-      "learning_rate": 0.00019998645138759552,
-      "loss": 46.0,
-      "step": 32548
-    },
-    {
-      "epoch": 5.2417971737992675,
-      "grad_norm": 0.010099576786160469,
-      "learning_rate": 0.0001999864505548145,
-      "loss": 46.0,
-      "step": 32549
-    },
-    {
-      "epoch": 5.241958210878055,
-      "grad_norm": 0.00549853453412652,
-      "learning_rate": 0.00019998644972200785,
-      "loss": 46.0,
-      "step": 32550
-    },
-    {
-      "epoch": 5.242119247956842,
-      "grad_norm": 0.002000484149903059,
-      "learning_rate": 0.00019998644888917565,
-      "loss": 46.0,
-      "step": 32551
-    },
-    {
-      "epoch": 5.24228028503563,
-      "grad_norm": 0.008779608644545078,
-      "learning_rate": 0.00019998644805631783,
-      "loss": 46.0,
-      "step": 32552
-    },
-    {
-      "epoch": 5.242441322114417,
-      "grad_norm": 0.004706146195530891,
-      "learning_rate": 0.00019998644722343443,
-      "loss": 46.0,
-      "step": 32553
-    },
-    {
-      "epoch": 5.242602359193204,
-      "grad_norm": 0.00858042761683464,
-      "learning_rate": 0.00019998644639052543,
-      "loss": 46.0,
-      "step": 32554
-    },
-    {
-      "epoch": 5.242763396271991,
-      "grad_norm": 0.006062607280910015,
-      "learning_rate": 0.00019998644555759088,
-      "loss": 46.0,
-      "step": 32555
-    },
-    {
-      "epoch": 5.242924433350779,
-      "grad_norm": 0.0026606328319758177,
-      "learning_rate": 0.0001999864447246307,
-      "loss": 46.0,
-      "step": 32556
-    },
-    {
-      "epoch": 5.243085470429566,
-      "grad_norm": 0.002813649130985141,
-      "learning_rate": 0.0001999864438916449,
-      "loss": 46.0,
-      "step": 32557
-    },
-    {
-      "epoch": 5.243246507508354,
-      "grad_norm": 0.005954826716333628,
-      "learning_rate": 0.00019998644305863357,
-      "loss": 46.0,
-      "step": 32558
-    },
-    {
-      "epoch": 5.243407544587141,
-      "grad_norm": 0.0022507337853312492,
-      "learning_rate": 0.00019998644222559664,
-      "loss": 46.0,
-      "step": 32559
-    },
-    {
-      "epoch": 5.2435685816659285,
-      "grad_norm": 0.007777234073728323,
-      "learning_rate": 0.00019998644139253407,
-      "loss": 46.0,
-      "step": 32560
-    },
-    {
-      "epoch": 5.243729618744716,
-      "grad_norm": 0.010295560583472252,
-      "learning_rate": 0.00019998644055944595,
-      "loss": 46.0,
-      "step": 32561
-    },
-    {
-      "epoch": 5.243890655823503,
-      "grad_norm": 0.005532537121325731,
-      "learning_rate": 0.00019998643972633223,
-      "loss": 46.0,
-      "step": 32562
-    },
-    {
-      "epoch": 5.244051692902291,
-      "grad_norm": 0.0031475473660975695,
-      "learning_rate": 0.00019998643889319293,
-      "loss": 46.0,
-      "step": 32563
-    },
-    {
-      "epoch": 5.244212729981078,
-      "grad_norm": 0.0032074186019599438,
-      "learning_rate": 0.00019998643806002804,
-      "loss": 46.0,
-      "step": 32564
-    },
-    {
-      "epoch": 5.244373767059866,
-      "grad_norm": 0.02486119233071804,
-      "learning_rate": 0.00019998643722683753,
-      "loss": 46.0,
-      "step": 32565
-    },
-    {
-      "epoch": 5.244534804138653,
-      "grad_norm": 0.010372549295425415,
-      "learning_rate": 0.0001999864363936214,
-      "loss": 46.0,
-      "step": 32566
-    },
-    {
-      "epoch": 5.244695841217441,
-      "grad_norm": 0.0032242003362625837,
-      "learning_rate": 0.00019998643556037976,
-      "loss": 46.0,
-      "step": 32567
-    },
-    {
-      "epoch": 5.244856878296227,
-      "grad_norm": 0.0028462721966207027,
-      "learning_rate": 0.0001999864347271125,
-      "loss": 46.0,
-      "step": 32568
-    },
-    {
-      "epoch": 5.245017915375015,
-      "grad_norm": 0.005855824798345566,
-      "learning_rate": 0.00019998643389381964,
-      "loss": 46.0,
-      "step": 32569
-    },
-    {
-      "epoch": 5.245178952453802,
-      "grad_norm": 0.0014157752739265561,
-      "learning_rate": 0.00019998643306050118,
-      "loss": 46.0,
-      "step": 32570
-    },
-    {
-      "epoch": 5.24533998953259,
-      "grad_norm": 0.001425734139047563,
-      "learning_rate": 0.00019998643222715715,
-      "loss": 46.0,
-      "step": 32571
-    },
-    {
-      "epoch": 5.245501026611377,
-      "grad_norm": 0.002818556036800146,
-      "learning_rate": 0.0001999864313937875,
-      "loss": 46.0,
-      "step": 32572
-    },
-    {
-      "epoch": 5.2456620636901645,
-      "grad_norm": 0.009476062841713428,
-      "learning_rate": 0.00019998643056039228,
-      "loss": 46.0,
-      "step": 32573
-    },
-    {
-      "epoch": 5.245823100768952,
-      "grad_norm": 0.002413427224382758,
-      "learning_rate": 0.00019998642972697146,
-      "loss": 46.0,
-      "step": 32574
-    },
-    {
-      "epoch": 5.245984137847739,
-      "grad_norm": 0.00271716364659369,
-      "learning_rate": 0.00019998642889352506,
-      "loss": 46.0,
-      "step": 32575
-    },
-    {
-      "epoch": 5.246145174926527,
-      "grad_norm": 0.00868601631373167,
-      "learning_rate": 0.00019998642806005307,
-      "loss": 46.0,
-      "step": 32576
-    },
-    {
-      "epoch": 5.246306212005314,
-      "grad_norm": 0.014675287529826164,
-      "learning_rate": 0.00019998642722655546,
-      "loss": 46.0,
-      "step": 32577
-    },
-    {
-      "epoch": 5.246467249084102,
-      "grad_norm": 0.01291351392865181,
-      "learning_rate": 0.00019998642639303227,
-      "loss": 46.0,
-      "step": 32578
-    },
-    {
-      "epoch": 5.246628286162889,
-      "grad_norm": 0.006505636032670736,
-      "learning_rate": 0.0001999864255594835,
-      "loss": 46.0,
-      "step": 32579
-    },
-    {
-      "epoch": 5.246789323241677,
-      "grad_norm": 0.003626520512625575,
-      "learning_rate": 0.00019998642472590913,
-      "loss": 46.0,
-      "step": 32580
-    },
-    {
-      "epoch": 5.246950360320464,
-      "grad_norm": 0.00314791570417583,
-      "learning_rate": 0.00019998642389230917,
-      "loss": 46.0,
-      "step": 32581
-    },
-    {
-      "epoch": 5.247111397399252,
-      "grad_norm": 0.0026092049665749073,
-      "learning_rate": 0.0001999864230586836,
-      "loss": 46.0,
-      "step": 32582
-    },
-    {
-      "epoch": 5.247272434478038,
-      "grad_norm": 0.003367555793374777,
-      "learning_rate": 0.00019998642222503248,
-      "loss": 46.0,
-      "step": 32583
-    },
-    {
-      "epoch": 5.247433471556826,
-      "grad_norm": 0.0025751274079084396,
-      "learning_rate": 0.00019998642139135576,
-      "loss": 46.0,
-      "step": 32584
-    },
-    {
-      "epoch": 5.247594508635613,
-      "grad_norm": 0.007634031120687723,
-      "learning_rate": 0.00019998642055765343,
-      "loss": 46.0,
-      "step": 32585
-    },
-    {
-      "epoch": 5.2477555457144005,
-      "grad_norm": 0.004878594074398279,
-      "learning_rate": 0.00019998641972392552,
-      "loss": 46.0,
-      "step": 32586
-    },
-    {
-      "epoch": 5.247916582793188,
-      "grad_norm": 0.01824226602911949,
-      "learning_rate": 0.000199986418890172,
-      "loss": 46.0,
-      "step": 32587
-    },
-    {
-      "epoch": 5.248077619871975,
-      "grad_norm": 0.005828591994941235,
-      "learning_rate": 0.0001999864180563929,
-      "loss": 46.0,
-      "step": 32588
-    },
-    {
-      "epoch": 5.248238656950763,
-      "grad_norm": 0.005131609737873077,
-      "learning_rate": 0.00019998641722258822,
-      "loss": 46.0,
-      "step": 32589
-    },
-    {
-      "epoch": 5.24839969402955,
-      "grad_norm": 0.005229932721704245,
-      "learning_rate": 0.00019998641638875792,
-      "loss": 46.0,
-      "step": 32590
-    },
-    {
-      "epoch": 5.248560731108338,
-      "grad_norm": 0.002291000448167324,
-      "learning_rate": 0.00019998641555490205,
-      "loss": 46.0,
-      "step": 32591
-    },
-    {
-      "epoch": 5.248721768187125,
-      "grad_norm": 0.02242285944521427,
-      "learning_rate": 0.0001999864147210206,
-      "loss": 46.0,
-      "step": 32592
-    },
-    {
-      "epoch": 5.248882805265913,
-      "grad_norm": 0.011595788411796093,
-      "learning_rate": 0.00019998641388711355,
-      "loss": 46.0,
-      "step": 32593
-    },
-    {
-      "epoch": 5.2490438423447,
-      "grad_norm": 0.003237377852201462,
-      "learning_rate": 0.0001999864130531809,
-      "loss": 46.0,
-      "step": 32594
-    },
-    {
-      "epoch": 5.249204879423488,
-      "grad_norm": 0.0025449118111282587,
-      "learning_rate": 0.00019998641221922268,
-      "loss": 46.0,
-      "step": 32595
-    },
-    {
-      "epoch": 5.249365916502275,
-      "grad_norm": 0.011958865448832512,
-      "learning_rate": 0.00019998641138523884,
-      "loss": 46.0,
-      "step": 32596
-    },
-    {
-      "epoch": 5.2495269535810625,
-      "grad_norm": 0.00785353034734726,
-      "learning_rate": 0.0001999864105512294,
-      "loss": 46.0,
-      "step": 32597
-    },
-    {
-      "epoch": 5.249687990659849,
-      "grad_norm": 0.005678877700120211,
-      "learning_rate": 0.0001999864097171944,
-      "loss": 46.0,
-      "step": 32598
-    },
-    {
-      "epoch": 5.2498490277386365,
-      "grad_norm": 0.0008249414386227727,
-      "learning_rate": 0.0001999864088831338,
-      "loss": 46.0,
-      "step": 32599
-    },
-    {
-      "epoch": 5.250010064817424,
-      "grad_norm": 0.005462795961648226,
-      "learning_rate": 0.0001999864080490476,
-      "loss": 46.0,
-      "step": 32600
-    },
-    {
-      "epoch": 5.250171101896211,
-      "grad_norm": 0.010967371053993702,
-      "learning_rate": 0.00019998640721493582,
-      "loss": 46.0,
-      "step": 32601
-    },
-    {
-      "epoch": 5.250332138974999,
-      "grad_norm": 0.033050745725631714,
-      "learning_rate": 0.00019998640638079843,
-      "loss": 46.0,
-      "step": 32602
-    },
-    {
-      "epoch": 5.250493176053786,
-      "grad_norm": 0.019316645339131355,
-      "learning_rate": 0.00019998640554663548,
-      "loss": 46.0,
-      "step": 32603
-    },
-    {
-      "epoch": 5.250654213132574,
-      "grad_norm": 0.004457741975784302,
-      "learning_rate": 0.0001999864047124469,
-      "loss": 46.0,
-      "step": 32604
-    },
-    {
-      "epoch": 5.250815250211361,
-      "grad_norm": 0.0012783827260136604,
-      "learning_rate": 0.00019998640387823278,
-      "loss": 46.0,
-      "step": 32605
-    },
-    {
-      "epoch": 5.250976287290149,
-      "grad_norm": 0.010230284184217453,
-      "learning_rate": 0.00019998640304399304,
-      "loss": 46.0,
-      "step": 32606
-    },
-    {
-      "epoch": 5.251137324368936,
-      "grad_norm": 0.006025716662406921,
-      "learning_rate": 0.0001999864022097277,
-      "loss": 46.0,
-      "step": 32607
-    },
-    {
-      "epoch": 5.2512983614477236,
-      "grad_norm": 0.008783235214650631,
-      "learning_rate": 0.00019998640137543677,
-      "loss": 46.0,
-      "step": 32608
-    },
-    {
-      "epoch": 5.251459398526511,
-      "grad_norm": 0.001808101311326027,
-      "learning_rate": 0.00019998640054112024,
-      "loss": 46.0,
-      "step": 32609
-    },
-    {
-      "epoch": 5.2516204356052985,
-      "grad_norm": 0.0016007708618417382,
-      "learning_rate": 0.00019998639970677813,
-      "loss": 46.0,
-      "step": 32610
-    },
-    {
-      "epoch": 5.251781472684086,
-      "grad_norm": 0.01337799709290266,
-      "learning_rate": 0.00019998639887241045,
-      "loss": 46.0,
-      "step": 32611
-    },
-    {
-      "epoch": 5.251942509762873,
-      "grad_norm": 0.002692866139113903,
-      "learning_rate": 0.00019998639803801713,
-      "loss": 46.0,
-      "step": 32612
-    },
-    {
-      "epoch": 5.25210354684166,
-      "grad_norm": 0.004638901446014643,
-      "learning_rate": 0.00019998639720359825,
-      "loss": 46.0,
-      "step": 32613
-    },
-    {
-      "epoch": 5.252264583920447,
-      "grad_norm": 0.008056741207838058,
-      "learning_rate": 0.00019998639636915376,
-      "loss": 46.0,
-      "step": 32614
-    },
-    {
-      "epoch": 5.252425620999235,
-      "grad_norm": 0.006555595900863409,
-      "learning_rate": 0.00019998639553468373,
-      "loss": 46.0,
-      "step": 32615
-    },
-    {
-      "epoch": 5.252586658078022,
-      "grad_norm": 0.005346778780221939,
-      "learning_rate": 0.00019998639470018804,
-      "loss": 46.0,
-      "step": 32616
-    },
-    {
-      "epoch": 5.25274769515681,
-      "grad_norm": 0.0016766526969149709,
-      "learning_rate": 0.0001999863938656668,
-      "loss": 46.0,
-      "step": 32617
-    },
-    {
-      "epoch": 5.252908732235597,
-      "grad_norm": 0.0030667639803141356,
-      "learning_rate": 0.00019998639303111997,
-      "loss": 46.0,
-      "step": 32618
-    },
-    {
-      "epoch": 5.253069769314385,
-      "grad_norm": 0.009963980875909328,
-      "learning_rate": 0.00019998639219654757,
-      "loss": 46.0,
-      "step": 32619
-    },
-    {
-      "epoch": 5.253230806393172,
-      "grad_norm": 0.007708756718784571,
-      "learning_rate": 0.00019998639136194953,
-      "loss": 46.0,
-      "step": 32620
-    },
-    {
-      "epoch": 5.2533918434719595,
-      "grad_norm": 0.002629245398566127,
-      "learning_rate": 0.0001999863905273259,
-      "loss": 46.0,
-      "step": 32621
-    },
-    {
-      "epoch": 5.253552880550747,
-      "grad_norm": 0.005356887821108103,
-      "learning_rate": 0.0001999863896926767,
-      "loss": 46.0,
-      "step": 32622
-    },
-    {
-      "epoch": 5.253713917629534,
-      "grad_norm": 0.0031015139538794756,
-      "learning_rate": 0.00019998638885800193,
-      "loss": 46.0,
-      "step": 32623
-    },
-    {
-      "epoch": 5.253874954708322,
-      "grad_norm": 0.0010287582408636808,
-      "learning_rate": 0.0001999863880233015,
-      "loss": 46.0,
-      "step": 32624
-    },
-    {
-      "epoch": 5.254035991787109,
-      "grad_norm": 0.015437331981956959,
-      "learning_rate": 0.00019998638718857553,
-      "loss": 46.0,
-      "step": 32625
-    },
-    {
-      "epoch": 5.254197028865896,
-      "grad_norm": 0.007384417578577995,
-      "learning_rate": 0.00019998638635382397,
-      "loss": 46.0,
-      "step": 32626
-    },
-    {
-      "epoch": 5.254358065944683,
-      "grad_norm": 0.012291919440031052,
-      "learning_rate": 0.00019998638551904681,
-      "loss": 46.0,
-      "step": 32627
-    },
-    {
-      "epoch": 5.254519103023471,
-      "grad_norm": 0.008701611310243607,
-      "learning_rate": 0.00019998638468424407,
-      "loss": 46.0,
-      "step": 32628
-    },
-    {
-      "epoch": 5.254680140102258,
-      "grad_norm": 0.002207002369686961,
-      "learning_rate": 0.00019998638384941572,
-      "loss": 46.0,
-      "step": 32629
-    },
-    {
-      "epoch": 5.254841177181046,
-      "grad_norm": 0.001916029374115169,
-      "learning_rate": 0.00019998638301456178,
-      "loss": 46.0,
-      "step": 32630
-    },
-    {
-      "epoch": 5.255002214259833,
-      "grad_norm": 0.013337378390133381,
-      "learning_rate": 0.00019998638217968228,
-      "loss": 46.0,
-      "step": 32631
-    },
-    {
-      "epoch": 5.255163251338621,
-      "grad_norm": 0.0057775224559009075,
-      "learning_rate": 0.00019998638134477713,
-      "loss": 46.0,
-      "step": 32632
-    },
-    {
-      "epoch": 5.255324288417408,
-      "grad_norm": 0.00446449825540185,
-      "learning_rate": 0.00019998638050984643,
-      "loss": 46.0,
-      "step": 32633
-    },
-    {
-      "epoch": 5.2554853254961955,
-      "grad_norm": 0.0013590480666607618,
-      "learning_rate": 0.0001999863796748901,
-      "loss": 46.0,
-      "step": 32634
-    },
-    {
-      "epoch": 5.255646362574983,
-      "grad_norm": 0.007231721188873053,
-      "learning_rate": 0.00019998637883990823,
-      "loss": 46.0,
-      "step": 32635
-    },
-    {
-      "epoch": 5.25580739965377,
-      "grad_norm": 0.008585846982896328,
-      "learning_rate": 0.00019998637800490074,
-      "loss": 46.0,
-      "step": 32636
-    },
-    {
-      "epoch": 5.255968436732558,
-      "grad_norm": 0.0019316069083288312,
-      "learning_rate": 0.00019998637716986766,
-      "loss": 46.0,
-      "step": 32637
-    },
-    {
-      "epoch": 5.256129473811345,
-      "grad_norm": 0.0039500887505710125,
-      "learning_rate": 0.00019998637633480902,
-      "loss": 46.0,
-      "step": 32638
-    },
-    {
-      "epoch": 5.256290510890133,
-      "grad_norm": 0.004407184664160013,
-      "learning_rate": 0.00019998637549972474,
-      "loss": 46.0,
-      "step": 32639
-    },
-    {
-      "epoch": 5.25645154796892,
-      "grad_norm": 0.0025003000628203154,
-      "learning_rate": 0.0001999863746646149,
-      "loss": 46.0,
-      "step": 32640
-    },
-    {
-      "epoch": 5.256612585047707,
-      "grad_norm": 0.016539763659238815,
-      "learning_rate": 0.00019998637382947948,
-      "loss": 46.0,
-      "step": 32641
-    },
-    {
-      "epoch": 5.256773622126494,
-      "grad_norm": 0.002038324251770973,
-      "learning_rate": 0.0001999863729943184,
-      "loss": 46.0,
-      "step": 32642
-    },
-    {
-      "epoch": 5.256934659205282,
-      "grad_norm": 0.011387698352336884,
-      "learning_rate": 0.00019998637215913178,
-      "loss": 46.0,
-      "step": 32643
-    },
-    {
-      "epoch": 5.257095696284069,
-      "grad_norm": 0.001516856369562447,
-      "learning_rate": 0.00019998637132391956,
-      "loss": 46.0,
-      "step": 32644
-    },
-    {
-      "epoch": 5.257256733362857,
-      "grad_norm": 0.010770903900265694,
-      "learning_rate": 0.00019998637048868179,
-      "loss": 46.0,
-      "step": 32645
-    },
-    {
-      "epoch": 5.257417770441644,
-      "grad_norm": 0.015435745939612389,
-      "learning_rate": 0.00019998636965341837,
-      "loss": 46.0,
-      "step": 32646
-    },
-    {
-      "epoch": 5.2575788075204315,
-      "grad_norm": 0.0016320951981469989,
-      "learning_rate": 0.0001999863688181294,
-      "loss": 46.0,
-      "step": 32647
-    },
-    {
-      "epoch": 5.257739844599219,
-      "grad_norm": 0.0033576814457774162,
-      "learning_rate": 0.0001999863679828148,
-      "loss": 46.0,
-      "step": 32648
-    },
-    {
-      "epoch": 5.257900881678006,
-      "grad_norm": 0.01032302901148796,
-      "learning_rate": 0.00019998636714747462,
-      "loss": 46.0,
-      "step": 32649
-    },
-    {
-      "epoch": 5.258061918756794,
-      "grad_norm": 0.00802572537213564,
-      "learning_rate": 0.00019998636631210888,
-      "loss": 46.0,
-      "step": 32650
-    },
-    {
-      "epoch": 5.258222955835581,
-      "grad_norm": 0.00871746614575386,
-      "learning_rate": 0.00019998636547671753,
-      "loss": 46.0,
-      "step": 32651
-    },
-    {
-      "epoch": 5.258383992914369,
-      "grad_norm": 0.007459382060915232,
-      "learning_rate": 0.00019998636464130059,
-      "loss": 46.0,
-      "step": 32652
-    },
-    {
-      "epoch": 5.258545029993156,
-      "grad_norm": 0.004042063374072313,
-      "learning_rate": 0.00019998636380585803,
-      "loss": 46.0,
-      "step": 32653
-    },
-    {
-      "epoch": 5.258706067071944,
-      "grad_norm": 0.0008590960642322898,
-      "learning_rate": 0.00019998636297038992,
-      "loss": 46.0,
-      "step": 32654
-    },
-    {
-      "epoch": 5.258867104150731,
-      "grad_norm": 0.003839415730908513,
-      "learning_rate": 0.00019998636213489616,
-      "loss": 46.0,
-      "step": 32655
-    },
-    {
-      "epoch": 5.259028141229518,
-      "grad_norm": 0.004814943298697472,
-      "learning_rate": 0.00019998636129937687,
-      "loss": 46.0,
-      "step": 32656
-    },
-    {
-      "epoch": 5.259189178308305,
-      "grad_norm": 0.0045902058482170105,
-      "learning_rate": 0.00019998636046383197,
-      "loss": 46.0,
-      "step": 32657
-    },
-    {
-      "epoch": 5.259350215387093,
-      "grad_norm": 0.001437442610040307,
-      "learning_rate": 0.00019998635962826148,
-      "loss": 46.0,
-      "step": 32658
-    },
-    {
-      "epoch": 5.25951125246588,
-      "grad_norm": 0.0025768957566469908,
-      "learning_rate": 0.0001999863587926654,
-      "loss": 46.0,
-      "step": 32659
-    },
-    {
-      "epoch": 5.2596722895446675,
-      "grad_norm": 0.003445117501541972,
-      "learning_rate": 0.0001999863579570437,
-      "loss": 46.0,
-      "step": 32660
-    },
-    {
-      "epoch": 5.259833326623455,
-      "grad_norm": 0.010923337191343307,
-      "learning_rate": 0.00019998635712139645,
-      "loss": 46.0,
-      "step": 32661
-    },
-    {
-      "epoch": 5.259994363702242,
-      "grad_norm": 0.006247815676033497,
-      "learning_rate": 0.00019998635628572359,
-      "loss": 46.0,
-      "step": 32662
-    },
-    {
-      "epoch": 5.26015540078103,
-      "grad_norm": 0.00402148487046361,
-      "learning_rate": 0.00019998635545002513,
-      "loss": 46.0,
-      "step": 32663
-    },
-    {
-      "epoch": 5.260316437859817,
-      "grad_norm": 0.004145434591919184,
-      "learning_rate": 0.0001999863546143011,
-      "loss": 46.0,
-      "step": 32664
-    },
-    {
-      "epoch": 5.260477474938605,
-      "grad_norm": 0.012223700992763042,
-      "learning_rate": 0.00019998635377855144,
-      "loss": 46.0,
-      "step": 32665
-    },
-    {
-      "epoch": 5.260638512017392,
-      "grad_norm": 0.007371906656771898,
-      "learning_rate": 0.00019998635294277622,
-      "loss": 46.0,
-      "step": 32666
-    },
-    {
-      "epoch": 5.26079954909618,
-      "grad_norm": 0.005936674773693085,
-      "learning_rate": 0.00019998635210697542,
-      "loss": 46.0,
-      "step": 32667
-    },
-    {
-      "epoch": 5.260960586174967,
-      "grad_norm": 0.0018563737394288182,
-      "learning_rate": 0.000199986351271149,
-      "loss": 46.0,
-      "step": 32668
-    },
-    {
-      "epoch": 5.2611216232537545,
-      "grad_norm": 0.003600636962801218,
-      "learning_rate": 0.000199986350435297,
-      "loss": 46.0,
-      "step": 32669
-    },
-    {
-      "epoch": 5.261282660332542,
-      "grad_norm": 0.004085914231836796,
-      "learning_rate": 0.00019998634959941943,
-      "loss": 46.0,
-      "step": 32670
-    },
-    {
-      "epoch": 5.261443697411329,
-      "grad_norm": 0.004710397217422724,
-      "learning_rate": 0.00019998634876351623,
-      "loss": 46.0,
-      "step": 32671
-    },
-    {
-      "epoch": 5.261604734490116,
-      "grad_norm": 0.0035617873072624207,
-      "learning_rate": 0.00019998634792758746,
-      "loss": 46.0,
-      "step": 32672
-    },
-    {
-      "epoch": 5.2617657715689035,
-      "grad_norm": 0.0027501003351062536,
-      "learning_rate": 0.00019998634709163308,
-      "loss": 46.0,
-      "step": 32673
-    },
-    {
-      "epoch": 5.261926808647691,
-      "grad_norm": 0.0017287508817389607,
-      "learning_rate": 0.00019998634625565314,
-      "loss": 46.0,
-      "step": 32674
-    },
-    {
-      "epoch": 5.262087845726478,
-      "grad_norm": 0.004268570337444544,
-      "learning_rate": 0.0001999863454196476,
-      "loss": 46.0,
-      "step": 32675
-    },
-    {
-      "epoch": 5.262248882805266,
-      "grad_norm": 0.014256948605179787,
-      "learning_rate": 0.00019998634458361645,
-      "loss": 46.0,
-      "step": 32676
-    },
-    {
-      "epoch": 5.262409919884053,
-      "grad_norm": 0.0046277130022645,
-      "learning_rate": 0.00019998634374755972,
-      "loss": 46.0,
-      "step": 32677
-    },
-    {
-      "epoch": 5.262570956962841,
-      "grad_norm": 0.009656930342316628,
-      "learning_rate": 0.0001999863429114774,
-      "loss": 46.0,
-      "step": 32678
-    },
-    {
-      "epoch": 5.262731994041628,
-      "grad_norm": 0.008416050113737583,
-      "learning_rate": 0.00019998634207536947,
-      "loss": 46.0,
-      "step": 32679
-    },
-    {
-      "epoch": 5.262893031120416,
-      "grad_norm": 0.016932500526309013,
-      "learning_rate": 0.00019998634123923598,
-      "loss": 46.0,
-      "step": 32680
-    },
-    {
-      "epoch": 5.263054068199203,
-      "grad_norm": 0.005687073338776827,
-      "learning_rate": 0.00019998634040307688,
-      "loss": 46.0,
-      "step": 32681
-    },
-    {
-      "epoch": 5.2632151052779905,
-      "grad_norm": 0.0048994203098118305,
-      "learning_rate": 0.0001999863395668922,
-      "loss": 46.0,
-      "step": 32682
-    },
-    {
-      "epoch": 5.263376142356778,
-      "grad_norm": 0.005206476431339979,
-      "learning_rate": 0.0001999863387306819,
-      "loss": 46.0,
-      "step": 32683
-    },
-    {
-      "epoch": 5.263537179435565,
-      "grad_norm": 0.0017812447622418404,
-      "learning_rate": 0.00019998633789444602,
-      "loss": 46.0,
-      "step": 32684
-    },
-    {
-      "epoch": 5.263698216514353,
-      "grad_norm": 0.02232889086008072,
-      "learning_rate": 0.00019998633705818456,
-      "loss": 46.0,
-      "step": 32685
-    },
-    {
-      "epoch": 5.263859253593139,
-      "grad_norm": 0.019658392295241356,
-      "learning_rate": 0.00019998633622189752,
-      "loss": 46.0,
-      "step": 32686
-    },
-    {
-      "epoch": 5.264020290671927,
-      "grad_norm": 0.003187777241691947,
-      "learning_rate": 0.00019998633538558487,
-      "loss": 46.0,
-      "step": 32687
-    },
-    {
-      "epoch": 5.264181327750714,
-      "grad_norm": 0.0042627472430467606,
-      "learning_rate": 0.00019998633454924666,
-      "loss": 46.0,
-      "step": 32688
-    },
-    {
-      "epoch": 5.264342364829502,
-      "grad_norm": 0.004050084855407476,
-      "learning_rate": 0.00019998633371288283,
-      "loss": 46.0,
-      "step": 32689
-    },
-    {
-      "epoch": 5.264503401908289,
-      "grad_norm": 0.006821055430918932,
-      "learning_rate": 0.0001999863328764934,
-      "loss": 46.0,
-      "step": 32690
-    },
-    {
-      "epoch": 5.264664438987077,
-      "grad_norm": 0.0021948502399027348,
-      "learning_rate": 0.00019998633204007838,
-      "loss": 46.0,
-      "step": 32691
-    },
-    {
-      "epoch": 5.264825476065864,
-      "grad_norm": 0.0024444195441901684,
-      "learning_rate": 0.0001999863312036378,
-      "loss": 46.0,
-      "step": 32692
-    },
-    {
-      "epoch": 5.264986513144652,
-      "grad_norm": 0.013647926039993763,
-      "learning_rate": 0.0001999863303671716,
-      "loss": 46.0,
-      "step": 32693
-    },
-    {
-      "epoch": 5.265147550223439,
-      "grad_norm": 0.0042956252582371235,
-      "learning_rate": 0.00019998632953067982,
-      "loss": 46.0,
-      "step": 32694
-    },
-    {
-      "epoch": 5.2653085873022265,
-      "grad_norm": 0.0008356008911505342,
-      "learning_rate": 0.00019998632869416242,
-      "loss": 46.0,
-      "step": 32695
-    },
-    {
-      "epoch": 5.265469624381014,
-      "grad_norm": 0.005223304498940706,
-      "learning_rate": 0.00019998632785761945,
-      "loss": 46.0,
-      "step": 32696
-    },
-    {
-      "epoch": 5.265630661459801,
-      "grad_norm": 0.005672079510986805,
-      "learning_rate": 0.00019998632702105092,
-      "loss": 46.0,
-      "step": 32697
-    },
-    {
-      "epoch": 5.265791698538589,
-      "grad_norm": 0.005248211324214935,
-      "learning_rate": 0.00019998632618445676,
-      "loss": 46.0,
-      "step": 32698
-    },
-    {
-      "epoch": 5.265952735617376,
-      "grad_norm": 0.002135050715878606,
-      "learning_rate": 0.00019998632534783703,
-      "loss": 46.0,
-      "step": 32699
-    },
-    {
-      "epoch": 5.266113772696163,
-      "grad_norm": 0.0070891049690544605,
-      "learning_rate": 0.00019998632451119171,
-      "loss": 46.0,
-      "step": 32700
-    },
-    {
-      "epoch": 5.26627480977495,
-      "grad_norm": 0.009737805463373661,
-      "learning_rate": 0.00019998632367452076,
-      "loss": 46.0,
-      "step": 32701
-    },
-    {
-      "epoch": 5.266435846853738,
-      "grad_norm": 0.002563453745096922,
-      "learning_rate": 0.00019998632283782427,
-      "loss": 46.0,
-      "step": 32702
-    },
-    {
-      "epoch": 5.266596883932525,
-      "grad_norm": 0.0024845018051564693,
-      "learning_rate": 0.00019998632200110214,
-      "loss": 46.0,
-      "step": 32703
-    },
-    {
-      "epoch": 5.266757921011313,
-      "grad_norm": 0.002924696309491992,
-      "learning_rate": 0.00019998632116435445,
-      "loss": 46.0,
-      "step": 32704
-    },
-    {
-      "epoch": 5.2669189580901,
-      "grad_norm": 0.002489969599992037,
-      "learning_rate": 0.00019998632032758114,
-      "loss": 46.0,
-      "step": 32705
-    },
-    {
-      "epoch": 5.267079995168888,
-      "grad_norm": 0.004896989557892084,
-      "learning_rate": 0.00019998631949078228,
-      "loss": 46.0,
-      "step": 32706
-    },
-    {
-      "epoch": 5.267241032247675,
-      "grad_norm": 0.007058939430862665,
-      "learning_rate": 0.0001999863186539578,
-      "loss": 46.0,
-      "step": 32707
-    },
-    {
-      "epoch": 5.2674020693264625,
-      "grad_norm": 0.007980616763234138,
-      "learning_rate": 0.00019998631781710776,
-      "loss": 46.0,
-      "step": 32708
-    },
-    {
-      "epoch": 5.26756310640525,
-      "grad_norm": 0.002323438646271825,
-      "learning_rate": 0.00019998631698023208,
-      "loss": 46.0,
-      "step": 32709
-    },
-    {
-      "epoch": 5.267724143484037,
-      "grad_norm": 0.017195073887705803,
-      "learning_rate": 0.00019998631614333087,
-      "loss": 46.0,
-      "step": 32710
-    },
-    {
-      "epoch": 5.267885180562825,
-      "grad_norm": 0.014264289289712906,
-      "learning_rate": 0.000199986315306404,
-      "loss": 46.0,
-      "step": 32711
-    },
-    {
-      "epoch": 5.268046217641612,
-      "grad_norm": 0.002280234592035413,
-      "learning_rate": 0.00019998631446945157,
-      "loss": 46.0,
-      "step": 32712
-    },
-    {
-      "epoch": 5.2682072547204,
-      "grad_norm": 0.004719855729490519,
-      "learning_rate": 0.00019998631363247354,
-      "loss": 46.0,
-      "step": 32713
-    },
-    {
-      "epoch": 5.268368291799186,
-      "grad_norm": 0.001047990401275456,
-      "learning_rate": 0.00019998631279546992,
-      "loss": 46.0,
-      "step": 32714
-    },
-    {
-      "epoch": 5.268529328877974,
-      "grad_norm": 0.008425416424870491,
-      "learning_rate": 0.00019998631195844072,
-      "loss": 46.0,
-      "step": 32715
-    },
-    {
-      "epoch": 5.268690365956761,
-      "grad_norm": 0.005264493636786938,
-      "learning_rate": 0.00019998631112138596,
-      "loss": 46.0,
-      "step": 32716
-    },
-    {
-      "epoch": 5.268851403035549,
-      "grad_norm": 0.005269521847367287,
-      "learning_rate": 0.00019998631028430555,
-      "loss": 46.0,
-      "step": 32717
-    },
-    {
-      "epoch": 5.269012440114336,
-      "grad_norm": 0.005105150863528252,
-      "learning_rate": 0.0001999863094471996,
-      "loss": 46.0,
-      "step": 32718
-    },
-    {
-      "epoch": 5.269173477193124,
-      "grad_norm": 0.008111500181257725,
-      "learning_rate": 0.000199986308610068,
-      "loss": 46.0,
-      "step": 32719
-    },
-    {
-      "epoch": 5.269334514271911,
-      "grad_norm": 0.002483420306816697,
-      "learning_rate": 0.00019998630777291084,
-      "loss": 46.0,
-      "step": 32720
-    },
-    {
-      "epoch": 5.2694955513506985,
-      "grad_norm": 0.009813596494495869,
-      "learning_rate": 0.00019998630693572811,
-      "loss": 46.0,
-      "step": 32721
-    },
-    {
-      "epoch": 5.269656588429486,
-      "grad_norm": 0.01022042240947485,
-      "learning_rate": 0.00019998630609851975,
-      "loss": 46.0,
-      "step": 32722
-    },
-    {
-      "epoch": 5.269817625508273,
-      "grad_norm": 0.008669614791870117,
-      "learning_rate": 0.00019998630526128585,
-      "loss": 46.0,
-      "step": 32723
-    },
-    {
-      "epoch": 5.269978662587061,
-      "grad_norm": 0.008839394897222519,
-      "learning_rate": 0.0001999863044240263,
-      "loss": 46.0,
-      "step": 32724
-    },
-    {
-      "epoch": 5.270139699665848,
-      "grad_norm": 0.00277986703440547,
-      "learning_rate": 0.00019998630358674117,
-      "loss": 46.0,
-      "step": 32725
-    },
-    {
-      "epoch": 5.270300736744636,
-      "grad_norm": 0.0019506255630403757,
-      "learning_rate": 0.00019998630274943046,
-      "loss": 46.0,
-      "step": 32726
-    },
-    {
-      "epoch": 5.270461773823423,
-      "grad_norm": 0.006148990243673325,
-      "learning_rate": 0.00019998630191209418,
-      "loss": 46.0,
-      "step": 32727
-    },
-    {
-      "epoch": 5.270622810902211,
-      "grad_norm": 0.007476849015802145,
-      "learning_rate": 0.0001999863010747323,
-      "loss": 46.0,
-      "step": 32728
-    },
-    {
-      "epoch": 5.270783847980997,
-      "grad_norm": 0.0024931083898991346,
-      "learning_rate": 0.0001999863002373448,
-      "loss": 46.0,
-      "step": 32729
-    },
-    {
-      "epoch": 5.270944885059785,
-      "grad_norm": 0.003910596948117018,
-      "learning_rate": 0.00019998629939993175,
-      "loss": 46.0,
-      "step": 32730
-    },
-    {
-      "epoch": 5.271105922138572,
-      "grad_norm": 0.010569185949862003,
-      "learning_rate": 0.00019998629856249307,
-      "loss": 46.0,
-      "step": 32731
-    },
-    {
-      "epoch": 5.2712669592173595,
-      "grad_norm": 0.0035121862310916185,
-      "learning_rate": 0.00019998629772502883,
-      "loss": 46.0,
-      "step": 32732
-    },
-    {
-      "epoch": 5.271427996296147,
-      "grad_norm": 0.010562081821262836,
-      "learning_rate": 0.00019998629688753897,
-      "loss": 46.0,
-      "step": 32733
-    },
-    {
-      "epoch": 5.2715890333749345,
-      "grad_norm": 0.0009269286529161036,
-      "learning_rate": 0.00019998629605002353,
-      "loss": 46.0,
-      "step": 32734
-    },
-    {
-      "epoch": 5.271750070453722,
-      "grad_norm": 0.0015503076137974858,
-      "learning_rate": 0.0001999862952124825,
-      "loss": 46.0,
-      "step": 32735
-    },
-    {
-      "epoch": 5.271911107532509,
-      "grad_norm": 0.006434526760131121,
-      "learning_rate": 0.00019998629437491586,
-      "loss": 46.0,
-      "step": 32736
-    },
-    {
-      "epoch": 5.272072144611297,
-      "grad_norm": 0.003989649936556816,
-      "learning_rate": 0.00019998629353732366,
-      "loss": 46.0,
-      "step": 32737
-    },
-    {
-      "epoch": 5.272233181690084,
-      "grad_norm": 0.011900383979082108,
-      "learning_rate": 0.00019998629269970587,
-      "loss": 46.0,
-      "step": 32738
-    },
-    {
-      "epoch": 5.272394218768872,
-      "grad_norm": 0.004666908178478479,
-      "learning_rate": 0.00019998629186206246,
-      "loss": 46.0,
-      "step": 32739
-    },
-    {
-      "epoch": 5.272555255847659,
-      "grad_norm": 0.0014495952054858208,
-      "learning_rate": 0.00019998629102439347,
-      "loss": 46.0,
-      "step": 32740
-    },
-    {
-      "epoch": 5.272716292926447,
-      "grad_norm": 0.002782973228022456,
-      "learning_rate": 0.0001999862901866989,
-      "loss": 46.0,
-      "step": 32741
-    },
-    {
-      "epoch": 5.272877330005234,
-      "grad_norm": 0.003483750391751528,
-      "learning_rate": 0.00019998628934897873,
-      "loss": 46.0,
-      "step": 32742
-    },
-    {
-      "epoch": 5.2730383670840215,
-      "grad_norm": 0.0021493907552212477,
-      "learning_rate": 0.00019998628851123295,
-      "loss": 46.0,
-      "step": 32743
-    },
-    {
-      "epoch": 5.273199404162808,
-      "grad_norm": 0.0012872567167505622,
-      "learning_rate": 0.0001999862876734616,
-      "loss": 46.0,
-      "step": 32744
-    },
-    {
-      "epoch": 5.2733604412415955,
-      "grad_norm": 0.00925049651414156,
-      "learning_rate": 0.00019998628683566468,
-      "loss": 46.0,
-      "step": 32745
-    },
-    {
-      "epoch": 5.273521478320383,
-      "grad_norm": 0.0046015591360628605,
-      "learning_rate": 0.00019998628599784211,
-      "loss": 46.0,
-      "step": 32746
-    },
-    {
-      "epoch": 5.27368251539917,
-      "grad_norm": 0.0036442552227526903,
-      "learning_rate": 0.000199986285159994,
-      "loss": 46.0,
-      "step": 32747
-    },
-    {
-      "epoch": 5.273843552477958,
-      "grad_norm": 0.0017966092564165592,
-      "learning_rate": 0.0001999862843221203,
-      "loss": 46.0,
-      "step": 32748
-    },
-    {
-      "epoch": 5.274004589556745,
-      "grad_norm": 0.00784636102616787,
-      "learning_rate": 0.000199986283484221,
-      "loss": 46.0,
-      "step": 32749
-    },
-    {
-      "epoch": 5.274165626635533,
-      "grad_norm": 0.0034893376287072897,
-      "learning_rate": 0.00019998628264629608,
-      "loss": 46.0,
-      "step": 32750
-    },
-    {
-      "epoch": 5.27432666371432,
-      "grad_norm": 0.0022212862968444824,
-      "learning_rate": 0.0001999862818083456,
-      "loss": 46.0,
-      "step": 32751
-    },
-    {
-      "epoch": 5.274487700793108,
-      "grad_norm": 0.015247160568833351,
-      "learning_rate": 0.0001999862809703695,
-      "loss": 46.0,
-      "step": 32752
-    },
-    {
-      "epoch": 5.274648737871895,
-      "grad_norm": 0.004423925653100014,
-      "learning_rate": 0.00019998628013236783,
-      "loss": 46.0,
-      "step": 32753
-    },
-    {
-      "epoch": 5.274809774950683,
-      "grad_norm": 0.0013157791690900922,
-      "learning_rate": 0.00019998627929434056,
-      "loss": 46.0,
-      "step": 32754
-    },
-    {
-      "epoch": 5.27497081202947,
-      "grad_norm": 0.003976835869252682,
-      "learning_rate": 0.0001999862784562877,
-      "loss": 46.0,
-      "step": 32755
-    },
-    {
-      "epoch": 5.2751318491082575,
-      "grad_norm": 0.0010143291437998414,
-      "learning_rate": 0.00019998627761820927,
-      "loss": 46.0,
-      "step": 32756
-    },
-    {
-      "epoch": 5.275292886187045,
-      "grad_norm": 0.00988712627440691,
-      "learning_rate": 0.00019998627678010522,
-      "loss": 46.0,
-      "step": 32757
-    },
-    {
-      "epoch": 5.275453923265832,
-      "grad_norm": 0.0009437710978090763,
-      "learning_rate": 0.00019998627594197557,
-      "loss": 46.0,
-      "step": 32758
-    },
-    {
-      "epoch": 5.275614960344619,
-      "grad_norm": 0.004082816652953625,
-      "learning_rate": 0.00019998627510382035,
-      "loss": 46.0,
-      "step": 32759
-    },
-    {
-      "epoch": 5.275775997423406,
-      "grad_norm": 0.011374269612133503,
-      "learning_rate": 0.00019998627426563953,
-      "loss": 46.0,
-      "step": 32760
-    },
-    {
-      "epoch": 5.275937034502194,
-      "grad_norm": 0.001038078567944467,
-      "learning_rate": 0.00019998627342743315,
-      "loss": 46.0,
-      "step": 32761
-    },
-    {
-      "epoch": 5.276098071580981,
-      "grad_norm": 0.009498587809503078,
-      "learning_rate": 0.00019998627258920116,
-      "loss": 46.0,
-      "step": 32762
-    },
-    {
-      "epoch": 5.276259108659769,
-      "grad_norm": 0.007078665774315596,
-      "learning_rate": 0.00019998627175094359,
-      "loss": 46.0,
-      "step": 32763
-    },
-    {
-      "epoch": 5.276420145738556,
-      "grad_norm": 0.0020795254968106747,
-      "learning_rate": 0.0001999862709126604,
-      "loss": 46.0,
-      "step": 32764
-    },
-    {
-      "epoch": 5.276581182817344,
-      "grad_norm": 0.0035221728030592203,
-      "learning_rate": 0.0001999862700743516,
-      "loss": 46.0,
-      "step": 32765
-    },
-    {
-      "epoch": 5.276742219896131,
-      "grad_norm": 0.0013312288792803884,
-      "learning_rate": 0.00019998626923601728,
-      "loss": 46.0,
-      "step": 32766
-    },
-    {
-      "epoch": 5.276903256974919,
-      "grad_norm": 0.004713039379566908,
-      "learning_rate": 0.0001999862683976573,
-      "loss": 46.0,
-      "step": 32767
-    },
-    {
-      "epoch": 5.277064294053706,
-      "grad_norm": 0.0017045966815203428,
-      "learning_rate": 0.00019998626755927175,
-      "loss": 46.0,
-      "step": 32768
-    },
-    {
-      "epoch": 5.2772253311324935,
-      "grad_norm": 0.004850388970226049,
-      "learning_rate": 0.0001999862667208606,
-      "loss": 46.0,
-      "step": 32769
-    },
-    {
-      "epoch": 5.277386368211281,
-      "grad_norm": 0.011895795352756977,
-      "learning_rate": 0.00019998626588242389,
-      "loss": 46.0,
-      "step": 32770
-    },
-    {
-      "epoch": 5.277547405290068,
-      "grad_norm": 0.0101715587079525,
-      "learning_rate": 0.00019998626504396158,
-      "loss": 46.0,
-      "step": 32771
-    },
-    {
-      "epoch": 5.277708442368856,
-      "grad_norm": 0.006214440334588289,
-      "learning_rate": 0.00019998626420547367,
-      "loss": 46.0,
-      "step": 32772
-    },
-    {
-      "epoch": 5.277869479447643,
-      "grad_norm": 0.0024516754783689976,
-      "learning_rate": 0.00019998626336696016,
-      "loss": 46.0,
-      "step": 32773
-    },
-    {
-      "epoch": 5.27803051652643,
-      "grad_norm": 0.0020385959651321173,
-      "learning_rate": 0.00019998626252842105,
-      "loss": 46.0,
-      "step": 32774
-    },
-    {
-      "epoch": 5.278191553605217,
-      "grad_norm": 0.024326873943209648,
-      "learning_rate": 0.00019998626168985637,
-      "loss": 46.0,
-      "step": 32775
-    },
-    {
-      "epoch": 5.278352590684005,
-      "grad_norm": 0.0008937088423408568,
-      "learning_rate": 0.0001999862608512661,
-      "loss": 46.0,
-      "step": 32776
-    },
-    {
-      "epoch": 5.278513627762792,
-      "grad_norm": 0.0014647595817223191,
-      "learning_rate": 0.00019998626001265025,
-      "loss": 46.0,
-      "step": 32777
-    },
-    {
-      "epoch": 5.27867466484158,
-      "grad_norm": 0.0017795716412365437,
-      "learning_rate": 0.00019998625917400878,
-      "loss": 46.0,
-      "step": 32778
-    },
-    {
-      "epoch": 5.278835701920367,
-      "grad_norm": 0.003916000947356224,
-      "learning_rate": 0.00019998625833534173,
-      "loss": 46.0,
-      "step": 32779
-    },
-    {
-      "epoch": 5.278996738999155,
-      "grad_norm": 0.005517369601875544,
-      "learning_rate": 0.00019998625749664906,
-      "loss": 46.0,
-      "step": 32780
-    },
-    {
-      "epoch": 5.279157776077942,
-      "grad_norm": 0.0024814626667648554,
-      "learning_rate": 0.00019998625665793086,
-      "loss": 46.0,
-      "step": 32781
-    },
-    {
-      "epoch": 5.2793188131567295,
-      "grad_norm": 0.0020449808798730373,
-      "learning_rate": 0.00019998625581918705,
-      "loss": 46.0,
-      "step": 32782
-    },
-    {
-      "epoch": 5.279479850235517,
-      "grad_norm": 0.005443553905934095,
-      "learning_rate": 0.00019998625498041762,
-      "loss": 46.0,
-      "step": 32783
-    },
-    {
-      "epoch": 5.279640887314304,
-      "grad_norm": 0.002270561410114169,
-      "learning_rate": 0.0001999862541416226,
-      "loss": 46.0,
-      "step": 32784
-    },
-    {
-      "epoch": 5.279801924393092,
-      "grad_norm": 0.003994714003056288,
-      "learning_rate": 0.00019998625330280203,
-      "loss": 46.0,
-      "step": 32785
-    },
-    {
-      "epoch": 5.279962961471879,
-      "grad_norm": 0.003503819927573204,
-      "learning_rate": 0.00019998625246395583,
-      "loss": 46.0,
-      "step": 32786
-    },
-    {
-      "epoch": 5.280123998550666,
-      "grad_norm": 0.0017188319470733404,
-      "learning_rate": 0.00019998625162508403,
-      "loss": 46.0,
-      "step": 32787
-    },
-    {
-      "epoch": 5.280285035629453,
-      "grad_norm": 0.0017572222277522087,
-      "learning_rate": 0.00019998625078618666,
-      "loss": 46.0,
-      "step": 32788
-    },
-    {
-      "epoch": 5.280446072708241,
-      "grad_norm": 0.0015798885142430663,
-      "learning_rate": 0.00019998624994726368,
-      "loss": 46.0,
-      "step": 32789
-    },
-    {
-      "epoch": 5.280607109787028,
-      "grad_norm": 0.020196184515953064,
-      "learning_rate": 0.00019998624910831514,
-      "loss": 46.0,
-      "step": 32790
-    },
-    {
-      "epoch": 5.280768146865816,
-      "grad_norm": 0.009554862044751644,
-      "learning_rate": 0.000199986248269341,
-      "loss": 46.0,
-      "step": 32791
-    },
-    {
-      "epoch": 5.280929183944603,
-      "grad_norm": 0.011458268389105797,
-      "learning_rate": 0.00019998624743034128,
-      "loss": 46.0,
-      "step": 32792
-    },
-    {
-      "epoch": 5.2810902210233905,
-      "grad_norm": 0.006587871816009283,
-      "learning_rate": 0.00019998624659131595,
-      "loss": 46.0,
-      "step": 32793
-    },
-    {
-      "epoch": 5.281251258102178,
-      "grad_norm": 0.004437935538589954,
-      "learning_rate": 0.00019998624575226503,
-      "loss": 46.0,
-      "step": 32794
-    },
-    {
-      "epoch": 5.281412295180965,
-      "grad_norm": 0.012687429785728455,
-      "learning_rate": 0.0001999862449131885,
-      "loss": 46.0,
-      "step": 32795
-    },
-    {
-      "epoch": 5.281573332259753,
-      "grad_norm": 0.009609921835362911,
-      "learning_rate": 0.00019998624407408641,
-      "loss": 46.0,
-      "step": 32796
-    },
-    {
-      "epoch": 5.28173436933854,
-      "grad_norm": 0.004458840936422348,
-      "learning_rate": 0.0001999862432349587,
-      "loss": 46.0,
-      "step": 32797
-    },
-    {
-      "epoch": 5.281895406417328,
-      "grad_norm": 0.010326903313398361,
-      "learning_rate": 0.00019998624239580545,
-      "loss": 46.0,
-      "step": 32798
-    },
-    {
-      "epoch": 5.282056443496115,
-      "grad_norm": 0.027878884226083755,
-      "learning_rate": 0.00019998624155662654,
-      "loss": 46.0,
-      "step": 32799
-    },
-    {
-      "epoch": 5.282217480574903,
-      "grad_norm": 0.014449832029640675,
-      "learning_rate": 0.0001999862407174221,
-      "loss": 46.0,
-      "step": 32800
-    },
-    {
-      "epoch": 5.28237851765369,
-      "grad_norm": 0.0024302948731929064,
-      "learning_rate": 0.00019998623987819202,
-      "loss": 46.0,
-      "step": 32801
-    },
-    {
-      "epoch": 5.282539554732477,
-      "grad_norm": 0.005840963684022427,
-      "learning_rate": 0.00019998623903893638,
-      "loss": 46.0,
-      "step": 32802
-    },
-    {
-      "epoch": 5.282700591811264,
-      "grad_norm": 0.015816377475857735,
-      "learning_rate": 0.00019998623819965513,
-      "loss": 46.0,
-      "step": 32803
-    },
-    {
-      "epoch": 5.282861628890052,
-      "grad_norm": 0.0071578435599803925,
-      "learning_rate": 0.0001999862373603483,
-      "loss": 46.0,
-      "step": 32804
-    },
-    {
-      "epoch": 5.283022665968839,
-      "grad_norm": 0.006389778107404709,
-      "learning_rate": 0.00019998623652101586,
-      "loss": 46.0,
-      "step": 32805
-    },
-    {
-      "epoch": 5.2831837030476265,
-      "grad_norm": 0.011697894893586636,
-      "learning_rate": 0.00019998623568165784,
-      "loss": 46.0,
-      "step": 32806
-    },
-    {
-      "epoch": 5.283344740126414,
-      "grad_norm": 0.004841446410864592,
-      "learning_rate": 0.00019998623484227427,
-      "loss": 46.0,
-      "step": 32807
-    },
-    {
-      "epoch": 5.283505777205201,
-      "grad_norm": 0.012552822008728981,
-      "learning_rate": 0.00019998623400286505,
-      "loss": 46.0,
-      "step": 32808
-    },
-    {
-      "epoch": 5.283666814283989,
-      "grad_norm": 0.014014051295816898,
-      "learning_rate": 0.00019998623316343025,
-      "loss": 46.0,
-      "step": 32809
-    },
-    {
-      "epoch": 5.283827851362776,
-      "grad_norm": 0.002752182772383094,
-      "learning_rate": 0.00019998623232396988,
-      "loss": 46.0,
-      "step": 32810
-    },
-    {
-      "epoch": 5.283988888441564,
-      "grad_norm": 0.011237689293920994,
-      "learning_rate": 0.0001999862314844839,
-      "loss": 46.0,
-      "step": 32811
-    },
-    {
-      "epoch": 5.284149925520351,
-      "grad_norm": 0.00717261852696538,
-      "learning_rate": 0.00019998623064497234,
-      "loss": 46.0,
-      "step": 32812
-    },
-    {
-      "epoch": 5.284310962599139,
-      "grad_norm": 0.0019925220403820276,
-      "learning_rate": 0.00019998622980543519,
-      "loss": 46.0,
-      "step": 32813
-    },
-    {
-      "epoch": 5.284471999677926,
-      "grad_norm": 0.0072485958226025105,
-      "learning_rate": 0.00019998622896587245,
-      "loss": 46.0,
-      "step": 32814
-    },
-    {
-      "epoch": 5.284633036756714,
-      "grad_norm": 0.0047632320784032345,
-      "learning_rate": 0.0001999862281262841,
-      "loss": 46.0,
-      "step": 32815
-    },
-    {
-      "epoch": 5.284794073835501,
-      "grad_norm": 0.004829026758670807,
-      "learning_rate": 0.00019998622728667015,
-      "loss": 46.0,
-      "step": 32816
-    },
-    {
-      "epoch": 5.284955110914288,
-      "grad_norm": 0.015597792342305183,
-      "learning_rate": 0.00019998622644703065,
-      "loss": 46.0,
-      "step": 32817
-    },
-    {
-      "epoch": 5.285116147993075,
-      "grad_norm": 0.0007577510550618172,
-      "learning_rate": 0.00019998622560736556,
-      "loss": 46.0,
-      "step": 32818
-    },
-    {
-      "epoch": 5.2852771850718625,
-      "grad_norm": 0.012728163041174412,
-      "learning_rate": 0.00019998622476767483,
-      "loss": 46.0,
-      "step": 32819
-    },
-    {
-      "epoch": 5.28543822215065,
-      "grad_norm": 0.002703931648284197,
-      "learning_rate": 0.00019998622392795854,
-      "loss": 46.0,
-      "step": 32820
-    },
-    {
-      "epoch": 5.285599259229437,
-      "grad_norm": 0.008138729259371758,
-      "learning_rate": 0.00019998622308821666,
-      "loss": 46.0,
-      "step": 32821
-    },
-    {
-      "epoch": 5.285760296308225,
-      "grad_norm": 0.0023652147501707077,
-      "learning_rate": 0.00019998622224844917,
-      "loss": 46.0,
-      "step": 32822
-    },
-    {
-      "epoch": 5.285921333387012,
-      "grad_norm": 0.004464826080948114,
-      "learning_rate": 0.00019998622140865612,
-      "loss": 46.0,
-      "step": 32823
-    },
-    {
-      "epoch": 5.2860823704658,
-      "grad_norm": 0.0026193042285740376,
-      "learning_rate": 0.00019998622056883745,
-      "loss": 46.0,
-      "step": 32824
-    },
-    {
-      "epoch": 5.286243407544587,
-      "grad_norm": 0.0034537853207439184,
-      "learning_rate": 0.00019998621972899317,
-      "loss": 46.0,
-      "step": 32825
-    },
-    {
-      "epoch": 5.286404444623375,
-      "grad_norm": 0.0020162579603493214,
-      "learning_rate": 0.00019998621888912336,
-      "loss": 46.0,
-      "step": 32826
-    },
-    {
-      "epoch": 5.286565481702162,
-      "grad_norm": 0.005515541881322861,
-      "learning_rate": 0.00019998621804922793,
-      "loss": 46.0,
-      "step": 32827
-    },
-    {
-      "epoch": 5.28672651878095,
-      "grad_norm": 0.012862896546721458,
-      "learning_rate": 0.00019998621720930692,
-      "loss": 46.0,
-      "step": 32828
-    },
-    {
-      "epoch": 5.286887555859737,
-      "grad_norm": 0.00212840992026031,
-      "learning_rate": 0.00019998621636936026,
-      "loss": 46.0,
-      "step": 32829
-    },
-    {
-      "epoch": 5.2870485929385245,
-      "grad_norm": 0.002766086021438241,
-      "learning_rate": 0.00019998621552938807,
-      "loss": 46.0,
-      "step": 32830
-    },
-    {
-      "epoch": 5.287209630017312,
-      "grad_norm": 0.009885932318866253,
-      "learning_rate": 0.00019998621468939027,
-      "loss": 46.0,
-      "step": 32831
-    },
-    {
-      "epoch": 5.2873706670960985,
-      "grad_norm": 0.0011209732620045543,
-      "learning_rate": 0.00019998621384936688,
-      "loss": 46.0,
-      "step": 32832
-    },
-    {
-      "epoch": 5.287531704174886,
-      "grad_norm": 0.006951188202947378,
-      "learning_rate": 0.00019998621300931787,
-      "loss": 46.0,
-      "step": 32833
-    },
-    {
-      "epoch": 5.287692741253673,
-      "grad_norm": 0.00442025950178504,
-      "learning_rate": 0.0001999862121692433,
-      "loss": 46.0,
-      "step": 32834
-    },
-    {
-      "epoch": 5.287853778332461,
-      "grad_norm": 0.0016486956737935543,
-      "learning_rate": 0.00019998621132914318,
-      "loss": 46.0,
-      "step": 32835
-    },
-    {
-      "epoch": 5.288014815411248,
-      "grad_norm": 0.0021979613229632378,
-      "learning_rate": 0.0001999862104890174,
-      "loss": 46.0,
-      "step": 32836
-    },
-    {
-      "epoch": 5.288175852490036,
-      "grad_norm": 0.011926851235330105,
-      "learning_rate": 0.00019998620964886606,
-      "loss": 46.0,
-      "step": 32837
-    },
-    {
-      "epoch": 5.288336889568823,
-      "grad_norm": 0.010933454148471355,
-      "learning_rate": 0.0001999862088086891,
-      "loss": 46.0,
-      "step": 32838
-    },
-    {
-      "epoch": 5.288497926647611,
-      "grad_norm": 0.00704097468405962,
-      "learning_rate": 0.00019998620796848657,
-      "loss": 46.0,
-      "step": 32839
-    },
-    {
-      "epoch": 5.288658963726398,
-      "grad_norm": 0.00409117853268981,
-      "learning_rate": 0.00019998620712825845,
-      "loss": 46.0,
-      "step": 32840
-    },
-    {
-      "epoch": 5.2888200008051856,
-      "grad_norm": 0.003234849777072668,
-      "learning_rate": 0.00019998620628800472,
-      "loss": 46.0,
-      "step": 32841
-    },
-    {
-      "epoch": 5.288981037883973,
-      "grad_norm": 0.007692200597375631,
-      "learning_rate": 0.00019998620544772543,
-      "loss": 46.0,
-      "step": 32842
-    },
-    {
-      "epoch": 5.2891420749627605,
-      "grad_norm": 0.006310525815933943,
-      "learning_rate": 0.00019998620460742056,
-      "loss": 46.0,
-      "step": 32843
-    },
-    {
-      "epoch": 5.289303112041548,
-      "grad_norm": 0.010792076587677002,
-      "learning_rate": 0.00019998620376709004,
-      "loss": 46.0,
-      "step": 32844
-    },
-    {
-      "epoch": 5.289464149120335,
-      "grad_norm": 0.0019435528665781021,
-      "learning_rate": 0.000199986202926734,
-      "loss": 46.0,
-      "step": 32845
-    },
-    {
-      "epoch": 5.289625186199123,
-      "grad_norm": 0.0055960011668503284,
-      "learning_rate": 0.0001999862020863523,
-      "loss": 46.0,
-      "step": 32846
-    },
-    {
-      "epoch": 5.289786223277909,
-      "grad_norm": 0.006108923349529505,
-      "learning_rate": 0.00019998620124594504,
-      "loss": 46.0,
-      "step": 32847
-    },
-    {
-      "epoch": 5.289947260356697,
-      "grad_norm": 0.003643910400569439,
-      "learning_rate": 0.0001999862004055122,
-      "loss": 46.0,
-      "step": 32848
-    },
-    {
-      "epoch": 5.290108297435484,
-      "grad_norm": 0.0015204141382128,
-      "learning_rate": 0.00019998619956505375,
-      "loss": 46.0,
-      "step": 32849
-    },
-    {
-      "epoch": 5.290269334514272,
-      "grad_norm": 0.006978069432079792,
-      "learning_rate": 0.0001999861987245697,
-      "loss": 46.0,
-      "step": 32850
-    },
-    {
-      "epoch": 5.290430371593059,
-      "grad_norm": 0.005346689838916063,
-      "learning_rate": 0.00019998619788406008,
-      "loss": 46.0,
-      "step": 32851
-    },
-    {
-      "epoch": 5.290591408671847,
-      "grad_norm": 0.0023616200778633356,
-      "learning_rate": 0.0001999861970435249,
-      "loss": 46.0,
-      "step": 32852
-    },
-    {
-      "epoch": 5.290752445750634,
-      "grad_norm": 0.002466344041749835,
-      "learning_rate": 0.00019998619620296407,
-      "loss": 46.0,
-      "step": 32853
-    },
-    {
-      "epoch": 5.2909134828294215,
-      "grad_norm": 0.005166475661098957,
-      "learning_rate": 0.00019998619536237765,
-      "loss": 46.0,
-      "step": 32854
-    },
-    {
-      "epoch": 5.291074519908209,
-      "grad_norm": 0.006704852916300297,
-      "learning_rate": 0.00019998619452176567,
-      "loss": 46.0,
-      "step": 32855
-    },
-    {
-      "epoch": 5.291235556986996,
-      "grad_norm": 0.0020694448612630367,
-      "learning_rate": 0.00019998619368112808,
-      "loss": 46.0,
-      "step": 32856
-    },
-    {
-      "epoch": 5.291396594065784,
-      "grad_norm": 0.009188308380544186,
-      "learning_rate": 0.0001999861928404649,
-      "loss": 46.0,
-      "step": 32857
-    },
-    {
-      "epoch": 5.291557631144571,
-      "grad_norm": 0.017101731151342392,
-      "learning_rate": 0.00019998619199977616,
-      "loss": 46.0,
-      "step": 32858
-    },
-    {
-      "epoch": 5.291718668223359,
-      "grad_norm": 0.0055967941880226135,
-      "learning_rate": 0.00019998619115906178,
-      "loss": 46.0,
-      "step": 32859
-    },
-    {
-      "epoch": 5.291879705302145,
-      "grad_norm": 0.005138530861586332,
-      "learning_rate": 0.00019998619031832184,
-      "loss": 46.0,
-      "step": 32860
-    },
-    {
-      "epoch": 5.292040742380933,
-      "grad_norm": 0.006875460501760244,
-      "learning_rate": 0.0001999861894775563,
-      "loss": 46.0,
-      "step": 32861
-    },
-    {
-      "epoch": 5.29220177945972,
-      "grad_norm": 0.0011748223332688212,
-      "learning_rate": 0.00019998618863676518,
-      "loss": 46.0,
-      "step": 32862
-    },
-    {
-      "epoch": 5.292362816538508,
-      "grad_norm": 0.007890964858233929,
-      "learning_rate": 0.00019998618779594845,
-      "loss": 46.0,
-      "step": 32863
-    },
-    {
-      "epoch": 5.292523853617295,
-      "grad_norm": 0.0018381194677203894,
-      "learning_rate": 0.00019998618695510613,
-      "loss": 46.0,
-      "step": 32864
-    },
-    {
-      "epoch": 5.292684890696083,
-      "grad_norm": 0.0024217022582888603,
-      "learning_rate": 0.00019998618611423823,
-      "loss": 46.0,
-      "step": 32865
-    },
-    {
-      "epoch": 5.29284592777487,
-      "grad_norm": 0.002413748065009713,
-      "learning_rate": 0.0001999861852733447,
-      "loss": 46.0,
-      "step": 32866
-    },
-    {
-      "epoch": 5.2930069648536575,
-      "grad_norm": 0.0053440299816429615,
-      "learning_rate": 0.00019998618443242564,
-      "loss": 46.0,
-      "step": 32867
-    },
-    {
-      "epoch": 5.293168001932445,
-      "grad_norm": 0.001821198151446879,
-      "learning_rate": 0.00019998618359148095,
-      "loss": 46.0,
-      "step": 32868
-    },
-    {
-      "epoch": 5.293329039011232,
-      "grad_norm": 0.002211799379438162,
-      "learning_rate": 0.0001999861827505107,
-      "loss": 46.0,
-      "step": 32869
-    },
-    {
-      "epoch": 5.29349007609002,
-      "grad_norm": 0.008944440633058548,
-      "learning_rate": 0.00019998618190951483,
-      "loss": 46.0,
-      "step": 32870
-    },
-    {
-      "epoch": 5.293651113168807,
-      "grad_norm": 0.004079732578247786,
-      "learning_rate": 0.00019998618106849338,
-      "loss": 46.0,
-      "step": 32871
-    },
-    {
-      "epoch": 5.293812150247595,
-      "grad_norm": 0.01905720867216587,
-      "learning_rate": 0.00019998618022744634,
-      "loss": 46.0,
-      "step": 32872
-    },
-    {
-      "epoch": 5.293973187326382,
-      "grad_norm": 0.0019842679612338543,
-      "learning_rate": 0.0001999861793863737,
-      "loss": 46.0,
-      "step": 32873
-    },
-    {
-      "epoch": 5.29413422440517,
-      "grad_norm": 0.005579811055213213,
-      "learning_rate": 0.00019998617854527547,
-      "loss": 46.0,
-      "step": 32874
-    },
-    {
-      "epoch": 5.294295261483956,
-      "grad_norm": 0.007021198980510235,
-      "learning_rate": 0.00019998617770415167,
-      "loss": 46.0,
-      "step": 32875
-    },
-    {
-      "epoch": 5.294456298562744,
-      "grad_norm": 0.0051645333878695965,
-      "learning_rate": 0.00019998617686300225,
-      "loss": 46.0,
-      "step": 32876
-    },
-    {
-      "epoch": 5.294617335641531,
-      "grad_norm": 0.011732934974133968,
-      "learning_rate": 0.00019998617602182725,
-      "loss": 46.0,
-      "step": 32877
-    },
-    {
-      "epoch": 5.294778372720319,
-      "grad_norm": 0.007981189526617527,
-      "learning_rate": 0.00019998617518062663,
-      "loss": 46.0,
-      "step": 32878
-    },
-    {
-      "epoch": 5.294939409799106,
-      "grad_norm": 0.0014728050446137786,
-      "learning_rate": 0.00019998617433940048,
-      "loss": 46.0,
-      "step": 32879
-    },
-    {
-      "epoch": 5.2951004468778935,
-      "grad_norm": 0.023625358939170837,
-      "learning_rate": 0.0001999861734981487,
-      "loss": 46.0,
-      "step": 32880
-    },
-    {
-      "epoch": 5.295261483956681,
-      "grad_norm": 0.003997601568698883,
-      "learning_rate": 0.0001999861726568713,
-      "loss": 46.0,
-      "step": 32881
-    },
-    {
-      "epoch": 5.295422521035468,
-      "grad_norm": 0.006560850888490677,
-      "learning_rate": 0.00019998617181556837,
-      "loss": 46.0,
-      "step": 32882
-    },
-    {
-      "epoch": 5.295583558114256,
-      "grad_norm": 0.0015688257990404963,
-      "learning_rate": 0.00019998617097423982,
-      "loss": 46.0,
-      "step": 32883
-    },
-    {
-      "epoch": 5.295744595193043,
-      "grad_norm": 0.010468300431966782,
-      "learning_rate": 0.00019998617013288568,
-      "loss": 46.0,
-      "step": 32884
-    },
-    {
-      "epoch": 5.295905632271831,
-      "grad_norm": 0.003485786961391568,
-      "learning_rate": 0.00019998616929150592,
-      "loss": 46.0,
-      "step": 32885
-    },
-    {
-      "epoch": 5.296066669350618,
-      "grad_norm": 0.005726935341954231,
-      "learning_rate": 0.00019998616845010064,
-      "loss": 46.0,
-      "step": 32886
-    },
-    {
-      "epoch": 5.296227706429406,
-      "grad_norm": 0.0030403868295252323,
-      "learning_rate": 0.0001999861676086697,
-      "loss": 46.0,
-      "step": 32887
-    },
-    {
-      "epoch": 5.296388743508193,
-      "grad_norm": 0.011329368688166142,
-      "learning_rate": 0.0001999861667672132,
-      "loss": 46.0,
-      "step": 32888
-    },
-    {
-      "epoch": 5.296549780586981,
-      "grad_norm": 0.013742944225668907,
-      "learning_rate": 0.00019998616592573111,
-      "loss": 46.0,
-      "step": 32889
-    },
-    {
-      "epoch": 5.296710817665767,
-      "grad_norm": 0.004545212723314762,
-      "learning_rate": 0.0001999861650842234,
-      "loss": 46.0,
-      "step": 32890
-    },
-    {
-      "epoch": 5.296871854744555,
-      "grad_norm": 0.006464786361902952,
-      "learning_rate": 0.00019998616424269012,
-      "loss": 46.0,
-      "step": 32891
-    },
-    {
-      "epoch": 5.297032891823342,
-      "grad_norm": 0.01642397791147232,
-      "learning_rate": 0.00019998616340113125,
-      "loss": 46.0,
-      "step": 32892
-    },
-    {
-      "epoch": 5.2971939289021295,
-      "grad_norm": 0.002626928733661771,
-      "learning_rate": 0.0001999861625595468,
-      "loss": 46.0,
-      "step": 32893
-    },
-    {
-      "epoch": 5.297354965980917,
-      "grad_norm": 0.003786102868616581,
-      "learning_rate": 0.00019998616171793674,
-      "loss": 46.0,
-      "step": 32894
-    },
-    {
-      "epoch": 5.297516003059704,
-      "grad_norm": 0.017067410051822662,
-      "learning_rate": 0.00019998616087630108,
-      "loss": 46.0,
-      "step": 32895
-    },
-    {
-      "epoch": 5.297677040138492,
-      "grad_norm": 0.0018080538138747215,
-      "learning_rate": 0.00019998616003463987,
-      "loss": 46.0,
-      "step": 32896
-    },
-    {
-      "epoch": 5.297838077217279,
-      "grad_norm": 0.017856232821941376,
-      "learning_rate": 0.00019998615919295304,
-      "loss": 46.0,
-      "step": 32897
-    },
-    {
-      "epoch": 5.297999114296067,
-      "grad_norm": 0.008416845463216305,
-      "learning_rate": 0.00019998615835124063,
-      "loss": 46.0,
-      "step": 32898
-    },
-    {
-      "epoch": 5.298160151374854,
-      "grad_norm": 0.0028227707371115685,
-      "learning_rate": 0.0001999861575095026,
-      "loss": 46.0,
-      "step": 32899
-    },
-    {
-      "epoch": 5.298321188453642,
-      "grad_norm": 0.0019845531787723303,
-      "learning_rate": 0.000199986156667739,
-      "loss": 46.0,
-      "step": 32900
-    },
-    {
-      "epoch": 5.298482225532429,
-      "grad_norm": 0.003557221032679081,
-      "learning_rate": 0.0001999861558259498,
-      "loss": 46.0,
-      "step": 32901
-    },
-    {
-      "epoch": 5.2986432626112165,
-      "grad_norm": 0.013394337147474289,
-      "learning_rate": 0.00019998615498413504,
-      "loss": 46.0,
-      "step": 32902
-    },
-    {
-      "epoch": 5.298804299690004,
-      "grad_norm": 0.011069035157561302,
-      "learning_rate": 0.00019998615414229466,
-      "loss": 46.0,
-      "step": 32903
-    },
-    {
-      "epoch": 5.2989653367687914,
-      "grad_norm": 0.006162405479699373,
-      "learning_rate": 0.00019998615330042867,
-      "loss": 46.0,
-      "step": 32904
-    },
-    {
-      "epoch": 5.299126373847578,
-      "grad_norm": 0.0018614601576700807,
-      "learning_rate": 0.00019998615245853715,
-      "loss": 46.0,
-      "step": 32905
-    },
-    {
-      "epoch": 5.2992874109263655,
-      "grad_norm": 0.024059470742940903,
-      "learning_rate": 0.00019998615161661998,
-      "loss": 46.0,
-      "step": 32906
-    },
-    {
-      "epoch": 5.299448448005153,
-      "grad_norm": 0.005213485099375248,
-      "learning_rate": 0.00019998615077467723,
-      "loss": 46.0,
-      "step": 32907
-    },
-    {
-      "epoch": 5.29960948508394,
-      "grad_norm": 0.0058577093295753,
-      "learning_rate": 0.0001999861499327089,
-      "loss": 46.0,
-      "step": 32908
-    },
-    {
-      "epoch": 5.299770522162728,
-      "grad_norm": 0.003455518279224634,
-      "learning_rate": 0.00019998614909071499,
-      "loss": 46.0,
-      "step": 32909
-    },
-    {
-      "epoch": 5.299931559241515,
-      "grad_norm": 0.0068109589628875256,
-      "learning_rate": 0.00019998614824869547,
-      "loss": 46.0,
-      "step": 32910
-    },
-    {
-      "epoch": 5.300092596320303,
-      "grad_norm": 0.012743260711431503,
-      "learning_rate": 0.00019998614740665037,
-      "loss": 46.0,
-      "step": 32911
-    },
-    {
-      "epoch": 5.30025363339909,
-      "grad_norm": 0.012862837873399258,
-      "learning_rate": 0.00019998614656457965,
-      "loss": 46.0,
-      "step": 32912
-    },
-    {
-      "epoch": 5.300414670477878,
-      "grad_norm": 0.005507150199264288,
-      "learning_rate": 0.00019998614572248338,
-      "loss": 46.0,
-      "step": 32913
-    },
-    {
-      "epoch": 5.300575707556665,
-      "grad_norm": 0.00125038786791265,
-      "learning_rate": 0.00019998614488036148,
-      "loss": 46.0,
-      "step": 32914
-    },
-    {
-      "epoch": 5.3007367446354525,
-      "grad_norm": 0.005946934223175049,
-      "learning_rate": 0.00019998614403821403,
-      "loss": 46.0,
-      "step": 32915
-    },
-    {
-      "epoch": 5.30089778171424,
-      "grad_norm": 0.007258351892232895,
-      "learning_rate": 0.00019998614319604094,
-      "loss": 46.0,
-      "step": 32916
-    },
-    {
-      "epoch": 5.301058818793027,
-      "grad_norm": 0.0035813457798212767,
-      "learning_rate": 0.0001999861423538423,
-      "loss": 46.0,
-      "step": 32917
-    },
-    {
-      "epoch": 5.301219855871815,
-      "grad_norm": 0.011239600367844105,
-      "learning_rate": 0.00019998614151161805,
-      "loss": 46.0,
-      "step": 32918
-    },
-    {
-      "epoch": 5.301380892950602,
-      "grad_norm": 0.002644438063725829,
-      "learning_rate": 0.0001999861406693682,
-      "loss": 46.0,
-      "step": 32919
-    },
-    {
-      "epoch": 5.301541930029389,
-      "grad_norm": 0.006767032667994499,
-      "learning_rate": 0.0001999861398270928,
-      "loss": 46.0,
-      "step": 32920
-    },
-    {
-      "epoch": 5.301702967108176,
-      "grad_norm": 0.0018406646559014916,
-      "learning_rate": 0.00019998613898479178,
-      "loss": 46.0,
-      "step": 32921
-    },
-    {
-      "epoch": 5.301864004186964,
-      "grad_norm": 0.002352233277633786,
-      "learning_rate": 0.00019998613814246517,
-      "loss": 46.0,
-      "step": 32922
-    },
-    {
-      "epoch": 5.302025041265751,
-      "grad_norm": 0.005606109742075205,
-      "learning_rate": 0.00019998613730011296,
-      "loss": 46.0,
-      "step": 32923
-    },
-    {
-      "epoch": 5.302186078344539,
-      "grad_norm": 0.0032901305239647627,
-      "learning_rate": 0.00019998613645773517,
-      "loss": 46.0,
-      "step": 32924
-    },
-    {
-      "epoch": 5.302347115423326,
-      "grad_norm": 0.005535468924790621,
-      "learning_rate": 0.00019998613561533177,
-      "loss": 46.0,
-      "step": 32925
-    },
-    {
-      "epoch": 5.302508152502114,
-      "grad_norm": 0.005546476226300001,
-      "learning_rate": 0.0001999861347729028,
-      "loss": 46.0,
-      "step": 32926
-    },
-    {
-      "epoch": 5.302669189580901,
-      "grad_norm": 0.009324170649051666,
-      "learning_rate": 0.00019998613393044825,
-      "loss": 46.0,
-      "step": 32927
-    },
-    {
-      "epoch": 5.3028302266596885,
-      "grad_norm": 0.0022420398890972137,
-      "learning_rate": 0.0001999861330879681,
-      "loss": 46.0,
-      "step": 32928
-    },
-    {
-      "epoch": 5.302991263738476,
-      "grad_norm": 0.013722854666411877,
-      "learning_rate": 0.0001999861322454623,
-      "loss": 46.0,
-      "step": 32929
-    },
-    {
-      "epoch": 5.303152300817263,
-      "grad_norm": 0.00519983097910881,
-      "learning_rate": 0.000199986131402931,
-      "loss": 46.0,
-      "step": 32930
-    },
-    {
-      "epoch": 5.303313337896051,
-      "grad_norm": 0.001633224543184042,
-      "learning_rate": 0.00019998613056037404,
-      "loss": 46.0,
-      "step": 32931
-    },
-    {
-      "epoch": 5.303474374974838,
-      "grad_norm": 0.004387850873172283,
-      "learning_rate": 0.00019998612971779153,
-      "loss": 46.0,
-      "step": 32932
-    },
-    {
-      "epoch": 5.303635412053625,
-      "grad_norm": 0.007314561866223812,
-      "learning_rate": 0.00019998612887518343,
-      "loss": 46.0,
-      "step": 32933
-    },
-    {
-      "epoch": 5.303796449132412,
-      "grad_norm": 0.021742820739746094,
-      "learning_rate": 0.00019998612803254974,
-      "loss": 46.0,
-      "step": 32934
-    },
-    {
-      "epoch": 5.3039574862112,
-      "grad_norm": 0.004345561843365431,
-      "learning_rate": 0.0001999861271898904,
-      "loss": 46.0,
-      "step": 32935
-    },
-    {
-      "epoch": 5.304118523289987,
-      "grad_norm": 0.012240979820489883,
-      "learning_rate": 0.00019998612634720552,
-      "loss": 46.0,
-      "step": 32936
-    },
-    {
-      "epoch": 5.304279560368775,
-      "grad_norm": 0.0026070650201290846,
-      "learning_rate": 0.000199986125504495,
-      "loss": 46.0,
-      "step": 32937
-    },
-    {
-      "epoch": 5.304440597447562,
-      "grad_norm": 0.01934000663459301,
-      "learning_rate": 0.00019998612466175898,
-      "loss": 46.0,
-      "step": 32938
-    },
-    {
-      "epoch": 5.30460163452635,
-      "grad_norm": 0.018276270478963852,
-      "learning_rate": 0.0001999861238189973,
-      "loss": 46.0,
-      "step": 32939
-    },
-    {
-      "epoch": 5.304762671605137,
-      "grad_norm": 0.010701451450586319,
-      "learning_rate": 0.00019998612297621006,
-      "loss": 46.0,
-      "step": 32940
-    },
-    {
-      "epoch": 5.3049237086839245,
-      "grad_norm": 0.002259456319734454,
-      "learning_rate": 0.0001999861221333972,
-      "loss": 46.0,
-      "step": 32941
-    },
-    {
-      "epoch": 5.305084745762712,
-      "grad_norm": 0.0013941318029537797,
-      "learning_rate": 0.00019998612129055877,
-      "loss": 46.0,
-      "step": 32942
-    },
-    {
-      "epoch": 5.305245782841499,
-      "grad_norm": 0.011542130261659622,
-      "learning_rate": 0.00019998612044769474,
-      "loss": 46.0,
-      "step": 32943
-    },
-    {
-      "epoch": 5.305406819920287,
-      "grad_norm": 0.005293803755193949,
-      "learning_rate": 0.00019998611960480512,
-      "loss": 46.0,
-      "step": 32944
-    },
-    {
-      "epoch": 5.305567856999074,
-      "grad_norm": 0.0014259928138926625,
-      "learning_rate": 0.0001999861187618899,
-      "loss": 46.0,
-      "step": 32945
-    },
-    {
-      "epoch": 5.305728894077862,
-      "grad_norm": 0.0016877994639798999,
-      "learning_rate": 0.0001999861179189491,
-      "loss": 46.0,
-      "step": 32946
-    },
-    {
-      "epoch": 5.305889931156649,
-      "grad_norm": 0.0017205652548000216,
-      "learning_rate": 0.0001999861170759827,
-      "loss": 46.0,
-      "step": 32947
-    },
-    {
-      "epoch": 5.306050968235436,
-      "grad_norm": 0.0020924003329128027,
-      "learning_rate": 0.00019998611623299074,
-      "loss": 46.0,
-      "step": 32948
-    },
-    {
-      "epoch": 5.306212005314223,
-      "grad_norm": 0.003963875584304333,
-      "learning_rate": 0.00019998611538997314,
-      "loss": 46.0,
-      "step": 32949
-    },
-    {
-      "epoch": 5.306373042393011,
-      "grad_norm": 0.0038015092723071575,
-      "learning_rate": 0.00019998611454693,
-      "loss": 46.0,
-      "step": 32950
-    },
-    {
-      "epoch": 5.306534079471798,
-      "grad_norm": 0.0036802717950195074,
-      "learning_rate": 0.00019998611370386122,
-      "loss": 46.0,
-      "step": 32951
-    },
-    {
-      "epoch": 5.306695116550586,
-      "grad_norm": 0.006565919611603022,
-      "learning_rate": 0.00019998611286076688,
-      "loss": 46.0,
-      "step": 32952
-    },
-    {
-      "epoch": 5.306856153629373,
-      "grad_norm": 0.003474402241408825,
-      "learning_rate": 0.00019998611201764695,
-      "loss": 46.0,
-      "step": 32953
-    },
-    {
-      "epoch": 5.3070171907081605,
-      "grad_norm": 0.006863272748887539,
-      "learning_rate": 0.0001999861111745014,
-      "loss": 46.0,
-      "step": 32954
-    },
-    {
-      "epoch": 5.307178227786948,
-      "grad_norm": 0.0032033929601311684,
-      "learning_rate": 0.00019998611033133026,
-      "loss": 46.0,
-      "step": 32955
-    },
-    {
-      "epoch": 5.307339264865735,
-      "grad_norm": 0.004401786718517542,
-      "learning_rate": 0.00019998610948813354,
-      "loss": 46.0,
-      "step": 32956
-    },
-    {
-      "epoch": 5.307500301944523,
-      "grad_norm": 0.002599339233711362,
-      "learning_rate": 0.00019998610864491124,
-      "loss": 46.0,
-      "step": 32957
-    },
-    {
-      "epoch": 5.30766133902331,
-      "grad_norm": 0.0025066251400858164,
-      "learning_rate": 0.00019998610780166333,
-      "loss": 46.0,
-      "step": 32958
-    },
-    {
-      "epoch": 5.307822376102098,
-      "grad_norm": 0.007181468885391951,
-      "learning_rate": 0.00019998610695838985,
-      "loss": 46.0,
-      "step": 32959
-    },
-    {
-      "epoch": 5.307983413180885,
-      "grad_norm": 0.0015719872899353504,
-      "learning_rate": 0.00019998610611509076,
-      "loss": 46.0,
-      "step": 32960
-    },
-    {
-      "epoch": 5.308144450259673,
-      "grad_norm": 0.003023274941369891,
-      "learning_rate": 0.0001999861052717661,
-      "loss": 46.0,
-      "step": 32961
-    },
-    {
-      "epoch": 5.30830548733846,
-      "grad_norm": 0.004649336915463209,
-      "learning_rate": 0.00019998610442841584,
-      "loss": 46.0,
-      "step": 32962
-    },
-    {
-      "epoch": 5.308466524417247,
-      "grad_norm": 0.007514502853155136,
-      "learning_rate": 0.00019998610358503997,
-      "loss": 46.0,
-      "step": 32963
-    },
-    {
-      "epoch": 5.308627561496034,
-      "grad_norm": 0.0021884969901293516,
-      "learning_rate": 0.00019998610274163853,
-      "loss": 46.0,
-      "step": 32964
-    },
-    {
-      "epoch": 5.3087885985748215,
-      "grad_norm": 0.0021241442300379276,
-      "learning_rate": 0.00019998610189821153,
-      "loss": 46.0,
-      "step": 32965
-    },
-    {
-      "epoch": 5.308949635653609,
-      "grad_norm": 0.021784642711281776,
-      "learning_rate": 0.0001999861010547589,
-      "loss": 46.0,
-      "step": 32966
-    },
-    {
-      "epoch": 5.3091106727323965,
-      "grad_norm": 0.0021011612843722105,
-      "learning_rate": 0.00019998610021128066,
-      "loss": 46.0,
-      "step": 32967
-    },
-    {
-      "epoch": 5.309271709811184,
-      "grad_norm": 0.005242939572781324,
-      "learning_rate": 0.00019998609936777687,
-      "loss": 46.0,
-      "step": 32968
-    },
-    {
-      "epoch": 5.309432746889971,
-      "grad_norm": 0.013063615188002586,
-      "learning_rate": 0.00019998609852424744,
-      "loss": 46.0,
-      "step": 32969
-    },
-    {
-      "epoch": 5.309593783968759,
-      "grad_norm": 0.005171630997210741,
-      "learning_rate": 0.00019998609768069245,
-      "loss": 46.0,
-      "step": 32970
-    },
-    {
-      "epoch": 5.309754821047546,
-      "grad_norm": 0.001569027896039188,
-      "learning_rate": 0.00019998609683711188,
-      "loss": 46.0,
-      "step": 32971
-    },
-    {
-      "epoch": 5.309915858126334,
-      "grad_norm": 0.010841806419193745,
-      "learning_rate": 0.00019998609599350569,
-      "loss": 46.0,
-      "step": 32972
-    },
-    {
-      "epoch": 5.310076895205121,
-      "grad_norm": 0.0023207219783216715,
-      "learning_rate": 0.0001999860951498739,
-      "loss": 46.0,
-      "step": 32973
-    },
-    {
-      "epoch": 5.310237932283909,
-      "grad_norm": 0.002600132254883647,
-      "learning_rate": 0.00019998609430621657,
-      "loss": 46.0,
-      "step": 32974
-    },
-    {
-      "epoch": 5.310398969362696,
-      "grad_norm": 0.007160381879657507,
-      "learning_rate": 0.00019998609346253362,
-      "loss": 46.0,
-      "step": 32975
-    },
-    {
-      "epoch": 5.3105600064414835,
-      "grad_norm": 0.0028567477129399776,
-      "learning_rate": 0.00019998609261882505,
-      "loss": 46.0,
-      "step": 32976
-    },
-    {
-      "epoch": 5.310721043520271,
-      "grad_norm": 0.005168521776795387,
-      "learning_rate": 0.00019998609177509092,
-      "loss": 46.0,
-      "step": 32977
-    },
-    {
-      "epoch": 5.3108820805990575,
-      "grad_norm": 0.006747628096491098,
-      "learning_rate": 0.0001999860909313312,
-      "loss": 46.0,
-      "step": 32978
-    },
-    {
-      "epoch": 5.311043117677845,
-      "grad_norm": 0.021432820707559586,
-      "learning_rate": 0.0001999860900875459,
-      "loss": 46.0,
-      "step": 32979
-    },
-    {
-      "epoch": 5.311204154756632,
-      "grad_norm": 0.0060675800777971745,
-      "learning_rate": 0.000199986089243735,
-      "loss": 46.0,
-      "step": 32980
-    },
-    {
-      "epoch": 5.31136519183542,
-      "grad_norm": 0.001670889905653894,
-      "learning_rate": 0.00019998608839989846,
-      "loss": 46.0,
-      "step": 32981
-    },
-    {
-      "epoch": 5.311526228914207,
-      "grad_norm": 0.0027938117273151875,
-      "learning_rate": 0.00019998608755603637,
-      "loss": 46.0,
-      "step": 32982
-    },
-    {
-      "epoch": 5.311687265992995,
-      "grad_norm": 0.002398216864094138,
-      "learning_rate": 0.00019998608671214872,
-      "loss": 46.0,
-      "step": 32983
-    },
-    {
-      "epoch": 5.311848303071782,
-      "grad_norm": 0.0028288995381444693,
-      "learning_rate": 0.00019998608586823543,
-      "loss": 46.0,
-      "step": 32984
-    },
-    {
-      "epoch": 5.31200934015057,
-      "grad_norm": 0.008504693396389484,
-      "learning_rate": 0.00019998608502429658,
-      "loss": 46.0,
-      "step": 32985
-    },
-    {
-      "epoch": 5.312170377229357,
-      "grad_norm": 0.0044979010708630085,
-      "learning_rate": 0.00019998608418033214,
-      "loss": 46.0,
-      "step": 32986
-    },
-    {
-      "epoch": 5.312331414308145,
-      "grad_norm": 0.0017711810069158673,
-      "learning_rate": 0.00019998608333634206,
-      "loss": 46.0,
-      "step": 32987
-    },
-    {
-      "epoch": 5.312492451386932,
-      "grad_norm": 0.0012784923892468214,
-      "learning_rate": 0.00019998608249232645,
-      "loss": 46.0,
-      "step": 32988
-    },
-    {
-      "epoch": 5.3126534884657195,
-      "grad_norm": 0.002647585002705455,
-      "learning_rate": 0.0001999860816482852,
-      "loss": 46.0,
-      "step": 32989
-    },
-    {
-      "epoch": 5.312814525544507,
-      "grad_norm": 0.0034546919632703066,
-      "learning_rate": 0.00019998608080421838,
-      "loss": 46.0,
-      "step": 32990
-    },
-    {
-      "epoch": 5.312975562623294,
-      "grad_norm": 0.002908692928031087,
-      "learning_rate": 0.00019998607996012598,
-      "loss": 46.0,
-      "step": 32991
-    },
-    {
-      "epoch": 5.313136599702082,
-      "grad_norm": 0.0047160726971924305,
-      "learning_rate": 0.00019998607911600797,
-      "loss": 46.0,
-      "step": 32992
-    },
-    {
-      "epoch": 5.313297636780868,
-      "grad_norm": 0.003456257516518235,
-      "learning_rate": 0.00019998607827186436,
-      "loss": 46.0,
-      "step": 32993
-    },
-    {
-      "epoch": 5.313458673859656,
-      "grad_norm": 0.0018826433224603534,
-      "learning_rate": 0.0001999860774276952,
-      "loss": 46.0,
-      "step": 32994
-    },
-    {
-      "epoch": 5.313619710938443,
-      "grad_norm": 0.0027917560655623674,
-      "learning_rate": 0.00019998607658350043,
-      "loss": 46.0,
-      "step": 32995
-    },
-    {
-      "epoch": 5.313780748017231,
-      "grad_norm": 0.0027280922513455153,
-      "learning_rate": 0.00019998607573928006,
-      "loss": 46.0,
-      "step": 32996
-    },
-    {
-      "epoch": 5.313941785096018,
-      "grad_norm": 0.004896465688943863,
-      "learning_rate": 0.00019998607489503408,
-      "loss": 46.0,
-      "step": 32997
-    },
-    {
-      "epoch": 5.314102822174806,
-      "grad_norm": 0.003834158182144165,
-      "learning_rate": 0.00019998607405076254,
-      "loss": 46.0,
-      "step": 32998
-    },
-    {
-      "epoch": 5.314263859253593,
-      "grad_norm": 0.0033229454420506954,
-      "learning_rate": 0.00019998607320646542,
-      "loss": 46.0,
-      "step": 32999
-    },
-    {
-      "epoch": 5.314424896332381,
-      "grad_norm": 0.002223525894805789,
-      "learning_rate": 0.00019998607236214268,
-      "loss": 46.0,
-      "step": 33000
-    },
-    {
-      "epoch": 5.314585933411168,
-      "grad_norm": 0.0014318267349153757,
-      "learning_rate": 0.00019998607151779435,
-      "loss": 46.0,
-      "step": 33001
-    },
-    {
-      "epoch": 5.3147469704899555,
-      "grad_norm": 0.002659878693521023,
-      "learning_rate": 0.00019998607067342044,
-      "loss": 46.0,
-      "step": 33002
-    },
-    {
-      "epoch": 5.314908007568743,
-      "grad_norm": 0.008306640200316906,
-      "learning_rate": 0.00019998606982902094,
-      "loss": 46.0,
-      "step": 33003
-    },
-    {
-      "epoch": 5.31506904464753,
-      "grad_norm": 0.0023147037718445063,
-      "learning_rate": 0.00019998606898459582,
-      "loss": 46.0,
-      "step": 33004
-    },
-    {
-      "epoch": 5.315230081726318,
-      "grad_norm": 0.006360264029353857,
-      "learning_rate": 0.00019998606814014514,
-      "loss": 46.0,
-      "step": 33005
-    },
-    {
-      "epoch": 5.315391118805105,
-      "grad_norm": 0.00375958695076406,
-      "learning_rate": 0.00019998606729566885,
-      "loss": 46.0,
-      "step": 33006
-    },
-    {
-      "epoch": 5.315552155883893,
-      "grad_norm": 0.008646082133054733,
-      "learning_rate": 0.00019998606645116698,
-      "loss": 46.0,
-      "step": 33007
-    },
-    {
-      "epoch": 5.315713192962679,
-      "grad_norm": 0.007140029687434435,
-      "learning_rate": 0.0001999860656066395,
-      "loss": 46.0,
-      "step": 33008
-    },
-    {
-      "epoch": 5.315874230041467,
-      "grad_norm": 0.0014294757274910808,
-      "learning_rate": 0.00019998606476208646,
-      "loss": 46.0,
-      "step": 33009
-    },
-    {
-      "epoch": 5.316035267120254,
-      "grad_norm": 0.009339352138340473,
-      "learning_rate": 0.00019998606391750782,
-      "loss": 46.0,
-      "step": 33010
-    },
-    {
-      "epoch": 5.316196304199042,
-      "grad_norm": 0.005555547308176756,
-      "learning_rate": 0.00019998606307290357,
-      "loss": 46.0,
-      "step": 33011
-    },
-    {
-      "epoch": 5.316357341277829,
-      "grad_norm": 0.0038346245419234037,
-      "learning_rate": 0.00019998606222827373,
-      "loss": 46.0,
-      "step": 33012
-    },
-    {
-      "epoch": 5.316518378356617,
-      "grad_norm": 0.00588664785027504,
-      "learning_rate": 0.00019998606138361833,
-      "loss": 46.0,
-      "step": 33013
-    },
-    {
-      "epoch": 5.316679415435404,
-      "grad_norm": 0.001739988918416202,
-      "learning_rate": 0.0001999860605389373,
-      "loss": 46.0,
-      "step": 33014
-    },
-    {
-      "epoch": 5.3168404525141915,
-      "grad_norm": 0.001739312196150422,
-      "learning_rate": 0.0001999860596942307,
-      "loss": 46.0,
-      "step": 33015
-    },
-    {
-      "epoch": 5.317001489592979,
-      "grad_norm": 0.008644703775644302,
-      "learning_rate": 0.00019998605884949852,
-      "loss": 46.0,
-      "step": 33016
-    },
-    {
-      "epoch": 5.317162526671766,
-      "grad_norm": 0.006621269043534994,
-      "learning_rate": 0.00019998605800474075,
-      "loss": 46.0,
-      "step": 33017
-    },
-    {
-      "epoch": 5.317323563750554,
-      "grad_norm": 0.0023716334253549576,
-      "learning_rate": 0.00019998605715995736,
-      "loss": 46.0,
-      "step": 33018
-    },
-    {
-      "epoch": 5.317484600829341,
-      "grad_norm": 0.007661569397896528,
-      "learning_rate": 0.00019998605631514838,
-      "loss": 46.0,
-      "step": 33019
-    },
-    {
-      "epoch": 5.317645637908129,
-      "grad_norm": 0.00661576958373189,
-      "learning_rate": 0.00019998605547031381,
-      "loss": 46.0,
-      "step": 33020
-    },
-    {
-      "epoch": 5.317806674986915,
-      "grad_norm": 0.007498806342482567,
-      "learning_rate": 0.0001999860546254537,
-      "loss": 46.0,
-      "step": 33021
-    },
-    {
-      "epoch": 5.317967712065703,
-      "grad_norm": 0.013822809793055058,
-      "learning_rate": 0.00019998605378056795,
-      "loss": 46.0,
-      "step": 33022
-    },
-    {
-      "epoch": 5.31812874914449,
-      "grad_norm": 0.00567164970561862,
-      "learning_rate": 0.00019998605293565662,
-      "loss": 46.0,
-      "step": 33023
-    },
-    {
-      "epoch": 5.318289786223278,
-      "grad_norm": 0.0007239525439217687,
-      "learning_rate": 0.00019998605209071968,
-      "loss": 46.0,
-      "step": 33024
-    },
-    {
-      "epoch": 5.318450823302065,
-      "grad_norm": 0.0006108058732934296,
-      "learning_rate": 0.00019998605124575716,
-      "loss": 46.0,
-      "step": 33025
-    },
-    {
-      "epoch": 5.3186118603808525,
-      "grad_norm": 0.007186190225183964,
-      "learning_rate": 0.00019998605040076904,
-      "loss": 46.0,
-      "step": 33026
-    },
-    {
-      "epoch": 5.31877289745964,
-      "grad_norm": 0.003068361897021532,
-      "learning_rate": 0.00019998604955575537,
-      "loss": 46.0,
-      "step": 33027
-    },
-    {
-      "epoch": 5.318933934538427,
-      "grad_norm": 0.0020151156932115555,
-      "learning_rate": 0.00019998604871071608,
-      "loss": 46.0,
-      "step": 33028
-    },
-    {
-      "epoch": 5.319094971617215,
-      "grad_norm": 0.0015516963321715593,
-      "learning_rate": 0.00019998604786565117,
-      "loss": 46.0,
-      "step": 33029
-    },
-    {
-      "epoch": 5.319256008696002,
-      "grad_norm": 0.004806454759091139,
-      "learning_rate": 0.0001999860470205607,
-      "loss": 46.0,
-      "step": 33030
-    },
-    {
-      "epoch": 5.31941704577479,
-      "grad_norm": 0.012646634131669998,
-      "learning_rate": 0.00019998604617544466,
-      "loss": 46.0,
-      "step": 33031
-    },
-    {
-      "epoch": 5.319578082853577,
-      "grad_norm": 0.004600579384714365,
-      "learning_rate": 0.000199986045330303,
-      "loss": 46.0,
-      "step": 33032
-    },
-    {
-      "epoch": 5.319739119932365,
-      "grad_norm": 0.0011752629652619362,
-      "learning_rate": 0.00019998604448513577,
-      "loss": 46.0,
-      "step": 33033
-    },
-    {
-      "epoch": 5.319900157011152,
-      "grad_norm": 0.013503649272024632,
-      "learning_rate": 0.0001999860436399429,
-      "loss": 46.0,
-      "step": 33034
-    },
-    {
-      "epoch": 5.32006119408994,
-      "grad_norm": 0.0034524749498814344,
-      "learning_rate": 0.00019998604279472448,
-      "loss": 46.0,
-      "step": 33035
-    },
-    {
-      "epoch": 5.320222231168726,
-      "grad_norm": 0.006999840494245291,
-      "learning_rate": 0.00019998604194948046,
-      "loss": 46.0,
-      "step": 33036
-    },
-    {
-      "epoch": 5.320383268247514,
-      "grad_norm": 0.002707788487896323,
-      "learning_rate": 0.00019998604110421083,
-      "loss": 46.0,
-      "step": 33037
-    },
-    {
-      "epoch": 5.320544305326301,
-      "grad_norm": 0.019953666254878044,
-      "learning_rate": 0.00019998604025891565,
-      "loss": 46.0,
-      "step": 33038
-    },
-    {
-      "epoch": 5.3207053424050885,
-      "grad_norm": 0.001057117129676044,
-      "learning_rate": 0.00019998603941359487,
-      "loss": 46.0,
-      "step": 33039
-    },
-    {
-      "epoch": 5.320866379483876,
-      "grad_norm": 0.0037325371522456408,
-      "learning_rate": 0.00019998603856824848,
-      "loss": 46.0,
-      "step": 33040
-    },
-    {
-      "epoch": 5.321027416562663,
-      "grad_norm": 0.004492900334298611,
-      "learning_rate": 0.0001999860377228765,
-      "loss": 46.0,
-      "step": 33041
-    },
-    {
-      "epoch": 5.321188453641451,
-      "grad_norm": 0.0019144448451697826,
-      "learning_rate": 0.00019998603687747891,
-      "loss": 46.0,
-      "step": 33042
-    },
-    {
-      "epoch": 5.321349490720238,
-      "grad_norm": 0.0017227960051968694,
-      "learning_rate": 0.00019998603603205576,
-      "loss": 46.0,
-      "step": 33043
-    },
-    {
-      "epoch": 5.321510527799026,
-      "grad_norm": 0.0015465219039469957,
-      "learning_rate": 0.00019998603518660702,
-      "loss": 46.0,
-      "step": 33044
-    },
-    {
-      "epoch": 5.321671564877813,
-      "grad_norm": 0.003507799468934536,
-      "learning_rate": 0.00019998603434113267,
-      "loss": 46.0,
-      "step": 33045
-    },
-    {
-      "epoch": 5.321832601956601,
-      "grad_norm": 0.0031565867830067873,
-      "learning_rate": 0.00019998603349563273,
-      "loss": 46.0,
-      "step": 33046
-    },
-    {
-      "epoch": 5.321993639035388,
-      "grad_norm": 0.014228135347366333,
-      "learning_rate": 0.0001999860326501072,
-      "loss": 46.0,
-      "step": 33047
-    },
-    {
-      "epoch": 5.322154676114176,
-      "grad_norm": 0.003176738740876317,
-      "learning_rate": 0.00019998603180455612,
-      "loss": 46.0,
-      "step": 33048
-    },
-    {
-      "epoch": 5.322315713192963,
-      "grad_norm": 0.007489599287509918,
-      "learning_rate": 0.0001999860309589794,
-      "loss": 46.0,
-      "step": 33049
-    },
-    {
-      "epoch": 5.3224767502717505,
-      "grad_norm": 0.006652691401541233,
-      "learning_rate": 0.00019998603011337708,
-      "loss": 46.0,
-      "step": 33050
-    },
-    {
-      "epoch": 5.322637787350537,
-      "grad_norm": 0.003964683040976524,
-      "learning_rate": 0.0001999860292677492,
-      "loss": 46.0,
-      "step": 33051
-    },
-    {
-      "epoch": 5.3227988244293245,
-      "grad_norm": 0.011925505474209785,
-      "learning_rate": 0.00019998602842209574,
-      "loss": 46.0,
-      "step": 33052
-    },
-    {
-      "epoch": 5.322959861508112,
-      "grad_norm": 0.0011837847996503115,
-      "learning_rate": 0.00019998602757641666,
-      "loss": 46.0,
-      "step": 33053
-    },
-    {
-      "epoch": 5.323120898586899,
-      "grad_norm": 0.005415219347923994,
-      "learning_rate": 0.000199986026730712,
-      "loss": 46.0,
-      "step": 33054
-    },
-    {
-      "epoch": 5.323281935665687,
-      "grad_norm": 0.00490936916321516,
-      "learning_rate": 0.00019998602588498172,
-      "loss": 46.0,
-      "step": 33055
-    },
-    {
-      "epoch": 5.323442972744474,
-      "grad_norm": 0.0021279919892549515,
-      "learning_rate": 0.0001999860250392259,
-      "loss": 46.0,
-      "step": 33056
-    },
-    {
-      "epoch": 5.323604009823262,
-      "grad_norm": 0.0008878706721588969,
-      "learning_rate": 0.00019998602419344445,
-      "loss": 46.0,
-      "step": 33057
-    },
-    {
-      "epoch": 5.323765046902049,
-      "grad_norm": 0.005314687266945839,
-      "learning_rate": 0.00019998602334763744,
-      "loss": 46.0,
-      "step": 33058
-    },
-    {
-      "epoch": 5.323926083980837,
-      "grad_norm": 0.0029676822014153004,
-      "learning_rate": 0.00019998602250180481,
-      "loss": 46.0,
-      "step": 33059
-    },
-    {
-      "epoch": 5.324087121059624,
-      "grad_norm": 0.003534886287525296,
-      "learning_rate": 0.0001999860216559466,
-      "loss": 46.0,
-      "step": 33060
-    },
-    {
-      "epoch": 5.324248158138412,
-      "grad_norm": 0.0021425921004265547,
-      "learning_rate": 0.0001999860208100628,
-      "loss": 46.0,
-      "step": 33061
-    },
-    {
-      "epoch": 5.324409195217199,
-      "grad_norm": 0.006580045446753502,
-      "learning_rate": 0.0001999860199641534,
-      "loss": 46.0,
-      "step": 33062
-    },
-    {
-      "epoch": 5.3245702322959865,
-      "grad_norm": 0.005584504920989275,
-      "learning_rate": 0.0001999860191182184,
-      "loss": 46.0,
-      "step": 33063
-    },
-    {
-      "epoch": 5.324731269374774,
-      "grad_norm": 0.0016489874105900526,
-      "learning_rate": 0.00019998601827225784,
-      "loss": 46.0,
-      "step": 33064
-    },
-    {
-      "epoch": 5.324892306453561,
-      "grad_norm": 0.007520326413214207,
-      "learning_rate": 0.00019998601742627167,
-      "loss": 46.0,
-      "step": 33065
-    },
-    {
-      "epoch": 5.325053343532348,
-      "grad_norm": 0.001698884880170226,
-      "learning_rate": 0.0001999860165802599,
-      "loss": 46.0,
-      "step": 33066
-    },
-    {
-      "epoch": 5.325214380611135,
-      "grad_norm": 0.007266270462423563,
-      "learning_rate": 0.00019998601573422255,
-      "loss": 46.0,
-      "step": 33067
-    },
-    {
-      "epoch": 5.325375417689923,
-      "grad_norm": 0.001486351597122848,
-      "learning_rate": 0.0001999860148881596,
-      "loss": 46.0,
-      "step": 33068
-    },
-    {
-      "epoch": 5.32553645476871,
-      "grad_norm": 0.010560727678239346,
-      "learning_rate": 0.00019998601404207106,
-      "loss": 46.0,
-      "step": 33069
-    },
-    {
-      "epoch": 5.325697491847498,
-      "grad_norm": 0.0055545661598443985,
-      "learning_rate": 0.00019998601319595695,
-      "loss": 46.0,
-      "step": 33070
-    },
-    {
-      "epoch": 5.325858528926285,
-      "grad_norm": 0.010793630965054035,
-      "learning_rate": 0.00019998601234981725,
-      "loss": 46.0,
-      "step": 33071
-    },
-    {
-      "epoch": 5.326019566005073,
-      "grad_norm": 0.003891882486641407,
-      "learning_rate": 0.00019998601150365196,
-      "loss": 46.0,
-      "step": 33072
-    },
-    {
-      "epoch": 5.32618060308386,
-      "grad_norm": 0.002866546157747507,
-      "learning_rate": 0.00019998601065746106,
-      "loss": 46.0,
-      "step": 33073
-    },
-    {
-      "epoch": 5.3263416401626476,
-      "grad_norm": 0.002724231919273734,
-      "learning_rate": 0.00019998600981124457,
-      "loss": 46.0,
-      "step": 33074
-    },
-    {
-      "epoch": 5.326502677241435,
-      "grad_norm": 0.015910878777503967,
-      "learning_rate": 0.0001999860089650025,
-      "loss": 46.0,
-      "step": 33075
-    },
-    {
-      "epoch": 5.3266637143202225,
-      "grad_norm": 0.00824980903416872,
-      "learning_rate": 0.0001999860081187348,
-      "loss": 46.0,
-      "step": 33076
-    },
-    {
-      "epoch": 5.32682475139901,
-      "grad_norm": 0.0013140970841050148,
-      "learning_rate": 0.00019998600727244155,
-      "loss": 46.0,
-      "step": 33077
-    },
-    {
-      "epoch": 5.326985788477797,
-      "grad_norm": 0.0077692368067801,
-      "learning_rate": 0.0001999860064261227,
-      "loss": 46.0,
-      "step": 33078
-    },
-    {
-      "epoch": 5.327146825556585,
-      "grad_norm": 0.008483261801302433,
-      "learning_rate": 0.00019998600557977824,
-      "loss": 46.0,
-      "step": 33079
-    },
-    {
-      "epoch": 5.327307862635372,
-      "grad_norm": 0.003456692909821868,
-      "learning_rate": 0.0001999860047334082,
-      "loss": 46.0,
-      "step": 33080
-    },
-    {
-      "epoch": 5.327468899714159,
-      "grad_norm": 0.004828238859772682,
-      "learning_rate": 0.0001999860038870126,
-      "loss": 46.0,
-      "step": 33081
-    },
-    {
-      "epoch": 5.327629936792946,
-      "grad_norm": 0.003765688044950366,
-      "learning_rate": 0.00019998600304059133,
-      "loss": 46.0,
-      "step": 33082
-    },
-    {
-      "epoch": 5.327790973871734,
-      "grad_norm": 0.012493998743593693,
-      "learning_rate": 0.00019998600219414453,
-      "loss": 46.0,
-      "step": 33083
-    },
-    {
-      "epoch": 5.327952010950521,
-      "grad_norm": 0.0010309262434020638,
-      "learning_rate": 0.00019998600134767214,
-      "loss": 46.0,
-      "step": 33084
-    },
-    {
-      "epoch": 5.328113048029309,
-      "grad_norm": 0.022467877715826035,
-      "learning_rate": 0.00019998600050117414,
-      "loss": 46.0,
-      "step": 33085
-    },
-    {
-      "epoch": 5.328274085108096,
-      "grad_norm": 0.0011528560426086187,
-      "learning_rate": 0.00019998599965465058,
-      "loss": 46.0,
-      "step": 33086
-    },
-    {
-      "epoch": 5.3284351221868835,
-      "grad_norm": 0.005552551243454218,
-      "learning_rate": 0.00019998599880810138,
-      "loss": 46.0,
-      "step": 33087
-    },
-    {
-      "epoch": 5.328596159265671,
-      "grad_norm": 0.006712593603879213,
-      "learning_rate": 0.0001999859979615266,
-      "loss": 46.0,
-      "step": 33088
-    },
-    {
-      "epoch": 5.328757196344458,
-      "grad_norm": 0.003801204962655902,
-      "learning_rate": 0.00019998599711492626,
-      "loss": 46.0,
-      "step": 33089
-    },
-    {
-      "epoch": 5.328918233423246,
-      "grad_norm": 0.016428135335445404,
-      "learning_rate": 0.0001999859962683003,
-      "loss": 46.0,
-      "step": 33090
-    },
-    {
-      "epoch": 5.329079270502033,
-      "grad_norm": 0.006544114556163549,
-      "learning_rate": 0.00019998599542164874,
-      "loss": 46.0,
-      "step": 33091
-    },
-    {
-      "epoch": 5.329240307580821,
-      "grad_norm": 0.004577435087412596,
-      "learning_rate": 0.00019998599457497163,
-      "loss": 46.0,
-      "step": 33092
-    },
-    {
-      "epoch": 5.329401344659608,
-      "grad_norm": 0.004981053061783314,
-      "learning_rate": 0.0001999859937282689,
-      "loss": 46.0,
-      "step": 33093
-    },
-    {
-      "epoch": 5.329562381738395,
-      "grad_norm": 0.0037103891372680664,
-      "learning_rate": 0.0001999859928815406,
-      "loss": 46.0,
-      "step": 33094
-    },
-    {
-      "epoch": 5.329723418817182,
-      "grad_norm": 0.014178406447172165,
-      "learning_rate": 0.0001999859920347867,
-      "loss": 46.0,
-      "step": 33095
-    },
-    {
-      "epoch": 5.32988445589597,
-      "grad_norm": 0.0023808954283595085,
-      "learning_rate": 0.00019998599118800718,
-      "loss": 46.0,
-      "step": 33096
-    },
-    {
-      "epoch": 5.330045492974757,
-      "grad_norm": 0.0042678555473685265,
-      "learning_rate": 0.0001999859903412021,
-      "loss": 46.0,
-      "step": 33097
-    },
-    {
-      "epoch": 5.330206530053545,
-      "grad_norm": 0.009074622765183449,
-      "learning_rate": 0.0001999859894943714,
-      "loss": 46.0,
-      "step": 33098
-    },
-    {
-      "epoch": 5.330367567132332,
-      "grad_norm": 0.0025333582889288664,
-      "learning_rate": 0.00019998598864751514,
-      "loss": 46.0,
-      "step": 33099
-    },
-    {
-      "epoch": 5.3305286042111195,
-      "grad_norm": 0.008307427167892456,
-      "learning_rate": 0.0001999859878006333,
-      "loss": 46.0,
-      "step": 33100
-    },
-    {
-      "epoch": 5.330689641289907,
-      "grad_norm": 0.015622492879629135,
-      "learning_rate": 0.00019998598695372582,
-      "loss": 46.0,
-      "step": 33101
-    },
-    {
-      "epoch": 5.330850678368694,
-      "grad_norm": 0.0032494382467120886,
-      "learning_rate": 0.0001999859861067928,
-      "loss": 46.0,
-      "step": 33102
-    },
-    {
-      "epoch": 5.331011715447482,
-      "grad_norm": 0.00774772185832262,
-      "learning_rate": 0.00019998598525983416,
-      "loss": 46.0,
-      "step": 33103
-    },
-    {
-      "epoch": 5.331172752526269,
-      "grad_norm": 0.009939687326550484,
-      "learning_rate": 0.00019998598441284992,
-      "loss": 46.0,
-      "step": 33104
-    },
-    {
-      "epoch": 5.331333789605057,
-      "grad_norm": 0.0057042501866817474,
-      "learning_rate": 0.00019998598356584012,
-      "loss": 46.0,
-      "step": 33105
-    },
-    {
-      "epoch": 5.331494826683844,
-      "grad_norm": 0.010075083002448082,
-      "learning_rate": 0.00019998598271880468,
-      "loss": 46.0,
-      "step": 33106
-    },
-    {
-      "epoch": 5.331655863762632,
-      "grad_norm": 0.0013022555503994226,
-      "learning_rate": 0.00019998598187174368,
-      "loss": 46.0,
-      "step": 33107
-    },
-    {
-      "epoch": 5.331816900841419,
-      "grad_norm": 0.003510661656036973,
-      "learning_rate": 0.00019998598102465712,
-      "loss": 46.0,
-      "step": 33108
-    },
-    {
-      "epoch": 5.331977937920206,
-      "grad_norm": 0.002145532751455903,
-      "learning_rate": 0.00019998598017754491,
-      "loss": 46.0,
-      "step": 33109
-    },
-    {
-      "epoch": 5.332138974998993,
-      "grad_norm": 0.0016985288821160793,
-      "learning_rate": 0.00019998597933040715,
-      "loss": 46.0,
-      "step": 33110
-    },
-    {
-      "epoch": 5.332300012077781,
-      "grad_norm": 0.003935207147151232,
-      "learning_rate": 0.00019998597848324377,
-      "loss": 46.0,
-      "step": 33111
-    },
-    {
-      "epoch": 5.332461049156568,
-      "grad_norm": 0.005572963505983353,
-      "learning_rate": 0.0001999859776360548,
-      "loss": 46.0,
-      "step": 33112
-    },
-    {
-      "epoch": 5.3326220862353555,
-      "grad_norm": 0.006349101662635803,
-      "learning_rate": 0.00019998597678884026,
-      "loss": 46.0,
-      "step": 33113
-    },
-    {
-      "epoch": 5.332783123314143,
-      "grad_norm": 0.006011417135596275,
-      "learning_rate": 0.00019998597594160012,
-      "loss": 46.0,
-      "step": 33114
-    },
-    {
-      "epoch": 5.33294416039293,
-      "grad_norm": 0.004160599783062935,
-      "learning_rate": 0.0001999859750943344,
-      "loss": 46.0,
-      "step": 33115
-    },
-    {
-      "epoch": 5.333105197471718,
-      "grad_norm": 0.007534564007073641,
-      "learning_rate": 0.00019998597424704308,
-      "loss": 46.0,
-      "step": 33116
-    },
-    {
-      "epoch": 5.333266234550505,
-      "grad_norm": 0.010425425134599209,
-      "learning_rate": 0.00019998597339972616,
-      "loss": 46.0,
-      "step": 33117
-    },
-    {
-      "epoch": 5.333427271629293,
-      "grad_norm": 0.01217585802078247,
-      "learning_rate": 0.00019998597255238364,
-      "loss": 46.0,
-      "step": 33118
-    },
-    {
-      "epoch": 5.33358830870808,
-      "grad_norm": 0.006327691487967968,
-      "learning_rate": 0.00019998597170501557,
-      "loss": 46.0,
-      "step": 33119
-    },
-    {
-      "epoch": 5.333749345786868,
-      "grad_norm": 0.012276839464902878,
-      "learning_rate": 0.00019998597085762185,
-      "loss": 46.0,
-      "step": 33120
-    },
-    {
-      "epoch": 5.333910382865655,
-      "grad_norm": 0.0010991720482707024,
-      "learning_rate": 0.00019998597001020258,
-      "loss": 46.0,
-      "step": 33121
-    },
-    {
-      "epoch": 5.334071419944443,
-      "grad_norm": 0.0010855909204110503,
-      "learning_rate": 0.00019998596916275771,
-      "loss": 46.0,
-      "step": 33122
-    },
-    {
-      "epoch": 5.33423245702323,
-      "grad_norm": 0.0035561115946620703,
-      "learning_rate": 0.00019998596831528724,
-      "loss": 46.0,
-      "step": 33123
-    },
-    {
-      "epoch": 5.334393494102017,
-      "grad_norm": 0.0018690096912905574,
-      "learning_rate": 0.0001999859674677912,
-      "loss": 46.0,
-      "step": 33124
-    },
-    {
-      "epoch": 5.334554531180804,
-      "grad_norm": 0.001971221761777997,
-      "learning_rate": 0.00019998596662026955,
-      "loss": 46.0,
-      "step": 33125
-    },
-    {
-      "epoch": 5.3347155682595915,
-      "grad_norm": 0.0030504302121698856,
-      "learning_rate": 0.00019998596577272234,
-      "loss": 46.0,
-      "step": 33126
-    },
-    {
-      "epoch": 5.334876605338379,
-      "grad_norm": 0.002618497470393777,
-      "learning_rate": 0.00019998596492514948,
-      "loss": 46.0,
-      "step": 33127
-    },
-    {
-      "epoch": 5.335037642417166,
-      "grad_norm": 0.0011711036786437035,
-      "learning_rate": 0.00019998596407755107,
-      "loss": 46.0,
-      "step": 33128
-    },
-    {
-      "epoch": 5.335198679495954,
-      "grad_norm": 0.02571200393140316,
-      "learning_rate": 0.00019998596322992707,
-      "loss": 46.0,
-      "step": 33129
-    },
-    {
-      "epoch": 5.335359716574741,
-      "grad_norm": 0.00836899597197771,
-      "learning_rate": 0.00019998596238227746,
-      "loss": 46.0,
-      "step": 33130
-    },
-    {
-      "epoch": 5.335520753653529,
-      "grad_norm": 0.0061898063868284225,
-      "learning_rate": 0.00019998596153460228,
-      "loss": 46.0,
-      "step": 33131
-    },
-    {
-      "epoch": 5.335681790732316,
-      "grad_norm": 0.006214721594005823,
-      "learning_rate": 0.0001999859606869015,
-      "loss": 46.0,
-      "step": 33132
-    },
-    {
-      "epoch": 5.335842827811104,
-      "grad_norm": 0.003894099500030279,
-      "learning_rate": 0.00019998595983917512,
-      "loss": 46.0,
-      "step": 33133
-    },
-    {
-      "epoch": 5.336003864889891,
-      "grad_norm": 0.0030999884475022554,
-      "learning_rate": 0.00019998595899142315,
-      "loss": 46.0,
-      "step": 33134
-    },
-    {
-      "epoch": 5.3361649019686785,
-      "grad_norm": 0.008983098901808262,
-      "learning_rate": 0.00019998595814364558,
-      "loss": 46.0,
-      "step": 33135
-    },
-    {
-      "epoch": 5.336325939047466,
-      "grad_norm": 0.015418984927237034,
-      "learning_rate": 0.00019998595729584244,
-      "loss": 46.0,
-      "step": 33136
-    },
-    {
-      "epoch": 5.3364869761262534,
-      "grad_norm": 0.0032537744846194983,
-      "learning_rate": 0.0001999859564480137,
-      "loss": 46.0,
-      "step": 33137
-    },
-    {
-      "epoch": 5.336648013205041,
-      "grad_norm": 0.001825700979679823,
-      "learning_rate": 0.00019998595560015935,
-      "loss": 46.0,
-      "step": 33138
-    },
-    {
-      "epoch": 5.3368090502838275,
-      "grad_norm": 0.0013571919407695532,
-      "learning_rate": 0.00019998595475227945,
-      "loss": 46.0,
-      "step": 33139
-    },
-    {
-      "epoch": 5.336970087362615,
-      "grad_norm": 0.00876959040760994,
-      "learning_rate": 0.00019998595390437394,
-      "loss": 46.0,
-      "step": 33140
-    },
-    {
-      "epoch": 5.337131124441402,
-      "grad_norm": 0.002108845626935363,
-      "learning_rate": 0.00019998595305644284,
-      "loss": 46.0,
-      "step": 33141
-    },
-    {
-      "epoch": 5.33729216152019,
-      "grad_norm": 0.01445204671472311,
-      "learning_rate": 0.00019998595220848615,
-      "loss": 46.0,
-      "step": 33142
-    },
-    {
-      "epoch": 5.337453198598977,
-      "grad_norm": 0.0016119669890031219,
-      "learning_rate": 0.00019998595136050385,
-      "loss": 46.0,
-      "step": 33143
-    },
-    {
-      "epoch": 5.337614235677765,
-      "grad_norm": 0.00953258853405714,
-      "learning_rate": 0.00019998595051249596,
-      "loss": 46.0,
-      "step": 33144
-    },
-    {
-      "epoch": 5.337775272756552,
-      "grad_norm": 0.001916867564432323,
-      "learning_rate": 0.00019998594966446248,
-      "loss": 46.0,
-      "step": 33145
-    },
-    {
-      "epoch": 5.33793630983534,
-      "grad_norm": 0.0016668703174218535,
-      "learning_rate": 0.00019998594881640345,
-      "loss": 46.0,
-      "step": 33146
-    },
-    {
-      "epoch": 5.338097346914127,
-      "grad_norm": 0.0018223979277536273,
-      "learning_rate": 0.00019998594796831877,
-      "loss": 46.0,
-      "step": 33147
-    },
-    {
-      "epoch": 5.3382583839929145,
-      "grad_norm": 0.0010866760276257992,
-      "learning_rate": 0.00019998594712020853,
-      "loss": 46.0,
-      "step": 33148
-    },
-    {
-      "epoch": 5.338419421071702,
-      "grad_norm": 0.005003968719393015,
-      "learning_rate": 0.0001999859462720727,
-      "loss": 46.0,
-      "step": 33149
-    },
-    {
-      "epoch": 5.338580458150489,
-      "grad_norm": 0.0035697175189852715,
-      "learning_rate": 0.0001999859454239113,
-      "loss": 46.0,
-      "step": 33150
-    },
-    {
-      "epoch": 5.338741495229277,
-      "grad_norm": 0.004712163005024195,
-      "learning_rate": 0.00019998594457572427,
-      "loss": 46.0,
-      "step": 33151
-    },
-    {
-      "epoch": 5.338902532308064,
-      "grad_norm": 0.0019653046038001776,
-      "learning_rate": 0.00019998594372751166,
-      "loss": 46.0,
-      "step": 33152
-    },
-    {
-      "epoch": 5.339063569386852,
-      "grad_norm": 0.0026867883279919624,
-      "learning_rate": 0.00019998594287927345,
-      "loss": 46.0,
-      "step": 33153
-    },
-    {
-      "epoch": 5.339224606465638,
-      "grad_norm": 0.00278069986961782,
-      "learning_rate": 0.00019998594203100964,
-      "loss": 46.0,
-      "step": 33154
-    },
-    {
-      "epoch": 5.339385643544426,
-      "grad_norm": 0.005466041620820761,
-      "learning_rate": 0.00019998594118272026,
-      "loss": 46.0,
-      "step": 33155
-    },
-    {
-      "epoch": 5.339546680623213,
-      "grad_norm": 0.0047329203225672245,
-      "learning_rate": 0.0001999859403344053,
-      "loss": 46.0,
-      "step": 33156
-    },
-    {
-      "epoch": 5.339707717702001,
-      "grad_norm": 0.002004955429583788,
-      "learning_rate": 0.00019998593948606473,
-      "loss": 46.0,
-      "step": 33157
-    },
-    {
-      "epoch": 5.339868754780788,
-      "grad_norm": 0.011458859778940678,
-      "learning_rate": 0.0001999859386376986,
-      "loss": 46.0,
-      "step": 33158
-    },
-    {
-      "epoch": 5.340029791859576,
-      "grad_norm": 0.0017322632484138012,
-      "learning_rate": 0.00019998593778930684,
-      "loss": 46.0,
-      "step": 33159
-    },
-    {
-      "epoch": 5.340190828938363,
-      "grad_norm": 0.0011169033823534846,
-      "learning_rate": 0.0001999859369408895,
-      "loss": 46.0,
-      "step": 33160
-    },
-    {
-      "epoch": 5.3403518660171505,
-      "grad_norm": 0.0022553205490112305,
-      "learning_rate": 0.00019998593609244655,
-      "loss": 46.0,
-      "step": 33161
-    },
-    {
-      "epoch": 5.340512903095938,
-      "grad_norm": 0.0006720060482621193,
-      "learning_rate": 0.00019998593524397804,
-      "loss": 46.0,
-      "step": 33162
-    },
-    {
-      "epoch": 5.340673940174725,
-      "grad_norm": 0.0028036385774612427,
-      "learning_rate": 0.0001999859343954839,
-      "loss": 46.0,
-      "step": 33163
-    },
-    {
-      "epoch": 5.340834977253513,
-      "grad_norm": 0.002368019660934806,
-      "learning_rate": 0.00019998593354696422,
-      "loss": 46.0,
-      "step": 33164
-    },
-    {
-      "epoch": 5.3409960143323,
-      "grad_norm": 0.0032712381798774004,
-      "learning_rate": 0.00019998593269841895,
-      "loss": 46.0,
-      "step": 33165
-    },
-    {
-      "epoch": 5.341157051411088,
-      "grad_norm": 0.0012793343048542738,
-      "learning_rate": 0.00019998593184984803,
-      "loss": 46.0,
-      "step": 33166
-    },
-    {
-      "epoch": 5.341318088489874,
-      "grad_norm": 0.003984276670962572,
-      "learning_rate": 0.00019998593100125156,
-      "loss": 46.0,
-      "step": 33167
-    },
-    {
-      "epoch": 5.341479125568662,
-      "grad_norm": 0.002494280692189932,
-      "learning_rate": 0.0001999859301526295,
-      "loss": 46.0,
-      "step": 33168
-    },
-    {
-      "epoch": 5.341640162647449,
-      "grad_norm": 0.0025377566926181316,
-      "learning_rate": 0.00019998592930398182,
-      "loss": 46.0,
-      "step": 33169
-    },
-    {
-      "epoch": 5.341801199726237,
-      "grad_norm": 0.002192853018641472,
-      "learning_rate": 0.00019998592845530858,
-      "loss": 46.0,
-      "step": 33170
-    },
-    {
-      "epoch": 5.341962236805024,
-      "grad_norm": 0.004914535209536552,
-      "learning_rate": 0.00019998592760660976,
-      "loss": 46.0,
-      "step": 33171
-    },
-    {
-      "epoch": 5.342123273883812,
-      "grad_norm": 0.00907733105123043,
-      "learning_rate": 0.0001999859267578853,
-      "loss": 46.0,
-      "step": 33172
-    },
-    {
-      "epoch": 5.342284310962599,
-      "grad_norm": 0.006224296521395445,
-      "learning_rate": 0.00019998592590913527,
-      "loss": 46.0,
-      "step": 33173
-    },
-    {
-      "epoch": 5.3424453480413865,
-      "grad_norm": 0.0022465987130999565,
-      "learning_rate": 0.00019998592506035966,
-      "loss": 46.0,
-      "step": 33174
-    },
-    {
-      "epoch": 5.342606385120174,
-      "grad_norm": 0.001446399255655706,
-      "learning_rate": 0.00019998592421155843,
-      "loss": 46.0,
-      "step": 33175
-    },
-    {
-      "epoch": 5.342767422198961,
-      "grad_norm": 0.008680148050189018,
-      "learning_rate": 0.00019998592336273164,
-      "loss": 46.0,
-      "step": 33176
-    },
-    {
-      "epoch": 5.342928459277749,
-      "grad_norm": 0.0034825035836547613,
-      "learning_rate": 0.00019998592251387927,
-      "loss": 46.0,
-      "step": 33177
-    },
-    {
-      "epoch": 5.343089496356536,
-      "grad_norm": 0.0027706236578524113,
-      "learning_rate": 0.00019998592166500128,
-      "loss": 46.0,
-      "step": 33178
-    },
-    {
-      "epoch": 5.343250533435324,
-      "grad_norm": 0.0012397044338285923,
-      "learning_rate": 0.0001999859208160977,
-      "loss": 46.0,
-      "step": 33179
-    },
-    {
-      "epoch": 5.343411570514111,
-      "grad_norm": 0.019820643588900566,
-      "learning_rate": 0.00019998591996716854,
-      "loss": 46.0,
-      "step": 33180
-    },
-    {
-      "epoch": 5.343572607592899,
-      "grad_norm": 0.0028394998516887426,
-      "learning_rate": 0.0001999859191182138,
-      "loss": 46.0,
-      "step": 33181
-    },
-    {
-      "epoch": 5.343733644671685,
-      "grad_norm": 0.01724991388618946,
-      "learning_rate": 0.00019998591826923346,
-      "loss": 46.0,
-      "step": 33182
-    },
-    {
-      "epoch": 5.343894681750473,
-      "grad_norm": 0.010076344013214111,
-      "learning_rate": 0.0001999859174202275,
-      "loss": 46.0,
-      "step": 33183
-    },
-    {
-      "epoch": 5.34405571882926,
-      "grad_norm": 0.006137934513390064,
-      "learning_rate": 0.00019998591657119597,
-      "loss": 46.0,
-      "step": 33184
-    },
-    {
-      "epoch": 5.344216755908048,
-      "grad_norm": 0.0028756221290677786,
-      "learning_rate": 0.00019998591572213884,
-      "loss": 46.0,
-      "step": 33185
-    },
-    {
-      "epoch": 5.344377792986835,
-      "grad_norm": 0.002660792088136077,
-      "learning_rate": 0.00019998591487305613,
-      "loss": 46.0,
-      "step": 33186
-    },
-    {
-      "epoch": 5.3445388300656225,
-      "grad_norm": 0.012917597778141499,
-      "learning_rate": 0.00019998591402394783,
-      "loss": 46.0,
-      "step": 33187
-    },
-    {
-      "epoch": 5.34469986714441,
-      "grad_norm": 0.003583114128559828,
-      "learning_rate": 0.00019998591317481394,
-      "loss": 46.0,
-      "step": 33188
-    },
-    {
-      "epoch": 5.344860904223197,
-      "grad_norm": 0.027246685698628426,
-      "learning_rate": 0.00019998591232565444,
-      "loss": 46.0,
-      "step": 33189
-    },
-    {
-      "epoch": 5.345021941301985,
-      "grad_norm": 0.005287294276058674,
-      "learning_rate": 0.00019998591147646938,
-      "loss": 46.0,
-      "step": 33190
-    },
-    {
-      "epoch": 5.345182978380772,
-      "grad_norm": 0.0030744103714823723,
-      "learning_rate": 0.0001999859106272587,
-      "loss": 46.0,
-      "step": 33191
-    },
-    {
-      "epoch": 5.34534401545956,
-      "grad_norm": 0.0018977083964273334,
-      "learning_rate": 0.00019998590977802244,
-      "loss": 46.0,
-      "step": 33192
-    },
-    {
-      "epoch": 5.345505052538347,
-      "grad_norm": 0.0036831132601946592,
-      "learning_rate": 0.00019998590892876062,
-      "loss": 46.0,
-      "step": 33193
-    },
-    {
-      "epoch": 5.345666089617135,
-      "grad_norm": 0.0054842764511704445,
-      "learning_rate": 0.00019998590807947315,
-      "loss": 46.0,
-      "step": 33194
-    },
-    {
-      "epoch": 5.345827126695922,
-      "grad_norm": 0.013415954075753689,
-      "learning_rate": 0.0001999859072301601,
-      "loss": 46.0,
-      "step": 33195
-    },
-    {
-      "epoch": 5.3459881637747095,
-      "grad_norm": 0.005169915035367012,
-      "learning_rate": 0.00019998590638082152,
-      "loss": 46.0,
-      "step": 33196
-    },
-    {
-      "epoch": 5.346149200853496,
-      "grad_norm": 0.0014741383492946625,
-      "learning_rate": 0.0001999859055314573,
-      "loss": 46.0,
-      "step": 33197
-    },
-    {
-      "epoch": 5.3463102379322835,
-      "grad_norm": 0.010834702290594578,
-      "learning_rate": 0.00019998590468206748,
-      "loss": 46.0,
-      "step": 33198
-    },
-    {
-      "epoch": 5.346471275011071,
-      "grad_norm": 0.0033167554065585136,
-      "learning_rate": 0.00019998590383265208,
-      "loss": 46.0,
-      "step": 33199
-    },
-    {
-      "epoch": 5.3466323120898585,
-      "grad_norm": 0.003050136147066951,
-      "learning_rate": 0.0001999859029832111,
-      "loss": 46.0,
-      "step": 33200
-    },
-    {
-      "epoch": 5.346793349168646,
-      "grad_norm": 0.005842990707606077,
-      "learning_rate": 0.00019998590213374451,
-      "loss": 46.0,
-      "step": 33201
-    },
-    {
-      "epoch": 5.346954386247433,
-      "grad_norm": 0.004670327063649893,
-      "learning_rate": 0.00019998590128425235,
-      "loss": 46.0,
-      "step": 33202
-    },
-    {
-      "epoch": 5.347115423326221,
-      "grad_norm": 0.009079623967409134,
-      "learning_rate": 0.00019998590043473455,
-      "loss": 46.0,
-      "step": 33203
-    },
-    {
-      "epoch": 5.347276460405008,
-      "grad_norm": 0.012257524766027927,
-      "learning_rate": 0.00019998589958519124,
-      "loss": 46.0,
-      "step": 33204
-    },
-    {
-      "epoch": 5.347437497483796,
-      "grad_norm": 0.003308637300506234,
-      "learning_rate": 0.0001999858987356223,
-      "loss": 46.0,
-      "step": 33205
-    },
-    {
-      "epoch": 5.347598534562583,
-      "grad_norm": 0.0025279151741415262,
-      "learning_rate": 0.00019998589788602773,
-      "loss": 46.0,
-      "step": 33206
-    },
-    {
-      "epoch": 5.347759571641371,
-      "grad_norm": 0.01778295822441578,
-      "learning_rate": 0.0001999858970364076,
-      "loss": 46.0,
-      "step": 33207
-    },
-    {
-      "epoch": 5.347920608720158,
-      "grad_norm": 0.001226358930580318,
-      "learning_rate": 0.00019998589618676186,
-      "loss": 46.0,
-      "step": 33208
-    },
-    {
-      "epoch": 5.3480816457989455,
-      "grad_norm": 0.002565372036769986,
-      "learning_rate": 0.00019998589533709056,
-      "loss": 46.0,
-      "step": 33209
-    },
-    {
-      "epoch": 5.348242682877733,
-      "grad_norm": 0.009788377210497856,
-      "learning_rate": 0.00019998589448739368,
-      "loss": 46.0,
-      "step": 33210
-    },
-    {
-      "epoch": 5.34840371995652,
-      "grad_norm": 0.003157887374982238,
-      "learning_rate": 0.00019998589363767118,
-      "loss": 46.0,
-      "step": 33211
-    },
-    {
-      "epoch": 5.348564757035307,
-      "grad_norm": 0.017140492796897888,
-      "learning_rate": 0.0001999858927879231,
-      "loss": 46.0,
-      "step": 33212
-    },
-    {
-      "epoch": 5.348725794114094,
-      "grad_norm": 0.008377748541533947,
-      "learning_rate": 0.00019998589193814944,
-      "loss": 46.0,
-      "step": 33213
-    },
-    {
-      "epoch": 5.348886831192882,
-      "grad_norm": 0.0018164287321269512,
-      "learning_rate": 0.00019998589108835013,
-      "loss": 46.0,
-      "step": 33214
-    },
-    {
-      "epoch": 5.349047868271669,
-      "grad_norm": 0.0025338446721434593,
-      "learning_rate": 0.0001999858902385253,
-      "loss": 46.0,
-      "step": 33215
-    },
-    {
-      "epoch": 5.349208905350457,
-      "grad_norm": 0.0017539168475195765,
-      "learning_rate": 0.00019998588938867484,
-      "loss": 46.0,
-      "step": 33216
-    },
-    {
-      "epoch": 5.349369942429244,
-      "grad_norm": 0.0034128446131944656,
-      "learning_rate": 0.00019998588853879882,
-      "loss": 46.0,
-      "step": 33217
-    },
-    {
-      "epoch": 5.349530979508032,
-      "grad_norm": 0.009461546316742897,
-      "learning_rate": 0.00019998588768889718,
-      "loss": 46.0,
-      "step": 33218
-    },
-    {
-      "epoch": 5.349692016586819,
-      "grad_norm": 0.0018438647966831923,
-      "learning_rate": 0.00019998588683896995,
-      "loss": 46.0,
-      "step": 33219
-    },
-    {
-      "epoch": 5.349853053665607,
-      "grad_norm": 0.005322962999343872,
-      "learning_rate": 0.00019998588598901714,
-      "loss": 46.0,
-      "step": 33220
-    },
-    {
-      "epoch": 5.350014090744394,
-      "grad_norm": 0.028859008103609085,
-      "learning_rate": 0.00019998588513903874,
-      "loss": 46.0,
-      "step": 33221
-    },
-    {
-      "epoch": 5.3501751278231815,
-      "grad_norm": 0.0017195470863953233,
-      "learning_rate": 0.0001999858842890347,
-      "loss": 46.0,
-      "step": 33222
-    },
-    {
-      "epoch": 5.350336164901969,
-      "grad_norm": 0.01269600447267294,
-      "learning_rate": 0.00019998588343900513,
-      "loss": 46.0,
-      "step": 33223
-    },
-    {
-      "epoch": 5.350497201980756,
-      "grad_norm": 0.009484685957431793,
-      "learning_rate": 0.00019998588258894994,
-      "loss": 46.0,
-      "step": 33224
-    },
-    {
-      "epoch": 5.350658239059544,
-      "grad_norm": 0.00613419059664011,
-      "learning_rate": 0.00019998588173886916,
-      "loss": 46.0,
-      "step": 33225
-    },
-    {
-      "epoch": 5.350819276138331,
-      "grad_norm": 0.01434311643242836,
-      "learning_rate": 0.00019998588088876283,
-      "loss": 46.0,
-      "step": 33226
-    },
-    {
-      "epoch": 5.350980313217118,
-      "grad_norm": 0.01462017185986042,
-      "learning_rate": 0.00019998588003863085,
-      "loss": 46.0,
-      "step": 33227
-    },
-    {
-      "epoch": 5.351141350295905,
-      "grad_norm": 0.009560765698552132,
-      "learning_rate": 0.00019998587918847332,
-      "loss": 46.0,
-      "step": 33228
-    },
-    {
-      "epoch": 5.351302387374693,
-      "grad_norm": 0.0227262731641531,
-      "learning_rate": 0.0001999858783382902,
-      "loss": 46.0,
-      "step": 33229
-    },
-    {
-      "epoch": 5.35146342445348,
-      "grad_norm": 0.001176685094833374,
-      "learning_rate": 0.00019998587748808145,
-      "loss": 46.0,
-      "step": 33230
-    },
-    {
-      "epoch": 5.351624461532268,
-      "grad_norm": 0.002229700330644846,
-      "learning_rate": 0.00019998587663784713,
-      "loss": 46.0,
-      "step": 33231
-    },
-    {
-      "epoch": 5.351785498611055,
-      "grad_norm": 0.009022231213748455,
-      "learning_rate": 0.00019998587578758722,
-      "loss": 46.0,
-      "step": 33232
-    },
-    {
-      "epoch": 5.351946535689843,
-      "grad_norm": 0.0059882174246013165,
-      "learning_rate": 0.00019998587493730172,
-      "loss": 46.0,
-      "step": 33233
-    },
-    {
-      "epoch": 5.35210757276863,
-      "grad_norm": 0.001762591302394867,
-      "learning_rate": 0.00019998587408699063,
-      "loss": 46.0,
-      "step": 33234
-    },
-    {
-      "epoch": 5.3522686098474175,
-      "grad_norm": 0.0027219848707318306,
-      "learning_rate": 0.00019998587323665393,
-      "loss": 46.0,
-      "step": 33235
-    },
-    {
-      "epoch": 5.352429646926205,
-      "grad_norm": 0.0027197706513106823,
-      "learning_rate": 0.00019998587238629167,
-      "loss": 46.0,
-      "step": 33236
-    },
-    {
-      "epoch": 5.352590684004992,
-      "grad_norm": 0.011602818965911865,
-      "learning_rate": 0.0001999858715359038,
-      "loss": 46.0,
-      "step": 33237
-    },
-    {
-      "epoch": 5.35275172108378,
-      "grad_norm": 0.0031015598215162754,
-      "learning_rate": 0.00019998587068549036,
-      "loss": 46.0,
-      "step": 33238
-    },
-    {
-      "epoch": 5.352912758162567,
-      "grad_norm": 0.006244780495762825,
-      "learning_rate": 0.0001999858698350513,
-      "loss": 46.0,
-      "step": 33239
-    },
-    {
-      "epoch": 5.353073795241355,
-      "grad_norm": 0.005492097232490778,
-      "learning_rate": 0.00019998586898458664,
-      "loss": 46.0,
-      "step": 33240
-    },
-    {
-      "epoch": 5.353234832320141,
-      "grad_norm": 0.01124728936702013,
-      "learning_rate": 0.00019998586813409642,
-      "loss": 46.0,
-      "step": 33241
-    },
-    {
-      "epoch": 5.353395869398929,
-      "grad_norm": 0.005954296328127384,
-      "learning_rate": 0.0001999858672835806,
-      "loss": 46.0,
-      "step": 33242
-    },
-    {
-      "epoch": 5.353556906477716,
-      "grad_norm": 0.030257176607847214,
-      "learning_rate": 0.00019998586643303918,
-      "loss": 46.0,
-      "step": 33243
-    },
-    {
-      "epoch": 5.353717943556504,
-      "grad_norm": 0.0056516434997320175,
-      "learning_rate": 0.0001999858655824722,
-      "loss": 46.0,
-      "step": 33244
-    },
-    {
-      "epoch": 5.353878980635291,
-      "grad_norm": 0.021140802651643753,
-      "learning_rate": 0.0001999858647318796,
-      "loss": 46.0,
-      "step": 33245
-    },
-    {
-      "epoch": 5.354040017714079,
-      "grad_norm": 0.006468849256634712,
-      "learning_rate": 0.0001999858638812614,
-      "loss": 46.0,
-      "step": 33246
-    },
-    {
-      "epoch": 5.354201054792866,
-      "grad_norm": 0.011477414518594742,
-      "learning_rate": 0.00019998586303061764,
-      "loss": 46.0,
-      "step": 33247
-    },
-    {
-      "epoch": 5.3543620918716535,
-      "grad_norm": 0.008933735080063343,
-      "learning_rate": 0.00019998586217994825,
-      "loss": 46.0,
-      "step": 33248
-    },
-    {
-      "epoch": 5.354523128950441,
-      "grad_norm": 0.0035963142290711403,
-      "learning_rate": 0.0001999858613292533,
-      "loss": 46.0,
-      "step": 33249
-    },
-    {
-      "epoch": 5.354684166029228,
-      "grad_norm": 0.002323634224012494,
-      "learning_rate": 0.00019998586047853274,
-      "loss": 46.0,
-      "step": 33250
-    },
-    {
-      "epoch": 5.354845203108016,
-      "grad_norm": 0.003420127322897315,
-      "learning_rate": 0.0001999858596277866,
-      "loss": 46.0,
-      "step": 33251
-    },
-    {
-      "epoch": 5.355006240186803,
-      "grad_norm": 0.0044273692183196545,
-      "learning_rate": 0.00019998585877701485,
-      "loss": 46.0,
-      "step": 33252
-    },
-    {
-      "epoch": 5.355167277265591,
-      "grad_norm": 0.01521384809166193,
-      "learning_rate": 0.00019998585792621753,
-      "loss": 46.0,
-      "step": 33253
-    },
-    {
-      "epoch": 5.355328314344378,
-      "grad_norm": 0.005429886281490326,
-      "learning_rate": 0.00019998585707539461,
-      "loss": 46.0,
-      "step": 33254
-    },
-    {
-      "epoch": 5.355489351423165,
-      "grad_norm": 0.008967528119683266,
-      "learning_rate": 0.00019998585622454612,
-      "loss": 46.0,
-      "step": 33255
-    },
-    {
-      "epoch": 5.355650388501952,
-      "grad_norm": 0.002316574566066265,
-      "learning_rate": 0.000199985855373672,
-      "loss": 46.0,
-      "step": 33256
-    },
-    {
-      "epoch": 5.35581142558074,
-      "grad_norm": 0.002046039793640375,
-      "learning_rate": 0.00019998585452277233,
-      "loss": 46.0,
-      "step": 33257
-    },
-    {
-      "epoch": 5.355972462659527,
-      "grad_norm": 0.013033757917582989,
-      "learning_rate": 0.00019998585367184704,
-      "loss": 46.0,
-      "step": 33258
-    },
-    {
-      "epoch": 5.3561334997383145,
-      "grad_norm": 0.00303908484056592,
-      "learning_rate": 0.00019998585282089617,
-      "loss": 46.0,
-      "step": 33259
-    },
-    {
-      "epoch": 5.356294536817102,
-      "grad_norm": 0.007970873266458511,
-      "learning_rate": 0.00019998585196991968,
-      "loss": 46.0,
-      "step": 33260
-    },
-    {
-      "epoch": 5.356455573895889,
-      "grad_norm": 0.02804405987262726,
-      "learning_rate": 0.00019998585111891763,
-      "loss": 46.0,
-      "step": 33261
-    },
-    {
-      "epoch": 5.356616610974677,
-      "grad_norm": 0.0022014323621988297,
-      "learning_rate": 0.00019998585026789,
-      "loss": 46.0,
-      "step": 33262
-    },
-    {
-      "epoch": 5.356777648053464,
-      "grad_norm": 0.006103958934545517,
-      "learning_rate": 0.00019998584941683674,
-      "loss": 46.0,
-      "step": 33263
-    },
-    {
-      "epoch": 5.356938685132252,
-      "grad_norm": 0.001840987242758274,
-      "learning_rate": 0.00019998584856575793,
-      "loss": 46.0,
-      "step": 33264
-    },
-    {
-      "epoch": 5.357099722211039,
-      "grad_norm": 0.009221870452165604,
-      "learning_rate": 0.0001999858477146535,
-      "loss": 46.0,
-      "step": 33265
-    },
-    {
-      "epoch": 5.357260759289827,
-      "grad_norm": 0.007611990440636873,
-      "learning_rate": 0.0001999858468635235,
-      "loss": 46.0,
-      "step": 33266
-    },
-    {
-      "epoch": 5.357421796368614,
-      "grad_norm": 0.018108505755662918,
-      "learning_rate": 0.0001999858460123679,
-      "loss": 46.0,
-      "step": 33267
-    },
-    {
-      "epoch": 5.357582833447402,
-      "grad_norm": 0.0021959838923066854,
-      "learning_rate": 0.00019998584516118669,
-      "loss": 46.0,
-      "step": 33268
-    },
-    {
-      "epoch": 5.357743870526189,
-      "grad_norm": 0.007607196923345327,
-      "learning_rate": 0.00019998584430997989,
-      "loss": 46.0,
-      "step": 33269
-    },
-    {
-      "epoch": 5.357904907604976,
-      "grad_norm": 0.0022318274714052677,
-      "learning_rate": 0.00019998584345874752,
-      "loss": 46.0,
-      "step": 33270
-    },
-    {
-      "epoch": 5.358065944683763,
-      "grad_norm": 0.0009972254047170281,
-      "learning_rate": 0.00019998584260748955,
-      "loss": 46.0,
-      "step": 33271
-    },
-    {
-      "epoch": 5.3582269817625505,
-      "grad_norm": 0.0031633200123906136,
-      "learning_rate": 0.000199985841756206,
-      "loss": 46.0,
-      "step": 33272
-    },
-    {
-      "epoch": 5.358388018841338,
-      "grad_norm": 0.0016663161804899573,
-      "learning_rate": 0.00019998584090489684,
-      "loss": 46.0,
-      "step": 33273
-    },
-    {
-      "epoch": 5.358549055920125,
-      "grad_norm": 0.004439766984432936,
-      "learning_rate": 0.0001999858400535621,
-      "loss": 46.0,
-      "step": 33274
-    },
-    {
-      "epoch": 5.358710092998913,
-      "grad_norm": 0.003509029047563672,
-      "learning_rate": 0.00019998583920220178,
-      "loss": 46.0,
-      "step": 33275
-    },
-    {
-      "epoch": 5.3588711300777,
-      "grad_norm": 0.006376531440764666,
-      "learning_rate": 0.00019998583835081584,
-      "loss": 46.0,
-      "step": 33276
-    },
-    {
-      "epoch": 5.359032167156488,
-      "grad_norm": 0.021703248843550682,
-      "learning_rate": 0.00019998583749940432,
-      "loss": 46.0,
-      "step": 33277
-    },
-    {
-      "epoch": 5.359193204235275,
-      "grad_norm": 0.0021995895076543093,
-      "learning_rate": 0.0001999858366479672,
-      "loss": 46.0,
-      "step": 33278
-    },
-    {
-      "epoch": 5.359354241314063,
-      "grad_norm": 0.0010908916592597961,
-      "learning_rate": 0.0001999858357965045,
-      "loss": 46.0,
-      "step": 33279
-    },
-    {
-      "epoch": 5.35951527839285,
-      "grad_norm": 0.004063006024807692,
-      "learning_rate": 0.00019998583494501625,
-      "loss": 46.0,
-      "step": 33280
-    },
-    {
-      "epoch": 5.359676315471638,
-      "grad_norm": 0.003869170555844903,
-      "learning_rate": 0.00019998583409350235,
-      "loss": 46.0,
-      "step": 33281
-    },
-    {
-      "epoch": 5.359837352550425,
-      "grad_norm": 0.00443886686116457,
-      "learning_rate": 0.00019998583324196289,
-      "loss": 46.0,
-      "step": 33282
-    },
-    {
-      "epoch": 5.3599983896292125,
-      "grad_norm": 0.0015142090851441026,
-      "learning_rate": 0.0001999858323903978,
-      "loss": 46.0,
-      "step": 33283
-    },
-    {
-      "epoch": 5.360159426708,
-      "grad_norm": 0.0017443251563236117,
-      "learning_rate": 0.00019998583153880715,
-      "loss": 46.0,
-      "step": 33284
-    },
-    {
-      "epoch": 5.3603204637867865,
-      "grad_norm": 0.003215186530724168,
-      "learning_rate": 0.0001999858306871909,
-      "loss": 46.0,
-      "step": 33285
-    },
-    {
-      "epoch": 5.360481500865574,
-      "grad_norm": 0.004337142687290907,
-      "learning_rate": 0.0001999858298355491,
-      "loss": 46.0,
-      "step": 33286
-    },
-    {
-      "epoch": 5.360642537944361,
-      "grad_norm": 0.003072574269026518,
-      "learning_rate": 0.00019998582898388164,
-      "loss": 46.0,
-      "step": 33287
-    },
-    {
-      "epoch": 5.360803575023149,
-      "grad_norm": 0.014539993368089199,
-      "learning_rate": 0.00019998582813218863,
-      "loss": 46.0,
-      "step": 33288
-    },
-    {
-      "epoch": 5.360964612101936,
-      "grad_norm": 0.004221667069941759,
-      "learning_rate": 0.00019998582728047003,
-      "loss": 46.0,
-      "step": 33289
-    },
-    {
-      "epoch": 5.361125649180724,
-      "grad_norm": 0.01120102871209383,
-      "learning_rate": 0.00019998582642872582,
-      "loss": 46.0,
-      "step": 33290
-    },
-    {
-      "epoch": 5.361286686259511,
-      "grad_norm": 0.0028298741672188044,
-      "learning_rate": 0.00019998582557695602,
-      "loss": 46.0,
-      "step": 33291
-    },
-    {
-      "epoch": 5.361447723338299,
-      "grad_norm": 0.0013431668048724532,
-      "learning_rate": 0.00019998582472516063,
-      "loss": 46.0,
-      "step": 33292
-    },
-    {
-      "epoch": 5.361608760417086,
-      "grad_norm": 0.008113016374409199,
-      "learning_rate": 0.00019998582387333966,
-      "loss": 46.0,
-      "step": 33293
-    },
-    {
-      "epoch": 5.361769797495874,
-      "grad_norm": 0.0029859375208616257,
-      "learning_rate": 0.0001999858230214931,
-      "loss": 46.0,
-      "step": 33294
-    },
-    {
-      "epoch": 5.361930834574661,
-      "grad_norm": 0.01254460122436285,
-      "learning_rate": 0.00019998582216962092,
-      "loss": 46.0,
-      "step": 33295
-    },
-    {
-      "epoch": 5.3620918716534485,
-      "grad_norm": 0.0012836339883506298,
-      "learning_rate": 0.00019998582131772316,
-      "loss": 46.0,
-      "step": 33296
-    },
-    {
-      "epoch": 5.362252908732236,
-      "grad_norm": 0.0015273225726559758,
-      "learning_rate": 0.00019998582046579984,
-      "loss": 46.0,
-      "step": 33297
-    },
-    {
-      "epoch": 5.362413945811023,
-      "grad_norm": 0.005086224526166916,
-      "learning_rate": 0.00019998581961385093,
-      "loss": 46.0,
-      "step": 33298
-    },
-    {
-      "epoch": 5.362574982889811,
-      "grad_norm": 0.006983438041061163,
-      "learning_rate": 0.00019998581876187638,
-      "loss": 46.0,
-      "step": 33299
-    },
-    {
-      "epoch": 5.362736019968597,
-      "grad_norm": 0.006965779233723879,
-      "learning_rate": 0.00019998581790987627,
-      "loss": 46.0,
-      "step": 33300
-    },
-    {
-      "epoch": 5.362897057047385,
-      "grad_norm": 0.008995898999273777,
-      "learning_rate": 0.00019998581705785057,
-      "loss": 46.0,
-      "step": 33301
-    },
-    {
-      "epoch": 5.363058094126172,
-      "grad_norm": 0.0008115133969113231,
-      "learning_rate": 0.00019998581620579925,
-      "loss": 46.0,
-      "step": 33302
-    },
-    {
-      "epoch": 5.36321913120496,
-      "grad_norm": 0.004873207304626703,
-      "learning_rate": 0.00019998581535372235,
-      "loss": 46.0,
-      "step": 33303
-    },
-    {
-      "epoch": 5.363380168283747,
-      "grad_norm": 0.003568889107555151,
-      "learning_rate": 0.0001999858145016199,
-      "loss": 46.0,
-      "step": 33304
-    },
-    {
-      "epoch": 5.363541205362535,
-      "grad_norm": 0.0034354953095316887,
-      "learning_rate": 0.00019998581364949182,
-      "loss": 46.0,
-      "step": 33305
-    },
-    {
-      "epoch": 5.363702242441322,
-      "grad_norm": 0.006653422024101019,
-      "learning_rate": 0.00019998581279733816,
-      "loss": 46.0,
-      "step": 33306
-    },
-    {
-      "epoch": 5.3638632795201096,
-      "grad_norm": 0.002001228742301464,
-      "learning_rate": 0.0001999858119451589,
-      "loss": 46.0,
-      "step": 33307
-    },
-    {
-      "epoch": 5.364024316598897,
-      "grad_norm": 0.004664026200771332,
-      "learning_rate": 0.00019998581109295408,
-      "loss": 46.0,
-      "step": 33308
-    },
-    {
-      "epoch": 5.3641853536776845,
-      "grad_norm": 0.0033457849640399218,
-      "learning_rate": 0.00019998581024072363,
-      "loss": 46.0,
-      "step": 33309
-    },
-    {
-      "epoch": 5.364346390756472,
-      "grad_norm": 0.0062686987221241,
-      "learning_rate": 0.00019998580938846762,
-      "loss": 46.0,
-      "step": 33310
-    },
-    {
-      "epoch": 5.364507427835259,
-      "grad_norm": 0.004319933243095875,
-      "learning_rate": 0.000199985808536186,
-      "loss": 46.0,
-      "step": 33311
-    },
-    {
-      "epoch": 5.364668464914047,
-      "grad_norm": 0.013152024708688259,
-      "learning_rate": 0.00019998580768387878,
-      "loss": 46.0,
-      "step": 33312
-    },
-    {
-      "epoch": 5.364829501992834,
-      "grad_norm": 0.004020859021693468,
-      "learning_rate": 0.00019998580683154598,
-      "loss": 46.0,
-      "step": 33313
-    },
-    {
-      "epoch": 5.364990539071622,
-      "grad_norm": 0.0056895967572927475,
-      "learning_rate": 0.00019998580597918757,
-      "loss": 46.0,
-      "step": 33314
-    },
-    {
-      "epoch": 5.365151576150408,
-      "grad_norm": 0.0030085165053606033,
-      "learning_rate": 0.0001999858051268036,
-      "loss": 46.0,
-      "step": 33315
-    },
-    {
-      "epoch": 5.365312613229196,
-      "grad_norm": 0.0010852243285626173,
-      "learning_rate": 0.000199985804274394,
-      "loss": 46.0,
-      "step": 33316
-    },
-    {
-      "epoch": 5.365473650307983,
-      "grad_norm": 0.002301767934113741,
-      "learning_rate": 0.00019998580342195886,
-      "loss": 46.0,
-      "step": 33317
-    },
-    {
-      "epoch": 5.365634687386771,
-      "grad_norm": 0.0026175789535045624,
-      "learning_rate": 0.0001999858025694981,
-      "loss": 46.0,
-      "step": 33318
-    },
-    {
-      "epoch": 5.365795724465558,
-      "grad_norm": 0.007056757807731628,
-      "learning_rate": 0.00019998580171701175,
-      "loss": 46.0,
-      "step": 33319
-    },
-    {
-      "epoch": 5.3659567615443455,
-      "grad_norm": 0.0018180046463385224,
-      "learning_rate": 0.00019998580086449982,
-      "loss": 46.0,
-      "step": 33320
-    },
-    {
-      "epoch": 5.366117798623133,
-      "grad_norm": 0.002870079828426242,
-      "learning_rate": 0.00019998580001196227,
-      "loss": 46.0,
-      "step": 33321
-    },
-    {
-      "epoch": 5.36627883570192,
-      "grad_norm": 0.003968583419919014,
-      "learning_rate": 0.00019998579915939916,
-      "loss": 46.0,
-      "step": 33322
-    },
-    {
-      "epoch": 5.366439872780708,
-      "grad_norm": 0.0013545351102948189,
-      "learning_rate": 0.00019998579830681043,
-      "loss": 46.0,
-      "step": 33323
-    },
-    {
-      "epoch": 5.366600909859495,
-      "grad_norm": 0.005956271663308144,
-      "learning_rate": 0.00019998579745419615,
-      "loss": 46.0,
-      "step": 33324
-    },
-    {
-      "epoch": 5.366761946938283,
-      "grad_norm": 0.003099706256762147,
-      "learning_rate": 0.00019998579660155622,
-      "loss": 46.0,
-      "step": 33325
-    },
-    {
-      "epoch": 5.36692298401707,
-      "grad_norm": 0.0016058270120993257,
-      "learning_rate": 0.00019998579574889076,
-      "loss": 46.0,
-      "step": 33326
-    },
-    {
-      "epoch": 5.367084021095858,
-      "grad_norm": 0.013404106721282005,
-      "learning_rate": 0.0001999857948961997,
-      "loss": 46.0,
-      "step": 33327
-    },
-    {
-      "epoch": 5.367245058174644,
-      "grad_norm": 0.003390738507732749,
-      "learning_rate": 0.000199985794043483,
-      "loss": 46.0,
-      "step": 33328
-    },
-    {
-      "epoch": 5.367406095253432,
-      "grad_norm": 0.002164378296583891,
-      "learning_rate": 0.00019998579319074075,
-      "loss": 46.0,
-      "step": 33329
-    },
-    {
-      "epoch": 5.367567132332219,
-      "grad_norm": 0.004207930061966181,
-      "learning_rate": 0.0001999857923379729,
-      "loss": 46.0,
-      "step": 33330
-    },
-    {
-      "epoch": 5.367728169411007,
-      "grad_norm": 0.004162882920354605,
-      "learning_rate": 0.00019998579148517944,
-      "loss": 46.0,
-      "step": 33331
-    },
-    {
-      "epoch": 5.367889206489794,
-      "grad_norm": 0.0037815652322024107,
-      "learning_rate": 0.00019998579063236043,
-      "loss": 46.0,
-      "step": 33332
-    },
-    {
-      "epoch": 5.3680502435685815,
-      "grad_norm": 0.008563963696360588,
-      "learning_rate": 0.00019998578977951578,
-      "loss": 46.0,
-      "step": 33333
-    },
-    {
-      "epoch": 5.368211280647369,
-      "grad_norm": 0.004082670900970697,
-      "learning_rate": 0.00019998578892664557,
-      "loss": 46.0,
-      "step": 33334
-    },
-    {
-      "epoch": 5.368372317726156,
-      "grad_norm": 0.002040287246927619,
-      "learning_rate": 0.00019998578807374978,
-      "loss": 46.0,
-      "step": 33335
-    },
-    {
-      "epoch": 5.368533354804944,
-      "grad_norm": 0.007208664435893297,
-      "learning_rate": 0.00019998578722082836,
-      "loss": 46.0,
-      "step": 33336
-    },
-    {
-      "epoch": 5.368694391883731,
-      "grad_norm": 0.011896511539816856,
-      "learning_rate": 0.0001999857863678814,
-      "loss": 46.0,
-      "step": 33337
-    },
-    {
-      "epoch": 5.368855428962519,
-      "grad_norm": 0.006968928035348654,
-      "learning_rate": 0.0001999857855149088,
-      "loss": 46.0,
-      "step": 33338
-    },
-    {
-      "epoch": 5.369016466041306,
-      "grad_norm": 0.010746875777840614,
-      "learning_rate": 0.00019998578466191063,
-      "loss": 46.0,
-      "step": 33339
-    },
-    {
-      "epoch": 5.369177503120094,
-      "grad_norm": 0.011602395214140415,
-      "learning_rate": 0.00019998578380888687,
-      "loss": 46.0,
-      "step": 33340
-    },
-    {
-      "epoch": 5.369338540198881,
-      "grad_norm": 0.0012585223885253072,
-      "learning_rate": 0.0001999857829558375,
-      "loss": 46.0,
-      "step": 33341
-    },
-    {
-      "epoch": 5.369499577277669,
-      "grad_norm": 0.004034592770040035,
-      "learning_rate": 0.00019998578210276256,
-      "loss": 46.0,
-      "step": 33342
-    },
-    {
-      "epoch": 5.369660614356455,
-      "grad_norm": 0.005447559989988804,
-      "learning_rate": 0.00019998578124966204,
-      "loss": 46.0,
-      "step": 33343
-    },
-    {
-      "epoch": 5.369821651435243,
-      "grad_norm": 0.01061705220490694,
-      "learning_rate": 0.00019998578039653588,
-      "loss": 46.0,
-      "step": 33344
-    },
-    {
-      "epoch": 5.36998268851403,
-      "grad_norm": 0.001415864797309041,
-      "learning_rate": 0.00019998577954338418,
-      "loss": 46.0,
-      "step": 33345
-    },
-    {
-      "epoch": 5.3701437255928175,
-      "grad_norm": 0.002299806335940957,
-      "learning_rate": 0.00019998577869020684,
-      "loss": 46.0,
-      "step": 33346
-    },
-    {
-      "epoch": 5.370304762671605,
-      "grad_norm": 0.005004958249628544,
-      "learning_rate": 0.00019998577783700397,
-      "loss": 46.0,
-      "step": 33347
-    },
-    {
-      "epoch": 5.370465799750392,
-      "grad_norm": 0.008812904357910156,
-      "learning_rate": 0.00019998577698377546,
-      "loss": 46.0,
-      "step": 33348
-    },
-    {
-      "epoch": 5.37062683682918,
-      "grad_norm": 0.0014168878551572561,
-      "learning_rate": 0.00019998577613052136,
-      "loss": 46.0,
-      "step": 33349
-    },
-    {
-      "epoch": 5.370787873907967,
-      "grad_norm": 0.0023052701726555824,
-      "learning_rate": 0.0001999857752772417,
-      "loss": 46.0,
-      "step": 33350
-    },
-    {
-      "epoch": 5.370948910986755,
-      "grad_norm": 0.004098802339285612,
-      "learning_rate": 0.00019998577442393645,
-      "loss": 46.0,
-      "step": 33351
-    },
-    {
-      "epoch": 5.371109948065542,
-      "grad_norm": 0.0010477675823494792,
-      "learning_rate": 0.00019998577357060557,
-      "loss": 46.0,
-      "step": 33352
-    },
-    {
-      "epoch": 5.37127098514433,
-      "grad_norm": 0.00528474198654294,
-      "learning_rate": 0.00019998577271724912,
-      "loss": 46.0,
-      "step": 33353
-    },
-    {
-      "epoch": 5.371432022223117,
-      "grad_norm": 0.0011248798109591007,
-      "learning_rate": 0.00019998577186386708,
-      "loss": 46.0,
-      "step": 33354
-    },
-    {
-      "epoch": 5.371593059301905,
-      "grad_norm": 0.004878604784607887,
-      "learning_rate": 0.00019998577101045946,
-      "loss": 46.0,
-      "step": 33355
-    },
-    {
-      "epoch": 5.371754096380692,
-      "grad_norm": 0.004384479019790888,
-      "learning_rate": 0.00019998577015702625,
-      "loss": 46.0,
-      "step": 33356
-    },
-    {
-      "epoch": 5.3719151334594795,
-      "grad_norm": 0.006826380733400583,
-      "learning_rate": 0.00019998576930356743,
-      "loss": 46.0,
-      "step": 33357
-    },
-    {
-      "epoch": 5.372076170538266,
-      "grad_norm": 0.004660832230001688,
-      "learning_rate": 0.00019998576845008302,
-      "loss": 46.0,
-      "step": 33358
-    },
-    {
-      "epoch": 5.3722372076170535,
-      "grad_norm": 0.01064602192491293,
-      "learning_rate": 0.00019998576759657302,
-      "loss": 46.0,
-      "step": 33359
-    },
-    {
-      "epoch": 5.372398244695841,
-      "grad_norm": 0.010420400649309158,
-      "learning_rate": 0.0001999857667430374,
-      "loss": 46.0,
-      "step": 33360
-    },
-    {
-      "epoch": 5.372559281774628,
-      "grad_norm": 0.002950671361759305,
-      "learning_rate": 0.00019998576588947623,
-      "loss": 46.0,
-      "step": 33361
-    },
-    {
-      "epoch": 5.372720318853416,
-      "grad_norm": 0.011195690371096134,
-      "learning_rate": 0.00019998576503588947,
-      "loss": 46.0,
-      "step": 33362
-    },
-    {
-      "epoch": 5.372881355932203,
-      "grad_norm": 0.00164573744405061,
-      "learning_rate": 0.0001999857641822771,
-      "loss": 46.0,
-      "step": 33363
-    },
-    {
-      "epoch": 5.373042393010991,
-      "grad_norm": 0.003164286958053708,
-      "learning_rate": 0.00019998576332863914,
-      "loss": 46.0,
-      "step": 33364
-    },
-    {
-      "epoch": 5.373203430089778,
-      "grad_norm": 0.0027676664758473635,
-      "learning_rate": 0.0001999857624749756,
-      "loss": 46.0,
-      "step": 33365
-    },
-    {
-      "epoch": 5.373364467168566,
-      "grad_norm": 0.004352774005383253,
-      "learning_rate": 0.00019998576162128648,
-      "loss": 46.0,
-      "step": 33366
-    },
-    {
-      "epoch": 5.373525504247353,
-      "grad_norm": 0.0033374002669006586,
-      "learning_rate": 0.00019998576076757173,
-      "loss": 46.0,
-      "step": 33367
-    },
-    {
-      "epoch": 5.3736865413261405,
-      "grad_norm": 0.005186352413147688,
-      "learning_rate": 0.00019998575991383142,
-      "loss": 46.0,
-      "step": 33368
-    },
-    {
-      "epoch": 5.373847578404928,
-      "grad_norm": 0.012580971233546734,
-      "learning_rate": 0.00019998575906006553,
-      "loss": 46.0,
-      "step": 33369
-    },
-    {
-      "epoch": 5.3740086154837154,
-      "grad_norm": 0.0006617176695726812,
-      "learning_rate": 0.00019998575820627401,
-      "loss": 46.0,
-      "step": 33370
-    },
-    {
-      "epoch": 5.374169652562503,
-      "grad_norm": 0.00987592339515686,
-      "learning_rate": 0.00019998575735245692,
-      "loss": 46.0,
-      "step": 33371
-    },
-    {
-      "epoch": 5.37433068964129,
-      "grad_norm": 0.008651567623019218,
-      "learning_rate": 0.00019998575649861423,
-      "loss": 46.0,
-      "step": 33372
-    },
-    {
-      "epoch": 5.374491726720077,
-      "grad_norm": 0.0022212076000869274,
-      "learning_rate": 0.00019998575564474598,
-      "loss": 46.0,
-      "step": 33373
-    },
-    {
-      "epoch": 5.374652763798864,
-      "grad_norm": 0.016480112448334694,
-      "learning_rate": 0.0001999857547908521,
-      "loss": 46.0,
-      "step": 33374
-    },
-    {
-      "epoch": 5.374813800877652,
-      "grad_norm": 0.0024351885076612234,
-      "learning_rate": 0.00019998575393693265,
-      "loss": 46.0,
-      "step": 33375
-    },
-    {
-      "epoch": 5.374974837956439,
-      "grad_norm": 0.004483151249587536,
-      "learning_rate": 0.0001999857530829876,
-      "loss": 46.0,
-      "step": 33376
-    },
-    {
-      "epoch": 5.375135875035227,
-      "grad_norm": 0.005428415257483721,
-      "learning_rate": 0.00019998575222901694,
-      "loss": 46.0,
-      "step": 33377
-    },
-    {
-      "epoch": 5.375296912114014,
-      "grad_norm": 0.0006876824190840125,
-      "learning_rate": 0.00019998575137502073,
-      "loss": 46.0,
-      "step": 33378
-    },
-    {
-      "epoch": 5.375457949192802,
-      "grad_norm": 0.011283421888947487,
-      "learning_rate": 0.0001999857505209989,
-      "loss": 46.0,
-      "step": 33379
-    },
-    {
-      "epoch": 5.375618986271589,
-      "grad_norm": 0.007189368829131126,
-      "learning_rate": 0.0001999857496669515,
-      "loss": 46.0,
-      "step": 33380
-    },
-    {
-      "epoch": 5.3757800233503765,
-      "grad_norm": 0.00192391371820122,
-      "learning_rate": 0.00019998574881287847,
-      "loss": 46.0,
-      "step": 33381
-    },
-    {
-      "epoch": 5.375941060429164,
-      "grad_norm": 0.002130111213773489,
-      "learning_rate": 0.0001999857479587799,
-      "loss": 46.0,
-      "step": 33382
-    },
-    {
-      "epoch": 5.376102097507951,
-      "grad_norm": 0.004121072124689817,
-      "learning_rate": 0.0001999857471046557,
-      "loss": 46.0,
-      "step": 33383
-    },
-    {
-      "epoch": 5.376263134586739,
-      "grad_norm": 0.01183114293962717,
-      "learning_rate": 0.00019998574625050593,
-      "loss": 46.0,
-      "step": 33384
-    },
-    {
-      "epoch": 5.376424171665526,
-      "grad_norm": 0.0017487617442384362,
-      "learning_rate": 0.00019998574539633056,
-      "loss": 46.0,
-      "step": 33385
-    },
-    {
-      "epoch": 5.376585208744314,
-      "grad_norm": 0.00510413059964776,
-      "learning_rate": 0.0001999857445421296,
-      "loss": 46.0,
-      "step": 33386
-    },
-    {
-      "epoch": 5.376746245823101,
-      "grad_norm": 0.0030885094311088324,
-      "learning_rate": 0.00019998574368790303,
-      "loss": 46.0,
-      "step": 33387
-    },
-    {
-      "epoch": 5.376907282901888,
-      "grad_norm": 0.0019977055490016937,
-      "learning_rate": 0.0001999857428336509,
-      "loss": 46.0,
-      "step": 33388
-    },
-    {
-      "epoch": 5.377068319980675,
-      "grad_norm": 0.0033095483668148518,
-      "learning_rate": 0.00019998574197937317,
-      "loss": 46.0,
-      "step": 33389
-    },
-    {
-      "epoch": 5.377229357059463,
-      "grad_norm": 0.009518763981759548,
-      "learning_rate": 0.00019998574112506986,
-      "loss": 46.0,
-      "step": 33390
-    },
-    {
-      "epoch": 5.37739039413825,
-      "grad_norm": 0.0059511540457606316,
-      "learning_rate": 0.00019998574027074094,
-      "loss": 46.0,
-      "step": 33391
-    },
-    {
-      "epoch": 5.377551431217038,
-      "grad_norm": 0.004025678150355816,
-      "learning_rate": 0.0001999857394163864,
-      "loss": 46.0,
-      "step": 33392
-    },
-    {
-      "epoch": 5.377712468295825,
-      "grad_norm": 0.007552312221378088,
-      "learning_rate": 0.0001999857385620063,
-      "loss": 46.0,
-      "step": 33393
-    },
-    {
-      "epoch": 5.3778735053746125,
-      "grad_norm": 0.0043281069956719875,
-      "learning_rate": 0.00019998573770760062,
-      "loss": 46.0,
-      "step": 33394
-    },
-    {
-      "epoch": 5.3780345424534,
-      "grad_norm": 0.009891233406960964,
-      "learning_rate": 0.00019998573685316932,
-      "loss": 46.0,
-      "step": 33395
-    },
-    {
-      "epoch": 5.378195579532187,
-      "grad_norm": 0.006546927150338888,
-      "learning_rate": 0.00019998573599871246,
-      "loss": 46.0,
-      "step": 33396
-    },
-    {
-      "epoch": 5.378356616610975,
-      "grad_norm": 0.0027265669777989388,
-      "learning_rate": 0.00019998573514423,
-      "loss": 46.0,
-      "step": 33397
-    },
-    {
-      "epoch": 5.378517653689762,
-      "grad_norm": 0.007394163403660059,
-      "learning_rate": 0.00019998573428972195,
-      "loss": 46.0,
-      "step": 33398
-    },
-    {
-      "epoch": 5.37867869076855,
-      "grad_norm": 0.014024234376847744,
-      "learning_rate": 0.0001999857334351883,
-      "loss": 46.0,
-      "step": 33399
-    },
-    {
-      "epoch": 5.378839727847337,
-      "grad_norm": 0.0036724803503602743,
-      "learning_rate": 0.00019998573258062904,
-      "loss": 46.0,
-      "step": 33400
-    },
-    {
-      "epoch": 5.379000764926124,
-      "grad_norm": 0.0011330376146361232,
-      "learning_rate": 0.00019998573172604425,
-      "loss": 46.0,
-      "step": 33401
-    },
-    {
-      "epoch": 5.379161802004911,
-      "grad_norm": 0.0036879756953567266,
-      "learning_rate": 0.0001999857308714338,
-      "loss": 46.0,
-      "step": 33402
-    },
-    {
-      "epoch": 5.379322839083699,
-      "grad_norm": 0.002011275850236416,
-      "learning_rate": 0.00019998573001679782,
-      "loss": 46.0,
-      "step": 33403
-    },
-    {
-      "epoch": 5.379483876162486,
-      "grad_norm": 0.010497014038264751,
-      "learning_rate": 0.0001999857291621362,
-      "loss": 46.0,
-      "step": 33404
-    },
-    {
-      "epoch": 5.379644913241274,
-      "grad_norm": 0.008969152346253395,
-      "learning_rate": 0.000199985728307449,
-      "loss": 46.0,
-      "step": 33405
-    },
-    {
-      "epoch": 5.379805950320061,
-      "grad_norm": 0.003393877064809203,
-      "learning_rate": 0.00019998572745273622,
-      "loss": 46.0,
-      "step": 33406
-    },
-    {
-      "epoch": 5.3799669873988485,
-      "grad_norm": 0.01113277580589056,
-      "learning_rate": 0.00019998572659799785,
-      "loss": 46.0,
-      "step": 33407
-    },
-    {
-      "epoch": 5.380128024477636,
-      "grad_norm": 0.001524369465187192,
-      "learning_rate": 0.00019998572574323386,
-      "loss": 46.0,
-      "step": 33408
-    },
-    {
-      "epoch": 5.380289061556423,
-      "grad_norm": 0.0029023520182818174,
-      "learning_rate": 0.00019998572488844432,
-      "loss": 46.0,
-      "step": 33409
-    },
-    {
-      "epoch": 5.380450098635211,
-      "grad_norm": 0.004046233836561441,
-      "learning_rate": 0.00019998572403362918,
-      "loss": 46.0,
-      "step": 33410
-    },
-    {
-      "epoch": 5.380611135713998,
-      "grad_norm": 0.0036837707739323378,
-      "learning_rate": 0.00019998572317878844,
-      "loss": 46.0,
-      "step": 33411
-    },
-    {
-      "epoch": 5.380772172792786,
-      "grad_norm": 0.003365278011187911,
-      "learning_rate": 0.0001999857223239221,
-      "loss": 46.0,
-      "step": 33412
-    },
-    {
-      "epoch": 5.380933209871573,
-      "grad_norm": 0.0011758033651858568,
-      "learning_rate": 0.00019998572146903015,
-      "loss": 46.0,
-      "step": 33413
-    },
-    {
-      "epoch": 5.381094246950361,
-      "grad_norm": 0.0019845296628773212,
-      "learning_rate": 0.00019998572061411264,
-      "loss": 46.0,
-      "step": 33414
-    },
-    {
-      "epoch": 5.381255284029148,
-      "grad_norm": 0.0005921475240029395,
-      "learning_rate": 0.00019998571975916955,
-      "loss": 46.0,
-      "step": 33415
-    },
-    {
-      "epoch": 5.381416321107935,
-      "grad_norm": 0.012319914065301418,
-      "learning_rate": 0.00019998571890420086,
-      "loss": 46.0,
-      "step": 33416
-    },
-    {
-      "epoch": 5.381577358186722,
-      "grad_norm": 0.005228742491453886,
-      "learning_rate": 0.0001999857180492066,
-      "loss": 46.0,
-      "step": 33417
-    },
-    {
-      "epoch": 5.38173839526551,
-      "grad_norm": 0.0036351035814732313,
-      "learning_rate": 0.00019998571719418668,
-      "loss": 46.0,
-      "step": 33418
-    },
-    {
-      "epoch": 5.381899432344297,
-      "grad_norm": 0.001702353940345347,
-      "learning_rate": 0.0001999857163391412,
-      "loss": 46.0,
-      "step": 33419
-    },
-    {
-      "epoch": 5.3820604694230845,
-      "grad_norm": 0.003618385409936309,
-      "learning_rate": 0.00019998571548407017,
-      "loss": 46.0,
-      "step": 33420
-    },
-    {
-      "epoch": 5.382221506501872,
-      "grad_norm": 0.0018738234648481011,
-      "learning_rate": 0.0001999857146289735,
-      "loss": 46.0,
-      "step": 33421
-    },
-    {
-      "epoch": 5.382382543580659,
-      "grad_norm": 0.0019335250835865736,
-      "learning_rate": 0.00019998571377385127,
-      "loss": 46.0,
-      "step": 33422
-    },
-    {
-      "epoch": 5.382543580659447,
-      "grad_norm": 0.0033097444102168083,
-      "learning_rate": 0.00019998571291870342,
-      "loss": 46.0,
-      "step": 33423
-    },
-    {
-      "epoch": 5.382704617738234,
-      "grad_norm": 0.00042688887333497405,
-      "learning_rate": 0.00019998571206353,
-      "loss": 46.0,
-      "step": 33424
-    },
-    {
-      "epoch": 5.382865654817022,
-      "grad_norm": 0.0016118204221129417,
-      "learning_rate": 0.000199985711208331,
-      "loss": 46.0,
-      "step": 33425
-    },
-    {
-      "epoch": 5.383026691895809,
-      "grad_norm": 0.0003310071479063481,
-      "learning_rate": 0.00019998571035310638,
-      "loss": 46.0,
-      "step": 33426
-    },
-    {
-      "epoch": 5.383187728974597,
-      "grad_norm": 0.0009101869654841721,
-      "learning_rate": 0.00019998570949785615,
-      "loss": 46.0,
-      "step": 33427
-    },
-    {
-      "epoch": 5.383348766053384,
-      "grad_norm": 0.0040672714821994305,
-      "learning_rate": 0.00019998570864258037,
-      "loss": 46.0,
-      "step": 33428
-    },
-    {
-      "epoch": 5.3835098031321715,
-      "grad_norm": 0.008558794856071472,
-      "learning_rate": 0.000199985707787279,
-      "loss": 46.0,
-      "step": 33429
-    },
-    {
-      "epoch": 5.383670840210959,
-      "grad_norm": 0.005282717291265726,
-      "learning_rate": 0.000199985706931952,
-      "loss": 46.0,
-      "step": 33430
-    },
-    {
-      "epoch": 5.3838318772897455,
-      "grad_norm": 0.01839902438223362,
-      "learning_rate": 0.00019998570607659947,
-      "loss": 46.0,
-      "step": 33431
-    },
-    {
-      "epoch": 5.383992914368533,
-      "grad_norm": 0.0026638892013579607,
-      "learning_rate": 0.00019998570522122128,
-      "loss": 46.0,
-      "step": 33432
-    },
-    {
-      "epoch": 5.3841539514473205,
-      "grad_norm": 0.002414526417851448,
-      "learning_rate": 0.00019998570436581753,
-      "loss": 46.0,
-      "step": 33433
-    },
-    {
-      "epoch": 5.384314988526108,
-      "grad_norm": 0.0034217049833387136,
-      "learning_rate": 0.0001999857035103882,
-      "loss": 46.0,
-      "step": 33434
-    },
-    {
-      "epoch": 5.384476025604895,
-      "grad_norm": 0.0033208851236850023,
-      "learning_rate": 0.00019998570265493328,
-      "loss": 46.0,
-      "step": 33435
-    },
-    {
-      "epoch": 5.384637062683683,
-      "grad_norm": 0.0019141914090141654,
-      "learning_rate": 0.00019998570179945277,
-      "loss": 46.0,
-      "step": 33436
-    },
-    {
-      "epoch": 5.38479809976247,
-      "grad_norm": 0.0010673184879124165,
-      "learning_rate": 0.00019998570094394665,
-      "loss": 46.0,
-      "step": 33437
-    },
-    {
-      "epoch": 5.384959136841258,
-      "grad_norm": 0.0030328824650496244,
-      "learning_rate": 0.00019998570008841494,
-      "loss": 46.0,
-      "step": 33438
-    },
-    {
-      "epoch": 5.385120173920045,
-      "grad_norm": 0.002872882643714547,
-      "learning_rate": 0.00019998569923285767,
-      "loss": 46.0,
-      "step": 33439
-    },
-    {
-      "epoch": 5.385281210998833,
-      "grad_norm": 0.005898167844861746,
-      "learning_rate": 0.00019998569837727476,
-      "loss": 46.0,
-      "step": 33440
-    },
-    {
-      "epoch": 5.38544224807762,
-      "grad_norm": 0.004220498725771904,
-      "learning_rate": 0.00019998569752166628,
-      "loss": 46.0,
-      "step": 33441
-    },
-    {
-      "epoch": 5.3856032851564075,
-      "grad_norm": 0.008398385718464851,
-      "learning_rate": 0.00019998569666603223,
-      "loss": 46.0,
-      "step": 33442
-    },
-    {
-      "epoch": 5.385764322235195,
-      "grad_norm": 0.0015079734148457646,
-      "learning_rate": 0.00019998569581037255,
-      "loss": 46.0,
-      "step": 33443
-    },
-    {
-      "epoch": 5.385925359313982,
-      "grad_norm": 0.00890716165304184,
-      "learning_rate": 0.00019998569495468732,
-      "loss": 46.0,
-      "step": 33444
-    },
-    {
-      "epoch": 5.38608639639277,
-      "grad_norm": 0.01060525979846716,
-      "learning_rate": 0.00019998569409897647,
-      "loss": 46.0,
-      "step": 33445
-    },
-    {
-      "epoch": 5.386247433471556,
-      "grad_norm": 0.0012454123934730887,
-      "learning_rate": 0.00019998569324324,
-      "loss": 46.0,
-      "step": 33446
-    },
-    {
-      "epoch": 5.386408470550344,
-      "grad_norm": 0.0034388345666229725,
-      "learning_rate": 0.00019998569238747802,
-      "loss": 46.0,
-      "step": 33447
-    },
-    {
-      "epoch": 5.386569507629131,
-      "grad_norm": 0.005685487762093544,
-      "learning_rate": 0.0001999856915316904,
-      "loss": 46.0,
-      "step": 33448
-    },
-    {
-      "epoch": 5.386730544707919,
-      "grad_norm": 0.004930159077048302,
-      "learning_rate": 0.00019998569067587719,
-      "loss": 46.0,
-      "step": 33449
-    },
-    {
-      "epoch": 5.386891581786706,
-      "grad_norm": 0.0255642831325531,
-      "learning_rate": 0.00019998568982003837,
-      "loss": 46.0,
-      "step": 33450
-    },
-    {
-      "epoch": 5.387052618865494,
-      "grad_norm": 0.003627521451562643,
-      "learning_rate": 0.00019998568896417398,
-      "loss": 46.0,
-      "step": 33451
-    },
-    {
-      "epoch": 5.387213655944281,
-      "grad_norm": 0.005542671773582697,
-      "learning_rate": 0.000199985688108284,
-      "loss": 46.0,
-      "step": 33452
-    },
-    {
-      "epoch": 5.387374693023069,
-      "grad_norm": 0.0017212245147675276,
-      "learning_rate": 0.00019998568725236845,
-      "loss": 46.0,
-      "step": 33453
-    },
-    {
-      "epoch": 5.387535730101856,
-      "grad_norm": 0.0063540819101035595,
-      "learning_rate": 0.00019998568639642726,
-      "loss": 46.0,
-      "step": 33454
-    },
-    {
-      "epoch": 5.3876967671806435,
-      "grad_norm": 0.004496101755648851,
-      "learning_rate": 0.00019998568554046054,
-      "loss": 46.0,
-      "step": 33455
-    },
-    {
-      "epoch": 5.387857804259431,
-      "grad_norm": 0.004551996476948261,
-      "learning_rate": 0.00019998568468446818,
-      "loss": 46.0,
-      "step": 33456
-    },
-    {
-      "epoch": 5.388018841338218,
-      "grad_norm": 0.0015107227955013514,
-      "learning_rate": 0.00019998568382845024,
-      "loss": 46.0,
-      "step": 33457
-    },
-    {
-      "epoch": 5.388179878417006,
-      "grad_norm": 0.004374251700937748,
-      "learning_rate": 0.00019998568297240673,
-      "loss": 46.0,
-      "step": 33458
-    },
-    {
-      "epoch": 5.388340915495793,
-      "grad_norm": 0.007724551949650049,
-      "learning_rate": 0.00019998568211633758,
-      "loss": 46.0,
-      "step": 33459
-    },
-    {
-      "epoch": 5.388501952574581,
-      "grad_norm": 0.008130398578941822,
-      "learning_rate": 0.0001999856812602429,
-      "loss": 46.0,
-      "step": 33460
-    },
-    {
-      "epoch": 5.388662989653367,
-      "grad_norm": 0.0009826701134443283,
-      "learning_rate": 0.00019998568040412257,
-      "loss": 46.0,
-      "step": 33461
-    },
-    {
-      "epoch": 5.388824026732155,
-      "grad_norm": 0.00276808044873178,
-      "learning_rate": 0.0001999856795479767,
-      "loss": 46.0,
-      "step": 33462
-    },
-    {
-      "epoch": 5.388985063810942,
-      "grad_norm": 0.007066166959702969,
-      "learning_rate": 0.0001999856786918052,
-      "loss": 46.0,
-      "step": 33463
-    },
-    {
-      "epoch": 5.38914610088973,
-      "grad_norm": 0.002974366070702672,
-      "learning_rate": 0.00019998567783560814,
-      "loss": 46.0,
-      "step": 33464
-    },
-    {
-      "epoch": 5.389307137968517,
-      "grad_norm": 0.0017695093993097544,
-      "learning_rate": 0.00019998567697938546,
-      "loss": 46.0,
-      "step": 33465
-    },
-    {
-      "epoch": 5.389468175047305,
-      "grad_norm": 0.0032727045472711325,
-      "learning_rate": 0.0001999856761231372,
-      "loss": 46.0,
-      "step": 33466
-    },
-    {
-      "epoch": 5.389629212126092,
-      "grad_norm": 0.002224060008302331,
-      "learning_rate": 0.00019998567526686336,
-      "loss": 46.0,
-      "step": 33467
-    },
-    {
-      "epoch": 5.3897902492048795,
-      "grad_norm": 0.009895727969706059,
-      "learning_rate": 0.00019998567441056392,
-      "loss": 46.0,
-      "step": 33468
-    },
-    {
-      "epoch": 5.389951286283667,
-      "grad_norm": 0.0024988381192088127,
-      "learning_rate": 0.00019998567355423888,
-      "loss": 46.0,
-      "step": 33469
-    },
-    {
-      "epoch": 5.390112323362454,
-      "grad_norm": 0.0025691697373986244,
-      "learning_rate": 0.00019998567269788827,
-      "loss": 46.0,
-      "step": 33470
-    },
-    {
-      "epoch": 5.390273360441242,
-      "grad_norm": 0.0020229837391525507,
-      "learning_rate": 0.00019998567184151205,
-      "loss": 46.0,
-      "step": 33471
-    },
-    {
-      "epoch": 5.390434397520029,
-      "grad_norm": 0.009341203607618809,
-      "learning_rate": 0.00019998567098511024,
-      "loss": 46.0,
-      "step": 33472
-    },
-    {
-      "epoch": 5.390595434598817,
-      "grad_norm": 0.021939948201179504,
-      "learning_rate": 0.00019998567012868287,
-      "loss": 46.0,
-      "step": 33473
-    },
-    {
-      "epoch": 5.390756471677603,
-      "grad_norm": 0.009470638819038868,
-      "learning_rate": 0.00019998566927222986,
-      "loss": 46.0,
-      "step": 33474
-    },
-    {
-      "epoch": 5.390917508756391,
-      "grad_norm": 0.007434885948896408,
-      "learning_rate": 0.0001999856684157513,
-      "loss": 46.0,
-      "step": 33475
-    },
-    {
-      "epoch": 5.391078545835178,
-      "grad_norm": 0.002830767072737217,
-      "learning_rate": 0.00019998566755924713,
-      "loss": 46.0,
-      "step": 33476
-    },
-    {
-      "epoch": 5.391239582913966,
-      "grad_norm": 0.0023723661433905363,
-      "learning_rate": 0.00019998566670271736,
-      "loss": 46.0,
-      "step": 33477
-    },
-    {
-      "epoch": 5.391400619992753,
-      "grad_norm": 0.0024492102675139904,
-      "learning_rate": 0.000199985665846162,
-      "loss": 46.0,
-      "step": 33478
-    },
-    {
-      "epoch": 5.391561657071541,
-      "grad_norm": 0.0027372061740607023,
-      "learning_rate": 0.00019998566498958105,
-      "loss": 46.0,
-      "step": 33479
-    },
-    {
-      "epoch": 5.391722694150328,
-      "grad_norm": 0.008980734273791313,
-      "learning_rate": 0.00019998566413297452,
-      "loss": 46.0,
-      "step": 33480
-    },
-    {
-      "epoch": 5.3918837312291155,
-      "grad_norm": 0.0062769996002316475,
-      "learning_rate": 0.0001999856632763424,
-      "loss": 46.0,
-      "step": 33481
-    },
-    {
-      "epoch": 5.392044768307903,
-      "grad_norm": 0.0015832630451768637,
-      "learning_rate": 0.0001999856624196847,
-      "loss": 46.0,
-      "step": 33482
-    },
-    {
-      "epoch": 5.39220580538669,
-      "grad_norm": 0.01112749520689249,
-      "learning_rate": 0.00019998566156300137,
-      "loss": 46.0,
-      "step": 33483
-    },
-    {
-      "epoch": 5.392366842465478,
-      "grad_norm": 0.0016363660106435418,
-      "learning_rate": 0.00019998566070629248,
-      "loss": 46.0,
-      "step": 33484
-    },
-    {
-      "epoch": 5.392527879544265,
-      "grad_norm": 0.0028556224424391985,
-      "learning_rate": 0.00019998565984955796,
-      "loss": 46.0,
-      "step": 33485
-    },
-    {
-      "epoch": 5.392688916623053,
-      "grad_norm": 0.0037885401397943497,
-      "learning_rate": 0.00019998565899279788,
-      "loss": 46.0,
-      "step": 33486
-    },
-    {
-      "epoch": 5.39284995370184,
-      "grad_norm": 0.012513256631791592,
-      "learning_rate": 0.00019998565813601223,
-      "loss": 46.0,
-      "step": 33487
-    },
-    {
-      "epoch": 5.393010990780628,
-      "grad_norm": 0.005354071035981178,
-      "learning_rate": 0.00019998565727920097,
-      "loss": 46.0,
-      "step": 33488
-    },
-    {
-      "epoch": 5.393172027859414,
-      "grad_norm": 0.00628530140966177,
-      "learning_rate": 0.0001999856564223641,
-      "loss": 46.0,
-      "step": 33489
-    },
-    {
-      "epoch": 5.393333064938202,
-      "grad_norm": 0.007867788895964622,
-      "learning_rate": 0.00019998565556550167,
-      "loss": 46.0,
-      "step": 33490
-    },
-    {
-      "epoch": 5.393494102016989,
-      "grad_norm": 0.002536516636610031,
-      "learning_rate": 0.00019998565470861362,
-      "loss": 46.0,
-      "step": 33491
-    },
-    {
-      "epoch": 5.3936551390957765,
-      "grad_norm": 0.0021895705722272396,
-      "learning_rate": 0.0001999856538517,
-      "loss": 46.0,
-      "step": 33492
-    },
-    {
-      "epoch": 5.393816176174564,
-      "grad_norm": 0.008843330666422844,
-      "learning_rate": 0.00019998565299476077,
-      "loss": 46.0,
-      "step": 33493
-    },
-    {
-      "epoch": 5.393977213253351,
-      "grad_norm": 0.001473539276048541,
-      "learning_rate": 0.00019998565213779596,
-      "loss": 46.0,
-      "step": 33494
-    },
-    {
-      "epoch": 5.394138250332139,
-      "grad_norm": 0.0046297828666865826,
-      "learning_rate": 0.00019998565128080556,
-      "loss": 46.0,
-      "step": 33495
-    },
-    {
-      "epoch": 5.394299287410926,
-      "grad_norm": 0.003924149554222822,
-      "learning_rate": 0.00019998565042378958,
-      "loss": 46.0,
-      "step": 33496
-    },
-    {
-      "epoch": 5.394460324489714,
-      "grad_norm": 0.01227219682186842,
-      "learning_rate": 0.00019998564956674798,
-      "loss": 46.0,
-      "step": 33497
-    },
-    {
-      "epoch": 5.394621361568501,
-      "grad_norm": 0.004981328267604113,
-      "learning_rate": 0.00019998564870968083,
-      "loss": 46.0,
-      "step": 33498
-    },
-    {
-      "epoch": 5.394782398647289,
-      "grad_norm": 0.0041697281412780285,
-      "learning_rate": 0.00019998564785258805,
-      "loss": 46.0,
-      "step": 33499
-    },
-    {
-      "epoch": 5.394943435726076,
-      "grad_norm": 0.0061863684095442295,
-      "learning_rate": 0.0001999856469954697,
-      "loss": 46.0,
-      "step": 33500
-    },
-    {
-      "epoch": 5.395104472804864,
-      "grad_norm": 0.007273610215634108,
-      "learning_rate": 0.00019998564613832572,
-      "loss": 46.0,
-      "step": 33501
-    },
-    {
-      "epoch": 5.395265509883651,
-      "grad_norm": 0.006181139498949051,
-      "learning_rate": 0.0001999856452811562,
-      "loss": 46.0,
-      "step": 33502
-    },
-    {
-      "epoch": 5.3954265469624385,
-      "grad_norm": 0.0028360106516629457,
-      "learning_rate": 0.00019998564442396107,
-      "loss": 46.0,
-      "step": 33503
-    },
-    {
-      "epoch": 5.395587584041225,
-      "grad_norm": 0.010671830736100674,
-      "learning_rate": 0.00019998564356674034,
-      "loss": 46.0,
-      "step": 33504
-    },
-    {
-      "epoch": 5.3957486211200125,
-      "grad_norm": 0.007207934278994799,
-      "learning_rate": 0.00019998564270949401,
-      "loss": 46.0,
-      "step": 33505
-    },
-    {
-      "epoch": 5.3959096581988,
-      "grad_norm": 0.00286873709410429,
-      "learning_rate": 0.00019998564185222213,
-      "loss": 46.0,
-      "step": 33506
-    },
-    {
-      "epoch": 5.396070695277587,
-      "grad_norm": 0.00712064141407609,
-      "learning_rate": 0.0001999856409949246,
-      "loss": 46.0,
-      "step": 33507
-    },
-    {
-      "epoch": 5.396231732356375,
-      "grad_norm": 0.009728697128593922,
-      "learning_rate": 0.00019998564013760155,
-      "loss": 46.0,
-      "step": 33508
-    },
-    {
-      "epoch": 5.396392769435162,
-      "grad_norm": 0.001330304890871048,
-      "learning_rate": 0.00019998563928025286,
-      "loss": 46.0,
-      "step": 33509
-    },
-    {
-      "epoch": 5.39655380651395,
-      "grad_norm": 0.0044160704128444195,
-      "learning_rate": 0.00019998563842287857,
-      "loss": 46.0,
-      "step": 33510
-    },
-    {
-      "epoch": 5.396714843592737,
-      "grad_norm": 0.006356305442750454,
-      "learning_rate": 0.00019998563756547873,
-      "loss": 46.0,
-      "step": 33511
-    },
-    {
-      "epoch": 5.396875880671525,
-      "grad_norm": 0.0037580919452011585,
-      "learning_rate": 0.00019998563670805327,
-      "loss": 46.0,
-      "step": 33512
-    },
-    {
-      "epoch": 5.397036917750312,
-      "grad_norm": 0.016473636031150818,
-      "learning_rate": 0.00019998563585060222,
-      "loss": 46.0,
-      "step": 33513
-    },
-    {
-      "epoch": 5.3971979548291,
-      "grad_norm": 0.001658030552789569,
-      "learning_rate": 0.0001999856349931256,
-      "loss": 46.0,
-      "step": 33514
-    },
-    {
-      "epoch": 5.397358991907887,
-      "grad_norm": 0.005189374089241028,
-      "learning_rate": 0.00019998563413562337,
-      "loss": 46.0,
-      "step": 33515
-    },
-    {
-      "epoch": 5.3975200289866745,
-      "grad_norm": 0.005965104792267084,
-      "learning_rate": 0.00019998563327809556,
-      "loss": 46.0,
-      "step": 33516
-    },
-    {
-      "epoch": 5.397681066065462,
-      "grad_norm": 0.0006654943572357297,
-      "learning_rate": 0.00019998563242054214,
-      "loss": 46.0,
-      "step": 33517
-    },
-    {
-      "epoch": 5.397842103144249,
-      "grad_norm": 0.004483413882553577,
-      "learning_rate": 0.00019998563156296313,
-      "loss": 46.0,
-      "step": 33518
-    },
-    {
-      "epoch": 5.398003140223036,
-      "grad_norm": 0.0023640641011297703,
-      "learning_rate": 0.00019998563070535856,
-      "loss": 46.0,
-      "step": 33519
-    },
-    {
-      "epoch": 5.398164177301823,
-      "grad_norm": 0.007390114013105631,
-      "learning_rate": 0.00019998562984772838,
-      "loss": 46.0,
-      "step": 33520
-    },
-    {
-      "epoch": 5.398325214380611,
-      "grad_norm": 0.0042753578163683414,
-      "learning_rate": 0.0001999856289900726,
-      "loss": 46.0,
-      "step": 33521
-    },
-    {
-      "epoch": 5.398486251459398,
-      "grad_norm": 0.005977501627057791,
-      "learning_rate": 0.00019998562813239122,
-      "loss": 46.0,
-      "step": 33522
-    },
-    {
-      "epoch": 5.398647288538186,
-      "grad_norm": 0.00928088091313839,
-      "learning_rate": 0.00019998562727468428,
-      "loss": 46.0,
-      "step": 33523
-    },
-    {
-      "epoch": 5.398808325616973,
-      "grad_norm": 0.01390144880861044,
-      "learning_rate": 0.00019998562641695172,
-      "loss": 46.0,
-      "step": 33524
-    },
-    {
-      "epoch": 5.398969362695761,
-      "grad_norm": 0.005800340324640274,
-      "learning_rate": 0.0001999856255591936,
-      "loss": 46.0,
-      "step": 33525
-    },
-    {
-      "epoch": 5.399130399774548,
-      "grad_norm": 0.012404757551848888,
-      "learning_rate": 0.00019998562470140984,
-      "loss": 46.0,
-      "step": 33526
-    },
-    {
-      "epoch": 5.399291436853336,
-      "grad_norm": 0.002577253384515643,
-      "learning_rate": 0.00019998562384360052,
-      "loss": 46.0,
-      "step": 33527
-    },
-    {
-      "epoch": 5.399452473932123,
-      "grad_norm": 0.002280876040458679,
-      "learning_rate": 0.00019998562298576564,
-      "loss": 46.0,
-      "step": 33528
-    },
-    {
-      "epoch": 5.3996135110109105,
-      "grad_norm": 0.003409678814932704,
-      "learning_rate": 0.00019998562212790512,
-      "loss": 46.0,
-      "step": 33529
-    },
-    {
-      "epoch": 5.399774548089698,
-      "grad_norm": 0.0028816857375204563,
-      "learning_rate": 0.00019998562127001904,
-      "loss": 46.0,
-      "step": 33530
-    },
-    {
-      "epoch": 5.399935585168485,
-      "grad_norm": 0.014008548110723495,
-      "learning_rate": 0.00019998562041210734,
-      "loss": 46.0,
-      "step": 33531
-    },
-    {
-      "epoch": 5.400096622247273,
-      "grad_norm": 0.008499814197421074,
-      "learning_rate": 0.00019998561955417006,
-      "loss": 46.0,
-      "step": 33532
-    },
-    {
-      "epoch": 5.40025765932606,
-      "grad_norm": 0.002058165380731225,
-      "learning_rate": 0.0001999856186962072,
-      "loss": 46.0,
-      "step": 33533
-    },
-    {
-      "epoch": 5.400418696404847,
-      "grad_norm": 0.0042491573840379715,
-      "learning_rate": 0.00019998561783821875,
-      "loss": 46.0,
-      "step": 33534
-    },
-    {
-      "epoch": 5.400579733483634,
-      "grad_norm": 0.005727290641516447,
-      "learning_rate": 0.0001999856169802047,
-      "loss": 46.0,
-      "step": 33535
-    },
-    {
-      "epoch": 5.400740770562422,
-      "grad_norm": 0.001044887350872159,
-      "learning_rate": 0.00019998561612216508,
-      "loss": 46.0,
-      "step": 33536
-    },
-    {
-      "epoch": 5.400901807641209,
-      "grad_norm": 0.00909384060651064,
-      "learning_rate": 0.0001999856152640998,
-      "loss": 46.0,
-      "step": 33537
-    },
-    {
-      "epoch": 5.401062844719997,
-      "grad_norm": 0.013593494892120361,
-      "learning_rate": 0.000199985614406009,
-      "loss": 46.0,
-      "step": 33538
-    },
-    {
-      "epoch": 5.401223881798784,
-      "grad_norm": 0.002498663729056716,
-      "learning_rate": 0.00019998561354789258,
-      "loss": 46.0,
-      "step": 33539
-    },
-    {
-      "epoch": 5.4013849188775716,
-      "grad_norm": 0.005663787480443716,
-      "learning_rate": 0.00019998561268975057,
-      "loss": 46.0,
-      "step": 33540
-    },
-    {
-      "epoch": 5.401545955956359,
-      "grad_norm": 0.0018130455864593387,
-      "learning_rate": 0.00019998561183158297,
-      "loss": 46.0,
-      "step": 33541
-    },
-    {
-      "epoch": 5.4017069930351465,
-      "grad_norm": 0.010311313904821873,
-      "learning_rate": 0.0001999856109733898,
-      "loss": 46.0,
-      "step": 33542
-    },
-    {
-      "epoch": 5.401868030113934,
-      "grad_norm": 0.0009938409784808755,
-      "learning_rate": 0.000199985610115171,
-      "loss": 46.0,
-      "step": 33543
-    },
-    {
-      "epoch": 5.402029067192721,
-      "grad_norm": 0.009713970124721527,
-      "learning_rate": 0.00019998560925692664,
-      "loss": 46.0,
-      "step": 33544
-    },
-    {
-      "epoch": 5.402190104271509,
-      "grad_norm": 0.01077380683273077,
-      "learning_rate": 0.00019998560839865667,
-      "loss": 46.0,
-      "step": 33545
-    },
-    {
-      "epoch": 5.402351141350296,
-      "grad_norm": 0.020101845264434814,
-      "learning_rate": 0.00019998560754036114,
-      "loss": 46.0,
-      "step": 33546
-    },
-    {
-      "epoch": 5.402512178429084,
-      "grad_norm": 0.008321388624608517,
-      "learning_rate": 0.00019998560668204,
-      "loss": 46.0,
-      "step": 33547
-    },
-    {
-      "epoch": 5.40267321550787,
-      "grad_norm": 0.0015650109853595495,
-      "learning_rate": 0.00019998560582369326,
-      "loss": 46.0,
-      "step": 33548
-    },
-    {
-      "epoch": 5.402834252586658,
-      "grad_norm": 0.0008906018920242786,
-      "learning_rate": 0.0001999856049653209,
-      "loss": 46.0,
-      "step": 33549
-    },
-    {
-      "epoch": 5.402995289665445,
-      "grad_norm": 0.0027846964076161385,
-      "learning_rate": 0.000199985604106923,
-      "loss": 46.0,
-      "step": 33550
-    },
-    {
-      "epoch": 5.403156326744233,
-      "grad_norm": 0.003964104223996401,
-      "learning_rate": 0.00019998560324849948,
-      "loss": 46.0,
-      "step": 33551
-    },
-    {
-      "epoch": 5.40331736382302,
-      "grad_norm": 0.005119128152728081,
-      "learning_rate": 0.00019998560239005038,
-      "loss": 46.0,
-      "step": 33552
-    },
-    {
-      "epoch": 5.4034784009018075,
-      "grad_norm": 0.0012736073695123196,
-      "learning_rate": 0.0001999856015315757,
-      "loss": 46.0,
-      "step": 33553
-    },
-    {
-      "epoch": 5.403639437980595,
-      "grad_norm": 0.020718751475214958,
-      "learning_rate": 0.00019998560067307542,
-      "loss": 46.0,
-      "step": 33554
-    },
-    {
-      "epoch": 5.403800475059382,
-      "grad_norm": 0.014644449576735497,
-      "learning_rate": 0.00019998559981454953,
-      "loss": 46.0,
-      "step": 33555
-    },
-    {
-      "epoch": 5.40396151213817,
-      "grad_norm": 0.023552147671580315,
-      "learning_rate": 0.0001999855989559981,
-      "loss": 46.0,
-      "step": 33556
-    },
-    {
-      "epoch": 5.404122549216957,
-      "grad_norm": 0.0076517341658473015,
-      "learning_rate": 0.000199985598097421,
-      "loss": 46.0,
-      "step": 33557
-    },
-    {
-      "epoch": 5.404283586295745,
-      "grad_norm": 0.0034961772616952658,
-      "learning_rate": 0.00019998559723881837,
-      "loss": 46.0,
-      "step": 33558
-    },
-    {
-      "epoch": 5.404444623374532,
-      "grad_norm": 0.0019426201470196247,
-      "learning_rate": 0.00019998559638019012,
-      "loss": 46.0,
-      "step": 33559
-    },
-    {
-      "epoch": 5.40460566045332,
-      "grad_norm": 0.003406703472137451,
-      "learning_rate": 0.00019998559552153632,
-      "loss": 46.0,
-      "step": 33560
-    },
-    {
-      "epoch": 5.404766697532107,
-      "grad_norm": 0.0010419911704957485,
-      "learning_rate": 0.00019998559466285687,
-      "loss": 46.0,
-      "step": 33561
-    },
-    {
-      "epoch": 5.404927734610894,
-      "grad_norm": 0.00535344984382391,
-      "learning_rate": 0.00019998559380415186,
-      "loss": 46.0,
-      "step": 33562
-    },
-    {
-      "epoch": 5.405088771689681,
-      "grad_norm": 0.0015867863548919559,
-      "learning_rate": 0.00019998559294542124,
-      "loss": 46.0,
-      "step": 33563
-    },
-    {
-      "epoch": 5.405249808768469,
-      "grad_norm": 0.0038380767218768597,
-      "learning_rate": 0.0001999855920866651,
-      "loss": 46.0,
-      "step": 33564
-    },
-    {
-      "epoch": 5.405410845847256,
-      "grad_norm": 0.004812820348888636,
-      "learning_rate": 0.00019998559122788326,
-      "loss": 46.0,
-      "step": 33565
-    },
-    {
-      "epoch": 5.4055718829260435,
-      "grad_norm": 0.0038598033133894205,
-      "learning_rate": 0.0001999855903690759,
-      "loss": 46.0,
-      "step": 33566
-    },
-    {
-      "epoch": 5.405732920004831,
-      "grad_norm": 0.010577513836324215,
-      "learning_rate": 0.00019998558951024294,
-      "loss": 46.0,
-      "step": 33567
-    },
-    {
-      "epoch": 5.405893957083618,
-      "grad_norm": 0.006066112779080868,
-      "learning_rate": 0.00019998558865138438,
-      "loss": 46.0,
-      "step": 33568
-    },
-    {
-      "epoch": 5.406054994162406,
-      "grad_norm": 0.006049091927707195,
-      "learning_rate": 0.00019998558779250024,
-      "loss": 46.0,
-      "step": 33569
-    },
-    {
-      "epoch": 5.406216031241193,
-      "grad_norm": 0.009319797158241272,
-      "learning_rate": 0.00019998558693359048,
-      "loss": 46.0,
-      "step": 33570
-    },
-    {
-      "epoch": 5.406377068319981,
-      "grad_norm": 0.004520068876445293,
-      "learning_rate": 0.00019998558607465516,
-      "loss": 46.0,
-      "step": 33571
-    },
-    {
-      "epoch": 5.406538105398768,
-      "grad_norm": 0.014994814060628414,
-      "learning_rate": 0.00019998558521569423,
-      "loss": 46.0,
-      "step": 33572
-    },
-    {
-      "epoch": 5.406699142477556,
-      "grad_norm": 0.008603583090007305,
-      "learning_rate": 0.00019998558435670773,
-      "loss": 46.0,
-      "step": 33573
-    },
-    {
-      "epoch": 5.406860179556343,
-      "grad_norm": 0.004327674861997366,
-      "learning_rate": 0.00019998558349769563,
-      "loss": 46.0,
-      "step": 33574
-    },
-    {
-      "epoch": 5.407021216635131,
-      "grad_norm": 0.010475815273821354,
-      "learning_rate": 0.00019998558263865793,
-      "loss": 46.0,
-      "step": 33575
-    },
-    {
-      "epoch": 5.407182253713918,
-      "grad_norm": 0.002505118492990732,
-      "learning_rate": 0.00019998558177959462,
-      "loss": 46.0,
-      "step": 33576
-    },
-    {
-      "epoch": 5.407343290792705,
-      "grad_norm": 0.006551137659698725,
-      "learning_rate": 0.00019998558092050575,
-      "loss": 46.0,
-      "step": 33577
-    },
-    {
-      "epoch": 5.407504327871492,
-      "grad_norm": 0.012180093675851822,
-      "learning_rate": 0.00019998558006139127,
-      "loss": 46.0,
-      "step": 33578
-    },
-    {
-      "epoch": 5.4076653649502795,
-      "grad_norm": 0.003010115586221218,
-      "learning_rate": 0.00019998557920225123,
-      "loss": 46.0,
-      "step": 33579
-    },
-    {
-      "epoch": 5.407826402029067,
-      "grad_norm": 0.011785740032792091,
-      "learning_rate": 0.00019998557834308554,
-      "loss": 46.0,
-      "step": 33580
-    },
-    {
-      "epoch": 5.407987439107854,
-      "grad_norm": 0.014025033451616764,
-      "learning_rate": 0.0001999855774838943,
-      "loss": 46.0,
-      "step": 33581
-    },
-    {
-      "epoch": 5.408148476186642,
-      "grad_norm": 0.002184830605983734,
-      "learning_rate": 0.00019998557662467747,
-      "loss": 46.0,
-      "step": 33582
-    },
-    {
-      "epoch": 5.408309513265429,
-      "grad_norm": 0.00212856475263834,
-      "learning_rate": 0.00019998557576543508,
-      "loss": 46.0,
-      "step": 33583
-    },
-    {
-      "epoch": 5.408470550344217,
-      "grad_norm": 0.005818785633891821,
-      "learning_rate": 0.00019998557490616701,
-      "loss": 46.0,
-      "step": 33584
-    },
-    {
-      "epoch": 5.408631587423004,
-      "grad_norm": 0.008384395390748978,
-      "learning_rate": 0.00019998557404687342,
-      "loss": 46.0,
-      "step": 33585
-    },
-    {
-      "epoch": 5.408792624501792,
-      "grad_norm": 0.004590184893459082,
-      "learning_rate": 0.00019998557318755421,
-      "loss": 46.0,
-      "step": 33586
-    },
-    {
-      "epoch": 5.408953661580579,
-      "grad_norm": 0.0016751501243561506,
-      "learning_rate": 0.00019998557232820945,
-      "loss": 46.0,
-      "step": 33587
-    },
-    {
-      "epoch": 5.409114698659367,
-      "grad_norm": 0.002438553376123309,
-      "learning_rate": 0.00019998557146883904,
-      "loss": 46.0,
-      "step": 33588
-    },
-    {
-      "epoch": 5.409275735738154,
-      "grad_norm": 0.0023840717040002346,
-      "learning_rate": 0.00019998557060944307,
-      "loss": 46.0,
-      "step": 33589
-    },
-    {
-      "epoch": 5.4094367728169415,
-      "grad_norm": 0.0013377969153225422,
-      "learning_rate": 0.0001999855697500215,
-      "loss": 46.0,
-      "step": 33590
-    },
-    {
-      "epoch": 5.409597809895729,
-      "grad_norm": 0.0029756140429526567,
-      "learning_rate": 0.00019998556889057437,
-      "loss": 46.0,
-      "step": 33591
-    },
-    {
-      "epoch": 5.4097588469745155,
-      "grad_norm": 0.0028415340930223465,
-      "learning_rate": 0.0001999855680311016,
-      "loss": 46.0,
-      "step": 33592
-    },
-    {
-      "epoch": 5.409919884053303,
-      "grad_norm": 0.005608447827398777,
-      "learning_rate": 0.0001999855671716033,
-      "loss": 46.0,
-      "step": 33593
-    },
-    {
-      "epoch": 5.41008092113209,
-      "grad_norm": 0.005532634910196066,
-      "learning_rate": 0.00019998556631207933,
-      "loss": 46.0,
-      "step": 33594
-    },
-    {
-      "epoch": 5.410241958210878,
-      "grad_norm": 0.0065854270942509174,
-      "learning_rate": 0.00019998556545252981,
-      "loss": 46.0,
-      "step": 33595
-    },
-    {
-      "epoch": 5.410402995289665,
-      "grad_norm": 0.00596965104341507,
-      "learning_rate": 0.0001999855645929547,
-      "loss": 46.0,
-      "step": 33596
-    },
-    {
-      "epoch": 5.410564032368453,
-      "grad_norm": 0.01035352237522602,
-      "learning_rate": 0.000199985563733354,
-      "loss": 46.0,
-      "step": 33597
-    },
-    {
-      "epoch": 5.41072506944724,
-      "grad_norm": 0.006125032436102629,
-      "learning_rate": 0.0001999855628737277,
-      "loss": 46.0,
-      "step": 33598
-    },
-    {
-      "epoch": 5.410886106526028,
-      "grad_norm": 0.0020886987913399935,
-      "learning_rate": 0.0001999855620140758,
-      "loss": 46.0,
-      "step": 33599
-    },
-    {
-      "epoch": 5.411047143604815,
-      "grad_norm": 0.012737304903566837,
-      "learning_rate": 0.00019998556115439836,
-      "loss": 46.0,
-      "step": 33600
-    },
-    {
-      "epoch": 5.4112081806836025,
-      "grad_norm": 0.004250716883689165,
-      "learning_rate": 0.0001999855602946953,
-      "loss": 46.0,
-      "step": 33601
-    },
-    {
-      "epoch": 5.41136921776239,
-      "grad_norm": 0.006766407750546932,
-      "learning_rate": 0.00019998555943496663,
-      "loss": 46.0,
-      "step": 33602
-    },
-    {
-      "epoch": 5.4115302548411774,
-      "grad_norm": 0.024289848282933235,
-      "learning_rate": 0.00019998555857521236,
-      "loss": 46.0,
-      "step": 33603
-    },
-    {
-      "epoch": 5.411691291919965,
-      "grad_norm": 0.0014451007591560483,
-      "learning_rate": 0.00019998555771543256,
-      "loss": 46.0,
-      "step": 33604
-    },
-    {
-      "epoch": 5.411852328998752,
-      "grad_norm": 0.008498722687363625,
-      "learning_rate": 0.0001999855568556271,
-      "loss": 46.0,
-      "step": 33605
-    },
-    {
-      "epoch": 5.41201336607754,
-      "grad_norm": 0.006658232305198908,
-      "learning_rate": 0.00019998555599579608,
-      "loss": 46.0,
-      "step": 33606
-    },
-    {
-      "epoch": 5.412174403156326,
-      "grad_norm": 0.010507320985198021,
-      "learning_rate": 0.00019998555513593946,
-      "loss": 46.0,
-      "step": 33607
-    },
-    {
-      "epoch": 5.412335440235114,
-      "grad_norm": 0.0025305356830358505,
-      "learning_rate": 0.00019998555427605725,
-      "loss": 46.0,
-      "step": 33608
-    },
-    {
-      "epoch": 5.412496477313901,
-      "grad_norm": 0.002887978684157133,
-      "learning_rate": 0.00019998555341614943,
-      "loss": 46.0,
-      "step": 33609
-    },
-    {
-      "epoch": 5.412657514392689,
-      "grad_norm": 0.0008898144005797803,
-      "learning_rate": 0.00019998555255621608,
-      "loss": 46.0,
-      "step": 33610
-    },
-    {
-      "epoch": 5.412818551471476,
-      "grad_norm": 0.007634243927896023,
-      "learning_rate": 0.00019998555169625708,
-      "loss": 46.0,
-      "step": 33611
-    },
-    {
-      "epoch": 5.412979588550264,
-      "grad_norm": 0.011333703994750977,
-      "learning_rate": 0.0001999855508362725,
-      "loss": 46.0,
-      "step": 33612
-    },
-    {
-      "epoch": 5.413140625629051,
-      "grad_norm": 0.011937910690903664,
-      "learning_rate": 0.00019998554997626236,
-      "loss": 46.0,
-      "step": 33613
-    },
-    {
-      "epoch": 5.4133016627078385,
-      "grad_norm": 0.005609564483165741,
-      "learning_rate": 0.0001999855491162266,
-      "loss": 46.0,
-      "step": 33614
-    },
-    {
-      "epoch": 5.413462699786626,
-      "grad_norm": 0.008082227781414986,
-      "learning_rate": 0.00019998554825616526,
-      "loss": 46.0,
-      "step": 33615
-    },
-    {
-      "epoch": 5.413623736865413,
-      "grad_norm": 0.0019346611807122827,
-      "learning_rate": 0.00019998554739607833,
-      "loss": 46.0,
-      "step": 33616
-    },
-    {
-      "epoch": 5.413784773944201,
-      "grad_norm": 0.009286635555326939,
-      "learning_rate": 0.0001999855465359658,
-      "loss": 46.0,
-      "step": 33617
-    },
-    {
-      "epoch": 5.413945811022988,
-      "grad_norm": 0.017126353457570076,
-      "learning_rate": 0.00019998554567582767,
-      "loss": 46.0,
-      "step": 33618
-    },
-    {
-      "epoch": 5.414106848101776,
-      "grad_norm": 0.004990117158740759,
-      "learning_rate": 0.00019998554481566395,
-      "loss": 46.0,
-      "step": 33619
-    },
-    {
-      "epoch": 5.414267885180563,
-      "grad_norm": 0.0036235779989510775,
-      "learning_rate": 0.00019998554395547467,
-      "loss": 46.0,
-      "step": 33620
-    },
-    {
-      "epoch": 5.414428922259351,
-      "grad_norm": 0.0021115224808454514,
-      "learning_rate": 0.00019998554309525978,
-      "loss": 46.0,
-      "step": 33621
-    },
-    {
-      "epoch": 5.414589959338137,
-      "grad_norm": 0.005655521061271429,
-      "learning_rate": 0.00019998554223501927,
-      "loss": 46.0,
-      "step": 33622
-    },
-    {
-      "epoch": 5.414750996416925,
-      "grad_norm": 0.017471928149461746,
-      "learning_rate": 0.0001999855413747532,
-      "loss": 46.0,
-      "step": 33623
-    },
-    {
-      "epoch": 5.414912033495712,
-      "grad_norm": 0.005181267857551575,
-      "learning_rate": 0.00019998554051446155,
-      "loss": 46.0,
-      "step": 33624
-    },
-    {
-      "epoch": 5.4150730705745,
-      "grad_norm": 0.0059936451725661755,
-      "learning_rate": 0.0001999855396541443,
-      "loss": 46.0,
-      "step": 33625
-    },
-    {
-      "epoch": 5.415234107653287,
-      "grad_norm": 0.005156735423952341,
-      "learning_rate": 0.00019998553879380145,
-      "loss": 46.0,
-      "step": 33626
-    },
-    {
-      "epoch": 5.4153951447320745,
-      "grad_norm": 0.006914381403476,
-      "learning_rate": 0.000199985537933433,
-      "loss": 46.0,
-      "step": 33627
-    },
-    {
-      "epoch": 5.415556181810862,
-      "grad_norm": 0.0009555213036946952,
-      "learning_rate": 0.00019998553707303897,
-      "loss": 46.0,
-      "step": 33628
-    },
-    {
-      "epoch": 5.415717218889649,
-      "grad_norm": 0.0022707798052579165,
-      "learning_rate": 0.00019998553621261938,
-      "loss": 46.0,
-      "step": 33629
-    },
-    {
-      "epoch": 5.415878255968437,
-      "grad_norm": 0.005456534679979086,
-      "learning_rate": 0.00019998553535217415,
-      "loss": 46.0,
-      "step": 33630
-    },
-    {
-      "epoch": 5.416039293047224,
-      "grad_norm": 0.005757391918450594,
-      "learning_rate": 0.00019998553449170333,
-      "loss": 46.0,
-      "step": 33631
-    },
-    {
-      "epoch": 5.416200330126012,
-      "grad_norm": 0.003124311799183488,
-      "learning_rate": 0.00019998553363120695,
-      "loss": 46.0,
-      "step": 33632
-    },
-    {
-      "epoch": 5.416361367204799,
-      "grad_norm": 0.005804627668112516,
-      "learning_rate": 0.00019998553277068498,
-      "loss": 46.0,
-      "step": 33633
-    },
-    {
-      "epoch": 5.416522404283587,
-      "grad_norm": 0.017441198229789734,
-      "learning_rate": 0.0001999855319101374,
-      "loss": 46.0,
-      "step": 33634
-    },
-    {
-      "epoch": 5.416683441362373,
-      "grad_norm": 0.0024849544279277325,
-      "learning_rate": 0.00019998553104956423,
-      "loss": 46.0,
-      "step": 33635
-    },
-    {
-      "epoch": 5.416844478441161,
-      "grad_norm": 0.00952769722789526,
-      "learning_rate": 0.00019998553018896547,
-      "loss": 46.0,
-      "step": 33636
-    },
-    {
-      "epoch": 5.417005515519948,
-      "grad_norm": 0.0021035310346633196,
-      "learning_rate": 0.0001999855293283411,
-      "loss": 46.0,
-      "step": 33637
-    },
-    {
-      "epoch": 5.417166552598736,
-      "grad_norm": 0.006762698292732239,
-      "learning_rate": 0.00019998552846769117,
-      "loss": 46.0,
-      "step": 33638
-    },
-    {
-      "epoch": 5.417327589677523,
-      "grad_norm": 0.002041331259533763,
-      "learning_rate": 0.00019998552760701563,
-      "loss": 46.0,
-      "step": 33639
-    },
-    {
-      "epoch": 5.4174886267563105,
-      "grad_norm": 0.00481139263138175,
-      "learning_rate": 0.00019998552674631453,
-      "loss": 46.0,
-      "step": 33640
-    },
-    {
-      "epoch": 5.417649663835098,
-      "grad_norm": 0.011447397992014885,
-      "learning_rate": 0.00019998552588558778,
-      "loss": 46.0,
-      "step": 33641
-    },
-    {
-      "epoch": 5.417810700913885,
-      "grad_norm": 0.0020422530360519886,
-      "learning_rate": 0.0001999855250248355,
-      "loss": 46.0,
-      "step": 33642
-    },
-    {
-      "epoch": 5.417971737992673,
-      "grad_norm": 0.013513357378542423,
-      "learning_rate": 0.0001999855241640576,
-      "loss": 46.0,
-      "step": 33643
-    },
-    {
-      "epoch": 5.41813277507146,
-      "grad_norm": 0.0046821534633636475,
-      "learning_rate": 0.0001999855233032541,
-      "loss": 46.0,
-      "step": 33644
-    },
-    {
-      "epoch": 5.418293812150248,
-      "grad_norm": 0.01117374375462532,
-      "learning_rate": 0.00019998552244242503,
-      "loss": 46.0,
-      "step": 33645
-    },
-    {
-      "epoch": 5.418454849229035,
-      "grad_norm": 0.008008531294763088,
-      "learning_rate": 0.00019998552158157038,
-      "loss": 46.0,
-      "step": 33646
-    },
-    {
-      "epoch": 5.418615886307823,
-      "grad_norm": 0.003920272924005985,
-      "learning_rate": 0.0001999855207206901,
-      "loss": 46.0,
-      "step": 33647
-    },
-    {
-      "epoch": 5.41877692338661,
-      "grad_norm": 0.019243530929088593,
-      "learning_rate": 0.00019998551985978425,
-      "loss": 46.0,
-      "step": 33648
-    },
-    {
-      "epoch": 5.4189379604653976,
-      "grad_norm": 0.006622105371206999,
-      "learning_rate": 0.0001999855189988528,
-      "loss": 46.0,
-      "step": 33649
-    },
-    {
-      "epoch": 5.419098997544184,
-      "grad_norm": 0.0025609456934034824,
-      "learning_rate": 0.00019998551813789578,
-      "loss": 46.0,
-      "step": 33650
-    },
-    {
-      "epoch": 5.419260034622972,
-      "grad_norm": 0.012442146427929401,
-      "learning_rate": 0.00019998551727691316,
-      "loss": 46.0,
-      "step": 33651
-    },
-    {
-      "epoch": 5.419421071701759,
-      "grad_norm": 0.01483453344553709,
-      "learning_rate": 0.00019998551641590493,
-      "loss": 46.0,
-      "step": 33652
-    },
-    {
-      "epoch": 5.4195821087805465,
-      "grad_norm": 0.0077059767208993435,
-      "learning_rate": 0.0001999855155548711,
-      "loss": 46.0,
-      "step": 33653
-    },
-    {
-      "epoch": 5.419743145859334,
-      "grad_norm": 0.0020480400417000055,
-      "learning_rate": 0.0001999855146938117,
-      "loss": 46.0,
-      "step": 33654
-    },
-    {
-      "epoch": 5.419904182938121,
-      "grad_norm": 0.004197259899228811,
-      "learning_rate": 0.0001999855138327267,
-      "loss": 46.0,
-      "step": 33655
-    },
-    {
-      "epoch": 5.420065220016909,
-      "grad_norm": 0.009529270231723785,
-      "learning_rate": 0.00019998551297161613,
-      "loss": 46.0,
-      "step": 33656
-    },
-    {
-      "epoch": 5.420226257095696,
-      "grad_norm": 0.007029429078102112,
-      "learning_rate": 0.00019998551211048,
-      "loss": 46.0,
-      "step": 33657
-    },
-    {
-      "epoch": 5.420387294174484,
-      "grad_norm": 0.0033611590042710304,
-      "learning_rate": 0.0001999855112493182,
-      "loss": 46.0,
-      "step": 33658
-    },
-    {
-      "epoch": 5.420548331253271,
-      "grad_norm": 0.0013261355925351381,
-      "learning_rate": 0.00019998551038813084,
-      "loss": 46.0,
-      "step": 33659
-    },
-    {
-      "epoch": 5.420709368332059,
-      "grad_norm": 0.005337535869330168,
-      "learning_rate": 0.0001999855095269179,
-      "loss": 46.0,
-      "step": 33660
-    },
-    {
-      "epoch": 5.420870405410846,
-      "grad_norm": 0.00729743717238307,
-      "learning_rate": 0.00019998550866567936,
-      "loss": 46.0,
-      "step": 33661
-    },
-    {
-      "epoch": 5.4210314424896335,
-      "grad_norm": 0.0034848027862608433,
-      "learning_rate": 0.00019998550780441523,
-      "loss": 46.0,
-      "step": 33662
-    },
-    {
-      "epoch": 5.421192479568421,
-      "grad_norm": 0.012427321635186672,
-      "learning_rate": 0.00019998550694312552,
-      "loss": 46.0,
-      "step": 33663
-    },
-    {
-      "epoch": 5.421353516647208,
-      "grad_norm": 0.011609414592385292,
-      "learning_rate": 0.00019998550608181018,
-      "loss": 46.0,
-      "step": 33664
-    },
-    {
-      "epoch": 5.421514553725995,
-      "grad_norm": 0.001503894105553627,
-      "learning_rate": 0.0001999855052204693,
-      "loss": 46.0,
-      "step": 33665
-    },
-    {
-      "epoch": 5.4216755908047825,
-      "grad_norm": 0.007099046837538481,
-      "learning_rate": 0.00019998550435910278,
-      "loss": 46.0,
-      "step": 33666
-    },
-    {
-      "epoch": 5.42183662788357,
-      "grad_norm": 0.001849538297392428,
-      "learning_rate": 0.00019998550349771072,
-      "loss": 46.0,
-      "step": 33667
-    },
-    {
-      "epoch": 5.421997664962357,
-      "grad_norm": 0.0016216576332226396,
-      "learning_rate": 0.00019998550263629304,
-      "loss": 46.0,
-      "step": 33668
-    },
-    {
-      "epoch": 5.422158702041145,
-      "grad_norm": 0.0057668848894536495,
-      "learning_rate": 0.00019998550177484977,
-      "loss": 46.0,
-      "step": 33669
-    },
-    {
-      "epoch": 5.422319739119932,
-      "grad_norm": 0.0019296335522085428,
-      "learning_rate": 0.00019998550091338091,
-      "loss": 46.0,
-      "step": 33670
-    },
-    {
-      "epoch": 5.42248077619872,
-      "grad_norm": 0.001681244932115078,
-      "learning_rate": 0.00019998550005188647,
-      "loss": 46.0,
-      "step": 33671
-    },
-    {
-      "epoch": 5.422641813277507,
-      "grad_norm": 0.0051642353646457195,
-      "learning_rate": 0.00019998549919036642,
-      "loss": 46.0,
-      "step": 33672
-    },
-    {
-      "epoch": 5.422802850356295,
-      "grad_norm": 0.0020802582148462534,
-      "learning_rate": 0.0001999854983288208,
-      "loss": 46.0,
-      "step": 33673
-    },
-    {
-      "epoch": 5.422963887435082,
-      "grad_norm": 0.0008842212264426053,
-      "learning_rate": 0.00019998549746724957,
-      "loss": 46.0,
-      "step": 33674
-    },
-    {
-      "epoch": 5.4231249245138695,
-      "grad_norm": 0.01572221890091896,
-      "learning_rate": 0.00019998549660565278,
-      "loss": 46.0,
-      "step": 33675
-    },
-    {
-      "epoch": 5.423285961592657,
-      "grad_norm": 0.0037692878395318985,
-      "learning_rate": 0.00019998549574403037,
-      "loss": 46.0,
-      "step": 33676
-    },
-    {
-      "epoch": 5.423446998671444,
-      "grad_norm": 0.0034856435377150774,
-      "learning_rate": 0.00019998549488238235,
-      "loss": 46.0,
-      "step": 33677
-    },
-    {
-      "epoch": 5.423608035750232,
-      "grad_norm": 0.013210601173341274,
-      "learning_rate": 0.0001999854940207088,
-      "loss": 46.0,
-      "step": 33678
-    },
-    {
-      "epoch": 5.423769072829019,
-      "grad_norm": 0.0023444397374987602,
-      "learning_rate": 0.00019998549315900958,
-      "loss": 46.0,
-      "step": 33679
-    },
-    {
-      "epoch": 5.423930109907806,
-      "grad_norm": 0.014169315807521343,
-      "learning_rate": 0.00019998549229728482,
-      "loss": 46.0,
-      "step": 33680
-    },
-    {
-      "epoch": 5.424091146986593,
-      "grad_norm": 0.007368167862296104,
-      "learning_rate": 0.00019998549143553446,
-      "loss": 46.0,
-      "step": 33681
-    },
-    {
-      "epoch": 5.424252184065381,
-      "grad_norm": 0.002794538391754031,
-      "learning_rate": 0.0001999854905737585,
-      "loss": 46.0,
-      "step": 33682
-    },
-    {
-      "epoch": 5.424413221144168,
-      "grad_norm": 0.011421543546020985,
-      "learning_rate": 0.00019998548971195696,
-      "loss": 46.0,
-      "step": 33683
-    },
-    {
-      "epoch": 5.424574258222956,
-      "grad_norm": 0.009053487330675125,
-      "learning_rate": 0.00019998548885012985,
-      "loss": 46.0,
-      "step": 33684
-    },
-    {
-      "epoch": 5.424735295301743,
-      "grad_norm": 0.002092099515721202,
-      "learning_rate": 0.0001999854879882771,
-      "loss": 46.0,
-      "step": 33685
-    },
-    {
-      "epoch": 5.424896332380531,
-      "grad_norm": 0.0028156281914561987,
-      "learning_rate": 0.00019998548712639878,
-      "loss": 46.0,
-      "step": 33686
-    },
-    {
-      "epoch": 5.425057369459318,
-      "grad_norm": 0.001339444424957037,
-      "learning_rate": 0.0001999854862644949,
-      "loss": 46.0,
-      "step": 33687
-    },
-    {
-      "epoch": 5.4252184065381055,
-      "grad_norm": 0.004757916089147329,
-      "learning_rate": 0.00019998548540256538,
-      "loss": 46.0,
-      "step": 33688
-    },
-    {
-      "epoch": 5.425379443616893,
-      "grad_norm": 0.005150027573108673,
-      "learning_rate": 0.0001999854845406103,
-      "loss": 46.0,
-      "step": 33689
-    },
-    {
-      "epoch": 5.42554048069568,
-      "grad_norm": 0.008297018706798553,
-      "learning_rate": 0.0001999854836786296,
-      "loss": 46.0,
-      "step": 33690
-    },
-    {
-      "epoch": 5.425701517774468,
-      "grad_norm": 0.0023368827532976866,
-      "learning_rate": 0.00019998548281662334,
-      "loss": 46.0,
-      "step": 33691
-    },
-    {
-      "epoch": 5.425862554853255,
-      "grad_norm": 0.0016465721419081092,
-      "learning_rate": 0.0001999854819545915,
-      "loss": 46.0,
-      "step": 33692
-    },
-    {
-      "epoch": 5.426023591932043,
-      "grad_norm": 0.004651474766433239,
-      "learning_rate": 0.00019998548109253405,
-      "loss": 46.0,
-      "step": 33693
-    },
-    {
-      "epoch": 5.42618462901083,
-      "grad_norm": 0.0022437949664890766,
-      "learning_rate": 0.00019998548023045096,
-      "loss": 46.0,
-      "step": 33694
-    },
-    {
-      "epoch": 5.426345666089617,
-      "grad_norm": 0.0022637040819972754,
-      "learning_rate": 0.00019998547936834232,
-      "loss": 46.0,
-      "step": 33695
-    },
-    {
-      "epoch": 5.426506703168404,
-      "grad_norm": 0.004307035356760025,
-      "learning_rate": 0.00019998547850620812,
-      "loss": 46.0,
-      "step": 33696
-    },
-    {
-      "epoch": 5.426667740247192,
-      "grad_norm": 0.0020192808005958796,
-      "learning_rate": 0.0001999854776440483,
-      "loss": 46.0,
-      "step": 33697
-    },
-    {
-      "epoch": 5.426828777325979,
-      "grad_norm": 0.0075362008064985275,
-      "learning_rate": 0.00019998547678186287,
-      "loss": 46.0,
-      "step": 33698
-    },
-    {
-      "epoch": 5.426989814404767,
-      "grad_norm": 0.0031508186366409063,
-      "learning_rate": 0.00019998547591965188,
-      "loss": 46.0,
-      "step": 33699
-    },
-    {
-      "epoch": 5.427150851483554,
-      "grad_norm": 0.014034881256520748,
-      "learning_rate": 0.00019998547505741527,
-      "loss": 46.0,
-      "step": 33700
-    },
-    {
-      "epoch": 5.4273118885623415,
-      "grad_norm": 0.003862165380269289,
-      "learning_rate": 0.0001999854741951531,
-      "loss": 46.0,
-      "step": 33701
-    },
-    {
-      "epoch": 5.427472925641129,
-      "grad_norm": 0.0027518896386027336,
-      "learning_rate": 0.00019998547333286532,
-      "loss": 46.0,
-      "step": 33702
-    },
-    {
-      "epoch": 5.427633962719916,
-      "grad_norm": 0.0015293208416551352,
-      "learning_rate": 0.00019998547247055198,
-      "loss": 46.0,
-      "step": 33703
-    },
-    {
-      "epoch": 5.427794999798704,
-      "grad_norm": 0.008325648494064808,
-      "learning_rate": 0.000199985471608213,
-      "loss": 46.0,
-      "step": 33704
-    },
-    {
-      "epoch": 5.427956036877491,
-      "grad_norm": 0.011661135591566563,
-      "learning_rate": 0.00019998547074584843,
-      "loss": 46.0,
-      "step": 33705
-    },
-    {
-      "epoch": 5.428117073956279,
-      "grad_norm": 0.01024138554930687,
-      "learning_rate": 0.00019998546988345833,
-      "loss": 46.0,
-      "step": 33706
-    },
-    {
-      "epoch": 5.428278111035066,
-      "grad_norm": 0.0030228830873966217,
-      "learning_rate": 0.0001999854690210426,
-      "loss": 46.0,
-      "step": 33707
-    },
-    {
-      "epoch": 5.428439148113853,
-      "grad_norm": 0.0019812029786407948,
-      "learning_rate": 0.00019998546815860126,
-      "loss": 46.0,
-      "step": 33708
-    },
-    {
-      "epoch": 5.42860018519264,
-      "grad_norm": 0.0019485143711790442,
-      "learning_rate": 0.00019998546729613434,
-      "loss": 46.0,
-      "step": 33709
-    },
-    {
-      "epoch": 5.428761222271428,
-      "grad_norm": 0.0027567846700549126,
-      "learning_rate": 0.00019998546643364184,
-      "loss": 46.0,
-      "step": 33710
-    },
-    {
-      "epoch": 5.428922259350215,
-      "grad_norm": 0.017840752378106117,
-      "learning_rate": 0.00019998546557112374,
-      "loss": 46.0,
-      "step": 33711
-    },
-    {
-      "epoch": 5.429083296429003,
-      "grad_norm": 0.017581719905138016,
-      "learning_rate": 0.00019998546470858007,
-      "loss": 46.0,
-      "step": 33712
-    },
-    {
-      "epoch": 5.42924433350779,
-      "grad_norm": 0.000688565953169018,
-      "learning_rate": 0.00019998546384601077,
-      "loss": 46.0,
-      "step": 33713
-    },
-    {
-      "epoch": 5.4294053705865775,
-      "grad_norm": 0.008608333766460419,
-      "learning_rate": 0.00019998546298341592,
-      "loss": 46.0,
-      "step": 33714
-    },
-    {
-      "epoch": 5.429566407665365,
-      "grad_norm": 0.002623354783281684,
-      "learning_rate": 0.00019998546212079545,
-      "loss": 46.0,
-      "step": 33715
-    },
-    {
-      "epoch": 5.429727444744152,
-      "grad_norm": 0.004323564935475588,
-      "learning_rate": 0.0001999854612581494,
-      "loss": 46.0,
-      "step": 33716
-    },
-    {
-      "epoch": 5.42988848182294,
-      "grad_norm": 0.002725014230236411,
-      "learning_rate": 0.00019998546039547775,
-      "loss": 46.0,
-      "step": 33717
-    },
-    {
-      "epoch": 5.430049518901727,
-      "grad_norm": 0.005183933302760124,
-      "learning_rate": 0.00019998545953278052,
-      "loss": 46.0,
-      "step": 33718
-    },
-    {
-      "epoch": 5.430210555980515,
-      "grad_norm": 0.01034574769437313,
-      "learning_rate": 0.0001999854586700577,
-      "loss": 46.0,
-      "step": 33719
-    },
-    {
-      "epoch": 5.430371593059302,
-      "grad_norm": 0.005405900999903679,
-      "learning_rate": 0.00019998545780730928,
-      "loss": 46.0,
-      "step": 33720
-    },
-    {
-      "epoch": 5.43053263013809,
-      "grad_norm": 0.008120122365653515,
-      "learning_rate": 0.0001999854569445353,
-      "loss": 46.0,
-      "step": 33721
-    },
-    {
-      "epoch": 5.430693667216877,
-      "grad_norm": 0.003856923431158066,
-      "learning_rate": 0.00019998545608173565,
-      "loss": 46.0,
-      "step": 33722
-    },
-    {
-      "epoch": 5.430854704295664,
-      "grad_norm": 0.006800223141908646,
-      "learning_rate": 0.0001999854552189105,
-      "loss": 46.0,
-      "step": 33723
-    },
-    {
-      "epoch": 5.431015741374451,
-      "grad_norm": 0.0028839276637881994,
-      "learning_rate": 0.0001999854543560597,
-      "loss": 46.0,
-      "step": 33724
-    },
-    {
-      "epoch": 5.4311767784532385,
-      "grad_norm": 0.01973271183669567,
-      "learning_rate": 0.00019998545349318332,
-      "loss": 46.0,
-      "step": 33725
-    },
-    {
-      "epoch": 5.431337815532026,
-      "grad_norm": 0.004148564301431179,
-      "learning_rate": 0.00019998545263028136,
-      "loss": 46.0,
-      "step": 33726
-    },
-    {
-      "epoch": 5.431498852610813,
-      "grad_norm": 0.004408187232911587,
-      "learning_rate": 0.00019998545176735382,
-      "loss": 46.0,
-      "step": 33727
-    },
-    {
-      "epoch": 5.431659889689601,
-      "grad_norm": 0.0015528157819062471,
-      "learning_rate": 0.00019998545090440067,
-      "loss": 46.0,
-      "step": 33728
-    },
-    {
-      "epoch": 5.431820926768388,
-      "grad_norm": 0.008553935214877129,
-      "learning_rate": 0.00019998545004142193,
-      "loss": 46.0,
-      "step": 33729
-    },
-    {
-      "epoch": 5.431981963847176,
-      "grad_norm": 0.006922516506165266,
-      "learning_rate": 0.00019998544917841762,
-      "loss": 46.0,
-      "step": 33730
-    },
-    {
-      "epoch": 5.432143000925963,
-      "grad_norm": 0.0024310853332281113,
-      "learning_rate": 0.0001999854483153877,
-      "loss": 46.0,
-      "step": 33731
-    },
-    {
-      "epoch": 5.432304038004751,
-      "grad_norm": 0.0049537308514118195,
-      "learning_rate": 0.00019998544745233218,
-      "loss": 46.0,
-      "step": 33732
-    },
-    {
-      "epoch": 5.432465075083538,
-      "grad_norm": 0.010817292146384716,
-      "learning_rate": 0.00019998544658925108,
-      "loss": 46.0,
-      "step": 33733
-    },
-    {
-      "epoch": 5.432626112162326,
-      "grad_norm": 0.016791680827736855,
-      "learning_rate": 0.0001999854457261444,
-      "loss": 46.0,
-      "step": 33734
-    },
-    {
-      "epoch": 5.432787149241113,
-      "grad_norm": 0.0012591202976182103,
-      "learning_rate": 0.00019998544486301211,
-      "loss": 46.0,
-      "step": 33735
-    },
-    {
-      "epoch": 5.4329481863199005,
-      "grad_norm": 0.003230729606002569,
-      "learning_rate": 0.0001999854439998542,
-      "loss": 46.0,
-      "step": 33736
-    },
-    {
-      "epoch": 5.433109223398688,
-      "grad_norm": 0.001971533754840493,
-      "learning_rate": 0.00019998544313667077,
-      "loss": 46.0,
-      "step": 33737
-    },
-    {
-      "epoch": 5.4332702604774745,
-      "grad_norm": 0.009771594777703285,
-      "learning_rate": 0.0001999854422734617,
-      "loss": 46.0,
-      "step": 33738
-    },
-    {
-      "epoch": 5.433431297556262,
-      "grad_norm": 0.007710725534707308,
-      "learning_rate": 0.00019998544141022707,
-      "loss": 46.0,
-      "step": 33739
-    },
-    {
-      "epoch": 5.433592334635049,
-      "grad_norm": 0.004870411939918995,
-      "learning_rate": 0.00019998544054696682,
-      "loss": 46.0,
-      "step": 33740
-    },
-    {
-      "epoch": 5.433753371713837,
-      "grad_norm": 0.006178237032145262,
-      "learning_rate": 0.00019998543968368097,
-      "loss": 46.0,
-      "step": 33741
-    },
-    {
-      "epoch": 5.433914408792624,
-      "grad_norm": 0.004385080188512802,
-      "learning_rate": 0.00019998543882036954,
-      "loss": 46.0,
-      "step": 33742
-    },
-    {
-      "epoch": 5.434075445871412,
-      "grad_norm": 0.0019855606369674206,
-      "learning_rate": 0.00019998543795703255,
-      "loss": 46.0,
-      "step": 33743
-    },
-    {
-      "epoch": 5.434236482950199,
-      "grad_norm": 0.01655840501189232,
-      "learning_rate": 0.00019998543709366995,
-      "loss": 46.0,
-      "step": 33744
-    },
-    {
-      "epoch": 5.434397520028987,
-      "grad_norm": 0.0004960960941389203,
-      "learning_rate": 0.00019998543623028176,
-      "loss": 46.0,
-      "step": 33745
-    },
-    {
-      "epoch": 5.434558557107774,
-      "grad_norm": 0.0033130559604614973,
-      "learning_rate": 0.00019998543536686798,
-      "loss": 46.0,
-      "step": 33746
-    },
-    {
-      "epoch": 5.434719594186562,
-      "grad_norm": 0.009820588864386082,
-      "learning_rate": 0.0001999854345034286,
-      "loss": 46.0,
-      "step": 33747
-    },
-    {
-      "epoch": 5.434880631265349,
-      "grad_norm": 0.0012613895814865828,
-      "learning_rate": 0.00019998543363996364,
-      "loss": 46.0,
-      "step": 33748
-    },
-    {
-      "epoch": 5.4350416683441365,
-      "grad_norm": 0.007635081652551889,
-      "learning_rate": 0.00019998543277647307,
-      "loss": 46.0,
-      "step": 33749
-    },
-    {
-      "epoch": 5.435202705422924,
-      "grad_norm": 0.0116058848798275,
-      "learning_rate": 0.00019998543191295691,
-      "loss": 46.0,
-      "step": 33750
-    },
-    {
-      "epoch": 5.435363742501711,
-      "grad_norm": 0.007447515148669481,
-      "learning_rate": 0.00019998543104941517,
-      "loss": 46.0,
-      "step": 33751
-    },
-    {
-      "epoch": 5.435524779580499,
-      "grad_norm": 0.004526468459516764,
-      "learning_rate": 0.00019998543018584784,
-      "loss": 46.0,
-      "step": 33752
-    },
-    {
-      "epoch": 5.435685816659285,
-      "grad_norm": 0.002396880416199565,
-      "learning_rate": 0.0001999854293222549,
-      "loss": 46.0,
-      "step": 33753
-    },
-    {
-      "epoch": 5.435846853738073,
-      "grad_norm": 0.010567798279225826,
-      "learning_rate": 0.0001999854284586364,
-      "loss": 46.0,
-      "step": 33754
-    },
-    {
-      "epoch": 5.43600789081686,
-      "grad_norm": 0.0034893634729087353,
-      "learning_rate": 0.0001999854275949923,
-      "loss": 46.0,
-      "step": 33755
-    },
-    {
-      "epoch": 5.436168927895648,
-      "grad_norm": 0.0030591634567826986,
-      "learning_rate": 0.00019998542673132258,
-      "loss": 46.0,
-      "step": 33756
-    },
-    {
-      "epoch": 5.436329964974435,
-      "grad_norm": 0.005990249570459127,
-      "learning_rate": 0.00019998542586762729,
-      "loss": 46.0,
-      "step": 33757
-    },
-    {
-      "epoch": 5.436491002053223,
-      "grad_norm": 0.01983432099223137,
-      "learning_rate": 0.00019998542500390643,
-      "loss": 46.0,
-      "step": 33758
-    },
-    {
-      "epoch": 5.43665203913201,
-      "grad_norm": 0.0014307056553661823,
-      "learning_rate": 0.00019998542414015997,
-      "loss": 46.0,
-      "step": 33759
-    },
-    {
-      "epoch": 5.436813076210798,
-      "grad_norm": 0.0015668855048716068,
-      "learning_rate": 0.0001999854232763879,
-      "loss": 46.0,
-      "step": 33760
-    },
-    {
-      "epoch": 5.436974113289585,
-      "grad_norm": 0.0019072664435952902,
-      "learning_rate": 0.00019998542241259025,
-      "loss": 46.0,
-      "step": 33761
-    },
-    {
-      "epoch": 5.4371351503683725,
-      "grad_norm": 0.005499883554875851,
-      "learning_rate": 0.000199985421548767,
-      "loss": 46.0,
-      "step": 33762
-    },
-    {
-      "epoch": 5.43729618744716,
-      "grad_norm": 0.0028436563443392515,
-      "learning_rate": 0.00019998542068491818,
-      "loss": 46.0,
-      "step": 33763
-    },
-    {
-      "epoch": 5.437457224525947,
-      "grad_norm": 0.00255585671402514,
-      "learning_rate": 0.00019998541982104375,
-      "loss": 46.0,
-      "step": 33764
-    },
-    {
-      "epoch": 5.437618261604735,
-      "grad_norm": 0.004707845393568277,
-      "learning_rate": 0.00019998541895714373,
-      "loss": 46.0,
-      "step": 33765
-    },
-    {
-      "epoch": 5.437779298683522,
-      "grad_norm": 0.017210451886057854,
-      "learning_rate": 0.0001999854180932181,
-      "loss": 46.0,
-      "step": 33766
-    },
-    {
-      "epoch": 5.43794033576231,
-      "grad_norm": 0.0029135614167898893,
-      "learning_rate": 0.00019998541722926691,
-      "loss": 46.0,
-      "step": 33767
-    },
-    {
-      "epoch": 5.438101372841096,
-      "grad_norm": 0.0022394165862351656,
-      "learning_rate": 0.0001999854163652901,
-      "loss": 46.0,
-      "step": 33768
-    },
-    {
-      "epoch": 5.438262409919884,
-      "grad_norm": 0.00235547567717731,
-      "learning_rate": 0.00019998541550128774,
-      "loss": 46.0,
-      "step": 33769
-    },
-    {
-      "epoch": 5.438423446998671,
-      "grad_norm": 0.0018850163323804736,
-      "learning_rate": 0.00019998541463725977,
-      "loss": 46.0,
-      "step": 33770
-    },
-    {
-      "epoch": 5.438584484077459,
-      "grad_norm": 0.0011621728772297502,
-      "learning_rate": 0.0001999854137732062,
-      "loss": 46.0,
-      "step": 33771
-    },
-    {
-      "epoch": 5.438745521156246,
-      "grad_norm": 0.003751841140910983,
-      "learning_rate": 0.00019998541290912707,
-      "loss": 46.0,
-      "step": 33772
-    },
-    {
-      "epoch": 5.4389065582350336,
-      "grad_norm": 0.01840587519109249,
-      "learning_rate": 0.0001999854120450223,
-      "loss": 46.0,
-      "step": 33773
-    },
-    {
-      "epoch": 5.439067595313821,
-      "grad_norm": 0.0015066039049997926,
-      "learning_rate": 0.00019998541118089195,
-      "loss": 46.0,
-      "step": 33774
-    },
-    {
-      "epoch": 5.4392286323926085,
-      "grad_norm": 0.004510980565100908,
-      "learning_rate": 0.00019998541031673606,
-      "loss": 46.0,
-      "step": 33775
-    },
-    {
-      "epoch": 5.439389669471396,
-      "grad_norm": 0.011435563676059246,
-      "learning_rate": 0.00019998540945255453,
-      "loss": 46.0,
-      "step": 33776
-    },
-    {
-      "epoch": 5.439550706550183,
-      "grad_norm": 0.013452005572617054,
-      "learning_rate": 0.00019998540858834742,
-      "loss": 46.0,
-      "step": 33777
-    },
-    {
-      "epoch": 5.439711743628971,
-      "grad_norm": 0.0035934315528720617,
-      "learning_rate": 0.0001999854077241147,
-      "loss": 46.0,
-      "step": 33778
-    },
-    {
-      "epoch": 5.439872780707758,
-      "grad_norm": 0.007829852402210236,
-      "learning_rate": 0.00019998540685985642,
-      "loss": 46.0,
-      "step": 33779
-    },
-    {
-      "epoch": 5.440033817786546,
-      "grad_norm": 0.006716438569128513,
-      "learning_rate": 0.00019998540599557255,
-      "loss": 46.0,
-      "step": 33780
-    },
-    {
-      "epoch": 5.440194854865332,
-      "grad_norm": 0.005890178959816694,
-      "learning_rate": 0.00019998540513126305,
-      "loss": 46.0,
-      "step": 33781
-    },
-    {
-      "epoch": 5.44035589194412,
-      "grad_norm": 0.0033761782106012106,
-      "learning_rate": 0.00019998540426692798,
-      "loss": 46.0,
-      "step": 33782
-    },
-    {
-      "epoch": 5.440516929022907,
-      "grad_norm": 0.00318462448194623,
-      "learning_rate": 0.00019998540340256734,
-      "loss": 46.0,
-      "step": 33783
-    },
-    {
-      "epoch": 5.440677966101695,
-      "grad_norm": 0.018436431884765625,
-      "learning_rate": 0.00019998540253818108,
-      "loss": 46.0,
-      "step": 33784
-    },
-    {
-      "epoch": 5.440839003180482,
-      "grad_norm": 0.005125865805894136,
-      "learning_rate": 0.00019998540167376921,
-      "loss": 46.0,
-      "step": 33785
-    },
-    {
-      "epoch": 5.4410000402592695,
-      "grad_norm": 0.00320666516199708,
-      "learning_rate": 0.00019998540080933181,
-      "loss": 46.0,
-      "step": 33786
-    },
-    {
-      "epoch": 5.441161077338057,
-      "grad_norm": 0.003564119804650545,
-      "learning_rate": 0.00019998539994486877,
-      "loss": 46.0,
-      "step": 33787
-    },
-    {
-      "epoch": 5.441322114416844,
-      "grad_norm": 0.007742627523839474,
-      "learning_rate": 0.00019998539908038017,
-      "loss": 46.0,
-      "step": 33788
-    },
-    {
-      "epoch": 5.441483151495632,
-      "grad_norm": 0.004321575164794922,
-      "learning_rate": 0.00019998539821586598,
-      "loss": 46.0,
-      "step": 33789
-    },
-    {
-      "epoch": 5.441644188574419,
-      "grad_norm": 0.001923218253068626,
-      "learning_rate": 0.00019998539735132618,
-      "loss": 46.0,
-      "step": 33790
-    },
-    {
-      "epoch": 5.441805225653207,
-      "grad_norm": 0.000759912480134517,
-      "learning_rate": 0.00019998539648676079,
-      "loss": 46.0,
-      "step": 33791
-    },
-    {
-      "epoch": 5.441966262731994,
-      "grad_norm": 0.0036928998306393623,
-      "learning_rate": 0.0001999853956221698,
-      "loss": 46.0,
-      "step": 33792
-    },
-    {
-      "epoch": 5.442127299810782,
-      "grad_norm": 0.0011880164965987206,
-      "learning_rate": 0.00019998539475755321,
-      "loss": 46.0,
-      "step": 33793
-    },
-    {
-      "epoch": 5.442288336889569,
-      "grad_norm": 0.0013276466634124517,
-      "learning_rate": 0.0001999853938929111,
-      "loss": 46.0,
-      "step": 33794
-    },
-    {
-      "epoch": 5.442449373968357,
-      "grad_norm": 0.0013939006021246314,
-      "learning_rate": 0.00019998539302824332,
-      "loss": 46.0,
-      "step": 33795
-    },
-    {
-      "epoch": 5.442610411047143,
-      "grad_norm": 0.012781567871570587,
-      "learning_rate": 0.00019998539216355,
-      "loss": 46.0,
-      "step": 33796
-    },
-    {
-      "epoch": 5.442771448125931,
-      "grad_norm": 0.0018839980475604534,
-      "learning_rate": 0.00019998539129883106,
-      "loss": 46.0,
-      "step": 33797
-    },
-    {
-      "epoch": 5.442932485204718,
-      "grad_norm": 0.0020724984351545572,
-      "learning_rate": 0.00019998539043408653,
-      "loss": 46.0,
-      "step": 33798
-    },
-    {
-      "epoch": 5.4430935222835055,
-      "grad_norm": 0.007911965250968933,
-      "learning_rate": 0.0001999853895693164,
-      "loss": 46.0,
-      "step": 33799
-    },
-    {
-      "epoch": 5.443254559362293,
-      "grad_norm": 0.0013278990518301725,
-      "learning_rate": 0.0001999853887045207,
-      "loss": 46.0,
-      "step": 33800
-    },
-    {
-      "epoch": 5.44341559644108,
-      "grad_norm": 0.007801434490829706,
-      "learning_rate": 0.0001999853878396994,
-      "loss": 46.0,
-      "step": 33801
-    },
-    {
-      "epoch": 5.443576633519868,
-      "grad_norm": 0.006084978114813566,
-      "learning_rate": 0.00019998538697485252,
-      "loss": 46.0,
-      "step": 33802
-    },
-    {
-      "epoch": 5.443737670598655,
-      "grad_norm": 0.006249023601412773,
-      "learning_rate": 0.00019998538610998002,
-      "loss": 46.0,
-      "step": 33803
-    },
-    {
-      "epoch": 5.443898707677443,
-      "grad_norm": 0.008042052388191223,
-      "learning_rate": 0.00019998538524508197,
-      "loss": 46.0,
-      "step": 33804
-    },
-    {
-      "epoch": 5.44405974475623,
-      "grad_norm": 0.00547713553532958,
-      "learning_rate": 0.0001999853843801583,
-      "loss": 46.0,
-      "step": 33805
-    },
-    {
-      "epoch": 5.444220781835018,
-      "grad_norm": 0.010139192454516888,
-      "learning_rate": 0.00019998538351520906,
-      "loss": 46.0,
-      "step": 33806
-    },
-    {
-      "epoch": 5.444381818913805,
-      "grad_norm": 0.0010297633707523346,
-      "learning_rate": 0.00019998538265023422,
-      "loss": 46.0,
-      "step": 33807
-    },
-    {
-      "epoch": 5.444542855992593,
-      "grad_norm": 0.01330942939966917,
-      "learning_rate": 0.00019998538178523376,
-      "loss": 46.0,
-      "step": 33808
-    },
-    {
-      "epoch": 5.44470389307138,
-      "grad_norm": 0.00419691763818264,
-      "learning_rate": 0.00019998538092020772,
-      "loss": 46.0,
-      "step": 33809
-    },
-    {
-      "epoch": 5.4448649301501675,
-      "grad_norm": 0.0010829964885488153,
-      "learning_rate": 0.0001999853800551561,
-      "loss": 46.0,
-      "step": 33810
-    },
-    {
-      "epoch": 5.445025967228954,
-      "grad_norm": 0.001953398808836937,
-      "learning_rate": 0.0001999853791900789,
-      "loss": 46.0,
-      "step": 33811
-    },
-    {
-      "epoch": 5.4451870043077415,
-      "grad_norm": 0.007270127534866333,
-      "learning_rate": 0.0001999853783249761,
-      "loss": 46.0,
-      "step": 33812
-    },
-    {
-      "epoch": 5.445348041386529,
-      "grad_norm": 0.003097108332440257,
-      "learning_rate": 0.0001999853774598477,
-      "loss": 46.0,
-      "step": 33813
-    },
-    {
-      "epoch": 5.445509078465316,
-      "grad_norm": 0.0017479805974289775,
-      "learning_rate": 0.00019998537659469373,
-      "loss": 46.0,
-      "step": 33814
-    },
-    {
-      "epoch": 5.445670115544104,
-      "grad_norm": 0.0027759966906160116,
-      "learning_rate": 0.00019998537572951417,
-      "loss": 46.0,
-      "step": 33815
-    },
-    {
-      "epoch": 5.445831152622891,
-      "grad_norm": 0.00909650232642889,
-      "learning_rate": 0.000199985374864309,
-      "loss": 46.0,
-      "step": 33816
-    },
-    {
-      "epoch": 5.445992189701679,
-      "grad_norm": 0.001967012882232666,
-      "learning_rate": 0.00019998537399907825,
-      "loss": 46.0,
-      "step": 33817
-    },
-    {
-      "epoch": 5.446153226780466,
-      "grad_norm": 0.0024580832105129957,
-      "learning_rate": 0.00019998537313382187,
-      "loss": 46.0,
-      "step": 33818
-    },
-    {
-      "epoch": 5.446314263859254,
-      "grad_norm": 0.006300802808254957,
-      "learning_rate": 0.00019998537226853996,
-      "loss": 46.0,
-      "step": 33819
-    },
-    {
-      "epoch": 5.446475300938041,
-      "grad_norm": 0.004440753720700741,
-      "learning_rate": 0.00019998537140323243,
-      "loss": 46.0,
-      "step": 33820
-    },
-    {
-      "epoch": 5.446636338016829,
-      "grad_norm": 0.00453588692471385,
-      "learning_rate": 0.0001999853705378993,
-      "loss": 46.0,
-      "step": 33821
-    },
-    {
-      "epoch": 5.446797375095616,
-      "grad_norm": 0.0036536843981593847,
-      "learning_rate": 0.0001999853696725406,
-      "loss": 46.0,
-      "step": 33822
-    },
-    {
-      "epoch": 5.4469584121744035,
-      "grad_norm": 0.012010179460048676,
-      "learning_rate": 0.00019998536880715627,
-      "loss": 46.0,
-      "step": 33823
-    },
-    {
-      "epoch": 5.447119449253191,
-      "grad_norm": 0.005867932457476854,
-      "learning_rate": 0.0001999853679417464,
-      "loss": 46.0,
-      "step": 33824
-    },
-    {
-      "epoch": 5.447280486331978,
-      "grad_norm": 0.005730391480028629,
-      "learning_rate": 0.0001999853670763109,
-      "loss": 46.0,
-      "step": 33825
-    },
-    {
-      "epoch": 5.447441523410765,
-      "grad_norm": 0.00212597637437284,
-      "learning_rate": 0.00019998536621084983,
-      "loss": 46.0,
-      "step": 33826
-    },
-    {
-      "epoch": 5.447602560489552,
-      "grad_norm": 0.0027515387628227472,
-      "learning_rate": 0.00019998536534536316,
-      "loss": 46.0,
-      "step": 33827
-    },
-    {
-      "epoch": 5.44776359756834,
-      "grad_norm": 0.008567631244659424,
-      "learning_rate": 0.0001999853644798509,
-      "loss": 46.0,
-      "step": 33828
-    },
-    {
-      "epoch": 5.447924634647127,
-      "grad_norm": 0.002860043430700898,
-      "learning_rate": 0.00019998536361431305,
-      "loss": 46.0,
-      "step": 33829
-    },
-    {
-      "epoch": 5.448085671725915,
-      "grad_norm": 0.01580500416457653,
-      "learning_rate": 0.00019998536274874962,
-      "loss": 46.0,
-      "step": 33830
-    },
-    {
-      "epoch": 5.448246708804702,
-      "grad_norm": 0.00769634498283267,
-      "learning_rate": 0.00019998536188316058,
-      "loss": 46.0,
-      "step": 33831
-    },
-    {
-      "epoch": 5.44840774588349,
-      "grad_norm": 0.0036289342679083347,
-      "learning_rate": 0.00019998536101754595,
-      "loss": 46.0,
-      "step": 33832
-    },
-    {
-      "epoch": 5.448568782962277,
-      "grad_norm": 0.014895744621753693,
-      "learning_rate": 0.00019998536015190574,
-      "loss": 46.0,
-      "step": 33833
-    },
-    {
-      "epoch": 5.4487298200410645,
-      "grad_norm": 0.016011957079172134,
-      "learning_rate": 0.00019998535928623994,
-      "loss": 46.0,
-      "step": 33834
-    },
-    {
-      "epoch": 5.448890857119852,
-      "grad_norm": 0.006428055465221405,
-      "learning_rate": 0.00019998535842054852,
-      "loss": 46.0,
-      "step": 33835
-    },
-    {
-      "epoch": 5.4490518941986394,
-      "grad_norm": 0.0018947329372167587,
-      "learning_rate": 0.00019998535755483157,
-      "loss": 46.0,
-      "step": 33836
-    },
-    {
-      "epoch": 5.449212931277427,
-      "grad_norm": 0.005177473183721304,
-      "learning_rate": 0.00019998535668908898,
-      "loss": 46.0,
-      "step": 33837
-    },
-    {
-      "epoch": 5.449373968356214,
-      "grad_norm": 0.013433219864964485,
-      "learning_rate": 0.0001999853558233208,
-      "loss": 46.0,
-      "step": 33838
-    },
-    {
-      "epoch": 5.449535005435002,
-      "grad_norm": 0.0031552989967167377,
-      "learning_rate": 0.00019998535495752707,
-      "loss": 46.0,
-      "step": 33839
-    },
-    {
-      "epoch": 5.449696042513789,
-      "grad_norm": 0.006568028125911951,
-      "learning_rate": 0.0001999853540917077,
-      "loss": 46.0,
-      "step": 33840
-    },
-    {
-      "epoch": 5.449857079592576,
-      "grad_norm": 0.020712271332740784,
-      "learning_rate": 0.00019998535322586275,
-      "loss": 46.0,
-      "step": 33841
-    },
-    {
-      "epoch": 5.450018116671363,
-      "grad_norm": 0.0020591234788298607,
-      "learning_rate": 0.0001999853523599922,
-      "loss": 46.0,
-      "step": 33842
-    },
-    {
-      "epoch": 5.450179153750151,
-      "grad_norm": 0.01104546058923006,
-      "learning_rate": 0.0001999853514940961,
-      "loss": 46.0,
-      "step": 33843
-    },
-    {
-      "epoch": 5.450340190828938,
-      "grad_norm": 0.00718472246080637,
-      "learning_rate": 0.00019998535062817435,
-      "loss": 46.0,
-      "step": 33844
-    },
-    {
-      "epoch": 5.450501227907726,
-      "grad_norm": 0.003475119825452566,
-      "learning_rate": 0.00019998534976222707,
-      "loss": 46.0,
-      "step": 33845
-    },
-    {
-      "epoch": 5.450662264986513,
-      "grad_norm": 0.010517102666199207,
-      "learning_rate": 0.00019998534889625416,
-      "loss": 46.0,
-      "step": 33846
-    },
-    {
-      "epoch": 5.4508233020653005,
-      "grad_norm": 0.004338243510574102,
-      "learning_rate": 0.00019998534803025567,
-      "loss": 46.0,
-      "step": 33847
-    },
-    {
-      "epoch": 5.450984339144088,
-      "grad_norm": 0.01050893496721983,
-      "learning_rate": 0.0001999853471642316,
-      "loss": 46.0,
-      "step": 33848
-    },
-    {
-      "epoch": 5.451145376222875,
-      "grad_norm": 0.0058682686649262905,
-      "learning_rate": 0.0001999853462981819,
-      "loss": 46.0,
-      "step": 33849
-    },
-    {
-      "epoch": 5.451306413301663,
-      "grad_norm": 0.02847227267920971,
-      "learning_rate": 0.00019998534543210666,
-      "loss": 46.0,
-      "step": 33850
-    },
-    {
-      "epoch": 5.45146745038045,
-      "grad_norm": 0.0028384150937199593,
-      "learning_rate": 0.00019998534456600582,
-      "loss": 46.0,
-      "step": 33851
-    },
-    {
-      "epoch": 5.451628487459238,
-      "grad_norm": 0.0020966320298612118,
-      "learning_rate": 0.00019998534369987937,
-      "loss": 46.0,
-      "step": 33852
-    },
-    {
-      "epoch": 5.451789524538025,
-      "grad_norm": 0.007990136742591858,
-      "learning_rate": 0.00019998534283372733,
-      "loss": 46.0,
-      "step": 33853
-    },
-    {
-      "epoch": 5.451950561616813,
-      "grad_norm": 0.020022640004754066,
-      "learning_rate": 0.0001999853419675497,
-      "loss": 46.0,
-      "step": 33854
-    },
-    {
-      "epoch": 5.452111598695599,
-      "grad_norm": 0.010539705865085125,
-      "learning_rate": 0.00019998534110134649,
-      "loss": 46.0,
-      "step": 33855
-    },
-    {
-      "epoch": 5.452272635774387,
-      "grad_norm": 0.019215842708945274,
-      "learning_rate": 0.00019998534023511768,
-      "loss": 46.0,
-      "step": 33856
-    },
-    {
-      "epoch": 5.452433672853174,
-      "grad_norm": 0.0017711084801703691,
-      "learning_rate": 0.00019998533936886327,
-      "loss": 46.0,
-      "step": 33857
-    },
-    {
-      "epoch": 5.452594709931962,
-      "grad_norm": 0.003719313768669963,
-      "learning_rate": 0.00019998533850258327,
-      "loss": 46.0,
-      "step": 33858
-    },
-    {
-      "epoch": 5.452755747010749,
-      "grad_norm": 0.006358031183481216,
-      "learning_rate": 0.0001999853376362777,
-      "loss": 46.0,
-      "step": 33859
-    },
-    {
-      "epoch": 5.4529167840895365,
-      "grad_norm": 0.0026636661496013403,
-      "learning_rate": 0.0001999853367699465,
-      "loss": 46.0,
-      "step": 33860
-    },
-    {
-      "epoch": 5.453077821168324,
-      "grad_norm": 0.00944479089230299,
-      "learning_rate": 0.00019998533590358974,
-      "loss": 46.0,
-      "step": 33861
-    },
-    {
-      "epoch": 5.453238858247111,
-      "grad_norm": 0.0041820695623755455,
-      "learning_rate": 0.0001999853350372074,
-      "loss": 46.0,
-      "step": 33862
-    },
-    {
-      "epoch": 5.453399895325899,
-      "grad_norm": 0.0037767766043543816,
-      "learning_rate": 0.00019998533417079942,
-      "loss": 46.0,
-      "step": 33863
-    },
-    {
-      "epoch": 5.453560932404686,
-      "grad_norm": 0.012879302725195885,
-      "learning_rate": 0.0001999853333043659,
-      "loss": 46.0,
-      "step": 33864
-    },
-    {
-      "epoch": 5.453721969483474,
-      "grad_norm": 0.005034186411648989,
-      "learning_rate": 0.00019998533243790678,
-      "loss": 46.0,
-      "step": 33865
-    },
-    {
-      "epoch": 5.453883006562261,
-      "grad_norm": 0.0017000120133161545,
-      "learning_rate": 0.00019998533157142206,
-      "loss": 46.0,
-      "step": 33866
-    },
-    {
-      "epoch": 5.454044043641049,
-      "grad_norm": 0.015321878716349602,
-      "learning_rate": 0.00019998533070491174,
-      "loss": 46.0,
-      "step": 33867
-    },
-    {
-      "epoch": 5.454205080719836,
-      "grad_norm": 0.004623544402420521,
-      "learning_rate": 0.00019998532983837587,
-      "loss": 46.0,
-      "step": 33868
-    },
-    {
-      "epoch": 5.454366117798623,
-      "grad_norm": 0.004284393507987261,
-      "learning_rate": 0.00019998532897181435,
-      "loss": 46.0,
-      "step": 33869
-    },
-    {
-      "epoch": 5.45452715487741,
-      "grad_norm": 0.007372124586254358,
-      "learning_rate": 0.00019998532810522725,
-      "loss": 46.0,
-      "step": 33870
-    },
-    {
-      "epoch": 5.454688191956198,
-      "grad_norm": 0.0013585810083895922,
-      "learning_rate": 0.0001999853272386146,
-      "loss": 46.0,
-      "step": 33871
-    },
-    {
-      "epoch": 5.454849229034985,
-      "grad_norm": 0.011057652533054352,
-      "learning_rate": 0.0001999853263719763,
-      "loss": 46.0,
-      "step": 33872
-    },
-    {
-      "epoch": 5.4550102661137725,
-      "grad_norm": 0.017047543078660965,
-      "learning_rate": 0.00019998532550531247,
-      "loss": 46.0,
-      "step": 33873
-    },
-    {
-      "epoch": 5.45517130319256,
-      "grad_norm": 0.007600443437695503,
-      "learning_rate": 0.000199985324638623,
-      "loss": 46.0,
-      "step": 33874
-    },
-    {
-      "epoch": 5.455332340271347,
-      "grad_norm": 0.0013674166984856129,
-      "learning_rate": 0.00019998532377190796,
-      "loss": 46.0,
-      "step": 33875
-    },
-    {
-      "epoch": 5.455493377350135,
-      "grad_norm": 0.0027655805461108685,
-      "learning_rate": 0.00019998532290516733,
-      "loss": 46.0,
-      "step": 33876
-    },
-    {
-      "epoch": 5.455654414428922,
-      "grad_norm": 0.002250152174383402,
-      "learning_rate": 0.0001999853220384011,
-      "loss": 46.0,
-      "step": 33877
-    },
-    {
-      "epoch": 5.45581545150771,
-      "grad_norm": 0.010605317540466785,
-      "learning_rate": 0.00019998532117160932,
-      "loss": 46.0,
-      "step": 33878
-    },
-    {
-      "epoch": 5.455976488586497,
-      "grad_norm": 0.0034706478472799063,
-      "learning_rate": 0.0001999853203047919,
-      "loss": 46.0,
-      "step": 33879
-    },
-    {
-      "epoch": 5.456137525665285,
-      "grad_norm": 0.009512580931186676,
-      "learning_rate": 0.0001999853194379489,
-      "loss": 46.0,
-      "step": 33880
-    },
-    {
-      "epoch": 5.456298562744072,
-      "grad_norm": 0.006329284515231848,
-      "learning_rate": 0.0001999853185710803,
-      "loss": 46.0,
-      "step": 33881
-    },
-    {
-      "epoch": 5.4564595998228596,
-      "grad_norm": 0.001741435145959258,
-      "learning_rate": 0.0001999853177041861,
-      "loss": 46.0,
-      "step": 33882
-    },
-    {
-      "epoch": 5.456620636901647,
-      "grad_norm": 0.000949910085182637,
-      "learning_rate": 0.00019998531683726635,
-      "loss": 46.0,
-      "step": 33883
-    },
-    {
-      "epoch": 5.456781673980434,
-      "grad_norm": 0.004449079278856516,
-      "learning_rate": 0.000199985315970321,
-      "loss": 46.0,
-      "step": 33884
-    },
-    {
-      "epoch": 5.456942711059221,
-      "grad_norm": 0.004924300126731396,
-      "learning_rate": 0.00019998531510335003,
-      "loss": 46.0,
-      "step": 33885
-    },
-    {
-      "epoch": 5.4571037481380085,
-      "grad_norm": 0.0030333097092807293,
-      "learning_rate": 0.0001999853142363535,
-      "loss": 46.0,
-      "step": 33886
-    },
-    {
-      "epoch": 5.457264785216796,
-      "grad_norm": 0.005165104288607836,
-      "learning_rate": 0.00019998531336933137,
-      "loss": 46.0,
-      "step": 33887
-    },
-    {
-      "epoch": 5.457425822295583,
-      "grad_norm": 0.0022031229455024004,
-      "learning_rate": 0.00019998531250228364,
-      "loss": 46.0,
-      "step": 33888
-    },
-    {
-      "epoch": 5.457586859374371,
-      "grad_norm": 0.00294051063247025,
-      "learning_rate": 0.00019998531163521033,
-      "loss": 46.0,
-      "step": 33889
-    },
-    {
-      "epoch": 5.457747896453158,
-      "grad_norm": 0.0017148663755506277,
-      "learning_rate": 0.00019998531076811143,
-      "loss": 46.0,
-      "step": 33890
-    },
-    {
-      "epoch": 5.457908933531946,
-      "grad_norm": 0.0173055287450552,
-      "learning_rate": 0.00019998530990098692,
-      "loss": 46.0,
-      "step": 33891
-    },
-    {
-      "epoch": 5.458069970610733,
-      "grad_norm": 0.002718935953453183,
-      "learning_rate": 0.00019998530903383684,
-      "loss": 46.0,
-      "step": 33892
-    },
-    {
-      "epoch": 5.458231007689521,
-      "grad_norm": 0.007467819377779961,
-      "learning_rate": 0.00019998530816666115,
-      "loss": 46.0,
-      "step": 33893
-    },
-    {
-      "epoch": 5.458392044768308,
-      "grad_norm": 0.0011812165612354875,
-      "learning_rate": 0.00019998530729945988,
-      "loss": 46.0,
-      "step": 33894
-    },
-    {
-      "epoch": 5.4585530818470955,
-      "grad_norm": 0.00619635172188282,
-      "learning_rate": 0.000199985306432233,
-      "loss": 46.0,
-      "step": 33895
-    },
-    {
-      "epoch": 5.458714118925883,
-      "grad_norm": 0.004404815845191479,
-      "learning_rate": 0.00019998530556498056,
-      "loss": 46.0,
-      "step": 33896
-    },
-    {
-      "epoch": 5.45887515600467,
-      "grad_norm": 0.0033379755914211273,
-      "learning_rate": 0.00019998530469770252,
-      "loss": 46.0,
-      "step": 33897
-    },
-    {
-      "epoch": 5.459036193083458,
-      "grad_norm": 0.002729051746428013,
-      "learning_rate": 0.00019998530383039887,
-      "loss": 46.0,
-      "step": 33898
-    },
-    {
-      "epoch": 5.4591972301622445,
-      "grad_norm": 0.0036032467614859343,
-      "learning_rate": 0.00019998530296306966,
-      "loss": 46.0,
-      "step": 33899
-    },
-    {
-      "epoch": 5.459358267241032,
-      "grad_norm": 0.0013720316346734762,
-      "learning_rate": 0.00019998530209571483,
-      "loss": 46.0,
-      "step": 33900
-    },
-    {
-      "epoch": 5.459519304319819,
-      "grad_norm": 0.004136795178055763,
-      "learning_rate": 0.00019998530122833442,
-      "loss": 46.0,
-      "step": 33901
-    },
-    {
-      "epoch": 5.459680341398607,
-      "grad_norm": 0.002127335174009204,
-      "learning_rate": 0.00019998530036092842,
-      "loss": 46.0,
-      "step": 33902
-    },
-    {
-      "epoch": 5.459841378477394,
-      "grad_norm": 0.003875572932884097,
-      "learning_rate": 0.00019998529949349683,
-      "loss": 46.0,
-      "step": 33903
-    },
-    {
-      "epoch": 5.460002415556182,
-      "grad_norm": 0.0017143464647233486,
-      "learning_rate": 0.00019998529862603963,
-      "loss": 46.0,
-      "step": 33904
-    },
-    {
-      "epoch": 5.460163452634969,
-      "grad_norm": 0.015358680859208107,
-      "learning_rate": 0.00019998529775855687,
-      "loss": 46.0,
-      "step": 33905
-    },
-    {
-      "epoch": 5.460324489713757,
-      "grad_norm": 0.00425465265288949,
-      "learning_rate": 0.0001999852968910485,
-      "loss": 46.0,
-      "step": 33906
-    },
-    {
-      "epoch": 5.460485526792544,
-      "grad_norm": 0.00239172694273293,
-      "learning_rate": 0.00019998529602351456,
-      "loss": 46.0,
-      "step": 33907
-    },
-    {
-      "epoch": 5.4606465638713315,
-      "grad_norm": 0.0067107947543263435,
-      "learning_rate": 0.000199985295155955,
-      "loss": 46.0,
-      "step": 33908
-    },
-    {
-      "epoch": 5.460807600950119,
-      "grad_norm": 0.001255196868441999,
-      "learning_rate": 0.00019998529428836984,
-      "loss": 46.0,
-      "step": 33909
-    },
-    {
-      "epoch": 5.460968638028906,
-      "grad_norm": 0.0015801206463947892,
-      "learning_rate": 0.00019998529342075912,
-      "loss": 46.0,
-      "step": 33910
-    },
-    {
-      "epoch": 5.461129675107694,
-      "grad_norm": 0.014534699730575085,
-      "learning_rate": 0.0001999852925531228,
-      "loss": 46.0,
-      "step": 33911
-    },
-    {
-      "epoch": 5.461290712186481,
-      "grad_norm": 0.0017297585727646947,
-      "learning_rate": 0.0001999852916854609,
-      "loss": 46.0,
-      "step": 33912
-    },
-    {
-      "epoch": 5.461451749265269,
-      "grad_norm": 0.013643634505569935,
-      "learning_rate": 0.00019998529081777336,
-      "loss": 46.0,
-      "step": 33913
-    },
-    {
-      "epoch": 5.461612786344055,
-      "grad_norm": 0.003494001692160964,
-      "learning_rate": 0.0001999852899500603,
-      "loss": 46.0,
-      "step": 33914
-    },
-    {
-      "epoch": 5.461773823422843,
-      "grad_norm": 0.0027977884747087955,
-      "learning_rate": 0.0001999852890823216,
-      "loss": 46.0,
-      "step": 33915
-    },
-    {
-      "epoch": 5.46193486050163,
-      "grad_norm": 0.002195589942857623,
-      "learning_rate": 0.00019998528821455733,
-      "loss": 46.0,
-      "step": 33916
-    },
-    {
-      "epoch": 5.462095897580418,
-      "grad_norm": 0.00214962475001812,
-      "learning_rate": 0.00019998528734676746,
-      "loss": 46.0,
-      "step": 33917
-    },
-    {
-      "epoch": 5.462256934659205,
-      "grad_norm": 0.009241893887519836,
-      "learning_rate": 0.00019998528647895201,
-      "loss": 46.0,
-      "step": 33918
-    },
-    {
-      "epoch": 5.462417971737993,
-      "grad_norm": 0.004307929426431656,
-      "learning_rate": 0.00019998528561111095,
-      "loss": 46.0,
-      "step": 33919
-    },
-    {
-      "epoch": 5.46257900881678,
-      "grad_norm": 0.006896978244185448,
-      "learning_rate": 0.00019998528474324433,
-      "loss": 46.0,
-      "step": 33920
-    },
-    {
-      "epoch": 5.4627400458955675,
-      "grad_norm": 0.0018596044974401593,
-      "learning_rate": 0.00019998528387535206,
-      "loss": 46.0,
-      "step": 33921
-    },
-    {
-      "epoch": 5.462901082974355,
-      "grad_norm": 0.012784102000296116,
-      "learning_rate": 0.00019998528300743424,
-      "loss": 46.0,
-      "step": 33922
-    },
-    {
-      "epoch": 5.463062120053142,
-      "grad_norm": 0.0023449386935681105,
-      "learning_rate": 0.00019998528213949082,
-      "loss": 46.0,
-      "step": 33923
-    },
-    {
-      "epoch": 5.46322315713193,
-      "grad_norm": 0.0060007041320204735,
-      "learning_rate": 0.00019998528127152185,
-      "loss": 46.0,
-      "step": 33924
-    },
-    {
-      "epoch": 5.463384194210717,
-      "grad_norm": 0.01186884194612503,
-      "learning_rate": 0.00019998528040352724,
-      "loss": 46.0,
-      "step": 33925
-    },
-    {
-      "epoch": 5.463545231289505,
-      "grad_norm": 0.00402445113286376,
-      "learning_rate": 0.00019998527953550706,
-      "loss": 46.0,
-      "step": 33926
-    },
-    {
-      "epoch": 5.463706268368292,
-      "grad_norm": 0.001791347865946591,
-      "learning_rate": 0.00019998527866746128,
-      "loss": 46.0,
-      "step": 33927
-    },
-    {
-      "epoch": 5.46386730544708,
-      "grad_norm": 0.00810669269412756,
-      "learning_rate": 0.0001999852777993899,
-      "loss": 46.0,
-      "step": 33928
-    },
-    {
-      "epoch": 5.464028342525866,
-      "grad_norm": 0.0035595502704381943,
-      "learning_rate": 0.00019998527693129294,
-      "loss": 46.0,
-      "step": 33929
-    },
-    {
-      "epoch": 5.464189379604654,
-      "grad_norm": 0.007522111292928457,
-      "learning_rate": 0.0001999852760631704,
-      "loss": 46.0,
-      "step": 33930
-    },
-    {
-      "epoch": 5.464350416683441,
-      "grad_norm": 0.006966003682464361,
-      "learning_rate": 0.00019998527519502222,
-      "loss": 46.0,
-      "step": 33931
-    },
-    {
-      "epoch": 5.464511453762229,
-      "grad_norm": 0.0027638860046863556,
-      "learning_rate": 0.00019998527432684853,
-      "loss": 46.0,
-      "step": 33932
-    },
-    {
-      "epoch": 5.464672490841016,
-      "grad_norm": 0.018949005752801895,
-      "learning_rate": 0.0001999852734586492,
-      "loss": 46.0,
-      "step": 33933
-    },
-    {
-      "epoch": 5.4648335279198035,
-      "grad_norm": 0.0020665456540882587,
-      "learning_rate": 0.0001999852725904243,
-      "loss": 46.0,
-      "step": 33934
-    },
-    {
-      "epoch": 5.464994564998591,
-      "grad_norm": 0.0008323253132402897,
-      "learning_rate": 0.00019998527172217378,
-      "loss": 46.0,
-      "step": 33935
-    },
-    {
-      "epoch": 5.465155602077378,
-      "grad_norm": 0.005588851403445005,
-      "learning_rate": 0.00019998527085389768,
-      "loss": 46.0,
-      "step": 33936
-    },
-    {
-      "epoch": 5.465316639156166,
-      "grad_norm": 0.0023333595599979162,
-      "learning_rate": 0.00019998526998559596,
-      "loss": 46.0,
-      "step": 33937
-    },
-    {
-      "epoch": 5.465477676234953,
-      "grad_norm": 0.0014875035267323256,
-      "learning_rate": 0.0001999852691172687,
-      "loss": 46.0,
-      "step": 33938
-    },
-    {
-      "epoch": 5.465638713313741,
-      "grad_norm": 0.03501440957188606,
-      "learning_rate": 0.00019998526824891583,
-      "loss": 46.0,
-      "step": 33939
-    },
-    {
-      "epoch": 5.465799750392528,
-      "grad_norm": 0.005613334476947784,
-      "learning_rate": 0.00019998526738053735,
-      "loss": 46.0,
-      "step": 33940
-    },
-    {
-      "epoch": 5.465960787471316,
-      "grad_norm": 0.004374946001917124,
-      "learning_rate": 0.0001999852665121333,
-      "loss": 46.0,
-      "step": 33941
-    },
-    {
-      "epoch": 5.466121824550102,
-      "grad_norm": 0.0014744155341759324,
-      "learning_rate": 0.00019998526564370367,
-      "loss": 46.0,
-      "step": 33942
-    },
-    {
-      "epoch": 5.46628286162889,
-      "grad_norm": 0.0032316846773028374,
-      "learning_rate": 0.00019998526477524843,
-      "loss": 46.0,
-      "step": 33943
-    },
-    {
-      "epoch": 5.466443898707677,
-      "grad_norm": 0.009230783209204674,
-      "learning_rate": 0.0001999852639067676,
-      "loss": 46.0,
-      "step": 33944
-    },
-    {
-      "epoch": 5.466604935786465,
-      "grad_norm": 0.003151049604639411,
-      "learning_rate": 0.00019998526303826117,
-      "loss": 46.0,
-      "step": 33945
-    },
-    {
-      "epoch": 5.466765972865252,
-      "grad_norm": 0.0030997558496892452,
-      "learning_rate": 0.00019998526216972917,
-      "loss": 46.0,
-      "step": 33946
-    },
-    {
-      "epoch": 5.4669270099440395,
-      "grad_norm": 0.006346467416733503,
-      "learning_rate": 0.00019998526130117155,
-      "loss": 46.0,
-      "step": 33947
-    },
-    {
-      "epoch": 5.467088047022827,
-      "grad_norm": 0.0038231804501265287,
-      "learning_rate": 0.00019998526043258838,
-      "loss": 46.0,
-      "step": 33948
-    },
-    {
-      "epoch": 5.467249084101614,
-      "grad_norm": 0.013598942197859287,
-      "learning_rate": 0.0001999852595639796,
-      "loss": 46.0,
-      "step": 33949
-    },
-    {
-      "epoch": 5.467410121180402,
-      "grad_norm": 0.0029380833730101585,
-      "learning_rate": 0.00019998525869534525,
-      "loss": 46.0,
-      "step": 33950
-    },
-    {
-      "epoch": 5.467571158259189,
-      "grad_norm": 0.01583113521337509,
-      "learning_rate": 0.00019998525782668526,
-      "loss": 46.0,
-      "step": 33951
-    },
-    {
-      "epoch": 5.467732195337977,
-      "grad_norm": 0.0023203128948807716,
-      "learning_rate": 0.0001999852569579997,
-      "loss": 46.0,
-      "step": 33952
-    },
-    {
-      "epoch": 5.467893232416764,
-      "grad_norm": 0.018242256715893745,
-      "learning_rate": 0.00019998525608928855,
-      "loss": 46.0,
-      "step": 33953
-    },
-    {
-      "epoch": 5.468054269495552,
-      "grad_norm": 0.009462394751608372,
-      "learning_rate": 0.0001999852552205518,
-      "loss": 46.0,
-      "step": 33954
-    },
-    {
-      "epoch": 5.468215306574339,
-      "grad_norm": 0.0072496808134019375,
-      "learning_rate": 0.00019998525435178949,
-      "loss": 46.0,
-      "step": 33955
-    },
-    {
-      "epoch": 5.4683763436531265,
-      "grad_norm": 0.004082279745489359,
-      "learning_rate": 0.00019998525348300156,
-      "loss": 46.0,
-      "step": 33956
-    },
-    {
-      "epoch": 5.468537380731913,
-      "grad_norm": 0.009565218351781368,
-      "learning_rate": 0.00019998525261418805,
-      "loss": 46.0,
-      "step": 33957
-    },
-    {
-      "epoch": 5.4686984178107005,
-      "grad_norm": 0.0022864476777613163,
-      "learning_rate": 0.00019998525174534898,
-      "loss": 46.0,
-      "step": 33958
-    },
-    {
-      "epoch": 5.468859454889488,
-      "grad_norm": 0.002576803555712104,
-      "learning_rate": 0.00019998525087648426,
-      "loss": 46.0,
-      "step": 33959
-    },
-    {
-      "epoch": 5.469020491968275,
-      "grad_norm": 0.004243740811944008,
-      "learning_rate": 0.00019998525000759396,
-      "loss": 46.0,
-      "step": 33960
-    },
-    {
-      "epoch": 5.469181529047063,
-      "grad_norm": 0.001701556844636798,
-      "learning_rate": 0.00019998524913867808,
-      "loss": 46.0,
-      "step": 33961
-    },
-    {
-      "epoch": 5.46934256612585,
-      "grad_norm": 0.0012116024736315012,
-      "learning_rate": 0.00019998524826973663,
-      "loss": 46.0,
-      "step": 33962
-    },
-    {
-      "epoch": 5.469503603204638,
-      "grad_norm": 0.0028388125356286764,
-      "learning_rate": 0.00019998524740076957,
-      "loss": 46.0,
-      "step": 33963
-    },
-    {
-      "epoch": 5.469664640283425,
-      "grad_norm": 0.006310044787824154,
-      "learning_rate": 0.00019998524653177692,
-      "loss": 46.0,
-      "step": 33964
-    },
-    {
-      "epoch": 5.469825677362213,
-      "grad_norm": 0.001129554002545774,
-      "learning_rate": 0.00019998524566275868,
-      "loss": 46.0,
-      "step": 33965
-    },
-    {
-      "epoch": 5.469986714441,
-      "grad_norm": 0.004442868288606405,
-      "learning_rate": 0.00019998524479371486,
-      "loss": 46.0,
-      "step": 33966
-    },
-    {
-      "epoch": 5.470147751519788,
-      "grad_norm": 0.002943875268101692,
-      "learning_rate": 0.00019998524392464542,
-      "loss": 46.0,
-      "step": 33967
-    },
-    {
-      "epoch": 5.470308788598575,
-      "grad_norm": 0.012347236275672913,
-      "learning_rate": 0.0001999852430555504,
-      "loss": 46.0,
-      "step": 33968
-    },
-    {
-      "epoch": 5.4704698256773625,
-      "grad_norm": 0.006594682112336159,
-      "learning_rate": 0.0001999852421864298,
-      "loss": 46.0,
-      "step": 33969
-    },
-    {
-      "epoch": 5.47063086275615,
-      "grad_norm": 0.010824074037373066,
-      "learning_rate": 0.0001999852413172836,
-      "loss": 46.0,
-      "step": 33970
-    },
-    {
-      "epoch": 5.470791899834937,
-      "grad_norm": 0.013155831955373287,
-      "learning_rate": 0.00019998524044811182,
-      "loss": 46.0,
-      "step": 33971
-    },
-    {
-      "epoch": 5.470952936913724,
-      "grad_norm": 0.004117035306990147,
-      "learning_rate": 0.00019998523957891445,
-      "loss": 46.0,
-      "step": 33972
-    },
-    {
-      "epoch": 5.471113973992511,
-      "grad_norm": 0.01561188604682684,
-      "learning_rate": 0.0001999852387096915,
-      "loss": 46.0,
-      "step": 33973
-    },
-    {
-      "epoch": 5.471275011071299,
-      "grad_norm": 0.0023450301960110664,
-      "learning_rate": 0.00019998523784044291,
-      "loss": 46.0,
-      "step": 33974
-    },
-    {
-      "epoch": 5.471436048150086,
-      "grad_norm": 0.013035872019827366,
-      "learning_rate": 0.00019998523697116878,
-      "loss": 46.0,
-      "step": 33975
-    },
-    {
-      "epoch": 5.471597085228874,
-      "grad_norm": 0.007042041979730129,
-      "learning_rate": 0.00019998523610186903,
-      "loss": 46.0,
-      "step": 33976
-    },
-    {
-      "epoch": 5.471758122307661,
-      "grad_norm": 0.004770055413246155,
-      "learning_rate": 0.0001999852352325437,
-      "loss": 46.0,
-      "step": 33977
-    },
-    {
-      "epoch": 5.471919159386449,
-      "grad_norm": 0.0034993013832718134,
-      "learning_rate": 0.00019998523436319277,
-      "loss": 46.0,
-      "step": 33978
-    },
-    {
-      "epoch": 5.472080196465236,
-      "grad_norm": 0.0034382238518446684,
-      "learning_rate": 0.00019998523349381623,
-      "loss": 46.0,
-      "step": 33979
-    },
-    {
-      "epoch": 5.472241233544024,
-      "grad_norm": 0.0015040583675727248,
-      "learning_rate": 0.00019998523262441413,
-      "loss": 46.0,
-      "step": 33980
-    },
-    {
-      "epoch": 5.472402270622811,
-      "grad_norm": 0.010898053646087646,
-      "learning_rate": 0.00019998523175498645,
-      "loss": 46.0,
-      "step": 33981
-    },
-    {
-      "epoch": 5.4725633077015985,
-      "grad_norm": 0.0020360820926725864,
-      "learning_rate": 0.00019998523088553315,
-      "loss": 46.0,
-      "step": 33982
-    },
-    {
-      "epoch": 5.472724344780386,
-      "grad_norm": 0.0016121412627398968,
-      "learning_rate": 0.0001999852300160543,
-      "loss": 46.0,
-      "step": 33983
-    },
-    {
-      "epoch": 5.472885381859173,
-      "grad_norm": 0.0034480844624340534,
-      "learning_rate": 0.0001999852291465498,
-      "loss": 46.0,
-      "step": 33984
-    },
-    {
-      "epoch": 5.473046418937961,
-      "grad_norm": 0.00642159441486001,
-      "learning_rate": 0.00019998522827701973,
-      "loss": 46.0,
-      "step": 33985
-    },
-    {
-      "epoch": 5.473207456016748,
-      "grad_norm": 0.004151217173784971,
-      "learning_rate": 0.0001999852274074641,
-      "loss": 46.0,
-      "step": 33986
-    },
-    {
-      "epoch": 5.473368493095535,
-      "grad_norm": 0.005182954482734203,
-      "learning_rate": 0.00019998522653788284,
-      "loss": 46.0,
-      "step": 33987
-    },
-    {
-      "epoch": 5.473529530174322,
-      "grad_norm": 0.0009014858515001833,
-      "learning_rate": 0.00019998522566827602,
-      "loss": 46.0,
-      "step": 33988
-    },
-    {
-      "epoch": 5.47369056725311,
-      "grad_norm": 0.014669469557702541,
-      "learning_rate": 0.00019998522479864358,
-      "loss": 46.0,
-      "step": 33989
-    },
-    {
-      "epoch": 5.473851604331897,
-      "grad_norm": 0.011654900386929512,
-      "learning_rate": 0.0001999852239289856,
-      "loss": 46.0,
-      "step": 33990
-    },
-    {
-      "epoch": 5.474012641410685,
-      "grad_norm": 0.002652890747413039,
-      "learning_rate": 0.00019998522305930195,
-      "loss": 46.0,
-      "step": 33991
-    },
-    {
-      "epoch": 5.474173678489472,
-      "grad_norm": 0.0024331207387149334,
-      "learning_rate": 0.00019998522218959278,
-      "loss": 46.0,
-      "step": 33992
-    },
-    {
-      "epoch": 5.47433471556826,
-      "grad_norm": 0.0052316877990961075,
-      "learning_rate": 0.00019998522131985794,
-      "loss": 46.0,
-      "step": 33993
-    },
-    {
-      "epoch": 5.474495752647047,
-      "grad_norm": 0.003056933870539069,
-      "learning_rate": 0.0001999852204500976,
-      "loss": 46.0,
-      "step": 33994
-    },
-    {
-      "epoch": 5.4746567897258345,
-      "grad_norm": 0.0026589741464704275,
-      "learning_rate": 0.00019998521958031163,
-      "loss": 46.0,
-      "step": 33995
-    },
-    {
-      "epoch": 5.474817826804622,
-      "grad_norm": 0.0057871281169354916,
-      "learning_rate": 0.00019998521871050003,
-      "loss": 46.0,
-      "step": 33996
-    },
-    {
-      "epoch": 5.474978863883409,
-      "grad_norm": 0.00477339792996645,
-      "learning_rate": 0.0001999852178406629,
-      "loss": 46.0,
-      "step": 33997
-    },
-    {
-      "epoch": 5.475139900962197,
-      "grad_norm": 0.005761266686022282,
-      "learning_rate": 0.00019998521697080015,
-      "loss": 46.0,
-      "step": 33998
-    },
-    {
-      "epoch": 5.475300938040984,
-      "grad_norm": 0.005985742434859276,
-      "learning_rate": 0.00019998521610091181,
-      "loss": 46.0,
-      "step": 33999
-    },
-    {
-      "epoch": 5.475461975119772,
-      "grad_norm": 0.003352626459673047,
-      "learning_rate": 0.00019998521523099786,
-      "loss": 46.0,
-      "step": 34000
-    },
-    {
-      "epoch": 5.475623012198559,
-      "grad_norm": 0.0029440601356327534,
-      "learning_rate": 0.00019998521436105838,
-      "loss": 46.0,
-      "step": 34001
-    },
-    {
-      "epoch": 5.475784049277346,
-      "grad_norm": 0.0032684379257261753,
-      "learning_rate": 0.00019998521349109326,
-      "loss": 46.0,
-      "step": 34002
-    },
-    {
-      "epoch": 5.475945086356133,
-      "grad_norm": 0.0021878937259316444,
-      "learning_rate": 0.00019998521262110255,
-      "loss": 46.0,
-      "step": 34003
-    },
-    {
-      "epoch": 5.476106123434921,
-      "grad_norm": 0.0016625227872282267,
-      "learning_rate": 0.00019998521175108627,
-      "loss": 46.0,
-      "step": 34004
-    },
-    {
-      "epoch": 5.476267160513708,
-      "grad_norm": 0.007223007269203663,
-      "learning_rate": 0.00019998521088104436,
-      "loss": 46.0,
-      "step": 34005
-    },
-    {
-      "epoch": 5.4764281975924956,
-      "grad_norm": 0.008779977448284626,
-      "learning_rate": 0.0001999852100109769,
-      "loss": 46.0,
-      "step": 34006
-    },
-    {
-      "epoch": 5.476589234671283,
-      "grad_norm": 0.004818316549062729,
-      "learning_rate": 0.00019998520914088383,
-      "loss": 46.0,
-      "step": 34007
-    },
-    {
-      "epoch": 5.4767502717500705,
-      "grad_norm": 0.0036331687588244677,
-      "learning_rate": 0.00019998520827076518,
-      "loss": 46.0,
-      "step": 34008
-    },
-    {
-      "epoch": 5.476911308828858,
-      "grad_norm": 0.007848354056477547,
-      "learning_rate": 0.00019998520740062092,
-      "loss": 46.0,
-      "step": 34009
-    },
-    {
-      "epoch": 5.477072345907645,
-      "grad_norm": 0.0012876796536147594,
-      "learning_rate": 0.0001999852065304511,
-      "loss": 46.0,
-      "step": 34010
-    },
-    {
-      "epoch": 5.477233382986433,
-      "grad_norm": 0.0017034618649631739,
-      "learning_rate": 0.00019998520566025566,
-      "loss": 46.0,
-      "step": 34011
-    },
-    {
-      "epoch": 5.47739442006522,
-      "grad_norm": 0.008430598303675652,
-      "learning_rate": 0.0001999852047900346,
-      "loss": 46.0,
-      "step": 34012
-    },
-    {
-      "epoch": 5.477555457144008,
-      "grad_norm": 0.0037775845266878605,
-      "learning_rate": 0.00019998520391978803,
-      "loss": 46.0,
-      "step": 34013
-    },
-    {
-      "epoch": 5.477716494222795,
-      "grad_norm": 0.0053141117095947266,
-      "learning_rate": 0.00019998520304951583,
-      "loss": 46.0,
-      "step": 34014
-    },
-    {
-      "epoch": 5.477877531301582,
-      "grad_norm": 0.016840746626257896,
-      "learning_rate": 0.00019998520217921805,
-      "loss": 46.0,
-      "step": 34015
-    },
-    {
-      "epoch": 5.478038568380369,
-      "grad_norm": 0.0034186423290520906,
-      "learning_rate": 0.00019998520130889462,
-      "loss": 46.0,
-      "step": 34016
-    },
-    {
-      "epoch": 5.478199605459157,
-      "grad_norm": 0.0020906454883515835,
-      "learning_rate": 0.00019998520043854566,
-      "loss": 46.0,
-      "step": 34017
-    },
-    {
-      "epoch": 5.478360642537944,
-      "grad_norm": 0.011682221665978432,
-      "learning_rate": 0.0001999851995681711,
-      "loss": 46.0,
-      "step": 34018
-    },
-    {
-      "epoch": 5.4785216796167315,
-      "grad_norm": 0.006353188306093216,
-      "learning_rate": 0.00019998519869777095,
-      "loss": 46.0,
-      "step": 34019
-    },
-    {
-      "epoch": 5.478682716695519,
-      "grad_norm": 0.013782745227217674,
-      "learning_rate": 0.00019998519782734518,
-      "loss": 46.0,
-      "step": 34020
-    },
-    {
-      "epoch": 5.478843753774306,
-      "grad_norm": 0.005086518824100494,
-      "learning_rate": 0.00019998519695689385,
-      "loss": 46.0,
-      "step": 34021
-    },
-    {
-      "epoch": 5.479004790853094,
-      "grad_norm": 0.002089346293359995,
-      "learning_rate": 0.0001999851960864169,
-      "loss": 46.0,
-      "step": 34022
-    },
-    {
-      "epoch": 5.479165827931881,
-      "grad_norm": 0.010834196582436562,
-      "learning_rate": 0.0001999851952159144,
-      "loss": 46.0,
-      "step": 34023
-    },
-    {
-      "epoch": 5.479326865010669,
-      "grad_norm": 0.006796369329094887,
-      "learning_rate": 0.00019998519434538626,
-      "loss": 46.0,
-      "step": 34024
-    },
-    {
-      "epoch": 5.479487902089456,
-      "grad_norm": 0.0034908675588667393,
-      "learning_rate": 0.00019998519347483255,
-      "loss": 46.0,
-      "step": 34025
-    },
-    {
-      "epoch": 5.479648939168244,
-      "grad_norm": 0.002998356707394123,
-      "learning_rate": 0.00019998519260425326,
-      "loss": 46.0,
-      "step": 34026
-    },
-    {
-      "epoch": 5.479809976247031,
-      "grad_norm": 0.0012734472984448075,
-      "learning_rate": 0.00019998519173364837,
-      "loss": 46.0,
-      "step": 34027
-    },
-    {
-      "epoch": 5.479971013325819,
-      "grad_norm": 0.016380852088332176,
-      "learning_rate": 0.0001999851908630179,
-      "loss": 46.0,
-      "step": 34028
-    },
-    {
-      "epoch": 5.480132050404606,
-      "grad_norm": 0.0009933009278029203,
-      "learning_rate": 0.00019998518999236184,
-      "loss": 46.0,
-      "step": 34029
-    },
-    {
-      "epoch": 5.480293087483393,
-      "grad_norm": 0.00865671131759882,
-      "learning_rate": 0.0001999851891216802,
-      "loss": 46.0,
-      "step": 34030
-    },
-    {
-      "epoch": 5.48045412456218,
-      "grad_norm": 0.002906026318669319,
-      "learning_rate": 0.00019998518825097293,
-      "loss": 46.0,
-      "step": 34031
-    },
-    {
-      "epoch": 5.4806151616409675,
-      "grad_norm": 0.004925843328237534,
-      "learning_rate": 0.00019998518738024009,
-      "loss": 46.0,
-      "step": 34032
-    },
-    {
-      "epoch": 5.480776198719755,
-      "grad_norm": 0.0007878330652602017,
-      "learning_rate": 0.00019998518650948165,
-      "loss": 46.0,
-      "step": 34033
-    },
-    {
-      "epoch": 5.480937235798542,
-      "grad_norm": 0.009944569319486618,
-      "learning_rate": 0.0001999851856386976,
-      "loss": 46.0,
-      "step": 34034
-    },
-    {
-      "epoch": 5.48109827287733,
-      "grad_norm": 0.0035611798521131277,
-      "learning_rate": 0.00019998518476788802,
-      "loss": 46.0,
-      "step": 34035
-    },
-    {
-      "epoch": 5.481259309956117,
-      "grad_norm": 0.0036387438885867596,
-      "learning_rate": 0.0001999851838970528,
-      "loss": 46.0,
-      "step": 34036
-    },
-    {
-      "epoch": 5.481420347034905,
-      "grad_norm": 0.002845211187377572,
-      "learning_rate": 0.00019998518302619198,
-      "loss": 46.0,
-      "step": 34037
-    },
-    {
-      "epoch": 5.481581384113692,
-      "grad_norm": 0.012023522518575191,
-      "learning_rate": 0.00019998518215530559,
-      "loss": 46.0,
-      "step": 34038
-    },
-    {
-      "epoch": 5.48174242119248,
-      "grad_norm": 0.005608272273093462,
-      "learning_rate": 0.00019998518128439363,
-      "loss": 46.0,
-      "step": 34039
-    },
-    {
-      "epoch": 5.481903458271267,
-      "grad_norm": 0.0010498077608644962,
-      "learning_rate": 0.00019998518041345603,
-      "loss": 46.0,
-      "step": 34040
-    },
-    {
-      "epoch": 5.482064495350055,
-      "grad_norm": 0.0070892926305532455,
-      "learning_rate": 0.00019998517954249287,
-      "loss": 46.0,
-      "step": 34041
-    },
-    {
-      "epoch": 5.482225532428842,
-      "grad_norm": 0.008261364884674549,
-      "learning_rate": 0.00019998517867150412,
-      "loss": 46.0,
-      "step": 34042
-    },
-    {
-      "epoch": 5.4823865695076295,
-      "grad_norm": 0.002518248511478305,
-      "learning_rate": 0.0001999851778004898,
-      "loss": 46.0,
-      "step": 34043
-    },
-    {
-      "epoch": 5.482547606586417,
-      "grad_norm": 0.0017912589246407151,
-      "learning_rate": 0.00019998517692944984,
-      "loss": 46.0,
-      "step": 34044
-    },
-    {
-      "epoch": 5.4827086436652035,
-      "grad_norm": 0.005765694193542004,
-      "learning_rate": 0.0001999851760583843,
-      "loss": 46.0,
-      "step": 34045
-    },
-    {
-      "epoch": 5.482869680743991,
-      "grad_norm": 0.0038748509250581264,
-      "learning_rate": 0.0001999851751872932,
-      "loss": 46.0,
-      "step": 34046
-    },
-    {
-      "epoch": 5.483030717822778,
-      "grad_norm": 0.0016907951794564724,
-      "learning_rate": 0.0001999851743161765,
-      "loss": 46.0,
-      "step": 34047
-    },
-    {
-      "epoch": 5.483191754901566,
-      "grad_norm": 0.0027127834036946297,
-      "learning_rate": 0.0001999851734450342,
-      "loss": 46.0,
-      "step": 34048
-    },
-    {
-      "epoch": 5.483352791980353,
-      "grad_norm": 0.0026182737201452255,
-      "learning_rate": 0.00019998517257386632,
-      "loss": 46.0,
-      "step": 34049
-    },
-    {
-      "epoch": 5.483513829059141,
-      "grad_norm": 0.004523037932813168,
-      "learning_rate": 0.00019998517170267282,
-      "loss": 46.0,
-      "step": 34050
-    },
-    {
-      "epoch": 5.483674866137928,
-      "grad_norm": 0.005244947038590908,
-      "learning_rate": 0.00019998517083145374,
-      "loss": 46.0,
-      "step": 34051
-    },
-    {
-      "epoch": 5.483835903216716,
-      "grad_norm": 0.001922972034662962,
-      "learning_rate": 0.0001999851699602091,
-      "loss": 46.0,
-      "step": 34052
-    },
-    {
-      "epoch": 5.483996940295503,
-      "grad_norm": 0.0017984440783038735,
-      "learning_rate": 0.00019998516908893883,
-      "loss": 46.0,
-      "step": 34053
-    },
-    {
-      "epoch": 5.484157977374291,
-      "grad_norm": 0.0021901780273765326,
-      "learning_rate": 0.00019998516821764296,
-      "loss": 46.0,
-      "step": 34054
-    },
-    {
-      "epoch": 5.484319014453078,
-      "grad_norm": 0.010198758915066719,
-      "learning_rate": 0.00019998516734632155,
-      "loss": 46.0,
-      "step": 34055
-    },
-    {
-      "epoch": 5.4844800515318655,
-      "grad_norm": 0.0030106448102742434,
-      "learning_rate": 0.00019998516647497453,
-      "loss": 46.0,
-      "step": 34056
-    },
-    {
-      "epoch": 5.484641088610653,
-      "grad_norm": 0.00413285568356514,
-      "learning_rate": 0.0001999851656036019,
-      "loss": 46.0,
-      "step": 34057
-    },
-    {
-      "epoch": 5.48480212568944,
-      "grad_norm": 0.003924929071217775,
-      "learning_rate": 0.0001999851647322037,
-      "loss": 46.0,
-      "step": 34058
-    },
-    {
-      "epoch": 5.484963162768228,
-      "grad_norm": 0.006923154927790165,
-      "learning_rate": 0.0001999851638607799,
-      "loss": 46.0,
-      "step": 34059
-    },
-    {
-      "epoch": 5.485124199847014,
-      "grad_norm": 0.0038515881169587374,
-      "learning_rate": 0.00019998516298933046,
-      "loss": 46.0,
-      "step": 34060
-    },
-    {
-      "epoch": 5.485285236925802,
-      "grad_norm": 0.0027746856212615967,
-      "learning_rate": 0.0001999851621178555,
-      "loss": 46.0,
-      "step": 34061
-    },
-    {
-      "epoch": 5.485446274004589,
-      "grad_norm": 0.01583748124539852,
-      "learning_rate": 0.00019998516124635494,
-      "loss": 46.0,
-      "step": 34062
-    },
-    {
-      "epoch": 5.485607311083377,
-      "grad_norm": 0.007230414543300867,
-      "learning_rate": 0.00019998516037482875,
-      "loss": 46.0,
-      "step": 34063
-    },
-    {
-      "epoch": 5.485768348162164,
-      "grad_norm": 0.006358353421092033,
-      "learning_rate": 0.000199985159503277,
-      "loss": 46.0,
-      "step": 34064
-    },
-    {
-      "epoch": 5.485929385240952,
-      "grad_norm": 0.008169732056558132,
-      "learning_rate": 0.00019998515863169964,
-      "loss": 46.0,
-      "step": 34065
-    },
-    {
-      "epoch": 5.486090422319739,
-      "grad_norm": 0.0013715194072574377,
-      "learning_rate": 0.00019998515776009672,
-      "loss": 46.0,
-      "step": 34066
-    },
-    {
-      "epoch": 5.4862514593985265,
-      "grad_norm": 0.001878093578852713,
-      "learning_rate": 0.00019998515688846816,
-      "loss": 46.0,
-      "step": 34067
-    },
-    {
-      "epoch": 5.486412496477314,
-      "grad_norm": 0.0031343852169811726,
-      "learning_rate": 0.00019998515601681407,
-      "loss": 46.0,
-      "step": 34068
-    },
-    {
-      "epoch": 5.4865735335561014,
-      "grad_norm": 0.004168075043708086,
-      "learning_rate": 0.00019998515514513433,
-      "loss": 46.0,
-      "step": 34069
-    },
-    {
-      "epoch": 5.486734570634889,
-      "grad_norm": 0.0060966731980443,
-      "learning_rate": 0.00019998515427342906,
-      "loss": 46.0,
-      "step": 34070
-    },
-    {
-      "epoch": 5.486895607713676,
-      "grad_norm": 0.013217343017458916,
-      "learning_rate": 0.00019998515340169815,
-      "loss": 46.0,
-      "step": 34071
-    },
-    {
-      "epoch": 5.487056644792464,
-      "grad_norm": 0.003502884414047003,
-      "learning_rate": 0.00019998515252994166,
-      "loss": 46.0,
-      "step": 34072
-    },
-    {
-      "epoch": 5.487217681871251,
-      "grad_norm": 0.0014455786440521479,
-      "learning_rate": 0.0001999851516581596,
-      "loss": 46.0,
-      "step": 34073
-    },
-    {
-      "epoch": 5.487378718950039,
-      "grad_norm": 0.00547865591943264,
-      "learning_rate": 0.00019998515078635193,
-      "loss": 46.0,
-      "step": 34074
-    },
-    {
-      "epoch": 5.487539756028825,
-      "grad_norm": 0.014122439548373222,
-      "learning_rate": 0.00019998514991451867,
-      "loss": 46.0,
-      "step": 34075
-    },
-    {
-      "epoch": 5.487700793107613,
-      "grad_norm": 0.0014552170177921653,
-      "learning_rate": 0.00019998514904265982,
-      "loss": 46.0,
-      "step": 34076
-    },
-    {
-      "epoch": 5.4878618301864,
-      "grad_norm": 0.0008031705510802567,
-      "learning_rate": 0.0001999851481707754,
-      "loss": 46.0,
-      "step": 34077
-    },
-    {
-      "epoch": 5.488022867265188,
-      "grad_norm": 0.000917566881980747,
-      "learning_rate": 0.00019998514729886532,
-      "loss": 46.0,
-      "step": 34078
-    },
-    {
-      "epoch": 5.488183904343975,
-      "grad_norm": 0.006410039961338043,
-      "learning_rate": 0.0001999851464269297,
-      "loss": 46.0,
-      "step": 34079
-    },
-    {
-      "epoch": 5.4883449414227625,
-      "grad_norm": 0.012301741167902946,
-      "learning_rate": 0.0001999851455549685,
-      "loss": 46.0,
-      "step": 34080
-    },
-    {
-      "epoch": 5.48850597850155,
-      "grad_norm": 0.0034094585571438074,
-      "learning_rate": 0.00019998514468298168,
-      "loss": 46.0,
-      "step": 34081
-    },
-    {
-      "epoch": 5.488667015580337,
-      "grad_norm": 0.005187683273106813,
-      "learning_rate": 0.0001999851438109693,
-      "loss": 46.0,
-      "step": 34082
-    },
-    {
-      "epoch": 5.488828052659125,
-      "grad_norm": 0.010926825925707817,
-      "learning_rate": 0.00019998514293893132,
-      "loss": 46.0,
-      "step": 34083
-    },
-    {
-      "epoch": 5.488989089737912,
-      "grad_norm": 0.0028332334477454424,
-      "learning_rate": 0.00019998514206686773,
-      "loss": 46.0,
-      "step": 34084
-    },
-    {
-      "epoch": 5.4891501268167,
-      "grad_norm": 0.005750021431595087,
-      "learning_rate": 0.00019998514119477857,
-      "loss": 46.0,
-      "step": 34085
-    },
-    {
-      "epoch": 5.489311163895487,
-      "grad_norm": 0.002164200646802783,
-      "learning_rate": 0.0001999851403226638,
-      "loss": 46.0,
-      "step": 34086
-    },
-    {
-      "epoch": 5.489472200974275,
-      "grad_norm": 0.015085523016750813,
-      "learning_rate": 0.00019998513945052344,
-      "loss": 46.0,
-      "step": 34087
-    },
-    {
-      "epoch": 5.489633238053061,
-      "grad_norm": 0.0060419607907533646,
-      "learning_rate": 0.0001999851385783575,
-      "loss": 46.0,
-      "step": 34088
-    },
-    {
-      "epoch": 5.489794275131849,
-      "grad_norm": 0.0045350040309131145,
-      "learning_rate": 0.00019998513770616599,
-      "loss": 46.0,
-      "step": 34089
-    },
-    {
-      "epoch": 5.489955312210636,
-      "grad_norm": 0.005301353987306356,
-      "learning_rate": 0.00019998513683394884,
-      "loss": 46.0,
-      "step": 34090
-    },
-    {
-      "epoch": 5.490116349289424,
-      "grad_norm": 0.0031123608350753784,
-      "learning_rate": 0.0001999851359617061,
-      "loss": 46.0,
-      "step": 34091
-    },
-    {
-      "epoch": 5.490277386368211,
-      "grad_norm": 0.013791980221867561,
-      "learning_rate": 0.0001999851350894378,
-      "loss": 46.0,
-      "step": 34092
-    },
-    {
-      "epoch": 5.4904384234469985,
-      "grad_norm": 0.0007826744113117456,
-      "learning_rate": 0.00019998513421714393,
-      "loss": 46.0,
-      "step": 34093
-    },
-    {
-      "epoch": 5.490599460525786,
-      "grad_norm": 0.004682152532041073,
-      "learning_rate": 0.00019998513334482443,
-      "loss": 46.0,
-      "step": 34094
-    },
-    {
-      "epoch": 5.490760497604573,
-      "grad_norm": 0.0020901490934193134,
-      "learning_rate": 0.00019998513247247935,
-      "loss": 46.0,
-      "step": 34095
-    },
-    {
-      "epoch": 5.490921534683361,
-      "grad_norm": 0.005975193809717894,
-      "learning_rate": 0.00019998513160010868,
-      "loss": 46.0,
-      "step": 34096
-    },
-    {
-      "epoch": 5.491082571762148,
-      "grad_norm": 0.005556042771786451,
-      "learning_rate": 0.00019998513072771245,
-      "loss": 46.0,
-      "step": 34097
-    },
-    {
-      "epoch": 5.491243608840936,
-      "grad_norm": 0.005330977495759726,
-      "learning_rate": 0.00019998512985529057,
-      "loss": 46.0,
-      "step": 34098
-    },
-    {
-      "epoch": 5.491404645919723,
-      "grad_norm": 0.023547105491161346,
-      "learning_rate": 0.00019998512898284311,
-      "loss": 46.0,
-      "step": 34099
-    },
-    {
-      "epoch": 5.491565682998511,
-      "grad_norm": 0.0007415079162456095,
-      "learning_rate": 0.0001999851281103701,
-      "loss": 46.0,
-      "step": 34100
-    },
-    {
-      "epoch": 5.491726720077298,
-      "grad_norm": 0.007302392739802599,
-      "learning_rate": 0.00019998512723787146,
-      "loss": 46.0,
-      "step": 34101
-    },
-    {
-      "epoch": 5.491887757156086,
-      "grad_norm": 0.001321967109106481,
-      "learning_rate": 0.00019998512636534724,
-      "loss": 46.0,
-      "step": 34102
-    },
-    {
-      "epoch": 5.492048794234872,
-      "grad_norm": 0.005085607059299946,
-      "learning_rate": 0.00019998512549279743,
-      "loss": 46.0,
-      "step": 34103
-    },
-    {
-      "epoch": 5.49220983131366,
-      "grad_norm": 0.0017957368399947882,
-      "learning_rate": 0.00019998512462022204,
-      "loss": 46.0,
-      "step": 34104
-    },
-    {
-      "epoch": 5.492370868392447,
-      "grad_norm": 0.007057764567434788,
-      "learning_rate": 0.00019998512374762103,
-      "loss": 46.0,
-      "step": 34105
-    },
-    {
-      "epoch": 5.4925319054712345,
-      "grad_norm": 0.0038359789177775383,
-      "learning_rate": 0.00019998512287499446,
-      "loss": 46.0,
-      "step": 34106
-    },
-    {
-      "epoch": 5.492692942550022,
-      "grad_norm": 0.015206251293420792,
-      "learning_rate": 0.00019998512200234227,
-      "loss": 46.0,
-      "step": 34107
-    },
-    {
-      "epoch": 5.492853979628809,
-      "grad_norm": 0.0011466221185401082,
-      "learning_rate": 0.00019998512112966453,
-      "loss": 46.0,
-      "step": 34108
-    },
-    {
-      "epoch": 5.493015016707597,
-      "grad_norm": 0.01089741475880146,
-      "learning_rate": 0.00019998512025696117,
-      "loss": 46.0,
-      "step": 34109
-    },
-    {
-      "epoch": 5.493176053786384,
-      "grad_norm": 0.004488585516810417,
-      "learning_rate": 0.0001999851193842322,
-      "loss": 46.0,
-      "step": 34110
-    },
-    {
-      "epoch": 5.493337090865172,
-      "grad_norm": 0.006881280802190304,
-      "learning_rate": 0.00019998511851147767,
-      "loss": 46.0,
-      "step": 34111
-    },
-    {
-      "epoch": 5.493498127943959,
-      "grad_norm": 0.003830853383988142,
-      "learning_rate": 0.00019998511763869755,
-      "loss": 46.0,
-      "step": 34112
-    },
-    {
-      "epoch": 5.493659165022747,
-      "grad_norm": 0.00198124791495502,
-      "learning_rate": 0.00019998511676589184,
-      "loss": 46.0,
-      "step": 34113
-    },
-    {
-      "epoch": 5.493820202101534,
-      "grad_norm": 0.010118413716554642,
-      "learning_rate": 0.00019998511589306052,
-      "loss": 46.0,
-      "step": 34114
-    },
-    {
-      "epoch": 5.4939812391803216,
-      "grad_norm": 0.00990296807140112,
-      "learning_rate": 0.00019998511502020364,
-      "loss": 46.0,
-      "step": 34115
-    },
-    {
-      "epoch": 5.494142276259109,
-      "grad_norm": 0.0018488084897398949,
-      "learning_rate": 0.00019998511414732112,
-      "loss": 46.0,
-      "step": 34116
-    },
-    {
-      "epoch": 5.4943033133378965,
-      "grad_norm": 0.004723068792372942,
-      "learning_rate": 0.00019998511327441306,
-      "loss": 46.0,
-      "step": 34117
-    },
-    {
-      "epoch": 5.494464350416683,
-      "grad_norm": 0.014449436217546463,
-      "learning_rate": 0.00019998511240147936,
-      "loss": 46.0,
-      "step": 34118
-    },
-    {
-      "epoch": 5.4946253874954705,
-      "grad_norm": 0.0053884778171777725,
-      "learning_rate": 0.0001999851115285201,
-      "loss": 46.0,
-      "step": 34119
-    },
-    {
-      "epoch": 5.494786424574258,
-      "grad_norm": 0.006324960384517908,
-      "learning_rate": 0.00019998511065553524,
-      "loss": 46.0,
-      "step": 34120
-    },
-    {
-      "epoch": 5.494947461653045,
-      "grad_norm": 0.009068763814866543,
-      "learning_rate": 0.0001999851097825248,
-      "loss": 46.0,
-      "step": 34121
-    },
-    {
-      "epoch": 5.495108498731833,
-      "grad_norm": 0.002731488784775138,
-      "learning_rate": 0.00019998510890948876,
-      "loss": 46.0,
-      "step": 34122
-    },
-    {
-      "epoch": 5.49526953581062,
-      "grad_norm": 0.004338014405220747,
-      "learning_rate": 0.00019998510803642713,
-      "loss": 46.0,
-      "step": 34123
-    },
-    {
-      "epoch": 5.495430572889408,
-      "grad_norm": 0.003702192334458232,
-      "learning_rate": 0.00019998510716333988,
-      "loss": 46.0,
-      "step": 34124
-    },
-    {
-      "epoch": 5.495591609968195,
-      "grad_norm": 0.004951875191181898,
-      "learning_rate": 0.0001999851062902271,
-      "loss": 46.0,
-      "step": 34125
-    },
-    {
-      "epoch": 5.495752647046983,
-      "grad_norm": 0.006439009215682745,
-      "learning_rate": 0.0001999851054170887,
-      "loss": 46.0,
-      "step": 34126
-    },
-    {
-      "epoch": 5.49591368412577,
-      "grad_norm": 0.0013956889742985368,
-      "learning_rate": 0.0001999851045439247,
-      "loss": 46.0,
-      "step": 34127
-    },
-    {
-      "epoch": 5.4960747212045575,
-      "grad_norm": 0.0022253713104873896,
-      "learning_rate": 0.0001999851036707351,
-      "loss": 46.0,
-      "step": 34128
-    },
-    {
-      "epoch": 5.496235758283345,
-      "grad_norm": 0.0068828389048576355,
-      "learning_rate": 0.00019998510279751992,
-      "loss": 46.0,
-      "step": 34129
-    },
-    {
-      "epoch": 5.496396795362132,
-      "grad_norm": 0.005418810062110424,
-      "learning_rate": 0.00019998510192427915,
-      "loss": 46.0,
-      "step": 34130
-    },
-    {
-      "epoch": 5.49655783244092,
-      "grad_norm": 0.0026387288235127926,
-      "learning_rate": 0.0001999851010510128,
-      "loss": 46.0,
-      "step": 34131
-    },
-    {
-      "epoch": 5.496718869519707,
-      "grad_norm": 0.0049411579966545105,
-      "learning_rate": 0.00019998510017772085,
-      "loss": 46.0,
-      "step": 34132
-    },
-    {
-      "epoch": 5.496879906598494,
-      "grad_norm": 0.005068782716989517,
-      "learning_rate": 0.00019998509930440331,
-      "loss": 46.0,
-      "step": 34133
-    },
-    {
-      "epoch": 5.497040943677281,
-      "grad_norm": 0.002494296757504344,
-      "learning_rate": 0.0001999850984310602,
-      "loss": 46.0,
-      "step": 34134
-    },
-    {
-      "epoch": 5.497201980756069,
-      "grad_norm": 0.01123699452728033,
-      "learning_rate": 0.00019998509755769146,
-      "loss": 46.0,
-      "step": 34135
-    },
-    {
-      "epoch": 5.497363017834856,
-      "grad_norm": 0.0022961741778999567,
-      "learning_rate": 0.00019998509668429714,
-      "loss": 46.0,
-      "step": 34136
-    },
-    {
-      "epoch": 5.497524054913644,
-      "grad_norm": 0.0012729689478874207,
-      "learning_rate": 0.00019998509581087723,
-      "loss": 46.0,
-      "step": 34137
-    },
-    {
-      "epoch": 5.497685091992431,
-      "grad_norm": 0.003151624696329236,
-      "learning_rate": 0.00019998509493743174,
-      "loss": 46.0,
-      "step": 34138
-    },
-    {
-      "epoch": 5.497846129071219,
-      "grad_norm": 0.005862434860318899,
-      "learning_rate": 0.00019998509406396066,
-      "loss": 46.0,
-      "step": 34139
-    },
-    {
-      "epoch": 5.498007166150006,
-      "grad_norm": 0.0068648867309093475,
-      "learning_rate": 0.00019998509319046396,
-      "loss": 46.0,
-      "step": 34140
-    },
-    {
-      "epoch": 5.4981682032287935,
-      "grad_norm": 0.0038339283782988787,
-      "learning_rate": 0.0001999850923169417,
-      "loss": 46.0,
-      "step": 34141
-    },
-    {
-      "epoch": 5.498329240307581,
-      "grad_norm": 0.0020844812970608473,
-      "learning_rate": 0.00019998509144339384,
-      "loss": 46.0,
-      "step": 34142
-    },
-    {
-      "epoch": 5.498490277386368,
-      "grad_norm": 0.0018930728547275066,
-      "learning_rate": 0.0001999850905698204,
-      "loss": 46.0,
-      "step": 34143
-    },
-    {
-      "epoch": 5.498651314465156,
-      "grad_norm": 0.006712261121720076,
-      "learning_rate": 0.00019998508969622133,
-      "loss": 46.0,
-      "step": 34144
-    },
-    {
-      "epoch": 5.498812351543943,
-      "grad_norm": 0.009071686305105686,
-      "learning_rate": 0.00019998508882259673,
-      "loss": 46.0,
-      "step": 34145
-    },
-    {
-      "epoch": 5.498973388622731,
-      "grad_norm": 0.0018597109010443091,
-      "learning_rate": 0.00019998508794894648,
-      "loss": 46.0,
-      "step": 34146
-    },
-    {
-      "epoch": 5.499134425701518,
-      "grad_norm": 0.007114425301551819,
-      "learning_rate": 0.00019998508707527068,
-      "loss": 46.0,
-      "step": 34147
-    },
-    {
-      "epoch": 5.499295462780305,
-      "grad_norm": 0.0013667274033650756,
-      "learning_rate": 0.00019998508620156926,
-      "loss": 46.0,
-      "step": 34148
-    },
-    {
-      "epoch": 5.499456499859092,
-      "grad_norm": 0.01908756047487259,
-      "learning_rate": 0.00019998508532784228,
-      "loss": 46.0,
-      "step": 34149
-    },
-    {
-      "epoch": 5.49961753693788,
-      "grad_norm": 0.005090340040624142,
-      "learning_rate": 0.00019998508445408968,
-      "loss": 46.0,
-      "step": 34150
-    },
-    {
-      "epoch": 5.499778574016667,
-      "grad_norm": 0.026592692360281944,
-      "learning_rate": 0.00019998508358031153,
-      "loss": 46.0,
-      "step": 34151
-    },
-    {
-      "epoch": 5.499939611095455,
-      "grad_norm": 0.002731078304350376,
-      "learning_rate": 0.00019998508270650773,
-      "loss": 46.0,
-      "step": 34152
-    },
-    {
-      "epoch": 5.500100648174242,
-      "grad_norm": 0.0017696896102279425,
-      "learning_rate": 0.00019998508183267837,
-      "loss": 46.0,
-      "step": 34153
-    },
-    {
-      "epoch": 5.5002616852530295,
-      "grad_norm": 0.0016494258306920528,
-      "learning_rate": 0.00019998508095882343,
-      "loss": 46.0,
-      "step": 34154
-    },
-    {
-      "epoch": 5.500422722331817,
-      "grad_norm": 0.005124740768224001,
-      "learning_rate": 0.00019998508008494287,
-      "loss": 46.0,
-      "step": 34155
-    },
-    {
-      "epoch": 5.500583759410604,
-      "grad_norm": 0.002132552908733487,
-      "learning_rate": 0.00019998507921103675,
-      "loss": 46.0,
-      "step": 34156
-    },
-    {
-      "epoch": 5.500744796489392,
-      "grad_norm": 0.010711484588682652,
-      "learning_rate": 0.00019998507833710502,
-      "loss": 46.0,
-      "step": 34157
-    },
-    {
-      "epoch": 5.500905833568179,
-      "grad_norm": 0.0009878241689875722,
-      "learning_rate": 0.0001999850774631477,
-      "loss": 46.0,
-      "step": 34158
-    },
-    {
-      "epoch": 5.501066870646967,
-      "grad_norm": 0.007199808955192566,
-      "learning_rate": 0.0001999850765891648,
-      "loss": 46.0,
-      "step": 34159
-    },
-    {
-      "epoch": 5.501227907725754,
-      "grad_norm": 0.01828848011791706,
-      "learning_rate": 0.00019998507571515628,
-      "loss": 46.0,
-      "step": 34160
-    },
-    {
-      "epoch": 5.501388944804541,
-      "grad_norm": 0.0034937537275254726,
-      "learning_rate": 0.0001999850748411222,
-      "loss": 46.0,
-      "step": 34161
-    },
-    {
-      "epoch": 5.501549981883329,
-      "grad_norm": 0.005627928301692009,
-      "learning_rate": 0.00019998507396706253,
-      "loss": 46.0,
-      "step": 34162
-    },
-    {
-      "epoch": 5.501711018962116,
-      "grad_norm": 0.0056982627138495445,
-      "learning_rate": 0.00019998507309297724,
-      "loss": 46.0,
-      "step": 34163
-    },
-    {
-      "epoch": 5.501872056040903,
-      "grad_norm": 0.006696680095046759,
-      "learning_rate": 0.0001999850722188664,
-      "loss": 46.0,
-      "step": 34164
-    },
-    {
-      "epoch": 5.502033093119691,
-      "grad_norm": 0.0043885246850550175,
-      "learning_rate": 0.00019998507134472995,
-      "loss": 46.0,
-      "step": 34165
-    },
-    {
-      "epoch": 5.502194130198478,
-      "grad_norm": 0.014521142467856407,
-      "learning_rate": 0.00019998507047056788,
-      "loss": 46.0,
-      "step": 34166
-    },
-    {
-      "epoch": 5.5023551672772655,
-      "grad_norm": 0.0067253378219902515,
-      "learning_rate": 0.00019998506959638027,
-      "loss": 46.0,
-      "step": 34167
-    },
-    {
-      "epoch": 5.502516204356053,
-      "grad_norm": 0.0022953529842197895,
-      "learning_rate": 0.00019998506872216703,
-      "loss": 46.0,
-      "step": 34168
-    },
-    {
-      "epoch": 5.50267724143484,
-      "grad_norm": 0.013663801364600658,
-      "learning_rate": 0.00019998506784792822,
-      "loss": 46.0,
-      "step": 34169
-    },
-    {
-      "epoch": 5.502838278513628,
-      "grad_norm": 0.00542009063065052,
-      "learning_rate": 0.0001999850669736638,
-      "loss": 46.0,
-      "step": 34170
-    },
-    {
-      "epoch": 5.502999315592415,
-      "grad_norm": 0.007828470319509506,
-      "learning_rate": 0.0001999850660993738,
-      "loss": 46.0,
-      "step": 34171
-    },
-    {
-      "epoch": 5.503160352671203,
-      "grad_norm": 0.0006252566818147898,
-      "learning_rate": 0.0001999850652250582,
-      "loss": 46.0,
-      "step": 34172
-    },
-    {
-      "epoch": 5.50332138974999,
-      "grad_norm": 0.010983610525727272,
-      "learning_rate": 0.00019998506435071702,
-      "loss": 46.0,
-      "step": 34173
-    },
-    {
-      "epoch": 5.503482426828778,
-      "grad_norm": 0.009040025062859058,
-      "learning_rate": 0.00019998506347635025,
-      "loss": 46.0,
-      "step": 34174
-    },
-    {
-      "epoch": 5.503643463907565,
-      "grad_norm": 0.0075152041390538216,
-      "learning_rate": 0.0001999850626019579,
-      "loss": 46.0,
-      "step": 34175
-    },
-    {
-      "epoch": 5.503804500986352,
-      "grad_norm": 0.0050583938136696815,
-      "learning_rate": 0.00019998506172753993,
-      "loss": 46.0,
-      "step": 34176
-    },
-    {
-      "epoch": 5.50396553806514,
-      "grad_norm": 0.0023125142324715853,
-      "learning_rate": 0.0001999850608530964,
-      "loss": 46.0,
-      "step": 34177
-    },
-    {
-      "epoch": 5.504126575143927,
-      "grad_norm": 0.003778290469199419,
-      "learning_rate": 0.00019998505997862723,
-      "loss": 46.0,
-      "step": 34178
-    },
-    {
-      "epoch": 5.504287612222714,
-      "grad_norm": 0.005179339088499546,
-      "learning_rate": 0.00019998505910413252,
-      "loss": 46.0,
-      "step": 34179
-    },
-    {
-      "epoch": 5.5044486493015015,
-      "grad_norm": 0.0011703349882736802,
-      "learning_rate": 0.0001999850582296122,
-      "loss": 46.0,
-      "step": 34180
-    },
-    {
-      "epoch": 5.504609686380289,
-      "grad_norm": 0.005269202403724194,
-      "learning_rate": 0.00019998505735506627,
-      "loss": 46.0,
-      "step": 34181
-    },
-    {
-      "epoch": 5.504770723459076,
-      "grad_norm": 0.004499434493482113,
-      "learning_rate": 0.00019998505648049478,
-      "loss": 46.0,
-      "step": 34182
-    },
-    {
-      "epoch": 5.504931760537864,
-      "grad_norm": 0.004150930792093277,
-      "learning_rate": 0.00019998505560589767,
-      "loss": 46.0,
-      "step": 34183
-    },
-    {
-      "epoch": 5.505092797616651,
-      "grad_norm": 0.002291294513270259,
-      "learning_rate": 0.00019998505473127498,
-      "loss": 46.0,
-      "step": 34184
-    },
-    {
-      "epoch": 5.505253834695439,
-      "grad_norm": 0.0016701340209692717,
-      "learning_rate": 0.00019998505385662672,
-      "loss": 46.0,
-      "step": 34185
-    },
-    {
-      "epoch": 5.505414871774226,
-      "grad_norm": 0.006673745345324278,
-      "learning_rate": 0.00019998505298195288,
-      "loss": 46.0,
-      "step": 34186
-    },
-    {
-      "epoch": 5.505575908853014,
-      "grad_norm": 0.006532181054353714,
-      "learning_rate": 0.0001999850521072534,
-      "loss": 46.0,
-      "step": 34187
-    },
-    {
-      "epoch": 5.505736945931801,
-      "grad_norm": 0.0020645789336413145,
-      "learning_rate": 0.00019998505123252836,
-      "loss": 46.0,
-      "step": 34188
-    },
-    {
-      "epoch": 5.5058979830105885,
-      "grad_norm": 0.009018348529934883,
-      "learning_rate": 0.0001999850503577777,
-      "loss": 46.0,
-      "step": 34189
-    },
-    {
-      "epoch": 5.506059020089376,
-      "grad_norm": 0.0013276365352794528,
-      "learning_rate": 0.00019998504948300148,
-      "loss": 46.0,
-      "step": 34190
-    },
-    {
-      "epoch": 5.5062200571681625,
-      "grad_norm": 0.005371210630983114,
-      "learning_rate": 0.00019998504860819965,
-      "loss": 46.0,
-      "step": 34191
-    },
-    {
-      "epoch": 5.50638109424695,
-      "grad_norm": 0.012877581641077995,
-      "learning_rate": 0.00019998504773337223,
-      "loss": 46.0,
-      "step": 34192
-    },
-    {
-      "epoch": 5.506542131325737,
-      "grad_norm": 0.0013488840777426958,
-      "learning_rate": 0.00019998504685851925,
-      "loss": 46.0,
-      "step": 34193
-    },
-    {
-      "epoch": 5.506703168404525,
-      "grad_norm": 0.006245528347790241,
-      "learning_rate": 0.00019998504598364063,
-      "loss": 46.0,
-      "step": 34194
-    },
-    {
-      "epoch": 5.506864205483312,
-      "grad_norm": 0.003025773447006941,
-      "learning_rate": 0.00019998504510873645,
-      "loss": 46.0,
-      "step": 34195
-    },
-    {
-      "epoch": 5.5070252425621,
-      "grad_norm": 0.00273637636564672,
-      "learning_rate": 0.00019998504423380668,
-      "loss": 46.0,
-      "step": 34196
-    },
-    {
-      "epoch": 5.507186279640887,
-      "grad_norm": 0.011829034425318241,
-      "learning_rate": 0.0001999850433588513,
-      "loss": 46.0,
-      "step": 34197
-    },
-    {
-      "epoch": 5.507347316719675,
-      "grad_norm": 0.003219072474166751,
-      "learning_rate": 0.00019998504248387036,
-      "loss": 46.0,
-      "step": 34198
-    },
-    {
-      "epoch": 5.507508353798462,
-      "grad_norm": 0.00635336060076952,
-      "learning_rate": 0.0001999850416088638,
-      "loss": 46.0,
-      "step": 34199
-    },
-    {
-      "epoch": 5.50766939087725,
-      "grad_norm": 0.001374226645566523,
-      "learning_rate": 0.00019998504073383166,
-      "loss": 46.0,
-      "step": 34200
-    },
-    {
-      "epoch": 5.507830427956037,
-      "grad_norm": 0.018123339861631393,
-      "learning_rate": 0.00019998503985877393,
-      "loss": 46.0,
-      "step": 34201
-    },
-    {
-      "epoch": 5.5079914650348245,
-      "grad_norm": 0.0018701660446822643,
-      "learning_rate": 0.00019998503898369061,
-      "loss": 46.0,
-      "step": 34202
-    },
-    {
-      "epoch": 5.508152502113612,
-      "grad_norm": 0.002980404533445835,
-      "learning_rate": 0.00019998503810858168,
-      "loss": 46.0,
-      "step": 34203
-    },
-    {
-      "epoch": 5.508313539192399,
-      "grad_norm": 0.002194474684074521,
-      "learning_rate": 0.0001999850372334472,
-      "loss": 46.0,
-      "step": 34204
-    },
-    {
-      "epoch": 5.508474576271187,
-      "grad_norm": 0.012783299200236797,
-      "learning_rate": 0.00019998503635828709,
-      "loss": 46.0,
-      "step": 34205
-    },
-    {
-      "epoch": 5.508635613349973,
-      "grad_norm": 0.002803035778924823,
-      "learning_rate": 0.0001999850354831014,
-      "loss": 46.0,
-      "step": 34206
-    },
-    {
-      "epoch": 5.508796650428761,
-      "grad_norm": 0.0049095768481493,
-      "learning_rate": 0.00019998503460789014,
-      "loss": 46.0,
-      "step": 34207
-    },
-    {
-      "epoch": 5.508957687507548,
-      "grad_norm": 0.0029583917930722237,
-      "learning_rate": 0.00019998503373265324,
-      "loss": 46.0,
-      "step": 34208
-    },
-    {
-      "epoch": 5.509118724586336,
-      "grad_norm": 0.011844675987958908,
-      "learning_rate": 0.00019998503285739082,
-      "loss": 46.0,
-      "step": 34209
-    },
-    {
-      "epoch": 5.509279761665123,
-      "grad_norm": 0.0039554196409881115,
-      "learning_rate": 0.00019998503198210275,
-      "loss": 46.0,
-      "step": 34210
-    },
-    {
-      "epoch": 5.509440798743911,
-      "grad_norm": 0.005298307631164789,
-      "learning_rate": 0.0001999850311067891,
-      "loss": 46.0,
-      "step": 34211
-    },
-    {
-      "epoch": 5.509601835822698,
-      "grad_norm": 0.010015979409217834,
-      "learning_rate": 0.00019998503023144987,
-      "loss": 46.0,
-      "step": 34212
-    },
-    {
-      "epoch": 5.509762872901486,
-      "grad_norm": 0.005829904228448868,
-      "learning_rate": 0.00019998502935608502,
-      "loss": 46.0,
-      "step": 34213
-    },
-    {
-      "epoch": 5.509923909980273,
-      "grad_norm": 0.003104221075773239,
-      "learning_rate": 0.00019998502848069463,
-      "loss": 46.0,
-      "step": 34214
-    },
-    {
-      "epoch": 5.5100849470590605,
-      "grad_norm": 0.0015328072477132082,
-      "learning_rate": 0.00019998502760527862,
-      "loss": 46.0,
-      "step": 34215
-    },
-    {
-      "epoch": 5.510245984137848,
-      "grad_norm": 0.004958477336913347,
-      "learning_rate": 0.000199985026729837,
-      "loss": 46.0,
-      "step": 34216
-    },
-    {
-      "epoch": 5.510407021216635,
-      "grad_norm": 0.007182045374065638,
-      "learning_rate": 0.00019998502585436985,
-      "loss": 46.0,
-      "step": 34217
-    },
-    {
-      "epoch": 5.510568058295423,
-      "grad_norm": 0.002235148102045059,
-      "learning_rate": 0.00019998502497887706,
-      "loss": 46.0,
-      "step": 34218
-    },
-    {
-      "epoch": 5.51072909537421,
-      "grad_norm": 0.0017434393521398306,
-      "learning_rate": 0.00019998502410335868,
-      "loss": 46.0,
-      "step": 34219
-    },
-    {
-      "epoch": 5.510890132452998,
-      "grad_norm": 0.01854533888399601,
-      "learning_rate": 0.00019998502322781474,
-      "loss": 46.0,
-      "step": 34220
-    },
-    {
-      "epoch": 5.511051169531784,
-      "grad_norm": 0.0016888438258320093,
-      "learning_rate": 0.00019998502235224515,
-      "loss": 46.0,
-      "step": 34221
-    },
-    {
-      "epoch": 5.511212206610572,
-      "grad_norm": 0.0008130050264298916,
-      "learning_rate": 0.00019998502147665,
-      "loss": 46.0,
-      "step": 34222
-    },
-    {
-      "epoch": 5.511373243689359,
-      "grad_norm": 0.012573973275721073,
-      "learning_rate": 0.00019998502060102928,
-      "loss": 46.0,
-      "step": 34223
-    },
-    {
-      "epoch": 5.511534280768147,
-      "grad_norm": 0.0010394317796453834,
-      "learning_rate": 0.00019998501972538297,
-      "loss": 46.0,
-      "step": 34224
-    },
-    {
-      "epoch": 5.511695317846934,
-      "grad_norm": 0.005306388717144728,
-      "learning_rate": 0.00019998501884971103,
-      "loss": 46.0,
-      "step": 34225
-    },
-    {
-      "epoch": 5.511856354925722,
-      "grad_norm": 0.002362685278058052,
-      "learning_rate": 0.00019998501797401354,
-      "loss": 46.0,
-      "step": 34226
-    },
-    {
-      "epoch": 5.512017392004509,
-      "grad_norm": 0.0027344832196831703,
-      "learning_rate": 0.00019998501709829044,
-      "loss": 46.0,
-      "step": 34227
-    },
-    {
-      "epoch": 5.5121784290832965,
-      "grad_norm": 0.012368762865662575,
-      "learning_rate": 0.00019998501622254175,
-      "loss": 46.0,
-      "step": 34228
-    },
-    {
-      "epoch": 5.512339466162084,
-      "grad_norm": 0.005386157892644405,
-      "learning_rate": 0.00019998501534676747,
-      "loss": 46.0,
-      "step": 34229
-    },
-    {
-      "epoch": 5.512500503240871,
-      "grad_norm": 0.01438825111836195,
-      "learning_rate": 0.0001999850144709676,
-      "loss": 46.0,
-      "step": 34230
-    },
-    {
-      "epoch": 5.512661540319659,
-      "grad_norm": 0.0038165715523064137,
-      "learning_rate": 0.00019998501359514212,
-      "loss": 46.0,
-      "step": 34231
-    },
-    {
-      "epoch": 5.512822577398446,
-      "grad_norm": 0.002818374428898096,
-      "learning_rate": 0.00019998501271929108,
-      "loss": 46.0,
-      "step": 34232
-    },
-    {
-      "epoch": 5.512983614477234,
-      "grad_norm": 0.0009560149046592414,
-      "learning_rate": 0.00019998501184341442,
-      "loss": 46.0,
-      "step": 34233
-    },
-    {
-      "epoch": 5.51314465155602,
-      "grad_norm": 0.009288566187024117,
-      "learning_rate": 0.00019998501096751218,
-      "loss": 46.0,
-      "step": 34234
-    },
-    {
-      "epoch": 5.513305688634809,
-      "grad_norm": 0.004407213069498539,
-      "learning_rate": 0.00019998501009158438,
-      "loss": 46.0,
-      "step": 34235
-    },
-    {
-      "epoch": 5.513466725713595,
-      "grad_norm": 0.005016713868826628,
-      "learning_rate": 0.00019998500921563096,
-      "loss": 46.0,
-      "step": 34236
-    },
-    {
-      "epoch": 5.513627762792383,
-      "grad_norm": 0.022559037432074547,
-      "learning_rate": 0.00019998500833965196,
-      "loss": 46.0,
-      "step": 34237
-    },
-    {
-      "epoch": 5.51378879987117,
-      "grad_norm": 0.003801714163273573,
-      "learning_rate": 0.00019998500746364734,
-      "loss": 46.0,
-      "step": 34238
-    },
-    {
-      "epoch": 5.5139498369499576,
-      "grad_norm": 0.0038948128931224346,
-      "learning_rate": 0.00019998500658761716,
-      "loss": 46.0,
-      "step": 34239
-    },
-    {
-      "epoch": 5.514110874028745,
-      "grad_norm": 0.0015044279862195253,
-      "learning_rate": 0.00019998500571156137,
-      "loss": 46.0,
-      "step": 34240
-    },
-    {
-      "epoch": 5.5142719111075325,
-      "grad_norm": 0.004923420492559671,
-      "learning_rate": 0.00019998500483548,
-      "loss": 46.0,
-      "step": 34241
-    },
-    {
-      "epoch": 5.51443294818632,
-      "grad_norm": 0.001664804178290069,
-      "learning_rate": 0.00019998500395937302,
-      "loss": 46.0,
-      "step": 34242
-    },
-    {
-      "epoch": 5.514593985265107,
-      "grad_norm": 0.0030872018542140722,
-      "learning_rate": 0.00019998500308324047,
-      "loss": 46.0,
-      "step": 34243
-    },
-    {
-      "epoch": 5.514755022343895,
-      "grad_norm": 0.0006278973305597901,
-      "learning_rate": 0.00019998500220708233,
-      "loss": 46.0,
-      "step": 34244
-    },
-    {
-      "epoch": 5.514916059422682,
-      "grad_norm": 0.008448035456240177,
-      "learning_rate": 0.0001999850013308986,
-      "loss": 46.0,
-      "step": 34245
-    },
-    {
-      "epoch": 5.51507709650147,
-      "grad_norm": 0.0005471177864819765,
-      "learning_rate": 0.00019998500045468926,
-      "loss": 46.0,
-      "step": 34246
-    },
-    {
-      "epoch": 5.515238133580257,
-      "grad_norm": 0.008499611169099808,
-      "learning_rate": 0.00019998499957845435,
-      "loss": 46.0,
-      "step": 34247
-    },
-    {
-      "epoch": 5.515399170659045,
-      "grad_norm": 0.004641486331820488,
-      "learning_rate": 0.00019998499870219384,
-      "loss": 46.0,
-      "step": 34248
-    },
-    {
-      "epoch": 5.515560207737831,
-      "grad_norm": 0.004723050631582737,
-      "learning_rate": 0.0001999849978259077,
-      "loss": 46.0,
-      "step": 34249
-    },
-    {
-      "epoch": 5.5157212448166195,
-      "grad_norm": 0.0019629092421382666,
-      "learning_rate": 0.00019998499694959604,
-      "loss": 46.0,
-      "step": 34250
-    },
-    {
-      "epoch": 5.515882281895406,
-      "grad_norm": 0.00206712051294744,
-      "learning_rate": 0.00019998499607325876,
-      "loss": 46.0,
-      "step": 34251
-    },
-    {
-      "epoch": 5.5160433189741935,
-      "grad_norm": 0.0007701151189394295,
-      "learning_rate": 0.00019998499519689587,
-      "loss": 46.0,
-      "step": 34252
-    },
-    {
-      "epoch": 5.516204356052981,
-      "grad_norm": 0.005060708150267601,
-      "learning_rate": 0.00019998499432050742,
-      "loss": 46.0,
-      "step": 34253
-    },
-    {
-      "epoch": 5.516365393131768,
-      "grad_norm": 0.008270849473774433,
-      "learning_rate": 0.00019998499344409335,
-      "loss": 46.0,
-      "step": 34254
-    },
-    {
-      "epoch": 5.516526430210556,
-      "grad_norm": 0.008350512012839317,
-      "learning_rate": 0.0001999849925676537,
-      "loss": 46.0,
-      "step": 34255
-    },
-    {
-      "epoch": 5.516687467289343,
-      "grad_norm": 0.003098027780652046,
-      "learning_rate": 0.00019998499169118848,
-      "loss": 46.0,
-      "step": 34256
-    },
-    {
-      "epoch": 5.516848504368131,
-      "grad_norm": 0.006756963208317757,
-      "learning_rate": 0.00019998499081469762,
-      "loss": 46.0,
-      "step": 34257
-    },
-    {
-      "epoch": 5.517009541446918,
-      "grad_norm": 0.0010933625744655728,
-      "learning_rate": 0.00019998498993818123,
-      "loss": 46.0,
-      "step": 34258
-    },
-    {
-      "epoch": 5.517170578525706,
-      "grad_norm": 0.005616801790893078,
-      "learning_rate": 0.0001999849890616392,
-      "loss": 46.0,
-      "step": 34259
-    },
-    {
-      "epoch": 5.517331615604493,
-      "grad_norm": 0.005372984334826469,
-      "learning_rate": 0.0001999849881850716,
-      "loss": 46.0,
-      "step": 34260
-    },
-    {
-      "epoch": 5.517492652683281,
-      "grad_norm": 0.008079294115304947,
-      "learning_rate": 0.0001999849873084784,
-      "loss": 46.0,
-      "step": 34261
-    },
-    {
-      "epoch": 5.517653689762068,
-      "grad_norm": 0.005586813669651747,
-      "learning_rate": 0.00019998498643185962,
-      "loss": 46.0,
-      "step": 34262
-    },
-    {
-      "epoch": 5.5178147268408555,
-      "grad_norm": 0.003906476777046919,
-      "learning_rate": 0.00019998498555521527,
-      "loss": 46.0,
-      "step": 34263
-    },
-    {
-      "epoch": 5.517975763919642,
-      "grad_norm": 0.0027365367859601974,
-      "learning_rate": 0.00019998498467854527,
-      "loss": 46.0,
-      "step": 34264
-    },
-    {
-      "epoch": 5.5181368009984295,
-      "grad_norm": 0.003583196084946394,
-      "learning_rate": 0.00019998498380184972,
-      "loss": 46.0,
-      "step": 34265
-    },
-    {
-      "epoch": 5.518297838077217,
-      "grad_norm": 0.001992402831092477,
-      "learning_rate": 0.00019998498292512858,
-      "loss": 46.0,
-      "step": 34266
-    },
-    {
-      "epoch": 5.518458875156004,
-      "grad_norm": 0.0035375088918954134,
-      "learning_rate": 0.00019998498204838183,
-      "loss": 46.0,
-      "step": 34267
-    },
-    {
-      "epoch": 5.518619912234792,
-      "grad_norm": 0.0033124389592558146,
-      "learning_rate": 0.0001999849811716095,
-      "loss": 46.0,
-      "step": 34268
-    },
-    {
-      "epoch": 5.518780949313579,
-      "grad_norm": 0.005209075752645731,
-      "learning_rate": 0.00019998498029481158,
-      "loss": 46.0,
-      "step": 34269
-    },
-    {
-      "epoch": 5.518941986392367,
-      "grad_norm": 0.001894405111670494,
-      "learning_rate": 0.00019998497941798807,
-      "loss": 46.0,
-      "step": 34270
-    },
-    {
-      "epoch": 5.519103023471154,
-      "grad_norm": 0.012381664477288723,
-      "learning_rate": 0.00019998497854113896,
-      "loss": 46.0,
-      "step": 34271
-    },
-    {
-      "epoch": 5.519264060549942,
-      "grad_norm": 0.0037883417680859566,
-      "learning_rate": 0.00019998497766426424,
-      "loss": 46.0,
-      "step": 34272
-    },
-    {
-      "epoch": 5.519425097628729,
-      "grad_norm": 0.0057252817787230015,
-      "learning_rate": 0.00019998497678736397,
-      "loss": 46.0,
-      "step": 34273
-    },
-    {
-      "epoch": 5.519586134707517,
-      "grad_norm": 0.0020538833923637867,
-      "learning_rate": 0.0001999849759104381,
-      "loss": 46.0,
-      "step": 34274
-    },
-    {
-      "epoch": 5.519747171786304,
-      "grad_norm": 0.025955431163311005,
-      "learning_rate": 0.00019998497503348662,
-      "loss": 46.0,
-      "step": 34275
-    },
-    {
-      "epoch": 5.5199082088650915,
-      "grad_norm": 0.009899445809423923,
-      "learning_rate": 0.00019998497415650958,
-      "loss": 46.0,
-      "step": 34276
-    },
-    {
-      "epoch": 5.520069245943879,
-      "grad_norm": 0.004798884503543377,
-      "learning_rate": 0.00019998497327950693,
-      "loss": 46.0,
-      "step": 34277
-    },
-    {
-      "epoch": 5.520230283022666,
-      "grad_norm": 0.0034558449406176805,
-      "learning_rate": 0.0001999849724024787,
-      "loss": 46.0,
-      "step": 34278
-    },
-    {
-      "epoch": 5.520391320101453,
-      "grad_norm": 0.0031368527561426163,
-      "learning_rate": 0.00019998497152542486,
-      "loss": 46.0,
-      "step": 34279
-    },
-    {
-      "epoch": 5.52055235718024,
-      "grad_norm": 0.006401309743523598,
-      "learning_rate": 0.00019998497064834542,
-      "loss": 46.0,
-      "step": 34280
-    },
-    {
-      "epoch": 5.520713394259028,
-      "grad_norm": 0.002318466315045953,
-      "learning_rate": 0.00019998496977124042,
-      "loss": 46.0,
-      "step": 34281
-    },
-    {
-      "epoch": 5.520874431337815,
-      "grad_norm": 0.003139452775940299,
-      "learning_rate": 0.0001999849688941098,
-      "loss": 46.0,
-      "step": 34282
-    },
-    {
-      "epoch": 5.521035468416603,
-      "grad_norm": 0.005002114921808243,
-      "learning_rate": 0.0001999849680169536,
-      "loss": 46.0,
-      "step": 34283
-    },
-    {
-      "epoch": 5.52119650549539,
-      "grad_norm": 0.006424926221370697,
-      "learning_rate": 0.0001999849671397718,
-      "loss": 46.0,
-      "step": 34284
-    },
-    {
-      "epoch": 5.521357542574178,
-      "grad_norm": 0.004735100083053112,
-      "learning_rate": 0.00019998496626256446,
-      "loss": 46.0,
-      "step": 34285
-    },
-    {
-      "epoch": 5.521518579652965,
-      "grad_norm": 0.012039123103022575,
-      "learning_rate": 0.00019998496538533146,
-      "loss": 46.0,
-      "step": 34286
-    },
-    {
-      "epoch": 5.521679616731753,
-      "grad_norm": 0.003961151000112295,
-      "learning_rate": 0.0001999849645080729,
-      "loss": 46.0,
-      "step": 34287
-    },
-    {
-      "epoch": 5.52184065381054,
-      "grad_norm": 0.004428533837199211,
-      "learning_rate": 0.00019998496363078875,
-      "loss": 46.0,
-      "step": 34288
-    },
-    {
-      "epoch": 5.5220016908893275,
-      "grad_norm": 0.008572126738727093,
-      "learning_rate": 0.00019998496275347902,
-      "loss": 46.0,
-      "step": 34289
-    },
-    {
-      "epoch": 5.522162727968115,
-      "grad_norm": 0.005892967339605093,
-      "learning_rate": 0.00019998496187614368,
-      "loss": 46.0,
-      "step": 34290
-    },
-    {
-      "epoch": 5.522323765046902,
-      "grad_norm": 0.002398570068180561,
-      "learning_rate": 0.00019998496099878278,
-      "loss": 46.0,
-      "step": 34291
-    },
-    {
-      "epoch": 5.52248480212569,
-      "grad_norm": 0.0023990217596292496,
-      "learning_rate": 0.00019998496012139623,
-      "loss": 46.0,
-      "step": 34292
-    },
-    {
-      "epoch": 5.522645839204477,
-      "grad_norm": 0.003208345966413617,
-      "learning_rate": 0.00019998495924398416,
-      "loss": 46.0,
-      "step": 34293
-    },
-    {
-      "epoch": 5.522806876283264,
-      "grad_norm": 0.009863338433206081,
-      "learning_rate": 0.00019998495836654642,
-      "loss": 46.0,
-      "step": 34294
-    },
-    {
-      "epoch": 5.522967913362051,
-      "grad_norm": 0.0046867383643984795,
-      "learning_rate": 0.00019998495748908314,
-      "loss": 46.0,
-      "step": 34295
-    },
-    {
-      "epoch": 5.523128950440839,
-      "grad_norm": 0.00847650971263647,
-      "learning_rate": 0.00019998495661159427,
-      "loss": 46.0,
-      "step": 34296
-    },
-    {
-      "epoch": 5.523289987519626,
-      "grad_norm": 0.0016790707595646381,
-      "learning_rate": 0.0001999849557340798,
-      "loss": 46.0,
-      "step": 34297
-    },
-    {
-      "epoch": 5.523451024598414,
-      "grad_norm": 0.0022495663724839687,
-      "learning_rate": 0.00019998495485653973,
-      "loss": 46.0,
-      "step": 34298
-    },
-    {
-      "epoch": 5.523612061677201,
-      "grad_norm": 0.0069077154621481895,
-      "learning_rate": 0.00019998495397897408,
-      "loss": 46.0,
-      "step": 34299
-    },
-    {
-      "epoch": 5.5237730987559885,
-      "grad_norm": 0.015760261565446854,
-      "learning_rate": 0.00019998495310138284,
-      "loss": 46.0,
-      "step": 34300
-    },
-    {
-      "epoch": 5.523934135834776,
-      "grad_norm": 0.004451438318938017,
-      "learning_rate": 0.000199984952223766,
-      "loss": 46.0,
-      "step": 34301
-    },
-    {
-      "epoch": 5.5240951729135634,
-      "grad_norm": 0.015721792355179787,
-      "learning_rate": 0.00019998495134612357,
-      "loss": 46.0,
-      "step": 34302
-    },
-    {
-      "epoch": 5.524256209992351,
-      "grad_norm": 0.005324992351233959,
-      "learning_rate": 0.00019998495046845557,
-      "loss": 46.0,
-      "step": 34303
-    },
-    {
-      "epoch": 5.524417247071138,
-      "grad_norm": 0.0029777069576084614,
-      "learning_rate": 0.00019998494959076195,
-      "loss": 46.0,
-      "step": 34304
-    },
-    {
-      "epoch": 5.524578284149926,
-      "grad_norm": 0.0026553133502602577,
-      "learning_rate": 0.00019998494871304275,
-      "loss": 46.0,
-      "step": 34305
-    },
-    {
-      "epoch": 5.524739321228713,
-      "grad_norm": 0.009386973455548286,
-      "learning_rate": 0.000199984947835298,
-      "loss": 46.0,
-      "step": 34306
-    },
-    {
-      "epoch": 5.5249003583075,
-      "grad_norm": 0.0059987870045006275,
-      "learning_rate": 0.00019998494695752758,
-      "loss": 46.0,
-      "step": 34307
-    },
-    {
-      "epoch": 5.525061395386288,
-      "grad_norm": 0.010576811619102955,
-      "learning_rate": 0.00019998494607973162,
-      "loss": 46.0,
-      "step": 34308
-    },
-    {
-      "epoch": 5.525222432465075,
-      "grad_norm": 0.002587325172498822,
-      "learning_rate": 0.00019998494520191004,
-      "loss": 46.0,
-      "step": 34309
-    },
-    {
-      "epoch": 5.525383469543862,
-      "grad_norm": 0.00582084758207202,
-      "learning_rate": 0.0001999849443240629,
-      "loss": 46.0,
-      "step": 34310
-    },
-    {
-      "epoch": 5.52554450662265,
-      "grad_norm": 0.008977241814136505,
-      "learning_rate": 0.00019998494344619015,
-      "loss": 46.0,
-      "step": 34311
-    },
-    {
-      "epoch": 5.525705543701437,
-      "grad_norm": 0.004376677796244621,
-      "learning_rate": 0.0001999849425682918,
-      "loss": 46.0,
-      "step": 34312
-    },
-    {
-      "epoch": 5.5258665807802245,
-      "grad_norm": 0.008413034491240978,
-      "learning_rate": 0.0001999849416903679,
-      "loss": 46.0,
-      "step": 34313
-    },
-    {
-      "epoch": 5.526027617859012,
-      "grad_norm": 0.011920742690563202,
-      "learning_rate": 0.00019998494081241837,
-      "loss": 46.0,
-      "step": 34314
-    },
-    {
-      "epoch": 5.526188654937799,
-      "grad_norm": 0.0039012304041534662,
-      "learning_rate": 0.00019998493993444324,
-      "loss": 46.0,
-      "step": 34315
-    },
-    {
-      "epoch": 5.526349692016587,
-      "grad_norm": 0.009865395724773407,
-      "learning_rate": 0.00019998493905644255,
-      "loss": 46.0,
-      "step": 34316
-    },
-    {
-      "epoch": 5.526510729095374,
-      "grad_norm": 0.003301802324131131,
-      "learning_rate": 0.00019998493817841628,
-      "loss": 46.0,
-      "step": 34317
-    },
-    {
-      "epoch": 5.526671766174162,
-      "grad_norm": 0.005590891465544701,
-      "learning_rate": 0.00019998493730036439,
-      "loss": 46.0,
-      "step": 34318
-    },
-    {
-      "epoch": 5.526832803252949,
-      "grad_norm": 0.005014342255890369,
-      "learning_rate": 0.0001999849364222869,
-      "loss": 46.0,
-      "step": 34319
-    },
-    {
-      "epoch": 5.526993840331737,
-      "grad_norm": 0.016980236396193504,
-      "learning_rate": 0.00019998493554418384,
-      "loss": 46.0,
-      "step": 34320
-    },
-    {
-      "epoch": 5.527154877410524,
-      "grad_norm": 0.004121170844882727,
-      "learning_rate": 0.0001999849346660552,
-      "loss": 46.0,
-      "step": 34321
-    },
-    {
-      "epoch": 5.527315914489311,
-      "grad_norm": 0.0013843679334968328,
-      "learning_rate": 0.00019998493378790095,
-      "loss": 46.0,
-      "step": 34322
-    },
-    {
-      "epoch": 5.527476951568099,
-      "grad_norm": 0.002124807331711054,
-      "learning_rate": 0.0001999849329097211,
-      "loss": 46.0,
-      "step": 34323
-    },
-    {
-      "epoch": 5.527637988646886,
-      "grad_norm": 0.014505774714052677,
-      "learning_rate": 0.0001999849320315157,
-      "loss": 46.0,
-      "step": 34324
-    },
-    {
-      "epoch": 5.527799025725673,
-      "grad_norm": 0.00169506692327559,
-      "learning_rate": 0.00019998493115328466,
-      "loss": 46.0,
-      "step": 34325
-    },
-    {
-      "epoch": 5.5279600628044605,
-      "grad_norm": 0.0019217573571950197,
-      "learning_rate": 0.00019998493027502805,
-      "loss": 46.0,
-      "step": 34326
-    },
-    {
-      "epoch": 5.528121099883248,
-      "grad_norm": 0.006825822405517101,
-      "learning_rate": 0.00019998492939674584,
-      "loss": 46.0,
-      "step": 34327
-    },
-    {
-      "epoch": 5.528282136962035,
-      "grad_norm": 0.006091851741075516,
-      "learning_rate": 0.00019998492851843803,
-      "loss": 46.0,
-      "step": 34328
-    },
-    {
-      "epoch": 5.528443174040823,
-      "grad_norm": 0.0034964075312018394,
-      "learning_rate": 0.00019998492764010468,
-      "loss": 46.0,
-      "step": 34329
-    },
-    {
-      "epoch": 5.52860421111961,
-      "grad_norm": 0.002492989879101515,
-      "learning_rate": 0.0001999849267617457,
-      "loss": 46.0,
-      "step": 34330
-    },
-    {
-      "epoch": 5.528765248198398,
-      "grad_norm": 0.0023658943828195333,
-      "learning_rate": 0.00019998492588336114,
-      "loss": 46.0,
-      "step": 34331
-    },
-    {
-      "epoch": 5.528926285277185,
-      "grad_norm": 0.001708652707748115,
-      "learning_rate": 0.00019998492500495097,
-      "loss": 46.0,
-      "step": 34332
-    },
-    {
-      "epoch": 5.529087322355973,
-      "grad_norm": 0.009595518000423908,
-      "learning_rate": 0.00019998492412651522,
-      "loss": 46.0,
-      "step": 34333
-    },
-    {
-      "epoch": 5.52924835943476,
-      "grad_norm": 0.009467966854572296,
-      "learning_rate": 0.00019998492324805388,
-      "loss": 46.0,
-      "step": 34334
-    },
-    {
-      "epoch": 5.529409396513548,
-      "grad_norm": 0.010365890339016914,
-      "learning_rate": 0.00019998492236956696,
-      "loss": 46.0,
-      "step": 34335
-    },
-    {
-      "epoch": 5.529570433592335,
-      "grad_norm": 0.002486846875399351,
-      "learning_rate": 0.00019998492149105444,
-      "loss": 46.0,
-      "step": 34336
-    },
-    {
-      "epoch": 5.529731470671122,
-      "grad_norm": 0.003119267988950014,
-      "learning_rate": 0.00019998492061251632,
-      "loss": 46.0,
-      "step": 34337
-    },
-    {
-      "epoch": 5.529892507749909,
-      "grad_norm": 0.021245073527097702,
-      "learning_rate": 0.00019998491973395263,
-      "loss": 46.0,
-      "step": 34338
-    },
-    {
-      "epoch": 5.5300535448286965,
-      "grad_norm": 0.001042731455527246,
-      "learning_rate": 0.00019998491885536335,
-      "loss": 46.0,
-      "step": 34339
-    },
-    {
-      "epoch": 5.530214581907484,
-      "grad_norm": 0.003292615059763193,
-      "learning_rate": 0.00019998491797674844,
-      "loss": 46.0,
-      "step": 34340
-    },
-    {
-      "epoch": 5.530375618986271,
-      "grad_norm": 0.017703954130411148,
-      "learning_rate": 0.000199984917098108,
-      "loss": 46.0,
-      "step": 34341
-    },
-    {
-      "epoch": 5.530536656065059,
-      "grad_norm": 0.006306701805442572,
-      "learning_rate": 0.0001999849162194419,
-      "loss": 46.0,
-      "step": 34342
-    },
-    {
-      "epoch": 5.530697693143846,
-      "grad_norm": 0.0068973214365541935,
-      "learning_rate": 0.00019998491534075025,
-      "loss": 46.0,
-      "step": 34343
-    },
-    {
-      "epoch": 5.530858730222634,
-      "grad_norm": 0.001817410229705274,
-      "learning_rate": 0.000199984914462033,
-      "loss": 46.0,
-      "step": 34344
-    },
-    {
-      "epoch": 5.531019767301421,
-      "grad_norm": 0.009167851880192757,
-      "learning_rate": 0.00019998491358329016,
-      "loss": 46.0,
-      "step": 34345
-    },
-    {
-      "epoch": 5.531180804380209,
-      "grad_norm": 0.005233730189502239,
-      "learning_rate": 0.00019998491270452174,
-      "loss": 46.0,
-      "step": 34346
-    },
-    {
-      "epoch": 5.531341841458996,
-      "grad_norm": 0.01128034945577383,
-      "learning_rate": 0.00019998491182572772,
-      "loss": 46.0,
-      "step": 34347
-    },
-    {
-      "epoch": 5.5315028785377836,
-      "grad_norm": 0.005587833467870951,
-      "learning_rate": 0.0001999849109469081,
-      "loss": 46.0,
-      "step": 34348
-    },
-    {
-      "epoch": 5.531663915616571,
-      "grad_norm": 0.00860118493437767,
-      "learning_rate": 0.0001999849100680629,
-      "loss": 46.0,
-      "step": 34349
-    },
-    {
-      "epoch": 5.5318249526953585,
-      "grad_norm": 0.018641334027051926,
-      "learning_rate": 0.00019998490918919212,
-      "loss": 46.0,
-      "step": 34350
-    },
-    {
-      "epoch": 5.531985989774146,
-      "grad_norm": 0.00677757803350687,
-      "learning_rate": 0.00019998490831029574,
-      "loss": 46.0,
-      "step": 34351
-    },
-    {
-      "epoch": 5.5321470268529325,
-      "grad_norm": 0.0046311113983392715,
-      "learning_rate": 0.00019998490743137372,
-      "loss": 46.0,
-      "step": 34352
-    },
-    {
-      "epoch": 5.53230806393172,
-      "grad_norm": 0.004090859089046717,
-      "learning_rate": 0.00019998490655242617,
-      "loss": 46.0,
-      "step": 34353
-    },
-    {
-      "epoch": 5.532469101010507,
-      "grad_norm": 0.008302649483084679,
-      "learning_rate": 0.000199984905673453,
-      "loss": 46.0,
-      "step": 34354
-    },
-    {
-      "epoch": 5.532630138089295,
-      "grad_norm": 0.012410673312842846,
-      "learning_rate": 0.00019998490479445426,
-      "loss": 46.0,
-      "step": 34355
-    },
-    {
-      "epoch": 5.532791175168082,
-      "grad_norm": 0.0005821784725412726,
-      "learning_rate": 0.00019998490391542992,
-      "loss": 46.0,
-      "step": 34356
-    },
-    {
-      "epoch": 5.53295221224687,
-      "grad_norm": 0.0010559582151472569,
-      "learning_rate": 0.00019998490303637997,
-      "loss": 46.0,
-      "step": 34357
-    },
-    {
-      "epoch": 5.533113249325657,
-      "grad_norm": 0.0035826722159981728,
-      "learning_rate": 0.00019998490215730446,
-      "loss": 46.0,
-      "step": 34358
-    },
-    {
-      "epoch": 5.533274286404445,
-      "grad_norm": 0.005177249666303396,
-      "learning_rate": 0.00019998490127820336,
-      "loss": 46.0,
-      "step": 34359
-    },
-    {
-      "epoch": 5.533435323483232,
-      "grad_norm": 0.0010994458571076393,
-      "learning_rate": 0.00019998490039907664,
-      "loss": 46.0,
-      "step": 34360
-    },
-    {
-      "epoch": 5.5335963605620195,
-      "grad_norm": 0.004649078939110041,
-      "learning_rate": 0.00019998489951992437,
-      "loss": 46.0,
-      "step": 34361
-    },
-    {
-      "epoch": 5.533757397640807,
-      "grad_norm": 0.0019365191692486405,
-      "learning_rate": 0.00019998489864074648,
-      "loss": 46.0,
-      "step": 34362
-    },
-    {
-      "epoch": 5.533918434719594,
-      "grad_norm": 0.0029365073423832655,
-      "learning_rate": 0.00019998489776154298,
-      "loss": 46.0,
-      "step": 34363
-    },
-    {
-      "epoch": 5.534079471798382,
-      "grad_norm": 0.004521762486547232,
-      "learning_rate": 0.00019998489688231392,
-      "loss": 46.0,
-      "step": 34364
-    },
-    {
-      "epoch": 5.534240508877169,
-      "grad_norm": 0.0012761177495121956,
-      "learning_rate": 0.00019998489600305927,
-      "loss": 46.0,
-      "step": 34365
-    },
-    {
-      "epoch": 5.534401545955957,
-      "grad_norm": 0.009701911360025406,
-      "learning_rate": 0.000199984895123779,
-      "loss": 46.0,
-      "step": 34366
-    },
-    {
-      "epoch": 5.534562583034743,
-      "grad_norm": 0.003976393491029739,
-      "learning_rate": 0.00019998489424447318,
-      "loss": 46.0,
-      "step": 34367
-    },
-    {
-      "epoch": 5.534723620113531,
-      "grad_norm": 0.01857377216219902,
-      "learning_rate": 0.0001999848933651417,
-      "loss": 46.0,
-      "step": 34368
-    },
-    {
-      "epoch": 5.534884657192318,
-      "grad_norm": 0.0061077275313436985,
-      "learning_rate": 0.0001999848924857847,
-      "loss": 46.0,
-      "step": 34369
-    },
-    {
-      "epoch": 5.535045694271106,
-      "grad_norm": 0.0031367880292236805,
-      "learning_rate": 0.00019998489160640207,
-      "loss": 46.0,
-      "step": 34370
-    },
-    {
-      "epoch": 5.535206731349893,
-      "grad_norm": 0.001604948891326785,
-      "learning_rate": 0.0001999848907269939,
-      "loss": 46.0,
-      "step": 34371
-    },
-    {
-      "epoch": 5.535367768428681,
-      "grad_norm": 0.007231251336634159,
-      "learning_rate": 0.0001999848898475601,
-      "loss": 46.0,
-      "step": 34372
-    },
-    {
-      "epoch": 5.535528805507468,
-      "grad_norm": 0.0014801690122112632,
-      "learning_rate": 0.0001999848889681007,
-      "loss": 46.0,
-      "step": 34373
-    },
-    {
-      "epoch": 5.5356898425862555,
-      "grad_norm": 0.0026634891983121634,
-      "learning_rate": 0.00019998488808861572,
-      "loss": 46.0,
-      "step": 34374
-    },
-    {
-      "epoch": 5.535850879665043,
-      "grad_norm": 0.0036713636945933104,
-      "learning_rate": 0.00019998488720910515,
-      "loss": 46.0,
-      "step": 34375
-    },
-    {
-      "epoch": 5.53601191674383,
-      "grad_norm": 0.01738077960908413,
-      "learning_rate": 0.000199984886329569,
-      "loss": 46.0,
-      "step": 34376
-    },
-    {
-      "epoch": 5.536172953822618,
-      "grad_norm": 0.0049221874214708805,
-      "learning_rate": 0.00019998488545000724,
-      "loss": 46.0,
-      "step": 34377
-    },
-    {
-      "epoch": 5.536333990901405,
-      "grad_norm": 0.006742923520505428,
-      "learning_rate": 0.0001999848845704199,
-      "loss": 46.0,
-      "step": 34378
-    },
-    {
-      "epoch": 5.536495027980193,
-      "grad_norm": 0.005495220422744751,
-      "learning_rate": 0.00019998488369080697,
-      "loss": 46.0,
-      "step": 34379
-    },
-    {
-      "epoch": 5.536656065058979,
-      "grad_norm": 0.009002111852169037,
-      "learning_rate": 0.00019998488281116843,
-      "loss": 46.0,
-      "step": 34380
-    },
-    {
-      "epoch": 5.536817102137768,
-      "grad_norm": 0.003474907483905554,
-      "learning_rate": 0.00019998488193150434,
-      "loss": 46.0,
-      "step": 34381
-    },
-    {
-      "epoch": 5.536978139216554,
-      "grad_norm": 0.006822897586971521,
-      "learning_rate": 0.00019998488105181462,
-      "loss": 46.0,
-      "step": 34382
-    },
-    {
-      "epoch": 5.537139176295342,
-      "grad_norm": 0.0038658790290355682,
-      "learning_rate": 0.00019998488017209932,
-      "loss": 46.0,
-      "step": 34383
-    },
-    {
-      "epoch": 5.537300213374129,
-      "grad_norm": 0.0008098231046460569,
-      "learning_rate": 0.00019998487929235844,
-      "loss": 46.0,
-      "step": 34384
-    },
-    {
-      "epoch": 5.537461250452917,
-      "grad_norm": 0.004747644532471895,
-      "learning_rate": 0.00019998487841259193,
-      "loss": 46.0,
-      "step": 34385
-    },
-    {
-      "epoch": 5.537622287531704,
-      "grad_norm": 0.004653476178646088,
-      "learning_rate": 0.00019998487753279987,
-      "loss": 46.0,
-      "step": 34386
-    },
-    {
-      "epoch": 5.5377833246104915,
-      "grad_norm": 0.002107733627781272,
-      "learning_rate": 0.00019998487665298222,
-      "loss": 46.0,
-      "step": 34387
-    },
-    {
-      "epoch": 5.537944361689279,
-      "grad_norm": 0.0020892168395221233,
-      "learning_rate": 0.000199984875773139,
-      "loss": 46.0,
-      "step": 34388
-    },
-    {
-      "epoch": 5.538105398768066,
-      "grad_norm": 0.01438012532889843,
-      "learning_rate": 0.00019998487489327014,
-      "loss": 46.0,
-      "step": 34389
-    },
-    {
-      "epoch": 5.538266435846854,
-      "grad_norm": 0.002079607453197241,
-      "learning_rate": 0.0001999848740133757,
-      "loss": 46.0,
-      "step": 34390
-    },
-    {
-      "epoch": 5.538427472925641,
-      "grad_norm": 0.002929080743342638,
-      "learning_rate": 0.00019998487313345567,
-      "loss": 46.0,
-      "step": 34391
-    },
-    {
-      "epoch": 5.538588510004429,
-      "grad_norm": 0.0011994106462225318,
-      "learning_rate": 0.00019998487225351004,
-      "loss": 46.0,
-      "step": 34392
-    },
-    {
-      "epoch": 5.538749547083216,
-      "grad_norm": 0.003827650099992752,
-      "learning_rate": 0.00019998487137353884,
-      "loss": 46.0,
-      "step": 34393
-    },
-    {
-      "epoch": 5.538910584162004,
-      "grad_norm": 0.002340164501219988,
-      "learning_rate": 0.00019998487049354205,
-      "loss": 46.0,
-      "step": 34394
-    },
-    {
-      "epoch": 5.53907162124079,
-      "grad_norm": 0.0025572828017175198,
-      "learning_rate": 0.00019998486961351965,
-      "loss": 46.0,
-      "step": 34395
-    },
-    {
-      "epoch": 5.539232658319579,
-      "grad_norm": 0.023076998069882393,
-      "learning_rate": 0.0001999848687334717,
-      "loss": 46.0,
-      "step": 34396
-    },
-    {
-      "epoch": 5.539393695398365,
-      "grad_norm": 0.005896314978599548,
-      "learning_rate": 0.0001999848678533981,
-      "loss": 46.0,
-      "step": 34397
-    },
-    {
-      "epoch": 5.539554732477153,
-      "grad_norm": 0.0021342081017792225,
-      "learning_rate": 0.00019998486697329893,
-      "loss": 46.0,
-      "step": 34398
-    },
-    {
-      "epoch": 5.53971576955594,
-      "grad_norm": 0.0015544709749519825,
-      "learning_rate": 0.0001999848660931742,
-      "loss": 46.0,
-      "step": 34399
-    },
-    {
-      "epoch": 5.5398768066347275,
-      "grad_norm": 0.011797095648944378,
-      "learning_rate": 0.00019998486521302384,
-      "loss": 46.0,
-      "step": 34400
-    },
-    {
-      "epoch": 5.540037843713515,
-      "grad_norm": 0.001448853756301105,
-      "learning_rate": 0.00019998486433284792,
-      "loss": 46.0,
-      "step": 34401
-    },
-    {
-      "epoch": 5.540198880792302,
-      "grad_norm": 0.003398359287530184,
-      "learning_rate": 0.00019998486345264638,
-      "loss": 46.0,
-      "step": 34402
-    },
-    {
-      "epoch": 5.54035991787109,
-      "grad_norm": 0.014259311370551586,
-      "learning_rate": 0.00019998486257241925,
-      "loss": 46.0,
-      "step": 34403
-    },
-    {
-      "epoch": 5.540520954949877,
-      "grad_norm": 0.0010322092566639185,
-      "learning_rate": 0.00019998486169216654,
-      "loss": 46.0,
-      "step": 34404
-    },
-    {
-      "epoch": 5.540681992028665,
-      "grad_norm": 0.010638273321092129,
-      "learning_rate": 0.00019998486081188824,
-      "loss": 46.0,
-      "step": 34405
-    },
-    {
-      "epoch": 5.540843029107452,
-      "grad_norm": 0.006411506328731775,
-      "learning_rate": 0.00019998485993158436,
-      "loss": 46.0,
-      "step": 34406
-    },
-    {
-      "epoch": 5.54100406618624,
-      "grad_norm": 0.003926464822143316,
-      "learning_rate": 0.00019998485905125488,
-      "loss": 46.0,
-      "step": 34407
-    },
-    {
-      "epoch": 5.541165103265027,
-      "grad_norm": 0.005655432585626841,
-      "learning_rate": 0.0001999848581708998,
-      "loss": 46.0,
-      "step": 34408
-    },
-    {
-      "epoch": 5.5413261403438145,
-      "grad_norm": 0.004261812195181847,
-      "learning_rate": 0.00019998485729051915,
-      "loss": 46.0,
-      "step": 34409
-    },
-    {
-      "epoch": 5.541487177422601,
-      "grad_norm": 0.009570648893713951,
-      "learning_rate": 0.00019998485641011289,
-      "loss": 46.0,
-      "step": 34410
-    },
-    {
-      "epoch": 5.541648214501389,
-      "grad_norm": 0.004239460453391075,
-      "learning_rate": 0.00019998485552968104,
-      "loss": 46.0,
-      "step": 34411
-    },
-    {
-      "epoch": 5.541809251580176,
-      "grad_norm": 0.006265928503125906,
-      "learning_rate": 0.0001999848546492236,
-      "loss": 46.0,
-      "step": 34412
-    },
-    {
-      "epoch": 5.5419702886589635,
-      "grad_norm": 0.008009640499949455,
-      "learning_rate": 0.00019998485376874058,
-      "loss": 46.0,
-      "step": 34413
-    },
-    {
-      "epoch": 5.542131325737751,
-      "grad_norm": 0.00547971623018384,
-      "learning_rate": 0.00019998485288823194,
-      "loss": 46.0,
-      "step": 34414
-    },
-    {
-      "epoch": 5.542292362816538,
-      "grad_norm": 0.005356329958885908,
-      "learning_rate": 0.00019998485200769777,
-      "loss": 46.0,
-      "step": 34415
-    },
-    {
-      "epoch": 5.542453399895326,
-      "grad_norm": 0.004759754985570908,
-      "learning_rate": 0.00019998485112713795,
-      "loss": 46.0,
-      "step": 34416
-    },
-    {
-      "epoch": 5.542614436974113,
-      "grad_norm": 0.0008380708168260753,
-      "learning_rate": 0.00019998485024655255,
-      "loss": 46.0,
-      "step": 34417
-    },
-    {
-      "epoch": 5.542775474052901,
-      "grad_norm": 0.0019018686143681407,
-      "learning_rate": 0.00019998484936594157,
-      "loss": 46.0,
-      "step": 34418
-    },
-    {
-      "epoch": 5.542936511131688,
-      "grad_norm": 0.002970636123791337,
-      "learning_rate": 0.000199984848485305,
-      "loss": 46.0,
-      "step": 34419
-    },
-    {
-      "epoch": 5.543097548210476,
-      "grad_norm": 0.0031251846812665462,
-      "learning_rate": 0.00019998484760464283,
-      "loss": 46.0,
-      "step": 34420
-    },
-    {
-      "epoch": 5.543258585289263,
-      "grad_norm": 0.006115109194070101,
-      "learning_rate": 0.00019998484672395508,
-      "loss": 46.0,
-      "step": 34421
-    },
-    {
-      "epoch": 5.5434196223680505,
-      "grad_norm": 0.006192250642925501,
-      "learning_rate": 0.00019998484584324172,
-      "loss": 46.0,
-      "step": 34422
-    },
-    {
-      "epoch": 5.543580659446838,
-      "grad_norm": 0.006695727817714214,
-      "learning_rate": 0.0001999848449625028,
-      "loss": 46.0,
-      "step": 34423
-    },
-    {
-      "epoch": 5.543741696525625,
-      "grad_norm": 0.0019366793567314744,
-      "learning_rate": 0.00019998484408173826,
-      "loss": 46.0,
-      "step": 34424
-    },
-    {
-      "epoch": 5.543902733604412,
-      "grad_norm": 0.01719304360449314,
-      "learning_rate": 0.00019998484320094814,
-      "loss": 46.0,
-      "step": 34425
-    },
-    {
-      "epoch": 5.544063770683199,
-      "grad_norm": 0.010537405498325825,
-      "learning_rate": 0.00019998484232013243,
-      "loss": 46.0,
-      "step": 34426
-    },
-    {
-      "epoch": 5.544224807761987,
-      "grad_norm": 0.0034111528657376766,
-      "learning_rate": 0.00019998484143929113,
-      "loss": 46.0,
-      "step": 34427
-    },
-    {
-      "epoch": 5.544385844840774,
-      "grad_norm": 0.00249963509850204,
-      "learning_rate": 0.00019998484055842424,
-      "loss": 46.0,
-      "step": 34428
-    },
-    {
-      "epoch": 5.544546881919562,
-      "grad_norm": 0.005336133763194084,
-      "learning_rate": 0.00019998483967753177,
-      "loss": 46.0,
-      "step": 34429
-    },
-    {
-      "epoch": 5.544707918998349,
-      "grad_norm": 0.002840744098648429,
-      "learning_rate": 0.00019998483879661368,
-      "loss": 46.0,
-      "step": 34430
-    },
-    {
-      "epoch": 5.544868956077137,
-      "grad_norm": 0.0067572928965091705,
-      "learning_rate": 0.00019998483791567,
-      "loss": 46.0,
-      "step": 34431
-    },
-    {
-      "epoch": 5.545029993155924,
-      "grad_norm": 0.004984838422387838,
-      "learning_rate": 0.00019998483703470078,
-      "loss": 46.0,
-      "step": 34432
-    },
-    {
-      "epoch": 5.545191030234712,
-      "grad_norm": 0.005045051220804453,
-      "learning_rate": 0.0001999848361537059,
-      "loss": 46.0,
-      "step": 34433
-    },
-    {
-      "epoch": 5.545352067313499,
-      "grad_norm": 0.010295151732861996,
-      "learning_rate": 0.00019998483527268546,
-      "loss": 46.0,
-      "step": 34434
-    },
-    {
-      "epoch": 5.5455131043922865,
-      "grad_norm": 0.0077599212527275085,
-      "learning_rate": 0.00019998483439163947,
-      "loss": 46.0,
-      "step": 34435
-    },
-    {
-      "epoch": 5.545674141471074,
-      "grad_norm": 0.004394874442368746,
-      "learning_rate": 0.00019998483351056783,
-      "loss": 46.0,
-      "step": 34436
-    },
-    {
-      "epoch": 5.545835178549861,
-      "grad_norm": 0.003599482821300626,
-      "learning_rate": 0.0001999848326294706,
-      "loss": 46.0,
-      "step": 34437
-    },
-    {
-      "epoch": 5.545996215628649,
-      "grad_norm": 0.0029495202470570803,
-      "learning_rate": 0.00019998483174834782,
-      "loss": 46.0,
-      "step": 34438
-    },
-    {
-      "epoch": 5.546157252707436,
-      "grad_norm": 0.005832471419125795,
-      "learning_rate": 0.00019998483086719942,
-      "loss": 46.0,
-      "step": 34439
-    },
-    {
-      "epoch": 5.546318289786223,
-      "grad_norm": 0.0029884595423936844,
-      "learning_rate": 0.00019998482998602544,
-      "loss": 46.0,
-      "step": 34440
-    },
-    {
-      "epoch": 5.54647932686501,
-      "grad_norm": 0.001863309764303267,
-      "learning_rate": 0.00019998482910482587,
-      "loss": 46.0,
-      "step": 34441
-    },
-    {
-      "epoch": 5.546640363943798,
-      "grad_norm": 0.0016161770327016711,
-      "learning_rate": 0.0001999848282236007,
-      "loss": 46.0,
-      "step": 34442
-    },
-    {
-      "epoch": 5.546801401022585,
-      "grad_norm": 0.006914549972862005,
-      "learning_rate": 0.00019998482734234993,
-      "loss": 46.0,
-      "step": 34443
-    },
-    {
-      "epoch": 5.546962438101373,
-      "grad_norm": 0.004871016833931208,
-      "learning_rate": 0.0001999848264610736,
-      "loss": 46.0,
-      "step": 34444
-    },
-    {
-      "epoch": 5.54712347518016,
-      "grad_norm": 0.010751190595328808,
-      "learning_rate": 0.00019998482557977165,
-      "loss": 46.0,
-      "step": 34445
-    },
-    {
-      "epoch": 5.547284512258948,
-      "grad_norm": 0.011315125040709972,
-      "learning_rate": 0.0001999848246984441,
-      "loss": 46.0,
-      "step": 34446
-    },
-    {
-      "epoch": 5.547445549337735,
-      "grad_norm": 0.0029595200903713703,
-      "learning_rate": 0.00019998482381709102,
-      "loss": 46.0,
-      "step": 34447
-    },
-    {
-      "epoch": 5.5476065864165225,
-      "grad_norm": 0.007874802686274052,
-      "learning_rate": 0.00019998482293571228,
-      "loss": 46.0,
-      "step": 34448
-    },
-    {
-      "epoch": 5.54776762349531,
-      "grad_norm": 0.007983098737895489,
-      "learning_rate": 0.000199984822054308,
-      "loss": 46.0,
-      "step": 34449
-    },
-    {
-      "epoch": 5.547928660574097,
-      "grad_norm": 0.003110006684437394,
-      "learning_rate": 0.0001999848211728781,
-      "loss": 46.0,
-      "step": 34450
-    },
-    {
-      "epoch": 5.548089697652885,
-      "grad_norm": 0.0019746567122638226,
-      "learning_rate": 0.0001999848202914226,
-      "loss": 46.0,
-      "step": 34451
-    },
-    {
-      "epoch": 5.548250734731672,
-      "grad_norm": 0.005718163680285215,
-      "learning_rate": 0.00019998481940994154,
-      "loss": 46.0,
-      "step": 34452
-    },
-    {
-      "epoch": 5.548411771810459,
-      "grad_norm": 0.004172750283032656,
-      "learning_rate": 0.00019998481852843484,
-      "loss": 46.0,
-      "step": 34453
-    },
-    {
-      "epoch": 5.548572808889247,
-      "grad_norm": 0.007584495469927788,
-      "learning_rate": 0.0001999848176469026,
-      "loss": 46.0,
-      "step": 34454
-    },
-    {
-      "epoch": 5.548733845968034,
-      "grad_norm": 0.005572207272052765,
-      "learning_rate": 0.00019998481676534476,
-      "loss": 46.0,
-      "step": 34455
-    },
-    {
-      "epoch": 5.548894883046821,
-      "grad_norm": 0.0016179116209968925,
-      "learning_rate": 0.0001999848158837613,
-      "loss": 46.0,
-      "step": 34456
-    },
-    {
-      "epoch": 5.549055920125609,
-      "grad_norm": 0.0063423095270991325,
-      "learning_rate": 0.0001999848150021523,
-      "loss": 46.0,
-      "step": 34457
-    },
-    {
-      "epoch": 5.549216957204396,
-      "grad_norm": 0.006661117076873779,
-      "learning_rate": 0.00019998481412051767,
-      "loss": 46.0,
-      "step": 34458
-    },
-    {
-      "epoch": 5.549377994283184,
-      "grad_norm": 0.00895021017640829,
-      "learning_rate": 0.00019998481323885744,
-      "loss": 46.0,
-      "step": 34459
-    },
-    {
-      "epoch": 5.549539031361971,
-      "grad_norm": 0.01310550607740879,
-      "learning_rate": 0.00019998481235717166,
-      "loss": 46.0,
-      "step": 34460
-    },
-    {
-      "epoch": 5.5497000684407585,
-      "grad_norm": 0.015178152360022068,
-      "learning_rate": 0.00019998481147546023,
-      "loss": 46.0,
-      "step": 34461
-    },
-    {
-      "epoch": 5.549861105519546,
-      "grad_norm": 0.0159327182918787,
-      "learning_rate": 0.00019998481059372325,
-      "loss": 46.0,
-      "step": 34462
-    },
-    {
-      "epoch": 5.550022142598333,
-      "grad_norm": 0.0018634002190083265,
-      "learning_rate": 0.00019998480971196068,
-      "loss": 46.0,
-      "step": 34463
-    },
-    {
-      "epoch": 5.550183179677121,
-      "grad_norm": 0.003102337010204792,
-      "learning_rate": 0.00019998480883017252,
-      "loss": 46.0,
-      "step": 34464
-    },
-    {
-      "epoch": 5.550344216755908,
-      "grad_norm": 0.004125271923840046,
-      "learning_rate": 0.00019998480794835877,
-      "loss": 46.0,
-      "step": 34465
-    },
-    {
-      "epoch": 5.550505253834696,
-      "grad_norm": 0.006804870907217264,
-      "learning_rate": 0.0001999848070665194,
-      "loss": 46.0,
-      "step": 34466
-    },
-    {
-      "epoch": 5.550666290913483,
-      "grad_norm": 0.005887329578399658,
-      "learning_rate": 0.00019998480618465446,
-      "loss": 46.0,
-      "step": 34467
-    },
-    {
-      "epoch": 5.55082732799227,
-      "grad_norm": 0.004895301535725594,
-      "learning_rate": 0.00019998480530276393,
-      "loss": 46.0,
-      "step": 34468
-    },
-    {
-      "epoch": 5.550988365071058,
-      "grad_norm": 0.006397101562470198,
-      "learning_rate": 0.0001999848044208478,
-      "loss": 46.0,
-      "step": 34469
-    },
-    {
-      "epoch": 5.551149402149845,
-      "grad_norm": 0.004611080978065729,
-      "learning_rate": 0.00019998480353890612,
-      "loss": 46.0,
-      "step": 34470
-    },
-    {
-      "epoch": 5.551310439228632,
-      "grad_norm": 0.012511455453932285,
-      "learning_rate": 0.0001999848026569388,
-      "loss": 46.0,
-      "step": 34471
-    },
-    {
-      "epoch": 5.5514714763074196,
-      "grad_norm": 0.00209240335971117,
-      "learning_rate": 0.0001999848017749459,
-      "loss": 46.0,
-      "step": 34472
-    },
-    {
-      "epoch": 5.551632513386207,
-      "grad_norm": 0.004890891723334789,
-      "learning_rate": 0.00019998480089292742,
-      "loss": 46.0,
-      "step": 34473
-    },
-    {
-      "epoch": 5.5517935504649945,
-      "grad_norm": 0.012718337588012218,
-      "learning_rate": 0.00019998480001088333,
-      "loss": 46.0,
-      "step": 34474
-    },
-    {
-      "epoch": 5.551954587543782,
-      "grad_norm": 0.004170960746705532,
-      "learning_rate": 0.00019998479912881366,
-      "loss": 46.0,
-      "step": 34475
-    },
-    {
-      "epoch": 5.552115624622569,
-      "grad_norm": 0.001989784650504589,
-      "learning_rate": 0.0001999847982467184,
-      "loss": 46.0,
-      "step": 34476
-    },
-    {
-      "epoch": 5.552276661701357,
-      "grad_norm": 0.004002275876700878,
-      "learning_rate": 0.00019998479736459755,
-      "loss": 46.0,
-      "step": 34477
-    },
-    {
-      "epoch": 5.552437698780144,
-      "grad_norm": 0.004416029900312424,
-      "learning_rate": 0.00019998479648245112,
-      "loss": 46.0,
-      "step": 34478
-    },
-    {
-      "epoch": 5.552598735858932,
-      "grad_norm": 0.005273669492453337,
-      "learning_rate": 0.00019998479560027907,
-      "loss": 46.0,
-      "step": 34479
-    },
-    {
-      "epoch": 5.552759772937719,
-      "grad_norm": 0.0024761478416621685,
-      "learning_rate": 0.00019998479471808146,
-      "loss": 46.0,
-      "step": 34480
-    },
-    {
-      "epoch": 5.552920810016507,
-      "grad_norm": 0.004912256728857756,
-      "learning_rate": 0.00019998479383585824,
-      "loss": 46.0,
-      "step": 34481
-    },
-    {
-      "epoch": 5.553081847095294,
-      "grad_norm": 0.012811501510441303,
-      "learning_rate": 0.00019998479295360943,
-      "loss": 46.0,
-      "step": 34482
-    },
-    {
-      "epoch": 5.553242884174081,
-      "grad_norm": 0.0008873852202668786,
-      "learning_rate": 0.00019998479207133504,
-      "loss": 46.0,
-      "step": 34483
-    },
-    {
-      "epoch": 5.553403921252869,
-      "grad_norm": 0.006089046597480774,
-      "learning_rate": 0.00019998479118903505,
-      "loss": 46.0,
-      "step": 34484
-    },
-    {
-      "epoch": 5.5535649583316555,
-      "grad_norm": 0.013763438910245895,
-      "learning_rate": 0.00019998479030670948,
-      "loss": 46.0,
-      "step": 34485
-    },
-    {
-      "epoch": 5.553725995410443,
-      "grad_norm": 0.0037245890125632286,
-      "learning_rate": 0.00019998478942435832,
-      "loss": 46.0,
-      "step": 34486
-    },
-    {
-      "epoch": 5.55388703248923,
-      "grad_norm": 0.0022490578703582287,
-      "learning_rate": 0.00019998478854198155,
-      "loss": 46.0,
-      "step": 34487
-    },
-    {
-      "epoch": 5.554048069568018,
-      "grad_norm": 0.0031325221061706543,
-      "learning_rate": 0.0001999847876595792,
-      "loss": 46.0,
-      "step": 34488
-    },
-    {
-      "epoch": 5.554209106646805,
-      "grad_norm": 0.001481261570006609,
-      "learning_rate": 0.00019998478677715124,
-      "loss": 46.0,
-      "step": 34489
-    },
-    {
-      "epoch": 5.554370143725593,
-      "grad_norm": 0.0010920444037765265,
-      "learning_rate": 0.0001999847858946977,
-      "loss": 46.0,
-      "step": 34490
-    },
-    {
-      "epoch": 5.55453118080438,
-      "grad_norm": 0.004955823067575693,
-      "learning_rate": 0.0001999847850122186,
-      "loss": 46.0,
-      "step": 34491
-    },
-    {
-      "epoch": 5.554692217883168,
-      "grad_norm": 0.003760437248274684,
-      "learning_rate": 0.00019998478412971385,
-      "loss": 46.0,
-      "step": 34492
-    },
-    {
-      "epoch": 5.554853254961955,
-      "grad_norm": 0.001445003435947001,
-      "learning_rate": 0.00019998478324718358,
-      "loss": 46.0,
-      "step": 34493
-    },
-    {
-      "epoch": 5.555014292040743,
-      "grad_norm": 0.002647166606038809,
-      "learning_rate": 0.00019998478236462767,
-      "loss": 46.0,
-      "step": 34494
-    },
-    {
-      "epoch": 5.55517532911953,
-      "grad_norm": 0.0009021577425301075,
-      "learning_rate": 0.00019998478148204618,
-      "loss": 46.0,
-      "step": 34495
-    },
-    {
-      "epoch": 5.5553363661983175,
-      "grad_norm": 0.006988700479269028,
-      "learning_rate": 0.00019998478059943912,
-      "loss": 46.0,
-      "step": 34496
-    },
-    {
-      "epoch": 5.555497403277105,
-      "grad_norm": 0.0035412353463470936,
-      "learning_rate": 0.00019998477971680645,
-      "loss": 46.0,
-      "step": 34497
-    },
-    {
-      "epoch": 5.5556584403558915,
-      "grad_norm": 0.005361685995012522,
-      "learning_rate": 0.0001999847788341482,
-      "loss": 46.0,
-      "step": 34498
-    },
-    {
-      "epoch": 5.555819477434679,
-      "grad_norm": 0.005229577422142029,
-      "learning_rate": 0.00019998477795146432,
-      "loss": 46.0,
-      "step": 34499
-    },
-    {
-      "epoch": 5.555980514513466,
-      "grad_norm": 0.002730033593252301,
-      "learning_rate": 0.00019998477706875488,
-      "loss": 46.0,
-      "step": 34500
-    },
-    {
-      "epoch": 5.556141551592254,
-      "grad_norm": 0.00528239319100976,
-      "learning_rate": 0.00019998477618601986,
-      "loss": 46.0,
-      "step": 34501
-    },
-    {
-      "epoch": 5.556302588671041,
-      "grad_norm": 0.00353110465221107,
-      "learning_rate": 0.00019998477530325923,
-      "loss": 46.0,
-      "step": 34502
-    },
-    {
-      "epoch": 5.556463625749829,
-      "grad_norm": 0.007440636400133371,
-      "learning_rate": 0.000199984774420473,
-      "loss": 46.0,
-      "step": 34503
-    },
-    {
-      "epoch": 5.556624662828616,
-      "grad_norm": 0.024559518322348595,
-      "learning_rate": 0.0001999847735376612,
-      "loss": 46.0,
-      "step": 34504
-    },
-    {
-      "epoch": 5.556785699907404,
-      "grad_norm": 0.006311963312327862,
-      "learning_rate": 0.0001999847726548238,
-      "loss": 46.0,
-      "step": 34505
-    },
-    {
-      "epoch": 5.556946736986191,
-      "grad_norm": 0.001077562803402543,
-      "learning_rate": 0.00019998477177196082,
-      "loss": 46.0,
-      "step": 34506
-    },
-    {
-      "epoch": 5.557107774064979,
-      "grad_norm": 0.002573074772953987,
-      "learning_rate": 0.00019998477088907222,
-      "loss": 46.0,
-      "step": 34507
-    },
-    {
-      "epoch": 5.557268811143766,
-      "grad_norm": 0.015399484895169735,
-      "learning_rate": 0.00019998477000615807,
-      "loss": 46.0,
-      "step": 34508
-    },
-    {
-      "epoch": 5.5574298482225535,
-      "grad_norm": 0.0031322527211159468,
-      "learning_rate": 0.0001999847691232183,
-      "loss": 46.0,
-      "step": 34509
-    },
-    {
-      "epoch": 5.557590885301341,
-      "grad_norm": 0.0032818394247442484,
-      "learning_rate": 0.00019998476824025296,
-      "loss": 46.0,
-      "step": 34510
-    },
-    {
-      "epoch": 5.557751922380128,
-      "grad_norm": 0.0019908598624169827,
-      "learning_rate": 0.000199984767357262,
-      "loss": 46.0,
-      "step": 34511
-    },
-    {
-      "epoch": 5.557912959458916,
-      "grad_norm": 0.006744795944541693,
-      "learning_rate": 0.00019998476647424546,
-      "loss": 46.0,
-      "step": 34512
-    },
-    {
-      "epoch": 5.558073996537702,
-      "grad_norm": 0.00100703164935112,
-      "learning_rate": 0.00019998476559120336,
-      "loss": 46.0,
-      "step": 34513
-    },
-    {
-      "epoch": 5.55823503361649,
-      "grad_norm": 0.01440840121358633,
-      "learning_rate": 0.00019998476470813563,
-      "loss": 46.0,
-      "step": 34514
-    },
-    {
-      "epoch": 5.558396070695277,
-      "grad_norm": 0.008395075798034668,
-      "learning_rate": 0.0001999847638250423,
-      "loss": 46.0,
-      "step": 34515
-    },
-    {
-      "epoch": 5.558557107774065,
-      "grad_norm": 0.001035574940033257,
-      "learning_rate": 0.0001999847629419234,
-      "loss": 46.0,
-      "step": 34516
-    },
-    {
-      "epoch": 5.558718144852852,
-      "grad_norm": 0.004762306343764067,
-      "learning_rate": 0.00019998476205877893,
-      "loss": 46.0,
-      "step": 34517
-    },
-    {
-      "epoch": 5.55887918193164,
-      "grad_norm": 0.011833344586193562,
-      "learning_rate": 0.00019998476117560885,
-      "loss": 46.0,
-      "step": 34518
-    },
-    {
-      "epoch": 5.559040219010427,
-      "grad_norm": 0.010235494002699852,
-      "learning_rate": 0.00019998476029241318,
-      "loss": 46.0,
-      "step": 34519
-    },
-    {
-      "epoch": 5.559201256089215,
-      "grad_norm": 0.006829431280493736,
-      "learning_rate": 0.00019998475940919192,
-      "loss": 46.0,
-      "step": 34520
-    },
-    {
-      "epoch": 5.559362293168002,
-      "grad_norm": 0.0015620003687217832,
-      "learning_rate": 0.00019998475852594507,
-      "loss": 46.0,
-      "step": 34521
-    },
-    {
-      "epoch": 5.5595233302467895,
-      "grad_norm": 0.0023274957202374935,
-      "learning_rate": 0.00019998475764267261,
-      "loss": 46.0,
-      "step": 34522
-    },
-    {
-      "epoch": 5.559684367325577,
-      "grad_norm": 0.004059371072798967,
-      "learning_rate": 0.00019998475675937457,
-      "loss": 46.0,
-      "step": 34523
-    },
-    {
-      "epoch": 5.559845404404364,
-      "grad_norm": 0.001856443239375949,
-      "learning_rate": 0.00019998475587605093,
-      "loss": 46.0,
-      "step": 34524
-    },
-    {
-      "epoch": 5.560006441483152,
-      "grad_norm": 0.003952109254896641,
-      "learning_rate": 0.00019998475499270174,
-      "loss": 46.0,
-      "step": 34525
-    },
-    {
-      "epoch": 5.560167478561939,
-      "grad_norm": 0.001939662266522646,
-      "learning_rate": 0.00019998475410932693,
-      "loss": 46.0,
-      "step": 34526
-    },
-    {
-      "epoch": 5.560328515640727,
-      "grad_norm": 0.009591889567673206,
-      "learning_rate": 0.00019998475322592654,
-      "loss": 46.0,
-      "step": 34527
-    },
-    {
-      "epoch": 5.560489552719513,
-      "grad_norm": 0.013927983120083809,
-      "learning_rate": 0.00019998475234250056,
-      "loss": 46.0,
-      "step": 34528
-    },
-    {
-      "epoch": 5.560650589798301,
-      "grad_norm": 0.008656861260533333,
-      "learning_rate": 0.00019998475145904896,
-      "loss": 46.0,
-      "step": 34529
-    },
-    {
-      "epoch": 5.560811626877088,
-      "grad_norm": 0.0023376161698251963,
-      "learning_rate": 0.00019998475057557178,
-      "loss": 46.0,
-      "step": 34530
-    },
-    {
-      "epoch": 5.560972663955876,
-      "grad_norm": 0.004913882352411747,
-      "learning_rate": 0.00019998474969206903,
-      "loss": 46.0,
-      "step": 34531
-    },
-    {
-      "epoch": 5.561133701034663,
-      "grad_norm": 0.002141357399523258,
-      "learning_rate": 0.00019998474880854067,
-      "loss": 46.0,
-      "step": 34532
-    },
-    {
-      "epoch": 5.5612947381134505,
-      "grad_norm": 0.016392672434449196,
-      "learning_rate": 0.00019998474792498673,
-      "loss": 46.0,
-      "step": 34533
-    },
-    {
-      "epoch": 5.561455775192238,
-      "grad_norm": 0.0009976379806175828,
-      "learning_rate": 0.0001999847470414072,
-      "loss": 46.0,
-      "step": 34534
-    },
-    {
-      "epoch": 5.5616168122710254,
-      "grad_norm": 0.0015318581135943532,
-      "learning_rate": 0.00019998474615780208,
-      "loss": 46.0,
-      "step": 34535
-    },
-    {
-      "epoch": 5.561777849349813,
-      "grad_norm": 0.0175799410790205,
-      "learning_rate": 0.00019998474527417134,
-      "loss": 46.0,
-      "step": 34536
-    },
-    {
-      "epoch": 5.5619388864286,
-      "grad_norm": 0.004948609042912722,
-      "learning_rate": 0.00019998474439051505,
-      "loss": 46.0,
-      "step": 34537
-    },
-    {
-      "epoch": 5.562099923507388,
-      "grad_norm": 0.007577365264296532,
-      "learning_rate": 0.00019998474350683314,
-      "loss": 46.0,
-      "step": 34538
-    },
-    {
-      "epoch": 5.562260960586175,
-      "grad_norm": 0.017973581328988075,
-      "learning_rate": 0.00019998474262312565,
-      "loss": 46.0,
-      "step": 34539
-    },
-    {
-      "epoch": 5.562421997664963,
-      "grad_norm": 0.0057105026207864285,
-      "learning_rate": 0.00019998474173939256,
-      "loss": 46.0,
-      "step": 34540
-    },
-    {
-      "epoch": 5.562583034743749,
-      "grad_norm": 0.008196820504963398,
-      "learning_rate": 0.0001999847408556339,
-      "loss": 46.0,
-      "step": 34541
-    },
-    {
-      "epoch": 5.562744071822538,
-      "grad_norm": 0.0031462498009204865,
-      "learning_rate": 0.00019998473997184964,
-      "loss": 46.0,
-      "step": 34542
-    },
-    {
-      "epoch": 5.562905108901324,
-      "grad_norm": 0.005572670139372349,
-      "learning_rate": 0.0001999847390880398,
-      "loss": 46.0,
-      "step": 34543
-    },
-    {
-      "epoch": 5.563066145980112,
-      "grad_norm": 0.0031304722651839256,
-      "learning_rate": 0.00019998473820420434,
-      "loss": 46.0,
-      "step": 34544
-    },
-    {
-      "epoch": 5.563227183058899,
-      "grad_norm": 0.010518571361899376,
-      "learning_rate": 0.00019998473732034332,
-      "loss": 46.0,
-      "step": 34545
-    },
-    {
-      "epoch": 5.5633882201376865,
-      "grad_norm": 0.01010617520660162,
-      "learning_rate": 0.00019998473643645666,
-      "loss": 46.0,
-      "step": 34546
-    },
-    {
-      "epoch": 5.563549257216474,
-      "grad_norm": 0.008721606805920601,
-      "learning_rate": 0.00019998473555254444,
-      "loss": 46.0,
-      "step": 34547
-    },
-    {
-      "epoch": 5.563710294295261,
-      "grad_norm": 0.01649344526231289,
-      "learning_rate": 0.00019998473466860663,
-      "loss": 46.0,
-      "step": 34548
-    },
-    {
-      "epoch": 5.563871331374049,
-      "grad_norm": 0.0019041125196963549,
-      "learning_rate": 0.00019998473378464324,
-      "loss": 46.0,
-      "step": 34549
-    },
-    {
-      "epoch": 5.564032368452836,
-      "grad_norm": 0.008882109075784683,
-      "learning_rate": 0.00019998473290065426,
-      "loss": 46.0,
-      "step": 34550
-    },
-    {
-      "epoch": 5.564193405531624,
-      "grad_norm": 0.004673162009567022,
-      "learning_rate": 0.00019998473201663966,
-      "loss": 46.0,
-      "step": 34551
-    },
-    {
-      "epoch": 5.564354442610411,
-      "grad_norm": 0.008114958181977272,
-      "learning_rate": 0.00019998473113259948,
-      "loss": 46.0,
-      "step": 34552
-    },
-    {
-      "epoch": 5.564515479689199,
-      "grad_norm": 0.01687239110469818,
-      "learning_rate": 0.00019998473024853374,
-      "loss": 46.0,
-      "step": 34553
-    },
-    {
-      "epoch": 5.564676516767986,
-      "grad_norm": 0.0021339314989745617,
-      "learning_rate": 0.00019998472936444238,
-      "loss": 46.0,
-      "step": 34554
-    },
-    {
-      "epoch": 5.564837553846774,
-      "grad_norm": 0.0014605579199269414,
-      "learning_rate": 0.00019998472848032543,
-      "loss": 46.0,
-      "step": 34555
-    },
-    {
-      "epoch": 5.56499859092556,
-      "grad_norm": 0.003656551940366626,
-      "learning_rate": 0.0001999847275961829,
-      "loss": 46.0,
-      "step": 34556
-    },
-    {
-      "epoch": 5.5651596280043485,
-      "grad_norm": 0.008041848428547382,
-      "learning_rate": 0.00019998472671201476,
-      "loss": 46.0,
-      "step": 34557
-    },
-    {
-      "epoch": 5.565320665083135,
-      "grad_norm": 0.001565627520903945,
-      "learning_rate": 0.00019998472582782105,
-      "loss": 46.0,
-      "step": 34558
-    },
-    {
-      "epoch": 5.5654817021619225,
-      "grad_norm": 0.00947917066514492,
-      "learning_rate": 0.00019998472494360173,
-      "loss": 46.0,
-      "step": 34559
-    },
-    {
-      "epoch": 5.56564273924071,
-      "grad_norm": 0.010613704100251198,
-      "learning_rate": 0.00019998472405935683,
-      "loss": 46.0,
-      "step": 34560
-    },
-    {
-      "epoch": 5.565803776319497,
-      "grad_norm": 0.007723473943769932,
-      "learning_rate": 0.00019998472317508633,
-      "loss": 46.0,
-      "step": 34561
-    },
-    {
-      "epoch": 5.565964813398285,
-      "grad_norm": 0.0053221858106553555,
-      "learning_rate": 0.00019998472229079025,
-      "loss": 46.0,
-      "step": 34562
-    },
-    {
-      "epoch": 5.566125850477072,
-      "grad_norm": 0.0020912366453558207,
-      "learning_rate": 0.00019998472140646858,
-      "loss": 46.0,
-      "step": 34563
-    },
-    {
-      "epoch": 5.56628688755586,
-      "grad_norm": 0.013028525747358799,
-      "learning_rate": 0.00019998472052212133,
-      "loss": 46.0,
-      "step": 34564
-    },
-    {
-      "epoch": 5.566447924634647,
-      "grad_norm": 0.0028623254038393497,
-      "learning_rate": 0.00019998471963774845,
-      "loss": 46.0,
-      "step": 34565
-    },
-    {
-      "epoch": 5.566608961713435,
-      "grad_norm": 0.008594634011387825,
-      "learning_rate": 0.00019998471875335,
-      "loss": 46.0,
-      "step": 34566
-    },
-    {
-      "epoch": 5.566769998792222,
-      "grad_norm": 0.0036207567900419235,
-      "learning_rate": 0.00019998471786892598,
-      "loss": 46.0,
-      "step": 34567
-    },
-    {
-      "epoch": 5.56693103587101,
-      "grad_norm": 0.015687158331274986,
-      "learning_rate": 0.00019998471698447635,
-      "loss": 46.0,
-      "step": 34568
-    },
-    {
-      "epoch": 5.567092072949797,
-      "grad_norm": 0.002217196160927415,
-      "learning_rate": 0.0001999847161000011,
-      "loss": 46.0,
-      "step": 34569
-    },
-    {
-      "epoch": 5.5672531100285845,
-      "grad_norm": 0.005931479390710592,
-      "learning_rate": 0.00019998471521550032,
-      "loss": 46.0,
-      "step": 34570
-    },
-    {
-      "epoch": 5.567414147107371,
-      "grad_norm": 0.003628320759162307,
-      "learning_rate": 0.00019998471433097393,
-      "loss": 46.0,
-      "step": 34571
-    },
-    {
-      "epoch": 5.5675751841861585,
-      "grad_norm": 0.00676200445741415,
-      "learning_rate": 0.00019998471344642192,
-      "loss": 46.0,
-      "step": 34572
-    },
-    {
-      "epoch": 5.567736221264946,
-      "grad_norm": 0.001319525996223092,
-      "learning_rate": 0.00019998471256184433,
-      "loss": 46.0,
-      "step": 34573
-    },
-    {
-      "epoch": 5.567897258343733,
-      "grad_norm": 0.0035766279324889183,
-      "learning_rate": 0.00019998471167724117,
-      "loss": 46.0,
-      "step": 34574
-    },
-    {
-      "epoch": 5.568058295422521,
-      "grad_norm": 0.004922098014503717,
-      "learning_rate": 0.0001999847107926124,
-      "loss": 46.0,
-      "step": 34575
-    },
-    {
-      "epoch": 5.568219332501308,
-      "grad_norm": 0.011862557381391525,
-      "learning_rate": 0.00019998470990795805,
-      "loss": 46.0,
-      "step": 34576
-    },
-    {
-      "epoch": 5.568380369580096,
-      "grad_norm": 0.0017253529513254762,
-      "learning_rate": 0.00019998470902327808,
-      "loss": 46.0,
-      "step": 34577
-    },
-    {
-      "epoch": 5.568541406658883,
-      "grad_norm": 0.001838596654124558,
-      "learning_rate": 0.00019998470813857255,
-      "loss": 46.0,
-      "step": 34578
-    },
-    {
-      "epoch": 5.568702443737671,
-      "grad_norm": 0.00573623925447464,
-      "learning_rate": 0.00019998470725384143,
-      "loss": 46.0,
-      "step": 34579
-    },
-    {
-      "epoch": 5.568863480816458,
-      "grad_norm": 0.006743662059307098,
-      "learning_rate": 0.0001999847063690847,
-      "loss": 46.0,
-      "step": 34580
-    },
-    {
-      "epoch": 5.5690245178952456,
-      "grad_norm": 0.0019299580017104745,
-      "learning_rate": 0.00019998470548430238,
-      "loss": 46.0,
-      "step": 34581
-    },
-    {
-      "epoch": 5.569185554974033,
-      "grad_norm": 0.0016555042238906026,
-      "learning_rate": 0.00019998470459949447,
-      "loss": 46.0,
-      "step": 34582
-    },
-    {
-      "epoch": 5.5693465920528205,
-      "grad_norm": 0.007202469278126955,
-      "learning_rate": 0.000199984703714661,
-      "loss": 46.0,
-      "step": 34583
-    },
-    {
-      "epoch": 5.569507629131608,
-      "grad_norm": 0.006697407457977533,
-      "learning_rate": 0.0001999847028298019,
-      "loss": 46.0,
-      "step": 34584
-    },
-    {
-      "epoch": 5.569668666210395,
-      "grad_norm": 0.007447336800396442,
-      "learning_rate": 0.00019998470194491726,
-      "loss": 46.0,
-      "step": 34585
-    },
-    {
-      "epoch": 5.569829703289182,
-      "grad_norm": 0.0025324360467493534,
-      "learning_rate": 0.00019998470106000698,
-      "loss": 46.0,
-      "step": 34586
-    },
-    {
-      "epoch": 5.569990740367969,
-      "grad_norm": 0.005339859053492546,
-      "learning_rate": 0.0001999847001750711,
-      "loss": 46.0,
-      "step": 34587
-    },
-    {
-      "epoch": 5.570151777446757,
-      "grad_norm": 0.0028478519525378942,
-      "learning_rate": 0.00019998469929010968,
-      "loss": 46.0,
-      "step": 34588
-    },
-    {
-      "epoch": 5.570312814525544,
-      "grad_norm": 0.0012775326613336802,
-      "learning_rate": 0.0001999846984051226,
-      "loss": 46.0,
-      "step": 34589
-    },
-    {
-      "epoch": 5.570473851604332,
-      "grad_norm": 0.0027482009027153254,
-      "learning_rate": 0.00019998469752011,
-      "loss": 46.0,
-      "step": 34590
-    },
-    {
-      "epoch": 5.570634888683119,
-      "grad_norm": 0.005615755449980497,
-      "learning_rate": 0.0001999846966350718,
-      "loss": 46.0,
-      "step": 34591
-    },
-    {
-      "epoch": 5.570795925761907,
-      "grad_norm": 0.0012348960153758526,
-      "learning_rate": 0.00019998469575000796,
-      "loss": 46.0,
-      "step": 34592
-    },
-    {
-      "epoch": 5.570956962840694,
-      "grad_norm": 0.0013820379972457886,
-      "learning_rate": 0.00019998469486491856,
-      "loss": 46.0,
-      "step": 34593
-    },
-    {
-      "epoch": 5.5711179999194815,
-      "grad_norm": 0.0035438458435237408,
-      "learning_rate": 0.00019998469397980358,
-      "loss": 46.0,
-      "step": 34594
-    },
-    {
-      "epoch": 5.571279036998269,
-      "grad_norm": 0.009423158131539822,
-      "learning_rate": 0.000199984693094663,
-      "loss": 46.0,
-      "step": 34595
-    },
-    {
-      "epoch": 5.571440074077056,
-      "grad_norm": 0.004822954535484314,
-      "learning_rate": 0.00019998469220949684,
-      "loss": 46.0,
-      "step": 34596
-    },
-    {
-      "epoch": 5.571601111155844,
-      "grad_norm": 0.0024205835070461035,
-      "learning_rate": 0.00019998469132430504,
-      "loss": 46.0,
-      "step": 34597
-    },
-    {
-      "epoch": 5.571762148234631,
-      "grad_norm": 0.0015072142705321312,
-      "learning_rate": 0.00019998469043908769,
-      "loss": 46.0,
-      "step": 34598
-    },
-    {
-      "epoch": 5.571923185313419,
-      "grad_norm": 0.0084876399487257,
-      "learning_rate": 0.00019998468955384472,
-      "loss": 46.0,
-      "step": 34599
-    },
-    {
-      "epoch": 5.572084222392206,
-      "grad_norm": 0.0026530916802585125,
-      "learning_rate": 0.00019998468866857621,
-      "loss": 46.0,
-      "step": 34600
-    },
-    {
-      "epoch": 5.572245259470993,
-      "grad_norm": 0.00257415440864861,
-      "learning_rate": 0.00019998468778328207,
-      "loss": 46.0,
-      "step": 34601
-    },
-    {
-      "epoch": 5.57240629654978,
-      "grad_norm": 0.009400319308042526,
-      "learning_rate": 0.00019998468689796234,
-      "loss": 46.0,
-      "step": 34602
-    },
-    {
-      "epoch": 5.572567333628568,
-      "grad_norm": 0.0017204693285748363,
-      "learning_rate": 0.00019998468601261705,
-      "loss": 46.0,
-      "step": 34603
-    },
-    {
-      "epoch": 5.572728370707355,
-      "grad_norm": 0.001527863903902471,
-      "learning_rate": 0.00019998468512724614,
-      "loss": 46.0,
-      "step": 34604
-    },
-    {
-      "epoch": 5.572889407786143,
-      "grad_norm": 0.009493887424468994,
-      "learning_rate": 0.00019998468424184965,
-      "loss": 46.0,
-      "step": 34605
-    },
-    {
-      "epoch": 5.57305044486493,
-      "grad_norm": 0.003111830912530422,
-      "learning_rate": 0.00019998468335642757,
-      "loss": 46.0,
-      "step": 34606
-    },
-    {
-      "epoch": 5.5732114819437175,
-      "grad_norm": 0.0009197515901178122,
-      "learning_rate": 0.00019998468247097988,
-      "loss": 46.0,
-      "step": 34607
-    },
-    {
-      "epoch": 5.573372519022505,
-      "grad_norm": 0.0014241928001865745,
-      "learning_rate": 0.0001999846815855066,
-      "loss": 46.0,
-      "step": 34608
-    },
-    {
-      "epoch": 5.573533556101292,
-      "grad_norm": 0.005001269746571779,
-      "learning_rate": 0.00019998468070000776,
-      "loss": 46.0,
-      "step": 34609
-    },
-    {
-      "epoch": 5.57369459318008,
-      "grad_norm": 0.005033680237829685,
-      "learning_rate": 0.00019998467981448333,
-      "loss": 46.0,
-      "step": 34610
-    },
-    {
-      "epoch": 5.573855630258867,
-      "grad_norm": 0.005431140307337046,
-      "learning_rate": 0.00019998467892893326,
-      "loss": 46.0,
-      "step": 34611
-    },
-    {
-      "epoch": 5.574016667337655,
-      "grad_norm": 0.001355149783194065,
-      "learning_rate": 0.00019998467804335763,
-      "loss": 46.0,
-      "step": 34612
-    },
-    {
-      "epoch": 5.574177704416442,
-      "grad_norm": 0.010457421652972698,
-      "learning_rate": 0.0001999846771577564,
-      "loss": 46.0,
-      "step": 34613
-    },
-    {
-      "epoch": 5.574338741495229,
-      "grad_norm": 0.0013484186492860317,
-      "learning_rate": 0.0001999846762721296,
-      "loss": 46.0,
-      "step": 34614
-    },
-    {
-      "epoch": 5.574499778574017,
-      "grad_norm": 0.0023423228412866592,
-      "learning_rate": 0.0001999846753864772,
-      "loss": 46.0,
-      "step": 34615
-    },
-    {
-      "epoch": 5.574660815652804,
-      "grad_norm": 0.0015337332151830196,
-      "learning_rate": 0.0001999846745007992,
-      "loss": 46.0,
-      "step": 34616
-    },
-    {
-      "epoch": 5.574821852731591,
-      "grad_norm": 0.04067227616906166,
-      "learning_rate": 0.00019998467361509562,
-      "loss": 46.0,
-      "step": 34617
-    },
-    {
-      "epoch": 5.574982889810379,
-      "grad_norm": 0.005900025833398104,
-      "learning_rate": 0.00019998467272936644,
-      "loss": 46.0,
-      "step": 34618
-    },
-    {
-      "epoch": 5.575143926889166,
-      "grad_norm": 0.0018213591538369656,
-      "learning_rate": 0.00019998467184361167,
-      "loss": 46.0,
-      "step": 34619
-    },
-    {
-      "epoch": 5.5753049639679535,
-      "grad_norm": 0.0012349592288956046,
-      "learning_rate": 0.00019998467095783134,
-      "loss": 46.0,
-      "step": 34620
-    },
-    {
-      "epoch": 5.575466001046741,
-      "grad_norm": 0.0075744627974927425,
-      "learning_rate": 0.00019998467007202537,
-      "loss": 46.0,
-      "step": 34621
-    },
-    {
-      "epoch": 5.575627038125528,
-      "grad_norm": 0.0007443575304932892,
-      "learning_rate": 0.00019998466918619382,
-      "loss": 46.0,
-      "step": 34622
-    },
-    {
-      "epoch": 5.575788075204316,
-      "grad_norm": 0.005646517965942621,
-      "learning_rate": 0.0001999846683003367,
-      "loss": 46.0,
-      "step": 34623
-    },
-    {
-      "epoch": 5.575949112283103,
-      "grad_norm": 0.0017934864154085517,
-      "learning_rate": 0.00019998466741445397,
-      "loss": 46.0,
-      "step": 34624
-    },
-    {
-      "epoch": 5.576110149361891,
-      "grad_norm": 0.00148340268060565,
-      "learning_rate": 0.00019998466652854566,
-      "loss": 46.0,
-      "step": 34625
-    },
-    {
-      "epoch": 5.576271186440678,
-      "grad_norm": 0.010009623132646084,
-      "learning_rate": 0.00019998466564261175,
-      "loss": 46.0,
-      "step": 34626
-    },
-    {
-      "epoch": 5.576432223519466,
-      "grad_norm": 0.012023997493088245,
-      "learning_rate": 0.00019998466475665226,
-      "loss": 46.0,
-      "step": 34627
-    },
-    {
-      "epoch": 5.576593260598253,
-      "grad_norm": 0.00648694159463048,
-      "learning_rate": 0.00019998466387066718,
-      "loss": 46.0,
-      "step": 34628
-    },
-    {
-      "epoch": 5.57675429767704,
-      "grad_norm": 0.011362219229340553,
-      "learning_rate": 0.0001999846629846565,
-      "loss": 46.0,
-      "step": 34629
-    },
-    {
-      "epoch": 5.576915334755828,
-      "grad_norm": 0.005906771868467331,
-      "learning_rate": 0.00019998466209862023,
-      "loss": 46.0,
-      "step": 34630
-    },
-    {
-      "epoch": 5.577076371834615,
-      "grad_norm": 0.005813187919557095,
-      "learning_rate": 0.0001999846612125584,
-      "loss": 46.0,
-      "step": 34631
-    },
-    {
-      "epoch": 5.577237408913402,
-      "grad_norm": 0.035298146307468414,
-      "learning_rate": 0.0001999846603264709,
-      "loss": 46.0,
-      "step": 34632
-    },
-    {
-      "epoch": 5.5773984459921895,
-      "grad_norm": 0.001976283034309745,
-      "learning_rate": 0.00019998465944035787,
-      "loss": 46.0,
-      "step": 34633
-    },
-    {
-      "epoch": 5.577559483070977,
-      "grad_norm": 0.01080408412963152,
-      "learning_rate": 0.00019998465855421924,
-      "loss": 46.0,
-      "step": 34634
-    },
-    {
-      "epoch": 5.577720520149764,
-      "grad_norm": 0.0017003521788865328,
-      "learning_rate": 0.00019998465766805503,
-      "loss": 46.0,
-      "step": 34635
-    },
-    {
-      "epoch": 5.577881557228552,
-      "grad_norm": 0.013147912919521332,
-      "learning_rate": 0.00019998465678186522,
-      "loss": 46.0,
-      "step": 34636
-    },
-    {
-      "epoch": 5.578042594307339,
-      "grad_norm": 0.0027957179117947817,
-      "learning_rate": 0.0001999846558956498,
-      "loss": 46.0,
-      "step": 34637
-    },
-    {
-      "epoch": 5.578203631386127,
-      "grad_norm": 0.02427743747830391,
-      "learning_rate": 0.00019998465500940883,
-      "loss": 46.0,
-      "step": 34638
-    },
-    {
-      "epoch": 5.578364668464914,
-      "grad_norm": 0.002231636783108115,
-      "learning_rate": 0.00019998465412314224,
-      "loss": 46.0,
-      "step": 34639
-    },
-    {
-      "epoch": 5.578525705543702,
-      "grad_norm": 0.007636507041752338,
-      "learning_rate": 0.00019998465323685006,
-      "loss": 46.0,
-      "step": 34640
-    },
-    {
-      "epoch": 5.578686742622489,
-      "grad_norm": 0.003328710561618209,
-      "learning_rate": 0.0001999846523505323,
-      "loss": 46.0,
-      "step": 34641
-    },
-    {
-      "epoch": 5.5788477797012765,
-      "grad_norm": 0.0014751259004697204,
-      "learning_rate": 0.00019998465146418894,
-      "loss": 46.0,
-      "step": 34642
-    },
-    {
-      "epoch": 5.579008816780064,
-      "grad_norm": 0.005650544073432684,
-      "learning_rate": 0.00019998465057781997,
-      "loss": 46.0,
-      "step": 34643
-    },
-    {
-      "epoch": 5.579169853858851,
-      "grad_norm": 0.001545062754303217,
-      "learning_rate": 0.00019998464969142544,
-      "loss": 46.0,
-      "step": 34644
-    },
-    {
-      "epoch": 5.579330890937638,
-      "grad_norm": 0.001414950587786734,
-      "learning_rate": 0.00019998464880500533,
-      "loss": 46.0,
-      "step": 34645
-    },
-    {
-      "epoch": 5.5794919280164255,
-      "grad_norm": 0.005363972391933203,
-      "learning_rate": 0.0001999846479185596,
-      "loss": 46.0,
-      "step": 34646
-    },
-    {
-      "epoch": 5.579652965095213,
-      "grad_norm": 0.004626765847206116,
-      "learning_rate": 0.00019998464703208826,
-      "loss": 46.0,
-      "step": 34647
-    },
-    {
-      "epoch": 5.579814002174,
-      "grad_norm": 0.003942246548831463,
-      "learning_rate": 0.00019998464614559138,
-      "loss": 46.0,
-      "step": 34648
-    },
-    {
-      "epoch": 5.579975039252788,
-      "grad_norm": 0.0029159795958548784,
-      "learning_rate": 0.0001999846452590689,
-      "loss": 46.0,
-      "step": 34649
-    },
-    {
-      "epoch": 5.580136076331575,
-      "grad_norm": 0.006351748947054148,
-      "learning_rate": 0.00019998464437252078,
-      "loss": 46.0,
-      "step": 34650
-    },
-    {
-      "epoch": 5.580297113410363,
-      "grad_norm": 0.00282082031480968,
-      "learning_rate": 0.0001999846434859471,
-      "loss": 46.0,
-      "step": 34651
-    },
-    {
-      "epoch": 5.58045815048915,
-      "grad_norm": 0.005532176699489355,
-      "learning_rate": 0.00019998464259934784,
-      "loss": 46.0,
-      "step": 34652
-    },
-    {
-      "epoch": 5.580619187567938,
-      "grad_norm": 0.026183107867836952,
-      "learning_rate": 0.00019998464171272297,
-      "loss": 46.0,
-      "step": 34653
-    },
-    {
-      "epoch": 5.580780224646725,
-      "grad_norm": 0.001215551164932549,
-      "learning_rate": 0.00019998464082607255,
-      "loss": 46.0,
-      "step": 34654
-    },
-    {
-      "epoch": 5.5809412617255125,
-      "grad_norm": 0.004192876163870096,
-      "learning_rate": 0.00019998463993939648,
-      "loss": 46.0,
-      "step": 34655
-    },
-    {
-      "epoch": 5.5811022988043,
-      "grad_norm": 0.0006772354245185852,
-      "learning_rate": 0.00019998463905269485,
-      "loss": 46.0,
-      "step": 34656
-    },
-    {
-      "epoch": 5.581263335883087,
-      "grad_norm": 0.007355592679232359,
-      "learning_rate": 0.0001999846381659676,
-      "loss": 46.0,
-      "step": 34657
-    },
-    {
-      "epoch": 5.581424372961875,
-      "grad_norm": 0.004486621357500553,
-      "learning_rate": 0.0001999846372792148,
-      "loss": 46.0,
-      "step": 34658
-    },
-    {
-      "epoch": 5.581585410040661,
-      "grad_norm": 0.0008767768740653992,
-      "learning_rate": 0.00019998463639243642,
-      "loss": 46.0,
-      "step": 34659
-    },
-    {
-      "epoch": 5.581746447119449,
-      "grad_norm": 0.020514369010925293,
-      "learning_rate": 0.0001999846355056324,
-      "loss": 46.0,
-      "step": 34660
-    },
-    {
-      "epoch": 5.581907484198236,
-      "grad_norm": 0.016994360834360123,
-      "learning_rate": 0.00019998463461880282,
-      "loss": 46.0,
-      "step": 34661
-    },
-    {
-      "epoch": 5.582068521277024,
-      "grad_norm": 0.008145978674292564,
-      "learning_rate": 0.00019998463373194767,
-      "loss": 46.0,
-      "step": 34662
-    },
-    {
-      "epoch": 5.582229558355811,
-      "grad_norm": 0.002680091420188546,
-      "learning_rate": 0.00019998463284506688,
-      "loss": 46.0,
-      "step": 34663
-    },
-    {
-      "epoch": 5.582390595434599,
-      "grad_norm": 0.005467541981488466,
-      "learning_rate": 0.00019998463195816053,
-      "loss": 46.0,
-      "step": 34664
-    },
-    {
-      "epoch": 5.582551632513386,
-      "grad_norm": 0.007323038298636675,
-      "learning_rate": 0.0001999846310712286,
-      "loss": 46.0,
-      "step": 34665
-    },
-    {
-      "epoch": 5.582712669592174,
-      "grad_norm": 0.00705747352913022,
-      "learning_rate": 0.00019998463018427103,
-      "loss": 46.0,
-      "step": 34666
-    },
-    {
-      "epoch": 5.582873706670961,
-      "grad_norm": 0.0039236838929355145,
-      "learning_rate": 0.0001999846292972879,
-      "loss": 46.0,
-      "step": 34667
-    },
-    {
-      "epoch": 5.5830347437497485,
-      "grad_norm": 0.006195586174726486,
-      "learning_rate": 0.0001999846284102792,
-      "loss": 46.0,
-      "step": 34668
-    },
-    {
-      "epoch": 5.583195780828536,
-      "grad_norm": 0.009123623371124268,
-      "learning_rate": 0.00019998462752324488,
-      "loss": 46.0,
-      "step": 34669
-    },
-    {
-      "epoch": 5.583356817907323,
-      "grad_norm": 0.0026506390422582626,
-      "learning_rate": 0.00019998462663618495,
-      "loss": 46.0,
-      "step": 34670
-    },
-    {
-      "epoch": 5.583517854986111,
-      "grad_norm": 0.007771295960992575,
-      "learning_rate": 0.00019998462574909946,
-      "loss": 46.0,
-      "step": 34671
-    },
-    {
-      "epoch": 5.583678892064898,
-      "grad_norm": 0.0029733378905802965,
-      "learning_rate": 0.00019998462486198838,
-      "loss": 46.0,
-      "step": 34672
-    },
-    {
-      "epoch": 5.583839929143686,
-      "grad_norm": 0.006273067090660334,
-      "learning_rate": 0.0001999846239748517,
-      "loss": 46.0,
-      "step": 34673
-    },
-    {
-      "epoch": 5.584000966222472,
-      "grad_norm": 0.004027407616376877,
-      "learning_rate": 0.00019998462308768944,
-      "loss": 46.0,
-      "step": 34674
-    },
-    {
-      "epoch": 5.58416200330126,
-      "grad_norm": 0.006385553162544966,
-      "learning_rate": 0.00019998462220050157,
-      "loss": 46.0,
-      "step": 34675
-    },
-    {
-      "epoch": 5.584323040380047,
-      "grad_norm": 0.001051747240126133,
-      "learning_rate": 0.0001999846213132881,
-      "loss": 46.0,
-      "step": 34676
-    },
-    {
-      "epoch": 5.584484077458835,
-      "grad_norm": 0.004667689558118582,
-      "learning_rate": 0.00019998462042604905,
-      "loss": 46.0,
-      "step": 34677
-    },
-    {
-      "epoch": 5.584645114537622,
-      "grad_norm": 0.0025501823984086514,
-      "learning_rate": 0.00019998461953878445,
-      "loss": 46.0,
-      "step": 34678
-    },
-    {
-      "epoch": 5.58480615161641,
-      "grad_norm": 0.015013315714895725,
-      "learning_rate": 0.0001999846186514942,
-      "loss": 46.0,
-      "step": 34679
-    },
-    {
-      "epoch": 5.584967188695197,
-      "grad_norm": 0.0013518910855054855,
-      "learning_rate": 0.0001999846177641784,
-      "loss": 46.0,
-      "step": 34680
-    },
-    {
-      "epoch": 5.5851282257739845,
-      "grad_norm": 0.0013906301464885473,
-      "learning_rate": 0.000199984616876837,
-      "loss": 46.0,
-      "step": 34681
-    },
-    {
-      "epoch": 5.585289262852772,
-      "grad_norm": 0.013587641529738903,
-      "learning_rate": 0.00019998461598947,
-      "loss": 46.0,
-      "step": 34682
-    },
-    {
-      "epoch": 5.585450299931559,
-      "grad_norm": 0.004045858047902584,
-      "learning_rate": 0.0001999846151020774,
-      "loss": 46.0,
-      "step": 34683
-    },
-    {
-      "epoch": 5.585611337010347,
-      "grad_norm": 0.012017167173326015,
-      "learning_rate": 0.00019998461421465925,
-      "loss": 46.0,
-      "step": 34684
-    },
-    {
-      "epoch": 5.585772374089134,
-      "grad_norm": 0.006830972619354725,
-      "learning_rate": 0.00019998461332721546,
-      "loss": 46.0,
-      "step": 34685
-    },
-    {
-      "epoch": 5.585933411167922,
-      "grad_norm": 0.0012241060612723231,
-      "learning_rate": 0.0001999846124397461,
-      "loss": 46.0,
-      "step": 34686
-    },
-    {
-      "epoch": 5.586094448246708,
-      "grad_norm": 0.006106555461883545,
-      "learning_rate": 0.00019998461155225114,
-      "loss": 46.0,
-      "step": 34687
-    },
-    {
-      "epoch": 5.586255485325497,
-      "grad_norm": 0.0015198610490188003,
-      "learning_rate": 0.0001999846106647306,
-      "loss": 46.0,
-      "step": 34688
-    },
-    {
-      "epoch": 5.586416522404283,
-      "grad_norm": 0.006363158114254475,
-      "learning_rate": 0.00019998460977718448,
-      "loss": 46.0,
-      "step": 34689
-    },
-    {
-      "epoch": 5.586577559483071,
-      "grad_norm": 0.002384186489507556,
-      "learning_rate": 0.00019998460888961278,
-      "loss": 46.0,
-      "step": 34690
-    },
-    {
-      "epoch": 5.586738596561858,
-      "grad_norm": 0.013466891832649708,
-      "learning_rate": 0.00019998460800201544,
-      "loss": 46.0,
-      "step": 34691
-    },
-    {
-      "epoch": 5.586899633640646,
-      "grad_norm": 0.0026554784271866083,
-      "learning_rate": 0.00019998460711439253,
-      "loss": 46.0,
-      "step": 34692
-    },
-    {
-      "epoch": 5.587060670719433,
-      "grad_norm": 0.010037070140242577,
-      "learning_rate": 0.00019998460622674404,
-      "loss": 46.0,
-      "step": 34693
-    },
-    {
-      "epoch": 5.5872217077982205,
-      "grad_norm": 0.0013583747204393148,
-      "learning_rate": 0.00019998460533906994,
-      "loss": 46.0,
-      "step": 34694
-    },
-    {
-      "epoch": 5.587382744877008,
-      "grad_norm": 0.003949266392737627,
-      "learning_rate": 0.00019998460445137028,
-      "loss": 46.0,
-      "step": 34695
-    },
-    {
-      "epoch": 5.587543781955795,
-      "grad_norm": 0.003091406775638461,
-      "learning_rate": 0.000199984603563645,
-      "loss": 46.0,
-      "step": 34696
-    },
-    {
-      "epoch": 5.587704819034583,
-      "grad_norm": 0.0037286831066012383,
-      "learning_rate": 0.00019998460267589414,
-      "loss": 46.0,
-      "step": 34697
-    },
-    {
-      "epoch": 5.58786585611337,
-      "grad_norm": 0.0180059764534235,
-      "learning_rate": 0.00019998460178811768,
-      "loss": 46.0,
-      "step": 34698
-    },
-    {
-      "epoch": 5.588026893192158,
-      "grad_norm": 0.0033999825827777386,
-      "learning_rate": 0.00019998460090031565,
-      "loss": 46.0,
-      "step": 34699
-    },
-    {
-      "epoch": 5.588187930270945,
-      "grad_norm": 0.004126907326281071,
-      "learning_rate": 0.00019998460001248802,
-      "loss": 46.0,
-      "step": 34700
-    },
-    {
-      "epoch": 5.588348967349733,
-      "grad_norm": 0.0020060690585523844,
-      "learning_rate": 0.00019998459912463478,
-      "loss": 46.0,
-      "step": 34701
-    },
-    {
-      "epoch": 5.588510004428519,
-      "grad_norm": 0.012995957396924496,
-      "learning_rate": 0.000199984598236756,
-      "loss": 46.0,
-      "step": 34702
-    },
-    {
-      "epoch": 5.5886710415073075,
-      "grad_norm": 0.012879100628197193,
-      "learning_rate": 0.00019998459734885156,
-      "loss": 46.0,
-      "step": 34703
-    },
-    {
-      "epoch": 5.588832078586094,
-      "grad_norm": 0.004682404454797506,
-      "learning_rate": 0.0001999845964609216,
-      "loss": 46.0,
-      "step": 34704
-    },
-    {
-      "epoch": 5.5889931156648816,
-      "grad_norm": 0.0017997967079281807,
-      "learning_rate": 0.000199984595572966,
-      "loss": 46.0,
-      "step": 34705
-    },
-    {
-      "epoch": 5.589154152743669,
-      "grad_norm": 0.0016788548091426492,
-      "learning_rate": 0.0001999845946849848,
-      "loss": 46.0,
-      "step": 34706
-    },
-    {
-      "epoch": 5.5893151898224565,
-      "grad_norm": 0.011671469546854496,
-      "learning_rate": 0.00019998459379697803,
-      "loss": 46.0,
-      "step": 34707
-    },
-    {
-      "epoch": 5.589476226901244,
-      "grad_norm": 0.01430121622979641,
-      "learning_rate": 0.00019998459290894566,
-      "loss": 46.0,
-      "step": 34708
-    },
-    {
-      "epoch": 5.589637263980031,
-      "grad_norm": 0.001300667878240347,
-      "learning_rate": 0.00019998459202088772,
-      "loss": 46.0,
-      "step": 34709
-    },
-    {
-      "epoch": 5.589798301058819,
-      "grad_norm": 0.01275206170976162,
-      "learning_rate": 0.0001999845911328042,
-      "loss": 46.0,
-      "step": 34710
-    },
-    {
-      "epoch": 5.589959338137606,
-      "grad_norm": 0.006128380540758371,
-      "learning_rate": 0.00019998459024469508,
-      "loss": 46.0,
-      "step": 34711
-    },
-    {
-      "epoch": 5.590120375216394,
-      "grad_norm": 0.012952849268913269,
-      "learning_rate": 0.00019998458935656033,
-      "loss": 46.0,
-      "step": 34712
-    },
-    {
-      "epoch": 5.590281412295181,
-      "grad_norm": 0.0019022960914298892,
-      "learning_rate": 0.00019998458846840001,
-      "loss": 46.0,
-      "step": 34713
-    },
-    {
-      "epoch": 5.590442449373969,
-      "grad_norm": 0.005407972261309624,
-      "learning_rate": 0.00019998458758021411,
-      "loss": 46.0,
-      "step": 34714
-    },
-    {
-      "epoch": 5.590603486452756,
-      "grad_norm": 0.0009761581895872951,
-      "learning_rate": 0.00019998458669200263,
-      "loss": 46.0,
-      "step": 34715
-    },
-    {
-      "epoch": 5.5907645235315435,
-      "grad_norm": 0.004773018881678581,
-      "learning_rate": 0.00019998458580376552,
-      "loss": 46.0,
-      "step": 34716
-    },
-    {
-      "epoch": 5.59092556061033,
-      "grad_norm": 0.0037342195864766836,
-      "learning_rate": 0.00019998458491550286,
-      "loss": 46.0,
-      "step": 34717
-    },
-    {
-      "epoch": 5.5910865976891175,
-      "grad_norm": 0.010398722253739834,
-      "learning_rate": 0.00019998458402721458,
-      "loss": 46.0,
-      "step": 34718
-    },
-    {
-      "epoch": 5.591247634767905,
-      "grad_norm": 0.002075908239930868,
-      "learning_rate": 0.00019998458313890072,
-      "loss": 46.0,
-      "step": 34719
-    },
-    {
-      "epoch": 5.591408671846692,
-      "grad_norm": 0.0018005879828706384,
-      "learning_rate": 0.00019998458225056127,
-      "loss": 46.0,
-      "step": 34720
-    },
-    {
-      "epoch": 5.59156970892548,
-      "grad_norm": 0.003978176973760128,
-      "learning_rate": 0.00019998458136219623,
-      "loss": 46.0,
-      "step": 34721
-    },
-    {
-      "epoch": 5.591730746004267,
-      "grad_norm": 0.001142253284342587,
-      "learning_rate": 0.0001999845804738056,
-      "loss": 46.0,
-      "step": 34722
-    },
-    {
-      "epoch": 5.591891783083055,
-      "grad_norm": 0.0032236415427178144,
-      "learning_rate": 0.00019998457958538937,
-      "loss": 46.0,
-      "step": 34723
-    },
-    {
-      "epoch": 5.592052820161842,
-      "grad_norm": 0.0014490430476143956,
-      "learning_rate": 0.00019998457869694757,
-      "loss": 46.0,
-      "step": 34724
-    },
-    {
-      "epoch": 5.59221385724063,
-      "grad_norm": 0.002264301525428891,
-      "learning_rate": 0.00019998457780848015,
-      "loss": 46.0,
-      "step": 34725
-    },
-    {
-      "epoch": 5.592374894319417,
-      "grad_norm": 0.0034744602162390947,
-      "learning_rate": 0.00019998457691998715,
-      "loss": 46.0,
-      "step": 34726
-    },
-    {
-      "epoch": 5.592535931398205,
-      "grad_norm": 0.0016214636852964759,
-      "learning_rate": 0.00019998457603146857,
-      "loss": 46.0,
-      "step": 34727
-    },
-    {
-      "epoch": 5.592696968476992,
-      "grad_norm": 0.0033249491825699806,
-      "learning_rate": 0.0001999845751429244,
-      "loss": 46.0,
-      "step": 34728
-    },
-    {
-      "epoch": 5.5928580055557795,
-      "grad_norm": 0.007058362010866404,
-      "learning_rate": 0.0001999845742543546,
-      "loss": 46.0,
-      "step": 34729
-    },
-    {
-      "epoch": 5.593019042634567,
-      "grad_norm": 0.006156631745398045,
-      "learning_rate": 0.00019998457336575925,
-      "loss": 46.0,
-      "step": 34730
-    },
-    {
-      "epoch": 5.593180079713354,
-      "grad_norm": 0.008450658991932869,
-      "learning_rate": 0.0001999845724771383,
-      "loss": 46.0,
-      "step": 34731
-    },
-    {
-      "epoch": 5.593341116792141,
-      "grad_norm": 0.003843887010589242,
-      "learning_rate": 0.00019998457158849176,
-      "loss": 46.0,
-      "step": 34732
-    },
-    {
-      "epoch": 5.593502153870928,
-      "grad_norm": 0.004342313855886459,
-      "learning_rate": 0.00019998457069981962,
-      "loss": 46.0,
-      "step": 34733
-    },
-    {
-      "epoch": 5.593663190949716,
-      "grad_norm": 0.002976346295326948,
-      "learning_rate": 0.00019998456981112187,
-      "loss": 46.0,
-      "step": 34734
-    },
-    {
-      "epoch": 5.593824228028503,
-      "grad_norm": 0.002610750962048769,
-      "learning_rate": 0.00019998456892239856,
-      "loss": 46.0,
-      "step": 34735
-    },
-    {
-      "epoch": 5.593985265107291,
-      "grad_norm": 0.0047075035981833935,
-      "learning_rate": 0.00019998456803364966,
-      "loss": 46.0,
-      "step": 34736
-    },
-    {
-      "epoch": 5.594146302186078,
-      "grad_norm": 0.0028581772930920124,
-      "learning_rate": 0.00019998456714487517,
-      "loss": 46.0,
-      "step": 34737
-    },
-    {
-      "epoch": 5.594307339264866,
-      "grad_norm": 0.0022644835989922285,
-      "learning_rate": 0.0001999845662560751,
-      "loss": 46.0,
-      "step": 34738
-    },
-    {
-      "epoch": 5.594468376343653,
-      "grad_norm": 0.004189268220216036,
-      "learning_rate": 0.0001999845653672494,
-      "loss": 46.0,
-      "step": 34739
-    },
-    {
-      "epoch": 5.594629413422441,
-      "grad_norm": 0.00970964040607214,
-      "learning_rate": 0.00019998456447839816,
-      "loss": 46.0,
-      "step": 34740
-    },
-    {
-      "epoch": 5.594790450501228,
-      "grad_norm": 0.003984611015766859,
-      "learning_rate": 0.00019998456358952127,
-      "loss": 46.0,
-      "step": 34741
-    },
-    {
-      "epoch": 5.5949514875800155,
-      "grad_norm": 0.0022746813483536243,
-      "learning_rate": 0.00019998456270061882,
-      "loss": 46.0,
-      "step": 34742
-    },
-    {
-      "epoch": 5.595112524658803,
-      "grad_norm": 0.002649534959346056,
-      "learning_rate": 0.00019998456181169079,
-      "loss": 46.0,
-      "step": 34743
-    },
-    {
-      "epoch": 5.59527356173759,
-      "grad_norm": 0.0070677027106285095,
-      "learning_rate": 0.00019998456092273713,
-      "loss": 46.0,
-      "step": 34744
-    },
-    {
-      "epoch": 5.595434598816378,
-      "grad_norm": 0.0032268110662698746,
-      "learning_rate": 0.00019998456003375792,
-      "loss": 46.0,
-      "step": 34745
-    },
-    {
-      "epoch": 5.595595635895165,
-      "grad_norm": 0.004400305915623903,
-      "learning_rate": 0.00019998455914475312,
-      "loss": 46.0,
-      "step": 34746
-    },
-    {
-      "epoch": 5.595756672973952,
-      "grad_norm": 0.0008275015861727297,
-      "learning_rate": 0.0001999845582557227,
-      "loss": 46.0,
-      "step": 34747
-    },
-    {
-      "epoch": 5.595917710052739,
-      "grad_norm": 0.009321720339357853,
-      "learning_rate": 0.0001999845573666667,
-      "loss": 46.0,
-      "step": 34748
-    },
-    {
-      "epoch": 5.596078747131527,
-      "grad_norm": 0.004384103696793318,
-      "learning_rate": 0.00019998455647758512,
-      "loss": 46.0,
-      "step": 34749
-    },
-    {
-      "epoch": 5.596239784210314,
-      "grad_norm": 0.0025092782452702522,
-      "learning_rate": 0.00019998455558847792,
-      "loss": 46.0,
-      "step": 34750
-    },
-    {
-      "epoch": 5.596400821289102,
-      "grad_norm": 0.004068761132657528,
-      "learning_rate": 0.00019998455469934516,
-      "loss": 46.0,
-      "step": 34751
-    },
-    {
-      "epoch": 5.596561858367889,
-      "grad_norm": 0.005694488994777203,
-      "learning_rate": 0.0001999845538101868,
-      "loss": 46.0,
-      "step": 34752
-    },
-    {
-      "epoch": 5.596722895446677,
-      "grad_norm": 0.005226957146078348,
-      "learning_rate": 0.00019998455292100285,
-      "loss": 46.0,
-      "step": 34753
-    },
-    {
-      "epoch": 5.596883932525464,
-      "grad_norm": 0.003106221556663513,
-      "learning_rate": 0.0001999845520317933,
-      "loss": 46.0,
-      "step": 34754
-    },
-    {
-      "epoch": 5.5970449696042515,
-      "grad_norm": 0.002458963543176651,
-      "learning_rate": 0.00019998455114255816,
-      "loss": 46.0,
-      "step": 34755
-    },
-    {
-      "epoch": 5.597206006683039,
-      "grad_norm": 0.0013307415647432208,
-      "learning_rate": 0.00019998455025329746,
-      "loss": 46.0,
-      "step": 34756
-    },
-    {
-      "epoch": 5.597367043761826,
-      "grad_norm": 0.008632835932075977,
-      "learning_rate": 0.00019998454936401113,
-      "loss": 46.0,
-      "step": 34757
-    },
-    {
-      "epoch": 5.597528080840614,
-      "grad_norm": 0.010196693241596222,
-      "learning_rate": 0.00019998454847469923,
-      "loss": 46.0,
-      "step": 34758
-    },
-    {
-      "epoch": 5.597689117919401,
-      "grad_norm": 0.012990818358957767,
-      "learning_rate": 0.0001999845475853617,
-      "loss": 46.0,
-      "step": 34759
-    },
-    {
-      "epoch": 5.597850154998189,
-      "grad_norm": 0.008843060582876205,
-      "learning_rate": 0.00019998454669599864,
-      "loss": 46.0,
-      "step": 34760
-    },
-    {
-      "epoch": 5.598011192076976,
-      "grad_norm": 0.004096719902008772,
-      "learning_rate": 0.00019998454580660995,
-      "loss": 46.0,
-      "step": 34761
-    },
-    {
-      "epoch": 5.598172229155763,
-      "grad_norm": 0.002911268500611186,
-      "learning_rate": 0.0001999845449171957,
-      "loss": 46.0,
-      "step": 34762
-    },
-    {
-      "epoch": 5.59833326623455,
-      "grad_norm": 0.013834027573466301,
-      "learning_rate": 0.00019998454402775582,
-      "loss": 46.0,
-      "step": 34763
-    },
-    {
-      "epoch": 5.598494303313338,
-      "grad_norm": 0.0027927388437092304,
-      "learning_rate": 0.0001999845431382904,
-      "loss": 46.0,
-      "step": 34764
-    },
-    {
-      "epoch": 5.598655340392125,
-      "grad_norm": 0.002818124135956168,
-      "learning_rate": 0.00019998454224879933,
-      "loss": 46.0,
-      "step": 34765
-    },
-    {
-      "epoch": 5.5988163774709125,
-      "grad_norm": 0.008320937864482403,
-      "learning_rate": 0.0001999845413592827,
-      "loss": 46.0,
-      "step": 34766
-    },
-    {
-      "epoch": 5.5989774145497,
-      "grad_norm": 0.010386254638433456,
-      "learning_rate": 0.00019998454046974047,
-      "loss": 46.0,
-      "step": 34767
-    },
-    {
-      "epoch": 5.5991384516284874,
-      "grad_norm": 0.0020796344615519047,
-      "learning_rate": 0.00019998453958017267,
-      "loss": 46.0,
-      "step": 34768
-    },
-    {
-      "epoch": 5.599299488707275,
-      "grad_norm": 0.001319212606176734,
-      "learning_rate": 0.00019998453869057926,
-      "loss": 46.0,
-      "step": 34769
-    },
-    {
-      "epoch": 5.599460525786062,
-      "grad_norm": 0.004777243360877037,
-      "learning_rate": 0.00019998453780096026,
-      "loss": 46.0,
-      "step": 34770
-    },
-    {
-      "epoch": 5.59962156286485,
-      "grad_norm": 0.0022300691343843937,
-      "learning_rate": 0.00019998453691131567,
-      "loss": 46.0,
-      "step": 34771
-    },
-    {
-      "epoch": 5.599782599943637,
-      "grad_norm": 0.009431141428649426,
-      "learning_rate": 0.00019998453602164547,
-      "loss": 46.0,
-      "step": 34772
-    },
-    {
-      "epoch": 5.599943637022425,
-      "grad_norm": 0.0006995851290412247,
-      "learning_rate": 0.0001999845351319497,
-      "loss": 46.0,
-      "step": 34773
-    },
-    {
-      "epoch": 5.600104674101212,
-      "grad_norm": 0.0018728980794548988,
-      "learning_rate": 0.00019998453424222837,
-      "loss": 46.0,
-      "step": 34774
-    },
-    {
-      "epoch": 5.600265711179999,
-      "grad_norm": 0.010138577781617641,
-      "learning_rate": 0.00019998453335248143,
-      "loss": 46.0,
-      "step": 34775
-    },
-    {
-      "epoch": 5.600426748258787,
-      "grad_norm": 0.006323046050965786,
-      "learning_rate": 0.00019998453246270885,
-      "loss": 46.0,
-      "step": 34776
-    },
-    {
-      "epoch": 5.600587785337574,
-      "grad_norm": 0.0011384043609723449,
-      "learning_rate": 0.00019998453157291072,
-      "loss": 46.0,
-      "step": 34777
-    },
-    {
-      "epoch": 5.600748822416361,
-      "grad_norm": 0.009958439506590366,
-      "learning_rate": 0.000199984530683087,
-      "loss": 46.0,
-      "step": 34778
-    },
-    {
-      "epoch": 5.6009098594951485,
-      "grad_norm": 0.007385083008557558,
-      "learning_rate": 0.00019998452979323768,
-      "loss": 46.0,
-      "step": 34779
-    },
-    {
-      "epoch": 5.601070896573936,
-      "grad_norm": 0.002337296027690172,
-      "learning_rate": 0.00019998452890336279,
-      "loss": 46.0,
-      "step": 34780
-    },
-    {
-      "epoch": 5.601231933652723,
-      "grad_norm": 0.0006477715214714408,
-      "learning_rate": 0.00019998452801346227,
-      "loss": 46.0,
-      "step": 34781
-    },
-    {
-      "epoch": 5.601392970731511,
-      "grad_norm": 0.00799001194536686,
-      "learning_rate": 0.0001999845271235362,
-      "loss": 46.0,
-      "step": 34782
-    },
-    {
-      "epoch": 5.601554007810298,
-      "grad_norm": 0.0018834640504792333,
-      "learning_rate": 0.00019998452623358451,
-      "loss": 46.0,
-      "step": 34783
-    },
-    {
-      "epoch": 5.601715044889086,
-      "grad_norm": 0.0038098585791885853,
-      "learning_rate": 0.00019998452534360724,
-      "loss": 46.0,
-      "step": 34784
-    },
-    {
-      "epoch": 5.601876081967873,
-      "grad_norm": 0.006432806141674519,
-      "learning_rate": 0.00019998452445360438,
-      "loss": 46.0,
-      "step": 34785
-    },
-    {
-      "epoch": 5.602037119046661,
-      "grad_norm": 0.0069917780347168446,
-      "learning_rate": 0.00019998452356357593,
-      "loss": 46.0,
-      "step": 34786
-    },
-    {
-      "epoch": 5.602198156125448,
-      "grad_norm": 0.006083559710532427,
-      "learning_rate": 0.00019998452267352187,
-      "loss": 46.0,
-      "step": 34787
-    },
-    {
-      "epoch": 5.602359193204236,
-      "grad_norm": 0.0023505445569753647,
-      "learning_rate": 0.00019998452178344227,
-      "loss": 46.0,
-      "step": 34788
-    },
-    {
-      "epoch": 5.602520230283023,
-      "grad_norm": 0.004209406673908234,
-      "learning_rate": 0.00019998452089333704,
-      "loss": 46.0,
-      "step": 34789
-    },
-    {
-      "epoch": 5.60268126736181,
-      "grad_norm": 0.00223302049562335,
-      "learning_rate": 0.0001999845200032062,
-      "loss": 46.0,
-      "step": 34790
-    },
-    {
-      "epoch": 5.602842304440598,
-      "grad_norm": 0.001660041161812842,
-      "learning_rate": 0.0001999845191130498,
-      "loss": 46.0,
-      "step": 34791
-    },
-    {
-      "epoch": 5.6030033415193845,
-      "grad_norm": 0.004566125106066465,
-      "learning_rate": 0.0001999845182228678,
-      "loss": 46.0,
-      "step": 34792
-    },
-    {
-      "epoch": 5.603164378598172,
-      "grad_norm": 0.00858062133193016,
-      "learning_rate": 0.00019998451733266022,
-      "loss": 46.0,
-      "step": 34793
-    },
-    {
-      "epoch": 5.603325415676959,
-      "grad_norm": 0.013128002174198627,
-      "learning_rate": 0.00019998451644242705,
-      "loss": 46.0,
-      "step": 34794
-    },
-    {
-      "epoch": 5.603486452755747,
-      "grad_norm": 0.0027794018387794495,
-      "learning_rate": 0.0001999845155521683,
-      "loss": 46.0,
-      "step": 34795
-    },
-    {
-      "epoch": 5.603647489834534,
-      "grad_norm": 0.009374003857374191,
-      "learning_rate": 0.0001999845146618839,
-      "loss": 46.0,
-      "step": 34796
-    },
-    {
-      "epoch": 5.603808526913322,
-      "grad_norm": 0.011360440403223038,
-      "learning_rate": 0.00019998451377157398,
-      "loss": 46.0,
-      "step": 34797
-    },
-    {
-      "epoch": 5.603969563992109,
-      "grad_norm": 0.005560627672821283,
-      "learning_rate": 0.0001999845128812384,
-      "loss": 46.0,
-      "step": 34798
-    },
-    {
-      "epoch": 5.604130601070897,
-      "grad_norm": 0.009031318128108978,
-      "learning_rate": 0.0001999845119908773,
-      "loss": 46.0,
-      "step": 34799
-    },
-    {
-      "epoch": 5.604291638149684,
-      "grad_norm": 0.007395155727863312,
-      "learning_rate": 0.00019998451110049054,
-      "loss": 46.0,
-      "step": 34800
-    },
-    {
-      "epoch": 5.604452675228472,
-      "grad_norm": 0.0023184281308203936,
-      "learning_rate": 0.00019998451021007824,
-      "loss": 46.0,
-      "step": 34801
-    },
-    {
-      "epoch": 5.604613712307259,
-      "grad_norm": 0.0031362087465822697,
-      "learning_rate": 0.0001999845093196403,
-      "loss": 46.0,
-      "step": 34802
-    },
-    {
-      "epoch": 5.6047747493860465,
-      "grad_norm": 0.0045433202758431435,
-      "learning_rate": 0.00019998450842917685,
-      "loss": 46.0,
-      "step": 34803
-    },
-    {
-      "epoch": 5.604935786464834,
-      "grad_norm": 0.012602468021214008,
-      "learning_rate": 0.00019998450753868773,
-      "loss": 46.0,
-      "step": 34804
-    },
-    {
-      "epoch": 5.6050968235436205,
-      "grad_norm": 0.0040380205027759075,
-      "learning_rate": 0.00019998450664817307,
-      "loss": 46.0,
-      "step": 34805
-    },
-    {
-      "epoch": 5.605257860622408,
-      "grad_norm": 0.00943379383534193,
-      "learning_rate": 0.0001999845057576328,
-      "loss": 46.0,
-      "step": 34806
-    },
-    {
-      "epoch": 5.605418897701195,
-      "grad_norm": 0.007022729143500328,
-      "learning_rate": 0.00019998450486706694,
-      "loss": 46.0,
-      "step": 34807
-    },
-    {
-      "epoch": 5.605579934779983,
-      "grad_norm": 0.0029596660751849413,
-      "learning_rate": 0.0001999845039764755,
-      "loss": 46.0,
-      "step": 34808
-    },
-    {
-      "epoch": 5.60574097185877,
-      "grad_norm": 0.01099966000765562,
-      "learning_rate": 0.00019998450308585843,
-      "loss": 46.0,
-      "step": 34809
-    },
-    {
-      "epoch": 5.605902008937558,
-      "grad_norm": 0.004348713904619217,
-      "learning_rate": 0.0001999845021952158,
-      "loss": 46.0,
-      "step": 34810
-    },
-    {
-      "epoch": 5.606063046016345,
-      "grad_norm": 0.0012874577660113573,
-      "learning_rate": 0.0001999845013045476,
-      "loss": 46.0,
-      "step": 34811
-    },
-    {
-      "epoch": 5.606224083095133,
-      "grad_norm": 0.002988834399729967,
-      "learning_rate": 0.00019998450041385378,
-      "loss": 46.0,
-      "step": 34812
-    },
-    {
-      "epoch": 5.60638512017392,
-      "grad_norm": 0.016629081219434738,
-      "learning_rate": 0.00019998449952313437,
-      "loss": 46.0,
-      "step": 34813
-    },
-    {
-      "epoch": 5.6065461572527076,
-      "grad_norm": 0.013277328573167324,
-      "learning_rate": 0.00019998449863238937,
-      "loss": 46.0,
-      "step": 34814
-    },
-    {
-      "epoch": 5.606707194331495,
-      "grad_norm": 0.0037772334180772305,
-      "learning_rate": 0.00019998449774161878,
-      "loss": 46.0,
-      "step": 34815
-    },
-    {
-      "epoch": 5.6068682314102825,
-      "grad_norm": 0.002887663198634982,
-      "learning_rate": 0.0001999844968508226,
-      "loss": 46.0,
-      "step": 34816
-    },
-    {
-      "epoch": 5.60702926848907,
-      "grad_norm": 0.003901789430528879,
-      "learning_rate": 0.00019998449596000085,
-      "loss": 46.0,
-      "step": 34817
-    },
-    {
-      "epoch": 5.607190305567857,
-      "grad_norm": 0.0009883488528430462,
-      "learning_rate": 0.00019998449506915345,
-      "loss": 46.0,
-      "step": 34818
-    },
-    {
-      "epoch": 5.607351342646645,
-      "grad_norm": 0.003526639426127076,
-      "learning_rate": 0.00019998449417828052,
-      "loss": 46.0,
-      "step": 34819
-    },
-    {
-      "epoch": 5.607512379725431,
-      "grad_norm": 0.014215493574738503,
-      "learning_rate": 0.00019998449328738197,
-      "loss": 46.0,
-      "step": 34820
-    },
-    {
-      "epoch": 5.607673416804219,
-      "grad_norm": 0.006592354737222195,
-      "learning_rate": 0.00019998449239645784,
-      "loss": 46.0,
-      "step": 34821
-    },
-    {
-      "epoch": 5.607834453883006,
-      "grad_norm": 0.0026498534716665745,
-      "learning_rate": 0.00019998449150550812,
-      "loss": 46.0,
-      "step": 34822
-    },
-    {
-      "epoch": 5.607995490961794,
-      "grad_norm": 0.0035986576694995165,
-      "learning_rate": 0.0001999844906145328,
-      "loss": 46.0,
-      "step": 34823
-    },
-    {
-      "epoch": 5.608156528040581,
-      "grad_norm": 0.0033094820100814104,
-      "learning_rate": 0.0001999844897235319,
-      "loss": 46.0,
-      "step": 34824
-    },
-    {
-      "epoch": 5.608317565119369,
-      "grad_norm": 0.004507584497332573,
-      "learning_rate": 0.0001999844888325054,
-      "loss": 46.0,
-      "step": 34825
-    },
-    {
-      "epoch": 5.608478602198156,
-      "grad_norm": 0.009295037016272545,
-      "learning_rate": 0.00019998448794145333,
-      "loss": 46.0,
-      "step": 34826
-    },
-    {
-      "epoch": 5.6086396392769435,
-      "grad_norm": 0.008341183885931969,
-      "learning_rate": 0.00019998448705037562,
-      "loss": 46.0,
-      "step": 34827
-    },
-    {
-      "epoch": 5.608800676355731,
-      "grad_norm": 0.0016650771722197533,
-      "learning_rate": 0.00019998448615927237,
-      "loss": 46.0,
-      "step": 34828
-    },
-    {
-      "epoch": 5.608961713434518,
-      "grad_norm": 0.01993122324347496,
-      "learning_rate": 0.0001999844852681435,
-      "loss": 46.0,
-      "step": 34829
-    },
-    {
-      "epoch": 5.609122750513306,
-      "grad_norm": 0.002327660797163844,
-      "learning_rate": 0.00019998448437698904,
-      "loss": 46.0,
-      "step": 34830
-    },
-    {
-      "epoch": 5.609283787592093,
-      "grad_norm": 0.00765073811635375,
-      "learning_rate": 0.000199984483485809,
-      "loss": 46.0,
-      "step": 34831
-    },
-    {
-      "epoch": 5.609444824670881,
-      "grad_norm": 0.011233063414692879,
-      "learning_rate": 0.0001999844825946034,
-      "loss": 46.0,
-      "step": 34832
-    },
-    {
-      "epoch": 5.609605861749668,
-      "grad_norm": 0.005589455831795931,
-      "learning_rate": 0.00019998448170337215,
-      "loss": 46.0,
-      "step": 34833
-    },
-    {
-      "epoch": 5.609766898828456,
-      "grad_norm": 0.01829095557332039,
-      "learning_rate": 0.00019998448081211533,
-      "loss": 46.0,
-      "step": 34834
-    },
-    {
-      "epoch": 5.609927935907242,
-      "grad_norm": 0.001708773779682815,
-      "learning_rate": 0.00019998447992083295,
-      "loss": 46.0,
-      "step": 34835
-    },
-    {
-      "epoch": 5.61008897298603,
-      "grad_norm": 0.010568246245384216,
-      "learning_rate": 0.00019998447902952492,
-      "loss": 46.0,
-      "step": 34836
-    },
-    {
-      "epoch": 5.610250010064817,
-      "grad_norm": 0.002612903481349349,
-      "learning_rate": 0.00019998447813819134,
-      "loss": 46.0,
-      "step": 34837
-    },
-    {
-      "epoch": 5.610411047143605,
-      "grad_norm": 0.007799685001373291,
-      "learning_rate": 0.00019998447724683217,
-      "loss": 46.0,
-      "step": 34838
-    },
-    {
-      "epoch": 5.610572084222392,
-      "grad_norm": 0.012057727202773094,
-      "learning_rate": 0.00019998447635544739,
-      "loss": 46.0,
-      "step": 34839
-    },
-    {
-      "epoch": 5.6107331213011795,
-      "grad_norm": 0.010302739217877388,
-      "learning_rate": 0.000199984475464037,
-      "loss": 46.0,
-      "step": 34840
-    },
-    {
-      "epoch": 5.610894158379967,
-      "grad_norm": 0.004059287719428539,
-      "learning_rate": 0.00019998447457260108,
-      "loss": 46.0,
-      "step": 34841
-    },
-    {
-      "epoch": 5.611055195458754,
-      "grad_norm": 0.010391690768301487,
-      "learning_rate": 0.00019998447368113956,
-      "loss": 46.0,
-      "step": 34842
-    },
-    {
-      "epoch": 5.611216232537542,
-      "grad_norm": 0.0031577947083860636,
-      "learning_rate": 0.0001999844727896524,
-      "loss": 46.0,
-      "step": 34843
-    },
-    {
-      "epoch": 5.611377269616329,
-      "grad_norm": 0.006777295842766762,
-      "learning_rate": 0.00019998447189813968,
-      "loss": 46.0,
-      "step": 34844
-    },
-    {
-      "epoch": 5.611538306695117,
-      "grad_norm": 0.004861742723733187,
-      "learning_rate": 0.00019998447100660137,
-      "loss": 46.0,
-      "step": 34845
-    },
-    {
-      "epoch": 5.611699343773904,
-      "grad_norm": 0.006205307319760323,
-      "learning_rate": 0.00019998447011503748,
-      "loss": 46.0,
-      "step": 34846
-    },
-    {
-      "epoch": 5.611860380852692,
-      "grad_norm": 0.002145164180546999,
-      "learning_rate": 0.00019998446922344797,
-      "loss": 46.0,
-      "step": 34847
-    },
-    {
-      "epoch": 5.612021417931478,
-      "grad_norm": 0.0015847224276512861,
-      "learning_rate": 0.00019998446833183287,
-      "loss": 46.0,
-      "step": 34848
-    },
-    {
-      "epoch": 5.612182455010267,
-      "grad_norm": 0.0029122214764356613,
-      "learning_rate": 0.0001999844674401922,
-      "loss": 46.0,
-      "step": 34849
-    },
-    {
-      "epoch": 5.612343492089053,
-      "grad_norm": 0.001439048908650875,
-      "learning_rate": 0.00019998446654852594,
-      "loss": 46.0,
-      "step": 34850
-    },
-    {
-      "epoch": 5.612504529167841,
-      "grad_norm": 0.007039692252874374,
-      "learning_rate": 0.00019998446565683406,
-      "loss": 46.0,
-      "step": 34851
-    },
-    {
-      "epoch": 5.612665566246628,
-      "grad_norm": 0.00641253124922514,
-      "learning_rate": 0.0001999844647651166,
-      "loss": 46.0,
-      "step": 34852
-    },
-    {
-      "epoch": 5.6128266033254155,
-      "grad_norm": 0.008297199383378029,
-      "learning_rate": 0.00019998446387337358,
-      "loss": 46.0,
-      "step": 34853
-    },
-    {
-      "epoch": 5.612987640404203,
-      "grad_norm": 0.014836383983492851,
-      "learning_rate": 0.00019998446298160493,
-      "loss": 46.0,
-      "step": 34854
-    },
-    {
-      "epoch": 5.61314867748299,
-      "grad_norm": 0.0032842091750353575,
-      "learning_rate": 0.00019998446208981073,
-      "loss": 46.0,
-      "step": 34855
-    },
-    {
-      "epoch": 5.613309714561778,
-      "grad_norm": 0.009159413166344166,
-      "learning_rate": 0.0001999844611979909,
-      "loss": 46.0,
-      "step": 34856
-    },
-    {
-      "epoch": 5.613470751640565,
-      "grad_norm": 0.018917739391326904,
-      "learning_rate": 0.0001999844603061455,
-      "loss": 46.0,
-      "step": 34857
-    },
-    {
-      "epoch": 5.613631788719353,
-      "grad_norm": 0.004317612387239933,
-      "learning_rate": 0.00019998445941427447,
-      "loss": 46.0,
-      "step": 34858
-    },
-    {
-      "epoch": 5.61379282579814,
-      "grad_norm": 0.001880093477666378,
-      "learning_rate": 0.0001999844585223779,
-      "loss": 46.0,
-      "step": 34859
-    },
-    {
-      "epoch": 5.613953862876928,
-      "grad_norm": 0.013252083212137222,
-      "learning_rate": 0.00019998445763045572,
-      "loss": 46.0,
-      "step": 34860
-    },
-    {
-      "epoch": 5.614114899955715,
-      "grad_norm": 0.0020481296814978123,
-      "learning_rate": 0.00019998445673850796,
-      "loss": 46.0,
-      "step": 34861
-    },
-    {
-      "epoch": 5.614275937034503,
-      "grad_norm": 0.0031048671808093786,
-      "learning_rate": 0.0001999844558465346,
-      "loss": 46.0,
-      "step": 34862
-    },
-    {
-      "epoch": 5.614436974113289,
-      "grad_norm": 0.00937010906636715,
-      "learning_rate": 0.00019998445495453563,
-      "loss": 46.0,
-      "step": 34863
-    },
-    {
-      "epoch": 5.6145980111920775,
-      "grad_norm": 0.0033450659830123186,
-      "learning_rate": 0.0001999844540625111,
-      "loss": 46.0,
-      "step": 34864
-    },
-    {
-      "epoch": 5.614759048270864,
-      "grad_norm": 0.007181660737842321,
-      "learning_rate": 0.00019998445317046098,
-      "loss": 46.0,
-      "step": 34865
-    },
-    {
-      "epoch": 5.6149200853496515,
-      "grad_norm": 0.00501789478585124,
-      "learning_rate": 0.00019998445227838524,
-      "loss": 46.0,
-      "step": 34866
-    },
-    {
-      "epoch": 5.615081122428439,
-      "grad_norm": 0.0010603123810142279,
-      "learning_rate": 0.00019998445138628396,
-      "loss": 46.0,
-      "step": 34867
-    },
-    {
-      "epoch": 5.615242159507226,
-      "grad_norm": 0.00131454027723521,
-      "learning_rate": 0.00019998445049415704,
-      "loss": 46.0,
-      "step": 34868
-    },
-    {
-      "epoch": 5.615403196586014,
-      "grad_norm": 0.002484876662492752,
-      "learning_rate": 0.00019998444960200453,
-      "loss": 46.0,
-      "step": 34869
-    },
-    {
-      "epoch": 5.615564233664801,
-      "grad_norm": 0.017979202792048454,
-      "learning_rate": 0.00019998444870982643,
-      "loss": 46.0,
-      "step": 34870
-    },
-    {
-      "epoch": 5.615725270743589,
-      "grad_norm": 0.008086731657385826,
-      "learning_rate": 0.00019998444781762275,
-      "loss": 46.0,
-      "step": 34871
-    },
-    {
-      "epoch": 5.615886307822376,
-      "grad_norm": 0.010307800024747849,
-      "learning_rate": 0.00019998444692539348,
-      "loss": 46.0,
-      "step": 34872
-    },
-    {
-      "epoch": 5.616047344901164,
-      "grad_norm": 0.005028403364121914,
-      "learning_rate": 0.00019998444603313865,
-      "loss": 46.0,
-      "step": 34873
-    },
-    {
-      "epoch": 5.616208381979951,
-      "grad_norm": 0.007739054970443249,
-      "learning_rate": 0.00019998444514085818,
-      "loss": 46.0,
-      "step": 34874
-    },
-    {
-      "epoch": 5.6163694190587385,
-      "grad_norm": 0.003156814491376281,
-      "learning_rate": 0.00019998444424855214,
-      "loss": 46.0,
-      "step": 34875
-    },
-    {
-      "epoch": 5.616530456137526,
-      "grad_norm": 0.00888319406658411,
-      "learning_rate": 0.0001999844433562205,
-      "loss": 46.0,
-      "step": 34876
-    },
-    {
-      "epoch": 5.6166914932163134,
-      "grad_norm": 0.017981408163905144,
-      "learning_rate": 0.00019998444246386327,
-      "loss": 46.0,
-      "step": 34877
-    },
-    {
-      "epoch": 5.6168525302951,
-      "grad_norm": 0.004737422801554203,
-      "learning_rate": 0.00019998444157148047,
-      "loss": 46.0,
-      "step": 34878
-    },
-    {
-      "epoch": 5.6170135673738875,
-      "grad_norm": 0.004558139014989138,
-      "learning_rate": 0.00019998444067907207,
-      "loss": 46.0,
-      "step": 34879
-    },
-    {
-      "epoch": 5.617174604452675,
-      "grad_norm": 0.004083253908902407,
-      "learning_rate": 0.00019998443978663807,
-      "loss": 46.0,
-      "step": 34880
-    },
-    {
-      "epoch": 5.617335641531462,
-      "grad_norm": 0.0034036540891975164,
-      "learning_rate": 0.0001999844388941785,
-      "loss": 46.0,
-      "step": 34881
-    },
-    {
-      "epoch": 5.61749667861025,
-      "grad_norm": 0.020808685570955276,
-      "learning_rate": 0.00019998443800169332,
-      "loss": 46.0,
-      "step": 34882
-    },
-    {
-      "epoch": 5.617657715689037,
-      "grad_norm": 0.004127429332584143,
-      "learning_rate": 0.00019998443710918254,
-      "loss": 46.0,
-      "step": 34883
-    },
-    {
-      "epoch": 5.617818752767825,
-      "grad_norm": 0.008937763050198555,
-      "learning_rate": 0.00019998443621664617,
-      "loss": 46.0,
-      "step": 34884
-    },
-    {
-      "epoch": 5.617979789846612,
-      "grad_norm": 0.0021199514158070087,
-      "learning_rate": 0.00019998443532408424,
-      "loss": 46.0,
-      "step": 34885
-    },
-    {
-      "epoch": 5.6181408269254,
-      "grad_norm": 0.009585599415004253,
-      "learning_rate": 0.0001999844344314967,
-      "loss": 46.0,
-      "step": 34886
-    },
-    {
-      "epoch": 5.618301864004187,
-      "grad_norm": 0.0030037083197385073,
-      "learning_rate": 0.00019998443353888353,
-      "loss": 46.0,
-      "step": 34887
-    },
-    {
-      "epoch": 5.6184629010829745,
-      "grad_norm": 0.008482609875500202,
-      "learning_rate": 0.00019998443264624481,
-      "loss": 46.0,
-      "step": 34888
-    },
-    {
-      "epoch": 5.618623938161762,
-      "grad_norm": 0.0035214766394346952,
-      "learning_rate": 0.00019998443175358053,
-      "loss": 46.0,
-      "step": 34889
-    },
-    {
-      "epoch": 5.618784975240549,
-      "grad_norm": 0.004729463253170252,
-      "learning_rate": 0.00019998443086089061,
-      "loss": 46.0,
-      "step": 34890
-    },
-    {
-      "epoch": 5.618946012319337,
-      "grad_norm": 0.0066327378153800964,
-      "learning_rate": 0.0001999844299681751,
-      "loss": 46.0,
-      "step": 34891
-    },
-    {
-      "epoch": 5.619107049398124,
-      "grad_norm": 0.0029992908239364624,
-      "learning_rate": 0.000199984429075434,
-      "loss": 46.0,
-      "step": 34892
-    },
-    {
-      "epoch": 5.619268086476911,
-      "grad_norm": 0.005086920224130154,
-      "learning_rate": 0.00019998442818266736,
-      "loss": 46.0,
-      "step": 34893
-    },
-    {
-      "epoch": 5.619429123555698,
-      "grad_norm": 0.0014695702120661736,
-      "learning_rate": 0.0001999844272898751,
-      "loss": 46.0,
-      "step": 34894
-    },
-    {
-      "epoch": 5.619590160634486,
-      "grad_norm": 0.002051666146144271,
-      "learning_rate": 0.0001999844263970572,
-      "loss": 46.0,
-      "step": 34895
-    },
-    {
-      "epoch": 5.619751197713273,
-      "grad_norm": 0.00768818985670805,
-      "learning_rate": 0.00019998442550421376,
-      "loss": 46.0,
-      "step": 34896
-    },
-    {
-      "epoch": 5.619912234792061,
-      "grad_norm": 0.006151929497718811,
-      "learning_rate": 0.00019998442461134473,
-      "loss": 46.0,
-      "step": 34897
-    },
-    {
-      "epoch": 5.620073271870848,
-      "grad_norm": 0.002680000849068165,
-      "learning_rate": 0.00019998442371845009,
-      "loss": 46.0,
-      "step": 34898
-    },
-    {
-      "epoch": 5.620234308949636,
-      "grad_norm": 0.005544870160520077,
-      "learning_rate": 0.00019998442282552988,
-      "loss": 46.0,
-      "step": 34899
-    },
-    {
-      "epoch": 5.620395346028423,
-      "grad_norm": 0.009820745326578617,
-      "learning_rate": 0.00019998442193258406,
-      "loss": 46.0,
-      "step": 34900
-    },
-    {
-      "epoch": 5.6205563831072105,
-      "grad_norm": 0.00863034836947918,
-      "learning_rate": 0.00019998442103961263,
-      "loss": 46.0,
-      "step": 34901
-    },
-    {
-      "epoch": 5.620717420185998,
-      "grad_norm": 0.0014620324363932014,
-      "learning_rate": 0.00019998442014661566,
-      "loss": 46.0,
-      "step": 34902
-    },
-    {
-      "epoch": 5.620878457264785,
-      "grad_norm": 0.012427587062120438,
-      "learning_rate": 0.00019998441925359308,
-      "loss": 46.0,
-      "step": 34903
-    },
-    {
-      "epoch": 5.621039494343573,
-      "grad_norm": 0.0019669944886118174,
-      "learning_rate": 0.00019998441836054488,
-      "loss": 46.0,
-      "step": 34904
-    },
-    {
-      "epoch": 5.62120053142236,
-      "grad_norm": 0.0017034054035320878,
-      "learning_rate": 0.00019998441746747113,
-      "loss": 46.0,
-      "step": 34905
-    },
-    {
-      "epoch": 5.621361568501148,
-      "grad_norm": 0.005487917456775904,
-      "learning_rate": 0.00019998441657437176,
-      "loss": 46.0,
-      "step": 34906
-    },
-    {
-      "epoch": 5.621522605579935,
-      "grad_norm": 0.0028911111876368523,
-      "learning_rate": 0.0001999844156812468,
-      "loss": 46.0,
-      "step": 34907
-    },
-    {
-      "epoch": 5.621683642658722,
-      "grad_norm": 0.005043532233685255,
-      "learning_rate": 0.00019998441478809626,
-      "loss": 46.0,
-      "step": 34908
-    },
-    {
-      "epoch": 5.621844679737509,
-      "grad_norm": 0.018376145511865616,
-      "learning_rate": 0.00019998441389492016,
-      "loss": 46.0,
-      "step": 34909
-    },
-    {
-      "epoch": 5.622005716816297,
-      "grad_norm": 0.002302020788192749,
-      "learning_rate": 0.0001999844130017184,
-      "loss": 46.0,
-      "step": 34910
-    },
-    {
-      "epoch": 5.622166753895084,
-      "grad_norm": 0.001525710104033351,
-      "learning_rate": 0.0001999844121084911,
-      "loss": 46.0,
-      "step": 34911
-    },
-    {
-      "epoch": 5.622327790973872,
-      "grad_norm": 0.004479615483433008,
-      "learning_rate": 0.0001999844112152382,
-      "loss": 46.0,
-      "step": 34912
-    },
-    {
-      "epoch": 5.622488828052659,
-      "grad_norm": 0.0035913384053856134,
-      "learning_rate": 0.00019998441032195968,
-      "loss": 46.0,
-      "step": 34913
-    },
-    {
-      "epoch": 5.6226498651314465,
-      "grad_norm": 0.0014180107973515987,
-      "learning_rate": 0.0001999844094286556,
-      "loss": 46.0,
-      "step": 34914
-    },
-    {
-      "epoch": 5.622810902210234,
-      "grad_norm": 0.010112438350915909,
-      "learning_rate": 0.00019998440853532593,
-      "loss": 46.0,
-      "step": 34915
-    },
-    {
-      "epoch": 5.622971939289021,
-      "grad_norm": 0.0008956961100921035,
-      "learning_rate": 0.00019998440764197066,
-      "loss": 46.0,
-      "step": 34916
-    },
-    {
-      "epoch": 5.623132976367809,
-      "grad_norm": 0.005347180180251598,
-      "learning_rate": 0.00019998440674858984,
-      "loss": 46.0,
-      "step": 34917
-    },
-    {
-      "epoch": 5.623294013446596,
-      "grad_norm": 0.0007655193912796676,
-      "learning_rate": 0.00019998440585518337,
-      "loss": 46.0,
-      "step": 34918
-    },
-    {
-      "epoch": 5.623455050525384,
-      "grad_norm": 0.009878878481686115,
-      "learning_rate": 0.00019998440496175134,
-      "loss": 46.0,
-      "step": 34919
-    },
-    {
-      "epoch": 5.623616087604171,
-      "grad_norm": 0.015162709169089794,
-      "learning_rate": 0.0001999844040682937,
-      "loss": 46.0,
-      "step": 34920
-    },
-    {
-      "epoch": 5.623777124682958,
-      "grad_norm": 0.008933644741773605,
-      "learning_rate": 0.00019998440317481046,
-      "loss": 46.0,
-      "step": 34921
-    },
-    {
-      "epoch": 5.623938161761746,
-      "grad_norm": 0.008423752151429653,
-      "learning_rate": 0.00019998440228130164,
-      "loss": 46.0,
-      "step": 34922
-    },
-    {
-      "epoch": 5.624099198840533,
-      "grad_norm": 0.010797766037285328,
-      "learning_rate": 0.00019998440138776724,
-      "loss": 46.0,
-      "step": 34923
-    },
-    {
-      "epoch": 5.62426023591932,
-      "grad_norm": 0.001507542678155005,
-      "learning_rate": 0.00019998440049420727,
-      "loss": 46.0,
-      "step": 34924
-    },
-    {
-      "epoch": 5.624421272998108,
-      "grad_norm": 0.0038560088723897934,
-      "learning_rate": 0.00019998439960062166,
-      "loss": 46.0,
-      "step": 34925
-    },
-    {
-      "epoch": 5.624582310076895,
-      "grad_norm": 0.0227101631462574,
-      "learning_rate": 0.0001999843987070105,
-      "loss": 46.0,
-      "step": 34926
-    },
-    {
-      "epoch": 5.6247433471556825,
-      "grad_norm": 0.00186068972107023,
-      "learning_rate": 0.00019998439781337372,
-      "loss": 46.0,
-      "step": 34927
-    },
-    {
-      "epoch": 5.62490438423447,
-      "grad_norm": 0.00811269972473383,
-      "learning_rate": 0.00019998439691971135,
-      "loss": 46.0,
-      "step": 34928
-    },
-    {
-      "epoch": 5.625065421313257,
-      "grad_norm": 0.0022010605316609144,
-      "learning_rate": 0.00019998439602602342,
-      "loss": 46.0,
-      "step": 34929
-    },
-    {
-      "epoch": 5.625226458392045,
-      "grad_norm": 0.005052570253610611,
-      "learning_rate": 0.00019998439513230988,
-      "loss": 46.0,
-      "step": 34930
-    },
-    {
-      "epoch": 5.625387495470832,
-      "grad_norm": 0.00520938728004694,
-      "learning_rate": 0.00019998439423857075,
-      "loss": 46.0,
-      "step": 34931
-    },
-    {
-      "epoch": 5.62554853254962,
-      "grad_norm": 0.0055502429604530334,
-      "learning_rate": 0.000199984393344806,
-      "loss": 46.0,
-      "step": 34932
-    },
-    {
-      "epoch": 5.625709569628407,
-      "grad_norm": 0.005287414416670799,
-      "learning_rate": 0.00019998439245101572,
-      "loss": 46.0,
-      "step": 34933
-    },
-    {
-      "epoch": 5.625870606707195,
-      "grad_norm": 0.007076835725456476,
-      "learning_rate": 0.0001999843915571998,
-      "loss": 46.0,
-      "step": 34934
-    },
-    {
-      "epoch": 5.626031643785982,
-      "grad_norm": 0.0077497451566159725,
-      "learning_rate": 0.00019998439066335833,
-      "loss": 46.0,
-      "step": 34935
-    },
-    {
-      "epoch": 5.626192680864769,
-      "grad_norm": 0.00195614667609334,
-      "learning_rate": 0.0001999843897694912,
-      "loss": 46.0,
-      "step": 34936
-    },
-    {
-      "epoch": 5.626353717943557,
-      "grad_norm": 0.00780531158670783,
-      "learning_rate": 0.00019998438887559855,
-      "loss": 46.0,
-      "step": 34937
-    },
-    {
-      "epoch": 5.6265147550223436,
-      "grad_norm": 0.0029682728927582502,
-      "learning_rate": 0.00019998438798168029,
-      "loss": 46.0,
-      "step": 34938
-    },
-    {
-      "epoch": 5.626675792101131,
-      "grad_norm": 0.0064385319128632545,
-      "learning_rate": 0.00019998438708773643,
-      "loss": 46.0,
-      "step": 34939
-    },
-    {
-      "epoch": 5.6268368291799185,
-      "grad_norm": 0.00434201629832387,
-      "learning_rate": 0.00019998438619376696,
-      "loss": 46.0,
-      "step": 34940
-    },
-    {
-      "epoch": 5.626997866258706,
-      "grad_norm": 0.002590473974123597,
-      "learning_rate": 0.00019998438529977196,
-      "loss": 46.0,
-      "step": 34941
-    },
-    {
-      "epoch": 5.627158903337493,
-      "grad_norm": 0.0050941938534379005,
-      "learning_rate": 0.00019998438440575132,
-      "loss": 46.0,
-      "step": 34942
-    },
-    {
-      "epoch": 5.627319940416281,
-      "grad_norm": 0.0027847914025187492,
-      "learning_rate": 0.0001999843835117051,
-      "loss": 46.0,
-      "step": 34943
-    },
-    {
-      "epoch": 5.627480977495068,
-      "grad_norm": 0.008297148160636425,
-      "learning_rate": 0.00019998438261763327,
-      "loss": 46.0,
-      "step": 34944
-    },
-    {
-      "epoch": 5.627642014573856,
-      "grad_norm": 0.0013164766132831573,
-      "learning_rate": 0.00019998438172353587,
-      "loss": 46.0,
-      "step": 34945
-    },
-    {
-      "epoch": 5.627803051652643,
-      "grad_norm": 0.014140353538095951,
-      "learning_rate": 0.00019998438082941287,
-      "loss": 46.0,
-      "step": 34946
-    },
-    {
-      "epoch": 5.627964088731431,
-      "grad_norm": 0.01257090549916029,
-      "learning_rate": 0.0001999843799352643,
-      "loss": 46.0,
-      "step": 34947
-    },
-    {
-      "epoch": 5.628125125810218,
-      "grad_norm": 0.012158217839896679,
-      "learning_rate": 0.0001999843790410901,
-      "loss": 46.0,
-      "step": 34948
-    },
-    {
-      "epoch": 5.6282861628890055,
-      "grad_norm": 0.005803667474538088,
-      "learning_rate": 0.00019998437814689032,
-      "loss": 46.0,
-      "step": 34949
-    },
-    {
-      "epoch": 5.628447199967793,
-      "grad_norm": 0.010137118399143219,
-      "learning_rate": 0.00019998437725266495,
-      "loss": 46.0,
-      "step": 34950
-    },
-    {
-      "epoch": 5.6286082370465795,
-      "grad_norm": 0.002603465225547552,
-      "learning_rate": 0.00019998437635841403,
-      "loss": 46.0,
-      "step": 34951
-    },
-    {
-      "epoch": 5.628769274125367,
-      "grad_norm": 0.008865314535796642,
-      "learning_rate": 0.0001999843754641375,
-      "loss": 46.0,
-      "step": 34952
-    },
-    {
-      "epoch": 5.628930311204154,
-      "grad_norm": 0.0072078160010278225,
-      "learning_rate": 0.00019998437456983536,
-      "loss": 46.0,
-      "step": 34953
-    },
-    {
-      "epoch": 5.629091348282942,
-      "grad_norm": 0.0025421488098800182,
-      "learning_rate": 0.00019998437367550764,
-      "loss": 46.0,
-      "step": 34954
-    },
-    {
-      "epoch": 5.629252385361729,
-      "grad_norm": 0.0029978107195347548,
-      "learning_rate": 0.00019998437278115434,
-      "loss": 46.0,
-      "step": 34955
-    },
-    {
-      "epoch": 5.629413422440517,
-      "grad_norm": 0.004747950471937656,
-      "learning_rate": 0.00019998437188677542,
-      "loss": 46.0,
-      "step": 34956
-    },
-    {
-      "epoch": 5.629574459519304,
-      "grad_norm": 0.02354920655488968,
-      "learning_rate": 0.00019998437099237094,
-      "loss": 46.0,
-      "step": 34957
-    },
-    {
-      "epoch": 5.629735496598092,
-      "grad_norm": 0.005115113221108913,
-      "learning_rate": 0.00019998437009794085,
-      "loss": 46.0,
-      "step": 34958
-    },
-    {
-      "epoch": 5.629896533676879,
-      "grad_norm": 0.0012572644045576453,
-      "learning_rate": 0.00019998436920348517,
-      "loss": 46.0,
-      "step": 34959
-    },
-    {
-      "epoch": 5.630057570755667,
-      "grad_norm": 0.001496284268796444,
-      "learning_rate": 0.0001999843683090039,
-      "loss": 46.0,
-      "step": 34960
-    },
-    {
-      "epoch": 5.630218607834454,
-      "grad_norm": 0.0006963298073969781,
-      "learning_rate": 0.00019998436741449705,
-      "loss": 46.0,
-      "step": 34961
-    },
-    {
-      "epoch": 5.6303796449132415,
-      "grad_norm": 0.017010528594255447,
-      "learning_rate": 0.00019998436651996458,
-      "loss": 46.0,
-      "step": 34962
-    },
-    {
-      "epoch": 5.630540681992029,
-      "grad_norm": 0.0035441096406430006,
-      "learning_rate": 0.00019998436562540658,
-      "loss": 46.0,
-      "step": 34963
-    },
-    {
-      "epoch": 5.630701719070816,
-      "grad_norm": 0.002389661269262433,
-      "learning_rate": 0.0001999843647308229,
-      "loss": 46.0,
-      "step": 34964
-    },
-    {
-      "epoch": 5.630862756149604,
-      "grad_norm": 0.0030964016914367676,
-      "learning_rate": 0.0001999843638362137,
-      "loss": 46.0,
-      "step": 34965
-    },
-    {
-      "epoch": 5.63102379322839,
-      "grad_norm": 0.0011341158533468843,
-      "learning_rate": 0.0001999843629415789,
-      "loss": 46.0,
-      "step": 34966
-    },
-    {
-      "epoch": 5.631184830307178,
-      "grad_norm": 0.0012163082137703896,
-      "learning_rate": 0.0001999843620469185,
-      "loss": 46.0,
-      "step": 34967
-    },
-    {
-      "epoch": 5.631345867385965,
-      "grad_norm": 0.006449075415730476,
-      "learning_rate": 0.00019998436115223252,
-      "loss": 46.0,
-      "step": 34968
-    },
-    {
-      "epoch": 5.631506904464753,
-      "grad_norm": 0.014941618777811527,
-      "learning_rate": 0.00019998436025752092,
-      "loss": 46.0,
-      "step": 34969
-    },
-    {
-      "epoch": 5.63166794154354,
-      "grad_norm": 0.005751014221459627,
-      "learning_rate": 0.00019998435936278375,
-      "loss": 46.0,
-      "step": 34970
-    },
-    {
-      "epoch": 5.631828978622328,
-      "grad_norm": 0.00354705355130136,
-      "learning_rate": 0.00019998435846802097,
-      "loss": 46.0,
-      "step": 34971
-    },
-    {
-      "epoch": 5.631990015701115,
-      "grad_norm": 0.0028325680177658796,
-      "learning_rate": 0.0001999843575732326,
-      "loss": 46.0,
-      "step": 34972
-    },
-    {
-      "epoch": 5.632151052779903,
-      "grad_norm": 0.004960860125720501,
-      "learning_rate": 0.00019998435667841865,
-      "loss": 46.0,
-      "step": 34973
-    },
-    {
-      "epoch": 5.63231208985869,
-      "grad_norm": 0.02809079922735691,
-      "learning_rate": 0.0001999843557835791,
-      "loss": 46.0,
-      "step": 34974
-    },
-    {
-      "epoch": 5.6324731269374775,
-      "grad_norm": 0.011901609599590302,
-      "learning_rate": 0.000199984354888714,
-      "loss": 46.0,
-      "step": 34975
-    },
-    {
-      "epoch": 5.632634164016265,
-      "grad_norm": 0.005198607221245766,
-      "learning_rate": 0.00019998435399382327,
-      "loss": 46.0,
-      "step": 34976
-    },
-    {
-      "epoch": 5.632795201095052,
-      "grad_norm": 0.020701995119452477,
-      "learning_rate": 0.00019998435309890697,
-      "loss": 46.0,
-      "step": 34977
-    },
-    {
-      "epoch": 5.63295623817384,
-      "grad_norm": 0.015023778192698956,
-      "learning_rate": 0.00019998435220396505,
-      "loss": 46.0,
-      "step": 34978
-    },
-    {
-      "epoch": 5.633117275252627,
-      "grad_norm": 0.014441955834627151,
-      "learning_rate": 0.00019998435130899757,
-      "loss": 46.0,
-      "step": 34979
-    },
-    {
-      "epoch": 5.633278312331415,
-      "grad_norm": 0.0021402090787887573,
-      "learning_rate": 0.00019998435041400448,
-      "loss": 46.0,
-      "step": 34980
-    },
-    {
-      "epoch": 5.633439349410201,
-      "grad_norm": 0.004062502179294825,
-      "learning_rate": 0.0001999843495189858,
-      "loss": 46.0,
-      "step": 34981
-    },
-    {
-      "epoch": 5.633600386488989,
-      "grad_norm": 0.006026627030223608,
-      "learning_rate": 0.00019998434862394157,
-      "loss": 46.0,
-      "step": 34982
-    },
-    {
-      "epoch": 5.633761423567776,
-      "grad_norm": 0.0017059509409591556,
-      "learning_rate": 0.00019998434772887169,
-      "loss": 46.0,
-      "step": 34983
-    },
-    {
-      "epoch": 5.633922460646564,
-      "grad_norm": 0.009850052185356617,
-      "learning_rate": 0.00019998434683377625,
-      "loss": 46.0,
-      "step": 34984
-    },
-    {
-      "epoch": 5.634083497725351,
-      "grad_norm": 0.0048669311217963696,
-      "learning_rate": 0.0001999843459386552,
-      "loss": 46.0,
-      "step": 34985
-    },
-    {
-      "epoch": 5.634244534804139,
-      "grad_norm": 0.004842791706323624,
-      "learning_rate": 0.00019998434504350858,
-      "loss": 46.0,
-      "step": 34986
-    },
-    {
-      "epoch": 5.634405571882926,
-      "grad_norm": 0.0023919185623526573,
-      "learning_rate": 0.00019998434414833635,
-      "loss": 46.0,
-      "step": 34987
-    },
-    {
-      "epoch": 5.6345666089617135,
-      "grad_norm": 0.005478554870933294,
-      "learning_rate": 0.00019998434325313856,
-      "loss": 46.0,
-      "step": 34988
-    },
-    {
-      "epoch": 5.634727646040501,
-      "grad_norm": 0.006147760897874832,
-      "learning_rate": 0.00019998434235791513,
-      "loss": 46.0,
-      "step": 34989
-    },
-    {
-      "epoch": 5.634888683119288,
-      "grad_norm": 0.002799938665702939,
-      "learning_rate": 0.00019998434146266617,
-      "loss": 46.0,
-      "step": 34990
-    },
-    {
-      "epoch": 5.635049720198076,
-      "grad_norm": 0.004861291963607073,
-      "learning_rate": 0.00019998434056739156,
-      "loss": 46.0,
-      "step": 34991
-    },
-    {
-      "epoch": 5.635210757276863,
-      "grad_norm": 0.01334360335022211,
-      "learning_rate": 0.0001999843396720914,
-      "loss": 46.0,
-      "step": 34992
-    },
-    {
-      "epoch": 5.635371794355651,
-      "grad_norm": 0.007166545372456312,
-      "learning_rate": 0.00019998433877676562,
-      "loss": 46.0,
-      "step": 34993
-    },
-    {
-      "epoch": 5.635532831434437,
-      "grad_norm": 0.0013273741351440549,
-      "learning_rate": 0.00019998433788141426,
-      "loss": 46.0,
-      "step": 34994
-    },
-    {
-      "epoch": 5.635693868513226,
-      "grad_norm": 0.0039773741737008095,
-      "learning_rate": 0.00019998433698603733,
-      "loss": 46.0,
-      "step": 34995
-    },
-    {
-      "epoch": 5.635854905592012,
-      "grad_norm": 0.004289644304662943,
-      "learning_rate": 0.0001999843360906348,
-      "loss": 46.0,
-      "step": 34996
-    },
-    {
-      "epoch": 5.6360159426708,
-      "grad_norm": 0.0035704439505934715,
-      "learning_rate": 0.00019998433519520666,
-      "loss": 46.0,
-      "step": 34997
-    },
-    {
-      "epoch": 5.636176979749587,
-      "grad_norm": 0.0017077274387702346,
-      "learning_rate": 0.00019998433429975295,
-      "loss": 46.0,
-      "step": 34998
-    },
-    {
-      "epoch": 5.6363380168283745,
-      "grad_norm": 0.008026796393096447,
-      "learning_rate": 0.00019998433340427365,
-      "loss": 46.0,
-      "step": 34999
-    },
-    {
-      "epoch": 5.636499053907162,
-      "grad_norm": 0.010748764500021935,
-      "learning_rate": 0.00019998433250876873,
-      "loss": 46.0,
-      "step": 35000
-    },
-    {
-      "epoch": 5.6366600909859494,
-      "grad_norm": 0.003786412300541997,
-      "learning_rate": 0.00019998433161323823,
-      "loss": 46.0,
-      "step": 35001
-    },
-    {
-      "epoch": 5.636821128064737,
-      "grad_norm": 0.006463125813752413,
-      "learning_rate": 0.00019998433071768216,
-      "loss": 46.0,
-      "step": 35002
-    },
-    {
-      "epoch": 5.636982165143524,
-      "grad_norm": 0.0012253804598003626,
-      "learning_rate": 0.00019998432982210046,
-      "loss": 46.0,
-      "step": 35003
-    },
-    {
-      "epoch": 5.637143202222312,
-      "grad_norm": 0.007236053701490164,
-      "learning_rate": 0.00019998432892649325,
-      "loss": 46.0,
-      "step": 35004
-    },
-    {
-      "epoch": 5.637304239301099,
-      "grad_norm": 0.0021753734908998013,
-      "learning_rate": 0.00019998432803086037,
-      "loss": 46.0,
-      "step": 35005
-    },
-    {
-      "epoch": 5.637465276379887,
-      "grad_norm": 0.0015616790624335408,
-      "learning_rate": 0.00019998432713520193,
-      "loss": 46.0,
-      "step": 35006
-    },
-    {
-      "epoch": 5.637626313458674,
-      "grad_norm": 0.0020285514183342457,
-      "learning_rate": 0.00019998432623951788,
-      "loss": 46.0,
-      "step": 35007
-    },
-    {
-      "epoch": 5.637787350537462,
-      "grad_norm": 0.0070483749732375145,
-      "learning_rate": 0.00019998432534380824,
-      "loss": 46.0,
-      "step": 35008
-    },
-    {
-      "epoch": 5.637948387616248,
-      "grad_norm": 0.002363958628848195,
-      "learning_rate": 0.000199984324448073,
-      "loss": 46.0,
-      "step": 35009
-    },
-    {
-      "epoch": 5.6381094246950365,
-      "grad_norm": 0.009358027018606663,
-      "learning_rate": 0.00019998432355231222,
-      "loss": 46.0,
-      "step": 35010
-    },
-    {
-      "epoch": 5.638270461773823,
-      "grad_norm": 0.0037781603168696165,
-      "learning_rate": 0.00019998432265652582,
-      "loss": 46.0,
-      "step": 35011
-    },
-    {
-      "epoch": 5.6384314988526105,
-      "grad_norm": 0.0015166779048740864,
-      "learning_rate": 0.00019998432176071383,
-      "loss": 46.0,
-      "step": 35012
-    },
-    {
-      "epoch": 5.638592535931398,
-      "grad_norm": 0.0035782321356236935,
-      "learning_rate": 0.00019998432086487626,
-      "loss": 46.0,
-      "step": 35013
-    },
-    {
-      "epoch": 5.638753573010185,
-      "grad_norm": 0.014220730401575565,
-      "learning_rate": 0.00019998431996901307,
-      "loss": 46.0,
-      "step": 35014
-    },
-    {
-      "epoch": 5.638914610088973,
-      "grad_norm": 0.0014554974623024464,
-      "learning_rate": 0.00019998431907312431,
-      "loss": 46.0,
-      "step": 35015
-    },
-    {
-      "epoch": 5.63907564716776,
-      "grad_norm": 0.010806080885231495,
-      "learning_rate": 0.00019998431817720995,
-      "loss": 46.0,
-      "step": 35016
-    },
-    {
-      "epoch": 5.639236684246548,
-      "grad_norm": 0.0052486443892121315,
-      "learning_rate": 0.00019998431728127,
-      "loss": 46.0,
-      "step": 35017
-    },
-    {
-      "epoch": 5.639397721325335,
-      "grad_norm": 0.006050060968846083,
-      "learning_rate": 0.00019998431638530449,
-      "loss": 46.0,
-      "step": 35018
-    },
-    {
-      "epoch": 5.639558758404123,
-      "grad_norm": 0.0034774153027683496,
-      "learning_rate": 0.00019998431548931333,
-      "loss": 46.0,
-      "step": 35019
-    },
-    {
-      "epoch": 5.63971979548291,
-      "grad_norm": 0.01162722334265709,
-      "learning_rate": 0.00019998431459329662,
-      "loss": 46.0,
-      "step": 35020
-    },
-    {
-      "epoch": 5.639880832561698,
-      "grad_norm": 0.0065436819568276405,
-      "learning_rate": 0.00019998431369725432,
-      "loss": 46.0,
-      "step": 35021
-    },
-    {
-      "epoch": 5.640041869640485,
-      "grad_norm": 0.0037487749941647053,
-      "learning_rate": 0.0001999843128011864,
-      "loss": 46.0,
-      "step": 35022
-    },
-    {
-      "epoch": 5.6402029067192725,
-      "grad_norm": 0.003316734917461872,
-      "learning_rate": 0.00019998431190509293,
-      "loss": 46.0,
-      "step": 35023
-    },
-    {
-      "epoch": 5.640363943798059,
-      "grad_norm": 0.0021634784061461687,
-      "learning_rate": 0.00019998431100897387,
-      "loss": 46.0,
-      "step": 35024
-    },
-    {
-      "epoch": 5.6405249808768465,
-      "grad_norm": 0.0069904625415802,
-      "learning_rate": 0.0001999843101128292,
-      "loss": 46.0,
-      "step": 35025
-    },
-    {
-      "epoch": 5.640686017955634,
-      "grad_norm": 0.006361029576510191,
-      "learning_rate": 0.00019998430921665893,
-      "loss": 46.0,
-      "step": 35026
-    },
-    {
-      "epoch": 5.640847055034421,
-      "grad_norm": 0.007781515829265118,
-      "learning_rate": 0.00019998430832046307,
-      "loss": 46.0,
-      "step": 35027
-    },
-    {
-      "epoch": 5.641008092113209,
-      "grad_norm": 0.0010851513361558318,
-      "learning_rate": 0.0001999843074242416,
-      "loss": 46.0,
-      "step": 35028
-    },
-    {
-      "epoch": 5.641169129191996,
-      "grad_norm": 0.009033090434968472,
-      "learning_rate": 0.00019998430652799458,
-      "loss": 46.0,
-      "step": 35029
-    },
-    {
-      "epoch": 5.641330166270784,
-      "grad_norm": 0.017299111932516098,
-      "learning_rate": 0.00019998430563172195,
-      "loss": 46.0,
-      "step": 35030
-    },
-    {
-      "epoch": 5.641491203349571,
-      "grad_norm": 0.004509763792157173,
-      "learning_rate": 0.00019998430473542375,
-      "loss": 46.0,
-      "step": 35031
-    },
-    {
-      "epoch": 5.641652240428359,
-      "grad_norm": 0.0031622734386473894,
-      "learning_rate": 0.0001999843038390999,
-      "loss": 46.0,
-      "step": 35032
-    },
-    {
-      "epoch": 5.641813277507146,
-      "grad_norm": 0.0007791734533384442,
-      "learning_rate": 0.00019998430294275053,
-      "loss": 46.0,
-      "step": 35033
-    },
-    {
-      "epoch": 5.641974314585934,
-      "grad_norm": 0.008482211269438267,
-      "learning_rate": 0.00019998430204637554,
-      "loss": 46.0,
-      "step": 35034
-    },
-    {
-      "epoch": 5.642135351664721,
-      "grad_norm": 0.0015810014447197318,
-      "learning_rate": 0.00019998430114997494,
-      "loss": 46.0,
-      "step": 35035
-    },
-    {
-      "epoch": 5.6422963887435085,
-      "grad_norm": 0.004794060252606869,
-      "learning_rate": 0.00019998430025354878,
-      "loss": 46.0,
-      "step": 35036
-    },
-    {
-      "epoch": 5.642457425822296,
-      "grad_norm": 0.0037337609101086855,
-      "learning_rate": 0.00019998429935709703,
-      "loss": 46.0,
-      "step": 35037
-    },
-    {
-      "epoch": 5.642618462901083,
-      "grad_norm": 0.004877662286162376,
-      "learning_rate": 0.0001999842984606197,
-      "loss": 46.0,
-      "step": 35038
-    },
-    {
-      "epoch": 5.64277949997987,
-      "grad_norm": 0.0019458206370472908,
-      "learning_rate": 0.0001999842975641167,
-      "loss": 46.0,
-      "step": 35039
-    },
-    {
-      "epoch": 5.642940537058657,
-      "grad_norm": 0.004025063943117857,
-      "learning_rate": 0.00019998429666758818,
-      "loss": 46.0,
-      "step": 35040
-    },
-    {
-      "epoch": 5.643101574137445,
-      "grad_norm": 0.013048715889453888,
-      "learning_rate": 0.00019998429577103405,
-      "loss": 46.0,
-      "step": 35041
-    },
-    {
-      "epoch": 5.643262611216232,
-      "grad_norm": 0.004395760595798492,
-      "learning_rate": 0.00019998429487445436,
-      "loss": 46.0,
-      "step": 35042
-    },
-    {
-      "epoch": 5.64342364829502,
-      "grad_norm": 0.001118919113650918,
-      "learning_rate": 0.00019998429397784904,
-      "loss": 46.0,
-      "step": 35043
-    },
-    {
-      "epoch": 5.643584685373807,
-      "grad_norm": 0.002619420178234577,
-      "learning_rate": 0.00019998429308121812,
-      "loss": 46.0,
-      "step": 35044
-    },
-    {
-      "epoch": 5.643745722452595,
-      "grad_norm": 0.015735985711216927,
-      "learning_rate": 0.00019998429218456165,
-      "loss": 46.0,
-      "step": 35045
-    },
-    {
-      "epoch": 5.643906759531382,
-      "grad_norm": 0.0015861950814723969,
-      "learning_rate": 0.00019998429128787956,
-      "loss": 46.0,
-      "step": 35046
-    },
-    {
-      "epoch": 5.6440677966101696,
-      "grad_norm": 0.004166461061686277,
-      "learning_rate": 0.0001999842903911719,
-      "loss": 46.0,
-      "step": 35047
-    },
-    {
-      "epoch": 5.644228833688957,
-      "grad_norm": 0.0022016323637217283,
-      "learning_rate": 0.00019998428949443865,
-      "loss": 46.0,
-      "step": 35048
-    },
-    {
-      "epoch": 5.6443898707677445,
-      "grad_norm": 0.00824781134724617,
-      "learning_rate": 0.0001999842885976798,
-      "loss": 46.0,
-      "step": 35049
-    },
-    {
-      "epoch": 5.644550907846532,
-      "grad_norm": 0.00235876371152699,
-      "learning_rate": 0.00019998428770089534,
-      "loss": 46.0,
-      "step": 35050
-    },
-    {
-      "epoch": 5.644711944925319,
-      "grad_norm": 0.009497219696640968,
-      "learning_rate": 0.00019998428680408528,
-      "loss": 46.0,
-      "step": 35051
-    },
-    {
-      "epoch": 5.644872982004107,
-      "grad_norm": 0.007655688561499119,
-      "learning_rate": 0.0001999842859072497,
-      "loss": 46.0,
-      "step": 35052
-    },
-    {
-      "epoch": 5.645034019082894,
-      "grad_norm": 0.00322348834015429,
-      "learning_rate": 0.00019998428501038847,
-      "loss": 46.0,
-      "step": 35053
-    },
-    {
-      "epoch": 5.645195056161681,
-      "grad_norm": 0.004492009524255991,
-      "learning_rate": 0.00019998428411350166,
-      "loss": 46.0,
-      "step": 35054
-    },
-    {
-      "epoch": 5.645356093240468,
-      "grad_norm": 0.0016542176017537713,
-      "learning_rate": 0.00019998428321658923,
-      "loss": 46.0,
-      "step": 35055
-    },
-    {
-      "epoch": 5.645517130319256,
-      "grad_norm": 0.006261840928345919,
-      "learning_rate": 0.00019998428231965125,
-      "loss": 46.0,
-      "step": 35056
-    },
-    {
-      "epoch": 5.645678167398043,
-      "grad_norm": 0.0060275401920080185,
-      "learning_rate": 0.00019998428142268767,
-      "loss": 46.0,
-      "step": 35057
-    },
-    {
-      "epoch": 5.645839204476831,
-      "grad_norm": 0.00190097582526505,
-      "learning_rate": 0.00019998428052569854,
-      "loss": 46.0,
-      "step": 35058
-    },
-    {
-      "epoch": 5.646000241555618,
-      "grad_norm": 0.00442290585488081,
-      "learning_rate": 0.00019998427962868376,
-      "loss": 46.0,
-      "step": 35059
-    },
-    {
-      "epoch": 5.6461612786344055,
-      "grad_norm": 0.0058402190916240215,
-      "learning_rate": 0.0001999842787316434,
-      "loss": 46.0,
-      "step": 35060
-    },
-    {
-      "epoch": 5.646322315713193,
-      "grad_norm": 0.0011229125084355474,
-      "learning_rate": 0.00019998427783457748,
-      "loss": 46.0,
-      "step": 35061
-    },
-    {
-      "epoch": 5.64648335279198,
-      "grad_norm": 0.011011681519448757,
-      "learning_rate": 0.0001999842769374859,
-      "loss": 46.0,
-      "step": 35062
-    },
-    {
-      "epoch": 5.646644389870768,
-      "grad_norm": 0.005655078683048487,
-      "learning_rate": 0.00019998427604036881,
-      "loss": 46.0,
-      "step": 35063
-    },
-    {
-      "epoch": 5.646805426949555,
-      "grad_norm": 0.003370630322024226,
-      "learning_rate": 0.00019998427514322608,
-      "loss": 46.0,
-      "step": 35064
-    },
-    {
-      "epoch": 5.646966464028343,
-      "grad_norm": 0.018145233392715454,
-      "learning_rate": 0.00019998427424605778,
-      "loss": 46.0,
-      "step": 35065
-    },
-    {
-      "epoch": 5.64712750110713,
-      "grad_norm": 0.003374564927071333,
-      "learning_rate": 0.0001999842733488639,
-      "loss": 46.0,
-      "step": 35066
-    },
-    {
-      "epoch": 5.647288538185918,
-      "grad_norm": 0.0029213225934654474,
-      "learning_rate": 0.0001999842724516444,
-      "loss": 46.0,
-      "step": 35067
-    },
-    {
-      "epoch": 5.647449575264705,
-      "grad_norm": 0.004805019125342369,
-      "learning_rate": 0.0001999842715543993,
-      "loss": 46.0,
-      "step": 35068
-    },
-    {
-      "epoch": 5.647610612343492,
-      "grad_norm": 0.0009276751079596579,
-      "learning_rate": 0.00019998427065712866,
-      "loss": 46.0,
-      "step": 35069
-    },
-    {
-      "epoch": 5.647771649422279,
-      "grad_norm": 0.0015234776074066758,
-      "learning_rate": 0.00019998426975983237,
-      "loss": 46.0,
-      "step": 35070
-    },
-    {
-      "epoch": 5.647932686501067,
-      "grad_norm": 0.010391012765467167,
-      "learning_rate": 0.00019998426886251052,
-      "loss": 46.0,
-      "step": 35071
-    },
-    {
-      "epoch": 5.648093723579854,
-      "grad_norm": 0.0023990606423467398,
-      "learning_rate": 0.00019998426796516308,
-      "loss": 46.0,
-      "step": 35072
-    },
-    {
-      "epoch": 5.6482547606586415,
-      "grad_norm": 0.0038860745262354612,
-      "learning_rate": 0.00019998426706779006,
-      "loss": 46.0,
-      "step": 35073
-    },
-    {
-      "epoch": 5.648415797737429,
-      "grad_norm": 0.005859003867954016,
-      "learning_rate": 0.00019998426617039142,
-      "loss": 46.0,
-      "step": 35074
-    },
-    {
-      "epoch": 5.648576834816216,
-      "grad_norm": 0.0016620560782030225,
-      "learning_rate": 0.00019998426527296722,
-      "loss": 46.0,
-      "step": 35075
-    },
-    {
-      "epoch": 5.648737871895004,
-      "grad_norm": 0.007439692039042711,
-      "learning_rate": 0.0001999842643755174,
-      "loss": 46.0,
-      "step": 35076
-    },
-    {
-      "epoch": 5.648898908973791,
-      "grad_norm": 0.0034636766649782658,
-      "learning_rate": 0.000199984263478042,
-      "loss": 46.0,
-      "step": 35077
-    },
-    {
-      "epoch": 5.649059946052579,
-      "grad_norm": 0.004421850200742483,
-      "learning_rate": 0.00019998426258054103,
-      "loss": 46.0,
-      "step": 35078
-    },
-    {
-      "epoch": 5.649220983131366,
-      "grad_norm": 0.004357545170933008,
-      "learning_rate": 0.00019998426168301445,
-      "loss": 46.0,
-      "step": 35079
-    },
-    {
-      "epoch": 5.649382020210154,
-      "grad_norm": 0.0015708237187936902,
-      "learning_rate": 0.0001999842607854623,
-      "loss": 46.0,
-      "step": 35080
-    },
-    {
-      "epoch": 5.649543057288941,
-      "grad_norm": 0.005285588093101978,
-      "learning_rate": 0.00019998425988788452,
-      "loss": 46.0,
-      "step": 35081
-    },
-    {
-      "epoch": 5.649704094367728,
-      "grad_norm": 0.010058462619781494,
-      "learning_rate": 0.00019998425899028118,
-      "loss": 46.0,
-      "step": 35082
-    },
-    {
-      "epoch": 5.649865131446516,
-      "grad_norm": 0.009400948882102966,
-      "learning_rate": 0.00019998425809265223,
-      "loss": 46.0,
-      "step": 35083
-    },
-    {
-      "epoch": 5.650026168525303,
-      "grad_norm": 0.003386115189641714,
-      "learning_rate": 0.0001999842571949977,
-      "loss": 46.0,
-      "step": 35084
-    },
-    {
-      "epoch": 5.65018720560409,
-      "grad_norm": 0.023702038452029228,
-      "learning_rate": 0.00019998425629731757,
-      "loss": 46.0,
-      "step": 35085
-    },
-    {
-      "epoch": 5.6503482426828775,
-      "grad_norm": 0.0018651328282430768,
-      "learning_rate": 0.00019998425539961186,
-      "loss": 46.0,
-      "step": 35086
-    },
-    {
-      "epoch": 5.650509279761665,
-      "grad_norm": 0.00218061450868845,
-      "learning_rate": 0.00019998425450188054,
-      "loss": 46.0,
-      "step": 35087
-    },
-    {
-      "epoch": 5.650670316840452,
-      "grad_norm": 0.002754851710051298,
-      "learning_rate": 0.00019998425360412365,
-      "loss": 46.0,
-      "step": 35088
-    },
-    {
-      "epoch": 5.65083135391924,
-      "grad_norm": 0.001908688573166728,
-      "learning_rate": 0.00019998425270634118,
-      "loss": 46.0,
-      "step": 35089
-    },
-    {
-      "epoch": 5.650992390998027,
-      "grad_norm": 0.011842873878777027,
-      "learning_rate": 0.0001999842518085331,
-      "loss": 46.0,
-      "step": 35090
-    },
-    {
-      "epoch": 5.651153428076815,
-      "grad_norm": 0.0058770552277565,
-      "learning_rate": 0.00019998425091069942,
-      "loss": 46.0,
-      "step": 35091
-    },
-    {
-      "epoch": 5.651314465155602,
-      "grad_norm": 0.011124215088784695,
-      "learning_rate": 0.00019998425001284016,
-      "loss": 46.0,
-      "step": 35092
-    },
-    {
-      "epoch": 5.65147550223439,
-      "grad_norm": 0.0013297568075358868,
-      "learning_rate": 0.00019998424911495534,
-      "loss": 46.0,
-      "step": 35093
-    },
-    {
-      "epoch": 5.651636539313177,
-      "grad_norm": 0.0024605095386505127,
-      "learning_rate": 0.00019998424821704487,
-      "loss": 46.0,
-      "step": 35094
-    },
-    {
-      "epoch": 5.651797576391965,
-      "grad_norm": 0.0023545182775706053,
-      "learning_rate": 0.00019998424731910882,
-      "loss": 46.0,
-      "step": 35095
-    },
-    {
-      "epoch": 5.651958613470752,
-      "grad_norm": 0.00208808365277946,
-      "learning_rate": 0.00019998424642114722,
-      "loss": 46.0,
-      "step": 35096
-    },
-    {
-      "epoch": 5.652119650549539,
-      "grad_norm": 0.006952462252229452,
-      "learning_rate": 0.00019998424552316002,
-      "loss": 46.0,
-      "step": 35097
-    },
-    {
-      "epoch": 5.652280687628327,
-      "grad_norm": 0.0029632486402988434,
-      "learning_rate": 0.0001999842446251472,
-      "loss": 46.0,
-      "step": 35098
-    },
-    {
-      "epoch": 5.6524417247071135,
-      "grad_norm": 0.006912900134921074,
-      "learning_rate": 0.0001999842437271088,
-      "loss": 46.0,
-      "step": 35099
-    },
-    {
-      "epoch": 5.652602761785901,
-      "grad_norm": 0.012486535124480724,
-      "learning_rate": 0.00019998424282904482,
-      "loss": 46.0,
-      "step": 35100
-    },
-    {
-      "epoch": 5.652763798864688,
-      "grad_norm": 0.002211576560512185,
-      "learning_rate": 0.00019998424193095525,
-      "loss": 46.0,
-      "step": 35101
-    },
-    {
-      "epoch": 5.652924835943476,
-      "grad_norm": 0.0016587079735472798,
-      "learning_rate": 0.0001999842410328401,
-      "loss": 46.0,
-      "step": 35102
-    },
-    {
-      "epoch": 5.653085873022263,
-      "grad_norm": 0.0007916645845398307,
-      "learning_rate": 0.00019998424013469932,
-      "loss": 46.0,
-      "step": 35103
-    },
-    {
-      "epoch": 5.653246910101051,
-      "grad_norm": 0.006694192532449961,
-      "learning_rate": 0.00019998423923653298,
-      "loss": 46.0,
-      "step": 35104
-    },
-    {
-      "epoch": 5.653407947179838,
-      "grad_norm": 0.006458644289523363,
-      "learning_rate": 0.000199984238338341,
-      "loss": 46.0,
-      "step": 35105
-    },
-    {
-      "epoch": 5.653568984258626,
-      "grad_norm": 0.0032324877101927996,
-      "learning_rate": 0.0001999842374401235,
-      "loss": 46.0,
-      "step": 35106
-    },
-    {
-      "epoch": 5.653730021337413,
-      "grad_norm": 0.0024469299241900444,
-      "learning_rate": 0.00019998423654188038,
-      "loss": 46.0,
-      "step": 35107
-    },
-    {
-      "epoch": 5.6538910584162005,
-      "grad_norm": 0.0029714880511164665,
-      "learning_rate": 0.00019998423564361167,
-      "loss": 46.0,
-      "step": 35108
-    },
-    {
-      "epoch": 5.654052095494988,
-      "grad_norm": 0.006573381833732128,
-      "learning_rate": 0.00019998423474531734,
-      "loss": 46.0,
-      "step": 35109
-    },
-    {
-      "epoch": 5.6542131325737754,
-      "grad_norm": 0.005313725210726261,
-      "learning_rate": 0.00019998423384699746,
-      "loss": 46.0,
-      "step": 35110
-    },
-    {
-      "epoch": 5.654374169652563,
-      "grad_norm": 0.00250240508466959,
-      "learning_rate": 0.000199984232948652,
-      "loss": 46.0,
-      "step": 35111
-    },
-    {
-      "epoch": 5.6545352067313495,
-      "grad_norm": 0.010545754805207253,
-      "learning_rate": 0.0001999842320502809,
-      "loss": 46.0,
-      "step": 35112
-    },
-    {
-      "epoch": 5.654696243810137,
-      "grad_norm": 0.009056560695171356,
-      "learning_rate": 0.00019998423115188426,
-      "loss": 46.0,
-      "step": 35113
-    },
-    {
-      "epoch": 5.654857280888924,
-      "grad_norm": 0.01301497407257557,
-      "learning_rate": 0.00019998423025346197,
-      "loss": 46.0,
-      "step": 35114
-    },
-    {
-      "epoch": 5.655018317967712,
-      "grad_norm": 0.0027683754451572895,
-      "learning_rate": 0.00019998422935501412,
-      "loss": 46.0,
-      "step": 35115
-    },
-    {
-      "epoch": 5.655179355046499,
-      "grad_norm": 0.0025818690191954374,
-      "learning_rate": 0.0001999842284565407,
-      "loss": 46.0,
-      "step": 35116
-    },
-    {
-      "epoch": 5.655340392125287,
-      "grad_norm": 0.009465574286878109,
-      "learning_rate": 0.00019998422755804164,
-      "loss": 46.0,
-      "step": 35117
-    },
-    {
-      "epoch": 5.655501429204074,
-      "grad_norm": 0.006048796698451042,
-      "learning_rate": 0.00019998422665951703,
-      "loss": 46.0,
-      "step": 35118
-    },
-    {
-      "epoch": 5.655662466282862,
-      "grad_norm": 0.0030552742537111044,
-      "learning_rate": 0.00019998422576096683,
-      "loss": 46.0,
-      "step": 35119
-    },
-    {
-      "epoch": 5.655823503361649,
-      "grad_norm": 0.012277650646865368,
-      "learning_rate": 0.00019998422486239102,
-      "loss": 46.0,
-      "step": 35120
-    },
-    {
-      "epoch": 5.6559845404404365,
-      "grad_norm": 0.004479520954191685,
-      "learning_rate": 0.00019998422396378965,
-      "loss": 46.0,
-      "step": 35121
-    },
-    {
-      "epoch": 5.656145577519224,
-      "grad_norm": 0.001483603729866445,
-      "learning_rate": 0.00019998422306516264,
-      "loss": 46.0,
-      "step": 35122
-    },
-    {
-      "epoch": 5.656306614598011,
-      "grad_norm": 0.002450986998155713,
-      "learning_rate": 0.00019998422216651007,
-      "loss": 46.0,
-      "step": 35123
-    },
-    {
-      "epoch": 5.656467651676799,
-      "grad_norm": 0.008994616568088531,
-      "learning_rate": 0.0001999842212678319,
-      "loss": 46.0,
-      "step": 35124
-    },
-    {
-      "epoch": 5.656628688755586,
-      "grad_norm": 0.014238941483199596,
-      "learning_rate": 0.00019998422036912814,
-      "loss": 46.0,
-      "step": 35125
-    },
-    {
-      "epoch": 5.656789725834374,
-      "grad_norm": 0.0024231786374002695,
-      "learning_rate": 0.0001999842194703988,
-      "loss": 46.0,
-      "step": 35126
-    },
-    {
-      "epoch": 5.65695076291316,
-      "grad_norm": 0.009421883150935173,
-      "learning_rate": 0.00019998421857164386,
-      "loss": 46.0,
-      "step": 35127
-    },
-    {
-      "epoch": 5.657111799991948,
-      "grad_norm": 0.011003999970853329,
-      "learning_rate": 0.00019998421767286332,
-      "loss": 46.0,
-      "step": 35128
-    },
-    {
-      "epoch": 5.657272837070735,
-      "grad_norm": 0.009684274904429913,
-      "learning_rate": 0.00019998421677405723,
-      "loss": 46.0,
-      "step": 35129
-    },
-    {
-      "epoch": 5.657433874149523,
-      "grad_norm": 0.0013297874247655272,
-      "learning_rate": 0.00019998421587522552,
-      "loss": 46.0,
-      "step": 35130
-    },
-    {
-      "epoch": 5.65759491122831,
-      "grad_norm": 0.006396141368895769,
-      "learning_rate": 0.00019998421497636822,
-      "loss": 46.0,
-      "step": 35131
-    },
-    {
-      "epoch": 5.657755948307098,
-      "grad_norm": 0.0022974489256739616,
-      "learning_rate": 0.00019998421407748531,
-      "loss": 46.0,
-      "step": 35132
-    },
-    {
-      "epoch": 5.657916985385885,
-      "grad_norm": 0.003934463020414114,
-      "learning_rate": 0.00019998421317857682,
-      "loss": 46.0,
-      "step": 35133
-    },
-    {
-      "epoch": 5.6580780224646725,
-      "grad_norm": 0.0018786523723974824,
-      "learning_rate": 0.00019998421227964276,
-      "loss": 46.0,
-      "step": 35134
-    },
-    {
-      "epoch": 5.65823905954346,
-      "grad_norm": 0.0068613323383033276,
-      "learning_rate": 0.0001999842113806831,
-      "loss": 46.0,
-      "step": 35135
-    },
-    {
-      "epoch": 5.658400096622247,
-      "grad_norm": 0.005072134081274271,
-      "learning_rate": 0.00019998421048169785,
-      "loss": 46.0,
-      "step": 35136
-    },
-    {
-      "epoch": 5.658561133701035,
-      "grad_norm": 0.010304098017513752,
-      "learning_rate": 0.00019998420958268698,
-      "loss": 46.0,
-      "step": 35137
-    },
-    {
-      "epoch": 5.658722170779822,
-      "grad_norm": 0.004165279679000378,
-      "learning_rate": 0.00019998420868365055,
-      "loss": 46.0,
-      "step": 35138
-    },
-    {
-      "epoch": 5.65888320785861,
-      "grad_norm": 0.01088697835803032,
-      "learning_rate": 0.0001999842077845885,
-      "loss": 46.0,
-      "step": 35139
-    },
-    {
-      "epoch": 5.659044244937397,
-      "grad_norm": 0.0009569205576553941,
-      "learning_rate": 0.0001999842068855009,
-      "loss": 46.0,
-      "step": 35140
-    },
-    {
-      "epoch": 5.659205282016185,
-      "grad_norm": 0.025378096848726273,
-      "learning_rate": 0.0001999842059863877,
-      "loss": 46.0,
-      "step": 35141
-    },
-    {
-      "epoch": 5.659366319094971,
-      "grad_norm": 0.002211647806689143,
-      "learning_rate": 0.0001999842050872489,
-      "loss": 46.0,
-      "step": 35142
-    },
-    {
-      "epoch": 5.659527356173759,
-      "grad_norm": 0.013512805104255676,
-      "learning_rate": 0.0001999842041880845,
-      "loss": 46.0,
-      "step": 35143
-    },
-    {
-      "epoch": 5.659688393252546,
-      "grad_norm": 0.0023886284325271845,
-      "learning_rate": 0.0001999842032888945,
-      "loss": 46.0,
-      "step": 35144
-    },
-    {
-      "epoch": 5.659849430331334,
-      "grad_norm": 0.004233606159687042,
-      "learning_rate": 0.00019998420238967894,
-      "loss": 46.0,
-      "step": 35145
-    },
-    {
-      "epoch": 5.660010467410121,
-      "grad_norm": 0.00212499825283885,
-      "learning_rate": 0.0001999842014904378,
-      "loss": 46.0,
-      "step": 35146
-    },
-    {
-      "epoch": 5.6601715044889085,
-      "grad_norm": 0.0006795204244554043,
-      "learning_rate": 0.00019998420059117104,
-      "loss": 46.0,
-      "step": 35147
-    },
-    {
-      "epoch": 5.660332541567696,
-      "grad_norm": 0.025922708213329315,
-      "learning_rate": 0.00019998419969187868,
-      "loss": 46.0,
-      "step": 35148
-    },
-    {
-      "epoch": 5.660493578646483,
-      "grad_norm": 0.026236068457365036,
-      "learning_rate": 0.00019998419879256073,
-      "loss": 46.0,
-      "step": 35149
-    },
-    {
-      "epoch": 5.660654615725271,
-      "grad_norm": 0.004925851244479418,
-      "learning_rate": 0.00019998419789321722,
-      "loss": 46.0,
-      "step": 35150
-    },
-    {
-      "epoch": 5.660815652804058,
-      "grad_norm": 0.004672850947827101,
-      "learning_rate": 0.0001999841969938481,
-      "loss": 46.0,
-      "step": 35151
-    },
-    {
-      "epoch": 5.660976689882846,
-      "grad_norm": 0.00997094251215458,
-      "learning_rate": 0.00019998419609445337,
-      "loss": 46.0,
-      "step": 35152
-    },
-    {
-      "epoch": 5.661137726961633,
-      "grad_norm": 0.005512923467904329,
-      "learning_rate": 0.0001999841951950331,
-      "loss": 46.0,
-      "step": 35153
-    },
-    {
-      "epoch": 5.661298764040421,
-      "grad_norm": 0.005155325401574373,
-      "learning_rate": 0.00019998419429558722,
-      "loss": 46.0,
-      "step": 35154
-    },
-    {
-      "epoch": 5.661459801119207,
-      "grad_norm": 0.0025486720260232687,
-      "learning_rate": 0.00019998419339611572,
-      "loss": 46.0,
-      "step": 35155
-    },
-    {
-      "epoch": 5.6616208381979956,
-      "grad_norm": 0.006066937930881977,
-      "learning_rate": 0.00019998419249661864,
-      "loss": 46.0,
-      "step": 35156
-    },
-    {
-      "epoch": 5.661781875276782,
-      "grad_norm": 0.005126535426825285,
-      "learning_rate": 0.000199984191597096,
-      "loss": 46.0,
-      "step": 35157
-    },
-    {
-      "epoch": 5.66194291235557,
-      "grad_norm": 0.010907979682087898,
-      "learning_rate": 0.00019998419069754773,
-      "loss": 46.0,
-      "step": 35158
-    },
-    {
-      "epoch": 5.662103949434357,
-      "grad_norm": 0.002863509114831686,
-      "learning_rate": 0.00019998418979797386,
-      "loss": 46.0,
-      "step": 35159
-    },
-    {
-      "epoch": 5.6622649865131445,
-      "grad_norm": 0.004635497462004423,
-      "learning_rate": 0.00019998418889837446,
-      "loss": 46.0,
-      "step": 35160
-    },
-    {
-      "epoch": 5.662426023591932,
-      "grad_norm": 0.0017995884409174323,
-      "learning_rate": 0.0001999841879987494,
-      "loss": 46.0,
-      "step": 35161
-    },
-    {
-      "epoch": 5.662587060670719,
-      "grad_norm": 0.0023648450151085854,
-      "learning_rate": 0.00019998418709909878,
-      "loss": 46.0,
-      "step": 35162
-    },
-    {
-      "epoch": 5.662748097749507,
-      "grad_norm": 0.007384988013654947,
-      "learning_rate": 0.00019998418619942258,
-      "loss": 46.0,
-      "step": 35163
-    },
-    {
-      "epoch": 5.662909134828294,
-      "grad_norm": 0.005359169561415911,
-      "learning_rate": 0.0001999841852997208,
-      "loss": 46.0,
-      "step": 35164
-    },
-    {
-      "epoch": 5.663070171907082,
-      "grad_norm": 0.0016436147270724177,
-      "learning_rate": 0.0001999841843999934,
-      "loss": 46.0,
-      "step": 35165
-    },
-    {
-      "epoch": 5.663231208985869,
-      "grad_norm": 0.002862224355340004,
-      "learning_rate": 0.00019998418350024042,
-      "loss": 46.0,
-      "step": 35166
-    },
-    {
-      "epoch": 5.663392246064657,
-      "grad_norm": 0.007666591554880142,
-      "learning_rate": 0.00019998418260046185,
-      "loss": 46.0,
-      "step": 35167
-    },
-    {
-      "epoch": 5.663553283143444,
-      "grad_norm": 0.002966654021292925,
-      "learning_rate": 0.0001999841817006577,
-      "loss": 46.0,
-      "step": 35168
-    },
-    {
-      "epoch": 5.6637143202222315,
-      "grad_norm": 0.0016130894655361772,
-      "learning_rate": 0.00019998418080082795,
-      "loss": 46.0,
-      "step": 35169
-    },
-    {
-      "epoch": 5.663875357301018,
-      "grad_norm": 0.003145878668874502,
-      "learning_rate": 0.0001999841799009726,
-      "loss": 46.0,
-      "step": 35170
-    },
-    {
-      "epoch": 5.664036394379806,
-      "grad_norm": 0.017086086794734,
-      "learning_rate": 0.00019998417900109165,
-      "loss": 46.0,
-      "step": 35171
-    },
-    {
-      "epoch": 5.664197431458593,
-      "grad_norm": 0.005891021341085434,
-      "learning_rate": 0.00019998417810118511,
-      "loss": 46.0,
-      "step": 35172
-    },
-    {
-      "epoch": 5.6643584685373805,
-      "grad_norm": 0.006925859954208136,
-      "learning_rate": 0.00019998417720125302,
-      "loss": 46.0,
-      "step": 35173
-    },
-    {
-      "epoch": 5.664519505616168,
-      "grad_norm": 0.0031833990942686796,
-      "learning_rate": 0.00019998417630129531,
-      "loss": 46.0,
-      "step": 35174
-    },
-    {
-      "epoch": 5.664680542694955,
-      "grad_norm": 0.006922585424035788,
-      "learning_rate": 0.000199984175401312,
-      "loss": 46.0,
-      "step": 35175
-    },
-    {
-      "epoch": 5.664841579773743,
-      "grad_norm": 0.0012990660034120083,
-      "learning_rate": 0.00019998417450130314,
-      "loss": 46.0,
-      "step": 35176
-    },
-    {
-      "epoch": 5.66500261685253,
-      "grad_norm": 0.002073422772809863,
-      "learning_rate": 0.00019998417360126864,
-      "loss": 46.0,
-      "step": 35177
-    },
-    {
-      "epoch": 5.665163653931318,
-      "grad_norm": 0.011119670234620571,
-      "learning_rate": 0.00019998417270120856,
-      "loss": 46.0,
-      "step": 35178
-    },
-    {
-      "epoch": 5.665324691010105,
-      "grad_norm": 0.004305449780076742,
-      "learning_rate": 0.00019998417180112292,
-      "loss": 46.0,
-      "step": 35179
-    },
-    {
-      "epoch": 5.665485728088893,
-      "grad_norm": 0.004370627924799919,
-      "learning_rate": 0.00019998417090101166,
-      "loss": 46.0,
-      "step": 35180
-    },
-    {
-      "epoch": 5.66564676516768,
-      "grad_norm": 0.002213856903836131,
-      "learning_rate": 0.00019998417000087482,
-      "loss": 46.0,
-      "step": 35181
-    },
-    {
-      "epoch": 5.6658078022464675,
-      "grad_norm": 0.015276379883289337,
-      "learning_rate": 0.00019998416910071238,
-      "loss": 46.0,
-      "step": 35182
-    },
-    {
-      "epoch": 5.665968839325255,
-      "grad_norm": 0.018438534811139107,
-      "learning_rate": 0.00019998416820052434,
-      "loss": 46.0,
-      "step": 35183
-    },
-    {
-      "epoch": 5.666129876404042,
-      "grad_norm": 0.0010997239733114839,
-      "learning_rate": 0.00019998416730031076,
-      "loss": 46.0,
-      "step": 35184
-    },
-    {
-      "epoch": 5.666290913482829,
-      "grad_norm": 0.015043821185827255,
-      "learning_rate": 0.00019998416640007154,
-      "loss": 46.0,
-      "step": 35185
-    },
-    {
-      "epoch": 5.666451950561616,
-      "grad_norm": 0.0033091683872044086,
-      "learning_rate": 0.00019998416549980673,
-      "loss": 46.0,
-      "step": 35186
-    },
-    {
-      "epoch": 5.666612987640404,
-      "grad_norm": 0.002052116673439741,
-      "learning_rate": 0.00019998416459951636,
-      "loss": 46.0,
-      "step": 35187
-    },
-    {
-      "epoch": 5.666774024719191,
-      "grad_norm": 0.004926176276057959,
-      "learning_rate": 0.00019998416369920036,
-      "loss": 46.0,
-      "step": 35188
-    },
-    {
-      "epoch": 5.666935061797979,
-      "grad_norm": 0.0025779197458177805,
-      "learning_rate": 0.00019998416279885881,
-      "loss": 46.0,
-      "step": 35189
-    },
-    {
-      "epoch": 5.667096098876766,
-      "grad_norm": 0.012536870315670967,
-      "learning_rate": 0.00019998416189849163,
-      "loss": 46.0,
-      "step": 35190
-    },
-    {
-      "epoch": 5.667257135955554,
-      "grad_norm": 0.004109465982764959,
-      "learning_rate": 0.0001999841609980989,
-      "loss": 46.0,
-      "step": 35191
-    },
-    {
-      "epoch": 5.667418173034341,
-      "grad_norm": 0.0008402433013543487,
-      "learning_rate": 0.00019998416009768056,
-      "loss": 46.0,
-      "step": 35192
-    },
-    {
-      "epoch": 5.667579210113129,
-      "grad_norm": 0.004334953147917986,
-      "learning_rate": 0.00019998415919723664,
-      "loss": 46.0,
-      "step": 35193
-    },
-    {
-      "epoch": 5.667740247191916,
-      "grad_norm": 0.004303778056055307,
-      "learning_rate": 0.0001999841582967671,
-      "loss": 46.0,
-      "step": 35194
-    },
-    {
-      "epoch": 5.6679012842707035,
-      "grad_norm": 0.008808623068034649,
-      "learning_rate": 0.00019998415739627196,
-      "loss": 46.0,
-      "step": 35195
-    },
-    {
-      "epoch": 5.668062321349491,
-      "grad_norm": 0.00121975876390934,
-      "learning_rate": 0.00019998415649575128,
-      "loss": 46.0,
-      "step": 35196
-    },
-    {
-      "epoch": 5.668223358428278,
-      "grad_norm": 0.002060552593320608,
-      "learning_rate": 0.000199984155595205,
-      "loss": 46.0,
-      "step": 35197
-    },
-    {
-      "epoch": 5.668384395507066,
-      "grad_norm": 0.0032270464580506086,
-      "learning_rate": 0.0001999841546946331,
-      "loss": 46.0,
-      "step": 35198
-    },
-    {
-      "epoch": 5.668545432585853,
-      "grad_norm": 0.002035292563959956,
-      "learning_rate": 0.00019998415379403561,
-      "loss": 46.0,
-      "step": 35199
-    },
-    {
-      "epoch": 5.66870646966464,
-      "grad_norm": 0.002801191294565797,
-      "learning_rate": 0.00019998415289341256,
-      "loss": 46.0,
-      "step": 35200
-    },
-    {
-      "epoch": 5.668867506743427,
-      "grad_norm": 0.011713326908648014,
-      "learning_rate": 0.0001999841519927639,
-      "loss": 46.0,
-      "step": 35201
-    },
-    {
-      "epoch": 5.669028543822215,
-      "grad_norm": 0.004615049343556166,
-      "learning_rate": 0.00019998415109208963,
-      "loss": 46.0,
-      "step": 35202
-    },
-    {
-      "epoch": 5.669189580901002,
-      "grad_norm": 0.01085527054965496,
-      "learning_rate": 0.00019998415019138982,
-      "loss": 46.0,
-      "step": 35203
-    },
-    {
-      "epoch": 5.66935061797979,
-      "grad_norm": 0.0012078335275873542,
-      "learning_rate": 0.00019998414929066436,
-      "loss": 46.0,
-      "step": 35204
-    },
-    {
-      "epoch": 5.669511655058577,
-      "grad_norm": 0.005233896430581808,
-      "learning_rate": 0.00019998414838991334,
-      "loss": 46.0,
-      "step": 35205
-    },
-    {
-      "epoch": 5.669672692137365,
-      "grad_norm": 0.001537130563519895,
-      "learning_rate": 0.0001999841474891367,
-      "loss": 46.0,
-      "step": 35206
-    },
-    {
-      "epoch": 5.669833729216152,
-      "grad_norm": 0.011775372549891472,
-      "learning_rate": 0.00019998414658833454,
-      "loss": 46.0,
-      "step": 35207
-    },
-    {
-      "epoch": 5.6699947662949395,
-      "grad_norm": 0.008319418877363205,
-      "learning_rate": 0.0001999841456875067,
-      "loss": 46.0,
-      "step": 35208
-    },
-    {
-      "epoch": 5.670155803373727,
-      "grad_norm": 0.0018707100534811616,
-      "learning_rate": 0.00019998414478665334,
-      "loss": 46.0,
-      "step": 35209
-    },
-    {
-      "epoch": 5.670316840452514,
-      "grad_norm": 0.0029868409037590027,
-      "learning_rate": 0.00019998414388577436,
-      "loss": 46.0,
-      "step": 35210
-    },
-    {
-      "epoch": 5.670477877531302,
-      "grad_norm": 0.01568084955215454,
-      "learning_rate": 0.0001999841429848698,
-      "loss": 46.0,
-      "step": 35211
-    },
-    {
-      "epoch": 5.670638914610089,
-      "grad_norm": 0.005840482655912638,
-      "learning_rate": 0.00019998414208393964,
-      "loss": 46.0,
-      "step": 35212
-    },
-    {
-      "epoch": 5.670799951688877,
-      "grad_norm": 0.0013728828635066748,
-      "learning_rate": 0.0001999841411829839,
-      "loss": 46.0,
-      "step": 35213
-    },
-    {
-      "epoch": 5.670960988767664,
-      "grad_norm": 0.006242575589567423,
-      "learning_rate": 0.00019998414028200254,
-      "loss": 46.0,
-      "step": 35214
-    },
-    {
-      "epoch": 5.671122025846451,
-      "grad_norm": 0.004143235739320517,
-      "learning_rate": 0.00019998413938099562,
-      "loss": 46.0,
-      "step": 35215
-    },
-    {
-      "epoch": 5.671283062925238,
-      "grad_norm": 0.005876758135855198,
-      "learning_rate": 0.0001999841384799631,
-      "loss": 46.0,
-      "step": 35216
-    },
-    {
-      "epoch": 5.671444100004026,
-      "grad_norm": 0.0059021892957389355,
-      "learning_rate": 0.000199984137578905,
-      "loss": 46.0,
-      "step": 35217
-    },
-    {
-      "epoch": 5.671605137082813,
-      "grad_norm": 0.012267417274415493,
-      "learning_rate": 0.00019998413667782127,
-      "loss": 46.0,
-      "step": 35218
-    },
-    {
-      "epoch": 5.671766174161601,
-      "grad_norm": 0.0027927905321121216,
-      "learning_rate": 0.000199984135776712,
-      "loss": 46.0,
-      "step": 35219
-    },
-    {
-      "epoch": 5.671927211240388,
-      "grad_norm": 0.0007919935742393136,
-      "learning_rate": 0.0001999841348755771,
-      "loss": 46.0,
-      "step": 35220
-    },
-    {
-      "epoch": 5.6720882483191755,
-      "grad_norm": 0.002068153116852045,
-      "learning_rate": 0.00019998413397441663,
-      "loss": 46.0,
-      "step": 35221
-    },
-    {
-      "epoch": 5.672249285397963,
-      "grad_norm": 0.006332676392048597,
-      "learning_rate": 0.00019998413307323055,
-      "loss": 46.0,
-      "step": 35222
-    },
-    {
-      "epoch": 5.67241032247675,
-      "grad_norm": 0.001940408954396844,
-      "learning_rate": 0.00019998413217201894,
-      "loss": 46.0,
-      "step": 35223
-    },
-    {
-      "epoch": 5.672571359555538,
-      "grad_norm": 0.0012789057800546288,
-      "learning_rate": 0.00019998413127078165,
-      "loss": 46.0,
-      "step": 35224
-    },
-    {
-      "epoch": 5.672732396634325,
-      "grad_norm": 0.01430367399007082,
-      "learning_rate": 0.0001999841303695188,
-      "loss": 46.0,
-      "step": 35225
-    },
-    {
-      "epoch": 5.672893433713113,
-      "grad_norm": 0.0017292052507400513,
-      "learning_rate": 0.0001999841294682304,
-      "loss": 46.0,
-      "step": 35226
-    },
-    {
-      "epoch": 5.6730544707919,
-      "grad_norm": 0.004862557630985975,
-      "learning_rate": 0.00019998412856691636,
-      "loss": 46.0,
-      "step": 35227
-    },
-    {
-      "epoch": 5.673215507870687,
-      "grad_norm": 0.0033276767935603857,
-      "learning_rate": 0.00019998412766557676,
-      "loss": 46.0,
-      "step": 35228
-    },
-    {
-      "epoch": 5.673376544949475,
-      "grad_norm": 0.001874297740869224,
-      "learning_rate": 0.00019998412676421157,
-      "loss": 46.0,
-      "step": 35229
-    },
-    {
-      "epoch": 5.673537582028262,
-      "grad_norm": 0.011564243584871292,
-      "learning_rate": 0.0001999841258628208,
-      "loss": 46.0,
-      "step": 35230
-    },
-    {
-      "epoch": 5.673698619107049,
-      "grad_norm": 0.01121477223932743,
-      "learning_rate": 0.00019998412496140437,
-      "loss": 46.0,
-      "step": 35231
-    },
-    {
-      "epoch": 5.6738596561858365,
-      "grad_norm": 0.002650787588208914,
-      "learning_rate": 0.00019998412405996242,
-      "loss": 46.0,
-      "step": 35232
-    },
-    {
-      "epoch": 5.674020693264624,
-      "grad_norm": 0.004733751993626356,
-      "learning_rate": 0.00019998412315849485,
-      "loss": 46.0,
-      "step": 35233
-    },
-    {
-      "epoch": 5.6741817303434114,
-      "grad_norm": 0.006988368928432465,
-      "learning_rate": 0.0001999841222570017,
-      "loss": 46.0,
-      "step": 35234
-    },
-    {
-      "epoch": 5.674342767422199,
-      "grad_norm": 0.0037831293884664774,
-      "learning_rate": 0.00019998412135548296,
-      "loss": 46.0,
-      "step": 35235
-    },
-    {
-      "epoch": 5.674503804500986,
-      "grad_norm": 0.024250324815511703,
-      "learning_rate": 0.0001999841204539386,
-      "loss": 46.0,
-      "step": 35236
-    },
-    {
-      "epoch": 5.674664841579774,
-      "grad_norm": 0.0014020624803379178,
-      "learning_rate": 0.00019998411955236866,
-      "loss": 46.0,
-      "step": 35237
-    },
-    {
-      "epoch": 5.674825878658561,
-      "grad_norm": 0.019380206242203712,
-      "learning_rate": 0.00019998411865077315,
-      "loss": 46.0,
-      "step": 35238
-    },
-    {
-      "epoch": 5.674986915737349,
-      "grad_norm": 0.006248376332223415,
-      "learning_rate": 0.00019998411774915204,
-      "loss": 46.0,
-      "step": 35239
-    },
-    {
-      "epoch": 5.675147952816136,
-      "grad_norm": 0.0027235010638833046,
-      "learning_rate": 0.00019998411684750536,
-      "loss": 46.0,
-      "step": 35240
-    },
-    {
-      "epoch": 5.675308989894924,
-      "grad_norm": 0.007138578686863184,
-      "learning_rate": 0.00019998411594583307,
-      "loss": 46.0,
-      "step": 35241
-    },
-    {
-      "epoch": 5.675470026973711,
-      "grad_norm": 0.006366404704749584,
-      "learning_rate": 0.0001999841150441352,
-      "loss": 46.0,
-      "step": 35242
-    },
-    {
-      "epoch": 5.675631064052498,
-      "grad_norm": 0.0036704582162201405,
-      "learning_rate": 0.0001999841141424117,
-      "loss": 46.0,
-      "step": 35243
-    },
-    {
-      "epoch": 5.675792101131286,
-      "grad_norm": 0.01133080292493105,
-      "learning_rate": 0.00019998411324066264,
-      "loss": 46.0,
-      "step": 35244
-    },
-    {
-      "epoch": 5.6759531382100725,
-      "grad_norm": 0.003278769087046385,
-      "learning_rate": 0.00019998411233888798,
-      "loss": 46.0,
-      "step": 35245
-    },
-    {
-      "epoch": 5.67611417528886,
-      "grad_norm": 0.007010200992226601,
-      "learning_rate": 0.00019998411143708772,
-      "loss": 46.0,
-      "step": 35246
-    },
-    {
-      "epoch": 5.676275212367647,
-      "grad_norm": 0.013183454982936382,
-      "learning_rate": 0.0001999841105352619,
-      "loss": 46.0,
-      "step": 35247
-    },
-    {
-      "epoch": 5.676436249446435,
-      "grad_norm": 0.0028443082701414824,
-      "learning_rate": 0.00019998410963341048,
-      "loss": 46.0,
-      "step": 35248
-    },
-    {
-      "epoch": 5.676597286525222,
-      "grad_norm": 0.0029535042122006416,
-      "learning_rate": 0.00019998410873153346,
-      "loss": 46.0,
-      "step": 35249
-    },
-    {
-      "epoch": 5.67675832360401,
-      "grad_norm": 0.0029641715809702873,
-      "learning_rate": 0.00019998410782963086,
-      "loss": 46.0,
-      "step": 35250
-    },
-    {
-      "epoch": 5.676919360682797,
-      "grad_norm": 0.0032623030710965395,
-      "learning_rate": 0.00019998410692770264,
-      "loss": 46.0,
-      "step": 35251
-    },
-    {
-      "epoch": 5.677080397761585,
-      "grad_norm": 0.006206602323800325,
-      "learning_rate": 0.00019998410602574886,
-      "loss": 46.0,
-      "step": 35252
-    },
-    {
-      "epoch": 5.677241434840372,
-      "grad_norm": 0.004512494429945946,
-      "learning_rate": 0.00019998410512376947,
-      "loss": 46.0,
-      "step": 35253
-    },
-    {
-      "epoch": 5.67740247191916,
-      "grad_norm": 0.005040169693529606,
-      "learning_rate": 0.0001999841042217645,
-      "loss": 46.0,
-      "step": 35254
-    },
-    {
-      "epoch": 5.677563508997947,
-      "grad_norm": 0.002800359157845378,
-      "learning_rate": 0.00019998410331973393,
-      "loss": 46.0,
-      "step": 35255
-    },
-    {
-      "epoch": 5.6777245460767345,
-      "grad_norm": 0.009813678450882435,
-      "learning_rate": 0.0001999841024176778,
-      "loss": 46.0,
-      "step": 35256
-    },
-    {
-      "epoch": 5.677885583155522,
-      "grad_norm": 0.0008741789497435093,
-      "learning_rate": 0.00019998410151559606,
-      "loss": 46.0,
-      "step": 35257
-    },
-    {
-      "epoch": 5.6780466202343085,
-      "grad_norm": 0.001780940336175263,
-      "learning_rate": 0.0001999841006134887,
-      "loss": 46.0,
-      "step": 35258
-    },
-    {
-      "epoch": 5.678207657313096,
-      "grad_norm": 0.003297362709417939,
-      "learning_rate": 0.0001999840997113558,
-      "loss": 46.0,
-      "step": 35259
-    },
-    {
-      "epoch": 5.678368694391883,
-      "grad_norm": 0.013721270486712456,
-      "learning_rate": 0.00019998409880919726,
-      "loss": 46.0,
-      "step": 35260
-    },
-    {
-      "epoch": 5.678529731470671,
-      "grad_norm": 0.003660049755126238,
-      "learning_rate": 0.00019998409790701315,
-      "loss": 46.0,
-      "step": 35261
-    },
-    {
-      "epoch": 5.678690768549458,
-      "grad_norm": 0.005717565305531025,
-      "learning_rate": 0.00019998409700480347,
-      "loss": 46.0,
-      "step": 35262
-    },
-    {
-      "epoch": 5.678851805628246,
-      "grad_norm": 0.002729204948991537,
-      "learning_rate": 0.00019998409610256815,
-      "loss": 46.0,
-      "step": 35263
-    },
-    {
-      "epoch": 5.679012842707033,
-      "grad_norm": 0.008691055700182915,
-      "learning_rate": 0.00019998409520030727,
-      "loss": 46.0,
-      "step": 35264
-    },
-    {
-      "epoch": 5.679173879785821,
-      "grad_norm": 0.011858818121254444,
-      "learning_rate": 0.0001999840942980208,
-      "loss": 46.0,
-      "step": 35265
-    },
-    {
-      "epoch": 5.679334916864608,
-      "grad_norm": 0.008600740693509579,
-      "learning_rate": 0.00019998409339570876,
-      "loss": 46.0,
-      "step": 35266
-    },
-    {
-      "epoch": 5.679495953943396,
-      "grad_norm": 0.010730131529271603,
-      "learning_rate": 0.00019998409249337112,
-      "loss": 46.0,
-      "step": 35267
-    },
-    {
-      "epoch": 5.679656991022183,
-      "grad_norm": 0.001740789390169084,
-      "learning_rate": 0.00019998409159100786,
-      "loss": 46.0,
-      "step": 35268
-    },
-    {
-      "epoch": 5.6798180281009705,
-      "grad_norm": 0.003853529691696167,
-      "learning_rate": 0.00019998409068861902,
-      "loss": 46.0,
-      "step": 35269
-    },
-    {
-      "epoch": 5.679979065179758,
-      "grad_norm": 0.003495418466627598,
-      "learning_rate": 0.0001999840897862046,
-      "loss": 46.0,
-      "step": 35270
-    },
-    {
-      "epoch": 5.680140102258545,
-      "grad_norm": 0.0079538868740201,
-      "learning_rate": 0.00019998408888376458,
-      "loss": 46.0,
-      "step": 35271
-    },
-    {
-      "epoch": 5.680301139337333,
-      "grad_norm": 0.0051788100972771645,
-      "learning_rate": 0.000199984087981299,
-      "loss": 46.0,
-      "step": 35272
-    },
-    {
-      "epoch": 5.680462176416119,
-      "grad_norm": 0.010263034142553806,
-      "learning_rate": 0.0001999840870788078,
-      "loss": 46.0,
-      "step": 35273
-    },
-    {
-      "epoch": 5.680623213494907,
-      "grad_norm": 0.003974579274654388,
-      "learning_rate": 0.00019998408617629099,
-      "loss": 46.0,
-      "step": 35274
-    },
-    {
-      "epoch": 5.680784250573694,
-      "grad_norm": 0.003546706400811672,
-      "learning_rate": 0.00019998408527374862,
-      "loss": 46.0,
-      "step": 35275
-    },
-    {
-      "epoch": 5.680945287652482,
-      "grad_norm": 0.012054373510181904,
-      "learning_rate": 0.00019998408437118067,
-      "loss": 46.0,
-      "step": 35276
-    },
-    {
-      "epoch": 5.681106324731269,
-      "grad_norm": 0.004586960654705763,
-      "learning_rate": 0.0001999840834685871,
-      "loss": 46.0,
-      "step": 35277
-    },
-    {
-      "epoch": 5.681267361810057,
-      "grad_norm": 0.0035653235390782356,
-      "learning_rate": 0.00019998408256596793,
-      "loss": 46.0,
-      "step": 35278
-    },
-    {
-      "epoch": 5.681428398888844,
-      "grad_norm": 0.0019957139156758785,
-      "learning_rate": 0.00019998408166332321,
-      "loss": 46.0,
-      "step": 35279
-    },
-    {
-      "epoch": 5.6815894359676316,
-      "grad_norm": 0.007851950824260712,
-      "learning_rate": 0.00019998408076065286,
-      "loss": 46.0,
-      "step": 35280
-    },
-    {
-      "epoch": 5.681750473046419,
-      "grad_norm": 0.015349440276622772,
-      "learning_rate": 0.00019998407985795695,
-      "loss": 46.0,
-      "step": 35281
-    },
-    {
-      "epoch": 5.6819115101252065,
-      "grad_norm": 0.0078217051923275,
-      "learning_rate": 0.00019998407895523544,
-      "loss": 46.0,
-      "step": 35282
-    },
-    {
-      "epoch": 5.682072547203994,
-      "grad_norm": 0.002218423644080758,
-      "learning_rate": 0.00019998407805248833,
-      "loss": 46.0,
-      "step": 35283
-    },
-    {
-      "epoch": 5.682233584282781,
-      "grad_norm": 0.0015029021305963397,
-      "learning_rate": 0.00019998407714971563,
-      "loss": 46.0,
-      "step": 35284
-    },
-    {
-      "epoch": 5.682394621361569,
-      "grad_norm": 0.0012599461479112506,
-      "learning_rate": 0.00019998407624691736,
-      "loss": 46.0,
-      "step": 35285
-    },
-    {
-      "epoch": 5.682555658440356,
-      "grad_norm": 0.004044000990688801,
-      "learning_rate": 0.00019998407534409343,
-      "loss": 46.0,
-      "step": 35286
-    },
-    {
-      "epoch": 5.682716695519144,
-      "grad_norm": 0.002554306061938405,
-      "learning_rate": 0.000199984074441244,
-      "loss": 46.0,
-      "step": 35287
-    },
-    {
-      "epoch": 5.68287773259793,
-      "grad_norm": 0.004146645776927471,
-      "learning_rate": 0.00019998407353836895,
-      "loss": 46.0,
-      "step": 35288
-    },
-    {
-      "epoch": 5.683038769676718,
-      "grad_norm": 0.009110577404499054,
-      "learning_rate": 0.0001999840726354683,
-      "loss": 46.0,
-      "step": 35289
-    },
-    {
-      "epoch": 5.683199806755505,
-      "grad_norm": 0.016336863860487938,
-      "learning_rate": 0.00019998407173254203,
-      "loss": 46.0,
-      "step": 35290
-    },
-    {
-      "epoch": 5.683360843834293,
-      "grad_norm": 0.01600031740963459,
-      "learning_rate": 0.00019998407082959021,
-      "loss": 46.0,
-      "step": 35291
-    },
-    {
-      "epoch": 5.68352188091308,
-      "grad_norm": 0.0050653861835598946,
-      "learning_rate": 0.00019998406992661281,
-      "loss": 46.0,
-      "step": 35292
-    },
-    {
-      "epoch": 5.6836829179918675,
-      "grad_norm": 0.00461761187762022,
-      "learning_rate": 0.00019998406902360977,
-      "loss": 46.0,
-      "step": 35293
-    },
-    {
-      "epoch": 5.683843955070655,
-      "grad_norm": 0.013072383590042591,
-      "learning_rate": 0.00019998406812058117,
-      "loss": 46.0,
-      "step": 35294
-    },
-    {
-      "epoch": 5.684004992149442,
-      "grad_norm": 0.0012824031291529536,
-      "learning_rate": 0.00019998406721752698,
-      "loss": 46.0,
-      "step": 35295
-    },
-    {
-      "epoch": 5.68416602922823,
-      "grad_norm": 0.00237314123660326,
-      "learning_rate": 0.00019998406631444718,
-      "loss": 46.0,
-      "step": 35296
-    },
-    {
-      "epoch": 5.684327066307017,
-      "grad_norm": 0.0017943476559594274,
-      "learning_rate": 0.0001999840654113418,
-      "loss": 46.0,
-      "step": 35297
-    },
-    {
-      "epoch": 5.684488103385805,
-      "grad_norm": 0.011635672301054,
-      "learning_rate": 0.00019998406450821087,
-      "loss": 46.0,
-      "step": 35298
-    },
-    {
-      "epoch": 5.684649140464592,
-      "grad_norm": 0.01504228450357914,
-      "learning_rate": 0.00019998406360505428,
-      "loss": 46.0,
-      "step": 35299
-    },
-    {
-      "epoch": 5.68481017754338,
-      "grad_norm": 0.003236522199586034,
-      "learning_rate": 0.00019998406270187212,
-      "loss": 46.0,
-      "step": 35300
-    },
-    {
-      "epoch": 5.684971214622166,
-      "grad_norm": 0.0021517241839319468,
-      "learning_rate": 0.0001999840617986644,
-      "loss": 46.0,
-      "step": 35301
-    },
-    {
-      "epoch": 5.685132251700955,
-      "grad_norm": 0.0018133294070139527,
-      "learning_rate": 0.0001999840608954311,
-      "loss": 46.0,
-      "step": 35302
-    },
-    {
-      "epoch": 5.685293288779741,
-      "grad_norm": 0.00377635401673615,
-      "learning_rate": 0.00019998405999217215,
-      "loss": 46.0,
-      "step": 35303
-    },
-    {
-      "epoch": 5.685454325858529,
-      "grad_norm": 0.002605409361422062,
-      "learning_rate": 0.00019998405908888765,
-      "loss": 46.0,
-      "step": 35304
-    },
-    {
-      "epoch": 5.685615362937316,
-      "grad_norm": 0.0031169280409812927,
-      "learning_rate": 0.00019998405818557756,
-      "loss": 46.0,
-      "step": 35305
-    },
-    {
-      "epoch": 5.6857764000161035,
-      "grad_norm": 0.0050070141442120075,
-      "learning_rate": 0.00019998405728224183,
-      "loss": 46.0,
-      "step": 35306
-    },
-    {
-      "epoch": 5.685937437094891,
-      "grad_norm": 0.005279757548123598,
-      "learning_rate": 0.00019998405637888057,
-      "loss": 46.0,
-      "step": 35307
-    },
-    {
-      "epoch": 5.686098474173678,
-      "grad_norm": 0.004055046942085028,
-      "learning_rate": 0.0001999840554754937,
-      "loss": 46.0,
-      "step": 35308
-    },
-    {
-      "epoch": 5.686259511252466,
-      "grad_norm": 0.002828560769557953,
-      "learning_rate": 0.00019998405457208123,
-      "loss": 46.0,
-      "step": 35309
-    },
-    {
-      "epoch": 5.686420548331253,
-      "grad_norm": 0.0008795047178864479,
-      "learning_rate": 0.00019998405366864318,
-      "loss": 46.0,
-      "step": 35310
-    },
-    {
-      "epoch": 5.686581585410041,
-      "grad_norm": 0.005033840425312519,
-      "learning_rate": 0.00019998405276517952,
-      "loss": 46.0,
-      "step": 35311
-    },
-    {
-      "epoch": 5.686742622488828,
-      "grad_norm": 0.003187753725796938,
-      "learning_rate": 0.00019998405186169026,
-      "loss": 46.0,
-      "step": 35312
-    },
-    {
-      "epoch": 5.686903659567616,
-      "grad_norm": 0.006427199114114046,
-      "learning_rate": 0.00019998405095817545,
-      "loss": 46.0,
-      "step": 35313
-    },
-    {
-      "epoch": 5.687064696646403,
-      "grad_norm": 0.0017722093034535646,
-      "learning_rate": 0.00019998405005463503,
-      "loss": 46.0,
-      "step": 35314
-    },
-    {
-      "epoch": 5.687225733725191,
-      "grad_norm": 0.015021676197648048,
-      "learning_rate": 0.00019998404915106904,
-      "loss": 46.0,
-      "step": 35315
-    },
-    {
-      "epoch": 5.687386770803977,
-      "grad_norm": 0.00340224988758564,
-      "learning_rate": 0.00019998404824747744,
-      "loss": 46.0,
-      "step": 35316
-    },
-    {
-      "epoch": 5.6875478078827655,
-      "grad_norm": 0.0029090321622788906,
-      "learning_rate": 0.00019998404734386022,
-      "loss": 46.0,
-      "step": 35317
-    },
-    {
-      "epoch": 5.687708844961552,
-      "grad_norm": 0.010697041638195515,
-      "learning_rate": 0.00019998404644021745,
-      "loss": 46.0,
-      "step": 35318
-    },
-    {
-      "epoch": 5.6878698820403395,
-      "grad_norm": 0.0056275296956300735,
-      "learning_rate": 0.00019998404553654906,
-      "loss": 46.0,
-      "step": 35319
-    },
-    {
-      "epoch": 5.688030919119127,
-      "grad_norm": 0.002728802850469947,
-      "learning_rate": 0.00019998404463285508,
-      "loss": 46.0,
-      "step": 35320
-    },
-    {
-      "epoch": 5.688191956197914,
-      "grad_norm": 0.007858066819608212,
-      "learning_rate": 0.00019998404372913555,
-      "loss": 46.0,
-      "step": 35321
-    },
-    {
-      "epoch": 5.688352993276702,
-      "grad_norm": 0.0014615998370572925,
-      "learning_rate": 0.0001999840428253904,
-      "loss": 46.0,
-      "step": 35322
-    },
-    {
-      "epoch": 5.688514030355489,
-      "grad_norm": 0.009267511777579784,
-      "learning_rate": 0.0001999840419216197,
-      "loss": 46.0,
-      "step": 35323
-    },
-    {
-      "epoch": 5.688675067434277,
-      "grad_norm": 0.0037768869660794735,
-      "learning_rate": 0.00019998404101782336,
-      "loss": 46.0,
-      "step": 35324
-    },
-    {
-      "epoch": 5.688836104513064,
-      "grad_norm": 0.0030597448348999023,
-      "learning_rate": 0.00019998404011400142,
-      "loss": 46.0,
-      "step": 35325
-    },
-    {
-      "epoch": 5.688997141591852,
-      "grad_norm": 0.010552025400102139,
-      "learning_rate": 0.00019998403921015392,
-      "loss": 46.0,
-      "step": 35326
-    },
-    {
-      "epoch": 5.689158178670639,
-      "grad_norm": 0.0018648904515430331,
-      "learning_rate": 0.0001999840383062808,
-      "loss": 46.0,
-      "step": 35327
-    },
-    {
-      "epoch": 5.689319215749427,
-      "grad_norm": 0.0015851183561608195,
-      "learning_rate": 0.0001999840374023821,
-      "loss": 46.0,
-      "step": 35328
-    },
-    {
-      "epoch": 5.689480252828214,
-      "grad_norm": 0.004243799950927496,
-      "learning_rate": 0.00019998403649845785,
-      "loss": 46.0,
-      "step": 35329
-    },
-    {
-      "epoch": 5.6896412899070015,
-      "grad_norm": 0.006032241974025965,
-      "learning_rate": 0.00019998403559450795,
-      "loss": 46.0,
-      "step": 35330
-    },
-    {
-      "epoch": 5.689802326985788,
-      "grad_norm": 0.0018145754002034664,
-      "learning_rate": 0.0001999840346905325,
-      "loss": 46.0,
-      "step": 35331
-    },
-    {
-      "epoch": 5.689963364064576,
-      "grad_norm": 0.0034648312721401453,
-      "learning_rate": 0.00019998403378653146,
-      "loss": 46.0,
-      "step": 35332
-    },
-    {
-      "epoch": 5.690124401143363,
-      "grad_norm": 0.0028540475759655237,
-      "learning_rate": 0.0001999840328825048,
-      "loss": 46.0,
-      "step": 35333
-    },
-    {
-      "epoch": 5.69028543822215,
-      "grad_norm": 0.022783741354942322,
-      "learning_rate": 0.0001999840319784526,
-      "loss": 46.0,
-      "step": 35334
-    },
-    {
-      "epoch": 5.690446475300938,
-      "grad_norm": 0.0026501058600842953,
-      "learning_rate": 0.00019998403107437476,
-      "loss": 46.0,
-      "step": 35335
-    },
-    {
-      "epoch": 5.690607512379725,
-      "grad_norm": 0.005448977928608656,
-      "learning_rate": 0.00019998403017027134,
-      "loss": 46.0,
-      "step": 35336
-    },
-    {
-      "epoch": 5.690768549458513,
-      "grad_norm": 0.000529517128597945,
-      "learning_rate": 0.00019998402926614233,
-      "loss": 46.0,
-      "step": 35337
-    },
-    {
-      "epoch": 5.6909295865373,
-      "grad_norm": 0.0025161313824355602,
-      "learning_rate": 0.00019998402836198773,
-      "loss": 46.0,
-      "step": 35338
-    },
-    {
-      "epoch": 5.691090623616088,
-      "grad_norm": 0.011775963939726353,
-      "learning_rate": 0.00019998402745780754,
-      "loss": 46.0,
-      "step": 35339
-    },
-    {
-      "epoch": 5.691251660694875,
-      "grad_norm": 0.017960337921977043,
-      "learning_rate": 0.00019998402655360177,
-      "loss": 46.0,
-      "step": 35340
-    },
-    {
-      "epoch": 5.6914126977736625,
-      "grad_norm": 0.0029451975133270025,
-      "learning_rate": 0.0001999840256493704,
-      "loss": 46.0,
-      "step": 35341
-    },
-    {
-      "epoch": 5.69157373485245,
-      "grad_norm": 0.003964627161622047,
-      "learning_rate": 0.00019998402474511343,
-      "loss": 46.0,
-      "step": 35342
-    },
-    {
-      "epoch": 5.6917347719312374,
-      "grad_norm": 0.0017763259820640087,
-      "learning_rate": 0.00019998402384083087,
-      "loss": 46.0,
-      "step": 35343
-    },
-    {
-      "epoch": 5.691895809010025,
-      "grad_norm": 0.005671083461493254,
-      "learning_rate": 0.00019998402293652272,
-      "loss": 46.0,
-      "step": 35344
-    },
-    {
-      "epoch": 5.692056846088812,
-      "grad_norm": 0.012492327019572258,
-      "learning_rate": 0.000199984022032189,
-      "loss": 46.0,
-      "step": 35345
-    },
-    {
-      "epoch": 5.692217883167599,
-      "grad_norm": 0.007014209404587746,
-      "learning_rate": 0.00019998402112782966,
-      "loss": 46.0,
-      "step": 35346
-    },
-    {
-      "epoch": 5.692378920246386,
-      "grad_norm": 0.010842365212738514,
-      "learning_rate": 0.00019998402022344475,
-      "loss": 46.0,
-      "step": 35347
-    },
-    {
-      "epoch": 5.692539957325174,
-      "grad_norm": 0.023855527862906456,
-      "learning_rate": 0.00019998401931903425,
-      "loss": 46.0,
-      "step": 35348
-    },
-    {
-      "epoch": 5.692700994403961,
-      "grad_norm": 0.0020728951785713434,
-      "learning_rate": 0.00019998401841459816,
-      "loss": 46.0,
-      "step": 35349
-    },
-    {
-      "epoch": 5.692862031482749,
-      "grad_norm": 0.0015600932529196143,
-      "learning_rate": 0.00019998401751013646,
-      "loss": 46.0,
-      "step": 35350
-    },
-    {
-      "epoch": 5.693023068561536,
-      "grad_norm": 0.00148162676487118,
-      "learning_rate": 0.0001999840166056492,
-      "loss": 46.0,
-      "step": 35351
-    },
-    {
-      "epoch": 5.693184105640324,
-      "grad_norm": 0.0023797908797860146,
-      "learning_rate": 0.00019998401570113633,
-      "loss": 46.0,
-      "step": 35352
-    },
-    {
-      "epoch": 5.693345142719111,
-      "grad_norm": 0.008714855648577213,
-      "learning_rate": 0.00019998401479659784,
-      "loss": 46.0,
-      "step": 35353
-    },
-    {
-      "epoch": 5.6935061797978985,
-      "grad_norm": 0.0029996950179338455,
-      "learning_rate": 0.0001999840138920338,
-      "loss": 46.0,
-      "step": 35354
-    },
-    {
-      "epoch": 5.693667216876686,
-      "grad_norm": 0.0018862314755097032,
-      "learning_rate": 0.00019998401298744418,
-      "loss": 46.0,
-      "step": 35355
-    },
-    {
-      "epoch": 5.693828253955473,
-      "grad_norm": 0.004545945208519697,
-      "learning_rate": 0.00019998401208282893,
-      "loss": 46.0,
-      "step": 35356
-    },
-    {
-      "epoch": 5.693989291034261,
-      "grad_norm": 0.008998136967420578,
-      "learning_rate": 0.0001999840111781881,
-      "loss": 46.0,
-      "step": 35357
-    },
-    {
-      "epoch": 5.694150328113048,
-      "grad_norm": 0.0057633561082184315,
-      "learning_rate": 0.0001999840102735217,
-      "loss": 46.0,
-      "step": 35358
-    },
-    {
-      "epoch": 5.694311365191836,
-      "grad_norm": 0.006195182912051678,
-      "learning_rate": 0.00019998400936882972,
-      "loss": 46.0,
-      "step": 35359
-    },
-    {
-      "epoch": 5.694472402270623,
-      "grad_norm": 0.002205311320722103,
-      "learning_rate": 0.0001999840084641121,
-      "loss": 46.0,
-      "step": 35360
-    },
-    {
-      "epoch": 5.69463343934941,
-      "grad_norm": 0.007667542900890112,
-      "learning_rate": 0.0001999840075593689,
-      "loss": 46.0,
-      "step": 35361
-    },
-    {
-      "epoch": 5.694794476428197,
-      "grad_norm": 0.008176380768418312,
-      "learning_rate": 0.00019998400665460013,
-      "loss": 46.0,
-      "step": 35362
-    },
-    {
-      "epoch": 5.694955513506985,
-      "grad_norm": 0.00937776267528534,
-      "learning_rate": 0.00019998400574980577,
-      "loss": 46.0,
-      "step": 35363
-    },
-    {
-      "epoch": 5.695116550585772,
-      "grad_norm": 0.0012654049787670374,
-      "learning_rate": 0.0001999840048449858,
-      "loss": 46.0,
-      "step": 35364
-    },
-    {
-      "epoch": 5.69527758766456,
-      "grad_norm": 0.003352086991071701,
-      "learning_rate": 0.00019998400394014026,
-      "loss": 46.0,
-      "step": 35365
-    },
-    {
-      "epoch": 5.695438624743347,
-      "grad_norm": 0.003260578028857708,
-      "learning_rate": 0.0001999840030352691,
-      "loss": 46.0,
-      "step": 35366
-    },
-    {
-      "epoch": 5.6955996618221345,
-      "grad_norm": 0.0036888134200125933,
-      "learning_rate": 0.00019998400213037238,
-      "loss": 46.0,
-      "step": 35367
-    },
-    {
-      "epoch": 5.695760698900922,
-      "grad_norm": 0.00892405305057764,
-      "learning_rate": 0.00019998400122545008,
-      "loss": 46.0,
-      "step": 35368
-    },
-    {
-      "epoch": 5.695921735979709,
-      "grad_norm": 0.003858861979097128,
-      "learning_rate": 0.00019998400032050214,
-      "loss": 46.0,
-      "step": 35369
-    },
-    {
-      "epoch": 5.696082773058497,
-      "grad_norm": 0.004568117670714855,
-      "learning_rate": 0.00019998399941552865,
-      "loss": 46.0,
-      "step": 35370
-    },
-    {
-      "epoch": 5.696243810137284,
-      "grad_norm": 0.008835841901600361,
-      "learning_rate": 0.00019998399851052956,
-      "loss": 46.0,
-      "step": 35371
-    },
-    {
-      "epoch": 5.696404847216072,
-      "grad_norm": 0.012812651693820953,
-      "learning_rate": 0.00019998399760550486,
-      "loss": 46.0,
-      "step": 35372
-    },
-    {
-      "epoch": 5.696565884294859,
-      "grad_norm": 0.002240388421341777,
-      "learning_rate": 0.0001999839967004546,
-      "loss": 46.0,
-      "step": 35373
-    },
-    {
-      "epoch": 5.696726921373647,
-      "grad_norm": 0.00512683903798461,
-      "learning_rate": 0.00019998399579537873,
-      "loss": 46.0,
-      "step": 35374
-    },
-    {
-      "epoch": 5.696887958452434,
-      "grad_norm": 0.005152825731784105,
-      "learning_rate": 0.00019998399489027727,
-      "loss": 46.0,
-      "step": 35375
-    },
-    {
-      "epoch": 5.697048995531221,
-      "grad_norm": 0.0022141211666166782,
-      "learning_rate": 0.00019998399398515022,
-      "loss": 46.0,
-      "step": 35376
-    },
-    {
-      "epoch": 5.697210032610008,
-      "grad_norm": 0.0014364334056153893,
-      "learning_rate": 0.0001999839930799976,
-      "loss": 46.0,
-      "step": 35377
-    },
-    {
-      "epoch": 5.697371069688796,
-      "grad_norm": 0.001976205036044121,
-      "learning_rate": 0.00019998399217481937,
-      "loss": 46.0,
-      "step": 35378
-    },
-    {
-      "epoch": 5.697532106767583,
-      "grad_norm": 0.0047792126424610615,
-      "learning_rate": 0.00019998399126961553,
-      "loss": 46.0,
-      "step": 35379
-    },
-    {
-      "epoch": 5.6976931438463705,
-      "grad_norm": 0.0037175470497459173,
-      "learning_rate": 0.00019998399036438613,
-      "loss": 46.0,
-      "step": 35380
-    },
-    {
-      "epoch": 5.697854180925158,
-      "grad_norm": 0.007630035746842623,
-      "learning_rate": 0.00019998398945913112,
-      "loss": 46.0,
-      "step": 35381
-    },
-    {
-      "epoch": 5.698015218003945,
-      "grad_norm": 0.0025430191308259964,
-      "learning_rate": 0.00019998398855385055,
-      "loss": 46.0,
-      "step": 35382
-    },
-    {
-      "epoch": 5.698176255082733,
-      "grad_norm": 0.001435289392247796,
-      "learning_rate": 0.00019998398764854437,
-      "loss": 46.0,
-      "step": 35383
-    },
-    {
-      "epoch": 5.69833729216152,
-      "grad_norm": 0.0018268882995471358,
-      "learning_rate": 0.00019998398674321257,
-      "loss": 46.0,
-      "step": 35384
-    },
-    {
-      "epoch": 5.698498329240308,
-      "grad_norm": 0.0006946501671336591,
-      "learning_rate": 0.0001999839858378552,
-      "loss": 46.0,
-      "step": 35385
-    },
-    {
-      "epoch": 5.698659366319095,
-      "grad_norm": 0.010200402699410915,
-      "learning_rate": 0.00019998398493247226,
-      "loss": 46.0,
-      "step": 35386
-    },
-    {
-      "epoch": 5.698820403397883,
-      "grad_norm": 0.0021359389647841454,
-      "learning_rate": 0.0001999839840270637,
-      "loss": 46.0,
-      "step": 35387
-    },
-    {
-      "epoch": 5.69898144047667,
-      "grad_norm": 0.008337379433214664,
-      "learning_rate": 0.00019998398312162956,
-      "loss": 46.0,
-      "step": 35388
-    },
-    {
-      "epoch": 5.699142477555457,
-      "grad_norm": 0.0247037410736084,
-      "learning_rate": 0.00019998398221616985,
-      "loss": 46.0,
-      "step": 35389
-    },
-    {
-      "epoch": 5.699303514634245,
-      "grad_norm": 0.00543738529086113,
-      "learning_rate": 0.00019998398131068453,
-      "loss": 46.0,
-      "step": 35390
-    },
-    {
-      "epoch": 5.699464551713032,
-      "grad_norm": 0.0012550480896607041,
-      "learning_rate": 0.00019998398040517362,
-      "loss": 46.0,
-      "step": 35391
-    },
-    {
-      "epoch": 5.699625588791819,
-      "grad_norm": 0.005019910167902708,
-      "learning_rate": 0.00019998397949963712,
-      "loss": 46.0,
-      "step": 35392
-    },
-    {
-      "epoch": 5.6997866258706065,
-      "grad_norm": 0.019118789583444595,
-      "learning_rate": 0.000199983978594075,
-      "loss": 46.0,
-      "step": 35393
-    },
-    {
-      "epoch": 5.699947662949394,
-      "grad_norm": 0.0034586978144943714,
-      "learning_rate": 0.00019998397768848737,
-      "loss": 46.0,
-      "step": 35394
-    },
-    {
-      "epoch": 5.700108700028181,
-      "grad_norm": 0.00217807712033391,
-      "learning_rate": 0.00019998397678287408,
-      "loss": 46.0,
-      "step": 35395
-    },
-    {
-      "epoch": 5.700269737106969,
-      "grad_norm": 0.0029005524702370167,
-      "learning_rate": 0.0001999839758772352,
-      "loss": 46.0,
-      "step": 35396
-    },
-    {
-      "epoch": 5.700430774185756,
-      "grad_norm": 0.0030601024627685547,
-      "learning_rate": 0.00019998397497157075,
-      "loss": 46.0,
-      "step": 35397
-    },
-    {
-      "epoch": 5.700591811264544,
-      "grad_norm": 0.002155889756977558,
-      "learning_rate": 0.0001999839740658807,
-      "loss": 46.0,
-      "step": 35398
-    },
-    {
-      "epoch": 5.700752848343331,
-      "grad_norm": 0.0024871493224054575,
-      "learning_rate": 0.00019998397316016507,
-      "loss": 46.0,
-      "step": 35399
-    },
-    {
-      "epoch": 5.700913885422119,
-      "grad_norm": 0.001718775718472898,
-      "learning_rate": 0.00019998397225442385,
-      "loss": 46.0,
-      "step": 35400
-    },
-    {
-      "epoch": 5.701074922500906,
-      "grad_norm": 0.0031118469778448343,
-      "learning_rate": 0.00019998397134865704,
-      "loss": 46.0,
-      "step": 35401
-    },
-    {
-      "epoch": 5.7012359595796935,
-      "grad_norm": 0.013429944403469563,
-      "learning_rate": 0.00019998397044286464,
-      "loss": 46.0,
-      "step": 35402
-    },
-    {
-      "epoch": 5.701396996658481,
-      "grad_norm": 0.005788861308246851,
-      "learning_rate": 0.00019998396953704663,
-      "loss": 46.0,
-      "step": 35403
-    },
-    {
-      "epoch": 5.7015580337372676,
-      "grad_norm": 0.002211272483691573,
-      "learning_rate": 0.00019998396863120304,
-      "loss": 46.0,
-      "step": 35404
-    },
-    {
-      "epoch": 5.701719070816056,
-      "grad_norm": 0.00576765276491642,
-      "learning_rate": 0.00019998396772533385,
-      "loss": 46.0,
-      "step": 35405
-    },
-    {
-      "epoch": 5.7018801078948425,
-      "grad_norm": 0.0010375359561294317,
-      "learning_rate": 0.00019998396681943908,
-      "loss": 46.0,
-      "step": 35406
-    },
-    {
-      "epoch": 5.70204114497363,
-      "grad_norm": 0.00444755470380187,
-      "learning_rate": 0.00019998396591351872,
-      "loss": 46.0,
-      "step": 35407
-    },
-    {
-      "epoch": 5.702202182052417,
-      "grad_norm": 0.008206761442124844,
-      "learning_rate": 0.00019998396500757278,
-      "loss": 46.0,
-      "step": 35408
-    },
-    {
-      "epoch": 5.702363219131205,
-      "grad_norm": 0.002685726620256901,
-      "learning_rate": 0.00019998396410160122,
-      "loss": 46.0,
-      "step": 35409
-    },
-    {
-      "epoch": 5.702524256209992,
-      "grad_norm": 0.0029550758190453053,
-      "learning_rate": 0.00019998396319560407,
-      "loss": 46.0,
-      "step": 35410
-    },
-    {
-      "epoch": 5.70268529328878,
-      "grad_norm": 0.00653796223923564,
-      "learning_rate": 0.00019998396228958137,
-      "loss": 46.0,
-      "step": 35411
-    },
-    {
-      "epoch": 5.702846330367567,
-      "grad_norm": 0.007673420011997223,
-      "learning_rate": 0.00019998396138353304,
-      "loss": 46.0,
-      "step": 35412
-    },
-    {
-      "epoch": 5.703007367446355,
-      "grad_norm": 0.008077666163444519,
-      "learning_rate": 0.00019998396047745916,
-      "loss": 46.0,
-      "step": 35413
-    },
-    {
-      "epoch": 5.703168404525142,
-      "grad_norm": 0.019116917625069618,
-      "learning_rate": 0.00019998395957135964,
-      "loss": 46.0,
-      "step": 35414
-    },
-    {
-      "epoch": 5.7033294416039295,
-      "grad_norm": 0.006062595173716545,
-      "learning_rate": 0.00019998395866523456,
-      "loss": 46.0,
-      "step": 35415
-    },
-    {
-      "epoch": 5.703490478682717,
-      "grad_norm": 0.007338200695812702,
-      "learning_rate": 0.00019998395775908386,
-      "loss": 46.0,
-      "step": 35416
-    },
-    {
-      "epoch": 5.703651515761504,
-      "grad_norm": 0.003945605829358101,
-      "learning_rate": 0.0001999839568529076,
-      "loss": 46.0,
-      "step": 35417
-    },
-    {
-      "epoch": 5.703812552840292,
-      "grad_norm": 0.004946451168507338,
-      "learning_rate": 0.00019998395594670573,
-      "loss": 46.0,
-      "step": 35418
-    },
-    {
-      "epoch": 5.703973589919078,
-      "grad_norm": 0.0020150833297520876,
-      "learning_rate": 0.0001999839550404783,
-      "loss": 46.0,
-      "step": 35419
-    },
-    {
-      "epoch": 5.704134626997866,
-      "grad_norm": 0.012988763861358166,
-      "learning_rate": 0.00019998395413422525,
-      "loss": 46.0,
-      "step": 35420
-    },
-    {
-      "epoch": 5.704295664076653,
-      "grad_norm": 0.004056087229400873,
-      "learning_rate": 0.0001999839532279466,
-      "loss": 46.0,
-      "step": 35421
-    },
-    {
-      "epoch": 5.704456701155441,
-      "grad_norm": 0.00551618542522192,
-      "learning_rate": 0.0001999839523216424,
-      "loss": 46.0,
-      "step": 35422
-    },
-    {
-      "epoch": 5.704617738234228,
-      "grad_norm": 0.0016281232237815857,
-      "learning_rate": 0.0001999839514153126,
-      "loss": 46.0,
-      "step": 35423
-    },
-    {
-      "epoch": 5.704778775313016,
-      "grad_norm": 0.0025894027203321457,
-      "learning_rate": 0.00019998395050895717,
-      "loss": 46.0,
-      "step": 35424
-    },
-    {
-      "epoch": 5.704939812391803,
-      "grad_norm": 0.0021113725379109383,
-      "learning_rate": 0.0001999839496025762,
-      "loss": 46.0,
-      "step": 35425
-    },
-    {
-      "epoch": 5.705100849470591,
-      "grad_norm": 0.013794437982141972,
-      "learning_rate": 0.00019998394869616962,
-      "loss": 46.0,
-      "step": 35426
-    },
-    {
-      "epoch": 5.705261886549378,
-      "grad_norm": 0.008361543528735638,
-      "learning_rate": 0.00019998394778973744,
-      "loss": 46.0,
-      "step": 35427
-    },
-    {
-      "epoch": 5.7054229236281655,
-      "grad_norm": 0.005546377040445805,
-      "learning_rate": 0.00019998394688327964,
-      "loss": 46.0,
-      "step": 35428
-    },
-    {
-      "epoch": 5.705583960706953,
-      "grad_norm": 0.003545439802110195,
-      "learning_rate": 0.0001999839459767963,
-      "loss": 46.0,
-      "step": 35429
-    },
-    {
-      "epoch": 5.70574499778574,
-      "grad_norm": 0.0045602587051689625,
-      "learning_rate": 0.00019998394507028734,
-      "loss": 46.0,
-      "step": 35430
-    },
-    {
-      "epoch": 5.705906034864528,
-      "grad_norm": 0.0029252043459564447,
-      "learning_rate": 0.0001999839441637528,
-      "loss": 46.0,
-      "step": 35431
-    },
-    {
-      "epoch": 5.706067071943315,
-      "grad_norm": 0.0065379999577999115,
-      "learning_rate": 0.00019998394325719266,
-      "loss": 46.0,
-      "step": 35432
-    },
-    {
-      "epoch": 5.706228109022103,
-      "grad_norm": 0.0016173160402104259,
-      "learning_rate": 0.00019998394235060696,
-      "loss": 46.0,
-      "step": 35433
-    },
-    {
-      "epoch": 5.706389146100889,
-      "grad_norm": 0.02710697054862976,
-      "learning_rate": 0.00019998394144399564,
-      "loss": 46.0,
-      "step": 35434
-    },
-    {
-      "epoch": 5.706550183179677,
-      "grad_norm": 0.008578195236623287,
-      "learning_rate": 0.00019998394053735873,
-      "loss": 46.0,
-      "step": 35435
-    },
-    {
-      "epoch": 5.706711220258464,
-      "grad_norm": 0.013228898867964745,
-      "learning_rate": 0.00019998393963069626,
-      "loss": 46.0,
-      "step": 35436
-    },
-    {
-      "epoch": 5.706872257337252,
-      "grad_norm": 0.003564579878002405,
-      "learning_rate": 0.00019998393872400815,
-      "loss": 46.0,
-      "step": 35437
-    },
-    {
-      "epoch": 5.707033294416039,
-      "grad_norm": 0.005520430859178305,
-      "learning_rate": 0.00019998393781729448,
-      "loss": 46.0,
-      "step": 35438
-    },
-    {
-      "epoch": 5.707194331494827,
-      "grad_norm": 0.004034971818327904,
-      "learning_rate": 0.00019998393691055523,
-      "loss": 46.0,
-      "step": 35439
-    },
-    {
-      "epoch": 5.707355368573614,
-      "grad_norm": 0.0052180602215230465,
-      "learning_rate": 0.00019998393600379038,
-      "loss": 46.0,
-      "step": 35440
-    },
-    {
-      "epoch": 5.7075164056524015,
-      "grad_norm": 0.0028647335711866617,
-      "learning_rate": 0.0001999839350969999,
-      "loss": 46.0,
-      "step": 35441
-    },
-    {
-      "epoch": 5.707677442731189,
-      "grad_norm": 0.00260867178440094,
-      "learning_rate": 0.00019998393419018385,
-      "loss": 46.0,
-      "step": 35442
-    },
-    {
-      "epoch": 5.707838479809976,
-      "grad_norm": 0.01264884416013956,
-      "learning_rate": 0.00019998393328334225,
-      "loss": 46.0,
-      "step": 35443
-    },
-    {
-      "epoch": 5.707999516888764,
-      "grad_norm": 0.0020286054350435734,
-      "learning_rate": 0.00019998393237647503,
-      "loss": 46.0,
-      "step": 35444
-    },
-    {
-      "epoch": 5.708160553967551,
-      "grad_norm": 0.005394205451011658,
-      "learning_rate": 0.0001999839314695822,
-      "loss": 46.0,
-      "step": 35445
-    },
-    {
-      "epoch": 5.708321591046339,
-      "grad_norm": 0.002989766653627157,
-      "learning_rate": 0.0001999839305626638,
-      "loss": 46.0,
-      "step": 35446
-    },
-    {
-      "epoch": 5.708482628125126,
-      "grad_norm": 0.0017508756136521697,
-      "learning_rate": 0.00019998392965571982,
-      "loss": 46.0,
-      "step": 35447
-    },
-    {
-      "epoch": 5.708643665203914,
-      "grad_norm": 0.010966694913804531,
-      "learning_rate": 0.00019998392874875026,
-      "loss": 46.0,
-      "step": 35448
-    },
-    {
-      "epoch": 5.7088047022827,
-      "grad_norm": 0.0068923430517315865,
-      "learning_rate": 0.00019998392784175505,
-      "loss": 46.0,
-      "step": 35449
-    },
-    {
-      "epoch": 5.708965739361488,
-      "grad_norm": 0.0124368192628026,
-      "learning_rate": 0.00019998392693473428,
-      "loss": 46.0,
-      "step": 35450
-    },
-    {
-      "epoch": 5.709126776440275,
-      "grad_norm": 0.01080751046538353,
-      "learning_rate": 0.00019998392602768795,
-      "loss": 46.0,
-      "step": 35451
-    },
-    {
-      "epoch": 5.709287813519063,
-      "grad_norm": 0.0024276564363390207,
-      "learning_rate": 0.000199983925120616,
-      "loss": 46.0,
-      "step": 35452
-    },
-    {
-      "epoch": 5.70944885059785,
-      "grad_norm": 0.0005492082564160228,
-      "learning_rate": 0.00019998392421351845,
-      "loss": 46.0,
-      "step": 35453
-    },
-    {
-      "epoch": 5.7096098876766375,
-      "grad_norm": 0.003871447639539838,
-      "learning_rate": 0.00019998392330639533,
-      "loss": 46.0,
-      "step": 35454
-    },
-    {
-      "epoch": 5.709770924755425,
-      "grad_norm": 0.004460901487618685,
-      "learning_rate": 0.00019998392239924662,
-      "loss": 46.0,
-      "step": 35455
-    },
-    {
-      "epoch": 5.709931961834212,
-      "grad_norm": 0.0016366359777748585,
-      "learning_rate": 0.00019998392149207233,
-      "loss": 46.0,
-      "step": 35456
-    },
-    {
-      "epoch": 5.710092998913,
-      "grad_norm": 0.0019701600540429354,
-      "learning_rate": 0.0001999839205848724,
-      "loss": 46.0,
-      "step": 35457
-    },
-    {
-      "epoch": 5.710254035991787,
-      "grad_norm": 0.003288273001089692,
-      "learning_rate": 0.0001999839196776469,
-      "loss": 46.0,
-      "step": 35458
-    },
-    {
-      "epoch": 5.710415073070575,
-      "grad_norm": 0.010506383143365383,
-      "learning_rate": 0.00019998391877039585,
-      "loss": 46.0,
-      "step": 35459
-    },
-    {
-      "epoch": 5.710576110149362,
-      "grad_norm": 0.011146649718284607,
-      "learning_rate": 0.00019998391786311915,
-      "loss": 46.0,
-      "step": 35460
-    },
-    {
-      "epoch": 5.71073714722815,
-      "grad_norm": 0.008965682238340378,
-      "learning_rate": 0.0001999839169558169,
-      "loss": 46.0,
-      "step": 35461
-    },
-    {
-      "epoch": 5.710898184306936,
-      "grad_norm": 0.0030571077950298786,
-      "learning_rate": 0.00019998391604848906,
-      "loss": 46.0,
-      "step": 35462
-    },
-    {
-      "epoch": 5.7110592213857245,
-      "grad_norm": 0.0015387525781989098,
-      "learning_rate": 0.00019998391514113563,
-      "loss": 46.0,
-      "step": 35463
-    },
-    {
-      "epoch": 5.711220258464511,
-      "grad_norm": 0.003867823863402009,
-      "learning_rate": 0.00019998391423375658,
-      "loss": 46.0,
-      "step": 35464
-    },
-    {
-      "epoch": 5.7113812955432985,
-      "grad_norm": 0.004653410520404577,
-      "learning_rate": 0.00019998391332635195,
-      "loss": 46.0,
-      "step": 35465
-    },
-    {
-      "epoch": 5.711542332622086,
-      "grad_norm": 0.0009822207503020763,
-      "learning_rate": 0.00019998391241892173,
-      "loss": 46.0,
-      "step": 35466
-    },
-    {
-      "epoch": 5.7117033697008734,
-      "grad_norm": 0.007330250460654497,
-      "learning_rate": 0.00019998391151146593,
-      "loss": 46.0,
-      "step": 35467
-    },
-    {
-      "epoch": 5.711864406779661,
-      "grad_norm": 0.003616988891735673,
-      "learning_rate": 0.0001999839106039845,
-      "loss": 46.0,
-      "step": 35468
-    },
-    {
-      "epoch": 5.712025443858448,
-      "grad_norm": 0.0022851997055113316,
-      "learning_rate": 0.00019998390969647753,
-      "loss": 46.0,
-      "step": 35469
-    },
-    {
-      "epoch": 5.712186480937236,
-      "grad_norm": 0.010202893987298012,
-      "learning_rate": 0.00019998390878894496,
-      "loss": 46.0,
-      "step": 35470
-    },
-    {
-      "epoch": 5.712347518016023,
-      "grad_norm": 0.016665631905198097,
-      "learning_rate": 0.00019998390788138678,
-      "loss": 46.0,
-      "step": 35471
-    },
-    {
-      "epoch": 5.712508555094811,
-      "grad_norm": 0.014059074223041534,
-      "learning_rate": 0.000199983906973803,
-      "loss": 46.0,
-      "step": 35472
-    },
-    {
-      "epoch": 5.712669592173598,
-      "grad_norm": 0.006730437278747559,
-      "learning_rate": 0.00019998390606619366,
-      "loss": 46.0,
-      "step": 35473
-    },
-    {
-      "epoch": 5.712830629252386,
-      "grad_norm": 0.006679036654531956,
-      "learning_rate": 0.00019998390515855871,
-      "loss": 46.0,
-      "step": 35474
-    },
-    {
-      "epoch": 5.712991666331173,
-      "grad_norm": 0.003336266614496708,
-      "learning_rate": 0.00019998390425089818,
-      "loss": 46.0,
-      "step": 35475
-    },
-    {
-      "epoch": 5.7131527034099605,
-      "grad_norm": 0.005979595240205526,
-      "learning_rate": 0.0001999839033432121,
-      "loss": 46.0,
-      "step": 35476
-    },
-    {
-      "epoch": 5.713313740488747,
-      "grad_norm": 0.018231607973575592,
-      "learning_rate": 0.00019998390243550036,
-      "loss": 46.0,
-      "step": 35477
-    },
-    {
-      "epoch": 5.713474777567535,
-      "grad_norm": 0.002581343287602067,
-      "learning_rate": 0.00019998390152776305,
-      "loss": 46.0,
-      "step": 35478
-    },
-    {
-      "epoch": 5.713635814646322,
-      "grad_norm": 0.004995923954993486,
-      "learning_rate": 0.00019998390062000017,
-      "loss": 46.0,
-      "step": 35479
-    },
-    {
-      "epoch": 5.713796851725109,
-      "grad_norm": 0.0024382679257541895,
-      "learning_rate": 0.00019998389971221167,
-      "loss": 46.0,
-      "step": 35480
-    },
-    {
-      "epoch": 5.713957888803897,
-      "grad_norm": 0.010854941792786121,
-      "learning_rate": 0.0001999838988043976,
-      "loss": 46.0,
-      "step": 35481
-    },
-    {
-      "epoch": 5.714118925882684,
-      "grad_norm": 0.01824612356722355,
-      "learning_rate": 0.00019998389789655793,
-      "loss": 46.0,
-      "step": 35482
-    },
-    {
-      "epoch": 5.714279962961472,
-      "grad_norm": 0.002907045418396592,
-      "learning_rate": 0.00019998389698869267,
-      "loss": 46.0,
-      "step": 35483
-    },
-    {
-      "epoch": 5.714441000040259,
-      "grad_norm": 0.010969860479235649,
-      "learning_rate": 0.0001999838960808018,
-      "loss": 46.0,
-      "step": 35484
-    },
-    {
-      "epoch": 5.714602037119047,
-      "grad_norm": 0.0015488305361941457,
-      "learning_rate": 0.00019998389517288535,
-      "loss": 46.0,
-      "step": 35485
-    },
-    {
-      "epoch": 5.714763074197834,
-      "grad_norm": 0.0008086160523816943,
-      "learning_rate": 0.00019998389426494333,
-      "loss": 46.0,
-      "step": 35486
-    },
-    {
-      "epoch": 5.714924111276622,
-      "grad_norm": 0.0008083141874521971,
-      "learning_rate": 0.0001999838933569757,
-      "loss": 46.0,
-      "step": 35487
-    },
-    {
-      "epoch": 5.715085148355409,
-      "grad_norm": 0.00237140990793705,
-      "learning_rate": 0.0001999838924489825,
-      "loss": 46.0,
-      "step": 35488
-    },
-    {
-      "epoch": 5.7152461854341965,
-      "grad_norm": 0.006972777657210827,
-      "learning_rate": 0.0001999838915409637,
-      "loss": 46.0,
-      "step": 35489
-    },
-    {
-      "epoch": 5.715407222512984,
-      "grad_norm": 0.006063186097890139,
-      "learning_rate": 0.00019998389063291932,
-      "loss": 46.0,
-      "step": 35490
-    },
-    {
-      "epoch": 5.715568259591771,
-      "grad_norm": 0.004452622961252928,
-      "learning_rate": 0.00019998388972484934,
-      "loss": 46.0,
-      "step": 35491
-    },
-    {
-      "epoch": 5.715729296670558,
-      "grad_norm": 0.00532748457044363,
-      "learning_rate": 0.00019998388881675374,
-      "loss": 46.0,
-      "step": 35492
-    },
-    {
-      "epoch": 5.715890333749345,
-      "grad_norm": 0.002177457557991147,
-      "learning_rate": 0.0001999838879086326,
-      "loss": 46.0,
-      "step": 35493
-    },
-    {
-      "epoch": 5.716051370828133,
-      "grad_norm": 0.007527903653681278,
-      "learning_rate": 0.00019998388700048582,
-      "loss": 46.0,
-      "step": 35494
-    },
-    {
-      "epoch": 5.71621240790692,
-      "grad_norm": 0.0010966056725010276,
-      "learning_rate": 0.00019998388609231347,
-      "loss": 46.0,
-      "step": 35495
-    },
-    {
-      "epoch": 5.716373444985708,
-      "grad_norm": 0.005752748344093561,
-      "learning_rate": 0.00019998388518411555,
-      "loss": 46.0,
-      "step": 35496
-    },
-    {
-      "epoch": 5.716534482064495,
-      "grad_norm": 0.003387514501810074,
-      "learning_rate": 0.000199983884275892,
-      "loss": 46.0,
-      "step": 35497
-    },
-    {
-      "epoch": 5.716695519143283,
-      "grad_norm": 0.007491863798350096,
-      "learning_rate": 0.00019998388336764288,
-      "loss": 46.0,
-      "step": 35498
-    },
-    {
-      "epoch": 5.71685655622207,
-      "grad_norm": 0.002125262515619397,
-      "learning_rate": 0.00019998388245936818,
-      "loss": 46.0,
-      "step": 35499
-    },
-    {
-      "epoch": 5.717017593300858,
-      "grad_norm": 0.0027601609472185373,
-      "learning_rate": 0.00019998388155106786,
-      "loss": 46.0,
-      "step": 35500
-    },
-    {
-      "epoch": 5.717178630379645,
-      "grad_norm": 0.005047936458140612,
-      "learning_rate": 0.00019998388064274198,
-      "loss": 46.0,
-      "step": 35501
-    },
-    {
-      "epoch": 5.7173396674584325,
-      "grad_norm": 0.0047386400401592255,
-      "learning_rate": 0.0001999838797343905,
-      "loss": 46.0,
-      "step": 35502
-    },
-    {
-      "epoch": 5.71750070453722,
-      "grad_norm": 0.009273567236959934,
-      "learning_rate": 0.0001999838788260134,
-      "loss": 46.0,
-      "step": 35503
-    },
-    {
-      "epoch": 5.717661741616007,
-      "grad_norm": 0.003139413660392165,
-      "learning_rate": 0.00019998387791761077,
-      "loss": 46.0,
-      "step": 35504
-    },
-    {
-      "epoch": 5.717822778694795,
-      "grad_norm": 0.0024542633909732103,
-      "learning_rate": 0.00019998387700918252,
-      "loss": 46.0,
-      "step": 35505
-    },
-    {
-      "epoch": 5.717983815773582,
-      "grad_norm": 0.004819979891180992,
-      "learning_rate": 0.00019998387610072865,
-      "loss": 46.0,
-      "step": 35506
-    },
-    {
-      "epoch": 5.718144852852369,
-      "grad_norm": 0.01143036037683487,
-      "learning_rate": 0.00019998387519224922,
-      "loss": 46.0,
-      "step": 35507
-    },
-    {
-      "epoch": 5.718305889931156,
-      "grad_norm": 0.013003865256905556,
-      "learning_rate": 0.00019998387428374418,
-      "loss": 46.0,
-      "step": 35508
-    },
-    {
-      "epoch": 5.718466927009944,
-      "grad_norm": 0.009061514399945736,
-      "learning_rate": 0.00019998387337521355,
-      "loss": 46.0,
-      "step": 35509
-    },
-    {
-      "epoch": 5.718627964088731,
-      "grad_norm": 0.028800861909985542,
-      "learning_rate": 0.00019998387246665736,
-      "loss": 46.0,
-      "step": 35510
-    },
-    {
-      "epoch": 5.718789001167519,
-      "grad_norm": 0.002898689592257142,
-      "learning_rate": 0.00019998387155807556,
-      "loss": 46.0,
-      "step": 35511
-    },
-    {
-      "epoch": 5.718950038246306,
-      "grad_norm": 0.0011271587572991848,
-      "learning_rate": 0.00019998387064946817,
-      "loss": 46.0,
-      "step": 35512
-    },
-    {
-      "epoch": 5.7191110753250936,
-      "grad_norm": 0.014553409069776535,
-      "learning_rate": 0.00019998386974083522,
-      "loss": 46.0,
-      "step": 35513
-    },
-    {
-      "epoch": 5.719272112403881,
-      "grad_norm": 0.004143801983445883,
-      "learning_rate": 0.00019998386883217663,
-      "loss": 46.0,
-      "step": 35514
-    },
-    {
-      "epoch": 5.7194331494826685,
-      "grad_norm": 0.0005808683345094323,
-      "learning_rate": 0.00019998386792349245,
-      "loss": 46.0,
-      "step": 35515
-    },
-    {
-      "epoch": 5.719594186561456,
-      "grad_norm": 0.006136034149676561,
-      "learning_rate": 0.0001999838670147827,
-      "loss": 46.0,
-      "step": 35516
-    },
-    {
-      "epoch": 5.719755223640243,
-      "grad_norm": 0.002585548674687743,
-      "learning_rate": 0.00019998386610604735,
-      "loss": 46.0,
-      "step": 35517
-    },
-    {
-      "epoch": 5.719916260719031,
-      "grad_norm": 0.005272881127893925,
-      "learning_rate": 0.00019998386519728641,
-      "loss": 46.0,
-      "step": 35518
-    },
-    {
-      "epoch": 5.720077297797818,
-      "grad_norm": 0.007837477140128613,
-      "learning_rate": 0.0001999838642884999,
-      "loss": 46.0,
-      "step": 35519
-    },
-    {
-      "epoch": 5.720238334876606,
-      "grad_norm": 0.0058410619385540485,
-      "learning_rate": 0.00019998386337968777,
-      "loss": 46.0,
-      "step": 35520
-    },
-    {
-      "epoch": 5.720399371955393,
-      "grad_norm": 0.004429119173437357,
-      "learning_rate": 0.00019998386247085007,
-      "loss": 46.0,
-      "step": 35521
-    },
-    {
-      "epoch": 5.72056040903418,
-      "grad_norm": 0.005085925571620464,
-      "learning_rate": 0.00019998386156198675,
-      "loss": 46.0,
-      "step": 35522
-    },
-    {
-      "epoch": 5.720721446112967,
-      "grad_norm": 0.00603752164170146,
-      "learning_rate": 0.00019998386065309787,
-      "loss": 46.0,
-      "step": 35523
-    },
-    {
-      "epoch": 5.720882483191755,
-      "grad_norm": 0.007471588905900717,
-      "learning_rate": 0.0001999838597441834,
-      "loss": 46.0,
-      "step": 35524
-    },
-    {
-      "epoch": 5.721043520270542,
-      "grad_norm": 0.0014969459734857082,
-      "learning_rate": 0.00019998385883524333,
-      "loss": 46.0,
-      "step": 35525
-    },
-    {
-      "epoch": 5.7212045573493295,
-      "grad_norm": 0.0057539427652955055,
-      "learning_rate": 0.00019998385792627767,
-      "loss": 46.0,
-      "step": 35526
-    },
-    {
-      "epoch": 5.721365594428117,
-      "grad_norm": 0.004889465402811766,
-      "learning_rate": 0.00019998385701728641,
-      "loss": 46.0,
-      "step": 35527
-    },
-    {
-      "epoch": 5.721526631506904,
-      "grad_norm": 0.010457836091518402,
-      "learning_rate": 0.0001999838561082696,
-      "loss": 46.0,
-      "step": 35528
-    },
-    {
-      "epoch": 5.721687668585692,
-      "grad_norm": 0.001264468184672296,
-      "learning_rate": 0.00019998385519922715,
-      "loss": 46.0,
-      "step": 35529
-    },
-    {
-      "epoch": 5.721848705664479,
-      "grad_norm": 0.0017279409803450108,
-      "learning_rate": 0.00019998385429015913,
-      "loss": 46.0,
-      "step": 35530
-    },
-    {
-      "epoch": 5.722009742743267,
-      "grad_norm": 0.0076323822140693665,
-      "learning_rate": 0.0001999838533810655,
-      "loss": 46.0,
-      "step": 35531
-    },
-    {
-      "epoch": 5.722170779822054,
-      "grad_norm": 0.01451136451214552,
-      "learning_rate": 0.00019998385247194632,
-      "loss": 46.0,
-      "step": 35532
-    },
-    {
-      "epoch": 5.722331816900842,
-      "grad_norm": 0.007001708727329969,
-      "learning_rate": 0.00019998385156280152,
-      "loss": 46.0,
-      "step": 35533
-    },
-    {
-      "epoch": 5.722492853979629,
-      "grad_norm": 0.001599399489350617,
-      "learning_rate": 0.00019998385065363113,
-      "loss": 46.0,
-      "step": 35534
-    },
-    {
-      "epoch": 5.722653891058416,
-      "grad_norm": 0.0036499109119176865,
-      "learning_rate": 0.00019998384974443515,
-      "loss": 46.0,
-      "step": 35535
-    },
-    {
-      "epoch": 5.722814928137204,
-      "grad_norm": 0.0013848617672920227,
-      "learning_rate": 0.00019998384883521358,
-      "loss": 46.0,
-      "step": 35536
-    },
-    {
-      "epoch": 5.722975965215991,
-      "grad_norm": 0.0069944728165864944,
-      "learning_rate": 0.0001999838479259664,
-      "loss": 46.0,
-      "step": 35537
-    },
-    {
-      "epoch": 5.723137002294778,
-      "grad_norm": 0.00514630414545536,
-      "learning_rate": 0.00019998384701669367,
-      "loss": 46.0,
-      "step": 35538
-    },
-    {
-      "epoch": 5.7232980393735655,
-      "grad_norm": 0.0030399570241570473,
-      "learning_rate": 0.00019998384610739534,
-      "loss": 46.0,
-      "step": 35539
-    },
-    {
-      "epoch": 5.723459076452353,
-      "grad_norm": 0.003940983675420284,
-      "learning_rate": 0.00019998384519807137,
-      "loss": 46.0,
-      "step": 35540
-    },
-    {
-      "epoch": 5.72362011353114,
-      "grad_norm": 0.021705487743020058,
-      "learning_rate": 0.00019998384428872185,
-      "loss": 46.0,
-      "step": 35541
-    },
-    {
-      "epoch": 5.723781150609928,
-      "grad_norm": 0.003187623806297779,
-      "learning_rate": 0.00019998384337934676,
-      "loss": 46.0,
-      "step": 35542
-    },
-    {
-      "epoch": 5.723942187688715,
-      "grad_norm": 0.0025115148164331913,
-      "learning_rate": 0.00019998384246994606,
-      "loss": 46.0,
-      "step": 35543
-    },
-    {
-      "epoch": 5.724103224767503,
-      "grad_norm": 0.019701754674315453,
-      "learning_rate": 0.00019998384156051974,
-      "loss": 46.0,
-      "step": 35544
-    },
-    {
-      "epoch": 5.72426426184629,
-      "grad_norm": 0.022417917847633362,
-      "learning_rate": 0.00019998384065106787,
-      "loss": 46.0,
-      "step": 35545
-    },
-    {
-      "epoch": 5.724425298925078,
-      "grad_norm": 0.010307450778782368,
-      "learning_rate": 0.00019998383974159038,
-      "loss": 46.0,
-      "step": 35546
-    },
-    {
-      "epoch": 5.724586336003865,
-      "grad_norm": 0.004518875852227211,
-      "learning_rate": 0.00019998383883208733,
-      "loss": 46.0,
-      "step": 35547
-    },
-    {
-      "epoch": 5.724747373082653,
-      "grad_norm": 0.013035601004958153,
-      "learning_rate": 0.00019998383792255866,
-      "loss": 46.0,
-      "step": 35548
-    },
-    {
-      "epoch": 5.72490841016144,
-      "grad_norm": 0.00627890694886446,
-      "learning_rate": 0.00019998383701300444,
-      "loss": 46.0,
-      "step": 35549
-    },
-    {
-      "epoch": 5.725069447240227,
-      "grad_norm": 0.001404904993250966,
-      "learning_rate": 0.00019998383610342457,
-      "loss": 46.0,
-      "step": 35550
-    },
-    {
-      "epoch": 5.725230484319015,
-      "grad_norm": 0.007445406634360552,
-      "learning_rate": 0.00019998383519381915,
-      "loss": 46.0,
-      "step": 35551
-    },
-    {
-      "epoch": 5.7253915213978015,
-      "grad_norm": 0.0019616286735981703,
-      "learning_rate": 0.00019998383428418813,
-      "loss": 46.0,
-      "step": 35552
-    },
-    {
-      "epoch": 5.725552558476589,
-      "grad_norm": 0.0012010824866592884,
-      "learning_rate": 0.00019998383337453153,
-      "loss": 46.0,
-      "step": 35553
-    },
-    {
-      "epoch": 5.725713595555376,
-      "grad_norm": 0.010856564156711102,
-      "learning_rate": 0.00019998383246484932,
-      "loss": 46.0,
-      "step": 35554
-    },
-    {
-      "epoch": 5.725874632634164,
-      "grad_norm": 0.009345943108201027,
-      "learning_rate": 0.00019998383155514154,
-      "loss": 46.0,
-      "step": 35555
-    },
-    {
-      "epoch": 5.726035669712951,
-      "grad_norm": 0.0037518595345318317,
-      "learning_rate": 0.00019998383064540813,
-      "loss": 46.0,
-      "step": 35556
-    },
-    {
-      "epoch": 5.726196706791739,
-      "grad_norm": 0.007128858473151922,
-      "learning_rate": 0.00019998382973564918,
-      "loss": 46.0,
-      "step": 35557
-    },
-    {
-      "epoch": 5.726357743870526,
-      "grad_norm": 0.007594176102429628,
-      "learning_rate": 0.0001999838288258646,
-      "loss": 46.0,
-      "step": 35558
-    },
-    {
-      "epoch": 5.726518780949314,
-      "grad_norm": 0.0034325586166232824,
-      "learning_rate": 0.00019998382791605447,
-      "loss": 46.0,
-      "step": 35559
-    },
-    {
-      "epoch": 5.726679818028101,
-      "grad_norm": 0.0015808983007445931,
-      "learning_rate": 0.0001999838270062187,
-      "loss": 46.0,
-      "step": 35560
-    },
-    {
-      "epoch": 5.726840855106889,
-      "grad_norm": 0.01398675236850977,
-      "learning_rate": 0.00019998382609635738,
-      "loss": 46.0,
-      "step": 35561
-    },
-    {
-      "epoch": 5.727001892185676,
-      "grad_norm": 0.010379220359027386,
-      "learning_rate": 0.00019998382518647044,
-      "loss": 46.0,
-      "step": 35562
-    },
-    {
-      "epoch": 5.7271629292644635,
-      "grad_norm": 0.0023658093996345997,
-      "learning_rate": 0.00019998382427655794,
-      "loss": 46.0,
-      "step": 35563
-    },
-    {
-      "epoch": 5.727323966343251,
-      "grad_norm": 0.003652440384030342,
-      "learning_rate": 0.0001999838233666198,
-      "loss": 46.0,
-      "step": 35564
-    },
-    {
-      "epoch": 5.7274850034220375,
-      "grad_norm": 0.007258872967213392,
-      "learning_rate": 0.0001999838224566561,
-      "loss": 46.0,
-      "step": 35565
-    },
-    {
-      "epoch": 5.727646040500825,
-      "grad_norm": 0.00116466183681041,
-      "learning_rate": 0.0001999838215466668,
-      "loss": 46.0,
-      "step": 35566
-    },
-    {
-      "epoch": 5.727807077579612,
-      "grad_norm": 0.0013377966824918985,
-      "learning_rate": 0.00019998382063665196,
-      "loss": 46.0,
-      "step": 35567
-    },
-    {
-      "epoch": 5.7279681146584,
-      "grad_norm": 0.0017657188000157475,
-      "learning_rate": 0.00019998381972661147,
-      "loss": 46.0,
-      "step": 35568
-    },
-    {
-      "epoch": 5.728129151737187,
-      "grad_norm": 0.0041077835485339165,
-      "learning_rate": 0.0001999838188165454,
-      "loss": 46.0,
-      "step": 35569
-    },
-    {
-      "epoch": 5.728290188815975,
-      "grad_norm": 0.008129140362143517,
-      "learning_rate": 0.00019998381790645376,
-      "loss": 46.0,
-      "step": 35570
-    },
-    {
-      "epoch": 5.728451225894762,
-      "grad_norm": 0.0026882204692810774,
-      "learning_rate": 0.00019998381699633654,
-      "loss": 46.0,
-      "step": 35571
-    },
-    {
-      "epoch": 5.72861226297355,
-      "grad_norm": 0.0014871583553031087,
-      "learning_rate": 0.00019998381608619367,
-      "loss": 46.0,
-      "step": 35572
-    },
-    {
-      "epoch": 5.728773300052337,
-      "grad_norm": 0.0032231139484792948,
-      "learning_rate": 0.00019998381517602525,
-      "loss": 46.0,
-      "step": 35573
-    },
-    {
-      "epoch": 5.7289343371311245,
-      "grad_norm": 0.0014955142978578806,
-      "learning_rate": 0.00019998381426583123,
-      "loss": 46.0,
-      "step": 35574
-    },
-    {
-      "epoch": 5.729095374209912,
-      "grad_norm": 0.005558602046221495,
-      "learning_rate": 0.00019998381335561163,
-      "loss": 46.0,
-      "step": 35575
-    },
-    {
-      "epoch": 5.7292564112886994,
-      "grad_norm": 0.0031127077527344227,
-      "learning_rate": 0.00019998381244536645,
-      "loss": 46.0,
-      "step": 35576
-    },
-    {
-      "epoch": 5.729417448367487,
-      "grad_norm": 0.0014645750634372234,
-      "learning_rate": 0.00019998381153509565,
-      "loss": 46.0,
-      "step": 35577
-    },
-    {
-      "epoch": 5.729578485446274,
-      "grad_norm": 0.005130200181156397,
-      "learning_rate": 0.00019998381062479926,
-      "loss": 46.0,
-      "step": 35578
-    },
-    {
-      "epoch": 5.729739522525062,
-      "grad_norm": 0.021106954663991928,
-      "learning_rate": 0.0001999838097144773,
-      "loss": 46.0,
-      "step": 35579
-    },
-    {
-      "epoch": 5.729900559603848,
-      "grad_norm": 0.017933879047632217,
-      "learning_rate": 0.00019998380880412975,
-      "loss": 46.0,
-      "step": 35580
-    },
-    {
-      "epoch": 5.730061596682636,
-      "grad_norm": 0.00688158068805933,
-      "learning_rate": 0.0001999838078937566,
-      "loss": 46.0,
-      "step": 35581
-    },
-    {
-      "epoch": 5.730222633761423,
-      "grad_norm": 0.0023649160284549,
-      "learning_rate": 0.00019998380698335784,
-      "loss": 46.0,
-      "step": 35582
-    },
-    {
-      "epoch": 5.730383670840211,
-      "grad_norm": 0.004906385205686092,
-      "learning_rate": 0.0001999838060729335,
-      "loss": 46.0,
-      "step": 35583
-    },
-    {
-      "epoch": 5.730544707918998,
-      "grad_norm": 0.0034793447703123093,
-      "learning_rate": 0.00019998380516248357,
-      "loss": 46.0,
-      "step": 35584
-    },
-    {
-      "epoch": 5.730705744997786,
-      "grad_norm": 0.008268315345048904,
-      "learning_rate": 0.00019998380425200808,
-      "loss": 46.0,
-      "step": 35585
-    },
-    {
-      "epoch": 5.730866782076573,
-      "grad_norm": 0.0066382382065057755,
-      "learning_rate": 0.00019998380334150694,
-      "loss": 46.0,
-      "step": 35586
-    },
-    {
-      "epoch": 5.7310278191553605,
-      "grad_norm": 0.0013370001688599586,
-      "learning_rate": 0.00019998380243098026,
-      "loss": 46.0,
-      "step": 35587
-    },
-    {
-      "epoch": 5.731188856234148,
-      "grad_norm": 0.016605084761977196,
-      "learning_rate": 0.00019998380152042798,
-      "loss": 46.0,
-      "step": 35588
-    },
-    {
-      "epoch": 5.731349893312935,
-      "grad_norm": 0.0036441953852772713,
-      "learning_rate": 0.0001999838006098501,
-      "loss": 46.0,
-      "step": 35589
-    },
-    {
-      "epoch": 5.731510930391723,
-      "grad_norm": 0.0015499975997954607,
-      "learning_rate": 0.00019998379969924664,
-      "loss": 46.0,
-      "step": 35590
-    },
-    {
-      "epoch": 5.73167196747051,
-      "grad_norm": 0.0052222684025764465,
-      "learning_rate": 0.0001999837987886176,
-      "loss": 46.0,
-      "step": 35591
-    },
-    {
-      "epoch": 5.731833004549298,
-      "grad_norm": 0.002013799035921693,
-      "learning_rate": 0.00019998379787796293,
-      "loss": 46.0,
-      "step": 35592
-    },
-    {
-      "epoch": 5.731994041628085,
-      "grad_norm": 0.001408213283866644,
-      "learning_rate": 0.00019998379696728268,
-      "loss": 46.0,
-      "step": 35593
-    },
-    {
-      "epoch": 5.732155078706873,
-      "grad_norm": 0.0074705276638269424,
-      "learning_rate": 0.00019998379605657687,
-      "loss": 46.0,
-      "step": 35594
-    },
-    {
-      "epoch": 5.732316115785659,
-      "grad_norm": 0.002812921768054366,
-      "learning_rate": 0.00019998379514584542,
-      "loss": 46.0,
-      "step": 35595
-    },
-    {
-      "epoch": 5.732477152864447,
-      "grad_norm": 0.0022542367223650217,
-      "learning_rate": 0.00019998379423508844,
-      "loss": 46.0,
-      "step": 35596
-    },
-    {
-      "epoch": 5.732638189943234,
-      "grad_norm": 0.002422074321657419,
-      "learning_rate": 0.00019998379332430584,
-      "loss": 46.0,
-      "step": 35597
-    },
-    {
-      "epoch": 5.732799227022022,
-      "grad_norm": 0.009186848066747189,
-      "learning_rate": 0.00019998379241349762,
-      "loss": 46.0,
-      "step": 35598
-    },
-    {
-      "epoch": 5.732960264100809,
-      "grad_norm": 0.0007388254161924124,
-      "learning_rate": 0.00019998379150266385,
-      "loss": 46.0,
-      "step": 35599
-    },
-    {
-      "epoch": 5.7331213011795965,
-      "grad_norm": 0.0061682178638875484,
-      "learning_rate": 0.00019998379059180447,
-      "loss": 46.0,
-      "step": 35600
-    },
-    {
-      "epoch": 5.733282338258384,
-      "grad_norm": 0.003723063040524721,
-      "learning_rate": 0.0001999837896809195,
-      "loss": 46.0,
-      "step": 35601
-    },
-    {
-      "epoch": 5.733443375337171,
-      "grad_norm": 0.0012781383702531457,
-      "learning_rate": 0.00019998378877000893,
-      "loss": 46.0,
-      "step": 35602
-    },
-    {
-      "epoch": 5.733604412415959,
-      "grad_norm": 0.0009361369884572923,
-      "learning_rate": 0.0001999837878590728,
-      "loss": 46.0,
-      "step": 35603
-    },
-    {
-      "epoch": 5.733765449494746,
-      "grad_norm": 0.00432397099211812,
-      "learning_rate": 0.00019998378694811107,
-      "loss": 46.0,
-      "step": 35604
-    },
-    {
-      "epoch": 5.733926486573534,
-      "grad_norm": 0.0022777237463742495,
-      "learning_rate": 0.00019998378603712372,
-      "loss": 46.0,
-      "step": 35605
-    },
-    {
-      "epoch": 5.734087523652321,
-      "grad_norm": 0.0024943219032138586,
-      "learning_rate": 0.0001999837851261108,
-      "loss": 46.0,
-      "step": 35606
-    },
-    {
-      "epoch": 5.734248560731109,
-      "grad_norm": 0.003432539524510503,
-      "learning_rate": 0.0001999837842150723,
-      "loss": 46.0,
-      "step": 35607
-    },
-    {
-      "epoch": 5.734409597809895,
-      "grad_norm": 0.010588981211185455,
-      "learning_rate": 0.0001999837833040082,
-      "loss": 46.0,
-      "step": 35608
-    },
-    {
-      "epoch": 5.734570634888684,
-      "grad_norm": 0.002912891795858741,
-      "learning_rate": 0.0001999837823929185,
-      "loss": 46.0,
-      "step": 35609
-    },
-    {
-      "epoch": 5.73473167196747,
-      "grad_norm": 0.006764731369912624,
-      "learning_rate": 0.00019998378148180322,
-      "loss": 46.0,
-      "step": 35610
-    },
-    {
-      "epoch": 5.734892709046258,
-      "grad_norm": 0.019105613231658936,
-      "learning_rate": 0.00019998378057066237,
-      "loss": 46.0,
-      "step": 35611
-    },
-    {
-      "epoch": 5.735053746125045,
-      "grad_norm": 0.005462784320116043,
-      "learning_rate": 0.00019998377965949589,
-      "loss": 46.0,
-      "step": 35612
-    },
-    {
-      "epoch": 5.7352147832038325,
-      "grad_norm": 0.005559720564633608,
-      "learning_rate": 0.00019998377874830384,
-      "loss": 46.0,
-      "step": 35613
-    },
-    {
-      "epoch": 5.73537582028262,
-      "grad_norm": 0.006342420820146799,
-      "learning_rate": 0.0001999837778370862,
-      "loss": 46.0,
-      "step": 35614
-    },
-    {
-      "epoch": 5.735536857361407,
-      "grad_norm": 0.0035092777106910944,
-      "learning_rate": 0.00019998377692584295,
-      "loss": 46.0,
-      "step": 35615
-    },
-    {
-      "epoch": 5.735697894440195,
-      "grad_norm": 0.0015751500613987446,
-      "learning_rate": 0.00019998377601457415,
-      "loss": 46.0,
-      "step": 35616
-    },
-    {
-      "epoch": 5.735858931518982,
-      "grad_norm": 0.013914305716753006,
-      "learning_rate": 0.00019998377510327972,
-      "loss": 46.0,
-      "step": 35617
-    },
-    {
-      "epoch": 5.73601996859777,
-      "grad_norm": 0.009146296419203281,
-      "learning_rate": 0.0001999837741919597,
-      "loss": 46.0,
-      "step": 35618
-    },
-    {
-      "epoch": 5.736181005676557,
-      "grad_norm": 0.003716784995049238,
-      "learning_rate": 0.00019998377328061411,
-      "loss": 46.0,
-      "step": 35619
-    },
-    {
-      "epoch": 5.736342042755345,
-      "grad_norm": 0.00288646318949759,
-      "learning_rate": 0.0001999837723692429,
-      "loss": 46.0,
-      "step": 35620
-    },
-    {
-      "epoch": 5.736503079834132,
-      "grad_norm": 0.003845304949209094,
-      "learning_rate": 0.00019998377145784613,
-      "loss": 46.0,
-      "step": 35621
-    },
-    {
-      "epoch": 5.7366641169129196,
-      "grad_norm": 0.0012538401642814279,
-      "learning_rate": 0.00019998377054642377,
-      "loss": 46.0,
-      "step": 35622
-    },
-    {
-      "epoch": 5.736825153991706,
-      "grad_norm": 0.0018852229695767164,
-      "learning_rate": 0.0001999837696349758,
-      "loss": 46.0,
-      "step": 35623
-    },
-    {
-      "epoch": 5.7369861910704945,
-      "grad_norm": 0.003224159125238657,
-      "learning_rate": 0.00019998376872350227,
-      "loss": 46.0,
-      "step": 35624
-    },
-    {
-      "epoch": 5.737147228149281,
-      "grad_norm": 0.00536219822242856,
-      "learning_rate": 0.00019998376781200312,
-      "loss": 46.0,
-      "step": 35625
-    },
-    {
-      "epoch": 5.7373082652280685,
-      "grad_norm": 0.016085563227534294,
-      "learning_rate": 0.00019998376690047838,
-      "loss": 46.0,
-      "step": 35626
-    },
-    {
-      "epoch": 5.737469302306856,
-      "grad_norm": 0.004196074325591326,
-      "learning_rate": 0.00019998376598892806,
-      "loss": 46.0,
-      "step": 35627
-    },
-    {
-      "epoch": 5.737630339385643,
-      "grad_norm": 0.006946604233235121,
-      "learning_rate": 0.00019998376507735212,
-      "loss": 46.0,
-      "step": 35628
-    },
-    {
-      "epoch": 5.737791376464431,
-      "grad_norm": 0.0011422872776165605,
-      "learning_rate": 0.00019998376416575063,
-      "loss": 46.0,
-      "step": 35629
-    },
-    {
-      "epoch": 5.737952413543218,
-      "grad_norm": 0.007601715158671141,
-      "learning_rate": 0.00019998376325412355,
-      "loss": 46.0,
-      "step": 35630
-    },
-    {
-      "epoch": 5.738113450622006,
-      "grad_norm": 0.0023788739927113056,
-      "learning_rate": 0.00019998376234247085,
-      "loss": 46.0,
-      "step": 35631
-    },
-    {
-      "epoch": 5.738274487700793,
-      "grad_norm": 0.004944286309182644,
-      "learning_rate": 0.0001999837614307926,
-      "loss": 46.0,
-      "step": 35632
-    },
-    {
-      "epoch": 5.738435524779581,
-      "grad_norm": 0.007749716751277447,
-      "learning_rate": 0.00019998376051908872,
-      "loss": 46.0,
-      "step": 35633
-    },
-    {
-      "epoch": 5.738596561858368,
-      "grad_norm": 0.004866868257522583,
-      "learning_rate": 0.00019998375960735923,
-      "loss": 46.0,
-      "step": 35634
-    },
-    {
-      "epoch": 5.7387575989371555,
-      "grad_norm": 0.001834242488257587,
-      "learning_rate": 0.00019998375869560418,
-      "loss": 46.0,
-      "step": 35635
-    },
-    {
-      "epoch": 5.738918636015943,
-      "grad_norm": 0.007009696215391159,
-      "learning_rate": 0.00019998375778382355,
-      "loss": 46.0,
-      "step": 35636
-    },
-    {
-      "epoch": 5.73907967309473,
-      "grad_norm": 0.009853570722043514,
-      "learning_rate": 0.0001999837568720173,
-      "loss": 46.0,
-      "step": 35637
-    },
-    {
-      "epoch": 5.739240710173517,
-      "grad_norm": 0.005257311277091503,
-      "learning_rate": 0.0001999837559601855,
-      "loss": 46.0,
-      "step": 35638
-    },
-    {
-      "epoch": 5.739401747252305,
-      "grad_norm": 0.0093264514580369,
-      "learning_rate": 0.00019998375504832807,
-      "loss": 46.0,
-      "step": 35639
-    },
-    {
-      "epoch": 5.739562784331092,
-      "grad_norm": 0.006017458625137806,
-      "learning_rate": 0.0001999837541364451,
-      "loss": 46.0,
-      "step": 35640
-    },
-    {
-      "epoch": 5.739723821409879,
-      "grad_norm": 0.005904871504753828,
-      "learning_rate": 0.0001999837532245365,
-      "loss": 46.0,
-      "step": 35641
-    },
-    {
-      "epoch": 5.739884858488667,
-      "grad_norm": 0.0026297178119421005,
-      "learning_rate": 0.00019998375231260228,
-      "loss": 46.0,
-      "step": 35642
-    },
-    {
-      "epoch": 5.740045895567454,
-      "grad_norm": 0.004746863152831793,
-      "learning_rate": 0.00019998375140064254,
-      "loss": 46.0,
-      "step": 35643
-    },
-    {
-      "epoch": 5.740206932646242,
-      "grad_norm": 0.007847358472645283,
-      "learning_rate": 0.00019998375048865715,
-      "loss": 46.0,
-      "step": 35644
-    },
-    {
-      "epoch": 5.740367969725029,
-      "grad_norm": 0.001639272435568273,
-      "learning_rate": 0.0001999837495766462,
-      "loss": 46.0,
-      "step": 35645
-    },
-    {
-      "epoch": 5.740529006803817,
-      "grad_norm": 0.006071700248867273,
-      "learning_rate": 0.00019998374866460965,
-      "loss": 46.0,
-      "step": 35646
-    },
-    {
-      "epoch": 5.740690043882604,
-      "grad_norm": 0.0010262627620249987,
-      "learning_rate": 0.0001999837477525475,
-      "loss": 46.0,
-      "step": 35647
-    },
-    {
-      "epoch": 5.7408510809613915,
-      "grad_norm": 0.0063830227591097355,
-      "learning_rate": 0.00019998374684045977,
-      "loss": 46.0,
-      "step": 35648
-    },
-    {
-      "epoch": 5.741012118040179,
-      "grad_norm": 0.0019633148331195116,
-      "learning_rate": 0.00019998374592834645,
-      "loss": 46.0,
-      "step": 35649
-    },
-    {
-      "epoch": 5.741173155118966,
-      "grad_norm": 0.005420413333922625,
-      "learning_rate": 0.00019998374501620757,
-      "loss": 46.0,
-      "step": 35650
-    },
-    {
-      "epoch": 5.741334192197754,
-      "grad_norm": 0.0025122330989688635,
-      "learning_rate": 0.00019998374410404304,
-      "loss": 46.0,
-      "step": 35651
-    },
-    {
-      "epoch": 5.741495229276541,
-      "grad_norm": 0.0027262959629297256,
-      "learning_rate": 0.00019998374319185296,
-      "loss": 46.0,
-      "step": 35652
-    },
-    {
-      "epoch": 5.741656266355328,
-      "grad_norm": 0.0015047615161165595,
-      "learning_rate": 0.00019998374227963726,
-      "loss": 46.0,
-      "step": 35653
-    },
-    {
-      "epoch": 5.741817303434115,
-      "grad_norm": 0.0015291034942492843,
-      "learning_rate": 0.00019998374136739598,
-      "loss": 46.0,
-      "step": 35654
-    },
-    {
-      "epoch": 5.741978340512903,
-      "grad_norm": 0.0024005360901355743,
-      "learning_rate": 0.00019998374045512914,
-      "loss": 46.0,
-      "step": 35655
-    },
-    {
-      "epoch": 5.74213937759169,
-      "grad_norm": 0.014162642881274223,
-      "learning_rate": 0.00019998373954283668,
-      "loss": 46.0,
-      "step": 35656
-    },
-    {
-      "epoch": 5.742300414670478,
-      "grad_norm": 0.002535266103222966,
-      "learning_rate": 0.00019998373863051866,
-      "loss": 46.0,
-      "step": 35657
-    },
-    {
-      "epoch": 5.742461451749265,
-      "grad_norm": 0.008849915117025375,
-      "learning_rate": 0.000199983737718175,
-      "loss": 46.0,
-      "step": 35658
-    },
-    {
-      "epoch": 5.742622488828053,
-      "grad_norm": 0.0026538653764873743,
-      "learning_rate": 0.00019998373680580578,
-      "loss": 46.0,
-      "step": 35659
-    },
-    {
-      "epoch": 5.74278352590684,
-      "grad_norm": 0.0032929491717368364,
-      "learning_rate": 0.00019998373589341095,
-      "loss": 46.0,
-      "step": 35660
-    },
-    {
-      "epoch": 5.7429445629856275,
-      "grad_norm": 0.013679970987141132,
-      "learning_rate": 0.00019998373498099055,
-      "loss": 46.0,
-      "step": 35661
-    },
-    {
-      "epoch": 5.743105600064415,
-      "grad_norm": 0.0015535237034782767,
-      "learning_rate": 0.00019998373406854457,
-      "loss": 46.0,
-      "step": 35662
-    },
-    {
-      "epoch": 5.743266637143202,
-      "grad_norm": 0.004194140899926424,
-      "learning_rate": 0.00019998373315607295,
-      "loss": 46.0,
-      "step": 35663
-    },
-    {
-      "epoch": 5.74342767422199,
-      "grad_norm": 0.00208999658934772,
-      "learning_rate": 0.00019998373224357576,
-      "loss": 46.0,
-      "step": 35664
-    },
-    {
-      "epoch": 5.743588711300777,
-      "grad_norm": 0.004345456603914499,
-      "learning_rate": 0.000199983731331053,
-      "loss": 46.0,
-      "step": 35665
-    },
-    {
-      "epoch": 5.743749748379565,
-      "grad_norm": 0.004026954062283039,
-      "learning_rate": 0.00019998373041850464,
-      "loss": 46.0,
-      "step": 35666
-    },
-    {
-      "epoch": 5.743910785458352,
-      "grad_norm": 0.003164366353303194,
-      "learning_rate": 0.00019998372950593067,
-      "loss": 46.0,
-      "step": 35667
-    },
-    {
-      "epoch": 5.744071822537139,
-      "grad_norm": 0.002806995762512088,
-      "learning_rate": 0.00019998372859333113,
-      "loss": 46.0,
-      "step": 35668
-    },
-    {
-      "epoch": 5.744232859615926,
-      "grad_norm": 0.004286506678909063,
-      "learning_rate": 0.00019998372768070602,
-      "loss": 46.0,
-      "step": 35669
-    },
-    {
-      "epoch": 5.744393896694714,
-      "grad_norm": 0.0030181854963302612,
-      "learning_rate": 0.00019998372676805528,
-      "loss": 46.0,
-      "step": 35670
-    },
-    {
-      "epoch": 5.744554933773501,
-      "grad_norm": 0.002859077649191022,
-      "learning_rate": 0.00019998372585537896,
-      "loss": 46.0,
-      "step": 35671
-    },
-    {
-      "epoch": 5.744715970852289,
-      "grad_norm": 0.021344603970646858,
-      "learning_rate": 0.00019998372494267708,
-      "loss": 46.0,
-      "step": 35672
-    },
-    {
-      "epoch": 5.744877007931076,
-      "grad_norm": 0.0027133887633681297,
-      "learning_rate": 0.0001999837240299496,
-      "loss": 46.0,
-      "step": 35673
-    },
-    {
-      "epoch": 5.7450380450098635,
-      "grad_norm": 0.003130651544779539,
-      "learning_rate": 0.00019998372311719645,
-      "loss": 46.0,
-      "step": 35674
-    },
-    {
-      "epoch": 5.745199082088651,
-      "grad_norm": 0.003432266879826784,
-      "learning_rate": 0.0001999837222044178,
-      "loss": 46.0,
-      "step": 35675
-    },
-    {
-      "epoch": 5.745360119167438,
-      "grad_norm": 0.007205731235444546,
-      "learning_rate": 0.00019998372129161353,
-      "loss": 46.0,
-      "step": 35676
-    },
-    {
-      "epoch": 5.745521156246226,
-      "grad_norm": 0.008296342566609383,
-      "learning_rate": 0.00019998372037878368,
-      "loss": 46.0,
-      "step": 35677
-    },
-    {
-      "epoch": 5.745682193325013,
-      "grad_norm": 0.002843861235305667,
-      "learning_rate": 0.00019998371946592823,
-      "loss": 46.0,
-      "step": 35678
-    },
-    {
-      "epoch": 5.745843230403801,
-      "grad_norm": 0.005493230652064085,
-      "learning_rate": 0.00019998371855304718,
-      "loss": 46.0,
-      "step": 35679
-    },
-    {
-      "epoch": 5.746004267482588,
-      "grad_norm": 0.01543631311506033,
-      "learning_rate": 0.00019998371764014055,
-      "loss": 46.0,
-      "step": 35680
-    },
-    {
-      "epoch": 5.746165304561376,
-      "grad_norm": 0.006336414720863104,
-      "learning_rate": 0.00019998371672720833,
-      "loss": 46.0,
-      "step": 35681
-    },
-    {
-      "epoch": 5.746326341640163,
-      "grad_norm": 0.0014595752581954002,
-      "learning_rate": 0.00019998371581425052,
-      "loss": 46.0,
-      "step": 35682
-    },
-    {
-      "epoch": 5.74648737871895,
-      "grad_norm": 0.01127290166914463,
-      "learning_rate": 0.0001999837149012671,
-      "loss": 46.0,
-      "step": 35683
-    },
-    {
-      "epoch": 5.746648415797737,
-      "grad_norm": 0.0035653971135616302,
-      "learning_rate": 0.0001999837139882581,
-      "loss": 46.0,
-      "step": 35684
-    },
-    {
-      "epoch": 5.746809452876525,
-      "grad_norm": 0.007793842814862728,
-      "learning_rate": 0.00019998371307522353,
-      "loss": 46.0,
-      "step": 35685
-    },
-    {
-      "epoch": 5.746970489955312,
-      "grad_norm": 0.008492912165820599,
-      "learning_rate": 0.00019998371216216337,
-      "loss": 46.0,
-      "step": 35686
-    },
-    {
-      "epoch": 5.7471315270340995,
-      "grad_norm": 0.0024247774854302406,
-      "learning_rate": 0.00019998371124907758,
-      "loss": 46.0,
-      "step": 35687
-    },
-    {
-      "epoch": 5.747292564112887,
-      "grad_norm": 0.0038225636817514896,
-      "learning_rate": 0.00019998371033596622,
-      "loss": 46.0,
-      "step": 35688
-    },
-    {
-      "epoch": 5.747453601191674,
-      "grad_norm": 0.001692612306214869,
-      "learning_rate": 0.00019998370942282928,
-      "loss": 46.0,
-      "step": 35689
-    },
-    {
-      "epoch": 5.747614638270462,
-      "grad_norm": 0.021703489124774933,
-      "learning_rate": 0.00019998370850966675,
-      "loss": 46.0,
-      "step": 35690
-    },
-    {
-      "epoch": 5.747775675349249,
-      "grad_norm": 0.0015950225060805678,
-      "learning_rate": 0.0001999837075964786,
-      "loss": 46.0,
-      "step": 35691
-    },
-    {
-      "epoch": 5.747936712428037,
-      "grad_norm": 0.004989693872630596,
-      "learning_rate": 0.0001999837066832649,
-      "loss": 46.0,
-      "step": 35692
-    },
-    {
-      "epoch": 5.748097749506824,
-      "grad_norm": 0.014336420223116875,
-      "learning_rate": 0.00019998370577002555,
-      "loss": 46.0,
-      "step": 35693
-    },
-    {
-      "epoch": 5.748258786585612,
-      "grad_norm": 0.014689628034830093,
-      "learning_rate": 0.00019998370485676067,
-      "loss": 46.0,
-      "step": 35694
-    },
-    {
-      "epoch": 5.748419823664399,
-      "grad_norm": 0.005761393811553717,
-      "learning_rate": 0.00019998370394347018,
-      "loss": 46.0,
-      "step": 35695
-    },
-    {
-      "epoch": 5.748580860743186,
-      "grad_norm": 0.0015850585186854005,
-      "learning_rate": 0.0001999837030301541,
-      "loss": 46.0,
-      "step": 35696
-    },
-    {
-      "epoch": 5.748741897821974,
-      "grad_norm": 0.012452409602701664,
-      "learning_rate": 0.00019998370211681243,
-      "loss": 46.0,
-      "step": 35697
-    },
-    {
-      "epoch": 5.7489029349007605,
-      "grad_norm": 0.0020173839293420315,
-      "learning_rate": 0.00019998370120344517,
-      "loss": 46.0,
-      "step": 35698
-    },
-    {
-      "epoch": 5.749063971979548,
-      "grad_norm": 0.020098872482776642,
-      "learning_rate": 0.0001999837002900523,
-      "loss": 46.0,
-      "step": 35699
-    },
-    {
-      "epoch": 5.7492250090583354,
-      "grad_norm": 0.002782142488285899,
-      "learning_rate": 0.00019998369937663385,
-      "loss": 46.0,
-      "step": 35700
-    },
-    {
-      "epoch": 5.749386046137123,
-      "grad_norm": 0.007855231873691082,
-      "learning_rate": 0.00019998369846318983,
-      "loss": 46.0,
-      "step": 35701
-    },
-    {
-      "epoch": 5.74954708321591,
-      "grad_norm": 0.0024741862434893847,
-      "learning_rate": 0.0001999836975497202,
-      "loss": 46.0,
-      "step": 35702
-    },
-    {
-      "epoch": 5.749708120294698,
-      "grad_norm": 0.02188832126557827,
-      "learning_rate": 0.00019998369663622498,
-      "loss": 46.0,
-      "step": 35703
-    },
-    {
-      "epoch": 5.749869157373485,
-      "grad_norm": 0.01413187850266695,
-      "learning_rate": 0.00019998369572270418,
-      "loss": 46.0,
-      "step": 35704
-    },
-    {
-      "epoch": 5.750030194452273,
-      "grad_norm": 0.0059817396104335785,
-      "learning_rate": 0.00019998369480915779,
-      "loss": 46.0,
-      "step": 35705
-    },
-    {
-      "epoch": 5.75019123153106,
-      "grad_norm": 0.0071333060041069984,
-      "learning_rate": 0.00019998369389558578,
-      "loss": 46.0,
-      "step": 35706
-    },
-    {
-      "epoch": 5.750352268609848,
-      "grad_norm": 0.009184440597891808,
-      "learning_rate": 0.0001999836929819882,
-      "loss": 46.0,
-      "step": 35707
-    },
-    {
-      "epoch": 5.750513305688635,
-      "grad_norm": 0.0009316659416072071,
-      "learning_rate": 0.00019998369206836503,
-      "loss": 46.0,
-      "step": 35708
-    },
-    {
-      "epoch": 5.7506743427674225,
-      "grad_norm": 0.016463695093989372,
-      "learning_rate": 0.00019998369115471626,
-      "loss": 46.0,
-      "step": 35709
-    },
-    {
-      "epoch": 5.75083537984621,
-      "grad_norm": 0.005868117790669203,
-      "learning_rate": 0.0001999836902410419,
-      "loss": 46.0,
-      "step": 35710
-    },
-    {
-      "epoch": 5.7509964169249965,
-      "grad_norm": 0.0034173561725765467,
-      "learning_rate": 0.00019998368932734197,
-      "loss": 46.0,
-      "step": 35711
-    },
-    {
-      "epoch": 5.751157454003785,
-      "grad_norm": 0.0017595123499631882,
-      "learning_rate": 0.00019998368841361644,
-      "loss": 46.0,
-      "step": 35712
-    },
-    {
-      "epoch": 5.751318491082571,
-      "grad_norm": 0.003993146121501923,
-      "learning_rate": 0.00019998368749986532,
-      "loss": 46.0,
-      "step": 35713
-    },
-    {
-      "epoch": 5.751479528161359,
-      "grad_norm": 0.005458069033920765,
-      "learning_rate": 0.0001999836865860886,
-      "loss": 46.0,
-      "step": 35714
-    },
-    {
-      "epoch": 5.751640565240146,
-      "grad_norm": 0.004291501361876726,
-      "learning_rate": 0.0001999836856722863,
-      "loss": 46.0,
-      "step": 35715
-    },
-    {
-      "epoch": 5.751801602318934,
-      "grad_norm": 0.004495375324040651,
-      "learning_rate": 0.0001999836847584584,
-      "loss": 46.0,
-      "step": 35716
-    },
-    {
-      "epoch": 5.751962639397721,
-      "grad_norm": 0.004737691953778267,
-      "learning_rate": 0.0001999836838446049,
-      "loss": 46.0,
-      "step": 35717
-    },
-    {
-      "epoch": 5.752123676476509,
-      "grad_norm": 0.014758976176381111,
-      "learning_rate": 0.00019998368293072582,
-      "loss": 46.0,
-      "step": 35718
-    },
-    {
-      "epoch": 5.752284713555296,
-      "grad_norm": 0.0030611951369792223,
-      "learning_rate": 0.00019998368201682115,
-      "loss": 46.0,
-      "step": 35719
-    },
-    {
-      "epoch": 5.752445750634084,
-      "grad_norm": 0.0074820262379944324,
-      "learning_rate": 0.0001999836811028909,
-      "loss": 46.0,
-      "step": 35720
-    },
-    {
-      "epoch": 5.752606787712871,
-      "grad_norm": 0.016567546874284744,
-      "learning_rate": 0.00019998368018893503,
-      "loss": 46.0,
-      "step": 35721
-    },
-    {
-      "epoch": 5.7527678247916585,
-      "grad_norm": 0.002446531318128109,
-      "learning_rate": 0.0001999836792749536,
-      "loss": 46.0,
-      "step": 35722
-    },
-    {
-      "epoch": 5.752928861870446,
-      "grad_norm": 0.002018690574914217,
-      "learning_rate": 0.00019998367836094656,
-      "loss": 46.0,
-      "step": 35723
-    },
-    {
-      "epoch": 5.753089898949233,
-      "grad_norm": 0.004184822551906109,
-      "learning_rate": 0.00019998367744691395,
-      "loss": 46.0,
-      "step": 35724
-    },
-    {
-      "epoch": 5.753250936028021,
-      "grad_norm": 0.004262144677340984,
-      "learning_rate": 0.00019998367653285574,
-      "loss": 46.0,
-      "step": 35725
-    },
-    {
-      "epoch": 5.753411973106807,
-      "grad_norm": 0.009587960317730904,
-      "learning_rate": 0.00019998367561877193,
-      "loss": 46.0,
-      "step": 35726
-    },
-    {
-      "epoch": 5.753573010185595,
-      "grad_norm": 0.0008531140047125518,
-      "learning_rate": 0.00019998367470466251,
-      "loss": 46.0,
-      "step": 35727
-    },
-    {
-      "epoch": 5.753734047264382,
-      "grad_norm": 0.001791357877664268,
-      "learning_rate": 0.00019998367379052754,
-      "loss": 46.0,
-      "step": 35728
-    },
-    {
-      "epoch": 5.75389508434317,
-      "grad_norm": 0.0028573365416377783,
-      "learning_rate": 0.00019998367287636697,
-      "loss": 46.0,
-      "step": 35729
-    },
-    {
-      "epoch": 5.754056121421957,
-      "grad_norm": 0.005356810986995697,
-      "learning_rate": 0.00019998367196218082,
-      "loss": 46.0,
-      "step": 35730
-    },
-    {
-      "epoch": 5.754217158500745,
-      "grad_norm": 0.0017005581175908446,
-      "learning_rate": 0.00019998367104796905,
-      "loss": 46.0,
-      "step": 35731
-    },
-    {
-      "epoch": 5.754378195579532,
-      "grad_norm": 0.0020110388286411762,
-      "learning_rate": 0.0001999836701337317,
-      "loss": 46.0,
-      "step": 35732
-    },
-    {
-      "epoch": 5.75453923265832,
-      "grad_norm": 0.0012521251337602735,
-      "learning_rate": 0.00019998366921946875,
-      "loss": 46.0,
-      "step": 35733
-    },
-    {
-      "epoch": 5.754700269737107,
-      "grad_norm": 0.009996892884373665,
-      "learning_rate": 0.00019998366830518025,
-      "loss": 46.0,
-      "step": 35734
-    },
-    {
-      "epoch": 5.7548613068158945,
-      "grad_norm": 0.0013321025762706995,
-      "learning_rate": 0.0001999836673908661,
-      "loss": 46.0,
-      "step": 35735
-    },
-    {
-      "epoch": 5.755022343894682,
-      "grad_norm": 0.0023752159904688597,
-      "learning_rate": 0.0001999836664765264,
-      "loss": 46.0,
-      "step": 35736
-    },
-    {
-      "epoch": 5.755183380973469,
-      "grad_norm": 0.002858733059838414,
-      "learning_rate": 0.00019998366556216112,
-      "loss": 46.0,
-      "step": 35737
-    },
-    {
-      "epoch": 5.755344418052257,
-      "grad_norm": 0.0037748587783426046,
-      "learning_rate": 0.0001999836646477702,
-      "loss": 46.0,
-      "step": 35738
-    },
-    {
-      "epoch": 5.755505455131044,
-      "grad_norm": 0.004789280705153942,
-      "learning_rate": 0.00019998366373335372,
-      "loss": 46.0,
-      "step": 35739
-    },
-    {
-      "epoch": 5.755666492209832,
-      "grad_norm": 0.007376205641776323,
-      "learning_rate": 0.00019998366281891167,
-      "loss": 46.0,
-      "step": 35740
-    },
-    {
-      "epoch": 5.755827529288618,
-      "grad_norm": 0.005412856582552195,
-      "learning_rate": 0.000199983661904444,
-      "loss": 46.0,
-      "step": 35741
-    },
-    {
-      "epoch": 5.755988566367406,
-      "grad_norm": 0.005582737270742655,
-      "learning_rate": 0.00019998366098995075,
-      "loss": 46.0,
-      "step": 35742
-    },
-    {
-      "epoch": 5.756149603446193,
-      "grad_norm": 0.019184790551662445,
-      "learning_rate": 0.0001999836600754319,
-      "loss": 46.0,
-      "step": 35743
-    },
-    {
-      "epoch": 5.756310640524981,
-      "grad_norm": 0.008804554119706154,
-      "learning_rate": 0.00019998365916088745,
-      "loss": 46.0,
-      "step": 35744
-    },
-    {
-      "epoch": 5.756471677603768,
-      "grad_norm": 0.008403409272432327,
-      "learning_rate": 0.00019998365824631744,
-      "loss": 46.0,
-      "step": 35745
-    },
-    {
-      "epoch": 5.7566327146825556,
-      "grad_norm": 0.0019840106833726168,
-      "learning_rate": 0.0001999836573317218,
-      "loss": 46.0,
-      "step": 35746
-    },
-    {
-      "epoch": 5.756793751761343,
-      "grad_norm": 0.00988280214369297,
-      "learning_rate": 0.0001999836564171006,
-      "loss": 46.0,
-      "step": 35747
-    },
-    {
-      "epoch": 5.7569547888401305,
-      "grad_norm": 0.0018794741481542587,
-      "learning_rate": 0.00019998365550245382,
-      "loss": 46.0,
-      "step": 35748
-    },
-    {
-      "epoch": 5.757115825918918,
-      "grad_norm": 0.002330838004127145,
-      "learning_rate": 0.0001999836545877814,
-      "loss": 46.0,
-      "step": 35749
-    },
-    {
-      "epoch": 5.757276862997705,
-      "grad_norm": 0.01236631441861391,
-      "learning_rate": 0.00019998365367308342,
-      "loss": 46.0,
-      "step": 35750
-    },
-    {
-      "epoch": 5.757437900076493,
-      "grad_norm": 0.001956024905666709,
-      "learning_rate": 0.00019998365275835986,
-      "loss": 46.0,
-      "step": 35751
-    },
-    {
-      "epoch": 5.75759893715528,
-      "grad_norm": 0.006627036724239588,
-      "learning_rate": 0.0001999836518436107,
-      "loss": 46.0,
-      "step": 35752
-    },
-    {
-      "epoch": 5.757759974234068,
-      "grad_norm": 0.002990341978147626,
-      "learning_rate": 0.00019998365092883594,
-      "loss": 46.0,
-      "step": 35753
-    },
-    {
-      "epoch": 5.757921011312855,
-      "grad_norm": 0.006379584316164255,
-      "learning_rate": 0.0001999836500140356,
-      "loss": 46.0,
-      "step": 35754
-    },
-    {
-      "epoch": 5.758082048391643,
-      "grad_norm": 0.004898947663605213,
-      "learning_rate": 0.00019998364909920967,
-      "loss": 46.0,
-      "step": 35755
-    },
-    {
-      "epoch": 5.758243085470429,
-      "grad_norm": 0.007700498215854168,
-      "learning_rate": 0.00019998364818435814,
-      "loss": 46.0,
-      "step": 35756
-    },
-    {
-      "epoch": 5.758404122549217,
-      "grad_norm": 0.006034488324075937,
-      "learning_rate": 0.00019998364726948103,
-      "loss": 46.0,
-      "step": 35757
-    },
-    {
-      "epoch": 5.758565159628004,
-      "grad_norm": 0.00562693364918232,
-      "learning_rate": 0.00019998364635457833,
-      "loss": 46.0,
-      "step": 35758
-    },
-    {
-      "epoch": 5.7587261967067915,
-      "grad_norm": 0.003450853517279029,
-      "learning_rate": 0.00019998364543965004,
-      "loss": 46.0,
-      "step": 35759
-    },
-    {
-      "epoch": 5.758887233785579,
-      "grad_norm": 0.0028900671750307083,
-      "learning_rate": 0.00019998364452469613,
-      "loss": 46.0,
-      "step": 35760
-    },
-    {
-      "epoch": 5.759048270864366,
-      "grad_norm": 0.009149179793894291,
-      "learning_rate": 0.00019998364360971667,
-      "loss": 46.0,
-      "step": 35761
-    },
-    {
-      "epoch": 5.759209307943154,
-      "grad_norm": 0.0018029846251010895,
-      "learning_rate": 0.0001999836426947116,
-      "loss": 46.0,
-      "step": 35762
-    },
-    {
-      "epoch": 5.759370345021941,
-      "grad_norm": 0.0027878256514668465,
-      "learning_rate": 0.00019998364177968095,
-      "loss": 46.0,
-      "step": 35763
-    },
-    {
-      "epoch": 5.759531382100729,
-      "grad_norm": 0.0033499824348837137,
-      "learning_rate": 0.0001999836408646247,
-      "loss": 46.0,
-      "step": 35764
-    },
-    {
-      "epoch": 5.759692419179516,
-      "grad_norm": 0.003248290391638875,
-      "learning_rate": 0.00019998363994954286,
-      "loss": 46.0,
-      "step": 35765
-    },
-    {
-      "epoch": 5.759853456258304,
-      "grad_norm": 0.004588032606989145,
-      "learning_rate": 0.0001999836390344354,
-      "loss": 46.0,
-      "step": 35766
-    },
-    {
-      "epoch": 5.760014493337091,
-      "grad_norm": 0.0025692398194223642,
-      "learning_rate": 0.0001999836381193024,
-      "loss": 46.0,
-      "step": 35767
-    },
-    {
-      "epoch": 5.760175530415879,
-      "grad_norm": 0.0012325189309194684,
-      "learning_rate": 0.00019998363720414376,
-      "loss": 46.0,
-      "step": 35768
-    },
-    {
-      "epoch": 5.760336567494665,
-      "grad_norm": 0.007198629900813103,
-      "learning_rate": 0.00019998363628895958,
-      "loss": 46.0,
-      "step": 35769
-    },
-    {
-      "epoch": 5.7604976045734535,
-      "grad_norm": 0.005350594874471426,
-      "learning_rate": 0.00019998363537374977,
-      "loss": 46.0,
-      "step": 35770
-    },
-    {
-      "epoch": 5.76065864165224,
-      "grad_norm": 0.008711155503988266,
-      "learning_rate": 0.00019998363445851438,
-      "loss": 46.0,
-      "step": 35771
-    },
-    {
-      "epoch": 5.7608196787310275,
-      "grad_norm": 0.002675512572750449,
-      "learning_rate": 0.0001999836335432534,
-      "loss": 46.0,
-      "step": 35772
-    },
-    {
-      "epoch": 5.760980715809815,
-      "grad_norm": 0.014440230093896389,
-      "learning_rate": 0.00019998363262796684,
-      "loss": 46.0,
-      "step": 35773
-    },
-    {
-      "epoch": 5.761141752888602,
-      "grad_norm": 0.0012159671168774366,
-      "learning_rate": 0.00019998363171265472,
-      "loss": 46.0,
-      "step": 35774
-    },
-    {
-      "epoch": 5.76130278996739,
-      "grad_norm": 0.0015782787231728435,
-      "learning_rate": 0.00019998363079731695,
-      "loss": 46.0,
-      "step": 35775
-    },
-    {
-      "epoch": 5.761463827046177,
-      "grad_norm": 0.001120796543546021,
-      "learning_rate": 0.0001999836298819536,
-      "loss": 46.0,
-      "step": 35776
-    },
-    {
-      "epoch": 5.761624864124965,
-      "grad_norm": 0.010463908314704895,
-      "learning_rate": 0.0001999836289665647,
-      "loss": 46.0,
-      "step": 35777
-    },
-    {
-      "epoch": 5.761785901203752,
-      "grad_norm": 0.0022101362701505423,
-      "learning_rate": 0.00019998362805115016,
-      "loss": 46.0,
-      "step": 35778
-    },
-    {
-      "epoch": 5.76194693828254,
-      "grad_norm": 0.0038292030803859234,
-      "learning_rate": 0.00019998362713571005,
-      "loss": 46.0,
-      "step": 35779
-    },
-    {
-      "epoch": 5.762107975361327,
-      "grad_norm": 0.010869057849049568,
-      "learning_rate": 0.00019998362622024435,
-      "loss": 46.0,
-      "step": 35780
-    },
-    {
-      "epoch": 5.762269012440115,
-      "grad_norm": 0.002579993335530162,
-      "learning_rate": 0.00019998362530475306,
-      "loss": 46.0,
-      "step": 35781
-    },
-    {
-      "epoch": 5.762430049518902,
-      "grad_norm": 0.012837952934205532,
-      "learning_rate": 0.00019998362438923616,
-      "loss": 46.0,
-      "step": 35782
-    },
-    {
-      "epoch": 5.7625910865976895,
-      "grad_norm": 0.006787641905248165,
-      "learning_rate": 0.0001999836234736937,
-      "loss": 46.0,
-      "step": 35783
-    },
-    {
-      "epoch": 5.762752123676476,
-      "grad_norm": 0.018284711986780167,
-      "learning_rate": 0.00019998362255812564,
-      "loss": 46.0,
-      "step": 35784
-    },
-    {
-      "epoch": 5.762913160755264,
-      "grad_norm": 0.009029623121023178,
-      "learning_rate": 0.00019998362164253198,
-      "loss": 46.0,
-      "step": 35785
-    },
-    {
-      "epoch": 5.763074197834051,
-      "grad_norm": 0.004047457128763199,
-      "learning_rate": 0.00019998362072691275,
-      "loss": 46.0,
-      "step": 35786
-    },
-    {
-      "epoch": 5.763235234912838,
-      "grad_norm": 0.0036357128992676735,
-      "learning_rate": 0.0001999836198112679,
-      "loss": 46.0,
-      "step": 35787
-    },
-    {
-      "epoch": 5.763396271991626,
-      "grad_norm": 0.012972780503332615,
-      "learning_rate": 0.00019998361889559746,
-      "loss": 46.0,
-      "step": 35788
-    },
-    {
-      "epoch": 5.763557309070413,
-      "grad_norm": 0.0011942959390580654,
-      "learning_rate": 0.00019998361797990145,
-      "loss": 46.0,
-      "step": 35789
-    },
-    {
-      "epoch": 5.763718346149201,
-      "grad_norm": 0.010222087614238262,
-      "learning_rate": 0.00019998361706417985,
-      "loss": 46.0,
-      "step": 35790
-    },
-    {
-      "epoch": 5.763879383227988,
-      "grad_norm": 0.0014351614518091083,
-      "learning_rate": 0.00019998361614843266,
-      "loss": 46.0,
-      "step": 35791
-    },
-    {
-      "epoch": 5.764040420306776,
-      "grad_norm": 0.03153061121702194,
-      "learning_rate": 0.00019998361523265986,
-      "loss": 46.0,
-      "step": 35792
-    },
-    {
-      "epoch": 5.764201457385563,
-      "grad_norm": 0.01333332434296608,
-      "learning_rate": 0.00019998361431686147,
-      "loss": 46.0,
-      "step": 35793
-    },
-    {
-      "epoch": 5.764362494464351,
-      "grad_norm": 0.0015930727822706103,
-      "learning_rate": 0.0001999836134010375,
-      "loss": 46.0,
-      "step": 35794
-    },
-    {
-      "epoch": 5.764523531543138,
-      "grad_norm": 0.00261515355668962,
-      "learning_rate": 0.00019998361248518796,
-      "loss": 46.0,
-      "step": 35795
-    },
-    {
-      "epoch": 5.7646845686219255,
-      "grad_norm": 0.003101665060967207,
-      "learning_rate": 0.0001999836115693128,
-      "loss": 46.0,
-      "step": 35796
-    },
-    {
-      "epoch": 5.764845605700713,
-      "grad_norm": 0.002982706530019641,
-      "learning_rate": 0.00019998361065341205,
-      "loss": 46.0,
-      "step": 35797
-    },
-    {
-      "epoch": 5.7650066427795,
-      "grad_norm": 0.003136037616059184,
-      "learning_rate": 0.00019998360973748572,
-      "loss": 46.0,
-      "step": 35798
-    },
-    {
-      "epoch": 5.765167679858287,
-      "grad_norm": 0.009190799668431282,
-      "learning_rate": 0.0001999836088215338,
-      "loss": 46.0,
-      "step": 35799
-    },
-    {
-      "epoch": 5.765328716937074,
-      "grad_norm": 0.00977477990090847,
-      "learning_rate": 0.00019998360790555628,
-      "loss": 46.0,
-      "step": 35800
-    },
-    {
-      "epoch": 5.765489754015862,
-      "grad_norm": 0.0027433226350694895,
-      "learning_rate": 0.0001999836069895532,
-      "loss": 46.0,
-      "step": 35801
-    },
-    {
-      "epoch": 5.765650791094649,
-      "grad_norm": 0.00669348007068038,
-      "learning_rate": 0.0001999836060735245,
-      "loss": 46.0,
-      "step": 35802
-    },
-    {
-      "epoch": 5.765811828173437,
-      "grad_norm": 0.007097512949258089,
-      "learning_rate": 0.0001999836051574702,
-      "loss": 46.0,
-      "step": 35803
-    },
-    {
-      "epoch": 5.765972865252224,
-      "grad_norm": 0.0016350100049749017,
-      "learning_rate": 0.00019998360424139033,
-      "loss": 46.0,
-      "step": 35804
-    },
-    {
-      "epoch": 5.766133902331012,
-      "grad_norm": 0.021562431007623672,
-      "learning_rate": 0.00019998360332528485,
-      "loss": 46.0,
-      "step": 35805
-    },
-    {
-      "epoch": 5.766294939409799,
-      "grad_norm": 0.012011711485683918,
-      "learning_rate": 0.0001999836024091538,
-      "loss": 46.0,
-      "step": 35806
-    },
-    {
-      "epoch": 5.7664559764885865,
-      "grad_norm": 0.017741721123456955,
-      "learning_rate": 0.00019998360149299713,
-      "loss": 46.0,
-      "step": 35807
-    },
-    {
-      "epoch": 5.766617013567374,
-      "grad_norm": 0.004392742179334164,
-      "learning_rate": 0.00019998360057681488,
-      "loss": 46.0,
-      "step": 35808
-    },
-    {
-      "epoch": 5.7667780506461614,
-      "grad_norm": 0.01222624909132719,
-      "learning_rate": 0.00019998359966060707,
-      "loss": 46.0,
-      "step": 35809
-    },
-    {
-      "epoch": 5.766939087724949,
-      "grad_norm": 0.009547607973217964,
-      "learning_rate": 0.00019998359874437367,
-      "loss": 46.0,
-      "step": 35810
-    },
-    {
-      "epoch": 5.767100124803736,
-      "grad_norm": 0.0027848128229379654,
-      "learning_rate": 0.00019998359782811463,
-      "loss": 46.0,
-      "step": 35811
-    },
-    {
-      "epoch": 5.767261161882524,
-      "grad_norm": 0.0015407265163958073,
-      "learning_rate": 0.00019998359691183004,
-      "loss": 46.0,
-      "step": 35812
-    },
-    {
-      "epoch": 5.767422198961311,
-      "grad_norm": 0.002375328214839101,
-      "learning_rate": 0.00019998359599551985,
-      "loss": 46.0,
-      "step": 35813
-    },
-    {
-      "epoch": 5.767583236040098,
-      "grad_norm": 0.003431180026382208,
-      "learning_rate": 0.00019998359507918405,
-      "loss": 46.0,
-      "step": 35814
-    },
-    {
-      "epoch": 5.767744273118885,
-      "grad_norm": 0.007888766936957836,
-      "learning_rate": 0.0001999835941628227,
-      "loss": 46.0,
-      "step": 35815
-    },
-    {
-      "epoch": 5.767905310197673,
-      "grad_norm": 0.008914914913475513,
-      "learning_rate": 0.0001999835932464357,
-      "loss": 46.0,
-      "step": 35816
-    },
-    {
-      "epoch": 5.76806634727646,
-      "grad_norm": 0.002494643209502101,
-      "learning_rate": 0.00019998359233002315,
-      "loss": 46.0,
-      "step": 35817
-    },
-    {
-      "epoch": 5.768227384355248,
-      "grad_norm": 0.004079343751072884,
-      "learning_rate": 0.00019998359141358503,
-      "loss": 46.0,
-      "step": 35818
-    },
-    {
-      "epoch": 5.768388421434035,
-      "grad_norm": 0.0044632237404584885,
-      "learning_rate": 0.00019998359049712127,
-      "loss": 46.0,
-      "step": 35819
-    },
-    {
-      "epoch": 5.7685494585128225,
-      "grad_norm": 0.007729186676442623,
-      "learning_rate": 0.00019998358958063194,
-      "loss": 46.0,
-      "step": 35820
-    },
-    {
-      "epoch": 5.76871049559161,
-      "grad_norm": 0.0020596070680767298,
-      "learning_rate": 0.00019998358866411703,
-      "loss": 46.0,
-      "step": 35821
-    },
-    {
-      "epoch": 5.768871532670397,
-      "grad_norm": 0.0034962501376867294,
-      "learning_rate": 0.00019998358774757653,
-      "loss": 46.0,
-      "step": 35822
-    },
-    {
-      "epoch": 5.769032569749185,
-      "grad_norm": 0.0007968193385750055,
-      "learning_rate": 0.00019998358683101045,
-      "loss": 46.0,
-      "step": 35823
-    },
-    {
-      "epoch": 5.769193606827972,
-      "grad_norm": 0.0038508260622620583,
-      "learning_rate": 0.00019998358591441875,
-      "loss": 46.0,
-      "step": 35824
-    },
-    {
-      "epoch": 5.76935464390676,
-      "grad_norm": 0.0031962452922016382,
-      "learning_rate": 0.00019998358499780146,
-      "loss": 46.0,
-      "step": 35825
-    },
-    {
-      "epoch": 5.769515680985547,
-      "grad_norm": 0.009239432401955128,
-      "learning_rate": 0.0001999835840811586,
-      "loss": 46.0,
-      "step": 35826
-    },
-    {
-      "epoch": 5.769676718064335,
-      "grad_norm": 0.002511995378881693,
-      "learning_rate": 0.0001999835831644901,
-      "loss": 46.0,
-      "step": 35827
-    },
-    {
-      "epoch": 5.769837755143122,
-      "grad_norm": 0.001880061929114163,
-      "learning_rate": 0.00019998358224779605,
-      "loss": 46.0,
-      "step": 35828
-    },
-    {
-      "epoch": 5.769998792221909,
-      "grad_norm": 0.011502335779368877,
-      "learning_rate": 0.00019998358133107642,
-      "loss": 46.0,
-      "step": 35829
-    },
-    {
-      "epoch": 5.770159829300696,
-      "grad_norm": 0.0020040462259203196,
-      "learning_rate": 0.0001999835804143312,
-      "loss": 46.0,
-      "step": 35830
-    },
-    {
-      "epoch": 5.770320866379484,
-      "grad_norm": 0.0027627963572740555,
-      "learning_rate": 0.00019998357949756036,
-      "loss": 46.0,
-      "step": 35831
-    },
-    {
-      "epoch": 5.770481903458271,
-      "grad_norm": 0.0052684820257127285,
-      "learning_rate": 0.00019998357858076396,
-      "loss": 46.0,
-      "step": 35832
-    },
-    {
-      "epoch": 5.7706429405370585,
-      "grad_norm": 0.00824128556996584,
-      "learning_rate": 0.00019998357766394195,
-      "loss": 46.0,
-      "step": 35833
-    },
-    {
-      "epoch": 5.770803977615846,
-      "grad_norm": 0.0081133171916008,
-      "learning_rate": 0.00019998357674709432,
-      "loss": 46.0,
-      "step": 35834
-    },
-    {
-      "epoch": 5.770965014694633,
-      "grad_norm": 0.00839622039347887,
-      "learning_rate": 0.00019998357583022117,
-      "loss": 46.0,
-      "step": 35835
-    },
-    {
-      "epoch": 5.771126051773421,
-      "grad_norm": 0.003653403138741851,
-      "learning_rate": 0.0001999835749133224,
-      "loss": 46.0,
-      "step": 35836
-    },
-    {
-      "epoch": 5.771287088852208,
-      "grad_norm": 0.007322989404201508,
-      "learning_rate": 0.000199983573996398,
-      "loss": 46.0,
-      "step": 35837
-    },
-    {
-      "epoch": 5.771448125930996,
-      "grad_norm": 0.010368783958256245,
-      "learning_rate": 0.00019998357307944806,
-      "loss": 46.0,
-      "step": 35838
-    },
-    {
-      "epoch": 5.771609163009783,
-      "grad_norm": 0.0215225238353014,
-      "learning_rate": 0.0001999835721624725,
-      "loss": 46.0,
-      "step": 35839
-    },
-    {
-      "epoch": 5.771770200088571,
-      "grad_norm": 0.01646846905350685,
-      "learning_rate": 0.00019998357124547135,
-      "loss": 46.0,
-      "step": 35840
-    },
-    {
-      "epoch": 5.771931237167358,
-      "grad_norm": 0.00138341065030545,
-      "learning_rate": 0.00019998357032844464,
-      "loss": 46.0,
-      "step": 35841
-    },
-    {
-      "epoch": 5.772092274246145,
-      "grad_norm": 0.005273793358355761,
-      "learning_rate": 0.0001999835694113923,
-      "loss": 46.0,
-      "step": 35842
-    },
-    {
-      "epoch": 5.772253311324933,
-      "grad_norm": 0.0074575068429112434,
-      "learning_rate": 0.00019998356849431438,
-      "loss": 46.0,
-      "step": 35843
-    },
-    {
-      "epoch": 5.77241434840372,
-      "grad_norm": 0.008774465881288052,
-      "learning_rate": 0.00019998356757721088,
-      "loss": 46.0,
-      "step": 35844
-    },
-    {
-      "epoch": 5.772575385482507,
-      "grad_norm": 0.001621044473722577,
-      "learning_rate": 0.00019998356666008177,
-      "loss": 46.0,
-      "step": 35845
-    },
-    {
-      "epoch": 5.7727364225612945,
-      "grad_norm": 0.004730668850243092,
-      "learning_rate": 0.00019998356574292713,
-      "loss": 46.0,
-      "step": 35846
-    },
-    {
-      "epoch": 5.772897459640082,
-      "grad_norm": 0.0032650940120220184,
-      "learning_rate": 0.00019998356482574682,
-      "loss": 46.0,
-      "step": 35847
-    },
-    {
-      "epoch": 5.773058496718869,
-      "grad_norm": 0.004273801576346159,
-      "learning_rate": 0.00019998356390854097,
-      "loss": 46.0,
-      "step": 35848
-    },
-    {
-      "epoch": 5.773219533797657,
-      "grad_norm": 0.007242194842547178,
-      "learning_rate": 0.0001999835629913095,
-      "loss": 46.0,
-      "step": 35849
-    },
-    {
-      "epoch": 5.773380570876444,
-      "grad_norm": 0.002563401823863387,
-      "learning_rate": 0.00019998356207405243,
-      "loss": 46.0,
-      "step": 35850
-    },
-    {
-      "epoch": 5.773541607955232,
-      "grad_norm": 0.001910101855173707,
-      "learning_rate": 0.00019998356115676983,
-      "loss": 46.0,
-      "step": 35851
-    },
-    {
-      "epoch": 5.773702645034019,
-      "grad_norm": 0.003323824843391776,
-      "learning_rate": 0.00019998356023946158,
-      "loss": 46.0,
-      "step": 35852
-    },
-    {
-      "epoch": 5.773863682112807,
-      "grad_norm": 0.01273073349148035,
-      "learning_rate": 0.00019998355932212774,
-      "loss": 46.0,
-      "step": 35853
-    },
-    {
-      "epoch": 5.774024719191594,
-      "grad_norm": 0.007262049242854118,
-      "learning_rate": 0.00019998355840476837,
-      "loss": 46.0,
-      "step": 35854
-    },
-    {
-      "epoch": 5.7741857562703816,
-      "grad_norm": 0.0019591215532273054,
-      "learning_rate": 0.00019998355748738336,
-      "loss": 46.0,
-      "step": 35855
-    },
-    {
-      "epoch": 5.774346793349169,
-      "grad_norm": 0.001541677862405777,
-      "learning_rate": 0.00019998355656997277,
-      "loss": 46.0,
-      "step": 35856
-    },
-    {
-      "epoch": 5.774507830427956,
-      "grad_norm": 0.003725646995007992,
-      "learning_rate": 0.00019998355565253658,
-      "loss": 46.0,
-      "step": 35857
-    },
-    {
-      "epoch": 5.774668867506744,
-      "grad_norm": 0.02196687087416649,
-      "learning_rate": 0.00019998355473507478,
-      "loss": 46.0,
-      "step": 35858
-    },
-    {
-      "epoch": 5.7748299045855305,
-      "grad_norm": 0.004947022069245577,
-      "learning_rate": 0.00019998355381758745,
-      "loss": 46.0,
-      "step": 35859
-    },
-    {
-      "epoch": 5.774990941664318,
-      "grad_norm": 0.004878837149590254,
-      "learning_rate": 0.00019998355290007448,
-      "loss": 46.0,
-      "step": 35860
-    },
-    {
-      "epoch": 5.775151978743105,
-      "grad_norm": 0.0014319252222776413,
-      "learning_rate": 0.00019998355198253595,
-      "loss": 46.0,
-      "step": 35861
-    },
-    {
-      "epoch": 5.775313015821893,
-      "grad_norm": 0.0016407968942075968,
-      "learning_rate": 0.00019998355106497182,
-      "loss": 46.0,
-      "step": 35862
-    },
-    {
-      "epoch": 5.77547405290068,
-      "grad_norm": 0.00904154498130083,
-      "learning_rate": 0.0001999835501473821,
-      "loss": 46.0,
-      "step": 35863
-    },
-    {
-      "epoch": 5.775635089979468,
-      "grad_norm": 0.002735934918746352,
-      "learning_rate": 0.00019998354922976677,
-      "loss": 46.0,
-      "step": 35864
-    },
-    {
-      "epoch": 5.775796127058255,
-      "grad_norm": 0.0019785065669566393,
-      "learning_rate": 0.00019998354831212586,
-      "loss": 46.0,
-      "step": 35865
-    },
-    {
-      "epoch": 5.775957164137043,
-      "grad_norm": 0.0018252504523843527,
-      "learning_rate": 0.00019998354739445934,
-      "loss": 46.0,
-      "step": 35866
-    },
-    {
-      "epoch": 5.77611820121583,
-      "grad_norm": 0.0024098071735352278,
-      "learning_rate": 0.00019998354647676728,
-      "loss": 46.0,
-      "step": 35867
-    },
-    {
-      "epoch": 5.7762792382946175,
-      "grad_norm": 0.0038775037974119186,
-      "learning_rate": 0.00019998354555904958,
-      "loss": 46.0,
-      "step": 35868
-    },
-    {
-      "epoch": 5.776440275373405,
-      "grad_norm": 0.003925892990082502,
-      "learning_rate": 0.0001999835446413063,
-      "loss": 46.0,
-      "step": 35869
-    },
-    {
-      "epoch": 5.776601312452192,
-      "grad_norm": 0.007865316234529018,
-      "learning_rate": 0.00019998354372353748,
-      "loss": 46.0,
-      "step": 35870
-    },
-    {
-      "epoch": 5.77676234953098,
-      "grad_norm": 0.0033923930022865534,
-      "learning_rate": 0.00019998354280574302,
-      "loss": 46.0,
-      "step": 35871
-    },
-    {
-      "epoch": 5.7769233866097665,
-      "grad_norm": 0.002170517109334469,
-      "learning_rate": 0.00019998354188792297,
-      "loss": 46.0,
-      "step": 35872
-    },
-    {
-      "epoch": 5.777084423688554,
-      "grad_norm": 0.002143422607332468,
-      "learning_rate": 0.00019998354097007734,
-      "loss": 46.0,
-      "step": 35873
-    },
-    {
-      "epoch": 5.777245460767341,
-      "grad_norm": 0.00580269331112504,
-      "learning_rate": 0.0001999835400522061,
-      "loss": 46.0,
-      "step": 35874
-    },
-    {
-      "epoch": 5.777406497846129,
-      "grad_norm": 0.013656793162226677,
-      "learning_rate": 0.00019998353913430929,
-      "loss": 46.0,
-      "step": 35875
-    },
-    {
-      "epoch": 5.777567534924916,
-      "grad_norm": 0.003275335766375065,
-      "learning_rate": 0.0001999835382163869,
-      "loss": 46.0,
-      "step": 35876
-    },
-    {
-      "epoch": 5.777728572003704,
-      "grad_norm": 0.0021338691003620625,
-      "learning_rate": 0.00019998353729843888,
-      "loss": 46.0,
-      "step": 35877
-    },
-    {
-      "epoch": 5.777889609082491,
-      "grad_norm": 0.0022901701740920544,
-      "learning_rate": 0.0001999835363804653,
-      "loss": 46.0,
-      "step": 35878
-    },
-    {
-      "epoch": 5.778050646161279,
-      "grad_norm": 0.0018274775939062238,
-      "learning_rate": 0.00019998353546246613,
-      "loss": 46.0,
-      "step": 35879
-    },
-    {
-      "epoch": 5.778211683240066,
-      "grad_norm": 0.0016587969148531556,
-      "learning_rate": 0.00019998353454444136,
-      "loss": 46.0,
-      "step": 35880
-    },
-    {
-      "epoch": 5.7783727203188535,
-      "grad_norm": 0.003965029958635569,
-      "learning_rate": 0.00019998353362639103,
-      "loss": 46.0,
-      "step": 35881
-    },
-    {
-      "epoch": 5.778533757397641,
-      "grad_norm": 0.005459026899188757,
-      "learning_rate": 0.00019998353270831505,
-      "loss": 46.0,
-      "step": 35882
-    },
-    {
-      "epoch": 5.778694794476428,
-      "grad_norm": 0.0031608333811163902,
-      "learning_rate": 0.00019998353179021352,
-      "loss": 46.0,
-      "step": 35883
-    },
-    {
-      "epoch": 5.778855831555216,
-      "grad_norm": 0.0010634392965584993,
-      "learning_rate": 0.00019998353087208638,
-      "loss": 46.0,
-      "step": 35884
-    },
-    {
-      "epoch": 5.779016868634003,
-      "grad_norm": 0.0027363006956875324,
-      "learning_rate": 0.00019998352995393367,
-      "loss": 46.0,
-      "step": 35885
-    },
-    {
-      "epoch": 5.779177905712791,
-      "grad_norm": 0.00758574390783906,
-      "learning_rate": 0.00019998352903575537,
-      "loss": 46.0,
-      "step": 35886
-    },
-    {
-      "epoch": 5.779338942791577,
-      "grad_norm": 0.002652771072462201,
-      "learning_rate": 0.00019998352811755147,
-      "loss": 46.0,
-      "step": 35887
-    },
-    {
-      "epoch": 5.779499979870365,
-      "grad_norm": 0.008481164462864399,
-      "learning_rate": 0.00019998352719932197,
-      "loss": 46.0,
-      "step": 35888
-    },
-    {
-      "epoch": 5.779661016949152,
-      "grad_norm": 0.0027792537584900856,
-      "learning_rate": 0.0001999835262810669,
-      "loss": 46.0,
-      "step": 35889
-    },
-    {
-      "epoch": 5.77982205402794,
-      "grad_norm": 0.033703941851854324,
-      "learning_rate": 0.0001999835253627862,
-      "loss": 46.0,
-      "step": 35890
-    },
-    {
-      "epoch": 5.779983091106727,
-      "grad_norm": 0.008566014468669891,
-      "learning_rate": 0.0001999835244444799,
-      "loss": 46.0,
-      "step": 35891
-    },
-    {
-      "epoch": 5.780144128185515,
-      "grad_norm": 0.0037682177498936653,
-      "learning_rate": 0.00019998352352614806,
-      "loss": 46.0,
-      "step": 35892
-    },
-    {
-      "epoch": 5.780305165264302,
-      "grad_norm": 0.002675719792023301,
-      "learning_rate": 0.00019998352260779063,
-      "loss": 46.0,
-      "step": 35893
-    },
-    {
-      "epoch": 5.7804662023430895,
-      "grad_norm": 0.013582653366029263,
-      "learning_rate": 0.0001999835216894076,
-      "loss": 46.0,
-      "step": 35894
-    },
-    {
-      "epoch": 5.780627239421877,
-      "grad_norm": 0.004690753761678934,
-      "learning_rate": 0.00019998352077099898,
-      "loss": 46.0,
-      "step": 35895
-    },
-    {
-      "epoch": 5.780788276500664,
-      "grad_norm": 0.004897431004792452,
-      "learning_rate": 0.00019998351985256476,
-      "loss": 46.0,
-      "step": 35896
-    },
-    {
-      "epoch": 5.780949313579452,
-      "grad_norm": 0.002507381374016404,
-      "learning_rate": 0.00019998351893410493,
-      "loss": 46.0,
-      "step": 35897
-    },
-    {
-      "epoch": 5.781110350658239,
-      "grad_norm": 0.0008928921306505799,
-      "learning_rate": 0.00019998351801561953,
-      "loss": 46.0,
-      "step": 35898
-    },
-    {
-      "epoch": 5.781271387737027,
-      "grad_norm": 0.037514083087444305,
-      "learning_rate": 0.00019998351709710855,
-      "loss": 46.0,
-      "step": 35899
-    },
-    {
-      "epoch": 5.781432424815814,
-      "grad_norm": 0.002529543824493885,
-      "learning_rate": 0.00019998351617857196,
-      "loss": 46.0,
-      "step": 35900
-    },
-    {
-      "epoch": 5.781593461894602,
-      "grad_norm": 0.017014116048812866,
-      "learning_rate": 0.0001999835152600098,
-      "loss": 46.0,
-      "step": 35901
-    },
-    {
-      "epoch": 5.781754498973388,
-      "grad_norm": 0.002385465195402503,
-      "learning_rate": 0.00019998351434142203,
-      "loss": 46.0,
-      "step": 35902
-    },
-    {
-      "epoch": 5.781915536052176,
-      "grad_norm": 0.00308678369037807,
-      "learning_rate": 0.0001999835134228087,
-      "loss": 46.0,
-      "step": 35903
-    },
-    {
-      "epoch": 5.782076573130963,
-      "grad_norm": 0.003813399001955986,
-      "learning_rate": 0.00019998351250416976,
-      "loss": 46.0,
-      "step": 35904
-    },
-    {
-      "epoch": 5.782237610209751,
-      "grad_norm": 0.011835162527859211,
-      "learning_rate": 0.0001999835115855052,
-      "loss": 46.0,
-      "step": 35905
-    },
-    {
-      "epoch": 5.782398647288538,
-      "grad_norm": 0.004578090272843838,
-      "learning_rate": 0.00019998351066681508,
-      "loss": 46.0,
-      "step": 35906
-    },
-    {
-      "epoch": 5.7825596843673255,
-      "grad_norm": 0.007764613721519709,
-      "learning_rate": 0.00019998350974809938,
-      "loss": 46.0,
-      "step": 35907
-    },
-    {
-      "epoch": 5.782720721446113,
-      "grad_norm": 0.002877760212868452,
-      "learning_rate": 0.00019998350882935806,
-      "loss": 46.0,
-      "step": 35908
-    },
-    {
-      "epoch": 5.7828817585249,
-      "grad_norm": 0.002632005140185356,
-      "learning_rate": 0.00019998350791059118,
-      "loss": 46.0,
-      "step": 35909
-    },
-    {
-      "epoch": 5.783042795603688,
-      "grad_norm": 0.0037167989648878574,
-      "learning_rate": 0.00019998350699179868,
-      "loss": 46.0,
-      "step": 35910
-    },
-    {
-      "epoch": 5.783203832682475,
-      "grad_norm": 0.003435267833992839,
-      "learning_rate": 0.0001999835060729806,
-      "loss": 46.0,
-      "step": 35911
-    },
-    {
-      "epoch": 5.783364869761263,
-      "grad_norm": 0.005022314842790365,
-      "learning_rate": 0.00019998350515413693,
-      "loss": 46.0,
-      "step": 35912
-    },
-    {
-      "epoch": 5.78352590684005,
-      "grad_norm": 0.008201629854738712,
-      "learning_rate": 0.00019998350423526765,
-      "loss": 46.0,
-      "step": 35913
-    },
-    {
-      "epoch": 5.783686943918838,
-      "grad_norm": 0.001541821868158877,
-      "learning_rate": 0.00019998350331637283,
-      "loss": 46.0,
-      "step": 35914
-    },
-    {
-      "epoch": 5.783847980997624,
-      "grad_norm": 0.006183285266160965,
-      "learning_rate": 0.00019998350239745237,
-      "loss": 46.0,
-      "step": 35915
-    },
-    {
-      "epoch": 5.7840090180764125,
-      "grad_norm": 0.018910681828856468,
-      "learning_rate": 0.00019998350147850636,
-      "loss": 46.0,
-      "step": 35916
-    },
-    {
-      "epoch": 5.784170055155199,
-      "grad_norm": 0.00981220230460167,
-      "learning_rate": 0.00019998350055953475,
-      "loss": 46.0,
-      "step": 35917
-    },
-    {
-      "epoch": 5.784331092233987,
-      "grad_norm": 0.006154462695121765,
-      "learning_rate": 0.0001999834996405375,
-      "loss": 46.0,
-      "step": 35918
-    },
-    {
-      "epoch": 5.784492129312774,
-      "grad_norm": 0.019950637593865395,
-      "learning_rate": 0.00019998349872151473,
-      "loss": 46.0,
-      "step": 35919
-    },
-    {
-      "epoch": 5.7846531663915615,
-      "grad_norm": 0.006219244562089443,
-      "learning_rate": 0.00019998349780246633,
-      "loss": 46.0,
-      "step": 35920
-    },
-    {
-      "epoch": 5.784814203470349,
-      "grad_norm": 0.0019006497459486127,
-      "learning_rate": 0.00019998349688339233,
-      "loss": 46.0,
-      "step": 35921
-    },
-    {
-      "epoch": 5.784975240549136,
-      "grad_norm": 0.0022500548511743546,
-      "learning_rate": 0.0001999834959642928,
-      "loss": 46.0,
-      "step": 35922
-    },
-    {
-      "epoch": 5.785136277627924,
-      "grad_norm": 0.006770461797714233,
-      "learning_rate": 0.0001999834950451676,
-      "loss": 46.0,
-      "step": 35923
-    },
-    {
-      "epoch": 5.785297314706711,
-      "grad_norm": 0.0028491022530943155,
-      "learning_rate": 0.00019998349412601684,
-      "loss": 46.0,
-      "step": 35924
-    },
-    {
-      "epoch": 5.785458351785499,
-      "grad_norm": 0.001747211441397667,
-      "learning_rate": 0.0001999834932068405,
-      "loss": 46.0,
-      "step": 35925
-    },
-    {
-      "epoch": 5.785619388864286,
-      "grad_norm": 0.000976066745352,
-      "learning_rate": 0.00019998349228763856,
-      "loss": 46.0,
-      "step": 35926
-    },
-    {
-      "epoch": 5.785780425943074,
-      "grad_norm": 0.004377948585897684,
-      "learning_rate": 0.000199983491368411,
-      "loss": 46.0,
-      "step": 35927
-    },
-    {
-      "epoch": 5.785941463021861,
-      "grad_norm": 0.0019161331001669168,
-      "learning_rate": 0.0001999834904491579,
-      "loss": 46.0,
-      "step": 35928
-    },
-    {
-      "epoch": 5.7861025001006485,
-      "grad_norm": 0.01532067358493805,
-      "learning_rate": 0.0001999834895298792,
-      "loss": 46.0,
-      "step": 35929
-    },
-    {
-      "epoch": 5.786263537179435,
-      "grad_norm": 0.0036150391679257154,
-      "learning_rate": 0.0001999834886105749,
-      "loss": 46.0,
-      "step": 35930
-    },
-    {
-      "epoch": 5.786424574258223,
-      "grad_norm": 0.0017206162447109818,
-      "learning_rate": 0.000199983487691245,
-      "loss": 46.0,
-      "step": 35931
-    },
-    {
-      "epoch": 5.78658561133701,
-      "grad_norm": 0.0028035107534378767,
-      "learning_rate": 0.00019998348677188955,
-      "loss": 46.0,
-      "step": 35932
-    },
-    {
-      "epoch": 5.7867466484157974,
-      "grad_norm": 0.0009451525984331965,
-      "learning_rate": 0.00019998348585250847,
-      "loss": 46.0,
-      "step": 35933
-    },
-    {
-      "epoch": 5.786907685494585,
-      "grad_norm": 0.012045511975884438,
-      "learning_rate": 0.0001999834849331018,
-      "loss": 46.0,
-      "step": 35934
-    },
-    {
-      "epoch": 5.787068722573372,
-      "grad_norm": 0.004879689309746027,
-      "learning_rate": 0.00019998348401366954,
-      "loss": 46.0,
-      "step": 35935
-    },
-    {
-      "epoch": 5.78722975965216,
-      "grad_norm": 0.0011944433208554983,
-      "learning_rate": 0.0001999834830942117,
-      "loss": 46.0,
-      "step": 35936
-    },
-    {
-      "epoch": 5.787390796730947,
-      "grad_norm": 0.009134566411376,
-      "learning_rate": 0.00019998348217472827,
-      "loss": 46.0,
-      "step": 35937
-    },
-    {
-      "epoch": 5.787551833809735,
-      "grad_norm": 0.0070863342843949795,
-      "learning_rate": 0.00019998348125521923,
-      "loss": 46.0,
-      "step": 35938
-    },
-    {
-      "epoch": 5.787712870888522,
-      "grad_norm": 0.0019568628631532192,
-      "learning_rate": 0.00019998348033568462,
-      "loss": 46.0,
-      "step": 35939
-    },
-    {
-      "epoch": 5.78787390796731,
-      "grad_norm": 0.007232495583593845,
-      "learning_rate": 0.00019998347941612443,
-      "loss": 46.0,
-      "step": 35940
-    },
-    {
-      "epoch": 5.788034945046097,
-      "grad_norm": 0.006487681530416012,
-      "learning_rate": 0.00019998347849653863,
-      "loss": 46.0,
-      "step": 35941
-    },
-    {
-      "epoch": 5.7881959821248845,
-      "grad_norm": 0.002533015562221408,
-      "learning_rate": 0.00019998347757692724,
-      "loss": 46.0,
-      "step": 35942
-    },
-    {
-      "epoch": 5.788357019203672,
-      "grad_norm": 0.002100419718772173,
-      "learning_rate": 0.00019998347665729026,
-      "loss": 46.0,
-      "step": 35943
-    },
-    {
-      "epoch": 5.788518056282459,
-      "grad_norm": 0.004103851970285177,
-      "learning_rate": 0.0001999834757376277,
-      "loss": 46.0,
-      "step": 35944
-    },
-    {
-      "epoch": 5.788679093361246,
-      "grad_norm": 0.002914935350418091,
-      "learning_rate": 0.00019998347481793954,
-      "loss": 46.0,
-      "step": 35945
-    },
-    {
-      "epoch": 5.788840130440034,
-      "grad_norm": 0.0027448509354144335,
-      "learning_rate": 0.00019998347389822577,
-      "loss": 46.0,
-      "step": 35946
-    },
-    {
-      "epoch": 5.789001167518821,
-      "grad_norm": 0.0037534795701503754,
-      "learning_rate": 0.00019998347297848644,
-      "loss": 46.0,
-      "step": 35947
-    },
-    {
-      "epoch": 5.789162204597608,
-      "grad_norm": 0.004226305056363344,
-      "learning_rate": 0.0001999834720587215,
-      "loss": 46.0,
-      "step": 35948
-    },
-    {
-      "epoch": 5.789323241676396,
-      "grad_norm": 0.001660239533521235,
-      "learning_rate": 0.00019998347113893097,
-      "loss": 46.0,
-      "step": 35949
-    },
-    {
-      "epoch": 5.789484278755183,
-      "grad_norm": 0.0029238113202154636,
-      "learning_rate": 0.00019998347021911488,
-      "loss": 46.0,
-      "step": 35950
-    },
-    {
-      "epoch": 5.789645315833971,
-      "grad_norm": 0.011482351459562778,
-      "learning_rate": 0.00019998346929927318,
-      "loss": 46.0,
-      "step": 35951
-    },
-    {
-      "epoch": 5.789806352912758,
-      "grad_norm": 0.004890913609415293,
-      "learning_rate": 0.00019998346837940586,
-      "loss": 46.0,
-      "step": 35952
-    },
-    {
-      "epoch": 5.789967389991546,
-      "grad_norm": 0.0017212658422067761,
-      "learning_rate": 0.000199983467459513,
-      "loss": 46.0,
-      "step": 35953
-    },
-    {
-      "epoch": 5.790128427070333,
-      "grad_norm": 0.009074416011571884,
-      "learning_rate": 0.00019998346653959449,
-      "loss": 46.0,
-      "step": 35954
-    },
-    {
-      "epoch": 5.7902894641491205,
-      "grad_norm": 0.004019788466393948,
-      "learning_rate": 0.00019998346561965043,
-      "loss": 46.0,
-      "step": 35955
-    },
-    {
-      "epoch": 5.790450501227908,
-      "grad_norm": 0.00943315215408802,
-      "learning_rate": 0.0001999834646996808,
-      "loss": 46.0,
-      "step": 35956
-    },
-    {
-      "epoch": 5.790611538306695,
-      "grad_norm": 0.003384312381967902,
-      "learning_rate": 0.00019998346377968554,
-      "loss": 46.0,
-      "step": 35957
-    },
-    {
-      "epoch": 5.790772575385483,
-      "grad_norm": 0.0022111611906439066,
-      "learning_rate": 0.0001999834628596647,
-      "loss": 46.0,
-      "step": 35958
-    },
-    {
-      "epoch": 5.79093361246427,
-      "grad_norm": 0.001141060027293861,
-      "learning_rate": 0.00019998346193961827,
-      "loss": 46.0,
-      "step": 35959
-    },
-    {
-      "epoch": 5.791094649543057,
-      "grad_norm": 0.007703473791480064,
-      "learning_rate": 0.00019998346101954625,
-      "loss": 46.0,
-      "step": 35960
-    },
-    {
-      "epoch": 5.791255686621844,
-      "grad_norm": 0.0030749852303415537,
-      "learning_rate": 0.00019998346009944862,
-      "loss": 46.0,
-      "step": 35961
-    },
-    {
-      "epoch": 5.791416723700632,
-      "grad_norm": 0.001100547262467444,
-      "learning_rate": 0.00019998345917932544,
-      "loss": 46.0,
-      "step": 35962
-    },
-    {
-      "epoch": 5.791577760779419,
-      "grad_norm": 0.007251678965985775,
-      "learning_rate": 0.00019998345825917663,
-      "loss": 46.0,
-      "step": 35963
-    },
-    {
-      "epoch": 5.791738797858207,
-      "grad_norm": 0.004754495806992054,
-      "learning_rate": 0.00019998345733900227,
-      "loss": 46.0,
-      "step": 35964
-    },
-    {
-      "epoch": 5.791899834936994,
-      "grad_norm": 0.002452654065564275,
-      "learning_rate": 0.0001999834564188023,
-      "loss": 46.0,
-      "step": 35965
-    },
-    {
-      "epoch": 5.792060872015782,
-      "grad_norm": 0.007526395842432976,
-      "learning_rate": 0.00019998345549857672,
-      "loss": 46.0,
-      "step": 35966
-    },
-    {
-      "epoch": 5.792221909094569,
-      "grad_norm": 0.007444023620337248,
-      "learning_rate": 0.0001999834545783256,
-      "loss": 46.0,
-      "step": 35967
-    },
-    {
-      "epoch": 5.7923829461733565,
-      "grad_norm": 0.0014156947145238519,
-      "learning_rate": 0.00019998345365804886,
-      "loss": 46.0,
-      "step": 35968
-    },
-    {
-      "epoch": 5.792543983252144,
-      "grad_norm": 0.0034335653763264418,
-      "learning_rate": 0.0001999834527377465,
-      "loss": 46.0,
-      "step": 35969
-    },
-    {
-      "epoch": 5.792705020330931,
-      "grad_norm": 0.0055056908167898655,
-      "learning_rate": 0.0001999834518174186,
-      "loss": 46.0,
-      "step": 35970
-    },
-    {
-      "epoch": 5.792866057409719,
-      "grad_norm": 0.011387539096176624,
-      "learning_rate": 0.00019998345089706506,
-      "loss": 46.0,
-      "step": 35971
-    },
-    {
-      "epoch": 5.793027094488506,
-      "grad_norm": 0.030026080086827278,
-      "learning_rate": 0.00019998344997668594,
-      "loss": 46.0,
-      "step": 35972
-    },
-    {
-      "epoch": 5.793188131567294,
-      "grad_norm": 0.002667773747816682,
-      "learning_rate": 0.00019998344905628127,
-      "loss": 46.0,
-      "step": 35973
-    },
-    {
-      "epoch": 5.793349168646081,
-      "grad_norm": 0.01391933299601078,
-      "learning_rate": 0.00019998344813585095,
-      "loss": 46.0,
-      "step": 35974
-    },
-    {
-      "epoch": 5.793510205724868,
-      "grad_norm": 0.0012938538566231728,
-      "learning_rate": 0.00019998344721539507,
-      "loss": 46.0,
-      "step": 35975
-    },
-    {
-      "epoch": 5.793671242803655,
-      "grad_norm": 0.002052146242931485,
-      "learning_rate": 0.0001999834462949136,
-      "loss": 46.0,
-      "step": 35976
-    },
-    {
-      "epoch": 5.793832279882443,
-      "grad_norm": 0.0018925995100289583,
-      "learning_rate": 0.00019998344537440653,
-      "loss": 46.0,
-      "step": 35977
-    },
-    {
-      "epoch": 5.79399331696123,
-      "grad_norm": 0.0006377461249940097,
-      "learning_rate": 0.00019998344445387392,
-      "loss": 46.0,
-      "step": 35978
-    },
-    {
-      "epoch": 5.7941543540400176,
-      "grad_norm": 0.011470134370028973,
-      "learning_rate": 0.00019998344353331567,
-      "loss": 46.0,
-      "step": 35979
-    },
-    {
-      "epoch": 5.794315391118805,
-      "grad_norm": 0.007166468072682619,
-      "learning_rate": 0.00019998344261273182,
-      "loss": 46.0,
-      "step": 35980
-    },
-    {
-      "epoch": 5.7944764281975925,
-      "grad_norm": 0.002545740222558379,
-      "learning_rate": 0.00019998344169212242,
-      "loss": 46.0,
-      "step": 35981
-    },
-    {
-      "epoch": 5.79463746527638,
-      "grad_norm": 0.0030606503132730722,
-      "learning_rate": 0.00019998344077148738,
-      "loss": 46.0,
-      "step": 35982
-    },
-    {
-      "epoch": 5.794798502355167,
-      "grad_norm": 0.002506427001208067,
-      "learning_rate": 0.00019998343985082678,
-      "loss": 46.0,
-      "step": 35983
-    },
-    {
-      "epoch": 5.794959539433955,
-      "grad_norm": 0.006900331936776638,
-      "learning_rate": 0.0001999834389301406,
-      "loss": 46.0,
-      "step": 35984
-    },
-    {
-      "epoch": 5.795120576512742,
-      "grad_norm": 0.015905145555734634,
-      "learning_rate": 0.00019998343800942882,
-      "loss": 46.0,
-      "step": 35985
-    },
-    {
-      "epoch": 5.79528161359153,
-      "grad_norm": 0.015668099746108055,
-      "learning_rate": 0.00019998343708869145,
-      "loss": 46.0,
-      "step": 35986
-    },
-    {
-      "epoch": 5.795442650670317,
-      "grad_norm": 0.002261025132611394,
-      "learning_rate": 0.00019998343616792848,
-      "loss": 46.0,
-      "step": 35987
-    },
-    {
-      "epoch": 5.795603687749105,
-      "grad_norm": 0.00249306159093976,
-      "learning_rate": 0.0001999834352471399,
-      "loss": 46.0,
-      "step": 35988
-    },
-    {
-      "epoch": 5.795764724827892,
-      "grad_norm": 0.003147156210616231,
-      "learning_rate": 0.00019998343432632579,
-      "loss": 46.0,
-      "step": 35989
-    },
-    {
-      "epoch": 5.795925761906679,
-      "grad_norm": 0.008631477132439613,
-      "learning_rate": 0.00019998343340548602,
-      "loss": 46.0,
-      "step": 35990
-    },
-    {
-      "epoch": 5.796086798985466,
-      "grad_norm": 0.001991981640458107,
-      "learning_rate": 0.0001999834324846207,
-      "loss": 46.0,
-      "step": 35991
-    },
-    {
-      "epoch": 5.7962478360642535,
-      "grad_norm": 0.005137927830219269,
-      "learning_rate": 0.00019998343156372978,
-      "loss": 46.0,
-      "step": 35992
-    },
-    {
-      "epoch": 5.796408873143041,
-      "grad_norm": 0.0016772891394793987,
-      "learning_rate": 0.00019998343064281325,
-      "loss": 46.0,
-      "step": 35993
-    },
-    {
-      "epoch": 5.796569910221828,
-      "grad_norm": 0.002501783426851034,
-      "learning_rate": 0.0001999834297218712,
-      "loss": 46.0,
-      "step": 35994
-    },
-    {
-      "epoch": 5.796730947300616,
-      "grad_norm": 0.004704023711383343,
-      "learning_rate": 0.0001999834288009035,
-      "loss": 46.0,
-      "step": 35995
-    },
-    {
-      "epoch": 5.796891984379403,
-      "grad_norm": 0.0033317310735583305,
-      "learning_rate": 0.0001999834278799102,
-      "loss": 46.0,
-      "step": 35996
-    },
-    {
-      "epoch": 5.797053021458191,
-      "grad_norm": 0.001641907263547182,
-      "learning_rate": 0.00019998342695889132,
-      "loss": 46.0,
-      "step": 35997
-    },
-    {
-      "epoch": 5.797214058536978,
-      "grad_norm": 0.0008138873963616788,
-      "learning_rate": 0.00019998342603784686,
-      "loss": 46.0,
-      "step": 35998
-    },
-    {
-      "epoch": 5.797375095615766,
-      "grad_norm": 0.011361321434378624,
-      "learning_rate": 0.00019998342511677678,
-      "loss": 46.0,
-      "step": 35999
-    },
-    {
-      "epoch": 5.797536132694553,
-      "grad_norm": 0.002221940318122506,
-      "learning_rate": 0.00019998342419568114,
-      "loss": 46.0,
-      "step": 36000
-    },
-    {
-      "epoch": 5.797697169773341,
-      "grad_norm": 0.005986008793115616,
-      "learning_rate": 0.00019998342327455992,
-      "loss": 46.0,
-      "step": 36001
-    },
-    {
-      "epoch": 5.797858206852128,
-      "grad_norm": 0.003031674772500992,
-      "learning_rate": 0.0001999834223534131,
-      "loss": 46.0,
-      "step": 36002
-    },
-    {
-      "epoch": 5.798019243930915,
-      "grad_norm": 0.0016920288326218724,
-      "learning_rate": 0.00019998342143224068,
-      "loss": 46.0,
-      "step": 36003
-    },
-    {
-      "epoch": 5.798180281009703,
-      "grad_norm": 0.030749212950468063,
-      "learning_rate": 0.0001999834205110427,
-      "loss": 46.0,
-      "step": 36004
-    },
-    {
-      "epoch": 5.7983413180884895,
-      "grad_norm": 0.008248284459114075,
-      "learning_rate": 0.00019998341958981907,
-      "loss": 46.0,
-      "step": 36005
-    },
-    {
-      "epoch": 5.798502355167277,
-      "grad_norm": 0.0041768429800868034,
-      "learning_rate": 0.00019998341866856988,
-      "loss": 46.0,
-      "step": 36006
-    },
-    {
-      "epoch": 5.798663392246064,
-      "grad_norm": 0.010321359150111675,
-      "learning_rate": 0.0001999834177472951,
-      "loss": 46.0,
-      "step": 36007
-    },
-    {
-      "epoch": 5.798824429324852,
-      "grad_norm": 0.013382167555391788,
-      "learning_rate": 0.00019998341682599474,
-      "loss": 46.0,
-      "step": 36008
-    },
-    {
-      "epoch": 5.798985466403639,
-      "grad_norm": 0.007584656123071909,
-      "learning_rate": 0.0001999834159046688,
-      "loss": 46.0,
-      "step": 36009
-    },
-    {
-      "epoch": 5.799146503482427,
-      "grad_norm": 0.0026863585226237774,
-      "learning_rate": 0.00019998341498331723,
-      "loss": 46.0,
-      "step": 36010
-    },
-    {
-      "epoch": 5.799307540561214,
-      "grad_norm": 0.007454379461705685,
-      "learning_rate": 0.0001999834140619401,
-      "loss": 46.0,
-      "step": 36011
-    },
-    {
-      "epoch": 5.799468577640002,
-      "grad_norm": 0.00596819119527936,
-      "learning_rate": 0.00019998341314053736,
-      "loss": 46.0,
-      "step": 36012
-    },
-    {
-      "epoch": 5.799629614718789,
-      "grad_norm": 0.002117754425853491,
-      "learning_rate": 0.00019998341221910904,
-      "loss": 46.0,
-      "step": 36013
-    },
-    {
-      "epoch": 5.799790651797577,
-      "grad_norm": 0.009177391417324543,
-      "learning_rate": 0.00019998341129765513,
-      "loss": 46.0,
-      "step": 36014
-    },
-    {
-      "epoch": 5.799951688876364,
-      "grad_norm": 0.006089060567319393,
-      "learning_rate": 0.0001999834103761756,
-      "loss": 46.0,
-      "step": 36015
-    },
-    {
-      "epoch": 5.8001127259551515,
-      "grad_norm": 0.0027072851080447435,
-      "learning_rate": 0.0001999834094546705,
-      "loss": 46.0,
-      "step": 36016
-    },
-    {
-      "epoch": 5.800273763033939,
-      "grad_norm": 0.008395212702453136,
-      "learning_rate": 0.00019998340853313984,
-      "loss": 46.0,
-      "step": 36017
-    },
-    {
-      "epoch": 5.8004348001127255,
-      "grad_norm": 0.0014350964920595288,
-      "learning_rate": 0.00019998340761158355,
-      "loss": 46.0,
-      "step": 36018
-    },
-    {
-      "epoch": 5.800595837191514,
-      "grad_norm": 0.006828527897596359,
-      "learning_rate": 0.0001999834066900017,
-      "loss": 46.0,
-      "step": 36019
-    },
-    {
-      "epoch": 5.8007568742703,
-      "grad_norm": 0.00606875866651535,
-      "learning_rate": 0.00019998340576839424,
-      "loss": 46.0,
-      "step": 36020
-    },
-    {
-      "epoch": 5.800917911349088,
-      "grad_norm": 0.0013484942028298974,
-      "learning_rate": 0.0001999834048467612,
-      "loss": 46.0,
-      "step": 36021
-    },
-    {
-      "epoch": 5.801078948427875,
-      "grad_norm": 0.0018275863258168101,
-      "learning_rate": 0.00019998340392510255,
-      "loss": 46.0,
-      "step": 36022
-    },
-    {
-      "epoch": 5.801239985506663,
-      "grad_norm": 0.00551186315715313,
-      "learning_rate": 0.0001999834030034183,
-      "loss": 46.0,
-      "step": 36023
-    },
-    {
-      "epoch": 5.80140102258545,
-      "grad_norm": 0.0013101280201226473,
-      "learning_rate": 0.0001999834020817085,
-      "loss": 46.0,
-      "step": 36024
-    },
-    {
-      "epoch": 5.801562059664238,
-      "grad_norm": 0.0013992469757795334,
-      "learning_rate": 0.0001999834011599731,
-      "loss": 46.0,
-      "step": 36025
-    },
-    {
-      "epoch": 5.801723096743025,
-      "grad_norm": 0.0021910525392740965,
-      "learning_rate": 0.0001999834002382121,
-      "loss": 46.0,
-      "step": 36026
-    },
-    {
-      "epoch": 5.801884133821813,
-      "grad_norm": 0.0008445768035016954,
-      "learning_rate": 0.0001999833993164255,
-      "loss": 46.0,
-      "step": 36027
-    },
-    {
-      "epoch": 5.8020451709006,
-      "grad_norm": 0.0015970169333741069,
-      "learning_rate": 0.00019998339839461332,
-      "loss": 46.0,
-      "step": 36028
-    },
-    {
-      "epoch": 5.8022062079793875,
-      "grad_norm": 0.0012416945537552238,
-      "learning_rate": 0.00019998339747277554,
-      "loss": 46.0,
-      "step": 36029
-    },
-    {
-      "epoch": 5.802367245058175,
-      "grad_norm": 0.00425972742959857,
-      "learning_rate": 0.00019998339655091218,
-      "loss": 46.0,
-      "step": 36030
-    },
-    {
-      "epoch": 5.802528282136962,
-      "grad_norm": 0.002603266155347228,
-      "learning_rate": 0.0001999833956290232,
-      "loss": 46.0,
-      "step": 36031
-    },
-    {
-      "epoch": 5.80268931921575,
-      "grad_norm": 0.013226257637143135,
-      "learning_rate": 0.00019998339470710867,
-      "loss": 46.0,
-      "step": 36032
-    },
-    {
-      "epoch": 5.802850356294536,
-      "grad_norm": 0.014850325882434845,
-      "learning_rate": 0.00019998339378516855,
-      "loss": 46.0,
-      "step": 36033
-    },
-    {
-      "epoch": 5.803011393373324,
-      "grad_norm": 0.0041097248904407024,
-      "learning_rate": 0.0001999833928632028,
-      "loss": 46.0,
-      "step": 36034
-    },
-    {
-      "epoch": 5.803172430452111,
-      "grad_norm": 0.002324201399460435,
-      "learning_rate": 0.00019998339194121151,
-      "loss": 46.0,
-      "step": 36035
-    },
-    {
-      "epoch": 5.803333467530899,
-      "grad_norm": 0.005739324726164341,
-      "learning_rate": 0.0001999833910191946,
-      "loss": 46.0,
-      "step": 36036
-    },
-    {
-      "epoch": 5.803494504609686,
-      "grad_norm": 0.012247636914253235,
-      "learning_rate": 0.0001999833900971521,
-      "loss": 46.0,
-      "step": 36037
-    },
-    {
-      "epoch": 5.803655541688474,
-      "grad_norm": 0.0008543399744667113,
-      "learning_rate": 0.00019998338917508402,
-      "loss": 46.0,
-      "step": 36038
-    },
-    {
-      "epoch": 5.803816578767261,
-      "grad_norm": 0.002412332920357585,
-      "learning_rate": 0.00019998338825299032,
-      "loss": 46.0,
-      "step": 36039
-    },
-    {
-      "epoch": 5.8039776158460485,
-      "grad_norm": 0.02353476732969284,
-      "learning_rate": 0.00019998338733087106,
-      "loss": 46.0,
-      "step": 36040
-    },
-    {
-      "epoch": 5.804138652924836,
-      "grad_norm": 0.006275770720094442,
-      "learning_rate": 0.0001999833864087262,
-      "loss": 46.0,
-      "step": 36041
-    },
-    {
-      "epoch": 5.8042996900036234,
-      "grad_norm": 0.008256159722805023,
-      "learning_rate": 0.00019998338548655578,
-      "loss": 46.0,
-      "step": 36042
-    },
-    {
-      "epoch": 5.804460727082411,
-      "grad_norm": 0.030541103333234787,
-      "learning_rate": 0.00019998338456435973,
-      "loss": 46.0,
-      "step": 36043
-    },
-    {
-      "epoch": 5.804621764161198,
-      "grad_norm": 0.0055891480296850204,
-      "learning_rate": 0.0001999833836421381,
-      "loss": 46.0,
-      "step": 36044
-    },
-    {
-      "epoch": 5.804782801239986,
-      "grad_norm": 0.011011507362127304,
-      "learning_rate": 0.00019998338271989087,
-      "loss": 46.0,
-      "step": 36045
-    },
-    {
-      "epoch": 5.804943838318773,
-      "grad_norm": 0.011860811151564121,
-      "learning_rate": 0.00019998338179761803,
-      "loss": 46.0,
-      "step": 36046
-    },
-    {
-      "epoch": 5.805104875397561,
-      "grad_norm": 0.01165796723216772,
-      "learning_rate": 0.00019998338087531966,
-      "loss": 46.0,
-      "step": 36047
-    },
-    {
-      "epoch": 5.805265912476347,
-      "grad_norm": 0.0028356791008263826,
-      "learning_rate": 0.00019998337995299565,
-      "loss": 46.0,
-      "step": 36048
-    },
-    {
-      "epoch": 5.805426949555135,
-      "grad_norm": 0.00123371253721416,
-      "learning_rate": 0.00019998337903064605,
-      "loss": 46.0,
-      "step": 36049
-    },
-    {
-      "epoch": 5.805587986633922,
-      "grad_norm": 0.01672019436955452,
-      "learning_rate": 0.0001999833781082709,
-      "loss": 46.0,
-      "step": 36050
-    },
-    {
-      "epoch": 5.80574902371271,
-      "grad_norm": 0.004154611844569445,
-      "learning_rate": 0.00019998337718587012,
-      "loss": 46.0,
-      "step": 36051
-    },
-    {
-      "epoch": 5.805910060791497,
-      "grad_norm": 0.007643844000995159,
-      "learning_rate": 0.00019998337626344375,
-      "loss": 46.0,
-      "step": 36052
-    },
-    {
-      "epoch": 5.8060710978702845,
-      "grad_norm": 0.0030058391857892275,
-      "learning_rate": 0.00019998337534099183,
-      "loss": 46.0,
-      "step": 36053
-    },
-    {
-      "epoch": 5.806232134949072,
-      "grad_norm": 0.006475790403783321,
-      "learning_rate": 0.00019998337441851427,
-      "loss": 46.0,
-      "step": 36054
-    },
-    {
-      "epoch": 5.806393172027859,
-      "grad_norm": 0.004651803057640791,
-      "learning_rate": 0.00019998337349601115,
-      "loss": 46.0,
-      "step": 36055
-    },
-    {
-      "epoch": 5.806554209106647,
-      "grad_norm": 0.0032809628173708916,
-      "learning_rate": 0.00019998337257348244,
-      "loss": 46.0,
-      "step": 36056
-    },
-    {
-      "epoch": 5.806715246185434,
-      "grad_norm": 0.004969972651451826,
-      "learning_rate": 0.00019998337165092812,
-      "loss": 46.0,
-      "step": 36057
-    },
-    {
-      "epoch": 5.806876283264222,
-      "grad_norm": 0.002350902184844017,
-      "learning_rate": 0.00019998337072834823,
-      "loss": 46.0,
-      "step": 36058
-    },
-    {
-      "epoch": 5.807037320343009,
-      "grad_norm": 0.006224953103810549,
-      "learning_rate": 0.00019998336980574274,
-      "loss": 46.0,
-      "step": 36059
-    },
-    {
-      "epoch": 5.807198357421797,
-      "grad_norm": 0.018788155168294907,
-      "learning_rate": 0.00019998336888311165,
-      "loss": 46.0,
-      "step": 36060
-    },
-    {
-      "epoch": 5.807359394500584,
-      "grad_norm": 0.0013146103592589498,
-      "learning_rate": 0.00019998336796045498,
-      "loss": 46.0,
-      "step": 36061
-    },
-    {
-      "epoch": 5.807520431579372,
-      "grad_norm": 0.002196866786107421,
-      "learning_rate": 0.00019998336703777272,
-      "loss": 46.0,
-      "step": 36062
-    },
-    {
-      "epoch": 5.807681468658158,
-      "grad_norm": 0.00149641465395689,
-      "learning_rate": 0.00019998336611506487,
-      "loss": 46.0,
-      "step": 36063
-    },
-    {
-      "epoch": 5.807842505736946,
-      "grad_norm": 0.005241080652922392,
-      "learning_rate": 0.00019998336519233144,
-      "loss": 46.0,
-      "step": 36064
-    },
-    {
-      "epoch": 5.808003542815733,
-      "grad_norm": 0.0010303921299055219,
-      "learning_rate": 0.0001999833642695724,
-      "loss": 46.0,
-      "step": 36065
-    },
-    {
-      "epoch": 5.8081645798945205,
-      "grad_norm": 0.004168648272752762,
-      "learning_rate": 0.00019998336334678776,
-      "loss": 46.0,
-      "step": 36066
-    },
-    {
-      "epoch": 5.808325616973308,
-      "grad_norm": 0.006564165931195021,
-      "learning_rate": 0.00019998336242397753,
-      "loss": 46.0,
-      "step": 36067
-    },
-    {
-      "epoch": 5.808486654052095,
-      "grad_norm": 0.005773855838924646,
-      "learning_rate": 0.00019998336150114175,
-      "loss": 46.0,
-      "step": 36068
-    },
-    {
-      "epoch": 5.808647691130883,
-      "grad_norm": 0.014673485420644283,
-      "learning_rate": 0.00019998336057828033,
-      "loss": 46.0,
-      "step": 36069
-    },
-    {
-      "epoch": 5.80880872820967,
-      "grad_norm": 0.006893686484545469,
-      "learning_rate": 0.00019998335965539337,
-      "loss": 46.0,
-      "step": 36070
-    },
-    {
-      "epoch": 5.808969765288458,
-      "grad_norm": 0.0011093391804024577,
-      "learning_rate": 0.00019998335873248077,
-      "loss": 46.0,
-      "step": 36071
-    },
-    {
-      "epoch": 5.809130802367245,
-      "grad_norm": 0.009555998258292675,
-      "learning_rate": 0.00019998335780954261,
-      "loss": 46.0,
-      "step": 36072
-    },
-    {
-      "epoch": 5.809291839446033,
-      "grad_norm": 0.005293944850564003,
-      "learning_rate": 0.00019998335688657884,
-      "loss": 46.0,
-      "step": 36073
-    },
-    {
-      "epoch": 5.80945287652482,
-      "grad_norm": 0.0034722699783742428,
-      "learning_rate": 0.00019998335596358948,
-      "loss": 46.0,
-      "step": 36074
-    },
-    {
-      "epoch": 5.809613913603608,
-      "grad_norm": 0.005422583315521479,
-      "learning_rate": 0.00019998335504057456,
-      "loss": 46.0,
-      "step": 36075
-    },
-    {
-      "epoch": 5.809774950682394,
-      "grad_norm": 0.0013842072803527117,
-      "learning_rate": 0.00019998335411753403,
-      "loss": 46.0,
-      "step": 36076
-    },
-    {
-      "epoch": 5.8099359877611825,
-      "grad_norm": 0.004719553515315056,
-      "learning_rate": 0.0001999833531944679,
-      "loss": 46.0,
-      "step": 36077
-    },
-    {
-      "epoch": 5.810097024839969,
-      "grad_norm": 0.0023871390148997307,
-      "learning_rate": 0.0001999833522713762,
-      "loss": 46.0,
-      "step": 36078
-    },
-    {
-      "epoch": 5.8102580619187565,
-      "grad_norm": 0.004307557828724384,
-      "learning_rate": 0.00019998335134825888,
-      "loss": 46.0,
-      "step": 36079
-    },
-    {
-      "epoch": 5.810419098997544,
-      "grad_norm": 0.003083202987909317,
-      "learning_rate": 0.00019998335042511597,
-      "loss": 46.0,
-      "step": 36080
-    },
-    {
-      "epoch": 5.810580136076331,
-      "grad_norm": 0.0027006755117326975,
-      "learning_rate": 0.0001999833495019475,
-      "loss": 46.0,
-      "step": 36081
-    },
-    {
-      "epoch": 5.810741173155119,
-      "grad_norm": 0.0022710133343935013,
-      "learning_rate": 0.00019998334857875344,
-      "loss": 46.0,
-      "step": 36082
-    },
-    {
-      "epoch": 5.810902210233906,
-      "grad_norm": 0.006473921705037355,
-      "learning_rate": 0.00019998334765553377,
-      "loss": 46.0,
-      "step": 36083
-    },
-    {
-      "epoch": 5.811063247312694,
-      "grad_norm": 0.009408203884959221,
-      "learning_rate": 0.0001999833467322885,
-      "loss": 46.0,
-      "step": 36084
-    },
-    {
-      "epoch": 5.811224284391481,
-      "grad_norm": 0.002478943672031164,
-      "learning_rate": 0.00019998334580901766,
-      "loss": 46.0,
-      "step": 36085
-    },
-    {
-      "epoch": 5.811385321470269,
-      "grad_norm": 0.0016110965516418219,
-      "learning_rate": 0.0001999833448857212,
-      "loss": 46.0,
-      "step": 36086
-    },
-    {
-      "epoch": 5.811546358549056,
-      "grad_norm": 0.0034495352301746607,
-      "learning_rate": 0.00019998334396239918,
-      "loss": 46.0,
-      "step": 36087
-    },
-    {
-      "epoch": 5.8117073956278436,
-      "grad_norm": 0.000996622839011252,
-      "learning_rate": 0.00019998334303905158,
-      "loss": 46.0,
-      "step": 36088
-    },
-    {
-      "epoch": 5.811868432706631,
-      "grad_norm": 0.02360527217388153,
-      "learning_rate": 0.00019998334211567836,
-      "loss": 46.0,
-      "step": 36089
-    },
-    {
-      "epoch": 5.8120294697854185,
-      "grad_norm": 0.008745023980736732,
-      "learning_rate": 0.00019998334119227957,
-      "loss": 46.0,
-      "step": 36090
-    },
-    {
-      "epoch": 5.812190506864205,
-      "grad_norm": 0.0018846254097297788,
-      "learning_rate": 0.00019998334026885518,
-      "loss": 46.0,
-      "step": 36091
-    },
-    {
-      "epoch": 5.812351543942993,
-      "grad_norm": 0.00911658350378275,
-      "learning_rate": 0.0001999833393454052,
-      "loss": 46.0,
-      "step": 36092
-    },
-    {
-      "epoch": 5.81251258102178,
-      "grad_norm": 0.0025152629241347313,
-      "learning_rate": 0.00019998333842192963,
-      "loss": 46.0,
-      "step": 36093
-    },
-    {
-      "epoch": 5.812673618100567,
-      "grad_norm": 0.004125799052417278,
-      "learning_rate": 0.00019998333749842847,
-      "loss": 46.0,
-      "step": 36094
-    },
-    {
-      "epoch": 5.812834655179355,
-      "grad_norm": 0.0018885155441239476,
-      "learning_rate": 0.00019998333657490172,
-      "loss": 46.0,
-      "step": 36095
-    },
-    {
-      "epoch": 5.812995692258142,
-      "grad_norm": 0.004321923479437828,
-      "learning_rate": 0.00019998333565134937,
-      "loss": 46.0,
-      "step": 36096
-    },
-    {
-      "epoch": 5.81315672933693,
-      "grad_norm": 0.006099647842347622,
-      "learning_rate": 0.00019998333472777145,
-      "loss": 46.0,
-      "step": 36097
-    },
-    {
-      "epoch": 5.813317766415717,
-      "grad_norm": 0.00210051448084414,
-      "learning_rate": 0.0001999833338041679,
-      "loss": 46.0,
-      "step": 36098
-    },
-    {
-      "epoch": 5.813478803494505,
-      "grad_norm": 0.002787800971418619,
-      "learning_rate": 0.0001999833328805388,
-      "loss": 46.0,
-      "step": 36099
-    },
-    {
-      "epoch": 5.813639840573292,
-      "grad_norm": 0.01858663558959961,
-      "learning_rate": 0.00019998333195688409,
-      "loss": 46.0,
-      "step": 36100
-    },
-    {
-      "epoch": 5.8138008776520795,
-      "grad_norm": 0.00549480551853776,
-      "learning_rate": 0.0001999833310332038,
-      "loss": 46.0,
-      "step": 36101
-    },
-    {
-      "epoch": 5.813961914730867,
-      "grad_norm": 0.002554436679929495,
-      "learning_rate": 0.0001999833301094979,
-      "loss": 46.0,
-      "step": 36102
-    },
-    {
-      "epoch": 5.814122951809654,
-      "grad_norm": 0.0012868873309344053,
-      "learning_rate": 0.00019998332918576641,
-      "loss": 46.0,
-      "step": 36103
-    },
-    {
-      "epoch": 5.814283988888442,
-      "grad_norm": 0.014549266546964645,
-      "learning_rate": 0.00019998332826200933,
-      "loss": 46.0,
-      "step": 36104
-    },
-    {
-      "epoch": 5.814445025967229,
-      "grad_norm": 0.0019923283252865076,
-      "learning_rate": 0.0001999833273382267,
-      "loss": 46.0,
-      "step": 36105
-    },
-    {
-      "epoch": 5.814606063046016,
-      "grad_norm": 0.0033107066992670298,
-      "learning_rate": 0.00019998332641441846,
-      "loss": 46.0,
-      "step": 36106
-    },
-    {
-      "epoch": 5.814767100124803,
-      "grad_norm": 0.003413548693060875,
-      "learning_rate": 0.0001999833254905846,
-      "loss": 46.0,
-      "step": 36107
-    },
-    {
-      "epoch": 5.814928137203591,
-      "grad_norm": 0.0015034214593470097,
-      "learning_rate": 0.00019998332456672518,
-      "loss": 46.0,
-      "step": 36108
-    },
-    {
-      "epoch": 5.815089174282378,
-      "grad_norm": 0.011482627131044865,
-      "learning_rate": 0.00019998332364284016,
-      "loss": 46.0,
-      "step": 36109
-    },
-    {
-      "epoch": 5.815250211361166,
-      "grad_norm": 0.0032607412431389093,
-      "learning_rate": 0.00019998332271892955,
-      "loss": 46.0,
-      "step": 36110
-    },
-    {
-      "epoch": 5.815411248439953,
-      "grad_norm": 0.00533420080319047,
-      "learning_rate": 0.00019998332179499336,
-      "loss": 46.0,
-      "step": 36111
-    },
-    {
-      "epoch": 5.815572285518741,
-      "grad_norm": 0.008268029429018497,
-      "learning_rate": 0.00019998332087103155,
-      "loss": 46.0,
-      "step": 36112
-    },
-    {
-      "epoch": 5.815733322597528,
-      "grad_norm": 0.0037971429992467165,
-      "learning_rate": 0.00019998331994704416,
-      "loss": 46.0,
-      "step": 36113
-    },
-    {
-      "epoch": 5.8158943596763155,
-      "grad_norm": 0.008767115883529186,
-      "learning_rate": 0.00019998331902303118,
-      "loss": 46.0,
-      "step": 36114
-    },
-    {
-      "epoch": 5.816055396755103,
-      "grad_norm": 0.0005402269889600575,
-      "learning_rate": 0.00019998331809899263,
-      "loss": 46.0,
-      "step": 36115
-    },
-    {
-      "epoch": 5.81621643383389,
-      "grad_norm": 0.003751007141545415,
-      "learning_rate": 0.00019998331717492845,
-      "loss": 46.0,
-      "step": 36116
-    },
-    {
-      "epoch": 5.816377470912678,
-      "grad_norm": 0.007433861494064331,
-      "learning_rate": 0.00019998331625083873,
-      "loss": 46.0,
-      "step": 36117
-    },
-    {
-      "epoch": 5.816538507991465,
-      "grad_norm": 0.0012924092588946223,
-      "learning_rate": 0.00019998331532672338,
-      "loss": 46.0,
-      "step": 36118
-    },
-    {
-      "epoch": 5.816699545070253,
-      "grad_norm": 0.009297915734350681,
-      "learning_rate": 0.00019998331440258243,
-      "loss": 46.0,
-      "step": 36119
-    },
-    {
-      "epoch": 5.81686058214904,
-      "grad_norm": 0.002273964462801814,
-      "learning_rate": 0.00019998331347841593,
-      "loss": 46.0,
-      "step": 36120
-    },
-    {
-      "epoch": 5.817021619227827,
-      "grad_norm": 0.01792997121810913,
-      "learning_rate": 0.00019998331255422383,
-      "loss": 46.0,
-      "step": 36121
-    },
-    {
-      "epoch": 5.817182656306614,
-      "grad_norm": 0.0037858830764889717,
-      "learning_rate": 0.00019998331163000616,
-      "loss": 46.0,
-      "step": 36122
-    },
-    {
-      "epoch": 5.817343693385402,
-      "grad_norm": 0.005455431528389454,
-      "learning_rate": 0.00019998331070576283,
-      "loss": 46.0,
-      "step": 36123
-    },
-    {
-      "epoch": 5.817504730464189,
-      "grad_norm": 0.0041198753751814365,
-      "learning_rate": 0.00019998330978149395,
-      "loss": 46.0,
-      "step": 36124
-    },
-    {
-      "epoch": 5.817665767542977,
-      "grad_norm": 0.003341143950819969,
-      "learning_rate": 0.00019998330885719949,
-      "loss": 46.0,
-      "step": 36125
-    },
-    {
-      "epoch": 5.817826804621764,
-      "grad_norm": 0.004304523579776287,
-      "learning_rate": 0.00019998330793287943,
-      "loss": 46.0,
-      "step": 36126
-    },
-    {
-      "epoch": 5.8179878417005515,
-      "grad_norm": 0.024700313806533813,
-      "learning_rate": 0.0001999833070085338,
-      "loss": 46.0,
-      "step": 36127
-    },
-    {
-      "epoch": 5.818148878779339,
-      "grad_norm": 0.0043274881318211555,
-      "learning_rate": 0.00019998330608416253,
-      "loss": 46.0,
-      "step": 36128
-    },
-    {
-      "epoch": 5.818309915858126,
-      "grad_norm": 0.0014909901656210423,
-      "learning_rate": 0.00019998330515976572,
-      "loss": 46.0,
-      "step": 36129
-    },
-    {
-      "epoch": 5.818470952936914,
-      "grad_norm": 0.015636015683412552,
-      "learning_rate": 0.00019998330423534326,
-      "loss": 46.0,
-      "step": 36130
-    },
-    {
-      "epoch": 5.818631990015701,
-      "grad_norm": 0.004163338802754879,
-      "learning_rate": 0.00019998330331089527,
-      "loss": 46.0,
-      "step": 36131
-    },
-    {
-      "epoch": 5.818793027094489,
-      "grad_norm": 0.001807473599910736,
-      "learning_rate": 0.00019998330238642166,
-      "loss": 46.0,
-      "step": 36132
-    },
-    {
-      "epoch": 5.818954064173276,
-      "grad_norm": 0.0048170145601034164,
-      "learning_rate": 0.00019998330146192247,
-      "loss": 46.0,
-      "step": 36133
-    },
-    {
-      "epoch": 5.819115101252064,
-      "grad_norm": 0.013081304728984833,
-      "learning_rate": 0.00019998330053739766,
-      "loss": 46.0,
-      "step": 36134
-    },
-    {
-      "epoch": 5.819276138330851,
-      "grad_norm": 0.0010845721699297428,
-      "learning_rate": 0.0001999832996128473,
-      "loss": 46.0,
-      "step": 36135
-    },
-    {
-      "epoch": 5.819437175409638,
-      "grad_norm": 0.015675462782382965,
-      "learning_rate": 0.0001999832986882713,
-      "loss": 46.0,
-      "step": 36136
-    },
-    {
-      "epoch": 5.819598212488425,
-      "grad_norm": 0.0058937156572937965,
-      "learning_rate": 0.00019998329776366977,
-      "loss": 46.0,
-      "step": 36137
-    },
-    {
-      "epoch": 5.819759249567213,
-      "grad_norm": 0.0021797495428472757,
-      "learning_rate": 0.00019998329683904262,
-      "loss": 46.0,
-      "step": 36138
-    },
-    {
-      "epoch": 5.819920286646,
-      "grad_norm": 0.009142951108515263,
-      "learning_rate": 0.00019998329591438987,
-      "loss": 46.0,
-      "step": 36139
-    },
-    {
-      "epoch": 5.8200813237247875,
-      "grad_norm": 0.007929841056466103,
-      "learning_rate": 0.00019998329498971154,
-      "loss": 46.0,
-      "step": 36140
-    },
-    {
-      "epoch": 5.820242360803575,
-      "grad_norm": 0.0022620921954512596,
-      "learning_rate": 0.00019998329406500763,
-      "loss": 46.0,
-      "step": 36141
-    },
-    {
-      "epoch": 5.820403397882362,
-      "grad_norm": 0.009755589067935944,
-      "learning_rate": 0.00019998329314027812,
-      "loss": 46.0,
-      "step": 36142
-    },
-    {
-      "epoch": 5.82056443496115,
-      "grad_norm": 0.01037575677037239,
-      "learning_rate": 0.000199983292215523,
-      "loss": 46.0,
-      "step": 36143
-    },
-    {
-      "epoch": 5.820725472039937,
-      "grad_norm": 0.0069565302692353725,
-      "learning_rate": 0.00019998329129074233,
-      "loss": 46.0,
-      "step": 36144
-    },
-    {
-      "epoch": 5.820886509118725,
-      "grad_norm": 0.008227053098380566,
-      "learning_rate": 0.00019998329036593603,
-      "loss": 46.0,
-      "step": 36145
-    },
-    {
-      "epoch": 5.821047546197512,
-      "grad_norm": 0.009839381091296673,
-      "learning_rate": 0.00019998328944110418,
-      "loss": 46.0,
-      "step": 36146
-    },
-    {
-      "epoch": 5.8212085832763,
-      "grad_norm": 0.009177645668387413,
-      "learning_rate": 0.00019998328851624671,
-      "loss": 46.0,
-      "step": 36147
-    },
-    {
-      "epoch": 5.821369620355087,
-      "grad_norm": 0.0018640889320522547,
-      "learning_rate": 0.00019998328759136363,
-      "loss": 46.0,
-      "step": 36148
-    },
-    {
-      "epoch": 5.821530657433874,
-      "grad_norm": 0.006727991160005331,
-      "learning_rate": 0.00019998328666645502,
-      "loss": 46.0,
-      "step": 36149
-    },
-    {
-      "epoch": 5.821691694512662,
-      "grad_norm": 0.0049310834147036076,
-      "learning_rate": 0.0001999832857415208,
-      "loss": 46.0,
-      "step": 36150
-    },
-    {
-      "epoch": 5.821852731591449,
-      "grad_norm": 0.001063590869307518,
-      "learning_rate": 0.00019998328481656095,
-      "loss": 46.0,
-      "step": 36151
-    },
-    {
-      "epoch": 5.822013768670236,
-      "grad_norm": 0.011167946271598339,
-      "learning_rate": 0.00019998328389157552,
-      "loss": 46.0,
-      "step": 36152
-    },
-    {
-      "epoch": 5.8221748057490235,
-      "grad_norm": 0.003994588274508715,
-      "learning_rate": 0.00019998328296656453,
-      "loss": 46.0,
-      "step": 36153
-    },
-    {
-      "epoch": 5.822335842827811,
-      "grad_norm": 0.012875864282250404,
-      "learning_rate": 0.00019998328204152792,
-      "loss": 46.0,
-      "step": 36154
-    },
-    {
-      "epoch": 5.822496879906598,
-      "grad_norm": 0.001265470520593226,
-      "learning_rate": 0.00019998328111646573,
-      "loss": 46.0,
-      "step": 36155
-    },
-    {
-      "epoch": 5.822657916985386,
-      "grad_norm": 0.0006662584492005408,
-      "learning_rate": 0.00019998328019137795,
-      "loss": 46.0,
-      "step": 36156
-    },
-    {
-      "epoch": 5.822818954064173,
-      "grad_norm": 0.0029247228521853685,
-      "learning_rate": 0.00019998327926626459,
-      "loss": 46.0,
-      "step": 36157
-    },
-    {
-      "epoch": 5.822979991142961,
-      "grad_norm": 0.00492109265178442,
-      "learning_rate": 0.00019998327834112563,
-      "loss": 46.0,
-      "step": 36158
-    },
-    {
-      "epoch": 5.823141028221748,
-      "grad_norm": 0.010264286771416664,
-      "learning_rate": 0.00019998327741596107,
-      "loss": 46.0,
-      "step": 36159
-    },
-    {
-      "epoch": 5.823302065300536,
-      "grad_norm": 0.0062038651667535305,
-      "learning_rate": 0.00019998327649077094,
-      "loss": 46.0,
-      "step": 36160
-    },
-    {
-      "epoch": 5.823463102379323,
-      "grad_norm": 0.006900831591337919,
-      "learning_rate": 0.0001999832755655552,
-      "loss": 46.0,
-      "step": 36161
-    },
-    {
-      "epoch": 5.8236241394581105,
-      "grad_norm": 0.007807180285453796,
-      "learning_rate": 0.00019998327464031387,
-      "loss": 46.0,
-      "step": 36162
-    },
-    {
-      "epoch": 5.823785176536898,
-      "grad_norm": 0.0009305981802754104,
-      "learning_rate": 0.00019998327371504698,
-      "loss": 46.0,
-      "step": 36163
-    },
-    {
-      "epoch": 5.8239462136156845,
-      "grad_norm": 0.004487654194235802,
-      "learning_rate": 0.00019998327278975445,
-      "loss": 46.0,
-      "step": 36164
-    },
-    {
-      "epoch": 5.824107250694473,
-      "grad_norm": 0.0012005011085420847,
-      "learning_rate": 0.00019998327186443639,
-      "loss": 46.0,
-      "step": 36165
-    },
-    {
-      "epoch": 5.8242682877732594,
-      "grad_norm": 0.005168136674910784,
-      "learning_rate": 0.00019998327093909268,
-      "loss": 46.0,
-      "step": 36166
-    },
-    {
-      "epoch": 5.824429324852047,
-      "grad_norm": 0.006324934773147106,
-      "learning_rate": 0.00019998327001372342,
-      "loss": 46.0,
-      "step": 36167
-    },
-    {
-      "epoch": 5.824590361930834,
-      "grad_norm": 0.004080827347934246,
-      "learning_rate": 0.00019998326908832854,
-      "loss": 46.0,
-      "step": 36168
-    },
-    {
-      "epoch": 5.824751399009622,
-      "grad_norm": 0.012046740390360355,
-      "learning_rate": 0.0001999832681629081,
-      "loss": 46.0,
-      "step": 36169
-    },
-    {
-      "epoch": 5.824912436088409,
-      "grad_norm": 0.012234746478497982,
-      "learning_rate": 0.00019998326723746202,
-      "loss": 46.0,
-      "step": 36170
-    },
-    {
-      "epoch": 5.825073473167197,
-      "grad_norm": 0.0036519921850413084,
-      "learning_rate": 0.0001999832663119904,
-      "loss": 46.0,
-      "step": 36171
-    },
-    {
-      "epoch": 5.825234510245984,
-      "grad_norm": 0.00487400870770216,
-      "learning_rate": 0.00019998326538649318,
-      "loss": 46.0,
-      "step": 36172
-    },
-    {
-      "epoch": 5.825395547324772,
-      "grad_norm": 0.022288190200924873,
-      "learning_rate": 0.00019998326446097036,
-      "loss": 46.0,
-      "step": 36173
-    },
-    {
-      "epoch": 5.825556584403559,
-      "grad_norm": 0.00806272029876709,
-      "learning_rate": 0.00019998326353542193,
-      "loss": 46.0,
-      "step": 36174
-    },
-    {
-      "epoch": 5.8257176214823465,
-      "grad_norm": 0.007910950109362602,
-      "learning_rate": 0.00019998326260984794,
-      "loss": 46.0,
-      "step": 36175
-    },
-    {
-      "epoch": 5.825878658561134,
-      "grad_norm": 0.0019355147378519177,
-      "learning_rate": 0.00019998326168424834,
-      "loss": 46.0,
-      "step": 36176
-    },
-    {
-      "epoch": 5.826039695639921,
-      "grad_norm": 0.0026842239312827587,
-      "learning_rate": 0.00019998326075862318,
-      "loss": 46.0,
-      "step": 36177
-    },
-    {
-      "epoch": 5.826200732718709,
-      "grad_norm": 0.012238512746989727,
-      "learning_rate": 0.0001999832598329724,
-      "loss": 46.0,
-      "step": 36178
-    },
-    {
-      "epoch": 5.826361769797495,
-      "grad_norm": 0.0280852522701025,
-      "learning_rate": 0.00019998325890729603,
-      "loss": 46.0,
-      "step": 36179
-    },
-    {
-      "epoch": 5.826522806876283,
-      "grad_norm": 0.00547018414363265,
-      "learning_rate": 0.00019998325798159408,
-      "loss": 46.0,
-      "step": 36180
-    },
-    {
-      "epoch": 5.82668384395507,
-      "grad_norm": 0.005220636259764433,
-      "learning_rate": 0.00019998325705586654,
-      "loss": 46.0,
-      "step": 36181
-    },
-    {
-      "epoch": 5.826844881033858,
-      "grad_norm": 0.007964136078953743,
-      "learning_rate": 0.0001999832561301134,
-      "loss": 46.0,
-      "step": 36182
-    },
-    {
-      "epoch": 5.827005918112645,
-      "grad_norm": 0.0066356235183775425,
-      "learning_rate": 0.00019998325520433467,
-      "loss": 46.0,
-      "step": 36183
-    },
-    {
-      "epoch": 5.827166955191433,
-      "grad_norm": 0.005015928763896227,
-      "learning_rate": 0.00019998325427853035,
-      "loss": 46.0,
-      "step": 36184
-    },
-    {
-      "epoch": 5.82732799227022,
-      "grad_norm": 0.005267851520329714,
-      "learning_rate": 0.00019998325335270046,
-      "loss": 46.0,
-      "step": 36185
-    },
-    {
-      "epoch": 5.827489029349008,
-      "grad_norm": 0.000808143406175077,
-      "learning_rate": 0.00019998325242684496,
-      "loss": 46.0,
-      "step": 36186
-    },
-    {
-      "epoch": 5.827650066427795,
-      "grad_norm": 0.010387449525296688,
-      "learning_rate": 0.00019998325150096387,
-      "loss": 46.0,
-      "step": 36187
-    },
-    {
-      "epoch": 5.8278111035065825,
-      "grad_norm": 0.015588668175041676,
-      "learning_rate": 0.0001999832505750572,
-      "loss": 46.0,
-      "step": 36188
-    },
-    {
-      "epoch": 5.82797214058537,
-      "grad_norm": 0.010332321748137474,
-      "learning_rate": 0.00019998324964912493,
-      "loss": 46.0,
-      "step": 36189
-    },
-    {
-      "epoch": 5.828133177664157,
-      "grad_norm": 0.0015874354867264628,
-      "learning_rate": 0.00019998324872316705,
-      "loss": 46.0,
-      "step": 36190
-    },
-    {
-      "epoch": 5.828294214742945,
-      "grad_norm": 0.0019170109881088138,
-      "learning_rate": 0.0001999832477971836,
-      "loss": 46.0,
-      "step": 36191
-    },
-    {
-      "epoch": 5.828455251821732,
-      "grad_norm": 0.0032974437344819307,
-      "learning_rate": 0.00019998324687117456,
-      "loss": 46.0,
-      "step": 36192
-    },
-    {
-      "epoch": 5.82861628890052,
-      "grad_norm": 0.003428238909691572,
-      "learning_rate": 0.00019998324594513992,
-      "loss": 46.0,
-      "step": 36193
-    },
-    {
-      "epoch": 5.828777325979306,
-      "grad_norm": 0.006415089126676321,
-      "learning_rate": 0.00019998324501907972,
-      "loss": 46.0,
-      "step": 36194
-    },
-    {
-      "epoch": 5.828938363058094,
-      "grad_norm": 0.0022081760689616203,
-      "learning_rate": 0.0001999832440929939,
-      "loss": 46.0,
-      "step": 36195
-    },
-    {
-      "epoch": 5.829099400136881,
-      "grad_norm": 0.002088650595396757,
-      "learning_rate": 0.00019998324316688248,
-      "loss": 46.0,
-      "step": 36196
-    },
-    {
-      "epoch": 5.829260437215669,
-      "grad_norm": 0.0030758059583604336,
-      "learning_rate": 0.0001999832422407455,
-      "loss": 46.0,
-      "step": 36197
-    },
-    {
-      "epoch": 5.829421474294456,
-      "grad_norm": 0.001338680856861174,
-      "learning_rate": 0.0001999832413145829,
-      "loss": 46.0,
-      "step": 36198
-    },
-    {
-      "epoch": 5.829582511373244,
-      "grad_norm": 0.0012243675300851464,
-      "learning_rate": 0.00019998324038839472,
-      "loss": 46.0,
-      "step": 36199
-    },
-    {
-      "epoch": 5.829743548452031,
-      "grad_norm": 0.005510313436388969,
-      "learning_rate": 0.00019998323946218097,
-      "loss": 46.0,
-      "step": 36200
-    },
-    {
-      "epoch": 5.8299045855308185,
-      "grad_norm": 0.004977809730917215,
-      "learning_rate": 0.0001999832385359416,
-      "loss": 46.0,
-      "step": 36201
-    },
-    {
-      "epoch": 5.830065622609606,
-      "grad_norm": 0.002307595917955041,
-      "learning_rate": 0.00019998323760967668,
-      "loss": 46.0,
-      "step": 36202
-    },
-    {
-      "epoch": 5.830226659688393,
-      "grad_norm": 0.0046867793425917625,
-      "learning_rate": 0.00019998323668338615,
-      "loss": 46.0,
-      "step": 36203
-    },
-    {
-      "epoch": 5.830387696767181,
-      "grad_norm": 0.0035534542985260487,
-      "learning_rate": 0.00019998323575707002,
-      "loss": 46.0,
-      "step": 36204
-    },
-    {
-      "epoch": 5.830548733845968,
-      "grad_norm": 0.006141949445009232,
-      "learning_rate": 0.00019998323483072828,
-      "loss": 46.0,
-      "step": 36205
-    },
-    {
-      "epoch": 5.830709770924756,
-      "grad_norm": 0.007672695908695459,
-      "learning_rate": 0.00019998323390436098,
-      "loss": 46.0,
-      "step": 36206
-    },
-    {
-      "epoch": 5.830870808003543,
-      "grad_norm": 0.001081821508705616,
-      "learning_rate": 0.0001999832329779681,
-      "loss": 46.0,
-      "step": 36207
-    },
-    {
-      "epoch": 5.831031845082331,
-      "grad_norm": 0.0015900125727057457,
-      "learning_rate": 0.00019998323205154962,
-      "loss": 46.0,
-      "step": 36208
-    },
-    {
-      "epoch": 5.831192882161117,
-      "grad_norm": 0.006711884867399931,
-      "learning_rate": 0.00019998323112510553,
-      "loss": 46.0,
-      "step": 36209
-    },
-    {
-      "epoch": 5.831353919239905,
-      "grad_norm": 0.006019486580044031,
-      "learning_rate": 0.00019998323019863586,
-      "loss": 46.0,
-      "step": 36210
-    },
-    {
-      "epoch": 5.831514956318692,
-      "grad_norm": 0.0029552823398262262,
-      "learning_rate": 0.0001999832292721406,
-      "loss": 46.0,
-      "step": 36211
-    },
-    {
-      "epoch": 5.8316759933974796,
-      "grad_norm": 0.005867298226803541,
-      "learning_rate": 0.00019998322834561974,
-      "loss": 46.0,
-      "step": 36212
-    },
-    {
-      "epoch": 5.831837030476267,
-      "grad_norm": 0.011626696214079857,
-      "learning_rate": 0.0001999832274190733,
-      "loss": 46.0,
-      "step": 36213
-    },
-    {
-      "epoch": 5.8319980675550545,
-      "grad_norm": 0.003413523081690073,
-      "learning_rate": 0.00019998322649250125,
-      "loss": 46.0,
-      "step": 36214
-    },
-    {
-      "epoch": 5.832159104633842,
-      "grad_norm": 0.005220881197601557,
-      "learning_rate": 0.00019998322556590364,
-      "loss": 46.0,
-      "step": 36215
-    },
-    {
-      "epoch": 5.832320141712629,
-      "grad_norm": 0.005811305250972509,
-      "learning_rate": 0.00019998322463928042,
-      "loss": 46.0,
-      "step": 36216
-    },
-    {
-      "epoch": 5.832481178791417,
-      "grad_norm": 0.0025376961566507816,
-      "learning_rate": 0.0001999832237126316,
-      "loss": 46.0,
-      "step": 36217
-    },
-    {
-      "epoch": 5.832642215870204,
-      "grad_norm": 0.0055940719321370125,
-      "learning_rate": 0.00019998322278595723,
-      "loss": 46.0,
-      "step": 36218
-    },
-    {
-      "epoch": 5.832803252948992,
-      "grad_norm": 0.004168991465121508,
-      "learning_rate": 0.00019998322185925722,
-      "loss": 46.0,
-      "step": 36219
-    },
-    {
-      "epoch": 5.832964290027779,
-      "grad_norm": 0.0012310562888160348,
-      "learning_rate": 0.00019998322093253167,
-      "loss": 46.0,
-      "step": 36220
-    },
-    {
-      "epoch": 5.833125327106567,
-      "grad_norm": 0.0031461971811950207,
-      "learning_rate": 0.00019998322000578048,
-      "loss": 46.0,
-      "step": 36221
-    },
-    {
-      "epoch": 5.833286364185353,
-      "grad_norm": 0.0010055401362478733,
-      "learning_rate": 0.00019998321907900373,
-      "loss": 46.0,
-      "step": 36222
-    },
-    {
-      "epoch": 5.8334474012641415,
-      "grad_norm": 0.004415613599121571,
-      "learning_rate": 0.00019998321815220137,
-      "loss": 46.0,
-      "step": 36223
-    },
-    {
-      "epoch": 5.833608438342928,
-      "grad_norm": 0.008054523728787899,
-      "learning_rate": 0.00019998321722537344,
-      "loss": 46.0,
-      "step": 36224
-    },
-    {
-      "epoch": 5.8337694754217155,
-      "grad_norm": 0.0010442027123644948,
-      "learning_rate": 0.00019998321629851993,
-      "loss": 46.0,
-      "step": 36225
-    },
-    {
-      "epoch": 5.833930512500503,
-      "grad_norm": 0.0011952792992815375,
-      "learning_rate": 0.0001999832153716408,
-      "loss": 46.0,
-      "step": 36226
-    },
-    {
-      "epoch": 5.83409154957929,
-      "grad_norm": 0.005662546958774328,
-      "learning_rate": 0.0001999832144447361,
-      "loss": 46.0,
-      "step": 36227
-    },
-    {
-      "epoch": 5.834252586658078,
-      "grad_norm": 0.006468662992119789,
-      "learning_rate": 0.00019998321351780577,
-      "loss": 46.0,
-      "step": 36228
-    },
-    {
-      "epoch": 5.834413623736865,
-      "grad_norm": 0.006029517389833927,
-      "learning_rate": 0.00019998321259084988,
-      "loss": 46.0,
-      "step": 36229
-    },
-    {
-      "epoch": 5.834574660815653,
-      "grad_norm": 0.00356446229852736,
-      "learning_rate": 0.0001999832116638684,
-      "loss": 46.0,
-      "step": 36230
-    },
-    {
-      "epoch": 5.83473569789444,
-      "grad_norm": 0.0048367357812821865,
-      "learning_rate": 0.00019998321073686132,
-      "loss": 46.0,
-      "step": 36231
-    },
-    {
-      "epoch": 5.834896734973228,
-      "grad_norm": 0.0046705519780516624,
-      "learning_rate": 0.00019998320980982868,
-      "loss": 46.0,
-      "step": 36232
-    },
-    {
-      "epoch": 5.835057772052015,
-      "grad_norm": 0.005171570926904678,
-      "learning_rate": 0.00019998320888277041,
-      "loss": 46.0,
-      "step": 36233
-    },
-    {
-      "epoch": 5.835218809130803,
-      "grad_norm": 0.0043447124771773815,
-      "learning_rate": 0.00019998320795568657,
-      "loss": 46.0,
-      "step": 36234
-    },
-    {
-      "epoch": 5.83537984620959,
-      "grad_norm": 0.000931622285861522,
-      "learning_rate": 0.00019998320702857713,
-      "loss": 46.0,
-      "step": 36235
-    },
-    {
-      "epoch": 5.8355408832883775,
-      "grad_norm": 0.015586710534989834,
-      "learning_rate": 0.00019998320610144213,
-      "loss": 46.0,
-      "step": 36236
-    },
-    {
-      "epoch": 5.835701920367164,
-      "grad_norm": 0.00594808254390955,
-      "learning_rate": 0.0001999832051742815,
-      "loss": 46.0,
-      "step": 36237
-    },
-    {
-      "epoch": 5.835862957445952,
-      "grad_norm": 0.004046519286930561,
-      "learning_rate": 0.0001999832042470953,
-      "loss": 46.0,
-      "step": 36238
-    },
-    {
-      "epoch": 5.836023994524739,
-      "grad_norm": 0.009893613867461681,
-      "learning_rate": 0.00019998320331988348,
-      "loss": 46.0,
-      "step": 36239
-    },
-    {
-      "epoch": 5.836185031603526,
-      "grad_norm": 0.009663325734436512,
-      "learning_rate": 0.00019998320239264609,
-      "loss": 46.0,
-      "step": 36240
-    },
-    {
-      "epoch": 5.836346068682314,
-      "grad_norm": 0.0025598008651286364,
-      "learning_rate": 0.00019998320146538313,
-      "loss": 46.0,
-      "step": 36241
-    },
-    {
-      "epoch": 5.836507105761101,
-      "grad_norm": 0.006545356009155512,
-      "learning_rate": 0.00019998320053809455,
-      "loss": 46.0,
-      "step": 36242
-    },
-    {
-      "epoch": 5.836668142839889,
-      "grad_norm": 0.01965847983956337,
-      "learning_rate": 0.0001999831996107804,
-      "loss": 46.0,
-      "step": 36243
-    },
-    {
-      "epoch": 5.836829179918676,
-      "grad_norm": 0.007745584473013878,
-      "learning_rate": 0.00019998319868344064,
-      "loss": 46.0,
-      "step": 36244
-    },
-    {
-      "epoch": 5.836990216997464,
-      "grad_norm": 0.003792156232520938,
-      "learning_rate": 0.0001999831977560753,
-      "loss": 46.0,
-      "step": 36245
-    },
-    {
-      "epoch": 5.837151254076251,
-      "grad_norm": 0.0022242029663175344,
-      "learning_rate": 0.00019998319682868436,
-      "loss": 46.0,
-      "step": 36246
-    },
-    {
-      "epoch": 5.837312291155039,
-      "grad_norm": 0.005726187955588102,
-      "learning_rate": 0.00019998319590126785,
-      "loss": 46.0,
-      "step": 36247
-    },
-    {
-      "epoch": 5.837473328233826,
-      "grad_norm": 0.006139697041362524,
-      "learning_rate": 0.00019998319497382573,
-      "loss": 46.0,
-      "step": 36248
-    },
-    {
-      "epoch": 5.8376343653126135,
-      "grad_norm": 0.0064367749728262424,
-      "learning_rate": 0.00019998319404635804,
-      "loss": 46.0,
-      "step": 36249
-    },
-    {
-      "epoch": 5.837795402391401,
-      "grad_norm": 0.0029867093544453382,
-      "learning_rate": 0.00019998319311886474,
-      "loss": 46.0,
-      "step": 36250
-    },
-    {
-      "epoch": 5.837956439470188,
-      "grad_norm": 0.002268476877361536,
-      "learning_rate": 0.00019998319219134586,
-      "loss": 46.0,
-      "step": 36251
-    },
-    {
-      "epoch": 5.838117476548975,
-      "grad_norm": 0.002606463385745883,
-      "learning_rate": 0.00019998319126380139,
-      "loss": 46.0,
-      "step": 36252
-    },
-    {
-      "epoch": 5.838278513627763,
-      "grad_norm": 0.013271412812173367,
-      "learning_rate": 0.00019998319033623133,
-      "loss": 46.0,
-      "step": 36253
-    },
-    {
-      "epoch": 5.83843955070655,
-      "grad_norm": 0.009845655411481857,
-      "learning_rate": 0.00019998318940863565,
-      "loss": 46.0,
-      "step": 36254
-    },
-    {
-      "epoch": 5.838600587785337,
-      "grad_norm": 0.006141947582364082,
-      "learning_rate": 0.00019998318848101442,
-      "loss": 46.0,
-      "step": 36255
-    },
-    {
-      "epoch": 5.838761624864125,
-      "grad_norm": 0.00165889086201787,
-      "learning_rate": 0.00019998318755336757,
-      "loss": 46.0,
-      "step": 36256
-    },
-    {
-      "epoch": 5.838922661942912,
-      "grad_norm": 0.0033974875696003437,
-      "learning_rate": 0.00019998318662569516,
-      "loss": 46.0,
-      "step": 36257
-    },
-    {
-      "epoch": 5.8390836990217,
-      "grad_norm": 0.000862635497469455,
-      "learning_rate": 0.00019998318569799714,
-      "loss": 46.0,
-      "step": 36258
-    },
-    {
-      "epoch": 5.839244736100487,
-      "grad_norm": 0.002606874331831932,
-      "learning_rate": 0.0001999831847702735,
-      "loss": 46.0,
-      "step": 36259
-    },
-    {
-      "epoch": 5.839405773179275,
-      "grad_norm": 0.01638873852789402,
-      "learning_rate": 0.00019998318384252433,
-      "loss": 46.0,
-      "step": 36260
-    },
-    {
-      "epoch": 5.839566810258062,
-      "grad_norm": 0.008939722552895546,
-      "learning_rate": 0.00019998318291474952,
-      "loss": 46.0,
-      "step": 36261
-    },
-    {
-      "epoch": 5.8397278473368495,
-      "grad_norm": 0.021137308329343796,
-      "learning_rate": 0.00019998318198694915,
-      "loss": 46.0,
-      "step": 36262
-    },
-    {
-      "epoch": 5.839888884415637,
-      "grad_norm": 0.00622069276869297,
-      "learning_rate": 0.0001999831810591232,
-      "loss": 46.0,
-      "step": 36263
-    },
-    {
-      "epoch": 5.840049921494424,
-      "grad_norm": 0.002902314765378833,
-      "learning_rate": 0.00019998318013127162,
-      "loss": 46.0,
-      "step": 36264
-    },
-    {
-      "epoch": 5.840210958573212,
-      "grad_norm": 0.0034124949015676975,
-      "learning_rate": 0.00019998317920339446,
-      "loss": 46.0,
-      "step": 36265
-    },
-    {
-      "epoch": 5.840371995651999,
-      "grad_norm": 0.0028374777175486088,
-      "learning_rate": 0.00019998317827549174,
-      "loss": 46.0,
-      "step": 36266
-    },
-    {
-      "epoch": 5.840533032730786,
-      "grad_norm": 0.008170254528522491,
-      "learning_rate": 0.00019998317734756338,
-      "loss": 46.0,
-      "step": 36267
-    },
-    {
-      "epoch": 5.840694069809573,
-      "grad_norm": 0.0020993370562791824,
-      "learning_rate": 0.00019998317641960946,
-      "loss": 46.0,
-      "step": 36268
-    },
-    {
-      "epoch": 5.840855106888361,
-      "grad_norm": 0.028592543676495552,
-      "learning_rate": 0.00019998317549162995,
-      "loss": 46.0,
-      "step": 36269
-    },
-    {
-      "epoch": 5.841016143967148,
-      "grad_norm": 0.0036306146066635847,
-      "learning_rate": 0.00019998317456362483,
-      "loss": 46.0,
-      "step": 36270
-    },
-    {
-      "epoch": 5.841177181045936,
-      "grad_norm": 0.007978690788149834,
-      "learning_rate": 0.00019998317363559417,
-      "loss": 46.0,
-      "step": 36271
-    },
-    {
-      "epoch": 5.841338218124723,
-      "grad_norm": 0.022357061505317688,
-      "learning_rate": 0.00019998317270753785,
-      "loss": 46.0,
-      "step": 36272
-    },
-    {
-      "epoch": 5.8414992552035105,
-      "grad_norm": 0.0019512284779921174,
-      "learning_rate": 0.000199983171779456,
-      "loss": 46.0,
-      "step": 36273
-    },
-    {
-      "epoch": 5.841660292282298,
-      "grad_norm": 0.015916481614112854,
-      "learning_rate": 0.00019998317085134852,
-      "loss": 46.0,
-      "step": 36274
-    },
-    {
-      "epoch": 5.8418213293610854,
-      "grad_norm": 0.013155380263924599,
-      "learning_rate": 0.00019998316992321546,
-      "loss": 46.0,
-      "step": 36275
-    },
-    {
-      "epoch": 5.841982366439873,
-      "grad_norm": 0.003157972823828459,
-      "learning_rate": 0.0001999831689950568,
-      "loss": 46.0,
-      "step": 36276
-    },
-    {
-      "epoch": 5.84214340351866,
-      "grad_norm": 0.0020603672601282597,
-      "learning_rate": 0.00019998316806687258,
-      "loss": 46.0,
-      "step": 36277
-    },
-    {
-      "epoch": 5.842304440597448,
-      "grad_norm": 0.001891984255053103,
-      "learning_rate": 0.00019998316713866273,
-      "loss": 46.0,
-      "step": 36278
-    },
-    {
-      "epoch": 5.842465477676235,
-      "grad_norm": 0.014637486077845097,
-      "learning_rate": 0.00019998316621042732,
-      "loss": 46.0,
-      "step": 36279
-    },
-    {
-      "epoch": 5.842626514755023,
-      "grad_norm": 0.005434535443782806,
-      "learning_rate": 0.00019998316528216633,
-      "loss": 46.0,
-      "step": 36280
-    },
-    {
-      "epoch": 5.84278755183381,
-      "grad_norm": 0.011293580755591393,
-      "learning_rate": 0.00019998316435387972,
-      "loss": 46.0,
-      "step": 36281
-    },
-    {
-      "epoch": 5.842948588912597,
-      "grad_norm": 0.001783306710422039,
-      "learning_rate": 0.0001999831634255675,
-      "loss": 46.0,
-      "step": 36282
-    },
-    {
-      "epoch": 5.843109625991384,
-      "grad_norm": 0.020579611882567406,
-      "learning_rate": 0.00019998316249722974,
-      "loss": 46.0,
-      "step": 36283
-    },
-    {
-      "epoch": 5.843270663070172,
-      "grad_norm": 0.0007040241616778076,
-      "learning_rate": 0.00019998316156886637,
-      "loss": 46.0,
-      "step": 36284
-    },
-    {
-      "epoch": 5.843431700148959,
-      "grad_norm": 0.012248336337506771,
-      "learning_rate": 0.0001999831606404774,
-      "loss": 46.0,
-      "step": 36285
-    },
-    {
-      "epoch": 5.8435927372277465,
-      "grad_norm": 0.013011324219405651,
-      "learning_rate": 0.00019998315971206284,
-      "loss": 46.0,
-      "step": 36286
-    },
-    {
-      "epoch": 5.843753774306534,
-      "grad_norm": 0.00786654930561781,
-      "learning_rate": 0.0001999831587836227,
-      "loss": 46.0,
-      "step": 36287
-    },
-    {
-      "epoch": 5.843914811385321,
-      "grad_norm": 0.0006335290381684899,
-      "learning_rate": 0.00019998315785515696,
-      "loss": 46.0,
-      "step": 36288
-    },
-    {
-      "epoch": 5.844075848464109,
-      "grad_norm": 0.0006969252717681229,
-      "learning_rate": 0.00019998315692666566,
-      "loss": 46.0,
-      "step": 36289
-    },
-    {
-      "epoch": 5.844236885542896,
-      "grad_norm": 0.006457506213337183,
-      "learning_rate": 0.00019998315599814874,
-      "loss": 46.0,
-      "step": 36290
-    },
-    {
-      "epoch": 5.844397922621684,
-      "grad_norm": 0.0030615711584687233,
-      "learning_rate": 0.0001999831550696062,
-      "loss": 46.0,
-      "step": 36291
-    },
-    {
-      "epoch": 5.844558959700471,
-      "grad_norm": 0.005610724911093712,
-      "learning_rate": 0.0001999831541410381,
-      "loss": 46.0,
-      "step": 36292
-    },
-    {
-      "epoch": 5.844719996779259,
-      "grad_norm": 0.002621626714244485,
-      "learning_rate": 0.00019998315321244443,
-      "loss": 46.0,
-      "step": 36293
-    },
-    {
-      "epoch": 5.844881033858046,
-      "grad_norm": 0.005308116786181927,
-      "learning_rate": 0.00019998315228382516,
-      "loss": 46.0,
-      "step": 36294
-    },
-    {
-      "epoch": 5.845042070936834,
-      "grad_norm": 0.0016062383074313402,
-      "learning_rate": 0.00019998315135518027,
-      "loss": 46.0,
-      "step": 36295
-    },
-    {
-      "epoch": 5.845203108015621,
-      "grad_norm": 0.002875788602977991,
-      "learning_rate": 0.0001999831504265098,
-      "loss": 46.0,
-      "step": 36296
-    },
-    {
-      "epoch": 5.845364145094408,
-      "grad_norm": 0.005451788194477558,
-      "learning_rate": 0.00019998314949781377,
-      "loss": 46.0,
-      "step": 36297
-    },
-    {
-      "epoch": 5.845525182173195,
-      "grad_norm": 0.0073241665959358215,
-      "learning_rate": 0.00019998314856909213,
-      "loss": 46.0,
-      "step": 36298
-    },
-    {
-      "epoch": 5.8456862192519825,
-      "grad_norm": 0.003614874789491296,
-      "learning_rate": 0.00019998314764034492,
-      "loss": 46.0,
-      "step": 36299
-    },
-    {
-      "epoch": 5.84584725633077,
-      "grad_norm": 0.003140663728117943,
-      "learning_rate": 0.00019998314671157208,
-      "loss": 46.0,
-      "step": 36300
-    },
-    {
-      "epoch": 5.846008293409557,
-      "grad_norm": 0.005265682470053434,
-      "learning_rate": 0.00019998314578277367,
-      "loss": 46.0,
-      "step": 36301
-    },
-    {
-      "epoch": 5.846169330488345,
-      "grad_norm": 0.005553533788770437,
-      "learning_rate": 0.00019998314485394968,
-      "loss": 46.0,
-      "step": 36302
-    },
-    {
-      "epoch": 5.846330367567132,
-      "grad_norm": 0.0012782574631273746,
-      "learning_rate": 0.00019998314392510007,
-      "loss": 46.0,
-      "step": 36303
-    },
-    {
-      "epoch": 5.84649140464592,
-      "grad_norm": 0.0017012894386425614,
-      "learning_rate": 0.0001999831429962249,
-      "loss": 46.0,
-      "step": 36304
-    },
-    {
-      "epoch": 5.846652441724707,
-      "grad_norm": 0.0031299844849854708,
-      "learning_rate": 0.00019998314206732412,
-      "loss": 46.0,
-      "step": 36305
-    },
-    {
-      "epoch": 5.846813478803495,
-      "grad_norm": 0.007620898075401783,
-      "learning_rate": 0.00019998314113839775,
-      "loss": 46.0,
-      "step": 36306
-    },
-    {
-      "epoch": 5.846974515882282,
-      "grad_norm": 0.005849129986017942,
-      "learning_rate": 0.0001999831402094458,
-      "loss": 46.0,
-      "step": 36307
-    },
-    {
-      "epoch": 5.84713555296107,
-      "grad_norm": 0.0035931537859141827,
-      "learning_rate": 0.00019998313928046825,
-      "loss": 46.0,
-      "step": 36308
-    },
-    {
-      "epoch": 5.847296590039857,
-      "grad_norm": 0.0025375315453857183,
-      "learning_rate": 0.00019998313835146512,
-      "loss": 46.0,
-      "step": 36309
-    },
-    {
-      "epoch": 5.847457627118644,
-      "grad_norm": 0.010508900508284569,
-      "learning_rate": 0.0001999831374224364,
-      "loss": 46.0,
-      "step": 36310
-    },
-    {
-      "epoch": 5.847618664197432,
-      "grad_norm": 0.0019658776000142097,
-      "learning_rate": 0.00019998313649338204,
-      "loss": 46.0,
-      "step": 36311
-    },
-    {
-      "epoch": 5.8477797012762185,
-      "grad_norm": 0.00748297106474638,
-      "learning_rate": 0.00019998313556430217,
-      "loss": 46.0,
-      "step": 36312
-    },
-    {
-      "epoch": 5.847940738355006,
-      "grad_norm": 0.005840287543833256,
-      "learning_rate": 0.00019998313463519667,
-      "loss": 46.0,
-      "step": 36313
-    },
-    {
-      "epoch": 5.848101775433793,
-      "grad_norm": 0.010568608529865742,
-      "learning_rate": 0.00019998313370606557,
-      "loss": 46.0,
-      "step": 36314
-    },
-    {
-      "epoch": 5.848262812512581,
-      "grad_norm": 0.0030749221332371235,
-      "learning_rate": 0.00019998313277690887,
-      "loss": 46.0,
-      "step": 36315
-    },
-    {
-      "epoch": 5.848423849591368,
-      "grad_norm": 0.002867255825549364,
-      "learning_rate": 0.00019998313184772662,
-      "loss": 46.0,
-      "step": 36316
-    },
-    {
-      "epoch": 5.848584886670156,
-      "grad_norm": 0.007032209541648626,
-      "learning_rate": 0.00019998313091851877,
-      "loss": 46.0,
-      "step": 36317
-    },
-    {
-      "epoch": 5.848745923748943,
-      "grad_norm": 0.007477450184524059,
-      "learning_rate": 0.0001999831299892853,
-      "loss": 46.0,
-      "step": 36318
-    },
-    {
-      "epoch": 5.848906960827731,
-      "grad_norm": 0.012234101071953773,
-      "learning_rate": 0.00019998312906002624,
-      "loss": 46.0,
-      "step": 36319
-    },
-    {
-      "epoch": 5.849067997906518,
-      "grad_norm": 0.005514687858521938,
-      "learning_rate": 0.00019998312813074163,
-      "loss": 46.0,
-      "step": 36320
-    },
-    {
-      "epoch": 5.8492290349853056,
-      "grad_norm": 0.002884516492486,
-      "learning_rate": 0.00019998312720143142,
-      "loss": 46.0,
-      "step": 36321
-    },
-    {
-      "epoch": 5.849390072064093,
-      "grad_norm": 0.019594764336943626,
-      "learning_rate": 0.00019998312627209558,
-      "loss": 46.0,
-      "step": 36322
-    },
-    {
-      "epoch": 5.8495511091428805,
-      "grad_norm": 0.004947365261614323,
-      "learning_rate": 0.00019998312534273417,
-      "loss": 46.0,
-      "step": 36323
-    },
-    {
-      "epoch": 5.849712146221668,
-      "grad_norm": 0.00443473132327199,
-      "learning_rate": 0.00019998312441334718,
-      "loss": 46.0,
-      "step": 36324
-    },
-    {
-      "epoch": 5.8498731833004545,
-      "grad_norm": 0.002974113682284951,
-      "learning_rate": 0.00019998312348393462,
-      "loss": 46.0,
-      "step": 36325
-    },
-    {
-      "epoch": 5.850034220379243,
-      "grad_norm": 0.010956228710711002,
-      "learning_rate": 0.00019998312255449643,
-      "loss": 46.0,
-      "step": 36326
-    },
-    {
-      "epoch": 5.850195257458029,
-      "grad_norm": 0.0021672870498150587,
-      "learning_rate": 0.00019998312162503265,
-      "loss": 46.0,
-      "step": 36327
-    },
-    {
-      "epoch": 5.850356294536817,
-      "grad_norm": 0.0028575530741363764,
-      "learning_rate": 0.0001999831206955433,
-      "loss": 46.0,
-      "step": 36328
-    },
-    {
-      "epoch": 5.850517331615604,
-      "grad_norm": 0.00432627834379673,
-      "learning_rate": 0.00019998311976602838,
-      "loss": 46.0,
-      "step": 36329
-    },
-    {
-      "epoch": 5.850678368694392,
-      "grad_norm": 0.0027518507558852434,
-      "learning_rate": 0.0001999831188364878,
-      "loss": 46.0,
-      "step": 36330
-    },
-    {
-      "epoch": 5.850839405773179,
-      "grad_norm": 0.004824016243219376,
-      "learning_rate": 0.00019998311790692168,
-      "loss": 46.0,
-      "step": 36331
-    },
-    {
-      "epoch": 5.851000442851967,
-      "grad_norm": 0.0047799102030694485,
-      "learning_rate": 0.00019998311697732996,
-      "loss": 46.0,
-      "step": 36332
-    },
-    {
-      "epoch": 5.851161479930754,
-      "grad_norm": 0.004516626708209515,
-      "learning_rate": 0.00019998311604771266,
-      "loss": 46.0,
-      "step": 36333
-    },
-    {
-      "epoch": 5.8513225170095415,
-      "grad_norm": 0.0026720231398940086,
-      "learning_rate": 0.00019998311511806977,
-      "loss": 46.0,
-      "step": 36334
-    },
-    {
-      "epoch": 5.851483554088329,
-      "grad_norm": 0.003077418077737093,
-      "learning_rate": 0.00019998311418840126,
-      "loss": 46.0,
-      "step": 36335
-    },
-    {
-      "epoch": 5.851644591167116,
-      "grad_norm": 0.0037507705856114626,
-      "learning_rate": 0.0001999831132587072,
-      "loss": 46.0,
-      "step": 36336
-    },
-    {
-      "epoch": 5.851805628245904,
-      "grad_norm": 0.006240047048777342,
-      "learning_rate": 0.00019998311232898751,
-      "loss": 46.0,
-      "step": 36337
-    },
-    {
-      "epoch": 5.851966665324691,
-      "grad_norm": 0.0036264536902308464,
-      "learning_rate": 0.00019998311139924227,
-      "loss": 46.0,
-      "step": 36338
-    },
-    {
-      "epoch": 5.852127702403479,
-      "grad_norm": 0.0033457425888627768,
-      "learning_rate": 0.0001999831104694714,
-      "loss": 46.0,
-      "step": 36339
-    },
-    {
-      "epoch": 5.852288739482265,
-      "grad_norm": 0.002295734593644738,
-      "learning_rate": 0.00019998310953967495,
-      "loss": 46.0,
-      "step": 36340
-    },
-    {
-      "epoch": 5.852449776561053,
-      "grad_norm": 0.014742361381649971,
-      "learning_rate": 0.00019998310860985292,
-      "loss": 46.0,
-      "step": 36341
-    },
-    {
-      "epoch": 5.85261081363984,
-      "grad_norm": 0.0023084236308932304,
-      "learning_rate": 0.0001999831076800053,
-      "loss": 46.0,
-      "step": 36342
-    },
-    {
-      "epoch": 5.852771850718628,
-      "grad_norm": 0.0025524224620312452,
-      "learning_rate": 0.00019998310675013208,
-      "loss": 46.0,
-      "step": 36343
-    },
-    {
-      "epoch": 5.852932887797415,
-      "grad_norm": 0.0047472259029746056,
-      "learning_rate": 0.00019998310582023329,
-      "loss": 46.0,
-      "step": 36344
-    },
-    {
-      "epoch": 5.853093924876203,
-      "grad_norm": 0.003503593150526285,
-      "learning_rate": 0.00019998310489030888,
-      "loss": 46.0,
-      "step": 36345
-    },
-    {
-      "epoch": 5.85325496195499,
-      "grad_norm": 0.004680312238633633,
-      "learning_rate": 0.0001999831039603589,
-      "loss": 46.0,
-      "step": 36346
-    },
-    {
-      "epoch": 5.8534159990337775,
-      "grad_norm": 0.002561883069574833,
-      "learning_rate": 0.0001999831030303833,
-      "loss": 46.0,
-      "step": 36347
-    },
-    {
-      "epoch": 5.853577036112565,
-      "grad_norm": 0.007297783624380827,
-      "learning_rate": 0.00019998310210038214,
-      "loss": 46.0,
-      "step": 36348
-    },
-    {
-      "epoch": 5.853738073191352,
-      "grad_norm": 0.003774170530959964,
-      "learning_rate": 0.00019998310117035536,
-      "loss": 46.0,
-      "step": 36349
-    },
-    {
-      "epoch": 5.85389911027014,
-      "grad_norm": 0.006983309984207153,
-      "learning_rate": 0.00019998310024030302,
-      "loss": 46.0,
-      "step": 36350
-    },
-    {
-      "epoch": 5.854060147348927,
-      "grad_norm": 0.0037038393784314394,
-      "learning_rate": 0.0001999830993102251,
-      "loss": 46.0,
-      "step": 36351
-    },
-    {
-      "epoch": 5.854221184427715,
-      "grad_norm": 0.003117919433861971,
-      "learning_rate": 0.00019998309838012158,
-      "loss": 46.0,
-      "step": 36352
-    },
-    {
-      "epoch": 5.854382221506502,
-      "grad_norm": 0.0026418811175972223,
-      "learning_rate": 0.00019998309744999245,
-      "loss": 46.0,
-      "step": 36353
-    },
-    {
-      "epoch": 5.85454325858529,
-      "grad_norm": 0.004526711534708738,
-      "learning_rate": 0.00019998309651983774,
-      "loss": 46.0,
-      "step": 36354
-    },
-    {
-      "epoch": 5.854704295664076,
-      "grad_norm": 0.002510344609618187,
-      "learning_rate": 0.00019998309558965743,
-      "loss": 46.0,
-      "step": 36355
-    },
-    {
-      "epoch": 5.854865332742864,
-      "grad_norm": 0.005148347932845354,
-      "learning_rate": 0.00019998309465945154,
-      "loss": 46.0,
-      "step": 36356
-    },
-    {
-      "epoch": 5.855026369821651,
-      "grad_norm": 0.010461927391588688,
-      "learning_rate": 0.00019998309372922006,
-      "loss": 46.0,
-      "step": 36357
-    },
-    {
-      "epoch": 5.855187406900439,
-      "grad_norm": 0.002579952822998166,
-      "learning_rate": 0.000199983092798963,
-      "loss": 46.0,
-      "step": 36358
-    },
-    {
-      "epoch": 5.855348443979226,
-      "grad_norm": 0.0033212583512067795,
-      "learning_rate": 0.00019998309186868032,
-      "loss": 46.0,
-      "step": 36359
-    },
-    {
-      "epoch": 5.8555094810580135,
-      "grad_norm": 0.0040848152711987495,
-      "learning_rate": 0.00019998309093837205,
-      "loss": 46.0,
-      "step": 36360
-    },
-    {
-      "epoch": 5.855670518136801,
-      "grad_norm": 0.007760020438581705,
-      "learning_rate": 0.0001999830900080382,
-      "loss": 46.0,
-      "step": 36361
-    },
-    {
-      "epoch": 5.855831555215588,
-      "grad_norm": 0.0017486640717834234,
-      "learning_rate": 0.00019998308907767879,
-      "loss": 46.0,
-      "step": 36362
-    },
-    {
-      "epoch": 5.855992592294376,
-      "grad_norm": 0.0010235154768452048,
-      "learning_rate": 0.00019998308814729376,
-      "loss": 46.0,
-      "step": 36363
-    },
-    {
-      "epoch": 5.856153629373163,
-      "grad_norm": 0.002583742141723633,
-      "learning_rate": 0.00019998308721688312,
-      "loss": 46.0,
-      "step": 36364
-    },
-    {
-      "epoch": 5.856314666451951,
-      "grad_norm": 0.0033911135978996754,
-      "learning_rate": 0.00019998308628644694,
-      "loss": 46.0,
-      "step": 36365
-    },
-    {
-      "epoch": 5.856475703530738,
-      "grad_norm": 0.006008876487612724,
-      "learning_rate": 0.00019998308535598512,
-      "loss": 46.0,
-      "step": 36366
-    },
-    {
-      "epoch": 5.856636740609526,
-      "grad_norm": 0.008264592848718166,
-      "learning_rate": 0.00019998308442549775,
-      "loss": 46.0,
-      "step": 36367
-    },
-    {
-      "epoch": 5.856797777688313,
-      "grad_norm": 0.0016596839996054769,
-      "learning_rate": 0.00019998308349498478,
-      "loss": 46.0,
-      "step": 36368
-    },
-    {
-      "epoch": 5.856958814767101,
-      "grad_norm": 0.0009562142658978701,
-      "learning_rate": 0.0001999830825644462,
-      "loss": 46.0,
-      "step": 36369
-    },
-    {
-      "epoch": 5.857119851845887,
-      "grad_norm": 0.00973709486424923,
-      "learning_rate": 0.00019998308163388204,
-      "loss": 46.0,
-      "step": 36370
-    },
-    {
-      "epoch": 5.857280888924675,
-      "grad_norm": 0.011221935041248798,
-      "learning_rate": 0.0001999830807032923,
-      "loss": 46.0,
-      "step": 36371
-    },
-    {
-      "epoch": 5.857441926003462,
-      "grad_norm": 0.019078610464930534,
-      "learning_rate": 0.00019998307977267695,
-      "loss": 46.0,
-      "step": 36372
-    },
-    {
-      "epoch": 5.8576029630822495,
-      "grad_norm": 0.016757983714342117,
-      "learning_rate": 0.000199983078842036,
-      "loss": 46.0,
-      "step": 36373
-    },
-    {
-      "epoch": 5.857764000161037,
-      "grad_norm": 0.008735565468668938,
-      "learning_rate": 0.00019998307791136948,
-      "loss": 46.0,
-      "step": 36374
-    },
-    {
-      "epoch": 5.857925037239824,
-      "grad_norm": 0.0015567387454211712,
-      "learning_rate": 0.00019998307698067738,
-      "loss": 46.0,
-      "step": 36375
-    },
-    {
-      "epoch": 5.858086074318612,
-      "grad_norm": 0.005044006276875734,
-      "learning_rate": 0.00019998307604995966,
-      "loss": 46.0,
-      "step": 36376
-    },
-    {
-      "epoch": 5.858247111397399,
-      "grad_norm": 0.0026960745453834534,
-      "learning_rate": 0.0001999830751192164,
-      "loss": 46.0,
-      "step": 36377
-    },
-    {
-      "epoch": 5.858408148476187,
-      "grad_norm": 0.017673548310995102,
-      "learning_rate": 0.0001999830741884475,
-      "loss": 46.0,
-      "step": 36378
-    },
-    {
-      "epoch": 5.858569185554974,
-      "grad_norm": 0.002498205518350005,
-      "learning_rate": 0.00019998307325765302,
-      "loss": 46.0,
-      "step": 36379
-    },
-    {
-      "epoch": 5.858730222633762,
-      "grad_norm": 0.0016205819556489587,
-      "learning_rate": 0.00019998307232683296,
-      "loss": 46.0,
-      "step": 36380
-    },
-    {
-      "epoch": 5.858891259712549,
-      "grad_norm": 0.0027848598547279835,
-      "learning_rate": 0.00019998307139598728,
-      "loss": 46.0,
-      "step": 36381
-    },
-    {
-      "epoch": 5.8590522967913365,
-      "grad_norm": 0.0014832587912678719,
-      "learning_rate": 0.00019998307046511604,
-      "loss": 46.0,
-      "step": 36382
-    },
-    {
-      "epoch": 5.859213333870123,
-      "grad_norm": 0.003027052618563175,
-      "learning_rate": 0.00019998306953421922,
-      "loss": 46.0,
-      "step": 36383
-    },
-    {
-      "epoch": 5.8593743709489114,
-      "grad_norm": 0.0014254313427954912,
-      "learning_rate": 0.00019998306860329678,
-      "loss": 46.0,
-      "step": 36384
-    },
-    {
-      "epoch": 5.859535408027698,
-      "grad_norm": 0.00411205505952239,
-      "learning_rate": 0.00019998306767234878,
-      "loss": 46.0,
-      "step": 36385
-    },
-    {
-      "epoch": 5.8596964451064855,
-      "grad_norm": 0.0035211164504289627,
-      "learning_rate": 0.00019998306674137516,
-      "loss": 46.0,
-      "step": 36386
-    },
-    {
-      "epoch": 5.859857482185273,
-      "grad_norm": 0.008908253163099289,
-      "learning_rate": 0.00019998306581037596,
-      "loss": 46.0,
-      "step": 36387
-    },
-    {
-      "epoch": 5.86001851926406,
-      "grad_norm": 0.011570955626666546,
-      "learning_rate": 0.00019998306487935117,
-      "loss": 46.0,
-      "step": 36388
-    },
-    {
-      "epoch": 5.860179556342848,
-      "grad_norm": 0.002055235905572772,
-      "learning_rate": 0.00019998306394830077,
-      "loss": 46.0,
-      "step": 36389
-    },
-    {
-      "epoch": 5.860340593421635,
-      "grad_norm": 0.010774900205433369,
-      "learning_rate": 0.0001999830630172248,
-      "loss": 46.0,
-      "step": 36390
-    },
-    {
-      "epoch": 5.860501630500423,
-      "grad_norm": 0.011273350566625595,
-      "learning_rate": 0.00019998306208612326,
-      "loss": 46.0,
-      "step": 36391
-    },
-    {
-      "epoch": 5.86066266757921,
-      "grad_norm": 0.008249950595200062,
-      "learning_rate": 0.00019998306115499612,
-      "loss": 46.0,
-      "step": 36392
-    },
-    {
-      "epoch": 5.860823704657998,
-      "grad_norm": 0.010889701545238495,
-      "learning_rate": 0.00019998306022384337,
-      "loss": 46.0,
-      "step": 36393
-    },
-    {
-      "epoch": 5.860984741736785,
-      "grad_norm": 0.0038173669017851353,
-      "learning_rate": 0.00019998305929266503,
-      "loss": 46.0,
-      "step": 36394
-    },
-    {
-      "epoch": 5.8611457788155725,
-      "grad_norm": 0.0039767855778336525,
-      "learning_rate": 0.00019998305836146113,
-      "loss": 46.0,
-      "step": 36395
-    },
-    {
-      "epoch": 5.86130681589436,
-      "grad_norm": 0.0031858470756560564,
-      "learning_rate": 0.0001999830574302316,
-      "loss": 46.0,
-      "step": 36396
-    },
-    {
-      "epoch": 5.861467852973147,
-      "grad_norm": 0.0016732686199247837,
-      "learning_rate": 0.0001999830564989765,
-      "loss": 46.0,
-      "step": 36397
-    },
-    {
-      "epoch": 5.861628890051934,
-      "grad_norm": 0.003488459624350071,
-      "learning_rate": 0.00019998305556769578,
-      "loss": 46.0,
-      "step": 36398
-    },
-    {
-      "epoch": 5.861789927130722,
-      "grad_norm": 0.004864083137363195,
-      "learning_rate": 0.0001999830546363895,
-      "loss": 46.0,
-      "step": 36399
-    },
-    {
-      "epoch": 5.861950964209509,
-      "grad_norm": 0.011952180415391922,
-      "learning_rate": 0.00019998305370505764,
-      "loss": 46.0,
-      "step": 36400
-    },
-    {
-      "epoch": 5.862112001288296,
-      "grad_norm": 0.009938381612300873,
-      "learning_rate": 0.00019998305277370016,
-      "loss": 46.0,
-      "step": 36401
-    },
-    {
-      "epoch": 5.862273038367084,
-      "grad_norm": 0.0026790171395987272,
-      "learning_rate": 0.0001999830518423171,
-      "loss": 46.0,
-      "step": 36402
-    },
-    {
-      "epoch": 5.862434075445871,
-      "grad_norm": 0.0014267965452745557,
-      "learning_rate": 0.00019998305091090845,
-      "loss": 46.0,
-      "step": 36403
-    },
-    {
-      "epoch": 5.862595112524659,
-      "grad_norm": 0.00914012175053358,
-      "learning_rate": 0.00019998304997947424,
-      "loss": 46.0,
-      "step": 36404
-    },
-    {
-      "epoch": 5.862756149603446,
-      "grad_norm": 0.002978983335196972,
-      "learning_rate": 0.0001999830490480144,
-      "loss": 46.0,
-      "step": 36405
-    },
-    {
-      "epoch": 5.862917186682234,
-      "grad_norm": 0.01818298175930977,
-      "learning_rate": 0.00019998304811652898,
-      "loss": 46.0,
-      "step": 36406
-    },
-    {
-      "epoch": 5.863078223761021,
-      "grad_norm": 0.002179454080760479,
-      "learning_rate": 0.00019998304718501798,
-      "loss": 46.0,
-      "step": 36407
-    },
-    {
-      "epoch": 5.8632392608398085,
-      "grad_norm": 0.008454645052552223,
-      "learning_rate": 0.00019998304625348136,
-      "loss": 46.0,
-      "step": 36408
-    },
-    {
-      "epoch": 5.863400297918596,
-      "grad_norm": 0.0026484730187803507,
-      "learning_rate": 0.0001999830453219192,
-      "loss": 46.0,
-      "step": 36409
-    },
-    {
-      "epoch": 5.863561334997383,
-      "grad_norm": 0.00836569257080555,
-      "learning_rate": 0.0001999830443903314,
-      "loss": 46.0,
-      "step": 36410
-    },
-    {
-      "epoch": 5.863722372076171,
-      "grad_norm": 0.004000257700681686,
-      "learning_rate": 0.000199983043458718,
-      "loss": 46.0,
-      "step": 36411
-    },
-    {
-      "epoch": 5.863883409154958,
-      "grad_norm": 0.0014719569589942694,
-      "learning_rate": 0.00019998304252707904,
-      "loss": 46.0,
-      "step": 36412
-    },
-    {
-      "epoch": 5.864044446233745,
-      "grad_norm": 0.009013321250677109,
-      "learning_rate": 0.0001999830415954145,
-      "loss": 46.0,
-      "step": 36413
-    },
-    {
-      "epoch": 5.864205483312532,
-      "grad_norm": 0.0021256129257380962,
-      "learning_rate": 0.00019998304066372435,
-      "loss": 46.0,
-      "step": 36414
-    },
-    {
-      "epoch": 5.86436652039132,
-      "grad_norm": 0.004540143068879843,
-      "learning_rate": 0.00019998303973200863,
-      "loss": 46.0,
-      "step": 36415
-    },
-    {
-      "epoch": 5.864527557470107,
-      "grad_norm": 0.003974375780671835,
-      "learning_rate": 0.0001999830388002673,
-      "loss": 46.0,
-      "step": 36416
-    },
-    {
-      "epoch": 5.864688594548895,
-      "grad_norm": 0.002385141095146537,
-      "learning_rate": 0.00019998303786850037,
-      "loss": 46.0,
-      "step": 36417
-    },
-    {
-      "epoch": 5.864849631627682,
-      "grad_norm": 0.010230368934571743,
-      "learning_rate": 0.00019998303693670788,
-      "loss": 46.0,
-      "step": 36418
-    },
-    {
-      "epoch": 5.86501066870647,
-      "grad_norm": 0.016091374680399895,
-      "learning_rate": 0.00019998303600488976,
-      "loss": 46.0,
-      "step": 36419
-    },
-    {
-      "epoch": 5.865171705785257,
-      "grad_norm": 0.003257193835452199,
-      "learning_rate": 0.0001999830350730461,
-      "loss": 46.0,
-      "step": 36420
-    },
-    {
-      "epoch": 5.8653327428640445,
-      "grad_norm": 0.004660800565034151,
-      "learning_rate": 0.0001999830341411768,
-      "loss": 46.0,
-      "step": 36421
-    },
-    {
-      "epoch": 5.865493779942832,
-      "grad_norm": 0.01390034519135952,
-      "learning_rate": 0.00019998303320928194,
-      "loss": 46.0,
-      "step": 36422
-    },
-    {
-      "epoch": 5.865654817021619,
-      "grad_norm": 0.005368242971599102,
-      "learning_rate": 0.0001999830322773615,
-      "loss": 46.0,
-      "step": 36423
-    },
-    {
-      "epoch": 5.865815854100407,
-      "grad_norm": 0.004282278008759022,
-      "learning_rate": 0.00019998303134541545,
-      "loss": 46.0,
-      "step": 36424
-    },
-    {
-      "epoch": 5.865976891179194,
-      "grad_norm": 0.0017107764724642038,
-      "learning_rate": 0.0001999830304134438,
-      "loss": 46.0,
-      "step": 36425
-    },
-    {
-      "epoch": 5.866137928257982,
-      "grad_norm": 0.011549933813512325,
-      "learning_rate": 0.00019998302948144657,
-      "loss": 46.0,
-      "step": 36426
-    },
-    {
-      "epoch": 5.866298965336769,
-      "grad_norm": 0.006075679790228605,
-      "learning_rate": 0.00019998302854942374,
-      "loss": 46.0,
-      "step": 36427
-    },
-    {
-      "epoch": 5.866460002415556,
-      "grad_norm": 0.00539898918941617,
-      "learning_rate": 0.00019998302761737536,
-      "loss": 46.0,
-      "step": 36428
-    },
-    {
-      "epoch": 5.866621039494343,
-      "grad_norm": 0.010720433667302132,
-      "learning_rate": 0.00019998302668530133,
-      "loss": 46.0,
-      "step": 36429
-    },
-    {
-      "epoch": 5.866782076573131,
-      "grad_norm": 0.003219319274649024,
-      "learning_rate": 0.00019998302575320175,
-      "loss": 46.0,
-      "step": 36430
-    },
-    {
-      "epoch": 5.866943113651918,
-      "grad_norm": 0.01537872664630413,
-      "learning_rate": 0.00019998302482107655,
-      "loss": 46.0,
-      "step": 36431
-    },
-    {
-      "epoch": 5.867104150730706,
-      "grad_norm": 0.003630524268373847,
-      "learning_rate": 0.0001999830238889258,
-      "loss": 46.0,
-      "step": 36432
-    },
-    {
-      "epoch": 5.867265187809493,
-      "grad_norm": 0.005842970684170723,
-      "learning_rate": 0.00019998302295674942,
-      "loss": 46.0,
-      "step": 36433
-    },
-    {
-      "epoch": 5.8674262248882805,
-      "grad_norm": 0.003947276622056961,
-      "learning_rate": 0.00019998302202454745,
-      "loss": 46.0,
-      "step": 36434
-    },
-    {
-      "epoch": 5.867587261967068,
-      "grad_norm": 0.006555921398103237,
-      "learning_rate": 0.00019998302109231993,
-      "loss": 46.0,
-      "step": 36435
-    },
-    {
-      "epoch": 5.867748299045855,
-      "grad_norm": 0.006180745083838701,
-      "learning_rate": 0.0001999830201600668,
-      "loss": 46.0,
-      "step": 36436
-    },
-    {
-      "epoch": 5.867909336124643,
-      "grad_norm": 0.024030642583966255,
-      "learning_rate": 0.00019998301922778805,
-      "loss": 46.0,
-      "step": 36437
-    },
-    {
-      "epoch": 5.86807037320343,
-      "grad_norm": 0.006887456867843866,
-      "learning_rate": 0.00019998301829548374,
-      "loss": 46.0,
-      "step": 36438
-    },
-    {
-      "epoch": 5.868231410282218,
-      "grad_norm": 0.020555047318339348,
-      "learning_rate": 0.00019998301736315384,
-      "loss": 46.0,
-      "step": 36439
-    },
-    {
-      "epoch": 5.868392447361005,
-      "grad_norm": 0.01198846660554409,
-      "learning_rate": 0.00019998301643079836,
-      "loss": 46.0,
-      "step": 36440
-    },
-    {
-      "epoch": 5.868553484439793,
-      "grad_norm": 0.010833866894245148,
-      "learning_rate": 0.00019998301549841726,
-      "loss": 46.0,
-      "step": 36441
-    },
-    {
-      "epoch": 5.86871452151858,
-      "grad_norm": 0.004562126472592354,
-      "learning_rate": 0.00019998301456601057,
-      "loss": 46.0,
-      "step": 36442
-    },
-    {
-      "epoch": 5.868875558597367,
-      "grad_norm": 0.006616507191210985,
-      "learning_rate": 0.0001999830136335783,
-      "loss": 46.0,
-      "step": 36443
-    },
-    {
-      "epoch": 5.869036595676154,
-      "grad_norm": 0.001832295092754066,
-      "learning_rate": 0.00019998301270112044,
-      "loss": 46.0,
-      "step": 36444
-    },
-    {
-      "epoch": 5.8691976327549416,
-      "grad_norm": 0.0022298151161521673,
-      "learning_rate": 0.000199983011768637,
-      "loss": 46.0,
-      "step": 36445
-    },
-    {
-      "epoch": 5.869358669833729,
-      "grad_norm": 0.007917076349258423,
-      "learning_rate": 0.00019998301083612796,
-      "loss": 46.0,
-      "step": 36446
-    },
-    {
-      "epoch": 5.8695197069125165,
-      "grad_norm": 0.011314750649034977,
-      "learning_rate": 0.00019998300990359334,
-      "loss": 46.0,
-      "step": 36447
-    },
-    {
-      "epoch": 5.869680743991304,
-      "grad_norm": 0.00800985749810934,
-      "learning_rate": 0.0001999830089710331,
-      "loss": 46.0,
-      "step": 36448
-    },
-    {
-      "epoch": 5.869841781070091,
-      "grad_norm": 0.0070522441528737545,
-      "learning_rate": 0.0001999830080384473,
-      "loss": 46.0,
-      "step": 36449
-    },
-    {
-      "epoch": 5.870002818148879,
-      "grad_norm": 0.008857863955199718,
-      "learning_rate": 0.0001999830071058359,
-      "loss": 46.0,
-      "step": 36450
-    },
-    {
-      "epoch": 5.870163855227666,
-      "grad_norm": 0.007319858763366938,
-      "learning_rate": 0.00019998300617319893,
-      "loss": 46.0,
-      "step": 36451
-    },
-    {
-      "epoch": 5.870324892306454,
-      "grad_norm": 0.002215184736996889,
-      "learning_rate": 0.00019998300524053632,
-      "loss": 46.0,
-      "step": 36452
-    },
-    {
-      "epoch": 5.870485929385241,
-      "grad_norm": 0.006677348166704178,
-      "learning_rate": 0.00019998300430784814,
-      "loss": 46.0,
-      "step": 36453
-    },
-    {
-      "epoch": 5.870646966464029,
-      "grad_norm": 0.00866650603711605,
-      "learning_rate": 0.00019998300337513439,
-      "loss": 46.0,
-      "step": 36454
-    },
-    {
-      "epoch": 5.870808003542816,
-      "grad_norm": 0.002989734522998333,
-      "learning_rate": 0.00019998300244239504,
-      "loss": 46.0,
-      "step": 36455
-    },
-    {
-      "epoch": 5.870969040621603,
-      "grad_norm": 0.014965535141527653,
-      "learning_rate": 0.00019998300150963008,
-      "loss": 46.0,
-      "step": 36456
-    },
-    {
-      "epoch": 5.871130077700391,
-      "grad_norm": 0.000794540683273226,
-      "learning_rate": 0.00019998300057683953,
-      "loss": 46.0,
-      "step": 36457
-    },
-    {
-      "epoch": 5.8712911147791775,
-      "grad_norm": 0.0042830901220440865,
-      "learning_rate": 0.0001999829996440234,
-      "loss": 46.0,
-      "step": 36458
-    },
-    {
-      "epoch": 5.871452151857965,
-      "grad_norm": 0.002588928211480379,
-      "learning_rate": 0.00019998299871118168,
-      "loss": 46.0,
-      "step": 36459
-    },
-    {
-      "epoch": 5.871613188936752,
-      "grad_norm": 0.001984506845474243,
-      "learning_rate": 0.0001999829977783144,
-      "loss": 46.0,
-      "step": 36460
-    },
-    {
-      "epoch": 5.87177422601554,
-      "grad_norm": 0.005372058600187302,
-      "learning_rate": 0.00019998299684542147,
-      "loss": 46.0,
-      "step": 36461
-    },
-    {
-      "epoch": 5.871935263094327,
-      "grad_norm": 0.00507509196177125,
-      "learning_rate": 0.000199982995912503,
-      "loss": 46.0,
-      "step": 36462
-    },
-    {
-      "epoch": 5.872096300173115,
-      "grad_norm": 0.004046759568154812,
-      "learning_rate": 0.00019998299497955892,
-      "loss": 46.0,
-      "step": 36463
-    },
-    {
-      "epoch": 5.872257337251902,
-      "grad_norm": 0.010963904671370983,
-      "learning_rate": 0.00019998299404658923,
-      "loss": 46.0,
-      "step": 36464
-    },
-    {
-      "epoch": 5.87241837433069,
-      "grad_norm": 0.012687266804277897,
-      "learning_rate": 0.00019998299311359402,
-      "loss": 46.0,
-      "step": 36465
-    },
-    {
-      "epoch": 5.872579411409477,
-      "grad_norm": 0.004060177132487297,
-      "learning_rate": 0.00019998299218057316,
-      "loss": 46.0,
-      "step": 36466
-    },
-    {
-      "epoch": 5.872740448488265,
-      "grad_norm": 0.0034510409459471703,
-      "learning_rate": 0.0001999829912475267,
-      "loss": 46.0,
-      "step": 36467
-    },
-    {
-      "epoch": 5.872901485567052,
-      "grad_norm": 0.013134202919900417,
-      "learning_rate": 0.00019998299031445468,
-      "loss": 46.0,
-      "step": 36468
-    },
-    {
-      "epoch": 5.8730625226458395,
-      "grad_norm": 0.0029541996773332357,
-      "learning_rate": 0.00019998298938135703,
-      "loss": 46.0,
-      "step": 36469
-    },
-    {
-      "epoch": 5.873223559724627,
-      "grad_norm": 0.007828726433217525,
-      "learning_rate": 0.00019998298844823382,
-      "loss": 46.0,
-      "step": 36470
-    },
-    {
-      "epoch": 5.8733845968034135,
-      "grad_norm": 0.017941521480679512,
-      "learning_rate": 0.00019998298751508503,
-      "loss": 46.0,
-      "step": 36471
-    },
-    {
-      "epoch": 5.873545633882202,
-      "grad_norm": 0.002676720265299082,
-      "learning_rate": 0.00019998298658191065,
-      "loss": 46.0,
-      "step": 36472
-    },
-    {
-      "epoch": 5.873706670960988,
-      "grad_norm": 0.004685257561504841,
-      "learning_rate": 0.00019998298564871065,
-      "loss": 46.0,
-      "step": 36473
-    },
-    {
-      "epoch": 5.873867708039776,
-      "grad_norm": 0.007790350820869207,
-      "learning_rate": 0.0001999829847154851,
-      "loss": 46.0,
-      "step": 36474
-    },
-    {
-      "epoch": 5.874028745118563,
-      "grad_norm": 0.005516765173524618,
-      "learning_rate": 0.00019998298378223392,
-      "loss": 46.0,
-      "step": 36475
-    },
-    {
-      "epoch": 5.874189782197351,
-      "grad_norm": 0.003555112751200795,
-      "learning_rate": 0.00019998298284895717,
-      "loss": 46.0,
-      "step": 36476
-    },
-    {
-      "epoch": 5.874350819276138,
-      "grad_norm": 0.005349715240299702,
-      "learning_rate": 0.00019998298191565482,
-      "loss": 46.0,
-      "step": 36477
-    },
-    {
-      "epoch": 5.874511856354926,
-      "grad_norm": 0.01286313682794571,
-      "learning_rate": 0.00019998298098232686,
-      "loss": 46.0,
-      "step": 36478
-    },
-    {
-      "epoch": 5.874672893433713,
-      "grad_norm": 0.003224862040951848,
-      "learning_rate": 0.00019998298004897334,
-      "loss": 46.0,
-      "step": 36479
-    },
-    {
-      "epoch": 5.874833930512501,
-      "grad_norm": 0.001877399394288659,
-      "learning_rate": 0.00019998297911559424,
-      "loss": 46.0,
-      "step": 36480
-    },
-    {
-      "epoch": 5.874994967591288,
-      "grad_norm": 0.0032142880372703075,
-      "learning_rate": 0.00019998297818218952,
-      "loss": 46.0,
-      "step": 36481
-    },
-    {
-      "epoch": 5.8751560046700755,
-      "grad_norm": 0.0013049393892288208,
-      "learning_rate": 0.00019998297724875923,
-      "loss": 46.0,
-      "step": 36482
-    },
-    {
-      "epoch": 5.875317041748863,
-      "grad_norm": 0.0024424907751381397,
-      "learning_rate": 0.00019998297631530334,
-      "loss": 46.0,
-      "step": 36483
-    },
-    {
-      "epoch": 5.87547807882765,
-      "grad_norm": 0.0020358997862786055,
-      "learning_rate": 0.00019998297538182186,
-      "loss": 46.0,
-      "step": 36484
-    },
-    {
-      "epoch": 5.875639115906438,
-      "grad_norm": 0.0046354178339242935,
-      "learning_rate": 0.0001999829744483148,
-      "loss": 46.0,
-      "step": 36485
-    },
-    {
-      "epoch": 5.875800152985224,
-      "grad_norm": 0.012167622335255146,
-      "learning_rate": 0.00019998297351478213,
-      "loss": 46.0,
-      "step": 36486
-    },
-    {
-      "epoch": 5.875961190064012,
-      "grad_norm": 0.0034470614045858383,
-      "learning_rate": 0.00019998297258122386,
-      "loss": 46.0,
-      "step": 36487
-    },
-    {
-      "epoch": 5.876122227142799,
-      "grad_norm": 0.001387792406603694,
-      "learning_rate": 0.00019998297164764003,
-      "loss": 46.0,
-      "step": 36488
-    },
-    {
-      "epoch": 5.876283264221587,
-      "grad_norm": 0.0025637776125222445,
-      "learning_rate": 0.0001999829707140306,
-      "loss": 46.0,
-      "step": 36489
-    },
-    {
-      "epoch": 5.876444301300374,
-      "grad_norm": 0.00794809591025114,
-      "learning_rate": 0.00019998296978039558,
-      "loss": 46.0,
-      "step": 36490
-    },
-    {
-      "epoch": 5.876605338379162,
-      "grad_norm": 0.00251396419480443,
-      "learning_rate": 0.00019998296884673496,
-      "loss": 46.0,
-      "step": 36491
-    },
-    {
-      "epoch": 5.876766375457949,
-      "grad_norm": 0.0030576088465750217,
-      "learning_rate": 0.00019998296791304875,
-      "loss": 46.0,
-      "step": 36492
-    },
-    {
-      "epoch": 5.876927412536737,
-      "grad_norm": 0.004708112683147192,
-      "learning_rate": 0.00019998296697933696,
-      "loss": 46.0,
-      "step": 36493
-    },
-    {
-      "epoch": 5.877088449615524,
-      "grad_norm": 0.002636307617649436,
-      "learning_rate": 0.00019998296604559958,
-      "loss": 46.0,
-      "step": 36494
-    },
-    {
-      "epoch": 5.8772494866943115,
-      "grad_norm": 0.011757121421396732,
-      "learning_rate": 0.0001999829651118366,
-      "loss": 46.0,
-      "step": 36495
-    },
-    {
-      "epoch": 5.877410523773099,
-      "grad_norm": 0.002685957122594118,
-      "learning_rate": 0.00019998296417804803,
-      "loss": 46.0,
-      "step": 36496
-    },
-    {
-      "epoch": 5.877571560851886,
-      "grad_norm": 0.0022678556852042675,
-      "learning_rate": 0.00019998296324423386,
-      "loss": 46.0,
-      "step": 36497
-    },
-    {
-      "epoch": 5.877732597930674,
-      "grad_norm": 0.0013451090781018138,
-      "learning_rate": 0.00019998296231039413,
-      "loss": 46.0,
-      "step": 36498
-    },
-    {
-      "epoch": 5.877893635009461,
-      "grad_norm": 0.006251571234315634,
-      "learning_rate": 0.00019998296137652878,
-      "loss": 46.0,
-      "step": 36499
-    },
-    {
-      "epoch": 5.878054672088249,
-      "grad_norm": 0.008110497146844864,
-      "learning_rate": 0.00019998296044263785,
-      "loss": 46.0,
-      "step": 36500
-    },
-    {
-      "epoch": 5.878215709167035,
-      "grad_norm": 0.0063213468529284,
-      "learning_rate": 0.00019998295950872133,
-      "loss": 46.0,
-      "step": 36501
-    },
-    {
-      "epoch": 5.878376746245823,
-      "grad_norm": 0.0010849563404917717,
-      "learning_rate": 0.00019998295857477923,
-      "loss": 46.0,
-      "step": 36502
-    },
-    {
-      "epoch": 5.87853778332461,
-      "grad_norm": 0.002168738516047597,
-      "learning_rate": 0.0001999829576408115,
-      "loss": 46.0,
-      "step": 36503
-    },
-    {
-      "epoch": 5.878698820403398,
-      "grad_norm": 0.008430509828031063,
-      "learning_rate": 0.00019998295670681823,
-      "loss": 46.0,
-      "step": 36504
-    },
-    {
-      "epoch": 5.878859857482185,
-      "grad_norm": 0.0007118963985703886,
-      "learning_rate": 0.00019998295577279936,
-      "loss": 46.0,
-      "step": 36505
-    },
-    {
-      "epoch": 5.8790208945609725,
-      "grad_norm": 0.003994202706962824,
-      "learning_rate": 0.00019998295483875488,
-      "loss": 46.0,
-      "step": 36506
-    },
-    {
-      "epoch": 5.87918193163976,
-      "grad_norm": 0.0018941318849101663,
-      "learning_rate": 0.0001999829539046848,
-      "loss": 46.0,
-      "step": 36507
-    },
-    {
-      "epoch": 5.8793429687185474,
-      "grad_norm": 0.0021580078173428774,
-      "learning_rate": 0.00019998295297058918,
-      "loss": 46.0,
-      "step": 36508
-    },
-    {
-      "epoch": 5.879504005797335,
-      "grad_norm": 0.0025449655950069427,
-      "learning_rate": 0.00019998295203646794,
-      "loss": 46.0,
-      "step": 36509
-    },
-    {
-      "epoch": 5.879665042876122,
-      "grad_norm": 0.0011550434865057468,
-      "learning_rate": 0.0001999829511023211,
-      "loss": 46.0,
-      "step": 36510
-    },
-    {
-      "epoch": 5.87982607995491,
-      "grad_norm": 0.0047856662422418594,
-      "learning_rate": 0.00019998295016814867,
-      "loss": 46.0,
-      "step": 36511
-    },
-    {
-      "epoch": 5.879987117033697,
-      "grad_norm": 0.002842686604708433,
-      "learning_rate": 0.00019998294923395066,
-      "loss": 46.0,
-      "step": 36512
-    },
-    {
-      "epoch": 5.880148154112485,
-      "grad_norm": 0.005742587614804506,
-      "learning_rate": 0.00019998294829972707,
-      "loss": 46.0,
-      "step": 36513
-    },
-    {
-      "epoch": 5.880309191191272,
-      "grad_norm": 0.004429700318723917,
-      "learning_rate": 0.00019998294736547787,
-      "loss": 46.0,
-      "step": 36514
-    },
-    {
-      "epoch": 5.88047022827006,
-      "grad_norm": 0.004376598633825779,
-      "learning_rate": 0.00019998294643120307,
-      "loss": 46.0,
-      "step": 36515
-    },
-    {
-      "epoch": 5.880631265348846,
-      "grad_norm": 0.002024495741352439,
-      "learning_rate": 0.00019998294549690272,
-      "loss": 46.0,
-      "step": 36516
-    },
-    {
-      "epoch": 5.880792302427634,
-      "grad_norm": 0.002057221019640565,
-      "learning_rate": 0.00019998294456257672,
-      "loss": 46.0,
-      "step": 36517
-    },
-    {
-      "epoch": 5.880953339506421,
-      "grad_norm": 0.006331522483378649,
-      "learning_rate": 0.00019998294362822517,
-      "loss": 46.0,
-      "step": 36518
-    },
-    {
-      "epoch": 5.8811143765852085,
-      "grad_norm": 0.0012722924584522843,
-      "learning_rate": 0.00019998294269384803,
-      "loss": 46.0,
-      "step": 36519
-    },
-    {
-      "epoch": 5.881275413663996,
-      "grad_norm": 0.0025560788344591856,
-      "learning_rate": 0.0001999829417594453,
-      "loss": 46.0,
-      "step": 36520
-    },
-    {
-      "epoch": 5.881436450742783,
-      "grad_norm": 0.0037808669731020927,
-      "learning_rate": 0.00019998294082501696,
-      "loss": 46.0,
-      "step": 36521
-    },
-    {
-      "epoch": 5.881597487821571,
-      "grad_norm": 0.006396201904863119,
-      "learning_rate": 0.00019998293989056305,
-      "loss": 46.0,
-      "step": 36522
-    },
-    {
-      "epoch": 5.881758524900358,
-      "grad_norm": 0.003589616622775793,
-      "learning_rate": 0.00019998293895608354,
-      "loss": 46.0,
-      "step": 36523
-    },
-    {
-      "epoch": 5.881919561979146,
-      "grad_norm": 0.0021946635097265244,
-      "learning_rate": 0.00019998293802157843,
-      "loss": 46.0,
-      "step": 36524
-    },
-    {
-      "epoch": 5.882080599057933,
-      "grad_norm": 0.006754905916750431,
-      "learning_rate": 0.00019998293708704774,
-      "loss": 46.0,
-      "step": 36525
-    },
-    {
-      "epoch": 5.882241636136721,
-      "grad_norm": 0.005507329013198614,
-      "learning_rate": 0.00019998293615249146,
-      "loss": 46.0,
-      "step": 36526
-    },
-    {
-      "epoch": 5.882402673215508,
-      "grad_norm": 0.001875278539955616,
-      "learning_rate": 0.0001999829352179096,
-      "loss": 46.0,
-      "step": 36527
-    },
-    {
-      "epoch": 5.882563710294296,
-      "grad_norm": 0.011401423253118992,
-      "learning_rate": 0.00019998293428330211,
-      "loss": 46.0,
-      "step": 36528
-    },
-    {
-      "epoch": 5.882724747373083,
-      "grad_norm": 0.005787578411400318,
-      "learning_rate": 0.0001999829333486691,
-      "loss": 46.0,
-      "step": 36529
-    },
-    {
-      "epoch": 5.8828857844518705,
-      "grad_norm": 0.003180295694619417,
-      "learning_rate": 0.00019998293241401042,
-      "loss": 46.0,
-      "step": 36530
-    },
-    {
-      "epoch": 5.883046821530657,
-      "grad_norm": 0.002754882210865617,
-      "learning_rate": 0.00019998293147932618,
-      "loss": 46.0,
-      "step": 36531
-    },
-    {
-      "epoch": 5.8832078586094445,
-      "grad_norm": 0.007766124326735735,
-      "learning_rate": 0.00019998293054461638,
-      "loss": 46.0,
-      "step": 36532
-    },
-    {
-      "epoch": 5.883368895688232,
-      "grad_norm": 0.0023852549493312836,
-      "learning_rate": 0.00019998292960988096,
-      "loss": 46.0,
-      "step": 36533
-    },
-    {
-      "epoch": 5.883529932767019,
-      "grad_norm": 0.002860397333279252,
-      "learning_rate": 0.00019998292867511993,
-      "loss": 46.0,
-      "step": 36534
-    },
-    {
-      "epoch": 5.883690969845807,
-      "grad_norm": 0.003329914528876543,
-      "learning_rate": 0.00019998292774033337,
-      "loss": 46.0,
-      "step": 36535
-    },
-    {
-      "epoch": 5.883852006924594,
-      "grad_norm": 0.012467294931411743,
-      "learning_rate": 0.00019998292680552116,
-      "loss": 46.0,
-      "step": 36536
-    },
-    {
-      "epoch": 5.884013044003382,
-      "grad_norm": 0.0048507400788366795,
-      "learning_rate": 0.0001999829258706834,
-      "loss": 46.0,
-      "step": 36537
-    },
-    {
-      "epoch": 5.884174081082169,
-      "grad_norm": 0.0010412726551294327,
-      "learning_rate": 0.00019998292493582002,
-      "loss": 46.0,
-      "step": 36538
-    },
-    {
-      "epoch": 5.884335118160957,
-      "grad_norm": 0.009102469310164452,
-      "learning_rate": 0.00019998292400093108,
-      "loss": 46.0,
-      "step": 36539
-    },
-    {
-      "epoch": 5.884496155239744,
-      "grad_norm": 0.009847925044596195,
-      "learning_rate": 0.00019998292306601652,
-      "loss": 46.0,
-      "step": 36540
-    },
-    {
-      "epoch": 5.884657192318532,
-      "grad_norm": 0.0020139359403401613,
-      "learning_rate": 0.00019998292213107636,
-      "loss": 46.0,
-      "step": 36541
-    },
-    {
-      "epoch": 5.884818229397319,
-      "grad_norm": 0.006518482230603695,
-      "learning_rate": 0.00019998292119611065,
-      "loss": 46.0,
-      "step": 36542
-    },
-    {
-      "epoch": 5.8849792664761065,
-      "grad_norm": 0.0015360381221398711,
-      "learning_rate": 0.00019998292026111934,
-      "loss": 46.0,
-      "step": 36543
-    },
-    {
-      "epoch": 5.885140303554893,
-      "grad_norm": 0.0035662108566612005,
-      "learning_rate": 0.00019998291932610244,
-      "loss": 46.0,
-      "step": 36544
-    },
-    {
-      "epoch": 5.885301340633681,
-      "grad_norm": 0.00753018818795681,
-      "learning_rate": 0.00019998291839105992,
-      "loss": 46.0,
-      "step": 36545
-    },
-    {
-      "epoch": 5.885462377712468,
-      "grad_norm": 0.003252539085224271,
-      "learning_rate": 0.00019998291745599182,
-      "loss": 46.0,
-      "step": 36546
-    },
-    {
-      "epoch": 5.885623414791255,
-      "grad_norm": 0.0040799705311656,
-      "learning_rate": 0.00019998291652089813,
-      "loss": 46.0,
-      "step": 36547
-    },
-    {
-      "epoch": 5.885784451870043,
-      "grad_norm": 0.0009041695157065988,
-      "learning_rate": 0.00019998291558577888,
-      "loss": 46.0,
-      "step": 36548
-    },
-    {
-      "epoch": 5.88594548894883,
-      "grad_norm": 0.01083538867533207,
-      "learning_rate": 0.000199982914650634,
-      "loss": 46.0,
-      "step": 36549
-    },
-    {
-      "epoch": 5.886106526027618,
-      "grad_norm": 0.012440397404134274,
-      "learning_rate": 0.00019998291371546356,
-      "loss": 46.0,
-      "step": 36550
-    },
-    {
-      "epoch": 5.886267563106405,
-      "grad_norm": 0.022902099415659904,
-      "learning_rate": 0.00019998291278026752,
-      "loss": 46.0,
-      "step": 36551
-    },
-    {
-      "epoch": 5.886428600185193,
-      "grad_norm": 0.003278498537838459,
-      "learning_rate": 0.0001999829118450459,
-      "loss": 46.0,
-      "step": 36552
-    },
-    {
-      "epoch": 5.88658963726398,
-      "grad_norm": 0.0028832892421633005,
-      "learning_rate": 0.00019998291090979865,
-      "loss": 46.0,
-      "step": 36553
-    },
-    {
-      "epoch": 5.8867506743427676,
-      "grad_norm": 0.007608143147081137,
-      "learning_rate": 0.00019998290997452585,
-      "loss": 46.0,
-      "step": 36554
-    },
-    {
-      "epoch": 5.886911711421555,
-      "grad_norm": 0.01458243653178215,
-      "learning_rate": 0.00019998290903922743,
-      "loss": 46.0,
-      "step": 36555
-    },
-    {
-      "epoch": 5.8870727485003425,
-      "grad_norm": 0.005875394679605961,
-      "learning_rate": 0.00019998290810390343,
-      "loss": 46.0,
-      "step": 36556
-    },
-    {
-      "epoch": 5.88723378557913,
-      "grad_norm": 0.007683324161916971,
-      "learning_rate": 0.00019998290716855387,
-      "loss": 46.0,
-      "step": 36557
-    },
-    {
-      "epoch": 5.887394822657917,
-      "grad_norm": 0.003210650524124503,
-      "learning_rate": 0.00019998290623317867,
-      "loss": 46.0,
-      "step": 36558
-    },
-    {
-      "epoch": 5.887555859736704,
-      "grad_norm": 0.008093958720564842,
-      "learning_rate": 0.0001999829052977779,
-      "loss": 46.0,
-      "step": 36559
-    },
-    {
-      "epoch": 5.887716896815492,
-      "grad_norm": 0.0030021965503692627,
-      "learning_rate": 0.00019998290436235155,
-      "loss": 46.0,
-      "step": 36560
-    },
-    {
-      "epoch": 5.887877933894279,
-      "grad_norm": 0.015265724621713161,
-      "learning_rate": 0.00019998290342689958,
-      "loss": 46.0,
-      "step": 36561
-    },
-    {
-      "epoch": 5.888038970973066,
-      "grad_norm": 0.003575454233214259,
-      "learning_rate": 0.00019998290249142206,
-      "loss": 46.0,
-      "step": 36562
-    },
-    {
-      "epoch": 5.888200008051854,
-      "grad_norm": 0.005558281671255827,
-      "learning_rate": 0.00019998290155591892,
-      "loss": 46.0,
-      "step": 36563
-    },
-    {
-      "epoch": 5.888361045130641,
-      "grad_norm": 0.0057060206308960915,
-      "learning_rate": 0.0001999829006203902,
-      "loss": 46.0,
-      "step": 36564
-    },
-    {
-      "epoch": 5.888522082209429,
-      "grad_norm": 0.018302863463759422,
-      "learning_rate": 0.0001999828996848359,
-      "loss": 46.0,
-      "step": 36565
-    },
-    {
-      "epoch": 5.888683119288216,
-      "grad_norm": 0.005606064572930336,
-      "learning_rate": 0.000199982898749256,
-      "loss": 46.0,
-      "step": 36566
-    },
-    {
-      "epoch": 5.8888441563670035,
-      "grad_norm": 0.0022443048655986786,
-      "learning_rate": 0.00019998289781365052,
-      "loss": 46.0,
-      "step": 36567
-    },
-    {
-      "epoch": 5.889005193445791,
-      "grad_norm": 0.009120387956500053,
-      "learning_rate": 0.00019998289687801944,
-      "loss": 46.0,
-      "step": 36568
-    },
-    {
-      "epoch": 5.889166230524578,
-      "grad_norm": 0.0007909274427220225,
-      "learning_rate": 0.00019998289594236275,
-      "loss": 46.0,
-      "step": 36569
-    },
-    {
-      "epoch": 5.889327267603366,
-      "grad_norm": 0.00211282423697412,
-      "learning_rate": 0.00019998289500668047,
-      "loss": 46.0,
-      "step": 36570
-    },
-    {
-      "epoch": 5.889488304682153,
-      "grad_norm": 0.0032763362396508455,
-      "learning_rate": 0.00019998289407097264,
-      "loss": 46.0,
-      "step": 36571
-    },
-    {
-      "epoch": 5.889649341760941,
-      "grad_norm": 0.006326448637992144,
-      "learning_rate": 0.00019998289313523918,
-      "loss": 46.0,
-      "step": 36572
-    },
-    {
-      "epoch": 5.889810378839728,
-      "grad_norm": 0.004175235517323017,
-      "learning_rate": 0.00019998289219948015,
-      "loss": 46.0,
-      "step": 36573
-    },
-    {
-      "epoch": 5.889971415918515,
-      "grad_norm": 0.0026226313784718513,
-      "learning_rate": 0.00019998289126369555,
-      "loss": 46.0,
-      "step": 36574
-    },
-    {
-      "epoch": 5.890132452997302,
-      "grad_norm": 0.007245667744427919,
-      "learning_rate": 0.0001999828903278853,
-      "loss": 46.0,
-      "step": 36575
-    },
-    {
-      "epoch": 5.89029349007609,
-      "grad_norm": 0.009717113338410854,
-      "learning_rate": 0.0001999828893920495,
-      "loss": 46.0,
-      "step": 36576
-    },
-    {
-      "epoch": 5.890454527154877,
-      "grad_norm": 0.010242221876978874,
-      "learning_rate": 0.00019998288845618812,
-      "loss": 46.0,
-      "step": 36577
-    },
-    {
-      "epoch": 5.890615564233665,
-      "grad_norm": 0.026576397940516472,
-      "learning_rate": 0.00019998288752030112,
-      "loss": 46.0,
-      "step": 36578
-    },
-    {
-      "epoch": 5.890776601312452,
-      "grad_norm": 0.001437407685443759,
-      "learning_rate": 0.00019998288658438856,
-      "loss": 46.0,
-      "step": 36579
-    },
-    {
-      "epoch": 5.8909376383912395,
-      "grad_norm": 0.009794313460588455,
-      "learning_rate": 0.00019998288564845038,
-      "loss": 46.0,
-      "step": 36580
-    },
-    {
-      "epoch": 5.891098675470027,
-      "grad_norm": 0.005004182457923889,
-      "learning_rate": 0.00019998288471248662,
-      "loss": 46.0,
-      "step": 36581
-    },
-    {
-      "epoch": 5.891259712548814,
-      "grad_norm": 0.01586543396115303,
-      "learning_rate": 0.00019998288377649727,
-      "loss": 46.0,
-      "step": 36582
-    },
-    {
-      "epoch": 5.891420749627602,
-      "grad_norm": 0.012091053649783134,
-      "learning_rate": 0.00019998288284048236,
-      "loss": 46.0,
-      "step": 36583
-    },
-    {
-      "epoch": 5.891581786706389,
-      "grad_norm": 0.006689284462481737,
-      "learning_rate": 0.0001999828819044418,
-      "loss": 46.0,
-      "step": 36584
-    },
-    {
-      "epoch": 5.891742823785177,
-      "grad_norm": 0.001232705544680357,
-      "learning_rate": 0.0001999828809683757,
-      "loss": 46.0,
-      "step": 36585
-    },
-    {
-      "epoch": 5.891903860863964,
-      "grad_norm": 0.003674765583127737,
-      "learning_rate": 0.00019998288003228397,
-      "loss": 46.0,
-      "step": 36586
-    },
-    {
-      "epoch": 5.892064897942752,
-      "grad_norm": 0.0030214886646717787,
-      "learning_rate": 0.00019998287909616665,
-      "loss": 46.0,
-      "step": 36587
-    },
-    {
-      "epoch": 5.892225935021539,
-      "grad_norm": 0.003527117194607854,
-      "learning_rate": 0.00019998287816002378,
-      "loss": 46.0,
-      "step": 36588
-    },
-    {
-      "epoch": 5.892386972100326,
-      "grad_norm": 0.002677063923329115,
-      "learning_rate": 0.0001999828772238553,
-      "loss": 46.0,
-      "step": 36589
-    },
-    {
-      "epoch": 5.892548009179113,
-      "grad_norm": 0.003265031846240163,
-      "learning_rate": 0.00019998287628766122,
-      "loss": 46.0,
-      "step": 36590
-    },
-    {
-      "epoch": 5.892709046257901,
-      "grad_norm": 0.00640291441231966,
-      "learning_rate": 0.00019998287535144156,
-      "loss": 46.0,
-      "step": 36591
-    },
-    {
-      "epoch": 5.892870083336688,
-      "grad_norm": 0.004319123458117247,
-      "learning_rate": 0.0001999828744151963,
-      "loss": 46.0,
-      "step": 36592
-    },
-    {
-      "epoch": 5.8930311204154755,
-      "grad_norm": 0.01202433928847313,
-      "learning_rate": 0.00019998287347892547,
-      "loss": 46.0,
-      "step": 36593
-    },
-    {
-      "epoch": 5.893192157494263,
-      "grad_norm": 0.0032576615922152996,
-      "learning_rate": 0.00019998287254262902,
-      "loss": 46.0,
-      "step": 36594
-    },
-    {
-      "epoch": 5.89335319457305,
-      "grad_norm": 0.011612916365265846,
-      "learning_rate": 0.000199982871606307,
-      "loss": 46.0,
-      "step": 36595
-    },
-    {
-      "epoch": 5.893514231651838,
-      "grad_norm": 0.00565600348636508,
-      "learning_rate": 0.0001999828706699594,
-      "loss": 46.0,
-      "step": 36596
-    },
-    {
-      "epoch": 5.893675268730625,
-      "grad_norm": 0.0027215436566621065,
-      "learning_rate": 0.00019998286973358617,
-      "loss": 46.0,
-      "step": 36597
-    },
-    {
-      "epoch": 5.893836305809413,
-      "grad_norm": 0.0009165478986687958,
-      "learning_rate": 0.00019998286879718737,
-      "loss": 46.0,
-      "step": 36598
-    },
-    {
-      "epoch": 5.8939973428882,
-      "grad_norm": 0.002467401558533311,
-      "learning_rate": 0.000199982867860763,
-      "loss": 46.0,
-      "step": 36599
-    },
-    {
-      "epoch": 5.894158379966988,
-      "grad_norm": 0.004900219384580851,
-      "learning_rate": 0.00019998286692431304,
-      "loss": 46.0,
-      "step": 36600
-    },
-    {
-      "epoch": 5.894319417045775,
-      "grad_norm": 0.01821213588118553,
-      "learning_rate": 0.00019998286598783745,
-      "loss": 46.0,
-      "step": 36601
-    },
-    {
-      "epoch": 5.894480454124563,
-      "grad_norm": 0.0020022073294967413,
-      "learning_rate": 0.00019998286505133628,
-      "loss": 46.0,
-      "step": 36602
-    },
-    {
-      "epoch": 5.89464149120335,
-      "grad_norm": 0.017472347244620323,
-      "learning_rate": 0.00019998286411480954,
-      "loss": 46.0,
-      "step": 36603
-    },
-    {
-      "epoch": 5.894802528282137,
-      "grad_norm": 0.002706969855353236,
-      "learning_rate": 0.0001999828631782572,
-      "loss": 46.0,
-      "step": 36604
-    },
-    {
-      "epoch": 5.894963565360924,
-      "grad_norm": 0.0010954283643513918,
-      "learning_rate": 0.00019998286224167928,
-      "loss": 46.0,
-      "step": 36605
-    },
-    {
-      "epoch": 5.8951246024397115,
-      "grad_norm": 0.008376320824027061,
-      "learning_rate": 0.00019998286130507573,
-      "loss": 46.0,
-      "step": 36606
-    },
-    {
-      "epoch": 5.895285639518499,
-      "grad_norm": 0.002078205579891801,
-      "learning_rate": 0.00019998286036844665,
-      "loss": 46.0,
-      "step": 36607
-    },
-    {
-      "epoch": 5.895446676597286,
-      "grad_norm": 0.00561330933123827,
-      "learning_rate": 0.00019998285943179192,
-      "loss": 46.0,
-      "step": 36608
-    },
-    {
-      "epoch": 5.895607713676074,
-      "grad_norm": 0.0016333928797394037,
-      "learning_rate": 0.00019998285849511164,
-      "loss": 46.0,
-      "step": 36609
-    },
-    {
-      "epoch": 5.895768750754861,
-      "grad_norm": 0.0012208586558699608,
-      "learning_rate": 0.00019998285755840577,
-      "loss": 46.0,
-      "step": 36610
-    },
-    {
-      "epoch": 5.895929787833649,
-      "grad_norm": 0.004889764357358217,
-      "learning_rate": 0.00019998285662167428,
-      "loss": 46.0,
-      "step": 36611
-    },
-    {
-      "epoch": 5.896090824912436,
-      "grad_norm": 0.015877557918429375,
-      "learning_rate": 0.0001999828556849172,
-      "loss": 46.0,
-      "step": 36612
-    },
-    {
-      "epoch": 5.896251861991224,
-      "grad_norm": 0.009460185654461384,
-      "learning_rate": 0.00019998285474813457,
-      "loss": 46.0,
-      "step": 36613
-    },
-    {
-      "epoch": 5.896412899070011,
-      "grad_norm": 0.0036141336895525455,
-      "learning_rate": 0.00019998285381132632,
-      "loss": 46.0,
-      "step": 36614
-    },
-    {
-      "epoch": 5.8965739361487985,
-      "grad_norm": 0.007342739962041378,
-      "learning_rate": 0.0001999828528744925,
-      "loss": 46.0,
-      "step": 36615
-    },
-    {
-      "epoch": 5.896734973227586,
-      "grad_norm": 0.006484126206487417,
-      "learning_rate": 0.00019998285193763307,
-      "loss": 46.0,
-      "step": 36616
-    },
-    {
-      "epoch": 5.896896010306373,
-      "grad_norm": 0.006829795893281698,
-      "learning_rate": 0.00019998285100074806,
-      "loss": 46.0,
-      "step": 36617
-    },
-    {
-      "epoch": 5.897057047385161,
-      "grad_norm": 0.001765860477462411,
-      "learning_rate": 0.00019998285006383743,
-      "loss": 46.0,
-      "step": 36618
-    },
-    {
-      "epoch": 5.8972180844639475,
-      "grad_norm": 0.000543889997061342,
-      "learning_rate": 0.00019998284912690125,
-      "loss": 46.0,
-      "step": 36619
-    },
-    {
-      "epoch": 5.897379121542735,
-      "grad_norm": 0.0029645240865647793,
-      "learning_rate": 0.00019998284818993945,
-      "loss": 46.0,
-      "step": 36620
-    },
-    {
-      "epoch": 5.897540158621522,
-      "grad_norm": 0.002774882595986128,
-      "learning_rate": 0.0001999828472529521,
-      "loss": 46.0,
-      "step": 36621
-    },
-    {
-      "epoch": 5.89770119570031,
-      "grad_norm": 0.0020182610023766756,
-      "learning_rate": 0.00019998284631593912,
-      "loss": 46.0,
-      "step": 36622
-    },
-    {
-      "epoch": 5.897862232779097,
-      "grad_norm": 0.00874936580657959,
-      "learning_rate": 0.00019998284537890056,
-      "loss": 46.0,
-      "step": 36623
-    },
-    {
-      "epoch": 5.898023269857885,
-      "grad_norm": 0.021953793242573738,
-      "learning_rate": 0.00019998284444183639,
-      "loss": 46.0,
-      "step": 36624
-    },
-    {
-      "epoch": 5.898184306936672,
-      "grad_norm": 0.006183766759932041,
-      "learning_rate": 0.00019998284350474665,
-      "loss": 46.0,
-      "step": 36625
-    },
-    {
-      "epoch": 5.89834534401546,
-      "grad_norm": 0.007318221498280764,
-      "learning_rate": 0.00019998284256763133,
-      "loss": 46.0,
-      "step": 36626
-    },
-    {
-      "epoch": 5.898506381094247,
-      "grad_norm": 0.007123200222849846,
-      "learning_rate": 0.00019998284163049042,
-      "loss": 46.0,
-      "step": 36627
-    },
-    {
-      "epoch": 5.8986674181730345,
-      "grad_norm": 0.003372452687472105,
-      "learning_rate": 0.00019998284069332393,
-      "loss": 46.0,
-      "step": 36628
-    },
-    {
-      "epoch": 5.898828455251822,
-      "grad_norm": 0.005449937656521797,
-      "learning_rate": 0.00019998283975613182,
-      "loss": 46.0,
-      "step": 36629
-    },
-    {
-      "epoch": 5.898989492330609,
-      "grad_norm": 0.003368463134393096,
-      "learning_rate": 0.00019998283881891412,
-      "loss": 46.0,
-      "step": 36630
-    },
-    {
-      "epoch": 5.899150529409397,
-      "grad_norm": 0.0032204920426011086,
-      "learning_rate": 0.00019998283788167084,
-      "loss": 46.0,
-      "step": 36631
-    },
-    {
-      "epoch": 5.8993115664881834,
-      "grad_norm": 0.0011918552918359637,
-      "learning_rate": 0.00019998283694440194,
-      "loss": 46.0,
-      "step": 36632
-    },
-    {
-      "epoch": 5.899472603566972,
-      "grad_norm": 0.003111568745225668,
-      "learning_rate": 0.00019998283600710748,
-      "loss": 46.0,
-      "step": 36633
-    },
-    {
-      "epoch": 5.899633640645758,
-      "grad_norm": 0.006619945168495178,
-      "learning_rate": 0.00019998283506978743,
-      "loss": 46.0,
-      "step": 36634
-    },
-    {
-      "epoch": 5.899794677724546,
-      "grad_norm": 0.005429692566394806,
-      "learning_rate": 0.0001999828341324418,
-      "loss": 46.0,
-      "step": 36635
-    },
-    {
-      "epoch": 5.899955714803333,
-      "grad_norm": 0.0039047219324856997,
-      "learning_rate": 0.00019998283319507055,
-      "loss": 46.0,
-      "step": 36636
-    },
-    {
-      "epoch": 5.900116751882121,
-      "grad_norm": 0.004827241878956556,
-      "learning_rate": 0.00019998283225767374,
-      "loss": 46.0,
-      "step": 36637
-    },
-    {
-      "epoch": 5.900277788960908,
-      "grad_norm": 0.008907639421522617,
-      "learning_rate": 0.0001999828313202513,
-      "loss": 46.0,
-      "step": 36638
-    },
-    {
-      "epoch": 5.900438826039696,
-      "grad_norm": 0.001424536225385964,
-      "learning_rate": 0.0001999828303828033,
-      "loss": 46.0,
-      "step": 36639
-    },
-    {
-      "epoch": 5.900599863118483,
-      "grad_norm": 0.003321142867207527,
-      "learning_rate": 0.00019998282944532972,
-      "loss": 46.0,
-      "step": 36640
-    },
-    {
-      "epoch": 5.9007609001972705,
-      "grad_norm": 0.002521553775295615,
-      "learning_rate": 0.00019998282850783054,
-      "loss": 46.0,
-      "step": 36641
-    },
-    {
-      "epoch": 5.900921937276058,
-      "grad_norm": 0.0020214428659528494,
-      "learning_rate": 0.00019998282757030574,
-      "loss": 46.0,
-      "step": 36642
-    },
-    {
-      "epoch": 5.901082974354845,
-      "grad_norm": 0.0028818012215197086,
-      "learning_rate": 0.00019998282663275538,
-      "loss": 46.0,
-      "step": 36643
-    },
-    {
-      "epoch": 5.901244011433633,
-      "grad_norm": 0.025667818263173103,
-      "learning_rate": 0.0001999828256951794,
-      "loss": 46.0,
-      "step": 36644
-    },
-    {
-      "epoch": 5.90140504851242,
-      "grad_norm": 0.002696142066270113,
-      "learning_rate": 0.00019998282475757785,
-      "loss": 46.0,
-      "step": 36645
-    },
-    {
-      "epoch": 5.901566085591208,
-      "grad_norm": 0.006760791875422001,
-      "learning_rate": 0.00019998282381995073,
-      "loss": 46.0,
-      "step": 36646
-    },
-    {
-      "epoch": 5.901727122669994,
-      "grad_norm": 0.0014532756758853793,
-      "learning_rate": 0.000199982822882298,
-      "loss": 46.0,
-      "step": 36647
-    },
-    {
-      "epoch": 5.901888159748782,
-      "grad_norm": 0.02876780927181244,
-      "learning_rate": 0.00019998282194461968,
-      "loss": 46.0,
-      "step": 36648
-    },
-    {
-      "epoch": 5.902049196827569,
-      "grad_norm": 0.00649761687964201,
-      "learning_rate": 0.00019998282100691577,
-      "loss": 46.0,
-      "step": 36649
-    },
-    {
-      "epoch": 5.902210233906357,
-      "grad_norm": 0.002604781650006771,
-      "learning_rate": 0.00019998282006918627,
-      "loss": 46.0,
-      "step": 36650
-    },
-    {
-      "epoch": 5.902371270985144,
-      "grad_norm": 0.0011523580178618431,
-      "learning_rate": 0.00019998281913143117,
-      "loss": 46.0,
-      "step": 36651
-    },
-    {
-      "epoch": 5.902532308063932,
-      "grad_norm": 0.007648960687220097,
-      "learning_rate": 0.0001999828181936505,
-      "loss": 46.0,
-      "step": 36652
-    },
-    {
-      "epoch": 5.902693345142719,
-      "grad_norm": 0.002044451190158725,
-      "learning_rate": 0.0001999828172558442,
-      "loss": 46.0,
-      "step": 36653
-    },
-    {
-      "epoch": 5.9028543822215065,
-      "grad_norm": 0.0013684118166565895,
-      "learning_rate": 0.00019998281631801234,
-      "loss": 46.0,
-      "step": 36654
-    },
-    {
-      "epoch": 5.903015419300294,
-      "grad_norm": 0.005389058962464333,
-      "learning_rate": 0.00019998281538015489,
-      "loss": 46.0,
-      "step": 36655
-    },
-    {
-      "epoch": 5.903176456379081,
-      "grad_norm": 0.0020386220421642065,
-      "learning_rate": 0.00019998281444227184,
-      "loss": 46.0,
-      "step": 36656
-    },
-    {
-      "epoch": 5.903337493457869,
-      "grad_norm": 0.004870891105383635,
-      "learning_rate": 0.00019998281350436324,
-      "loss": 46.0,
-      "step": 36657
-    },
-    {
-      "epoch": 5.903498530536656,
-      "grad_norm": 0.011061694473028183,
-      "learning_rate": 0.000199982812566429,
-      "loss": 46.0,
-      "step": 36658
-    },
-    {
-      "epoch": 5.903659567615444,
-      "grad_norm": 0.006413934286683798,
-      "learning_rate": 0.00019998281162846916,
-      "loss": 46.0,
-      "step": 36659
-    },
-    {
-      "epoch": 5.903820604694231,
-      "grad_norm": 0.002223755931481719,
-      "learning_rate": 0.00019998281069048376,
-      "loss": 46.0,
-      "step": 36660
-    },
-    {
-      "epoch": 5.903981641773019,
-      "grad_norm": 0.001216917997226119,
-      "learning_rate": 0.00019998280975247275,
-      "loss": 46.0,
-      "step": 36661
-    },
-    {
-      "epoch": 5.904142678851805,
-      "grad_norm": 0.0033759952057152987,
-      "learning_rate": 0.00019998280881443619,
-      "loss": 46.0,
-      "step": 36662
-    },
-    {
-      "epoch": 5.904303715930593,
-      "grad_norm": 0.009596017189323902,
-      "learning_rate": 0.000199982807876374,
-      "loss": 46.0,
-      "step": 36663
-    },
-    {
-      "epoch": 5.90446475300938,
-      "grad_norm": 0.00442937295883894,
-      "learning_rate": 0.00019998280693828623,
-      "loss": 46.0,
-      "step": 36664
-    },
-    {
-      "epoch": 5.904625790088168,
-      "grad_norm": 0.0015244719106703997,
-      "learning_rate": 0.0001999828060001729,
-      "loss": 46.0,
-      "step": 36665
-    },
-    {
-      "epoch": 5.904786827166955,
-      "grad_norm": 0.008727251552045345,
-      "learning_rate": 0.00019998280506203393,
-      "loss": 46.0,
-      "step": 36666
-    },
-    {
-      "epoch": 5.9049478642457425,
-      "grad_norm": 0.024323971942067146,
-      "learning_rate": 0.00019998280412386938,
-      "loss": 46.0,
-      "step": 36667
-    },
-    {
-      "epoch": 5.90510890132453,
-      "grad_norm": 0.0024902704171836376,
-      "learning_rate": 0.00019998280318567926,
-      "loss": 46.0,
-      "step": 36668
-    },
-    {
-      "epoch": 5.905269938403317,
-      "grad_norm": 0.0018617215100675821,
-      "learning_rate": 0.00019998280224746355,
-      "loss": 46.0,
-      "step": 36669
-    },
-    {
-      "epoch": 5.905430975482105,
-      "grad_norm": 0.002323055872693658,
-      "learning_rate": 0.00019998280130922223,
-      "loss": 46.0,
-      "step": 36670
-    },
-    {
-      "epoch": 5.905592012560892,
-      "grad_norm": 0.013949236832559109,
-      "learning_rate": 0.00019998280037095533,
-      "loss": 46.0,
-      "step": 36671
-    },
-    {
-      "epoch": 5.90575304963968,
-      "grad_norm": 0.0015396467642858624,
-      "learning_rate": 0.00019998279943266286,
-      "loss": 46.0,
-      "step": 36672
-    },
-    {
-      "epoch": 5.905914086718467,
-      "grad_norm": 0.0022673136554658413,
-      "learning_rate": 0.00019998279849434475,
-      "loss": 46.0,
-      "step": 36673
-    },
-    {
-      "epoch": 5.906075123797255,
-      "grad_norm": 0.005111353471875191,
-      "learning_rate": 0.00019998279755600108,
-      "loss": 46.0,
-      "step": 36674
-    },
-    {
-      "epoch": 5.906236160876042,
-      "grad_norm": 0.0019955916795879602,
-      "learning_rate": 0.00019998279661763183,
-      "loss": 46.0,
-      "step": 36675
-    },
-    {
-      "epoch": 5.9063971979548295,
-      "grad_norm": 0.004853942431509495,
-      "learning_rate": 0.00019998279567923696,
-      "loss": 46.0,
-      "step": 36676
-    },
-    {
-      "epoch": 5.906558235033616,
-      "grad_norm": 0.0062437900342047215,
-      "learning_rate": 0.0001999827947408165,
-      "loss": 46.0,
-      "step": 36677
-    },
-    {
-      "epoch": 5.9067192721124036,
-      "grad_norm": 0.0019858789164572954,
-      "learning_rate": 0.00019998279380237049,
-      "loss": 46.0,
-      "step": 36678
-    },
-    {
-      "epoch": 5.906880309191191,
-      "grad_norm": 0.0011909750755876303,
-      "learning_rate": 0.00019998279286389888,
-      "loss": 46.0,
-      "step": 36679
-    },
-    {
-      "epoch": 5.9070413462699785,
-      "grad_norm": 0.0011144229210913181,
-      "learning_rate": 0.00019998279192540164,
-      "loss": 46.0,
-      "step": 36680
-    },
-    {
-      "epoch": 5.907202383348766,
-      "grad_norm": 0.0036076551768928766,
-      "learning_rate": 0.00019998279098687883,
-      "loss": 46.0,
-      "step": 36681
-    },
-    {
-      "epoch": 5.907363420427553,
-      "grad_norm": 0.005558940581977367,
-      "learning_rate": 0.00019998279004833044,
-      "loss": 46.0,
-      "step": 36682
-    },
-    {
-      "epoch": 5.907524457506341,
-      "grad_norm": 0.002356515731662512,
-      "learning_rate": 0.00019998278910975646,
-      "loss": 46.0,
-      "step": 36683
-    },
-    {
-      "epoch": 5.907685494585128,
-      "grad_norm": 0.00231341365724802,
-      "learning_rate": 0.00019998278817115686,
-      "loss": 46.0,
-      "step": 36684
-    },
-    {
-      "epoch": 5.907846531663916,
-      "grad_norm": 0.001637893496081233,
-      "learning_rate": 0.0001999827872325317,
-      "loss": 46.0,
-      "step": 36685
-    },
-    {
-      "epoch": 5.908007568742703,
-      "grad_norm": 0.0010874888394027948,
-      "learning_rate": 0.00019998278629388097,
-      "loss": 46.0,
-      "step": 36686
-    },
-    {
-      "epoch": 5.908168605821491,
-      "grad_norm": 0.0027469471096992493,
-      "learning_rate": 0.00019998278535520459,
-      "loss": 46.0,
-      "step": 36687
-    },
-    {
-      "epoch": 5.908329642900278,
-      "grad_norm": 0.0027042501606047153,
-      "learning_rate": 0.00019998278441650267,
-      "loss": 46.0,
-      "step": 36688
-    },
-    {
-      "epoch": 5.9084906799790655,
-      "grad_norm": 0.006202050019055605,
-      "learning_rate": 0.0001999827834777751,
-      "loss": 46.0,
-      "step": 36689
-    },
-    {
-      "epoch": 5.908651717057852,
-      "grad_norm": 0.016852544620633125,
-      "learning_rate": 0.000199982782539022,
-      "loss": 46.0,
-      "step": 36690
-    },
-    {
-      "epoch": 5.90881275413664,
-      "grad_norm": 0.003232463961467147,
-      "learning_rate": 0.00019998278160024332,
-      "loss": 46.0,
-      "step": 36691
-    },
-    {
-      "epoch": 5.908973791215427,
-      "grad_norm": 0.0029984028078615665,
-      "learning_rate": 0.00019998278066143897,
-      "loss": 46.0,
-      "step": 36692
-    },
-    {
-      "epoch": 5.909134828294214,
-      "grad_norm": 0.002600082429125905,
-      "learning_rate": 0.0001999827797226091,
-      "loss": 46.0,
-      "step": 36693
-    },
-    {
-      "epoch": 5.909295865373002,
-      "grad_norm": 0.011613850481808186,
-      "learning_rate": 0.00019998277878375363,
-      "loss": 46.0,
-      "step": 36694
-    },
-    {
-      "epoch": 5.909456902451789,
-      "grad_norm": 0.009321661666035652,
-      "learning_rate": 0.00019998277784487255,
-      "loss": 46.0,
-      "step": 36695
-    },
-    {
-      "epoch": 5.909617939530577,
-      "grad_norm": 0.0020441017113626003,
-      "learning_rate": 0.0001999827769059659,
-      "loss": 46.0,
-      "step": 36696
-    },
-    {
-      "epoch": 5.909778976609364,
-      "grad_norm": 0.002522408729419112,
-      "learning_rate": 0.00019998277596703363,
-      "loss": 46.0,
-      "step": 36697
-    },
-    {
-      "epoch": 5.909940013688152,
-      "grad_norm": 0.019608579576015472,
-      "learning_rate": 0.00019998277502807579,
-      "loss": 46.0,
-      "step": 36698
-    },
-    {
-      "epoch": 5.910101050766939,
-      "grad_norm": 0.003051810199394822,
-      "learning_rate": 0.00019998277408909236,
-      "loss": 46.0,
-      "step": 36699
-    },
-    {
-      "epoch": 5.910262087845727,
-      "grad_norm": 0.002684236504137516,
-      "learning_rate": 0.0001999827731500833,
-      "loss": 46.0,
-      "step": 36700
-    },
-    {
-      "epoch": 5.910423124924514,
-      "grad_norm": 0.013379686512053013,
-      "learning_rate": 0.00019998277221104874,
-      "loss": 46.0,
-      "step": 36701
-    },
-    {
-      "epoch": 5.9105841620033015,
-      "grad_norm": 0.0016601013485342264,
-      "learning_rate": 0.00019998277127198852,
-      "loss": 46.0,
-      "step": 36702
-    },
-    {
-      "epoch": 5.910745199082089,
-      "grad_norm": 0.004295967984944582,
-      "learning_rate": 0.00019998277033290272,
-      "loss": 46.0,
-      "step": 36703
-    },
-    {
-      "epoch": 5.910906236160876,
-      "grad_norm": 0.002593794371932745,
-      "learning_rate": 0.00019998276939379132,
-      "loss": 46.0,
-      "step": 36704
-    },
-    {
-      "epoch": 5.911067273239663,
-      "grad_norm": 0.0030752660240978003,
-      "learning_rate": 0.00019998276845465434,
-      "loss": 46.0,
-      "step": 36705
-    },
-    {
-      "epoch": 5.911228310318451,
-      "grad_norm": 0.011294949799776077,
-      "learning_rate": 0.00019998276751549175,
-      "loss": 46.0,
-      "step": 36706
-    },
-    {
-      "epoch": 5.911389347397238,
-      "grad_norm": 0.011487973853945732,
-      "learning_rate": 0.00019998276657630363,
-      "loss": 46.0,
-      "step": 36707
-    },
-    {
-      "epoch": 5.911550384476025,
-      "grad_norm": 0.00791802629828453,
-      "learning_rate": 0.00019998276563708989,
-      "loss": 46.0,
-      "step": 36708
-    },
-    {
-      "epoch": 5.911711421554813,
-      "grad_norm": 0.0009943817276507616,
-      "learning_rate": 0.00019998276469785053,
-      "loss": 46.0,
-      "step": 36709
-    },
-    {
-      "epoch": 5.9118724586336,
-      "grad_norm": 0.005660542752593756,
-      "learning_rate": 0.0001999827637585856,
-      "loss": 46.0,
-      "step": 36710
-    },
-    {
-      "epoch": 5.912033495712388,
-      "grad_norm": 0.0018545347265899181,
-      "learning_rate": 0.0001999827628192951,
-      "loss": 46.0,
-      "step": 36711
-    },
-    {
-      "epoch": 5.912194532791175,
-      "grad_norm": 0.0020563986618071795,
-      "learning_rate": 0.00019998276187997897,
-      "loss": 46.0,
-      "step": 36712
-    },
-    {
-      "epoch": 5.912355569869963,
-      "grad_norm": 0.0018423276487737894,
-      "learning_rate": 0.0001999827609406373,
-      "loss": 46.0,
-      "step": 36713
-    },
-    {
-      "epoch": 5.91251660694875,
-      "grad_norm": 0.003212996758520603,
-      "learning_rate": 0.00019998276000126998,
-      "loss": 46.0,
-      "step": 36714
-    },
-    {
-      "epoch": 5.9126776440275375,
-      "grad_norm": 0.0018341336399316788,
-      "learning_rate": 0.0001999827590618771,
-      "loss": 46.0,
-      "step": 36715
-    },
-    {
-      "epoch": 5.912838681106325,
-      "grad_norm": 0.0019206481520086527,
-      "learning_rate": 0.0001999827581224586,
-      "loss": 46.0,
-      "step": 36716
-    },
-    {
-      "epoch": 5.912999718185112,
-      "grad_norm": 0.01229285728186369,
-      "learning_rate": 0.00019998275718301458,
-      "loss": 46.0,
-      "step": 36717
-    },
-    {
-      "epoch": 5.9131607552639,
-      "grad_norm": 0.002880864543840289,
-      "learning_rate": 0.00019998275624354492,
-      "loss": 46.0,
-      "step": 36718
-    },
-    {
-      "epoch": 5.913321792342687,
-      "grad_norm": 0.006600085645914078,
-      "learning_rate": 0.00019998275530404966,
-      "loss": 46.0,
-      "step": 36719
-    },
-    {
-      "epoch": 5.913482829421474,
-      "grad_norm": 0.004796166904270649,
-      "learning_rate": 0.00019998275436452882,
-      "loss": 46.0,
-      "step": 36720
-    },
-    {
-      "epoch": 5.913643866500261,
-      "grad_norm": 0.016062308102846146,
-      "learning_rate": 0.0001999827534249824,
-      "loss": 46.0,
-      "step": 36721
-    },
-    {
-      "epoch": 5.913804903579049,
-      "grad_norm": 0.0097150644287467,
-      "learning_rate": 0.0001999827524854104,
-      "loss": 46.0,
-      "step": 36722
-    },
-    {
-      "epoch": 5.913965940657836,
-      "grad_norm": 0.004015443846583366,
-      "learning_rate": 0.00019998275154581278,
-      "loss": 46.0,
-      "step": 36723
-    },
-    {
-      "epoch": 5.914126977736624,
-      "grad_norm": 0.0021828943863511086,
-      "learning_rate": 0.0001999827506061896,
-      "loss": 46.0,
-      "step": 36724
-    },
-    {
-      "epoch": 5.914288014815411,
-      "grad_norm": 0.005539830774068832,
-      "learning_rate": 0.0001999827496665408,
-      "loss": 46.0,
-      "step": 36725
-    },
-    {
-      "epoch": 5.914449051894199,
-      "grad_norm": 0.0038171743508428335,
-      "learning_rate": 0.00019998274872686645,
-      "loss": 46.0,
-      "step": 36726
-    },
-    {
-      "epoch": 5.914610088972986,
-      "grad_norm": 0.005330363754183054,
-      "learning_rate": 0.00019998274778716647,
-      "loss": 46.0,
-      "step": 36727
-    },
-    {
-      "epoch": 5.9147711260517735,
-      "grad_norm": 0.00901678204536438,
-      "learning_rate": 0.0001999827468474409,
-      "loss": 46.0,
-      "step": 36728
-    },
-    {
-      "epoch": 5.914932163130561,
-      "grad_norm": 0.004908686503767967,
-      "learning_rate": 0.00019998274590768975,
-      "loss": 46.0,
-      "step": 36729
-    },
-    {
-      "epoch": 5.915093200209348,
-      "grad_norm": 0.001657033571973443,
-      "learning_rate": 0.00019998274496791304,
-      "loss": 46.0,
-      "step": 36730
-    },
-    {
-      "epoch": 5.915254237288136,
-      "grad_norm": 0.0011588465422391891,
-      "learning_rate": 0.0001999827440281107,
-      "loss": 46.0,
-      "step": 36731
-    },
-    {
-      "epoch": 5.915415274366923,
-      "grad_norm": 0.0031610203441232443,
-      "learning_rate": 0.00019998274308828274,
-      "loss": 46.0,
-      "step": 36732
-    },
-    {
-      "epoch": 5.915576311445711,
-      "grad_norm": 0.0034914512652903795,
-      "learning_rate": 0.00019998274214842927,
-      "loss": 46.0,
-      "step": 36733
-    },
-    {
-      "epoch": 5.915737348524498,
-      "grad_norm": 0.006047399248927832,
-      "learning_rate": 0.00019998274120855015,
-      "loss": 46.0,
-      "step": 36734
-    },
-    {
-      "epoch": 5.915898385603285,
-      "grad_norm": 0.0038502642419189215,
-      "learning_rate": 0.00019998274026864548,
-      "loss": 46.0,
-      "step": 36735
-    },
-    {
-      "epoch": 5.916059422682072,
-      "grad_norm": 0.010237429291009903,
-      "learning_rate": 0.0001999827393287152,
-      "loss": 46.0,
-      "step": 36736
-    },
-    {
-      "epoch": 5.91622045976086,
-      "grad_norm": 0.0006746245780959725,
-      "learning_rate": 0.0001999827383887593,
-      "loss": 46.0,
-      "step": 36737
-    },
-    {
-      "epoch": 5.916381496839647,
-      "grad_norm": 0.0018468481721356511,
-      "learning_rate": 0.00019998273744877785,
-      "loss": 46.0,
-      "step": 36738
-    },
-    {
-      "epoch": 5.9165425339184345,
-      "grad_norm": 0.005950854625552893,
-      "learning_rate": 0.0001999827365087708,
-      "loss": 46.0,
-      "step": 36739
-    },
-    {
-      "epoch": 5.916703570997222,
-      "grad_norm": 0.004224453587085009,
-      "learning_rate": 0.00019998273556873816,
-      "loss": 46.0,
-      "step": 36740
-    },
-    {
-      "epoch": 5.9168646080760094,
-      "grad_norm": 0.002298058243468404,
-      "learning_rate": 0.0001999827346286799,
-      "loss": 46.0,
-      "step": 36741
-    },
-    {
-      "epoch": 5.917025645154797,
-      "grad_norm": 0.0024822931736707687,
-      "learning_rate": 0.00019998273368859607,
-      "loss": 46.0,
-      "step": 36742
-    },
-    {
-      "epoch": 5.917186682233584,
-      "grad_norm": 0.002132893307134509,
-      "learning_rate": 0.00019998273274848667,
-      "loss": 46.0,
-      "step": 36743
-    },
-    {
-      "epoch": 5.917347719312372,
-      "grad_norm": 0.0025236632209271193,
-      "learning_rate": 0.00019998273180835168,
-      "loss": 46.0,
-      "step": 36744
-    },
-    {
-      "epoch": 5.917508756391159,
-      "grad_norm": 0.008523834869265556,
-      "learning_rate": 0.00019998273086819108,
-      "loss": 46.0,
-      "step": 36745
-    },
-    {
-      "epoch": 5.917669793469947,
-      "grad_norm": 0.0021781628020107746,
-      "learning_rate": 0.00019998272992800492,
-      "loss": 46.0,
-      "step": 36746
-    },
-    {
-      "epoch": 5.917830830548734,
-      "grad_norm": 0.003650036873295903,
-      "learning_rate": 0.00019998272898779314,
-      "loss": 46.0,
-      "step": 36747
-    },
-    {
-      "epoch": 5.917991867627522,
-      "grad_norm": 0.013499385677278042,
-      "learning_rate": 0.00019998272804755575,
-      "loss": 46.0,
-      "step": 36748
-    },
-    {
-      "epoch": 5.918152904706309,
-      "grad_norm": 0.005816685501486063,
-      "learning_rate": 0.0001999827271072928,
-      "loss": 46.0,
-      "step": 36749
-    },
-    {
-      "epoch": 5.918313941785096,
-      "grad_norm": 0.019862115383148193,
-      "learning_rate": 0.00019998272616700424,
-      "loss": 46.0,
-      "step": 36750
-    },
-    {
-      "epoch": 5.918474978863883,
-      "grad_norm": 0.006577814929187298,
-      "learning_rate": 0.00019998272522669011,
-      "loss": 46.0,
-      "step": 36751
-    },
-    {
-      "epoch": 5.9186360159426705,
-      "grad_norm": 0.006244547665119171,
-      "learning_rate": 0.00019998272428635037,
-      "loss": 46.0,
-      "step": 36752
-    },
-    {
-      "epoch": 5.918797053021458,
-      "grad_norm": 0.006949910894036293,
-      "learning_rate": 0.00019998272334598505,
-      "loss": 46.0,
-      "step": 36753
-    },
-    {
-      "epoch": 5.918958090100245,
-      "grad_norm": 0.021270664408802986,
-      "learning_rate": 0.00019998272240559414,
-      "loss": 46.0,
-      "step": 36754
-    },
-    {
-      "epoch": 5.919119127179033,
-      "grad_norm": 0.007246372755616903,
-      "learning_rate": 0.00019998272146517764,
-      "loss": 46.0,
-      "step": 36755
-    },
-    {
-      "epoch": 5.91928016425782,
-      "grad_norm": 0.008000325411558151,
-      "learning_rate": 0.00019998272052473555,
-      "loss": 46.0,
-      "step": 36756
-    },
-    {
-      "epoch": 5.919441201336608,
-      "grad_norm": 0.0015457691624760628,
-      "learning_rate": 0.00019998271958426787,
-      "loss": 46.0,
-      "step": 36757
-    },
-    {
-      "epoch": 5.919602238415395,
-      "grad_norm": 0.009535088203847408,
-      "learning_rate": 0.0001999827186437746,
-      "loss": 46.0,
-      "step": 36758
-    },
-    {
-      "epoch": 5.919763275494183,
-      "grad_norm": 0.013317562639713287,
-      "learning_rate": 0.00019998271770325574,
-      "loss": 46.0,
-      "step": 36759
-    },
-    {
-      "epoch": 5.91992431257297,
-      "grad_norm": 0.008625675924122334,
-      "learning_rate": 0.0001999827167627113,
-      "loss": 46.0,
-      "step": 36760
-    },
-    {
-      "epoch": 5.920085349651758,
-      "grad_norm": 0.002901580650359392,
-      "learning_rate": 0.00019998271582214125,
-      "loss": 46.0,
-      "step": 36761
-    },
-    {
-      "epoch": 5.920246386730545,
-      "grad_norm": 0.006856206338852644,
-      "learning_rate": 0.00019998271488154564,
-      "loss": 46.0,
-      "step": 36762
-    },
-    {
-      "epoch": 5.920407423809332,
-      "grad_norm": 0.005245402920991182,
-      "learning_rate": 0.00019998271394092442,
-      "loss": 46.0,
-      "step": 36763
-    },
-    {
-      "epoch": 5.92056846088812,
-      "grad_norm": 0.002144214464351535,
-      "learning_rate": 0.00019998271300027758,
-      "loss": 46.0,
-      "step": 36764
-    },
-    {
-      "epoch": 5.9207294979669065,
-      "grad_norm": 0.008015332743525505,
-      "learning_rate": 0.00019998271205960518,
-      "loss": 46.0,
-      "step": 36765
-    },
-    {
-      "epoch": 5.920890535045694,
-      "grad_norm": 0.014490757137537003,
-      "learning_rate": 0.0001999827111189072,
-      "loss": 46.0,
-      "step": 36766
-    },
-    {
-      "epoch": 5.921051572124481,
-      "grad_norm": 0.00442791823297739,
-      "learning_rate": 0.0001999827101781836,
-      "loss": 46.0,
-      "step": 36767
-    },
-    {
-      "epoch": 5.921212609203269,
-      "grad_norm": 0.006829854566603899,
-      "learning_rate": 0.0001999827092374344,
-      "loss": 46.0,
-      "step": 36768
-    },
-    {
-      "epoch": 5.921373646282056,
-      "grad_norm": 0.008199771866202354,
-      "learning_rate": 0.00019998270829665966,
-      "loss": 46.0,
-      "step": 36769
-    },
-    {
-      "epoch": 5.921534683360844,
-      "grad_norm": 0.015690838918089867,
-      "learning_rate": 0.0001999827073558593,
-      "loss": 46.0,
-      "step": 36770
-    },
-    {
-      "epoch": 5.921695720439631,
-      "grad_norm": 0.0020068727899342775,
-      "learning_rate": 0.00019998270641503335,
-      "loss": 46.0,
-      "step": 36771
-    },
-    {
-      "epoch": 5.921856757518419,
-      "grad_norm": 0.011578435078263283,
-      "learning_rate": 0.0001999827054741818,
-      "loss": 46.0,
-      "step": 36772
-    },
-    {
-      "epoch": 5.922017794597206,
-      "grad_norm": 0.009915011003613472,
-      "learning_rate": 0.00019998270453330469,
-      "loss": 46.0,
-      "step": 36773
-    },
-    {
-      "epoch": 5.922178831675994,
-      "grad_norm": 0.00173856969922781,
-      "learning_rate": 0.00019998270359240198,
-      "loss": 46.0,
-      "step": 36774
-    },
-    {
-      "epoch": 5.922339868754781,
-      "grad_norm": 0.006825334392488003,
-      "learning_rate": 0.00019998270265147365,
-      "loss": 46.0,
-      "step": 36775
-    },
-    {
-      "epoch": 5.9225009058335685,
-      "grad_norm": 0.0027432385832071304,
-      "learning_rate": 0.00019998270171051977,
-      "loss": 46.0,
-      "step": 36776
-    },
-    {
-      "epoch": 5.922661942912356,
-      "grad_norm": 0.008600936271250248,
-      "learning_rate": 0.00019998270076954027,
-      "loss": 46.0,
-      "step": 36777
-    },
-    {
-      "epoch": 5.9228229799911425,
-      "grad_norm": 0.0015825724694877863,
-      "learning_rate": 0.00019998269982853518,
-      "loss": 46.0,
-      "step": 36778
-    },
-    {
-      "epoch": 5.922984017069931,
-      "grad_norm": 0.008640332147479057,
-      "learning_rate": 0.0001999826988875045,
-      "loss": 46.0,
-      "step": 36779
-    },
-    {
-      "epoch": 5.923145054148717,
-      "grad_norm": 0.00999928917735815,
-      "learning_rate": 0.00019998269794644827,
-      "loss": 46.0,
-      "step": 36780
-    },
-    {
-      "epoch": 5.923306091227505,
-      "grad_norm": 0.0022006940562278032,
-      "learning_rate": 0.0001999826970053664,
-      "loss": 46.0,
-      "step": 36781
-    },
-    {
-      "epoch": 5.923467128306292,
-      "grad_norm": 0.0022133244201540947,
-      "learning_rate": 0.00019998269606425896,
-      "loss": 46.0,
-      "step": 36782
-    },
-    {
-      "epoch": 5.92362816538508,
-      "grad_norm": 0.005816758144646883,
-      "learning_rate": 0.00019998269512312594,
-      "loss": 46.0,
-      "step": 36783
-    },
-    {
-      "epoch": 5.923789202463867,
-      "grad_norm": 0.010484855622053146,
-      "learning_rate": 0.0001999826941819673,
-      "loss": 46.0,
-      "step": 36784
-    },
-    {
-      "epoch": 5.923950239542655,
-      "grad_norm": 0.005769622046500444,
-      "learning_rate": 0.00019998269324078308,
-      "loss": 46.0,
-      "step": 36785
-    },
-    {
-      "epoch": 5.924111276621442,
-      "grad_norm": 0.0020913099870085716,
-      "learning_rate": 0.0001999826922995733,
-      "loss": 46.0,
-      "step": 36786
-    },
-    {
-      "epoch": 5.9242723137002296,
-      "grad_norm": 0.001544525264762342,
-      "learning_rate": 0.0001999826913583379,
-      "loss": 46.0,
-      "step": 36787
-    },
-    {
-      "epoch": 5.924433350779017,
-      "grad_norm": 0.005578508134931326,
-      "learning_rate": 0.00019998269041707691,
-      "loss": 46.0,
-      "step": 36788
-    },
-    {
-      "epoch": 5.9245943878578045,
-      "grad_norm": 0.00633410457521677,
-      "learning_rate": 0.00019998268947579034,
-      "loss": 46.0,
-      "step": 36789
-    },
-    {
-      "epoch": 5.924755424936592,
-      "grad_norm": 0.008016320876777172,
-      "learning_rate": 0.00019998268853447818,
-      "loss": 46.0,
-      "step": 36790
-    },
-    {
-      "epoch": 5.924916462015379,
-      "grad_norm": 0.0036888362374156713,
-      "learning_rate": 0.0001999826875931404,
-      "loss": 46.0,
-      "step": 36791
-    },
-    {
-      "epoch": 5.925077499094167,
-      "grad_norm": 0.004931919742375612,
-      "learning_rate": 0.00019998268665177707,
-      "loss": 46.0,
-      "step": 36792
-    },
-    {
-      "epoch": 5.925238536172953,
-      "grad_norm": 0.00144600635394454,
-      "learning_rate": 0.00019998268571038815,
-      "loss": 46.0,
-      "step": 36793
-    },
-    {
-      "epoch": 5.925399573251741,
-      "grad_norm": 0.002845043782144785,
-      "learning_rate": 0.0001999826847689736,
-      "loss": 46.0,
-      "step": 36794
-    },
-    {
-      "epoch": 5.925560610330528,
-      "grad_norm": 0.01663287729024887,
-      "learning_rate": 0.0001999826838275335,
-      "loss": 46.0,
-      "step": 36795
-    },
-    {
-      "epoch": 5.925721647409316,
-      "grad_norm": 0.005727509967982769,
-      "learning_rate": 0.00019998268288606779,
-      "loss": 46.0,
-      "step": 36796
-    },
-    {
-      "epoch": 5.925882684488103,
-      "grad_norm": 0.0028591693844646215,
-      "learning_rate": 0.0001999826819445765,
-      "loss": 46.0,
-      "step": 36797
-    },
-    {
-      "epoch": 5.926043721566891,
-      "grad_norm": 0.004529683385044336,
-      "learning_rate": 0.00019998268100305958,
-      "loss": 46.0,
-      "step": 36798
-    },
-    {
-      "epoch": 5.926204758645678,
-      "grad_norm": 0.0013791258679702878,
-      "learning_rate": 0.00019998268006151713,
-      "loss": 46.0,
-      "step": 36799
-    },
-    {
-      "epoch": 5.9263657957244655,
-      "grad_norm": 0.002873591845855117,
-      "learning_rate": 0.00019998267911994905,
-      "loss": 46.0,
-      "step": 36800
-    },
-    {
-      "epoch": 5.926526832803253,
-      "grad_norm": 0.0020084616262465715,
-      "learning_rate": 0.00019998267817835538,
-      "loss": 46.0,
-      "step": 36801
-    },
-    {
-      "epoch": 5.92668786988204,
-      "grad_norm": 0.009874220006167889,
-      "learning_rate": 0.00019998267723673614,
-      "loss": 46.0,
-      "step": 36802
-    },
-    {
-      "epoch": 5.926848906960828,
-      "grad_norm": 0.0038444348610937595,
-      "learning_rate": 0.0001999826762950913,
-      "loss": 46.0,
-      "step": 36803
-    },
-    {
-      "epoch": 5.927009944039615,
-      "grad_norm": 0.002023317152634263,
-      "learning_rate": 0.00019998267535342086,
-      "loss": 46.0,
-      "step": 36804
-    },
-    {
-      "epoch": 5.927170981118403,
-      "grad_norm": 0.0018444063607603312,
-      "learning_rate": 0.00019998267441172484,
-      "loss": 46.0,
-      "step": 36805
-    },
-    {
-      "epoch": 5.92733201819719,
-      "grad_norm": 0.009191123768687248,
-      "learning_rate": 0.0001999826734700032,
-      "loss": 46.0,
-      "step": 36806
-    },
-    {
-      "epoch": 5.927493055275978,
-      "grad_norm": 0.003837514203041792,
-      "learning_rate": 0.000199982672528256,
-      "loss": 46.0,
-      "step": 36807
-    },
-    {
-      "epoch": 5.927654092354764,
-      "grad_norm": 0.005431676749140024,
-      "learning_rate": 0.00019998267158648323,
-      "loss": 46.0,
-      "step": 36808
-    },
-    {
-      "epoch": 5.927815129433552,
-      "grad_norm": 0.005784062203019857,
-      "learning_rate": 0.00019998267064468483,
-      "loss": 46.0,
-      "step": 36809
-    },
-    {
-      "epoch": 5.927976166512339,
-      "grad_norm": 0.012924952432513237,
-      "learning_rate": 0.00019998266970286087,
-      "loss": 46.0,
-      "step": 36810
-    },
-    {
-      "epoch": 5.928137203591127,
-      "grad_norm": 0.006054857280105352,
-      "learning_rate": 0.0001999826687610113,
-      "loss": 46.0,
-      "step": 36811
-    },
-    {
-      "epoch": 5.928298240669914,
-      "grad_norm": 0.01029572170227766,
-      "learning_rate": 0.00019998266781913614,
-      "loss": 46.0,
-      "step": 36812
-    },
-    {
-      "epoch": 5.9284592777487015,
-      "grad_norm": 0.0015591337578371167,
-      "learning_rate": 0.00019998266687723542,
-      "loss": 46.0,
-      "step": 36813
-    },
-    {
-      "epoch": 5.928620314827489,
-      "grad_norm": 0.008079933002591133,
-      "learning_rate": 0.00019998266593530906,
-      "loss": 46.0,
-      "step": 36814
-    },
-    {
-      "epoch": 5.928781351906276,
-      "grad_norm": 0.005458934232592583,
-      "learning_rate": 0.00019998266499335717,
-      "loss": 46.0,
-      "step": 36815
-    },
-    {
-      "epoch": 5.928942388985064,
-      "grad_norm": 0.002612989628687501,
-      "learning_rate": 0.00019998266405137964,
-      "loss": 46.0,
-      "step": 36816
-    },
-    {
-      "epoch": 5.929103426063851,
-      "grad_norm": 0.002193618332967162,
-      "learning_rate": 0.00019998266310937651,
-      "loss": 46.0,
-      "step": 36817
-    },
-    {
-      "epoch": 5.929264463142639,
-      "grad_norm": 0.020804710686206818,
-      "learning_rate": 0.00019998266216734783,
-      "loss": 46.0,
-      "step": 36818
-    },
-    {
-      "epoch": 5.929425500221426,
-      "grad_norm": 0.0009054761030711234,
-      "learning_rate": 0.0001999826612252935,
-      "loss": 46.0,
-      "step": 36819
-    },
-    {
-      "epoch": 5.929586537300214,
-      "grad_norm": 0.003950676415115595,
-      "learning_rate": 0.00019998266028321365,
-      "loss": 46.0,
-      "step": 36820
-    },
-    {
-      "epoch": 5.929747574379001,
-      "grad_norm": 0.0007985651027411222,
-      "learning_rate": 0.00019998265934110818,
-      "loss": 46.0,
-      "step": 36821
-    },
-    {
-      "epoch": 5.929908611457789,
-      "grad_norm": 0.001724847941659391,
-      "learning_rate": 0.00019998265839897715,
-      "loss": 46.0,
-      "step": 36822
-    },
-    {
-      "epoch": 5.930069648536575,
-      "grad_norm": 0.0027348280418664217,
-      "learning_rate": 0.00019998265745682048,
-      "loss": 46.0,
-      "step": 36823
-    },
-    {
-      "epoch": 5.930230685615363,
-      "grad_norm": 0.012773178517818451,
-      "learning_rate": 0.00019998265651463822,
-      "loss": 46.0,
-      "step": 36824
-    },
-    {
-      "epoch": 5.93039172269415,
-      "grad_norm": 0.004951681476086378,
-      "learning_rate": 0.0001999826555724304,
-      "loss": 46.0,
-      "step": 36825
-    },
-    {
-      "epoch": 5.9305527597729375,
-      "grad_norm": 0.010246754623949528,
-      "learning_rate": 0.000199982654630197,
-      "loss": 46.0,
-      "step": 36826
-    },
-    {
-      "epoch": 5.930713796851725,
-      "grad_norm": 0.008923030458390713,
-      "learning_rate": 0.00019998265368793798,
-      "loss": 46.0,
-      "step": 36827
-    },
-    {
-      "epoch": 5.930874833930512,
-      "grad_norm": 0.0013812516117468476,
-      "learning_rate": 0.00019998265274565337,
-      "loss": 46.0,
-      "step": 36828
-    },
-    {
-      "epoch": 5.9310358710093,
-      "grad_norm": 0.010209945961833,
-      "learning_rate": 0.0001999826518033432,
-      "loss": 46.0,
-      "step": 36829
-    },
-    {
-      "epoch": 5.931196908088087,
-      "grad_norm": 0.0032221318688243628,
-      "learning_rate": 0.0001999826508610074,
-      "loss": 46.0,
-      "step": 36830
-    },
-    {
-      "epoch": 5.931357945166875,
-      "grad_norm": 0.00614152243360877,
-      "learning_rate": 0.00019998264991864602,
-      "loss": 46.0,
-      "step": 36831
-    },
-    {
-      "epoch": 5.931518982245662,
-      "grad_norm": 0.003793375799432397,
-      "learning_rate": 0.00019998264897625907,
-      "loss": 46.0,
-      "step": 36832
-    },
-    {
-      "epoch": 5.93168001932445,
-      "grad_norm": 0.006623846013098955,
-      "learning_rate": 0.00019998264803384653,
-      "loss": 46.0,
-      "step": 36833
-    },
-    {
-      "epoch": 5.931841056403237,
-      "grad_norm": 0.0018425935413688421,
-      "learning_rate": 0.00019998264709140834,
-      "loss": 46.0,
-      "step": 36834
-    },
-    {
-      "epoch": 5.932002093482025,
-      "grad_norm": 0.003296756185591221,
-      "learning_rate": 0.00019998264614894465,
-      "loss": 46.0,
-      "step": 36835
-    },
-    {
-      "epoch": 5.932163130560812,
-      "grad_norm": 0.0028366323094815016,
-      "learning_rate": 0.00019998264520645532,
-      "loss": 46.0,
-      "step": 36836
-    },
-    {
-      "epoch": 5.9323241676395995,
-      "grad_norm": 0.0022995176259428263,
-      "learning_rate": 0.0001999826442639404,
-      "loss": 46.0,
-      "step": 36837
-    },
-    {
-      "epoch": 5.932485204718386,
-      "grad_norm": 0.016457993537187576,
-      "learning_rate": 0.0001999826433213999,
-      "loss": 46.0,
-      "step": 36838
-    },
-    {
-      "epoch": 5.9326462417971735,
-      "grad_norm": 0.0038181720301508904,
-      "learning_rate": 0.0001999826423788338,
-      "loss": 46.0,
-      "step": 36839
-    },
-    {
-      "epoch": 5.932807278875961,
-      "grad_norm": 0.01116947177797556,
-      "learning_rate": 0.0001999826414362421,
-      "loss": 46.0,
-      "step": 36840
-    },
-    {
-      "epoch": 5.932968315954748,
-      "grad_norm": 0.010643711313605309,
-      "learning_rate": 0.00019998264049362482,
-      "loss": 46.0,
-      "step": 36841
-    },
-    {
-      "epoch": 5.933129353033536,
-      "grad_norm": 0.0033071911893785,
-      "learning_rate": 0.00019998263955098194,
-      "loss": 46.0,
-      "step": 36842
-    },
-    {
-      "epoch": 5.933290390112323,
-      "grad_norm": 0.004507561679929495,
-      "learning_rate": 0.0001999826386083135,
-      "loss": 46.0,
-      "step": 36843
-    },
-    {
-      "epoch": 5.933451427191111,
-      "grad_norm": 0.007893814705312252,
-      "learning_rate": 0.00019998263766561944,
-      "loss": 46.0,
-      "step": 36844
-    },
-    {
-      "epoch": 5.933612464269898,
-      "grad_norm": 0.005530776921659708,
-      "learning_rate": 0.0001999826367228998,
-      "loss": 46.0,
-      "step": 36845
-    },
-    {
-      "epoch": 5.933773501348686,
-      "grad_norm": 0.007638771086931229,
-      "learning_rate": 0.0001999826357801546,
-      "loss": 46.0,
-      "step": 36846
-    },
-    {
-      "epoch": 5.933934538427473,
-      "grad_norm": 0.0024144684430211782,
-      "learning_rate": 0.00019998263483738375,
-      "loss": 46.0,
-      "step": 36847
-    },
-    {
-      "epoch": 5.9340955755062605,
-      "grad_norm": 0.01121516153216362,
-      "learning_rate": 0.00019998263389458734,
-      "loss": 46.0,
-      "step": 36848
-    },
-    {
-      "epoch": 5.934256612585048,
-      "grad_norm": 0.0031699230894446373,
-      "learning_rate": 0.00019998263295176532,
-      "loss": 46.0,
-      "step": 36849
-    },
-    {
-      "epoch": 5.9344176496638354,
-      "grad_norm": 0.001713825506158173,
-      "learning_rate": 0.00019998263200891775,
-      "loss": 46.0,
-      "step": 36850
-    },
-    {
-      "epoch": 5.934578686742622,
-      "grad_norm": 0.012295176275074482,
-      "learning_rate": 0.00019998263106604455,
-      "loss": 46.0,
-      "step": 36851
-    },
-    {
-      "epoch": 5.93473972382141,
-      "grad_norm": 0.0014337149914354086,
-      "learning_rate": 0.0001999826301231458,
-      "loss": 46.0,
-      "step": 36852
-    },
-    {
-      "epoch": 5.934900760900197,
-      "grad_norm": 0.007024618797004223,
-      "learning_rate": 0.00019998262918022143,
-      "loss": 46.0,
-      "step": 36853
-    },
-    {
-      "epoch": 5.935061797978984,
-      "grad_norm": 0.00954799260944128,
-      "learning_rate": 0.00019998262823727147,
-      "loss": 46.0,
-      "step": 36854
-    },
-    {
-      "epoch": 5.935222835057772,
-      "grad_norm": 0.005930591840296984,
-      "learning_rate": 0.00019998262729429596,
-      "loss": 46.0,
-      "step": 36855
-    },
-    {
-      "epoch": 5.935383872136559,
-      "grad_norm": 0.006837198976427317,
-      "learning_rate": 0.0001999826263512948,
-      "loss": 46.0,
-      "step": 36856
-    },
-    {
-      "epoch": 5.935544909215347,
-      "grad_norm": 0.003982583526521921,
-      "learning_rate": 0.00019998262540826809,
-      "loss": 46.0,
-      "step": 36857
-    },
-    {
-      "epoch": 5.935705946294134,
-      "grad_norm": 0.0175002571195364,
-      "learning_rate": 0.00019998262446521578,
-      "loss": 46.0,
-      "step": 36858
-    },
-    {
-      "epoch": 5.935866983372922,
-      "grad_norm": 0.0042568654753267765,
-      "learning_rate": 0.00019998262352213786,
-      "loss": 46.0,
-      "step": 36859
-    },
-    {
-      "epoch": 5.936028020451709,
-      "grad_norm": 0.0033177966251969337,
-      "learning_rate": 0.00019998262257903438,
-      "loss": 46.0,
-      "step": 36860
-    },
-    {
-      "epoch": 5.9361890575304965,
-      "grad_norm": 0.0018974541453644633,
-      "learning_rate": 0.0001999826216359053,
-      "loss": 46.0,
-      "step": 36861
-    },
-    {
-      "epoch": 5.936350094609284,
-      "grad_norm": 0.005490519572049379,
-      "learning_rate": 0.0001999826206927506,
-      "loss": 46.0,
-      "step": 36862
-    },
-    {
-      "epoch": 5.936511131688071,
-      "grad_norm": 0.004988665692508221,
-      "learning_rate": 0.00019998261974957035,
-      "loss": 46.0,
-      "step": 36863
-    },
-    {
-      "epoch": 5.936672168766859,
-      "grad_norm": 0.014877923764288425,
-      "learning_rate": 0.0001999826188063645,
-      "loss": 46.0,
-      "step": 36864
-    },
-    {
-      "epoch": 5.936833205845646,
-      "grad_norm": 0.00861698854714632,
-      "learning_rate": 0.00019998261786313305,
-      "loss": 46.0,
-      "step": 36865
-    },
-    {
-      "epoch": 5.936994242924433,
-      "grad_norm": 0.002893635770305991,
-      "learning_rate": 0.00019998261691987602,
-      "loss": 46.0,
-      "step": 36866
-    },
-    {
-      "epoch": 5.937155280003221,
-      "grad_norm": 0.010076380334794521,
-      "learning_rate": 0.00019998261597659335,
-      "loss": 46.0,
-      "step": 36867
-    },
-    {
-      "epoch": 5.937316317082008,
-      "grad_norm": 0.008594764396548271,
-      "learning_rate": 0.00019998261503328515,
-      "loss": 46.0,
-      "step": 36868
-    },
-    {
-      "epoch": 5.937477354160795,
-      "grad_norm": 0.006651430856436491,
-      "learning_rate": 0.00019998261408995133,
-      "loss": 46.0,
-      "step": 36869
-    },
-    {
-      "epoch": 5.937638391239583,
-      "grad_norm": 0.025923868641257286,
-      "learning_rate": 0.00019998261314659193,
-      "loss": 46.0,
-      "step": 36870
-    },
-    {
-      "epoch": 5.93779942831837,
-      "grad_norm": 0.0025881456676870584,
-      "learning_rate": 0.00019998261220320696,
-      "loss": 46.0,
-      "step": 36871
-    },
-    {
-      "epoch": 5.937960465397158,
-      "grad_norm": 0.0168552715331316,
-      "learning_rate": 0.00019998261125979636,
-      "loss": 46.0,
-      "step": 36872
-    },
-    {
-      "epoch": 5.938121502475945,
-      "grad_norm": 0.002950269030407071,
-      "learning_rate": 0.0001999826103163602,
-      "loss": 46.0,
-      "step": 36873
-    },
-    {
-      "epoch": 5.9382825395547325,
-      "grad_norm": 0.01831338368356228,
-      "learning_rate": 0.00019998260937289844,
-      "loss": 46.0,
-      "step": 36874
-    },
-    {
-      "epoch": 5.93844357663352,
-      "grad_norm": 0.00851343385875225,
-      "learning_rate": 0.0001999826084294111,
-      "loss": 46.0,
-      "step": 36875
-    },
-    {
-      "epoch": 5.938604613712307,
-      "grad_norm": 0.0019877853337675333,
-      "learning_rate": 0.00019998260748589814,
-      "loss": 46.0,
-      "step": 36876
-    },
-    {
-      "epoch": 5.938765650791095,
-      "grad_norm": 0.0016741901636123657,
-      "learning_rate": 0.0001999826065423596,
-      "loss": 46.0,
-      "step": 36877
-    },
-    {
-      "epoch": 5.938926687869882,
-      "grad_norm": 0.0024467746261507273,
-      "learning_rate": 0.0001999826055987955,
-      "loss": 46.0,
-      "step": 36878
-    },
-    {
-      "epoch": 5.93908772494867,
-      "grad_norm": 0.002907415386289358,
-      "learning_rate": 0.00019998260465520578,
-      "loss": 46.0,
-      "step": 36879
-    },
-    {
-      "epoch": 5.939248762027457,
-      "grad_norm": 0.004205035045742989,
-      "learning_rate": 0.00019998260371159048,
-      "loss": 46.0,
-      "step": 36880
-    },
-    {
-      "epoch": 5.939409799106244,
-      "grad_norm": 0.0022435863502323627,
-      "learning_rate": 0.0001999826027679496,
-      "loss": 46.0,
-      "step": 36881
-    },
-    {
-      "epoch": 5.939570836185031,
-      "grad_norm": 0.0032009012065827847,
-      "learning_rate": 0.0001999826018242831,
-      "loss": 46.0,
-      "step": 36882
-    },
-    {
-      "epoch": 5.939731873263819,
-      "grad_norm": 0.0007960580987855792,
-      "learning_rate": 0.00019998260088059102,
-      "loss": 46.0,
-      "step": 36883
-    },
-    {
-      "epoch": 5.939892910342606,
-      "grad_norm": 0.006090050563216209,
-      "learning_rate": 0.00019998259993687337,
-      "loss": 46.0,
-      "step": 36884
-    },
-    {
-      "epoch": 5.940053947421394,
-      "grad_norm": 0.003619585884734988,
-      "learning_rate": 0.0001999825989931301,
-      "loss": 46.0,
-      "step": 36885
-    },
-    {
-      "epoch": 5.940214984500181,
-      "grad_norm": 0.006734329741448164,
-      "learning_rate": 0.00019998259804936125,
-      "loss": 46.0,
-      "step": 36886
-    },
-    {
-      "epoch": 5.9403760215789685,
-      "grad_norm": 0.003483704524114728,
-      "learning_rate": 0.0001999825971055668,
-      "loss": 46.0,
-      "step": 36887
-    },
-    {
-      "epoch": 5.940537058657756,
-      "grad_norm": 0.0039081210270524025,
-      "learning_rate": 0.00019998259616174678,
-      "loss": 46.0,
-      "step": 36888
-    },
-    {
-      "epoch": 5.940698095736543,
-      "grad_norm": 0.001193365198560059,
-      "learning_rate": 0.00019998259521790117,
-      "loss": 46.0,
-      "step": 36889
-    },
-    {
-      "epoch": 5.940859132815331,
-      "grad_norm": 0.0035276266280561686,
-      "learning_rate": 0.00019998259427402997,
-      "loss": 46.0,
-      "step": 36890
-    },
-    {
-      "epoch": 5.941020169894118,
-      "grad_norm": 0.013093777000904083,
-      "learning_rate": 0.00019998259333013315,
-      "loss": 46.0,
-      "step": 36891
-    },
-    {
-      "epoch": 5.941181206972906,
-      "grad_norm": 0.003916393965482712,
-      "learning_rate": 0.00019998259238621077,
-      "loss": 46.0,
-      "step": 36892
-    },
-    {
-      "epoch": 5.941342244051693,
-      "grad_norm": 0.004663741681724787,
-      "learning_rate": 0.00019998259144226278,
-      "loss": 46.0,
-      "step": 36893
-    },
-    {
-      "epoch": 5.941503281130481,
-      "grad_norm": 0.004135070368647575,
-      "learning_rate": 0.00019998259049828923,
-      "loss": 46.0,
-      "step": 36894
-    },
-    {
-      "epoch": 5.941664318209268,
-      "grad_norm": 0.009576123207807541,
-      "learning_rate": 0.00019998258955429007,
-      "loss": 46.0,
-      "step": 36895
-    },
-    {
-      "epoch": 5.941825355288055,
-      "grad_norm": 0.0024867786560207605,
-      "learning_rate": 0.00019998258861026532,
-      "loss": 46.0,
-      "step": 36896
-    },
-    {
-      "epoch": 5.941986392366842,
-      "grad_norm": 0.001111357705667615,
-      "learning_rate": 0.00019998258766621495,
-      "loss": 46.0,
-      "step": 36897
-    },
-    {
-      "epoch": 5.94214742944563,
-      "grad_norm": 0.0033197253942489624,
-      "learning_rate": 0.00019998258672213902,
-      "loss": 46.0,
-      "step": 36898
-    },
-    {
-      "epoch": 5.942308466524417,
-      "grad_norm": 0.0030090510845184326,
-      "learning_rate": 0.0001999825857780375,
-      "loss": 46.0,
-      "step": 36899
-    },
-    {
-      "epoch": 5.9424695036032045,
-      "grad_norm": 0.004534139297902584,
-      "learning_rate": 0.00019998258483391038,
-      "loss": 46.0,
-      "step": 36900
-    },
-    {
-      "epoch": 5.942630540681992,
-      "grad_norm": 0.0017294470453634858,
-      "learning_rate": 0.0001999825838897577,
-      "loss": 46.0,
-      "step": 36901
-    },
-    {
-      "epoch": 5.942791577760779,
-      "grad_norm": 0.001450419775210321,
-      "learning_rate": 0.00019998258294557942,
-      "loss": 46.0,
-      "step": 36902
-    },
-    {
-      "epoch": 5.942952614839567,
-      "grad_norm": 0.0018378933891654015,
-      "learning_rate": 0.00019998258200137553,
-      "loss": 46.0,
-      "step": 36903
-    },
-    {
-      "epoch": 5.943113651918354,
-      "grad_norm": 0.0013845686335116625,
-      "learning_rate": 0.00019998258105714605,
-      "loss": 46.0,
-      "step": 36904
-    },
-    {
-      "epoch": 5.943274688997142,
-      "grad_norm": 0.003157664556056261,
-      "learning_rate": 0.000199982580112891,
-      "loss": 46.0,
-      "step": 36905
-    },
-    {
-      "epoch": 5.943435726075929,
-      "grad_norm": 0.0026509626768529415,
-      "learning_rate": 0.00019998257916861034,
-      "loss": 46.0,
-      "step": 36906
-    },
-    {
-      "epoch": 5.943596763154717,
-      "grad_norm": 0.016884757205843925,
-      "learning_rate": 0.00019998257822430407,
-      "loss": 46.0,
-      "step": 36907
-    },
-    {
-      "epoch": 5.943757800233504,
-      "grad_norm": 0.0024585712235420942,
-      "learning_rate": 0.00019998257727997225,
-      "loss": 46.0,
-      "step": 36908
-    },
-    {
-      "epoch": 5.9439188373122915,
-      "grad_norm": 0.0050835562869906425,
-      "learning_rate": 0.0001999825763356148,
-      "loss": 46.0,
-      "step": 36909
-    },
-    {
-      "epoch": 5.944079874391079,
-      "grad_norm": 0.006497082766145468,
-      "learning_rate": 0.0001999825753912318,
-      "loss": 46.0,
-      "step": 36910
-    },
-    {
-      "epoch": 5.9442409114698656,
-      "grad_norm": 0.002081113401800394,
-      "learning_rate": 0.0001999825744468232,
-      "loss": 46.0,
-      "step": 36911
-    },
-    {
-      "epoch": 5.944401948548653,
-      "grad_norm": 0.0036203069612383842,
-      "learning_rate": 0.000199982573502389,
-      "loss": 46.0,
-      "step": 36912
-    },
-    {
-      "epoch": 5.9445629856274405,
-      "grad_norm": 0.003072316525503993,
-      "learning_rate": 0.00019998257255792923,
-      "loss": 46.0,
-      "step": 36913
-    },
-    {
-      "epoch": 5.944724022706228,
-      "grad_norm": 0.004336175974458456,
-      "learning_rate": 0.00019998257161344386,
-      "loss": 46.0,
-      "step": 36914
-    },
-    {
-      "epoch": 5.944885059785015,
-      "grad_norm": 0.004530665930360556,
-      "learning_rate": 0.00019998257066893287,
-      "loss": 46.0,
-      "step": 36915
-    },
-    {
-      "epoch": 5.945046096863803,
-      "grad_norm": 0.004490611143410206,
-      "learning_rate": 0.00019998256972439632,
-      "loss": 46.0,
-      "step": 36916
-    },
-    {
-      "epoch": 5.94520713394259,
-      "grad_norm": 0.009018462151288986,
-      "learning_rate": 0.00019998256877983416,
-      "loss": 46.0,
-      "step": 36917
-    },
-    {
-      "epoch": 5.945368171021378,
-      "grad_norm": 0.007181808818131685,
-      "learning_rate": 0.00019998256783524643,
-      "loss": 46.0,
-      "step": 36918
-    },
-    {
-      "epoch": 5.945529208100165,
-      "grad_norm": 0.009374646469950676,
-      "learning_rate": 0.0001999825668906331,
-      "loss": 46.0,
-      "step": 36919
-    },
-    {
-      "epoch": 5.945690245178953,
-      "grad_norm": 0.0029697518330067396,
-      "learning_rate": 0.0001999825659459942,
-      "loss": 46.0,
-      "step": 36920
-    },
-    {
-      "epoch": 5.94585128225774,
-      "grad_norm": 0.011085838079452515,
-      "learning_rate": 0.00019998256500132965,
-      "loss": 46.0,
-      "step": 36921
-    },
-    {
-      "epoch": 5.9460123193365275,
-      "grad_norm": 0.0015818714164197445,
-      "learning_rate": 0.00019998256405663956,
-      "loss": 46.0,
-      "step": 36922
-    },
-    {
-      "epoch": 5.946173356415315,
-      "grad_norm": 0.0025518531911075115,
-      "learning_rate": 0.00019998256311192387,
-      "loss": 46.0,
-      "step": 36923
-    },
-    {
-      "epoch": 5.9463343934941015,
-      "grad_norm": 0.007699660491198301,
-      "learning_rate": 0.0001999825621671826,
-      "loss": 46.0,
-      "step": 36924
-    },
-    {
-      "epoch": 5.94649543057289,
-      "grad_norm": 0.001772611285559833,
-      "learning_rate": 0.00019998256122241573,
-      "loss": 46.0,
-      "step": 36925
-    },
-    {
-      "epoch": 5.946656467651676,
-      "grad_norm": 0.006378230173140764,
-      "learning_rate": 0.00019998256027762326,
-      "loss": 46.0,
-      "step": 36926
-    },
-    {
-      "epoch": 5.946817504730464,
-      "grad_norm": 0.00555045111104846,
-      "learning_rate": 0.00019998255933280522,
-      "loss": 46.0,
-      "step": 36927
-    },
-    {
-      "epoch": 5.946978541809251,
-      "grad_norm": 0.0040244730189442635,
-      "learning_rate": 0.00019998255838796157,
-      "loss": 46.0,
-      "step": 36928
-    },
-    {
-      "epoch": 5.947139578888039,
-      "grad_norm": 0.0026228982023894787,
-      "learning_rate": 0.00019998255744309233,
-      "loss": 46.0,
-      "step": 36929
-    },
-    {
-      "epoch": 5.947300615966826,
-      "grad_norm": 0.005417807027697563,
-      "learning_rate": 0.0001999825564981975,
-      "loss": 46.0,
-      "step": 36930
-    },
-    {
-      "epoch": 5.947461653045614,
-      "grad_norm": 0.018967410549521446,
-      "learning_rate": 0.0001999825555532771,
-      "loss": 46.0,
-      "step": 36931
-    },
-    {
-      "epoch": 5.947622690124401,
-      "grad_norm": 0.001580257317982614,
-      "learning_rate": 0.00019998255460833107,
-      "loss": 46.0,
-      "step": 36932
-    },
-    {
-      "epoch": 5.947783727203189,
-      "grad_norm": 0.00166042847558856,
-      "learning_rate": 0.0001999825536633595,
-      "loss": 46.0,
-      "step": 36933
-    },
-    {
-      "epoch": 5.947944764281976,
-      "grad_norm": 0.004260808229446411,
-      "learning_rate": 0.00019998255271836231,
-      "loss": 46.0,
-      "step": 36934
-    },
-    {
-      "epoch": 5.9481058013607635,
-      "grad_norm": 0.003028763923794031,
-      "learning_rate": 0.00019998255177333953,
-      "loss": 46.0,
-      "step": 36935
-    },
-    {
-      "epoch": 5.948266838439551,
-      "grad_norm": 0.006037257146090269,
-      "learning_rate": 0.00019998255082829113,
-      "loss": 46.0,
-      "step": 36936
-    },
-    {
-      "epoch": 5.948427875518338,
-      "grad_norm": 0.007757402490824461,
-      "learning_rate": 0.00019998254988321722,
-      "loss": 46.0,
-      "step": 36937
-    },
-    {
-      "epoch": 5.948588912597126,
-      "grad_norm": 0.0016534540336579084,
-      "learning_rate": 0.00019998254893811764,
-      "loss": 46.0,
-      "step": 36938
-    },
-    {
-      "epoch": 5.948749949675912,
-      "grad_norm": 0.02142789401113987,
-      "learning_rate": 0.0001999825479929925,
-      "loss": 46.0,
-      "step": 36939
-    },
-    {
-      "epoch": 5.948910986754701,
-      "grad_norm": 0.006063102278858423,
-      "learning_rate": 0.00019998254704784178,
-      "loss": 46.0,
-      "step": 36940
-    },
-    {
-      "epoch": 5.949072023833487,
-      "grad_norm": 0.00894453190267086,
-      "learning_rate": 0.00019998254610266547,
-      "loss": 46.0,
-      "step": 36941
-    },
-    {
-      "epoch": 5.949233060912275,
-      "grad_norm": 0.008720950223505497,
-      "learning_rate": 0.00019998254515746355,
-      "loss": 46.0,
-      "step": 36942
-    },
-    {
-      "epoch": 5.949394097991062,
-      "grad_norm": 0.0016925291856750846,
-      "learning_rate": 0.00019998254421223606,
-      "loss": 46.0,
-      "step": 36943
-    },
-    {
-      "epoch": 5.94955513506985,
-      "grad_norm": 0.003567629959434271,
-      "learning_rate": 0.00019998254326698296,
-      "loss": 46.0,
-      "step": 36944
-    },
-    {
-      "epoch": 5.949716172148637,
-      "grad_norm": 0.003626785473898053,
-      "learning_rate": 0.00019998254232170428,
-      "loss": 46.0,
-      "step": 36945
-    },
-    {
-      "epoch": 5.949877209227425,
-      "grad_norm": 0.004043857101351023,
-      "learning_rate": 0.0001999825413764,
-      "loss": 46.0,
-      "step": 36946
-    },
-    {
-      "epoch": 5.950038246306212,
-      "grad_norm": 0.005927152466028929,
-      "learning_rate": 0.00019998254043107015,
-      "loss": 46.0,
-      "step": 36947
-    },
-    {
-      "epoch": 5.9501992833849995,
-      "grad_norm": 0.010291928425431252,
-      "learning_rate": 0.0001999825394857147,
-      "loss": 46.0,
-      "step": 36948
-    },
-    {
-      "epoch": 5.950360320463787,
-      "grad_norm": 0.0007531038136221468,
-      "learning_rate": 0.00019998253854033364,
-      "loss": 46.0,
-      "step": 36949
-    },
-    {
-      "epoch": 5.950521357542574,
-      "grad_norm": 0.00466937618330121,
-      "learning_rate": 0.00019998253759492701,
-      "loss": 46.0,
-      "step": 36950
-    },
-    {
-      "epoch": 5.950682394621362,
-      "grad_norm": 0.008026626892387867,
-      "learning_rate": 0.0001999825366494948,
-      "loss": 46.0,
-      "step": 36951
-    },
-    {
-      "epoch": 5.950843431700149,
-      "grad_norm": 0.003214562311768532,
-      "learning_rate": 0.00019998253570403698,
-      "loss": 46.0,
-      "step": 36952
-    },
-    {
-      "epoch": 5.951004468778937,
-      "grad_norm": 0.001777904573827982,
-      "learning_rate": 0.0001999825347585536,
-      "loss": 46.0,
-      "step": 36953
-    },
-    {
-      "epoch": 5.951165505857723,
-      "grad_norm": 0.018588803708553314,
-      "learning_rate": 0.0001999825338130446,
-      "loss": 46.0,
-      "step": 36954
-    },
-    {
-      "epoch": 5.951326542936511,
-      "grad_norm": 0.00618468364700675,
-      "learning_rate": 0.00019998253286751002,
-      "loss": 46.0,
-      "step": 36955
-    },
-    {
-      "epoch": 5.951487580015298,
-      "grad_norm": 0.002803102135658264,
-      "learning_rate": 0.00019998253192194984,
-      "loss": 46.0,
-      "step": 36956
-    },
-    {
-      "epoch": 5.951648617094086,
-      "grad_norm": 0.017444560304284096,
-      "learning_rate": 0.00019998253097636406,
-      "loss": 46.0,
-      "step": 36957
-    },
-    {
-      "epoch": 5.951809654172873,
-      "grad_norm": 0.002624739659950137,
-      "learning_rate": 0.0001999825300307527,
-      "loss": 46.0,
-      "step": 36958
-    },
-    {
-      "epoch": 5.951970691251661,
-      "grad_norm": 0.007176046725362539,
-      "learning_rate": 0.00019998252908511578,
-      "loss": 46.0,
-      "step": 36959
-    },
-    {
-      "epoch": 5.952131728330448,
-      "grad_norm": 0.006515739485621452,
-      "learning_rate": 0.00019998252813945323,
-      "loss": 46.0,
-      "step": 36960
-    },
-    {
-      "epoch": 5.9522927654092355,
-      "grad_norm": 0.002790396334603429,
-      "learning_rate": 0.00019998252719376512,
-      "loss": 46.0,
-      "step": 36961
-    },
-    {
-      "epoch": 5.952453802488023,
-      "grad_norm": 0.0010005622170865536,
-      "learning_rate": 0.00019998252624805137,
-      "loss": 46.0,
-      "step": 36962
-    },
-    {
-      "epoch": 5.95261483956681,
-      "grad_norm": 0.0025750664062798023,
-      "learning_rate": 0.00019998252530231206,
-      "loss": 46.0,
-      "step": 36963
-    },
-    {
-      "epoch": 5.952775876645598,
-      "grad_norm": 0.0009480998269282281,
-      "learning_rate": 0.0001999825243565472,
-      "loss": 46.0,
-      "step": 36964
-    },
-    {
-      "epoch": 5.952936913724385,
-      "grad_norm": 0.02478196658194065,
-      "learning_rate": 0.00019998252341075668,
-      "loss": 46.0,
-      "step": 36965
-    },
-    {
-      "epoch": 5.953097950803173,
-      "grad_norm": 0.002616604557260871,
-      "learning_rate": 0.0001999825224649406,
-      "loss": 46.0,
-      "step": 36966
-    },
-    {
-      "epoch": 5.95325898788196,
-      "grad_norm": 0.0112298633903265,
-      "learning_rate": 0.00019998252151909895,
-      "loss": 46.0,
-      "step": 36967
-    },
-    {
-      "epoch": 5.953420024960748,
-      "grad_norm": 0.004851947538554668,
-      "learning_rate": 0.00019998252057323168,
-      "loss": 46.0,
-      "step": 36968
-    },
-    {
-      "epoch": 5.953581062039534,
-      "grad_norm": 0.00239943852648139,
-      "learning_rate": 0.00019998251962733882,
-      "loss": 46.0,
-      "step": 36969
-    },
-    {
-      "epoch": 5.953742099118322,
-      "grad_norm": 0.007052809931337833,
-      "learning_rate": 0.0001999825186814204,
-      "loss": 46.0,
-      "step": 36970
-    },
-    {
-      "epoch": 5.953903136197109,
-      "grad_norm": 0.008446740917861462,
-      "learning_rate": 0.00019998251773547637,
-      "loss": 46.0,
-      "step": 36971
-    },
-    {
-      "epoch": 5.9540641732758965,
-      "grad_norm": 0.004200976341962814,
-      "learning_rate": 0.00019998251678950672,
-      "loss": 46.0,
-      "step": 36972
-    },
-    {
-      "epoch": 5.954225210354684,
-      "grad_norm": 0.024789215996861458,
-      "learning_rate": 0.00019998251584351154,
-      "loss": 46.0,
-      "step": 36973
-    },
-    {
-      "epoch": 5.9543862474334714,
-      "grad_norm": 0.006637731567025185,
-      "learning_rate": 0.00019998251489749072,
-      "loss": 46.0,
-      "step": 36974
-    },
-    {
-      "epoch": 5.954547284512259,
-      "grad_norm": 0.032918959856033325,
-      "learning_rate": 0.0001999825139514443,
-      "loss": 46.0,
-      "step": 36975
-    },
-    {
-      "epoch": 5.954708321591046,
-      "grad_norm": 0.00895368866622448,
-      "learning_rate": 0.00019998251300537234,
-      "loss": 46.0,
-      "step": 36976
-    },
-    {
-      "epoch": 5.954869358669834,
-      "grad_norm": 0.0028805043548345566,
-      "learning_rate": 0.00019998251205927478,
-      "loss": 46.0,
-      "step": 36977
-    },
-    {
-      "epoch": 5.955030395748621,
-      "grad_norm": 0.007396654691547155,
-      "learning_rate": 0.0001999825111131516,
-      "loss": 46.0,
-      "step": 36978
-    },
-    {
-      "epoch": 5.955191432827409,
-      "grad_norm": 0.0033771449234336615,
-      "learning_rate": 0.00019998251016700285,
-      "loss": 46.0,
-      "step": 36979
-    },
-    {
-      "epoch": 5.955352469906196,
-      "grad_norm": 0.0005558221018873155,
-      "learning_rate": 0.0001999825092208285,
-      "loss": 46.0,
-      "step": 36980
-    },
-    {
-      "epoch": 5.955513506984984,
-      "grad_norm": 0.00286436197347939,
-      "learning_rate": 0.0001999825082746286,
-      "loss": 46.0,
-      "step": 36981
-    },
-    {
-      "epoch": 5.955674544063771,
-      "grad_norm": 0.007379957474768162,
-      "learning_rate": 0.00019998250732840305,
-      "loss": 46.0,
-      "step": 36982
-    },
-    {
-      "epoch": 5.9558355811425585,
-      "grad_norm": 0.006225490011274815,
-      "learning_rate": 0.00019998250638215192,
-      "loss": 46.0,
-      "step": 36983
-    },
-    {
-      "epoch": 5.955996618221345,
-      "grad_norm": 0.0007428998360410333,
-      "learning_rate": 0.00019998250543587523,
-      "loss": 46.0,
-      "step": 36984
-    },
-    {
-      "epoch": 5.9561576553001325,
-      "grad_norm": 0.027918454259634018,
-      "learning_rate": 0.00019998250448957294,
-      "loss": 46.0,
-      "step": 36985
-    },
-    {
-      "epoch": 5.95631869237892,
-      "grad_norm": 0.0026623853482306004,
-      "learning_rate": 0.00019998250354324505,
-      "loss": 46.0,
-      "step": 36986
-    },
-    {
-      "epoch": 5.956479729457707,
-      "grad_norm": 0.007374635897576809,
-      "learning_rate": 0.00019998250259689157,
-      "loss": 46.0,
-      "step": 36987
-    },
-    {
-      "epoch": 5.956640766536495,
-      "grad_norm": 0.002996884984895587,
-      "learning_rate": 0.00019998250165051252,
-      "loss": 46.0,
-      "step": 36988
-    },
-    {
-      "epoch": 5.956801803615282,
-      "grad_norm": 0.004277043975889683,
-      "learning_rate": 0.00019998250070410784,
-      "loss": 46.0,
-      "step": 36989
-    },
-    {
-      "epoch": 5.95696284069407,
-      "grad_norm": 0.016431516036391258,
-      "learning_rate": 0.0001999824997576776,
-      "loss": 46.0,
-      "step": 36990
-    },
-    {
-      "epoch": 5.957123877772857,
-      "grad_norm": 0.0027599839959293604,
-      "learning_rate": 0.00019998249881122176,
-      "loss": 46.0,
-      "step": 36991
-    },
-    {
-      "epoch": 5.957284914851645,
-      "grad_norm": 0.006952356547117233,
-      "learning_rate": 0.00019998249786474034,
-      "loss": 46.0,
-      "step": 36992
-    },
-    {
-      "epoch": 5.957445951930432,
-      "grad_norm": 0.0024024515878409147,
-      "learning_rate": 0.00019998249691823328,
-      "loss": 46.0,
-      "step": 36993
-    },
-    {
-      "epoch": 5.95760698900922,
-      "grad_norm": 0.01071472093462944,
-      "learning_rate": 0.00019998249597170072,
-      "loss": 46.0,
-      "step": 36994
-    },
-    {
-      "epoch": 5.957768026088007,
-      "grad_norm": 0.007149725221097469,
-      "learning_rate": 0.0001999824950251425,
-      "loss": 46.0,
-      "step": 36995
-    },
-    {
-      "epoch": 5.9579290631667945,
-      "grad_norm": 0.00957890972495079,
-      "learning_rate": 0.00019998249407855872,
-      "loss": 46.0,
-      "step": 36996
-    },
-    {
-      "epoch": 5.958090100245581,
-      "grad_norm": 0.001530267996713519,
-      "learning_rate": 0.00019998249313194933,
-      "loss": 46.0,
-      "step": 36997
-    },
-    {
-      "epoch": 5.958251137324369,
-      "grad_norm": 0.0027556815184652805,
-      "learning_rate": 0.00019998249218531434,
-      "loss": 46.0,
-      "step": 36998
-    },
-    {
-      "epoch": 5.958412174403156,
-      "grad_norm": 0.0062414417043328285,
-      "learning_rate": 0.0001999824912386538,
-      "loss": 46.0,
-      "step": 36999
-    },
-    {
-      "epoch": 5.958573211481943,
-      "grad_norm": 0.007867907173931599,
-      "learning_rate": 0.00019998249029196764,
-      "loss": 46.0,
-      "step": 37000
-    },
-    {
-      "epoch": 5.958734248560731,
-      "grad_norm": 0.0024836256634444,
-      "learning_rate": 0.0001999824893452559,
-      "loss": 46.0,
-      "step": 37001
-    },
-    {
-      "epoch": 5.958895285639518,
-      "grad_norm": 0.0028169762808829546,
-      "learning_rate": 0.0001999824883985186,
-      "loss": 46.0,
-      "step": 37002
-    },
-    {
-      "epoch": 5.959056322718306,
-      "grad_norm": 0.004039568826556206,
-      "learning_rate": 0.00019998248745175566,
-      "loss": 46.0,
-      "step": 37003
-    },
-    {
-      "epoch": 5.959217359797093,
-      "grad_norm": 0.0026104410644620657,
-      "learning_rate": 0.00019998248650496714,
-      "loss": 46.0,
-      "step": 37004
-    },
-    {
-      "epoch": 5.959378396875881,
-      "grad_norm": 0.00978656392544508,
-      "learning_rate": 0.00019998248555815304,
-      "loss": 46.0,
-      "step": 37005
-    },
-    {
-      "epoch": 5.959539433954668,
-      "grad_norm": 0.0026953667402267456,
-      "learning_rate": 0.00019998248461131332,
-      "loss": 46.0,
-      "step": 37006
-    },
-    {
-      "epoch": 5.959700471033456,
-      "grad_norm": 0.00448788283392787,
-      "learning_rate": 0.00019998248366444806,
-      "loss": 46.0,
-      "step": 37007
-    },
-    {
-      "epoch": 5.959861508112243,
-      "grad_norm": 0.013963552191853523,
-      "learning_rate": 0.00019998248271755717,
-      "loss": 46.0,
-      "step": 37008
-    },
-    {
-      "epoch": 5.9600225451910305,
-      "grad_norm": 0.001601981115527451,
-      "learning_rate": 0.00019998248177064069,
-      "loss": 46.0,
-      "step": 37009
-    },
-    {
-      "epoch": 5.960183582269818,
-      "grad_norm": 0.003871679538860917,
-      "learning_rate": 0.00019998248082369867,
-      "loss": 46.0,
-      "step": 37010
-    },
-    {
-      "epoch": 5.960344619348605,
-      "grad_norm": 0.004306553862988949,
-      "learning_rate": 0.000199982479876731,
-      "loss": 46.0,
-      "step": 37011
-    },
-    {
-      "epoch": 5.960505656427392,
-      "grad_norm": 0.00464608846232295,
-      "learning_rate": 0.00019998247892973777,
-      "loss": 46.0,
-      "step": 37012
-    },
-    {
-      "epoch": 5.96066669350618,
-      "grad_norm": 0.009025861509144306,
-      "learning_rate": 0.00019998247798271894,
-      "loss": 46.0,
-      "step": 37013
-    },
-    {
-      "epoch": 5.960827730584967,
-      "grad_norm": 0.0028407550416886806,
-      "learning_rate": 0.00019998247703567452,
-      "loss": 46.0,
-      "step": 37014
-    },
-    {
-      "epoch": 5.960988767663754,
-      "grad_norm": 0.020244669169187546,
-      "learning_rate": 0.00019998247608860452,
-      "loss": 46.0,
-      "step": 37015
-    },
-    {
-      "epoch": 5.961149804742542,
-      "grad_norm": 0.004608507733792067,
-      "learning_rate": 0.00019998247514150893,
-      "loss": 46.0,
-      "step": 37016
-    },
-    {
-      "epoch": 5.961310841821329,
-      "grad_norm": 0.001804869738407433,
-      "learning_rate": 0.00019998247419438775,
-      "loss": 46.0,
-      "step": 37017
-    },
-    {
-      "epoch": 5.961471878900117,
-      "grad_norm": 0.003487489651888609,
-      "learning_rate": 0.00019998247324724095,
-      "loss": 46.0,
-      "step": 37018
-    },
-    {
-      "epoch": 5.961632915978904,
-      "grad_norm": 0.0036999143194407225,
-      "learning_rate": 0.0001999824723000686,
-      "loss": 46.0,
-      "step": 37019
-    },
-    {
-      "epoch": 5.9617939530576916,
-      "grad_norm": 0.001381033449433744,
-      "learning_rate": 0.00019998247135287063,
-      "loss": 46.0,
-      "step": 37020
-    },
-    {
-      "epoch": 5.961954990136479,
-      "grad_norm": 0.006018578540533781,
-      "learning_rate": 0.00019998247040564708,
-      "loss": 46.0,
-      "step": 37021
-    },
-    {
-      "epoch": 5.9621160272152665,
-      "grad_norm": 0.027820903807878494,
-      "learning_rate": 0.00019998246945839793,
-      "loss": 46.0,
-      "step": 37022
-    },
-    {
-      "epoch": 5.962277064294054,
-      "grad_norm": 0.002910071285441518,
-      "learning_rate": 0.0001999824685111232,
-      "loss": 46.0,
-      "step": 37023
-    },
-    {
-      "epoch": 5.962438101372841,
-      "grad_norm": 0.010540717281401157,
-      "learning_rate": 0.0001999824675638229,
-      "loss": 46.0,
-      "step": 37024
-    },
-    {
-      "epoch": 5.962599138451629,
-      "grad_norm": 0.017062803730368614,
-      "learning_rate": 0.00019998246661649698,
-      "loss": 46.0,
-      "step": 37025
-    },
-    {
-      "epoch": 5.962760175530416,
-      "grad_norm": 0.002759263152256608,
-      "learning_rate": 0.00019998246566914547,
-      "loss": 46.0,
-      "step": 37026
-    },
-    {
-      "epoch": 5.962921212609203,
-      "grad_norm": 0.0048307436518371105,
-      "learning_rate": 0.0001999824647217684,
-      "loss": 46.0,
-      "step": 37027
-    },
-    {
-      "epoch": 5.96308224968799,
-      "grad_norm": 0.004624025896191597,
-      "learning_rate": 0.0001999824637743657,
-      "loss": 46.0,
-      "step": 37028
-    },
-    {
-      "epoch": 5.963243286766778,
-      "grad_norm": 0.005874814465641975,
-      "learning_rate": 0.00019998246282693742,
-      "loss": 46.0,
-      "step": 37029
-    },
-    {
-      "epoch": 5.963404323845565,
-      "grad_norm": 0.004562532529234886,
-      "learning_rate": 0.00019998246187948355,
-      "loss": 46.0,
-      "step": 37030
-    },
-    {
-      "epoch": 5.963565360924353,
-      "grad_norm": 0.004294287413358688,
-      "learning_rate": 0.0001999824609320041,
-      "loss": 46.0,
-      "step": 37031
-    },
-    {
-      "epoch": 5.96372639800314,
-      "grad_norm": 0.0021024267189204693,
-      "learning_rate": 0.00019998245998449906,
-      "loss": 46.0,
-      "step": 37032
-    },
-    {
-      "epoch": 5.9638874350819275,
-      "grad_norm": 0.01449536345899105,
-      "learning_rate": 0.00019998245903696843,
-      "loss": 46.0,
-      "step": 37033
-    },
-    {
-      "epoch": 5.964048472160715,
-      "grad_norm": 0.009961480274796486,
-      "learning_rate": 0.0001999824580894122,
-      "loss": 46.0,
-      "step": 37034
-    },
-    {
-      "epoch": 5.964209509239502,
-      "grad_norm": 0.005282219965010881,
-      "learning_rate": 0.00019998245714183038,
-      "loss": 46.0,
-      "step": 37035
-    },
-    {
-      "epoch": 5.96437054631829,
-      "grad_norm": 0.00541432062163949,
-      "learning_rate": 0.00019998245619422297,
-      "loss": 46.0,
-      "step": 37036
-    },
-    {
-      "epoch": 5.964531583397077,
-      "grad_norm": 0.010922852903604507,
-      "learning_rate": 0.00019998245524658996,
-      "loss": 46.0,
-      "step": 37037
-    },
-    {
-      "epoch": 5.964692620475865,
-      "grad_norm": 0.0033942132722586393,
-      "learning_rate": 0.0001999824542989314,
-      "loss": 46.0,
-      "step": 37038
-    },
-    {
-      "epoch": 5.964853657554652,
-      "grad_norm": 0.007351012900471687,
-      "learning_rate": 0.00019998245335124722,
-      "loss": 46.0,
-      "step": 37039
-    },
-    {
-      "epoch": 5.96501469463344,
-      "grad_norm": 0.004245967138558626,
-      "learning_rate": 0.00019998245240353745,
-      "loss": 46.0,
-      "step": 37040
-    },
-    {
-      "epoch": 5.965175731712227,
-      "grad_norm": 0.01678537391126156,
-      "learning_rate": 0.0001999824514558021,
-      "loss": 46.0,
-      "step": 37041
-    },
-    {
-      "epoch": 5.965336768791014,
-      "grad_norm": 0.006228059530258179,
-      "learning_rate": 0.00019998245050804113,
-      "loss": 46.0,
-      "step": 37042
-    },
-    {
-      "epoch": 5.965497805869801,
-      "grad_norm": 0.00493715750053525,
-      "learning_rate": 0.0001999824495602546,
-      "loss": 46.0,
-      "step": 37043
-    },
-    {
-      "epoch": 5.965658842948589,
-      "grad_norm": 0.003997992258518934,
-      "learning_rate": 0.00019998244861244247,
-      "loss": 46.0,
-      "step": 37044
-    },
-    {
-      "epoch": 5.965819880027376,
-      "grad_norm": 0.0010850883554667234,
-      "learning_rate": 0.00019998244766460474,
-      "loss": 46.0,
-      "step": 37045
-    },
-    {
-      "epoch": 5.9659809171061635,
-      "grad_norm": 0.019312787801027298,
-      "learning_rate": 0.00019998244671674145,
-      "loss": 46.0,
-      "step": 37046
-    },
-    {
-      "epoch": 5.966141954184951,
-      "grad_norm": 0.008613995276391506,
-      "learning_rate": 0.00019998244576885255,
-      "loss": 46.0,
-      "step": 37047
-    },
-    {
-      "epoch": 5.966302991263738,
-      "grad_norm": 0.005646351259201765,
-      "learning_rate": 0.00019998244482093806,
-      "loss": 46.0,
-      "step": 37048
-    },
-    {
-      "epoch": 5.966464028342526,
-      "grad_norm": 0.00948098860681057,
-      "learning_rate": 0.00019998244387299798,
-      "loss": 46.0,
-      "step": 37049
-    },
-    {
-      "epoch": 5.966625065421313,
-      "grad_norm": 0.0033022197894752026,
-      "learning_rate": 0.0001999824429250323,
-      "loss": 46.0,
-      "step": 37050
-    },
-    {
-      "epoch": 5.966786102500101,
-      "grad_norm": 0.005580211989581585,
-      "learning_rate": 0.00019998244197704104,
-      "loss": 46.0,
-      "step": 37051
-    },
-    {
-      "epoch": 5.966947139578888,
-      "grad_norm": 0.0010011555859819055,
-      "learning_rate": 0.00019998244102902417,
-      "loss": 46.0,
-      "step": 37052
-    },
-    {
-      "epoch": 5.967108176657676,
-      "grad_norm": 0.0023300573229789734,
-      "learning_rate": 0.00019998244008098174,
-      "loss": 46.0,
-      "step": 37053
-    },
-    {
-      "epoch": 5.967269213736463,
-      "grad_norm": 0.007257194723933935,
-      "learning_rate": 0.0001999824391329137,
-      "loss": 46.0,
-      "step": 37054
-    },
-    {
-      "epoch": 5.967430250815251,
-      "grad_norm": 0.002618317026644945,
-      "learning_rate": 0.0001999824381848201,
-      "loss": 46.0,
-      "step": 37055
-    },
-    {
-      "epoch": 5.967591287894038,
-      "grad_norm": 0.002119618933647871,
-      "learning_rate": 0.00019998243723670086,
-      "loss": 46.0,
-      "step": 37056
-    },
-    {
-      "epoch": 5.967752324972825,
-      "grad_norm": 0.009573266841471195,
-      "learning_rate": 0.00019998243628855606,
-      "loss": 46.0,
-      "step": 37057
-    },
-    {
-      "epoch": 5.967913362051612,
-      "grad_norm": 0.020647570490837097,
-      "learning_rate": 0.00019998243534038567,
-      "loss": 46.0,
-      "step": 37058
-    },
-    {
-      "epoch": 5.9680743991303995,
-      "grad_norm": 0.008657350204885006,
-      "learning_rate": 0.0001999824343921897,
-      "loss": 46.0,
-      "step": 37059
-    },
-    {
-      "epoch": 5.968235436209187,
-      "grad_norm": 0.0030234032310545444,
-      "learning_rate": 0.00019998243344396813,
-      "loss": 46.0,
-      "step": 37060
-    },
-    {
-      "epoch": 5.968396473287974,
-      "grad_norm": 0.00368187646381557,
-      "learning_rate": 0.00019998243249572096,
-      "loss": 46.0,
-      "step": 37061
-    },
-    {
-      "epoch": 5.968557510366762,
-      "grad_norm": 0.0033871724735945463,
-      "learning_rate": 0.0001999824315474482,
-      "loss": 46.0,
-      "step": 37062
-    },
-    {
-      "epoch": 5.968718547445549,
-      "grad_norm": 0.007835880853235722,
-      "learning_rate": 0.00019998243059914984,
-      "loss": 46.0,
-      "step": 37063
-    },
-    {
-      "epoch": 5.968879584524337,
-      "grad_norm": 0.008382249623537064,
-      "learning_rate": 0.00019998242965082593,
-      "loss": 46.0,
-      "step": 37064
-    },
-    {
-      "epoch": 5.969040621603124,
-      "grad_norm": 0.0015771297039464116,
-      "learning_rate": 0.00019998242870247637,
-      "loss": 46.0,
-      "step": 37065
-    },
-    {
-      "epoch": 5.969201658681912,
-      "grad_norm": 0.001699367188848555,
-      "learning_rate": 0.0001999824277541013,
-      "loss": 46.0,
-      "step": 37066
-    },
-    {
-      "epoch": 5.969362695760699,
-      "grad_norm": 0.0034891997929662466,
-      "learning_rate": 0.00019998242680570056,
-      "loss": 46.0,
-      "step": 37067
-    },
-    {
-      "epoch": 5.969523732839487,
-      "grad_norm": 0.0017445620615035295,
-      "learning_rate": 0.00019998242585727427,
-      "loss": 46.0,
-      "step": 37068
-    },
-    {
-      "epoch": 5.969684769918274,
-      "grad_norm": 0.014812956564128399,
-      "learning_rate": 0.00019998242490882237,
-      "loss": 46.0,
-      "step": 37069
-    },
-    {
-      "epoch": 5.969845806997061,
-      "grad_norm": 0.011132532730698586,
-      "learning_rate": 0.0001999824239603449,
-      "loss": 46.0,
-      "step": 37070
-    },
-    {
-      "epoch": 5.970006844075849,
-      "grad_norm": 0.009903768077492714,
-      "learning_rate": 0.00019998242301184183,
-      "loss": 46.0,
-      "step": 37071
-    },
-    {
-      "epoch": 5.9701678811546355,
-      "grad_norm": 0.002271889941766858,
-      "learning_rate": 0.00019998242206331317,
-      "loss": 46.0,
-      "step": 37072
-    },
-    {
-      "epoch": 5.970328918233423,
-      "grad_norm": 0.009553348645567894,
-      "learning_rate": 0.00019998242111475892,
-      "loss": 46.0,
-      "step": 37073
-    },
-    {
-      "epoch": 5.97048995531221,
-      "grad_norm": 0.020357292145490646,
-      "learning_rate": 0.00019998242016617908,
-      "loss": 46.0,
-      "step": 37074
-    },
-    {
-      "epoch": 5.970650992390998,
-      "grad_norm": 0.0046440488658845425,
-      "learning_rate": 0.00019998241921757366,
-      "loss": 46.0,
-      "step": 37075
-    },
-    {
-      "epoch": 5.970812029469785,
-      "grad_norm": 0.006508314982056618,
-      "learning_rate": 0.00019998241826894262,
-      "loss": 46.0,
-      "step": 37076
-    },
-    {
-      "epoch": 5.970973066548573,
-      "grad_norm": 0.0016414884012192488,
-      "learning_rate": 0.00019998241732028602,
-      "loss": 46.0,
-      "step": 37077
-    },
-    {
-      "epoch": 5.97113410362736,
-      "grad_norm": 0.0027092830277979374,
-      "learning_rate": 0.0001999824163716038,
-      "loss": 46.0,
-      "step": 37078
-    },
-    {
-      "epoch": 5.971295140706148,
-      "grad_norm": 0.0013686578022316098,
-      "learning_rate": 0.00019998241542289603,
-      "loss": 46.0,
-      "step": 37079
-    },
-    {
-      "epoch": 5.971456177784935,
-      "grad_norm": 0.0020064592827111483,
-      "learning_rate": 0.00019998241447416264,
-      "loss": 46.0,
-      "step": 37080
-    },
-    {
-      "epoch": 5.9716172148637225,
-      "grad_norm": 0.0025119762867689133,
-      "learning_rate": 0.00019998241352540367,
-      "loss": 46.0,
-      "step": 37081
-    },
-    {
-      "epoch": 5.97177825194251,
-      "grad_norm": 0.006823559291660786,
-      "learning_rate": 0.0001999824125766191,
-      "loss": 46.0,
-      "step": 37082
-    },
-    {
-      "epoch": 5.9719392890212974,
-      "grad_norm": 0.004893995355814695,
-      "learning_rate": 0.00019998241162780896,
-      "loss": 46.0,
-      "step": 37083
-    },
-    {
-      "epoch": 5.972100326100085,
-      "grad_norm": 0.0027168591041117907,
-      "learning_rate": 0.00019998241067897322,
-      "loss": 46.0,
-      "step": 37084
-    },
-    {
-      "epoch": 5.9722613631788715,
-      "grad_norm": 0.003251564921811223,
-      "learning_rate": 0.00019998240973011187,
-      "loss": 46.0,
-      "step": 37085
-    },
-    {
-      "epoch": 5.97242240025766,
-      "grad_norm": 0.001785908592864871,
-      "learning_rate": 0.00019998240878122493,
-      "loss": 46.0,
-      "step": 37086
-    },
-    {
-      "epoch": 5.972583437336446,
-      "grad_norm": 0.006836011540144682,
-      "learning_rate": 0.00019998240783231243,
-      "loss": 46.0,
-      "step": 37087
-    },
-    {
-      "epoch": 5.972744474415234,
-      "grad_norm": 0.0011833381140604615,
-      "learning_rate": 0.00019998240688337432,
-      "loss": 46.0,
-      "step": 37088
-    },
-    {
-      "epoch": 5.972905511494021,
-      "grad_norm": 0.007628599647432566,
-      "learning_rate": 0.00019998240593441065,
-      "loss": 46.0,
-      "step": 37089
-    },
-    {
-      "epoch": 5.973066548572809,
-      "grad_norm": 0.0030606070067733526,
-      "learning_rate": 0.00019998240498542136,
-      "loss": 46.0,
-      "step": 37090
-    },
-    {
-      "epoch": 5.973227585651596,
-      "grad_norm": 0.002758630784228444,
-      "learning_rate": 0.00019998240403640646,
-      "loss": 46.0,
-      "step": 37091
-    },
-    {
-      "epoch": 5.973388622730384,
-      "grad_norm": 0.0008674160926602781,
-      "learning_rate": 0.000199982403087366,
-      "loss": 46.0,
-      "step": 37092
-    },
-    {
-      "epoch": 5.973549659809171,
-      "grad_norm": 0.003769145580008626,
-      "learning_rate": 0.00019998240213829992,
-      "loss": 46.0,
-      "step": 37093
-    },
-    {
-      "epoch": 5.9737106968879585,
-      "grad_norm": 0.011612356640398502,
-      "learning_rate": 0.0001999824011892083,
-      "loss": 46.0,
-      "step": 37094
-    },
-    {
-      "epoch": 5.973871733966746,
-      "grad_norm": 0.0016333921812474728,
-      "learning_rate": 0.00019998240024009104,
-      "loss": 46.0,
-      "step": 37095
-    },
-    {
-      "epoch": 5.974032771045533,
-      "grad_norm": 0.015395691618323326,
-      "learning_rate": 0.00019998239929094823,
-      "loss": 46.0,
-      "step": 37096
-    },
-    {
-      "epoch": 5.974193808124321,
-      "grad_norm": 0.005219616461545229,
-      "learning_rate": 0.0001999823983417798,
-      "loss": 46.0,
-      "step": 37097
-    },
-    {
-      "epoch": 5.974354845203108,
-      "grad_norm": 0.005606296006590128,
-      "learning_rate": 0.0001999823973925858,
-      "loss": 46.0,
-      "step": 37098
-    },
-    {
-      "epoch": 5.974515882281896,
-      "grad_norm": 0.001775104203261435,
-      "learning_rate": 0.0001999823964433662,
-      "loss": 46.0,
-      "step": 37099
-    },
-    {
-      "epoch": 5.974676919360682,
-      "grad_norm": 0.004449504427611828,
-      "learning_rate": 0.000199982395494121,
-      "loss": 46.0,
-      "step": 37100
-    },
-    {
-      "epoch": 5.974837956439471,
-      "grad_norm": 0.0008433775510638952,
-      "learning_rate": 0.00019998239454485024,
-      "loss": 46.0,
-      "step": 37101
-    },
-    {
-      "epoch": 5.974998993518257,
-      "grad_norm": 0.001981128705665469,
-      "learning_rate": 0.00019998239359555385,
-      "loss": 46.0,
-      "step": 37102
-    },
-    {
-      "epoch": 5.975160030597045,
-      "grad_norm": 0.01373693160712719,
-      "learning_rate": 0.0001999823926462319,
-      "loss": 46.0,
-      "step": 37103
-    },
-    {
-      "epoch": 5.975321067675832,
-      "grad_norm": 0.002771165454760194,
-      "learning_rate": 0.00019998239169688434,
-      "loss": 46.0,
-      "step": 37104
-    },
-    {
-      "epoch": 5.97548210475462,
-      "grad_norm": 0.004061555489897728,
-      "learning_rate": 0.00019998239074751122,
-      "loss": 46.0,
-      "step": 37105
-    },
-    {
-      "epoch": 5.975643141833407,
-      "grad_norm": 0.012783958576619625,
-      "learning_rate": 0.00019998238979811249,
-      "loss": 46.0,
-      "step": 37106
-    },
-    {
-      "epoch": 5.9758041789121945,
-      "grad_norm": 0.0038586093578487635,
-      "learning_rate": 0.00019998238884868816,
-      "loss": 46.0,
-      "step": 37107
-    },
-    {
-      "epoch": 5.975965215990982,
-      "grad_norm": 0.013749733567237854,
-      "learning_rate": 0.00019998238789923825,
-      "loss": 46.0,
-      "step": 37108
-    },
-    {
-      "epoch": 5.976126253069769,
-      "grad_norm": 0.0018850447377189994,
-      "learning_rate": 0.00019998238694976276,
-      "loss": 46.0,
-      "step": 37109
-    },
-    {
-      "epoch": 5.976287290148557,
-      "grad_norm": 0.005399515386670828,
-      "learning_rate": 0.00019998238600026164,
-      "loss": 46.0,
-      "step": 37110
-    },
-    {
-      "epoch": 5.976448327227344,
-      "grad_norm": 0.004808348137885332,
-      "learning_rate": 0.00019998238505073497,
-      "loss": 46.0,
-      "step": 37111
-    },
-    {
-      "epoch": 5.976609364306132,
-      "grad_norm": 0.0017414601752534509,
-      "learning_rate": 0.0001999823841011827,
-      "loss": 46.0,
-      "step": 37112
-    },
-    {
-      "epoch": 5.976770401384919,
-      "grad_norm": 0.010008828714489937,
-      "learning_rate": 0.00019998238315160481,
-      "loss": 46.0,
-      "step": 37113
-    },
-    {
-      "epoch": 5.976931438463707,
-      "grad_norm": 0.008640378713607788,
-      "learning_rate": 0.00019998238220200135,
-      "loss": 46.0,
-      "step": 37114
-    },
-    {
-      "epoch": 5.977092475542493,
-      "grad_norm": 0.0008898033411242068,
-      "learning_rate": 0.0001999823812523723,
-      "loss": 46.0,
-      "step": 37115
-    },
-    {
-      "epoch": 5.977253512621281,
-      "grad_norm": 0.005171169526875019,
-      "learning_rate": 0.0001999823803027177,
-      "loss": 46.0,
-      "step": 37116
-    },
-    {
-      "epoch": 5.977414549700068,
-      "grad_norm": 0.004170972388237715,
-      "learning_rate": 0.00019998237935303745,
-      "loss": 46.0,
-      "step": 37117
-    },
-    {
-      "epoch": 5.977575586778856,
-      "grad_norm": 0.0012544617056846619,
-      "learning_rate": 0.00019998237840333164,
-      "loss": 46.0,
-      "step": 37118
-    },
-    {
-      "epoch": 5.977736623857643,
-      "grad_norm": 0.0035998127423226833,
-      "learning_rate": 0.00019998237745360025,
-      "loss": 46.0,
-      "step": 37119
-    },
-    {
-      "epoch": 5.9778976609364305,
-      "grad_norm": 0.003074868582189083,
-      "learning_rate": 0.00019998237650384324,
-      "loss": 46.0,
-      "step": 37120
-    },
-    {
-      "epoch": 5.978058698015218,
-      "grad_norm": 0.0025523509830236435,
-      "learning_rate": 0.00019998237555406067,
-      "loss": 46.0,
-      "step": 37121
-    },
-    {
-      "epoch": 5.978219735094005,
-      "grad_norm": 0.005547038745135069,
-      "learning_rate": 0.00019998237460425248,
-      "loss": 46.0,
-      "step": 37122
-    },
-    {
-      "epoch": 5.978380772172793,
-      "grad_norm": 0.0020832493901252747,
-      "learning_rate": 0.0001999823736544187,
-      "loss": 46.0,
-      "step": 37123
-    },
-    {
-      "epoch": 5.97854180925158,
-      "grad_norm": 0.005689549725502729,
-      "learning_rate": 0.00019998237270455935,
-      "loss": 46.0,
-      "step": 37124
-    },
-    {
-      "epoch": 5.978702846330368,
-      "grad_norm": 0.0017920630052685738,
-      "learning_rate": 0.0001999823717546744,
-      "loss": 46.0,
-      "step": 37125
-    },
-    {
-      "epoch": 5.978863883409155,
-      "grad_norm": 0.017536768689751625,
-      "learning_rate": 0.00019998237080476384,
-      "loss": 46.0,
-      "step": 37126
-    },
-    {
-      "epoch": 5.979024920487943,
-      "grad_norm": 0.005700399167835712,
-      "learning_rate": 0.00019998236985482772,
-      "loss": 46.0,
-      "step": 37127
-    },
-    {
-      "epoch": 5.97918595756673,
-      "grad_norm": 0.007136261556297541,
-      "learning_rate": 0.00019998236890486602,
-      "loss": 46.0,
-      "step": 37128
-    },
-    {
-      "epoch": 5.9793469946455176,
-      "grad_norm": 0.0017615235410630703,
-      "learning_rate": 0.0001999823679548787,
-      "loss": 46.0,
-      "step": 37129
-    },
-    {
-      "epoch": 5.979508031724304,
-      "grad_norm": 0.00639650272205472,
-      "learning_rate": 0.0001999823670048658,
-      "loss": 46.0,
-      "step": 37130
-    },
-    {
-      "epoch": 5.979669068803092,
-      "grad_norm": 0.003200503997504711,
-      "learning_rate": 0.00019998236605482732,
-      "loss": 46.0,
-      "step": 37131
-    },
-    {
-      "epoch": 5.979830105881879,
-      "grad_norm": 0.002541875932365656,
-      "learning_rate": 0.00019998236510476326,
-      "loss": 46.0,
-      "step": 37132
-    },
-    {
-      "epoch": 5.9799911429606665,
-      "grad_norm": 0.0013374063419178128,
-      "learning_rate": 0.00019998236415467356,
-      "loss": 46.0,
-      "step": 37133
-    },
-    {
-      "epoch": 5.980152180039454,
-      "grad_norm": 0.0021549889352172613,
-      "learning_rate": 0.0001999823632045583,
-      "loss": 46.0,
-      "step": 37134
-    },
-    {
-      "epoch": 5.980313217118241,
-      "grad_norm": 0.0028978127520531416,
-      "learning_rate": 0.00019998236225441743,
-      "loss": 46.0,
-      "step": 37135
-    },
-    {
-      "epoch": 5.980474254197029,
-      "grad_norm": 0.003962274640798569,
-      "learning_rate": 0.000199982361304251,
-      "loss": 46.0,
-      "step": 37136
-    },
-    {
-      "epoch": 5.980635291275816,
-      "grad_norm": 0.0036189004313200712,
-      "learning_rate": 0.00019998236035405898,
-      "loss": 46.0,
-      "step": 37137
-    },
-    {
-      "epoch": 5.980796328354604,
-      "grad_norm": 0.006551253143697977,
-      "learning_rate": 0.00019998235940384135,
-      "loss": 46.0,
-      "step": 37138
-    },
-    {
-      "epoch": 5.980957365433391,
-      "grad_norm": 0.003585867816582322,
-      "learning_rate": 0.00019998235845359813,
-      "loss": 46.0,
-      "step": 37139
-    },
-    {
-      "epoch": 5.981118402512179,
-      "grad_norm": 0.006087068468332291,
-      "learning_rate": 0.00019998235750332932,
-      "loss": 46.0,
-      "step": 37140
-    },
-    {
-      "epoch": 5.981279439590966,
-      "grad_norm": 0.005042713135480881,
-      "learning_rate": 0.00019998235655303495,
-      "loss": 46.0,
-      "step": 37141
-    },
-    {
-      "epoch": 5.9814404766697535,
-      "grad_norm": 0.00267610396258533,
-      "learning_rate": 0.00019998235560271494,
-      "loss": 46.0,
-      "step": 37142
-    },
-    {
-      "epoch": 5.981601513748541,
-      "grad_norm": 0.003711590077728033,
-      "learning_rate": 0.00019998235465236937,
-      "loss": 46.0,
-      "step": 37143
-    },
-    {
-      "epoch": 5.981762550827328,
-      "grad_norm": 0.005382280331104994,
-      "learning_rate": 0.00019998235370199822,
-      "loss": 46.0,
-      "step": 37144
-    },
-    {
-      "epoch": 5.981923587906115,
-      "grad_norm": 0.0033533256500959396,
-      "learning_rate": 0.00019998235275160147,
-      "loss": 46.0,
-      "step": 37145
-    },
-    {
-      "epoch": 5.9820846249849025,
-      "grad_norm": 0.002165958285331726,
-      "learning_rate": 0.00019998235180117911,
-      "loss": 46.0,
-      "step": 37146
-    },
-    {
-      "epoch": 5.98224566206369,
-      "grad_norm": 0.0034255855716764927,
-      "learning_rate": 0.00019998235085073117,
-      "loss": 46.0,
-      "step": 37147
-    },
-    {
-      "epoch": 5.982406699142477,
-      "grad_norm": 0.0010856261942535639,
-      "learning_rate": 0.00019998234990025764,
-      "loss": 46.0,
-      "step": 37148
-    },
-    {
-      "epoch": 5.982567736221265,
-      "grad_norm": 0.003644917393103242,
-      "learning_rate": 0.00019998234894975854,
-      "loss": 46.0,
-      "step": 37149
-    },
-    {
-      "epoch": 5.982728773300052,
-      "grad_norm": 0.01214093342423439,
-      "learning_rate": 0.0001999823479992338,
-      "loss": 46.0,
-      "step": 37150
-    },
-    {
-      "epoch": 5.98288981037884,
-      "grad_norm": 0.0041720070876181126,
-      "learning_rate": 0.00019998234704868352,
-      "loss": 46.0,
-      "step": 37151
-    },
-    {
-      "epoch": 5.983050847457627,
-      "grad_norm": 0.0014115235535427928,
-      "learning_rate": 0.0001999823460981076,
-      "loss": 46.0,
-      "step": 37152
-    },
-    {
-      "epoch": 5.983211884536415,
-      "grad_norm": 0.004376129247248173,
-      "learning_rate": 0.00019998234514750617,
-      "loss": 46.0,
-      "step": 37153
-    },
-    {
-      "epoch": 5.983372921615202,
-      "grad_norm": 0.002607706468552351,
-      "learning_rate": 0.00019998234419687908,
-      "loss": 46.0,
-      "step": 37154
-    },
-    {
-      "epoch": 5.9835339586939895,
-      "grad_norm": 0.011401246301829815,
-      "learning_rate": 0.00019998234324622642,
-      "loss": 46.0,
-      "step": 37155
-    },
-    {
-      "epoch": 5.983694995772777,
-      "grad_norm": 0.0013237452367320657,
-      "learning_rate": 0.00019998234229554819,
-      "loss": 46.0,
-      "step": 37156
-    },
-    {
-      "epoch": 5.983856032851564,
-      "grad_norm": 0.002852878998965025,
-      "learning_rate": 0.00019998234134484431,
-      "loss": 46.0,
-      "step": 37157
-    },
-    {
-      "epoch": 5.984017069930351,
-      "grad_norm": 0.0027569483499974012,
-      "learning_rate": 0.0001999823403941149,
-      "loss": 46.0,
-      "step": 37158
-    },
-    {
-      "epoch": 5.984178107009139,
-      "grad_norm": 0.02459539659321308,
-      "learning_rate": 0.0001999823394433599,
-      "loss": 46.0,
-      "step": 37159
-    },
-    {
-      "epoch": 5.984339144087926,
-      "grad_norm": 0.007951420731842518,
-      "learning_rate": 0.0001999823384925793,
-      "loss": 46.0,
-      "step": 37160
-    },
-    {
-      "epoch": 5.984500181166713,
-      "grad_norm": 0.0024845865555107594,
-      "learning_rate": 0.0001999823375417731,
-      "loss": 46.0,
-      "step": 37161
-    },
-    {
-      "epoch": 5.984661218245501,
-      "grad_norm": 0.007075507193803787,
-      "learning_rate": 0.0001999823365909413,
-      "loss": 46.0,
-      "step": 37162
-    },
-    {
-      "epoch": 5.984822255324288,
-      "grad_norm": 0.0025554453022778034,
-      "learning_rate": 0.0001999823356400839,
-      "loss": 46.0,
-      "step": 37163
-    },
-    {
-      "epoch": 5.984983292403076,
-      "grad_norm": 0.001774023869074881,
-      "learning_rate": 0.0001999823346892009,
-      "loss": 46.0,
-      "step": 37164
-    },
-    {
-      "epoch": 5.985144329481863,
-      "grad_norm": 0.003452279604971409,
-      "learning_rate": 0.00019998233373829237,
-      "loss": 46.0,
-      "step": 37165
-    },
-    {
-      "epoch": 5.985305366560651,
-      "grad_norm": 0.005397127475589514,
-      "learning_rate": 0.00019998233278735821,
-      "loss": 46.0,
-      "step": 37166
-    },
-    {
-      "epoch": 5.985466403639438,
-      "grad_norm": 0.004143360070884228,
-      "learning_rate": 0.00019998233183639847,
-      "loss": 46.0,
-      "step": 37167
-    },
-    {
-      "epoch": 5.9856274407182255,
-      "grad_norm": 0.0029504646081477404,
-      "learning_rate": 0.00019998233088541314,
-      "loss": 46.0,
-      "step": 37168
-    },
-    {
-      "epoch": 5.985788477797013,
-      "grad_norm": 0.003240946214646101,
-      "learning_rate": 0.00019998232993440222,
-      "loss": 46.0,
-      "step": 37169
-    },
-    {
-      "epoch": 5.9859495148758,
-      "grad_norm": 0.0009449656354263425,
-      "learning_rate": 0.0001999823289833657,
-      "loss": 46.0,
-      "step": 37170
-    },
-    {
-      "epoch": 5.986110551954588,
-      "grad_norm": 0.010652769356966019,
-      "learning_rate": 0.0001999823280323036,
-      "loss": 46.0,
-      "step": 37171
-    },
-    {
-      "epoch": 5.986271589033375,
-      "grad_norm": 0.0018262845696881413,
-      "learning_rate": 0.0001999823270812159,
-      "loss": 46.0,
-      "step": 37172
-    },
-    {
-      "epoch": 5.986432626112162,
-      "grad_norm": 0.01686927303671837,
-      "learning_rate": 0.0001999823261301026,
-      "loss": 46.0,
-      "step": 37173
-    },
-    {
-      "epoch": 5.98659366319095,
-      "grad_norm": 0.009832118637859821,
-      "learning_rate": 0.00019998232517896375,
-      "loss": 46.0,
-      "step": 37174
-    },
-    {
-      "epoch": 5.986754700269737,
-      "grad_norm": 0.0014192365342751145,
-      "learning_rate": 0.00019998232422779926,
-      "loss": 46.0,
-      "step": 37175
-    },
-    {
-      "epoch": 5.986915737348524,
-      "grad_norm": 0.005346552934497595,
-      "learning_rate": 0.0001999823232766092,
-      "loss": 46.0,
-      "step": 37176
-    },
-    {
-      "epoch": 5.987076774427312,
-      "grad_norm": 0.009062749333679676,
-      "learning_rate": 0.00019998232232539356,
-      "loss": 46.0,
-      "step": 37177
-    },
-    {
-      "epoch": 5.987237811506099,
-      "grad_norm": 0.003989992197602987,
-      "learning_rate": 0.00019998232137415233,
-      "loss": 46.0,
-      "step": 37178
-    },
-    {
-      "epoch": 5.987398848584887,
-      "grad_norm": 0.003291300730779767,
-      "learning_rate": 0.0001999823204228855,
-      "loss": 46.0,
-      "step": 37179
-    },
-    {
-      "epoch": 5.987559885663674,
-      "grad_norm": 0.001711393124423921,
-      "learning_rate": 0.00019998231947159308,
-      "loss": 46.0,
-      "step": 37180
-    },
-    {
-      "epoch": 5.9877209227424615,
-      "grad_norm": 0.0007577283540740609,
-      "learning_rate": 0.00019998231852027507,
-      "loss": 46.0,
-      "step": 37181
-    },
-    {
-      "epoch": 5.987881959821249,
-      "grad_norm": 0.0029843361116945744,
-      "learning_rate": 0.0001999823175689315,
-      "loss": 46.0,
-      "step": 37182
-    },
-    {
-      "epoch": 5.988042996900036,
-      "grad_norm": 0.007334083318710327,
-      "learning_rate": 0.00019998231661756227,
-      "loss": 46.0,
-      "step": 37183
-    },
-    {
-      "epoch": 5.988204033978824,
-      "grad_norm": 0.0019836369901895523,
-      "learning_rate": 0.0001999823156661675,
-      "loss": 46.0,
-      "step": 37184
-    },
-    {
-      "epoch": 5.988365071057611,
-      "grad_norm": 0.003181385574862361,
-      "learning_rate": 0.00019998231471474713,
-      "loss": 46.0,
-      "step": 37185
-    },
-    {
-      "epoch": 5.988526108136399,
-      "grad_norm": 0.008767525665462017,
-      "learning_rate": 0.0001999823137633012,
-      "loss": 46.0,
-      "step": 37186
-    },
-    {
-      "epoch": 5.988687145215186,
-      "grad_norm": 0.0035331561230123043,
-      "learning_rate": 0.00019998231281182963,
-      "loss": 46.0,
-      "step": 37187
-    },
-    {
-      "epoch": 5.988848182293973,
-      "grad_norm": 0.006100935861468315,
-      "learning_rate": 0.00019998231186033248,
-      "loss": 46.0,
-      "step": 37188
-    },
-    {
-      "epoch": 5.98900921937276,
-      "grad_norm": 0.005503537133336067,
-      "learning_rate": 0.00019998231090880974,
-      "loss": 46.0,
-      "step": 37189
-    },
-    {
-      "epoch": 5.989170256451548,
-      "grad_norm": 0.01521372701972723,
-      "learning_rate": 0.00019998230995726143,
-      "loss": 46.0,
-      "step": 37190
-    },
-    {
-      "epoch": 5.989331293530335,
-      "grad_norm": 0.006318739615380764,
-      "learning_rate": 0.00019998230900568752,
-      "loss": 46.0,
-      "step": 37191
-    },
-    {
-      "epoch": 5.989492330609123,
-      "grad_norm": 0.005000052507966757,
-      "learning_rate": 0.00019998230805408801,
-      "loss": 46.0,
-      "step": 37192
-    },
-    {
-      "epoch": 5.98965336768791,
-      "grad_norm": 0.004043553490191698,
-      "learning_rate": 0.00019998230710246292,
-      "loss": 46.0,
-      "step": 37193
-    },
-    {
-      "epoch": 5.9898144047666975,
-      "grad_norm": 0.010761475190520287,
-      "learning_rate": 0.00019998230615081225,
-      "loss": 46.0,
-      "step": 37194
-    },
-    {
-      "epoch": 5.989975441845485,
-      "grad_norm": 0.01122052501887083,
-      "learning_rate": 0.00019998230519913595,
-      "loss": 46.0,
-      "step": 37195
-    },
-    {
-      "epoch": 5.990136478924272,
-      "grad_norm": 0.00972563587129116,
-      "learning_rate": 0.00019998230424743413,
-      "loss": 46.0,
-      "step": 37196
-    },
-    {
-      "epoch": 5.99029751600306,
-      "grad_norm": 0.004708815831691027,
-      "learning_rate": 0.00019998230329570666,
-      "loss": 46.0,
-      "step": 37197
-    },
-    {
-      "epoch": 5.990458553081847,
-      "grad_norm": 0.001252274145372212,
-      "learning_rate": 0.0001999823023439536,
-      "loss": 46.0,
-      "step": 37198
-    },
-    {
-      "epoch": 5.990619590160635,
-      "grad_norm": 0.0019713020883500576,
-      "learning_rate": 0.00019998230139217497,
-      "loss": 46.0,
-      "step": 37199
-    },
-    {
-      "epoch": 5.990780627239422,
-      "grad_norm": 0.0017733210697770119,
-      "learning_rate": 0.00019998230044037074,
-      "loss": 46.0,
-      "step": 37200
-    },
-    {
-      "epoch": 5.99094166431821,
-      "grad_norm": 0.003067875513806939,
-      "learning_rate": 0.00019998229948854092,
-      "loss": 46.0,
-      "step": 37201
-    },
-    {
-      "epoch": 5.991102701396997,
-      "grad_norm": 0.002472830470651388,
-      "learning_rate": 0.00019998229853668552,
-      "loss": 46.0,
-      "step": 37202
-    },
-    {
-      "epoch": 5.991263738475784,
-      "grad_norm": 0.0011308375978842378,
-      "learning_rate": 0.00019998229758480453,
-      "loss": 46.0,
-      "step": 37203
-    },
-    {
-      "epoch": 5.991424775554571,
-      "grad_norm": 0.003935275599360466,
-      "learning_rate": 0.00019998229663289796,
-      "loss": 46.0,
-      "step": 37204
-    },
-    {
-      "epoch": 5.9915858126333585,
-      "grad_norm": 0.014076292514801025,
-      "learning_rate": 0.0001999822956809658,
-      "loss": 46.0,
-      "step": 37205
-    },
-    {
-      "epoch": 5.991746849712146,
-      "grad_norm": 0.004145600367337465,
-      "learning_rate": 0.00019998229472900801,
-      "loss": 46.0,
-      "step": 37206
-    },
-    {
-      "epoch": 5.9919078867909334,
-      "grad_norm": 0.0025334605015814304,
-      "learning_rate": 0.00019998229377702465,
-      "loss": 46.0,
-      "step": 37207
-    },
-    {
-      "epoch": 5.992068923869721,
-      "grad_norm": 0.0030865399166941643,
-      "learning_rate": 0.00019998229282501567,
-      "loss": 46.0,
-      "step": 37208
-    },
-    {
-      "epoch": 5.992229960948508,
-      "grad_norm": 0.004648580215871334,
-      "learning_rate": 0.00019998229187298116,
-      "loss": 46.0,
-      "step": 37209
-    },
-    {
-      "epoch": 5.992390998027296,
-      "grad_norm": 0.02252069115638733,
-      "learning_rate": 0.00019998229092092103,
-      "loss": 46.0,
-      "step": 37210
-    },
-    {
-      "epoch": 5.992552035106083,
-      "grad_norm": 0.010720000602304935,
-      "learning_rate": 0.00019998228996883532,
-      "loss": 46.0,
-      "step": 37211
-    },
-    {
-      "epoch": 5.992713072184871,
-      "grad_norm": 0.0017160896677523851,
-      "learning_rate": 0.000199982289016724,
-      "loss": 46.0,
-      "step": 37212
-    },
-    {
-      "epoch": 5.992874109263658,
-      "grad_norm": 0.02023484744131565,
-      "learning_rate": 0.0001999822880645871,
-      "loss": 46.0,
-      "step": 37213
-    },
-    {
-      "epoch": 5.993035146342446,
-      "grad_norm": 0.0030344570986926556,
-      "learning_rate": 0.00019998228711242463,
-      "loss": 46.0,
-      "step": 37214
-    },
-    {
-      "epoch": 5.993196183421233,
-      "grad_norm": 0.0024063277523964643,
-      "learning_rate": 0.00019998228616023654,
-      "loss": 46.0,
-      "step": 37215
-    },
-    {
-      "epoch": 5.9933572205000205,
-      "grad_norm": 0.0023261159658432007,
-      "learning_rate": 0.00019998228520802286,
-      "loss": 46.0,
-      "step": 37216
-    },
-    {
-      "epoch": 5.993518257578808,
-      "grad_norm": 0.007904346100986004,
-      "learning_rate": 0.0001999822842557836,
-      "loss": 46.0,
-      "step": 37217
-    },
-    {
-      "epoch": 5.9936792946575945,
-      "grad_norm": 0.006069988943636417,
-      "learning_rate": 0.00019998228330351877,
-      "loss": 46.0,
-      "step": 37218
-    },
-    {
-      "epoch": 5.993840331736382,
-      "grad_norm": 0.0035625228192657232,
-      "learning_rate": 0.0001999822823512283,
-      "loss": 46.0,
-      "step": 37219
-    },
-    {
-      "epoch": 5.994001368815169,
-      "grad_norm": 0.0026505119167268276,
-      "learning_rate": 0.00019998228139891228,
-      "loss": 46.0,
-      "step": 37220
-    },
-    {
-      "epoch": 5.994162405893957,
-      "grad_norm": 0.009775853715837002,
-      "learning_rate": 0.00019998228044657067,
-      "loss": 46.0,
-      "step": 37221
-    },
-    {
-      "epoch": 5.994323442972744,
-      "grad_norm": 0.008968714624643326,
-      "learning_rate": 0.00019998227949420344,
-      "loss": 46.0,
-      "step": 37222
-    },
-    {
-      "epoch": 5.994484480051532,
-      "grad_norm": 0.0030200222972780466,
-      "learning_rate": 0.00019998227854181065,
-      "loss": 46.0,
-      "step": 37223
-    },
-    {
-      "epoch": 5.994645517130319,
-      "grad_norm": 0.0058414386585354805,
-      "learning_rate": 0.00019998227758939225,
-      "loss": 46.0,
-      "step": 37224
-    },
-    {
-      "epoch": 5.994806554209107,
-      "grad_norm": 0.008276640437543392,
-      "learning_rate": 0.00019998227663694826,
-      "loss": 46.0,
-      "step": 37225
-    },
-    {
-      "epoch": 5.994967591287894,
-      "grad_norm": 0.019536063075065613,
-      "learning_rate": 0.0001999822756844787,
-      "loss": 46.0,
-      "step": 37226
-    },
-    {
-      "epoch": 5.995128628366682,
-      "grad_norm": 0.004572566598653793,
-      "learning_rate": 0.00019998227473198355,
-      "loss": 46.0,
-      "step": 37227
-    },
-    {
-      "epoch": 5.995289665445469,
-      "grad_norm": 0.009302785620093346,
-      "learning_rate": 0.00019998227377946278,
-      "loss": 46.0,
-      "step": 37228
-    },
-    {
-      "epoch": 5.9954507025242565,
-      "grad_norm": 0.0041602598503232,
-      "learning_rate": 0.0001999822728269164,
-      "loss": 46.0,
-      "step": 37229
-    },
-    {
-      "epoch": 5.995611739603044,
-      "grad_norm": 0.019374659284949303,
-      "learning_rate": 0.00019998227187434449,
-      "loss": 46.0,
-      "step": 37230
-    },
-    {
-      "epoch": 5.9957727766818305,
-      "grad_norm": 0.008892161771655083,
-      "learning_rate": 0.00019998227092174697,
-      "loss": 46.0,
-      "step": 37231
-    },
-    {
-      "epoch": 5.995933813760619,
-      "grad_norm": 0.006749188527464867,
-      "learning_rate": 0.00019998226996912385,
-      "loss": 46.0,
-      "step": 37232
-    },
-    {
-      "epoch": 5.996094850839405,
-      "grad_norm": 0.003060927614569664,
-      "learning_rate": 0.00019998226901647516,
-      "loss": 46.0,
-      "step": 37233
-    },
-    {
-      "epoch": 5.996255887918193,
-      "grad_norm": 0.004948275163769722,
-      "learning_rate": 0.00019998226806380086,
-      "loss": 46.0,
-      "step": 37234
-    },
-    {
-      "epoch": 5.99641692499698,
-      "grad_norm": 0.0046185594983398914,
-      "learning_rate": 0.00019998226711110098,
-      "loss": 46.0,
-      "step": 37235
-    },
-    {
-      "epoch": 5.996577962075768,
-      "grad_norm": 0.0008127544424496591,
-      "learning_rate": 0.0001999822661583755,
-      "loss": 46.0,
-      "step": 37236
-    },
-    {
-      "epoch": 5.996738999154555,
-      "grad_norm": 0.00691266218200326,
-      "learning_rate": 0.0001999822652056244,
-      "loss": 46.0,
-      "step": 37237
-    },
-    {
-      "epoch": 5.996900036233343,
-      "grad_norm": 0.0009220805368386209,
-      "learning_rate": 0.00019998226425284776,
-      "loss": 46.0,
-      "step": 37238
-    },
-    {
-      "epoch": 5.99706107331213,
-      "grad_norm": 0.012285149656236172,
-      "learning_rate": 0.00019998226330004553,
-      "loss": 46.0,
-      "step": 37239
-    },
-    {
-      "epoch": 5.997222110390918,
-      "grad_norm": 0.002424470614641905,
-      "learning_rate": 0.00019998226234721768,
-      "loss": 46.0,
-      "step": 37240
-    },
-    {
-      "epoch": 5.997383147469705,
-      "grad_norm": 0.004506948869675398,
-      "learning_rate": 0.00019998226139436427,
-      "loss": 46.0,
-      "step": 37241
-    },
-    {
-      "epoch": 5.9975441845484925,
-      "grad_norm": 0.002975679701194167,
-      "learning_rate": 0.00019998226044148524,
-      "loss": 46.0,
-      "step": 37242
-    },
-    {
-      "epoch": 5.99770522162728,
-      "grad_norm": 0.006455249153077602,
-      "learning_rate": 0.00019998225948858063,
-      "loss": 46.0,
-      "step": 37243
-    },
-    {
-      "epoch": 5.997866258706067,
-      "grad_norm": 0.0023767005186527967,
-      "learning_rate": 0.00019998225853565043,
-      "loss": 46.0,
-      "step": 37244
-    },
-    {
-      "epoch": 5.998027295784855,
-      "grad_norm": 0.003752077929675579,
-      "learning_rate": 0.00019998225758269462,
-      "loss": 46.0,
-      "step": 37245
-    },
-    {
-      "epoch": 5.998188332863641,
-      "grad_norm": 0.0031229937449097633,
-      "learning_rate": 0.00019998225662971327,
-      "loss": 46.0,
-      "step": 37246
-    },
-    {
-      "epoch": 5.99834936994243,
-      "grad_norm": 0.002404554048553109,
-      "learning_rate": 0.00019998225567670629,
-      "loss": 46.0,
-      "step": 37247
-    },
-    {
-      "epoch": 5.998510407021216,
-      "grad_norm": 0.009118358604609966,
-      "learning_rate": 0.00019998225472367374,
-      "loss": 46.0,
-      "step": 37248
-    },
-    {
-      "epoch": 5.998671444100004,
-      "grad_norm": 0.0017667836509644985,
-      "learning_rate": 0.00019998225377061558,
-      "loss": 46.0,
-      "step": 37249
-    },
-    {
-      "epoch": 5.998832481178791,
-      "grad_norm": 0.004705061204731464,
-      "learning_rate": 0.00019998225281753183,
-      "loss": 46.0,
-      "step": 37250
-    },
-    {
-      "epoch": 5.998993518257579,
-      "grad_norm": 0.001341478549875319,
-      "learning_rate": 0.0001999822518644225,
-      "loss": 46.0,
-      "step": 37251
-    },
-    {
-      "epoch": 5.999154555336366,
-      "grad_norm": 0.006101112347096205,
-      "learning_rate": 0.00019998225091128757,
-      "loss": 46.0,
-      "step": 37252
-    },
-    {
-      "epoch": 5.9993155924151536,
-      "grad_norm": 0.004955587908625603,
-      "learning_rate": 0.00019998224995812706,
-      "loss": 46.0,
-      "step": 37253
-    },
-    {
-      "epoch": 5.999476629493941,
-      "grad_norm": 0.009894069284200668,
-      "learning_rate": 0.00019998224900494096,
-      "loss": 46.0,
-      "step": 37254
-    },
-    {
-      "epoch": 5.9996376665727285,
-      "grad_norm": 0.0034555376041680574,
-      "learning_rate": 0.00019998224805172927,
-      "loss": 46.0,
-      "step": 37255
-    },
-    {
-      "epoch": 5.999798703651516,
-      "grad_norm": 0.003928289748728275,
-      "learning_rate": 0.00019998224709849197,
-      "loss": 46.0,
-      "step": 37256
-    },
-    {
-      "epoch": 5.999959740730303,
-      "grad_norm": 0.0034505431540310383,
-      "learning_rate": 0.00019998224614522911,
-      "loss": 46.0,
-      "step": 37257
-    },
-    {
-      "epoch": 5.999959740730303,
-      "eval_loss": 11.5,
-      "eval_runtime": 15.137,
-      "eval_samples_per_second": 172.756,
-      "eval_steps_per_second": 86.411,
-      "step": 37257
-    },
-    {
-      "epoch": 6.0001610370787875,
-      "grad_norm": 0.001225961372256279,
-      "learning_rate": 0.00019998224519194064,
-      "loss": 46.0,
-      "step": 37258
-    },
-    {
-      "epoch": 6.000322074157575,
-      "grad_norm": 0.00869645643979311,
-      "learning_rate": 0.00019998224423862658,
-      "loss": 46.0,
-      "step": 37259
-    },
-    {
-      "epoch": 6.000483111236362,
-      "grad_norm": 0.009907534345984459,
-      "learning_rate": 0.00019998224328528696,
-      "loss": 46.0,
-      "step": 37260
-    },
-    {
-      "epoch": 6.00064414831515,
-      "grad_norm": 0.000844583788421005,
-      "learning_rate": 0.0001999822423319217,
-      "loss": 46.0,
-      "step": 37261
-    },
-    {
-      "epoch": 6.000805185393937,
-      "grad_norm": 0.005630886647850275,
-      "learning_rate": 0.00019998224137853088,
-      "loss": 46.0,
-      "step": 37262
-    },
-    {
-      "epoch": 6.000966222472725,
-      "grad_norm": 0.0012189788976684213,
-      "learning_rate": 0.00019998224042511447,
-      "loss": 46.0,
-      "step": 37263
-    },
-    {
-      "epoch": 6.001127259551512,
-      "grad_norm": 0.0024373335763812065,
-      "learning_rate": 0.00019998223947167247,
-      "loss": 46.0,
-      "step": 37264
-    },
-    {
-      "epoch": 6.001288296630299,
-      "grad_norm": 0.0018727300921455026,
-      "learning_rate": 0.00019998223851820486,
-      "loss": 46.0,
-      "step": 37265
-    },
-    {
-      "epoch": 6.001449333709086,
-      "grad_norm": 0.008964837528765202,
-      "learning_rate": 0.0001999822375647117,
-      "loss": 46.0,
-      "step": 37266
-    },
-    {
-      "epoch": 6.001610370787874,
-      "grad_norm": 0.0017164716264232993,
-      "learning_rate": 0.0001999822366111929,
-      "loss": 46.0,
-      "step": 37267
-    },
-    {
-      "epoch": 6.001771407866661,
-      "grad_norm": 0.0040796780958771706,
-      "learning_rate": 0.00019998223565764853,
-      "loss": 46.0,
-      "step": 37268
-    },
-    {
-      "epoch": 6.0019324449454485,
-      "grad_norm": 0.0067762043327093124,
-      "learning_rate": 0.0001999822347040786,
-      "loss": 46.0,
-      "step": 37269
-    },
-    {
-      "epoch": 6.002093482024236,
-      "grad_norm": 0.0016794345574453473,
-      "learning_rate": 0.00019998223375048305,
-      "loss": 46.0,
-      "step": 37270
-    },
-    {
-      "epoch": 6.002254519103023,
-      "grad_norm": 0.015847422182559967,
-      "learning_rate": 0.0001999822327968619,
-      "loss": 46.0,
-      "step": 37271
-    },
-    {
-      "epoch": 6.002415556181811,
-      "grad_norm": 0.004540848545730114,
-      "learning_rate": 0.00019998223184321517,
-      "loss": 46.0,
-      "step": 37272
-    },
-    {
-      "epoch": 6.002576593260598,
-      "grad_norm": 0.0037385665345937014,
-      "learning_rate": 0.00019998223088954283,
-      "loss": 46.0,
-      "step": 37273
-    },
-    {
-      "epoch": 6.002737630339386,
-      "grad_norm": 0.006277099717408419,
-      "learning_rate": 0.0001999822299358449,
-      "loss": 46.0,
-      "step": 37274
-    },
-    {
-      "epoch": 6.002898667418173,
-      "grad_norm": 0.0035424153320491314,
-      "learning_rate": 0.00019998222898212146,
-      "loss": 46.0,
-      "step": 37275
-    },
-    {
-      "epoch": 6.003059704496961,
-      "grad_norm": 0.009347725659608841,
-      "learning_rate": 0.00019998222802837233,
-      "loss": 46.0,
-      "step": 37276
-    },
-    {
-      "epoch": 6.003220741575748,
-      "grad_norm": 0.003090252634137869,
-      "learning_rate": 0.00019998222707459767,
-      "loss": 46.0,
-      "step": 37277
-    },
-    {
-      "epoch": 6.003381778654536,
-      "grad_norm": 0.011063123121857643,
-      "learning_rate": 0.0001999822261207974,
-      "loss": 46.0,
-      "step": 37278
-    },
-    {
-      "epoch": 6.003542815733322,
-      "grad_norm": 0.014631282538175583,
-      "learning_rate": 0.00019998222516697155,
-      "loss": 46.0,
-      "step": 37279
-    },
-    {
-      "epoch": 6.00370385281211,
-      "grad_norm": 0.002892265561968088,
-      "learning_rate": 0.0001999822242131201,
-      "loss": 46.0,
-      "step": 37280
-    },
-    {
-      "epoch": 6.003864889890897,
-      "grad_norm": 0.002990832319483161,
-      "learning_rate": 0.00019998222325924304,
-      "loss": 46.0,
-      "step": 37281
-    },
-    {
-      "epoch": 6.0040259269696845,
-      "grad_norm": 0.012151021510362625,
-      "learning_rate": 0.00019998222230534042,
-      "loss": 46.0,
-      "step": 37282
-    },
-    {
-      "epoch": 6.004186964048472,
-      "grad_norm": 0.0014796452596783638,
-      "learning_rate": 0.0001999822213514122,
-      "loss": 46.0,
-      "step": 37283
-    },
-    {
-      "epoch": 6.004348001127259,
-      "grad_norm": 0.0038918552454560995,
-      "learning_rate": 0.00019998222039745836,
-      "loss": 46.0,
-      "step": 37284
-    },
-    {
-      "epoch": 6.004509038206047,
-      "grad_norm": 0.006824355572462082,
-      "learning_rate": 0.00019998221944347896,
-      "loss": 46.0,
-      "step": 37285
-    },
-    {
-      "epoch": 6.004670075284834,
-      "grad_norm": 0.0050840736366808414,
-      "learning_rate": 0.000199982218489474,
-      "loss": 46.0,
-      "step": 37286
-    },
-    {
-      "epoch": 6.004831112363622,
-      "grad_norm": 0.004692744929343462,
-      "learning_rate": 0.0001999822175354434,
-      "loss": 46.0,
-      "step": 37287
-    },
-    {
-      "epoch": 6.004992149442409,
-      "grad_norm": 0.0027085093315690756,
-      "learning_rate": 0.00019998221658138723,
-      "loss": 46.0,
-      "step": 37288
-    },
-    {
-      "epoch": 6.005153186521197,
-      "grad_norm": 0.02194962278008461,
-      "learning_rate": 0.00019998221562730545,
-      "loss": 46.0,
-      "step": 37289
-    },
-    {
-      "epoch": 6.005314223599984,
-      "grad_norm": 0.007821271196007729,
-      "learning_rate": 0.0001999822146731981,
-      "loss": 46.0,
-      "step": 37290
-    },
-    {
-      "epoch": 6.005475260678772,
-      "grad_norm": 0.005405803211033344,
-      "learning_rate": 0.00019998221371906515,
-      "loss": 46.0,
-      "step": 37291
-    },
-    {
-      "epoch": 6.005636297757559,
-      "grad_norm": 0.008932650089263916,
-      "learning_rate": 0.00019998221276490663,
-      "loss": 46.0,
-      "step": 37292
-    },
-    {
-      "epoch": 6.0057973348363465,
-      "grad_norm": 0.005691902246326208,
-      "learning_rate": 0.0001999822118107225,
-      "loss": 46.0,
-      "step": 37293
-    },
-    {
-      "epoch": 6.005958371915133,
-      "grad_norm": 0.007117095403373241,
-      "learning_rate": 0.00019998221085651278,
-      "loss": 46.0,
-      "step": 37294
-    },
-    {
-      "epoch": 6.0061194089939205,
-      "grad_norm": 0.006958093028515577,
-      "learning_rate": 0.0001999822099022775,
-      "loss": 46.0,
-      "step": 37295
-    },
-    {
-      "epoch": 6.006280446072708,
-      "grad_norm": 0.012638378888368607,
-      "learning_rate": 0.00019998220894801657,
-      "loss": 46.0,
-      "step": 37296
-    },
-    {
-      "epoch": 6.006441483151495,
-      "grad_norm": 0.0037501745391637087,
-      "learning_rate": 0.0001999822079937301,
-      "loss": 46.0,
-      "step": 37297
-    },
-    {
-      "epoch": 6.006602520230283,
-      "grad_norm": 0.006044814828783274,
-      "learning_rate": 0.00019998220703941802,
-      "loss": 46.0,
-      "step": 37298
-    },
-    {
-      "epoch": 6.00676355730907,
-      "grad_norm": 0.009478239342570305,
-      "learning_rate": 0.00019998220608508037,
-      "loss": 46.0,
-      "step": 37299
-    },
-    {
-      "epoch": 6.006924594387858,
-      "grad_norm": 0.003994721919298172,
-      "learning_rate": 0.0001999822051307171,
-      "loss": 46.0,
-      "step": 37300
-    },
-    {
-      "epoch": 6.007085631466645,
-      "grad_norm": 0.0020560252014547586,
-      "learning_rate": 0.00019998220417632824,
-      "loss": 46.0,
-      "step": 37301
-    },
-    {
-      "epoch": 6.007246668545433,
-      "grad_norm": 0.009916729293763638,
-      "learning_rate": 0.00019998220322191382,
-      "loss": 46.0,
-      "step": 37302
-    },
-    {
-      "epoch": 6.00740770562422,
-      "grad_norm": 0.0034186893608421087,
-      "learning_rate": 0.00019998220226747376,
-      "loss": 46.0,
-      "step": 37303
-    },
-    {
-      "epoch": 6.007568742703008,
-      "grad_norm": 0.0007954526809044182,
-      "learning_rate": 0.00019998220131300814,
-      "loss": 46.0,
-      "step": 37304
-    },
-    {
-      "epoch": 6.007729779781795,
-      "grad_norm": 0.004026367329061031,
-      "learning_rate": 0.00019998220035851697,
-      "loss": 46.0,
-      "step": 37305
-    },
-    {
-      "epoch": 6.0078908168605825,
-      "grad_norm": 0.011674848385155201,
-      "learning_rate": 0.00019998219940400017,
-      "loss": 46.0,
-      "step": 37306
-    },
-    {
-      "epoch": 6.00805185393937,
-      "grad_norm": 0.003670050762593746,
-      "learning_rate": 0.0001999821984494578,
-      "loss": 46.0,
-      "step": 37307
-    },
-    {
-      "epoch": 6.008212891018157,
-      "grad_norm": 0.011337011121213436,
-      "learning_rate": 0.0001999821974948898,
-      "loss": 46.0,
-      "step": 37308
-    },
-    {
-      "epoch": 6.008373928096944,
-      "grad_norm": 0.005951997358351946,
-      "learning_rate": 0.00019998219654029621,
-      "loss": 46.0,
-      "step": 37309
-    },
-    {
-      "epoch": 6.008534965175731,
-      "grad_norm": 0.008224834688007832,
-      "learning_rate": 0.00019998219558567705,
-      "loss": 46.0,
-      "step": 37310
-    },
-    {
-      "epoch": 6.008696002254519,
-      "grad_norm": 0.004736274015158415,
-      "learning_rate": 0.0001999821946310323,
-      "loss": 46.0,
-      "step": 37311
-    },
-    {
-      "epoch": 6.008857039333306,
-      "grad_norm": 0.0023458453360944986,
-      "learning_rate": 0.00019998219367636195,
-      "loss": 46.0,
-      "step": 37312
-    },
-    {
-      "epoch": 6.009018076412094,
-      "grad_norm": 0.006603058893233538,
-      "learning_rate": 0.00019998219272166602,
-      "loss": 46.0,
-      "step": 37313
-    },
-    {
-      "epoch": 6.009179113490881,
-      "grad_norm": 0.010915643535554409,
-      "learning_rate": 0.0001999821917669445,
-      "loss": 46.0,
-      "step": 37314
-    },
-    {
-      "epoch": 6.009340150569669,
-      "grad_norm": 0.004482037387788296,
-      "learning_rate": 0.0001999821908121974,
-      "loss": 46.0,
-      "step": 37315
-    },
-    {
-      "epoch": 6.009501187648456,
-      "grad_norm": 0.015618345700204372,
-      "learning_rate": 0.0001999821898574247,
-      "loss": 46.0,
-      "step": 37316
-    },
-    {
-      "epoch": 6.0096622247272435,
-      "grad_norm": 0.01698637567460537,
-      "learning_rate": 0.00019998218890262642,
-      "loss": 46.0,
-      "step": 37317
-    },
-    {
-      "epoch": 6.009823261806031,
-      "grad_norm": 0.0017972791101783514,
-      "learning_rate": 0.0001999821879478025,
-      "loss": 46.0,
-      "step": 37318
-    },
-    {
-      "epoch": 6.009984298884818,
-      "grad_norm": 0.005839345045387745,
-      "learning_rate": 0.00019998218699295305,
-      "loss": 46.0,
-      "step": 37319
-    },
-    {
-      "epoch": 6.010145335963606,
-      "grad_norm": 0.013213645666837692,
-      "learning_rate": 0.00019998218603807798,
-      "loss": 46.0,
-      "step": 37320
-    },
-    {
-      "epoch": 6.010306373042393,
-      "grad_norm": 0.010846849530935287,
-      "learning_rate": 0.00019998218508317735,
-      "loss": 46.0,
-      "step": 37321
-    },
-    {
-      "epoch": 6.010467410121181,
-      "grad_norm": 0.002234281040728092,
-      "learning_rate": 0.00019998218412825108,
-      "loss": 46.0,
-      "step": 37322
-    },
-    {
-      "epoch": 6.010628447199967,
-      "grad_norm": 0.0065218741074204445,
-      "learning_rate": 0.00019998218317329925,
-      "loss": 46.0,
-      "step": 37323
-    },
-    {
-      "epoch": 6.010789484278755,
-      "grad_norm": 0.0009722562972456217,
-      "learning_rate": 0.00019998218221832184,
-      "loss": 46.0,
-      "step": 37324
-    },
-    {
-      "epoch": 6.010950521357542,
-      "grad_norm": 0.005423685535788536,
-      "learning_rate": 0.0001999821812633188,
-      "loss": 46.0,
-      "step": 37325
-    },
-    {
-      "epoch": 6.01111155843633,
-      "grad_norm": 0.001530548557639122,
-      "learning_rate": 0.0001999821803082902,
-      "loss": 46.0,
-      "step": 37326
-    },
-    {
-      "epoch": 6.011272595515117,
-      "grad_norm": 0.0016949146520346403,
-      "learning_rate": 0.000199982179353236,
-      "loss": 46.0,
-      "step": 37327
-    },
-    {
-      "epoch": 6.011433632593905,
-      "grad_norm": 0.008777761831879616,
-      "learning_rate": 0.00019998217839815622,
-      "loss": 46.0,
-      "step": 37328
-    },
-    {
-      "epoch": 6.011594669672692,
-      "grad_norm": 0.002952597802504897,
-      "learning_rate": 0.00019998217744305086,
-      "loss": 46.0,
-      "step": 37329
-    },
-    {
-      "epoch": 6.0117557067514795,
-      "grad_norm": 0.01073518954217434,
-      "learning_rate": 0.00019998217648791987,
-      "loss": 46.0,
-      "step": 37330
-    },
-    {
-      "epoch": 6.011916743830267,
-      "grad_norm": 0.005928761791437864,
-      "learning_rate": 0.00019998217553276332,
-      "loss": 46.0,
-      "step": 37331
-    },
-    {
-      "epoch": 6.012077780909054,
-      "grad_norm": 0.03993905708193779,
-      "learning_rate": 0.00019998217457758118,
-      "loss": 46.0,
-      "step": 37332
-    },
-    {
-      "epoch": 6.012238817987842,
-      "grad_norm": 0.0022269103210419416,
-      "learning_rate": 0.00019998217362237345,
-      "loss": 46.0,
-      "step": 37333
-    },
-    {
-      "epoch": 6.012399855066629,
-      "grad_norm": 0.0030804877169430256,
-      "learning_rate": 0.00019998217266714013,
-      "loss": 46.0,
-      "step": 37334
-    },
-    {
-      "epoch": 6.012560892145417,
-      "grad_norm": 0.0035887672565877438,
-      "learning_rate": 0.00019998217171188117,
-      "loss": 46.0,
-      "step": 37335
-    },
-    {
-      "epoch": 6.012721929224204,
-      "grad_norm": 0.0018088629003614187,
-      "learning_rate": 0.00019998217075659666,
-      "loss": 46.0,
-      "step": 37336
-    },
-    {
-      "epoch": 6.012882966302992,
-      "grad_norm": 0.004458155948668718,
-      "learning_rate": 0.00019998216980128655,
-      "loss": 46.0,
-      "step": 37337
-    },
-    {
-      "epoch": 6.013044003381778,
-      "grad_norm": 0.009427427314221859,
-      "learning_rate": 0.0001999821688459509,
-      "loss": 46.0,
-      "step": 37338
-    },
-    {
-      "epoch": 6.013205040460566,
-      "grad_norm": 0.0040281061083078384,
-      "learning_rate": 0.0001999821678905896,
-      "loss": 46.0,
-      "step": 37339
-    },
-    {
-      "epoch": 6.013366077539353,
-      "grad_norm": 0.0010320403380319476,
-      "learning_rate": 0.00019998216693520275,
-      "loss": 46.0,
-      "step": 37340
-    },
-    {
-      "epoch": 6.013527114618141,
-      "grad_norm": 0.007825557142496109,
-      "learning_rate": 0.0001999821659797903,
-      "loss": 46.0,
-      "step": 37341
-    },
-    {
-      "epoch": 6.013688151696928,
-      "grad_norm": 0.0043025859631598,
-      "learning_rate": 0.00019998216502435223,
-      "loss": 46.0,
-      "step": 37342
-    },
-    {
-      "epoch": 6.0138491887757155,
-      "grad_norm": 0.0018057033885270357,
-      "learning_rate": 0.0001999821640688886,
-      "loss": 46.0,
-      "step": 37343
-    },
-    {
-      "epoch": 6.014010225854503,
-      "grad_norm": 0.002024852205067873,
-      "learning_rate": 0.00019998216311339936,
-      "loss": 46.0,
-      "step": 37344
-    },
-    {
-      "epoch": 6.01417126293329,
-      "grad_norm": 0.007777228951454163,
-      "learning_rate": 0.00019998216215788453,
-      "loss": 46.0,
-      "step": 37345
-    },
-    {
-      "epoch": 6.014332300012078,
-      "grad_norm": 0.009013590402901173,
-      "learning_rate": 0.00019998216120234411,
-      "loss": 46.0,
-      "step": 37346
-    },
-    {
-      "epoch": 6.014493337090865,
-      "grad_norm": 0.0059803444892168045,
-      "learning_rate": 0.00019998216024677814,
-      "loss": 46.0,
-      "step": 37347
-    },
-    {
-      "epoch": 6.014654374169653,
-      "grad_norm": 0.001572663546539843,
-      "learning_rate": 0.00019998215929118652,
-      "loss": 46.0,
-      "step": 37348
-    },
-    {
-      "epoch": 6.01481541124844,
-      "grad_norm": 0.002022368600592017,
-      "learning_rate": 0.00019998215833556934,
-      "loss": 46.0,
-      "step": 37349
-    },
-    {
-      "epoch": 6.014976448327228,
-      "grad_norm": 0.006453935522586107,
-      "learning_rate": 0.00019998215737992658,
-      "loss": 46.0,
-      "step": 37350
-    },
-    {
-      "epoch": 6.015137485406015,
-      "grad_norm": 0.006935174111276865,
-      "learning_rate": 0.00019998215642425823,
-      "loss": 46.0,
-      "step": 37351
-    },
-    {
-      "epoch": 6.015298522484803,
-      "grad_norm": 0.017548134550452232,
-      "learning_rate": 0.00019998215546856426,
-      "loss": 46.0,
-      "step": 37352
-    },
-    {
-      "epoch": 6.015459559563589,
-      "grad_norm": 0.008491123095154762,
-      "learning_rate": 0.0001999821545128447,
-      "loss": 46.0,
-      "step": 37353
-    },
-    {
-      "epoch": 6.015620596642377,
-      "grad_norm": 0.0016087363474071026,
-      "learning_rate": 0.0001999821535570996,
-      "loss": 46.0,
-      "step": 37354
-    },
-    {
-      "epoch": 6.015781633721164,
-      "grad_norm": 0.0031094527803361416,
-      "learning_rate": 0.00019998215260132884,
-      "loss": 46.0,
-      "step": 37355
-    },
-    {
-      "epoch": 6.0159426707999515,
-      "grad_norm": 0.0021818201057612896,
-      "learning_rate": 0.00019998215164553255,
-      "loss": 46.0,
-      "step": 37356
-    },
-    {
-      "epoch": 6.016103707878739,
-      "grad_norm": 0.003377706278115511,
-      "learning_rate": 0.00019998215068971065,
-      "loss": 46.0,
-      "step": 37357
-    },
-    {
-      "epoch": 6.016264744957526,
-      "grad_norm": 0.00046303318231366575,
-      "learning_rate": 0.00019998214973386316,
-      "loss": 46.0,
-      "step": 37358
-    },
-    {
-      "epoch": 6.016425782036314,
-      "grad_norm": 0.006079382263123989,
-      "learning_rate": 0.00019998214877799003,
-      "loss": 46.0,
-      "step": 37359
-    },
-    {
-      "epoch": 6.016586819115101,
-      "grad_norm": 0.011122734285891056,
-      "learning_rate": 0.0001999821478220914,
-      "loss": 46.0,
-      "step": 37360
-    },
-    {
-      "epoch": 6.016747856193889,
-      "grad_norm": 0.012124674394726753,
-      "learning_rate": 0.00019998214686616712,
-      "loss": 46.0,
-      "step": 37361
-    },
-    {
-      "epoch": 6.016908893272676,
-      "grad_norm": 0.008392972871661186,
-      "learning_rate": 0.00019998214591021726,
-      "loss": 46.0,
-      "step": 37362
-    },
-    {
-      "epoch": 6.017069930351464,
-      "grad_norm": 0.010684572160243988,
-      "learning_rate": 0.0001999821449542418,
-      "loss": 46.0,
-      "step": 37363
-    },
-    {
-      "epoch": 6.017230967430251,
-      "grad_norm": 0.0042912717908620834,
-      "learning_rate": 0.0001999821439982408,
-      "loss": 46.0,
-      "step": 37364
-    },
-    {
-      "epoch": 6.0173920045090385,
-      "grad_norm": 0.016456574201583862,
-      "learning_rate": 0.00019998214304221414,
-      "loss": 46.0,
-      "step": 37365
-    },
-    {
-      "epoch": 6.017553041587826,
-      "grad_norm": 0.001738523948006332,
-      "learning_rate": 0.00019998214208616193,
-      "loss": 46.0,
-      "step": 37366
-    },
-    {
-      "epoch": 6.017714078666613,
-      "grad_norm": 0.0020229408983141184,
-      "learning_rate": 0.00019998214113008412,
-      "loss": 46.0,
-      "step": 37367
-    },
-    {
-      "epoch": 6.0178751157454,
-      "grad_norm": 0.0016072492580860853,
-      "learning_rate": 0.0001999821401739807,
-      "loss": 46.0,
-      "step": 37368
-    },
-    {
-      "epoch": 6.0180361528241875,
-      "grad_norm": 0.011316402815282345,
-      "learning_rate": 0.0001999821392178517,
-      "loss": 46.0,
-      "step": 37369
-    },
-    {
-      "epoch": 6.018197189902975,
-      "grad_norm": 0.001191994408145547,
-      "learning_rate": 0.00019998213826169715,
-      "loss": 46.0,
-      "step": 37370
-    },
-    {
-      "epoch": 6.018358226981762,
-      "grad_norm": 0.012513085268437862,
-      "learning_rate": 0.00019998213730551697,
-      "loss": 46.0,
-      "step": 37371
-    },
-    {
-      "epoch": 6.01851926406055,
-      "grad_norm": 0.0015407592291012406,
-      "learning_rate": 0.00019998213634931123,
-      "loss": 46.0,
-      "step": 37372
-    },
-    {
-      "epoch": 6.018680301139337,
-      "grad_norm": 0.0031680718529969454,
-      "learning_rate": 0.00019998213539307988,
-      "loss": 46.0,
-      "step": 37373
-    },
-    {
-      "epoch": 6.018841338218125,
-      "grad_norm": 0.01608908548951149,
-      "learning_rate": 0.00019998213443682294,
-      "loss": 46.0,
-      "step": 37374
-    },
-    {
-      "epoch": 6.019002375296912,
-      "grad_norm": 0.0027943188324570656,
-      "learning_rate": 0.0001999821334805404,
-      "loss": 46.0,
-      "step": 37375
-    },
-    {
-      "epoch": 6.0191634123757,
-      "grad_norm": 0.0034422928001731634,
-      "learning_rate": 0.00019998213252423228,
-      "loss": 46.0,
-      "step": 37376
-    },
-    {
-      "epoch": 6.019324449454487,
-      "grad_norm": 0.005031601991504431,
-      "learning_rate": 0.00019998213156789858,
-      "loss": 46.0,
-      "step": 37377
-    },
-    {
-      "epoch": 6.0194854865332745,
-      "grad_norm": 0.006252941209822893,
-      "learning_rate": 0.00019998213061153927,
-      "loss": 46.0,
-      "step": 37378
-    },
-    {
-      "epoch": 6.019646523612062,
-      "grad_norm": 0.0052984305657446384,
-      "learning_rate": 0.00019998212965515437,
-      "loss": 46.0,
-      "step": 37379
-    },
-    {
-      "epoch": 6.019807560690849,
-      "grad_norm": 0.007897310890257359,
-      "learning_rate": 0.0001999821286987439,
-      "loss": 46.0,
-      "step": 37380
-    },
-    {
-      "epoch": 6.019968597769637,
-      "grad_norm": 0.0014272683765739202,
-      "learning_rate": 0.00019998212774230783,
-      "loss": 46.0,
-      "step": 37381
-    },
-    {
-      "epoch": 6.0201296348484234,
-      "grad_norm": 0.005481986328959465,
-      "learning_rate": 0.00019998212678584617,
-      "loss": 46.0,
-      "step": 37382
-    },
-    {
-      "epoch": 6.020290671927211,
-      "grad_norm": 0.001721731387078762,
-      "learning_rate": 0.00019998212582935892,
-      "loss": 46.0,
-      "step": 37383
-    },
-    {
-      "epoch": 6.020451709005998,
-      "grad_norm": 0.000984185840934515,
-      "learning_rate": 0.00019998212487284606,
-      "loss": 46.0,
-      "step": 37384
-    },
-    {
-      "epoch": 6.020612746084786,
-      "grad_norm": 0.0015970334643498063,
-      "learning_rate": 0.00019998212391630763,
-      "loss": 46.0,
-      "step": 37385
-    },
-    {
-      "epoch": 6.020773783163573,
-      "grad_norm": 0.0028565407264977694,
-      "learning_rate": 0.00019998212295974362,
-      "loss": 46.0,
-      "step": 37386
-    },
-    {
-      "epoch": 6.020934820242361,
-      "grad_norm": 0.0026087737642228603,
-      "learning_rate": 0.000199982122003154,
-      "loss": 46.0,
-      "step": 37387
-    },
-    {
-      "epoch": 6.021095857321148,
-      "grad_norm": 0.0014479245292022824,
-      "learning_rate": 0.00019998212104653878,
-      "loss": 46.0,
-      "step": 37388
-    },
-    {
-      "epoch": 6.021256894399936,
-      "grad_norm": 0.002667824039235711,
-      "learning_rate": 0.000199982120089898,
-      "loss": 46.0,
-      "step": 37389
-    },
-    {
-      "epoch": 6.021417931478723,
-      "grad_norm": 0.006317758467048407,
-      "learning_rate": 0.00019998211913323162,
-      "loss": 46.0,
-      "step": 37390
-    },
-    {
-      "epoch": 6.0215789685575105,
-      "grad_norm": 0.0035046732518821955,
-      "learning_rate": 0.00019998211817653965,
-      "loss": 46.0,
-      "step": 37391
-    },
-    {
-      "epoch": 6.021740005636298,
-      "grad_norm": 0.0036133877001702785,
-      "learning_rate": 0.00019998211721982206,
-      "loss": 46.0,
-      "step": 37392
-    },
-    {
-      "epoch": 6.021901042715085,
-      "grad_norm": 0.0047370134852826595,
-      "learning_rate": 0.00019998211626307891,
-      "loss": 46.0,
-      "step": 37393
-    },
-    {
-      "epoch": 6.022062079793873,
-      "grad_norm": 0.0017004030523821712,
-      "learning_rate": 0.00019998211530631018,
-      "loss": 46.0,
-      "step": 37394
-    },
-    {
-      "epoch": 6.02222311687266,
-      "grad_norm": 0.007283463142812252,
-      "learning_rate": 0.00019998211434951583,
-      "loss": 46.0,
-      "step": 37395
-    },
-    {
-      "epoch": 6.022384153951447,
-      "grad_norm": 0.002013358287513256,
-      "learning_rate": 0.00019998211339269592,
-      "loss": 46.0,
-      "step": 37396
-    },
-    {
-      "epoch": 6.022545191030234,
-      "grad_norm": 0.004235814791172743,
-      "learning_rate": 0.0001999821124358504,
-      "loss": 46.0,
-      "step": 37397
-    },
-    {
-      "epoch": 6.022706228109022,
-      "grad_norm": 0.014187032356858253,
-      "learning_rate": 0.00019998211147897928,
-      "loss": 46.0,
-      "step": 37398
-    },
-    {
-      "epoch": 6.022867265187809,
-      "grad_norm": 0.0019778998102992773,
-      "learning_rate": 0.00019998211052208259,
-      "loss": 46.0,
-      "step": 37399
-    },
-    {
-      "epoch": 6.023028302266597,
-      "grad_norm": 0.021550370380282402,
-      "learning_rate": 0.0001999821095651603,
-      "loss": 46.0,
-      "step": 37400
-    },
-    {
-      "epoch": 6.023189339345384,
-      "grad_norm": 0.00854356400668621,
-      "learning_rate": 0.00019998210860821243,
-      "loss": 46.0,
-      "step": 37401
-    },
-    {
-      "epoch": 6.023350376424172,
-      "grad_norm": 0.0035862955264747143,
-      "learning_rate": 0.00019998210765123894,
-      "loss": 46.0,
-      "step": 37402
-    },
-    {
-      "epoch": 6.023511413502959,
-      "grad_norm": 0.006736627779901028,
-      "learning_rate": 0.0001999821066942399,
-      "loss": 46.0,
-      "step": 37403
-    },
-    {
-      "epoch": 6.0236724505817465,
-      "grad_norm": 0.008893050253391266,
-      "learning_rate": 0.00019998210573721523,
-      "loss": 46.0,
-      "step": 37404
-    },
-    {
-      "epoch": 6.023833487660534,
-      "grad_norm": 0.00908861868083477,
-      "learning_rate": 0.000199982104780165,
-      "loss": 46.0,
-      "step": 37405
-    },
-    {
-      "epoch": 6.023994524739321,
-      "grad_norm": 0.01650627702474594,
-      "learning_rate": 0.00019998210382308917,
-      "loss": 46.0,
-      "step": 37406
-    },
-    {
-      "epoch": 6.024155561818109,
-      "grad_norm": 0.006853667553514242,
-      "learning_rate": 0.00019998210286598775,
-      "loss": 46.0,
-      "step": 37407
-    },
-    {
-      "epoch": 6.024316598896896,
-      "grad_norm": 0.002855593804270029,
-      "learning_rate": 0.00019998210190886074,
-      "loss": 46.0,
-      "step": 37408
-    },
-    {
-      "epoch": 6.024477635975684,
-      "grad_norm": 0.006694091949611902,
-      "learning_rate": 0.00019998210095170814,
-      "loss": 46.0,
-      "step": 37409
-    },
-    {
-      "epoch": 6.024638673054471,
-      "grad_norm": 0.0018880751449614763,
-      "learning_rate": 0.00019998209999452996,
-      "loss": 46.0,
-      "step": 37410
-    },
-    {
-      "epoch": 6.024799710133258,
-      "grad_norm": 0.0077400715090334415,
-      "learning_rate": 0.0001999820990373262,
-      "loss": 46.0,
-      "step": 37411
-    },
-    {
-      "epoch": 6.024960747212045,
-      "grad_norm": 0.01441923901438713,
-      "learning_rate": 0.0001999820980800968,
-      "loss": 46.0,
-      "step": 37412
-    },
-    {
-      "epoch": 6.025121784290833,
-      "grad_norm": 0.005400239489972591,
-      "learning_rate": 0.00019998209712284183,
-      "loss": 46.0,
-      "step": 37413
-    },
-    {
-      "epoch": 6.02528282136962,
-      "grad_norm": 0.0047841304913163185,
-      "learning_rate": 0.0001999820961655613,
-      "loss": 46.0,
-      "step": 37414
-    },
-    {
-      "epoch": 6.025443858448408,
-      "grad_norm": 0.00750749371945858,
-      "learning_rate": 0.00019998209520825515,
-      "loss": 46.0,
-      "step": 37415
-    },
-    {
-      "epoch": 6.025604895527195,
-      "grad_norm": 0.006577526684850454,
-      "learning_rate": 0.00019998209425092342,
-      "loss": 46.0,
-      "step": 37416
-    },
-    {
-      "epoch": 6.0257659326059825,
-      "grad_norm": 0.002485182136297226,
-      "learning_rate": 0.0001999820932935661,
-      "loss": 46.0,
-      "step": 37417
-    },
-    {
-      "epoch": 6.02592696968477,
-      "grad_norm": 0.0026798786129802465,
-      "learning_rate": 0.00019998209233618316,
-      "loss": 46.0,
-      "step": 37418
-    },
-    {
-      "epoch": 6.026088006763557,
-      "grad_norm": 0.003676310181617737,
-      "learning_rate": 0.00019998209137877466,
-      "loss": 46.0,
-      "step": 37419
-    },
-    {
-      "epoch": 6.026249043842345,
-      "grad_norm": 0.0023098753299564123,
-      "learning_rate": 0.00019998209042134058,
-      "loss": 46.0,
-      "step": 37420
-    },
-    {
-      "epoch": 6.026410080921132,
-      "grad_norm": 0.012121430598199368,
-      "learning_rate": 0.0001999820894638809,
-      "loss": 46.0,
-      "step": 37421
-    },
-    {
-      "epoch": 6.02657111799992,
-      "grad_norm": 0.0013131614541634917,
-      "learning_rate": 0.00019998208850639562,
-      "loss": 46.0,
-      "step": 37422
-    },
-    {
-      "epoch": 6.026732155078707,
-      "grad_norm": 0.00855423603206873,
-      "learning_rate": 0.00019998208754888478,
-      "loss": 46.0,
-      "step": 37423
-    },
-    {
-      "epoch": 6.026893192157495,
-      "grad_norm": 0.0023163682781159878,
-      "learning_rate": 0.0001999820865913483,
-      "loss": 46.0,
-      "step": 37424
-    },
-    {
-      "epoch": 6.027054229236282,
-      "grad_norm": 0.009444854222238064,
-      "learning_rate": 0.00019998208563378627,
-      "loss": 46.0,
-      "step": 37425
-    },
-    {
-      "epoch": 6.027215266315069,
-      "grad_norm": 0.001725554233416915,
-      "learning_rate": 0.0001999820846761986,
-      "loss": 46.0,
-      "step": 37426
-    },
-    {
-      "epoch": 6.027376303393856,
-      "grad_norm": 0.0016803030157461762,
-      "learning_rate": 0.0001999820837185854,
-      "loss": 46.0,
-      "step": 37427
-    },
-    {
-      "epoch": 6.027537340472644,
-      "grad_norm": 0.006890660617500544,
-      "learning_rate": 0.00019998208276094658,
-      "loss": 46.0,
-      "step": 37428
-    },
-    {
-      "epoch": 6.027698377551431,
-      "grad_norm": 0.014234675094485283,
-      "learning_rate": 0.00019998208180328216,
-      "loss": 46.0,
-      "step": 37429
-    },
-    {
-      "epoch": 6.0278594146302185,
-      "grad_norm": 0.002693300601094961,
-      "learning_rate": 0.00019998208084559218,
-      "loss": 46.0,
-      "step": 37430
-    },
-    {
-      "epoch": 6.028020451709006,
-      "grad_norm": 0.00880126841366291,
-      "learning_rate": 0.0001999820798878766,
-      "loss": 46.0,
-      "step": 37431
-    },
-    {
-      "epoch": 6.028181488787793,
-      "grad_norm": 0.011463573202490807,
-      "learning_rate": 0.00019998207893013543,
-      "loss": 46.0,
-      "step": 37432
-    },
-    {
-      "epoch": 6.028342525866581,
-      "grad_norm": 0.010493949986994267,
-      "learning_rate": 0.00019998207797236866,
-      "loss": 46.0,
-      "step": 37433
-    },
-    {
-      "epoch": 6.028503562945368,
-      "grad_norm": 0.002940594684332609,
-      "learning_rate": 0.0001999820770145763,
-      "loss": 46.0,
-      "step": 37434
-    },
-    {
-      "epoch": 6.028664600024156,
-      "grad_norm": 0.0030396052170544863,
-      "learning_rate": 0.00019998207605675833,
-      "loss": 46.0,
-      "step": 37435
-    },
-    {
-      "epoch": 6.028825637102943,
-      "grad_norm": 0.002695484086871147,
-      "learning_rate": 0.0001999820750989148,
-      "loss": 46.0,
-      "step": 37436
-    },
-    {
-      "epoch": 6.028986674181731,
-      "grad_norm": 0.013518051244318485,
-      "learning_rate": 0.00019998207414104568,
-      "loss": 46.0,
-      "step": 37437
-    },
-    {
-      "epoch": 6.029147711260518,
-      "grad_norm": 0.002834593178704381,
-      "learning_rate": 0.00019998207318315097,
-      "loss": 46.0,
-      "step": 37438
-    },
-    {
-      "epoch": 6.0293087483393055,
-      "grad_norm": 0.00237145833671093,
-      "learning_rate": 0.00019998207222523068,
-      "loss": 46.0,
-      "step": 37439
-    },
-    {
-      "epoch": 6.029469785418092,
-      "grad_norm": 0.011666586622595787,
-      "learning_rate": 0.00019998207126728477,
-      "loss": 46.0,
-      "step": 37440
-    },
-    {
-      "epoch": 6.0296308224968795,
-      "grad_norm": 0.0017858650535345078,
-      "learning_rate": 0.00019998207030931327,
-      "loss": 46.0,
-      "step": 37441
-    },
-    {
-      "epoch": 6.029791859575667,
-      "grad_norm": 0.005706263706088066,
-      "learning_rate": 0.0001999820693513162,
-      "loss": 46.0,
-      "step": 37442
-    },
-    {
-      "epoch": 6.029952896654454,
-      "grad_norm": 0.00561361201107502,
-      "learning_rate": 0.00019998206839329352,
-      "loss": 46.0,
-      "step": 37443
-    },
-    {
-      "epoch": 6.030113933733242,
-      "grad_norm": 0.005115280393511057,
-      "learning_rate": 0.00019998206743524526,
-      "loss": 46.0,
-      "step": 37444
-    },
-    {
-      "epoch": 6.030274970812029,
-      "grad_norm": 0.0019947404507547617,
-      "learning_rate": 0.0001999820664771714,
-      "loss": 46.0,
-      "step": 37445
-    },
-    {
-      "epoch": 6.030436007890817,
-      "grad_norm": 0.006450181361287832,
-      "learning_rate": 0.00019998206551907196,
-      "loss": 46.0,
-      "step": 37446
-    },
-    {
-      "epoch": 6.030597044969604,
-      "grad_norm": 0.009363310411572456,
-      "learning_rate": 0.00019998206456094692,
-      "loss": 46.0,
-      "step": 37447
-    },
-    {
-      "epoch": 6.030758082048392,
-      "grad_norm": 0.00960224587470293,
-      "learning_rate": 0.0001999820636027963,
-      "loss": 46.0,
-      "step": 37448
-    },
-    {
-      "epoch": 6.030919119127179,
-      "grad_norm": 0.0018297643400728703,
-      "learning_rate": 0.0001999820626446201,
-      "loss": 46.0,
-      "step": 37449
-    },
-    {
-      "epoch": 6.031080156205967,
-      "grad_norm": 0.006229907739907503,
-      "learning_rate": 0.0001999820616864183,
-      "loss": 46.0,
-      "step": 37450
-    },
-    {
-      "epoch": 6.031241193284754,
-      "grad_norm": 0.010183049365878105,
-      "learning_rate": 0.0001999820607281909,
-      "loss": 46.0,
-      "step": 37451
-    },
-    {
-      "epoch": 6.0314022303635415,
-      "grad_norm": 0.014703231863677502,
-      "learning_rate": 0.0001999820597699379,
-      "loss": 46.0,
-      "step": 37452
-    },
-    {
-      "epoch": 6.031563267442329,
-      "grad_norm": 0.0016461131162941456,
-      "learning_rate": 0.00019998205881165934,
-      "loss": 46.0,
-      "step": 37453
-    },
-    {
-      "epoch": 6.031724304521116,
-      "grad_norm": 0.0014284674543887377,
-      "learning_rate": 0.0001999820578533552,
-      "loss": 46.0,
-      "step": 37454
-    },
-    {
-      "epoch": 6.031885341599903,
-      "grad_norm": 0.005886481609195471,
-      "learning_rate": 0.00019998205689502542,
-      "loss": 46.0,
-      "step": 37455
-    },
-    {
-      "epoch": 6.03204637867869,
-      "grad_norm": 0.01851947233080864,
-      "learning_rate": 0.0001999820559366701,
-      "loss": 46.0,
-      "step": 37456
-    },
-    {
-      "epoch": 6.032207415757478,
-      "grad_norm": 0.00692877359688282,
-      "learning_rate": 0.00019998205497828914,
-      "loss": 46.0,
-      "step": 37457
-    },
-    {
-      "epoch": 6.032368452836265,
-      "grad_norm": 0.0025417099241167307,
-      "learning_rate": 0.00019998205401988264,
-      "loss": 46.0,
-      "step": 37458
-    },
-    {
-      "epoch": 6.032529489915053,
-      "grad_norm": 0.007743894588202238,
-      "learning_rate": 0.0001999820530614505,
-      "loss": 46.0,
-      "step": 37459
-    },
-    {
-      "epoch": 6.03269052699384,
-      "grad_norm": 0.004280034452676773,
-      "learning_rate": 0.00019998205210299282,
-      "loss": 46.0,
-      "step": 37460
-    },
-    {
-      "epoch": 6.032851564072628,
-      "grad_norm": 0.004860276356339455,
-      "learning_rate": 0.00019998205114450952,
-      "loss": 46.0,
-      "step": 37461
-    },
-    {
-      "epoch": 6.033012601151415,
-      "grad_norm": 0.010296354070305824,
-      "learning_rate": 0.00019998205018600062,
-      "loss": 46.0,
-      "step": 37462
-    },
-    {
-      "epoch": 6.033173638230203,
-      "grad_norm": 0.004620649851858616,
-      "learning_rate": 0.00019998204922746618,
-      "loss": 46.0,
-      "step": 37463
-    },
-    {
-      "epoch": 6.03333467530899,
-      "grad_norm": 0.0031424169428646564,
-      "learning_rate": 0.0001999820482689061,
-      "loss": 46.0,
-      "step": 37464
-    },
-    {
-      "epoch": 6.0334957123877775,
-      "grad_norm": 0.0029387453105300665,
-      "learning_rate": 0.00019998204731032045,
-      "loss": 46.0,
-      "step": 37465
-    },
-    {
-      "epoch": 6.033656749466565,
-      "grad_norm": 0.0012476368574425578,
-      "learning_rate": 0.0001999820463517092,
-      "loss": 46.0,
-      "step": 37466
-    },
-    {
-      "epoch": 6.033817786545352,
-      "grad_norm": 0.0011241765459999442,
-      "learning_rate": 0.00019998204539307235,
-      "loss": 46.0,
-      "step": 37467
-    },
-    {
-      "epoch": 6.03397882362414,
-      "grad_norm": 0.005471441429108381,
-      "learning_rate": 0.0001999820444344099,
-      "loss": 46.0,
-      "step": 37468
-    },
-    {
-      "epoch": 6.034139860702927,
-      "grad_norm": 0.003955712541937828,
-      "learning_rate": 0.00019998204347572188,
-      "loss": 46.0,
-      "step": 37469
-    },
-    {
-      "epoch": 6.034300897781714,
-      "grad_norm": 0.001570369116961956,
-      "learning_rate": 0.0001999820425170083,
-      "loss": 46.0,
-      "step": 37470
-    },
-    {
-      "epoch": 6.034461934860501,
-      "grad_norm": 0.0023459214717149734,
-      "learning_rate": 0.0001999820415582691,
-      "loss": 46.0,
-      "step": 37471
-    },
-    {
-      "epoch": 6.034622971939289,
-      "grad_norm": 0.0036348237190395594,
-      "learning_rate": 0.0001999820405995043,
-      "loss": 46.0,
-      "step": 37472
-    },
-    {
-      "epoch": 6.034784009018076,
-      "grad_norm": 0.003946497570723295,
-      "learning_rate": 0.00019998203964071394,
-      "loss": 46.0,
-      "step": 37473
-    },
-    {
-      "epoch": 6.034945046096864,
-      "grad_norm": 0.009683527983725071,
-      "learning_rate": 0.000199982038681898,
-      "loss": 46.0,
-      "step": 37474
-    },
-    {
-      "epoch": 6.035106083175651,
-      "grad_norm": 0.008714226074516773,
-      "learning_rate": 0.00019998203772305642,
-      "loss": 46.0,
-      "step": 37475
-    },
-    {
-      "epoch": 6.035267120254439,
-      "grad_norm": 0.004451780114322901,
-      "learning_rate": 0.00019998203676418926,
-      "loss": 46.0,
-      "step": 37476
-    },
-    {
-      "epoch": 6.035428157333226,
-      "grad_norm": 0.004312592092901468,
-      "learning_rate": 0.00019998203580529652,
-      "loss": 46.0,
-      "step": 37477
-    },
-    {
-      "epoch": 6.0355891944120135,
-      "grad_norm": 0.0028346767649054527,
-      "learning_rate": 0.0001999820348463782,
-      "loss": 46.0,
-      "step": 37478
-    },
-    {
-      "epoch": 6.035750231490801,
-      "grad_norm": 0.010172472335398197,
-      "learning_rate": 0.00019998203388743425,
-      "loss": 46.0,
-      "step": 37479
-    },
-    {
-      "epoch": 6.035911268569588,
-      "grad_norm": 0.019719993695616722,
-      "learning_rate": 0.00019998203292846475,
-      "loss": 46.0,
-      "step": 37480
-    },
-    {
-      "epoch": 6.036072305648376,
-      "grad_norm": 0.0034994955640286207,
-      "learning_rate": 0.00019998203196946968,
-      "loss": 46.0,
-      "step": 37481
-    },
-    {
-      "epoch": 6.036233342727163,
-      "grad_norm": 0.009868305176496506,
-      "learning_rate": 0.00019998203101044898,
-      "loss": 46.0,
-      "step": 37482
-    },
-    {
-      "epoch": 6.036394379805951,
-      "grad_norm": 0.0043615433387458324,
-      "learning_rate": 0.0001999820300514027,
-      "loss": 46.0,
-      "step": 37483
-    },
-    {
-      "epoch": 6.036555416884737,
-      "grad_norm": 0.010640202090144157,
-      "learning_rate": 0.0001999820290923308,
-      "loss": 46.0,
-      "step": 37484
-    },
-    {
-      "epoch": 6.036716453963525,
-      "grad_norm": 0.0031366292387247086,
-      "learning_rate": 0.00019998202813323337,
-      "loss": 46.0,
-      "step": 37485
-    },
-    {
-      "epoch": 6.036877491042312,
-      "grad_norm": 0.001333284075371921,
-      "learning_rate": 0.00019998202717411032,
-      "loss": 46.0,
-      "step": 37486
-    },
-    {
-      "epoch": 6.0370385281211,
-      "grad_norm": 0.004389138892292976,
-      "learning_rate": 0.00019998202621496165,
-      "loss": 46.0,
-      "step": 37487
-    },
-    {
-      "epoch": 6.037199565199887,
-      "grad_norm": 0.0037388508208096027,
-      "learning_rate": 0.00019998202525578745,
-      "loss": 46.0,
-      "step": 37488
-    },
-    {
-      "epoch": 6.0373606022786745,
-      "grad_norm": 0.0016469616675749421,
-      "learning_rate": 0.0001999820242965876,
-      "loss": 46.0,
-      "step": 37489
-    },
-    {
-      "epoch": 6.037521639357462,
-      "grad_norm": 0.0033098775893449783,
-      "learning_rate": 0.0001999820233373622,
-      "loss": 46.0,
-      "step": 37490
-    },
-    {
-      "epoch": 6.0376826764362495,
-      "grad_norm": 0.021230148151516914,
-      "learning_rate": 0.0001999820223781112,
-      "loss": 46.0,
-      "step": 37491
-    },
-    {
-      "epoch": 6.037843713515037,
-      "grad_norm": 0.0025015573482960463,
-      "learning_rate": 0.00019998202141883459,
-      "loss": 46.0,
-      "step": 37492
-    },
-    {
-      "epoch": 6.038004750593824,
-      "grad_norm": 0.00358877913095057,
-      "learning_rate": 0.0001999820204595324,
-      "loss": 46.0,
-      "step": 37493
-    },
-    {
-      "epoch": 6.038165787672612,
-      "grad_norm": 0.006047093775123358,
-      "learning_rate": 0.00019998201950020464,
-      "loss": 46.0,
-      "step": 37494
-    },
-    {
-      "epoch": 6.038326824751399,
-      "grad_norm": 0.0024844740983098745,
-      "learning_rate": 0.00019998201854085128,
-      "loss": 46.0,
-      "step": 37495
-    },
-    {
-      "epoch": 6.038487861830187,
-      "grad_norm": 0.0014714018907397985,
-      "learning_rate": 0.00019998201758147233,
-      "loss": 46.0,
-      "step": 37496
-    },
-    {
-      "epoch": 6.038648898908974,
-      "grad_norm": 0.0034114255104213953,
-      "learning_rate": 0.0001999820166220678,
-      "loss": 46.0,
-      "step": 37497
-    },
-    {
-      "epoch": 6.038809935987762,
-      "grad_norm": 0.0029032016173005104,
-      "learning_rate": 0.00019998201566263766,
-      "loss": 46.0,
-      "step": 37498
-    },
-    {
-      "epoch": 6.038970973066548,
-      "grad_norm": 0.002625077962875366,
-      "learning_rate": 0.00019998201470318192,
-      "loss": 46.0,
-      "step": 37499
-    },
-    {
-      "epoch": 6.039132010145336,
-      "grad_norm": 0.001698369043879211,
-      "learning_rate": 0.0001999820137437006,
-      "loss": 46.0,
-      "step": 37500
-    },
-    {
-      "epoch": 6.039293047224123,
-      "grad_norm": 0.0052072396501898766,
-      "learning_rate": 0.00019998201278419368,
-      "loss": 46.0,
-      "step": 37501
-    },
-    {
-      "epoch": 6.0394540843029105,
-      "grad_norm": 0.003848006948828697,
-      "learning_rate": 0.00019998201182466117,
-      "loss": 46.0,
-      "step": 37502
-    },
-    {
-      "epoch": 6.039615121381698,
-      "grad_norm": 0.010110526345670223,
-      "learning_rate": 0.0001999820108651031,
-      "loss": 46.0,
-      "step": 37503
-    },
-    {
-      "epoch": 6.039776158460485,
-      "grad_norm": 0.006718708202242851,
-      "learning_rate": 0.00019998200990551946,
-      "loss": 46.0,
-      "step": 37504
-    },
-    {
-      "epoch": 6.039937195539273,
-      "grad_norm": 0.001223905710503459,
-      "learning_rate": 0.00019998200894591017,
-      "loss": 46.0,
-      "step": 37505
-    },
-    {
-      "epoch": 6.04009823261806,
-      "grad_norm": 0.022507235407829285,
-      "learning_rate": 0.00019998200798627532,
-      "loss": 46.0,
-      "step": 37506
-    },
-    {
-      "epoch": 6.040259269696848,
-      "grad_norm": 0.004028702154755592,
-      "learning_rate": 0.00019998200702661488,
-      "loss": 46.0,
-      "step": 37507
-    },
-    {
-      "epoch": 6.040420306775635,
-      "grad_norm": 0.0016088886186480522,
-      "learning_rate": 0.00019998200606692883,
-      "loss": 46.0,
-      "step": 37508
-    },
-    {
-      "epoch": 6.040581343854423,
-      "grad_norm": 0.01807459071278572,
-      "learning_rate": 0.00019998200510721722,
-      "loss": 46.0,
-      "step": 37509
-    },
-    {
-      "epoch": 6.04074238093321,
-      "grad_norm": 0.01649654284119606,
-      "learning_rate": 0.00019998200414748,
-      "loss": 46.0,
-      "step": 37510
-    },
-    {
-      "epoch": 6.040903418011998,
-      "grad_norm": 0.004719622898846865,
-      "learning_rate": 0.00019998200318771718,
-      "loss": 46.0,
-      "step": 37511
-    },
-    {
-      "epoch": 6.041064455090785,
-      "grad_norm": 0.0028916921000927687,
-      "learning_rate": 0.00019998200222792878,
-      "loss": 46.0,
-      "step": 37512
-    },
-    {
-      "epoch": 6.041225492169572,
-      "grad_norm": 0.0018338811350986362,
-      "learning_rate": 0.0001999820012681148,
-      "loss": 46.0,
-      "step": 37513
-    },
-    {
-      "epoch": 6.041386529248359,
-      "grad_norm": 0.0019286073511466384,
-      "learning_rate": 0.0001999820003082752,
-      "loss": 46.0,
-      "step": 37514
-    },
-    {
-      "epoch": 6.0415475663271465,
-      "grad_norm": 0.0032685145270079374,
-      "learning_rate": 0.00019998199934841005,
-      "loss": 46.0,
-      "step": 37515
-    },
-    {
-      "epoch": 6.041708603405934,
-      "grad_norm": 0.0031629721634089947,
-      "learning_rate": 0.00019998199838851927,
-      "loss": 46.0,
-      "step": 37516
-    },
-    {
-      "epoch": 6.041869640484721,
-      "grad_norm": 0.002542752306908369,
-      "learning_rate": 0.00019998199742860294,
-      "loss": 46.0,
-      "step": 37517
-    },
-    {
-      "epoch": 6.042030677563509,
-      "grad_norm": 0.003851477289572358,
-      "learning_rate": 0.00019998199646866098,
-      "loss": 46.0,
-      "step": 37518
-    },
-    {
-      "epoch": 6.042191714642296,
-      "grad_norm": 0.007247148547321558,
-      "learning_rate": 0.00019998199550869345,
-      "loss": 46.0,
-      "step": 37519
-    },
-    {
-      "epoch": 6.042352751721084,
-      "grad_norm": 0.004933567717671394,
-      "learning_rate": 0.00019998199454870035,
-      "loss": 46.0,
-      "step": 37520
-    },
-    {
-      "epoch": 6.042513788799871,
-      "grad_norm": 0.01611747033894062,
-      "learning_rate": 0.0001999819935886816,
-      "loss": 46.0,
-      "step": 37521
-    },
-    {
-      "epoch": 6.042674825878659,
-      "grad_norm": 0.004654871765524149,
-      "learning_rate": 0.00019998199262863734,
-      "loss": 46.0,
-      "step": 37522
-    },
-    {
-      "epoch": 6.042835862957446,
-      "grad_norm": 0.010619490407407284,
-      "learning_rate": 0.00019998199166856745,
-      "loss": 46.0,
-      "step": 37523
-    },
-    {
-      "epoch": 6.042996900036234,
-      "grad_norm": 0.0028765748720616102,
-      "learning_rate": 0.00019998199070847195,
-      "loss": 46.0,
-      "step": 37524
-    },
-    {
-      "epoch": 6.043157937115021,
-      "grad_norm": 0.010387152433395386,
-      "learning_rate": 0.00019998198974835089,
-      "loss": 46.0,
-      "step": 37525
-    },
-    {
-      "epoch": 6.0433189741938085,
-      "grad_norm": 0.0020469878800213337,
-      "learning_rate": 0.0001999819887882042,
-      "loss": 46.0,
-      "step": 37526
-    },
-    {
-      "epoch": 6.043480011272596,
-      "grad_norm": 0.010006871074438095,
-      "learning_rate": 0.00019998198782803197,
-      "loss": 46.0,
-      "step": 37527
-    },
-    {
-      "epoch": 6.0436410483513825,
-      "grad_norm": 0.0013885344378650188,
-      "learning_rate": 0.00019998198686783412,
-      "loss": 46.0,
-      "step": 37528
-    },
-    {
-      "epoch": 6.04380208543017,
-      "grad_norm": 0.0037922165356576443,
-      "learning_rate": 0.0001999819859076107,
-      "loss": 46.0,
-      "step": 37529
-    },
-    {
-      "epoch": 6.043963122508957,
-      "grad_norm": 0.008302551694214344,
-      "learning_rate": 0.00019998198494736166,
-      "loss": 46.0,
-      "step": 37530
-    },
-    {
-      "epoch": 6.044124159587745,
-      "grad_norm": 0.006932658609002829,
-      "learning_rate": 0.00019998198398708705,
-      "loss": 46.0,
-      "step": 37531
-    },
-    {
-      "epoch": 6.044285196666532,
-      "grad_norm": 0.003112620674073696,
-      "learning_rate": 0.00019998198302678682,
-      "loss": 46.0,
-      "step": 37532
-    },
-    {
-      "epoch": 6.04444623374532,
-      "grad_norm": 0.009254147298634052,
-      "learning_rate": 0.00019998198206646107,
-      "loss": 46.0,
-      "step": 37533
-    },
-    {
-      "epoch": 6.044607270824107,
-      "grad_norm": 0.001310255378484726,
-      "learning_rate": 0.00019998198110610967,
-      "loss": 46.0,
-      "step": 37534
-    },
-    {
-      "epoch": 6.044768307902895,
-      "grad_norm": 0.0075689353980124,
-      "learning_rate": 0.00019998198014573268,
-      "loss": 46.0,
-      "step": 37535
-    },
-    {
-      "epoch": 6.044929344981682,
-      "grad_norm": 0.004640243481844664,
-      "learning_rate": 0.00019998197918533013,
-      "loss": 46.0,
-      "step": 37536
-    },
-    {
-      "epoch": 6.04509038206047,
-      "grad_norm": 0.0066511379554867744,
-      "learning_rate": 0.00019998197822490197,
-      "loss": 46.0,
-      "step": 37537
-    },
-    {
-      "epoch": 6.045251419139257,
-      "grad_norm": 0.004504088778048754,
-      "learning_rate": 0.00019998197726444822,
-      "loss": 46.0,
-      "step": 37538
-    },
-    {
-      "epoch": 6.0454124562180445,
-      "grad_norm": 0.0021557160653173923,
-      "learning_rate": 0.00019998197630396888,
-      "loss": 46.0,
-      "step": 37539
-    },
-    {
-      "epoch": 6.045573493296832,
-      "grad_norm": 0.006787695921957493,
-      "learning_rate": 0.00019998197534346396,
-      "loss": 46.0,
-      "step": 37540
-    },
-    {
-      "epoch": 6.045734530375619,
-      "grad_norm": 0.0007721335859969258,
-      "learning_rate": 0.00019998197438293345,
-      "loss": 46.0,
-      "step": 37541
-    },
-    {
-      "epoch": 6.045895567454407,
-      "grad_norm": 0.0205103550106287,
-      "learning_rate": 0.00019998197342237735,
-      "loss": 46.0,
-      "step": 37542
-    },
-    {
-      "epoch": 6.046056604533193,
-      "grad_norm": 0.008283515460789204,
-      "learning_rate": 0.00019998197246179564,
-      "loss": 46.0,
-      "step": 37543
-    },
-    {
-      "epoch": 6.046217641611981,
-      "grad_norm": 0.0037953618448227644,
-      "learning_rate": 0.00019998197150118837,
-      "loss": 46.0,
-      "step": 37544
-    },
-    {
-      "epoch": 6.046378678690768,
-      "grad_norm": 0.01417340524494648,
-      "learning_rate": 0.00019998197054055548,
-      "loss": 46.0,
-      "step": 37545
-    },
-    {
-      "epoch": 6.046539715769556,
-      "grad_norm": 0.00409560464322567,
-      "learning_rate": 0.000199981969579897,
-      "loss": 46.0,
-      "step": 37546
-    },
-    {
-      "epoch": 6.046700752848343,
-      "grad_norm": 0.0052768150344491005,
-      "learning_rate": 0.00019998196861921295,
-      "loss": 46.0,
-      "step": 37547
-    },
-    {
-      "epoch": 6.046861789927131,
-      "grad_norm": 0.0019367403583601117,
-      "learning_rate": 0.0001999819676585033,
-      "loss": 46.0,
-      "step": 37548
-    },
-    {
-      "epoch": 6.047022827005918,
-      "grad_norm": 0.003626108868047595,
-      "learning_rate": 0.00019998196669776807,
-      "loss": 46.0,
-      "step": 37549
-    },
-    {
-      "epoch": 6.0471838640847055,
-      "grad_norm": 0.008510523475706577,
-      "learning_rate": 0.00019998196573700722,
-      "loss": 46.0,
-      "step": 37550
-    },
-    {
-      "epoch": 6.047344901163493,
-      "grad_norm": 0.0049221389926970005,
-      "learning_rate": 0.0001999819647762208,
-      "loss": 46.0,
-      "step": 37551
-    },
-    {
-      "epoch": 6.04750593824228,
-      "grad_norm": 0.0011162991868332028,
-      "learning_rate": 0.0001999819638154088,
-      "loss": 46.0,
-      "step": 37552
-    },
-    {
-      "epoch": 6.047666975321068,
-      "grad_norm": 0.003094008192420006,
-      "learning_rate": 0.0001999819628545712,
-      "loss": 46.0,
-      "step": 37553
-    },
-    {
-      "epoch": 6.047828012399855,
-      "grad_norm": 0.006014491431415081,
-      "learning_rate": 0.00019998196189370803,
-      "loss": 46.0,
-      "step": 37554
-    },
-    {
-      "epoch": 6.047989049478643,
-      "grad_norm": 0.007106460630893707,
-      "learning_rate": 0.00019998196093281925,
-      "loss": 46.0,
-      "step": 37555
-    },
-    {
-      "epoch": 6.04815008655743,
-      "grad_norm": 0.0056846062652766705,
-      "learning_rate": 0.00019998195997190487,
-      "loss": 46.0,
-      "step": 37556
-    },
-    {
-      "epoch": 6.048311123636217,
-      "grad_norm": 0.0012455809628590941,
-      "learning_rate": 0.00019998195901096491,
-      "loss": 46.0,
-      "step": 37557
-    },
-    {
-      "epoch": 6.048472160715004,
-      "grad_norm": 0.0029519947711378336,
-      "learning_rate": 0.00019998195804999934,
-      "loss": 46.0,
-      "step": 37558
-    },
-    {
-      "epoch": 6.048633197793792,
-      "grad_norm": 0.02495420351624489,
-      "learning_rate": 0.0001999819570890082,
-      "loss": 46.0,
-      "step": 37559
-    },
-    {
-      "epoch": 6.048794234872579,
-      "grad_norm": 0.00945096742361784,
-      "learning_rate": 0.00019998195612799149,
-      "loss": 46.0,
-      "step": 37560
-    },
-    {
-      "epoch": 6.048955271951367,
-      "grad_norm": 0.007004577666521072,
-      "learning_rate": 0.00019998195516694918,
-      "loss": 46.0,
-      "step": 37561
-    },
-    {
-      "epoch": 6.049116309030154,
-      "grad_norm": 0.006936834659427404,
-      "learning_rate": 0.00019998195420588126,
-      "loss": 46.0,
-      "step": 37562
-    },
-    {
-      "epoch": 6.0492773461089415,
-      "grad_norm": 0.005849672947078943,
-      "learning_rate": 0.00019998195324478775,
-      "loss": 46.0,
-      "step": 37563
-    },
-    {
-      "epoch": 6.049438383187729,
-      "grad_norm": 0.0059710959903895855,
-      "learning_rate": 0.00019998195228366865,
-      "loss": 46.0,
-      "step": 37564
-    },
-    {
-      "epoch": 6.049599420266516,
-      "grad_norm": 0.0028126032557338476,
-      "learning_rate": 0.00019998195132252397,
-      "loss": 46.0,
-      "step": 37565
-    },
-    {
-      "epoch": 6.049760457345304,
-      "grad_norm": 0.00363634480163455,
-      "learning_rate": 0.00019998195036135372,
-      "loss": 46.0,
-      "step": 37566
-    },
-    {
-      "epoch": 6.049921494424091,
-      "grad_norm": 0.0015998026356101036,
-      "learning_rate": 0.00019998194940015784,
-      "loss": 46.0,
-      "step": 37567
-    },
-    {
-      "epoch": 6.050082531502879,
-      "grad_norm": 0.0032095424830913544,
-      "learning_rate": 0.0001999819484389364,
-      "loss": 46.0,
-      "step": 37568
-    },
-    {
-      "epoch": 6.050243568581666,
-      "grad_norm": 0.0014049247838556767,
-      "learning_rate": 0.00019998194747768936,
-      "loss": 46.0,
-      "step": 37569
-    },
-    {
-      "epoch": 6.050404605660454,
-      "grad_norm": 0.001702752080745995,
-      "learning_rate": 0.00019998194651641674,
-      "loss": 46.0,
-      "step": 37570
-    },
-    {
-      "epoch": 6.050565642739241,
-      "grad_norm": 0.010074181482195854,
-      "learning_rate": 0.00019998194555511848,
-      "loss": 46.0,
-      "step": 37571
-    },
-    {
-      "epoch": 6.050726679818028,
-      "grad_norm": 0.0052128564566373825,
-      "learning_rate": 0.00019998194459379466,
-      "loss": 46.0,
-      "step": 37572
-    },
-    {
-      "epoch": 6.050887716896815,
-      "grad_norm": 0.002733284141868353,
-      "learning_rate": 0.00019998194363244528,
-      "loss": 46.0,
-      "step": 37573
-    },
-    {
-      "epoch": 6.051048753975603,
-      "grad_norm": 0.0018818244570866227,
-      "learning_rate": 0.00019998194267107028,
-      "loss": 46.0,
-      "step": 37574
-    },
-    {
-      "epoch": 6.05120979105439,
-      "grad_norm": 0.007615710608661175,
-      "learning_rate": 0.0001999819417096697,
-      "loss": 46.0,
-      "step": 37575
-    },
-    {
-      "epoch": 6.0513708281331775,
-      "grad_norm": 0.008013997226953506,
-      "learning_rate": 0.00019998194074824353,
-      "loss": 46.0,
-      "step": 37576
-    },
-    {
-      "epoch": 6.051531865211965,
-      "grad_norm": 0.003439794760197401,
-      "learning_rate": 0.00019998193978679177,
-      "loss": 46.0,
-      "step": 37577
-    },
-    {
-      "epoch": 6.051692902290752,
-      "grad_norm": 0.0008201716700568795,
-      "learning_rate": 0.0001999819388253144,
-      "loss": 46.0,
-      "step": 37578
-    },
-    {
-      "epoch": 6.05185393936954,
-      "grad_norm": 0.013601507991552353,
-      "learning_rate": 0.00019998193786381144,
-      "loss": 46.0,
-      "step": 37579
-    },
-    {
-      "epoch": 6.052014976448327,
-      "grad_norm": 0.017344433814287186,
-      "learning_rate": 0.00019998193690228292,
-      "loss": 46.0,
-      "step": 37580
-    },
-    {
-      "epoch": 6.052176013527115,
-      "grad_norm": 0.017316481098532677,
-      "learning_rate": 0.00019998193594072882,
-      "loss": 46.0,
-      "step": 37581
-    },
-    {
-      "epoch": 6.052337050605902,
-      "grad_norm": 0.00462355138733983,
-      "learning_rate": 0.00019998193497914907,
-      "loss": 46.0,
-      "step": 37582
-    },
-    {
-      "epoch": 6.05249808768469,
-      "grad_norm": 0.010756057687103748,
-      "learning_rate": 0.00019998193401754376,
-      "loss": 46.0,
-      "step": 37583
-    },
-    {
-      "epoch": 6.052659124763477,
-      "grad_norm": 0.0032620029523968697,
-      "learning_rate": 0.00019998193305591287,
-      "loss": 46.0,
-      "step": 37584
-    },
-    {
-      "epoch": 6.052820161842265,
-      "grad_norm": 0.0014358529588207603,
-      "learning_rate": 0.0001999819320942564,
-      "loss": 46.0,
-      "step": 37585
-    },
-    {
-      "epoch": 6.052981198921051,
-      "grad_norm": 0.01641983352601528,
-      "learning_rate": 0.00019998193113257432,
-      "loss": 46.0,
-      "step": 37586
-    },
-    {
-      "epoch": 6.053142235999839,
-      "grad_norm": 0.001161521882750094,
-      "learning_rate": 0.00019998193017086664,
-      "loss": 46.0,
-      "step": 37587
-    },
-    {
-      "epoch": 6.053303273078626,
-      "grad_norm": 0.00905858725309372,
-      "learning_rate": 0.0001999819292091334,
-      "loss": 46.0,
-      "step": 37588
-    },
-    {
-      "epoch": 6.0534643101574135,
-      "grad_norm": 0.004667037632316351,
-      "learning_rate": 0.00019998192824737454,
-      "loss": 46.0,
-      "step": 37589
-    },
-    {
-      "epoch": 6.053625347236201,
-      "grad_norm": 0.005883590783923864,
-      "learning_rate": 0.00019998192728559012,
-      "loss": 46.0,
-      "step": 37590
-    },
-    {
-      "epoch": 6.053786384314988,
-      "grad_norm": 0.019329693168401718,
-      "learning_rate": 0.00019998192632378006,
-      "loss": 46.0,
-      "step": 37591
-    },
-    {
-      "epoch": 6.053947421393776,
-      "grad_norm": 0.003590578678995371,
-      "learning_rate": 0.00019998192536194444,
-      "loss": 46.0,
-      "step": 37592
-    },
-    {
-      "epoch": 6.054108458472563,
-      "grad_norm": 0.0036193940322846174,
-      "learning_rate": 0.00019998192440008326,
-      "loss": 46.0,
-      "step": 37593
-    },
-    {
-      "epoch": 6.054269495551351,
-      "grad_norm": 0.002063893247395754,
-      "learning_rate": 0.00019998192343819644,
-      "loss": 46.0,
-      "step": 37594
-    },
-    {
-      "epoch": 6.054430532630138,
-      "grad_norm": 0.003849130356684327,
-      "learning_rate": 0.00019998192247628403,
-      "loss": 46.0,
-      "step": 37595
-    },
-    {
-      "epoch": 6.054591569708926,
-      "grad_norm": 0.007457686122506857,
-      "learning_rate": 0.00019998192151434607,
-      "loss": 46.0,
-      "step": 37596
-    },
-    {
-      "epoch": 6.054752606787713,
-      "grad_norm": 0.010821538977324963,
-      "learning_rate": 0.00019998192055238249,
-      "loss": 46.0,
-      "step": 37597
-    },
-    {
-      "epoch": 6.0549136438665005,
-      "grad_norm": 0.009901958517730236,
-      "learning_rate": 0.00019998191959039334,
-      "loss": 46.0,
-      "step": 37598
-    },
-    {
-      "epoch": 6.055074680945288,
-      "grad_norm": 0.004473851528018713,
-      "learning_rate": 0.0001999819186283786,
-      "loss": 46.0,
-      "step": 37599
-    },
-    {
-      "epoch": 6.0552357180240755,
-      "grad_norm": 0.0010103661334142089,
-      "learning_rate": 0.00019998191766633824,
-      "loss": 46.0,
-      "step": 37600
-    },
-    {
-      "epoch": 6.055396755102862,
-      "grad_norm": 0.00481271930038929,
-      "learning_rate": 0.00019998191670427234,
-      "loss": 46.0,
-      "step": 37601
-    },
-    {
-      "epoch": 6.0555577921816495,
-      "grad_norm": 0.005023716017603874,
-      "learning_rate": 0.0001999819157421808,
-      "loss": 46.0,
-      "step": 37602
-    },
-    {
-      "epoch": 6.055718829260437,
-      "grad_norm": 0.006178389303386211,
-      "learning_rate": 0.00019998191478006366,
-      "loss": 46.0,
-      "step": 37603
-    },
-    {
-      "epoch": 6.055879866339224,
-      "grad_norm": 0.005437683314085007,
-      "learning_rate": 0.000199981913817921,
-      "loss": 46.0,
-      "step": 37604
-    },
-    {
-      "epoch": 6.056040903418012,
-      "grad_norm": 0.011029353365302086,
-      "learning_rate": 0.0001999819128557527,
-      "loss": 46.0,
-      "step": 37605
-    },
-    {
-      "epoch": 6.056201940496799,
-      "grad_norm": 0.008159259334206581,
-      "learning_rate": 0.00019998191189355883,
-      "loss": 46.0,
-      "step": 37606
-    },
-    {
-      "epoch": 6.056362977575587,
-      "grad_norm": 0.012174089439213276,
-      "learning_rate": 0.00019998191093133932,
-      "loss": 46.0,
-      "step": 37607
-    },
-    {
-      "epoch": 6.056524014654374,
-      "grad_norm": 0.0017896260833367705,
-      "learning_rate": 0.00019998190996909428,
-      "loss": 46.0,
-      "step": 37608
-    },
-    {
-      "epoch": 6.056685051733162,
-      "grad_norm": 0.005163069348782301,
-      "learning_rate": 0.0001999819090068236,
-      "loss": 46.0,
-      "step": 37609
-    },
-    {
-      "epoch": 6.056846088811949,
-      "grad_norm": 0.004615572281181812,
-      "learning_rate": 0.00019998190804452736,
-      "loss": 46.0,
-      "step": 37610
-    },
-    {
-      "epoch": 6.0570071258907365,
-      "grad_norm": 0.010767842642962933,
-      "learning_rate": 0.00019998190708220555,
-      "loss": 46.0,
-      "step": 37611
-    },
-    {
-      "epoch": 6.057168162969524,
-      "grad_norm": 0.003302559722214937,
-      "learning_rate": 0.0001999819061198581,
-      "loss": 46.0,
-      "step": 37612
-    },
-    {
-      "epoch": 6.057329200048311,
-      "grad_norm": 0.0014210636727511883,
-      "learning_rate": 0.0001999819051574851,
-      "loss": 46.0,
-      "step": 37613
-    },
-    {
-      "epoch": 6.057490237127099,
-      "grad_norm": 0.004590318538248539,
-      "learning_rate": 0.00019998190419508652,
-      "loss": 46.0,
-      "step": 37614
-    },
-    {
-      "epoch": 6.057651274205886,
-      "grad_norm": 0.0017021981766447425,
-      "learning_rate": 0.0001999819032326623,
-      "loss": 46.0,
-      "step": 37615
-    },
-    {
-      "epoch": 6.057812311284673,
-      "grad_norm": 0.0070489319041371346,
-      "learning_rate": 0.00019998190227021252,
-      "loss": 46.0,
-      "step": 37616
-    },
-    {
-      "epoch": 6.05797334836346,
-      "grad_norm": 0.005635182838886976,
-      "learning_rate": 0.00019998190130773714,
-      "loss": 46.0,
-      "step": 37617
-    },
-    {
-      "epoch": 6.058134385442248,
-      "grad_norm": 0.011269645765423775,
-      "learning_rate": 0.00019998190034523617,
-      "loss": 46.0,
-      "step": 37618
-    },
-    {
-      "epoch": 6.058295422521035,
-      "grad_norm": 0.007693962659686804,
-      "learning_rate": 0.00019998189938270962,
-      "loss": 46.0,
-      "step": 37619
-    },
-    {
-      "epoch": 6.058456459599823,
-      "grad_norm": 0.003759726881980896,
-      "learning_rate": 0.00019998189842015748,
-      "loss": 46.0,
-      "step": 37620
-    },
-    {
-      "epoch": 6.05861749667861,
-      "grad_norm": 0.0019457685993984342,
-      "learning_rate": 0.00019998189745757975,
-      "loss": 46.0,
-      "step": 37621
-    },
-    {
-      "epoch": 6.058778533757398,
-      "grad_norm": 0.004904123488813639,
-      "learning_rate": 0.0001999818964949764,
-      "loss": 46.0,
-      "step": 37622
-    },
-    {
-      "epoch": 6.058939570836185,
-      "grad_norm": 0.006958994548767805,
-      "learning_rate": 0.0001999818955323475,
-      "loss": 46.0,
-      "step": 37623
-    },
-    {
-      "epoch": 6.0591006079149725,
-      "grad_norm": 0.0025455632712692022,
-      "learning_rate": 0.00019998189456969302,
-      "loss": 46.0,
-      "step": 37624
-    },
-    {
-      "epoch": 6.05926164499376,
-      "grad_norm": 0.004838835448026657,
-      "learning_rate": 0.00019998189360701288,
-      "loss": 46.0,
-      "step": 37625
-    },
-    {
-      "epoch": 6.059422682072547,
-      "grad_norm": 0.003193289041519165,
-      "learning_rate": 0.0001999818926443072,
-      "loss": 46.0,
-      "step": 37626
-    },
-    {
-      "epoch": 6.059583719151335,
-      "grad_norm": 0.009384668432176113,
-      "learning_rate": 0.00019998189168157592,
-      "loss": 46.0,
-      "step": 37627
-    },
-    {
-      "epoch": 6.059744756230122,
-      "grad_norm": 0.001866149134002626,
-      "learning_rate": 0.00019998189071881908,
-      "loss": 46.0,
-      "step": 37628
-    },
-    {
-      "epoch": 6.05990579330891,
-      "grad_norm": 0.009719938039779663,
-      "learning_rate": 0.0001999818897560366,
-      "loss": 46.0,
-      "step": 37629
-    },
-    {
-      "epoch": 6.060066830387696,
-      "grad_norm": 0.003988360054790974,
-      "learning_rate": 0.00019998188879322856,
-      "loss": 46.0,
-      "step": 37630
-    },
-    {
-      "epoch": 6.060227867466484,
-      "grad_norm": 0.0032057813368737698,
-      "learning_rate": 0.0001999818878303949,
-      "loss": 46.0,
-      "step": 37631
-    },
-    {
-      "epoch": 6.060388904545271,
-      "grad_norm": 0.0011375918984413147,
-      "learning_rate": 0.0001999818868675357,
-      "loss": 46.0,
-      "step": 37632
-    },
-    {
-      "epoch": 6.060549941624059,
-      "grad_norm": 0.004053414333611727,
-      "learning_rate": 0.0001999818859046509,
-      "loss": 46.0,
-      "step": 37633
-    },
-    {
-      "epoch": 6.060710978702846,
-      "grad_norm": 0.013996666297316551,
-      "learning_rate": 0.00019998188494174047,
-      "loss": 46.0,
-      "step": 37634
-    },
-    {
-      "epoch": 6.060872015781634,
-      "grad_norm": 0.006290460005402565,
-      "learning_rate": 0.0001999818839788045,
-      "loss": 46.0,
-      "step": 37635
-    },
-    {
-      "epoch": 6.061033052860421,
-      "grad_norm": 0.013546953909099102,
-      "learning_rate": 0.00019998188301584285,
-      "loss": 46.0,
-      "step": 37636
-    },
-    {
-      "epoch": 6.0611940899392085,
-      "grad_norm": 0.0071923574432730675,
-      "learning_rate": 0.0001999818820528557,
-      "loss": 46.0,
-      "step": 37637
-    },
-    {
-      "epoch": 6.061355127017996,
-      "grad_norm": 0.008371626026928425,
-      "learning_rate": 0.0001999818810898429,
-      "loss": 46.0,
-      "step": 37638
-    },
-    {
-      "epoch": 6.061516164096783,
-      "grad_norm": 0.0030042158905416727,
-      "learning_rate": 0.00019998188012680456,
-      "loss": 46.0,
-      "step": 37639
-    },
-    {
-      "epoch": 6.061677201175571,
-      "grad_norm": 0.009203464724123478,
-      "learning_rate": 0.00019998187916374062,
-      "loss": 46.0,
-      "step": 37640
-    },
-    {
-      "epoch": 6.061838238254358,
-      "grad_norm": 0.0011970112100243568,
-      "learning_rate": 0.00019998187820065107,
-      "loss": 46.0,
-      "step": 37641
-    },
-    {
-      "epoch": 6.061999275333146,
-      "grad_norm": 0.010034559294581413,
-      "learning_rate": 0.00019998187723753596,
-      "loss": 46.0,
-      "step": 37642
-    },
-    {
-      "epoch": 6.062160312411933,
-      "grad_norm": 0.0034387779887765646,
-      "learning_rate": 0.00019998187627439523,
-      "loss": 46.0,
-      "step": 37643
-    },
-    {
-      "epoch": 6.062321349490721,
-      "grad_norm": 0.002076078439131379,
-      "learning_rate": 0.00019998187531122892,
-      "loss": 46.0,
-      "step": 37644
-    },
-    {
-      "epoch": 6.062482386569507,
-      "grad_norm": 0.0006307397852651775,
-      "learning_rate": 0.00019998187434803701,
-      "loss": 46.0,
-      "step": 37645
-    },
-    {
-      "epoch": 6.062643423648295,
-      "grad_norm": 0.002080549718812108,
-      "learning_rate": 0.0001999818733848195,
-      "loss": 46.0,
-      "step": 37646
-    },
-    {
-      "epoch": 6.062804460727082,
-      "grad_norm": 0.006836682092398405,
-      "learning_rate": 0.00019998187242157645,
-      "loss": 46.0,
-      "step": 37647
-    },
-    {
-      "epoch": 6.06296549780587,
-      "grad_norm": 0.0017179992282763124,
-      "learning_rate": 0.00019998187145830776,
-      "loss": 46.0,
-      "step": 37648
-    },
-    {
-      "epoch": 6.063126534884657,
-      "grad_norm": 0.0013875626027584076,
-      "learning_rate": 0.0001999818704950135,
-      "loss": 46.0,
-      "step": 37649
-    },
-    {
-      "epoch": 6.0632875719634445,
-      "grad_norm": 0.0037537780590355396,
-      "learning_rate": 0.00019998186953169365,
-      "loss": 46.0,
-      "step": 37650
-    },
-    {
-      "epoch": 6.063448609042232,
-      "grad_norm": 0.0062209502793848515,
-      "learning_rate": 0.0001999818685683482,
-      "loss": 46.0,
-      "step": 37651
-    },
-    {
-      "epoch": 6.063609646121019,
-      "grad_norm": 0.014454977586865425,
-      "learning_rate": 0.00019998186760497718,
-      "loss": 46.0,
-      "step": 37652
-    },
-    {
-      "epoch": 6.063770683199807,
-      "grad_norm": 0.002053608885034919,
-      "learning_rate": 0.00019998186664158056,
-      "loss": 46.0,
-      "step": 37653
-    },
-    {
-      "epoch": 6.063931720278594,
-      "grad_norm": 0.0029052826575934887,
-      "learning_rate": 0.00019998186567815834,
-      "loss": 46.0,
-      "step": 37654
-    },
-    {
-      "epoch": 6.064092757357382,
-      "grad_norm": 0.010470651090145111,
-      "learning_rate": 0.00019998186471471054,
-      "loss": 46.0,
-      "step": 37655
-    },
-    {
-      "epoch": 6.064253794436169,
-      "grad_norm": 0.004357074853032827,
-      "learning_rate": 0.00019998186375123713,
-      "loss": 46.0,
-      "step": 37656
-    },
-    {
-      "epoch": 6.064414831514957,
-      "grad_norm": 0.02475530281662941,
-      "learning_rate": 0.00019998186278773815,
-      "loss": 46.0,
-      "step": 37657
-    },
-    {
-      "epoch": 6.064575868593744,
-      "grad_norm": 0.00314020412042737,
-      "learning_rate": 0.0001999818618242136,
-      "loss": 46.0,
-      "step": 37658
-    },
-    {
-      "epoch": 6.064736905672531,
-      "grad_norm": 0.005822691135108471,
-      "learning_rate": 0.00019998186086066342,
-      "loss": 46.0,
-      "step": 37659
-    },
-    {
-      "epoch": 6.064897942751318,
-      "grad_norm": 0.002460055286064744,
-      "learning_rate": 0.00019998185989708765,
-      "loss": 46.0,
-      "step": 37660
-    },
-    {
-      "epoch": 6.065058979830106,
-      "grad_norm": 0.0021120214369148016,
-      "learning_rate": 0.0001999818589334863,
-      "loss": 46.0,
-      "step": 37661
-    },
-    {
-      "epoch": 6.065220016908893,
-      "grad_norm": 0.007058229297399521,
-      "learning_rate": 0.0001999818579698594,
-      "loss": 46.0,
-      "step": 37662
-    },
-    {
-      "epoch": 6.0653810539876805,
-      "grad_norm": 0.008226689882576466,
-      "learning_rate": 0.00019998185700620687,
-      "loss": 46.0,
-      "step": 37663
-    },
-    {
-      "epoch": 6.065542091066468,
-      "grad_norm": 0.0018811551854014397,
-      "learning_rate": 0.00019998185604252876,
-      "loss": 46.0,
-      "step": 37664
-    },
-    {
-      "epoch": 6.065703128145255,
-      "grad_norm": 0.007397475652396679,
-      "learning_rate": 0.00019998185507882506,
-      "loss": 46.0,
-      "step": 37665
-    },
-    {
-      "epoch": 6.065864165224043,
-      "grad_norm": 0.0032358323223888874,
-      "learning_rate": 0.00019998185411509574,
-      "loss": 46.0,
-      "step": 37666
-    },
-    {
-      "epoch": 6.06602520230283,
-      "grad_norm": 0.014195479452610016,
-      "learning_rate": 0.00019998185315134087,
-      "loss": 46.0,
-      "step": 37667
-    },
-    {
-      "epoch": 6.066186239381618,
-      "grad_norm": 0.0020239283330738544,
-      "learning_rate": 0.00019998185218756038,
-      "loss": 46.0,
-      "step": 37668
-    },
-    {
-      "epoch": 6.066347276460405,
-      "grad_norm": 0.0025001477915793657,
-      "learning_rate": 0.0001999818512237543,
-      "loss": 46.0,
-      "step": 37669
-    },
-    {
-      "epoch": 6.066508313539193,
-      "grad_norm": 0.005226124078035355,
-      "learning_rate": 0.00019998185025992267,
-      "loss": 46.0,
-      "step": 37670
-    },
-    {
-      "epoch": 6.06666935061798,
-      "grad_norm": 0.008876755833625793,
-      "learning_rate": 0.0001999818492960654,
-      "loss": 46.0,
-      "step": 37671
-    },
-    {
-      "epoch": 6.0668303876967675,
-      "grad_norm": 0.0037230125162750483,
-      "learning_rate": 0.00019998184833218256,
-      "loss": 46.0,
-      "step": 37672
-    },
-    {
-      "epoch": 6.066991424775555,
-      "grad_norm": 0.009621241129934788,
-      "learning_rate": 0.00019998184736827414,
-      "loss": 46.0,
-      "step": 37673
-    },
-    {
-      "epoch": 6.0671524618543415,
-      "grad_norm": 0.000965080747846514,
-      "learning_rate": 0.00019998184640434012,
-      "loss": 46.0,
-      "step": 37674
-    },
-    {
-      "epoch": 6.067313498933129,
-      "grad_norm": 0.0017883339896798134,
-      "learning_rate": 0.00019998184544038053,
-      "loss": 46.0,
-      "step": 37675
-    },
-    {
-      "epoch": 6.067474536011916,
-      "grad_norm": 0.010907691903412342,
-      "learning_rate": 0.00019998184447639531,
-      "loss": 46.0,
-      "step": 37676
-    },
-    {
-      "epoch": 6.067635573090704,
-      "grad_norm": 0.00790534820407629,
-      "learning_rate": 0.00019998184351238454,
-      "loss": 46.0,
-      "step": 37677
-    },
-    {
-      "epoch": 6.067796610169491,
-      "grad_norm": 0.013911493122577667,
-      "learning_rate": 0.00019998184254834816,
-      "loss": 46.0,
-      "step": 37678
-    },
-    {
-      "epoch": 6.067957647248279,
-      "grad_norm": 0.0084577901288867,
-      "learning_rate": 0.00019998184158428618,
-      "loss": 46.0,
-      "step": 37679
-    },
-    {
-      "epoch": 6.068118684327066,
-      "grad_norm": 0.005075663328170776,
-      "learning_rate": 0.00019998184062019862,
-      "loss": 46.0,
-      "step": 37680
-    },
-    {
-      "epoch": 6.068279721405854,
-      "grad_norm": 0.002144649624824524,
-      "learning_rate": 0.00019998183965608547,
-      "loss": 46.0,
-      "step": 37681
-    },
-    {
-      "epoch": 6.068440758484641,
-      "grad_norm": 0.0026876521296799183,
-      "learning_rate": 0.00019998183869194674,
-      "loss": 46.0,
-      "step": 37682
-    },
-    {
-      "epoch": 6.068601795563429,
-      "grad_norm": 0.0033802534453570843,
-      "learning_rate": 0.00019998183772778242,
-      "loss": 46.0,
-      "step": 37683
-    },
-    {
-      "epoch": 6.068762832642216,
-      "grad_norm": 0.017919015139341354,
-      "learning_rate": 0.00019998183676359248,
-      "loss": 46.0,
-      "step": 37684
-    },
-    {
-      "epoch": 6.0689238697210035,
-      "grad_norm": 0.004853111691772938,
-      "learning_rate": 0.00019998183579937698,
-      "loss": 46.0,
-      "step": 37685
-    },
-    {
-      "epoch": 6.069084906799791,
-      "grad_norm": 0.011810258962213993,
-      "learning_rate": 0.00019998183483513587,
-      "loss": 46.0,
-      "step": 37686
-    },
-    {
-      "epoch": 6.069245943878578,
-      "grad_norm": 0.024830302223563194,
-      "learning_rate": 0.00019998183387086917,
-      "loss": 46.0,
-      "step": 37687
-    },
-    {
-      "epoch": 6.069406980957366,
-      "grad_norm": 0.006081919185817242,
-      "learning_rate": 0.0001999818329065769,
-      "loss": 46.0,
-      "step": 37688
-    },
-    {
-      "epoch": 6.069568018036152,
-      "grad_norm": 0.001091701677069068,
-      "learning_rate": 0.00019998183194225902,
-      "loss": 46.0,
-      "step": 37689
-    },
-    {
-      "epoch": 6.06972905511494,
-      "grad_norm": 0.0021871202625334263,
-      "learning_rate": 0.00019998183097791556,
-      "loss": 46.0,
-      "step": 37690
-    },
-    {
-      "epoch": 6.069890092193727,
-      "grad_norm": 0.0007331203669309616,
-      "learning_rate": 0.0001999818300135465,
-      "loss": 46.0,
-      "step": 37691
-    },
-    {
-      "epoch": 6.070051129272515,
-      "grad_norm": 0.0017588506452739239,
-      "learning_rate": 0.00019998182904915187,
-      "loss": 46.0,
-      "step": 37692
-    },
-    {
-      "epoch": 6.070212166351302,
-      "grad_norm": 0.007409881334751844,
-      "learning_rate": 0.00019998182808473165,
-      "loss": 46.0,
-      "step": 37693
-    },
-    {
-      "epoch": 6.07037320343009,
-      "grad_norm": 0.0028296681120991707,
-      "learning_rate": 0.00019998182712028582,
-      "loss": 46.0,
-      "step": 37694
-    },
-    {
-      "epoch": 6.070534240508877,
-      "grad_norm": 0.003882122691720724,
-      "learning_rate": 0.0001999818261558144,
-      "loss": 46.0,
-      "step": 37695
-    },
-    {
-      "epoch": 6.070695277587665,
-      "grad_norm": 0.015723485499620438,
-      "learning_rate": 0.00019998182519131739,
-      "loss": 46.0,
-      "step": 37696
-    },
-    {
-      "epoch": 6.070856314666452,
-      "grad_norm": 0.007811098359525204,
-      "learning_rate": 0.00019998182422679482,
-      "loss": 46.0,
-      "step": 37697
-    },
-    {
-      "epoch": 6.0710173517452395,
-      "grad_norm": 0.00732022151350975,
-      "learning_rate": 0.0001999818232622466,
-      "loss": 46.0,
-      "step": 37698
-    },
-    {
-      "epoch": 6.071178388824027,
-      "grad_norm": 0.005656701512634754,
-      "learning_rate": 0.00019998182229767286,
-      "loss": 46.0,
-      "step": 37699
-    },
-    {
-      "epoch": 6.071339425902814,
-      "grad_norm": 0.0034385446924716234,
-      "learning_rate": 0.00019998182133307347,
-      "loss": 46.0,
-      "step": 37700
-    },
-    {
-      "epoch": 6.071500462981602,
-      "grad_norm": 0.008008560165762901,
-      "learning_rate": 0.00019998182036844853,
-      "loss": 46.0,
-      "step": 37701
-    },
-    {
-      "epoch": 6.071661500060389,
-      "grad_norm": 0.00456809950992465,
-      "learning_rate": 0.000199981819403798,
-      "loss": 46.0,
-      "step": 37702
-    },
-    {
-      "epoch": 6.071822537139177,
-      "grad_norm": 0.007869865745306015,
-      "learning_rate": 0.00019998181843912188,
-      "loss": 46.0,
-      "step": 37703
-    },
-    {
-      "epoch": 6.071983574217963,
-      "grad_norm": 0.001352488063275814,
-      "learning_rate": 0.00019998181747442014,
-      "loss": 46.0,
-      "step": 37704
-    },
-    {
-      "epoch": 6.072144611296751,
-      "grad_norm": 0.005654243752360344,
-      "learning_rate": 0.00019998181650969282,
-      "loss": 46.0,
-      "step": 37705
-    },
-    {
-      "epoch": 6.072305648375538,
-      "grad_norm": 0.0026084836572408676,
-      "learning_rate": 0.0001999818155449399,
-      "loss": 46.0,
-      "step": 37706
-    },
-    {
-      "epoch": 6.072466685454326,
-      "grad_norm": 0.0034040119498968124,
-      "learning_rate": 0.00019998181458016144,
-      "loss": 46.0,
-      "step": 37707
-    },
-    {
-      "epoch": 6.072627722533113,
-      "grad_norm": 0.007144168950617313,
-      "learning_rate": 0.00019998181361535733,
-      "loss": 46.0,
-      "step": 37708
-    },
-    {
-      "epoch": 6.072788759611901,
-      "grad_norm": 0.0042244018986821175,
-      "learning_rate": 0.00019998181265052764,
-      "loss": 46.0,
-      "step": 37709
-    },
-    {
-      "epoch": 6.072949796690688,
-      "grad_norm": 0.0025879214517772198,
-      "learning_rate": 0.00019998181168567238,
-      "loss": 46.0,
-      "step": 37710
-    },
-    {
-      "epoch": 6.0731108337694755,
-      "grad_norm": 0.00336763565428555,
-      "learning_rate": 0.00019998181072079153,
-      "loss": 46.0,
-      "step": 37711
-    },
-    {
-      "epoch": 6.073271870848263,
-      "grad_norm": 0.008521047420799732,
-      "learning_rate": 0.00019998180975588507,
-      "loss": 46.0,
-      "step": 37712
-    },
-    {
-      "epoch": 6.07343290792705,
-      "grad_norm": 0.007442219648510218,
-      "learning_rate": 0.00019998180879095306,
-      "loss": 46.0,
-      "step": 37713
-    },
-    {
-      "epoch": 6.073593945005838,
-      "grad_norm": 0.01680023968219757,
-      "learning_rate": 0.00019998180782599542,
-      "loss": 46.0,
-      "step": 37714
-    },
-    {
-      "epoch": 6.073754982084625,
-      "grad_norm": 0.016661720350384712,
-      "learning_rate": 0.0001999818068610122,
-      "loss": 46.0,
-      "step": 37715
-    },
-    {
-      "epoch": 6.073916019163413,
-      "grad_norm": 0.003566613420844078,
-      "learning_rate": 0.0001999818058960034,
-      "loss": 46.0,
-      "step": 37716
-    },
-    {
-      "epoch": 6.0740770562422,
-      "grad_norm": 0.002581510692834854,
-      "learning_rate": 0.000199981804930969,
-      "loss": 46.0,
-      "step": 37717
-    },
-    {
-      "epoch": 6.074238093320987,
-      "grad_norm": 0.0032831125427037477,
-      "learning_rate": 0.00019998180396590902,
-      "loss": 46.0,
-      "step": 37718
-    },
-    {
-      "epoch": 6.074399130399774,
-      "grad_norm": 0.0018455300014466047,
-      "learning_rate": 0.00019998180300082342,
-      "loss": 46.0,
-      "step": 37719
-    },
-    {
-      "epoch": 6.074560167478562,
-      "grad_norm": 0.0010420371545478702,
-      "learning_rate": 0.00019998180203571227,
-      "loss": 46.0,
-      "step": 37720
-    },
-    {
-      "epoch": 6.074721204557349,
-      "grad_norm": 0.00391122093424201,
-      "learning_rate": 0.00019998180107057552,
-      "loss": 46.0,
-      "step": 37721
-    },
-    {
-      "epoch": 6.0748822416361365,
-      "grad_norm": 0.01678929477930069,
-      "learning_rate": 0.00019998180010541317,
-      "loss": 46.0,
-      "step": 37722
-    },
-    {
-      "epoch": 6.075043278714924,
-      "grad_norm": 0.0030893736984580755,
-      "learning_rate": 0.00019998179914022522,
-      "loss": 46.0,
-      "step": 37723
-    },
-    {
-      "epoch": 6.0752043157937115,
-      "grad_norm": 0.0034665707498788834,
-      "learning_rate": 0.00019998179817501172,
-      "loss": 46.0,
-      "step": 37724
-    },
-    {
-      "epoch": 6.075365352872499,
-      "grad_norm": 0.010268663056194782,
-      "learning_rate": 0.00019998179720977257,
-      "loss": 46.0,
-      "step": 37725
-    },
-    {
-      "epoch": 6.075526389951286,
-      "grad_norm": 0.006135433446615934,
-      "learning_rate": 0.00019998179624450786,
-      "loss": 46.0,
-      "step": 37726
-    },
-    {
-      "epoch": 6.075687427030074,
-      "grad_norm": 0.0077285682782530785,
-      "learning_rate": 0.00019998179527921757,
-      "loss": 46.0,
-      "step": 37727
-    },
-    {
-      "epoch": 6.075848464108861,
-      "grad_norm": 0.014268687926232815,
-      "learning_rate": 0.00019998179431390166,
-      "loss": 46.0,
-      "step": 37728
-    },
-    {
-      "epoch": 6.076009501187649,
-      "grad_norm": 0.006600626278668642,
-      "learning_rate": 0.0001999817933485602,
-      "loss": 46.0,
-      "step": 37729
-    },
-    {
-      "epoch": 6.076170538266436,
-      "grad_norm": 0.007557873148471117,
-      "learning_rate": 0.0001999817923831931,
-      "loss": 46.0,
-      "step": 37730
-    },
-    {
-      "epoch": 6.076331575345224,
-      "grad_norm": 0.003869457636028528,
-      "learning_rate": 0.00019998179141780047,
-      "loss": 46.0,
-      "step": 37731
-    },
-    {
-      "epoch": 6.076492612424011,
-      "grad_norm": 0.004075285978615284,
-      "learning_rate": 0.00019998179045238221,
-      "loss": 46.0,
-      "step": 37732
-    },
-    {
-      "epoch": 6.076653649502798,
-      "grad_norm": 0.006278911139816046,
-      "learning_rate": 0.00019998178948693837,
-      "loss": 46.0,
-      "step": 37733
-    },
-    {
-      "epoch": 6.076814686581585,
-      "grad_norm": 0.01255898829549551,
-      "learning_rate": 0.00019998178852146894,
-      "loss": 46.0,
-      "step": 37734
-    },
-    {
-      "epoch": 6.0769757236603725,
-      "grad_norm": 0.0029725776985287666,
-      "learning_rate": 0.00019998178755597392,
-      "loss": 46.0,
-      "step": 37735
-    },
-    {
-      "epoch": 6.07713676073916,
-      "grad_norm": 0.0010472597787156701,
-      "learning_rate": 0.00019998178659045332,
-      "loss": 46.0,
-      "step": 37736
-    },
-    {
-      "epoch": 6.077297797817947,
-      "grad_norm": 0.0033789826557040215,
-      "learning_rate": 0.0001999817856249071,
-      "loss": 46.0,
-      "step": 37737
-    },
-    {
-      "epoch": 6.077458834896735,
-      "grad_norm": 0.003187508787959814,
-      "learning_rate": 0.0001999817846593353,
-      "loss": 46.0,
-      "step": 37738
-    },
-    {
-      "epoch": 6.077619871975522,
-      "grad_norm": 0.010816927067935467,
-      "learning_rate": 0.00019998178369373792,
-      "loss": 46.0,
-      "step": 37739
-    },
-    {
-      "epoch": 6.07778090905431,
-      "grad_norm": 0.002791301580145955,
-      "learning_rate": 0.00019998178272811497,
-      "loss": 46.0,
-      "step": 37740
-    },
-    {
-      "epoch": 6.077941946133097,
-      "grad_norm": 0.005094537511467934,
-      "learning_rate": 0.0001999817817624664,
-      "loss": 46.0,
-      "step": 37741
-    },
-    {
-      "epoch": 6.078102983211885,
-      "grad_norm": 0.006306060589849949,
-      "learning_rate": 0.00019998178079679227,
-      "loss": 46.0,
-      "step": 37742
-    },
-    {
-      "epoch": 6.078264020290672,
-      "grad_norm": 0.0030474953819066286,
-      "learning_rate": 0.00019998177983109253,
-      "loss": 46.0,
-      "step": 37743
-    },
-    {
-      "epoch": 6.07842505736946,
-      "grad_norm": 0.0017909746384248137,
-      "learning_rate": 0.00019998177886536718,
-      "loss": 46.0,
-      "step": 37744
-    },
-    {
-      "epoch": 6.078586094448247,
-      "grad_norm": 0.008043472655117512,
-      "learning_rate": 0.00019998177789961626,
-      "loss": 46.0,
-      "step": 37745
-    },
-    {
-      "epoch": 6.0787471315270345,
-      "grad_norm": 0.005664821248501539,
-      "learning_rate": 0.00019998177693383973,
-      "loss": 46.0,
-      "step": 37746
-    },
-    {
-      "epoch": 6.078908168605821,
-      "grad_norm": 0.001391304307617247,
-      "learning_rate": 0.0001999817759680376,
-      "loss": 46.0,
-      "step": 37747
-    },
-    {
-      "epoch": 6.0790692056846085,
-      "grad_norm": 0.0021558445878326893,
-      "learning_rate": 0.00019998177500220993,
-      "loss": 46.0,
-      "step": 37748
-    },
-    {
-      "epoch": 6.079230242763396,
-      "grad_norm": 0.0063851396553218365,
-      "learning_rate": 0.00019998177403635664,
-      "loss": 46.0,
-      "step": 37749
-    },
-    {
-      "epoch": 6.079391279842183,
-      "grad_norm": 0.010128156282007694,
-      "learning_rate": 0.00019998177307047779,
-      "loss": 46.0,
-      "step": 37750
-    },
-    {
-      "epoch": 6.079552316920971,
-      "grad_norm": 0.005564019083976746,
-      "learning_rate": 0.00019998177210457332,
-      "loss": 46.0,
-      "step": 37751
-    },
-    {
-      "epoch": 6.079713353999758,
-      "grad_norm": 0.0013689086772501469,
-      "learning_rate": 0.00019998177113864327,
-      "loss": 46.0,
-      "step": 37752
-    },
-    {
-      "epoch": 6.079874391078546,
-      "grad_norm": 0.010868123732507229,
-      "learning_rate": 0.00019998177017268762,
-      "loss": 46.0,
-      "step": 37753
-    },
-    {
-      "epoch": 6.080035428157333,
-      "grad_norm": 0.00566981453448534,
-      "learning_rate": 0.0001999817692067064,
-      "loss": 46.0,
-      "step": 37754
-    },
-    {
-      "epoch": 6.080196465236121,
-      "grad_norm": 0.00451606186106801,
-      "learning_rate": 0.00019998176824069955,
-      "loss": 46.0,
-      "step": 37755
-    },
-    {
-      "epoch": 6.080357502314908,
-      "grad_norm": 0.0017864335095509887,
-      "learning_rate": 0.00019998176727466715,
-      "loss": 46.0,
-      "step": 37756
-    },
-    {
-      "epoch": 6.080518539393696,
-      "grad_norm": 0.00178652280010283,
-      "learning_rate": 0.00019998176630860913,
-      "loss": 46.0,
-      "step": 37757
-    },
-    {
-      "epoch": 6.080679576472483,
-      "grad_norm": 0.0012568294769153,
-      "learning_rate": 0.00019998176534252556,
-      "loss": 46.0,
-      "step": 37758
-    },
-    {
-      "epoch": 6.0808406135512705,
-      "grad_norm": 0.0038692532107234,
-      "learning_rate": 0.00019998176437641636,
-      "loss": 46.0,
-      "step": 37759
-    },
-    {
-      "epoch": 6.081001650630058,
-      "grad_norm": 0.01501297764480114,
-      "learning_rate": 0.00019998176341028159,
-      "loss": 46.0,
-      "step": 37760
-    },
-    {
-      "epoch": 6.081162687708845,
-      "grad_norm": 0.0016528075793758035,
-      "learning_rate": 0.00019998176244412122,
-      "loss": 46.0,
-      "step": 37761
-    },
-    {
-      "epoch": 6.081323724787632,
-      "grad_norm": 0.015143604017794132,
-      "learning_rate": 0.00019998176147793527,
-      "loss": 46.0,
-      "step": 37762
-    },
-    {
-      "epoch": 6.081484761866419,
-      "grad_norm": 0.004088354762643576,
-      "learning_rate": 0.0001999817605117237,
-      "loss": 46.0,
-      "step": 37763
-    },
-    {
-      "epoch": 6.081645798945207,
-      "grad_norm": 0.0016328999772667885,
-      "learning_rate": 0.00019998175954548657,
-      "loss": 46.0,
-      "step": 37764
-    },
-    {
-      "epoch": 6.081806836023994,
-      "grad_norm": 0.001362946117296815,
-      "learning_rate": 0.00019998175857922386,
-      "loss": 46.0,
-      "step": 37765
-    },
-    {
-      "epoch": 6.081967873102782,
-      "grad_norm": 0.005000079050660133,
-      "learning_rate": 0.0001999817576129355,
-      "loss": 46.0,
-      "step": 37766
-    },
-    {
-      "epoch": 6.082128910181569,
-      "grad_norm": 0.0035223346203565598,
-      "learning_rate": 0.00019998175664662161,
-      "loss": 46.0,
-      "step": 37767
-    },
-    {
-      "epoch": 6.082289947260357,
-      "grad_norm": 0.002147382590919733,
-      "learning_rate": 0.0001999817556802821,
-      "loss": 46.0,
-      "step": 37768
-    },
-    {
-      "epoch": 6.082450984339144,
-      "grad_norm": 0.0032049333676695824,
-      "learning_rate": 0.00019998175471391702,
-      "loss": 46.0,
-      "step": 37769
-    },
-    {
-      "epoch": 6.082612021417932,
-      "grad_norm": 0.002670903457328677,
-      "learning_rate": 0.00019998175374752637,
-      "loss": 46.0,
-      "step": 37770
-    },
-    {
-      "epoch": 6.082773058496719,
-      "grad_norm": 0.002350566443055868,
-      "learning_rate": 0.00019998175278111008,
-      "loss": 46.0,
-      "step": 37771
-    },
-    {
-      "epoch": 6.0829340955755065,
-      "grad_norm": 0.010421018116176128,
-      "learning_rate": 0.00019998175181466823,
-      "loss": 46.0,
-      "step": 37772
-    },
-    {
-      "epoch": 6.083095132654294,
-      "grad_norm": 0.009646703489124775,
-      "learning_rate": 0.0001999817508482008,
-      "loss": 46.0,
-      "step": 37773
-    },
-    {
-      "epoch": 6.083256169733081,
-      "grad_norm": 0.0058188787661492825,
-      "learning_rate": 0.00019998174988170774,
-      "loss": 46.0,
-      "step": 37774
-    },
-    {
-      "epoch": 6.083417206811869,
-      "grad_norm": 0.007185832131654024,
-      "learning_rate": 0.0001999817489151891,
-      "loss": 46.0,
-      "step": 37775
-    },
-    {
-      "epoch": 6.083578243890656,
-      "grad_norm": 0.0030367723666131496,
-      "learning_rate": 0.0001999817479486449,
-      "loss": 46.0,
-      "step": 37776
-    },
-    {
-      "epoch": 6.083739280969443,
-      "grad_norm": 0.008109228685498238,
-      "learning_rate": 0.00019998174698207508,
-      "loss": 46.0,
-      "step": 37777
-    },
-    {
-      "epoch": 6.08390031804823,
-      "grad_norm": 0.002837846055626869,
-      "learning_rate": 0.00019998174601547968,
-      "loss": 46.0,
-      "step": 37778
-    },
-    {
-      "epoch": 6.084061355127018,
-      "grad_norm": 0.0011027230648323894,
-      "learning_rate": 0.0001999817450488587,
-      "loss": 46.0,
-      "step": 37779
-    },
-    {
-      "epoch": 6.084222392205805,
-      "grad_norm": 0.00266228336840868,
-      "learning_rate": 0.0001999817440822121,
-      "loss": 46.0,
-      "step": 37780
-    },
-    {
-      "epoch": 6.084383429284593,
-      "grad_norm": 0.0011304491199553013,
-      "learning_rate": 0.00019998174311553995,
-      "loss": 46.0,
-      "step": 37781
-    },
-    {
-      "epoch": 6.08454446636338,
-      "grad_norm": 0.0015038179699331522,
-      "learning_rate": 0.00019998174214884217,
-      "loss": 46.0,
-      "step": 37782
-    },
-    {
-      "epoch": 6.0847055034421675,
-      "grad_norm": 0.006239565555006266,
-      "learning_rate": 0.00019998174118211883,
-      "loss": 46.0,
-      "step": 37783
-    },
-    {
-      "epoch": 6.084866540520955,
-      "grad_norm": 0.005426181945949793,
-      "learning_rate": 0.00019998174021536988,
-      "loss": 46.0,
-      "step": 37784
-    },
-    {
-      "epoch": 6.085027577599742,
-      "grad_norm": 0.011212630197405815,
-      "learning_rate": 0.00019998173924859534,
-      "loss": 46.0,
-      "step": 37785
-    },
-    {
-      "epoch": 6.08518861467853,
-      "grad_norm": 0.007488547824323177,
-      "learning_rate": 0.00019998173828179522,
-      "loss": 46.0,
-      "step": 37786
-    },
-    {
-      "epoch": 6.085349651757317,
-      "grad_norm": 0.0012200287310406566,
-      "learning_rate": 0.00019998173731496953,
-      "loss": 46.0,
-      "step": 37787
-    },
-    {
-      "epoch": 6.085510688836105,
-      "grad_norm": 0.0040877568535506725,
-      "learning_rate": 0.00019998173634811823,
-      "loss": 46.0,
-      "step": 37788
-    },
-    {
-      "epoch": 6.085671725914892,
-      "grad_norm": 0.0018878281116485596,
-      "learning_rate": 0.00019998173538124134,
-      "loss": 46.0,
-      "step": 37789
-    },
-    {
-      "epoch": 6.08583276299368,
-      "grad_norm": 0.00496339239180088,
-      "learning_rate": 0.00019998173441433884,
-      "loss": 46.0,
-      "step": 37790
-    },
-    {
-      "epoch": 6.085993800072466,
-      "grad_norm": 0.0024393887724727392,
-      "learning_rate": 0.00019998173344741078,
-      "loss": 46.0,
-      "step": 37791
-    },
-    {
-      "epoch": 6.086154837151254,
-      "grad_norm": 0.008046883158385754,
-      "learning_rate": 0.0001999817324804571,
-      "loss": 46.0,
-      "step": 37792
-    },
-    {
-      "epoch": 6.086315874230041,
-      "grad_norm": 0.0015929220244288445,
-      "learning_rate": 0.00019998173151347787,
-      "loss": 46.0,
-      "step": 37793
-    },
-    {
-      "epoch": 6.086476911308829,
-      "grad_norm": 0.017275795340538025,
-      "learning_rate": 0.00019998173054647301,
-      "loss": 46.0,
-      "step": 37794
-    },
-    {
-      "epoch": 6.086637948387616,
-      "grad_norm": 0.002468364778906107,
-      "learning_rate": 0.00019998172957944258,
-      "loss": 46.0,
-      "step": 37795
-    },
-    {
-      "epoch": 6.0867989854664035,
-      "grad_norm": 0.006003603804856539,
-      "learning_rate": 0.00019998172861238655,
-      "loss": 46.0,
-      "step": 37796
-    },
-    {
-      "epoch": 6.086960022545191,
-      "grad_norm": 0.0015118923038244247,
-      "learning_rate": 0.00019998172764530497,
-      "loss": 46.0,
-      "step": 37797
-    },
-    {
-      "epoch": 6.087121059623978,
-      "grad_norm": 0.004045466426759958,
-      "learning_rate": 0.00019998172667819774,
-      "loss": 46.0,
-      "step": 37798
-    },
-    {
-      "epoch": 6.087282096702766,
-      "grad_norm": 0.010266249068081379,
-      "learning_rate": 0.00019998172571106495,
-      "loss": 46.0,
-      "step": 37799
-    },
-    {
-      "epoch": 6.087443133781553,
-      "grad_norm": 0.0012398279504850507,
-      "learning_rate": 0.00019998172474390655,
-      "loss": 46.0,
-      "step": 37800
-    },
-    {
-      "epoch": 6.087604170860341,
-      "grad_norm": 0.0013336322735995054,
-      "learning_rate": 0.00019998172377672256,
-      "loss": 46.0,
-      "step": 37801
-    },
-    {
-      "epoch": 6.087765207939128,
-      "grad_norm": 0.00229376507923007,
-      "learning_rate": 0.00019998172280951302,
-      "loss": 46.0,
-      "step": 37802
-    },
-    {
-      "epoch": 6.087926245017916,
-      "grad_norm": 0.0028291502967476845,
-      "learning_rate": 0.00019998172184227788,
-      "loss": 46.0,
-      "step": 37803
-    },
-    {
-      "epoch": 6.088087282096703,
-      "grad_norm": 0.010053534992039204,
-      "learning_rate": 0.00019998172087501713,
-      "loss": 46.0,
-      "step": 37804
-    },
-    {
-      "epoch": 6.088248319175491,
-      "grad_norm": 0.0007450601551681757,
-      "learning_rate": 0.00019998171990773077,
-      "loss": 46.0,
-      "step": 37805
-    },
-    {
-      "epoch": 6.088409356254277,
-      "grad_norm": 0.0018764289561659098,
-      "learning_rate": 0.00019998171894041887,
-      "loss": 46.0,
-      "step": 37806
-    },
-    {
-      "epoch": 6.088570393333065,
-      "grad_norm": 0.0032143276184797287,
-      "learning_rate": 0.00019998171797308136,
-      "loss": 46.0,
-      "step": 37807
-    },
-    {
-      "epoch": 6.088731430411852,
-      "grad_norm": 0.0026210385840386152,
-      "learning_rate": 0.00019998171700571823,
-      "loss": 46.0,
-      "step": 37808
-    },
-    {
-      "epoch": 6.0888924674906395,
-      "grad_norm": 0.001016295631416142,
-      "learning_rate": 0.00019998171603832955,
-      "loss": 46.0,
-      "step": 37809
-    },
-    {
-      "epoch": 6.089053504569427,
-      "grad_norm": 0.0027335411868989468,
-      "learning_rate": 0.00019998171507091525,
-      "loss": 46.0,
-      "step": 37810
-    },
-    {
-      "epoch": 6.089214541648214,
-      "grad_norm": 0.004869597498327494,
-      "learning_rate": 0.0001999817141034754,
-      "loss": 46.0,
-      "step": 37811
-    },
-    {
-      "epoch": 6.089375578727002,
-      "grad_norm": 0.0014778736513108015,
-      "learning_rate": 0.00019998171313600992,
-      "loss": 46.0,
-      "step": 37812
-    },
-    {
-      "epoch": 6.089536615805789,
-      "grad_norm": 0.0025663163978606462,
-      "learning_rate": 0.00019998171216851885,
-      "loss": 46.0,
-      "step": 37813
-    },
-    {
-      "epoch": 6.089697652884577,
-      "grad_norm": 0.00807054340839386,
-      "learning_rate": 0.00019998171120100223,
-      "loss": 46.0,
-      "step": 37814
-    },
-    {
-      "epoch": 6.089858689963364,
-      "grad_norm": 0.0015795142389833927,
-      "learning_rate": 0.00019998171023345997,
-      "loss": 46.0,
-      "step": 37815
-    },
-    {
-      "epoch": 6.090019727042152,
-      "grad_norm": 0.0025860497262328863,
-      "learning_rate": 0.00019998170926589215,
-      "loss": 46.0,
-      "step": 37816
-    },
-    {
-      "epoch": 6.090180764120939,
-      "grad_norm": 0.001016231719404459,
-      "learning_rate": 0.00019998170829829874,
-      "loss": 46.0,
-      "step": 37817
-    },
-    {
-      "epoch": 6.090341801199727,
-      "grad_norm": 0.003120674518868327,
-      "learning_rate": 0.00019998170733067974,
-      "loss": 46.0,
-      "step": 37818
-    },
-    {
-      "epoch": 6.090502838278514,
-      "grad_norm": 0.009864549152553082,
-      "learning_rate": 0.00019998170636303513,
-      "loss": 46.0,
-      "step": 37819
-    },
-    {
-      "epoch": 6.090663875357301,
-      "grad_norm": 0.0027071491349488497,
-      "learning_rate": 0.00019998170539536493,
-      "loss": 46.0,
-      "step": 37820
-    },
-    {
-      "epoch": 6.090824912436088,
-      "grad_norm": 0.003821437945589423,
-      "learning_rate": 0.00019998170442766915,
-      "loss": 46.0,
-      "step": 37821
-    },
-    {
-      "epoch": 6.0909859495148755,
-      "grad_norm": 0.005640721879899502,
-      "learning_rate": 0.0001999817034599478,
-      "loss": 46.0,
-      "step": 37822
-    },
-    {
-      "epoch": 6.091146986593663,
-      "grad_norm": 0.008255884982645512,
-      "learning_rate": 0.00019998170249220084,
-      "loss": 46.0,
-      "step": 37823
-    },
-    {
-      "epoch": 6.09130802367245,
-      "grad_norm": 0.0023857494816184044,
-      "learning_rate": 0.0001999817015244283,
-      "loss": 46.0,
-      "step": 37824
-    },
-    {
-      "epoch": 6.091469060751238,
-      "grad_norm": 0.0019795005209743977,
-      "learning_rate": 0.00019998170055663016,
-      "loss": 46.0,
-      "step": 37825
-    },
-    {
-      "epoch": 6.091630097830025,
-      "grad_norm": 0.009732520207762718,
-      "learning_rate": 0.0001999816995888064,
-      "loss": 46.0,
-      "step": 37826
-    },
-    {
-      "epoch": 6.091791134908813,
-      "grad_norm": 0.007448509335517883,
-      "learning_rate": 0.0001999816986209571,
-      "loss": 46.0,
-      "step": 37827
-    },
-    {
-      "epoch": 6.0919521719876,
-      "grad_norm": 0.001298447954468429,
-      "learning_rate": 0.00019998169765308218,
-      "loss": 46.0,
-      "step": 37828
-    },
-    {
-      "epoch": 6.092113209066388,
-      "grad_norm": 0.006628180388361216,
-      "learning_rate": 0.0001999816966851817,
-      "loss": 46.0,
-      "step": 37829
-    },
-    {
-      "epoch": 6.092274246145175,
-      "grad_norm": 0.001536199008114636,
-      "learning_rate": 0.0001999816957172556,
-      "loss": 46.0,
-      "step": 37830
-    },
-    {
-      "epoch": 6.0924352832239625,
-      "grad_norm": 0.00138020608574152,
-      "learning_rate": 0.00019998169474930394,
-      "loss": 46.0,
-      "step": 37831
-    },
-    {
-      "epoch": 6.09259632030275,
-      "grad_norm": 0.006614008452743292,
-      "learning_rate": 0.00019998169378132664,
-      "loss": 46.0,
-      "step": 37832
-    },
-    {
-      "epoch": 6.0927573573815375,
-      "grad_norm": 0.003222718834877014,
-      "learning_rate": 0.0001999816928133238,
-      "loss": 46.0,
-      "step": 37833
-    },
-    {
-      "epoch": 6.092918394460325,
-      "grad_norm": 0.008102389052510262,
-      "learning_rate": 0.00019998169184529534,
-      "loss": 46.0,
-      "step": 37834
-    },
-    {
-      "epoch": 6.0930794315391115,
-      "grad_norm": 0.005741816014051437,
-      "learning_rate": 0.0001999816908772413,
-      "loss": 46.0,
-      "step": 37835
-    },
-    {
-      "epoch": 6.093240468617899,
-      "grad_norm": 0.005830100737512112,
-      "learning_rate": 0.00019998168990916168,
-      "loss": 46.0,
-      "step": 37836
-    },
-    {
-      "epoch": 6.093401505696686,
-      "grad_norm": 0.002829157281666994,
-      "learning_rate": 0.00019998168894105645,
-      "loss": 46.0,
-      "step": 37837
-    },
-    {
-      "epoch": 6.093562542775474,
-      "grad_norm": 0.004213452804833651,
-      "learning_rate": 0.00019998168797292563,
-      "loss": 46.0,
-      "step": 37838
-    },
-    {
-      "epoch": 6.093723579854261,
-      "grad_norm": 0.002485365839675069,
-      "learning_rate": 0.00019998168700476922,
-      "loss": 46.0,
-      "step": 37839
-    },
-    {
-      "epoch": 6.093884616933049,
-      "grad_norm": 0.0038056441117078066,
-      "learning_rate": 0.00019998168603658725,
-      "loss": 46.0,
-      "step": 37840
-    },
-    {
-      "epoch": 6.094045654011836,
-      "grad_norm": 0.00824055913835764,
-      "learning_rate": 0.00019998168506837966,
-      "loss": 46.0,
-      "step": 37841
-    },
-    {
-      "epoch": 6.094206691090624,
-      "grad_norm": 0.0017792948056012392,
-      "learning_rate": 0.0001999816841001465,
-      "loss": 46.0,
-      "step": 37842
-    },
-    {
-      "epoch": 6.094367728169411,
-      "grad_norm": 0.004388670902699232,
-      "learning_rate": 0.00019998168313188774,
-      "loss": 46.0,
-      "step": 37843
-    },
-    {
-      "epoch": 6.0945287652481985,
-      "grad_norm": 0.017673298716545105,
-      "learning_rate": 0.0001999816821636034,
-      "loss": 46.0,
-      "step": 37844
-    },
-    {
-      "epoch": 6.094689802326986,
-      "grad_norm": 0.0008767392719164491,
-      "learning_rate": 0.00019998168119529343,
-      "loss": 46.0,
-      "step": 37845
-    },
-    {
-      "epoch": 6.094850839405773,
-      "grad_norm": 0.002628319663926959,
-      "learning_rate": 0.00019998168022695789,
-      "loss": 46.0,
-      "step": 37846
-    },
-    {
-      "epoch": 6.095011876484561,
-      "grad_norm": 0.003136137267574668,
-      "learning_rate": 0.00019998167925859675,
-      "loss": 46.0,
-      "step": 37847
-    },
-    {
-      "epoch": 6.095172913563348,
-      "grad_norm": 0.007929238490760326,
-      "learning_rate": 0.00019998167829021006,
-      "loss": 46.0,
-      "step": 37848
-    },
-    {
-      "epoch": 6.095333950642136,
-      "grad_norm": 0.0013684015721082687,
-      "learning_rate": 0.00019998167732179778,
-      "loss": 46.0,
-      "step": 37849
-    },
-    {
-      "epoch": 6.095494987720922,
-      "grad_norm": 0.007586033549159765,
-      "learning_rate": 0.00019998167635335988,
-      "loss": 46.0,
-      "step": 37850
-    },
-    {
-      "epoch": 6.09565602479971,
-      "grad_norm": 0.007154947612434626,
-      "learning_rate": 0.00019998167538489637,
-      "loss": 46.0,
-      "step": 37851
-    },
-    {
-      "epoch": 6.095817061878497,
-      "grad_norm": 0.0011864241678267717,
-      "learning_rate": 0.0001999816744164073,
-      "loss": 46.0,
-      "step": 37852
-    },
-    {
-      "epoch": 6.095978098957285,
-      "grad_norm": 0.005761478561908007,
-      "learning_rate": 0.00019998167344789265,
-      "loss": 46.0,
-      "step": 37853
-    },
-    {
-      "epoch": 6.096139136036072,
-      "grad_norm": 0.029655126854777336,
-      "learning_rate": 0.0001999816724793524,
-      "loss": 46.0,
-      "step": 37854
-    },
-    {
-      "epoch": 6.09630017311486,
-      "grad_norm": 0.0015494056278839707,
-      "learning_rate": 0.00019998167151078655,
-      "loss": 46.0,
-      "step": 37855
-    },
-    {
-      "epoch": 6.096461210193647,
-      "grad_norm": 0.004338191822171211,
-      "learning_rate": 0.00019998167054219513,
-      "loss": 46.0,
-      "step": 37856
-    },
-    {
-      "epoch": 6.0966222472724345,
-      "grad_norm": 0.001258274307474494,
-      "learning_rate": 0.0001999816695735781,
-      "loss": 46.0,
-      "step": 37857
-    },
-    {
-      "epoch": 6.096783284351222,
-      "grad_norm": 0.0016012617852538824,
-      "learning_rate": 0.00019998166860493548,
-      "loss": 46.0,
-      "step": 37858
-    },
-    {
-      "epoch": 6.096944321430009,
-      "grad_norm": 0.0020765161607414484,
-      "learning_rate": 0.00019998166763626727,
-      "loss": 46.0,
-      "step": 37859
-    },
-    {
-      "epoch": 6.097105358508797,
-      "grad_norm": 0.0010396366706117988,
-      "learning_rate": 0.00019998166666757348,
-      "loss": 46.0,
-      "step": 37860
-    },
-    {
-      "epoch": 6.097266395587584,
-      "grad_norm": 0.0017461135284975171,
-      "learning_rate": 0.0001999816656988541,
-      "loss": 46.0,
-      "step": 37861
-    },
-    {
-      "epoch": 6.097427432666372,
-      "grad_norm": 0.0015936428681015968,
-      "learning_rate": 0.0001999816647301091,
-      "loss": 46.0,
-      "step": 37862
-    },
-    {
-      "epoch": 6.097588469745159,
-      "grad_norm": 0.007107043173164129,
-      "learning_rate": 0.00019998166376133854,
-      "loss": 46.0,
-      "step": 37863
-    },
-    {
-      "epoch": 6.097749506823946,
-      "grad_norm": 0.009870864450931549,
-      "learning_rate": 0.00019998166279254238,
-      "loss": 46.0,
-      "step": 37864
-    },
-    {
-      "epoch": 6.097910543902733,
-      "grad_norm": 0.00346820754930377,
-      "learning_rate": 0.00019998166182372065,
-      "loss": 46.0,
-      "step": 37865
-    },
-    {
-      "epoch": 6.098071580981521,
-      "grad_norm": 0.0031371782533824444,
-      "learning_rate": 0.0001999816608548733,
-      "loss": 46.0,
-      "step": 37866
-    },
-    {
-      "epoch": 6.098232618060308,
-      "grad_norm": 0.013296880759298801,
-      "learning_rate": 0.0001999816598860004,
-      "loss": 46.0,
-      "step": 37867
-    },
-    {
-      "epoch": 6.098393655139096,
-      "grad_norm": 0.005946441553533077,
-      "learning_rate": 0.00019998165891710188,
-      "loss": 46.0,
-      "step": 37868
-    },
-    {
-      "epoch": 6.098554692217883,
-      "grad_norm": 0.011022618040442467,
-      "learning_rate": 0.00019998165794817775,
-      "loss": 46.0,
-      "step": 37869
-    },
-    {
-      "epoch": 6.0987157292966705,
-      "grad_norm": 0.0025238527450710535,
-      "learning_rate": 0.00019998165697922805,
-      "loss": 46.0,
-      "step": 37870
-    },
-    {
-      "epoch": 6.098876766375458,
-      "grad_norm": 0.0042534396052360535,
-      "learning_rate": 0.00019998165601025275,
-      "loss": 46.0,
-      "step": 37871
-    },
-    {
-      "epoch": 6.099037803454245,
-      "grad_norm": 0.0019194354536011815,
-      "learning_rate": 0.0001999816550412519,
-      "loss": 46.0,
-      "step": 37872
-    },
-    {
-      "epoch": 6.099198840533033,
-      "grad_norm": 0.006690456531941891,
-      "learning_rate": 0.00019998165407222543,
-      "loss": 46.0,
-      "step": 37873
-    },
-    {
-      "epoch": 6.09935987761182,
-      "grad_norm": 0.0009905594633892179,
-      "learning_rate": 0.00019998165310317338,
-      "loss": 46.0,
-      "step": 37874
-    },
-    {
-      "epoch": 6.099520914690608,
-      "grad_norm": 0.003276173258200288,
-      "learning_rate": 0.00019998165213409573,
-      "loss": 46.0,
-      "step": 37875
-    },
-    {
-      "epoch": 6.099681951769395,
-      "grad_norm": 0.0015897850971668959,
-      "learning_rate": 0.0001999816511649925,
-      "loss": 46.0,
-      "step": 37876
-    },
-    {
-      "epoch": 6.099842988848183,
-      "grad_norm": 0.00707613630220294,
-      "learning_rate": 0.00019998165019586366,
-      "loss": 46.0,
-      "step": 37877
-    },
-    {
-      "epoch": 6.10000402592697,
-      "grad_norm": 0.0074333129450678825,
-      "learning_rate": 0.00019998164922670927,
-      "loss": 46.0,
-      "step": 37878
-    },
-    {
-      "epoch": 6.100165063005757,
-      "grad_norm": 0.008302539587020874,
-      "learning_rate": 0.00019998164825752924,
-      "loss": 46.0,
-      "step": 37879
-    },
-    {
-      "epoch": 6.100326100084544,
-      "grad_norm": 0.001487349858507514,
-      "learning_rate": 0.00019998164728832362,
-      "loss": 46.0,
-      "step": 37880
-    },
-    {
-      "epoch": 6.100487137163332,
-      "grad_norm": 0.006328992545604706,
-      "learning_rate": 0.00019998164631909244,
-      "loss": 46.0,
-      "step": 37881
-    },
-    {
-      "epoch": 6.100648174242119,
-      "grad_norm": 0.005400301888585091,
-      "learning_rate": 0.00019998164534983567,
-      "loss": 46.0,
-      "step": 37882
-    },
-    {
-      "epoch": 6.1008092113209065,
-      "grad_norm": 0.008416292257606983,
-      "learning_rate": 0.0001999816443805533,
-      "loss": 46.0,
-      "step": 37883
-    },
-    {
-      "epoch": 6.100970248399694,
-      "grad_norm": 0.0014171391958370805,
-      "learning_rate": 0.00019998164341124533,
-      "loss": 46.0,
-      "step": 37884
-    },
-    {
-      "epoch": 6.101131285478481,
-      "grad_norm": 0.01370865199714899,
-      "learning_rate": 0.00019998164244191177,
-      "loss": 46.0,
-      "step": 37885
-    },
-    {
-      "epoch": 6.101292322557269,
-      "grad_norm": 0.0027628070674836636,
-      "learning_rate": 0.00019998164147255266,
-      "loss": 46.0,
-      "step": 37886
-    },
-    {
-      "epoch": 6.101453359636056,
-      "grad_norm": 0.008990590460598469,
-      "learning_rate": 0.00019998164050316793,
-      "loss": 46.0,
-      "step": 37887
-    },
-    {
-      "epoch": 6.101614396714844,
-      "grad_norm": 0.011891982518136501,
-      "learning_rate": 0.0001999816395337576,
-      "loss": 46.0,
-      "step": 37888
-    },
-    {
-      "epoch": 6.101775433793631,
-      "grad_norm": 0.017663370817899704,
-      "learning_rate": 0.00019998163856432168,
-      "loss": 46.0,
-      "step": 37889
-    },
-    {
-      "epoch": 6.101936470872419,
-      "grad_norm": 0.002410690300166607,
-      "learning_rate": 0.00019998163759486016,
-      "loss": 46.0,
-      "step": 37890
-    },
-    {
-      "epoch": 6.102097507951206,
-      "grad_norm": 0.0016873523127287626,
-      "learning_rate": 0.00019998163662537311,
-      "loss": 46.0,
-      "step": 37891
-    },
-    {
-      "epoch": 6.1022585450299935,
-      "grad_norm": 0.005430019460618496,
-      "learning_rate": 0.00019998163565586042,
-      "loss": 46.0,
-      "step": 37892
-    },
-    {
-      "epoch": 6.10241958210878,
-      "grad_norm": 0.005881018005311489,
-      "learning_rate": 0.00019998163468632214,
-      "loss": 46.0,
-      "step": 37893
-    },
-    {
-      "epoch": 6.102580619187568,
-      "grad_norm": 0.00956530962139368,
-      "learning_rate": 0.00019998163371675828,
-      "loss": 46.0,
-      "step": 37894
-    },
-    {
-      "epoch": 6.102741656266355,
-      "grad_norm": 0.019478866830468178,
-      "learning_rate": 0.00019998163274716882,
-      "loss": 46.0,
-      "step": 37895
-    },
-    {
-      "epoch": 6.1029026933451425,
-      "grad_norm": 0.0012692884774878621,
-      "learning_rate": 0.0001999816317775538,
-      "loss": 46.0,
-      "step": 37896
-    },
-    {
-      "epoch": 6.10306373042393,
-      "grad_norm": 0.01564978063106537,
-      "learning_rate": 0.00019998163080791316,
-      "loss": 46.0,
-      "step": 37897
-    },
-    {
-      "epoch": 6.103224767502717,
-      "grad_norm": 0.003363940631970763,
-      "learning_rate": 0.00019998162983824694,
-      "loss": 46.0,
-      "step": 37898
-    },
-    {
-      "epoch": 6.103385804581505,
-      "grad_norm": 0.007944186218082905,
-      "learning_rate": 0.0001999816288685551,
-      "loss": 46.0,
-      "step": 37899
-    },
-    {
-      "epoch": 6.103546841660292,
-      "grad_norm": 0.013608613982796669,
-      "learning_rate": 0.00019998162789883772,
-      "loss": 46.0,
-      "step": 37900
-    },
-    {
-      "epoch": 6.10370787873908,
-      "grad_norm": 0.0018901459407061338,
-      "learning_rate": 0.00019998162692909475,
-      "loss": 46.0,
-      "step": 37901
-    },
-    {
-      "epoch": 6.103868915817867,
-      "grad_norm": 0.002593145938590169,
-      "learning_rate": 0.00019998162595932616,
-      "loss": 46.0,
-      "step": 37902
-    },
-    {
-      "epoch": 6.104029952896655,
-      "grad_norm": 0.006367926951497793,
-      "learning_rate": 0.00019998162498953198,
-      "loss": 46.0,
-      "step": 37903
-    },
-    {
-      "epoch": 6.104190989975442,
-      "grad_norm": 0.005814204923808575,
-      "learning_rate": 0.00019998162401971221,
-      "loss": 46.0,
-      "step": 37904
-    },
-    {
-      "epoch": 6.1043520270542295,
-      "grad_norm": 0.009547271765768528,
-      "learning_rate": 0.00019998162304986686,
-      "loss": 46.0,
-      "step": 37905
-    },
-    {
-      "epoch": 6.104513064133017,
-      "grad_norm": 0.004241904243826866,
-      "learning_rate": 0.00019998162207999592,
-      "loss": 46.0,
-      "step": 37906
-    },
-    {
-      "epoch": 6.104674101211804,
-      "grad_norm": 0.0032758968882262707,
-      "learning_rate": 0.00019998162111009937,
-      "loss": 46.0,
-      "step": 37907
-    },
-    {
-      "epoch": 6.104835138290591,
-      "grad_norm": 0.003191495081409812,
-      "learning_rate": 0.00019998162014017728,
-      "loss": 46.0,
-      "step": 37908
-    },
-    {
-      "epoch": 6.104996175369378,
-      "grad_norm": 0.0013137803180143237,
-      "learning_rate": 0.00019998161917022955,
-      "loss": 46.0,
-      "step": 37909
-    },
-    {
-      "epoch": 6.105157212448166,
-      "grad_norm": 0.0024282927624881268,
-      "learning_rate": 0.00019998161820025624,
-      "loss": 46.0,
-      "step": 37910
-    },
-    {
-      "epoch": 6.105318249526953,
-      "grad_norm": 0.0022701954003423452,
-      "learning_rate": 0.00019998161723025736,
-      "loss": 46.0,
-      "step": 37911
-    },
-    {
-      "epoch": 6.105479286605741,
-      "grad_norm": 0.017151150852441788,
-      "learning_rate": 0.00019998161626023288,
-      "loss": 46.0,
-      "step": 37912
-    },
-    {
-      "epoch": 6.105640323684528,
-      "grad_norm": 0.0027103072497993708,
-      "learning_rate": 0.00019998161529018277,
-      "loss": 46.0,
-      "step": 37913
-    },
-    {
-      "epoch": 6.105801360763316,
-      "grad_norm": 0.0013941014185547829,
-      "learning_rate": 0.0001999816143201071,
-      "loss": 46.0,
-      "step": 37914
-    },
-    {
-      "epoch": 6.105962397842103,
-      "grad_norm": 0.002070285379886627,
-      "learning_rate": 0.00019998161335000586,
-      "loss": 46.0,
-      "step": 37915
-    },
-    {
-      "epoch": 6.106123434920891,
-      "grad_norm": 0.0026621525175869465,
-      "learning_rate": 0.00019998161237987902,
-      "loss": 46.0,
-      "step": 37916
-    },
-    {
-      "epoch": 6.106284471999678,
-      "grad_norm": 0.01080483477562666,
-      "learning_rate": 0.0001999816114097266,
-      "loss": 46.0,
-      "step": 37917
-    },
-    {
-      "epoch": 6.1064455090784655,
-      "grad_norm": 0.006349895615130663,
-      "learning_rate": 0.00019998161043954858,
-      "loss": 46.0,
-      "step": 37918
-    },
-    {
-      "epoch": 6.106606546157253,
-      "grad_norm": 0.0034800455905497074,
-      "learning_rate": 0.00019998160946934496,
-      "loss": 46.0,
-      "step": 37919
-    },
-    {
-      "epoch": 6.10676758323604,
-      "grad_norm": 0.010056184604763985,
-      "learning_rate": 0.00019998160849911574,
-      "loss": 46.0,
-      "step": 37920
-    },
-    {
-      "epoch": 6.106928620314828,
-      "grad_norm": 0.009464414790272713,
-      "learning_rate": 0.00019998160752886094,
-      "loss": 46.0,
-      "step": 37921
-    },
-    {
-      "epoch": 6.107089657393615,
-      "grad_norm": 0.003048659535124898,
-      "learning_rate": 0.00019998160655858058,
-      "loss": 46.0,
-      "step": 37922
-    },
-    {
-      "epoch": 6.107250694472402,
-      "grad_norm": 0.0032577086240053177,
-      "learning_rate": 0.00019998160558827458,
-      "loss": 46.0,
-      "step": 37923
-    },
-    {
-      "epoch": 6.107411731551189,
-      "grad_norm": 0.0035832293797284365,
-      "learning_rate": 0.00019998160461794301,
-      "loss": 46.0,
-      "step": 37924
-    },
-    {
-      "epoch": 6.107572768629977,
-      "grad_norm": 0.0037689015734940767,
-      "learning_rate": 0.00019998160364758586,
-      "loss": 46.0,
-      "step": 37925
-    },
-    {
-      "epoch": 6.107733805708764,
-      "grad_norm": 0.002485231263563037,
-      "learning_rate": 0.00019998160267720313,
-      "loss": 46.0,
-      "step": 37926
-    },
-    {
-      "epoch": 6.107894842787552,
-      "grad_norm": 0.002768041333183646,
-      "learning_rate": 0.00019998160170679478,
-      "loss": 46.0,
-      "step": 37927
-    },
-    {
-      "epoch": 6.108055879866339,
-      "grad_norm": 0.017291713505983353,
-      "learning_rate": 0.00019998160073636087,
-      "loss": 46.0,
-      "step": 37928
-    },
-    {
-      "epoch": 6.108216916945127,
-      "grad_norm": 0.007592449430376291,
-      "learning_rate": 0.00019998159976590134,
-      "loss": 46.0,
-      "step": 37929
-    },
-    {
-      "epoch": 6.108377954023914,
-      "grad_norm": 0.002873070538043976,
-      "learning_rate": 0.00019998159879541625,
-      "loss": 46.0,
-      "step": 37930
-    },
-    {
-      "epoch": 6.1085389911027015,
-      "grad_norm": 0.010722826234996319,
-      "learning_rate": 0.00019998159782490555,
-      "loss": 46.0,
-      "step": 37931
-    },
-    {
-      "epoch": 6.108700028181489,
-      "grad_norm": 0.014389226213097572,
-      "learning_rate": 0.00019998159685436927,
-      "loss": 46.0,
-      "step": 37932
-    },
-    {
-      "epoch": 6.108861065260276,
-      "grad_norm": 0.02572016417980194,
-      "learning_rate": 0.0001999815958838074,
-      "loss": 46.0,
-      "step": 37933
-    },
-    {
-      "epoch": 6.109022102339064,
-      "grad_norm": 0.0034549778793007135,
-      "learning_rate": 0.0001999815949132199,
-      "loss": 46.0,
-      "step": 37934
-    },
-    {
-      "epoch": 6.109183139417851,
-      "grad_norm": 0.0053646741434931755,
-      "learning_rate": 0.00019998159394260686,
-      "loss": 46.0,
-      "step": 37935
-    },
-    {
-      "epoch": 6.109344176496639,
-      "grad_norm": 0.005037750583142042,
-      "learning_rate": 0.00019998159297196822,
-      "loss": 46.0,
-      "step": 37936
-    },
-    {
-      "epoch": 6.109505213575425,
-      "grad_norm": 0.002898403676226735,
-      "learning_rate": 0.00019998159200130394,
-      "loss": 46.0,
-      "step": 37937
-    },
-    {
-      "epoch": 6.109666250654213,
-      "grad_norm": 0.008576362393796444,
-      "learning_rate": 0.00019998159103061413,
-      "loss": 46.0,
-      "step": 37938
-    },
-    {
-      "epoch": 6.109827287733,
-      "grad_norm": 0.009034862741827965,
-      "learning_rate": 0.0001999815900598987,
-      "loss": 46.0,
-      "step": 37939
-    },
-    {
-      "epoch": 6.109988324811788,
-      "grad_norm": 0.010549517348408699,
-      "learning_rate": 0.0001999815890891577,
-      "loss": 46.0,
-      "step": 37940
-    },
-    {
-      "epoch": 6.110149361890575,
-      "grad_norm": 0.0023834872990846634,
-      "learning_rate": 0.00019998158811839113,
-      "loss": 46.0,
-      "step": 37941
-    },
-    {
-      "epoch": 6.110310398969363,
-      "grad_norm": 0.005191650707274675,
-      "learning_rate": 0.00019998158714759891,
-      "loss": 46.0,
-      "step": 37942
-    },
-    {
-      "epoch": 6.11047143604815,
-      "grad_norm": 0.0058679780922830105,
-      "learning_rate": 0.00019998158617678114,
-      "loss": 46.0,
-      "step": 37943
-    },
-    {
-      "epoch": 6.1106324731269375,
-      "grad_norm": 0.005831942893564701,
-      "learning_rate": 0.00019998158520593778,
-      "loss": 46.0,
-      "step": 37944
-    },
-    {
-      "epoch": 6.110793510205725,
-      "grad_norm": 0.0051593054085969925,
-      "learning_rate": 0.0001999815842350688,
-      "loss": 46.0,
-      "step": 37945
-    },
-    {
-      "epoch": 6.110954547284512,
-      "grad_norm": 0.003682398470118642,
-      "learning_rate": 0.00019998158326417427,
-      "loss": 46.0,
-      "step": 37946
-    },
-    {
-      "epoch": 6.1111155843633,
-      "grad_norm": 0.003645791672170162,
-      "learning_rate": 0.00019998158229325412,
-      "loss": 46.0,
-      "step": 37947
-    },
-    {
-      "epoch": 6.111276621442087,
-      "grad_norm": 0.006710568908601999,
-      "learning_rate": 0.0001999815813223084,
-      "loss": 46.0,
-      "step": 37948
-    },
-    {
-      "epoch": 6.111437658520875,
-      "grad_norm": 0.033028069883584976,
-      "learning_rate": 0.00019998158035133706,
-      "loss": 46.0,
-      "step": 37949
-    },
-    {
-      "epoch": 6.111598695599662,
-      "grad_norm": 0.0022661553230136633,
-      "learning_rate": 0.00019998157938034015,
-      "loss": 46.0,
-      "step": 37950
-    },
-    {
-      "epoch": 6.11175973267845,
-      "grad_norm": 0.0008625874179415405,
-      "learning_rate": 0.00019998157840931766,
-      "loss": 46.0,
-      "step": 37951
-    },
-    {
-      "epoch": 6.111920769757236,
-      "grad_norm": 0.0030962605960667133,
-      "learning_rate": 0.00019998157743826957,
-      "loss": 46.0,
-      "step": 37952
-    },
-    {
-      "epoch": 6.112081806836024,
-      "grad_norm": 0.003905230201780796,
-      "learning_rate": 0.0001999815764671959,
-      "loss": 46.0,
-      "step": 37953
-    },
-    {
-      "epoch": 6.112242843914811,
-      "grad_norm": 0.006993682123720646,
-      "learning_rate": 0.00019998157549609661,
-      "loss": 46.0,
-      "step": 37954
-    },
-    {
-      "epoch": 6.1124038809935985,
-      "grad_norm": 0.0074691143818199635,
-      "learning_rate": 0.00019998157452497174,
-      "loss": 46.0,
-      "step": 37955
-    },
-    {
-      "epoch": 6.112564918072386,
-      "grad_norm": 0.006067709997296333,
-      "learning_rate": 0.0001999815735538213,
-      "loss": 46.0,
-      "step": 37956
-    },
-    {
-      "epoch": 6.1127259551511735,
-      "grad_norm": 0.0025282027199864388,
-      "learning_rate": 0.00019998157258264526,
-      "loss": 46.0,
-      "step": 37957
-    },
-    {
-      "epoch": 6.112886992229961,
-      "grad_norm": 0.0023204137105494738,
-      "learning_rate": 0.00019998157161144365,
-      "loss": 46.0,
-      "step": 37958
-    },
-    {
-      "epoch": 6.113048029308748,
-      "grad_norm": 0.008157733827829361,
-      "learning_rate": 0.00019998157064021643,
-      "loss": 46.0,
-      "step": 37959
-    },
-    {
-      "epoch": 6.113209066387536,
-      "grad_norm": 0.0015601470367982984,
-      "learning_rate": 0.00019998156966896362,
-      "loss": 46.0,
-      "step": 37960
-    },
-    {
-      "epoch": 6.113370103466323,
-      "grad_norm": 0.01031606923788786,
-      "learning_rate": 0.0001999815686976852,
-      "loss": 46.0,
-      "step": 37961
-    },
-    {
-      "epoch": 6.113531140545111,
-      "grad_norm": 0.004116305150091648,
-      "learning_rate": 0.00019998156772638121,
-      "loss": 46.0,
-      "step": 37962
-    },
-    {
-      "epoch": 6.113692177623898,
-      "grad_norm": 0.005379362031817436,
-      "learning_rate": 0.00019998156675505162,
-      "loss": 46.0,
-      "step": 37963
-    },
-    {
-      "epoch": 6.113853214702686,
-      "grad_norm": 0.0032583598513156176,
-      "learning_rate": 0.00019998156578369643,
-      "loss": 46.0,
-      "step": 37964
-    },
-    {
-      "epoch": 6.114014251781473,
-      "grad_norm": 0.008800827898085117,
-      "learning_rate": 0.0001999815648123157,
-      "loss": 46.0,
-      "step": 37965
-    },
-    {
-      "epoch": 6.11417528886026,
-      "grad_norm": 0.0035629295744001865,
-      "learning_rate": 0.00019998156384090935,
-      "loss": 46.0,
-      "step": 37966
-    },
-    {
-      "epoch": 6.114336325939047,
-      "grad_norm": 0.0010251258499920368,
-      "learning_rate": 0.0001999815628694774,
-      "loss": 46.0,
-      "step": 37967
-    },
-    {
-      "epoch": 6.1144973630178345,
-      "grad_norm": 0.006093880161643028,
-      "learning_rate": 0.00019998156189801987,
-      "loss": 46.0,
-      "step": 37968
-    },
-    {
-      "epoch": 6.114658400096622,
-      "grad_norm": 0.0019650463946163654,
-      "learning_rate": 0.00019998156092653675,
-      "loss": 46.0,
-      "step": 37969
-    },
-    {
-      "epoch": 6.114819437175409,
-      "grad_norm": 0.002851185156032443,
-      "learning_rate": 0.00019998155995502802,
-      "loss": 46.0,
-      "step": 37970
-    },
-    {
-      "epoch": 6.114980474254197,
-      "grad_norm": 0.004890704061836004,
-      "learning_rate": 0.00019998155898349372,
-      "loss": 46.0,
-      "step": 37971
-    },
-    {
-      "epoch": 6.115141511332984,
-      "grad_norm": 0.029467392712831497,
-      "learning_rate": 0.00019998155801193384,
-      "loss": 46.0,
-      "step": 37972
-    },
-    {
-      "epoch": 6.115302548411772,
-      "grad_norm": 0.001138051156885922,
-      "learning_rate": 0.00019998155704034835,
-      "loss": 46.0,
-      "step": 37973
-    },
-    {
-      "epoch": 6.115463585490559,
-      "grad_norm": 0.002168205101042986,
-      "learning_rate": 0.00019998155606873726,
-      "loss": 46.0,
-      "step": 37974
-    },
-    {
-      "epoch": 6.115624622569347,
-      "grad_norm": 0.0041888016276061535,
-      "learning_rate": 0.00019998155509710062,
-      "loss": 46.0,
-      "step": 37975
-    },
-    {
-      "epoch": 6.115785659648134,
-      "grad_norm": 0.021223602816462517,
-      "learning_rate": 0.00019998155412543836,
-      "loss": 46.0,
-      "step": 37976
-    },
-    {
-      "epoch": 6.115946696726922,
-      "grad_norm": 0.0009306090651080012,
-      "learning_rate": 0.00019998155315375051,
-      "loss": 46.0,
-      "step": 37977
-    },
-    {
-      "epoch": 6.116107733805709,
-      "grad_norm": 0.0015372381312772632,
-      "learning_rate": 0.00019998155218203708,
-      "loss": 46.0,
-      "step": 37978
-    },
-    {
-      "epoch": 6.1162687708844965,
-      "grad_norm": 0.017244672402739525,
-      "learning_rate": 0.00019998155121029804,
-      "loss": 46.0,
-      "step": 37979
-    },
-    {
-      "epoch": 6.116429807963284,
-      "grad_norm": 0.003151554614305496,
-      "learning_rate": 0.00019998155023853346,
-      "loss": 46.0,
-      "step": 37980
-    },
-    {
-      "epoch": 6.1165908450420705,
-      "grad_norm": 0.004102529492229223,
-      "learning_rate": 0.00019998154926674324,
-      "loss": 46.0,
-      "step": 37981
-    },
-    {
-      "epoch": 6.116751882120858,
-      "grad_norm": 0.002577023347839713,
-      "learning_rate": 0.00019998154829492743,
-      "loss": 46.0,
-      "step": 37982
-    },
-    {
-      "epoch": 6.116912919199645,
-      "grad_norm": 0.005130618344992399,
-      "learning_rate": 0.00019998154732308606,
-      "loss": 46.0,
-      "step": 37983
-    },
-    {
-      "epoch": 6.117073956278433,
-      "grad_norm": 0.002902435604482889,
-      "learning_rate": 0.00019998154635121908,
-      "loss": 46.0,
-      "step": 37984
-    },
-    {
-      "epoch": 6.11723499335722,
-      "grad_norm": 0.0036257412284612656,
-      "learning_rate": 0.0001999815453793265,
-      "loss": 46.0,
-      "step": 37985
-    },
-    {
-      "epoch": 6.117396030436008,
-      "grad_norm": 0.0008864043047651649,
-      "learning_rate": 0.00019998154440740838,
-      "loss": 46.0,
-      "step": 37986
-    },
-    {
-      "epoch": 6.117557067514795,
-      "grad_norm": 0.0033855861984193325,
-      "learning_rate": 0.0001999815434354646,
-      "loss": 46.0,
-      "step": 37987
-    },
-    {
-      "epoch": 6.117718104593583,
-      "grad_norm": 0.013077301904559135,
-      "learning_rate": 0.00019998154246349527,
-      "loss": 46.0,
-      "step": 37988
-    },
-    {
-      "epoch": 6.11787914167237,
-      "grad_norm": 0.0028082553762942553,
-      "learning_rate": 0.00019998154149150036,
-      "loss": 46.0,
-      "step": 37989
-    },
-    {
-      "epoch": 6.118040178751158,
-      "grad_norm": 0.005463637411594391,
-      "learning_rate": 0.00019998154051947985,
-      "loss": 46.0,
-      "step": 37990
-    },
-    {
-      "epoch": 6.118201215829945,
-      "grad_norm": 0.02385544776916504,
-      "learning_rate": 0.00019998153954743373,
-      "loss": 46.0,
-      "step": 37991
-    },
-    {
-      "epoch": 6.1183622529087325,
-      "grad_norm": 0.01385748852044344,
-      "learning_rate": 0.00019998153857536205,
-      "loss": 46.0,
-      "step": 37992
-    },
-    {
-      "epoch": 6.11852328998752,
-      "grad_norm": 0.007933791726827621,
-      "learning_rate": 0.00019998153760326476,
-      "loss": 46.0,
-      "step": 37993
-    },
-    {
-      "epoch": 6.118684327066307,
-      "grad_norm": 0.0036477777175605297,
-      "learning_rate": 0.00019998153663114188,
-      "loss": 46.0,
-      "step": 37994
-    },
-    {
-      "epoch": 6.118845364145095,
-      "grad_norm": 0.010108874179422855,
-      "learning_rate": 0.0001999815356589934,
-      "loss": 46.0,
-      "step": 37995
-    },
-    {
-      "epoch": 6.119006401223881,
-      "grad_norm": 0.012677617371082306,
-      "learning_rate": 0.00019998153468681935,
-      "loss": 46.0,
-      "step": 37996
-    },
-    {
-      "epoch": 6.119167438302669,
-      "grad_norm": 0.011315323412418365,
-      "learning_rate": 0.0001999815337146197,
-      "loss": 46.0,
-      "step": 37997
-    },
-    {
-      "epoch": 6.119328475381456,
-      "grad_norm": 0.003727993695065379,
-      "learning_rate": 0.00019998153274239445,
-      "loss": 46.0,
-      "step": 37998
-    },
-    {
-      "epoch": 6.119489512460244,
-      "grad_norm": 0.005289892200380564,
-      "learning_rate": 0.00019998153177014363,
-      "loss": 46.0,
-      "step": 37999
-    },
-    {
-      "epoch": 6.119650549539031,
-      "grad_norm": 0.01043575257062912,
-      "learning_rate": 0.00019998153079786723,
-      "loss": 46.0,
-      "step": 38000
-    },
-    {
-      "epoch": 6.119811586617819,
-      "grad_norm": 0.013002244755625725,
-      "learning_rate": 0.0001999815298255652,
-      "loss": 46.0,
-      "step": 38001
-    },
-    {
-      "epoch": 6.119972623696606,
-      "grad_norm": 0.0046134768053889275,
-      "learning_rate": 0.00019998152885323763,
-      "loss": 46.0,
-      "step": 38002
-    },
-    {
-      "epoch": 6.120133660775394,
-      "grad_norm": 0.01637146808207035,
-      "learning_rate": 0.0001999815278808844,
-      "loss": 46.0,
-      "step": 38003
-    },
-    {
-      "epoch": 6.120294697854181,
-      "grad_norm": 0.002225615317001939,
-      "learning_rate": 0.00019998152690850563,
-      "loss": 46.0,
-      "step": 38004
-    },
-    {
-      "epoch": 6.1204557349329685,
-      "grad_norm": 0.002993962960317731,
-      "learning_rate": 0.00019998152593610126,
-      "loss": 46.0,
-      "step": 38005
-    },
-    {
-      "epoch": 6.120616772011756,
-      "grad_norm": 0.002434214809909463,
-      "learning_rate": 0.0001999815249636713,
-      "loss": 46.0,
-      "step": 38006
-    },
-    {
-      "epoch": 6.120777809090543,
-      "grad_norm": 0.0043135639280080795,
-      "learning_rate": 0.00019998152399121577,
-      "loss": 46.0,
-      "step": 38007
-    },
-    {
-      "epoch": 6.120938846169331,
-      "grad_norm": 0.0027210870757699013,
-      "learning_rate": 0.00019998152301873464,
-      "loss": 46.0,
-      "step": 38008
-    },
-    {
-      "epoch": 6.121099883248118,
-      "grad_norm": 0.012294287793338299,
-      "learning_rate": 0.0001999815220462279,
-      "loss": 46.0,
-      "step": 38009
-    },
-    {
-      "epoch": 6.121260920326906,
-      "grad_norm": 0.0053977444767951965,
-      "learning_rate": 0.0001999815210736956,
-      "loss": 46.0,
-      "step": 38010
-    },
-    {
-      "epoch": 6.121421957405692,
-      "grad_norm": 0.007198524661362171,
-      "learning_rate": 0.00019998152010113768,
-      "loss": 46.0,
-      "step": 38011
-    },
-    {
-      "epoch": 6.12158299448448,
-      "grad_norm": 0.0027359111700206995,
-      "learning_rate": 0.00019998151912855417,
-      "loss": 46.0,
-      "step": 38012
-    },
-    {
-      "epoch": 6.121744031563267,
-      "grad_norm": 0.00808628648519516,
-      "learning_rate": 0.0001999815181559451,
-      "loss": 46.0,
-      "step": 38013
-    },
-    {
-      "epoch": 6.121905068642055,
-      "grad_norm": 0.0009916462004184723,
-      "learning_rate": 0.00019998151718331043,
-      "loss": 46.0,
-      "step": 38014
-    },
-    {
-      "epoch": 6.122066105720842,
-      "grad_norm": 0.003694161307066679,
-      "learning_rate": 0.00019998151621065016,
-      "loss": 46.0,
-      "step": 38015
-    },
-    {
-      "epoch": 6.1222271427996295,
-      "grad_norm": 0.0021007994655519724,
-      "learning_rate": 0.00019998151523796425,
-      "loss": 46.0,
-      "step": 38016
-    },
-    {
-      "epoch": 6.122388179878417,
-      "grad_norm": 0.002220150548964739,
-      "learning_rate": 0.00019998151426525284,
-      "loss": 46.0,
-      "step": 38017
-    },
-    {
-      "epoch": 6.122549216957204,
-      "grad_norm": 0.015249320305883884,
-      "learning_rate": 0.00019998151329251579,
-      "loss": 46.0,
-      "step": 38018
-    },
-    {
-      "epoch": 6.122710254035992,
-      "grad_norm": 0.009781640022993088,
-      "learning_rate": 0.00019998151231975317,
-      "loss": 46.0,
-      "step": 38019
-    },
-    {
-      "epoch": 6.122871291114779,
-      "grad_norm": 0.016867421567440033,
-      "learning_rate": 0.00019998151134696494,
-      "loss": 46.0,
-      "step": 38020
-    },
-    {
-      "epoch": 6.123032328193567,
-      "grad_norm": 0.00385505729354918,
-      "learning_rate": 0.00019998151037415113,
-      "loss": 46.0,
-      "step": 38021
-    },
-    {
-      "epoch": 6.123193365272354,
-      "grad_norm": 0.0023131840862333775,
-      "learning_rate": 0.00019998150940131172,
-      "loss": 46.0,
-      "step": 38022
-    },
-    {
-      "epoch": 6.123354402351142,
-      "grad_norm": 0.0061761774122715,
-      "learning_rate": 0.00019998150842844673,
-      "loss": 46.0,
-      "step": 38023
-    },
-    {
-      "epoch": 6.123515439429929,
-      "grad_norm": 0.007513138931244612,
-      "learning_rate": 0.00019998150745555618,
-      "loss": 46.0,
-      "step": 38024
-    },
-    {
-      "epoch": 6.123676476508716,
-      "grad_norm": 0.012906387448310852,
-      "learning_rate": 0.00019998150648264,
-      "loss": 46.0,
-      "step": 38025
-    },
-    {
-      "epoch": 6.123837513587503,
-      "grad_norm": 0.0005248477682471275,
-      "learning_rate": 0.0001999815055096982,
-      "loss": 46.0,
-      "step": 38026
-    },
-    {
-      "epoch": 6.123998550666291,
-      "grad_norm": 0.002935335971415043,
-      "learning_rate": 0.00019998150453673084,
-      "loss": 46.0,
-      "step": 38027
-    },
-    {
-      "epoch": 6.124159587745078,
-      "grad_norm": 0.002595342695713043,
-      "learning_rate": 0.00019998150356373792,
-      "loss": 46.0,
-      "step": 38028
-    },
-    {
-      "epoch": 6.1243206248238655,
-      "grad_norm": 0.012992041185498238,
-      "learning_rate": 0.0001999815025907194,
-      "loss": 46.0,
-      "step": 38029
-    },
-    {
-      "epoch": 6.124481661902653,
-      "grad_norm": 0.002713417401537299,
-      "learning_rate": 0.00019998150161767528,
-      "loss": 46.0,
-      "step": 38030
-    },
-    {
-      "epoch": 6.12464269898144,
-      "grad_norm": 0.026003146544098854,
-      "learning_rate": 0.00019998150064460556,
-      "loss": 46.0,
-      "step": 38031
-    },
-    {
-      "epoch": 6.124803736060228,
-      "grad_norm": 0.0016140779480338097,
-      "learning_rate": 0.00019998149967151029,
-      "loss": 46.0,
-      "step": 38032
-    },
-    {
-      "epoch": 6.124964773139015,
-      "grad_norm": 0.002056755591183901,
-      "learning_rate": 0.00019998149869838937,
-      "loss": 46.0,
-      "step": 38033
-    },
-    {
-      "epoch": 6.125125810217803,
-      "grad_norm": 0.0034807310439646244,
-      "learning_rate": 0.0001999814977252429,
-      "loss": 46.0,
-      "step": 38034
-    },
-    {
-      "epoch": 6.12528684729659,
-      "grad_norm": 0.003984707407653332,
-      "learning_rate": 0.00019998149675207083,
-      "loss": 46.0,
-      "step": 38035
-    },
-    {
-      "epoch": 6.125447884375378,
-      "grad_norm": 0.0059023951180279255,
-      "learning_rate": 0.00019998149577887315,
-      "loss": 46.0,
-      "step": 38036
-    },
-    {
-      "epoch": 6.125608921454165,
-      "grad_norm": 0.00479265907779336,
-      "learning_rate": 0.00019998149480564989,
-      "loss": 46.0,
-      "step": 38037
-    },
-    {
-      "epoch": 6.125769958532953,
-      "grad_norm": 0.0049417889676988125,
-      "learning_rate": 0.00019998149383240106,
-      "loss": 46.0,
-      "step": 38038
-    },
-    {
-      "epoch": 6.125930995611739,
-      "grad_norm": 0.009702890180051327,
-      "learning_rate": 0.00019998149285912665,
-      "loss": 46.0,
-      "step": 38039
-    },
-    {
-      "epoch": 6.126092032690527,
-      "grad_norm": 0.003543219994753599,
-      "learning_rate": 0.00019998149188582662,
-      "loss": 46.0,
-      "step": 38040
-    },
-    {
-      "epoch": 6.126253069769314,
-      "grad_norm": 0.004731372464448214,
-      "learning_rate": 0.00019998149091250098,
-      "loss": 46.0,
-      "step": 38041
-    },
-    {
-      "epoch": 6.1264141068481015,
-      "grad_norm": 0.002965356456115842,
-      "learning_rate": 0.0001999814899391498,
-      "loss": 46.0,
-      "step": 38042
-    },
-    {
-      "epoch": 6.126575143926889,
-      "grad_norm": 0.00771909998729825,
-      "learning_rate": 0.000199981488965773,
-      "loss": 46.0,
-      "step": 38043
-    },
-    {
-      "epoch": 6.126736181005676,
-      "grad_norm": 0.0014029525918886065,
-      "learning_rate": 0.00019998148799237062,
-      "loss": 46.0,
-      "step": 38044
-    },
-    {
-      "epoch": 6.126897218084464,
-      "grad_norm": 0.008475450798869133,
-      "learning_rate": 0.00019998148701894263,
-      "loss": 46.0,
-      "step": 38045
-    },
-    {
-      "epoch": 6.127058255163251,
-      "grad_norm": 0.010319489054381847,
-      "learning_rate": 0.00019998148604548908,
-      "loss": 46.0,
-      "step": 38046
-    },
-    {
-      "epoch": 6.127219292242039,
-      "grad_norm": 0.008587724529206753,
-      "learning_rate": 0.0001999814850720099,
-      "loss": 46.0,
-      "step": 38047
-    },
-    {
-      "epoch": 6.127380329320826,
-      "grad_norm": 0.002170801628381014,
-      "learning_rate": 0.00019998148409850516,
-      "loss": 46.0,
-      "step": 38048
-    },
-    {
-      "epoch": 6.127541366399614,
-      "grad_norm": 0.003980133216828108,
-      "learning_rate": 0.00019998148312497485,
-      "loss": 46.0,
-      "step": 38049
-    },
-    {
-      "epoch": 6.127702403478401,
-      "grad_norm": 0.003032906446605921,
-      "learning_rate": 0.00019998148215141892,
-      "loss": 46.0,
-      "step": 38050
-    },
-    {
-      "epoch": 6.127863440557189,
-      "grad_norm": 0.005040084943175316,
-      "learning_rate": 0.0001999814811778374,
-      "loss": 46.0,
-      "step": 38051
-    },
-    {
-      "epoch": 6.128024477635976,
-      "grad_norm": 0.0038813622668385506,
-      "learning_rate": 0.0001999814802042303,
-      "loss": 46.0,
-      "step": 38052
-    },
-    {
-      "epoch": 6.1281855147147635,
-      "grad_norm": 0.03539165109395981,
-      "learning_rate": 0.0001999814792305976,
-      "loss": 46.0,
-      "step": 38053
-    },
-    {
-      "epoch": 6.12834655179355,
-      "grad_norm": 0.0008202372118830681,
-      "learning_rate": 0.00019998147825693932,
-      "loss": 46.0,
-      "step": 38054
-    },
-    {
-      "epoch": 6.1285075888723375,
-      "grad_norm": 0.00880894809961319,
-      "learning_rate": 0.00019998147728325543,
-      "loss": 46.0,
-      "step": 38055
-    },
-    {
-      "epoch": 6.128668625951125,
-      "grad_norm": 0.001970510231330991,
-      "learning_rate": 0.00019998147630954599,
-      "loss": 46.0,
-      "step": 38056
-    },
-    {
-      "epoch": 6.128829663029912,
-      "grad_norm": 0.0038770800456404686,
-      "learning_rate": 0.0001999814753358109,
-      "loss": 46.0,
-      "step": 38057
-    },
-    {
-      "epoch": 6.1289907001087,
-      "grad_norm": 0.0012005168246105313,
-      "learning_rate": 0.00019998147436205027,
-      "loss": 46.0,
-      "step": 38058
-    },
-    {
-      "epoch": 6.129151737187487,
-      "grad_norm": 0.006247636396437883,
-      "learning_rate": 0.00019998147338826404,
-      "loss": 46.0,
-      "step": 38059
-    },
-    {
-      "epoch": 6.129312774266275,
-      "grad_norm": 0.002642413368448615,
-      "learning_rate": 0.0001999814724144522,
-      "loss": 46.0,
-      "step": 38060
-    },
-    {
-      "epoch": 6.129473811345062,
-      "grad_norm": 0.0021298436913639307,
-      "learning_rate": 0.0001999814714406148,
-      "loss": 46.0,
-      "step": 38061
-    },
-    {
-      "epoch": 6.12963484842385,
-      "grad_norm": 0.002107681706547737,
-      "learning_rate": 0.0001999814704667518,
-      "loss": 46.0,
-      "step": 38062
-    },
-    {
-      "epoch": 6.129795885502637,
-      "grad_norm": 0.004172795917838812,
-      "learning_rate": 0.00019998146949286321,
-      "loss": 46.0,
-      "step": 38063
-    },
-    {
-      "epoch": 6.1299569225814245,
-      "grad_norm": 0.003007931634783745,
-      "learning_rate": 0.00019998146851894901,
-      "loss": 46.0,
-      "step": 38064
-    },
-    {
-      "epoch": 6.130117959660212,
-      "grad_norm": 0.01600167155265808,
-      "learning_rate": 0.00019998146754500923,
-      "loss": 46.0,
-      "step": 38065
-    },
-    {
-      "epoch": 6.1302789967389995,
-      "grad_norm": 0.008139573968946934,
-      "learning_rate": 0.00019998146657104385,
-      "loss": 46.0,
-      "step": 38066
-    },
-    {
-      "epoch": 6.130440033817787,
-      "grad_norm": 0.002752548549324274,
-      "learning_rate": 0.00019998146559705292,
-      "loss": 46.0,
-      "step": 38067
-    },
-    {
-      "epoch": 6.130601070896574,
-      "grad_norm": 0.008915587328374386,
-      "learning_rate": 0.00019998146462303637,
-      "loss": 46.0,
-      "step": 38068
-    },
-    {
-      "epoch": 6.130762107975361,
-      "grad_norm": 0.006481254007667303,
-      "learning_rate": 0.00019998146364899423,
-      "loss": 46.0,
-      "step": 38069
-    },
-    {
-      "epoch": 6.130923145054148,
-      "grad_norm": 0.0029557249508798122,
-      "learning_rate": 0.0001999814626749265,
-      "loss": 46.0,
-      "step": 38070
-    },
-    {
-      "epoch": 6.131084182132936,
-      "grad_norm": 0.005246481858193874,
-      "learning_rate": 0.00019998146170083317,
-      "loss": 46.0,
-      "step": 38071
-    },
-    {
-      "epoch": 6.131245219211723,
-      "grad_norm": 0.020238157361745834,
-      "learning_rate": 0.00019998146072671428,
-      "loss": 46.0,
-      "step": 38072
-    },
-    {
-      "epoch": 6.131406256290511,
-      "grad_norm": 0.0030776364728808403,
-      "learning_rate": 0.0001999814597525698,
-      "loss": 46.0,
-      "step": 38073
-    },
-    {
-      "epoch": 6.131567293369298,
-      "grad_norm": 0.005505464971065521,
-      "learning_rate": 0.0001999814587783997,
-      "loss": 46.0,
-      "step": 38074
-    },
-    {
-      "epoch": 6.131728330448086,
-      "grad_norm": 0.008040811866521835,
-      "learning_rate": 0.00019998145780420403,
-      "loss": 46.0,
-      "step": 38075
-    },
-    {
-      "epoch": 6.131889367526873,
-      "grad_norm": 0.005472471471875906,
-      "learning_rate": 0.00019998145682998276,
-      "loss": 46.0,
-      "step": 38076
-    },
-    {
-      "epoch": 6.1320504046056605,
-      "grad_norm": 0.007615697104483843,
-      "learning_rate": 0.0001999814558557359,
-      "loss": 46.0,
-      "step": 38077
-    },
-    {
-      "epoch": 6.132211441684448,
-      "grad_norm": 0.002209107857197523,
-      "learning_rate": 0.00019998145488146345,
-      "loss": 46.0,
-      "step": 38078
-    },
-    {
-      "epoch": 6.132372478763235,
-      "grad_norm": 0.00337052159011364,
-      "learning_rate": 0.00019998145390716542,
-      "loss": 46.0,
-      "step": 38079
-    },
-    {
-      "epoch": 6.132533515842023,
-      "grad_norm": 0.002775853266939521,
-      "learning_rate": 0.0001999814529328418,
-      "loss": 46.0,
-      "step": 38080
-    },
-    {
-      "epoch": 6.13269455292081,
-      "grad_norm": 0.01040705107152462,
-      "learning_rate": 0.00019998145195849256,
-      "loss": 46.0,
-      "step": 38081
-    },
-    {
-      "epoch": 6.132855589999598,
-      "grad_norm": 0.002455855254083872,
-      "learning_rate": 0.00019998145098411776,
-      "loss": 46.0,
-      "step": 38082
-    },
-    {
-      "epoch": 6.133016627078385,
-      "grad_norm": 0.002750950865447521,
-      "learning_rate": 0.00019998145000971735,
-      "loss": 46.0,
-      "step": 38083
-    },
-    {
-      "epoch": 6.133177664157172,
-      "grad_norm": 0.0018378561362624168,
-      "learning_rate": 0.00019998144903529138,
-      "loss": 46.0,
-      "step": 38084
-    },
-    {
-      "epoch": 6.133338701235959,
-      "grad_norm": 0.018560731783509254,
-      "learning_rate": 0.0001999814480608398,
-      "loss": 46.0,
-      "step": 38085
-    },
-    {
-      "epoch": 6.133499738314747,
-      "grad_norm": 0.011354668997228146,
-      "learning_rate": 0.00019998144708636262,
-      "loss": 46.0,
-      "step": 38086
-    },
-    {
-      "epoch": 6.133660775393534,
-      "grad_norm": 0.002525091404095292,
-      "learning_rate": 0.00019998144611185986,
-      "loss": 46.0,
-      "step": 38087
-    },
-    {
-      "epoch": 6.133821812472322,
-      "grad_norm": 0.004146402236074209,
-      "learning_rate": 0.0001999814451373315,
-      "loss": 46.0,
-      "step": 38088
-    },
-    {
-      "epoch": 6.133982849551109,
-      "grad_norm": 0.0017427769489586353,
-      "learning_rate": 0.00019998144416277758,
-      "loss": 46.0,
-      "step": 38089
-    },
-    {
-      "epoch": 6.1341438866298965,
-      "grad_norm": 0.0190053079277277,
-      "learning_rate": 0.00019998144318819804,
-      "loss": 46.0,
-      "step": 38090
-    },
-    {
-      "epoch": 6.134304923708684,
-      "grad_norm": 0.003641800954937935,
-      "learning_rate": 0.00019998144221359293,
-      "loss": 46.0,
-      "step": 38091
-    },
-    {
-      "epoch": 6.134465960787471,
-      "grad_norm": 0.0028951247222721577,
-      "learning_rate": 0.00019998144123896223,
-      "loss": 46.0,
-      "step": 38092
-    },
-    {
-      "epoch": 6.134626997866259,
-      "grad_norm": 0.008201111108064651,
-      "learning_rate": 0.00019998144026430592,
-      "loss": 46.0,
-      "step": 38093
-    },
-    {
-      "epoch": 6.134788034945046,
-      "grad_norm": 0.006603895220905542,
-      "learning_rate": 0.00019998143928962403,
-      "loss": 46.0,
-      "step": 38094
-    },
-    {
-      "epoch": 6.134949072023834,
-      "grad_norm": 0.01317819207906723,
-      "learning_rate": 0.00019998143831491657,
-      "loss": 46.0,
-      "step": 38095
-    },
-    {
-      "epoch": 6.135110109102621,
-      "grad_norm": 0.00565582700073719,
-      "learning_rate": 0.00019998143734018347,
-      "loss": 46.0,
-      "step": 38096
-    },
-    {
-      "epoch": 6.135271146181409,
-      "grad_norm": 0.004093275871127844,
-      "learning_rate": 0.00019998143636542481,
-      "loss": 46.0,
-      "step": 38097
-    },
-    {
-      "epoch": 6.135432183260195,
-      "grad_norm": 0.016649968922138214,
-      "learning_rate": 0.00019998143539064057,
-      "loss": 46.0,
-      "step": 38098
-    },
-    {
-      "epoch": 6.135593220338983,
-      "grad_norm": 0.0034655649214982986,
-      "learning_rate": 0.00019998143441583074,
-      "loss": 46.0,
-      "step": 38099
-    },
-    {
-      "epoch": 6.13575425741777,
-      "grad_norm": 0.0032079655211418867,
-      "learning_rate": 0.0001999814334409953,
-      "loss": 46.0,
-      "step": 38100
-    },
-    {
-      "epoch": 6.135915294496558,
-      "grad_norm": 0.002937356708571315,
-      "learning_rate": 0.00019998143246613428,
-      "loss": 46.0,
-      "step": 38101
-    },
-    {
-      "epoch": 6.136076331575345,
-      "grad_norm": 0.005748485215008259,
-      "learning_rate": 0.00019998143149124766,
-      "loss": 46.0,
-      "step": 38102
-    },
-    {
-      "epoch": 6.1362373686541325,
-      "grad_norm": 0.0035205793101340532,
-      "learning_rate": 0.00019998143051633545,
-      "loss": 46.0,
-      "step": 38103
-    },
-    {
-      "epoch": 6.13639840573292,
-      "grad_norm": 0.01791226677596569,
-      "learning_rate": 0.00019998142954139766,
-      "loss": 46.0,
-      "step": 38104
-    },
-    {
-      "epoch": 6.136559442811707,
-      "grad_norm": 0.0010290952632203698,
-      "learning_rate": 0.0001999814285664343,
-      "loss": 46.0,
-      "step": 38105
-    },
-    {
-      "epoch": 6.136720479890495,
-      "grad_norm": 0.005278922617435455,
-      "learning_rate": 0.0001999814275914453,
-      "loss": 46.0,
-      "step": 38106
-    },
-    {
-      "epoch": 6.136881516969282,
-      "grad_norm": 0.0071410625241696835,
-      "learning_rate": 0.00019998142661643075,
-      "loss": 46.0,
-      "step": 38107
-    },
-    {
-      "epoch": 6.13704255404807,
-      "grad_norm": 0.005684057250618935,
-      "learning_rate": 0.0001999814256413906,
-      "loss": 46.0,
-      "step": 38108
-    },
-    {
-      "epoch": 6.137203591126857,
-      "grad_norm": 0.010827167890965939,
-      "learning_rate": 0.00019998142466632487,
-      "loss": 46.0,
-      "step": 38109
-    },
-    {
-      "epoch": 6.137364628205645,
-      "grad_norm": 0.0033071995712816715,
-      "learning_rate": 0.00019998142369123353,
-      "loss": 46.0,
-      "step": 38110
-    },
-    {
-      "epoch": 6.137525665284432,
-      "grad_norm": 0.001631106366403401,
-      "learning_rate": 0.00019998142271611662,
-      "loss": 46.0,
-      "step": 38111
-    },
-    {
-      "epoch": 6.13768670236322,
-      "grad_norm": 0.004384367261081934,
-      "learning_rate": 0.0001999814217409741,
-      "loss": 46.0,
-      "step": 38112
-    },
-    {
-      "epoch": 6.137847739442006,
-      "grad_norm": 0.0061812326312065125,
-      "learning_rate": 0.000199981420765806,
-      "loss": 46.0,
-      "step": 38113
-    },
-    {
-      "epoch": 6.138008776520794,
-      "grad_norm": 0.005505761131644249,
-      "learning_rate": 0.0001999814197906123,
-      "loss": 46.0,
-      "step": 38114
-    },
-    {
-      "epoch": 6.138169813599581,
-      "grad_norm": 0.0031309358309954405,
-      "learning_rate": 0.00019998141881539302,
-      "loss": 46.0,
-      "step": 38115
-    },
-    {
-      "epoch": 6.1383308506783685,
-      "grad_norm": 0.015083485282957554,
-      "learning_rate": 0.00019998141784014815,
-      "loss": 46.0,
-      "step": 38116
-    },
-    {
-      "epoch": 6.138491887757156,
-      "grad_norm": 0.010768556967377663,
-      "learning_rate": 0.0001999814168648777,
-      "loss": 46.0,
-      "step": 38117
-    },
-    {
-      "epoch": 6.138652924835943,
-      "grad_norm": 0.016163639724254608,
-      "learning_rate": 0.00019998141588958162,
-      "loss": 46.0,
-      "step": 38118
-    },
-    {
-      "epoch": 6.138813961914731,
-      "grad_norm": 0.0023822453804314137,
-      "learning_rate": 0.00019998141491426,
-      "loss": 46.0,
-      "step": 38119
-    },
-    {
-      "epoch": 6.138974998993518,
-      "grad_norm": 0.007294551003724337,
-      "learning_rate": 0.00019998141393891278,
-      "loss": 46.0,
-      "step": 38120
-    },
-    {
-      "epoch": 6.139136036072306,
-      "grad_norm": 0.0016139112412929535,
-      "learning_rate": 0.00019998141296353995,
-      "loss": 46.0,
-      "step": 38121
-    },
-    {
-      "epoch": 6.139297073151093,
-      "grad_norm": 0.005083042196929455,
-      "learning_rate": 0.00019998141198814153,
-      "loss": 46.0,
-      "step": 38122
-    },
-    {
-      "epoch": 6.139458110229881,
-      "grad_norm": 0.008838369511067867,
-      "learning_rate": 0.00019998141101271752,
-      "loss": 46.0,
-      "step": 38123
-    },
-    {
-      "epoch": 6.139619147308668,
-      "grad_norm": 0.002758219838142395,
-      "learning_rate": 0.00019998141003726793,
-      "loss": 46.0,
-      "step": 38124
-    },
-    {
-      "epoch": 6.1397801843874555,
-      "grad_norm": 0.0021110151428729296,
-      "learning_rate": 0.00019998140906179275,
-      "loss": 46.0,
-      "step": 38125
-    },
-    {
-      "epoch": 6.139941221466243,
-      "grad_norm": 0.003262303536757827,
-      "learning_rate": 0.00019998140808629198,
-      "loss": 46.0,
-      "step": 38126
-    },
-    {
-      "epoch": 6.14010225854503,
-      "grad_norm": 0.010376634076237679,
-      "learning_rate": 0.0001999814071107656,
-      "loss": 46.0,
-      "step": 38127
-    },
-    {
-      "epoch": 6.140263295623817,
-      "grad_norm": 0.008289509452879429,
-      "learning_rate": 0.00019998140613521366,
-      "loss": 46.0,
-      "step": 38128
-    },
-    {
-      "epoch": 6.1404243327026045,
-      "grad_norm": 0.002768499543890357,
-      "learning_rate": 0.00019998140515963613,
-      "loss": 46.0,
-      "step": 38129
-    },
-    {
-      "epoch": 6.140585369781392,
-      "grad_norm": 0.0010839044116437435,
-      "learning_rate": 0.00019998140418403299,
-      "loss": 46.0,
-      "step": 38130
-    },
-    {
-      "epoch": 6.140746406860179,
-      "grad_norm": 0.009540265426039696,
-      "learning_rate": 0.00019998140320840423,
-      "loss": 46.0,
-      "step": 38131
-    },
-    {
-      "epoch": 6.140907443938967,
-      "grad_norm": 0.022692784667015076,
-      "learning_rate": 0.00019998140223274994,
-      "loss": 46.0,
-      "step": 38132
-    },
-    {
-      "epoch": 6.141068481017754,
-      "grad_norm": 0.003273993730545044,
-      "learning_rate": 0.00019998140125707003,
-      "loss": 46.0,
-      "step": 38133
-    },
-    {
-      "epoch": 6.141229518096542,
-      "grad_norm": 0.009144116193056107,
-      "learning_rate": 0.00019998140028136454,
-      "loss": 46.0,
-      "step": 38134
-    },
-    {
-      "epoch": 6.141390555175329,
-      "grad_norm": 0.005073470063507557,
-      "learning_rate": 0.00019998139930563346,
-      "loss": 46.0,
-      "step": 38135
-    },
-    {
-      "epoch": 6.141551592254117,
-      "grad_norm": 0.02522943541407585,
-      "learning_rate": 0.0001999813983298768,
-      "loss": 46.0,
-      "step": 38136
-    },
-    {
-      "epoch": 6.141712629332904,
-      "grad_norm": 0.01788550801575184,
-      "learning_rate": 0.00019998139735409451,
-      "loss": 46.0,
-      "step": 38137
-    },
-    {
-      "epoch": 6.1418736664116915,
-      "grad_norm": 0.0052560726180672646,
-      "learning_rate": 0.00019998139637828665,
-      "loss": 46.0,
-      "step": 38138
-    },
-    {
-      "epoch": 6.142034703490479,
-      "grad_norm": 0.005746690556406975,
-      "learning_rate": 0.00019998139540245322,
-      "loss": 46.0,
-      "step": 38139
-    },
-    {
-      "epoch": 6.142195740569266,
-      "grad_norm": 0.00949172955006361,
-      "learning_rate": 0.00019998139442659418,
-      "loss": 46.0,
-      "step": 38140
-    },
-    {
-      "epoch": 6.142356777648054,
-      "grad_norm": 0.0020084786228835583,
-      "learning_rate": 0.00019998139345070955,
-      "loss": 46.0,
-      "step": 38141
-    },
-    {
-      "epoch": 6.14251781472684,
-      "grad_norm": 0.010748251341283321,
-      "learning_rate": 0.00019998139247479933,
-      "loss": 46.0,
-      "step": 38142
-    },
-    {
-      "epoch": 6.142678851805628,
-      "grad_norm": 0.0010066042887046933,
-      "learning_rate": 0.00019998139149886353,
-      "loss": 46.0,
-      "step": 38143
-    },
-    {
-      "epoch": 6.142839888884415,
-      "grad_norm": 0.005377229768782854,
-      "learning_rate": 0.00019998139052290214,
-      "loss": 46.0,
-      "step": 38144
-    },
-    {
-      "epoch": 6.143000925963203,
-      "grad_norm": 0.00917661003768444,
-      "learning_rate": 0.00019998138954691513,
-      "loss": 46.0,
-      "step": 38145
-    },
-    {
-      "epoch": 6.14316196304199,
-      "grad_norm": 0.0022325965110212564,
-      "learning_rate": 0.0001999813885709026,
-      "loss": 46.0,
-      "step": 38146
-    },
-    {
-      "epoch": 6.143323000120778,
-      "grad_norm": 0.016158388927578926,
-      "learning_rate": 0.0001999813875948644,
-      "loss": 46.0,
-      "step": 38147
-    },
-    {
-      "epoch": 6.143484037199565,
-      "grad_norm": 0.007938831113278866,
-      "learning_rate": 0.00019998138661880065,
-      "loss": 46.0,
-      "step": 38148
-    },
-    {
-      "epoch": 6.143645074278353,
-      "grad_norm": 0.005012881942093372,
-      "learning_rate": 0.00019998138564271132,
-      "loss": 46.0,
-      "step": 38149
-    },
-    {
-      "epoch": 6.14380611135714,
-      "grad_norm": 0.007556302472949028,
-      "learning_rate": 0.00019998138466659638,
-      "loss": 46.0,
-      "step": 38150
-    },
-    {
-      "epoch": 6.1439671484359275,
-      "grad_norm": 0.0013723070733249187,
-      "learning_rate": 0.00019998138369045585,
-      "loss": 46.0,
-      "step": 38151
-    },
-    {
-      "epoch": 6.144128185514715,
-      "grad_norm": 0.006414035335183144,
-      "learning_rate": 0.00019998138271428974,
-      "loss": 46.0,
-      "step": 38152
-    },
-    {
-      "epoch": 6.144289222593502,
-      "grad_norm": 0.0014599489513784647,
-      "learning_rate": 0.00019998138173809804,
-      "loss": 46.0,
-      "step": 38153
-    },
-    {
-      "epoch": 6.14445025967229,
-      "grad_norm": 0.009400549344718456,
-      "learning_rate": 0.00019998138076188072,
-      "loss": 46.0,
-      "step": 38154
-    },
-    {
-      "epoch": 6.144611296751077,
-      "grad_norm": 0.0049391454085707664,
-      "learning_rate": 0.00019998137978563784,
-      "loss": 46.0,
-      "step": 38155
-    },
-    {
-      "epoch": 6.144772333829865,
-      "grad_norm": 0.02154109999537468,
-      "learning_rate": 0.00019998137880936935,
-      "loss": 46.0,
-      "step": 38156
-    },
-    {
-      "epoch": 6.144933370908651,
-      "grad_norm": 0.010342485271394253,
-      "learning_rate": 0.0001999813778330753,
-      "loss": 46.0,
-      "step": 38157
-    },
-    {
-      "epoch": 6.145094407987439,
-      "grad_norm": 0.0028692493215203285,
-      "learning_rate": 0.00019998137685675564,
-      "loss": 46.0,
-      "step": 38158
-    },
-    {
-      "epoch": 6.145255445066226,
-      "grad_norm": 0.012828550301492214,
-      "learning_rate": 0.0001999813758804104,
-      "loss": 46.0,
-      "step": 38159
-    },
-    {
-      "epoch": 6.145416482145014,
-      "grad_norm": 0.0009911274537444115,
-      "learning_rate": 0.00019998137490403957,
-      "loss": 46.0,
-      "step": 38160
-    },
-    {
-      "epoch": 6.145577519223801,
-      "grad_norm": 0.0066237300634384155,
-      "learning_rate": 0.00019998137392764314,
-      "loss": 46.0,
-      "step": 38161
-    },
-    {
-      "epoch": 6.145738556302589,
-      "grad_norm": 0.003932784777134657,
-      "learning_rate": 0.00019998137295122113,
-      "loss": 46.0,
-      "step": 38162
-    },
-    {
-      "epoch": 6.145899593381376,
-      "grad_norm": 0.0034112115390598774,
-      "learning_rate": 0.0001999813719747735,
-      "loss": 46.0,
-      "step": 38163
-    },
-    {
-      "epoch": 6.1460606304601635,
-      "grad_norm": 0.0019163420656695962,
-      "learning_rate": 0.0001999813709983003,
-      "loss": 46.0,
-      "step": 38164
-    },
-    {
-      "epoch": 6.146221667538951,
-      "grad_norm": 0.0012701962841674685,
-      "learning_rate": 0.00019998137002180154,
-      "loss": 46.0,
-      "step": 38165
-    },
-    {
-      "epoch": 6.146382704617738,
-      "grad_norm": 0.007408374920487404,
-      "learning_rate": 0.00019998136904527715,
-      "loss": 46.0,
-      "step": 38166
-    },
-    {
-      "epoch": 6.146543741696526,
-      "grad_norm": 0.00444489112123847,
-      "learning_rate": 0.00019998136806872717,
-      "loss": 46.0,
-      "step": 38167
-    },
-    {
-      "epoch": 6.146704778775313,
-      "grad_norm": 0.009543496184051037,
-      "learning_rate": 0.00019998136709215163,
-      "loss": 46.0,
-      "step": 38168
-    },
-    {
-      "epoch": 6.146865815854101,
-      "grad_norm": 0.0064465063624084,
-      "learning_rate": 0.00019998136611555048,
-      "loss": 46.0,
-      "step": 38169
-    },
-    {
-      "epoch": 6.147026852932888,
-      "grad_norm": 0.004339747130870819,
-      "learning_rate": 0.00019998136513892374,
-      "loss": 46.0,
-      "step": 38170
-    },
-    {
-      "epoch": 6.147187890011675,
-      "grad_norm": 0.008034233003854752,
-      "learning_rate": 0.00019998136416227142,
-      "loss": 46.0,
-      "step": 38171
-    },
-    {
-      "epoch": 6.147348927090462,
-      "grad_norm": 0.007363523822277784,
-      "learning_rate": 0.00019998136318559348,
-      "loss": 46.0,
-      "step": 38172
-    },
-    {
-      "epoch": 6.14750996416925,
-      "grad_norm": 0.013469934463500977,
-      "learning_rate": 0.00019998136220888998,
-      "loss": 46.0,
-      "step": 38173
-    },
-    {
-      "epoch": 6.147671001248037,
-      "grad_norm": 0.013557642698287964,
-      "learning_rate": 0.00019998136123216086,
-      "loss": 46.0,
-      "step": 38174
-    },
-    {
-      "epoch": 6.147832038326825,
-      "grad_norm": 0.004900000989437103,
-      "learning_rate": 0.0001999813602554062,
-      "loss": 46.0,
-      "step": 38175
-    },
-    {
-      "epoch": 6.147993075405612,
-      "grad_norm": 0.0017434413312003016,
-      "learning_rate": 0.0001999813592786259,
-      "loss": 46.0,
-      "step": 38176
-    },
-    {
-      "epoch": 6.1481541124843995,
-      "grad_norm": 0.0008235702407546341,
-      "learning_rate": 0.00019998135830182005,
-      "loss": 46.0,
-      "step": 38177
-    },
-    {
-      "epoch": 6.148315149563187,
-      "grad_norm": 0.006556585896760225,
-      "learning_rate": 0.0001999813573249886,
-      "loss": 46.0,
-      "step": 38178
-    },
-    {
-      "epoch": 6.148476186641974,
-      "grad_norm": 0.017589643597602844,
-      "learning_rate": 0.00019998135634813157,
-      "loss": 46.0,
-      "step": 38179
-    },
-    {
-      "epoch": 6.148637223720762,
-      "grad_norm": 0.0028431490063667297,
-      "learning_rate": 0.0001999813553712489,
-      "loss": 46.0,
-      "step": 38180
-    },
-    {
-      "epoch": 6.148798260799549,
-      "grad_norm": 0.0018631118582561612,
-      "learning_rate": 0.00019998135439434068,
-      "loss": 46.0,
-      "step": 38181
-    },
-    {
-      "epoch": 6.148959297878337,
-      "grad_norm": 0.00662678899243474,
-      "learning_rate": 0.00019998135341740684,
-      "loss": 46.0,
-      "step": 38182
-    },
-    {
-      "epoch": 6.149120334957124,
-      "grad_norm": 0.012830279767513275,
-      "learning_rate": 0.00019998135244044744,
-      "loss": 46.0,
-      "step": 38183
-    },
-    {
-      "epoch": 6.149281372035912,
-      "grad_norm": 0.004706101026386023,
-      "learning_rate": 0.00019998135146346243,
-      "loss": 46.0,
-      "step": 38184
-    },
-    {
-      "epoch": 6.149442409114699,
-      "grad_norm": 0.008113808929920197,
-      "learning_rate": 0.00019998135048645185,
-      "loss": 46.0,
-      "step": 38185
-    },
-    {
-      "epoch": 6.149603446193486,
-      "grad_norm": 0.018631508573889732,
-      "learning_rate": 0.00019998134950941567,
-      "loss": 46.0,
-      "step": 38186
-    },
-    {
-      "epoch": 6.149764483272273,
-      "grad_norm": 0.0024354676716029644,
-      "learning_rate": 0.00019998134853235392,
-      "loss": 46.0,
-      "step": 38187
-    },
-    {
-      "epoch": 6.1499255203510605,
-      "grad_norm": 0.0007428895914927125,
-      "learning_rate": 0.00019998134755526653,
-      "loss": 46.0,
-      "step": 38188
-    },
-    {
-      "epoch": 6.150086557429848,
-      "grad_norm": 0.017941290512681007,
-      "learning_rate": 0.00019998134657815358,
-      "loss": 46.0,
-      "step": 38189
-    },
-    {
-      "epoch": 6.1502475945086355,
-      "grad_norm": 0.004254871513694525,
-      "learning_rate": 0.00019998134560101507,
-      "loss": 46.0,
-      "step": 38190
-    },
-    {
-      "epoch": 6.150408631587423,
-      "grad_norm": 0.002839587163180113,
-      "learning_rate": 0.00019998134462385092,
-      "loss": 46.0,
-      "step": 38191
-    },
-    {
-      "epoch": 6.15056966866621,
-      "grad_norm": 0.005300450138747692,
-      "learning_rate": 0.00019998134364666118,
-      "loss": 46.0,
-      "step": 38192
-    },
-    {
-      "epoch": 6.150730705744998,
-      "grad_norm": 0.0015842207940295339,
-      "learning_rate": 0.00019998134266944591,
-      "loss": 46.0,
-      "step": 38193
-    },
-    {
-      "epoch": 6.150891742823785,
-      "grad_norm": 0.006529216188937426,
-      "learning_rate": 0.000199981341692205,
-      "loss": 46.0,
-      "step": 38194
-    },
-    {
-      "epoch": 6.151052779902573,
-      "grad_norm": 0.0021917361300438643,
-      "learning_rate": 0.0001999813407149385,
-      "loss": 46.0,
-      "step": 38195
-    },
-    {
-      "epoch": 6.15121381698136,
-      "grad_norm": 0.015276730060577393,
-      "learning_rate": 0.00019998133973764644,
-      "loss": 46.0,
-      "step": 38196
-    },
-    {
-      "epoch": 6.151374854060148,
-      "grad_norm": 0.0172035600990057,
-      "learning_rate": 0.00019998133876032874,
-      "loss": 46.0,
-      "step": 38197
-    },
-    {
-      "epoch": 6.151535891138935,
-      "grad_norm": 0.008346655406057835,
-      "learning_rate": 0.00019998133778298548,
-      "loss": 46.0,
-      "step": 38198
-    },
-    {
-      "epoch": 6.1516969282177225,
-      "grad_norm": 0.011320813558995724,
-      "learning_rate": 0.00019998133680561664,
-      "loss": 46.0,
-      "step": 38199
-    },
-    {
-      "epoch": 6.151857965296509,
-      "grad_norm": 0.004653043579310179,
-      "learning_rate": 0.0001999813358282222,
-      "loss": 46.0,
-      "step": 38200
-    },
-    {
-      "epoch": 6.1520190023752965,
-      "grad_norm": 0.0027833282947540283,
-      "learning_rate": 0.00019998133485080218,
-      "loss": 46.0,
-      "step": 38201
-    },
-    {
-      "epoch": 6.152180039454084,
-      "grad_norm": 0.007442052476108074,
-      "learning_rate": 0.00019998133387335654,
-      "loss": 46.0,
-      "step": 38202
-    },
-    {
-      "epoch": 6.152341076532871,
-      "grad_norm": 0.0020572366192936897,
-      "learning_rate": 0.00019998133289588532,
-      "loss": 46.0,
-      "step": 38203
-    },
-    {
-      "epoch": 6.152502113611659,
-      "grad_norm": 0.005303728394210339,
-      "learning_rate": 0.00019998133191838853,
-      "loss": 46.0,
-      "step": 38204
-    },
-    {
-      "epoch": 6.152663150690446,
-      "grad_norm": 0.02952386625111103,
-      "learning_rate": 0.00019998133094086616,
-      "loss": 46.0,
-      "step": 38205
-    },
-    {
-      "epoch": 6.152824187769234,
-      "grad_norm": 0.0012008358025923371,
-      "learning_rate": 0.00019998132996331815,
-      "loss": 46.0,
-      "step": 38206
-    },
-    {
-      "epoch": 6.152985224848021,
-      "grad_norm": 0.01046083401888609,
-      "learning_rate": 0.00019998132898574458,
-      "loss": 46.0,
-      "step": 38207
-    },
-    {
-      "epoch": 6.153146261926809,
-      "grad_norm": 0.004147620871663094,
-      "learning_rate": 0.00019998132800814542,
-      "loss": 46.0,
-      "step": 38208
-    },
-    {
-      "epoch": 6.153307299005596,
-      "grad_norm": 0.002657199278473854,
-      "learning_rate": 0.00019998132703052067,
-      "loss": 46.0,
-      "step": 38209
-    },
-    {
-      "epoch": 6.153468336084384,
-      "grad_norm": 0.010750516317784786,
-      "learning_rate": 0.0001999813260528703,
-      "loss": 46.0,
-      "step": 38210
-    },
-    {
-      "epoch": 6.153629373163171,
-      "grad_norm": 0.00933627039194107,
-      "learning_rate": 0.0001999813250751944,
-      "loss": 46.0,
-      "step": 38211
-    },
-    {
-      "epoch": 6.1537904102419585,
-      "grad_norm": 0.0017052008770406246,
-      "learning_rate": 0.00019998132409749288,
-      "loss": 46.0,
-      "step": 38212
-    },
-    {
-      "epoch": 6.153951447320746,
-      "grad_norm": 0.0023985765874385834,
-      "learning_rate": 0.00019998132311976576,
-      "loss": 46.0,
-      "step": 38213
-    },
-    {
-      "epoch": 6.154112484399533,
-      "grad_norm": 0.0024245830718427896,
-      "learning_rate": 0.00019998132214201308,
-      "loss": 46.0,
-      "step": 38214
-    },
-    {
-      "epoch": 6.15427352147832,
-      "grad_norm": 0.004561722278594971,
-      "learning_rate": 0.00019998132116423478,
-      "loss": 46.0,
-      "step": 38215
-    },
-    {
-      "epoch": 6.154434558557107,
-      "grad_norm": 0.004155233968049288,
-      "learning_rate": 0.0001999813201864309,
-      "loss": 46.0,
-      "step": 38216
-    },
-    {
-      "epoch": 6.154595595635895,
-      "grad_norm": 0.017093494534492493,
-      "learning_rate": 0.00019998131920860143,
-      "loss": 46.0,
-      "step": 38217
-    },
-    {
-      "epoch": 6.154756632714682,
-      "grad_norm": 0.003951186314225197,
-      "learning_rate": 0.00019998131823074634,
-      "loss": 46.0,
-      "step": 38218
-    },
-    {
-      "epoch": 6.15491766979347,
-      "grad_norm": 0.008156186901032925,
-      "learning_rate": 0.0001999813172528657,
-      "loss": 46.0,
-      "step": 38219
-    },
-    {
-      "epoch": 6.155078706872257,
-      "grad_norm": 0.0031261479016393423,
-      "learning_rate": 0.00019998131627495946,
-      "loss": 46.0,
-      "step": 38220
-    },
-    {
-      "epoch": 6.155239743951045,
-      "grad_norm": 0.005263959988951683,
-      "learning_rate": 0.00019998131529702762,
-      "loss": 46.0,
-      "step": 38221
-    },
-    {
-      "epoch": 6.155400781029832,
-      "grad_norm": 0.014790375716984272,
-      "learning_rate": 0.0001999813143190702,
-      "loss": 46.0,
-      "step": 38222
-    },
-    {
-      "epoch": 6.15556181810862,
-      "grad_norm": 0.0013469933765009046,
-      "learning_rate": 0.0001999813133410872,
-      "loss": 46.0,
-      "step": 38223
-    },
-    {
-      "epoch": 6.155722855187407,
-      "grad_norm": 0.0006598061299882829,
-      "learning_rate": 0.00019998131236307858,
-      "loss": 46.0,
-      "step": 38224
-    },
-    {
-      "epoch": 6.1558838922661945,
-      "grad_norm": 0.0046853115782141685,
-      "learning_rate": 0.00019998131138504439,
-      "loss": 46.0,
-      "step": 38225
-    },
-    {
-      "epoch": 6.156044929344982,
-      "grad_norm": 0.011621733196079731,
-      "learning_rate": 0.00019998131040698463,
-      "loss": 46.0,
-      "step": 38226
-    },
-    {
-      "epoch": 6.156205966423769,
-      "grad_norm": 0.02308325655758381,
-      "learning_rate": 0.00019998130942889923,
-      "loss": 46.0,
-      "step": 38227
-    },
-    {
-      "epoch": 6.156367003502557,
-      "grad_norm": 0.004738691728562117,
-      "learning_rate": 0.00019998130845078828,
-      "loss": 46.0,
-      "step": 38228
-    },
-    {
-      "epoch": 6.156528040581344,
-      "grad_norm": 0.008279988542199135,
-      "learning_rate": 0.0001999813074726517,
-      "loss": 46.0,
-      "step": 38229
-    },
-    {
-      "epoch": 6.156689077660131,
-      "grad_norm": 0.001596157904714346,
-      "learning_rate": 0.00019998130649448957,
-      "loss": 46.0,
-      "step": 38230
-    },
-    {
-      "epoch": 6.156850114738918,
-      "grad_norm": 0.014751922339200974,
-      "learning_rate": 0.00019998130551630183,
-      "loss": 46.0,
-      "step": 38231
-    },
-    {
-      "epoch": 6.157011151817706,
-      "grad_norm": 0.014020483009517193,
-      "learning_rate": 0.00019998130453808852,
-      "loss": 46.0,
-      "step": 38232
-    },
-    {
-      "epoch": 6.157172188896493,
-      "grad_norm": 0.0033833379857242107,
-      "learning_rate": 0.0001999813035598496,
-      "loss": 46.0,
-      "step": 38233
-    },
-    {
-      "epoch": 6.157333225975281,
-      "grad_norm": 0.0010672836797311902,
-      "learning_rate": 0.0001999813025815851,
-      "loss": 46.0,
-      "step": 38234
-    },
-    {
-      "epoch": 6.157494263054068,
-      "grad_norm": 0.0030781817622482777,
-      "learning_rate": 0.00019998130160329503,
-      "loss": 46.0,
-      "step": 38235
-    },
-    {
-      "epoch": 6.157655300132856,
-      "grad_norm": 0.01249680481851101,
-      "learning_rate": 0.00019998130062497932,
-      "loss": 46.0,
-      "step": 38236
-    },
-    {
-      "epoch": 6.157816337211643,
-      "grad_norm": 0.018243996426463127,
-      "learning_rate": 0.00019998129964663805,
-      "loss": 46.0,
-      "step": 38237
-    },
-    {
-      "epoch": 6.1579773742904305,
-      "grad_norm": 0.009122656658291817,
-      "learning_rate": 0.0001999812986682712,
-      "loss": 46.0,
-      "step": 38238
-    },
-    {
-      "epoch": 6.158138411369218,
-      "grad_norm": 0.007188413292169571,
-      "learning_rate": 0.00019998129768987872,
-      "loss": 46.0,
-      "step": 38239
-    },
-    {
-      "epoch": 6.158299448448005,
-      "grad_norm": 0.007149084471166134,
-      "learning_rate": 0.0001999812967114607,
-      "loss": 46.0,
-      "step": 38240
-    },
-    {
-      "epoch": 6.158460485526793,
-      "grad_norm": 0.0060903532430529594,
-      "learning_rate": 0.00019998129573301708,
-      "loss": 46.0,
-      "step": 38241
-    },
-    {
-      "epoch": 6.15862152260558,
-      "grad_norm": 0.006014631595462561,
-      "learning_rate": 0.00019998129475454784,
-      "loss": 46.0,
-      "step": 38242
-    },
-    {
-      "epoch": 6.158782559684368,
-      "grad_norm": 0.00504056503996253,
-      "learning_rate": 0.00019998129377605303,
-      "loss": 46.0,
-      "step": 38243
-    },
-    {
-      "epoch": 6.158943596763155,
-      "grad_norm": 0.0018291643355041742,
-      "learning_rate": 0.00019998129279753262,
-      "loss": 46.0,
-      "step": 38244
-    },
-    {
-      "epoch": 6.159104633841942,
-      "grad_norm": 0.001553820795379579,
-      "learning_rate": 0.00019998129181898663,
-      "loss": 46.0,
-      "step": 38245
-    },
-    {
-      "epoch": 6.159265670920729,
-      "grad_norm": 0.004942046478390694,
-      "learning_rate": 0.00019998129084041505,
-      "loss": 46.0,
-      "step": 38246
-    },
-    {
-      "epoch": 6.159426707999517,
-      "grad_norm": 0.003221700666472316,
-      "learning_rate": 0.00019998128986181785,
-      "loss": 46.0,
-      "step": 38247
-    },
-    {
-      "epoch": 6.159587745078304,
-      "grad_norm": 0.0010458199540153146,
-      "learning_rate": 0.0001999812888831951,
-      "loss": 46.0,
-      "step": 38248
-    },
-    {
-      "epoch": 6.1597487821570915,
-      "grad_norm": 0.009848294779658318,
-      "learning_rate": 0.00019998128790454673,
-      "loss": 46.0,
-      "step": 38249
-    },
-    {
-      "epoch": 6.159909819235879,
-      "grad_norm": 0.0064690001308918,
-      "learning_rate": 0.0001999812869258728,
-      "loss": 46.0,
-      "step": 38250
-    },
-    {
-      "epoch": 6.160070856314666,
-      "grad_norm": 0.0045645879581570625,
-      "learning_rate": 0.00019998128594717326,
-      "loss": 46.0,
-      "step": 38251
-    },
-    {
-      "epoch": 6.160231893393454,
-      "grad_norm": 0.005330066196620464,
-      "learning_rate": 0.00019998128496844815,
-      "loss": 46.0,
-      "step": 38252
-    },
-    {
-      "epoch": 6.160392930472241,
-      "grad_norm": 0.006402960047125816,
-      "learning_rate": 0.0001999812839896974,
-      "loss": 46.0,
-      "step": 38253
-    },
-    {
-      "epoch": 6.160553967551029,
-      "grad_norm": 0.01867598108947277,
-      "learning_rate": 0.0001999812830109211,
-      "loss": 46.0,
-      "step": 38254
-    },
-    {
-      "epoch": 6.160715004629816,
-      "grad_norm": 0.0015556865837424994,
-      "learning_rate": 0.0001999812820321192,
-      "loss": 46.0,
-      "step": 38255
-    },
-    {
-      "epoch": 6.160876041708604,
-      "grad_norm": 0.002903979504480958,
-      "learning_rate": 0.00019998128105329173,
-      "loss": 46.0,
-      "step": 38256
-    },
-    {
-      "epoch": 6.161037078787391,
-      "grad_norm": 0.00995730608701706,
-      "learning_rate": 0.00019998128007443864,
-      "loss": 46.0,
-      "step": 38257
-    },
-    {
-      "epoch": 6.161198115866179,
-      "grad_norm": 0.001089159632101655,
-      "learning_rate": 0.00019998127909555999,
-      "loss": 46.0,
-      "step": 38258
-    },
-    {
-      "epoch": 6.161359152944965,
-      "grad_norm": 0.0021435206290334463,
-      "learning_rate": 0.00019998127811665572,
-      "loss": 46.0,
-      "step": 38259
-    },
-    {
-      "epoch": 6.161520190023753,
-      "grad_norm": 0.006833336781710386,
-      "learning_rate": 0.0001999812771377259,
-      "loss": 46.0,
-      "step": 38260
-    },
-    {
-      "epoch": 6.16168122710254,
-      "grad_norm": 0.0033831140026450157,
-      "learning_rate": 0.00019998127615877045,
-      "loss": 46.0,
-      "step": 38261
-    },
-    {
-      "epoch": 6.1618422641813275,
-      "grad_norm": 0.002358735539019108,
-      "learning_rate": 0.00019998127517978942,
-      "loss": 46.0,
-      "step": 38262
-    },
-    {
-      "epoch": 6.162003301260115,
-      "grad_norm": 0.002337033860385418,
-      "learning_rate": 0.0001999812742007828,
-      "loss": 46.0,
-      "step": 38263
-    },
-    {
-      "epoch": 6.162164338338902,
-      "grad_norm": 0.009295708499848843,
-      "learning_rate": 0.0001999812732217506,
-      "loss": 46.0,
-      "step": 38264
-    },
-    {
-      "epoch": 6.16232537541769,
-      "grad_norm": 0.011148395016789436,
-      "learning_rate": 0.0001999812722426928,
-      "loss": 46.0,
-      "step": 38265
-    },
-    {
-      "epoch": 6.162486412496477,
-      "grad_norm": 0.004165736027061939,
-      "learning_rate": 0.0001999812712636094,
-      "loss": 46.0,
-      "step": 38266
-    },
-    {
-      "epoch": 6.162647449575265,
-      "grad_norm": 0.0015171512495726347,
-      "learning_rate": 0.00019998127028450044,
-      "loss": 46.0,
-      "step": 38267
-    },
-    {
-      "epoch": 6.162808486654052,
-      "grad_norm": 0.001808301080018282,
-      "learning_rate": 0.00019998126930536586,
-      "loss": 46.0,
-      "step": 38268
-    },
-    {
-      "epoch": 6.16296952373284,
-      "grad_norm": 0.002995187882333994,
-      "learning_rate": 0.00019998126832620572,
-      "loss": 46.0,
-      "step": 38269
-    },
-    {
-      "epoch": 6.163130560811627,
-      "grad_norm": 0.005926748272031546,
-      "learning_rate": 0.00019998126734701996,
-      "loss": 46.0,
-      "step": 38270
-    },
-    {
-      "epoch": 6.163291597890415,
-      "grad_norm": 0.002057367004454136,
-      "learning_rate": 0.00019998126636780862,
-      "loss": 46.0,
-      "step": 38271
-    },
-    {
-      "epoch": 6.163452634969202,
-      "grad_norm": 0.0014211698435246944,
-      "learning_rate": 0.00019998126538857172,
-      "loss": 46.0,
-      "step": 38272
-    },
-    {
-      "epoch": 6.163613672047989,
-      "grad_norm": 0.0036514876410365105,
-      "learning_rate": 0.00019998126440930918,
-      "loss": 46.0,
-      "step": 38273
-    },
-    {
-      "epoch": 6.163774709126776,
-      "grad_norm": 0.006783165503293276,
-      "learning_rate": 0.00019998126343002108,
-      "loss": 46.0,
-      "step": 38274
-    },
-    {
-      "epoch": 6.1639357462055635,
-      "grad_norm": 0.007316098548471928,
-      "learning_rate": 0.0001999812624507074,
-      "loss": 46.0,
-      "step": 38275
-    },
-    {
-      "epoch": 6.164096783284351,
-      "grad_norm": 0.0031602047383785248,
-      "learning_rate": 0.00019998126147136808,
-      "loss": 46.0,
-      "step": 38276
-    },
-    {
-      "epoch": 6.164257820363138,
-      "grad_norm": 0.00608556205406785,
-      "learning_rate": 0.00019998126049200322,
-      "loss": 46.0,
-      "step": 38277
-    },
-    {
-      "epoch": 6.164418857441926,
-      "grad_norm": 0.006487439386546612,
-      "learning_rate": 0.00019998125951261274,
-      "loss": 46.0,
-      "step": 38278
-    },
-    {
-      "epoch": 6.164579894520713,
-      "grad_norm": 0.0024090849328786135,
-      "learning_rate": 0.0001999812585331967,
-      "loss": 46.0,
-      "step": 38279
-    },
-    {
-      "epoch": 6.164740931599501,
-      "grad_norm": 0.0014213065151125193,
-      "learning_rate": 0.00019998125755375505,
-      "loss": 46.0,
-      "step": 38280
-    },
-    {
-      "epoch": 6.164901968678288,
-      "grad_norm": 0.0029977899976074696,
-      "learning_rate": 0.0001999812565742878,
-      "loss": 46.0,
-      "step": 38281
-    },
-    {
-      "epoch": 6.165063005757076,
-      "grad_norm": 0.0020964420400559902,
-      "learning_rate": 0.00019998125559479498,
-      "loss": 46.0,
-      "step": 38282
-    },
-    {
-      "epoch": 6.165224042835863,
-      "grad_norm": 0.006059358827769756,
-      "learning_rate": 0.00019998125461527657,
-      "loss": 46.0,
-      "step": 38283
-    },
-    {
-      "epoch": 6.165385079914651,
-      "grad_norm": 0.003123324830085039,
-      "learning_rate": 0.00019998125363573257,
-      "loss": 46.0,
-      "step": 38284
-    },
-    {
-      "epoch": 6.165546116993438,
-      "grad_norm": 0.002225477248430252,
-      "learning_rate": 0.00019998125265616295,
-      "loss": 46.0,
-      "step": 38285
-    },
-    {
-      "epoch": 6.1657071540722255,
-      "grad_norm": 0.008417419157922268,
-      "learning_rate": 0.00019998125167656778,
-      "loss": 46.0,
-      "step": 38286
-    },
-    {
-      "epoch": 6.165868191151013,
-      "grad_norm": 0.003415049985051155,
-      "learning_rate": 0.000199981250696947,
-      "loss": 46.0,
-      "step": 38287
-    },
-    {
-      "epoch": 6.1660292282297995,
-      "grad_norm": 0.0019182567484676838,
-      "learning_rate": 0.00019998124971730064,
-      "loss": 46.0,
-      "step": 38288
-    },
-    {
-      "epoch": 6.166190265308587,
-      "grad_norm": 0.0014812789158895612,
-      "learning_rate": 0.00019998124873762867,
-      "loss": 46.0,
-      "step": 38289
-    },
-    {
-      "epoch": 6.166351302387374,
-      "grad_norm": 0.005335068330168724,
-      "learning_rate": 0.00019998124775793112,
-      "loss": 46.0,
-      "step": 38290
-    },
-    {
-      "epoch": 6.166512339466162,
-      "grad_norm": 0.0017525829607620835,
-      "learning_rate": 0.00019998124677820798,
-      "loss": 46.0,
-      "step": 38291
-    },
-    {
-      "epoch": 6.166673376544949,
-      "grad_norm": 0.0102470638230443,
-      "learning_rate": 0.00019998124579845926,
-      "loss": 46.0,
-      "step": 38292
-    },
-    {
-      "epoch": 6.166834413623737,
-      "grad_norm": 0.004462293814867735,
-      "learning_rate": 0.00019998124481868495,
-      "loss": 46.0,
-      "step": 38293
-    },
-    {
-      "epoch": 6.166995450702524,
-      "grad_norm": 0.011110853403806686,
-      "learning_rate": 0.00019998124383888502,
-      "loss": 46.0,
-      "step": 38294
-    },
-    {
-      "epoch": 6.167156487781312,
-      "grad_norm": 0.00249012210406363,
-      "learning_rate": 0.00019998124285905953,
-      "loss": 46.0,
-      "step": 38295
-    },
-    {
-      "epoch": 6.167317524860099,
-      "grad_norm": 0.00370098277926445,
-      "learning_rate": 0.00019998124187920843,
-      "loss": 46.0,
-      "step": 38296
-    },
-    {
-      "epoch": 6.1674785619388865,
-      "grad_norm": 0.00599149102345109,
-      "learning_rate": 0.00019998124089933177,
-      "loss": 46.0,
-      "step": 38297
-    },
-    {
-      "epoch": 6.167639599017674,
-      "grad_norm": 0.007958956994116306,
-      "learning_rate": 0.0001999812399194295,
-      "loss": 46.0,
-      "step": 38298
-    },
-    {
-      "epoch": 6.1678006360964615,
-      "grad_norm": 0.0014243399491533637,
-      "learning_rate": 0.00019998123893950163,
-      "loss": 46.0,
-      "step": 38299
-    },
-    {
-      "epoch": 6.167961673175249,
-      "grad_norm": 0.010591837577521801,
-      "learning_rate": 0.00019998123795954818,
-      "loss": 46.0,
-      "step": 38300
-    },
-    {
-      "epoch": 6.168122710254036,
-      "grad_norm": 0.007675534579902887,
-      "learning_rate": 0.00019998123697956914,
-      "loss": 46.0,
-      "step": 38301
-    },
-    {
-      "epoch": 6.168283747332824,
-      "grad_norm": 0.0066420529037714005,
-      "learning_rate": 0.00019998123599956454,
-      "loss": 46.0,
-      "step": 38302
-    },
-    {
-      "epoch": 6.16844478441161,
-      "grad_norm": 0.0013729004422202706,
-      "learning_rate": 0.0001999812350195343,
-      "loss": 46.0,
-      "step": 38303
-    },
-    {
-      "epoch": 6.168605821490398,
-      "grad_norm": 0.0026360023766756058,
-      "learning_rate": 0.00019998123403947848,
-      "loss": 46.0,
-      "step": 38304
-    },
-    {
-      "epoch": 6.168766858569185,
-      "grad_norm": 0.001710473676212132,
-      "learning_rate": 0.0001999812330593971,
-      "loss": 46.0,
-      "step": 38305
-    },
-    {
-      "epoch": 6.168927895647973,
-      "grad_norm": 0.00616045854985714,
-      "learning_rate": 0.00019998123207929012,
-      "loss": 46.0,
-      "step": 38306
-    },
-    {
-      "epoch": 6.16908893272676,
-      "grad_norm": 0.0023090096656233072,
-      "learning_rate": 0.0001999812310991575,
-      "loss": 46.0,
-      "step": 38307
-    },
-    {
-      "epoch": 6.169249969805548,
-      "grad_norm": 0.007716451771557331,
-      "learning_rate": 0.00019998123011899933,
-      "loss": 46.0,
-      "step": 38308
-    },
-    {
-      "epoch": 6.169411006884335,
-      "grad_norm": 0.002946835244074464,
-      "learning_rate": 0.0001999812291388156,
-      "loss": 46.0,
-      "step": 38309
-    },
-    {
-      "epoch": 6.1695720439631225,
-      "grad_norm": 0.005844874307513237,
-      "learning_rate": 0.00019998122815860625,
-      "loss": 46.0,
-      "step": 38310
-    },
-    {
-      "epoch": 6.16973308104191,
-      "grad_norm": 0.005737279541790485,
-      "learning_rate": 0.00019998122717837128,
-      "loss": 46.0,
-      "step": 38311
-    },
-    {
-      "epoch": 6.169894118120697,
-      "grad_norm": 0.01215299405157566,
-      "learning_rate": 0.00019998122619811076,
-      "loss": 46.0,
-      "step": 38312
-    },
-    {
-      "epoch": 6.170055155199485,
-      "grad_norm": 0.012855935841798782,
-      "learning_rate": 0.00019998122521782465,
-      "loss": 46.0,
-      "step": 38313
-    },
-    {
-      "epoch": 6.170216192278272,
-      "grad_norm": 0.0019637797959148884,
-      "learning_rate": 0.00019998122423751293,
-      "loss": 46.0,
-      "step": 38314
-    },
-    {
-      "epoch": 6.17037722935706,
-      "grad_norm": 0.011320100165903568,
-      "learning_rate": 0.00019998122325717564,
-      "loss": 46.0,
-      "step": 38315
-    },
-    {
-      "epoch": 6.170538266435847,
-      "grad_norm": 0.0019749843049794436,
-      "learning_rate": 0.00019998122227681274,
-      "loss": 46.0,
-      "step": 38316
-    },
-    {
-      "epoch": 6.170699303514635,
-      "grad_norm": 0.0028116346802562475,
-      "learning_rate": 0.00019998122129642428,
-      "loss": 46.0,
-      "step": 38317
-    },
-    {
-      "epoch": 6.170860340593421,
-      "grad_norm": 0.0025257417000830173,
-      "learning_rate": 0.0001999812203160102,
-      "loss": 46.0,
-      "step": 38318
-    },
-    {
-      "epoch": 6.171021377672209,
-      "grad_norm": 0.007057895418256521,
-      "learning_rate": 0.00019998121933557055,
-      "loss": 46.0,
-      "step": 38319
-    },
-    {
-      "epoch": 6.171182414750996,
-      "grad_norm": 0.0011920987162739038,
-      "learning_rate": 0.0001999812183551053,
-      "loss": 46.0,
-      "step": 38320
-    },
-    {
-      "epoch": 6.171343451829784,
-      "grad_norm": 0.0020475653000175953,
-      "learning_rate": 0.00019998121737461447,
-      "loss": 46.0,
-      "step": 38321
-    },
-    {
-      "epoch": 6.171504488908571,
-      "grad_norm": 0.0026233065873384476,
-      "learning_rate": 0.00019998121639409804,
-      "loss": 46.0,
-      "step": 38322
-    },
-    {
-      "epoch": 6.1716655259873585,
-      "grad_norm": 0.006268750876188278,
-      "learning_rate": 0.00019998121541355598,
-      "loss": 46.0,
-      "step": 38323
-    },
-    {
-      "epoch": 6.171826563066146,
-      "grad_norm": 0.00309157301671803,
-      "learning_rate": 0.0001999812144329884,
-      "loss": 46.0,
-      "step": 38324
-    },
-    {
-      "epoch": 6.171987600144933,
-      "grad_norm": 0.0013755416730418801,
-      "learning_rate": 0.0001999812134523952,
-      "loss": 46.0,
-      "step": 38325
-    },
-    {
-      "epoch": 6.172148637223721,
-      "grad_norm": 0.0016376717248931527,
-      "learning_rate": 0.0001999812124717764,
-      "loss": 46.0,
-      "step": 38326
-    },
-    {
-      "epoch": 6.172309674302508,
-      "grad_norm": 0.008242752403020859,
-      "learning_rate": 0.00019998121149113202,
-      "loss": 46.0,
-      "step": 38327
-    },
-    {
-      "epoch": 6.172470711381296,
-      "grad_norm": 0.0016679811524227262,
-      "learning_rate": 0.00019998121051046207,
-      "loss": 46.0,
-      "step": 38328
-    },
-    {
-      "epoch": 6.172631748460083,
-      "grad_norm": 0.006396996323019266,
-      "learning_rate": 0.00019998120952976648,
-      "loss": 46.0,
-      "step": 38329
-    },
-    {
-      "epoch": 6.172792785538871,
-      "grad_norm": 0.00371205760166049,
-      "learning_rate": 0.00019998120854904534,
-      "loss": 46.0,
-      "step": 38330
-    },
-    {
-      "epoch": 6.172953822617658,
-      "grad_norm": 0.0054909847676754,
-      "learning_rate": 0.0001999812075682986,
-      "loss": 46.0,
-      "step": 38331
-    },
-    {
-      "epoch": 6.173114859696445,
-      "grad_norm": 0.004582086578011513,
-      "learning_rate": 0.00019998120658752626,
-      "loss": 46.0,
-      "step": 38332
-    },
-    {
-      "epoch": 6.173275896775232,
-      "grad_norm": 0.0013593441108241677,
-      "learning_rate": 0.00019998120560672835,
-      "loss": 46.0,
-      "step": 38333
-    },
-    {
-      "epoch": 6.17343693385402,
-      "grad_norm": 0.01412205770611763,
-      "learning_rate": 0.00019998120462590485,
-      "loss": 46.0,
-      "step": 38334
-    },
-    {
-      "epoch": 6.173597970932807,
-      "grad_norm": 0.007406009826809168,
-      "learning_rate": 0.00019998120364505572,
-      "loss": 46.0,
-      "step": 38335
-    },
-    {
-      "epoch": 6.1737590080115945,
-      "grad_norm": 0.002195616252720356,
-      "learning_rate": 0.00019998120266418105,
-      "loss": 46.0,
-      "step": 38336
-    },
-    {
-      "epoch": 6.173920045090382,
-      "grad_norm": 0.00959303230047226,
-      "learning_rate": 0.00019998120168328076,
-      "loss": 46.0,
-      "step": 38337
-    },
-    {
-      "epoch": 6.174081082169169,
-      "grad_norm": 0.003236561780795455,
-      "learning_rate": 0.0001999812007023549,
-      "loss": 46.0,
-      "step": 38338
-    },
-    {
-      "epoch": 6.174242119247957,
-      "grad_norm": 0.006811128929257393,
-      "learning_rate": 0.00019998119972140343,
-      "loss": 46.0,
-      "step": 38339
-    },
-    {
-      "epoch": 6.174403156326744,
-      "grad_norm": 0.008931980468332767,
-      "learning_rate": 0.0001999811987404264,
-      "loss": 46.0,
-      "step": 38340
-    },
-    {
-      "epoch": 6.174564193405532,
-      "grad_norm": 0.007095687557011843,
-      "learning_rate": 0.00019998119775942375,
-      "loss": 46.0,
-      "step": 38341
-    },
-    {
-      "epoch": 6.174725230484319,
-      "grad_norm": 0.002567967865616083,
-      "learning_rate": 0.0001999811967783955,
-      "loss": 46.0,
-      "step": 38342
-    },
-    {
-      "epoch": 6.174886267563107,
-      "grad_norm": 0.002297481754794717,
-      "learning_rate": 0.0001999811957973417,
-      "loss": 46.0,
-      "step": 38343
-    },
-    {
-      "epoch": 6.175047304641894,
-      "grad_norm": 0.0025986151304095984,
-      "learning_rate": 0.00019998119481626228,
-      "loss": 46.0,
-      "step": 38344
-    },
-    {
-      "epoch": 6.175208341720682,
-      "grad_norm": 0.0014254905981943011,
-      "learning_rate": 0.00019998119383515727,
-      "loss": 46.0,
-      "step": 38345
-    },
-    {
-      "epoch": 6.175369378799468,
-      "grad_norm": 0.006875585298985243,
-      "learning_rate": 0.00019998119285402668,
-      "loss": 46.0,
-      "step": 38346
-    },
-    {
-      "epoch": 6.175530415878256,
-      "grad_norm": 0.003062509698793292,
-      "learning_rate": 0.0001999811918728705,
-      "loss": 46.0,
-      "step": 38347
-    },
-    {
-      "epoch": 6.175691452957043,
-      "grad_norm": 0.0026650105137377977,
-      "learning_rate": 0.00019998119089168875,
-      "loss": 46.0,
-      "step": 38348
-    },
-    {
-      "epoch": 6.1758524900358305,
-      "grad_norm": 0.00251230550929904,
-      "learning_rate": 0.00019998118991048137,
-      "loss": 46.0,
-      "step": 38349
-    },
-    {
-      "epoch": 6.176013527114618,
-      "grad_norm": 0.00517770042642951,
-      "learning_rate": 0.0001999811889292484,
-      "loss": 46.0,
-      "step": 38350
-    },
-    {
-      "epoch": 6.176174564193405,
-      "grad_norm": 0.01145362202078104,
-      "learning_rate": 0.00019998118794798986,
-      "loss": 46.0,
-      "step": 38351
-    },
-    {
-      "epoch": 6.176335601272193,
-      "grad_norm": 0.007486908230930567,
-      "learning_rate": 0.00019998118696670574,
-      "loss": 46.0,
-      "step": 38352
-    },
-    {
-      "epoch": 6.17649663835098,
-      "grad_norm": 0.002026529749855399,
-      "learning_rate": 0.000199981185985396,
-      "loss": 46.0,
-      "step": 38353
-    },
-    {
-      "epoch": 6.176657675429768,
-      "grad_norm": 0.016162317246198654,
-      "learning_rate": 0.00019998118500406072,
-      "loss": 46.0,
-      "step": 38354
-    },
-    {
-      "epoch": 6.176818712508555,
-      "grad_norm": 0.0048370095901191235,
-      "learning_rate": 0.0001999811840226998,
-      "loss": 46.0,
-      "step": 38355
-    },
-    {
-      "epoch": 6.176979749587343,
-      "grad_norm": 0.0014495229115709662,
-      "learning_rate": 0.00019998118304131332,
-      "loss": 46.0,
-      "step": 38356
-    },
-    {
-      "epoch": 6.17714078666613,
-      "grad_norm": 0.020785022526979446,
-      "learning_rate": 0.0001999811820599012,
-      "loss": 46.0,
-      "step": 38357
-    },
-    {
-      "epoch": 6.1773018237449175,
-      "grad_norm": 0.006474633235484362,
-      "learning_rate": 0.00019998118107846357,
-      "loss": 46.0,
-      "step": 38358
-    },
-    {
-      "epoch": 6.177462860823705,
-      "grad_norm": 0.002027590526267886,
-      "learning_rate": 0.0001999811800970003,
-      "loss": 46.0,
-      "step": 38359
-    },
-    {
-      "epoch": 6.1776238979024924,
-      "grad_norm": 0.002121510449796915,
-      "learning_rate": 0.00019998117911551144,
-      "loss": 46.0,
-      "step": 38360
-    },
-    {
-      "epoch": 6.177784934981279,
-      "grad_norm": 0.0012588793179020286,
-      "learning_rate": 0.00019998117813399698,
-      "loss": 46.0,
-      "step": 38361
-    },
-    {
-      "epoch": 6.1779459720600665,
-      "grad_norm": 0.00857162568718195,
-      "learning_rate": 0.00019998117715245696,
-      "loss": 46.0,
-      "step": 38362
-    },
-    {
-      "epoch": 6.178107009138854,
-      "grad_norm": 0.00472683971747756,
-      "learning_rate": 0.00019998117617089133,
-      "loss": 46.0,
-      "step": 38363
-    },
-    {
-      "epoch": 6.178268046217641,
-      "grad_norm": 0.00392599543556571,
-      "learning_rate": 0.00019998117518930014,
-      "loss": 46.0,
-      "step": 38364
-    },
-    {
-      "epoch": 6.178429083296429,
-      "grad_norm": 0.01352886762470007,
-      "learning_rate": 0.0001999811742076833,
-      "loss": 46.0,
-      "step": 38365
-    },
-    {
-      "epoch": 6.178590120375216,
-      "grad_norm": 0.0017806334653869271,
-      "learning_rate": 0.00019998117322604094,
-      "loss": 46.0,
-      "step": 38366
-    },
-    {
-      "epoch": 6.178751157454004,
-      "grad_norm": 0.0024008865002542734,
-      "learning_rate": 0.00019998117224437296,
-      "loss": 46.0,
-      "step": 38367
-    },
-    {
-      "epoch": 6.178912194532791,
-      "grad_norm": 0.002225179923698306,
-      "learning_rate": 0.0001999811712626794,
-      "loss": 46.0,
-      "step": 38368
-    },
-    {
-      "epoch": 6.179073231611579,
-      "grad_norm": 0.003912065178155899,
-      "learning_rate": 0.0001999811702809602,
-      "loss": 46.0,
-      "step": 38369
-    },
-    {
-      "epoch": 6.179234268690366,
-      "grad_norm": 0.01413965318351984,
-      "learning_rate": 0.00019998116929921547,
-      "loss": 46.0,
-      "step": 38370
-    },
-    {
-      "epoch": 6.1793953057691535,
-      "grad_norm": 0.003520033787935972,
-      "learning_rate": 0.00019998116831744511,
-      "loss": 46.0,
-      "step": 38371
-    },
-    {
-      "epoch": 6.179556342847941,
-      "grad_norm": 0.0037289022002369165,
-      "learning_rate": 0.00019998116733564917,
-      "loss": 46.0,
-      "step": 38372
-    },
-    {
-      "epoch": 6.179717379926728,
-      "grad_norm": 0.003721510758623481,
-      "learning_rate": 0.00019998116635382767,
-      "loss": 46.0,
-      "step": 38373
-    },
-    {
-      "epoch": 6.179878417005516,
-      "grad_norm": 0.005485042463988066,
-      "learning_rate": 0.00019998116537198055,
-      "loss": 46.0,
-      "step": 38374
-    },
-    {
-      "epoch": 6.180039454084303,
-      "grad_norm": 0.004296879284083843,
-      "learning_rate": 0.00019998116439010784,
-      "loss": 46.0,
-      "step": 38375
-    },
-    {
-      "epoch": 6.18020049116309,
-      "grad_norm": 0.0034149251878261566,
-      "learning_rate": 0.00019998116340820955,
-      "loss": 46.0,
-      "step": 38376
-    },
-    {
-      "epoch": 6.180361528241877,
-      "grad_norm": 0.018693335354328156,
-      "learning_rate": 0.00019998116242628567,
-      "loss": 46.0,
-      "step": 38377
-    },
-    {
-      "epoch": 6.180522565320665,
-      "grad_norm": 0.004584385547786951,
-      "learning_rate": 0.00019998116144433618,
-      "loss": 46.0,
-      "step": 38378
-    },
-    {
-      "epoch": 6.180683602399452,
-      "grad_norm": 0.011237303726375103,
-      "learning_rate": 0.00019998116046236115,
-      "loss": 46.0,
-      "step": 38379
-    },
-    {
-      "epoch": 6.18084463947824,
-      "grad_norm": 0.006295708008110523,
-      "learning_rate": 0.00019998115948036049,
-      "loss": 46.0,
-      "step": 38380
-    },
-    {
-      "epoch": 6.181005676557027,
-      "grad_norm": 0.005004039034247398,
-      "learning_rate": 0.00019998115849833426,
-      "loss": 46.0,
-      "step": 38381
-    },
-    {
-      "epoch": 6.181166713635815,
-      "grad_norm": 0.002871039556339383,
-      "learning_rate": 0.0001999811575162824,
-      "loss": 46.0,
-      "step": 38382
-    },
-    {
-      "epoch": 6.181327750714602,
-      "grad_norm": 0.00879332609474659,
-      "learning_rate": 0.00019998115653420499,
-      "loss": 46.0,
-      "step": 38383
-    },
-    {
-      "epoch": 6.1814887877933895,
-      "grad_norm": 0.003851083805784583,
-      "learning_rate": 0.00019998115555210197,
-      "loss": 46.0,
-      "step": 38384
-    },
-    {
-      "epoch": 6.181649824872177,
-      "grad_norm": 0.005037369206547737,
-      "learning_rate": 0.00019998115456997337,
-      "loss": 46.0,
-      "step": 38385
-    },
-    {
-      "epoch": 6.181810861950964,
-      "grad_norm": 0.004178883507847786,
-      "learning_rate": 0.00019998115358781917,
-      "loss": 46.0,
-      "step": 38386
-    },
-    {
-      "epoch": 6.181971899029752,
-      "grad_norm": 0.008065703324973583,
-      "learning_rate": 0.0001999811526056394,
-      "loss": 46.0,
-      "step": 38387
-    },
-    {
-      "epoch": 6.182132936108539,
-      "grad_norm": 0.0018702278612181544,
-      "learning_rate": 0.000199981151623434,
-      "loss": 46.0,
-      "step": 38388
-    },
-    {
-      "epoch": 6.182293973187327,
-      "grad_norm": 0.0022260923869907856,
-      "learning_rate": 0.00019998115064120305,
-      "loss": 46.0,
-      "step": 38389
-    },
-    {
-      "epoch": 6.182455010266114,
-      "grad_norm": 0.008190924301743507,
-      "learning_rate": 0.00019998114965894649,
-      "loss": 46.0,
-      "step": 38390
-    },
-    {
-      "epoch": 6.182616047344901,
-      "grad_norm": 0.015463664196431637,
-      "learning_rate": 0.00019998114867666436,
-      "loss": 46.0,
-      "step": 38391
-    },
-    {
-      "epoch": 6.182777084423688,
-      "grad_norm": 0.0030463470611721277,
-      "learning_rate": 0.00019998114769435662,
-      "loss": 46.0,
-      "step": 38392
-    },
-    {
-      "epoch": 6.182938121502476,
-      "grad_norm": 0.004487467464059591,
-      "learning_rate": 0.00019998114671202332,
-      "loss": 46.0,
-      "step": 38393
-    },
-    {
-      "epoch": 6.183099158581263,
-      "grad_norm": 0.004162515047937632,
-      "learning_rate": 0.00019998114572966437,
-      "loss": 46.0,
-      "step": 38394
-    },
-    {
-      "epoch": 6.183260195660051,
-      "grad_norm": 0.009179658256471157,
-      "learning_rate": 0.00019998114474727987,
-      "loss": 46.0,
-      "step": 38395
-    },
-    {
-      "epoch": 6.183421232738838,
-      "grad_norm": 0.008634859696030617,
-      "learning_rate": 0.00019998114376486978,
-      "loss": 46.0,
-      "step": 38396
-    },
-    {
-      "epoch": 6.1835822698176255,
-      "grad_norm": 0.0035388132091611624,
-      "learning_rate": 0.0001999811427824341,
-      "loss": 46.0,
-      "step": 38397
-    },
-    {
-      "epoch": 6.183743306896413,
-      "grad_norm": 0.0043466719798743725,
-      "learning_rate": 0.00019998114179997281,
-      "loss": 46.0,
-      "step": 38398
-    },
-    {
-      "epoch": 6.1839043439752,
-      "grad_norm": 0.0103925671428442,
-      "learning_rate": 0.00019998114081748596,
-      "loss": 46.0,
-      "step": 38399
-    },
-    {
-      "epoch": 6.184065381053988,
-      "grad_norm": 0.007809177972376347,
-      "learning_rate": 0.0001999811398349735,
-      "loss": 46.0,
-      "step": 38400
-    },
-    {
-      "epoch": 6.184226418132775,
-      "grad_norm": 0.009312433190643787,
-      "learning_rate": 0.00019998113885243544,
-      "loss": 46.0,
-      "step": 38401
-    },
-    {
-      "epoch": 6.184387455211563,
-      "grad_norm": 0.005295824259519577,
-      "learning_rate": 0.00019998113786987183,
-      "loss": 46.0,
-      "step": 38402
-    },
-    {
-      "epoch": 6.18454849229035,
-      "grad_norm": 0.018275827169418335,
-      "learning_rate": 0.0001999811368872826,
-      "loss": 46.0,
-      "step": 38403
-    },
-    {
-      "epoch": 6.184709529369138,
-      "grad_norm": 0.013024595566093922,
-      "learning_rate": 0.0001999811359046678,
-      "loss": 46.0,
-      "step": 38404
-    },
-    {
-      "epoch": 6.184870566447924,
-      "grad_norm": 0.005584705621004105,
-      "learning_rate": 0.0001999811349220274,
-      "loss": 46.0,
-      "step": 38405
-    },
-    {
-      "epoch": 6.185031603526712,
-      "grad_norm": 0.006218624766916037,
-      "learning_rate": 0.0001999811339393614,
-      "loss": 46.0,
-      "step": 38406
-    },
-    {
-      "epoch": 6.185192640605499,
-      "grad_norm": 0.007102572359144688,
-      "learning_rate": 0.0001999811329566698,
-      "loss": 46.0,
-      "step": 38407
-    },
-    {
-      "epoch": 6.185353677684287,
-      "grad_norm": 0.0029884851537644863,
-      "learning_rate": 0.00019998113197395264,
-      "loss": 46.0,
-      "step": 38408
-    },
-    {
-      "epoch": 6.185514714763074,
-      "grad_norm": 0.004121674224734306,
-      "learning_rate": 0.00019998113099120986,
-      "loss": 46.0,
-      "step": 38409
-    },
-    {
-      "epoch": 6.1856757518418615,
-      "grad_norm": 0.0016099418280646205,
-      "learning_rate": 0.0001999811300084415,
-      "loss": 46.0,
-      "step": 38410
-    },
-    {
-      "epoch": 6.185836788920649,
-      "grad_norm": 0.005504402797669172,
-      "learning_rate": 0.00019998112902564757,
-      "loss": 46.0,
-      "step": 38411
-    },
-    {
-      "epoch": 6.185997825999436,
-      "grad_norm": 0.009299728088080883,
-      "learning_rate": 0.00019998112804282803,
-      "loss": 46.0,
-      "step": 38412
-    },
-    {
-      "epoch": 6.186158863078224,
-      "grad_norm": 0.00526196975260973,
-      "learning_rate": 0.0001999811270599829,
-      "loss": 46.0,
-      "step": 38413
-    },
-    {
-      "epoch": 6.186319900157011,
-      "grad_norm": 0.004188033752143383,
-      "learning_rate": 0.0001999811260771122,
-      "loss": 46.0,
-      "step": 38414
-    },
-    {
-      "epoch": 6.186480937235799,
-      "grad_norm": 0.002466693054884672,
-      "learning_rate": 0.0001999811250942159,
-      "loss": 46.0,
-      "step": 38415
-    },
-    {
-      "epoch": 6.186641974314586,
-      "grad_norm": 0.0014106532325968146,
-      "learning_rate": 0.00019998112411129398,
-      "loss": 46.0,
-      "step": 38416
-    },
-    {
-      "epoch": 6.186803011393374,
-      "grad_norm": 0.009123711846768856,
-      "learning_rate": 0.0001999811231283465,
-      "loss": 46.0,
-      "step": 38417
-    },
-    {
-      "epoch": 6.186964048472161,
-      "grad_norm": 0.0063522448763251305,
-      "learning_rate": 0.00019998112214537344,
-      "loss": 46.0,
-      "step": 38418
-    },
-    {
-      "epoch": 6.1871250855509485,
-      "grad_norm": 0.0025015079881995916,
-      "learning_rate": 0.00019998112116237477,
-      "loss": 46.0,
-      "step": 38419
-    },
-    {
-      "epoch": 6.187286122629735,
-      "grad_norm": 0.011544137261807919,
-      "learning_rate": 0.0001999811201793505,
-      "loss": 46.0,
-      "step": 38420
-    },
-    {
-      "epoch": 6.1874471597085225,
-      "grad_norm": 0.004128925036638975,
-      "learning_rate": 0.00019998111919630068,
-      "loss": 46.0,
-      "step": 38421
-    },
-    {
-      "epoch": 6.18760819678731,
-      "grad_norm": 0.009844529442489147,
-      "learning_rate": 0.00019998111821322521,
-      "loss": 46.0,
-      "step": 38422
-    },
-    {
-      "epoch": 6.1877692338660975,
-      "grad_norm": 0.004908531438559294,
-      "learning_rate": 0.00019998111723012422,
-      "loss": 46.0,
-      "step": 38423
-    },
-    {
-      "epoch": 6.187930270944885,
-      "grad_norm": 0.0027955572586506605,
-      "learning_rate": 0.0001999811162469976,
-      "loss": 46.0,
-      "step": 38424
-    },
-    {
-      "epoch": 6.188091308023672,
-      "grad_norm": 0.011891791597008705,
-      "learning_rate": 0.00019998111526384538,
-      "loss": 46.0,
-      "step": 38425
-    },
-    {
-      "epoch": 6.18825234510246,
-      "grad_norm": 0.00636469479650259,
-      "learning_rate": 0.0001999811142806676,
-      "loss": 46.0,
-      "step": 38426
-    },
-    {
-      "epoch": 6.188413382181247,
-      "grad_norm": 0.003043847158551216,
-      "learning_rate": 0.00019998111329746422,
-      "loss": 46.0,
-      "step": 38427
-    },
-    {
-      "epoch": 6.188574419260035,
-      "grad_norm": 0.0017767632380127907,
-      "learning_rate": 0.00019998111231423523,
-      "loss": 46.0,
-      "step": 38428
-    },
-    {
-      "epoch": 6.188735456338822,
-      "grad_norm": 0.012975356541574001,
-      "learning_rate": 0.00019998111133098068,
-      "loss": 46.0,
-      "step": 38429
-    },
-    {
-      "epoch": 6.18889649341761,
-      "grad_norm": 0.002072844421491027,
-      "learning_rate": 0.0001999811103477005,
-      "loss": 46.0,
-      "step": 38430
-    },
-    {
-      "epoch": 6.189057530496397,
-      "grad_norm": 0.003157053841277957,
-      "learning_rate": 0.00019998110936439475,
-      "loss": 46.0,
-      "step": 38431
-    },
-    {
-      "epoch": 6.1892185675751845,
-      "grad_norm": 0.0028113776352256536,
-      "learning_rate": 0.00019998110838106344,
-      "loss": 46.0,
-      "step": 38432
-    },
-    {
-      "epoch": 6.189379604653972,
-      "grad_norm": 0.0035995894577354193,
-      "learning_rate": 0.0001999811073977065,
-      "loss": 46.0,
-      "step": 38433
-    },
-    {
-      "epoch": 6.1895406417327585,
-      "grad_norm": 0.0022093728184700012,
-      "learning_rate": 0.000199981106414324,
-      "loss": 46.0,
-      "step": 38434
-    },
-    {
-      "epoch": 6.189701678811546,
-      "grad_norm": 0.0016370511148124933,
-      "learning_rate": 0.00019998110543091588,
-      "loss": 46.0,
-      "step": 38435
-    },
-    {
-      "epoch": 6.189862715890333,
-      "grad_norm": 0.0027529362123459578,
-      "learning_rate": 0.0001999811044474822,
-      "loss": 46.0,
-      "step": 38436
-    },
-    {
-      "epoch": 6.190023752969121,
-      "grad_norm": 0.00694660097360611,
-      "learning_rate": 0.0001999811034640229,
-      "loss": 46.0,
-      "step": 38437
-    },
-    {
-      "epoch": 6.190184790047908,
-      "grad_norm": 0.013354429043829441,
-      "learning_rate": 0.00019998110248053803,
-      "loss": 46.0,
-      "step": 38438
-    },
-    {
-      "epoch": 6.190345827126696,
-      "grad_norm": 0.012662060558795929,
-      "learning_rate": 0.00019998110149702756,
-      "loss": 46.0,
-      "step": 38439
-    },
-    {
-      "epoch": 6.190506864205483,
-      "grad_norm": 0.020456044003367424,
-      "learning_rate": 0.0001999811005134915,
-      "loss": 46.0,
-      "step": 38440
-    },
-    {
-      "epoch": 6.190667901284271,
-      "grad_norm": 0.002556009916588664,
-      "learning_rate": 0.00019998109952992985,
-      "loss": 46.0,
-      "step": 38441
-    },
-    {
-      "epoch": 6.190828938363058,
-      "grad_norm": 0.0012579491594806314,
-      "learning_rate": 0.00019998109854634262,
-      "loss": 46.0,
-      "step": 38442
-    },
-    {
-      "epoch": 6.190989975441846,
-      "grad_norm": 0.0013837008737027645,
-      "learning_rate": 0.00019998109756272977,
-      "loss": 46.0,
-      "step": 38443
-    },
-    {
-      "epoch": 6.191151012520633,
-      "grad_norm": 0.004684407729655504,
-      "learning_rate": 0.00019998109657909136,
-      "loss": 46.0,
-      "step": 38444
-    },
-    {
-      "epoch": 6.1913120495994205,
-      "grad_norm": 0.003011419204995036,
-      "learning_rate": 0.00019998109559542736,
-      "loss": 46.0,
-      "step": 38445
-    },
-    {
-      "epoch": 6.191473086678208,
-      "grad_norm": 0.0025610574521124363,
-      "learning_rate": 0.00019998109461173778,
-      "loss": 46.0,
-      "step": 38446
-    },
-    {
-      "epoch": 6.191634123756995,
-      "grad_norm": 0.005681574810296297,
-      "learning_rate": 0.0001999810936280226,
-      "loss": 46.0,
-      "step": 38447
-    },
-    {
-      "epoch": 6.191795160835783,
-      "grad_norm": 0.0061127422377467155,
-      "learning_rate": 0.00019998109264428182,
-      "loss": 46.0,
-      "step": 38448
-    },
-    {
-      "epoch": 6.191956197914569,
-      "grad_norm": 0.006921328604221344,
-      "learning_rate": 0.00019998109166051545,
-      "loss": 46.0,
-      "step": 38449
-    },
-    {
-      "epoch": 6.192117234993357,
-      "grad_norm": 0.0032270397059619427,
-      "learning_rate": 0.0001999810906767235,
-      "loss": 46.0,
-      "step": 38450
-    },
-    {
-      "epoch": 6.192278272072144,
-      "grad_norm": 0.0013061941135674715,
-      "learning_rate": 0.00019998108969290594,
-      "loss": 46.0,
-      "step": 38451
-    },
-    {
-      "epoch": 6.192439309150932,
-      "grad_norm": 0.011984949931502342,
-      "learning_rate": 0.0001999810887090628,
-      "loss": 46.0,
-      "step": 38452
-    },
-    {
-      "epoch": 6.192600346229719,
-      "grad_norm": 0.010736663825809956,
-      "learning_rate": 0.0001999810877251941,
-      "loss": 46.0,
-      "step": 38453
-    },
-    {
-      "epoch": 6.192761383308507,
-      "grad_norm": 0.0023626014590263367,
-      "learning_rate": 0.00019998108674129975,
-      "loss": 46.0,
-      "step": 38454
-    },
-    {
-      "epoch": 6.192922420387294,
-      "grad_norm": 0.0017383425729349256,
-      "learning_rate": 0.00019998108575737986,
-      "loss": 46.0,
-      "step": 38455
-    },
-    {
-      "epoch": 6.193083457466082,
-      "grad_norm": 0.008774983696639538,
-      "learning_rate": 0.00019998108477343435,
-      "loss": 46.0,
-      "step": 38456
-    },
-    {
-      "epoch": 6.193244494544869,
-      "grad_norm": 0.004909793380647898,
-      "learning_rate": 0.00019998108378946328,
-      "loss": 46.0,
-      "step": 38457
-    },
-    {
-      "epoch": 6.1934055316236565,
-      "grad_norm": 0.0026797510217875242,
-      "learning_rate": 0.0001999810828054666,
-      "loss": 46.0,
-      "step": 38458
-    },
-    {
-      "epoch": 6.193566568702444,
-      "grad_norm": 0.0020990953780710697,
-      "learning_rate": 0.00019998108182144432,
-      "loss": 46.0,
-      "step": 38459
-    },
-    {
-      "epoch": 6.193727605781231,
-      "grad_norm": 0.00247512711212039,
-      "learning_rate": 0.0001999810808373965,
-      "loss": 46.0,
-      "step": 38460
-    },
-    {
-      "epoch": 6.193888642860019,
-      "grad_norm": 0.006691396702080965,
-      "learning_rate": 0.00019998107985332305,
-      "loss": 46.0,
-      "step": 38461
-    },
-    {
-      "epoch": 6.194049679938806,
-      "grad_norm": 0.0046486384235322475,
-      "learning_rate": 0.00019998107886922399,
-      "loss": 46.0,
-      "step": 38462
-    },
-    {
-      "epoch": 6.194210717017594,
-      "grad_norm": 0.004219725262373686,
-      "learning_rate": 0.00019998107788509937,
-      "loss": 46.0,
-      "step": 38463
-    },
-    {
-      "epoch": 6.19437175409638,
-      "grad_norm": 0.0061407554894685745,
-      "learning_rate": 0.00019998107690094913,
-      "loss": 46.0,
-      "step": 38464
-    },
-    {
-      "epoch": 6.194532791175168,
-      "grad_norm": 0.004228383302688599,
-      "learning_rate": 0.00019998107591677334,
-      "loss": 46.0,
-      "step": 38465
-    },
-    {
-      "epoch": 6.194693828253955,
-      "grad_norm": 0.0012823771685361862,
-      "learning_rate": 0.00019998107493257196,
-      "loss": 46.0,
-      "step": 38466
-    },
-    {
-      "epoch": 6.194854865332743,
-      "grad_norm": 0.013044308871030807,
-      "learning_rate": 0.000199981073948345,
-      "loss": 46.0,
-      "step": 38467
-    },
-    {
-      "epoch": 6.19501590241153,
-      "grad_norm": 0.005286875646561384,
-      "learning_rate": 0.00019998107296409238,
-      "loss": 46.0,
-      "step": 38468
-    },
-    {
-      "epoch": 6.195176939490318,
-      "grad_norm": 0.0065002660267055035,
-      "learning_rate": 0.0001999810719798142,
-      "loss": 46.0,
-      "step": 38469
-    },
-    {
-      "epoch": 6.195337976569105,
-      "grad_norm": 0.007993276230990887,
-      "learning_rate": 0.00019998107099551048,
-      "loss": 46.0,
-      "step": 38470
-    },
-    {
-      "epoch": 6.1954990136478925,
-      "grad_norm": 0.002286696108058095,
-      "learning_rate": 0.0001999810700111811,
-      "loss": 46.0,
-      "step": 38471
-    },
-    {
-      "epoch": 6.19566005072668,
-      "grad_norm": 0.0018333852058276534,
-      "learning_rate": 0.00019998106902682617,
-      "loss": 46.0,
-      "step": 38472
-    },
-    {
-      "epoch": 6.195821087805467,
-      "grad_norm": 0.0015674241585657,
-      "learning_rate": 0.00019998106804244566,
-      "loss": 46.0,
-      "step": 38473
-    },
-    {
-      "epoch": 6.195982124884255,
-      "grad_norm": 0.0017585881287232041,
-      "learning_rate": 0.00019998106705803955,
-      "loss": 46.0,
-      "step": 38474
-    },
-    {
-      "epoch": 6.196143161963042,
-      "grad_norm": 0.003924446180462837,
-      "learning_rate": 0.00019998106607360783,
-      "loss": 46.0,
-      "step": 38475
-    },
-    {
-      "epoch": 6.19630419904183,
-      "grad_norm": 0.008062554523348808,
-      "learning_rate": 0.00019998106508915052,
-      "loss": 46.0,
-      "step": 38476
-    },
-    {
-      "epoch": 6.196465236120617,
-      "grad_norm": 0.0027739075012505054,
-      "learning_rate": 0.00019998106410466763,
-      "loss": 46.0,
-      "step": 38477
-    },
-    {
-      "epoch": 6.196626273199404,
-      "grad_norm": 0.0020640636794269085,
-      "learning_rate": 0.00019998106312015917,
-      "loss": 46.0,
-      "step": 38478
-    },
-    {
-      "epoch": 6.196787310278191,
-      "grad_norm": 0.004370938055217266,
-      "learning_rate": 0.0001999810621356251,
-      "loss": 46.0,
-      "step": 38479
-    },
-    {
-      "epoch": 6.196948347356979,
-      "grad_norm": 0.002564073074609041,
-      "learning_rate": 0.00019998106115106542,
-      "loss": 46.0,
-      "step": 38480
-    },
-    {
-      "epoch": 6.197109384435766,
-      "grad_norm": 0.002289981348440051,
-      "learning_rate": 0.0001999810601664802,
-      "loss": 46.0,
-      "step": 38481
-    },
-    {
-      "epoch": 6.1972704215145535,
-      "grad_norm": 0.008379199542105198,
-      "learning_rate": 0.00019998105918186935,
-      "loss": 46.0,
-      "step": 38482
-    },
-    {
-      "epoch": 6.197431458593341,
-      "grad_norm": 0.007253784220665693,
-      "learning_rate": 0.00019998105819723293,
-      "loss": 46.0,
-      "step": 38483
-    },
-    {
-      "epoch": 6.197592495672128,
-      "grad_norm": 0.002788061508908868,
-      "learning_rate": 0.00019998105721257093,
-      "loss": 46.0,
-      "step": 38484
-    },
-    {
-      "epoch": 6.197753532750916,
-      "grad_norm": 0.0016887930687516928,
-      "learning_rate": 0.0001999810562278833,
-      "loss": 46.0,
-      "step": 38485
-    },
-    {
-      "epoch": 6.197914569829703,
-      "grad_norm": 0.007803408429026604,
-      "learning_rate": 0.0001999810552431701,
-      "loss": 46.0,
-      "step": 38486
-    },
-    {
-      "epoch": 6.198075606908491,
-      "grad_norm": 0.0014974446967244148,
-      "learning_rate": 0.00019998105425843133,
-      "loss": 46.0,
-      "step": 38487
-    },
-    {
-      "epoch": 6.198236643987278,
-      "grad_norm": 0.003115519415587187,
-      "learning_rate": 0.00019998105327366695,
-      "loss": 46.0,
-      "step": 38488
-    },
-    {
-      "epoch": 6.198397681066066,
-      "grad_norm": 0.002230840502306819,
-      "learning_rate": 0.00019998105228887699,
-      "loss": 46.0,
-      "step": 38489
-    },
-    {
-      "epoch": 6.198558718144853,
-      "grad_norm": 0.002150146523490548,
-      "learning_rate": 0.00019998105130406143,
-      "loss": 46.0,
-      "step": 38490
-    },
-    {
-      "epoch": 6.198719755223641,
-      "grad_norm": 0.024488558992743492,
-      "learning_rate": 0.0001999810503192203,
-      "loss": 46.0,
-      "step": 38491
-    },
-    {
-      "epoch": 6.198880792302428,
-      "grad_norm": 0.006840047426521778,
-      "learning_rate": 0.00019998104933435356,
-      "loss": 46.0,
-      "step": 38492
-    },
-    {
-      "epoch": 6.199041829381215,
-      "grad_norm": 0.00801506545394659,
-      "learning_rate": 0.00019998104834946122,
-      "loss": 46.0,
-      "step": 38493
-    },
-    {
-      "epoch": 6.199202866460002,
-      "grad_norm": 0.005811032839119434,
-      "learning_rate": 0.00019998104736454329,
-      "loss": 46.0,
-      "step": 38494
-    },
-    {
-      "epoch": 6.1993639035387895,
-      "grad_norm": 0.003501306753605604,
-      "learning_rate": 0.00019998104637959977,
-      "loss": 46.0,
-      "step": 38495
-    },
-    {
-      "epoch": 6.199524940617577,
-      "grad_norm": 0.00610383041203022,
-      "learning_rate": 0.0001999810453946307,
-      "loss": 46.0,
-      "step": 38496
-    },
-    {
-      "epoch": 6.199685977696364,
-      "grad_norm": 0.0011243073968216777,
-      "learning_rate": 0.000199981044409636,
-      "loss": 46.0,
-      "step": 38497
-    },
-    {
-      "epoch": 6.199847014775152,
-      "grad_norm": 0.0010643979767337441,
-      "learning_rate": 0.00019998104342461572,
-      "loss": 46.0,
-      "step": 38498
-    },
-    {
-      "epoch": 6.200008051853939,
-      "grad_norm": 0.0012480011209845543,
-      "learning_rate": 0.00019998104243956985,
-      "loss": 46.0,
-      "step": 38499
-    },
-    {
-      "epoch": 6.200169088932727,
-      "grad_norm": 0.007588703650981188,
-      "learning_rate": 0.0001999810414544984,
-      "loss": 46.0,
-      "step": 38500
-    },
-    {
-      "epoch": 6.200330126011514,
-      "grad_norm": 0.021159600466489792,
-      "learning_rate": 0.00019998104046940136,
-      "loss": 46.0,
-      "step": 38501
-    },
-    {
-      "epoch": 6.200491163090302,
-      "grad_norm": 0.0031840221490710974,
-      "learning_rate": 0.0001999810394842787,
-      "loss": 46.0,
-      "step": 38502
-    },
-    {
-      "epoch": 6.200652200169089,
-      "grad_norm": 0.006811012048274279,
-      "learning_rate": 0.0001999810384991305,
-      "loss": 46.0,
-      "step": 38503
-    },
-    {
-      "epoch": 6.200813237247877,
-      "grad_norm": 0.0030008992180228233,
-      "learning_rate": 0.00019998103751395666,
-      "loss": 46.0,
-      "step": 38504
-    },
-    {
-      "epoch": 6.200974274326664,
-      "grad_norm": 0.0083366883918643,
-      "learning_rate": 0.00019998103652875727,
-      "loss": 46.0,
-      "step": 38505
-    },
-    {
-      "epoch": 6.2011353114054515,
-      "grad_norm": 0.011031761765480042,
-      "learning_rate": 0.00019998103554353226,
-      "loss": 46.0,
-      "step": 38506
-    },
-    {
-      "epoch": 6.201296348484238,
-      "grad_norm": 0.008430404588580132,
-      "learning_rate": 0.00019998103455828167,
-      "loss": 46.0,
-      "step": 38507
-    },
-    {
-      "epoch": 6.2014573855630255,
-      "grad_norm": 0.004892644938081503,
-      "learning_rate": 0.0001999810335730055,
-      "loss": 46.0,
-      "step": 38508
-    },
-    {
-      "epoch": 6.201618422641813,
-      "grad_norm": 0.00163461163174361,
-      "learning_rate": 0.00019998103258770373,
-      "loss": 46.0,
-      "step": 38509
-    },
-    {
-      "epoch": 6.2017794597206,
-      "grad_norm": 0.0021717629861086607,
-      "learning_rate": 0.00019998103160237638,
-      "loss": 46.0,
-      "step": 38510
-    },
-    {
-      "epoch": 6.201940496799388,
-      "grad_norm": 0.0017033193726092577,
-      "learning_rate": 0.0001999810306170234,
-      "loss": 46.0,
-      "step": 38511
-    },
-    {
-      "epoch": 6.202101533878175,
-      "grad_norm": 0.0027894889935851097,
-      "learning_rate": 0.00019998102963164488,
-      "loss": 46.0,
-      "step": 38512
-    },
-    {
-      "epoch": 6.202262570956963,
-      "grad_norm": 0.0014997628750279546,
-      "learning_rate": 0.00019998102864624077,
-      "loss": 46.0,
-      "step": 38513
-    },
-    {
-      "epoch": 6.20242360803575,
-      "grad_norm": 0.004330724012106657,
-      "learning_rate": 0.00019998102766081104,
-      "loss": 46.0,
-      "step": 38514
-    },
-    {
-      "epoch": 6.202584645114538,
-      "grad_norm": 0.008996432647109032,
-      "learning_rate": 0.00019998102667535575,
-      "loss": 46.0,
-      "step": 38515
-    },
-    {
-      "epoch": 6.202745682193325,
-      "grad_norm": 0.0032568250317126513,
-      "learning_rate": 0.00019998102568987485,
-      "loss": 46.0,
-      "step": 38516
-    },
-    {
-      "epoch": 6.202906719272113,
-      "grad_norm": 0.008566852658987045,
-      "learning_rate": 0.00019998102470436833,
-      "loss": 46.0,
-      "step": 38517
-    },
-    {
-      "epoch": 6.2030677563509,
-      "grad_norm": 0.01192668080329895,
-      "learning_rate": 0.00019998102371883628,
-      "loss": 46.0,
-      "step": 38518
-    },
-    {
-      "epoch": 6.2032287934296875,
-      "grad_norm": 0.003461219370365143,
-      "learning_rate": 0.00019998102273327862,
-      "loss": 46.0,
-      "step": 38519
-    },
-    {
-      "epoch": 6.203389830508475,
-      "grad_norm": 0.012207056395709515,
-      "learning_rate": 0.00019998102174769534,
-      "loss": 46.0,
-      "step": 38520
-    },
-    {
-      "epoch": 6.203550867587262,
-      "grad_norm": 0.002491791034117341,
-      "learning_rate": 0.0001999810207620865,
-      "loss": 46.0,
-      "step": 38521
-    },
-    {
-      "epoch": 6.203711904666049,
-      "grad_norm": 0.006606719456613064,
-      "learning_rate": 0.00019998101977645208,
-      "loss": 46.0,
-      "step": 38522
-    },
-    {
-      "epoch": 6.203872941744836,
-      "grad_norm": 0.006626618094742298,
-      "learning_rate": 0.00019998101879079206,
-      "loss": 46.0,
-      "step": 38523
-    },
-    {
-      "epoch": 6.204033978823624,
-      "grad_norm": 0.009684987366199493,
-      "learning_rate": 0.00019998101780510644,
-      "loss": 46.0,
-      "step": 38524
-    },
-    {
-      "epoch": 6.204195015902411,
-      "grad_norm": 0.017381057143211365,
-      "learning_rate": 0.00019998101681939522,
-      "loss": 46.0,
-      "step": 38525
-    },
-    {
-      "epoch": 6.204356052981199,
-      "grad_norm": 0.01152043417096138,
-      "learning_rate": 0.0001999810158336584,
-      "loss": 46.0,
-      "step": 38526
-    },
-    {
-      "epoch": 6.204517090059986,
-      "grad_norm": 0.0025890732649713755,
-      "learning_rate": 0.00019998101484789603,
-      "loss": 46.0,
-      "step": 38527
-    },
-    {
-      "epoch": 6.204678127138774,
-      "grad_norm": 0.0017062181141227484,
-      "learning_rate": 0.00019998101386210806,
-      "loss": 46.0,
-      "step": 38528
-    },
-    {
-      "epoch": 6.204839164217561,
-      "grad_norm": 0.0017474015476182103,
-      "learning_rate": 0.0001999810128762945,
-      "loss": 46.0,
-      "step": 38529
-    },
-    {
-      "epoch": 6.2050002012963485,
-      "grad_norm": 0.0033452387433499098,
-      "learning_rate": 0.00019998101189045532,
-      "loss": 46.0,
-      "step": 38530
-    },
-    {
-      "epoch": 6.205161238375136,
-      "grad_norm": 0.0010951875010505319,
-      "learning_rate": 0.00019998101090459058,
-      "loss": 46.0,
-      "step": 38531
-    },
-    {
-      "epoch": 6.2053222754539235,
-      "grad_norm": 0.010053624399006367,
-      "learning_rate": 0.00019998100991870023,
-      "loss": 46.0,
-      "step": 38532
-    },
-    {
-      "epoch": 6.205483312532711,
-      "grad_norm": 0.005480147432535887,
-      "learning_rate": 0.00019998100893278432,
-      "loss": 46.0,
-      "step": 38533
-    },
-    {
-      "epoch": 6.205644349611498,
-      "grad_norm": 0.003845823463052511,
-      "learning_rate": 0.0001999810079468428,
-      "loss": 46.0,
-      "step": 38534
-    },
-    {
-      "epoch": 6.205805386690286,
-      "grad_norm": 0.002510937163606286,
-      "learning_rate": 0.0001999810069608757,
-      "loss": 46.0,
-      "step": 38535
-    },
-    {
-      "epoch": 6.205966423769073,
-      "grad_norm": 0.005096937529742718,
-      "learning_rate": 0.000199981005974883,
-      "loss": 46.0,
-      "step": 38536
-    },
-    {
-      "epoch": 6.20612746084786,
-      "grad_norm": 0.004791811108589172,
-      "learning_rate": 0.00019998100498886472,
-      "loss": 46.0,
-      "step": 38537
-    },
-    {
-      "epoch": 6.206288497926647,
-      "grad_norm": 0.0035096872597932816,
-      "learning_rate": 0.00019998100400282082,
-      "loss": 46.0,
-      "step": 38538
-    },
-    {
-      "epoch": 6.206449535005435,
-      "grad_norm": 0.009524703957140446,
-      "learning_rate": 0.00019998100301675135,
-      "loss": 46.0,
-      "step": 38539
-    },
-    {
-      "epoch": 6.206610572084222,
-      "grad_norm": 0.010632744058966637,
-      "learning_rate": 0.00019998100203065628,
-      "loss": 46.0,
-      "step": 38540
-    },
-    {
-      "epoch": 6.20677160916301,
-      "grad_norm": 0.003830992616713047,
-      "learning_rate": 0.00019998100104453564,
-      "loss": 46.0,
-      "step": 38541
-    },
-    {
-      "epoch": 6.206932646241797,
-      "grad_norm": 0.0018759763333946466,
-      "learning_rate": 0.0001999810000583894,
-      "loss": 46.0,
-      "step": 38542
-    },
-    {
-      "epoch": 6.2070936833205845,
-      "grad_norm": 0.02378779463469982,
-      "learning_rate": 0.00019998099907221758,
-      "loss": 46.0,
-      "step": 38543
-    },
-    {
-      "epoch": 6.207254720399372,
-      "grad_norm": 0.011196010746061802,
-      "learning_rate": 0.00019998099808602016,
-      "loss": 46.0,
-      "step": 38544
-    },
-    {
-      "epoch": 6.207415757478159,
-      "grad_norm": 0.006778963375836611,
-      "learning_rate": 0.00019998099709979717,
-      "loss": 46.0,
-      "step": 38545
-    },
-    {
-      "epoch": 6.207576794556947,
-      "grad_norm": 0.0018767776200547814,
-      "learning_rate": 0.00019998099611354857,
-      "loss": 46.0,
-      "step": 38546
-    },
-    {
-      "epoch": 6.207737831635734,
-      "grad_norm": 0.006590351928025484,
-      "learning_rate": 0.00019998099512727439,
-      "loss": 46.0,
-      "step": 38547
-    },
-    {
-      "epoch": 6.207898868714522,
-      "grad_norm": 0.0036499181296676397,
-      "learning_rate": 0.00019998099414097459,
-      "loss": 46.0,
-      "step": 38548
-    },
-    {
-      "epoch": 6.208059905793309,
-      "grad_norm": 0.013853631913661957,
-      "learning_rate": 0.00019998099315464925,
-      "loss": 46.0,
-      "step": 38549
-    },
-    {
-      "epoch": 6.208220942872097,
-      "grad_norm": 0.003738798899576068,
-      "learning_rate": 0.00019998099216829828,
-      "loss": 46.0,
-      "step": 38550
-    },
-    {
-      "epoch": 6.208381979950884,
-      "grad_norm": 0.005554336588829756,
-      "learning_rate": 0.00019998099118192172,
-      "loss": 46.0,
-      "step": 38551
-    },
-    {
-      "epoch": 6.208543017029671,
-      "grad_norm": 0.002461962401866913,
-      "learning_rate": 0.0001999809901955196,
-      "loss": 46.0,
-      "step": 38552
-    },
-    {
-      "epoch": 6.208704054108458,
-      "grad_norm": 0.002006252994760871,
-      "learning_rate": 0.00019998098920909186,
-      "loss": 46.0,
-      "step": 38553
-    },
-    {
-      "epoch": 6.208865091187246,
-      "grad_norm": 0.0008995594689622521,
-      "learning_rate": 0.00019998098822263853,
-      "loss": 46.0,
-      "step": 38554
-    },
-    {
-      "epoch": 6.209026128266033,
-      "grad_norm": 0.0034110061824321747,
-      "learning_rate": 0.00019998098723615962,
-      "loss": 46.0,
-      "step": 38555
-    },
-    {
-      "epoch": 6.2091871653448205,
-      "grad_norm": 0.0011905698338523507,
-      "learning_rate": 0.00019998098624965512,
-      "loss": 46.0,
-      "step": 38556
-    },
-    {
-      "epoch": 6.209348202423608,
-      "grad_norm": 0.002167128724977374,
-      "learning_rate": 0.00019998098526312507,
-      "loss": 46.0,
-      "step": 38557
-    },
-    {
-      "epoch": 6.209509239502395,
-      "grad_norm": 0.00407895864918828,
-      "learning_rate": 0.00019998098427656937,
-      "loss": 46.0,
-      "step": 38558
-    },
-    {
-      "epoch": 6.209670276581183,
-      "grad_norm": 0.005671153776347637,
-      "learning_rate": 0.0001999809832899881,
-      "loss": 46.0,
-      "step": 38559
-    },
-    {
-      "epoch": 6.20983131365997,
-      "grad_norm": 0.0096208481118083,
-      "learning_rate": 0.00019998098230338126,
-      "loss": 46.0,
-      "step": 38560
-    },
-    {
-      "epoch": 6.209992350738758,
-      "grad_norm": 0.008138995617628098,
-      "learning_rate": 0.0001999809813167488,
-      "loss": 46.0,
-      "step": 38561
-    },
-    {
-      "epoch": 6.210153387817545,
-      "grad_norm": 0.006476081442087889,
-      "learning_rate": 0.00019998098033009075,
-      "loss": 46.0,
-      "step": 38562
-    },
-    {
-      "epoch": 6.210314424896333,
-      "grad_norm": 0.005781737621873617,
-      "learning_rate": 0.00019998097934340714,
-      "loss": 46.0,
-      "step": 38563
-    },
-    {
-      "epoch": 6.21047546197512,
-      "grad_norm": 0.0018086661584675312,
-      "learning_rate": 0.00019998097835669792,
-      "loss": 46.0,
-      "step": 38564
-    },
-    {
-      "epoch": 6.210636499053908,
-      "grad_norm": 0.0038158376701176167,
-      "learning_rate": 0.0001999809773699631,
-      "loss": 46.0,
-      "step": 38565
-    },
-    {
-      "epoch": 6.210797536132694,
-      "grad_norm": 0.0017599390121176839,
-      "learning_rate": 0.0001999809763832027,
-      "loss": 46.0,
-      "step": 38566
-    },
-    {
-      "epoch": 6.210958573211482,
-      "grad_norm": 0.005088081583380699,
-      "learning_rate": 0.0001999809753964167,
-      "loss": 46.0,
-      "step": 38567
-    },
-    {
-      "epoch": 6.211119610290269,
-      "grad_norm": 0.00866482499986887,
-      "learning_rate": 0.00019998097440960516,
-      "loss": 46.0,
-      "step": 38568
-    },
-    {
-      "epoch": 6.2112806473690565,
-      "grad_norm": 0.003718878608196974,
-      "learning_rate": 0.00019998097342276795,
-      "loss": 46.0,
-      "step": 38569
-    },
-    {
-      "epoch": 6.211441684447844,
-      "grad_norm": 0.0012801260454580188,
-      "learning_rate": 0.0001999809724359052,
-      "loss": 46.0,
-      "step": 38570
-    },
-    {
-      "epoch": 6.211602721526631,
-      "grad_norm": 0.008715521544218063,
-      "learning_rate": 0.0001999809714490169,
-      "loss": 46.0,
-      "step": 38571
-    },
-    {
-      "epoch": 6.211763758605419,
-      "grad_norm": 0.002537169260904193,
-      "learning_rate": 0.00019998097046210292,
-      "loss": 46.0,
-      "step": 38572
-    },
-    {
-      "epoch": 6.211924795684206,
-      "grad_norm": 0.0015906566986814141,
-      "learning_rate": 0.00019998096947516342,
-      "loss": 46.0,
-      "step": 38573
-    },
-    {
-      "epoch": 6.212085832762994,
-      "grad_norm": 0.0034709463361650705,
-      "learning_rate": 0.00019998096848819827,
-      "loss": 46.0,
-      "step": 38574
-    },
-    {
-      "epoch": 6.212246869841781,
-      "grad_norm": 0.003196131670847535,
-      "learning_rate": 0.00019998096750120756,
-      "loss": 46.0,
-      "step": 38575
-    },
-    {
-      "epoch": 6.212407906920569,
-      "grad_norm": 0.0008856742642819881,
-      "learning_rate": 0.00019998096651419126,
-      "loss": 46.0,
-      "step": 38576
-    },
-    {
-      "epoch": 6.212568943999356,
-      "grad_norm": 0.011310459114611149,
-      "learning_rate": 0.0001999809655271494,
-      "loss": 46.0,
-      "step": 38577
-    },
-    {
-      "epoch": 6.212729981078144,
-      "grad_norm": 0.0015778564848005772,
-      "learning_rate": 0.0001999809645400819,
-      "loss": 46.0,
-      "step": 38578
-    },
-    {
-      "epoch": 6.212891018156931,
-      "grad_norm": 0.004972171038389206,
-      "learning_rate": 0.00019998096355298883,
-      "loss": 46.0,
-      "step": 38579
-    },
-    {
-      "epoch": 6.213052055235718,
-      "grad_norm": 0.0064807613380253315,
-      "learning_rate": 0.00019998096256587018,
-      "loss": 46.0,
-      "step": 38580
-    },
-    {
-      "epoch": 6.213213092314505,
-      "grad_norm": 0.002693760907277465,
-      "learning_rate": 0.00019998096157872595,
-      "loss": 46.0,
-      "step": 38581
-    },
-    {
-      "epoch": 6.2133741293932925,
-      "grad_norm": 0.007887729443609715,
-      "learning_rate": 0.00019998096059155608,
-      "loss": 46.0,
-      "step": 38582
-    },
-    {
-      "epoch": 6.21353516647208,
-      "grad_norm": 0.00348364794626832,
-      "learning_rate": 0.00019998095960436067,
-      "loss": 46.0,
-      "step": 38583
-    },
-    {
-      "epoch": 6.213696203550867,
-      "grad_norm": 0.005970843601971865,
-      "learning_rate": 0.00019998095861713965,
-      "loss": 46.0,
-      "step": 38584
-    },
-    {
-      "epoch": 6.213857240629655,
-      "grad_norm": 0.0008491827175021172,
-      "learning_rate": 0.00019998095762989305,
-      "loss": 46.0,
-      "step": 38585
-    },
-    {
-      "epoch": 6.214018277708442,
-      "grad_norm": 0.01671157404780388,
-      "learning_rate": 0.00019998095664262085,
-      "loss": 46.0,
-      "step": 38586
-    },
-    {
-      "epoch": 6.21417931478723,
-      "grad_norm": 0.010379582643508911,
-      "learning_rate": 0.00019998095565532307,
-      "loss": 46.0,
-      "step": 38587
-    },
-    {
-      "epoch": 6.214340351866017,
-      "grad_norm": 0.009952492080628872,
-      "learning_rate": 0.00019998095466799968,
-      "loss": 46.0,
-      "step": 38588
-    },
-    {
-      "epoch": 6.214501388944805,
-      "grad_norm": 0.004560786299407482,
-      "learning_rate": 0.00019998095368065072,
-      "loss": 46.0,
-      "step": 38589
-    },
-    {
-      "epoch": 6.214662426023592,
-      "grad_norm": 0.007111283019185066,
-      "learning_rate": 0.00019998095269327615,
-      "loss": 46.0,
-      "step": 38590
-    },
-    {
-      "epoch": 6.2148234631023795,
-      "grad_norm": 0.007550255861133337,
-      "learning_rate": 0.00019998095170587602,
-      "loss": 46.0,
-      "step": 38591
-    },
-    {
-      "epoch": 6.214984500181167,
-      "grad_norm": 0.0056282998993992805,
-      "learning_rate": 0.0001999809507184503,
-      "loss": 46.0,
-      "step": 38592
-    },
-    {
-      "epoch": 6.2151455372599544,
-      "grad_norm": 0.001895827241241932,
-      "learning_rate": 0.00019998094973099894,
-      "loss": 46.0,
-      "step": 38593
-    },
-    {
-      "epoch": 6.215306574338742,
-      "grad_norm": 0.002043043030425906,
-      "learning_rate": 0.00019998094874352202,
-      "loss": 46.0,
-      "step": 38594
-    },
-    {
-      "epoch": 6.2154676114175285,
-      "grad_norm": 0.0031424269545823336,
-      "learning_rate": 0.00019998094775601952,
-      "loss": 46.0,
-      "step": 38595
-    },
-    {
-      "epoch": 6.215628648496316,
-      "grad_norm": 0.0006763574783690274,
-      "learning_rate": 0.00019998094676849145,
-      "loss": 46.0,
-      "step": 38596
-    },
-    {
-      "epoch": 6.215789685575103,
-      "grad_norm": 0.0025461832992732525,
-      "learning_rate": 0.00019998094578093774,
-      "loss": 46.0,
-      "step": 38597
-    },
-    {
-      "epoch": 6.215950722653891,
-      "grad_norm": 0.0018364378483965993,
-      "learning_rate": 0.00019998094479335845,
-      "loss": 46.0,
-      "step": 38598
-    },
-    {
-      "epoch": 6.216111759732678,
-      "grad_norm": 0.0012988555245101452,
-      "learning_rate": 0.0001999809438057536,
-      "loss": 46.0,
-      "step": 38599
-    },
-    {
-      "epoch": 6.216272796811466,
-      "grad_norm": 0.006263650022447109,
-      "learning_rate": 0.00019998094281812315,
-      "loss": 46.0,
-      "step": 38600
-    },
-    {
-      "epoch": 6.216433833890253,
-      "grad_norm": 0.0025493402499705553,
-      "learning_rate": 0.0001999809418304671,
-      "loss": 46.0,
-      "step": 38601
-    },
-    {
-      "epoch": 6.216594870969041,
-      "grad_norm": 0.004002938512712717,
-      "learning_rate": 0.00019998094084278545,
-      "loss": 46.0,
-      "step": 38602
-    },
-    {
-      "epoch": 6.216755908047828,
-      "grad_norm": 0.00786288920789957,
-      "learning_rate": 0.00019998093985507822,
-      "loss": 46.0,
-      "step": 38603
-    },
-    {
-      "epoch": 6.2169169451266155,
-      "grad_norm": 0.002239077817648649,
-      "learning_rate": 0.00019998093886734538,
-      "loss": 46.0,
-      "step": 38604
-    },
-    {
-      "epoch": 6.217077982205403,
-      "grad_norm": 0.013029851019382477,
-      "learning_rate": 0.00019998093787958697,
-      "loss": 46.0,
-      "step": 38605
-    },
-    {
-      "epoch": 6.21723901928419,
-      "grad_norm": 0.0038467231206595898,
-      "learning_rate": 0.00019998093689180298,
-      "loss": 46.0,
-      "step": 38606
-    },
-    {
-      "epoch": 6.217400056362978,
-      "grad_norm": 0.0027490751817822456,
-      "learning_rate": 0.00019998093590399343,
-      "loss": 46.0,
-      "step": 38607
-    },
-    {
-      "epoch": 6.217561093441765,
-      "grad_norm": 0.0025106542743742466,
-      "learning_rate": 0.00019998093491615824,
-      "loss": 46.0,
-      "step": 38608
-    },
-    {
-      "epoch": 6.217722130520553,
-      "grad_norm": 0.0023332072887569666,
-      "learning_rate": 0.00019998093392829746,
-      "loss": 46.0,
-      "step": 38609
-    },
-    {
-      "epoch": 6.217883167599339,
-      "grad_norm": 0.010310209356248379,
-      "learning_rate": 0.00019998093294041112,
-      "loss": 46.0,
-      "step": 38610
-    },
-    {
-      "epoch": 6.218044204678127,
-      "grad_norm": 0.005291286390274763,
-      "learning_rate": 0.00019998093195249916,
-      "loss": 46.0,
-      "step": 38611
-    },
-    {
-      "epoch": 6.218205241756914,
-      "grad_norm": 0.006393376272171736,
-      "learning_rate": 0.00019998093096456162,
-      "loss": 46.0,
-      "step": 38612
-    },
-    {
-      "epoch": 6.218366278835702,
-      "grad_norm": 0.0024698018096387386,
-      "learning_rate": 0.0001999809299765985,
-      "loss": 46.0,
-      "step": 38613
-    },
-    {
-      "epoch": 6.218527315914489,
-      "grad_norm": 0.002322337357327342,
-      "learning_rate": 0.00019998092898860977,
-      "loss": 46.0,
-      "step": 38614
-    },
-    {
-      "epoch": 6.218688352993277,
-      "grad_norm": 0.004036011174321175,
-      "learning_rate": 0.00019998092800059547,
-      "loss": 46.0,
-      "step": 38615
-    },
-    {
-      "epoch": 6.218849390072064,
-      "grad_norm": 0.002421214710921049,
-      "learning_rate": 0.00019998092701255558,
-      "loss": 46.0,
-      "step": 38616
-    },
-    {
-      "epoch": 6.2190104271508515,
-      "grad_norm": 0.010165990330278873,
-      "learning_rate": 0.00019998092602449007,
-      "loss": 46.0,
-      "step": 38617
-    },
-    {
-      "epoch": 6.219171464229639,
-      "grad_norm": 0.0024456074461340904,
-      "learning_rate": 0.000199980925036399,
-      "loss": 46.0,
-      "step": 38618
-    },
-    {
-      "epoch": 6.219332501308426,
-      "grad_norm": 0.01061293762177229,
-      "learning_rate": 0.00019998092404828236,
-      "loss": 46.0,
-      "step": 38619
-    },
-    {
-      "epoch": 6.219493538387214,
-      "grad_norm": 0.0008300011395476758,
-      "learning_rate": 0.00019998092306014006,
-      "loss": 46.0,
-      "step": 38620
-    },
-    {
-      "epoch": 6.219654575466001,
-      "grad_norm": 0.003805123269557953,
-      "learning_rate": 0.00019998092207197224,
-      "loss": 46.0,
-      "step": 38621
-    },
-    {
-      "epoch": 6.219815612544789,
-      "grad_norm": 0.013111239299178123,
-      "learning_rate": 0.0001999809210837788,
-      "loss": 46.0,
-      "step": 38622
-    },
-    {
-      "epoch": 6.219976649623576,
-      "grad_norm": 0.0020245127379894257,
-      "learning_rate": 0.00019998092009555977,
-      "loss": 46.0,
-      "step": 38623
-    },
-    {
-      "epoch": 6.220137686702364,
-      "grad_norm": 0.006429504137486219,
-      "learning_rate": 0.00019998091910731515,
-      "loss": 46.0,
-      "step": 38624
-    },
-    {
-      "epoch": 6.22029872378115,
-      "grad_norm": 0.01914132386445999,
-      "learning_rate": 0.00019998091811904495,
-      "loss": 46.0,
-      "step": 38625
-    },
-    {
-      "epoch": 6.220459760859938,
-      "grad_norm": 0.02236207015812397,
-      "learning_rate": 0.00019998091713074913,
-      "loss": 46.0,
-      "step": 38626
-    },
-    {
-      "epoch": 6.220620797938725,
-      "grad_norm": 0.02192418836057186,
-      "learning_rate": 0.00019998091614242776,
-      "loss": 46.0,
-      "step": 38627
-    },
-    {
-      "epoch": 6.220781835017513,
-      "grad_norm": 0.006403478793799877,
-      "learning_rate": 0.0001999809151540808,
-      "loss": 46.0,
-      "step": 38628
-    },
-    {
-      "epoch": 6.2209428720963,
-      "grad_norm": 0.0026490618474781513,
-      "learning_rate": 0.0001999809141657082,
-      "loss": 46.0,
-      "step": 38629
-    },
-    {
-      "epoch": 6.2211039091750875,
-      "grad_norm": 0.008898897096514702,
-      "learning_rate": 0.00019998091317731005,
-      "loss": 46.0,
-      "step": 38630
-    },
-    {
-      "epoch": 6.221264946253875,
-      "grad_norm": 0.006643269676715136,
-      "learning_rate": 0.0001999809121888863,
-      "loss": 46.0,
-      "step": 38631
-    },
-    {
-      "epoch": 6.221425983332662,
-      "grad_norm": 0.002111181616783142,
-      "learning_rate": 0.00019998091120043698,
-      "loss": 46.0,
-      "step": 38632
-    },
-    {
-      "epoch": 6.22158702041145,
-      "grad_norm": 0.004600047133862972,
-      "learning_rate": 0.00019998091021196203,
-      "loss": 46.0,
-      "step": 38633
-    },
-    {
-      "epoch": 6.221748057490237,
-      "grad_norm": 0.000806512136477977,
-      "learning_rate": 0.00019998090922346154,
-      "loss": 46.0,
-      "step": 38634
-    },
-    {
-      "epoch": 6.221909094569025,
-      "grad_norm": 0.018649008125066757,
-      "learning_rate": 0.00019998090823493544,
-      "loss": 46.0,
-      "step": 38635
-    },
-    {
-      "epoch": 6.222070131647812,
-      "grad_norm": 0.010169747285544872,
-      "learning_rate": 0.00019998090724638372,
-      "loss": 46.0,
-      "step": 38636
-    },
-    {
-      "epoch": 6.2222311687266,
-      "grad_norm": 0.0023480055388063192,
-      "learning_rate": 0.00019998090625780642,
-      "loss": 46.0,
-      "step": 38637
-    },
-    {
-      "epoch": 6.222392205805387,
-      "grad_norm": 0.007084401324391365,
-      "learning_rate": 0.00019998090526920356,
-      "loss": 46.0,
-      "step": 38638
-    },
-    {
-      "epoch": 6.222553242884174,
-      "grad_norm": 0.00897884089499712,
-      "learning_rate": 0.0001999809042805751,
-      "loss": 46.0,
-      "step": 38639
-    },
-    {
-      "epoch": 6.222714279962961,
-      "grad_norm": 0.009737041778862476,
-      "learning_rate": 0.00019998090329192105,
-      "loss": 46.0,
-      "step": 38640
-    },
-    {
-      "epoch": 6.222875317041749,
-      "grad_norm": 0.002869783900678158,
-      "learning_rate": 0.00019998090230324137,
-      "loss": 46.0,
-      "step": 38641
-    },
-    {
-      "epoch": 6.223036354120536,
-      "grad_norm": 0.0036697534378618,
-      "learning_rate": 0.00019998090131453616,
-      "loss": 46.0,
-      "step": 38642
-    },
-    {
-      "epoch": 6.2231973911993235,
-      "grad_norm": 0.004103336948901415,
-      "learning_rate": 0.00019998090032580533,
-      "loss": 46.0,
-      "step": 38643
-    },
-    {
-      "epoch": 6.223358428278111,
-      "grad_norm": 0.011155062355101109,
-      "learning_rate": 0.00019998089933704892,
-      "loss": 46.0,
-      "step": 38644
-    },
-    {
-      "epoch": 6.223519465356898,
-      "grad_norm": 0.005551514215767384,
-      "learning_rate": 0.0001999808983482669,
-      "loss": 46.0,
-      "step": 38645
-    },
-    {
-      "epoch": 6.223680502435686,
-      "grad_norm": 0.00575565779581666,
-      "learning_rate": 0.0001999808973594593,
-      "loss": 46.0,
-      "step": 38646
-    },
-    {
-      "epoch": 6.223841539514473,
-      "grad_norm": 0.008430094458162785,
-      "learning_rate": 0.0001999808963706261,
-      "loss": 46.0,
-      "step": 38647
-    },
-    {
-      "epoch": 6.224002576593261,
-      "grad_norm": 0.015129799023270607,
-      "learning_rate": 0.00019998089538176734,
-      "loss": 46.0,
-      "step": 38648
-    },
-    {
-      "epoch": 6.224163613672048,
-      "grad_norm": 0.004835840314626694,
-      "learning_rate": 0.00019998089439288296,
-      "loss": 46.0,
-      "step": 38649
-    },
-    {
-      "epoch": 6.224324650750836,
-      "grad_norm": 0.0026272109244018793,
-      "learning_rate": 0.000199980893403973,
-      "loss": 46.0,
-      "step": 38650
-    },
-    {
-      "epoch": 6.224485687829623,
-      "grad_norm": 0.009347965009510517,
-      "learning_rate": 0.00019998089241503748,
-      "loss": 46.0,
-      "step": 38651
-    },
-    {
-      "epoch": 6.2246467249084105,
-      "grad_norm": 0.023988954722881317,
-      "learning_rate": 0.00019998089142607634,
-      "loss": 46.0,
-      "step": 38652
-    },
-    {
-      "epoch": 6.224807761987197,
-      "grad_norm": 0.0048380945809185505,
-      "learning_rate": 0.00019998089043708962,
-      "loss": 46.0,
-      "step": 38653
-    },
-    {
-      "epoch": 6.2249687990659845,
-      "grad_norm": 0.0059645334258675575,
-      "learning_rate": 0.0001999808894480773,
-      "loss": 46.0,
-      "step": 38654
-    },
-    {
-      "epoch": 6.225129836144772,
-      "grad_norm": 0.01953182741999626,
-      "learning_rate": 0.00019998088845903938,
-      "loss": 46.0,
-      "step": 38655
-    },
-    {
-      "epoch": 6.2252908732235595,
-      "grad_norm": 0.009990277700126171,
-      "learning_rate": 0.00019998088746997586,
-      "loss": 46.0,
-      "step": 38656
-    },
-    {
-      "epoch": 6.225451910302347,
-      "grad_norm": 0.0029023620299994946,
-      "learning_rate": 0.0001999808864808868,
-      "loss": 46.0,
-      "step": 38657
-    },
-    {
-      "epoch": 6.225612947381134,
-      "grad_norm": 0.0032614143565297127,
-      "learning_rate": 0.00019998088549177213,
-      "loss": 46.0,
-      "step": 38658
-    },
-    {
-      "epoch": 6.225773984459922,
-      "grad_norm": 0.001424444722943008,
-      "learning_rate": 0.00019998088450263186,
-      "loss": 46.0,
-      "step": 38659
-    },
-    {
-      "epoch": 6.225935021538709,
-      "grad_norm": 0.0017049384769052267,
-      "learning_rate": 0.000199980883513466,
-      "loss": 46.0,
-      "step": 38660
-    },
-    {
-      "epoch": 6.226096058617497,
-      "grad_norm": 0.008877153508365154,
-      "learning_rate": 0.00019998088252427457,
-      "loss": 46.0,
-      "step": 38661
-    },
-    {
-      "epoch": 6.226257095696284,
-      "grad_norm": 0.0033927957993000746,
-      "learning_rate": 0.0001999808815350575,
-      "loss": 46.0,
-      "step": 38662
-    },
-    {
-      "epoch": 6.226418132775072,
-      "grad_norm": 0.01044750027358532,
-      "learning_rate": 0.00019998088054581489,
-      "loss": 46.0,
-      "step": 38663
-    },
-    {
-      "epoch": 6.226579169853859,
-      "grad_norm": 0.0124536557123065,
-      "learning_rate": 0.00019998087955654667,
-      "loss": 46.0,
-      "step": 38664
-    },
-    {
-      "epoch": 6.2267402069326465,
-      "grad_norm": 0.002921351930126548,
-      "learning_rate": 0.00019998087856725288,
-      "loss": 46.0,
-      "step": 38665
-    },
-    {
-      "epoch": 6.226901244011434,
-      "grad_norm": 0.00365428626537323,
-      "learning_rate": 0.00019998087757793347,
-      "loss": 46.0,
-      "step": 38666
-    },
-    {
-      "epoch": 6.227062281090221,
-      "grad_norm": 0.007008520420640707,
-      "learning_rate": 0.0001999808765885885,
-      "loss": 46.0,
-      "step": 38667
-    },
-    {
-      "epoch": 6.227223318169008,
-      "grad_norm": 0.004778375383466482,
-      "learning_rate": 0.0001999808755992179,
-      "loss": 46.0,
-      "step": 38668
-    },
-    {
-      "epoch": 6.227384355247795,
-      "grad_norm": 0.006018646992743015,
-      "learning_rate": 0.00019998087460982173,
-      "loss": 46.0,
-      "step": 38669
-    },
-    {
-      "epoch": 6.227545392326583,
-      "grad_norm": 0.004066370893269777,
-      "learning_rate": 0.0001999808736204,
-      "loss": 46.0,
-      "step": 38670
-    },
-    {
-      "epoch": 6.22770642940537,
-      "grad_norm": 0.0016448433743789792,
-      "learning_rate": 0.00019998087263095262,
-      "loss": 46.0,
-      "step": 38671
-    },
-    {
-      "epoch": 6.227867466484158,
-      "grad_norm": 0.0017839446663856506,
-      "learning_rate": 0.00019998087164147972,
-      "loss": 46.0,
-      "step": 38672
-    },
-    {
-      "epoch": 6.228028503562945,
-      "grad_norm": 0.014859459362924099,
-      "learning_rate": 0.0001999808706519812,
-      "loss": 46.0,
-      "step": 38673
-    },
-    {
-      "epoch": 6.228189540641733,
-      "grad_norm": 0.002171889878809452,
-      "learning_rate": 0.00019998086966245706,
-      "loss": 46.0,
-      "step": 38674
-    },
-    {
-      "epoch": 6.22835057772052,
-      "grad_norm": 0.012175188399851322,
-      "learning_rate": 0.00019998086867290736,
-      "loss": 46.0,
-      "step": 38675
-    },
-    {
-      "epoch": 6.228511614799308,
-      "grad_norm": 0.012284522876143456,
-      "learning_rate": 0.00019998086768333205,
-      "loss": 46.0,
-      "step": 38676
-    },
-    {
-      "epoch": 6.228672651878095,
-      "grad_norm": 0.0025695443619042635,
-      "learning_rate": 0.00019998086669373118,
-      "loss": 46.0,
-      "step": 38677
-    },
-    {
-      "epoch": 6.2288336889568825,
-      "grad_norm": 0.009514286182820797,
-      "learning_rate": 0.00019998086570410467,
-      "loss": 46.0,
-      "step": 38678
-    },
-    {
-      "epoch": 6.22899472603567,
-      "grad_norm": 0.0032513632904738188,
-      "learning_rate": 0.00019998086471445262,
-      "loss": 46.0,
-      "step": 38679
-    },
-    {
-      "epoch": 6.229155763114457,
-      "grad_norm": 0.002580565633252263,
-      "learning_rate": 0.00019998086372477496,
-      "loss": 46.0,
-      "step": 38680
-    },
-    {
-      "epoch": 6.229316800193245,
-      "grad_norm": 0.007044296711683273,
-      "learning_rate": 0.00019998086273507172,
-      "loss": 46.0,
-      "step": 38681
-    },
-    {
-      "epoch": 6.229477837272032,
-      "grad_norm": 0.004722347483038902,
-      "learning_rate": 0.00019998086174534288,
-      "loss": 46.0,
-      "step": 38682
-    },
-    {
-      "epoch": 6.229638874350819,
-      "grad_norm": 0.01471523568034172,
-      "learning_rate": 0.00019998086075558846,
-      "loss": 46.0,
-      "step": 38683
-    },
-    {
-      "epoch": 6.229799911429606,
-      "grad_norm": 0.002049632603302598,
-      "learning_rate": 0.00019998085976580845,
-      "loss": 46.0,
-      "step": 38684
-    },
-    {
-      "epoch": 6.229960948508394,
-      "grad_norm": 0.007146105635911226,
-      "learning_rate": 0.00019998085877600286,
-      "loss": 46.0,
-      "step": 38685
-    },
-    {
-      "epoch": 6.230121985587181,
-      "grad_norm": 0.006557625252753496,
-      "learning_rate": 0.00019998085778617165,
-      "loss": 46.0,
-      "step": 38686
-    },
-    {
-      "epoch": 6.230283022665969,
-      "grad_norm": 0.0206775963306427,
-      "learning_rate": 0.00019998085679631485,
-      "loss": 46.0,
-      "step": 38687
-    },
-    {
-      "epoch": 6.230444059744756,
-      "grad_norm": 0.00934943649917841,
-      "learning_rate": 0.0001999808558064325,
-      "loss": 46.0,
-      "step": 38688
-    },
-    {
-      "epoch": 6.230605096823544,
-      "grad_norm": 0.02179321087896824,
-      "learning_rate": 0.0001999808548165245,
-      "loss": 46.0,
-      "step": 38689
-    },
-    {
-      "epoch": 6.230766133902331,
-      "grad_norm": 0.0011299765901640058,
-      "learning_rate": 0.00019998085382659097,
-      "loss": 46.0,
-      "step": 38690
-    },
-    {
-      "epoch": 6.2309271709811185,
-      "grad_norm": 0.012291212566196918,
-      "learning_rate": 0.00019998085283663182,
-      "loss": 46.0,
-      "step": 38691
-    },
-    {
-      "epoch": 6.231088208059906,
-      "grad_norm": 0.005116683896631002,
-      "learning_rate": 0.0001999808518466471,
-      "loss": 46.0,
-      "step": 38692
-    },
-    {
-      "epoch": 6.231249245138693,
-      "grad_norm": 0.020011739805340767,
-      "learning_rate": 0.00019998085085663677,
-      "loss": 46.0,
-      "step": 38693
-    },
-    {
-      "epoch": 6.231410282217481,
-      "grad_norm": 0.010386831127107143,
-      "learning_rate": 0.00019998084986660083,
-      "loss": 46.0,
-      "step": 38694
-    },
-    {
-      "epoch": 6.231571319296268,
-      "grad_norm": 0.0034894703421741724,
-      "learning_rate": 0.00019998084887653934,
-      "loss": 46.0,
-      "step": 38695
-    },
-    {
-      "epoch": 6.231732356375056,
-      "grad_norm": 0.005942290183156729,
-      "learning_rate": 0.00019998084788645223,
-      "loss": 46.0,
-      "step": 38696
-    },
-    {
-      "epoch": 6.231893393453843,
-      "grad_norm": 0.00561217125505209,
-      "learning_rate": 0.00019998084689633956,
-      "loss": 46.0,
-      "step": 38697
-    },
-    {
-      "epoch": 6.23205443053263,
-      "grad_norm": 0.01627948507666588,
-      "learning_rate": 0.00019998084590620128,
-      "loss": 46.0,
-      "step": 38698
-    },
-    {
-      "epoch": 6.232215467611417,
-      "grad_norm": 0.009304588660597801,
-      "learning_rate": 0.0001999808449160374,
-      "loss": 46.0,
-      "step": 38699
-    },
-    {
-      "epoch": 6.232376504690205,
-      "grad_norm": 0.003932025283575058,
-      "learning_rate": 0.00019998084392584795,
-      "loss": 46.0,
-      "step": 38700
-    },
-    {
-      "epoch": 6.232537541768992,
-      "grad_norm": 0.0026350137777626514,
-      "learning_rate": 0.0001999808429356329,
-      "loss": 46.0,
-      "step": 38701
-    },
-    {
-      "epoch": 6.23269857884778,
-      "grad_norm": 0.0015443023294210434,
-      "learning_rate": 0.00019998084194539225,
-      "loss": 46.0,
-      "step": 38702
-    },
-    {
-      "epoch": 6.232859615926567,
-      "grad_norm": 0.0015425975434482098,
-      "learning_rate": 0.00019998084095512603,
-      "loss": 46.0,
-      "step": 38703
-    },
-    {
-      "epoch": 6.2330206530053545,
-      "grad_norm": 0.006231886334717274,
-      "learning_rate": 0.00019998083996483423,
-      "loss": 46.0,
-      "step": 38704
-    },
-    {
-      "epoch": 6.233181690084142,
-      "grad_norm": 0.0019597583450376987,
-      "learning_rate": 0.00019998083897451683,
-      "loss": 46.0,
-      "step": 38705
-    },
-    {
-      "epoch": 6.233342727162929,
-      "grad_norm": 0.004525311291217804,
-      "learning_rate": 0.00019998083798417385,
-      "loss": 46.0,
-      "step": 38706
-    },
-    {
-      "epoch": 6.233503764241717,
-      "grad_norm": 0.006644236855208874,
-      "learning_rate": 0.00019998083699380526,
-      "loss": 46.0,
-      "step": 38707
-    },
-    {
-      "epoch": 6.233664801320504,
-      "grad_norm": 0.0036862134002149105,
-      "learning_rate": 0.00019998083600341108,
-      "loss": 46.0,
-      "step": 38708
-    },
-    {
-      "epoch": 6.233825838399292,
-      "grad_norm": 0.0038347700610756874,
-      "learning_rate": 0.0001999808350129913,
-      "loss": 46.0,
-      "step": 38709
-    },
-    {
-      "epoch": 6.233986875478079,
-      "grad_norm": 0.0073648057878017426,
-      "learning_rate": 0.00019998083402254595,
-      "loss": 46.0,
-      "step": 38710
-    },
-    {
-      "epoch": 6.234147912556867,
-      "grad_norm": 0.00634078960865736,
-      "learning_rate": 0.000199980833032075,
-      "loss": 46.0,
-      "step": 38711
-    },
-    {
-      "epoch": 6.234308949635653,
-      "grad_norm": 0.00382929272018373,
-      "learning_rate": 0.00019998083204157848,
-      "loss": 46.0,
-      "step": 38712
-    },
-    {
-      "epoch": 6.234469986714441,
-      "grad_norm": 0.001045657554641366,
-      "learning_rate": 0.00019998083105105633,
-      "loss": 46.0,
-      "step": 38713
-    },
-    {
-      "epoch": 6.234631023793228,
-      "grad_norm": 0.0015196145977824926,
-      "learning_rate": 0.00019998083006050863,
-      "loss": 46.0,
-      "step": 38714
-    },
-    {
-      "epoch": 6.2347920608720155,
-      "grad_norm": 0.003375179599970579,
-      "learning_rate": 0.00019998082906993534,
-      "loss": 46.0,
-      "step": 38715
-    },
-    {
-      "epoch": 6.234953097950803,
-      "grad_norm": 0.001859193667769432,
-      "learning_rate": 0.00019998082807933643,
-      "loss": 46.0,
-      "step": 38716
-    },
-    {
-      "epoch": 6.23511413502959,
-      "grad_norm": 0.0022404841147363186,
-      "learning_rate": 0.00019998082708871194,
-      "loss": 46.0,
-      "step": 38717
-    },
-    {
-      "epoch": 6.235275172108378,
-      "grad_norm": 0.005487280897796154,
-      "learning_rate": 0.00019998082609806188,
-      "loss": 46.0,
-      "step": 38718
-    },
-    {
-      "epoch": 6.235436209187165,
-      "grad_norm": 0.0018628500401973724,
-      "learning_rate": 0.00019998082510738622,
-      "loss": 46.0,
-      "step": 38719
-    },
-    {
-      "epoch": 6.235597246265953,
-      "grad_norm": 0.0027922329027205706,
-      "learning_rate": 0.00019998082411668496,
-      "loss": 46.0,
-      "step": 38720
-    },
-    {
-      "epoch": 6.23575828334474,
-      "grad_norm": 0.0011255184654146433,
-      "learning_rate": 0.00019998082312595812,
-      "loss": 46.0,
-      "step": 38721
-    },
-    {
-      "epoch": 6.235919320423528,
-      "grad_norm": 0.014780014753341675,
-      "learning_rate": 0.00019998082213520566,
-      "loss": 46.0,
-      "step": 38722
-    },
-    {
-      "epoch": 6.236080357502315,
-      "grad_norm": 0.002269803313538432,
-      "learning_rate": 0.00019998082114442765,
-      "loss": 46.0,
-      "step": 38723
-    },
-    {
-      "epoch": 6.236241394581103,
-      "grad_norm": 0.0055558690801262856,
-      "learning_rate": 0.00019998082015362402,
-      "loss": 46.0,
-      "step": 38724
-    },
-    {
-      "epoch": 6.23640243165989,
-      "grad_norm": 0.007153920829296112,
-      "learning_rate": 0.00019998081916279483,
-      "loss": 46.0,
-      "step": 38725
-    },
-    {
-      "epoch": 6.2365634687386775,
-      "grad_norm": 0.004320100415498018,
-      "learning_rate": 0.00019998081817194002,
-      "loss": 46.0,
-      "step": 38726
-    },
-    {
-      "epoch": 6.236724505817464,
-      "grad_norm": 0.0030976238194853067,
-      "learning_rate": 0.00019998081718105963,
-      "loss": 46.0,
-      "step": 38727
-    },
-    {
-      "epoch": 6.2368855428962515,
-      "grad_norm": 0.0020188416820019484,
-      "learning_rate": 0.00019998081619015368,
-      "loss": 46.0,
-      "step": 38728
-    },
-    {
-      "epoch": 6.237046579975039,
-      "grad_norm": 0.0012540711322799325,
-      "learning_rate": 0.00019998081519922208,
-      "loss": 46.0,
-      "step": 38729
-    },
-    {
-      "epoch": 6.237207617053826,
-      "grad_norm": 0.004225165583193302,
-      "learning_rate": 0.00019998081420826496,
-      "loss": 46.0,
-      "step": 38730
-    },
-    {
-      "epoch": 6.237368654132614,
-      "grad_norm": 0.0022625618148595095,
-      "learning_rate": 0.0001999808132172822,
-      "loss": 46.0,
-      "step": 38731
-    },
-    {
-      "epoch": 6.237529691211401,
-      "grad_norm": 0.006432642694562674,
-      "learning_rate": 0.00019998081222627386,
-      "loss": 46.0,
-      "step": 38732
-    },
-    {
-      "epoch": 6.237690728290189,
-      "grad_norm": 0.006834803149104118,
-      "learning_rate": 0.00019998081123523995,
-      "loss": 46.0,
-      "step": 38733
-    },
-    {
-      "epoch": 6.237851765368976,
-      "grad_norm": 0.008049764670431614,
-      "learning_rate": 0.00019998081024418042,
-      "loss": 46.0,
-      "step": 38734
-    },
-    {
-      "epoch": 6.238012802447764,
-      "grad_norm": 0.003339129965752363,
-      "learning_rate": 0.00019998080925309535,
-      "loss": 46.0,
-      "step": 38735
-    },
-    {
-      "epoch": 6.238173839526551,
-      "grad_norm": 0.005171710625290871,
-      "learning_rate": 0.00019998080826198465,
-      "loss": 46.0,
-      "step": 38736
-    },
-    {
-      "epoch": 6.238334876605339,
-      "grad_norm": 0.0026569857727736235,
-      "learning_rate": 0.00019998080727084836,
-      "loss": 46.0,
-      "step": 38737
-    },
-    {
-      "epoch": 6.238495913684126,
-      "grad_norm": 0.003168070688843727,
-      "learning_rate": 0.00019998080627968648,
-      "loss": 46.0,
-      "step": 38738
-    },
-    {
-      "epoch": 6.2386569507629135,
-      "grad_norm": 0.005243937484920025,
-      "learning_rate": 0.00019998080528849902,
-      "loss": 46.0,
-      "step": 38739
-    },
-    {
-      "epoch": 6.238817987841701,
-      "grad_norm": 0.0015928170178085566,
-      "learning_rate": 0.00019998080429728596,
-      "loss": 46.0,
-      "step": 38740
-    },
-    {
-      "epoch": 6.2389790249204875,
-      "grad_norm": 0.00266423006542027,
-      "learning_rate": 0.00019998080330604733,
-      "loss": 46.0,
-      "step": 38741
-    },
-    {
-      "epoch": 6.239140061999275,
-      "grad_norm": 0.0020855246111750603,
-      "learning_rate": 0.00019998080231478307,
-      "loss": 46.0,
-      "step": 38742
-    },
-    {
-      "epoch": 6.239301099078062,
-      "grad_norm": 0.009634270332753658,
-      "learning_rate": 0.00019998080132349326,
-      "loss": 46.0,
-      "step": 38743
-    },
-    {
-      "epoch": 6.23946213615685,
-      "grad_norm": 0.005434541497379541,
-      "learning_rate": 0.00019998080033217783,
-      "loss": 46.0,
-      "step": 38744
-    },
-    {
-      "epoch": 6.239623173235637,
-      "grad_norm": 0.0017689937958493829,
-      "learning_rate": 0.00019998079934083684,
-      "loss": 46.0,
-      "step": 38745
-    },
-    {
-      "epoch": 6.239784210314425,
-      "grad_norm": 0.015064570121467113,
-      "learning_rate": 0.00019998079834947024,
-      "loss": 46.0,
-      "step": 38746
-    },
-    {
-      "epoch": 6.239945247393212,
-      "grad_norm": 0.00765609135851264,
-      "learning_rate": 0.00019998079735807808,
-      "loss": 46.0,
-      "step": 38747
-    },
-    {
-      "epoch": 6.240106284472,
-      "grad_norm": 0.003262442070990801,
-      "learning_rate": 0.0001999807963666603,
-      "loss": 46.0,
-      "step": 38748
-    },
-    {
-      "epoch": 6.240267321550787,
-      "grad_norm": 0.0124131478369236,
-      "learning_rate": 0.00019998079537521693,
-      "loss": 46.0,
-      "step": 38749
-    },
-    {
-      "epoch": 6.240428358629575,
-      "grad_norm": 0.008744933642446995,
-      "learning_rate": 0.00019998079438374798,
-      "loss": 46.0,
-      "step": 38750
-    },
-    {
-      "epoch": 6.240589395708362,
-      "grad_norm": 0.002951371716335416,
-      "learning_rate": 0.00019998079339225342,
-      "loss": 46.0,
-      "step": 38751
-    },
-    {
-      "epoch": 6.2407504327871495,
-      "grad_norm": 0.0077109141275286674,
-      "learning_rate": 0.00019998079240073332,
-      "loss": 46.0,
-      "step": 38752
-    },
-    {
-      "epoch": 6.240911469865937,
-      "grad_norm": 0.0024209748953580856,
-      "learning_rate": 0.00019998079140918758,
-      "loss": 46.0,
-      "step": 38753
-    },
-    {
-      "epoch": 6.241072506944724,
-      "grad_norm": 0.0029037443455308676,
-      "learning_rate": 0.00019998079041761625,
-      "loss": 46.0,
-      "step": 38754
-    },
-    {
-      "epoch": 6.241233544023512,
-      "grad_norm": 0.0021141646429896355,
-      "learning_rate": 0.00019998078942601937,
-      "loss": 46.0,
-      "step": 38755
-    },
-    {
-      "epoch": 6.241394581102298,
-      "grad_norm": 0.0038508321158587933,
-      "learning_rate": 0.00019998078843439684,
-      "loss": 46.0,
-      "step": 38756
-    },
-    {
-      "epoch": 6.241555618181086,
-      "grad_norm": 0.02414633333683014,
-      "learning_rate": 0.0001999807874427488,
-      "loss": 46.0,
-      "step": 38757
-    },
-    {
-      "epoch": 6.241716655259873,
-      "grad_norm": 0.010605452582240105,
-      "learning_rate": 0.00019998078645107513,
-      "loss": 46.0,
-      "step": 38758
-    },
-    {
-      "epoch": 6.241877692338661,
-      "grad_norm": 0.003035190049558878,
-      "learning_rate": 0.00019998078545937584,
-      "loss": 46.0,
-      "step": 38759
-    },
-    {
-      "epoch": 6.242038729417448,
-      "grad_norm": 0.0018731292802840471,
-      "learning_rate": 0.000199980784467651,
-      "loss": 46.0,
-      "step": 38760
-    },
-    {
-      "epoch": 6.242199766496236,
-      "grad_norm": 0.0007259367848746479,
-      "learning_rate": 0.00019998078347590055,
-      "loss": 46.0,
-      "step": 38761
-    },
-    {
-      "epoch": 6.242360803575023,
-      "grad_norm": 0.0016311903018504381,
-      "learning_rate": 0.00019998078248412453,
-      "loss": 46.0,
-      "step": 38762
-    },
-    {
-      "epoch": 6.2425218406538105,
-      "grad_norm": 0.00248343450948596,
-      "learning_rate": 0.0001999807814923229,
-      "loss": 46.0,
-      "step": 38763
-    },
-    {
-      "epoch": 6.242682877732598,
-      "grad_norm": 0.0009264177642762661,
-      "learning_rate": 0.00019998078050049566,
-      "loss": 46.0,
-      "step": 38764
-    },
-    {
-      "epoch": 6.2428439148113855,
-      "grad_norm": 0.009981558658182621,
-      "learning_rate": 0.0001999807795086429,
-      "loss": 46.0,
-      "step": 38765
-    },
-    {
-      "epoch": 6.243004951890173,
-      "grad_norm": 0.013788151554763317,
-      "learning_rate": 0.00019998077851676448,
-      "loss": 46.0,
-      "step": 38766
-    },
-    {
-      "epoch": 6.24316598896896,
-      "grad_norm": 0.003375322325155139,
-      "learning_rate": 0.00019998077752486052,
-      "loss": 46.0,
-      "step": 38767
-    },
-    {
-      "epoch": 6.243327026047748,
-      "grad_norm": 0.014961320906877518,
-      "learning_rate": 0.0001999807765329309,
-      "loss": 46.0,
-      "step": 38768
-    },
-    {
-      "epoch": 6.243488063126535,
-      "grad_norm": 0.0045212144032120705,
-      "learning_rate": 0.00019998077554097575,
-      "loss": 46.0,
-      "step": 38769
-    },
-    {
-      "epoch": 6.243649100205323,
-      "grad_norm": 0.009469741024076939,
-      "learning_rate": 0.000199980774548995,
-      "loss": 46.0,
-      "step": 38770
-    },
-    {
-      "epoch": 6.243810137284109,
-      "grad_norm": 0.014732648618519306,
-      "learning_rate": 0.00019998077355698867,
-      "loss": 46.0,
-      "step": 38771
-    },
-    {
-      "epoch": 6.243971174362897,
-      "grad_norm": 0.00553680257871747,
-      "learning_rate": 0.00019998077256495674,
-      "loss": 46.0,
-      "step": 38772
-    },
-    {
-      "epoch": 6.244132211441684,
-      "grad_norm": 0.011354890652000904,
-      "learning_rate": 0.00019998077157289923,
-      "loss": 46.0,
-      "step": 38773
-    },
-    {
-      "epoch": 6.244293248520472,
-      "grad_norm": 0.002924802014604211,
-      "learning_rate": 0.0001999807705808161,
-      "loss": 46.0,
-      "step": 38774
-    },
-    {
-      "epoch": 6.244454285599259,
-      "grad_norm": 0.012061532586812973,
-      "learning_rate": 0.0001999807695887074,
-      "loss": 46.0,
-      "step": 38775
-    },
-    {
-      "epoch": 6.2446153226780465,
-      "grad_norm": 0.0025552804581820965,
-      "learning_rate": 0.0001999807685965731,
-      "loss": 46.0,
-      "step": 38776
-    },
-    {
-      "epoch": 6.244776359756834,
-      "grad_norm": 0.00809148233383894,
-      "learning_rate": 0.0001999807676044132,
-      "loss": 46.0,
-      "step": 38777
-    },
-    {
-      "epoch": 6.244937396835621,
-      "grad_norm": 0.004028466064482927,
-      "learning_rate": 0.00019998076661222777,
-      "loss": 46.0,
-      "step": 38778
-    },
-    {
-      "epoch": 6.245098433914409,
-      "grad_norm": 0.004658353049308062,
-      "learning_rate": 0.00019998076562001668,
-      "loss": 46.0,
-      "step": 38779
-    },
-    {
-      "epoch": 6.245259470993196,
-      "grad_norm": 0.012473132461309433,
-      "learning_rate": 0.00019998076462778003,
-      "loss": 46.0,
-      "step": 38780
-    },
-    {
-      "epoch": 6.245420508071984,
-      "grad_norm": 0.008844288997352123,
-      "learning_rate": 0.00019998076363551782,
-      "loss": 46.0,
-      "step": 38781
-    },
-    {
-      "epoch": 6.245581545150771,
-      "grad_norm": 0.006389549933373928,
-      "learning_rate": 0.00019998076264322995,
-      "loss": 46.0,
-      "step": 38782
-    },
-    {
-      "epoch": 6.245742582229559,
-      "grad_norm": 0.0013734856620430946,
-      "learning_rate": 0.00019998076165091654,
-      "loss": 46.0,
-      "step": 38783
-    },
-    {
-      "epoch": 6.245903619308346,
-      "grad_norm": 0.007066596299409866,
-      "learning_rate": 0.00019998076065857754,
-      "loss": 46.0,
-      "step": 38784
-    },
-    {
-      "epoch": 6.246064656387133,
-      "grad_norm": 0.00653229421004653,
-      "learning_rate": 0.00019998075966621293,
-      "loss": 46.0,
-      "step": 38785
-    },
-    {
-      "epoch": 6.24622569346592,
-      "grad_norm": 0.0075106387957930565,
-      "learning_rate": 0.00019998075867382276,
-      "loss": 46.0,
-      "step": 38786
-    },
-    {
-      "epoch": 6.246386730544708,
-      "grad_norm": 0.008485783822834492,
-      "learning_rate": 0.00019998075768140695,
-      "loss": 46.0,
-      "step": 38787
-    },
-    {
-      "epoch": 6.246547767623495,
-      "grad_norm": 0.005509641487151384,
-      "learning_rate": 0.00019998075668896558,
-      "loss": 46.0,
-      "step": 38788
-    },
-    {
-      "epoch": 6.2467088047022825,
-      "grad_norm": 0.005409873090684414,
-      "learning_rate": 0.00019998075569649864,
-      "loss": 46.0,
-      "step": 38789
-    },
-    {
-      "epoch": 6.24686984178107,
-      "grad_norm": 0.009433098137378693,
-      "learning_rate": 0.00019998075470400607,
-      "loss": 46.0,
-      "step": 38790
-    },
-    {
-      "epoch": 6.247030878859857,
-      "grad_norm": 0.013554275035858154,
-      "learning_rate": 0.00019998075371148794,
-      "loss": 46.0,
-      "step": 38791
-    },
-    {
-      "epoch": 6.247191915938645,
-      "grad_norm": 0.006548671051859856,
-      "learning_rate": 0.00019998075271894422,
-      "loss": 46.0,
-      "step": 38792
-    },
-    {
-      "epoch": 6.247352953017432,
-      "grad_norm": 0.008764144964516163,
-      "learning_rate": 0.00019998075172637488,
-      "loss": 46.0,
-      "step": 38793
-    },
-    {
-      "epoch": 6.24751399009622,
-      "grad_norm": 0.023204488679766655,
-      "learning_rate": 0.00019998075073377996,
-      "loss": 46.0,
-      "step": 38794
-    },
-    {
-      "epoch": 6.247675027175007,
-      "grad_norm": 0.022843537852168083,
-      "learning_rate": 0.00019998074974115948,
-      "loss": 46.0,
-      "step": 38795
-    },
-    {
-      "epoch": 6.247836064253795,
-      "grad_norm": 0.005780502688139677,
-      "learning_rate": 0.00019998074874851338,
-      "loss": 46.0,
-      "step": 38796
-    },
-    {
-      "epoch": 6.247997101332582,
-      "grad_norm": 0.007769168820232153,
-      "learning_rate": 0.0001999807477558417,
-      "loss": 46.0,
-      "step": 38797
-    },
-    {
-      "epoch": 6.24815813841137,
-      "grad_norm": 0.0015627213288098574,
-      "learning_rate": 0.00019998074676314442,
-      "loss": 46.0,
-      "step": 38798
-    },
-    {
-      "epoch": 6.248319175490157,
-      "grad_norm": 0.0024867020547389984,
-      "learning_rate": 0.00019998074577042157,
-      "loss": 46.0,
-      "step": 38799
-    },
-    {
-      "epoch": 6.248480212568944,
-      "grad_norm": 0.004033821634948254,
-      "learning_rate": 0.0001999807447776731,
-      "loss": 46.0,
-      "step": 38800
-    },
-    {
-      "epoch": 6.248641249647731,
-      "grad_norm": 0.005027530714869499,
-      "learning_rate": 0.00019998074378489906,
-      "loss": 46.0,
-      "step": 38801
-    },
-    {
-      "epoch": 6.2488022867265185,
-      "grad_norm": 0.0021902115549892187,
-      "learning_rate": 0.00019998074279209944,
-      "loss": 46.0,
-      "step": 38802
-    },
-    {
-      "epoch": 6.248963323805306,
-      "grad_norm": 0.003065049648284912,
-      "learning_rate": 0.0001999807417992742,
-      "loss": 46.0,
-      "step": 38803
-    },
-    {
-      "epoch": 6.249124360884093,
-      "grad_norm": 0.010861421003937721,
-      "learning_rate": 0.0001999807408064234,
-      "loss": 46.0,
-      "step": 38804
-    },
-    {
-      "epoch": 6.249285397962881,
-      "grad_norm": 0.003233975963667035,
-      "learning_rate": 0.000199980739813547,
-      "loss": 46.0,
-      "step": 38805
-    },
-    {
-      "epoch": 6.249446435041668,
-      "grad_norm": 0.005381418392062187,
-      "learning_rate": 0.00019998073882064504,
-      "loss": 46.0,
-      "step": 38806
-    },
-    {
-      "epoch": 6.249607472120456,
-      "grad_norm": 0.001557415584102273,
-      "learning_rate": 0.00019998073782771743,
-      "loss": 46.0,
-      "step": 38807
-    },
-    {
-      "epoch": 6.249768509199243,
-      "grad_norm": 0.015568535774946213,
-      "learning_rate": 0.00019998073683476428,
-      "loss": 46.0,
-      "step": 38808
-    },
-    {
-      "epoch": 6.249929546278031,
-      "grad_norm": 0.002914362819865346,
-      "learning_rate": 0.00019998073584178553,
-      "loss": 46.0,
-      "step": 38809
-    },
-    {
-      "epoch": 6.250090583356818,
-      "grad_norm": 0.007278406526893377,
-      "learning_rate": 0.00019998073484878118,
-      "loss": 46.0,
-      "step": 38810
-    },
-    {
-      "epoch": 6.250251620435606,
-      "grad_norm": 0.000835315091535449,
-      "learning_rate": 0.0001999807338557512,
-      "loss": 46.0,
-      "step": 38811
-    },
-    {
-      "epoch": 6.250412657514393,
-      "grad_norm": 0.004674816038459539,
-      "learning_rate": 0.0001999807328626957,
-      "loss": 46.0,
-      "step": 38812
-    },
-    {
-      "epoch": 6.2505736945931805,
-      "grad_norm": 0.004328897222876549,
-      "learning_rate": 0.00019998073186961457,
-      "loss": 46.0,
-      "step": 38813
-    },
-    {
-      "epoch": 6.250734731671967,
-      "grad_norm": 0.0014253858244046569,
-      "learning_rate": 0.00019998073087650785,
-      "loss": 46.0,
-      "step": 38814
-    },
-    {
-      "epoch": 6.2508957687507545,
-      "grad_norm": 0.004605847876518965,
-      "learning_rate": 0.00019998072988337555,
-      "loss": 46.0,
-      "step": 38815
-    },
-    {
-      "epoch": 6.251056805829542,
-      "grad_norm": 0.004812745843082666,
-      "learning_rate": 0.00019998072889021765,
-      "loss": 46.0,
-      "step": 38816
-    },
-    {
-      "epoch": 6.251217842908329,
-      "grad_norm": 0.005110305733978748,
-      "learning_rate": 0.00019998072789703417,
-      "loss": 46.0,
-      "step": 38817
-    },
-    {
-      "epoch": 6.251378879987117,
-      "grad_norm": 0.0016644394490867853,
-      "learning_rate": 0.0001999807269038251,
-      "loss": 46.0,
-      "step": 38818
-    },
-    {
-      "epoch": 6.251539917065904,
-      "grad_norm": 0.009690712206065655,
-      "learning_rate": 0.00019998072591059045,
-      "loss": 46.0,
-      "step": 38819
-    },
-    {
-      "epoch": 6.251700954144692,
-      "grad_norm": 0.006749164313077927,
-      "learning_rate": 0.00019998072491733023,
-      "loss": 46.0,
-      "step": 38820
-    },
-    {
-      "epoch": 6.251861991223479,
-      "grad_norm": 0.009052845649421215,
-      "learning_rate": 0.00019998072392404437,
-      "loss": 46.0,
-      "step": 38821
-    },
-    {
-      "epoch": 6.252023028302267,
-      "grad_norm": 0.016530882567167282,
-      "learning_rate": 0.00019998072293073293,
-      "loss": 46.0,
-      "step": 38822
-    },
-    {
-      "epoch": 6.252184065381054,
-      "grad_norm": 0.00537888752296567,
-      "learning_rate": 0.00019998072193739592,
-      "loss": 46.0,
-      "step": 38823
-    },
-    {
-      "epoch": 6.2523451024598415,
-      "grad_norm": 0.005701381713151932,
-      "learning_rate": 0.0001999807209440333,
-      "loss": 46.0,
-      "step": 38824
-    },
-    {
-      "epoch": 6.252506139538629,
-      "grad_norm": 0.0031267041340470314,
-      "learning_rate": 0.0001999807199506451,
-      "loss": 46.0,
-      "step": 38825
-    },
-    {
-      "epoch": 6.2526671766174164,
-      "grad_norm": 0.004520677030086517,
-      "learning_rate": 0.0001999807189572313,
-      "loss": 46.0,
-      "step": 38826
-    },
-    {
-      "epoch": 6.252828213696204,
-      "grad_norm": 0.0032734551932662725,
-      "learning_rate": 0.00019998071796379195,
-      "loss": 46.0,
-      "step": 38827
-    },
-    {
-      "epoch": 6.252989250774991,
-      "grad_norm": 0.0007657860987819731,
-      "learning_rate": 0.000199980716970327,
-      "loss": 46.0,
-      "step": 38828
-    },
-    {
-      "epoch": 6.253150287853778,
-      "grad_norm": 0.006113671697676182,
-      "learning_rate": 0.0001999807159768364,
-      "loss": 46.0,
-      "step": 38829
-    },
-    {
-      "epoch": 6.253311324932565,
-      "grad_norm": 0.0017102460842579603,
-      "learning_rate": 0.00019998071498332026,
-      "loss": 46.0,
-      "step": 38830
-    },
-    {
-      "epoch": 6.253472362011353,
-      "grad_norm": 0.002060038037598133,
-      "learning_rate": 0.0001999807139897785,
-      "loss": 46.0,
-      "step": 38831
-    },
-    {
-      "epoch": 6.25363339909014,
-      "grad_norm": 0.0160819161683321,
-      "learning_rate": 0.0001999807129962112,
-      "loss": 46.0,
-      "step": 38832
-    },
-    {
-      "epoch": 6.253794436168928,
-      "grad_norm": 0.0016725462628528476,
-      "learning_rate": 0.0001999807120026183,
-      "loss": 46.0,
-      "step": 38833
-    },
-    {
-      "epoch": 6.253955473247715,
-      "grad_norm": 0.0012638554908335209,
-      "learning_rate": 0.00019998071100899977,
-      "loss": 46.0,
-      "step": 38834
-    },
-    {
-      "epoch": 6.254116510326503,
-      "grad_norm": 0.005010468885302544,
-      "learning_rate": 0.00019998071001535567,
-      "loss": 46.0,
-      "step": 38835
-    },
-    {
-      "epoch": 6.25427754740529,
-      "grad_norm": 0.008706933818757534,
-      "learning_rate": 0.00019998070902168597,
-      "loss": 46.0,
-      "step": 38836
-    },
-    {
-      "epoch": 6.2544385844840775,
-      "grad_norm": 0.0068745906464755535,
-      "learning_rate": 0.0001999807080279907,
-      "loss": 46.0,
-      "step": 38837
-    },
-    {
-      "epoch": 6.254599621562865,
-      "grad_norm": 0.002690599299967289,
-      "learning_rate": 0.00019998070703426983,
-      "loss": 46.0,
-      "step": 38838
-    },
-    {
-      "epoch": 6.254760658641652,
-      "grad_norm": 0.004016812890768051,
-      "learning_rate": 0.00019998070604052338,
-      "loss": 46.0,
-      "step": 38839
-    },
-    {
-      "epoch": 6.25492169572044,
-      "grad_norm": 0.0023593066725879908,
-      "learning_rate": 0.00019998070504675133,
-      "loss": 46.0,
-      "step": 38840
-    },
-    {
-      "epoch": 6.255082732799227,
-      "grad_norm": 0.0017565255984663963,
-      "learning_rate": 0.00019998070405295368,
-      "loss": 46.0,
-      "step": 38841
-    },
-    {
-      "epoch": 6.255243769878015,
-      "grad_norm": 0.002819111803546548,
-      "learning_rate": 0.00019998070305913046,
-      "loss": 46.0,
-      "step": 38842
-    },
-    {
-      "epoch": 6.255404806956802,
-      "grad_norm": 0.012306607328355312,
-      "learning_rate": 0.00019998070206528166,
-      "loss": 46.0,
-      "step": 38843
-    },
-    {
-      "epoch": 6.255565844035589,
-      "grad_norm": 0.002566410694271326,
-      "learning_rate": 0.00019998070107140722,
-      "loss": 46.0,
-      "step": 38844
-    },
-    {
-      "epoch": 6.255726881114376,
-      "grad_norm": 0.0021592844277620316,
-      "learning_rate": 0.00019998070007750722,
-      "loss": 46.0,
-      "step": 38845
-    },
-    {
-      "epoch": 6.255887918193164,
-      "grad_norm": 0.0027225178200751543,
-      "learning_rate": 0.00019998069908358165,
-      "loss": 46.0,
-      "step": 38846
-    },
-    {
-      "epoch": 6.256048955271951,
-      "grad_norm": 0.0011230995878577232,
-      "learning_rate": 0.00019998069808963047,
-      "loss": 46.0,
-      "step": 38847
-    },
-    {
-      "epoch": 6.256209992350739,
-      "grad_norm": 0.002933935495093465,
-      "learning_rate": 0.0001999806970956537,
-      "loss": 46.0,
-      "step": 38848
-    },
-    {
-      "epoch": 6.256371029429526,
-      "grad_norm": 0.0029875149484723806,
-      "learning_rate": 0.00019998069610165136,
-      "loss": 46.0,
-      "step": 38849
-    },
-    {
-      "epoch": 6.2565320665083135,
-      "grad_norm": 0.003300453070551157,
-      "learning_rate": 0.0001999806951076234,
-      "loss": 46.0,
-      "step": 38850
-    },
-    {
-      "epoch": 6.256693103587101,
-      "grad_norm": 0.003974775783717632,
-      "learning_rate": 0.00019998069411356986,
-      "loss": 46.0,
-      "step": 38851
-    },
-    {
-      "epoch": 6.256854140665888,
-      "grad_norm": 0.0083570946007967,
-      "learning_rate": 0.00019998069311949075,
-      "loss": 46.0,
-      "step": 38852
-    },
-    {
-      "epoch": 6.257015177744676,
-      "grad_norm": 0.0014552467036992311,
-      "learning_rate": 0.00019998069212538602,
-      "loss": 46.0,
-      "step": 38853
-    },
-    {
-      "epoch": 6.257176214823463,
-      "grad_norm": 0.002063014777377248,
-      "learning_rate": 0.0001999806911312557,
-      "loss": 46.0,
-      "step": 38854
-    },
-    {
-      "epoch": 6.257337251902251,
-      "grad_norm": 0.0020061584655195475,
-      "learning_rate": 0.0001999806901370998,
-      "loss": 46.0,
-      "step": 38855
-    },
-    {
-      "epoch": 6.257498288981038,
-      "grad_norm": 0.0025122964289039373,
-      "learning_rate": 0.00019998068914291832,
-      "loss": 46.0,
-      "step": 38856
-    },
-    {
-      "epoch": 6.257659326059826,
-      "grad_norm": 0.00746725732460618,
-      "learning_rate": 0.00019998068814871124,
-      "loss": 46.0,
-      "step": 38857
-    },
-    {
-      "epoch": 6.257820363138613,
-      "grad_norm": 0.009549075737595558,
-      "learning_rate": 0.00019998068715447857,
-      "loss": 46.0,
-      "step": 38858
-    },
-    {
-      "epoch": 6.2579814002174,
-      "grad_norm": 0.003968052566051483,
-      "learning_rate": 0.00019998068616022032,
-      "loss": 46.0,
-      "step": 38859
-    },
-    {
-      "epoch": 6.258142437296187,
-      "grad_norm": 0.006248748395591974,
-      "learning_rate": 0.00019998068516593649,
-      "loss": 46.0,
-      "step": 38860
-    },
-    {
-      "epoch": 6.258303474374975,
-      "grad_norm": 0.003635324537754059,
-      "learning_rate": 0.00019998068417162706,
-      "loss": 46.0,
-      "step": 38861
-    },
-    {
-      "epoch": 6.258464511453762,
-      "grad_norm": 0.0016183869447559118,
-      "learning_rate": 0.000199980683177292,
-      "loss": 46.0,
-      "step": 38862
-    },
-    {
-      "epoch": 6.2586255485325495,
-      "grad_norm": 0.0033688012044876814,
-      "learning_rate": 0.0001999806821829314,
-      "loss": 46.0,
-      "step": 38863
-    },
-    {
-      "epoch": 6.258786585611337,
-      "grad_norm": 0.003631132422015071,
-      "learning_rate": 0.00019998068118854518,
-      "loss": 46.0,
-      "step": 38864
-    },
-    {
-      "epoch": 6.258947622690124,
-      "grad_norm": 0.003034138586372137,
-      "learning_rate": 0.00019998068019413338,
-      "loss": 46.0,
-      "step": 38865
-    },
-    {
-      "epoch": 6.259108659768912,
-      "grad_norm": 0.002651434624567628,
-      "learning_rate": 0.00019998067919969602,
-      "loss": 46.0,
-      "step": 38866
-    },
-    {
-      "epoch": 6.259269696847699,
-      "grad_norm": 0.014678088948130608,
-      "learning_rate": 0.00019998067820523304,
-      "loss": 46.0,
-      "step": 38867
-    },
-    {
-      "epoch": 6.259430733926487,
-      "grad_norm": 0.008156015537679195,
-      "learning_rate": 0.00019998067721074448,
-      "loss": 46.0,
-      "step": 38868
-    },
-    {
-      "epoch": 6.259591771005274,
-      "grad_norm": 0.005732891149818897,
-      "learning_rate": 0.0001999806762162303,
-      "loss": 46.0,
-      "step": 38869
-    },
-    {
-      "epoch": 6.259752808084062,
-      "grad_norm": 0.0012494615511968732,
-      "learning_rate": 0.00019998067522169056,
-      "loss": 46.0,
-      "step": 38870
-    },
-    {
-      "epoch": 6.259913845162849,
-      "grad_norm": 0.002612861804664135,
-      "learning_rate": 0.00019998067422712521,
-      "loss": 46.0,
-      "step": 38871
-    },
-    {
-      "epoch": 6.2600748822416366,
-      "grad_norm": 0.00813622958958149,
-      "learning_rate": 0.0001999806732325343,
-      "loss": 46.0,
-      "step": 38872
-    },
-    {
-      "epoch": 6.260235919320423,
-      "grad_norm": 0.016974350437521935,
-      "learning_rate": 0.00019998067223791778,
-      "loss": 46.0,
-      "step": 38873
-    },
-    {
-      "epoch": 6.260396956399211,
-      "grad_norm": 0.0010822816984727979,
-      "learning_rate": 0.0001999806712432757,
-      "loss": 46.0,
-      "step": 38874
-    },
-    {
-      "epoch": 6.260557993477998,
-      "grad_norm": 0.003296394133940339,
-      "learning_rate": 0.000199980670248608,
-      "loss": 46.0,
-      "step": 38875
-    },
-    {
-      "epoch": 6.2607190305567855,
-      "grad_norm": 0.008714224211871624,
-      "learning_rate": 0.0001999806692539147,
-      "loss": 46.0,
-      "step": 38876
-    },
-    {
-      "epoch": 6.260880067635573,
-      "grad_norm": 0.006531689781695604,
-      "learning_rate": 0.0001999806682591958,
-      "loss": 46.0,
-      "step": 38877
-    },
-    {
-      "epoch": 6.26104110471436,
-      "grad_norm": 0.008106173016130924,
-      "learning_rate": 0.00019998066726445134,
-      "loss": 46.0,
-      "step": 38878
-    },
-    {
-      "epoch": 6.261202141793148,
-      "grad_norm": 0.0029440741054713726,
-      "learning_rate": 0.0001999806662696813,
-      "loss": 46.0,
-      "step": 38879
-    },
-    {
-      "epoch": 6.261363178871935,
-      "grad_norm": 0.004526916425675154,
-      "learning_rate": 0.00019998066527488563,
-      "loss": 46.0,
-      "step": 38880
-    },
-    {
-      "epoch": 6.261524215950723,
-      "grad_norm": 0.006189214531332254,
-      "learning_rate": 0.0001999806642800644,
-      "loss": 46.0,
-      "step": 38881
-    },
-    {
-      "epoch": 6.26168525302951,
-      "grad_norm": 0.0036093282978981733,
-      "learning_rate": 0.0001999806632852176,
-      "loss": 46.0,
-      "step": 38882
-    },
-    {
-      "epoch": 6.261846290108298,
-      "grad_norm": 0.0014533509965986013,
-      "learning_rate": 0.00019998066229034517,
-      "loss": 46.0,
-      "step": 38883
-    },
-    {
-      "epoch": 6.262007327187085,
-      "grad_norm": 0.003021940356120467,
-      "learning_rate": 0.00019998066129544716,
-      "loss": 46.0,
-      "step": 38884
-    },
-    {
-      "epoch": 6.2621683642658725,
-      "grad_norm": 0.003992132376879454,
-      "learning_rate": 0.0001999806603005236,
-      "loss": 46.0,
-      "step": 38885
-    },
-    {
-      "epoch": 6.26232940134466,
-      "grad_norm": 0.006618468556553125,
-      "learning_rate": 0.0001999806593055744,
-      "loss": 46.0,
-      "step": 38886
-    },
-    {
-      "epoch": 6.2624904384234465,
-      "grad_norm": 0.013644207268953323,
-      "learning_rate": 0.00019998065831059963,
-      "loss": 46.0,
-      "step": 38887
-    },
-    {
-      "epoch": 6.262651475502234,
-      "grad_norm": 0.005208782851696014,
-      "learning_rate": 0.00019998065731559924,
-      "loss": 46.0,
-      "step": 38888
-    },
-    {
-      "epoch": 6.2628125125810215,
-      "grad_norm": 0.018910543993115425,
-      "learning_rate": 0.0001999806563205733,
-      "loss": 46.0,
-      "step": 38889
-    },
-    {
-      "epoch": 6.262973549659809,
-      "grad_norm": 0.008549229241907597,
-      "learning_rate": 0.00019998065532552176,
-      "loss": 46.0,
-      "step": 38890
-    },
-    {
-      "epoch": 6.263134586738596,
-      "grad_norm": 0.01707020401954651,
-      "learning_rate": 0.0001999806543304446,
-      "loss": 46.0,
-      "step": 38891
-    },
-    {
-      "epoch": 6.263295623817384,
-      "grad_norm": 0.005410929676145315,
-      "learning_rate": 0.00019998065333534187,
-      "loss": 46.0,
-      "step": 38892
-    },
-    {
-      "epoch": 6.263456660896171,
-      "grad_norm": 0.002450622385367751,
-      "learning_rate": 0.0001999806523402136,
-      "loss": 46.0,
-      "step": 38893
-    },
-    {
-      "epoch": 6.263617697974959,
-      "grad_norm": 0.01842244155704975,
-      "learning_rate": 0.00019998065134505967,
-      "loss": 46.0,
-      "step": 38894
-    },
-    {
-      "epoch": 6.263778735053746,
-      "grad_norm": 0.01023776363581419,
-      "learning_rate": 0.00019998065034988017,
-      "loss": 46.0,
-      "step": 38895
-    },
-    {
-      "epoch": 6.263939772132534,
-      "grad_norm": 0.006754412315785885,
-      "learning_rate": 0.0001999806493546751,
-      "loss": 46.0,
-      "step": 38896
-    },
-    {
-      "epoch": 6.264100809211321,
-      "grad_norm": 0.00219578854739666,
-      "learning_rate": 0.0001999806483594444,
-      "loss": 46.0,
-      "step": 38897
-    },
-    {
-      "epoch": 6.2642618462901085,
-      "grad_norm": 0.0015779847744852304,
-      "learning_rate": 0.00019998064736418815,
-      "loss": 46.0,
-      "step": 38898
-    },
-    {
-      "epoch": 6.264422883368896,
-      "grad_norm": 0.006254156120121479,
-      "learning_rate": 0.0001999806463689063,
-      "loss": 46.0,
-      "step": 38899
-    },
-    {
-      "epoch": 6.264583920447683,
-      "grad_norm": 0.004142298363149166,
-      "learning_rate": 0.00019998064537359887,
-      "loss": 46.0,
-      "step": 38900
-    },
-    {
-      "epoch": 6.264744957526471,
-      "grad_norm": 0.004489119164645672,
-      "learning_rate": 0.00019998064437826583,
-      "loss": 46.0,
-      "step": 38901
-    },
-    {
-      "epoch": 6.264905994605257,
-      "grad_norm": 0.0037455197889357805,
-      "learning_rate": 0.00019998064338290722,
-      "loss": 46.0,
-      "step": 38902
-    },
-    {
-      "epoch": 6.265067031684045,
-      "grad_norm": 0.0023001465015113354,
-      "learning_rate": 0.000199980642387523,
-      "loss": 46.0,
-      "step": 38903
-    },
-    {
-      "epoch": 6.265228068762832,
-      "grad_norm": 0.006872283760458231,
-      "learning_rate": 0.0001999806413921132,
-      "loss": 46.0,
-      "step": 38904
-    },
-    {
-      "epoch": 6.26538910584162,
-      "grad_norm": 0.0023588535841554403,
-      "learning_rate": 0.0001999806403966778,
-      "loss": 46.0,
-      "step": 38905
-    },
-    {
-      "epoch": 6.265550142920407,
-      "grad_norm": 0.007021674886345863,
-      "learning_rate": 0.0001999806394012168,
-      "loss": 46.0,
-      "step": 38906
-    },
-    {
-      "epoch": 6.265711179999195,
-      "grad_norm": 0.0018633791478350759,
-      "learning_rate": 0.00019998063840573024,
-      "loss": 46.0,
-      "step": 38907
-    },
-    {
-      "epoch": 6.265872217077982,
-      "grad_norm": 0.006334179081022739,
-      "learning_rate": 0.00019998063741021808,
-      "loss": 46.0,
-      "step": 38908
-    },
-    {
-      "epoch": 6.26603325415677,
-      "grad_norm": 0.0021135439164936543,
-      "learning_rate": 0.00019998063641468033,
-      "loss": 46.0,
-      "step": 38909
-    },
-    {
-      "epoch": 6.266194291235557,
-      "grad_norm": 0.016040639951825142,
-      "learning_rate": 0.00019998063541911697,
-      "loss": 46.0,
-      "step": 38910
-    },
-    {
-      "epoch": 6.2663553283143445,
-      "grad_norm": 0.018864290788769722,
-      "learning_rate": 0.00019998063442352803,
-      "loss": 46.0,
-      "step": 38911
-    },
-    {
-      "epoch": 6.266516365393132,
-      "grad_norm": 0.0028109773993492126,
-      "learning_rate": 0.00019998063342791352,
-      "loss": 46.0,
-      "step": 38912
-    },
-    {
-      "epoch": 6.266677402471919,
-      "grad_norm": 0.00144783069845289,
-      "learning_rate": 0.0001999806324322734,
-      "loss": 46.0,
-      "step": 38913
-    },
-    {
-      "epoch": 6.266838439550707,
-      "grad_norm": 0.0012966226786375046,
-      "learning_rate": 0.00019998063143660772,
-      "loss": 46.0,
-      "step": 38914
-    },
-    {
-      "epoch": 6.266999476629494,
-      "grad_norm": 0.006295338273048401,
-      "learning_rate": 0.00019998063044091642,
-      "loss": 46.0,
-      "step": 38915
-    },
-    {
-      "epoch": 6.267160513708282,
-      "grad_norm": 0.002131564309820533,
-      "learning_rate": 0.00019998062944519954,
-      "loss": 46.0,
-      "step": 38916
-    },
-    {
-      "epoch": 6.267321550787068,
-      "grad_norm": 0.0032865824177861214,
-      "learning_rate": 0.00019998062844945707,
-      "loss": 46.0,
-      "step": 38917
-    },
-    {
-      "epoch": 6.267482587865856,
-      "grad_norm": 0.008777833543717861,
-      "learning_rate": 0.000199980627453689,
-      "loss": 46.0,
-      "step": 38918
-    },
-    {
-      "epoch": 6.267643624944643,
-      "grad_norm": 0.006253014784306288,
-      "learning_rate": 0.00019998062645789534,
-      "loss": 46.0,
-      "step": 38919
-    },
-    {
-      "epoch": 6.267804662023431,
-      "grad_norm": 0.0031594850588589907,
-      "learning_rate": 0.00019998062546207609,
-      "loss": 46.0,
-      "step": 38920
-    },
-    {
-      "epoch": 6.267965699102218,
-      "grad_norm": 0.002162312623113394,
-      "learning_rate": 0.00019998062446623127,
-      "loss": 46.0,
-      "step": 38921
-    },
-    {
-      "epoch": 6.268126736181006,
-      "grad_norm": 0.001528203021734953,
-      "learning_rate": 0.00019998062347036086,
-      "loss": 46.0,
-      "step": 38922
-    },
-    {
-      "epoch": 6.268287773259793,
-      "grad_norm": 0.004104269668459892,
-      "learning_rate": 0.00019998062247446484,
-      "loss": 46.0,
-      "step": 38923
-    },
-    {
-      "epoch": 6.2684488103385805,
-      "grad_norm": 0.013286665081977844,
-      "learning_rate": 0.00019998062147854324,
-      "loss": 46.0,
-      "step": 38924
-    },
-    {
-      "epoch": 6.268609847417368,
-      "grad_norm": 0.0030473757069557905,
-      "learning_rate": 0.00019998062048259604,
-      "loss": 46.0,
-      "step": 38925
-    },
-    {
-      "epoch": 6.268770884496155,
-      "grad_norm": 0.0043985131196677685,
-      "learning_rate": 0.00019998061948662326,
-      "loss": 46.0,
-      "step": 38926
-    },
-    {
-      "epoch": 6.268931921574943,
-      "grad_norm": 0.0037587396800518036,
-      "learning_rate": 0.0001999806184906249,
-      "loss": 46.0,
-      "step": 38927
-    },
-    {
-      "epoch": 6.26909295865373,
-      "grad_norm": 0.003596738213673234,
-      "learning_rate": 0.00019998061749460094,
-      "loss": 46.0,
-      "step": 38928
-    },
-    {
-      "epoch": 6.269253995732518,
-      "grad_norm": 0.006315917707979679,
-      "learning_rate": 0.00019998061649855137,
-      "loss": 46.0,
-      "step": 38929
-    },
-    {
-      "epoch": 6.269415032811305,
-      "grad_norm": 0.002405857667326927,
-      "learning_rate": 0.00019998061550247624,
-      "loss": 46.0,
-      "step": 38930
-    },
-    {
-      "epoch": 6.269576069890093,
-      "grad_norm": 0.015180880203843117,
-      "learning_rate": 0.00019998061450637552,
-      "loss": 46.0,
-      "step": 38931
-    },
-    {
-      "epoch": 6.269737106968879,
-      "grad_norm": 0.0017666129861027002,
-      "learning_rate": 0.0001999806135102492,
-      "loss": 46.0,
-      "step": 38932
-    },
-    {
-      "epoch": 6.269898144047667,
-      "grad_norm": 0.0018145734211429954,
-      "learning_rate": 0.00019998061251409727,
-      "loss": 46.0,
-      "step": 38933
-    },
-    {
-      "epoch": 6.270059181126454,
-      "grad_norm": 0.0033974708057940006,
-      "learning_rate": 0.0001999806115179198,
-      "loss": 46.0,
-      "step": 38934
-    },
-    {
-      "epoch": 6.270220218205242,
-      "grad_norm": 0.009783491492271423,
-      "learning_rate": 0.00019998061052171667,
-      "loss": 46.0,
-      "step": 38935
-    },
-    {
-      "epoch": 6.270381255284029,
-      "grad_norm": 0.0019134997855871916,
-      "learning_rate": 0.00019998060952548802,
-      "loss": 46.0,
-      "step": 38936
-    },
-    {
-      "epoch": 6.2705422923628165,
-      "grad_norm": 0.0031892128754407167,
-      "learning_rate": 0.00019998060852923372,
-      "loss": 46.0,
-      "step": 38937
-    },
-    {
-      "epoch": 6.270703329441604,
-      "grad_norm": 0.004307840019464493,
-      "learning_rate": 0.00019998060753295387,
-      "loss": 46.0,
-      "step": 38938
-    },
-    {
-      "epoch": 6.270864366520391,
-      "grad_norm": 0.005807558540254831,
-      "learning_rate": 0.00019998060653664843,
-      "loss": 46.0,
-      "step": 38939
-    },
-    {
-      "epoch": 6.271025403599179,
-      "grad_norm": 0.022980840876698494,
-      "learning_rate": 0.0001999806055403174,
-      "loss": 46.0,
-      "step": 38940
-    },
-    {
-      "epoch": 6.271186440677966,
-      "grad_norm": 0.001879759831354022,
-      "learning_rate": 0.00019998060454396078,
-      "loss": 46.0,
-      "step": 38941
-    },
-    {
-      "epoch": 6.271347477756754,
-      "grad_norm": 0.009212806820869446,
-      "learning_rate": 0.00019998060354757855,
-      "loss": 46.0,
-      "step": 38942
-    },
-    {
-      "epoch": 6.271508514835541,
-      "grad_norm": 0.005267132073640823,
-      "learning_rate": 0.00019998060255117074,
-      "loss": 46.0,
-      "step": 38943
-    },
-    {
-      "epoch": 6.271669551914329,
-      "grad_norm": 0.00745200552046299,
-      "learning_rate": 0.00019998060155473733,
-      "loss": 46.0,
-      "step": 38944
-    },
-    {
-      "epoch": 6.271830588993116,
-      "grad_norm": 0.00822604913264513,
-      "learning_rate": 0.00019998060055827834,
-      "loss": 46.0,
-      "step": 38945
-    },
-    {
-      "epoch": 6.2719916260719035,
-      "grad_norm": 0.003954382613301277,
-      "learning_rate": 0.00019998059956179376,
-      "loss": 46.0,
-      "step": 38946
-    },
-    {
-      "epoch": 6.27215266315069,
-      "grad_norm": 0.0016620233654975891,
-      "learning_rate": 0.00019998059856528357,
-      "loss": 46.0,
-      "step": 38947
-    },
-    {
-      "epoch": 6.2723137002294775,
-      "grad_norm": 0.009317819029092789,
-      "learning_rate": 0.00019998059756874784,
-      "loss": 46.0,
-      "step": 38948
-    },
-    {
-      "epoch": 6.272474737308265,
-      "grad_norm": 0.009431381709873676,
-      "learning_rate": 0.00019998059657218647,
-      "loss": 46.0,
-      "step": 38949
-    },
-    {
-      "epoch": 6.272635774387052,
-      "grad_norm": 0.007788951508700848,
-      "learning_rate": 0.00019998059557559955,
-      "loss": 46.0,
-      "step": 38950
-    },
-    {
-      "epoch": 6.27279681146584,
-      "grad_norm": 0.0017576408572494984,
-      "learning_rate": 0.000199980594578987,
-      "loss": 46.0,
-      "step": 38951
-    },
-    {
-      "epoch": 6.272957848544627,
-      "grad_norm": 0.0009234852041117847,
-      "learning_rate": 0.00019998059358234887,
-      "loss": 46.0,
-      "step": 38952
-    },
-    {
-      "epoch": 6.273118885623415,
-      "grad_norm": 0.0018828288884833455,
-      "learning_rate": 0.00019998059258568519,
-      "loss": 46.0,
-      "step": 38953
-    },
-    {
-      "epoch": 6.273279922702202,
-      "grad_norm": 0.005209984723478556,
-      "learning_rate": 0.00019998059158899588,
-      "loss": 46.0,
-      "step": 38954
-    },
-    {
-      "epoch": 6.27344095978099,
-      "grad_norm": 0.010661369189620018,
-      "learning_rate": 0.000199980590592281,
-      "loss": 46.0,
-      "step": 38955
-    },
-    {
-      "epoch": 6.273601996859777,
-      "grad_norm": 0.009893474169075489,
-      "learning_rate": 0.00019998058959554051,
-      "loss": 46.0,
-      "step": 38956
-    },
-    {
-      "epoch": 6.273763033938565,
-      "grad_norm": 0.002609674585983157,
-      "learning_rate": 0.00019998058859877445,
-      "loss": 46.0,
-      "step": 38957
-    },
-    {
-      "epoch": 6.273924071017352,
-      "grad_norm": 0.008573015220463276,
-      "learning_rate": 0.0001999805876019828,
-      "loss": 46.0,
-      "step": 38958
-    },
-    {
-      "epoch": 6.2740851080961395,
-      "grad_norm": 0.002283828565850854,
-      "learning_rate": 0.00019998058660516553,
-      "loss": 46.0,
-      "step": 38959
-    },
-    {
-      "epoch": 6.274246145174926,
-      "grad_norm": 0.007836409844458103,
-      "learning_rate": 0.00019998058560832268,
-      "loss": 46.0,
-      "step": 38960
-    },
-    {
-      "epoch": 6.2744071822537135,
-      "grad_norm": 0.001190313370898366,
-      "learning_rate": 0.0001999805846114543,
-      "loss": 46.0,
-      "step": 38961
-    },
-    {
-      "epoch": 6.274568219332501,
-      "grad_norm": 0.003508426481857896,
-      "learning_rate": 0.00019998058361456024,
-      "loss": 46.0,
-      "step": 38962
-    },
-    {
-      "epoch": 6.274729256411288,
-      "grad_norm": 0.0013165523996576667,
-      "learning_rate": 0.00019998058261764062,
-      "loss": 46.0,
-      "step": 38963
-    },
-    {
-      "epoch": 6.274890293490076,
-      "grad_norm": 0.003525273874402046,
-      "learning_rate": 0.00019998058162069545,
-      "loss": 46.0,
-      "step": 38964
-    },
-    {
-      "epoch": 6.275051330568863,
-      "grad_norm": 0.012274498119950294,
-      "learning_rate": 0.00019998058062372466,
-      "loss": 46.0,
-      "step": 38965
-    },
-    {
-      "epoch": 6.275212367647651,
-      "grad_norm": 0.001048225094564259,
-      "learning_rate": 0.00019998057962672828,
-      "loss": 46.0,
-      "step": 38966
-    },
-    {
-      "epoch": 6.275373404726438,
-      "grad_norm": 0.0024400344118475914,
-      "learning_rate": 0.00019998057862970632,
-      "loss": 46.0,
-      "step": 38967
-    },
-    {
-      "epoch": 6.275534441805226,
-      "grad_norm": 0.005804820451885462,
-      "learning_rate": 0.00019998057763265877,
-      "loss": 46.0,
-      "step": 38968
-    },
-    {
-      "epoch": 6.275695478884013,
-      "grad_norm": 0.00299848266877234,
-      "learning_rate": 0.0001999805766355856,
-      "loss": 46.0,
-      "step": 38969
-    },
-    {
-      "epoch": 6.275856515962801,
-      "grad_norm": 0.010823872871696949,
-      "learning_rate": 0.00019998057563848688,
-      "loss": 46.0,
-      "step": 38970
-    },
-    {
-      "epoch": 6.276017553041588,
-      "grad_norm": 0.00453021377325058,
-      "learning_rate": 0.00019998057464136256,
-      "loss": 46.0,
-      "step": 38971
-    },
-    {
-      "epoch": 6.2761785901203755,
-      "grad_norm": 0.0029733553528785706,
-      "learning_rate": 0.00019998057364421264,
-      "loss": 46.0,
-      "step": 38972
-    },
-    {
-      "epoch": 6.276339627199163,
-      "grad_norm": 0.0019039901671931148,
-      "learning_rate": 0.00019998057264703712,
-      "loss": 46.0,
-      "step": 38973
-    },
-    {
-      "epoch": 6.27650066427795,
-      "grad_norm": 0.0020593050867319107,
-      "learning_rate": 0.00019998057164983602,
-      "loss": 46.0,
-      "step": 38974
-    },
-    {
-      "epoch": 6.276661701356737,
-      "grad_norm": 0.0016014946158975363,
-      "learning_rate": 0.00019998057065260936,
-      "loss": 46.0,
-      "step": 38975
-    },
-    {
-      "epoch": 6.276822738435524,
-      "grad_norm": 0.010088545270264149,
-      "learning_rate": 0.00019998056965535709,
-      "loss": 46.0,
-      "step": 38976
-    },
-    {
-      "epoch": 6.276983775514312,
-      "grad_norm": 0.004582440946251154,
-      "learning_rate": 0.00019998056865807922,
-      "loss": 46.0,
-      "step": 38977
-    },
-    {
-      "epoch": 6.277144812593099,
-      "grad_norm": 0.0009791270131245255,
-      "learning_rate": 0.00019998056766077572,
-      "loss": 46.0,
-      "step": 38978
-    },
-    {
-      "epoch": 6.277305849671887,
-      "grad_norm": 0.018884582445025444,
-      "learning_rate": 0.0001999805666634467,
-      "loss": 46.0,
-      "step": 38979
-    },
-    {
-      "epoch": 6.277466886750674,
-      "grad_norm": 0.008716803975403309,
-      "learning_rate": 0.00019998056566609206,
-      "loss": 46.0,
-      "step": 38980
-    },
-    {
-      "epoch": 6.277627923829462,
-      "grad_norm": 0.007112538442015648,
-      "learning_rate": 0.00019998056466871185,
-      "loss": 46.0,
-      "step": 38981
-    },
-    {
-      "epoch": 6.277788960908249,
-      "grad_norm": 0.010333537124097347,
-      "learning_rate": 0.00019998056367130602,
-      "loss": 46.0,
-      "step": 38982
-    },
-    {
-      "epoch": 6.277949997987037,
-      "grad_norm": 0.003959720488637686,
-      "learning_rate": 0.0001999805626738746,
-      "loss": 46.0,
-      "step": 38983
-    },
-    {
-      "epoch": 6.278111035065824,
-      "grad_norm": 0.0013928828993812203,
-      "learning_rate": 0.0001999805616764176,
-      "loss": 46.0,
-      "step": 38984
-    },
-    {
-      "epoch": 6.2782720721446115,
-      "grad_norm": 0.001907423953525722,
-      "learning_rate": 0.00019998056067893502,
-      "loss": 46.0,
-      "step": 38985
-    },
-    {
-      "epoch": 6.278433109223399,
-      "grad_norm": 0.0045247795060276985,
-      "learning_rate": 0.00019998055968142685,
-      "loss": 46.0,
-      "step": 38986
-    },
-    {
-      "epoch": 6.278594146302186,
-      "grad_norm": 0.003401419846341014,
-      "learning_rate": 0.00019998055868389309,
-      "loss": 46.0,
-      "step": 38987
-    },
-    {
-      "epoch": 6.278755183380974,
-      "grad_norm": 0.007197286002337933,
-      "learning_rate": 0.00019998055768633374,
-      "loss": 46.0,
-      "step": 38988
-    },
-    {
-      "epoch": 6.278916220459761,
-      "grad_norm": 0.013942079618573189,
-      "learning_rate": 0.0001999805566887488,
-      "loss": 46.0,
-      "step": 38989
-    },
-    {
-      "epoch": 6.279077257538548,
-      "grad_norm": 0.005511273164302111,
-      "learning_rate": 0.00019998055569113828,
-      "loss": 46.0,
-      "step": 38990
-    },
-    {
-      "epoch": 6.279238294617335,
-      "grad_norm": 0.002227319171652198,
-      "learning_rate": 0.00019998055469350212,
-      "loss": 46.0,
-      "step": 38991
-    },
-    {
-      "epoch": 6.279399331696123,
-      "grad_norm": 0.0025035315193235874,
-      "learning_rate": 0.00019998055369584042,
-      "loss": 46.0,
-      "step": 38992
-    },
-    {
-      "epoch": 6.27956036877491,
-      "grad_norm": 0.0009496422135271132,
-      "learning_rate": 0.00019998055269815313,
-      "loss": 46.0,
-      "step": 38993
-    },
-    {
-      "epoch": 6.279721405853698,
-      "grad_norm": 0.004296082071959972,
-      "learning_rate": 0.0001999805517004402,
-      "loss": 46.0,
-      "step": 38994
-    },
-    {
-      "epoch": 6.279882442932485,
-      "grad_norm": 0.0049510616809129715,
-      "learning_rate": 0.00019998055070270172,
-      "loss": 46.0,
-      "step": 38995
-    },
-    {
-      "epoch": 6.2800434800112725,
-      "grad_norm": 0.004195664543658495,
-      "learning_rate": 0.00019998054970493768,
-      "loss": 46.0,
-      "step": 38996
-    },
-    {
-      "epoch": 6.28020451709006,
-      "grad_norm": 0.003316165180876851,
-      "learning_rate": 0.000199980548707148,
-      "loss": 46.0,
-      "step": 38997
-    },
-    {
-      "epoch": 6.2803655541688475,
-      "grad_norm": 0.005499253049492836,
-      "learning_rate": 0.00019998054770933274,
-      "loss": 46.0,
-      "step": 38998
-    },
-    {
-      "epoch": 6.280526591247635,
-      "grad_norm": 0.025381211191415787,
-      "learning_rate": 0.00019998054671149188,
-      "loss": 46.0,
-      "step": 38999
-    },
-    {
-      "epoch": 6.280687628326422,
-      "grad_norm": 0.010825708508491516,
-      "learning_rate": 0.00019998054571362546,
-      "loss": 46.0,
-      "step": 39000
-    },
-    {
-      "epoch": 6.28084866540521,
-      "grad_norm": 0.0010732181835919619,
-      "learning_rate": 0.00019998054471573345,
-      "loss": 46.0,
-      "step": 39001
-    },
-    {
-      "epoch": 6.281009702483997,
-      "grad_norm": 0.005935711786150932,
-      "learning_rate": 0.00019998054371781583,
-      "loss": 46.0,
-      "step": 39002
-    },
-    {
-      "epoch": 6.281170739562785,
-      "grad_norm": 0.0020413235761225224,
-      "learning_rate": 0.00019998054271987262,
-      "loss": 46.0,
-      "step": 39003
-    },
-    {
-      "epoch": 6.281331776641572,
-      "grad_norm": 0.009944637306034565,
-      "learning_rate": 0.00019998054172190385,
-      "loss": 46.0,
-      "step": 39004
-    },
-    {
-      "epoch": 6.281492813720359,
-      "grad_norm": 0.008419917896389961,
-      "learning_rate": 0.00019998054072390943,
-      "loss": 46.0,
-      "step": 39005
-    },
-    {
-      "epoch": 6.281653850799146,
-      "grad_norm": 0.007590579334646463,
-      "learning_rate": 0.00019998053972588946,
-      "loss": 46.0,
-      "step": 39006
-    },
-    {
-      "epoch": 6.281814887877934,
-      "grad_norm": 0.0017160444986075163,
-      "learning_rate": 0.0001999805387278439,
-      "loss": 46.0,
-      "step": 39007
-    },
-    {
-      "epoch": 6.281975924956721,
-      "grad_norm": 0.01751594804227352,
-      "learning_rate": 0.00019998053772977276,
-      "loss": 46.0,
-      "step": 39008
-    },
-    {
-      "epoch": 6.2821369620355085,
-      "grad_norm": 0.017763033509254456,
-      "learning_rate": 0.000199980536731676,
-      "loss": 46.0,
-      "step": 39009
-    },
-    {
-      "epoch": 6.282297999114296,
-      "grad_norm": 0.0032215043902397156,
-      "learning_rate": 0.00019998053573355368,
-      "loss": 46.0,
-      "step": 39010
-    },
-    {
-      "epoch": 6.282459036193083,
-      "grad_norm": 0.006471071392297745,
-      "learning_rate": 0.00019998053473540574,
-      "loss": 46.0,
-      "step": 39011
-    },
-    {
-      "epoch": 6.282620073271871,
-      "grad_norm": 0.008333157747983932,
-      "learning_rate": 0.00019998053373723225,
-      "loss": 46.0,
-      "step": 39012
-    },
-    {
-      "epoch": 6.282781110350658,
-      "grad_norm": 0.0006876584375277162,
-      "learning_rate": 0.00019998053273903314,
-      "loss": 46.0,
-      "step": 39013
-    },
-    {
-      "epoch": 6.282942147429446,
-      "grad_norm": 0.007667304016649723,
-      "learning_rate": 0.00019998053174080844,
-      "loss": 46.0,
-      "step": 39014
-    },
-    {
-      "epoch": 6.283103184508233,
-      "grad_norm": 0.006231206003576517,
-      "learning_rate": 0.00019998053074255819,
-      "loss": 46.0,
-      "step": 39015
-    },
-    {
-      "epoch": 6.283264221587021,
-      "grad_norm": 0.0014082818524912,
-      "learning_rate": 0.0001999805297442823,
-      "loss": 46.0,
-      "step": 39016
-    },
-    {
-      "epoch": 6.283425258665808,
-      "grad_norm": 0.005389024503529072,
-      "learning_rate": 0.00019998052874598083,
-      "loss": 46.0,
-      "step": 39017
-    },
-    {
-      "epoch": 6.283586295744596,
-      "grad_norm": 0.01694563962519169,
-      "learning_rate": 0.00019998052774765376,
-      "loss": 46.0,
-      "step": 39018
-    },
-    {
-      "epoch": 6.283747332823383,
-      "grad_norm": 0.006275527644902468,
-      "learning_rate": 0.00019998052674930113,
-      "loss": 46.0,
-      "step": 39019
-    },
-    {
-      "epoch": 6.28390836990217,
-      "grad_norm": 0.002231972524896264,
-      "learning_rate": 0.0001999805257509229,
-      "loss": 46.0,
-      "step": 39020
-    },
-    {
-      "epoch": 6.284069406980957,
-      "grad_norm": 0.006555341184139252,
-      "learning_rate": 0.0001999805247525191,
-      "loss": 46.0,
-      "step": 39021
-    },
-    {
-      "epoch": 6.2842304440597445,
-      "grad_norm": 0.010288205929100513,
-      "learning_rate": 0.00019998052375408968,
-      "loss": 46.0,
-      "step": 39022
-    },
-    {
-      "epoch": 6.284391481138532,
-      "grad_norm": 0.002767049940302968,
-      "learning_rate": 0.00019998052275563467,
-      "loss": 46.0,
-      "step": 39023
-    },
-    {
-      "epoch": 6.284552518217319,
-      "grad_norm": 0.0039021295960992575,
-      "learning_rate": 0.00019998052175715408,
-      "loss": 46.0,
-      "step": 39024
-    },
-    {
-      "epoch": 6.284713555296107,
-      "grad_norm": 0.0073213232681155205,
-      "learning_rate": 0.0001999805207586479,
-      "loss": 46.0,
-      "step": 39025
-    },
-    {
-      "epoch": 6.284874592374894,
-      "grad_norm": 0.002442700555548072,
-      "learning_rate": 0.00019998051976011612,
-      "loss": 46.0,
-      "step": 39026
-    },
-    {
-      "epoch": 6.285035629453682,
-      "grad_norm": 0.00956632848829031,
-      "learning_rate": 0.00019998051876155877,
-      "loss": 46.0,
-      "step": 39027
-    },
-    {
-      "epoch": 6.285196666532469,
-      "grad_norm": 0.0088145537301898,
-      "learning_rate": 0.00019998051776297582,
-      "loss": 46.0,
-      "step": 39028
-    },
-    {
-      "epoch": 6.285357703611257,
-      "grad_norm": 0.0022800287697464228,
-      "learning_rate": 0.00019998051676436727,
-      "loss": 46.0,
-      "step": 39029
-    },
-    {
-      "epoch": 6.285518740690044,
-      "grad_norm": 0.013208272866904736,
-      "learning_rate": 0.00019998051576573315,
-      "loss": 46.0,
-      "step": 39030
-    },
-    {
-      "epoch": 6.285679777768832,
-      "grad_norm": 0.012782255187630653,
-      "learning_rate": 0.00019998051476707342,
-      "loss": 46.0,
-      "step": 39031
-    },
-    {
-      "epoch": 6.285840814847619,
-      "grad_norm": 0.0017857920611277223,
-      "learning_rate": 0.00019998051376838812,
-      "loss": 46.0,
-      "step": 39032
-    },
-    {
-      "epoch": 6.286001851926406,
-      "grad_norm": 0.006807293277233839,
-      "learning_rate": 0.00019998051276967722,
-      "loss": 46.0,
-      "step": 39033
-    },
-    {
-      "epoch": 6.286162889005193,
-      "grad_norm": 0.003250164445489645,
-      "learning_rate": 0.00019998051177094072,
-      "loss": 46.0,
-      "step": 39034
-    },
-    {
-      "epoch": 6.2863239260839805,
-      "grad_norm": 0.0012515768175944686,
-      "learning_rate": 0.00019998051077217864,
-      "loss": 46.0,
-      "step": 39035
-    },
-    {
-      "epoch": 6.286484963162768,
-      "grad_norm": 0.0008437428041361272,
-      "learning_rate": 0.00019998050977339097,
-      "loss": 46.0,
-      "step": 39036
-    },
-    {
-      "epoch": 6.286646000241555,
-      "grad_norm": 0.002129937754943967,
-      "learning_rate": 0.00019998050877457772,
-      "loss": 46.0,
-      "step": 39037
-    },
-    {
-      "epoch": 6.286807037320343,
-      "grad_norm": 0.005391864571720362,
-      "learning_rate": 0.00019998050777573888,
-      "loss": 46.0,
-      "step": 39038
-    },
-    {
-      "epoch": 6.28696807439913,
-      "grad_norm": 0.0026512891054153442,
-      "learning_rate": 0.00019998050677687445,
-      "loss": 46.0,
-      "step": 39039
-    },
-    {
-      "epoch": 6.287129111477918,
-      "grad_norm": 0.006737624295055866,
-      "learning_rate": 0.0001999805057779844,
-      "loss": 46.0,
-      "step": 39040
-    },
-    {
-      "epoch": 6.287290148556705,
-      "grad_norm": 0.0065542832016944885,
-      "learning_rate": 0.00019998050477906877,
-      "loss": 46.0,
-      "step": 39041
-    },
-    {
-      "epoch": 6.287451185635493,
-      "grad_norm": 0.009030701592564583,
-      "learning_rate": 0.00019998050378012758,
-      "loss": 46.0,
-      "step": 39042
-    },
-    {
-      "epoch": 6.28761222271428,
-      "grad_norm": 0.015139184892177582,
-      "learning_rate": 0.00019998050278116077,
-      "loss": 46.0,
-      "step": 39043
-    },
-    {
-      "epoch": 6.287773259793068,
-      "grad_norm": 0.025006920099258423,
-      "learning_rate": 0.00019998050178216838,
-      "loss": 46.0,
-      "step": 39044
-    },
-    {
-      "epoch": 6.287934296871855,
-      "grad_norm": 0.0011401117080822587,
-      "learning_rate": 0.0001999805007831504,
-      "loss": 46.0,
-      "step": 39045
-    },
-    {
-      "epoch": 6.2880953339506425,
-      "grad_norm": 0.0029454021714627743,
-      "learning_rate": 0.00019998049978410683,
-      "loss": 46.0,
-      "step": 39046
-    },
-    {
-      "epoch": 6.28825637102943,
-      "grad_norm": 0.0033295098692178726,
-      "learning_rate": 0.00019998049878503768,
-      "loss": 46.0,
-      "step": 39047
-    },
-    {
-      "epoch": 6.2884174081082165,
-      "grad_norm": 0.0015547252260148525,
-      "learning_rate": 0.0001999804977859429,
-      "loss": 46.0,
-      "step": 39048
-    },
-    {
-      "epoch": 6.288578445187004,
-      "grad_norm": 0.02438298799097538,
-      "learning_rate": 0.00019998049678682258,
-      "loss": 46.0,
-      "step": 39049
-    },
-    {
-      "epoch": 6.288739482265791,
-      "grad_norm": 0.002501065144315362,
-      "learning_rate": 0.00019998049578767667,
-      "loss": 46.0,
-      "step": 39050
-    },
-    {
-      "epoch": 6.288900519344579,
-      "grad_norm": 0.0012736415956169367,
-      "learning_rate": 0.00019998049478850514,
-      "loss": 46.0,
-      "step": 39051
-    },
-    {
-      "epoch": 6.289061556423366,
-      "grad_norm": 0.009752868674695492,
-      "learning_rate": 0.00019998049378930802,
-      "loss": 46.0,
-      "step": 39052
-    },
-    {
-      "epoch": 6.289222593502154,
-      "grad_norm": 0.000991630950011313,
-      "learning_rate": 0.00019998049279008532,
-      "loss": 46.0,
-      "step": 39053
-    },
-    {
-      "epoch": 6.289383630580941,
-      "grad_norm": 0.005731829907745123,
-      "learning_rate": 0.00019998049179083705,
-      "loss": 46.0,
-      "step": 39054
-    },
-    {
-      "epoch": 6.289544667659729,
-      "grad_norm": 0.009796880185604095,
-      "learning_rate": 0.00019998049079156317,
-      "loss": 46.0,
-      "step": 39055
-    },
-    {
-      "epoch": 6.289705704738516,
-      "grad_norm": 0.0018510626396164298,
-      "learning_rate": 0.00019998048979226368,
-      "loss": 46.0,
-      "step": 39056
-    },
-    {
-      "epoch": 6.2898667418173035,
-      "grad_norm": 0.001044785720296204,
-      "learning_rate": 0.00019998048879293865,
-      "loss": 46.0,
-      "step": 39057
-    },
-    {
-      "epoch": 6.290027778896091,
-      "grad_norm": 0.004461322911083698,
-      "learning_rate": 0.000199980487793588,
-      "loss": 46.0,
-      "step": 39058
-    },
-    {
-      "epoch": 6.2901888159748784,
-      "grad_norm": 0.0017150011844933033,
-      "learning_rate": 0.00019998048679421176,
-      "loss": 46.0,
-      "step": 39059
-    },
-    {
-      "epoch": 6.290349853053666,
-      "grad_norm": 0.005361329764127731,
-      "learning_rate": 0.00019998048579480994,
-      "loss": 46.0,
-      "step": 39060
-    },
-    {
-      "epoch": 6.290510890132453,
-      "grad_norm": 0.013194623403251171,
-      "learning_rate": 0.00019998048479538251,
-      "loss": 46.0,
-      "step": 39061
-    },
-    {
-      "epoch": 6.290671927211241,
-      "grad_norm": 0.003736241487786174,
-      "learning_rate": 0.0001999804837959295,
-      "loss": 46.0,
-      "step": 39062
-    },
-    {
-      "epoch": 6.290832964290027,
-      "grad_norm": 0.006789695471525192,
-      "learning_rate": 0.00019998048279645092,
-      "loss": 46.0,
-      "step": 39063
-    },
-    {
-      "epoch": 6.290994001368815,
-      "grad_norm": 0.005020936485379934,
-      "learning_rate": 0.00019998048179694673,
-      "loss": 46.0,
-      "step": 39064
-    },
-    {
-      "epoch": 6.291155038447602,
-      "grad_norm": 0.0038886622060090303,
-      "learning_rate": 0.00019998048079741695,
-      "loss": 46.0,
-      "step": 39065
-    },
-    {
-      "epoch": 6.29131607552639,
-      "grad_norm": 0.002402149373665452,
-      "learning_rate": 0.0001999804797978616,
-      "loss": 46.0,
-      "step": 39066
-    },
-    {
-      "epoch": 6.291477112605177,
-      "grad_norm": 0.003210574621334672,
-      "learning_rate": 0.00019998047879828064,
-      "loss": 46.0,
-      "step": 39067
-    },
-    {
-      "epoch": 6.291638149683965,
-      "grad_norm": 0.00162588432431221,
-      "learning_rate": 0.00019998047779867407,
-      "loss": 46.0,
-      "step": 39068
-    },
-    {
-      "epoch": 6.291799186762752,
-      "grad_norm": 0.008654010482132435,
-      "learning_rate": 0.00019998047679904195,
-      "loss": 46.0,
-      "step": 39069
-    },
-    {
-      "epoch": 6.2919602238415395,
-      "grad_norm": 0.001888230093754828,
-      "learning_rate": 0.0001999804757993842,
-      "loss": 46.0,
-      "step": 39070
-    },
-    {
-      "epoch": 6.292121260920327,
-      "grad_norm": 0.007668881211429834,
-      "learning_rate": 0.0001999804747997009,
-      "loss": 46.0,
-      "step": 39071
-    },
-    {
-      "epoch": 6.292282297999114,
-      "grad_norm": 0.004537229426205158,
-      "learning_rate": 0.00019998047379999202,
-      "loss": 46.0,
-      "step": 39072
-    },
-    {
-      "epoch": 6.292443335077902,
-      "grad_norm": 0.020649900659918785,
-      "learning_rate": 0.0001999804728002575,
-      "loss": 46.0,
-      "step": 39073
-    },
-    {
-      "epoch": 6.292604372156689,
-      "grad_norm": 0.0029868492856621742,
-      "learning_rate": 0.00019998047180049743,
-      "loss": 46.0,
-      "step": 39074
-    },
-    {
-      "epoch": 6.292765409235477,
-      "grad_norm": 0.014783100225031376,
-      "learning_rate": 0.00019998047080071175,
-      "loss": 46.0,
-      "step": 39075
-    },
-    {
-      "epoch": 6.292926446314264,
-      "grad_norm": 0.0035305072087794542,
-      "learning_rate": 0.00019998046980090049,
-      "loss": 46.0,
-      "step": 39076
-    },
-    {
-      "epoch": 6.293087483393052,
-      "grad_norm": 0.009474482387304306,
-      "learning_rate": 0.00019998046880106364,
-      "loss": 46.0,
-      "step": 39077
-    },
-    {
-      "epoch": 6.293248520471838,
-      "grad_norm": 0.010321266017854214,
-      "learning_rate": 0.00019998046780120117,
-      "loss": 46.0,
-      "step": 39078
-    },
-    {
-      "epoch": 6.293409557550626,
-      "grad_norm": 0.006980300415307283,
-      "learning_rate": 0.00019998046680131315,
-      "loss": 46.0,
-      "step": 39079
-    },
-    {
-      "epoch": 6.293570594629413,
-      "grad_norm": 0.0057964324951171875,
-      "learning_rate": 0.0001999804658013995,
-      "loss": 46.0,
-      "step": 39080
-    },
-    {
-      "epoch": 6.293731631708201,
-      "grad_norm": 0.01205200981348753,
-      "learning_rate": 0.0001999804648014603,
-      "loss": 46.0,
-      "step": 39081
-    },
-    {
-      "epoch": 6.293892668786988,
-      "grad_norm": 0.00474933348596096,
-      "learning_rate": 0.0001999804638014955,
-      "loss": 46.0,
-      "step": 39082
-    },
-    {
-      "epoch": 6.2940537058657755,
-      "grad_norm": 0.004265707451850176,
-      "learning_rate": 0.00019998046280150512,
-      "loss": 46.0,
-      "step": 39083
-    },
-    {
-      "epoch": 6.294214742944563,
-      "grad_norm": 0.007605432998389006,
-      "learning_rate": 0.00019998046180148913,
-      "loss": 46.0,
-      "step": 39084
-    },
-    {
-      "epoch": 6.29437578002335,
-      "grad_norm": 0.007893330417573452,
-      "learning_rate": 0.00019998046080144756,
-      "loss": 46.0,
-      "step": 39085
-    },
-    {
-      "epoch": 6.294536817102138,
-      "grad_norm": 0.001544732484035194,
-      "learning_rate": 0.0001999804598013804,
-      "loss": 46.0,
-      "step": 39086
-    },
-    {
-      "epoch": 6.294697854180925,
-      "grad_norm": 0.017420900985598564,
-      "learning_rate": 0.00019998045880128764,
-      "loss": 46.0,
-      "step": 39087
-    },
-    {
-      "epoch": 6.294858891259713,
-      "grad_norm": 0.013849043287336826,
-      "learning_rate": 0.0001999804578011693,
-      "loss": 46.0,
-      "step": 39088
-    },
-    {
-      "epoch": 6.2950199283385,
-      "grad_norm": 0.013542650267481804,
-      "learning_rate": 0.00019998045680102538,
-      "loss": 46.0,
-      "step": 39089
-    },
-    {
-      "epoch": 6.295180965417288,
-      "grad_norm": 0.003099643625319004,
-      "learning_rate": 0.00019998045580085582,
-      "loss": 46.0,
-      "step": 39090
-    },
-    {
-      "epoch": 6.295342002496075,
-      "grad_norm": 0.003130297176539898,
-      "learning_rate": 0.00019998045480066072,
-      "loss": 46.0,
-      "step": 39091
-    },
-    {
-      "epoch": 6.295503039574863,
-      "grad_norm": 0.004167810548096895,
-      "learning_rate": 0.00019998045380044,
-      "loss": 46.0,
-      "step": 39092
-    },
-    {
-      "epoch": 6.295664076653649,
-      "grad_norm": 0.0021290292497724295,
-      "learning_rate": 0.00019998045280019374,
-      "loss": 46.0,
-      "step": 39093
-    },
-    {
-      "epoch": 6.295825113732437,
-      "grad_norm": 0.0017451278399676085,
-      "learning_rate": 0.00019998045179992185,
-      "loss": 46.0,
-      "step": 39094
-    },
-    {
-      "epoch": 6.295986150811224,
-      "grad_norm": 0.0014398435596376657,
-      "learning_rate": 0.00019998045079962437,
-      "loss": 46.0,
-      "step": 39095
-    },
-    {
-      "epoch": 6.2961471878900115,
-      "grad_norm": 0.011341184377670288,
-      "learning_rate": 0.00019998044979930131,
-      "loss": 46.0,
-      "step": 39096
-    },
-    {
-      "epoch": 6.296308224968799,
-      "grad_norm": 0.002223289804533124,
-      "learning_rate": 0.00019998044879895267,
-      "loss": 46.0,
-      "step": 39097
-    },
-    {
-      "epoch": 6.296469262047586,
-      "grad_norm": 0.0019719330593943596,
-      "learning_rate": 0.0001999804477985784,
-      "loss": 46.0,
-      "step": 39098
-    },
-    {
-      "epoch": 6.296630299126374,
-      "grad_norm": 0.009261532686650753,
-      "learning_rate": 0.0001999804467981786,
-      "loss": 46.0,
-      "step": 39099
-    },
-    {
-      "epoch": 6.296791336205161,
-      "grad_norm": 0.001704698195680976,
-      "learning_rate": 0.00019998044579775317,
-      "loss": 46.0,
-      "step": 39100
-    },
-    {
-      "epoch": 6.296952373283949,
-      "grad_norm": 0.0019331870134919882,
-      "learning_rate": 0.00019998044479730215,
-      "loss": 46.0,
-      "step": 39101
-    },
-    {
-      "epoch": 6.297113410362736,
-      "grad_norm": 0.0017456094501540065,
-      "learning_rate": 0.00019998044379682553,
-      "loss": 46.0,
-      "step": 39102
-    },
-    {
-      "epoch": 6.297274447441524,
-      "grad_norm": 0.004795679822564125,
-      "learning_rate": 0.00019998044279632336,
-      "loss": 46.0,
-      "step": 39103
-    },
-    {
-      "epoch": 6.297435484520311,
-      "grad_norm": 0.02229963429272175,
-      "learning_rate": 0.00019998044179579558,
-      "loss": 46.0,
-      "step": 39104
-    },
-    {
-      "epoch": 6.2975965215990986,
-      "grad_norm": 0.018068289384245872,
-      "learning_rate": 0.0001999804407952422,
-      "loss": 46.0,
-      "step": 39105
-    },
-    {
-      "epoch": 6.297757558677885,
-      "grad_norm": 0.017323555424809456,
-      "learning_rate": 0.00019998043979466325,
-      "loss": 46.0,
-      "step": 39106
-    },
-    {
-      "epoch": 6.297918595756673,
-      "grad_norm": 0.001503427280113101,
-      "learning_rate": 0.0001999804387940587,
-      "loss": 46.0,
-      "step": 39107
-    },
-    {
-      "epoch": 6.29807963283546,
-      "grad_norm": 0.003334401175379753,
-      "learning_rate": 0.00019998043779342854,
-      "loss": 46.0,
-      "step": 39108
-    },
-    {
-      "epoch": 6.2982406699142475,
-      "grad_norm": 0.011444994248449802,
-      "learning_rate": 0.00019998043679277282,
-      "loss": 46.0,
-      "step": 39109
-    },
-    {
-      "epoch": 6.298401706993035,
-      "grad_norm": 0.011853308416903019,
-      "learning_rate": 0.00019998043579209148,
-      "loss": 46.0,
-      "step": 39110
-    },
-    {
-      "epoch": 6.298562744071822,
-      "grad_norm": 0.0029568809550255537,
-      "learning_rate": 0.00019998043479138458,
-      "loss": 46.0,
-      "step": 39111
-    },
-    {
-      "epoch": 6.29872378115061,
-      "grad_norm": 0.0021663238294422626,
-      "learning_rate": 0.00019998043379065207,
-      "loss": 46.0,
-      "step": 39112
-    },
-    {
-      "epoch": 6.298884818229397,
-      "grad_norm": 0.004826546646654606,
-      "learning_rate": 0.000199980432789894,
-      "loss": 46.0,
-      "step": 39113
-    },
-    {
-      "epoch": 6.299045855308185,
-      "grad_norm": 0.001618703012354672,
-      "learning_rate": 0.00019998043178911032,
-      "loss": 46.0,
-      "step": 39114
-    },
-    {
-      "epoch": 6.299206892386972,
-      "grad_norm": 0.006082882639020681,
-      "learning_rate": 0.00019998043078830105,
-      "loss": 46.0,
-      "step": 39115
-    },
-    {
-      "epoch": 6.29936792946576,
-      "grad_norm": 0.0039020765107125044,
-      "learning_rate": 0.00019998042978746622,
-      "loss": 46.0,
-      "step": 39116
-    },
-    {
-      "epoch": 6.299528966544547,
-      "grad_norm": 0.006699398625642061,
-      "learning_rate": 0.00019998042878660574,
-      "loss": 46.0,
-      "step": 39117
-    },
-    {
-      "epoch": 6.2996900036233345,
-      "grad_norm": 0.004362224135547876,
-      "learning_rate": 0.0001999804277857197,
-      "loss": 46.0,
-      "step": 39118
-    },
-    {
-      "epoch": 6.299851040702122,
-      "grad_norm": 0.004655822180211544,
-      "learning_rate": 0.0001999804267848081,
-      "loss": 46.0,
-      "step": 39119
-    },
-    {
-      "epoch": 6.300012077780909,
-      "grad_norm": 0.011351593770086765,
-      "learning_rate": 0.00019998042578387085,
-      "loss": 46.0,
-      "step": 39120
-    },
-    {
-      "epoch": 6.300173114859696,
-      "grad_norm": 0.0015688723651692271,
-      "learning_rate": 0.00019998042478290806,
-      "loss": 46.0,
-      "step": 39121
-    },
-    {
-      "epoch": 6.3003341519384835,
-      "grad_norm": 0.01027495414018631,
-      "learning_rate": 0.00019998042378191965,
-      "loss": 46.0,
-      "step": 39122
-    },
-    {
-      "epoch": 6.300495189017271,
-      "grad_norm": 0.008149495348334312,
-      "learning_rate": 0.00019998042278090568,
-      "loss": 46.0,
-      "step": 39123
-    },
-    {
-      "epoch": 6.300656226096058,
-      "grad_norm": 0.005291405133903027,
-      "learning_rate": 0.00019998042177986607,
-      "loss": 46.0,
-      "step": 39124
-    },
-    {
-      "epoch": 6.300817263174846,
-      "grad_norm": 0.013073574751615524,
-      "learning_rate": 0.00019998042077880093,
-      "loss": 46.0,
-      "step": 39125
-    },
-    {
-      "epoch": 6.300978300253633,
-      "grad_norm": 0.0036625864449888468,
-      "learning_rate": 0.00019998041977771017,
-      "loss": 46.0,
-      "step": 39126
-    },
-    {
-      "epoch": 6.301139337332421,
-      "grad_norm": 0.003950038924813271,
-      "learning_rate": 0.00019998041877659383,
-      "loss": 46.0,
-      "step": 39127
-    },
-    {
-      "epoch": 6.301300374411208,
-      "grad_norm": 0.008592743426561356,
-      "learning_rate": 0.00019998041777545187,
-      "loss": 46.0,
-      "step": 39128
-    },
-    {
-      "epoch": 6.301461411489996,
-      "grad_norm": 0.0017043410334736109,
-      "learning_rate": 0.00019998041677428435,
-      "loss": 46.0,
-      "step": 39129
-    },
-    {
-      "epoch": 6.301622448568783,
-      "grad_norm": 0.007966981269419193,
-      "learning_rate": 0.00019998041577309121,
-      "loss": 46.0,
-      "step": 39130
-    },
-    {
-      "epoch": 6.3017834856475705,
-      "grad_norm": 0.01137902494519949,
-      "learning_rate": 0.0001999804147718725,
-      "loss": 46.0,
-      "step": 39131
-    },
-    {
-      "epoch": 6.301944522726358,
-      "grad_norm": 0.0053497301414608955,
-      "learning_rate": 0.0001999804137706282,
-      "loss": 46.0,
-      "step": 39132
-    },
-    {
-      "epoch": 6.302105559805145,
-      "grad_norm": 0.009508968330919743,
-      "learning_rate": 0.00019998041276935835,
-      "loss": 46.0,
-      "step": 39133
-    },
-    {
-      "epoch": 6.302266596883933,
-      "grad_norm": 0.013140874914824963,
-      "learning_rate": 0.00019998041176806286,
-      "loss": 46.0,
-      "step": 39134
-    },
-    {
-      "epoch": 6.30242763396272,
-      "grad_norm": 0.0019670783076435328,
-      "learning_rate": 0.0001999804107667418,
-      "loss": 46.0,
-      "step": 39135
-    },
-    {
-      "epoch": 6.302588671041507,
-      "grad_norm": 0.005829277448356152,
-      "learning_rate": 0.00019998040976539514,
-      "loss": 46.0,
-      "step": 39136
-    },
-    {
-      "epoch": 6.302749708120294,
-      "grad_norm": 0.0085930610075593,
-      "learning_rate": 0.0001999804087640229,
-      "loss": 46.0,
-      "step": 39137
-    },
-    {
-      "epoch": 6.302910745199082,
-      "grad_norm": 0.0019664703868329525,
-      "learning_rate": 0.00019998040776262506,
-      "loss": 46.0,
-      "step": 39138
-    },
-    {
-      "epoch": 6.303071782277869,
-      "grad_norm": 0.008131389506161213,
-      "learning_rate": 0.00019998040676120162,
-      "loss": 46.0,
-      "step": 39139
-    },
-    {
-      "epoch": 6.303232819356657,
-      "grad_norm": 0.0008929395698942244,
-      "learning_rate": 0.0001999804057597526,
-      "loss": 46.0,
-      "step": 39140
-    },
-    {
-      "epoch": 6.303393856435444,
-      "grad_norm": 0.0014199292054399848,
-      "learning_rate": 0.00019998040475827802,
-      "loss": 46.0,
-      "step": 39141
-    },
-    {
-      "epoch": 6.303554893514232,
-      "grad_norm": 0.007445950526744127,
-      "learning_rate": 0.00019998040375677781,
-      "loss": 46.0,
-      "step": 39142
-    },
-    {
-      "epoch": 6.303715930593019,
-      "grad_norm": 0.004941760562360287,
-      "learning_rate": 0.000199980402755252,
-      "loss": 46.0,
-      "step": 39143
-    },
-    {
-      "epoch": 6.3038769676718065,
-      "grad_norm": 0.0027029714547097683,
-      "learning_rate": 0.00019998040175370064,
-      "loss": 46.0,
-      "step": 39144
-    },
-    {
-      "epoch": 6.304038004750594,
-      "grad_norm": 0.003458069171756506,
-      "learning_rate": 0.00019998040075212367,
-      "loss": 46.0,
-      "step": 39145
-    },
-    {
-      "epoch": 6.304199041829381,
-      "grad_norm": 0.005030560307204723,
-      "learning_rate": 0.00019998039975052114,
-      "loss": 46.0,
-      "step": 39146
-    },
-    {
-      "epoch": 6.304360078908169,
-      "grad_norm": 0.014450975693762302,
-      "learning_rate": 0.00019998039874889297,
-      "loss": 46.0,
-      "step": 39147
-    },
-    {
-      "epoch": 6.304521115986956,
-      "grad_norm": 0.00146191893145442,
-      "learning_rate": 0.00019998039774723924,
-      "loss": 46.0,
-      "step": 39148
-    },
-    {
-      "epoch": 6.304682153065744,
-      "grad_norm": 0.005476540885865688,
-      "learning_rate": 0.00019998039674555993,
-      "loss": 46.0,
-      "step": 39149
-    },
-    {
-      "epoch": 6.304843190144531,
-      "grad_norm": 0.004657631739974022,
-      "learning_rate": 0.000199980395743855,
-      "loss": 46.0,
-      "step": 39150
-    },
-    {
-      "epoch": 6.305004227223318,
-      "grad_norm": 0.005235560238361359,
-      "learning_rate": 0.0001999803947421245,
-      "loss": 46.0,
-      "step": 39151
-    },
-    {
-      "epoch": 6.305165264302105,
-      "grad_norm": 0.001904643140733242,
-      "learning_rate": 0.0001999803937403684,
-      "loss": 46.0,
-      "step": 39152
-    },
-    {
-      "epoch": 6.305326301380893,
-      "grad_norm": 0.00708166416734457,
-      "learning_rate": 0.00019998039273858673,
-      "loss": 46.0,
-      "step": 39153
-    },
-    {
-      "epoch": 6.30548733845968,
-      "grad_norm": 0.0015886577311903238,
-      "learning_rate": 0.00019998039173677946,
-      "loss": 46.0,
-      "step": 39154
-    },
-    {
-      "epoch": 6.305648375538468,
-      "grad_norm": 0.012389445677399635,
-      "learning_rate": 0.0001999803907349466,
-      "loss": 46.0,
-      "step": 39155
-    },
-    {
-      "epoch": 6.305809412617255,
-      "grad_norm": 0.00206724158488214,
-      "learning_rate": 0.00019998038973308813,
-      "loss": 46.0,
-      "step": 39156
-    },
-    {
-      "epoch": 6.3059704496960425,
-      "grad_norm": 0.0034249427262693644,
-      "learning_rate": 0.0001999803887312041,
-      "loss": 46.0,
-      "step": 39157
-    },
-    {
-      "epoch": 6.30613148677483,
-      "grad_norm": 0.003982074558734894,
-      "learning_rate": 0.00019998038772929446,
-      "loss": 46.0,
-      "step": 39158
-    },
-    {
-      "epoch": 6.306292523853617,
-      "grad_norm": 0.008811620064079762,
-      "learning_rate": 0.00019998038672735925,
-      "loss": 46.0,
-      "step": 39159
-    },
-    {
-      "epoch": 6.306453560932405,
-      "grad_norm": 0.00203324225731194,
-      "learning_rate": 0.00019998038572539845,
-      "loss": 46.0,
-      "step": 39160
-    },
-    {
-      "epoch": 6.306614598011192,
-      "grad_norm": 0.0016288454644382,
-      "learning_rate": 0.00019998038472341203,
-      "loss": 46.0,
-      "step": 39161
-    },
-    {
-      "epoch": 6.30677563508998,
-      "grad_norm": 0.005029772873967886,
-      "learning_rate": 0.00019998038372140003,
-      "loss": 46.0,
-      "step": 39162
-    },
-    {
-      "epoch": 6.306936672168767,
-      "grad_norm": 0.0026036540511995554,
-      "learning_rate": 0.00019998038271936243,
-      "loss": 46.0,
-      "step": 39163
-    },
-    {
-      "epoch": 6.307097709247555,
-      "grad_norm": 0.008446864783763885,
-      "learning_rate": 0.00019998038171729928,
-      "loss": 46.0,
-      "step": 39164
-    },
-    {
-      "epoch": 6.307258746326342,
-      "grad_norm": 0.004770705010741949,
-      "learning_rate": 0.0001999803807152105,
-      "loss": 46.0,
-      "step": 39165
-    },
-    {
-      "epoch": 6.307419783405129,
-      "grad_norm": 0.006333968602120876,
-      "learning_rate": 0.00019998037971309616,
-      "loss": 46.0,
-      "step": 39166
-    },
-    {
-      "epoch": 6.307580820483916,
-      "grad_norm": 0.0029384917579591274,
-      "learning_rate": 0.0001999803787109562,
-      "loss": 46.0,
-      "step": 39167
-    },
-    {
-      "epoch": 6.307741857562704,
-      "grad_norm": 0.001456237630918622,
-      "learning_rate": 0.00019998037770879067,
-      "loss": 46.0,
-      "step": 39168
-    },
-    {
-      "epoch": 6.307902894641491,
-      "grad_norm": 0.0021201788913458586,
-      "learning_rate": 0.00019998037670659958,
-      "loss": 46.0,
-      "step": 39169
-    },
-    {
-      "epoch": 6.3080639317202785,
-      "grad_norm": 0.027082212269306183,
-      "learning_rate": 0.00019998037570438285,
-      "loss": 46.0,
-      "step": 39170
-    },
-    {
-      "epoch": 6.308224968799066,
-      "grad_norm": 0.003020569682121277,
-      "learning_rate": 0.00019998037470214054,
-      "loss": 46.0,
-      "step": 39171
-    },
-    {
-      "epoch": 6.308386005877853,
-      "grad_norm": 0.005190604366362095,
-      "learning_rate": 0.00019998037369987266,
-      "loss": 46.0,
-      "step": 39172
-    },
-    {
-      "epoch": 6.308547042956641,
-      "grad_norm": 0.007044903002679348,
-      "learning_rate": 0.00019998037269757917,
-      "loss": 46.0,
-      "step": 39173
-    },
-    {
-      "epoch": 6.308708080035428,
-      "grad_norm": 0.0012601400958374143,
-      "learning_rate": 0.0001999803716952601,
-      "loss": 46.0,
-      "step": 39174
-    },
-    {
-      "epoch": 6.308869117114216,
-      "grad_norm": 0.0019907911773771048,
-      "learning_rate": 0.00019998037069291543,
-      "loss": 46.0,
-      "step": 39175
-    },
-    {
-      "epoch": 6.309030154193003,
-      "grad_norm": 0.009525091387331486,
-      "learning_rate": 0.00019998036969054518,
-      "loss": 46.0,
-      "step": 39176
-    },
-    {
-      "epoch": 6.309191191271791,
-      "grad_norm": 0.00317505095154047,
-      "learning_rate": 0.00019998036868814937,
-      "loss": 46.0,
-      "step": 39177
-    },
-    {
-      "epoch": 6.309352228350578,
-      "grad_norm": 0.015424644574522972,
-      "learning_rate": 0.0001999803676857279,
-      "loss": 46.0,
-      "step": 39178
-    },
-    {
-      "epoch": 6.3095132654293655,
-      "grad_norm": 0.003346010809764266,
-      "learning_rate": 0.0001999803666832809,
-      "loss": 46.0,
-      "step": 39179
-    },
-    {
-      "epoch": 6.309674302508152,
-      "grad_norm": 0.0050209881737828255,
-      "learning_rate": 0.0001999803656808083,
-      "loss": 46.0,
-      "step": 39180
-    },
-    {
-      "epoch": 6.3098353395869395,
-      "grad_norm": 0.002247745404019952,
-      "learning_rate": 0.00019998036467831008,
-      "loss": 46.0,
-      "step": 39181
-    },
-    {
-      "epoch": 6.309996376665727,
-      "grad_norm": 0.0019059209153056145,
-      "learning_rate": 0.0001999803636757863,
-      "loss": 46.0,
-      "step": 39182
-    },
-    {
-      "epoch": 6.310157413744514,
-      "grad_norm": 0.007723457179963589,
-      "learning_rate": 0.00019998036267323692,
-      "loss": 46.0,
-      "step": 39183
-    },
-    {
-      "epoch": 6.310318450823302,
-      "grad_norm": 0.0029650733340531588,
-      "learning_rate": 0.00019998036167066197,
-      "loss": 46.0,
-      "step": 39184
-    },
-    {
-      "epoch": 6.310479487902089,
-      "grad_norm": 0.005114555358886719,
-      "learning_rate": 0.00019998036066806138,
-      "loss": 46.0,
-      "step": 39185
-    },
-    {
-      "epoch": 6.310640524980877,
-      "grad_norm": 0.001683143200352788,
-      "learning_rate": 0.00019998035966543526,
-      "loss": 46.0,
-      "step": 39186
-    },
-    {
-      "epoch": 6.310801562059664,
-      "grad_norm": 0.0021252150181680918,
-      "learning_rate": 0.0001999803586627835,
-      "loss": 46.0,
-      "step": 39187
-    },
-    {
-      "epoch": 6.310962599138452,
-      "grad_norm": 0.005112302955240011,
-      "learning_rate": 0.00019998035766010617,
-      "loss": 46.0,
-      "step": 39188
-    },
-    {
-      "epoch": 6.311123636217239,
-      "grad_norm": 0.0043867845088243484,
-      "learning_rate": 0.00019998035665740326,
-      "loss": 46.0,
-      "step": 39189
-    },
-    {
-      "epoch": 6.311284673296027,
-      "grad_norm": 0.00221996963955462,
-      "learning_rate": 0.00019998035565467476,
-      "loss": 46.0,
-      "step": 39190
-    },
-    {
-      "epoch": 6.311445710374814,
-      "grad_norm": 0.0011288755340501666,
-      "learning_rate": 0.00019998035465192064,
-      "loss": 46.0,
-      "step": 39191
-    },
-    {
-      "epoch": 6.3116067474536015,
-      "grad_norm": 0.0027596564032137394,
-      "learning_rate": 0.00019998035364914097,
-      "loss": 46.0,
-      "step": 39192
-    },
-    {
-      "epoch": 6.311767784532389,
-      "grad_norm": 0.007949800230562687,
-      "learning_rate": 0.00019998035264633568,
-      "loss": 46.0,
-      "step": 39193
-    },
-    {
-      "epoch": 6.3119288216111755,
-      "grad_norm": 0.001189812202937901,
-      "learning_rate": 0.0001999803516435048,
-      "loss": 46.0,
-      "step": 39194
-    },
-    {
-      "epoch": 6.312089858689963,
-      "grad_norm": 0.002585930284112692,
-      "learning_rate": 0.00019998035064064835,
-      "loss": 46.0,
-      "step": 39195
-    },
-    {
-      "epoch": 6.31225089576875,
-      "grad_norm": 0.009255031123757362,
-      "learning_rate": 0.0001999803496377663,
-      "loss": 46.0,
-      "step": 39196
-    },
-    {
-      "epoch": 6.312411932847538,
-      "grad_norm": 0.006625410169363022,
-      "learning_rate": 0.0001999803486348587,
-      "loss": 46.0,
-      "step": 39197
-    },
-    {
-      "epoch": 6.312572969926325,
-      "grad_norm": 0.005916438531130552,
-      "learning_rate": 0.00019998034763192544,
-      "loss": 46.0,
-      "step": 39198
-    },
-    {
-      "epoch": 6.312734007005113,
-      "grad_norm": 0.001434377976693213,
-      "learning_rate": 0.00019998034662896663,
-      "loss": 46.0,
-      "step": 39199
-    },
-    {
-      "epoch": 6.3128950440839,
-      "grad_norm": 0.0037752208299934864,
-      "learning_rate": 0.00019998034562598223,
-      "loss": 46.0,
-      "step": 39200
-    },
-    {
-      "epoch": 6.313056081162688,
-      "grad_norm": 0.003529967973008752,
-      "learning_rate": 0.00019998034462297221,
-      "loss": 46.0,
-      "step": 39201
-    },
-    {
-      "epoch": 6.313217118241475,
-      "grad_norm": 0.011132239364087582,
-      "learning_rate": 0.00019998034361993664,
-      "loss": 46.0,
-      "step": 39202
-    },
-    {
-      "epoch": 6.313378155320263,
-      "grad_norm": 0.016028204932808876,
-      "learning_rate": 0.00019998034261687546,
-      "loss": 46.0,
-      "step": 39203
-    },
-    {
-      "epoch": 6.31353919239905,
-      "grad_norm": 0.006546396296471357,
-      "learning_rate": 0.00019998034161378868,
-      "loss": 46.0,
-      "step": 39204
-    },
-    {
-      "epoch": 6.3137002294778375,
-      "grad_norm": 0.009632307104766369,
-      "learning_rate": 0.00019998034061067635,
-      "loss": 46.0,
-      "step": 39205
-    },
-    {
-      "epoch": 6.313861266556625,
-      "grad_norm": 0.0013545771362259984,
-      "learning_rate": 0.0001999803396075384,
-      "loss": 46.0,
-      "step": 39206
-    },
-    {
-      "epoch": 6.314022303635412,
-      "grad_norm": 0.014357620850205421,
-      "learning_rate": 0.00019998033860437486,
-      "loss": 46.0,
-      "step": 39207
-    },
-    {
-      "epoch": 6.3141833407142,
-      "grad_norm": 0.007033585105091333,
-      "learning_rate": 0.00019998033760118574,
-      "loss": 46.0,
-      "step": 39208
-    },
-    {
-      "epoch": 6.314344377792986,
-      "grad_norm": 0.0013322340091690421,
-      "learning_rate": 0.000199980336597971,
-      "loss": 46.0,
-      "step": 39209
-    },
-    {
-      "epoch": 6.314505414871774,
-      "grad_norm": 0.014954268001019955,
-      "learning_rate": 0.00019998033559473074,
-      "loss": 46.0,
-      "step": 39210
-    },
-    {
-      "epoch": 6.314666451950561,
-      "grad_norm": 0.005739069543778896,
-      "learning_rate": 0.00019998033459146482,
-      "loss": 46.0,
-      "step": 39211
-    },
-    {
-      "epoch": 6.314827489029349,
-      "grad_norm": 0.0020559667609632015,
-      "learning_rate": 0.00019998033358817333,
-      "loss": 46.0,
-      "step": 39212
-    },
-    {
-      "epoch": 6.314988526108136,
-      "grad_norm": 0.0015947798965498805,
-      "learning_rate": 0.00019998033258485624,
-      "loss": 46.0,
-      "step": 39213
-    },
-    {
-      "epoch": 6.315149563186924,
-      "grad_norm": 0.015269514173269272,
-      "learning_rate": 0.0001999803315815136,
-      "loss": 46.0,
-      "step": 39214
-    },
-    {
-      "epoch": 6.315310600265711,
-      "grad_norm": 0.006795963272452354,
-      "learning_rate": 0.00019998033057814534,
-      "loss": 46.0,
-      "step": 39215
-    },
-    {
-      "epoch": 6.315471637344499,
-      "grad_norm": 0.0033634265419095755,
-      "learning_rate": 0.0001999803295747515,
-      "loss": 46.0,
-      "step": 39216
-    },
-    {
-      "epoch": 6.315632674423286,
-      "grad_norm": 0.001218912424519658,
-      "learning_rate": 0.00019998032857133205,
-      "loss": 46.0,
-      "step": 39217
-    },
-    {
-      "epoch": 6.3157937115020735,
-      "grad_norm": 0.007960383780300617,
-      "learning_rate": 0.00019998032756788706,
-      "loss": 46.0,
-      "step": 39218
-    },
-    {
-      "epoch": 6.315954748580861,
-      "grad_norm": 0.0032012341544032097,
-      "learning_rate": 0.0001999803265644164,
-      "loss": 46.0,
-      "step": 39219
-    },
-    {
-      "epoch": 6.316115785659648,
-      "grad_norm": 0.002882528118789196,
-      "learning_rate": 0.00019998032556092023,
-      "loss": 46.0,
-      "step": 39220
-    },
-    {
-      "epoch": 6.316276822738436,
-      "grad_norm": 0.013656115159392357,
-      "learning_rate": 0.00019998032455739845,
-      "loss": 46.0,
-      "step": 39221
-    },
-    {
-      "epoch": 6.316437859817223,
-      "grad_norm": 0.0015731171006336808,
-      "learning_rate": 0.00019998032355385105,
-      "loss": 46.0,
-      "step": 39222
-    },
-    {
-      "epoch": 6.316598896896011,
-      "grad_norm": 0.0029132100753486156,
-      "learning_rate": 0.00019998032255027806,
-      "loss": 46.0,
-      "step": 39223
-    },
-    {
-      "epoch": 6.316759933974797,
-      "grad_norm": 0.007392663508653641,
-      "learning_rate": 0.00019998032154667952,
-      "loss": 46.0,
-      "step": 39224
-    },
-    {
-      "epoch": 6.316920971053585,
-      "grad_norm": 0.01143241673707962,
-      "learning_rate": 0.00019998032054305536,
-      "loss": 46.0,
-      "step": 39225
-    },
-    {
-      "epoch": 6.317082008132372,
-      "grad_norm": 0.0049977474845945835,
-      "learning_rate": 0.00019998031953940562,
-      "loss": 46.0,
-      "step": 39226
-    },
-    {
-      "epoch": 6.31724304521116,
-      "grad_norm": 0.014875965192914009,
-      "learning_rate": 0.00019998031853573028,
-      "loss": 46.0,
-      "step": 39227
-    },
-    {
-      "epoch": 6.317404082289947,
-      "grad_norm": 0.010098783299326897,
-      "learning_rate": 0.00019998031753202934,
-      "loss": 46.0,
-      "step": 39228
-    },
-    {
-      "epoch": 6.3175651193687345,
-      "grad_norm": 0.0035059915389865637,
-      "learning_rate": 0.00019998031652830286,
-      "loss": 46.0,
-      "step": 39229
-    },
-    {
-      "epoch": 6.317726156447522,
-      "grad_norm": 0.00809943862259388,
-      "learning_rate": 0.00019998031552455073,
-      "loss": 46.0,
-      "step": 39230
-    },
-    {
-      "epoch": 6.3178871935263095,
-      "grad_norm": 0.01116069033741951,
-      "learning_rate": 0.00019998031452077303,
-      "loss": 46.0,
-      "step": 39231
-    },
-    {
-      "epoch": 6.318048230605097,
-      "grad_norm": 0.0016322049777954817,
-      "learning_rate": 0.00019998031351696978,
-      "loss": 46.0,
-      "step": 39232
-    },
-    {
-      "epoch": 6.318209267683884,
-      "grad_norm": 0.005741297733038664,
-      "learning_rate": 0.0001999803125131409,
-      "loss": 46.0,
-      "step": 39233
-    },
-    {
-      "epoch": 6.318370304762672,
-      "grad_norm": 0.0017672909889370203,
-      "learning_rate": 0.00019998031150928643,
-      "loss": 46.0,
-      "step": 39234
-    },
-    {
-      "epoch": 6.318531341841459,
-      "grad_norm": 0.015608440153300762,
-      "learning_rate": 0.00019998031050540637,
-      "loss": 46.0,
-      "step": 39235
-    },
-    {
-      "epoch": 6.318692378920247,
-      "grad_norm": 0.0031857292633503675,
-      "learning_rate": 0.00019998030950150076,
-      "loss": 46.0,
-      "step": 39236
-    },
-    {
-      "epoch": 6.318853415999034,
-      "grad_norm": 0.0018248548731207848,
-      "learning_rate": 0.0001999803084975695,
-      "loss": 46.0,
-      "step": 39237
-    },
-    {
-      "epoch": 6.319014453077822,
-      "grad_norm": 0.0016317018307745457,
-      "learning_rate": 0.0001999803074936127,
-      "loss": 46.0,
-      "step": 39238
-    },
-    {
-      "epoch": 6.319175490156608,
-      "grad_norm": 0.0008817727211862803,
-      "learning_rate": 0.0001999803064896303,
-      "loss": 46.0,
-      "step": 39239
-    },
-    {
-      "epoch": 6.319336527235396,
-      "grad_norm": 0.01828228496015072,
-      "learning_rate": 0.00019998030548562228,
-      "loss": 46.0,
-      "step": 39240
-    },
-    {
-      "epoch": 6.319497564314183,
-      "grad_norm": 0.007376500405371189,
-      "learning_rate": 0.0001999803044815887,
-      "loss": 46.0,
-      "step": 39241
-    },
-    {
-      "epoch": 6.3196586013929705,
-      "grad_norm": 0.0017238880973309278,
-      "learning_rate": 0.00019998030347752953,
-      "loss": 46.0,
-      "step": 39242
-    },
-    {
-      "epoch": 6.319819638471758,
-      "grad_norm": 0.0020580957643687725,
-      "learning_rate": 0.00019998030247344475,
-      "loss": 46.0,
-      "step": 39243
-    },
-    {
-      "epoch": 6.319980675550545,
-      "grad_norm": 0.014273291453719139,
-      "learning_rate": 0.0001999803014693344,
-      "loss": 46.0,
-      "step": 39244
-    },
-    {
-      "epoch": 6.320141712629333,
-      "grad_norm": 0.0007183046545833349,
-      "learning_rate": 0.00019998030046519842,
-      "loss": 46.0,
-      "step": 39245
-    },
-    {
-      "epoch": 6.32030274970812,
-      "grad_norm": 0.00794605165719986,
-      "learning_rate": 0.00019998029946103688,
-      "loss": 46.0,
-      "step": 39246
-    },
-    {
-      "epoch": 6.320463786786908,
-      "grad_norm": 0.011247053742408752,
-      "learning_rate": 0.00019998029845684978,
-      "loss": 46.0,
-      "step": 39247
-    },
-    {
-      "epoch": 6.320624823865695,
-      "grad_norm": 0.001349453697912395,
-      "learning_rate": 0.00019998029745263703,
-      "loss": 46.0,
-      "step": 39248
-    },
-    {
-      "epoch": 6.320785860944483,
-      "grad_norm": 0.0016149523435160518,
-      "learning_rate": 0.00019998029644839873,
-      "loss": 46.0,
-      "step": 39249
-    },
-    {
-      "epoch": 6.32094689802327,
-      "grad_norm": 0.005839682184159756,
-      "learning_rate": 0.00019998029544413483,
-      "loss": 46.0,
-      "step": 39250
-    },
-    {
-      "epoch": 6.321107935102058,
-      "grad_norm": 0.007284276653081179,
-      "learning_rate": 0.00019998029443984536,
-      "loss": 46.0,
-      "step": 39251
-    },
-    {
-      "epoch": 6.321268972180845,
-      "grad_norm": 0.024734221398830414,
-      "learning_rate": 0.0001999802934355303,
-      "loss": 46.0,
-      "step": 39252
-    },
-    {
-      "epoch": 6.3214300092596325,
-      "grad_norm": 0.0008910219185054302,
-      "learning_rate": 0.0001999802924311896,
-      "loss": 46.0,
-      "step": 39253
-    },
-    {
-      "epoch": 6.321591046338419,
-      "grad_norm": 0.004034031648188829,
-      "learning_rate": 0.00019998029142682334,
-      "loss": 46.0,
-      "step": 39254
-    },
-    {
-      "epoch": 6.3217520834172065,
-      "grad_norm": 0.018770208582282066,
-      "learning_rate": 0.00019998029042243148,
-      "loss": 46.0,
-      "step": 39255
-    },
-    {
-      "epoch": 6.321913120495994,
-      "grad_norm": 0.0029557356610894203,
-      "learning_rate": 0.00019998028941801404,
-      "loss": 46.0,
-      "step": 39256
-    },
-    {
-      "epoch": 6.322074157574781,
-      "grad_norm": 0.008802195079624653,
-      "learning_rate": 0.00019998028841357104,
-      "loss": 46.0,
-      "step": 39257
-    },
-    {
-      "epoch": 6.322235194653569,
-      "grad_norm": 0.004535462241619825,
-      "learning_rate": 0.00019998028740910242,
-      "loss": 46.0,
-      "step": 39258
-    },
-    {
-      "epoch": 6.322396231732356,
-      "grad_norm": 0.001260081771761179,
-      "learning_rate": 0.00019998028640460822,
-      "loss": 46.0,
-      "step": 39259
-    },
-    {
-      "epoch": 6.322557268811144,
-      "grad_norm": 0.0010555196786299348,
-      "learning_rate": 0.00019998028540008843,
-      "loss": 46.0,
-      "step": 39260
-    },
-    {
-      "epoch": 6.322718305889931,
-      "grad_norm": 0.005778071470558643,
-      "learning_rate": 0.00019998028439554305,
-      "loss": 46.0,
-      "step": 39261
-    },
-    {
-      "epoch": 6.322879342968719,
-      "grad_norm": 0.0042562601156532764,
-      "learning_rate": 0.00019998028339097206,
-      "loss": 46.0,
-      "step": 39262
-    },
-    {
-      "epoch": 6.323040380047506,
-      "grad_norm": 0.02027692273259163,
-      "learning_rate": 0.0001999802823863755,
-      "loss": 46.0,
-      "step": 39263
-    },
-    {
-      "epoch": 6.323201417126294,
-      "grad_norm": 0.004296464845538139,
-      "learning_rate": 0.00019998028138175334,
-      "loss": 46.0,
-      "step": 39264
-    },
-    {
-      "epoch": 6.323362454205081,
-      "grad_norm": 0.0060869259759783745,
-      "learning_rate": 0.00019998028037710559,
-      "loss": 46.0,
-      "step": 39265
-    },
-    {
-      "epoch": 6.3235234912838685,
-      "grad_norm": 0.020527543500065804,
-      "learning_rate": 0.00019998027937243224,
-      "loss": 46.0,
-      "step": 39266
-    },
-    {
-      "epoch": 6.323684528362655,
-      "grad_norm": 0.0066842553205788136,
-      "learning_rate": 0.00019998027836773332,
-      "loss": 46.0,
-      "step": 39267
-    },
-    {
-      "epoch": 6.3238455654414425,
-      "grad_norm": 0.01725761592388153,
-      "learning_rate": 0.0001999802773630088,
-      "loss": 46.0,
-      "step": 39268
-    },
-    {
-      "epoch": 6.32400660252023,
-      "grad_norm": 0.009204615838825703,
-      "learning_rate": 0.0001999802763582587,
-      "loss": 46.0,
-      "step": 39269
-    },
-    {
-      "epoch": 6.324167639599017,
-      "grad_norm": 0.006052333861589432,
-      "learning_rate": 0.000199980275353483,
-      "loss": 46.0,
-      "step": 39270
-    },
-    {
-      "epoch": 6.324328676677805,
-      "grad_norm": 0.006382647901773453,
-      "learning_rate": 0.00019998027434868173,
-      "loss": 46.0,
-      "step": 39271
-    },
-    {
-      "epoch": 6.324489713756592,
-      "grad_norm": 0.00808104407042265,
-      "learning_rate": 0.00019998027334385484,
-      "loss": 46.0,
-      "step": 39272
-    },
-    {
-      "epoch": 6.32465075083538,
-      "grad_norm": 0.003572257002815604,
-      "learning_rate": 0.0001999802723390024,
-      "loss": 46.0,
-      "step": 39273
-    },
-    {
-      "epoch": 6.324811787914167,
-      "grad_norm": 0.002168485429137945,
-      "learning_rate": 0.00019998027133412432,
-      "loss": 46.0,
-      "step": 39274
-    },
-    {
-      "epoch": 6.324972824992955,
-      "grad_norm": 0.0169981736689806,
-      "learning_rate": 0.0001999802703292207,
-      "loss": 46.0,
-      "step": 39275
-    },
-    {
-      "epoch": 6.325133862071742,
-      "grad_norm": 0.002466046018525958,
-      "learning_rate": 0.00019998026932429146,
-      "loss": 46.0,
-      "step": 39276
-    },
-    {
-      "epoch": 6.32529489915053,
-      "grad_norm": 0.004602525848895311,
-      "learning_rate": 0.00019998026831933663,
-      "loss": 46.0,
-      "step": 39277
-    },
-    {
-      "epoch": 6.325455936229317,
-      "grad_norm": 0.004470365587621927,
-      "learning_rate": 0.00019998026731435622,
-      "loss": 46.0,
-      "step": 39278
-    },
-    {
-      "epoch": 6.3256169733081045,
-      "grad_norm": 0.0031690194737166166,
-      "learning_rate": 0.00019998026630935022,
-      "loss": 46.0,
-      "step": 39279
-    },
-    {
-      "epoch": 6.325778010386892,
-      "grad_norm": 0.0015667417319491506,
-      "learning_rate": 0.00019998026530431863,
-      "loss": 46.0,
-      "step": 39280
-    },
-    {
-      "epoch": 6.325939047465679,
-      "grad_norm": 0.008357924409210682,
-      "learning_rate": 0.00019998026429926142,
-      "loss": 46.0,
-      "step": 39281
-    },
-    {
-      "epoch": 6.326100084544466,
-      "grad_norm": 0.009117667563259602,
-      "learning_rate": 0.00019998026329417866,
-      "loss": 46.0,
-      "step": 39282
-    },
-    {
-      "epoch": 6.326261121623253,
-      "grad_norm": 0.01177041232585907,
-      "learning_rate": 0.00019998026228907028,
-      "loss": 46.0,
-      "step": 39283
-    },
-    {
-      "epoch": 6.326422158702041,
-      "grad_norm": 0.005420915316790342,
-      "learning_rate": 0.00019998026128393637,
-      "loss": 46.0,
-      "step": 39284
-    },
-    {
-      "epoch": 6.326583195780828,
-      "grad_norm": 0.00805660430341959,
-      "learning_rate": 0.00019998026027877682,
-      "loss": 46.0,
-      "step": 39285
-    },
-    {
-      "epoch": 6.326744232859616,
-      "grad_norm": 0.005308833438903093,
-      "learning_rate": 0.00019998025927359168,
-      "loss": 46.0,
-      "step": 39286
-    },
-    {
-      "epoch": 6.326905269938403,
-      "grad_norm": 0.009006757289171219,
-      "learning_rate": 0.00019998025826838096,
-      "loss": 46.0,
-      "step": 39287
-    },
-    {
-      "epoch": 6.327066307017191,
-      "grad_norm": 0.001884071039967239,
-      "learning_rate": 0.00019998025726314465,
-      "loss": 46.0,
-      "step": 39288
-    },
-    {
-      "epoch": 6.327227344095978,
-      "grad_norm": 0.0026832083240151405,
-      "learning_rate": 0.00019998025625788275,
-      "loss": 46.0,
-      "step": 39289
-    },
-    {
-      "epoch": 6.3273883811747655,
-      "grad_norm": 0.003901432501152158,
-      "learning_rate": 0.00019998025525259526,
-      "loss": 46.0,
-      "step": 39290
-    },
-    {
-      "epoch": 6.327549418253553,
-      "grad_norm": 0.004649568349123001,
-      "learning_rate": 0.00019998025424728218,
-      "loss": 46.0,
-      "step": 39291
-    },
-    {
-      "epoch": 6.3277104553323404,
-      "grad_norm": 0.01862914301455021,
-      "learning_rate": 0.0001999802532419435,
-      "loss": 46.0,
-      "step": 39292
-    },
-    {
-      "epoch": 6.327871492411128,
-      "grad_norm": 0.004360104911029339,
-      "learning_rate": 0.00019998025223657925,
-      "loss": 46.0,
-      "step": 39293
-    },
-    {
-      "epoch": 6.328032529489915,
-      "grad_norm": 0.00408626114949584,
-      "learning_rate": 0.00019998025123118938,
-      "loss": 46.0,
-      "step": 39294
-    },
-    {
-      "epoch": 6.328193566568703,
-      "grad_norm": 0.0039005533326417208,
-      "learning_rate": 0.00019998025022577393,
-      "loss": 46.0,
-      "step": 39295
-    },
-    {
-      "epoch": 6.32835460364749,
-      "grad_norm": 0.011384246870875359,
-      "learning_rate": 0.00019998024922033292,
-      "loss": 46.0,
-      "step": 39296
-    },
-    {
-      "epoch": 6.328515640726277,
-      "grad_norm": 0.007530677132308483,
-      "learning_rate": 0.0001999802482148663,
-      "loss": 46.0,
-      "step": 39297
-    },
-    {
-      "epoch": 6.328676677805064,
-      "grad_norm": 0.004398074001073837,
-      "learning_rate": 0.00019998024720937408,
-      "loss": 46.0,
-      "step": 39298
-    },
-    {
-      "epoch": 6.328837714883852,
-      "grad_norm": 0.0016032342100515962,
-      "learning_rate": 0.00019998024620385629,
-      "loss": 46.0,
-      "step": 39299
-    },
-    {
-      "epoch": 6.328998751962639,
-      "grad_norm": 0.012873155064880848,
-      "learning_rate": 0.0001999802451983129,
-      "loss": 46.0,
-      "step": 39300
-    },
-    {
-      "epoch": 6.329159789041427,
-      "grad_norm": 0.032686855643987656,
-      "learning_rate": 0.0001999802441927439,
-      "loss": 46.0,
-      "step": 39301
-    },
-    {
-      "epoch": 6.329320826120214,
-      "grad_norm": 0.005882898345589638,
-      "learning_rate": 0.00019998024318714937,
-      "loss": 46.0,
-      "step": 39302
-    },
-    {
-      "epoch": 6.3294818631990015,
-      "grad_norm": 0.007807684130966663,
-      "learning_rate": 0.0001999802421815292,
-      "loss": 46.0,
-      "step": 39303
-    },
-    {
-      "epoch": 6.329642900277789,
-      "grad_norm": 0.008720343932509422,
-      "learning_rate": 0.00019998024117588346,
-      "loss": 46.0,
-      "step": 39304
-    },
-    {
-      "epoch": 6.329803937356576,
-      "grad_norm": 0.0032543563283979893,
-      "learning_rate": 0.00019998024017021208,
-      "loss": 46.0,
-      "step": 39305
-    },
-    {
-      "epoch": 6.329964974435364,
-      "grad_norm": 0.0028530405834317207,
-      "learning_rate": 0.00019998023916451517,
-      "loss": 46.0,
-      "step": 39306
-    },
-    {
-      "epoch": 6.330126011514151,
-      "grad_norm": 0.004464778117835522,
-      "learning_rate": 0.00019998023815879265,
-      "loss": 46.0,
-      "step": 39307
-    },
-    {
-      "epoch": 6.330287048592939,
-      "grad_norm": 0.0033770210575312376,
-      "learning_rate": 0.00019998023715304454,
-      "loss": 46.0,
-      "step": 39308
-    },
-    {
-      "epoch": 6.330448085671726,
-      "grad_norm": 0.019810356199741364,
-      "learning_rate": 0.00019998023614727087,
-      "loss": 46.0,
-      "step": 39309
-    },
-    {
-      "epoch": 6.330609122750514,
-      "grad_norm": 0.0027148162480443716,
-      "learning_rate": 0.00019998023514147155,
-      "loss": 46.0,
-      "step": 39310
-    },
-    {
-      "epoch": 6.330770159829301,
-      "grad_norm": 0.011955147609114647,
-      "learning_rate": 0.00019998023413564668,
-      "loss": 46.0,
-      "step": 39311
-    },
-    {
-      "epoch": 6.330931196908088,
-      "grad_norm": 0.005633179098367691,
-      "learning_rate": 0.00019998023312979622,
-      "loss": 46.0,
-      "step": 39312
-    },
-    {
-      "epoch": 6.331092233986875,
-      "grad_norm": 0.007518400903791189,
-      "learning_rate": 0.00019998023212392018,
-      "loss": 46.0,
-      "step": 39313
-    },
-    {
-      "epoch": 6.331253271065663,
-      "grad_norm": 0.0015899842837825418,
-      "learning_rate": 0.00019998023111801851,
-      "loss": 46.0,
-      "step": 39314
-    },
-    {
-      "epoch": 6.33141430814445,
-      "grad_norm": 0.0027054939419031143,
-      "learning_rate": 0.0001999802301120913,
-      "loss": 46.0,
-      "step": 39315
-    },
-    {
-      "epoch": 6.3315753452232375,
-      "grad_norm": 0.006814792286604643,
-      "learning_rate": 0.00019998022910613846,
-      "loss": 46.0,
-      "step": 39316
-    },
-    {
-      "epoch": 6.331736382302025,
-      "grad_norm": 0.0038679130375385284,
-      "learning_rate": 0.00019998022810016004,
-      "loss": 46.0,
-      "step": 39317
-    },
-    {
-      "epoch": 6.331897419380812,
-      "grad_norm": 0.0044594211503863335,
-      "learning_rate": 0.00019998022709415603,
-      "loss": 46.0,
-      "step": 39318
-    },
-    {
-      "epoch": 6.3320584564596,
-      "grad_norm": 0.002873303834348917,
-      "learning_rate": 0.00019998022608812643,
-      "loss": 46.0,
-      "step": 39319
-    },
-    {
-      "epoch": 6.332219493538387,
-      "grad_norm": 0.00246586580760777,
-      "learning_rate": 0.00019998022508207125,
-      "loss": 46.0,
-      "step": 39320
-    },
-    {
-      "epoch": 6.332380530617175,
-      "grad_norm": 0.014758276753127575,
-      "learning_rate": 0.00019998022407599048,
-      "loss": 46.0,
-      "step": 39321
-    },
-    {
-      "epoch": 6.332541567695962,
-      "grad_norm": 0.004454613663256168,
-      "learning_rate": 0.00019998022306988412,
-      "loss": 46.0,
-      "step": 39322
-    },
-    {
-      "epoch": 6.33270260477475,
-      "grad_norm": 0.0018376229563727975,
-      "learning_rate": 0.00019998022206375215,
-      "loss": 46.0,
-      "step": 39323
-    },
-    {
-      "epoch": 6.332863641853537,
-      "grad_norm": 0.01380782388150692,
-      "learning_rate": 0.00019998022105759459,
-      "loss": 46.0,
-      "step": 39324
-    },
-    {
-      "epoch": 6.333024678932325,
-      "grad_norm": 0.002713521709665656,
-      "learning_rate": 0.00019998022005141147,
-      "loss": 46.0,
-      "step": 39325
-    },
-    {
-      "epoch": 6.333185716011112,
-      "grad_norm": 0.028691653162240982,
-      "learning_rate": 0.00019998021904520273,
-      "loss": 46.0,
-      "step": 39326
-    },
-    {
-      "epoch": 6.333346753089899,
-      "grad_norm": 0.003941669128835201,
-      "learning_rate": 0.00019998021803896844,
-      "loss": 46.0,
-      "step": 39327
-    },
-    {
-      "epoch": 6.333507790168686,
-      "grad_norm": 0.00348378112539649,
-      "learning_rate": 0.00019998021703270853,
-      "loss": 46.0,
-      "step": 39328
-    },
-    {
-      "epoch": 6.3336688272474735,
-      "grad_norm": 0.007795641198754311,
-      "learning_rate": 0.00019998021602642303,
-      "loss": 46.0,
-      "step": 39329
-    },
-    {
-      "epoch": 6.333829864326261,
-      "grad_norm": 0.0017936760559678078,
-      "learning_rate": 0.00019998021502011195,
-      "loss": 46.0,
-      "step": 39330
-    },
-    {
-      "epoch": 6.333990901405048,
-      "grad_norm": 0.007852043956518173,
-      "learning_rate": 0.00019998021401377528,
-      "loss": 46.0,
-      "step": 39331
-    },
-    {
-      "epoch": 6.334151938483836,
-      "grad_norm": 0.002989057218655944,
-      "learning_rate": 0.000199980213007413,
-      "loss": 46.0,
-      "step": 39332
-    },
-    {
-      "epoch": 6.334312975562623,
-      "grad_norm": 0.005599173717200756,
-      "learning_rate": 0.00019998021200102515,
-      "loss": 46.0,
-      "step": 39333
-    },
-    {
-      "epoch": 6.334474012641411,
-      "grad_norm": 0.0017831216100603342,
-      "learning_rate": 0.0001999802109946117,
-      "loss": 46.0,
-      "step": 39334
-    },
-    {
-      "epoch": 6.334635049720198,
-      "grad_norm": 0.0014196459669619799,
-      "learning_rate": 0.00019998020998817268,
-      "loss": 46.0,
-      "step": 39335
-    },
-    {
-      "epoch": 6.334796086798986,
-      "grad_norm": 0.011581974104046822,
-      "learning_rate": 0.00019998020898170807,
-      "loss": 46.0,
-      "step": 39336
-    },
-    {
-      "epoch": 6.334957123877773,
-      "grad_norm": 0.003051828360185027,
-      "learning_rate": 0.00019998020797521785,
-      "loss": 46.0,
-      "step": 39337
-    },
-    {
-      "epoch": 6.3351181609565606,
-      "grad_norm": 0.0050719608552753925,
-      "learning_rate": 0.00019998020696870204,
-      "loss": 46.0,
-      "step": 39338
-    },
-    {
-      "epoch": 6.335279198035348,
-      "grad_norm": 0.0036676533054560423,
-      "learning_rate": 0.00019998020596216068,
-      "loss": 46.0,
-      "step": 39339
-    },
-    {
-      "epoch": 6.335440235114135,
-      "grad_norm": 0.00894926581531763,
-      "learning_rate": 0.00019998020495559367,
-      "loss": 46.0,
-      "step": 39340
-    },
-    {
-      "epoch": 6.335601272192922,
-      "grad_norm": 0.005566160194575787,
-      "learning_rate": 0.0001999802039490011,
-      "loss": 46.0,
-      "step": 39341
-    },
-    {
-      "epoch": 6.3357623092717095,
-      "grad_norm": 0.01981467381119728,
-      "learning_rate": 0.00019998020294238294,
-      "loss": 46.0,
-      "step": 39342
-    },
-    {
-      "epoch": 6.335923346350497,
-      "grad_norm": 0.005660607013851404,
-      "learning_rate": 0.00019998020193573917,
-      "loss": 46.0,
-      "step": 39343
-    },
-    {
-      "epoch": 6.336084383429284,
-      "grad_norm": 0.0038558386731892824,
-      "learning_rate": 0.00019998020092906984,
-      "loss": 46.0,
-      "step": 39344
-    },
-    {
-      "epoch": 6.336245420508072,
-      "grad_norm": 0.004187397658824921,
-      "learning_rate": 0.00019998019992237493,
-      "loss": 46.0,
-      "step": 39345
-    },
-    {
-      "epoch": 6.336406457586859,
-      "grad_norm": 0.0016036182641983032,
-      "learning_rate": 0.0001999801989156544,
-      "loss": 46.0,
-      "step": 39346
-    },
-    {
-      "epoch": 6.336567494665647,
-      "grad_norm": 0.0014665999915450811,
-      "learning_rate": 0.0001999801979089083,
-      "loss": 46.0,
-      "step": 39347
-    },
-    {
-      "epoch": 6.336728531744434,
-      "grad_norm": 0.0037546316161751747,
-      "learning_rate": 0.0001999801969021366,
-      "loss": 46.0,
-      "step": 39348
-    },
-    {
-      "epoch": 6.336889568823222,
-      "grad_norm": 0.004217116162180901,
-      "learning_rate": 0.0001999801958953393,
-      "loss": 46.0,
-      "step": 39349
-    },
-    {
-      "epoch": 6.337050605902009,
-      "grad_norm": 0.005435938481241465,
-      "learning_rate": 0.0001999801948885164,
-      "loss": 46.0,
-      "step": 39350
-    },
-    {
-      "epoch": 6.3372116429807965,
-      "grad_norm": 0.005066473037004471,
-      "learning_rate": 0.00019998019388166796,
-      "loss": 46.0,
-      "step": 39351
-    },
-    {
-      "epoch": 6.337372680059584,
-      "grad_norm": 0.017243310809135437,
-      "learning_rate": 0.00019998019287479387,
-      "loss": 46.0,
-      "step": 39352
-    },
-    {
-      "epoch": 6.337533717138371,
-      "grad_norm": 0.01188422366976738,
-      "learning_rate": 0.00019998019186789426,
-      "loss": 46.0,
-      "step": 39353
-    },
-    {
-      "epoch": 6.337694754217159,
-      "grad_norm": 0.010011446662247181,
-      "learning_rate": 0.000199980190860969,
-      "loss": 46.0,
-      "step": 39354
-    },
-    {
-      "epoch": 6.3378557912959455,
-      "grad_norm": 0.0011682201875373721,
-      "learning_rate": 0.00019998018985401816,
-      "loss": 46.0,
-      "step": 39355
-    },
-    {
-      "epoch": 6.338016828374733,
-      "grad_norm": 0.006333307828754187,
-      "learning_rate": 0.00019998018884704176,
-      "loss": 46.0,
-      "step": 39356
-    },
-    {
-      "epoch": 6.33817786545352,
-      "grad_norm": 0.007472933270037174,
-      "learning_rate": 0.00019998018784003977,
-      "loss": 46.0,
-      "step": 39357
-    },
-    {
-      "epoch": 6.338338902532308,
-      "grad_norm": 0.009217468090355396,
-      "learning_rate": 0.00019998018683301213,
-      "loss": 46.0,
-      "step": 39358
-    },
-    {
-      "epoch": 6.338499939611095,
-      "grad_norm": 0.005128510762006044,
-      "learning_rate": 0.00019998018582595897,
-      "loss": 46.0,
-      "step": 39359
-    },
-    {
-      "epoch": 6.338660976689883,
-      "grad_norm": 0.003001793520525098,
-      "learning_rate": 0.00019998018481888016,
-      "loss": 46.0,
-      "step": 39360
-    },
-    {
-      "epoch": 6.33882201376867,
-      "grad_norm": 0.001012038323096931,
-      "learning_rate": 0.00019998018381177582,
-      "loss": 46.0,
-      "step": 39361
-    },
-    {
-      "epoch": 6.338983050847458,
-      "grad_norm": 0.006456390488892794,
-      "learning_rate": 0.00019998018280464584,
-      "loss": 46.0,
-      "step": 39362
-    },
-    {
-      "epoch": 6.339144087926245,
-      "grad_norm": 0.018388397991657257,
-      "learning_rate": 0.00019998018179749033,
-      "loss": 46.0,
-      "step": 39363
-    },
-    {
-      "epoch": 6.3393051250050325,
-      "grad_norm": 0.004710391163825989,
-      "learning_rate": 0.00019998018079030918,
-      "loss": 46.0,
-      "step": 39364
-    },
-    {
-      "epoch": 6.33946616208382,
-      "grad_norm": 0.0038208451587706804,
-      "learning_rate": 0.00019998017978310246,
-      "loss": 46.0,
-      "step": 39365
-    },
-    {
-      "epoch": 6.339627199162607,
-      "grad_norm": 0.004799526650458574,
-      "learning_rate": 0.00019998017877587013,
-      "loss": 46.0,
-      "step": 39366
-    },
-    {
-      "epoch": 6.339788236241395,
-      "grad_norm": 0.0019337047124281526,
-      "learning_rate": 0.00019998017776861221,
-      "loss": 46.0,
-      "step": 39367
-    },
-    {
-      "epoch": 6.339949273320182,
-      "grad_norm": 0.0025381832383573055,
-      "learning_rate": 0.00019998017676132874,
-      "loss": 46.0,
-      "step": 39368
-    },
-    {
-      "epoch": 6.34011031039897,
-      "grad_norm": 0.0052637062035501,
-      "learning_rate": 0.00019998017575401965,
-      "loss": 46.0,
-      "step": 39369
-    },
-    {
-      "epoch": 6.340271347477756,
-      "grad_norm": 0.006857589352875948,
-      "learning_rate": 0.00019998017474668494,
-      "loss": 46.0,
-      "step": 39370
-    },
-    {
-      "epoch": 6.340432384556544,
-      "grad_norm": 0.007295440416783094,
-      "learning_rate": 0.0001999801737393247,
-      "loss": 46.0,
-      "step": 39371
-    },
-    {
-      "epoch": 6.340593421635331,
-      "grad_norm": 0.00410598935559392,
-      "learning_rate": 0.00019998017273193885,
-      "loss": 46.0,
-      "step": 39372
-    },
-    {
-      "epoch": 6.340754458714119,
-      "grad_norm": 0.016875099390745163,
-      "learning_rate": 0.0001999801717245274,
-      "loss": 46.0,
-      "step": 39373
-    },
-    {
-      "epoch": 6.340915495792906,
-      "grad_norm": 0.002704192651435733,
-      "learning_rate": 0.00019998017071709036,
-      "loss": 46.0,
-      "step": 39374
-    },
-    {
-      "epoch": 6.341076532871694,
-      "grad_norm": 0.00840501394122839,
-      "learning_rate": 0.00019998016970962775,
-      "loss": 46.0,
-      "step": 39375
-    },
-    {
-      "epoch": 6.341237569950481,
-      "grad_norm": 0.0050213211216032505,
-      "learning_rate": 0.00019998016870213952,
-      "loss": 46.0,
-      "step": 39376
-    },
-    {
-      "epoch": 6.3413986070292685,
-      "grad_norm": 0.0030093619134277105,
-      "learning_rate": 0.0001999801676946257,
-      "loss": 46.0,
-      "step": 39377
-    },
-    {
-      "epoch": 6.341559644108056,
-      "grad_norm": 0.0020738153252750635,
-      "learning_rate": 0.00019998016668708633,
-      "loss": 46.0,
-      "step": 39378
-    },
-    {
-      "epoch": 6.341720681186843,
-      "grad_norm": 0.00840769987553358,
-      "learning_rate": 0.00019998016567952136,
-      "loss": 46.0,
-      "step": 39379
-    },
-    {
-      "epoch": 6.341881718265631,
-      "grad_norm": 0.0038362813647836447,
-      "learning_rate": 0.00019998016467193076,
-      "loss": 46.0,
-      "step": 39380
-    },
-    {
-      "epoch": 6.342042755344418,
-      "grad_norm": 0.009474311023950577,
-      "learning_rate": 0.0001999801636643146,
-      "loss": 46.0,
-      "step": 39381
-    },
-    {
-      "epoch": 6.342203792423206,
-      "grad_norm": 0.007288126274943352,
-      "learning_rate": 0.00019998016265667285,
-      "loss": 46.0,
-      "step": 39382
-    },
-    {
-      "epoch": 6.342364829501993,
-      "grad_norm": 0.008051365613937378,
-      "learning_rate": 0.00019998016164900553,
-      "loss": 46.0,
-      "step": 39383
-    },
-    {
-      "epoch": 6.342525866580781,
-      "grad_norm": 0.005412889178842306,
-      "learning_rate": 0.00019998016064131258,
-      "loss": 46.0,
-      "step": 39384
-    },
-    {
-      "epoch": 6.342686903659567,
-      "grad_norm": 0.005608964711427689,
-      "learning_rate": 0.00019998015963359404,
-      "loss": 46.0,
-      "step": 39385
-    },
-    {
-      "epoch": 6.342847940738355,
-      "grad_norm": 0.005815706681460142,
-      "learning_rate": 0.00019998015862584994,
-      "loss": 46.0,
-      "step": 39386
-    },
-    {
-      "epoch": 6.343008977817142,
-      "grad_norm": 0.0014363507507368922,
-      "learning_rate": 0.00019998015761808026,
-      "loss": 46.0,
-      "step": 39387
-    },
-    {
-      "epoch": 6.34317001489593,
-      "grad_norm": 0.00316497590392828,
-      "learning_rate": 0.00019998015661028495,
-      "loss": 46.0,
-      "step": 39388
-    },
-    {
-      "epoch": 6.343331051974717,
-      "grad_norm": 0.005221744999289513,
-      "learning_rate": 0.0001999801556024641,
-      "loss": 46.0,
-      "step": 39389
-    },
-    {
-      "epoch": 6.3434920890535045,
-      "grad_norm": 0.003934514243155718,
-      "learning_rate": 0.00019998015459461762,
-      "loss": 46.0,
-      "step": 39390
-    },
-    {
-      "epoch": 6.343653126132292,
-      "grad_norm": 0.0018451712094247341,
-      "learning_rate": 0.00019998015358674553,
-      "loss": 46.0,
-      "step": 39391
-    },
-    {
-      "epoch": 6.343814163211079,
-      "grad_norm": 0.030934782698750496,
-      "learning_rate": 0.00019998015257884788,
-      "loss": 46.0,
-      "step": 39392
-    },
-    {
-      "epoch": 6.343975200289867,
-      "grad_norm": 0.007448768708854914,
-      "learning_rate": 0.00019998015157092464,
-      "loss": 46.0,
-      "step": 39393
-    },
-    {
-      "epoch": 6.344136237368654,
-      "grad_norm": 0.001728096860460937,
-      "learning_rate": 0.00019998015056297582,
-      "loss": 46.0,
-      "step": 39394
-    },
-    {
-      "epoch": 6.344297274447442,
-      "grad_norm": 0.0027998778969049454,
-      "learning_rate": 0.0001999801495550014,
-      "loss": 46.0,
-      "step": 39395
-    },
-    {
-      "epoch": 6.344458311526229,
-      "grad_norm": 0.002442327793687582,
-      "learning_rate": 0.00019998014854700138,
-      "loss": 46.0,
-      "step": 39396
-    },
-    {
-      "epoch": 6.344619348605017,
-      "grad_norm": 0.005001716315746307,
-      "learning_rate": 0.0001999801475389758,
-      "loss": 46.0,
-      "step": 39397
-    },
-    {
-      "epoch": 6.344780385683804,
-      "grad_norm": 0.009232645854353905,
-      "learning_rate": 0.0001999801465309246,
-      "loss": 46.0,
-      "step": 39398
-    },
-    {
-      "epoch": 6.3449414227625915,
-      "grad_norm": 0.007224122527986765,
-      "learning_rate": 0.0001999801455228478,
-      "loss": 46.0,
-      "step": 39399
-    },
-    {
-      "epoch": 6.345102459841378,
-      "grad_norm": 0.004758620634675026,
-      "learning_rate": 0.00019998014451474543,
-      "loss": 46.0,
-      "step": 39400
-    },
-    {
-      "epoch": 6.345263496920166,
-      "grad_norm": 0.007991883903741837,
-      "learning_rate": 0.0001999801435066175,
-      "loss": 46.0,
-      "step": 39401
-    },
-    {
-      "epoch": 6.345424533998953,
-      "grad_norm": 0.0038770984392613173,
-      "learning_rate": 0.00019998014249846392,
-      "loss": 46.0,
-      "step": 39402
-    },
-    {
-      "epoch": 6.3455855710777405,
-      "grad_norm": 0.0015964569756761193,
-      "learning_rate": 0.00019998014149028479,
-      "loss": 46.0,
-      "step": 39403
-    },
-    {
-      "epoch": 6.345746608156528,
-      "grad_norm": 0.00449296273291111,
-      "learning_rate": 0.00019998014048208006,
-      "loss": 46.0,
-      "step": 39404
-    },
-    {
-      "epoch": 6.345907645235315,
-      "grad_norm": 0.0014050485333427787,
-      "learning_rate": 0.00019998013947384975,
-      "loss": 46.0,
-      "step": 39405
-    },
-    {
-      "epoch": 6.346068682314103,
-      "grad_norm": 0.008442426100373268,
-      "learning_rate": 0.00019998013846559386,
-      "loss": 46.0,
-      "step": 39406
-    },
-    {
-      "epoch": 6.34622971939289,
-      "grad_norm": 0.003294851863756776,
-      "learning_rate": 0.00019998013745731234,
-      "loss": 46.0,
-      "step": 39407
-    },
-    {
-      "epoch": 6.346390756471678,
-      "grad_norm": 0.0032214017119258642,
-      "learning_rate": 0.00019998013644900527,
-      "loss": 46.0,
-      "step": 39408
-    },
-    {
-      "epoch": 6.346551793550465,
-      "grad_norm": 0.0018650633282959461,
-      "learning_rate": 0.00019998013544067259,
-      "loss": 46.0,
-      "step": 39409
-    },
-    {
-      "epoch": 6.346712830629253,
-      "grad_norm": 0.004513530060648918,
-      "learning_rate": 0.0001999801344323143,
-      "loss": 46.0,
-      "step": 39410
-    },
-    {
-      "epoch": 6.34687386770804,
-      "grad_norm": 0.0024002757854759693,
-      "learning_rate": 0.00019998013342393045,
-      "loss": 46.0,
-      "step": 39411
-    },
-    {
-      "epoch": 6.3470349047868275,
-      "grad_norm": 0.006897768937051296,
-      "learning_rate": 0.000199980132415521,
-      "loss": 46.0,
-      "step": 39412
-    },
-    {
-      "epoch": 6.347195941865614,
-      "grad_norm": 0.021896084770560265,
-      "learning_rate": 0.00019998013140708597,
-      "loss": 46.0,
-      "step": 39413
-    },
-    {
-      "epoch": 6.3473569789444015,
-      "grad_norm": 0.012006690725684166,
-      "learning_rate": 0.00019998013039862532,
-      "loss": 46.0,
-      "step": 39414
-    },
-    {
-      "epoch": 6.347518016023189,
-      "grad_norm": 0.0020825478713959455,
-      "learning_rate": 0.00019998012939013914,
-      "loss": 46.0,
-      "step": 39415
-    },
-    {
-      "epoch": 6.347679053101976,
-      "grad_norm": 0.0020682106260210276,
-      "learning_rate": 0.00019998012838162732,
-      "loss": 46.0,
-      "step": 39416
-    },
-    {
-      "epoch": 6.347840090180764,
-      "grad_norm": 0.012185588479042053,
-      "learning_rate": 0.0001999801273730899,
-      "loss": 46.0,
-      "step": 39417
-    },
-    {
-      "epoch": 6.348001127259551,
-      "grad_norm": 0.001970587996765971,
-      "learning_rate": 0.00019998012636452694,
-      "loss": 46.0,
-      "step": 39418
-    },
-    {
-      "epoch": 6.348162164338339,
-      "grad_norm": 0.005669783800840378,
-      "learning_rate": 0.00019998012535593835,
-      "loss": 46.0,
-      "step": 39419
-    },
-    {
-      "epoch": 6.348323201417126,
-      "grad_norm": 0.005162478890269995,
-      "learning_rate": 0.00019998012434732418,
-      "loss": 46.0,
-      "step": 39420
-    },
-    {
-      "epoch": 6.348484238495914,
-      "grad_norm": 0.0027079936116933823,
-      "learning_rate": 0.00019998012333868445,
-      "loss": 46.0,
-      "step": 39421
-    },
-    {
-      "epoch": 6.348645275574701,
-      "grad_norm": 0.007538971025496721,
-      "learning_rate": 0.0001999801223300191,
-      "loss": 46.0,
-      "step": 39422
-    },
-    {
-      "epoch": 6.348806312653489,
-      "grad_norm": 0.002288009040057659,
-      "learning_rate": 0.00019998012132132814,
-      "loss": 46.0,
-      "step": 39423
-    },
-    {
-      "epoch": 6.348967349732276,
-      "grad_norm": 0.001883052522316575,
-      "learning_rate": 0.00019998012031261165,
-      "loss": 46.0,
-      "step": 39424
-    },
-    {
-      "epoch": 6.3491283868110635,
-      "grad_norm": 0.003149421652778983,
-      "learning_rate": 0.00019998011930386954,
-      "loss": 46.0,
-      "step": 39425
-    },
-    {
-      "epoch": 6.349289423889851,
-      "grad_norm": 0.0018784533021971583,
-      "learning_rate": 0.00019998011829510182,
-      "loss": 46.0,
-      "step": 39426
-    },
-    {
-      "epoch": 6.349450460968638,
-      "grad_norm": 0.002980217570438981,
-      "learning_rate": 0.00019998011728630853,
-      "loss": 46.0,
-      "step": 39427
-    },
-    {
-      "epoch": 6.349611498047425,
-      "grad_norm": 0.005143294110894203,
-      "learning_rate": 0.00019998011627748964,
-      "loss": 46.0,
-      "step": 39428
-    },
-    {
-      "epoch": 6.349772535126212,
-      "grad_norm": 0.0020320769399404526,
-      "learning_rate": 0.00019998011526864518,
-      "loss": 46.0,
-      "step": 39429
-    },
-    {
-      "epoch": 6.349933572205,
-      "grad_norm": 0.002204232383519411,
-      "learning_rate": 0.0001999801142597751,
-      "loss": 46.0,
-      "step": 39430
-    },
-    {
-      "epoch": 6.350094609283787,
-      "grad_norm": 0.016811968758702278,
-      "learning_rate": 0.00019998011325087945,
-      "loss": 46.0,
-      "step": 39431
-    },
-    {
-      "epoch": 6.350255646362575,
-      "grad_norm": 0.0065052201971411705,
-      "learning_rate": 0.00019998011224195823,
-      "loss": 46.0,
-      "step": 39432
-    },
-    {
-      "epoch": 6.350416683441362,
-      "grad_norm": 0.017053255811333656,
-      "learning_rate": 0.0001999801112330114,
-      "loss": 46.0,
-      "step": 39433
-    },
-    {
-      "epoch": 6.35057772052015,
-      "grad_norm": 0.002551349811255932,
-      "learning_rate": 0.00019998011022403898,
-      "loss": 46.0,
-      "step": 39434
-    },
-    {
-      "epoch": 6.350738757598937,
-      "grad_norm": 0.0014461820246651769,
-      "learning_rate": 0.00019998010921504092,
-      "loss": 46.0,
-      "step": 39435
-    },
-    {
-      "epoch": 6.350899794677725,
-      "grad_norm": 0.019337797537446022,
-      "learning_rate": 0.00019998010820601735,
-      "loss": 46.0,
-      "step": 39436
-    },
-    {
-      "epoch": 6.351060831756512,
-      "grad_norm": 0.003692854195833206,
-      "learning_rate": 0.00019998010719696815,
-      "loss": 46.0,
-      "step": 39437
-    },
-    {
-      "epoch": 6.3512218688352995,
-      "grad_norm": 0.008299127221107483,
-      "learning_rate": 0.00019998010618789338,
-      "loss": 46.0,
-      "step": 39438
-    },
-    {
-      "epoch": 6.351382905914087,
-      "grad_norm": 0.001452952390536666,
-      "learning_rate": 0.000199980105178793,
-      "loss": 46.0,
-      "step": 39439
-    },
-    {
-      "epoch": 6.351543942992874,
-      "grad_norm": 0.007712093181908131,
-      "learning_rate": 0.00019998010416966703,
-      "loss": 46.0,
-      "step": 39440
-    },
-    {
-      "epoch": 6.351704980071662,
-      "grad_norm": 0.010283350944519043,
-      "learning_rate": 0.00019998010316051547,
-      "loss": 46.0,
-      "step": 39441
-    },
-    {
-      "epoch": 6.351866017150449,
-      "grad_norm": 0.0124505078420043,
-      "learning_rate": 0.00019998010215133835,
-      "loss": 46.0,
-      "step": 39442
-    },
-    {
-      "epoch": 6.352027054229236,
-      "grad_norm": 0.016991138458251953,
-      "learning_rate": 0.00019998010114213562,
-      "loss": 46.0,
-      "step": 39443
-    },
-    {
-      "epoch": 6.352188091308023,
-      "grad_norm": 0.0026531871408224106,
-      "learning_rate": 0.0001999801001329073,
-      "loss": 46.0,
-      "step": 39444
-    },
-    {
-      "epoch": 6.352349128386811,
-      "grad_norm": 0.006383951287716627,
-      "learning_rate": 0.0001999800991236534,
-      "loss": 46.0,
-      "step": 39445
-    },
-    {
-      "epoch": 6.352510165465598,
-      "grad_norm": 0.007048714440315962,
-      "learning_rate": 0.00019998009811437388,
-      "loss": 46.0,
-      "step": 39446
-    },
-    {
-      "epoch": 6.352671202544386,
-      "grad_norm": 0.005653482396155596,
-      "learning_rate": 0.0001999800971050688,
-      "loss": 46.0,
-      "step": 39447
-    },
-    {
-      "epoch": 6.352832239623173,
-      "grad_norm": 0.0026763854548335075,
-      "learning_rate": 0.00019998009609573813,
-      "loss": 46.0,
-      "step": 39448
-    },
-    {
-      "epoch": 6.352993276701961,
-      "grad_norm": 0.003750694915652275,
-      "learning_rate": 0.00019998009508638185,
-      "loss": 46.0,
-      "step": 39449
-    },
-    {
-      "epoch": 6.353154313780748,
-      "grad_norm": 0.0018128384836018085,
-      "learning_rate": 0.000199980094077,
-      "loss": 46.0,
-      "step": 39450
-    },
-    {
-      "epoch": 6.3533153508595355,
-      "grad_norm": 0.0016510699642822146,
-      "learning_rate": 0.00019998009306759253,
-      "loss": 46.0,
-      "step": 39451
-    },
-    {
-      "epoch": 6.353476387938323,
-      "grad_norm": 0.014423185028135777,
-      "learning_rate": 0.0001999800920581595,
-      "loss": 46.0,
-      "step": 39452
-    },
-    {
-      "epoch": 6.35363742501711,
-      "grad_norm": 0.0018934386316686869,
-      "learning_rate": 0.00019998009104870088,
-      "loss": 46.0,
-      "step": 39453
-    },
-    {
-      "epoch": 6.353798462095898,
-      "grad_norm": 0.002123798942193389,
-      "learning_rate": 0.00019998009003921666,
-      "loss": 46.0,
-      "step": 39454
-    },
-    {
-      "epoch": 6.353959499174685,
-      "grad_norm": 0.004711429122835398,
-      "learning_rate": 0.00019998008902970686,
-      "loss": 46.0,
-      "step": 39455
-    },
-    {
-      "epoch": 6.354120536253473,
-      "grad_norm": 0.0009010710055008531,
-      "learning_rate": 0.00019998008802017147,
-      "loss": 46.0,
-      "step": 39456
-    },
-    {
-      "epoch": 6.35428157333226,
-      "grad_norm": 0.0024187364615499973,
-      "learning_rate": 0.00019998008701061046,
-      "loss": 46.0,
-      "step": 39457
-    },
-    {
-      "epoch": 6.354442610411047,
-      "grad_norm": 0.011012710630893707,
-      "learning_rate": 0.00019998008600102392,
-      "loss": 46.0,
-      "step": 39458
-    },
-    {
-      "epoch": 6.354603647489834,
-      "grad_norm": 0.003747508628293872,
-      "learning_rate": 0.00019998008499141174,
-      "loss": 46.0,
-      "step": 39459
-    },
-    {
-      "epoch": 6.354764684568622,
-      "grad_norm": 0.009693090803921223,
-      "learning_rate": 0.000199980083981774,
-      "loss": 46.0,
-      "step": 39460
-    },
-    {
-      "epoch": 6.354925721647409,
-      "grad_norm": 0.0014709543902426958,
-      "learning_rate": 0.00019998008297211062,
-      "loss": 46.0,
-      "step": 39461
-    },
-    {
-      "epoch": 6.3550867587261965,
-      "grad_norm": 0.003294868627563119,
-      "learning_rate": 0.0001999800819624217,
-      "loss": 46.0,
-      "step": 39462
-    },
-    {
-      "epoch": 6.355247795804984,
-      "grad_norm": 0.005450052209198475,
-      "learning_rate": 0.00019998008095270718,
-      "loss": 46.0,
-      "step": 39463
-    },
-    {
-      "epoch": 6.3554088328837715,
-      "grad_norm": 0.002521320478990674,
-      "learning_rate": 0.00019998007994296706,
-      "loss": 46.0,
-      "step": 39464
-    },
-    {
-      "epoch": 6.355569869962559,
-      "grad_norm": 0.0077145956456661224,
-      "learning_rate": 0.00019998007893320133,
-      "loss": 46.0,
-      "step": 39465
-    },
-    {
-      "epoch": 6.355730907041346,
-      "grad_norm": 0.002833607839420438,
-      "learning_rate": 0.00019998007792341004,
-      "loss": 46.0,
-      "step": 39466
-    },
-    {
-      "epoch": 6.355891944120134,
-      "grad_norm": 0.016394183039665222,
-      "learning_rate": 0.00019998007691359316,
-      "loss": 46.0,
-      "step": 39467
-    },
-    {
-      "epoch": 6.356052981198921,
-      "grad_norm": 0.012824208475649357,
-      "learning_rate": 0.0001999800759037507,
-      "loss": 46.0,
-      "step": 39468
-    },
-    {
-      "epoch": 6.356214018277709,
-      "grad_norm": 0.0016593087930232286,
-      "learning_rate": 0.00019998007489388262,
-      "loss": 46.0,
-      "step": 39469
-    },
-    {
-      "epoch": 6.356375055356496,
-      "grad_norm": 0.006278193090111017,
-      "learning_rate": 0.00019998007388398898,
-      "loss": 46.0,
-      "step": 39470
-    },
-    {
-      "epoch": 6.356536092435284,
-      "grad_norm": 0.006127243861556053,
-      "learning_rate": 0.00019998007287406973,
-      "loss": 46.0,
-      "step": 39471
-    },
-    {
-      "epoch": 6.356697129514071,
-      "grad_norm": 0.0033390542957931757,
-      "learning_rate": 0.0001999800718641249,
-      "loss": 46.0,
-      "step": 39472
-    },
-    {
-      "epoch": 6.356858166592858,
-      "grad_norm": 0.007892763242125511,
-      "learning_rate": 0.00019998007085415446,
-      "loss": 46.0,
-      "step": 39473
-    },
-    {
-      "epoch": 6.357019203671645,
-      "grad_norm": 0.008341191336512566,
-      "learning_rate": 0.00019998006984415847,
-      "loss": 46.0,
-      "step": 39474
-    },
-    {
-      "epoch": 6.3571802407504325,
-      "grad_norm": 0.0009239570936188102,
-      "learning_rate": 0.00019998006883413684,
-      "loss": 46.0,
-      "step": 39475
-    },
-    {
-      "epoch": 6.35734127782922,
-      "grad_norm": 0.002829850185662508,
-      "learning_rate": 0.00019998006782408968,
-      "loss": 46.0,
-      "step": 39476
-    },
-    {
-      "epoch": 6.357502314908007,
-      "grad_norm": 0.005340429488569498,
-      "learning_rate": 0.00019998006681401688,
-      "loss": 46.0,
-      "step": 39477
-    },
-    {
-      "epoch": 6.357663351986795,
-      "grad_norm": 0.0010472108842805028,
-      "learning_rate": 0.00019998006580391851,
-      "loss": 46.0,
-      "step": 39478
-    },
-    {
-      "epoch": 6.357824389065582,
-      "grad_norm": 0.002105482155457139,
-      "learning_rate": 0.00019998006479379456,
-      "loss": 46.0,
-      "step": 39479
-    },
-    {
-      "epoch": 6.35798542614437,
-      "grad_norm": 0.002203951356932521,
-      "learning_rate": 0.000199980063783645,
-      "loss": 46.0,
-      "step": 39480
-    },
-    {
-      "epoch": 6.358146463223157,
-      "grad_norm": 0.0024880466517060995,
-      "learning_rate": 0.00019998006277346985,
-      "loss": 46.0,
-      "step": 39481
-    },
-    {
-      "epoch": 6.358307500301945,
-      "grad_norm": 0.0020039756782352924,
-      "learning_rate": 0.00019998006176326913,
-      "loss": 46.0,
-      "step": 39482
-    },
-    {
-      "epoch": 6.358468537380732,
-      "grad_norm": 0.022428035736083984,
-      "learning_rate": 0.0001999800607530428,
-      "loss": 46.0,
-      "step": 39483
-    },
-    {
-      "epoch": 6.35862957445952,
-      "grad_norm": 0.0015898382989689708,
-      "learning_rate": 0.0001999800597427909,
-      "loss": 46.0,
-      "step": 39484
-    },
-    {
-      "epoch": 6.358790611538307,
-      "grad_norm": 0.007732732221484184,
-      "learning_rate": 0.0001999800587325134,
-      "loss": 46.0,
-      "step": 39485
-    },
-    {
-      "epoch": 6.3589516486170945,
-      "grad_norm": 0.002958668163046241,
-      "learning_rate": 0.0001999800577222103,
-      "loss": 46.0,
-      "step": 39486
-    },
-    {
-      "epoch": 6.359112685695881,
-      "grad_norm": 0.009270669892430305,
-      "learning_rate": 0.00019998005671188163,
-      "loss": 46.0,
-      "step": 39487
-    },
-    {
-      "epoch": 6.3592737227746685,
-      "grad_norm": 0.003521893871948123,
-      "learning_rate": 0.00019998005570152737,
-      "loss": 46.0,
-      "step": 39488
-    },
-    {
-      "epoch": 6.359434759853456,
-      "grad_norm": 0.007394652813673019,
-      "learning_rate": 0.0001999800546911475,
-      "loss": 46.0,
-      "step": 39489
-    },
-    {
-      "epoch": 6.359595796932243,
-      "grad_norm": 0.0035296056885272264,
-      "learning_rate": 0.00019998005368074205,
-      "loss": 46.0,
-      "step": 39490
-    },
-    {
-      "epoch": 6.359756834011031,
-      "grad_norm": 0.0020039433147758245,
-      "learning_rate": 0.000199980052670311,
-      "loss": 46.0,
-      "step": 39491
-    },
-    {
-      "epoch": 6.359917871089818,
-      "grad_norm": 0.013090742751955986,
-      "learning_rate": 0.0001999800516598544,
-      "loss": 46.0,
-      "step": 39492
-    },
-    {
-      "epoch": 6.360078908168606,
-      "grad_norm": 0.00434614485129714,
-      "learning_rate": 0.00019998005064937217,
-      "loss": 46.0,
-      "step": 39493
-    },
-    {
-      "epoch": 6.360239945247393,
-      "grad_norm": 0.005496986676007509,
-      "learning_rate": 0.00019998004963886436,
-      "loss": 46.0,
-      "step": 39494
-    },
-    {
-      "epoch": 6.360400982326181,
-      "grad_norm": 0.002192819956690073,
-      "learning_rate": 0.00019998004862833096,
-      "loss": 46.0,
-      "step": 39495
-    },
-    {
-      "epoch": 6.360562019404968,
-      "grad_norm": 0.02494894340634346,
-      "learning_rate": 0.00019998004761777197,
-      "loss": 46.0,
-      "step": 39496
-    },
-    {
-      "epoch": 6.360723056483756,
-      "grad_norm": 0.006052169483155012,
-      "learning_rate": 0.0001999800466071874,
-      "loss": 46.0,
-      "step": 39497
-    },
-    {
-      "epoch": 6.360884093562543,
-      "grad_norm": 0.016569942235946655,
-      "learning_rate": 0.00019998004559657723,
-      "loss": 46.0,
-      "step": 39498
-    },
-    {
-      "epoch": 6.3610451306413305,
-      "grad_norm": 0.005270159803330898,
-      "learning_rate": 0.00019998004458594146,
-      "loss": 46.0,
-      "step": 39499
-    },
-    {
-      "epoch": 6.361206167720118,
-      "grad_norm": 0.004607352893799543,
-      "learning_rate": 0.0001999800435752801,
-      "loss": 46.0,
-      "step": 39500
-    },
-    {
-      "epoch": 6.3613672047989045,
-      "grad_norm": 0.0014795513125136495,
-      "learning_rate": 0.0001999800425645932,
-      "loss": 46.0,
-      "step": 39501
-    },
-    {
-      "epoch": 6.361528241877692,
-      "grad_norm": 0.00941571407020092,
-      "learning_rate": 0.00019998004155388066,
-      "loss": 46.0,
-      "step": 39502
-    },
-    {
-      "epoch": 6.361689278956479,
-      "grad_norm": 0.0019849573727697134,
-      "learning_rate": 0.00019998004054314254,
-      "loss": 46.0,
-      "step": 39503
-    },
-    {
-      "epoch": 6.361850316035267,
-      "grad_norm": 0.004644973669201136,
-      "learning_rate": 0.00019998003953237886,
-      "loss": 46.0,
-      "step": 39504
-    },
-    {
-      "epoch": 6.362011353114054,
-      "grad_norm": 0.0060126567259430885,
-      "learning_rate": 0.00019998003852158953,
-      "loss": 46.0,
-      "step": 39505
-    },
-    {
-      "epoch": 6.362172390192842,
-      "grad_norm": 0.006208830047398806,
-      "learning_rate": 0.00019998003751077464,
-      "loss": 46.0,
-      "step": 39506
-    },
-    {
-      "epoch": 6.362333427271629,
-      "grad_norm": 0.004187299869954586,
-      "learning_rate": 0.00019998003649993417,
-      "loss": 46.0,
-      "step": 39507
-    },
-    {
-      "epoch": 6.362494464350417,
-      "grad_norm": 0.003422250971198082,
-      "learning_rate": 0.00019998003548906808,
-      "loss": 46.0,
-      "step": 39508
-    },
-    {
-      "epoch": 6.362655501429204,
-      "grad_norm": 0.003779014805331826,
-      "learning_rate": 0.00019998003447817646,
-      "loss": 46.0,
-      "step": 39509
-    },
-    {
-      "epoch": 6.362816538507992,
-      "grad_norm": 0.0063432808965444565,
-      "learning_rate": 0.0001999800334672592,
-      "loss": 46.0,
-      "step": 39510
-    },
-    {
-      "epoch": 6.362977575586779,
-      "grad_norm": 0.0014192431699484587,
-      "learning_rate": 0.00019998003245631638,
-      "loss": 46.0,
-      "step": 39511
-    },
-    {
-      "epoch": 6.3631386126655665,
-      "grad_norm": 0.001706940121948719,
-      "learning_rate": 0.00019998003144534797,
-      "loss": 46.0,
-      "step": 39512
-    },
-    {
-      "epoch": 6.363299649744354,
-      "grad_norm": 0.01604827307164669,
-      "learning_rate": 0.00019998003043435395,
-      "loss": 46.0,
-      "step": 39513
-    },
-    {
-      "epoch": 6.363460686823141,
-      "grad_norm": 0.013224104419350624,
-      "learning_rate": 0.00019998002942333434,
-      "loss": 46.0,
-      "step": 39514
-    },
-    {
-      "epoch": 6.363621723901929,
-      "grad_norm": 0.004361344501376152,
-      "learning_rate": 0.00019998002841228914,
-      "loss": 46.0,
-      "step": 39515
-    },
-    {
-      "epoch": 6.363782760980715,
-      "grad_norm": 0.0021434673108160496,
-      "learning_rate": 0.00019998002740121839,
-      "loss": 46.0,
-      "step": 39516
-    },
-    {
-      "epoch": 6.363943798059503,
-      "grad_norm": 0.002538310829550028,
-      "learning_rate": 0.000199980026390122,
-      "loss": 46.0,
-      "step": 39517
-    },
-    {
-      "epoch": 6.36410483513829,
-      "grad_norm": 0.0021049438510090113,
-      "learning_rate": 0.00019998002537900003,
-      "loss": 46.0,
-      "step": 39518
-    },
-    {
-      "epoch": 6.364265872217078,
-      "grad_norm": 0.002298015169799328,
-      "learning_rate": 0.00019998002436785246,
-      "loss": 46.0,
-      "step": 39519
-    },
-    {
-      "epoch": 6.364426909295865,
-      "grad_norm": 0.0041793943382799625,
-      "learning_rate": 0.00019998002335667935,
-      "loss": 46.0,
-      "step": 39520
-    },
-    {
-      "epoch": 6.364587946374653,
-      "grad_norm": 0.0015891504008322954,
-      "learning_rate": 0.0001999800223454806,
-      "loss": 46.0,
-      "step": 39521
-    },
-    {
-      "epoch": 6.36474898345344,
-      "grad_norm": 0.010709723457694054,
-      "learning_rate": 0.0001999800213342563,
-      "loss": 46.0,
-      "step": 39522
-    },
-    {
-      "epoch": 6.3649100205322275,
-      "grad_norm": 0.0045154886320233345,
-      "learning_rate": 0.00019998002032300635,
-      "loss": 46.0,
-      "step": 39523
-    },
-    {
-      "epoch": 6.365071057611015,
-      "grad_norm": 0.005094493739306927,
-      "learning_rate": 0.00019998001931173086,
-      "loss": 46.0,
-      "step": 39524
-    },
-    {
-      "epoch": 6.3652320946898024,
-      "grad_norm": 0.013623693957924843,
-      "learning_rate": 0.00019998001830042977,
-      "loss": 46.0,
-      "step": 39525
-    },
-    {
-      "epoch": 6.36539313176859,
-      "grad_norm": 0.010770655237138271,
-      "learning_rate": 0.00019998001728910308,
-      "loss": 46.0,
-      "step": 39526
-    },
-    {
-      "epoch": 6.365554168847377,
-      "grad_norm": 0.004940268117934465,
-      "learning_rate": 0.00019998001627775081,
-      "loss": 46.0,
-      "step": 39527
-    },
-    {
-      "epoch": 6.365715205926165,
-      "grad_norm": 0.0054796659387648106,
-      "learning_rate": 0.00019998001526637298,
-      "loss": 46.0,
-      "step": 39528
-    },
-    {
-      "epoch": 6.365876243004952,
-      "grad_norm": 0.008328541181981564,
-      "learning_rate": 0.00019998001425496954,
-      "loss": 46.0,
-      "step": 39529
-    },
-    {
-      "epoch": 6.36603728008374,
-      "grad_norm": 0.01888740248978138,
-      "learning_rate": 0.00019998001324354048,
-      "loss": 46.0,
-      "step": 39530
-    },
-    {
-      "epoch": 6.366198317162526,
-      "grad_norm": 0.00396219827234745,
-      "learning_rate": 0.00019998001223208583,
-      "loss": 46.0,
-      "step": 39531
-    },
-    {
-      "epoch": 6.366359354241314,
-      "grad_norm": 0.012622925452888012,
-      "learning_rate": 0.0001999800112206056,
-      "loss": 46.0,
-      "step": 39532
-    },
-    {
-      "epoch": 6.366520391320101,
-      "grad_norm": 0.015233221463859081,
-      "learning_rate": 0.00019998001020909983,
-      "loss": 46.0,
-      "step": 39533
-    },
-    {
-      "epoch": 6.366681428398889,
-      "grad_norm": 0.00618281913921237,
-      "learning_rate": 0.0001999800091975684,
-      "loss": 46.0,
-      "step": 39534
-    },
-    {
-      "epoch": 6.366842465477676,
-      "grad_norm": 0.004907556343823671,
-      "learning_rate": 0.00019998000818601143,
-      "loss": 46.0,
-      "step": 39535
-    },
-    {
-      "epoch": 6.3670035025564635,
-      "grad_norm": 0.0021349135786294937,
-      "learning_rate": 0.00019998000717442885,
-      "loss": 46.0,
-      "step": 39536
-    },
-    {
-      "epoch": 6.367164539635251,
-      "grad_norm": 0.0018690564902499318,
-      "learning_rate": 0.00019998000616282068,
-      "loss": 46.0,
-      "step": 39537
-    },
-    {
-      "epoch": 6.367325576714038,
-      "grad_norm": 0.003193999407812953,
-      "learning_rate": 0.00019998000515118692,
-      "loss": 46.0,
-      "step": 39538
-    },
-    {
-      "epoch": 6.367486613792826,
-      "grad_norm": 0.002504838164895773,
-      "learning_rate": 0.00019998000413952758,
-      "loss": 46.0,
-      "step": 39539
-    },
-    {
-      "epoch": 6.367647650871613,
-      "grad_norm": 0.0014109468320384622,
-      "learning_rate": 0.00019998000312784265,
-      "loss": 46.0,
-      "step": 39540
-    },
-    {
-      "epoch": 6.367808687950401,
-      "grad_norm": 0.000765151169616729,
-      "learning_rate": 0.00019998000211613213,
-      "loss": 46.0,
-      "step": 39541
-    },
-    {
-      "epoch": 6.367969725029188,
-      "grad_norm": 0.0011773803271353245,
-      "learning_rate": 0.000199980001104396,
-      "loss": 46.0,
-      "step": 39542
-    },
-    {
-      "epoch": 6.368130762107976,
-      "grad_norm": 0.0008933040080592036,
-      "learning_rate": 0.00019998000009263428,
-      "loss": 46.0,
-      "step": 39543
-    },
-    {
-      "epoch": 6.368291799186763,
-      "grad_norm": 0.0034444809425622225,
-      "learning_rate": 0.000199979999080847,
-      "loss": 46.0,
-      "step": 39544
-    },
-    {
-      "epoch": 6.368452836265551,
-      "grad_norm": 0.0028496696613729,
-      "learning_rate": 0.0001999799980690341,
-      "loss": 46.0,
-      "step": 39545
-    },
-    {
-      "epoch": 6.368613873344337,
-      "grad_norm": 0.01163305714726448,
-      "learning_rate": 0.00019997999705719563,
-      "loss": 46.0,
-      "step": 39546
-    },
-    {
-      "epoch": 6.368774910423125,
-      "grad_norm": 0.0006827443139627576,
-      "learning_rate": 0.00019997999604533159,
-      "loss": 46.0,
-      "step": 39547
-    },
-    {
-      "epoch": 6.368935947501912,
-      "grad_norm": 0.003109041368588805,
-      "learning_rate": 0.0001999799950334419,
-      "loss": 46.0,
-      "step": 39548
-    },
-    {
-      "epoch": 6.3690969845806995,
-      "grad_norm": 0.006627863738685846,
-      "learning_rate": 0.00019997999402152666,
-      "loss": 46.0,
-      "step": 39549
-    },
-    {
-      "epoch": 6.369258021659487,
-      "grad_norm": 0.011682174168527126,
-      "learning_rate": 0.00019997999300958583,
-      "loss": 46.0,
-      "step": 39550
-    },
-    {
-      "epoch": 6.369419058738274,
-      "grad_norm": 0.0022269426845014095,
-      "learning_rate": 0.00019997999199761941,
-      "loss": 46.0,
-      "step": 39551
-    },
-    {
-      "epoch": 6.369580095817062,
-      "grad_norm": 0.020668867975473404,
-      "learning_rate": 0.00019997999098562738,
-      "loss": 46.0,
-      "step": 39552
-    },
-    {
-      "epoch": 6.369741132895849,
-      "grad_norm": 0.0013435936998575926,
-      "learning_rate": 0.0001999799899736098,
-      "loss": 46.0,
-      "step": 39553
-    },
-    {
-      "epoch": 6.369902169974637,
-      "grad_norm": 0.005933010950684547,
-      "learning_rate": 0.0001999799889615666,
-      "loss": 46.0,
-      "step": 39554
-    },
-    {
-      "epoch": 6.370063207053424,
-      "grad_norm": 0.006973384879529476,
-      "learning_rate": 0.00019997998794949782,
-      "loss": 46.0,
-      "step": 39555
-    },
-    {
-      "epoch": 6.370224244132212,
-      "grad_norm": 0.005510346964001656,
-      "learning_rate": 0.00019997998693740344,
-      "loss": 46.0,
-      "step": 39556
-    },
-    {
-      "epoch": 6.370385281210999,
-      "grad_norm": 0.004914383869618177,
-      "learning_rate": 0.00019997998592528348,
-      "loss": 46.0,
-      "step": 39557
-    },
-    {
-      "epoch": 6.370546318289787,
-      "grad_norm": 0.009859933517873287,
-      "learning_rate": 0.00019997998491313795,
-      "loss": 46.0,
-      "step": 39558
-    },
-    {
-      "epoch": 6.370707355368574,
-      "grad_norm": 0.012067588977515697,
-      "learning_rate": 0.00019997998390096678,
-      "loss": 46.0,
-      "step": 39559
-    },
-    {
-      "epoch": 6.3708683924473615,
-      "grad_norm": 0.005373584106564522,
-      "learning_rate": 0.00019997998288877008,
-      "loss": 46.0,
-      "step": 39560
-    },
-    {
-      "epoch": 6.371029429526148,
-      "grad_norm": 0.009342162869870663,
-      "learning_rate": 0.0001999799818765477,
-      "loss": 46.0,
-      "step": 39561
-    },
-    {
-      "epoch": 6.3711904666049355,
-      "grad_norm": 0.00786909181624651,
-      "learning_rate": 0.00019997998086429983,
-      "loss": 46.0,
-      "step": 39562
-    },
-    {
-      "epoch": 6.371351503683723,
-      "grad_norm": 0.00831774715334177,
-      "learning_rate": 0.00019997997985202632,
-      "loss": 46.0,
-      "step": 39563
-    },
-    {
-      "epoch": 6.37151254076251,
-      "grad_norm": 0.006877412088215351,
-      "learning_rate": 0.00019997997883972724,
-      "loss": 46.0,
-      "step": 39564
-    },
-    {
-      "epoch": 6.371673577841298,
-      "grad_norm": 0.02035493776202202,
-      "learning_rate": 0.00019997997782740255,
-      "loss": 46.0,
-      "step": 39565
-    },
-    {
-      "epoch": 6.371834614920085,
-      "grad_norm": 0.0028133559972047806,
-      "learning_rate": 0.00019997997681505227,
-      "loss": 46.0,
-      "step": 39566
-    },
-    {
-      "epoch": 6.371995651998873,
-      "grad_norm": 0.014005298726260662,
-      "learning_rate": 0.0001999799758026764,
-      "loss": 46.0,
-      "step": 39567
-    },
-    {
-      "epoch": 6.37215668907766,
-      "grad_norm": 0.0031042341142892838,
-      "learning_rate": 0.00019997997479027495,
-      "loss": 46.0,
-      "step": 39568
-    },
-    {
-      "epoch": 6.372317726156448,
-      "grad_norm": 0.0029636898543685675,
-      "learning_rate": 0.0001999799737778479,
-      "loss": 46.0,
-      "step": 39569
-    },
-    {
-      "epoch": 6.372478763235235,
-      "grad_norm": 0.0021732032764703035,
-      "learning_rate": 0.0001999799727653953,
-      "loss": 46.0,
-      "step": 39570
-    },
-    {
-      "epoch": 6.3726398003140226,
-      "grad_norm": 0.010055347345769405,
-      "learning_rate": 0.00019997997175291707,
-      "loss": 46.0,
-      "step": 39571
-    },
-    {
-      "epoch": 6.37280083739281,
-      "grad_norm": 0.001629504607990384,
-      "learning_rate": 0.00019997997074041327,
-      "loss": 46.0,
-      "step": 39572
-    },
-    {
-      "epoch": 6.3729618744715975,
-      "grad_norm": 0.0034732099156826735,
-      "learning_rate": 0.00019997996972788385,
-      "loss": 46.0,
-      "step": 39573
-    },
-    {
-      "epoch": 6.373122911550384,
-      "grad_norm": 0.0012738791992887855,
-      "learning_rate": 0.0001999799687153289,
-      "loss": 46.0,
-      "step": 39574
-    },
-    {
-      "epoch": 6.3732839486291715,
-      "grad_norm": 0.008240816183388233,
-      "learning_rate": 0.0001999799677027483,
-      "loss": 46.0,
-      "step": 39575
-    },
-    {
-      "epoch": 6.373444985707959,
-      "grad_norm": 0.002950426656752825,
-      "learning_rate": 0.0001999799666901421,
-      "loss": 46.0,
-      "step": 39576
-    },
-    {
-      "epoch": 6.373606022786746,
-      "grad_norm": 0.01027581375092268,
-      "learning_rate": 0.00019997996567751037,
-      "loss": 46.0,
-      "step": 39577
-    },
-    {
-      "epoch": 6.373767059865534,
-      "grad_norm": 0.014655972830951214,
-      "learning_rate": 0.00019997996466485302,
-      "loss": 46.0,
-      "step": 39578
-    },
-    {
-      "epoch": 6.373928096944321,
-      "grad_norm": 0.00414480222389102,
-      "learning_rate": 0.00019997996365217008,
-      "loss": 46.0,
-      "step": 39579
-    },
-    {
-      "epoch": 6.374089134023109,
-      "grad_norm": 0.0041672587394714355,
-      "learning_rate": 0.00019997996263946158,
-      "loss": 46.0,
-      "step": 39580
-    },
-    {
-      "epoch": 6.374250171101896,
-      "grad_norm": 0.01014675386250019,
-      "learning_rate": 0.00019997996162672744,
-      "loss": 46.0,
-      "step": 39581
-    },
-    {
-      "epoch": 6.374411208180684,
-      "grad_norm": 0.008969432674348354,
-      "learning_rate": 0.00019997996061396774,
-      "loss": 46.0,
-      "step": 39582
-    },
-    {
-      "epoch": 6.374572245259471,
-      "grad_norm": 0.002961366903036833,
-      "learning_rate": 0.00019997995960118243,
-      "loss": 46.0,
-      "step": 39583
-    },
-    {
-      "epoch": 6.3747332823382585,
-      "grad_norm": 0.003803764469921589,
-      "learning_rate": 0.00019997995858837158,
-      "loss": 46.0,
-      "step": 39584
-    },
-    {
-      "epoch": 6.374894319417046,
-      "grad_norm": 0.0049678306095302105,
-      "learning_rate": 0.0001999799575755351,
-      "loss": 46.0,
-      "step": 39585
-    },
-    {
-      "epoch": 6.375055356495833,
-      "grad_norm": 0.008770765736699104,
-      "learning_rate": 0.00019997995656267304,
-      "loss": 46.0,
-      "step": 39586
-    },
-    {
-      "epoch": 6.375216393574621,
-      "grad_norm": 0.0038183191791176796,
-      "learning_rate": 0.00019997995554978535,
-      "loss": 46.0,
-      "step": 39587
-    },
-    {
-      "epoch": 6.375377430653408,
-      "grad_norm": 0.008817486464977264,
-      "learning_rate": 0.00019997995453687213,
-      "loss": 46.0,
-      "step": 39588
-    },
-    {
-      "epoch": 6.375538467732195,
-      "grad_norm": 0.004258319735527039,
-      "learning_rate": 0.0001999799535239333,
-      "loss": 46.0,
-      "step": 39589
-    },
-    {
-      "epoch": 6.375699504810982,
-      "grad_norm": 0.0024216873571276665,
-      "learning_rate": 0.00019997995251096887,
-      "loss": 46.0,
-      "step": 39590
-    },
-    {
-      "epoch": 6.37586054188977,
-      "grad_norm": 0.0012228693813085556,
-      "learning_rate": 0.00019997995149797885,
-      "loss": 46.0,
-      "step": 39591
-    },
-    {
-      "epoch": 6.376021578968557,
-      "grad_norm": 0.0012588078388944268,
-      "learning_rate": 0.00019997995048496323,
-      "loss": 46.0,
-      "step": 39592
-    },
-    {
-      "epoch": 6.376182616047345,
-      "grad_norm": 0.013182600028812885,
-      "learning_rate": 0.00019997994947192204,
-      "loss": 46.0,
-      "step": 39593
-    },
-    {
-      "epoch": 6.376343653126132,
-      "grad_norm": 0.0024391391780227423,
-      "learning_rate": 0.00019997994845885527,
-      "loss": 46.0,
-      "step": 39594
-    },
-    {
-      "epoch": 6.37650469020492,
-      "grad_norm": 0.0009883253369480371,
-      "learning_rate": 0.0001999799474457629,
-      "loss": 46.0,
-      "step": 39595
-    },
-    {
-      "epoch": 6.376665727283707,
-      "grad_norm": 0.0027416495140641928,
-      "learning_rate": 0.00019997994643264493,
-      "loss": 46.0,
-      "step": 39596
-    },
-    {
-      "epoch": 6.3768267643624945,
-      "grad_norm": 0.01627153903245926,
-      "learning_rate": 0.00019997994541950137,
-      "loss": 46.0,
-      "step": 39597
-    },
-    {
-      "epoch": 6.376987801441282,
-      "grad_norm": 0.008926328271627426,
-      "learning_rate": 0.00019997994440633222,
-      "loss": 46.0,
-      "step": 39598
-    },
-    {
-      "epoch": 6.377148838520069,
-      "grad_norm": 0.005098787136375904,
-      "learning_rate": 0.0001999799433931375,
-      "loss": 46.0,
-      "step": 39599
-    },
-    {
-      "epoch": 6.377309875598857,
-      "grad_norm": 0.00829018373042345,
-      "learning_rate": 0.00019997994237991716,
-      "loss": 46.0,
-      "step": 39600
-    },
-    {
-      "epoch": 6.377470912677644,
-      "grad_norm": 0.003187085734680295,
-      "learning_rate": 0.00019997994136667125,
-      "loss": 46.0,
-      "step": 39601
-    },
-    {
-      "epoch": 6.377631949756432,
-      "grad_norm": 0.02408532053232193,
-      "learning_rate": 0.00019997994035339978,
-      "loss": 46.0,
-      "step": 39602
-    },
-    {
-      "epoch": 6.377792986835219,
-      "grad_norm": 0.0139978202059865,
-      "learning_rate": 0.0001999799393401027,
-      "loss": 46.0,
-      "step": 39603
-    },
-    {
-      "epoch": 6.377954023914006,
-      "grad_norm": 0.0023415570612996817,
-      "learning_rate": 0.00019997993832678,
-      "loss": 46.0,
-      "step": 39604
-    },
-    {
-      "epoch": 6.378115060992793,
-      "grad_norm": 0.007179120555520058,
-      "learning_rate": 0.0001999799373134317,
-      "loss": 46.0,
-      "step": 39605
-    },
-    {
-      "epoch": 6.378276098071581,
-      "grad_norm": 0.008959531784057617,
-      "learning_rate": 0.00019997993630005786,
-      "loss": 46.0,
-      "step": 39606
-    },
-    {
-      "epoch": 6.378437135150368,
-      "grad_norm": 0.001087911194190383,
-      "learning_rate": 0.00019997993528665843,
-      "loss": 46.0,
-      "step": 39607
-    },
-    {
-      "epoch": 6.378598172229156,
-      "grad_norm": 0.008189628832042217,
-      "learning_rate": 0.00019997993427323338,
-      "loss": 46.0,
-      "step": 39608
-    },
-    {
-      "epoch": 6.378759209307943,
-      "grad_norm": 0.0056917001493275166,
-      "learning_rate": 0.00019997993325978274,
-      "loss": 46.0,
-      "step": 39609
-    },
-    {
-      "epoch": 6.3789202463867305,
-      "grad_norm": 0.005968928802758455,
-      "learning_rate": 0.00019997993224630652,
-      "loss": 46.0,
-      "step": 39610
-    },
-    {
-      "epoch": 6.379081283465518,
-      "grad_norm": 0.00679363077506423,
-      "learning_rate": 0.0001999799312328047,
-      "loss": 46.0,
-      "step": 39611
-    },
-    {
-      "epoch": 6.379242320544305,
-      "grad_norm": 0.0041407993994653225,
-      "learning_rate": 0.00019997993021927732,
-      "loss": 46.0,
-      "step": 39612
-    },
-    {
-      "epoch": 6.379403357623093,
-      "grad_norm": 0.00477386312559247,
-      "learning_rate": 0.00019997992920572433,
-      "loss": 46.0,
-      "step": 39613
-    },
-    {
-      "epoch": 6.37956439470188,
-      "grad_norm": 0.02691279724240303,
-      "learning_rate": 0.00019997992819214576,
-      "loss": 46.0,
-      "step": 39614
-    },
-    {
-      "epoch": 6.379725431780668,
-      "grad_norm": 0.004952750168740749,
-      "learning_rate": 0.0001999799271785416,
-      "loss": 46.0,
-      "step": 39615
-    },
-    {
-      "epoch": 6.379886468859455,
-      "grad_norm": 0.014814638532698154,
-      "learning_rate": 0.00019997992616491183,
-      "loss": 46.0,
-      "step": 39616
-    },
-    {
-      "epoch": 6.380047505938243,
-      "grad_norm": 0.0016894518630579114,
-      "learning_rate": 0.0001999799251512565,
-      "loss": 46.0,
-      "step": 39617
-    },
-    {
-      "epoch": 6.38020854301703,
-      "grad_norm": 0.0017963474383577704,
-      "learning_rate": 0.00019997992413757552,
-      "loss": 46.0,
-      "step": 39618
-    },
-    {
-      "epoch": 6.380369580095817,
-      "grad_norm": 0.0059068575501441956,
-      "learning_rate": 0.00019997992312386902,
-      "loss": 46.0,
-      "step": 39619
-    },
-    {
-      "epoch": 6.380530617174604,
-      "grad_norm": 0.008470118045806885,
-      "learning_rate": 0.00019997992211013692,
-      "loss": 46.0,
-      "step": 39620
-    },
-    {
-      "epoch": 6.380691654253392,
-      "grad_norm": 0.0018899007700383663,
-      "learning_rate": 0.0001999799210963792,
-      "loss": 46.0,
-      "step": 39621
-    },
-    {
-      "epoch": 6.380852691332179,
-      "grad_norm": 0.001159351202659309,
-      "learning_rate": 0.0001999799200825959,
-      "loss": 46.0,
-      "step": 39622
-    },
-    {
-      "epoch": 6.3810137284109665,
-      "grad_norm": 0.007207023445516825,
-      "learning_rate": 0.000199979919068787,
-      "loss": 46.0,
-      "step": 39623
-    },
-    {
-      "epoch": 6.381174765489754,
-      "grad_norm": 0.004233041312545538,
-      "learning_rate": 0.00019997991805495257,
-      "loss": 46.0,
-      "step": 39624
-    },
-    {
-      "epoch": 6.381335802568541,
-      "grad_norm": 0.00571990804746747,
-      "learning_rate": 0.00019997991704109248,
-      "loss": 46.0,
-      "step": 39625
-    },
-    {
-      "epoch": 6.381496839647329,
-      "grad_norm": 0.002765171229839325,
-      "learning_rate": 0.0001999799160272068,
-      "loss": 46.0,
-      "step": 39626
-    },
-    {
-      "epoch": 6.381657876726116,
-      "grad_norm": 0.007528841495513916,
-      "learning_rate": 0.00019997991501329558,
-      "loss": 46.0,
-      "step": 39627
-    },
-    {
-      "epoch": 6.381818913804904,
-      "grad_norm": 0.005357755813747644,
-      "learning_rate": 0.00019997991399935873,
-      "loss": 46.0,
-      "step": 39628
-    },
-    {
-      "epoch": 6.381979950883691,
-      "grad_norm": 0.004060770384967327,
-      "learning_rate": 0.0001999799129853963,
-      "loss": 46.0,
-      "step": 39629
-    },
-    {
-      "epoch": 6.382140987962479,
-      "grad_norm": 0.008046452887356281,
-      "learning_rate": 0.0001999799119714083,
-      "loss": 46.0,
-      "step": 39630
-    },
-    {
-      "epoch": 6.382302025041266,
-      "grad_norm": 0.006763724144548178,
-      "learning_rate": 0.0001999799109573947,
-      "loss": 46.0,
-      "step": 39631
-    },
-    {
-      "epoch": 6.3824630621200535,
-      "grad_norm": 0.006087682209908962,
-      "learning_rate": 0.0001999799099433555,
-      "loss": 46.0,
-      "step": 39632
-    },
-    {
-      "epoch": 6.382624099198841,
-      "grad_norm": 0.003705762792378664,
-      "learning_rate": 0.00019997990892929072,
-      "loss": 46.0,
-      "step": 39633
-    },
-    {
-      "epoch": 6.382785136277628,
-      "grad_norm": 0.006714996881783009,
-      "learning_rate": 0.00019997990791520035,
-      "loss": 46.0,
-      "step": 39634
-    },
-    {
-      "epoch": 6.382946173356415,
-      "grad_norm": 0.005154167767614126,
-      "learning_rate": 0.0001999799069010844,
-      "loss": 46.0,
-      "step": 39635
-    },
-    {
-      "epoch": 6.3831072104352025,
-      "grad_norm": 0.00879506766796112,
-      "learning_rate": 0.00019997990588694283,
-      "loss": 46.0,
-      "step": 39636
-    },
-    {
-      "epoch": 6.38326824751399,
-      "grad_norm": 0.022557592019438744,
-      "learning_rate": 0.0001999799048727757,
-      "loss": 46.0,
-      "step": 39637
-    },
-    {
-      "epoch": 6.383429284592777,
-      "grad_norm": 0.0012223473750054836,
-      "learning_rate": 0.00019997990385858295,
-      "loss": 46.0,
-      "step": 39638
-    },
-    {
-      "epoch": 6.383590321671565,
-      "grad_norm": 0.0034364170860499144,
-      "learning_rate": 0.00019997990284436465,
-      "loss": 46.0,
-      "step": 39639
-    },
-    {
-      "epoch": 6.383751358750352,
-      "grad_norm": 0.014220552518963814,
-      "learning_rate": 0.00019997990183012073,
-      "loss": 46.0,
-      "step": 39640
-    },
-    {
-      "epoch": 6.38391239582914,
-      "grad_norm": 0.0046723405830562115,
-      "learning_rate": 0.00019997990081585122,
-      "loss": 46.0,
-      "step": 39641
-    },
-    {
-      "epoch": 6.384073432907927,
-      "grad_norm": 0.007212279364466667,
-      "learning_rate": 0.00019997989980155613,
-      "loss": 46.0,
-      "step": 39642
-    },
-    {
-      "epoch": 6.384234469986715,
-      "grad_norm": 0.002519486704841256,
-      "learning_rate": 0.00019997989878723548,
-      "loss": 46.0,
-      "step": 39643
-    },
-    {
-      "epoch": 6.384395507065502,
-      "grad_norm": 0.0023943139240145683,
-      "learning_rate": 0.00019997989777288918,
-      "loss": 46.0,
-      "step": 39644
-    },
-    {
-      "epoch": 6.3845565441442895,
-      "grad_norm": 0.0026246444322168827,
-      "learning_rate": 0.00019997989675851733,
-      "loss": 46.0,
-      "step": 39645
-    },
-    {
-      "epoch": 6.384717581223077,
-      "grad_norm": 0.01296649593859911,
-      "learning_rate": 0.0001999798957441199,
-      "loss": 46.0,
-      "step": 39646
-    },
-    {
-      "epoch": 6.3848786183018635,
-      "grad_norm": 0.004096482414752245,
-      "learning_rate": 0.00019997989472969686,
-      "loss": 46.0,
-      "step": 39647
-    },
-    {
-      "epoch": 6.385039655380651,
-      "grad_norm": 0.00932546891272068,
-      "learning_rate": 0.00019997989371524821,
-      "loss": 46.0,
-      "step": 39648
-    },
-    {
-      "epoch": 6.385200692459438,
-      "grad_norm": 0.005613727029412985,
-      "learning_rate": 0.00019997989270077398,
-      "loss": 46.0,
-      "step": 39649
-    },
-    {
-      "epoch": 6.385361729538226,
-      "grad_norm": 0.004359958693385124,
-      "learning_rate": 0.0001999798916862742,
-      "loss": 46.0,
-      "step": 39650
-    },
-    {
-      "epoch": 6.385522766617013,
-      "grad_norm": 0.0027462400030344725,
-      "learning_rate": 0.00019997989067174882,
-      "loss": 46.0,
-      "step": 39651
-    },
-    {
-      "epoch": 6.385683803695801,
-      "grad_norm": 0.0024416567757725716,
-      "learning_rate": 0.0001999798896571978,
-      "loss": 46.0,
-      "step": 39652
-    },
-    {
-      "epoch": 6.385844840774588,
-      "grad_norm": 0.007831353694200516,
-      "learning_rate": 0.00019997988864262122,
-      "loss": 46.0,
-      "step": 39653
-    },
-    {
-      "epoch": 6.386005877853376,
-      "grad_norm": 0.01293005608022213,
-      "learning_rate": 0.00019997988762801905,
-      "loss": 46.0,
-      "step": 39654
-    },
-    {
-      "epoch": 6.386166914932163,
-      "grad_norm": 0.013343794271349907,
-      "learning_rate": 0.0001999798866133913,
-      "loss": 46.0,
-      "step": 39655
-    },
-    {
-      "epoch": 6.386327952010951,
-      "grad_norm": 0.002792490879073739,
-      "learning_rate": 0.00019997988559873796,
-      "loss": 46.0,
-      "step": 39656
-    },
-    {
-      "epoch": 6.386488989089738,
-      "grad_norm": 0.005933757405728102,
-      "learning_rate": 0.00019997988458405903,
-      "loss": 46.0,
-      "step": 39657
-    },
-    {
-      "epoch": 6.3866500261685255,
-      "grad_norm": 0.0017501752590760589,
-      "learning_rate": 0.00019997988356935449,
-      "loss": 46.0,
-      "step": 39658
-    },
-    {
-      "epoch": 6.386811063247313,
-      "grad_norm": 0.002971403067931533,
-      "learning_rate": 0.00019997988255462436,
-      "loss": 46.0,
-      "step": 39659
-    },
-    {
-      "epoch": 6.3869721003261,
-      "grad_norm": 0.006849338300526142,
-      "learning_rate": 0.00019997988153986867,
-      "loss": 46.0,
-      "step": 39660
-    },
-    {
-      "epoch": 6.387133137404888,
-      "grad_norm": 0.008996205404400826,
-      "learning_rate": 0.0001999798805250874,
-      "loss": 46.0,
-      "step": 39661
-    },
-    {
-      "epoch": 6.387294174483674,
-      "grad_norm": 0.005194155965000391,
-      "learning_rate": 0.00019997987951028047,
-      "loss": 46.0,
-      "step": 39662
-    },
-    {
-      "epoch": 6.387455211562462,
-      "grad_norm": 0.0013714199885725975,
-      "learning_rate": 0.00019997987849544802,
-      "loss": 46.0,
-      "step": 39663
-    },
-    {
-      "epoch": 6.387616248641249,
-      "grad_norm": 0.009962350130081177,
-      "learning_rate": 0.00019997987748058996,
-      "loss": 46.0,
-      "step": 39664
-    },
-    {
-      "epoch": 6.387777285720037,
-      "grad_norm": 0.004367387853562832,
-      "learning_rate": 0.00019997987646570628,
-      "loss": 46.0,
-      "step": 39665
-    },
-    {
-      "epoch": 6.387938322798824,
-      "grad_norm": 0.008136908523738384,
-      "learning_rate": 0.00019997987545079706,
-      "loss": 46.0,
-      "step": 39666
-    },
-    {
-      "epoch": 6.388099359877612,
-      "grad_norm": 0.0021178254391998053,
-      "learning_rate": 0.0001999798744358622,
-      "loss": 46.0,
-      "step": 39667
-    },
-    {
-      "epoch": 6.388260396956399,
-      "grad_norm": 0.020958187058568,
-      "learning_rate": 0.0001999798734209018,
-      "loss": 46.0,
-      "step": 39668
-    },
-    {
-      "epoch": 6.388421434035187,
-      "grad_norm": 0.0017117172246798873,
-      "learning_rate": 0.00019997987240591577,
-      "loss": 46.0,
-      "step": 39669
-    },
-    {
-      "epoch": 6.388582471113974,
-      "grad_norm": 0.004571861587464809,
-      "learning_rate": 0.0001999798713909042,
-      "loss": 46.0,
-      "step": 39670
-    },
-    {
-      "epoch": 6.3887435081927615,
-      "grad_norm": 0.006385928951203823,
-      "learning_rate": 0.00019997987037586698,
-      "loss": 46.0,
-      "step": 39671
-    },
-    {
-      "epoch": 6.388904545271549,
-      "grad_norm": 0.026783538982272148,
-      "learning_rate": 0.0001999798693608042,
-      "loss": 46.0,
-      "step": 39672
-    },
-    {
-      "epoch": 6.389065582350336,
-      "grad_norm": 0.0008000983507372439,
-      "learning_rate": 0.00019997986834571584,
-      "loss": 46.0,
-      "step": 39673
-    },
-    {
-      "epoch": 6.389226619429124,
-      "grad_norm": 0.0019062068313360214,
-      "learning_rate": 0.00019997986733060187,
-      "loss": 46.0,
-      "step": 39674
-    },
-    {
-      "epoch": 6.389387656507911,
-      "grad_norm": 0.01616768166422844,
-      "learning_rate": 0.00019997986631546232,
-      "loss": 46.0,
-      "step": 39675
-    },
-    {
-      "epoch": 6.389548693586699,
-      "grad_norm": 0.010500124655663967,
-      "learning_rate": 0.00019997986530029719,
-      "loss": 46.0,
-      "step": 39676
-    },
-    {
-      "epoch": 6.389709730665485,
-      "grad_norm": 0.004461308009922504,
-      "learning_rate": 0.00019997986428510646,
-      "loss": 46.0,
-      "step": 39677
-    },
-    {
-      "epoch": 6.389870767744273,
-      "grad_norm": 0.001676552346907556,
-      "learning_rate": 0.00019997986326989012,
-      "loss": 46.0,
-      "step": 39678
-    },
-    {
-      "epoch": 6.39003180482306,
-      "grad_norm": 0.00176837551407516,
-      "learning_rate": 0.00019997986225464822,
-      "loss": 46.0,
-      "step": 39679
-    },
-    {
-      "epoch": 6.390192841901848,
-      "grad_norm": 0.004148111212998629,
-      "learning_rate": 0.00019997986123938073,
-      "loss": 46.0,
-      "step": 39680
-    },
-    {
-      "epoch": 6.390353878980635,
-      "grad_norm": 0.0026615133974701166,
-      "learning_rate": 0.00019997986022408763,
-      "loss": 46.0,
-      "step": 39681
-    },
-    {
-      "epoch": 6.390514916059423,
-      "grad_norm": 0.0034285816363990307,
-      "learning_rate": 0.00019997985920876897,
-      "loss": 46.0,
-      "step": 39682
-    },
-    {
-      "epoch": 6.39067595313821,
-      "grad_norm": 0.007380757946521044,
-      "learning_rate": 0.0001999798581934247,
-      "loss": 46.0,
-      "step": 39683
-    },
-    {
-      "epoch": 6.3908369902169975,
-      "grad_norm": 0.0022231696639209986,
-      "learning_rate": 0.00019997985717805483,
-      "loss": 46.0,
-      "step": 39684
-    },
-    {
-      "epoch": 6.390998027295785,
-      "grad_norm": 0.003195648081600666,
-      "learning_rate": 0.0001999798561626594,
-      "loss": 46.0,
-      "step": 39685
-    },
-    {
-      "epoch": 6.391159064374572,
-      "grad_norm": 0.0016613835468888283,
-      "learning_rate": 0.00019997985514723834,
-      "loss": 46.0,
-      "step": 39686
-    },
-    {
-      "epoch": 6.39132010145336,
-      "grad_norm": 0.002278456464409828,
-      "learning_rate": 0.0001999798541317917,
-      "loss": 46.0,
-      "step": 39687
-    },
-    {
-      "epoch": 6.391481138532147,
-      "grad_norm": 0.001075371983461082,
-      "learning_rate": 0.0001999798531163195,
-      "loss": 46.0,
-      "step": 39688
-    },
-    {
-      "epoch": 6.391642175610935,
-      "grad_norm": 0.014691442251205444,
-      "learning_rate": 0.00019997985210082168,
-      "loss": 46.0,
-      "step": 39689
-    },
-    {
-      "epoch": 6.391803212689722,
-      "grad_norm": 0.009903491474688053,
-      "learning_rate": 0.0001999798510852983,
-      "loss": 46.0,
-      "step": 39690
-    },
-    {
-      "epoch": 6.39196424976851,
-      "grad_norm": 0.0012265342520549893,
-      "learning_rate": 0.0001999798500697493,
-      "loss": 46.0,
-      "step": 39691
-    },
-    {
-      "epoch": 6.392125286847296,
-      "grad_norm": 0.00846842210739851,
-      "learning_rate": 0.00019997984905417474,
-      "loss": 46.0,
-      "step": 39692
-    },
-    {
-      "epoch": 6.392286323926084,
-      "grad_norm": 0.004979741759598255,
-      "learning_rate": 0.00019997984803857456,
-      "loss": 46.0,
-      "step": 39693
-    },
-    {
-      "epoch": 6.392447361004871,
-      "grad_norm": 0.00590496463701129,
-      "learning_rate": 0.0001999798470229488,
-      "loss": 46.0,
-      "step": 39694
-    },
-    {
-      "epoch": 6.3926083980836585,
-      "grad_norm": 0.009984750300645828,
-      "learning_rate": 0.00019997984600729748,
-      "loss": 46.0,
-      "step": 39695
-    },
-    {
-      "epoch": 6.392769435162446,
-      "grad_norm": 0.01124607678502798,
-      "learning_rate": 0.00019997984499162052,
-      "loss": 46.0,
-      "step": 39696
-    },
-    {
-      "epoch": 6.3929304722412335,
-      "grad_norm": 0.002240577945485711,
-      "learning_rate": 0.000199979843975918,
-      "loss": 46.0,
-      "step": 39697
-    },
-    {
-      "epoch": 6.393091509320021,
-      "grad_norm": 0.013211602345108986,
-      "learning_rate": 0.00019997984296018988,
-      "loss": 46.0,
-      "step": 39698
-    },
-    {
-      "epoch": 6.393252546398808,
-      "grad_norm": 0.00550662400200963,
-      "learning_rate": 0.00019997984194443618,
-      "loss": 46.0,
-      "step": 39699
-    },
-    {
-      "epoch": 6.393413583477596,
-      "grad_norm": 0.00861713569611311,
-      "learning_rate": 0.0001999798409286569,
-      "loss": 46.0,
-      "step": 39700
-    },
-    {
-      "epoch": 6.393574620556383,
-      "grad_norm": 0.004874045494943857,
-      "learning_rate": 0.000199979839912852,
-      "loss": 46.0,
-      "step": 39701
-    },
-    {
-      "epoch": 6.393735657635171,
-      "grad_norm": 0.007115269545465708,
-      "learning_rate": 0.00019997983889702154,
-      "loss": 46.0,
-      "step": 39702
-    },
-    {
-      "epoch": 6.393896694713958,
-      "grad_norm": 0.011768237687647343,
-      "learning_rate": 0.00019997983788116547,
-      "loss": 46.0,
-      "step": 39703
-    },
-    {
-      "epoch": 6.394057731792746,
-      "grad_norm": 0.0038908589631319046,
-      "learning_rate": 0.00019997983686528384,
-      "loss": 46.0,
-      "step": 39704
-    },
-    {
-      "epoch": 6.394218768871533,
-      "grad_norm": 0.0010088979033753276,
-      "learning_rate": 0.00019997983584937656,
-      "loss": 46.0,
-      "step": 39705
-    },
-    {
-      "epoch": 6.3943798059503205,
-      "grad_norm": 0.0034306878224015236,
-      "learning_rate": 0.00019997983483344375,
-      "loss": 46.0,
-      "step": 39706
-    },
-    {
-      "epoch": 6.394540843029107,
-      "grad_norm": 0.0018780187238007784,
-      "learning_rate": 0.00019997983381748533,
-      "loss": 46.0,
-      "step": 39707
-    },
-    {
-      "epoch": 6.3947018801078945,
-      "grad_norm": 0.003269060980528593,
-      "learning_rate": 0.00019997983280150132,
-      "loss": 46.0,
-      "step": 39708
-    },
-    {
-      "epoch": 6.394862917186682,
-      "grad_norm": 0.0015705307014286518,
-      "learning_rate": 0.0001999798317854917,
-      "loss": 46.0,
-      "step": 39709
-    },
-    {
-      "epoch": 6.395023954265469,
-      "grad_norm": 0.005281734745949507,
-      "learning_rate": 0.00019997983076945651,
-      "loss": 46.0,
-      "step": 39710
-    },
-    {
-      "epoch": 6.395184991344257,
-      "grad_norm": 0.0012669767020270228,
-      "learning_rate": 0.00019997982975339574,
-      "loss": 46.0,
-      "step": 39711
-    },
-    {
-      "epoch": 6.395346028423044,
-      "grad_norm": 0.006262979935854673,
-      "learning_rate": 0.00019997982873730936,
-      "loss": 46.0,
-      "step": 39712
-    },
-    {
-      "epoch": 6.395507065501832,
-      "grad_norm": 0.002279779640957713,
-      "learning_rate": 0.00019997982772119741,
-      "loss": 46.0,
-      "step": 39713
-    },
-    {
-      "epoch": 6.395668102580619,
-      "grad_norm": 0.0008431622409261763,
-      "learning_rate": 0.00019997982670505985,
-      "loss": 46.0,
-      "step": 39714
-    },
-    {
-      "epoch": 6.395829139659407,
-      "grad_norm": 0.004218118730932474,
-      "learning_rate": 0.00019997982568889674,
-      "loss": 46.0,
-      "step": 39715
-    },
-    {
-      "epoch": 6.395990176738194,
-      "grad_norm": 0.00430978974327445,
-      "learning_rate": 0.000199979824672708,
-      "loss": 46.0,
-      "step": 39716
-    },
-    {
-      "epoch": 6.396151213816982,
-      "grad_norm": 0.002378294011577964,
-      "learning_rate": 0.00019997982365649368,
-      "loss": 46.0,
-      "step": 39717
-    },
-    {
-      "epoch": 6.396312250895769,
-      "grad_norm": 0.003344185883179307,
-      "learning_rate": 0.00019997982264025375,
-      "loss": 46.0,
-      "step": 39718
-    },
-    {
-      "epoch": 6.3964732879745565,
-      "grad_norm": 0.0019271261990070343,
-      "learning_rate": 0.00019997982162398828,
-      "loss": 46.0,
-      "step": 39719
-    },
-    {
-      "epoch": 6.396634325053343,
-      "grad_norm": 0.008766323328018188,
-      "learning_rate": 0.00019997982060769717,
-      "loss": 46.0,
-      "step": 39720
-    },
-    {
-      "epoch": 6.3967953621321305,
-      "grad_norm": 0.0019945718813687563,
-      "learning_rate": 0.0001999798195913805,
-      "loss": 46.0,
-      "step": 39721
-    },
-    {
-      "epoch": 6.396956399210918,
-      "grad_norm": 0.014293872751295567,
-      "learning_rate": 0.00019997981857503821,
-      "loss": 46.0,
-      "step": 39722
-    },
-    {
-      "epoch": 6.397117436289705,
-      "grad_norm": 0.008086971007287502,
-      "learning_rate": 0.00019997981755867037,
-      "loss": 46.0,
-      "step": 39723
-    },
-    {
-      "epoch": 6.397278473368493,
-      "grad_norm": 0.012430197559297085,
-      "learning_rate": 0.0001999798165422769,
-      "loss": 46.0,
-      "step": 39724
-    },
-    {
-      "epoch": 6.39743951044728,
-      "grad_norm": 0.0034204518888145685,
-      "learning_rate": 0.00019997981552585787,
-      "loss": 46.0,
-      "step": 39725
-    },
-    {
-      "epoch": 6.397600547526068,
-      "grad_norm": 0.01530166994780302,
-      "learning_rate": 0.00019997981450941326,
-      "loss": 46.0,
-      "step": 39726
-    },
-    {
-      "epoch": 6.397761584604855,
-      "grad_norm": 0.005163129884749651,
-      "learning_rate": 0.00019997981349294307,
-      "loss": 46.0,
-      "step": 39727
-    },
-    {
-      "epoch": 6.397922621683643,
-      "grad_norm": 0.0057404255494475365,
-      "learning_rate": 0.00019997981247644724,
-      "loss": 46.0,
-      "step": 39728
-    },
-    {
-      "epoch": 6.39808365876243,
-      "grad_norm": 0.004381768871098757,
-      "learning_rate": 0.00019997981145992584,
-      "loss": 46.0,
-      "step": 39729
-    },
-    {
-      "epoch": 6.398244695841218,
-      "grad_norm": 0.009724602103233337,
-      "learning_rate": 0.00019997981044337886,
-      "loss": 46.0,
-      "step": 39730
-    },
-    {
-      "epoch": 6.398405732920005,
-      "grad_norm": 0.015879547223448753,
-      "learning_rate": 0.0001999798094268063,
-      "loss": 46.0,
-      "step": 39731
-    },
-    {
-      "epoch": 6.3985667699987925,
-      "grad_norm": 0.006917417980730534,
-      "learning_rate": 0.0001999798084102081,
-      "loss": 46.0,
-      "step": 39732
-    },
-    {
-      "epoch": 6.39872780707758,
-      "grad_norm": 0.00795740820467472,
-      "learning_rate": 0.00019997980739358434,
-      "loss": 46.0,
-      "step": 39733
-    },
-    {
-      "epoch": 6.398888844156367,
-      "grad_norm": 0.0063113742507994175,
-      "learning_rate": 0.000199979806376935,
-      "loss": 46.0,
-      "step": 39734
-    },
-    {
-      "epoch": 6.399049881235154,
-      "grad_norm": 0.0043883859179914,
-      "learning_rate": 0.0001999798053602601,
-      "loss": 46.0,
-      "step": 39735
-    },
-    {
-      "epoch": 6.399210918313941,
-      "grad_norm": 0.0016365100163966417,
-      "learning_rate": 0.00019997980434355956,
-      "loss": 46.0,
-      "step": 39736
-    },
-    {
-      "epoch": 6.399371955392729,
-      "grad_norm": 0.005349626764655113,
-      "learning_rate": 0.00019997980332683347,
-      "loss": 46.0,
-      "step": 39737
-    },
-    {
-      "epoch": 6.399532992471516,
-      "grad_norm": 0.024173546582460403,
-      "learning_rate": 0.00019997980231008177,
-      "loss": 46.0,
-      "step": 39738
-    },
-    {
-      "epoch": 6.399694029550304,
-      "grad_norm": 0.0015852623619139194,
-      "learning_rate": 0.00019997980129330447,
-      "loss": 46.0,
-      "step": 39739
-    },
-    {
-      "epoch": 6.399855066629091,
-      "grad_norm": 0.0010524904355406761,
-      "learning_rate": 0.00019997980027650157,
-      "loss": 46.0,
-      "step": 39740
-    },
-    {
-      "epoch": 6.400016103707879,
-      "grad_norm": 0.0010647393064573407,
-      "learning_rate": 0.00019997979925967313,
-      "loss": 46.0,
-      "step": 39741
-    },
-    {
-      "epoch": 6.400177140786666,
-      "grad_norm": 0.02489626407623291,
-      "learning_rate": 0.00019997979824281904,
-      "loss": 46.0,
-      "step": 39742
-    },
-    {
-      "epoch": 6.400338177865454,
-      "grad_norm": 0.010901028290390968,
-      "learning_rate": 0.0001999797972259394,
-      "loss": 46.0,
-      "step": 39743
-    },
-    {
-      "epoch": 6.400499214944241,
-      "grad_norm": 0.0010789092630147934,
-      "learning_rate": 0.00019997979620903415,
-      "loss": 46.0,
-      "step": 39744
-    },
-    {
-      "epoch": 6.4006602520230285,
-      "grad_norm": 0.004068130627274513,
-      "learning_rate": 0.00019997979519210333,
-      "loss": 46.0,
-      "step": 39745
-    },
-    {
-      "epoch": 6.400821289101816,
-      "grad_norm": 0.001501579419709742,
-      "learning_rate": 0.0001999797941751469,
-      "loss": 46.0,
-      "step": 39746
-    },
-    {
-      "epoch": 6.400982326180603,
-      "grad_norm": 0.0024193590506911278,
-      "learning_rate": 0.00019997979315816488,
-      "loss": 46.0,
-      "step": 39747
-    },
-    {
-      "epoch": 6.401143363259391,
-      "grad_norm": 0.005171597935259342,
-      "learning_rate": 0.0001999797921411573,
-      "loss": 46.0,
-      "step": 39748
-    },
-    {
-      "epoch": 6.401304400338178,
-      "grad_norm": 0.003315950045362115,
-      "learning_rate": 0.0001999797911241241,
-      "loss": 46.0,
-      "step": 39749
-    },
-    {
-      "epoch": 6.401465437416965,
-      "grad_norm": 0.0035181797575205564,
-      "learning_rate": 0.00019997979010706534,
-      "loss": 46.0,
-      "step": 39750
-    },
-    {
-      "epoch": 6.401626474495752,
-      "grad_norm": 0.004503350239247084,
-      "learning_rate": 0.00019997978908998097,
-      "loss": 46.0,
-      "step": 39751
-    },
-    {
-      "epoch": 6.40178751157454,
-      "grad_norm": 0.004382152575999498,
-      "learning_rate": 0.000199979788072871,
-      "loss": 46.0,
-      "step": 39752
-    },
-    {
-      "epoch": 6.401948548653327,
-      "grad_norm": 0.000647125591058284,
-      "learning_rate": 0.00019997978705573542,
-      "loss": 46.0,
-      "step": 39753
-    },
-    {
-      "epoch": 6.402109585732115,
-      "grad_norm": 0.007898072712123394,
-      "learning_rate": 0.0001999797860385743,
-      "loss": 46.0,
-      "step": 39754
-    },
-    {
-      "epoch": 6.402270622810902,
-      "grad_norm": 0.0011270168470218778,
-      "learning_rate": 0.00019997978502138758,
-      "loss": 46.0,
-      "step": 39755
-    },
-    {
-      "epoch": 6.4024316598896895,
-      "grad_norm": 0.01105918362736702,
-      "learning_rate": 0.00019997978400417528,
-      "loss": 46.0,
-      "step": 39756
-    },
-    {
-      "epoch": 6.402592696968477,
-      "grad_norm": 0.004094884265214205,
-      "learning_rate": 0.00019997978298693736,
-      "loss": 46.0,
-      "step": 39757
-    },
-    {
-      "epoch": 6.4027537340472644,
-      "grad_norm": 0.018097953870892525,
-      "learning_rate": 0.00019997978196967386,
-      "loss": 46.0,
-      "step": 39758
-    },
-    {
-      "epoch": 6.402914771126052,
-      "grad_norm": 0.0019281319109722972,
-      "learning_rate": 0.00019997978095238477,
-      "loss": 46.0,
-      "step": 39759
-    },
-    {
-      "epoch": 6.403075808204839,
-      "grad_norm": 0.0011446558637544513,
-      "learning_rate": 0.0001999797799350701,
-      "loss": 46.0,
-      "step": 39760
-    },
-    {
-      "epoch": 6.403236845283627,
-      "grad_norm": 0.004522460047155619,
-      "learning_rate": 0.00019997977891772983,
-      "loss": 46.0,
-      "step": 39761
-    },
-    {
-      "epoch": 6.403397882362414,
-      "grad_norm": 0.002561889123171568,
-      "learning_rate": 0.00019997977790036397,
-      "loss": 46.0,
-      "step": 39762
-    },
-    {
-      "epoch": 6.403558919441202,
-      "grad_norm": 0.0010753024835139513,
-      "learning_rate": 0.00019997977688297254,
-      "loss": 46.0,
-      "step": 39763
-    },
-    {
-      "epoch": 6.403719956519989,
-      "grad_norm": 0.01704479567706585,
-      "learning_rate": 0.0001999797758655555,
-      "loss": 46.0,
-      "step": 39764
-    },
-    {
-      "epoch": 6.403880993598776,
-      "grad_norm": 0.00651383213698864,
-      "learning_rate": 0.00019997977484811287,
-      "loss": 46.0,
-      "step": 39765
-    },
-    {
-      "epoch": 6.404042030677563,
-      "grad_norm": 0.005670061334967613,
-      "learning_rate": 0.00019997977383064467,
-      "loss": 46.0,
-      "step": 39766
-    },
-    {
-      "epoch": 6.404203067756351,
-      "grad_norm": 0.007568455766886473,
-      "learning_rate": 0.00019997977281315085,
-      "loss": 46.0,
-      "step": 39767
-    },
-    {
-      "epoch": 6.404364104835138,
-      "grad_norm": 0.003849287983030081,
-      "learning_rate": 0.00019997977179563145,
-      "loss": 46.0,
-      "step": 39768
-    },
-    {
-      "epoch": 6.4045251419139255,
-      "grad_norm": 0.0026378552429378033,
-      "learning_rate": 0.00019997977077808646,
-      "loss": 46.0,
-      "step": 39769
-    },
-    {
-      "epoch": 6.404686178992713,
-      "grad_norm": 0.00456943828612566,
-      "learning_rate": 0.0001999797697605159,
-      "loss": 46.0,
-      "step": 39770
-    },
-    {
-      "epoch": 6.4048472160715,
-      "grad_norm": 0.001680535264313221,
-      "learning_rate": 0.00019997976874291972,
-      "loss": 46.0,
-      "step": 39771
-    },
-    {
-      "epoch": 6.405008253150288,
-      "grad_norm": 0.014071165584027767,
-      "learning_rate": 0.00019997976772529797,
-      "loss": 46.0,
-      "step": 39772
-    },
-    {
-      "epoch": 6.405169290229075,
-      "grad_norm": 0.0013729599304497242,
-      "learning_rate": 0.00019997976670765063,
-      "loss": 46.0,
-      "step": 39773
-    },
-    {
-      "epoch": 6.405330327307863,
-      "grad_norm": 0.014479031786322594,
-      "learning_rate": 0.0001999797656899777,
-      "loss": 46.0,
-      "step": 39774
-    },
-    {
-      "epoch": 6.40549136438665,
-      "grad_norm": 0.0012981047620996833,
-      "learning_rate": 0.0001999797646722792,
-      "loss": 46.0,
-      "step": 39775
-    },
-    {
-      "epoch": 6.405652401465438,
-      "grad_norm": 0.003990803845226765,
-      "learning_rate": 0.0001999797636545551,
-      "loss": 46.0,
-      "step": 39776
-    },
-    {
-      "epoch": 6.405813438544225,
-      "grad_norm": 0.006770376581698656,
-      "learning_rate": 0.00019997976263680538,
-      "loss": 46.0,
-      "step": 39777
-    },
-    {
-      "epoch": 6.405974475623013,
-      "grad_norm": 0.014279181137681007,
-      "learning_rate": 0.00019997976161903008,
-      "loss": 46.0,
-      "step": 39778
-    },
-    {
-      "epoch": 6.4061355127018,
-      "grad_norm": 0.014250132255256176,
-      "learning_rate": 0.0001999797606012292,
-      "loss": 46.0,
-      "step": 39779
-    },
-    {
-      "epoch": 6.406296549780587,
-      "grad_norm": 0.002729963744059205,
-      "learning_rate": 0.00019997975958340275,
-      "loss": 46.0,
-      "step": 39780
-    },
-    {
-      "epoch": 6.406457586859374,
-      "grad_norm": 0.003831196343526244,
-      "learning_rate": 0.00019997975856555068,
-      "loss": 46.0,
-      "step": 39781
-    },
-    {
-      "epoch": 6.4066186239381615,
-      "grad_norm": 0.0032163283322006464,
-      "learning_rate": 0.00019997975754767303,
-      "loss": 46.0,
-      "step": 39782
-    },
-    {
-      "epoch": 6.406779661016949,
-      "grad_norm": 0.005565740168094635,
-      "learning_rate": 0.0001999797565297698,
-      "loss": 46.0,
-      "step": 39783
-    },
-    {
-      "epoch": 6.406940698095736,
-      "grad_norm": 0.0036747378762811422,
-      "learning_rate": 0.00019997975551184097,
-      "loss": 46.0,
-      "step": 39784
-    },
-    {
-      "epoch": 6.407101735174524,
-      "grad_norm": 0.006656508427113295,
-      "learning_rate": 0.00019997975449388653,
-      "loss": 46.0,
-      "step": 39785
-    },
-    {
-      "epoch": 6.407262772253311,
-      "grad_norm": 0.00356505555100739,
-      "learning_rate": 0.00019997975347590654,
-      "loss": 46.0,
-      "step": 39786
-    },
-    {
-      "epoch": 6.407423809332099,
-      "grad_norm": 0.001133436686359346,
-      "learning_rate": 0.00019997975245790092,
-      "loss": 46.0,
-      "step": 39787
-    },
-    {
-      "epoch": 6.407584846410886,
-      "grad_norm": 0.0012598708271980286,
-      "learning_rate": 0.00019997975143986975,
-      "loss": 46.0,
-      "step": 39788
-    },
-    {
-      "epoch": 6.407745883489674,
-      "grad_norm": 0.008294720202684402,
-      "learning_rate": 0.00019997975042181296,
-      "loss": 46.0,
-      "step": 39789
-    },
-    {
-      "epoch": 6.407906920568461,
-      "grad_norm": 0.010146544314920902,
-      "learning_rate": 0.00019997974940373062,
-      "loss": 46.0,
-      "step": 39790
-    },
-    {
-      "epoch": 6.408067957647249,
-      "grad_norm": 0.001885155332274735,
-      "learning_rate": 0.00019997974838562266,
-      "loss": 46.0,
-      "step": 39791
-    },
-    {
-      "epoch": 6.408228994726036,
-      "grad_norm": 0.00269310618750751,
-      "learning_rate": 0.0001999797473674891,
-      "loss": 46.0,
-      "step": 39792
-    },
-    {
-      "epoch": 6.4083900318048235,
-      "grad_norm": 0.008755331858992577,
-      "learning_rate": 0.00019997974634932997,
-      "loss": 46.0,
-      "step": 39793
-    },
-    {
-      "epoch": 6.40855106888361,
-      "grad_norm": 0.002146894345059991,
-      "learning_rate": 0.00019997974533114525,
-      "loss": 46.0,
-      "step": 39794
-    },
-    {
-      "epoch": 6.4087121059623975,
-      "grad_norm": 0.008886881172657013,
-      "learning_rate": 0.00019997974431293494,
-      "loss": 46.0,
-      "step": 39795
-    },
-    {
-      "epoch": 6.408873143041185,
-      "grad_norm": 0.001966224517673254,
-      "learning_rate": 0.00019997974329469904,
-      "loss": 46.0,
-      "step": 39796
-    },
-    {
-      "epoch": 6.409034180119972,
-      "grad_norm": 0.008079875260591507,
-      "learning_rate": 0.00019997974227643753,
-      "loss": 46.0,
-      "step": 39797
-    },
-    {
-      "epoch": 6.40919521719876,
-      "grad_norm": 0.0045585534535348415,
-      "learning_rate": 0.00019997974125815043,
-      "loss": 46.0,
-      "step": 39798
-    },
-    {
-      "epoch": 6.409356254277547,
-      "grad_norm": 0.00700960960239172,
-      "learning_rate": 0.00019997974023983777,
-      "loss": 46.0,
-      "step": 39799
-    },
-    {
-      "epoch": 6.409517291356335,
-      "grad_norm": 0.01103118248283863,
-      "learning_rate": 0.0001999797392214995,
-      "loss": 46.0,
-      "step": 39800
-    },
-    {
-      "epoch": 6.409678328435122,
-      "grad_norm": 0.0008251548861153424,
-      "learning_rate": 0.00019997973820313564,
-      "loss": 46.0,
-      "step": 39801
-    },
-    {
-      "epoch": 6.40983936551391,
-      "grad_norm": 0.00289416266605258,
-      "learning_rate": 0.00019997973718474622,
-      "loss": 46.0,
-      "step": 39802
-    },
-    {
-      "epoch": 6.410000402592697,
-      "grad_norm": 0.0012698157224804163,
-      "learning_rate": 0.00019997973616633119,
-      "loss": 46.0,
-      "step": 39803
-    },
-    {
-      "epoch": 6.4101614396714846,
-      "grad_norm": 0.0016321815783157945,
-      "learning_rate": 0.00019997973514789056,
-      "loss": 46.0,
-      "step": 39804
-    },
-    {
-      "epoch": 6.410322476750272,
-      "grad_norm": 0.012155700474977493,
-      "learning_rate": 0.00019997973412942436,
-      "loss": 46.0,
-      "step": 39805
-    },
-    {
-      "epoch": 6.4104835138290595,
-      "grad_norm": 0.0021846024319529533,
-      "learning_rate": 0.00019997973311093253,
-      "loss": 46.0,
-      "step": 39806
-    },
-    {
-      "epoch": 6.410644550907847,
-      "grad_norm": 0.002425346989184618,
-      "learning_rate": 0.00019997973209241515,
-      "loss": 46.0,
-      "step": 39807
-    },
-    {
-      "epoch": 6.4108055879866335,
-      "grad_norm": 0.003928133752197027,
-      "learning_rate": 0.00019997973107387215,
-      "loss": 46.0,
-      "step": 39808
-    },
-    {
-      "epoch": 6.410966625065421,
-      "grad_norm": 0.0027051239740103483,
-      "learning_rate": 0.0001999797300553036,
-      "loss": 46.0,
-      "step": 39809
-    },
-    {
-      "epoch": 6.411127662144208,
-      "grad_norm": 0.0018706123810261488,
-      "learning_rate": 0.00019997972903670945,
-      "loss": 46.0,
-      "step": 39810
-    },
-    {
-      "epoch": 6.411288699222996,
-      "grad_norm": 0.002114361384883523,
-      "learning_rate": 0.0001999797280180897,
-      "loss": 46.0,
-      "step": 39811
-    },
-    {
-      "epoch": 6.411449736301783,
-      "grad_norm": 0.00814177468419075,
-      "learning_rate": 0.00019997972699944435,
-      "loss": 46.0,
-      "step": 39812
-    },
-    {
-      "epoch": 6.411610773380571,
-      "grad_norm": 0.0031662913970649242,
-      "learning_rate": 0.0001999797259807734,
-      "loss": 46.0,
-      "step": 39813
-    },
-    {
-      "epoch": 6.411771810459358,
-      "grad_norm": 0.0016604793490841985,
-      "learning_rate": 0.00019997972496207692,
-      "loss": 46.0,
-      "step": 39814
-    },
-    {
-      "epoch": 6.411932847538146,
-      "grad_norm": 0.003207709640264511,
-      "learning_rate": 0.00019997972394335478,
-      "loss": 46.0,
-      "step": 39815
-    },
-    {
-      "epoch": 6.412093884616933,
-      "grad_norm": 0.01981075294315815,
-      "learning_rate": 0.0001999797229246071,
-      "loss": 46.0,
-      "step": 39816
-    },
-    {
-      "epoch": 6.4122549216957205,
-      "grad_norm": 0.00438704015687108,
-      "learning_rate": 0.0001999797219058338,
-      "loss": 46.0,
-      "step": 39817
-    },
-    {
-      "epoch": 6.412415958774508,
-      "grad_norm": 0.0021715660113841295,
-      "learning_rate": 0.0001999797208870349,
-      "loss": 46.0,
-      "step": 39818
-    },
-    {
-      "epoch": 6.412576995853295,
-      "grad_norm": 0.009480034932494164,
-      "learning_rate": 0.00019997971986821046,
-      "loss": 46.0,
-      "step": 39819
-    },
-    {
-      "epoch": 6.412738032932083,
-      "grad_norm": 0.004029309842735529,
-      "learning_rate": 0.00019997971884936039,
-      "loss": 46.0,
-      "step": 39820
-    },
-    {
-      "epoch": 6.41289907001087,
-      "grad_norm": 0.003662655595690012,
-      "learning_rate": 0.00019997971783048475,
-      "loss": 46.0,
-      "step": 39821
-    },
-    {
-      "epoch": 6.413060107089658,
-      "grad_norm": 0.0010933639714494348,
-      "learning_rate": 0.0001999797168115835,
-      "loss": 46.0,
-      "step": 39822
-    },
-    {
-      "epoch": 6.413221144168444,
-      "grad_norm": 0.0026792031712830067,
-      "learning_rate": 0.0001999797157926567,
-      "loss": 46.0,
-      "step": 39823
-    },
-    {
-      "epoch": 6.413382181247232,
-      "grad_norm": 0.0029935420025140047,
-      "learning_rate": 0.00019997971477370429,
-      "loss": 46.0,
-      "step": 39824
-    },
-    {
-      "epoch": 6.413543218326019,
-      "grad_norm": 0.01949482224881649,
-      "learning_rate": 0.00019997971375472625,
-      "loss": 46.0,
-      "step": 39825
-    },
-    {
-      "epoch": 6.413704255404807,
-      "grad_norm": 0.01633744314312935,
-      "learning_rate": 0.00019997971273572266,
-      "loss": 46.0,
-      "step": 39826
-    },
-    {
-      "epoch": 6.413865292483594,
-      "grad_norm": 0.002060030587017536,
-      "learning_rate": 0.0001999797117166935,
-      "loss": 46.0,
-      "step": 39827
-    },
-    {
-      "epoch": 6.414026329562382,
-      "grad_norm": 0.005133353173732758,
-      "learning_rate": 0.0001999797106976387,
-      "loss": 46.0,
-      "step": 39828
-    },
-    {
-      "epoch": 6.414187366641169,
-      "grad_norm": 0.0031170567963272333,
-      "learning_rate": 0.00019997970967855838,
-      "loss": 46.0,
-      "step": 39829
-    },
-    {
-      "epoch": 6.4143484037199565,
-      "grad_norm": 0.0056282877922058105,
-      "learning_rate": 0.0001999797086594524,
-      "loss": 46.0,
-      "step": 39830
-    },
-    {
-      "epoch": 6.414509440798744,
-      "grad_norm": 0.00829550065100193,
-      "learning_rate": 0.00019997970764032086,
-      "loss": 46.0,
-      "step": 39831
-    },
-    {
-      "epoch": 6.414670477877531,
-      "grad_norm": 0.0027705435641109943,
-      "learning_rate": 0.0001999797066211637,
-      "loss": 46.0,
-      "step": 39832
-    },
-    {
-      "epoch": 6.414831514956319,
-      "grad_norm": 0.003135720267891884,
-      "learning_rate": 0.00019997970560198098,
-      "loss": 46.0,
-      "step": 39833
-    },
-    {
-      "epoch": 6.414992552035106,
-      "grad_norm": 0.009653245098888874,
-      "learning_rate": 0.0001999797045827727,
-      "loss": 46.0,
-      "step": 39834
-    },
-    {
-      "epoch": 6.415153589113894,
-      "grad_norm": 0.0020820614881813526,
-      "learning_rate": 0.00019997970356353876,
-      "loss": 46.0,
-      "step": 39835
-    },
-    {
-      "epoch": 6.415314626192681,
-      "grad_norm": 0.0022661357652395964,
-      "learning_rate": 0.0001999797025442793,
-      "loss": 46.0,
-      "step": 39836
-    },
-    {
-      "epoch": 6.415475663271469,
-      "grad_norm": 0.003800329752266407,
-      "learning_rate": 0.00019997970152499421,
-      "loss": 46.0,
-      "step": 39837
-    },
-    {
-      "epoch": 6.415636700350255,
-      "grad_norm": 0.0021906327456235886,
-      "learning_rate": 0.00019997970050568355,
-      "loss": 46.0,
-      "step": 39838
-    },
-    {
-      "epoch": 6.415797737429043,
-      "grad_norm": 0.0018767487490549684,
-      "learning_rate": 0.0001999796994863473,
-      "loss": 46.0,
-      "step": 39839
-    },
-    {
-      "epoch": 6.41595877450783,
-      "grad_norm": 0.00743872718885541,
-      "learning_rate": 0.0001999796984669854,
-      "loss": 46.0,
-      "step": 39840
-    },
-    {
-      "epoch": 6.416119811586618,
-      "grad_norm": 0.008345180191099644,
-      "learning_rate": 0.000199979697447598,
-      "loss": 46.0,
-      "step": 39841
-    },
-    {
-      "epoch": 6.416280848665405,
-      "grad_norm": 0.002836718689650297,
-      "learning_rate": 0.00019997969642818495,
-      "loss": 46.0,
-      "step": 39842
-    },
-    {
-      "epoch": 6.4164418857441925,
-      "grad_norm": 0.0029085404239594936,
-      "learning_rate": 0.00019997969540874633,
-      "loss": 46.0,
-      "step": 39843
-    },
-    {
-      "epoch": 6.41660292282298,
-      "grad_norm": 0.0067632547579705715,
-      "learning_rate": 0.00019997969438928214,
-      "loss": 46.0,
-      "step": 39844
-    },
-    {
-      "epoch": 6.416763959901767,
-      "grad_norm": 0.0033712266013026237,
-      "learning_rate": 0.00019997969336979233,
-      "loss": 46.0,
-      "step": 39845
-    },
-    {
-      "epoch": 6.416924996980555,
-      "grad_norm": 0.005762216169387102,
-      "learning_rate": 0.00019997969235027694,
-      "loss": 46.0,
-      "step": 39846
-    },
-    {
-      "epoch": 6.417086034059342,
-      "grad_norm": 0.005931038875132799,
-      "learning_rate": 0.00019997969133073596,
-      "loss": 46.0,
-      "step": 39847
-    },
-    {
-      "epoch": 6.41724707113813,
-      "grad_norm": 0.013277573511004448,
-      "learning_rate": 0.0001999796903111694,
-      "loss": 46.0,
-      "step": 39848
-    },
-    {
-      "epoch": 6.417408108216917,
-      "grad_norm": 0.008250647224485874,
-      "learning_rate": 0.00019997968929157724,
-      "loss": 46.0,
-      "step": 39849
-    },
-    {
-      "epoch": 6.417569145295705,
-      "grad_norm": 0.008172634989023209,
-      "learning_rate": 0.00019997968827195948,
-      "loss": 46.0,
-      "step": 39850
-    },
-    {
-      "epoch": 6.417730182374492,
-      "grad_norm": 0.004523251671344042,
-      "learning_rate": 0.00019997968725231615,
-      "loss": 46.0,
-      "step": 39851
-    },
-    {
-      "epoch": 6.41789121945328,
-      "grad_norm": 0.023154499009251595,
-      "learning_rate": 0.00019997968623264724,
-      "loss": 46.0,
-      "step": 39852
-    },
-    {
-      "epoch": 6.418052256532066,
-      "grad_norm": 0.002058671787381172,
-      "learning_rate": 0.0001999796852129527,
-      "loss": 46.0,
-      "step": 39853
-    },
-    {
-      "epoch": 6.418213293610854,
-      "grad_norm": 0.005009278655052185,
-      "learning_rate": 0.00019997968419323262,
-      "loss": 46.0,
-      "step": 39854
-    },
-    {
-      "epoch": 6.418374330689641,
-      "grad_norm": 0.004390972200781107,
-      "learning_rate": 0.0001999796831734869,
-      "loss": 46.0,
-      "step": 39855
-    },
-    {
-      "epoch": 6.4185353677684285,
-      "grad_norm": 0.0029138128738850355,
-      "learning_rate": 0.0001999796821537156,
-      "loss": 46.0,
-      "step": 39856
-    },
-    {
-      "epoch": 6.418696404847216,
-      "grad_norm": 0.006636871490627527,
-      "learning_rate": 0.00019997968113391875,
-      "loss": 46.0,
-      "step": 39857
-    },
-    {
-      "epoch": 6.418857441926003,
-      "grad_norm": 0.003436867380514741,
-      "learning_rate": 0.00019997968011409628,
-      "loss": 46.0,
-      "step": 39858
-    },
-    {
-      "epoch": 6.419018479004791,
-      "grad_norm": 0.009712686762213707,
-      "learning_rate": 0.0001999796790942482,
-      "loss": 46.0,
-      "step": 39859
-    },
-    {
-      "epoch": 6.419179516083578,
-      "grad_norm": 0.004144848790019751,
-      "learning_rate": 0.00019997967807437457,
-      "loss": 46.0,
-      "step": 39860
-    },
-    {
-      "epoch": 6.419340553162366,
-      "grad_norm": 0.01024557463824749,
-      "learning_rate": 0.00019997967705447534,
-      "loss": 46.0,
-      "step": 39861
-    },
-    {
-      "epoch": 6.419501590241153,
-      "grad_norm": 0.008240933530032635,
-      "learning_rate": 0.00019997967603455053,
-      "loss": 46.0,
-      "step": 39862
-    },
-    {
-      "epoch": 6.419662627319941,
-      "grad_norm": 0.009815323166549206,
-      "learning_rate": 0.0001999796750146001,
-      "loss": 46.0,
-      "step": 39863
-    },
-    {
-      "epoch": 6.419823664398728,
-      "grad_norm": 0.0018769773887470365,
-      "learning_rate": 0.0001999796739946241,
-      "loss": 46.0,
-      "step": 39864
-    },
-    {
-      "epoch": 6.4199847014775155,
-      "grad_norm": 0.02064286358654499,
-      "learning_rate": 0.0001999796729746225,
-      "loss": 46.0,
-      "step": 39865
-    },
-    {
-      "epoch": 6.420145738556303,
-      "grad_norm": 0.0028559728525578976,
-      "learning_rate": 0.00019997967195459535,
-      "loss": 46.0,
-      "step": 39866
-    },
-    {
-      "epoch": 6.4203067756350904,
-      "grad_norm": 0.004231594502925873,
-      "learning_rate": 0.00019997967093454255,
-      "loss": 46.0,
-      "step": 39867
-    },
-    {
-      "epoch": 6.420467812713877,
-      "grad_norm": 0.0037256081122905016,
-      "learning_rate": 0.00019997966991446418,
-      "loss": 46.0,
-      "step": 39868
-    },
-    {
-      "epoch": 6.4206288497926645,
-      "grad_norm": 0.004277756903320551,
-      "learning_rate": 0.00019997966889436026,
-      "loss": 46.0,
-      "step": 39869
-    },
-    {
-      "epoch": 6.420789886871452,
-      "grad_norm": 0.003637191839516163,
-      "learning_rate": 0.0001999796678742307,
-      "loss": 46.0,
-      "step": 39870
-    },
-    {
-      "epoch": 6.420950923950239,
-      "grad_norm": 0.005565167870372534,
-      "learning_rate": 0.00019997966685407557,
-      "loss": 46.0,
-      "step": 39871
-    },
-    {
-      "epoch": 6.421111961029027,
-      "grad_norm": 0.00239057675935328,
-      "learning_rate": 0.00019997966583389486,
-      "loss": 46.0,
-      "step": 39872
-    },
-    {
-      "epoch": 6.421272998107814,
-      "grad_norm": 0.0013104741228744388,
-      "learning_rate": 0.00019997966481368856,
-      "loss": 46.0,
-      "step": 39873
-    },
-    {
-      "epoch": 6.421434035186602,
-      "grad_norm": 0.014932210557162762,
-      "learning_rate": 0.00019997966379345665,
-      "loss": 46.0,
-      "step": 39874
-    },
-    {
-      "epoch": 6.421595072265389,
-      "grad_norm": 0.0022315820679068565,
-      "learning_rate": 0.00019997966277319917,
-      "loss": 46.0,
-      "step": 39875
-    },
-    {
-      "epoch": 6.421756109344177,
-      "grad_norm": 0.001118012354709208,
-      "learning_rate": 0.00019997966175291606,
-      "loss": 46.0,
-      "step": 39876
-    },
-    {
-      "epoch": 6.421917146422964,
-      "grad_norm": 0.0058967643417418,
-      "learning_rate": 0.0001999796607326074,
-      "loss": 46.0,
-      "step": 39877
-    },
-    {
-      "epoch": 6.4220781835017515,
-      "grad_norm": 0.0029340407345443964,
-      "learning_rate": 0.00019997965971227315,
-      "loss": 46.0,
-      "step": 39878
-    },
-    {
-      "epoch": 6.422239220580539,
-      "grad_norm": 0.005983872339129448,
-      "learning_rate": 0.0001999796586919133,
-      "loss": 46.0,
-      "step": 39879
-    },
-    {
-      "epoch": 6.422400257659326,
-      "grad_norm": 0.007284213323146105,
-      "learning_rate": 0.00019997965767152786,
-      "loss": 46.0,
-      "step": 39880
-    },
-    {
-      "epoch": 6.422561294738113,
-      "grad_norm": 0.005100995302200317,
-      "learning_rate": 0.00019997965665111684,
-      "loss": 46.0,
-      "step": 39881
-    },
-    {
-      "epoch": 6.4227223318169,
-      "grad_norm": 0.0020110623445361853,
-      "learning_rate": 0.00019997965563068023,
-      "loss": 46.0,
-      "step": 39882
-    },
-    {
-      "epoch": 6.422883368895688,
-      "grad_norm": 0.0035027451813220978,
-      "learning_rate": 0.000199979654610218,
-      "loss": 46.0,
-      "step": 39883
-    },
-    {
-      "epoch": 6.423044405974475,
-      "grad_norm": 0.0021621331106871367,
-      "learning_rate": 0.00019997965358973022,
-      "loss": 46.0,
-      "step": 39884
-    },
-    {
-      "epoch": 6.423205443053263,
-      "grad_norm": 0.012578175403177738,
-      "learning_rate": 0.00019997965256921685,
-      "loss": 46.0,
-      "step": 39885
-    },
-    {
-      "epoch": 6.42336648013205,
-      "grad_norm": 0.005133448168635368,
-      "learning_rate": 0.00019997965154867786,
-      "loss": 46.0,
-      "step": 39886
-    },
-    {
-      "epoch": 6.423527517210838,
-      "grad_norm": 0.0025536410976201296,
-      "learning_rate": 0.0001999796505281133,
-      "loss": 46.0,
-      "step": 39887
-    },
-    {
-      "epoch": 6.423688554289625,
-      "grad_norm": 0.00344303366728127,
-      "learning_rate": 0.00019997964950752315,
-      "loss": 46.0,
-      "step": 39888
-    },
-    {
-      "epoch": 6.423849591368413,
-      "grad_norm": 0.004066459368914366,
-      "learning_rate": 0.0001999796484869074,
-      "loss": 46.0,
-      "step": 39889
-    },
-    {
-      "epoch": 6.4240106284472,
-      "grad_norm": 0.015413583256304264,
-      "learning_rate": 0.00019997964746626607,
-      "loss": 46.0,
-      "step": 39890
-    },
-    {
-      "epoch": 6.4241716655259875,
-      "grad_norm": 0.02832922898232937,
-      "learning_rate": 0.00019997964644559915,
-      "loss": 46.0,
-      "step": 39891
-    },
-    {
-      "epoch": 6.424332702604775,
-      "grad_norm": 0.0025692095514386892,
-      "learning_rate": 0.00019997964542490664,
-      "loss": 46.0,
-      "step": 39892
-    },
-    {
-      "epoch": 6.424493739683562,
-      "grad_norm": 0.002435144269838929,
-      "learning_rate": 0.00019997964440418852,
-      "loss": 46.0,
-      "step": 39893
-    },
-    {
-      "epoch": 6.42465477676235,
-      "grad_norm": 0.007873868569731712,
-      "learning_rate": 0.00019997964338344483,
-      "loss": 46.0,
-      "step": 39894
-    },
-    {
-      "epoch": 6.424815813841137,
-      "grad_norm": 0.003546998370438814,
-      "learning_rate": 0.00019997964236267553,
-      "loss": 46.0,
-      "step": 39895
-    },
-    {
-      "epoch": 6.424976850919924,
-      "grad_norm": 0.002073277486488223,
-      "learning_rate": 0.00019997964134188068,
-      "loss": 46.0,
-      "step": 39896
-    },
-    {
-      "epoch": 6.425137887998711,
-      "grad_norm": 0.014415815472602844,
-      "learning_rate": 0.00019997964032106023,
-      "loss": 46.0,
-      "step": 39897
-    },
-    {
-      "epoch": 6.425298925077499,
-      "grad_norm": 0.0028904308564960957,
-      "learning_rate": 0.00019997963930021414,
-      "loss": 46.0,
-      "step": 39898
-    },
-    {
-      "epoch": 6.425459962156286,
-      "grad_norm": 0.008536191657185555,
-      "learning_rate": 0.00019997963827934252,
-      "loss": 46.0,
-      "step": 39899
-    },
-    {
-      "epoch": 6.425620999235074,
-      "grad_norm": 0.0025488147512078285,
-      "learning_rate": 0.00019997963725844532,
-      "loss": 46.0,
-      "step": 39900
-    },
-    {
-      "epoch": 6.425782036313861,
-      "grad_norm": 0.0025260413531214,
-      "learning_rate": 0.00019997963623752247,
-      "loss": 46.0,
-      "step": 39901
-    },
-    {
-      "epoch": 6.425943073392649,
-      "grad_norm": 0.005069076549261808,
-      "learning_rate": 0.00019997963521657406,
-      "loss": 46.0,
-      "step": 39902
-    },
-    {
-      "epoch": 6.426104110471436,
-      "grad_norm": 0.0027852696366608143,
-      "learning_rate": 0.00019997963419560007,
-      "loss": 46.0,
-      "step": 39903
-    },
-    {
-      "epoch": 6.4262651475502235,
-      "grad_norm": 0.007520048413425684,
-      "learning_rate": 0.00019997963317460046,
-      "loss": 46.0,
-      "step": 39904
-    },
-    {
-      "epoch": 6.426426184629011,
-      "grad_norm": 0.008179187774658203,
-      "learning_rate": 0.00019997963215357529,
-      "loss": 46.0,
-      "step": 39905
-    },
-    {
-      "epoch": 6.426587221707798,
-      "grad_norm": 0.008504759520292282,
-      "learning_rate": 0.0001999796311325245,
-      "loss": 46.0,
-      "step": 39906
-    },
-    {
-      "epoch": 6.426748258786586,
-      "grad_norm": 0.0013578677317127585,
-      "learning_rate": 0.00019997963011144818,
-      "loss": 46.0,
-      "step": 39907
-    },
-    {
-      "epoch": 6.426909295865373,
-      "grad_norm": 0.0014628483913838863,
-      "learning_rate": 0.00019997962909034623,
-      "loss": 46.0,
-      "step": 39908
-    },
-    {
-      "epoch": 6.427070332944161,
-      "grad_norm": 0.006598500534892082,
-      "learning_rate": 0.00019997962806921868,
-      "loss": 46.0,
-      "step": 39909
-    },
-    {
-      "epoch": 6.427231370022948,
-      "grad_norm": 0.00779604772105813,
-      "learning_rate": 0.00019997962704806558,
-      "loss": 46.0,
-      "step": 39910
-    },
-    {
-      "epoch": 6.427392407101735,
-      "grad_norm": 0.009120805189013481,
-      "learning_rate": 0.00019997962602688683,
-      "loss": 46.0,
-      "step": 39911
-    },
-    {
-      "epoch": 6.427553444180522,
-      "grad_norm": 0.022817475721240044,
-      "learning_rate": 0.00019997962500568255,
-      "loss": 46.0,
-      "step": 39912
-    },
-    {
-      "epoch": 6.42771448125931,
-      "grad_norm": 0.020378459244966507,
-      "learning_rate": 0.00019997962398445265,
-      "loss": 46.0,
-      "step": 39913
-    },
-    {
-      "epoch": 6.427875518338097,
-      "grad_norm": 0.007933182641863823,
-      "learning_rate": 0.00019997962296319715,
-      "loss": 46.0,
-      "step": 39914
-    },
-    {
-      "epoch": 6.428036555416885,
-      "grad_norm": 0.0112680085003376,
-      "learning_rate": 0.00019997962194191605,
-      "loss": 46.0,
-      "step": 39915
-    },
-    {
-      "epoch": 6.428197592495672,
-      "grad_norm": 0.001529692322947085,
-      "learning_rate": 0.00019997962092060942,
-      "loss": 46.0,
-      "step": 39916
-    },
-    {
-      "epoch": 6.4283586295744595,
-      "grad_norm": 0.008812527172267437,
-      "learning_rate": 0.00019997961989927715,
-      "loss": 46.0,
-      "step": 39917
-    },
-    {
-      "epoch": 6.428519666653247,
-      "grad_norm": 0.004057199228554964,
-      "learning_rate": 0.00019997961887791932,
-      "loss": 46.0,
-      "step": 39918
-    },
-    {
-      "epoch": 6.428680703732034,
-      "grad_norm": 0.0029255924746394157,
-      "learning_rate": 0.00019997961785653588,
-      "loss": 46.0,
-      "step": 39919
-    },
-    {
-      "epoch": 6.428841740810822,
-      "grad_norm": 0.012478676624596119,
-      "learning_rate": 0.00019997961683512684,
-      "loss": 46.0,
-      "step": 39920
-    },
-    {
-      "epoch": 6.429002777889609,
-      "grad_norm": 0.0013470868580043316,
-      "learning_rate": 0.00019997961581369223,
-      "loss": 46.0,
-      "step": 39921
-    },
-    {
-      "epoch": 6.429163814968397,
-      "grad_norm": 0.0019315013196319342,
-      "learning_rate": 0.00019997961479223205,
-      "loss": 46.0,
-      "step": 39922
-    },
-    {
-      "epoch": 6.429324852047184,
-      "grad_norm": 0.0019911068957298994,
-      "learning_rate": 0.00019997961377074623,
-      "loss": 46.0,
-      "step": 39923
-    },
-    {
-      "epoch": 6.429485889125972,
-      "grad_norm": 0.007775294128805399,
-      "learning_rate": 0.00019997961274923487,
-      "loss": 46.0,
-      "step": 39924
-    },
-    {
-      "epoch": 6.429646926204759,
-      "grad_norm": 0.0005506537854671478,
-      "learning_rate": 0.00019997961172769788,
-      "loss": 46.0,
-      "step": 39925
-    },
-    {
-      "epoch": 6.429807963283546,
-      "grad_norm": 0.005360723938792944,
-      "learning_rate": 0.00019997961070613532,
-      "loss": 46.0,
-      "step": 39926
-    },
-    {
-      "epoch": 6.429969000362333,
-      "grad_norm": 0.0028473397251218557,
-      "learning_rate": 0.00019997960968454716,
-      "loss": 46.0,
-      "step": 39927
-    },
-    {
-      "epoch": 6.4301300374411205,
-      "grad_norm": 0.00832462403923273,
-      "learning_rate": 0.00019997960866293343,
-      "loss": 46.0,
-      "step": 39928
-    },
-    {
-      "epoch": 6.430291074519908,
-      "grad_norm": 0.003507880959659815,
-      "learning_rate": 0.00019997960764129408,
-      "loss": 46.0,
-      "step": 39929
-    },
-    {
-      "epoch": 6.4304521115986955,
-      "grad_norm": 0.0037783945444971323,
-      "learning_rate": 0.00019997960661962918,
-      "loss": 46.0,
-      "step": 39930
-    },
-    {
-      "epoch": 6.430613148677483,
-      "grad_norm": 0.0014754581497982144,
-      "learning_rate": 0.00019997960559793866,
-      "loss": 46.0,
-      "step": 39931
-    },
-    {
-      "epoch": 6.43077418575627,
-      "grad_norm": 0.02546934224665165,
-      "learning_rate": 0.00019997960457622259,
-      "loss": 46.0,
-      "step": 39932
-    },
-    {
-      "epoch": 6.430935222835058,
-      "grad_norm": 0.003229832975193858,
-      "learning_rate": 0.0001999796035544809,
-      "loss": 46.0,
-      "step": 39933
-    },
-    {
-      "epoch": 6.431096259913845,
-      "grad_norm": 0.004819541238248348,
-      "learning_rate": 0.0001999796025327136,
-      "loss": 46.0,
-      "step": 39934
-    },
-    {
-      "epoch": 6.431257296992633,
-      "grad_norm": 0.01111472025513649,
-      "learning_rate": 0.00019997960151092072,
-      "loss": 46.0,
-      "step": 39935
-    },
-    {
-      "epoch": 6.43141833407142,
-      "grad_norm": 0.0025102586951106787,
-      "learning_rate": 0.00019997960048910227,
-      "loss": 46.0,
-      "step": 39936
-    },
-    {
-      "epoch": 6.431579371150208,
-      "grad_norm": 0.004387294873595238,
-      "learning_rate": 0.00019997959946725823,
-      "loss": 46.0,
-      "step": 39937
-    },
-    {
-      "epoch": 6.431740408228995,
-      "grad_norm": 0.0065118856728076935,
-      "learning_rate": 0.00019997959844538857,
-      "loss": 46.0,
-      "step": 39938
-    },
-    {
-      "epoch": 6.4319014453077825,
-      "grad_norm": 0.0028932224959135056,
-      "learning_rate": 0.00019997959742349336,
-      "loss": 46.0,
-      "step": 39939
-    },
-    {
-      "epoch": 6.43206248238657,
-      "grad_norm": 0.0025285137817263603,
-      "learning_rate": 0.00019997959640157255,
-      "loss": 46.0,
-      "step": 39940
-    },
-    {
-      "epoch": 6.4322235194653565,
-      "grad_norm": 0.0064513226971030235,
-      "learning_rate": 0.00019997959537962614,
-      "loss": 46.0,
-      "step": 39941
-    },
-    {
-      "epoch": 6.432384556544144,
-      "grad_norm": 0.0023720781318843365,
-      "learning_rate": 0.00019997959435765413,
-      "loss": 46.0,
-      "step": 39942
-    },
-    {
-      "epoch": 6.432545593622931,
-      "grad_norm": 0.0040083895437419415,
-      "learning_rate": 0.00019997959333565657,
-      "loss": 46.0,
-      "step": 39943
-    },
-    {
-      "epoch": 6.432706630701719,
-      "grad_norm": 0.005140585359185934,
-      "learning_rate": 0.0001999795923136334,
-      "loss": 46.0,
-      "step": 39944
-    },
-    {
-      "epoch": 6.432867667780506,
-      "grad_norm": 0.0031324916053563356,
-      "learning_rate": 0.00019997959129158463,
-      "loss": 46.0,
-      "step": 39945
-    },
-    {
-      "epoch": 6.433028704859294,
-      "grad_norm": 0.008028978481888771,
-      "learning_rate": 0.00019997959026951025,
-      "loss": 46.0,
-      "step": 39946
-    },
-    {
-      "epoch": 6.433189741938081,
-      "grad_norm": 0.006418431177735329,
-      "learning_rate": 0.00019997958924741034,
-      "loss": 46.0,
-      "step": 39947
-    },
-    {
-      "epoch": 6.433350779016869,
-      "grad_norm": 0.012938342057168484,
-      "learning_rate": 0.0001999795882252848,
-      "loss": 46.0,
-      "step": 39948
-    },
-    {
-      "epoch": 6.433511816095656,
-      "grad_norm": 0.008282676339149475,
-      "learning_rate": 0.00019997958720313367,
-      "loss": 46.0,
-      "step": 39949
-    },
-    {
-      "epoch": 6.433672853174444,
-      "grad_norm": 0.0019169311271980405,
-      "learning_rate": 0.00019997958618095697,
-      "loss": 46.0,
-      "step": 39950
-    },
-    {
-      "epoch": 6.433833890253231,
-      "grad_norm": 0.003588273422792554,
-      "learning_rate": 0.00019997958515875465,
-      "loss": 46.0,
-      "step": 39951
-    },
-    {
-      "epoch": 6.4339949273320185,
-      "grad_norm": 0.002353373449295759,
-      "learning_rate": 0.00019997958413652675,
-      "loss": 46.0,
-      "step": 39952
-    },
-    {
-      "epoch": 6.434155964410806,
-      "grad_norm": 0.00157739722635597,
-      "learning_rate": 0.00019997958311427326,
-      "loss": 46.0,
-      "step": 39953
-    },
-    {
-      "epoch": 6.4343170014895925,
-      "grad_norm": 0.0011100948322564363,
-      "learning_rate": 0.0001999795820919942,
-      "loss": 46.0,
-      "step": 39954
-    },
-    {
-      "epoch": 6.43447803856838,
-      "grad_norm": 0.016226520761847496,
-      "learning_rate": 0.00019997958106968952,
-      "loss": 46.0,
-      "step": 39955
-    },
-    {
-      "epoch": 6.434639075647167,
-      "grad_norm": 0.018520256504416466,
-      "learning_rate": 0.0001999795800473593,
-      "loss": 46.0,
-      "step": 39956
-    },
-    {
-      "epoch": 6.434800112725955,
-      "grad_norm": 0.0020396767649799585,
-      "learning_rate": 0.00019997957902500346,
-      "loss": 46.0,
-      "step": 39957
-    },
-    {
-      "epoch": 6.434961149804742,
-      "grad_norm": 0.01737462542951107,
-      "learning_rate": 0.000199979578002622,
-      "loss": 46.0,
-      "step": 39958
-    },
-    {
-      "epoch": 6.43512218688353,
-      "grad_norm": 0.02390354871749878,
-      "learning_rate": 0.000199979576980215,
-      "loss": 46.0,
-      "step": 39959
-    },
-    {
-      "epoch": 6.435283223962317,
-      "grad_norm": 0.005807912442833185,
-      "learning_rate": 0.0001999795759577824,
-      "loss": 46.0,
-      "step": 39960
-    },
-    {
-      "epoch": 6.435444261041105,
-      "grad_norm": 0.009375504218041897,
-      "learning_rate": 0.0001999795749353242,
-      "loss": 46.0,
-      "step": 39961
-    },
-    {
-      "epoch": 6.435605298119892,
-      "grad_norm": 0.009242935106158257,
-      "learning_rate": 0.0001999795739128404,
-      "loss": 46.0,
-      "step": 39962
-    },
-    {
-      "epoch": 6.43576633519868,
-      "grad_norm": 0.0029400717467069626,
-      "learning_rate": 0.00019997957289033102,
-      "loss": 46.0,
-      "step": 39963
-    },
-    {
-      "epoch": 6.435927372277467,
-      "grad_norm": 0.014346753247082233,
-      "learning_rate": 0.00019997957186779607,
-      "loss": 46.0,
-      "step": 39964
-    },
-    {
-      "epoch": 6.4360884093562545,
-      "grad_norm": 0.010555626824498177,
-      "learning_rate": 0.00019997957084523548,
-      "loss": 46.0,
-      "step": 39965
-    },
-    {
-      "epoch": 6.436249446435042,
-      "grad_norm": 0.005844416096806526,
-      "learning_rate": 0.00019997956982264933,
-      "loss": 46.0,
-      "step": 39966
-    },
-    {
-      "epoch": 6.436410483513829,
-      "grad_norm": 0.014791611582040787,
-      "learning_rate": 0.00019997956880003762,
-      "loss": 46.0,
-      "step": 39967
-    },
-    {
-      "epoch": 6.436571520592617,
-      "grad_norm": 0.0034442765172570944,
-      "learning_rate": 0.0001999795677774003,
-      "loss": 46.0,
-      "step": 39968
-    },
-    {
-      "epoch": 6.436732557671403,
-      "grad_norm": 0.0019294825615361333,
-      "learning_rate": 0.00019997956675473736,
-      "loss": 46.0,
-      "step": 39969
-    },
-    {
-      "epoch": 6.436893594750191,
-      "grad_norm": 0.0016740956343710423,
-      "learning_rate": 0.00019997956573204886,
-      "loss": 46.0,
-      "step": 39970
-    },
-    {
-      "epoch": 6.437054631828978,
-      "grad_norm": 0.015755239874124527,
-      "learning_rate": 0.00019997956470933477,
-      "loss": 46.0,
-      "step": 39971
-    },
-    {
-      "epoch": 6.437215668907766,
-      "grad_norm": 0.0018250519642606378,
-      "learning_rate": 0.0001999795636865951,
-      "loss": 46.0,
-      "step": 39972
-    },
-    {
-      "epoch": 6.437376705986553,
-      "grad_norm": 0.0018212461145594716,
-      "learning_rate": 0.0001999795626638298,
-      "loss": 46.0,
-      "step": 39973
-    },
-    {
-      "epoch": 6.437537743065341,
-      "grad_norm": 0.0012948332587257028,
-      "learning_rate": 0.00019997956164103894,
-      "loss": 46.0,
-      "step": 39974
-    },
-    {
-      "epoch": 6.437698780144128,
-      "grad_norm": 0.017226548865437508,
-      "learning_rate": 0.0001999795606182225,
-      "loss": 46.0,
-      "step": 39975
-    },
-    {
-      "epoch": 6.437859817222916,
-      "grad_norm": 0.006045111455023289,
-      "learning_rate": 0.00019997955959538043,
-      "loss": 46.0,
-      "step": 39976
-    },
-    {
-      "epoch": 6.438020854301703,
-      "grad_norm": 0.004506081808358431,
-      "learning_rate": 0.00019997955857251282,
-      "loss": 46.0,
-      "step": 39977
-    },
-    {
-      "epoch": 6.4381818913804905,
-      "grad_norm": 0.002988075837492943,
-      "learning_rate": 0.0001999795575496196,
-      "loss": 46.0,
-      "step": 39978
-    },
-    {
-      "epoch": 6.438342928459278,
-      "grad_norm": 0.00566342007368803,
-      "learning_rate": 0.00019997955652670079,
-      "loss": 46.0,
-      "step": 39979
-    },
-    {
-      "epoch": 6.438503965538065,
-      "grad_norm": 0.010630440898239613,
-      "learning_rate": 0.0001999795555037564,
-      "loss": 46.0,
-      "step": 39980
-    },
-    {
-      "epoch": 6.438665002616853,
-      "grad_norm": 0.026092644780874252,
-      "learning_rate": 0.00019997955448078638,
-      "loss": 46.0,
-      "step": 39981
-    },
-    {
-      "epoch": 6.43882603969564,
-      "grad_norm": 0.0012930453522130847,
-      "learning_rate": 0.0001999795534577908,
-      "loss": 46.0,
-      "step": 39982
-    },
-    {
-      "epoch": 6.438987076774428,
-      "grad_norm": 0.009602982550859451,
-      "learning_rate": 0.00019997955243476962,
-      "loss": 46.0,
-      "step": 39983
-    },
-    {
-      "epoch": 6.439148113853214,
-      "grad_norm": 0.005496717058122158,
-      "learning_rate": 0.00019997955141172287,
-      "loss": 46.0,
-      "step": 39984
-    },
-    {
-      "epoch": 6.439309150932002,
-      "grad_norm": 0.0011985086603090167,
-      "learning_rate": 0.00019997955038865054,
-      "loss": 46.0,
-      "step": 39985
-    },
-    {
-      "epoch": 6.439470188010789,
-      "grad_norm": 0.00453329598531127,
-      "learning_rate": 0.00019997954936555256,
-      "loss": 46.0,
-      "step": 39986
-    },
-    {
-      "epoch": 6.439631225089577,
-      "grad_norm": 0.0018771899631246924,
-      "learning_rate": 0.00019997954834242906,
-      "loss": 46.0,
-      "step": 39987
-    },
-    {
-      "epoch": 6.439792262168364,
-      "grad_norm": 0.0022936861496418715,
-      "learning_rate": 0.00019997954731927994,
-      "loss": 46.0,
-      "step": 39988
-    },
-    {
-      "epoch": 6.4399532992471515,
-      "grad_norm": 0.00746731273829937,
-      "learning_rate": 0.0001999795462961052,
-      "loss": 46.0,
-      "step": 39989
-    },
-    {
-      "epoch": 6.440114336325939,
-      "grad_norm": 0.003333126427605748,
-      "learning_rate": 0.00019997954527290493,
-      "loss": 46.0,
-      "step": 39990
-    },
-    {
-      "epoch": 6.4402753734047264,
-      "grad_norm": 0.005318079609423876,
-      "learning_rate": 0.00019997954424967902,
-      "loss": 46.0,
-      "step": 39991
-    },
-    {
-      "epoch": 6.440436410483514,
-      "grad_norm": 0.00217154691927135,
-      "learning_rate": 0.00019997954322642755,
-      "loss": 46.0,
-      "step": 39992
-    },
-    {
-      "epoch": 6.440597447562301,
-      "grad_norm": 0.007333929650485516,
-      "learning_rate": 0.00019997954220315046,
-      "loss": 46.0,
-      "step": 39993
-    },
-    {
-      "epoch": 6.440758484641089,
-      "grad_norm": 0.003449036506935954,
-      "learning_rate": 0.00019997954117984782,
-      "loss": 46.0,
-      "step": 39994
-    },
-    {
-      "epoch": 6.440919521719876,
-      "grad_norm": 0.011705799959599972,
-      "learning_rate": 0.0001999795401565196,
-      "loss": 46.0,
-      "step": 39995
-    },
-    {
-      "epoch": 6.441080558798664,
-      "grad_norm": 0.010781088843941689,
-      "learning_rate": 0.00019997953913316577,
-      "loss": 46.0,
-      "step": 39996
-    },
-    {
-      "epoch": 6.441241595877451,
-      "grad_norm": 0.015286247245967388,
-      "learning_rate": 0.00019997953810978633,
-      "loss": 46.0,
-      "step": 39997
-    },
-    {
-      "epoch": 6.441402632956239,
-      "grad_norm": 0.007600422948598862,
-      "learning_rate": 0.00019997953708638131,
-      "loss": 46.0,
-      "step": 39998
-    },
-    {
-      "epoch": 6.441563670035025,
-      "grad_norm": 0.013189624063670635,
-      "learning_rate": 0.0001999795360629507,
-      "loss": 46.0,
-      "step": 39999
-    },
-    {
-      "epoch": 6.441724707113813,
-      "grad_norm": 0.0012945410562679172,
-      "learning_rate": 0.00019997953503949448,
-      "loss": 46.0,
-      "step": 40000
-    },
-    {
-      "epoch": 6.4418857441926,
-      "grad_norm": 0.022465655580163002,
-      "learning_rate": 0.0001999795340160127,
-      "loss": 46.0,
-      "step": 40001
-    },
-    {
-      "epoch": 6.4420467812713875,
-      "grad_norm": 0.004609566181898117,
-      "learning_rate": 0.00019997953299250533,
-      "loss": 46.0,
-      "step": 40002
-    },
-    {
-      "epoch": 6.442207818350175,
-      "grad_norm": 0.0034489461686462164,
-      "learning_rate": 0.00019997953196897238,
-      "loss": 46.0,
-      "step": 40003
-    },
-    {
-      "epoch": 6.442368855428962,
-      "grad_norm": 0.001429785043001175,
-      "learning_rate": 0.0001999795309454138,
-      "loss": 46.0,
-      "step": 40004
-    },
-    {
-      "epoch": 6.44252989250775,
-      "grad_norm": 0.0014718134189024568,
-      "learning_rate": 0.00019997952992182967,
-      "loss": 46.0,
-      "step": 40005
-    },
-    {
-      "epoch": 6.442690929586537,
-      "grad_norm": 0.013667715713381767,
-      "learning_rate": 0.00019997952889821993,
-      "loss": 46.0,
-      "step": 40006
-    },
-    {
-      "epoch": 6.442851966665325,
-      "grad_norm": 0.00549883721396327,
-      "learning_rate": 0.00019997952787458462,
-      "loss": 46.0,
-      "step": 40007
-    },
-    {
-      "epoch": 6.443013003744112,
-      "grad_norm": 0.0064135026186704636,
-      "learning_rate": 0.0001999795268509237,
-      "loss": 46.0,
-      "step": 40008
-    },
-    {
-      "epoch": 6.4431740408229,
-      "grad_norm": 0.004383997060358524,
-      "learning_rate": 0.00019997952582723723,
-      "loss": 46.0,
-      "step": 40009
-    },
-    {
-      "epoch": 6.443335077901687,
-      "grad_norm": 0.003669844940304756,
-      "learning_rate": 0.00019997952480352513,
-      "loss": 46.0,
-      "step": 40010
-    },
-    {
-      "epoch": 6.443496114980475,
-      "grad_norm": 0.0008407138520851731,
-      "learning_rate": 0.00019997952377978742,
-      "loss": 46.0,
-      "step": 40011
-    },
-    {
-      "epoch": 6.443657152059262,
-      "grad_norm": 0.004744230303913355,
-      "learning_rate": 0.00019997952275602416,
-      "loss": 46.0,
-      "step": 40012
-    },
-    {
-      "epoch": 6.4438181891380495,
-      "grad_norm": 0.005047994665801525,
-      "learning_rate": 0.0001999795217322353,
-      "loss": 46.0,
-      "step": 40013
-    },
-    {
-      "epoch": 6.443979226216836,
-      "grad_norm": 0.00792969111353159,
-      "learning_rate": 0.00019997952070842088,
-      "loss": 46.0,
-      "step": 40014
-    },
-    {
-      "epoch": 6.4441402632956235,
-      "grad_norm": 0.0027087705675512552,
-      "learning_rate": 0.00019997951968458083,
-      "loss": 46.0,
-      "step": 40015
-    },
-    {
-      "epoch": 6.444301300374411,
-      "grad_norm": 0.006329684052616358,
-      "learning_rate": 0.00019997951866071518,
-      "loss": 46.0,
-      "step": 40016
-    },
-    {
-      "epoch": 6.444462337453198,
-      "grad_norm": 0.0010234903311356902,
-      "learning_rate": 0.00019997951763682398,
-      "loss": 46.0,
-      "step": 40017
-    },
-    {
-      "epoch": 6.444623374531986,
-      "grad_norm": 0.008537469431757927,
-      "learning_rate": 0.00019997951661290716,
-      "loss": 46.0,
-      "step": 40018
-    },
-    {
-      "epoch": 6.444784411610773,
-      "grad_norm": 0.006302362307906151,
-      "learning_rate": 0.00019997951558896478,
-      "loss": 46.0,
-      "step": 40019
-    },
-    {
-      "epoch": 6.444945448689561,
-      "grad_norm": 0.0013192294863983989,
-      "learning_rate": 0.0001999795145649968,
-      "loss": 46.0,
-      "step": 40020
-    },
-    {
-      "epoch": 6.445106485768348,
-      "grad_norm": 0.0026684666518121958,
-      "learning_rate": 0.00019997951354100318,
-      "loss": 46.0,
-      "step": 40021
-    },
-    {
-      "epoch": 6.445267522847136,
-      "grad_norm": 0.007107526995241642,
-      "learning_rate": 0.00019997951251698404,
-      "loss": 46.0,
-      "step": 40022
-    },
-    {
-      "epoch": 6.445428559925923,
-      "grad_norm": 0.005974226165562868,
-      "learning_rate": 0.00019997951149293926,
-      "loss": 46.0,
-      "step": 40023
-    },
-    {
-      "epoch": 6.445589597004711,
-      "grad_norm": 0.001770784961991012,
-      "learning_rate": 0.00019997951046886895,
-      "loss": 46.0,
-      "step": 40024
-    },
-    {
-      "epoch": 6.445750634083498,
-      "grad_norm": 0.0035143611021339893,
-      "learning_rate": 0.000199979509444773,
-      "loss": 46.0,
-      "step": 40025
-    },
-    {
-      "epoch": 6.4459116711622855,
-      "grad_norm": 0.010423265397548676,
-      "learning_rate": 0.00019997950842065148,
-      "loss": 46.0,
-      "step": 40026
-    },
-    {
-      "epoch": 6.446072708241072,
-      "grad_norm": 0.0038291343953460455,
-      "learning_rate": 0.00019997950739650435,
-      "loss": 46.0,
-      "step": 40027
-    },
-    {
-      "epoch": 6.4462337453198595,
-      "grad_norm": 0.0021179742179811,
-      "learning_rate": 0.00019997950637233166,
-      "loss": 46.0,
-      "step": 40028
-    },
-    {
-      "epoch": 6.446394782398647,
-      "grad_norm": 0.0076445299200713634,
-      "learning_rate": 0.00019997950534813335,
-      "loss": 46.0,
-      "step": 40029
-    },
-    {
-      "epoch": 6.446555819477434,
-      "grad_norm": 0.0016426180955022573,
-      "learning_rate": 0.0001999795043239095,
-      "loss": 46.0,
-      "step": 40030
-    },
-    {
-      "epoch": 6.446716856556222,
-      "grad_norm": 0.0028022995684295893,
-      "learning_rate": 0.00019997950329966,
-      "loss": 46.0,
-      "step": 40031
-    },
-    {
-      "epoch": 6.446877893635009,
-      "grad_norm": 0.003427430521696806,
-      "learning_rate": 0.00019997950227538494,
-      "loss": 46.0,
-      "step": 40032
-    },
-    {
-      "epoch": 6.447038930713797,
-      "grad_norm": 0.006693064235150814,
-      "learning_rate": 0.0001999795012510843,
-      "loss": 46.0,
-      "step": 40033
-    },
-    {
-      "epoch": 6.447199967792584,
-      "grad_norm": 0.0017944596474990249,
-      "learning_rate": 0.00019997950022675805,
-      "loss": 46.0,
-      "step": 40034
-    },
-    {
-      "epoch": 6.447361004871372,
-      "grad_norm": 0.015099226497113705,
-      "learning_rate": 0.00019997949920240622,
-      "loss": 46.0,
-      "step": 40035
-    },
-    {
-      "epoch": 6.447522041950159,
-      "grad_norm": 0.017275149002671242,
-      "learning_rate": 0.00019997949817802878,
-      "loss": 46.0,
-      "step": 40036
-    },
-    {
-      "epoch": 6.4476830790289466,
-      "grad_norm": 0.0040924581699073315,
-      "learning_rate": 0.00019997949715362578,
-      "loss": 46.0,
-      "step": 40037
-    },
-    {
-      "epoch": 6.447844116107734,
-      "grad_norm": 0.003378939349204302,
-      "learning_rate": 0.00019997949612919716,
-      "loss": 46.0,
-      "step": 40038
-    },
-    {
-      "epoch": 6.4480051531865215,
-      "grad_norm": 0.002016445156186819,
-      "learning_rate": 0.000199979495104743,
-      "loss": 46.0,
-      "step": 40039
-    },
-    {
-      "epoch": 6.448166190265309,
-      "grad_norm": 0.0035596839152276516,
-      "learning_rate": 0.0001999794940802632,
-      "loss": 46.0,
-      "step": 40040
-    },
-    {
-      "epoch": 6.448327227344096,
-      "grad_norm": 0.0029883934184908867,
-      "learning_rate": 0.00019997949305575782,
-      "loss": 46.0,
-      "step": 40041
-    },
-    {
-      "epoch": 6.448488264422883,
-      "grad_norm": 0.002941165817901492,
-      "learning_rate": 0.00019997949203122688,
-      "loss": 46.0,
-      "step": 40042
-    },
-    {
-      "epoch": 6.44864930150167,
-      "grad_norm": 0.004230792168527842,
-      "learning_rate": 0.00019997949100667033,
-      "loss": 46.0,
-      "step": 40043
-    },
-    {
-      "epoch": 6.448810338580458,
-      "grad_norm": 0.0031707226298749447,
-      "learning_rate": 0.0001999794899820882,
-      "loss": 46.0,
-      "step": 40044
-    },
-    {
-      "epoch": 6.448971375659245,
-      "grad_norm": 0.0026096939109265804,
-      "learning_rate": 0.00019997948895748047,
-      "loss": 46.0,
-      "step": 40045
-    },
-    {
-      "epoch": 6.449132412738033,
-      "grad_norm": 0.0022774513345211744,
-      "learning_rate": 0.00019997948793284715,
-      "loss": 46.0,
-      "step": 40046
-    },
-    {
-      "epoch": 6.44929344981682,
-      "grad_norm": 0.005252016242593527,
-      "learning_rate": 0.00019997948690818825,
-      "loss": 46.0,
-      "step": 40047
-    },
-    {
-      "epoch": 6.449454486895608,
-      "grad_norm": 0.004806511104106903,
-      "learning_rate": 0.00019997948588350374,
-      "loss": 46.0,
-      "step": 40048
-    },
-    {
-      "epoch": 6.449615523974395,
-      "grad_norm": 0.0052448175847530365,
-      "learning_rate": 0.00019997948485879366,
-      "loss": 46.0,
-      "step": 40049
-    },
-    {
-      "epoch": 6.4497765610531825,
-      "grad_norm": 0.003529990091919899,
-      "learning_rate": 0.00019997948383405797,
-      "loss": 46.0,
-      "step": 40050
-    },
-    {
-      "epoch": 6.44993759813197,
-      "grad_norm": 0.004390238784253597,
-      "learning_rate": 0.00019997948280929672,
-      "loss": 46.0,
-      "step": 40051
-    },
-    {
-      "epoch": 6.450098635210757,
-      "grad_norm": 0.005980095360428095,
-      "learning_rate": 0.00019997948178450986,
-      "loss": 46.0,
-      "step": 40052
-    },
-    {
-      "epoch": 6.450259672289545,
-      "grad_norm": 0.005523684434592724,
-      "learning_rate": 0.0001999794807596974,
-      "loss": 46.0,
-      "step": 40053
-    },
-    {
-      "epoch": 6.450420709368332,
-      "grad_norm": 0.0012795368675142527,
-      "learning_rate": 0.00019997947973485937,
-      "loss": 46.0,
-      "step": 40054
-    },
-    {
-      "epoch": 6.45058174644712,
-      "grad_norm": 0.0007410626276396215,
-      "learning_rate": 0.00019997947870999575,
-      "loss": 46.0,
-      "step": 40055
-    },
-    {
-      "epoch": 6.450742783525907,
-      "grad_norm": 0.0017558034742251039,
-      "learning_rate": 0.00019997947768510654,
-      "loss": 46.0,
-      "step": 40056
-    },
-    {
-      "epoch": 6.450903820604694,
-      "grad_norm": 0.0038226377218961716,
-      "learning_rate": 0.00019997947666019174,
-      "loss": 46.0,
-      "step": 40057
-    },
-    {
-      "epoch": 6.451064857683481,
-      "grad_norm": 0.01705530285835266,
-      "learning_rate": 0.00019997947563525132,
-      "loss": 46.0,
-      "step": 40058
-    },
-    {
-      "epoch": 6.451225894762269,
-      "grad_norm": 0.009139101020991802,
-      "learning_rate": 0.00019997947461028535,
-      "loss": 46.0,
-      "step": 40059
-    },
-    {
-      "epoch": 6.451386931841056,
-      "grad_norm": 0.0032983715645968914,
-      "learning_rate": 0.0001999794735852938,
-      "loss": 46.0,
-      "step": 40060
-    },
-    {
-      "epoch": 6.451547968919844,
-      "grad_norm": 0.0030108338687568903,
-      "learning_rate": 0.0001999794725602766,
-      "loss": 46.0,
-      "step": 40061
-    },
-    {
-      "epoch": 6.451709005998631,
-      "grad_norm": 0.013655808754265308,
-      "learning_rate": 0.00019997947153523388,
-      "loss": 46.0,
-      "step": 40062
-    },
-    {
-      "epoch": 6.4518700430774185,
-      "grad_norm": 0.0029608544427901506,
-      "learning_rate": 0.00019997947051016553,
-      "loss": 46.0,
-      "step": 40063
-    },
-    {
-      "epoch": 6.452031080156206,
-      "grad_norm": 0.004351452458649874,
-      "learning_rate": 0.0001999794694850716,
-      "loss": 46.0,
-      "step": 40064
-    },
-    {
-      "epoch": 6.452192117234993,
-      "grad_norm": 0.005431823898106813,
-      "learning_rate": 0.00019997946845995207,
-      "loss": 46.0,
-      "step": 40065
-    },
-    {
-      "epoch": 6.452353154313781,
-      "grad_norm": 0.011322613805532455,
-      "learning_rate": 0.00019997946743480696,
-      "loss": 46.0,
-      "step": 40066
-    },
-    {
-      "epoch": 6.452514191392568,
-      "grad_norm": 0.002415593946352601,
-      "learning_rate": 0.00019997946640963626,
-      "loss": 46.0,
-      "step": 40067
-    },
-    {
-      "epoch": 6.452675228471356,
-      "grad_norm": 0.0038729137741029263,
-      "learning_rate": 0.00019997946538443997,
-      "loss": 46.0,
-      "step": 40068
-    },
-    {
-      "epoch": 6.452836265550143,
-      "grad_norm": 0.009057389572262764,
-      "learning_rate": 0.0001999794643592181,
-      "loss": 46.0,
-      "step": 40069
-    },
-    {
-      "epoch": 6.452997302628931,
-      "grad_norm": 0.022179776802659035,
-      "learning_rate": 0.00019997946333397061,
-      "loss": 46.0,
-      "step": 40070
-    },
-    {
-      "epoch": 6.453158339707718,
-      "grad_norm": 0.00583422789350152,
-      "learning_rate": 0.00019997946230869757,
-      "loss": 46.0,
-      "step": 40071
-    },
-    {
-      "epoch": 6.453319376786505,
-      "grad_norm": 0.005262425635010004,
-      "learning_rate": 0.00019997946128339893,
-      "loss": 46.0,
-      "step": 40072
-    },
-    {
-      "epoch": 6.453480413865292,
-      "grad_norm": 0.0029457639902830124,
-      "learning_rate": 0.00019997946025807469,
-      "loss": 46.0,
-      "step": 40073
-    },
-    {
-      "epoch": 6.45364145094408,
-      "grad_norm": 0.004011599812656641,
-      "learning_rate": 0.00019997945923272488,
-      "loss": 46.0,
-      "step": 40074
-    },
-    {
-      "epoch": 6.453802488022867,
-      "grad_norm": 0.002502941060811281,
-      "learning_rate": 0.00019997945820734945,
-      "loss": 46.0,
-      "step": 40075
-    },
-    {
-      "epoch": 6.4539635251016545,
-      "grad_norm": 0.0024388267192989588,
-      "learning_rate": 0.00019997945718194844,
-      "loss": 46.0,
-      "step": 40076
-    },
-    {
-      "epoch": 6.454124562180442,
-      "grad_norm": 0.00433287862688303,
-      "learning_rate": 0.00019997945615652185,
-      "loss": 46.0,
-      "step": 40077
-    },
-    {
-      "epoch": 6.454285599259229,
-      "grad_norm": 0.0031550717540085316,
-      "learning_rate": 0.00019997945513106966,
-      "loss": 46.0,
-      "step": 40078
-    },
-    {
-      "epoch": 6.454446636338017,
-      "grad_norm": 0.006372411735355854,
-      "learning_rate": 0.0001999794541055919,
-      "loss": 46.0,
-      "step": 40079
-    },
-    {
-      "epoch": 6.454607673416804,
-      "grad_norm": 0.0012201223289594054,
-      "learning_rate": 0.0001999794530800885,
-      "loss": 46.0,
-      "step": 40080
-    },
-    {
-      "epoch": 6.454768710495592,
-      "grad_norm": 0.001294377027079463,
-      "learning_rate": 0.00019997945205455956,
-      "loss": 46.0,
-      "step": 40081
-    },
-    {
-      "epoch": 6.454929747574379,
-      "grad_norm": 0.004865513648837805,
-      "learning_rate": 0.00019997945102900503,
-      "loss": 46.0,
-      "step": 40082
-    },
-    {
-      "epoch": 6.455090784653167,
-      "grad_norm": 0.0015484819887205958,
-      "learning_rate": 0.00019997945000342488,
-      "loss": 46.0,
-      "step": 40083
-    },
-    {
-      "epoch": 6.455251821731954,
-      "grad_norm": 0.0014008554862812161,
-      "learning_rate": 0.00019997944897781915,
-      "loss": 46.0,
-      "step": 40084
-    },
-    {
-      "epoch": 6.455412858810742,
-      "grad_norm": 0.004493103828281164,
-      "learning_rate": 0.00019997944795218783,
-      "loss": 46.0,
-      "step": 40085
-    },
-    {
-      "epoch": 6.455573895889529,
-      "grad_norm": 0.00622917665168643,
-      "learning_rate": 0.00019997944692653094,
-      "loss": 46.0,
-      "step": 40086
-    },
-    {
-      "epoch": 6.455734932968316,
-      "grad_norm": 0.0022607920691370964,
-      "learning_rate": 0.00019997944590084845,
-      "loss": 46.0,
-      "step": 40087
-    },
-    {
-      "epoch": 6.455895970047103,
-      "grad_norm": 0.002559753367677331,
-      "learning_rate": 0.00019997944487514037,
-      "loss": 46.0,
-      "step": 40088
-    },
-    {
-      "epoch": 6.4560570071258905,
-      "grad_norm": 0.0028425741475075483,
-      "learning_rate": 0.0001999794438494067,
-      "loss": 46.0,
-      "step": 40089
-    },
-    {
-      "epoch": 6.456218044204678,
-      "grad_norm": 0.0009906552731990814,
-      "learning_rate": 0.00019997944282364744,
-      "loss": 46.0,
-      "step": 40090
-    },
-    {
-      "epoch": 6.456379081283465,
-      "grad_norm": 0.019082501530647278,
-      "learning_rate": 0.0001999794417978626,
-      "loss": 46.0,
-      "step": 40091
-    },
-    {
-      "epoch": 6.456540118362253,
-      "grad_norm": 0.008088387548923492,
-      "learning_rate": 0.00019997944077205216,
-      "loss": 46.0,
-      "step": 40092
-    },
-    {
-      "epoch": 6.45670115544104,
-      "grad_norm": 0.0030749980360269547,
-      "learning_rate": 0.00019997943974621612,
-      "loss": 46.0,
-      "step": 40093
-    },
-    {
-      "epoch": 6.456862192519828,
-      "grad_norm": 0.0019023667555302382,
-      "learning_rate": 0.00019997943872035448,
-      "loss": 46.0,
-      "step": 40094
-    },
-    {
-      "epoch": 6.457023229598615,
-      "grad_norm": 0.00077101617353037,
-      "learning_rate": 0.0001999794376944673,
-      "loss": 46.0,
-      "step": 40095
-    },
-    {
-      "epoch": 6.457184266677403,
-      "grad_norm": 0.01156501192599535,
-      "learning_rate": 0.0001999794366685545,
-      "loss": 46.0,
-      "step": 40096
-    },
-    {
-      "epoch": 6.45734530375619,
-      "grad_norm": 0.003724921029061079,
-      "learning_rate": 0.00019997943564261612,
-      "loss": 46.0,
-      "step": 40097
-    },
-    {
-      "epoch": 6.4575063408349775,
-      "grad_norm": 0.013687696307897568,
-      "learning_rate": 0.00019997943461665213,
-      "loss": 46.0,
-      "step": 40098
-    },
-    {
-      "epoch": 6.457667377913765,
-      "grad_norm": 0.007870011031627655,
-      "learning_rate": 0.00019997943359066256,
-      "loss": 46.0,
-      "step": 40099
-    },
-    {
-      "epoch": 6.4578284149925524,
-      "grad_norm": 0.0033407360315322876,
-      "learning_rate": 0.0001999794325646474,
-      "loss": 46.0,
-      "step": 40100
-    },
-    {
-      "epoch": 6.45798945207134,
-      "grad_norm": 0.015944834798574448,
-      "learning_rate": 0.00019997943153860667,
-      "loss": 46.0,
-      "step": 40101
-    },
-    {
-      "epoch": 6.4581504891501265,
-      "grad_norm": 0.0017989519983530045,
-      "learning_rate": 0.00019997943051254033,
-      "loss": 46.0,
-      "step": 40102
-    },
-    {
-      "epoch": 6.458311526228914,
-      "grad_norm": 0.010646210052073002,
-      "learning_rate": 0.00019997942948644842,
-      "loss": 46.0,
-      "step": 40103
-    },
-    {
-      "epoch": 6.458472563307701,
-      "grad_norm": 0.004232795909047127,
-      "learning_rate": 0.00019997942846033086,
-      "loss": 46.0,
-      "step": 40104
-    },
-    {
-      "epoch": 6.458633600386489,
-      "grad_norm": 0.0007176382350735366,
-      "learning_rate": 0.0001999794274341878,
-      "loss": 46.0,
-      "step": 40105
-    },
-    {
-      "epoch": 6.458794637465276,
-      "grad_norm": 0.0025290539488196373,
-      "learning_rate": 0.0001999794264080191,
-      "loss": 46.0,
-      "step": 40106
-    },
-    {
-      "epoch": 6.458955674544064,
-      "grad_norm": 0.013825730420649052,
-      "learning_rate": 0.0001999794253818248,
-      "loss": 46.0,
-      "step": 40107
-    },
-    {
-      "epoch": 6.459116711622851,
-      "grad_norm": 0.002387827495113015,
-      "learning_rate": 0.00019997942435560494,
-      "loss": 46.0,
-      "step": 40108
-    },
-    {
-      "epoch": 6.459277748701639,
-      "grad_norm": 0.010204867459833622,
-      "learning_rate": 0.00019997942332935947,
-      "loss": 46.0,
-      "step": 40109
-    },
-    {
-      "epoch": 6.459438785780426,
-      "grad_norm": 0.003758529433980584,
-      "learning_rate": 0.00019997942230308842,
-      "loss": 46.0,
-      "step": 40110
-    },
-    {
-      "epoch": 6.4595998228592135,
-      "grad_norm": 0.004909013397991657,
-      "learning_rate": 0.00019997942127679178,
-      "loss": 46.0,
-      "step": 40111
-    },
-    {
-      "epoch": 6.459760859938001,
-      "grad_norm": 0.001407152391038835,
-      "learning_rate": 0.00019997942025046955,
-      "loss": 46.0,
-      "step": 40112
-    },
-    {
-      "epoch": 6.459921897016788,
-      "grad_norm": 0.005716997664421797,
-      "learning_rate": 0.0001999794192241217,
-      "loss": 46.0,
-      "step": 40113
-    },
-    {
-      "epoch": 6.460082934095576,
-      "grad_norm": 0.0028513907454907894,
-      "learning_rate": 0.0001999794181977483,
-      "loss": 46.0,
-      "step": 40114
-    },
-    {
-      "epoch": 6.460243971174362,
-      "grad_norm": 0.0013694504741579294,
-      "learning_rate": 0.00019997941717134928,
-      "loss": 46.0,
-      "step": 40115
-    },
-    {
-      "epoch": 6.46040500825315,
-      "grad_norm": 0.002183281583711505,
-      "learning_rate": 0.0001999794161449247,
-      "loss": 46.0,
-      "step": 40116
-    },
-    {
-      "epoch": 6.460566045331937,
-      "grad_norm": 0.014641164802014828,
-      "learning_rate": 0.00019997941511847454,
-      "loss": 46.0,
-      "step": 40117
-    },
-    {
-      "epoch": 6.460727082410725,
-      "grad_norm": 0.002976014045998454,
-      "learning_rate": 0.00019997941409199876,
-      "loss": 46.0,
-      "step": 40118
-    },
-    {
-      "epoch": 6.460888119489512,
-      "grad_norm": 0.006201675161719322,
-      "learning_rate": 0.00019997941306549742,
-      "loss": 46.0,
-      "step": 40119
-    },
-    {
-      "epoch": 6.4610491565683,
-      "grad_norm": 0.002512093400582671,
-      "learning_rate": 0.00019997941203897047,
-      "loss": 46.0,
-      "step": 40120
-    },
-    {
-      "epoch": 6.461210193647087,
-      "grad_norm": 0.0016711759380996227,
-      "learning_rate": 0.0001999794110124179,
-      "loss": 46.0,
-      "step": 40121
-    },
-    {
-      "epoch": 6.461371230725875,
-      "grad_norm": 0.0021573281846940517,
-      "learning_rate": 0.00019997940998583977,
-      "loss": 46.0,
-      "step": 40122
-    },
-    {
-      "epoch": 6.461532267804662,
-      "grad_norm": 0.003349128644913435,
-      "learning_rate": 0.00019997940895923608,
-      "loss": 46.0,
-      "step": 40123
-    },
-    {
-      "epoch": 6.4616933048834495,
-      "grad_norm": 0.006997830234467983,
-      "learning_rate": 0.00019997940793260678,
-      "loss": 46.0,
-      "step": 40124
-    },
-    {
-      "epoch": 6.461854341962237,
-      "grad_norm": 0.0030931050423532724,
-      "learning_rate": 0.00019997940690595186,
-      "loss": 46.0,
-      "step": 40125
-    },
-    {
-      "epoch": 6.462015379041024,
-      "grad_norm": 0.008127564564347267,
-      "learning_rate": 0.00019997940587927139,
-      "loss": 46.0,
-      "step": 40126
-    },
-    {
-      "epoch": 6.462176416119812,
-      "grad_norm": 0.005348661914467812,
-      "learning_rate": 0.0001999794048525653,
-      "loss": 46.0,
-      "step": 40127
-    },
-    {
-      "epoch": 6.462337453198599,
-      "grad_norm": 0.0026755311992019415,
-      "learning_rate": 0.00019997940382583362,
-      "loss": 46.0,
-      "step": 40128
-    },
-    {
-      "epoch": 6.462498490277387,
-      "grad_norm": 0.0032196207903325558,
-      "learning_rate": 0.00019997940279907638,
-      "loss": 46.0,
-      "step": 40129
-    },
-    {
-      "epoch": 6.462659527356173,
-      "grad_norm": 0.003318954259157181,
-      "learning_rate": 0.00019997940177229353,
-      "loss": 46.0,
-      "step": 40130
-    },
-    {
-      "epoch": 6.462820564434961,
-      "grad_norm": 0.002483679447323084,
-      "learning_rate": 0.0001999794007454851,
-      "loss": 46.0,
-      "step": 40131
-    },
-    {
-      "epoch": 6.462981601513748,
-      "grad_norm": 0.004994211718440056,
-      "learning_rate": 0.00019997939971865106,
-      "loss": 46.0,
-      "step": 40132
-    },
-    {
-      "epoch": 6.463142638592536,
-      "grad_norm": 0.014873869717121124,
-      "learning_rate": 0.00019997939869179145,
-      "loss": 46.0,
-      "step": 40133
-    },
-    {
-      "epoch": 6.463303675671323,
-      "grad_norm": 0.0029470722656697035,
-      "learning_rate": 0.00019997939766490624,
-      "loss": 46.0,
-      "step": 40134
-    },
-    {
-      "epoch": 6.463464712750111,
-      "grad_norm": 0.0030978533904999495,
-      "learning_rate": 0.00019997939663799548,
-      "loss": 46.0,
-      "step": 40135
-    },
-    {
-      "epoch": 6.463625749828898,
-      "grad_norm": 0.002906282665207982,
-      "learning_rate": 0.00019997939561105908,
-      "loss": 46.0,
-      "step": 40136
-    },
-    {
-      "epoch": 6.4637867869076855,
-      "grad_norm": 0.0015427491161972284,
-      "learning_rate": 0.00019997939458409712,
-      "loss": 46.0,
-      "step": 40137
-    },
-    {
-      "epoch": 6.463947823986473,
-      "grad_norm": 0.0049125440418720245,
-      "learning_rate": 0.00019997939355710954,
-      "loss": 46.0,
-      "step": 40138
-    },
-    {
-      "epoch": 6.46410886106526,
-      "grad_norm": 0.012352939695119858,
-      "learning_rate": 0.00019997939253009638,
-      "loss": 46.0,
-      "step": 40139
-    },
-    {
-      "epoch": 6.464269898144048,
-      "grad_norm": 0.005172920413315296,
-      "learning_rate": 0.00019997939150305768,
-      "loss": 46.0,
-      "step": 40140
-    },
-    {
-      "epoch": 6.464430935222835,
-      "grad_norm": 0.001561078941449523,
-      "learning_rate": 0.0001999793904759933,
-      "loss": 46.0,
-      "step": 40141
-    },
-    {
-      "epoch": 6.464591972301623,
-      "grad_norm": 0.0011646532220765948,
-      "learning_rate": 0.0001999793894489034,
-      "loss": 46.0,
-      "step": 40142
-    },
-    {
-      "epoch": 6.46475300938041,
-      "grad_norm": 0.0011256361613050103,
-      "learning_rate": 0.0001999793884217879,
-      "loss": 46.0,
-      "step": 40143
-    },
-    {
-      "epoch": 6.464914046459198,
-      "grad_norm": 0.0028008469380438328,
-      "learning_rate": 0.0001999793873946468,
-      "loss": 46.0,
-      "step": 40144
-    },
-    {
-      "epoch": 6.465075083537984,
-      "grad_norm": 0.006941705942153931,
-      "learning_rate": 0.00019997938636748011,
-      "loss": 46.0,
-      "step": 40145
-    },
-    {
-      "epoch": 6.465236120616772,
-      "grad_norm": 0.00401967391371727,
-      "learning_rate": 0.0001999793853402878,
-      "loss": 46.0,
-      "step": 40146
-    },
-    {
-      "epoch": 6.465397157695559,
-      "grad_norm": 0.017262045294046402,
-      "learning_rate": 0.00019997938431306995,
-      "loss": 46.0,
-      "step": 40147
-    },
-    {
-      "epoch": 6.465558194774347,
-      "grad_norm": 0.0008813973981887102,
-      "learning_rate": 0.0001999793832858265,
-      "loss": 46.0,
-      "step": 40148
-    },
-    {
-      "epoch": 6.465719231853134,
-      "grad_norm": 0.008579926565289497,
-      "learning_rate": 0.00019997938225855747,
-      "loss": 46.0,
-      "step": 40149
-    },
-    {
-      "epoch": 6.4658802689319215,
-      "grad_norm": 0.004942048341035843,
-      "learning_rate": 0.00019997938123126282,
-      "loss": 46.0,
-      "step": 40150
-    },
-    {
-      "epoch": 6.466041306010709,
-      "grad_norm": 0.0009261569357477129,
-      "learning_rate": 0.0001999793802039426,
-      "loss": 46.0,
-      "step": 40151
-    },
-    {
-      "epoch": 6.466202343089496,
-      "grad_norm": 0.008591445162892342,
-      "learning_rate": 0.00019997937917659678,
-      "loss": 46.0,
-      "step": 40152
-    },
-    {
-      "epoch": 6.466363380168284,
-      "grad_norm": 0.0028812282253056765,
-      "learning_rate": 0.00019997937814922537,
-      "loss": 46.0,
-      "step": 40153
-    },
-    {
-      "epoch": 6.466524417247071,
-      "grad_norm": 0.0044649397023022175,
-      "learning_rate": 0.00019997937712182837,
-      "loss": 46.0,
-      "step": 40154
-    },
-    {
-      "epoch": 6.466685454325859,
-      "grad_norm": 0.0053070043213665485,
-      "learning_rate": 0.0001999793760944058,
-      "loss": 46.0,
-      "step": 40155
-    },
-    {
-      "epoch": 6.466846491404646,
-      "grad_norm": 0.0022988743148744106,
-      "learning_rate": 0.00019997937506695764,
-      "loss": 46.0,
-      "step": 40156
-    },
-    {
-      "epoch": 6.467007528483434,
-      "grad_norm": 0.0038261639419943094,
-      "learning_rate": 0.00019997937403948385,
-      "loss": 46.0,
-      "step": 40157
-    },
-    {
-      "epoch": 6.467168565562221,
-      "grad_norm": 0.004226604010909796,
-      "learning_rate": 0.0001999793730119845,
-      "loss": 46.0,
-      "step": 40158
-    },
-    {
-      "epoch": 6.4673296026410085,
-      "grad_norm": 0.002454396802932024,
-      "learning_rate": 0.00019997937198445954,
-      "loss": 46.0,
-      "step": 40159
-    },
-    {
-      "epoch": 6.467490639719795,
-      "grad_norm": 0.002462859731167555,
-      "learning_rate": 0.00019997937095690902,
-      "loss": 46.0,
-      "step": 40160
-    },
-    {
-      "epoch": 6.4676516767985825,
-      "grad_norm": 0.0024687927216291428,
-      "learning_rate": 0.0001999793699293329,
-      "loss": 46.0,
-      "step": 40161
-    },
-    {
-      "epoch": 6.46781271387737,
-      "grad_norm": 0.004976825788617134,
-      "learning_rate": 0.0001999793689017312,
-      "loss": 46.0,
-      "step": 40162
-    },
-    {
-      "epoch": 6.4679737509561575,
-      "grad_norm": 0.0007814789423719049,
-      "learning_rate": 0.0001999793678741039,
-      "loss": 46.0,
-      "step": 40163
-    },
-    {
-      "epoch": 6.468134788034945,
-      "grad_norm": 0.001508883317001164,
-      "learning_rate": 0.000199979366846451,
-      "loss": 46.0,
-      "step": 40164
-    },
-    {
-      "epoch": 6.468295825113732,
-      "grad_norm": 0.0031199741642922163,
-      "learning_rate": 0.00019997936581877252,
-      "loss": 46.0,
-      "step": 40165
-    },
-    {
-      "epoch": 6.46845686219252,
-      "grad_norm": 0.0024616036098450422,
-      "learning_rate": 0.00019997936479106845,
-      "loss": 46.0,
-      "step": 40166
-    },
-    {
-      "epoch": 6.468617899271307,
-      "grad_norm": 0.013303125277161598,
-      "learning_rate": 0.0001999793637633388,
-      "loss": 46.0,
-      "step": 40167
-    },
-    {
-      "epoch": 6.468778936350095,
-      "grad_norm": 0.002328054280951619,
-      "learning_rate": 0.00019997936273558352,
-      "loss": 46.0,
-      "step": 40168
-    },
-    {
-      "epoch": 6.468939973428882,
-      "grad_norm": 0.001860763761214912,
-      "learning_rate": 0.00019997936170780268,
-      "loss": 46.0,
-      "step": 40169
-    },
-    {
-      "epoch": 6.46910101050767,
-      "grad_norm": 0.0043235779739916325,
-      "learning_rate": 0.00019997936067999626,
-      "loss": 46.0,
-      "step": 40170
-    },
-    {
-      "epoch": 6.469262047586457,
-      "grad_norm": 0.0044525377452373505,
-      "learning_rate": 0.00019997935965216426,
-      "loss": 46.0,
-      "step": 40171
-    },
-    {
-      "epoch": 6.4694230846652445,
-      "grad_norm": 0.0010528952116146684,
-      "learning_rate": 0.00019997935862430666,
-      "loss": 46.0,
-      "step": 40172
-    },
-    {
-      "epoch": 6.469584121744032,
-      "grad_norm": 0.0032428554259240627,
-      "learning_rate": 0.00019997935759642345,
-      "loss": 46.0,
-      "step": 40173
-    },
-    {
-      "epoch": 6.469745158822819,
-      "grad_norm": 0.0042559741996228695,
-      "learning_rate": 0.00019997935656851468,
-      "loss": 46.0,
-      "step": 40174
-    },
-    {
-      "epoch": 6.469906195901606,
-      "grad_norm": 0.0012639843625947833,
-      "learning_rate": 0.00019997935554058027,
-      "loss": 46.0,
-      "step": 40175
-    },
-    {
-      "epoch": 6.470067232980393,
-      "grad_norm": 0.001522873411886394,
-      "learning_rate": 0.00019997935451262033,
-      "loss": 46.0,
-      "step": 40176
-    },
-    {
-      "epoch": 6.470228270059181,
-      "grad_norm": 0.009718555957078934,
-      "learning_rate": 0.00019997935348463477,
-      "loss": 46.0,
-      "step": 40177
-    },
-    {
-      "epoch": 6.470389307137968,
-      "grad_norm": 0.005866462830454111,
-      "learning_rate": 0.00019997935245662362,
-      "loss": 46.0,
-      "step": 40178
-    },
-    {
-      "epoch": 6.470550344216756,
-      "grad_norm": 0.0020871348679065704,
-      "learning_rate": 0.0001999793514285869,
-      "loss": 46.0,
-      "step": 40179
-    },
-    {
-      "epoch": 6.470711381295543,
-      "grad_norm": 0.00416685501113534,
-      "learning_rate": 0.00019997935040052455,
-      "loss": 46.0,
-      "step": 40180
-    },
-    {
-      "epoch": 6.470872418374331,
-      "grad_norm": 0.006891305558383465,
-      "learning_rate": 0.00019997934937243664,
-      "loss": 46.0,
-      "step": 40181
-    },
-    {
-      "epoch": 6.471033455453118,
-      "grad_norm": 0.005611291155219078,
-      "learning_rate": 0.00019997934834432312,
-      "loss": 46.0,
-      "step": 40182
-    },
-    {
-      "epoch": 6.471194492531906,
-      "grad_norm": 0.0030021113343536854,
-      "learning_rate": 0.00019997934731618404,
-      "loss": 46.0,
-      "step": 40183
-    },
-    {
-      "epoch": 6.471355529610693,
-      "grad_norm": 0.005294486880302429,
-      "learning_rate": 0.00019997934628801937,
-      "loss": 46.0,
-      "step": 40184
-    },
-    {
-      "epoch": 6.4715165666894805,
-      "grad_norm": 0.002153391018509865,
-      "learning_rate": 0.00019997934525982909,
-      "loss": 46.0,
-      "step": 40185
-    },
-    {
-      "epoch": 6.471677603768268,
-      "grad_norm": 0.004230473656207323,
-      "learning_rate": 0.00019997934423161322,
-      "loss": 46.0,
-      "step": 40186
-    },
-    {
-      "epoch": 6.471838640847055,
-      "grad_norm": 0.0013075951719656587,
-      "learning_rate": 0.00019997934320337176,
-      "loss": 46.0,
-      "step": 40187
-    },
-    {
-      "epoch": 6.471999677925842,
-      "grad_norm": 0.015328380279242992,
-      "learning_rate": 0.00019997934217510474,
-      "loss": 46.0,
-      "step": 40188
-    },
-    {
-      "epoch": 6.472160715004629,
-      "grad_norm": 0.0030476991087198257,
-      "learning_rate": 0.00019997934114681208,
-      "loss": 46.0,
-      "step": 40189
-    },
-    {
-      "epoch": 6.472321752083417,
-      "grad_norm": 0.00534987635910511,
-      "learning_rate": 0.0001999793401184939,
-      "loss": 46.0,
-      "step": 40190
-    },
-    {
-      "epoch": 6.472482789162204,
-      "grad_norm": 0.00708777317777276,
-      "learning_rate": 0.0001999793390901501,
-      "loss": 46.0,
-      "step": 40191
-    },
-    {
-      "epoch": 6.472643826240992,
-      "grad_norm": 0.0036839821841567755,
-      "learning_rate": 0.00019997933806178067,
-      "loss": 46.0,
-      "step": 40192
-    },
-    {
-      "epoch": 6.472804863319779,
-      "grad_norm": 0.0024663747753947973,
-      "learning_rate": 0.0001999793370333857,
-      "loss": 46.0,
-      "step": 40193
-    },
-    {
-      "epoch": 6.472965900398567,
-      "grad_norm": 0.006557401269674301,
-      "learning_rate": 0.0001999793360049651,
-      "loss": 46.0,
-      "step": 40194
-    },
-    {
-      "epoch": 6.473126937477354,
-      "grad_norm": 0.006940734572708607,
-      "learning_rate": 0.00019997933497651894,
-      "loss": 46.0,
-      "step": 40195
-    },
-    {
-      "epoch": 6.473287974556142,
-      "grad_norm": 0.004655441269278526,
-      "learning_rate": 0.00019997933394804717,
-      "loss": 46.0,
-      "step": 40196
-    },
-    {
-      "epoch": 6.473449011634929,
-      "grad_norm": 0.002498186891898513,
-      "learning_rate": 0.00019997933291954984,
-      "loss": 46.0,
-      "step": 40197
-    },
-    {
-      "epoch": 6.4736100487137165,
-      "grad_norm": 0.006033514626324177,
-      "learning_rate": 0.0001999793318910269,
-      "loss": 46.0,
-      "step": 40198
-    },
-    {
-      "epoch": 6.473771085792504,
-      "grad_norm": 0.002992918249219656,
-      "learning_rate": 0.00019997933086247837,
-      "loss": 46.0,
-      "step": 40199
-    },
-    {
-      "epoch": 6.473932122871291,
-      "grad_norm": 0.0022429730743169785,
-      "learning_rate": 0.00019997932983390425,
-      "loss": 46.0,
-      "step": 40200
-    },
-    {
-      "epoch": 6.474093159950079,
-      "grad_norm": 0.012457989156246185,
-      "learning_rate": 0.00019997932880530455,
-      "loss": 46.0,
-      "step": 40201
-    },
-    {
-      "epoch": 6.474254197028866,
-      "grad_norm": 0.0008719643228687346,
-      "learning_rate": 0.00019997932777667928,
-      "loss": 46.0,
-      "step": 40202
-    },
-    {
-      "epoch": 6.474415234107653,
-      "grad_norm": 0.0030901897698640823,
-      "learning_rate": 0.00019997932674802838,
-      "loss": 46.0,
-      "step": 40203
-    },
-    {
-      "epoch": 6.47457627118644,
-      "grad_norm": 0.016922269016504288,
-      "learning_rate": 0.00019997932571935189,
-      "loss": 46.0,
-      "step": 40204
-    },
-    {
-      "epoch": 6.474737308265228,
-      "grad_norm": 0.0035538964439183474,
-      "learning_rate": 0.00019997932469064983,
-      "loss": 46.0,
-      "step": 40205
-    },
-    {
-      "epoch": 6.474898345344015,
-      "grad_norm": 0.008031846955418587,
-      "learning_rate": 0.0001999793236619222,
-      "loss": 46.0,
-      "step": 40206
-    },
-    {
-      "epoch": 6.475059382422803,
-      "grad_norm": 0.004215062130242586,
-      "learning_rate": 0.00019997932263316894,
-      "loss": 46.0,
-      "step": 40207
-    },
-    {
-      "epoch": 6.47522041950159,
-      "grad_norm": 0.006000776309520006,
-      "learning_rate": 0.0001999793216043901,
-      "loss": 46.0,
-      "step": 40208
-    },
-    {
-      "epoch": 6.475381456580378,
-      "grad_norm": 0.005052017979323864,
-      "learning_rate": 0.0001999793205755857,
-      "loss": 46.0,
-      "step": 40209
-    },
-    {
-      "epoch": 6.475542493659165,
-      "grad_norm": 0.009588254615664482,
-      "learning_rate": 0.00019997931954675568,
-      "loss": 46.0,
-      "step": 40210
-    },
-    {
-      "epoch": 6.4757035307379525,
-      "grad_norm": 0.0022454005666077137,
-      "learning_rate": 0.00019997931851790005,
-      "loss": 46.0,
-      "step": 40211
-    },
-    {
-      "epoch": 6.47586456781674,
-      "grad_norm": 0.0036979801952838898,
-      "learning_rate": 0.00019997931748901888,
-      "loss": 46.0,
-      "step": 40212
-    },
-    {
-      "epoch": 6.476025604895527,
-      "grad_norm": 0.006309863179922104,
-      "learning_rate": 0.0001999793164601121,
-      "loss": 46.0,
-      "step": 40213
-    },
-    {
-      "epoch": 6.476186641974315,
-      "grad_norm": 0.008912652730941772,
-      "learning_rate": 0.00019997931543117972,
-      "loss": 46.0,
-      "step": 40214
-    },
-    {
-      "epoch": 6.476347679053102,
-      "grad_norm": 0.002049471717327833,
-      "learning_rate": 0.00019997931440222176,
-      "loss": 46.0,
-      "step": 40215
-    },
-    {
-      "epoch": 6.47650871613189,
-      "grad_norm": 0.003810411086305976,
-      "learning_rate": 0.0001999793133732382,
-      "loss": 46.0,
-      "step": 40216
-    },
-    {
-      "epoch": 6.476669753210677,
-      "grad_norm": 0.009678413160145283,
-      "learning_rate": 0.00019997931234422907,
-      "loss": 46.0,
-      "step": 40217
-    },
-    {
-      "epoch": 6.476830790289464,
-      "grad_norm": 0.007415224332362413,
-      "learning_rate": 0.00019997931131519433,
-      "loss": 46.0,
-      "step": 40218
-    },
-    {
-      "epoch": 6.476991827368251,
-      "grad_norm": 0.0014222897589206696,
-      "learning_rate": 0.00019997931028613403,
-      "loss": 46.0,
-      "step": 40219
-    },
-    {
-      "epoch": 6.477152864447039,
-      "grad_norm": 0.003049605991691351,
-      "learning_rate": 0.00019997930925704812,
-      "loss": 46.0,
-      "step": 40220
-    },
-    {
-      "epoch": 6.477313901525826,
-      "grad_norm": 0.0019180566305294633,
-      "learning_rate": 0.0001999793082279366,
-      "loss": 46.0,
-      "step": 40221
-    },
-    {
-      "epoch": 6.4774749386046135,
-      "grad_norm": 0.0035815176088362932,
-      "learning_rate": 0.00019997930719879952,
-      "loss": 46.0,
-      "step": 40222
-    },
-    {
-      "epoch": 6.477635975683401,
-      "grad_norm": 0.004851811565458775,
-      "learning_rate": 0.00019997930616963685,
-      "loss": 46.0,
-      "step": 40223
-    },
-    {
-      "epoch": 6.4777970127621884,
-      "grad_norm": 0.007060496136546135,
-      "learning_rate": 0.0001999793051404486,
-      "loss": 46.0,
-      "step": 40224
-    },
-    {
-      "epoch": 6.477958049840976,
-      "grad_norm": 0.007451748475432396,
-      "learning_rate": 0.00019997930411123473,
-      "loss": 46.0,
-      "step": 40225
-    },
-    {
-      "epoch": 6.478119086919763,
-      "grad_norm": 0.002310977317392826,
-      "learning_rate": 0.0001999793030819953,
-      "loss": 46.0,
-      "step": 40226
-    },
-    {
-      "epoch": 6.478280123998551,
-      "grad_norm": 0.0038187282625585794,
-      "learning_rate": 0.00019997930205273024,
-      "loss": 46.0,
-      "step": 40227
-    },
-    {
-      "epoch": 6.478441161077338,
-      "grad_norm": 0.005842072889208794,
-      "learning_rate": 0.00019997930102343966,
-      "loss": 46.0,
-      "step": 40228
-    },
-    {
-      "epoch": 6.478602198156126,
-      "grad_norm": 0.006124195642769337,
-      "learning_rate": 0.0001999792999941234,
-      "loss": 46.0,
-      "step": 40229
-    },
-    {
-      "epoch": 6.478763235234913,
-      "grad_norm": 0.0029630945064127445,
-      "learning_rate": 0.00019997929896478162,
-      "loss": 46.0,
-      "step": 40230
-    },
-    {
-      "epoch": 6.478924272313701,
-      "grad_norm": 0.0009804241126403213,
-      "learning_rate": 0.00019997929793541425,
-      "loss": 46.0,
-      "step": 40231
-    },
-    {
-      "epoch": 6.479085309392488,
-      "grad_norm": 0.0014537015231326222,
-      "learning_rate": 0.00019997929690602123,
-      "loss": 46.0,
-      "step": 40232
-    },
-    {
-      "epoch": 6.479246346471275,
-      "grad_norm": 0.0010318144923076034,
-      "learning_rate": 0.00019997929587660268,
-      "loss": 46.0,
-      "step": 40233
-    },
-    {
-      "epoch": 6.479407383550062,
-      "grad_norm": 0.0032097308430820704,
-      "learning_rate": 0.00019997929484715852,
-      "loss": 46.0,
-      "step": 40234
-    },
-    {
-      "epoch": 6.4795684206288495,
-      "grad_norm": 0.0027899506967514753,
-      "learning_rate": 0.00019997929381768877,
-      "loss": 46.0,
-      "step": 40235
-    },
-    {
-      "epoch": 6.479729457707637,
-      "grad_norm": 0.007092115934938192,
-      "learning_rate": 0.00019997929278819343,
-      "loss": 46.0,
-      "step": 40236
-    },
-    {
-      "epoch": 6.479890494786424,
-      "grad_norm": 0.010679597035050392,
-      "learning_rate": 0.00019997929175867248,
-      "loss": 46.0,
-      "step": 40237
-    },
-    {
-      "epoch": 6.480051531865212,
-      "grad_norm": 0.0014577755937352777,
-      "learning_rate": 0.00019997929072912597,
-      "loss": 46.0,
-      "step": 40238
-    },
-    {
-      "epoch": 6.480212568943999,
-      "grad_norm": 0.004734776448458433,
-      "learning_rate": 0.00019997928969955387,
-      "loss": 46.0,
-      "step": 40239
-    },
-    {
-      "epoch": 6.480373606022787,
-      "grad_norm": 0.011812287382781506,
-      "learning_rate": 0.00019997928866995616,
-      "loss": 46.0,
-      "step": 40240
-    },
-    {
-      "epoch": 6.480534643101574,
-      "grad_norm": 0.006357528269290924,
-      "learning_rate": 0.00019997928764033286,
-      "loss": 46.0,
-      "step": 40241
-    },
-    {
-      "epoch": 6.480695680180362,
-      "grad_norm": 0.01355251669883728,
-      "learning_rate": 0.000199979286610684,
-      "loss": 46.0,
-      "step": 40242
-    },
-    {
-      "epoch": 6.480856717259149,
-      "grad_norm": 0.001603866694495082,
-      "learning_rate": 0.0001999792855810095,
-      "loss": 46.0,
-      "step": 40243
-    },
-    {
-      "epoch": 6.481017754337937,
-      "grad_norm": 0.02170816995203495,
-      "learning_rate": 0.00019997928455130947,
-      "loss": 46.0,
-      "step": 40244
-    },
-    {
-      "epoch": 6.481178791416724,
-      "grad_norm": 0.00755128962919116,
-      "learning_rate": 0.00019997928352158382,
-      "loss": 46.0,
-      "step": 40245
-    },
-    {
-      "epoch": 6.4813398284955115,
-      "grad_norm": 0.005113938823342323,
-      "learning_rate": 0.0001999792824918326,
-      "loss": 46.0,
-      "step": 40246
-    },
-    {
-      "epoch": 6.481500865574299,
-      "grad_norm": 0.0018067056080326438,
-      "learning_rate": 0.00019997928146205576,
-      "loss": 46.0,
-      "step": 40247
-    },
-    {
-      "epoch": 6.4816619026530855,
-      "grad_norm": 0.005931834224611521,
-      "learning_rate": 0.00019997928043225332,
-      "loss": 46.0,
-      "step": 40248
-    },
-    {
-      "epoch": 6.481822939731873,
-      "grad_norm": 0.0011589303612709045,
-      "learning_rate": 0.00019997927940242533,
-      "loss": 46.0,
-      "step": 40249
-    },
-    {
-      "epoch": 6.48198397681066,
-      "grad_norm": 0.005806329660117626,
-      "learning_rate": 0.00019997927837257174,
-      "loss": 46.0,
-      "step": 40250
-    },
-    {
-      "epoch": 6.482145013889448,
-      "grad_norm": 0.003034749301150441,
-      "learning_rate": 0.00019997927734269254,
-      "loss": 46.0,
-      "step": 40251
-    },
-    {
-      "epoch": 6.482306050968235,
-      "grad_norm": 0.009186151437461376,
-      "learning_rate": 0.00019997927631278779,
-      "loss": 46.0,
-      "step": 40252
-    },
-    {
-      "epoch": 6.482467088047023,
-      "grad_norm": 0.008191969245672226,
-      "learning_rate": 0.0001999792752828574,
-      "loss": 46.0,
-      "step": 40253
-    },
-    {
-      "epoch": 6.48262812512581,
-      "grad_norm": 0.0043816883116960526,
-      "learning_rate": 0.00019997927425290145,
-      "loss": 46.0,
-      "step": 40254
-    },
-    {
-      "epoch": 6.482789162204598,
-      "grad_norm": 0.016037199646234512,
-      "learning_rate": 0.00019997927322291993,
-      "loss": 46.0,
-      "step": 40255
-    },
-    {
-      "epoch": 6.482950199283385,
-      "grad_norm": 0.012546764686703682,
-      "learning_rate": 0.0001999792721929128,
-      "loss": 46.0,
-      "step": 40256
-    },
-    {
-      "epoch": 6.483111236362173,
-      "grad_norm": 0.0015752743929624557,
-      "learning_rate": 0.00019997927116288008,
-      "loss": 46.0,
-      "step": 40257
-    },
-    {
-      "epoch": 6.48327227344096,
-      "grad_norm": 0.0016135724727064371,
-      "learning_rate": 0.00019997927013282177,
-      "loss": 46.0,
-      "step": 40258
-    },
-    {
-      "epoch": 6.4834333105197475,
-      "grad_norm": 0.011241168715059757,
-      "learning_rate": 0.00019997926910273787,
-      "loss": 46.0,
-      "step": 40259
-    },
-    {
-      "epoch": 6.483594347598535,
-      "grad_norm": 0.0033009974285960197,
-      "learning_rate": 0.00019997926807262836,
-      "loss": 46.0,
-      "step": 40260
-    },
-    {
-      "epoch": 6.4837553846773215,
-      "grad_norm": 0.0037363627925515175,
-      "learning_rate": 0.00019997926704249327,
-      "loss": 46.0,
-      "step": 40261
-    },
-    {
-      "epoch": 6.483916421756109,
-      "grad_norm": 0.01179394032806158,
-      "learning_rate": 0.0001999792660123326,
-      "loss": 46.0,
-      "step": 40262
-    },
-    {
-      "epoch": 6.484077458834896,
-      "grad_norm": 0.0009958798764273524,
-      "learning_rate": 0.00019997926498214636,
-      "loss": 46.0,
-      "step": 40263
-    },
-    {
-      "epoch": 6.484238495913684,
-      "grad_norm": 0.004711686633527279,
-      "learning_rate": 0.0001999792639519345,
-      "loss": 46.0,
-      "step": 40264
-    },
-    {
-      "epoch": 6.484399532992471,
-      "grad_norm": 0.002110701985657215,
-      "learning_rate": 0.00019997926292169709,
-      "loss": 46.0,
-      "step": 40265
-    },
-    {
-      "epoch": 6.484560570071259,
-      "grad_norm": 0.014985272660851479,
-      "learning_rate": 0.00019997926189143405,
-      "loss": 46.0,
-      "step": 40266
-    },
-    {
-      "epoch": 6.484721607150046,
-      "grad_norm": 0.006289361044764519,
-      "learning_rate": 0.00019997926086114543,
-      "loss": 46.0,
-      "step": 40267
-    },
-    {
-      "epoch": 6.484882644228834,
-      "grad_norm": 0.0019607581198215485,
-      "learning_rate": 0.0001999792598308312,
-      "loss": 46.0,
-      "step": 40268
-    },
-    {
-      "epoch": 6.485043681307621,
-      "grad_norm": 0.005681016948074102,
-      "learning_rate": 0.00019997925880049143,
-      "loss": 46.0,
-      "step": 40269
-    },
-    {
-      "epoch": 6.4852047183864086,
-      "grad_norm": 0.0031340315472334623,
-      "learning_rate": 0.00019997925777012605,
-      "loss": 46.0,
-      "step": 40270
-    },
-    {
-      "epoch": 6.485365755465196,
-      "grad_norm": 0.003052378771826625,
-      "learning_rate": 0.00019997925673973508,
-      "loss": 46.0,
-      "step": 40271
-    },
-    {
-      "epoch": 6.4855267925439835,
-      "grad_norm": 0.00330049148760736,
-      "learning_rate": 0.0001999792557093185,
-      "loss": 46.0,
-      "step": 40272
-    },
-    {
-      "epoch": 6.485687829622771,
-      "grad_norm": 0.009420718997716904,
-      "learning_rate": 0.00019997925467887635,
-      "loss": 46.0,
-      "step": 40273
-    },
-    {
-      "epoch": 6.485848866701558,
-      "grad_norm": 0.0036219116300344467,
-      "learning_rate": 0.0001999792536484086,
-      "loss": 46.0,
-      "step": 40274
-    },
-    {
-      "epoch": 6.486009903780346,
-      "grad_norm": 0.0015857815742492676,
-      "learning_rate": 0.00019997925261791525,
-      "loss": 46.0,
-      "step": 40275
-    },
-    {
-      "epoch": 6.486170940859132,
-      "grad_norm": 0.003069506725296378,
-      "learning_rate": 0.00019997925158739635,
-      "loss": 46.0,
-      "step": 40276
-    },
-    {
-      "epoch": 6.48633197793792,
-      "grad_norm": 0.003559347242116928,
-      "learning_rate": 0.00019997925055685185,
-      "loss": 46.0,
-      "step": 40277
-    },
-    {
-      "epoch": 6.486493015016707,
-      "grad_norm": 0.012691457755863667,
-      "learning_rate": 0.00019997924952628172,
-      "loss": 46.0,
-      "step": 40278
-    },
-    {
-      "epoch": 6.486654052095495,
-      "grad_norm": 0.011235514655709267,
-      "learning_rate": 0.00019997924849568603,
-      "loss": 46.0,
-      "step": 40279
-    },
-    {
-      "epoch": 6.486815089174282,
-      "grad_norm": 0.0017168110935017467,
-      "learning_rate": 0.00019997924746506475,
-      "loss": 46.0,
-      "step": 40280
-    },
-    {
-      "epoch": 6.48697612625307,
-      "grad_norm": 0.0033350728917866945,
-      "learning_rate": 0.0001999792464344179,
-      "loss": 46.0,
-      "step": 40281
-    },
-    {
-      "epoch": 6.487137163331857,
-      "grad_norm": 0.0035972122568637133,
-      "learning_rate": 0.00019997924540374542,
-      "loss": 46.0,
-      "step": 40282
-    },
-    {
-      "epoch": 6.4872982004106445,
-      "grad_norm": 0.0012851665960624814,
-      "learning_rate": 0.00019997924437304738,
-      "loss": 46.0,
-      "step": 40283
-    },
-    {
-      "epoch": 6.487459237489432,
-      "grad_norm": 0.010535992681980133,
-      "learning_rate": 0.00019997924334232375,
-      "loss": 46.0,
-      "step": 40284
-    },
-    {
-      "epoch": 6.487620274568219,
-      "grad_norm": 0.0014540920965373516,
-      "learning_rate": 0.0001999792423115745,
-      "loss": 46.0,
-      "step": 40285
-    },
-    {
-      "epoch": 6.487781311647007,
-      "grad_norm": 0.005983424838632345,
-      "learning_rate": 0.0001999792412807997,
-      "loss": 46.0,
-      "step": 40286
-    },
-    {
-      "epoch": 6.487942348725794,
-      "grad_norm": 0.014725212939083576,
-      "learning_rate": 0.0001999792402499993,
-      "loss": 46.0,
-      "step": 40287
-    },
-    {
-      "epoch": 6.488103385804582,
-      "grad_norm": 0.0010673404904082417,
-      "learning_rate": 0.00019997923921917328,
-      "loss": 46.0,
-      "step": 40288
-    },
-    {
-      "epoch": 6.488264422883369,
-      "grad_norm": 0.001559184631332755,
-      "learning_rate": 0.0001999792381883217,
-      "loss": 46.0,
-      "step": 40289
-    },
-    {
-      "epoch": 6.488425459962157,
-      "grad_norm": 0.0036646516527980566,
-      "learning_rate": 0.00019997923715744454,
-      "loss": 46.0,
-      "step": 40290
-    },
-    {
-      "epoch": 6.488586497040943,
-      "grad_norm": 0.0033899792470037937,
-      "learning_rate": 0.00019997923612654174,
-      "loss": 46.0,
-      "step": 40291
-    },
-    {
-      "epoch": 6.488747534119731,
-      "grad_norm": 0.00300061353482306,
-      "learning_rate": 0.0001999792350956134,
-      "loss": 46.0,
-      "step": 40292
-    },
-    {
-      "epoch": 6.488908571198518,
-      "grad_norm": 0.0012679887004196644,
-      "learning_rate": 0.00019997923406465945,
-      "loss": 46.0,
-      "step": 40293
-    },
-    {
-      "epoch": 6.489069608277306,
-      "grad_norm": 0.0013386975042521954,
-      "learning_rate": 0.00019997923303367992,
-      "loss": 46.0,
-      "step": 40294
-    },
-    {
-      "epoch": 6.489230645356093,
-      "grad_norm": 0.0008013815386220813,
-      "learning_rate": 0.00019997923200267478,
-      "loss": 46.0,
-      "step": 40295
-    },
-    {
-      "epoch": 6.4893916824348805,
-      "grad_norm": 0.004258880857378244,
-      "learning_rate": 0.00019997923097164408,
-      "loss": 46.0,
-      "step": 40296
-    },
-    {
-      "epoch": 6.489552719513668,
-      "grad_norm": 0.0037058277521282434,
-      "learning_rate": 0.0001999792299405878,
-      "loss": 46.0,
-      "step": 40297
-    },
-    {
-      "epoch": 6.489713756592455,
-      "grad_norm": 0.0032727655488997698,
-      "learning_rate": 0.00019997922890950588,
-      "loss": 46.0,
-      "step": 40298
-    },
-    {
-      "epoch": 6.489874793671243,
-      "grad_norm": 0.009909887798130512,
-      "learning_rate": 0.0001999792278783984,
-      "loss": 46.0,
-      "step": 40299
-    },
-    {
-      "epoch": 6.49003583075003,
-      "grad_norm": 0.004497320391237736,
-      "learning_rate": 0.00019997922684726534,
-      "loss": 46.0,
-      "step": 40300
-    },
-    {
-      "epoch": 6.490196867828818,
-      "grad_norm": 0.0013308433117344975,
-      "learning_rate": 0.00019997922581610668,
-      "loss": 46.0,
-      "step": 40301
-    },
-    {
-      "epoch": 6.490357904907605,
-      "grad_norm": 0.0034019825980067253,
-      "learning_rate": 0.00019997922478492242,
-      "loss": 46.0,
-      "step": 40302
-    },
-    {
-      "epoch": 6.490518941986393,
-      "grad_norm": 0.001946385600604117,
-      "learning_rate": 0.00019997922375371256,
-      "loss": 46.0,
-      "step": 40303
-    },
-    {
-      "epoch": 6.49067997906518,
-      "grad_norm": 0.005960382521152496,
-      "learning_rate": 0.00019997922272247713,
-      "loss": 46.0,
-      "step": 40304
-    },
-    {
-      "epoch": 6.490841016143968,
-      "grad_norm": 0.0016100500943139195,
-      "learning_rate": 0.0001999792216912161,
-      "loss": 46.0,
-      "step": 40305
-    },
-    {
-      "epoch": 6.491002053222754,
-      "grad_norm": 0.0014602484880015254,
-      "learning_rate": 0.00019997922065992951,
-      "loss": 46.0,
-      "step": 40306
-    },
-    {
-      "epoch": 6.491163090301542,
-      "grad_norm": 0.004667380359023809,
-      "learning_rate": 0.0001999792196286173,
-      "loss": 46.0,
-      "step": 40307
-    },
-    {
-      "epoch": 6.491324127380329,
-      "grad_norm": 0.001152698416262865,
-      "learning_rate": 0.00019997921859727952,
-      "loss": 46.0,
-      "step": 40308
-    },
-    {
-      "epoch": 6.4914851644591165,
-      "grad_norm": 0.0025428500957787037,
-      "learning_rate": 0.00019997921756591613,
-      "loss": 46.0,
-      "step": 40309
-    },
-    {
-      "epoch": 6.491646201537904,
-      "grad_norm": 0.01012470480054617,
-      "learning_rate": 0.00019997921653452716,
-      "loss": 46.0,
-      "step": 40310
-    },
-    {
-      "epoch": 6.491807238616691,
-      "grad_norm": 0.010339552536606789,
-      "learning_rate": 0.0001999792155031126,
-      "loss": 46.0,
-      "step": 40311
-    },
-    {
-      "epoch": 6.491968275695479,
-      "grad_norm": 0.002963364589959383,
-      "learning_rate": 0.00019997921447167247,
-      "loss": 46.0,
-      "step": 40312
-    },
-    {
-      "epoch": 6.492129312774266,
-      "grad_norm": 0.0012313439510762691,
-      "learning_rate": 0.0001999792134402067,
-      "loss": 46.0,
-      "step": 40313
-    },
-    {
-      "epoch": 6.492290349853054,
-      "grad_norm": 0.003133671125397086,
-      "learning_rate": 0.0001999792124087154,
-      "loss": 46.0,
-      "step": 40314
-    },
-    {
-      "epoch": 6.492451386931841,
-      "grad_norm": 0.003556988202035427,
-      "learning_rate": 0.0001999792113771985,
-      "loss": 46.0,
-      "step": 40315
-    },
-    {
-      "epoch": 6.492612424010629,
-      "grad_norm": 0.0022010449320077896,
-      "learning_rate": 0.000199979210345656,
-      "loss": 46.0,
-      "step": 40316
-    },
-    {
-      "epoch": 6.492773461089416,
-      "grad_norm": 0.023579290136694908,
-      "learning_rate": 0.00019997920931408788,
-      "loss": 46.0,
-      "step": 40317
-    },
-    {
-      "epoch": 6.492934498168204,
-      "grad_norm": 0.004673734772950411,
-      "learning_rate": 0.0001999792082824942,
-      "loss": 46.0,
-      "step": 40318
-    },
-    {
-      "epoch": 6.493095535246991,
-      "grad_norm": 0.019197355955839157,
-      "learning_rate": 0.00019997920725087494,
-      "loss": 46.0,
-      "step": 40319
-    },
-    {
-      "epoch": 6.4932565723257785,
-      "grad_norm": 0.00511762173846364,
-      "learning_rate": 0.00019997920621923006,
-      "loss": 46.0,
-      "step": 40320
-    },
-    {
-      "epoch": 6.493417609404565,
-      "grad_norm": 0.006581800524145365,
-      "learning_rate": 0.00019997920518755963,
-      "loss": 46.0,
-      "step": 40321
-    },
-    {
-      "epoch": 6.4935786464833525,
-      "grad_norm": 0.003833992872387171,
-      "learning_rate": 0.00019997920415586355,
-      "loss": 46.0,
-      "step": 40322
-    },
-    {
-      "epoch": 6.49373968356214,
-      "grad_norm": 0.008301000110805035,
-      "learning_rate": 0.00019997920312414194,
-      "loss": 46.0,
-      "step": 40323
-    },
-    {
-      "epoch": 6.493900720640927,
-      "grad_norm": 0.003949692007154226,
-      "learning_rate": 0.00019997920209239472,
-      "loss": 46.0,
-      "step": 40324
-    },
-    {
-      "epoch": 6.494061757719715,
-      "grad_norm": 0.0062357839196920395,
-      "learning_rate": 0.00019997920106062188,
-      "loss": 46.0,
-      "step": 40325
-    },
-    {
-      "epoch": 6.494222794798502,
-      "grad_norm": 0.017794236540794373,
-      "learning_rate": 0.0001999792000288235,
-      "loss": 46.0,
-      "step": 40326
-    },
-    {
-      "epoch": 6.49438383187729,
-      "grad_norm": 0.009414306841790676,
-      "learning_rate": 0.0001999791989969995,
-      "loss": 46.0,
-      "step": 40327
-    },
-    {
-      "epoch": 6.494544868956077,
-      "grad_norm": 0.006454518996179104,
-      "learning_rate": 0.00019997919796514992,
-      "loss": 46.0,
-      "step": 40328
-    },
-    {
-      "epoch": 6.494705906034865,
-      "grad_norm": 0.00879088044166565,
-      "learning_rate": 0.00019997919693327476,
-      "loss": 46.0,
-      "step": 40329
-    },
-    {
-      "epoch": 6.494866943113652,
-      "grad_norm": 0.0038317013531923294,
-      "learning_rate": 0.000199979195901374,
-      "loss": 46.0,
-      "step": 40330
-    },
-    {
-      "epoch": 6.4950279801924395,
-      "grad_norm": 0.005642174277454615,
-      "learning_rate": 0.00019997919486944765,
-      "loss": 46.0,
-      "step": 40331
-    },
-    {
-      "epoch": 6.495189017271227,
-      "grad_norm": 0.008032823912799358,
-      "learning_rate": 0.0001999791938374957,
-      "loss": 46.0,
-      "step": 40332
-    },
-    {
-      "epoch": 6.4953500543500144,
-      "grad_norm": 0.012297971174120903,
-      "learning_rate": 0.0001999791928055182,
-      "loss": 46.0,
-      "step": 40333
-    },
-    {
-      "epoch": 6.495511091428802,
-      "grad_norm": 0.0021798356901854277,
-      "learning_rate": 0.00019997919177351507,
-      "loss": 46.0,
-      "step": 40334
-    },
-    {
-      "epoch": 6.4956721285075885,
-      "grad_norm": 0.0048887706361711025,
-      "learning_rate": 0.00019997919074148636,
-      "loss": 46.0,
-      "step": 40335
-    },
-    {
-      "epoch": 6.495833165586376,
-      "grad_norm": 0.0012530314270406961,
-      "learning_rate": 0.0001999791897094321,
-      "loss": 46.0,
-      "step": 40336
-    },
-    {
-      "epoch": 6.495994202665163,
-      "grad_norm": 0.009131227619946003,
-      "learning_rate": 0.0001999791886773522,
-      "loss": 46.0,
-      "step": 40337
-    },
-    {
-      "epoch": 6.496155239743951,
-      "grad_norm": 0.0027114255353808403,
-      "learning_rate": 0.00019997918764524673,
-      "loss": 46.0,
-      "step": 40338
-    },
-    {
-      "epoch": 6.496316276822738,
-      "grad_norm": 0.004454668611288071,
-      "learning_rate": 0.00019997918661311564,
-      "loss": 46.0,
-      "step": 40339
-    },
-    {
-      "epoch": 6.496477313901526,
-      "grad_norm": 0.0008246162324212492,
-      "learning_rate": 0.00019997918558095897,
-      "loss": 46.0,
-      "step": 40340
-    },
-    {
-      "epoch": 6.496638350980313,
-      "grad_norm": 0.001442980719730258,
-      "learning_rate": 0.00019997918454877676,
-      "loss": 46.0,
-      "step": 40341
-    },
-    {
-      "epoch": 6.496799388059101,
-      "grad_norm": 0.0023324722424149513,
-      "learning_rate": 0.00019997918351656892,
-      "loss": 46.0,
-      "step": 40342
-    },
-    {
-      "epoch": 6.496960425137888,
-      "grad_norm": 0.008202840574085712,
-      "learning_rate": 0.00019997918248433548,
-      "loss": 46.0,
-      "step": 40343
-    },
-    {
-      "epoch": 6.4971214622166755,
-      "grad_norm": 0.003734815865755081,
-      "learning_rate": 0.00019997918145207649,
-      "loss": 46.0,
-      "step": 40344
-    },
-    {
-      "epoch": 6.497282499295463,
-      "grad_norm": 0.003976847976446152,
-      "learning_rate": 0.00019997918041979188,
-      "loss": 46.0,
-      "step": 40345
-    },
-    {
-      "epoch": 6.49744353637425,
-      "grad_norm": 0.01052770484238863,
-      "learning_rate": 0.00019997917938748168,
-      "loss": 46.0,
-      "step": 40346
-    },
-    {
-      "epoch": 6.497604573453038,
-      "grad_norm": 0.003329983912408352,
-      "learning_rate": 0.0001999791783551459,
-      "loss": 46.0,
-      "step": 40347
-    },
-    {
-      "epoch": 6.497765610531825,
-      "grad_norm": 0.01705816015601158,
-      "learning_rate": 0.00019997917732278453,
-      "loss": 46.0,
-      "step": 40348
-    },
-    {
-      "epoch": 6.497926647610612,
-      "grad_norm": 0.003354690968990326,
-      "learning_rate": 0.00019997917629039757,
-      "loss": 46.0,
-      "step": 40349
-    },
-    {
-      "epoch": 6.498087684689399,
-      "grad_norm": 0.010628162883222103,
-      "learning_rate": 0.00019997917525798502,
-      "loss": 46.0,
-      "step": 40350
-    },
-    {
-      "epoch": 6.498248721768187,
-      "grad_norm": 0.0010940140346065164,
-      "learning_rate": 0.0001999791742255469,
-      "loss": 46.0,
-      "step": 40351
-    },
-    {
-      "epoch": 6.498409758846974,
-      "grad_norm": 0.0038282989989966154,
-      "learning_rate": 0.00019997917319308314,
-      "loss": 46.0,
-      "step": 40352
-    },
-    {
-      "epoch": 6.498570795925762,
-      "grad_norm": 0.006773098837584257,
-      "learning_rate": 0.00019997917216059384,
-      "loss": 46.0,
-      "step": 40353
-    },
-    {
-      "epoch": 6.498731833004549,
-      "grad_norm": 0.002765289042145014,
-      "learning_rate": 0.00019997917112807892,
-      "loss": 46.0,
-      "step": 40354
-    },
-    {
-      "epoch": 6.498892870083337,
-      "grad_norm": 0.0015978373121470213,
-      "learning_rate": 0.0001999791700955384,
-      "loss": 46.0,
-      "step": 40355
-    },
-    {
-      "epoch": 6.499053907162124,
-      "grad_norm": 0.005281936377286911,
-      "learning_rate": 0.00019997916906297234,
-      "loss": 46.0,
-      "step": 40356
-    },
-    {
-      "epoch": 6.4992149442409115,
-      "grad_norm": 0.00275397184304893,
-      "learning_rate": 0.00019997916803038066,
-      "loss": 46.0,
-      "step": 40357
-    },
-    {
-      "epoch": 6.499375981319699,
-      "grad_norm": 0.01011024322360754,
-      "learning_rate": 0.00019997916699776341,
-      "loss": 46.0,
-      "step": 40358
-    },
-    {
-      "epoch": 6.499537018398486,
-      "grad_norm": 0.014196864329278469,
-      "learning_rate": 0.00019997916596512053,
-      "loss": 46.0,
-      "step": 40359
-    },
-    {
-      "epoch": 6.499698055477274,
-      "grad_norm": 0.004001722205430269,
-      "learning_rate": 0.0001999791649324521,
-      "loss": 46.0,
-      "step": 40360
-    },
-    {
-      "epoch": 6.499859092556061,
-      "grad_norm": 0.0022927215322852135,
-      "learning_rate": 0.00019997916389975805,
-      "loss": 46.0,
-      "step": 40361
-    },
-    {
-      "epoch": 6.500020129634849,
-      "grad_norm": 0.002157264156267047,
-      "learning_rate": 0.00019997916286703844,
-      "loss": 46.0,
-      "step": 40362
-    },
-    {
-      "epoch": 6.500181166713636,
-      "grad_norm": 0.007294053211808205,
-      "learning_rate": 0.0001999791618342932,
-      "loss": 46.0,
-      "step": 40363
-    },
-    {
-      "epoch": 6.500342203792423,
-      "grad_norm": 0.0013782167807221413,
-      "learning_rate": 0.0001999791608015224,
-      "loss": 46.0,
-      "step": 40364
-    },
-    {
-      "epoch": 6.50050324087121,
-      "grad_norm": 0.005053127650171518,
-      "learning_rate": 0.00019997915976872603,
-      "loss": 46.0,
-      "step": 40365
-    },
-    {
-      "epoch": 6.500664277949998,
-      "grad_norm": 0.00581505848094821,
-      "learning_rate": 0.00019997915873590404,
-      "loss": 46.0,
-      "step": 40366
-    },
-    {
-      "epoch": 6.500825315028785,
-      "grad_norm": 0.0025002227630466223,
-      "learning_rate": 0.00019997915770305648,
-      "loss": 46.0,
-      "step": 40367
-    },
-    {
-      "epoch": 6.500986352107573,
-      "grad_norm": 0.005692105740308762,
-      "learning_rate": 0.00019997915667018331,
-      "loss": 46.0,
-      "step": 40368
-    },
-    {
-      "epoch": 6.50114738918636,
-      "grad_norm": 0.009045046754181385,
-      "learning_rate": 0.00019997915563728456,
-      "loss": 46.0,
-      "step": 40369
-    },
-    {
-      "epoch": 6.5013084262651475,
-      "grad_norm": 0.0019253044156357646,
-      "learning_rate": 0.00019997915460436022,
-      "loss": 46.0,
-      "step": 40370
-    },
-    {
-      "epoch": 6.501469463343935,
-      "grad_norm": 0.0029128703754395247,
-      "learning_rate": 0.0001999791535714103,
-      "loss": 46.0,
-      "step": 40371
-    },
-    {
-      "epoch": 6.501630500422722,
-      "grad_norm": 0.010929145850241184,
-      "learning_rate": 0.00019997915253843477,
-      "loss": 46.0,
-      "step": 40372
-    },
-    {
-      "epoch": 6.50179153750151,
-      "grad_norm": 0.005774811841547489,
-      "learning_rate": 0.00019997915150543366,
-      "loss": 46.0,
-      "step": 40373
-    },
-    {
-      "epoch": 6.501952574580297,
-      "grad_norm": 0.006323961541056633,
-      "learning_rate": 0.00019997915047240697,
-      "loss": 46.0,
-      "step": 40374
-    },
-    {
-      "epoch": 6.502113611659085,
-      "grad_norm": 0.006913771387189627,
-      "learning_rate": 0.00019997914943935467,
-      "loss": 46.0,
-      "step": 40375
-    },
-    {
-      "epoch": 6.502274648737872,
-      "grad_norm": 0.00869273766875267,
-      "learning_rate": 0.00019997914840627677,
-      "loss": 46.0,
-      "step": 40376
-    },
-    {
-      "epoch": 6.50243568581666,
-      "grad_norm": 0.0034256496001034975,
-      "learning_rate": 0.00019997914737317332,
-      "loss": 46.0,
-      "step": 40377
-    },
-    {
-      "epoch": 6.502596722895447,
-      "grad_norm": 0.011104143224656582,
-      "learning_rate": 0.00019997914634004425,
-      "loss": 46.0,
-      "step": 40378
-    },
-    {
-      "epoch": 6.502757759974234,
-      "grad_norm": 0.005923696327954531,
-      "learning_rate": 0.00019997914530688963,
-      "loss": 46.0,
-      "step": 40379
-    },
-    {
-      "epoch": 6.502918797053021,
-      "grad_norm": 0.003140954999253154,
-      "learning_rate": 0.00019997914427370938,
-      "loss": 46.0,
-      "step": 40380
-    },
-    {
-      "epoch": 6.503079834131809,
-      "grad_norm": 0.0022571368608623743,
-      "learning_rate": 0.00019997914324050356,
-      "loss": 46.0,
-      "step": 40381
-    },
-    {
-      "epoch": 6.503240871210596,
-      "grad_norm": 0.012748190201818943,
-      "learning_rate": 0.00019997914220727214,
-      "loss": 46.0,
-      "step": 40382
-    },
-    {
-      "epoch": 6.5034019082893835,
-      "grad_norm": 0.004408635199069977,
-      "learning_rate": 0.00019997914117401516,
-      "loss": 46.0,
-      "step": 40383
-    },
-    {
-      "epoch": 6.503562945368171,
-      "grad_norm": 0.0014002801617607474,
-      "learning_rate": 0.00019997914014073255,
-      "loss": 46.0,
-      "step": 40384
-    },
-    {
-      "epoch": 6.503723982446958,
-      "grad_norm": 0.004015210550278425,
-      "learning_rate": 0.00019997913910742437,
-      "loss": 46.0,
-      "step": 40385
-    },
-    {
-      "epoch": 6.503885019525746,
-      "grad_norm": 0.013162247836589813,
-      "learning_rate": 0.00019997913807409058,
-      "loss": 46.0,
-      "step": 40386
-    },
-    {
-      "epoch": 6.504046056604533,
-      "grad_norm": 0.0035329218953847885,
-      "learning_rate": 0.00019997913704073122,
-      "loss": 46.0,
-      "step": 40387
-    },
-    {
-      "epoch": 6.504207093683321,
-      "grad_norm": 0.011753002181649208,
-      "learning_rate": 0.00019997913600734629,
-      "loss": 46.0,
-      "step": 40388
-    },
-    {
-      "epoch": 6.504368130762108,
-      "grad_norm": 0.005099029745906591,
-      "learning_rate": 0.00019997913497393573,
-      "loss": 46.0,
-      "step": 40389
-    },
-    {
-      "epoch": 6.504529167840896,
-      "grad_norm": 0.01428051944822073,
-      "learning_rate": 0.00019997913394049962,
-      "loss": 46.0,
-      "step": 40390
-    },
-    {
-      "epoch": 6.504690204919683,
-      "grad_norm": 0.0017496683867648244,
-      "learning_rate": 0.00019997913290703786,
-      "loss": 46.0,
-      "step": 40391
-    },
-    {
-      "epoch": 6.5048512419984705,
-      "grad_norm": 0.00535649573430419,
-      "learning_rate": 0.00019997913187355058,
-      "loss": 46.0,
-      "step": 40392
-    },
-    {
-      "epoch": 6.505012279077258,
-      "grad_norm": 0.001099920365959406,
-      "learning_rate": 0.00019997913084003767,
-      "loss": 46.0,
-      "step": 40393
-    },
-    {
-      "epoch": 6.5051733161560445,
-      "grad_norm": 0.006725079379975796,
-      "learning_rate": 0.0001999791298064992,
-      "loss": 46.0,
-      "step": 40394
-    },
-    {
-      "epoch": 6.505334353234832,
-      "grad_norm": 0.0021010739728808403,
-      "learning_rate": 0.0001999791287729351,
-      "loss": 46.0,
-      "step": 40395
-    },
-    {
-      "epoch": 6.5054953903136195,
-      "grad_norm": 0.006272085011005402,
-      "learning_rate": 0.00019997912773934545,
-      "loss": 46.0,
-      "step": 40396
-    },
-    {
-      "epoch": 6.505656427392407,
-      "grad_norm": 0.004059297032654285,
-      "learning_rate": 0.00019997912670573017,
-      "loss": 46.0,
-      "step": 40397
-    },
-    {
-      "epoch": 6.505817464471194,
-      "grad_norm": 0.002523328410461545,
-      "learning_rate": 0.00019997912567208933,
-      "loss": 46.0,
-      "step": 40398
-    },
-    {
-      "epoch": 6.505978501549982,
-      "grad_norm": 0.0030419593676924706,
-      "learning_rate": 0.0001999791246384229,
-      "loss": 46.0,
-      "step": 40399
-    },
-    {
-      "epoch": 6.506139538628769,
-      "grad_norm": 0.005724390037357807,
-      "learning_rate": 0.00019997912360473087,
-      "loss": 46.0,
-      "step": 40400
-    },
-    {
-      "epoch": 6.506300575707557,
-      "grad_norm": 0.004784403834491968,
-      "learning_rate": 0.00019997912257101327,
-      "loss": 46.0,
-      "step": 40401
-    },
-    {
-      "epoch": 6.506461612786344,
-      "grad_norm": 0.001489435089752078,
-      "learning_rate": 0.00019997912153727003,
-      "loss": 46.0,
-      "step": 40402
-    },
-    {
-      "epoch": 6.506622649865132,
-      "grad_norm": 0.0024986187927424908,
-      "learning_rate": 0.00019997912050350125,
-      "loss": 46.0,
-      "step": 40403
-    },
-    {
-      "epoch": 6.506783686943919,
-      "grad_norm": 0.005360976792871952,
-      "learning_rate": 0.00019997911946970686,
-      "loss": 46.0,
-      "step": 40404
-    },
-    {
-      "epoch": 6.5069447240227065,
-      "grad_norm": 0.0036494999658316374,
-      "learning_rate": 0.0001999791184358869,
-      "loss": 46.0,
-      "step": 40405
-    },
-    {
-      "epoch": 6.507105761101494,
-      "grad_norm": 0.014332235790789127,
-      "learning_rate": 0.00019997911740204133,
-      "loss": 46.0,
-      "step": 40406
-    },
-    {
-      "epoch": 6.5072667981802805,
-      "grad_norm": 0.01938583329319954,
-      "learning_rate": 0.00019997911636817018,
-      "loss": 46.0,
-      "step": 40407
-    },
-    {
-      "epoch": 6.507427835259069,
-      "grad_norm": 0.001076909713447094,
-      "learning_rate": 0.0001999791153342734,
-      "loss": 46.0,
-      "step": 40408
-    },
-    {
-      "epoch": 6.507588872337855,
-      "grad_norm": 0.0043213204480707645,
-      "learning_rate": 0.0001999791143003511,
-      "loss": 46.0,
-      "step": 40409
-    },
-    {
-      "epoch": 6.507749909416643,
-      "grad_norm": 0.02043306455016136,
-      "learning_rate": 0.0001999791132664032,
-      "loss": 46.0,
-      "step": 40410
-    },
-    {
-      "epoch": 6.50791094649543,
-      "grad_norm": 0.001814353629015386,
-      "learning_rate": 0.00019997911223242965,
-      "loss": 46.0,
-      "step": 40411
-    },
-    {
-      "epoch": 6.508071983574218,
-      "grad_norm": 0.014954142272472382,
-      "learning_rate": 0.00019997911119843057,
-      "loss": 46.0,
-      "step": 40412
-    },
-    {
-      "epoch": 6.508233020653005,
-      "grad_norm": 0.008600250817835331,
-      "learning_rate": 0.00019997911016440587,
-      "loss": 46.0,
-      "step": 40413
-    },
-    {
-      "epoch": 6.508394057731793,
-      "grad_norm": 0.0025109057314693928,
-      "learning_rate": 0.0001999791091303556,
-      "loss": 46.0,
-      "step": 40414
-    },
-    {
-      "epoch": 6.50855509481058,
-      "grad_norm": 0.0009861724684014916,
-      "learning_rate": 0.00019997910809627973,
-      "loss": 46.0,
-      "step": 40415
-    },
-    {
-      "epoch": 6.508716131889368,
-      "grad_norm": 0.005474052857607603,
-      "learning_rate": 0.00019997910706217827,
-      "loss": 46.0,
-      "step": 40416
-    },
-    {
-      "epoch": 6.508877168968155,
-      "grad_norm": 0.002236713655292988,
-      "learning_rate": 0.00019997910602805122,
-      "loss": 46.0,
-      "step": 40417
-    },
-    {
-      "epoch": 6.5090382060469425,
-      "grad_norm": 0.0008153865928761661,
-      "learning_rate": 0.0001999791049938986,
-      "loss": 46.0,
-      "step": 40418
-    },
-    {
-      "epoch": 6.50919924312573,
-      "grad_norm": 0.0026476015336811543,
-      "learning_rate": 0.00019997910395972034,
-      "loss": 46.0,
-      "step": 40419
-    },
-    {
-      "epoch": 6.509360280204517,
-      "grad_norm": 0.0034569473937153816,
-      "learning_rate": 0.00019997910292551653,
-      "loss": 46.0,
-      "step": 40420
-    },
-    {
-      "epoch": 6.509521317283305,
-      "grad_norm": 0.002157674403861165,
-      "learning_rate": 0.00019997910189128713,
-      "loss": 46.0,
-      "step": 40421
-    },
-    {
-      "epoch": 6.509682354362091,
-      "grad_norm": 0.004206405952572823,
-      "learning_rate": 0.00019997910085703212,
-      "loss": 46.0,
-      "step": 40422
-    },
-    {
-      "epoch": 6.50984339144088,
-      "grad_norm": 0.0036905494052916765,
-      "learning_rate": 0.00019997909982275155,
-      "loss": 46.0,
-      "step": 40423
-    },
-    {
-      "epoch": 6.510004428519666,
-      "grad_norm": 0.0018899135757237673,
-      "learning_rate": 0.0001999790987884454,
-      "loss": 46.0,
-      "step": 40424
-    },
-    {
-      "epoch": 6.510165465598454,
-      "grad_norm": 0.002122113946825266,
-      "learning_rate": 0.00019997909775411362,
-      "loss": 46.0,
-      "step": 40425
-    },
-    {
-      "epoch": 6.510326502677241,
-      "grad_norm": 0.0012016886612400413,
-      "learning_rate": 0.00019997909671975626,
-      "loss": 46.0,
-      "step": 40426
-    },
-    {
-      "epoch": 6.510487539756029,
-      "grad_norm": 0.004437318537384272,
-      "learning_rate": 0.0001999790956853733,
-      "loss": 46.0,
-      "step": 40427
-    },
-    {
-      "epoch": 6.510648576834816,
-      "grad_norm": 0.002166911493986845,
-      "learning_rate": 0.00019997909465096478,
-      "loss": 46.0,
-      "step": 40428
-    },
-    {
-      "epoch": 6.510809613913604,
-      "grad_norm": 0.007263000123202801,
-      "learning_rate": 0.00019997909361653066,
-      "loss": 46.0,
-      "step": 40429
-    },
-    {
-      "epoch": 6.510970650992391,
-      "grad_norm": 0.0013375879498198628,
-      "learning_rate": 0.00019997909258207095,
-      "loss": 46.0,
-      "step": 40430
-    },
-    {
-      "epoch": 6.5111316880711785,
-      "grad_norm": 0.002949443878605962,
-      "learning_rate": 0.00019997909154758565,
-      "loss": 46.0,
-      "step": 40431
-    },
-    {
-      "epoch": 6.511292725149966,
-      "grad_norm": 0.01013131532818079,
-      "learning_rate": 0.00019997909051307474,
-      "loss": 46.0,
-      "step": 40432
-    },
-    {
-      "epoch": 6.511453762228753,
-      "grad_norm": 0.005717073101550341,
-      "learning_rate": 0.00019997908947853827,
-      "loss": 46.0,
-      "step": 40433
-    },
-    {
-      "epoch": 6.511614799307541,
-      "grad_norm": 0.0020720921456813812,
-      "learning_rate": 0.0001999790884439762,
-      "loss": 46.0,
-      "step": 40434
-    },
-    {
-      "epoch": 6.511775836386328,
-      "grad_norm": 0.004989050328731537,
-      "learning_rate": 0.00019997908740938854,
-      "loss": 46.0,
-      "step": 40435
-    },
-    {
-      "epoch": 6.511936873465116,
-      "grad_norm": 0.017240235581994057,
-      "learning_rate": 0.00019997908637477528,
-      "loss": 46.0,
-      "step": 40436
-    },
-    {
-      "epoch": 6.512097910543902,
-      "grad_norm": 0.0009644894162192941,
-      "learning_rate": 0.00019997908534013646,
-      "loss": 46.0,
-      "step": 40437
-    },
-    {
-      "epoch": 6.51225894762269,
-      "grad_norm": 0.003865724429488182,
-      "learning_rate": 0.00019997908430547203,
-      "loss": 46.0,
-      "step": 40438
-    },
-    {
-      "epoch": 6.512419984701477,
-      "grad_norm": 0.0022453987039625645,
-      "learning_rate": 0.000199979083270782,
-      "loss": 46.0,
-      "step": 40439
-    },
-    {
-      "epoch": 6.512581021780265,
-      "grad_norm": 0.007251511327922344,
-      "learning_rate": 0.0001999790822360664,
-      "loss": 46.0,
-      "step": 40440
-    },
-    {
-      "epoch": 6.512742058859052,
-      "grad_norm": 0.0015736159402877092,
-      "learning_rate": 0.0001999790812013252,
-      "loss": 46.0,
-      "step": 40441
-    },
-    {
-      "epoch": 6.51290309593784,
-      "grad_norm": 0.0032892003655433655,
-      "learning_rate": 0.00019997908016655843,
-      "loss": 46.0,
-      "step": 40442
-    },
-    {
-      "epoch": 6.513064133016627,
-      "grad_norm": 0.008809706196188927,
-      "learning_rate": 0.00019997907913176603,
-      "loss": 46.0,
-      "step": 40443
-    },
-    {
-      "epoch": 6.5132251700954145,
-      "grad_norm": 0.0073125348426401615,
-      "learning_rate": 0.00019997907809694808,
-      "loss": 46.0,
-      "step": 40444
-    },
-    {
-      "epoch": 6.513386207174202,
-      "grad_norm": 0.0027208244428038597,
-      "learning_rate": 0.00019997907706210453,
-      "loss": 46.0,
-      "step": 40445
-    },
-    {
-      "epoch": 6.513547244252989,
-      "grad_norm": 0.004564281087368727,
-      "learning_rate": 0.00019997907602723537,
-      "loss": 46.0,
-      "step": 40446
-    },
-    {
-      "epoch": 6.513708281331777,
-      "grad_norm": 0.0024781201500445604,
-      "learning_rate": 0.00019997907499234066,
-      "loss": 46.0,
-      "step": 40447
-    },
-    {
-      "epoch": 6.513869318410564,
-      "grad_norm": 0.00413032341748476,
-      "learning_rate": 0.00019997907395742032,
-      "loss": 46.0,
-      "step": 40448
-    },
-    {
-      "epoch": 6.514030355489352,
-      "grad_norm": 0.007044166326522827,
-      "learning_rate": 0.00019997907292247443,
-      "loss": 46.0,
-      "step": 40449
-    },
-    {
-      "epoch": 6.514191392568139,
-      "grad_norm": 0.004675073549151421,
-      "learning_rate": 0.0001999790718875029,
-      "loss": 46.0,
-      "step": 40450
-    },
-    {
-      "epoch": 6.514352429646927,
-      "grad_norm": 0.0032668865751475096,
-      "learning_rate": 0.00019997907085250583,
-      "loss": 46.0,
-      "step": 40451
-    },
-    {
-      "epoch": 6.514513466725713,
-      "grad_norm": 0.001038560178130865,
-      "learning_rate": 0.00019997906981748312,
-      "loss": 46.0,
-      "step": 40452
-    },
-    {
-      "epoch": 6.514674503804501,
-      "grad_norm": 0.002219228306785226,
-      "learning_rate": 0.00019997906878243486,
-      "loss": 46.0,
-      "step": 40453
-    },
-    {
-      "epoch": 6.514835540883288,
-      "grad_norm": 0.007459376938641071,
-      "learning_rate": 0.000199979067747361,
-      "loss": 46.0,
-      "step": 40454
-    },
-    {
-      "epoch": 6.5149965779620755,
-      "grad_norm": 0.006110853981226683,
-      "learning_rate": 0.00019997906671226153,
-      "loss": 46.0,
-      "step": 40455
-    },
-    {
-      "epoch": 6.515157615040863,
-      "grad_norm": 0.003304006764665246,
-      "learning_rate": 0.0001999790656771365,
-      "loss": 46.0,
-      "step": 40456
-    },
-    {
-      "epoch": 6.5153186521196504,
-      "grad_norm": 0.0014726794324815273,
-      "learning_rate": 0.0001999790646419859,
-      "loss": 46.0,
-      "step": 40457
-    },
-    {
-      "epoch": 6.515479689198438,
-      "grad_norm": 0.008875620551407337,
-      "learning_rate": 0.00019997906360680966,
-      "loss": 46.0,
-      "step": 40458
-    },
-    {
-      "epoch": 6.515640726277225,
-      "grad_norm": 0.023572100326418877,
-      "learning_rate": 0.00019997906257160787,
-      "loss": 46.0,
-      "step": 40459
-    },
-    {
-      "epoch": 6.515801763356013,
-      "grad_norm": 0.004309433978050947,
-      "learning_rate": 0.00019997906153638043,
-      "loss": 46.0,
-      "step": 40460
-    },
-    {
-      "epoch": 6.5159628004348,
-      "grad_norm": 0.001962625188753009,
-      "learning_rate": 0.00019997906050112747,
-      "loss": 46.0,
-      "step": 40461
-    },
-    {
-      "epoch": 6.516123837513588,
-      "grad_norm": 0.0007694255327805877,
-      "learning_rate": 0.0001999790594658489,
-      "loss": 46.0,
-      "step": 40462
-    },
-    {
-      "epoch": 6.516284874592375,
-      "grad_norm": 0.005986039061099291,
-      "learning_rate": 0.00019997905843054472,
-      "loss": 46.0,
-      "step": 40463
-    },
-    {
-      "epoch": 6.516445911671163,
-      "grad_norm": 0.0012874397216364741,
-      "learning_rate": 0.00019997905739521497,
-      "loss": 46.0,
-      "step": 40464
-    },
-    {
-      "epoch": 6.51660694874995,
-      "grad_norm": 0.005189476069062948,
-      "learning_rate": 0.00019997905635985963,
-      "loss": 46.0,
-      "step": 40465
-    },
-    {
-      "epoch": 6.5167679858287375,
-      "grad_norm": 0.006326170638203621,
-      "learning_rate": 0.0001999790553244787,
-      "loss": 46.0,
-      "step": 40466
-    },
-    {
-      "epoch": 6.516929022907524,
-      "grad_norm": 0.0013067203108221292,
-      "learning_rate": 0.00019997905428907218,
-      "loss": 46.0,
-      "step": 40467
-    },
-    {
-      "epoch": 6.5170900599863115,
-      "grad_norm": 0.0038871399592608213,
-      "learning_rate": 0.00019997905325364003,
-      "loss": 46.0,
-      "step": 40468
-    },
-    {
-      "epoch": 6.517251097065099,
-      "grad_norm": 0.0070176864974200726,
-      "learning_rate": 0.00019997905221818236,
-      "loss": 46.0,
-      "step": 40469
-    },
-    {
-      "epoch": 6.517412134143886,
-      "grad_norm": 0.005611415486782789,
-      "learning_rate": 0.00019997905118269906,
-      "loss": 46.0,
-      "step": 40470
-    },
-    {
-      "epoch": 6.517573171222674,
-      "grad_norm": 0.006022224202752113,
-      "learning_rate": 0.0001999790501471902,
-      "loss": 46.0,
-      "step": 40471
-    },
-    {
-      "epoch": 6.517734208301461,
-      "grad_norm": 0.001275284099392593,
-      "learning_rate": 0.0001999790491116557,
-      "loss": 46.0,
-      "step": 40472
-    },
-    {
-      "epoch": 6.517895245380249,
-      "grad_norm": 0.012299781665205956,
-      "learning_rate": 0.00019997904807609565,
-      "loss": 46.0,
-      "step": 40473
-    },
-    {
-      "epoch": 6.518056282459036,
-      "grad_norm": 0.002479087794199586,
-      "learning_rate": 0.00019997904704050997,
-      "loss": 46.0,
-      "step": 40474
-    },
-    {
-      "epoch": 6.518217319537824,
-      "grad_norm": 0.01013909187167883,
-      "learning_rate": 0.00019997904600489873,
-      "loss": 46.0,
-      "step": 40475
-    },
-    {
-      "epoch": 6.518378356616611,
-      "grad_norm": 0.0018155794823542237,
-      "learning_rate": 0.0001999790449692619,
-      "loss": 46.0,
-      "step": 40476
-    },
-    {
-      "epoch": 6.518539393695399,
-      "grad_norm": 0.0007866375381127,
-      "learning_rate": 0.0001999790439335995,
-      "loss": 46.0,
-      "step": 40477
-    },
-    {
-      "epoch": 6.518700430774186,
-      "grad_norm": 0.015021715313196182,
-      "learning_rate": 0.0001999790428979115,
-      "loss": 46.0,
-      "step": 40478
-    },
-    {
-      "epoch": 6.5188614678529735,
-      "grad_norm": 0.0028334539383649826,
-      "learning_rate": 0.0001999790418621979,
-      "loss": 46.0,
-      "step": 40479
-    },
-    {
-      "epoch": 6.51902250493176,
-      "grad_norm": 0.0025506368838250637,
-      "learning_rate": 0.0001999790408264587,
-      "loss": 46.0,
-      "step": 40480
-    },
-    {
-      "epoch": 6.519183542010548,
-      "grad_norm": 0.006063203793019056,
-      "learning_rate": 0.00019997903979069394,
-      "loss": 46.0,
-      "step": 40481
-    },
-    {
-      "epoch": 6.519344579089335,
-      "grad_norm": 0.013854878954589367,
-      "learning_rate": 0.00019997903875490356,
-      "loss": 46.0,
-      "step": 40482
-    },
-    {
-      "epoch": 6.519505616168122,
-      "grad_norm": 0.004554451443254948,
-      "learning_rate": 0.0001999790377190876,
-      "loss": 46.0,
-      "step": 40483
-    },
-    {
-      "epoch": 6.51966665324691,
-      "grad_norm": 0.002264037961140275,
-      "learning_rate": 0.00019997903668324604,
-      "loss": 46.0,
-      "step": 40484
-    },
-    {
-      "epoch": 6.519827690325697,
-      "grad_norm": 0.0024320161901414394,
-      "learning_rate": 0.00019997903564737893,
-      "loss": 46.0,
-      "step": 40485
-    },
-    {
-      "epoch": 6.519988727404485,
-      "grad_norm": 0.0007251136703416705,
-      "learning_rate": 0.0001999790346114862,
-      "loss": 46.0,
-      "step": 40486
-    },
-    {
-      "epoch": 6.520149764483272,
-      "grad_norm": 0.0049898019060492516,
-      "learning_rate": 0.0001999790335755679,
-      "loss": 46.0,
-      "step": 40487
-    },
-    {
-      "epoch": 6.52031080156206,
-      "grad_norm": 0.012271677143871784,
-      "learning_rate": 0.00019997903253962397,
-      "loss": 46.0,
-      "step": 40488
-    },
-    {
-      "epoch": 6.520471838640847,
-      "grad_norm": 0.009346774779260159,
-      "learning_rate": 0.00019997903150365448,
-      "loss": 46.0,
-      "step": 40489
-    },
-    {
-      "epoch": 6.520632875719635,
-      "grad_norm": 0.014756559394299984,
-      "learning_rate": 0.0001999790304676594,
-      "loss": 46.0,
-      "step": 40490
-    },
-    {
-      "epoch": 6.520793912798422,
-      "grad_norm": 0.0011654084082692862,
-      "learning_rate": 0.00019997902943163874,
-      "loss": 46.0,
-      "step": 40491
-    },
-    {
-      "epoch": 6.5209549498772095,
-      "grad_norm": 0.003624498378485441,
-      "learning_rate": 0.00019997902839559247,
-      "loss": 46.0,
-      "step": 40492
-    },
-    {
-      "epoch": 6.521115986955997,
-      "grad_norm": 0.004516322165727615,
-      "learning_rate": 0.0001999790273595206,
-      "loss": 46.0,
-      "step": 40493
-    },
-    {
-      "epoch": 6.521277024034784,
-      "grad_norm": 0.012934727594256401,
-      "learning_rate": 0.00019997902632342318,
-      "loss": 46.0,
-      "step": 40494
-    },
-    {
-      "epoch": 6.521438061113571,
-      "grad_norm": 0.001190581708215177,
-      "learning_rate": 0.00019997902528730014,
-      "loss": 46.0,
-      "step": 40495
-    },
-    {
-      "epoch": 6.521599098192359,
-      "grad_norm": 0.009363081306219101,
-      "learning_rate": 0.00019997902425115155,
-      "loss": 46.0,
-      "step": 40496
-    },
-    {
-      "epoch": 6.521760135271146,
-      "grad_norm": 0.007522832602262497,
-      "learning_rate": 0.00019997902321497734,
-      "loss": 46.0,
-      "step": 40497
-    },
-    {
-      "epoch": 6.521921172349933,
-      "grad_norm": 0.003615233115851879,
-      "learning_rate": 0.00019997902217877754,
-      "loss": 46.0,
-      "step": 40498
-    },
-    {
-      "epoch": 6.522082209428721,
-      "grad_norm": 0.0009744943236000836,
-      "learning_rate": 0.00019997902114255212,
-      "loss": 46.0,
-      "step": 40499
-    },
-    {
-      "epoch": 6.522243246507508,
-      "grad_norm": 0.007887487299740314,
-      "learning_rate": 0.00019997902010630115,
-      "loss": 46.0,
-      "step": 40500
-    },
-    {
-      "epoch": 6.522404283586296,
-      "grad_norm": 0.0020922317635267973,
-      "learning_rate": 0.0001999790190700246,
-      "loss": 46.0,
-      "step": 40501
-    },
-    {
-      "epoch": 6.522565320665083,
-      "grad_norm": 0.00752679817378521,
-      "learning_rate": 0.00019997901803372244,
-      "loss": 46.0,
-      "step": 40502
-    },
-    {
-      "epoch": 6.5227263577438706,
-      "grad_norm": 0.013033091090619564,
-      "learning_rate": 0.0001999790169973947,
-      "loss": 46.0,
-      "step": 40503
-    },
-    {
-      "epoch": 6.522887394822658,
-      "grad_norm": 0.007432173006236553,
-      "learning_rate": 0.00019997901596104133,
-      "loss": 46.0,
-      "step": 40504
-    },
-    {
-      "epoch": 6.5230484319014455,
-      "grad_norm": 0.0029655941762030125,
-      "learning_rate": 0.00019997901492466242,
-      "loss": 46.0,
-      "step": 40505
-    },
-    {
-      "epoch": 6.523209468980233,
-      "grad_norm": 0.0011716659646481276,
-      "learning_rate": 0.0001999790138882579,
-      "loss": 46.0,
-      "step": 40506
-    },
-    {
-      "epoch": 6.52337050605902,
-      "grad_norm": 0.0024206924717873335,
-      "learning_rate": 0.00019997901285182781,
-      "loss": 46.0,
-      "step": 40507
-    },
-    {
-      "epoch": 6.523531543137808,
-      "grad_norm": 0.005756554193794727,
-      "learning_rate": 0.00019997901181537212,
-      "loss": 46.0,
-      "step": 40508
-    },
-    {
-      "epoch": 6.523692580216595,
-      "grad_norm": 0.0036053687799721956,
-      "learning_rate": 0.00019997901077889083,
-      "loss": 46.0,
-      "step": 40509
-    },
-    {
-      "epoch": 6.523853617295382,
-      "grad_norm": 0.003992556128650904,
-      "learning_rate": 0.00019997900974238396,
-      "loss": 46.0,
-      "step": 40510
-    },
-    {
-      "epoch": 6.524014654374169,
-      "grad_norm": 0.003620494157075882,
-      "learning_rate": 0.0001999790087058515,
-      "loss": 46.0,
-      "step": 40511
-    },
-    {
-      "epoch": 6.524175691452957,
-      "grad_norm": 0.016325142234563828,
-      "learning_rate": 0.00019997900766929343,
-      "loss": 46.0,
-      "step": 40512
-    },
-    {
-      "epoch": 6.524336728531744,
-      "grad_norm": 0.010594401508569717,
-      "learning_rate": 0.00019997900663270982,
-      "loss": 46.0,
-      "step": 40513
-    },
-    {
-      "epoch": 6.524497765610532,
-      "grad_norm": 0.01333087868988514,
-      "learning_rate": 0.00019997900559610057,
-      "loss": 46.0,
-      "step": 40514
-    },
-    {
-      "epoch": 6.524658802689319,
-      "grad_norm": 0.003551728092133999,
-      "learning_rate": 0.00019997900455946574,
-      "loss": 46.0,
-      "step": 40515
-    },
-    {
-      "epoch": 6.5248198397681065,
-      "grad_norm": 0.010311528109014034,
-      "learning_rate": 0.00019997900352280534,
-      "loss": 46.0,
-      "step": 40516
-    },
-    {
-      "epoch": 6.524980876846894,
-      "grad_norm": 0.00210933992639184,
-      "learning_rate": 0.00019997900248611933,
-      "loss": 46.0,
-      "step": 40517
-    },
-    {
-      "epoch": 6.525141913925681,
-      "grad_norm": 0.0017669072840362787,
-      "learning_rate": 0.00019997900144940776,
-      "loss": 46.0,
-      "step": 40518
-    },
-    {
-      "epoch": 6.525302951004469,
-      "grad_norm": 0.003850339911878109,
-      "learning_rate": 0.00019997900041267058,
-      "loss": 46.0,
-      "step": 40519
-    },
-    {
-      "epoch": 6.525463988083256,
-      "grad_norm": 0.009675009176135063,
-      "learning_rate": 0.0001999789993759078,
-      "loss": 46.0,
-      "step": 40520
-    },
-    {
-      "epoch": 6.525625025162044,
-      "grad_norm": 0.005492130294442177,
-      "learning_rate": 0.00019997899833911945,
-      "loss": 46.0,
-      "step": 40521
-    },
-    {
-      "epoch": 6.525786062240831,
-      "grad_norm": 0.0045510828495025635,
-      "learning_rate": 0.0001999789973023055,
-      "loss": 46.0,
-      "step": 40522
-    },
-    {
-      "epoch": 6.525947099319619,
-      "grad_norm": 0.005213772412389517,
-      "learning_rate": 0.00019997899626546597,
-      "loss": 46.0,
-      "step": 40523
-    },
-    {
-      "epoch": 6.526108136398406,
-      "grad_norm": 0.00515779247507453,
-      "learning_rate": 0.00019997899522860085,
-      "loss": 46.0,
-      "step": 40524
-    },
-    {
-      "epoch": 6.526269173477193,
-      "grad_norm": 0.005757125560194254,
-      "learning_rate": 0.00019997899419171012,
-      "loss": 46.0,
-      "step": 40525
-    },
-    {
-      "epoch": 6.52643021055598,
-      "grad_norm": 0.005522201303392649,
-      "learning_rate": 0.00019997899315479383,
-      "loss": 46.0,
-      "step": 40526
-    },
-    {
-      "epoch": 6.526591247634768,
-      "grad_norm": 0.01856265403330326,
-      "learning_rate": 0.00019997899211785192,
-      "loss": 46.0,
-      "step": 40527
-    },
-    {
-      "epoch": 6.526752284713555,
-      "grad_norm": 0.004763638600707054,
-      "learning_rate": 0.00019997899108088442,
-      "loss": 46.0,
-      "step": 40528
-    },
-    {
-      "epoch": 6.5269133217923425,
-      "grad_norm": 0.018956495448946953,
-      "learning_rate": 0.00019997899004389137,
-      "loss": 46.0,
-      "step": 40529
-    },
-    {
-      "epoch": 6.52707435887113,
-      "grad_norm": 0.0008553876541554928,
-      "learning_rate": 0.0001999789890068727,
-      "loss": 46.0,
-      "step": 40530
-    },
-    {
-      "epoch": 6.527235395949917,
-      "grad_norm": 0.0019558309577405453,
-      "learning_rate": 0.00019997898796982847,
-      "loss": 46.0,
-      "step": 40531
-    },
-    {
-      "epoch": 6.527396433028705,
-      "grad_norm": 0.002452574437484145,
-      "learning_rate": 0.00019997898693275862,
-      "loss": 46.0,
-      "step": 40532
-    },
-    {
-      "epoch": 6.527557470107492,
-      "grad_norm": 0.0070543475449085236,
-      "learning_rate": 0.0001999789858956632,
-      "loss": 46.0,
-      "step": 40533
-    },
-    {
-      "epoch": 6.52771850718628,
-      "grad_norm": 0.0019224740099161863,
-      "learning_rate": 0.00019997898485854217,
-      "loss": 46.0,
-      "step": 40534
-    },
-    {
-      "epoch": 6.527879544265067,
-      "grad_norm": 0.0032826277893036604,
-      "learning_rate": 0.00019997898382139554,
-      "loss": 46.0,
-      "step": 40535
-    },
-    {
-      "epoch": 6.528040581343855,
-      "grad_norm": 0.0015773138729855418,
-      "learning_rate": 0.00019997898278422335,
-      "loss": 46.0,
-      "step": 40536
-    },
-    {
-      "epoch": 6.528201618422642,
-      "grad_norm": 0.002164094476029277,
-      "learning_rate": 0.00019997898174702557,
-      "loss": 46.0,
-      "step": 40537
-    },
-    {
-      "epoch": 6.52836265550143,
-      "grad_norm": 0.011215342208743095,
-      "learning_rate": 0.0001999789807098022,
-      "loss": 46.0,
-      "step": 40538
-    },
-    {
-      "epoch": 6.528523692580217,
-      "grad_norm": 0.008242777548730373,
-      "learning_rate": 0.00019997897967255322,
-      "loss": 46.0,
-      "step": 40539
-    },
-    {
-      "epoch": 6.528684729659004,
-      "grad_norm": 0.009067090228199959,
-      "learning_rate": 0.00019997897863527865,
-      "loss": 46.0,
-      "step": 40540
-    },
-    {
-      "epoch": 6.528845766737791,
-      "grad_norm": 0.004935994744300842,
-      "learning_rate": 0.00019997897759797852,
-      "loss": 46.0,
-      "step": 40541
-    },
-    {
-      "epoch": 6.5290068038165785,
-      "grad_norm": 0.0013682785211130977,
-      "learning_rate": 0.00019997897656065278,
-      "loss": 46.0,
-      "step": 40542
-    },
-    {
-      "epoch": 6.529167840895366,
-      "grad_norm": 0.0028611968737095594,
-      "learning_rate": 0.00019997897552330145,
-      "loss": 46.0,
-      "step": 40543
-    },
-    {
-      "epoch": 6.529328877974153,
-      "grad_norm": 0.004627167247235775,
-      "learning_rate": 0.00019997897448592453,
-      "loss": 46.0,
-      "step": 40544
-    },
-    {
-      "epoch": 6.529489915052941,
-      "grad_norm": 0.007628864608705044,
-      "learning_rate": 0.000199978973448522,
-      "loss": 46.0,
-      "step": 40545
-    },
-    {
-      "epoch": 6.529650952131728,
-      "grad_norm": 0.011871740221977234,
-      "learning_rate": 0.00019997897241109394,
-      "loss": 46.0,
-      "step": 40546
-    },
-    {
-      "epoch": 6.529811989210516,
-      "grad_norm": 0.003553625661879778,
-      "learning_rate": 0.00019997897137364023,
-      "loss": 46.0,
-      "step": 40547
-    },
-    {
-      "epoch": 6.529973026289303,
-      "grad_norm": 0.0022385010961443186,
-      "learning_rate": 0.00019997897033616096,
-      "loss": 46.0,
-      "step": 40548
-    },
-    {
-      "epoch": 6.530134063368091,
-      "grad_norm": 0.004715350456535816,
-      "learning_rate": 0.0001999789692986561,
-      "loss": 46.0,
-      "step": 40549
-    },
-    {
-      "epoch": 6.530295100446878,
-      "grad_norm": 0.0049905963242053986,
-      "learning_rate": 0.00019997896826112567,
-      "loss": 46.0,
-      "step": 40550
-    },
-    {
-      "epoch": 6.530456137525666,
-      "grad_norm": 0.0024281549267470837,
-      "learning_rate": 0.00019997896722356962,
-      "loss": 46.0,
-      "step": 40551
-    },
-    {
-      "epoch": 6.530617174604453,
-      "grad_norm": 0.004370152018964291,
-      "learning_rate": 0.000199978966185988,
-      "loss": 46.0,
-      "step": 40552
-    },
-    {
-      "epoch": 6.53077821168324,
-      "grad_norm": 0.008099764585494995,
-      "learning_rate": 0.00019997896514838077,
-      "loss": 46.0,
-      "step": 40553
-    },
-    {
-      "epoch": 6.530939248762028,
-      "grad_norm": 0.0015924018807709217,
-      "learning_rate": 0.00019997896411074796,
-      "loss": 46.0,
-      "step": 40554
-    },
-    {
-      "epoch": 6.5311002858408145,
-      "grad_norm": 0.0010014927247539163,
-      "learning_rate": 0.00019997896307308955,
-      "loss": 46.0,
-      "step": 40555
-    },
-    {
-      "epoch": 6.531261322919602,
-      "grad_norm": 0.003553119720891118,
-      "learning_rate": 0.00019997896203540556,
-      "loss": 46.0,
-      "step": 40556
-    },
-    {
-      "epoch": 6.531422359998389,
-      "grad_norm": 0.0023748748935759068,
-      "learning_rate": 0.000199978960997696,
-      "loss": 46.0,
-      "step": 40557
-    },
-    {
-      "epoch": 6.531583397077177,
-      "grad_norm": 0.005541473627090454,
-      "learning_rate": 0.00019997895995996082,
-      "loss": 46.0,
-      "step": 40558
-    },
-    {
-      "epoch": 6.531744434155964,
-      "grad_norm": 0.015427596867084503,
-      "learning_rate": 0.00019997895892220007,
-      "loss": 46.0,
-      "step": 40559
-    },
-    {
-      "epoch": 6.531905471234752,
-      "grad_norm": 0.006583834998309612,
-      "learning_rate": 0.00019997895788441373,
-      "loss": 46.0,
-      "step": 40560
-    },
-    {
-      "epoch": 6.532066508313539,
-      "grad_norm": 0.005457608494907618,
-      "learning_rate": 0.00019997895684660178,
-      "loss": 46.0,
-      "step": 40561
-    },
-    {
-      "epoch": 6.532227545392327,
-      "grad_norm": 0.00598899507895112,
-      "learning_rate": 0.00019997895580876426,
-      "loss": 46.0,
-      "step": 40562
-    },
-    {
-      "epoch": 6.532388582471114,
-      "grad_norm": 0.0022378363646566868,
-      "learning_rate": 0.00019997895477090114,
-      "loss": 46.0,
-      "step": 40563
-    },
-    {
-      "epoch": 6.5325496195499015,
-      "grad_norm": 0.0029661315493285656,
-      "learning_rate": 0.00019997895373301242,
-      "loss": 46.0,
-      "step": 40564
-    },
-    {
-      "epoch": 6.532710656628689,
-      "grad_norm": 0.0018811047775670886,
-      "learning_rate": 0.00019997895269509815,
-      "loss": 46.0,
-      "step": 40565
-    },
-    {
-      "epoch": 6.5328716937074764,
-      "grad_norm": 0.010170129127800465,
-      "learning_rate": 0.00019997895165715826,
-      "loss": 46.0,
-      "step": 40566
-    },
-    {
-      "epoch": 6.533032730786264,
-      "grad_norm": 0.0016801892779767513,
-      "learning_rate": 0.00019997895061919278,
-      "loss": 46.0,
-      "step": 40567
-    },
-    {
-      "epoch": 6.5331937678650505,
-      "grad_norm": 0.0027235024608671665,
-      "learning_rate": 0.00019997894958120172,
-      "loss": 46.0,
-      "step": 40568
-    },
-    {
-      "epoch": 6.533354804943839,
-      "grad_norm": 0.015057950280606747,
-      "learning_rate": 0.00019997894854318507,
-      "loss": 46.0,
-      "step": 40569
-    },
-    {
-      "epoch": 6.533515842022625,
-      "grad_norm": 0.01454103458672762,
-      "learning_rate": 0.00019997894750514283,
-      "loss": 46.0,
-      "step": 40570
-    },
-    {
-      "epoch": 6.533676879101413,
-      "grad_norm": 0.002031525131314993,
-      "learning_rate": 0.00019997894646707498,
-      "loss": 46.0,
-      "step": 40571
-    },
-    {
-      "epoch": 6.5338379161802,
-      "grad_norm": 0.005130839999765158,
-      "learning_rate": 0.00019997894542898156,
-      "loss": 46.0,
-      "step": 40572
-    },
-    {
-      "epoch": 6.533998953258988,
-      "grad_norm": 0.0019787601195275784,
-      "learning_rate": 0.00019997894439086257,
-      "loss": 46.0,
-      "step": 40573
-    },
-    {
-      "epoch": 6.534159990337775,
-      "grad_norm": 0.003488605609163642,
-      "learning_rate": 0.00019997894335271795,
-      "loss": 46.0,
-      "step": 40574
-    },
-    {
-      "epoch": 6.534321027416563,
-      "grad_norm": 0.01663118042051792,
-      "learning_rate": 0.00019997894231454778,
-      "loss": 46.0,
-      "step": 40575
-    },
-    {
-      "epoch": 6.53448206449535,
-      "grad_norm": 0.0049020093865692616,
-      "learning_rate": 0.000199978941276352,
-      "loss": 46.0,
-      "step": 40576
-    },
-    {
-      "epoch": 6.5346431015741375,
-      "grad_norm": 0.0044265445321798325,
-      "learning_rate": 0.00019997894023813061,
-      "loss": 46.0,
-      "step": 40577
-    },
-    {
-      "epoch": 6.534804138652925,
-      "grad_norm": 0.008141937665641308,
-      "learning_rate": 0.00019997893919988365,
-      "loss": 46.0,
-      "step": 40578
-    },
-    {
-      "epoch": 6.534965175731712,
-      "grad_norm": 0.009648614563047886,
-      "learning_rate": 0.0001999789381616111,
-      "loss": 46.0,
-      "step": 40579
-    },
-    {
-      "epoch": 6.5351262128105,
-      "grad_norm": 0.015393831767141819,
-      "learning_rate": 0.000199978937123313,
-      "loss": 46.0,
-      "step": 40580
-    },
-    {
-      "epoch": 6.535287249889287,
-      "grad_norm": 0.006348058115690947,
-      "learning_rate": 0.00019997893608498924,
-      "loss": 46.0,
-      "step": 40581
-    },
-    {
-      "epoch": 6.535448286968075,
-      "grad_norm": 0.006249545142054558,
-      "learning_rate": 0.00019997893504663993,
-      "loss": 46.0,
-      "step": 40582
-    },
-    {
-      "epoch": 6.535609324046861,
-      "grad_norm": 0.004322050604969263,
-      "learning_rate": 0.00019997893400826503,
-      "loss": 46.0,
-      "step": 40583
-    },
-    {
-      "epoch": 6.535770361125649,
-      "grad_norm": 0.011012271046638489,
-      "learning_rate": 0.00019997893296986455,
-      "loss": 46.0,
-      "step": 40584
-    },
-    {
-      "epoch": 6.535931398204436,
-      "grad_norm": 0.0016725219320505857,
-      "learning_rate": 0.00019997893193143847,
-      "loss": 46.0,
-      "step": 40585
-    },
-    {
-      "epoch": 6.536092435283224,
-      "grad_norm": 0.0028530291747301817,
-      "learning_rate": 0.0001999789308929868,
-      "loss": 46.0,
-      "step": 40586
-    },
-    {
-      "epoch": 6.536253472362011,
-      "grad_norm": 0.00382414273917675,
-      "learning_rate": 0.00019997892985450951,
-      "loss": 46.0,
-      "step": 40587
-    },
-    {
-      "epoch": 6.536414509440799,
-      "grad_norm": 0.009705050848424435,
-      "learning_rate": 0.00019997892881600665,
-      "loss": 46.0,
-      "step": 40588
-    },
-    {
-      "epoch": 6.536575546519586,
-      "grad_norm": 0.011702614836394787,
-      "learning_rate": 0.00019997892777747823,
-      "loss": 46.0,
-      "step": 40589
-    },
-    {
-      "epoch": 6.5367365835983735,
-      "grad_norm": 0.012955775484442711,
-      "learning_rate": 0.00019997892673892417,
-      "loss": 46.0,
-      "step": 40590
-    },
-    {
-      "epoch": 6.536897620677161,
-      "grad_norm": 0.0059370798990130424,
-      "learning_rate": 0.00019997892570034455,
-      "loss": 46.0,
-      "step": 40591
-    },
-    {
-      "epoch": 6.537058657755948,
-      "grad_norm": 0.002202889183536172,
-      "learning_rate": 0.00019997892466173936,
-      "loss": 46.0,
-      "step": 40592
-    },
-    {
-      "epoch": 6.537219694834736,
-      "grad_norm": 0.003232336137443781,
-      "learning_rate": 0.00019997892362310857,
-      "loss": 46.0,
-      "step": 40593
-    },
-    {
-      "epoch": 6.537380731913523,
-      "grad_norm": 0.003538022981956601,
-      "learning_rate": 0.00019997892258445216,
-      "loss": 46.0,
-      "step": 40594
-    },
-    {
-      "epoch": 6.537541768992311,
-      "grad_norm": 0.006802130024880171,
-      "learning_rate": 0.00019997892154577018,
-      "loss": 46.0,
-      "step": 40595
-    },
-    {
-      "epoch": 6.537702806071098,
-      "grad_norm": 0.0020270701497793198,
-      "learning_rate": 0.00019997892050706263,
-      "loss": 46.0,
-      "step": 40596
-    },
-    {
-      "epoch": 6.537863843149886,
-      "grad_norm": 0.011947513557970524,
-      "learning_rate": 0.00019997891946832948,
-      "loss": 46.0,
-      "step": 40597
-    },
-    {
-      "epoch": 6.538024880228672,
-      "grad_norm": 0.004161762539297342,
-      "learning_rate": 0.0001999789184295707,
-      "loss": 46.0,
-      "step": 40598
-    },
-    {
-      "epoch": 6.53818591730746,
-      "grad_norm": 0.001710326294414699,
-      "learning_rate": 0.00019997891739078637,
-      "loss": 46.0,
-      "step": 40599
-    },
-    {
-      "epoch": 6.538346954386247,
-      "grad_norm": 0.008246785029768944,
-      "learning_rate": 0.00019997891635197646,
-      "loss": 46.0,
-      "step": 40600
-    },
-    {
-      "epoch": 6.538507991465035,
-      "grad_norm": 0.001023536897264421,
-      "learning_rate": 0.00019997891531314092,
-      "loss": 46.0,
-      "step": 40601
-    },
-    {
-      "epoch": 6.538669028543822,
-      "grad_norm": 0.005823458544909954,
-      "learning_rate": 0.00019997891427427983,
-      "loss": 46.0,
-      "step": 40602
-    },
-    {
-      "epoch": 6.5388300656226095,
-      "grad_norm": 0.0016505605308339,
-      "learning_rate": 0.00019997891323539314,
-      "loss": 46.0,
-      "step": 40603
-    },
-    {
-      "epoch": 6.538991102701397,
-      "grad_norm": 0.004465322941541672,
-      "learning_rate": 0.00019997891219648083,
-      "loss": 46.0,
-      "step": 40604
-    },
-    {
-      "epoch": 6.539152139780184,
-      "grad_norm": 0.002746766898781061,
-      "learning_rate": 0.00019997891115754296,
-      "loss": 46.0,
-      "step": 40605
-    },
-    {
-      "epoch": 6.539313176858972,
-      "grad_norm": 0.010930893942713737,
-      "learning_rate": 0.00019997891011857953,
-      "loss": 46.0,
-      "step": 40606
-    },
-    {
-      "epoch": 6.539474213937759,
-      "grad_norm": 0.005336800590157509,
-      "learning_rate": 0.00019997890907959046,
-      "loss": 46.0,
-      "step": 40607
-    },
-    {
-      "epoch": 6.539635251016547,
-      "grad_norm": 0.001621581264771521,
-      "learning_rate": 0.00019997890804057582,
-      "loss": 46.0,
-      "step": 40608
-    },
-    {
-      "epoch": 6.539796288095334,
-      "grad_norm": 0.005950243212282658,
-      "learning_rate": 0.00019997890700153558,
-      "loss": 46.0,
-      "step": 40609
-    },
-    {
-      "epoch": 6.539957325174122,
-      "grad_norm": 0.008815617300570011,
-      "learning_rate": 0.00019997890596246975,
-      "loss": 46.0,
-      "step": 40610
-    },
-    {
-      "epoch": 6.540118362252909,
-      "grad_norm": 0.002849416807293892,
-      "learning_rate": 0.00019997890492337835,
-      "loss": 46.0,
-      "step": 40611
-    },
-    {
-      "epoch": 6.5402793993316966,
-      "grad_norm": 0.0018635190790519118,
-      "learning_rate": 0.00019997890388426135,
-      "loss": 46.0,
-      "step": 40612
-    },
-    {
-      "epoch": 6.540440436410483,
-      "grad_norm": 0.008092408068478107,
-      "learning_rate": 0.00019997890284511875,
-      "loss": 46.0,
-      "step": 40613
-    },
-    {
-      "epoch": 6.540601473489271,
-      "grad_norm": 0.0025586779229342937,
-      "learning_rate": 0.00019997890180595057,
-      "loss": 46.0,
-      "step": 40614
-    },
-    {
-      "epoch": 6.540762510568058,
-      "grad_norm": 0.0016096540493890643,
-      "learning_rate": 0.00019997890076675683,
-      "loss": 46.0,
-      "step": 40615
-    },
-    {
-      "epoch": 6.5409235476468455,
-      "grad_norm": 0.01131502166390419,
-      "learning_rate": 0.00019997889972753744,
-      "loss": 46.0,
-      "step": 40616
-    },
-    {
-      "epoch": 6.541084584725633,
-      "grad_norm": 0.00605749199166894,
-      "learning_rate": 0.00019997889868829253,
-      "loss": 46.0,
-      "step": 40617
-    },
-    {
-      "epoch": 6.54124562180442,
-      "grad_norm": 0.007547752000391483,
-      "learning_rate": 0.00019997889764902197,
-      "loss": 46.0,
-      "step": 40618
-    },
-    {
-      "epoch": 6.541406658883208,
-      "grad_norm": 0.00610667048022151,
-      "learning_rate": 0.00019997889660972585,
-      "loss": 46.0,
-      "step": 40619
-    },
-    {
-      "epoch": 6.541567695961995,
-      "grad_norm": 0.013484025374054909,
-      "learning_rate": 0.00019997889557040415,
-      "loss": 46.0,
-      "step": 40620
-    },
-    {
-      "epoch": 6.541728733040783,
-      "grad_norm": 0.00537380576133728,
-      "learning_rate": 0.00019997889453105683,
-      "loss": 46.0,
-      "step": 40621
-    },
-    {
-      "epoch": 6.54188977011957,
-      "grad_norm": 0.002446510596200824,
-      "learning_rate": 0.00019997889349168392,
-      "loss": 46.0,
-      "step": 40622
-    },
-    {
-      "epoch": 6.542050807198358,
-      "grad_norm": 0.003181547624990344,
-      "learning_rate": 0.00019997889245228546,
-      "loss": 46.0,
-      "step": 40623
-    },
-    {
-      "epoch": 6.542211844277145,
-      "grad_norm": 0.017466126009821892,
-      "learning_rate": 0.0001999788914128614,
-      "loss": 46.0,
-      "step": 40624
-    },
-    {
-      "epoch": 6.5423728813559325,
-      "grad_norm": 0.0033442063722759485,
-      "learning_rate": 0.0001999788903734117,
-      "loss": 46.0,
-      "step": 40625
-    },
-    {
-      "epoch": 6.542533918434719,
-      "grad_norm": 0.0013541518710553646,
-      "learning_rate": 0.00019997888933393645,
-      "loss": 46.0,
-      "step": 40626
-    },
-    {
-      "epoch": 6.542694955513507,
-      "grad_norm": 0.0019404829945415258,
-      "learning_rate": 0.0001999788882944356,
-      "loss": 46.0,
-      "step": 40627
-    },
-    {
-      "epoch": 6.542855992592294,
-      "grad_norm": 0.003663154784590006,
-      "learning_rate": 0.00019997888725490918,
-      "loss": 46.0,
-      "step": 40628
-    },
-    {
-      "epoch": 6.5430170296710815,
-      "grad_norm": 0.0036188082303851843,
-      "learning_rate": 0.00019997888621535716,
-      "loss": 46.0,
-      "step": 40629
-    },
-    {
-      "epoch": 6.543178066749869,
-      "grad_norm": 0.01687401905655861,
-      "learning_rate": 0.00019997888517577956,
-      "loss": 46.0,
-      "step": 40630
-    },
-    {
-      "epoch": 6.543339103828656,
-      "grad_norm": 0.0010863416828215122,
-      "learning_rate": 0.00019997888413617637,
-      "loss": 46.0,
-      "step": 40631
-    },
-    {
-      "epoch": 6.543500140907444,
-      "grad_norm": 0.011693691834807396,
-      "learning_rate": 0.00019997888309654756,
-      "loss": 46.0,
-      "step": 40632
-    },
-    {
-      "epoch": 6.543661177986231,
-      "grad_norm": 0.00698266364634037,
-      "learning_rate": 0.0001999788820568932,
-      "loss": 46.0,
-      "step": 40633
-    },
-    {
-      "epoch": 6.543822215065019,
-      "grad_norm": 0.0019774027168750763,
-      "learning_rate": 0.00019997888101721322,
-      "loss": 46.0,
-      "step": 40634
-    },
-    {
-      "epoch": 6.543983252143806,
-      "grad_norm": 0.006216689944267273,
-      "learning_rate": 0.00019997887997750765,
-      "loss": 46.0,
-      "step": 40635
-    },
-    {
-      "epoch": 6.544144289222594,
-      "grad_norm": 0.015553610399365425,
-      "learning_rate": 0.00019997887893777653,
-      "loss": 46.0,
-      "step": 40636
-    },
-    {
-      "epoch": 6.544305326301381,
-      "grad_norm": 0.004962827544659376,
-      "learning_rate": 0.00019997887789801978,
-      "loss": 46.0,
-      "step": 40637
-    },
-    {
-      "epoch": 6.5444663633801685,
-      "grad_norm": 0.003268804866820574,
-      "learning_rate": 0.00019997887685823743,
-      "loss": 46.0,
-      "step": 40638
-    },
-    {
-      "epoch": 6.544627400458956,
-      "grad_norm": 0.019117962568998337,
-      "learning_rate": 0.00019997887581842954,
-      "loss": 46.0,
-      "step": 40639
-    },
-    {
-      "epoch": 6.544788437537743,
-      "grad_norm": 0.0073148044757544994,
-      "learning_rate": 0.00019997887477859604,
-      "loss": 46.0,
-      "step": 40640
-    },
-    {
-      "epoch": 6.54494947461653,
-      "grad_norm": 0.0012962461914867163,
-      "learning_rate": 0.00019997887373873695,
-      "loss": 46.0,
-      "step": 40641
-    },
-    {
-      "epoch": 6.545110511695318,
-      "grad_norm": 0.005489963572472334,
-      "learning_rate": 0.00019997887269885225,
-      "loss": 46.0,
-      "step": 40642
-    },
-    {
-      "epoch": 6.545271548774105,
-      "grad_norm": 0.005328271072357893,
-      "learning_rate": 0.00019997887165894198,
-      "loss": 46.0,
-      "step": 40643
-    },
-    {
-      "epoch": 6.545432585852892,
-      "grad_norm": 0.010754041373729706,
-      "learning_rate": 0.00019997887061900613,
-      "loss": 46.0,
-      "step": 40644
-    },
-    {
-      "epoch": 6.54559362293168,
-      "grad_norm": 0.0083598168566823,
-      "learning_rate": 0.00019997886957904466,
-      "loss": 46.0,
-      "step": 40645
-    },
-    {
-      "epoch": 6.545754660010467,
-      "grad_norm": 0.001120619010180235,
-      "learning_rate": 0.0001999788685390576,
-      "loss": 46.0,
-      "step": 40646
-    },
-    {
-      "epoch": 6.545915697089255,
-      "grad_norm": 0.0011444613337516785,
-      "learning_rate": 0.000199978867499045,
-      "loss": 46.0,
-      "step": 40647
-    },
-    {
-      "epoch": 6.546076734168042,
-      "grad_norm": 0.0014339227927848697,
-      "learning_rate": 0.00019997886645900677,
-      "loss": 46.0,
-      "step": 40648
-    },
-    {
-      "epoch": 6.54623777124683,
-      "grad_norm": 0.0035265947226434946,
-      "learning_rate": 0.00019997886541894296,
-      "loss": 46.0,
-      "step": 40649
-    },
-    {
-      "epoch": 6.546398808325617,
-      "grad_norm": 0.004440826829522848,
-      "learning_rate": 0.00019997886437885356,
-      "loss": 46.0,
-      "step": 40650
-    },
-    {
-      "epoch": 6.5465598454044045,
-      "grad_norm": 0.004230854567140341,
-      "learning_rate": 0.00019997886333873857,
-      "loss": 46.0,
-      "step": 40651
-    },
-    {
-      "epoch": 6.546720882483192,
-      "grad_norm": 0.0028550291899591684,
-      "learning_rate": 0.000199978862298598,
-      "loss": 46.0,
-      "step": 40652
-    },
-    {
-      "epoch": 6.546881919561979,
-      "grad_norm": 0.0029974381905049086,
-      "learning_rate": 0.0001999788612584318,
-      "loss": 46.0,
-      "step": 40653
-    },
-    {
-      "epoch": 6.547042956640767,
-      "grad_norm": 0.006866373587399721,
-      "learning_rate": 0.00019997886021824005,
-      "loss": 46.0,
-      "step": 40654
-    },
-    {
-      "epoch": 6.547203993719554,
-      "grad_norm": 0.0030630505643785,
-      "learning_rate": 0.00019997885917802272,
-      "loss": 46.0,
-      "step": 40655
-    },
-    {
-      "epoch": 6.547365030798341,
-      "grad_norm": 0.002616151934489608,
-      "learning_rate": 0.00019997885813777976,
-      "loss": 46.0,
-      "step": 40656
-    },
-    {
-      "epoch": 6.547526067877129,
-      "grad_norm": 0.006250202655792236,
-      "learning_rate": 0.00019997885709751125,
-      "loss": 46.0,
-      "step": 40657
-    },
-    {
-      "epoch": 6.547687104955916,
-      "grad_norm": 0.0033784371335059404,
-      "learning_rate": 0.00019997885605721715,
-      "loss": 46.0,
-      "step": 40658
-    },
-    {
-      "epoch": 6.547848142034703,
-      "grad_norm": 0.005065184086561203,
-      "learning_rate": 0.00019997885501689744,
-      "loss": 46.0,
-      "step": 40659
-    },
-    {
-      "epoch": 6.548009179113491,
-      "grad_norm": 0.0019097983604297042,
-      "learning_rate": 0.00019997885397655214,
-      "loss": 46.0,
-      "step": 40660
-    },
-    {
-      "epoch": 6.548170216192278,
-      "grad_norm": 0.02277885004878044,
-      "learning_rate": 0.00019997885293618125,
-      "loss": 46.0,
-      "step": 40661
-    },
-    {
-      "epoch": 6.548331253271066,
-      "grad_norm": 0.0019294701050966978,
-      "learning_rate": 0.00019997885189578478,
-      "loss": 46.0,
-      "step": 40662
-    },
-    {
-      "epoch": 6.548492290349853,
-      "grad_norm": 0.007393193896859884,
-      "learning_rate": 0.00019997885085536272,
-      "loss": 46.0,
-      "step": 40663
-    },
-    {
-      "epoch": 6.5486533274286405,
-      "grad_norm": 0.005524695850908756,
-      "learning_rate": 0.00019997884981491507,
-      "loss": 46.0,
-      "step": 40664
-    },
-    {
-      "epoch": 6.548814364507428,
-      "grad_norm": 0.022860087454319,
-      "learning_rate": 0.00019997884877444183,
-      "loss": 46.0,
-      "step": 40665
-    },
-    {
-      "epoch": 6.548975401586215,
-      "grad_norm": 0.008511630818247795,
-      "learning_rate": 0.000199978847733943,
-      "loss": 46.0,
-      "step": 40666
-    },
-    {
-      "epoch": 6.549136438665003,
-      "grad_norm": 0.0018915238324552774,
-      "learning_rate": 0.00019997884669341857,
-      "loss": 46.0,
-      "step": 40667
-    },
-    {
-      "epoch": 6.54929747574379,
-      "grad_norm": 0.00618374440819025,
-      "learning_rate": 0.00019997884565286857,
-      "loss": 46.0,
-      "step": 40668
-    },
-    {
-      "epoch": 6.549458512822578,
-      "grad_norm": 0.00811982061713934,
-      "learning_rate": 0.00019997884461229296,
-      "loss": 46.0,
-      "step": 40669
-    },
-    {
-      "epoch": 6.549619549901365,
-      "grad_norm": 0.015774866566061974,
-      "learning_rate": 0.0001999788435716918,
-      "loss": 46.0,
-      "step": 40670
-    },
-    {
-      "epoch": 6.549780586980152,
-      "grad_norm": 0.005287447478622198,
-      "learning_rate": 0.000199978842531065,
-      "loss": 46.0,
-      "step": 40671
-    },
-    {
-      "epoch": 6.549941624058939,
-      "grad_norm": 0.0025843321345746517,
-      "learning_rate": 0.00019997884149041263,
-      "loss": 46.0,
-      "step": 40672
-    },
-    {
-      "epoch": 6.550102661137727,
-      "grad_norm": 0.011869542300701141,
-      "learning_rate": 0.00019997884044973467,
-      "loss": 46.0,
-      "step": 40673
-    },
-    {
-      "epoch": 6.550263698216514,
-      "grad_norm": 0.004435858689248562,
-      "learning_rate": 0.00019997883940903112,
-      "loss": 46.0,
-      "step": 40674
-    },
-    {
-      "epoch": 6.550424735295302,
-      "grad_norm": 0.0013995266053825617,
-      "learning_rate": 0.000199978838368302,
-      "loss": 46.0,
-      "step": 40675
-    },
-    {
-      "epoch": 6.550585772374089,
-      "grad_norm": 0.003281451528891921,
-      "learning_rate": 0.00019997883732754724,
-      "loss": 46.0,
-      "step": 40676
-    },
-    {
-      "epoch": 6.5507468094528765,
-      "grad_norm": 0.009259171783924103,
-      "learning_rate": 0.00019997883628676696,
-      "loss": 46.0,
-      "step": 40677
-    },
-    {
-      "epoch": 6.550907846531664,
-      "grad_norm": 0.0045952280052006245,
-      "learning_rate": 0.00019997883524596106,
-      "loss": 46.0,
-      "step": 40678
-    },
-    {
-      "epoch": 6.551068883610451,
-      "grad_norm": 0.012239092960953712,
-      "learning_rate": 0.00019997883420512955,
-      "loss": 46.0,
-      "step": 40679
-    },
-    {
-      "epoch": 6.551229920689239,
-      "grad_norm": 0.0015869359485805035,
-      "learning_rate": 0.00019997883316427245,
-      "loss": 46.0,
-      "step": 40680
-    },
-    {
-      "epoch": 6.551390957768026,
-      "grad_norm": 0.003501494647935033,
-      "learning_rate": 0.0001999788321233898,
-      "loss": 46.0,
-      "step": 40681
-    },
-    {
-      "epoch": 6.551551994846814,
-      "grad_norm": 0.0037833626847714186,
-      "learning_rate": 0.00019997883108248155,
-      "loss": 46.0,
-      "step": 40682
-    },
-    {
-      "epoch": 6.551713031925601,
-      "grad_norm": 0.0044454666785895824,
-      "learning_rate": 0.00019997883004154766,
-      "loss": 46.0,
-      "step": 40683
-    },
-    {
-      "epoch": 6.551874069004389,
-      "grad_norm": 0.004663343541324139,
-      "learning_rate": 0.00019997882900058824,
-      "loss": 46.0,
-      "step": 40684
-    },
-    {
-      "epoch": 6.552035106083176,
-      "grad_norm": 0.0018624671502038836,
-      "learning_rate": 0.0001999788279596032,
-      "loss": 46.0,
-      "step": 40685
-    },
-    {
-      "epoch": 6.552196143161963,
-      "grad_norm": 0.01254362240433693,
-      "learning_rate": 0.0001999788269185926,
-      "loss": 46.0,
-      "step": 40686
-    },
-    {
-      "epoch": 6.55235718024075,
-      "grad_norm": 0.004956195130944252,
-      "learning_rate": 0.00019997882587755638,
-      "loss": 46.0,
-      "step": 40687
-    },
-    {
-      "epoch": 6.5525182173195375,
-      "grad_norm": 0.002800451824441552,
-      "learning_rate": 0.00019997882483649456,
-      "loss": 46.0,
-      "step": 40688
-    },
-    {
-      "epoch": 6.552679254398325,
-      "grad_norm": 0.0033860914409160614,
-      "learning_rate": 0.0001999788237954072,
-      "loss": 46.0,
-      "step": 40689
-    },
-    {
-      "epoch": 6.5528402914771124,
-      "grad_norm": 0.007571641821414232,
-      "learning_rate": 0.0001999788227542942,
-      "loss": 46.0,
-      "step": 40690
-    },
-    {
-      "epoch": 6.5530013285559,
-      "grad_norm": 0.004056733567267656,
-      "learning_rate": 0.00019997882171315565,
-      "loss": 46.0,
-      "step": 40691
-    },
-    {
-      "epoch": 6.553162365634687,
-      "grad_norm": 0.007771741598844528,
-      "learning_rate": 0.0001999788206719915,
-      "loss": 46.0,
-      "step": 40692
-    },
-    {
-      "epoch": 6.553323402713475,
-      "grad_norm": 0.006224946118891239,
-      "learning_rate": 0.00019997881963080175,
-      "loss": 46.0,
-      "step": 40693
-    },
-    {
-      "epoch": 6.553484439792262,
-      "grad_norm": 0.013901463709771633,
-      "learning_rate": 0.00019997881858958643,
-      "loss": 46.0,
-      "step": 40694
-    },
-    {
-      "epoch": 6.55364547687105,
-      "grad_norm": 0.0015252531738951802,
-      "learning_rate": 0.0001999788175483455,
-      "loss": 46.0,
-      "step": 40695
-    },
-    {
-      "epoch": 6.553806513949837,
-      "grad_norm": 0.011995387263596058,
-      "learning_rate": 0.00019997881650707898,
-      "loss": 46.0,
-      "step": 40696
-    },
-    {
-      "epoch": 6.553967551028625,
-      "grad_norm": 0.0017018553335219622,
-      "learning_rate": 0.00019997881546578684,
-      "loss": 46.0,
-      "step": 40697
-    },
-    {
-      "epoch": 6.554128588107412,
-      "grad_norm": 0.006945356726646423,
-      "learning_rate": 0.00019997881442446918,
-      "loss": 46.0,
-      "step": 40698
-    },
-    {
-      "epoch": 6.5542896251861995,
-      "grad_norm": 0.0076630376279354095,
-      "learning_rate": 0.0001999788133831259,
-      "loss": 46.0,
-      "step": 40699
-    },
-    {
-      "epoch": 6.554450662264987,
-      "grad_norm": 0.00933765433728695,
-      "learning_rate": 0.00019997881234175703,
-      "loss": 46.0,
-      "step": 40700
-    },
-    {
-      "epoch": 6.5546116993437735,
-      "grad_norm": 0.005538893863558769,
-      "learning_rate": 0.00019997881130036255,
-      "loss": 46.0,
-      "step": 40701
-    },
-    {
-      "epoch": 6.554772736422561,
-      "grad_norm": 0.0009473536629229784,
-      "learning_rate": 0.00019997881025894253,
-      "loss": 46.0,
-      "step": 40702
-    },
-    {
-      "epoch": 6.554933773501348,
-      "grad_norm": 0.0023595639504492283,
-      "learning_rate": 0.00019997880921749687,
-      "loss": 46.0,
-      "step": 40703
-    },
-    {
-      "epoch": 6.555094810580136,
-      "grad_norm": 0.005580158904194832,
-      "learning_rate": 0.00019997880817602563,
-      "loss": 46.0,
-      "step": 40704
-    },
-    {
-      "epoch": 6.555255847658923,
-      "grad_norm": 0.011074107140302658,
-      "learning_rate": 0.00019997880713452882,
-      "loss": 46.0,
-      "step": 40705
-    },
-    {
-      "epoch": 6.555416884737711,
-      "grad_norm": 0.009842784143984318,
-      "learning_rate": 0.0001999788060930064,
-      "loss": 46.0,
-      "step": 40706
-    },
-    {
-      "epoch": 6.555577921816498,
-      "grad_norm": 0.004351463634520769,
-      "learning_rate": 0.0001999788050514584,
-      "loss": 46.0,
-      "step": 40707
-    },
-    {
-      "epoch": 6.555738958895286,
-      "grad_norm": 0.002302167471498251,
-      "learning_rate": 0.00019997880400988484,
-      "loss": 46.0,
-      "step": 40708
-    },
-    {
-      "epoch": 6.555899995974073,
-      "grad_norm": 0.013627825304865837,
-      "learning_rate": 0.00019997880296828563,
-      "loss": 46.0,
-      "step": 40709
-    },
-    {
-      "epoch": 6.556061033052861,
-      "grad_norm": 0.007735341787338257,
-      "learning_rate": 0.0001999788019266609,
-      "loss": 46.0,
-      "step": 40710
-    },
-    {
-      "epoch": 6.556222070131648,
-      "grad_norm": 0.002870397176593542,
-      "learning_rate": 0.00019997880088501053,
-      "loss": 46.0,
-      "step": 40711
-    },
-    {
-      "epoch": 6.5563831072104355,
-      "grad_norm": 0.011340837925672531,
-      "learning_rate": 0.0001999787998433346,
-      "loss": 46.0,
-      "step": 40712
-    },
-    {
-      "epoch": 6.556544144289223,
-      "grad_norm": 0.005011400673538446,
-      "learning_rate": 0.00019997879880163304,
-      "loss": 46.0,
-      "step": 40713
-    },
-    {
-      "epoch": 6.5567051813680095,
-      "grad_norm": 0.0035711501259356737,
-      "learning_rate": 0.00019997879775990592,
-      "loss": 46.0,
-      "step": 40714
-    },
-    {
-      "epoch": 6.556866218446798,
-      "grad_norm": 0.0013839555904269218,
-      "learning_rate": 0.00019997879671815322,
-      "loss": 46.0,
-      "step": 40715
-    },
-    {
-      "epoch": 6.557027255525584,
-      "grad_norm": 0.006828649435192347,
-      "learning_rate": 0.0001999787956763749,
-      "loss": 46.0,
-      "step": 40716
-    },
-    {
-      "epoch": 6.557188292604372,
-      "grad_norm": 0.013376781716942787,
-      "learning_rate": 0.000199978794634571,
-      "loss": 46.0,
-      "step": 40717
-    },
-    {
-      "epoch": 6.557349329683159,
-      "grad_norm": 0.003747333539649844,
-      "learning_rate": 0.00019997879359274153,
-      "loss": 46.0,
-      "step": 40718
-    },
-    {
-      "epoch": 6.557510366761947,
-      "grad_norm": 0.0020217036362737417,
-      "learning_rate": 0.00019997879255088645,
-      "loss": 46.0,
-      "step": 40719
-    },
-    {
-      "epoch": 6.557671403840734,
-      "grad_norm": 0.0012212367728352547,
-      "learning_rate": 0.00019997879150900581,
-      "loss": 46.0,
-      "step": 40720
-    },
-    {
-      "epoch": 6.557832440919522,
-      "grad_norm": 0.003769966773688793,
-      "learning_rate": 0.00019997879046709956,
-      "loss": 46.0,
-      "step": 40721
-    },
-    {
-      "epoch": 6.557993477998309,
-      "grad_norm": 0.005456156097352505,
-      "learning_rate": 0.0001999787894251677,
-      "loss": 46.0,
-      "step": 40722
-    },
-    {
-      "epoch": 6.558154515077097,
-      "grad_norm": 0.005896203685551882,
-      "learning_rate": 0.00019997878838321027,
-      "loss": 46.0,
-      "step": 40723
-    },
-    {
-      "epoch": 6.558315552155884,
-      "grad_norm": 0.014687170274555683,
-      "learning_rate": 0.00019997878734122722,
-      "loss": 46.0,
-      "step": 40724
-    },
-    {
-      "epoch": 6.5584765892346715,
-      "grad_norm": 0.001518176170065999,
-      "learning_rate": 0.00019997878629921865,
-      "loss": 46.0,
-      "step": 40725
-    },
-    {
-      "epoch": 6.558637626313459,
-      "grad_norm": 0.013105345889925957,
-      "learning_rate": 0.00019997878525718443,
-      "loss": 46.0,
-      "step": 40726
-    },
-    {
-      "epoch": 6.558798663392246,
-      "grad_norm": 0.007250647991895676,
-      "learning_rate": 0.00019997878421512466,
-      "loss": 46.0,
-      "step": 40727
-    },
-    {
-      "epoch": 6.558959700471034,
-      "grad_norm": 0.00250617996789515,
-      "learning_rate": 0.00019997878317303927,
-      "loss": 46.0,
-      "step": 40728
-    },
-    {
-      "epoch": 6.55912073754982,
-      "grad_norm": 0.0013489233097061515,
-      "learning_rate": 0.00019997878213092832,
-      "loss": 46.0,
-      "step": 40729
-    },
-    {
-      "epoch": 6.559281774628609,
-      "grad_norm": 0.009942890144884586,
-      "learning_rate": 0.00019997878108879175,
-      "loss": 46.0,
-      "step": 40730
-    },
-    {
-      "epoch": 6.559442811707395,
-      "grad_norm": 0.006347054615616798,
-      "learning_rate": 0.0001999787800466296,
-      "loss": 46.0,
-      "step": 40731
-    },
-    {
-      "epoch": 6.559603848786183,
-      "grad_norm": 0.0033637864980846643,
-      "learning_rate": 0.0001999787790044419,
-      "loss": 46.0,
-      "step": 40732
-    },
-    {
-      "epoch": 6.55976488586497,
-      "grad_norm": 0.006357844918966293,
-      "learning_rate": 0.00019997877796222853,
-      "loss": 46.0,
-      "step": 40733
-    },
-    {
-      "epoch": 6.559925922943758,
-      "grad_norm": 0.005246862303465605,
-      "learning_rate": 0.00019997877691998962,
-      "loss": 46.0,
-      "step": 40734
-    },
-    {
-      "epoch": 6.560086960022545,
-      "grad_norm": 0.0019441910553723574,
-      "learning_rate": 0.00019997877587772512,
-      "loss": 46.0,
-      "step": 40735
-    },
-    {
-      "epoch": 6.5602479971013326,
-      "grad_norm": 0.002969289431348443,
-      "learning_rate": 0.00019997877483543506,
-      "loss": 46.0,
-      "step": 40736
-    },
-    {
-      "epoch": 6.56040903418012,
-      "grad_norm": 0.010823934338986874,
-      "learning_rate": 0.00019997877379311936,
-      "loss": 46.0,
-      "step": 40737
-    },
-    {
-      "epoch": 6.5605700712589075,
-      "grad_norm": 0.0025764349848031998,
-      "learning_rate": 0.00019997877275077807,
-      "loss": 46.0,
-      "step": 40738
-    },
-    {
-      "epoch": 6.560731108337695,
-      "grad_norm": 0.0011528560426086187,
-      "learning_rate": 0.00019997877170841122,
-      "loss": 46.0,
-      "step": 40739
-    },
-    {
-      "epoch": 6.560892145416482,
-      "grad_norm": 0.004093565978109837,
-      "learning_rate": 0.00019997877066601878,
-      "loss": 46.0,
-      "step": 40740
-    },
-    {
-      "epoch": 6.56105318249527,
-      "grad_norm": 0.0027801082469522953,
-      "learning_rate": 0.00019997876962360073,
-      "loss": 46.0,
-      "step": 40741
-    },
-    {
-      "epoch": 6.561214219574057,
-      "grad_norm": 0.0014597009867429733,
-      "learning_rate": 0.00019997876858115712,
-      "loss": 46.0,
-      "step": 40742
-    },
-    {
-      "epoch": 6.561375256652845,
-      "grad_norm": 0.006947056390345097,
-      "learning_rate": 0.0001999787675386879,
-      "loss": 46.0,
-      "step": 40743
-    },
-    {
-      "epoch": 6.561536293731631,
-      "grad_norm": 0.0034314082004129887,
-      "learning_rate": 0.00019997876649619308,
-      "loss": 46.0,
-      "step": 40744
-    },
-    {
-      "epoch": 6.561697330810419,
-      "grad_norm": 0.0020651633385568857,
-      "learning_rate": 0.00019997876545367268,
-      "loss": 46.0,
-      "step": 40745
-    },
-    {
-      "epoch": 6.561858367889206,
-      "grad_norm": 0.018291432410478592,
-      "learning_rate": 0.0001999787644111267,
-      "loss": 46.0,
-      "step": 40746
-    },
-    {
-      "epoch": 6.562019404967994,
-      "grad_norm": 0.009931794367730618,
-      "learning_rate": 0.00019997876336855512,
-      "loss": 46.0,
-      "step": 40747
-    },
-    {
-      "epoch": 6.562180442046781,
-      "grad_norm": 0.006406045518815517,
-      "learning_rate": 0.00019997876232595796,
-      "loss": 46.0,
-      "step": 40748
-    },
-    {
-      "epoch": 6.5623414791255685,
-      "grad_norm": 0.0034832200035452843,
-      "learning_rate": 0.00019997876128333521,
-      "loss": 46.0,
-      "step": 40749
-    },
-    {
-      "epoch": 6.562502516204356,
-      "grad_norm": 0.004517795518040657,
-      "learning_rate": 0.00019997876024068685,
-      "loss": 46.0,
-      "step": 40750
-    },
-    {
-      "epoch": 6.562663553283143,
-      "grad_norm": 0.006442656274884939,
-      "learning_rate": 0.00019997875919801296,
-      "loss": 46.0,
-      "step": 40751
-    },
-    {
-      "epoch": 6.562824590361931,
-      "grad_norm": 0.004893993493169546,
-      "learning_rate": 0.00019997875815531342,
-      "loss": 46.0,
-      "step": 40752
-    },
-    {
-      "epoch": 6.562985627440718,
-      "grad_norm": 0.002450728788971901,
-      "learning_rate": 0.0001999787571125883,
-      "loss": 46.0,
-      "step": 40753
-    },
-    {
-      "epoch": 6.563146664519506,
-      "grad_norm": 0.004116914700716734,
-      "learning_rate": 0.00019997875606983759,
-      "loss": 46.0,
-      "step": 40754
-    },
-    {
-      "epoch": 6.563307701598293,
-      "grad_norm": 0.003956758417189121,
-      "learning_rate": 0.00019997875502706131,
-      "loss": 46.0,
-      "step": 40755
-    },
-    {
-      "epoch": 6.563468738677081,
-      "grad_norm": 0.017810363322496414,
-      "learning_rate": 0.00019997875398425943,
-      "loss": 46.0,
-      "step": 40756
-    },
-    {
-      "epoch": 6.563629775755868,
-      "grad_norm": 0.0020648043137043715,
-      "learning_rate": 0.00019997875294143196,
-      "loss": 46.0,
-      "step": 40757
-    },
-    {
-      "epoch": 6.563790812834656,
-      "grad_norm": 0.032018031924963,
-      "learning_rate": 0.0001999787518985789,
-      "loss": 46.0,
-      "step": 40758
-    },
-    {
-      "epoch": 6.563951849913442,
-      "grad_norm": 0.006671628449112177,
-      "learning_rate": 0.00019997875085570025,
-      "loss": 46.0,
-      "step": 40759
-    },
-    {
-      "epoch": 6.56411288699223,
-      "grad_norm": 0.016330774873495102,
-      "learning_rate": 0.000199978749812796,
-      "loss": 46.0,
-      "step": 40760
-    },
-    {
-      "epoch": 6.564273924071017,
-      "grad_norm": 0.0022147325798869133,
-      "learning_rate": 0.00019997874876986617,
-      "loss": 46.0,
-      "step": 40761
-    },
-    {
-      "epoch": 6.5644349611498045,
-      "grad_norm": 0.004885328933596611,
-      "learning_rate": 0.00019997874772691073,
-      "loss": 46.0,
-      "step": 40762
-    },
-    {
-      "epoch": 6.564595998228592,
-      "grad_norm": 0.003634854219853878,
-      "learning_rate": 0.00019997874668392976,
-      "loss": 46.0,
-      "step": 40763
-    },
-    {
-      "epoch": 6.564757035307379,
-      "grad_norm": 0.004276583902537823,
-      "learning_rate": 0.00019997874564092313,
-      "loss": 46.0,
-      "step": 40764
-    },
-    {
-      "epoch": 6.564918072386167,
-      "grad_norm": 0.010841818526387215,
-      "learning_rate": 0.00019997874459789096,
-      "loss": 46.0,
-      "step": 40765
-    },
-    {
-      "epoch": 6.565079109464954,
-      "grad_norm": 0.004987781867384911,
-      "learning_rate": 0.0001999787435548332,
-      "loss": 46.0,
-      "step": 40766
-    },
-    {
-      "epoch": 6.565240146543742,
-      "grad_norm": 0.007482773624360561,
-      "learning_rate": 0.00019997874251174983,
-      "loss": 46.0,
-      "step": 40767
-    },
-    {
-      "epoch": 6.565401183622529,
-      "grad_norm": 0.0031461124308407307,
-      "learning_rate": 0.0001999787414686409,
-      "loss": 46.0,
-      "step": 40768
-    },
-    {
-      "epoch": 6.565562220701317,
-      "grad_norm": 0.006653795950114727,
-      "learning_rate": 0.00019997874042550632,
-      "loss": 46.0,
-      "step": 40769
-    },
-    {
-      "epoch": 6.565723257780104,
-      "grad_norm": 0.02060164138674736,
-      "learning_rate": 0.0001999787393823462,
-      "loss": 46.0,
-      "step": 40770
-    },
-    {
-      "epoch": 6.565884294858892,
-      "grad_norm": 0.0013250625925138593,
-      "learning_rate": 0.00019997873833916047,
-      "loss": 46.0,
-      "step": 40771
-    },
-    {
-      "epoch": 6.566045331937679,
-      "grad_norm": 0.0012524082558229566,
-      "learning_rate": 0.00019997873729594916,
-      "loss": 46.0,
-      "step": 40772
-    },
-    {
-      "epoch": 6.5662063690164665,
-      "grad_norm": 0.01475952286273241,
-      "learning_rate": 0.00019997873625271227,
-      "loss": 46.0,
-      "step": 40773
-    },
-    {
-      "epoch": 6.566367406095253,
-      "grad_norm": 0.0017552569042891264,
-      "learning_rate": 0.00019997873520944976,
-      "loss": 46.0,
-      "step": 40774
-    },
-    {
-      "epoch": 6.5665284431740405,
-      "grad_norm": 0.002845402341336012,
-      "learning_rate": 0.0001999787341661617,
-      "loss": 46.0,
-      "step": 40775
-    },
-    {
-      "epoch": 6.566689480252828,
-      "grad_norm": 0.010622103698551655,
-      "learning_rate": 0.00019997873312284803,
-      "loss": 46.0,
-      "step": 40776
-    },
-    {
-      "epoch": 6.566850517331615,
-      "grad_norm": 0.0050599114038050175,
-      "learning_rate": 0.00019997873207950874,
-      "loss": 46.0,
-      "step": 40777
-    },
-    {
-      "epoch": 6.567011554410403,
-      "grad_norm": 0.00894957035779953,
-      "learning_rate": 0.00019997873103614393,
-      "loss": 46.0,
-      "step": 40778
-    },
-    {
-      "epoch": 6.56717259148919,
-      "grad_norm": 0.009361461736261845,
-      "learning_rate": 0.0001999787299927535,
-      "loss": 46.0,
-      "step": 40779
-    },
-    {
-      "epoch": 6.567333628567978,
-      "grad_norm": 0.003837777068838477,
-      "learning_rate": 0.00019997872894933746,
-      "loss": 46.0,
-      "step": 40780
-    },
-    {
-      "epoch": 6.567494665646765,
-      "grad_norm": 0.0020335065200924873,
-      "learning_rate": 0.0001999787279058958,
-      "loss": 46.0,
-      "step": 40781
-    },
-    {
-      "epoch": 6.567655702725553,
-      "grad_norm": 0.010169140063226223,
-      "learning_rate": 0.00019997872686242863,
-      "loss": 46.0,
-      "step": 40782
-    },
-    {
-      "epoch": 6.56781673980434,
-      "grad_norm": 0.004751738626509905,
-      "learning_rate": 0.00019997872581893584,
-      "loss": 46.0,
-      "step": 40783
-    },
-    {
-      "epoch": 6.567977776883128,
-      "grad_norm": 0.003848387161269784,
-      "learning_rate": 0.00019997872477541746,
-      "loss": 46.0,
-      "step": 40784
-    },
-    {
-      "epoch": 6.568138813961915,
-      "grad_norm": 0.001586927566677332,
-      "learning_rate": 0.00019997872373187346,
-      "loss": 46.0,
-      "step": 40785
-    },
-    {
-      "epoch": 6.5682998510407025,
-      "grad_norm": 0.003316056914627552,
-      "learning_rate": 0.0001999787226883039,
-      "loss": 46.0,
-      "step": 40786
-    },
-    {
-      "epoch": 6.568460888119489,
-      "grad_norm": 0.019763285294175148,
-      "learning_rate": 0.00019997872164470874,
-      "loss": 46.0,
-      "step": 40787
-    },
-    {
-      "epoch": 6.568621925198277,
-      "grad_norm": 0.01035550981760025,
-      "learning_rate": 0.000199978720601088,
-      "loss": 46.0,
-      "step": 40788
-    },
-    {
-      "epoch": 6.568782962277064,
-      "grad_norm": 0.0027674254961311817,
-      "learning_rate": 0.00019997871955744167,
-      "loss": 46.0,
-      "step": 40789
-    },
-    {
-      "epoch": 6.568943999355851,
-      "grad_norm": 0.010693095624446869,
-      "learning_rate": 0.00019997871851376976,
-      "loss": 46.0,
-      "step": 40790
-    },
-    {
-      "epoch": 6.569105036434639,
-      "grad_norm": 0.012854852713644505,
-      "learning_rate": 0.00019997871747007225,
-      "loss": 46.0,
-      "step": 40791
-    },
-    {
-      "epoch": 6.569266073513426,
-      "grad_norm": 0.0011087863240391016,
-      "learning_rate": 0.00019997871642634914,
-      "loss": 46.0,
-      "step": 40792
-    },
-    {
-      "epoch": 6.569427110592214,
-      "grad_norm": 0.0018523853505030274,
-      "learning_rate": 0.00019997871538260045,
-      "loss": 46.0,
-      "step": 40793
-    },
-    {
-      "epoch": 6.569588147671001,
-      "grad_norm": 0.002040497027337551,
-      "learning_rate": 0.00019997871433882617,
-      "loss": 46.0,
-      "step": 40794
-    },
-    {
-      "epoch": 6.569749184749789,
-      "grad_norm": 0.0021707494743168354,
-      "learning_rate": 0.0001999787132950263,
-      "loss": 46.0,
-      "step": 40795
-    },
-    {
-      "epoch": 6.569910221828576,
-      "grad_norm": 0.0050607346929609776,
-      "learning_rate": 0.00019997871225120085,
-      "loss": 46.0,
-      "step": 40796
-    },
-    {
-      "epoch": 6.5700712589073635,
-      "grad_norm": 0.003341927193105221,
-      "learning_rate": 0.0001999787112073498,
-      "loss": 46.0,
-      "step": 40797
-    },
-    {
-      "epoch": 6.570232295986151,
-      "grad_norm": 0.007981318980455399,
-      "learning_rate": 0.00019997871016347318,
-      "loss": 46.0,
-      "step": 40798
-    },
-    {
-      "epoch": 6.5703933330649384,
-      "grad_norm": 0.008813790045678616,
-      "learning_rate": 0.00019997870911957094,
-      "loss": 46.0,
-      "step": 40799
-    },
-    {
-      "epoch": 6.570554370143726,
-      "grad_norm": 0.013091310858726501,
-      "learning_rate": 0.00019997870807564314,
-      "loss": 46.0,
-      "step": 40800
-    },
-    {
-      "epoch": 6.570715407222513,
-      "grad_norm": 0.0035188747569918633,
-      "learning_rate": 0.00019997870703168972,
-      "loss": 46.0,
-      "step": 40801
-    },
-    {
-      "epoch": 6.5708764443013,
-      "grad_norm": 0.00641377829015255,
-      "learning_rate": 0.00019997870598771074,
-      "loss": 46.0,
-      "step": 40802
-    },
-    {
-      "epoch": 6.571037481380088,
-      "grad_norm": 0.004212986212223768,
-      "learning_rate": 0.00019997870494370615,
-      "loss": 46.0,
-      "step": 40803
-    },
-    {
-      "epoch": 6.571198518458875,
-      "grad_norm": 0.007577687036246061,
-      "learning_rate": 0.00019997870389967598,
-      "loss": 46.0,
-      "step": 40804
-    },
-    {
-      "epoch": 6.571359555537662,
-      "grad_norm": 0.004961410071700811,
-      "learning_rate": 0.00019997870285562018,
-      "loss": 46.0,
-      "step": 40805
-    },
-    {
-      "epoch": 6.57152059261645,
-      "grad_norm": 0.002276793820783496,
-      "learning_rate": 0.00019997870181153883,
-      "loss": 46.0,
-      "step": 40806
-    },
-    {
-      "epoch": 6.571681629695237,
-      "grad_norm": 0.01013907603919506,
-      "learning_rate": 0.0001999787007674319,
-      "loss": 46.0,
-      "step": 40807
-    },
-    {
-      "epoch": 6.571842666774025,
-      "grad_norm": 0.008978056721389294,
-      "learning_rate": 0.0001999786997232994,
-      "loss": 46.0,
-      "step": 40808
-    },
-    {
-      "epoch": 6.572003703852812,
-      "grad_norm": 0.005078437272459269,
-      "learning_rate": 0.00019997869867914125,
-      "loss": 46.0,
-      "step": 40809
-    },
-    {
-      "epoch": 6.5721647409315995,
-      "grad_norm": 0.0006928339716978371,
-      "learning_rate": 0.00019997869763495752,
-      "loss": 46.0,
-      "step": 40810
-    },
-    {
-      "epoch": 6.572325778010387,
-      "grad_norm": 0.001591209787875414,
-      "learning_rate": 0.00019997869659074823,
-      "loss": 46.0,
-      "step": 40811
-    },
-    {
-      "epoch": 6.572486815089174,
-      "grad_norm": 0.006118203978985548,
-      "learning_rate": 0.00019997869554651336,
-      "loss": 46.0,
-      "step": 40812
-    },
-    {
-      "epoch": 6.572647852167962,
-      "grad_norm": 0.004776508081704378,
-      "learning_rate": 0.0001999786945022529,
-      "loss": 46.0,
-      "step": 40813
-    },
-    {
-      "epoch": 6.572808889246749,
-      "grad_norm": 0.00453460868448019,
-      "learning_rate": 0.00019997869345796682,
-      "loss": 46.0,
-      "step": 40814
-    },
-    {
-      "epoch": 6.572969926325537,
-      "grad_norm": 0.004105367232114077,
-      "learning_rate": 0.00019997869241365516,
-      "loss": 46.0,
-      "step": 40815
-    },
-    {
-      "epoch": 6.573130963404324,
-      "grad_norm": 0.004433273337781429,
-      "learning_rate": 0.00019997869136931793,
-      "loss": 46.0,
-      "step": 40816
-    },
-    {
-      "epoch": 6.573292000483111,
-      "grad_norm": 0.0016228772001340985,
-      "learning_rate": 0.00019997869032495507,
-      "loss": 46.0,
-      "step": 40817
-    },
-    {
-      "epoch": 6.573453037561898,
-      "grad_norm": 0.009247311390936375,
-      "learning_rate": 0.00019997868928056667,
-      "loss": 46.0,
-      "step": 40818
-    },
-    {
-      "epoch": 6.573614074640686,
-      "grad_norm": 0.028416039422154427,
-      "learning_rate": 0.00019997868823615265,
-      "loss": 46.0,
-      "step": 40819
-    },
-    {
-      "epoch": 6.573775111719473,
-      "grad_norm": 0.005283160600811243,
-      "learning_rate": 0.00019997868719171305,
-      "loss": 46.0,
-      "step": 40820
-    },
-    {
-      "epoch": 6.573936148798261,
-      "grad_norm": 0.010821589268743992,
-      "learning_rate": 0.00019997868614724784,
-      "loss": 46.0,
-      "step": 40821
-    },
-    {
-      "epoch": 6.574097185877048,
-      "grad_norm": 0.0017294847639277577,
-      "learning_rate": 0.00019997868510275707,
-      "loss": 46.0,
-      "step": 40822
-    },
-    {
-      "epoch": 6.5742582229558355,
-      "grad_norm": 0.005353229120373726,
-      "learning_rate": 0.00019997868405824068,
-      "loss": 46.0,
-      "step": 40823
-    },
-    {
-      "epoch": 6.574419260034623,
-      "grad_norm": 0.0020743291825056076,
-      "learning_rate": 0.0001999786830136987,
-      "loss": 46.0,
-      "step": 40824
-    },
-    {
-      "epoch": 6.57458029711341,
-      "grad_norm": 0.004140579141676426,
-      "learning_rate": 0.00019997868196913117,
-      "loss": 46.0,
-      "step": 40825
-    },
-    {
-      "epoch": 6.574741334192198,
-      "grad_norm": 0.012167209759354591,
-      "learning_rate": 0.00019997868092453804,
-      "loss": 46.0,
-      "step": 40826
-    },
-    {
-      "epoch": 6.574902371270985,
-      "grad_norm": 0.0013769238721579313,
-      "learning_rate": 0.0001999786798799193,
-      "loss": 46.0,
-      "step": 40827
-    },
-    {
-      "epoch": 6.575063408349773,
-      "grad_norm": 0.007982422597706318,
-      "learning_rate": 0.00019997867883527498,
-      "loss": 46.0,
-      "step": 40828
-    },
-    {
-      "epoch": 6.57522444542856,
-      "grad_norm": 0.010315612889826298,
-      "learning_rate": 0.00019997867779060507,
-      "loss": 46.0,
-      "step": 40829
-    },
-    {
-      "epoch": 6.575385482507348,
-      "grad_norm": 0.001964476192370057,
-      "learning_rate": 0.00019997867674590957,
-      "loss": 46.0,
-      "step": 40830
-    },
-    {
-      "epoch": 6.575546519586135,
-      "grad_norm": 0.007544730789959431,
-      "learning_rate": 0.00019997867570118849,
-      "loss": 46.0,
-      "step": 40831
-    },
-    {
-      "epoch": 6.575707556664922,
-      "grad_norm": 0.0013011571718379855,
-      "learning_rate": 0.0001999786746564418,
-      "loss": 46.0,
-      "step": 40832
-    },
-    {
-      "epoch": 6.575868593743709,
-      "grad_norm": 0.00497397780418396,
-      "learning_rate": 0.00019997867361166955,
-      "loss": 46.0,
-      "step": 40833
-    },
-    {
-      "epoch": 6.576029630822497,
-      "grad_norm": 0.003266011830419302,
-      "learning_rate": 0.0001999786725668717,
-      "loss": 46.0,
-      "step": 40834
-    },
-    {
-      "epoch": 6.576190667901284,
-      "grad_norm": 0.025336232036352158,
-      "learning_rate": 0.00019997867152204824,
-      "loss": 46.0,
-      "step": 40835
-    },
-    {
-      "epoch": 6.5763517049800715,
-      "grad_norm": 0.0018570650136098266,
-      "learning_rate": 0.00019997867047719922,
-      "loss": 46.0,
-      "step": 40836
-    },
-    {
-      "epoch": 6.576512742058859,
-      "grad_norm": 0.00737730972468853,
-      "learning_rate": 0.00019997866943232458,
-      "loss": 46.0,
-      "step": 40837
-    },
-    {
-      "epoch": 6.576673779137646,
-      "grad_norm": 0.0024157322477549314,
-      "learning_rate": 0.0001999786683874244,
-      "loss": 46.0,
-      "step": 40838
-    },
-    {
-      "epoch": 6.576834816216434,
-      "grad_norm": 0.00592298386618495,
-      "learning_rate": 0.00019997866734249858,
-      "loss": 46.0,
-      "step": 40839
-    },
-    {
-      "epoch": 6.576995853295221,
-      "grad_norm": 0.001229961053468287,
-      "learning_rate": 0.00019997866629754718,
-      "loss": 46.0,
-      "step": 40840
-    },
-    {
-      "epoch": 6.577156890374009,
-      "grad_norm": 0.006123790983110666,
-      "learning_rate": 0.0001999786652525702,
-      "loss": 46.0,
-      "step": 40841
-    },
-    {
-      "epoch": 6.577317927452796,
-      "grad_norm": 0.007423703093081713,
-      "learning_rate": 0.00019997866420756762,
-      "loss": 46.0,
-      "step": 40842
-    },
-    {
-      "epoch": 6.577478964531584,
-      "grad_norm": 0.01447257585823536,
-      "learning_rate": 0.00019997866316253947,
-      "loss": 46.0,
-      "step": 40843
-    },
-    {
-      "epoch": 6.577640001610371,
-      "grad_norm": 0.0025015114806592464,
-      "learning_rate": 0.00019997866211748572,
-      "loss": 46.0,
-      "step": 40844
-    },
-    {
-      "epoch": 6.5778010386891586,
-      "grad_norm": 0.007844620384275913,
-      "learning_rate": 0.00019997866107240636,
-      "loss": 46.0,
-      "step": 40845
-    },
-    {
-      "epoch": 6.577962075767946,
-      "grad_norm": 0.007799843791872263,
-      "learning_rate": 0.00019997866002730144,
-      "loss": 46.0,
-      "step": 40846
-    },
-    {
-      "epoch": 6.578123112846733,
-      "grad_norm": 0.003350872313603759,
-      "learning_rate": 0.00019997865898217093,
-      "loss": 46.0,
-      "step": 40847
-    },
-    {
-      "epoch": 6.57828414992552,
-      "grad_norm": 0.0009966054931282997,
-      "learning_rate": 0.00019997865793701484,
-      "loss": 46.0,
-      "step": 40848
-    },
-    {
-      "epoch": 6.5784451870043075,
-      "grad_norm": 0.00975534226745367,
-      "learning_rate": 0.00019997865689183313,
-      "loss": 46.0,
-      "step": 40849
-    },
-    {
-      "epoch": 6.578606224083095,
-      "grad_norm": 0.0023474418558180332,
-      "learning_rate": 0.00019997865584662586,
-      "loss": 46.0,
-      "step": 40850
-    },
-    {
-      "epoch": 6.578767261161882,
-      "grad_norm": 0.004771187901496887,
-      "learning_rate": 0.00019997865480139297,
-      "loss": 46.0,
-      "step": 40851
-    },
-    {
-      "epoch": 6.57892829824067,
-      "grad_norm": 0.0008967381436377764,
-      "learning_rate": 0.0001999786537561345,
-      "loss": 46.0,
-      "step": 40852
-    },
-    {
-      "epoch": 6.579089335319457,
-      "grad_norm": 0.011502484790980816,
-      "learning_rate": 0.00019997865271085044,
-      "loss": 46.0,
-      "step": 40853
-    },
-    {
-      "epoch": 6.579250372398245,
-      "grad_norm": 0.0028328821063041687,
-      "learning_rate": 0.00019997865166554083,
-      "loss": 46.0,
-      "step": 40854
-    },
-    {
-      "epoch": 6.579411409477032,
-      "grad_norm": 0.019505729898810387,
-      "learning_rate": 0.00019997865062020557,
-      "loss": 46.0,
-      "step": 40855
-    },
-    {
-      "epoch": 6.57957244655582,
-      "grad_norm": 0.005499968770891428,
-      "learning_rate": 0.00019997864957484475,
-      "loss": 46.0,
-      "step": 40856
-    },
-    {
-      "epoch": 6.579733483634607,
-      "grad_norm": 0.008117297664284706,
-      "learning_rate": 0.00019997864852945834,
-      "loss": 46.0,
-      "step": 40857
-    },
-    {
-      "epoch": 6.5798945207133945,
-      "grad_norm": 0.0017934462521225214,
-      "learning_rate": 0.00019997864748404632,
-      "loss": 46.0,
-      "step": 40858
-    },
-    {
-      "epoch": 6.580055557792182,
-      "grad_norm": 0.0010216551600024104,
-      "learning_rate": 0.00019997864643860874,
-      "loss": 46.0,
-      "step": 40859
-    },
-    {
-      "epoch": 6.5802165948709685,
-      "grad_norm": 0.008551942184567451,
-      "learning_rate": 0.00019997864539314554,
-      "loss": 46.0,
-      "step": 40860
-    },
-    {
-      "epoch": 6.580377631949757,
-      "grad_norm": 0.003321589669212699,
-      "learning_rate": 0.0001999786443476568,
-      "loss": 46.0,
-      "step": 40861
-    },
-    {
-      "epoch": 6.5805386690285435,
-      "grad_norm": 0.0028386928606778383,
-      "learning_rate": 0.00019997864330214245,
-      "loss": 46.0,
-      "step": 40862
-    },
-    {
-      "epoch": 6.580699706107331,
-      "grad_norm": 0.007586246356368065,
-      "learning_rate": 0.0001999786422566025,
-      "loss": 46.0,
-      "step": 40863
-    },
-    {
-      "epoch": 6.580860743186118,
-      "grad_norm": 0.0038805403746664524,
-      "learning_rate": 0.00019997864121103692,
-      "loss": 46.0,
-      "step": 40864
-    },
-    {
-      "epoch": 6.581021780264906,
-      "grad_norm": 0.002877336461097002,
-      "learning_rate": 0.00019997864016544582,
-      "loss": 46.0,
-      "step": 40865
-    },
-    {
-      "epoch": 6.581182817343693,
-      "grad_norm": 0.01757013238966465,
-      "learning_rate": 0.0001999786391198291,
-      "loss": 46.0,
-      "step": 40866
-    },
-    {
-      "epoch": 6.581343854422481,
-      "grad_norm": 0.0022511135321110487,
-      "learning_rate": 0.0001999786380741868,
-      "loss": 46.0,
-      "step": 40867
-    },
-    {
-      "epoch": 6.581504891501268,
-      "grad_norm": 0.0058492147363722324,
-      "learning_rate": 0.0001999786370285189,
-      "loss": 46.0,
-      "step": 40868
-    },
-    {
-      "epoch": 6.581665928580056,
-      "grad_norm": 0.009627859108150005,
-      "learning_rate": 0.00019997863598282542,
-      "loss": 46.0,
-      "step": 40869
-    },
-    {
-      "epoch": 6.581826965658843,
-      "grad_norm": 0.0015005981549620628,
-      "learning_rate": 0.00019997863493710633,
-      "loss": 46.0,
-      "step": 40870
-    },
-    {
-      "epoch": 6.5819880027376305,
-      "grad_norm": 0.0010363651672378182,
-      "learning_rate": 0.00019997863389136167,
-      "loss": 46.0,
-      "step": 40871
-    },
-    {
-      "epoch": 6.582149039816418,
-      "grad_norm": 0.004977967590093613,
-      "learning_rate": 0.00019997863284559143,
-      "loss": 46.0,
-      "step": 40872
-    },
-    {
-      "epoch": 6.582310076895205,
-      "grad_norm": 0.0015544480411335826,
-      "learning_rate": 0.0001999786317997956,
-      "loss": 46.0,
-      "step": 40873
-    },
-    {
-      "epoch": 6.582471113973993,
-      "grad_norm": 0.01028465386480093,
-      "learning_rate": 0.00019997863075397416,
-      "loss": 46.0,
-      "step": 40874
-    },
-    {
-      "epoch": 6.582632151052779,
-      "grad_norm": 0.0031032762490212917,
-      "learning_rate": 0.00019997862970812713,
-      "loss": 46.0,
-      "step": 40875
-    },
-    {
-      "epoch": 6.582793188131568,
-      "grad_norm": 0.0034985090605914593,
-      "learning_rate": 0.0001999786286622545,
-      "loss": 46.0,
-      "step": 40876
-    },
-    {
-      "epoch": 6.582954225210354,
-      "grad_norm": 0.004831480793654919,
-      "learning_rate": 0.0001999786276163563,
-      "loss": 46.0,
-      "step": 40877
-    },
-    {
-      "epoch": 6.583115262289142,
-      "grad_norm": 0.016321226954460144,
-      "learning_rate": 0.00019997862657043252,
-      "loss": 46.0,
-      "step": 40878
-    },
-    {
-      "epoch": 6.583276299367929,
-      "grad_norm": 0.0049707540310919285,
-      "learning_rate": 0.00019997862552448314,
-      "loss": 46.0,
-      "step": 40879
-    },
-    {
-      "epoch": 6.583437336446717,
-      "grad_norm": 0.013099361211061478,
-      "learning_rate": 0.0001999786244785082,
-      "loss": 46.0,
-      "step": 40880
-    },
-    {
-      "epoch": 6.583598373525504,
-      "grad_norm": 0.0033694191370159388,
-      "learning_rate": 0.00019997862343250762,
-      "loss": 46.0,
-      "step": 40881
-    },
-    {
-      "epoch": 6.583759410604292,
-      "grad_norm": 0.0035773268900811672,
-      "learning_rate": 0.00019997862238648145,
-      "loss": 46.0,
-      "step": 40882
-    },
-    {
-      "epoch": 6.583920447683079,
-      "grad_norm": 0.002976084128022194,
-      "learning_rate": 0.00019997862134042973,
-      "loss": 46.0,
-      "step": 40883
-    },
-    {
-      "epoch": 6.5840814847618665,
-      "grad_norm": 0.017070339992642403,
-      "learning_rate": 0.0001999786202943524,
-      "loss": 46.0,
-      "step": 40884
-    },
-    {
-      "epoch": 6.584242521840654,
-      "grad_norm": 0.0022377576678991318,
-      "learning_rate": 0.0001999786192482495,
-      "loss": 46.0,
-      "step": 40885
-    },
-    {
-      "epoch": 6.584403558919441,
-      "grad_norm": 0.005276334472000599,
-      "learning_rate": 0.000199978618202121,
-      "loss": 46.0,
-      "step": 40886
-    },
-    {
-      "epoch": 6.584564595998229,
-      "grad_norm": 0.0013971786247566342,
-      "learning_rate": 0.00019997861715596692,
-      "loss": 46.0,
-      "step": 40887
-    },
-    {
-      "epoch": 6.584725633077016,
-      "grad_norm": 0.0033143588807433844,
-      "learning_rate": 0.0001999786161097872,
-      "loss": 46.0,
-      "step": 40888
-    },
-    {
-      "epoch": 6.584886670155804,
-      "grad_norm": 0.001610062550753355,
-      "learning_rate": 0.00019997861506358192,
-      "loss": 46.0,
-      "step": 40889
-    },
-    {
-      "epoch": 6.58504770723459,
-      "grad_norm": 0.013162648305296898,
-      "learning_rate": 0.00019997861401735109,
-      "loss": 46.0,
-      "step": 40890
-    },
-    {
-      "epoch": 6.585208744313378,
-      "grad_norm": 0.00695267878472805,
-      "learning_rate": 0.00019997861297109463,
-      "loss": 46.0,
-      "step": 40891
-    },
-    {
-      "epoch": 6.585369781392165,
-      "grad_norm": 0.004841205198317766,
-      "learning_rate": 0.0001999786119248126,
-      "loss": 46.0,
-      "step": 40892
-    },
-    {
-      "epoch": 6.585530818470953,
-      "grad_norm": 0.003625884186476469,
-      "learning_rate": 0.00019997861087850497,
-      "loss": 46.0,
-      "step": 40893
-    },
-    {
-      "epoch": 6.58569185554974,
-      "grad_norm": 0.011560887098312378,
-      "learning_rate": 0.00019997860983217173,
-      "loss": 46.0,
-      "step": 40894
-    },
-    {
-      "epoch": 6.585852892628528,
-      "grad_norm": 0.0012081846361979842,
-      "learning_rate": 0.00019997860878581293,
-      "loss": 46.0,
-      "step": 40895
-    },
-    {
-      "epoch": 6.586013929707315,
-      "grad_norm": 0.013927397318184376,
-      "learning_rate": 0.0001999786077394285,
-      "loss": 46.0,
-      "step": 40896
-    },
-    {
-      "epoch": 6.5861749667861025,
-      "grad_norm": 0.0018666948890313506,
-      "learning_rate": 0.00019997860669301854,
-      "loss": 46.0,
-      "step": 40897
-    },
-    {
-      "epoch": 6.58633600386489,
-      "grad_norm": 0.003324862103909254,
-      "learning_rate": 0.00019997860564658298,
-      "loss": 46.0,
-      "step": 40898
-    },
-    {
-      "epoch": 6.586497040943677,
-      "grad_norm": 0.018707364797592163,
-      "learning_rate": 0.0001999786046001218,
-      "loss": 46.0,
-      "step": 40899
-    },
-    {
-      "epoch": 6.586658078022465,
-      "grad_norm": 0.0038193203508853912,
-      "learning_rate": 0.00019997860355363504,
-      "loss": 46.0,
-      "step": 40900
-    },
-    {
-      "epoch": 6.586819115101252,
-      "grad_norm": 0.002704898826777935,
-      "learning_rate": 0.0001999786025071227,
-      "loss": 46.0,
-      "step": 40901
-    },
-    {
-      "epoch": 6.58698015218004,
-      "grad_norm": 0.0006604773225262761,
-      "learning_rate": 0.00019997860146058475,
-      "loss": 46.0,
-      "step": 40902
-    },
-    {
-      "epoch": 6.587141189258827,
-      "grad_norm": 0.008635904639959335,
-      "learning_rate": 0.00019997860041402122,
-      "loss": 46.0,
-      "step": 40903
-    },
-    {
-      "epoch": 6.587302226337615,
-      "grad_norm": 0.001925706397742033,
-      "learning_rate": 0.00019997859936743214,
-      "loss": 46.0,
-      "step": 40904
-    },
-    {
-      "epoch": 6.587463263416401,
-      "grad_norm": 0.005759078543633223,
-      "learning_rate": 0.0001999785983208174,
-      "loss": 46.0,
-      "step": 40905
-    },
-    {
-      "epoch": 6.587624300495189,
-      "grad_norm": 0.006487691309303045,
-      "learning_rate": 0.00019997859727417713,
-      "loss": 46.0,
-      "step": 40906
-    },
-    {
-      "epoch": 6.587785337573976,
-      "grad_norm": 0.006073189899325371,
-      "learning_rate": 0.00019997859622751123,
-      "loss": 46.0,
-      "step": 40907
-    },
-    {
-      "epoch": 6.587946374652764,
-      "grad_norm": 0.006661845836788416,
-      "learning_rate": 0.0001999785951808198,
-      "loss": 46.0,
-      "step": 40908
-    },
-    {
-      "epoch": 6.588107411731551,
-      "grad_norm": 0.01026956457644701,
-      "learning_rate": 0.00019997859413410272,
-      "loss": 46.0,
-      "step": 40909
-    },
-    {
-      "epoch": 6.5882684488103385,
-      "grad_norm": 0.007520552258938551,
-      "learning_rate": 0.00019997859308736006,
-      "loss": 46.0,
-      "step": 40910
-    },
-    {
-      "epoch": 6.588429485889126,
-      "grad_norm": 0.007870912551879883,
-      "learning_rate": 0.0001999785920405918,
-      "loss": 46.0,
-      "step": 40911
-    },
-    {
-      "epoch": 6.588590522967913,
-      "grad_norm": 0.019782068207859993,
-      "learning_rate": 0.000199978590993798,
-      "loss": 46.0,
-      "step": 40912
-    },
-    {
-      "epoch": 6.588751560046701,
-      "grad_norm": 0.002928605070337653,
-      "learning_rate": 0.00019997858994697857,
-      "loss": 46.0,
-      "step": 40913
-    },
-    {
-      "epoch": 6.588912597125488,
-      "grad_norm": 0.006589285098016262,
-      "learning_rate": 0.00019997858890013356,
-      "loss": 46.0,
-      "step": 40914
-    },
-    {
-      "epoch": 6.589073634204276,
-      "grad_norm": 0.013120392337441444,
-      "learning_rate": 0.00019997858785326296,
-      "loss": 46.0,
-      "step": 40915
-    },
-    {
-      "epoch": 6.589234671283063,
-      "grad_norm": 0.0015728623839095235,
-      "learning_rate": 0.00019997858680636678,
-      "loss": 46.0,
-      "step": 40916
-    },
-    {
-      "epoch": 6.589395708361851,
-      "grad_norm": 0.0145568223670125,
-      "learning_rate": 0.00019997858575944498,
-      "loss": 46.0,
-      "step": 40917
-    },
-    {
-      "epoch": 6.589556745440638,
-      "grad_norm": 0.005859727505594492,
-      "learning_rate": 0.00019997858471249765,
-      "loss": 46.0,
-      "step": 40918
-    },
-    {
-      "epoch": 6.5897177825194255,
-      "grad_norm": 0.0022197493817657232,
-      "learning_rate": 0.0001999785836655247,
-      "loss": 46.0,
-      "step": 40919
-    },
-    {
-      "epoch": 6.589878819598212,
-      "grad_norm": 0.004570066463202238,
-      "learning_rate": 0.00019997858261852614,
-      "loss": 46.0,
-      "step": 40920
-    },
-    {
-      "epoch": 6.5900398566769995,
-      "grad_norm": 0.0077544027008116245,
-      "learning_rate": 0.00019997858157150202,
-      "loss": 46.0,
-      "step": 40921
-    },
-    {
-      "epoch": 6.590200893755787,
-      "grad_norm": 0.005387146957218647,
-      "learning_rate": 0.00019997858052445228,
-      "loss": 46.0,
-      "step": 40922
-    },
-    {
-      "epoch": 6.5903619308345744,
-      "grad_norm": 0.0050185625441372395,
-      "learning_rate": 0.00019997857947737696,
-      "loss": 46.0,
-      "step": 40923
-    },
-    {
-      "epoch": 6.590522967913362,
-      "grad_norm": 0.010602330788969994,
-      "learning_rate": 0.00019997857843027608,
-      "loss": 46.0,
-      "step": 40924
-    },
-    {
-      "epoch": 6.590684004992149,
-      "grad_norm": 0.0017769129481166601,
-      "learning_rate": 0.00019997857738314958,
-      "loss": 46.0,
-      "step": 40925
-    },
-    {
-      "epoch": 6.590845042070937,
-      "grad_norm": 0.0017785043455660343,
-      "learning_rate": 0.0001999785763359975,
-      "loss": 46.0,
-      "step": 40926
-    },
-    {
-      "epoch": 6.591006079149724,
-      "grad_norm": 0.002208475489169359,
-      "learning_rate": 0.00019997857528881983,
-      "loss": 46.0,
-      "step": 40927
-    },
-    {
-      "epoch": 6.591167116228512,
-      "grad_norm": 0.0021965792402625084,
-      "learning_rate": 0.00019997857424161657,
-      "loss": 46.0,
-      "step": 40928
-    },
-    {
-      "epoch": 6.591328153307299,
-      "grad_norm": 0.006657461170107126,
-      "learning_rate": 0.00019997857319438772,
-      "loss": 46.0,
-      "step": 40929
-    },
-    {
-      "epoch": 6.591489190386087,
-      "grad_norm": 0.004748990293592215,
-      "learning_rate": 0.00019997857214713326,
-      "loss": 46.0,
-      "step": 40930
-    },
-    {
-      "epoch": 6.591650227464874,
-      "grad_norm": 0.005602410063147545,
-      "learning_rate": 0.00019997857109985327,
-      "loss": 46.0,
-      "step": 40931
-    },
-    {
-      "epoch": 6.5918112645436615,
-      "grad_norm": 0.010183610953390598,
-      "learning_rate": 0.00019997857005254766,
-      "loss": 46.0,
-      "step": 40932
-    },
-    {
-      "epoch": 6.591972301622448,
-      "grad_norm": 0.005670685321092606,
-      "learning_rate": 0.00019997856900521644,
-      "loss": 46.0,
-      "step": 40933
-    },
-    {
-      "epoch": 6.592133338701236,
-      "grad_norm": 0.0024473739322274923,
-      "learning_rate": 0.00019997856795785963,
-      "loss": 46.0,
-      "step": 40934
-    },
-    {
-      "epoch": 6.592294375780023,
-      "grad_norm": 0.014006580226123333,
-      "learning_rate": 0.00019997856691047724,
-      "loss": 46.0,
-      "step": 40935
-    },
-    {
-      "epoch": 6.59245541285881,
-      "grad_norm": 0.0041218772530555725,
-      "learning_rate": 0.00019997856586306928,
-      "loss": 46.0,
-      "step": 40936
-    },
-    {
-      "epoch": 6.592616449937598,
-      "grad_norm": 0.0031477452721446753,
-      "learning_rate": 0.00019997856481563574,
-      "loss": 46.0,
-      "step": 40937
-    },
-    {
-      "epoch": 6.592777487016385,
-      "grad_norm": 0.026103513315320015,
-      "learning_rate": 0.00019997856376817655,
-      "loss": 46.0,
-      "step": 40938
-    },
-    {
-      "epoch": 6.592938524095173,
-      "grad_norm": 0.00866468995809555,
-      "learning_rate": 0.0001999785627206918,
-      "loss": 46.0,
-      "step": 40939
-    },
-    {
-      "epoch": 6.59309956117396,
-      "grad_norm": 0.0032765506766736507,
-      "learning_rate": 0.00019997856167318148,
-      "loss": 46.0,
-      "step": 40940
-    },
-    {
-      "epoch": 6.593260598252748,
-      "grad_norm": 0.0032905288971960545,
-      "learning_rate": 0.00019997856062564556,
-      "loss": 46.0,
-      "step": 40941
-    },
-    {
-      "epoch": 6.593421635331535,
-      "grad_norm": 0.003337105503305793,
-      "learning_rate": 0.00019997855957808405,
-      "loss": 46.0,
-      "step": 40942
-    },
-    {
-      "epoch": 6.593582672410323,
-      "grad_norm": 0.006013059988617897,
-      "learning_rate": 0.00019997855853049696,
-      "loss": 46.0,
-      "step": 40943
-    },
-    {
-      "epoch": 6.59374370948911,
-      "grad_norm": 0.007998323999345303,
-      "learning_rate": 0.00019997855748288425,
-      "loss": 46.0,
-      "step": 40944
-    },
-    {
-      "epoch": 6.5939047465678975,
-      "grad_norm": 0.005047450307756662,
-      "learning_rate": 0.00019997855643524599,
-      "loss": 46.0,
-      "step": 40945
-    },
-    {
-      "epoch": 6.594065783646685,
-      "grad_norm": 0.013005949556827545,
-      "learning_rate": 0.00019997855538758213,
-      "loss": 46.0,
-      "step": 40946
-    },
-    {
-      "epoch": 6.594226820725472,
-      "grad_norm": 0.008321783505380154,
-      "learning_rate": 0.00019997855433989266,
-      "loss": 46.0,
-      "step": 40947
-    },
-    {
-      "epoch": 6.594387857804259,
-      "grad_norm": 0.003362649353221059,
-      "learning_rate": 0.00019997855329217763,
-      "loss": 46.0,
-      "step": 40948
-    },
-    {
-      "epoch": 6.594548894883047,
-      "grad_norm": 0.006200097035616636,
-      "learning_rate": 0.000199978552244437,
-      "loss": 46.0,
-      "step": 40949
-    },
-    {
-      "epoch": 6.594709931961834,
-      "grad_norm": 0.015620601363480091,
-      "learning_rate": 0.00019997855119667076,
-      "loss": 46.0,
-      "step": 40950
-    },
-    {
-      "epoch": 6.594870969040621,
-      "grad_norm": 0.004222431220114231,
-      "learning_rate": 0.00019997855014887894,
-      "loss": 46.0,
-      "step": 40951
-    },
-    {
-      "epoch": 6.595032006119409,
-      "grad_norm": 0.003103445516899228,
-      "learning_rate": 0.00019997854910106156,
-      "loss": 46.0,
-      "step": 40952
-    },
-    {
-      "epoch": 6.595193043198196,
-      "grad_norm": 0.013094968162477016,
-      "learning_rate": 0.00019997854805321855,
-      "loss": 46.0,
-      "step": 40953
-    },
-    {
-      "epoch": 6.595354080276984,
-      "grad_norm": 0.004147983621805906,
-      "learning_rate": 0.00019997854700534994,
-      "loss": 46.0,
-      "step": 40954
-    },
-    {
-      "epoch": 6.595515117355771,
-      "grad_norm": 0.004095536656677723,
-      "learning_rate": 0.00019997854595745577,
-      "loss": 46.0,
-      "step": 40955
-    },
-    {
-      "epoch": 6.595676154434559,
-      "grad_norm": 0.010508733801543713,
-      "learning_rate": 0.00019997854490953602,
-      "loss": 46.0,
-      "step": 40956
-    },
-    {
-      "epoch": 6.595837191513346,
-      "grad_norm": 0.006186679471284151,
-      "learning_rate": 0.00019997854386159068,
-      "loss": 46.0,
-      "step": 40957
-    },
-    {
-      "epoch": 6.5959982285921335,
-      "grad_norm": 0.016042983159422874,
-      "learning_rate": 0.00019997854281361975,
-      "loss": 46.0,
-      "step": 40958
-    },
-    {
-      "epoch": 6.596159265670921,
-      "grad_norm": 0.0016949169803410769,
-      "learning_rate": 0.0001999785417656232,
-      "loss": 46.0,
-      "step": 40959
-    },
-    {
-      "epoch": 6.596320302749708,
-      "grad_norm": 0.004380188882350922,
-      "learning_rate": 0.00019997854071760108,
-      "loss": 46.0,
-      "step": 40960
-    },
-    {
-      "epoch": 6.596481339828496,
-      "grad_norm": 0.0059052156284451485,
-      "learning_rate": 0.00019997853966955336,
-      "loss": 46.0,
-      "step": 40961
-    },
-    {
-      "epoch": 6.596642376907283,
-      "grad_norm": 0.009181967005133629,
-      "learning_rate": 0.0001999785386214801,
-      "loss": 46.0,
-      "step": 40962
-    },
-    {
-      "epoch": 6.59680341398607,
-      "grad_norm": 0.0032888425048440695,
-      "learning_rate": 0.0001999785375733812,
-      "loss": 46.0,
-      "step": 40963
-    },
-    {
-      "epoch": 6.596964451064858,
-      "grad_norm": 0.006101379171013832,
-      "learning_rate": 0.00019997853652525672,
-      "loss": 46.0,
-      "step": 40964
-    },
-    {
-      "epoch": 6.597125488143645,
-      "grad_norm": 0.014859865419566631,
-      "learning_rate": 0.00019997853547710665,
-      "loss": 46.0,
-      "step": 40965
-    },
-    {
-      "epoch": 6.597286525222432,
-      "grad_norm": 0.005723286885768175,
-      "learning_rate": 0.000199978534428931,
-      "loss": 46.0,
-      "step": 40966
-    },
-    {
-      "epoch": 6.59744756230122,
-      "grad_norm": 0.0016400391468778253,
-      "learning_rate": 0.00019997853338072976,
-      "loss": 46.0,
-      "step": 40967
-    },
-    {
-      "epoch": 6.597608599380007,
-      "grad_norm": 0.003897425951436162,
-      "learning_rate": 0.0001999785323325029,
-      "loss": 46.0,
-      "step": 40968
-    },
-    {
-      "epoch": 6.5977696364587946,
-      "grad_norm": 0.00242862687446177,
-      "learning_rate": 0.0001999785312842505,
-      "loss": 46.0,
-      "step": 40969
-    },
-    {
-      "epoch": 6.597930673537582,
-      "grad_norm": 0.011331310495734215,
-      "learning_rate": 0.00019997853023597247,
-      "loss": 46.0,
-      "step": 40970
-    },
-    {
-      "epoch": 6.5980917106163695,
-      "grad_norm": 0.008717311546206474,
-      "learning_rate": 0.00019997852918766888,
-      "loss": 46.0,
-      "step": 40971
-    },
-    {
-      "epoch": 6.598252747695157,
-      "grad_norm": 0.009111875668168068,
-      "learning_rate": 0.00019997852813933968,
-      "loss": 46.0,
-      "step": 40972
-    },
-    {
-      "epoch": 6.598413784773944,
-      "grad_norm": 0.0011732830898836255,
-      "learning_rate": 0.00019997852709098492,
-      "loss": 46.0,
-      "step": 40973
-    },
-    {
-      "epoch": 6.598574821852732,
-      "grad_norm": 0.0075826495885849,
-      "learning_rate": 0.00019997852604260454,
-      "loss": 46.0,
-      "step": 40974
-    },
-    {
-      "epoch": 6.598735858931519,
-      "grad_norm": 0.013696345500648022,
-      "learning_rate": 0.00019997852499419855,
-      "loss": 46.0,
-      "step": 40975
-    },
-    {
-      "epoch": 6.598896896010307,
-      "grad_norm": 0.0021244550589472055,
-      "learning_rate": 0.00019997852394576702,
-      "loss": 46.0,
-      "step": 40976
-    },
-    {
-      "epoch": 6.599057933089094,
-      "grad_norm": 0.0035381598863750696,
-      "learning_rate": 0.00019997852289730989,
-      "loss": 46.0,
-      "step": 40977
-    },
-    {
-      "epoch": 6.599218970167881,
-      "grad_norm": 0.0014341837959364057,
-      "learning_rate": 0.00019997852184882716,
-      "loss": 46.0,
-      "step": 40978
-    },
-    {
-      "epoch": 6.599380007246668,
-      "grad_norm": 0.00937070045620203,
-      "learning_rate": 0.00019997852080031882,
-      "loss": 46.0,
-      "step": 40979
-    },
-    {
-      "epoch": 6.599541044325456,
-      "grad_norm": 0.0015439860289916396,
-      "learning_rate": 0.00019997851975178495,
-      "loss": 46.0,
-      "step": 40980
-    },
-    {
-      "epoch": 6.599702081404243,
-      "grad_norm": 0.004059721250087023,
-      "learning_rate": 0.00019997851870322543,
-      "loss": 46.0,
-      "step": 40981
-    },
-    {
-      "epoch": 6.5998631184830305,
-      "grad_norm": 0.0023277876898646355,
-      "learning_rate": 0.00019997851765464036,
-      "loss": 46.0,
-      "step": 40982
-    },
-    {
-      "epoch": 6.600024155561818,
-      "grad_norm": 0.006967457942664623,
-      "learning_rate": 0.0001999785166060297,
-      "loss": 46.0,
-      "step": 40983
-    },
-    {
-      "epoch": 6.600185192640605,
-      "grad_norm": 0.0016623716801404953,
-      "learning_rate": 0.0001999785155573934,
-      "loss": 46.0,
-      "step": 40984
-    },
-    {
-      "epoch": 6.600346229719393,
-      "grad_norm": 0.011348284780979156,
-      "learning_rate": 0.00019997851450873156,
-      "loss": 46.0,
-      "step": 40985
-    },
-    {
-      "epoch": 6.60050726679818,
-      "grad_norm": 0.005747534334659576,
-      "learning_rate": 0.00019997851346004414,
-      "loss": 46.0,
-      "step": 40986
-    },
-    {
-      "epoch": 6.600668303876968,
-      "grad_norm": 0.0016448006499558687,
-      "learning_rate": 0.00019997851241133107,
-      "loss": 46.0,
-      "step": 40987
-    },
-    {
-      "epoch": 6.600829340955755,
-      "grad_norm": 0.009472008794546127,
-      "learning_rate": 0.00019997851136259245,
-      "loss": 46.0,
-      "step": 40988
-    },
-    {
-      "epoch": 6.600990378034543,
-      "grad_norm": 0.004862797446548939,
-      "learning_rate": 0.00019997851031382827,
-      "loss": 46.0,
-      "step": 40989
-    },
-    {
-      "epoch": 6.60115141511333,
-      "grad_norm": 0.010619417764246464,
-      "learning_rate": 0.00019997850926503844,
-      "loss": 46.0,
-      "step": 40990
-    },
-    {
-      "epoch": 6.601312452192118,
-      "grad_norm": 0.0033544304315000772,
-      "learning_rate": 0.00019997850821622305,
-      "loss": 46.0,
-      "step": 40991
-    },
-    {
-      "epoch": 6.601473489270905,
-      "grad_norm": 0.01468256488442421,
-      "learning_rate": 0.00019997850716738208,
-      "loss": 46.0,
-      "step": 40992
-    },
-    {
-      "epoch": 6.601634526349692,
-      "grad_norm": 0.009828968904912472,
-      "learning_rate": 0.0001999785061185155,
-      "loss": 46.0,
-      "step": 40993
-    },
-    {
-      "epoch": 6.601795563428479,
-      "grad_norm": 0.00132093601860106,
-      "learning_rate": 0.00019997850506962335,
-      "loss": 46.0,
-      "step": 40994
-    },
-    {
-      "epoch": 6.6019566005072665,
-      "grad_norm": 0.0014363840455189347,
-      "learning_rate": 0.0001999785040207056,
-      "loss": 46.0,
-      "step": 40995
-    },
-    {
-      "epoch": 6.602117637586054,
-      "grad_norm": 0.004275656770914793,
-      "learning_rate": 0.00019997850297176226,
-      "loss": 46.0,
-      "step": 40996
-    },
-    {
-      "epoch": 6.602278674664841,
-      "grad_norm": 0.006909401621669531,
-      "learning_rate": 0.00019997850192279333,
-      "loss": 46.0,
-      "step": 40997
-    },
-    {
-      "epoch": 6.602439711743629,
-      "grad_norm": 0.0018044457538053393,
-      "learning_rate": 0.00019997850087379883,
-      "loss": 46.0,
-      "step": 40998
-    },
-    {
-      "epoch": 6.602600748822416,
-      "grad_norm": 0.012258270755410194,
-      "learning_rate": 0.00019997849982477872,
-      "loss": 46.0,
-      "step": 40999
-    },
-    {
-      "epoch": 6.602761785901204,
-      "grad_norm": 0.01446872390806675,
-      "learning_rate": 0.00019997849877573302,
-      "loss": 46.0,
-      "step": 41000
-    },
-    {
-      "epoch": 6.602922822979991,
-      "grad_norm": 0.0036528734490275383,
-      "learning_rate": 0.00019997849772666174,
-      "loss": 46.0,
-      "step": 41001
-    },
-    {
-      "epoch": 6.603083860058779,
-      "grad_norm": 0.0015007511246949434,
-      "learning_rate": 0.00019997849667756484,
-      "loss": 46.0,
-      "step": 41002
-    },
-    {
-      "epoch": 6.603244897137566,
-      "grad_norm": 0.0034847212955355644,
-      "learning_rate": 0.0001999784956284424,
-      "loss": 46.0,
-      "step": 41003
-    },
-    {
-      "epoch": 6.603405934216354,
-      "grad_norm": 0.003368258010596037,
-      "learning_rate": 0.00019997849457929436,
-      "loss": 46.0,
-      "step": 41004
-    },
-    {
-      "epoch": 6.603566971295141,
-      "grad_norm": 0.0019945099484175444,
-      "learning_rate": 0.00019997849353012073,
-      "loss": 46.0,
-      "step": 41005
-    },
-    {
-      "epoch": 6.6037280083739285,
-      "grad_norm": 0.002101441379636526,
-      "learning_rate": 0.00019997849248092148,
-      "loss": 46.0,
-      "step": 41006
-    },
-    {
-      "epoch": 6.603889045452716,
-      "grad_norm": 0.013742894865572453,
-      "learning_rate": 0.00019997849143169668,
-      "loss": 46.0,
-      "step": 41007
-    },
-    {
-      "epoch": 6.6040500825315025,
-      "grad_norm": 0.022664552554488182,
-      "learning_rate": 0.00019997849038244625,
-      "loss": 46.0,
-      "step": 41008
-    },
-    {
-      "epoch": 6.60421111961029,
-      "grad_norm": 0.006416258867830038,
-      "learning_rate": 0.00019997848933317025,
-      "loss": 46.0,
-      "step": 41009
-    },
-    {
-      "epoch": 6.604372156689077,
-      "grad_norm": 0.0036397199146449566,
-      "learning_rate": 0.00019997848828386868,
-      "loss": 46.0,
-      "step": 41010
-    },
-    {
-      "epoch": 6.604533193767865,
-      "grad_norm": 0.012458142824470997,
-      "learning_rate": 0.0001999784872345415,
-      "loss": 46.0,
-      "step": 41011
-    },
-    {
-      "epoch": 6.604694230846652,
-      "grad_norm": 0.009527890011668205,
-      "learning_rate": 0.00019997848618518872,
-      "loss": 46.0,
-      "step": 41012
-    },
-    {
-      "epoch": 6.60485526792544,
-      "grad_norm": 0.009535655379295349,
-      "learning_rate": 0.00019997848513581036,
-      "loss": 46.0,
-      "step": 41013
-    },
-    {
-      "epoch": 6.605016305004227,
-      "grad_norm": 0.007483928930014372,
-      "learning_rate": 0.00019997848408640642,
-      "loss": 46.0,
-      "step": 41014
-    },
-    {
-      "epoch": 6.605177342083015,
-      "grad_norm": 0.0066192299127578735,
-      "learning_rate": 0.00019997848303697691,
-      "loss": 46.0,
-      "step": 41015
-    },
-    {
-      "epoch": 6.605338379161802,
-      "grad_norm": 0.004171983804553747,
-      "learning_rate": 0.00019997848198752177,
-      "loss": 46.0,
-      "step": 41016
-    },
-    {
-      "epoch": 6.60549941624059,
-      "grad_norm": 0.004527946934103966,
-      "learning_rate": 0.00019997848093804104,
-      "loss": 46.0,
-      "step": 41017
-    },
-    {
-      "epoch": 6.605660453319377,
-      "grad_norm": 0.004235798958688974,
-      "learning_rate": 0.00019997847988853474,
-      "loss": 46.0,
-      "step": 41018
-    },
-    {
-      "epoch": 6.6058214903981645,
-      "grad_norm": 0.00447089271619916,
-      "learning_rate": 0.00019997847883900283,
-      "loss": 46.0,
-      "step": 41019
-    },
-    {
-      "epoch": 6.605982527476952,
-      "grad_norm": 0.002794915111735463,
-      "learning_rate": 0.0001999784777894454,
-      "loss": 46.0,
-      "step": 41020
-    },
-    {
-      "epoch": 6.6061435645557385,
-      "grad_norm": 0.003605864243581891,
-      "learning_rate": 0.00019997847673986228,
-      "loss": 46.0,
-      "step": 41021
-    },
-    {
-      "epoch": 6.606304601634527,
-      "grad_norm": 0.007824013940989971,
-      "learning_rate": 0.00019997847569025364,
-      "loss": 46.0,
-      "step": 41022
-    },
-    {
-      "epoch": 6.606465638713313,
-      "grad_norm": 0.012308187782764435,
-      "learning_rate": 0.00019997847464061939,
-      "loss": 46.0,
-      "step": 41023
-    },
-    {
-      "epoch": 6.606626675792101,
-      "grad_norm": 0.004476560279726982,
-      "learning_rate": 0.00019997847359095957,
-      "loss": 46.0,
-      "step": 41024
-    },
-    {
-      "epoch": 6.606787712870888,
-      "grad_norm": 0.0016742307925596833,
-      "learning_rate": 0.00019997847254127414,
-      "loss": 46.0,
-      "step": 41025
-    },
-    {
-      "epoch": 6.606948749949676,
-      "grad_norm": 0.006550919730216265,
-      "learning_rate": 0.00019997847149156312,
-      "loss": 46.0,
-      "step": 41026
-    },
-    {
-      "epoch": 6.607109787028463,
-      "grad_norm": 0.001879259361885488,
-      "learning_rate": 0.00019997847044182651,
-      "loss": 46.0,
-      "step": 41027
-    },
-    {
-      "epoch": 6.607270824107251,
-      "grad_norm": 0.001983418595045805,
-      "learning_rate": 0.0001999784693920643,
-      "loss": 46.0,
-      "step": 41028
-    },
-    {
-      "epoch": 6.607431861186038,
-      "grad_norm": 0.002833044156432152,
-      "learning_rate": 0.00019997846834227652,
-      "loss": 46.0,
-      "step": 41029
-    },
-    {
-      "epoch": 6.6075928982648255,
-      "grad_norm": 0.0032925051636993885,
-      "learning_rate": 0.00019997846729246315,
-      "loss": 46.0,
-      "step": 41030
-    },
-    {
-      "epoch": 6.607753935343613,
-      "grad_norm": 0.0021400924306362867,
-      "learning_rate": 0.0001999784662426242,
-      "loss": 46.0,
-      "step": 41031
-    },
-    {
-      "epoch": 6.6079149724224004,
-      "grad_norm": 0.00787995383143425,
-      "learning_rate": 0.00019997846519275963,
-      "loss": 46.0,
-      "step": 41032
-    },
-    {
-      "epoch": 6.608076009501188,
-      "grad_norm": 0.004491230007261038,
-      "learning_rate": 0.00019997846414286947,
-      "loss": 46.0,
-      "step": 41033
-    },
-    {
-      "epoch": 6.608237046579975,
-      "grad_norm": 0.0006392892100848258,
-      "learning_rate": 0.00019997846309295376,
-      "loss": 46.0,
-      "step": 41034
-    },
-    {
-      "epoch": 6.608398083658763,
-      "grad_norm": 0.0030184241477400064,
-      "learning_rate": 0.00019997846204301243,
-      "loss": 46.0,
-      "step": 41035
-    },
-    {
-      "epoch": 6.608559120737549,
-      "grad_norm": 0.0017763242358341813,
-      "learning_rate": 0.0001999784609930455,
-      "loss": 46.0,
-      "step": 41036
-    },
-    {
-      "epoch": 6.608720157816338,
-      "grad_norm": 0.021407047286629677,
-      "learning_rate": 0.00019997845994305303,
-      "loss": 46.0,
-      "step": 41037
-    },
-    {
-      "epoch": 6.608881194895124,
-      "grad_norm": 0.006149849388748407,
-      "learning_rate": 0.00019997845889303492,
-      "loss": 46.0,
-      "step": 41038
-    },
-    {
-      "epoch": 6.609042231973912,
-      "grad_norm": 0.011370550841093063,
-      "learning_rate": 0.00019997845784299126,
-      "loss": 46.0,
-      "step": 41039
-    },
-    {
-      "epoch": 6.609203269052699,
-      "grad_norm": 0.0022126734256744385,
-      "learning_rate": 0.00019997845679292197,
-      "loss": 46.0,
-      "step": 41040
-    },
-    {
-      "epoch": 6.609364306131487,
-      "grad_norm": 0.008452162146568298,
-      "learning_rate": 0.00019997845574282712,
-      "loss": 46.0,
-      "step": 41041
-    },
-    {
-      "epoch": 6.609525343210274,
-      "grad_norm": 0.005821246653795242,
-      "learning_rate": 0.00019997845469270668,
-      "loss": 46.0,
-      "step": 41042
-    },
-    {
-      "epoch": 6.6096863802890615,
-      "grad_norm": 0.018352357670664787,
-      "learning_rate": 0.00019997845364256062,
-      "loss": 46.0,
-      "step": 41043
-    },
-    {
-      "epoch": 6.609847417367849,
-      "grad_norm": 0.008488418534398079,
-      "learning_rate": 0.000199978452592389,
-      "loss": 46.0,
-      "step": 41044
-    },
-    {
-      "epoch": 6.610008454446636,
-      "grad_norm": 0.008687576279044151,
-      "learning_rate": 0.0001999784515421918,
-      "loss": 46.0,
-      "step": 41045
-    },
-    {
-      "epoch": 6.610169491525424,
-      "grad_norm": 0.0065894159488379955,
-      "learning_rate": 0.000199978450491969,
-      "loss": 46.0,
-      "step": 41046
-    },
-    {
-      "epoch": 6.610330528604211,
-      "grad_norm": 0.003859002608805895,
-      "learning_rate": 0.0001999784494417206,
-      "loss": 46.0,
-      "step": 41047
-    },
-    {
-      "epoch": 6.610491565682999,
-      "grad_norm": 0.0011809971183538437,
-      "learning_rate": 0.00019997844839144663,
-      "loss": 46.0,
-      "step": 41048
-    },
-    {
-      "epoch": 6.610652602761786,
-      "grad_norm": 0.020778413861989975,
-      "learning_rate": 0.00019997844734114705,
-      "loss": 46.0,
-      "step": 41049
-    },
-    {
-      "epoch": 6.610813639840574,
-      "grad_norm": 0.006073105148971081,
-      "learning_rate": 0.0001999784462908219,
-      "loss": 46.0,
-      "step": 41050
-    },
-    {
-      "epoch": 6.61097467691936,
-      "grad_norm": 0.00774040911346674,
-      "learning_rate": 0.0001999784452404711,
-      "loss": 46.0,
-      "step": 41051
-    },
-    {
-      "epoch": 6.611135713998148,
-      "grad_norm": 0.0050262147560715675,
-      "learning_rate": 0.0001999784441900948,
-      "loss": 46.0,
-      "step": 41052
-    },
-    {
-      "epoch": 6.611296751076935,
-      "grad_norm": 0.006304901093244553,
-      "learning_rate": 0.00019997844313969284,
-      "loss": 46.0,
-      "step": 41053
-    },
-    {
-      "epoch": 6.611457788155723,
-      "grad_norm": 0.00688542565330863,
-      "learning_rate": 0.00019997844208926533,
-      "loss": 46.0,
-      "step": 41054
-    },
-    {
-      "epoch": 6.61161882523451,
-      "grad_norm": 0.0029033017344772816,
-      "learning_rate": 0.00019997844103881223,
-      "loss": 46.0,
-      "step": 41055
-    },
-    {
-      "epoch": 6.6117798623132975,
-      "grad_norm": 0.004210735205560923,
-      "learning_rate": 0.00019997843998833354,
-      "loss": 46.0,
-      "step": 41056
-    },
-    {
-      "epoch": 6.611940899392085,
-      "grad_norm": 0.005546304862946272,
-      "learning_rate": 0.00019997843893782924,
-      "loss": 46.0,
-      "step": 41057
-    },
-    {
-      "epoch": 6.612101936470872,
-      "grad_norm": 0.0018246410181745887,
-      "learning_rate": 0.00019997843788729935,
-      "loss": 46.0,
-      "step": 41058
-    },
-    {
-      "epoch": 6.61226297354966,
-      "grad_norm": 0.018975183367729187,
-      "learning_rate": 0.00019997843683674388,
-      "loss": 46.0,
-      "step": 41059
-    },
-    {
-      "epoch": 6.612424010628447,
-      "grad_norm": 0.01114060077816248,
-      "learning_rate": 0.00019997843578616284,
-      "loss": 46.0,
-      "step": 41060
-    },
-    {
-      "epoch": 6.612585047707235,
-      "grad_norm": 0.00270092417486012,
-      "learning_rate": 0.0001999784347355562,
-      "loss": 46.0,
-      "step": 41061
-    },
-    {
-      "epoch": 6.612746084786022,
-      "grad_norm": 0.004035099409520626,
-      "learning_rate": 0.00019997843368492398,
-      "loss": 46.0,
-      "step": 41062
-    },
-    {
-      "epoch": 6.61290712186481,
-      "grad_norm": 0.005571081768721342,
-      "learning_rate": 0.00019997843263426616,
-      "loss": 46.0,
-      "step": 41063
-    },
-    {
-      "epoch": 6.613068158943597,
-      "grad_norm": 0.02523900754749775,
-      "learning_rate": 0.00019997843158358274,
-      "loss": 46.0,
-      "step": 41064
-    },
-    {
-      "epoch": 6.613229196022385,
-      "grad_norm": 0.01615004613995552,
-      "learning_rate": 0.00019997843053287372,
-      "loss": 46.0,
-      "step": 41065
-    },
-    {
-      "epoch": 6.613390233101171,
-      "grad_norm": 0.008677395060658455,
-      "learning_rate": 0.00019997842948213913,
-      "loss": 46.0,
-      "step": 41066
-    },
-    {
-      "epoch": 6.613551270179959,
-      "grad_norm": 0.02636909857392311,
-      "learning_rate": 0.00019997842843137896,
-      "loss": 46.0,
-      "step": 41067
-    },
-    {
-      "epoch": 6.613712307258746,
-      "grad_norm": 0.0016285745659843087,
-      "learning_rate": 0.00019997842738059317,
-      "loss": 46.0,
-      "step": 41068
-    },
-    {
-      "epoch": 6.6138733443375335,
-      "grad_norm": 0.0035759760066866875,
-      "learning_rate": 0.00019997842632978182,
-      "loss": 46.0,
-      "step": 41069
-    },
-    {
-      "epoch": 6.614034381416321,
-      "grad_norm": 0.001556383678689599,
-      "learning_rate": 0.00019997842527894486,
-      "loss": 46.0,
-      "step": 41070
-    },
-    {
-      "epoch": 6.614195418495108,
-      "grad_norm": 0.006428986322134733,
-      "learning_rate": 0.00019997842422808234,
-      "loss": 46.0,
-      "step": 41071
-    },
-    {
-      "epoch": 6.614356455573896,
-      "grad_norm": 0.002071127062663436,
-      "learning_rate": 0.00019997842317719423,
-      "loss": 46.0,
-      "step": 41072
-    },
-    {
-      "epoch": 6.614517492652683,
-      "grad_norm": 0.0017701896140351892,
-      "learning_rate": 0.00019997842212628048,
-      "loss": 46.0,
-      "step": 41073
-    },
-    {
-      "epoch": 6.614678529731471,
-      "grad_norm": 0.017276708036661148,
-      "learning_rate": 0.00019997842107534117,
-      "loss": 46.0,
-      "step": 41074
-    },
-    {
-      "epoch": 6.614839566810258,
-      "grad_norm": 0.0033349853474646807,
-      "learning_rate": 0.0001999784200243763,
-      "loss": 46.0,
-      "step": 41075
-    },
-    {
-      "epoch": 6.615000603889046,
-      "grad_norm": 0.004386052954941988,
-      "learning_rate": 0.0001999784189733858,
-      "loss": 46.0,
-      "step": 41076
-    },
-    {
-      "epoch": 6.615161640967833,
-      "grad_norm": 0.019192369654774666,
-      "learning_rate": 0.00019997841792236974,
-      "loss": 46.0,
-      "step": 41077
-    },
-    {
-      "epoch": 6.6153226780466206,
-      "grad_norm": 0.011368497274816036,
-      "learning_rate": 0.00019997841687132808,
-      "loss": 46.0,
-      "step": 41078
-    },
-    {
-      "epoch": 6.615483715125408,
-      "grad_norm": 0.0017469103913754225,
-      "learning_rate": 0.0001999784158202608,
-      "loss": 46.0,
-      "step": 41079
-    },
-    {
-      "epoch": 6.6156447522041955,
-      "grad_norm": 0.0031596715562045574,
-      "learning_rate": 0.00019997841476916797,
-      "loss": 46.0,
-      "step": 41080
-    },
-    {
-      "epoch": 6.615805789282982,
-      "grad_norm": 0.0018933404935523868,
-      "learning_rate": 0.00019997841371804955,
-      "loss": 46.0,
-      "step": 41081
-    },
-    {
-      "epoch": 6.6159668263617695,
-      "grad_norm": 0.0034992152359336615,
-      "learning_rate": 0.00019997841266690552,
-      "loss": 46.0,
-      "step": 41082
-    },
-    {
-      "epoch": 6.616127863440557,
-      "grad_norm": 0.0057185981422662735,
-      "learning_rate": 0.0001999784116157359,
-      "loss": 46.0,
-      "step": 41083
-    },
-    {
-      "epoch": 6.616288900519344,
-      "grad_norm": 0.001386741641908884,
-      "learning_rate": 0.00019997841056454071,
-      "loss": 46.0,
-      "step": 41084
-    },
-    {
-      "epoch": 6.616449937598132,
-      "grad_norm": 0.0051393453031778336,
-      "learning_rate": 0.00019997840951331992,
-      "loss": 46.0,
-      "step": 41085
-    },
-    {
-      "epoch": 6.616610974676919,
-      "grad_norm": 0.0008212599204853177,
-      "learning_rate": 0.00019997840846207353,
-      "loss": 46.0,
-      "step": 41086
-    },
-    {
-      "epoch": 6.616772011755707,
-      "grad_norm": 0.01296071894466877,
-      "learning_rate": 0.0001999784074108016,
-      "loss": 46.0,
-      "step": 41087
-    },
-    {
-      "epoch": 6.616933048834494,
-      "grad_norm": 0.002062899060547352,
-      "learning_rate": 0.000199978406359504,
-      "loss": 46.0,
-      "step": 41088
-    },
-    {
-      "epoch": 6.617094085913282,
-      "grad_norm": 0.0179916899651289,
-      "learning_rate": 0.00019997840530818086,
-      "loss": 46.0,
-      "step": 41089
-    },
-    {
-      "epoch": 6.617255122992069,
-      "grad_norm": 0.002371684880927205,
-      "learning_rate": 0.00019997840425683215,
-      "loss": 46.0,
-      "step": 41090
-    },
-    {
-      "epoch": 6.6174161600708565,
-      "grad_norm": 0.0025576360058039427,
-      "learning_rate": 0.0001999784032054578,
-      "loss": 46.0,
-      "step": 41091
-    },
-    {
-      "epoch": 6.617577197149644,
-      "grad_norm": 0.0023558472748845816,
-      "learning_rate": 0.00019997840215405787,
-      "loss": 46.0,
-      "step": 41092
-    },
-    {
-      "epoch": 6.617738234228431,
-      "grad_norm": 0.003779823426157236,
-      "learning_rate": 0.0001999784011026324,
-      "loss": 46.0,
-      "step": 41093
-    },
-    {
-      "epoch": 6.617899271307218,
-      "grad_norm": 0.0016630009049549699,
-      "learning_rate": 0.0001999784000511813,
-      "loss": 46.0,
-      "step": 41094
-    },
-    {
-      "epoch": 6.618060308386006,
-      "grad_norm": 0.0050531914457678795,
-      "learning_rate": 0.0001999783989997046,
-      "loss": 46.0,
-      "step": 41095
-    },
-    {
-      "epoch": 6.618221345464793,
-      "grad_norm": 0.007217828184366226,
-      "learning_rate": 0.00019997839794820235,
-      "loss": 46.0,
-      "step": 41096
-    },
-    {
-      "epoch": 6.61838238254358,
-      "grad_norm": 0.004020967986434698,
-      "learning_rate": 0.00019997839689667445,
-      "loss": 46.0,
-      "step": 41097
-    },
-    {
-      "epoch": 6.618543419622368,
-      "grad_norm": 0.0027252137660980225,
-      "learning_rate": 0.00019997839584512105,
-      "loss": 46.0,
-      "step": 41098
-    },
-    {
-      "epoch": 6.618704456701155,
-      "grad_norm": 0.008718476630747318,
-      "learning_rate": 0.000199978394793542,
-      "loss": 46.0,
-      "step": 41099
-    },
-    {
-      "epoch": 6.618865493779943,
-      "grad_norm": 0.004874077625572681,
-      "learning_rate": 0.00019997839374193737,
-      "loss": 46.0,
-      "step": 41100
-    },
-    {
-      "epoch": 6.61902653085873,
-      "grad_norm": 0.001031780382618308,
-      "learning_rate": 0.00019997839269030713,
-      "loss": 46.0,
-      "step": 41101
-    },
-    {
-      "epoch": 6.619187567937518,
-      "grad_norm": 0.0021287561394274235,
-      "learning_rate": 0.00019997839163865132,
-      "loss": 46.0,
-      "step": 41102
-    },
-    {
-      "epoch": 6.619348605016305,
-      "grad_norm": 0.005838374141603708,
-      "learning_rate": 0.00019997839058696993,
-      "loss": 46.0,
-      "step": 41103
-    },
-    {
-      "epoch": 6.6195096420950925,
-      "grad_norm": 0.01185868214815855,
-      "learning_rate": 0.00019997838953526292,
-      "loss": 46.0,
-      "step": 41104
-    },
-    {
-      "epoch": 6.61967067917388,
-      "grad_norm": 0.002571073593571782,
-      "learning_rate": 0.00019997838848353035,
-      "loss": 46.0,
-      "step": 41105
-    },
-    {
-      "epoch": 6.619831716252667,
-      "grad_norm": 0.008395260199904442,
-      "learning_rate": 0.0001999783874317722,
-      "loss": 46.0,
-      "step": 41106
-    },
-    {
-      "epoch": 6.619992753331455,
-      "grad_norm": 0.0031780200079083443,
-      "learning_rate": 0.00019997838637998846,
-      "loss": 46.0,
-      "step": 41107
-    },
-    {
-      "epoch": 6.620153790410242,
-      "grad_norm": 0.0061030020006000996,
-      "learning_rate": 0.0001999783853281791,
-      "loss": 46.0,
-      "step": 41108
-    },
-    {
-      "epoch": 6.620314827489029,
-      "grad_norm": 0.0018235614988952875,
-      "learning_rate": 0.00019997838427634416,
-      "loss": 46.0,
-      "step": 41109
-    },
-    {
-      "epoch": 6.620475864567817,
-      "grad_norm": 0.003984051290899515,
-      "learning_rate": 0.00019997838322448366,
-      "loss": 46.0,
-      "step": 41110
-    },
-    {
-      "epoch": 6.620636901646604,
-      "grad_norm": 0.0016804715851321816,
-      "learning_rate": 0.0001999783821725975,
-      "loss": 46.0,
-      "step": 41111
-    },
-    {
-      "epoch": 6.620797938725391,
-      "grad_norm": 0.0058973077684640884,
-      "learning_rate": 0.00019997838112068583,
-      "loss": 46.0,
-      "step": 41112
-    },
-    {
-      "epoch": 6.620958975804179,
-      "grad_norm": 0.0025775195099413395,
-      "learning_rate": 0.00019997838006874854,
-      "loss": 46.0,
-      "step": 41113
-    },
-    {
-      "epoch": 6.621120012882966,
-      "grad_norm": 0.0027494679670780897,
-      "learning_rate": 0.00019997837901678566,
-      "loss": 46.0,
-      "step": 41114
-    },
-    {
-      "epoch": 6.621281049961754,
-      "grad_norm": 0.002431021071970463,
-      "learning_rate": 0.0001999783779647972,
-      "loss": 46.0,
-      "step": 41115
-    },
-    {
-      "epoch": 6.621442087040541,
-      "grad_norm": 0.0037709027528762817,
-      "learning_rate": 0.00019997837691278312,
-      "loss": 46.0,
-      "step": 41116
-    },
-    {
-      "epoch": 6.6216031241193285,
-      "grad_norm": 0.0024342546239495277,
-      "learning_rate": 0.00019997837586074348,
-      "loss": 46.0,
-      "step": 41117
-    },
-    {
-      "epoch": 6.621764161198116,
-      "grad_norm": 0.0029722130857408047,
-      "learning_rate": 0.00019997837480867825,
-      "loss": 46.0,
-      "step": 41118
-    },
-    {
-      "epoch": 6.621925198276903,
-      "grad_norm": 0.006364222150295973,
-      "learning_rate": 0.0001999783737565874,
-      "loss": 46.0,
-      "step": 41119
-    },
-    {
-      "epoch": 6.622086235355691,
-      "grad_norm": 0.0027700962964445353,
-      "learning_rate": 0.000199978372704471,
-      "loss": 46.0,
-      "step": 41120
-    },
-    {
-      "epoch": 6.622247272434478,
-      "grad_norm": 0.003449861891567707,
-      "learning_rate": 0.000199978371652329,
-      "loss": 46.0,
-      "step": 41121
-    },
-    {
-      "epoch": 6.622408309513266,
-      "grad_norm": 0.00423266040161252,
-      "learning_rate": 0.00019997837060016141,
-      "loss": 46.0,
-      "step": 41122
-    },
-    {
-      "epoch": 6.622569346592053,
-      "grad_norm": 0.007254303898662329,
-      "learning_rate": 0.00019997836954796822,
-      "loss": 46.0,
-      "step": 41123
-    },
-    {
-      "epoch": 6.62273038367084,
-      "grad_norm": 0.0058316076174378395,
-      "learning_rate": 0.00019997836849574945,
-      "loss": 46.0,
-      "step": 41124
-    },
-    {
-      "epoch": 6.622891420749627,
-      "grad_norm": 0.008839089423418045,
-      "learning_rate": 0.00019997836744350508,
-      "loss": 46.0,
-      "step": 41125
-    },
-    {
-      "epoch": 6.623052457828415,
-      "grad_norm": 0.004016009159386158,
-      "learning_rate": 0.00019997836639123513,
-      "loss": 46.0,
-      "step": 41126
-    },
-    {
-      "epoch": 6.623213494907202,
-      "grad_norm": 0.012918100692331791,
-      "learning_rate": 0.0001999783653389396,
-      "loss": 46.0,
-      "step": 41127
-    },
-    {
-      "epoch": 6.62337453198599,
-      "grad_norm": 0.02669021673500538,
-      "learning_rate": 0.00019997836428661846,
-      "loss": 46.0,
-      "step": 41128
-    },
-    {
-      "epoch": 6.623535569064777,
-      "grad_norm": 0.002213901374489069,
-      "learning_rate": 0.00019997836323427172,
-      "loss": 46.0,
-      "step": 41129
-    },
-    {
-      "epoch": 6.6236966061435645,
-      "grad_norm": 0.0016288220649585128,
-      "learning_rate": 0.00019997836218189942,
-      "loss": 46.0,
-      "step": 41130
-    },
-    {
-      "epoch": 6.623857643222352,
-      "grad_norm": 0.007245820015668869,
-      "learning_rate": 0.00019997836112950154,
-      "loss": 46.0,
-      "step": 41131
-    },
-    {
-      "epoch": 6.624018680301139,
-      "grad_norm": 0.00630686990916729,
-      "learning_rate": 0.00019997836007707803,
-      "loss": 46.0,
-      "step": 41132
-    },
-    {
-      "epoch": 6.624179717379927,
-      "grad_norm": 0.002153727924451232,
-      "learning_rate": 0.00019997835902462897,
-      "loss": 46.0,
-      "step": 41133
-    },
-    {
-      "epoch": 6.624340754458714,
-      "grad_norm": 0.001160130137577653,
-      "learning_rate": 0.0001999783579721543,
-      "loss": 46.0,
-      "step": 41134
-    },
-    {
-      "epoch": 6.624501791537502,
-      "grad_norm": 0.0025597941130399704,
-      "learning_rate": 0.00019997835691965406,
-      "loss": 46.0,
-      "step": 41135
-    },
-    {
-      "epoch": 6.624662828616289,
-      "grad_norm": 0.005943742580711842,
-      "learning_rate": 0.0001999783558671282,
-      "loss": 46.0,
-      "step": 41136
-    },
-    {
-      "epoch": 6.624823865695077,
-      "grad_norm": 0.0011956996750086546,
-      "learning_rate": 0.00019997835481457677,
-      "loss": 46.0,
-      "step": 41137
-    },
-    {
-      "epoch": 6.624984902773864,
-      "grad_norm": 0.008160853758454323,
-      "learning_rate": 0.00019997835376199974,
-      "loss": 46.0,
-      "step": 41138
-    },
-    {
-      "epoch": 6.625145939852651,
-      "grad_norm": 0.0016330907819792628,
-      "learning_rate": 0.0001999783527093971,
-      "loss": 46.0,
-      "step": 41139
-    },
-    {
-      "epoch": 6.625306976931438,
-      "grad_norm": 0.0023520688991993666,
-      "learning_rate": 0.00019997835165676893,
-      "loss": 46.0,
-      "step": 41140
-    },
-    {
-      "epoch": 6.625468014010226,
-      "grad_norm": 0.0037449656520038843,
-      "learning_rate": 0.00019997835060411512,
-      "loss": 46.0,
-      "step": 41141
-    },
-    {
-      "epoch": 6.625629051089013,
-      "grad_norm": 0.013129186816513538,
-      "learning_rate": 0.00019997834955143575,
-      "loss": 46.0,
-      "step": 41142
-    },
-    {
-      "epoch": 6.6257900881678005,
-      "grad_norm": 0.001787198125384748,
-      "learning_rate": 0.00019997834849873076,
-      "loss": 46.0,
-      "step": 41143
-    },
-    {
-      "epoch": 6.625951125246588,
-      "grad_norm": 0.003374287160113454,
-      "learning_rate": 0.0001999783474460002,
-      "loss": 46.0,
-      "step": 41144
-    },
-    {
-      "epoch": 6.626112162325375,
-      "grad_norm": 0.005721678491681814,
-      "learning_rate": 0.00019997834639324407,
-      "loss": 46.0,
-      "step": 41145
-    },
-    {
-      "epoch": 6.626273199404163,
-      "grad_norm": 0.0050403098575770855,
-      "learning_rate": 0.00019997834534046232,
-      "loss": 46.0,
-      "step": 41146
-    },
-    {
-      "epoch": 6.62643423648295,
-      "grad_norm": 0.0048050605691969395,
-      "learning_rate": 0.00019997834428765501,
-      "loss": 46.0,
-      "step": 41147
-    },
-    {
-      "epoch": 6.626595273561738,
-      "grad_norm": 0.012133794836699963,
-      "learning_rate": 0.00019997834323482206,
-      "loss": 46.0,
-      "step": 41148
-    },
-    {
-      "epoch": 6.626756310640525,
-      "grad_norm": 0.007100074551999569,
-      "learning_rate": 0.00019997834218196355,
-      "loss": 46.0,
-      "step": 41149
-    },
-    {
-      "epoch": 6.626917347719313,
-      "grad_norm": 0.006903825327754021,
-      "learning_rate": 0.00019997834112907948,
-      "loss": 46.0,
-      "step": 41150
-    },
-    {
-      "epoch": 6.6270783847981,
-      "grad_norm": 0.006233078893274069,
-      "learning_rate": 0.0001999783400761698,
-      "loss": 46.0,
-      "step": 41151
-    },
-    {
-      "epoch": 6.6272394218768875,
-      "grad_norm": 0.015550930052995682,
-      "learning_rate": 0.0001999783390232345,
-      "loss": 46.0,
-      "step": 41152
-    },
-    {
-      "epoch": 6.627400458955675,
-      "grad_norm": 0.004609704948961735,
-      "learning_rate": 0.00019997833797027366,
-      "loss": 46.0,
-      "step": 41153
-    },
-    {
-      "epoch": 6.6275614960344615,
-      "grad_norm": 0.006760106887668371,
-      "learning_rate": 0.0001999783369172872,
-      "loss": 46.0,
-      "step": 41154
-    },
-    {
-      "epoch": 6.627722533113249,
-      "grad_norm": 0.002083009108901024,
-      "learning_rate": 0.00019997833586427516,
-      "loss": 46.0,
-      "step": 41155
-    },
-    {
-      "epoch": 6.6278835701920364,
-      "grad_norm": 0.0013089803978800774,
-      "learning_rate": 0.00019997833481123753,
-      "loss": 46.0,
-      "step": 41156
-    },
-    {
-      "epoch": 6.628044607270824,
-      "grad_norm": 0.0077623240649700165,
-      "learning_rate": 0.0001999783337581743,
-      "loss": 46.0,
-      "step": 41157
-    },
-    {
-      "epoch": 6.628205644349611,
-      "grad_norm": 0.0020460295490920544,
-      "learning_rate": 0.00019997833270508548,
-      "loss": 46.0,
-      "step": 41158
-    },
-    {
-      "epoch": 6.628366681428399,
-      "grad_norm": 0.0013743790332227945,
-      "learning_rate": 0.0001999783316519711,
-      "loss": 46.0,
-      "step": 41159
-    },
-    {
-      "epoch": 6.628527718507186,
-      "grad_norm": 0.004074026830494404,
-      "learning_rate": 0.00019997833059883112,
-      "loss": 46.0,
-      "step": 41160
-    },
-    {
-      "epoch": 6.628688755585974,
-      "grad_norm": 0.003076382912695408,
-      "learning_rate": 0.00019997832954566554,
-      "loss": 46.0,
-      "step": 41161
-    },
-    {
-      "epoch": 6.628849792664761,
-      "grad_norm": 0.0028957887552678585,
-      "learning_rate": 0.00019997832849247437,
-      "loss": 46.0,
-      "step": 41162
-    },
-    {
-      "epoch": 6.629010829743549,
-      "grad_norm": 0.0035534012131392956,
-      "learning_rate": 0.0001999783274392576,
-      "loss": 46.0,
-      "step": 41163
-    },
-    {
-      "epoch": 6.629171866822336,
-      "grad_norm": 0.008254371583461761,
-      "learning_rate": 0.00019997832638601526,
-      "loss": 46.0,
-      "step": 41164
-    },
-    {
-      "epoch": 6.6293329039011235,
-      "grad_norm": 0.004110126290470362,
-      "learning_rate": 0.00019997832533274733,
-      "loss": 46.0,
-      "step": 41165
-    },
-    {
-      "epoch": 6.629493940979911,
-      "grad_norm": 0.0065147290006279945,
-      "learning_rate": 0.0001999783242794538,
-      "loss": 46.0,
-      "step": 41166
-    },
-    {
-      "epoch": 6.6296549780586975,
-      "grad_norm": 0.007026699371635914,
-      "learning_rate": 0.00019997832322613468,
-      "loss": 46.0,
-      "step": 41167
-    },
-    {
-      "epoch": 6.629816015137486,
-      "grad_norm": 0.010335328988730907,
-      "learning_rate": 0.00019997832217278998,
-      "loss": 46.0,
-      "step": 41168
-    },
-    {
-      "epoch": 6.629977052216272,
-      "grad_norm": 0.0031468693632632494,
-      "learning_rate": 0.00019997832111941967,
-      "loss": 46.0,
-      "step": 41169
-    },
-    {
-      "epoch": 6.63013808929506,
-      "grad_norm": 0.0037579922936856747,
-      "learning_rate": 0.0001999783200660238,
-      "loss": 46.0,
-      "step": 41170
-    },
-    {
-      "epoch": 6.630299126373847,
-      "grad_norm": 0.015171720646321774,
-      "learning_rate": 0.00019997831901260232,
-      "loss": 46.0,
-      "step": 41171
-    },
-    {
-      "epoch": 6.630460163452635,
-      "grad_norm": 0.0034958517644554377,
-      "learning_rate": 0.00019997831795915528,
-      "loss": 46.0,
-      "step": 41172
-    },
-    {
-      "epoch": 6.630621200531422,
-      "grad_norm": 0.0050625279545784,
-      "learning_rate": 0.00019997831690568262,
-      "loss": 46.0,
-      "step": 41173
-    },
-    {
-      "epoch": 6.63078223761021,
-      "grad_norm": 0.0009731936734169722,
-      "learning_rate": 0.00019997831585218435,
-      "loss": 46.0,
-      "step": 41174
-    },
-    {
-      "epoch": 6.630943274688997,
-      "grad_norm": 0.013226950541138649,
-      "learning_rate": 0.00019997831479866057,
-      "loss": 46.0,
-      "step": 41175
-    },
-    {
-      "epoch": 6.631104311767785,
-      "grad_norm": 0.00621444545686245,
-      "learning_rate": 0.00019997831374511112,
-      "loss": 46.0,
-      "step": 41176
-    },
-    {
-      "epoch": 6.631265348846572,
-      "grad_norm": 0.0019267479656264186,
-      "learning_rate": 0.00019997831269153612,
-      "loss": 46.0,
-      "step": 41177
-    },
-    {
-      "epoch": 6.6314263859253595,
-      "grad_norm": 0.0015656901523470879,
-      "learning_rate": 0.00019997831163793552,
-      "loss": 46.0,
-      "step": 41178
-    },
-    {
-      "epoch": 6.631587423004147,
-      "grad_norm": 0.00155361695215106,
-      "learning_rate": 0.00019997831058430934,
-      "loss": 46.0,
-      "step": 41179
-    },
-    {
-      "epoch": 6.631748460082934,
-      "grad_norm": 0.006082166451960802,
-      "learning_rate": 0.00019997830953065755,
-      "loss": 46.0,
-      "step": 41180
-    },
-    {
-      "epoch": 6.631909497161722,
-      "grad_norm": 0.004673150833696127,
-      "learning_rate": 0.0001999783084769802,
-      "loss": 46.0,
-      "step": 41181
-    },
-    {
-      "epoch": 6.632070534240508,
-      "grad_norm": 0.0023002601228654385,
-      "learning_rate": 0.00019997830742327723,
-      "loss": 46.0,
-      "step": 41182
-    },
-    {
-      "epoch": 6.632231571319297,
-      "grad_norm": 0.003658955218270421,
-      "learning_rate": 0.0001999783063695487,
-      "loss": 46.0,
-      "step": 41183
-    },
-    {
-      "epoch": 6.632392608398083,
-      "grad_norm": 0.0015056031988933682,
-      "learning_rate": 0.00019997830531579455,
-      "loss": 46.0,
-      "step": 41184
-    },
-    {
-      "epoch": 6.632553645476871,
-      "grad_norm": 0.004236246459186077,
-      "learning_rate": 0.00019997830426201485,
-      "loss": 46.0,
-      "step": 41185
-    },
-    {
-      "epoch": 6.632714682555658,
-      "grad_norm": 0.0032955333590507507,
-      "learning_rate": 0.00019997830320820953,
-      "loss": 46.0,
-      "step": 41186
-    },
-    {
-      "epoch": 6.632875719634446,
-      "grad_norm": 0.003155897604301572,
-      "learning_rate": 0.00019997830215437863,
-      "loss": 46.0,
-      "step": 41187
-    },
-    {
-      "epoch": 6.633036756713233,
-      "grad_norm": 0.0054049682803452015,
-      "learning_rate": 0.00019997830110052216,
-      "loss": 46.0,
-      "step": 41188
-    },
-    {
-      "epoch": 6.633197793792021,
-      "grad_norm": 0.006927745882421732,
-      "learning_rate": 0.00019997830004664006,
-      "loss": 46.0,
-      "step": 41189
-    },
-    {
-      "epoch": 6.633358830870808,
-      "grad_norm": 0.003276033326983452,
-      "learning_rate": 0.0001999782989927324,
-      "loss": 46.0,
-      "step": 41190
-    },
-    {
-      "epoch": 6.6335198679495955,
-      "grad_norm": 0.003885750425979495,
-      "learning_rate": 0.00019997829793879914,
-      "loss": 46.0,
-      "step": 41191
-    },
-    {
-      "epoch": 6.633680905028383,
-      "grad_norm": 0.026463288813829422,
-      "learning_rate": 0.0001999782968848403,
-      "loss": 46.0,
-      "step": 41192
-    },
-    {
-      "epoch": 6.63384194210717,
-      "grad_norm": 0.004569042939692736,
-      "learning_rate": 0.00019997829583085587,
-      "loss": 46.0,
-      "step": 41193
-    },
-    {
-      "epoch": 6.634002979185958,
-      "grad_norm": 0.003324605990201235,
-      "learning_rate": 0.00019997829477684582,
-      "loss": 46.0,
-      "step": 41194
-    },
-    {
-      "epoch": 6.634164016264745,
-      "grad_norm": 0.002257521264255047,
-      "learning_rate": 0.0001999782937228102,
-      "loss": 46.0,
-      "step": 41195
-    },
-    {
-      "epoch": 6.634325053343533,
-      "grad_norm": 0.0006985857035033405,
-      "learning_rate": 0.000199978292668749,
-      "loss": 46.0,
-      "step": 41196
-    },
-    {
-      "epoch": 6.634486090422319,
-      "grad_norm": 0.004117724485695362,
-      "learning_rate": 0.0001999782916146622,
-      "loss": 46.0,
-      "step": 41197
-    },
-    {
-      "epoch": 6.634647127501107,
-      "grad_norm": 0.005097700282931328,
-      "learning_rate": 0.00019997829056054984,
-      "loss": 46.0,
-      "step": 41198
-    },
-    {
-      "epoch": 6.634808164579894,
-      "grad_norm": 0.02677321434020996,
-      "learning_rate": 0.00019997828950641186,
-      "loss": 46.0,
-      "step": 41199
-    },
-    {
-      "epoch": 6.634969201658682,
-      "grad_norm": 0.0101937772706151,
-      "learning_rate": 0.0001999782884522483,
-      "loss": 46.0,
-      "step": 41200
-    },
-    {
-      "epoch": 6.635130238737469,
-      "grad_norm": 0.0020081985276192427,
-      "learning_rate": 0.00019997828739805912,
-      "loss": 46.0,
-      "step": 41201
-    },
-    {
-      "epoch": 6.6352912758162566,
-      "grad_norm": 0.002573740668594837,
-      "learning_rate": 0.0001999782863438444,
-      "loss": 46.0,
-      "step": 41202
-    },
-    {
-      "epoch": 6.635452312895044,
-      "grad_norm": 0.009583241306245327,
-      "learning_rate": 0.00019997828528960408,
-      "loss": 46.0,
-      "step": 41203
-    },
-    {
-      "epoch": 6.6356133499738315,
-      "grad_norm": 0.007890709675848484,
-      "learning_rate": 0.00019997828423533816,
-      "loss": 46.0,
-      "step": 41204
-    },
-    {
-      "epoch": 6.635774387052619,
-      "grad_norm": 0.0036622032057493925,
-      "learning_rate": 0.00019997828318104666,
-      "loss": 46.0,
-      "step": 41205
-    },
-    {
-      "epoch": 6.635935424131406,
-      "grad_norm": 0.0033045532181859016,
-      "learning_rate": 0.00019997828212672955,
-      "loss": 46.0,
-      "step": 41206
-    },
-    {
-      "epoch": 6.636096461210194,
-      "grad_norm": 0.003874134737998247,
-      "learning_rate": 0.00019997828107238684,
-      "loss": 46.0,
-      "step": 41207
-    },
-    {
-      "epoch": 6.636257498288981,
-      "grad_norm": 0.005551476962864399,
-      "learning_rate": 0.00019997828001801858,
-      "loss": 46.0,
-      "step": 41208
-    },
-    {
-      "epoch": 6.636418535367769,
-      "grad_norm": 0.0020345996599644423,
-      "learning_rate": 0.00019997827896362473,
-      "loss": 46.0,
-      "step": 41209
-    },
-    {
-      "epoch": 6.636579572446556,
-      "grad_norm": 0.006998830940574408,
-      "learning_rate": 0.00019997827790920524,
-      "loss": 46.0,
-      "step": 41210
-    },
-    {
-      "epoch": 6.636740609525344,
-      "grad_norm": 0.005958764813840389,
-      "learning_rate": 0.00019997827685476022,
-      "loss": 46.0,
-      "step": 41211
-    },
-    {
-      "epoch": 6.63690164660413,
-      "grad_norm": 0.005613343324512243,
-      "learning_rate": 0.00019997827580028958,
-      "loss": 46.0,
-      "step": 41212
-    },
-    {
-      "epoch": 6.637062683682918,
-      "grad_norm": 0.0014376823091879487,
-      "learning_rate": 0.00019997827474579335,
-      "loss": 46.0,
-      "step": 41213
-    },
-    {
-      "epoch": 6.637223720761705,
-      "grad_norm": 0.007475843653082848,
-      "learning_rate": 0.00019997827369127157,
-      "loss": 46.0,
-      "step": 41214
-    },
-    {
-      "epoch": 6.6373847578404925,
-      "grad_norm": 0.0028920944314450026,
-      "learning_rate": 0.00019997827263672414,
-      "loss": 46.0,
-      "step": 41215
-    },
-    {
-      "epoch": 6.63754579491928,
-      "grad_norm": 0.003980892710387707,
-      "learning_rate": 0.00019997827158215118,
-      "loss": 46.0,
-      "step": 41216
-    },
-    {
-      "epoch": 6.637706831998067,
-      "grad_norm": 0.018523935228586197,
-      "learning_rate": 0.00019997827052755258,
-      "loss": 46.0,
-      "step": 41217
-    },
-    {
-      "epoch": 6.637867869076855,
-      "grad_norm": 0.003589930944144726,
-      "learning_rate": 0.00019997826947292842,
-      "loss": 46.0,
-      "step": 41218
-    },
-    {
-      "epoch": 6.638028906155642,
-      "grad_norm": 0.0012552345870062709,
-      "learning_rate": 0.00019997826841827864,
-      "loss": 46.0,
-      "step": 41219
-    },
-    {
-      "epoch": 6.63818994323443,
-      "grad_norm": 0.006195634603500366,
-      "learning_rate": 0.0001999782673636033,
-      "loss": 46.0,
-      "step": 41220
-    },
-    {
-      "epoch": 6.638350980313217,
-      "grad_norm": 0.0015473488019779325,
-      "learning_rate": 0.00019997826630890235,
-      "loss": 46.0,
-      "step": 41221
-    },
-    {
-      "epoch": 6.638512017392005,
-      "grad_norm": 0.01875738613307476,
-      "learning_rate": 0.00019997826525417582,
-      "loss": 46.0,
-      "step": 41222
-    },
-    {
-      "epoch": 6.638673054470792,
-      "grad_norm": 0.0023227131459861994,
-      "learning_rate": 0.00019997826419942372,
-      "loss": 46.0,
-      "step": 41223
-    },
-    {
-      "epoch": 6.63883409154958,
-      "grad_norm": 0.0035696225240826607,
-      "learning_rate": 0.00019997826314464604,
-      "loss": 46.0,
-      "step": 41224
-    },
-    {
-      "epoch": 6.638995128628367,
-      "grad_norm": 0.0018136863363906741,
-      "learning_rate": 0.0001999782620898427,
-      "loss": 46.0,
-      "step": 41225
-    },
-    {
-      "epoch": 6.6391561657071545,
-      "grad_norm": 0.0017965954029932618,
-      "learning_rate": 0.00019997826103501382,
-      "loss": 46.0,
-      "step": 41226
-    },
-    {
-      "epoch": 6.639317202785941,
-      "grad_norm": 0.002883186796680093,
-      "learning_rate": 0.00019997825998015935,
-      "loss": 46.0,
-      "step": 41227
-    },
-    {
-      "epoch": 6.6394782398647285,
-      "grad_norm": 0.006738876923918724,
-      "learning_rate": 0.00019997825892527932,
-      "loss": 46.0,
-      "step": 41228
-    },
-    {
-      "epoch": 6.639639276943516,
-      "grad_norm": 0.0017054156633093953,
-      "learning_rate": 0.00019997825787037364,
-      "loss": 46.0,
-      "step": 41229
-    },
-    {
-      "epoch": 6.639800314022303,
-      "grad_norm": 0.015789788216352463,
-      "learning_rate": 0.0001999782568154424,
-      "loss": 46.0,
-      "step": 41230
-    },
-    {
-      "epoch": 6.639961351101091,
-      "grad_norm": 0.0012283873511478305,
-      "learning_rate": 0.0001999782557604856,
-      "loss": 46.0,
-      "step": 41231
-    },
-    {
-      "epoch": 6.640122388179878,
-      "grad_norm": 0.0022991145960986614,
-      "learning_rate": 0.00019997825470550315,
-      "loss": 46.0,
-      "step": 41232
-    },
-    {
-      "epoch": 6.640283425258666,
-      "grad_norm": 0.007703986018896103,
-      "learning_rate": 0.00019997825365049513,
-      "loss": 46.0,
-      "step": 41233
-    },
-    {
-      "epoch": 6.640444462337453,
-      "grad_norm": 0.006665561348199844,
-      "learning_rate": 0.00019997825259546154,
-      "loss": 46.0,
-      "step": 41234
-    },
-    {
-      "epoch": 6.640605499416241,
-      "grad_norm": 0.006929056718945503,
-      "learning_rate": 0.00019997825154040237,
-      "loss": 46.0,
-      "step": 41235
-    },
-    {
-      "epoch": 6.640766536495028,
-      "grad_norm": 0.0023154662922024727,
-      "learning_rate": 0.00019997825048531756,
-      "loss": 46.0,
-      "step": 41236
-    },
-    {
-      "epoch": 6.640927573573816,
-      "grad_norm": 0.011944940313696861,
-      "learning_rate": 0.0001999782494302072,
-      "loss": 46.0,
-      "step": 41237
-    },
-    {
-      "epoch": 6.641088610652603,
-      "grad_norm": 0.006466531660407782,
-      "learning_rate": 0.00019997824837507126,
-      "loss": 46.0,
-      "step": 41238
-    },
-    {
-      "epoch": 6.6412496477313905,
-      "grad_norm": 0.013151127845048904,
-      "learning_rate": 0.0001999782473199097,
-      "loss": 46.0,
-      "step": 41239
-    },
-    {
-      "epoch": 6.641410684810177,
-      "grad_norm": 0.03186663240194321,
-      "learning_rate": 0.00019997824626472258,
-      "loss": 46.0,
-      "step": 41240
-    },
-    {
-      "epoch": 6.641571721888965,
-      "grad_norm": 0.0044538117945194244,
-      "learning_rate": 0.00019997824520950986,
-      "loss": 46.0,
-      "step": 41241
-    },
-    {
-      "epoch": 6.641732758967752,
-      "grad_norm": 0.012548520229756832,
-      "learning_rate": 0.00019997824415427155,
-      "loss": 46.0,
-      "step": 41242
-    },
-    {
-      "epoch": 6.641893796046539,
-      "grad_norm": 0.011111288331449032,
-      "learning_rate": 0.00019997824309900763,
-      "loss": 46.0,
-      "step": 41243
-    },
-    {
-      "epoch": 6.642054833125327,
-      "grad_norm": 0.003127770032733679,
-      "learning_rate": 0.00019997824204371815,
-      "loss": 46.0,
-      "step": 41244
-    },
-    {
-      "epoch": 6.642215870204114,
-      "grad_norm": 0.008705846033990383,
-      "learning_rate": 0.00019997824098840305,
-      "loss": 46.0,
-      "step": 41245
-    },
-    {
-      "epoch": 6.642376907282902,
-      "grad_norm": 0.0007114775362424552,
-      "learning_rate": 0.0001999782399330624,
-      "loss": 46.0,
-      "step": 41246
-    },
-    {
-      "epoch": 6.642537944361689,
-      "grad_norm": 0.0087512182071805,
-      "learning_rate": 0.00019997823887769615,
-      "loss": 46.0,
-      "step": 41247
-    },
-    {
-      "epoch": 6.642698981440477,
-      "grad_norm": 0.0017549317562952638,
-      "learning_rate": 0.0001999782378223043,
-      "loss": 46.0,
-      "step": 41248
-    },
-    {
-      "epoch": 6.642860018519264,
-      "grad_norm": 0.007432641927152872,
-      "learning_rate": 0.00019997823676688685,
-      "loss": 46.0,
-      "step": 41249
-    },
-    {
-      "epoch": 6.643021055598052,
-      "grad_norm": 0.003654295578598976,
-      "learning_rate": 0.00019997823571144384,
-      "loss": 46.0,
-      "step": 41250
-    },
-    {
-      "epoch": 6.643182092676839,
-      "grad_norm": 0.009870448149740696,
-      "learning_rate": 0.0001999782346559752,
-      "loss": 46.0,
-      "step": 41251
-    },
-    {
-      "epoch": 6.6433431297556265,
-      "grad_norm": 0.0028439168818295,
-      "learning_rate": 0.00019997823360048102,
-      "loss": 46.0,
-      "step": 41252
-    },
-    {
-      "epoch": 6.643504166834414,
-      "grad_norm": 0.005359800532460213,
-      "learning_rate": 0.0001999782325449612,
-      "loss": 46.0,
-      "step": 41253
-    },
-    {
-      "epoch": 6.643665203913201,
-      "grad_norm": 0.003559744916856289,
-      "learning_rate": 0.00019997823148941581,
-      "loss": 46.0,
-      "step": 41254
-    },
-    {
-      "epoch": 6.643826240991988,
-      "grad_norm": 0.0015891611110419035,
-      "learning_rate": 0.00019997823043384485,
-      "loss": 46.0,
-      "step": 41255
-    },
-    {
-      "epoch": 6.643987278070776,
-      "grad_norm": 0.001680420245975256,
-      "learning_rate": 0.0001999782293782483,
-      "loss": 46.0,
-      "step": 41256
-    },
-    {
-      "epoch": 6.644148315149563,
-      "grad_norm": 0.003986607771366835,
-      "learning_rate": 0.00019997822832262615,
-      "loss": 46.0,
-      "step": 41257
-    },
-    {
-      "epoch": 6.64430935222835,
-      "grad_norm": 0.005513923708349466,
-      "learning_rate": 0.0001999782272669784,
-      "loss": 46.0,
-      "step": 41258
-    },
-    {
-      "epoch": 6.644470389307138,
-      "grad_norm": 0.0016328264027833939,
-      "learning_rate": 0.00019997822621130507,
-      "loss": 46.0,
-      "step": 41259
-    },
-    {
-      "epoch": 6.644631426385925,
-      "grad_norm": 0.008640449494123459,
-      "learning_rate": 0.00019997822515560614,
-      "loss": 46.0,
-      "step": 41260
-    },
-    {
-      "epoch": 6.644792463464713,
-      "grad_norm": 0.008669750764966011,
-      "learning_rate": 0.00019997822409988163,
-      "loss": 46.0,
-      "step": 41261
-    },
-    {
-      "epoch": 6.6449535005435,
-      "grad_norm": 0.004570953082293272,
-      "learning_rate": 0.00019997822304413152,
-      "loss": 46.0,
-      "step": 41262
-    },
-    {
-      "epoch": 6.6451145376222875,
-      "grad_norm": 0.0025238145608454943,
-      "learning_rate": 0.00019997822198835583,
-      "loss": 46.0,
-      "step": 41263
-    },
-    {
-      "epoch": 6.645275574701075,
-      "grad_norm": 0.00553938327357173,
-      "learning_rate": 0.00019997822093255455,
-      "loss": 46.0,
-      "step": 41264
-    },
-    {
-      "epoch": 6.6454366117798624,
-      "grad_norm": 0.006852875463664532,
-      "learning_rate": 0.0001999782198767277,
-      "loss": 46.0,
-      "step": 41265
-    },
-    {
-      "epoch": 6.64559764885865,
-      "grad_norm": 0.004326727241277695,
-      "learning_rate": 0.00019997821882087525,
-      "loss": 46.0,
-      "step": 41266
-    },
-    {
-      "epoch": 6.645758685937437,
-      "grad_norm": 0.0020337312016636133,
-      "learning_rate": 0.00019997821776499719,
-      "loss": 46.0,
-      "step": 41267
-    },
-    {
-      "epoch": 6.645919723016225,
-      "grad_norm": 0.009487706236541271,
-      "learning_rate": 0.00019997821670909356,
-      "loss": 46.0,
-      "step": 41268
-    },
-    {
-      "epoch": 6.646080760095012,
-      "grad_norm": 0.0014569255290552974,
-      "learning_rate": 0.00019997821565316434,
-      "loss": 46.0,
-      "step": 41269
-    },
-    {
-      "epoch": 6.646241797173799,
-      "grad_norm": 0.005698317661881447,
-      "learning_rate": 0.0001999782145972095,
-      "loss": 46.0,
-      "step": 41270
-    },
-    {
-      "epoch": 6.646402834252587,
-      "grad_norm": 0.019783319905400276,
-      "learning_rate": 0.00019997821354122912,
-      "loss": 46.0,
-      "step": 41271
-    },
-    {
-      "epoch": 6.646563871331374,
-      "grad_norm": 0.007254421710968018,
-      "learning_rate": 0.00019997821248522314,
-      "loss": 46.0,
-      "step": 41272
-    },
-    {
-      "epoch": 6.646724908410161,
-      "grad_norm": 0.0017364928498864174,
-      "learning_rate": 0.00019997821142919155,
-      "loss": 46.0,
-      "step": 41273
-    },
-    {
-      "epoch": 6.646885945488949,
-      "grad_norm": 0.0014317418681457639,
-      "learning_rate": 0.00019997821037313437,
-      "loss": 46.0,
-      "step": 41274
-    },
-    {
-      "epoch": 6.647046982567736,
-      "grad_norm": 0.010185652412474155,
-      "learning_rate": 0.00019997820931705164,
-      "loss": 46.0,
-      "step": 41275
-    },
-    {
-      "epoch": 6.6472080196465235,
-      "grad_norm": 0.002763714175671339,
-      "learning_rate": 0.00019997820826094328,
-      "loss": 46.0,
-      "step": 41276
-    },
-    {
-      "epoch": 6.647369056725311,
-      "grad_norm": 0.0007979915826581419,
-      "learning_rate": 0.00019997820720480934,
-      "loss": 46.0,
-      "step": 41277
-    },
-    {
-      "epoch": 6.647530093804098,
-      "grad_norm": 0.02134522795677185,
-      "learning_rate": 0.0001999782061486498,
-      "loss": 46.0,
-      "step": 41278
-    },
-    {
-      "epoch": 6.647691130882886,
-      "grad_norm": 0.003303972538560629,
-      "learning_rate": 0.0001999782050924647,
-      "loss": 46.0,
-      "step": 41279
-    },
-    {
-      "epoch": 6.647852167961673,
-      "grad_norm": 0.005645588971674442,
-      "learning_rate": 0.00019997820403625397,
-      "loss": 46.0,
-      "step": 41280
-    },
-    {
-      "epoch": 6.648013205040461,
-      "grad_norm": 0.005770730786025524,
-      "learning_rate": 0.00019997820298001768,
-      "loss": 46.0,
-      "step": 41281
-    },
-    {
-      "epoch": 6.648174242119248,
-      "grad_norm": 0.0007078899652697146,
-      "learning_rate": 0.0001999782019237558,
-      "loss": 46.0,
-      "step": 41282
-    },
-    {
-      "epoch": 6.648335279198036,
-      "grad_norm": 0.0071189431473612785,
-      "learning_rate": 0.00019997820086746834,
-      "loss": 46.0,
-      "step": 41283
-    },
-    {
-      "epoch": 6.648496316276823,
-      "grad_norm": 0.0028874436393380165,
-      "learning_rate": 0.00019997819981115527,
-      "loss": 46.0,
-      "step": 41284
-    },
-    {
-      "epoch": 6.64865735335561,
-      "grad_norm": 0.030310943722724915,
-      "learning_rate": 0.0001999781987548166,
-      "loss": 46.0,
-      "step": 41285
-    },
-    {
-      "epoch": 6.648818390434397,
-      "grad_norm": 0.003774383570998907,
-      "learning_rate": 0.00019997819769845235,
-      "loss": 46.0,
-      "step": 41286
-    },
-    {
-      "epoch": 6.648979427513185,
-      "grad_norm": 0.00117953447625041,
-      "learning_rate": 0.00019997819664206254,
-      "loss": 46.0,
-      "step": 41287
-    },
-    {
-      "epoch": 6.649140464591972,
-      "grad_norm": 0.003956024069339037,
-      "learning_rate": 0.0001999781955856471,
-      "loss": 46.0,
-      "step": 41288
-    },
-    {
-      "epoch": 6.6493015016707595,
-      "grad_norm": 0.0027459862176328897,
-      "learning_rate": 0.00019997819452920613,
-      "loss": 46.0,
-      "step": 41289
-    },
-    {
-      "epoch": 6.649462538749547,
-      "grad_norm": 0.0014338220935314894,
-      "learning_rate": 0.00019997819347273953,
-      "loss": 46.0,
-      "step": 41290
-    },
-    {
-      "epoch": 6.649623575828334,
-      "grad_norm": 0.007602814119309187,
-      "learning_rate": 0.00019997819241624734,
-      "loss": 46.0,
-      "step": 41291
-    },
-    {
-      "epoch": 6.649784612907122,
-      "grad_norm": 0.016116492450237274,
-      "learning_rate": 0.00019997819135972954,
-      "loss": 46.0,
-      "step": 41292
-    },
-    {
-      "epoch": 6.649945649985909,
-      "grad_norm": 0.0049855369143188,
-      "learning_rate": 0.00019997819030318618,
-      "loss": 46.0,
-      "step": 41293
-    },
-    {
-      "epoch": 6.650106687064697,
-      "grad_norm": 0.016336875036358833,
-      "learning_rate": 0.00019997818924661723,
-      "loss": 46.0,
-      "step": 41294
-    },
-    {
-      "epoch": 6.650267724143484,
-      "grad_norm": 0.012784531340003014,
-      "learning_rate": 0.00019997818819002266,
-      "loss": 46.0,
-      "step": 41295
-    },
-    {
-      "epoch": 6.650428761222272,
-      "grad_norm": 0.012742854654788971,
-      "learning_rate": 0.00019997818713340254,
-      "loss": 46.0,
-      "step": 41296
-    },
-    {
-      "epoch": 6.650589798301059,
-      "grad_norm": 0.01026222575455904,
-      "learning_rate": 0.00019997818607675683,
-      "loss": 46.0,
-      "step": 41297
-    },
-    {
-      "epoch": 6.650750835379847,
-      "grad_norm": 0.02060762047767639,
-      "learning_rate": 0.00019997818502008553,
-      "loss": 46.0,
-      "step": 41298
-    },
-    {
-      "epoch": 6.650911872458634,
-      "grad_norm": 0.0023718683514744043,
-      "learning_rate": 0.00019997818396338862,
-      "loss": 46.0,
-      "step": 41299
-    },
-    {
-      "epoch": 6.651072909537421,
-      "grad_norm": 0.026104917749762535,
-      "learning_rate": 0.00019997818290666612,
-      "loss": 46.0,
-      "step": 41300
-    },
-    {
-      "epoch": 6.651233946616208,
-      "grad_norm": 0.009435472078621387,
-      "learning_rate": 0.00019997818184991803,
-      "loss": 46.0,
-      "step": 41301
-    },
-    {
-      "epoch": 6.6513949836949955,
-      "grad_norm": 0.006782672833651304,
-      "learning_rate": 0.00019997818079314436,
-      "loss": 46.0,
-      "step": 41302
-    },
-    {
-      "epoch": 6.651556020773783,
-      "grad_norm": 0.0013868138194084167,
-      "learning_rate": 0.0001999781797363451,
-      "loss": 46.0,
-      "step": 41303
-    },
-    {
-      "epoch": 6.65171705785257,
-      "grad_norm": 0.004889416974037886,
-      "learning_rate": 0.00019997817867952025,
-      "loss": 46.0,
-      "step": 41304
-    },
-    {
-      "epoch": 6.651878094931358,
-      "grad_norm": 0.002161378040909767,
-      "learning_rate": 0.00019997817762266982,
-      "loss": 46.0,
-      "step": 41305
-    },
-    {
-      "epoch": 6.652039132010145,
-      "grad_norm": 0.00307559035718441,
-      "learning_rate": 0.0001999781765657938,
-      "loss": 46.0,
-      "step": 41306
-    },
-    {
-      "epoch": 6.652200169088933,
-      "grad_norm": 0.021258361637592316,
-      "learning_rate": 0.00019997817550889218,
-      "loss": 46.0,
-      "step": 41307
-    },
-    {
-      "epoch": 6.65236120616772,
-      "grad_norm": 0.0028325102757662535,
-      "learning_rate": 0.00019997817445196496,
-      "loss": 46.0,
-      "step": 41308
-    },
-    {
-      "epoch": 6.652522243246508,
-      "grad_norm": 0.001658924506045878,
-      "learning_rate": 0.00019997817339501218,
-      "loss": 46.0,
-      "step": 41309
-    },
-    {
-      "epoch": 6.652683280325295,
-      "grad_norm": 0.0026529680471867323,
-      "learning_rate": 0.0001999781723380338,
-      "loss": 46.0,
-      "step": 41310
-    },
-    {
-      "epoch": 6.6528443174040826,
-      "grad_norm": 0.003789222100749612,
-      "learning_rate": 0.0001999781712810298,
-      "loss": 46.0,
-      "step": 41311
-    },
-    {
-      "epoch": 6.65300535448287,
-      "grad_norm": 0.015022702515125275,
-      "learning_rate": 0.00019997817022400025,
-      "loss": 46.0,
-      "step": 41312
-    },
-    {
-      "epoch": 6.6531663915616575,
-      "grad_norm": 0.005480973049998283,
-      "learning_rate": 0.0001999781691669451,
-      "loss": 46.0,
-      "step": 41313
-    },
-    {
-      "epoch": 6.653327428640445,
-      "grad_norm": 0.00444228108972311,
-      "learning_rate": 0.00019997816810986434,
-      "loss": 46.0,
-      "step": 41314
-    },
-    {
-      "epoch": 6.6534884657192315,
-      "grad_norm": 0.002165646757930517,
-      "learning_rate": 0.00019997816705275804,
-      "loss": 46.0,
-      "step": 41315
-    },
-    {
-      "epoch": 6.653649502798019,
-      "grad_norm": 0.002898496575653553,
-      "learning_rate": 0.0001999781659956261,
-      "loss": 46.0,
-      "step": 41316
-    },
-    {
-      "epoch": 6.653810539876806,
-      "grad_norm": 0.005737585015594959,
-      "learning_rate": 0.0001999781649384686,
-      "loss": 46.0,
-      "step": 41317
-    },
-    {
-      "epoch": 6.653971576955594,
-      "grad_norm": 0.0014970627380535007,
-      "learning_rate": 0.00019997816388128548,
-      "loss": 46.0,
-      "step": 41318
-    },
-    {
-      "epoch": 6.654132614034381,
-      "grad_norm": 0.008356812410056591,
-      "learning_rate": 0.0001999781628240768,
-      "loss": 46.0,
-      "step": 41319
-    },
-    {
-      "epoch": 6.654293651113169,
-      "grad_norm": 0.0011939759133383632,
-      "learning_rate": 0.00019997816176684253,
-      "loss": 46.0,
-      "step": 41320
-    },
-    {
-      "epoch": 6.654454688191956,
-      "grad_norm": 0.0013400658499449492,
-      "learning_rate": 0.00019997816070958267,
-      "loss": 46.0,
-      "step": 41321
-    },
-    {
-      "epoch": 6.654615725270744,
-      "grad_norm": 0.003531099995598197,
-      "learning_rate": 0.0001999781596522972,
-      "loss": 46.0,
-      "step": 41322
-    },
-    {
-      "epoch": 6.654776762349531,
-      "grad_norm": 0.0037914314307272434,
-      "learning_rate": 0.00019997815859498617,
-      "loss": 46.0,
-      "step": 41323
-    },
-    {
-      "epoch": 6.6549377994283185,
-      "grad_norm": 0.003836628282442689,
-      "learning_rate": 0.00019997815753764953,
-      "loss": 46.0,
-      "step": 41324
-    },
-    {
-      "epoch": 6.655098836507106,
-      "grad_norm": 0.003639188129454851,
-      "learning_rate": 0.0001999781564802873,
-      "loss": 46.0,
-      "step": 41325
-    },
-    {
-      "epoch": 6.655259873585893,
-      "grad_norm": 0.004865992348641157,
-      "learning_rate": 0.00019997815542289947,
-      "loss": 46.0,
-      "step": 41326
-    },
-    {
-      "epoch": 6.655420910664681,
-      "grad_norm": 0.00623518880456686,
-      "learning_rate": 0.0001999781543654861,
-      "loss": 46.0,
-      "step": 41327
-    },
-    {
-      "epoch": 6.6555819477434675,
-      "grad_norm": 0.018353795632719994,
-      "learning_rate": 0.0001999781533080471,
-      "loss": 46.0,
-      "step": 41328
-    },
-    {
-      "epoch": 6.655742984822256,
-      "grad_norm": 0.001177751924842596,
-      "learning_rate": 0.0001999781522505825,
-      "loss": 46.0,
-      "step": 41329
-    },
-    {
-      "epoch": 6.655904021901042,
-      "grad_norm": 0.003544985083863139,
-      "learning_rate": 0.00019997815119309235,
-      "loss": 46.0,
-      "step": 41330
-    },
-    {
-      "epoch": 6.65606505897983,
-      "grad_norm": 0.006752661429345608,
-      "learning_rate": 0.0001999781501355766,
-      "loss": 46.0,
-      "step": 41331
-    },
-    {
-      "epoch": 6.656226096058617,
-      "grad_norm": 0.008299527689814568,
-      "learning_rate": 0.00019997814907803522,
-      "loss": 46.0,
-      "step": 41332
-    },
-    {
-      "epoch": 6.656387133137405,
-      "grad_norm": 0.006667337380349636,
-      "learning_rate": 0.0001999781480204683,
-      "loss": 46.0,
-      "step": 41333
-    },
-    {
-      "epoch": 6.656548170216192,
-      "grad_norm": 0.0055719828233122826,
-      "learning_rate": 0.00019997814696287578,
-      "loss": 46.0,
-      "step": 41334
-    },
-    {
-      "epoch": 6.65670920729498,
-      "grad_norm": 0.0019611932802945375,
-      "learning_rate": 0.00019997814590525765,
-      "loss": 46.0,
-      "step": 41335
-    },
-    {
-      "epoch": 6.656870244373767,
-      "grad_norm": 0.0023944592103362083,
-      "learning_rate": 0.00019997814484761395,
-      "loss": 46.0,
-      "step": 41336
-    },
-    {
-      "epoch": 6.6570312814525545,
-      "grad_norm": 0.003260157536715269,
-      "learning_rate": 0.00019997814378994465,
-      "loss": 46.0,
-      "step": 41337
-    },
-    {
-      "epoch": 6.657192318531342,
-      "grad_norm": 0.0038181780837476254,
-      "learning_rate": 0.00019997814273224975,
-      "loss": 46.0,
-      "step": 41338
-    },
-    {
-      "epoch": 6.657353355610129,
-      "grad_norm": 0.002535885199904442,
-      "learning_rate": 0.00019997814167452927,
-      "loss": 46.0,
-      "step": 41339
-    },
-    {
-      "epoch": 6.657514392688917,
-      "grad_norm": 0.0018500848673284054,
-      "learning_rate": 0.00019997814061678323,
-      "loss": 46.0,
-      "step": 41340
-    },
-    {
-      "epoch": 6.657675429767704,
-      "grad_norm": 0.010041836649179459,
-      "learning_rate": 0.00019997813955901158,
-      "loss": 46.0,
-      "step": 41341
-    },
-    {
-      "epoch": 6.657836466846492,
-      "grad_norm": 0.0033802934922277927,
-      "learning_rate": 0.0001999781385012143,
-      "loss": 46.0,
-      "step": 41342
-    },
-    {
-      "epoch": 6.657997503925278,
-      "grad_norm": 0.007863650098443031,
-      "learning_rate": 0.0001999781374433915,
-      "loss": 46.0,
-      "step": 41343
-    },
-    {
-      "epoch": 6.658158541004067,
-      "grad_norm": 0.0029994079377502203,
-      "learning_rate": 0.0001999781363855431,
-      "loss": 46.0,
-      "step": 41344
-    },
-    {
-      "epoch": 6.658319578082853,
-      "grad_norm": 0.004304221365600824,
-      "learning_rate": 0.00019997813532766906,
-      "loss": 46.0,
-      "step": 41345
-    },
-    {
-      "epoch": 6.658480615161641,
-      "grad_norm": 0.004520541988313198,
-      "learning_rate": 0.00019997813426976947,
-      "loss": 46.0,
-      "step": 41346
-    },
-    {
-      "epoch": 6.658641652240428,
-      "grad_norm": 0.0008796003530733287,
-      "learning_rate": 0.0001999781332118443,
-      "loss": 46.0,
-      "step": 41347
-    },
-    {
-      "epoch": 6.658802689319216,
-      "grad_norm": 0.008355273865163326,
-      "learning_rate": 0.0001999781321538935,
-      "loss": 46.0,
-      "step": 41348
-    },
-    {
-      "epoch": 6.658963726398003,
-      "grad_norm": 0.0015010188799351454,
-      "learning_rate": 0.00019997813109591715,
-      "loss": 46.0,
-      "step": 41349
-    },
-    {
-      "epoch": 6.6591247634767905,
-      "grad_norm": 0.01182329747825861,
-      "learning_rate": 0.00019997813003791518,
-      "loss": 46.0,
-      "step": 41350
-    },
-    {
-      "epoch": 6.659285800555578,
-      "grad_norm": 0.0036371885798871517,
-      "learning_rate": 0.00019997812897988763,
-      "loss": 46.0,
-      "step": 41351
-    },
-    {
-      "epoch": 6.659446837634365,
-      "grad_norm": 0.003482918255031109,
-      "learning_rate": 0.0001999781279218345,
-      "loss": 46.0,
-      "step": 41352
-    },
-    {
-      "epoch": 6.659607874713153,
-      "grad_norm": 0.00894628930836916,
-      "learning_rate": 0.00019997812686375576,
-      "loss": 46.0,
-      "step": 41353
-    },
-    {
-      "epoch": 6.65976891179194,
-      "grad_norm": 0.0015628489200025797,
-      "learning_rate": 0.00019997812580565148,
-      "loss": 46.0,
-      "step": 41354
-    },
-    {
-      "epoch": 6.659929948870728,
-      "grad_norm": 0.008763380348682404,
-      "learning_rate": 0.00019997812474752155,
-      "loss": 46.0,
-      "step": 41355
-    },
-    {
-      "epoch": 6.660090985949515,
-      "grad_norm": 0.006190703250467777,
-      "learning_rate": 0.00019997812368936608,
-      "loss": 46.0,
-      "step": 41356
-    },
-    {
-      "epoch": 6.660252023028303,
-      "grad_norm": 0.006778019946068525,
-      "learning_rate": 0.00019997812263118498,
-      "loss": 46.0,
-      "step": 41357
-    },
-    {
-      "epoch": 6.660413060107089,
-      "grad_norm": 0.001629995065741241,
-      "learning_rate": 0.00019997812157297832,
-      "loss": 46.0,
-      "step": 41358
-    },
-    {
-      "epoch": 6.660574097185877,
-      "grad_norm": 0.00679272273555398,
-      "learning_rate": 0.00019997812051474607,
-      "loss": 46.0,
-      "step": 41359
-    },
-    {
-      "epoch": 6.660735134264664,
-      "grad_norm": 0.003945086617022753,
-      "learning_rate": 0.0001999781194564882,
-      "loss": 46.0,
-      "step": 41360
-    },
-    {
-      "epoch": 6.660896171343452,
-      "grad_norm": 0.0028573949821293354,
-      "learning_rate": 0.0001999781183982048,
-      "loss": 46.0,
-      "step": 41361
-    },
-    {
-      "epoch": 6.661057208422239,
-      "grad_norm": 0.010052439756691456,
-      "learning_rate": 0.00019997811733989576,
-      "loss": 46.0,
-      "step": 41362
-    },
-    {
-      "epoch": 6.6612182455010265,
-      "grad_norm": 0.0043817260302603245,
-      "learning_rate": 0.00019997811628156114,
-      "loss": 46.0,
-      "step": 41363
-    },
-    {
-      "epoch": 6.661379282579814,
-      "grad_norm": 0.005444617010653019,
-      "learning_rate": 0.00019997811522320092,
-      "loss": 46.0,
-      "step": 41364
-    },
-    {
-      "epoch": 6.661540319658601,
-      "grad_norm": 0.004544877912849188,
-      "learning_rate": 0.00019997811416481512,
-      "loss": 46.0,
-      "step": 41365
-    },
-    {
-      "epoch": 6.661701356737389,
-      "grad_norm": 0.003842040430754423,
-      "learning_rate": 0.00019997811310640376,
-      "loss": 46.0,
-      "step": 41366
-    },
-    {
-      "epoch": 6.661862393816176,
-      "grad_norm": 0.007663982454687357,
-      "learning_rate": 0.00019997811204796676,
-      "loss": 46.0,
-      "step": 41367
-    },
-    {
-      "epoch": 6.662023430894964,
-      "grad_norm": 0.0035739578306674957,
-      "learning_rate": 0.0001999781109895042,
-      "loss": 46.0,
-      "step": 41368
-    },
-    {
-      "epoch": 6.662184467973751,
-      "grad_norm": 0.0126936174929142,
-      "learning_rate": 0.00019997810993101607,
-      "loss": 46.0,
-      "step": 41369
-    },
-    {
-      "epoch": 6.662345505052539,
-      "grad_norm": 0.007733174134045839,
-      "learning_rate": 0.0001999781088725023,
-      "loss": 46.0,
-      "step": 41370
-    },
-    {
-      "epoch": 6.662506542131326,
-      "grad_norm": 0.002514298539608717,
-      "learning_rate": 0.00019997810781396301,
-      "loss": 46.0,
-      "step": 41371
-    },
-    {
-      "epoch": 6.6626675792101135,
-      "grad_norm": 0.0024871560744941235,
-      "learning_rate": 0.00019997810675539808,
-      "loss": 46.0,
-      "step": 41372
-    },
-    {
-      "epoch": 6.6628286162889,
-      "grad_norm": 0.006003184709697962,
-      "learning_rate": 0.00019997810569680758,
-      "loss": 46.0,
-      "step": 41373
-    },
-    {
-      "epoch": 6.662989653367688,
-      "grad_norm": 0.012818343006074429,
-      "learning_rate": 0.00019997810463819146,
-      "loss": 46.0,
-      "step": 41374
-    },
-    {
-      "epoch": 6.663150690446475,
-      "grad_norm": 0.00463420944288373,
-      "learning_rate": 0.0001999781035795498,
-      "loss": 46.0,
-      "step": 41375
-    },
-    {
-      "epoch": 6.6633117275252625,
-      "grad_norm": 0.005430162884294987,
-      "learning_rate": 0.0001999781025208825,
-      "loss": 46.0,
-      "step": 41376
-    },
-    {
-      "epoch": 6.66347276460405,
-      "grad_norm": 0.0032714323606342077,
-      "learning_rate": 0.00019997810146218963,
-      "loss": 46.0,
-      "step": 41377
-    },
-    {
-      "epoch": 6.663633801682837,
-      "grad_norm": 0.003485821420326829,
-      "learning_rate": 0.00019997810040347117,
-      "loss": 46.0,
-      "step": 41378
-    },
-    {
-      "epoch": 6.663794838761625,
-      "grad_norm": 0.0026943886186927557,
-      "learning_rate": 0.00019997809934472712,
-      "loss": 46.0,
-      "step": 41379
-    },
-    {
-      "epoch": 6.663955875840412,
-      "grad_norm": 0.0012250575236976147,
-      "learning_rate": 0.0001999780982859575,
-      "loss": 46.0,
-      "step": 41380
-    },
-    {
-      "epoch": 6.6641169129192,
-      "grad_norm": 0.0010569618316367269,
-      "learning_rate": 0.00019997809722716226,
-      "loss": 46.0,
-      "step": 41381
-    },
-    {
-      "epoch": 6.664277949997987,
-      "grad_norm": 0.006684210617095232,
-      "learning_rate": 0.00019997809616834145,
-      "loss": 46.0,
-      "step": 41382
-    },
-    {
-      "epoch": 6.664438987076775,
-      "grad_norm": 0.0024920450523495674,
-      "learning_rate": 0.00019997809510949508,
-      "loss": 46.0,
-      "step": 41383
-    },
-    {
-      "epoch": 6.664600024155562,
-      "grad_norm": 0.00532752787694335,
-      "learning_rate": 0.00019997809405062307,
-      "loss": 46.0,
-      "step": 41384
-    },
-    {
-      "epoch": 6.6647610612343495,
-      "grad_norm": 0.004439402371644974,
-      "learning_rate": 0.0001999780929917255,
-      "loss": 46.0,
-      "step": 41385
-    },
-    {
-      "epoch": 6.664922098313137,
-      "grad_norm": 0.006457884330302477,
-      "learning_rate": 0.00019997809193280232,
-      "loss": 46.0,
-      "step": 41386
-    },
-    {
-      "epoch": 6.665083135391924,
-      "grad_norm": 0.013806230388581753,
-      "learning_rate": 0.00019997809087385354,
-      "loss": 46.0,
-      "step": 41387
-    },
-    {
-      "epoch": 6.665244172470711,
-      "grad_norm": 0.002769937738776207,
-      "learning_rate": 0.0001999780898148792,
-      "loss": 46.0,
-      "step": 41388
-    },
-    {
-      "epoch": 6.6654052095494984,
-      "grad_norm": 0.0013071835273876786,
-      "learning_rate": 0.00019997808875587926,
-      "loss": 46.0,
-      "step": 41389
-    },
-    {
-      "epoch": 6.665566246628286,
-      "grad_norm": 0.0019575459882616997,
-      "learning_rate": 0.00019997808769685375,
-      "loss": 46.0,
-      "step": 41390
-    },
-    {
-      "epoch": 6.665727283707073,
-      "grad_norm": 0.0015985085628926754,
-      "learning_rate": 0.00019997808663780263,
-      "loss": 46.0,
-      "step": 41391
-    },
-    {
-      "epoch": 6.665888320785861,
-      "grad_norm": 0.009649241343140602,
-      "learning_rate": 0.00019997808557872592,
-      "loss": 46.0,
-      "step": 41392
-    },
-    {
-      "epoch": 6.666049357864648,
-      "grad_norm": 0.0017637225100770593,
-      "learning_rate": 0.00019997808451962366,
-      "loss": 46.0,
-      "step": 41393
-    },
-    {
-      "epoch": 6.666210394943436,
-      "grad_norm": 0.004917894024401903,
-      "learning_rate": 0.00019997808346049575,
-      "loss": 46.0,
-      "step": 41394
-    },
-    {
-      "epoch": 6.666371432022223,
-      "grad_norm": 0.007446544244885445,
-      "learning_rate": 0.00019997808240134228,
-      "loss": 46.0,
-      "step": 41395
-    },
-    {
-      "epoch": 6.666532469101011,
-      "grad_norm": 0.0043029217049479485,
-      "learning_rate": 0.00019997808134216322,
-      "loss": 46.0,
-      "step": 41396
-    },
-    {
-      "epoch": 6.666693506179798,
-      "grad_norm": 0.001443596906028688,
-      "learning_rate": 0.00019997808028295857,
-      "loss": 46.0,
-      "step": 41397
-    },
-    {
-      "epoch": 6.6668545432585855,
-      "grad_norm": 0.0046377237886190414,
-      "learning_rate": 0.00019997807922372834,
-      "loss": 46.0,
-      "step": 41398
-    },
-    {
-      "epoch": 6.667015580337373,
-      "grad_norm": 0.008895192295312881,
-      "learning_rate": 0.0001999780781644725,
-      "loss": 46.0,
-      "step": 41399
-    },
-    {
-      "epoch": 6.66717661741616,
-      "grad_norm": 0.0017852013697847724,
-      "learning_rate": 0.00019997807710519106,
-      "loss": 46.0,
-      "step": 41400
-    },
-    {
-      "epoch": 6.667337654494947,
-      "grad_norm": 0.009846286848187447,
-      "learning_rate": 0.00019997807604588407,
-      "loss": 46.0,
-      "step": 41401
-    },
-    {
-      "epoch": 6.667498691573735,
-      "grad_norm": 0.0038804770447313786,
-      "learning_rate": 0.0001999780749865515,
-      "loss": 46.0,
-      "step": 41402
-    },
-    {
-      "epoch": 6.667659728652522,
-      "grad_norm": 0.003125979797914624,
-      "learning_rate": 0.0001999780739271933,
-      "loss": 46.0,
-      "step": 41403
-    },
-    {
-      "epoch": 6.667820765731309,
-      "grad_norm": 0.01097989734262228,
-      "learning_rate": 0.0001999780728678095,
-      "loss": 46.0,
-      "step": 41404
-    },
-    {
-      "epoch": 6.667981802810097,
-      "grad_norm": 0.003038732334971428,
-      "learning_rate": 0.00019997807180840017,
-      "loss": 46.0,
-      "step": 41405
-    },
-    {
-      "epoch": 6.668142839888884,
-      "grad_norm": 0.00931643694639206,
-      "learning_rate": 0.0001999780707489652,
-      "loss": 46.0,
-      "step": 41406
-    },
-    {
-      "epoch": 6.668303876967672,
-      "grad_norm": 0.00387683417648077,
-      "learning_rate": 0.00019997806968950464,
-      "loss": 46.0,
-      "step": 41407
-    },
-    {
-      "epoch": 6.668464914046459,
-      "grad_norm": 0.0029038600623607635,
-      "learning_rate": 0.0001999780686300185,
-      "loss": 46.0,
-      "step": 41408
-    },
-    {
-      "epoch": 6.668625951125247,
-      "grad_norm": 0.002887092065066099,
-      "learning_rate": 0.0001999780675705068,
-      "loss": 46.0,
-      "step": 41409
-    },
-    {
-      "epoch": 6.668786988204034,
-      "grad_norm": 0.005947868339717388,
-      "learning_rate": 0.0001999780665109695,
-      "loss": 46.0,
-      "step": 41410
-    },
-    {
-      "epoch": 6.6689480252828215,
-      "grad_norm": 0.0013755843974649906,
-      "learning_rate": 0.00019997806545140657,
-      "loss": 46.0,
-      "step": 41411
-    },
-    {
-      "epoch": 6.669109062361609,
-      "grad_norm": 0.007008018903434277,
-      "learning_rate": 0.0001999780643918181,
-      "loss": 46.0,
-      "step": 41412
-    },
-    {
-      "epoch": 6.669270099440396,
-      "grad_norm": 0.0030276132747530937,
-      "learning_rate": 0.00019997806333220403,
-      "loss": 46.0,
-      "step": 41413
-    },
-    {
-      "epoch": 6.669431136519184,
-      "grad_norm": 0.009313974529504776,
-      "learning_rate": 0.00019997806227256435,
-      "loss": 46.0,
-      "step": 41414
-    },
-    {
-      "epoch": 6.669592173597971,
-      "grad_norm": 0.0027604824863374233,
-      "learning_rate": 0.00019997806121289908,
-      "loss": 46.0,
-      "step": 41415
-    },
-    {
-      "epoch": 6.669753210676758,
-      "grad_norm": 0.006833921652287245,
-      "learning_rate": 0.00019997806015320822,
-      "loss": 46.0,
-      "step": 41416
-    },
-    {
-      "epoch": 6.669914247755546,
-      "grad_norm": 0.0011460887035354972,
-      "learning_rate": 0.00019997805909349178,
-      "loss": 46.0,
-      "step": 41417
-    },
-    {
-      "epoch": 6.670075284834333,
-      "grad_norm": 0.002286622766405344,
-      "learning_rate": 0.00019997805803374975,
-      "loss": 46.0,
-      "step": 41418
-    },
-    {
-      "epoch": 6.67023632191312,
-      "grad_norm": 0.002973241964355111,
-      "learning_rate": 0.00019997805697398216,
-      "loss": 46.0,
-      "step": 41419
-    },
-    {
-      "epoch": 6.670397358991908,
-      "grad_norm": 0.010130366310477257,
-      "learning_rate": 0.00019997805591418896,
-      "loss": 46.0,
-      "step": 41420
-    },
-    {
-      "epoch": 6.670558396070695,
-      "grad_norm": 0.0011209291405975819,
-      "learning_rate": 0.00019997805485437017,
-      "loss": 46.0,
-      "step": 41421
-    },
-    {
-      "epoch": 6.670719433149483,
-      "grad_norm": 0.004139076452702284,
-      "learning_rate": 0.00019997805379452576,
-      "loss": 46.0,
-      "step": 41422
-    },
-    {
-      "epoch": 6.67088047022827,
-      "grad_norm": 0.006929545197635889,
-      "learning_rate": 0.0001999780527346558,
-      "loss": 46.0,
-      "step": 41423
-    },
-    {
-      "epoch": 6.6710415073070575,
-      "grad_norm": 0.002197990193963051,
-      "learning_rate": 0.00019997805167476024,
-      "loss": 46.0,
-      "step": 41424
-    },
-    {
-      "epoch": 6.671202544385845,
-      "grad_norm": 0.00770388450473547,
-      "learning_rate": 0.0001999780506148391,
-      "loss": 46.0,
-      "step": 41425
-    },
-    {
-      "epoch": 6.671363581464632,
-      "grad_norm": 0.003448836738243699,
-      "learning_rate": 0.00019997804955489238,
-      "loss": 46.0,
-      "step": 41426
-    },
-    {
-      "epoch": 6.67152461854342,
-      "grad_norm": 0.0026357972528785467,
-      "learning_rate": 0.00019997804849492004,
-      "loss": 46.0,
-      "step": 41427
-    },
-    {
-      "epoch": 6.671685655622207,
-      "grad_norm": 0.0041018882766366005,
-      "learning_rate": 0.00019997804743492214,
-      "loss": 46.0,
-      "step": 41428
-    },
-    {
-      "epoch": 6.671846692700995,
-      "grad_norm": 0.0043464601039886475,
-      "learning_rate": 0.0001999780463748986,
-      "loss": 46.0,
-      "step": 41429
-    },
-    {
-      "epoch": 6.672007729779782,
-      "grad_norm": 0.002962283091619611,
-      "learning_rate": 0.00019997804531484952,
-      "loss": 46.0,
-      "step": 41430
-    },
-    {
-      "epoch": 6.672168766858569,
-      "grad_norm": 0.0074380990117788315,
-      "learning_rate": 0.00019997804425477483,
-      "loss": 46.0,
-      "step": 41431
-    },
-    {
-      "epoch": 6.672329803937356,
-      "grad_norm": 0.002464896999299526,
-      "learning_rate": 0.00019997804319467455,
-      "loss": 46.0,
-      "step": 41432
-    },
-    {
-      "epoch": 6.672490841016144,
-      "grad_norm": 0.003040800103917718,
-      "learning_rate": 0.0001999780421345487,
-      "loss": 46.0,
-      "step": 41433
-    },
-    {
-      "epoch": 6.672651878094931,
-      "grad_norm": 0.009164533577859402,
-      "learning_rate": 0.0001999780410743972,
-      "loss": 46.0,
-      "step": 41434
-    },
-    {
-      "epoch": 6.6728129151737186,
-      "grad_norm": 0.006473027169704437,
-      "learning_rate": 0.00019997804001422017,
-      "loss": 46.0,
-      "step": 41435
-    },
-    {
-      "epoch": 6.672973952252506,
-      "grad_norm": 0.002698664553463459,
-      "learning_rate": 0.00019997803895401757,
-      "loss": 46.0,
-      "step": 41436
-    },
-    {
-      "epoch": 6.6731349893312935,
-      "grad_norm": 0.009651648811995983,
-      "learning_rate": 0.00019997803789378933,
-      "loss": 46.0,
-      "step": 41437
-    },
-    {
-      "epoch": 6.673296026410081,
-      "grad_norm": 0.007370231673121452,
-      "learning_rate": 0.0001999780368335355,
-      "loss": 46.0,
-      "step": 41438
-    },
-    {
-      "epoch": 6.673457063488868,
-      "grad_norm": 0.0070519051514565945,
-      "learning_rate": 0.00019997803577325614,
-      "loss": 46.0,
-      "step": 41439
-    },
-    {
-      "epoch": 6.673618100567656,
-      "grad_norm": 0.002306445501744747,
-      "learning_rate": 0.00019997803471295114,
-      "loss": 46.0,
-      "step": 41440
-    },
-    {
-      "epoch": 6.673779137646443,
-      "grad_norm": 0.009618941694498062,
-      "learning_rate": 0.00019997803365262055,
-      "loss": 46.0,
-      "step": 41441
-    },
-    {
-      "epoch": 6.673940174725231,
-      "grad_norm": 0.008067110553383827,
-      "learning_rate": 0.0001999780325922644,
-      "loss": 46.0,
-      "step": 41442
-    },
-    {
-      "epoch": 6.674101211804018,
-      "grad_norm": 0.01169788371771574,
-      "learning_rate": 0.00019997803153188262,
-      "loss": 46.0,
-      "step": 41443
-    },
-    {
-      "epoch": 6.674262248882806,
-      "grad_norm": 0.009598050266504288,
-      "learning_rate": 0.0001999780304714753,
-      "loss": 46.0,
-      "step": 41444
-    },
-    {
-      "epoch": 6.674423285961593,
-      "grad_norm": 0.01065739430487156,
-      "learning_rate": 0.00019997802941104236,
-      "loss": 46.0,
-      "step": 41445
-    },
-    {
-      "epoch": 6.67458432304038,
-      "grad_norm": 0.006738572847098112,
-      "learning_rate": 0.00019997802835058383,
-      "loss": 46.0,
-      "step": 41446
-    },
-    {
-      "epoch": 6.674745360119167,
-      "grad_norm": 0.0030417251400649548,
-      "learning_rate": 0.00019997802729009972,
-      "loss": 46.0,
-      "step": 41447
-    },
-    {
-      "epoch": 6.6749063971979545,
-      "grad_norm": 0.0015606951201334596,
-      "learning_rate": 0.00019997802622959002,
-      "loss": 46.0,
-      "step": 41448
-    },
-    {
-      "epoch": 6.675067434276742,
-      "grad_norm": 0.012461822479963303,
-      "learning_rate": 0.0001999780251690547,
-      "loss": 46.0,
-      "step": 41449
-    },
-    {
-      "epoch": 6.675228471355529,
-      "grad_norm": 0.01988958939909935,
-      "learning_rate": 0.00019997802410849384,
-      "loss": 46.0,
-      "step": 41450
-    },
-    {
-      "epoch": 6.675389508434317,
-      "grad_norm": 0.005897939670830965,
-      "learning_rate": 0.00019997802304790735,
-      "loss": 46.0,
-      "step": 41451
-    },
-    {
-      "epoch": 6.675550545513104,
-      "grad_norm": 0.0061839912086725235,
-      "learning_rate": 0.00019997802198729533,
-      "loss": 46.0,
-      "step": 41452
-    },
-    {
-      "epoch": 6.675711582591892,
-      "grad_norm": 0.001188946538604796,
-      "learning_rate": 0.00019997802092665764,
-      "loss": 46.0,
-      "step": 41453
-    },
-    {
-      "epoch": 6.675872619670679,
-      "grad_norm": 0.007058894261717796,
-      "learning_rate": 0.0001999780198659944,
-      "loss": 46.0,
-      "step": 41454
-    },
-    {
-      "epoch": 6.676033656749467,
-      "grad_norm": 0.0012896310072392225,
-      "learning_rate": 0.00019997801880530558,
-      "loss": 46.0,
-      "step": 41455
-    },
-    {
-      "epoch": 6.676194693828254,
-      "grad_norm": 0.002240738831460476,
-      "learning_rate": 0.00019997801774459119,
-      "loss": 46.0,
-      "step": 41456
-    },
-    {
-      "epoch": 6.676355730907042,
-      "grad_norm": 0.001775287906639278,
-      "learning_rate": 0.00019997801668385117,
-      "loss": 46.0,
-      "step": 41457
-    },
-    {
-      "epoch": 6.676516767985829,
-      "grad_norm": 0.005676504224538803,
-      "learning_rate": 0.00019997801562308558,
-      "loss": 46.0,
-      "step": 41458
-    },
-    {
-      "epoch": 6.6766778050646165,
-      "grad_norm": 0.004192544613033533,
-      "learning_rate": 0.0001999780145622944,
-      "loss": 46.0,
-      "step": 41459
-    },
-    {
-      "epoch": 6.676838842143404,
-      "grad_norm": 0.021272780373692513,
-      "learning_rate": 0.00019997801350147762,
-      "loss": 46.0,
-      "step": 41460
-    },
-    {
-      "epoch": 6.6769998792221905,
-      "grad_norm": 0.004423180595040321,
-      "learning_rate": 0.00019997801244063523,
-      "loss": 46.0,
-      "step": 41461
-    },
-    {
-      "epoch": 6.677160916300978,
-      "grad_norm": 0.0009819597471505404,
-      "learning_rate": 0.00019997801137976732,
-      "loss": 46.0,
-      "step": 41462
-    },
-    {
-      "epoch": 6.677321953379765,
-      "grad_norm": 0.009840531274676323,
-      "learning_rate": 0.00019997801031887375,
-      "loss": 46.0,
-      "step": 41463
-    },
-    {
-      "epoch": 6.677482990458553,
-      "grad_norm": 0.01274296548217535,
-      "learning_rate": 0.00019997800925795463,
-      "loss": 46.0,
-      "step": 41464
-    },
-    {
-      "epoch": 6.67764402753734,
-      "grad_norm": 0.0029992652125656605,
-      "learning_rate": 0.00019997800819700993,
-      "loss": 46.0,
-      "step": 41465
-    },
-    {
-      "epoch": 6.677805064616128,
-      "grad_norm": 0.0011716948356479406,
-      "learning_rate": 0.00019997800713603958,
-      "loss": 46.0,
-      "step": 41466
-    },
-    {
-      "epoch": 6.677966101694915,
-      "grad_norm": 0.0017167023615911603,
-      "learning_rate": 0.0001999780060750437,
-      "loss": 46.0,
-      "step": 41467
-    },
-    {
-      "epoch": 6.678127138773703,
-      "grad_norm": 0.005630992352962494,
-      "learning_rate": 0.00019997800501402223,
-      "loss": 46.0,
-      "step": 41468
-    },
-    {
-      "epoch": 6.67828817585249,
-      "grad_norm": 0.006417365744709969,
-      "learning_rate": 0.00019997800395297512,
-      "loss": 46.0,
-      "step": 41469
-    },
-    {
-      "epoch": 6.678449212931278,
-      "grad_norm": 0.008727450855076313,
-      "learning_rate": 0.00019997800289190244,
-      "loss": 46.0,
-      "step": 41470
-    },
-    {
-      "epoch": 6.678610250010065,
-      "grad_norm": 0.004421089310199022,
-      "learning_rate": 0.00019997800183080421,
-      "loss": 46.0,
-      "step": 41471
-    },
-    {
-      "epoch": 6.6787712870888525,
-      "grad_norm": 0.0031029530800879,
-      "learning_rate": 0.00019997800076968037,
-      "loss": 46.0,
-      "step": 41472
-    },
-    {
-      "epoch": 6.67893232416764,
-      "grad_norm": 0.006378788035362959,
-      "learning_rate": 0.00019997799970853094,
-      "loss": 46.0,
-      "step": 41473
-    },
-    {
-      "epoch": 6.6790933612464265,
-      "grad_norm": 0.0020963973365724087,
-      "learning_rate": 0.00019997799864735592,
-      "loss": 46.0,
-      "step": 41474
-    },
-    {
-      "epoch": 6.679254398325215,
-      "grad_norm": 0.002083128783851862,
-      "learning_rate": 0.00019997799758615528,
-      "loss": 46.0,
-      "step": 41475
-    },
-    {
-      "epoch": 6.679415435404001,
-      "grad_norm": 0.003655440639704466,
-      "learning_rate": 0.0001999779965249291,
-      "loss": 46.0,
-      "step": 41476
-    },
-    {
-      "epoch": 6.679576472482789,
-      "grad_norm": 0.028048155829310417,
-      "learning_rate": 0.0001999779954636773,
-      "loss": 46.0,
-      "step": 41477
-    },
-    {
-      "epoch": 6.679737509561576,
-      "grad_norm": 0.002458154456689954,
-      "learning_rate": 0.00019997799440239991,
-      "loss": 46.0,
-      "step": 41478
-    },
-    {
-      "epoch": 6.679898546640364,
-      "grad_norm": 0.007947004400193691,
-      "learning_rate": 0.00019997799334109696,
-      "loss": 46.0,
-      "step": 41479
-    },
-    {
-      "epoch": 6.680059583719151,
-      "grad_norm": 0.0083960285410285,
-      "learning_rate": 0.0001999779922797684,
-      "loss": 46.0,
-      "step": 41480
-    },
-    {
-      "epoch": 6.680220620797939,
-      "grad_norm": 0.00619344599545002,
-      "learning_rate": 0.00019997799121841423,
-      "loss": 46.0,
-      "step": 41481
-    },
-    {
-      "epoch": 6.680381657876726,
-      "grad_norm": 0.0034263241104781628,
-      "learning_rate": 0.0001999779901570345,
-      "loss": 46.0,
-      "step": 41482
-    },
-    {
-      "epoch": 6.680542694955514,
-      "grad_norm": 0.0025248266756534576,
-      "learning_rate": 0.00019997798909562916,
-      "loss": 46.0,
-      "step": 41483
-    },
-    {
-      "epoch": 6.680703732034301,
-      "grad_norm": 0.01644541509449482,
-      "learning_rate": 0.00019997798803419827,
-      "loss": 46.0,
-      "step": 41484
-    },
-    {
-      "epoch": 6.6808647691130885,
-      "grad_norm": 0.0025362572632730007,
-      "learning_rate": 0.00019997798697274176,
-      "loss": 46.0,
-      "step": 41485
-    },
-    {
-      "epoch": 6.681025806191876,
-      "grad_norm": 0.0027080951258540154,
-      "learning_rate": 0.00019997798591125964,
-      "loss": 46.0,
-      "step": 41486
-    },
-    {
-      "epoch": 6.681186843270663,
-      "grad_norm": 0.0037105679512023926,
-      "learning_rate": 0.00019997798484975196,
-      "loss": 46.0,
-      "step": 41487
-    },
-    {
-      "epoch": 6.681347880349451,
-      "grad_norm": 0.005108132492750883,
-      "learning_rate": 0.0001999779837882187,
-      "loss": 46.0,
-      "step": 41488
-    },
-    {
-      "epoch": 6.681508917428237,
-      "grad_norm": 0.009482454508543015,
-      "learning_rate": 0.00019997798272665984,
-      "loss": 46.0,
-      "step": 41489
-    },
-    {
-      "epoch": 6.681669954507026,
-      "grad_norm": 0.0033914861269295216,
-      "learning_rate": 0.0001999779816650754,
-      "loss": 46.0,
-      "step": 41490
-    },
-    {
-      "epoch": 6.681830991585812,
-      "grad_norm": 0.0020544622093439102,
-      "learning_rate": 0.00019997798060346534,
-      "loss": 46.0,
-      "step": 41491
-    },
-    {
-      "epoch": 6.6819920286646,
-      "grad_norm": 0.0039254650473594666,
-      "learning_rate": 0.00019997797954182973,
-      "loss": 46.0,
-      "step": 41492
-    },
-    {
-      "epoch": 6.682153065743387,
-      "grad_norm": 0.002754541812464595,
-      "learning_rate": 0.0001999779784801685,
-      "loss": 46.0,
-      "step": 41493
-    },
-    {
-      "epoch": 6.682314102822175,
-      "grad_norm": 0.0027967761270701885,
-      "learning_rate": 0.0001999779774184817,
-      "loss": 46.0,
-      "step": 41494
-    },
-    {
-      "epoch": 6.682475139900962,
-      "grad_norm": 0.007121903821825981,
-      "learning_rate": 0.0001999779763567693,
-      "loss": 46.0,
-      "step": 41495
-    },
-    {
-      "epoch": 6.6826361769797495,
-      "grad_norm": 0.004999418742954731,
-      "learning_rate": 0.0001999779752950313,
-      "loss": 46.0,
-      "step": 41496
-    },
-    {
-      "epoch": 6.682797214058537,
-      "grad_norm": 0.00398630416020751,
-      "learning_rate": 0.00019997797423326773,
-      "loss": 46.0,
-      "step": 41497
-    },
-    {
-      "epoch": 6.6829582511373244,
-      "grad_norm": 0.0035232247319072485,
-      "learning_rate": 0.00019997797317147857,
-      "loss": 46.0,
-      "step": 41498
-    },
-    {
-      "epoch": 6.683119288216112,
-      "grad_norm": 0.008806174620985985,
-      "learning_rate": 0.00019997797210966382,
-      "loss": 46.0,
-      "step": 41499
-    },
-    {
-      "epoch": 6.683280325294899,
-      "grad_norm": 0.0030216407030820847,
-      "learning_rate": 0.00019997797104782348,
-      "loss": 46.0,
-      "step": 41500
-    },
-    {
-      "epoch": 6.683441362373687,
-      "grad_norm": 0.007164289243519306,
-      "learning_rate": 0.00019997796998595755,
-      "loss": 46.0,
-      "step": 41501
-    },
-    {
-      "epoch": 6.683602399452474,
-      "grad_norm": 0.007627043407410383,
-      "learning_rate": 0.000199977968924066,
-      "loss": 46.0,
-      "step": 41502
-    },
-    {
-      "epoch": 6.683763436531262,
-      "grad_norm": 0.006922471337020397,
-      "learning_rate": 0.0001999779678621489,
-      "loss": 46.0,
-      "step": 41503
-    },
-    {
-      "epoch": 6.683924473610048,
-      "grad_norm": 0.0027551406528800726,
-      "learning_rate": 0.0001999779668002062,
-      "loss": 46.0,
-      "step": 41504
-    },
-    {
-      "epoch": 6.684085510688836,
-      "grad_norm": 0.0064051453955471516,
-      "learning_rate": 0.00019997796573823791,
-      "loss": 46.0,
-      "step": 41505
-    },
-    {
-      "epoch": 6.684246547767623,
-      "grad_norm": 0.0018681363435462117,
-      "learning_rate": 0.00019997796467624402,
-      "loss": 46.0,
-      "step": 41506
-    },
-    {
-      "epoch": 6.684407584846411,
-      "grad_norm": 0.0018531293608248234,
-      "learning_rate": 0.00019997796361422457,
-      "loss": 46.0,
-      "step": 41507
-    },
-    {
-      "epoch": 6.684568621925198,
-      "grad_norm": 0.011441242881119251,
-      "learning_rate": 0.0001999779625521795,
-      "loss": 46.0,
-      "step": 41508
-    },
-    {
-      "epoch": 6.6847296590039855,
-      "grad_norm": 0.0034289825707674026,
-      "learning_rate": 0.00019997796149010886,
-      "loss": 46.0,
-      "step": 41509
-    },
-    {
-      "epoch": 6.684890696082773,
-      "grad_norm": 0.0196436308324337,
-      "learning_rate": 0.00019997796042801262,
-      "loss": 46.0,
-      "step": 41510
-    },
-    {
-      "epoch": 6.68505173316156,
-      "grad_norm": 0.00993046909570694,
-      "learning_rate": 0.00019997795936589082,
-      "loss": 46.0,
-      "step": 41511
-    },
-    {
-      "epoch": 6.685212770240348,
-      "grad_norm": 0.0011435869382694364,
-      "learning_rate": 0.00019997795830374338,
-      "loss": 46.0,
-      "step": 41512
-    },
-    {
-      "epoch": 6.685373807319135,
-      "grad_norm": 0.012656190432608128,
-      "learning_rate": 0.0001999779572415704,
-      "loss": 46.0,
-      "step": 41513
-    },
-    {
-      "epoch": 6.685534844397923,
-      "grad_norm": 0.02194136008620262,
-      "learning_rate": 0.00019997795617937182,
-      "loss": 46.0,
-      "step": 41514
-    },
-    {
-      "epoch": 6.68569588147671,
-      "grad_norm": 0.001996238250285387,
-      "learning_rate": 0.00019997795511714762,
-      "loss": 46.0,
-      "step": 41515
-    },
-    {
-      "epoch": 6.685856918555498,
-      "grad_norm": 0.0031546689569950104,
-      "learning_rate": 0.00019997795405489783,
-      "loss": 46.0,
-      "step": 41516
-    },
-    {
-      "epoch": 6.686017955634285,
-      "grad_norm": 0.002523136092349887,
-      "learning_rate": 0.00019997795299262248,
-      "loss": 46.0,
-      "step": 41517
-    },
-    {
-      "epoch": 6.686178992713073,
-      "grad_norm": 0.0032731450628489256,
-      "learning_rate": 0.00019997795193032154,
-      "loss": 46.0,
-      "step": 41518
-    },
-    {
-      "epoch": 6.686340029791859,
-      "grad_norm": 0.003066750941798091,
-      "learning_rate": 0.000199977950867995,
-      "loss": 46.0,
-      "step": 41519
-    },
-    {
-      "epoch": 6.686501066870647,
-      "grad_norm": 0.007604025304317474,
-      "learning_rate": 0.00019997794980564285,
-      "loss": 46.0,
-      "step": 41520
-    },
-    {
-      "epoch": 6.686662103949434,
-      "grad_norm": 0.010633991099894047,
-      "learning_rate": 0.00019997794874326513,
-      "loss": 46.0,
-      "step": 41521
-    },
-    {
-      "epoch": 6.6868231410282215,
-      "grad_norm": 0.0023399614728987217,
-      "learning_rate": 0.00019997794768086184,
-      "loss": 46.0,
-      "step": 41522
-    },
-    {
-      "epoch": 6.686984178107009,
-      "grad_norm": 0.008343911729753017,
-      "learning_rate": 0.00019997794661843295,
-      "loss": 46.0,
-      "step": 41523
-    },
-    {
-      "epoch": 6.687145215185796,
-      "grad_norm": 0.0035168544854968786,
-      "learning_rate": 0.00019997794555597846,
-      "loss": 46.0,
-      "step": 41524
-    },
-    {
-      "epoch": 6.687306252264584,
-      "grad_norm": 0.0018394142389297485,
-      "learning_rate": 0.00019997794449349839,
-      "loss": 46.0,
-      "step": 41525
-    },
-    {
-      "epoch": 6.687467289343371,
-      "grad_norm": 0.003639424219727516,
-      "learning_rate": 0.00019997794343099272,
-      "loss": 46.0,
-      "step": 41526
-    },
-    {
-      "epoch": 6.687628326422159,
-      "grad_norm": 0.00887287687510252,
-      "learning_rate": 0.00019997794236846148,
-      "loss": 46.0,
-      "step": 41527
-    },
-    {
-      "epoch": 6.687789363500946,
-      "grad_norm": 0.0015143896453082561,
-      "learning_rate": 0.00019997794130590464,
-      "loss": 46.0,
-      "step": 41528
-    },
-    {
-      "epoch": 6.687950400579734,
-      "grad_norm": 0.006364369299262762,
-      "learning_rate": 0.0001999779402433222,
-      "loss": 46.0,
-      "step": 41529
-    },
-    {
-      "epoch": 6.688111437658521,
-      "grad_norm": 0.004334803204983473,
-      "learning_rate": 0.00019997793918071416,
-      "loss": 46.0,
-      "step": 41530
-    },
-    {
-      "epoch": 6.688272474737309,
-      "grad_norm": 0.0014573076041415334,
-      "learning_rate": 0.00019997793811808056,
-      "loss": 46.0,
-      "step": 41531
-    },
-    {
-      "epoch": 6.688433511816096,
-      "grad_norm": 0.00799465086311102,
-      "learning_rate": 0.00019997793705542138,
-      "loss": 46.0,
-      "step": 41532
-    },
-    {
-      "epoch": 6.6885945488948835,
-      "grad_norm": 0.001455390825867653,
-      "learning_rate": 0.00019997793599273658,
-      "loss": 46.0,
-      "step": 41533
-    },
-    {
-      "epoch": 6.68875558597367,
-      "grad_norm": 0.010175321251153946,
-      "learning_rate": 0.0001999779349300262,
-      "loss": 46.0,
-      "step": 41534
-    },
-    {
-      "epoch": 6.6889166230524575,
-      "grad_norm": 0.0028319982811808586,
-      "learning_rate": 0.00019997793386729025,
-      "loss": 46.0,
-      "step": 41535
-    },
-    {
-      "epoch": 6.689077660131245,
-      "grad_norm": 0.0052672214806079865,
-      "learning_rate": 0.0001999779328045287,
-      "loss": 46.0,
-      "step": 41536
-    },
-    {
-      "epoch": 6.689238697210032,
-      "grad_norm": 0.017750782892107964,
-      "learning_rate": 0.00019997793174174154,
-      "loss": 46.0,
-      "step": 41537
-    },
-    {
-      "epoch": 6.68939973428882,
-      "grad_norm": 0.010550031438469887,
-      "learning_rate": 0.0001999779306789288,
-      "loss": 46.0,
-      "step": 41538
-    },
-    {
-      "epoch": 6.689560771367607,
-      "grad_norm": 0.00702440133318305,
-      "learning_rate": 0.00019997792961609049,
-      "loss": 46.0,
-      "step": 41539
-    },
-    {
-      "epoch": 6.689721808446395,
-      "grad_norm": 0.00806315429508686,
-      "learning_rate": 0.00019997792855322658,
-      "loss": 46.0,
-      "step": 41540
-    },
-    {
-      "epoch": 6.689882845525182,
-      "grad_norm": 0.004118355456739664,
-      "learning_rate": 0.00019997792749033706,
-      "loss": 46.0,
-      "step": 41541
-    },
-    {
-      "epoch": 6.69004388260397,
-      "grad_norm": 0.00675570871680975,
-      "learning_rate": 0.00019997792642742197,
-      "loss": 46.0,
-      "step": 41542
-    },
-    {
-      "epoch": 6.690204919682757,
-      "grad_norm": 0.003215635195374489,
-      "learning_rate": 0.0001999779253644813,
-      "loss": 46.0,
-      "step": 41543
-    },
-    {
-      "epoch": 6.6903659567615446,
-      "grad_norm": 0.0011074589565396309,
-      "learning_rate": 0.00019997792430151502,
-      "loss": 46.0,
-      "step": 41544
-    },
-    {
-      "epoch": 6.690526993840332,
-      "grad_norm": 0.003002199809998274,
-      "learning_rate": 0.00019997792323852317,
-      "loss": 46.0,
-      "step": 41545
-    },
-    {
-      "epoch": 6.6906880309191195,
-      "grad_norm": 0.003767826361581683,
-      "learning_rate": 0.00019997792217550572,
-      "loss": 46.0,
-      "step": 41546
-    },
-    {
-      "epoch": 6.690849067997906,
-      "grad_norm": 0.008559972047805786,
-      "learning_rate": 0.0001999779211124627,
-      "loss": 46.0,
-      "step": 41547
-    },
-    {
-      "epoch": 6.691010105076694,
-      "grad_norm": 0.01038906816393137,
-      "learning_rate": 0.00019997792004939404,
-      "loss": 46.0,
-      "step": 41548
-    },
-    {
-      "epoch": 6.691171142155481,
-      "grad_norm": 0.00894656777381897,
-      "learning_rate": 0.00019997791898629985,
-      "loss": 46.0,
-      "step": 41549
-    },
-    {
-      "epoch": 6.691332179234268,
-      "grad_norm": 0.002707582665607333,
-      "learning_rate": 0.00019997791792318004,
-      "loss": 46.0,
-      "step": 41550
-    }
-  ],
-  "logging_steps": 1,
-  "max_steps": 6209750,
-  "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": 829708196511744.0,
-  "train_batch_size": 2,
-  "trial_name": null,
-  "trial_params": null
-}
+version https://git-lfs.github.com/spec/v1
+oid sha256:142d481d5f81d6911ad6da3cb2e30569135f1d5ed220426e31d4e642bba94e24
+size 11039870